diff --git "a/checkpoint-2049/trainer_state.json" "b/checkpoint-2049/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2049/trainer_state.json"
@@ -0,0 +1,14376 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998373454782042,
+  "eval_steps": 500,
+  "global_step": 2049,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00048796356538711777,
+      "grad_norm": 0.4446243345737457,
+      "learning_rate": 0.0001,
+      "loss": 1.8998,
+      "step": 1
+    },
+    {
+      "epoch": 0.0009759271307742355,
+      "grad_norm": 0.443472683429718,
+      "learning_rate": 0.0001,
+      "loss": 2.146,
+      "step": 2
+    },
+    {
+      "epoch": 0.0014638906961613532,
+      "grad_norm": 0.246729776263237,
+      "learning_rate": 0.0001,
+      "loss": 1.8931,
+      "step": 3
+    },
+    {
+      "epoch": 0.001951854261548471,
+      "grad_norm": 0.3018186688423157,
+      "learning_rate": 0.0001,
+      "loss": 1.984,
+      "step": 4
+    },
+    {
+      "epoch": 0.002439817826935589,
+      "grad_norm": 0.2850761413574219,
+      "learning_rate": 0.0001,
+      "loss": 1.863,
+      "step": 5
+    },
+    {
+      "epoch": 0.0029277813923227064,
+      "grad_norm": 0.23705212771892548,
+      "learning_rate": 0.0001,
+      "loss": 1.8384,
+      "step": 6
+    },
+    {
+      "epoch": 0.0034157449577098243,
+      "grad_norm": 0.24392390251159668,
+      "learning_rate": 0.0001,
+      "loss": 1.8743,
+      "step": 7
+    },
+    {
+      "epoch": 0.003903708523096942,
+      "grad_norm": 0.24215014278888702,
+      "learning_rate": 0.0001,
+      "loss": 1.8048,
+      "step": 8
+    },
+    {
+      "epoch": 0.00439167208848406,
+      "grad_norm": 0.22235405445098877,
+      "learning_rate": 0.0001,
+      "loss": 1.8098,
+      "step": 9
+    },
+    {
+      "epoch": 0.004879635653871178,
+      "grad_norm": 0.1880388706922531,
+      "learning_rate": 0.0001,
+      "loss": 1.7519,
+      "step": 10
+    },
+    {
+      "epoch": 0.005367599219258295,
+      "grad_norm": 0.2197292149066925,
+      "learning_rate": 0.0001,
+      "loss": 1.905,
+      "step": 11
+    },
+    {
+      "epoch": 0.005855562784645413,
+      "grad_norm": 0.20583945512771606,
+      "learning_rate": 0.0001,
+      "loss": 1.8143,
+      "step": 12
+    },
+    {
+      "epoch": 0.006343526350032531,
+      "grad_norm": 0.20737111568450928,
+      "learning_rate": 0.0001,
+      "loss": 1.8505,
+      "step": 13
+    },
+    {
+      "epoch": 0.0068314899154196486,
+      "grad_norm": 0.19384053349494934,
+      "learning_rate": 0.0001,
+      "loss": 1.7528,
+      "step": 14
+    },
+    {
+      "epoch": 0.007319453480806766,
+      "grad_norm": 0.23753000795841217,
+      "learning_rate": 0.0001,
+      "loss": 1.7206,
+      "step": 15
+    },
+    {
+      "epoch": 0.007807417046193884,
+      "grad_norm": 0.1946115642786026,
+      "learning_rate": 0.0001,
+      "loss": 1.7562,
+      "step": 16
+    },
+    {
+      "epoch": 0.008295380611581003,
+      "grad_norm": 0.18985839188098907,
+      "learning_rate": 0.0001,
+      "loss": 1.6665,
+      "step": 17
+    },
+    {
+      "epoch": 0.00878334417696812,
+      "grad_norm": 0.20499983429908752,
+      "learning_rate": 0.0001,
+      "loss": 1.9491,
+      "step": 18
+    },
+    {
+      "epoch": 0.009271307742355238,
+      "grad_norm": 0.1874532699584961,
+      "learning_rate": 0.0001,
+      "loss": 1.7975,
+      "step": 19
+    },
+    {
+      "epoch": 0.009759271307742356,
+      "grad_norm": 0.18048429489135742,
+      "learning_rate": 0.0001,
+      "loss": 1.7799,
+      "step": 20
+    },
+    {
+      "epoch": 0.010247234873129472,
+      "grad_norm": 0.1777779906988144,
+      "learning_rate": 0.0001,
+      "loss": 1.7816,
+      "step": 21
+    },
+    {
+      "epoch": 0.01073519843851659,
+      "grad_norm": 0.17349651455879211,
+      "learning_rate": 0.0001,
+      "loss": 1.7431,
+      "step": 22
+    },
+    {
+      "epoch": 0.011223162003903709,
+      "grad_norm": 0.18479375541210175,
+      "learning_rate": 0.0001,
+      "loss": 1.903,
+      "step": 23
+    },
+    {
+      "epoch": 0.011711125569290826,
+      "grad_norm": 0.1918632984161377,
+      "learning_rate": 0.0001,
+      "loss": 1.7957,
+      "step": 24
+    },
+    {
+      "epoch": 0.012199089134677944,
+      "grad_norm": 0.18239013850688934,
+      "learning_rate": 0.0001,
+      "loss": 1.8039,
+      "step": 25
+    },
+    {
+      "epoch": 0.012687052700065062,
+      "grad_norm": 0.17392802238464355,
+      "learning_rate": 0.0001,
+      "loss": 1.7022,
+      "step": 26
+    },
+    {
+      "epoch": 0.013175016265452179,
+      "grad_norm": 0.1769259124994278,
+      "learning_rate": 0.0001,
+      "loss": 1.7131,
+      "step": 27
+    },
+    {
+      "epoch": 0.013662979830839297,
+      "grad_norm": 0.17371872067451477,
+      "learning_rate": 0.0001,
+      "loss": 1.7657,
+      "step": 28
+    },
+    {
+      "epoch": 0.014150943396226415,
+      "grad_norm": 0.19897091388702393,
+      "learning_rate": 0.0001,
+      "loss": 1.8791,
+      "step": 29
+    },
+    {
+      "epoch": 0.014638906961613532,
+      "grad_norm": 0.17471033334732056,
+      "learning_rate": 0.0001,
+      "loss": 1.8765,
+      "step": 30
+    },
+    {
+      "epoch": 0.01512687052700065,
+      "grad_norm": 0.17650161683559418,
+      "learning_rate": 0.0001,
+      "loss": 1.8181,
+      "step": 31
+    },
+    {
+      "epoch": 0.015614834092387769,
+      "grad_norm": 0.18008925020694733,
+      "learning_rate": 0.0001,
+      "loss": 1.8138,
+      "step": 32
+    },
+    {
+      "epoch": 0.016102797657774885,
+      "grad_norm": 0.18406356871128082,
+      "learning_rate": 0.0001,
+      "loss": 1.907,
+      "step": 33
+    },
+    {
+      "epoch": 0.016590761223162005,
+      "grad_norm": 0.18869489431381226,
+      "learning_rate": 0.0001,
+      "loss": 1.9043,
+      "step": 34
+    },
+    {
+      "epoch": 0.017078724788549122,
+      "grad_norm": 0.18416965007781982,
+      "learning_rate": 0.0001,
+      "loss": 1.7695,
+      "step": 35
+    },
+    {
+      "epoch": 0.01756668835393624,
+      "grad_norm": 0.18121257424354553,
+      "learning_rate": 0.0001,
+      "loss": 1.8342,
+      "step": 36
+    },
+    {
+      "epoch": 0.01805465191932336,
+      "grad_norm": 0.18426860868930817,
+      "learning_rate": 0.0001,
+      "loss": 1.818,
+      "step": 37
+    },
+    {
+      "epoch": 0.018542615484710475,
+      "grad_norm": 0.18800823390483856,
+      "learning_rate": 0.0001,
+      "loss": 1.8019,
+      "step": 38
+    },
+    {
+      "epoch": 0.01903057905009759,
+      "grad_norm": 0.18787121772766113,
+      "learning_rate": 0.0001,
+      "loss": 1.8052,
+      "step": 39
+    },
+    {
+      "epoch": 0.01951854261548471,
+      "grad_norm": 0.18341200053691864,
+      "learning_rate": 0.0001,
+      "loss": 1.7288,
+      "step": 40
+    },
+    {
+      "epoch": 0.020006506180871828,
+      "grad_norm": 0.18460282683372498,
+      "learning_rate": 0.0001,
+      "loss": 1.9984,
+      "step": 41
+    },
+    {
+      "epoch": 0.020494469746258945,
+      "grad_norm": 0.17212441563606262,
+      "learning_rate": 0.0001,
+      "loss": 1.7928,
+      "step": 42
+    },
+    {
+      "epoch": 0.020982433311646065,
+      "grad_norm": 0.18548350036144257,
+      "learning_rate": 0.0001,
+      "loss": 1.9719,
+      "step": 43
+    },
+    {
+      "epoch": 0.02147039687703318,
+      "grad_norm": 0.18035617470741272,
+      "learning_rate": 0.0001,
+      "loss": 1.9265,
+      "step": 44
+    },
+    {
+      "epoch": 0.021958360442420298,
+      "grad_norm": 0.16300201416015625,
+      "learning_rate": 0.0001,
+      "loss": 1.6821,
+      "step": 45
+    },
+    {
+      "epoch": 0.022446324007807418,
+      "grad_norm": 0.1797887086868286,
+      "learning_rate": 0.0001,
+      "loss": 1.8276,
+      "step": 46
+    },
+    {
+      "epoch": 0.022934287573194535,
+      "grad_norm": 0.18614032864570618,
+      "learning_rate": 0.0001,
+      "loss": 1.769,
+      "step": 47
+    },
+    {
+      "epoch": 0.02342225113858165,
+      "grad_norm": 0.18762686848640442,
+      "learning_rate": 0.0001,
+      "loss": 1.7716,
+      "step": 48
+    },
+    {
+      "epoch": 0.02391021470396877,
+      "grad_norm": 0.1779824048280716,
+      "learning_rate": 0.0001,
+      "loss": 1.7047,
+      "step": 49
+    },
+    {
+      "epoch": 0.024398178269355888,
+      "grad_norm": 0.1713806688785553,
+      "learning_rate": 0.0001,
+      "loss": 1.7085,
+      "step": 50
+    },
+    {
+      "epoch": 0.024886141834743004,
+      "grad_norm": 0.17888174951076508,
+      "learning_rate": 0.0001,
+      "loss": 1.8539,
+      "step": 51
+    },
+    {
+      "epoch": 0.025374105400130124,
+      "grad_norm": 0.18366138637065887,
+      "learning_rate": 0.0001,
+      "loss": 1.7948,
+      "step": 52
+    },
+    {
+      "epoch": 0.02586206896551724,
+      "grad_norm": 0.1684766262769699,
+      "learning_rate": 0.0001,
+      "loss": 1.7752,
+      "step": 53
+    },
+    {
+      "epoch": 0.026350032530904358,
+      "grad_norm": 0.18316026031970978,
+      "learning_rate": 0.0001,
+      "loss": 1.8153,
+      "step": 54
+    },
+    {
+      "epoch": 0.026837996096291478,
+      "grad_norm": 0.1712900847196579,
+      "learning_rate": 0.0001,
+      "loss": 1.8209,
+      "step": 55
+    },
+    {
+      "epoch": 0.027325959661678594,
+      "grad_norm": 0.17653001844882965,
+      "learning_rate": 0.0001,
+      "loss": 1.7142,
+      "step": 56
+    },
+    {
+      "epoch": 0.02781392322706571,
+      "grad_norm": 0.17115001380443573,
+      "learning_rate": 0.0001,
+      "loss": 1.7014,
+      "step": 57
+    },
+    {
+      "epoch": 0.02830188679245283,
+      "grad_norm": 0.19934123754501343,
+      "learning_rate": 0.0001,
+      "loss": 1.8184,
+      "step": 58
+    },
+    {
+      "epoch": 0.028789850357839947,
+      "grad_norm": 0.20567697286605835,
+      "learning_rate": 0.0001,
+      "loss": 1.9174,
+      "step": 59
+    },
+    {
+      "epoch": 0.029277813923227064,
+      "grad_norm": 0.17345917224884033,
+      "learning_rate": 0.0001,
+      "loss": 1.7448,
+      "step": 60
+    },
+    {
+      "epoch": 0.029765777488614184,
+      "grad_norm": 0.24353067576885223,
+      "learning_rate": 0.0001,
+      "loss": 1.7974,
+      "step": 61
+    },
+    {
+      "epoch": 0.0302537410540013,
+      "grad_norm": 0.18949398398399353,
+      "learning_rate": 0.0001,
+      "loss": 1.8231,
+      "step": 62
+    },
+    {
+      "epoch": 0.03074170461938842,
+      "grad_norm": 0.22029712796211243,
+      "learning_rate": 0.0001,
+      "loss": 1.8535,
+      "step": 63
+    },
+    {
+      "epoch": 0.031229668184775537,
+      "grad_norm": 0.16962048411369324,
+      "learning_rate": 0.0001,
+      "loss": 1.7686,
+      "step": 64
+    },
+    {
+      "epoch": 0.03171763175016266,
+      "grad_norm": 0.19039765000343323,
+      "learning_rate": 0.0001,
+      "loss": 1.8303,
+      "step": 65
+    },
+    {
+      "epoch": 0.03220559531554977,
+      "grad_norm": 0.20166978240013123,
+      "learning_rate": 0.0001,
+      "loss": 1.768,
+      "step": 66
+    },
+    {
+      "epoch": 0.03269355888093689,
+      "grad_norm": 0.173394113779068,
+      "learning_rate": 0.0001,
+      "loss": 1.8253,
+      "step": 67
+    },
+    {
+      "epoch": 0.03318152244632401,
+      "grad_norm": 0.19260728359222412,
+      "learning_rate": 0.0001,
+      "loss": 1.7589,
+      "step": 68
+    },
+    {
+      "epoch": 0.033669486011711124,
+      "grad_norm": 0.19539032876491547,
+      "learning_rate": 0.0001,
+      "loss": 1.749,
+      "step": 69
+    },
+    {
+      "epoch": 0.034157449577098244,
+      "grad_norm": 0.16770870983600616,
+      "learning_rate": 0.0001,
+      "loss": 1.7132,
+      "step": 70
+    },
+    {
+      "epoch": 0.034645413142485364,
+      "grad_norm": 0.19755178689956665,
+      "learning_rate": 0.0001,
+      "loss": 1.8323,
+      "step": 71
+    },
+    {
+      "epoch": 0.03513337670787248,
+      "grad_norm": 0.18038292229175568,
+      "learning_rate": 0.0001,
+      "loss": 1.7599,
+      "step": 72
+    },
+    {
+      "epoch": 0.0356213402732596,
+      "grad_norm": 0.17995433509349823,
+      "learning_rate": 0.0001,
+      "loss": 1.9183,
+      "step": 73
+    },
+    {
+      "epoch": 0.03610930383864672,
+      "grad_norm": 0.19222807884216309,
+      "learning_rate": 0.0001,
+      "loss": 1.8642,
+      "step": 74
+    },
+    {
+      "epoch": 0.03659726740403383,
+      "grad_norm": 0.16965682804584503,
+      "learning_rate": 0.0001,
+      "loss": 1.7271,
+      "step": 75
+    },
+    {
+      "epoch": 0.03708523096942095,
+      "grad_norm": 0.17662999033927917,
+      "learning_rate": 0.0001,
+      "loss": 1.8263,
+      "step": 76
+    },
+    {
+      "epoch": 0.03757319453480807,
+      "grad_norm": 0.1699201613664627,
+      "learning_rate": 0.0001,
+      "loss": 1.6818,
+      "step": 77
+    },
+    {
+      "epoch": 0.03806115810019518,
+      "grad_norm": 0.17309829592704773,
+      "learning_rate": 0.0001,
+      "loss": 1.7424,
+      "step": 78
+    },
+    {
+      "epoch": 0.0385491216655823,
+      "grad_norm": 0.18537020683288574,
+      "learning_rate": 0.0001,
+      "loss": 1.7986,
+      "step": 79
+    },
+    {
+      "epoch": 0.03903708523096942,
+      "grad_norm": 0.1709861010313034,
+      "learning_rate": 0.0001,
+      "loss": 1.6091,
+      "step": 80
+    },
+    {
+      "epoch": 0.039525048796356536,
+      "grad_norm": 0.17050296068191528,
+      "learning_rate": 0.0001,
+      "loss": 1.6904,
+      "step": 81
+    },
+    {
+      "epoch": 0.040013012361743656,
+      "grad_norm": 0.17640157043933868,
+      "learning_rate": 0.0001,
+      "loss": 1.7087,
+      "step": 82
+    },
+    {
+      "epoch": 0.040500975927130776,
+      "grad_norm": 0.1919400542974472,
+      "learning_rate": 0.0001,
+      "loss": 1.8223,
+      "step": 83
+    },
+    {
+      "epoch": 0.04098893949251789,
+      "grad_norm": 0.19427765905857086,
+      "learning_rate": 0.0001,
+      "loss": 1.7443,
+      "step": 84
+    },
+    {
+      "epoch": 0.04147690305790501,
+      "grad_norm": 0.19496281445026398,
+      "learning_rate": 0.0001,
+      "loss": 1.8336,
+      "step": 85
+    },
+    {
+      "epoch": 0.04196486662329213,
+      "grad_norm": 0.18101565539836884,
+      "learning_rate": 0.0001,
+      "loss": 1.8422,
+      "step": 86
+    },
+    {
+      "epoch": 0.04245283018867924,
+      "grad_norm": 0.19941496849060059,
+      "learning_rate": 0.0001,
+      "loss": 1.7168,
+      "step": 87
+    },
+    {
+      "epoch": 0.04294079375406636,
+      "grad_norm": 0.1963973492383957,
+      "learning_rate": 0.0001,
+      "loss": 1.7558,
+      "step": 88
+    },
+    {
+      "epoch": 0.04342875731945348,
+      "grad_norm": 0.17694450914859772,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 89
+    },
+    {
+      "epoch": 0.043916720884840596,
+      "grad_norm": 0.19362711906433105,
+      "learning_rate": 0.0001,
+      "loss": 1.8165,
+      "step": 90
+    },
+    {
+      "epoch": 0.044404684450227716,
+      "grad_norm": 0.1736024022102356,
+      "learning_rate": 0.0001,
+      "loss": 1.777,
+      "step": 91
+    },
+    {
+      "epoch": 0.044892648015614836,
+      "grad_norm": 0.17649488151073456,
+      "learning_rate": 0.0001,
+      "loss": 1.7507,
+      "step": 92
+    },
+    {
+      "epoch": 0.04538061158100195,
+      "grad_norm": 0.2002265304327011,
+      "learning_rate": 0.0001,
+      "loss": 1.8796,
+      "step": 93
+    },
+    {
+      "epoch": 0.04586857514638907,
+      "grad_norm": 0.1667991429567337,
+      "learning_rate": 0.0001,
+      "loss": 1.7051,
+      "step": 94
+    },
+    {
+      "epoch": 0.04635653871177619,
+      "grad_norm": 0.1868171989917755,
+      "learning_rate": 0.0001,
+      "loss": 1.747,
+      "step": 95
+    },
+    {
+      "epoch": 0.0468445022771633,
+      "grad_norm": 0.18312174081802368,
+      "learning_rate": 0.0001,
+      "loss": 1.7835,
+      "step": 96
+    },
+    {
+      "epoch": 0.04733246584255042,
+      "grad_norm": 0.1762659102678299,
+      "learning_rate": 0.0001,
+      "loss": 1.6517,
+      "step": 97
+    },
+    {
+      "epoch": 0.04782042940793754,
+      "grad_norm": 0.19766494631767273,
+      "learning_rate": 0.0001,
+      "loss": 1.826,
+      "step": 98
+    },
+    {
+      "epoch": 0.048308392973324656,
+      "grad_norm": 0.17331789433956146,
+      "learning_rate": 0.0001,
+      "loss": 1.7506,
+      "step": 99
+    },
+    {
+      "epoch": 0.048796356538711776,
+      "grad_norm": 0.16851170361042023,
+      "learning_rate": 0.0001,
+      "loss": 1.744,
+      "step": 100
+    },
+    {
+      "epoch": 0.049284320104098896,
+      "grad_norm": 0.17572622001171112,
+      "learning_rate": 0.0001,
+      "loss": 1.6986,
+      "step": 101
+    },
+    {
+      "epoch": 0.04977228366948601,
+      "grad_norm": 0.1850849688053131,
+      "learning_rate": 0.0001,
+      "loss": 1.7895,
+      "step": 102
+    },
+    {
+      "epoch": 0.05026024723487313,
+      "grad_norm": 0.18450362980365753,
+      "learning_rate": 0.0001,
+      "loss": 1.8234,
+      "step": 103
+    },
+    {
+      "epoch": 0.05074821080026025,
+      "grad_norm": 0.1832476705312729,
+      "learning_rate": 0.0001,
+      "loss": 1.7986,
+      "step": 104
+    },
+    {
+      "epoch": 0.05123617436564736,
+      "grad_norm": 0.1809314638376236,
+      "learning_rate": 0.0001,
+      "loss": 1.7923,
+      "step": 105
+    },
+    {
+      "epoch": 0.05172413793103448,
+      "grad_norm": 0.17974039912223816,
+      "learning_rate": 0.0001,
+      "loss": 1.7095,
+      "step": 106
+    },
+    {
+      "epoch": 0.0522121014964216,
+      "grad_norm": 0.16436076164245605,
+      "learning_rate": 0.0001,
+      "loss": 1.6873,
+      "step": 107
+    },
+    {
+      "epoch": 0.052700065061808715,
+      "grad_norm": 0.16344858705997467,
+      "learning_rate": 0.0001,
+      "loss": 1.6991,
+      "step": 108
+    },
+    {
+      "epoch": 0.053188028627195835,
+      "grad_norm": 0.17950277030467987,
+      "learning_rate": 0.0001,
+      "loss": 1.8591,
+      "step": 109
+    },
+    {
+      "epoch": 0.053675992192582955,
+      "grad_norm": 0.18337760865688324,
+      "learning_rate": 0.0001,
+      "loss": 1.784,
+      "step": 110
+    },
+    {
+      "epoch": 0.05416395575797007,
+      "grad_norm": 0.1895488053560257,
+      "learning_rate": 0.0001,
+      "loss": 1.7853,
+      "step": 111
+    },
+    {
+      "epoch": 0.05465191932335719,
+      "grad_norm": 0.17522425949573517,
+      "learning_rate": 0.0001,
+      "loss": 1.7127,
+      "step": 112
+    },
+    {
+      "epoch": 0.05513988288874431,
+      "grad_norm": 0.17943814396858215,
+      "learning_rate": 0.0001,
+      "loss": 1.755,
+      "step": 113
+    },
+    {
+      "epoch": 0.05562784645413142,
+      "grad_norm": 0.1815492808818817,
+      "learning_rate": 0.0001,
+      "loss": 1.7687,
+      "step": 114
+    },
+    {
+      "epoch": 0.05611581001951854,
+      "grad_norm": 0.16954658925533295,
+      "learning_rate": 0.0001,
+      "loss": 1.7562,
+      "step": 115
+    },
+    {
+      "epoch": 0.05660377358490566,
+      "grad_norm": 0.17870648205280304,
+      "learning_rate": 0.0001,
+      "loss": 1.841,
+      "step": 116
+    },
+    {
+      "epoch": 0.057091737150292775,
+      "grad_norm": 0.17044954001903534,
+      "learning_rate": 0.0001,
+      "loss": 1.7118,
+      "step": 117
+    },
+    {
+      "epoch": 0.057579700715679895,
+      "grad_norm": 0.17524173855781555,
+      "learning_rate": 0.0001,
+      "loss": 1.6045,
+      "step": 118
+    },
+    {
+      "epoch": 0.058067664281067015,
+      "grad_norm": 0.17537613213062286,
+      "learning_rate": 0.0001,
+      "loss": 1.8018,
+      "step": 119
+    },
+    {
+      "epoch": 0.05855562784645413,
+      "grad_norm": 0.17819495499134064,
+      "learning_rate": 0.0001,
+      "loss": 1.7723,
+      "step": 120
+    },
+    {
+      "epoch": 0.05904359141184125,
+      "grad_norm": 0.17807795107364655,
+      "learning_rate": 0.0001,
+      "loss": 1.8558,
+      "step": 121
+    },
+    {
+      "epoch": 0.05953155497722837,
+      "grad_norm": 0.1687198132276535,
+      "learning_rate": 0.0001,
+      "loss": 1.7673,
+      "step": 122
+    },
+    {
+      "epoch": 0.06001951854261549,
+      "grad_norm": 0.17069241404533386,
+      "learning_rate": 0.0001,
+      "loss": 1.7561,
+      "step": 123
+    },
+    {
+      "epoch": 0.0605074821080026,
+      "grad_norm": 0.1655956506729126,
+      "learning_rate": 0.0001,
+      "loss": 1.6607,
+      "step": 124
+    },
+    {
+      "epoch": 0.06099544567338972,
+      "grad_norm": 0.1846679002046585,
+      "learning_rate": 0.0001,
+      "loss": 1.8676,
+      "step": 125
+    },
+    {
+      "epoch": 0.06148340923877684,
+      "grad_norm": 0.17344145476818085,
+      "learning_rate": 0.0001,
+      "loss": 1.7427,
+      "step": 126
+    },
+    {
+      "epoch": 0.061971372804163954,
+      "grad_norm": 0.17264996469020844,
+      "learning_rate": 0.0001,
+      "loss": 1.7279,
+      "step": 127
+    },
+    {
+      "epoch": 0.062459336369551074,
+      "grad_norm": 0.18628281354904175,
+      "learning_rate": 0.0001,
+      "loss": 1.6708,
+      "step": 128
+    },
+    {
+      "epoch": 0.0629472999349382,
+      "grad_norm": 0.178174689412117,
+      "learning_rate": 0.0001,
+      "loss": 1.7931,
+      "step": 129
+    },
+    {
+      "epoch": 0.06343526350032531,
+      "grad_norm": 0.17690585553646088,
+      "learning_rate": 0.0001,
+      "loss": 1.7647,
+      "step": 130
+    },
+    {
+      "epoch": 0.06392322706571242,
+      "grad_norm": 0.18117444217205048,
+      "learning_rate": 0.0001,
+      "loss": 1.7376,
+      "step": 131
+    },
+    {
+      "epoch": 0.06441119063109954,
+      "grad_norm": 0.17523089051246643,
+      "learning_rate": 0.0001,
+      "loss": 1.8403,
+      "step": 132
+    },
+    {
+      "epoch": 0.06489915419648666,
+      "grad_norm": 0.16988244652748108,
+      "learning_rate": 0.0001,
+      "loss": 1.6958,
+      "step": 133
+    },
+    {
+      "epoch": 0.06538711776187378,
+      "grad_norm": 0.1890041083097458,
+      "learning_rate": 0.0001,
+      "loss": 1.7388,
+      "step": 134
+    },
+    {
+      "epoch": 0.0658750813272609,
+      "grad_norm": 0.1703094244003296,
+      "learning_rate": 0.0001,
+      "loss": 1.6424,
+      "step": 135
+    },
+    {
+      "epoch": 0.06636304489264802,
+      "grad_norm": 0.17852698266506195,
+      "learning_rate": 0.0001,
+      "loss": 1.7786,
+      "step": 136
+    },
+    {
+      "epoch": 0.06685100845803513,
+      "grad_norm": 0.17648550868034363,
+      "learning_rate": 0.0001,
+      "loss": 1.7172,
+      "step": 137
+    },
+    {
+      "epoch": 0.06733897202342225,
+      "grad_norm": 0.18284566700458527,
+      "learning_rate": 0.0001,
+      "loss": 1.7491,
+      "step": 138
+    },
+    {
+      "epoch": 0.06782693558880937,
+      "grad_norm": 0.1686737835407257,
+      "learning_rate": 0.0001,
+      "loss": 1.7218,
+      "step": 139
+    },
+    {
+      "epoch": 0.06831489915419649,
+      "grad_norm": 0.1741771250963211,
+      "learning_rate": 0.0001,
+      "loss": 1.7534,
+      "step": 140
+    },
+    {
+      "epoch": 0.06880286271958361,
+      "grad_norm": 0.1778876781463623,
+      "learning_rate": 0.0001,
+      "loss": 1.7388,
+      "step": 141
+    },
+    {
+      "epoch": 0.06929082628497073,
+      "grad_norm": 0.1860485076904297,
+      "learning_rate": 0.0001,
+      "loss": 1.8109,
+      "step": 142
+    },
+    {
+      "epoch": 0.06977878985035783,
+      "grad_norm": 0.17966079711914062,
+      "learning_rate": 0.0001,
+      "loss": 1.7171,
+      "step": 143
+    },
+    {
+      "epoch": 0.07026675341574495,
+      "grad_norm": 0.19341900944709778,
+      "learning_rate": 0.0001,
+      "loss": 1.7911,
+      "step": 144
+    },
+    {
+      "epoch": 0.07075471698113207,
+      "grad_norm": 0.1968701183795929,
+      "learning_rate": 0.0001,
+      "loss": 1.858,
+      "step": 145
+    },
+    {
+      "epoch": 0.0712426805465192,
+      "grad_norm": 0.17585061490535736,
+      "learning_rate": 0.0001,
+      "loss": 1.6731,
+      "step": 146
+    },
+    {
+      "epoch": 0.07173064411190631,
+      "grad_norm": 0.17294664680957794,
+      "learning_rate": 0.0001,
+      "loss": 1.7284,
+      "step": 147
+    },
+    {
+      "epoch": 0.07221860767729343,
+      "grad_norm": 0.18245872855186462,
+      "learning_rate": 0.0001,
+      "loss": 1.7595,
+      "step": 148
+    },
+    {
+      "epoch": 0.07270657124268054,
+      "grad_norm": 0.16850219666957855,
+      "learning_rate": 0.0001,
+      "loss": 1.73,
+      "step": 149
+    },
+    {
+      "epoch": 0.07319453480806766,
+      "grad_norm": 0.16891759634017944,
+      "learning_rate": 0.0001,
+      "loss": 1.7434,
+      "step": 150
+    },
+    {
+      "epoch": 0.07368249837345478,
+      "grad_norm": 0.17363204061985016,
+      "learning_rate": 0.0001,
+      "loss": 1.738,
+      "step": 151
+    },
+    {
+      "epoch": 0.0741704619388419,
+      "grad_norm": 0.16307075321674347,
+      "learning_rate": 0.0001,
+      "loss": 1.6285,
+      "step": 152
+    },
+    {
+      "epoch": 0.07465842550422902,
+      "grad_norm": 0.1735111027956009,
+      "learning_rate": 0.0001,
+      "loss": 1.5711,
+      "step": 153
+    },
+    {
+      "epoch": 0.07514638906961614,
+      "grad_norm": 0.18169796466827393,
+      "learning_rate": 0.0001,
+      "loss": 1.7395,
+      "step": 154
+    },
+    {
+      "epoch": 0.07563435263500325,
+      "grad_norm": 0.16926725208759308,
+      "learning_rate": 0.0001,
+      "loss": 1.7534,
+      "step": 155
+    },
+    {
+      "epoch": 0.07612231620039037,
+      "grad_norm": 0.19919319450855255,
+      "learning_rate": 0.0001,
+      "loss": 1.6975,
+      "step": 156
+    },
+    {
+      "epoch": 0.07661027976577749,
+      "grad_norm": 0.19146177172660828,
+      "learning_rate": 0.0001,
+      "loss": 1.8272,
+      "step": 157
+    },
+    {
+      "epoch": 0.0770982433311646,
+      "grad_norm": 0.19453231990337372,
+      "learning_rate": 0.0001,
+      "loss": 1.8229,
+      "step": 158
+    },
+    {
+      "epoch": 0.07758620689655173,
+      "grad_norm": 0.20597495138645172,
+      "learning_rate": 0.0001,
+      "loss": 1.8567,
+      "step": 159
+    },
+    {
+      "epoch": 0.07807417046193885,
+      "grad_norm": 0.18599432706832886,
+      "learning_rate": 0.0001,
+      "loss": 1.7587,
+      "step": 160
+    },
+    {
+      "epoch": 0.07856213402732595,
+      "grad_norm": 0.21232162415981293,
+      "learning_rate": 0.0001,
+      "loss": 1.7179,
+      "step": 161
+    },
+    {
+      "epoch": 0.07905009759271307,
+      "grad_norm": 0.1712743043899536,
+      "learning_rate": 0.0001,
+      "loss": 1.678,
+      "step": 162
+    },
+    {
+      "epoch": 0.07953806115810019,
+      "grad_norm": 0.18402481079101562,
+      "learning_rate": 0.0001,
+      "loss": 1.7731,
+      "step": 163
+    },
+    {
+      "epoch": 0.08002602472348731,
+      "grad_norm": 0.18908202648162842,
+      "learning_rate": 0.0001,
+      "loss": 1.841,
+      "step": 164
+    },
+    {
+      "epoch": 0.08051398828887443,
+      "grad_norm": 0.17370882630348206,
+      "learning_rate": 0.0001,
+      "loss": 1.6713,
+      "step": 165
+    },
+    {
+      "epoch": 0.08100195185426155,
+      "grad_norm": 0.1881919503211975,
+      "learning_rate": 0.0001,
+      "loss": 1.8285,
+      "step": 166
+    },
+    {
+      "epoch": 0.08148991541964867,
+      "grad_norm": 0.1770172417163849,
+      "learning_rate": 0.0001,
+      "loss": 1.7292,
+      "step": 167
+    },
+    {
+      "epoch": 0.08197787898503578,
+      "grad_norm": 0.1822032779455185,
+      "learning_rate": 0.0001,
+      "loss": 1.6977,
+      "step": 168
+    },
+    {
+      "epoch": 0.0824658425504229,
+      "grad_norm": 0.19020989537239075,
+      "learning_rate": 0.0001,
+      "loss": 1.6964,
+      "step": 169
+    },
+    {
+      "epoch": 0.08295380611581002,
+      "grad_norm": 0.17227591574192047,
+      "learning_rate": 0.0001,
+      "loss": 1.703,
+      "step": 170
+    },
+    {
+      "epoch": 0.08344176968119714,
+      "grad_norm": 0.19228717684745789,
+      "learning_rate": 0.0001,
+      "loss": 1.7247,
+      "step": 171
+    },
+    {
+      "epoch": 0.08392973324658426,
+      "grad_norm": 0.1909552961587906,
+      "learning_rate": 0.0001,
+      "loss": 1.7973,
+      "step": 172
+    },
+    {
+      "epoch": 0.08441769681197138,
+      "grad_norm": 0.18189294636249542,
+      "learning_rate": 0.0001,
+      "loss": 1.7579,
+      "step": 173
+    },
+    {
+      "epoch": 0.08490566037735849,
+      "grad_norm": 0.19137217104434967,
+      "learning_rate": 0.0001,
+      "loss": 1.7198,
+      "step": 174
+    },
+    {
+      "epoch": 0.0853936239427456,
+      "grad_norm": 0.18612581491470337,
+      "learning_rate": 0.0001,
+      "loss": 1.7585,
+      "step": 175
+    },
+    {
+      "epoch": 0.08588158750813273,
+      "grad_norm": 0.1759909838438034,
+      "learning_rate": 0.0001,
+      "loss": 1.6732,
+      "step": 176
+    },
+    {
+      "epoch": 0.08636955107351985,
+      "grad_norm": 0.18982531130313873,
+      "learning_rate": 0.0001,
+      "loss": 1.8301,
+      "step": 177
+    },
+    {
+      "epoch": 0.08685751463890697,
+      "grad_norm": 0.16662733256816864,
+      "learning_rate": 0.0001,
+      "loss": 1.6799,
+      "step": 178
+    },
+    {
+      "epoch": 0.08734547820429409,
+      "grad_norm": 0.17956425249576569,
+      "learning_rate": 0.0001,
+      "loss": 1.671,
+      "step": 179
+    },
+    {
+      "epoch": 0.08783344176968119,
+      "grad_norm": 0.18416181206703186,
+      "learning_rate": 0.0001,
+      "loss": 1.7922,
+      "step": 180
+    },
+    {
+      "epoch": 0.08832140533506831,
+      "grad_norm": 0.16633754968643188,
+      "learning_rate": 0.0001,
+      "loss": 1.7096,
+      "step": 181
+    },
+    {
+      "epoch": 0.08880936890045543,
+      "grad_norm": 0.19759412109851837,
+      "learning_rate": 0.0001,
+      "loss": 1.8402,
+      "step": 182
+    },
+    {
+      "epoch": 0.08929733246584255,
+      "grad_norm": 0.17006362974643707,
+      "learning_rate": 0.0001,
+      "loss": 1.6922,
+      "step": 183
+    },
+    {
+      "epoch": 0.08978529603122967,
+      "grad_norm": 0.16919896006584167,
+      "learning_rate": 0.0001,
+      "loss": 1.6657,
+      "step": 184
+    },
+    {
+      "epoch": 0.09027325959661679,
+      "grad_norm": 0.20307502150535583,
+      "learning_rate": 0.0001,
+      "loss": 1.8772,
+      "step": 185
+    },
+    {
+      "epoch": 0.0907612231620039,
+      "grad_norm": 0.17572732269763947,
+      "learning_rate": 0.0001,
+      "loss": 1.7666,
+      "step": 186
+    },
+    {
+      "epoch": 0.09124918672739102,
+      "grad_norm": 0.17327293753623962,
+      "learning_rate": 0.0001,
+      "loss": 1.8206,
+      "step": 187
+    },
+    {
+      "epoch": 0.09173715029277814,
+      "grad_norm": 0.18354281783103943,
+      "learning_rate": 0.0001,
+      "loss": 1.8013,
+      "step": 188
+    },
+    {
+      "epoch": 0.09222511385816526,
+      "grad_norm": 0.16821032762527466,
+      "learning_rate": 0.0001,
+      "loss": 1.6893,
+      "step": 189
+    },
+    {
+      "epoch": 0.09271307742355238,
+      "grad_norm": 0.17506404221057892,
+      "learning_rate": 0.0001,
+      "loss": 1.7657,
+      "step": 190
+    },
+    {
+      "epoch": 0.0932010409889395,
+      "grad_norm": 0.1758153885602951,
+      "learning_rate": 0.0001,
+      "loss": 1.7095,
+      "step": 191
+    },
+    {
+      "epoch": 0.0936890045543266,
+      "grad_norm": 0.18787072598934174,
+      "learning_rate": 0.0001,
+      "loss": 1.7312,
+      "step": 192
+    },
+    {
+      "epoch": 0.09417696811971372,
+      "grad_norm": 0.1803017109632492,
+      "learning_rate": 0.0001,
+      "loss": 1.7521,
+      "step": 193
+    },
+    {
+      "epoch": 0.09466493168510084,
+      "grad_norm": 0.18097610771656036,
+      "learning_rate": 0.0001,
+      "loss": 1.6861,
+      "step": 194
+    },
+    {
+      "epoch": 0.09515289525048796,
+      "grad_norm": 0.1760302186012268,
+      "learning_rate": 0.0001,
+      "loss": 1.6703,
+      "step": 195
+    },
+    {
+      "epoch": 0.09564085881587508,
+      "grad_norm": 0.17225316166877747,
+      "learning_rate": 0.0001,
+      "loss": 1.73,
+      "step": 196
+    },
+    {
+      "epoch": 0.0961288223812622,
+      "grad_norm": 0.1856345683336258,
+      "learning_rate": 0.0001,
+      "loss": 1.6828,
+      "step": 197
+    },
+    {
+      "epoch": 0.09661678594664931,
+      "grad_norm": 0.18595090508460999,
+      "learning_rate": 0.0001,
+      "loss": 1.7136,
+      "step": 198
+    },
+    {
+      "epoch": 0.09710474951203643,
+      "grad_norm": 0.1780211329460144,
+      "learning_rate": 0.0001,
+      "loss": 1.8146,
+      "step": 199
+    },
+    {
+      "epoch": 0.09759271307742355,
+      "grad_norm": 0.17781271040439606,
+      "learning_rate": 0.0001,
+      "loss": 1.6679,
+      "step": 200
+    },
+    {
+      "epoch": 0.09808067664281067,
+      "grad_norm": 0.17124401032924652,
+      "learning_rate": 0.0001,
+      "loss": 1.7077,
+      "step": 201
+    },
+    {
+      "epoch": 0.09856864020819779,
+      "grad_norm": 0.18443076312541962,
+      "learning_rate": 0.0001,
+      "loss": 1.8058,
+      "step": 202
+    },
+    {
+      "epoch": 0.09905660377358491,
+      "grad_norm": 0.1758834272623062,
+      "learning_rate": 0.0001,
+      "loss": 1.81,
+      "step": 203
+    },
+    {
+      "epoch": 0.09954456733897202,
+      "grad_norm": 0.17878177762031555,
+      "learning_rate": 0.0001,
+      "loss": 1.7515,
+      "step": 204
+    },
+    {
+      "epoch": 0.10003253090435914,
+      "grad_norm": 0.18028298020362854,
+      "learning_rate": 0.0001,
+      "loss": 1.7733,
+      "step": 205
+    },
+    {
+      "epoch": 0.10052049446974626,
+      "grad_norm": 0.17935384809970856,
+      "learning_rate": 0.0001,
+      "loss": 1.8011,
+      "step": 206
+    },
+    {
+      "epoch": 0.10100845803513338,
+      "grad_norm": 0.19665150344371796,
+      "learning_rate": 0.0001,
+      "loss": 1.7667,
+      "step": 207
+    },
+    {
+      "epoch": 0.1014964216005205,
+      "grad_norm": 0.16669659316539764,
+      "learning_rate": 0.0001,
+      "loss": 1.7046,
+      "step": 208
+    },
+    {
+      "epoch": 0.10198438516590762,
+      "grad_norm": 0.17783086001873016,
+      "learning_rate": 0.0001,
+      "loss": 1.6424,
+      "step": 209
+    },
+    {
+      "epoch": 0.10247234873129472,
+      "grad_norm": 0.1761302351951599,
+      "learning_rate": 0.0001,
+      "loss": 1.726,
+      "step": 210
+    },
+    {
+      "epoch": 0.10296031229668184,
+      "grad_norm": 0.17417997121810913,
+      "learning_rate": 0.0001,
+      "loss": 1.7181,
+      "step": 211
+    },
+    {
+      "epoch": 0.10344827586206896,
+      "grad_norm": 0.17537769675254822,
+      "learning_rate": 0.0001,
+      "loss": 1.6876,
+      "step": 212
+    },
+    {
+      "epoch": 0.10393623942745608,
+      "grad_norm": 0.16924896836280823,
+      "learning_rate": 0.0001,
+      "loss": 1.768,
+      "step": 213
+    },
+    {
+      "epoch": 0.1044242029928432,
+      "grad_norm": 0.20247921347618103,
+      "learning_rate": 0.0001,
+      "loss": 1.9159,
+      "step": 214
+    },
+    {
+      "epoch": 0.10491216655823032,
+      "grad_norm": 0.16506172716617584,
+      "learning_rate": 0.0001,
+      "loss": 1.6491,
+      "step": 215
+    },
+    {
+      "epoch": 0.10540013012361743,
+      "grad_norm": 0.17558075487613678,
+      "learning_rate": 0.0001,
+      "loss": 1.7169,
+      "step": 216
+    },
+    {
+      "epoch": 0.10588809368900455,
+      "grad_norm": 0.17124514281749725,
+      "learning_rate": 0.0001,
+      "loss": 1.6931,
+      "step": 217
+    },
+    {
+      "epoch": 0.10637605725439167,
+      "grad_norm": 0.16885621845722198,
+      "learning_rate": 0.0001,
+      "loss": 1.6946,
+      "step": 218
+    },
+    {
+      "epoch": 0.10686402081977879,
+      "grad_norm": 0.17787247896194458,
+      "learning_rate": 0.0001,
+      "loss": 1.7477,
+      "step": 219
+    },
+    {
+      "epoch": 0.10735198438516591,
+      "grad_norm": 0.17979493737220764,
+      "learning_rate": 0.0001,
+      "loss": 1.7215,
+      "step": 220
+    },
+    {
+      "epoch": 0.10783994795055303,
+      "grad_norm": 0.187989741563797,
+      "learning_rate": 0.0001,
+      "loss": 1.6946,
+      "step": 221
+    },
+    {
+      "epoch": 0.10832791151594014,
+      "grad_norm": 0.18497705459594727,
+      "learning_rate": 0.0001,
+      "loss": 1.7725,
+      "step": 222
+    },
+    {
+      "epoch": 0.10881587508132726,
+      "grad_norm": 0.1895315796136856,
+      "learning_rate": 0.0001,
+      "loss": 1.7455,
+      "step": 223
+    },
+    {
+      "epoch": 0.10930383864671438,
+      "grad_norm": 0.17897574603557587,
+      "learning_rate": 0.0001,
+      "loss": 1.7297,
+      "step": 224
+    },
+    {
+      "epoch": 0.1097918022121015,
+      "grad_norm": 0.18770314753055573,
+      "learning_rate": 0.0001,
+      "loss": 1.7948,
+      "step": 225
+    },
+    {
+      "epoch": 0.11027976577748862,
+      "grad_norm": 0.1812209188938141,
+      "learning_rate": 0.0001,
+      "loss": 1.8229,
+      "step": 226
+    },
+    {
+      "epoch": 0.11076772934287574,
+      "grad_norm": 0.17030760645866394,
+      "learning_rate": 0.0001,
+      "loss": 1.6029,
+      "step": 227
+    },
+    {
+      "epoch": 0.11125569290826284,
+      "grad_norm": 0.18503767251968384,
+      "learning_rate": 0.0001,
+      "loss": 1.644,
+      "step": 228
+    },
+    {
+      "epoch": 0.11174365647364996,
+      "grad_norm": 0.17443233728408813,
+      "learning_rate": 0.0001,
+      "loss": 1.7024,
+      "step": 229
+    },
+    {
+      "epoch": 0.11223162003903708,
+      "grad_norm": 0.1859743744134903,
+      "learning_rate": 0.0001,
+      "loss": 1.7859,
+      "step": 230
+    },
+    {
+      "epoch": 0.1127195836044242,
+      "grad_norm": 0.1692182421684265,
+      "learning_rate": 0.0001,
+      "loss": 1.6996,
+      "step": 231
+    },
+    {
+      "epoch": 0.11320754716981132,
+      "grad_norm": 0.16695043444633484,
+      "learning_rate": 0.0001,
+      "loss": 1.7185,
+      "step": 232
+    },
+    {
+      "epoch": 0.11369551073519844,
+      "grad_norm": 0.18184787034988403,
+      "learning_rate": 0.0001,
+      "loss": 1.712,
+      "step": 233
+    },
+    {
+      "epoch": 0.11418347430058555,
+      "grad_norm": 0.19107092916965485,
+      "learning_rate": 0.0001,
+      "loss": 1.8902,
+      "step": 234
+    },
+    {
+      "epoch": 0.11467143786597267,
+      "grad_norm": 0.1724960058927536,
+      "learning_rate": 0.0001,
+      "loss": 1.7464,
+      "step": 235
+    },
+    {
+      "epoch": 0.11515940143135979,
+      "grad_norm": 0.17673127353191376,
+      "learning_rate": 0.0001,
+      "loss": 1.785,
+      "step": 236
+    },
+    {
+      "epoch": 0.11564736499674691,
+      "grad_norm": 0.18474438786506653,
+      "learning_rate": 0.0001,
+      "loss": 1.8143,
+      "step": 237
+    },
+    {
+      "epoch": 0.11613532856213403,
+      "grad_norm": 0.17361678183078766,
+      "learning_rate": 0.0001,
+      "loss": 1.7558,
+      "step": 238
+    },
+    {
+      "epoch": 0.11662329212752115,
+      "grad_norm": 0.17701455950737,
+      "learning_rate": 0.0001,
+      "loss": 1.5568,
+      "step": 239
+    },
+    {
+      "epoch": 0.11711125569290826,
+      "grad_norm": 0.18372413516044617,
+      "learning_rate": 0.0001,
+      "loss": 1.7913,
+      "step": 240
+    },
+    {
+      "epoch": 0.11759921925829538,
+      "grad_norm": 0.17780154943466187,
+      "learning_rate": 0.0001,
+      "loss": 1.668,
+      "step": 241
+    },
+    {
+      "epoch": 0.1180871828236825,
+      "grad_norm": 0.17763271927833557,
+      "learning_rate": 0.0001,
+      "loss": 1.7006,
+      "step": 242
+    },
+    {
+      "epoch": 0.11857514638906962,
+      "grad_norm": 0.17323441803455353,
+      "learning_rate": 0.0001,
+      "loss": 1.5985,
+      "step": 243
+    },
+    {
+      "epoch": 0.11906310995445674,
+      "grad_norm": 0.1981297731399536,
+      "learning_rate": 0.0001,
+      "loss": 1.7938,
+      "step": 244
+    },
+    {
+      "epoch": 0.11955107351984386,
+      "grad_norm": 0.1856129914522171,
+      "learning_rate": 0.0001,
+      "loss": 1.7469,
+      "step": 245
+    },
+    {
+      "epoch": 0.12003903708523098,
+      "grad_norm": 0.17878711223602295,
+      "learning_rate": 0.0001,
+      "loss": 1.7156,
+      "step": 246
+    },
+    {
+      "epoch": 0.12052700065061808,
+      "grad_norm": 0.18860337138175964,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 247
+    },
+    {
+      "epoch": 0.1210149642160052,
+      "grad_norm": 0.17960023880004883,
+      "learning_rate": 0.0001,
+      "loss": 1.7484,
+      "step": 248
+    },
+    {
+      "epoch": 0.12150292778139232,
+      "grad_norm": 0.21390804648399353,
+      "learning_rate": 0.0001,
+      "loss": 1.7815,
+      "step": 249
+    },
+    {
+      "epoch": 0.12199089134677944,
+      "grad_norm": 0.18213345110416412,
+      "learning_rate": 0.0001,
+      "loss": 1.8368,
+      "step": 250
+    },
+    {
+      "epoch": 0.12247885491216656,
+      "grad_norm": 0.19667306542396545,
+      "learning_rate": 0.0001,
+      "loss": 1.7547,
+      "step": 251
+    },
+    {
+      "epoch": 0.12296681847755368,
+      "grad_norm": 0.18796378374099731,
+      "learning_rate": 0.0001,
+      "loss": 1.6831,
+      "step": 252
+    },
+    {
+      "epoch": 0.12345478204294079,
+      "grad_norm": 0.18432985246181488,
+      "learning_rate": 0.0001,
+      "loss": 1.8219,
+      "step": 253
+    },
+    {
