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": 2.9966722129783694,
+  "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": 0.0,
+      "learning_rate": 0,
+      "loss": 0.6446,
+      "step": 1
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.608,
+      "step": 2
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.5698,
+      "step": 3
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.6129,
+      "step": 4
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.65,
+      "step": 5
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.7315,
+      "step": 6
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.7678,
+      "step": 7
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 0.0,
+      "learning_rate": 0,
+      "loss": 0.6479,
+      "step": 8
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 985.1844482421875,
+      "learning_rate": 0.0,
+      "loss": 0.627,
+      "step": 9
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 985.1844482421875,
+      "learning_rate": 0.0,
+      "loss": 0.9046,
+      "step": 10
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 985.1844482421875,
+      "learning_rate": 0.0,
+      "loss": 0.9398,
+      "step": 11
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 985.1844482421875,
+      "learning_rate": 0.0,
+      "loss": 0.9316,
+      "step": 12
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 3791.072509765625,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 0.9978,
+      "step": 13
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 6438.296875,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 0.9518,
+      "step": 14
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 6438.296875,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 0.8972,
+      "step": 15
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 5852.2099609375,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 0.9943,
+      "step": 16
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 5731.96142578125,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.9127,
+      "step": 17
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 5856.81689453125,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 0.7472,
+      "step": 18
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 2891.84814453125,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 0.8537,
+      "step": 19
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 1052.528076171875,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 0.7046,
+      "step": 20
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 676.8657836914062,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.6422,
+      "step": 21
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 1723.6614990234375,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 0.6902,
+      "step": 22
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 4462.20703125,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 0.7075,
+      "step": 23
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 2511.7978515625,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 0.6761,
+      "step": 24
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 4069.682373046875,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.7377,
+      "step": 25
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 4274.97998046875,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 0.7237,
+      "step": 26
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 2575.159912109375,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 0.7748,
+      "step": 27
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 3277.528564453125,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 0.5484,
+      "step": 28
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 1533.4759521484375,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.6053,
+      "step": 29
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 1472.3345947265625,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5325,
+      "step": 30
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 3179.1162109375,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 0.6581,
+      "step": 31
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 5084.49462890625,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 0.6065,
+      "step": 32
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 1920.1943359375,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.648,
+      "step": 33
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 3079.337646484375,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 0.5473,
+      "step": 34
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 6452.27294921875,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.6385,
+      "step": 35
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 2025.9248046875,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 0.5735,
+      "step": 36
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 2414.412841796875,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.5707,
+      "step": 37
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 2923.263427734375,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 0.7164,
+      "step": 38
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 1430.942626953125,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 0.6238,
+      "step": 39
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 1992.25732421875,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 0.6567,
+      "step": 40
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 3866.493408203125,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 0.6258,
+      "step": 41
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 3555.39013671875,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 0.5909,
+      "step": 42
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 806.6991577148438,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 0.5843,
+      "step": 43
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 848.513916015625,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 0.6521,
+      "step": 44
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 2022.592041015625,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.6524,
+      "step": 45
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 4125.48779296875,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 0.736,
+      "step": 46
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 1471.8026123046875,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7231,
+      "step": 47
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 12320.845703125,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 0.7395,
+      "step": 48
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 1674.9398193359375,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.5851,
+      "step": 49
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 1337.0648193359375,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 0.7459,
+      "step": 50
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 3396.963623046875,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 0.5984,
+      "step": 51
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 4167.63525390625,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 0.648,
+      "step": 52
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 776.7545166015625,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.5785,
+      "step": 53
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 1553.5216064453125,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 0.6041,
+      "step": 54
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 1646.2296142578125,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 0.7255,
+      "step": 55
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 757.6424560546875,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 0.5279,
+      "step": 56
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 821.070556640625,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 0.7083,
+      "step": 57
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 1315.488037109375,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 0.6612,
+      "step": 58
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 2176.86669921875,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 0.6532,
+      "step": 59
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 1099.84521484375,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 0.8234,
+      "step": 60
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 4432.42822265625,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.6966,
+      "step": 61
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 1902.2816162109375,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 0.6414,
+      "step": 62
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 1139.3065185546875,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 0.6221,
+      "step": 63
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 1309.5606689453125,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.6014,
+      "step": 64
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 2304.3125,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 0.6888,
+      "step": 65
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 2410.008544921875,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 0.6205,
+      "step": 66
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 1492.7923583984375,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 0.6922,
+      "step": 67
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 5913.8779296875,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 0.5485,
+      "step": 68
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 3758.1943359375,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.7329,
+      "step": 69
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 1962.16259765625,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 0.6837,
+      "step": 70
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 3625.859375,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 0.6277,
+      "step": 71
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 2340.0546875,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 0.6523,
+      "step": 72
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 2891.10986328125,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.6439,
+      "step": 73
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 528.99853515625,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 0.5562,
+      "step": 74
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 3599.318603515625,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 0.6656,
+      "step": 75
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 1144.25048828125,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 0.6646,
+      "step": 76
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 6799.154296875,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.7175,
+      "step": 77
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 1644.26171875,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 0.7234,
+      "step": 78
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 1231.1402587890625,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 0.6788,
+      "step": 79
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 6014.7138671875,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 0.5721,
+      "step": 80
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 412.0801696777344,
+      "learning_rate": 5e-05,
+      "loss": 0.7156,
+      "step": 81
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 4178.47412109375,
+      "learning_rate": 4.99176276771005e-05,
+      "loss": 0.6006,
+      "step": 82
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 1196.9801025390625,
+      "learning_rate": 4.9835255354200987e-05,
+      "loss": 0.6413,
+      "step": 83
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 1778.488037109375,
+      "learning_rate": 4.975288303130148e-05,
+      "loss": 0.6398,
+      "step": 84
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 2187.544677734375,
+      "learning_rate": 4.967051070840198e-05,
+      "loss": 0.638,
+      "step": 85
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 2087.809326171875,
+      "learning_rate": 4.958813838550247e-05,
+      "loss": 0.6895,
+      "step": 86
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 4670.61474609375,
+      "learning_rate": 4.950576606260297e-05,
+      "loss": 0.7886,
+      "step": 87
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 3102.271728515625,
+      "learning_rate": 4.9423393739703464e-05,
+      "loss": 0.6897,
+      "step": 88
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 1334.5616455078125,
+      "learning_rate": 4.934102141680396e-05,
+      "loss": 0.7002,
+      "step": 89
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 2719.68359375,
+      "learning_rate": 4.9258649093904455e-05,
+      "loss": 0.7352,
+      "step": 90
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 2433.34912109375,
+      "learning_rate": 4.9176276771004943e-05,
+      "loss": 0.6917,
+      "step": 91
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 3023.9150390625,
+      "learning_rate": 4.909390444810544e-05,
+      "loss": 0.6818,
+      "step": 92
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 1653.8729248046875,
+      "learning_rate": 4.9011532125205934e-05,
+      "loss": 0.6679,
+      "step": 93
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 2969.647216796875,
+      "learning_rate": 4.892915980230643e-05,
+      "loss": 0.6882,
+      "step": 94
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 910.2950439453125,
+      "learning_rate": 4.884678747940692e-05,
+      "loss": 0.7884,
+      "step": 95
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 2126.17236328125,
+      "learning_rate": 4.8764415156507414e-05,
+      "loss": 0.6489,
+      "step": 96
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 1509.8497314453125,
+      "learning_rate": 4.868204283360791e-05,
+      "loss": 0.6107,
+      "step": 97
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 1818.816650390625,
