diff --git "a/checkpoint-1060/trainer_state.json" "b/checkpoint-1060/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1060/trainer_state.json"
@@ -0,0 +1,7453 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.996865203761756,
+  "eval_steps": 500,
+  "global_step": 1060,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004702194357366771,
+      "grad_norm": 3.308954954147339,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 0.9758,
+      "step": 1
+    },
+    {
+      "epoch": 0.009404388714733543,
+      "grad_norm": 3.0945961475372314,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.9784,
+      "step": 2
+    },
+    {
+      "epoch": 0.014106583072100314,
+      "grad_norm": 2.6266396045684814,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 0.9407,
+      "step": 3
+    },
+    {
+      "epoch": 0.018808777429467086,
+      "grad_norm": 2.781346321105957,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.983,
+      "step": 4
+    },
+    {
+      "epoch": 0.023510971786833857,
+      "grad_norm": 2.8360986709594727,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 0.941,
+      "step": 5
+    },
+    {
+      "epoch": 0.02821316614420063,
+      "grad_norm": 2.814222812652588,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.9675,
+      "step": 6
+    },
+    {
+      "epoch": 0.032915360501567396,
+      "grad_norm": 2.821528196334839,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 0.8856,
+      "step": 7
+    },
+    {
+      "epoch": 0.03761755485893417,
+      "grad_norm": 2.7562029361724854,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.95,
+      "step": 8
+    },
+    {
+      "epoch": 0.04231974921630094,
+      "grad_norm": 2.9743621349334717,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 0.9654,
+      "step": 9
+    },
+    {
+      "epoch": 0.047021943573667714,
+      "grad_norm": 3.0444681644439697,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.9133,
+      "step": 10
+    },
+    {
+      "epoch": 0.05172413793103448,
+      "grad_norm": 2.9696431159973145,
+      "learning_rate": 5.5e-07,
+      "loss": 0.9425,
+      "step": 11
+    },
+    {
+      "epoch": 0.05642633228840126,
+      "grad_norm": 2.8478922843933105,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.9743,
+      "step": 12
+    },
+    {
+      "epoch": 0.061128526645768025,
+      "grad_norm": 2.664818525314331,
+      "learning_rate": 6.5e-07,
+      "loss": 0.9276,
+      "step": 13
+    },
+    {
+      "epoch": 0.06583072100313479,
+      "grad_norm": 2.727060556411743,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.9221,
+      "step": 14
+    },
+    {
+      "epoch": 0.07053291536050156,
+      "grad_norm": 2.8166537284851074,
+      "learning_rate": 7.5e-07,
+      "loss": 0.9406,
+      "step": 15
+    },
+    {
+      "epoch": 0.07523510971786834,
+      "grad_norm": 2.8412551879882812,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.901,
+      "step": 16
+    },
+    {
+      "epoch": 0.07993730407523511,
+      "grad_norm": 2.6464271545410156,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 0.9586,
+      "step": 17
+    },
+    {
+      "epoch": 0.08463949843260188,
+      "grad_norm": 2.506002187728882,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 0.8895,
+      "step": 18
+    },
+    {
+      "epoch": 0.08934169278996865,
+      "grad_norm": 2.663196086883545,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 0.8833,
+      "step": 19
+    },
+    {
+      "epoch": 0.09404388714733543,
+      "grad_norm": 2.5434982776641846,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.8737,
+      "step": 20
+    },
+    {
+      "epoch": 0.0987460815047022,
+      "grad_norm": 2.2520358562469482,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 0.8895,
+      "step": 21
+    },
+    {
+      "epoch": 0.10344827586206896,
+      "grad_norm": 2.3946690559387207,
+      "learning_rate": 1.1e-06,
+      "loss": 0.8405,
+      "step": 22
+    },
+    {
+      "epoch": 0.10815047021943573,
+      "grad_norm": 2.209516763687134,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 0.8441,
+      "step": 23
+    },
+    {
+      "epoch": 0.11285266457680251,
+      "grad_norm": 2.3084263801574707,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.8103,
+      "step": 24
+    },
+    {
+      "epoch": 0.11755485893416928,
+      "grad_norm": 2.4108471870422363,
+      "learning_rate": 1.25e-06,
+      "loss": 0.808,
+      "step": 25
+    },
+    {
+      "epoch": 0.12225705329153605,
+      "grad_norm": 2.116116762161255,
+      "learning_rate": 1.3e-06,
+      "loss": 0.7901,
+      "step": 26
+    },
+    {
+      "epoch": 0.12695924764890282,
+      "grad_norm": 2.2457115650177,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 0.7737,
+      "step": 27
+    },
+    {
+      "epoch": 0.13166144200626959,
+      "grad_norm": 1.9436711072921753,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.7465,
+      "step": 28
+    },
+    {
+      "epoch": 0.13636363636363635,
+      "grad_norm": 2.1678073406219482,
+      "learning_rate": 1.45e-06,
+      "loss": 0.7819,
+      "step": 29
+    },