+      "epoch": 0.12394274560832791,
+      "grad_norm": 0.19263121485710144,
+      "learning_rate": 0.0001,
+      "loss": 1.7033,
+      "step": 254
+    },
+    {
+      "epoch": 0.12443070917371503,
+      "grad_norm": 0.19383201003074646,
+      "learning_rate": 0.0001,
+      "loss": 1.723,
+      "step": 255
+    },
+    {
+      "epoch": 0.12491867273910215,
+      "grad_norm": 0.17456290125846863,
+      "learning_rate": 0.0001,
+      "loss": 1.7354,
+      "step": 256
+    },
+    {
+      "epoch": 0.12540663630448926,
+      "grad_norm": 0.2073334902524948,
+      "learning_rate": 0.0001,
+      "loss": 1.7359,
+      "step": 257
+    },
+    {
+      "epoch": 0.1258945998698764,
+      "grad_norm": 0.1819145232439041,
+      "learning_rate": 0.0001,
+      "loss": 1.661,
+      "step": 258
+    },
+    {
+      "epoch": 0.1263825634352635,
+      "grad_norm": 0.18823570013046265,
+      "learning_rate": 0.0001,
+      "loss": 1.7093,
+      "step": 259
+    },
+    {
+      "epoch": 0.12687052700065063,
+      "grad_norm": 0.2142113894224167,
+      "learning_rate": 0.0001,
+      "loss": 1.7367,
+      "step": 260
+    },
+    {
+      "epoch": 0.12735849056603774,
+      "grad_norm": 0.17133839428424835,
+      "learning_rate": 0.0001,
+      "loss": 1.7257,
+      "step": 261
+    },
+    {
+      "epoch": 0.12784645413142484,
+      "grad_norm": 0.20852066576480865,
+      "learning_rate": 0.0001,
+      "loss": 1.7453,
+      "step": 262
+    },
+    {
+      "epoch": 0.12833441769681198,
+      "grad_norm": 0.19172458350658417,
+      "learning_rate": 0.0001,
+      "loss": 1.817,
+      "step": 263
+    },
+    {
+      "epoch": 0.12882238126219908,
+      "grad_norm": 0.1805960088968277,
+      "learning_rate": 0.0001,
+      "loss": 1.7679,
+      "step": 264
+    },
+    {
+      "epoch": 0.12931034482758622,
+      "grad_norm": 0.2055218368768692,
+      "learning_rate": 0.0001,
+      "loss": 1.7874,
+      "step": 265
+    },
+    {
+      "epoch": 0.12979830839297332,
+      "grad_norm": 0.16831174492835999,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 266
+    },
+    {
+      "epoch": 0.13028627195836046,
+      "grad_norm": 0.17563872039318085,
+      "learning_rate": 0.0001,
+      "loss": 1.7768,
+      "step": 267
+    },
+    {
+      "epoch": 0.13077423552374756,
+      "grad_norm": 0.1891409158706665,
+      "learning_rate": 0.0001,
+      "loss": 1.7653,
+      "step": 268
+    },
+    {
+      "epoch": 0.13126219908913467,
+      "grad_norm": 0.2160748541355133,
+      "learning_rate": 0.0001,
+      "loss": 1.6957,
+      "step": 269
+    },
+    {
+      "epoch": 0.1317501626545218,
+      "grad_norm": 0.16802331805229187,
+      "learning_rate": 0.0001,
+      "loss": 1.6474,
+      "step": 270
+    },
+    {
+      "epoch": 0.1322381262199089,
+      "grad_norm": 0.21498991549015045,
+      "learning_rate": 0.0001,
+      "loss": 1.7201,
+      "step": 271
+    },
+    {
+      "epoch": 0.13272608978529604,
+      "grad_norm": 0.1941365897655487,
+      "learning_rate": 0.0001,
+      "loss": 1.7387,
+      "step": 272
+    },
+    {
+      "epoch": 0.13321405335068315,
+      "grad_norm": 0.19020740687847137,
+      "learning_rate": 0.0001,
+      "loss": 1.6985,
+      "step": 273
+    },
+    {
+      "epoch": 0.13370201691607025,
+      "grad_norm": 0.18627683818340302,
+      "learning_rate": 0.0001,
+      "loss": 1.7752,
+      "step": 274
+    },
+    {
+      "epoch": 0.1341899804814574,
+      "grad_norm": 0.1916990429162979,
+      "learning_rate": 0.0001,
+      "loss": 1.7438,
+      "step": 275
+    },
+    {
+      "epoch": 0.1346779440468445,
+      "grad_norm": 0.18649545311927795,
+      "learning_rate": 0.0001,
+      "loss": 1.663,
+      "step": 276
+    },
+    {
+      "epoch": 0.13516590761223163,
+      "grad_norm": 0.17986956238746643,
+      "learning_rate": 0.0001,
+      "loss": 1.7905,
+      "step": 277
+    },
+    {
+      "epoch": 0.13565387117761873,
+      "grad_norm": 0.18601469695568085,
+      "learning_rate": 0.0001,
+      "loss": 1.5608,
+      "step": 278
+    },
+    {
+      "epoch": 0.13614183474300587,
+      "grad_norm": 0.19612380862236023,
+      "learning_rate": 0.0001,
+      "loss": 1.7317,
+      "step": 279
+    },
+    {
+      "epoch": 0.13662979830839297,
+      "grad_norm": 0.17528840899467468,
+      "learning_rate": 0.0001,
+      "loss": 1.7114,
+      "step": 280
+    },
+    {
+      "epoch": 0.13711776187378008,
+      "grad_norm": 0.196456179022789,
+      "learning_rate": 0.0001,
+      "loss": 1.674,
+      "step": 281
+    },
+    {
+      "epoch": 0.13760572543916721,
+      "grad_norm": 0.18218737840652466,
+      "learning_rate": 0.0001,
+      "loss": 1.6971,
+      "step": 282
+    },
+    {
+      "epoch": 0.13809368900455432,
+      "grad_norm": 0.18146923184394836,
+      "learning_rate": 0.0001,
+      "loss": 1.7656,
+      "step": 283
+    },
+    {
+      "epoch": 0.13858165256994145,
+      "grad_norm": 0.17707045376300812,
+      "learning_rate": 0.0001,
+      "loss": 1.6322,
+      "step": 284
+    },
+    {
+      "epoch": 0.13906961613532856,
+      "grad_norm": 0.18990135192871094,
+      "learning_rate": 0.0001,
+      "loss": 1.7412,
+      "step": 285
+    },
+    {
+      "epoch": 0.13955757970071567,
+      "grad_norm": 0.17993967235088348,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 286
+    },
+    {
+      "epoch": 0.1400455432661028,
+      "grad_norm": 0.20445284247398376,
+      "learning_rate": 0.0001,
+      "loss": 1.9164,
+      "step": 287
+    },
+    {
+      "epoch": 0.1405335068314899,
+      "grad_norm": 0.18881991505622864,
+      "learning_rate": 0.0001,
+      "loss": 1.8395,
+      "step": 288
+    },
+    {
+      "epoch": 0.14102147039687704,
+      "grad_norm": 0.17268231511116028,
+      "learning_rate": 0.0001,
+      "loss": 1.6494,
+      "step": 289
+    },
+    {
+      "epoch": 0.14150943396226415,
+      "grad_norm": 0.17375007271766663,
+      "learning_rate": 0.0001,
+      "loss": 1.6968,
+      "step": 290
+    },
+    {
+      "epoch": 0.14199739752765128,
+      "grad_norm": 0.17844517529010773,
+      "learning_rate": 0.0001,
+      "loss": 1.8686,
+      "step": 291
+    },
+    {
+      "epoch": 0.1424853610930384,
+      "grad_norm": 0.18538935482501984,
+      "learning_rate": 0.0001,
+      "loss": 1.8035,
+      "step": 292
+    },
+    {
+      "epoch": 0.1429733246584255,
+      "grad_norm": 0.18314018845558167,
+      "learning_rate": 0.0001,
+      "loss": 1.8051,
+      "step": 293
+    },
+    {
+      "epoch": 0.14346128822381263,
+      "grad_norm": 0.18008261919021606,
+      "learning_rate": 0.0001,
+      "loss": 1.7992,
+      "step": 294
+    },
+    {
+      "epoch": 0.14394925178919973,
+      "grad_norm": 0.19243541359901428,
+      "learning_rate": 0.0001,
+      "loss": 1.7394,
+      "step": 295
+    },
+    {
+      "epoch": 0.14443721535458687,
+      "grad_norm": 0.18523713946342468,
+      "learning_rate": 0.0001,
+      "loss": 1.7845,
+      "step": 296
+    },
+    {
+      "epoch": 0.14492517891997397,
+      "grad_norm": 0.1781051605939865,
+      "learning_rate": 0.0001,
+      "loss": 1.6748,
+      "step": 297
+    },
+    {
+      "epoch": 0.14541314248536108,
+      "grad_norm": 0.18994836509227753,
+      "learning_rate": 0.0001,
+      "loss": 1.704,
+      "step": 298
+    },
+    {
+      "epoch": 0.1459011060507482,
+      "grad_norm": 0.17285694181919098,
+      "learning_rate": 0.0001,
+      "loss": 1.7832,
+      "step": 299
+    },
+    {
+      "epoch": 0.14638906961613532,
+      "grad_norm": 0.20339974761009216,
+      "learning_rate": 0.0001,
+      "loss": 1.7191,
+      "step": 300
+    },
+    {
+      "epoch": 0.14687703318152245,
+      "grad_norm": 0.17608943581581116,
+      "learning_rate": 0.0001,
+      "loss": 1.6315,
+      "step": 301
+    },
+    {
+      "epoch": 0.14736499674690956,
+      "grad_norm": 0.17653749883174896,
+      "learning_rate": 0.0001,
+      "loss": 1.6948,
+      "step": 302
+    },
+    {
+      "epoch": 0.1478529603122967,
+      "grad_norm": 0.1792931854724884,
+      "learning_rate": 0.0001,
+      "loss": 1.7027,
+      "step": 303
+    },
+    {
+      "epoch": 0.1483409238776838,
+      "grad_norm": 0.18247826397418976,
+      "learning_rate": 0.0001,
+      "loss": 1.7433,
+      "step": 304
+    },
+    {
+      "epoch": 0.1488288874430709,
+      "grad_norm": 0.1712041050195694,
+      "learning_rate": 0.0001,
+      "loss": 1.6548,
+      "step": 305
+    },
+    {
+      "epoch": 0.14931685100845804,
+      "grad_norm": 0.184691920876503,
+      "learning_rate": 0.0001,
+      "loss": 1.7226,
+      "step": 306
+    },
+    {
+      "epoch": 0.14980481457384515,
+      "grad_norm": 0.1834600865840912,
+      "learning_rate": 0.0001,
+      "loss": 1.7894,
+      "step": 307
+    },
+    {
+      "epoch": 0.15029277813923228,
+      "grad_norm": 0.1753443032503128,
+      "learning_rate": 0.0001,
+      "loss": 1.636,
+      "step": 308
+    },
+    {
+      "epoch": 0.1507807417046194,
+      "grad_norm": 0.16590848565101624,
+      "learning_rate": 0.0001,
+      "loss": 1.6802,
+      "step": 309
+    },
+    {
+      "epoch": 0.1512687052700065,
+      "grad_norm": 0.17210128903388977,
+      "learning_rate": 0.0001,
+      "loss": 1.758,
+      "step": 310
+    },
+    {
+      "epoch": 0.15175666883539363,
+      "grad_norm": 0.19016823172569275,
+      "learning_rate": 0.0001,
+      "loss": 1.8243,
+      "step": 311
+    },
+    {
+      "epoch": 0.15224463240078073,
+      "grad_norm": 0.1756354421377182,
+      "learning_rate": 0.0001,
+      "loss": 1.7666,
+      "step": 312
+    },
+    {
+      "epoch": 0.15273259596616787,
+      "grad_norm": 0.19266565144062042,
+      "learning_rate": 0.0001,
+      "loss": 1.7856,
+      "step": 313
+    },
+    {
+      "epoch": 0.15322055953155497,
+      "grad_norm": 0.17626765370368958,
+      "learning_rate": 0.0001,
+      "loss": 1.7453,
+      "step": 314
+    },
+    {
+      "epoch": 0.1537085230969421,
+      "grad_norm": 0.1796361356973648,
+      "learning_rate": 0.0001,
+      "loss": 1.8428,
+      "step": 315
+    },
+    {
+      "epoch": 0.1541964866623292,
+      "grad_norm": 0.1971481889486313,
+      "learning_rate": 0.0001,
+      "loss": 1.8298,
+      "step": 316
+    },
+    {
+      "epoch": 0.15468445022771632,
+      "grad_norm": 0.17479249835014343,
+      "learning_rate": 0.0001,
+      "loss": 1.7243,
+      "step": 317
+    },
+    {
+      "epoch": 0.15517241379310345,
+      "grad_norm": 0.18558745086193085,
+      "learning_rate": 0.0001,
+      "loss": 1.8265,
+      "step": 318
+    },
+    {
+      "epoch": 0.15566037735849056,
+      "grad_norm": 0.17821088433265686,
+      "learning_rate": 0.0001,
+      "loss": 1.6735,
+      "step": 319
+    },
+    {
+      "epoch": 0.1561483409238777,
+      "grad_norm": 0.17939302325248718,
+      "learning_rate": 0.0001,
+      "loss": 1.7158,
+      "step": 320
+    },
+    {
+      "epoch": 0.1566363044892648,
+      "grad_norm": 0.17538347840309143,
+      "learning_rate": 0.0001,
+      "loss": 1.7467,
+      "step": 321
+    },
+    {
+      "epoch": 0.1571242680546519,
+      "grad_norm": 0.1796545684337616,
+      "learning_rate": 0.0001,
+      "loss": 1.7148,
+      "step": 322
+    },
+    {
+      "epoch": 0.15761223162003904,
+      "grad_norm": 0.19828006625175476,
+      "learning_rate": 0.0001,
+      "loss": 1.8431,
+      "step": 323
+    },
+    {
+      "epoch": 0.15810019518542615,
+      "grad_norm": 0.17246133089065552,
+      "learning_rate": 0.0001,
+      "loss": 1.7291,
+      "step": 324
+    },
+    {
+      "epoch": 0.15858815875081328,
+      "grad_norm": 0.1835339218378067,
+      "learning_rate": 0.0001,
+      "loss": 1.7319,
+      "step": 325
+    },
+    {
+      "epoch": 0.15907612231620039,
+      "grad_norm": 0.18122561275959015,
+      "learning_rate": 0.0001,
+      "loss": 1.7263,
+      "step": 326
+    },
+    {
+      "epoch": 0.15956408588158752,
+      "grad_norm": 0.19297321140766144,
+      "learning_rate": 0.0001,
+      "loss": 1.8792,
+      "step": 327
+    },
+    {
+      "epoch": 0.16005204944697463,
+      "grad_norm": 0.1762656420469284,
+      "learning_rate": 0.0001,
+      "loss": 1.7495,
+      "step": 328
+    },
+    {
+      "epoch": 0.16054001301236173,
+      "grad_norm": 0.17146944999694824,
+      "learning_rate": 0.0001,
+      "loss": 1.7089,
+      "step": 329
+    },
+    {
+      "epoch": 0.16102797657774887,
+      "grad_norm": 0.17192597687244415,
+      "learning_rate": 0.0001,
+      "loss": 1.694,
+      "step": 330
+    },
+    {
+      "epoch": 0.16151594014313597,
+      "grad_norm": 0.17271386086940765,
+      "learning_rate": 0.0001,
+      "loss": 1.6223,
+      "step": 331
+    },
+    {
+      "epoch": 0.1620039037085231,
+      "grad_norm": 0.17589011788368225,
+      "learning_rate": 0.0001,
+      "loss": 1.7123,
+      "step": 332
+    },
+    {
+      "epoch": 0.1624918672739102,
+      "grad_norm": 0.17920418083667755,
+      "learning_rate": 0.0001,
+      "loss": 1.6938,
+      "step": 333
+    },
+    {
+      "epoch": 0.16297983083929735,
+      "grad_norm": 0.16645678877830505,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 334
+    },
+    {
+      "epoch": 0.16346779440468445,
+      "grad_norm": 0.1698988974094391,
+      "learning_rate": 0.0001,
+      "loss": 1.7562,
+      "step": 335
+    },
+    {
+      "epoch": 0.16395575797007156,
+      "grad_norm": 0.17255748808383942,
+      "learning_rate": 0.0001,
+      "loss": 1.7408,
+      "step": 336
+    },
+    {
+      "epoch": 0.1644437215354587,
+      "grad_norm": 0.16908328235149384,
+      "learning_rate": 0.0001,
+      "loss": 1.711,
+      "step": 337
+    },
+    {
+      "epoch": 0.1649316851008458,
+      "grad_norm": 0.17891424894332886,
+      "learning_rate": 0.0001,
+      "loss": 1.7199,
+      "step": 338
+    },
+    {
+      "epoch": 0.16541964866623293,
+      "grad_norm": 0.17500531673431396,
+      "learning_rate": 0.0001,
+      "loss": 1.8027,
+      "step": 339
+    },
+    {
+      "epoch": 0.16590761223162004,
+      "grad_norm": 0.1908222734928131,
+      "learning_rate": 0.0001,
+      "loss": 1.7267,
+      "step": 340
+    },
+    {
+      "epoch": 0.16639557579700714,
+      "grad_norm": 0.16457560658454895,
+      "learning_rate": 0.0001,
+      "loss": 1.6551,
+      "step": 341
+    },
+    {
+      "epoch": 0.16688353936239428,
+      "grad_norm": 0.17455148696899414,
+      "learning_rate": 0.0001,
+      "loss": 1.7536,
+      "step": 342
+    },
+    {
+      "epoch": 0.16737150292778138,
+      "grad_norm": 0.24865932762622833,
+      "learning_rate": 0.0001,
+      "loss": 1.7038,
+      "step": 343
+    },
+    {
+      "epoch": 0.16785946649316852,
+      "grad_norm": 0.16769102215766907,
+      "learning_rate": 0.0001,
+      "loss": 1.6666,
+      "step": 344
+    },
+    {
+      "epoch": 0.16834743005855562,
+      "grad_norm": 0.17845629155635834,
+      "learning_rate": 0.0001,
+      "loss": 1.7729,
+      "step": 345
+    },
+    {
+      "epoch": 0.16883539362394276,
+      "grad_norm": 0.18893101811408997,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 346
+    },
+    {
+      "epoch": 0.16932335718932987,
+      "grad_norm": 0.17489705979824066,
+      "learning_rate": 0.0001,
+      "loss": 1.6451,
+      "step": 347
+    },
+    {
+      "epoch": 0.16981132075471697,
+      "grad_norm": 0.1895252764225006,
+      "learning_rate": 0.0001,
+      "loss": 1.6664,
+      "step": 348
+    },
+    {
+      "epoch": 0.1702992843201041,
+      "grad_norm": 0.18796460330486298,
+      "learning_rate": 0.0001,
+      "loss": 1.8179,
+      "step": 349
+    },
+    {
+      "epoch": 0.1707872478854912,
+      "grad_norm": 0.18239444494247437,
+      "learning_rate": 0.0001,
+      "loss": 1.7895,
+      "step": 350
+    },
+    {
+      "epoch": 0.17127521145087835,
+      "grad_norm": 0.18578602373600006,
+      "learning_rate": 0.0001,
+      "loss": 1.7201,
+      "step": 351
+    },
+    {
+      "epoch": 0.17176317501626545,
+      "grad_norm": 0.17505811154842377,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 352
+    },
+    {
+      "epoch": 0.17225113858165256,
+      "grad_norm": 0.16880185902118683,
+      "learning_rate": 0.0001,
+      "loss": 1.7064,
+      "step": 353
+    },
+    {
+      "epoch": 0.1727391021470397,
+      "grad_norm": 0.1847655326128006,
+      "learning_rate": 0.0001,
+      "loss": 1.6227,
+      "step": 354
+    },
+    {
+      "epoch": 0.1732270657124268,
+      "grad_norm": 0.18033885955810547,
+      "learning_rate": 0.0001,
+      "loss": 1.7613,
+      "step": 355
+    },
+    {
+      "epoch": 0.17371502927781393,
+      "grad_norm": 0.2022799551486969,
+      "learning_rate": 0.0001,
+      "loss": 1.6975,
+      "step": 356
+    },
+    {
+      "epoch": 0.17420299284320104,
+      "grad_norm": 0.18487118184566498,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 357
+    },
+    {
+      "epoch": 0.17469095640858817,
+      "grad_norm": 0.18200282752513885,
+      "learning_rate": 0.0001,
+      "loss": 1.8013,
+      "step": 358
+    },
+    {
+      "epoch": 0.17517891997397528,
+      "grad_norm": 0.16840700805187225,
+      "learning_rate": 0.0001,
+      "loss": 1.6904,
+      "step": 359
+    },
+    {
+      "epoch": 0.17566688353936238,
+      "grad_norm": 0.17556121945381165,
+      "learning_rate": 0.0001,
+      "loss": 1.7331,
+      "step": 360
+    },
+    {
+      "epoch": 0.17615484710474952,
+      "grad_norm": 0.18641792237758636,
+      "learning_rate": 0.0001,
+      "loss": 1.8248,
+      "step": 361
+    },
+    {
+      "epoch": 0.17664281067013662,
+      "grad_norm": 0.16753801703453064,
+      "learning_rate": 0.0001,
+      "loss": 1.591,
+      "step": 362
+    },
+    {
+      "epoch": 0.17713077423552376,
+      "grad_norm": 0.16265541315078735,
+      "learning_rate": 0.0001,
+      "loss": 1.5814,
+      "step": 363
+    },
+    {
+      "epoch": 0.17761873780091086,
+      "grad_norm": 0.17881396412849426,
+      "learning_rate": 0.0001,
+      "loss": 1.8452,
+      "step": 364
+    },
+    {
+      "epoch": 0.17810670136629797,
+      "grad_norm": 0.18160590529441833,
+      "learning_rate": 0.0001,
+      "loss": 1.7977,
+      "step": 365
+    },
+    {
+      "epoch": 0.1785946649316851,
+      "grad_norm": 0.1778435856103897,
+      "learning_rate": 0.0001,
+      "loss": 1.7319,
+      "step": 366
+    },
+    {
+      "epoch": 0.1790826284970722,
+      "grad_norm": 0.17236903309822083,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 367
+    },
+    {
+      "epoch": 0.17957059206245934,
+      "grad_norm": 0.16980677843093872,
+      "learning_rate": 0.0001,
+      "loss": 1.6814,
+      "step": 368
+    },
+    {
+      "epoch": 0.18005855562784645,
+      "grad_norm": 0.17113539576530457,
+      "learning_rate": 0.0001,
+      "loss": 1.5835,
+      "step": 369
+    },
+    {
+      "epoch": 0.18054651919323358,
+      "grad_norm": 0.22926300764083862,
+      "learning_rate": 0.0001,
+      "loss": 1.7127,
+      "step": 370
+    },
+    {
+      "epoch": 0.1810344827586207,
+      "grad_norm": 0.1766396313905716,
+      "learning_rate": 0.0001,
+      "loss": 1.8002,
+      "step": 371
+    },
+    {
+      "epoch": 0.1815224463240078,
+      "grad_norm": 0.1911155730485916,
+      "learning_rate": 0.0001,
+      "loss": 1.7287,
+      "step": 372
+    },
+    {
+      "epoch": 0.18201040988939493,
+      "grad_norm": 0.1996450275182724,
+      "learning_rate": 0.0001,
+      "loss": 1.5601,
+      "step": 373
+    },
+    {
+      "epoch": 0.18249837345478204,
+      "grad_norm": 0.17531970143318176,
+      "learning_rate": 0.0001,
+      "loss": 1.674,
+      "step": 374
+    },
+    {
+      "epoch": 0.18298633702016917,
+      "grad_norm": 0.19017955660820007,
+      "learning_rate": 0.0001,
+      "loss": 1.8052,
+      "step": 375
+    },
+    {
+      "epoch": 0.18347430058555628,
+      "grad_norm": 0.195291206240654,
+      "learning_rate": 0.0001,
+      "loss": 1.6787,
+      "step": 376
+    },
+    {
+      "epoch": 0.18396226415094338,
+      "grad_norm": 0.18030132353305817,
+      "learning_rate": 0.0001,
+      "loss": 1.6931,
+      "step": 377
+    },
+    {
+      "epoch": 0.18445022771633052,
+      "grad_norm": 0.1725359857082367,
+      "learning_rate": 0.0001,
+      "loss": 1.5814,
+      "step": 378
+    },
+    {
+      "epoch": 0.18493819128171762,
+      "grad_norm": 0.18235339224338531,
+      "learning_rate": 0.0001,
+      "loss": 1.7759,
+      "step": 379
+    },
+    {
+      "epoch": 0.18542615484710476,
+      "grad_norm": 0.19052359461784363,
+      "learning_rate": 0.0001,
+      "loss": 1.7898,
+      "step": 380
+    },
+    {
+      "epoch": 0.18591411841249186,
+      "grad_norm": 0.1713322550058365,
+      "learning_rate": 0.0001,
+      "loss": 1.623,
+      "step": 381
+    },
+    {
+      "epoch": 0.186402081977879,
+      "grad_norm": 0.19699741899967194,
+      "learning_rate": 0.0001,
+      "loss": 1.7517,
+      "step": 382
+    },
+    {
+      "epoch": 0.1868900455432661,
+      "grad_norm": 0.17510955035686493,
+      "learning_rate": 0.0001,
+      "loss": 1.7045,
+      "step": 383
+    },
+    {
+      "epoch": 0.1873780091086532,
+      "grad_norm": 0.17883911728858948,
+      "learning_rate": 0.0001,
+      "loss": 1.6763,
+      "step": 384
+    },
+    {
+      "epoch": 0.18786597267404034,
+      "grad_norm": 0.18562713265419006,
+      "learning_rate": 0.0001,
+      "loss": 1.6603,
+      "step": 385
+    },
+    {
+      "epoch": 0.18835393623942745,
+      "grad_norm": 0.18200963735580444,
+      "learning_rate": 0.0001,
+      "loss": 1.7698,
+      "step": 386
+    },
+    {
+      "epoch": 0.18884189980481458,
+      "grad_norm": 0.192865788936615,
+      "learning_rate": 0.0001,
+      "loss": 1.8058,
+      "step": 387
+    },
+    {
+      "epoch": 0.1893298633702017,
+      "grad_norm": 0.17498141527175903,
+      "learning_rate": 0.0001,
+      "loss": 1.657,
+      "step": 388
+    },
+    {
+      "epoch": 0.1898178269355888,
+      "grad_norm": 0.17550218105316162,
+      "learning_rate": 0.0001,
+      "loss": 1.7638,
+      "step": 389
+    },
+    {
+      "epoch": 0.19030579050097593,
+      "grad_norm": 0.19263967871665955,
+      "learning_rate": 0.0001,
+      "loss": 1.7375,
+      "step": 390
+    },
+    {
+      "epoch": 0.19079375406636304,
+      "grad_norm": 0.1728338897228241,
+      "learning_rate": 0.0001,
+      "loss": 1.7467,
+      "step": 391
+    },
+    {
+      "epoch": 0.19128171763175017,
+      "grad_norm": 0.17929600179195404,
+      "learning_rate": 0.0001,
+      "loss": 1.6489,
+      "step": 392
+    },
+    {
+      "epoch": 0.19176968119713728,
+      "grad_norm": 0.18325988948345184,
+      "learning_rate": 0.0001,
+      "loss": 1.8676,
+      "step": 393
+    },
+    {
+      "epoch": 0.1922576447625244,
+      "grad_norm": 0.17365989089012146,
+      "learning_rate": 0.0001,
+      "loss": 1.6916,
+      "step": 394
+    },
+    {
+      "epoch": 0.19274560832791152,
+      "grad_norm": 0.17361170053482056,
+      "learning_rate": 0.0001,
+      "loss": 1.7118,
+      "step": 395
+    },
+    {
+      "epoch": 0.19323357189329862,
+      "grad_norm": 0.181492879986763,
+      "learning_rate": 0.0001,
+      "loss": 1.7197,
+      "step": 396
+    },
+    {
+      "epoch": 0.19372153545868576,
+      "grad_norm": 0.19113008677959442,
+      "learning_rate": 0.0001,
+      "loss": 1.788,
+      "step": 397
+    },
+    {
+      "epoch": 0.19420949902407286,
+      "grad_norm": 0.173355832695961,
+      "learning_rate": 0.0001,
+      "loss": 1.6866,
+      "step": 398
+    },
+    {
+      "epoch": 0.19469746258946,
+      "grad_norm": 0.1797139197587967,
+      "learning_rate": 0.0001,
+      "loss": 1.7505,
+      "step": 399
+    },
+    {
+      "epoch": 0.1951854261548471,
+      "grad_norm": 0.18337444961071014,
+      "learning_rate": 0.0001,
+      "loss": 1.7099,
+      "step": 400
+    },
+    {
+      "epoch": 0.1956733897202342,
+      "grad_norm": 0.17387695610523224,
+      "learning_rate": 0.0001,
+      "loss": 1.737,
+      "step": 401
+    },
+    {
+      "epoch": 0.19616135328562134,
+      "grad_norm": 0.1695685237646103,
+      "learning_rate": 0.0001,
+      "loss": 1.6916,
+      "step": 402
+    },
+    {
+      "epoch": 0.19664931685100845,
+      "grad_norm": 0.1874959021806717,
+      "learning_rate": 0.0001,
+      "loss": 1.6919,
+      "step": 403
+    },
+    {
+      "epoch": 0.19713728041639558,
+      "grad_norm": 0.17886492609977722,
+      "learning_rate": 0.0001,
+      "loss": 1.737,
+      "step": 404
+    },
+    {
+      "epoch": 0.1976252439817827,
+      "grad_norm": 0.19390465319156647,
+      "learning_rate": 0.0001,
+      "loss": 1.8003,
+      "step": 405
+    },
+    {
+      "epoch": 0.19811320754716982,
+      "grad_norm": 0.17292645573616028,
+      "learning_rate": 0.0001,
+      "loss": 1.6714,
+      "step": 406
+    },
+    {
+      "epoch": 0.19860117111255693,
+      "grad_norm": 0.16998599469661713,
+      "learning_rate": 0.0001,
+      "loss": 1.7242,
+      "step": 407
+    },
+    {
+      "epoch": 0.19908913467794404,
+      "grad_norm": 0.18668459355831146,
+      "learning_rate": 0.0001,
+      "loss": 1.7025,
+      "step": 408
+    },
+    {
+      "epoch": 0.19957709824333117,
+      "grad_norm": 0.16807502508163452,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 409
+    },
+    {
+      "epoch": 0.20006506180871828,
+      "grad_norm": 0.1849876344203949,
+      "learning_rate": 0.0001,
+      "loss": 1.8173,
+      "step": 410
+    },
+    {
+      "epoch": 0.2005530253741054,
+      "grad_norm": 0.18935902416706085,
+      "learning_rate": 0.0001,
+      "loss": 1.7108,
+      "step": 411
+    },
+    {
+      "epoch": 0.20104098893949252,
+      "grad_norm": 0.17630939185619354,
+      "learning_rate": 0.0001,
+      "loss": 1.7023,
+      "step": 412
+    },
+    {
+      "epoch": 0.20152895250487965,
+      "grad_norm": 0.19990061223506927,
+      "learning_rate": 0.0001,
+      "loss": 1.6862,
+      "step": 413
+    },
+    {
+      "epoch": 0.20201691607026676,
+      "grad_norm": 0.18538086116313934,
+      "learning_rate": 0.0001,
+      "loss": 1.796,
+      "step": 414
+    },
+    {
+      "epoch": 0.20250487963565386,
+      "grad_norm": 0.18812508881092072,
+      "learning_rate": 0.0001,
+      "loss": 1.7034,
+      "step": 415
+    },
+    {
+      "epoch": 0.202992843201041,
+      "grad_norm": 0.19069646298885345,
+      "learning_rate": 0.0001,
+      "loss": 1.7504,
+      "step": 416
+    },
+    {
+      "epoch": 0.2034808067664281,
+      "grad_norm": 0.17794154584407806,
+      "learning_rate": 0.0001,
+      "loss": 1.6469,
+      "step": 417
+    },
+    {
+      "epoch": 0.20396877033181524,
+      "grad_norm": 0.17641998827457428,
+      "learning_rate": 0.0001,
+      "loss": 1.7526,
+      "step": 418
+    },
+    {
+      "epoch": 0.20445673389720234,
+      "grad_norm": 0.19693951308727264,
+      "learning_rate": 0.0001,
+      "loss": 1.7007,
+      "step": 419
+    },
+    {
+      "epoch": 0.20494469746258945,
+      "grad_norm": 0.1921786069869995,
+      "learning_rate": 0.0001,
+      "loss": 1.7514,
+      "step": 420
+    },
+    {
+      "epoch": 0.20543266102797658,
+      "grad_norm": 0.1899469792842865,
+      "learning_rate": 0.0001,
+      "loss": 1.7508,
+      "step": 421
+    },
+    {
+      "epoch": 0.2059206245933637,
+      "grad_norm": 0.16994713246822357,
+      "learning_rate": 0.0001,
+      "loss": 1.6313,
+      "step": 422
+    },
+    {
+      "epoch": 0.20640858815875082,
+      "grad_norm": 0.20480570197105408,
+      "learning_rate": 0.0001,
+      "loss": 1.7714,
+      "step": 423
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 0.20870919525623322,
+      "learning_rate": 0.0001,
+      "loss": 1.7782,
+      "step": 424
+    },
+    {
+      "epoch": 0.20738451528952506,
+      "grad_norm": 0.18410471081733704,
+      "learning_rate": 0.0001,
+      "loss": 1.72,
+      "step": 425
+    },
+    {
+      "epoch": 0.20787247885491217,
+      "grad_norm": 0.23531974852085114,
+      "learning_rate": 0.0001,
+      "loss": 1.8923,
+      "step": 426
+    },
+    {
+      "epoch": 0.20836044242029927,
+      "grad_norm": 0.18552608788013458,
+      "learning_rate": 0.0001,
+      "loss": 1.7272,
+      "step": 427
+    },
+    {
+      "epoch": 0.2088484059856864,
+      "grad_norm": 0.2085346281528473,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 428
+    },
+    {
+      "epoch": 0.20933636955107351,
+      "grad_norm": 0.1959279626607895,
+      "learning_rate": 0.0001,
+      "loss": 1.6288,
+      "step": 429
+    },
+    {
+      "epoch": 0.20982433311646065,
+      "grad_norm": 0.17610879242420197,
+      "learning_rate": 0.0001,
+      "loss": 1.7151,
+      "step": 430
+    },
+    {
+      "epoch": 0.21031229668184775,
+      "grad_norm": 0.1928284466266632,
+      "learning_rate": 0.0001,
+      "loss": 1.687,
+      "step": 431
+    },
+    {
+      "epoch": 0.21080026024723486,
+      "grad_norm": 0.199452742934227,
+      "learning_rate": 0.0001,
+      "loss": 1.7704,
+      "step": 432
+    },
+    {
+      "epoch": 0.211288223812622,
+      "grad_norm": 0.18074338138103485,
+      "learning_rate": 0.0001,
+      "loss": 1.7899,
+      "step": 433
+    },
+    {
+      "epoch": 0.2117761873780091,
+      "grad_norm": 0.19121356308460236,
+      "learning_rate": 0.0001,
+      "loss": 1.694,
+      "step": 434
+    },
+    {
+      "epoch": 0.21226415094339623,
+      "grad_norm": 0.18307030200958252,
+      "learning_rate": 0.0001,
+      "loss": 1.6335,
+      "step": 435
+    },
+    {
+      "epoch": 0.21275211450878334,
+      "grad_norm": 0.18400311470031738,
+      "learning_rate": 0.0001,
+      "loss": 1.7526,
+      "step": 436
+    },
+    {
+      "epoch": 0.21324007807417047,
+      "grad_norm": 0.1944567859172821,
+      "learning_rate": 0.0001,
+      "loss": 1.7884,
+      "step": 437
+    },
+    {
+      "epoch": 0.21372804163955758,
+      "grad_norm": 0.18847782909870148,
+      "learning_rate": 0.0001,
+      "loss": 1.6859,
+      "step": 438
+    },
+    {
+      "epoch": 0.2142160052049447,
+      "grad_norm": 0.17663119733333588,
+      "learning_rate": 0.0001,
+      "loss": 1.615,
+      "step": 439
+    },
+    {
+      "epoch": 0.21470396877033182,
+      "grad_norm": 0.18704909086227417,
+      "learning_rate": 0.0001,
+      "loss": 1.7352,
+      "step": 440
+    },
+    {
+      "epoch": 0.21519193233571893,
+      "grad_norm": 0.19525641202926636,
+      "learning_rate": 0.0001,
+      "loss": 1.6241,
+      "step": 441
+    },
+    {
+      "epoch": 0.21567989590110606,
+      "grad_norm": 0.19030174612998962,
+      "learning_rate": 0.0001,
+      "loss": 1.7425,
+      "step": 442
+    },
+    {
+      "epoch": 0.21616785946649317,
+      "grad_norm": 0.18872150778770447,
+      "learning_rate": 0.0001,
+      "loss": 1.7177,
+      "step": 443
+    },
+    {
+      "epoch": 0.21665582303188027,
+      "grad_norm": 0.17374157905578613,
+      "learning_rate": 0.0001,
+      "loss": 1.7236,
+      "step": 444
+    },
+    {
+      "epoch": 0.2171437865972674,
+      "grad_norm": 0.18159011006355286,
+      "learning_rate": 0.0001,
+      "loss": 1.6885,
+      "step": 445
+    },
+    {
+      "epoch": 0.2176317501626545,
+      "grad_norm": 0.18726180493831635,
+      "learning_rate": 0.0001,
+      "loss": 1.8226,
+      "step": 446
+    },
+    {
+      "epoch": 0.21811971372804165,
+      "grad_norm": 0.193464457988739,
+      "learning_rate": 0.0001,
+      "loss": 1.7834,
+      "step": 447
+    },
+    {
+      "epoch": 0.21860767729342875,
+      "grad_norm": 0.19700440764427185,
+      "learning_rate": 0.0001,
+      "loss": 1.6766,
+      "step": 448
+    },
+    {
+      "epoch": 0.2190956408588159,
+      "grad_norm": 0.16808220744132996,
+      "learning_rate": 0.0001,
+      "loss": 1.6773,
+      "step": 449
+    },
+    {
+      "epoch": 0.219583604424203,
+      "grad_norm": 0.1885610967874527,
+      "learning_rate": 0.0001,
+      "loss": 1.7195,
+      "step": 450
+    },
+    {
+      "epoch": 0.2200715679895901,
+      "grad_norm": 0.17235183715820312,
+      "learning_rate": 0.0001,
+      "loss": 1.6651,
+      "step": 451
+    },
+    {
+      "epoch": 0.22055953155497723,
+      "grad_norm": 0.17667032778263092,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 452
+    },
+    {
+      "epoch": 0.22104749512036434,
+      "grad_norm": 0.17659679055213928,
+      "learning_rate": 0.0001,
+      "loss": 1.8337,
+      "step": 453
+    },
+    {
+      "epoch": 0.22153545868575147,
+      "grad_norm": 0.17201969027519226,
+      "learning_rate": 0.0001,
+      "loss": 1.7385,
+      "step": 454
+    },
+    {
+      "epoch": 0.22202342225113858,
+      "grad_norm": 0.17937779426574707,
+      "learning_rate": 0.0001,
+      "loss": 1.7864,
+      "step": 455
+    },
+    {
+      "epoch": 0.2225113858165257,
+      "grad_norm": 0.1681385189294815,
+      "learning_rate": 0.0001,
+      "loss": 1.636,
+      "step": 456
+    },
+    {
+      "epoch": 0.22299934938191282,
+      "grad_norm": 0.17030152678489685,
+      "learning_rate": 0.0001,
+      "loss": 1.7613,
+      "step": 457
+    },
+    {
+      "epoch": 0.22348731294729993,
+      "grad_norm": 0.18430882692337036,
+      "learning_rate": 0.0001,
+      "loss": 1.7746,
+      "step": 458
+    },
+    {
+      "epoch": 0.22397527651268706,
+      "grad_norm": 0.17070208489894867,
+      "learning_rate": 0.0001,
+      "loss": 1.619,
+      "step": 459
+    },
+    {
+      "epoch": 0.22446324007807417,
+      "grad_norm": 0.1672583520412445,
+      "learning_rate": 0.0001,
+      "loss": 1.6935,
+      "step": 460
+    },
+    {
+      "epoch": 0.2249512036434613,
+      "grad_norm": 0.18070879578590393,
+      "learning_rate": 0.0001,
+      "loss": 1.7752,
+      "step": 461
+    },
+    {
+      "epoch": 0.2254391672088484,
+      "grad_norm": 0.17931310832500458,
+      "learning_rate": 0.0001,
+      "loss": 1.8331,
+      "step": 462
+    },
+    {
+      "epoch": 0.2259271307742355,
+      "grad_norm": 0.18687482178211212,
+      "learning_rate": 0.0001,
+      "loss": 1.7745,
+      "step": 463
+    },
+    {
+      "epoch": 0.22641509433962265,
+      "grad_norm": 0.18673428893089294,
+      "learning_rate": 0.0001,
+      "loss": 1.8001,
+      "step": 464
+    },
+    {
+      "epoch": 0.22690305790500975,
+      "grad_norm": 0.18758326768875122,
+      "learning_rate": 0.0001,
+      "loss": 1.8024,
+      "step": 465
+    },
+    {
+      "epoch": 0.2273910214703969,
+      "grad_norm": 0.17651711404323578,
+      "learning_rate": 0.0001,
+      "loss": 1.6348,
+      "step": 466
+    },
+    {
+      "epoch": 0.227878985035784,
+      "grad_norm": 0.17466424405574799,
+      "learning_rate": 0.0001,
+      "loss": 1.6529,
+      "step": 467
+    },
+    {
+      "epoch": 0.2283669486011711,
+      "grad_norm": 0.17049545049667358,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 468
+    },
+    {
+      "epoch": 0.22885491216655823,
+      "grad_norm": 0.19238895177841187,
+      "learning_rate": 0.0001,
+      "loss": 1.7262,
+      "step": 469
+    },
+    {
+      "epoch": 0.22934287573194534,
+      "grad_norm": 0.183549702167511,
+      "learning_rate": 0.0001,
+      "loss": 1.6949,
+      "step": 470
+    },
+    {
+      "epoch": 0.22983083929733247,
+      "grad_norm": 0.19222155213356018,
+      "learning_rate": 0.0001,
+      "loss": 1.7727,
+      "step": 471
+    },
+    {
+      "epoch": 0.23031880286271958,
+      "grad_norm": 0.18078762292861938,
+      "learning_rate": 0.0001,
+      "loss": 1.8166,
+      "step": 472
+    },
+    {
+      "epoch": 0.2308067664281067,
+      "grad_norm": 0.17769628763198853,
+      "learning_rate": 0.0001,
+      "loss": 1.7215,
+      "step": 473
+    },
+    {
+      "epoch": 0.23129472999349382,
+      "grad_norm": 0.1750006526708603,
+      "learning_rate": 0.0001,
+      "loss": 1.7311,
+      "step": 474
+    },
+    {
+      "epoch": 0.23178269355888093,
+      "grad_norm": 0.1803676038980484,
+      "learning_rate": 0.0001,
+      "loss": 1.7596,
+      "step": 475
+    },
+    {
+      "epoch": 0.23227065712426806,
+      "grad_norm": 0.18478356301784515,
+      "learning_rate": 0.0001,
+      "loss": 1.7262,
+      "step": 476
+    },
+    {
+      "epoch": 0.23275862068965517,
+      "grad_norm": 0.16509763896465302,
+      "learning_rate": 0.0001,
+      "loss": 1.623,
+      "step": 477
+    },
+    {
+      "epoch": 0.2332465842550423,
+      "grad_norm": 0.19317001104354858,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 478
+    },
+    {
+      "epoch": 0.2337345478204294,
+      "grad_norm": 0.18081186711788177,
+      "learning_rate": 0.0001,
+      "loss": 1.6959,
+      "step": 479
+    },
+    {
+      "epoch": 0.2342225113858165,
+      "grad_norm": 0.18306545913219452,
+      "learning_rate": 0.0001,
+      "loss": 1.7328,
+      "step": 480
+    },
+    {
+      "epoch": 0.23471047495120365,
+      "grad_norm": 0.18552261590957642,
+      "learning_rate": 0.0001,
+      "loss": 1.6847,
+      "step": 481
+    },
+    {
+      "epoch": 0.23519843851659075,
+      "grad_norm": 0.17930322885513306,
+      "learning_rate": 0.0001,
+      "loss": 1.7678,
+      "step": 482
+    },
+    {
+      "epoch": 0.23568640208197789,
+      "grad_norm": 0.17558367550373077,
+      "learning_rate": 0.0001,
+      "loss": 1.6756,
+      "step": 483
+    },
+    {
+      "epoch": 0.236174365647365,
+      "grad_norm": 0.18899041414260864,
+      "learning_rate": 0.0001,
+      "loss": 1.7778,
+      "step": 484
+    },
+    {
+      "epoch": 0.23666232921275213,
+      "grad_norm": 0.17528998851776123,
+      "learning_rate": 0.0001,
+      "loss": 1.6651,
+      "step": 485
+    },
+    {
+      "epoch": 0.23715029277813923,
+      "grad_norm": 0.16732053458690643,
+      "learning_rate": 0.0001,
+      "loss": 1.6796,
+      "step": 486
+    },
+    {
+      "epoch": 0.23763825634352634,
+      "grad_norm": 0.1849820613861084,
+      "learning_rate": 0.0001,
+      "loss": 1.737,
+      "step": 487
+    },
+    {
+      "epoch": 0.23812621990891347,
+      "grad_norm": 0.1789163500070572,
+      "learning_rate": 0.0001,
+      "loss": 1.6919,
+      "step": 488
+    },
+    {
+      "epoch": 0.23861418347430058,
+      "grad_norm": 0.1739804446697235,
+      "learning_rate": 0.0001,
+      "loss": 1.8225,
+      "step": 489
+    },
+    {
+      "epoch": 0.2391021470396877,
+      "grad_norm": 0.18246984481811523,
+      "learning_rate": 0.0001,
+      "loss": 1.734,
+      "step": 490
+    },
+    {
+      "epoch": 0.23959011060507482,
+      "grad_norm": 0.17464157938957214,
+      "learning_rate": 0.0001,
+      "loss": 1.7442,
+      "step": 491
+    },
+    {
+      "epoch": 0.24007807417046195,
+      "grad_norm": 0.19501306116580963,
+      "learning_rate": 0.0001,
+      "loss": 1.7521,
+      "step": 492
+    },
+    {
+      "epoch": 0.24056603773584906,
+      "grad_norm": 0.17958857119083405,
+      "learning_rate": 0.0001,
+      "loss": 1.8191,
+      "step": 493
+    },
+    {
+      "epoch": 0.24105400130123616,
+      "grad_norm": 0.18241986632347107,
+      "learning_rate": 0.0001,
+      "loss": 1.7709,
+      "step": 494
+    },
+    {
+      "epoch": 0.2415419648666233,
+      "grad_norm": 0.18529468774795532,
+      "learning_rate": 0.0001,
+      "loss": 1.6871,
+      "step": 495
+    },
+    {
+      "epoch": 0.2420299284320104,
+      "grad_norm": 0.18519562482833862,
+      "learning_rate": 0.0001,
+      "loss": 1.7605,
+      "step": 496
+    },
+    {
+      "epoch": 0.24251789199739754,
+      "grad_norm": 0.17868764698505402,
+      "learning_rate": 0.0001,
+      "loss": 1.725,
+      "step": 497
+    },
+    {
+      "epoch": 0.24300585556278465,
+      "grad_norm": 0.17040537297725677,
+      "learning_rate": 0.0001,
+      "loss": 1.6161,
+      "step": 498
+    },
+    {
+      "epoch": 0.24349381912817175,
+      "grad_norm": 0.1820056289434433,
+      "learning_rate": 0.0001,
+      "loss": 1.7249,
+      "step": 499
+    },
+    {
+      "epoch": 0.24398178269355889,
+      "grad_norm": 0.1877366453409195,
+      "learning_rate": 0.0001,
+      "loss": 1.6976,
+      "step": 500
+    },
+    {
+      "epoch": 0.244469746258946,
+      "grad_norm": 0.1717415153980255,
+      "learning_rate": 0.0001,
+      "loss": 1.6109,
+      "step": 501
+    },
+    {
+      "epoch": 0.24495770982433313,
+      "grad_norm": 0.17338915169239044,
+      "learning_rate": 0.0001,
+      "loss": 1.7433,
+      "step": 502
+    },
+    {
+      "epoch": 0.24544567338972023,
+      "grad_norm": 0.18489517271518707,
+      "learning_rate": 0.0001,
+      "loss": 1.7283,
+      "step": 503
+    },
+    {
+      "epoch": 0.24593363695510737,
+      "grad_norm": 0.17153921723365784,
+      "learning_rate": 0.0001,
+      "loss": 1.7261,
+      "step": 504
+    },
+    {
+      "epoch": 0.24642160052049447,
+      "grad_norm": 0.19024662673473358,
+      "learning_rate": 0.0001,
+      "loss": 1.8498,
+      "step": 505
+    },
+    {
+      "epoch": 0.24690956408588158,
+      "grad_norm": 0.1675989329814911,
+      "learning_rate": 0.0001,
+      "loss": 1.5903,
+      "step": 506
+    },
+    {
+      "epoch": 0.2473975276512687,
+      "grad_norm": 0.18422546982765198,
+      "learning_rate": 0.0001,
+      "loss": 1.7294,
+      "step": 507
+    },
+    {
+      "epoch": 0.24788549121665582,
+      "grad_norm": 0.17943088710308075,
+      "learning_rate": 0.0001,
+      "loss": 1.6842,
+      "step": 508
+    },
+    {
+      "epoch": 0.24837345478204295,
+      "grad_norm": 0.18048308789730072,
+      "learning_rate": 0.0001,
+      "loss": 1.677,
+      "step": 509
+    },
+    {
+      "epoch": 0.24886141834743006,
