diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4767 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 675,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004437049362174155,
+      "grad_norm": 3.4363257049624876,
+      "learning_rate": 0.0,
+      "loss": 0.6212,
+      "step": 1
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 3.4363257049624876,
+      "learning_rate": 0.0,
+      "loss": 1.447,
+      "step": 2
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 3.4363257049624876,
+      "learning_rate": 0.0,
+      "loss": 1.4008,
+      "step": 3
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 16.868642103790727,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 1.4622,
+      "step": 4
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 16.643551679254042,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 1.449,
+      "step": 5
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 10.319968368718369,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 1.2176,
+      "step": 6
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 6.838063285541478,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.1671,
+      "step": 7
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 3.524165786619175,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 0.8897,
+      "step": 8
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 2.692583820309877,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 0.7701,
+      "step": 9
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 2.1523694488597362,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 0.6826,
+      "step": 10
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 2.065704399579686,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.6885,
+      "step": 11
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 1.8642136901781687,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 0.6381,
+      "step": 12
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 1.9903385333507209,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 0.6594,
+      "step": 13
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 1.5186915884833811,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 0.6485,
+      "step": 14
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 1.7115993526016227,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.6417,
+      "step": 15
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 1.7099641443698512,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 0.6591,
+      "step": 16
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 1.850613121296057,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 0.639,
+      "step": 17
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 1.7329331324202384,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 0.5688,
+      "step": 18
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 1.7006038939523738,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.7069,
+      "step": 19
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 1.7294014767466337,
+      "learning_rate": 1.25e-05,
+      "loss": 0.6022,
+      "step": 20
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 1.6255794372391963,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 0.5901,
+      "step": 21
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 1.6321097143730798,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 0.6678,
+      "step": 22
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 2.412341050362017,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.6245,
+      "step": 23
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 1.6139308554481517,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 0.6382,
+      "step": 24
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 1.6049208251343587,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.7061,
+      "step": 25
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 1.5780465890270454,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 0.7123,
+      "step": 26
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 1.696241273794309,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.7704,
+      "step": 27
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 1.5906252428215757,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 0.5403,
+      "step": 28
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 1.5414524788746833,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 0.5979,
+      "step": 29
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 1.596561047173864,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 0.5242,
+      "step": 30
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 1.7145694174342805,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 0.6417,
+      "step": 31
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 1.433522900901715,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 0.6031,
+      "step": 32
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 1.4685539764106288,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 0.6575,
+      "step": 33
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 1.4483896850463052,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 0.5516,
+      "step": 34
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 1.7283584437052888,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.644,
+      "step": 35
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 1.5105552434381186,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 0.5804,
+      "step": 36
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 1.5464090743766123,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5736,
+      "step": 37
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 1.5477842490785196,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 0.7317,
+      "step": 38
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 1.8091612351926254,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.6621,
+      "step": 39
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 1.785226632698615,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 0.6575,
+      "step": 40
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 1.4735924454223048,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 0.6695,
+      "step": 41
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 2.236068403995869,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 0.6331,
+      "step": 42
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 2.2307337187664844,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.6197,
+      "step": 43
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 1.298781361132413,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 0.652,
+      "step": 44
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 1.707441058167839,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 0.664,
+      "step": 45
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 1.7003456325349169,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 0.7874,
+      "step": 46
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 1.3961934673643377,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 0.7497,
+      "step": 47
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 1.946072288342776,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 0.7855,
+      "step": 48
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 1.5449525851779393,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 0.6254,
+      "step": 49
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 1.6317975115632137,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 0.7916,
+      "step": 50
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 1.4264993609758905,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.6275,
+      "step": 51
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 1.872068656876876,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 0.6932,
+      "step": 52
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 1.5201820999787976,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 0.6139,
+      "step": 53
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 1.3257820106913285,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.6663,
+      "step": 54
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 1.4712901773857452,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 0.7818,
+      "step": 55
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 1.6491941189067307,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 0.5963,
+      "step": 56
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 1.6066731114607802,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 0.7589,
+      "step": 57
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 1.5803395963838538,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 0.6909,
+      "step": 58
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 1.8348490633721526,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.6927,
+      "step": 59
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 1.7772939821758158,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 0.8725,
+      "step": 60
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 1.570535065170522,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 0.752,
+      "step": 61
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 1.4213107320379403,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 0.664,
+      "step": 62
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 1.4010270932182938,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.6503,
+      "step": 63
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 1.559712020751588,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 0.6395,
+      "step": 64
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 1.398449689929195,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 0.7274,
+      "step": 65
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 1.5012715830907073,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 0.6339,
+      "step": 66
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 1.3953862067188367,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.7273,
+      "step": 67
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 1.7813022003555992,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 0.6098,
+      "step": 68
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 1.6629697408532516,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 0.7984,
+      "step": 69
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 1.5104460965501205,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 0.7231,
+      "step": 70
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 1.601897860515358,
+      "learning_rate": 5e-05,
+      "loss": 0.6644,
+      "step": 71
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 1.4486607713844175,
+      "learning_rate": 4.99176276771005e-05,
+      "loss": 0.7003,
+      "step": 72
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 1.493309438051621,
+      "learning_rate": 4.9835255354200987e-05,
+      "loss": 0.6949,
+      "step": 73
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 1.361584063296398,
+      "learning_rate": 4.975288303130148e-05,
+      "loss": 0.6169,
+      "step": 74
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 1.458929630258479,
+      "learning_rate": 4.967051070840198e-05,
+      "loss": 0.732,
+      "step": 75
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 1.8222299717464396,
+      "learning_rate": 4.958813838550247e-05,
+      "loss": 0.7367,
+      "step": 76
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 1.5631573629403719,
+      "learning_rate": 4.950576606260297e-05,
+      "loss": 0.7863,
+      "step": 77
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 1.5057181404666258,
+      "learning_rate": 4.9423393739703464e-05,
+      "loss": 0.784,
+      "step": 78
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 1.5000204243858912,
+      "learning_rate": 4.934102141680396e-05,
+      "loss": 0.7517,
+      "step": 79
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 1.3679094424808862,
+      "learning_rate": 4.9258649093904455e-05,
+      "loss": 0.5966,
+      "step": 80
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 1.509099539505429,
+      "learning_rate": 4.9176276771004943e-05,
+      "loss": 0.7955,
+      "step": 81
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 1.2782656174885776,
+      "learning_rate": 4.909390444810544e-05,
+      "loss": 0.6774,
+      "step": 82
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 1.4750850491330671,
+      "learning_rate": 4.9011532125205934e-05,
+      "loss": 0.6925,
+      "step": 83
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 1.4230772531949174,
+      "learning_rate": 4.892915980230643e-05,
+      "loss": 0.7124,
+      "step": 84
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 1.3933968511428394,
+      "learning_rate": 4.884678747940692e-05,
+      "loss": 0.6607,
+      "step": 85
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 1.4883118473622319,
+      "learning_rate": 4.8764415156507414e-05,
+      "loss": 0.7622,
+      "step": 86
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 1.4552755541921092,
+      "learning_rate": 4.868204283360791e-05,
+      "loss": 0.8021,
+      "step": 87
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 1.5206313674109841,
+      "learning_rate": 4.8599670510708405e-05,
+      "loss": 0.7699,
+      "step": 88
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 1.3870986632973126,
+      "learning_rate": 4.8517298187808894e-05,
+      "loss": 0.7337,
+      "step": 89
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 1.2888477493348762,
+      "learning_rate": 4.843492586490939e-05,
+      "loss": 0.7701,
+      "step": 90
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 1.3593434735731116,
+      "learning_rate": 4.8352553542009884e-05,
+      "loss": 0.7312,
+      "step": 91
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 1.4612922441757445,
+      "learning_rate": 4.827018121911038e-05,
+      "loss": 0.723,
+      "step": 92
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 1.2774829380366661,