+      "learning_rate": 4.8599670510708405e-05,
+      "loss": 0.7265,
+      "step": 98
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 830.9838256835938,
+      "learning_rate": 4.8517298187808894e-05,
+      "loss": 0.7198,
+      "step": 99
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 2239.420654296875,
+      "learning_rate": 4.843492586490939e-05,
+      "loss": 0.7251,
+      "step": 100
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 2107.39501953125,
+      "learning_rate": 4.8352553542009884e-05,
+      "loss": 0.6304,
+      "step": 101
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 2922.1005859375,
+      "learning_rate": 4.827018121911038e-05,
+      "loss": 0.8164,
+      "step": 102
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 3320.2626953125,
+      "learning_rate": 4.8187808896210875e-05,
+      "loss": 0.7315,
+      "step": 103
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 1545.265625,
+      "learning_rate": 4.810543657331137e-05,
+      "loss": 0.6998,
+      "step": 104
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 1031.28125,
+      "learning_rate": 4.8023064250411866e-05,
+      "loss": 0.6325,
+      "step": 105
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 2164.244873046875,
+      "learning_rate": 4.794069192751236e-05,
+      "loss": 0.6603,
+      "step": 106
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 3971.90869140625,
+      "learning_rate": 4.785831960461285e-05,
+      "loss": 0.6953,
+      "step": 107
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 2538.948486328125,
+      "learning_rate": 4.7775947281713346e-05,
+      "loss": 0.666,
+      "step": 108
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 2034.2662353515625,
+      "learning_rate": 4.769357495881384e-05,
+      "loss": 0.6555,
+      "step": 109
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 2194.614013671875,
+      "learning_rate": 4.761120263591434e-05,
+      "loss": 0.6283,
+      "step": 110
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 1166.9786376953125,
+      "learning_rate": 4.7528830313014825e-05,
+      "loss": 0.7168,
+      "step": 111
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 1481.7020263671875,
+      "learning_rate": 4.744645799011532e-05,
+      "loss": 0.7567,
+      "step": 112
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 1393.0477294921875,
+      "learning_rate": 4.7364085667215816e-05,
+      "loss": 0.6893,
+      "step": 113
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 2169.02001953125,
+      "learning_rate": 4.728171334431631e-05,
+      "loss": 0.7607,
+      "step": 114
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 2963.00634765625,
+      "learning_rate": 4.719934102141681e-05,
+      "loss": 0.7069,
+      "step": 115
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 933.7859497070312,
+      "learning_rate": 4.71169686985173e-05,
+      "loss": 0.6205,
+      "step": 116
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 653.1210327148438,
+      "learning_rate": 4.70345963756178e-05,
+      "loss": 0.6556,
+      "step": 117
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 1581.3846435546875,
+      "learning_rate": 4.6952224052718294e-05,
+      "loss": 0.5847,
+      "step": 118
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 4526.40771484375,
+      "learning_rate": 4.686985172981878e-05,
+      "loss": 0.694,
+      "step": 119
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 1564.104248046875,
+      "learning_rate": 4.678747940691928e-05,
+      "loss": 0.5909,
+      "step": 120
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 1162.020263671875,
+      "learning_rate": 4.670510708401977e-05,
+      "loss": 0.6552,
+      "step": 121
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 2731.7685546875,
+      "learning_rate": 4.662273476112027e-05,
+      "loss": 0.6151,
+      "step": 122
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 4384.37890625,
+      "learning_rate": 4.654036243822076e-05,
+      "loss": 0.6345,
+      "step": 123
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 1888.7584228515625,
+      "learning_rate": 4.645799011532125e-05,
+      "loss": 0.5943,
+      "step": 124
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 960.8678588867188,
+      "learning_rate": 4.637561779242175e-05,
+      "loss": 0.5883,
+      "step": 125
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 3856.2646484375,
+      "learning_rate": 4.6293245469522244e-05,
+      "loss": 0.6573,
+      "step": 126
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 1998.6790771484375,
+      "learning_rate": 4.621087314662273e-05,
+      "loss": 0.7223,
+      "step": 127
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 2028.89453125,
+      "learning_rate": 4.612850082372323e-05,
+      "loss": 0.6032,
+      "step": 128
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 2696.942138671875,
+      "learning_rate": 4.6046128500823723e-05,
+      "loss": 0.6947,
+      "step": 129
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 1404.772705078125,
+      "learning_rate": 4.596375617792422e-05,
+      "loss": 0.6662,
+      "step": 130
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 1059.7689208984375,
+      "learning_rate": 4.5881383855024714e-05,
+      "loss": 0.649,
+      "step": 131
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 2828.816162109375,
+      "learning_rate": 4.579901153212521e-05,
+      "loss": 0.6374,
+      "step": 132
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 2688.14697265625,
+      "learning_rate": 4.5716639209225705e-05,
+      "loss": 0.6867,
+      "step": 133
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 3709.8515625,
+      "learning_rate": 4.56342668863262e-05,
+      "loss": 0.7148,
+      "step": 134
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 1540.975830078125,
+      "learning_rate": 4.555189456342669e-05,
+      "loss": 0.7632,
+      "step": 135
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 2370.242919921875,
+      "learning_rate": 4.5469522240527185e-05,
+      "loss": 0.6181,
+      "step": 136
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 8659.658203125,
+      "learning_rate": 4.538714991762768e-05,
+      "loss": 0.7432,
+      "step": 137
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 922.9805908203125,
+      "learning_rate": 4.5304777594728176e-05,
+      "loss": 0.5984,
+      "step": 138
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 2811.553955078125,
+      "learning_rate": 4.5222405271828664e-05,
+      "loss": 0.6318,
+      "step": 139
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 2402.760986328125,
+      "learning_rate": 4.514003294892916e-05,
+      "loss": 0.5695,
+      "step": 140
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 1656.7237548828125,
+      "learning_rate": 4.5057660626029655e-05,
+      "loss": 0.5897,
+      "step": 141
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 1129.016357421875,
+      "learning_rate": 4.497528830313015e-05,
+      "loss": 0.5077,
+      "step": 142
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 2030.2449951171875,
+      "learning_rate": 4.4892915980230646e-05,
+      "loss": 0.6291,
+      "step": 143
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 2495.515869140625,
+      "learning_rate": 4.481054365733114e-05,
+      "loss": 0.6125,
+      "step": 144
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 2553.15283203125,
+      "learning_rate": 4.472817133443164e-05,
+      "loss": 0.7118,
+      "step": 145
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 1253.3089599609375,
+      "learning_rate": 4.464579901153213e-05,
+      "loss": 0.7211,
+      "step": 146
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 273.9736022949219,
+      "learning_rate": 4.456342668863262e-05,
+      "loss": 0.6347,
+      "step": 147
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 1711.0601806640625,
+      "learning_rate": 4.448105436573312e-05,
+      "loss": 0.6328,
+      "step": 148
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 2826.65380859375,
+      "learning_rate": 4.439868204283361e-05,
+      "loss": 0.5791,
+      "step": 149
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 2472.78515625,
+      "learning_rate": 4.43163097199341e-05,
+      "loss": 0.8158,
+      "step": 150
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 2284.592529296875,
+      "learning_rate": 4.4233937397034596e-05,
+      "loss": 0.6566,
+      "step": 151
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 476.1126708984375,
+      "learning_rate": 4.415156507413509e-05,
+      "loss": 0.5967,
+      "step": 152
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 954.8773193359375,
+      "learning_rate": 4.406919275123559e-05,
+      "loss": 0.6395,
+      "step": 153
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 3078.819580078125,
+      "learning_rate": 4.3986820428336076e-05,
+      "loss": 0.7615,
+      "step": 154
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 8495.021484375,
+      "learning_rate": 4.390444810543657e-05,
+      "loss": 0.6093,
+      "step": 155
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 4457.1103515625,
+      "learning_rate": 4.382207578253707e-05,
+      "loss": 0.7578,
+      "step": 156
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 2622.5693359375,
+      "learning_rate": 4.373970345963756e-05,
+      "loss": 0.7021,
+      "step": 157
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 4072.7666015625,
+      "learning_rate": 4.365733113673806e-05,
+      "loss": 0.6512,
+      "step": 158
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 2668.179443359375,
+      "learning_rate": 4.357495881383855e-05,
+      "loss": 0.5573,
+      "step": 159
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 3620.03857421875,
+      "learning_rate": 4.349258649093905e-05,
+      "loss": 0.7856,
+      "step": 160
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 3201.927978515625,
+      "learning_rate": 4.3410214168039544e-05,
+      "loss": 0.6921,
+      "step": 161
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 3264.18896484375,
+      "learning_rate": 4.332784184514003e-05,
+      "loss": 0.7532,
+      "step": 162
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 1550.6214599609375,
+      "learning_rate": 4.324546952224053e-05,
+      "loss": 0.6269,
+      "step": 163
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 1520.86279296875,
+      "learning_rate": 4.3163097199341024e-05,
+      "loss": 0.7289,
+      "step": 164
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 2983.39306640625,
+      "learning_rate": 4.308072487644152e-05,
+      "loss": 0.6573,
+      "step": 165
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 3470.577880859375,
+      "learning_rate": 4.299835255354201e-05,
+      "loss": 0.7208,
+      "step": 166
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 885.436767578125,
+      "learning_rate": 4.29159802306425e-05,
+      "loss": 0.6361,
+      "step": 167
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 1379.3741455078125,
+      "learning_rate": 4.2833607907743e-05,
+      "loss": 0.7297,
+      "step": 168
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 1982.2476806640625,
+      "learning_rate": 4.2751235584843494e-05,
+      "loss": 0.7277,
+      "step": 169
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 875.0794067382812,
+      "learning_rate": 4.266886326194399e-05,
+      "loss": 0.7361,
+      "step": 170
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 2365.578857421875,
+      "learning_rate": 4.2586490939044485e-05,
+      "loss": 0.6047,
+      "step": 171
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 1938.486572265625,
+      "learning_rate": 4.250411861614498e-05,
+      "loss": 0.6051,
+      "step": 172
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 3185.165283203125,
+      "learning_rate": 4.2421746293245476e-05,
+      "loss": 0.6951,
+      "step": 173
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 1933.4111328125,
+      "learning_rate": 4.2339373970345965e-05,
+      "loss": 0.5925,
+      "step": 174
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 2535.295654296875,
+      "learning_rate": 4.225700164744646e-05,
+      "loss": 0.7072,
+      "step": 175
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 701.6828002929688,
+      "learning_rate": 4.2174629324546956e-05,
+      "loss": 0.5946,
+      "step": 176
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 3727.28662109375,
+      "learning_rate": 4.209225700164745e-05,
+      "loss": 0.6708,
+      "step": 177
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 3325.718505859375,
+      "learning_rate": 4.200988467874794e-05,
+      "loss": 0.7776,
+      "step": 178
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 3831.520751953125,
+      "learning_rate": 4.1927512355848435e-05,
+      "loss": 0.6708,
+      "step": 179
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 1416.1466064453125,
+      "learning_rate": 4.184514003294893e-05,
+      "loss": 0.7087,
+      "step": 180
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 7736.84521484375,
+      "learning_rate": 4.1762767710049426e-05,
+      "loss": 0.5992,
+      "step": 181
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 2687.558349609375,
+      "learning_rate": 4.1680395387149915e-05,
+      "loss": 0.7365,
+      "step": 182
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 1648.83056640625,
+      "learning_rate": 4.159802306425041e-05,
+      "loss": 0.6928,
+      "step": 183
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 1713.8175048828125,
+      "learning_rate": 4.1515650741350906e-05,
+      "loss": 0.7601,
+      "step": 184
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 1627.3001708984375,
+      "learning_rate": 4.14332784184514e-05,
+      "loss": 0.7547,
+      "step": 185
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 1992.2027587890625,
+      "learning_rate": 4.13509060955519e-05,
+      "loss": 0.6496,
+      "step": 186
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 3589.783447265625,