+    {
+      "epoch": 0.14106583072100312,
+      "grad_norm": 1.8869282007217407,
+      "learning_rate": 1.5e-06,
+      "loss": 0.7653,
+      "step": 30
+    },
+    {
+      "epoch": 0.14576802507836992,
+      "grad_norm": 1.6957731246948242,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 0.7659,
+      "step": 31
+    },
+    {
+      "epoch": 0.15047021943573669,
+      "grad_norm": 1.4701424837112427,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.7553,
+      "step": 32
+    },
+    {
+      "epoch": 0.15517241379310345,
+      "grad_norm": 1.3234986066818237,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 0.6875,
+      "step": 33
+    },
+    {
+      "epoch": 0.15987460815047022,
+      "grad_norm": 1.3325831890106201,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.6681,
+      "step": 34
+    },
+    {
+      "epoch": 0.164576802507837,
+      "grad_norm": 1.3700138330459595,
+      "learning_rate": 1.75e-06,
+      "loss": 0.7635,
+      "step": 35
+    },
+    {
+      "epoch": 0.16927899686520376,
+      "grad_norm": 1.2316628694534302,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.6776,
+      "step": 36
+    },
+    {
+      "epoch": 0.17398119122257052,
+      "grad_norm": 1.170616865158081,
+      "learning_rate": 1.85e-06,
+      "loss": 0.7073,
+      "step": 37
+    },
+    {
+      "epoch": 0.1786833855799373,
+      "grad_norm": 1.1797659397125244,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.7572,
+      "step": 38
+    },
+    {
+      "epoch": 0.1833855799373041,
+      "grad_norm": 1.0123047828674316,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 0.6798,
+      "step": 39
+    },
+    {
+      "epoch": 0.18808777429467086,
+      "grad_norm": 0.975169837474823,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7088,
+      "step": 40
+    },
+    {
+      "epoch": 0.19278996865203762,
+      "grad_norm": 0.891377866268158,
+      "learning_rate": 2.05e-06,
+      "loss": 0.6696,
+      "step": 41
+    },
+    {
+      "epoch": 0.1974921630094044,
+      "grad_norm": 0.8997182250022888,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.7135,
+      "step": 42
+    },
+    {
+      "epoch": 0.20219435736677116,
+      "grad_norm": 0.9378617405891418,
+      "learning_rate": 2.15e-06,
+      "loss": 0.6827,
+      "step": 43
+    },
+    {
+      "epoch": 0.20689655172413793,
+      "grad_norm": 0.8060838580131531,
+      "learning_rate": 2.2e-06,
+      "loss": 0.6704,
+      "step": 44
+    },
+    {
+      "epoch": 0.2115987460815047,
+      "grad_norm": 0.7708892822265625,
+      "learning_rate": 2.25e-06,
+      "loss": 0.6499,
+      "step": 45
+    },
+    {
+      "epoch": 0.21630094043887146,
+      "grad_norm": 0.7236770987510681,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.657,
+      "step": 46
+    },
+    {
+      "epoch": 0.22100313479623823,
+      "grad_norm": 0.6666117906570435,
+      "learning_rate": 2.35e-06,
+      "loss": 0.634,
+      "step": 47
+    },
+    {
+      "epoch": 0.22570532915360503,
+      "grad_norm": 0.6475918292999268,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.6237,
+      "step": 48
+    },
+    {
+      "epoch": 0.2304075235109718,
+      "grad_norm": 0.7277692556381226,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.6707,
+      "step": 49
+    },
+    {
+      "epoch": 0.23510971786833856,
+      "grad_norm": 0.6572731733322144,
+      "learning_rate": 2.5e-06,
+      "loss": 0.6812,
+      "step": 50
+    },
+    {
+      "epoch": 0.23981191222570533,
+      "grad_norm": 0.6716486811637878,
+      "learning_rate": 2.55e-06,
+      "loss": 0.6803,
+      "step": 51
+    },
+    {
+      "epoch": 0.2445141065830721,
+      "grad_norm": 0.6065218448638916,
+      "learning_rate": 2.6e-06,
+      "loss": 0.6353,
+      "step": 52
+    },
+    {
+      "epoch": 0.24921630094043887,
+      "grad_norm": 0.6749820709228516,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 0.6393,
+      "step": 53
+    },
+    {
+      "epoch": 0.25391849529780564,
+      "grad_norm": 0.6187207102775574,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.6318,
+      "step": 54
+    },
+    {
+      "epoch": 0.25862068965517243,
+      "grad_norm": 0.6700596809387207,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.6762,
+      "step": 55
+    },
+    {
+      "epoch": 0.26332288401253917,
+      "grad_norm": 0.6460264921188354,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.6388,
+      "step": 56
+    },
+    {
+      "epoch": 0.26802507836990597,
+      "grad_norm": 0.7263085842132568,
+      "learning_rate": 2.85e-06,
+      "loss": 0.6413,
+      "step": 57
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 0.674784779548645,
+      "learning_rate": 2.9e-06,
+      "loss": 0.6134,
+      "step": 58
+    },
+    {
+      "epoch": 0.2774294670846395,
+      "grad_norm": 0.6552851796150208,
+      "learning_rate": 2.95e-06,
+      "loss": 0.6508,
+      "step": 59
+    },
+    {
+      "epoch": 0.28213166144200624,
+      "grad_norm": 0.7160854339599609,
+      "learning_rate": 3e-06,
+      "loss": 0.677,
+      "step": 60
+    },
+    {