+      "grad_norm": 0.17185211181640625,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 510
+    },
+    {
+      "epoch": 0.24934938191281716,
+      "grad_norm": 0.1717991977930069,
+      "learning_rate": 0.0001,
+      "loss": 1.7077,
+      "step": 511
+    },
+    {
+      "epoch": 0.2498373454782043,
+      "grad_norm": 0.18661388754844666,
+      "learning_rate": 0.0001,
+      "loss": 1.8163,
+      "step": 512
+    },
+    {
+      "epoch": 0.2503253090435914,
+      "grad_norm": 0.19672876596450806,
+      "learning_rate": 0.0001,
+      "loss": 1.7733,
+      "step": 513
+    },
+    {
+      "epoch": 0.2508132726089785,
+      "grad_norm": 0.18052315711975098,
+      "learning_rate": 0.0001,
+      "loss": 1.7242,
+      "step": 514
+    },
+    {
+      "epoch": 0.25130123617436567,
+      "grad_norm": 0.17241713404655457,
+      "learning_rate": 0.0001,
+      "loss": 1.6513,
+      "step": 515
+    },
+    {
+      "epoch": 0.2517891997397528,
+      "grad_norm": 0.1861806958913803,
+      "learning_rate": 0.0001,
+      "loss": 1.7189,
+      "step": 516
+    },
+    {
+      "epoch": 0.2522771633051399,
+      "grad_norm": 0.17267678678035736,
+      "learning_rate": 0.0001,
+      "loss": 1.5993,
+      "step": 517
+    },
+    {
+      "epoch": 0.252765126870527,
+      "grad_norm": 0.16948658227920532,
+      "learning_rate": 0.0001,
+      "loss": 1.5733,
+      "step": 518
+    },
+    {
+      "epoch": 0.2532530904359141,
+      "grad_norm": 0.18075625598430634,
+      "learning_rate": 0.0001,
+      "loss": 1.7755,
+      "step": 519
+    },
+    {
+      "epoch": 0.25374105400130126,
+      "grad_norm": 0.17203836143016815,
+      "learning_rate": 0.0001,
+      "loss": 1.6755,
+      "step": 520
+    },
+    {
+      "epoch": 0.25422901756668836,
+      "grad_norm": 0.1631672978401184,
+      "learning_rate": 0.0001,
+      "loss": 1.5949,
+      "step": 521
+    },
+    {
+      "epoch": 0.25471698113207547,
+      "grad_norm": 0.1776244342327118,
+      "learning_rate": 0.0001,
+      "loss": 1.7231,
+      "step": 522
+    },
+    {
+      "epoch": 0.2552049446974626,
+      "grad_norm": 0.18010790646076202,
+      "learning_rate": 0.0001,
+      "loss": 1.7575,
+      "step": 523
+    },
+    {
+      "epoch": 0.2556929082628497,
+      "grad_norm": 0.16827166080474854,
+      "learning_rate": 0.0001,
+      "loss": 1.6907,
+      "step": 524
+    },
+    {
+      "epoch": 0.25618087182823684,
+      "grad_norm": 0.19028151035308838,
+      "learning_rate": 0.0001,
+      "loss": 1.6602,
+      "step": 525
+    },
+    {
+      "epoch": 0.25666883539362395,
+      "grad_norm": 0.17831748723983765,
+      "learning_rate": 0.0001,
+      "loss": 1.7746,
+      "step": 526
+    },
+    {
+      "epoch": 0.25715679895901106,
+      "grad_norm": 0.19768738746643066,
+      "learning_rate": 0.0001,
+      "loss": 1.7111,
+      "step": 527
+    },
+    {
+      "epoch": 0.25764476252439816,
+      "grad_norm": 0.1869453638792038,
+      "learning_rate": 0.0001,
+      "loss": 1.7493,
+      "step": 528
+    },
+    {
+      "epoch": 0.25813272608978527,
+      "grad_norm": 0.17493435740470886,
+      "learning_rate": 0.0001,
+      "loss": 1.6401,
+      "step": 529
+    },
+    {
+      "epoch": 0.25862068965517243,
+      "grad_norm": 0.1741894632577896,
+      "learning_rate": 0.0001,
+      "loss": 1.6737,
+      "step": 530
+    },
+    {
+      "epoch": 0.25910865322055954,
+      "grad_norm": 0.19671699404716492,
+      "learning_rate": 0.0001,
+      "loss": 1.7265,
+      "step": 531
+    },
+    {
+      "epoch": 0.25959661678594664,
+      "grad_norm": 0.1766589730978012,
+      "learning_rate": 0.0001,
+      "loss": 1.6655,
+      "step": 532
+    },
+    {
+      "epoch": 0.26008458035133375,
+      "grad_norm": 0.17494948208332062,
+      "learning_rate": 0.0001,
+      "loss": 1.6571,
+      "step": 533
+    },
+    {
+      "epoch": 0.2605725439167209,
+      "grad_norm": 0.20303772389888763,
+      "learning_rate": 0.0001,
+      "loss": 1.7987,
+      "step": 534
+    },
+    {
+      "epoch": 0.261060507482108,
+      "grad_norm": 0.18097007274627686,
+      "learning_rate": 0.0001,
+      "loss": 1.6341,
+      "step": 535
+    },
+    {
+      "epoch": 0.2615484710474951,
+      "grad_norm": 0.20877449214458466,
+      "learning_rate": 0.0001,
+      "loss": 1.7057,
+      "step": 536
+    },
+    {
+      "epoch": 0.26203643461288223,
+      "grad_norm": 0.19047099351882935,
+      "learning_rate": 0.0001,
+      "loss": 1.7048,
+      "step": 537
+    },
+    {
+      "epoch": 0.26252439817826934,
+      "grad_norm": 0.18251296877861023,
+      "learning_rate": 0.0001,
+      "loss": 1.6979,
+      "step": 538
+    },
+    {
+      "epoch": 0.2630123617436565,
+      "grad_norm": 0.18078570067882538,
+      "learning_rate": 0.0001,
+      "loss": 1.801,
+      "step": 539
+    },
+    {
+      "epoch": 0.2635003253090436,
+      "grad_norm": 0.18725551664829254,
+      "learning_rate": 0.0001,
+      "loss": 1.7638,
+      "step": 540
+    },
+    {
+      "epoch": 0.2639882888744307,
+      "grad_norm": 0.20769141614437103,
+      "learning_rate": 0.0001,
+      "loss": 1.8201,
+      "step": 541
+    },
+    {
+      "epoch": 0.2644762524398178,
+      "grad_norm": 0.16759508848190308,
+      "learning_rate": 0.0001,
+      "loss": 1.6739,
+      "step": 542
+    },
+    {
+      "epoch": 0.2649642160052049,
+      "grad_norm": 0.20297077298164368,
+      "learning_rate": 0.0001,
+      "loss": 1.8241,
+      "step": 543
+    },
+    {
+      "epoch": 0.2654521795705921,
+      "grad_norm": 0.17038699984550476,
+      "learning_rate": 0.0001,
+      "loss": 1.6566,
+      "step": 544
+    },
+    {
+      "epoch": 0.2659401431359792,
+      "grad_norm": 0.17414064705371857,
+      "learning_rate": 0.0001,
+      "loss": 1.5866,
+      "step": 545
+    },
+    {
+      "epoch": 0.2664281067013663,
+      "grad_norm": 0.1856188178062439,
+      "learning_rate": 0.0001,
+      "loss": 1.7166,
+      "step": 546
+    },
+    {
+      "epoch": 0.2669160702667534,
+      "grad_norm": 0.17565833032131195,
+      "learning_rate": 0.0001,
+      "loss": 1.7206,
+      "step": 547
+    },
+    {
+      "epoch": 0.2674040338321405,
+      "grad_norm": 0.18267709016799927,
+      "learning_rate": 0.0001,
+      "loss": 1.6728,
+      "step": 548
+    },
+    {
+      "epoch": 0.26789199739752767,
+      "grad_norm": 0.18981780111789703,
+      "learning_rate": 0.0001,
+      "loss": 1.7425,
+      "step": 549
+    },
+    {
+      "epoch": 0.2683799609629148,
+      "grad_norm": 0.18254795670509338,
+      "learning_rate": 0.0001,
+      "loss": 1.6948,
+      "step": 550
+    },
+    {
+      "epoch": 0.2688679245283019,
+      "grad_norm": 0.18846552073955536,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 551
+    },
+    {
+      "epoch": 0.269355888093689,
+      "grad_norm": 0.1776316910982132,
+      "learning_rate": 0.0001,
+      "loss": 1.618,
+      "step": 552
+    },
+    {
+      "epoch": 0.2698438516590761,
+      "grad_norm": 0.1822226643562317,
+      "learning_rate": 0.0001,
+      "loss": 1.8876,
+      "step": 553
+    },
+    {
+      "epoch": 0.27033181522446326,
+      "grad_norm": 0.1873788982629776,
+      "learning_rate": 0.0001,
+      "loss": 1.7301,
+      "step": 554
+    },
+    {
+      "epoch": 0.27081977878985036,
+      "grad_norm": 0.19234952330589294,
+      "learning_rate": 0.0001,
+      "loss": 1.7235,
+      "step": 555
+    },
+    {
+      "epoch": 0.27130774235523747,
+      "grad_norm": 0.17642012238502502,
+      "learning_rate": 0.0001,
+      "loss": 1.7258,
+      "step": 556
+    },
+    {
+      "epoch": 0.2717957059206246,
+      "grad_norm": 0.21255896985530853,
+      "learning_rate": 0.0001,
+      "loss": 1.6937,
+      "step": 557
+    },
+    {
+      "epoch": 0.27228366948601174,
+      "grad_norm": 0.2181590497493744,
+      "learning_rate": 0.0001,
+      "loss": 1.9076,
+      "step": 558
+    },
+    {
+      "epoch": 0.27277163305139884,
+      "grad_norm": 0.16595962643623352,
+      "learning_rate": 0.0001,
+      "loss": 1.5664,
+      "step": 559
+    },
+    {
+      "epoch": 0.27325959661678595,
+      "grad_norm": 0.1832776963710785,
+      "learning_rate": 0.0001,
+      "loss": 1.658,
+      "step": 560
+    },
+    {
+      "epoch": 0.27374756018217306,
+      "grad_norm": 0.18969666957855225,
+      "learning_rate": 0.0001,
+      "loss": 1.8031,
+      "step": 561
+    },
+    {
+      "epoch": 0.27423552374756016,
+      "grad_norm": 0.1813500076532364,
+      "learning_rate": 0.0001,
+      "loss": 1.7209,
+      "step": 562
+    },
+    {
+      "epoch": 0.2747234873129473,
+      "grad_norm": 0.18055056035518646,
+      "learning_rate": 0.0001,
+      "loss": 1.7658,
+      "step": 563
+    },
+    {
+      "epoch": 0.27521145087833443,
+      "grad_norm": 0.17362233996391296,
+      "learning_rate": 0.0001,
+      "loss": 1.7746,
+      "step": 564
+    },
+    {
+      "epoch": 0.27569941444372154,
+      "grad_norm": 0.19305916130542755,
+      "learning_rate": 0.0001,
+      "loss": 1.9062,
+      "step": 565
+    },
+    {
+      "epoch": 0.27618737800910864,
+      "grad_norm": 0.17458635568618774,
+      "learning_rate": 0.0001,
+      "loss": 1.6339,
+      "step": 566
+    },
+    {
+      "epoch": 0.27667534157449575,
+      "grad_norm": 0.18760624527931213,
+      "learning_rate": 0.0001,
+      "loss": 1.6433,
+      "step": 567
+    },
+    {
+      "epoch": 0.2771633051398829,
+      "grad_norm": 0.17057117819786072,
+      "learning_rate": 0.0001,
+      "loss": 1.6318,
+      "step": 568
+    },
+    {
+      "epoch": 0.27765126870527,
+      "grad_norm": 0.17930074036121368,
+      "learning_rate": 0.0001,
+      "loss": 1.7227,
+      "step": 569
+    },
+    {
+      "epoch": 0.2781392322706571,
+      "grad_norm": 0.17012158036231995,
+      "learning_rate": 0.0001,
+      "loss": 1.6309,
+      "step": 570
+    },
+    {
+      "epoch": 0.27862719583604423,
+      "grad_norm": 0.17562495172023773,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 571
+    },
+    {
+      "epoch": 0.27911515940143133,
+      "grad_norm": 0.18494853377342224,
+      "learning_rate": 0.0001,
+      "loss": 1.8355,
+      "step": 572
+    },
+    {
+      "epoch": 0.2796031229668185,
+      "grad_norm": 0.18261797726154327,
+      "learning_rate": 0.0001,
+      "loss": 1.6015,
+      "step": 573
+    },
+    {
+      "epoch": 0.2800910865322056,
+      "grad_norm": 0.18148979544639587,
+      "learning_rate": 0.0001,
+      "loss": 1.797,
+      "step": 574
+    },
+    {
+      "epoch": 0.2805790500975927,
+      "grad_norm": 0.16941653192043304,
+      "learning_rate": 0.0001,
+      "loss": 1.6382,
+      "step": 575
+    },
+    {
+      "epoch": 0.2810670136629798,
+      "grad_norm": 0.18611697852611542,
+      "learning_rate": 0.0001,
+      "loss": 1.6595,
+      "step": 576
+    },
+    {
+      "epoch": 0.281554977228367,
+      "grad_norm": 0.16945675015449524,
+      "learning_rate": 0.0001,
+      "loss": 1.6678,
+      "step": 577
+    },
+    {
+      "epoch": 0.2820429407937541,
+      "grad_norm": 0.17999336123466492,
+      "learning_rate": 0.0001,
+      "loss": 1.7161,
+      "step": 578
+    },
+    {
+      "epoch": 0.2825309043591412,
+      "grad_norm": 0.185410276055336,
+      "learning_rate": 0.0001,
+      "loss": 1.6731,
+      "step": 579
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 0.1757509708404541,
+      "learning_rate": 0.0001,
+      "loss": 1.7162,
+      "step": 580
+    },
+    {
+      "epoch": 0.2835068314899154,
+      "grad_norm": 0.1721939593553543,
+      "learning_rate": 0.0001,
+      "loss": 1.6374,
+      "step": 581
+    },
+    {
+      "epoch": 0.28399479505530256,
+      "grad_norm": 0.17961697280406952,
+      "learning_rate": 0.0001,
+      "loss": 1.5798,
+      "step": 582
+    },
+    {
+      "epoch": 0.28448275862068967,
+      "grad_norm": 0.18612822890281677,
+      "learning_rate": 0.0001,
+      "loss": 1.7694,
+      "step": 583
+    },
+    {
+      "epoch": 0.2849707221860768,
+      "grad_norm": 0.18089883029460907,
+      "learning_rate": 0.0001,
+      "loss": 1.7426,
+      "step": 584
+    },
+    {
+      "epoch": 0.2854586857514639,
+      "grad_norm": 0.19402338564395905,
+      "learning_rate": 0.0001,
+      "loss": 1.7604,
+      "step": 585
+    },
+    {
+      "epoch": 0.285946649316851,
+      "grad_norm": 0.18208986520767212,
+      "learning_rate": 0.0001,
+      "loss": 1.6998,
+      "step": 586
+    },
+    {
+      "epoch": 0.28643461288223815,
+      "grad_norm": 0.19270221889019012,
+      "learning_rate": 0.0001,
+      "loss": 1.6564,
+      "step": 587
+    },
+    {
+      "epoch": 0.28692257644762525,
+      "grad_norm": 0.17604075372219086,
+      "learning_rate": 0.0001,
+      "loss": 1.653,
+      "step": 588
+    },
+    {
+      "epoch": 0.28741054001301236,
+      "grad_norm": 0.17964652180671692,
+      "learning_rate": 0.0001,
+      "loss": 1.7613,
+      "step": 589
+    },
+    {
+      "epoch": 0.28789850357839947,
+      "grad_norm": 0.18317797780036926,
+      "learning_rate": 0.0001,
+      "loss": 1.6621,
+      "step": 590
+    },
+    {
+      "epoch": 0.2883864671437866,
+      "grad_norm": 0.18271799385547638,
+      "learning_rate": 0.0001,
+      "loss": 1.8067,
+      "step": 591
+    },
+    {
+      "epoch": 0.28887443070917374,
+      "grad_norm": 0.19613641500473022,
+      "learning_rate": 0.0001,
+      "loss": 1.8544,
+      "step": 592
+    },
+    {
+      "epoch": 0.28936239427456084,
+      "grad_norm": 0.19165842235088348,
+      "learning_rate": 0.0001,
+      "loss": 1.8834,
+      "step": 593
+    },
+    {
+      "epoch": 0.28985035783994795,
+      "grad_norm": 0.18238607048988342,
+      "learning_rate": 0.0001,
+      "loss": 1.7776,
+      "step": 594
+    },
+    {
+      "epoch": 0.29033832140533505,
+      "grad_norm": 0.16585291922092438,
+      "learning_rate": 0.0001,
+      "loss": 1.5959,
+      "step": 595
+    },
+    {
+      "epoch": 0.29082628497072216,
+      "grad_norm": 0.1774480640888214,
+      "learning_rate": 0.0001,
+      "loss": 1.6114,
+      "step": 596
+    },
+    {
+      "epoch": 0.2913142485361093,
+      "grad_norm": 0.17970281839370728,
+      "learning_rate": 0.0001,
+      "loss": 1.79,
+      "step": 597
+    },
+    {
+      "epoch": 0.2918022121014964,
+      "grad_norm": 0.18806995451450348,
+      "learning_rate": 0.0001,
+      "loss": 1.7842,
+      "step": 598
+    },
+    {
+      "epoch": 0.29229017566688353,
+      "grad_norm": 0.16845998167991638,
+      "learning_rate": 0.0001,
+      "loss": 1.6788,
+      "step": 599
+    },
+    {
+      "epoch": 0.29277813923227064,
+      "grad_norm": 0.18506960570812225,
+      "learning_rate": 0.0001,
+      "loss": 1.758,
+      "step": 600
+    },
+    {
+      "epoch": 0.2932661027976578,
+      "grad_norm": 0.1771155744791031,
+      "learning_rate": 0.0001,
+      "loss": 1.7259,
+      "step": 601
+    },
+    {
+      "epoch": 0.2937540663630449,
+      "grad_norm": 0.1760523021221161,
+      "learning_rate": 0.0001,
+      "loss": 1.7807,
+      "step": 602
+    },
+    {
+      "epoch": 0.294242029928432,
+      "grad_norm": 0.1765487641096115,
+      "learning_rate": 0.0001,
+      "loss": 1.5886,
+      "step": 603
+    },
+    {
+      "epoch": 0.2947299934938191,
+      "grad_norm": 0.17646710574626923,
+      "learning_rate": 0.0001,
+      "loss": 1.6508,
+      "step": 604
+    },
+    {
+      "epoch": 0.2952179570592062,
+      "grad_norm": 0.18383362889289856,
+      "learning_rate": 0.0001,
+      "loss": 1.7049,
+      "step": 605
+    },
+    {
+      "epoch": 0.2957059206245934,
+      "grad_norm": 0.18808609247207642,
+      "learning_rate": 0.0001,
+      "loss": 1.6948,
+      "step": 606
+    },
+    {
+      "epoch": 0.2961938841899805,
+      "grad_norm": 0.18178711831569672,
+      "learning_rate": 0.0001,
+      "loss": 1.7306,
+      "step": 607
+    },
+    {
+      "epoch": 0.2966818477553676,
+      "grad_norm": 0.18499815464019775,
+      "learning_rate": 0.0001,
+      "loss": 1.6072,
+      "step": 608
+    },
+    {
+      "epoch": 0.2971698113207547,
+      "grad_norm": 0.18511821329593658,
+      "learning_rate": 0.0001,
+      "loss": 1.6383,
+      "step": 609
+    },
+    {
+      "epoch": 0.2976577748861418,
+      "grad_norm": 0.17731331288814545,
+      "learning_rate": 0.0001,
+      "loss": 1.738,
+      "step": 610
+    },
+    {
+      "epoch": 0.298145738451529,
+      "grad_norm": 0.19273065030574799,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 611
+    },
+    {
+      "epoch": 0.2986337020169161,
+      "grad_norm": 0.1858029067516327,
+      "learning_rate": 0.0001,
+      "loss": 1.6565,
+      "step": 612
+    },
+    {
+      "epoch": 0.2991216655823032,
+      "grad_norm": 0.18791264295578003,
+      "learning_rate": 0.0001,
+      "loss": 1.6857,
+      "step": 613
+    },
+    {
+      "epoch": 0.2996096291476903,
+      "grad_norm": 0.19478711485862732,
+      "learning_rate": 0.0001,
+      "loss": 1.6655,
+      "step": 614
+    },
+    {
+      "epoch": 0.3000975927130774,
+      "grad_norm": 0.18538743257522583,
+      "learning_rate": 0.0001,
+      "loss": 1.701,
+      "step": 615
+    },
+    {
+      "epoch": 0.30058555627846456,
+      "grad_norm": 0.1899065524339676,
+      "learning_rate": 0.0001,
+      "loss": 1.7014,
+      "step": 616
+    },
+    {
+      "epoch": 0.30107351984385167,
+      "grad_norm": 0.19550780951976776,
+      "learning_rate": 0.0001,
+      "loss": 1.8021,
+      "step": 617
+    },
+    {
+      "epoch": 0.3015614834092388,
+      "grad_norm": 0.1695028841495514,
+      "learning_rate": 0.0001,
+      "loss": 1.6423,
+      "step": 618
+    },
+    {
+      "epoch": 0.3020494469746259,
+      "grad_norm": 0.18605121970176697,
+      "learning_rate": 0.0001,
+      "loss": 1.7441,
+      "step": 619
+    },
+    {
+      "epoch": 0.302537410540013,
+      "grad_norm": 0.20526890456676483,
+      "learning_rate": 0.0001,
+      "loss": 1.7878,
+      "step": 620
+    },
+    {
+      "epoch": 0.30302537410540015,
+      "grad_norm": 0.17033647000789642,
+      "learning_rate": 0.0001,
+      "loss": 1.688,
+      "step": 621
+    },
+    {
+      "epoch": 0.30351333767078725,
+      "grad_norm": 0.1756584197282791,
+      "learning_rate": 0.0001,
+      "loss": 1.6914,
+      "step": 622
+    },
+    {
+      "epoch": 0.30400130123617436,
+      "grad_norm": 0.18451380729675293,
+      "learning_rate": 0.0001,
+      "loss": 1.6135,
+      "step": 623
+    },
+    {
+      "epoch": 0.30448926480156147,
+      "grad_norm": 0.17828862369060516,
+      "learning_rate": 0.0001,
+      "loss": 1.677,
+      "step": 624
+    },
+    {
+      "epoch": 0.3049772283669486,
+      "grad_norm": 0.17056816816329956,
+      "learning_rate": 0.0001,
+      "loss": 1.647,
+      "step": 625
+    },
+    {
+      "epoch": 0.30546519193233573,
+      "grad_norm": 0.1786261945962906,
+      "learning_rate": 0.0001,
+      "loss": 1.7212,
+      "step": 626
+    },
+    {
+      "epoch": 0.30595315549772284,
+      "grad_norm": 0.1788036823272705,
+      "learning_rate": 0.0001,
+      "loss": 1.6646,
+      "step": 627
+    },
+    {
+      "epoch": 0.30644111906310995,
+      "grad_norm": 0.17864547669887543,
+      "learning_rate": 0.0001,
+      "loss": 1.7123,
+      "step": 628
+    },
+    {
+      "epoch": 0.30692908262849705,
+      "grad_norm": 0.19462743401527405,
+      "learning_rate": 0.0001,
+      "loss": 1.7975,
+      "step": 629
+    },
+    {
+      "epoch": 0.3074170461938842,
+      "grad_norm": 0.17800424993038177,
+      "learning_rate": 0.0001,
+      "loss": 1.5499,
+      "step": 630
+    },
+    {
+      "epoch": 0.3079050097592713,
+      "grad_norm": 0.1856238692998886,
+      "learning_rate": 0.0001,
+      "loss": 1.9104,
+      "step": 631
+    },
+    {
+      "epoch": 0.3083929733246584,
+      "grad_norm": 0.17673279345035553,
+      "learning_rate": 0.0001,
+      "loss": 1.6382,
+      "step": 632
+    },
+    {
+      "epoch": 0.30888093689004553,
+      "grad_norm": 0.18032853305339813,
+      "learning_rate": 0.0001,
+      "loss": 1.7374,
+      "step": 633
+    },
+    {
+      "epoch": 0.30936890045543264,
+      "grad_norm": 0.17968174815177917,
+      "learning_rate": 0.0001,
+      "loss": 1.662,
+      "step": 634
+    },
+    {
+      "epoch": 0.3098568640208198,
+      "grad_norm": 0.1789749562740326,
+      "learning_rate": 0.0001,
+      "loss": 1.6044,
+      "step": 635
+    },
+    {
+      "epoch": 0.3103448275862069,
+      "grad_norm": 0.175074502825737,
+      "learning_rate": 0.0001,
+      "loss": 1.7047,
+      "step": 636
+    },
+    {
+      "epoch": 0.310832791151594,
+      "grad_norm": 0.17318876087665558,
+      "learning_rate": 0.0001,
+      "loss": 1.6148,
+      "step": 637
+    },
+    {
+      "epoch": 0.3113207547169811,
+      "grad_norm": 0.20739412307739258,
+      "learning_rate": 0.0001,
+      "loss": 1.9162,
+      "step": 638
+    },
+    {
+      "epoch": 0.3118087182823682,
+      "grad_norm": 0.1787186861038208,
+      "learning_rate": 0.0001,
+      "loss": 1.6657,
+      "step": 639
+    },
+    {
+      "epoch": 0.3122966818477554,
+      "grad_norm": 0.1855590045452118,
+      "learning_rate": 0.0001,
+      "loss": 1.7058,
+      "step": 640
+    },
+    {
+      "epoch": 0.3127846454131425,
+      "grad_norm": 0.17939618229866028,
+      "learning_rate": 0.0001,
+      "loss": 1.7663,
+      "step": 641
+    },
+    {
+      "epoch": 0.3132726089785296,
+      "grad_norm": 0.17440925538539886,
+      "learning_rate": 0.0001,
+      "loss": 1.6337,
+      "step": 642
+    },
+    {
+      "epoch": 0.3137605725439167,
+      "grad_norm": 0.19695165753364563,
+      "learning_rate": 0.0001,
+      "loss": 1.6048,
+      "step": 643
+    },
+    {
+      "epoch": 0.3142485361093038,
+      "grad_norm": 0.16877804696559906,
+      "learning_rate": 0.0001,
+      "loss": 1.6677,
+      "step": 644
+    },
+    {
+      "epoch": 0.314736499674691,
+      "grad_norm": 0.1742711365222931,
+      "learning_rate": 0.0001,
+      "loss": 1.6459,
+      "step": 645
+    },
+    {
+      "epoch": 0.3152244632400781,
+      "grad_norm": 0.18073154985904694,
+      "learning_rate": 0.0001,
+      "loss": 1.7392,
+      "step": 646
+    },
+    {
+      "epoch": 0.3157124268054652,
+      "grad_norm": 0.1714729368686676,
+      "learning_rate": 0.0001,
+      "loss": 1.6981,
+      "step": 647
+    },
+    {
+      "epoch": 0.3162003903708523,
+      "grad_norm": 0.17316888272762299,
+      "learning_rate": 0.0001,
+      "loss": 1.6746,
+      "step": 648
+    },
+    {
+      "epoch": 0.31668835393623945,
+      "grad_norm": 0.1779533475637436,
+      "learning_rate": 0.0001,
+      "loss": 1.7709,
+      "step": 649
+    },
+    {
+      "epoch": 0.31717631750162656,
+      "grad_norm": 0.1709679216146469,
+      "learning_rate": 0.0001,
+      "loss": 1.5822,
+      "step": 650
+    },
+    {
+      "epoch": 0.31766428106701367,
+      "grad_norm": 0.17804761230945587,
+      "learning_rate": 0.0001,
+      "loss": 1.7638,
+      "step": 651
+    },
+    {
+      "epoch": 0.31815224463240077,
+      "grad_norm": 0.18509989976882935,
+      "learning_rate": 0.0001,
+      "loss": 1.8712,
+      "step": 652
+    },
+    {
+      "epoch": 0.3186402081977879,
+      "grad_norm": 0.1751030832529068,
+      "learning_rate": 0.0001,
+      "loss": 1.7032,
+      "step": 653
+    },
+    {
+      "epoch": 0.31912817176317504,
+      "grad_norm": 0.17232050001621246,
+      "learning_rate": 0.0001,
+      "loss": 1.6331,
+      "step": 654
+    },
+    {
+      "epoch": 0.31961613532856215,
+      "grad_norm": 0.17198053002357483,
+      "learning_rate": 0.0001,
+      "loss": 1.7067,
+      "step": 655
+    },
+    {
+      "epoch": 0.32010409889394925,
+      "grad_norm": 0.1797952950000763,
+      "learning_rate": 0.0001,
+      "loss": 1.687,
+      "step": 656
+    },
+    {
+      "epoch": 0.32059206245933636,
+      "grad_norm": 0.1817045360803604,
+      "learning_rate": 0.0001,
+      "loss": 1.7448,
+      "step": 657
+    },
+    {
+      "epoch": 0.32108002602472346,
+      "grad_norm": 0.1710105687379837,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 658
+    },
+    {
+      "epoch": 0.3215679895901106,
+      "grad_norm": 0.19661752879619598,
+      "learning_rate": 0.0001,
+      "loss": 1.7867,
+      "step": 659
+    },
+    {
+      "epoch": 0.32205595315549773,
+      "grad_norm": 0.1723627746105194,
+      "learning_rate": 0.0001,
+      "loss": 1.5887,
+      "step": 660
+    },
+    {
+      "epoch": 0.32254391672088484,
+      "grad_norm": 0.21364371478557587,
+      "learning_rate": 0.0001,
+      "loss": 1.8418,
+      "step": 661
+    },
+    {
+      "epoch": 0.32303188028627194,
+      "grad_norm": 0.17605622112751007,
+      "learning_rate": 0.0001,
+      "loss": 1.6892,
+      "step": 662
+    },
+    {
+      "epoch": 0.32351984385165905,
+      "grad_norm": 0.17851850390434265,
+      "learning_rate": 0.0001,
+      "loss": 1.7639,
+      "step": 663
+    },
+    {
+      "epoch": 0.3240078074170462,
+      "grad_norm": 0.1816173940896988,
+      "learning_rate": 0.0001,
+      "loss": 1.6567,
+      "step": 664
+    },
+    {
+      "epoch": 0.3244957709824333,
+      "grad_norm": 0.17529702186584473,
+      "learning_rate": 0.0001,
+      "loss": 1.6945,
+      "step": 665
+    },
+    {
+      "epoch": 0.3249837345478204,
+      "grad_norm": 0.16997535526752472,
+      "learning_rate": 0.0001,
+      "loss": 1.6833,
+      "step": 666
+    },
+    {
+      "epoch": 0.32547169811320753,
+      "grad_norm": 0.18423834443092346,
+      "learning_rate": 0.0001,
+      "loss": 1.7486,
+      "step": 667
+    },
+    {
+      "epoch": 0.3259596616785947,
+      "grad_norm": 0.18737761676311493,
+      "learning_rate": 0.0001,
+      "loss": 1.7561,
+      "step": 668
+    },
+    {
+      "epoch": 0.3264476252439818,
+      "grad_norm": 0.17731069028377533,
+      "learning_rate": 0.0001,
+      "loss": 1.5679,
+      "step": 669
+    },
+    {
+      "epoch": 0.3269355888093689,
+      "grad_norm": 0.197565495967865,
+      "learning_rate": 0.0001,
+      "loss": 1.7457,
+      "step": 670
+    },
+    {
+      "epoch": 0.327423552374756,
+      "grad_norm": 0.19319871068000793,
+      "learning_rate": 0.0001,
+      "loss": 1.8458,
+      "step": 671
+    },
+    {
+      "epoch": 0.3279115159401431,
+      "grad_norm": 0.18049995601177216,
+      "learning_rate": 0.0001,
+      "loss": 1.7076,
+      "step": 672
+    },
+    {
+      "epoch": 0.3283994795055303,
+      "grad_norm": 0.18907921016216278,
+      "learning_rate": 0.0001,
+      "loss": 1.7031,
+      "step": 673
+    },
+    {
+      "epoch": 0.3288874430709174,
+      "grad_norm": 0.18252240121364594,
+      "learning_rate": 0.0001,
+      "loss": 1.6304,
+      "step": 674
+    },
+    {
+      "epoch": 0.3293754066363045,
+      "grad_norm": 0.1798553168773651,
+      "learning_rate": 0.0001,
+      "loss": 1.6504,
+      "step": 675
+    },
+    {
+      "epoch": 0.3298633702016916,
+      "grad_norm": 0.1712959110736847,
+      "learning_rate": 0.0001,
+      "loss": 1.6827,
+      "step": 676
+    },
+    {
+      "epoch": 0.3303513337670787,
+      "grad_norm": 0.169499009847641,
+      "learning_rate": 0.0001,
+      "loss": 1.67,
+      "step": 677
+    },
+    {
+      "epoch": 0.33083929733246586,
+      "grad_norm": 0.17921562492847443,
+      "learning_rate": 0.0001,
+      "loss": 1.6913,
+      "step": 678
+    },
+    {
+      "epoch": 0.33132726089785297,
+      "grad_norm": 0.16730189323425293,
+      "learning_rate": 0.0001,
+      "loss": 1.6585,
+      "step": 679
+    },
+    {
+      "epoch": 0.3318152244632401,
+      "grad_norm": 0.1731245219707489,
+      "learning_rate": 0.0001,
+      "loss": 1.6891,
+      "step": 680
+    },
+    {
+      "epoch": 0.3323031880286272,
+      "grad_norm": 0.18989908695220947,
+      "learning_rate": 0.0001,
+      "loss": 1.8335,
+      "step": 681
+    },
+    {
+      "epoch": 0.3327911515940143,
+      "grad_norm": 0.17079797387123108,
+      "learning_rate": 0.0001,
+      "loss": 1.6074,
+      "step": 682
+    },
+    {
+      "epoch": 0.33327911515940145,
+      "grad_norm": 0.1855732947587967,
+      "learning_rate": 0.0001,
+      "loss": 1.8051,
+      "step": 683
+    },
+    {
+      "epoch": 0.33376707872478856,
+      "grad_norm": 0.19362801313400269,
+      "learning_rate": 0.0001,
+      "loss": 1.7934,
+      "step": 684
+    },
+    {
+      "epoch": 0.33425504229017566,
+      "grad_norm": 0.18407447636127472,
+      "learning_rate": 0.0001,
+      "loss": 1.7676,
+      "step": 685
+    },
+    {
+      "epoch": 0.33474300585556277,
+      "grad_norm": 0.17326807975769043,
+      "learning_rate": 0.0001,
+      "loss": 1.6867,
+      "step": 686
+    },
+    {
+      "epoch": 0.3352309694209499,
+      "grad_norm": 0.18629767000675201,
+      "learning_rate": 0.0001,
+      "loss": 1.7577,
+      "step": 687
+    },
+    {
+      "epoch": 0.33571893298633704,
+      "grad_norm": 0.19202108681201935,
+      "learning_rate": 0.0001,
+      "loss": 1.7742,
+      "step": 688
+    },
+    {
+      "epoch": 0.33620689655172414,
+      "grad_norm": 0.1923230141401291,
+      "learning_rate": 0.0001,
+      "loss": 1.7646,
+      "step": 689
+    },
+    {
+      "epoch": 0.33669486011711125,
+      "grad_norm": 0.1855097860097885,
+      "learning_rate": 0.0001,
+      "loss": 1.7189,
+      "step": 690
+    },
+    {
+      "epoch": 0.33718282368249836,
+      "grad_norm": 0.17661595344543457,
+      "learning_rate": 0.0001,
+      "loss": 1.6404,
+      "step": 691
+    },
+    {
+      "epoch": 0.3376707872478855,
+      "grad_norm": 0.19284093379974365,
+      "learning_rate": 0.0001,
+      "loss": 1.7621,
+      "step": 692
+    },
+    {
+      "epoch": 0.3381587508132726,
+      "grad_norm": 0.18006063997745514,
+      "learning_rate": 0.0001,
+      "loss": 1.6163,
+      "step": 693
+    },
+    {
+      "epoch": 0.33864671437865973,
+      "grad_norm": 0.1881456822156906,
+      "learning_rate": 0.0001,
+      "loss": 1.732,
+      "step": 694
+    },
+    {
+      "epoch": 0.33913467794404684,
+      "grad_norm": 0.17196986079216003,
+      "learning_rate": 0.0001,
+      "loss": 1.7099,
+      "step": 695
+    },
+    {
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.186056986451149,
+      "learning_rate": 0.0001,
+      "loss": 1.8247,
+      "step": 696
+    },
+    {
+      "epoch": 0.3401106050748211,
+      "grad_norm": 0.18548524379730225,
+      "learning_rate": 0.0001,
+      "loss": 1.7185,
+      "step": 697
+    },
+    {
+      "epoch": 0.3405985686402082,
+      "grad_norm": 0.182390958070755,
+      "learning_rate": 0.0001,
+      "loss": 1.8278,
+      "step": 698
+    },
+    {
+      "epoch": 0.3410865322055953,
+      "grad_norm": 0.18355803191661835,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 699
+    },
+    {
+      "epoch": 0.3415744957709824,
+      "grad_norm": 0.176362544298172,
+      "learning_rate": 0.0001,
+      "loss": 1.71,
+      "step": 700
+    },
+    {
+      "epoch": 0.34206245933636953,
+      "grad_norm": 0.1753791868686676,
+      "learning_rate": 0.0001,
+      "loss": 1.7079,
+      "step": 701
+    },
+    {
+      "epoch": 0.3425504229017567,
+      "grad_norm": 0.17833958566188812,
+      "learning_rate": 0.0001,
+      "loss": 1.6155,
+      "step": 702
+    },
+    {
+      "epoch": 0.3430383864671438,
+      "grad_norm": 0.18626241385936737,
+      "learning_rate": 0.0001,
+      "loss": 1.8164,
+      "step": 703
+    },
+    {
+      "epoch": 0.3435263500325309,
+      "grad_norm": 0.18040528893470764,
+      "learning_rate": 0.0001,
+      "loss": 1.7061,
+      "step": 704
+    },
+    {
+      "epoch": 0.344014313597918,
+      "grad_norm": 0.18248948454856873,
+      "learning_rate": 0.0001,
+      "loss": 1.7002,
+      "step": 705
+    },
+    {
+      "epoch": 0.3445022771633051,
+      "grad_norm": 0.18155597150325775,
+      "learning_rate": 0.0001,
+      "loss": 1.7623,
+      "step": 706
+    },
+    {
+      "epoch": 0.3449902407286923,
+      "grad_norm": 0.18167854845523834,
+      "learning_rate": 0.0001,
+      "loss": 1.7209,
+      "step": 707
+    },
+    {
+      "epoch": 0.3454782042940794,
+      "grad_norm": 0.18228544294834137,
+      "learning_rate": 0.0001,
+      "loss": 1.7166,
+      "step": 708
+    },
+    {
+      "epoch": 0.3459661678594665,
+      "grad_norm": 0.1872456818819046,
+      "learning_rate": 0.0001,
+      "loss": 1.8073,
+      "step": 709
+    },
+    {
+      "epoch": 0.3464541314248536,
+      "grad_norm": 0.17062440514564514,
+      "learning_rate": 0.0001,
+      "loss": 1.653,
+      "step": 710
+    },
+    {
+      "epoch": 0.3469420949902407,
+      "grad_norm": 0.17459101974964142,
+      "learning_rate": 0.0001,
+      "loss": 1.6982,
+      "step": 711
+    },
+    {
+      "epoch": 0.34743005855562786,
+      "grad_norm": 0.1724562644958496,
+      "learning_rate": 0.0001,
+      "loss": 1.7638,
+      "step": 712
+    },
+    {
+      "epoch": 0.34791802212101497,
+      "grad_norm": 0.16791169345378876,
+      "learning_rate": 0.0001,
+      "loss": 1.5451,
+      "step": 713
+    },
+    {
+      "epoch": 0.3484059856864021,
+      "grad_norm": 0.17250396311283112,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 714
+    },
+    {
+      "epoch": 0.3488939492517892,
+      "grad_norm": 0.17893101274967194,
+      "learning_rate": 0.0001,
+      "loss": 1.7786,
+      "step": 715
+    },
+    {
+      "epoch": 0.34938191281717634,
+      "grad_norm": 0.1739955097436905,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 716
+    },
+    {
+      "epoch": 0.34986987638256345,
+      "grad_norm": 0.183289036154747,
+      "learning_rate": 0.0001,
+      "loss": 1.7026,
+      "step": 717
+    },
+    {
+      "epoch": 0.35035783994795056,
+      "grad_norm": 0.1769326776266098,
+      "learning_rate": 0.0001,
+      "loss": 1.7008,
+      "step": 718
+    },
+    {
+      "epoch": 0.35084580351333766,
+      "grad_norm": 0.1857866495847702,
+      "learning_rate": 0.0001,
+      "loss": 1.6844,
+      "step": 719
+    },
+    {
+      "epoch": 0.35133376707872477,
+      "grad_norm": 0.18651182949543,
+      "learning_rate": 0.0001,
+      "loss": 1.7033,
+      "step": 720
+    },
+    {
+      "epoch": 0.35182173064411193,
+      "grad_norm": 0.18966244161128998,
+      "learning_rate": 0.0001,
+      "loss": 1.7673,
+      "step": 721
+    },
+    {
+      "epoch": 0.35230969420949904,
+      "grad_norm": 0.1810387372970581,
+      "learning_rate": 0.0001,
+      "loss": 1.7161,
+      "step": 722
+    },
+    {
+      "epoch": 0.35279765777488614,
+      "grad_norm": 0.17334793508052826,
+      "learning_rate": 0.0001,
+      "loss": 1.5957,
+      "step": 723
+    },
+    {
+      "epoch": 0.35328562134027325,
+      "grad_norm": 0.18044047057628632,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 724
+    },
+    {
+      "epoch": 0.35377358490566035,
+      "grad_norm": 0.18923179805278778,
+      "learning_rate": 0.0001,
+      "loss": 1.7244,
+      "step": 725
+    },
+    {
+      "epoch": 0.3542615484710475,
+      "grad_norm": 0.18003158271312714,
+      "learning_rate": 0.0001,
+      "loss": 1.7655,
+      "step": 726
+    },
+    {
+      "epoch": 0.3547495120364346,
+      "grad_norm": 0.18161289393901825,
+      "learning_rate": 0.0001,
+      "loss": 1.7199,
+      "step": 727
+    },
+    {
+      "epoch": 0.35523747560182173,
+      "grad_norm": 0.19969268143177032,
+      "learning_rate": 0.0001,
+      "loss": 1.7138,
+      "step": 728
+    },
+    {
+      "epoch": 0.35572543916720883,
+      "grad_norm": 0.1782398670911789,
+      "learning_rate": 0.0001,
+      "loss": 1.6231,
+      "step": 729
+    },
+    {
+      "epoch": 0.35621340273259594,
+      "grad_norm": 0.20619311928749084,
+      "learning_rate": 0.0001,
+      "loss": 1.7745,
+      "step": 730
+    },
+    {
+      "epoch": 0.3567013662979831,
+      "grad_norm": 0.1790829598903656,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 731
+    },
+    {
+      "epoch": 0.3571893298633702,
+      "grad_norm": 0.17978286743164062,
+      "learning_rate": 0.0001,
+      "loss": 1.6495,
+      "step": 732
+    },
+    {
+      "epoch": 0.3576772934287573,
+      "grad_norm": 0.20410868525505066,
+      "learning_rate": 0.0001,
+      "loss": 1.7264,
+      "step": 733
+    },
+    {
+      "epoch": 0.3581652569941444,
+      "grad_norm": 0.18116474151611328,
+      "learning_rate": 0.0001,
+      "loss": 1.7379,
+      "step": 734
+    },
+    {
+      "epoch": 0.3586532205595316,
+      "grad_norm": 0.20212259888648987,
+      "learning_rate": 0.0001,
+      "loss": 1.6964,
+      "step": 735
+    },
+    {
+      "epoch": 0.3591411841249187,
+      "grad_norm": 0.17794452607631683,
+      "learning_rate": 0.0001,
+      "loss": 1.6666,
+      "step": 736
+    },
+    {
+      "epoch": 0.3596291476903058,
+      "grad_norm": 0.17267604172229767,
+      "learning_rate": 0.0001,
+      "loss": 1.5783,
+      "step": 737
+    },
+    {
+      "epoch": 0.3601171112556929,
+      "grad_norm": 0.21285639703273773,
+      "learning_rate": 0.0001,
+      "loss": 1.7575,
+      "step": 738
+    },
+    {
+      "epoch": 0.36060507482108,
+      "grad_norm": 0.1822413057088852,
+      "learning_rate": 0.0001,
+      "loss": 1.7244,
+      "step": 739
+    },
+    {
+      "epoch": 0.36109303838646717,
+      "grad_norm": 0.1909700185060501,
+      "learning_rate": 0.0001,
+      "loss": 1.7614,
+      "step": 740
+    },
+    {
+      "epoch": 0.3615810019518543,
+      "grad_norm": 0.19396358728408813,
+      "learning_rate": 0.0001,
+      "loss": 1.701,
+      "step": 741
+    },
+    {
+      "epoch": 0.3620689655172414,
+      "grad_norm": 0.18860898911952972,
+      "learning_rate": 0.0001,
+      "loss": 1.7215,
+      "step": 742
+    },
+    {
+      "epoch": 0.3625569290826285,
+      "grad_norm": 0.1891864836215973,
+      "learning_rate": 0.0001,
+      "loss": 1.7127,
+      "step": 743
+    },
+    {
+      "epoch": 0.3630448926480156,
+      "grad_norm": 0.18963932991027832,
+      "learning_rate": 0.0001,
+      "loss": 1.6591,
+      "step": 744
+    },
+    {
+      "epoch": 0.36353285621340276,
+      "grad_norm": 0.17823189496994019,
+      "learning_rate": 0.0001,
+      "loss": 1.7356,
+      "step": 745
+    },
+    {
+      "epoch": 0.36402081977878986,
+      "grad_norm": 0.19020548462867737,
+      "learning_rate": 0.0001,
+      "loss": 1.7591,
+      "step": 746
+    },
+    {
+      "epoch": 0.36450878334417697,
+      "grad_norm": 0.1983988732099533,
+      "learning_rate": 0.0001,
+      "loss": 1.6688,
+      "step": 747
+    },
+    {
+      "epoch": 0.3649967469095641,
+      "grad_norm": 0.17455948889255524,
+      "learning_rate": 0.0001,
+      "loss": 1.6981,
+      "step": 748
+    },
+    {
+      "epoch": 0.3654847104749512,
+      "grad_norm": 0.19214113056659698,
+      "learning_rate": 0.0001,
+      "loss": 1.6858,
+      "step": 749
+    },
+    {
+      "epoch": 0.36597267404033834,
+      "grad_norm": 0.19815075397491455,
+      "learning_rate": 0.0001,
+      "loss": 1.7088,
+      "step": 750
+    },
+    {
+      "epoch": 0.36646063760572545,
+      "grad_norm": 0.18052172660827637,
+      "learning_rate": 0.0001,
+      "loss": 1.7046,
+      "step": 751
+    },
+    {
+      "epoch": 0.36694860117111255,
+      "grad_norm": 0.19308723509311676,
+      "learning_rate": 0.0001,
+      "loss": 1.7145,
+      "step": 752
+    },
+    {
+      "epoch": 0.36743656473649966,
+      "grad_norm": 0.20036271214485168,
+      "learning_rate": 0.0001,
+      "loss": 1.6666,
+      "step": 753
+    },
+    {
+      "epoch": 0.36792452830188677,
+      "grad_norm": 0.18619637191295624,
+      "learning_rate": 0.0001,
+      "loss": 1.7144,
+      "step": 754
+    },
+    {
+      "epoch": 0.36841249186727393,
+      "grad_norm": 0.19576376676559448,
+      "learning_rate": 0.0001,
+      "loss": 1.7653,
+      "step": 755
+    },
+    {
+      "epoch": 0.36890045543266103,
+      "grad_norm": 0.18974775075912476,
+      "learning_rate": 0.0001,
+      "loss": 1.7836,
+      "step": 756
+    },
+    {
+      "epoch": 0.36938841899804814,
+      "grad_norm": 0.17752085626125336,
+      "learning_rate": 0.0001,
+      "loss": 1.6496,
+      "step": 757
+    },
+    {
+      "epoch": 0.36987638256343525,
+      "grad_norm": 0.1844092309474945,
+      "learning_rate": 0.0001,
+      "loss": 1.6863,
+      "step": 758
+    },
+    {
+      "epoch": 0.3703643461288224,
+      "grad_norm": 0.18102730810642242,
+      "learning_rate": 0.0001,
+      "loss": 1.5805,
+      "step": 759
+    },
+    {
+      "epoch": 0.3708523096942095,
+      "grad_norm": 0.1773853898048401,
+      "learning_rate": 0.0001,
+      "loss": 1.7169,
+      "step": 760
+    },
+    {
+      "epoch": 0.3713402732595966,
+      "grad_norm": 0.17917506396770477,
+      "learning_rate": 0.0001,
+      "loss": 1.705,
+      "step": 761
+    },
+    {
+      "epoch": 0.3718282368249837,
+      "grad_norm": 0.1869056671857834,
+      "learning_rate": 0.0001,
+      "loss": 1.5653,
+      "step": 762
+    },
+    {
+      "epoch": 0.37231620039037083,
+      "grad_norm": 0.1744174063205719,
+      "learning_rate": 0.0001,
+      "loss": 1.7014,
+      "step": 763
+    },
+    {
+      "epoch": 0.372804163955758,
+      "grad_norm": 0.18072061240673065,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 764
+    },
+    {
+      "epoch": 0.3732921275211451,
+      "grad_norm": 0.17331485450267792,
+      "learning_rate": 0.0001,
+      "loss": 1.6642,
+      "step": 765
+    },
+    {
+      "epoch": 0.3737800910865322,
+      "grad_norm": 0.1780969500541687,
+      "learning_rate": 0.0001,