+      "learning_rate": 4.8187808896210875e-05,
+      "loss": 0.6934,
+      "step": 93
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 1.6465782781320284,
+      "learning_rate": 4.810543657331137e-05,
+      "loss": 0.7358,
+      "step": 94
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 1.6370311793933272,
+      "learning_rate": 4.8023064250411866e-05,
+      "loss": 0.8459,
+      "step": 95
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 1.371967960762082,
+      "learning_rate": 4.794069192751236e-05,
+      "loss": 0.6993,
+      "step": 96
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 1.2628676908270764,
+      "learning_rate": 4.785831960461285e-05,
+      "loss": 0.6787,
+      "step": 97
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 1.4005074977653627,
+      "learning_rate": 4.7775947281713346e-05,
+      "loss": 0.8011,
+      "step": 98
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 1.2543277210397603,
+      "learning_rate": 4.769357495881384e-05,
+      "loss": 0.7595,
+      "step": 99
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 1.1805964120015806,
+      "learning_rate": 4.761120263591434e-05,
+      "loss": 0.7692,
+      "step": 100
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 1.2276416089327171,
+      "learning_rate": 4.7528830313014825e-05,
+      "loss": 0.6908,
+      "step": 101
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 1.3507566309682646,
+      "learning_rate": 4.744645799011532e-05,
+      "loss": 0.8863,
+      "step": 102
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 1.4097027247458103,
+      "learning_rate": 4.7364085667215816e-05,
+      "loss": 0.789,
+      "step": 103
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 1.1723423343870047,
+      "learning_rate": 4.728171334431631e-05,
+      "loss": 0.7641,
+      "step": 104
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 1.1380218692079394,
+      "learning_rate": 4.719934102141681e-05,
+      "loss": 0.6747,
+      "step": 105
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 1.1778272399222929,
+      "learning_rate": 4.71169686985173e-05,
+      "loss": 0.727,
+      "step": 106
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 1.4296109538614465,
+      "learning_rate": 4.70345963756178e-05,
+      "loss": 0.766,
+      "step": 107
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 1.5045893715366057,
+      "learning_rate": 4.6952224052718294e-05,
+      "loss": 0.7547,
+      "step": 108
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 1.2423295713548035,
+      "learning_rate": 4.686985172981878e-05,
+      "loss": 0.7361,
+      "step": 109
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 1.2453380433721188,
+      "learning_rate": 4.678747940691928e-05,
+      "loss": 0.693,
+      "step": 110
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 1.1167979061286064,
+      "learning_rate": 4.670510708401977e-05,
+      "loss": 0.7904,
+      "step": 111
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 1.1003429073432243,
+      "learning_rate": 4.662273476112027e-05,
+      "loss": 0.8332,
+      "step": 112
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 1.479955257048743,
+      "learning_rate": 4.654036243822076e-05,
+      "loss": 0.7635,
+      "step": 113
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 1.1210095310809869,
+      "learning_rate": 4.645799011532125e-05,
+      "loss": 0.8422,
+      "step": 114
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 1.1554116999100634,
+      "learning_rate": 4.637561779242175e-05,
+      "loss": 0.777,
+      "step": 115
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 1.3394475857232722,
+      "learning_rate": 4.6293245469522244e-05,
+      "loss": 0.7022,
+      "step": 116
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 1.4296884171946767,
+      "learning_rate": 4.621087314662273e-05,
+      "loss": 0.7253,
+      "step": 117
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 1.6262478072366437,
+      "learning_rate": 4.612850082372323e-05,
+      "loss": 0.688,
+      "step": 118
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 1.3776166899700177,
+      "learning_rate": 4.6046128500823723e-05,
+      "loss": 0.7821,
+      "step": 119
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 1.1789329591510056,
+      "learning_rate": 4.596375617792422e-05,
+      "loss": 0.6633,
+      "step": 120
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 1.3628007976726708,
+      "learning_rate": 4.5881383855024714e-05,
+      "loss": 0.7478,
+      "step": 121
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 1.2320996813979046,
+      "learning_rate": 4.579901153212521e-05,
+      "loss": 0.6536,
+      "step": 122
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 1.3977794270959722,
+      "learning_rate": 4.5716639209225705e-05,
+      "loss": 0.7028,
+      "step": 123
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 1.0468264326530035,
+      "learning_rate": 4.56342668863262e-05,
+      "loss": 0.6335,
+      "step": 124
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 1.113360378732564,
+      "learning_rate": 4.555189456342669e-05,
+      "loss": 0.6417,
+      "step": 125
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 1.3100109567984795,
+      "learning_rate": 4.5469522240527185e-05,
+      "loss": 0.7237,
+      "step": 126
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 1.2716899184171513,
+      "learning_rate": 4.538714991762768e-05,
+      "loss": 0.8029,
+      "step": 127
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 1.1535260875316555,
+      "learning_rate": 4.5304777594728176e-05,
+      "loss": 0.6549,
+      "step": 128
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 1.3185370883549747,
+      "learning_rate": 4.5222405271828664e-05,
+      "loss": 0.7478,
+      "step": 129
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 1.0101101138265811,
+      "learning_rate": 4.514003294892916e-05,
+      "loss": 0.7266,
+      "step": 130
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 1.2160341863689144,
+      "learning_rate": 4.5057660626029655e-05,
+      "loss": 0.7094,
+      "step": 131
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 1.28669160074244,
+      "learning_rate": 4.497528830313015e-05,
+      "loss": 0.6842,
+      "step": 132
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 1.2305138473723627,
+      "learning_rate": 4.4892915980230646e-05,
+      "loss": 0.7785,
+      "step": 133
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 1.1839778264392269,
+      "learning_rate": 4.481054365733114e-05,
+      "loss": 0.7821,
+      "step": 134
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 1.13330078125,
+      "learning_rate": 4.472817133443164e-05,
+      "loss": 0.8602,
+      "step": 135
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 1.3982152922135496,
+      "learning_rate": 4.464579901153213e-05,
+      "loss": 0.7126,
+      "step": 136
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 1.2006974935017933,
+      "learning_rate": 4.456342668863262e-05,
+      "loss": 0.8415,
+      "step": 137
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 0.9033962898263505,
+      "learning_rate": 4.448105436573312e-05,
+      "loss": 0.6684,
+      "step": 138
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 1.104182597111398,
+      "learning_rate": 4.439868204283361e-05,
+      "loss": 0.728,
+      "step": 139
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 0.8660290859416853,
+      "learning_rate": 4.43163097199341e-05,
+      "loss": 0.6536,
+      "step": 140
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 1.2196199295675134,
+      "learning_rate": 4.4233937397034596e-05,
+      "loss": 0.6733,
+      "step": 141
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 0.9899389063162056,
+      "learning_rate": 4.415156507413509e-05,
+      "loss": 0.5878,
+      "step": 142
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 1.0461409543179492,
+      "learning_rate": 4.406919275123559e-05,
+      "loss": 0.6893,
+      "step": 143
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 1.6991287382993012,
+      "learning_rate": 4.3986820428336076e-05,
+      "loss": 0.7363,
+      "step": 144
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 1.8572584014098694,
+      "learning_rate": 4.390444810543657e-05,
+      "loss": 0.7767,
+      "step": 145
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 1.1577735348630036,
+      "learning_rate": 4.382207578253707e-05,
+      "loss": 0.8126,
+      "step": 146
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 1.1352947006279315,
+      "learning_rate": 4.373970345963756e-05,
+      "loss": 0.7144,
+      "step": 147
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 1.3015549566776177,
+      "learning_rate": 4.365733113673806e-05,
+      "loss": 0.7324,
+      "step": 148
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 1.094701516565074,
+      "learning_rate": 4.357495881383855e-05,
+      "loss": 0.6737,
+      "step": 149
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 1.2602135150382683,
+      "learning_rate": 4.349258649093905e-05,
+      "loss": 0.8867,
+      "step": 150
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 1.2826388205488253,
+      "learning_rate": 4.3410214168039544e-05,
+      "loss": 0.7138,
+      "step": 151
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 0.914123989547532,
+      "learning_rate": 4.332784184514003e-05,
+      "loss": 0.6167,
+      "step": 152
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 1.0897851196706918,
+      "learning_rate": 4.324546952224053e-05,
+      "loss": 0.6871,
+      "step": 153
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 1.4113414148232353,
+      "learning_rate": 4.3163097199341024e-05,
+      "loss": 0.8047,
+      "step": 154
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 1.3141605227239495,
+      "learning_rate": 4.308072487644152e-05,
+      "loss": 0.6614,
+      "step": 155
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 1.0685581949797351,
+      "learning_rate": 4.299835255354201e-05,
+      "loss": 0.7868,
+      "step": 156
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 1.0776774818305639,
+      "learning_rate": 4.29159802306425e-05,
+      "loss": 0.7193,
+      "step": 157
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 0.9757258073435284,
+      "learning_rate": 4.2833607907743e-05,
+      "loss": 0.6609,
+      "step": 158
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 1.091813252826064,
+      "learning_rate": 4.2751235584843494e-05,
+      "loss": 0.5548,
+      "step": 159
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 1.2464545992047698,
+      "learning_rate": 4.266886326194399e-05,
+      "loss": 0.8032,
+      "step": 160
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 1.148898492789776,
+      "learning_rate": 4.2586490939044485e-05,
+      "loss": 0.7182,
+      "step": 161
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 1.3851632003480712,
+      "learning_rate": 4.250411861614498e-05,
+      "loss": 0.7686,
+      "step": 162
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 1.072810467152235,
+      "learning_rate": 4.2421746293245476e-05,
+      "loss": 0.6567,
+      "step": 163
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 1.0450828356970694,
+      "learning_rate": 4.2339373970345965e-05,
+      "loss": 0.7796,
+      "step": 164
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 1.5079605741380997,
+      "learning_rate": 4.225700164744646e-05,
+      "loss": 0.6934,
+      "step": 165
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 1.104207535953022,
+      "learning_rate": 4.2174629324546956e-05,
+      "loss": 0.7709,
+      "step": 166
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 1.2299912752446822,
+      "learning_rate": 4.209225700164745e-05,
+      "loss": 0.6813,
+      "step": 167
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 1.2594723855505108,
+      "learning_rate": 4.200988467874794e-05,
+      "loss": 0.7787,
+      "step": 168
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 1.0866567850443722,
+      "learning_rate": 4.1927512355848435e-05,
+      "loss": 0.7779,
+      "step": 169
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 1.2387755457364094,
+      "learning_rate": 4.184514003294893e-05,
+      "loss": 0.8369,
+      "step": 170
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 1.2877205076548874,
+      "learning_rate": 4.1762767710049426e-05,
+      "loss": 0.6673,
+      "step": 171
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 1.141038323333023,
+      "learning_rate": 4.1680395387149915e-05,
+      "loss": 0.6762,
+      "step": 172
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 1.1145196819253242,
+      "learning_rate": 4.159802306425041e-05,
+      "loss": 0.7632,
+      "step": 173
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 1.0668930786960935,
+      "learning_rate": 4.1515650741350906e-05,
+      "loss": 0.6362,
+      "step": 174
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 1.2744678939031877,
+      "learning_rate": 4.14332784184514e-05,
+      "loss": 0.7398,
+      "step": 175
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 1.149177416713596,
+      "learning_rate": 4.13509060955519e-05,
+      "loss": 0.6689,
+      "step": 176
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 1.1000077659159404,
+      "learning_rate": 4.126853377265239e-05,
+      "loss": 0.7335,
+      "step": 177
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 1.2669367644465137,
+      "learning_rate": 4.118616144975289e-05,
+      "loss": 0.8656,
+      "step": 178
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 1.1232511915026113,
+      "learning_rate": 4.110378912685338e-05,
+      "loss": 0.7366,
+      "step": 179
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 1.1200223581602873,
+      "learning_rate": 4.102141680395387e-05,
+      "loss": 0.7661,
+      "step": 180
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 1.0756192043044124,