+      "epoch": 0.28683385579937304,
+      "grad_norm": 0.682802677154541,
+      "learning_rate": 3.05e-06,
+      "loss": 0.5983,
+      "step": 61
+    },
+    {
+      "epoch": 0.29153605015673983,
+      "grad_norm": 0.5360945463180542,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.5928,
+      "step": 62
+    },
+    {
+      "epoch": 0.2962382445141066,
+      "grad_norm": 0.6875145435333252,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 0.6366,
+      "step": 63
+    },
+    {
+      "epoch": 0.30094043887147337,
+      "grad_norm": 0.5564191937446594,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.6466,
+      "step": 64
+    },
+    {
+      "epoch": 0.3056426332288401,
+      "grad_norm": 0.8959600925445557,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.5875,
+      "step": 65
+    },
+    {
+      "epoch": 0.3103448275862069,
+      "grad_norm": 0.5513110160827637,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.6042,
+      "step": 66
+    },
+    {
+      "epoch": 0.31504702194357365,
+      "grad_norm": 0.5638250112533569,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.6343,
+      "step": 67
+    },
+    {
+      "epoch": 0.31974921630094044,
+      "grad_norm": 0.5723473429679871,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.6007,
+      "step": 68
+    },
+    {
+      "epoch": 0.32445141065830724,
+      "grad_norm": 0.5193734765052795,
+      "learning_rate": 3.45e-06,
+      "loss": 0.5877,
+      "step": 69
+    },
+    {
+      "epoch": 0.329153605015674,
+      "grad_norm": 0.7437443733215332,
+      "learning_rate": 3.5e-06,
+      "loss": 0.6579,
+      "step": 70
+    },
+    {
+      "epoch": 0.3338557993730408,
+      "grad_norm": 0.5810403823852539,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 0.625,
+      "step": 71
+    },
+    {
+      "epoch": 0.3385579937304075,
+      "grad_norm": 0.5594515204429626,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.6039,
+      "step": 72
+    },
+    {
+      "epoch": 0.3432601880877743,
+      "grad_norm": 0.5781659483909607,
+      "learning_rate": 3.65e-06,
+      "loss": 0.5936,
+      "step": 73
+    },
+    {
+      "epoch": 0.34796238244514105,
+      "grad_norm": 0.5736168622970581,
+      "learning_rate": 3.7e-06,
+      "loss": 0.61,
+      "step": 74
+    },
+    {
+      "epoch": 0.35266457680250785,
+      "grad_norm": 0.5082181096076965,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.6031,
+      "step": 75
+    },
+    {
+      "epoch": 0.3573667711598746,
+      "grad_norm": 0.5188243985176086,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.6048,
+      "step": 76
+    },
+    {
+      "epoch": 0.3620689655172414,
+      "grad_norm": 0.678931713104248,
+      "learning_rate": 3.85e-06,
+      "loss": 0.615,
+      "step": 77
+    },
+    {
+      "epoch": 0.3667711598746082,
+      "grad_norm": 0.5580578446388245,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.5622,
+      "step": 78
+    },
+    {
+      "epoch": 0.3714733542319749,
+      "grad_norm": 0.5222985744476318,
+      "learning_rate": 3.95e-06,
+      "loss": 0.6233,
+      "step": 79
+    },
+    {
+      "epoch": 0.3761755485893417,
+      "grad_norm": 0.6105766892433167,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5448,
+      "step": 80
+    },
+    {
+      "epoch": 0.38087774294670845,
+      "grad_norm": 0.4837906062602997,
+      "learning_rate": 4.05e-06,
+      "loss": 0.6058,
+      "step": 81
+    },
+    {
+      "epoch": 0.38557993730407525,
+      "grad_norm": 0.5260081887245178,
+      "learning_rate": 4.1e-06,
+      "loss": 0.5753,
+      "step": 82
+    },
+    {
+      "epoch": 0.390282131661442,
+      "grad_norm": 0.5424107909202576,
+      "learning_rate": 4.15e-06,
+      "loss": 0.5836,
+      "step": 83
+    },
+    {
+      "epoch": 0.3949843260188088,
+      "grad_norm": 0.5307353734970093,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.5891,
+      "step": 84
+    },
+    {
+      "epoch": 0.3996865203761755,
+      "grad_norm": 0.5606550574302673,
+      "learning_rate": 4.25e-06,
+      "loss": 0.6097,
+      "step": 85
+    },
+    {
+      "epoch": 0.4043887147335423,
+      "grad_norm": 0.5380442142486572,
+      "learning_rate": 4.3e-06,
+      "loss": 0.6161,
+      "step": 86
+    },
+    {
+      "epoch": 0.4090909090909091,
+      "grad_norm": 0.7110352516174316,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 0.5681,
+      "step": 87
+    },
+    {
+      "epoch": 0.41379310344827586,
+      "grad_norm": 0.525680422782898,
+      "learning_rate": 4.4e-06,
+      "loss": 0.5942,
+      "step": 88
+    },
+    {
+      "epoch": 0.41849529780564265,
+      "grad_norm": 0.5567077398300171,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 0.5929,
+      "step": 89
+    },
+    {
+      "epoch": 0.4231974921630094,
+      "grad_norm": 0.5358735918998718,
+      "learning_rate": 4.5e-06,
+      "loss": 0.5717,
+      "step": 90
+    },
+    {
+      "epoch": 0.4278996865203762,
+      "grad_norm": 0.48081454634666443,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 0.5984,
+      "step": 91
+    },
+    {