+      "learning_rate": 4.126853377265239e-05,
+      "loss": 0.6458,
+      "step": 187
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 1722.544677734375,
+      "learning_rate": 4.118616144975289e-05,
+      "loss": 0.7085,
+      "step": 188
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 3329.778564453125,
+      "learning_rate": 4.110378912685338e-05,
+      "loss": 0.8311,
+      "step": 189
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 5714.1708984375,
+      "learning_rate": 4.102141680395387e-05,
+      "loss": 0.6773,
+      "step": 190
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 2157.65576171875,
+      "learning_rate": 4.093904448105437e-05,
+      "loss": 0.6263,
+      "step": 191
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 4037.8818359375,
+      "learning_rate": 4.085667215815486e-05,
+      "loss": 0.5867,
+      "step": 192
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 2322.66015625,
+      "learning_rate": 4.077429983525536e-05,
+      "loss": 0.6167,
+      "step": 193
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 566.5298461914062,
+      "learning_rate": 4.069192751235585e-05,
+      "loss": 0.7564,
+      "step": 194
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 2636.25390625,
+      "learning_rate": 4.060955518945634e-05,
+      "loss": 0.6569,
+      "step": 195
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 484.79052734375,
+      "learning_rate": 4.052718286655684e-05,
+      "loss": 0.6649,
+      "step": 196
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 1665.9478759765625,
+      "learning_rate": 4.044481054365733e-05,
+      "loss": 0.6222,
+      "step": 197
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 2254.69384765625,
+      "learning_rate": 4.036243822075783e-05,
+      "loss": 0.618,
+      "step": 198
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 2466.4140625,
+      "learning_rate": 4.0280065897858324e-05,
+      "loss": 0.7012,
+      "step": 199
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 1356.300537109375,
+      "learning_rate": 4.019769357495882e-05,
+      "loss": 0.7464,
+      "step": 200
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 2215.750732421875,
+      "learning_rate": 4.0115321252059315e-05,
+      "loss": 0.6169,
+      "step": 201
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 2143.20751953125,
+      "learning_rate": 4.0032948929159804e-05,
+      "loss": 0.6607,
+      "step": 202
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 3176.353515625,
+      "learning_rate": 3.99505766062603e-05,
+      "loss": 0.5413,
+      "step": 203
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 1526.5035400390625,
+      "learning_rate": 3.9868204283360795e-05,
+      "loss": 0.7633,
+      "step": 204
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 2677.9453125,
+      "learning_rate": 3.978583196046129e-05,
+      "loss": 0.738,
+      "step": 205
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 1329.2694091796875,
+      "learning_rate": 3.970345963756178e-05,
+      "loss": 0.6432,
+      "step": 206
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 5269.51806640625,
+      "learning_rate": 3.9621087314662274e-05,
+      "loss": 0.6111,
+      "step": 207
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 1229.02783203125,
+      "learning_rate": 3.953871499176277e-05,
+      "loss": 0.6212,
+      "step": 208
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 2995.77783203125,
+      "learning_rate": 3.9456342668863265e-05,
+      "loss": 0.694,
+      "step": 209
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 2244.182373046875,
+      "learning_rate": 3.9373970345963754e-05,
+      "loss": 0.6843,
+      "step": 210
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 1751.8253173828125,
+      "learning_rate": 3.929159802306425e-05,
+      "loss": 0.7096,
+      "step": 211
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 1167.339111328125,
+      "learning_rate": 3.9209225700164745e-05,
+      "loss": 0.713,
+      "step": 212
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 1482.8048095703125,
+      "learning_rate": 3.912685337726524e-05,
+      "loss": 0.7545,
+      "step": 213
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 2629.055419921875,
+      "learning_rate": 3.9044481054365736e-05,
+      "loss": 0.6763,
+      "step": 214
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 94.62659454345703,
+      "learning_rate": 3.896210873146623e-05,
+      "loss": 0.6067,
+      "step": 215
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 3104.61083984375,
+      "learning_rate": 3.8879736408566727e-05,
+      "loss": 0.7046,
+      "step": 216
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 848.708740234375,
+      "learning_rate": 3.8797364085667215e-05,
+      "loss": 0.8404,
+      "step": 217
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 1472.7178955078125,
+      "learning_rate": 3.871499176276771e-05,
+      "loss": 0.7695,
+      "step": 218
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 4443.62060546875,
+      "learning_rate": 3.8632619439868206e-05,
+      "loss": 0.7434,
+      "step": 219
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 6470.4375,
+      "learning_rate": 3.85502471169687e-05,
+      "loss": 0.9423,
+      "step": 220
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 1003.2015380859375,
+      "learning_rate": 3.846787479406919e-05,
+      "loss": 0.8267,
+      "step": 221
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 6435.31494140625,
+      "learning_rate": 3.8385502471169686e-05,
+      "loss": 0.9124,
+      "step": 222
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 4398.57177734375,
+      "learning_rate": 3.830313014827018e-05,
+      "loss": 0.8491,
+      "step": 223
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 2149.6123046875,
+      "learning_rate": 3.822075782537068e-05,
+      "loss": 0.7811,
+      "step": 224
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 2641.676513671875,
+      "learning_rate": 3.813838550247117e-05,
+      "loss": 0.9134,
+      "step": 225
+    },
+    {
+      "epoch": 1.0044370493621742,
+      "grad_norm": 3034.025390625,
+      "learning_rate": 3.805601317957167e-05,
+      "loss": 0.7079,
+      "step": 226
+    },
+    {
+      "epoch": 1.0088740987243483,
+      "grad_norm": 11870.8291015625,
+      "learning_rate": 3.797364085667216e-05,
+      "loss": 0.5958,
+      "step": 227
+    },
+    {
+      "epoch": 1.0133111480865225,
+      "grad_norm": 6084.44921875,
+      "learning_rate": 3.789126853377266e-05,
+      "loss": 0.594,
+      "step": 228
+    },
+    {
+      "epoch": 1.0177481974486966,
+      "grad_norm": 5187.701171875,
+      "learning_rate": 3.780889621087315e-05,
+      "loss": 0.5405,
+      "step": 229
+    },
+    {
+      "epoch": 1.0221852468108708,
+      "grad_norm": 17869.20703125,
+      "learning_rate": 3.772652388797364e-05,
+      "loss": 0.5,
+      "step": 230
+    },
+    {
+      "epoch": 1.026622296173045,
+      "grad_norm": 9684.33203125,
+      "learning_rate": 3.764415156507414e-05,
+      "loss": 0.5654,
+      "step": 231
+    },
+    {
+      "epoch": 1.031059345535219,
+      "grad_norm": 10450.5830078125,
+      "learning_rate": 3.7561779242174634e-05,
+      "loss": 0.5737,
+      "step": 232
+    },
+    {
+      "epoch": 1.0354963948973932,
+      "grad_norm": 12497.962890625,
+      "learning_rate": 3.747940691927512e-05,
+      "loss": 0.6335,
+      "step": 233
+    },
+    {
+      "epoch": 1.0399334442595674,
+      "grad_norm": 15563.8681640625,
+      "learning_rate": 3.739703459637562e-05,
+      "loss": 0.7054,
+      "step": 234
+    },
+    {
+      "epoch": 1.0443704936217415,
+      "grad_norm": 5878.6826171875,
+      "learning_rate": 3.731466227347611e-05,
+      "loss": 0.563,
+      "step": 235
+    },
+    {
+      "epoch": 1.0488075429839157,
+      "grad_norm": 8078.8916015625,
+      "learning_rate": 3.723228995057661e-05,
+      "loss": 0.68,
+      "step": 236
+    },
+    {
+      "epoch": 1.0532445923460898,
+      "grad_norm": 8727.3701171875,
+      "learning_rate": 3.71499176276771e-05,
+      "loss": 0.5903,
+      "step": 237
+    },
+    {
+      "epoch": 1.057681641708264,
+      "grad_norm": 8727.3701171875,
+      "learning_rate": 3.71499176276771e-05,
+      "loss": 0.4867,
+      "step": 238
+    },
+    {
+      "epoch": 1.062118691070438,
+      "grad_norm": 12285.748046875,
+      "learning_rate": 3.706754530477759e-05,
+      "loss": 0.5137,
+      "step": 239
+    },
+    {
+      "epoch": 1.0665557404326123,
+      "grad_norm": 12795.34765625,
+      "learning_rate": 3.698517298187809e-05,
+      "loss": 0.5026,
+      "step": 240
+    },
+    {
+      "epoch": 1.0709927897947864,
+      "grad_norm": 7992.96142578125,
+      "learning_rate": 3.6902800658978584e-05,
+      "loss": 0.7298,
+      "step": 241
+    },
+    {
+      "epoch": 1.0754298391569606,
+      "grad_norm": 8621.84375,
+      "learning_rate": 3.682042833607908e-05,
+      "loss": 0.5946,
+      "step": 242
+    },
+    {
+      "epoch": 1.0798668885191347,
+      "grad_norm": 7196.5087890625,
+      "learning_rate": 3.6738056013179575e-05,
+      "loss": 0.5162,
+      "step": 243
+    },
+    {
+      "epoch": 1.0843039378813089,
+      "grad_norm": 6808.27978515625,
+      "learning_rate": 3.665568369028007e-05,
+      "loss": 0.6237,
+      "step": 244
+    },
+    {
+      "epoch": 1.088740987243483,
+      "grad_norm": 38360.1328125,
+      "learning_rate": 3.6573311367380566e-05,
+      "loss": 0.6606,
+      "step": 245
+    },
+    {
+      "epoch": 1.0931780366056572,
+      "grad_norm": 43912.79296875,
+      "learning_rate": 3.6490939044481054e-05,
+      "loss": 0.6049,
+      "step": 246
+    },
+    {
+      "epoch": 1.0976150859678313,
+      "grad_norm": 6522.84375,
+      "learning_rate": 3.640856672158155e-05,
+      "loss": 0.5802,
+      "step": 247
+    },
+    {
+      "epoch": 1.1020521353300055,
+      "grad_norm": 4078.390625,
+      "learning_rate": 3.6326194398682045e-05,
+      "loss": 0.6246,
+      "step": 248
+    },
+    {
+      "epoch": 1.1064891846921796,
+      "grad_norm": 4014.811767578125,
+      "learning_rate": 3.624382207578254e-05,
+      "loss": 0.5434,
+      "step": 249
+    },
+    {
+      "epoch": 1.1109262340543538,
+      "grad_norm": 15712.125,
+      "learning_rate": 3.616144975288303e-05,
+      "loss": 0.5444,
+      "step": 250
+    },
+    {
+      "epoch": 1.115363283416528,
+      "grad_norm": 20902.935546875,
+      "learning_rate": 3.6079077429983525e-05,
+      "loss": 0.5814,
+      "step": 251
+    },
+    {
+      "epoch": 1.119800332778702,
+      "grad_norm": 17141.220703125,
+      "learning_rate": 3.599670510708402e-05,
+      "loss": 0.5793,
+      "step": 252
+    },
+    {
+      "epoch": 1.1242373821408762,
+      "grad_norm": 5549.111328125,
+      "learning_rate": 3.5914332784184516e-05,
+      "loss": 0.5132,
+      "step": 253
+    },
+    {
+      "epoch": 1.1286744315030504,
+      "grad_norm": 17512.890625,
+      "learning_rate": 3.583196046128501e-05,
+      "loss": 0.5001,
+      "step": 254
+    },
+    {
+      "epoch": 1.1331114808652245,
+      "grad_norm": 8071.10888671875,
+      "learning_rate": 3.5749588138385507e-05,
+      "loss": 0.5851,
+      "step": 255
+    },
+    {
+      "epoch": 1.1375485302273987,
+      "grad_norm": 6785.5986328125,
+      "learning_rate": 3.5667215815486e-05,
+      "loss": 0.6491,
+      "step": 256
+    },
+    {
+      "epoch": 1.141985579589573,
+      "grad_norm": 10344.1728515625,
+      "learning_rate": 3.55848434925865e-05,
+      "loss": 0.5594,
+      "step": 257
+    },
+    {
+      "epoch": 1.1464226289517472,
+      "grad_norm": 7005.60888671875,
+      "learning_rate": 3.5502471169686986e-05,
+      "loss": 0.5298,
+      "step": 258
+    },
+    {
+      "epoch": 1.1508596783139213,
+      "grad_norm": 8260.1533203125,
+      "learning_rate": 3.542009884678748e-05,
+      "loss": 0.6922,
+      "step": 259
+    },
+    {
+      "epoch": 1.1552967276760955,
+      "grad_norm": 4516.599609375,
+      "learning_rate": 3.533772652388798e-05,
+      "loss": 0.4805,
+      "step": 260
+    },
+    {
+      "epoch": 1.1597337770382696,
+      "grad_norm": 4140.3701171875,
+      "learning_rate": 3.525535420098847e-05,
+      "loss": 0.5088,
+      "step": 261
+    },
+    {
+      "epoch": 1.1641708264004438,
+      "grad_norm": 4980.150390625,
+      "learning_rate": 3.517298187808896e-05,
+      "loss": 0.6193,
+      "step": 262
+    },
+    {
+      "epoch": 1.168607875762618,
+      "grad_norm": 8093.62060546875,
+      "learning_rate": 3.509060955518946e-05,
+      "loss": 0.5324,
+      "step": 263
+    },
+    {
+      "epoch": 1.173044925124792,
+      "grad_norm": 10816.904296875,
+      "learning_rate": 3.500823723228995e-05,
+      "loss": 0.533,
+      "step": 264
+    },
+    {
+      "epoch": 1.1774819744869662,
+      "grad_norm": 12252.0087890625,
+      "learning_rate": 3.492586490939045e-05,
+      "loss": 0.4386,
+      "step": 265
+    },
+    {
+      "epoch": 1.1819190238491404,
+      "grad_norm": 7755.14208984375,
+      "learning_rate": 3.4843492586490936e-05,
+      "loss": 0.505,
+      "step": 266
+    },
+    {
+      "epoch": 1.1863560732113145,
+      "grad_norm": 4057.841796875,
+      "learning_rate": 3.476112026359143e-05,
+      "loss": 0.4629,
+      "step": 267
+    },
+    {
+      "epoch": 1.1907931225734887,
+      "grad_norm": 7117.71484375,
+      "learning_rate": 3.467874794069193e-05,
+      "loss": 0.5573,
+      "step": 268
+    },
+    {
+      "epoch": 1.1952301719356628,
+      "grad_norm": 4871.728515625,
+      "learning_rate": 3.459637561779242e-05,
+      "loss": 0.4815,
+      "step": 269
+    },
+    {
+      "epoch": 1.199667221297837,
+      "grad_norm": 5573.029296875,
+      "learning_rate": 3.451400329489292e-05,
+      "loss": 0.4741,
+      "step": 270
+    },
+    {
+      "epoch": 1.2041042706600111,
+      "grad_norm": 24566.0859375,
+      "learning_rate": 3.4431630971993414e-05,
+      "loss": 0.446,
+      "step": 271
+    },
+    {
