diff --git "a/run-0/checkpoint-620/trainer_state.json" "b/run-0/checkpoint-620/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/run-0/checkpoint-620/trainer_state.json"
@@ -0,0 +1,4440 @@
+{
+  "best_metric": 1.6532543369745685,
+  "best_model_checkpoint": "./modernBERT-content-regression/run-0/checkpoint-620",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 620,
+  "is_hyper_param_search": true,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008064516129032258,
+      "grad_norm": 312.8164978027344,
+      "learning_rate": 1.9405916792639294e-05,
+      "loss": 21.3087,
+      "step": 1
+    },
+    {
+      "epoch": 0.016129032258064516,
+      "grad_norm": 21.297748565673828,
+      "learning_rate": 1.9374566361633415e-05,
+      "loss": 0.3159,
+      "step": 2
+    },
+    {
+      "epoch": 0.024193548387096774,
+      "grad_norm": 378.6422424316406,
+      "learning_rate": 1.9343215930627533e-05,
+      "loss": 138.0821,
+      "step": 3
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 98.11734008789062,
+      "learning_rate": 1.9311865499621654e-05,
+      "loss": 13.477,
+      "step": 4
+    },
+    {
+      "epoch": 0.04032258064516129,
+      "grad_norm": 55.27867126464844,
+      "learning_rate": 1.9280515068615775e-05,
+      "loss": 1.8628,
+      "step": 5
+    },
+    {
+      "epoch": 0.04838709677419355,
+      "grad_norm": 116.12382507324219,
+      "learning_rate": 1.92491646376099e-05,
+      "loss": 2.5156,
+      "step": 6
+    },
+    {
+      "epoch": 0.056451612903225805,
+      "grad_norm": 44.64065170288086,
+      "learning_rate": 1.921781420660402e-05,
+      "loss": 4.839,
+      "step": 7
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 444.887451171875,
+      "learning_rate": 1.9186463775598138e-05,
+      "loss": 10.6158,
+      "step": 8
+    },
+    {
+      "epoch": 0.07258064516129033,
+      "grad_norm": 29.839136123657227,
+      "learning_rate": 1.915511334459226e-05,
+      "loss": 0.2628,
+      "step": 9
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 233.91651916503906,
+      "learning_rate": 1.912376291358638e-05,
+      "loss": 2.1308,
+      "step": 10
+    },
+    {
+      "epoch": 0.08870967741935484,
+      "grad_norm": 169.41172790527344,
+      "learning_rate": 1.90924124825805e-05,
+      "loss": 10.3841,
+      "step": 11
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 25.190792083740234,
+      "learning_rate": 1.906106205157462e-05,
+      "loss": 0.3598,
+      "step": 12
+    },
+    {
+      "epoch": 0.10483870967741936,
+      "grad_norm": 112.92284393310547,
+      "learning_rate": 1.902971162056874e-05,
+      "loss": 1.3362,
+      "step": 13
+    },
+    {
+      "epoch": 0.11290322580645161,
+      "grad_norm": 29.17025375366211,
+      "learning_rate": 1.899836118956286e-05,
+      "loss": 1.1826,
+      "step": 14
+    },
+    {
+      "epoch": 0.12096774193548387,
+      "grad_norm": 53.96384048461914,
+      "learning_rate": 1.8967010758556982e-05,
+      "loss": 0.3421,
+      "step": 15
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 201.16738891601562,
+      "learning_rate": 1.8935660327551106e-05,
+      "loss": 9.4231,
+      "step": 16
+    },
+    {
+      "epoch": 0.13709677419354838,
+      "grad_norm": 52.72154235839844,
+      "learning_rate": 1.8904309896545224e-05,
+      "loss": 1.9206,
+      "step": 17
+    },
+    {
+      "epoch": 0.14516129032258066,
+      "grad_norm": 127.67330932617188,
+      "learning_rate": 1.8872959465539345e-05,
+      "loss": 2.6729,
+      "step": 18
+    },
+    {
+      "epoch": 0.1532258064516129,
+      "grad_norm": 36.01234436035156,
+      "learning_rate": 1.8841609034533466e-05,
+      "loss": 1.044,
+      "step": 19
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 291.8822021484375,
+      "learning_rate": 1.8810258603527587e-05,
+      "loss": 9.1818,
+      "step": 20
+    },
+    {
+      "epoch": 0.1693548387096774,
+      "grad_norm": 270.27142333984375,
+      "learning_rate": 1.8778908172521708e-05,
+      "loss": 3.1568,
+      "step": 21
+    },
+    {
+      "epoch": 0.1774193548387097,
+      "grad_norm": 527.8693237304688,
+      "learning_rate": 1.8747557741515826e-05,
+      "loss": 7.3642,
+      "step": 22
+    },
+    {
+      "epoch": 0.18548387096774194,
+      "grad_norm": 135.9112548828125,
+      "learning_rate": 1.8716207310509947e-05,
+      "loss": 15.6669,
+      "step": 23
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 490.7304382324219,
+      "learning_rate": 1.8684856879504068e-05,
+      "loss": 6.4283,
+      "step": 24
+    },
+    {
+      "epoch": 0.20161290322580644,
+      "grad_norm": 159.22994995117188,
+      "learning_rate": 1.865350644849819e-05,
+      "loss": 1.1921,
+      "step": 25
+    },
+    {
+      "epoch": 0.20967741935483872,
+      "grad_norm": 183.89662170410156,
+      "learning_rate": 1.862215601749231e-05,
+      "loss": 1.6719,
+      "step": 26
+    },
+    {
+      "epoch": 0.21774193548387097,
+      "grad_norm": 89.63036346435547,
+      "learning_rate": 1.859080558648643e-05,
+      "loss": 2.104,
+      "step": 27
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 108.4155044555664,
+      "learning_rate": 1.8559455155480552e-05,
+      "loss": 1.345,
+      "step": 28
+    },
+    {
+      "epoch": 0.23387096774193547,
+      "grad_norm": 25.60169792175293,
+      "learning_rate": 1.8528104724474673e-05,
+      "loss": 0.6057,
+      "step": 29
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 64.74250793457031,
+      "learning_rate": 1.8496754293468794e-05,
+      "loss": 0.5283,
+      "step": 30
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 48.2547721862793,
+      "learning_rate": 1.8465403862462912e-05,
+      "loss": 1.0487,
+      "step": 31
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 85.43180847167969,
+      "learning_rate": 1.8434053431457033e-05,
+      "loss": 1.1554,
+      "step": 32
+    },
+    {
+      "epoch": 0.2661290322580645,
+      "grad_norm": 23.75019645690918,
+      "learning_rate": 1.8402703000451154e-05,
+      "loss": 0.6364,
+      "step": 33
+    },
+    {
+      "epoch": 0.27419354838709675,
+      "grad_norm": 60.279544830322266,
+      "learning_rate": 1.8371352569445275e-05,
+      "loss": 1.4024,
+      "step": 34
+    },
+    {
+      "epoch": 0.28225806451612906,
+      "grad_norm": 256.9117126464844,
+      "learning_rate": 1.8340002138439396e-05,
+      "loss": 21.5309,
+      "step": 35
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 83.0962142944336,
+      "learning_rate": 1.8308651707433517e-05,
+      "loss": 1.4266,
+      "step": 36
+    },
+    {
+      "epoch": 0.29838709677419356,
+      "grad_norm": 31.391490936279297,
+      "learning_rate": 1.827730127642764e-05,
+      "loss": 0.1593,
+      "step": 37
+    },
+    {
+      "epoch": 0.3064516129032258,
+      "grad_norm": 196.5593719482422,
+      "learning_rate": 1.824595084542176e-05,
+      "loss": 18.6657,
+      "step": 38
+    },
+    {
+      "epoch": 0.31451612903225806,
+      "grad_norm": 526.9293823242188,
+      "learning_rate": 1.821460041441588e-05,
+      "loss": 2.932,
+      "step": 39
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 398.2054138183594,
+      "learning_rate": 1.8183249983409998e-05,
+      "loss": 17.9616,
+      "step": 40
+    },
+    {
+      "epoch": 0.33064516129032256,
+      "grad_norm": 720.282958984375,
+      "learning_rate": 1.815189955240412e-05,
+      "loss": 13.228,
+      "step": 41
+    },
+    {
+      "epoch": 0.3387096774193548,
+      "grad_norm": 163.12245178222656,
+      "learning_rate": 1.812054912139824e-05,
+      "loss": 19.4242,
+      "step": 42
+    },
+    {
+      "epoch": 0.3467741935483871,
+      "grad_norm": 355.48846435546875,
+      "learning_rate": 1.808919869039236e-05,
+      "loss": 26.5729,
+      "step": 43
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 241.5015411376953,
+      "learning_rate": 1.8057848259386482e-05,
+      "loss": 3.7127,
+      "step": 44
+    },
+    {
+      "epoch": 0.3629032258064516,
+      "grad_norm": 185.13877868652344,
+      "learning_rate": 1.8026497828380604e-05,
+      "loss": 3.4602,
+      "step": 45
+    },
+    {
+      "epoch": 0.3709677419354839,
+      "grad_norm": 63.099327087402344,
+      "learning_rate": 1.7995147397374725e-05,
+      "loss": 2.2782,
+      "step": 46
+    },
+    {
+      "epoch": 0.3790322580645161,
+      "grad_norm": 52.98133087158203,
+      "learning_rate": 1.7963796966368846e-05,
+      "loss": 2.8164,
+      "step": 47
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 72.33665466308594,
+      "learning_rate": 1.7932446535362967e-05,
+      "loss": 1.1977,
+      "step": 48
+    },
+    {
+      "epoch": 0.3951612903225806,
+      "grad_norm": 79.54340362548828,
+      "learning_rate": 1.7901096104357088e-05,
+      "loss": 2.3691,
+      "step": 49
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 112.45056915283203,
+      "learning_rate": 1.7869745673351205e-05,
+      "loss": 10.6579,
+      "step": 50
+    },
+    {
+      "epoch": 0.4112903225806452,
+      "grad_norm": 9.006159782409668,
+      "learning_rate": 1.7838395242345327e-05,
+      "loss": 0.1436,
+      "step": 51
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 62.99940490722656,
+      "learning_rate": 1.7807044811339448e-05,
+      "loss": 1.9241,
+      "step": 52
+    },
+    {
+      "epoch": 0.4274193548387097,
+      "grad_norm": 205.84600830078125,
+      "learning_rate": 1.777569438033357e-05,
+      "loss": 21.9755,
+      "step": 53
+    },
+    {
+      "epoch": 0.43548387096774194,
+      "grad_norm": 18.60399055480957,
+      "learning_rate": 1.774434394932769e-05,
+      "loss": 0.6385,
+      "step": 54
+    },
+    {
+      "epoch": 0.4435483870967742,
+      "grad_norm": 60.74576187133789,
+      "learning_rate": 1.771299351832181e-05,
+      "loss": 6.0776,
+      "step": 55
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 55.38773727416992,
+      "learning_rate": 1.7681643087315932e-05,
+      "loss": 1.713,
+      "step": 56
+    },
+    {
+      "epoch": 0.4596774193548387,
+      "grad_norm": 22.658430099487305,
+      "learning_rate": 1.7650292656310053e-05,
+      "loss": 0.2931,
+      "step": 57
+    },
+    {
+      "epoch": 0.46774193548387094,
+      "grad_norm": 103.3204574584961,
+      "learning_rate": 1.7618942225304174e-05,
+      "loss": 9.7618,
+      "step": 58
+    },
+    {
+      "epoch": 0.47580645161290325,
+      "grad_norm": 22.80617332458496,
+      "learning_rate": 1.758759179429829e-05,
+      "loss": 0.3063,
+      "step": 59
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 126.92974090576172,
+      "learning_rate": 1.7556241363292413e-05,
+      "loss": 14.4092,
+      "step": 60
+    },
+    {
+      "epoch": 0.49193548387096775,
+      "grad_norm": 43.025787353515625,
+      "learning_rate": 1.7524890932286534e-05,
+      "loss": 0.7731,
+      "step": 61
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 80.46817779541016,
+      "learning_rate": 1.7493540501280655e-05,
+      "loss": 21.7639,
+      "step": 62
+    },
+    {
+      "epoch": 0.5080645161290323,
+      "grad_norm": 59.49717712402344,
+      "learning_rate": 1.7462190070274776e-05,
+      "loss": 1.618,
+      "step": 63
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 15.44653606414795,
+      "learning_rate": 1.7430839639268897e-05,
+      "loss": 2.4747,
+      "step": 64
+    },
+    {
+      "epoch": 0.5241935483870968,
+      "grad_norm": 89.72061920166016,
+      "learning_rate": 1.7399489208263018e-05,
+      "loss": 13.3349,
+      "step": 65
+    },
+    {
+      "epoch": 0.532258064516129,
+      "grad_norm": 16.540616989135742,
+      "learning_rate": 1.736813877725714e-05,
+      "loss": 1.4963,
+      "step": 66
+    },
+    {
+      "epoch": 0.5403225806451613,
+      "grad_norm": 50.684844970703125,
+      "learning_rate": 1.733678834625126e-05,
+      "loss": 2.0989,
+      "step": 67
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 56.257877349853516,
+      "learning_rate": 1.730543791524538e-05,
+      "loss": 1.6343,
+      "step": 68
+    },
+    {
+      "epoch": 0.5564516129032258,
+      "grad_norm": 87.75580596923828,
+      "learning_rate": 1.72740874842395e-05,
+      "loss": 3.4088,
+      "step": 69
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 14.779145240783691,
+      "learning_rate": 1.724273705323362e-05,
+      "loss": 0.4494,
+      "step": 70
+    },
+    {
+      "epoch": 0.5725806451612904,
+      "grad_norm": 20.206077575683594,
+      "learning_rate": 1.721138662222774e-05,
+      "loss": 2.2875,
+      "step": 71
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 81.59163665771484,
+      "learning_rate": 1.7180036191221862e-05,
+      "loss": 16.2722,
+      "step": 72
+    },
+    {
+      "epoch": 0.5887096774193549,
+      "grad_norm": 20.435260772705078,
+      "learning_rate": 1.7148685760215983e-05,
+      "loss": 0.2519,
+      "step": 73
+    },
+    {
+      "epoch": 0.5967741935483871,
+      "grad_norm": 77.98759460449219,
+      "learning_rate": 1.7117335329210104e-05,
+      "loss": 2.4472,
+      "step": 74
+    },
+    {
+      "epoch": 0.6048387096774194,
+      "grad_norm": 75.45653533935547,
+      "learning_rate": 1.7085984898204225e-05,
+      "loss": 9.1756,
+      "step": 75
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 16.75188636779785,
+      "learning_rate": 1.7054634467198346e-05,
+      "loss": 0.25,
+      "step": 76
+    },
+    {
+      "epoch": 0.6209677419354839,
+      "grad_norm": 21.303180694580078,
+      "learning_rate": 1.7023284036192467e-05,
+      "loss": 0.3106,
+      "step": 77
+    },
+    {
+      "epoch": 0.6290322580645161,
+      "grad_norm": 56.08082962036133,
+      "learning_rate": 1.6991933605186585e-05,
+      "loss": 13.2157,
+      "step": 78
+    },
+    {
+      "epoch": 0.6370967741935484,
