diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4704 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9943851768669285,
+  "eval_steps": 500,
+  "global_step": 666,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004491858506457047,
+      "grad_norm": 5.744765553310083,
+      "learning_rate": 4.2215328358208957e-07,
+      "loss": 0.8286,
+      "step": 1
+    },
+    {
+      "epoch": 0.008983717012914094,
+      "grad_norm": 5.938427190328152,
+      "learning_rate": 8.443065671641791e-07,
+      "loss": 0.8731,
+      "step": 2
+    },
+    {
+      "epoch": 0.01347557551937114,
+      "grad_norm": 5.7787941236611164,
+      "learning_rate": 1.2664598507462685e-06,
+      "loss": 0.8549,
+      "step": 3
+    },
+    {
+      "epoch": 0.017967434025828188,
+      "grad_norm": 5.730303354635296,
+      "learning_rate": 1.6886131343283583e-06,
+      "loss": 0.8641,
+      "step": 4
+    },
+    {
+      "epoch": 0.022459292532285232,
+      "grad_norm": 5.463956060746314,
+      "learning_rate": 2.1107664179104476e-06,
+      "loss": 0.859,
+      "step": 5
+    },
+    {
+      "epoch": 0.02695115103874228,
+      "grad_norm": 4.359222218466504,
+      "learning_rate": 2.532919701492537e-06,
+      "loss": 0.8086,
+      "step": 6
+    },
+    {
+      "epoch": 0.031443009545199324,
+      "grad_norm": 3.9314742540418877,
+      "learning_rate": 2.955072985074627e-06,
+      "loss": 0.7987,
+      "step": 7
+    },
+    {
+      "epoch": 0.035934868051656375,
+      "grad_norm": 2.2948652467354904,
+      "learning_rate": 3.3772262686567165e-06,
+      "loss": 0.768,
+      "step": 8
+    },
+    {
+      "epoch": 0.04042672655811342,
+      "grad_norm": 1.9963513578345484,
+      "learning_rate": 3.799379552238806e-06,
+      "loss": 0.7609,
+      "step": 9
+    },
+    {
+      "epoch": 0.044918585064570464,
+      "grad_norm": 1.7047152492699387,
+      "learning_rate": 4.221532835820895e-06,
+      "loss": 0.7504,
+      "step": 10
+    },
+    {
+      "epoch": 0.049410443571027515,
+      "grad_norm": 3.8017193994614322,
+      "learning_rate": 4.6436861194029855e-06,
+      "loss": 0.7503,
+      "step": 11
+    },
+    {
+      "epoch": 0.05390230207748456,
+      "grad_norm": 4.022894465311438,
+      "learning_rate": 5.065839402985074e-06,
+      "loss": 0.7455,
+      "step": 12
+    },
+    {
+      "epoch": 0.058394160583941604,
+      "grad_norm": 3.9288586911775494,
+      "learning_rate": 5.4879926865671645e-06,
+      "loss": 0.7295,
+      "step": 13
+    },
+    {
+      "epoch": 0.06288601909039865,
+      "grad_norm": 3.209589214583598,
+      "learning_rate": 5.910145970149254e-06,
+      "loss": 0.7024,
+      "step": 14
+    },
+    {
+      "epoch": 0.06737787759685569,
+      "grad_norm": 3.011893288172853,
+      "learning_rate": 6.3322992537313436e-06,
+      "loss": 0.6876,
+      "step": 15
+    },
+    {
+      "epoch": 0.07186973610331275,
+      "grad_norm": 2.414862038960305,
+      "learning_rate": 6.754452537313433e-06,
+      "loss": 0.6812,
+      "step": 16
+    },
+    {
+      "epoch": 0.0763615946097698,
+      "grad_norm": 1.6414783160484103,
+      "learning_rate": 7.176605820895523e-06,
+      "loss": 0.6814,
+      "step": 17
+    },
+    {
+      "epoch": 0.08085345311622684,
+      "grad_norm": 1.369475527415624,
+      "learning_rate": 7.598759104477612e-06,
+      "loss": 0.659,
+      "step": 18
+    },
+    {
+      "epoch": 0.08534531162268388,
+      "grad_norm": 1.6743202955313934,
+      "learning_rate": 8.0209123880597e-06,
+      "loss": 0.6456,
+      "step": 19
+    },
+    {
+      "epoch": 0.08983717012914093,
+      "grad_norm": 1.8668298838147386,
+      "learning_rate": 8.44306567164179e-06,
+      "loss": 0.6516,
+      "step": 20
+    },
+    {
+      "epoch": 0.09432902863559797,
+      "grad_norm": 1.4484409449750104,
+      "learning_rate": 8.865218955223881e-06,
+      "loss": 0.6304,
+      "step": 21
+    },
+    {
+      "epoch": 0.09882088714205503,
+      "grad_norm": 1.0037439652654327,
+      "learning_rate": 9.287372238805971e-06,
+      "loss": 0.6245,
+      "step": 22
+    },
+    {
+      "epoch": 0.10331274564851207,
+      "grad_norm": 1.2035026989163056,
+      "learning_rate": 9.70952552238806e-06,
+      "loss": 0.621,
+      "step": 23
+    },
+    {
+      "epoch": 0.10780460415496912,
+      "grad_norm": 1.2996895063130545,
+      "learning_rate": 1.0131678805970148e-05,
+      "loss": 0.6092,
+      "step": 24
+    },
+    {
+      "epoch": 0.11229646266142616,
+      "grad_norm": 0.9219635804670414,
+      "learning_rate": 1.055383208955224e-05,
+      "loss": 0.6082,
+      "step": 25
+    },
+    {
+      "epoch": 0.11678832116788321,
+      "grad_norm": 0.8154512278475434,
+      "learning_rate": 1.0975985373134329e-05,
+      "loss": 0.585,
+      "step": 26
+    },
+    {
+      "epoch": 0.12128017967434025,
+      "grad_norm": 0.9247980377052808,
+      "learning_rate": 1.1398138656716419e-05,
+      "loss": 0.5938,
+      "step": 27
+    },
+    {
+      "epoch": 0.1257720381807973,
+      "grad_norm": 0.8351405332062142,
+      "learning_rate": 1.1820291940298508e-05,
+      "loss": 0.5915,
+      "step": 28
+    },
+    {
+      "epoch": 0.13026389668725435,
+      "grad_norm": 0.7133220578891004,
+      "learning_rate": 1.2242445223880596e-05,
+      "loss": 0.588,
+      "step": 29
+    },
+    {
+      "epoch": 0.13475575519371139,
+      "grad_norm": 0.7504890368369718,
+      "learning_rate": 1.2664598507462687e-05,
+      "loss": 0.5871,
+      "step": 30
+    },
+    {
+      "epoch": 0.13924761370016844,
+      "grad_norm": 0.7315285717243533,
+      "learning_rate": 1.3086751791044777e-05,
+      "loss": 0.5858,
+      "step": 31
+    },
+    {
+      "epoch": 0.1437394722066255,
+      "grad_norm": 0.5441735042606766,
+      "learning_rate": 1.3508905074626866e-05,
+      "loss": 0.575,
+      "step": 32
+    },
+    {
+      "epoch": 0.14823133071308253,
+      "grad_norm": 0.6473935352696674,
+      "learning_rate": 1.3931058358208956e-05,
+      "loss": 0.5615,
+      "step": 33
+    },
+    {
+      "epoch": 0.1527231892195396,
+      "grad_norm": 0.6223091865317807,
+      "learning_rate": 1.4353211641791045e-05,
+      "loss": 0.5686,
+      "step": 34
+    },
+    {
+      "epoch": 0.15721504772599662,
+      "grad_norm": 0.5869026473489213,
+      "learning_rate": 1.4775364925373133e-05,
+      "loss": 0.5632,
+      "step": 35
+    },
+    {
+      "epoch": 0.16170690623245368,
+      "grad_norm": 0.6697261661322835,
+      "learning_rate": 1.5197518208955224e-05,
+      "loss": 0.5538,
+      "step": 36
+    },
+    {
+      "epoch": 0.16619876473891074,
+      "grad_norm": 0.5523340874893558,
+      "learning_rate": 1.5619671492537315e-05,
+      "loss": 0.5515,
+      "step": 37
+    },
+    {
+      "epoch": 0.17069062324536777,
+      "grad_norm": 0.5402779001437888,
+      "learning_rate": 1.60418247761194e-05,
+      "loss": 0.5585,
+      "step": 38
+    },
+    {
+      "epoch": 0.17518248175182483,
+      "grad_norm": 0.6393793878404187,
+      "learning_rate": 1.6463978059701494e-05,
+      "loss": 0.5545,
+      "step": 39
+    },
+    {
+      "epoch": 0.17967434025828186,
+      "grad_norm": 0.4041240036369504,
+      "learning_rate": 1.688613134328358e-05,
+      "loss": 0.5578,
+      "step": 40
+    },
+    {
+      "epoch": 0.18416619876473891,
+      "grad_norm": 0.6086298220416736,
+      "learning_rate": 1.7308284626865673e-05,
+      "loss": 0.554,
+      "step": 41
+    },
+    {
+      "epoch": 0.18865805727119594,
+      "grad_norm": 0.3999870115396727,
+      "learning_rate": 1.7730437910447763e-05,
+      "loss": 0.5399,
+      "step": 42
+    },
+    {
+      "epoch": 0.193149915777653,
+      "grad_norm": 0.5006786688823199,
+      "learning_rate": 1.815259119402985e-05,
+      "loss": 0.5388,
+      "step": 43
+    },
+    {
+      "epoch": 0.19764177428411006,
+      "grad_norm": 0.5268906883702746,
+      "learning_rate": 1.8574744477611942e-05,
+      "loss": 0.5458,
+      "step": 44
+    },
+    {
+      "epoch": 0.2021336327905671,
+      "grad_norm": 0.36114074862039075,
+      "learning_rate": 1.899689776119403e-05,
+      "loss": 0.5318,
+      "step": 45
+    },
+    {
+      "epoch": 0.20662549129702415,
+      "grad_norm": 0.49219557077110226,
+      "learning_rate": 1.941905104477612e-05,
+      "loss": 0.538,
+      "step": 46
+    },
+    {
+      "epoch": 0.21111734980348118,
+      "grad_norm": 0.3898606142294089,
+      "learning_rate": 1.984120432835821e-05,
+      "loss": 0.5303,
+      "step": 47
+    },
+    {
+      "epoch": 0.21560920830993824,
+      "grad_norm": 0.4747550212627052,
+      "learning_rate": 2.0263357611940297e-05,
+      "loss": 0.5359,
+      "step": 48
+    },
+    {
+      "epoch": 0.2201010668163953,
+      "grad_norm": 0.356291193847858,
+      "learning_rate": 2.068551089552239e-05,
+      "loss": 0.5225,
+      "step": 49
+    },
+    {
+      "epoch": 0.22459292532285233,
+      "grad_norm": 0.5154384369701829,
+      "learning_rate": 2.110766417910448e-05,
+      "loss": 0.5432,
+      "step": 50
+    },
+    {
+      "epoch": 0.22908478382930939,
+      "grad_norm": 0.4156042173356652,
+      "learning_rate": 2.152981746268657e-05,
+      "loss": 0.5269,
+      "step": 51
+    },
+    {
+      "epoch": 0.23357664233576642,
+      "grad_norm": 0.34458484379208343,
+      "learning_rate": 2.1951970746268658e-05,
+      "loss": 0.5318,
+      "step": 52
+    },
+    {
+      "epoch": 0.23806850084222347,
+      "grad_norm": 0.45421913764112787,
+      "learning_rate": 2.2374124029850744e-05,
+      "loss": 0.5245,
+      "step": 53
+    },
+    {
+      "epoch": 0.2425603593486805,
+      "grad_norm": 0.567239004859855,
+      "learning_rate": 2.2796277313432837e-05,
+      "loss": 0.5289,
+      "step": 54
+    },
+    {
+      "epoch": 0.24705221785513756,
+      "grad_norm": 0.6740910902425077,
+      "learning_rate": 2.3218430597014927e-05,
+      "loss": 0.5084,
+      "step": 55
+    },
+    {
+      "epoch": 0.2515440763615946,
+      "grad_norm": 0.7752068297960009,
+      "learning_rate": 2.3640583880597016e-05,
+      "loss": 0.5226,
+      "step": 56
+    },
+    {
+      "epoch": 0.25603593486805165,
+      "grad_norm": 0.9552601356444903,
+      "learning_rate": 2.4062737164179106e-05,
+      "loss": 0.5196,
+      "step": 57
+    },
+    {
+      "epoch": 0.2605277933745087,
+      "grad_norm": 0.9379110300025817,
+      "learning_rate": 2.4484890447761192e-05,
+      "loss": 0.5105,
+      "step": 58
+    },
+    {
+      "epoch": 0.26501965188096577,
+      "grad_norm": 0.7772460143850226,
+      "learning_rate": 2.4907043731343285e-05,
+      "loss": 0.5242,
+      "step": 59
+    },
+    {
+      "epoch": 0.26951151038742277,
+      "grad_norm": 0.8788446944945193,
+      "learning_rate": 2.5329197014925374e-05,
+      "loss": 0.5139,
+      "step": 60
+    },
+    {
+      "epoch": 0.27400336889387983,
+      "grad_norm": 0.7472788464484211,
+      "learning_rate": 2.5751350298507464e-05,
+      "loss": 0.5087,
+      "step": 61
+    },
+    {
+      "epoch": 0.2784952274003369,
+      "grad_norm": 0.5206157571508017,
+      "learning_rate": 2.6173503582089553e-05,
+      "loss": 0.5135,
+      "step": 62
+    },
+    {
+      "epoch": 0.28298708590679394,
+      "grad_norm": 0.4158324584939451,
+      "learning_rate": 2.6595656865671643e-05,
+      "loss": 0.5221,
+      "step": 63
+    },
+    {
+      "epoch": 0.287478944413251,
+      "grad_norm": 0.6550327478832779,
+      "learning_rate": 2.7017810149253732e-05,
+      "loss": 0.5076,
+      "step": 64
+    },
+    {
+      "epoch": 0.291970802919708,
+      "grad_norm": 0.9256302053482004,
+      "learning_rate": 2.7439963432835822e-05,
+      "loss": 0.5127,
+      "step": 65
+    },
+    {
+      "epoch": 0.29646266142616506,
+      "grad_norm": 0.9775983835454853,
+      "learning_rate": 2.786211671641791e-05,
+      "loss": 0.5026,
+      "step": 66
+    },
+    {
+      "epoch": 0.3009545199326221,
+      "grad_norm": 1.51113067675059,
+      "learning_rate": 2.828427e-05,
+      "loss": 0.521,
+      "step": 67
+    },
+    {
+      "epoch": 0.3054463784390792,
+      "grad_norm": 0.8192228855206966,
+      "learning_rate": 2.8284075495305353e-05,
+      "loss": 0.501,
+      "step": 68
+    },
+    {
+      "epoch": 0.30993823694553624,
+      "grad_norm": 1.2300181702860546,
+      "learning_rate": 2.8283491986571674e-05,
+      "loss": 0.5127,
+      "step": 69
+    },
+    {
+      "epoch": 0.31443009545199324,
+      "grad_norm": 1.1329832003075657,
+      "learning_rate": 2.8282519489849603e-05,
+      "loss": 0.5071,
+      "step": 70
+    },
+    {
+      "epoch": 0.3189219539584503,
+      "grad_norm": 1.2646909380634277,
+      "learning_rate": 2.8281158031889725e-05,
+      "loss": 0.5053,
+      "step": 71
+    },
+    {
+      "epoch": 0.32341381246490736,
+      "grad_norm": 0.9837732589410637,
+      "learning_rate": 2.8279407650141834e-05,
+      "loss": 0.5197,
+      "step": 72
+    },
+    {
+      "epoch": 0.3279056709713644,
+      "grad_norm": 1.224216460317663,
+      "learning_rate": 2.8277268392753874e-05,
+      "loss": 0.5066,
+      "step": 73
+    },
+    {
+      "epoch": 0.3323975294778215,
+      "grad_norm": 0.8794559027371804,
+      "learning_rate": 2.827474031857067e-05,
+      "loss": 0.5064,
+      "step": 74
+    },
+    {
+      "epoch": 0.3368893879842785,
+      "grad_norm": 1.028650146078467,
+      "learning_rate": 2.8271823497132245e-05,
+      "loss": 0.5117,
+      "step": 75
+    },
+    {
+      "epoch": 0.34138124649073553,
+      "grad_norm": 1.0164722337507859,
+      "learning_rate": 2.8268518008671955e-05,
+      "loss": 0.4991,
+      "step": 76
+    },
+    {
+      "epoch": 0.3458731049971926,
+      "grad_norm": 1.024636665199281,
+      "learning_rate": 2.8264823944114274e-05,
+      "loss": 0.5125,
+      "step": 77
+    },
+    {
+      "epoch": 0.35036496350364965,
+      "grad_norm": 0.726482732351472,
+      "learning_rate": 2.826074140507226e-05,
+      "loss": 0.4926,
+      "step": 78
+    },
+    {
+      "epoch": 0.35485682201010665,
+      "grad_norm": 0.953436081285762,
+      "learning_rate": 2.825627050384482e-05,
+      "loss": 0.5061,
+      "step": 79
+    },
+    {
+      "epoch": 0.3593486805165637,
+      "grad_norm": 0.8241055859758966,
+      "learning_rate": 2.8251411363413542e-05,
+      "loss": 0.5074,
+      "step": 80
+    },
+    {
+      "epoch": 0.36384053902302077,
+      "grad_norm": 0.8014142170976486,
+      "learning_rate": 2.8246164117439403e-05,
+      "loss": 0.5065,
+      "step": 81
+    },
+    {
+      "epoch": 0.36833239752947783,
+      "grad_norm": 0.8877417141780828,
+      "learning_rate": 2.8240528910259e-05,
+      "loss": 0.5106,
+      "step": 82
+    },
+    {
+      "epoch": 0.3728242560359349,
+      "grad_norm": 0.7288874323602634,
+      "learning_rate": 2.823450589688067e-05,