+      "epoch": 1.2085413200221853,
+      "grad_norm": 8788.73828125,
+      "learning_rate": 3.434925864909391e-05,
+      "loss": 0.4925,
+      "step": 272
+    },
+    {
+      "epoch": 1.2129783693843594,
+      "grad_norm": 3944.056396484375,
+      "learning_rate": 3.4266886326194404e-05,
+      "loss": 0.4393,
+      "step": 273
+    },
+    {
+      "epoch": 1.2174154187465336,
+      "grad_norm": 16148.9814453125,
+      "learning_rate": 3.418451400329489e-05,
+      "loss": 0.532,
+      "step": 274
+    },
+    {
+      "epoch": 1.2218524681087077,
+      "grad_norm": 11594.6630859375,
+      "learning_rate": 3.410214168039539e-05,
+      "loss": 0.6125,
+      "step": 275
+    },
+    {
+      "epoch": 1.2262895174708819,
+      "grad_norm": 6368.61572265625,
+      "learning_rate": 3.4019769357495884e-05,
+      "loss": 0.5429,
+      "step": 276
+    },
+    {
+      "epoch": 1.230726566833056,
+      "grad_norm": 3425.814453125,
+      "learning_rate": 3.393739703459638e-05,
+      "loss": 0.501,
+      "step": 277
+    },
+    {
+      "epoch": 1.2351636161952302,
+      "grad_norm": 14671.6650390625,
+      "learning_rate": 3.385502471169687e-05,
+      "loss": 0.5479,
+      "step": 278
+    },
+    {
+      "epoch": 1.2396006655574043,
+      "grad_norm": 9015.1904296875,
+      "learning_rate": 3.3772652388797364e-05,
+      "loss": 0.4545,
+      "step": 279
+    },
+    {
+      "epoch": 1.2440377149195785,
+      "grad_norm": 3020.54443359375,
+      "learning_rate": 3.369028006589786e-05,
+      "loss": 0.4921,
+      "step": 280
+    },
+    {
+      "epoch": 1.2484747642817526,
+      "grad_norm": 7199.0546875,
+      "learning_rate": 3.3607907742998355e-05,
+      "loss": 0.5344,
+      "step": 281
+    },
+    {
+      "epoch": 1.2529118136439268,
+      "grad_norm": 6765.45068359375,
+      "learning_rate": 3.352553542009885e-05,
+      "loss": 0.4384,
+      "step": 282
+    },
+    {
+      "epoch": 1.257348863006101,
+      "grad_norm": 6560.703125,
+      "learning_rate": 3.3443163097199346e-05,
+      "loss": 0.59,
+      "step": 283
+    },
+    {
+      "epoch": 1.261785912368275,
+      "grad_norm": 9844.828125,
+      "learning_rate": 3.336079077429984e-05,
+      "loss": 0.533,
+      "step": 284
+    },
+    {
+      "epoch": 1.2662229617304492,
+      "grad_norm": 3519.1572265625,
+      "learning_rate": 3.327841845140033e-05,
+      "loss": 0.447,
+      "step": 285
+    },
+    {
+      "epoch": 1.2706600110926234,
+      "grad_norm": 8746.982421875,
+      "learning_rate": 3.3196046128500825e-05,
+      "loss": 0.5225,
+      "step": 286
+    },
+    {
+      "epoch": 1.2750970604547975,
+      "grad_norm": 7445.09912109375,
+      "learning_rate": 3.311367380560132e-05,
+      "loss": 0.5536,
+      "step": 287
+    },
+    {
+      "epoch": 1.2795341098169717,
+      "grad_norm": 3131.3369140625,
+      "learning_rate": 3.3031301482701816e-05,
+      "loss": 0.5469,
+      "step": 288
+    },
+    {
+      "epoch": 1.2839711591791458,
+      "grad_norm": 4435.88427734375,
+      "learning_rate": 3.2948929159802305e-05,
+      "loss": 0.4651,
+      "step": 289
+    },
+    {
+      "epoch": 1.28840820854132,
+      "grad_norm": 5754.47900390625,
+      "learning_rate": 3.28665568369028e-05,
+      "loss": 0.5002,
+      "step": 290
+    },
+    {
+      "epoch": 1.2928452579034941,
+      "grad_norm": 4102.96240234375,
+      "learning_rate": 3.2784184514003296e-05,
+      "loss": 0.5085,
+      "step": 291
+    },
+    {
+      "epoch": 1.2972823072656683,
+      "grad_norm": 11771.892578125,
+      "learning_rate": 3.270181219110379e-05,
+      "loss": 0.4349,
+      "step": 292
+    },
+    {
+      "epoch": 1.3017193566278424,
+      "grad_norm": 26589.9296875,
+      "learning_rate": 3.261943986820428e-05,
+      "loss": 0.4491,
+      "step": 293
+    },
+    {
+      "epoch": 1.3061564059900166,
+      "grad_norm": 9183.0146484375,
+      "learning_rate": 3.2537067545304775e-05,
+      "loss": 0.4526,
+      "step": 294
+    },
+    {
+      "epoch": 1.3105934553521907,
+      "grad_norm": 12892.5400390625,
+      "learning_rate": 3.245469522240527e-05,
+      "loss": 0.4456,
+      "step": 295
+    },
+    {
+      "epoch": 1.315030504714365,
+      "grad_norm": 28994.1484375,
+      "learning_rate": 3.2372322899505766e-05,
+      "loss": 0.4788,
+      "step": 296
+    },
+    {
+      "epoch": 1.319467554076539,
+      "grad_norm": 12730.1240234375,
+      "learning_rate": 3.228995057660626e-05,
+      "loss": 0.7489,
+      "step": 297
+    },
+    {
+      "epoch": 1.3239046034387132,
+      "grad_norm": 15733.169921875,
+      "learning_rate": 3.220757825370676e-05,
+      "loss": 0.4152,
+      "step": 298
+    },
+    {
+      "epoch": 1.3283416528008873,
+      "grad_norm": 2216.435791015625,
+      "learning_rate": 3.212520593080725e-05,
+      "loss": 0.3614,
+      "step": 299
+    },
+    {
+      "epoch": 1.3327787021630615,
+      "grad_norm": 3411.817138671875,
+      "learning_rate": 3.204283360790775e-05,
+      "loss": 0.4807,
+      "step": 300
+    },
+    {
+      "epoch": 1.3372157515252356,
+      "grad_norm": 10712.48046875,
+      "learning_rate": 3.196046128500824e-05,
+      "loss": 0.4707,
+      "step": 301
+    },
+    {
+      "epoch": 1.3416528008874098,
+      "grad_norm": 4834.8603515625,
+      "learning_rate": 3.187808896210873e-05,
+      "loss": 0.468,
+      "step": 302
+    },
+    {
+      "epoch": 1.3460898502495842,
+      "grad_norm": 4424.07373046875,
+      "learning_rate": 3.179571663920923e-05,
+      "loss": 0.5807,
+      "step": 303
+    },
+    {
+      "epoch": 1.3505268996117583,
+      "grad_norm": 9875.8291015625,
+      "learning_rate": 3.171334431630972e-05,
+      "loss": 0.5514,
+      "step": 304
+    },
+    {
+      "epoch": 1.3549639489739325,
+      "grad_norm": 8319.3427734375,
+      "learning_rate": 3.163097199341021e-05,
+      "loss": 0.399,
+      "step": 305
+    },
+    {
+      "epoch": 1.3594009983361066,
+      "grad_norm": 5627.841796875,
+      "learning_rate": 3.154859967051071e-05,
+      "loss": 0.4483,
+      "step": 306
+    },
+    {
+      "epoch": 1.3638380476982808,
+      "grad_norm": 12529.3662109375,
+      "learning_rate": 3.14662273476112e-05,
+      "loss": 0.4402,
+      "step": 307
+    },
+    {
+      "epoch": 1.368275097060455,
+      "grad_norm": 4284.75244140625,
+      "learning_rate": 3.13838550247117e-05,
+      "loss": 0.495,
+      "step": 308
+    },
+    {
+      "epoch": 1.372712146422629,
+      "grad_norm": 13047.9990234375,
+      "learning_rate": 3.1301482701812194e-05,
+      "loss": 0.484,
+      "step": 309
+    },
+    {
+      "epoch": 1.3771491957848032,
+      "grad_norm": 11393.1103515625,
+      "learning_rate": 3.121911037891269e-05,
+      "loss": 0.5429,
+      "step": 310
+    },
+    {
+      "epoch": 1.3815862451469774,
+      "grad_norm": 7537.15576171875,
+      "learning_rate": 3.1136738056013184e-05,
+      "loss": 0.4303,
+      "step": 311
+    },
+    {
+      "epoch": 1.3860232945091515,
+      "grad_norm": 10372.8466796875,
+      "learning_rate": 3.105436573311368e-05,
+      "loss": 0.4128,
+      "step": 312
+    },
+    {
+      "epoch": 1.3904603438713257,
+      "grad_norm": 16802.3046875,
+      "learning_rate": 3.097199341021417e-05,
+      "loss": 0.4362,
+      "step": 313
+    },
+    {
+      "epoch": 1.3948973932334998,
+      "grad_norm": 6575.572265625,
+      "learning_rate": 3.0889621087314664e-05,
+      "loss": 0.457,
+      "step": 314
+    },
+    {
+      "epoch": 1.399334442595674,
+      "grad_norm": 15999.4423828125,
+      "learning_rate": 3.080724876441516e-05,
+      "loss": 0.4625,
+      "step": 315
+    },
+    {
+      "epoch": 1.4037714919578481,
+      "grad_norm": 11838.2890625,
+      "learning_rate": 3.0724876441515655e-05,
+      "loss": 0.4789,
+      "step": 316
+    },
+    {
+      "epoch": 1.4082085413200223,
+      "grad_norm": 7021.83154296875,
+      "learning_rate": 3.0642504118616144e-05,
+      "loss": 0.4171,
+      "step": 317
+    },
+    {
+      "epoch": 1.4126455906821964,
+      "grad_norm": 9889.8447265625,
+      "learning_rate": 3.056013179571664e-05,
+      "loss": 0.4594,
+      "step": 318
+    },
+    {
+      "epoch": 1.4170826400443706,
+      "grad_norm": 7274.470703125,
+      "learning_rate": 3.0477759472817135e-05,
+      "loss": 0.4472,
+      "step": 319
+    },
+    {
+      "epoch": 1.4215196894065447,
+      "grad_norm": 7417.62109375,
+      "learning_rate": 3.039538714991763e-05,
+      "loss": 0.4204,
+      "step": 320
+    },
+    {
+      "epoch": 1.4259567387687189,
+      "grad_norm": 2274.059326171875,
+      "learning_rate": 3.0313014827018122e-05,
+      "loss": 0.4137,
+      "step": 321
+    },
+    {
+      "epoch": 1.430393788130893,
+      "grad_norm": 10090.2900390625,
+      "learning_rate": 3.0230642504118618e-05,
+      "loss": 0.4605,
+      "step": 322
+    },
+    {
+      "epoch": 1.4348308374930672,
+      "grad_norm": 7575.30908203125,
+      "learning_rate": 3.0148270181219113e-05,
+      "loss": 0.3987,
+      "step": 323
+    },
+    {
+      "epoch": 1.4392678868552413,
+      "grad_norm": 6146.14794921875,
+      "learning_rate": 3.006589785831961e-05,
+      "loss": 0.457,
+      "step": 324
+    },
+    {
+      "epoch": 1.4437049362174155,
+      "grad_norm": 8637.0048828125,
+      "learning_rate": 2.9983525535420097e-05,
+      "loss": 0.3743,
+      "step": 325
+    },
+    {
+      "epoch": 1.4481419855795896,
+      "grad_norm": 10008.0849609375,
+      "learning_rate": 2.9901153212520593e-05,
+      "loss": 0.4113,
+      "step": 326
+    },
+    {
+      "epoch": 1.4525790349417638,
+      "grad_norm": 3893.683837890625,
+      "learning_rate": 2.9818780889621088e-05,
+      "loss": 0.3641,
+      "step": 327
+    },
+    {
+      "epoch": 1.457016084303938,
+      "grad_norm": 11483.0849609375,
+      "learning_rate": 2.9736408566721587e-05,
+      "loss": 0.4293,
+      "step": 328
+    },
+    {
+      "epoch": 1.461453133666112,
+      "grad_norm": 9978.671875,
+      "learning_rate": 2.9654036243822076e-05,
+      "loss": 0.5509,
+      "step": 329
+    },
+    {
+      "epoch": 1.4658901830282862,
+      "grad_norm": 8842.6669921875,
+      "learning_rate": 2.957166392092257e-05,
+      "loss": 0.4588,
+      "step": 330
+    },
+    {
+      "epoch": 1.4703272323904604,
+      "grad_norm": 4001.705810546875,
+      "learning_rate": 2.9489291598023067e-05,
+      "loss": 0.4695,
+      "step": 331
+    },
+    {
+      "epoch": 1.4747642817526345,
+      "grad_norm": 10248.3017578125,
+      "learning_rate": 2.9406919275123562e-05,
+      "loss": 0.4749,
+      "step": 332
+    },
+    {
+      "epoch": 1.4792013311148087,
+      "grad_norm": 7325.65478515625,
+      "learning_rate": 2.9324546952224054e-05,
+      "loss": 0.5116,
+      "step": 333
+    },
+    {
+      "epoch": 1.4836383804769828,
+      "grad_norm": 7472.83056640625,
+      "learning_rate": 2.924217462932455e-05,
+      "loss": 0.4268,
+      "step": 334
+    },
+    {
+      "epoch": 1.488075429839157,
+      "grad_norm": 3410.792236328125,
+      "learning_rate": 2.9159802306425045e-05,
+      "loss": 0.4117,
+      "step": 335
+    },
+    {
+      "epoch": 1.4925124792013311,
+      "grad_norm": 7307.14453125,
+      "learning_rate": 2.907742998352554e-05,
+      "loss": 0.4716,
+      "step": 336
+    },
+    {
+      "epoch": 1.4969495285635053,
+      "grad_norm": 6047.51416015625,
+      "learning_rate": 2.899505766062603e-05,
+      "loss": 0.4066,
+      "step": 337
+    },
+    {
+      "epoch": 1.5013865779256794,
+      "grad_norm": 10396.7392578125,
+      "learning_rate": 2.8912685337726525e-05,
+      "loss": 0.4411,
+      "step": 338
+    },
+    {
+      "epoch": 1.5058236272878536,
+      "grad_norm": 6540.0595703125,
+      "learning_rate": 2.883031301482702e-05,
+      "loss": 0.3979,
+      "step": 339
+    },
+    {
+      "epoch": 1.5102606766500277,
+      "grad_norm": 3863.45849609375,
+      "learning_rate": 2.8747940691927515e-05,
+      "loss": 0.3906,
+      "step": 340
+    },
+    {
+      "epoch": 1.5146977260122019,
+      "grad_norm": 13261.5732421875,
+      "learning_rate": 2.8665568369028008e-05,
+      "loss": 0.554,
+      "step": 341
+    },
+    {
+      "epoch": 1.519134775374376,
+      "grad_norm": 5675.54248046875,
+      "learning_rate": 2.8583196046128503e-05,
+      "loss": 0.4381,
+      "step": 342
+    },
+    {
+      "epoch": 1.5235718247365502,
+      "grad_norm": 7444.20947265625,
+      "learning_rate": 2.8500823723229e-05,
+      "loss": 0.4225,
+      "step": 343
+    },
+    {
+      "epoch": 1.5280088740987243,
+      "grad_norm": 10648.59765625,
+      "learning_rate": 2.8418451400329494e-05,
+      "loss": 0.5227,
+      "step": 344
+    },
+    {
+      "epoch": 1.5324459234608985,
+      "grad_norm": 8689.919921875,
+      "learning_rate": 2.8336079077429983e-05,
+      "loss": 0.4427,
+      "step": 345
+    },
+    {
+      "epoch": 1.5368829728230726,
+      "grad_norm": 13844.3818359375,
+      "learning_rate": 2.8253706754530478e-05,
+      "loss": 0.4912,
+      "step": 346
+    },
+    {
+      "epoch": 1.5413200221852468,
+      "grad_norm": 4550.9091796875,
+      "learning_rate": 2.8171334431630973e-05,
+      "loss": 0.4494,
+      "step": 347
+    },
+    {
+      "epoch": 1.545757071547421,
+      "grad_norm": 5387.0693359375,
+      "learning_rate": 2.808896210873147e-05,
+      "loss": 0.4182,
+      "step": 348
+    },
+    {
+      "epoch": 1.550194120909595,
+      "grad_norm": 2335.486572265625,
+      "learning_rate": 2.800658978583196e-05,
+      "loss": 0.4114,
+      "step": 349
+    },
+    {
+      "epoch": 1.5546311702717692,
+      "grad_norm": 6499.111328125,
+      "learning_rate": 2.7924217462932456e-05,
+      "loss": 0.3942,
+      "step": 350
+    },
+    {
+      "epoch": 1.5590682196339434,
+      "grad_norm": 9013.1005859375,
+      "learning_rate": 2.7841845140032952e-05,
+      "loss": 0.5319,
+      "step": 351
+    },
+    {
+      "epoch": 1.5635052689961175,
+      "grad_norm": 27906.298828125,
+      "learning_rate": 2.775947281713344e-05,
+      "loss": 0.4856,
+      "step": 352
+    },
+    {
+      "epoch": 1.5679423183582917,