+      "grad_norm": 115.4497299194336,
+      "learning_rate": 1.6960583174180706e-05,
+      "loss": 24.3069,
+      "step": 79
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 18.160301208496094,
+      "learning_rate": 1.6929232743174827e-05,
+      "loss": 1.2587,
+      "step": 80
+    },
+    {
+      "epoch": 0.6532258064516129,
+      "grad_norm": 15.181955337524414,
+      "learning_rate": 1.6897882312168948e-05,
+      "loss": 0.317,
+      "step": 81
+    },
+    {
+      "epoch": 0.6612903225806451,
+      "grad_norm": 50.316402435302734,
+      "learning_rate": 1.686653188116307e-05,
+      "loss": 2.1468,
+      "step": 82
+    },
+    {
+      "epoch": 0.6693548387096774,
+      "grad_norm": 24.807689666748047,
+      "learning_rate": 1.683518145015719e-05,
+      "loss": 0.4887,
+      "step": 83
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 7.860809803009033,
+      "learning_rate": 1.680383101915131e-05,
+      "loss": 0.7155,
+      "step": 84
+    },
+    {
+      "epoch": 0.6854838709677419,
+      "grad_norm": 152.7879180908203,
+      "learning_rate": 1.6772480588145432e-05,
+      "loss": 18.5123,
+      "step": 85
+    },
+    {
+      "epoch": 0.6935483870967742,
+      "grad_norm": 46.42924499511719,
+      "learning_rate": 1.6741130157139553e-05,
+      "loss": 14.8221,
+      "step": 86
+    },
+    {
+      "epoch": 0.7016129032258065,
+      "grad_norm": 55.30247497558594,
+      "learning_rate": 1.670977972613367e-05,
+      "loss": 1.3708,
+      "step": 87
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 64.38847351074219,
+      "learning_rate": 1.6678429295127792e-05,
+      "loss": 1.5522,
+      "step": 88
+    },
+    {
+      "epoch": 0.717741935483871,
+      "grad_norm": 71.2874755859375,
+      "learning_rate": 1.6647078864121913e-05,
+      "loss": 2.576,
+      "step": 89
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 81.83346557617188,
+      "learning_rate": 1.6615728433116034e-05,
+      "loss": 2.3827,
+      "step": 90
+    },
+    {
+      "epoch": 0.7338709677419355,
+      "grad_norm": 133.49618530273438,
+      "learning_rate": 1.6584378002110155e-05,
+      "loss": 18.9283,
+      "step": 91
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 58.27260971069336,
+      "learning_rate": 1.6553027571104276e-05,
+      "loss": 5.9736,
+      "step": 92
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.685044288635254,
+      "learning_rate": 1.6521677140098398e-05,
+      "loss": 0.1484,
+      "step": 93
+    },
+    {
+      "epoch": 0.7580645161290323,
+      "grad_norm": 100.5199966430664,
+      "learning_rate": 1.649032670909252e-05,
+      "loss": 4.3167,
+      "step": 94
+    },
+    {
+      "epoch": 0.7661290322580645,
+      "grad_norm": 78.9014663696289,
+      "learning_rate": 1.645897627808664e-05,
+      "loss": 9.7868,
+      "step": 95
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 90.67410278320312,
+      "learning_rate": 1.642762584708076e-05,
+      "loss": 3.1579,
+      "step": 96
+    },
+    {
+      "epoch": 0.782258064516129,
+      "grad_norm": 64.82438659667969,
+      "learning_rate": 1.639627541607488e-05,
+      "loss": 1.3801,
+      "step": 97
+    },
+    {
+      "epoch": 0.7903225806451613,
+      "grad_norm": 54.49703598022461,
+      "learning_rate": 1.6364924985069e-05,
+      "loss": 1.1181,
+      "step": 98
+    },
+    {
+      "epoch": 0.7983870967741935,
+      "grad_norm": 48.997737884521484,
+      "learning_rate": 1.633357455406312e-05,
+      "loss": 0.8717,
+      "step": 99
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 163.54615783691406,
+      "learning_rate": 1.630222412305724e-05,
+      "loss": 20.7014,
+      "step": 100
+    },
+    {
+      "epoch": 0.8145161290322581,
+      "grad_norm": 40.45456314086914,
+      "learning_rate": 1.6270873692051363e-05,
+      "loss": 1.3114,
+      "step": 101
+    },
+    {
+      "epoch": 0.8225806451612904,
+      "grad_norm": 10.254913330078125,
+      "learning_rate": 1.6239523261045484e-05,
+      "loss": 0.2994,
+      "step": 102
+    },
+    {
+      "epoch": 0.8306451612903226,
+      "grad_norm": 97.43340301513672,
+      "learning_rate": 1.6208172830039605e-05,
+      "loss": 13.0826,
+      "step": 103
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 26.264678955078125,
+      "learning_rate": 1.6176822399033726e-05,
+      "loss": 0.8864,
+      "step": 104
+    },
+    {
+      "epoch": 0.8467741935483871,
+      "grad_norm": 52.89865493774414,
+      "learning_rate": 1.6145471968027847e-05,
+      "loss": 0.9729,
+      "step": 105
+    },
+    {
+      "epoch": 0.8548387096774194,
+      "grad_norm": 57.24679183959961,
+      "learning_rate": 1.6114121537021965e-05,
+      "loss": 2.4299,
+      "step": 106
+    },
+    {
+      "epoch": 0.8629032258064516,
+      "grad_norm": 50.82107925415039,
+      "learning_rate": 1.6082771106016086e-05,
+      "loss": 1.7577,
+      "step": 107
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 71.07525634765625,
+      "learning_rate": 1.6051420675010207e-05,
+      "loss": 1.3853,
+      "step": 108
+    },
+    {
+      "epoch": 0.8790322580645161,
+      "grad_norm": 102.36954498291016,
+      "learning_rate": 1.6020070244004328e-05,
+      "loss": 3.1053,
+      "step": 109
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 36.77400207519531,
+      "learning_rate": 1.598871981299845e-05,
+      "loss": 1.1706,
+      "step": 110
+    },
+    {
+      "epoch": 0.8951612903225806,
+      "grad_norm": 75.524658203125,
+      "learning_rate": 1.595736938199257e-05,
+      "loss": 2.8564,
+      "step": 111
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 22.996549606323242,
+      "learning_rate": 1.592601895098669e-05,
+      "loss": 1.1411,
+      "step": 112
+    },
+    {
+      "epoch": 0.9112903225806451,
+      "grad_norm": 56.16508102416992,
+      "learning_rate": 1.5894668519980812e-05,
+      "loss": 1.1069,
+      "step": 113
+    },
+    {
+      "epoch": 0.9193548387096774,
+      "grad_norm": 45.19282913208008,
+      "learning_rate": 1.5863318088974933e-05,
+      "loss": 1.7093,
+      "step": 114
+    },
+    {
+      "epoch": 0.9274193548387096,
+      "grad_norm": 66.1149673461914,
+      "learning_rate": 1.5831967657969054e-05,
+      "loss": 4.0799,
+      "step": 115
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 48.84867477416992,
+      "learning_rate": 1.5800617226963172e-05,
+      "loss": 2.2224,
+      "step": 116
+    },
+    {
+      "epoch": 0.9435483870967742,
+      "grad_norm": 59.682125091552734,
+      "learning_rate": 1.5769266795957293e-05,
+      "loss": 0.9111,
+      "step": 117
+    },
+    {
+      "epoch": 0.9516129032258065,
+      "grad_norm": 76.44542694091797,
+      "learning_rate": 1.5737916364951414e-05,
+      "loss": 14.3846,
+      "step": 118
+    },
+    {
+      "epoch": 0.9596774193548387,
+      "grad_norm": 28.405059814453125,
+      "learning_rate": 1.5706565933945535e-05,
+      "loss": 1.9018,
+      "step": 119
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 10.535357475280762,
+      "learning_rate": 1.5675215502939656e-05,
+      "loss": 0.9783,
+      "step": 120
+    },
+    {
+      "epoch": 0.9758064516129032,
+      "grad_norm": 21.8922176361084,
+      "learning_rate": 1.5643865071933777e-05,
+      "loss": 0.3224,
+      "step": 121
+    },
+    {
+      "epoch": 0.9838709677419355,
+      "grad_norm": 15.08030891418457,
+      "learning_rate": 1.5612514640927898e-05,
+      "loss": 0.4911,
+      "step": 122
+    },
+    {
+      "epoch": 0.9919354838709677,
+      "grad_norm": 3.00480580329895,
+      "learning_rate": 1.558116420992202e-05,
+      "loss": 0.1242,
+      "step": 123
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 26.523765563964844,
+      "learning_rate": 1.554981377891614e-05,
+      "loss": 0.238,
+      "step": 124
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 4.573007583618164,
+      "eval_mae": 1.3245397806167603,
+      "eval_mse": 4.573008060455322,
+      "eval_r2": -0.2520101070404053,
+      "eval_rmse": 2.138459272573439,
+      "eval_runtime": 1.736,
+      "eval_samples_per_second": 31.682,
+      "eval_smape": 54.24200892448425,
+      "eval_steps_per_second": 8.064,
+      "step": 124
+    },
+    {
+      "epoch": 1.0080645161290323,
+      "grad_norm": 23.402734756469727,
+      "learning_rate": 1.5518463347910258e-05,
+      "loss": 0.8834,
+      "step": 125
+    },
+    {
+      "epoch": 1.0161290322580645,
+      "grad_norm": 40.177024841308594,
+      "learning_rate": 1.548711291690438e-05,
+      "loss": 1.2453,
+      "step": 126
+    },
+    {
+      "epoch": 1.0241935483870968,
+      "grad_norm": 86.14007568359375,
+      "learning_rate": 1.54557624858985e-05,
+      "loss": 5.4819,
+      "step": 127
+    },
+    {
+      "epoch": 1.032258064516129,
+      "grad_norm": 10.090241432189941,
+      "learning_rate": 1.542441205489262e-05,
+      "loss": 0.357,
+      "step": 128
+    },
+    {
+      "epoch": 1.0403225806451613,
+      "grad_norm": 29.933624267578125,
+      "learning_rate": 1.5393061623886742e-05,
+      "loss": 1.0975,
+      "step": 129
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 20.847654342651367,
+      "learning_rate": 1.5361711192880863e-05,
+      "loss": 2.23,
+      "step": 130
+    },
+    {
+      "epoch": 1.0564516129032258,
+      "grad_norm": 41.63291549682617,
+      "learning_rate": 1.5330360761874984e-05,
+      "loss": 0.6496,
+      "step": 131
+    },
+    {
+      "epoch": 1.064516129032258,
+      "grad_norm": 49.841773986816406,
+      "learning_rate": 1.5299010330869105e-05,
+      "loss": 3.59,
+      "step": 132
+    },
+    {
+      "epoch": 1.0725806451612903,
+      "grad_norm": 55.52513122558594,
+      "learning_rate": 1.5267659899863226e-05,
+      "loss": 1.1932,
+      "step": 133
+    },
+    {
+      "epoch": 1.0806451612903225,
+      "grad_norm": 46.58870315551758,
+      "learning_rate": 1.5236309468857344e-05,
+      "loss": 0.8727,
+      "step": 134
+    },
+    {
+      "epoch": 1.0887096774193548,
+      "grad_norm": 52.691673278808594,
+      "learning_rate": 1.5204959037851465e-05,
+      "loss": 5.618,
+      "step": 135
+    },
+    {
+      "epoch": 1.096774193548387,
+      "grad_norm": 94.9619140625,
+      "learning_rate": 1.5173608606845586e-05,
+      "loss": 13.7161,
+      "step": 136
+    },
+    {
+      "epoch": 1.1048387096774193,
+      "grad_norm": 105.40327453613281,
+      "learning_rate": 1.5142258175839707e-05,
+      "loss": 7.7381,
+      "step": 137
+    },
+    {
+      "epoch": 1.1129032258064515,
+      "grad_norm": 90.10973358154297,
+      "learning_rate": 1.5110907744833828e-05,
+      "loss": 1.8823,
+      "step": 138
+    },
+    {
+      "epoch": 1.120967741935484,
+      "grad_norm": 220.77796936035156,
+      "learning_rate": 1.5079557313827948e-05,
+      "loss": 35.8495,
+      "step": 139
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 122.45404815673828,
+      "learning_rate": 1.5048206882822069e-05,
+      "loss": 2.8625,
+      "step": 140
+    },
+    {
+      "epoch": 1.1370967741935485,
+      "grad_norm": 87.10842895507812,
+      "learning_rate": 1.501685645181619e-05,
+      "loss": 2.4731,
+      "step": 141
+    },
+    {
+      "epoch": 1.1451612903225807,
+      "grad_norm": 35.40389633178711,
+      "learning_rate": 1.498550602081031e-05,
+      "loss": 0.7125,
+      "step": 142
+    },
+    {
+      "epoch": 1.153225806451613,
+      "grad_norm": 45.49386215209961,
+      "learning_rate": 1.4954155589804432e-05,
+      "loss": 0.7262,
+      "step": 143
+    },
+    {
+      "epoch": 1.1612903225806452,
+      "grad_norm": 52.729427337646484,
+      "learning_rate": 1.4922805158798551e-05,
+      "loss": 0.9633,
+      "step": 144
+    },
+    {
+      "epoch": 1.1693548387096775,
+      "grad_norm": 40.62730407714844,
+      "learning_rate": 1.4891454727792672e-05,
+      "loss": 0.619,
+      "step": 145
+    },
+    {
+      "epoch": 1.1774193548387097,
+      "grad_norm": 63.05773162841797,
+      "learning_rate": 1.4860104296786793e-05,
+      "loss": 1.575,
+      "step": 146
+    },
+    {
+      "epoch": 1.185483870967742,
+      "grad_norm": 22.308488845825195,
+      "learning_rate": 1.4828753865780914e-05,
+      "loss": 0.4164,
+      "step": 147
+    },
+    {
+      "epoch": 1.1935483870967742,
+      "grad_norm": 20.031757354736328,
+      "learning_rate": 1.4797403434775034e-05,
+      "loss": 0.6238,
+      "step": 148
+    },
+    {
+      "epoch": 1.2016129032258065,
+      "grad_norm": 165.66102600097656,
+      "learning_rate": 1.4766053003769155e-05,
+      "loss": 11.1981,
+      "step": 149
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 184.57411193847656,
+      "learning_rate": 1.4734702572763276e-05,
+      "loss": 35.5336,
+      "step": 150
+    },
+    {
+      "epoch": 1.217741935483871,
+      "grad_norm": 97.44104766845703,
+      "learning_rate": 1.4703352141757397e-05,
+      "loss": 16.5436,
+      "step": 151
+    },
+    {
+      "epoch": 1.2258064516129032,
+      "grad_norm": 55.624298095703125,
+      "learning_rate": 1.4672001710751518e-05,
+      "loss": 1.5125,
+      "step": 152
+    },
+    {
+      "epoch": 1.2338709677419355,
+      "grad_norm": 69.92327117919922,
+      "learning_rate": 1.4640651279745637e-05,
+      "loss": 9.3334,
+      "step": 153
+    },
+    {
+      "epoch": 1.2419354838709677,
+      "grad_norm": 85.40164184570312,
+      "learning_rate": 1.4609300848739758e-05,
+      "loss": 11.2524,
+      "step": 154
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 39.81537628173828,
+      "learning_rate": 1.457795041773388e-05,
+      "loss": 0.6493,
+      "step": 155
+    },
+    {
+      "epoch": 1.2580645161290323,
+      "grad_norm": 47.04484939575195,
+      "learning_rate": 1.4546599986728e-05,
+      "loss": 1.4688,
+      "step": 156
+    },
+    {
+      "epoch": 1.2661290322580645,