+      "loss": 0.5025,
+      "step": 83
+    },
+    {
+      "epoch": 0.3773161145423919,
+      "grad_norm": 0.6009546992075813,
+      "learning_rate": 2.8228095242980137e-05,
+      "loss": 0.5006,
+      "step": 84
+    },
+    {
+      "epoch": 0.38180797304884895,
+      "grad_norm": 0.7490335675774521,
+      "learning_rate": 2.822129712489603e-05,
+      "loss": 0.5044,
+      "step": 85
+    },
+    {
+      "epoch": 0.386299831555306,
+      "grad_norm": 0.779289911199876,
+      "learning_rate": 2.8214111729624994e-05,
+      "loss": 0.487,
+      "step": 86
+    },
+    {
+      "epoch": 0.39079169006176306,
+      "grad_norm": 0.5030837162189737,
+      "learning_rate": 2.8206539254816545e-05,
+      "loss": 0.4859,
+      "step": 87
+    },
+    {
+      "epoch": 0.3952835485682201,
+      "grad_norm": 0.6689905622885237,
+      "learning_rate": 2.8198579908767657e-05,
+      "loss": 0.4912,
+      "step": 88
+    },
+    {
+      "epoch": 0.3997754070746771,
+      "grad_norm": 0.591172274153396,
+      "learning_rate": 2.8190233910416995e-05,
+      "loss": 0.4996,
+      "step": 89
+    },
+    {
+      "epoch": 0.4042672655811342,
+      "grad_norm": 0.49390500427611916,
+      "learning_rate": 2.8181501489338944e-05,
+      "loss": 0.4972,
+      "step": 90
+    },
+    {
+      "epoch": 0.40875912408759124,
+      "grad_norm": 0.5414477218971624,
+      "learning_rate": 2.8172382885737236e-05,
+      "loss": 0.4981,
+      "step": 91
+    },
+    {
+      "epoch": 0.4132509825940483,
+      "grad_norm": 0.43235732060188803,
+      "learning_rate": 2.8162878350438392e-05,
+      "loss": 0.4895,
+      "step": 92
+    },
+    {
+      "epoch": 0.41774284110050536,
+      "grad_norm": 0.7027519999992445,
+      "learning_rate": 2.8152988144884793e-05,
+      "loss": 0.4916,
+      "step": 93
+    },
+    {
+      "epoch": 0.42223469960696236,
+      "grad_norm": 0.7103554223289062,
+      "learning_rate": 2.8142712541127497e-05,
+      "loss": 0.4992,
+      "step": 94
+    },
+    {
+      "epoch": 0.4267265581134194,
+      "grad_norm": 0.727736158392093,
+      "learning_rate": 2.813205182181876e-05,
+      "loss": 0.4985,
+      "step": 95
+    },
+    {
+      "epoch": 0.4312184166198765,
+      "grad_norm": 0.9555199352525765,
+      "learning_rate": 2.812100628020426e-05,
+      "loss": 0.4798,
+      "step": 96
+    },
+    {
+      "epoch": 0.43571027512633353,
+      "grad_norm": 0.868910581986562,
+      "learning_rate": 2.8109576220115017e-05,
+      "loss": 0.4981,
+      "step": 97
+    },
+    {
+      "epoch": 0.4402021336327906,
+      "grad_norm": 0.6529644140315375,
+      "learning_rate": 2.8097761955959054e-05,
+      "loss": 0.4919,
+      "step": 98
+    },
+    {
+      "epoch": 0.4446939921392476,
+      "grad_norm": 0.5671052329679708,
+      "learning_rate": 2.8085563812712747e-05,
+      "loss": 0.4712,
+      "step": 99
+    },
+    {
+      "epoch": 0.44918585064570465,
+      "grad_norm": 0.5136497684738213,
+      "learning_rate": 2.8072982125911862e-05,
+      "loss": 0.4822,
+      "step": 100
+    },
+    {
+      "epoch": 0.4536777091521617,
+      "grad_norm": 0.473175867183367,
+      "learning_rate": 2.8060017241642368e-05,
+      "loss": 0.4947,
+      "step": 101
+    },
+    {
+      "epoch": 0.45816956765861877,
+      "grad_norm": 0.6552480677126711,
+      "learning_rate": 2.8046669516530868e-05,
+      "loss": 0.4906,
+      "step": 102
+    },
+    {
+      "epoch": 0.4626614261650758,
+      "grad_norm": 0.659650419201284,
+      "learning_rate": 2.8032939317734826e-05,
+      "loss": 0.4909,
+      "step": 103
+    },
+    {
+      "epoch": 0.46715328467153283,
+      "grad_norm": 0.7383804714557528,
+      "learning_rate": 2.8018827022932455e-05,
+      "loss": 0.4804,
+      "step": 104
+    },
+    {
+      "epoch": 0.4716451431779899,
+      "grad_norm": 0.5224334871193078,
+      "learning_rate": 2.8004333020312324e-05,
+      "loss": 0.476,
+      "step": 105
+    },
+    {
+      "epoch": 0.47613700168444695,
+      "grad_norm": 0.5088529143014228,
+      "learning_rate": 2.798945770856269e-05,
+      "loss": 0.4838,
+      "step": 106
+    },
+    {
+      "epoch": 0.480628860190904,
+      "grad_norm": 0.5964414382468598,
+      "learning_rate": 2.7974201496860518e-05,
+      "loss": 0.4828,
+      "step": 107
+    },
+    {
+      "epoch": 0.485120718697361,
+      "grad_norm": 0.6360776918845367,
+      "learning_rate": 2.7958564804860243e-05,
+      "loss": 0.4931,
+      "step": 108
+    },
+    {
+      "epoch": 0.48961257720381807,
+      "grad_norm": 0.44711939823413394,
+      "learning_rate": 2.7942548062682207e-05,
+      "loss": 0.4931,
+      "step": 109
+    },
+    {
+      "epoch": 0.4941044357102751,
+      "grad_norm": 0.39966821409323977,
+      "learning_rate": 2.7926151710900844e-05,
+      "loss": 0.4877,
+      "step": 110
+    },
+    {
+      "epoch": 0.4985962942167322,
+      "grad_norm": 0.5350994599031005,
+      "learning_rate": 2.7909376200532555e-05,
+      "loss": 0.489,
+      "step": 111
+    },
+    {
+      "epoch": 0.5030881527231892,
+      "grad_norm": 0.6783310634204751,
+      "learning_rate": 2.7892221993023292e-05,
+      "loss": 0.4787,
+      "step": 112
+    },
+    {
+      "epoch": 0.5075800112296462,
+      "grad_norm": 0.6347156840740921,
+      "learning_rate": 2.7874689560235885e-05,
+      "loss": 0.4947,
+      "step": 113
+    },
+    {
+      "epoch": 0.5120718697361033,
+      "grad_norm": 0.6856811377347193,
+      "learning_rate": 2.785677938443705e-05,
+      "loss": 0.4895,
+      "step": 114
+    },
+    {
+      "epoch": 0.5165637282425604,
+      "grad_norm": 0.5537101587480877,
+      "learning_rate": 2.7838491958284118e-05,
+      "loss": 0.4766,
+      "step": 115
+    },
+    {
+      "epoch": 0.5210555867490174,
+      "grad_norm": 0.4282052718202538,
+      "learning_rate": 2.78198277848115e-05,
+      "loss": 0.497,
+      "step": 116
+    },
+    {
+      "epoch": 0.5255474452554745,
+      "grad_norm": 0.3393944725268088,
+      "learning_rate": 2.780078737741683e-05,
+      "loss": 0.4824,
+      "step": 117
+    },
+    {
+      "epoch": 0.5300393037619315,
+      "grad_norm": 0.44536797037412856,
+      "learning_rate": 2.7781371259846857e-05,
+      "loss": 0.4889,
+      "step": 118
+    },
+    {
+      "epoch": 0.5345311622683886,
+      "grad_norm": 0.6227643022770726,
+      "learning_rate": 2.7761579966183044e-05,
+      "loss": 0.4819,
+      "step": 119
+    },
+    {
+      "epoch": 0.5390230207748455,
+      "grad_norm": 0.6149075794124363,
+      "learning_rate": 2.774141404082686e-05,
+      "loss": 0.4823,
+      "step": 120
+    },
+    {
+      "epoch": 0.5435148792813026,
+      "grad_norm": 0.5251207353470196,
+      "learning_rate": 2.7720874038484803e-05,
+      "loss": 0.4896,
+      "step": 121
+    },
+    {
+      "epoch": 0.5480067377877597,
+      "grad_norm": 0.540780427167242,
+      "learning_rate": 2.7699960524153162e-05,
+      "loss": 0.4817,
+      "step": 122
+    },
+    {
+      "epoch": 0.5524985962942167,
+      "grad_norm": 0.5891410499193173,
+      "learning_rate": 2.767867407310247e-05,
+      "loss": 0.4662,
+      "step": 123
+    },
+    {
+      "epoch": 0.5569904548006738,
+      "grad_norm": 0.6370627313155826,
+      "learning_rate": 2.765701527086166e-05,
+      "loss": 0.4934,
+      "step": 124
+    },
+    {
+      "epoch": 0.5614823133071308,
+      "grad_norm": 0.6977215530471076,
+      "learning_rate": 2.7634984713201984e-05,
+      "loss": 0.4945,
+      "step": 125
+    },
+    {
+      "epoch": 0.5659741718135879,
+      "grad_norm": 0.6200258417785485,
+      "learning_rate": 2.7612583006120617e-05,
+      "loss": 0.4722,
+      "step": 126
+    },
+    {
+      "epoch": 0.570466030320045,
+      "grad_norm": 0.5002600730923096,
+      "learning_rate": 2.758981076582397e-05,
+      "loss": 0.4743,
+      "step": 127
+    },
+    {
+      "epoch": 0.574957888826502,
+      "grad_norm": 0.5784959376735728,
+      "learning_rate": 2.7566668618710772e-05,
+      "loss": 0.4785,
+      "step": 128
+    },
+    {
+      "epoch": 0.5794497473329591,
+      "grad_norm": 0.6395956953557737,
+      "learning_rate": 2.754315720135481e-05,
+      "loss": 0.4844,
+      "step": 129
+    },
+    {
+      "epoch": 0.583941605839416,
+      "grad_norm": 0.5350847012934989,
+      "learning_rate": 2.751927716048744e-05,
+      "loss": 0.4692,
+      "step": 130
+    },
+    {
+      "epoch": 0.5884334643458731,
+      "grad_norm": 0.43903297472349573,
+      "learning_rate": 2.749502915297979e-05,
+      "loss": 0.4842,
+      "step": 131
+    },
+    {
+      "epoch": 0.5929253228523301,
+      "grad_norm": 0.45348394184058516,
+      "learning_rate": 2.747041384582468e-05,
+      "loss": 0.4673,
+      "step": 132
+    },
+    {
+      "epoch": 0.5974171813587872,
+      "grad_norm": 0.5537640504259822,
+      "learning_rate": 2.7445431916118297e-05,
+      "loss": 0.4735,
+      "step": 133
+    },
+    {
+      "epoch": 0.6019090398652442,
+      "grad_norm": 0.44994353104607454,
+      "learning_rate": 2.742008405104155e-05,
+      "loss": 0.4797,
+      "step": 134
+    },
+    {
+      "epoch": 0.6064008983717013,
+      "grad_norm": 0.4544214314383658,
+      "learning_rate": 2.7394370947841175e-05,
+      "loss": 0.4796,
+      "step": 135
+    },
+    {
+      "epoch": 0.6108927568781584,
+      "grad_norm": 0.5054093825095585,
+      "learning_rate": 2.7368293313810566e-05,
+      "loss": 0.483,
+      "step": 136
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7187663549969284,
+      "learning_rate": 2.7341851866270298e-05,
+      "loss": 0.4773,
+      "step": 137
+    },
+    {
+      "epoch": 0.6198764738910725,
+      "grad_norm": 0.699515058479278,
+      "learning_rate": 2.7315047332548415e-05,
+      "loss": 0.4736,
+      "step": 138
+    },
+    {
+      "epoch": 0.6243683323975294,
+      "grad_norm": 0.5165553858169959,
+      "learning_rate": 2.728788044996041e-05,
+      "loss": 0.479,
+      "step": 139
+    },
+    {
+      "epoch": 0.6288601909039865,
+      "grad_norm": 0.4327740748377208,
+      "learning_rate": 2.7260351965788962e-05,
+      "loss": 0.4715,
+      "step": 140
+    },
+    {
+      "epoch": 0.6333520494104435,
+      "grad_norm": 0.5038382774885327,
+      "learning_rate": 2.7232462637263354e-05,
+      "loss": 0.4821,
+      "step": 141
+    },
+    {
+      "epoch": 0.6378439079169006,
+      "grad_norm": 0.5193208261847068,
+      "learning_rate": 2.720421323153866e-05,
+      "loss": 0.4721,
+      "step": 142
+    },
+    {
+      "epoch": 0.6423357664233577,
+      "grad_norm": 0.5055322576434895,
+      "learning_rate": 2.7175604525674646e-05,
+      "loss": 0.4656,
+      "step": 143
+    },
+    {
+      "epoch": 0.6468276249298147,
+      "grad_norm": 0.46251349642636685,
+      "learning_rate": 2.7146637306614383e-05,
+      "loss": 0.4676,
+      "step": 144
+    },
+    {
+      "epoch": 0.6513194834362718,
+      "grad_norm": 0.5012429738213727,
+      "learning_rate": 2.711731237116262e-05,
+      "loss": 0.4831,
+      "step": 145
+    },
+    {
+      "epoch": 0.6558113419427288,
+      "grad_norm": 0.5946163957927814,
+      "learning_rate": 2.708763052596383e-05,
+      "loss": 0.4761,
+      "step": 146
+    },
+    {
+      "epoch": 0.6603032004491859,
+      "grad_norm": 0.6959852968009713,
+      "learning_rate": 2.7057592587480068e-05,
+      "loss": 0.4749,
+      "step": 147
+    },
+    {
+      "epoch": 0.664795058955643,
+      "grad_norm": 0.6454980147287459,
+      "learning_rate": 2.702719938196848e-05,
+      "loss": 0.4743,
+      "step": 148
+    },
+    {
+      "epoch": 0.6692869174620999,
+      "grad_norm": 0.48108943458946607,
+      "learning_rate": 2.6996451745458583e-05,
+      "loss": 0.4773,
+      "step": 149
+    },
+    {
+      "epoch": 0.673778775968557,
+      "grad_norm": 0.3774779921512963,
+      "learning_rate": 2.696535052372927e-05,
+      "loss": 0.47,
+      "step": 150
+    },
+    {
+      "epoch": 0.678270634475014,
+      "grad_norm": 0.4392920410610574,
+      "learning_rate": 2.6933896572285545e-05,
+      "loss": 0.4848,
+      "step": 151
+    },
+    {
+      "epoch": 0.6827624929814711,
+      "grad_norm": 0.5769958845705692,
+      "learning_rate": 2.6902090756334994e-05,
+      "loss": 0.4923,
+      "step": 152
+    },
+    {
+      "epoch": 0.6872543514879281,
+      "grad_norm": 0.5757560343292677,
+      "learning_rate": 2.686993395076398e-05,
+      "loss": 0.4791,
+      "step": 153
+    },
+    {
+      "epoch": 0.6917462099943852,
+      "grad_norm": 0.5671105946152095,
+      "learning_rate": 2.683742704011357e-05,
+      "loss": 0.4729,
+      "step": 154
+    },
+    {
+      "epoch": 0.6962380685008422,
+      "grad_norm": 0.5782184489398106,
+      "learning_rate": 2.6804570918555232e-05,
+      "loss": 0.4779,
+      "step": 155
+    },
+    {
+      "epoch": 0.7007299270072993,
+      "grad_norm": 0.6301887119444084,
+      "learning_rate": 2.67713664898662e-05,
+      "loss": 0.4757,
+      "step": 156
+    },
+    {
+      "epoch": 0.7052217855137564,
+      "grad_norm": 0.6399273153805983,
+      "learning_rate": 2.673781466740465e-05,
+      "loss": 0.4683,
+      "step": 157
+    },
+    {
+      "epoch": 0.7097136440202133,
+      "grad_norm": 0.5751139777508563,
+      "learning_rate": 2.6703916374084546e-05,
+      "loss": 0.474,
+      "step": 158
+    },
+    {
+      "epoch": 0.7142055025266704,
+      "grad_norm": 0.5237392026559451,
+      "learning_rate": 2.6669672542350276e-05,
+      "loss": 0.4868,
+      "step": 159
+    },
+    {
+      "epoch": 0.7186973610331274,
+      "grad_norm": 0.546643971635381,
+      "learning_rate": 2.6635084114150996e-05,
+      "loss": 0.4804,
+      "step": 160
+    },
+    {
+      "epoch": 0.7231892195395845,
+      "grad_norm": 0.5550996062832795,
+      "learning_rate": 2.6600152040914712e-05,
+      "loss": 0.4697,
+      "step": 161
+    },
+    {
+      "epoch": 0.7276810780460415,
+      "grad_norm": 0.5376518959056129,
+      "learning_rate": 2.6564877283522115e-05,
+      "loss": 0.4722,
+      "step": 162
+    },
+    {
+      "epoch": 0.7321729365524986,
+      "grad_norm": 0.4777569065930761,
+      "learning_rate": 2.6529260812280152e-05,
+      "loss": 0.4555,
+      "step": 163
+    },
+    {
+      "epoch": 0.7366647950589557,
+      "grad_norm": 0.46167672523022346,
+      "learning_rate": 2.6493303606895333e-05,
+      "loss": 0.4606,
+      "step": 164
+    },
+    {
+      "epoch": 0.7411566535654127,
+      "grad_norm": 0.41976167794536035,
+      "learning_rate": 2.6457006656446777e-05,
+      "loss": 0.4771,
+      "step": 165
+    },
+    {
+      "epoch": 0.7456485120718698,
+      "grad_norm": 0.6674527791762378,
+      "learning_rate": 2.642037095935903e-05,
+      "loss": 0.4772,
+      "step": 166
+    },
+    {
+      "epoch": 0.7501403705783268,
+      "grad_norm": 0.6290119968833247,