+      "learning_rate": 4.093904448105437e-05,
+      "loss": 0.6768,
+      "step": 181
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 1.2291643282765705,
+      "learning_rate": 4.085667215815486e-05,
+      "loss": 0.827,
+      "step": 182
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 1.2924700463843568,
+      "learning_rate": 4.077429983525536e-05,
+      "loss": 0.751,
+      "step": 183
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 1.1406485150147252,
+      "learning_rate": 4.069192751235585e-05,
+      "loss": 0.8197,
+      "step": 184
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 1.6746923648103702,
+      "learning_rate": 4.060955518945634e-05,
+      "loss": 0.8636,
+      "step": 185
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 1.0226322529704257,
+      "learning_rate": 4.052718286655684e-05,
+      "loss": 0.7296,
+      "step": 186
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 1.2317174479211401,
+      "learning_rate": 4.044481054365733e-05,
+      "loss": 0.7331,
+      "step": 187
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 1.1765805692665705,
+      "learning_rate": 4.036243822075783e-05,
+      "loss": 0.7811,
+      "step": 188
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 1.1237666999331317,
+      "learning_rate": 4.0280065897858324e-05,
+      "loss": 0.8865,
+      "step": 189
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 1.0807313237302771,
+      "learning_rate": 4.019769357495882e-05,
+      "loss": 0.7731,
+      "step": 190
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 1.149091469506126,
+      "learning_rate": 4.0115321252059315e-05,
+      "loss": 0.6796,
+      "step": 191
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 1.5166262436115496,
+      "learning_rate": 4.0032948929159804e-05,
+      "loss": 0.6694,
+      "step": 192
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 1.3121738482730616,
+      "learning_rate": 3.99505766062603e-05,
+      "loss": 0.6903,
+      "step": 193
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 1.0641393356477866,
+      "learning_rate": 3.9868204283360795e-05,
+      "loss": 0.8431,
+      "step": 194
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 1.2395748277046463,
+      "learning_rate": 3.978583196046129e-05,
+      "loss": 0.7475,
+      "step": 195
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 1.1301999187009746,
+      "learning_rate": 3.970345963756178e-05,
+      "loss": 0.7522,
+      "step": 196
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 0.9439853147479779,
+      "learning_rate": 3.9621087314662274e-05,
+      "loss": 0.7204,
+      "step": 197
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 1.0228506837425515,
+      "learning_rate": 3.953871499176277e-05,
+      "loss": 0.6889,
+      "step": 198
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 1.1469914681234548,
+      "learning_rate": 3.9456342668863265e-05,
+      "loss": 0.791,
+      "step": 199
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 1.0061181781739899,
+      "learning_rate": 3.9373970345963754e-05,
+      "loss": 0.8283,
+      "step": 200
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 1.0773855797362921,
+      "learning_rate": 3.929159802306425e-05,
+      "loss": 0.6699,
+      "step": 201
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 1.0711644255778545,
+      "learning_rate": 3.9209225700164745e-05,
+      "loss": 0.7143,
+      "step": 202
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 0.9061474248779531,
+      "learning_rate": 3.912685337726524e-05,
+      "loss": 0.5905,
+      "step": 203
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 1.1783259536605601,
+      "learning_rate": 3.9044481054365736e-05,
+      "loss": 0.8541,
+      "step": 204
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 0.9644733598331529,
+      "learning_rate": 3.896210873146623e-05,
+      "loss": 0.8005,
+      "step": 205
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 1.1284868457857407,
+      "learning_rate": 3.8879736408566727e-05,
+      "loss": 0.7008,
+      "step": 206
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 1.1875985757168446,
+      "learning_rate": 3.8797364085667215e-05,
+      "loss": 0.6928,
+      "step": 207
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 1.0014509165672858,
+      "learning_rate": 3.871499176276771e-05,
+      "loss": 0.6631,
+      "step": 208
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 1.1995148929913273,
+      "learning_rate": 3.8632619439868206e-05,
+      "loss": 0.75,
+      "step": 209
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 1.0930321653891049,
+      "learning_rate": 3.85502471169687e-05,
+      "loss": 0.7021,
+      "step": 210
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 1.035845142234787,
+      "learning_rate": 3.846787479406919e-05,
+      "loss": 0.7832,
+      "step": 211
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 1.1155766965522822,
+      "learning_rate": 3.8385502471169686e-05,
+      "loss": 0.8157,
+      "step": 212
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 1.161790307104464,
+      "learning_rate": 3.830313014827018e-05,
+      "loss": 0.8157,
+      "step": 213
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 1.1603177661401252,
+      "learning_rate": 3.822075782537068e-05,
+      "loss": 0.7666,
+      "step": 214
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 1.0978324450063657,
+      "learning_rate": 3.813838550247117e-05,
+      "loss": 0.6614,
+      "step": 215
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 1.0604673464692589,
+      "learning_rate": 3.805601317957167e-05,
+      "loss": 0.779,
+      "step": 216
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 1.1550506351155287,
+      "learning_rate": 3.797364085667216e-05,
+      "loss": 0.8661,
+      "step": 217
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 1.0968998919757251,
+      "learning_rate": 3.789126853377266e-05,
+      "loss": 0.7193,
+      "step": 218
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 1.0406620649088993,
+      "learning_rate": 3.780889621087315e-05,
+      "loss": 0.648,
+      "step": 219
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 1.0409014497880078,
+      "learning_rate": 3.772652388797364e-05,
+      "loss": 0.8307,
+      "step": 220
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 1.0170270778648376,
+      "learning_rate": 3.764415156507414e-05,
+      "loss": 0.683,
+      "step": 221
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 1.0712048228829765,
+      "learning_rate": 3.7561779242174634e-05,
+      "loss": 0.7853,
+      "step": 222
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 1.0226822024346145,
+      "learning_rate": 3.747940691927512e-05,
+      "loss": 0.7272,
+      "step": 223
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 1.04540416959913,
+      "learning_rate": 3.739703459637562e-05,
+      "loss": 0.6587,
+      "step": 224
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 1.4481215938022614,
+      "learning_rate": 3.731466227347611e-05,
+      "loss": 0.8142,
+      "step": 225
+    },
+    {
+      "epoch": 1.0027731558513588,
+      "grad_norm": 0.9916722919935,
+      "learning_rate": 3.723228995057661e-05,
+      "loss": 0.6207,
+      "step": 226
+    },
+    {
+      "epoch": 1.007210205213533,
+      "grad_norm": 0.7317840240122286,
+      "learning_rate": 3.71499176276771e-05,
+      "loss": 0.3225,
+      "step": 227
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 0.7507249983753904,
+      "learning_rate": 3.706754530477759e-05,
+      "loss": 0.3261,
+      "step": 228
+    },
+    {
+      "epoch": 1.0160843039378813,
+      "grad_norm": 0.8379545387487033,
+      "learning_rate": 3.698517298187809e-05,
+      "loss": 0.2845,
+      "step": 229
+    },
+    {
+      "epoch": 1.0205213533000554,
+      "grad_norm": 0.8473111342241626,
+      "learning_rate": 3.6902800658978584e-05,
+      "loss": 0.2527,
+      "step": 230
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 1.6604069778497936,
+      "learning_rate": 3.682042833607908e-05,
+      "loss": 0.3043,
+      "step": 231
+    },
+    {
+      "epoch": 1.0293954520244037,
+      "grad_norm": 0.9140945820191196,
+      "learning_rate": 3.6738056013179575e-05,
+      "loss": 0.2774,
+      "step": 232
+    },
+    {
+      "epoch": 1.033832501386578,
+      "grad_norm": 1.0852895216811833,
+      "learning_rate": 3.665568369028007e-05,
+      "loss": 0.3232,
+      "step": 233
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 1.0861523093312244,
+      "learning_rate": 3.6573311367380566e-05,
+      "loss": 0.3223,
+      "step": 234
+    },
+    {
+      "epoch": 1.0427066001109262,
+      "grad_norm": 0.9068528505087753,
+      "learning_rate": 3.6490939044481054e-05,
+      "loss": 0.2583,
+      "step": 235
+    },
+    {
+      "epoch": 1.0471436494731003,
+      "grad_norm": 1.1612832089901859,
+      "learning_rate": 3.640856672158155e-05,
+      "loss": 0.3444,
+      "step": 236
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 0.9997441441335119,
+      "learning_rate": 3.6326194398682045e-05,
+      "loss": 0.3421,
+      "step": 237
+    },
+    {
+      "epoch": 1.0560177481974486,
+      "grad_norm": 0.9958660089779533,
+      "learning_rate": 3.624382207578254e-05,
+      "loss": 0.3039,
+      "step": 238
+    },
+    {
+      "epoch": 1.0604547975596228,
+      "grad_norm": 0.8133162286658734,
+      "learning_rate": 3.616144975288303e-05,
+      "loss": 0.2468,
+      "step": 239
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 1.0163785046625053,
+      "learning_rate": 3.6079077429983525e-05,
+      "loss": 0.3126,
+      "step": 240
+    },
+    {
+      "epoch": 1.069328896283971,
+      "grad_norm": 0.9544507171632961,
+      "learning_rate": 3.599670510708402e-05,
+      "loss": 0.3053,
+      "step": 241
+    },
+    {
+      "epoch": 1.0737659456461452,
+      "grad_norm": 0.9019798436756871,
+      "learning_rate": 3.5914332784184516e-05,
+      "loss": 0.2968,
+      "step": 242
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 0.7217732576614062,
+      "learning_rate": 3.583196046128501e-05,
+      "loss": 0.2845,
+      "step": 243
+    },
+    {
+      "epoch": 1.0826400443704935,
+      "grad_norm": 1.061456392166015,
+      "learning_rate": 3.5749588138385507e-05,
+      "loss": 0.2893,
+      "step": 244
+    },
+    {
+      "epoch": 1.0870770937326677,
+      "grad_norm": 0.9467418677779301,
+      "learning_rate": 3.5667215815486e-05,
+      "loss": 0.3549,
+      "step": 245
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 0.8491210235543027,
+      "learning_rate": 3.55848434925865e-05,
+      "loss": 0.2585,
+      "step": 246
+    },
+    {
+      "epoch": 1.095951192457016,
+      "grad_norm": 1.1423342713536853,
+      "learning_rate": 3.5502471169686986e-05,
+      "loss": 0.3952,
+      "step": 247
+    },
+    {
+      "epoch": 1.1003882418191901,
+      "grad_norm": 0.9288000150333899,
+      "learning_rate": 3.542009884678748e-05,
+      "loss": 0.3183,
+      "step": 248
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 0.9360465226654853,
+      "learning_rate": 3.533772652388798e-05,
+      "loss": 0.2987,
+      "step": 249
+    },
+    {
+      "epoch": 1.1092623405435384,
+      "grad_norm": 0.8683516789900543,
+      "learning_rate": 3.525535420098847e-05,
+      "loss": 0.2772,
+      "step": 250
+    },
+    {
+      "epoch": 1.1136993899057126,
+      "grad_norm": 1.0308466035630688,
+      "learning_rate": 3.517298187808896e-05,
+      "loss": 0.3035,
+      "step": 251
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 0.8014850407999317,
+      "learning_rate": 3.509060955518946e-05,
+      "loss": 0.2814,
+      "step": 252
+    },
+    {
+      "epoch": 1.122573488630061,
+      "grad_norm": 0.7585536046508811,
+      "learning_rate": 3.500823723228995e-05,
+      "loss": 0.2757,
+      "step": 253
+    },
+    {
+      "epoch": 1.127010537992235,
+      "grad_norm": 0.8396995243167018,
+      "learning_rate": 3.492586490939045e-05,
+      "loss": 0.2495,
+      "step": 254
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 0.7893894386934217,
+      "learning_rate": 3.4843492586490936e-05,
+      "loss": 0.2749,
+      "step": 255
+    },
+    {
+      "epoch": 1.1358846367165834,
+      "grad_norm": 1.0932045802816392,
+      "learning_rate": 3.476112026359143e-05,
+      "loss": 0.3878,
+      "step": 256
+    },
+    {
+      "epoch": 1.1403216860787575,
+      "grad_norm": 0.7575049808703544,
+      "learning_rate": 3.467874794069193e-05,
+      "loss": 0.2722,
+      "step": 257
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 1.159473051053268,
+      "learning_rate": 3.459637561779242e-05,
+      "loss": 0.3658,
+      "step": 258
+    },
+    {
+      "epoch": 1.149195784803106,
+      "grad_norm": 1.053227183651025,
+      "learning_rate": 3.451400329489292e-05,
+      "loss": 0.311,
+      "step": 259
+    },
+    {
+      "epoch": 1.1536328341652802,
+      "grad_norm": 1.2917391182463467,
+      "learning_rate": 3.4431630971993414e-05,
+      "loss": 0.2715,
+      "step": 260
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 0.801097416684275,
+      "learning_rate": 3.434925864909391e-05,
+      "loss": 0.3288,
+      "step": 261
+    },
+    {
+      "epoch": 1.1625069328896285,
+      "grad_norm": 1.1338875766376697,
+      "learning_rate": 3.4266886326194404e-05,
+      "loss": 0.3412,
+      "step": 262
+    },
+    {
+      "epoch": 1.1669439822518026,
+      "grad_norm": 0.9350602192279442,
+      "learning_rate": 3.418451400329489e-05,
+      "loss": 0.3275,
+      "step": 263
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 0.7845508271984446,
+      "learning_rate": 3.410214168039539e-05,
+      "loss": 0.2728,
+      "step": 264
+    },
+    {
+      "epoch": 1.175818080976151,
+      "grad_norm": 0.858609847246771,
+      "learning_rate": 3.4019769357495884e-05,
+      "loss": 0.2967,
+      "step": 265
+    },
+    {
+      "epoch": 1.180255130338325,
+      "grad_norm": 0.9073453400448334,
+      "learning_rate": 3.393739703459638e-05,