+      "epoch": 0.43260188087774293,
+      "grad_norm": 0.4900401830673218,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.6017,
+      "step": 92
+    },
+    {
+      "epoch": 0.4373040752351097,
+      "grad_norm": 0.48369699716567993,
+      "learning_rate": 4.65e-06,
+      "loss": 0.5295,
+      "step": 93
+    },
+    {
+      "epoch": 0.44200626959247646,
+      "grad_norm": 0.4992128312587738,
+      "learning_rate": 4.7e-06,
+      "loss": 0.5495,
+      "step": 94
+    },
+    {
+      "epoch": 0.44670846394984326,
+      "grad_norm": 0.5139058828353882,
+      "learning_rate": 4.75e-06,
+      "loss": 0.5528,
+      "step": 95
+    },
+    {
+      "epoch": 0.45141065830721006,
+      "grad_norm": 0.485097199678421,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.5528,
+      "step": 96
+    },
+    {
+      "epoch": 0.4561128526645768,
+      "grad_norm": 0.5944838523864746,
+      "learning_rate": 4.85e-06,
+      "loss": 0.5577,
+      "step": 97
+    },
+    {
+      "epoch": 0.4608150470219436,
+      "grad_norm": 0.5313893556594849,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.5744,
+      "step": 98
+    },
+    {
+      "epoch": 0.46551724137931033,
+      "grad_norm": 0.49062076210975647,
+      "learning_rate": 4.95e-06,
+      "loss": 0.5563,
+      "step": 99
+    },
+    {
+      "epoch": 0.4702194357366771,
+      "grad_norm": 0.5196151733398438,
+      "learning_rate": 5e-06,
+      "loss": 0.5817,
+      "step": 100
+    },
+    {
+      "epoch": 0.47492163009404387,
+      "grad_norm": 0.507022500038147,
+      "learning_rate": 4.9999910183883085e-06,
+      "loss": 0.5608,
+      "step": 101
+    },
+    {
+      "epoch": 0.47962382445141066,
+      "grad_norm": 0.494018018245697,
+      "learning_rate": 4.999964073617768e-06,
+      "loss": 0.57,
+      "step": 102
+    },
+    {
+      "epoch": 0.4843260188087774,
+      "grad_norm": 0.48901236057281494,
+      "learning_rate": 4.999919165881985e-06,
+      "loss": 0.5568,
+      "step": 103
+    },
+    {
+      "epoch": 0.4890282131661442,
+      "grad_norm": 0.5166047811508179,
+      "learning_rate": 4.999856295503635e-06,
+      "loss": 0.5676,
+      "step": 104
+    },
+    {
+      "epoch": 0.493730407523511,
+      "grad_norm": 0.4952854812145233,
+      "learning_rate": 4.9997754629344596e-06,
+      "loss": 0.5454,
+      "step": 105
+    },
+    {
+      "epoch": 0.49843260188087773,
+      "grad_norm": 0.48603224754333496,
+      "learning_rate": 4.999676668755263e-06,
+      "loss": 0.5351,
+      "step": 106
+    },
+    {
+      "epoch": 0.5031347962382445,
+      "grad_norm": 0.4922390878200531,
+      "learning_rate": 4.999559913675912e-06,
+      "loss": 0.5672,
+      "step": 107
+    },
+    {
+      "epoch": 0.5078369905956113,
+      "grad_norm": 0.4846023619174957,
+      "learning_rate": 4.999425198535325e-06,
+      "loss": 0.5472,
+      "step": 108
+    },
+    {
+      "epoch": 0.512539184952978,
+      "grad_norm": 0.4974515736103058,
+      "learning_rate": 4.999272524301469e-06,
+      "loss": 0.5507,
+      "step": 109
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.5394945740699768,
+      "learning_rate": 4.9991018920713505e-06,
+      "loss": 0.5667,
+      "step": 110
+    },
+    {
+      "epoch": 0.5219435736677116,
+      "grad_norm": 0.5393733978271484,
+      "learning_rate": 4.9989133030710154e-06,
+      "loss": 0.5505,
+      "step": 111
+    },
+    {
+      "epoch": 0.5266457680250783,
+      "grad_norm": 0.7595535516738892,
+      "learning_rate": 4.9987067586555275e-06,
+      "loss": 0.5484,
+      "step": 112
+    },
+    {
+      "epoch": 0.5313479623824452,
+      "grad_norm": 0.5122578144073486,
+      "learning_rate": 4.998482260308969e-06,
+      "loss": 0.5669,
+      "step": 113
+    },
+    {
+      "epoch": 0.5360501567398119,
+      "grad_norm": 0.5233834981918335,
+      "learning_rate": 4.998239809644427e-06,
+      "loss": 0.5613,
+      "step": 114
+    },
+    {
+      "epoch": 0.5407523510971787,
+      "grad_norm": 3.08868408203125,
+      "learning_rate": 4.9979794084039755e-06,
+      "loss": 0.5746,
+      "step": 115
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.49716249108314514,
+      "learning_rate": 4.997701058458677e-06,
+      "loss": 0.5349,
+      "step": 116
+    },
+    {
+      "epoch": 0.5501567398119123,
+      "grad_norm": 0.5552968978881836,
+      "learning_rate": 4.997404761808554e-06,
+      "loss": 0.569,
+      "step": 117
+    },
+    {
+      "epoch": 0.554858934169279,
+      "grad_norm": 0.5029860734939575,
+      "learning_rate": 4.9970905205825845e-06,
+      "loss": 0.5755,
+      "step": 118
+    },
+    {
+      "epoch": 0.5595611285266457,
+      "grad_norm": 0.5006077885627747,
+      "learning_rate": 4.996758337038683e-06,
+      "loss": 0.5654,
+      "step": 119
+    },
+    {
+      "epoch": 0.5642633228840125,
+      "grad_norm": 0.4900127053260803,
+      "learning_rate": 4.996408213563684e-06,
+      "loss": 0.5575,
+      "step": 120
+    },
+    {
+      "epoch": 0.5689655172413793,
+      "grad_norm": 0.5142834186553955,
+      "learning_rate": 4.996040152673326e-06,
+      "loss": 0.5247,