+      "learning_rate": 2.638339752337456e-05,
+      "loss": 0.482,
+      "step": 167
+    },
+    {
+      "epoch": 0.7546322290847838,
+      "grad_norm": 0.40285509227429195,
+      "learning_rate": 2.634608736552607e-05,
+      "loss": 0.4653,
+      "step": 168
+    },
+    {
+      "epoch": 0.7591240875912408,
+      "grad_norm": 0.4164094017051228,
+      "learning_rate": 2.630844151210851e-05,
+      "loss": 0.4709,
+      "step": 169
+    },
+    {
+      "epoch": 0.7636159460976979,
+      "grad_norm": 0.4467679602252483,
+      "learning_rate": 2.6270460998650838e-05,
+      "loss": 0.4698,
+      "step": 170
+    },
+    {
+      "epoch": 0.768107804604155,
+      "grad_norm": 0.5462945507817383,
+      "learning_rate": 2.6232146869887565e-05,
+      "loss": 0.4778,
+      "step": 171
+    },
+    {
+      "epoch": 0.772599663110612,
+      "grad_norm": 0.3685789574173616,
+      "learning_rate": 2.6193500179729982e-05,
+      "loss": 0.4681,
+      "step": 172
+    },
+    {
+      "epoch": 0.7770915216170691,
+      "grad_norm": 0.4490725514023042,
+      "learning_rate": 2.6154521991237195e-05,
+      "loss": 0.4586,
+      "step": 173
+    },
+    {
+      "epoch": 0.7815833801235261,
+      "grad_norm": 0.46279499032609167,
+      "learning_rate": 2.6115213376586873e-05,
+      "loss": 0.4731,
+      "step": 174
+    },
+    {
+      "epoch": 0.7860752386299832,
+      "grad_norm": 0.4997523270309222,
+      "learning_rate": 2.6075575417045745e-05,
+      "loss": 0.4769,
+      "step": 175
+    },
+    {
+      "epoch": 0.7905670971364402,
+      "grad_norm": 0.5505245764097036,
+      "learning_rate": 2.603560920293988e-05,
+      "loss": 0.4758,
+      "step": 176
+    },
+    {
+      "epoch": 0.7950589556428973,
+      "grad_norm": 0.363463467720154,
+      "learning_rate": 2.5995315833624693e-05,
+      "loss": 0.4561,
+      "step": 177
+    },
+    {
+      "epoch": 0.7995508141493542,
+      "grad_norm": 0.3917739564310974,
+      "learning_rate": 2.595469641745467e-05,
+      "loss": 0.4641,
+      "step": 178
+    },
+    {
+      "epoch": 0.8040426726558113,
+      "grad_norm": 0.45375939875892646,
+      "learning_rate": 2.5913752071752935e-05,
+      "loss": 0.4679,
+      "step": 179
+    },
+    {
+      "epoch": 0.8085345311622684,
+      "grad_norm": 0.4072903389108503,
+      "learning_rate": 2.5872483922780468e-05,
+      "loss": 0.4736,
+      "step": 180
+    },
+    {
+      "epoch": 0.8130263896687254,
+      "grad_norm": 0.5207413079308953,
+      "learning_rate": 2.583089310570515e-05,
+      "loss": 0.4823,
+      "step": 181
+    },
+    {
+      "epoch": 0.8175182481751825,
+      "grad_norm": 0.47973720569304484,
+      "learning_rate": 2.578898076457053e-05,
+      "loss": 0.4649,
+      "step": 182
+    },
+    {
+      "epoch": 0.8220101066816395,
+      "grad_norm": 0.3987214245331553,
+      "learning_rate": 2.5746748052264372e-05,
+      "loss": 0.4697,
+      "step": 183
+    },
+    {
+      "epoch": 0.8265019651880966,
+      "grad_norm": 0.40108553854919987,
+      "learning_rate": 2.570419613048691e-05,
+      "loss": 0.4622,
+      "step": 184
+    },
+    {
+      "epoch": 0.8309938236945537,
+      "grad_norm": 0.521581370496541,
+      "learning_rate": 2.566132616971891e-05,
+      "loss": 0.4619,
+      "step": 185
+    },
+    {
+      "epoch": 0.8354856822010107,
+      "grad_norm": 0.5047605572020768,
+      "learning_rate": 2.5618139349189486e-05,
+      "loss": 0.4671,
+      "step": 186
+    },
+    {
+      "epoch": 0.8399775407074677,
+      "grad_norm": 0.5364692556005867,
+      "learning_rate": 2.5574636856843648e-05,
+      "loss": 0.4689,
+      "step": 187
+    },
+    {
+      "epoch": 0.8444693992139247,
+      "grad_norm": 0.4492847766801117,
+      "learning_rate": 2.5530819889309624e-05,
+      "loss": 0.4599,
+      "step": 188
+    },
+    {
+      "epoch": 0.8489612577203818,
+      "grad_norm": 0.38696429925917464,
+      "learning_rate": 2.548668965186596e-05,
+      "loss": 0.4606,
+      "step": 189
+    },
+    {
+      "epoch": 0.8534531162268388,
+      "grad_norm": 0.4268656959845891,
+      "learning_rate": 2.5442247358408328e-05,
+      "loss": 0.4755,
+      "step": 190
+    },
+    {
+      "epoch": 0.8579449747332959,
+      "grad_norm": 0.5247596649132644,
+      "learning_rate": 2.539749423141619e-05,
+      "loss": 0.477,
+      "step": 191
+    },
+    {
+      "epoch": 0.862436833239753,
+      "grad_norm": 0.4921984452162432,
+      "learning_rate": 2.5352431501919134e-05,
+      "loss": 0.4731,
+      "step": 192
+    },
+    {
+      "epoch": 0.86692869174621,
+      "grad_norm": 0.48240116746472755,
+      "learning_rate": 2.5307060409463016e-05,
+      "loss": 0.4689,
+      "step": 193
+    },
+    {
+      "epoch": 0.8714205502526671,
+      "grad_norm": 0.4931496661169666,
+      "learning_rate": 2.526138220207587e-05,
+      "loss": 0.4803,
+      "step": 194
+    },
+    {
+      "epoch": 0.8759124087591241,
+      "grad_norm": 0.582045224133066,
+      "learning_rate": 2.5215398136233585e-05,
+      "loss": 0.4557,
+      "step": 195
+    },
+    {
+      "epoch": 0.8804042672655812,
+      "grad_norm": 0.5411459162367669,
+      "learning_rate": 2.5169109476825326e-05,
+      "loss": 0.4589,
+      "step": 196
+    },
+    {
+      "epoch": 0.8848961257720381,
+      "grad_norm": 0.39863691240843,
+      "learning_rate": 2.512251749711875e-05,
+      "loss": 0.4668,
+      "step": 197
+    },
+    {
+      "epoch": 0.8893879842784952,
+      "grad_norm": 0.4593569962235714,
+      "learning_rate": 2.5075623478724984e-05,
+      "loss": 0.4753,
+      "step": 198
+    },
+    {
+      "epoch": 0.8938798427849522,
+      "grad_norm": 0.4287957363960024,
+      "learning_rate": 2.502842871156337e-05,
+      "loss": 0.465,
+      "step": 199
+    },
+    {
+      "epoch": 0.8983717012914093,
+      "grad_norm": 0.4228370536587653,
+      "learning_rate": 2.498093449382598e-05,
+      "loss": 0.4519,
+      "step": 200
+    },
+    {
+      "epoch": 0.9028635597978664,
+      "grad_norm": 0.375019534593526,
+      "learning_rate": 2.4933142131941917e-05,
+      "loss": 0.4645,
+      "step": 201
+    },
+    {
+      "epoch": 0.9073554183043234,
+      "grad_norm": 0.43124335474679265,
+      "learning_rate": 2.488505294054135e-05,
+      "loss": 0.4804,
+      "step": 202
+    },
+    {
+      "epoch": 0.9118472768107805,
+      "grad_norm": 0.45579971562245586,
+      "learning_rate": 2.4836668242419398e-05,
+      "loss": 0.4589,
+      "step": 203
+    },
+    {
+      "epoch": 0.9163391353172375,
+      "grad_norm": 0.36653038951983824,
+      "learning_rate": 2.47879893684997e-05,
+      "loss": 0.4593,
+      "step": 204
+    },
+    {
+      "epoch": 0.9208309938236946,
+      "grad_norm": 0.38913683153964257,
+      "learning_rate": 2.4739017657797826e-05,
+      "loss": 0.469,
+      "step": 205
+    },
+    {
+      "epoch": 0.9253228523301515,
+      "grad_norm": 0.3716699947181807,
+      "learning_rate": 2.4689754457384455e-05,
+      "loss": 0.474,
+      "step": 206
+    },
+    {
+      "epoch": 0.9298147108366086,
+      "grad_norm": 0.4537303687953904,
+      "learning_rate": 2.4640201122348293e-05,
+      "loss": 0.4624,
+      "step": 207
+    },
+    {
+      "epoch": 0.9343065693430657,
+      "grad_norm": 0.3637412162756154,
+      "learning_rate": 2.4590359015758823e-05,
+      "loss": 0.4602,
+      "step": 208
+    },
+    {
+      "epoch": 0.9387984278495227,
+      "grad_norm": 0.45311583807808,
+      "learning_rate": 2.45402295086288e-05,
+      "loss": 0.4699,
+      "step": 209
+    },
+    {
+      "epoch": 0.9432902863559798,
+      "grad_norm": 0.4006232500664817,
+      "learning_rate": 2.4489813979876532e-05,
+      "loss": 0.4734,
+      "step": 210
+    },
+    {
+      "epoch": 0.9477821448624368,
+      "grad_norm": 0.3550939989105772,
+      "learning_rate": 2.443911381628797e-05,
+      "loss": 0.4635,
+      "step": 211
+    },
+    {
+      "epoch": 0.9522740033688939,
+      "grad_norm": 0.3358080281360604,
+      "learning_rate": 2.4388130412478547e-05,
+      "loss": 0.4572,
+      "step": 212
+    },
+    {
+      "epoch": 0.956765861875351,
+      "grad_norm": 0.4055886952871211,
+      "learning_rate": 2.4336865170854804e-05,
+      "loss": 0.4569,
+      "step": 213
+    },
+    {
+      "epoch": 0.961257720381808,
+      "grad_norm": 0.4576518855442424,
+      "learning_rate": 2.4285319501575847e-05,
+      "loss": 0.4671,
+      "step": 214
+    },
+    {
+      "epoch": 0.9657495788882651,
+      "grad_norm": 0.40899913469666216,
+      "learning_rate": 2.4233494822514523e-05,
+      "loss": 0.4458,
+      "step": 215
+    },
+    {
+      "epoch": 0.970241437394722,
+      "grad_norm": 0.28945530496891764,
+      "learning_rate": 2.4181392559218453e-05,
+      "loss": 0.4618,
+      "step": 216
+    },
+    {
+      "epoch": 0.9747332959011791,
+      "grad_norm": 0.35107378537689937,
+      "learning_rate": 2.4129014144870783e-05,
+      "loss": 0.4574,
+      "step": 217
+    },
+    {
+      "epoch": 0.9792251544076361,
+      "grad_norm": 0.4020814076094774,
+      "learning_rate": 2.4076361020250788e-05,
+      "loss": 0.4646,
+      "step": 218
+    },
+    {
+      "epoch": 0.9837170129140932,
+      "grad_norm": 0.4291582741080788,
+      "learning_rate": 2.4023434633694226e-05,
+      "loss": 0.4637,
+      "step": 219
+    },
+    {
+      "epoch": 0.9882088714205502,
+      "grad_norm": 0.40682515531336805,
+      "learning_rate": 2.3970236441053506e-05,
+      "loss": 0.4772,
+      "step": 220
+    },
+    {
+      "epoch": 0.9927007299270073,
+      "grad_norm": 0.3749932255250272,
+      "learning_rate": 2.3916767905657643e-05,
+      "loss": 0.465,
+      "step": 221
+    },
+    {
+      "epoch": 0.9971925884334644,
+      "grad_norm": 0.4077210425304697,
+      "learning_rate": 2.3863030498271994e-05,
+      "loss": 0.4628,
+      "step": 222
+    },
+    {
+      "epoch": 1.0030881527231892,
+      "grad_norm": 0.6159591297183691,
+      "learning_rate": 2.3809025697057822e-05,
+      "loss": 0.7494,
+      "step": 223
+    },
+    {
+      "epoch": 1.0075800112296462,
+      "grad_norm": 0.743081485050473,
+      "learning_rate": 2.3754754987531616e-05,
+      "loss": 0.4526,
+      "step": 224
+    },
+    {
+      "epoch": 1.0120718697361033,
+      "grad_norm": 0.9936672942774801,
+      "learning_rate": 2.3700219862524227e-05,
+      "loss": 0.4246,
+      "step": 225
+    },
+    {
+      "epoch": 1.0165637282425604,
+      "grad_norm": 0.798262818791095,
+      "learning_rate": 2.364542182213983e-05,
+      "loss": 0.4375,
+      "step": 226
+    },
+    {
+      "epoch": 1.0210555867490174,
+      "grad_norm": 0.6667703200360748,
+      "learning_rate": 2.3590362373714646e-05,
+      "loss": 0.4336,
+      "step": 227
+    },
+    {
+      "epoch": 1.0255474452554745,
+      "grad_norm": 0.8477945932709707,
+      "learning_rate": 2.3535043031775457e-05,
+      "loss": 0.4555,
+      "step": 228
+    },
+    {
+      "epoch": 1.0300393037619315,
+      "grad_norm": 0.7922389974719279,
+      "learning_rate": 2.3479465317997993e-05,
+      "loss": 0.4365,
+      "step": 229
+    },
+    {
+      "epoch": 1.0345311622683886,
+      "grad_norm": 0.7126148195703437,
+      "learning_rate": 2.342363076116503e-05,
+      "loss": 0.4523,
+      "step": 230
+    },
+    {
+      "epoch": 1.0390230207748457,
+      "grad_norm": 0.8802731555951069,
+      "learning_rate": 2.336754089712436e-05,
+      "loss": 0.4491,
+      "step": 231
+    },
+    {
+      "epoch": 1.0435148792813027,
+      "grad_norm": 0.9077015045585495,
+      "learning_rate": 2.331119726874655e-05,
+      "loss": 0.4564,
+      "step": 232
+    },
+    {
+      "epoch": 1.0480067377877598,
+      "grad_norm": 0.8853843133720011,
+      "learning_rate": 2.3254601425882486e-05,
+      "loss": 0.442,
+      "step": 233
+    },
+    {
+      "epoch": 1.0524985962942168,
+      "grad_norm": 0.7675340514208551,
+      "learning_rate": 2.3197754925320744e-05,
+      "loss": 0.4386,
+      "step": 234
+    },
+    {
+      "epoch": 1.0569904548006739,
+      "grad_norm": 0.795855384998558,
+      "learning_rate": 2.314065933074478e-05,
+      "loss": 0.4483,
+      "step": 235
+    },
+    {
+      "epoch": 1.0614823133071307,
+      "grad_norm": 0.8688332898542778,
+      "learning_rate": 2.30833162126899e-05,
+      "loss": 0.4401,
+      "step": 236
+    },
+    {
+      "epoch": 1.0659741718135878,
+      "grad_norm": 0.6912025881093444,
+      "learning_rate": 2.3025727148500078e-05,
+      "loss": 0.4258,
+      "step": 237
+    },
+    {
+      "epoch": 1.0704660303200448,
+      "grad_norm": 0.6210071932563501,
+      "learning_rate": 2.296789372228454e-05,
+      "loss": 0.4446,
+      "step": 238
+    },
+    {
+      "epoch": 1.074957888826502,
+      "grad_norm": 0.7612886989218541,
+      "learning_rate": 2.290981752487423e-05,
+      "loss": 0.4371,
+      "step": 239
+    },
+    {
+      "epoch": 1.079449747332959,
+      "grad_norm": 0.49304669277389745,
+      "learning_rate": 2.285150015377801e-05,
+      "loss": 0.4313,
+      "step": 240
+    },
+    {
+      "epoch": 1.083941605839416,
+      "grad_norm": 0.6740461248297163,
+      "learning_rate": 2.279294321313875e-05,
+      "loss": 0.4347,
+      "step": 241
+    },
+    {
+      "epoch": 1.088433464345873,
+      "grad_norm": 0.47110764302293234,
+      "learning_rate": 2.2734148313689173e-05,
+      "loss": 0.4472,
+      "step": 242
+    },
+    {
+      "epoch": 1.0929253228523301,
+      "grad_norm": 0.6319912881250679,
+      "learning_rate": 2.2675117072707572e-05,
+      "loss": 0.438,
+      "step": 243
+    },
+    {
+      "epoch": 1.0974171813587872,
+      "grad_norm": 0.48989174750287556,
+      "learning_rate": 2.2615851113973312e-05,
+      "loss": 0.4266,
+      "step": 244
+    },
+    {
+      "epoch": 1.1019090398652442,
+      "grad_norm": 0.4886585461236165,
+      "learning_rate": 2.255635206772217e-05,
+      "loss": 0.4359,
+      "step": 245
+    },
+    {
+      "epoch": 1.1064008983717013,
+      "grad_norm": 0.5509821980510767,
+      "learning_rate": 2.2496621570601487e-05,
+      "loss": 0.4598,
+      "step": 246
+    },
+    {
+      "epoch": 1.1108927568781584,
+      "grad_norm": 0.366525891107272,
+      "learning_rate": 2.243666126562515e-05,
+      "loss": 0.4199,
+      "step": 247
+    },
+    {
+      "epoch": 1.1153846153846154,
+      "grad_norm": 0.5181695608354417,
+      "learning_rate": 2.2376472802128402e-05,
+      "loss": 0.4375,
+      "step": 248
+    },
+    {
+      "epoch": 1.1198764738910725,
+      "grad_norm": 0.38286389285013006,
+      "learning_rate": 2.231605783572246e-05,
+      "loss": 0.4417,
+      "step": 249
+    },
+    {
+      "epoch": 1.1243683323975295,
+      "grad_norm": 0.5620717664471987,
+      "learning_rate": 2.2255418028248992e-05,
+      "loss": 0.4593,
+      "step": 250
+    },
+    {
+      "epoch": 1.1288601909039866,