+      "loss": 0.2576,
+      "step": 266
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 0.8136112976233186,
+      "learning_rate": 3.385502471169687e-05,
+      "loss": 0.2419,
+      "step": 267
+    },
+    {
+      "epoch": 1.1891292290626734,
+      "grad_norm": 0.8282445065621016,
+      "learning_rate": 3.3772652388797364e-05,
+      "loss": 0.2645,
+      "step": 268
+    },
+    {
+      "epoch": 1.1935662784248475,
+      "grad_norm": 1.2341381763275305,
+      "learning_rate": 3.369028006589786e-05,
+      "loss": 0.3302,
+      "step": 269
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 1.0081006606605467,
+      "learning_rate": 3.3607907742998355e-05,
+      "loss": 0.2697,
+      "step": 270
+    },
+    {
+      "epoch": 1.2024403771491958,
+      "grad_norm": 0.8126944529320962,
+      "learning_rate": 3.352553542009885e-05,
+      "loss": 0.2788,
+      "step": 271
+    },
+    {
+      "epoch": 1.20687742651137,
+      "grad_norm": 0.7731330349380529,
+      "learning_rate": 3.3443163097199346e-05,
+      "loss": 0.2507,
+      "step": 272
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 0.8577813370399663,
+      "learning_rate": 3.336079077429984e-05,
+      "loss": 0.2703,
+      "step": 273
+    },
+    {
+      "epoch": 1.2157515252357183,
+      "grad_norm": 1.070830505209032,
+      "learning_rate": 3.327841845140033e-05,
+      "loss": 0.3063,
+      "step": 274
+    },
+    {
+      "epoch": 1.2201885745978924,
+      "grad_norm": 1.0377982342895966,
+      "learning_rate": 3.3196046128500825e-05,
+      "loss": 0.3407,
+      "step": 275
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 1.210461824573832,
+      "learning_rate": 3.311367380560132e-05,
+      "loss": 0.2878,
+      "step": 276
+    },
+    {
+      "epoch": 1.2290626733222407,
+      "grad_norm": 0.8504328102649177,
+      "learning_rate": 3.3031301482701816e-05,
+      "loss": 0.3108,
+      "step": 277
+    },
+    {
+      "epoch": 1.2334997226844149,
+      "grad_norm": 0.9184666358814657,
+      "learning_rate": 3.2948929159802305e-05,
+      "loss": 0.3501,
+      "step": 278
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 1.0056092182673133,
+      "learning_rate": 3.28665568369028e-05,
+      "loss": 0.2707,
+      "step": 279
+    },
+    {
+      "epoch": 1.2423738214087632,
+      "grad_norm": 0.994694343852553,
+      "learning_rate": 3.2784184514003296e-05,
+      "loss": 0.3204,
+      "step": 280
+    },
+    {
+      "epoch": 1.2468108707709373,
+      "grad_norm": 0.8917410700644745,
+      "learning_rate": 3.270181219110379e-05,
+      "loss": 0.2532,
+      "step": 281
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 0.9639140603708609,
+      "learning_rate": 3.261943986820428e-05,
+      "loss": 0.304,
+      "step": 282
+    },
+    {
+      "epoch": 1.2556849694952856,
+      "grad_norm": 0.8265452521963786,
+      "learning_rate": 3.2537067545304775e-05,
+      "loss": 0.3003,
+      "step": 283
+    },
+    {
+      "epoch": 1.2601220188574598,
+      "grad_norm": 0.9524436766825125,
+      "learning_rate": 3.245469522240527e-05,
+      "loss": 0.3842,
+      "step": 284
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 0.8633449410120922,
+      "learning_rate": 3.2372322899505766e-05,
+      "loss": 0.2975,
+      "step": 285
+    },
+    {
+      "epoch": 1.268996117581808,
+      "grad_norm": 0.8701473620456235,
+      "learning_rate": 3.228995057660626e-05,
+      "loss": 0.278,
+      "step": 286
+    },
+    {
+      "epoch": 1.2734331669439822,
+      "grad_norm": 0.8051498714278014,
+      "learning_rate": 3.220757825370676e-05,
+      "loss": 0.2895,
+      "step": 287
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 0.9415134036547899,
+      "learning_rate": 3.212520593080725e-05,
+      "loss": 0.322,
+      "step": 288
+    },
+    {
+      "epoch": 1.2823072656683305,
+      "grad_norm": 0.8176229071894334,
+      "learning_rate": 3.204283360790775e-05,
+      "loss": 0.3009,
+      "step": 289
+    },
+    {
+      "epoch": 1.2867443150305047,
+      "grad_norm": 1.2048221854237993,
+      "learning_rate": 3.196046128500824e-05,
+      "loss": 0.3218,
+      "step": 290
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 0.8829288912315847,
+      "learning_rate": 3.187808896210873e-05,
+      "loss": 0.2953,
+      "step": 291
+    },
+    {
+      "epoch": 1.295618413754853,
+      "grad_norm": 0.8975883783589373,
+      "learning_rate": 3.179571663920923e-05,
+      "loss": 0.2623,
+      "step": 292
+    },
+    {
+      "epoch": 1.3000554631170271,
+      "grad_norm": 0.8529012812904146,
+      "learning_rate": 3.171334431630972e-05,
+      "loss": 0.2307,
+      "step": 293
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 0.9528461736356937,
+      "learning_rate": 3.163097199341021e-05,
+      "loss": 0.2976,
+      "step": 294
+    },
+    {
+      "epoch": 1.3089295618413754,
+      "grad_norm": 0.7594351159934947,
+      "learning_rate": 3.154859967051071e-05,
+      "loss": 0.2556,
+      "step": 295
+    },
+    {
+      "epoch": 1.3133666112035496,
+      "grad_norm": 0.7686742667754073,
+      "learning_rate": 3.14662273476112e-05,
+      "loss": 0.2054,
+      "step": 296
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 0.8745361869963002,
+      "learning_rate": 3.13838550247117e-05,
+      "loss": 0.3749,
+      "step": 297
+    },
+    {
+      "epoch": 1.3222407099278979,
+      "grad_norm": 0.8835353086591043,
+      "learning_rate": 3.1301482701812194e-05,
+      "loss": 0.3486,
+      "step": 298
+    },
+    {
+      "epoch": 1.326677759290072,
+      "grad_norm": 0.9474340544727997,
+      "learning_rate": 3.121911037891269e-05,
+      "loss": 0.2235,
+      "step": 299
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 0.9114246289515612,
+      "learning_rate": 3.1136738056013184e-05,
+      "loss": 0.2845,
+      "step": 300
+    },
+    {
+      "epoch": 1.3355518580144203,
+      "grad_norm": 0.9417669467925294,
+      "learning_rate": 3.105436573311368e-05,
+      "loss": 0.3097,
+      "step": 301
+    },
+    {
+      "epoch": 1.3399889073765945,
+      "grad_norm": 1.030762933091569,
+      "learning_rate": 3.097199341021417e-05,
+      "loss": 0.3131,
+      "step": 302
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 0.9372533791726105,
+      "learning_rate": 3.0889621087314664e-05,
+      "loss": 0.376,
+      "step": 303
+    },
+    {
+      "epoch": 1.348863006100943,
+      "grad_norm": 0.9500920552274774,
+      "learning_rate": 3.080724876441516e-05,
+      "loss": 0.3784,
+      "step": 304
+    },
+    {
+      "epoch": 1.3533000554631172,
+      "grad_norm": 0.7514858230527693,
+      "learning_rate": 3.0724876441515655e-05,
+      "loss": 0.2833,
+      "step": 305
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 0.7939288335875365,
+      "learning_rate": 3.0642504118616144e-05,
+      "loss": 0.2497,
+      "step": 306
+    },
+    {
+      "epoch": 1.3621741541874655,
+      "grad_norm": 0.8358984964843912,
+      "learning_rate": 3.056013179571664e-05,
+      "loss": 0.2741,
+      "step": 307
+    },
+    {
+      "epoch": 1.3666112035496396,
+      "grad_norm": 0.9062743019265684,
+      "learning_rate": 3.0477759472817135e-05,
+      "loss": 0.2993,
+      "step": 308
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 0.9332367282212933,
+      "learning_rate": 3.039538714991763e-05,
+      "loss": 0.3015,
+      "step": 309
+    },
+    {
+      "epoch": 1.375485302273988,
+      "grad_norm": 1.0069126577219902,
+      "learning_rate": 3.0313014827018122e-05,
+      "loss": 0.2868,
+      "step": 310
+    },
+    {
+      "epoch": 1.379922351636162,
+      "grad_norm": 0.9033547884573856,
+      "learning_rate": 3.0230642504118618e-05,
+      "loss": 0.2841,
+      "step": 311
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 0.8078728747356874,
+      "learning_rate": 3.0148270181219113e-05,
+      "loss": 0.2782,
+      "step": 312
+    },
+    {
+      "epoch": 1.3887964503605104,
+      "grad_norm": 0.9084694748675116,
+      "learning_rate": 3.006589785831961e-05,
+      "loss": 0.2259,
+      "step": 313
+    },
+    {
+      "epoch": 1.3932334997226845,
+      "grad_norm": 0.7881935547213544,
+      "learning_rate": 2.9983525535420097e-05,
+      "loss": 0.2487,
+      "step": 314
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 0.8016907528003615,
+      "learning_rate": 2.9901153212520593e-05,
+      "loss": 0.2626,
+      "step": 315
+    },
+    {
+      "epoch": 1.4021075984470328,
+      "grad_norm": 0.8279080376625739,
+      "learning_rate": 2.9818780889621088e-05,
+      "loss": 0.2477,
+      "step": 316
+    },
+    {
+      "epoch": 1.406544647809207,
+      "grad_norm": 0.8147915290345366,
+      "learning_rate": 2.9736408566721587e-05,
+      "loss": 0.2526,
+      "step": 317
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 1.1702039884776705,
+      "learning_rate": 2.9654036243822076e-05,
+      "loss": 0.2434,
+      "step": 318
+    },
+    {
+      "epoch": 1.4154187465335553,
+      "grad_norm": 0.7850168968010128,
+      "learning_rate": 2.957166392092257e-05,
+      "loss": 0.2621,
+      "step": 319
+    },
+    {
+      "epoch": 1.4198557958957294,
+      "grad_norm": 1.0163017952307563,
+      "learning_rate": 2.9489291598023067e-05,
+      "loss": 0.2982,
+      "step": 320
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 0.7909751599169546,
+      "learning_rate": 2.9406919275123562e-05,
+      "loss": 0.281,
+      "step": 321
+    },
+    {
+      "epoch": 1.4287298946200777,
+      "grad_norm": 0.8777170910425888,
+      "learning_rate": 2.9324546952224054e-05,
+      "loss": 0.2668,
+      "step": 322
+    },
+    {
+      "epoch": 1.4331669439822519,
+      "grad_norm": 1.0099323658912123,
+      "learning_rate": 2.924217462932455e-05,
+      "loss": 0.2783,
+      "step": 323
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 0.8585319545399854,
+      "learning_rate": 2.9159802306425045e-05,
+      "loss": 0.2639,
+      "step": 324
+    },
+    {
+      "epoch": 1.4420410427066002,
+      "grad_norm": 1.2373091194055394,
+      "learning_rate": 2.907742998352554e-05,
+      "loss": 0.3356,
+      "step": 325
+    },
+    {
+      "epoch": 1.4464780920687743,
+      "grad_norm": 0.8724563274001284,
+      "learning_rate": 2.899505766062603e-05,
+      "loss": 0.2682,
+      "step": 326
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 0.8894584444512684,
+      "learning_rate": 2.8912685337726525e-05,
+      "loss": 0.2345,
+      "step": 327
+    },
+    {
+      "epoch": 1.4553521907931226,
+      "grad_norm": 0.791290988794545,
+      "learning_rate": 2.883031301482702e-05,
+      "loss": 0.227,
+      "step": 328
+    },
+    {
+      "epoch": 1.4597892401552968,
+      "grad_norm": 0.9471555980217089,
+      "learning_rate": 2.8747940691927515e-05,
+      "loss": 0.3117,
+      "step": 329
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 0.9461745437913269,
+      "learning_rate": 2.8665568369028008e-05,
+      "loss": 0.2994,
+      "step": 330
+    },
+    {
+      "epoch": 1.468663338879645,
+      "grad_norm": 0.9585732629602921,
+      "learning_rate": 2.8583196046128503e-05,
+      "loss": 0.3601,
+      "step": 331
+    },
+    {
+      "epoch": 1.4731003882418192,
+      "grad_norm": 0.9797419555715378,
+      "learning_rate": 2.8500823723229e-05,
+      "loss": 0.2985,
+      "step": 332
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 0.9041880795511308,
+      "learning_rate": 2.8418451400329494e-05,
+      "loss": 0.2879,
+      "step": 333
+    },
+    {
+      "epoch": 1.4819744869661675,
+      "grad_norm": 0.7579890566566851,
+      "learning_rate": 2.8336079077429983e-05,
+      "loss": 0.2363,
+      "step": 334
+    },
+    {
+      "epoch": 1.4864115363283417,
+      "grad_norm": 1.0136081330061153,
+      "learning_rate": 2.8253706754530478e-05,
+      "loss": 0.3412,
+      "step": 335
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 0.8575878634355566,
+      "learning_rate": 2.8171334431630973e-05,
+      "loss": 0.3048,
+      "step": 336
+    },
+    {
+      "epoch": 1.49528563505269,
+      "grad_norm": 1.0070902284052874,
+      "learning_rate": 2.808896210873147e-05,
+      "loss": 0.2816,
+      "step": 337
+    },
+    {
+      "epoch": 1.4997226844148641,
+      "grad_norm": 1.0785926689088374,
+      "learning_rate": 2.800658978583196e-05,
+      "loss": 0.2806,
+      "step": 338
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 0.8276318215194345,
+      "learning_rate": 2.7924217462932456e-05,
+      "loss": 0.2752,
+      "step": 339
+    },
+    {
+      "epoch": 1.5085967831392124,
+      "grad_norm": 0.9757991401247934,
+      "learning_rate": 2.7841845140032952e-05,
+      "loss": 0.2357,
+      "step": 340
+    },
+    {
+      "epoch": 1.5130338325013866,
+      "grad_norm": 1.8061211906359669,
+      "learning_rate": 2.775947281713344e-05,
+      "loss": 0.3929,
+      "step": 341
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 0.8440439630511615,
+      "learning_rate": 2.7677100494233936e-05,
+      "loss": 0.2631,
+      "step": 342
+    },
+    {
+      "epoch": 1.5219079312257349,
+      "grad_norm": 0.8785557026624231,
+      "learning_rate": 2.759472817133443e-05,
+      "loss": 0.2579,
+      "step": 343
+    },
+    {
+      "epoch": 1.526344980587909,
+      "grad_norm": 0.9168099414234008,
+      "learning_rate": 2.7512355848434927e-05,
+      "loss": 0.3854,
+      "step": 344
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 0.9170947195302683,
+      "learning_rate": 2.742998352553542e-05,
+      "loss": 0.3013,
+      "step": 345
+    },
+    {
+      "epoch": 1.5352190793122573,
+      "grad_norm": 0.8619100460371237,
+      "learning_rate": 2.7347611202635915e-05,
+      "loss": 0.3073,
+      "step": 346
+    },
+    {
+      "epoch": 1.5396561286744315,
+      "grad_norm": 0.9818485779002657,
+      "learning_rate": 2.726523887973641e-05,
+      "loss": 0.3411,
+      "step": 347
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 0.9725562967318679,
+      "learning_rate": 2.7182866556836905e-05,
+      "loss": 0.2584,
+      "step": 348
+    },
+    {