+      "grad_norm": 17134.248046875,
+      "learning_rate": 2.7677100494233936e-05,
+      "loss": 0.4635,
+      "step": 353
+    },
+    {
+      "epoch": 1.5723793677204658,
+      "grad_norm": 2359.93603515625,
+      "learning_rate": 2.759472817133443e-05,
+      "loss": 0.3834,
+      "step": 354
+    },
+    {
+      "epoch": 1.57681641708264,
+      "grad_norm": 8103.72705078125,
+      "learning_rate": 2.7512355848434927e-05,
+      "loss": 0.4369,
+      "step": 355
+    },
+    {
+      "epoch": 1.5812534664448141,
+      "grad_norm": 5756.72998046875,
+      "learning_rate": 2.742998352553542e-05,
+      "loss": 0.4076,
+      "step": 356
+    },
+    {
+      "epoch": 1.5856905158069883,
+      "grad_norm": 11778.0205078125,
+      "learning_rate": 2.7347611202635915e-05,
+      "loss": 0.4564,
+      "step": 357
+    },
+    {
+      "epoch": 1.5901275651691624,
+      "grad_norm": 3563.62158203125,
+      "learning_rate": 2.726523887973641e-05,
+      "loss": 0.4679,
+      "step": 358
+    },
+    {
+      "epoch": 1.5945646145313366,
+      "grad_norm": 13851.51171875,
+      "learning_rate": 2.7182866556836905e-05,
+      "loss": 0.4988,
+      "step": 359
+    },
+    {
+      "epoch": 1.5990016638935107,
+      "grad_norm": 9023.9638671875,
+      "learning_rate": 2.7100494233937394e-05,
+      "loss": 0.4526,
+      "step": 360
+    },
+    {
+      "epoch": 1.6034387132556849,
+      "grad_norm": 25059.20703125,
+      "learning_rate": 2.7018121911037893e-05,
+      "loss": 0.4261,
+      "step": 361
+    },
+    {
+      "epoch": 1.607875762617859,
+      "grad_norm": 10355.134765625,
+      "learning_rate": 2.693574958813839e-05,
+      "loss": 0.4966,
+      "step": 362
+    },
+    {
+      "epoch": 1.6123128119800332,
+      "grad_norm": 9313.6435546875,
+      "learning_rate": 2.6853377265238884e-05,
+      "loss": 0.4347,
+      "step": 363
+    },
+    {
+      "epoch": 1.6167498613422073,
+      "grad_norm": 4487.12255859375,
+      "learning_rate": 2.6771004942339373e-05,
+      "loss": 0.4244,
+      "step": 364
+    },
+    {
+      "epoch": 1.6211869107043815,
+      "grad_norm": 10358.0615234375,
+      "learning_rate": 2.6688632619439868e-05,
+      "loss": 0.4222,
+      "step": 365
+    },
+    {
+      "epoch": 1.6256239600665556,
+      "grad_norm": 10428.6669921875,
+      "learning_rate": 2.6606260296540363e-05,
+      "loss": 0.4184,
+      "step": 366
+    },
+    {
+      "epoch": 1.6300610094287298,
+      "grad_norm": 7052.03564453125,
+      "learning_rate": 2.652388797364086e-05,
+      "loss": 0.4261,
+      "step": 367
+    },
+    {
+      "epoch": 1.634498058790904,
+      "grad_norm": 9667.5341796875,
+      "learning_rate": 2.644151565074135e-05,
+      "loss": 0.3969,
+      "step": 368
+    },
+    {
+      "epoch": 1.638935108153078,
+      "grad_norm": 6778.982421875,
+      "learning_rate": 2.6359143327841846e-05,
+      "loss": 0.41,
+      "step": 369
+    },
+    {
+      "epoch": 1.6433721575152522,
+      "grad_norm": 51579.52734375,
+      "learning_rate": 2.6276771004942342e-05,
+      "loss": 0.4775,
+      "step": 370
+    },
+    {
+      "epoch": 1.6478092068774264,
+      "grad_norm": 10998.0263671875,
+      "learning_rate": 2.6194398682042837e-05,
+      "loss": 0.4095,
+      "step": 371
+    },
+    {
+      "epoch": 1.6522462562396005,
+      "grad_norm": 26724.208984375,
+      "learning_rate": 2.6112026359143326e-05,
+      "loss": 0.4638,
+      "step": 372
+    },
+    {
+      "epoch": 1.6566833056017747,
+      "grad_norm": 8755.763671875,
+      "learning_rate": 2.602965403624382e-05,
+      "loss": 0.5112,
+      "step": 373
+    },
+    {
+      "epoch": 1.6611203549639488,
+      "grad_norm": 2200.8251953125,
+      "learning_rate": 2.5947281713344317e-05,
+      "loss": 0.4041,
+      "step": 374
+    },
+    {
+      "epoch": 1.665557404326123,
+      "grad_norm": 4841.39111328125,
+      "learning_rate": 2.5864909390444812e-05,
+      "loss": 0.3594,
+      "step": 375
+    },
+    {
+      "epoch": 1.6699944536882971,
+      "grad_norm": 6945.18212890625,
+      "learning_rate": 2.5782537067545305e-05,
+      "loss": 0.3593,
+      "step": 376
+    },
+    {
+      "epoch": 1.6744315030504713,
+      "grad_norm": 24270.375,
+      "learning_rate": 2.57001647446458e-05,
+      "loss": 0.4364,
+      "step": 377
+    },
+    {
+      "epoch": 1.6788685524126454,
+      "grad_norm": 8844.3515625,
+      "learning_rate": 2.5617792421746295e-05,
+      "loss": 0.5474,
+      "step": 378
+    },
+    {
+      "epoch": 1.6833056017748196,
+      "grad_norm": 2044.9327392578125,
+      "learning_rate": 2.553542009884679e-05,
+      "loss": 0.3915,
+      "step": 379
+    },
+    {
+      "epoch": 1.687742651136994,
+      "grad_norm": 8917.6083984375,
+      "learning_rate": 2.545304777594728e-05,
+      "loss": 0.4519,
+      "step": 380
+    },
+    {
+      "epoch": 1.692179700499168,
+      "grad_norm": 8733.7265625,
+      "learning_rate": 2.5370675453047775e-05,
+      "loss": 0.3886,
+      "step": 381
+    },
+    {
+      "epoch": 1.6966167498613423,
+      "grad_norm": 16858.744140625,
+      "learning_rate": 2.528830313014827e-05,
+      "loss": 0.5291,
+      "step": 382
+    },
+    {
+      "epoch": 1.7010537992235164,
+      "grad_norm": 8900.0537109375,
+      "learning_rate": 2.520593080724877e-05,
+      "loss": 0.4109,
+      "step": 383
+    },
+    {
+      "epoch": 1.7054908485856906,
+      "grad_norm": 18292.86328125,
+      "learning_rate": 2.5123558484349258e-05,
+      "loss": 0.4471,
+      "step": 384
+    },
+    {
+      "epoch": 1.7099278979478647,
+      "grad_norm": 16270.12109375,
+      "learning_rate": 2.5041186161449753e-05,
+      "loss": 0.4718,
+      "step": 385
+    },
+    {
+      "epoch": 1.7143649473100389,
+      "grad_norm": 12952.6103515625,
+      "learning_rate": 2.495881383855025e-05,
+      "loss": 0.419,
+      "step": 386
+    },
+    {
+      "epoch": 1.718801996672213,
+      "grad_norm": 21410.482421875,
+      "learning_rate": 2.487644151565074e-05,
+      "loss": 0.4258,
+      "step": 387
+    },
+    {
+      "epoch": 1.7232390460343872,
+      "grad_norm": 6131.7421875,
+      "learning_rate": 2.4794069192751236e-05,
+      "loss": 0.5263,
+      "step": 388
+    },
+    {
+      "epoch": 1.7276760953965613,
+      "grad_norm": 7797.50439453125,
+      "learning_rate": 2.4711696869851732e-05,
+      "loss": 0.5074,
+      "step": 389
+    },
+    {
+      "epoch": 1.7321131447587355,
+      "grad_norm": 7881.93212890625,
+      "learning_rate": 2.4629324546952227e-05,
+      "loss": 0.3924,
+      "step": 390
+    },
+    {
+      "epoch": 1.7365501941209096,
+      "grad_norm": 10285.5458984375,
+      "learning_rate": 2.454695222405272e-05,
+      "loss": 0.3844,
+      "step": 391
+    },
+    {
+      "epoch": 1.7409872434830838,
+      "grad_norm": 7698.091796875,
+      "learning_rate": 2.4464579901153215e-05,
+      "loss": 0.4452,
+      "step": 392
+    },
+    {
+      "epoch": 1.745424292845258,
+      "grad_norm": 11063.9072265625,
+      "learning_rate": 2.4382207578253707e-05,
+      "loss": 0.4668,
+      "step": 393
+    },
+    {
+      "epoch": 1.749861342207432,
+      "grad_norm": 5692.85498046875,
+      "learning_rate": 2.4299835255354202e-05,
+      "loss": 0.519,
+      "step": 394
+    },
+    {
+      "epoch": 1.7542983915696062,
+      "grad_norm": 16777.486328125,
+      "learning_rate": 2.4217462932454694e-05,
+      "loss": 0.3775,
+      "step": 395
+    },
+    {
+      "epoch": 1.7587354409317804,
+      "grad_norm": 4603.4892578125,
+      "learning_rate": 2.413509060955519e-05,
+      "loss": 0.3691,
+      "step": 396
+    },
+    {
+      "epoch": 1.7631724902939545,
+      "grad_norm": 6875.03759765625,
+      "learning_rate": 2.4052718286655685e-05,
+      "loss": 0.409,
+      "step": 397
+    },
+    {
+      "epoch": 1.7676095396561287,
+      "grad_norm": 2404.7490234375,
+      "learning_rate": 2.397034596375618e-05,
+      "loss": 0.4091,
+      "step": 398
+    },
+    {
+      "epoch": 1.7720465890183028,
+      "grad_norm": 9926.2451171875,
+      "learning_rate": 2.3887973640856673e-05,
+      "loss": 0.4961,
+      "step": 399
+    },
+    {
+      "epoch": 1.776483638380477,
+      "grad_norm": 12885.69921875,
+      "learning_rate": 2.380560131795717e-05,
+      "loss": 0.4083,
+      "step": 400
+    },
+    {
+      "epoch": 1.780920687742651,
+      "grad_norm": 10893.0986328125,
+      "learning_rate": 2.372322899505766e-05,
+      "loss": 0.3934,
+      "step": 401
+    },
+    {
+      "epoch": 1.7853577371048253,
+      "grad_norm": 4749.9833984375,
+      "learning_rate": 2.3640856672158156e-05,
+      "loss": 0.4328,
+      "step": 402
+    },
+    {
+      "epoch": 1.7897947864669994,
+      "grad_norm": 8398.91796875,
+      "learning_rate": 2.355848434925865e-05,
+      "loss": 0.3771,
+      "step": 403
+    },
+    {
+      "epoch": 1.7942318358291736,
+      "grad_norm": 11402.8720703125,
+      "learning_rate": 2.3476112026359147e-05,
+      "loss": 0.4252,
+      "step": 404
+    },
+    {
+      "epoch": 1.7986688851913477,
+      "grad_norm": 4041.045654296875,
+      "learning_rate": 2.339373970345964e-05,
+      "loss": 0.4316,
+      "step": 405
+    },
+    {
+      "epoch": 1.8031059345535219,
+      "grad_norm": 4436.52392578125,
+      "learning_rate": 2.3311367380560134e-05,
+      "loss": 0.5129,
+      "step": 406
+    },
+    {
+      "epoch": 1.807542983915696,
+      "grad_norm": 3380.321044921875,
+      "learning_rate": 2.3228995057660626e-05,
+      "loss": 0.4431,
+      "step": 407
+    },
+    {
+      "epoch": 1.8119800332778702,
+      "grad_norm": 6277.3369140625,
+      "learning_rate": 2.3146622734761122e-05,
+      "loss": 0.4627,
+      "step": 408
+    },
+    {
+      "epoch": 1.8164170826400445,
+      "grad_norm": 10034.5419921875,
+      "learning_rate": 2.3064250411861614e-05,
+      "loss": 0.4617,
+      "step": 409
+    },
+    {
+      "epoch": 1.8208541320022187,
+      "grad_norm": 12608.28125,
+      "learning_rate": 2.298187808896211e-05,
+      "loss": 0.4003,
+      "step": 410
+    },
+    {
+      "epoch": 1.8252911813643928,
+      "grad_norm": 11147.564453125,
+      "learning_rate": 2.2899505766062605e-05,
+      "loss": 0.4123,
+      "step": 411
+    },
+    {
+      "epoch": 1.829728230726567,
+      "grad_norm": 5768.64697265625,
+      "learning_rate": 2.28171334431631e-05,
+      "loss": 0.4655,
+      "step": 412
+    },
+    {
+      "epoch": 1.8341652800887411,
+      "grad_norm": 5285.51806640625,
+      "learning_rate": 2.2734761120263592e-05,
+      "loss": 0.5375,
+      "step": 413
+    },
+    {
+      "epoch": 1.8386023294509153,
+      "grad_norm": 3518.06591796875,
+      "learning_rate": 2.2652388797364088e-05,
+      "loss": 0.4281,
+      "step": 414
+    },
+    {
+      "epoch": 1.8430393788130894,
+      "grad_norm": 3987.496337890625,
+      "learning_rate": 2.257001647446458e-05,
+      "loss": 0.4506,
+      "step": 415
+    },
+    {
+      "epoch": 1.8474764281752636,
+      "grad_norm": 13528.0654296875,
+      "learning_rate": 2.2487644151565075e-05,
+      "loss": 0.4163,
+      "step": 416
+    },
+    {
+      "epoch": 1.8519134775374377,
+      "grad_norm": 13044.259765625,
+      "learning_rate": 2.240527182866557e-05,
+      "loss": 0.4181,
+      "step": 417
+    },
+    {
+      "epoch": 1.8563505268996119,
+      "grad_norm": 7670.06787109375,
+      "learning_rate": 2.2322899505766066e-05,
+      "loss": 0.4794,
+      "step": 418
+    },
+    {
+      "epoch": 1.860787576261786,
+      "grad_norm": 4279.3955078125,
+      "learning_rate": 2.224052718286656e-05,
+      "loss": 0.3708,
+      "step": 419
+    },
+    {
+      "epoch": 1.8652246256239602,
+      "grad_norm": 18060.40234375,
+      "learning_rate": 2.215815485996705e-05,
+      "loss": 0.4386,
+      "step": 420
+    },
+    {
+      "epoch": 1.8696616749861343,
+      "grad_norm": 6616.28271484375,
+      "learning_rate": 2.2075782537067546e-05,
+      "loss": 0.4706,
+      "step": 421
+    },
+    {
+      "epoch": 1.8740987243483085,
+      "grad_norm": 7691.806640625,
+      "learning_rate": 2.1993410214168038e-05,
+      "loss": 0.3925,
+      "step": 422
+    },
+    {
+      "epoch": 1.8785357737104826,
+      "grad_norm": 3910.61474609375,
+      "learning_rate": 2.1911037891268533e-05,
+      "loss": 0.4692,
+      "step": 423
+    },
+    {
+      "epoch": 1.8829728230726568,
+      "grad_norm": 6454.8330078125,
+      "learning_rate": 2.182866556836903e-05,
+      "loss": 0.3806,
+      "step": 424
+    },
+    {
+      "epoch": 1.887409872434831,
+      "grad_norm": 11075.0537109375,
+      "learning_rate": 2.1746293245469524e-05,
+      "loss": 0.49,
+      "step": 425
+    },
+    {
+      "epoch": 1.891846921797005,
+      "grad_norm": 5856.767578125,
+      "learning_rate": 2.1663920922570016e-05,
+      "loss": 0.4335,
+      "step": 426
+    },
+    {
+      "epoch": 1.8962839711591792,
+      "grad_norm": 7504.56396484375,
+      "learning_rate": 2.1581548599670512e-05,
+      "loss": 0.4641,
+      "step": 427
+    },
+    {
+      "epoch": 1.9007210205213534,
+      "grad_norm": 8157.1943359375,
+      "learning_rate": 2.1499176276771004e-05,
+      "loss": 0.4111,
+      "step": 428
+    },
+    {
+      "epoch": 1.9051580698835275,
+      "grad_norm": 10462.9794921875,
+      "learning_rate": 2.14168039538715e-05,
+      "loss": 0.4593,
+      "step": 429
+    },
+    {
+      "epoch": 1.9095951192457017,
+      "grad_norm": 13179.11328125,
+      "learning_rate": 2.1334431630971995e-05,
+      "loss": 0.6586,
+      "step": 430
+    },
+    {
+      "epoch": 1.9140321686078758,
+      "grad_norm": 4364.4208984375,
+      "learning_rate": 2.125205930807249e-05,
+      "loss": 0.3736,
+      "step": 431
+    },
+    {
+      "epoch": 1.91846921797005,
+      "grad_norm": 4098.35791015625,
+      "learning_rate": 2.1169686985172982e-05,
+      "loss": 0.3934,
+      "step": 432
+    },
+    {
+      "epoch": 1.9229062673322241,
+      "grad_norm": 3337.504638671875,
+      "learning_rate": 2.1087314662273478e-05,
+      "loss": 0.416,
+      "step": 433
+    },
+    {
+      "epoch": 1.9273433166943983,