+      "grad_norm": 0.4462443722479522,
+      "learning_rate": 2.21945550477344e-05,
+      "loss": 0.4374,
+      "step": 251
+    },
+    {
+      "epoch": 1.1333520494104437,
+      "grad_norm": 0.4272534160396403,
+      "learning_rate": 2.213347056834392e-05,
+      "loss": 0.4131,
+      "step": 252
+    },
+    {
+      "epoch": 1.1378439079169007,
+      "grad_norm": 0.5008773829340849,
+      "learning_rate": 2.2072166270335592e-05,
+      "loss": 0.4586,
+      "step": 253
+    },
+    {
+      "epoch": 1.1423357664233578,
+      "grad_norm": 0.3792872396771057,
+      "learning_rate": 2.2010643840014032e-05,
+      "loss": 0.4219,
+      "step": 254
+    },
+    {
+      "epoch": 1.1468276249298146,
+      "grad_norm": 0.44576445227452827,
+      "learning_rate": 2.1948904969684047e-05,
+      "loss": 0.4512,
+      "step": 255
+    },
+    {
+      "epoch": 1.1513194834362717,
+      "grad_norm": 0.3963785617647081,
+      "learning_rate": 2.188695135760409e-05,
+      "loss": 0.4187,
+      "step": 256
+    },
+    {
+      "epoch": 1.1558113419427287,
+      "grad_norm": 0.3745162442173085,
+      "learning_rate": 2.1824784707939534e-05,
+      "loss": 0.4527,
+      "step": 257
+    },
+    {
+      "epoch": 1.1603032004491858,
+      "grad_norm": 0.4302552912637187,
+      "learning_rate": 2.17624067307158e-05,
+      "loss": 0.4338,
+      "step": 258
+    },
+    {
+      "epoch": 1.1647950589556428,
+      "grad_norm": 0.35964883159886635,
+      "learning_rate": 2.1699819141771334e-05,
+      "loss": 0.4503,
+      "step": 259
+    },
+    {
+      "epoch": 1.1692869174621,
+      "grad_norm": 0.46269656933781855,
+      "learning_rate": 2.1637023662710378e-05,
+      "loss": 0.4226,
+      "step": 260
+    },
+    {
+      "epoch": 1.173778775968557,
+      "grad_norm": 0.3412078867716096,
+      "learning_rate": 2.1574022020855653e-05,
+      "loss": 0.4147,
+      "step": 261
+    },
+    {
+      "epoch": 1.178270634475014,
+      "grad_norm": 0.4572735310680308,
+      "learning_rate": 2.1510815949200806e-05,
+      "loss": 0.4611,
+      "step": 262
+    },
+    {
+      "epoch": 1.182762492981471,
+      "grad_norm": 0.37564713470069194,
+      "learning_rate": 2.1447407186362773e-05,
+      "loss": 0.4317,
+      "step": 263
+    },
+    {
+      "epoch": 1.1872543514879281,
+      "grad_norm": 0.3853481594309711,
+      "learning_rate": 2.138379747653393e-05,
+      "loss": 0.4413,
+      "step": 264
+    },
+    {
+      "epoch": 1.1917462099943852,
+      "grad_norm": 0.3203618610490023,
+      "learning_rate": 2.131998856943413e-05,
+      "loss": 0.4273,
+      "step": 265
+    },
+    {
+      "epoch": 1.1962380685008422,
+      "grad_norm": 0.4418711444396078,
+      "learning_rate": 2.125598222026257e-05,
+      "loss": 0.4434,
+      "step": 266
+    },
+    {
+      "epoch": 1.2007299270072993,
+      "grad_norm": 0.4573587827702061,
+      "learning_rate": 2.1191780189649513e-05,
+      "loss": 0.4367,
+      "step": 267
+    },
+    {
+      "epoch": 1.2052217855137564,
+      "grad_norm": 0.38776707741248617,
+      "learning_rate": 2.1127384243607844e-05,
+      "loss": 0.4357,
+      "step": 268
+    },
+    {
+      "epoch": 1.2097136440202134,
+      "grad_norm": 0.3259763037218784,
+      "learning_rate": 2.106279615348451e-05,
+      "loss": 0.4446,
+      "step": 269
+    },
+    {
+      "epoch": 1.2142055025266705,
+      "grad_norm": 0.34557251617943774,
+      "learning_rate": 2.099801769591179e-05,
+      "loss": 0.4204,
+      "step": 270
+    },
+    {
+      "epoch": 1.2186973610331275,
+      "grad_norm": 0.4181566384819826,
+      "learning_rate": 2.0933050652758417e-05,
+      "loss": 0.4551,
+      "step": 271
+    },
+    {
+      "epoch": 1.2231892195395846,
+      "grad_norm": 0.2875529069270976,
+      "learning_rate": 2.0867896811080576e-05,
+      "loss": 0.4155,
+      "step": 272
+    },
+    {
+      "epoch": 1.2276810780460417,
+      "grad_norm": 0.32586319219897486,
+      "learning_rate": 2.0802557963072736e-05,
+      "loss": 0.4315,
+      "step": 273
+    },
+    {
+      "epoch": 1.2321729365524985,
+      "grad_norm": 0.3014143337486855,
+      "learning_rate": 2.073703590601836e-05,
+      "loss": 0.445,
+      "step": 274
+    },
+    {
+      "epoch": 1.2366647950589555,
+      "grad_norm": 0.3366952207644692,
+      "learning_rate": 2.067133244224047e-05,
+      "loss": 0.4522,
+      "step": 275
+    },
+    {
+      "epoch": 1.2411566535654126,
+      "grad_norm": 0.2981657265784339,
+      "learning_rate": 2.0605449379052047e-05,
+      "loss": 0.4128,
+      "step": 276
+    },
+    {
+      "epoch": 1.2456485120718697,
+      "grad_norm": 0.2741164199668628,
+      "learning_rate": 2.0539388528706354e-05,
+      "loss": 0.4465,
+      "step": 277
+    },
+    {
+      "epoch": 1.2501403705783267,
+      "grad_norm": 0.28116445398834183,
+      "learning_rate": 2.0473151708347052e-05,
+      "loss": 0.4214,
+      "step": 278
+    },
+    {
+      "epoch": 1.2546322290847838,
+      "grad_norm": 0.26743411373744025,
+      "learning_rate": 2.040674073995824e-05,
+      "loss": 0.4378,
+      "step": 279
+    },
+    {
+      "epoch": 1.2591240875912408,
+      "grad_norm": 0.349163241377986,
+      "learning_rate": 2.0340157450314317e-05,
+      "loss": 0.4328,
+      "step": 280
+    },
+    {
+      "epoch": 1.263615946097698,
+      "grad_norm": 0.3478461930112901,
+      "learning_rate": 2.0273403670929756e-05,
+      "loss": 0.4373,
+      "step": 281
+    },
+    {
+      "epoch": 1.268107804604155,
+      "grad_norm": 0.35369927623819397,
+      "learning_rate": 2.020648123800871e-05,
+      "loss": 0.4356,
+      "step": 282
+    },
+    {
+      "epoch": 1.272599663110612,
+      "grad_norm": 0.4318259227052038,
+      "learning_rate": 2.0139391992394488e-05,
+      "loss": 0.4337,
+      "step": 283
+    },
+    {
+      "epoch": 1.277091521617069,
+      "grad_norm": 0.3341822520627836,
+      "learning_rate": 2.0072137779518955e-05,
+      "loss": 0.4364,
+      "step": 284
+    },
+    {
+      "epoch": 1.2815833801235261,
+      "grad_norm": 0.32765945183019074,
+      "learning_rate": 2.000472044935174e-05,
+      "loss": 0.4376,
+      "step": 285
+    },
+    {
+      "epoch": 1.2860752386299832,
+      "grad_norm": 0.29486663216008824,
+      "learning_rate": 1.9937141856349362e-05,
+      "loss": 0.4362,
+      "step": 286
+    },
+    {
+      "epoch": 1.2905670971364402,
+      "grad_norm": 0.3708118576046198,
+      "learning_rate": 1.9869403859404217e-05,
+      "loss": 0.4505,
+      "step": 287
+    },
+    {
+      "epoch": 1.2950589556428973,
+      "grad_norm": 0.2647734977958948,
+      "learning_rate": 1.9801508321793443e-05,
+      "loss": 0.4351,
+      "step": 288
+    },
+    {
+      "epoch": 1.2995508141493544,
+      "grad_norm": 0.29786540958009505,
+      "learning_rate": 1.973345711112766e-05,
+      "loss": 0.4148,
+      "step": 289
+    },
+    {
+      "epoch": 1.3040426726558114,
+      "grad_norm": 0.34111071679248295,
+      "learning_rate": 1.9665252099299626e-05,
+      "loss": 0.4369,
+      "step": 290
+    },
+    {
+      "epoch": 1.3085345311622683,
+      "grad_norm": 0.2825874052298496,
+      "learning_rate": 1.95968951624327e-05,
+      "loss": 0.4375,
+      "step": 291
+    },
+    {
+      "epoch": 1.3130263896687255,
+      "grad_norm": 0.35110084669077885,
+      "learning_rate": 1.9528388180829284e-05,
+      "loss": 0.4296,
+      "step": 292
+    },
+    {
+      "epoch": 1.3175182481751824,
+      "grad_norm": 0.26124827557042796,
+      "learning_rate": 1.9459733038919063e-05,
+      "loss": 0.4404,
+      "step": 293
+    },
+    {
+      "epoch": 1.3220101066816397,
+      "grad_norm": 0.3319179737449958,
+      "learning_rate": 1.939093162520719e-05,
+      "loss": 0.4342,
+      "step": 294
+    },
+    {
+      "epoch": 1.3265019651880965,
+      "grad_norm": 0.3292095376226441,
+      "learning_rate": 1.9321985832222337e-05,
+      "loss": 0.4433,
+      "step": 295
+    },
+    {
+      "epoch": 1.3309938236945535,
+      "grad_norm": 0.24475291813862662,
+      "learning_rate": 1.9252897556464622e-05,
+      "loss": 0.4368,
+      "step": 296
+    },
+    {
+      "epoch": 1.3354856822010106,
+      "grad_norm": 0.3056744142638689,
+      "learning_rate": 1.9183668698353474e-05,
+      "loss": 0.4284,
+      "step": 297
+    },
+    {
+      "epoch": 1.3399775407074677,
+      "grad_norm": 0.28821906996690305,
+      "learning_rate": 1.9114301162175324e-05,
+      "loss": 0.4297,
+      "step": 298
+    },
+    {
+      "epoch": 1.3444693992139247,
+      "grad_norm": 0.26046990859779695,
+      "learning_rate": 1.9044796856031236e-05,
+      "loss": 0.4289,
+      "step": 299
+    },
+    {
+      "epoch": 1.3489612577203818,
+      "grad_norm": 0.3281548899006919,
+      "learning_rate": 1.8975157691784422e-05,
+      "loss": 0.446,
+      "step": 300
+    },
+    {
+      "epoch": 1.3534531162268388,
+      "grad_norm": 0.3101251116335806,
+      "learning_rate": 1.890538558500766e-05,
+      "loss": 0.4381,
+      "step": 301
+    },
+    {
+      "epoch": 1.357944974733296,
+      "grad_norm": 0.27697116298210783,
+      "learning_rate": 1.8835482454930595e-05,
+      "loss": 0.4303,
+      "step": 302
+    },
+    {
+      "epoch": 1.362436833239753,
+      "grad_norm": 0.26183856441501985,
+      "learning_rate": 1.8765450224386924e-05,
+      "loss": 0.4271,
+      "step": 303
+    },
+    {
+      "epoch": 1.36692869174621,
+      "grad_norm": 0.24441781947836166,
+      "learning_rate": 1.8695290819761552e-05,
+      "loss": 0.426,
+      "step": 304
+    },
+    {
+      "epoch": 1.371420550252667,
+      "grad_norm": 0.2637655350752813,
+      "learning_rate": 1.8625006170937556e-05,
+      "loss": 0.4255,
+      "step": 305
+    },
+    {
+      "epoch": 1.3759124087591241,
+      "grad_norm": 0.3417693190160526,
+      "learning_rate": 1.855459821124314e-05,
+      "loss": 0.4304,
+      "step": 306
+    },
+    {
+      "epoch": 1.3804042672655812,
+      "grad_norm": 0.24824305610064504,
+      "learning_rate": 1.8484068877398422e-05,
+      "loss": 0.4149,
+      "step": 307
+    },
+    {
+      "epoch": 1.3848961257720382,
+      "grad_norm": 0.2983957956329138,
+      "learning_rate": 1.8413420109462173e-05,
+      "loss": 0.4459,
+      "step": 308
+    },
+    {
+      "epoch": 1.3893879842784953,
+      "grad_norm": 0.29141281712707245,
+      "learning_rate": 1.8342653850778457e-05,
+      "loss": 0.4261,
+      "step": 309
+    },
+    {
+      "epoch": 1.3938798427849521,
+      "grad_norm": 0.31643097989145,
+      "learning_rate": 1.8271772047923166e-05,
+      "loss": 0.4276,
+      "step": 310
+    },
+    {
+      "epoch": 1.3983717012914094,
+      "grad_norm": 0.26661817104489643,
+      "learning_rate": 1.820077665065048e-05,
+      "loss": 0.413,
+      "step": 311
+    },
+    {
+      "epoch": 1.4028635597978663,
+      "grad_norm": 0.3061227775995877,
+      "learning_rate": 1.812966961183924e-05,
+      "loss": 0.4329,
+      "step": 312
+    },
+    {
+      "epoch": 1.4073554183043235,
+      "grad_norm": 0.29634237723914175,
+      "learning_rate": 1.805845288743921e-05,
+      "loss": 0.4249,
+      "step": 313
+    },
+    {
+      "epoch": 1.4118472768107804,
+      "grad_norm": 0.3301297214638871,
+      "learning_rate": 1.7987128436417303e-05,
+      "loss": 0.4588,
+      "step": 314
+    },
+    {
+      "epoch": 1.4163391353172374,
+      "grad_norm": 0.3698658191349894,
+      "learning_rate": 1.791569822070368e-05,
+      "loss": 0.4245,
+      "step": 315
+    },
+    {
+      "epoch": 1.4208309938236945,
+      "grad_norm": 0.22262993712499923,
+      "learning_rate": 1.7844164205137772e-05,
+      "loss": 0.424,
+      "step": 316
+    },
+    {
+      "epoch": 1.4253228523301515,
+      "grad_norm": 0.37654108374839934,
+      "learning_rate": 1.7772528357414252e-05,
+      "loss": 0.453,
+      "step": 317
+    },
+    {
+      "epoch": 1.4298147108366086,
+      "grad_norm": 0.2594395877577525,
+      "learning_rate": 1.770079264802891e-05,
+      "loss": 0.4296,
+      "step": 318
+    },
+    {
+      "epoch": 1.4343065693430657,
+      "grad_norm": 0.3514445535834096,
+      "learning_rate": 1.762895905022443e-05,
+      "loss": 0.4407,
+      "step": 319
+    },
+    {
+      "epoch": 1.4387984278495227,
+      "grad_norm": 0.2476040970236891,
+      "learning_rate": 1.7557029539936132e-05,
+      "loss": 0.4324,
+      "step": 320
+    },
+    {
+      "epoch": 1.4432902863559798,
+      "grad_norm": 0.31874415328292965,
+      "learning_rate": 1.7485006095737602e-05,
+      "loss": 0.445,
+      "step": 321
+    },
+    {
+      "epoch": 1.4477821448624368,
+      "grad_norm": 0.2583979409843575,
+      "learning_rate": 1.7412890698786298e-05,
+      "loss": 0.4244,
+      "step": 322
+    },
+    {
+      "epoch": 1.452274003368894,
+      "grad_norm": 0.3144942018984089,
+      "learning_rate": 1.734068533276901e-05,
+      "loss": 0.4443,
+      "step": 323
+    },
+    {
+      "epoch": 1.456765861875351,
+      "grad_norm": 0.27690020636097235,
+      "learning_rate": 1.726839198384733e-05,
+      "loss": 0.4364,
+      "step": 324
+    },
+    {
+      "epoch": 1.461257720381808,
+      "grad_norm": 0.26426874913937026,
+      "learning_rate": 1.7196012640603007e-05,
+      "loss": 0.4383,
+      "step": 325
+    },
+    {
+      "epoch": 1.465749578888265,
+      "grad_norm": 0.24541716474302672,
+      "learning_rate": 1.7123549293983246e-05,
+      "loss": 0.4374,
+      "step": 326
+    },
+    {
+      "epoch": 1.4702414373947221,
+      "grad_norm": 0.24834945996912894,
+      "learning_rate": 1.7051003937245932e-05,
+      "loss": 0.4126,
+      "step": 327
+    },
+    {
+      "epoch": 1.4747332959011792,
+      "grad_norm": 0.2624637386600607,
+      "learning_rate": 1.6978378565904823e-05,
+      "loss": 0.453,
+      "step": 328
+    },
+    {
+      "epoch": 1.479225154407636,
+      "grad_norm": 0.27142087832902706,
+      "learning_rate": 1.6905675177674652e-05,
+      "loss": 0.4194,
+      "step": 329
+    },
+    {
+      "epoch": 1.4837170129140933,
+      "grad_norm": 0.27566892739796234,
+      "learning_rate": 1.683289577241615e-05,
+      "loss": 0.4456,
+      "step": 330
+    },
+    {
+      "epoch": 1.4882088714205501,
+      "grad_norm": 0.2700242393817183,
+      "learning_rate": 1.676004235208109e-05,
+      "loss": 0.4367,
+      "step": 331
+    },
+    {
+      "epoch": 1.4927007299270074,
+      "grad_norm": 0.3142461399016386,
+      "learning_rate": 1.668711692065716e-05,
+      "loss": 0.4247,
+      "step": 332
+    },
+    {
+      "epoch": 1.4971925884334643,
+      "grad_norm": 0.25043376604704976,
+      "learning_rate": 1.661412148411288e-05,
+      "loss": 0.4326,
+      "step": 333
+    },
+    {
+      "epoch": 1.5016844469399215,
+      "grad_norm": 0.2826302795105161,
+      "learning_rate": 1.6541058050342402e-05,