+      "grad_norm": 30.717687606811523,
+      "learning_rate": 1.4515249555722122e-05,
+      "loss": 2.073,
+      "step": 157
+    },
+    {
+      "epoch": 1.2741935483870968,
+      "grad_norm": 84.80261993408203,
+      "learning_rate": 1.4483899124716241e-05,
+      "loss": 11.3152,
+      "step": 158
+    },
+    {
+      "epoch": 1.282258064516129,
+      "grad_norm": 339.03363037109375,
+      "learning_rate": 1.4452548693710362e-05,
+      "loss": 19.9553,
+      "step": 159
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 48.307281494140625,
+      "learning_rate": 1.4421198262704483e-05,
+      "loss": 8.5167,
+      "step": 160
+    },
+    {
+      "epoch": 1.2983870967741935,
+      "grad_norm": 253.1395263671875,
+      "learning_rate": 1.4389847831698604e-05,
+      "loss": 22.7371,
+      "step": 161
+    },
+    {
+      "epoch": 1.3064516129032258,
+      "grad_norm": 120.206787109375,
+      "learning_rate": 1.4358497400692725e-05,
+      "loss": 3.3325,
+      "step": 162
+    },
+    {
+      "epoch": 1.314516129032258,
+      "grad_norm": 134.07962036132812,
+      "learning_rate": 1.4327146969686845e-05,
+      "loss": 2.6457,
+      "step": 163
+    },
+    {
+      "epoch": 1.3225806451612903,
+      "grad_norm": 85.90950775146484,
+      "learning_rate": 1.4295796538680966e-05,
+      "loss": 4.7217,
+      "step": 164
+    },
+    {
+      "epoch": 1.3306451612903225,
+      "grad_norm": 349.7077331542969,
+      "learning_rate": 1.4264446107675087e-05,
+      "loss": 9.5306,
+      "step": 165
+    },
+    {
+      "epoch": 1.3387096774193548,
+      "grad_norm": 175.317138671875,
+      "learning_rate": 1.4233095676669208e-05,
+      "loss": 4.5035,
+      "step": 166
+    },
+    {
+      "epoch": 1.346774193548387,
+      "grad_norm": 59.4417839050293,
+      "learning_rate": 1.4201745245663327e-05,
+      "loss": 0.9392,
+      "step": 167
+    },
+    {
+      "epoch": 1.3548387096774195,
+      "grad_norm": 51.75499725341797,
+      "learning_rate": 1.4170394814657448e-05,
+      "loss": 0.7847,
+      "step": 168
+    },
+    {
+      "epoch": 1.3629032258064515,
+      "grad_norm": 98.38888549804688,
+      "learning_rate": 1.413904438365157e-05,
+      "loss": 2.4999,
+      "step": 169
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 430.9798278808594,
+      "learning_rate": 1.410769395264569e-05,
+      "loss": 124.0953,
+      "step": 170
+    },
+    {
+      "epoch": 1.379032258064516,
+      "grad_norm": 24.690670013427734,
+      "learning_rate": 1.4076343521639811e-05,
+      "loss": 0.8053,
+      "step": 171
+    },
+    {
+      "epoch": 1.3870967741935485,
+      "grad_norm": 198.97019958496094,
+      "learning_rate": 1.404499309063393e-05,
+      "loss": 12.995,
+      "step": 172
+    },
+    {
+      "epoch": 1.3951612903225805,
+      "grad_norm": 13.825026512145996,
+      "learning_rate": 1.4013642659628052e-05,
+      "loss": 0.4482,
+      "step": 173
+    },
+    {
+      "epoch": 1.403225806451613,
+      "grad_norm": 94.05176544189453,
+      "learning_rate": 1.3982292228622173e-05,
+      "loss": 8.6515,
+      "step": 174
+    },
+    {
+      "epoch": 1.4112903225806452,
+      "grad_norm": 64.00646209716797,
+      "learning_rate": 1.3950941797616294e-05,
+      "loss": 1.366,
+      "step": 175
+    },
+    {
+      "epoch": 1.4193548387096775,
+      "grad_norm": 259.6784973144531,
+      "learning_rate": 1.3919591366610415e-05,
+      "loss": 20.5985,
+      "step": 176
+    },
+    {
+      "epoch": 1.4274193548387097,
+      "grad_norm": 87.15518951416016,
+      "learning_rate": 1.3888240935604534e-05,
+      "loss": 2.9658,
+      "step": 177
+    },
+    {
+      "epoch": 1.435483870967742,
+      "grad_norm": 29.206682205200195,
+      "learning_rate": 1.3856890504598655e-05,
+      "loss": 0.4756,
+      "step": 178
+    },
+    {
+      "epoch": 1.4435483870967742,
+      "grad_norm": 53.03056716918945,
+      "learning_rate": 1.3825540073592777e-05,
+      "loss": 1.38,
+      "step": 179
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 23.307886123657227,
+      "learning_rate": 1.3794189642586898e-05,
+      "loss": 0.3993,
+      "step": 180
+    },
+    {
+      "epoch": 1.4596774193548387,
+      "grad_norm": 29.451797485351562,
+      "learning_rate": 1.3762839211581017e-05,
+      "loss": 0.3372,
+      "step": 181
+    },
+    {
+      "epoch": 1.467741935483871,
+      "grad_norm": 13.200406074523926,
+      "learning_rate": 1.3731488780575138e-05,
+      "loss": 0.4087,
+      "step": 182
+    },
+    {
+      "epoch": 1.4758064516129032,
+      "grad_norm": 51.59567642211914,
+      "learning_rate": 1.3700138349569259e-05,
+      "loss": 2.83,
+      "step": 183
+    },
+    {
+      "epoch": 1.4838709677419355,
+      "grad_norm": 49.19482421875,
+      "learning_rate": 1.366878791856338e-05,
+      "loss": 4.3183,
+      "step": 184
+    },
+    {
+      "epoch": 1.4919354838709677,
+      "grad_norm": 62.753883361816406,
+      "learning_rate": 1.3637437487557501e-05,
+      "loss": 2.1211,
+      "step": 185
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 13.07942008972168,
+      "learning_rate": 1.360608705655162e-05,
+      "loss": 0.2284,
+      "step": 186
+    },
+    {
+      "epoch": 1.5080645161290323,
+      "grad_norm": 57.539710998535156,
+      "learning_rate": 1.3574736625545742e-05,
+      "loss": 4.917,
+      "step": 187
+    },
+    {
+      "epoch": 1.5161290322580645,
+      "grad_norm": 21.182952880859375,
+      "learning_rate": 1.3543386194539863e-05,
+      "loss": 1.2659,
+      "step": 188
+    },
+    {
+      "epoch": 1.5241935483870968,
+      "grad_norm": 63.780704498291016,
+      "learning_rate": 1.3512035763533984e-05,
+      "loss": 1.9342,
+      "step": 189
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 39.30858612060547,
+      "learning_rate": 1.3480685332528105e-05,
+      "loss": 0.6736,
+      "step": 190
+    },
+    {
+      "epoch": 1.5403225806451613,
+      "grad_norm": 10.713706970214844,
+      "learning_rate": 1.3449334901522224e-05,
+      "loss": 0.2435,
+      "step": 191
+    },
+    {
+      "epoch": 1.5483870967741935,
+      "grad_norm": 26.50084686279297,
+      "learning_rate": 1.3417984470516345e-05,
+      "loss": 0.4394,
+      "step": 192
+    },
+    {
+      "epoch": 1.5564516129032258,
+      "grad_norm": 61.80959701538086,
+      "learning_rate": 1.3386634039510466e-05,
+      "loss": 1.6371,
+      "step": 193
+    },
+    {
+      "epoch": 1.564516129032258,
+      "grad_norm": 44.66097640991211,
+      "learning_rate": 1.3355283608504587e-05,
+      "loss": 0.9796,
+      "step": 194
+    },
+    {
+      "epoch": 1.5725806451612905,
+      "grad_norm": 37.348472595214844,
+      "learning_rate": 1.3323933177498707e-05,
+      "loss": 0.7044,
+      "step": 195
+    },
+    {
+      "epoch": 1.5806451612903225,
+      "grad_norm": 120.70457458496094,
+      "learning_rate": 1.3292582746492828e-05,
+      "loss": 4.7388,
+      "step": 196
+    },
+    {
+      "epoch": 1.588709677419355,
+      "grad_norm": 10.883711814880371,
+      "learning_rate": 1.3261232315486949e-05,
+      "loss": 0.6842,
+      "step": 197
+    },
+    {
+      "epoch": 1.596774193548387,
+      "grad_norm": 9.519349098205566,
+      "learning_rate": 1.322988188448107e-05,
+      "loss": 0.4115,
+      "step": 198
+    },
+    {
+      "epoch": 1.6048387096774195,
+      "grad_norm": 21.985918045043945,
+      "learning_rate": 1.3198531453475191e-05,
+      "loss": 0.1746,
+      "step": 199
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 51.3434944152832,
+      "learning_rate": 1.316718102246931e-05,
+      "loss": 0.8559,
+      "step": 200
+    },
+    {
+      "epoch": 1.620967741935484,
+      "grad_norm": 38.343318939208984,
+      "learning_rate": 1.3135830591463431e-05,
+      "loss": 0.5729,
+      "step": 201
+    },
+    {
+      "epoch": 1.629032258064516,
+      "grad_norm": 31.334217071533203,
+      "learning_rate": 1.3104480160457552e-05,
+      "loss": 0.8461,
+      "step": 202
+    },
+    {
+      "epoch": 1.6370967741935485,
+      "grad_norm": 28.43387222290039,
+      "learning_rate": 1.3073129729451674e-05,
+      "loss": 0.5074,
+      "step": 203
+    },
+    {
+      "epoch": 1.6451612903225805,
+      "grad_norm": 12.179628372192383,
+      "learning_rate": 1.3041779298445795e-05,
+      "loss": 0.5357,
+      "step": 204
+    },
+    {
+      "epoch": 1.653225806451613,
+      "grad_norm": 160.48968505859375,
+      "learning_rate": 1.3010428867439914e-05,
+      "loss": 15.9211,
+      "step": 205
+    },
+    {
+      "epoch": 1.661290322580645,
+      "grad_norm": 8.148846626281738,
+      "learning_rate": 1.2979078436434035e-05,
+      "loss": 0.074,
+      "step": 206
+    },
+    {
+      "epoch": 1.6693548387096775,
+      "grad_norm": 8.479569435119629,
+      "learning_rate": 1.2947728005428156e-05,
+      "loss": 0.1534,
+      "step": 207
+    },
+    {
+      "epoch": 1.6774193548387095,
+      "grad_norm": 19.190534591674805,
+      "learning_rate": 1.2916377574422277e-05,
+      "loss": 0.3617,
+      "step": 208
+    },
+    {
+      "epoch": 1.685483870967742,
+      "grad_norm": 7.419437408447266,
+      "learning_rate": 1.2885027143416398e-05,
+      "loss": 0.372,
+      "step": 209
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 36.760528564453125,
+      "learning_rate": 1.2853676712410518e-05,
+      "loss": 0.6221,
+      "step": 210
+    },
+    {
+      "epoch": 1.7016129032258065,
+      "grad_norm": 46.47172546386719,
+      "learning_rate": 1.2822326281404639e-05,
+      "loss": 1.0012,
+      "step": 211
+    },
+    {
+      "epoch": 1.7096774193548387,
+      "grad_norm": 21.93129539489746,
+      "learning_rate": 1.279097585039876e-05,
+      "loss": 0.9615,
+      "step": 212
+    },
+    {
+      "epoch": 1.717741935483871,
+      "grad_norm": 4.738753795623779,
+      "learning_rate": 1.275962541939288e-05,
+      "loss": 0.0968,
+      "step": 213
+    },
+    {
+      "epoch": 1.7258064516129032,
+      "grad_norm": 144.14422607421875,
+      "learning_rate": 1.2728274988386998e-05,
+      "loss": 24.8741,
+      "step": 214
+    },
+    {
+      "epoch": 1.7338709677419355,
+      "grad_norm": 98.76880645751953,
+      "learning_rate": 1.2696924557381121e-05,
+      "loss": 15.2355,
+      "step": 215
+    },
+    {
+      "epoch": 1.7419354838709677,
+      "grad_norm": 10.59051513671875,
+      "learning_rate": 1.2665574126375242e-05,
+      "loss": 0.3515,
+      "step": 216
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 131.96859741210938,
+      "learning_rate": 1.2634223695369363e-05,
+      "loss": 7.7135,
+      "step": 217
+    },
+    {
+      "epoch": 1.7580645161290323,
+      "grad_norm": 25.29785919189453,
+      "learning_rate": 1.2602873264363484e-05,
+      "loss": 0.2482,
+      "step": 218
+    },
+    {
+      "epoch": 1.7661290322580645,
+      "grad_norm": 177.49879455566406,
+      "learning_rate": 1.2571522833357602e-05,
+      "loss": 15.0218,
+      "step": 219
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 30.580156326293945,
+      "learning_rate": 1.2540172402351725e-05,
+      "loss": 0.4993,
+      "step": 220
+    },
+    {
+      "epoch": 1.782258064516129,
+      "grad_norm": 82.16755676269531,
+      "learning_rate": 1.2508821971345846e-05,
+      "loss": 0.8773,
+      "step": 221
+    },
+    {
+      "epoch": 1.7903225806451613,
+      "grad_norm": 13.509096145629883,
+      "learning_rate": 1.2477471540339967e-05,
+      "loss": 0.598,
+      "step": 222
+    },
+    {
+      "epoch": 1.7983870967741935,
+      "grad_norm": 31.23018455505371,
+      "learning_rate": 1.2446121109334088e-05,
+      "loss": 0.2144,
+      "step": 223
+    },
+    {
+      "epoch": 1.8064516129032258,
+      "grad_norm": 146.499267578125,
+      "learning_rate": 1.2414770678328206e-05,
+      "loss": 14.024,
+      "step": 224
+    },
+    {
+      "epoch": 1.814516129032258,
+      "grad_norm": 411.3276672363281,
+      "learning_rate": 1.2383420247322328e-05,
+      "loss": 6.1271,
+      "step": 225
+    },
+    {
+      "epoch": 1.8225806451612905,
+      "grad_norm": 36.80350112915039,
+      "learning_rate": 1.235206981631645e-05,
+      "loss": 1.4717,
+      "step": 226
+    },
+    {
+      "epoch": 1.8306451612903225,
+      "grad_norm": 81.75621795654297,
+      "learning_rate": 1.232071938531057e-05,
+      "loss": 1.1353,
+      "step": 227
+    },
+    {
+      "epoch": 1.838709677419355,
+      "grad_norm": 64.03866577148438,
+      "learning_rate": 1.2289368954304688e-05,
+      "loss": 1.1949,
+      "step": 228
+    },
+    {
+      "epoch": 1.846774193548387,
+      "grad_norm": 38.97941207885742,
+      "learning_rate": 1.225801852329881e-05,
+      "loss": 0.7342,
+      "step": 229
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 247.11788940429688,
+      "learning_rate": 1.2226668092292932e-05,
+      "loss": 7.2037,
+      "step": 230
+    },
+    {
+      "epoch": 1.8629032258064515,
+      "grad_norm": 62.207733154296875,
+      "learning_rate": 1.2195317661287053e-05,
+      "loss": 1.5858,
+      "step": 231
+    },
+    {
+      "epoch": 1.870967741935484,
+      "grad_norm": 15.489973068237305,
+      "learning_rate": 1.2163967230281174e-05,
+      "loss": 0.3868,
+      "step": 232
+    },
+    {
+      "epoch": 1.879032258064516,
+      "grad_norm": 29.43061637878418,
+      "learning_rate": 1.2132616799275292e-05,
+      "loss": 1.1257,
+      "step": 233
+    },
+    {
+      "epoch": 1.8870967741935485,
+      "grad_norm": 10.211771011352539,
+      "learning_rate": 1.2101266368269413e-05,
+      "loss": 0.1734,
+      "step": 234
+    },
+    {
+      "epoch": 1.8951612903225805,
+      "grad_norm": 11.279986381530762,
+      "learning_rate": 1.2069915937263534e-05,
+      "loss": 0.4312,
+      "step": 235
+    },
+    {
+      "epoch": 1.903225806451613,