+      "loss": 1.6563,
+      "step": 766
+    },
+    {
+      "epoch": 0.3742680546519193,
+      "grad_norm": 0.1959829479455948,
+      "learning_rate": 0.0001,
+      "loss": 1.8421,
+      "step": 767
+    },
+    {
+      "epoch": 0.3747560182173064,
+      "grad_norm": 0.18532420694828033,
+      "learning_rate": 0.0001,
+      "loss": 1.7752,
+      "step": 768
+    },
+    {
+      "epoch": 0.3752439817826936,
+      "grad_norm": 0.1861323118209839,
+      "learning_rate": 0.0001,
+      "loss": 1.6672,
+      "step": 769
+    },
+    {
+      "epoch": 0.3757319453480807,
+      "grad_norm": 0.17399415373802185,
+      "learning_rate": 0.0001,
+      "loss": 1.506,
+      "step": 770
+    },
+    {
+      "epoch": 0.3762199089134678,
+      "grad_norm": 0.1861727237701416,
+      "learning_rate": 0.0001,
+      "loss": 1.7164,
+      "step": 771
+    },
+    {
+      "epoch": 0.3767078724788549,
+      "grad_norm": 0.17571841180324554,
+      "learning_rate": 0.0001,
+      "loss": 1.6256,
+      "step": 772
+    },
+    {
+      "epoch": 0.377195836044242,
+      "grad_norm": 0.1843421310186386,
+      "learning_rate": 0.0001,
+      "loss": 1.7273,
+      "step": 773
+    },
+    {
+      "epoch": 0.37768379960962917,
+      "grad_norm": 0.17336313426494598,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 774
+    },
+    {
+      "epoch": 0.3781717631750163,
+      "grad_norm": 0.173604816198349,
+      "learning_rate": 0.0001,
+      "loss": 1.6492,
+      "step": 775
+    },
+    {
+      "epoch": 0.3786597267404034,
+      "grad_norm": 0.19042102992534637,
+      "learning_rate": 0.0001,
+      "loss": 1.7671,
+      "step": 776
+    },
+    {
+      "epoch": 0.3791476903057905,
+      "grad_norm": 0.19237715005874634,
+      "learning_rate": 0.0001,
+      "loss": 1.6948,
+      "step": 777
+    },
+    {
+      "epoch": 0.3796356538711776,
+      "grad_norm": 0.1934320628643036,
+      "learning_rate": 0.0001,
+      "loss": 1.7704,
+      "step": 778
+    },
+    {
+      "epoch": 0.38012361743656475,
+      "grad_norm": 0.18237414956092834,
+      "learning_rate": 0.0001,
+      "loss": 1.7163,
+      "step": 779
+    },
+    {
+      "epoch": 0.38061158100195186,
+      "grad_norm": 0.1750539243221283,
+      "learning_rate": 0.0001,
+      "loss": 1.675,
+      "step": 780
+    },
+    {
+      "epoch": 0.38109954456733897,
+      "grad_norm": 0.18425478041172028,
+      "learning_rate": 0.0001,
+      "loss": 1.803,
+      "step": 781
+    },
+    {
+      "epoch": 0.38158750813272607,
+      "grad_norm": 0.17386333644390106,
+      "learning_rate": 0.0001,
+      "loss": 1.5968,
+      "step": 782
+    },
+    {
+      "epoch": 0.38207547169811323,
+      "grad_norm": 0.1958070695400238,
+      "learning_rate": 0.0001,
+      "loss": 1.7117,
+      "step": 783
+    },
+    {
+      "epoch": 0.38256343526350034,
+      "grad_norm": 0.18313884735107422,
+      "learning_rate": 0.0001,
+      "loss": 1.7634,
+      "step": 784
+    },
+    {
+      "epoch": 0.38305139882888745,
+      "grad_norm": 0.1904529333114624,
+      "learning_rate": 0.0001,
+      "loss": 1.7944,
+      "step": 785
+    },
+    {
+      "epoch": 0.38353936239427455,
+      "grad_norm": 0.18762192130088806,
+      "learning_rate": 0.0001,
+      "loss": 1.6575,
+      "step": 786
+    },
+    {
+      "epoch": 0.38402732595966166,
+      "grad_norm": 0.1828492432832718,
+      "learning_rate": 0.0001,
+      "loss": 1.6451,
+      "step": 787
+    },
+    {
+      "epoch": 0.3845152895250488,
+      "grad_norm": 0.19027890264987946,
+      "learning_rate": 0.0001,
+      "loss": 1.7919,
+      "step": 788
+    },
+    {
+      "epoch": 0.3850032530904359,
+      "grad_norm": 0.17186413705348969,
+      "learning_rate": 0.0001,
+      "loss": 1.6794,
+      "step": 789
+    },
+    {
+      "epoch": 0.38549121665582303,
+      "grad_norm": 0.1878061145544052,
+      "learning_rate": 0.0001,
+      "loss": 1.6987,
+      "step": 790
+    },
+    {
+      "epoch": 0.38597918022121014,
+      "grad_norm": 0.18121576309204102,
+      "learning_rate": 0.0001,
+      "loss": 1.796,
+      "step": 791
+    },
+    {
+      "epoch": 0.38646714378659724,
+      "grad_norm": 0.19097453355789185,
+      "learning_rate": 0.0001,
+      "loss": 1.7155,
+      "step": 792
+    },
+    {
+      "epoch": 0.3869551073519844,
+      "grad_norm": 0.18126630783081055,
+      "learning_rate": 0.0001,
+      "loss": 1.7499,
+      "step": 793
+    },
+    {
+      "epoch": 0.3874430709173715,
+      "grad_norm": 0.1922173947095871,
+      "learning_rate": 0.0001,
+      "loss": 1.7447,
+      "step": 794
+    },
+    {
+      "epoch": 0.3879310344827586,
+      "grad_norm": 0.17474421858787537,
+      "learning_rate": 0.0001,
+      "loss": 1.6234,
+      "step": 795
+    },
+    {
+      "epoch": 0.3884189980481457,
+      "grad_norm": 0.19023337960243225,
+      "learning_rate": 0.0001,
+      "loss": 1.7285,
+      "step": 796
+    },
+    {
+      "epoch": 0.38890696161353283,
+      "grad_norm": 0.17856378853321075,
+      "learning_rate": 0.0001,
+      "loss": 1.598,
+      "step": 797
+    },
+    {
+      "epoch": 0.38939492517892,
+      "grad_norm": 0.17470918595790863,
+      "learning_rate": 0.0001,
+      "loss": 1.7021,
+      "step": 798
+    },
+    {
+      "epoch": 0.3898828887443071,
+      "grad_norm": 0.20127350091934204,
+      "learning_rate": 0.0001,
+      "loss": 1.6433,
+      "step": 799
+    },
+    {
+      "epoch": 0.3903708523096942,
+      "grad_norm": 0.17676322162151337,
+      "learning_rate": 0.0001,
+      "loss": 1.6967,
+      "step": 800
+    },
+    {
+      "epoch": 0.3908588158750813,
+      "grad_norm": 0.17519530653953552,
+      "learning_rate": 0.0001,
+      "loss": 1.7357,
+      "step": 801
+    },
+    {
+      "epoch": 0.3913467794404684,
+      "grad_norm": 0.19061584770679474,
+      "learning_rate": 0.0001,
+      "loss": 1.7182,
+      "step": 802
+    },
+    {
+      "epoch": 0.3918347430058556,
+      "grad_norm": 0.18246081471443176,
+      "learning_rate": 0.0001,
+      "loss": 1.7688,
+      "step": 803
+    },
+    {
+      "epoch": 0.3923227065712427,
+      "grad_norm": 0.20583999156951904,
+      "learning_rate": 0.0001,
+      "loss": 1.8205,
+      "step": 804
+    },
+    {
+      "epoch": 0.3928106701366298,
+      "grad_norm": 0.18392029404640198,
+      "learning_rate": 0.0001,
+      "loss": 1.7499,
+      "step": 805
+    },
+    {
+      "epoch": 0.3932986337020169,
+      "grad_norm": 0.18296070396900177,
+      "learning_rate": 0.0001,
+      "loss": 1.7422,
+      "step": 806
+    },
+    {
+      "epoch": 0.39378659726740406,
+      "grad_norm": 0.176628977060318,
+      "learning_rate": 0.0001,
+      "loss": 1.6818,
+      "step": 807
+    },
+    {
+      "epoch": 0.39427456083279117,
+      "grad_norm": 0.17783887684345245,
+      "learning_rate": 0.0001,
+      "loss": 1.6935,
+      "step": 808
+    },
+    {
+      "epoch": 0.39476252439817827,
+      "grad_norm": 0.18225261569023132,
+      "learning_rate": 0.0001,
+      "loss": 1.7117,
+      "step": 809
+    },
+    {
+      "epoch": 0.3952504879635654,
+      "grad_norm": 0.18413884937763214,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 810
+    },
+    {
+      "epoch": 0.3957384515289525,
+      "grad_norm": 0.18847863376140594,
+      "learning_rate": 0.0001,
+      "loss": 1.6942,
+      "step": 811
+    },
+    {
+      "epoch": 0.39622641509433965,
+      "grad_norm": 0.177464559674263,
+      "learning_rate": 0.0001,
+      "loss": 1.7731,
+      "step": 812
+    },
+    {
+      "epoch": 0.39671437865972675,
+      "grad_norm": 0.18517576158046722,
+      "learning_rate": 0.0001,
+      "loss": 1.709,
+      "step": 813
+    },
+    {
+      "epoch": 0.39720234222511386,
+      "grad_norm": 0.18677739799022675,
+      "learning_rate": 0.0001,
+      "loss": 1.709,
+      "step": 814
+    },
+    {
+      "epoch": 0.39769030579050096,
+      "grad_norm": 0.1786472350358963,
+      "learning_rate": 0.0001,
+      "loss": 1.6966,
+      "step": 815
+    },
+    {
+      "epoch": 0.39817826935588807,
+      "grad_norm": 0.18321356177330017,
+      "learning_rate": 0.0001,
+      "loss": 1.6611,
+      "step": 816
+    },
+    {
+      "epoch": 0.39866623292127523,
+      "grad_norm": 0.19883863627910614,
+      "learning_rate": 0.0001,
+      "loss": 1.7824,
+      "step": 817
+    },
+    {
+      "epoch": 0.39915419648666234,
+      "grad_norm": 0.18374767899513245,
+      "learning_rate": 0.0001,
+      "loss": 1.8102,
+      "step": 818
+    },
+    {
+      "epoch": 0.39964216005204944,
+      "grad_norm": 0.1768617182970047,
+      "learning_rate": 0.0001,
+      "loss": 1.6278,
+      "step": 819
+    },
+    {
+      "epoch": 0.40013012361743655,
+      "grad_norm": 0.17839239537715912,
+      "learning_rate": 0.0001,
+      "loss": 1.5887,
+      "step": 820
+    },
+    {
+      "epoch": 0.40061808718282366,
+      "grad_norm": 0.18420036137104034,
+      "learning_rate": 0.0001,
+      "loss": 1.7334,
+      "step": 821
+    },
+    {
+      "epoch": 0.4011060507482108,
+      "grad_norm": 0.18662692606449127,
+      "learning_rate": 0.0001,
+      "loss": 1.7035,
+      "step": 822
+    },
+    {
+      "epoch": 0.4015940143135979,
+      "grad_norm": 0.1809212863445282,
+      "learning_rate": 0.0001,
+      "loss": 1.6425,
+      "step": 823
+    },
+    {
+      "epoch": 0.40208197787898503,
+      "grad_norm": 0.18343691527843475,
+      "learning_rate": 0.0001,
+      "loss": 1.6915,
+      "step": 824
+    },
+    {
+      "epoch": 0.40256994144437214,
+      "grad_norm": 0.19546520709991455,
+      "learning_rate": 0.0001,
+      "loss": 1.5398,
+      "step": 825
+    },
+    {
+      "epoch": 0.4030579050097593,
+      "grad_norm": 0.18498557806015015,
+      "learning_rate": 0.0001,
+      "loss": 1.76,
+      "step": 826
+    },
+    {
+      "epoch": 0.4035458685751464,
+      "grad_norm": 0.1787293255329132,
+      "learning_rate": 0.0001,
+      "loss": 1.7072,
+      "step": 827
+    },
+    {
+      "epoch": 0.4040338321405335,
+      "grad_norm": 0.18626105785369873,
+      "learning_rate": 0.0001,
+      "loss": 1.6154,
+      "step": 828
+    },
+    {
+      "epoch": 0.4045217957059206,
+      "grad_norm": 0.18181754648685455,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 829
+    },
+    {
+      "epoch": 0.4050097592713077,
+      "grad_norm": 0.1738763153553009,
+      "learning_rate": 0.0001,
+      "loss": 1.6003,
+      "step": 830
+    },
+    {
+      "epoch": 0.4054977228366949,
+      "grad_norm": 0.19205868244171143,
+      "learning_rate": 0.0001,
+      "loss": 1.6516,
+      "step": 831
+    },
+    {
+      "epoch": 0.405985686402082,
+      "grad_norm": 0.17389516532421112,
+      "learning_rate": 0.0001,
+      "loss": 1.6675,
+      "step": 832
+    },
+    {
+      "epoch": 0.4064736499674691,
+      "grad_norm": 0.17901460826396942,
+      "learning_rate": 0.0001,
+      "loss": 1.7835,
+      "step": 833
+    },
+    {
+      "epoch": 0.4069616135328562,
+      "grad_norm": 0.16918572783470154,
+      "learning_rate": 0.0001,
+      "loss": 1.5688,
+      "step": 834
+    },
+    {
+      "epoch": 0.4074495770982433,
+      "grad_norm": 0.17327755689620972,
+      "learning_rate": 0.0001,
+      "loss": 1.612,
+      "step": 835
+    },
+    {
+      "epoch": 0.40793754066363047,
+      "grad_norm": 0.17260931432247162,
+      "learning_rate": 0.0001,
+      "loss": 1.5631,
+      "step": 836
+    },
+    {
+      "epoch": 0.4084255042290176,
+      "grad_norm": 0.18616695702075958,
+      "learning_rate": 0.0001,
+      "loss": 1.8026,
+      "step": 837
+    },
+    {
+      "epoch": 0.4089134677944047,
+      "grad_norm": 0.1833159476518631,
+      "learning_rate": 0.0001,
+      "loss": 1.7407,
+      "step": 838
+    },
+    {
+      "epoch": 0.4094014313597918,
+      "grad_norm": 0.17563556134700775,
+      "learning_rate": 0.0001,
+      "loss": 1.6497,
+      "step": 839
+    },
+    {
+      "epoch": 0.4098893949251789,
+      "grad_norm": 0.1728363335132599,
+      "learning_rate": 0.0001,
+      "loss": 1.7369,
+      "step": 840
+    },
+    {
+      "epoch": 0.41037735849056606,
+      "grad_norm": 0.16742554306983948,
+      "learning_rate": 0.0001,
+      "loss": 1.5323,
+      "step": 841
+    },
+    {
+      "epoch": 0.41086532205595316,
+      "grad_norm": 0.18149816989898682,
+      "learning_rate": 0.0001,
+      "loss": 1.6658,
+      "step": 842
+    },
+    {
+      "epoch": 0.41135328562134027,
+      "grad_norm": 0.1730806678533554,
+      "learning_rate": 0.0001,
+      "loss": 1.6736,
+      "step": 843
+    },
+    {
+      "epoch": 0.4118412491867274,
+      "grad_norm": 0.19350793957710266,
+      "learning_rate": 0.0001,
+      "loss": 1.7305,
+      "step": 844
+    },
+    {
+      "epoch": 0.4123292127521145,
+      "grad_norm": 0.17669609189033508,
+      "learning_rate": 0.0001,
+      "loss": 1.7208,
+      "step": 845
+    },
+    {
+      "epoch": 0.41281717631750164,
+      "grad_norm": 0.18896430730819702,
+      "learning_rate": 0.0001,
+      "loss": 1.7677,
+      "step": 846
+    },
+    {
+      "epoch": 0.41330513988288875,
+      "grad_norm": 0.18296490609645844,
+      "learning_rate": 0.0001,
+      "loss": 1.7551,
+      "step": 847
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 0.18311992287635803,
+      "learning_rate": 0.0001,
+      "loss": 1.6724,
+      "step": 848
+    },
+    {
+      "epoch": 0.41428106701366296,
+      "grad_norm": 0.1732887476682663,
+      "learning_rate": 0.0001,
+      "loss": 1.6779,
+      "step": 849
+    },
+    {
+      "epoch": 0.4147690305790501,
+      "grad_norm": 0.18442484736442566,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 850
+    },
+    {
+      "epoch": 0.41525699414443723,
+      "grad_norm": 0.18358947336673737,
+      "learning_rate": 0.0001,
+      "loss": 1.7059,
+      "step": 851
+    },
+    {
+      "epoch": 0.41574495770982434,
+      "grad_norm": 0.17849397659301758,
+      "learning_rate": 0.0001,
+      "loss": 1.6633,
+      "step": 852
+    },
+    {
+      "epoch": 0.41623292127521144,
+      "grad_norm": 0.17558790743350983,
+      "learning_rate": 0.0001,
+      "loss": 1.7351,
+      "step": 853
+    },
+    {
+      "epoch": 0.41672088484059855,
+      "grad_norm": 0.18554963171482086,
+      "learning_rate": 0.0001,
+      "loss": 1.722,
+      "step": 854
+    },
+    {
+      "epoch": 0.4172088484059857,
+      "grad_norm": 0.17529337108135223,
+      "learning_rate": 0.0001,
+      "loss": 1.7565,
+      "step": 855
+    },
+    {
+      "epoch": 0.4176968119713728,
+      "grad_norm": 0.1806408166885376,
+      "learning_rate": 0.0001,
+      "loss": 1.6164,
+      "step": 856
+    },
+    {
+      "epoch": 0.4181847755367599,
+      "grad_norm": 0.17640672624111176,
+      "learning_rate": 0.0001,
+      "loss": 1.6622,
+      "step": 857
+    },
+    {
+      "epoch": 0.41867273910214703,
+      "grad_norm": 0.18511973321437836,
+      "learning_rate": 0.0001,
+      "loss": 1.7708,
+      "step": 858
+    },
+    {
+      "epoch": 0.41916070266753414,
+      "grad_norm": 0.17402327060699463,
+      "learning_rate": 0.0001,
+      "loss": 1.5703,
+      "step": 859
+    },
+    {
+      "epoch": 0.4196486662329213,
+      "grad_norm": 0.1716722548007965,
+      "learning_rate": 0.0001,
+      "loss": 1.6326,
+      "step": 860
+    },
+    {
+      "epoch": 0.4201366297983084,
+      "grad_norm": 0.18517763912677765,
+      "learning_rate": 0.0001,
+      "loss": 1.638,
+      "step": 861
+    },
+    {
+      "epoch": 0.4206245933636955,
+      "grad_norm": 0.18149396777153015,
+      "learning_rate": 0.0001,
+      "loss": 1.772,
+      "step": 862
+    },
+    {
+      "epoch": 0.4211125569290826,
+      "grad_norm": 0.1842370480298996,
+      "learning_rate": 0.0001,
+      "loss": 1.7326,
+      "step": 863
+    },
+    {
+      "epoch": 0.4216005204944697,
+      "grad_norm": 0.1832754909992218,
+      "learning_rate": 0.0001,
+      "loss": 1.571,
+      "step": 864
+    },
+    {
+      "epoch": 0.4220884840598569,
+      "grad_norm": 0.18610063195228577,
+      "learning_rate": 0.0001,
+      "loss": 1.6853,
+      "step": 865
+    },
+    {
+      "epoch": 0.422576447625244,
+      "grad_norm": 0.18227741122245789,
+      "learning_rate": 0.0001,
+      "loss": 1.7299,
+      "step": 866
+    },
+    {
+      "epoch": 0.4230644111906311,
+      "grad_norm": 0.1710875779390335,
+      "learning_rate": 0.0001,
+      "loss": 1.6311,
+      "step": 867
+    },
+    {
+      "epoch": 0.4235523747560182,
+      "grad_norm": 0.1772422045469284,
+      "learning_rate": 0.0001,
+      "loss": 1.6997,
+      "step": 868
+    },
+    {
+      "epoch": 0.4240403383214053,
+      "grad_norm": 0.18706001341342926,
+      "learning_rate": 0.0001,
+      "loss": 1.7453,
+      "step": 869
+    },
+    {
+      "epoch": 0.42452830188679247,
+      "grad_norm": 0.18400168418884277,
+      "learning_rate": 0.0001,
+      "loss": 1.7748,
+      "step": 870
+    },
+    {
+      "epoch": 0.4250162654521796,
+      "grad_norm": 0.1813107579946518,
+      "learning_rate": 0.0001,
+      "loss": 1.6386,
+      "step": 871
+    },
+    {
+      "epoch": 0.4255042290175667,
+      "grad_norm": 0.18432138860225677,
+      "learning_rate": 0.0001,
+      "loss": 1.6548,
+      "step": 872
+    },
+    {
+      "epoch": 0.4259921925829538,
+      "grad_norm": 0.1701667755842209,
+      "learning_rate": 0.0001,
+      "loss": 1.7228,
+      "step": 873
+    },
+    {
+      "epoch": 0.42648015614834095,
+      "grad_norm": 0.17490911483764648,
+      "learning_rate": 0.0001,
+      "loss": 1.6574,
+      "step": 874
+    },
+    {
+      "epoch": 0.42696811971372806,
+      "grad_norm": 0.1863052397966385,
+      "learning_rate": 0.0001,
+      "loss": 1.6902,
+      "step": 875
+    },
+    {
+      "epoch": 0.42745608327911516,
+      "grad_norm": 0.17869678139686584,
+      "learning_rate": 0.0001,
+      "loss": 1.7961,
+      "step": 876
+    },
+    {
+      "epoch": 0.42794404684450227,
+      "grad_norm": 0.17393270134925842,
+      "learning_rate": 0.0001,
+      "loss": 1.6968,
+      "step": 877
+    },
+    {
+      "epoch": 0.4284320104098894,
+      "grad_norm": 0.1801164150238037,
+      "learning_rate": 0.0001,
+      "loss": 1.8419,
+      "step": 878
+    },
+    {
+      "epoch": 0.42891997397527654,
+      "grad_norm": 0.17271965742111206,
+      "learning_rate": 0.0001,
+      "loss": 1.6948,
+      "step": 879
+    },
+    {
+      "epoch": 0.42940793754066364,
+      "grad_norm": 0.18875744938850403,
+      "learning_rate": 0.0001,
+      "loss": 1.7529,
+      "step": 880
+    },
+    {
+      "epoch": 0.42989590110605075,
+      "grad_norm": 0.18350331485271454,
+      "learning_rate": 0.0001,
+      "loss": 1.7162,
+      "step": 881
+    },
+    {
+      "epoch": 0.43038386467143785,
+      "grad_norm": 0.18316605687141418,
+      "learning_rate": 0.0001,
+      "loss": 1.7071,
+      "step": 882
+    },
+    {
+      "epoch": 0.43087182823682496,
+      "grad_norm": 0.17159631848335266,
+      "learning_rate": 0.0001,
+      "loss": 1.5494,
+      "step": 883
+    },
+    {
+      "epoch": 0.4313597918022121,
+      "grad_norm": 0.1835523098707199,
+      "learning_rate": 0.0001,
+      "loss": 1.7773,
+      "step": 884
+    },
+    {
+      "epoch": 0.43184775536759923,
+      "grad_norm": 0.18305568397045135,
+      "learning_rate": 0.0001,
+      "loss": 1.6616,
+      "step": 885
+    },
+    {
+      "epoch": 0.43233571893298633,
+      "grad_norm": 0.18325333297252655,
+      "learning_rate": 0.0001,
+      "loss": 1.71,
+      "step": 886
+    },
+    {
+      "epoch": 0.43282368249837344,
+      "grad_norm": 0.16807565093040466,
+      "learning_rate": 0.0001,
+      "loss": 1.5946,
+      "step": 887
+    },
+    {
+      "epoch": 0.43331164606376055,
+      "grad_norm": 0.17560525238513947,
+      "learning_rate": 0.0001,
+      "loss": 1.573,
+      "step": 888
+    },
+    {
+      "epoch": 0.4337996096291477,
+      "grad_norm": 0.1823277622461319,
+      "learning_rate": 0.0001,
+      "loss": 1.616,
+      "step": 889
+    },
+    {
+      "epoch": 0.4342875731945348,
+      "grad_norm": 0.17946025729179382,
+      "learning_rate": 0.0001,
+      "loss": 1.5907,
+      "step": 890
+    },
+    {
+      "epoch": 0.4347755367599219,
+      "grad_norm": 0.18940189480781555,
+      "learning_rate": 0.0001,
+      "loss": 1.6697,
+      "step": 891
+    },
+    {
+      "epoch": 0.435263500325309,
+      "grad_norm": 0.17899388074874878,
+      "learning_rate": 0.0001,
+      "loss": 1.6849,
+      "step": 892
+    },
+    {
+      "epoch": 0.4357514638906962,
+      "grad_norm": 0.1885358840227127,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 893
+    },
+    {
+      "epoch": 0.4362394274560833,
+      "grad_norm": 0.1721390187740326,
+      "learning_rate": 0.0001,
+      "loss": 1.6514,
+      "step": 894
+    },
+    {
+      "epoch": 0.4367273910214704,
+      "grad_norm": 0.19019658863544464,
+      "learning_rate": 0.0001,
+      "loss": 1.7234,
+      "step": 895
+    },
+    {
+      "epoch": 0.4372153545868575,
+      "grad_norm": 0.17101971805095673,
+      "learning_rate": 0.0001,
+      "loss": 1.6003,
+      "step": 896
+    },
+    {
+      "epoch": 0.4377033181522446,
+      "grad_norm": 0.192877396941185,
+      "learning_rate": 0.0001,
+      "loss": 1.8151,
+      "step": 897
+    },
+    {
+      "epoch": 0.4381912817176318,
+      "grad_norm": 0.17775356769561768,
+      "learning_rate": 0.0001,
+      "loss": 1.5926,
+      "step": 898
+    },
+    {
+      "epoch": 0.4386792452830189,
+      "grad_norm": 0.19545124471187592,
+      "learning_rate": 0.0001,
+      "loss": 1.7123,
+      "step": 899
+    },
+    {
+      "epoch": 0.439167208848406,
+      "grad_norm": 0.17418169975280762,
+      "learning_rate": 0.0001,
+      "loss": 1.6774,
+      "step": 900
+    },
+    {
+      "epoch": 0.4396551724137931,
+      "grad_norm": 0.19206389784812927,
+      "learning_rate": 0.0001,
+      "loss": 1.7278,
+      "step": 901
+    },
+    {
+      "epoch": 0.4401431359791802,
+      "grad_norm": 0.18674510717391968,
+      "learning_rate": 0.0001,
+      "loss": 1.6049,
+      "step": 902
+    },
+    {
+      "epoch": 0.44063109954456736,
+      "grad_norm": 0.18307790160179138,
+      "learning_rate": 0.0001,
+      "loss": 1.6985,
+      "step": 903
+    },
+    {
+      "epoch": 0.44111906310995447,
+      "grad_norm": 0.1894843429327011,
+      "learning_rate": 0.0001,
+      "loss": 1.676,
+      "step": 904
+    },
+    {
+      "epoch": 0.4416070266753416,
+      "grad_norm": 0.17619220912456512,
+      "learning_rate": 0.0001,
+      "loss": 1.6807,
+      "step": 905
+    },
+    {
+      "epoch": 0.4420949902407287,
+      "grad_norm": 0.1805913895368576,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 906
+    },
+    {
+      "epoch": 0.4425829538061158,
+      "grad_norm": 0.17293816804885864,
+      "learning_rate": 0.0001,
+      "loss": 1.597,
+      "step": 907
+    },
+    {
+      "epoch": 0.44307091737150295,
+      "grad_norm": 0.17609193921089172,
+      "learning_rate": 0.0001,
+      "loss": 1.6562,
+      "step": 908
+    },
+    {
+      "epoch": 0.44355888093689005,
+      "grad_norm": 0.17432111501693726,
+      "learning_rate": 0.0001,
+      "loss": 1.594,
+      "step": 909
+    },
+    {
+      "epoch": 0.44404684450227716,
+      "grad_norm": 0.17889589071273804,
+      "learning_rate": 0.0001,
+      "loss": 1.8029,
+      "step": 910
+    },
+    {
+      "epoch": 0.44453480806766427,
+      "grad_norm": 0.17299845814704895,
+      "learning_rate": 0.0001,
+      "loss": 1.6116,
+      "step": 911
+    },
+    {
+      "epoch": 0.4450227716330514,
+      "grad_norm": 0.17839674651622772,
+      "learning_rate": 0.0001,
+      "loss": 1.7055,
+      "step": 912
+    },
+    {
+      "epoch": 0.44551073519843853,
+      "grad_norm": 0.1751437783241272,
+      "learning_rate": 0.0001,
+      "loss": 1.6218,
+      "step": 913
+    },
+    {
+      "epoch": 0.44599869876382564,
+      "grad_norm": 0.1901925653219223,
+      "learning_rate": 0.0001,
+      "loss": 1.6578,
+      "step": 914
+    },
+    {
+      "epoch": 0.44648666232921275,
+      "grad_norm": 0.17236626148223877,
+      "learning_rate": 0.0001,
+      "loss": 1.6951,
+      "step": 915
+    },
+    {
+      "epoch": 0.44697462589459985,
+      "grad_norm": 0.17387427389621735,
+      "learning_rate": 0.0001,
+      "loss": 1.5922,
+      "step": 916
+    },
+    {
+      "epoch": 0.447462589459987,
+      "grad_norm": 0.1684548258781433,
+      "learning_rate": 0.0001,
+      "loss": 1.5566,
+      "step": 917
+    },
+    {
+      "epoch": 0.4479505530253741,
+      "grad_norm": 0.18070632219314575,
+      "learning_rate": 0.0001,
+      "loss": 1.6904,
+      "step": 918
+    },
+    {
+      "epoch": 0.4484385165907612,
+      "grad_norm": 0.1905713975429535,
+      "learning_rate": 0.0001,
+      "loss": 1.8206,
+      "step": 919
+    },
+    {
+      "epoch": 0.44892648015614833,
+      "grad_norm": 0.1828422248363495,
+      "learning_rate": 0.0001,
+      "loss": 1.7974,
+      "step": 920
+    },
+    {
+      "epoch": 0.44941444372153544,
+      "grad_norm": 0.17595981061458588,
+      "learning_rate": 0.0001,
+      "loss": 1.7308,
+      "step": 921
+    },
+    {
+      "epoch": 0.4499024072869226,
+      "grad_norm": 0.18210361897945404,
+      "learning_rate": 0.0001,
+      "loss": 1.6915,
+      "step": 922
+    },
+    {
+      "epoch": 0.4503903708523097,
+      "grad_norm": 0.18826089799404144,
+      "learning_rate": 0.0001,
+      "loss": 1.7588,
+      "step": 923
+    },
+    {
+      "epoch": 0.4508783344176968,
+      "grad_norm": 0.17665328085422516,
+      "learning_rate": 0.0001,
+      "loss": 1.6797,
+      "step": 924
+    },
+    {
+      "epoch": 0.4513662979830839,
+      "grad_norm": 0.17838731408119202,
+      "learning_rate": 0.0001,
+      "loss": 1.6644,
+      "step": 925
+    },
+    {
+      "epoch": 0.451854261548471,
+      "grad_norm": 0.18045654892921448,
+      "learning_rate": 0.0001,
+      "loss": 1.689,
+      "step": 926
+    },
+    {
+      "epoch": 0.4523422251138582,
+      "grad_norm": 0.18226969242095947,
+      "learning_rate": 0.0001,
+      "loss": 1.8157,
+      "step": 927
+    },
+    {
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.17917855083942413,
+      "learning_rate": 0.0001,
+      "loss": 1.7772,
+      "step": 928
+    },
+    {
+      "epoch": 0.4533181522446324,
+      "grad_norm": 0.1778966784477234,
+      "learning_rate": 0.0001,
+      "loss": 1.6912,
+      "step": 929
+    },
+    {
+      "epoch": 0.4538061158100195,
+      "grad_norm": 0.18105091154575348,
+      "learning_rate": 0.0001,
+      "loss": 1.7072,
+      "step": 930
+    },
+    {
+      "epoch": 0.4542940793754066,
+      "grad_norm": 0.17502936720848083,
+      "learning_rate": 0.0001,
+      "loss": 1.6462,
+      "step": 931
+    },
+    {
+      "epoch": 0.4547820429407938,
+      "grad_norm": 0.1830134093761444,
+      "learning_rate": 0.0001,
+      "loss": 1.6876,
+      "step": 932
+    },
+    {
+      "epoch": 0.4552700065061809,
+      "grad_norm": 0.18607327342033386,
+      "learning_rate": 0.0001,
+      "loss": 1.7082,
+      "step": 933
+    },
+    {
+      "epoch": 0.455757970071568,
+      "grad_norm": 0.18888945877552032,
+      "learning_rate": 0.0001,
+      "loss": 1.7509,
+      "step": 934
+    },
+    {
+      "epoch": 0.4562459336369551,
+      "grad_norm": 0.1867811232805252,
+      "learning_rate": 0.0001,
+      "loss": 1.7233,
+      "step": 935
+    },
+    {
+      "epoch": 0.4567338972023422,
+      "grad_norm": 0.1898915022611618,
+      "learning_rate": 0.0001,
+      "loss": 1.6237,
+      "step": 936
+    },
+    {
+      "epoch": 0.45722186076772936,
+      "grad_norm": 0.1797095388174057,
+      "learning_rate": 0.0001,
+      "loss": 1.7404,
+      "step": 937
+    },
+    {
+      "epoch": 0.45770982433311647,
+      "grad_norm": 0.17534306645393372,
+      "learning_rate": 0.0001,
+      "loss": 1.6726,
+      "step": 938
+    },
+    {
+      "epoch": 0.4581977878985036,
+      "grad_norm": 0.19073282182216644,
+      "learning_rate": 0.0001,
+      "loss": 1.8081,
+      "step": 939
+    },
+    {
+      "epoch": 0.4586857514638907,
+      "grad_norm": 0.1878473460674286,
+      "learning_rate": 0.0001,
+      "loss": 1.6855,
+      "step": 940
+    },
+    {
+      "epoch": 0.45917371502927784,
+      "grad_norm": 0.18376657366752625,
+      "learning_rate": 0.0001,
+      "loss": 1.6833,
+      "step": 941
+    },
+    {
+      "epoch": 0.45966167859466495,
+      "grad_norm": 0.18948735296726227,
+      "learning_rate": 0.0001,
+      "loss": 1.7525,
+      "step": 942
+    },
+    {
+      "epoch": 0.46014964216005205,
+      "grad_norm": 0.18738175928592682,
+      "learning_rate": 0.0001,
+      "loss": 1.752,
+      "step": 943
+    },
+    {
+      "epoch": 0.46063760572543916,
+      "grad_norm": 0.1765458881855011,
+      "learning_rate": 0.0001,
+      "loss": 1.696,
+      "step": 944
+    },
+    {
+      "epoch": 0.46112556929082626,
+      "grad_norm": 0.18650664389133453,
+      "learning_rate": 0.0001,
+      "loss": 1.7409,
+      "step": 945
+    },
+    {
+      "epoch": 0.4616135328562134,
+      "grad_norm": 0.1759469360113144,
+      "learning_rate": 0.0001,
+      "loss": 1.6119,
+      "step": 946
+    },
+    {
+      "epoch": 0.46210149642160053,
+      "grad_norm": 0.18343883752822876,
+      "learning_rate": 0.0001,
+      "loss": 1.7091,
+      "step": 947
+    },
+    {
+      "epoch": 0.46258945998698764,
+      "grad_norm": 0.1964959353208542,
+      "learning_rate": 0.0001,
+      "loss": 1.7388,
+      "step": 948
+    },
+    {
+      "epoch": 0.46307742355237475,
+      "grad_norm": 0.18265226483345032,
+      "learning_rate": 0.0001,
+      "loss": 1.7036,
+      "step": 949
+    },
+    {
+      "epoch": 0.46356538711776185,
+      "grad_norm": 0.18132254481315613,
+      "learning_rate": 0.0001,
+      "loss": 1.688,
+      "step": 950
+    },
+    {
+      "epoch": 0.464053350683149,
+      "grad_norm": 0.18742497265338898,
+      "learning_rate": 0.0001,
+      "loss": 1.6212,
+      "step": 951
+    },
+    {
+      "epoch": 0.4645413142485361,
+      "grad_norm": 0.1776818335056305,
+      "learning_rate": 0.0001,
+      "loss": 1.5739,
+      "step": 952
+    },
+    {
+      "epoch": 0.4650292778139232,
+      "grad_norm": 0.193990558385849,
+      "learning_rate": 0.0001,
+      "loss": 1.6852,
+      "step": 953
+    },
+    {
+      "epoch": 0.46551724137931033,
+      "grad_norm": 0.1853352040052414,
+      "learning_rate": 0.0001,
+      "loss": 1.6057,
+      "step": 954
+    },
+    {
+      "epoch": 0.46600520494469744,
+      "grad_norm": 0.2000368982553482,
+      "learning_rate": 0.0001,
+      "loss": 1.7329,
+      "step": 955
+    },
+    {
+      "epoch": 0.4664931685100846,
+      "grad_norm": 0.20909981429576874,
+      "learning_rate": 0.0001,
+      "loss": 1.687,
+      "step": 956
+    },
+    {
+      "epoch": 0.4669811320754717,
+      "grad_norm": 0.21065653860569,
+      "learning_rate": 0.0001,
+      "loss": 1.7239,
+      "step": 957
+    },
+    {
+      "epoch": 0.4674690956408588,
+      "grad_norm": 0.1819789707660675,
+      "learning_rate": 0.0001,
+      "loss": 1.7258,
+      "step": 958
+    },
+    {
+      "epoch": 0.4679570592062459,
+      "grad_norm": 0.20444951951503754,
+      "learning_rate": 0.0001,
+      "loss": 1.679,
+      "step": 959
+    },
+    {
+      "epoch": 0.468445022771633,
+      "grad_norm": 0.19722609221935272,
+      "learning_rate": 0.0001,
+      "loss": 1.6114,
+      "step": 960
+    },
+    {
+      "epoch": 0.4689329863370202,
+      "grad_norm": 0.18290160596370697,
+      "learning_rate": 0.0001,
+      "loss": 1.7676,
+      "step": 961
+    },
+    {
+      "epoch": 0.4694209499024073,
+      "grad_norm": 0.20910906791687012,
+      "learning_rate": 0.0001,
+      "loss": 1.6688,
+      "step": 962
+    },
+    {
+      "epoch": 0.4699089134677944,
+      "grad_norm": 0.2053229659795761,
+      "learning_rate": 0.0001,
+      "loss": 1.7208,
+      "step": 963
+    },
+    {
+      "epoch": 0.4703968770331815,
+      "grad_norm": 0.18317236006259918,
+      "learning_rate": 0.0001,
+      "loss": 1.6808,
+      "step": 964
+    },
+    {
+      "epoch": 0.47088484059856867,
+      "grad_norm": 0.20331262052059174,
+      "learning_rate": 0.0001,
+      "loss": 1.7621,
+      "step": 965
+    },
+    {
+      "epoch": 0.47137280416395577,
+      "grad_norm": 0.194210484623909,
+      "learning_rate": 0.0001,
+      "loss": 1.7045,
+      "step": 966
+    },
+    {
+      "epoch": 0.4718607677293429,
+      "grad_norm": 0.18274177610874176,
+      "learning_rate": 0.0001,
+      "loss": 1.7462,
+      "step": 967
+    },
+    {
+      "epoch": 0.47234873129473,
+      "grad_norm": 0.211595356464386,
+      "learning_rate": 0.0001,
+      "loss": 1.7322,
+      "step": 968
+    },
+    {
+      "epoch": 0.4728366948601171,
+      "grad_norm": 0.1885220855474472,
+      "learning_rate": 0.0001,
+      "loss": 1.6825,
+      "step": 969
+    },
+    {
+      "epoch": 0.47332465842550425,
+      "grad_norm": 0.17875580489635468,
+      "learning_rate": 0.0001,
+      "loss": 1.6192,
+      "step": 970
+    },
+    {
+      "epoch": 0.47381262199089136,
+      "grad_norm": 0.1805390864610672,
+      "learning_rate": 0.0001,
+      "loss": 1.6668,
+      "step": 971
+    },
+    {
+      "epoch": 0.47430058555627846,
+      "grad_norm": 0.19222760200500488,
+      "learning_rate": 0.0001,
+      "loss": 1.7478,
+      "step": 972
+    },
+    {
+      "epoch": 0.47478854912166557,
+      "grad_norm": 0.18637999892234802,
+      "learning_rate": 0.0001,
+      "loss": 1.7773,
+      "step": 973
+    },
+    {
+      "epoch": 0.4752765126870527,
+      "grad_norm": 0.18341195583343506,
+      "learning_rate": 0.0001,
+      "loss": 1.7021,
+      "step": 974
+    },
+    {
+      "epoch": 0.47576447625243984,
+      "grad_norm": 0.17885076999664307,
+      "learning_rate": 0.0001,
+      "loss": 1.6424,
+      "step": 975
+    },
+    {
+      "epoch": 0.47625243981782694,
+      "grad_norm": 0.1952183097600937,
+      "learning_rate": 0.0001,
+      "loss": 1.9142,
+      "step": 976
+    },
+    {
+      "epoch": 0.47674040338321405,
+      "grad_norm": 0.18243496119976044,
+      "learning_rate": 0.0001,
+      "loss": 1.6983,
+      "step": 977
+    },
+    {
+      "epoch": 0.47722836694860116,
+      "grad_norm": 0.18224705755710602,
+      "learning_rate": 0.0001,
+      "loss": 1.5847,
+      "step": 978
+    },
+    {
+      "epoch": 0.47771633051398826,
+      "grad_norm": 0.25170522928237915,
+      "learning_rate": 0.0001,
+      "loss": 1.9113,
+      "step": 979
+    },
+    {
+      "epoch": 0.4782042940793754,
+      "grad_norm": 0.18615500628948212,
+      "learning_rate": 0.0001,
+      "loss": 1.7893,
+      "step": 980
+    },
+    {
+      "epoch": 0.47869225764476253,
+      "grad_norm": 0.18177960813045502,
+      "learning_rate": 0.0001,
+      "loss": 1.753,
+      "step": 981
+    },
+    {
+      "epoch": 0.47918022121014964,
+      "grad_norm": 0.17566373944282532,
+      "learning_rate": 0.0001,
+      "loss": 1.749,
+      "step": 982
+    },
+    {
+      "epoch": 0.47966818477553674,
+      "grad_norm": 0.18363641202449799,
+      "learning_rate": 0.0001,
+      "loss": 1.7202,
+      "step": 983
+    },
+    {
+      "epoch": 0.4801561483409239,
+      "grad_norm": 0.18019676208496094,
+      "learning_rate": 0.0001,
+      "loss": 1.7756,
+      "step": 984
+    },
+    {
+      "epoch": 0.480644111906311,
+      "grad_norm": 0.18838275969028473,
+      "learning_rate": 0.0001,
+      "loss": 1.6533,
+      "step": 985
+    },
+    {
+      "epoch": 0.4811320754716981,
+      "grad_norm": 0.17840002477169037,
+      "learning_rate": 0.0001,
+      "loss": 1.6495,
+      "step": 986
+    },
+    {
+      "epoch": 0.4816200390370852,
+      "grad_norm": 0.18629398941993713,
+      "learning_rate": 0.0001,
+      "loss": 1.746,
+      "step": 987
+    },
+    {
+      "epoch": 0.48210800260247233,
+      "grad_norm": 0.19068728387355804,
+      "learning_rate": 0.0001,
+      "loss": 1.7956,
+      "step": 988
+    },
+    {
+      "epoch": 0.4825959661678595,
+      "grad_norm": 0.17752403020858765,
+      "learning_rate": 0.0001,
+      "loss": 1.6085,
+      "step": 989
+    },
+    {
+      "epoch": 0.4830839297332466,
+      "grad_norm": 0.17869940400123596,
+      "learning_rate": 0.0001,
+      "loss": 1.687,
+      "step": 990
+    },
+    {
+      "epoch": 0.4835718932986337,
+      "grad_norm": 0.19462576508522034,
+      "learning_rate": 0.0001,
+      "loss": 1.766,
+      "step": 991
+    },
+    {
+      "epoch": 0.4840598568640208,
+      "grad_norm": 0.17635509371757507,
+      "learning_rate": 0.0001,
+      "loss": 1.6512,
+      "step": 992
+    },
+    {
+      "epoch": 0.4845478204294079,
+      "grad_norm": 0.18457075953483582,
+      "learning_rate": 0.0001,
+      "loss": 1.6829,
+      "step": 993
+    },
+    {
+      "epoch": 0.4850357839947951,
+      "grad_norm": 0.19008415937423706,
+      "learning_rate": 0.0001,
+      "loss": 1.8335,
+      "step": 994
+    },
+    {
+      "epoch": 0.4855237475601822,
+      "grad_norm": 0.1748104840517044,
+      "learning_rate": 0.0001,
+      "loss": 1.6822,
+      "step": 995
+    },
+    {
+      "epoch": 0.4860117111255693,
+      "grad_norm": 0.18871375918388367,
+      "learning_rate": 0.0001,
+      "loss": 1.7749,
+      "step": 996
+    },
+    {
+      "epoch": 0.4864996746909564,
+      "grad_norm": 0.19204716384410858,
+      "learning_rate": 0.0001,
+      "loss": 1.7027,
+      "step": 997
+    },
+    {
+      "epoch": 0.4869876382563435,
+      "grad_norm": 0.17363031208515167,
+      "learning_rate": 0.0001,
+      "loss": 1.6329,
+      "step": 998
+    },
+    {
+      "epoch": 0.48747560182173066,
+      "grad_norm": 0.18046556413173676,
+      "learning_rate": 0.0001,
+      "loss": 1.6251,
+      "step": 999
+    },
+    {
+      "epoch": 0.48796356538711777,
+      "grad_norm": 0.18280474841594696,
+      "learning_rate": 0.0001,
+      "loss": 1.7468,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4884515289525049,
+      "grad_norm": 0.1856307089328766,
+      "learning_rate": 0.0001,
+      "loss": 1.8059,
+      "step": 1001
+    },
+    {
+      "epoch": 0.488939492517892,
+      "grad_norm": 0.18734587728977203,
+      "learning_rate": 0.0001,
+      "loss": 1.7482,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4894274560832791,
+      "grad_norm": 0.18201518058776855,
+      "learning_rate": 0.0001,
+      "loss": 1.6618,
+      "step": 1003
+    },
+    {
+      "epoch": 0.48991541964866625,
+      "grad_norm": 0.18317224085330963,
+      "learning_rate": 0.0001,
+      "loss": 1.6556,
+      "step": 1004
+    },
+    {
+      "epoch": 0.49040338321405336,
+      "grad_norm": 0.18233336508274078,
+      "learning_rate": 0.0001,
+      "loss": 1.7073,
+      "step": 1005
+    },
+    {
+      "epoch": 0.49089134677944046,
+      "grad_norm": 0.19454477727413177,
+      "learning_rate": 0.0001,
+      "loss": 1.5993,
+      "step": 1006
+    },
+    {
+      "epoch": 0.49137931034482757,
+      "grad_norm": 0.1874353140592575,
+      "learning_rate": 0.0001,
+      "loss": 1.6976,
+      "step": 1007
+    },
+    {
+      "epoch": 0.49186727391021473,
+      "grad_norm": 0.18378609418869019,
+      "learning_rate": 0.0001,
+      "loss": 1.7292,
+      "step": 1008
+    },
+    {
+      "epoch": 0.49235523747560184,
+      "grad_norm": 0.18301472067832947,
+      "learning_rate": 0.0001,
+      "loss": 1.6702,
+      "step": 1009
+    },
+    {
+      "epoch": 0.49284320104098894,
+      "grad_norm": 0.18581345677375793,
+      "learning_rate": 0.0001,
+      "loss": 1.769,
+      "step": 1010
+    },
+    {
+      "epoch": 0.49333116460637605,
+      "grad_norm": 0.18604816496372223,
+      "learning_rate": 0.0001,
+      "loss": 1.7022,
+      "step": 1011
+    },
+    {
+      "epoch": 0.49381912817176316,
+      "grad_norm": 0.1670636236667633,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 1012
+    },
+    {
+      "epoch": 0.4943070917371503,
+      "grad_norm": 0.18545298278331757,
+      "learning_rate": 0.0001,
+      "loss": 1.777,
+      "step": 1013
+    },
+    {
+      "epoch": 0.4947950553025374,
+      "grad_norm": 0.18108947575092316,
+      "learning_rate": 0.0001,
+      "loss": 1.7066,
+      "step": 1014
+    },
+    {
+      "epoch": 0.49528301886792453,
+      "grad_norm": 0.18042118847370148,
+      "learning_rate": 0.0001,
+      "loss": 1.6393,
+      "step": 1015
+    },
+    {
+      "epoch": 0.49577098243331164,
+      "grad_norm": 0.19193610548973083,
+      "learning_rate": 0.0001,
+      "loss": 1.8438,
+      "step": 1016
+    },
+    {
+      "epoch": 0.49625894599869874,
+      "grad_norm": 0.18542861938476562,
+      "learning_rate": 0.0001,
+      "loss": 1.8076,
+      "step": 1017
+    },
+    {
+      "epoch": 0.4967469095640859,
+      "grad_norm": 0.17646706104278564,
+      "learning_rate": 0.0001,
+      "loss": 1.4699,
+      "step": 1018
+    },
+    {
+      "epoch": 0.497234873129473,
+      "grad_norm": 0.18862095475196838,
+      "learning_rate": 0.0001,
+      "loss": 1.7165,
+      "step": 1019
+    },
+    {
+      "epoch": 0.4977228366948601,
+      "grad_norm": 0.18618489801883698,
+      "learning_rate": 0.0001,
+      "loss": 1.7683,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4982108002602472,
+      "grad_norm": 0.18750105798244476,
+      "learning_rate": 0.0001,
+      "loss": 1.6681,
+      "step": 1021
+    },
+    {
+      "epoch": 0.49869876382563433,
+      "grad_norm": 0.1942930370569229,
+      "learning_rate": 0.0001,