+      "loss": 0.438,
+      "step": 334
+    },
+    {
+      "epoch": 1.5061763054463784,
+      "grad_norm": 0.27506460881743333,
+      "learning_rate": 1.646792862911028e-05,
+      "loss": 0.4358,
+      "step": 335
+    },
+    {
+      "epoch": 1.5106681639528357,
+      "grad_norm": 0.30976807555293234,
+      "learning_rate": 1.6394735231996212e-05,
+      "loss": 0.443,
+      "step": 336
+    },
+    {
+      "epoch": 1.5151600224592925,
+      "grad_norm": 0.3182926560081222,
+      "learning_rate": 1.632147987233967e-05,
+      "loss": 0.4284,
+      "step": 337
+    },
+    {
+      "epoch": 1.5196518809657495,
+      "grad_norm": 0.3073893367627004,
+      "learning_rate": 1.624816456518455e-05,
+      "loss": 0.421,
+      "step": 338
+    },
+    {
+      "epoch": 1.5241437394722066,
+      "grad_norm": 0.3026151787871035,
+      "learning_rate": 1.617479132722372e-05,
+      "loss": 0.4345,
+      "step": 339
+    },
+    {
+      "epoch": 1.5286355979786637,
+      "grad_norm": 0.3488262304344424,
+      "learning_rate": 1.6101362176743562e-05,
+      "loss": 0.4304,
+      "step": 340
+    },
+    {
+      "epoch": 1.5331274564851207,
+      "grad_norm": 0.31909400236183205,
+      "learning_rate": 1.6027879133568455e-05,
+      "loss": 0.4367,
+      "step": 341
+    },
+    {
+      "epoch": 1.5376193149915778,
+      "grad_norm": 0.3507006208659429,
+      "learning_rate": 1.59543442190052e-05,
+      "loss": 0.4286,
+      "step": 342
+    },
+    {
+      "epoch": 1.5421111734980348,
+      "grad_norm": 0.3247486352951706,
+      "learning_rate": 1.5880759455787446e-05,
+      "loss": 0.444,
+      "step": 343
+    },
+    {
+      "epoch": 1.546603032004492,
+      "grad_norm": 0.3157749827732778,
+      "learning_rate": 1.5807126868020016e-05,
+      "loss": 0.4354,
+      "step": 344
+    },
+    {
+      "epoch": 1.551094890510949,
+      "grad_norm": 0.3169873921812676,
+      "learning_rate": 1.5733448481123264e-05,
+      "loss": 0.4236,
+      "step": 345
+    },
+    {
+      "epoch": 1.5555867490174058,
+      "grad_norm": 0.33198535545378055,
+      "learning_rate": 1.565972632177734e-05,
+      "loss": 0.4386,
+      "step": 346
+    },
+    {
+      "epoch": 1.560078607523863,
+      "grad_norm": 0.3083260440048532,
+      "learning_rate": 1.558596241786645e-05,
+      "loss": 0.4282,
+      "step": 347
+    },
+    {
+      "epoch": 1.56457046603032,
+      "grad_norm": 0.3357526579603315,
+      "learning_rate": 1.551215879842307e-05,
+      "loss": 0.4178,
+      "step": 348
+    },
+    {
+      "epoch": 1.5690623245367772,
+      "grad_norm": 0.3287950186329193,
+      "learning_rate": 1.543831749357214e-05,
+      "loss": 0.431,
+      "step": 349
+    },
+    {
+      "epoch": 1.573554183043234,
+      "grad_norm": 0.3260432629951177,
+      "learning_rate": 1.5364440534475205e-05,
+      "loss": 0.4486,
+      "step": 350
+    },
+    {
+      "epoch": 1.5780460415496913,
+      "grad_norm": 0.29274164579590756,
+      "learning_rate": 1.5290529953274573e-05,
+      "loss": 0.4348,
+      "step": 351
+    },
+    {
+      "epoch": 1.5825379000561481,
+      "grad_norm": 0.2793617167300319,
+      "learning_rate": 1.521658778303739e-05,
+      "loss": 0.4339,
+      "step": 352
+    },
+    {
+      "epoch": 1.5870297585626054,
+      "grad_norm": 0.22729692344689448,
+      "learning_rate": 1.5142616057699725e-05,
+      "loss": 0.4088,
+      "step": 353
+    },
+    {
+      "epoch": 1.5915216170690623,
+      "grad_norm": 0.2706527199288221,
+      "learning_rate": 1.506861681201062e-05,
+      "loss": 0.4487,
+      "step": 354
+    },
+    {
+      "epoch": 1.5960134755755195,
+      "grad_norm": 0.23763623498472133,
+      "learning_rate": 1.4994592081476132e-05,
+      "loss": 0.4399,
+      "step": 355
+    },
+    {
+      "epoch": 1.6005053340819764,
+      "grad_norm": 0.2289251582859145,
+      "learning_rate": 1.4920543902303326e-05,
+      "loss": 0.4114,
+      "step": 356
+    },
+    {
+      "epoch": 1.6049971925884334,
+      "grad_norm": 0.22999268807574383,
+      "learning_rate": 1.4846474311344276e-05,
+      "loss": 0.4249,
+      "step": 357
+    },
+    {
+      "epoch": 1.6094890510948905,
+      "grad_norm": 0.2550402742123251,
+      "learning_rate": 1.4772385346040023e-05,
+      "loss": 0.4414,
+      "step": 358
+    },
+    {
+      "epoch": 1.6139809096013475,
+      "grad_norm": 0.28794877883820574,
+      "learning_rate": 1.4698279044364549e-05,
+      "loss": 0.4418,
+      "step": 359
+    },
+    {
+      "epoch": 1.6184727681078046,
+      "grad_norm": 0.2587286922700844,
+      "learning_rate": 1.4624157444768707e-05,
+      "loss": 0.4279,
+      "step": 360
+    },
+    {
+      "epoch": 1.6229646266142617,
+      "grad_norm": 0.26948710334654885,
+      "learning_rate": 1.4550022586124148e-05,
+      "loss": 0.4315,
+      "step": 361
+    },
+    {
+      "epoch": 1.6274564851207187,
+      "grad_norm": 0.26259124764610464,
+      "learning_rate": 1.4475876507667255e-05,
+      "loss": 0.4349,
+      "step": 362
+    },
+    {
+      "epoch": 1.6319483436271758,
+      "grad_norm": 0.23935177475002697,
+      "learning_rate": 1.4401721248943016e-05,
+      "loss": 0.4092,
+      "step": 363
+    },
+    {
+      "epoch": 1.6364402021336328,
+      "grad_norm": 0.23992737233823272,
+      "learning_rate": 1.432755884974896e-05,
+      "loss": 0.437,
+      "step": 364
+    },
+    {
+      "epoch": 1.6409320606400897,
+      "grad_norm": 0.22177963951273988,
+      "learning_rate": 1.4253391350079015e-05,
+      "loss": 0.4175,
+      "step": 365
+    },
+    {
+      "epoch": 1.645423919146547,
+      "grad_norm": 0.2670621296838188,
+      "learning_rate": 1.417922079006742e-05,
+      "loss": 0.4287,
+      "step": 366
+    },
+    {
+      "epoch": 1.6499157776530038,
+      "grad_norm": 0.23007364031192343,
+      "learning_rate": 1.4105049209932585e-05,
+      "loss": 0.4315,
+      "step": 367
+    },
+    {
+      "epoch": 1.654407636159461,
+      "grad_norm": 0.2862891319622476,
+      "learning_rate": 1.4030878649920989e-05,
+      "loss": 0.4105,
+      "step": 368
+    },
+    {
+      "epoch": 1.658899494665918,
+      "grad_norm": 0.2542826948774142,
+      "learning_rate": 1.3956711150251044e-05,
+      "loss": 0.4418,
+      "step": 369
+    },
+    {
+      "epoch": 1.6633913531723752,
+      "grad_norm": 0.25016704641271026,
+      "learning_rate": 1.3882548751056987e-05,
+      "loss": 0.4234,
+      "step": 370
+    },
+    {
+      "epoch": 1.667883211678832,
+      "grad_norm": 0.31248484686645767,
+      "learning_rate": 1.380839349233275e-05,
+      "loss": 0.4246,
+      "step": 371
+    },
+    {
+      "epoch": 1.6723750701852893,
+      "grad_norm": 0.2493758279331931,
+      "learning_rate": 1.3734247413875855e-05,
+      "loss": 0.4417,
+      "step": 372
+    },
+    {
+      "epoch": 1.6768669286917461,
+      "grad_norm": 0.33089735282060806,
+      "learning_rate": 1.36601125552313e-05,
+      "loss": 0.4453,
+      "step": 373
+    },
+    {
+      "epoch": 1.6813587871982034,
+      "grad_norm": 0.2358526660603175,
+      "learning_rate": 1.3585990955635459e-05,
+      "loss": 0.4138,
+      "step": 374
+    },
+    {
+      "epoch": 1.6858506457046603,
+      "grad_norm": 0.2836876613895396,
+      "learning_rate": 1.3511884653959981e-05,
+      "loss": 0.4235,
+      "step": 375
+    },
+    {
+      "epoch": 1.6903425042111173,
+      "grad_norm": 0.2796457407843088,
+      "learning_rate": 1.3437795688655726e-05,
+      "loss": 0.4355,
+      "step": 376
+    },
+    {
+      "epoch": 1.6948343627175744,
+      "grad_norm": 0.252086707823967,
+      "learning_rate": 1.3363726097696673e-05,
+      "loss": 0.433,
+      "step": 377
+    },
+    {
+      "epoch": 1.6993262212240314,
+      "grad_norm": 0.3504937797572051,
+      "learning_rate": 1.3289677918523868e-05,
+      "loss": 0.4314,
+      "step": 378
+    },
+    {
+      "epoch": 1.7038180797304885,
+      "grad_norm": 0.23392274241262292,
+      "learning_rate": 1.321565318798938e-05,
+      "loss": 0.4234,
+      "step": 379
+    },
+    {
+      "epoch": 1.7083099382369455,
+      "grad_norm": 0.2527199944902928,
+      "learning_rate": 1.3141653942300279e-05,
+      "loss": 0.4265,
+      "step": 380
+    },
+    {
+      "epoch": 1.7128017967434026,
+      "grad_norm": 0.27218994035462163,
+      "learning_rate": 1.3067682216962613e-05,
+      "loss": 0.4428,
+      "step": 381
+    },
+    {
+      "epoch": 1.7172936552498597,
+      "grad_norm": 0.24181872866822765,
+      "learning_rate": 1.2993740046725429e-05,
+      "loss": 0.4251,
+      "step": 382
+    },
+    {
+      "epoch": 1.7217855137563167,
+      "grad_norm": 0.23454523482773984,
+      "learning_rate": 1.2919829465524797e-05,
+      "loss": 0.4508,
+      "step": 383
+    },
+    {
+      "epoch": 1.7262773722627736,
+      "grad_norm": 0.2141716112253175,
+      "learning_rate": 1.2845952506427866e-05,
+      "loss": 0.4209,
+      "step": 384
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.21483571836516002,
+      "learning_rate": 1.2772111201576932e-05,
+      "loss": 0.4297,
+      "step": 385
+    },
+    {
+      "epoch": 1.7352610892756877,
+      "grad_norm": 0.22676478933835803,
+      "learning_rate": 1.2698307582133551e-05,
+      "loss": 0.4436,
+      "step": 386
+    },
+    {
+      "epoch": 1.739752947782145,
+      "grad_norm": 0.21804047909429286,
+      "learning_rate": 1.2624543678222662e-05,
+      "loss": 0.4284,
+      "step": 387
+    },
+    {
+      "epoch": 1.7442448062886018,
+      "grad_norm": 0.2161320011302351,
+      "learning_rate": 1.255082151887674e-05,
+      "loss": 0.4199,
+      "step": 388
+    },
+    {
+      "epoch": 1.748736664795059,
+      "grad_norm": 0.23536261615237247,
+      "learning_rate": 1.2477143131979989e-05,
+      "loss": 0.4416,
+      "step": 389
+    },
+    {
+      "epoch": 1.753228523301516,
+      "grad_norm": 0.19516762515369254,
+      "learning_rate": 1.2403510544212563e-05,
+      "loss": 0.422,
+      "step": 390
+    },
+    {
+      "epoch": 1.7577203818079732,
+      "grad_norm": 0.2597761337222779,
+      "learning_rate": 1.2329925780994805e-05,
+      "loss": 0.4555,
+      "step": 391
+    },
+    {
+      "epoch": 1.76221224031443,
+      "grad_norm": 0.2365793711701312,
+      "learning_rate": 1.225639086643155e-05,
+      "loss": 0.4198,
+      "step": 392
+    },
+    {
+      "epoch": 1.7667040988208873,
+      "grad_norm": 0.2697456739133074,
+      "learning_rate": 1.218290782325644e-05,
+      "loss": 0.4356,
+      "step": 393
+    },
+    {
+      "epoch": 1.7711959573273441,
+      "grad_norm": 0.27620361441234015,
+      "learning_rate": 1.2109478672776284e-05,
+      "loss": 0.4203,
+      "step": 394
+    },
+    {
+      "epoch": 1.7756878158338012,
+      "grad_norm": 0.24512868512343508,
+      "learning_rate": 1.2036105434815453e-05,
+      "loss": 0.4203,
+      "step": 395
+    },
+    {
+      "epoch": 1.7801796743402583,
+      "grad_norm": 0.26248183198436537,
+      "learning_rate": 1.196279012766033e-05,
+      "loss": 0.4171,
+      "step": 396
+    },
+    {
+      "epoch": 1.7846715328467153,
+      "grad_norm": 0.22075536899568315,
+      "learning_rate": 1.1889534768003789e-05,
+      "loss": 0.417,
+      "step": 397
+    },
+    {
+      "epoch": 1.7891633913531724,
+      "grad_norm": 0.22534931235643543,
+      "learning_rate": 1.1816341370889721e-05,
+      "loss": 0.4307,
+      "step": 398
+    },
+    {
+      "epoch": 1.7936552498596294,
+      "grad_norm": 0.226631593581966,
+      "learning_rate": 1.1743211949657602e-05,
+      "loss": 0.4188,
+      "step": 399
+    },
+    {
+      "epoch": 1.7981471083660865,
+      "grad_norm": 0.23848047414834528,
+      "learning_rate": 1.1670148515887122e-05,
+      "loss": 0.4356,
+      "step": 400
+    },
+    {
+      "epoch": 1.8026389668725435,
+      "grad_norm": 0.19151331669734864,
+      "learning_rate": 1.1597153079342841e-05,
+      "loss": 0.436,
+      "step": 401
+    },
+    {
+      "epoch": 1.8071308253790006,
+      "grad_norm": 0.22870877041290116,
+      "learning_rate": 1.1524227647918911e-05,
+      "loss": 0.419,
+      "step": 402
+    },
+    {
+      "epoch": 1.8116226838854577,
+      "grad_norm": 0.21466616093632906,
+      "learning_rate": 1.1451374227583852e-05,
+      "loss": 0.4424,
+      "step": 403
+    },
+    {
+      "epoch": 1.8161145423919147,
+      "grad_norm": 0.25318885652943574,
+      "learning_rate": 1.1378594822325357e-05,
+      "loss": 0.4406,
+      "step": 404
+    },
+    {
+      "epoch": 1.8206064008983716,
+      "grad_norm": 0.20250073619587594,
+      "learning_rate": 1.1305891434095182e-05,
+      "loss": 0.4238,
+      "step": 405
+    },
+    {
+      "epoch": 1.8250982594048288,
+      "grad_norm": 0.2259040741342042,
+      "learning_rate": 1.1233266062754074e-05,
+      "loss": 0.4287,
+      "step": 406
+    },
+    {
+      "epoch": 1.8295901179112857,
+      "grad_norm": 0.21613984026340174,
+      "learning_rate": 1.116072070601676e-05,
+      "loss": 0.429,
+      "step": 407
+    },
+    {
+      "epoch": 1.834081976417743,
+      "grad_norm": 0.1986928299960945,
+      "learning_rate": 1.1088257359396997e-05,
+      "loss": 0.412,
+      "step": 408
+    },
+    {
+      "epoch": 1.8385738349241998,
+      "grad_norm": 0.23767036451906237,
+      "learning_rate": 1.1015878016152666e-05,
+      "loss": 0.4235,
+      "step": 409
+    },
+    {
+      "epoch": 1.843065693430657,
+      "grad_norm": 0.22229388428754168,
+      "learning_rate": 1.094358466723099e-05,
+      "loss": 0.4434,
+      "step": 410
+    },
+    {
+      "epoch": 1.847557551937114,
+      "grad_norm": 0.2652264081143693,
+      "learning_rate": 1.0871379301213705e-05,
+      "loss": 0.4181,
+      "step": 411
+    },
+    {
+      "epoch": 1.8520494104435712,
+      "grad_norm": 0.26380236575167754,
+      "learning_rate": 1.0799263904262397e-05,
+      "loss": 0.4242,
+      "step": 412
+    },
+    {
+      "epoch": 1.856541268950028,
+      "grad_norm": 0.23160769920164645,
+      "learning_rate": 1.0727240460063872e-05,
+      "loss": 0.4461,
+      "step": 413
+    },
+    {
+      "epoch": 1.861033127456485,
+      "grad_norm": 0.2277570708946221,
+      "learning_rate": 1.0655310949775572e-05,
+      "loss": 0.405,
+      "step": 414
+    },
+    {
+      "epoch": 1.8655249859629421,
+      "grad_norm": 0.21522197135984833,
+      "learning_rate": 1.0583477351971092e-05,
+      "loss": 0.4244,
+      "step": 415
+    },
+    {
+      "epoch": 1.8700168444693992,
+      "grad_norm": 0.24137253643395323,
+      "learning_rate": 1.0511741642585747e-05,
+      "loss": 0.4569,
+      "step": 416
+    },
+    {
+      "epoch": 1.8745087029758563,
+      "grad_norm": 0.22284276780245038,
+      "learning_rate": 1.044010579486223e-05,
+      "loss": 0.4209,
+      "step": 417
+    },
+    {
+      "epoch": 1.8790005614823133,