+      "step": 121
+    },
+    {
+      "epoch": 0.5736677115987461,
+      "grad_norm": 0.5097452998161316,
+      "learning_rate": 4.995654157012233e-06,
+      "loss": 0.538,
+      "step": 122
+    },
+    {
+      "epoch": 0.5783699059561128,
+      "grad_norm": 0.5581697225570679,
+      "learning_rate": 4.995250229353895e-06,
+      "loss": 0.5324,
+      "step": 123
+    },
+    {
+      "epoch": 0.5830721003134797,
+      "grad_norm": 0.5943841934204102,
+      "learning_rate": 4.99482837260065e-06,
+      "loss": 0.547,
+      "step": 124
+    },
+    {
+      "epoch": 0.5877742946708464,
+      "grad_norm": 0.541631281375885,
+      "learning_rate": 4.99438858978366e-06,
+      "loss": 0.577,
+      "step": 125
+    },
+    {
+      "epoch": 0.5924764890282131,
+      "grad_norm": 0.6008974313735962,
+      "learning_rate": 4.993930884062892e-06,
+      "loss": 0.5579,
+      "step": 126
+    },
+    {
+      "epoch": 0.5971786833855799,
+      "grad_norm": 0.7009158730506897,
+      "learning_rate": 4.993455258727094e-06,
+      "loss": 0.5579,
+      "step": 127
+    },
+    {
+      "epoch": 0.6018808777429467,
+      "grad_norm": 0.5942498445510864,
+      "learning_rate": 4.992961717193773e-06,
+      "loss": 0.5599,
+      "step": 128
+    },
+    {
+      "epoch": 0.6065830721003135,
+      "grad_norm": 0.4936768710613251,
+      "learning_rate": 4.9924502630091655e-06,
+      "loss": 0.5676,
+      "step": 129
+    },
+    {
+      "epoch": 0.6112852664576802,
+      "grad_norm": 0.5035172700881958,
+      "learning_rate": 4.99192089984822e-06,
+      "loss": 0.5519,
+      "step": 130
+    },
+    {
+      "epoch": 0.6159874608150471,
+      "grad_norm": 0.8178919553756714,
+      "learning_rate": 4.9913736315145614e-06,
+      "loss": 0.5594,
+      "step": 131
+    },
+    {
+      "epoch": 0.6206896551724138,
+      "grad_norm": 0.4977147877216339,
+      "learning_rate": 4.990808461940474e-06,
+      "loss": 0.54,
+      "step": 132
+    },
+    {
+      "epoch": 0.6253918495297806,
+      "grad_norm": 0.7395517826080322,
+      "learning_rate": 4.990225395186862e-06,
+      "loss": 0.5467,
+      "step": 133
+    },
+    {
+      "epoch": 0.6300940438871473,
+      "grad_norm": 1.1424250602722168,
+      "learning_rate": 4.9896244354432314e-06,
+      "loss": 0.5432,
+      "step": 134
+    },
+    {
+      "epoch": 0.6347962382445141,
+      "grad_norm": 0.5002970695495605,
+      "learning_rate": 4.98900558702765e-06,
+      "loss": 0.5526,
+      "step": 135
+    },
+    {
+      "epoch": 0.6394984326018809,
+      "grad_norm": 0.49464455246925354,
+      "learning_rate": 4.9883688543867225e-06,
+      "loss": 0.563,
+      "step": 136
+    },
+    {
+      "epoch": 0.6442006269592476,
+      "grad_norm": 0.48660922050476074,
+      "learning_rate": 4.987714242095558e-06,
+      "loss": 0.5328,
+      "step": 137
+    },
+    {
+      "epoch": 0.6489028213166145,
+      "grad_norm": 0.5086668133735657,
+      "learning_rate": 4.9870417548577355e-06,
+      "loss": 0.5373,
+      "step": 138
+    },
+    {
+      "epoch": 0.6536050156739812,
+      "grad_norm": 0.4638510048389435,
+      "learning_rate": 4.9863513975052696e-06,
+      "loss": 0.5459,
+      "step": 139
+    },
+    {
+      "epoch": 0.658307210031348,
+      "grad_norm": 0.5742124915122986,
+      "learning_rate": 4.985643174998578e-06,
+      "loss": 0.5681,
+      "step": 140
+    },
+    {
+      "epoch": 0.6630094043887147,
+      "grad_norm": 0.5262806415557861,
+      "learning_rate": 4.984917092426445e-06,
+      "loss": 0.5549,
+      "step": 141
+    },
+    {
+      "epoch": 0.6677115987460815,
+      "grad_norm": 0.4765531122684479,
+      "learning_rate": 4.984173155005982e-06,
+      "loss": 0.5435,
+      "step": 142
+    },
+    {
+      "epoch": 0.6724137931034483,
+      "grad_norm": 0.4731754660606384,
+      "learning_rate": 4.983411368082597e-06,
+      "loss": 0.5484,
+      "step": 143
+    },
+    {
+      "epoch": 0.677115987460815,
+      "grad_norm": 0.5015071630477905,
+      "learning_rate": 4.982631737129948e-06,
+      "loss": 0.5314,
+      "step": 144
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 0.5013841986656189,
+      "learning_rate": 4.98183426774991e-06,
+      "loss": 0.5705,
+      "step": 145
+    },
+    {
+      "epoch": 0.6865203761755486,
+      "grad_norm": 0.5398444533348083,
+      "learning_rate": 4.981018965672529e-06,
+      "loss": 0.5713,
+      "step": 146
+    },
+    {
+      "epoch": 0.6912225705329154,
+      "grad_norm": 0.5247324109077454,
+      "learning_rate": 4.98018583675599e-06,
+      "loss": 0.5512,
+      "step": 147
+    },
+    {
+      "epoch": 0.6959247648902821,
+      "grad_norm": 0.488935649394989,
+      "learning_rate": 4.979334886986562e-06,
+      "loss": 0.5532,
+      "step": 148
+    },
+    {
+      "epoch": 0.700626959247649,
+      "grad_norm": 0.5056779384613037,
+      "learning_rate": 4.978466122478567e-06,
+      "loss": 0.5668,
+      "step": 149
+    },
+    {
+      "epoch": 0.7053291536050157,
+      "grad_norm": 0.5102982521057129,
+      "learning_rate": 4.97757954947433e-06,