+      "epoch": 1.5485302273987798,
+      "grad_norm": 1.0259249212520478,
+      "learning_rate": 2.7100494233937394e-05,
+      "loss": 0.3058,
+      "step": 349
+    },
+    {
+      "epoch": 1.552967276760954,
+      "grad_norm": 1.0405768351821034,
+      "learning_rate": 2.7018121911037893e-05,
+      "loss": 0.2496,
+      "step": 350
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 0.953800399868498,
+      "learning_rate": 2.693574958813839e-05,
+      "loss": 0.3542,
+      "step": 351
+    },
+    {
+      "epoch": 1.5618413754853022,
+      "grad_norm": 0.8755003656509656,
+      "learning_rate": 2.6853377265238884e-05,
+      "loss": 0.3325,
+      "step": 352
+    },
+    {
+      "epoch": 1.5662784248474764,
+      "grad_norm": 1.1163203468668474,
+      "learning_rate": 2.6771004942339373e-05,
+      "loss": 0.31,
+      "step": 353
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 0.7249833861453431,
+      "learning_rate": 2.6688632619439868e-05,
+      "loss": 0.2298,
+      "step": 354
+    },
+    {
+      "epoch": 1.5751525235718247,
+      "grad_norm": 0.8321543857634233,
+      "learning_rate": 2.6606260296540363e-05,
+      "loss": 0.3034,
+      "step": 355
+    },
+    {
+      "epoch": 1.5795895729339988,
+      "grad_norm": 0.9329128575612846,
+      "learning_rate": 2.652388797364086e-05,
+      "loss": 0.3591,
+      "step": 356
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 0.7888539104661904,
+      "learning_rate": 2.644151565074135e-05,
+      "loss": 0.2383,
+      "step": 357
+    },
+    {
+      "epoch": 1.5884636716583471,
+      "grad_norm": 0.8482609363146597,
+      "learning_rate": 2.6359143327841846e-05,
+      "loss": 0.3259,
+      "step": 358
+    },
+    {
+      "epoch": 1.5929007210205213,
+      "grad_norm": 0.8128427369485877,
+      "learning_rate": 2.6276771004942342e-05,
+      "loss": 0.2471,
+      "step": 359
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 1.1424089877149042,
+      "learning_rate": 2.6194398682042837e-05,
+      "loss": 0.3053,
+      "step": 360
+    },
+    {
+      "epoch": 1.6017748197448696,
+      "grad_norm": 0.797133684244717,
+      "learning_rate": 2.6112026359143326e-05,
+      "loss": 0.3085,
+      "step": 361
+    },
+    {
+      "epoch": 1.6062118691070437,
+      "grad_norm": 1.0599037313830935,
+      "learning_rate": 2.602965403624382e-05,
+      "loss": 0.3954,
+      "step": 362
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 0.9704756594970683,
+      "learning_rate": 2.5947281713344317e-05,
+      "loss": 0.29,
+      "step": 363
+    },
+    {
+      "epoch": 1.615085967831392,
+      "grad_norm": 0.8971954716853453,
+      "learning_rate": 2.5864909390444812e-05,
+      "loss": 0.2839,
+      "step": 364
+    },
+    {
+      "epoch": 1.6195230171935662,
+      "grad_norm": 0.8698279245703869,
+      "learning_rate": 2.5782537067545305e-05,
+      "loss": 0.2789,
+      "step": 365
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 1.0854765098802885,
+      "learning_rate": 2.57001647446458e-05,
+      "loss": 0.3104,
+      "step": 366
+    },
+    {
+      "epoch": 1.6283971159179145,
+      "grad_norm": 0.8386426987212547,
+      "learning_rate": 2.5617792421746295e-05,
+      "loss": 0.2445,
+      "step": 367
+    },
+    {
+      "epoch": 1.6328341652800886,
+      "grad_norm": 0.8092227277052146,
+      "learning_rate": 2.553542009884679e-05,
+      "loss": 0.2425,
+      "step": 368
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 1.0627030571275184,
+      "learning_rate": 2.545304777594728e-05,
+      "loss": 0.2771,
+      "step": 369
+    },
+    {
+      "epoch": 1.641708264004437,
+      "grad_norm": 0.7848551917908129,
+      "learning_rate": 2.5370675453047775e-05,
+      "loss": 0.2914,
+      "step": 370
+    },
+    {
+      "epoch": 1.646145313366611,
+      "grad_norm": 0.9757022578302557,
+      "learning_rate": 2.528830313014827e-05,
+      "loss": 0.2373,
+      "step": 371
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 1.027171131325082,
+      "learning_rate": 2.520593080724877e-05,
+      "loss": 0.2909,
+      "step": 372
+    },
+    {
+      "epoch": 1.6550194120909594,
+      "grad_norm": 0.8259234404332476,
+      "learning_rate": 2.5123558484349258e-05,
+      "loss": 0.2698,
+      "step": 373
+    },
+    {
+      "epoch": 1.6594564614531335,
+      "grad_norm": 0.9243264336301544,
+      "learning_rate": 2.5041186161449753e-05,
+      "loss": 0.3099,
+      "step": 374
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 0.7192692746737803,
+      "learning_rate": 2.495881383855025e-05,
+      "loss": 0.2423,
+      "step": 375
+    },
+    {
+      "epoch": 1.6683305601774818,
+      "grad_norm": 0.9214747901796418,
+      "learning_rate": 2.487644151565074e-05,
+      "loss": 0.2449,
+      "step": 376
+    },
+    {
+      "epoch": 1.672767609539656,
+      "grad_norm": 0.8653940436672877,
+      "learning_rate": 2.4794069192751236e-05,
+      "loss": 0.2449,
+      "step": 377
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 0.8719790283817912,
+      "learning_rate": 2.4711696869851732e-05,
+      "loss": 0.3459,
+      "step": 378
+    },
+    {
+      "epoch": 1.6816417082640043,
+      "grad_norm": 0.884484185262341,
+      "learning_rate": 2.4629324546952227e-05,
+      "loss": 0.2936,
+      "step": 379
+    },
+    {
+      "epoch": 1.6860787576261784,
+      "grad_norm": 0.9556852567947143,
+      "learning_rate": 2.454695222405272e-05,
+      "loss": 0.258,
+      "step": 380
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 0.9006107747994894,
+      "learning_rate": 2.4464579901153215e-05,
+      "loss": 0.2868,
+      "step": 381
+    },
+    {
+      "epoch": 1.694952856350527,
+      "grad_norm": 1.0573331639957975,
+      "learning_rate": 2.4382207578253707e-05,
+      "loss": 0.2647,
+      "step": 382
+    },
+    {
+      "epoch": 1.699389905712701,
+      "grad_norm": 0.8786757442415969,
+      "learning_rate": 2.4299835255354202e-05,
+      "loss": 0.2474,
+      "step": 383
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 1.034443390305425,
+      "learning_rate": 2.4217462932454694e-05,
+      "loss": 0.3275,
+      "step": 384
+    },
+    {
+      "epoch": 1.7082640044370494,
+      "grad_norm": 1.0336686295354667,
+      "learning_rate": 2.413509060955519e-05,
+      "loss": 0.2602,
+      "step": 385
+    },
+    {
+      "epoch": 1.7127010537992235,
+      "grad_norm": 0.9454431364731563,
+      "learning_rate": 2.4052718286655685e-05,
+      "loss": 0.2715,
+      "step": 386
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 0.8850143116801344,
+      "learning_rate": 2.397034596375618e-05,
+      "loss": 0.2814,
+      "step": 387
+    },
+    {
+      "epoch": 1.7215751525235718,
+      "grad_norm": 0.8386031102638419,
+      "learning_rate": 2.3887973640856673e-05,
+      "loss": 0.263,
+      "step": 388
+    },
+    {
+      "epoch": 1.726012201885746,
+      "grad_norm": 1.061756996212574,
+      "learning_rate": 2.380560131795717e-05,
+      "loss": 0.3846,
+      "step": 389
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 0.8173241087183969,
+      "learning_rate": 2.372322899505766e-05,
+      "loss": 0.2794,
+      "step": 390
+    },
+    {
+      "epoch": 1.7348863006100943,
+      "grad_norm": 0.7094044767464892,
+      "learning_rate": 2.3640856672158156e-05,
+      "loss": 0.24,
+      "step": 391
+    },
+    {
+      "epoch": 1.7393233499722685,
+      "grad_norm": 1.0181711412890253,
+      "learning_rate": 2.355848434925865e-05,
+      "loss": 0.2293,
+      "step": 392
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 0.7974819041384473,
+      "learning_rate": 2.3476112026359147e-05,
+      "loss": 0.299,
+      "step": 393
+    },
+    {
+      "epoch": 1.7481974486966168,
+      "grad_norm": 1.0842945651693483,
+      "learning_rate": 2.339373970345964e-05,
+      "loss": 0.4007,
+      "step": 394
+    },
+    {
+      "epoch": 1.752634498058791,
+      "grad_norm": 1.0029823415690775,
+      "learning_rate": 2.3311367380560134e-05,
+      "loss": 0.2896,
+      "step": 395
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 0.8037712287769038,
+      "learning_rate": 2.3228995057660626e-05,
+      "loss": 0.2463,
+      "step": 396
+    },
+    {
+      "epoch": 1.7615085967831392,
+      "grad_norm": 0.8595851294504453,
+      "learning_rate": 2.3146622734761122e-05,
+      "loss": 0.2451,
+      "step": 397
+    },
+    {
+      "epoch": 1.7659456461453134,
+      "grad_norm": 0.8154261057883538,
+      "learning_rate": 2.3064250411861614e-05,
+      "loss": 0.2669,
+      "step": 398
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 0.8657303673690703,
+      "learning_rate": 2.298187808896211e-05,
+      "loss": 0.33,
+      "step": 399
+    },
+    {
+      "epoch": 1.7748197448696617,
+      "grad_norm": 0.8175680846114033,
+      "learning_rate": 2.2899505766062605e-05,
+      "loss": 0.2579,
+      "step": 400
+    },
+    {
+      "epoch": 1.7792567942318358,
+      "grad_norm": 0.8055971799890129,
+      "learning_rate": 2.28171334431631e-05,
+      "loss": 0.2466,
+      "step": 401
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 0.9723359465312749,
+      "learning_rate": 2.2734761120263592e-05,
+      "loss": 0.2923,
+      "step": 402
+    },
+    {
+      "epoch": 1.788130892956184,
+      "grad_norm": 0.7871930477829718,
+      "learning_rate": 2.2652388797364088e-05,
+      "loss": 0.245,
+      "step": 403
+    },
+    {
+      "epoch": 1.7925679423183583,
+      "grad_norm": 0.9587060093143946,
+      "learning_rate": 2.257001647446458e-05,
+      "loss": 0.327,
+      "step": 404
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 0.7518155058726878,
+      "learning_rate": 2.2487644151565075e-05,
+      "loss": 0.248,
+      "step": 405
+    },
+    {
+      "epoch": 1.8014420410427066,
+      "grad_norm": 0.8452159720710262,
+      "learning_rate": 2.240527182866557e-05,
+      "loss": 0.2898,
+      "step": 406
+    },
+    {
+      "epoch": 1.8058790904048807,
+      "grad_norm": 0.8077753320403622,
+      "learning_rate": 2.2322899505766066e-05,
+      "loss": 0.3047,
+      "step": 407
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 0.883161222736862,
+      "learning_rate": 2.224052718286656e-05,
+      "loss": 0.2765,
+      "step": 408
+    },
+    {
+      "epoch": 1.8147531891292292,
+      "grad_norm": 1.019587665034733,
+      "learning_rate": 2.215815485996705e-05,
+      "loss": 0.3072,
+      "step": 409
+    },
+    {
+      "epoch": 1.8191902384914034,
+      "grad_norm": 0.973902291030218,
+      "learning_rate": 2.2075782537067546e-05,
+      "loss": 0.2474,
+      "step": 410
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 0.8247021166224197,
+      "learning_rate": 2.1993410214168038e-05,
+      "loss": 0.2281,
+      "step": 411
+    },
+    {
+      "epoch": 1.8280643372157517,
+      "grad_norm": 0.8467350889953564,
+      "learning_rate": 2.1911037891268533e-05,
+      "loss": 0.2551,
+      "step": 412
+    },
+    {
+      "epoch": 1.8325013865779258,
+      "grad_norm": 0.9629678753241201,
+      "learning_rate": 2.182866556836903e-05,
+      "loss": 0.319,
+      "step": 413
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 0.9004484807622022,
+      "learning_rate": 2.1746293245469524e-05,
+      "loss": 0.3164,
+      "step": 414
+    },
+    {
+      "epoch": 1.8413754853022741,
+      "grad_norm": 0.9259354668143355,
+      "learning_rate": 2.1663920922570016e-05,
+      "loss": 0.255,
+      "step": 415
+    },
+    {
+      "epoch": 1.8458125346644483,
+      "grad_norm": 1.0792016996983913,
+      "learning_rate": 2.1581548599670512e-05,
+      "loss": 0.2869,
+      "step": 416
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 0.7780691624718593,
+      "learning_rate": 2.1499176276771004e-05,
+      "loss": 0.2631,
+      "step": 417
+    },
+    {
+      "epoch": 1.8546866333887966,
+      "grad_norm": 1.0150115287764934,
+      "learning_rate": 2.14168039538715e-05,
+      "loss": 0.2905,
+      "step": 418
+    },
+    {
+      "epoch": 1.8591236827509707,
+      "grad_norm": 0.9414552541301019,
+      "learning_rate": 2.1334431630971995e-05,
+      "loss": 0.2175,
+      "step": 419
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 1.022383285240764,
+      "learning_rate": 2.125205930807249e-05,
+      "loss": 0.2797,
+      "step": 420
+    },
+    {
+      "epoch": 1.867997781475319,
+      "grad_norm": 0.8661251950323628,
+      "learning_rate": 2.1169686985172982e-05,
+      "loss": 0.3511,
+      "step": 421
+    },
+    {
+      "epoch": 1.8724348308374932,
+      "grad_norm": 1.0114859530406477,
+      "learning_rate": 2.1087314662273478e-05,
+      "loss": 0.2536,
+      "step": 422
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 0.8061046528471709,
+      "learning_rate": 2.100494233937397e-05,
+      "loss": 0.3064,
+      "step": 423
+    },
+    {
+      "epoch": 1.8813089295618415,
+      "grad_norm": 0.8968122294188483,
+      "learning_rate": 2.0922570016474465e-05,
+      "loss": 0.2891,
+      "step": 424
+    },
+    {
+      "epoch": 1.8857459789240156,
+      "grad_norm": 0.9586335140521199,
+      "learning_rate": 2.0840197693574957e-05,
+      "loss": 0.2613,
+      "step": 425
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 0.9543990078587583,
+      "learning_rate": 2.0757825370675453e-05,
+      "loss": 0.3193,
+      "step": 426
+    },
+    {
+      "epoch": 1.894620077648364,
+      "grad_norm": 0.964414122524041,
+      "learning_rate": 2.067545304777595e-05,
+      "loss": 0.2762,
+      "step": 427
+    },
+    {
+      "epoch": 1.899057127010538,
+      "grad_norm": 0.7175021327050176,
+      "learning_rate": 2.0593080724876444e-05,
+      "loss": 0.3267,
+      "step": 428
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 1.0475358727442101,
+      "learning_rate": 2.0510708401976936e-05,
+      "loss": 0.2623,
+      "step": 429
+    },
+    {
+      "epoch": 1.9079312257348864,
+      "grad_norm": 0.9951684880533599,
+      "learning_rate": 2.042833607907743e-05,
+      "loss": 0.3282,
+      "step": 430
+    },
+    {
+      "epoch": 1.9123682750970605,