+      "grad_norm": 119.58721160888672,
+      "learning_rate": 1.2038565506257657e-05,
+      "loss": 9.7858,
+      "step": 236
+    },
+    {
+      "epoch": 1.911290322580645,
+      "grad_norm": 201.3207244873047,
+      "learning_rate": 1.2007215075251778e-05,
+      "loss": 4.2142,
+      "step": 237
+    },
+    {
+      "epoch": 1.9193548387096775,
+      "grad_norm": 37.971343994140625,
+      "learning_rate": 1.1975864644245895e-05,
+      "loss": 0.9522,
+      "step": 238
+    },
+    {
+      "epoch": 1.9274193548387095,
+      "grad_norm": 34.468807220458984,
+      "learning_rate": 1.1944514213240016e-05,
+      "loss": 0.7723,
+      "step": 239
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 39.51136016845703,
+      "learning_rate": 1.1913163782234138e-05,
+      "loss": 2.6126,
+      "step": 240
+    },
+    {
+      "epoch": 1.9435483870967742,
+      "grad_norm": 27.13111686706543,
+      "learning_rate": 1.188181335122826e-05,
+      "loss": 0.6028,
+      "step": 241
+    },
+    {
+      "epoch": 1.9516129032258065,
+      "grad_norm": 17.496204376220703,
+      "learning_rate": 1.1850462920222378e-05,
+      "loss": 0.733,
+      "step": 242
+    },
+    {
+      "epoch": 1.9596774193548387,
+      "grad_norm": 43.25605773925781,
+      "learning_rate": 1.1819112489216499e-05,
+      "loss": 1.8264,
+      "step": 243
+    },
+    {
+      "epoch": 1.967741935483871,
+      "grad_norm": 144.58236694335938,
+      "learning_rate": 1.178776205821062e-05,
+      "loss": 4.7865,
+      "step": 244
+    },
+    {
+      "epoch": 1.9758064516129032,
+      "grad_norm": 71.39822387695312,
+      "learning_rate": 1.1756411627204741e-05,
+      "loss": 1.7471,
+      "step": 245
+    },
+    {
+      "epoch": 1.9838709677419355,
+      "grad_norm": 418.1564025878906,
+      "learning_rate": 1.1725061196198864e-05,
+      "loss": 15.5524,
+      "step": 246
+    },
+    {
+      "epoch": 1.9919354838709677,
+      "grad_norm": 103.1528091430664,
+      "learning_rate": 1.1693710765192982e-05,
+      "loss": 2.4682,
+      "step": 247
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 129.07899475097656,
+      "learning_rate": 1.1662360334187103e-05,
+      "loss": 3.7685,
+      "step": 248
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 4.093452453613281,
+      "eval_mae": 1.4580568075180054,
+      "eval_mse": 4.093452453613281,
+      "eval_r2": -0.12071609497070312,
+      "eval_rmse": 2.02322822578504,
+      "eval_runtime": 1.3208,
+      "eval_samples_per_second": 41.641,
+      "eval_smape": 53.77084016799927,
+      "eval_steps_per_second": 10.599,
+      "step": 248
+    },
+    {
+      "epoch": 2.0080645161290325,
+      "grad_norm": 176.1205291748047,
+      "learning_rate": 1.1631009903181224e-05,
+      "loss": 3.3068,
+      "step": 249
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 84.84161376953125,
+      "learning_rate": 1.1599659472175345e-05,
+      "loss": 2.8326,
+      "step": 250
+    },
+    {
+      "epoch": 2.024193548387097,
+      "grad_norm": 128.6629638671875,
+      "learning_rate": 1.1568309041169468e-05,
+      "loss": 4.0163,
+      "step": 251
+    },
+    {
+      "epoch": 2.032258064516129,
+      "grad_norm": 70.34894561767578,
+      "learning_rate": 1.1536958610163585e-05,
+      "loss": 3.7917,
+      "step": 252
+    },
+    {
+      "epoch": 2.0403225806451615,
+      "grad_norm": 18.99244499206543,
+      "learning_rate": 1.1505608179157706e-05,
+      "loss": 0.2877,
+      "step": 253
+    },
+    {
+      "epoch": 2.0483870967741935,
+      "grad_norm": 189.7742156982422,
+      "learning_rate": 1.1474257748151827e-05,
+      "loss": 6.5608,
+      "step": 254
+    },
+    {
+      "epoch": 2.056451612903226,
+      "grad_norm": 19.993907928466797,
+      "learning_rate": 1.1442907317145948e-05,
+      "loss": 0.2394,
+      "step": 255
+    },
+    {
+      "epoch": 2.064516129032258,
+      "grad_norm": 174.26377868652344,
+      "learning_rate": 1.1411556886140068e-05,
+      "loss": 19.2623,
+      "step": 256
+    },
+    {
+      "epoch": 2.0725806451612905,
+      "grad_norm": 20.718603134155273,
+      "learning_rate": 1.1380206455134189e-05,
+      "loss": 0.2838,
+      "step": 257
+    },
+    {
+      "epoch": 2.0806451612903225,
+      "grad_norm": 55.44133377075195,
+      "learning_rate": 1.134885602412831e-05,
+      "loss": 0.9125,
+      "step": 258
+    },
+    {
+      "epoch": 2.088709677419355,
+      "grad_norm": 113.99673461914062,
+      "learning_rate": 1.1317505593122431e-05,
+      "loss": 23.6723,
+      "step": 259
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 23.634668350219727,
+      "learning_rate": 1.1286155162116552e-05,
+      "loss": 0.3163,
+      "step": 260
+    },
+    {
+      "epoch": 2.1048387096774195,
+      "grad_norm": 142.3331756591797,
+      "learning_rate": 1.1254804731110671e-05,
+      "loss": 24.4121,
+      "step": 261
+    },
+    {
+      "epoch": 2.1129032258064515,
+      "grad_norm": 97.85147857666016,
+      "learning_rate": 1.1223454300104792e-05,
+      "loss": 0.8764,
+      "step": 262
+    },
+    {
+      "epoch": 2.120967741935484,
+      "grad_norm": 33.50703430175781,
+      "learning_rate": 1.1192103869098913e-05,
+      "loss": 0.2653,
+      "step": 263
+    },
+    {
+      "epoch": 2.129032258064516,
+      "grad_norm": 25.053003311157227,
+      "learning_rate": 1.1160753438093035e-05,
+      "loss": 0.4152,
+      "step": 264
+    },
+    {
+      "epoch": 2.1370967741935485,
+      "grad_norm": 55.80361557006836,
+      "learning_rate": 1.1129403007087156e-05,
+      "loss": 0.8291,
+      "step": 265
+    },
+    {
+      "epoch": 2.1451612903225805,
+      "grad_norm": 161.3894805908203,
+      "learning_rate": 1.1098052576081275e-05,
+      "loss": 4.3344,
+      "step": 266
+    },
+    {
+      "epoch": 2.153225806451613,
+      "grad_norm": 12.537243843078613,
+      "learning_rate": 1.1066702145075396e-05,
+      "loss": 0.2252,
+      "step": 267
+    },
+    {
+      "epoch": 2.161290322580645,
+      "grad_norm": 60.92879867553711,
+      "learning_rate": 1.1035351714069517e-05,
+      "loss": 1.7349,
+      "step": 268
+    },
+    {
+      "epoch": 2.1693548387096775,
+      "grad_norm": 52.16379165649414,
+      "learning_rate": 1.1004001283063638e-05,
+      "loss": 2.7655,
+      "step": 269
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 13.385435104370117,
+      "learning_rate": 1.097265085205776e-05,
+      "loss": 0.7339,
+      "step": 270
+    },
+    {
+      "epoch": 2.185483870967742,
+      "grad_norm": 82.908935546875,
+      "learning_rate": 1.0941300421051879e-05,
+      "loss": 10.5988,
+      "step": 271
+    },
+    {
+      "epoch": 2.193548387096774,
+      "grad_norm": 114.64527130126953,
+      "learning_rate": 1.0909949990046e-05,
+      "loss": 14.9863,
+      "step": 272
+    },
+    {
+      "epoch": 2.2016129032258065,
+      "grad_norm": 90.68907928466797,
+      "learning_rate": 1.087859955904012e-05,
+      "loss": 12.8366,
+      "step": 273
+    },
+    {
+      "epoch": 2.2096774193548385,
+      "grad_norm": 24.137863159179688,
+      "learning_rate": 1.0847249128034242e-05,
+      "loss": 0.6209,
+      "step": 274
+    },
+    {
+      "epoch": 2.217741935483871,
+      "grad_norm": 86.06786346435547,
+      "learning_rate": 1.0815898697028361e-05,
+      "loss": 0.6642,
+      "step": 275
+    },
+    {
+      "epoch": 2.225806451612903,
+      "grad_norm": 34.535762786865234,
+      "learning_rate": 1.0784548266022482e-05,
+      "loss": 0.3586,
+      "step": 276
+    },
+    {
+      "epoch": 2.2338709677419355,
+      "grad_norm": 55.70576095581055,
+      "learning_rate": 1.0753197835016603e-05,
+      "loss": 0.7619,
+      "step": 277
+    },
+    {
+      "epoch": 2.241935483870968,
+      "grad_norm": 76.06886291503906,
+      "learning_rate": 1.0721847404010724e-05,
+      "loss": 2.1191,
+      "step": 278
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 28.001771926879883,
+      "learning_rate": 1.0690496973004845e-05,
+      "loss": 0.8164,
+      "step": 279
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 103.4825439453125,
+      "learning_rate": 1.0659146541998965e-05,
+      "loss": 10.475,
+      "step": 280
+    },
+    {
+      "epoch": 2.2661290322580645,
+      "grad_norm": 33.585960388183594,
+      "learning_rate": 1.0627796110993086e-05,
+      "loss": 1.6235,
+      "step": 281
+    },
+    {
+      "epoch": 2.274193548387097,
+      "grad_norm": 232.9297332763672,
+      "learning_rate": 1.0596445679987207e-05,
+      "loss": 13.3261,
+      "step": 282
+    },
+    {
+      "epoch": 2.282258064516129,
+      "grad_norm": 96.77767944335938,
+      "learning_rate": 1.0565095248981328e-05,
+      "loss": 3.1002,
+      "step": 283
+    },
+    {
+      "epoch": 2.2903225806451615,
+      "grad_norm": 116.28626251220703,
+      "learning_rate": 1.0533744817975449e-05,
+      "loss": 5.5134,
+      "step": 284
+    },
+    {
+      "epoch": 2.2983870967741935,
+      "grad_norm": 69.49893951416016,
+      "learning_rate": 1.0502394386969568e-05,
+      "loss": 1.413,
+      "step": 285
+    },
+    {
+      "epoch": 2.306451612903226,
+      "grad_norm": 36.76485824584961,
+      "learning_rate": 1.047104395596369e-05,
+      "loss": 0.3031,
+      "step": 286
+    },
+    {
+      "epoch": 2.314516129032258,
+      "grad_norm": 62.41328048706055,
+      "learning_rate": 1.043969352495781e-05,
+      "loss": 1.5162,
+      "step": 287
+    },
+    {
+      "epoch": 2.3225806451612905,
+      "grad_norm": 30.905513763427734,
+      "learning_rate": 1.0408343093951932e-05,
+      "loss": 0.6394,
+      "step": 288
+    },
+    {
+      "epoch": 2.3306451612903225,
+      "grad_norm": 48.83860778808594,
+      "learning_rate": 1.0376992662946051e-05,
+      "loss": 1.7204,
+      "step": 289
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 95.79818725585938,
+      "learning_rate": 1.0345642231940172e-05,
+      "loss": 9.2122,
+      "step": 290
+    },
+    {
+      "epoch": 2.346774193548387,
+      "grad_norm": 45.55128860473633,
+      "learning_rate": 1.0314291800934293e-05,
+      "loss": 0.6816,
+      "step": 291
+    },
+    {
+      "epoch": 2.3548387096774195,
+      "grad_norm": 170.74098205566406,
+      "learning_rate": 1.0282941369928414e-05,
+      "loss": 5.1995,
+      "step": 292
+    },
+    {
+      "epoch": 2.3629032258064515,
+      "grad_norm": 27.760730743408203,
+      "learning_rate": 1.0251590938922535e-05,
+      "loss": 0.2949,
+      "step": 293
+    },
+    {
+      "epoch": 2.370967741935484,
+      "grad_norm": 61.26620864868164,
+      "learning_rate": 1.0220240507916654e-05,
+      "loss": 12.498,
+      "step": 294
+    },
+    {
+      "epoch": 2.379032258064516,
+      "grad_norm": 88.63400268554688,
+      "learning_rate": 1.0188890076910776e-05,
+      "loss": 14.3648,
+      "step": 295
+    },
+    {
+      "epoch": 2.3870967741935485,
+      "grad_norm": 15.192766189575195,
+      "learning_rate": 1.0157539645904897e-05,
+      "loss": 0.7797,
+      "step": 296
+    },
+    {
+      "epoch": 2.3951612903225805,
+      "grad_norm": 14.215178489685059,
+      "learning_rate": 1.0126189214899018e-05,
+      "loss": 0.3113,
+      "step": 297
+    },
+    {
+      "epoch": 2.403225806451613,
+      "grad_norm": 83.61798858642578,
+      "learning_rate": 1.0094838783893139e-05,
+      "loss": 2.0366,
+      "step": 298
+    },
+    {
+      "epoch": 2.411290322580645,
+      "grad_norm": 16.302412033081055,
+      "learning_rate": 1.0063488352887258e-05,
+      "loss": 0.7133,
+      "step": 299
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 128.92327880859375,
+      "learning_rate": 1.0032137921881379e-05,
+      "loss": 3.2273,
+      "step": 300
+    },
+    {
+      "epoch": 2.4274193548387095,
+      "grad_norm": 42.748626708984375,
+      "learning_rate": 1.00007874908755e-05,
+      "loss": 0.66,
+      "step": 301
+    },
+    {
+      "epoch": 2.435483870967742,
+      "grad_norm": 30.09368133544922,
+      "learning_rate": 9.969437059869621e-06,
+      "loss": 0.2957,
+      "step": 302
+    },
+    {
+      "epoch": 2.443548387096774,
+      "grad_norm": 98.78866577148438,
+      "learning_rate": 9.93808662886374e-06,
+      "loss": 6.8945,
+      "step": 303
+    },
+    {
+      "epoch": 2.4516129032258065,
+      "grad_norm": 47.38784408569336,
+      "learning_rate": 9.906736197857862e-06,
+      "loss": 0.5275,
+      "step": 304
+    },
+    {
+      "epoch": 2.4596774193548385,
+      "grad_norm": 91.3726806640625,
+      "learning_rate": 9.875385766851983e-06,
+      "loss": 12.6285,
+      "step": 305
+    },
+    {
+      "epoch": 2.467741935483871,
+      "grad_norm": 14.849919319152832,
+      "learning_rate": 9.844035335846104e-06,
+      "loss": 0.1922,
+      "step": 306
+    },
+    {
+      "epoch": 2.475806451612903,
+      "grad_norm": 27.1463565826416,
+      "learning_rate": 9.812684904840225e-06,
+      "loss": 0.4246,
+      "step": 307
+    },
+    {
+      "epoch": 2.4838709677419355,
+      "grad_norm": 305.164306640625,
+      "learning_rate": 9.781334473834344e-06,
+      "loss": 11.3922,
+      "step": 308
+    },
+    {
+      "epoch": 2.491935483870968,
+      "grad_norm": 20.79343605041504,
+      "learning_rate": 9.749984042828465e-06,
+      "loss": 0.448,
+      "step": 309
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 23.24410629272461,
+      "learning_rate": 9.718633611822586e-06,
+      "loss": 0.4327,
+      "step": 310
+    },
+    {
+      "epoch": 2.508064516129032,
+      "grad_norm": 9.029376029968262,
+      "learning_rate": 9.687283180816707e-06,
+      "loss": 0.3693,
+      "step": 311
+    },
+    {
+      "epoch": 2.5161290322580645,
+      "grad_norm": 9.926502227783203,
+      "learning_rate": 9.655932749810827e-06,
+      "loss": 0.1628,
+      "step": 312
+    },
+    {
+      "epoch": 2.524193548387097,
