diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4662 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9977298524404086,
+  "eval_steps": 500,
+  "global_step": 660,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003026863412788498,
+      "grad_norm": 3.2617823024465724,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.6874,
+      "step": 1
+    },
+    {
+      "epoch": 0.006053726825576996,
+      "grad_norm": 3.3690474841208706,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.7308,
+      "step": 2
+    },
+    {
+      "epoch": 0.009080590238365494,
+      "grad_norm": 3.079655114317389,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6854,
+      "step": 3
+    },
+    {
+      "epoch": 0.012107453651153992,
+      "grad_norm": 3.015745021914065,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.6928,
+      "step": 4
+    },
+    {
+      "epoch": 0.01513431706394249,
+      "grad_norm": 3.2201683330665576,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.6829,
+      "step": 5
+    },
+    {
+      "epoch": 0.018161180476730987,
+      "grad_norm": 2.967326603871486,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.6867,
+      "step": 6
+    },
+    {
+      "epoch": 0.021188043889519486,
+      "grad_norm": 2.600967128837199,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.6519,
+      "step": 7
+    },
+    {
+      "epoch": 0.024214907302307985,
+      "grad_norm": 2.4173834772799396,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.6402,
+      "step": 8
+    },
+    {
+      "epoch": 0.02724177071509648,
+      "grad_norm": 1.7857041054313472,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.6453,
+      "step": 9
+    },
+    {
+      "epoch": 0.03026863412788498,
+      "grad_norm": 1.567457708313361,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.6074,
+      "step": 10
+    },
+    {
+      "epoch": 0.03329549754067348,
+      "grad_norm": 1.5599868656010474,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.601,
+      "step": 11
+    },
+    {
+      "epoch": 0.036322360953461974,
+      "grad_norm": 1.8068108336570274,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.5934,
+      "step": 12
+    },
+    {
+      "epoch": 0.03934922436625047,
+      "grad_norm": 2.760356706664217,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.6364,
+      "step": 13
+    },
+    {
+      "epoch": 0.04237608777903897,
+      "grad_norm": 2.3369742847198838,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.5749,
+      "step": 14
+    },
+    {
+      "epoch": 0.04540295119182747,
+      "grad_norm": 1.7116730174232366,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.5639,
+      "step": 15
+    },
+    {
+      "epoch": 0.04842981460461597,
+      "grad_norm": 1.466912263807418,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.5847,
+      "step": 16
+    },
+    {
+      "epoch": 0.051456678017404466,
+      "grad_norm": 1.355933841830451,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.5752,
+      "step": 17
+    },
+    {
+      "epoch": 0.05448354143019296,
+      "grad_norm": 1.4264844165916435,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.5678,
+      "step": 18
+    },
+    {
+      "epoch": 0.057510404842981463,
+      "grad_norm": 1.312953420797211,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.5392,
+      "step": 19
+    },
+    {
+      "epoch": 0.06053726825576996,
+      "grad_norm": 1.1684304133038967,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.5576,
+      "step": 20
+    },
+    {
+      "epoch": 0.06356413166855845,
+      "grad_norm": 1.049925959788892,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.551,
+      "step": 21
+    },
+    {
+      "epoch": 0.06659099508134696,
+      "grad_norm": 1.138340312827715,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.5555,
+      "step": 22
+    },
+    {
+      "epoch": 0.06961785849413545,
+      "grad_norm": 1.066004625629727,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.5196,
+      "step": 23
+    },
+    {
+      "epoch": 0.07264472190692395,
+      "grad_norm": 0.9444207836805502,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.5434,
+      "step": 24
+    },
+    {
+      "epoch": 0.07567158531971245,
+      "grad_norm": 1.1412191315568117,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.5039,
+      "step": 25
+    },
+    {
+      "epoch": 0.07869844873250094,
+      "grad_norm": 0.9440313044733073,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.5227,
+      "step": 26
+    },
+    {
+      "epoch": 0.08172531214528944,
+      "grad_norm": 0.9518208812908499,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.4899,
+      "step": 27
+    },
+    {
+      "epoch": 0.08475217555807794,
+      "grad_norm": 0.9319294673291787,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.5177,
+      "step": 28
+    },
+    {
+      "epoch": 0.08777903897086645,
+      "grad_norm": 0.8901571501878328,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.486,
+      "step": 29
+    },
+    {
+      "epoch": 0.09080590238365494,
+      "grad_norm": 0.9857706084521647,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.5281,
+      "step": 30
+    },
+    {
+      "epoch": 0.09383276579644344,
+      "grad_norm": 0.8878307026963279,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 0.4869,
+      "step": 31
+    },
+    {
+      "epoch": 0.09685962920923194,
+      "grad_norm": 0.9349198843070594,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.49,
+      "step": 32
+    },
+    {
+      "epoch": 0.09988649262202043,
+      "grad_norm": 0.9865801860520347,
+      "learning_rate": 1e-05,
+      "loss": 0.4641,
+      "step": 33
+    },
+    {
+      "epoch": 0.10291335603480893,
+      "grad_norm": 0.9215133867240487,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.5066,
+      "step": 34
+    },
+    {
+      "epoch": 0.10594021944759743,
+      "grad_norm": 0.9259211229683464,
+      "learning_rate": 1.0606060606060606e-05,
+      "loss": 0.5056,
+      "step": 35
+    },
+    {
+      "epoch": 0.10896708286038592,
+      "grad_norm": 0.9730936003260622,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.5361,
+      "step": 36
+    },
+    {
+      "epoch": 0.11199394627317442,
+      "grad_norm": 0.8882513069422203,
+      "learning_rate": 1.1212121212121212e-05,
+      "loss": 0.5099,
+      "step": 37
+    },
+    {
+      "epoch": 0.11502080968596293,
+      "grad_norm": 1.0101113876435475,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 0.4904,
+      "step": 38
+    },
+    {
+      "epoch": 0.11804767309875142,
+      "grad_norm": 0.8163259346265521,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 0.5113,
+      "step": 39
+    },
+    {
+      "epoch": 0.12107453651153992,
+      "grad_norm": 0.8808059665770918,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.473,
+      "step": 40
+    },
+    {
+      "epoch": 0.12410139992432842,
+      "grad_norm": 1.009008579133104,
+      "learning_rate": 1.2424242424242425e-05,
+      "loss": 0.4735,
+      "step": 41
+    },
+    {
+      "epoch": 0.1271282633371169,
+      "grad_norm": 0.9007629999697133,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.4881,
+      "step": 42
+    },
+    {
+      "epoch": 0.1301551267499054,
+      "grad_norm": 0.9075998896945083,
+      "learning_rate": 1.3030303030303032e-05,
+      "loss": 0.5171,
+      "step": 43
+    },
+    {
+      "epoch": 0.13318199016269391,
+      "grad_norm": 1.0236420309009964,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4893,
+      "step": 44
+    },
+    {
+      "epoch": 0.1362088535754824,
+      "grad_norm": 0.8948081421637436,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.4915,
+      "step": 45
+    },
+    {
+      "epoch": 0.1392357169882709,
+      "grad_norm": 0.8858319184362027,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.4822,
+      "step": 46
+    },
+    {
+      "epoch": 0.1422625804010594,
+      "grad_norm": 0.9451549700147779,
+      "learning_rate": 1.4242424242424245e-05,
+      "loss": 0.4833,
+      "step": 47
+    },
+    {
+      "epoch": 0.1452894438138479,
+      "grad_norm": 0.8269190081158121,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.4824,
+      "step": 48
+    },
+    {
+      "epoch": 0.14831630722663638,
+      "grad_norm": 0.9098347785779631,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 0.4935,
+      "step": 49
+    },
+    {
+      "epoch": 0.1513431706394249,
+      "grad_norm": 0.8426410172802636,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.4795,
+      "step": 50
+    },
+    {
+      "epoch": 0.1543700340522134,
+      "grad_norm": 0.8707110392162887,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.5074,
+      "step": 51
+    },
+    {
+      "epoch": 0.15739689746500188,
+      "grad_norm": 0.9822734739737572,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 0.4668,
+      "step": 52
+    },
+    {
+      "epoch": 0.1604237608777904,
+      "grad_norm": 0.8026531632038805,
+      "learning_rate": 1.606060606060606e-05,
+      "loss": 0.4748,
+      "step": 53
+    },
+    {
+      "epoch": 0.16345062429057888,
+      "grad_norm": 0.9446747307067135,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.4902,
+      "step": 54
+    },
+    {
+      "epoch": 0.16647748770336737,
+      "grad_norm": 0.8743244602267946,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4626,
+      "step": 55
+    },
+    {
+      "epoch": 0.1695043511161559,
+      "grad_norm": 0.9600857878239794,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.4988,
+      "step": 56
+    },
+    {
+      "epoch": 0.17253121452894438,
+      "grad_norm": 0.9772063517587551,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.5004,
+      "step": 57
+    },
+    {
+      "epoch": 0.1755580779417329,
+      "grad_norm": 0.8848387258173624,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.519,
+      "step": 58
+    },
+    {
+      "epoch": 0.17858494135452138,
+      "grad_norm": 1.3487701254810844,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 0.5007,
+      "step": 59
+    },
+    {
+      "epoch": 0.18161180476730987,
+      "grad_norm": 1.2211725292532742,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.4982,
+      "step": 60
+    },
+    {
+      "epoch": 0.1846386681800984,
+      "grad_norm": 1.1262226590900428,
+      "learning_rate": 1.8484848484848487e-05,
+      "loss": 0.4924,
+      "step": 61
+    },
+    {
+      "epoch": 0.18766553159288688,
+      "grad_norm": 1.2094768905224471,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 0.4882,
+      "step": 62
+    },
+    {
+      "epoch": 0.19069239500567536,
+      "grad_norm": 0.893648317551491,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.4809,
+      "step": 63
+    },
+    {
+      "epoch": 0.19371925841846388,
+      "grad_norm": 1.1692921101941405,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.4806,
+      "step": 64
+    },
+    {
+      "epoch": 0.19674612183125237,
+      "grad_norm": 1.0198119083792836,
+      "learning_rate": 1.96969696969697e-05,
+      "loss": 0.463,
+      "step": 65
+    },
+    {
+      "epoch": 0.19977298524404086,
+      "grad_norm": 0.855634974190042,
+      "learning_rate": 2e-05,
+      "loss": 0.4591,
+      "step": 66
+    },
+    {
+      "epoch": 0.20279984865682937,
+      "grad_norm": 1.2946891078133405,
+      "learning_rate": 1.9999860139251737e-05,
+      "loss": 0.4768,
+      "step": 67
+    },
+    {
+      "epoch": 0.20582671206961786,
+      "grad_norm": 0.9108206295621432,
+      "learning_rate": 1.9999440560919153e-05,
+      "loss": 0.4922,
+      "step": 68
+    },
+    {
+      "epoch": 0.20885357548240635,
+      "grad_norm": 1.0340084911150065,
+      "learning_rate": 1.9998741276738753e-05,
+      "loss": 0.4946,
+      "step": 69
+    },
+    {
+      "epoch": 0.21188043889519487,
+      "grad_norm": 1.0048595755204783,
+      "learning_rate": 1.999776230627102e-05,
+      "loss": 0.4737,
+      "step": 70
+    },
+    {
+      "epoch": 0.21490730230798336,
+      "grad_norm": 0.8940099277439812,
+      "learning_rate": 1.9996503676899863e-05,
+      "loss": 0.4945,
+      "step": 71
+    },
+    {
+      "epoch": 0.21793416572077184,
+      "grad_norm": 1.0994430419795238,
+      "learning_rate": 1.9994965423831853e-05,
+      "loss": 0.4742,
+      "step": 72
+    },
+    {
+      "epoch": 0.22096102913356036,
+      "grad_norm": 0.8726676978656456,
+      "learning_rate": 1.9993147590095232e-05,
+      "loss": 0.4803,
+      "step": 73
+    },
+    {
+      "epoch": 0.22398789254634885,
+      "grad_norm": 0.8354080150234044,
+      "learning_rate": 1.999105022653872e-05,
+      "loss": 0.46,
+      "step": 74
+    },
+    {
+      "epoch": 0.22701475595913734,
+      "grad_norm": 0.9404851227112417,
+      "learning_rate": 1.9988673391830082e-05,
+      "loss": 0.48,
+      "step": 75
+    },
+    {
+      "epoch": 0.23004161937192585,
+      "grad_norm": 0.857980992745153,
+      "learning_rate": 1.9986017152454497e-05,
+      "loss": 0.4762,
+      "step": 76
+    },
+    {
+      "epoch": 0.23306848278471434,
+      "grad_norm": 0.920503468589322,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.473,
+      "step": 77
+    },
+    {
+      "epoch": 0.23609534619750283,
+      "grad_norm": 0.9632278096731705,
+      "learning_rate": 1.9979866764718846e-05,
+      "loss": 0.4587,
+      "step": 78
+    },
+    {
+      "epoch": 0.23912220961029135,
+      "grad_norm": 0.8717356338753535,
+      "learning_rate": 1.997637278839835e-05,
+      "loss": 0.4773,
+      "step": 79
+    },
+    {
+      "epoch": 0.24214907302307984,
+      "grad_norm": 0.9286733642442924,
+      "learning_rate": 1.9972599751485225e-05,
+      "loss": 0.4726,
+      "step": 80
+    },
+    {
+      "epoch": 0.24517593643586832,
+      "grad_norm": 0.9052954076534013,
+      "learning_rate": 1.9968547759519426e-05,
+      "loss": 0.4807,
+      "step": 81
+    },
+    {
+      "epoch": 0.24820279984865684,
+      "grad_norm": 0.8988276165138743,
+      "learning_rate": 1.9964216925843876e-05,
+      "loss": 0.4936,
+      "step": 82
+    },
+    {
+      "epoch": 0.2512296632614453,
+      "grad_norm": 0.8570360743115466,
+      "learning_rate": 1.9959607371601303e-05,
+      "loss": 0.47,
+      "step": 83
+    },
+    {
+      "epoch": 0.2542565266742338,
+      "grad_norm": 0.8500638128902182,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 0.4775,
+      "step": 84
+    },
+    {
+      "epoch": 0.25728339008702233,
+      "grad_norm": 0.9158074401855775,