+      "grad_norm": 0.8125336713416275,
+      "learning_rate": 2.0345963756177923e-05,
+      "loss": 0.4212,
+      "step": 431
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 0.89389068456606,
+      "learning_rate": 2.026359143327842e-05,
+      "loss": 0.2463,
+      "step": 432
+    },
+    {
+      "epoch": 1.9212423738214088,
+      "grad_norm": 0.8712168721998037,
+      "learning_rate": 2.0181219110378914e-05,
+      "loss": 0.2754,
+      "step": 433
+    },
+    {
+      "epoch": 1.925679423183583,
+      "grad_norm": 0.8748220535261405,
+      "learning_rate": 2.009884678747941e-05,
+      "loss": 0.2562,
+      "step": 434
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 0.9056723496594914,
+      "learning_rate": 2.0016474464579902e-05,
+      "loss": 0.3349,
+      "step": 435
+    },
+    {
+      "epoch": 1.9345535219079313,
+      "grad_norm": 0.767429638349272,
+      "learning_rate": 1.9934102141680397e-05,
+      "loss": 0.2061,
+      "step": 436
+    },
+    {
+      "epoch": 1.9389905712701054,
+      "grad_norm": 0.8320915665949056,
+      "learning_rate": 1.985172981878089e-05,
+      "loss": 0.3067,
+      "step": 437
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 0.7555880983869324,
+      "learning_rate": 1.9769357495881385e-05,
+      "loss": 0.2607,
+      "step": 438
+    },
+    {
+      "epoch": 1.9478646699944537,
+      "grad_norm": 0.7105796667986897,
+      "learning_rate": 1.9686985172981877e-05,
+      "loss": 0.2464,
+      "step": 439
+    },
+    {
+      "epoch": 1.9523017193566279,
+      "grad_norm": 0.9767889752038426,
+      "learning_rate": 1.9604612850082372e-05,
+      "loss": 0.2808,
+      "step": 440
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 0.7118591789785065,
+      "learning_rate": 1.9522240527182868e-05,
+      "loss": 0.2548,
+      "step": 441
+    },
+    {
+      "epoch": 1.9611758180809762,
+      "grad_norm": 0.9356831429456314,
+      "learning_rate": 1.9439868204283363e-05,
+      "loss": 0.2863,
+      "step": 442
+    },
+    {
+      "epoch": 1.9656128674431503,
+      "grad_norm": 0.9251371952073134,
+      "learning_rate": 1.9357495881383855e-05,
+      "loss": 0.2446,
+      "step": 443
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 0.9581332032312233,
+      "learning_rate": 1.927512355848435e-05,
+      "loss": 0.2707,
+      "step": 444
+    },
+    {
+      "epoch": 1.9744869661674986,
+      "grad_norm": 0.8201545926653439,
+      "learning_rate": 1.9192751235584843e-05,
+      "loss": 0.2358,
+      "step": 445
+    },
+    {
+      "epoch": 1.9789240155296728,
+      "grad_norm": 0.8917348538527614,
+      "learning_rate": 1.911037891268534e-05,
+      "loss": 0.2327,
+      "step": 446
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 0.793870009713998,
+      "learning_rate": 1.9028006589785834e-05,
+      "loss": 0.2351,
+      "step": 447
+    },
+    {
+      "epoch": 1.987798114254021,
+      "grad_norm": 0.7696144330140515,
+      "learning_rate": 1.894563426688633e-05,
+      "loss": 0.2435,
+      "step": 448
+    },
+    {
+      "epoch": 1.9922351636161952,
+      "grad_norm": 0.8950314518394308,
+      "learning_rate": 1.886326194398682e-05,
+      "loss": 0.2186,
+      "step": 449
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 0.937115749450349,
+      "learning_rate": 1.8780889621087317e-05,
+      "loss": 0.2396,
+      "step": 450
+    },
+    {
+      "epoch": 2.0011092623405435,
+      "grad_norm": 0.7675418990810471,
+      "learning_rate": 1.869851729818781e-05,
+      "loss": 0.2239,
+      "step": 451
+    },
+    {
+      "epoch": 2.0055463117027177,
+      "grad_norm": 0.7037873856826813,
+      "learning_rate": 1.8616144975288304e-05,
+      "loss": 0.0884,
+      "step": 452
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 1.0684471303545413,
+      "learning_rate": 1.8533772652388796e-05,
+      "loss": 0.0896,
+      "step": 453
+    },
+    {
+      "epoch": 2.014420410427066,
+      "grad_norm": 0.4397641924768517,
+      "learning_rate": 1.8451400329489292e-05,
+      "loss": 0.0987,
+      "step": 454
+    },
+    {
+      "epoch": 2.01885745978924,
+      "grad_norm": 0.753924710536916,
+      "learning_rate": 1.8369028006589787e-05,
+      "loss": 0.108,
+      "step": 455
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 0.7470844342262113,
+      "learning_rate": 1.8286655683690283e-05,
+      "loss": 0.0982,
+      "step": 456
+    },
+    {
+      "epoch": 2.0277315585135884,
+      "grad_norm": 0.4854531440890792,
+      "learning_rate": 1.8204283360790775e-05,
+      "loss": 0.0931,
+      "step": 457
+    },
+    {
+      "epoch": 2.0321686078757626,
+      "grad_norm": 1.039503591889987,
+      "learning_rate": 1.812191103789127e-05,
+      "loss": 0.1163,
+      "step": 458
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 0.9666652397167154,
+      "learning_rate": 1.8039538714991762e-05,
+      "loss": 0.1476,
+      "step": 459
+    },
+    {
+      "epoch": 2.041042706600111,
+      "grad_norm": 0.4388264574776795,
+      "learning_rate": 1.7957166392092258e-05,
+      "loss": 0.0621,
+      "step": 460
+    },
+    {
+      "epoch": 2.045479755962285,
+      "grad_norm": 0.5621865511807232,
+      "learning_rate": 1.7874794069192753e-05,
+      "loss": 0.0812,
+      "step": 461
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 0.41065498184531973,
+      "learning_rate": 1.779242174629325e-05,
+      "loss": 0.0753,
+      "step": 462
+    },
+    {
+      "epoch": 2.0543538546866333,
+      "grad_norm": 0.5116349654001895,
+      "learning_rate": 1.771004942339374e-05,
+      "loss": 0.0839,
+      "step": 463
+    },
+    {
+      "epoch": 2.0587909040488075,
+      "grad_norm": 0.5144947264005313,
+      "learning_rate": 1.7627677100494236e-05,
+      "loss": 0.095,
+      "step": 464
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 0.5109739926339697,
+      "learning_rate": 1.754530477759473e-05,
+      "loss": 0.0632,
+      "step": 465
+    },
+    {
+      "epoch": 2.067665002773156,
+      "grad_norm": 0.6532249894330738,
+      "learning_rate": 1.7462932454695224e-05,
+      "loss": 0.0853,
+      "step": 466
+    },
+    {
+      "epoch": 2.07210205213533,
+      "grad_norm": 1.5461723966868925,
+      "learning_rate": 1.7380560131795716e-05,
+      "loss": 0.1146,
+      "step": 467
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 0.7694204027923975,
+      "learning_rate": 1.729818780889621e-05,
+      "loss": 0.0851,
+      "step": 468
+    },
+    {
+      "epoch": 2.0809761508596782,
+      "grad_norm": 0.781619518629895,
+      "learning_rate": 1.7215815485996707e-05,
+      "loss": 0.1058,
+      "step": 469
+    },
+    {
+      "epoch": 2.0854132002218524,
+      "grad_norm": 0.48288750667703845,
+      "learning_rate": 1.7133443163097202e-05,
+      "loss": 0.0711,
+      "step": 470
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 0.6270734249696698,
+      "learning_rate": 1.7051070840197694e-05,
+      "loss": 0.0771,
+      "step": 471
+    },
+    {
+      "epoch": 2.0942872989462007,
+      "grad_norm": 0.5509850686277749,
+      "learning_rate": 1.696869851729819e-05,
+      "loss": 0.076,
+      "step": 472
+    },
+    {
+      "epoch": 2.098724348308375,
+      "grad_norm": 0.6074745179942609,
+      "learning_rate": 1.6886326194398682e-05,
+      "loss": 0.0692,
+      "step": 473
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 0.4698384998207071,
+      "learning_rate": 1.6803953871499177e-05,
+      "loss": 0.0677,
+      "step": 474
+    },
+    {
+      "epoch": 2.107598447032723,
+      "grad_norm": 0.5091839976360598,
+      "learning_rate": 1.6721581548599673e-05,
+      "loss": 0.082,
+      "step": 475
+    },
+    {
+      "epoch": 2.1120354963948973,
+      "grad_norm": 0.6306552852091889,
+      "learning_rate": 1.6639209225700165e-05,
+      "loss": 0.0799,
+      "step": 476
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 0.5453074741951704,
+      "learning_rate": 1.655683690280066e-05,
+      "loss": 0.087,
+      "step": 477
+    },
+    {
+      "epoch": 2.1209095951192456,
+      "grad_norm": 0.8333096421370929,
+      "learning_rate": 1.6474464579901152e-05,
+      "loss": 0.1027,
+      "step": 478
+    },
+    {
+      "epoch": 2.1253466444814197,
+      "grad_norm": 0.6588226255524119,
+      "learning_rate": 1.6392092257001648e-05,
+      "loss": 0.0946,
+      "step": 479
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 0.5146614087549416,
+      "learning_rate": 1.630971993410214e-05,
+      "loss": 0.0849,
+      "step": 480
+    },
+    {
+      "epoch": 2.134220743205768,
+      "grad_norm": 0.6728810828478047,
+      "learning_rate": 1.6227347611202635e-05,
+      "loss": 0.1113,
+      "step": 481
+    },
+    {
+      "epoch": 2.138657792567942,
+      "grad_norm": 0.6846333038038229,
+      "learning_rate": 1.614497528830313e-05,
+      "loss": 0.079,
+      "step": 482
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 0.5156807435854919,
+      "learning_rate": 1.6062602965403626e-05,
+      "loss": 0.0666,
+      "step": 483
+    },
+    {
+      "epoch": 2.1475318912922905,
+      "grad_norm": 0.44976733338742386,
+      "learning_rate": 1.598023064250412e-05,
+      "loss": 0.0717,
+      "step": 484
+    },
+    {
+      "epoch": 2.1519689406544646,
+      "grad_norm": 0.4132489182745962,
+      "learning_rate": 1.5897858319604614e-05,
+      "loss": 0.0589,
+      "step": 485
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 1.332057382097956,
+      "learning_rate": 1.5815485996705106e-05,
+      "loss": 0.1179,
+      "step": 486
+    },
+    {
+      "epoch": 2.160843039378813,
+      "grad_norm": 0.8673309345016713,
+      "learning_rate": 1.57331136738056e-05,
+      "loss": 0.0864,
+      "step": 487
+    },
+    {
+      "epoch": 2.165280088740987,
+      "grad_norm": 0.5207558160952449,
+      "learning_rate": 1.5650741350906097e-05,
+      "loss": 0.0828,
+      "step": 488
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 0.5697716016101573,
+      "learning_rate": 1.5568369028006592e-05,
+      "loss": 0.0721,
+      "step": 489
+    },
+    {
+      "epoch": 2.1741541874653354,
+      "grad_norm": 0.6143986429843966,
+      "learning_rate": 1.5485996705107084e-05,
+      "loss": 0.089,
+      "step": 490
+    },
+    {
+      "epoch": 2.1785912368275095,
+      "grad_norm": 0.6728794883834863,
+      "learning_rate": 1.540362438220758e-05,
+      "loss": 0.0724,
+      "step": 491
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 0.5646442235039268,
+      "learning_rate": 1.5321252059308072e-05,
+      "loss": 0.0903,
+      "step": 492
+    },
+    {
+      "epoch": 2.187465335551858,
+      "grad_norm": 0.5635051753086566,
+      "learning_rate": 1.5238879736408567e-05,
+      "loss": 0.063,
+      "step": 493
+    },
+    {
+      "epoch": 2.191902384914032,
+      "grad_norm": 0.5104055662959266,
+      "learning_rate": 1.5156507413509061e-05,
+      "loss": 0.0493,
+      "step": 494
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 0.4011144468810002,
+      "learning_rate": 1.5074135090609557e-05,
+      "loss": 0.0451,
+      "step": 495
+    },
+    {
+      "epoch": 2.2007764836383803,
+      "grad_norm": 0.3690753051448543,
+      "learning_rate": 1.4991762767710049e-05,
+      "loss": 0.0542,
+      "step": 496
+    },
+    {
+      "epoch": 2.2052135330005544,
+      "grad_norm": 0.646653951203632,
+      "learning_rate": 1.4909390444810544e-05,
+      "loss": 0.0957,
+      "step": 497
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 0.6739316998918211,
+      "learning_rate": 1.4827018121911038e-05,
+      "loss": 0.0931,
+      "step": 498
+    },
+    {
+      "epoch": 2.2140876317249027,
+      "grad_norm": 0.45490674242141194,
+      "learning_rate": 1.4744645799011533e-05,
+      "loss": 0.0595,
+      "step": 499
+    },
+    {
+      "epoch": 2.218524681087077,
+      "grad_norm": 0.5700139871353489,
+      "learning_rate": 1.4662273476112027e-05,
+      "loss": 0.0583,
+      "step": 500
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 0.6867079073209452,
+      "learning_rate": 1.4579901153212522e-05,
+      "loss": 0.1612,
+      "step": 501
+    },
+    {
+      "epoch": 2.227398779811425,
+      "grad_norm": 0.5329755919416289,
+      "learning_rate": 1.4497528830313015e-05,
+      "loss": 0.077,
+      "step": 502
+    },
+    {
+      "epoch": 2.2318358291735994,
+      "grad_norm": 0.8264761290105969,
+      "learning_rate": 1.441515650741351e-05,
+      "loss": 0.1016,
+      "step": 503
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 0.8494572645802404,
+      "learning_rate": 1.4332784184514004e-05,
+      "loss": 0.1514,
+      "step": 504
+    },
+    {
+      "epoch": 2.2407099278979477,
+      "grad_norm": 0.7267350842992685,
+      "learning_rate": 1.42504118616145e-05,
+      "loss": 0.09,
+      "step": 505
+    },
+    {
+      "epoch": 2.245146977260122,
+      "grad_norm": 0.42344046588154616,
+      "learning_rate": 1.4168039538714991e-05,
+      "loss": 0.066,
+      "step": 506
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 0.48137381063317836,
+      "learning_rate": 1.4085667215815487e-05,
+      "loss": 0.0738,
+      "step": 507
+    },
+    {
+      "epoch": 2.25402107598447,
+      "grad_norm": 0.5369929239439096,
+      "learning_rate": 1.400329489291598e-05,
+      "loss": 0.0838,
+      "step": 508
+    },
+    {
+      "epoch": 2.2584581253466443,
+      "grad_norm": 0.5755565396332807,
+      "learning_rate": 1.3920922570016476e-05,
+      "loss": 0.1193,
+      "step": 509
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 0.7425527376654243,
+      "learning_rate": 1.3838550247116968e-05,
+      "loss": 0.159,
+      "step": 510
+    },
+    {
+      "epoch": 2.2673322240709926,
+      "grad_norm": 0.6678896851621685,
+      "learning_rate": 1.3756177924217463e-05,
+      "loss": 0.0704,
+      "step": 511
+    },
+    {
+      "epoch": 2.2717692734331667,
+      "grad_norm": 0.6407874761956576,
+      "learning_rate": 1.3673805601317957e-05,
+      "loss": 0.0673,
+      "step": 512
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 0.606625884811036,
+      "learning_rate": 1.3591433278418453e-05,