+      "grad_norm": 5.0813703536987305,
+      "learning_rate": 9.62458231880495e-06,
+      "loss": 0.0807,
+      "step": 313
+    },
+    {
+      "epoch": 2.532258064516129,
+      "grad_norm": 14.10543441772461,
+      "learning_rate": 9.593231887799069e-06,
+      "loss": 0.4584,
+      "step": 314
+    },
+    {
+      "epoch": 2.540322580645161,
+      "grad_norm": 112.93350219726562,
+      "learning_rate": 9.56188145679319e-06,
+      "loss": 10.0475,
+      "step": 315
+    },
+    {
+      "epoch": 2.5483870967741935,
+      "grad_norm": 8.689790725708008,
+      "learning_rate": 9.53053102578731e-06,
+      "loss": 0.2067,
+      "step": 316
+    },
+    {
+      "epoch": 2.556451612903226,
+      "grad_norm": 28.200075149536133,
+      "learning_rate": 9.49918059478143e-06,
+      "loss": 1.2775,
+      "step": 317
+    },
+    {
+      "epoch": 2.564516129032258,
+      "grad_norm": 15.749256134033203,
+      "learning_rate": 9.467830163775553e-06,
+      "loss": 0.4091,
+      "step": 318
+    },
+    {
+      "epoch": 2.5725806451612905,
+      "grad_norm": 8.130678176879883,
+      "learning_rate": 9.436479732769673e-06,
+      "loss": 0.7939,
+      "step": 319
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 13.053906440734863,
+      "learning_rate": 9.405129301763794e-06,
+      "loss": 0.8335,
+      "step": 320
+    },
+    {
+      "epoch": 2.588709677419355,
+      "grad_norm": 166.3572540283203,
+      "learning_rate": 9.373778870757913e-06,
+      "loss": 6.0399,
+      "step": 321
+    },
+    {
+      "epoch": 2.596774193548387,
+      "grad_norm": 9.214460372924805,
+      "learning_rate": 9.342428439752034e-06,
+      "loss": 0.3924,
+      "step": 322
+    },
+    {
+      "epoch": 2.6048387096774195,
+      "grad_norm": 143.34503173828125,
+      "learning_rate": 9.311078008746155e-06,
+      "loss": 13.3153,
+      "step": 323
+    },
+    {
+      "epoch": 2.6129032258064515,
+      "grad_norm": 23.62826156616211,
+      "learning_rate": 9.279727577740276e-06,
+      "loss": 0.3648,
+      "step": 324
+    },
+    {
+      "epoch": 2.620967741935484,
+      "grad_norm": 12.584870338439941,
+      "learning_rate": 9.248377146734397e-06,
+      "loss": 0.1898,
+      "step": 325
+    },
+    {
+      "epoch": 2.629032258064516,
+      "grad_norm": 20.08246612548828,
+      "learning_rate": 9.217026715728517e-06,
+      "loss": 0.6812,
+      "step": 326
+    },
+    {
+      "epoch": 2.6370967741935485,
+      "grad_norm": 15.498703956604004,
+      "learning_rate": 9.185676284722638e-06,
+      "loss": 0.8188,
+      "step": 327
+    },
+    {
+      "epoch": 2.6451612903225805,
+      "grad_norm": 26.546886444091797,
+      "learning_rate": 9.154325853716759e-06,
+      "loss": 0.2763,
+      "step": 328
+    },
+    {
+      "epoch": 2.653225806451613,
+      "grad_norm": 381.572021484375,
+      "learning_rate": 9.12297542271088e-06,
+      "loss": 126.325,
+      "step": 329
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 92.21846771240234,
+      "learning_rate": 9.091624991704999e-06,
+      "loss": 6.1702,
+      "step": 330
+    },
+    {
+      "epoch": 2.6693548387096775,
+      "grad_norm": 16.447362899780273,
+      "learning_rate": 9.06027456069912e-06,
+      "loss": 0.8041,
+      "step": 331
+    },
+    {
+      "epoch": 2.6774193548387095,
+      "grad_norm": 19.612592697143555,
+      "learning_rate": 9.028924129693241e-06,
+      "loss": 1.1898,
+      "step": 332
+    },
+    {
+      "epoch": 2.685483870967742,
+      "grad_norm": 40.966026306152344,
+      "learning_rate": 8.997573698687362e-06,
+      "loss": 0.2561,
+      "step": 333
+    },
+    {
+      "epoch": 2.693548387096774,
+      "grad_norm": 164.29673767089844,
+      "learning_rate": 8.966223267681483e-06,
+      "loss": 10.4936,
+      "step": 334
+    },
+    {
+      "epoch": 2.7016129032258065,
+      "grad_norm": 20.719127655029297,
+      "learning_rate": 8.934872836675603e-06,
+      "loss": 1.0469,
+      "step": 335
+    },
+    {
+      "epoch": 2.709677419354839,
+      "grad_norm": 46.28770446777344,
+      "learning_rate": 8.903522405669724e-06,
+      "loss": 1.2291,
+      "step": 336
+    },
+    {
+      "epoch": 2.717741935483871,
+      "grad_norm": 53.001338958740234,
+      "learning_rate": 8.872171974663845e-06,
+      "loss": 1.0995,
+      "step": 337
+    },
+    {
+      "epoch": 2.725806451612903,
+      "grad_norm": 80.97772979736328,
+      "learning_rate": 8.840821543657966e-06,
+      "loss": 1.4318,
+      "step": 338
+    },
+    {
+      "epoch": 2.7338709677419355,
+      "grad_norm": 81.25788116455078,
+      "learning_rate": 8.809471112652087e-06,
+      "loss": 10.1732,
+      "step": 339
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 70.82359313964844,
+      "learning_rate": 8.778120681646206e-06,
+      "loss": 1.1948,
+      "step": 340
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 28.553592681884766,
+      "learning_rate": 8.746770250640327e-06,
+      "loss": 0.3979,
+      "step": 341
+    },
+    {
+      "epoch": 2.758064516129032,
+      "grad_norm": 58.9395751953125,
+      "learning_rate": 8.715419819634448e-06,
+      "loss": 8.2167,
+      "step": 342
+    },
+    {
+      "epoch": 2.7661290322580645,
+      "grad_norm": 36.5241584777832,
+      "learning_rate": 8.68406938862857e-06,
+      "loss": 0.5727,
+      "step": 343
+    },
+    {
+      "epoch": 2.774193548387097,
+      "grad_norm": 12.758500099182129,
+      "learning_rate": 8.65271895762269e-06,
+      "loss": 0.5117,
+      "step": 344
+    },
+    {
+      "epoch": 2.782258064516129,
+      "grad_norm": 83.17584228515625,
+      "learning_rate": 8.62136852661681e-06,
+      "loss": 2.2623,
+      "step": 345
+    },
+    {
+      "epoch": 2.790322580645161,
+      "grad_norm": 26.435558319091797,
+      "learning_rate": 8.590018095610931e-06,
+      "loss": 0.2321,
+      "step": 346
+    },
+    {
+      "epoch": 2.7983870967741935,
+      "grad_norm": 50.01573181152344,
+      "learning_rate": 8.558667664605052e-06,
+      "loss": 0.9035,
+      "step": 347
+    },
+    {
+      "epoch": 2.806451612903226,
+      "grad_norm": 13.123492240905762,
+      "learning_rate": 8.527317233599173e-06,
+      "loss": 0.274,
+      "step": 348
+    },
+    {
+      "epoch": 2.814516129032258,
+      "grad_norm": 62.45608901977539,
+      "learning_rate": 8.495966802593292e-06,
+      "loss": 6.3547,
+      "step": 349
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 78.55066680908203,
+      "learning_rate": 8.464616371587414e-06,
+      "loss": 2.2642,
+      "step": 350
+    },
+    {
+      "epoch": 2.8306451612903225,
+      "grad_norm": 56.115962982177734,
+      "learning_rate": 8.433265940581535e-06,
+      "loss": 0.8852,
+      "step": 351
+    },
+    {
+      "epoch": 2.838709677419355,
+      "grad_norm": 36.07647705078125,
+      "learning_rate": 8.401915509575656e-06,
+      "loss": 2.3096,
+      "step": 352
+    },
+    {
+      "epoch": 2.846774193548387,
+      "grad_norm": 37.17245864868164,
+      "learning_rate": 8.370565078569777e-06,
+      "loss": 0.8445,
+      "step": 353
+    },
+    {
+      "epoch": 2.8548387096774195,
+      "grad_norm": 146.50592041015625,
+      "learning_rate": 8.339214647563896e-06,
+      "loss": 7.5497,
+      "step": 354
+    },
+    {
+      "epoch": 2.8629032258064515,
+      "grad_norm": 122.4879150390625,
+      "learning_rate": 8.307864216558017e-06,
+      "loss": 6.0334,
+      "step": 355
+    },
+    {
+      "epoch": 2.870967741935484,
+      "grad_norm": 32.15070724487305,
+      "learning_rate": 8.276513785552138e-06,
+      "loss": 0.3793,
+      "step": 356
+    },
+    {
+      "epoch": 2.879032258064516,
+      "grad_norm": 143.58071899414062,
+      "learning_rate": 8.24516335454626e-06,
+      "loss": 16.1946,
+      "step": 357
+    },
+    {
+      "epoch": 2.8870967741935485,
+      "grad_norm": 24.55172348022461,
+      "learning_rate": 8.21381292354038e-06,
+      "loss": 0.178,
+      "step": 358
+    },
+    {
+      "epoch": 2.8951612903225805,
+      "grad_norm": 24.15800666809082,
+      "learning_rate": 8.1824624925345e-06,
+      "loss": 0.5304,
+      "step": 359
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 16.40801239013672,
+      "learning_rate": 8.15111206152862e-06,
+      "loss": 0.7086,
+      "step": 360
+    },
+    {
+      "epoch": 2.911290322580645,
+      "grad_norm": 144.3972930908203,
+      "learning_rate": 8.119761630522742e-06,
+      "loss": 15.8248,
+      "step": 361
+    },
+    {
+      "epoch": 2.9193548387096775,
+      "grad_norm": 56.63901901245117,
+      "learning_rate": 8.088411199516863e-06,
+      "loss": 0.8383,
+      "step": 362
+    },
+    {
+      "epoch": 2.9274193548387095,
+      "grad_norm": 36.86946487426758,
+      "learning_rate": 8.057060768510982e-06,
+      "loss": 0.4044,
+      "step": 363
+    },
+    {
+      "epoch": 2.935483870967742,
+      "grad_norm": 54.56624984741211,
+      "learning_rate": 8.025710337505103e-06,
+      "loss": 0.8564,
+      "step": 364
+    },
+    {
+      "epoch": 2.943548387096774,
+      "grad_norm": 45.62172317504883,
+      "learning_rate": 7.994359906499224e-06,
+      "loss": 0.8052,
+      "step": 365
+    },
+    {
+      "epoch": 2.9516129032258065,
+      "grad_norm": 28.312416076660156,
+      "learning_rate": 7.963009475493345e-06,
+      "loss": 0.6061,
+      "step": 366
+    },
+    {
+      "epoch": 2.959677419354839,
+      "grad_norm": 47.32063293457031,
+      "learning_rate": 7.931659044487467e-06,
+      "loss": 0.7651,
+      "step": 367
+    },
+    {
+      "epoch": 2.967741935483871,
+      "grad_norm": 93.73538208007812,
+      "learning_rate": 7.900308613481586e-06,
+      "loss": 1.4781,
+      "step": 368
+    },
+    {
+      "epoch": 2.975806451612903,
+      "grad_norm": 30.8656063079834,
+      "learning_rate": 7.868958182475707e-06,
+      "loss": 0.3955,
+      "step": 369
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 11.793440818786621,
+      "learning_rate": 7.837607751469828e-06,
+      "loss": 0.3443,
+      "step": 370
+    },
+    {
+      "epoch": 2.991935483870968,
+      "grad_norm": 47.234580993652344,
+      "learning_rate": 7.806257320463949e-06,
+      "loss": 2.2459,
+      "step": 371
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 330.5805969238281,
+      "learning_rate": 7.77490688945807e-06,
+      "loss": 27.661,
+      "step": 372
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 3.361875057220459,
+      "eval_mae": 1.126670479774475,
+      "eval_mse": 3.3618738651275635,
+      "eval_r2": 0.07957738637924194,
+      "eval_rmse": 1.8335413453553655,
+      "eval_runtime": 1.3777,
+      "eval_samples_per_second": 39.921,
+      "eval_smape": 52.641284465789795,
+      "eval_steps_per_second": 10.162,
+      "step": 372
+    },
+    {
+      "epoch": 3.0080645161290325,
+      "grad_norm": 75.24702453613281,
+      "learning_rate": 7.74355645845219e-06,
+      "loss": 1.7123,
+      "step": 373
+    },
+    {
+      "epoch": 3.0161290322580645,
+      "grad_norm": 127.73233795166016,
+      "learning_rate": 7.71220602744631e-06,
+      "loss": 7.3851,
+      "step": 374
+    },
+    {
+      "epoch": 3.024193548387097,
+      "grad_norm": 13.305014610290527,
+      "learning_rate": 7.680855596440432e-06,
+      "loss": 0.2886,
+      "step": 375
+    },
+    {
+      "epoch": 3.032258064516129,
+      "grad_norm": 41.47939682006836,
+      "learning_rate": 7.649505165434553e-06,
+      "loss": 1.5766,
+      "step": 376
+    },
+    {
+      "epoch": 3.0403225806451615,
+      "grad_norm": 5.5650177001953125,
+      "learning_rate": 7.618154734428672e-06,
+      "loss": 0.1177,
+      "step": 377
+    },
+    {
+      "epoch": 3.0483870967741935,
+      "grad_norm": 60.55559539794922,
+      "learning_rate": 7.586804303422793e-06,
+      "loss": 1.4604,
+      "step": 378
+    },
+    {
+      "epoch": 3.056451612903226,
+      "grad_norm": 45.52365493774414,
+      "learning_rate": 7.555453872416914e-06,
+      "loss": 0.946,
+      "step": 379
+    },
+    {
+      "epoch": 3.064516129032258,
+      "grad_norm": 20.433025360107422,
+      "learning_rate": 7.524103441411034e-06,
+      "loss": 0.6702,
+      "step": 380
+    },
+    {
+      "epoch": 3.0725806451612905,
+      "grad_norm": 218.40994262695312,
+      "learning_rate": 7.492753010405155e-06,
+      "loss": 12.6873,
+      "step": 381
+    },
+    {
+      "epoch": 3.0806451612903225,
+      "grad_norm": 50.64395523071289,
+      "learning_rate": 7.461402579399276e-06,
+      "loss": 1.67,
+      "step": 382
+    },
+    {
+      "epoch": 3.088709677419355,
+      "grad_norm": 29.460969924926758,
+      "learning_rate": 7.430052148393397e-06,
+      "loss": 0.5095,
+      "step": 383
+    },
+    {
+      "epoch": 3.096774193548387,
+      "grad_norm": 31.18248176574707,
+      "learning_rate": 7.398701717387517e-06,
+      "loss": 1.3069,
+      "step": 384
+    },
+    {
+      "epoch": 3.1048387096774195,
+      "grad_norm": 71.00679779052734,
+      "learning_rate": 7.367351286381638e-06,
+      "loss": 2.2479,
+      "step": 385
+    },
+    {
+      "epoch": 3.1129032258064515,
+      "grad_norm": 23.04047393798828,
+      "learning_rate": 7.336000855375759e-06,
+      "loss": 0.2883,
+      "step": 386
+    },
+    {
+      "epoch": 3.120967741935484,
+      "grad_norm": 30.270566940307617,
+      "learning_rate": 7.304650424369879e-06,
+      "loss": 0.4375,
+      "step": 387
+    },
+    {
+      "epoch": 3.129032258064516,
+      "grad_norm": 19.1580867767334,
+      "learning_rate": 7.273299993364e-06,
+      "loss": 0.5482,
+      "step": 388
+    },
+    {
+      "epoch": 3.1370967741935485,
+      "grad_norm": 26.573993682861328,
+      "learning_rate": 7.2419495623581205e-06,
+      "loss": 0.3888,
+      "step": 389
+    },
+    {
+      "epoch": 3.1451612903225805,
+      "grad_norm": 43.726409912109375,