+      "learning_rate": 1.994955262496446e-05,
+      "loss": 0.488,
+      "step": 85
+    },
+    {
+      "epoch": 0.2603102534998108,
+      "grad_norm": 0.8432785556665732,
+      "learning_rate": 1.9944107713823068e-05,
+      "loss": 0.4879,
+      "step": 86
+    },
+    {
+      "epoch": 0.2633371169125993,
+      "grad_norm": 0.9355269732255403,
+      "learning_rate": 1.9938384644612542e-05,
+      "loss": 0.4649,
+      "step": 87
+    },
+    {
+      "epoch": 0.26636398032538783,
+      "grad_norm": 0.8209147775987876,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.4678,
+      "step": 88
+    },
+    {
+      "epoch": 0.2693908437381763,
+      "grad_norm": 0.8390349330105009,
+      "learning_rate": 1.9926104680106484e-05,
+      "loss": 0.4744,
+      "step": 89
+    },
+    {
+      "epoch": 0.2724177071509648,
+      "grad_norm": 0.9274522138379951,
+      "learning_rate": 1.9919548128307954e-05,
+      "loss": 0.4884,
+      "step": 90
+    },
+    {
+      "epoch": 0.2754445705637533,
+      "grad_norm": 0.8143501827269861,
+      "learning_rate": 1.9912714105424694e-05,
+      "loss": 0.4583,
+      "step": 91
+    },
+    {
+      "epoch": 0.2784714339765418,
+      "grad_norm": 0.8139364596789473,
+      "learning_rate": 1.990560280261901e-05,
+      "loss": 0.4814,
+      "step": 92
+    },
+    {
+      "epoch": 0.2814982973893303,
+      "grad_norm": 0.8750935629111116,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.4585,
+      "step": 93
+    },
+    {
+      "epoch": 0.2845251608021188,
+      "grad_norm": 0.81088027895057,
+      "learning_rate": 1.9890549160664633e-05,
+      "loss": 0.4663,
+      "step": 94
+    },
+    {
+      "epoch": 0.2875520242149073,
+      "grad_norm": 0.8019033008592847,
+      "learning_rate": 1.9882607242598663e-05,
+      "loss": 0.4403,
+      "step": 95
+    },
+    {
+      "epoch": 0.2905788876276958,
+      "grad_norm": 0.8683807091372306,
+      "learning_rate": 1.9874388886763944e-05,
+      "loss": 0.4587,
+      "step": 96
+    },
+    {
+      "epoch": 0.2936057510404843,
+      "grad_norm": 0.8403228245920575,
+      "learning_rate": 1.9865894323045558e-05,
+      "loss": 0.4716,
+      "step": 97
+    },
+    {
+      "epoch": 0.29663261445327277,
+      "grad_norm": 0.8988013123998222,
+      "learning_rate": 1.9857123789054707e-05,
+      "loss": 0.506,
+      "step": 98
+    },
+    {
+      "epoch": 0.2996594778660613,
+      "grad_norm": 0.8495011461418398,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.4747,
+      "step": 99
+    },
+    {
+      "epoch": 0.3026863412788498,
+      "grad_norm": 0.8370130288185559,
+      "learning_rate": 1.9838755799290993e-05,
+      "loss": 0.4698,
+      "step": 100
+    },
+    {
+      "epoch": 0.30571320469163826,
+      "grad_norm": 0.88343417084434,
+      "learning_rate": 1.9829158857310288e-05,
+      "loss": 0.4737,
+      "step": 101
+    },
+    {
+      "epoch": 0.3087400681044268,
+      "grad_norm": 0.8050898110543138,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.4834,
+      "step": 102
+    },
+    {
+      "epoch": 0.3117669315172153,
+      "grad_norm": 0.9372086576374935,
+      "learning_rate": 1.9809140421379168e-05,
+      "loss": 0.516,
+      "step": 103
+    },
+    {
+      "epoch": 0.31479379493000376,
+      "grad_norm": 0.8963337264370949,
+      "learning_rate": 1.979871948738743e-05,
+      "loss": 0.4712,
+      "step": 104
+    },
+    {
+      "epoch": 0.3178206583427923,
+      "grad_norm": 0.9817530495719915,
+      "learning_rate": 1.978802446214779e-05,
+      "loss": 0.4703,
+      "step": 105
+    },
+    {
+      "epoch": 0.3208475217555808,
+      "grad_norm": 0.8772809063936746,
+      "learning_rate": 1.9777055644823087e-05,
+      "loss": 0.4651,
+      "step": 106
+    },
+    {
+      "epoch": 0.32387438516836925,
+      "grad_norm": 0.8434330822498869,
+      "learning_rate": 1.9765813342234726e-05,
+      "loss": 0.4872,
+      "step": 107
+    },
+    {
+      "epoch": 0.32690124858115777,
+      "grad_norm": 1.0277262156736706,
+      "learning_rate": 1.9754297868854075e-05,
+      "loss": 0.485,
+      "step": 108
+    },
+    {
+      "epoch": 0.3299281119939463,
+      "grad_norm": 0.8417103807033413,
+      "learning_rate": 1.9742509546793673e-05,
+      "loss": 0.4427,
+      "step": 109
+    },
+    {
+      "epoch": 0.33295497540673474,
+      "grad_norm": 0.8742684564611128,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.4682,
+      "step": 110
+    },
+    {
+      "epoch": 0.33598183881952326,
+      "grad_norm": 0.935213150088787,
+      "learning_rate": 1.9718115683235418e-05,
+      "loss": 0.4568,
+      "step": 111
+    },
+    {
+      "epoch": 0.3390087022323118,
+      "grad_norm": 0.790857302447283,
+      "learning_rate": 1.970551082408636e-05,
+      "loss": 0.4558,
+      "step": 112
+    },
+    {
+      "epoch": 0.3420355656451003,
+      "grad_norm": 0.8149909991003174,
+      "learning_rate": 1.969263448093608e-05,
+      "loss": 0.4651,
+      "step": 113
+    },
+    {
+      "epoch": 0.34506242905788875,
+      "grad_norm": 0.7870840357209562,
+      "learning_rate": 1.9679487013963566e-05,
+      "loss": 0.4695,
+      "step": 114
+    },
+    {
+      "epoch": 0.34808929247067727,
+      "grad_norm": 0.801469010654098,
+      "learning_rate": 1.9666068790931733e-05,
+      "loss": 0.4815,
+      "step": 115
+    },
+    {
+      "epoch": 0.3511161558834658,
+      "grad_norm": 0.7737656156152036,
+      "learning_rate": 1.9652380187177128e-05,
+      "loss": 0.4672,
+      "step": 116
+    },
+    {
+      "epoch": 0.35414301929625425,
+      "grad_norm": 0.8164098118854404,
+      "learning_rate": 1.9638421585599422e-05,
+      "loss": 0.4973,
+      "step": 117
+    },
+    {
+      "epoch": 0.35716988270904276,
+      "grad_norm": 0.800648561968564,
+      "learning_rate": 1.9624193376650708e-05,
+      "loss": 0.469,
+      "step": 118
+    },
+    {
+      "epoch": 0.3601967461218313,
+      "grad_norm": 0.768721797881389,
+      "learning_rate": 1.960969595832457e-05,
+      "loss": 0.459,
+      "step": 119
+    },
+    {
+      "epoch": 0.36322360953461974,
+      "grad_norm": 0.7907394188148069,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.4828,
+      "step": 120
+    },
+    {
+      "epoch": 0.36625047294740826,
+      "grad_norm": 0.7782969504525207,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.4442,
+      "step": 121
+    },
+    {
+      "epoch": 0.3692773363601968,
+      "grad_norm": 0.7630840444418707,
+      "learning_rate": 1.956459253990476e-05,
+      "loss": 0.466,
+      "step": 122
+    },
+    {
+      "epoch": 0.37230419977298523,
+      "grad_norm": 0.8107488594564556,
+      "learning_rate": 1.9549022414440738e-05,
+      "loss": 0.4645,
+      "step": 123
+    },
+    {
+      "epoch": 0.37533106318577375,
+      "grad_norm": 0.8053190774822062,
+      "learning_rate": 1.9533185182292705e-05,
+      "loss": 0.4761,
+      "step": 124
+    },
+    {
+      "epoch": 0.37835792659856227,
+      "grad_norm": 0.8065169189667375,
+      "learning_rate": 1.9517081286462082e-05,
+      "loss": 0.4722,
+      "step": 125
+    },
+    {
+      "epoch": 0.3813847900113507,
+      "grad_norm": 0.8303870745866065,
+      "learning_rate": 1.9500711177409456e-05,
+      "loss": 0.4975,
+      "step": 126
+    },
+    {
+      "epoch": 0.38441165342413924,
+      "grad_norm": 0.8117027952054194,
+      "learning_rate": 1.9484075313041968e-05,
+      "loss": 0.4726,
+      "step": 127
+    },
+    {
+      "epoch": 0.38743851683692776,
+      "grad_norm": 0.7727800328990134,
+      "learning_rate": 1.9467174158700507e-05,
+      "loss": 0.454,
+      "step": 128
+    },
+    {
+      "epoch": 0.3904653802497162,
+      "grad_norm": 0.7664083678601097,
+      "learning_rate": 1.9450008187146685e-05,
+      "loss": 0.4525,
+      "step": 129
+    },
+    {
+      "epoch": 0.39349224366250474,
+      "grad_norm": 0.7743145990618222,
+      "learning_rate": 1.9432577878549635e-05,
+      "loss": 0.4662,
+      "step": 130
+    },
+    {
+      "epoch": 0.39651910707529325,
+      "grad_norm": 0.7785850112923786,
+      "learning_rate": 1.9414883720472557e-05,
+      "loss": 0.4627,
+      "step": 131
+    },
+    {
+      "epoch": 0.3995459704880817,
+      "grad_norm": 0.839723541762426,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.4809,
+      "step": 132
+    },
+    {
+      "epoch": 0.40257283390087023,
+      "grad_norm": 0.7710622320959767,
+      "learning_rate": 1.937870584301945e-05,
+      "loss": 0.4612,
+      "step": 133
+    },
+    {
+      "epoch": 0.40559969731365875,
+      "grad_norm": 0.7953161734918524,
+      "learning_rate": 1.9360223135616423e-05,
+      "loss": 0.4776,
+      "step": 134
+    },
+    {
+      "epoch": 0.4086265607264472,
+      "grad_norm": 0.7905189423449387,
+      "learning_rate": 1.9341478602651068e-05,
+      "loss": 0.4791,
+      "step": 135
+    },
+    {
+      "epoch": 0.4116534241392357,
+      "grad_norm": 0.7936861949300658,
+      "learning_rate": 1.932247276844826e-05,
+      "loss": 0.4299,
+      "step": 136
+    },
+    {
+      "epoch": 0.41468028755202424,
+      "grad_norm": 0.7876379140464452,
+      "learning_rate": 1.9303206164642037e-05,
+      "loss": 0.4808,
+      "step": 137
+    },
+    {
+      "epoch": 0.4177071509648127,
+      "grad_norm": 0.8905286081361297,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 0.4799,
+      "step": 138
+    },
+    {
+      "epoch": 0.4207340143776012,
+      "grad_norm": 0.7999359194199922,
+      "learning_rate": 1.9263892811211865e-05,
+      "loss": 0.4665,
+      "step": 139
+    },
+    {
+      "epoch": 0.42376087779038973,
+      "grad_norm": 0.8408863767054626,
+      "learning_rate": 1.9243847161266924e-05,
+      "loss": 0.4625,
+      "step": 140
+    },
+    {
+      "epoch": 0.4267877412031782,
+      "grad_norm": 0.885608428330326,
+      "learning_rate": 1.9223542941045817e-05,
+      "loss": 0.488,
+      "step": 141
+    },
+    {
+      "epoch": 0.4298146046159667,
+      "grad_norm": 0.7299543835801919,
+      "learning_rate": 1.920298071850123e-05,
+      "loss": 0.4639,
+      "step": 142
+    },
+    {
+      "epoch": 0.4328414680287552,
+      "grad_norm": 0.9052298980188906,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.4542,
+      "step": 143
+    },
+    {
+      "epoch": 0.4358683314415437,
+      "grad_norm": 0.8846186987133594,
+      "learning_rate": 1.9161084574320696e-05,
+      "loss": 0.4589,
+      "step": 144
+    },
+    {
+      "epoch": 0.4388951948543322,
+      "grad_norm": 1.017471249576828,
+      "learning_rate": 1.913975182460996e-05,
+      "loss": 0.4604,
+      "step": 145
+    },
+    {
+      "epoch": 0.4419220582671207,
+      "grad_norm": 0.9273124367739143,
+      "learning_rate": 1.9118163416393392e-05,
+      "loss": 0.453,
+      "step": 146
+    },
+    {
+      "epoch": 0.4449489216799092,
+      "grad_norm": 0.9686730387778019,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.4751,
+      "step": 147
+    },
+    {
+      "epoch": 0.4479757850926977,
+      "grad_norm": 1.0350833064002376,
+      "learning_rate": 1.9074222047073945e-05,
+      "loss": 0.4884,
+      "step": 148
+    },
+    {
+      "epoch": 0.4510026485054862,
+      "grad_norm": 0.9164842852912688,
+      "learning_rate": 1.9051870315105626e-05,
+      "loss": 0.4786,
+      "step": 149
+    },
+    {
+      "epoch": 0.4540295119182747,
+      "grad_norm": 1.0008813447192486,
+      "learning_rate": 1.9029265382866216e-05,
+      "loss": 0.4659,
+      "step": 150
+    },
+    {
+      "epoch": 0.4570563753310632,
+      "grad_norm": 0.8349941094748187,
+      "learning_rate": 1.9006407882664256e-05,
+      "loss": 0.4488,
+      "step": 151
+    },
+    {
+      "epoch": 0.4600832387438517,
+      "grad_norm": 1.1058846123960806,
+      "learning_rate": 1.8983298453873172e-05,
+      "loss": 0.4642,
+      "step": 152
+    },
+    {
+      "epoch": 0.46311010215664017,
+      "grad_norm": 0.7930394449330936,
+      "learning_rate": 1.895993774291336e-05,
+      "loss": 0.4652,
+      "step": 153
+    },
+    {
+      "epoch": 0.4661369655694287,
+      "grad_norm": 0.901012412606741,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.4351,
+      "step": 154
+    },
+    {
+      "epoch": 0.4691638289822172,
+      "grad_norm": 0.870222775443807,
+      "learning_rate": 1.891246509529539e-05,
+      "loss": 0.4588,
+      "step": 155
+    },
+    {
+      "epoch": 0.47219069239500566,
+      "grad_norm": 0.9011781930146978,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.4754,
+      "step": 156
+    },
+    {
+      "epoch": 0.4752175558077942,
+      "grad_norm": 0.8028453343220097,
+      "learning_rate": 1.886399525142122e-05,
+      "loss": 0.4507,
+      "step": 157
+    },
+    {
+      "epoch": 0.4782444192205827,
+      "grad_norm": 0.7378154495157554,
+      "learning_rate": 1.8839388071291506e-05,
+      "loss": 0.4643,
+      "step": 158
+    },
+    {
+      "epoch": 0.48127128263337116,
+      "grad_norm": 0.7883445374716177,
+      "learning_rate": 1.881453363447582e-05,
+      "loss": 0.4665,
+      "step": 159
+    },
+    {
+      "epoch": 0.48429814604615967,
+      "grad_norm": 0.7652184135249859,
+      "learning_rate": 1.8789432636206197e-05,
+      "loss": 0.4336,
+      "step": 160
+    },
+    {
+      "epoch": 0.4873250094589482,
+      "grad_norm": 0.7961220993584102,
+      "learning_rate": 1.8764085778611507e-05,
+      "loss": 0.4781,
+      "step": 161
+    },
+    {
+      "epoch": 0.49035187287173665,
+      "grad_norm": 0.7966953848750636,
+      "learning_rate": 1.873849377069785e-05,
+      "loss": 0.4232,
+      "step": 162
+    },
+    {
+      "epoch": 0.49337873628452517,
+      "grad_norm": 0.8125786350883972,
+      "learning_rate": 1.87126573283287e-05,
+      "loss": 0.4664,
+      "step": 163
+    },
+    {
+      "epoch": 0.4964055996973137,
+      "grad_norm": 0.7809266231536923,
+      "learning_rate": 1.8686577174204887e-05,
+      "loss": 0.4654,
+      "step": 164
+    },
+    {
+      "epoch": 0.49943246311010214,
+      "grad_norm": 0.7314653002307387,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.4473,
+      "step": 165
+    },
+    {
+      "epoch": 0.5024593265228906,
+      "grad_norm": 0.7520181725999131,
+      "learning_rate": 1.863368865556191e-05,
+      "loss": 0.4473,
+      "step": 166
+    },
+    {
+      "epoch": 0.5054861899356792,
+      "grad_norm": 0.8095889704222708,
+      "learning_rate": 1.8606881770448305e-05,
+      "loss": 0.4577,
+      "step": 167
+    },