+      "grad_norm": 8191.77587890625,
+      "learning_rate": 2.100494233937397e-05,
+      "loss": 0.4885,
+      "step": 434
+    },
+    {
+      "epoch": 1.9317803660565724,
+      "grad_norm": 5037.484375,
+      "learning_rate": 2.0922570016474465e-05,
+      "loss": 0.4832,
+      "step": 435
+    },
+    {
+      "epoch": 1.9362174154187466,
+      "grad_norm": 17356.53515625,
+      "learning_rate": 2.0840197693574957e-05,
+      "loss": 0.4344,
+      "step": 436
+    },
+    {
+      "epoch": 1.9406544647809207,
+      "grad_norm": 9798.25390625,
+      "learning_rate": 2.0757825370675453e-05,
+      "loss": 0.4935,
+      "step": 437
+    },
+    {
+      "epoch": 1.945091514143095,
+      "grad_norm": 15715.357421875,
+      "learning_rate": 2.067545304777595e-05,
+      "loss": 0.3853,
+      "step": 438
+    },
+    {
+      "epoch": 1.949528563505269,
+      "grad_norm": 7445.16748046875,
+      "learning_rate": 2.0593080724876444e-05,
+      "loss": 0.4336,
+      "step": 439
+    },
+    {
+      "epoch": 1.9539656128674432,
+      "grad_norm": 8402.876953125,
+      "learning_rate": 2.0510708401976936e-05,
+      "loss": 0.3743,
+      "step": 440
+    },
+    {
+      "epoch": 1.9584026622296173,
+      "grad_norm": 4802.0849609375,
+      "learning_rate": 2.042833607907743e-05,
+      "loss": 0.4453,
+      "step": 441
+    },
+    {
+      "epoch": 1.9628397115917915,
+      "grad_norm": 8812.525390625,
+      "learning_rate": 2.0345963756177923e-05,
+      "loss": 0.4641,
+      "step": 442
+    },
+    {
+      "epoch": 1.9672767609539656,
+      "grad_norm": 4904.171875,
+      "learning_rate": 2.026359143327842e-05,
+      "loss": 0.4196,
+      "step": 443
+    },
+    {
+      "epoch": 1.9717138103161398,
+      "grad_norm": 13073.9609375,
+      "learning_rate": 2.0181219110378914e-05,
+      "loss": 0.4024,
+      "step": 444
+    },
+    {
+      "epoch": 1.976150859678314,
+      "grad_norm": 5855.8720703125,
+      "learning_rate": 2.009884678747941e-05,
+      "loss": 0.4116,
+      "step": 445
+    },
+    {
+      "epoch": 1.980587909040488,
+      "grad_norm": 13108.2412109375,
+      "learning_rate": 2.0016474464579902e-05,
+      "loss": 0.3518,
+      "step": 446
+    },
+    {
+      "epoch": 1.9850249584026622,
+      "grad_norm": 8249.966796875,
+      "learning_rate": 1.9934102141680397e-05,
+      "loss": 0.4499,
+      "step": 447
+    },
+    {
+      "epoch": 1.9894620077648364,
+      "grad_norm": 7402.30517578125,
+      "learning_rate": 1.985172981878089e-05,
+      "loss": 0.3638,
+      "step": 448
+    },
+    {
+      "epoch": 1.9938990571270105,
+      "grad_norm": 9077.05859375,
+      "learning_rate": 1.9769357495881385e-05,
+      "loss": 0.3974,
+      "step": 449
+    },
+    {
+      "epoch": 1.9983361064891847,
+      "grad_norm": 6565.54296875,
+      "learning_rate": 1.9686985172981877e-05,
+      "loss": 0.4431,
+      "step": 450
+    },
+    {
+      "epoch": 2.002773155851359,
+      "grad_norm": 9013.576171875,
+      "learning_rate": 1.9604612850082372e-05,
+      "loss": 0.3477,
+      "step": 451
+    },
+    {
+      "epoch": 2.007210205213533,
+      "grad_norm": 11014.0927734375,
+      "learning_rate": 1.9522240527182868e-05,
+      "loss": 0.359,
+      "step": 452
+    },
+    {
+      "epoch": 2.011647254575707,
+      "grad_norm": 9727.5078125,
+      "learning_rate": 1.9439868204283363e-05,
+      "loss": 0.3128,
+      "step": 453
+    },
+    {
+      "epoch": 2.0160843039378813,
+      "grad_norm": 9441.8505859375,
+      "learning_rate": 1.9357495881383855e-05,
+      "loss": 0.3684,
+      "step": 454
+    },
+    {
+      "epoch": 2.0205213533000554,
+      "grad_norm": 5378.3115234375,
+      "learning_rate": 1.927512355848435e-05,
+      "loss": 0.3266,
+      "step": 455
+    },
+    {
+      "epoch": 2.0249584026622296,
+      "grad_norm": 14302.375,
+      "learning_rate": 1.9192751235584843e-05,
+      "loss": 0.3005,
+      "step": 456
+    },
+    {
+      "epoch": 2.0293954520244037,
+      "grad_norm": 1865.003662109375,
+      "learning_rate": 1.911037891268534e-05,
+      "loss": 0.2982,
+      "step": 457
+    },
+    {
+      "epoch": 2.033832501386578,
+      "grad_norm": 6775.93505859375,
+      "learning_rate": 1.9028006589785834e-05,
+      "loss": 0.2929,
+      "step": 458
+    },
+    {
+      "epoch": 2.038269550748752,
+      "grad_norm": 6535.48388671875,
+      "learning_rate": 1.894563426688633e-05,
+      "loss": 0.3043,
+      "step": 459
+    },
+    {
+      "epoch": 2.042706600110926,
+      "grad_norm": 11312.7392578125,
+      "learning_rate": 1.886326194398682e-05,
+      "loss": 0.2944,
+      "step": 460
+    },
+    {
+      "epoch": 2.0471436494731003,
+      "grad_norm": 12864.8671875,
+      "learning_rate": 1.8780889621087317e-05,
+      "loss": 0.362,
+      "step": 461
+    },
+    {
+      "epoch": 2.0515806988352745,
+      "grad_norm": 8311.9443359375,
+      "learning_rate": 1.869851729818781e-05,
+      "loss": 0.3062,
+      "step": 462
+    },
+    {
+      "epoch": 2.0560177481974486,
+      "grad_norm": 6701.55322265625,
+      "learning_rate": 1.8616144975288304e-05,
+      "loss": 0.3472,
+      "step": 463
+    },
+    {
+      "epoch": 2.060454797559623,
+      "grad_norm": 4350.64892578125,
+      "learning_rate": 1.8533772652388796e-05,
+      "loss": 0.3278,
+      "step": 464
+    },
+    {
+      "epoch": 2.064891846921797,
+      "grad_norm": 9409.6240234375,
+      "learning_rate": 1.8451400329489292e-05,
+      "loss": 0.436,
+      "step": 465
+    },
+    {
+      "epoch": 2.069328896283971,
+      "grad_norm": 4991.9775390625,
+      "learning_rate": 1.8369028006589787e-05,
+      "loss": 0.312,
+      "step": 466
+    },
+    {
+      "epoch": 2.0737659456461452,
+      "grad_norm": 14262.6591796875,
+      "learning_rate": 1.8286655683690283e-05,
+      "loss": 0.3671,
+      "step": 467
+    },
+    {
+      "epoch": 2.0782029950083194,
+      "grad_norm": 5293.71728515625,
+      "learning_rate": 1.8204283360790775e-05,
+      "loss": 0.3683,
+      "step": 468
+    },
+    {
+      "epoch": 2.0826400443704935,
+      "grad_norm": 11404.3525390625,
+      "learning_rate": 1.812191103789127e-05,
+      "loss": 0.3306,
+      "step": 469
+    },
+    {
+      "epoch": 2.0870770937326677,
+      "grad_norm": 11855.4306640625,
+      "learning_rate": 1.8039538714991762e-05,
+      "loss": 0.2957,
+      "step": 470
+    },
+    {
+      "epoch": 2.091514143094842,
+      "grad_norm": 11855.4306640625,
+      "learning_rate": 1.8039538714991762e-05,
+      "loss": 0.3321,
+      "step": 471
+    },
+    {
+      "epoch": 2.095951192457016,
+      "grad_norm": 8311.7265625,
+      "learning_rate": 1.7957166392092258e-05,
+      "loss": 0.3195,
+      "step": 472
+    },
+    {
+      "epoch": 2.10038824181919,
+      "grad_norm": 4663.2587890625,
+      "learning_rate": 1.7874794069192753e-05,
+      "loss": 0.3096,
+      "step": 473
+    },
+    {
+      "epoch": 2.1048252911813643,
+      "grad_norm": 3929.626708984375,
+      "learning_rate": 1.779242174629325e-05,
+      "loss": 0.312,
+      "step": 474
+    },
+    {
+      "epoch": 2.1092623405435384,
+      "grad_norm": 5304.91552734375,
+      "learning_rate": 1.771004942339374e-05,
+      "loss": 0.3003,
+      "step": 475
+    },
+    {
+      "epoch": 2.1136993899057126,
+      "grad_norm": 5304.91552734375,
+      "learning_rate": 1.771004942339374e-05,
+      "loss": 0.2915,
+      "step": 476
+    },
+    {
+      "epoch": 2.1181364392678868,
+      "grad_norm": 4898.5791015625,
+      "learning_rate": 1.7627677100494236e-05,
+      "loss": 0.3322,
+      "step": 477
+    },
+    {
+      "epoch": 2.122573488630061,
+      "grad_norm": 3694.562744140625,
+      "learning_rate": 1.754530477759473e-05,
+      "loss": 0.3538,
+      "step": 478
+    },
+    {
+      "epoch": 2.127010537992235,
+      "grad_norm": 2915.031982421875,
+      "learning_rate": 1.7462932454695224e-05,
+      "loss": 0.2978,
+      "step": 479
+    },
+    {
+      "epoch": 2.131447587354409,
+      "grad_norm": 6939.525390625,
+      "learning_rate": 1.7380560131795716e-05,
+      "loss": 0.3504,
+      "step": 480
+    },
+    {
+      "epoch": 2.1358846367165834,
+      "grad_norm": 8300.626953125,
+      "learning_rate": 1.729818780889621e-05,
+      "loss": 0.4364,
+      "step": 481
+    },
+    {
+      "epoch": 2.1403216860787575,
+      "grad_norm": 8347.9306640625,
+      "learning_rate": 1.7215815485996707e-05,
+      "loss": 0.3631,
+      "step": 482
+    },
+    {
+      "epoch": 2.1447587354409317,
+      "grad_norm": 7728.7734375,
+      "learning_rate": 1.7133443163097202e-05,
+      "loss": 0.3726,
+      "step": 483
+    },
+    {
+      "epoch": 2.149195784803106,
+      "grad_norm": 32386.455078125,
+      "learning_rate": 1.7051070840197694e-05,
+      "loss": 0.3709,
+      "step": 484
+    },
+    {
+      "epoch": 2.15363283416528,
+      "grad_norm": 6187.22314453125,
+      "learning_rate": 1.696869851729819e-05,
+      "loss": 0.374,
+      "step": 485
+    },
+    {
+      "epoch": 2.158069883527454,
+      "grad_norm": 7320.59619140625,
+      "learning_rate": 1.6886326194398682e-05,
+      "loss": 0.3368,
+      "step": 486
+    },
+    {
+      "epoch": 2.1625069328896283,
+      "grad_norm": 14155.255859375,
+      "learning_rate": 1.6803953871499177e-05,
+      "loss": 0.3474,
+      "step": 487
+    },
+    {
+      "epoch": 2.1669439822518024,
+      "grad_norm": 4782.28857421875,
+      "learning_rate": 1.6721581548599673e-05,
+      "loss": 0.3044,
+      "step": 488
+    },
+    {
+      "epoch": 2.1713810316139766,
+      "grad_norm": 6132.2705078125,
+      "learning_rate": 1.6639209225700165e-05,
+      "loss": 0.4248,
+      "step": 489
+    },
+    {
+      "epoch": 2.1758180809761507,
+      "grad_norm": 7415.359375,
+      "learning_rate": 1.655683690280066e-05,
+      "loss": 0.3069,
+      "step": 490
+    },
+    {
+      "epoch": 2.180255130338325,
+      "grad_norm": 10917.0244140625,
+      "learning_rate": 1.6474464579901152e-05,
+      "loss": 0.325,
+      "step": 491
+    },
+    {
+      "epoch": 2.184692179700499,
+      "grad_norm": 12631.1201171875,
+      "learning_rate": 1.6392092257001648e-05,
+      "loss": 0.3204,
+      "step": 492
+    },
+    {
+      "epoch": 2.189129229062673,
+      "grad_norm": 3821.25927734375,
+      "learning_rate": 1.630971993410214e-05,
+      "loss": 0.2765,
+      "step": 493
+    },
+    {
+      "epoch": 2.1935662784248473,
+      "grad_norm": 10865.91796875,
+      "learning_rate": 1.6227347611202635e-05,
+      "loss": 0.254,
+      "step": 494
+    },
+    {
+      "epoch": 2.1980033277870215,
+      "grad_norm": 10243.947265625,
+      "learning_rate": 1.614497528830313e-05,
+      "loss": 0.2624,
+      "step": 495
+    },
+    {
+      "epoch": 2.2024403771491956,
+      "grad_norm": 7441.53271484375,
+      "learning_rate": 1.6062602965403626e-05,
+      "loss": 0.3426,
+      "step": 496
+    },
+    {
+      "epoch": 2.2068774265113698,
+      "grad_norm": 9427.314453125,
+      "learning_rate": 1.598023064250412e-05,
+      "loss": 0.3022,
+      "step": 497
+    },
+    {
+      "epoch": 2.211314475873544,
+      "grad_norm": 14851.20703125,
+      "learning_rate": 1.5897858319604614e-05,
+      "loss": 0.3354,
+      "step": 498
+    },
+    {
+      "epoch": 2.215751525235718,
+      "grad_norm": 13256.5078125,
+      "learning_rate": 1.5815485996705106e-05,
+      "loss": 0.3098,
+      "step": 499
+    },
+    {
+      "epoch": 2.220188574597892,
+      "grad_norm": 5459.83203125,
+      "learning_rate": 1.57331136738056e-05,
+      "loss": 0.2898,
+      "step": 500
+    },
+    {
+      "epoch": 2.2246256239600664,
+      "grad_norm": 8498.4716796875,
+      "learning_rate": 1.5650741350906097e-05,
+      "loss": 0.3458,
+      "step": 501
+    },
+    {
+      "epoch": 2.2290626733222405,
+      "grad_norm": 6474.73779296875,
+      "learning_rate": 1.5568369028006592e-05,
+      "loss": 0.2876,
+      "step": 502
+    },
+    {
+      "epoch": 2.2334997226844147,
+      "grad_norm": 11498.4736328125,
+      "learning_rate": 1.5485996705107084e-05,
+      "loss": 0.3115,
+      "step": 503
+    },
+    {
+      "epoch": 2.237936772046589,
+      "grad_norm": 9938.4833984375,
+      "learning_rate": 1.540362438220758e-05,
+      "loss": 0.3881,
+      "step": 504
+    },
+    {
+      "epoch": 2.242373821408763,
+      "grad_norm": 7235.1689453125,
+      "learning_rate": 1.5321252059308072e-05,
+      "loss": 0.2764,
+      "step": 505
+    },
+    {
+      "epoch": 2.246810870770937,
+      "grad_norm": 3982.308349609375,
+      "learning_rate": 1.5238879736408567e-05,
+      "loss": 0.2793,
+      "step": 506
+    },
+    {
+      "epoch": 2.2512479201331113,
+      "grad_norm": 10787.408203125,
+      "learning_rate": 1.5156507413509061e-05,
+      "loss": 0.3168,
+      "step": 507
+    },
+    {
+      "epoch": 2.2556849694952854,
+      "grad_norm": 12415.60546875,
+      "learning_rate": 1.5074135090609557e-05,
+      "loss": 0.4268,
+      "step": 508
+    },
+    {
+      "epoch": 2.2601220188574596,
+      "grad_norm": 7812.7578125,
+      "learning_rate": 1.4991762767710049e-05,
+      "loss": 0.4256,
+      "step": 509
+    },
+    {
+      "epoch": 2.2645590682196337,
+      "grad_norm": 9457.279296875,
+      "learning_rate": 1.4909390444810544e-05,
+      "loss": 0.3838,
+      "step": 510
+    },
+    {
+      "epoch": 2.268996117581808,
+      "grad_norm": 4898.96337890625,
+      "learning_rate": 1.4827018121911038e-05,
+      "loss": 0.2913,
+      "step": 511
+    },
+    {
+      "epoch": 2.273433166943982,
+      "grad_norm": 8582.72265625,
+      "learning_rate": 1.4744645799011533e-05,
+      "loss": 0.3204,
+      "step": 512
+    },
+    {
+      "epoch": 2.277870216306156,
+      "grad_norm": 3687.293212890625,
+      "learning_rate": 1.4662273476112027e-05,
+      "loss": 0.3006,
+      "step": 513
+    },
+    {
+      "epoch": 2.2823072656683308,
+      "grad_norm": 7922.71728515625,
+      "learning_rate": 1.4579901153212522e-05,
+      "loss": 0.2804,
+      "step": 514
+    },
+    {
+      "epoch": 2.286744315030505,
+      "grad_norm": 5794.0966796875,