+      "learning_rate": 7.210599131352242e-06,
+      "loss": 0.6747,
+      "step": 390
+    },
+    {
+      "epoch": 3.153225806451613,
+      "grad_norm": 100.00196075439453,
+      "learning_rate": 7.179248700346363e-06,
+      "loss": 10.1746,
+      "step": 391
+    },
+    {
+      "epoch": 3.161290322580645,
+      "grad_norm": 17.870851516723633,
+      "learning_rate": 7.147898269340483e-06,
+      "loss": 0.8346,
+      "step": 392
+    },
+    {
+      "epoch": 3.1693548387096775,
+      "grad_norm": 11.3032865524292,
+      "learning_rate": 7.116547838334604e-06,
+      "loss": 0.0478,
+      "step": 393
+    },
+    {
+      "epoch": 3.1774193548387095,
+      "grad_norm": 8.371842384338379,
+      "learning_rate": 7.085197407328724e-06,
+      "loss": 0.1455,
+      "step": 394
+    },
+    {
+      "epoch": 3.185483870967742,
+      "grad_norm": 29.164669036865234,
+      "learning_rate": 7.053846976322845e-06,
+      "loss": 0.5291,
+      "step": 395
+    },
+    {
+      "epoch": 3.193548387096774,
+      "grad_norm": 16.50055694580078,
+      "learning_rate": 7.022496545316965e-06,
+      "loss": 0.0949,
+      "step": 396
+    },
+    {
+      "epoch": 3.2016129032258065,
+      "grad_norm": 11.366058349609375,
+      "learning_rate": 6.9911461143110865e-06,
+      "loss": 0.4745,
+      "step": 397
+    },
+    {
+      "epoch": 3.2096774193548385,
+      "grad_norm": 50.085636138916016,
+      "learning_rate": 6.9597956833052075e-06,
+      "loss": 1.0656,
+      "step": 398
+    },
+    {
+      "epoch": 3.217741935483871,
+      "grad_norm": 175.33628845214844,
+      "learning_rate": 6.928445252299328e-06,
+      "loss": 17.953,
+      "step": 399
+    },
+    {
+      "epoch": 3.225806451612903,
+      "grad_norm": 166.4563446044922,
+      "learning_rate": 6.897094821293449e-06,
+      "loss": 10.9431,
+      "step": 400
+    },
+    {
+      "epoch": 3.2338709677419355,
+      "grad_norm": 49.0633430480957,
+      "learning_rate": 6.865744390287569e-06,
+      "loss": 0.9611,
+      "step": 401
+    },
+    {
+      "epoch": 3.241935483870968,
+      "grad_norm": 185.53343200683594,
+      "learning_rate": 6.83439395928169e-06,
+      "loss": 5.934,
+      "step": 402
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 146.576171875,
+      "learning_rate": 6.80304352827581e-06,
+      "loss": 3.7093,
+      "step": 403
+    },
+    {
+      "epoch": 3.258064516129032,
+      "grad_norm": 9.762494087219238,
+      "learning_rate": 6.771693097269931e-06,
+      "loss": 0.125,
+      "step": 404
+    },
+    {
+      "epoch": 3.2661290322580645,
+      "grad_norm": 37.09442901611328,
+      "learning_rate": 6.740342666264052e-06,
+      "loss": 0.3642,
+      "step": 405
+    },
+    {
+      "epoch": 3.274193548387097,
+      "grad_norm": 196.52310180664062,
+      "learning_rate": 6.708992235258173e-06,
+      "loss": 14.157,
+      "step": 406
+    },
+    {
+      "epoch": 3.282258064516129,
+      "grad_norm": 576.5252075195312,
+      "learning_rate": 6.677641804252294e-06,
+      "loss": 3.4606,
+      "step": 407
+    },
+    {
+      "epoch": 3.2903225806451615,
+      "grad_norm": 70.45011901855469,
+      "learning_rate": 6.646291373246414e-06,
+      "loss": 4.9324,
+      "step": 408
+    },
+    {
+      "epoch": 3.2983870967741935,
+      "grad_norm": 37.71321487426758,
+      "learning_rate": 6.614940942240535e-06,
+      "loss": 1.5522,
+      "step": 409
+    },
+    {
+      "epoch": 3.306451612903226,
+      "grad_norm": 19.980318069458008,
+      "learning_rate": 6.583590511234655e-06,
+      "loss": 0.7383,
+      "step": 410
+    },
+    {
+      "epoch": 3.314516129032258,
+      "grad_norm": 29.11577606201172,
+      "learning_rate": 6.552240080228776e-06,
+      "loss": 0.587,
+      "step": 411
+    },
+    {
+      "epoch": 3.3225806451612905,
+      "grad_norm": 63.494163513183594,
+      "learning_rate": 6.520889649222897e-06,
+      "loss": 1.7971,
+      "step": 412
+    },
+    {
+      "epoch": 3.3306451612903225,
+      "grad_norm": 11.22016429901123,
+      "learning_rate": 6.4895392182170175e-06,
+      "loss": 0.0712,
+      "step": 413
+    },
+    {
+      "epoch": 3.338709677419355,
+      "grad_norm": 83.33367919921875,
+      "learning_rate": 6.4581887872111386e-06,
+      "loss": 1.6527,
+      "step": 414
+    },
+    {
+      "epoch": 3.346774193548387,
+      "grad_norm": 14.314624786376953,
+      "learning_rate": 6.426838356205259e-06,
+      "loss": 0.2666,
+      "step": 415
+    },
+    {
+      "epoch": 3.3548387096774195,
+      "grad_norm": 49.528526306152344,
+      "learning_rate": 6.39548792519938e-06,
+      "loss": 0.4909,
+      "step": 416
+    },
+    {
+      "epoch": 3.3629032258064515,
+      "grad_norm": 101.2572021484375,
+      "learning_rate": 6.364137494193499e-06,
+      "loss": 0.998,
+      "step": 417
+    },
+    {
+      "epoch": 3.370967741935484,
+      "grad_norm": 42.62125778198242,
+      "learning_rate": 6.332787063187621e-06,
+      "loss": 0.5947,
+      "step": 418
+    },
+    {
+      "epoch": 3.379032258064516,
+      "grad_norm": 38.499935150146484,
+      "learning_rate": 6.301436632181742e-06,
+      "loss": 0.3158,
+      "step": 419
+    },
+    {
+      "epoch": 3.3870967741935485,
+      "grad_norm": 163.30111694335938,
+      "learning_rate": 6.270086201175862e-06,
+      "loss": 6.147,
+      "step": 420
+    },
+    {
+      "epoch": 3.3951612903225805,
+      "grad_norm": 34.82032012939453,
+      "learning_rate": 6.2387357701699835e-06,
+      "loss": 0.7495,
+      "step": 421
+    },
+    {
+      "epoch": 3.403225806451613,
+      "grad_norm": 24.731409072875977,
+      "learning_rate": 6.207385339164103e-06,
+      "loss": 0.876,
+      "step": 422
+    },
+    {
+      "epoch": 3.411290322580645,
+      "grad_norm": 218.20574951171875,
+      "learning_rate": 6.176034908158225e-06,
+      "loss": 10.9933,
+      "step": 423
+    },
+    {
+      "epoch": 3.4193548387096775,
+      "grad_norm": 288.42236328125,
+      "learning_rate": 6.144684477152344e-06,
+      "loss": 16.9427,
+      "step": 424
+    },
+    {
+      "epoch": 3.4274193548387095,
+      "grad_norm": 12.138355255126953,
+      "learning_rate": 6.113334046146466e-06,
+      "loss": 0.2813,
+      "step": 425
+    },
+    {
+      "epoch": 3.435483870967742,
+      "grad_norm": 66.90897369384766,
+      "learning_rate": 6.081983615140587e-06,
+      "loss": 1.1357,
+      "step": 426
+    },
+    {
+      "epoch": 3.443548387096774,
+      "grad_norm": 32.00901412963867,
+      "learning_rate": 6.0506331841347064e-06,
+      "loss": 0.3149,
+      "step": 427
+    },
+    {
+      "epoch": 3.4516129032258065,
+      "grad_norm": 28.868900299072266,
+      "learning_rate": 6.019282753128828e-06,
+      "loss": 0.3611,
+      "step": 428
+    },
+    {
+      "epoch": 3.4596774193548385,
+      "grad_norm": 15.651339530944824,
+      "learning_rate": 5.987932322122948e-06,
+      "loss": 0.119,
+      "step": 429
+    },
+    {
+      "epoch": 3.467741935483871,
+      "grad_norm": 11.72746753692627,
+      "learning_rate": 5.956581891117069e-06,
+      "loss": 0.1911,
+      "step": 430
+    },
+    {
+      "epoch": 3.475806451612903,
+      "grad_norm": 218.60215759277344,
+      "learning_rate": 5.925231460111189e-06,
+      "loss": 9.1296,
+      "step": 431
+    },
+    {
+      "epoch": 3.4838709677419355,
+      "grad_norm": 124.23365783691406,
+      "learning_rate": 5.89388102910531e-06,
+      "loss": 3.0086,
+      "step": 432
+    },
+    {
+      "epoch": 3.491935483870968,
+      "grad_norm": 15.99737548828125,
+      "learning_rate": 5.862530598099432e-06,
+      "loss": 0.4271,
+      "step": 433
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 100.8034439086914,
+      "learning_rate": 5.831180167093551e-06,
+      "loss": 1.3045,
+      "step": 434
+    },
+    {
+      "epoch": 3.508064516129032,
+      "grad_norm": 16.803627014160156,
+      "learning_rate": 5.799829736087672e-06,
+      "loss": 0.3571,
+      "step": 435
+    },
+    {
+      "epoch": 3.5161290322580645,
+      "grad_norm": 9.734840393066406,
+      "learning_rate": 5.768479305081793e-06,
+      "loss": 0.0472,
+      "step": 436
+    },
+    {
+      "epoch": 3.524193548387097,
+      "grad_norm": 19.890609741210938,
+      "learning_rate": 5.737128874075914e-06,
+      "loss": 0.2308,
+      "step": 437
+    },
+    {
+      "epoch": 3.532258064516129,
+      "grad_norm": 59.4325065612793,
+      "learning_rate": 5.705778443070034e-06,
+      "loss": 0.8228,
+      "step": 438
+    },
+    {
+      "epoch": 3.540322580645161,
+      "grad_norm": 29.798025131225586,
+      "learning_rate": 5.674428012064155e-06,
+      "loss": 0.3622,
+      "step": 439
+    },
+    {
+      "epoch": 3.5483870967741935,
+      "grad_norm": 77.82572937011719,
+      "learning_rate": 5.643077581058276e-06,
+      "loss": 4.0208,
+      "step": 440
+    },
+    {
+      "epoch": 3.556451612903226,
+      "grad_norm": 54.52943420410156,
+      "learning_rate": 5.611727150052396e-06,
+      "loss": 0.9624,
+      "step": 441
+    },
+    {
+      "epoch": 3.564516129032258,
+      "grad_norm": 45.591339111328125,
+      "learning_rate": 5.580376719046517e-06,
+      "loss": 0.5806,
+      "step": 442
+    },
+    {
+      "epoch": 3.5725806451612905,
+      "grad_norm": 135.01614379882812,
+      "learning_rate": 5.5490262880406375e-06,
+      "loss": 1.8653,
+      "step": 443
+    },
+    {
+      "epoch": 3.5806451612903225,
+      "grad_norm": 17.62562370300293,
+      "learning_rate": 5.5176758570347585e-06,
+      "loss": 0.565,
+      "step": 444
+    },
+    {
+      "epoch": 3.588709677419355,
+      "grad_norm": 15.431583404541016,
+      "learning_rate": 5.48632542602888e-06,
+      "loss": 0.8572,
+      "step": 445
+    },
+    {
+      "epoch": 3.596774193548387,
+      "grad_norm": 228.9459991455078,
+      "learning_rate": 5.454974995023e-06,
+      "loss": 12.251,
+      "step": 446
+    },
+    {
+      "epoch": 3.6048387096774195,
+      "grad_norm": 92.89791107177734,
+      "learning_rate": 5.423624564017121e-06,
+      "loss": 6.3119,
+      "step": 447
+    },
+    {
+      "epoch": 3.6129032258064515,
+      "grad_norm": 117.79043579101562,
+      "learning_rate": 5.392274133011241e-06,
+      "loss": 1.9855,
+      "step": 448
+    },
+    {
+      "epoch": 3.620967741935484,
+      "grad_norm": 59.39741516113281,
+      "learning_rate": 5.360923702005362e-06,
+      "loss": 0.574,
+      "step": 449
+    },
+    {
+      "epoch": 3.629032258064516,
+      "grad_norm": 30.700626373291016,
+      "learning_rate": 5.329573270999482e-06,
+      "loss": 0.7384,
+      "step": 450
+    },
+    {
+      "epoch": 3.6370967741935485,
+      "grad_norm": 6.502930641174316,
+      "learning_rate": 5.298222839993603e-06,
+      "loss": 0.0293,
+      "step": 451
+    },
+    {
+      "epoch": 3.6451612903225805,
+      "grad_norm": 59.104400634765625,
+      "learning_rate": 5.2668724089877245e-06,
+      "loss": 0.5975,
+      "step": 452
+    },
+    {
+      "epoch": 3.653225806451613,
+      "grad_norm": 29.07902717590332,
+      "learning_rate": 5.235521977981845e-06,
+      "loss": 0.3861,
+      "step": 453
+    },
+    {
+      "epoch": 3.661290322580645,
+      "grad_norm": 154.84112548828125,
+      "learning_rate": 5.204171546975966e-06,
+      "loss": 10.357,
+      "step": 454
+    },
+    {
+      "epoch": 3.6693548387096775,
+      "grad_norm": 252.0116424560547,
+      "learning_rate": 5.172821115970086e-06,
+      "loss": 14.5254,
+      "step": 455
+    },
+    {
+      "epoch": 3.6774193548387095,
+      "grad_norm": 153.28329467773438,
+      "learning_rate": 5.141470684964207e-06,
+      "loss": 2.4753,
+      "step": 456
+    },
+    {
+      "epoch": 3.685483870967742,
+      "grad_norm": 22.697925567626953,
+      "learning_rate": 5.110120253958327e-06,
+      "loss": 0.1632,
+      "step": 457
+    },
+    {
+      "epoch": 3.693548387096774,
+      "grad_norm": 153.0172882080078,
+      "learning_rate": 5.078769822952448e-06,
+      "loss": 7.6211,
+      "step": 458
+    },
+    {
+      "epoch": 3.7016129032258065,
+      "grad_norm": 10.542808532714844,
+      "learning_rate": 5.047419391946569e-06,
+      "loss": 0.1378,
+      "step": 459
+    },
+    {
+      "epoch": 3.709677419354839,
+      "grad_norm": 50.96455001831055,
+      "learning_rate": 5.0160689609406896e-06,
+      "loss": 0.4241,
+      "step": 460
+    },
+    {
+      "epoch": 3.717741935483871,
+      "grad_norm": 41.19651412963867,
+      "learning_rate": 4.984718529934811e-06,
+      "loss": 0.3353,
+      "step": 461
+    },
+    {
+      "epoch": 3.725806451612903,
+      "grad_norm": 62.61945343017578,
+      "learning_rate": 4.953368098928931e-06,
+      "loss": 1.7132,
+      "step": 462
+    },
+    {
+      "epoch": 3.7338709677419355,
+      "grad_norm": 50.97965621948242,
+      "learning_rate": 4.922017667923052e-06,
+      "loss": 0.9171,
+      "step": 463
+    },
+    {
+      "epoch": 3.741935483870968,
+      "grad_norm": 171.7479248046875,
+      "learning_rate": 4.890667236917172e-06,
+      "loss": 11.7453,
+      "step": 464
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 17.77037239074707,
+      "learning_rate": 4.859316805911293e-06,
+      "loss": 0.2725,
+      "step": 465
+    },
+    {
+      "epoch": 3.758064516129032,
+      "grad_norm": 1127.8482666015625,
+      "learning_rate": 4.827966374905413e-06,
+      "loss": 116.4141,
+      "step": 466
+    },
+    {
+      "epoch": 3.7661290322580645,
+      "grad_norm": 67.08070373535156,
+      "learning_rate": 4.7966159438995345e-06,
+      "loss": 1.17,
+      "step": 467
+    },
+    {
+      "epoch": 3.774193548387097,
+      "grad_norm": 182.09410095214844,
+      "learning_rate": 4.765265512893655e-06,
+      "loss": 7.6538,
+      "step": 468
+    },
+    {
+      "epoch": 3.782258064516129,