+    {
+      "epoch": 0.5085130533484676,
+      "grad_norm": 0.8722798950517553,
+      "learning_rate": 1.8579834132349773e-05,
+      "loss": 0.4484,
+      "step": 168
+    },
+    {
+      "epoch": 0.5115399167612561,
+      "grad_norm": 0.7124617799867115,
+      "learning_rate": 1.8552546497846893e-05,
+      "loss": 0.4569,
+      "step": 169
+    },
+    {
+      "epoch": 0.5145667801740447,
+      "grad_norm": 0.832658001424768,
+      "learning_rate": 1.8525019630233463e-05,
+      "loss": 0.4745,
+      "step": 170
+    },
+    {
+      "epoch": 0.5175936435868331,
+      "grad_norm": 0.8509261922026864,
+      "learning_rate": 1.8497254299495147e-05,
+      "loss": 0.4349,
+      "step": 171
+    },
+    {
+      "epoch": 0.5206205069996216,
+      "grad_norm": 0.7661706427905607,
+      "learning_rate": 1.8469251282287925e-05,
+      "loss": 0.4673,
+      "step": 172
+    },
+    {
+      "epoch": 0.5236473704124102,
+      "grad_norm": 0.799519497593696,
+      "learning_rate": 1.8441011361916387e-05,
+      "loss": 0.4465,
+      "step": 173
+    },
+    {
+      "epoch": 0.5266742338251986,
+      "grad_norm": 0.8162880332000012,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.4593,
+      "step": 174
+    },
+    {
+      "epoch": 0.5297010972379871,
+      "grad_norm": 0.728312757903625,
+      "learning_rate": 1.8383823978010077e-05,
+      "loss": 0.4642,
+      "step": 175
+    },
+    {
+      "epoch": 0.5327279606507757,
+      "grad_norm": 0.8245075000448732,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.4342,
+      "step": 176
+    },
+    {
+      "epoch": 0.5357548240635641,
+      "grad_norm": 0.7671492488151214,
+      "learning_rate": 1.8325698546347714e-05,
+      "loss": 0.4767,
+      "step": 177
+    },
+    {
+      "epoch": 0.5387816874763526,
+      "grad_norm": 0.7044308662036874,
+      "learning_rate": 1.8296286090880362e-05,
+      "loss": 0.4402,
+      "step": 178
+    },
+    {
+      "epoch": 0.5418085508891411,
+      "grad_norm": 0.805564438963134,
+      "learning_rate": 1.8266641570456915e-05,
+      "loss": 0.4476,
+      "step": 179
+    },
+    {
+      "epoch": 0.5448354143019296,
+      "grad_norm": 0.7332920416802393,
+      "learning_rate": 1.8236765814298328e-05,
+      "loss": 0.452,
+      "step": 180
+    },
+    {
+      "epoch": 0.5478622777147181,
+      "grad_norm": 0.8289941190848701,
+      "learning_rate": 1.820665965809373e-05,
+      "loss": 0.4504,
+      "step": 181
+    },
+    {
+      "epoch": 0.5508891411275066,
+      "grad_norm": 0.7796999019065849,
+      "learning_rate": 1.8176323943977034e-05,
+      "loss": 0.4453,
+      "step": 182
+    },
+    {
+      "epoch": 0.5539160045402951,
+      "grad_norm": 0.849125107166502,
+      "learning_rate": 1.814575952050336e-05,
+      "loss": 0.4486,
+      "step": 183
+    },
+    {
+      "epoch": 0.5569428679530836,
+      "grad_norm": 0.7465741122297203,
+      "learning_rate": 1.8114967242625342e-05,
+      "loss": 0.4477,
+      "step": 184
+    },
+    {
+      "epoch": 0.5599697313658721,
+      "grad_norm": 0.7882319160754231,
+      "learning_rate": 1.808394797166919e-05,
+      "loss": 0.4617,
+      "step": 185
+    },
+    {
+      "epoch": 0.5629965947786606,
+      "grad_norm": 0.7474838003818545,
+      "learning_rate": 1.8052702575310588e-05,
+      "loss": 0.4288,
+      "step": 186
+    },
+    {
+      "epoch": 0.5660234581914491,
+      "grad_norm": 0.8155013775731236,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.466,
+      "step": 187
+    },
+    {
+      "epoch": 0.5690503216042376,
+      "grad_norm": 0.7460068580250709,
+      "learning_rate": 1.7989536908690413e-05,
+      "loss": 0.4348,
+      "step": 188
+    },
+    {
+      "epoch": 0.5720771850170261,
+      "grad_norm": 0.7656442159970734,
+      "learning_rate": 1.7957618405308323e-05,
+      "loss": 0.4486,
+      "step": 189
+    },
+    {
+      "epoch": 0.5751040484298146,
+      "grad_norm": 0.6770144632514713,
+      "learning_rate": 1.792547731023332e-05,
+      "loss": 0.4275,
+      "step": 190
+    },
+    {
+      "epoch": 0.5781309118426031,
+      "grad_norm": 0.7465392275602387,
+      "learning_rate": 1.789311452252092e-05,
+      "loss": 0.4429,
+      "step": 191
+    },
+    {
+      "epoch": 0.5811577752553916,
+      "grad_norm": 0.7127281486505666,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.4514,
+      "step": 192
+    },
+    {
+      "epoch": 0.58418463866818,
+      "grad_norm": 0.7591266199009,
+      "learning_rate": 1.782772749638682e-05,
+      "loss": 0.4495,
+      "step": 193
+    },
+    {
+      "epoch": 0.5872115020809686,
+      "grad_norm": 0.7570441611594431,
+      "learning_rate": 1.779470508698079e-05,
+      "loss": 0.4628,
+      "step": 194
+    },
+    {
+      "epoch": 0.5902383654937571,
+      "grad_norm": 0.688156110718093,
+      "learning_rate": 1.776146464291757e-05,
+      "loss": 0.4444,
+      "step": 195
+    },
+    {
+      "epoch": 0.5932652289065455,
+      "grad_norm": 0.7959250209764478,
+      "learning_rate": 1.772800709400383e-05,
+      "loss": 0.4522,
+      "step": 196
+    },
+    {
+      "epoch": 0.5962920923193341,
+      "grad_norm": 0.7251620422974576,
+      "learning_rate": 1.7694333376119144e-05,
+      "loss": 0.4416,
+      "step": 197
+    },
+    {
+      "epoch": 0.5993189557321226,
+      "grad_norm": 0.7310841603682728,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.4498,
+      "step": 198
+    },
+    {
+      "epoch": 0.602345819144911,
+      "grad_norm": 0.7368747012865285,
+      "learning_rate": 1.762634120716238e-05,
+      "loss": 0.4569,
+      "step": 199
+    },
+    {
+      "epoch": 0.6053726825576996,
+      "grad_norm": 0.760764074141423,
+      "learning_rate": 1.7592024657977432e-05,
+      "loss": 0.4763,
+      "step": 200
+    },
+    {
+      "epoch": 0.6083995459704881,
+      "grad_norm": 0.7617915436660051,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.4605,
+      "step": 201
+    },
+    {
+      "epoch": 0.6114264093832765,
+      "grad_norm": 0.7507559372049951,
+      "learning_rate": 1.75227554297058e-05,
+      "loss": 0.4421,
+      "step": 202
+    },
+    {
+      "epoch": 0.6144532727960651,
+      "grad_norm": 0.8064212793444883,
+      "learning_rate": 1.7487804688228327e-05,
+      "loss": 0.4454,
+      "step": 203
+    },
+    {
+      "epoch": 0.6174801362088536,
+      "grad_norm": 0.8296397137069347,
+      "learning_rate": 1.745264449675755e-05,
+      "loss": 0.4531,
+      "step": 204
+    },
+    {
+      "epoch": 0.620506999621642,
+      "grad_norm": 0.8330341913883574,
+      "learning_rate": 1.7417275838799596e-05,
+      "loss": 0.4424,
+      "step": 205
+    },
+    {
+      "epoch": 0.6235338630344306,
+      "grad_norm": 0.8438848743704817,
+      "learning_rate": 1.7381699703691866e-05,
+      "loss": 0.4507,
+      "step": 206
+    },
+    {
+      "epoch": 0.626560726447219,
+      "grad_norm": 0.6953421056635204,
+      "learning_rate": 1.734591708657533e-05,
+      "loss": 0.4377,
+      "step": 207
+    },
+    {
+      "epoch": 0.6295875898600075,
+      "grad_norm": 0.905281887746711,
+      "learning_rate": 1.730992898836672e-05,
+      "loss": 0.4314,
+      "step": 208
+    },
+    {
+      "epoch": 0.6326144532727961,
+      "grad_norm": 0.7547147802874563,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.4275,
+      "step": 209
+    },
+    {
+      "epoch": 0.6356413166855845,
+      "grad_norm": 0.7957354859392501,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 0.4617,
+      "step": 210
+    },
+    {
+      "epoch": 0.638668180098373,
+      "grad_norm": 0.7259637538533981,
+      "learning_rate": 1.720074190240269e-05,
+      "loss": 0.4561,
+      "step": 211
+    },
+    {
+      "epoch": 0.6416950435111616,
+      "grad_norm": 0.834407678885038,
+      "learning_rate": 1.7163942003524574e-05,
+      "loss": 0.4628,
+      "step": 212
+    },
+    {
+      "epoch": 0.64472190692395,
+      "grad_norm": 0.7690032232197687,
+      "learning_rate": 1.7126941713788633e-05,
+      "loss": 0.4788,
+      "step": 213
+    },
+    {
+      "epoch": 0.6477487703367385,
+      "grad_norm": 0.8656077860154962,
+      "learning_rate": 1.70897420681725e-05,
+      "loss": 0.4537,
+      "step": 214
+    },
+    {
+      "epoch": 0.6507756337495271,
+      "grad_norm": 0.8004404206976288,
+      "learning_rate": 1.7052344107230244e-05,
+      "loss": 0.4342,
+      "step": 215
+    },
+    {
+      "epoch": 0.6538024971623155,
+      "grad_norm": 0.9163003135187244,
+      "learning_rate": 1.7014748877063212e-05,
+      "loss": 0.4494,
+      "step": 216
+    },
+    {
+      "epoch": 0.656829360575104,
+      "grad_norm": 0.7674258807265918,
+      "learning_rate": 1.697695742929082e-05,
+      "loss": 0.4254,
+      "step": 217
+    },
+    {
+      "epoch": 0.6598562239878926,
+      "grad_norm": 0.9372266682877208,
+      "learning_rate": 1.693897082102109e-05,
+      "loss": 0.4399,
+      "step": 218
+    },
+    {
+      "epoch": 0.662883087400681,
+      "grad_norm": 0.6982228363586895,
+      "learning_rate": 1.6900790114821122e-05,
+      "loss": 0.4305,
+      "step": 219
+    },
+    {
+      "epoch": 0.6659099508134695,
+      "grad_norm": 0.9650079768565273,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.4457,
+      "step": 220
+    },
+    {
+      "epoch": 0.6689368142262581,
+      "grad_norm": 0.7194707566096896,
+      "learning_rate": 1.682385068601563e-05,
+      "loss": 0.4342,
+      "step": 221
+    },
+    {
+      "epoch": 0.6719636776390465,
+      "grad_norm": 0.7365471354012435,
+      "learning_rate": 1.6785094115571323e-05,
+      "loss": 0.4372,
+      "step": 222
+    },
+    {
+      "epoch": 0.6749905410518351,
+      "grad_norm": 0.7270254121997651,
+      "learning_rate": 1.674614775145901e-05,
+      "loss": 0.4441,
+      "step": 223
+    },
+    {
+      "epoch": 0.6780174044646236,
+      "grad_norm": 0.8106563709770243,
+      "learning_rate": 1.670701268309221e-05,
+      "loss": 0.4283,
+      "step": 224
+    },
+    {
+      "epoch": 0.681044267877412,
+      "grad_norm": 0.7912012412436614,
+      "learning_rate": 1.666769000516292e-05,
+      "loss": 0.4467,
+      "step": 225
+    },
+    {
+      "epoch": 0.6840711312902006,
+      "grad_norm": 0.7863808512420861,
+      "learning_rate": 1.6628180817610963e-05,
+      "loss": 0.4406,
+      "step": 226
+    },
+    {
+      "epoch": 0.687097994702989,
+      "grad_norm": 0.8389148297166515,
+      "learning_rate": 1.658848622559325e-05,
+      "loss": 0.4145,
+      "step": 227
+    },
+    {
+      "epoch": 0.6901248581157775,
+      "grad_norm": 0.6875543345483303,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.4428,
+      "step": 228
+    },
+    {
+      "epoch": 0.6931517215285661,
+      "grad_norm": 0.9484360141211251,
+      "learning_rate": 1.6508545274687936e-05,
+      "loss": 0.4658,
+      "step": 229
+    },
+    {
+      "epoch": 0.6961785849413545,
+      "grad_norm": 0.7501609956382691,
+      "learning_rate": 1.6468301151920576e-05,
+      "loss": 0.4431,
+      "step": 230
+    },
+    {
+      "epoch": 0.699205448354143,
+      "grad_norm": 1.0266396097675403,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.4449,
+      "step": 231
+    },
+    {
+      "epoch": 0.7022323117669316,
+      "grad_norm": 0.7753361795663771,
+      "learning_rate": 1.6387271240298082e-05,
+      "loss": 0.4502,
+      "step": 232
+    },
+    {
+      "epoch": 0.70525917517972,
+      "grad_norm": 0.8882028777114376,
+      "learning_rate": 1.6346487718023762e-05,
+      "loss": 0.4555,
+      "step": 233
+    },
+    {
+      "epoch": 0.7082860385925085,
+      "grad_norm": 0.8039721247378876,
+      "learning_rate": 1.6305526670845225e-05,
+      "loss": 0.4562,
+      "step": 234
+    },
+    {
+      "epoch": 0.7113129020052971,
+      "grad_norm": 0.756859557207153,
+      "learning_rate": 1.6264389244531015e-05,
+      "loss": 0.4452,
+      "step": 235
+    },
+    {
+      "epoch": 0.7143397654180855,
+      "grad_norm": 0.8436817311275543,
+      "learning_rate": 1.6223076589783368e-05,
+      "loss": 0.4341,
+      "step": 236
+    },
+    {
+      "epoch": 0.717366628830874,
+      "grad_norm": 0.8099884708083491,
+      "learning_rate": 1.6181589862206053e-05,
+      "loss": 0.4587,
+      "step": 237
+    },
+    {
+      "epoch": 0.7203934922436626,
+      "grad_norm": 0.82705715653206,
+      "learning_rate": 1.613993022227202e-05,
+      "loss": 0.4785,
+      "step": 238
+    },
+    {
+      "epoch": 0.723420355656451,
+      "grad_norm": 0.7406400851101517,
+      "learning_rate": 1.6098098835290955e-05,
+      "loss": 0.4105,
+      "step": 239
+    },
+    {
+      "epoch": 0.7264472190692395,
+      "grad_norm": 0.7845145840282317,
+      "learning_rate": 1.6056096871376667e-05,
+      "loss": 0.4253,
+      "step": 240
+    },
+    {
+      "epoch": 0.729474082482028,
+      "grad_norm": 0.6755769263820055,
+      "learning_rate": 1.6013925505414386e-05,
+      "loss": 0.4543,
+      "step": 241
+    },
+    {
+      "epoch": 0.7325009458948165,
+      "grad_norm": 0.759726933133282,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.4572,
+      "step": 242
+    },
+    {
+      "epoch": 0.735527809307605,
+      "grad_norm": 0.740880688090326,
+      "learning_rate": 1.5929079290546408e-05,
+      "loss": 0.433,
+      "step": 243
+    },
+    {
+      "epoch": 0.7385546727203935,
+      "grad_norm": 0.7106269957214116,
+      "learning_rate": 1.5886406814971728e-05,
+      "loss": 0.4435,
+      "step": 244
+    },
+    {
+      "epoch": 0.741581536133182,
+      "grad_norm": 0.7342036050639433,
+      "learning_rate": 1.584356968394471e-05,
+      "loss": 0.4344,
+      "step": 245
+    },
+    {
+      "epoch": 0.7446083995459705,
+      "grad_norm": 0.7356744561318753,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.4694,
+      "step": 246
+    },
+    {
+      "epoch": 0.747635262958759,
+      "grad_norm": 0.7845237686020454,
+      "learning_rate": 1.575740625309244e-05,
+      "loss": 0.488,
+      "step": 247
+    },
+    {
+      "epoch": 0.7506621263715475,
+      "grad_norm": 0.6631531205318687,
+      "learning_rate": 1.5714082363443576e-05,
+      "loss": 0.427,
+      "step": 248
+    },
+    {
+      "epoch": 0.753688989784336,
+      "grad_norm": 0.694352953714639,
+      "learning_rate": 1.5670598638627707e-05,
+      "loss": 0.4461,
+      "step": 249
+    },
+    {
+      "epoch": 0.7567158531971245,
+      "grad_norm": 0.7098118755989453,
+      "learning_rate": 1.5626956294978103e-05,