+      "loss": 0.053,
+      "step": 513
+    },
+    {
+      "epoch": 2.280643372157515,
+      "grad_norm": 0.43153445634045084,
+      "learning_rate": 1.3509060955518946e-05,
+      "loss": 0.05,
+      "step": 514
+    },
+    {
+      "epoch": 2.2850804215196896,
+      "grad_norm": 0.5168983746170788,
+      "learning_rate": 1.3426688632619442e-05,
+      "loss": 0.0562,
+      "step": 515
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 0.6170294354031257,
+      "learning_rate": 1.3344316309719934e-05,
+      "loss": 0.0699,
+      "step": 516
+    },
+    {
+      "epoch": 2.293954520244038,
+      "grad_norm": 0.4461452327489392,
+      "learning_rate": 1.326194398682043e-05,
+      "loss": 0.0585,
+      "step": 517
+    },
+    {
+      "epoch": 2.298391569606212,
+      "grad_norm": 0.47529950612582456,
+      "learning_rate": 1.3179571663920923e-05,
+      "loss": 0.0743,
+      "step": 518
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 0.6245852524785371,
+      "learning_rate": 1.3097199341021419e-05,
+      "loss": 0.0913,
+      "step": 519
+    },
+    {
+      "epoch": 2.3072656683305603,
+      "grad_norm": 0.5049104959656808,
+      "learning_rate": 1.301482701812191e-05,
+      "loss": 0.0693,
+      "step": 520
+    },
+    {
+      "epoch": 2.3117027176927345,
+      "grad_norm": 0.5192151075769776,
+      "learning_rate": 1.2932454695222406e-05,
+      "loss": 0.063,
+      "step": 521
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 0.8438188030306165,
+      "learning_rate": 1.28500823723229e-05,
+      "loss": 0.0765,
+      "step": 522
+    },
+    {
+      "epoch": 2.320576816417083,
+      "grad_norm": 0.7458599183514235,
+      "learning_rate": 1.2767710049423395e-05,
+      "loss": 0.0883,
+      "step": 523
+    },
+    {
+      "epoch": 2.325013865779257,
+      "grad_norm": 0.4773691261151876,
+      "learning_rate": 1.2685337726523888e-05,
+      "loss": 0.0825,
+      "step": 524
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 0.6631184767996374,
+      "learning_rate": 1.2602965403624385e-05,
+      "loss": 0.0678,
+      "step": 525
+    },
+    {
+      "epoch": 2.3338879645036053,
+      "grad_norm": 0.48514902659274023,
+      "learning_rate": 1.2520593080724877e-05,
+      "loss": 0.0629,
+      "step": 526
+    },
+    {
+      "epoch": 2.3383250138657794,
+      "grad_norm": 0.6363169332495404,
+      "learning_rate": 1.243822075782537e-05,
+      "loss": 0.076,
+      "step": 527
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 0.39497748718207804,
+      "learning_rate": 1.2355848434925866e-05,
+      "loss": 0.082,
+      "step": 528
+    },
+    {
+      "epoch": 2.3471991125901277,
+      "grad_norm": 0.520740707426262,
+      "learning_rate": 1.227347611202636e-05,
+      "loss": 0.0696,
+      "step": 529
+    },
+    {
+      "epoch": 2.351636161952302,
+      "grad_norm": 0.46006806051497967,
+      "learning_rate": 1.2191103789126853e-05,
+      "loss": 0.0608,
+      "step": 530
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 0.450141042884803,
+      "learning_rate": 1.2108731466227347e-05,
+      "loss": 0.0719,
+      "step": 531
+    },
+    {
+      "epoch": 2.36051026067665,
+      "grad_norm": 0.4490820967102202,
+      "learning_rate": 1.2026359143327843e-05,
+      "loss": 0.0582,
+      "step": 532
+    },
+    {
+      "epoch": 2.3649473100388243,
+      "grad_norm": 0.3785612364255548,
+      "learning_rate": 1.1943986820428336e-05,
+      "loss": 0.0371,
+      "step": 533
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 0.597793712889077,
+      "learning_rate": 1.186161449752883e-05,
+      "loss": 0.0704,
+      "step": 534
+    },
+    {
+      "epoch": 2.3738214087631726,
+      "grad_norm": 0.5339908021820534,
+      "learning_rate": 1.1779242174629326e-05,
+      "loss": 0.064,
+      "step": 535
+    },
+    {
+      "epoch": 2.3782584581253468,
+      "grad_norm": 0.545507738200127,
+      "learning_rate": 1.169686985172982e-05,
+      "loss": 0.0823,
+      "step": 536
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 0.683046307023061,
+      "learning_rate": 1.1614497528830313e-05,
+      "loss": 0.0811,
+      "step": 537
+    },
+    {
+      "epoch": 2.387132556849695,
+      "grad_norm": 0.6888771352760809,
+      "learning_rate": 1.1532125205930807e-05,
+      "loss": 0.0437,
+      "step": 538
+    },
+    {
+      "epoch": 2.391569606211869,
+      "grad_norm": 0.4496631580203644,
+      "learning_rate": 1.1449752883031302e-05,
+      "loss": 0.078,
+      "step": 539
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 0.5088280426732856,
+      "learning_rate": 1.1367380560131796e-05,
+      "loss": 0.0561,
+      "step": 540
+    },
+    {
+      "epoch": 2.4004437049362175,
+      "grad_norm": 0.4440276029808717,
+      "learning_rate": 1.128500823723229e-05,
+      "loss": 0.0646,
+      "step": 541
+    },
+    {
+      "epoch": 2.4048807542983917,
+      "grad_norm": 0.4119316549510094,
+      "learning_rate": 1.1202635914332785e-05,
+      "loss": 0.0766,
+      "step": 542
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 0.4575063078607008,
+      "learning_rate": 1.112026359143328e-05,
+      "loss": 0.0566,
+      "step": 543
+    },
+    {
+      "epoch": 2.41375485302274,
+      "grad_norm": 0.5761313731295636,
+      "learning_rate": 1.1037891268533773e-05,
+      "loss": 0.0532,
+      "step": 544
+    },
+    {
+      "epoch": 2.418191902384914,
+      "grad_norm": 0.5218773265032963,
+      "learning_rate": 1.0955518945634267e-05,
+      "loss": 0.0776,
+      "step": 545
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 0.4300647466728647,
+      "learning_rate": 1.0873146622734762e-05,
+      "loss": 0.0613,
+      "step": 546
+    },
+    {
+      "epoch": 2.4270660011092624,
+      "grad_norm": 0.6984842374384344,
+      "learning_rate": 1.0790774299835256e-05,
+      "loss": 0.0872,
+      "step": 547
+    },
+    {
+      "epoch": 2.4315030504714366,
+      "grad_norm": 0.407745232277647,
+      "learning_rate": 1.070840197693575e-05,
+      "loss": 0.0655,
+      "step": 548
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 0.4468363238317229,
+      "learning_rate": 1.0626029654036245e-05,
+      "loss": 0.0545,
+      "step": 549
+    },
+    {
+      "epoch": 2.440377149195785,
+      "grad_norm": 0.41888501993897503,
+      "learning_rate": 1.0543657331136739e-05,
+      "loss": 0.0506,
+      "step": 550
+    },
+    {
+      "epoch": 2.444814198557959,
+      "grad_norm": 0.614258637298855,
+      "learning_rate": 1.0461285008237233e-05,
+      "loss": 0.0638,
+      "step": 551
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 0.5383878715788591,
+      "learning_rate": 1.0378912685337726e-05,
+      "loss": 0.0714,
+      "step": 552
+    },
+    {
+      "epoch": 2.4536882972823073,
+      "grad_norm": 0.49440474858341443,
+      "learning_rate": 1.0296540362438222e-05,
+      "loss": 0.0774,
+      "step": 553
+    },
+    {
+      "epoch": 2.4581253466444815,
+      "grad_norm": 0.5396107083695306,
+      "learning_rate": 1.0214168039538716e-05,
+      "loss": 0.0605,
+      "step": 554
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 0.6227835454493174,
+      "learning_rate": 1.013179571663921e-05,
+      "loss": 0.0591,
+      "step": 555
+    },
+    {
+      "epoch": 2.4669994453688298,
+      "grad_norm": 0.6709203813676109,
+      "learning_rate": 1.0049423393739705e-05,
+      "loss": 0.0636,
+      "step": 556
+    },
+    {
+      "epoch": 2.471436494731004,
+      "grad_norm": 0.884123107839097,
+      "learning_rate": 9.967051070840199e-06,
+      "loss": 0.0967,
+      "step": 557
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 0.557280216515346,
+      "learning_rate": 9.884678747940692e-06,
+      "loss": 0.0527,
+      "step": 558
+    },
+    {
+      "epoch": 2.480310593455352,
+      "grad_norm": 0.4521203578197039,
+      "learning_rate": 9.802306425041186e-06,
+      "loss": 0.0624,
+      "step": 559
+    },
+    {
+      "epoch": 2.4847476428175264,
+      "grad_norm": 0.47421592133383833,
+      "learning_rate": 9.719934102141682e-06,
+      "loss": 0.0638,
+      "step": 560
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 0.44367777948603515,
+      "learning_rate": 9.637561779242175e-06,
+      "loss": 0.0435,
+      "step": 561
+    },
+    {
+      "epoch": 2.4936217415418747,
+      "grad_norm": 0.5122972378060052,
+      "learning_rate": 9.55518945634267e-06,
+      "loss": 0.0616,
+      "step": 562
+    },
+    {
+      "epoch": 2.498058790904049,
+      "grad_norm": 0.6076303106035497,
+      "learning_rate": 9.472817133443165e-06,
+      "loss": 0.0891,
+      "step": 563
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 0.45183570543231943,
+      "learning_rate": 9.390444810543658e-06,
+      "loss": 0.0706,
+      "step": 564
+    },
+    {
+      "epoch": 2.506932889628397,
+      "grad_norm": 0.5822435030412974,
+      "learning_rate": 9.308072487644152e-06,
+      "loss": 0.0862,
+      "step": 565
+    },
+    {
+      "epoch": 2.5113699389905713,
+      "grad_norm": 0.34218349933462927,
+      "learning_rate": 9.225700164744646e-06,
+      "loss": 0.0488,
+      "step": 566
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 0.5334752544681166,
+      "learning_rate": 9.143327841845141e-06,
+      "loss": 0.0732,
+      "step": 567
+    },
+    {
+      "epoch": 2.5202440377149196,
+      "grad_norm": 0.7441077437858243,
+      "learning_rate": 9.060955518945635e-06,
+      "loss": 0.0945,
+      "step": 568
+    },
+    {
+      "epoch": 2.5246810870770937,
+      "grad_norm": 0.5803895364880287,
+      "learning_rate": 8.978583196046129e-06,
+      "loss": 0.0628,
+      "step": 569
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 0.7023908809340375,
+      "learning_rate": 8.896210873146624e-06,
+      "loss": 0.0808,
+      "step": 570
+    },
+    {
+      "epoch": 2.533555185801442,
+      "grad_norm": 0.7513922246654061,
+      "learning_rate": 8.813838550247118e-06,
+      "loss": 0.0675,
+      "step": 571
+    },
+    {
+      "epoch": 2.537992235163616,
+      "grad_norm": 0.5272545067896169,
+      "learning_rate": 8.731466227347612e-06,
+      "loss": 0.0663,
+      "step": 572
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 0.34520250916196227,
+      "learning_rate": 8.649093904448106e-06,
+      "loss": 0.0467,
+      "step": 573
+    },
+    {
+      "epoch": 2.5468663338879645,
+      "grad_norm": 0.5582395644470308,
+      "learning_rate": 8.566721581548601e-06,
+      "loss": 0.0945,
+      "step": 574
+    },
+    {
+      "epoch": 2.5513033832501386,
+      "grad_norm": 0.47035592281087324,
+      "learning_rate": 8.484349258649095e-06,
+      "loss": 0.0594,
+      "step": 575
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 0.5237242354563192,
+      "learning_rate": 8.401976935749589e-06,
+      "loss": 0.0719,
+      "step": 576
+    },
+    {
+      "epoch": 2.560177481974487,
+      "grad_norm": 0.6226672507813605,
+      "learning_rate": 8.319604612850082e-06,
+      "loss": 0.0737,
+      "step": 577
+    },
+    {
+      "epoch": 2.564614531336661,
+      "grad_norm": 0.6005362637454409,
+      "learning_rate": 8.237232289950576e-06,
+      "loss": 0.066,
+      "step": 578
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 0.8960722597267446,
+      "learning_rate": 8.15485996705107e-06,
+      "loss": 0.0523,
+      "step": 579
+    },
+    {
+      "epoch": 2.5734886300610094,
+      "grad_norm": 0.5358038554925384,
+      "learning_rate": 8.072487644151565e-06,
+      "loss": 0.0819,
+      "step": 580
+    },
+    {
+      "epoch": 2.5779256794231835,
+      "grad_norm": 0.6011761750085439,
+      "learning_rate": 7.99011532125206e-06,
+      "loss": 0.0983,
+      "step": 581
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 0.5709607346457392,
+      "learning_rate": 7.907742998352553e-06,
+      "loss": 0.0593,
+      "step": 582
+    },
+    {
+      "epoch": 2.586799778147532,
+      "grad_norm": 0.5049364662806708,
+      "learning_rate": 7.825370675453048e-06,
+      "loss": 0.0704,
+      "step": 583
+    },
+    {
+      "epoch": 2.591236827509706,
+      "grad_norm": 0.4517422988015621,
+      "learning_rate": 7.742998352553542e-06,
+      "loss": 0.0471,
+      "step": 584
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 0.6858449217380069,
+      "learning_rate": 7.660626029654036e-06,
+      "loss": 0.0919,
+      "step": 585
+    },
+    {
+      "epoch": 2.6001109262340543,
+      "grad_norm": 0.5095035566275075,
+      "learning_rate": 7.5782537067545305e-06,
+      "loss": 0.07,
+      "step": 586
+    },
+    {
+      "epoch": 2.6045479755962284,
+      "grad_norm": 0.40903359219934565,
+      "learning_rate": 7.495881383855024e-06,
+      "loss": 0.05,
+      "step": 587
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 0.6623676887680296,
+      "learning_rate": 7.413509060955519e-06,
+      "loss": 0.0607,
+      "step": 588
+    },
+    {
+      "epoch": 2.6134220743205767,
+      "grad_norm": 0.44794662138853475,
+      "learning_rate": 7.3311367380560135e-06,
+      "loss": 0.0428,
+      "step": 589
+    },
+    {
+      "epoch": 2.617859123682751,
+      "grad_norm": 0.4648021230571587,
+      "learning_rate": 7.248764415156507e-06,
+      "loss": 0.0683,
+      "step": 590
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 0.41227107920751704,
+      "learning_rate": 7.166392092257002e-06,
+      "loss": 0.0601,
+      "step": 591
+    },
+    {
+      "epoch": 2.626733222407099,
+      "grad_norm": 0.682180480536477,
+      "learning_rate": 7.084019769357496e-06,
+      "loss": 0.061,
+      "step": 592
+    },
+    {
+      "epoch": 2.6311702717692733,
+      "grad_norm": 0.4296358597675649,
+      "learning_rate": 7.00164744645799e-06,
+      "loss": 0.059,
+      "step": 593
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 0.39973459230303815,
+      "learning_rate": 6.919275123558484e-06,
+      "loss": 0.0422,
+      "step": 594
+    },
+    {
+      "epoch": 2.6400443704936216,
+      "grad_norm": 0.3776539035356485,
+      "learning_rate": 6.836902800658979e-06,
+      "loss": 0.0564,
+      "step": 595
+    },
+    {