+      "loss": 1.6555,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4991867273910215,
+      "grad_norm": 0.18165245652198792,
+      "learning_rate": 0.0001,
+      "loss": 1.7059,
+      "step": 1023
+    },
+    {
+      "epoch": 0.4996746909564086,
+      "grad_norm": 0.18349111080169678,
+      "learning_rate": 0.0001,
+      "loss": 1.7165,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5001626545217958,
+      "grad_norm": 0.17459173500537872,
+      "learning_rate": 0.0001,
+      "loss": 1.6784,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5006506180871828,
+      "grad_norm": 0.19236469268798828,
+      "learning_rate": 0.0001,
+      "loss": 1.6727,
+      "step": 1026
+    },
+    {
+      "epoch": 0.50113858165257,
+      "grad_norm": 0.18120145797729492,
+      "learning_rate": 0.0001,
+      "loss": 1.7109,
+      "step": 1027
+    },
+    {
+      "epoch": 0.501626545217957,
+      "grad_norm": 0.18319325149059296,
+      "learning_rate": 0.0001,
+      "loss": 1.6353,
+      "step": 1028
+    },
+    {
+      "epoch": 0.5021145087833442,
+      "grad_norm": 0.1807912439107895,
+      "learning_rate": 0.0001,
+      "loss": 1.6866,
+      "step": 1029
+    },
+    {
+      "epoch": 0.5026024723487313,
+      "grad_norm": 0.1748090237379074,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 1030
+    },
+    {
+      "epoch": 0.5030904359141184,
+      "grad_norm": 0.1822468489408493,
+      "learning_rate": 0.0001,
+      "loss": 1.7539,
+      "step": 1031
+    },
+    {
+      "epoch": 0.5035783994795056,
+      "grad_norm": 0.18360479176044464,
+      "learning_rate": 0.0001,
+      "loss": 1.6853,
+      "step": 1032
+    },
+    {
+      "epoch": 0.5040663630448926,
+      "grad_norm": 0.18836341798305511,
+      "learning_rate": 0.0001,
+      "loss": 1.6796,
+      "step": 1033
+    },
+    {
+      "epoch": 0.5045543266102798,
+      "grad_norm": 0.18044047057628632,
+      "learning_rate": 0.0001,
+      "loss": 1.6929,
+      "step": 1034
+    },
+    {
+      "epoch": 0.5050422901756669,
+      "grad_norm": 0.18836145102977753,
+      "learning_rate": 0.0001,
+      "loss": 1.8204,
+      "step": 1035
+    },
+    {
+      "epoch": 0.505530253741054,
+      "grad_norm": 0.1829444319009781,
+      "learning_rate": 0.0001,
+      "loss": 1.7364,
+      "step": 1036
+    },
+    {
+      "epoch": 0.5060182173064411,
+      "grad_norm": 0.1847165822982788,
+      "learning_rate": 0.0001,
+      "loss": 1.6792,
+      "step": 1037
+    },
+    {
+      "epoch": 0.5065061808718282,
+      "grad_norm": 0.17972713708877563,
+      "learning_rate": 0.0001,
+      "loss": 1.5694,
+      "step": 1038
+    },
+    {
+      "epoch": 0.5069941444372154,
+      "grad_norm": 0.1910099983215332,
+      "learning_rate": 0.0001,
+      "loss": 1.6189,
+      "step": 1039
+    },
+    {
+      "epoch": 0.5074821080026025,
+      "grad_norm": 0.18901146948337555,
+      "learning_rate": 0.0001,
+      "loss": 1.7515,
+      "step": 1040
+    },
+    {
+      "epoch": 0.5079700715679896,
+      "grad_norm": 0.18210864067077637,
+      "learning_rate": 0.0001,
+      "loss": 1.729,
+      "step": 1041
+    },
+    {
+      "epoch": 0.5084580351333767,
+      "grad_norm": 0.18417298793792725,
+      "learning_rate": 0.0001,
+      "loss": 1.7392,
+      "step": 1042
+    },
+    {
+      "epoch": 0.5089459986987638,
+      "grad_norm": 0.18548882007598877,
+      "learning_rate": 0.0001,
+      "loss": 1.7452,
+      "step": 1043
+    },
+    {
+      "epoch": 0.5094339622641509,
+      "grad_norm": 0.17644409835338593,
+      "learning_rate": 0.0001,
+      "loss": 1.5658,
+      "step": 1044
+    },
+    {
+      "epoch": 0.5099219258295381,
+      "grad_norm": 0.18809697031974792,
+      "learning_rate": 0.0001,
+      "loss": 1.6806,
+      "step": 1045
+    },
+    {
+      "epoch": 0.5104098893949252,
+      "grad_norm": 0.18309113383293152,
+      "learning_rate": 0.0001,
+      "loss": 1.7068,
+      "step": 1046
+    },
+    {
+      "epoch": 0.5108978529603123,
+      "grad_norm": 0.1873452365398407,
+      "learning_rate": 0.0001,
+      "loss": 1.7401,
+      "step": 1047
+    },
+    {
+      "epoch": 0.5113858165256994,
+      "grad_norm": 0.18118296563625336,
+      "learning_rate": 0.0001,
+      "loss": 1.6853,
+      "step": 1048
+    },
+    {
+      "epoch": 0.5118737800910865,
+      "grad_norm": 0.19551081955432892,
+      "learning_rate": 0.0001,
+      "loss": 1.6851,
+      "step": 1049
+    },
+    {
+      "epoch": 0.5123617436564737,
+      "grad_norm": 0.19051168859004974,
+      "learning_rate": 0.0001,
+      "loss": 1.7153,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5128497072218607,
+      "grad_norm": 0.1723107546567917,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 1051
+    },
+    {
+      "epoch": 0.5133376707872479,
+      "grad_norm": 0.18448057770729065,
+      "learning_rate": 0.0001,
+      "loss": 1.6798,
+      "step": 1052
+    },
+    {
+      "epoch": 0.513825634352635,
+      "grad_norm": 0.1888912320137024,
+      "learning_rate": 0.0001,
+      "loss": 1.7696,
+      "step": 1053
+    },
+    {
+      "epoch": 0.5143135979180221,
+      "grad_norm": 0.19481922686100006,
+      "learning_rate": 0.0001,
+      "loss": 1.6657,
+      "step": 1054
+    },
+    {
+      "epoch": 0.5148015614834093,
+      "grad_norm": 0.17614057660102844,
+      "learning_rate": 0.0001,
+      "loss": 1.6758,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5152895250487963,
+      "grad_norm": 0.1752062737941742,
+      "learning_rate": 0.0001,
+      "loss": 1.644,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5157774886141835,
+      "grad_norm": 0.1882951855659485,
+      "learning_rate": 0.0001,
+      "loss": 1.6644,
+      "step": 1057
+    },
+    {
+      "epoch": 0.5162654521795705,
+      "grad_norm": 0.20255088806152344,
+      "learning_rate": 0.0001,
+      "loss": 1.7119,
+      "step": 1058
+    },
+    {
+      "epoch": 0.5167534157449577,
+      "grad_norm": 0.181501105427742,
+      "learning_rate": 0.0001,
+      "loss": 1.662,
+      "step": 1059
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.1865651160478592,
+      "learning_rate": 0.0001,
+      "loss": 1.7279,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5177293428757319,
+      "grad_norm": 0.1911836862564087,
+      "learning_rate": 0.0001,
+      "loss": 1.6795,
+      "step": 1061
+    },
+    {
+      "epoch": 0.5182173064411191,
+      "grad_norm": 0.18534213304519653,
+      "learning_rate": 0.0001,
+      "loss": 1.7126,
+      "step": 1062
+    },
+    {
+      "epoch": 0.5187052700065062,
+      "grad_norm": 0.1829744428396225,
+      "learning_rate": 0.0001,
+      "loss": 1.6598,
+      "step": 1063
+    },
+    {
+      "epoch": 0.5191932335718933,
+      "grad_norm": 0.17899416387081146,
+      "learning_rate": 0.0001,
+      "loss": 1.6293,
+      "step": 1064
+    },
+    {
+      "epoch": 0.5196811971372804,
+      "grad_norm": 0.17233431339263916,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5201691607026675,
+      "grad_norm": 0.1891251802444458,
+      "learning_rate": 0.0001,
+      "loss": 1.72,
+      "step": 1066
+    },
+    {
+      "epoch": 0.5206571242680547,
+      "grad_norm": 0.19288107752799988,
+      "learning_rate": 0.0001,
+      "loss": 1.8331,
+      "step": 1067
+    },
+    {
+      "epoch": 0.5211450878334418,
+      "grad_norm": 0.18534426391124725,
+      "learning_rate": 0.0001,
+      "loss": 1.6229,
+      "step": 1068
+    },
+    {
+      "epoch": 0.5216330513988289,
+      "grad_norm": 0.19013041257858276,
+      "learning_rate": 0.0001,
+      "loss": 1.7331,
+      "step": 1069
+    },
+    {
+      "epoch": 0.522121014964216,
+      "grad_norm": 0.18765857815742493,
+      "learning_rate": 0.0001,
+      "loss": 1.6951,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5226089785296031,
+      "grad_norm": 0.17150448262691498,
+      "learning_rate": 0.0001,
+      "loss": 1.6581,
+      "step": 1071
+    },
+    {
+      "epoch": 0.5230969420949902,
+      "grad_norm": 0.20504555106163025,
+      "learning_rate": 0.0001,
+      "loss": 1.7247,
+      "step": 1072
+    },
+    {
+      "epoch": 0.5235849056603774,
+      "grad_norm": 0.17816084623336792,
+      "learning_rate": 0.0001,
+      "loss": 1.5654,
+      "step": 1073
+    },
+    {
+      "epoch": 0.5240728692257645,
+      "grad_norm": 0.1842648684978485,
+      "learning_rate": 0.0001,
+      "loss": 1.601,
+      "step": 1074
+    },
+    {
+      "epoch": 0.5245608327911516,
+      "grad_norm": 0.18370290100574493,
+      "learning_rate": 0.0001,
+      "loss": 1.6369,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5250487963565387,
+      "grad_norm": 0.18270552158355713,
+      "learning_rate": 0.0001,
+      "loss": 1.6541,
+      "step": 1076
+    },
+    {
+      "epoch": 0.5255367599219258,
+      "grad_norm": 0.1808508038520813,
+      "learning_rate": 0.0001,
+      "loss": 1.6598,
+      "step": 1077
+    },
+    {
+      "epoch": 0.526024723487313,
+      "grad_norm": 0.17794300615787506,
+      "learning_rate": 0.0001,
+      "loss": 1.7294,
+      "step": 1078
+    },
+    {
+      "epoch": 0.5265126870527,
+      "grad_norm": 0.18382461369037628,
+      "learning_rate": 0.0001,
+      "loss": 1.6901,
+      "step": 1079
+    },
+    {
+      "epoch": 0.5270006506180872,
+      "grad_norm": 0.1806422621011734,
+      "learning_rate": 0.0001,
+      "loss": 1.6193,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5274886141834743,
+      "grad_norm": 0.18108539283275604,
+      "learning_rate": 0.0001,
+      "loss": 1.6911,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5279765777488614,
+      "grad_norm": 0.18681305646896362,
+      "learning_rate": 0.0001,
+      "loss": 1.726,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5284645413142486,
+      "grad_norm": 0.18909889459609985,
+      "learning_rate": 0.0001,
+      "loss": 1.6857,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5289525048796356,
+      "grad_norm": 0.18421509861946106,
+      "learning_rate": 0.0001,
+      "loss": 1.6564,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5294404684450228,
+      "grad_norm": 0.18811306357383728,
+      "learning_rate": 0.0001,
+      "loss": 1.7817,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5299284320104098,
+      "grad_norm": 0.17478449642658234,
+      "learning_rate": 0.0001,
+      "loss": 1.681,
+      "step": 1086
+    },
+    {
+      "epoch": 0.530416395575797,
+      "grad_norm": 0.1789132058620453,
+      "learning_rate": 0.0001,
+      "loss": 1.6906,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5309043591411842,
+      "grad_norm": 0.18358959257602692,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5313923227065712,
+      "grad_norm": 0.18565410375595093,
+      "learning_rate": 0.0001,
+      "loss": 1.7078,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5318802862719584,
+      "grad_norm": 0.19210746884346008,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5323682498373454,
+      "grad_norm": 0.18205370008945465,
+      "learning_rate": 0.0001,
+      "loss": 1.6541,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5328562134027326,
+      "grad_norm": 0.19181987643241882,
+      "learning_rate": 0.0001,
+      "loss": 1.8033,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5333441769681198,
+      "grad_norm": 0.20362940430641174,
+      "learning_rate": 0.0001,
+      "loss": 1.7497,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5338321405335068,
+      "grad_norm": 0.1858234405517578,
+      "learning_rate": 0.0001,
+      "loss": 1.6342,
+      "step": 1094
+    },
+    {
+      "epoch": 0.534320104098894,
+      "grad_norm": 0.19925346970558167,
+      "learning_rate": 0.0001,
+      "loss": 1.686,
+      "step": 1095
+    },
+    {
+      "epoch": 0.534808067664281,
+      "grad_norm": 0.19114282727241516,
+      "learning_rate": 0.0001,
+      "loss": 1.7186,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5352960312296682,
+      "grad_norm": 0.1771971732378006,
+      "learning_rate": 0.0001,
+      "loss": 1.776,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5357839947950553,
+      "grad_norm": 0.18942809104919434,
+      "learning_rate": 0.0001,
+      "loss": 1.7179,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5362719583604424,
+      "grad_norm": 0.1868084967136383,
+      "learning_rate": 0.0001,
+      "loss": 1.6454,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5367599219258296,
+      "grad_norm": 0.18689820170402527,
+      "learning_rate": 0.0001,
+      "loss": 1.6196,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5372478854912166,
+      "grad_norm": 0.1820572018623352,
+      "learning_rate": 0.0001,
+      "loss": 1.6673,
+      "step": 1101
+    },
+    {
+      "epoch": 0.5377358490566038,
+      "grad_norm": 0.17870689928531647,
+      "learning_rate": 0.0001,
+      "loss": 1.5968,
+      "step": 1102
+    },
+    {
+      "epoch": 0.5382238126219909,
+      "grad_norm": 0.18118569254875183,
+      "learning_rate": 0.0001,
+      "loss": 1.7227,
+      "step": 1103
+    },
+    {
+      "epoch": 0.538711776187378,
+      "grad_norm": 0.1880924552679062,
+      "learning_rate": 0.0001,
+      "loss": 1.6108,
+      "step": 1104
+    },
+    {
+      "epoch": 0.5391997397527651,
+      "grad_norm": 0.18598206341266632,
+      "learning_rate": 0.0001,
+      "loss": 1.6542,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5396877033181522,
+      "grad_norm": 0.1872934103012085,
+      "learning_rate": 0.0001,
+      "loss": 1.737,
+      "step": 1106
+    },
+    {
+      "epoch": 0.5401756668835394,
+      "grad_norm": 0.1890784651041031,
+      "learning_rate": 0.0001,
+      "loss": 1.6661,
+      "step": 1107
+    },
+    {
+      "epoch": 0.5406636304489265,
+      "grad_norm": 0.18039381504058838,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 1108
+    },
+    {
+      "epoch": 0.5411515940143136,
+      "grad_norm": 0.18550348281860352,
+      "learning_rate": 0.0001,
+      "loss": 1.6828,
+      "step": 1109
+    },
+    {
+      "epoch": 0.5416395575797007,
+      "grad_norm": 0.17449964582920074,
+      "learning_rate": 0.0001,
+      "loss": 1.5034,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5421275211450879,
+      "grad_norm": 0.18202394247055054,
+      "learning_rate": 0.0001,
+      "loss": 1.6561,
+      "step": 1111
+    },
+    {
+      "epoch": 0.5426154847104749,
+      "grad_norm": 0.19365155696868896,
+      "learning_rate": 0.0001,
+      "loss": 1.5,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5431034482758621,
+      "grad_norm": 0.17744717001914978,
+      "learning_rate": 0.0001,
+      "loss": 1.5921,
+      "step": 1113
+    },
+    {
+      "epoch": 0.5435914118412492,
+      "grad_norm": 0.17965885996818542,
+      "learning_rate": 0.0001,
+      "loss": 1.6819,
+      "step": 1114
+    },
+    {
+      "epoch": 0.5440793754066363,
+      "grad_norm": 0.17675574123859406,
+      "learning_rate": 0.0001,
+      "loss": 1.6471,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5445673389720235,
+      "grad_norm": 0.17376431822776794,
+      "learning_rate": 0.0001,
+      "loss": 1.7007,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5450553025374105,
+      "grad_norm": 0.18188650906085968,
+      "learning_rate": 0.0001,
+      "loss": 1.774,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5455432661027977,
+      "grad_norm": 0.17877081036567688,
+      "learning_rate": 0.0001,
+      "loss": 1.6535,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5460312296681847,
+      "grad_norm": 0.17933769524097443,
+      "learning_rate": 0.0001,
+      "loss": 1.7362,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5465191932335719,
+      "grad_norm": 0.1805192083120346,
+      "learning_rate": 0.0001,
+      "loss": 1.7321,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5470071567989591,
+      "grad_norm": 0.17312046885490417,
+      "learning_rate": 0.0001,
+      "loss": 1.6415,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5474951203643461,
+      "grad_norm": 0.18119437992572784,
+      "learning_rate": 0.0001,
+      "loss": 1.7104,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5479830839297333,
+      "grad_norm": 0.182356595993042,
+      "learning_rate": 0.0001,
+      "loss": 1.6866,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5484710474951203,
+      "grad_norm": 0.1846156120300293,
+      "learning_rate": 0.0001,
+      "loss": 1.6612,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5489590110605075,
+      "grad_norm": 0.17960377037525177,
+      "learning_rate": 0.0001,
+      "loss": 1.6848,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5494469746258946,
+      "grad_norm": 0.17133495211601257,
+      "learning_rate": 0.0001,
+      "loss": 1.5885,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5499349381912817,
+      "grad_norm": 0.18075834214687347,
+      "learning_rate": 0.0001,
+      "loss": 1.7428,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5504229017566689,
+      "grad_norm": 0.18319405615329742,
+      "learning_rate": 0.0001,
+      "loss": 1.5856,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5509108653220559,
+      "grad_norm": 0.17644239962100983,
+      "learning_rate": 0.0001,
+      "loss": 1.6198,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5513988288874431,
+      "grad_norm": 0.18394580483436584,
+      "learning_rate": 0.0001,
+      "loss": 1.6435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5518867924528302,
+      "grad_norm": 0.1763201355934143,
+      "learning_rate": 0.0001,
+      "loss": 1.6975,
+      "step": 1131
+    },
+    {
+      "epoch": 0.5523747560182173,
+      "grad_norm": 0.16742850840091705,
+      "learning_rate": 0.0001,
+      "loss": 1.5377,
+      "step": 1132
+    },
+    {
+      "epoch": 0.5528627195836044,
+      "grad_norm": 0.1892685890197754,
+      "learning_rate": 0.0001,
+      "loss": 1.6111,
+      "step": 1133
+    },
+    {
+      "epoch": 0.5533506831489915,
+      "grad_norm": 0.18346691131591797,
+      "learning_rate": 0.0001,
+      "loss": 1.6844,
+      "step": 1134
+    },
+    {
+      "epoch": 0.5538386467143787,
+      "grad_norm": 0.1796543449163437,
+      "learning_rate": 0.0001,
+      "loss": 1.7746,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5543266102797658,
+      "grad_norm": 0.18673722445964813,
+      "learning_rate": 0.0001,
+      "loss": 1.6464,
+      "step": 1136
+    },
+    {
+      "epoch": 0.5548145738451529,
+      "grad_norm": 0.17763900756835938,
+      "learning_rate": 0.0001,
+      "loss": 1.6237,
+      "step": 1137
+    },
+    {
+      "epoch": 0.55530253741054,
+      "grad_norm": 0.17686204612255096,
+      "learning_rate": 0.0001,
+      "loss": 1.5131,
+      "step": 1138
+    },
+    {
+      "epoch": 0.5557905009759271,
+      "grad_norm": 0.18360872566699982,
+      "learning_rate": 0.0001,
+      "loss": 1.6699,
+      "step": 1139
+    },
+    {
+      "epoch": 0.5562784645413142,
+      "grad_norm": 0.1827259063720703,
+      "learning_rate": 0.0001,
+      "loss": 1.746,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5567664281067014,
+      "grad_norm": 0.17962484061717987,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 1141
+    },
+    {
+      "epoch": 0.5572543916720885,
+      "grad_norm": 0.18114878237247467,
+      "learning_rate": 0.0001,
+      "loss": 1.6737,
+      "step": 1142
+    },
+    {
+      "epoch": 0.5577423552374756,
+      "grad_norm": 0.18968282639980316,
+      "learning_rate": 0.0001,
+      "loss": 1.7798,
+      "step": 1143
+    },
+    {
+      "epoch": 0.5582303188028627,
+      "grad_norm": 0.18505877256393433,
+      "learning_rate": 0.0001,
+      "loss": 1.708,
+      "step": 1144
+    },
+    {
+      "epoch": 0.5587182823682498,
+      "grad_norm": 0.1776040643453598,
+      "learning_rate": 0.0001,
+      "loss": 1.7424,
+      "step": 1145
+    },
+    {
+      "epoch": 0.559206245933637,
+      "grad_norm": 0.17982693016529083,
+      "learning_rate": 0.0001,
+      "loss": 1.6197,
+      "step": 1146
+    },
+    {
+      "epoch": 0.559694209499024,
+      "grad_norm": 0.19187504053115845,
+      "learning_rate": 0.0001,
+      "loss": 1.7451,
+      "step": 1147
+    },
+    {
+      "epoch": 0.5601821730644112,
+      "grad_norm": 0.17975229024887085,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 1148
+    },
+    {
+      "epoch": 0.5606701366297983,
+      "grad_norm": 0.18996664881706238,
+      "learning_rate": 0.0001,
+      "loss": 1.7377,
+      "step": 1149
+    },
+    {
+      "epoch": 0.5611581001951854,
+      "grad_norm": 0.18252383172512054,
+      "learning_rate": 0.0001,
+      "loss": 1.628,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5616460637605726,
+      "grad_norm": 0.18448345363140106,
+      "learning_rate": 0.0001,
+      "loss": 1.7109,
+      "step": 1151
+    },
+    {
+      "epoch": 0.5621340273259596,
+      "grad_norm": 0.17741243541240692,
+      "learning_rate": 0.0001,
+      "loss": 1.6088,
+      "step": 1152
+    },
+    {
+      "epoch": 0.5626219908913468,
+      "grad_norm": 0.19825778901576996,
+      "learning_rate": 0.0001,
+      "loss": 1.5972,
+      "step": 1153
+    },
+    {
+      "epoch": 0.563109954456734,
+      "grad_norm": 0.18595324456691742,
+      "learning_rate": 0.0001,
+      "loss": 1.672,
+      "step": 1154
+    },
+    {
+      "epoch": 0.563597918022121,
+      "grad_norm": 0.18176652491092682,
+      "learning_rate": 0.0001,
+      "loss": 1.6216,
+      "step": 1155
+    },
+    {
+      "epoch": 0.5640858815875082,
+      "grad_norm": 0.1950223743915558,
+      "learning_rate": 0.0001,
+      "loss": 1.705,
+      "step": 1156
+    },
+    {
+      "epoch": 0.5645738451528952,
+      "grad_norm": 0.1990990787744522,
+      "learning_rate": 0.0001,
+      "loss": 1.7031,
+      "step": 1157
+    },
+    {
+      "epoch": 0.5650618087182824,
+      "grad_norm": 0.1937246173620224,
+      "learning_rate": 0.0001,
+      "loss": 1.6838,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5655497722836695,
+      "grad_norm": 0.1884077787399292,
+      "learning_rate": 0.0001,
+      "loss": 1.5994,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.19293847680091858,
+      "learning_rate": 0.0001,
+      "loss": 1.7657,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5665256994144438,
+      "grad_norm": 0.18362392485141754,
+      "learning_rate": 0.0001,
+      "loss": 1.7443,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5670136629798308,
+      "grad_norm": 0.17800559103488922,
+      "learning_rate": 0.0001,
+      "loss": 1.6433,
+      "step": 1162
+    },
+    {
+      "epoch": 0.567501626545218,
+      "grad_norm": 0.1774267852306366,
+      "learning_rate": 0.0001,
+      "loss": 1.6468,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5679895901106051,
+      "grad_norm": 0.18834517896175385,
+      "learning_rate": 0.0001,
+      "loss": 1.7715,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5684775536759922,
+      "grad_norm": 0.1841384768486023,
+      "learning_rate": 0.0001,
+      "loss": 1.7604,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5689655172413793,
+      "grad_norm": 0.18285635113716125,
+      "learning_rate": 0.0001,
+      "loss": 1.6941,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5694534808067664,
+      "grad_norm": 0.1796160191297531,
+      "learning_rate": 0.0001,
+      "loss": 1.6835,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5699414443721535,
+      "grad_norm": 0.18359331786632538,
+      "learning_rate": 0.0001,
+      "loss": 1.6658,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5704294079375407,
+      "grad_norm": 0.17833665013313293,
+      "learning_rate": 0.0001,
+      "loss": 1.6455,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5709173715029278,
+      "grad_norm": 0.17929013073444366,
+      "learning_rate": 0.0001,
+      "loss": 1.5912,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5714053350683149,
+      "grad_norm": 0.18901382386684418,
+      "learning_rate": 0.0001,
+      "loss": 1.7305,
+      "step": 1171
+    },
+    {
+      "epoch": 0.571893298633702,
+      "grad_norm": 0.18040084838867188,
+      "learning_rate": 0.0001,
+      "loss": 1.6239,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5723812621990891,
+      "grad_norm": 0.1832232028245926,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5728692257644763,
+      "grad_norm": 0.1900448203086853,
+      "learning_rate": 0.0001,
+      "loss": 1.7176,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5733571893298633,
+      "grad_norm": 0.1859886199235916,
+      "learning_rate": 0.0001,
+      "loss": 1.6823,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5738451528952505,
+      "grad_norm": 0.1816965788602829,
+      "learning_rate": 0.0001,
+      "loss": 1.6936,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5743331164606376,
+      "grad_norm": 0.1927751749753952,
+      "learning_rate": 0.0001,
+      "loss": 1.7069,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5748210800260247,
+      "grad_norm": 0.20290379226207733,
+      "learning_rate": 0.0001,
+      "loss": 1.7987,
+      "step": 1178
+    },
+    {
+      "epoch": 0.5753090435914119,
+      "grad_norm": 0.1756032556295395,
+      "learning_rate": 0.0001,
+      "loss": 1.656,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5757970071567989,
+      "grad_norm": 0.19676676392555237,
+      "learning_rate": 0.0001,
+      "loss": 1.8415,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5762849707221861,
+      "grad_norm": 0.18112622201442719,
+      "learning_rate": 0.0001,
+      "loss": 1.6081,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5767729342875731,
+      "grad_norm": 0.20109887421131134,
+      "learning_rate": 0.0001,
+      "loss": 1.7772,
+      "step": 1182
+    },
+    {
+      "epoch": 0.5772608978529603,
+      "grad_norm": 0.191656693816185,
+      "learning_rate": 0.0001,
+      "loss": 1.6869,
+      "step": 1183
+    },
+    {
+      "epoch": 0.5777488614183475,
+      "grad_norm": 0.17886236310005188,
+      "learning_rate": 0.0001,
+      "loss": 1.5931,
+      "step": 1184
+    },
+    {
+      "epoch": 0.5782368249837345,
+      "grad_norm": 0.18148286640644073,
+      "learning_rate": 0.0001,
+      "loss": 1.6056,
+      "step": 1185
+    },
+    {
+      "epoch": 0.5787247885491217,
+      "grad_norm": 0.20596817135810852,
+      "learning_rate": 0.0001,
+      "loss": 1.6129,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5792127521145087,
+      "grad_norm": 0.17900511622428894,
+      "learning_rate": 0.0001,
+      "loss": 1.6487,
+      "step": 1187
+    },
+    {
+      "epoch": 0.5797007156798959,
+      "grad_norm": 0.1893642693758011,
+      "learning_rate": 0.0001,
+      "loss": 1.7566,
+      "step": 1188
+    },
+    {
+      "epoch": 0.5801886792452831,
+      "grad_norm": 0.19354504346847534,
+      "learning_rate": 0.0001,
+      "loss": 1.6665,
+      "step": 1189
+    },
+    {
+      "epoch": 0.5806766428106701,
+      "grad_norm": 0.18692192435264587,
+      "learning_rate": 0.0001,
+      "loss": 1.7069,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5811646063760573,
+      "grad_norm": 0.204212948679924,
+      "learning_rate": 0.0001,
+      "loss": 1.7943,
+      "step": 1191
+    },
+    {
+      "epoch": 0.5816525699414443,
+      "grad_norm": 0.18666908144950867,
+      "learning_rate": 0.0001,
+      "loss": 1.7031,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5821405335068315,
+      "grad_norm": 0.1859620362520218,
+      "learning_rate": 0.0001,
+      "loss": 1.7443,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5826284970722186,
+      "grad_norm": 0.1774389147758484,
+      "learning_rate": 0.0001,
+      "loss": 1.6697,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5831164606376057,
+      "grad_norm": 0.17645440995693207,
+      "learning_rate": 0.0001,
+      "loss": 1.7566,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5836044242029929,
+      "grad_norm": 0.17927305400371552,
+      "learning_rate": 0.0001,
+      "loss": 1.5341,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5840923877683799,
+      "grad_norm": 0.19179411232471466,
+      "learning_rate": 0.0001,
+      "loss": 1.706,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5845803513337671,
+      "grad_norm": 0.18921273946762085,
+      "learning_rate": 0.0001,
+      "loss": 1.6651,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5850683148991542,
+      "grad_norm": 0.20988748967647552,
+      "learning_rate": 0.0001,
+      "loss": 1.8307,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5855562784645413,
+      "grad_norm": 0.1767909973859787,
+      "learning_rate": 0.0001,
+      "loss": 1.7116,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5860442420299284,
+      "grad_norm": 0.18889738619327545,
+      "learning_rate": 0.0001,
+      "loss": 1.6623,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5865322055953156,
+      "grad_norm": 0.17658928036689758,
+      "learning_rate": 0.0001,
+      "loss": 1.6147,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5870201691607027,
+      "grad_norm": 0.181167870759964,
+      "learning_rate": 0.0001,
+      "loss": 1.658,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5875081327260898,
+      "grad_norm": 0.18597833812236786,
+      "learning_rate": 0.0001,
+      "loss": 1.6392,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5879960962914769,
+      "grad_norm": 0.1838957518339157,
+      "learning_rate": 0.0001,
+      "loss": 1.699,
+      "step": 1205
+    },
+    {
+      "epoch": 0.588484059856864,
+      "grad_norm": 0.18274423480033875,
+      "learning_rate": 0.0001,
+      "loss": 1.7159,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5889720234222512,
+      "grad_norm": 0.19154992699623108,
+      "learning_rate": 0.0001,
+      "loss": 1.7488,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5894599869876382,
+      "grad_norm": 0.17971384525299072,
+      "learning_rate": 0.0001,
+      "loss": 1.5762,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5899479505530254,
+      "grad_norm": 0.17908671498298645,
+      "learning_rate": 0.0001,
+      "loss": 1.6836,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5904359141184125,
+      "grad_norm": 0.17960527539253235,
+      "learning_rate": 0.0001,
+      "loss": 1.6949,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5909238776837996,
+      "grad_norm": 0.18325302004814148,
+      "learning_rate": 0.0001,
+      "loss": 1.636,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5914118412491868,
+      "grad_norm": 0.18727539479732513,
+      "learning_rate": 0.0001,
+      "loss": 1.6365,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5918998048145738,
+      "grad_norm": 0.1794605702161789,
+      "learning_rate": 0.0001,
+      "loss": 1.6403,
+      "step": 1213
+    },
+    {
+      "epoch": 0.592387768379961,
+      "grad_norm": 0.17613062262535095,
+      "learning_rate": 0.0001,
+      "loss": 1.5417,
+      "step": 1214
+    },
+    {
+      "epoch": 0.592875731945348,
+      "grad_norm": 0.1804569661617279,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5933636955107352,
+      "grad_norm": 0.1809697449207306,
+      "learning_rate": 0.0001,
+      "loss": 1.6587,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5938516590761224,
+      "grad_norm": 0.19974054396152496,
+      "learning_rate": 0.0001,
+      "loss": 1.6771,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5943396226415094,
+      "grad_norm": 0.1866472214460373,
+      "learning_rate": 0.0001,
+      "loss": 1.7422,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5948275862068966,
+      "grad_norm": 0.20864078402519226,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5953155497722836,
+      "grad_norm": 0.18205468356609344,
+      "learning_rate": 0.0001,
+      "loss": 1.7434,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5958035133376708,
+      "grad_norm": 0.19407768547534943,
+      "learning_rate": 0.0001,
+      "loss": 1.752,
+      "step": 1221
+    },
+    {
+      "epoch": 0.596291476903058,
+      "grad_norm": 0.1877565234899521,
+      "learning_rate": 0.0001,
+      "loss": 1.711,
+      "step": 1222
+    },
+    {
+      "epoch": 0.596779440468445,
+      "grad_norm": 0.18702515959739685,
+      "learning_rate": 0.0001,
+      "loss": 1.7298,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5972674040338322,
+      "grad_norm": 0.17825458943843842,
+      "learning_rate": 0.0001,
+      "loss": 1.7,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5977553675992192,
+      "grad_norm": 0.18612068891525269,
+      "learning_rate": 0.0001,
+      "loss": 1.7307,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5982433311646064,
+      "grad_norm": 0.1892668455839157,
+      "learning_rate": 0.0001,
+      "loss": 1.7661,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5987312947299935,
+      "grad_norm": 0.18714402616024017,
+      "learning_rate": 0.0001,
+      "loss": 1.7071,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5992192582953806,
+      "grad_norm": 0.21308167278766632,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5997072218607677,
+      "grad_norm": 0.21097207069396973,
+      "learning_rate": 0.0001,
+      "loss": 1.8387,
+      "step": 1229
+    },
+    {
+      "epoch": 0.6001951854261548,
+      "grad_norm": 0.19734272360801697,
+      "learning_rate": 0.0001,
+      "loss": 1.6462,
+      "step": 1230
+    },
+    {
+      "epoch": 0.600683148991542,
+      "grad_norm": 0.17935802042484283,
+      "learning_rate": 0.0001,
+      "loss": 1.6129,
+      "step": 1231
+    },
+    {
+      "epoch": 0.6011711125569291,
+      "grad_norm": 0.1758161038160324,
+      "learning_rate": 0.0001,
+      "loss": 1.5895,
+      "step": 1232
+    },
+    {
+      "epoch": 0.6016590761223162,
+      "grad_norm": 0.19049344956874847,
+      "learning_rate": 0.0001,
+      "loss": 1.5845,
+      "step": 1233
+    },
+    {
+      "epoch": 0.6021470396877033,
+      "grad_norm": 0.19208753108978271,
+      "learning_rate": 0.0001,
+      "loss": 1.5719,
+      "step": 1234
+    },
+    {
+      "epoch": 0.6026350032530904,
+      "grad_norm": 0.18235936760902405,
+      "learning_rate": 0.0001,
+      "loss": 1.7095,
+      "step": 1235
+    },
+    {
+      "epoch": 0.6031229668184775,
+      "grad_norm": 0.19607332348823547,
+      "learning_rate": 0.0001,
+      "loss": 1.6312,
+      "step": 1236
+    },
+    {
+      "epoch": 0.6036109303838647,
+      "grad_norm": 0.18990549445152283,
+      "learning_rate": 0.0001,
+      "loss": 1.5929,
+      "step": 1237
+    },
+    {
+      "epoch": 0.6040988939492518,
+      "grad_norm": 0.1892758309841156,
+      "learning_rate": 0.0001,
+      "loss": 1.6401,
+      "step": 1238
+    },
+    {
+      "epoch": 0.6045868575146389,
+      "grad_norm": 0.19703449308872223,
+      "learning_rate": 0.0001,
+      "loss": 1.7751,
+      "step": 1239
+    },
+    {
+      "epoch": 0.605074821080026,
+      "grad_norm": 0.18029844760894775,
+      "learning_rate": 0.0001,
+      "loss": 1.5848,
+      "step": 1240
+    },
+    {
+      "epoch": 0.6055627846454131,
+      "grad_norm": 0.18993432819843292,
+      "learning_rate": 0.0001,
+      "loss": 1.6906,
+      "step": 1241
+    },
+    {
+      "epoch": 0.6060507482108003,
+      "grad_norm": 0.18150927126407623,
+      "learning_rate": 0.0001,
+      "loss": 1.6235,
+      "step": 1242
+    },
+    {
+      "epoch": 0.6065387117761873,
+      "grad_norm": 0.1799500286579132,
+      "learning_rate": 0.0001,
+      "loss": 1.6826,
+      "step": 1243
+    },
+    {
+      "epoch": 0.6070266753415745,
+      "grad_norm": 0.18196311593055725,
+      "learning_rate": 0.0001,
+      "loss": 1.5883,
+      "step": 1244
+    },
+    {
+      "epoch": 0.6075146389069617,
+      "grad_norm": 0.19016975164413452,
+      "learning_rate": 0.0001,
+      "loss": 1.685,
+      "step": 1245
+    },
+    {
+      "epoch": 0.6080026024723487,
+      "grad_norm": 0.18666522204875946,
+      "learning_rate": 0.0001,
+      "loss": 1.7852,
+      "step": 1246
+    },
+    {
+      "epoch": 0.6084905660377359,
+      "grad_norm": 0.18055546283721924,
+      "learning_rate": 0.0001,
+      "loss": 1.6591,
+      "step": 1247
+    },
+    {
+      "epoch": 0.6089785296031229,
+      "grad_norm": 0.18201833963394165,
+      "learning_rate": 0.0001,
+      "loss": 1.7492,
+      "step": 1248
+    },
+    {
+      "epoch": 0.6094664931685101,
+      "grad_norm": 0.18506184220314026,
+      "learning_rate": 0.0001,
+      "loss": 1.7207,
+      "step": 1249
+    },
+    {
+      "epoch": 0.6099544567338973,
+      "grad_norm": 0.17904452979564667,
+      "learning_rate": 0.0001,
+      "loss": 1.6761,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6104424202992843,
+      "grad_norm": 0.18653587996959686,
+      "learning_rate": 0.0001,
+      "loss": 1.6684,
+      "step": 1251
+    },
+    {
+      "epoch": 0.6109303838646715,
+      "grad_norm": 0.19012029469013214,
+      "learning_rate": 0.0001,
+      "loss": 1.7476,
+      "step": 1252
+    },
+    {
+      "epoch": 0.6114183474300585,
+      "grad_norm": 0.17272864282131195,
+      "learning_rate": 0.0001,
+      "loss": 1.6047,
+      "step": 1253
+    },
+    {
+      "epoch": 0.6119063109954457,
+      "grad_norm": 0.19090582430362701,
+      "learning_rate": 0.0001,
+      "loss": 1.7275,
+      "step": 1254
+    },
+    {
+      "epoch": 0.6123942745608328,
+      "grad_norm": 0.18830102682113647,
+      "learning_rate": 0.0001,
+      "loss": 1.6075,
+      "step": 1255
+    },
+    {
+      "epoch": 0.6128822381262199,
+      "grad_norm": 0.1959345042705536,
+      "learning_rate": 0.0001,
+      "loss": 1.77,
+      "step": 1256
+    },
+    {
+      "epoch": 0.613370201691607,
+      "grad_norm": 0.18547998368740082,
+      "learning_rate": 0.0001,
+      "loss": 1.6256,
+      "step": 1257
+    },
+    {
+      "epoch": 0.6138581652569941,
+      "grad_norm": 0.18706414103507996,
+      "learning_rate": 0.0001,
+      "loss": 1.6926,
+      "step": 1258
+    },
+    {
+      "epoch": 0.6143461288223813,
+      "grad_norm": 0.18563984334468842,
+      "learning_rate": 0.0001,
+      "loss": 1.6137,
+      "step": 1259
+    },
+    {
+      "epoch": 0.6148340923877684,
+      "grad_norm": 0.18717099726200104,
+      "learning_rate": 0.0001,
+      "loss": 1.717,
+      "step": 1260
+    },
+    {
+      "epoch": 0.6153220559531555,
+      "grad_norm": 0.18817085027694702,
+      "learning_rate": 0.0001,
+      "loss": 1.765,
+      "step": 1261
+    },
+    {
+      "epoch": 0.6158100195185426,
+      "grad_norm": 0.18568897247314453,
+      "learning_rate": 0.0001,
+      "loss": 1.6533,
+      "step": 1262
+    },
+    {
+      "epoch": 0.6162979830839297,
+      "grad_norm": 0.18605171144008636,
+      "learning_rate": 0.0001,
+      "loss": 1.602,
+      "step": 1263
+    },
+    {
+      "epoch": 0.6167859466493169,
+      "grad_norm": 0.19337786734104156,
+      "learning_rate": 0.0001,
+      "loss": 1.7735,
+      "step": 1264
+    },
+    {
+      "epoch": 0.617273910214704,
+      "grad_norm": 0.1964695155620575,
+      "learning_rate": 0.0001,
+      "loss": 1.6991,
+      "step": 1265
+    },
+    {
+      "epoch": 0.6177618737800911,
+      "grad_norm": 0.19506755471229553,
+      "learning_rate": 0.0001,
+      "loss": 1.6753,
+      "step": 1266
+    },
+    {
+      "epoch": 0.6182498373454782,
+      "grad_norm": 0.19231939315795898,
+      "learning_rate": 0.0001,
+      "loss": 1.6501,
+      "step": 1267
+    },
+    {
+      "epoch": 0.6187378009108653,
+      "grad_norm": 0.1804661899805069,
+      "learning_rate": 0.0001,
+      "loss": 1.7394,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6192257644762524,
+      "grad_norm": 0.1843184381723404,
+      "learning_rate": 0.0001,
+      "loss": 1.7195,
+      "step": 1269
+    },
+    {
+      "epoch": 0.6197137280416396,
+      "grad_norm": 0.18432582914829254,
+      "learning_rate": 0.0001,
+      "loss": 1.6494,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6202016916070267,
+      "grad_norm": 0.18056923151016235,
+      "learning_rate": 0.0001,
+      "loss": 1.6848,
+      "step": 1271
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 0.18145865201950073,
+      "learning_rate": 0.0001,
+      "loss": 1.5716,
+      "step": 1272
+    },
+    {
+      "epoch": 0.6211776187378009,
+      "grad_norm": 0.1727244108915329,
+      "learning_rate": 0.0001,
+      "loss": 1.5528,
+      "step": 1273
+    },
+    {
+      "epoch": 0.621665582303188,
+      "grad_norm": 0.17910513281822205,
+      "learning_rate": 0.0001,
+      "loss": 1.58,
+      "step": 1274
+    },
+    {
+      "epoch": 0.6221535458685752,
+      "grad_norm": 0.19149386882781982,
+      "learning_rate": 0.0001,
+      "loss": 1.7281,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6226415094339622,
+      "grad_norm": 0.1800622195005417,
+      "learning_rate": 0.0001,
+      "loss": 1.7371,
+      "step": 1276
+    },
+    {
+      "epoch": 0.6231294729993494,
+      "grad_norm": 0.19336798787117004,
+      "learning_rate": 0.0001,
+      "loss": 1.7774,
+      "step": 1277
+    },
+    {
+      "epoch": 0.6236174365647364,
+      "grad_norm": 0.18681703507900238,
+      "learning_rate": 0.0001,
+      "loss": 1.7692,
+      "step": 1278
+    },
+    {
+      "epoch": 0.6241054001301236,
+      "grad_norm": 0.1942637413740158,
+      "learning_rate": 0.0001,
+      "loss": 1.6347,
+      "step": 1279
+    },
+    {
+      "epoch": 0.6245933636955108,
+      "grad_norm": 0.18045265972614288,
+      "learning_rate": 0.0001,
+      "loss": 1.6797,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6250813272608978,