+      "loss": 0.4202,
+      "step": 250
+    },
+    {
+      "epoch": 0.759742716609913,
+      "grad_norm": 0.6777025659890347,
+      "learning_rate": 1.5583156553264923e-05,
+      "loss": 0.424,
+      "step": 251
+    },
+    {
+      "epoch": 0.7627695800227015,
+      "grad_norm": 0.8129576983983366,
+      "learning_rate": 1.5539200638661106e-05,
+      "loss": 0.4412,
+      "step": 252
+    },
+    {
+      "epoch": 0.76579644343549,
+      "grad_norm": 0.6925882905996416,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.4545,
+      "step": 253
+    },
+    {
+      "epoch": 0.7688233068482785,
+      "grad_norm": 0.6884701542742249,
+      "learning_rate": 1.5450825213281317e-05,
+      "loss": 0.4438,
+      "step": 254
+    },
+    {
+      "epoch": 0.771850170261067,
+      "grad_norm": 0.7599219919859785,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.4396,
+      "step": 255
+    },
+    {
+      "epoch": 0.7748770336738555,
+      "grad_norm": 0.699029405304382,
+      "learning_rate": 1.5361839906972095e-05,
+      "loss": 0.4301,
+      "step": 256
+    },
+    {
+      "epoch": 0.777903897086644,
+      "grad_norm": 0.7089603504326589,
+      "learning_rate": 1.531712165719992e-05,
+      "loss": 0.4353,
+      "step": 257
+    },
+    {
+      "epoch": 0.7809307604994324,
+      "grad_norm": 0.7061985795347577,
+      "learning_rate": 1.5272254676105026e-05,
+      "loss": 0.4261,
+      "step": 258
+    },
+    {
+      "epoch": 0.783957623912221,
+      "grad_norm": 0.7613587379124639,
+      "learning_rate": 1.5227240218713326e-05,
+      "loss": 0.4376,
+      "step": 259
+    },
+    {
+      "epoch": 0.7869844873250095,
+      "grad_norm": 0.7315093630311442,
+      "learning_rate": 1.5182079544175957e-05,
+      "loss": 0.4578,
+      "step": 260
+    },
+    {
+      "epoch": 0.7900113507377979,
+      "grad_norm": 0.8191178110122554,
+      "learning_rate": 1.5136773915734067e-05,
+      "loss": 0.4519,
+      "step": 261
+    },
+    {
+      "epoch": 0.7930382141505865,
+      "grad_norm": 0.6930974174992991,
+      "learning_rate": 1.5091324600683472e-05,
+      "loss": 0.4115,
+      "step": 262
+    },
+    {
+      "epoch": 0.796065077563375,
+      "grad_norm": 0.8629331859634798,
+      "learning_rate": 1.5045732870339213e-05,
+      "loss": 0.4054,
+      "step": 263
+    },
+    {
+      "epoch": 0.7990919409761634,
+      "grad_norm": 0.7374373654099501,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4375,
+      "step": 264
+    },
+    {
+      "epoch": 0.802118804388952,
+      "grad_norm": 0.7890493920457685,
+      "learning_rate": 1.4954127268912525e-05,
+      "loss": 0.4288,
+      "step": 265
+    },
+    {
+      "epoch": 0.8051456678017405,
+      "grad_norm": 0.8346059814614823,
+      "learning_rate": 1.4908115960235683e-05,
+      "loss": 0.4512,
+      "step": 266
+    },
+    {
+      "epoch": 0.8081725312145289,
+      "grad_norm": 0.7295245916317982,
+      "learning_rate": 1.4861967361004687e-05,
+      "loss": 0.4491,
+      "step": 267
+    },
+    {
+      "epoch": 0.8111993946273175,
+      "grad_norm": 0.7743684662446152,
+      "learning_rate": 1.4815682762095065e-05,
+      "loss": 0.4201,
+      "step": 268
+    },
+    {
+      "epoch": 0.814226258040106,
+      "grad_norm": 0.6817181259751697,
+      "learning_rate": 1.476926345818654e-05,
+      "loss": 0.4365,
+      "step": 269
+    },
+    {
+      "epoch": 0.8172531214528944,
+      "grad_norm": 0.6938342569799029,
+      "learning_rate": 1.472271074772683e-05,
+      "loss": 0.4509,
+      "step": 270
+    },
+    {
+      "epoch": 0.820279984865683,
+      "grad_norm": 0.7373554923574489,
+      "learning_rate": 1.4676025932895315e-05,
+      "loss": 0.4294,
+      "step": 271
+    },
+    {
+      "epoch": 0.8233068482784714,
+      "grad_norm": 0.8298273234090764,
+      "learning_rate": 1.4629210319566626e-05,
+      "loss": 0.42,
+      "step": 272
+    },
+    {
+      "epoch": 0.8263337116912599,
+      "grad_norm": 0.6947235130122069,
+      "learning_rate": 1.4582265217274105e-05,
+      "loss": 0.4433,
+      "step": 273
+    },
+    {
+      "epoch": 0.8293605751040485,
+      "grad_norm": 0.7922243565981998,
+      "learning_rate": 1.4535191939173179e-05,
+      "loss": 0.436,
+      "step": 274
+    },
+    {
+      "epoch": 0.8323874385168369,
+      "grad_norm": 0.8133101571507579,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.4246,
+      "step": 275
+    },
+    {
+      "epoch": 0.8354143019296254,
+      "grad_norm": 0.7428009505245705,
+      "learning_rate": 1.4440666126057743e-05,
+      "loss": 0.4428,
+      "step": 276
+    },
+    {
+      "epoch": 0.838441165342414,
+      "grad_norm": 0.8357663896029989,
+      "learning_rate": 1.4393216235133427e-05,
+      "loss": 0.4312,
+      "step": 277
+    },
+    {
+      "epoch": 0.8414680287552024,
+      "grad_norm": 0.7234646010807861,
+      "learning_rate": 1.4345643456507126e-05,
+      "loss": 0.421,
+      "step": 278
+    },
+    {
+      "epoch": 0.8444948921679909,
+      "grad_norm": 0.865197864439475,
+      "learning_rate": 1.4297949120891718e-05,
+      "loss": 0.4059,
+      "step": 279
+    },
+    {
+      "epoch": 0.8475217555807795,
+      "grad_norm": 0.7993096087256234,
+      "learning_rate": 1.4250134562400301e-05,
+      "loss": 0.4393,
+      "step": 280
+    },
+    {
+      "epoch": 0.8505486189935679,
+      "grad_norm": 0.7183725929268479,
+      "learning_rate": 1.4202201118508863e-05,
+      "loss": 0.4191,
+      "step": 281
+    },
+    {
+      "epoch": 0.8535754824063564,
+      "grad_norm": 0.7799591969636352,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.4428,
+      "step": 282
+    },
+    {
+      "epoch": 0.856602345819145,
+      "grad_norm": 0.8084982205737231,
+      "learning_rate": 1.4105982941019751e-05,
+      "loss": 0.4663,
+      "step": 283
+    },
+    {
+      "epoch": 0.8596292092319334,
+      "grad_norm": 0.7673484437306302,
+      "learning_rate": 1.405770089885134e-05,
+      "loss": 0.4392,
+      "step": 284
+    },
+    {
+      "epoch": 0.8626560726447219,
+      "grad_norm": 0.8469599715720034,
+      "learning_rate": 1.4009305354066138e-05,
+      "loss": 0.4519,
+      "step": 285
+    },
+    {
+      "epoch": 0.8656829360575105,
+      "grad_norm": 0.7675388048532539,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.4052,
+      "step": 286
+    },
+    {
+      "epoch": 0.8687097994702989,
+      "grad_norm": 0.7454449958664093,
+      "learning_rate": 1.39121791746921e-05,
+      "loss": 0.4508,
+      "step": 287
+    },
+    {
+      "epoch": 0.8717366628830874,
+      "grad_norm": 0.7270938085775619,
+      "learning_rate": 1.3863451256931286e-05,
+      "loss": 0.4192,
+      "step": 288
+    },
+    {
+      "epoch": 0.874763526295876,
+      "grad_norm": 0.6806851832638964,
+      "learning_rate": 1.381461527013374e-05,
+      "loss": 0.4306,
+      "step": 289
+    },
+    {
+      "epoch": 0.8777903897086644,
+      "grad_norm": 0.7135120496300053,
+      "learning_rate": 1.3765672580346986e-05,
+      "loss": 0.4341,
+      "step": 290
+    },
+    {
+      "epoch": 0.8808172531214529,
+      "grad_norm": 0.6594815841959984,
+      "learning_rate": 1.3716624556603275e-05,
+      "loss": 0.4389,
+      "step": 291
+    },
+    {
+      "epoch": 0.8838441165342414,
+      "grad_norm": 0.6954999080358071,
+      "learning_rate": 1.3667472570881264e-05,
+      "loss": 0.4428,
+      "step": 292
+    },
+    {
+      "epoch": 0.8868709799470299,
+      "grad_norm": 0.7084448062363153,
+      "learning_rate": 1.361821799806765e-05,
+      "loss": 0.4365,
+      "step": 293
+    },
+    {
+      "epoch": 0.8898978433598184,
+      "grad_norm": 0.700444102312848,
+      "learning_rate": 1.356886221591872e-05,
+      "loss": 0.4304,
+      "step": 294
+    },
+    {
+      "epoch": 0.8929247067726069,
+      "grad_norm": 0.7284176901894484,
+      "learning_rate": 1.3519406605021797e-05,
+      "loss": 0.4158,
+      "step": 295
+    },
+    {
+      "epoch": 0.8959515701853954,
+      "grad_norm": 0.7068665986057537,
+      "learning_rate": 1.3469852548756626e-05,
+      "loss": 0.4138,
+      "step": 296
+    },
+    {
+      "epoch": 0.8989784335981839,
+      "grad_norm": 0.7816899454943492,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.4588,
+      "step": 297
+    },
+    {
+      "epoch": 0.9020052970109724,
+      "grad_norm": 0.7279483877475228,
+      "learning_rate": 1.3370454647370418e-05,
+      "loss": 0.4297,
+      "step": 298
+    },
+    {
+      "epoch": 0.9050321604237609,
+      "grad_norm": 0.7073461095832493,
+      "learning_rate": 1.3320613582622354e-05,
+      "loss": 0.4305,
+      "step": 299
+    },
+    {
+      "epoch": 0.9080590238365494,
+      "grad_norm": 0.7259761696820024,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.4355,
+      "step": 300
+    },
+    {
+      "epoch": 0.9110858872493379,
+      "grad_norm": 0.7256657716076117,
+      "learning_rate": 1.3220654195785917e-05,
+      "loss": 0.4515,
+      "step": 301
+    },
+    {
+      "epoch": 0.9141127506621264,
+      "grad_norm": 0.6987249515334674,
+      "learning_rate": 1.3170538669776469e-05,
+      "loss": 0.4331,
+      "step": 302
+    },
+    {
+      "epoch": 0.9171396140749148,
+      "grad_norm": 0.7057928111575967,
+      "learning_rate": 1.3120334456984871e-05,
+      "loss": 0.4237,
+      "step": 303
+    },
+    {
+      "epoch": 0.9201664774877034,
+      "grad_norm": 0.7083077772724465,
+      "learning_rate": 1.3070042961730878e-05,
+      "loss": 0.4418,
+      "step": 304
+    },
+    {
+      "epoch": 0.9231933409004919,
+      "grad_norm": 0.669110184385481,
+      "learning_rate": 1.3019665590775717e-05,
+      "loss": 0.4032,
+      "step": 305
+    },
+    {
+      "epoch": 0.9262202043132803,
+      "grad_norm": 0.7526085301594759,
+      "learning_rate": 1.296920375328275e-05,
+      "loss": 0.4463,
+      "step": 306
+    },
+    {
+      "epoch": 0.9292470677260689,
+      "grad_norm": 0.705554904281121,
+      "learning_rate": 1.2918658860778046e-05,
+      "loss": 0.4271,
+      "step": 307
+    },
+    {
+      "epoch": 0.9322739311388574,
+      "grad_norm": 0.6974611600129492,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.4232,
+      "step": 308
+    },
+    {
+      "epoch": 0.9353007945516458,
+      "grad_norm": 0.7442501624491912,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.4458,
+      "step": 309
+    },
+    {
+      "epoch": 0.9383276579644344,
+      "grad_norm": 0.6872876974178388,
+      "learning_rate": 1.2766540003065272e-05,
+      "loss": 0.3994,
+      "step": 310
+    },
+    {
+      "epoch": 0.9413545213772229,
+      "grad_norm": 0.6788679963863709,
+      "learning_rate": 1.2715677051645259e-05,
+      "loss": 0.4354,
+      "step": 311
+    },
+    {
+      "epoch": 0.9443813847900113,
+      "grad_norm": 0.6816191348058505,
+      "learning_rate": 1.266473813690035e-05,
+      "loss": 0.4105,
+      "step": 312
+    },
+    {
+      "epoch": 0.9474082482027999,
+      "grad_norm": 0.7040493758531391,
+      "learning_rate": 1.2613724683701491e-05,
+      "loss": 0.4438,
+      "step": 313
+    },
+    {
+      "epoch": 0.9504351116155884,
+      "grad_norm": 0.7081516835005162,
+      "learning_rate": 1.2562638119004627e-05,
+      "loss": 0.4258,
+      "step": 314
+    },
+    {
+      "epoch": 0.9534619750283768,
+      "grad_norm": 0.696461069865041,
+      "learning_rate": 1.2511479871810792e-05,
+      "loss": 0.4291,
+      "step": 315
+    },
+    {
+      "epoch": 0.9564888384411654,
+      "grad_norm": 0.6672562409645458,
+      "learning_rate": 1.2460251373126136e-05,
+      "loss": 0.4156,
+      "step": 316
+    },
+    {
+      "epoch": 0.9595157018539539,
+      "grad_norm": 0.7000392885286649,
+      "learning_rate": 1.2408954055921884e-05,
+      "loss": 0.4274,
+      "step": 317
+    },
+    {
+      "epoch": 0.9625425652667423,
+      "grad_norm": 0.6918111306378716,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.4315,
+      "step": 318
+    },
+    {
+      "epoch": 0.9655694286795309,
+      "grad_norm": 0.681085041129214,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.4306,
+      "step": 319
+    },
+    {
+      "epoch": 0.9685962920923193,
+      "grad_norm": 0.6712604426953142,
+      "learning_rate": 1.2254663551538047e-05,
+      "loss": 0.4223,
+      "step": 320
+    },
+    {
+      "epoch": 0.9716231555051078,
+      "grad_norm": 0.6991416187590415,
+      "learning_rate": 1.2203105327865407e-05,
+      "loss": 0.4338,
+      "step": 321
+    },
+    {
+      "epoch": 0.9746500189178964,
+      "grad_norm": 0.7337406612065791,
+      "learning_rate": 1.215148547860084e-05,
+      "loss": 0.4292,
+      "step": 322
+    },
+    {
+      "epoch": 0.9776768823306848,
+      "grad_norm": 0.6974181205899826,
+      "learning_rate": 1.2099805447662485e-05,
+      "loss": 0.3933,
+      "step": 323
+    },
+    {
+      "epoch": 0.9807037457434733,
+      "grad_norm": 0.6761622017462692,
+      "learning_rate": 1.2048066680651908e-05,
+      "loss": 0.4343,
+      "step": 324
+    },
+    {
+      "epoch": 0.9837306091562619,
+      "grad_norm": 0.7293395639381434,
+      "learning_rate": 1.1996270624813642e-05,
+      "loss": 0.4144,
+      "step": 325
+    },
+    {
+      "epoch": 0.9867574725690503,
+      "grad_norm": 0.7108445862832423,
+      "learning_rate": 1.194441872899471e-05,
+      "loss": 0.4262,
+      "step": 326
+    },
+    {
+      "epoch": 0.9897843359818388,
+      "grad_norm": 0.6579869158952266,
+      "learning_rate": 1.1892512443604103e-05,
+      "loss": 0.393,
+      "step": 327
+    },
+    {
+      "epoch": 0.9928111993946274,
+      "grad_norm": 0.6939934077016605,
+      "learning_rate": 1.1840553220572204e-05,
+      "loss": 0.4331,
+      "step": 328
+    },
+    {
+      "epoch": 0.9958380628074158,
+      "grad_norm": 0.6933796518297354,
+      "learning_rate": 1.1788542513310178e-05,
+      "loss": 0.4496,
+      "step": 329
+    },
+    {
+      "epoch": 0.9988649262202043,
+      "grad_norm": 0.686821863583047,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.4361,
+      "step": 330
+    },
+    {
+      "epoch": 1.0018917896329929,
+      "grad_norm": 0.8195905721346842,
+      "learning_rate": 1.1684372466900306e-05,
+      "loss": 0.3728,
+      "step": 331
+    },
+    {
+      "epoch": 1.0049186530457812,
+      "grad_norm": 0.7931235959195142,
+      "learning_rate": 1.1632216041612595e-05,
+      "loss": 0.3182,
+      "step": 332
+    },
+    {
+      "epoch": 1.0079455164585698,