+      "grad_norm": 53.98783493041992,
+      "learning_rate": 4.733915081887777e-06,
+      "loss": 0.8628,
+      "step": 469
+    },
+    {
+      "epoch": 3.790322580645161,
+      "grad_norm": 28.54851722717285,
+      "learning_rate": 4.702564650881897e-06,
+      "loss": 0.1568,
+      "step": 470
+    },
+    {
+      "epoch": 3.7983870967741935,
+      "grad_norm": 73.85025024414062,
+      "learning_rate": 4.671214219876017e-06,
+      "loss": 3.8978,
+      "step": 471
+    },
+    {
+      "epoch": 3.806451612903226,
+      "grad_norm": 20.776636123657227,
+      "learning_rate": 4.639863788870138e-06,
+      "loss": 0.2737,
+      "step": 472
+    },
+    {
+      "epoch": 3.814516129032258,
+      "grad_norm": 109.15306854248047,
+      "learning_rate": 4.608513357864258e-06,
+      "loss": 0.9342,
+      "step": 473
+    },
+    {
+      "epoch": 3.8225806451612905,
+      "grad_norm": 67.26899719238281,
+      "learning_rate": 4.577162926858379e-06,
+      "loss": 1.0592,
+      "step": 474
+    },
+    {
+      "epoch": 3.8306451612903225,
+      "grad_norm": 61.645511627197266,
+      "learning_rate": 4.5458124958524996e-06,
+      "loss": 1.381,
+      "step": 475
+    },
+    {
+      "epoch": 3.838709677419355,
+      "grad_norm": 357.5131530761719,
+      "learning_rate": 4.514462064846621e-06,
+      "loss": 6.9558,
+      "step": 476
+    },
+    {
+      "epoch": 3.846774193548387,
+      "grad_norm": 52.92768478393555,
+      "learning_rate": 4.483111633840742e-06,
+      "loss": 1.4706,
+      "step": 477
+    },
+    {
+      "epoch": 3.8548387096774195,
+      "grad_norm": 145.08580017089844,
+      "learning_rate": 4.451761202834862e-06,
+      "loss": 4.1149,
+      "step": 478
+    },
+    {
+      "epoch": 3.8629032258064515,
+      "grad_norm": 12.895330429077148,
+      "learning_rate": 4.420410771828983e-06,
+      "loss": 0.2032,
+      "step": 479
+    },
+    {
+      "epoch": 3.870967741935484,
+      "grad_norm": 99.95780181884766,
+      "learning_rate": 4.389060340823103e-06,
+      "loss": 3.0421,
+      "step": 480
+    },
+    {
+      "epoch": 3.879032258064516,
+      "grad_norm": 13.724507331848145,
+      "learning_rate": 4.357709909817224e-06,
+      "loss": 0.1361,
+      "step": 481
+    },
+    {
+      "epoch": 3.8870967741935485,
+      "grad_norm": 179.6150665283203,
+      "learning_rate": 4.326359478811345e-06,
+      "loss": 2.9513,
+      "step": 482
+    },
+    {
+      "epoch": 3.8951612903225805,
+      "grad_norm": 37.32778549194336,
+      "learning_rate": 4.2950090478054655e-06,
+      "loss": 1.0036,
+      "step": 483
+    },
+    {
+      "epoch": 3.903225806451613,
+      "grad_norm": 121.20023345947266,
+      "learning_rate": 4.2636586167995866e-06,
+      "loss": 12.377,
+      "step": 484
+    },
+    {
+      "epoch": 3.911290322580645,
+      "grad_norm": 77.21897888183594,
+      "learning_rate": 4.232308185793707e-06,
+      "loss": 0.9028,
+      "step": 485
+    },
+    {
+      "epoch": 3.9193548387096775,
+      "grad_norm": 28.58183479309082,
+      "learning_rate": 4.200957754787828e-06,
+      "loss": 0.2499,
+      "step": 486
+    },
+    {
+      "epoch": 3.9274193548387095,
+      "grad_norm": 40.57847595214844,
+      "learning_rate": 4.169607323781948e-06,
+      "loss": 0.5026,
+      "step": 487
+    },
+    {
+      "epoch": 3.935483870967742,
+      "grad_norm": 29.6917781829834,
+      "learning_rate": 4.138256892776069e-06,
+      "loss": 0.4407,
+      "step": 488
+    },
+    {
+      "epoch": 3.943548387096774,
+      "grad_norm": 207.91818237304688,
+      "learning_rate": 4.10690646177019e-06,
+      "loss": 8.9606,
+      "step": 489
+    },
+    {
+      "epoch": 3.9516129032258065,
+      "grad_norm": 57.388397216796875,
+      "learning_rate": 4.07555603076431e-06,
+      "loss": 1.1955,
+      "step": 490
+    },
+    {
+      "epoch": 3.959677419354839,
+      "grad_norm": 39.93169403076172,
+      "learning_rate": 4.0442055997584314e-06,
+      "loss": 0.6821,
+      "step": 491
+    },
+    {
+      "epoch": 3.967741935483871,
+      "grad_norm": 115.96873474121094,
+      "learning_rate": 4.012855168752552e-06,
+      "loss": 1.2737,
+      "step": 492
+    },
+    {
+      "epoch": 3.975806451612903,
+      "grad_norm": 4.700145721435547,
+      "learning_rate": 3.981504737746673e-06,
+      "loss": 0.1307,
+      "step": 493
+    },
+    {
+      "epoch": 3.9838709677419355,
+      "grad_norm": 226.87930297851562,
+      "learning_rate": 3.950154306740793e-06,
+      "loss": 16.7857,
+      "step": 494
+    },
+    {
+      "epoch": 3.991935483870968,
+      "grad_norm": 22.20536994934082,
+      "learning_rate": 3.918803875734914e-06,
+      "loss": 0.4253,
+      "step": 495
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 22.875497817993164,
+      "learning_rate": 3.887453444729035e-06,
+      "loss": 0.0923,
+      "step": 496
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 2.7594592571258545,
+      "eval_mae": 1.0400739908218384,
+      "eval_mse": 2.7594590187072754,
+      "eval_r2": 0.24450808763504028,
+      "eval_rmse": 1.6611619483684532,
+      "eval_runtime": 1.3594,
+      "eval_samples_per_second": 40.46,
+      "eval_smape": 53.00933122634888,
+      "eval_steps_per_second": 10.299,
+      "step": 496
+    },
+    {
+      "epoch": 4.008064516129032,
+      "grad_norm": 12.457742691040039,
+      "learning_rate": 3.856103013723155e-06,
+      "loss": 0.1641,
+      "step": 497
+    },
+    {
+      "epoch": 4.016129032258065,
+      "grad_norm": 14.920323371887207,
+      "learning_rate": 3.824752582717276e-06,
+      "loss": 0.2707,
+      "step": 498
+    },
+    {
+      "epoch": 4.024193548387097,
+      "grad_norm": 29.544240951538086,
+      "learning_rate": 3.7934021517113965e-06,
+      "loss": 0.3924,
+      "step": 499
+    },
+    {
+      "epoch": 4.032258064516129,
+      "grad_norm": 38.91734313964844,
+      "learning_rate": 3.762051720705517e-06,
+      "loss": 0.9084,
+      "step": 500
+    },
+    {
+      "epoch": 4.040322580645161,
+      "grad_norm": 31.240724563598633,
+      "learning_rate": 3.730701289699638e-06,
+      "loss": 0.5555,
+      "step": 501
+    },
+    {
+      "epoch": 4.048387096774194,
+      "grad_norm": 94.12042236328125,
+      "learning_rate": 3.6993508586937585e-06,
+      "loss": 1.486,
+      "step": 502
+    },
+    {
+      "epoch": 4.056451612903226,
+      "grad_norm": 53.03417205810547,
+      "learning_rate": 3.6680004276878795e-06,
+      "loss": 0.4751,
+      "step": 503
+    },
+    {
+      "epoch": 4.064516129032258,
+      "grad_norm": 14.563098907470703,
+      "learning_rate": 3.636649996682e-06,
+      "loss": 0.1335,
+      "step": 504
+    },
+    {
+      "epoch": 4.07258064516129,
+      "grad_norm": 12.984607696533203,
+      "learning_rate": 3.605299565676121e-06,
+      "loss": 0.2579,
+      "step": 505
+    },
+    {
+      "epoch": 4.080645161290323,
+      "grad_norm": 13.699532508850098,
+      "learning_rate": 3.5739491346702414e-06,
+      "loss": 0.2477,
+      "step": 506
+    },
+    {
+      "epoch": 4.088709677419355,
+      "grad_norm": 94.13684844970703,
+      "learning_rate": 3.542598703664362e-06,
+      "loss": 3.815,
+      "step": 507
+    },
+    {
+      "epoch": 4.096774193548387,
+      "grad_norm": 19.701648712158203,
+      "learning_rate": 3.5112482726584827e-06,
+      "loss": 0.106,
+      "step": 508
+    },
+    {
+      "epoch": 4.104838709677419,
+      "grad_norm": 29.861709594726562,
+      "learning_rate": 3.4798978416526038e-06,
+      "loss": 0.3138,
+      "step": 509
+    },
+    {
+      "epoch": 4.112903225806452,
+      "grad_norm": 55.61738967895508,
+      "learning_rate": 3.4485474106467244e-06,
+      "loss": 0.6359,
+      "step": 510
+    },
+    {
+      "epoch": 4.120967741935484,
+      "grad_norm": 93.4219970703125,
+      "learning_rate": 3.417196979640845e-06,
+      "loss": 1.7865,
+      "step": 511
+    },
+    {
+      "epoch": 4.129032258064516,
+      "grad_norm": 85.9775619506836,
+      "learning_rate": 3.3858465486349657e-06,
+      "loss": 0.8047,
+      "step": 512
+    },
+    {
+      "epoch": 4.137096774193548,
+      "grad_norm": 183.13255310058594,
+      "learning_rate": 3.3544961176290863e-06,
+      "loss": 7.5678,
+      "step": 513
+    },
+    {
+      "epoch": 4.145161290322581,
+      "grad_norm": 248.67733764648438,
+      "learning_rate": 3.323145686623207e-06,
+      "loss": 7.1086,
+      "step": 514
+    },
+    {
+      "epoch": 4.153225806451613,
+      "grad_norm": 20.28827667236328,
+      "learning_rate": 3.2917952556173276e-06,
+      "loss": 0.1826,
+      "step": 515
+    },
+    {
+      "epoch": 4.161290322580645,
+      "grad_norm": 43.15050506591797,
+      "learning_rate": 3.2604448246114486e-06,
+      "loss": 0.4589,
+      "step": 516
+    },
+    {
+      "epoch": 4.169354838709677,
+      "grad_norm": 57.90827941894531,
+      "learning_rate": 3.2290943936055693e-06,
+      "loss": 0.7951,
+      "step": 517
+    },
+    {
+      "epoch": 4.17741935483871,
+      "grad_norm": 122.35932922363281,
+      "learning_rate": 3.19774396259969e-06,
+      "loss": 3.4445,
+      "step": 518
+    },
+    {
+      "epoch": 4.185483870967742,
+      "grad_norm": 34.796329498291016,
+      "learning_rate": 3.1663935315938106e-06,
+      "loss": 1.4262,
+      "step": 519
+    },
+    {
+      "epoch": 4.193548387096774,
+      "grad_norm": 165.61900329589844,
+      "learning_rate": 3.135043100587931e-06,
+      "loss": 11.7323,
+      "step": 520
+    },
+    {
+      "epoch": 4.201612903225806,
+      "grad_norm": 22.81295394897461,
+      "learning_rate": 3.1036926695820514e-06,
+      "loss": 0.2421,
+      "step": 521
+    },
+    {
+      "epoch": 4.209677419354839,
+      "grad_norm": 29.325931549072266,
+      "learning_rate": 3.072342238576172e-06,
+      "loss": 1.1239,
+      "step": 522
+    },
+    {
+      "epoch": 4.217741935483871,
+      "grad_norm": 51.42033386230469,
+      "learning_rate": 3.0409918075702935e-06,
+      "loss": 0.7285,
+      "step": 523
+    },
+    {
+      "epoch": 4.225806451612903,
+      "grad_norm": 29.262659072875977,
+      "learning_rate": 3.009641376564414e-06,
+      "loss": 0.7601,
+      "step": 524
+    },
+    {
+      "epoch": 4.233870967741935,
+      "grad_norm": 94.11527252197266,
+      "learning_rate": 2.9782909455585344e-06,
+      "loss": 4.2293,
+      "step": 525
+    },
+    {
+      "epoch": 4.241935483870968,
+      "grad_norm": 42.665855407714844,
+      "learning_rate": 2.946940514552655e-06,
+      "loss": 0.4546,
+      "step": 526
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 40.39009475708008,
+      "learning_rate": 2.9155900835467757e-06,
+      "loss": 0.3513,
+      "step": 527
+    },
+    {
+      "epoch": 4.258064516129032,
+      "grad_norm": 11.640003204345703,
+      "learning_rate": 2.8842396525408963e-06,
+      "loss": 0.4406,
+      "step": 528
+    },
+    {
+      "epoch": 4.266129032258064,
+      "grad_norm": 102.39125061035156,
+      "learning_rate": 2.852889221535017e-06,
+      "loss": 2.7261,
+      "step": 529
+    },
+    {
+      "epoch": 4.274193548387097,
+      "grad_norm": 26.152294158935547,
+      "learning_rate": 2.821538790529138e-06,
+      "loss": 0.1906,
+      "step": 530
+    },
+    {
+      "epoch": 4.282258064516129,
+      "grad_norm": 20.90414047241211,
+      "learning_rate": 2.7901883595232586e-06,
+      "loss": 0.2463,
+      "step": 531
+    },
+    {
+      "epoch": 4.290322580645161,
+      "grad_norm": 62.55961990356445,
+      "learning_rate": 2.7588379285173793e-06,
+      "loss": 0.4813,
+      "step": 532
+    },
+    {
+      "epoch": 4.298387096774194,
+      "grad_norm": 53.30845260620117,
+      "learning_rate": 2.7274874975115e-06,
+      "loss": 1.1314,
+      "step": 533
+    },
+    {
+      "epoch": 4.306451612903226,
+      "grad_norm": 14.514229774475098,
+      "learning_rate": 2.6961370665056205e-06,
+      "loss": 0.5306,
+      "step": 534
+    },
+    {
+      "epoch": 4.314516129032258,
+      "grad_norm": 42.68838119506836,
+      "learning_rate": 2.664786635499741e-06,
+      "loss": 0.6835,
+      "step": 535
+    },
+    {
+      "epoch": 4.32258064516129,
+      "grad_norm": 52.57284164428711,
+      "learning_rate": 2.6334362044938622e-06,
+      "loss": 0.8247,
+      "step": 536
+    },
+    {
+      "epoch": 4.330645161290323,
+      "grad_norm": 65.6913070678711,
+      "learning_rate": 2.602085773487983e-06,
+      "loss": 1.043,
+      "step": 537
+    },
+    {
+      "epoch": 4.338709677419355,
+      "grad_norm": 26.719701766967773,
+      "learning_rate": 2.5707353424821035e-06,
+      "loss": 0.2364,
+      "step": 538
+    },
+    {
+      "epoch": 4.346774193548387,
+      "grad_norm": 64.74887084960938,
+      "learning_rate": 2.539384911476224e-06,
+      "loss": 0.5677,
+      "step": 539
+    },
+    {
+      "epoch": 4.354838709677419,
+      "grad_norm": 34.147438049316406,
+      "learning_rate": 2.5080344804703448e-06,
+      "loss": 0.6611,
+      "step": 540
+    },
+    {
+      "epoch": 4.362903225806452,
+      "grad_norm": 6.538976669311523,
+      "learning_rate": 2.4766840494644654e-06,
+      "loss": 0.0663,
+      "step": 541
+    },
+    {
+      "epoch": 4.370967741935484,
+      "grad_norm": 23.64634895324707,
+      "learning_rate": 2.445333618458586e-06,
+      "loss": 0.2443,
+      "step": 542
+    },
+    {
+      "epoch": 4.379032258064516,
+      "grad_norm": 42.18586349487305,
+      "learning_rate": 2.4139831874527067e-06,
+      "loss": 0.4006,
+      "step": 543
+    },
+    {
+      "epoch": 4.387096774193548,
+      "grad_norm": 213.9248504638672,
+      "learning_rate": 2.3826327564468273e-06,
+      "loss": 8.2247,
+      "step": 544
+    },
+    {
+      "epoch": 4.395161290322581,
+      "grad_norm": 7.639063835144043,
+      "learning_rate": 2.3512823254409484e-06,
+      "loss": 0.0266,