+      "grad_norm": 0.20682215213756386,
+      "learning_rate": 1.0368571779296324e-05,
+      "loss": 0.427,
+      "step": 418
+    },
+    {
+      "epoch": 1.8834924199887704,
+      "grad_norm": 0.1903967688448084,
+      "learning_rate": 1.0297141563582698e-05,
+      "loss": 0.439,
+      "step": 419
+    },
+    {
+      "epoch": 1.8879842784952274,
+      "grad_norm": 0.22460948249947274,
+      "learning_rate": 1.0225817112560795e-05,
+      "loss": 0.4154,
+      "step": 420
+    },
+    {
+      "epoch": 1.8924761370016845,
+      "grad_norm": 0.20928856005143603,
+      "learning_rate": 1.0154600388160767e-05,
+      "loss": 0.4267,
+      "step": 421
+    },
+    {
+      "epoch": 1.8969679955081415,
+      "grad_norm": 0.1970852478249293,
+      "learning_rate": 1.0083493349349523e-05,
+      "loss": 0.424,
+      "step": 422
+    },
+    {
+      "epoch": 1.9014598540145986,
+      "grad_norm": 0.2189181076915757,
+      "learning_rate": 1.0012497952076838e-05,
+      "loss": 0.427,
+      "step": 423
+    },
+    {
+      "epoch": 1.9059517125210554,
+      "grad_norm": 0.2009004896673229,
+      "learning_rate": 9.941616149221549e-06,
+      "loss": 0.4413,
+      "step": 424
+    },
+    {
+      "epoch": 1.9104435710275127,
+      "grad_norm": 0.20621935633909672,
+      "learning_rate": 9.870849890537826e-06,
+      "loss": 0.4133,
+      "step": 425
+    },
+    {
+      "epoch": 1.9149354295339696,
+      "grad_norm": 0.20677347142692576,
+      "learning_rate": 9.800201122601579e-06,
+      "loss": 0.4177,
+      "step": 426
+    },
+    {
+      "epoch": 1.9194272880404268,
+      "grad_norm": 0.216610474906858,
+      "learning_rate": 9.72967178875686e-06,
+      "loss": 0.4493,
+      "step": 427
+    },
+    {
+      "epoch": 1.9239191465468837,
+      "grad_norm": 0.19615798611811022,
+      "learning_rate": 9.659263829062443e-06,
+      "loss": 0.4235,
+      "step": 428
+    },
+    {
+      "epoch": 1.928411005053341,
+      "grad_norm": 0.21389343121593873,
+      "learning_rate": 9.588979180238454e-06,
+      "loss": 0.4275,
+      "step": 429
+    },
+    {
+      "epoch": 1.9329028635597978,
+      "grad_norm": 0.1783568108760251,
+      "learning_rate": 9.518819775613077e-06,
+      "loss": 0.417,
+      "step": 430
+    },
+    {
+      "epoch": 1.937394722066255,
+      "grad_norm": 0.19382064589217926,
+      "learning_rate": 9.44878754506941e-06,
+      "loss": 0.427,
+      "step": 431
+    },
+    {
+      "epoch": 1.941886580572712,
+      "grad_norm": 0.1868776486518385,
+      "learning_rate": 9.37888441499234e-06,
+      "loss": 0.4344,
+      "step": 432
+    },
+    {
+      "epoch": 1.946378439079169,
+      "grad_norm": 0.19196173137999686,
+      "learning_rate": 9.309112308215579e-06,
+      "loss": 0.4218,
+      "step": 433
+    },
+    {
+      "epoch": 1.950870297585626,
+      "grad_norm": 0.18168839883358026,
+      "learning_rate": 9.239473143968768e-06,
+      "loss": 0.4164,
+      "step": 434
+    },
+    {
+      "epoch": 1.955362156092083,
+      "grad_norm": 0.19586388090243548,
+      "learning_rate": 9.169968837824679e-06,
+      "loss": 0.4346,
+      "step": 435
+    },
+    {
+      "epoch": 1.9598540145985401,
+      "grad_norm": 0.1968196503268063,
+      "learning_rate": 9.100601301646527e-06,
+      "loss": 0.435,
+      "step": 436
+    },
+    {
+      "epoch": 1.9643458731049972,
+      "grad_norm": 0.1892076438600969,
+      "learning_rate": 9.031372443535382e-06,
+      "loss": 0.4277,
+      "step": 437
+    },
+    {
+      "epoch": 1.9688377316114543,
+      "grad_norm": 0.17997215604618058,
+      "learning_rate": 8.96228416777767e-06,
+      "loss": 0.4246,
+      "step": 438
+    },
+    {
+      "epoch": 1.9733295901179113,
+      "grad_norm": 0.21280790454241325,
+      "learning_rate": 8.893338374792817e-06,
+      "loss": 0.4191,
+      "step": 439
+    },
+    {
+      "epoch": 1.9778214486243684,
+      "grad_norm": 0.2122734568816957,
+      "learning_rate": 8.824536961080943e-06,
+      "loss": 0.4326,
+      "step": 440
+    },
+    {
+      "epoch": 1.9823133071308254,
+      "grad_norm": 0.18374860642807206,
+      "learning_rate": 8.75588181917072e-06,
+      "loss": 0.4201,
+      "step": 441
+    },
+    {
+      "epoch": 1.9868051656372825,
+      "grad_norm": 0.19354659406561456,
+      "learning_rate": 8.687374837567298e-06,
+      "loss": 0.4138,
+      "step": 442
+    },
+    {
+      "epoch": 1.9912970241437393,
+      "grad_norm": 0.21445379315505658,
+      "learning_rate": 8.619017900700374e-06,
+      "loss": 0.4393,
+      "step": 443
+    },
+    {
+      "epoch": 1.9957888826501966,
+      "grad_norm": 0.20362548247073203,
+      "learning_rate": 8.550812888872337e-06,
+      "loss": 0.4239,
+      "step": 444
+    },
+    {
+      "epoch": 2.0016844469399215,
+      "grad_norm": 0.38025114081538913,
+      "learning_rate": 8.48276167820656e-06,
+      "loss": 0.6928,
+      "step": 445
+    },
+    {
+      "epoch": 2.0061763054463784,
+      "grad_norm": 0.24372193822660385,
+      "learning_rate": 8.414866140595784e-06,
+      "loss": 0.4081,
+      "step": 446
+    },
+    {
+      "epoch": 2.0106681639528357,
+      "grad_norm": 0.2776148147113284,
+      "learning_rate": 8.347128143650638e-06,
+      "loss": 0.4048,
+      "step": 447
+    },
+    {
+      "epoch": 2.0151600224592925,
+      "grad_norm": 0.23373713524039486,
+      "learning_rate": 8.279549550648262e-06,
+      "loss": 0.3931,
+      "step": 448
+    },
+    {
+      "epoch": 2.0196518809657498,
+      "grad_norm": 0.22958395982377883,
+      "learning_rate": 8.212132220481046e-06,
+      "loss": 0.4008,
+      "step": 449
+    },
+    {
+      "epoch": 2.0241437394722066,
+      "grad_norm": 0.29063552467815107,
+      "learning_rate": 8.144878007605514e-06,
+      "loss": 0.4154,
+      "step": 450
+    },
+    {
+      "epoch": 2.028635597978664,
+      "grad_norm": 0.21727860645522265,
+      "learning_rate": 8.077788761991291e-06,
+      "loss": 0.4162,
+      "step": 451
+    },
+    {
+      "epoch": 2.0331274564851207,
+      "grad_norm": 0.24646687235941006,
+      "learning_rate": 8.01086632907024e-06,
+      "loss": 0.4003,
+      "step": 452
+    },
+    {
+      "epoch": 2.0376193149915776,
+      "grad_norm": 0.26443217492234267,
+      "learning_rate": 7.944112549685684e-06,
+      "loss": 0.3955,
+      "step": 453
+    },
+    {
+      "epoch": 2.042111173498035,
+      "grad_norm": 0.2351417166729191,
+      "learning_rate": 7.877529260041764e-06,
+      "loss": 0.4102,
+      "step": 454
+    },
+    {
+      "epoch": 2.0466030320044917,
+      "grad_norm": 0.23415549490836263,
+      "learning_rate": 7.811118291652952e-06,
+      "loss": 0.4138,
+      "step": 455
+    },
+    {
+      "epoch": 2.051094890510949,
+      "grad_norm": 0.23596205188013428,
+      "learning_rate": 7.744881471293649e-06,
+      "loss": 0.3935,
+      "step": 456
+    },
+    {
+      "epoch": 2.055586749017406,
+      "grad_norm": 0.2535757872007115,
+      "learning_rate": 7.678820620947954e-06,
+      "loss": 0.4042,
+      "step": 457
+    },
+    {
+      "epoch": 2.060078607523863,
+      "grad_norm": 0.20639323573803794,
+      "learning_rate": 7.612937557759534e-06,
+      "loss": 0.4161,
+      "step": 458
+    },
+    {
+      "epoch": 2.06457046603032,
+      "grad_norm": 0.22362539726003386,
+      "learning_rate": 7.54723409398164e-06,
+      "loss": 0.421,
+      "step": 459
+    },
+    {
+      "epoch": 2.069062324536777,
+      "grad_norm": 0.22127045680743251,
+      "learning_rate": 7.481712036927267e-06,
+      "loss": 0.3861,
+      "step": 460
+    },
+    {
+      "epoch": 2.073554183043234,
+      "grad_norm": 0.19232753420525026,
+      "learning_rate": 7.416373188919427e-06,
+      "loss": 0.3922,
+      "step": 461
+    },
+    {
+      "epoch": 2.0780460415496913,
+      "grad_norm": 0.20966827389795048,
+      "learning_rate": 7.351219347241587e-06,
+      "loss": 0.4049,
+      "step": 462
+    },
+    {
+      "epoch": 2.082537900056148,
+      "grad_norm": 0.19401074372527943,
+      "learning_rate": 7.286252304088215e-06,
+      "loss": 0.4038,
+      "step": 463
+    },
+    {
+      "epoch": 2.0870297585626054,
+      "grad_norm": 0.2033592701308335,
+      "learning_rate": 7.221473846515494e-06,
+      "loss": 0.402,
+      "step": 464
+    },
+    {
+      "epoch": 2.0915216170690623,
+      "grad_norm": 0.19991172839172033,
+      "learning_rate": 7.156885756392161e-06,
+      "loss": 0.3956,
+      "step": 465
+    },
+    {
+      "epoch": 2.0960134755755195,
+      "grad_norm": 0.19027112767750493,
+      "learning_rate": 7.092489810350491e-06,
+      "loss": 0.3912,
+      "step": 466
+    },
+    {
+      "epoch": 2.1005053340819764,
+      "grad_norm": 0.18837141345128516,
+      "learning_rate": 7.028287779737433e-06,
+      "loss": 0.4066,
+      "step": 467
+    },
+    {
+      "epoch": 2.1049971925884337,
+      "grad_norm": 0.18954944621965775,
+      "learning_rate": 6.964281430565874e-06,
+      "loss": 0.4048,
+      "step": 468
+    },
+    {
+      "epoch": 2.1094890510948905,
+      "grad_norm": 0.19737625715167087,
+      "learning_rate": 6.9004725234660754e-06,
+      "loss": 0.3929,
+      "step": 469
+    },
+    {
+      "epoch": 2.1139809096013478,
+      "grad_norm": 0.20640221374882523,
+      "learning_rate": 6.836862813637232e-06,
+      "loss": 0.4214,
+      "step": 470
+    },
+    {
+      "epoch": 2.1184727681078046,
+      "grad_norm": 0.19729058573447977,
+      "learning_rate": 6.773454050799198e-06,
+      "loss": 0.3956,
+      "step": 471
+    },
+    {
+      "epoch": 2.1229646266142614,
+      "grad_norm": 0.2210511604280809,
+      "learning_rate": 6.710247979144353e-06,
+      "loss": 0.4067,
+      "step": 472
+    },
+    {
+      "epoch": 2.1274564851207187,
+      "grad_norm": 0.21144716690556348,
+      "learning_rate": 6.6472463372896255e-06,
+      "loss": 0.4087,
+      "step": 473
+    },
+    {
+      "epoch": 2.1319483436271756,
+      "grad_norm": 0.21401394951272074,
+      "learning_rate": 6.584450858228674e-06,
+      "loss": 0.4155,
+      "step": 474
+    },
+    {
+      "epoch": 2.136440202133633,
+      "grad_norm": 0.2110307270757492,
+      "learning_rate": 6.5218632692842e-06,
+      "loss": 0.3982,
+      "step": 475
+    },
+    {
+      "epoch": 2.1409320606400897,
+      "grad_norm": 0.2188562401970234,
+      "learning_rate": 6.459485292060468e-06,
+      "loss": 0.4012,
+      "step": 476
+    },
+    {
+      "epoch": 2.145423919146547,
+      "grad_norm": 0.20122965285822791,
+      "learning_rate": 6.397318642395911e-06,
+      "loss": 0.4009,
+      "step": 477
+    },
+    {
+      "epoch": 2.149915777653004,
+      "grad_norm": 0.18963676557334538,
+      "learning_rate": 6.335365030315953e-06,
+      "loss": 0.3972,
+      "step": 478
+    },
+    {
+      "epoch": 2.154407636159461,
+      "grad_norm": 0.1936342170939391,
+      "learning_rate": 6.273626159985971e-06,
+      "loss": 0.4017,
+      "step": 479
+    },
+    {
+      "epoch": 2.158899494665918,
+      "grad_norm": 0.1617872571717289,
+      "learning_rate": 6.2121037296644114e-06,
+      "loss": 0.4028,
+      "step": 480
+    },
+    {
+      "epoch": 2.163391353172375,
+      "grad_norm": 0.1927846186688275,
+      "learning_rate": 6.1507994316560836e-06,
+      "loss": 0.4188,
+      "step": 481
+    },
+    {
+      "epoch": 2.167883211678832,
+      "grad_norm": 0.18495924225832375,
+      "learning_rate": 6.089714952265603e-06,
+      "loss": 0.37,
+      "step": 482
+    },
+    {
+      "epoch": 2.1723750701852893,
+      "grad_norm": 0.188904946062925,
+      "learning_rate": 6.028851971751007e-06,
+      "loss": 0.4305,
+      "step": 483
+    },
+    {
+      "epoch": 2.176866928691746,
+      "grad_norm": 0.18764565684075873,
+      "learning_rate": 5.968212164277541e-06,
+      "loss": 0.3958,
+      "step": 484
+    },
+    {
+      "epoch": 2.1813587871982034,
+      "grad_norm": 0.17630565500194345,
+      "learning_rate": 5.9077971978716e-06,
+      "loss": 0.4198,
+      "step": 485
+    },
+    {
+      "epoch": 2.1858506457046603,
+      "grad_norm": 0.1773620762541041,
+      "learning_rate": 5.8476087343748505e-06,
+      "loss": 0.4195,
+      "step": 486
+    },
+    {
+      "epoch": 2.1903425042111175,
+      "grad_norm": 0.1747002407714538,
+      "learning_rate": 5.787648429398515e-06,
+      "loss": 0.3914,
+      "step": 487
+    },
+    {
+      "epoch": 2.1948343627175744,
+      "grad_norm": 0.16985548830481717,
+      "learning_rate": 5.727917932277831e-06,
+      "loss": 0.414,
+      "step": 488
+    },
+    {
+      "epoch": 2.199326221224031,
+      "grad_norm": 0.18212297018180695,
+      "learning_rate": 5.668418886026689e-06,
+      "loss": 0.3993,
+      "step": 489
+    },
+    {
+      "epoch": 2.2038180797304885,
+      "grad_norm": 0.16983942237475746,
+      "learning_rate": 5.609152927292432e-06,
+      "loss": 0.3934,
+      "step": 490
+    },
+    {
+      "epoch": 2.2083099382369458,
+      "grad_norm": 0.17190288036434193,
+      "learning_rate": 5.55012168631083e-06,
+      "loss": 0.3982,
+      "step": 491
+    },
+    {
+      "epoch": 2.2128017967434026,
+      "grad_norm": 0.18111711688919532,
+      "learning_rate": 5.491326786861248e-06,
+      "loss": 0.3956,
+      "step": 492
+    },
+    {
+      "epoch": 2.2172936552498594,
+      "grad_norm": 0.18152369105301266,
+      "learning_rate": 5.432769846221986e-06,
+      "loss": 0.4166,
+      "step": 493
+    },
+    {
+      "epoch": 2.2217855137563167,
+      "grad_norm": 0.1633243062989284,
+      "learning_rate": 5.374452475125769e-06,
+      "loss": 0.3881,
+      "step": 494
+    },
+    {
+      "epoch": 2.2262773722627736,
+      "grad_norm": 0.17295838406368827,
+      "learning_rate": 5.316376277715458e-06,
+      "loss": 0.3912,
+      "step": 495
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.17468906348146815,
+      "learning_rate": 5.258542851499929e-06,
+      "loss": 0.3956,
+      "step": 496
+    },
+    {
+      "epoch": 2.2352610892756877,
+      "grad_norm": 0.17244392435386519,
+      "learning_rate": 5.200953787310104e-06,
+      "loss": 0.4026,
+      "step": 497
+    },
+    {
+      "epoch": 2.239752947782145,
+      "grad_norm": 0.1596913829094359,
+      "learning_rate": 5.143610669255224e-06,
+      "loss": 0.3952,
+      "step": 498
+    },
+    {
+      "epoch": 2.244244806288602,
+      "grad_norm": 0.1725969837782697,
+      "learning_rate": 5.08651507467926e-06,
+      "loss": 0.3885,
+      "step": 499
+    },
+    {
+      "epoch": 2.248736664795059,
+      "grad_norm": 0.17719497977550294,
+      "learning_rate": 5.02966857411752e-06,
+      "loss": 0.4158,
+      "step": 500
+    },
+    {
+      "epoch": 2.253228523301516,
+      "grad_norm": 0.1564882926046542,
+      "learning_rate": 4.9730727312534544e-06,
+      "loss": 0.3998,
+      "step": 501
+    },
+    {