+      "grad_norm": 0.7318761706726017,
+      "learning_rate": 1.15800139597335e-05,
+      "loss": 0.325,
+      "step": 333
+    },
+    {
+      "epoch": 1.0109723798713584,
+      "grad_norm": 0.702460624427335,
+      "learning_rate": 1.1527767681467472e-05,
+      "loss": 0.3396,
+      "step": 334
+    },
+    {
+      "epoch": 1.0139992432841467,
+      "grad_norm": 0.8336829672884268,
+      "learning_rate": 1.1475478668255223e-05,
+      "loss": 0.3204,
+      "step": 335
+    },
+    {
+      "epoch": 1.0170261066969353,
+      "grad_norm": 0.7657232416758061,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.3224,
+      "step": 336
+    },
+    {
+      "epoch": 1.0200529701097238,
+      "grad_norm": 0.7939559098772329,
+      "learning_rate": 1.1370778288690947e-05,
+      "loss": 0.3162,
+      "step": 337
+    },
+    {
+      "epoch": 1.0230798335225122,
+      "grad_norm": 0.779202132777816,
+      "learning_rate": 1.1318369851033604e-05,
+      "loss": 0.3247,
+      "step": 338
+    },
+    {
+      "epoch": 1.0261066969353008,
+      "grad_norm": 0.715531171677848,
+      "learning_rate": 1.1265924535737494e-05,
+      "loss": 0.3095,
+      "step": 339
+    },
+    {
+      "epoch": 1.0291335603480893,
+      "grad_norm": 0.7165436797889654,
+      "learning_rate": 1.121344380981082e-05,
+      "loss": 0.3294,
+      "step": 340
+    },
+    {
+      "epoch": 1.0321604237608777,
+      "grad_norm": 0.8198583253890138,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.312,
+      "step": 341
+    },
+    {
+      "epoch": 1.0351872871736663,
+      "grad_norm": 0.7376023826689638,
+      "learning_rate": 1.1108381999010111e-05,
+      "loss": 0.3091,
+      "step": 342
+    },
+    {
+      "epoch": 1.0382141505864548,
+      "grad_norm": 0.8147297360498046,
+      "learning_rate": 1.1055803852940772e-05,
+      "loss": 0.3165,
+      "step": 343
+    },
+    {
+      "epoch": 1.0412410139992432,
+      "grad_norm": 0.7681053268800154,
+      "learning_rate": 1.1003196173768051e-05,
+      "loss": 0.3107,
+      "step": 344
+    },
+    {
+      "epoch": 1.0442678774120318,
+      "grad_norm": 0.804590538599873,
+      "learning_rate": 1.0950560433041825e-05,
+      "loss": 0.3139,
+      "step": 345
+    },
+    {
+      "epoch": 1.0472947408248203,
+      "grad_norm": 0.7485917341127918,
+      "learning_rate": 1.0897898103096917e-05,
+      "loss": 0.3144,
+      "step": 346
+    },
+    {
+      "epoch": 1.0503216042376087,
+      "grad_norm": 0.789495549417792,
+      "learning_rate": 1.0845210657011893e-05,
+      "loss": 0.3064,
+      "step": 347
+    },
+    {
+      "epoch": 1.0533484676503972,
+      "grad_norm": 0.8451806541906273,
+      "learning_rate": 1.0792499568567885e-05,
+      "loss": 0.3124,
+      "step": 348
+    },
+    {
+      "epoch": 1.0563753310631858,
+      "grad_norm": 0.7527830249588188,
+      "learning_rate": 1.0739766312207344e-05,
+      "loss": 0.3019,
+      "step": 349
+    },
+    {
+      "epoch": 1.0594021944759742,
+      "grad_norm": 0.8070796981497186,
+      "learning_rate": 1.068701236299281e-05,
+      "loss": 0.3134,
+      "step": 350
+    },
+    {
+      "epoch": 1.0624290578887627,
+      "grad_norm": 0.822488207588396,
+      "learning_rate": 1.0634239196565646e-05,
+      "loss": 0.3016,
+      "step": 351
+    },
+    {
+      "epoch": 1.0654559213015513,
+      "grad_norm": 0.7643798559209546,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.3136,
+      "step": 352
+    },
+    {
+      "epoch": 1.0684827847143397,
+      "grad_norm": 0.7533823001352534,
+      "learning_rate": 1.0528641117285315e-05,
+      "loss": 0.3205,
+      "step": 353
+    },
+    {
+      "epoch": 1.0715096481271282,
+      "grad_norm": 0.7105604004763676,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.2934,
+      "step": 354
+    },
+    {
+      "epoch": 1.0745365115399168,
+      "grad_norm": 0.7500819047045877,
+      "learning_rate": 1.0422983889504831e-05,
+      "loss": 0.3297,
+      "step": 355
+    },
+    {
+      "epoch": 1.0775633749527052,
+      "grad_norm": 0.7072289300014105,
+      "learning_rate": 1.0370136789003582e-05,
+      "loss": 0.3188,
+      "step": 356
+    },
+    {
+      "epoch": 1.0805902383654937,
+      "grad_norm": 0.7097705291249344,
+      "learning_rate": 1.031727933498068e-05,
+      "loss": 0.309,
+      "step": 357
+    },
+    {
+      "epoch": 1.0836171017782823,
+      "grad_norm": 0.678427961193918,
+      "learning_rate": 1.0264413005972736e-05,
+      "loss": 0.3176,
+      "step": 358
+    },
+    {
+      "epoch": 1.0866439651910706,
+      "grad_norm": 0.7853255555483389,
+      "learning_rate": 1.0211539280764617e-05,
+      "loss": 0.3121,
+      "step": 359
+    },
+    {
+      "epoch": 1.0896708286038592,
+      "grad_norm": 0.769437606325394,
+      "learning_rate": 1.015865963834808e-05,
+      "loss": 0.3113,
+      "step": 360
+    },
+    {
+      "epoch": 1.0926976920166478,
+      "grad_norm": 0.7247955320636588,
+      "learning_rate": 1.0105775557880398e-05,
+      "loss": 0.3231,
+      "step": 361
+    },
+    {
+      "epoch": 1.0957245554294364,
+      "grad_norm": 0.7608072142812601,
+      "learning_rate": 1.0052888518642978e-05,
+      "loss": 0.3233,
+      "step": 362
+    },
+    {
+      "epoch": 1.0987514188422247,
+      "grad_norm": 0.7071040856748567,
+      "learning_rate": 1e-05,
+      "loss": 0.3132,
+      "step": 363
+    },
+    {
+      "epoch": 1.1017782822550133,
+      "grad_norm": 0.7683687307161858,
+      "learning_rate": 9.947111481357023e-06,
+      "loss": 0.3073,
+      "step": 364
+    },
+    {
+      "epoch": 1.1048051456678016,
+      "grad_norm": 0.7843985585456426,
+      "learning_rate": 9.894224442119606e-06,
+      "loss": 0.3014,
+      "step": 365
+    },
+    {
+      "epoch": 1.1078320090805902,
+      "grad_norm": 0.7814186608586865,
+      "learning_rate": 9.841340361651921e-06,
+      "loss": 0.2998,
+      "step": 366
+    },
+    {
+      "epoch": 1.1108588724933788,
+      "grad_norm": 0.7630324172276898,
+      "learning_rate": 9.788460719235386e-06,
+      "loss": 0.31,
+      "step": 367
+    },
+    {
+      "epoch": 1.1138857359061674,
+      "grad_norm": 0.7802088373052618,
+      "learning_rate": 9.735586994027267e-06,
+      "loss": 0.307,
+      "step": 368
+    },
+    {
+      "epoch": 1.1169125993189557,
+      "grad_norm": 0.927996810592608,
+      "learning_rate": 9.682720665019325e-06,
+      "loss": 0.3213,
+      "step": 369
+    },
+    {
+      "epoch": 1.1199394627317443,
+      "grad_norm": 0.7653541588481467,
+      "learning_rate": 9.62986321099642e-06,
+      "loss": 0.2968,
+      "step": 370
+    },
+    {
+      "epoch": 1.1229663261445326,
+      "grad_norm": 0.8520826408392281,
+      "learning_rate": 9.57701611049517e-06,
+      "loss": 0.3103,
+      "step": 371
+    },
+    {
+      "epoch": 1.1259931895573212,
+      "grad_norm": 0.726618863678328,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.3098,
+      "step": 372
+    },
+    {
+      "epoch": 1.1290200529701098,
+      "grad_norm": 0.8179596957051172,
+      "learning_rate": 9.471358882714687e-06,
+      "loss": 0.3092,
+      "step": 373
+    },
+    {
+      "epoch": 1.1320469163828983,
+      "grad_norm": 0.7818665559671562,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.319,
+      "step": 374
+    },
+    {
+      "epoch": 1.1350737797956867,
+      "grad_norm": 0.8239256823755685,
+      "learning_rate": 9.365760803434356e-06,
+      "loss": 0.3196,
+      "step": 375
+    },
+    {
+      "epoch": 1.1381006432084753,
+      "grad_norm": 0.8082030153316792,
+      "learning_rate": 9.312987637007191e-06,
+      "loss": 0.3201,
+      "step": 376
+    },
+    {
+      "epoch": 1.1411275066212636,
+      "grad_norm": 0.7119970790457402,
+      "learning_rate": 9.260233687792657e-06,
+      "loss": 0.3046,
+      "step": 377
+    },
+    {
+      "epoch": 1.1441543700340522,
+      "grad_norm": 0.7788930783503916,
+      "learning_rate": 9.207500431432115e-06,
+      "loss": 0.3072,
+      "step": 378
+    },
+    {
+      "epoch": 1.1471812334468408,
+      "grad_norm": 0.775297955840287,
+      "learning_rate": 9.154789342988108e-06,
+      "loss": 0.3115,
+      "step": 379
+    },
+    {
+      "epoch": 1.1502080968596293,
+      "grad_norm": 0.759688379261278,
+      "learning_rate": 9.102101896903084e-06,
+      "loss": 0.3582,
+      "step": 380
+    },
+    {
+      "epoch": 1.1532349602724177,
+      "grad_norm": 0.7835125430105545,
+      "learning_rate": 9.049439566958176e-06,
+      "loss": 0.3266,
+      "step": 381
+    },
+    {
+      "epoch": 1.1562618236852062,
+      "grad_norm": 0.7741338988673719,
+      "learning_rate": 8.99680382623195e-06,
+      "loss": 0.3104,
+      "step": 382
+    },
+    {
+      "epoch": 1.1592886870979946,
+      "grad_norm": 0.6850740666145212,
+      "learning_rate": 8.944196147059233e-06,
+      "loss": 0.2939,
+      "step": 383
+    },
+    {
+      "epoch": 1.1623155505107832,
+      "grad_norm": 0.7784201161953679,
+      "learning_rate": 8.89161800098989e-06,
+      "loss": 0.3227,
+      "step": 384
+    },
+    {
+      "epoch": 1.1653424139235717,
+      "grad_norm": 0.7634364871910098,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.3158,
+      "step": 385
+    },
+    {
+      "epoch": 1.1683692773363603,
+      "grad_norm": 0.6766912944283272,
+      "learning_rate": 8.786556190189183e-06,
+      "loss": 0.2876,
+      "step": 386
+    },
+    {
+      "epoch": 1.1713961407491487,
+      "grad_norm": 0.6856957410882696,
+      "learning_rate": 8.734075464262507e-06,
+      "loss": 0.2992,
+      "step": 387
+    },
+    {
+      "epoch": 1.1744230041619372,
+      "grad_norm": 0.7449092617029119,
+      "learning_rate": 8.681630148966397e-06,
+      "loss": 0.3139,
+      "step": 388
+    },
+    {
+      "epoch": 1.1774498675747256,
+      "grad_norm": 0.6922348334364381,
+      "learning_rate": 8.629221711309056e-06,
+      "loss": 0.3046,
+      "step": 389
+    },
+    {
+      "epoch": 1.1804767309875142,
+      "grad_norm": 0.7370659424447896,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.3052,
+      "step": 390
+    },
+    {
+      "epoch": 1.1835035944003027,
+      "grad_norm": 0.684757302411277,
+      "learning_rate": 8.52452133174478e-06,
+      "loss": 0.2973,
+      "step": 391
+    },
+    {
+      "epoch": 1.1865304578130913,
+      "grad_norm": 0.6946037809770749,
+      "learning_rate": 8.472232318532531e-06,
+      "loss": 0.3024,
+      "step": 392
+    },
+    {
+      "epoch": 1.1895573212258796,
+      "grad_norm": 0.7521647824962547,
+      "learning_rate": 8.419986040266502e-06,
+      "loss": 0.3047,
+      "step": 393
+    },
+    {
+      "epoch": 1.1925841846386682,
+      "grad_norm": 0.7247277049420847,
+      "learning_rate": 8.367783958387407e-06,
+      "loss": 0.2949,
+      "step": 394
+    },
+    {
+      "epoch": 1.1956110480514566,
+      "grad_norm": 0.6889462368971019,
+      "learning_rate": 8.315627533099697e-06,
+      "loss": 0.3005,
+      "step": 395
+    },
+    {
+      "epoch": 1.1986379114642451,
+      "grad_norm": 0.727966245526417,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.3126,
+      "step": 396
+    },
+    {
+      "epoch": 1.2016647748770337,
+      "grad_norm": 0.7440372879494809,
+      "learning_rate": 8.211457486689829e-06,
+      "loss": 0.2991,
+      "step": 397
+    },
+    {
+      "epoch": 1.2046916382898223,
+      "grad_norm": 0.7104899915035207,
+      "learning_rate": 8.159446779427798e-06,
+      "loss": 0.3073,
+      "step": 398
+    },
+    {
+      "epoch": 1.2077185017026106,
+      "grad_norm": 0.6949187614660589,
+      "learning_rate": 8.107487556395902e-06,
+      "loss": 0.294,
+      "step": 399
+    },
+    {
+      "epoch": 1.2107453651153992,
+      "grad_norm": 0.7192907858802814,
+      "learning_rate": 8.055581271005292e-06,
+      "loss": 0.3062,
+      "step": 400
+    },
+    {
+      "epoch": 1.2137722285281876,
+      "grad_norm": 0.7169073202413152,
+      "learning_rate": 8.00372937518636e-06,
+      "loss": 0.3049,
+      "step": 401
+    },
+    {
+      "epoch": 1.2167990919409761,
+      "grad_norm": 0.8024554846710744,
+      "learning_rate": 7.951933319348095e-06,
+      "loss": 0.3098,
+      "step": 402
+    },
+    {
+      "epoch": 1.2198259553537647,
+      "grad_norm": 0.7161385906662583,
+      "learning_rate": 7.900194552337516e-06,
+      "loss": 0.3204,
+      "step": 403
+    },
+    {
+      "epoch": 1.2228528187665533,
+      "grad_norm": 0.7372316359915377,
+      "learning_rate": 7.848514521399167e-06,
+      "loss": 0.32,
+      "step": 404
+    },
+    {
+      "epoch": 1.2258796821793416,
+      "grad_norm": 0.6769196543314566,
+      "learning_rate": 7.796894672134594e-06,
+      "loss": 0.3017,
+      "step": 405
+    },
+    {
+      "epoch": 1.2289065455921302,
+      "grad_norm": 0.6739286272657946,
+      "learning_rate": 7.745336448461958e-06,
+      "loss": 0.2952,
+      "step": 406
+    },
+    {
+      "epoch": 1.2319334090049185,
+      "grad_norm": 0.6894304362747847,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.2987,
+      "step": 407
+    },
+    {
+      "epoch": 1.2349602724177071,
+      "grad_norm": 0.7665718249824524,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.3112,
+      "step": 408
+    },
+    {
+      "epoch": 1.2379871358304957,
+      "grad_norm": 0.7099302787076011,
+      "learning_rate": 7.591045944078119e-06,
+      "loss": 0.3034,
+      "step": 409
+    },
+    {
+      "epoch": 1.2410139992432843,
+      "grad_norm": 0.7500616055255146,
+      "learning_rate": 7.539748626873866e-06,
+      "loss": 0.3052,
+      "step": 410
+    },
+    {
+      "epoch": 1.2440408626560726,
+      "grad_norm": 0.7353764731484195,
+      "learning_rate": 7.488520128189209e-06,
+      "loss": 0.3145,
+      "step": 411
+    },
+    {
+      "epoch": 1.2470677260688612,
+      "grad_norm": 0.7278381405034842,
+      "learning_rate": 7.4373618809953755e-06,
+      "loss": 0.2942,
+      "step": 412
+    },
+    {
+      "epoch": 1.2500945894816495,
+      "grad_norm": 0.7477255719967081,
+      "learning_rate": 7.386275316298513e-06,
+      "loss": 0.2965,
+      "step": 413
+    },
+    {
+      "epoch": 1.253121452894438,
+      "grad_norm": 0.7004379186447255,
+      "learning_rate": 7.335261863099652e-06,
+      "loss": 0.2843,
+      "step": 414
+    },
+    {
+      "epoch": 1.2561483163072267,
+      "grad_norm": 0.7426177532567828,
+      "learning_rate": 7.2843229483547405e-06,
+      "loss": 0.3167,