+      "step": 545
+    },
+    {
+      "epoch": 4.403225806451613,
+      "grad_norm": 13.752813339233398,
+      "learning_rate": 2.319931894435069e-06,
+      "loss": 0.1823,
+      "step": 546
+    },
+    {
+      "epoch": 4.411290322580645,
+      "grad_norm": 47.70906066894531,
+      "learning_rate": 2.2885814634291897e-06,
+      "loss": 0.9809,
+      "step": 547
+    },
+    {
+      "epoch": 4.419354838709677,
+      "grad_norm": 45.133907318115234,
+      "learning_rate": 2.2572310324233103e-06,
+      "loss": 1.2876,
+      "step": 548
+    },
+    {
+      "epoch": 4.42741935483871,
+      "grad_norm": 45.30345153808594,
+      "learning_rate": 2.225880601417431e-06,
+      "loss": 0.4589,
+      "step": 549
+    },
+    {
+      "epoch": 4.435483870967742,
+      "grad_norm": 143.42005920410156,
+      "learning_rate": 2.1945301704115516e-06,
+      "loss": 2.8116,
+      "step": 550
+    },
+    {
+      "epoch": 4.443548387096774,
+      "grad_norm": 43.46016311645508,
+      "learning_rate": 2.1631797394056726e-06,
+      "loss": 0.4703,
+      "step": 551
+    },
+    {
+      "epoch": 4.451612903225806,
+      "grad_norm": 129.6931610107422,
+      "learning_rate": 2.1318293083997933e-06,
+      "loss": 3.5121,
+      "step": 552
+    },
+    {
+      "epoch": 4.459677419354839,
+      "grad_norm": 21.558963775634766,
+      "learning_rate": 2.100478877393914e-06,
+      "loss": 0.3465,
+      "step": 553
+    },
+    {
+      "epoch": 4.467741935483871,
+      "grad_norm": 36.43219757080078,
+      "learning_rate": 2.0691284463880346e-06,
+      "loss": 0.6322,
+      "step": 554
+    },
+    {
+      "epoch": 4.475806451612903,
+      "grad_norm": 35.63532257080078,
+      "learning_rate": 2.037778015382155e-06,
+      "loss": 0.2085,
+      "step": 555
+    },
+    {
+      "epoch": 4.483870967741936,
+      "grad_norm": 19.665674209594727,
+      "learning_rate": 2.006427584376276e-06,
+      "loss": 0.1428,
+      "step": 556
+    },
+    {
+      "epoch": 4.491935483870968,
+      "grad_norm": 26.67451286315918,
+      "learning_rate": 1.9750771533703965e-06,
+      "loss": 0.2116,
+      "step": 557
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 231.28211975097656,
+      "learning_rate": 1.9437267223645175e-06,
+      "loss": 10.8706,
+      "step": 558
+    },
+    {
+      "epoch": 4.508064516129032,
+      "grad_norm": 16.69761085510254,
+      "learning_rate": 1.912376291358638e-06,
+      "loss": 0.0591,
+      "step": 559
+    },
+    {
+      "epoch": 4.516129032258064,
+      "grad_norm": 142.25442504882812,
+      "learning_rate": 1.8810258603527586e-06,
+      "loss": 6.6835,
+      "step": 560
+    },
+    {
+      "epoch": 4.524193548387097,
+      "grad_norm": 19.17723274230957,
+      "learning_rate": 1.8496754293468792e-06,
+      "loss": 0.0427,
+      "step": 561
+    },
+    {
+      "epoch": 4.532258064516129,
+      "grad_norm": 33.36142349243164,
+      "learning_rate": 1.818324998341e-06,
+      "loss": 0.5271,
+      "step": 562
+    },
+    {
+      "epoch": 4.540322580645161,
+      "grad_norm": 279.6239929199219,
+      "learning_rate": 1.7869745673351207e-06,
+      "loss": 7.1391,
+      "step": 563
+    },
+    {
+      "epoch": 4.548387096774194,
+      "grad_norm": 9.864730834960938,
+      "learning_rate": 1.7556241363292414e-06,
+      "loss": 0.2604,
+      "step": 564
+    },
+    {
+      "epoch": 4.556451612903226,
+      "grad_norm": 53.1966438293457,
+      "learning_rate": 1.7242737053233622e-06,
+      "loss": 0.9147,
+      "step": 565
+    },
+    {
+      "epoch": 4.564516129032258,
+      "grad_norm": 123.42327880859375,
+      "learning_rate": 1.6929232743174828e-06,
+      "loss": 1.8798,
+      "step": 566
+    },
+    {
+      "epoch": 4.57258064516129,
+      "grad_norm": 135.15972900390625,
+      "learning_rate": 1.6615728433116035e-06,
+      "loss": 1.6816,
+      "step": 567
+    },
+    {
+      "epoch": 4.580645161290323,
+      "grad_norm": 145.2718963623047,
+      "learning_rate": 1.6302224123057243e-06,
+      "loss": 7.8278,
+      "step": 568
+    },
+    {
+      "epoch": 4.588709677419355,
+      "grad_norm": 53.99338912963867,
+      "learning_rate": 1.598871981299845e-06,
+      "loss": 0.6349,
+      "step": 569
+    },
+    {
+      "epoch": 4.596774193548387,
+      "grad_norm": 196.27590942382812,
+      "learning_rate": 1.5675215502939656e-06,
+      "loss": 4.1074,
+      "step": 570
+    },
+    {
+      "epoch": 4.604838709677419,
+      "grad_norm": 18.781539916992188,
+      "learning_rate": 1.536171119288086e-06,
+      "loss": 0.8987,
+      "step": 571
+    },
+    {
+      "epoch": 4.612903225806452,
+      "grad_norm": 15.959029197692871,
+      "learning_rate": 1.504820688282207e-06,
+      "loss": 0.3374,
+      "step": 572
+    },
+    {
+      "epoch": 4.620967741935484,
+      "grad_norm": 77.43952178955078,
+      "learning_rate": 1.4734702572763275e-06,
+      "loss": 2.507,
+      "step": 573
+    },
+    {
+      "epoch": 4.629032258064516,
+      "grad_norm": 12.534307479858398,
+      "learning_rate": 1.4421198262704481e-06,
+      "loss": 0.07,
+      "step": 574
+    },
+    {
+      "epoch": 4.637096774193548,
+      "grad_norm": 62.35894775390625,
+      "learning_rate": 1.410769395264569e-06,
+      "loss": 1.2078,
+      "step": 575
+    },
+    {
+      "epoch": 4.645161290322581,
+      "grad_norm": 28.931297302246094,
+      "learning_rate": 1.3794189642586896e-06,
+      "loss": 0.4481,
+      "step": 576
+    },
+    {
+      "epoch": 4.653225806451613,
+      "grad_norm": 22.307531356811523,
+      "learning_rate": 1.3480685332528103e-06,
+      "loss": 0.2238,
+      "step": 577
+    },
+    {
+      "epoch": 4.661290322580645,
+      "grad_norm": 235.91934204101562,
+      "learning_rate": 1.3167181022469311e-06,
+      "loss": 12.7111,
+      "step": 578
+    },
+    {
+      "epoch": 4.669354838709677,
+      "grad_norm": 12.003568649291992,
+      "learning_rate": 1.2853676712410518e-06,
+      "loss": 0.3155,
+      "step": 579
+    },
+    {
+      "epoch": 4.67741935483871,
+      "grad_norm": 115.58039855957031,
+      "learning_rate": 1.2540172402351724e-06,
+      "loss": 2.4697,
+      "step": 580
+    },
+    {
+      "epoch": 4.685483870967742,
+      "grad_norm": 7.92896032333374,
+      "learning_rate": 1.222666809229293e-06,
+      "loss": 0.1071,
+      "step": 581
+    },
+    {
+      "epoch": 4.693548387096774,
+      "grad_norm": 62.47394943237305,
+      "learning_rate": 1.1913163782234137e-06,
+      "loss": 0.4068,
+      "step": 582
+    },
+    {
+      "epoch": 4.701612903225806,
+      "grad_norm": 64.04085540771484,
+      "learning_rate": 1.1599659472175345e-06,
+      "loss": 0.5876,
+      "step": 583
+    },
+    {
+      "epoch": 4.709677419354839,
+      "grad_norm": 167.1731719970703,
+      "learning_rate": 1.1286155162116552e-06,
+      "loss": 7.7562,
+      "step": 584
+    },
+    {
+      "epoch": 4.717741935483871,
+      "grad_norm": 110.1922836303711,
+      "learning_rate": 1.0972650852057758e-06,
+      "loss": 1.7847,
+      "step": 585
+    },
+    {
+      "epoch": 4.725806451612903,
+      "grad_norm": 144.95281982421875,
+      "learning_rate": 1.0659146541998966e-06,
+      "loss": 9.9987,
+      "step": 586
+    },
+    {
+      "epoch": 4.733870967741936,
+      "grad_norm": 26.671306610107422,
+      "learning_rate": 1.0345642231940173e-06,
+      "loss": 0.1145,
+      "step": 587
+    },
+    {
+      "epoch": 4.741935483870968,
+      "grad_norm": 14.88308048248291,
+      "learning_rate": 1.003213792188138e-06,
+      "loss": 0.2181,
+      "step": 588
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 117.62973022460938,
+      "learning_rate": 9.718633611822588e-07,
+      "loss": 1.5278,
+      "step": 589
+    },
+    {
+      "epoch": 4.758064516129032,
+      "grad_norm": 130.04318237304688,
+      "learning_rate": 9.405129301763793e-07,
+      "loss": 0.4744,
+      "step": 590
+    },
+    {
+      "epoch": 4.766129032258064,
+      "grad_norm": 184.98939514160156,
+      "learning_rate": 9.091624991705e-07,
+      "loss": 3.6068,
+      "step": 591
+    },
+    {
+      "epoch": 4.774193548387097,
+      "grad_norm": 48.318660736083984,
+      "learning_rate": 8.778120681646207e-07,
+      "loss": 0.4334,
+      "step": 592
+    },
+    {
+      "epoch": 4.782258064516129,
+      "grad_norm": 70.92566680908203,
+      "learning_rate": 8.464616371587414e-07,
+      "loss": 0.6742,
+      "step": 593
+    },
+    {
+      "epoch": 4.790322580645161,
+      "grad_norm": 48.69011306762695,
+      "learning_rate": 8.151112061528622e-07,
+      "loss": 0.403,
+      "step": 594
+    },
+    {
+      "epoch": 4.798387096774194,
+      "grad_norm": 168.84817504882812,
+      "learning_rate": 7.837607751469828e-07,
+      "loss": 2.1519,
+      "step": 595
+    },
+    {
+      "epoch": 4.806451612903226,
+      "grad_norm": 12.404576301574707,
+      "learning_rate": 7.524103441411035e-07,
+      "loss": 0.0748,
+      "step": 596
+    },
+    {
+      "epoch": 4.814516129032258,
+      "grad_norm": 11.056885719299316,
+      "learning_rate": 7.210599131352241e-07,
+      "loss": 0.1587,
+      "step": 597
+    },
+    {
+      "epoch": 4.82258064516129,
+      "grad_norm": 176.21572875976562,
+      "learning_rate": 6.897094821293448e-07,
+      "loss": 2.582,
+      "step": 598
+    },
+    {
+      "epoch": 4.830645161290323,
+      "grad_norm": 233.09788513183594,
+      "learning_rate": 6.583590511234656e-07,
+      "loss": 2.2739,
+      "step": 599
+    },
+    {
+      "epoch": 4.838709677419355,
+      "grad_norm": 64.253662109375,
+      "learning_rate": 6.270086201175862e-07,
+      "loss": 3.2137,
+      "step": 600
+    },
+    {
+      "epoch": 4.846774193548387,
+      "grad_norm": 91.62824249267578,
+      "learning_rate": 5.956581891117068e-07,
+      "loss": 1.013,
+      "step": 601
+    },
+    {
+      "epoch": 4.854838709677419,
+      "grad_norm": 8.819452285766602,
+      "learning_rate": 5.643077581058276e-07,
+      "loss": 0.1805,
+      "step": 602
+    },
+    {
+      "epoch": 4.862903225806452,
+      "grad_norm": 166.83380126953125,
+      "learning_rate": 5.329573270999483e-07,
+      "loss": 1.2785,
+      "step": 603
+    },
+    {
+      "epoch": 4.870967741935484,
+      "grad_norm": 21.492931365966797,
+      "learning_rate": 5.01606896094069e-07,
+      "loss": 0.154,
+      "step": 604
+    },
+    {
+      "epoch": 4.879032258064516,
+      "grad_norm": 50.13455581665039,
+      "learning_rate": 4.7025646508818965e-07,
+      "loss": 0.406,
+      "step": 605
+    },
+    {
+      "epoch": 4.887096774193548,
+      "grad_norm": 259.16168212890625,
+      "learning_rate": 4.3890603408231034e-07,
+      "loss": 10.8365,
+      "step": 606
+    },
+    {
+      "epoch": 4.895161290322581,
+      "grad_norm": 119.88123321533203,
+      "learning_rate": 4.075556030764311e-07,
+      "loss": 1.7068,
+      "step": 607
+    },
+    {
+      "epoch": 4.903225806451613,
+      "grad_norm": 60.6947135925293,
+      "learning_rate": 3.7620517207055177e-07,
+      "loss": 0.558,
+      "step": 608
+    },
+    {
+      "epoch": 4.911290322580645,
+      "grad_norm": 760.4828491210938,
+      "learning_rate": 3.448547410646724e-07,
+      "loss": 108.7089,
+      "step": 609
+    },
+    {
+      "epoch": 4.919354838709677,
+      "grad_norm": 27.476360321044922,
+      "learning_rate": 3.135043100587931e-07,
+      "loss": 0.1452,
+      "step": 610
+    },
+    {
+      "epoch": 4.92741935483871,
+      "grad_norm": 47.5902214050293,
+      "learning_rate": 2.821538790529138e-07,
+      "loss": 0.8969,
+      "step": 611
+    },
+    {
+      "epoch": 4.935483870967742,
+      "grad_norm": 128.73582458496094,
+      "learning_rate": 2.508034480470345e-07,
+      "loss": 9.6038,
+      "step": 612
+    },
+    {
+      "epoch": 4.943548387096774,
+      "grad_norm": 288.2920227050781,
+      "learning_rate": 2.1945301704115517e-07,
+      "loss": 5.6554,
+      "step": 613
+    },
+    {
+      "epoch": 4.951612903225806,
+      "grad_norm": 22.636474609375,
+      "learning_rate": 1.8810258603527589e-07,
+      "loss": 0.6051,
+      "step": 614
+    },
+    {
+      "epoch": 4.959677419354839,
+      "grad_norm": 17.700489044189453,
+      "learning_rate": 1.5675215502939655e-07,
+      "loss": 0.3193,
+      "step": 615
+    },
+    {
+      "epoch": 4.967741935483871,
+      "grad_norm": 20.34910774230957,
+      "learning_rate": 1.2540172402351724e-07,
+      "loss": 0.0564,
+      "step": 616
+    },
+    {
+      "epoch": 4.975806451612903,
+      "grad_norm": 21.71570587158203,
+      "learning_rate": 9.405129301763794e-08,
+      "loss": 0.1759,
+      "step": 617
+    },
+    {
+      "epoch": 4.983870967741936,
+      "grad_norm": 9.695738792419434,
+      "learning_rate": 6.270086201175862e-08,
+      "loss": 0.1146,
+      "step": 618
+    },
+    {
+      "epoch": 4.991935483870968,
+      "grad_norm": 246.5233612060547,
+      "learning_rate": 3.135043100587931e-08,
+      "loss": 10.0593,
+      "step": 619
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 9.63797378540039,
+      "learning_rate": 0.0,
+      "loss": 0.0203,
+      "step": 620
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 2.733250379562378,
+      "eval_mae": 1.0786534547805786,
+      "eval_mse": 2.7332499027252197,
+      "eval_r2": 0.25168371200561523,
+      "eval_rmse": 1.6532543369745685,
+      "eval_runtime": 1.3402,
+      "eval_samples_per_second": 41.038,
+      "eval_smape": 54.187166690826416,
+      "eval_steps_per_second": 10.446,
+      "step": 620
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 620,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "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": 3359849068769280.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": {
+    "learning_rate": 1.9437267223645173e-05
+  }
+}