+      "grad_norm": 0.20641352236270905,
+      "learning_rate": 0.0001,
+      "loss": 1.669,
+      "step": 1281
+    },
+    {
+      "epoch": 0.625569290826285,
+      "grad_norm": 0.1820315718650818,
+      "learning_rate": 0.0001,
+      "loss": 1.6098,
+      "step": 1282
+    },
+    {
+      "epoch": 0.626057254391672,
+      "grad_norm": 0.1736179143190384,
+      "learning_rate": 0.0001,
+      "loss": 1.4763,
+      "step": 1283
+    },
+    {
+      "epoch": 0.6265452179570592,
+      "grad_norm": 0.18899646401405334,
+      "learning_rate": 0.0001,
+      "loss": 1.6033,
+      "step": 1284
+    },
+    {
+      "epoch": 0.6270331815224464,
+      "grad_norm": 0.18059246242046356,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6275211450878334,
+      "grad_norm": 0.1879289597272873,
+      "learning_rate": 0.0001,
+      "loss": 1.7003,
+      "step": 1286
+    },
+    {
+      "epoch": 0.6280091086532206,
+      "grad_norm": 0.1910688430070877,
+      "learning_rate": 0.0001,
+      "loss": 1.64,
+      "step": 1287
+    },
+    {
+      "epoch": 0.6284970722186076,
+      "grad_norm": 0.18205375969409943,
+      "learning_rate": 0.0001,
+      "loss": 1.6633,
+      "step": 1288
+    },
+    {
+      "epoch": 0.6289850357839948,
+      "grad_norm": 0.1857621818780899,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 1289
+    },
+    {
+      "epoch": 0.629472999349382,
+      "grad_norm": 0.18360133469104767,
+      "learning_rate": 0.0001,
+      "loss": 1.6645,
+      "step": 1290
+    },
+    {
+      "epoch": 0.629960962914769,
+      "grad_norm": 0.1781957596540451,
+      "learning_rate": 0.0001,
+      "loss": 1.6996,
+      "step": 1291
+    },
+    {
+      "epoch": 0.6304489264801562,
+      "grad_norm": 0.16980469226837158,
+      "learning_rate": 0.0001,
+      "loss": 1.5369,
+      "step": 1292
+    },
+    {
+      "epoch": 0.6309368900455433,
+      "grad_norm": 0.19171112775802612,
+      "learning_rate": 0.0001,
+      "loss": 1.7421,
+      "step": 1293
+    },
+    {
+      "epoch": 0.6314248536109304,
+      "grad_norm": 0.1753898411989212,
+      "learning_rate": 0.0001,
+      "loss": 1.67,
+      "step": 1294
+    },
+    {
+      "epoch": 0.6319128171763175,
+      "grad_norm": 0.1746547669172287,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 1295
+    },
+    {
+      "epoch": 0.6324007807417046,
+      "grad_norm": 0.19930396974086761,
+      "learning_rate": 0.0001,
+      "loss": 1.7406,
+      "step": 1296
+    },
+    {
+      "epoch": 0.6328887443070917,
+      "grad_norm": 0.18658863008022308,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 1297
+    },
+    {
+      "epoch": 0.6333767078724789,
+      "grad_norm": 0.19694779813289642,
+      "learning_rate": 0.0001,
+      "loss": 1.6941,
+      "step": 1298
+    },
+    {
+      "epoch": 0.633864671437866,
+      "grad_norm": 0.18741701543331146,
+      "learning_rate": 0.0001,
+      "loss": 1.7258,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6343526350032531,
+      "grad_norm": 0.18276375532150269,
+      "learning_rate": 0.0001,
+      "loss": 1.5375,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6348405985686402,
+      "grad_norm": 0.19398179650306702,
+      "learning_rate": 0.0001,
+      "loss": 1.7516,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6353285621340273,
+      "grad_norm": 0.18019351363182068,
+      "learning_rate": 0.0001,
+      "loss": 1.6373,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6358165256994145,
+      "grad_norm": 0.17721763253211975,
+      "learning_rate": 0.0001,
+      "loss": 1.518,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6363044892648015,
+      "grad_norm": 0.20623503625392914,
+      "learning_rate": 0.0001,
+      "loss": 1.7557,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6367924528301887,
+      "grad_norm": 0.17641960084438324,
+      "learning_rate": 0.0001,
+      "loss": 1.6621,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6372804163955758,
+      "grad_norm": 0.18402546644210815,
+      "learning_rate": 0.0001,
+      "loss": 1.7142,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6377683799609629,
+      "grad_norm": 0.18915514647960663,
+      "learning_rate": 0.0001,
+      "loss": 1.7845,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6382563435263501,
+      "grad_norm": 0.17517401278018951,
+      "learning_rate": 0.0001,
+      "loss": 1.6117,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6387443070917371,
+      "grad_norm": 0.1806933432817459,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6392322706571243,
+      "grad_norm": 0.1815006136894226,
+      "learning_rate": 0.0001,
+      "loss": 1.6428,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6397202342225113,
+      "grad_norm": 0.18762600421905518,
+      "learning_rate": 0.0001,
+      "loss": 1.6741,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6402081977878985,
+      "grad_norm": 0.17323340475559235,
+      "learning_rate": 0.0001,
+      "loss": 1.5812,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6406961613532857,
+      "grad_norm": 0.18929678201675415,
+      "learning_rate": 0.0001,
+      "loss": 1.6747,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6411841249186727,
+      "grad_norm": 0.18279722332954407,
+      "learning_rate": 0.0001,
+      "loss": 1.6429,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6416720884840599,
+      "grad_norm": 0.1740237921476364,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6421600520494469,
+      "grad_norm": 0.18474610149860382,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6426480156148341,
+      "grad_norm": 0.18610845506191254,
+      "learning_rate": 0.0001,
+      "loss": 1.7036,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6431359791802213,
+      "grad_norm": 0.18621689081192017,
+      "learning_rate": 0.0001,
+      "loss": 1.7495,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6436239427456083,
+      "grad_norm": 0.1806156188249588,
+      "learning_rate": 0.0001,
+      "loss": 1.6756,
+      "step": 1319
+    },
+    {
+      "epoch": 0.6441119063109955,
+      "grad_norm": 0.18515653908252716,
+      "learning_rate": 0.0001,
+      "loss": 1.5946,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6445998698763825,
+      "grad_norm": 0.17863605916500092,
+      "learning_rate": 0.0001,
+      "loss": 1.6519,
+      "step": 1321
+    },
+    {
+      "epoch": 0.6450878334417697,
+      "grad_norm": 0.17926158010959625,
+      "learning_rate": 0.0001,
+      "loss": 1.6493,
+      "step": 1322
+    },
+    {
+      "epoch": 0.6455757970071568,
+      "grad_norm": 0.19456753134727478,
+      "learning_rate": 0.0001,
+      "loss": 1.7178,
+      "step": 1323
+    },
+    {
+      "epoch": 0.6460637605725439,
+      "grad_norm": 0.17429687082767487,
+      "learning_rate": 0.0001,
+      "loss": 1.5933,
+      "step": 1324
+    },
+    {
+      "epoch": 0.646551724137931,
+      "grad_norm": 0.18286961317062378,
+      "learning_rate": 0.0001,
+      "loss": 1.7709,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6470396877033181,
+      "grad_norm": 0.18863536417484283,
+      "learning_rate": 0.0001,
+      "loss": 1.707,
+      "step": 1326
+    },
+    {
+      "epoch": 0.6475276512687053,
+      "grad_norm": 0.18341319262981415,
+      "learning_rate": 0.0001,
+      "loss": 1.6812,
+      "step": 1327
+    },
+    {
+      "epoch": 0.6480156148340924,
+      "grad_norm": 0.2063741534948349,
+      "learning_rate": 0.0001,
+      "loss": 1.7651,
+      "step": 1328
+    },
+    {
+      "epoch": 0.6485035783994795,
+      "grad_norm": 0.18247577548027039,
+      "learning_rate": 0.0001,
+      "loss": 1.6966,
+      "step": 1329
+    },
+    {
+      "epoch": 0.6489915419648666,
+      "grad_norm": 0.18382957577705383,
+      "learning_rate": 0.0001,
+      "loss": 1.6692,
+      "step": 1330
+    },
+    {
+      "epoch": 0.6494795055302537,
+      "grad_norm": 0.19317232072353363,
+      "learning_rate": 0.0001,
+      "loss": 1.6594,
+      "step": 1331
+    },
+    {
+      "epoch": 0.6499674690956408,
+      "grad_norm": 0.18904589116573334,
+      "learning_rate": 0.0001,
+      "loss": 1.7119,
+      "step": 1332
+    },
+    {
+      "epoch": 0.650455432661028,
+      "grad_norm": 0.19716934859752655,
+      "learning_rate": 0.0001,
+      "loss": 1.639,
+      "step": 1333
+    },
+    {
+      "epoch": 0.6509433962264151,
+      "grad_norm": 0.18755610287189484,
+      "learning_rate": 0.0001,
+      "loss": 1.7698,
+      "step": 1334
+    },
+    {
+      "epoch": 0.6514313597918022,
+      "grad_norm": 0.18484559655189514,
+      "learning_rate": 0.0001,
+      "loss": 1.6491,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6519193233571894,
+      "grad_norm": 0.1828099489212036,
+      "learning_rate": 0.0001,
+      "loss": 1.7007,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6524072869225764,
+      "grad_norm": 0.1784173548221588,
+      "learning_rate": 0.0001,
+      "loss": 1.7014,
+      "step": 1337
+    },
+    {
+      "epoch": 0.6528952504879636,
+      "grad_norm": 0.18998552858829498,
+      "learning_rate": 0.0001,
+      "loss": 1.7141,
+      "step": 1338
+    },
+    {
+      "epoch": 0.6533832140533506,
+      "grad_norm": 0.18050242960453033,
+      "learning_rate": 0.0001,
+      "loss": 1.5419,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6538711776187378,
+      "grad_norm": 0.17193758487701416,
+      "learning_rate": 0.0001,
+      "loss": 1.6478,
+      "step": 1340
+    },
+    {
+      "epoch": 0.654359141184125,
+      "grad_norm": 0.1884956806898117,
+      "learning_rate": 0.0001,
+      "loss": 1.6398,
+      "step": 1341
+    },
+    {
+      "epoch": 0.654847104749512,
+      "grad_norm": 0.18393546342849731,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6553350683148992,
+      "grad_norm": 0.17768503725528717,
+      "learning_rate": 0.0001,
+      "loss": 1.6551,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6558230318802862,
+      "grad_norm": 0.1898057609796524,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6563109954456734,
+      "grad_norm": 0.18605998158454895,
+      "learning_rate": 0.0001,
+      "loss": 1.6671,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6567989590110606,
+      "grad_norm": 0.17978940904140472,
+      "learning_rate": 0.0001,
+      "loss": 1.6983,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6572869225764476,
+      "grad_norm": 0.18616552650928497,
+      "learning_rate": 0.0001,
+      "loss": 1.7164,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6577748861418348,
+      "grad_norm": 0.1831643134355545,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6582628497072218,
+      "grad_norm": 0.18316183984279633,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 1349
+    },
+    {
+      "epoch": 0.658750813272609,
+      "grad_norm": 0.1927359700202942,
+      "learning_rate": 0.0001,
+      "loss": 1.7889,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6592387768379961,
+      "grad_norm": 0.17629116773605347,
+      "learning_rate": 0.0001,
+      "loss": 1.6653,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6597267404033832,
+      "grad_norm": 0.17519547045230865,
+      "learning_rate": 0.0001,
+      "loss": 1.7317,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6602147039687704,
+      "grad_norm": 0.1796695441007614,
+      "learning_rate": 0.0001,
+      "loss": 1.6736,
+      "step": 1353
+    },
+    {
+      "epoch": 0.6607026675341574,
+      "grad_norm": 0.17477834224700928,
+      "learning_rate": 0.0001,
+      "loss": 1.593,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6611906310995446,
+      "grad_norm": 0.1775176078081131,
+      "learning_rate": 0.0001,
+      "loss": 1.6762,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6616785946649317,
+      "grad_norm": 0.18086951971054077,
+      "learning_rate": 0.0001,
+      "loss": 1.7373,
+      "step": 1356
+    },
+    {
+      "epoch": 0.6621665582303188,
+      "grad_norm": 0.18718330562114716,
+      "learning_rate": 0.0001,
+      "loss": 1.74,
+      "step": 1357
+    },
+    {
+      "epoch": 0.6626545217957059,
+      "grad_norm": 0.19314613938331604,
+      "learning_rate": 0.0001,
+      "loss": 1.7523,
+      "step": 1358
+    },
+    {
+      "epoch": 0.663142485361093,
+      "grad_norm": 0.18622739613056183,
+      "learning_rate": 0.0001,
+      "loss": 1.6772,
+      "step": 1359
+    },
+    {
+      "epoch": 0.6636304489264802,
+      "grad_norm": 0.18634377419948578,
+      "learning_rate": 0.0001,
+      "loss": 1.7391,
+      "step": 1360
+    },
+    {
+      "epoch": 0.6641184124918673,
+      "grad_norm": 0.19122491776943207,
+      "learning_rate": 0.0001,
+      "loss": 1.6955,
+      "step": 1361
+    },
+    {
+      "epoch": 0.6646063760572544,
+      "grad_norm": 0.18408794701099396,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 1362
+    },
+    {
+      "epoch": 0.6650943396226415,
+      "grad_norm": 0.17737893760204315,
+      "learning_rate": 0.0001,
+      "loss": 1.6305,
+      "step": 1363
+    },
+    {
+      "epoch": 0.6655823031880286,
+      "grad_norm": 0.19772779941558838,
+      "learning_rate": 0.0001,
+      "loss": 1.5194,
+      "step": 1364
+    },
+    {
+      "epoch": 0.6660702667534157,
+      "grad_norm": 0.18935418128967285,
+      "learning_rate": 0.0001,
+      "loss": 1.759,
+      "step": 1365
+    },
+    {
+      "epoch": 0.6665582303188029,
+      "grad_norm": 0.1936458796262741,
+      "learning_rate": 0.0001,
+      "loss": 1.672,
+      "step": 1366
+    },
+    {
+      "epoch": 0.66704619388419,
+      "grad_norm": 0.18454033136367798,
+      "learning_rate": 0.0001,
+      "loss": 1.6954,
+      "step": 1367
+    },
+    {
+      "epoch": 0.6675341574495771,
+      "grad_norm": 0.18430058658123016,
+      "learning_rate": 0.0001,
+      "loss": 1.5498,
+      "step": 1368
+    },
+    {
+      "epoch": 0.6680221210149642,
+      "grad_norm": 0.1890435814857483,
+      "learning_rate": 0.0001,
+      "loss": 1.6311,
+      "step": 1369
+    },
+    {
+      "epoch": 0.6685100845803513,
+      "grad_norm": 0.20457454025745392,
+      "learning_rate": 0.0001,
+      "loss": 1.8362,
+      "step": 1370
+    },
+    {
+      "epoch": 0.6689980481457385,
+      "grad_norm": 0.18559999763965607,
+      "learning_rate": 0.0001,
+      "loss": 1.6873,
+      "step": 1371
+    },
+    {
+      "epoch": 0.6694860117111255,
+      "grad_norm": 0.1795533299446106,
+      "learning_rate": 0.0001,
+      "loss": 1.5727,
+      "step": 1372
+    },
+    {
+      "epoch": 0.6699739752765127,
+      "grad_norm": 0.17595787346363068,
+      "learning_rate": 0.0001,
+      "loss": 1.6365,
+      "step": 1373
+    },
+    {
+      "epoch": 0.6704619388418998,
+      "grad_norm": 0.18084342777729034,
+      "learning_rate": 0.0001,
+      "loss": 1.728,
+      "step": 1374
+    },
+    {
+      "epoch": 0.6709499024072869,
+      "grad_norm": 0.18217813968658447,
+      "learning_rate": 0.0001,
+      "loss": 1.5737,
+      "step": 1375
+    },
+    {
+      "epoch": 0.6714378659726741,
+      "grad_norm": 0.19856606423854828,
+      "learning_rate": 0.0001,
+      "loss": 1.7181,
+      "step": 1376
+    },
+    {
+      "epoch": 0.6719258295380611,
+      "grad_norm": 0.18344613909721375,
+      "learning_rate": 0.0001,
+      "loss": 1.6762,
+      "step": 1377
+    },
+    {
+      "epoch": 0.6724137931034483,
+      "grad_norm": 0.1860368400812149,
+      "learning_rate": 0.0001,
+      "loss": 1.7163,
+      "step": 1378
+    },
+    {
+      "epoch": 0.6729017566688354,
+      "grad_norm": 0.1970399022102356,
+      "learning_rate": 0.0001,
+      "loss": 1.6938,
+      "step": 1379
+    },
+    {
+      "epoch": 0.6733897202342225,
+      "grad_norm": 0.18704956769943237,
+      "learning_rate": 0.0001,
+      "loss": 1.68,
+      "step": 1380
+    },
+    {
+      "epoch": 0.6738776837996097,
+      "grad_norm": 0.18241243064403534,
+      "learning_rate": 0.0001,
+      "loss": 1.556,
+      "step": 1381
+    },
+    {
+      "epoch": 0.6743656473649967,
+      "grad_norm": 0.1883855015039444,
+      "learning_rate": 0.0001,
+      "loss": 1.6499,
+      "step": 1382
+    },
+    {
+      "epoch": 0.6748536109303839,
+      "grad_norm": 0.19078750908374786,
+      "learning_rate": 0.0001,
+      "loss": 1.6102,
+      "step": 1383
+    },
+    {
+      "epoch": 0.675341574495771,
+      "grad_norm": 0.18574097752571106,
+      "learning_rate": 0.0001,
+      "loss": 1.6377,
+      "step": 1384
+    },
+    {
+      "epoch": 0.6758295380611581,
+      "grad_norm": 0.18113356828689575,
+      "learning_rate": 0.0001,
+      "loss": 1.6558,
+      "step": 1385
+    },
+    {
+      "epoch": 0.6763175016265452,
+      "grad_norm": 0.1948838084936142,
+      "learning_rate": 0.0001,
+      "loss": 1.6773,
+      "step": 1386
+    },
+    {
+      "epoch": 0.6768054651919323,
+      "grad_norm": 0.18788839876651764,
+      "learning_rate": 0.0001,
+      "loss": 1.6001,
+      "step": 1387
+    },
+    {
+      "epoch": 0.6772934287573195,
+      "grad_norm": 0.18287061154842377,
+      "learning_rate": 0.0001,
+      "loss": 1.6765,
+      "step": 1388
+    },
+    {
+      "epoch": 0.6777813923227066,
+      "grad_norm": 0.1944217085838318,
+      "learning_rate": 0.0001,
+      "loss": 1.7176,
+      "step": 1389
+    },
+    {
+      "epoch": 0.6782693558880937,
+      "grad_norm": 0.19344620406627655,
+      "learning_rate": 0.0001,
+      "loss": 1.7028,
+      "step": 1390
+    },
+    {
+      "epoch": 0.6787573194534808,
+      "grad_norm": 0.18965065479278564,
+      "learning_rate": 0.0001,
+      "loss": 1.6866,
+      "step": 1391
+    },
+    {
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.18736335635185242,
+      "learning_rate": 0.0001,
+      "loss": 1.6367,
+      "step": 1392
+    },
+    {
+      "epoch": 0.679733246584255,
+      "grad_norm": 0.19293658435344696,
+      "learning_rate": 0.0001,
+      "loss": 1.6772,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6802212101496422,
+      "grad_norm": 0.1829851269721985,
+      "learning_rate": 0.0001,
+      "loss": 1.5907,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6807091737150293,
+      "grad_norm": 0.18881766498088837,
+      "learning_rate": 0.0001,
+      "loss": 1.6568,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6811971372804164,
+      "grad_norm": 0.1797613501548767,
+      "learning_rate": 0.0001,
+      "loss": 1.7498,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6816851008458035,
+      "grad_norm": 0.19345468282699585,
+      "learning_rate": 0.0001,
+      "loss": 1.7604,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6821730644111906,
+      "grad_norm": 0.1797412633895874,
+      "learning_rate": 0.0001,
+      "loss": 1.6099,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6826610279765778,
+      "grad_norm": 0.17828069627285004,
+      "learning_rate": 0.0001,
+      "loss": 1.5906,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6831489915419648,
+      "grad_norm": 0.17425017058849335,
+      "learning_rate": 0.0001,
+      "loss": 1.655,
+      "step": 1400
+    },
+    {
+      "epoch": 0.683636955107352,
+      "grad_norm": 0.1832888275384903,
+      "learning_rate": 0.0001,
+      "loss": 1.6294,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6841249186727391,
+      "grad_norm": 0.17417742311954498,
+      "learning_rate": 0.0001,
+      "loss": 1.6001,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6846128822381262,
+      "grad_norm": 0.17659293115139008,
+      "learning_rate": 0.0001,
+      "loss": 1.6074,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6851008458035134,
+      "grad_norm": 0.18741555511951447,
+      "learning_rate": 0.0001,
+      "loss": 1.6052,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6855888093689004,
+      "grad_norm": 0.20556053519248962,
+      "learning_rate": 0.0001,
+      "loss": 1.8142,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6860767729342876,
+      "grad_norm": 0.1742892563343048,
+      "learning_rate": 0.0001,
+      "loss": 1.5287,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6865647364996746,
+      "grad_norm": 0.17847485840320587,
+      "learning_rate": 0.0001,
+      "loss": 1.5856,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6870527000650618,
+      "grad_norm": 0.17445972561836243,
+      "learning_rate": 0.0001,
+      "loss": 1.6276,
+      "step": 1408
+    },
+    {
+      "epoch": 0.687540663630449,
+      "grad_norm": 0.20477156341075897,
+      "learning_rate": 0.0001,
+      "loss": 1.8143,
+      "step": 1409
+    },
+    {
+      "epoch": 0.688028627195836,
+      "grad_norm": 0.19618012011051178,
+      "learning_rate": 0.0001,
+      "loss": 1.7586,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6885165907612232,
+      "grad_norm": 0.18680322170257568,
+      "learning_rate": 0.0001,
+      "loss": 1.5995,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6890045543266102,
+      "grad_norm": 0.18768328428268433,
+      "learning_rate": 0.0001,
+      "loss": 1.6376,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6894925178919974,
+      "grad_norm": 0.1906110793352127,
+      "learning_rate": 0.0001,
+      "loss": 1.6941,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6899804814573846,
+      "grad_norm": 0.182110995054245,
+      "learning_rate": 0.0001,
+      "loss": 1.7078,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6904684450227716,
+      "grad_norm": 0.182440385222435,
+      "learning_rate": 0.0001,
+      "loss": 1.5876,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6909564085881588,
+      "grad_norm": 0.1977446973323822,
+      "learning_rate": 0.0001,
+      "loss": 1.7053,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6914443721535458,
+      "grad_norm": 0.18622663617134094,
+      "learning_rate": 0.0001,
+      "loss": 1.6786,
+      "step": 1417
+    },
+    {
+      "epoch": 0.691932335718933,
+      "grad_norm": 0.1755010485649109,
+      "learning_rate": 0.0001,
+      "loss": 1.6419,
+      "step": 1418
+    },
+    {
+      "epoch": 0.6924202992843201,
+      "grad_norm": 0.18327659368515015,
+      "learning_rate": 0.0001,
+      "loss": 1.6468,
+      "step": 1419
+    },
+    {
+      "epoch": 0.6929082628497072,
+      "grad_norm": 0.19392666220664978,
+      "learning_rate": 0.0001,
+      "loss": 1.7399,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6933962264150944,
+      "grad_norm": 0.17711861431598663,
+      "learning_rate": 0.0001,
+      "loss": 1.5096,
+      "step": 1421
+    },
+    {
+      "epoch": 0.6938841899804814,
+      "grad_norm": 0.19558769464492798,
+      "learning_rate": 0.0001,
+      "loss": 1.6734,
+      "step": 1422
+    },
+    {
+      "epoch": 0.6943721535458686,
+      "grad_norm": 0.18583737313747406,
+      "learning_rate": 0.0001,
+      "loss": 1.5908,
+      "step": 1423
+    },
+    {
+      "epoch": 0.6948601171112557,
+      "grad_norm": 0.19811202585697174,
+      "learning_rate": 0.0001,
+      "loss": 1.685,
+      "step": 1424
+    },
+    {
+      "epoch": 0.6953480806766428,
+      "grad_norm": 0.1776018738746643,
+      "learning_rate": 0.0001,
+      "loss": 1.5341,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6958360442420299,
+      "grad_norm": 0.18943732976913452,
+      "learning_rate": 0.0001,
+      "loss": 1.7079,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6963240078074171,
+      "grad_norm": 0.1881314069032669,
+      "learning_rate": 0.0001,
+      "loss": 1.7115,
+      "step": 1427
+    },
+    {
+      "epoch": 0.6968119713728042,
+      "grad_norm": 0.17805875837802887,
+      "learning_rate": 0.0001,
+      "loss": 1.7172,
+      "step": 1428
+    },
+    {
+      "epoch": 0.6972999349381913,
+      "grad_norm": 0.18958143889904022,
+      "learning_rate": 0.0001,
+      "loss": 1.6772,
+      "step": 1429
+    },
+    {
+      "epoch": 0.6977878985035784,
+      "grad_norm": 0.18394924700260162,
+      "learning_rate": 0.0001,
+      "loss": 1.7144,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6982758620689655,
+      "grad_norm": 0.1808011531829834,
+      "learning_rate": 0.0001,
+      "loss": 1.6236,
+      "step": 1431
+    },
+    {
+      "epoch": 0.6987638256343527,
+      "grad_norm": 0.19299408793449402,
+      "learning_rate": 0.0001,
+      "loss": 1.7708,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6992517891997397,
+      "grad_norm": 0.18791189789772034,
+      "learning_rate": 0.0001,
+      "loss": 1.6471,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6997397527651269,
+      "grad_norm": 0.18509036302566528,
+      "learning_rate": 0.0001,
+      "loss": 1.6443,
+      "step": 1434
+    },
+    {
+      "epoch": 0.700227716330514,
+      "grad_norm": 0.181147500872612,
+      "learning_rate": 0.0001,
+      "loss": 1.6402,
+      "step": 1435
+    },
+    {
+      "epoch": 0.7007156798959011,
+      "grad_norm": 0.18755219876766205,
+      "learning_rate": 0.0001,
+      "loss": 1.616,
+      "step": 1436
+    },
+    {
+      "epoch": 0.7012036434612883,
+      "grad_norm": 0.1845817118883133,
+      "learning_rate": 0.0001,
+      "loss": 1.6785,
+      "step": 1437
+    },
+    {
+      "epoch": 0.7016916070266753,
+      "grad_norm": 0.1759112924337387,
+      "learning_rate": 0.0001,
+      "loss": 1.5915,
+      "step": 1438
+    },
+    {
+      "epoch": 0.7021795705920625,
+      "grad_norm": 0.19621609151363373,
+      "learning_rate": 0.0001,
+      "loss": 1.7141,
+      "step": 1439
+    },
+    {
+      "epoch": 0.7026675341574495,
+      "grad_norm": 0.19262106716632843,
+      "learning_rate": 0.0001,
+      "loss": 1.6559,
+      "step": 1440
+    },
+    {
+      "epoch": 0.7031554977228367,
+      "grad_norm": 0.18579107522964478,
+      "learning_rate": 0.0001,
+      "loss": 1.7255,
+      "step": 1441
+    },
+    {
+      "epoch": 0.7036434612882239,
+      "grad_norm": 0.1896117478609085,
+      "learning_rate": 0.0001,
+      "loss": 1.6592,
+      "step": 1442
+    },
+    {
+      "epoch": 0.7041314248536109,
+      "grad_norm": 0.19127638638019562,
+      "learning_rate": 0.0001,
+      "loss": 1.6569,
+      "step": 1443
+    },
+    {
+      "epoch": 0.7046193884189981,
+      "grad_norm": 0.18615250289440155,
+      "learning_rate": 0.0001,
+      "loss": 1.6458,
+      "step": 1444
+    },
+    {
+      "epoch": 0.7051073519843851,
+      "grad_norm": 0.20700882375240326,
+      "learning_rate": 0.0001,
+      "loss": 1.7548,
+      "step": 1445
+    },
+    {
+      "epoch": 0.7055953155497723,
+      "grad_norm": 0.18569111824035645,
+      "learning_rate": 0.0001,
+      "loss": 1.6141,
+      "step": 1446
+    },
+    {
+      "epoch": 0.7060832791151594,
+      "grad_norm": 0.1929357945919037,
+      "learning_rate": 0.0001,
+      "loss": 1.6579,
+      "step": 1447
+    },
+    {
+      "epoch": 0.7065712426805465,
+      "grad_norm": 0.19292321801185608,
+      "learning_rate": 0.0001,
+      "loss": 1.7026,
+      "step": 1448
+    },
+    {
+      "epoch": 0.7070592062459337,
+      "grad_norm": 0.1860547810792923,
+      "learning_rate": 0.0001,
+      "loss": 1.7159,
+      "step": 1449
+    },
+    {
+      "epoch": 0.7075471698113207,
+      "grad_norm": 0.18196798861026764,
+      "learning_rate": 0.0001,
+      "loss": 1.6683,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7080351333767079,
+      "grad_norm": 0.1784311681985855,
+      "learning_rate": 0.0001,
+      "loss": 1.591,
+      "step": 1451
+    },
+    {
+      "epoch": 0.708523096942095,
+      "grad_norm": 0.1859826296567917,
+      "learning_rate": 0.0001,
+      "loss": 1.686,
+      "step": 1452
+    },
+    {
+      "epoch": 0.7090110605074821,
+      "grad_norm": 0.18661972880363464,
+      "learning_rate": 0.0001,
+      "loss": 1.6035,
+      "step": 1453
+    },
+    {
+      "epoch": 0.7094990240728692,
+      "grad_norm": 0.18923242390155792,
+      "learning_rate": 0.0001,
+      "loss": 1.7313,
+      "step": 1454
+    },
+    {
+      "epoch": 0.7099869876382563,
+      "grad_norm": 0.1933595836162567,
+      "learning_rate": 0.0001,
+      "loss": 1.7847,
+      "step": 1455
+    },
+    {
+      "epoch": 0.7104749512036435,
+      "grad_norm": 0.19189570844173431,
+      "learning_rate": 0.0001,
+      "loss": 1.6822,
+      "step": 1456
+    },
+    {
+      "epoch": 0.7109629147690306,
+      "grad_norm": 0.18658806383609772,
+      "learning_rate": 0.0001,
+      "loss": 1.7259,
+      "step": 1457
+    },
+    {
+      "epoch": 0.7114508783344177,
+      "grad_norm": 0.19276390969753265,
+      "learning_rate": 0.0001,
+      "loss": 1.7137,
+      "step": 1458
+    },
+    {
+      "epoch": 0.7119388418998048,
+      "grad_norm": 0.19343958795070648,
+      "learning_rate": 0.0001,
+      "loss": 1.85,
+      "step": 1459
+    },
+    {
+      "epoch": 0.7124268054651919,
+      "grad_norm": 0.18348975479602814,
+      "learning_rate": 0.0001,
+      "loss": 1.6551,
+      "step": 1460
+    },
+    {
+      "epoch": 0.712914769030579,
+      "grad_norm": 0.18842749297618866,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 1461
+    },
+    {
+      "epoch": 0.7134027325959662,
+      "grad_norm": 0.1927134394645691,
+      "learning_rate": 0.0001,
+      "loss": 1.7096,
+      "step": 1462
+    },
+    {
+      "epoch": 0.7138906961613533,
+      "grad_norm": 0.18802672624588013,
+      "learning_rate": 0.0001,
+      "loss": 1.6056,
+      "step": 1463
+    },
+    {
+      "epoch": 0.7143786597267404,
+      "grad_norm": 0.19619612395763397,
+      "learning_rate": 0.0001,
+      "loss": 1.774,
+      "step": 1464
+    },
+    {
+      "epoch": 0.7148666232921275,
+      "grad_norm": 0.18309350311756134,
+      "learning_rate": 0.0001,
+      "loss": 1.7173,
+      "step": 1465
+    },
+    {
+      "epoch": 0.7153545868575146,
+      "grad_norm": 0.190412700176239,
+      "learning_rate": 0.0001,
+      "loss": 1.6976,
+      "step": 1466
+    },
+    {
+      "epoch": 0.7158425504229018,
+      "grad_norm": 0.1945875585079193,
+      "learning_rate": 0.0001,
+      "loss": 1.6,
+      "step": 1467
+    },
+    {
+      "epoch": 0.7163305139882888,
+      "grad_norm": 0.19393402338027954,
+      "learning_rate": 0.0001,
+      "loss": 1.6569,
+      "step": 1468
+    },
+    {
+      "epoch": 0.716818477553676,
+      "grad_norm": 0.19331271946430206,
+      "learning_rate": 0.0001,
+      "loss": 1.6866,
+      "step": 1469
+    },
+    {
+      "epoch": 0.7173064411190632,
+      "grad_norm": 0.18295609951019287,
+      "learning_rate": 0.0001,
+      "loss": 1.6991,
+      "step": 1470
+    },
+    {
+      "epoch": 0.7177944046844502,
+      "grad_norm": 0.1889050453901291,
+      "learning_rate": 0.0001,
+      "loss": 1.6893,
+      "step": 1471
+    },
+    {
+      "epoch": 0.7182823682498374,
+      "grad_norm": 0.18971437215805054,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 1472
+    },
+    {
+      "epoch": 0.7187703318152244,
+      "grad_norm": 0.19291648268699646,
+      "learning_rate": 0.0001,
+      "loss": 1.5201,
+      "step": 1473
+    },
+    {
+      "epoch": 0.7192582953806116,
+      "grad_norm": 0.18243971467018127,
+      "learning_rate": 0.0001,
+      "loss": 1.6197,
+      "step": 1474
+    },
+    {
+      "epoch": 0.7197462589459988,
+      "grad_norm": 0.2129974663257599,
+      "learning_rate": 0.0001,
+      "loss": 1.6374,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7202342225113858,
+      "grad_norm": 0.23039822280406952,
+      "learning_rate": 0.0001,
+      "loss": 1.7185,
+      "step": 1476
+    },
+    {
+      "epoch": 0.720722186076773,
+      "grad_norm": 0.19735904037952423,
+      "learning_rate": 0.0001,
+      "loss": 1.7196,
+      "step": 1477
+    },
+    {
+      "epoch": 0.72121014964216,
+      "grad_norm": 0.19186371564865112,
+      "learning_rate": 0.0001,
+      "loss": 1.6343,
+      "step": 1478
+    },
+    {
+      "epoch": 0.7216981132075472,
+      "grad_norm": 0.1926344484090805,
+      "learning_rate": 0.0001,
+      "loss": 1.5566,
+      "step": 1479
+    },
+    {
+      "epoch": 0.7221860767729343,
+      "grad_norm": 0.18684318661689758,
+      "learning_rate": 0.0001,
+      "loss": 1.6809,
+      "step": 1480
+    },
+    {
+      "epoch": 0.7226740403383214,
+      "grad_norm": 0.17722366750240326,
+      "learning_rate": 0.0001,
+      "loss": 1.5618,
+      "step": 1481
+    },
+    {
+      "epoch": 0.7231620039037086,
+      "grad_norm": 0.17876088619232178,
+      "learning_rate": 0.0001,
+      "loss": 1.5692,
+      "step": 1482
+    },
+    {
+      "epoch": 0.7236499674690956,
+      "grad_norm": 0.19231794774532318,
+      "learning_rate": 0.0001,
+      "loss": 1.7443,
+      "step": 1483
+    },
+    {
+      "epoch": 0.7241379310344828,
+      "grad_norm": 0.19395712018013,
+      "learning_rate": 0.0001,
+      "loss": 1.6295,
+      "step": 1484
+    },
+    {
+      "epoch": 0.7246258945998699,
+      "grad_norm": 0.18045175075531006,
+      "learning_rate": 0.0001,
+      "loss": 1.636,
+      "step": 1485
+    },
+    {
+      "epoch": 0.725113858165257,
+      "grad_norm": 0.19451959431171417,
+      "learning_rate": 0.0001,
+      "loss": 1.7643,
+      "step": 1486
+    },
+    {
+      "epoch": 0.7256018217306441,
+      "grad_norm": 0.18595635890960693,
+      "learning_rate": 0.0001,
+      "loss": 1.7065,
+      "step": 1487
+    },
+    {
+      "epoch": 0.7260897852960312,
+      "grad_norm": 0.1827252060174942,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 1488
+    },
+    {
+      "epoch": 0.7265777488614183,
+      "grad_norm": 0.18135575950145721,
+      "learning_rate": 0.0001,
+      "loss": 1.5711,
+      "step": 1489
+    },
+    {
+      "epoch": 0.7270657124268055,
+      "grad_norm": 0.17621076107025146,
+      "learning_rate": 0.0001,
+      "loss": 1.4973,
+      "step": 1490
+    },
+    {
+      "epoch": 0.7275536759921926,
+      "grad_norm": 0.17968687415122986,
+      "learning_rate": 0.0001,
+      "loss": 1.5858,
+      "step": 1491
+    },
+    {
+      "epoch": 0.7280416395575797,
+      "grad_norm": 0.19981656968593597,
+      "learning_rate": 0.0001,
+      "loss": 1.6636,
+      "step": 1492
+    },
+    {
+      "epoch": 0.7285296031229668,
+      "grad_norm": 0.18831577897071838,
+      "learning_rate": 0.0001,
+      "loss": 1.6519,
+      "step": 1493
+    },
+    {
+      "epoch": 0.7290175666883539,
+      "grad_norm": 0.18130916357040405,
+      "learning_rate": 0.0001,
+      "loss": 1.7075,
+      "step": 1494
+    },
+    {
+      "epoch": 0.7295055302537411,
+      "grad_norm": 0.18603913486003876,
+      "learning_rate": 0.0001,
+      "loss": 1.6399,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7299934938191281,
+      "grad_norm": 0.18082112073898315,
+      "learning_rate": 0.0001,
+      "loss": 1.5542,
+      "step": 1496
+    },
+    {
+      "epoch": 0.7304814573845153,
+      "grad_norm": 0.19021904468536377,
+      "learning_rate": 0.0001,
+      "loss": 1.6491,
+      "step": 1497
+    },
+    {
+      "epoch": 0.7309694209499024,
+      "grad_norm": 0.18982046842575073,
+      "learning_rate": 0.0001,
+      "loss": 1.7551,
+      "step": 1498
+    },
+    {
+      "epoch": 0.7314573845152895,
+      "grad_norm": 0.17388087511062622,
+      "learning_rate": 0.0001,
+      "loss": 1.5713,
+      "step": 1499
+    },
+    {
+      "epoch": 0.7319453480806767,
+      "grad_norm": 0.18898619711399078,
+      "learning_rate": 0.0001,
+      "loss": 1.6261,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7324333116460637,
+      "grad_norm": 0.1856594830751419,
+      "learning_rate": 0.0001,
+      "loss": 1.7416,
+      "step": 1501
+    },
+    {
+      "epoch": 0.7329212752114509,
+      "grad_norm": 0.1952165812253952,
+      "learning_rate": 0.0001,
+      "loss": 1.8623,
+      "step": 1502
+    },
+    {
+      "epoch": 0.733409238776838,
+      "grad_norm": 0.18985500931739807,
+      "learning_rate": 0.0001,
+      "loss": 1.6428,
+      "step": 1503
+    },
+    {
+      "epoch": 0.7338972023422251,
+      "grad_norm": 0.19051025807857513,
+      "learning_rate": 0.0001,
+      "loss": 1.6933,
+      "step": 1504
+    },
+    {
+      "epoch": 0.7343851659076123,
+      "grad_norm": 0.19088402390480042,
+      "learning_rate": 0.0001,
+      "loss": 1.7145,
+      "step": 1505
+    },
+    {
+      "epoch": 0.7348731294729993,
+      "grad_norm": 0.20936012268066406,
+      "learning_rate": 0.0001,
+      "loss": 1.854,
+      "step": 1506
+    },
+    {
+      "epoch": 0.7353610930383865,
+      "grad_norm": 0.17536798119544983,
+      "learning_rate": 0.0001,
+      "loss": 1.6205,
+      "step": 1507
+    },
+    {
+      "epoch": 0.7358490566037735,
+      "grad_norm": 0.1752844750881195,
+      "learning_rate": 0.0001,
+      "loss": 1.6902,
+      "step": 1508
+    },
+    {
+      "epoch": 0.7363370201691607,
+      "grad_norm": 0.17726869881153107,
+      "learning_rate": 0.0001,
+      "loss": 1.6562,
+      "step": 1509
+    },
+    {
+      "epoch": 0.7368249837345479,
+      "grad_norm": 0.17950955033302307,
+      "learning_rate": 0.0001,
+      "loss": 1.6322,
+      "step": 1510
+    },
+    {
+      "epoch": 0.7373129472999349,
+      "grad_norm": 0.18425075709819794,
+      "learning_rate": 0.0001,
+      "loss": 1.6205,
+      "step": 1511
+    },
+    {
+      "epoch": 0.7378009108653221,
+      "grad_norm": 0.18774688243865967,
+      "learning_rate": 0.0001,
+      "loss": 1.6976,
+      "step": 1512
+    },
+    {
+      "epoch": 0.7382888744307091,
+      "grad_norm": 0.18913634121418,
+      "learning_rate": 0.0001,
+      "loss": 1.6605,
+      "step": 1513
+    },
+    {
+      "epoch": 0.7387768379960963,
+      "grad_norm": 0.18106456100940704,
+      "learning_rate": 0.0001,
+      "loss": 1.7201,
+      "step": 1514
+    },
+    {
+      "epoch": 0.7392648015614834,
+      "grad_norm": 0.1875046044588089,
+      "learning_rate": 0.0001,
+      "loss": 1.7115,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7397527651268705,
+      "grad_norm": 0.1848473697900772,
+      "learning_rate": 0.0001,
+      "loss": 1.6694,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7402407286922577,
+      "grad_norm": 0.18459069728851318,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7407286922576448,
+      "grad_norm": 0.1830248087644577,
+      "learning_rate": 0.0001,
+      "loss": 1.6973,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7412166558230319,
+      "grad_norm": 0.18021319806575775,
+      "learning_rate": 0.0001,
+      "loss": 1.7161,
+      "step": 1519
+    },
+    {
+      "epoch": 0.741704619388419,
+      "grad_norm": 0.1852198839187622,
+      "learning_rate": 0.0001,
+      "loss": 1.6411,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7421925829538061,
+      "grad_norm": 0.18931609392166138,
+      "learning_rate": 0.0001,
+      "loss": 1.7385,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7426805465191932,
+      "grad_norm": 0.18388409912586212,
+      "learning_rate": 0.0001,
+      "loss": 1.6473,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7431685100845804,
+      "grad_norm": 0.19164806604385376,
+      "learning_rate": 0.0001,
+      "loss": 1.7588,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7436564736499675,
+      "grad_norm": 0.19370153546333313,
+      "learning_rate": 0.0001,
+      "loss": 1.6054,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7441444372153546,
+      "grad_norm": 0.17590291798114777,
+      "learning_rate": 0.0001,
+      "loss": 1.6625,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7446324007807417,
+      "grad_norm": 0.18713389337062836,
+      "learning_rate": 0.0001,
+      "loss": 1.6697,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7451203643461288,
+      "grad_norm": 0.18523730337619781,
+      "learning_rate": 0.0001,
+      "loss": 1.613,
+      "step": 1527
+    },
+    {
+      "epoch": 0.745608327911516,
+      "grad_norm": 0.1839686632156372,
+      "learning_rate": 0.0001,
+      "loss": 1.6729,
+      "step": 1528
+    },
+    {
+      "epoch": 0.746096291476903,
+      "grad_norm": 0.20028969645500183,
+      "learning_rate": 0.0001,
+      "loss": 1.6504,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7465842550422902,
+      "grad_norm": 0.18220870196819305,
+      "learning_rate": 0.0001,
+      "loss": 1.6921,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7470722186076773,
+      "grad_norm": 0.18175910413265228,
+      "learning_rate": 0.0001,
+      "loss": 1.6284,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7475601821730644,
+      "grad_norm": 0.18016168475151062,
+      "learning_rate": 0.0001,
+      "loss": 1.5405,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7480481457384516,
+      "grad_norm": 0.1960187703371048,
+      "learning_rate": 0.0001,
+      "loss": 1.6983,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7485361093038386,
+      "grad_norm": 0.1788274049758911,
+      "learning_rate": 0.0001,
+      "loss": 1.6387,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7490240728692258,
+      "grad_norm": 0.19441407918930054,
+      "learning_rate": 0.0001,
+      "loss": 1.6093,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7495120364346128,
+      "grad_norm": 0.19135618209838867,
+      "learning_rate": 0.0001,
+      "loss": 1.6645,
+      "step": 1536
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1894136369228363,
+      "learning_rate": 0.0001,
+      "loss": 1.618,
+      "step": 1537
+    },
+    {
+      "epoch": 0.7504879635653872,
+      "grad_norm": 0.1781785488128662,
+      "learning_rate": 0.0001,
+      "loss": 1.6687,
+      "step": 1538
+    },
+    {
+      "epoch": 0.7509759271307742,
+      "grad_norm": 0.18362712860107422,
+      "learning_rate": 0.0001,
+      "loss": 1.7104,
+      "step": 1539