+      "step": 415
+    },
+    {
+      "epoch": 1.2591751797200152,
+      "grad_norm": 0.6810230548909262,
+      "learning_rate": 7.233459996934731e-06,
+      "loss": 0.3191,
+      "step": 416
+    },
+    {
+      "epoch": 1.2622020431328036,
+      "grad_norm": 0.763150110283304,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.3071,
+      "step": 417
+    },
+    {
+      "epoch": 1.2652289065455922,
+      "grad_norm": 0.7770950993962036,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.3114,
+      "step": 418
+    },
+    {
+      "epoch": 1.2682557699583805,
+      "grad_norm": 0.7973805095749931,
+      "learning_rate": 7.081341139221955e-06,
+      "loss": 0.3253,
+      "step": 419
+    },
+    {
+      "epoch": 1.271282633371169,
+      "grad_norm": 0.6658189305246623,
+      "learning_rate": 7.0307962467172555e-06,
+      "loss": 0.3069,
+      "step": 420
+    },
+    {
+      "epoch": 1.2743094967839577,
+      "grad_norm": 0.7900843685093274,
+      "learning_rate": 6.9803344092242855e-06,
+      "loss": 0.3194,
+      "step": 421
+    },
+    {
+      "epoch": 1.2773363601967462,
+      "grad_norm": 0.7593287661689743,
+      "learning_rate": 6.929957038269123e-06,
+      "loss": 0.3112,
+      "step": 422
+    },
+    {
+      "epoch": 1.2803632236095346,
+      "grad_norm": 0.7476270854671544,
+      "learning_rate": 6.87966554301513e-06,
+      "loss": 0.32,
+      "step": 423
+    },
+    {
+      "epoch": 1.2833900870223232,
+      "grad_norm": 0.7197879153788757,
+      "learning_rate": 6.8294613302235325e-06,
+      "loss": 0.2909,
+      "step": 424
+    },
+    {
+      "epoch": 1.2864169504351115,
+      "grad_norm": 0.7038909833180136,
+      "learning_rate": 6.779345804214088e-06,
+      "loss": 0.3036,
+      "step": 425
+    },
+    {
+      "epoch": 1.2894438138479,
+      "grad_norm": 0.7699317952309939,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.2987,
+      "step": 426
+    },
+    {
+      "epoch": 1.2924706772606886,
+      "grad_norm": 0.7439102778498503,
+      "learning_rate": 6.679386417377649e-06,
+      "loss": 0.3053,
+      "step": 427
+    },
+    {
+      "epoch": 1.2954975406734772,
+      "grad_norm": 0.724704097937217,
+      "learning_rate": 6.629545352629583e-06,
+      "loss": 0.3095,
+      "step": 428
+    },
+    {
+      "epoch": 1.2985244040862656,
+      "grad_norm": 0.7128341162875507,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.2999,
+      "step": 429
+    },
+    {
+      "epoch": 1.3015512674990541,
+      "grad_norm": 0.7494357439641058,
+      "learning_rate": 6.530147451243377e-06,
+      "loss": 0.3201,
+      "step": 430
+    },
+    {
+      "epoch": 1.3045781309118425,
+      "grad_norm": 0.8475818603266285,
+      "learning_rate": 6.480593394978208e-06,
+      "loss": 0.3315,
+      "step": 431
+    },
+    {
+      "epoch": 1.307604994324631,
+      "grad_norm": 0.7610619827644427,
+      "learning_rate": 6.431137784081283e-06,
+      "loss": 0.3113,
+      "step": 432
+    },
+    {
+      "epoch": 1.3106318577374196,
+      "grad_norm": 0.8066020643232051,
+      "learning_rate": 6.381782001932352e-06,
+      "loss": 0.3019,
+      "step": 433
+    },
+    {
+      "epoch": 1.3136587211502082,
+      "grad_norm": 0.8072942059486607,
+      "learning_rate": 6.33252742911874e-06,
+      "loss": 0.3086,
+      "step": 434
+    },
+    {
+      "epoch": 1.3166855845629966,
+      "grad_norm": 0.79640310211253,
+      "learning_rate": 6.283375443396726e-06,
+      "loss": 0.3053,
+      "step": 435
+    },
+    {
+      "epoch": 1.3197124479757851,
+      "grad_norm": 0.7402456600510771,
+      "learning_rate": 6.234327419653013e-06,
+      "loss": 0.3364,
+      "step": 436
+    },
+    {
+      "epoch": 1.3227393113885735,
+      "grad_norm": 0.7705827896648257,
+      "learning_rate": 6.185384729866264e-06,
+      "loss": 0.3019,
+      "step": 437
+    },
+    {
+      "epoch": 1.325766174801362,
+      "grad_norm": 0.7872719953440453,
+      "learning_rate": 6.136548743068713e-06,
+      "loss": 0.3103,
+      "step": 438
+    },
+    {
+      "epoch": 1.3287930382141506,
+      "grad_norm": 0.7702288430902404,
+      "learning_rate": 6.087820825307904e-06,
+      "loss": 0.3047,
+      "step": 439
+    },
+    {
+      "epoch": 1.3318199016269392,
+      "grad_norm": 0.7255015032562324,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.3108,
+      "step": 440
+    },
+    {
+      "epoch": 1.3348467650397275,
+      "grad_norm": 0.7372082487711056,
+      "learning_rate": 5.990694645933866e-06,
+      "loss": 0.2957,
+      "step": 441
+    },
+    {
+      "epoch": 1.3378736284525161,
+      "grad_norm": 0.7967190910457529,
+      "learning_rate": 5.9422991011486635e-06,
+      "loss": 0.2924,
+      "step": 442
+    },
+    {
+      "epoch": 1.3409004918653045,
+      "grad_norm": 0.7417852505135866,
+      "learning_rate": 5.894017058980249e-06,
+      "loss": 0.2864,
+      "step": 443
+    },
+    {
+      "epoch": 1.343927355278093,
+      "grad_norm": 0.7289948405053147,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.2995,
+      "step": 444
+    },
+    {
+      "epoch": 1.3469542186908816,
+      "grad_norm": 0.7342788124970788,
+      "learning_rate": 5.797798881491138e-06,
+      "loss": 0.3138,
+      "step": 445
+    },
+    {
+      "epoch": 1.3499810821036702,
+      "grad_norm": 0.7224803534024413,
+      "learning_rate": 5.749865437599703e-06,
+      "loss": 0.3209,
+      "step": 446
+    },
+    {
+      "epoch": 1.3530079455164585,
+      "grad_norm": 0.7218116352555798,
+      "learning_rate": 5.702050879108284e-06,
+      "loss": 0.2919,
+      "step": 447
+    },
+    {
+      "epoch": 1.356034808929247,
+      "grad_norm": 0.7097099662676319,
+      "learning_rate": 5.654356543492883e-06,
+      "loss": 0.3027,
+      "step": 448
+    },
+    {
+      "epoch": 1.3590616723420355,
+      "grad_norm": 0.7017842050686611,
+      "learning_rate": 5.606783764866576e-06,
+      "loss": 0.2892,
+      "step": 449
+    },
+    {
+      "epoch": 1.362088535754824,
+      "grad_norm": 0.7101909671181817,
+      "learning_rate": 5.559333873942259e-06,
+      "loss": 0.2809,
+      "step": 450
+    },
+    {
+      "epoch": 1.3651153991676126,
+      "grad_norm": 0.7200315851322242,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.3064,
+      "step": 451
+    },
+    {
+      "epoch": 1.3681422625804012,
+      "grad_norm": 0.7374570123311738,
+      "learning_rate": 5.464808060826825e-06,
+      "loss": 0.3037,
+      "step": 452
+    },
+    {
+      "epoch": 1.3711691259931895,
+      "grad_norm": 0.7196533485506698,
+      "learning_rate": 5.417734782725896e-06,
+      "loss": 0.2917,
+      "step": 453
+    },
+    {
+      "epoch": 1.374195989405978,
+      "grad_norm": 0.7716392107905302,
+      "learning_rate": 5.370789680433376e-06,
+      "loss": 0.3063,
+      "step": 454
+    },
+    {
+      "epoch": 1.3772228528187664,
+      "grad_norm": 0.7470317135563169,
+      "learning_rate": 5.323974067104687e-06,
+      "loss": 0.2969,
+      "step": 455
+    },
+    {
+      "epoch": 1.380249716231555,
+      "grad_norm": 0.7347758763914756,
+      "learning_rate": 5.277289252273175e-06,
+      "loss": 0.3141,
+      "step": 456
+    },
+    {
+      "epoch": 1.3832765796443436,
+      "grad_norm": 0.6975451401489193,
+      "learning_rate": 5.230736541813463e-06,
+      "loss": 0.3131,
+      "step": 457
+    },
+    {
+      "epoch": 1.3863034430571322,
+      "grad_norm": 0.7669601419271628,
+      "learning_rate": 5.184317237904939e-06,
+      "loss": 0.2986,
+      "step": 458
+    },
+    {
+      "epoch": 1.3893303064699205,
+      "grad_norm": 0.7660536618774079,
+      "learning_rate": 5.138032638995315e-06,
+      "loss": 0.3152,
+      "step": 459
+    },
+    {
+      "epoch": 1.392357169882709,
+      "grad_norm": 0.7795254627435152,
+      "learning_rate": 5.091884039764321e-06,
+      "loss": 0.2836,
+      "step": 460
+    },
+    {
+      "epoch": 1.3953840332954974,
+      "grad_norm": 0.7059557313555045,
+      "learning_rate": 5.045872731087479e-06,
+      "loss": 0.3094,
+      "step": 461
+    },
+    {
+      "epoch": 1.398410896708286,
+      "grad_norm": 0.6967246887596856,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.3085,
+      "step": 462
+    },
+    {
+      "epoch": 1.4014377601210746,
+      "grad_norm": 0.6968747497932463,
+      "learning_rate": 4.954267129660789e-06,
+      "loss": 0.3041,
+      "step": 463
+    },
+    {
+      "epoch": 1.4044646235338631,
+      "grad_norm": 0.7279670101137755,
+      "learning_rate": 4.908675399316534e-06,
+      "loss": 0.3019,
+      "step": 464
+    },
+    {
+      "epoch": 1.4074914869466515,
+      "grad_norm": 1.2600618763313924,
+      "learning_rate": 4.863226084265939e-06,
+      "loss": 0.3136,
+      "step": 465
+    },
+    {
+      "epoch": 1.41051835035944,
+      "grad_norm": 0.6773494354805526,
+      "learning_rate": 4.817920455824045e-06,
+      "loss": 0.2947,
+      "step": 466
+    },
+    {
+      "epoch": 1.4135452137722284,
+      "grad_norm": 0.7672986257708132,
+      "learning_rate": 4.772759781286679e-06,
+      "loss": 0.3129,
+      "step": 467
+    },
+    {
+      "epoch": 1.416572077185017,
+      "grad_norm": 0.6972809472933599,
+      "learning_rate": 4.727745323894976e-06,
+      "loss": 0.2962,
+      "step": 468
+    },
+    {
+      "epoch": 1.4195989405978056,
+      "grad_norm": 0.7085333988411966,
+      "learning_rate": 4.682878342800087e-06,
+      "loss": 0.2872,
+      "step": 469
+    },
+    {
+      "epoch": 1.4226258040105941,
+      "grad_norm": 0.761812631060059,
+      "learning_rate": 4.638160093027908e-06,
+      "loss": 0.2999,
+      "step": 470
+    },
+    {
+      "epoch": 1.4256526674233825,
+      "grad_norm": 0.692839337346362,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.3108,
+      "step": 471
+    },
+    {
+      "epoch": 1.428679530836171,
+      "grad_norm": 0.7060175335375108,
+      "learning_rate": 4.549174786718684e-06,
+      "loss": 0.2952,
+      "step": 472
+    },
+    {
+      "epoch": 1.4317063942489594,
+      "grad_norm": 0.7179712607147729,
+      "learning_rate": 4.504910219291941e-06,
+      "loss": 0.3036,
+      "step": 473
+    },
+    {
+      "epoch": 1.434733257661748,
+      "grad_norm": 0.7341286293360819,
+      "learning_rate": 4.460799361338898e-06,
+      "loss": 0.3072,
+      "step": 474
+    },
+    {
+      "epoch": 1.4377601210745365,
+      "grad_norm": 0.7068358588632465,
+      "learning_rate": 4.416843446735077e-06,
+      "loss": 0.2927,
+      "step": 475
+    },
+    {
+      "epoch": 1.4407869844873251,
+      "grad_norm": 0.7014414370936332,
+      "learning_rate": 4.373043705021899e-06,
+      "loss": 0.2792,
+      "step": 476
+    },
+    {
+      "epoch": 1.4438138479001135,
+      "grad_norm": 0.6932814196655317,
+      "learning_rate": 4.3294013613722944e-06,
+      "loss": 0.3004,
+      "step": 477
+    },
+    {
+      "epoch": 1.446840711312902,
+      "grad_norm": 0.7104733050465204,
+      "learning_rate": 4.2859176365564294e-06,
+      "loss": 0.3019,
+      "step": 478
+    },
+    {
+      "epoch": 1.4498675747256904,
+      "grad_norm": 0.681839341345162,
+      "learning_rate": 4.2425937469075626e-06,
+      "loss": 0.2991,
+      "step": 479
+    },
+    {
+      "epoch": 1.452894438138479,
+      "grad_norm": 0.7045670978776379,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.3093,
+      "step": 480
+    },
+    {
+      "epoch": 1.4559213015512675,
+      "grad_norm": 0.7162157338746798,
+      "learning_rate": 4.1564303160552935e-06,
+      "loss": 0.2996,
+      "step": 481
+    },
+    {
+      "epoch": 1.458948164964056,
+      "grad_norm": 0.7125304866191854,
+      "learning_rate": 4.113593185028273e-06,
+      "loss": 0.3074,
+      "step": 482
+    },
+    {
+      "epoch": 1.4619750283768445,
+      "grad_norm": 0.6771311380281757,
+      "learning_rate": 4.070920709453597e-06,
+      "loss": 0.2986,
+      "step": 483
+    },
+    {
+      "epoch": 1.465001891789633,
+      "grad_norm": 0.7040241292554285,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.2931,
+      "step": 484
+    },
+    {
+      "epoch": 1.4680287552024214,
+      "grad_norm": 0.7309274386644186,
+      "learning_rate": 3.986074494585619e-06,
+      "loss": 0.3234,
+      "step": 485
+    },
+    {
+      "epoch": 1.47105561861521,
+      "grad_norm": 0.6829437135863815,
+      "learning_rate": 3.943903128623336e-06,
+      "loss": 0.284,
+      "step": 486
+    },
+    {
+      "epoch": 1.4740824820279985,
+      "grad_norm": 0.7794304721658968,
+      "learning_rate": 3.9019011647090465e-06,
+      "loss": 0.3081,
+      "step": 487
+    },
+    {
+      "epoch": 1.477109345440787,
+      "grad_norm": 0.7555116536191332,
+      "learning_rate": 3.860069777727983e-06,
+      "loss": 0.2923,
+      "step": 488
+    },
+    {
+      "epoch": 1.4801362088535754,
+      "grad_norm": 0.6703912141554664,
+      "learning_rate": 3.818410137793947e-06,
+      "loss": 0.2962,
+      "step": 489
+    },
+    {
+      "epoch": 1.483163072266364,
+      "grad_norm": 0.698048853364522,
+      "learning_rate": 3.7769234102166365e-06,
+      "loss": 0.2958,
+      "step": 490
+    },
+    {
+      "epoch": 1.4861899356791524,
+      "grad_norm": 0.6985918932177311,
+      "learning_rate": 3.735610755468988e-06,
+      "loss": 0.292,
+      "step": 491
+    },
+    {
+      "epoch": 1.489216799091941,
+      "grad_norm": 0.7536616637145266,
+      "learning_rate": 3.6944733291547784e-06,
+      "loss": 0.3162,
+      "step": 492
+    },
+    {
+      "epoch": 1.4922436625047295,
+      "grad_norm": 0.6986329345064591,
+      "learning_rate": 3.653512281976238e-06,
+      "loss": 0.2935,
+      "step": 493
+    },
+    {
+      "epoch": 1.495270525917518,
+      "grad_norm": 0.7054014773958908,
+      "learning_rate": 3.612728759701919e-06,
+      "loss": 0.3032,
+      "step": 494
+    },
+    {
+      "epoch": 1.4982973893303064,
+      "grad_norm": 0.7242941265735119,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.3066,
+      "step": 495
+    },
+    {
+      "epoch": 1.501324252743095,
+      "grad_norm": 0.6732093586241171,
+      "learning_rate": 3.5316988480794255e-06,
+      "loss": 0.2931,
+      "step": 496
+    },
+    {
+      "epoch": 1.5043511161558833,
+      "grad_norm": 0.7694492798695871,
+      "learning_rate": 3.4914547253120655e-06,
+      "loss": 0.3131,
+      "step": 497
+    },
+    {
+      "epoch": 1.507377979568672,
+      "grad_norm": 0.7105887277912891,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.2903,