+      "epoch": 2.6444814198557958,
+      "grad_norm": 0.448340677315833,
+      "learning_rate": 6.754530477759473e-06,
+      "loss": 0.0733,
+      "step": 596
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 0.44326019858338933,
+      "learning_rate": 6.672158154859967e-06,
+      "loss": 0.0585,
+      "step": 597
+    },
+    {
+      "epoch": 2.653355518580144,
+      "grad_norm": 0.4393932216930158,
+      "learning_rate": 6.589785831960462e-06,
+      "loss": 0.0585,
+      "step": 598
+    },
+    {
+      "epoch": 2.657792567942318,
+      "grad_norm": 0.6003175957864102,
+      "learning_rate": 6.507413509060955e-06,
+      "loss": 0.0578,
+      "step": 599
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 0.6888817210512455,
+      "learning_rate": 6.42504118616145e-06,
+      "loss": 0.0622,
+      "step": 600
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5133451866147913,
+      "learning_rate": 6.342668863261944e-06,
+      "loss": 0.085,
+      "step": 601
+    },
+    {
+      "epoch": 2.6711037160288407,
+      "grad_norm": 0.6264705999619853,
+      "learning_rate": 6.260296540362438e-06,
+      "loss": 0.0774,
+      "step": 602
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 0.5217647870120197,
+      "learning_rate": 6.177924217462933e-06,
+      "loss": 0.0605,
+      "step": 603
+    },
+    {
+      "epoch": 2.679977814753189,
+      "grad_norm": 0.5906476738023915,
+      "learning_rate": 6.095551894563427e-06,
+      "loss": 0.0725,
+      "step": 604
+    },
+    {
+      "epoch": 2.684414864115363,
+      "grad_norm": 0.5933742338006437,
+      "learning_rate": 6.013179571663921e-06,
+      "loss": 0.0651,
+      "step": 605
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 0.4941534410103864,
+      "learning_rate": 5.930807248764415e-06,
+      "loss": 0.0642,
+      "step": 606
+    },
+    {
+      "epoch": 2.693288962839712,
+      "grad_norm": 0.5281822602419853,
+      "learning_rate": 5.84843492586491e-06,
+      "loss": 0.0638,
+      "step": 607
+    },
+    {
+      "epoch": 2.697726012201886,
+      "grad_norm": 0.7662992816902268,
+      "learning_rate": 5.7660626029654035e-06,
+      "loss": 0.0564,
+      "step": 608
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 0.4372832919599054,
+      "learning_rate": 5.683690280065898e-06,
+      "loss": 0.0352,
+      "step": 609
+    },
+    {
+      "epoch": 2.7066001109262343,
+      "grad_norm": 0.5487768889163155,
+      "learning_rate": 5.601317957166393e-06,
+      "loss": 0.0675,
+      "step": 610
+    },
+    {
+      "epoch": 2.7110371602884085,
+      "grad_norm": 0.4992506103616956,
+      "learning_rate": 5.5189456342668865e-06,
+      "loss": 0.0675,
+      "step": 611
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 0.5265814463662151,
+      "learning_rate": 5.436573311367381e-06,
+      "loss": 0.0737,
+      "step": 612
+    },
+    {
+      "epoch": 2.7199112590127568,
+      "grad_norm": 0.49921904610160917,
+      "learning_rate": 5.354200988467875e-06,
+      "loss": 0.0367,
+      "step": 613
+    },
+    {
+      "epoch": 2.724348308374931,
+      "grad_norm": 0.5023294366722539,
+      "learning_rate": 5.2718286655683695e-06,
+      "loss": 0.0661,
+      "step": 614
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 0.585411853250103,
+      "learning_rate": 5.189456342668863e-06,
+      "loss": 0.0718,
+      "step": 615
+    },
+    {
+      "epoch": 2.733222407099279,
+      "grad_norm": 0.46309840693552984,
+      "learning_rate": 5.107084019769358e-06,
+      "loss": 0.0589,
+      "step": 616
+    },
+    {
+      "epoch": 2.7376594564614534,
+      "grad_norm": 0.42236199351208137,
+      "learning_rate": 5.0247116968698524e-06,
+      "loss": 0.0442,
+      "step": 617
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 0.6546125646782767,
+      "learning_rate": 4.942339373970346e-06,
+      "loss": 0.0558,
+      "step": 618
+    },
+    {
+      "epoch": 2.7465335551858017,
+      "grad_norm": 0.8323427432114843,
+      "learning_rate": 4.859967051070841e-06,
+      "loss": 0.0953,
+      "step": 619
+    },
+    {
+      "epoch": 2.750970604547976,
+      "grad_norm": 0.5960678234936589,
+      "learning_rate": 4.777594728171335e-06,
+      "loss": 0.0846,
+      "step": 620
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 0.6092122667608001,
+      "learning_rate": 4.695222405271829e-06,
+      "loss": 0.0847,
+      "step": 621
+    },
+    {
+      "epoch": 2.759844703272324,
+      "grad_norm": 0.48532761451387557,
+      "learning_rate": 4.612850082372323e-06,
+      "loss": 0.0688,
+      "step": 622
+    },
+    {
+      "epoch": 2.7642817526344983,
+      "grad_norm": 0.4948244150935661,
+      "learning_rate": 4.5304777594728176e-06,
+      "loss": 0.0752,
+      "step": 623
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 0.5544094678362775,
+      "learning_rate": 4.448105436573312e-06,
+      "loss": 0.0774,
+      "step": 624
+    },
+    {
+      "epoch": 2.7731558513588466,
+      "grad_norm": 0.6077609572338848,
+      "learning_rate": 4.365733113673806e-06,
+      "loss": 0.0862,
+      "step": 625
+    },
+    {
+      "epoch": 2.7775929007210207,
+      "grad_norm": 0.5485551637243942,
+      "learning_rate": 4.2833607907743006e-06,
+      "loss": 0.0688,
+      "step": 626
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 0.6388005118357853,
+      "learning_rate": 4.200988467874794e-06,
+      "loss": 0.0735,
+      "step": 627
+    },
+    {
+      "epoch": 2.786466999445369,
+      "grad_norm": 0.45218177161710815,
+      "learning_rate": 4.118616144975288e-06,
+      "loss": 0.0545,
+      "step": 628
+    },
+    {
+      "epoch": 2.790904048807543,
+      "grad_norm": 0.4346717630330878,
+      "learning_rate": 4.036243822075783e-06,
+      "loss": 0.0505,
+      "step": 629
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 0.5175320298883663,
+      "learning_rate": 3.9538714991762765e-06,
+      "loss": 0.0534,
+      "step": 630
+    },
+    {
+      "epoch": 2.7997781475318915,
+      "grad_norm": 0.44823500683586665,
+      "learning_rate": 3.871499176276771e-06,
+      "loss": 0.0517,
+      "step": 631
+    },
+    {
+      "epoch": 2.8042151968940656,
+      "grad_norm": 0.350213801698505,
+      "learning_rate": 3.7891268533772653e-06,
+      "loss": 0.051,
+      "step": 632
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 0.6663699731646126,
+      "learning_rate": 3.7067545304777594e-06,
+      "loss": 0.0942,
+      "step": 633
+    },
+    {
+      "epoch": 2.813089295618414,
+      "grad_norm": 0.6932216453765204,
+      "learning_rate": 3.6243822075782536e-06,
+      "loss": 0.1499,
+      "step": 634
+    },
+    {
+      "epoch": 2.817526344980588,
+      "grad_norm": 0.4397429125368117,
+      "learning_rate": 3.542009884678748e-06,
+      "loss": 0.0616,
+      "step": 635
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 0.48755041742041916,
+      "learning_rate": 3.459637561779242e-06,
+      "loss": 0.056,
+      "step": 636
+    },
+    {
+      "epoch": 2.8264004437049364,
+      "grad_norm": 0.4803019598534695,
+      "learning_rate": 3.3772652388797366e-06,
+      "loss": 0.046,
+      "step": 637
+    },
+    {
+      "epoch": 2.8308374930671105,
+      "grad_norm": 0.48878390570205,
+      "learning_rate": 3.294892915980231e-06,
+      "loss": 0.0455,
+      "step": 638
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 0.4355736625884251,
+      "learning_rate": 3.212520593080725e-06,
+      "loss": 0.0513,
+      "step": 639
+    },
+    {
+      "epoch": 2.839711591791459,
+      "grad_norm": 0.4806382415296703,
+      "learning_rate": 3.130148270181219e-06,
+      "loss": 0.0691,
+      "step": 640
+    },
+    {
+      "epoch": 2.844148641153633,
+      "grad_norm": 0.7085503544920213,
+      "learning_rate": 3.0477759472817134e-06,
+      "loss": 0.092,
+      "step": 641
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 0.4539527895470607,
+      "learning_rate": 2.9654036243822076e-06,
+      "loss": 0.0771,
+      "step": 642
+    },
+    {
+      "epoch": 2.8530227398779813,
+      "grad_norm": 0.4315965379665383,
+      "learning_rate": 2.8830313014827017e-06,
+      "loss": 0.0546,
+      "step": 643
+    },
+    {
+      "epoch": 2.8574597892401554,
+      "grad_norm": 0.5233086384678962,
+      "learning_rate": 2.8006589785831964e-06,
+      "loss": 0.0658,
+      "step": 644
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 0.6225381047486279,
+      "learning_rate": 2.7182866556836905e-06,
+      "loss": 0.0572,
+      "step": 645
+    },
+    {
+      "epoch": 2.8663338879645037,
+      "grad_norm": 0.7265111023145802,
+      "learning_rate": 2.6359143327841847e-06,
+      "loss": 0.0801,
+      "step": 646
+    },
+    {
+      "epoch": 2.870770937326678,
+      "grad_norm": 0.478098847415718,
+      "learning_rate": 2.553542009884679e-06,
+      "loss": 0.0653,
+      "step": 647
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 0.46170161646311497,
+      "learning_rate": 2.471169686985173e-06,
+      "loss": 0.0676,
+      "step": 648
+    },
+    {
+      "epoch": 2.879645036051026,
+      "grad_norm": 0.5335652166420076,
+      "learning_rate": 2.3887973640856673e-06,
+      "loss": 0.051,
+      "step": 649
+    },
+    {
+      "epoch": 2.8840820854132003,
+      "grad_norm": 0.43710767662120653,
+      "learning_rate": 2.3064250411861615e-06,
+      "loss": 0.065,
+      "step": 650
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 0.593841244563058,
+      "learning_rate": 2.224052718286656e-06,
+      "loss": 0.0567,
+      "step": 651
+    },
+    {
+      "epoch": 2.8929561841375486,
+      "grad_norm": 0.4648433332681791,
+      "learning_rate": 2.1416803953871503e-06,
+      "loss": 0.0609,
+      "step": 652
+    },
+    {
+      "epoch": 2.8973932334997228,
+      "grad_norm": 0.6244898382891827,
+      "learning_rate": 2.059308072487644e-06,
+      "loss": 0.0546,
+      "step": 653
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 0.4721974124032116,
+      "learning_rate": 1.9769357495881382e-06,
+      "loss": 0.0655,
+      "step": 654
+    },
+    {
+      "epoch": 2.906267332224071,
+      "grad_norm": 0.4897283701099957,
+      "learning_rate": 1.8945634266886326e-06,
+      "loss": 0.0615,
+      "step": 655
+    },
+    {
+      "epoch": 2.9107043815862452,
+      "grad_norm": 0.4779585729427744,
+      "learning_rate": 1.8121911037891268e-06,
+      "loss": 0.0584,
+      "step": 656
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 0.6361478340546802,
+      "learning_rate": 1.729818780889621e-06,
+      "loss": 0.0599,
+      "step": 657
+    },
+    {
+      "epoch": 2.9195784803105935,
+      "grad_norm": 0.5243759283482696,
+      "learning_rate": 1.6474464579901154e-06,
+      "loss": 0.0605,
+      "step": 658
+    },
+    {
+      "epoch": 2.9240155296727677,
+      "grad_norm": 0.546129345040773,
+      "learning_rate": 1.5650741350906096e-06,
+      "loss": 0.0705,
+      "step": 659
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 0.4469212701491516,
+      "learning_rate": 1.4827018121911038e-06,
+      "loss": 0.0598,
+      "step": 660
+    },
+    {
+      "epoch": 2.932889628397116,
+      "grad_norm": 0.5182427511559341,
+      "learning_rate": 1.4003294892915982e-06,
+      "loss": 0.0606,
+      "step": 661
+    },
+    {
+      "epoch": 2.93732667775929,
+      "grad_norm": 0.4125175985281365,
+      "learning_rate": 1.3179571663920924e-06,
+      "loss": 0.0569,
+      "step": 662
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 0.4624383382674929,
+      "learning_rate": 1.2355848434925866e-06,
+      "loss": 0.0631,
+      "step": 663
+    },
+    {
+      "epoch": 2.9462007764836384,
+      "grad_norm": 0.6619390758156715,
+      "learning_rate": 1.1532125205930807e-06,
+      "loss": 0.0629,
+      "step": 664
+    },
+    {
+      "epoch": 2.9506378258458126,
+      "grad_norm": 0.4984503364703818,
+      "learning_rate": 1.0708401976935751e-06,
+      "loss": 0.0575,
+      "step": 665
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 0.4221022311828792,
+      "learning_rate": 9.884678747940691e-07,
+      "loss": 0.0546,
+      "step": 666
+    },
+    {
+      "epoch": 2.959511924570161,
+      "grad_norm": 0.812292219150222,
+      "learning_rate": 9.060955518945634e-07,
+      "loss": 0.0711,
+      "step": 667
+    },
+    {
+      "epoch": 2.963948973932335,
+      "grad_norm": 0.5341856849794002,
+      "learning_rate": 8.237232289950577e-07,
+      "loss": 0.0639,
+      "step": 668
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 0.36564535589692126,
+      "learning_rate": 7.413509060955519e-07,
+      "loss": 0.0495,
+      "step": 669
+    },
+    {
+      "epoch": 2.9728230726566833,
+      "grad_norm": 0.5936093916793973,
+      "learning_rate": 6.589785831960462e-07,
+      "loss": 0.0703,
+      "step": 670
+    },
+    {
+      "epoch": 2.9772601220188575,
+      "grad_norm": 0.4459384400665918,
+      "learning_rate": 5.766062602965404e-07,
+      "loss": 0.0477,
+      "step": 671
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 0.5018832384433306,
+      "learning_rate": 4.942339373970346e-07,
+      "loss": 0.0561,
+      "step": 672
+    },
+    {
+      "epoch": 2.986134220743206,
+      "grad_norm": 0.582965689563923,
+      "learning_rate": 4.1186161449752885e-07,
+      "loss": 0.0746,
+      "step": 673
+    },
+    {
+      "epoch": 2.99057127010538,
+      "grad_norm": 0.9209740325547386,
+      "learning_rate": 3.294892915980231e-07,
+      "loss": 0.075,
+      "step": 674
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 0.5632266014139005,
+      "learning_rate": 2.471169686985173e-07,
+      "loss": 0.0554,
+      "step": 675
+    },
+    {
+      "epoch": 2.995008319467554,
+      "step": 675,
+      "total_flos": 116490238033920.0,
+      "train_loss": 0.36549280296873166,
+      "train_runtime": 96569.5267,
+      "train_samples_per_second": 0.224,
+      "train_steps_per_second": 0.007
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 675,
+  "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": 116490238033920.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}