+      "learning_rate": 1.4497528830313015e-05,
+      "loss": 0.3516,
+      "step": 515
+    },
+    {
+      "epoch": 2.291181364392679,
+      "grad_norm": 8164.35009765625,
+      "learning_rate": 1.441515650741351e-05,
+      "loss": 0.27,
+      "step": 516
+    },
+    {
+      "epoch": 2.295618413754853,
+      "grad_norm": 4838.0634765625,
+      "learning_rate": 1.4332784184514004e-05,
+      "loss": 0.3082,
+      "step": 517
+    },
+    {
+      "epoch": 2.3000554631170274,
+      "grad_norm": 13034.2333984375,
+      "learning_rate": 1.42504118616145e-05,
+      "loss": 0.3339,
+      "step": 518
+    },
+    {
+      "epoch": 2.3044925124792015,
+      "grad_norm": 6675.81884765625,
+      "learning_rate": 1.4168039538714991e-05,
+      "loss": 0.3691,
+      "step": 519
+    },
+    {
+      "epoch": 2.3089295618413757,
+      "grad_norm": 3921.32861328125,
+      "learning_rate": 1.4085667215815487e-05,
+      "loss": 0.3651,
+      "step": 520
+    },
+    {
+      "epoch": 2.31336661120355,
+      "grad_norm": 18242.69921875,
+      "learning_rate": 1.400329489291598e-05,
+      "loss": 0.2662,
+      "step": 521
+    },
+    {
+      "epoch": 2.317803660565724,
+      "grad_norm": 6525.142578125,
+      "learning_rate": 1.3920922570016476e-05,
+      "loss": 0.3322,
+      "step": 522
+    },
+    {
+      "epoch": 2.322240709927898,
+      "grad_norm": 6435.427734375,
+      "learning_rate": 1.3838550247116968e-05,
+      "loss": 0.3606,
+      "step": 523
+    },
+    {
+      "epoch": 2.3266777592900723,
+      "grad_norm": 3723.80712890625,
+      "learning_rate": 1.3756177924217463e-05,
+      "loss": 0.3307,
+      "step": 524
+    },
+    {
+      "epoch": 2.3311148086522464,
+      "grad_norm": 17771.19921875,
+      "learning_rate": 1.3673805601317957e-05,
+      "loss": 0.3151,
+      "step": 525
+    },
+    {
+      "epoch": 2.3355518580144206,
+      "grad_norm": 7537.1484375,
+      "learning_rate": 1.3591433278418453e-05,
+      "loss": 0.333,
+      "step": 526
+    },
+    {
+      "epoch": 2.3399889073765947,
+      "grad_norm": 18066.201171875,
+      "learning_rate": 1.3509060955518946e-05,
+      "loss": 0.2967,
+      "step": 527
+    },
+    {
+      "epoch": 2.344425956738769,
+      "grad_norm": 6065.0693359375,
+      "learning_rate": 1.3426688632619442e-05,
+      "loss": 0.3885,
+      "step": 528
+    },
+    {
+      "epoch": 2.348863006100943,
+      "grad_norm": 4104.03076171875,
+      "learning_rate": 1.3344316309719934e-05,
+      "loss": 0.2722,
+      "step": 529
+    },
+    {
+      "epoch": 2.353300055463117,
+      "grad_norm": 3189.949462890625,
+      "learning_rate": 1.326194398682043e-05,
+      "loss": 0.2994,
+      "step": 530
+    },
+    {
+      "epoch": 2.3577371048252913,
+      "grad_norm": 3484.22607421875,
+      "learning_rate": 1.3179571663920923e-05,
+      "loss": 0.3065,
+      "step": 531
+    },
+    {
+      "epoch": 2.3621741541874655,
+      "grad_norm": 12326.0732421875,
+      "learning_rate": 1.3097199341021419e-05,
+      "loss": 0.2309,
+      "step": 532
+    },
+    {
+      "epoch": 2.3666112035496396,
+      "grad_norm": 6139.84326171875,
+      "learning_rate": 1.301482701812191e-05,
+      "loss": 0.3397,
+      "step": 533
+    },
+    {
+      "epoch": 2.3710482529118138,
+      "grad_norm": 6288.798828125,
+      "learning_rate": 1.2932454695222406e-05,
+      "loss": 0.3414,
+      "step": 534
+    },
+    {
+      "epoch": 2.375485302273988,
+      "grad_norm": 54391.0,
+      "learning_rate": 1.28500823723229e-05,
+      "loss": 0.3345,
+      "step": 535
+    },
+    {
+      "epoch": 2.379922351636162,
+      "grad_norm": 3919.669189453125,
+      "learning_rate": 1.2767710049423395e-05,
+      "loss": 0.3739,
+      "step": 536
+    },
+    {
+      "epoch": 2.384359400998336,
+      "grad_norm": 7220.857421875,
+      "learning_rate": 1.2685337726523888e-05,
+      "loss": 0.3661,
+      "step": 537
+    },
+    {
+      "epoch": 2.3887964503605104,
+      "grad_norm": 6530.623046875,
+      "learning_rate": 1.2602965403624385e-05,
+      "loss": 0.3154,
+      "step": 538
+    },
+    {
+      "epoch": 2.3932334997226845,
+      "grad_norm": 12763.6337890625,
+      "learning_rate": 1.2520593080724877e-05,
+      "loss": 0.3999,
+      "step": 539
+    },
+    {
+      "epoch": 2.3976705490848587,
+      "grad_norm": 3606.4228515625,
+      "learning_rate": 1.243822075782537e-05,
+      "loss": 0.2806,
+      "step": 540
+    },
+    {
+      "epoch": 2.402107598447033,
+      "grad_norm": 2511.720703125,
+      "learning_rate": 1.2355848434925866e-05,
+      "loss": 0.34,
+      "step": 541
+    },
+    {
+      "epoch": 2.406544647809207,
+      "grad_norm": 6545.259765625,
+      "learning_rate": 1.227347611202636e-05,
+      "loss": 0.3057,
+      "step": 542
+    },
+    {
+      "epoch": 2.410981697171381,
+      "grad_norm": 7115.1259765625,
+      "learning_rate": 1.2191103789126853e-05,
+      "loss": 0.2434,
+      "step": 543
+    },
+    {
+      "epoch": 2.4154187465335553,
+      "grad_norm": 2655.114501953125,
+      "learning_rate": 1.2108731466227347e-05,
+      "loss": 0.2718,
+      "step": 544
+    },
+    {
+      "epoch": 2.4198557958957294,
+      "grad_norm": 4364.2646484375,
+      "learning_rate": 1.2026359143327843e-05,
+      "loss": 0.366,
+      "step": 545
+    },
+    {
+      "epoch": 2.4242928452579036,
+      "grad_norm": 10678.7783203125,
+      "learning_rate": 1.1943986820428336e-05,
+      "loss": 0.3293,
+      "step": 546
+    },
+    {
+      "epoch": 2.4287298946200777,
+      "grad_norm": 2421.989013671875,
+      "learning_rate": 1.186161449752883e-05,
+      "loss": 0.2941,
+      "step": 547
+    },
+    {
+      "epoch": 2.433166943982252,
+      "grad_norm": 2855.467041015625,
+      "learning_rate": 1.1779242174629326e-05,
+      "loss": 0.374,
+      "step": 548
+    },
+    {
+      "epoch": 2.437603993344426,
+      "grad_norm": 13584.5498046875,
+      "learning_rate": 1.169686985172982e-05,
+      "loss": 0.3483,
+      "step": 549
+    },
+    {
+      "epoch": 2.4420410427066,
+      "grad_norm": 8351.8916015625,
+      "learning_rate": 1.1614497528830313e-05,
+      "loss": 0.2939,
+      "step": 550
+    },
+    {
+      "epoch": 2.4464780920687743,
+      "grad_norm": 9228.5634765625,
+      "learning_rate": 1.1532125205930807e-05,
+      "loss": 0.3247,
+      "step": 551
+    },
+    {
+      "epoch": 2.4509151414309485,
+      "grad_norm": 6275.79052734375,
+      "learning_rate": 1.1449752883031302e-05,
+      "loss": 0.2678,
+      "step": 552
+    },
+    {
+      "epoch": 2.4553521907931226,
+      "grad_norm": 13641.73828125,
+      "learning_rate": 1.1367380560131796e-05,
+      "loss": 0.2737,
+      "step": 553
+    },
+    {
+      "epoch": 2.4597892401552968,
+      "grad_norm": 6121.5791015625,
+      "learning_rate": 1.128500823723229e-05,
+      "loss": 0.2077,
+      "step": 554
+    },
+    {
+      "epoch": 2.464226289517471,
+      "grad_norm": 3010.915283203125,
+      "learning_rate": 1.1202635914332785e-05,
+      "loss": 0.2821,
+      "step": 555
+    },
+    {
+      "epoch": 2.468663338879645,
+      "grad_norm": 6493.578125,
+      "learning_rate": 1.112026359143328e-05,
+      "loss": 0.2959,
+      "step": 556
+    },
+    {
+      "epoch": 2.473100388241819,
+      "grad_norm": 6420.4365234375,
+      "learning_rate": 1.1037891268533773e-05,
+      "loss": 0.2998,
+      "step": 557
+    },
+    {
+      "epoch": 2.4775374376039934,
+      "grad_norm": 6401.4375,
+      "learning_rate": 1.0955518945634267e-05,
+      "loss": 0.3112,
+      "step": 558
+    },
+    {
+      "epoch": 2.4819744869661675,
+      "grad_norm": 8719.298828125,
+      "learning_rate": 1.0873146622734762e-05,
+      "loss": 0.2996,
+      "step": 559
+    },
+    {
+      "epoch": 2.4864115363283417,
+      "grad_norm": 9697.6103515625,
+      "learning_rate": 1.0790774299835256e-05,
+      "loss": 0.3148,
+      "step": 560
+    },
+    {
+      "epoch": 2.490848585690516,
+      "grad_norm": 9796.7890625,
+      "learning_rate": 1.070840197693575e-05,
+      "loss": 0.324,
+      "step": 561
+    },
+    {
+      "epoch": 2.49528563505269,
+      "grad_norm": 8435.099609375,
+      "learning_rate": 1.0626029654036245e-05,
+      "loss": 0.392,
+      "step": 562
+    },
+    {
+      "epoch": 2.499722684414864,
+      "grad_norm": 5161.76220703125,
+      "learning_rate": 1.0543657331136739e-05,
+      "loss": 0.3134,
+      "step": 563
+    },
+    {
+      "epoch": 2.5041597337770383,
+      "grad_norm": 6516.07763671875,
+      "learning_rate": 1.0461285008237233e-05,
+      "loss": 0.3281,
+      "step": 564
+    },
+    {
+      "epoch": 2.5085967831392124,
+      "grad_norm": 8268.9140625,
+      "learning_rate": 1.0378912685337726e-05,
+      "loss": 0.2645,
+      "step": 565
+    },
+    {
+      "epoch": 2.5130338325013866,
+      "grad_norm": 11513.44921875,
+      "learning_rate": 1.0296540362438222e-05,
+      "loss": 0.4041,
+      "step": 566
+    },
+    {
+      "epoch": 2.5174708818635607,
+      "grad_norm": 3220.71533203125,
+      "learning_rate": 1.0214168039538716e-05,
+      "loss": 0.3071,
+      "step": 567
+    },
+    {
+      "epoch": 2.521907931225735,
+      "grad_norm": 6902.10791015625,
+      "learning_rate": 1.013179571663921e-05,
+      "loss": 0.334,
+      "step": 568
+    },
+    {
+      "epoch": 2.526344980587909,
+      "grad_norm": 13407.0771484375,
+      "learning_rate": 1.0049423393739705e-05,
+      "loss": 0.3135,
+      "step": 569
+    },
+    {
+      "epoch": 2.530782029950083,
+      "grad_norm": 23491.77734375,
+      "learning_rate": 9.967051070840199e-06,
+      "loss": 0.3494,
+      "step": 570
+    },
+    {
+      "epoch": 2.5352190793122573,
+      "grad_norm": 12857.658203125,
+      "learning_rate": 9.884678747940692e-06,
+      "loss": 0.333,
+      "step": 571
+    },
+    {
+      "epoch": 2.5396561286744315,
+      "grad_norm": 10788.306640625,
+      "learning_rate": 9.802306425041186e-06,
+      "loss": 0.4399,
+      "step": 572
+    },
+    {
+      "epoch": 2.5440931780366056,
+      "grad_norm": 7962.6728515625,
+      "learning_rate": 9.719934102141682e-06,
+      "loss": 0.3411,
+      "step": 573
+    },
+    {
+      "epoch": 2.5485302273987798,
+      "grad_norm": 13886.4052734375,
+      "learning_rate": 9.637561779242175e-06,
+      "loss": 0.3667,
+      "step": 574
+    },
+    {
+      "epoch": 2.552967276760954,
+      "grad_norm": 4301.244140625,
+      "learning_rate": 9.55518945634267e-06,
+      "loss": 0.2927,
+      "step": 575
+    },
+    {
+      "epoch": 2.557404326123128,
+      "grad_norm": 9445.1884765625,
+      "learning_rate": 9.472817133443165e-06,
+      "loss": 0.3176,
+      "step": 576
+    },
+    {
+      "epoch": 2.561841375485302,
+      "grad_norm": 5979.98583984375,
+      "learning_rate": 9.390444810543658e-06,
+      "loss": 0.2872,
+      "step": 577
+    },
+    {
+      "epoch": 2.5662784248474764,
+      "grad_norm": 10551.0068359375,
+      "learning_rate": 9.308072487644152e-06,
+      "loss": 0.2918,
+      "step": 578
+    },
+    {
+      "epoch": 2.5707154742096505,
+      "grad_norm": 7856.26513671875,
+      "learning_rate": 9.225700164744646e-06,
+      "loss": 0.3165,
+      "step": 579
+    },
+    {
+      "epoch": 2.5751525235718247,
+      "grad_norm": 9975.029296875,
+      "learning_rate": 9.143327841845141e-06,
+      "loss": 0.3306,
+      "step": 580
+    },
+    {
+      "epoch": 2.579589572933999,
+      "grad_norm": 8852.380859375,
+      "learning_rate": 9.060955518945635e-06,
+      "loss": 0.3497,
+      "step": 581
+    },
+    {
+      "epoch": 2.584026622296173,
+      "grad_norm": 17077.798828125,
+      "learning_rate": 8.978583196046129e-06,
+      "loss": 0.2911,
+      "step": 582
+    },
+    {
+      "epoch": 2.588463671658347,
+      "grad_norm": 6085.35009765625,
+      "learning_rate": 8.896210873146624e-06,
+      "loss": 0.2635,
+      "step": 583
+    },
+    {
+      "epoch": 2.5929007210205213,
+      "grad_norm": 7225.9951171875,
+      "learning_rate": 8.813838550247118e-06,
+      "loss": 0.2898,
+      "step": 584
+    },
+    {
+      "epoch": 2.5973377703826954,
+      "grad_norm": 7063.8291015625,
+      "learning_rate": 8.731466227347612e-06,
+      "loss": 0.3104,
+      "step": 585
+    },
+    {
+      "epoch": 2.6017748197448696,
+      "grad_norm": 11306.908203125,
+      "learning_rate": 8.649093904448106e-06,
+      "loss": 0.3806,
+      "step": 586
+    },
+    {
+      "epoch": 2.6062118691070437,
+      "grad_norm": 8992.98046875,
+      "learning_rate": 8.566721581548601e-06,
+      "loss": 0.2919,
+      "step": 587
+    },
+    {
+      "epoch": 2.610648918469218,
+      "grad_norm": 3013.324462890625,
+      "learning_rate": 8.484349258649095e-06,
+      "loss": 0.3299,
+      "step": 588
+    },
+    {
+      "epoch": 2.615085967831392,
+      "grad_norm": 11653.1875,
+      "learning_rate": 8.401976935749589e-06,
+      "loss": 0.3308,
+      "step": 589
+    },
+    {
+      "epoch": 2.619523017193566,
+      "grad_norm": 9767.2587890625,
+      "learning_rate": 8.319604612850082e-06,
+      "loss": 0.3387,
+      "step": 590
+    },
+    {
+      "epoch": 2.6239600665557403,
+      "grad_norm": 16572.38671875,
+      "learning_rate": 8.237232289950576e-06,
+      "loss": 0.2576,
+      "step": 591
+    },
+    {
+      "epoch": 2.6283971159179145,
+      "grad_norm": 7412.2734375,
+      "learning_rate": 8.15485996705107e-06,
+      "loss": 0.314,
+      "step": 592
+    },
+    {
+      "epoch": 2.6328341652800886,
+      "grad_norm": 9435.06640625,
+      "learning_rate": 8.072487644151565e-06,
+      "loss": 0.2829,
+      "step": 593
+    },
+    {
+      "epoch": 2.6372712146422628,
+      "grad_norm": 10025.59765625,
+      "learning_rate": 7.99011532125206e-06,
+      "loss": 0.2445,
+      "step": 594
+    },
+    {
+      "epoch": 2.641708264004437,
+      "grad_norm": 12398.07421875,
+      "learning_rate": 7.907742998352553e-06,
+      "loss": 0.2961,
+      "step": 595
+    },
+    {
+      "epoch": 2.646145313366611,
+      "grad_norm": 4652.546875,
+      "learning_rate": 7.825370675453048e-06,
+      "loss": 0.3698,