+      "loss": 0.5346,
+      "step": 150
+    },
+    {
+      "epoch": 0.7100313479623824,
+      "grad_norm": 0.4989202320575714,
+      "learning_rate": 4.976675174344132e-06,
+      "loss": 0.5488,
+      "step": 151
+    },
+    {
+      "epoch": 0.7147335423197492,
+      "grad_norm": 0.4996865689754486,
+      "learning_rate": 4.975753003586172e-06,
+      "loss": 0.528,
+      "step": 152
+    },
+    {
+      "epoch": 0.719435736677116,
+      "grad_norm": 0.5011430978775024,
+      "learning_rate": 4.974813043826513e-06,
+      "loss": 0.5221,
+      "step": 153
+    },
+    {
+      "epoch": 0.7241379310344828,
+      "grad_norm": 0.530082643032074,
+      "learning_rate": 4.973855301819039e-06,
+      "loss": 0.5398,
+      "step": 154
+    },
+    {
+      "epoch": 0.7288401253918495,
+      "grad_norm": 0.5243843197822571,
+      "learning_rate": 4.972879784445402e-06,
+      "loss": 0.5215,
+      "step": 155
+    },
+    {
+      "epoch": 0.7335423197492164,
+      "grad_norm": 0.5321754217147827,
+      "learning_rate": 4.971886498714978e-06,
+      "loss": 0.5432,
+      "step": 156
+    },
+    {
+      "epoch": 0.7382445141065831,
+      "grad_norm": 0.5054258704185486,
+      "learning_rate": 4.97087545176481e-06,
+      "loss": 0.5197,
+      "step": 157
+    },
+    {
+      "epoch": 0.7429467084639498,
+      "grad_norm": 0.5209529995918274,
+      "learning_rate": 4.9698466508595655e-06,
+      "loss": 0.5486,
+      "step": 158
+    },
+    {
+      "epoch": 0.7476489028213166,
+      "grad_norm": 0.47057369351387024,
+      "learning_rate": 4.9688001033914756e-06,
+      "loss": 0.5358,
+      "step": 159
+    },
+    {
+      "epoch": 0.7523510971786834,
+      "grad_norm": 0.5286029577255249,
+      "learning_rate": 4.967735816880286e-06,
+      "loss": 0.5471,
+      "step": 160
+    },
+    {
+      "epoch": 0.7570532915360502,
+      "grad_norm": 0.7403479218482971,
+      "learning_rate": 4.966653798973205e-06,
+      "loss": 0.5264,
+      "step": 161
+    },
+    {
+      "epoch": 0.7617554858934169,
+      "grad_norm": 0.5945819020271301,
+      "learning_rate": 4.965554057444842e-06,
+      "loss": 0.5194,
+      "step": 162
+    },
+    {
+      "epoch": 0.7664576802507836,
+      "grad_norm": 0.48147913813591003,
+      "learning_rate": 4.964436600197161e-06,
+      "loss": 0.5425,
+      "step": 163
+    },
+    {
+      "epoch": 0.7711598746081505,
+      "grad_norm": 0.5892655849456787,
+      "learning_rate": 4.963301435259413e-06,
+      "loss": 0.5134,
+      "step": 164
+    },
+    {
+      "epoch": 0.7758620689655172,
+      "grad_norm": 0.546521782875061,
+      "learning_rate": 4.962148570788088e-06,
+      "loss": 0.5334,
+      "step": 165
+    },
+    {
+      "epoch": 0.780564263322884,
+      "grad_norm": 0.5432559251785278,
+      "learning_rate": 4.96097801506685e-06,
+      "loss": 0.5214,
+      "step": 166
+    },
+    {
+      "epoch": 0.7852664576802508,
+      "grad_norm": 0.4892060458660126,
+      "learning_rate": 4.959789776506482e-06,
+      "loss": 0.5307,
+      "step": 167
+    },
+    {
+      "epoch": 0.7899686520376176,
+      "grad_norm": 0.5296177268028259,
+      "learning_rate": 4.958583863644821e-06,
+      "loss": 0.5539,
+      "step": 168
+    },
+    {
+      "epoch": 0.7946708463949843,
+      "grad_norm": 0.4970422685146332,
+      "learning_rate": 4.9573602851466985e-06,
+      "loss": 0.5193,
+      "step": 169
+    },
+    {
+      "epoch": 0.799373040752351,
+      "grad_norm": 0.5201449990272522,
+      "learning_rate": 4.9561190498038815e-06,
+      "loss": 0.542,
+      "step": 170
+    },
+    {
+      "epoch": 0.8040752351097179,
+      "grad_norm": 0.6067378520965576,
+      "learning_rate": 4.954860166535005e-06,
+      "loss": 0.5319,
+      "step": 171
+    },
+    {
+      "epoch": 0.8087774294670846,
+      "grad_norm": 0.4754677712917328,
+      "learning_rate": 4.95358364438551e-06,
+      "loss": 0.5366,
+      "step": 172
+    },
+    {
+      "epoch": 0.8134796238244514,
+      "grad_norm": 0.4970039129257202,
+      "learning_rate": 4.952289492527576e-06,
+      "loss": 0.5626,
+      "step": 173
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.5602061748504639,
+      "learning_rate": 4.9509777202600605e-06,
+      "loss": 0.4951,
+      "step": 174
+    },
+    {
+      "epoch": 0.822884012539185,
+      "grad_norm": 0.5069930553436279,
+      "learning_rate": 4.949648337008425e-06,
+      "loss": 0.5419,
+      "step": 175
+    },
+    {
+      "epoch": 0.8275862068965517,
+      "grad_norm": 2.1186537742614746,
+      "learning_rate": 4.948301352324674e-06,
+      "loss": 0.5361,
+      "step": 176
+    },
+    {
+      "epoch": 0.8322884012539185,
+      "grad_norm": 0.4984792470932007,
+      "learning_rate": 4.946936775887281e-06,
+      "loss": 0.5315,
+      "step": 177
+    },
+    {
+      "epoch": 0.8369905956112853,
+      "grad_norm": 0.49590349197387695,
+      "learning_rate": 4.945554617501124e-06,
+      "loss": 0.5258,
+      "step": 178
+    },
+    {
+      "epoch": 0.841692789968652,
+      "grad_norm": 0.48108625411987305,
+      "learning_rate": 4.944154887097411e-06,