+      "step": 498
+    },
+    {
+      "epoch": 1.5104048429814605,
+      "grad_norm": 0.6882742078109515,
+      "learning_rate": 3.4115137744067516e-06,
+      "loss": 0.2988,
+      "step": 499
+    },
+    {
+      "epoch": 1.513431706394249,
+      "grad_norm": 0.7346187512280352,
+      "learning_rate": 3.37181918238904e-06,
+      "loss": 0.3016,
+      "step": 500
+    },
+    {
+      "epoch": 1.5164585698070374,
+      "grad_norm": 0.7087019316452556,
+      "learning_rate": 3.3323099948370853e-06,
+      "loss": 0.3163,
+      "step": 501
+    },
+    {
+      "epoch": 1.519485433219826,
+      "grad_norm": 0.7391978786960687,
+      "learning_rate": 3.292987316907792e-06,
+      "loss": 0.2982,
+      "step": 502
+    },
+    {
+      "epoch": 1.5225122966326143,
+      "grad_norm": 0.7223515176362528,
+      "learning_rate": 3.253852248540994e-06,
+      "loss": 0.2955,
+      "step": 503
+    },
+    {
+      "epoch": 1.525539160045403,
+      "grad_norm": 0.7372263853461524,
+      "learning_rate": 3.2149058844286796e-06,
+      "loss": 0.3142,
+      "step": 504
+    },
+    {
+      "epoch": 1.5285660234581915,
+      "grad_norm": 0.6769419925047941,
+      "learning_rate": 3.1761493139843734e-06,
+      "loss": 0.2926,
+      "step": 505
+    },
+    {
+      "epoch": 1.53159288687098,
+      "grad_norm": 0.7092278996541448,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.2827,
+      "step": 506
+    },
+    {
+      "epoch": 1.5346197502837684,
+      "grad_norm": 0.6868020861182853,
+      "learning_rate": 3.099209885178882e-06,
+      "loss": 0.2907,
+      "step": 507
+    },
+    {
+      "epoch": 1.537646613696557,
+      "grad_norm": 0.6937755067727701,
+      "learning_rate": 3.0610291789789094e-06,
+      "loss": 0.2944,
+      "step": 508
+    },
+    {
+      "epoch": 1.5406734771093453,
+      "grad_norm": 0.7074866378922333,
+      "learning_rate": 3.023042570709185e-06,
+      "loss": 0.2951,
+      "step": 509
+    },
+    {
+      "epoch": 1.543700340522134,
+      "grad_norm": 0.7404988430837182,
+      "learning_rate": 2.9852511229367862e-06,
+      "loss": 0.284,
+      "step": 510
+    },
+    {
+      "epoch": 1.5467272039349225,
+      "grad_norm": 0.7503407646480187,
+      "learning_rate": 2.9476558927697605e-06,
+      "loss": 0.307,
+      "step": 511
+    },
+    {
+      "epoch": 1.549754067347711,
+      "grad_norm": 0.6888703921269884,
+      "learning_rate": 2.9102579318274994e-06,
+      "loss": 0.3013,
+      "step": 512
+    },
+    {
+      "epoch": 1.5527809307604994,
+      "grad_norm": 0.7417086609687752,
+      "learning_rate": 2.8730582862113743e-06,
+      "loss": 0.3112,
+      "step": 513
+    },
+    {
+      "epoch": 1.555807794173288,
+      "grad_norm": 0.7334068736581798,
+      "learning_rate": 2.8360579964754277e-06,
+      "loss": 0.2916,
+      "step": 514
+    },
+    {
+      "epoch": 1.5588346575860763,
+      "grad_norm": 0.7241681794639987,
+      "learning_rate": 2.7992580975973136e-06,
+      "loss": 0.3184,
+      "step": 515
+    },
+    {
+      "epoch": 1.5618615209988649,
+      "grad_norm": 0.7307358918764819,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.3091,
+      "step": 516
+    },
+    {
+      "epoch": 1.5648883844116535,
+      "grad_norm": 0.7401827808278196,
+      "learning_rate": 2.726263584269513e-06,
+      "loss": 0.3032,
+      "step": 517
+    },
+    {
+      "epoch": 1.567915247824442,
+      "grad_norm": 0.7248108610692715,
+      "learning_rate": 2.690071011633284e-06,
+      "loss": 0.2779,
+      "step": 518
+    },
+    {
+      "epoch": 1.5709421112372304,
+      "grad_norm": 0.7408999654099044,
+      "learning_rate": 2.6540829134246683e-06,
+      "loss": 0.2988,
+      "step": 519
+    },
+    {
+      "epoch": 1.573968974650019,
+      "grad_norm": 0.7463671533640204,
+      "learning_rate": 2.618300296308135e-06,
+      "loss": 0.3026,
+      "step": 520
+    },
+    {
+      "epoch": 1.5769958380628073,
+      "grad_norm": 0.7639062080518795,
+      "learning_rate": 2.582724161200405e-06,
+      "loss": 0.3048,
+      "step": 521
+    },
+    {
+      "epoch": 1.5800227014755959,
+      "grad_norm": 0.6910474292894525,
+      "learning_rate": 2.5473555032424534e-06,
+      "loss": 0.3033,
+      "step": 522
+    },
+    {
+      "epoch": 1.5830495648883844,
+      "grad_norm": 0.7686751926779183,
+      "learning_rate": 2.5121953117716744e-06,
+      "loss": 0.3013,
+      "step": 523
+    },
+    {
+      "epoch": 1.586076428301173,
+      "grad_norm": 0.7224058005697255,
+      "learning_rate": 2.477244570294206e-06,
+      "loss": 0.2944,
+      "step": 524
+    },
+    {
+      "epoch": 1.5891032917139614,
+      "grad_norm": 0.6896927360290549,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.3079,
+      "step": 525
+    },
+    {
+      "epoch": 1.59213015512675,
+      "grad_norm": 0.7337988601604027,
+      "learning_rate": 2.4079753420225694e-06,
+      "loss": 0.2941,
+      "step": 526
+    },
+    {
+      "epoch": 1.5951570185395383,
+      "grad_norm": 0.6957005760492644,
+      "learning_rate": 2.3736587928376197e-06,
+      "loss": 0.2873,
+      "step": 527
+    },
+    {
+      "epoch": 1.5981838819523269,
+      "grad_norm": 0.6986713088928649,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.2869,
+      "step": 528
+    },
+    {
+      "epoch": 1.6012107453651154,
+      "grad_norm": 0.7588906677745827,
+      "learning_rate": 2.305666623880858e-06,
+      "loss": 0.3072,
+      "step": 529
+    },
+    {
+      "epoch": 1.604237608777904,
+      "grad_norm": 0.7143611012932533,
+      "learning_rate": 2.27199290599617e-06,
+      "loss": 0.2981,
+      "step": 530
+    },
+    {
+      "epoch": 1.6072644721906924,
+      "grad_norm": 0.7174267656357006,
+      "learning_rate": 2.2385353570824308e-06,
+      "loss": 0.3053,
+      "step": 531
+    },
+    {
+      "epoch": 1.610291335603481,
+      "grad_norm": 0.7472229032211092,
+      "learning_rate": 2.2052949130192136e-06,
+      "loss": 0.3109,
+      "step": 532
+    },
+    {
+      "epoch": 1.6133181990162693,
+      "grad_norm": 0.7243280849543813,
+      "learning_rate": 2.172272503613183e-06,
+      "loss": 0.2864,
+      "step": 533
+    },
+    {
+      "epoch": 1.6163450624290578,
+      "grad_norm": 0.6963565921378788,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.3023,
+      "step": 534
+    },
+    {
+      "epoch": 1.6193719258418464,
+      "grad_norm": 0.7416290021314224,
+      "learning_rate": 2.1068854774790783e-06,
+      "loss": 0.2973,
+      "step": 535
+    },
+    {
+      "epoch": 1.622398789254635,
+      "grad_norm": 0.7103282645790373,
+      "learning_rate": 2.0745226897666858e-06,
+      "loss": 0.3068,
+      "step": 536
+    },
+    {
+      "epoch": 1.6254256526674233,
+      "grad_norm": 0.7267520796486289,
+      "learning_rate": 2.0423815946916783e-06,
+      "loss": 0.2959,
+      "step": 537
+    },
+    {
+      "epoch": 1.628452516080212,
+      "grad_norm": 0.710761498179922,
+      "learning_rate": 2.010463091309587e-06,
+      "loss": 0.2952,
+      "step": 538
+    },
+    {
+      "epoch": 1.6314793794930003,
+      "grad_norm": 0.7225059726738401,
+      "learning_rate": 1.9787680724495617e-06,
+      "loss": 0.2924,
+      "step": 539
+    },
+    {
+      "epoch": 1.6345062429057888,
+      "grad_norm": 0.7082709480679593,
+      "learning_rate": 1.947297424689414e-06,
+      "loss": 0.2941,
+      "step": 540
+    },
+    {
+      "epoch": 1.6375331063185774,
+      "grad_norm": 0.7168677192021707,
+      "learning_rate": 1.9160520283308115e-06,
+      "loss": 0.3125,
+      "step": 541
+    },
+    {
+      "epoch": 1.640559969731366,
+      "grad_norm": 0.7525633676785706,
+      "learning_rate": 1.8850327573746584e-06,
+      "loss": 0.3068,
+      "step": 542
+    },
+    {
+      "epoch": 1.6435868331441543,
+      "grad_norm": 0.703377537564333,
+      "learning_rate": 1.854240479496643e-06,
+      "loss": 0.2974,
+      "step": 543
+    },
+    {
+      "epoch": 1.646613696556943,
+      "grad_norm": 0.798482687887647,
+      "learning_rate": 1.8236760560229715e-06,
+      "loss": 0.3115,
+      "step": 544
+    },
+    {
+      "epoch": 1.6496405599697312,
+      "grad_norm": 0.7302538792945725,
+      "learning_rate": 1.7933403419062689e-06,
+      "loss": 0.3008,
+      "step": 545
+    },
+    {
+      "epoch": 1.6526674233825198,
+      "grad_norm": 0.7443847337165238,
+      "learning_rate": 1.7632341857016733e-06,
+      "loss": 0.309,
+      "step": 546
+    },
+    {
+      "epoch": 1.6556942867953084,
+      "grad_norm": 0.7357073462172213,
+      "learning_rate": 1.7333584295430894e-06,
+      "loss": 0.2911,
+      "step": 547
+    },
+    {
+      "epoch": 1.658721150208097,
+      "grad_norm": 0.6958958649402552,
+      "learning_rate": 1.7037139091196396e-06,
+      "loss": 0.2963,
+      "step": 548
+    },
+    {
+      "epoch": 1.6617480136208853,
+      "grad_norm": 0.6812795446591681,
+      "learning_rate": 1.6743014536522872e-06,
+      "loss": 0.3022,
+      "step": 549
+    },
+    {
+      "epoch": 1.6647748770336739,
+      "grad_norm": 0.7287015051650604,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.2961,
+      "step": 550
+    },
+    {
+      "epoch": 1.6678017404464622,
+      "grad_norm": 0.7557057551479847,
+      "learning_rate": 1.616176021989926e-06,
+      "loss": 0.2903,
+      "step": 551
+    },
+    {
+      "epoch": 1.6708286038592508,
+      "grad_norm": 0.6862037732941804,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.2872,
+      "step": 552
+    },
+    {
+      "epoch": 1.6738554672720394,
+      "grad_norm": 0.7616328821517183,
+      "learning_rate": 1.558988638083616e-06,
+      "loss": 0.3091,
+      "step": 553
+    },
+    {
+      "epoch": 1.676882330684828,
+      "grad_norm": 0.7308032264614859,
+      "learning_rate": 1.5307487177120773e-06,
+      "loss": 0.2927,
+      "step": 554
+    },
+    {
+      "epoch": 1.6799091940976163,
+      "grad_norm": 0.8013527063032867,
+      "learning_rate": 1.5027457005048573e-06,
+      "loss": 0.3045,
+      "step": 555
+    },
+    {
+      "epoch": 1.6829360575104049,
+      "grad_norm": 0.7197738825060199,
+      "learning_rate": 1.4749803697665366e-06,
+      "loss": 0.2907,
+      "step": 556
+    },
+    {
+      "epoch": 1.6859629209231932,
+      "grad_norm": 0.710276295134691,
+      "learning_rate": 1.4474535021531099e-06,
+      "loss": 0.2871,
+      "step": 557
+    },
+    {
+      "epoch": 1.6889897843359818,
+      "grad_norm": 0.6912938150098465,
+      "learning_rate": 1.4201658676502294e-06,
+      "loss": 0.2837,
+      "step": 558
+    },
+    {
+      "epoch": 1.6920166477487704,
+      "grad_norm": 0.7186640870545189,
+      "learning_rate": 1.3931182295516965e-06,
+      "loss": 0.2988,
+      "step": 559
+    },
+    {
+      "epoch": 1.695043511161559,
+      "grad_norm": 0.7584668240225306,
+      "learning_rate": 1.3663113444380905e-06,
+      "loss": 0.3047,
+      "step": 560
+    },
+    {
+      "epoch": 1.6980703745743473,
+      "grad_norm": 0.7095343578668626,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2942,
+      "step": 561
+    },
+    {
+      "epoch": 1.7010972379871359,
+      "grad_norm": 1.1529852794262418,
+      "learning_rate": 1.3134228257951142e-06,
+      "loss": 0.292,
+      "step": 562
+    },
+    {
+      "epoch": 1.7041241013999242,
+      "grad_norm": 0.6867090436587207,
+      "learning_rate": 1.2873426716713012e-06,
+      "loss": 0.2916,
+      "step": 563
+    },
+    {
+      "epoch": 1.7071509648127128,
+      "grad_norm": 0.7109074940818713,
+      "learning_rate": 1.2615062293021508e-06,
+      "loss": 0.2931,
+      "step": 564
+    },
+    {
+      "epoch": 1.7101778282255014,
+      "grad_norm": 0.6745459510890884,
+      "learning_rate": 1.2359142213884933e-06,
+      "loss": 0.2917,
+      "step": 565
+    },
+    {
+      "epoch": 1.71320469163829,
+      "grad_norm": 0.719647607531641,
+      "learning_rate": 1.2105673637938054e-06,
+      "loss": 0.2936,
+      "step": 566
+    },
+    {
+      "epoch": 1.7162315550510783,
+      "grad_norm": 0.7134213512474908,
+      "learning_rate": 1.1854663655241804e-06,
+      "loss": 0.3059,
+      "step": 567
+    },
+    {
+      "epoch": 1.7192584184638668,
+      "grad_norm": 0.7378885819959448,
+      "learning_rate": 1.1606119287084982e-06,
+      "loss": 0.3173,
+      "step": 568
+    },
+    {
+      "epoch": 1.7222852818766552,
+      "grad_norm": 0.7142493283676818,
+      "learning_rate": 1.136004748578785e-06,
+      "loss": 0.2975,
+      "step": 569
+    },
+    {
+      "epoch": 1.7253121452894438,
+      "grad_norm": 0.7127066025048728,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.2925,
+      "step": 570
+    },
+    {
+      "epoch": 1.7283390087022323,
+      "grad_norm": 0.7108316248913543,
+      "learning_rate": 1.0875349047046113e-06,
+      "loss": 0.2871,
+      "step": 571
+    },
+    {
+      "epoch": 1.731365872115021,
+      "grad_norm": 0.7113635464687907,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.2895,
+      "step": 572
+    },
+    {
+      "epoch": 1.7343927355278093,
+      "grad_norm": 0.7041888031890677,
+      "learning_rate": 1.0400622570866426e-06,
+      "loss": 0.2958,
+      "step": 573
+    },
+    {
+      "epoch": 1.7374195989405978,
+      "grad_norm": 0.6879565600787596,
+      "learning_rate": 1.0167015461268303e-06,
+      "loss": 0.2919,
+      "step": 574
+    },
+    {
+      "epoch": 1.7404464623533862,
+      "grad_norm": 0.6818544506089125,
+      "learning_rate": 9.935921173357444e-07,
+      "loss": 0.2901,
+      "step": 575
+    },
+    {
+      "epoch": 1.7434733257661748,
+      "grad_norm": 0.7232202682049852,
+      "learning_rate": 9.707346171337895e-07,
+      "loss": 0.2927,
+      "step": 576
+    },
+    {
+      "epoch": 1.7465001891789633,
+      "grad_norm": 0.69155835293231,
+      "learning_rate": 9.481296848943744e-07,
+      "loss": 0.2886,
+      "step": 577
+    },
+    {
+      "epoch": 1.749527052591752,
+      "grad_norm": 0.6990918933428112,
+      "learning_rate": 9.257779529260558e-07,
+      "loss": 0.3027,
+      "step": 578
+    },
+    {
+      "epoch": 1.7525539160045402,
+      "grad_norm": 0.6876891649313901,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.2811,
+      "step": 579
+    },
+    {
+      "epoch": 1.7555807794173288,
+      "grad_norm": 0.6741361597647003,
+      "learning_rate": 8.818365836066101e-07,
+      "loss": 0.2895,
+      "step": 580
+    },
+    {
+      "epoch": 1.7586076428301172,