+      "step": 596
+    },
+    {
+      "epoch": 2.6505823627287852,
+      "grad_norm": 8505.6650390625,
+      "learning_rate": 7.742998352553542e-06,
+      "loss": 0.2797,
+      "step": 597
+    },
+    {
+      "epoch": 2.6550194120909594,
+      "grad_norm": 4741.369140625,
+      "learning_rate": 7.660626029654036e-06,
+      "loss": 0.2623,
+      "step": 598
+    },
+    {
+      "epoch": 2.6594564614531335,
+      "grad_norm": 4993.56689453125,
+      "learning_rate": 7.5782537067545305e-06,
+      "loss": 0.3256,
+      "step": 599
+    },
+    {
+      "epoch": 2.6638935108153077,
+      "grad_norm": 3874.74755859375,
+      "learning_rate": 7.495881383855024e-06,
+      "loss": 0.2814,
+      "step": 600
+    },
+    {
+      "epoch": 2.668330560177482,
+      "grad_norm": 6255.4013671875,
+      "learning_rate": 7.413509060955519e-06,
+      "loss": 0.4212,
+      "step": 601
+    },
+    {
+      "epoch": 2.672767609539656,
+      "grad_norm": 4960.4990234375,
+      "learning_rate": 7.3311367380560135e-06,
+      "loss": 0.3611,
+      "step": 602
+    },
+    {
+      "epoch": 2.67720465890183,
+      "grad_norm": 11081.822265625,
+      "learning_rate": 7.248764415156507e-06,
+      "loss": 0.3339,
+      "step": 603
+    },
+    {
+      "epoch": 2.6816417082640043,
+      "grad_norm": 9572.4462890625,
+      "learning_rate": 7.166392092257002e-06,
+      "loss": 0.2685,
+      "step": 604
+    },
+    {
+      "epoch": 2.6860787576261784,
+      "grad_norm": 8953.3935546875,
+      "learning_rate": 7.084019769357496e-06,
+      "loss": 0.3243,
+      "step": 605
+    },
+    {
+      "epoch": 2.690515806988353,
+      "grad_norm": 7798.02392578125,
+      "learning_rate": 7.00164744645799e-06,
+      "loss": 0.3569,
+      "step": 606
+    },
+    {
+      "epoch": 2.694952856350527,
+      "grad_norm": 3023.310546875,
+      "learning_rate": 6.919275123558484e-06,
+      "loss": 0.2798,
+      "step": 607
+    },
+    {
+      "epoch": 2.6993899057127013,
+      "grad_norm": 5004.8828125,
+      "learning_rate": 6.836902800658979e-06,
+      "loss": 0.3464,
+      "step": 608
+    },
+    {
+      "epoch": 2.7038269550748755,
+      "grad_norm": 5235.65869140625,
+      "learning_rate": 6.754530477759473e-06,
+      "loss": 0.2856,
+      "step": 609
+    },
+    {
+      "epoch": 2.7082640044370496,
+      "grad_norm": 13685.0625,
+      "learning_rate": 6.672158154859967e-06,
+      "loss": 0.33,
+      "step": 610
+    },
+    {
+      "epoch": 2.7127010537992238,
+      "grad_norm": 1542.4080810546875,
+      "learning_rate": 6.589785831960462e-06,
+      "loss": 0.2955,
+      "step": 611
+    },
+    {
+      "epoch": 2.717138103161398,
+      "grad_norm": 10128.8994140625,
+      "learning_rate": 6.507413509060955e-06,
+      "loss": 0.4246,
+      "step": 612
+    },
+    {
+      "epoch": 2.721575152523572,
+      "grad_norm": 6730.404296875,
+      "learning_rate": 6.42504118616145e-06,
+      "loss": 0.3176,
+      "step": 613
+    },
+    {
+      "epoch": 2.7260122018857462,
+      "grad_norm": 8910.18359375,
+      "learning_rate": 6.342668863261944e-06,
+      "loss": 0.3465,
+      "step": 614
+    },
+    {
+      "epoch": 2.7304492512479204,
+      "grad_norm": 7740.04150390625,
+      "learning_rate": 6.260296540362438e-06,
+      "loss": 0.3097,
+      "step": 615
+    },
+    {
+      "epoch": 2.7348863006100945,
+      "grad_norm": 3845.580810546875,
+      "learning_rate": 6.177924217462933e-06,
+      "loss": 0.2703,
+      "step": 616
+    },
+    {
+      "epoch": 2.7393233499722687,
+      "grad_norm": 20928.3203125,
+      "learning_rate": 6.095551894563427e-06,
+      "loss": 0.2786,
+      "step": 617
+    },
+    {
+      "epoch": 2.743760399334443,
+      "grad_norm": 5004.64404296875,
+      "learning_rate": 6.013179571663921e-06,
+      "loss": 0.3049,
+      "step": 618
+    },
+    {
+      "epoch": 2.748197448696617,
+      "grad_norm": 6614.98095703125,
+      "learning_rate": 5.930807248764415e-06,
+      "loss": 0.3095,
+      "step": 619
+    },
+    {
+      "epoch": 2.752634498058791,
+      "grad_norm": 4118.806640625,
+      "learning_rate": 5.84843492586491e-06,
+      "loss": 0.3274,
+      "step": 620
+    },
+    {
+      "epoch": 2.7570715474209653,
+      "grad_norm": 6724.39892578125,
+      "learning_rate": 5.7660626029654035e-06,
+      "loss": 0.3081,
+      "step": 621
+    },
+    {
+      "epoch": 2.7615085967831394,
+      "grad_norm": 20886.7265625,
+      "learning_rate": 5.683690280065898e-06,
+      "loss": 0.4122,
+      "step": 622
+    },
+    {
+      "epoch": 2.7659456461453136,
+      "grad_norm": 12638.28125,
+      "learning_rate": 5.601317957166393e-06,
+      "loss": 0.3199,
+      "step": 623
+    },
+    {
+      "epoch": 2.7703826955074877,
+      "grad_norm": 17803.33984375,
+      "learning_rate": 5.5189456342668865e-06,
+      "loss": 0.3383,
+      "step": 624
+    },
+    {
+      "epoch": 2.774819744869662,
+      "grad_norm": 6198.55810546875,
+      "learning_rate": 5.436573311367381e-06,
+      "loss": 0.3577,
+      "step": 625
+    },
+    {
+      "epoch": 2.779256794231836,
+      "grad_norm": 8318.6513671875,
+      "learning_rate": 5.354200988467875e-06,
+      "loss": 0.3414,
+      "step": 626
+    },
+    {
+      "epoch": 2.78369384359401,
+      "grad_norm": 12702.873046875,
+      "learning_rate": 5.2718286655683695e-06,
+      "loss": 0.3262,
+      "step": 627
+    },
+    {
+      "epoch": 2.7881308929561843,
+      "grad_norm": 8501.73046875,
+      "learning_rate": 5.189456342668863e-06,
+      "loss": 0.3186,
+      "step": 628
+    },
+    {
+      "epoch": 2.7925679423183585,
+      "grad_norm": 11130.0322265625,
+      "learning_rate": 5.107084019769358e-06,
+      "loss": 0.3866,
+      "step": 629
+    },
+    {
+      "epoch": 2.7970049916805326,
+      "grad_norm": 15967.6943359375,
+      "learning_rate": 5.0247116968698524e-06,
+      "loss": 0.3142,
+      "step": 630
+    },
+    {
+      "epoch": 2.8014420410427068,
+      "grad_norm": 5288.306640625,
+      "learning_rate": 4.942339373970346e-06,
+      "loss": 0.3106,
+      "step": 631
+    },
+    {
+      "epoch": 2.805879090404881,
+      "grad_norm": 7328.09716796875,
+      "learning_rate": 4.859967051070841e-06,
+      "loss": 0.3427,
+      "step": 632
+    },
+    {
+      "epoch": 2.810316139767055,
+      "grad_norm": 11073.931640625,
+      "learning_rate": 4.777594728171335e-06,
+      "loss": 0.3447,
+      "step": 633
+    },
+    {
+      "epoch": 2.8147531891292292,
+      "grad_norm": 6741.98291015625,
+      "learning_rate": 4.695222405271829e-06,
+      "loss": 0.2998,
+      "step": 634
+    },
+    {
+      "epoch": 2.8191902384914034,
+      "grad_norm": 6070.072265625,
+      "learning_rate": 4.612850082372323e-06,
+      "loss": 0.2639,
+      "step": 635
+    },
+    {
+      "epoch": 2.8236272878535775,
+      "grad_norm": 5818.5439453125,
+      "learning_rate": 4.5304777594728176e-06,
+      "loss": 0.3684,
+      "step": 636
+    },
+    {
+      "epoch": 2.8280643372157517,
+      "grad_norm": 4406.9482421875,
+      "learning_rate": 4.448105436573312e-06,
+      "loss": 0.3343,
+      "step": 637
+    },
+    {
+      "epoch": 2.832501386577926,
+      "grad_norm": 10366.3935546875,
+      "learning_rate": 4.365733113673806e-06,
+      "loss": 0.3558,
+      "step": 638
+    },
+    {
+      "epoch": 2.8369384359401,
+      "grad_norm": 10846.302734375,
+      "learning_rate": 4.2833607907743006e-06,
+      "loss": 0.2879,
+      "step": 639
+    },
+    {
+      "epoch": 2.841375485302274,
+      "grad_norm": 8078.3984375,
+      "learning_rate": 4.200988467874794e-06,
+      "loss": 0.2548,
+      "step": 640
+    },
+    {
+      "epoch": 2.8458125346644483,
+      "grad_norm": 7038.9365234375,
+      "learning_rate": 4.118616144975288e-06,
+      "loss": 0.3659,
+      "step": 641
+    },
+    {
+      "epoch": 2.8502495840266224,
+      "grad_norm": 5697.41162109375,
+      "learning_rate": 4.036243822075783e-06,
+      "loss": 0.3234,
+      "step": 642
+    },
+    {
+      "epoch": 2.8546866333887966,
+      "grad_norm": 23884.875,
+      "learning_rate": 3.9538714991762765e-06,
+      "loss": 0.2487,
+      "step": 643
+    },
+    {
+      "epoch": 2.8591236827509707,
+      "grad_norm": 6687.5869140625,
+      "learning_rate": 3.871499176276771e-06,
+      "loss": 0.3185,
+      "step": 644
+    },
+    {
+      "epoch": 2.863560732113145,
+      "grad_norm": 13451.443359375,
+      "learning_rate": 3.7891268533772653e-06,
+      "loss": 0.3085,
+      "step": 645
+    },
+    {
+      "epoch": 2.867997781475319,
+      "grad_norm": 10280.291015625,
+      "learning_rate": 3.7067545304777594e-06,
+      "loss": 0.2995,
+      "step": 646
+    },
+    {
+      "epoch": 2.872434830837493,
+      "grad_norm": 7421.79296875,
+      "learning_rate": 3.6243822075782536e-06,
+      "loss": 0.4275,
+      "step": 647
+    },
+    {
+      "epoch": 2.8768718801996673,
+      "grad_norm": 6324.1015625,
+      "learning_rate": 3.542009884678748e-06,
+      "loss": 0.349,
+      "step": 648
+    },
+    {
+      "epoch": 2.8813089295618415,
+      "grad_norm": 15548.0849609375,
+      "learning_rate": 3.459637561779242e-06,
+      "loss": 0.4313,
+      "step": 649
+    },
+    {
+      "epoch": 2.8857459789240156,
+      "grad_norm": 7416.5673828125,
+      "learning_rate": 3.3772652388797366e-06,
+      "loss": 0.356,
+      "step": 650
+    },
+    {
+      "epoch": 2.89018302828619,
+      "grad_norm": 6816.31591796875,
+      "learning_rate": 3.294892915980231e-06,
+      "loss": 0.244,
+      "step": 651
+    },
+    {
+      "epoch": 2.894620077648364,
+      "grad_norm": 9225.623046875,
+      "learning_rate": 3.212520593080725e-06,
+      "loss": 0.2713,
+      "step": 652
+    },
+    {
+      "epoch": 2.899057127010538,
+      "grad_norm": 9805.4033203125,
+      "learning_rate": 3.130148270181219e-06,
+      "loss": 0.3101,
+      "step": 653
+    },
+    {
+      "epoch": 2.9034941763727122,
+      "grad_norm": 16117.62890625,
+      "learning_rate": 3.0477759472817134e-06,
+      "loss": 0.3553,
+      "step": 654
+    },
+    {
+      "epoch": 2.9079312257348864,
+      "grad_norm": 3125.488525390625,
+      "learning_rate": 2.9654036243822076e-06,
+      "loss": 0.2807,
+      "step": 655
+    },
+    {
+      "epoch": 2.9123682750970605,
+      "grad_norm": 8943.958984375,
+      "learning_rate": 2.8830313014827017e-06,
+      "loss": 0.314,
+      "step": 656
+    },
+    {
+      "epoch": 2.9168053244592347,
+      "grad_norm": 7321.2939453125,
+      "learning_rate": 2.8006589785831964e-06,
+      "loss": 0.2709,
+      "step": 657
+    },
+    {
+      "epoch": 2.921242373821409,
+      "grad_norm": 4118.58544921875,
+      "learning_rate": 2.7182866556836905e-06,
+      "loss": 0.4568,
+      "step": 658
+    },
+    {
+      "epoch": 2.925679423183583,
+      "grad_norm": 4775.123046875,
+      "learning_rate": 2.6359143327841847e-06,
+      "loss": 0.317,
+      "step": 659
+    },
+    {
+      "epoch": 2.930116472545757,
+      "grad_norm": 4293.9443359375,
+      "learning_rate": 2.553542009884679e-06,
+      "loss": 0.3883,
+      "step": 660
+    },
+    {
+      "epoch": 2.9345535219079313,
+      "grad_norm": 3168.8193359375,
+      "learning_rate": 2.471169686985173e-06,
+      "loss": 0.262,
+      "step": 661
+    },
+    {
+      "epoch": 2.9389905712701054,
+      "grad_norm": 12145.2041015625,
+      "learning_rate": 2.3887973640856673e-06,
+      "loss": 0.2491,
+      "step": 662
+    },
+    {
+      "epoch": 2.9434276206322796,
+      "grad_norm": 7698.5869140625,
+      "learning_rate": 2.3064250411861615e-06,
+      "loss": 0.3309,
+      "step": 663
+    },
+    {
+      "epoch": 2.9478646699944537,
+      "grad_norm": 13563.0078125,
+      "learning_rate": 2.224052718286656e-06,
+      "loss": 0.369,
+      "step": 664
+    },
+    {
+      "epoch": 2.952301719356628,
+      "grad_norm": 9722.326171875,
+      "learning_rate": 2.1416803953871503e-06,
+      "loss": 0.3291,
+      "step": 665
+    },
+    {
+      "epoch": 2.956738768718802,
+      "grad_norm": 6289.4462890625,
+      "learning_rate": 2.059308072487644e-06,
+      "loss": 0.3573,
+      "step": 666
+    },
+    {
+      "epoch": 2.961175818080976,
+      "grad_norm": 14025.2451171875,
+      "learning_rate": 1.9769357495881382e-06,
+      "loss": 0.3194,
+      "step": 667
+    },
+    {
+      "epoch": 2.9656128674431503,
+      "grad_norm": 3258.3564453125,
+      "learning_rate": 1.8945634266886326e-06,
+      "loss": 0.3254,
+      "step": 668
+    },
+    {
+      "epoch": 2.9700499168053245,
+      "grad_norm": 8908.5693359375,
+      "learning_rate": 1.8121911037891268e-06,
+      "loss": 0.2773,
+      "step": 669
+    },
+    {
+      "epoch": 2.9744869661674986,
+      "grad_norm": 4715.8173828125,
+      "learning_rate": 1.729818780889621e-06,
+      "loss": 0.3859,
+      "step": 670
+    },
+    {
+      "epoch": 2.978924015529673,
+      "grad_norm": 24816.103515625,
+      "learning_rate": 1.6474464579901154e-06,
+      "loss": 0.3396,
+      "step": 671
+    },
+    {
+      "epoch": 2.983361064891847,
+      "grad_norm": 5145.150390625,
+      "learning_rate": 1.5650741350906096e-06,
+      "loss": 0.436,
+      "step": 672
+    },
+    {
+      "epoch": 2.987798114254021,
+      "grad_norm": 8119.482421875,
+      "learning_rate": 1.4827018121911038e-06,
+      "loss": 0.3883,
+      "step": 673
+    },
+    {
+      "epoch": 2.9922351636161952,
+      "grad_norm": 6871.0234375,
+      "learning_rate": 1.4003294892915982e-06,
+      "loss": 0.3212,
+      "step": 674
+    },
+    {
+      "epoch": 2.9966722129783694,
+      "grad_norm": 9786.31640625,
+      "learning_rate": 1.3179571663920924e-06,
+      "loss": 0.2953,
+      "step": 675
+    },
+    {
+      "epoch": 2.9966722129783694,
+      "step": 675,
+      "total_flos": 5.085141987879813e+17,
+      "train_loss": 0.26676949086012663,
+      "train_runtime": 25733.113,
+      "train_samples_per_second": 0.841,
+      "train_steps_per_second": 0.026
+    }
+  ],
+  "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": 5.085141987879813e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}