+    },
+    {
+      "epoch": 0.7514638906961614,
+      "grad_norm": 0.20387201011180878,
+      "learning_rate": 0.0001,
+      "loss": 1.7899,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7519518542615484,
+      "grad_norm": 0.18107970058918,
+      "learning_rate": 0.0001,
+      "loss": 1.5952,
+      "step": 1541
+    },
+    {
+      "epoch": 0.7524398178269356,
+      "grad_norm": 0.18185783922672272,
+      "learning_rate": 0.0001,
+      "loss": 1.6294,
+      "step": 1542
+    },
+    {
+      "epoch": 0.7529277813923227,
+      "grad_norm": 0.17853549122810364,
+      "learning_rate": 0.0001,
+      "loss": 1.5997,
+      "step": 1543
+    },
+    {
+      "epoch": 0.7534157449577098,
+      "grad_norm": 0.17627951502799988,
+      "learning_rate": 0.0001,
+      "loss": 1.5845,
+      "step": 1544
+    },
+    {
+      "epoch": 0.753903708523097,
+      "grad_norm": 0.19762729108333588,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 1545
+    },
+    {
+      "epoch": 0.754391672088484,
+      "grad_norm": 0.20241263508796692,
+      "learning_rate": 0.0001,
+      "loss": 1.794,
+      "step": 1546
+    },
+    {
+      "epoch": 0.7548796356538712,
+      "grad_norm": 0.1798173040151596,
+      "learning_rate": 0.0001,
+      "loss": 1.6637,
+      "step": 1547
+    },
+    {
+      "epoch": 0.7553675992192583,
+      "grad_norm": 0.1928299218416214,
+      "learning_rate": 0.0001,
+      "loss": 1.7751,
+      "step": 1548
+    },
+    {
+      "epoch": 0.7558555627846454,
+      "grad_norm": 0.18737445771694183,
+      "learning_rate": 0.0001,
+      "loss": 1.76,
+      "step": 1549
+    },
+    {
+      "epoch": 0.7563435263500325,
+      "grad_norm": 0.1899656057357788,
+      "learning_rate": 0.0001,
+      "loss": 1.7796,
+      "step": 1550
+    },
+    {
+      "epoch": 0.7568314899154196,
+      "grad_norm": 0.18091318011283875,
+      "learning_rate": 0.0001,
+      "loss": 1.6277,
+      "step": 1551
+    },
+    {
+      "epoch": 0.7573194534808068,
+      "grad_norm": 0.18001140654087067,
+      "learning_rate": 0.0001,
+      "loss": 1.5968,
+      "step": 1552
+    },
+    {
+      "epoch": 0.7578074170461939,
+      "grad_norm": 0.17739026248455048,
+      "learning_rate": 0.0001,
+      "loss": 1.5413,
+      "step": 1553
+    },
+    {
+      "epoch": 0.758295380611581,
+      "grad_norm": 0.183801531791687,
+      "learning_rate": 0.0001,
+      "loss": 1.6924,
+      "step": 1554
+    },
+    {
+      "epoch": 0.7587833441769681,
+      "grad_norm": 0.19029074907302856,
+      "learning_rate": 0.0001,
+      "loss": 1.7264,
+      "step": 1555
+    },
+    {
+      "epoch": 0.7592713077423552,
+      "grad_norm": 0.18088550865650177,
+      "learning_rate": 0.0001,
+      "loss": 1.648,
+      "step": 1556
+    },
+    {
+      "epoch": 0.7597592713077423,
+      "grad_norm": 0.182144433259964,
+      "learning_rate": 0.0001,
+      "loss": 1.6971,
+      "step": 1557
+    },
+    {
+      "epoch": 0.7602472348731295,
+      "grad_norm": 0.1943800002336502,
+      "learning_rate": 0.0001,
+      "loss": 1.8083,
+      "step": 1558
+    },
+    {
+      "epoch": 0.7607351984385166,
+      "grad_norm": 0.19364763796329498,
+      "learning_rate": 0.0001,
+      "loss": 1.8159,
+      "step": 1559
+    },
+    {
+      "epoch": 0.7612231620039037,
+      "grad_norm": 0.19187410175800323,
+      "learning_rate": 0.0001,
+      "loss": 1.6442,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7617111255692909,
+      "grad_norm": 0.1859988272190094,
+      "learning_rate": 0.0001,
+      "loss": 1.643,
+      "step": 1561
+    },
+    {
+      "epoch": 0.7621990891346779,
+      "grad_norm": 0.18581219017505646,
+      "learning_rate": 0.0001,
+      "loss": 1.6164,
+      "step": 1562
+    },
+    {
+      "epoch": 0.7626870527000651,
+      "grad_norm": 0.1905384510755539,
+      "learning_rate": 0.0001,
+      "loss": 1.6672,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7631750162654521,
+      "grad_norm": 0.19391977787017822,
+      "learning_rate": 0.0001,
+      "loss": 1.7094,
+      "step": 1564
+    },
+    {
+      "epoch": 0.7636629798308393,
+      "grad_norm": 0.18060721457004547,
+      "learning_rate": 0.0001,
+      "loss": 1.5947,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7641509433962265,
+      "grad_norm": 0.19189974665641785,
+      "learning_rate": 0.0001,
+      "loss": 1.6929,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7646389069616135,
+      "grad_norm": 0.1866428405046463,
+      "learning_rate": 0.0001,
+      "loss": 1.6223,
+      "step": 1567
+    },
+    {
+      "epoch": 0.7651268705270007,
+      "grad_norm": 0.1927812546491623,
+      "learning_rate": 0.0001,
+      "loss": 1.6787,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7656148340923877,
+      "grad_norm": 0.1899099200963974,
+      "learning_rate": 0.0001,
+      "loss": 1.7085,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7661027976577749,
+      "grad_norm": 0.19770196080207825,
+      "learning_rate": 0.0001,
+      "loss": 1.6935,
+      "step": 1570
+    },
+    {
+      "epoch": 0.766590761223162,
+      "grad_norm": 0.1968637853860855,
+      "learning_rate": 0.0001,
+      "loss": 1.6972,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7670787247885491,
+      "grad_norm": 0.19224700331687927,
+      "learning_rate": 0.0001,
+      "loss": 1.6879,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7675666883539363,
+      "grad_norm": 0.18638695776462555,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7680546519193233,
+      "grad_norm": 0.18514102697372437,
+      "learning_rate": 0.0001,
+      "loss": 1.6404,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7685426154847105,
+      "grad_norm": 0.1898544877767563,
+      "learning_rate": 0.0001,
+      "loss": 1.6941,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7690305790500976,
+      "grad_norm": 0.1941719949245453,
+      "learning_rate": 0.0001,
+      "loss": 1.8447,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7695185426154847,
+      "grad_norm": 0.19126634299755096,
+      "learning_rate": 0.0001,
+      "loss": 1.7318,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7700065061808719,
+      "grad_norm": 0.2004380077123642,
+      "learning_rate": 0.0001,
+      "loss": 1.7452,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7704944697462589,
+      "grad_norm": 0.18317647278308868,
+      "learning_rate": 0.0001,
+      "loss": 1.6582,
+      "step": 1579
+    },
+    {
+      "epoch": 0.7709824333116461,
+      "grad_norm": 0.17985877394676208,
+      "learning_rate": 0.0001,
+      "loss": 1.5736,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7714703968770332,
+      "grad_norm": 0.19441930949687958,
+      "learning_rate": 0.0001,
+      "loss": 1.6716,
+      "step": 1581
+    },
+    {
+      "epoch": 0.7719583604424203,
+      "grad_norm": 0.18969953060150146,
+      "learning_rate": 0.0001,
+      "loss": 1.6254,
+      "step": 1582
+    },
+    {
+      "epoch": 0.7724463240078074,
+      "grad_norm": 0.19708946347236633,
+      "learning_rate": 0.0001,
+      "loss": 1.7267,
+      "step": 1583
+    },
+    {
+      "epoch": 0.7729342875731945,
+      "grad_norm": 0.17867428064346313,
+      "learning_rate": 0.0001,
+      "loss": 1.5539,
+      "step": 1584
+    },
+    {
+      "epoch": 0.7734222511385817,
+      "grad_norm": 0.1882658153772354,
+      "learning_rate": 0.0001,
+      "loss": 1.6716,
+      "step": 1585
+    },
+    {
+      "epoch": 0.7739102147039688,
+      "grad_norm": 0.19716131687164307,
+      "learning_rate": 0.0001,
+      "loss": 1.7124,
+      "step": 1586
+    },
+    {
+      "epoch": 0.7743981782693559,
+      "grad_norm": 0.1901741474866867,
+      "learning_rate": 0.0001,
+      "loss": 1.5683,
+      "step": 1587
+    },
+    {
+      "epoch": 0.774886141834743,
+      "grad_norm": 0.1997768133878708,
+      "learning_rate": 0.0001,
+      "loss": 1.7426,
+      "step": 1588
+    },
+    {
+      "epoch": 0.7753741054001301,
+      "grad_norm": 0.20431944727897644,
+      "learning_rate": 0.0001,
+      "loss": 1.6736,
+      "step": 1589
+    },
+    {
+      "epoch": 0.7758620689655172,
+      "grad_norm": 0.19220104813575745,
+      "learning_rate": 0.0001,
+      "loss": 1.613,
+      "step": 1590
+    },
+    {
+      "epoch": 0.7763500325309044,
+      "grad_norm": 0.21191276609897614,
+      "learning_rate": 0.0001,
+      "loss": 1.7514,
+      "step": 1591
+    },
+    {
+      "epoch": 0.7768379960962914,
+      "grad_norm": 0.1914515644311905,
+      "learning_rate": 0.0001,
+      "loss": 1.6661,
+      "step": 1592
+    },
+    {
+      "epoch": 0.7773259596616786,
+      "grad_norm": 0.20354917645454407,
+      "learning_rate": 0.0001,
+      "loss": 1.623,
+      "step": 1593
+    },
+    {
+      "epoch": 0.7778139232270657,
+      "grad_norm": 0.18612132966518402,
+      "learning_rate": 0.0001,
+      "loss": 1.654,
+      "step": 1594
+    },
+    {
+      "epoch": 0.7783018867924528,
+      "grad_norm": 0.18054398894309998,
+      "learning_rate": 0.0001,
+      "loss": 1.5979,
+      "step": 1595
+    },
+    {
+      "epoch": 0.77878985035784,
+      "grad_norm": 0.18446581065654755,
+      "learning_rate": 0.0001,
+      "loss": 1.6939,
+      "step": 1596
+    },
+    {
+      "epoch": 0.779277813923227,
+      "grad_norm": 0.19891038537025452,
+      "learning_rate": 0.0001,
+      "loss": 1.598,
+      "step": 1597
+    },
+    {
+      "epoch": 0.7797657774886142,
+      "grad_norm": 0.1845475286245346,
+      "learning_rate": 0.0001,
+      "loss": 1.6092,
+      "step": 1598
+    },
+    {
+      "epoch": 0.7802537410540012,
+      "grad_norm": 0.19764572381973267,
+      "learning_rate": 0.0001,
+      "loss": 1.7236,
+      "step": 1599
+    },
+    {
+      "epoch": 0.7807417046193884,
+      "grad_norm": 0.18992526829242706,
+      "learning_rate": 0.0001,
+      "loss": 1.659,
+      "step": 1600
+    },
+    {
+      "epoch": 0.7812296681847756,
+      "grad_norm": 0.19778114557266235,
+      "learning_rate": 0.0001,
+      "loss": 1.7526,
+      "step": 1601
+    },
+    {
+      "epoch": 0.7817176317501626,
+      "grad_norm": 0.18773706257343292,
+      "learning_rate": 0.0001,
+      "loss": 1.747,
+      "step": 1602
+    },
+    {
+      "epoch": 0.7822055953155498,
+      "grad_norm": 0.1825982630252838,
+      "learning_rate": 0.0001,
+      "loss": 1.6572,
+      "step": 1603
+    },
+    {
+      "epoch": 0.7826935588809368,
+      "grad_norm": 0.1806437224149704,
+      "learning_rate": 0.0001,
+      "loss": 1.6724,
+      "step": 1604
+    },
+    {
+      "epoch": 0.783181522446324,
+      "grad_norm": 0.20853163301944733,
+      "learning_rate": 0.0001,
+      "loss": 1.7719,
+      "step": 1605
+    },
+    {
+      "epoch": 0.7836694860117112,
+      "grad_norm": 0.17729510366916656,
+      "learning_rate": 0.0001,
+      "loss": 1.5767,
+      "step": 1606
+    },
+    {
+      "epoch": 0.7841574495770982,
+      "grad_norm": 0.1899496465921402,
+      "learning_rate": 0.0001,
+      "loss": 1.6542,
+      "step": 1607
+    },
+    {
+      "epoch": 0.7846454131424854,
+      "grad_norm": 0.1904434859752655,
+      "learning_rate": 0.0001,
+      "loss": 1.6901,
+      "step": 1608
+    },
+    {
+      "epoch": 0.7851333767078725,
+      "grad_norm": 0.19352108240127563,
+      "learning_rate": 0.0001,
+      "loss": 1.6535,
+      "step": 1609
+    },
+    {
+      "epoch": 0.7856213402732596,
+      "grad_norm": 0.18233756721019745,
+      "learning_rate": 0.0001,
+      "loss": 1.6878,
+      "step": 1610
+    },
+    {
+      "epoch": 0.7861093038386467,
+      "grad_norm": 0.17833392322063446,
+      "learning_rate": 0.0001,
+      "loss": 1.5601,
+      "step": 1611
+    },
+    {
+      "epoch": 0.7865972674040338,
+      "grad_norm": 0.18793350458145142,
+      "learning_rate": 0.0001,
+      "loss": 1.6703,
+      "step": 1612
+    },
+    {
+      "epoch": 0.787085230969421,
+      "grad_norm": 0.2092360109090805,
+      "learning_rate": 0.0001,
+      "loss": 1.7353,
+      "step": 1613
+    },
+    {
+      "epoch": 0.7875731945348081,
+      "grad_norm": 0.18878163397312164,
+      "learning_rate": 0.0001,
+      "loss": 1.6139,
+      "step": 1614
+    },
+    {
+      "epoch": 0.7880611581001952,
+      "grad_norm": 0.17977482080459595,
+      "learning_rate": 0.0001,
+      "loss": 1.5614,
+      "step": 1615
+    },
+    {
+      "epoch": 0.7885491216655823,
+      "grad_norm": 0.1935320347547531,
+      "learning_rate": 0.0001,
+      "loss": 1.6885,
+      "step": 1616
+    },
+    {
+      "epoch": 0.7890370852309694,
+      "grad_norm": 0.19334746897220612,
+      "learning_rate": 0.0001,
+      "loss": 1.7766,
+      "step": 1617
+    },
+    {
+      "epoch": 0.7895250487963565,
+      "grad_norm": 0.2163378894329071,
+      "learning_rate": 0.0001,
+      "loss": 1.7319,
+      "step": 1618
+    },
+    {
+      "epoch": 0.7900130123617437,
+      "grad_norm": 0.1957041621208191,
+      "learning_rate": 0.0001,
+      "loss": 1.7022,
+      "step": 1619
+    },
+    {
+      "epoch": 0.7905009759271308,
+      "grad_norm": 0.18318067491054535,
+      "learning_rate": 0.0001,
+      "loss": 1.5488,
+      "step": 1620
+    },
+    {
+      "epoch": 0.7909889394925179,
+      "grad_norm": 0.2046167552471161,
+      "learning_rate": 0.0001,
+      "loss": 1.5745,
+      "step": 1621
+    },
+    {
+      "epoch": 0.791476903057905,
+      "grad_norm": 0.1877659112215042,
+      "learning_rate": 0.0001,
+      "loss": 1.6708,
+      "step": 1622
+    },
+    {
+      "epoch": 0.7919648666232921,
+      "grad_norm": 0.21634037792682648,
+      "learning_rate": 0.0001,
+      "loss": 1.8028,
+      "step": 1623
+    },
+    {
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.22510355710983276,
+      "learning_rate": 0.0001,
+      "loss": 1.6316,
+      "step": 1624
+    },
+    {
+      "epoch": 0.7929407937540663,
+      "grad_norm": 0.21921491622924805,
+      "learning_rate": 0.0001,
+      "loss": 1.6614,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7934287573194535,
+      "grad_norm": 0.1849048137664795,
+      "learning_rate": 0.0001,
+      "loss": 1.7025,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7939167208848406,
+      "grad_norm": 0.20738188922405243,
+      "learning_rate": 0.0001,
+      "loss": 1.5814,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7944046844502277,
+      "grad_norm": 0.227822944521904,
+      "learning_rate": 0.0001,
+      "loss": 1.6831,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7948926480156149,
+      "grad_norm": 0.1745869219303131,
+      "learning_rate": 0.0001,
+      "loss": 1.5815,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7953806115810019,
+      "grad_norm": 0.23134981095790863,
+      "learning_rate": 0.0001,
+      "loss": 1.6855,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7958685751463891,
+      "grad_norm": 0.2177252322435379,
+      "learning_rate": 0.0001,
+      "loss": 1.6097,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7963565387117761,
+      "grad_norm": 0.19471696019172668,
+      "learning_rate": 0.0001,
+      "loss": 1.7006,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7968445022771633,
+      "grad_norm": 0.212845116853714,
+      "learning_rate": 0.0001,
+      "loss": 1.6917,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7973324658425505,
+      "grad_norm": 0.22464518249034882,
+      "learning_rate": 0.0001,
+      "loss": 1.5826,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7978204294079375,
+      "grad_norm": 0.185982346534729,
+      "learning_rate": 0.0001,
+      "loss": 1.7345,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7983083929733247,
+      "grad_norm": 0.20566463470458984,
+      "learning_rate": 0.0001,
+      "loss": 1.6962,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7987963565387117,
+      "grad_norm": 0.20042772591114044,
+      "learning_rate": 0.0001,
+      "loss": 1.6684,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7992843201040989,
+      "grad_norm": 0.1849479079246521,
+      "learning_rate": 0.0001,
+      "loss": 1.7162,
+      "step": 1638
+    },
+    {
+      "epoch": 0.799772283669486,
+      "grad_norm": 0.199243426322937,
+      "learning_rate": 0.0001,
+      "loss": 1.8008,
+      "step": 1639
+    },
+    {
+      "epoch": 0.8002602472348731,
+      "grad_norm": 0.18501123785972595,
+      "learning_rate": 0.0001,
+      "loss": 1.7044,
+      "step": 1640
+    },
+    {
+      "epoch": 0.8007482108002603,
+      "grad_norm": 0.18791933357715607,
+      "learning_rate": 0.0001,
+      "loss": 1.6895,
+      "step": 1641
+    },
+    {
+      "epoch": 0.8012361743656473,
+      "grad_norm": 0.20649796724319458,
+      "learning_rate": 0.0001,
+      "loss": 1.7186,
+      "step": 1642
+    },
+    {
+      "epoch": 0.8017241379310345,
+      "grad_norm": 0.18353037536144257,
+      "learning_rate": 0.0001,
+      "loss": 1.5924,
+      "step": 1643
+    },
+    {
+      "epoch": 0.8022121014964216,
+      "grad_norm": 0.19343356788158417,
+      "learning_rate": 0.0001,
+      "loss": 1.7828,
+      "step": 1644
+    },
+    {
+      "epoch": 0.8027000650618087,
+      "grad_norm": 0.18914443254470825,
+      "learning_rate": 0.0001,
+      "loss": 1.736,
+      "step": 1645
+    },
+    {
+      "epoch": 0.8031880286271958,
+      "grad_norm": 0.19788146018981934,
+      "learning_rate": 0.0001,
+      "loss": 1.8165,
+      "step": 1646
+    },
+    {
+      "epoch": 0.8036759921925829,
+      "grad_norm": 0.19266889989376068,
+      "learning_rate": 0.0001,
+      "loss": 1.6921,
+      "step": 1647
+    },
+    {
+      "epoch": 0.8041639557579701,
+      "grad_norm": 0.1988595426082611,
+      "learning_rate": 0.0001,
+      "loss": 1.6117,
+      "step": 1648
+    },
+    {
+      "epoch": 0.8046519193233572,
+      "grad_norm": 0.1922290027141571,
+      "learning_rate": 0.0001,
+      "loss": 1.6717,
+      "step": 1649
+    },
+    {
+      "epoch": 0.8051398828887443,
+      "grad_norm": 0.1773114949464798,
+      "learning_rate": 0.0001,
+      "loss": 1.5806,
+      "step": 1650
+    },
+    {
+      "epoch": 0.8056278464541314,
+      "grad_norm": 0.18884365260601044,
+      "learning_rate": 0.0001,
+      "loss": 1.6146,
+      "step": 1651
+    },
+    {
+      "epoch": 0.8061158100195186,
+      "grad_norm": 0.1919168382883072,
+      "learning_rate": 0.0001,
+      "loss": 1.6228,
+      "step": 1652
+    },
+    {
+      "epoch": 0.8066037735849056,
+      "grad_norm": 0.19377237558364868,
+      "learning_rate": 0.0001,
+      "loss": 1.6811,
+      "step": 1653
+    },
+    {
+      "epoch": 0.8070917371502928,
+      "grad_norm": 0.20503416657447815,
+      "learning_rate": 0.0001,
+      "loss": 1.6838,
+      "step": 1654
+    },
+    {
+      "epoch": 0.8075797007156799,
+      "grad_norm": 0.1844811588525772,
+      "learning_rate": 0.0001,
+      "loss": 1.5669,
+      "step": 1655
+    },
+    {
+      "epoch": 0.808067664281067,
+      "grad_norm": 0.19252845644950867,
+      "learning_rate": 0.0001,
+      "loss": 1.7096,
+      "step": 1656
+    },
+    {
+      "epoch": 0.8085556278464542,
+      "grad_norm": 0.1933499425649643,
+      "learning_rate": 0.0001,
+      "loss": 1.6825,
+      "step": 1657
+    },
+    {
+      "epoch": 0.8090435914118412,
+      "grad_norm": 0.1752086579799652,
+      "learning_rate": 0.0001,
+      "loss": 1.6121,
+      "step": 1658
+    },
+    {
+      "epoch": 0.8095315549772284,
+      "grad_norm": 0.1772938370704651,
+      "learning_rate": 0.0001,
+      "loss": 1.7315,
+      "step": 1659
+    },
+    {
+      "epoch": 0.8100195185426154,
+      "grad_norm": 0.18318156898021698,
+      "learning_rate": 0.0001,
+      "loss": 1.7374,
+      "step": 1660
+    },
+    {
+      "epoch": 0.8105074821080026,
+      "grad_norm": 0.1868072748184204,
+      "learning_rate": 0.0001,
+      "loss": 1.6066,
+      "step": 1661
+    },
+    {
+      "epoch": 0.8109954456733898,
+      "grad_norm": 0.1911938190460205,
+      "learning_rate": 0.0001,
+      "loss": 1.74,
+      "step": 1662
+    },
+    {
+      "epoch": 0.8114834092387768,
+      "grad_norm": 0.19108733534812927,
+      "learning_rate": 0.0001,
+      "loss": 1.7353,
+      "step": 1663
+    },
+    {
+      "epoch": 0.811971372804164,
+      "grad_norm": 0.1893574446439743,
+      "learning_rate": 0.0001,
+      "loss": 1.6413,
+      "step": 1664
+    },
+    {
+      "epoch": 0.812459336369551,
+      "grad_norm": 0.18296094238758087,
+      "learning_rate": 0.0001,
+      "loss": 1.679,
+      "step": 1665
+    },
+    {
+      "epoch": 0.8129472999349382,
+      "grad_norm": 0.18871071934700012,
+      "learning_rate": 0.0001,
+      "loss": 1.6738,
+      "step": 1666
+    },
+    {
+      "epoch": 0.8134352635003254,
+      "grad_norm": 0.192779541015625,
+      "learning_rate": 0.0001,
+      "loss": 1.7312,
+      "step": 1667
+    },
+    {
+      "epoch": 0.8139232270657124,
+      "grad_norm": 0.19144397974014282,
+      "learning_rate": 0.0001,
+      "loss": 1.649,
+      "step": 1668
+    },
+    {
+      "epoch": 0.8144111906310996,
+      "grad_norm": 0.1832989603281021,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 1669
+    },
+    {
+      "epoch": 0.8148991541964866,
+      "grad_norm": 0.1771543025970459,
+      "learning_rate": 0.0001,
+      "loss": 1.547,
+      "step": 1670
+    },
+    {
+      "epoch": 0.8153871177618738,
+      "grad_norm": 0.2000136524438858,
+      "learning_rate": 0.0001,
+      "loss": 1.7604,
+      "step": 1671
+    },
+    {
+      "epoch": 0.8158750813272609,
+      "grad_norm": 0.17948202788829803,
+      "learning_rate": 0.0001,
+      "loss": 1.6072,
+      "step": 1672
+    },
+    {
+      "epoch": 0.816363044892648,
+      "grad_norm": 0.18728068470954895,
+      "learning_rate": 0.0001,
+      "loss": 1.6382,
+      "step": 1673
+    },
+    {
+      "epoch": 0.8168510084580352,
+      "grad_norm": 0.17753037810325623,
+      "learning_rate": 0.0001,
+      "loss": 1.58,
+      "step": 1674
+    },
+    {
+      "epoch": 0.8173389720234222,
+      "grad_norm": 0.19952118396759033,
+      "learning_rate": 0.0001,
+      "loss": 1.7201,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8178269355888094,
+      "grad_norm": 0.18967780470848083,
+      "learning_rate": 0.0001,
+      "loss": 1.7637,
+      "step": 1676
+    },
+    {
+      "epoch": 0.8183148991541965,
+      "grad_norm": 0.18781159818172455,
+      "learning_rate": 0.0001,
+      "loss": 1.6385,
+      "step": 1677
+    },
+    {
+      "epoch": 0.8188028627195836,
+      "grad_norm": 0.18192321062088013,
+      "learning_rate": 0.0001,
+      "loss": 1.6262,
+      "step": 1678
+    },
+    {
+      "epoch": 0.8192908262849707,
+      "grad_norm": 0.1921571046113968,
+      "learning_rate": 0.0001,
+      "loss": 1.7492,
+      "step": 1679
+    },
+    {
+      "epoch": 0.8197787898503578,
+      "grad_norm": 0.19465389847755432,
+      "learning_rate": 0.0001,
+      "loss": 1.6658,
+      "step": 1680
+    },
+    {
+      "epoch": 0.820266753415745,
+      "grad_norm": 0.19218620657920837,
+      "learning_rate": 0.0001,
+      "loss": 1.6639,
+      "step": 1681
+    },
+    {
+      "epoch": 0.8207547169811321,
+      "grad_norm": 0.18591156601905823,
+      "learning_rate": 0.0001,
+      "loss": 1.5869,
+      "step": 1682
+    },
+    {
+      "epoch": 0.8212426805465192,
+      "grad_norm": 0.2010912001132965,
+      "learning_rate": 0.0001,
+      "loss": 1.7304,
+      "step": 1683
+    },
+    {
+      "epoch": 0.8217306441119063,
+      "grad_norm": 0.18728473782539368,
+      "learning_rate": 0.0001,
+      "loss": 1.6596,
+      "step": 1684
+    },
+    {
+      "epoch": 0.8222186076772934,
+      "grad_norm": 0.1863940954208374,
+      "learning_rate": 0.0001,
+      "loss": 1.5735,
+      "step": 1685
+    },
+    {
+      "epoch": 0.8227065712426805,
+      "grad_norm": 0.18404056131839752,
+      "learning_rate": 0.0001,
+      "loss": 1.5921,
+      "step": 1686
+    },
+    {
+      "epoch": 0.8231945348080677,
+      "grad_norm": 0.19402609765529633,
+      "learning_rate": 0.0001,
+      "loss": 1.6873,
+      "step": 1687
+    },
+    {
+      "epoch": 0.8236824983734548,
+      "grad_norm": 0.1880168318748474,
+      "learning_rate": 0.0001,
+      "loss": 1.6991,
+      "step": 1688
+    },
+    {
+      "epoch": 0.8241704619388419,
+      "grad_norm": 0.18674950301647186,
+      "learning_rate": 0.0001,
+      "loss": 1.7284,
+      "step": 1689
+    },
+    {
+      "epoch": 0.824658425504229,
+      "grad_norm": 0.18363825976848602,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 1690
+    },
+    {
+      "epoch": 0.8251463890696161,
+      "grad_norm": 0.19676996767520905,
+      "learning_rate": 0.0001,
+      "loss": 1.8167,
+      "step": 1691
+    },
+    {
+      "epoch": 0.8256343526350033,
+      "grad_norm": 0.17890886962413788,
+      "learning_rate": 0.0001,
+      "loss": 1.6515,
+      "step": 1692
+    },
+    {
+      "epoch": 0.8261223162003903,
+      "grad_norm": 0.19220395386219025,
+      "learning_rate": 0.0001,
+      "loss": 1.6924,
+      "step": 1693
+    },
+    {
+      "epoch": 0.8266102797657775,
+      "grad_norm": 0.17939774692058563,
+      "learning_rate": 0.0001,
+      "loss": 1.6658,
+      "step": 1694
+    },
+    {
+      "epoch": 0.8270982433311646,
+      "grad_norm": 0.18299368023872375,
+      "learning_rate": 0.0001,
+      "loss": 1.6109,
+      "step": 1695
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 0.19559091329574585,
+      "learning_rate": 0.0001,
+      "loss": 1.71,
+      "step": 1696
+    },
+    {
+      "epoch": 0.8280741704619389,
+      "grad_norm": 0.18175216019153595,
+      "learning_rate": 0.0001,
+      "loss": 1.579,
+      "step": 1697
+    },
+    {
+      "epoch": 0.8285621340273259,
+      "grad_norm": 0.17916333675384521,
+      "learning_rate": 0.0001,
+      "loss": 1.6226,
+      "step": 1698
+    },
+    {
+      "epoch": 0.8290500975927131,
+      "grad_norm": 0.19414590299129486,
+      "learning_rate": 0.0001,
+      "loss": 1.6539,
+      "step": 1699
+    },
+    {
+      "epoch": 0.8295380611581002,
+      "grad_norm": 0.190690279006958,
+      "learning_rate": 0.0001,
+      "loss": 1.6743,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8300260247234873,
+      "grad_norm": 0.1775364726781845,
+      "learning_rate": 0.0001,
+      "loss": 1.6156,
+      "step": 1701
+    },
+    {
+      "epoch": 0.8305139882888745,
+      "grad_norm": 0.19147178530693054,
+      "learning_rate": 0.0001,
+      "loss": 1.6988,
+      "step": 1702
+    },
+    {
+      "epoch": 0.8310019518542615,
+      "grad_norm": 0.191775843501091,
+      "learning_rate": 0.0001,
+      "loss": 1.6263,
+      "step": 1703
+    },
+    {
+      "epoch": 0.8314899154196487,
+      "grad_norm": 0.18212559819221497,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 1704
+    },
+    {
+      "epoch": 0.8319778789850358,
+      "grad_norm": 0.1973716914653778,
+      "learning_rate": 0.0001,
+      "loss": 1.6681,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8324658425504229,
+      "grad_norm": 0.19240307807922363,
+      "learning_rate": 0.0001,
+      "loss": 1.6565,
+      "step": 1706
+    },
+    {
+      "epoch": 0.83295380611581,
+      "grad_norm": 0.18731878697872162,
+      "learning_rate": 0.0001,
+      "loss": 1.7008,
+      "step": 1707
+    },
+    {
+      "epoch": 0.8334417696811971,
+      "grad_norm": 0.18686629831790924,
+      "learning_rate": 0.0001,
+      "loss": 1.6936,
+      "step": 1708
+    },
+    {
+      "epoch": 0.8339297332465843,
+      "grad_norm": 0.19418823719024658,
+      "learning_rate": 0.0001,
+      "loss": 1.6749,
+      "step": 1709
+    },
+    {
+      "epoch": 0.8344176968119714,
+      "grad_norm": 0.19065696001052856,
+      "learning_rate": 0.0001,
+      "loss": 1.5825,
+      "step": 1710
+    },
+    {
+      "epoch": 0.8349056603773585,
+      "grad_norm": 0.19086980819702148,
+      "learning_rate": 0.0001,
+      "loss": 1.6586,
+      "step": 1711
+    },
+    {
+      "epoch": 0.8353936239427456,
+      "grad_norm": 0.18461991846561432,
+      "learning_rate": 0.0001,
+      "loss": 1.6877,
+      "step": 1712
+    },
+    {
+      "epoch": 0.8358815875081327,
+      "grad_norm": 0.18750816583633423,
+      "learning_rate": 0.0001,
+      "loss": 1.5869,
+      "step": 1713
+    },
+    {
+      "epoch": 0.8363695510735198,
+      "grad_norm": 0.2001444548368454,
+      "learning_rate": 0.0001,
+      "loss": 1.7659,
+      "step": 1714
+    },
+    {
+      "epoch": 0.836857514638907,
+      "grad_norm": 0.195534348487854,
+      "learning_rate": 0.0001,
+      "loss": 1.6863,
+      "step": 1715
+    },
+    {
+      "epoch": 0.8373454782042941,
+      "grad_norm": 0.18564340472221375,
+      "learning_rate": 0.0001,
+      "loss": 1.6825,
+      "step": 1716
+    },
+    {
+      "epoch": 0.8378334417696812,
+      "grad_norm": 0.1830897182226181,
+      "learning_rate": 0.0001,
+      "loss": 1.5911,
+      "step": 1717
+    },
+    {
+      "epoch": 0.8383214053350683,
+      "grad_norm": 0.18405020236968994,
+      "learning_rate": 0.0001,
+      "loss": 1.6177,
+      "step": 1718
+    },
+    {
+      "epoch": 0.8388093689004554,
+      "grad_norm": 0.1829666793346405,
+      "learning_rate": 0.0001,
+      "loss": 1.5921,
+      "step": 1719
+    },
+    {
+      "epoch": 0.8392973324658426,
+      "grad_norm": 0.19332802295684814,
+      "learning_rate": 0.0001,
+      "loss": 1.6575,
+      "step": 1720
+    },
+    {
+      "epoch": 0.8397852960312296,
+      "grad_norm": 0.19402368366718292,
+      "learning_rate": 0.0001,
+      "loss": 1.6052,
+      "step": 1721
+    },
+    {
+      "epoch": 0.8402732595966168,
+      "grad_norm": 0.1924324482679367,
+      "learning_rate": 0.0001,
+      "loss": 1.7398,
+      "step": 1722
+    },
+    {
+      "epoch": 0.8407612231620039,
+      "grad_norm": 0.1912868320941925,
+      "learning_rate": 0.0001,
+      "loss": 1.6186,
+      "step": 1723
+    },
+    {
+      "epoch": 0.841249186727391,
+      "grad_norm": 0.19338296353816986,
+      "learning_rate": 0.0001,
+      "loss": 1.7595,
+      "step": 1724
+    },
+    {
+      "epoch": 0.8417371502927782,
+      "grad_norm": 0.19593428075313568,
+      "learning_rate": 0.0001,
+      "loss": 1.754,
+      "step": 1725
+    },
+    {
+      "epoch": 0.8422251138581652,
+      "grad_norm": 0.18348434567451477,
+      "learning_rate": 0.0001,
+      "loss": 1.6831,
+      "step": 1726
+    },
+    {
+      "epoch": 0.8427130774235524,
+      "grad_norm": 0.18937769532203674,
+      "learning_rate": 0.0001,
+      "loss": 1.6921,
+      "step": 1727
+    },
+    {
+      "epoch": 0.8432010409889394,
+      "grad_norm": 0.1873343288898468,
+      "learning_rate": 0.0001,
+      "loss": 1.6638,
+      "step": 1728
+    },
+    {
+      "epoch": 0.8436890045543266,
+      "grad_norm": 0.18602168560028076,
+      "learning_rate": 0.0001,
+      "loss": 1.6805,
+      "step": 1729
+    },
+    {
+      "epoch": 0.8441769681197138,
+      "grad_norm": 0.1878175288438797,
+      "learning_rate": 0.0001,
+      "loss": 1.6887,
+      "step": 1730
+    },
+    {
+      "epoch": 0.8446649316851008,
+      "grad_norm": 0.18829582631587982,
+      "learning_rate": 0.0001,
+      "loss": 1.6766,
+      "step": 1731
+    },
+    {
+      "epoch": 0.845152895250488,
+      "grad_norm": 0.19731226563453674,
+      "learning_rate": 0.0001,
+      "loss": 1.7651,
+      "step": 1732
+    },
+    {
+      "epoch": 0.845640858815875,
+      "grad_norm": 0.1821230798959732,
+      "learning_rate": 0.0001,
+      "loss": 1.6899,
+      "step": 1733
+    },
+    {
+      "epoch": 0.8461288223812622,
+      "grad_norm": 0.19442221522331238,
+      "learning_rate": 0.0001,
+      "loss": 1.6838,
+      "step": 1734
+    },
+    {
+      "epoch": 0.8466167859466494,
+      "grad_norm": 0.18161530792713165,
+      "learning_rate": 0.0001,
+      "loss": 1.5891,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8471047495120364,
+      "grad_norm": 0.18692250549793243,
+      "learning_rate": 0.0001,
+      "loss": 1.6704,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8475927130774236,
+      "grad_norm": 0.1943616271018982,
+      "learning_rate": 0.0001,
+      "loss": 1.5322,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8480806766428106,
+      "grad_norm": 0.1985720843076706,
+      "learning_rate": 0.0001,
+      "loss": 1.6677,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8485686402081978,
+      "grad_norm": 0.18838457763195038,
+      "learning_rate": 0.0001,
+      "loss": 1.6649,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.19150035083293915,
+      "learning_rate": 0.0001,
+      "loss": 1.6901,
+      "step": 1740
+    },
+    {
+      "epoch": 0.849544567338972,
+      "grad_norm": 0.18909554183483124,
+      "learning_rate": 0.0001,
+      "loss": 1.6245,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8500325309043592,
+      "grad_norm": 0.19308097660541534,
+      "learning_rate": 0.0001,
+      "loss": 1.7253,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8505204944697463,
+      "grad_norm": 0.19270172715187073,
+      "learning_rate": 0.0001,
+      "loss": 1.7028,
+      "step": 1743
+    },
+    {
+      "epoch": 0.8510084580351334,
+      "grad_norm": 0.1880318969488144,
+      "learning_rate": 0.0001,
+      "loss": 1.6327,
+      "step": 1744
+    },
+    {
+      "epoch": 0.8514964216005205,
+      "grad_norm": 0.1749912053346634,
+      "learning_rate": 0.0001,
+      "loss": 1.6776,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8519843851659076,
+      "grad_norm": 0.18535734713077545,
+      "learning_rate": 0.0001,
+      "loss": 1.7433,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8524723487312947,
+      "grad_norm": 0.18681146204471588,
+      "learning_rate": 0.0001,
+      "loss": 1.615,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8529603122966819,
+      "grad_norm": 0.21550102531909943,
+      "learning_rate": 0.0001,
+      "loss": 1.6953,
+      "step": 1748
+    },
+    {
+      "epoch": 0.853448275862069,
+      "grad_norm": 0.1936476081609726,
+      "learning_rate": 0.0001,
+      "loss": 1.6796,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8539362394274561,
+      "grad_norm": 0.18475957214832306,
+      "learning_rate": 0.0001,
+      "loss": 1.7073,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8544242029928432,
+      "grad_norm": 0.1896783858537674,
+      "learning_rate": 0.0001,
+      "loss": 1.6556,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8549121665582303,
+      "grad_norm": 0.1936115324497223,
+      "learning_rate": 0.0001,
+      "loss": 1.6614,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8554001301236175,
+      "grad_norm": 0.18598994612693787,
+      "learning_rate": 0.0001,
+      "loss": 1.708,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8558880936890045,
+      "grad_norm": 0.1906435489654541,
+      "learning_rate": 0.0001,
+      "loss": 1.6371,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8563760572543917,
+      "grad_norm": 0.199356809258461,
+      "learning_rate": 0.0001,
+      "loss": 1.55,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8568640208197787,
+      "grad_norm": 0.1791318953037262,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 1756
+    },
+    {
+      "epoch": 0.8573519843851659,
+      "grad_norm": 0.2023209035396576,
+      "learning_rate": 0.0001,
+      "loss": 1.7851,
+      "step": 1757
+    },
+    {
+      "epoch": 0.8578399479505531,
+      "grad_norm": 0.18916785717010498,
+      "learning_rate": 0.0001,
+      "loss": 1.6051,
+      "step": 1758
+    },
+    {
+      "epoch": 0.8583279115159401,
+      "grad_norm": 0.19388586282730103,
+      "learning_rate": 0.0001,
+      "loss": 1.736,
+      "step": 1759
+    },
+    {
+      "epoch": 0.8588158750813273,
+      "grad_norm": 0.1888210028409958,
+      "learning_rate": 0.0001,
+      "loss": 1.6526,
+      "step": 1760
+    },
+    {
+      "epoch": 0.8593038386467143,
+      "grad_norm": 0.18460845947265625,
+      "learning_rate": 0.0001,
+      "loss": 1.657,
+      "step": 1761
+    },
+    {
+      "epoch": 0.8597918022121015,
+      "grad_norm": 0.18918482959270477,
+      "learning_rate": 0.0001,
+      "loss": 1.6517,
+      "step": 1762
+    },
+    {
+      "epoch": 0.8602797657774887,
+      "grad_norm": 0.2215425670146942,
+      "learning_rate": 0.0001,
+      "loss": 1.8116,
+      "step": 1763
+    },
+    {
+      "epoch": 0.8607677293428757,
+      "grad_norm": 0.1845925748348236,
+      "learning_rate": 0.0001,
+      "loss": 1.5914,
+      "step": 1764
+    },
+    {
+      "epoch": 0.8612556929082629,
+      "grad_norm": 0.1800438016653061,
+      "learning_rate": 0.0001,
+      "loss": 1.6074,
+      "step": 1765
+    },
+    {
+      "epoch": 0.8617436564736499,
+      "grad_norm": 0.1770068109035492,
+      "learning_rate": 0.0001,
+      "loss": 1.6869,
+      "step": 1766
+    },
+    {
+      "epoch": 0.8622316200390371,
+      "grad_norm": 0.19672898948192596,
+      "learning_rate": 0.0001,
+      "loss": 1.5741,
+      "step": 1767
+    },
+    {
+      "epoch": 0.8627195836044242,
+      "grad_norm": 0.1833876222372055,
+      "learning_rate": 0.0001,
+      "loss": 1.6364,
+      "step": 1768
+    },
+    {
+      "epoch": 0.8632075471698113,
+      "grad_norm": 0.18901808559894562,
+      "learning_rate": 0.0001,
+      "loss": 1.7703,
+      "step": 1769
+    },
+    {
+      "epoch": 0.8636955107351985,
+      "grad_norm": 0.18848566710948944,
+      "learning_rate": 0.0001,
+      "loss": 1.7137,
+      "step": 1770
+    },
+    {
+      "epoch": 0.8641834743005855,
+      "grad_norm": 0.1907043308019638,
+      "learning_rate": 0.0001,
+      "loss": 1.6075,
+      "step": 1771
+    },
+    {
+      "epoch": 0.8646714378659727,
+      "grad_norm": 0.20296499133110046,
+      "learning_rate": 0.0001,
+      "loss": 1.7779,
+      "step": 1772
+    },
+    {
+      "epoch": 0.8651594014313598,
+      "grad_norm": 0.18933746218681335,
+      "learning_rate": 0.0001,
+      "loss": 1.7367,
+      "step": 1773
+    },
+    {
+      "epoch": 0.8656473649967469,
+      "grad_norm": 0.19736367464065552,
+      "learning_rate": 0.0001,
+      "loss": 1.7314,
+      "step": 1774
+    },
+    {
+      "epoch": 0.866135328562134,
+      "grad_norm": 0.18523859977722168,
+      "learning_rate": 0.0001,
+      "loss": 1.6182,
+      "step": 1775
+    },
+    {
+      "epoch": 0.8666232921275211,
+      "grad_norm": 0.200862318277359,
+      "learning_rate": 0.0001,
+      "loss": 1.72,
+      "step": 1776
+    },
+    {
+      "epoch": 0.8671112556929083,
+      "grad_norm": 0.18352118134498596,
+      "learning_rate": 0.0001,
+      "loss": 1.5903,
+      "step": 1777
+    },
+    {
+      "epoch": 0.8675992192582954,
+      "grad_norm": 0.18788601458072662,
+      "learning_rate": 0.0001,
+      "loss": 1.6115,
+      "step": 1778
+    },
+    {
+      "epoch": 0.8680871828236825,
+      "grad_norm": 0.2129855901002884,
+      "learning_rate": 0.0001,
+      "loss": 1.6931,
+      "step": 1779
+    },
+    {
+      "epoch": 0.8685751463890696,
+      "grad_norm": 0.18848247826099396,
+      "learning_rate": 0.0001,
+      "loss": 1.7186,
+      "step": 1780
+    },
+    {
+      "epoch": 0.8690631099544567,
+      "grad_norm": 0.18020255863666534,
+      "learning_rate": 0.0001,
+      "loss": 1.5447,
+      "step": 1781
+    },
+    {
+      "epoch": 0.8695510735198438,
+      "grad_norm": 0.19755859673023224,
+      "learning_rate": 0.0001,
+      "loss": 1.6055,
+      "step": 1782
+    },
+    {
+      "epoch": 0.870039037085231,
+      "grad_norm": 0.19791275262832642,
+      "learning_rate": 0.0001,
+      "loss": 1.681,
+      "step": 1783
+    },
+    {
+      "epoch": 0.870527000650618,
+      "grad_norm": 0.194001704454422,
+      "learning_rate": 0.0001,
+      "loss": 1.7187,
+      "step": 1784
+    },
+    {
+      "epoch": 0.8710149642160052,
+      "grad_norm": 0.19635316729545593,
+      "learning_rate": 0.0001,
+      "loss": 1.6656,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8715029277813924,
+      "grad_norm": 0.18945664167404175,
+      "learning_rate": 0.0001,
+      "loss": 1.6149,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8719908913467794,
+      "grad_norm": 0.19505362212657928,
+      "learning_rate": 0.0001,
+      "loss": 1.7541,
+      "step": 1787
+    },
+    {
+      "epoch": 0.8724788549121666,
+      "grad_norm": 0.2048032581806183,
+      "learning_rate": 0.0001,
+      "loss": 1.7539,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8729668184775536,
+      "grad_norm": 0.1928759068250656,
+      "learning_rate": 0.0001,
+      "loss": 1.6836,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8734547820429408,
+      "grad_norm": 0.1996418982744217,
+      "learning_rate": 0.0001,
+      "loss": 1.5969,
+      "step": 1790
+    },
+    {
+      "epoch": 0.873942745608328,
+      "grad_norm": 0.1895206868648529,
+      "learning_rate": 0.0001,
+      "loss": 1.6577,
+      "step": 1791
+    },
+    {
+      "epoch": 0.874430709173715,
+      "grad_norm": 0.19562388956546783,
+      "learning_rate": 0.0001,
+      "loss": 1.6688,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8749186727391022,
+      "grad_norm": 0.1907225251197815,
+      "learning_rate": 0.0001,
+      "loss": 1.6491,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8754066363044892,
+      "grad_norm": 0.1887628734111786,
+      "learning_rate": 0.0001,
+      "loss": 1.657,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8758945998698764,
+      "grad_norm": 0.1809932142496109,
+      "learning_rate": 0.0001,
+      "loss": 1.5484,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8763825634352636,