+      "grad_norm": 0.6836853522200241,
+      "learning_rate": 8.602481753900427e-07,
+      "loss": 0.291,
+      "step": 581
+    },
+    {
+      "epoch": 1.7616345062429057,
+      "grad_norm": 0.6759290662031785,
+      "learning_rate": 8.389154256793042e-07,
+      "loss": 0.2968,
+      "step": 582
+    },
+    {
+      "epoch": 1.7646613696556943,
+      "grad_norm": 0.7126253318598841,
+      "learning_rate": 8.178389311972612e-07,
+      "loss": 0.3259,
+      "step": 583
+    },
+    {
+      "epoch": 1.7676882330684829,
+      "grad_norm": 0.7313812388298458,
+      "learning_rate": 7.970192814987676e-07,
+      "loss": 0.2946,
+      "step": 584
+    },
+    {
+      "epoch": 1.7707150964812712,
+      "grad_norm": 0.694619325361865,
+      "learning_rate": 7.764570589541876e-07,
+      "loss": 0.2825,
+      "step": 585
+    },
+    {
+      "epoch": 1.7737419598940598,
+      "grad_norm": 0.6834714077924963,
+      "learning_rate": 7.561528387330797e-07,
+      "loss": 0.2831,
+      "step": 586
+    },
+    {
+      "epoch": 1.7767688233068482,
+      "grad_norm": 0.7150338151105461,
+      "learning_rate": 7.361071887881376e-07,
+      "loss": 0.2928,
+      "step": 587
+    },
+    {
+      "epoch": 1.7797956867196367,
+      "grad_norm": 0.7071827809077127,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.2743,
+      "step": 588
+    },
+    {
+      "epoch": 1.7828225501324253,
+      "grad_norm": 0.7459136029575344,
+      "learning_rate": 6.96793835357964e-07,
+      "loss": 0.2978,
+      "step": 589
+    },
+    {
+      "epoch": 1.7858494135452139,
+      "grad_norm": 0.7309069849490312,
+      "learning_rate": 6.775272315517423e-07,
+      "loss": 0.288,
+      "step": 590
+    },
+    {
+      "epoch": 1.7888762769580022,
+      "grad_norm": 0.7399741226892932,
+      "learning_rate": 6.585213973489335e-07,
+      "loss": 0.3167,
+      "step": 591
+    },
+    {
+      "epoch": 1.7919031403707908,
+      "grad_norm": 0.7115733254116231,
+      "learning_rate": 6.397768643835755e-07,
+      "loss": 0.3009,
+      "step": 592
+    },
+    {
+      "epoch": 1.7949300037835791,
+      "grad_norm": 0.7856145719315364,
+      "learning_rate": 6.212941569805508e-07,
+      "loss": 0.306,
+      "step": 593
+    },
+    {
+      "epoch": 1.7979568671963677,
+      "grad_norm": 0.715686601104013,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.3007,
+      "step": 594
+    },
+    {
+      "epoch": 1.8009837306091563,
+      "grad_norm": 0.669073783674399,
+      "learning_rate": 5.851162795274445e-07,
+      "loss": 0.2935,
+      "step": 595
+    },
+    {
+      "epoch": 1.8040105940219449,
+      "grad_norm": 0.7155545014531404,
+      "learning_rate": 5.674221214503639e-07,
+      "loss": 0.2831,
+      "step": 596
+    },
+    {
+      "epoch": 1.8070374574347332,
+      "grad_norm": 0.6891705275950715,
+      "learning_rate": 5.499918128533155e-07,
+      "loss": 0.2772,
+      "step": 597
+    },
+    {
+      "epoch": 1.8100643208475218,
+      "grad_norm": 0.7319475663825057,
+      "learning_rate": 5.328258412994958e-07,
+      "loss": 0.3004,
+      "step": 598
+    },
+    {
+      "epoch": 1.8130911842603101,
+      "grad_norm": 0.7241713967891835,
+      "learning_rate": 5.159246869580348e-07,
+      "loss": 0.3022,
+      "step": 599
+    },
+    {
+      "epoch": 1.8161180476730987,
+      "grad_norm": 0.7354410284141902,
+      "learning_rate": 4.992888225905467e-07,
+      "loss": 0.2979,
+      "step": 600
+    },
+    {
+      "epoch": 1.8191449110858873,
+      "grad_norm": 0.730431130347255,
+      "learning_rate": 4.829187135379221e-07,
+      "loss": 0.3089,
+      "step": 601
+    },
+    {
+      "epoch": 1.8221717744986758,
+      "grad_norm": 0.7179239705338737,
+      "learning_rate": 4.6681481770729844e-07,
+      "loss": 0.2969,
+      "step": 602
+    },
+    {
+      "epoch": 1.8251986379114642,
+      "grad_norm": 0.6849092906508351,
+      "learning_rate": 4.509775855592613e-07,
+      "loss": 0.2947,
+      "step": 603
+    },
+    {
+      "epoch": 1.8282255013242528,
+      "grad_norm": 0.708829154601995,
+      "learning_rate": 4.354074600952407e-07,
+      "loss": 0.2953,
+      "step": 604
+    },
+    {
+      "epoch": 1.8312523647370411,
+      "grad_norm": 0.7108524471804767,
+      "learning_rate": 4.2010487684511105e-07,
+      "loss": 0.2853,
+      "step": 605
+    },
+    {
+      "epoch": 1.8342792281498297,
+      "grad_norm": 0.7146567350827009,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.2876,
+      "step": 606
+    },
+    {
+      "epoch": 1.8373060915626183,
+      "grad_norm": 0.710615158492881,
+      "learning_rate": 3.9030404167542777e-07,
+      "loss": 0.2986,
+      "step": 607
+    },
+    {
+      "epoch": 1.8403329549754068,
+      "grad_norm": 0.6693821527608284,
+      "learning_rate": 3.7580662334929517e-07,
+      "loss": 0.303,
+      "step": 608
+    },
+    {
+      "epoch": 1.8433598183881952,
+      "grad_norm": 0.7067047477092284,
+      "learning_rate": 3.615784144005796e-07,
+      "loss": 0.3073,
+      "step": 609
+    },
+    {
+      "epoch": 1.8463866818009838,
+      "grad_norm": 0.6815004193960651,
+      "learning_rate": 3.476198128228736e-07,
+      "loss": 0.302,
+      "step": 610
+    },
+    {
+      "epoch": 1.849413545213772,
+      "grad_norm": 0.7063903487339278,
+      "learning_rate": 3.339312090682689e-07,
+      "loss": 0.3003,
+      "step": 611
+    },
+    {
+      "epoch": 1.8524404086265607,
+      "grad_norm": 0.7073219383959698,
+      "learning_rate": 3.2051298603643754e-07,
+      "loss": 0.2988,
+      "step": 612
+    },
+    {
+      "epoch": 1.8554672720393492,
+      "grad_norm": 0.7249923323440401,
+      "learning_rate": 3.0736551906392354e-07,
+      "loss": 0.3054,
+      "step": 613
+    },
+    {
+      "epoch": 1.8584941354521378,
+      "grad_norm": 0.6844121153556538,
+      "learning_rate": 2.9448917591363923e-07,
+      "loss": 0.2871,
+      "step": 614
+    },
+    {
+      "epoch": 1.8615209988649262,
+      "grad_norm": 0.7580067785082366,
+      "learning_rate": 2.818843167645835e-07,
+      "loss": 0.3022,
+      "step": 615
+    },
+    {
+      "epoch": 1.8645478622777147,
+      "grad_norm": 0.6978591007092542,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.2904,
+      "step": 616
+    },
+    {
+      "epoch": 1.867574725690503,
+      "grad_norm": 0.6837773994195777,
+      "learning_rate": 2.5749045320632824e-07,
+      "loss": 0.2817,
+      "step": 617
+    },
+    {
+      "epoch": 1.8706015891032917,
+      "grad_norm": 0.6778843651927179,
+      "learning_rate": 2.4570213114592957e-07,
+      "loss": 0.2866,
+      "step": 618
+    },
+    {
+      "epoch": 1.8736284525160802,
+      "grad_norm": 0.6923387088970009,
+      "learning_rate": 2.3418665776527738e-07,
+      "loss": 0.2883,
+      "step": 619
+    },
+    {
+      "epoch": 1.8766553159288688,
+      "grad_norm": 0.7089706175428833,
+      "learning_rate": 2.2294435517691504e-07,
+      "loss": 0.2818,
+      "step": 620
+    },
+    {
+      "epoch": 1.8796821793416572,
+      "grad_norm": 0.6644536539357546,
+      "learning_rate": 2.119755378522137e-07,
+      "loss": 0.2837,
+      "step": 621
+    },
+    {
+      "epoch": 1.8827090427544457,
+      "grad_norm": 0.650857864844326,
+      "learning_rate": 2.0128051261257165e-07,
+      "loss": 0.2834,
+      "step": 622
+    },
+    {
+      "epoch": 1.885735906167234,
+      "grad_norm": 0.6759244102401449,
+      "learning_rate": 1.908595786208367e-07,
+      "loss": 0.2928,
+      "step": 623
+    },
+    {
+      "epoch": 1.8887627695800226,
+      "grad_norm": 0.6945468876304339,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.2927,
+      "step": 624
+    },
+    {
+      "epoch": 1.8917896329928112,
+      "grad_norm": 0.7451967396793971,
+      "learning_rate": 1.7084114268971275e-07,
+      "loss": 0.291,
+      "step": 625
+    },
+    {
+      "epoch": 1.8948164964055998,
+      "grad_norm": 0.668783430393399,
+      "learning_rate": 1.612442007090076e-07,
+      "loss": 0.2848,
+      "step": 626
+    },
+    {
+      "epoch": 1.8978433598183881,
+      "grad_norm": 0.6842657264088914,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.2997,
+      "step": 627
+    },
+    {
+      "epoch": 1.9008702232311767,
+      "grad_norm": 0.6702983212122103,
+      "learning_rate": 1.4287621094529524e-07,
+      "loss": 0.2868,
+      "step": 628
+    },
+    {
+      "epoch": 1.903897086643965,
+      "grad_norm": 0.7392644456548253,
+      "learning_rate": 1.3410567695444576e-07,
+      "loss": 0.3185,
+      "step": 629
+    },
+    {
+      "epoch": 1.9069239500567536,
+      "grad_norm": 0.6889954899593312,
+      "learning_rate": 1.2561111323605714e-07,
+      "loss": 0.2913,
+      "step": 630
+    },
+    {
+      "epoch": 1.9099508134695422,
+      "grad_norm": 0.7007697277913568,
+      "learning_rate": 1.1739275740134004e-07,
+      "loss": 0.2996,
+      "step": 631
+    },
+    {
+      "epoch": 1.9129776768823308,
+      "grad_norm": 0.7236788393431746,
+      "learning_rate": 1.0945083933537104e-07,
+      "loss": 0.2995,
+      "step": 632
+    },
+    {
+      "epoch": 1.9160045402951191,
+      "grad_norm": 0.6747758951171987,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.2728,
+      "step": 633
+    },
+    {
+      "epoch": 1.9190314037079077,
+      "grad_norm": 0.7077771949380657,
+      "learning_rate": 9.439719738099318e-08,
+      "loss": 0.2856,
+      "step": 634
+    },
+    {
+      "epoch": 1.922058267120696,
+      "grad_norm": 0.6816029057845483,
+      "learning_rate": 8.728589457530857e-08,
+      "loss": 0.282,
+      "step": 635
+    },
+    {
+      "epoch": 1.9250851305334846,
+      "grad_norm": 0.7184255804353783,
+      "learning_rate": 8.04518716920466e-08,
+      "loss": 0.2891,
+      "step": 636
+    },
+    {
+      "epoch": 1.9281119939462732,
+      "grad_norm": 0.6688615354435729,
+      "learning_rate": 7.389531989351773e-08,
+      "loss": 0.279,
+      "step": 637
+    },
+    {
+      "epoch": 1.9311388573590618,
+      "grad_norm": 0.7101628632784439,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.2948,
+      "step": 638
+    },
+    {
+      "epoch": 1.9341657207718501,
+      "grad_norm": 0.6804338668723345,
+      "learning_rate": 6.161535538745877e-08,
+      "loss": 0.2994,
+      "step": 639
+    },
+    {
+      "epoch": 1.9371925841846387,
+      "grad_norm": 0.7115154792090796,
+      "learning_rate": 5.5892286176932875e-08,
+      "loss": 0.2881,
+      "step": 640
+    },
+    {
+      "epoch": 1.940219447597427,
+      "grad_norm": 0.7047672989854594,
+      "learning_rate": 5.044737503554165e-08,
+      "loss": 0.293,
+      "step": 641
+    },
+    {
+      "epoch": 1.9432463110102156,
+      "grad_norm": 0.7016964857731067,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.3108,
+      "step": 642
+    },
+    {
+      "epoch": 1.9462731744230042,
+      "grad_norm": 0.7169015700233877,
+      "learning_rate": 4.0392628398699954e-08,
+      "loss": 0.3014,
+      "step": 643
+    },
+    {
+      "epoch": 1.9493000378357928,
+      "grad_norm": 0.639909663936784,
+      "learning_rate": 3.578307415612714e-08,
+      "loss": 0.2799,
+      "step": 644
+    },
+    {
+      "epoch": 1.952326901248581,
+      "grad_norm": 0.7006438001807862,
+      "learning_rate": 3.1452240480577265e-08,
+      "loss": 0.2864,
+      "step": 645
+    },
+    {
+      "epoch": 1.9553537646613697,
+      "grad_norm": 0.7256079310044938,
+      "learning_rate": 2.7400248514776184e-08,
+      "loss": 0.3034,
+      "step": 646
+    },
+    {
+      "epoch": 1.958380628074158,
+      "grad_norm": 0.7310594465635177,
+      "learning_rate": 2.3627211601651157e-08,
+      "loss": 0.3015,
+      "step": 647
+    },
+    {
+      "epoch": 1.9614074914869466,
+      "grad_norm": 0.7416711370266051,
+      "learning_rate": 2.013323528115674e-08,
+      "loss": 0.3062,
+      "step": 648
+    },
+    {
+      "epoch": 1.9644343548997352,
+      "grad_norm": 0.720502019257089,
+      "learning_rate": 1.6918417287318245e-08,
+      "loss": 0.2968,
+      "step": 649
+    },
+    {
+      "epoch": 1.9674612183125237,
+      "grad_norm": 0.6850885412184405,
+      "learning_rate": 1.3982847545507271e-08,
+      "loss": 0.2856,
+      "step": 650
+    },
+    {
+      "epoch": 1.970488081725312,
+      "grad_norm": 0.714073427722188,
+      "learning_rate": 1.1326608169920373e-08,
+      "loss": 0.2886,
+      "step": 651
+    },
+    {
+      "epoch": 1.9735149451381007,
+      "grad_norm": 0.7044775938150841,
+      "learning_rate": 8.949773461282008e-09,
+      "loss": 0.2963,
+      "step": 652
+    },
+    {
+      "epoch": 1.976541808550889,
+      "grad_norm": 0.6763258276454628,
+      "learning_rate": 6.8524099047695415e-09,
+      "loss": 0.3003,
+      "step": 653
+    },
+    {
+      "epoch": 1.9795686719636776,
+      "grad_norm": 0.6710834842238353,
+      "learning_rate": 5.034576168149175e-09,
+      "loss": 0.2882,
+      "step": 654
+    },
+    {
+      "epoch": 1.9825955353764662,
+      "grad_norm": 0.6708507303155626,
+      "learning_rate": 3.4963231001383657e-09,
+      "loss": 0.2844,
+      "step": 655
+    },
+    {
+      "epoch": 1.9856223987892547,
+      "grad_norm": 0.655868400304347,
+      "learning_rate": 2.237693728981416e-09,
+      "loss": 0.2832,
+      "step": 656
+    },
+    {
+      "epoch": 1.988649262202043,
+      "grad_norm": 0.6947030072617233,
+      "learning_rate": 1.2587232612493172e-09,
+      "loss": 0.2865,
+      "step": 657
+    },
+    {
+      "epoch": 1.9916761256148316,
+      "grad_norm": 0.6987494144831641,
+      "learning_rate": 5.594390808494332e-10,
+      "loss": 0.3034,
+      "step": 658
+    },
+    {
+      "epoch": 1.99470298902762,
+      "grad_norm": 0.6670751771753797,
+      "learning_rate": 1.3986074826388697e-10,
+      "loss": 0.2795,
+      "step": 659
+    },
+    {
+      "epoch": 1.9977298524404086,
+      "grad_norm": 0.7013681670676795,
+      "learning_rate": 0.0,
+      "loss": 0.304,
+      "step": 660
+    },
+    {
+      "epoch": 1.9977298524404086,
+      "step": 660,
+      "total_flos": 73127279247360.0,
+      "train_loss": 0.38489366110527157,
+      "train_runtime": 2379.8317,
+      "train_samples_per_second": 35.527,
+      "train_steps_per_second": 0.277
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 660,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "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": 73127279247360.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}