+      "loss": 0.5566,
+      "step": 179
+    },
+    {
+      "epoch": 0.8463949843260188,
+      "grad_norm": 0.5167516469955444,
+      "learning_rate": 4.942737594733608e-06,
+      "loss": 0.5285,
+      "step": 180
+    },
+    {
+      "epoch": 0.8510971786833855,
+      "grad_norm": 0.4649943709373474,
+      "learning_rate": 4.941302750593373e-06,
+      "loss": 0.5443,
+      "step": 181
+    },
+    {
+      "epoch": 0.8557993730407524,
+      "grad_norm": 0.5184140205383301,
+      "learning_rate": 4.939850364986475e-06,
+      "loss": 0.4859,
+      "step": 182
+    },
+    {
+      "epoch": 0.8605015673981191,
+      "grad_norm": 0.6746440529823303,
+      "learning_rate": 4.938380448348725e-06,
+      "loss": 0.4947,
+      "step": 183
+    },
+    {
+      "epoch": 0.8652037617554859,
+      "grad_norm": 0.5150278806686401,
+      "learning_rate": 4.9368930112419e-06,
+      "loss": 0.5351,
+      "step": 184
+    },
+    {
+      "epoch": 0.8699059561128527,
+      "grad_norm": 0.47614121437072754,
+      "learning_rate": 4.935388064353665e-06,
+      "loss": 0.5362,
+      "step": 185
+    },
+    {
+      "epoch": 0.8746081504702194,
+      "grad_norm": 0.5044179558753967,
+      "learning_rate": 4.9338656184975e-06,
+      "loss": 0.532,
+      "step": 186
+    },
+    {
+      "epoch": 0.8793103448275862,
+      "grad_norm": 0.4887774884700775,
+      "learning_rate": 4.932325684612618e-06,
+      "loss": 0.5448,
+      "step": 187
+    },
+    {
+      "epoch": 0.8840125391849529,
+      "grad_norm": 0.49574583768844604,
+      "learning_rate": 4.93076827376389e-06,
+      "loss": 0.5485,
+      "step": 188
+    },
+    {
+      "epoch": 0.8887147335423198,
+      "grad_norm": 0.5294933915138245,
+      "learning_rate": 4.9291933971417635e-06,
+      "loss": 0.5441,
+      "step": 189
+    },
+    {
+      "epoch": 0.8934169278996865,
+      "grad_norm": 0.4594694972038269,
+      "learning_rate": 4.9276010660621835e-06,
+      "loss": 0.5322,
+      "step": 190
+    },
+    {
+      "epoch": 0.8981191222570533,
+      "grad_norm": 0.47507232427597046,
+      "learning_rate": 4.925991291966508e-06,
+      "loss": 0.5103,
+      "step": 191
+    },
+    {
+      "epoch": 0.9028213166144201,
+      "grad_norm": 0.4808727204799652,
+      "learning_rate": 4.92436408642143e-06,
+      "loss": 0.5479,
+      "step": 192
+    },
+    {
+      "epoch": 0.9075235109717869,
+      "grad_norm": 0.481921523809433,
+      "learning_rate": 4.9227194611188934e-06,
+      "loss": 0.5242,
+      "step": 193
+    },
+    {
+      "epoch": 0.9122257053291536,
+      "grad_norm": 0.6387396454811096,
+      "learning_rate": 4.921057427876007e-06,
+      "loss": 0.4937,
+      "step": 194
+    },
+    {
+      "epoch": 0.9169278996865203,
+      "grad_norm": 0.5185182094573975,
+      "learning_rate": 4.919377998634959e-06,
+      "loss": 0.5515,
+      "step": 195
+    },
+    {
+      "epoch": 0.9216300940438872,
+      "grad_norm": 0.5029557943344116,
+      "learning_rate": 4.917681185462934e-06,
+      "loss": 0.5371,
+      "step": 196
+    },
+    {
+      "epoch": 0.9263322884012539,
+      "grad_norm": 0.5311322808265686,
+      "learning_rate": 4.915967000552028e-06,
+      "loss": 0.5275,
+      "step": 197
+    },
+    {
+      "epoch": 0.9310344827586207,
+      "grad_norm": 0.46653756499290466,
+      "learning_rate": 4.914235456219154e-06,
+      "loss": 0.5208,
+      "step": 198
+    },
+    {
+      "epoch": 0.9357366771159875,
+      "grad_norm": 0.4970494508743286,
+      "learning_rate": 4.912486564905959e-06,
+      "loss": 0.549,
+      "step": 199
+    },
+    {
+      "epoch": 0.9404388714733543,
+      "grad_norm": 0.5315521359443665,
+      "learning_rate": 4.910720339178735e-06,
+      "loss": 0.5314,
+      "step": 200
+    },
+    {
+      "epoch": 0.945141065830721,
+      "grad_norm": 0.5144534707069397,
+      "learning_rate": 4.908936791728323e-06,
+      "loss": 0.5372,
+      "step": 201
+    },
+    {
+      "epoch": 0.9498432601880877,
+      "grad_norm": 0.6076200008392334,
+      "learning_rate": 4.907135935370027e-06,
+      "loss": 0.5353,
+      "step": 202
+    },
+    {
+      "epoch": 0.9545454545454546,
+      "grad_norm": 0.5259833931922913,
+      "learning_rate": 4.905317783043523e-06,
+      "loss": 0.5442,
+      "step": 203
+    },
+    {
+      "epoch": 0.9592476489028213,
+      "grad_norm": 0.4931281507015228,
+      "learning_rate": 4.9034823478127605e-06,
+      "loss": 0.5246,
+      "step": 204
+    },
+    {
+      "epoch": 0.9639498432601881,
+      "grad_norm": 0.4909270405769348,
+      "learning_rate": 4.901629642865872e-06,
+      "loss": 0.5011,
+      "step": 205
+    },
+    {
+      "epoch": 0.9686520376175548,
+      "grad_norm": 0.581427276134491,
+      "learning_rate": 4.89975968151508e-06,
+      "loss": 0.5253,
+      "step": 206
+    },
+    {
+      "epoch": 0.9733542319749217,
+      "grad_norm": 0.5055932998657227,
+      "learning_rate": 4.8978724771965965e-06,
+      "loss": 0.5152,
+      "step": 207
+    },
+    {
+      "epoch": 0.9780564263322884,
+      "grad_norm": 0.5337440371513367,