+      "grad_norm": 0.18477503955364227,
+      "learning_rate": 0.0001,
+      "loss": 1.5783,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8768705270006506,
+      "grad_norm": 0.2000272572040558,
+      "learning_rate": 0.0001,
+      "loss": 1.7027,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8773584905660378,
+      "grad_norm": 0.18049073219299316,
+      "learning_rate": 0.0001,
+      "loss": 1.6178,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8778464541314248,
+      "grad_norm": 0.19112178683280945,
+      "learning_rate": 0.0001,
+      "loss": 1.7403,
+      "step": 1799
+    },
+    {
+      "epoch": 0.878334417696812,
+      "grad_norm": 0.1985938549041748,
+      "learning_rate": 0.0001,
+      "loss": 1.5708,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8788223812621991,
+      "grad_norm": 0.1869334578514099,
+      "learning_rate": 0.0001,
+      "loss": 1.7111,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8793103448275862,
+      "grad_norm": 0.20291414856910706,
+      "learning_rate": 0.0001,
+      "loss": 1.753,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8797983083929733,
+      "grad_norm": 0.19386352598667145,
+      "learning_rate": 0.0001,
+      "loss": 1.6402,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8802862719583604,
+      "grad_norm": 0.18778564035892487,
+      "learning_rate": 0.0001,
+      "loss": 1.6885,
+      "step": 1804
+    },
+    {
+      "epoch": 0.8807742355237476,
+      "grad_norm": 0.19215920567512512,
+      "learning_rate": 0.0001,
+      "loss": 1.6659,
+      "step": 1805
+    },
+    {
+      "epoch": 0.8812621990891347,
+      "grad_norm": 0.2166108340024948,
+      "learning_rate": 0.0001,
+      "loss": 1.6867,
+      "step": 1806
+    },
+    {
+      "epoch": 0.8817501626545218,
+      "grad_norm": 0.18133436143398285,
+      "learning_rate": 0.0001,
+      "loss": 1.5926,
+      "step": 1807
+    },
+    {
+      "epoch": 0.8822381262199089,
+      "grad_norm": 0.18868204951286316,
+      "learning_rate": 0.0001,
+      "loss": 1.6172,
+      "step": 1808
+    },
+    {
+      "epoch": 0.882726089785296,
+      "grad_norm": 0.20519724488258362,
+      "learning_rate": 0.0001,
+      "loss": 1.7143,
+      "step": 1809
+    },
+    {
+      "epoch": 0.8832140533506831,
+      "grad_norm": 0.190599262714386,
+      "learning_rate": 0.0001,
+      "loss": 1.7854,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8837020169160703,
+      "grad_norm": 0.1950819492340088,
+      "learning_rate": 0.0001,
+      "loss": 1.5849,
+      "step": 1811
+    },
+    {
+      "epoch": 0.8841899804814574,
+      "grad_norm": 0.1758696287870407,
+      "learning_rate": 0.0001,
+      "loss": 1.5769,
+      "step": 1812
+    },
+    {
+      "epoch": 0.8846779440468445,
+      "grad_norm": 0.19683918356895447,
+      "learning_rate": 0.0001,
+      "loss": 1.7032,
+      "step": 1813
+    },
+    {
+      "epoch": 0.8851659076122316,
+      "grad_norm": 0.17837479710578918,
+      "learning_rate": 0.0001,
+      "loss": 1.5567,
+      "step": 1814
+    },
+    {
+      "epoch": 0.8856538711776187,
+      "grad_norm": 0.19034849107265472,
+      "learning_rate": 0.0001,
+      "loss": 1.7274,
+      "step": 1815
+    },
+    {
+      "epoch": 0.8861418347430059,
+      "grad_norm": 0.19284600019454956,
+      "learning_rate": 0.0001,
+      "loss": 1.6933,
+      "step": 1816
+    },
+    {
+      "epoch": 0.886629798308393,
+      "grad_norm": 0.18302425742149353,
+      "learning_rate": 0.0001,
+      "loss": 1.6091,
+      "step": 1817
+    },
+    {
+      "epoch": 0.8871177618737801,
+      "grad_norm": 0.1840258687734604,
+      "learning_rate": 0.0001,
+      "loss": 1.6178,
+      "step": 1818
+    },
+    {
+      "epoch": 0.8876057254391672,
+      "grad_norm": 0.19161662459373474,
+      "learning_rate": 0.0001,
+      "loss": 1.7768,
+      "step": 1819
+    },
+    {
+      "epoch": 0.8880936890045543,
+      "grad_norm": 0.19132988154888153,
+      "learning_rate": 0.0001,
+      "loss": 1.6508,
+      "step": 1820
+    },
+    {
+      "epoch": 0.8885816525699415,
+      "grad_norm": 0.19872814416885376,
+      "learning_rate": 0.0001,
+      "loss": 1.756,
+      "step": 1821
+    },
+    {
+      "epoch": 0.8890696161353285,
+      "grad_norm": 0.19945646822452545,
+      "learning_rate": 0.0001,
+      "loss": 1.6164,
+      "step": 1822
+    },
+    {
+      "epoch": 0.8895575797007157,
+      "grad_norm": 0.1971243917942047,
+      "learning_rate": 0.0001,
+      "loss": 1.718,
+      "step": 1823
+    },
+    {
+      "epoch": 0.8900455432661027,
+      "grad_norm": 0.1854260414838791,
+      "learning_rate": 0.0001,
+      "loss": 1.6402,
+      "step": 1824
+    },
+    {
+      "epoch": 0.8905335068314899,
+      "grad_norm": 0.19041430950164795,
+      "learning_rate": 0.0001,
+      "loss": 1.5898,
+      "step": 1825
+    },
+    {
+      "epoch": 0.8910214703968771,
+      "grad_norm": 0.20725248754024506,
+      "learning_rate": 0.0001,
+      "loss": 1.6835,
+      "step": 1826
+    },
+    {
+      "epoch": 0.8915094339622641,
+      "grad_norm": 0.19040203094482422,
+      "learning_rate": 0.0001,
+      "loss": 1.7338,
+      "step": 1827
+    },
+    {
+      "epoch": 0.8919973975276513,
+      "grad_norm": 0.19236230850219727,
+      "learning_rate": 0.0001,
+      "loss": 1.6475,
+      "step": 1828
+    },
+    {
+      "epoch": 0.8924853610930383,
+      "grad_norm": 0.19737380743026733,
+      "learning_rate": 0.0001,
+      "loss": 1.6187,
+      "step": 1829
+    },
+    {
+      "epoch": 0.8929733246584255,
+      "grad_norm": 0.1852233111858368,
+      "learning_rate": 0.0001,
+      "loss": 1.5997,
+      "step": 1830
+    },
+    {
+      "epoch": 0.8934612882238127,
+      "grad_norm": 0.19599072635173798,
+      "learning_rate": 0.0001,
+      "loss": 1.7316,
+      "step": 1831
+    },
+    {
+      "epoch": 0.8939492517891997,
+      "grad_norm": 0.20338813960552216,
+      "learning_rate": 0.0001,
+      "loss": 1.7148,
+      "step": 1832
+    },
+    {
+      "epoch": 0.8944372153545869,
+      "grad_norm": 0.19061818718910217,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 1833
+    },
+    {
+      "epoch": 0.894925178919974,
+      "grad_norm": 0.1853945404291153,
+      "learning_rate": 0.0001,
+      "loss": 1.6853,
+      "step": 1834
+    },
+    {
+      "epoch": 0.8954131424853611,
+      "grad_norm": 0.1979200690984726,
+      "learning_rate": 0.0001,
+      "loss": 1.6889,
+      "step": 1835
+    },
+    {
+      "epoch": 0.8959011060507482,
+      "grad_norm": 0.1888495832681656,
+      "learning_rate": 0.0001,
+      "loss": 1.6354,
+      "step": 1836
+    },
+    {
+      "epoch": 0.8963890696161353,
+      "grad_norm": 0.1901407241821289,
+      "learning_rate": 0.0001,
+      "loss": 1.642,
+      "step": 1837
+    },
+    {
+      "epoch": 0.8968770331815225,
+      "grad_norm": 0.20139391720294952,
+      "learning_rate": 0.0001,
+      "loss": 1.7404,
+      "step": 1838
+    },
+    {
+      "epoch": 0.8973649967469096,
+      "grad_norm": 0.18233637511730194,
+      "learning_rate": 0.0001,
+      "loss": 1.6359,
+      "step": 1839
+    },
+    {
+      "epoch": 0.8978529603122967,
+      "grad_norm": 0.18822279572486877,
+      "learning_rate": 0.0001,
+      "loss": 1.6719,
+      "step": 1840
+    },
+    {
+      "epoch": 0.8983409238776838,
+      "grad_norm": 0.18992972373962402,
+      "learning_rate": 0.0001,
+      "loss": 1.6534,
+      "step": 1841
+    },
+    {
+      "epoch": 0.8988288874430709,
+      "grad_norm": 0.18185953795909882,
+      "learning_rate": 0.0001,
+      "loss": 1.5577,
+      "step": 1842
+    },
+    {
+      "epoch": 0.899316851008458,
+      "grad_norm": 0.18909883499145508,
+      "learning_rate": 0.0001,
+      "loss": 1.755,
+      "step": 1843
+    },
+    {
+      "epoch": 0.8998048145738452,
+      "grad_norm": 0.1953660547733307,
+      "learning_rate": 0.0001,
+      "loss": 1.6148,
+      "step": 1844
+    },
+    {
+      "epoch": 0.9002927781392323,
+      "grad_norm": 0.19910429418087006,
+      "learning_rate": 0.0001,
+      "loss": 1.6864,
+      "step": 1845
+    },
+    {
+      "epoch": 0.9007807417046194,
+      "grad_norm": 0.19126242399215698,
+      "learning_rate": 0.0001,
+      "loss": 1.6264,
+      "step": 1846
+    },
+    {
+      "epoch": 0.9012687052700065,
+      "grad_norm": 0.18766577541828156,
+      "learning_rate": 0.0001,
+      "loss": 1.7142,
+      "step": 1847
+    },
+    {
+      "epoch": 0.9017566688353936,
+      "grad_norm": 0.18595442175865173,
+      "learning_rate": 0.0001,
+      "loss": 1.6663,
+      "step": 1848
+    },
+    {
+      "epoch": 0.9022446324007808,
+      "grad_norm": 0.1945996880531311,
+      "learning_rate": 0.0001,
+      "loss": 1.7243,
+      "step": 1849
+    },
+    {
+      "epoch": 0.9027325959661678,
+      "grad_norm": 0.1816486269235611,
+      "learning_rate": 0.0001,
+      "loss": 1.632,
+      "step": 1850
+    },
+    {
+      "epoch": 0.903220559531555,
+      "grad_norm": 0.18732072412967682,
+      "learning_rate": 0.0001,
+      "loss": 1.5309,
+      "step": 1851
+    },
+    {
+      "epoch": 0.903708523096942,
+      "grad_norm": 0.18456104397773743,
+      "learning_rate": 0.0001,
+      "loss": 1.6329,
+      "step": 1852
+    },
+    {
+      "epoch": 0.9041964866623292,
+      "grad_norm": 0.1911781132221222,
+      "learning_rate": 0.0001,
+      "loss": 1.5973,
+      "step": 1853
+    },
+    {
+      "epoch": 0.9046844502277164,
+      "grad_norm": 0.19040006399154663,
+      "learning_rate": 0.0001,
+      "loss": 1.6354,
+      "step": 1854
+    },
+    {
+      "epoch": 0.9051724137931034,
+      "grad_norm": 0.17650040984153748,
+      "learning_rate": 0.0001,
+      "loss": 1.5285,
+      "step": 1855
+    },
+    {
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.19496093690395355,
+      "learning_rate": 0.0001,
+      "loss": 1.7319,
+      "step": 1856
+    },
+    {
+      "epoch": 0.9061483409238776,
+      "grad_norm": 0.18969886004924774,
+      "learning_rate": 0.0001,
+      "loss": 1.591,
+      "step": 1857
+    },
+    {
+      "epoch": 0.9066363044892648,
+      "grad_norm": 0.1851990967988968,
+      "learning_rate": 0.0001,
+      "loss": 1.6195,
+      "step": 1858
+    },
+    {
+      "epoch": 0.907124268054652,
+      "grad_norm": 0.2054198980331421,
+      "learning_rate": 0.0001,
+      "loss": 1.779,
+      "step": 1859
+    },
+    {
+      "epoch": 0.907612231620039,
+      "grad_norm": 0.19475503265857697,
+      "learning_rate": 0.0001,
+      "loss": 1.6826,
+      "step": 1860
+    },
+    {
+      "epoch": 0.9081001951854262,
+      "grad_norm": 0.18214543163776398,
+      "learning_rate": 0.0001,
+      "loss": 1.6816,
+      "step": 1861
+    },
+    {
+      "epoch": 0.9085881587508132,
+      "grad_norm": 0.18886341154575348,
+      "learning_rate": 0.0001,
+      "loss": 1.5819,
+      "step": 1862
+    },
+    {
+      "epoch": 0.9090761223162004,
+      "grad_norm": 0.17963244020938873,
+      "learning_rate": 0.0001,
+      "loss": 1.6782,
+      "step": 1863
+    },
+    {
+      "epoch": 0.9095640858815875,
+      "grad_norm": 0.18377065658569336,
+      "learning_rate": 0.0001,
+      "loss": 1.6856,
+      "step": 1864
+    },
+    {
+      "epoch": 0.9100520494469746,
+      "grad_norm": 0.2315450757741928,
+      "learning_rate": 0.0001,
+      "loss": 1.6973,
+      "step": 1865
+    },
+    {
+      "epoch": 0.9105400130123618,
+      "grad_norm": 0.19719429314136505,
+      "learning_rate": 0.0001,
+      "loss": 1.7254,
+      "step": 1866
+    },
+    {
+      "epoch": 0.9110279765777488,
+      "grad_norm": 0.1869468241930008,
+      "learning_rate": 0.0001,
+      "loss": 1.7119,
+      "step": 1867
+    },
+    {
+      "epoch": 0.911515940143136,
+      "grad_norm": 0.19327637553215027,
+      "learning_rate": 0.0001,
+      "loss": 1.6643,
+      "step": 1868
+    },
+    {
+      "epoch": 0.9120039037085231,
+      "grad_norm": 0.1934322863817215,
+      "learning_rate": 0.0001,
+      "loss": 1.7806,
+      "step": 1869
+    },
+    {
+      "epoch": 0.9124918672739102,
+      "grad_norm": 0.1793750673532486,
+      "learning_rate": 0.0001,
+      "loss": 1.5899,
+      "step": 1870
+    },
+    {
+      "epoch": 0.9129798308392973,
+      "grad_norm": 0.1910194307565689,
+      "learning_rate": 0.0001,
+      "loss": 1.676,
+      "step": 1871
+    },
+    {
+      "epoch": 0.9134677944046844,
+      "grad_norm": 0.19140726327896118,
+      "learning_rate": 0.0001,
+      "loss": 1.609,
+      "step": 1872
+    },
+    {
+      "epoch": 0.9139557579700716,
+      "grad_norm": 0.18246062099933624,
+      "learning_rate": 0.0001,
+      "loss": 1.5884,
+      "step": 1873
+    },
+    {
+      "epoch": 0.9144437215354587,
+      "grad_norm": 0.19457212090492249,
+      "learning_rate": 0.0001,
+      "loss": 1.7357,
+      "step": 1874
+    },
+    {
+      "epoch": 0.9149316851008458,
+      "grad_norm": 0.1949836015701294,
+      "learning_rate": 0.0001,
+      "loss": 1.805,
+      "step": 1875
+    },
+    {
+      "epoch": 0.9154196486662329,
+      "grad_norm": 0.18788614869117737,
+      "learning_rate": 0.0001,
+      "loss": 1.6812,
+      "step": 1876
+    },
+    {
+      "epoch": 0.9159076122316201,
+      "grad_norm": 0.1801442950963974,
+      "learning_rate": 0.0001,
+      "loss": 1.6334,
+      "step": 1877
+    },
+    {
+      "epoch": 0.9163955757970071,
+      "grad_norm": 0.18876299262046814,
+      "learning_rate": 0.0001,
+      "loss": 1.7086,
+      "step": 1878
+    },
+    {
+      "epoch": 0.9168835393623943,
+      "grad_norm": 0.20244908332824707,
+      "learning_rate": 0.0001,
+      "loss": 1.5959,
+      "step": 1879
+    },
+    {
+      "epoch": 0.9173715029277814,
+      "grad_norm": 0.1789156198501587,
+      "learning_rate": 0.0001,
+      "loss": 1.5349,
+      "step": 1880
+    },
+    {
+      "epoch": 0.9178594664931685,
+      "grad_norm": 0.2068023681640625,
+      "learning_rate": 0.0001,
+      "loss": 1.7403,
+      "step": 1881
+    },
+    {
+      "epoch": 0.9183474300585557,
+      "grad_norm": 0.18543741106987,
+      "learning_rate": 0.0001,
+      "loss": 1.6918,
+      "step": 1882
+    },
+    {
+      "epoch": 0.9188353936239427,
+      "grad_norm": 0.18919646739959717,
+      "learning_rate": 0.0001,
+      "loss": 1.7332,
+      "step": 1883
+    },
+    {
+      "epoch": 0.9193233571893299,
+      "grad_norm": 0.18657784163951874,
+      "learning_rate": 0.0001,
+      "loss": 1.7097,
+      "step": 1884
+    },
+    {
+      "epoch": 0.9198113207547169,
+      "grad_norm": 0.18775001168251038,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 1885
+    },
+    {
+      "epoch": 0.9202992843201041,
+      "grad_norm": 0.19378246366977692,
+      "learning_rate": 0.0001,
+      "loss": 1.6625,
+      "step": 1886
+    },
+    {
+      "epoch": 0.9207872478854913,
+      "grad_norm": 0.1863091140985489,
+      "learning_rate": 0.0001,
+      "loss": 1.7335,
+      "step": 1887
+    },
+    {
+      "epoch": 0.9212752114508783,
+      "grad_norm": 0.20152948796749115,
+      "learning_rate": 0.0001,
+      "loss": 1.7632,
+      "step": 1888
+    },
+    {
+      "epoch": 0.9217631750162655,
+      "grad_norm": 0.1818552315235138,
+      "learning_rate": 0.0001,
+      "loss": 1.5801,
+      "step": 1889
+    },
+    {
+      "epoch": 0.9222511385816525,
+      "grad_norm": 0.19405800104141235,
+      "learning_rate": 0.0001,
+      "loss": 1.6958,
+      "step": 1890
+    },
+    {
+      "epoch": 0.9227391021470397,
+      "grad_norm": 0.1904546320438385,
+      "learning_rate": 0.0001,
+      "loss": 1.7213,
+      "step": 1891
+    },
+    {
+      "epoch": 0.9232270657124269,
+      "grad_norm": 0.18888206779956818,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 1892
+    },
+    {
+      "epoch": 0.9237150292778139,
+      "grad_norm": 0.1954280138015747,
+      "learning_rate": 0.0001,
+      "loss": 1.6506,
+      "step": 1893
+    },
+    {
+      "epoch": 0.9242029928432011,
+      "grad_norm": 0.19753329455852509,
+      "learning_rate": 0.0001,
+      "loss": 1.5673,
+      "step": 1894
+    },
+    {
+      "epoch": 0.9246909564085881,
+      "grad_norm": 0.1822923868894577,
+      "learning_rate": 0.0001,
+      "loss": 1.6265,
+      "step": 1895
+    },
+    {
+      "epoch": 0.9251789199739753,
+      "grad_norm": 0.18649877607822418,
+      "learning_rate": 0.0001,
+      "loss": 1.6065,
+      "step": 1896
+    },
+    {
+      "epoch": 0.9256668835393624,
+      "grad_norm": 0.19342879951000214,
+      "learning_rate": 0.0001,
+      "loss": 1.6686,
+      "step": 1897
+    },
+    {
+      "epoch": 0.9261548471047495,
+      "grad_norm": 0.1949494183063507,
+      "learning_rate": 0.0001,
+      "loss": 1.7297,
+      "step": 1898
+    },
+    {
+      "epoch": 0.9266428106701367,
+      "grad_norm": 0.18507172167301178,
+      "learning_rate": 0.0001,
+      "loss": 1.7023,
+      "step": 1899
+    },
+    {
+      "epoch": 0.9271307742355237,
+      "grad_norm": 0.19422446191310883,
+      "learning_rate": 0.0001,
+      "loss": 1.7484,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9276187378009109,
+      "grad_norm": 0.18911758065223694,
+      "learning_rate": 0.0001,
+      "loss": 1.6291,
+      "step": 1901
+    },
+    {
+      "epoch": 0.928106701366298,
+      "grad_norm": 0.18500325083732605,
+      "learning_rate": 0.0001,
+      "loss": 1.6286,
+      "step": 1902
+    },
+    {
+      "epoch": 0.9285946649316851,
+      "grad_norm": 0.1928795874118805,
+      "learning_rate": 0.0001,
+      "loss": 1.6231,
+      "step": 1903
+    },
+    {
+      "epoch": 0.9290826284970722,
+      "grad_norm": 0.187940314412117,
+      "learning_rate": 0.0001,
+      "loss": 1.4776,
+      "step": 1904
+    },
+    {
+      "epoch": 0.9295705920624593,
+      "grad_norm": 0.19200646877288818,
+      "learning_rate": 0.0001,
+      "loss": 1.6558,
+      "step": 1905
+    },
+    {
+      "epoch": 0.9300585556278465,
+      "grad_norm": 0.19187362492084503,
+      "learning_rate": 0.0001,
+      "loss": 1.732,
+      "step": 1906
+    },
+    {
+      "epoch": 0.9305465191932336,
+      "grad_norm": 0.18200846016407013,
+      "learning_rate": 0.0001,
+      "loss": 1.6865,
+      "step": 1907
+    },
+    {
+      "epoch": 0.9310344827586207,
+      "grad_norm": 0.20386971533298492,
+      "learning_rate": 0.0001,
+      "loss": 1.6716,
+      "step": 1908
+    },
+    {
+      "epoch": 0.9315224463240078,
+      "grad_norm": 0.18020214140415192,
+      "learning_rate": 0.0001,
+      "loss": 1.5867,
+      "step": 1909
+    },
+    {
+      "epoch": 0.9320104098893949,
+      "grad_norm": 0.19435757398605347,
+      "learning_rate": 0.0001,
+      "loss": 1.7309,
+      "step": 1910
+    },
+    {
+      "epoch": 0.932498373454782,
+      "grad_norm": 0.19285395741462708,
+      "learning_rate": 0.0001,
+      "loss": 1.6832,
+      "step": 1911
+    },
+    {
+      "epoch": 0.9329863370201692,
+      "grad_norm": 0.19747254252433777,
+      "learning_rate": 0.0001,
+      "loss": 1.6816,
+      "step": 1912
+    },
+    {
+      "epoch": 0.9334743005855562,
+      "grad_norm": 0.18839652836322784,
+      "learning_rate": 0.0001,
+      "loss": 1.7361,
+      "step": 1913
+    },
+    {
+      "epoch": 0.9339622641509434,
+      "grad_norm": 0.18727315962314606,
+      "learning_rate": 0.0001,
+      "loss": 1.6448,
+      "step": 1914
+    },
+    {
+      "epoch": 0.9344502277163305,
+      "grad_norm": 0.17452287673950195,
+      "learning_rate": 0.0001,
+      "loss": 1.555,
+      "step": 1915
+    },
+    {
+      "epoch": 0.9349381912817176,
+      "grad_norm": 0.19202987849712372,
+      "learning_rate": 0.0001,
+      "loss": 1.7024,
+      "step": 1916
+    },
+    {
+      "epoch": 0.9354261548471048,
+      "grad_norm": 0.18424364924430847,
+      "learning_rate": 0.0001,
+      "loss": 1.7268,
+      "step": 1917
+    },
+    {
+      "epoch": 0.9359141184124918,
+      "grad_norm": 0.1959637701511383,
+      "learning_rate": 0.0001,
+      "loss": 1.7705,
+      "step": 1918
+    },
+    {
+      "epoch": 0.936402081977879,
+      "grad_norm": 0.19204477965831757,
+      "learning_rate": 0.0001,
+      "loss": 1.6194,
+      "step": 1919
+    },
+    {
+      "epoch": 0.936890045543266,
+      "grad_norm": 0.22303852438926697,
+      "learning_rate": 0.0001,
+      "loss": 1.6484,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9373780091086532,
+      "grad_norm": 0.1938052773475647,
+      "learning_rate": 0.0001,
+      "loss": 1.6094,
+      "step": 1921
+    },
+    {
+      "epoch": 0.9378659726740404,
+      "grad_norm": 0.20095346868038177,
+      "learning_rate": 0.0001,
+      "loss": 1.6189,
+      "step": 1922
+    },
+    {
+      "epoch": 0.9383539362394274,
+      "grad_norm": 0.20438328385353088,
+      "learning_rate": 0.0001,
+      "loss": 1.6793,
+      "step": 1923
+    },
+    {
+      "epoch": 0.9388418998048146,
+      "grad_norm": 0.1934322714805603,
+      "learning_rate": 0.0001,
+      "loss": 1.6434,
+      "step": 1924
+    },
+    {
+      "epoch": 0.9393298633702017,
+      "grad_norm": 0.1885659545660019,
+      "learning_rate": 0.0001,
+      "loss": 1.4937,
+      "step": 1925
+    },
+    {
+      "epoch": 0.9398178269355888,
+      "grad_norm": 0.18867121636867523,
+      "learning_rate": 0.0001,
+      "loss": 1.6838,
+      "step": 1926
+    },
+    {
+      "epoch": 0.940305790500976,
+      "grad_norm": 0.19884291291236877,
+      "learning_rate": 0.0001,
+      "loss": 1.7175,
+      "step": 1927
+    },
+    {
+      "epoch": 0.940793754066363,
+      "grad_norm": 0.19754834473133087,
+      "learning_rate": 0.0001,
+      "loss": 1.6299,
+      "step": 1928
+    },
+    {
+      "epoch": 0.9412817176317502,
+      "grad_norm": 0.20804598927497864,
+      "learning_rate": 0.0001,
+      "loss": 1.7577,
+      "step": 1929
+    },
+    {
+      "epoch": 0.9417696811971373,
+      "grad_norm": 0.1934799700975418,
+      "learning_rate": 0.0001,
+      "loss": 1.6434,
+      "step": 1930
+    },
+    {
+      "epoch": 0.9422576447625244,
+      "grad_norm": 0.19872422516345978,
+      "learning_rate": 0.0001,
+      "loss": 1.5745,
+      "step": 1931
+    },
+    {
+      "epoch": 0.9427456083279115,
+      "grad_norm": 0.19481562077999115,
+      "learning_rate": 0.0001,
+      "loss": 1.683,
+      "step": 1932
+    },
+    {
+      "epoch": 0.9432335718932986,
+      "grad_norm": 0.18682962656021118,
+      "learning_rate": 0.0001,
+      "loss": 1.5869,
+      "step": 1933
+    },
+    {
+      "epoch": 0.9437215354586858,
+      "grad_norm": 0.19617128372192383,
+      "learning_rate": 0.0001,
+      "loss": 1.6752,
+      "step": 1934
+    },
+    {
+      "epoch": 0.9442094990240729,
+      "grad_norm": 0.18968826532363892,
+      "learning_rate": 0.0001,
+      "loss": 1.637,
+      "step": 1935
+    },
+    {
+      "epoch": 0.94469746258946,
+      "grad_norm": 0.20613940060138702,
+      "learning_rate": 0.0001,
+      "loss": 1.7039,
+      "step": 1936
+    },
+    {
+      "epoch": 0.9451854261548471,
+      "grad_norm": 0.2038343995809555,
+      "learning_rate": 0.0001,
+      "loss": 1.7396,
+      "step": 1937
+    },
+    {
+      "epoch": 0.9456733897202342,
+      "grad_norm": 0.19330835342407227,
+      "learning_rate": 0.0001,
+      "loss": 1.6096,
+      "step": 1938
+    },
+    {
+      "epoch": 0.9461613532856213,
+      "grad_norm": 0.21375514566898346,
+      "learning_rate": 0.0001,
+      "loss": 1.7174,
+      "step": 1939
+    },
+    {
+      "epoch": 0.9466493168510085,
+      "grad_norm": 0.20083387196063995,
+      "learning_rate": 0.0001,
+      "loss": 1.5922,
+      "step": 1940
+    },
+    {
+      "epoch": 0.9471372804163956,
+      "grad_norm": 0.18656747043132782,
+      "learning_rate": 0.0001,
+      "loss": 1.6689,
+      "step": 1941
+    },
+    {
+      "epoch": 0.9476252439817827,
+      "grad_norm": 0.19367806613445282,
+      "learning_rate": 0.0001,
+      "loss": 1.6574,
+      "step": 1942
+    },
+    {
+      "epoch": 0.9481132075471698,
+      "grad_norm": 0.19045981764793396,
+      "learning_rate": 0.0001,
+      "loss": 1.6072,
+      "step": 1943
+    },
+    {
+      "epoch": 0.9486011711125569,
+      "grad_norm": 0.19302068650722504,
+      "learning_rate": 0.0001,
+      "loss": 1.6066,
+      "step": 1944
+    },
+    {
+      "epoch": 0.9490891346779441,
+      "grad_norm": 0.1919882893562317,
+      "learning_rate": 0.0001,
+      "loss": 1.6674,
+      "step": 1945
+    },
+    {
+      "epoch": 0.9495770982433311,
+      "grad_norm": 0.18873807787895203,
+      "learning_rate": 0.0001,
+      "loss": 1.5804,
+      "step": 1946
+    },
+    {
+      "epoch": 0.9500650618087183,
+      "grad_norm": 0.18615473806858063,
+      "learning_rate": 0.0001,
+      "loss": 1.4677,
+      "step": 1947
+    },
+    {
+      "epoch": 0.9505530253741054,
+      "grad_norm": 0.196221262216568,
+      "learning_rate": 0.0001,
+      "loss": 1.7643,
+      "step": 1948
+    },
+    {
+      "epoch": 0.9510409889394925,
+      "grad_norm": 0.19638247787952423,
+      "learning_rate": 0.0001,
+      "loss": 1.7174,
+      "step": 1949
+    },
+    {
+      "epoch": 0.9515289525048797,
+      "grad_norm": 0.18977941572666168,
+      "learning_rate": 0.0001,
+      "loss": 1.7463,
+      "step": 1950
+    },
+    {
+      "epoch": 0.9520169160702667,
+      "grad_norm": 0.18567734956741333,
+      "learning_rate": 0.0001,
+      "loss": 1.6243,
+      "step": 1951
+    },
+    {
+      "epoch": 0.9525048796356539,
+      "grad_norm": 0.1937309354543686,
+      "learning_rate": 0.0001,
+      "loss": 1.6272,
+      "step": 1952
+    },
+    {
+      "epoch": 0.9529928432010409,
+      "grad_norm": 0.18985025584697723,
+      "learning_rate": 0.0001,
+      "loss": 1.5859,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9534808067664281,
+      "grad_norm": 0.19192877411842346,
+      "learning_rate": 0.0001,
+      "loss": 1.6615,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9539687703318153,
+      "grad_norm": 0.18933430314064026,
+      "learning_rate": 0.0001,
+      "loss": 1.6925,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9544567338972023,
+      "grad_norm": 0.18384099006652832,
+      "learning_rate": 0.0001,
+      "loss": 1.6519,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9549446974625895,
+      "grad_norm": 0.187970370054245,
+      "learning_rate": 0.0001,
+      "loss": 1.6444,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9554326610279765,
+      "grad_norm": 0.18869946897029877,
+      "learning_rate": 0.0001,
+      "loss": 1.661,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9559206245933637,
+      "grad_norm": 0.18695271015167236,
+      "learning_rate": 0.0001,
+      "loss": 1.711,
+      "step": 1959
+    },
+    {
+      "epoch": 0.9564085881587508,
+      "grad_norm": 0.18627239763736725,
+      "learning_rate": 0.0001,
+      "loss": 1.6396,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9568965517241379,
+      "grad_norm": 0.1865292340517044,
+      "learning_rate": 0.0001,
+      "loss": 1.6333,
+      "step": 1961
+    },
+    {
+      "epoch": 0.9573845152895251,
+      "grad_norm": 0.19713494181632996,
+      "learning_rate": 0.0001,
+      "loss": 1.6997,
+      "step": 1962
+    },
+    {
+      "epoch": 0.9578724788549121,
+      "grad_norm": 0.19183889031410217,
+      "learning_rate": 0.0001,
+      "loss": 1.6611,
+      "step": 1963
+    },
+    {
+      "epoch": 0.9583604424202993,
+      "grad_norm": 0.1824401468038559,
+      "learning_rate": 0.0001,
+      "loss": 1.6543,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9588484059856864,
+      "grad_norm": 0.19251850247383118,
+      "learning_rate": 0.0001,
+      "loss": 1.6621,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9593363695510735,
+      "grad_norm": 0.188092902302742,
+      "learning_rate": 0.0001,
+      "loss": 1.5799,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9598243331164606,
+      "grad_norm": 0.2004861980676651,
+      "learning_rate": 0.0001,
+      "loss": 1.829,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9603122966818478,
+      "grad_norm": 0.18226411938667297,
+      "learning_rate": 0.0001,
+      "loss": 1.6785,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9608002602472349,
+      "grad_norm": 0.18180397152900696,
+      "learning_rate": 0.0001,
+      "loss": 1.6726,
+      "step": 1969
+    },
+    {
+      "epoch": 0.961288223812622,
+      "grad_norm": 0.18457727134227753,
+      "learning_rate": 0.0001,
+      "loss": 1.6192,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9617761873780091,
+      "grad_norm": 0.18507972359657288,
+      "learning_rate": 0.0001,
+      "loss": 1.4573,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9622641509433962,
+      "grad_norm": 0.17920102179050446,
+      "learning_rate": 0.0001,
+      "loss": 1.5538,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9627521145087834,
+      "grad_norm": 0.18464797735214233,
+      "learning_rate": 0.0001,
+      "loss": 1.6241,
+      "step": 1973
+    },
+    {
+      "epoch": 0.9632400780741704,
+      "grad_norm": 0.18835680186748505,
+      "learning_rate": 0.0001,
+      "loss": 1.6425,
+      "step": 1974
+    },
+    {
+      "epoch": 0.9637280416395576,
+      "grad_norm": 0.19201530516147614,
+      "learning_rate": 0.0001,
+      "loss": 1.5636,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9642160052049447,
+      "grad_norm": 0.1867157518863678,
+      "learning_rate": 0.0001,
+      "loss": 1.7102,
+      "step": 1976
+    },
+    {
+      "epoch": 0.9647039687703318,
+      "grad_norm": 0.18558953702449799,
+      "learning_rate": 0.0001,
+      "loss": 1.6129,
+      "step": 1977
+    },
+    {
+      "epoch": 0.965191932335719,
+      "grad_norm": 0.18594802916049957,
+      "learning_rate": 0.0001,
+      "loss": 1.6562,
+      "step": 1978
+    },
+    {
+      "epoch": 0.965679895901106,
+      "grad_norm": 0.18944233655929565,
+      "learning_rate": 0.0001,
+      "loss": 1.6166,
+      "step": 1979
+    },
+    {
+      "epoch": 0.9661678594664932,
+      "grad_norm": 0.19270744919776917,
+      "learning_rate": 0.0001,
+      "loss": 1.7154,
+      "step": 1980
+    },
+    {
+      "epoch": 0.9666558230318802,
+      "grad_norm": 0.18543404340744019,
+      "learning_rate": 0.0001,
+      "loss": 1.6351,
+      "step": 1981
+    },
+    {
+      "epoch": 0.9671437865972674,
+      "grad_norm": 0.1873311698436737,
+      "learning_rate": 0.0001,
+      "loss": 1.6802,
+      "step": 1982
+    },
+    {
+      "epoch": 0.9676317501626546,
+      "grad_norm": 0.20122510194778442,
+      "learning_rate": 0.0001,
+      "loss": 1.6775,
+      "step": 1983
+    },
+    {
+      "epoch": 0.9681197137280416,
+      "grad_norm": 0.18540111184120178,
+      "learning_rate": 0.0001,
+      "loss": 1.71,
+      "step": 1984
+    },
+    {
+      "epoch": 0.9686076772934288,
+      "grad_norm": 0.19041728973388672,
+      "learning_rate": 0.0001,
+      "loss": 1.6158,
+      "step": 1985
+    },
+    {
+      "epoch": 0.9690956408588158,
+      "grad_norm": 0.18193362653255463,
+      "learning_rate": 0.0001,
+      "loss": 1.5784,
+      "step": 1986
+    },
+    {
+      "epoch": 0.969583604424203,
+      "grad_norm": 0.18553735315799713,
+      "learning_rate": 0.0001,
+      "loss": 1.5646,
+      "step": 1987
+    },
+    {
+      "epoch": 0.9700715679895902,
+      "grad_norm": 0.18669581413269043,
+      "learning_rate": 0.0001,
+      "loss": 1.6766,
+      "step": 1988
+    },
+    {
+      "epoch": 0.9705595315549772,
+      "grad_norm": 0.18920449912548065,
+      "learning_rate": 0.0001,
+      "loss": 1.6766,
+      "step": 1989
+    },
+    {
+      "epoch": 0.9710474951203644,
+      "grad_norm": 0.20458273589611053,
+      "learning_rate": 0.0001,
+      "loss": 1.7894,
+      "step": 1990
+    },
+    {
+      "epoch": 0.9715354586857514,
+      "grad_norm": 0.1898808777332306,
+      "learning_rate": 0.0001,
+      "loss": 1.6446,
+      "step": 1991
+    },
+    {
+      "epoch": 0.9720234222511386,
+      "grad_norm": 0.2061244398355484,
+      "learning_rate": 0.0001,
+      "loss": 1.7023,
+      "step": 1992
+    },
+    {
+      "epoch": 0.9725113858165257,
+      "grad_norm": 0.1939939558506012,
+      "learning_rate": 0.0001,
+      "loss": 1.6788,
+      "step": 1993
+    },
+    {
+      "epoch": 0.9729993493819128,
+      "grad_norm": 0.18010011315345764,
+      "learning_rate": 0.0001,
+      "loss": 1.6432,
+      "step": 1994
+    },
+    {
+      "epoch": 0.9734873129473,
+      "grad_norm": 0.1993561089038849,
+      "learning_rate": 0.0001,
+      "loss": 1.703,
+      "step": 1995
+    },
+    {
+      "epoch": 0.973975276512687,
+      "grad_norm": 0.2100955992937088,
+      "learning_rate": 0.0001,
+      "loss": 1.8338,
+      "step": 1996
+    },
+    {
+      "epoch": 0.9744632400780742,
+      "grad_norm": 0.18871144950389862,
+      "learning_rate": 0.0001,
+      "loss": 1.5936,
+      "step": 1997
+    },
+    {
+      "epoch": 0.9749512036434613,
+      "grad_norm": 0.18759900331497192,
+      "learning_rate": 0.0001,
+      "loss": 1.5765,
+      "step": 1998
+    },
+    {
+      "epoch": 0.9754391672088484,
+      "grad_norm": 0.18255159258842468,
+      "learning_rate": 0.0001,
+      "loss": 1.565,
+      "step": 1999
+    },
+    {
+      "epoch": 0.9759271307742355,
+      "grad_norm": 0.19834022223949432,
+      "learning_rate": 0.0001,
+      "loss": 1.7178,
+      "step": 2000
+    },
+    {
+      "epoch": 0.9764150943396226,
+      "grad_norm": 0.1984955072402954,
+      "learning_rate": 0.0001,
+      "loss": 1.6893,
+      "step": 2001
+    },
+    {
+      "epoch": 0.9769030579050098,
+      "grad_norm": 0.19068273901939392,
+      "learning_rate": 0.0001,
+      "loss": 1.5748,
+      "step": 2002
+    },
+    {
+      "epoch": 0.9773910214703969,
+      "grad_norm": 0.18630477786064148,
+      "learning_rate": 0.0001,
+      "loss": 1.524,
+      "step": 2003
+    },
+    {
+      "epoch": 0.977878985035784,
+      "grad_norm": 0.19503630697727203,
+      "learning_rate": 0.0001,
+      "loss": 1.8031,
+      "step": 2004
+    },
+    {
+      "epoch": 0.9783669486011711,
+      "grad_norm": 0.19002202153205872,
+      "learning_rate": 0.0001,
+      "loss": 1.7034,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9788549121665582,
+      "grad_norm": 0.18310774862766266,
+      "learning_rate": 0.0001,
+      "loss": 1.6308,
+      "step": 2006
+    },
+    {
+      "epoch": 0.9793428757319453,
+      "grad_norm": 0.19140413403511047,
+      "learning_rate": 0.0001,
+      "loss": 1.6739,
+      "step": 2007
+    },
+    {
+      "epoch": 0.9798308392973325,
+      "grad_norm": 0.19118629395961761,
+      "learning_rate": 0.0001,
+      "loss": 1.8085,
+      "step": 2008
+    },
+    {
+      "epoch": 0.9803188028627196,
+      "grad_norm": 0.180838942527771,
+      "learning_rate": 0.0001,
+      "loss": 1.6047,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9808067664281067,
+      "grad_norm": 0.19258812069892883,
+      "learning_rate": 0.0001,
+      "loss": 1.6151,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9812947299934938,
+      "grad_norm": 0.18637025356292725,
+      "learning_rate": 0.0001,
+      "loss": 1.6767,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9817826935588809,
+      "grad_norm": 0.19936662912368774,
+      "learning_rate": 0.0001,
+      "loss": 1.6625,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9822706571242681,
+      "grad_norm": 0.194144606590271,
+      "learning_rate": 0.0001,
+      "loss": 1.7133,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9827586206896551,
+      "grad_norm": 0.20041455328464508,
+      "learning_rate": 0.0001,
+      "loss": 1.6546,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9832465842550423,
+      "grad_norm": 0.19850897789001465,
+      "learning_rate": 0.0001,
+      "loss": 1.7541,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9837345478204295,
+      "grad_norm": 0.18957051634788513,
+      "learning_rate": 0.0001,
+      "loss": 1.5824,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9842225113858165,
+      "grad_norm": 0.19494563341140747,
+      "learning_rate": 0.0001,
+      "loss": 1.6266,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9847104749512037,
+      "grad_norm": 0.1976606249809265,
+      "learning_rate": 0.0001,
+      "loss": 1.7235,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9851984385165907,
+      "grad_norm": 0.1937410980463028,
+      "learning_rate": 0.0001,
+      "loss": 1.635,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9856864020819779,
+      "grad_norm": 0.2040870636701584,
+      "learning_rate": 0.0001,
+      "loss": 1.5867,
+      "step": 2020
+    },
+    {
+      "epoch": 0.986174365647365,
+      "grad_norm": 0.18701307475566864,
+      "learning_rate": 0.0001,
+      "loss": 1.7137,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9866623292127521,
+      "grad_norm": 0.19137197732925415,
+      "learning_rate": 0.0001,
+      "loss": 1.7141,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9871502927781393,
+      "grad_norm": 0.18723750114440918,
+      "learning_rate": 0.0001,
+      "loss": 1.5555,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9876382563435263,
+      "grad_norm": 0.19908587634563446,
+      "learning_rate": 0.0001,
+      "loss": 1.7694,
+      "step": 2024
+    },
+    {
+      "epoch": 0.9881262199089135,
+      "grad_norm": 0.18599991500377655,
+      "learning_rate": 0.0001,
+      "loss": 1.6416,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9886141834743006,
+      "grad_norm": 0.19032098352909088,
+      "learning_rate": 0.0001,
+      "loss": 1.725,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9891021470396877,
+      "grad_norm": 0.19751659035682678,
+      "learning_rate": 0.0001,
+      "loss": 1.6269,
+      "step": 2027
+    },
+    {
+      "epoch": 0.9895901106050748,
+      "grad_norm": 0.19484581053256989,
+      "learning_rate": 0.0001,
+      "loss": 1.6075,
+      "step": 2028
+    },
+    {
+      "epoch": 0.9900780741704619,
+      "grad_norm": 0.18925638496875763,
+      "learning_rate": 0.0001,
+      "loss": 1.6035,
+      "step": 2029
+    },
+    {
+      "epoch": 0.9905660377358491,
+      "grad_norm": 0.1926533430814743,
+      "learning_rate": 0.0001,
+      "loss": 1.6842,
+      "step": 2030
+    },
+    {
+      "epoch": 0.9910540013012362,
+      "grad_norm": 0.18332500755786896,
+      "learning_rate": 0.0001,
+      "loss": 1.6314,
+      "step": 2031
+    },
+    {
+      "epoch": 0.9915419648666233,
+      "grad_norm": 0.19438926875591278,
+      "learning_rate": 0.0001,
+      "loss": 1.7524,
+      "step": 2032
+    },
+    {
+      "epoch": 0.9920299284320104,
+      "grad_norm": 0.18711014091968536,
+      "learning_rate": 0.0001,
+      "loss": 1.5707,
+      "step": 2033
+    },
+    {
+      "epoch": 0.9925178919973975,
+      "grad_norm": 0.1909669041633606,
+      "learning_rate": 0.0001,
+      "loss": 1.6926,
+      "step": 2034
+    },
+    {
+      "epoch": 0.9930058555627846,
+      "grad_norm": 0.187711700797081,
+      "learning_rate": 0.0001,
+      "loss": 1.571,
+      "step": 2035
+    },
+    {
+      "epoch": 0.9934938191281718,
+      "grad_norm": 0.18141263723373413,
+      "learning_rate": 0.0001,
+      "loss": 1.6505,
+      "step": 2036
+    },
+    {
+      "epoch": 0.9939817826935589,
+      "grad_norm": 0.18751592934131622,
+      "learning_rate": 0.0001,
+      "loss": 1.5944,
+      "step": 2037
+    },
+    {
+      "epoch": 0.994469746258946,
+      "grad_norm": 0.18346115946769714,
+      "learning_rate": 0.0001,
+      "loss": 1.6837,
+      "step": 2038
+    },
+    {
+      "epoch": 0.9949577098243331,
+      "grad_norm": 0.1962192952632904,
+      "learning_rate": 0.0001,
+      "loss": 1.7056,
+      "step": 2039
+    },
+    {
+      "epoch": 0.9954456733897202,
+      "grad_norm": 0.19639593362808228,
+      "learning_rate": 0.0001,
+      "loss": 1.6419,
+      "step": 2040
+    },
+    {
+      "epoch": 0.9959336369551074,
+      "grad_norm": 0.19418394565582275,
+      "learning_rate": 0.0001,
+      "loss": 1.7652,
+      "step": 2041
+    },
+    {
+      "epoch": 0.9964216005204944,
+      "grad_norm": 0.19044318795204163,
+      "learning_rate": 0.0001,
+      "loss": 1.6707,
+      "step": 2042
+    },
+    {
+      "epoch": 0.9969095640858816,
+      "grad_norm": 0.1870298981666565,
+      "learning_rate": 0.0001,
+      "loss": 1.521,
+      "step": 2043
+    },
+    {
+      "epoch": 0.9973975276512687,
+      "grad_norm": 0.20216675102710724,
+      "learning_rate": 0.0001,
+      "loss": 1.6267,
+      "step": 2044
+    },
+    {
+      "epoch": 0.9978854912166558,
+      "grad_norm": 0.1920004040002823,
+      "learning_rate": 0.0001,
+      "loss": 1.626,
+      "step": 2045
+    },
+    {
+      "epoch": 0.998373454782043,
+      "grad_norm": 0.20293602347373962,
+      "learning_rate": 0.0001,
+      "loss": 1.5053,
+      "step": 2046
+    },
+    {
+      "epoch": 0.99886141834743,
+      "grad_norm": 0.2009141743183136,
+      "learning_rate": 0.0001,
+      "loss": 1.6191,
+      "step": 2047
+    },
+    {
+      "epoch": 0.9993493819128172,
+      "grad_norm": 0.1968158483505249,
+      "learning_rate": 0.0001,
+      "loss": 1.6945,
+      "step": 2048
+    },
+    {
+      "epoch": 0.9998373454782042,
+      "grad_norm": 0.2032029926776886,
+      "learning_rate": 0.0001,
+      "loss": 1.6321,
+      "step": 2049
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2049,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 300,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.374213438189863e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}