+      "epoch": 2.257720381807973,
+      "grad_norm": 0.1562726027662319,
+      "learning_rate": 4.916729102875645e-06,
+      "loss": 0.418,
+      "step": 502
+    },
+    {
+      "epoch": 2.26221224031443,
+      "grad_norm": 0.16350678522915907,
+      "learning_rate": 4.860639238834977e-06,
+      "loss": 0.3897,
+      "step": 503
+    },
+    {
+      "epoch": 2.2667040988208873,
+      "grad_norm": 0.1590466726049414,
+      "learning_rate": 4.8048046820020125e-06,
+      "loss": 0.4283,
+      "step": 504
+    },
+    {
+      "epoch": 2.271195957327344,
+      "grad_norm": 0.1598494755149621,
+      "learning_rate": 4.7492269682245465e-06,
+      "loss": 0.3976,
+      "step": 505
+    },
+    {
+      "epoch": 2.2756878158338014,
+      "grad_norm": 0.1564777609984641,
+      "learning_rate": 4.693907626285361e-06,
+      "loss": 0.3925,
+      "step": 506
+    },
+    {
+      "epoch": 2.2801796743402583,
+      "grad_norm": 0.15431022450548962,
+      "learning_rate": 4.638848177860173e-06,
+      "loss": 0.4004,
+      "step": 507
+    },
+    {
+      "epoch": 2.2846715328467155,
+      "grad_norm": 0.16548691474486177,
+      "learning_rate": 4.584050137475774e-06,
+      "loss": 0.4313,
+      "step": 508
+    },
+    {
+      "epoch": 2.2891633913531724,
+      "grad_norm": 0.1559165936066942,
+      "learning_rate": 4.529515012468387e-06,
+      "loss": 0.3981,
+      "step": 509
+    },
+    {
+      "epoch": 2.293655249859629,
+      "grad_norm": 0.1502614692328062,
+      "learning_rate": 4.475244302942176e-06,
+      "loss": 0.409,
+      "step": 510
+    },
+    {
+      "epoch": 2.2981471083660865,
+      "grad_norm": 0.17002770838666154,
+      "learning_rate": 4.421239501728004e-06,
+      "loss": 0.4111,
+      "step": 511
+    },
+    {
+      "epoch": 2.3026389668725433,
+      "grad_norm": 0.15405846526605765,
+      "learning_rate": 4.367502094342358e-06,
+      "loss": 0.4148,
+      "step": 512
+    },
+    {
+      "epoch": 2.3071308253790006,
+      "grad_norm": 0.16443857327327438,
+      "learning_rate": 4.3140335589464955e-06,
+      "loss": 0.4175,
+      "step": 513
+    },
+    {
+      "epoch": 2.3116226838854574,
+      "grad_norm": 0.16108894575419452,
+      "learning_rate": 4.260835366305778e-06,
+      "loss": 0.3954,
+      "step": 514
+    },
+    {
+      "epoch": 2.3161145423919147,
+      "grad_norm": 0.16581098420880921,
+      "learning_rate": 4.2079089797492155e-06,
+      "loss": 0.4029,
+      "step": 515
+    },
+    {
+      "epoch": 2.3206064008983716,
+      "grad_norm": 0.1462496764056089,
+      "learning_rate": 4.1552558551292194e-06,
+      "loss": 0.3914,
+      "step": 516
+    },
+    {
+      "epoch": 2.325098259404829,
+      "grad_norm": 0.16257688465964204,
+      "learning_rate": 4.1028774407815484e-06,
+      "loss": 0.4087,
+      "step": 517
+    },
+    {
+      "epoch": 2.3295901179112857,
+      "grad_norm": 0.15447353531940203,
+      "learning_rate": 4.050775177485476e-06,
+      "loss": 0.4116,
+      "step": 518
+    },
+    {
+      "epoch": 2.334081976417743,
+      "grad_norm": 0.15272812111654316,
+      "learning_rate": 3.998950498424157e-06,
+      "loss": 0.4011,
+      "step": 519
+    },
+    {
+      "epoch": 2.3385738349242,
+      "grad_norm": 0.14233004337736202,
+      "learning_rate": 3.947404829145199e-06,
+      "loss": 0.3923,
+      "step": 520
+    },
+    {
+      "epoch": 2.343065693430657,
+      "grad_norm": 0.1703317244482347,
+      "learning_rate": 3.896139587521457e-06,
+      "loss": 0.4405,
+      "step": 521
+    },
+    {
+      "epoch": 2.347557551937114,
+      "grad_norm": 0.14563012327083313,
+      "learning_rate": 3.84515618371203e-06,
+      "loss": 0.385,
+      "step": 522
+    },
+    {
+      "epoch": 2.352049410443571,
+      "grad_norm": 0.17742686794643436,
+      "learning_rate": 3.794456020123469e-06,
+      "loss": 0.4128,
+      "step": 523
+    },
+    {
+      "epoch": 2.356541268950028,
+      "grad_norm": 0.14926460855891024,
+      "learning_rate": 3.744040491371204e-06,
+      "loss": 0.3956,
+      "step": 524
+    },
+    {
+      "epoch": 2.3610331274564853,
+      "grad_norm": 0.14734023096974366,
+      "learning_rate": 3.693910984241176e-06,
+      "loss": 0.4134,
+      "step": 525
+    },
+    {
+      "epoch": 2.365524985962942,
+      "grad_norm": 0.17461277226346902,
+      "learning_rate": 3.6440688776517076e-06,
+      "loss": 0.3955,
+      "step": 526
+    },
+    {
+      "epoch": 2.370016844469399,
+      "grad_norm": 0.1829426578142121,
+      "learning_rate": 3.5945155426155476e-06,
+      "loss": 0.3952,
+      "step": 527
+    },
+    {
+      "epoch": 2.3745087029758563,
+      "grad_norm": 0.1473193865232448,
+      "learning_rate": 3.5452523422021744e-06,
+      "loss": 0.4198,
+      "step": 528
+    },
+    {
+      "epoch": 2.3790005614823135,
+      "grad_norm": 0.15970567054534546,
+      "learning_rate": 3.496280631500304e-06,
+      "loss": 0.4071,
+      "step": 529
+    },
+    {
+      "epoch": 2.3834924199887704,
+      "grad_norm": 0.16964921917368456,
+      "learning_rate": 3.447601757580604e-06,
+      "loss": 0.4254,
+      "step": 530
+    },
+    {
+      "epoch": 2.387984278495227,
+      "grad_norm": 0.16200193328776716,
+      "learning_rate": 3.399217059458649e-06,
+      "loss": 0.3874,
+      "step": 531
+    },
+    {
+      "epoch": 2.3924761370016845,
+      "grad_norm": 0.147997015119327,
+      "learning_rate": 3.3511278680580843e-06,
+      "loss": 0.4191,
+      "step": 532
+    },
+    {
+      "epoch": 2.3969679955081413,
+      "grad_norm": 0.15472131311543866,
+      "learning_rate": 3.303335506174018e-06,
+      "loss": 0.3967,
+      "step": 533
+    },
+    {
+      "epoch": 2.4014598540145986,
+      "grad_norm": 0.15792003965676266,
+      "learning_rate": 3.2558412884366303e-06,
+      "loss": 0.404,
+      "step": 534
+    },
+    {
+      "epoch": 2.4059517125210554,
+      "grad_norm": 0.14885399084267711,
+      "learning_rate": 3.2086465212750174e-06,
+      "loss": 0.419,
+      "step": 535
+    },
+    {
+      "epoch": 2.4104435710275127,
+      "grad_norm": 0.14435379373559862,
+      "learning_rate": 3.1617525028812533e-06,
+      "loss": 0.4005,
+      "step": 536
+    },
+    {
+      "epoch": 2.4149354295339696,
+      "grad_norm": 0.1419888428470599,
+      "learning_rate": 3.115160523174681e-06,
+      "loss": 0.4078,
+      "step": 537
+    },
+    {
+      "epoch": 2.419427288040427,
+      "grad_norm": 0.15641078822340915,
+      "learning_rate": 3.0688718637664205e-06,
+      "loss": 0.4163,
+      "step": 538
+    },
+    {
+      "epoch": 2.4239191465468837,
+      "grad_norm": 0.14302287152301948,
+      "learning_rate": 3.0228877979241345e-06,
+      "loss": 0.3743,
+      "step": 539
+    },
+    {
+      "epoch": 2.428411005053341,
+      "grad_norm": 0.14838140644695952,
+      "learning_rate": 2.9772095905369913e-06,
+      "loss": 0.4152,
+      "step": 540
+    },
+    {
+      "epoch": 2.432902863559798,
+      "grad_norm": 0.14296491653836946,
+      "learning_rate": 2.9318384980808727e-06,
+      "loss": 0.4059,
+      "step": 541
+    },
+    {
+      "epoch": 2.437394722066255,
+      "grad_norm": 0.15008286020132178,
+      "learning_rate": 2.8867757685838118e-06,
+      "loss": 0.3939,
+      "step": 542
+    },
+    {
+      "epoch": 2.441886580572712,
+      "grad_norm": 0.1403709429341766,
+      "learning_rate": 2.8420226415916757e-06,
+      "loss": 0.3997,
+      "step": 543
+    },
+    {
+      "epoch": 2.446378439079169,
+      "grad_norm": 0.15082563258882675,
+      "learning_rate": 2.7975803481340463e-06,
+      "loss": 0.4023,
+      "step": 544
+    },
+    {
+      "epoch": 2.450870297585626,
+      "grad_norm": 0.15784705073081262,
+      "learning_rate": 2.753450110690375e-06,
+      "loss": 0.4052,
+      "step": 545
+    },
+    {
+      "epoch": 2.4553621560920833,
+      "grad_norm": 0.13636735237201847,
+      "learning_rate": 2.709633143156353e-06,
+      "loss": 0.3799,
+      "step": 546
+    },
+    {
+      "epoch": 2.45985401459854,
+      "grad_norm": 0.1593373076088822,
+      "learning_rate": 2.666130650810515e-06,
+      "loss": 0.4042,
+      "step": 547
+    },
+    {
+      "epoch": 2.464345873104997,
+      "grad_norm": 0.1580770346070772,
+      "learning_rate": 2.6229438302810933e-06,
+      "loss": 0.4253,
+      "step": 548
+    },
+    {
+      "epoch": 2.4688377316114543,
+      "grad_norm": 0.13423927554603976,
+      "learning_rate": 2.580073869513094e-06,
+      "loss": 0.3916,
+      "step": 549
+    },
+    {
+      "epoch": 2.473329590117911,
+      "grad_norm": 0.16720772837034956,
+      "learning_rate": 2.5375219477356294e-06,
+      "loss": 0.4193,
+      "step": 550
+    },
+    {
+      "epoch": 2.4778214486243684,
+      "grad_norm": 0.1493859768063124,
+      "learning_rate": 2.4952892354294682e-06,
+      "loss": 0.4168,
+      "step": 551
+    },
+    {
+      "epoch": 2.482313307130825,
+      "grad_norm": 0.13772252373465832,
+      "learning_rate": 2.4533768942948534e-06,
+      "loss": 0.3986,
+      "step": 552
+    },
+    {
+      "epoch": 2.4868051656372825,
+      "grad_norm": 0.15653288394332063,
+      "learning_rate": 2.4117860772195375e-06,
+      "loss": 0.4066,
+      "step": 553
+    },
+    {
+      "epoch": 2.4912970241437393,
+      "grad_norm": 0.15949884029658656,
+      "learning_rate": 2.3705179282470687e-06,
+      "loss": 0.3951,
+      "step": 554
+    },
+    {
+      "epoch": 2.4957888826501966,
+      "grad_norm": 0.1429803919129004,
+      "learning_rate": 2.329573582545332e-06,
+      "loss": 0.4096,
+      "step": 555
+    },
+    {
+      "epoch": 2.5002807411566534,
+      "grad_norm": 0.1367359661730616,
+      "learning_rate": 2.288954166375312e-06,
+      "loss": 0.3814,
+      "step": 556
+    },
+    {
+      "epoch": 2.5047725996631107,
+      "grad_norm": 0.1513111966124078,
+      "learning_rate": 2.2486607970601196e-06,
+      "loss": 0.4286,
+      "step": 557
+    },
+    {
+      "epoch": 2.5092644581695676,
+      "grad_norm": 0.14045368969817978,
+      "learning_rate": 2.2086945829542548e-06,
+      "loss": 0.375,
+      "step": 558
+    },
+    {
+      "epoch": 2.513756316676025,
+      "grad_norm": 0.15528683121885756,
+      "learning_rate": 2.1690566234131268e-06,
+      "loss": 0.3948,
+      "step": 559
+    },
+    {
+      "epoch": 2.5182481751824817,
+      "grad_norm": 0.14551379178457585,
+      "learning_rate": 2.1297480087628017e-06,
+      "loss": 0.4195,
+      "step": 560
+    },
+    {
+      "epoch": 2.522740033688939,
+      "grad_norm": 0.14434811868307373,
+      "learning_rate": 2.090769820270016e-06,
+      "loss": 0.394,
+      "step": 561
+    },
+    {
+      "epoch": 2.527231892195396,
+      "grad_norm": 0.14183706063818433,
+      "learning_rate": 2.0521231301124356e-06,
+      "loss": 0.4015,
+      "step": 562
+    },
+    {
+      "epoch": 2.531723750701853,
+      "grad_norm": 0.13841495180500263,
+      "learning_rate": 2.013809001349163e-06,
+      "loss": 0.39,
+      "step": 563
+    },
+    {
+      "epoch": 2.53621560920831,
+      "grad_norm": 0.1405110514240103,
+      "learning_rate": 1.9758284878914944e-06,
+      "loss": 0.4132,
+      "step": 564
+    },
+    {
+      "epoch": 2.5407074677147667,
+      "grad_norm": 0.1582397797574471,
+      "learning_rate": 1.9381826344739317e-06,
+      "loss": 0.4211,
+      "step": 565
+    },
+    {
+      "epoch": 2.545199326221224,
+      "grad_norm": 0.14844762464251526,
+      "learning_rate": 1.9008724766254398e-06,
+      "loss": 0.4131,
+      "step": 566
+    },
+    {
+      "epoch": 2.5496911847276813,
+      "grad_norm": 0.13912350401808557,
+      "learning_rate": 1.8638990406409697e-06,
+      "loss": 0.3899,
+      "step": 567
+    },
+    {
+      "epoch": 2.554183043234138,
+      "grad_norm": 0.1348047186536698,
+      "learning_rate": 1.8272633435532205e-06,
+      "loss": 0.3861,
+      "step": 568
+    },
+    {
+      "epoch": 2.558674901740595,
+      "grad_norm": 0.1509435397791266,
+      "learning_rate": 1.7909663931046709e-06,
+      "loss": 0.4472,
+      "step": 569
+    },
+    {
+      "epoch": 2.5631667602470523,
+      "grad_norm": 0.14173089587097376,
+      "learning_rate": 1.7550091877198515e-06,
+      "loss": 0.3958,
+      "step": 570
+    },
+    {
+      "epoch": 2.5676586187535095,
+      "grad_norm": 0.14302822127986503,
+      "learning_rate": 1.719392716477887e-06,
+      "loss": 0.3979,
+      "step": 571
+    },
+    {
+      "epoch": 2.5721504772599664,
+      "grad_norm": 0.15922994001113902,
+      "learning_rate": 1.6841179590852887e-06,
+      "loss": 0.4047,
+      "step": 572
+    },
+    {
+      "epoch": 2.576642335766423,
+      "grad_norm": 0.12883007341555425,
+      "learning_rate": 1.6491858858490027e-06,
+      "loss": 0.3919,
+      "step": 573
+    },
+    {
+      "epoch": 2.5811341942728805,
+      "grad_norm": 0.1431322016396387,
+      "learning_rate": 1.614597457649723e-06,
+      "loss": 0.4361,
+      "step": 574
+    },
+    {
+      "epoch": 2.5856260527793373,
+      "grad_norm": 0.16005846399946025,
+      "learning_rate": 1.5803536259154552e-06,
+      "loss": 0.3871,
+      "step": 575
+    },
+    {
+      "epoch": 2.5901179112857946,
+      "grad_norm": 0.1438806473172459,
+      "learning_rate": 1.546455332595352e-06,
+      "loss": 0.4031,
+      "step": 576
+    },
+    {
+      "epoch": 2.5946097697922514,
+      "grad_norm": 0.14492270187306566,
+      "learning_rate": 1.5129035101338013e-06,
+      "loss": 0.41,
+      "step": 577
+    },
+    {
+      "epoch": 2.5991016282987087,
+      "grad_norm": 0.13992088460829535,
+      "learning_rate": 1.479699081444771e-06,
+      "loss": 0.3911,
+      "step": 578
+    },
+    {
+      "epoch": 2.6035934868051656,
+      "grad_norm": 0.15034483503496646,
+      "learning_rate": 1.44684295988643e-06,
+      "loss": 0.4091,
+      "step": 579
+    },
+    {
+      "epoch": 2.608085345311623,
+      "grad_norm": 0.14079803397340293,
+      "learning_rate": 1.4143360492360238e-06,
+      "loss": 0.4048,
+      "step": 580
+    },
+    {
+      "epoch": 2.6125772038180797,
+      "grad_norm": 0.1433534420740642,
+      "learning_rate": 1.3821792436650067e-06,
+      "loss": 0.4065,
+      "step": 581
+    },
+    {
+      "epoch": 2.6170690623245365,
+      "grad_norm": 0.14477932815556296,
+      "learning_rate": 1.350373427714457e-06,
+      "loss": 0.4163,
+      "step": 582
+    },
+    {
+      "epoch": 2.621560920830994,
+      "grad_norm": 0.12446547733516874,
+      "learning_rate": 1.3189194762707326e-06,
+      "loss": 0.3724,
+      "step": 583
+    },
+    {
+      "epoch": 2.626052779337451,
+      "grad_norm": 0.13398055360321232,
+      "learning_rate": 1.2878182545414191e-06,
+      "loss": 0.4149,
+      "step": 584
+    },
+    {
+      "epoch": 2.630544637843908,
+      "grad_norm": 0.13177864587555388,
+      "learning_rate": 1.2570706180315225e-06,
+      "loss": 0.3933,
+      "step": 585
+    },
+    {
+      "epoch": 2.6350364963503647,
+      "grad_norm": 0.1436166683524555,
+      "learning_rate": 1.2266774125199337e-06,
+      "loss": 0.4214,
+      "step": 586
+    },
+    {
+      "epoch": 2.639528354856822,
+      "grad_norm": 0.13383337224950775,
+      "learning_rate": 1.1966394740361719e-06,
+      "loss": 0.4042,
+      "step": 587
+    },
+    {
+      "epoch": 2.6440202133632793,
+      "grad_norm": 0.13908228590146968,
+      "learning_rate": 1.1669576288373846e-06,
+      "loss": 0.4081,
+      "step": 588
+    },
+    {
+      "epoch": 2.648512071869736,
+      "grad_norm": 0.13706113871707354,
+      "learning_rate": 1.1376326933856181e-06,
+      "loss": 0.3906,
+      "step": 589
+    },
+    {
+      "epoch": 2.653003930376193,
+      "grad_norm": 0.14637537593074662,
+      "learning_rate": 1.1086654743253583e-06,
+      "loss": 0.4137,
+      "step": 590
+    },
+    {
+      "epoch": 2.6574957888826503,
+      "grad_norm": 0.13803972231031073,
+      "learning_rate": 1.0800567684613445e-06,
+      "loss": 0.4087,
+      "step": 591
+    },
+    {
+      "epoch": 2.661987647389107,
+      "grad_norm": 0.13332822270628664,
+      "learning_rate": 1.0518073627366485e-06,
+      "loss": 0.4019,
+      "step": 592
+    },
+    {
+      "epoch": 2.6664795058955644,
+      "grad_norm": 0.13077083513715798,
+      "learning_rate": 1.0239180342110379e-06,
+      "loss": 0.3952,
+      "step": 593
+    },
+    {
+      "epoch": 2.670971364402021,
+      "grad_norm": 0.1336310148960128,
+      "learning_rate": 9.963895500395894e-07,
+      "loss": 0.4065,
+      "step": 594
+    },
+    {
+      "epoch": 2.6754632229084785,
+      "grad_norm": 0.1432123548055435,
+      "learning_rate": 9.692226674515869e-07,
+      "loss": 0.4102,
+      "step": 595
+    },
+    {
+      "epoch": 2.6799550814149353,
+      "grad_norm": 0.13782458019797403,
+      "learning_rate": 9.424181337297035e-07,
+      "loss": 0.402,
+      "step": 596
+    },
+    {
+      "epoch": 2.6844469399213926,
+      "grad_norm": 0.1407930193717778,
+      "learning_rate": 9.159766861894357e-07,
+      "loss": 0.4067,
+      "step": 597
+    },
+    {
+      "epoch": 2.6889387984278494,
+      "grad_norm": 0.1309505801207627,
+      "learning_rate": 8.89899052158824e-07,
+      "loss": 0.395,
+      "step": 598
+    },
+    {
+      "epoch": 2.6934306569343067,
+      "grad_norm": 0.13372502807779402,
+      "learning_rate": 8.641859489584498e-07,
+      "loss": 0.4073,
+      "step": 599
+    },
+    {
+      "epoch": 2.6979225154407636,
+      "grad_norm": 0.1341983283881159,
+      "learning_rate": 8.388380838817022e-07,
+      "loss": 0.387,
+      "step": 600
+    },
+    {
+      "epoch": 2.702414373947221,
+      "grad_norm": 0.13533167423280718,
+      "learning_rate": 8.138561541753177e-07,
+      "loss": 0.411,
+      "step": 601
+    },
+    {
+      "epoch": 2.7069062324536777,
+      "grad_norm": 0.12945926261780205,
+      "learning_rate": 7.892408470202099e-07,
+      "loss": 0.3968,
+      "step": 602
+    },
+    {
+      "epoch": 2.7113980909601345,
+      "grad_norm": 0.13778972221216923,
+      "learning_rate": 7.649928395125592e-07,
+      "loss": 0.4057,
+      "step": 603
+    },
+    {
+      "epoch": 2.715889949466592,
+      "grad_norm": 0.13968457328358516,
+      "learning_rate": 7.411127986451912e-07,
+      "loss": 0.4037,
+      "step": 604
+    },
+    {
+      "epoch": 2.720381807973049,
+      "grad_norm": 0.1397235533395217,
+      "learning_rate": 7.176013812892307e-07,
+      "loss": 0.4194,
+      "step": 605
+    },
+    {
+      "epoch": 2.724873666479506,
+      "grad_norm": 0.12677793388517908,
+      "learning_rate": 6.944592341760311e-07,
+      "loss": 0.3818,
+      "step": 606
+    },
+    {
+      "epoch": 2.7293655249859627,
+      "grad_norm": 0.13425231454213735,
+      "learning_rate": 6.716869938793857e-07,
+      "loss": 0.3984,
+      "step": 607
+    },
+    {
+      "epoch": 2.73385738349242,
+      "grad_norm": 0.12619942214267516,
+      "learning_rate": 6.492852867980155e-07,
+      "loss": 0.398,
+      "step": 608
+    },
+    {
+      "epoch": 2.7383492419988773,
+      "grad_norm": 0.13488281757393125,
+      "learning_rate": 6.272547291383403e-07,
+      "loss": 0.4086,
+      "step": 609
+    },
+    {
+      "epoch": 2.742841100505334,
+      "grad_norm": 0.1291152392505185,
+      "learning_rate": 6.055959268975332e-07,
+      "loss": 0.4031,
+      "step": 610
+    },
+    {
+      "epoch": 2.747332959011791,
+      "grad_norm": 0.13610320289234296,
+      "learning_rate": 5.843094758468389e-07,
+      "loss": 0.3861,
+      "step": 611
+    },
+    {
+      "epoch": 2.7518248175182483,
+      "grad_norm": 0.13227498539439772,
+      "learning_rate": 5.633959615152007e-07,
+      "loss": 0.4037,
+      "step": 612
+    },
+    {
+      "epoch": 2.756316676024705,
+      "grad_norm": 0.13186795971842452,
+      "learning_rate": 5.428559591731436e-07,
+      "loss": 0.3994,
+      "step": 613
+    },
+    {
+      "epoch": 2.7608085345311624,
+      "grad_norm": 0.1333721489625571,
+      "learning_rate": 5.226900338169552e-07,
+      "loss": 0.3973,
+      "step": 614
+    },
+    {
+      "epoch": 2.765300393037619,
+      "grad_norm": 0.1338646249689998,
+      "learning_rate": 5.028987401531437e-07,
+      "loss": 0.4056,
+      "step": 615
+    },
+    {
+      "epoch": 2.7697922515440765,
+      "grad_norm": 0.1382552063190408,
+      "learning_rate": 4.834826225831736e-07,
+      "loss": 0.3863,
+      "step": 616
+    },
+    {
+      "epoch": 2.7742841100505333,
+      "grad_norm": 0.1323165241089875,
+      "learning_rate": 4.644422151885038e-07,
+      "loss": 0.4194,
+      "step": 617
+    },
+    {
+      "epoch": 2.7787759685569906,
+      "grad_norm": 0.12553692047901485,
+      "learning_rate": 4.457780417158842e-07,
+      "loss": 0.4066,
+      "step": 618
+    },
+    {
+      "epoch": 2.7832678270634474,
+      "grad_norm": 0.12956844890475527,
+      "learning_rate": 4.274906155629513e-07,
+      "loss": 0.4004,
+      "step": 619
+    },
+    {
+      "epoch": 2.7877596855699043,
+      "grad_norm": 0.14538864839626559,
+      "learning_rate": 4.095804397641149e-07,
+      "loss": 0.3835,
+      "step": 620
+    },
+    {
+      "epoch": 2.7922515440763616,
+      "grad_norm": 0.13248803294494046,
+      "learning_rate": 3.9204800697670965e-07,
+      "loss": 0.4147,
+      "step": 621
+    },
+    {
+      "epoch": 2.796743402582819,
+      "grad_norm": 0.12885381135549048,
+      "learning_rate": 3.7489379946744694e-07,
+      "loss": 0.4056,
+      "step": 622
+    },
+    {
+      "epoch": 2.8012352610892757,
+      "grad_norm": 0.14078461225962682,
+      "learning_rate": 3.5811828909915534e-07,
+      "loss": 0.4149,
+      "step": 623
+    },
+    {
+      "epoch": 2.8057271195957325,
+      "grad_norm": 0.13097000904248593,
+      "learning_rate": 3.417219373177928e-07,
+      "loss": 0.3966,
+      "step": 624
+    },
+    {
+      "epoch": 2.81021897810219,
+      "grad_norm": 0.1376316131786541,
+      "learning_rate": 3.257051951397572e-07,
+      "loss": 0.3997,
+      "step": 625
+    },
+    {
+      "epoch": 2.814710836608647,
+      "grad_norm": 0.13657095261268032,
+      "learning_rate": 3.1006850313948107e-07,
+      "loss": 0.4033,
+      "step": 626
+    },
+    {
+      "epoch": 2.819202695115104,
+      "grad_norm": 0.13207939504527372,
+      "learning_rate": 2.9481229143731197e-07,
+      "loss": 0.3886,
+      "step": 627
+    },
+    {
+      "epoch": 2.8236945536215607,
+      "grad_norm": 0.13018989583008223,
+      "learning_rate": 2.7993697968767725e-07,
+      "loss": 0.4158,
+      "step": 628
+    },
+    {
+      "epoch": 2.828186412128018,
+      "grad_norm": 0.13040240548669169,
+      "learning_rate": 2.6544297706754683e-07,
+      "loss": 0.4068,
+      "step": 629
+    },
+    {
+      "epoch": 2.832678270634475,
+      "grad_norm": 0.13392224545793752,
+      "learning_rate": 2.5133068226517584e-07,
+      "loss": 0.398,
+      "step": 630
+    },
+    {
+      "epoch": 2.837170129140932,
+      "grad_norm": 0.1278432929244879,
+      "learning_rate": 2.3760048346913432e-07,
+      "loss": 0.4131,
+      "step": 631
+    },
+    {
+      "epoch": 2.841661987647389,
+      "grad_norm": 0.125465501573195,
+      "learning_rate": 2.2425275835763363e-07,
+      "loss": 0.3974,
+      "step": 632
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.13434683909908915,
+      "learning_rate": 2.1128787408813722e-07,
+      "loss": 0.4061,
+      "step": 633
+    },
+    {
+      "epoch": 2.850645704660303,
+      "grad_norm": 0.12917014219042902,
+      "learning_rate": 1.9870618728725558e-07,
+      "loss": 0.4119,
+      "step": 634
+    },
+    {
+      "epoch": 2.8551375631667604,
+      "grad_norm": 0.12927799197589832,
+      "learning_rate": 1.8650804404094718e-07,
+      "loss": 0.4165,
+      "step": 635
+    },
+    {
+      "epoch": 2.859629421673217,
+      "grad_norm": 0.12828667749491696,
+      "learning_rate": 1.7469377988498657e-07,
+      "loss": 0.3936,
+      "step": 636
+    },
+    {
+      "epoch": 2.8641212801796745,
+      "grad_norm": 0.13007544913143448,
+      "learning_rate": 1.632637197957446e-07,
+      "loss": 0.3946,
+      "step": 637
+    },
+    {
+      "epoch": 2.8686131386861313,
+      "grad_norm": 0.1354126849121975,
+      "learning_rate": 1.522181781812407e-07,
+      "loss": 0.4082,
+      "step": 638
+    },
+    {
+      "epoch": 2.8731049971925886,
+      "grad_norm": 0.13152253429926178,
+      "learning_rate": 1.415574588725056e-07,
+      "loss": 0.4165,
+      "step": 639
+    },
+    {
+      "epoch": 2.8775968556990454,
+      "grad_norm": 0.12461687039628878,
+      "learning_rate": 1.3128185511520974e-07,
+      "loss": 0.3951,
+      "step": 640
+    },
+    {
+      "epoch": 2.8820887142055023,
+      "grad_norm": 0.12537645712273682,
+      "learning_rate": 1.2139164956160858e-07,
+      "loss": 0.3838,
+      "step": 641
+    },
+    {
+      "epoch": 2.8865805727119596,
+      "grad_norm": 0.1304530973255801,
+      "learning_rate": 1.1188711426276294e-07,
+      "loss": 0.4242,
+      "step": 642
+    },
+    {
+      "epoch": 2.891072431218417,
+      "grad_norm": 0.1270688998552854,
+      "learning_rate": 1.0276851066105732e-07,
+      "loss": 0.3999,
+      "step": 643
+    },
+    {
+      "epoch": 2.8955642897248737,
+      "grad_norm": 0.12479392286448229,
+      "learning_rate": 9.403608958300276e-08,
+      "loss": 0.3986,
+      "step": 644
+    },
+    {
+      "epoch": 2.9000561482313305,
+      "grad_norm": 0.12433792246453539,
+      "learning_rate": 8.569009123234562e-08,
+      "loss": 0.4052,
+      "step": 645
+    },
+    {
+      "epoch": 2.904548006737788,
+      "grad_norm": 0.13028761337968897,
+      "learning_rate": 7.773074518345438e-08,
+      "loss": 0.4053,
+      "step": 646
+    },
+    {
+      "epoch": 2.909039865244245,
+      "grad_norm": 0.1245947432502766,
+      "learning_rate": 7.015827037500785e-08,
+      "loss": 0.393,
+      "step": 647
+    },
+    {
+      "epoch": 2.913531723750702,
+      "grad_norm": 0.13031905992918363,
+      "learning_rate": 6.297287510397075e-08,
+      "loss": 0.4043,
+      "step": 648
+    },
+    {
+      "epoch": 2.9180235822571587,
+      "grad_norm": 0.12590059661457523,
+      "learning_rate": 5.6174757019866e-08,
+      "loss": 0.4062,
+      "step": 649
+    },
+    {
+      "epoch": 2.922515440763616,
+      "grad_norm": 0.13389326831674256,
+      "learning_rate": 4.976410311933432e-08,
+      "loss": 0.4024,
+      "step": 650
+    },
+    {
+      "epoch": 2.927007299270073,
+      "grad_norm": 0.1250548459939687,
+      "learning_rate": 4.3741089740998524e-08,
+      "loss": 0.4023,
+      "step": 651
+    },
+    {
+      "epoch": 2.93149915777653,
+      "grad_norm": 0.12270077200805969,
+      "learning_rate": 3.810588256060089e-08,
+      "loss": 0.3967,
+      "step": 652
+    },
+    {
+      "epoch": 2.935991016282987,
+      "grad_norm": 0.12624611049254067,
+      "learning_rate": 3.2858636586456185e-08,
+      "loss": 0.4006,
+      "step": 653
+    },
+    {
+      "epoch": 2.9404828747894443,
+      "grad_norm": 0.1306767931158808,
+      "learning_rate": 2.7999496155184166e-08,
+      "loss": 0.39,
+      "step": 654
+    },
+    {
+      "epoch": 2.944974733295901,
+      "grad_norm": 0.12512189399104776,
+      "learning_rate": 2.3528594927737242e-08,
+      "loss": 0.3861,
+      "step": 655
+    },
+    {
+      "epoch": 2.9494665918023584,
+      "grad_norm": 0.1331701408947349,
+      "learning_rate": 1.9446055885728008e-08,
+      "loss": 0.4219,
+      "step": 656
+    },
+    {
+      "epoch": 2.953958450308815,
+      "grad_norm": 0.1269883743014457,
+      "learning_rate": 1.5751991328044162e-08,
+      "loss": 0.3958,
+      "step": 657
+    },
+    {
+      "epoch": 2.958450308815272,
+      "grad_norm": 0.12769990540091625,
+      "learning_rate": 1.2446502867756959e-08,
+      "loss": 0.4025,
+      "step": 658
+    },
+    {
+      "epoch": 2.9629421673217293,
+      "grad_norm": 0.13209320420166154,
+      "learning_rate": 9.529681429332745e-09,
+      "loss": 0.4144,
+      "step": 659
+    },
+    {
+      "epoch": 2.9674340258281866,
+      "grad_norm": 0.13117251039825162,
+      "learning_rate": 7.0016072461239484e-09,
+      "loss": 0.3971,
+      "step": 660
+    },
+    {
+      "epoch": 2.9719258843346434,
+      "grad_norm": 0.12741148550926698,
+      "learning_rate": 4.862349858167804e-09,
+      "loss": 0.4035,
+      "step": 661
+    },
+    {
+      "epoch": 2.9764177428411003,
+      "grad_norm": 0.13192324038184064,
+      "learning_rate": 3.111968110273985e-09,
+      "loss": 0.4031,
+      "step": 662
+    },
+    {
+      "epoch": 2.9809096013475576,
+      "grad_norm": 0.13150035056861856,
+      "learning_rate": 1.7505101503979872e-09,
+      "loss": 0.4044,
+      "step": 663
+    },
+    {
+      "epoch": 2.985401459854015,
+      "grad_norm": 0.12410294041882683,
+      "learning_rate": 7.780134283285303e-10,
+      "loss": 0.3951,
+      "step": 664
+    },
+    {
+      "epoch": 2.9898933183604717,
+      "grad_norm": 0.12559324324833065,
+      "learning_rate": 1.945046946481569e-10,
+      "loss": 0.3995,
+      "step": 665
+    },
+    {
+      "epoch": 2.9943851768669285,
+      "grad_norm": 0.13271897553792253,
+      "learning_rate": 0.0,
+      "loss": 0.4058,
+      "step": 666
+    },
+    {
+      "epoch": 2.9943851768669285,
+      "step": 666,
+      "total_flos": 7.06867120956244e+18,
+      "train_loss": 0.4525994020360368,
+      "train_runtime": 109632.7326,
+      "train_samples_per_second": 3.118,
+      "train_steps_per_second": 0.006
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 666,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.06867120956244e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}