+      "learning_rate": 4.895968043470532e-06,
+      "loss": 0.5321,
+      "step": 208
+    },
+    {
+      "epoch": 0.9827586206896551,
+      "grad_norm": 0.5171258449554443,
+      "learning_rate": 4.894046394020794e-06,
+      "loss": 0.4973,
+      "step": 209
+    },
+    {
+      "epoch": 0.987460815047022,
+      "grad_norm": 0.4880523085594177,
+      "learning_rate": 4.892107542654988e-06,
+      "loss": 0.529,
+      "step": 210
+    },
+    {
+      "epoch": 0.9921630094043887,
+      "grad_norm": 0.4594219923019409,
+      "learning_rate": 4.890151503304325e-06,
+      "loss": 0.5502,
+      "step": 211
+    },
+    {
+      "epoch": 0.9968652037617555,
+      "grad_norm": 0.5071013569831848,
+      "learning_rate": 4.88817829002351e-06,
+      "loss": 0.5258,
+      "step": 212
+    },
+    {
+      "epoch": 1.0047021943573669,
+      "grad_norm": 0.953562319278717,
+      "learning_rate": 4.886187916990653e-06,
+      "loss": 1.0585,
+      "step": 213
+    },
+    {
+      "epoch": 1.0094043887147335,
+      "grad_norm": 0.46260955929756165,
+      "learning_rate": 4.884180398507163e-06,
+      "loss": 0.5036,
+      "step": 214
+    },
+    {
+      "epoch": 1.0141065830721003,
+      "grad_norm": 0.46361738443374634,
+      "learning_rate": 4.882155748997636e-06,
+      "loss": 0.4975,
+      "step": 215
+    },
+    {
+      "epoch": 1.0188087774294672,
+      "grad_norm": 0.47986194491386414,
+      "learning_rate": 4.8801139830097685e-06,
+      "loss": 0.5039,
+      "step": 216
+    },
+    {
+      "epoch": 1.0235109717868338,
+      "grad_norm": 0.49743711948394775,
+      "learning_rate": 4.878055115214238e-06,
+      "loss": 0.5139,
+      "step": 217
+    },
+    {
+      "epoch": 1.0282131661442007,
+      "grad_norm": 0.5503370761871338,
+      "learning_rate": 4.875979160404607e-06,
+      "loss": 0.5117,
+      "step": 218
+    },
+    {
+      "epoch": 1.0329153605015673,
+      "grad_norm": 0.47498467564582825,
+      "learning_rate": 4.873886133497209e-06,
+      "loss": 0.5195,
+      "step": 219
+    },
+    {
+      "epoch": 1.0376175548589341,
+      "grad_norm": 0.4586140513420105,
+      "learning_rate": 4.87177604953105e-06,
+      "loss": 0.5168,
+      "step": 220
+    },
+    {
+      "epoch": 1.042319749216301,
+      "grad_norm": 0.5042305588722229,
+      "learning_rate": 4.869648923667694e-06,
+      "loss": 0.4693,
+      "step": 221
+    },
+    {
+      "epoch": 1.0470219435736676,
+      "grad_norm": 0.6340722441673279,
+      "learning_rate": 4.867504771191154e-06,
+      "loss": 0.4945,
+      "step": 222
+    },
+    {
+      "epoch": 1.0517241379310345,
+      "grad_norm": 0.4741908311843872,
+      "learning_rate": 4.865343607507788e-06,
+      "loss": 0.5027,
+      "step": 223
+    },
+    {
+      "epoch": 1.0564263322884013,
+      "grad_norm": 0.6638283133506775,
+      "learning_rate": 4.86316544814618e-06,
+      "loss": 0.5191,
+      "step": 224
+    },
+    {
+      "epoch": 1.061128526645768,
+      "grad_norm": 0.7720419764518738,
+      "learning_rate": 4.860970308757038e-06,
+      "loss": 0.4634,
+      "step": 225
+    },
+    {
+      "epoch": 1.0658307210031348,
+      "grad_norm": 0.5258729457855225,
+      "learning_rate": 4.858758205113072e-06,
+      "loss": 0.4924,
+      "step": 226
+    },
+    {
+      "epoch": 1.0705329153605017,
+      "grad_norm": 0.5001716017723083,
+      "learning_rate": 4.856529153108888e-06,
+      "loss": 0.5241,
+      "step": 227
+    },
+    {
+      "epoch": 1.0752351097178683,
+      "grad_norm": 0.5368038415908813,
+      "learning_rate": 4.854283168760868e-06,
+      "loss": 0.5031,
+      "step": 228
+    },
+    {
+      "epoch": 1.0799373040752351,
+      "grad_norm": 0.5357080101966858,
+      "learning_rate": 4.85202026820706e-06,
+      "loss": 0.5003,
+      "step": 229
+    },
+    {
+      "epoch": 1.084639498432602,
+      "grad_norm": 0.5052481293678284,
+      "learning_rate": 4.84974046770706e-06,
+      "loss": 0.5375,
+      "step": 230
+    },
+    {
+      "epoch": 1.0893416927899686,
+      "grad_norm": 0.5544849634170532,
+      "learning_rate": 4.847443783641893e-06,
+      "loss": 0.4472,
+      "step": 231
+    },
+    {
+      "epoch": 1.0940438871473355,
+      "grad_norm": 0.49668335914611816,
+      "learning_rate": 4.845130232513901e-06,
+      "loss": 0.4934,
+      "step": 232
+    },
+    {
+      "epoch": 1.098746081504702,
+      "grad_norm": 0.5318028330802917,
+      "learning_rate": 4.842799830946615e-06,
+      "loss": 0.4903,
+      "step": 233
+    },
+    {
+      "epoch": 1.103448275862069,
+      "grad_norm": 0.48978278040885925,
+      "learning_rate": 4.840452595684646e-06,
+      "loss": 0.4864,
+      "step": 234
+    },
+    {
+      "epoch": 1.1081504702194358,
+      "grad_norm": 0.47105321288108826,
+      "learning_rate": 4.83808854359356e-06,
+      "loss": 0.4805,
+      "step": 235
+    },
+    {
+      "epoch": 1.1128526645768024,
+      "grad_norm": 0.6040552854537964,
+      "learning_rate": 4.835707691659753e-06,
+      "loss": 0.4839,
+      "step": 236
+    },
+    {
+      "epoch": 1.1175548589341693,
+      "grad_norm": 0.501312255859375,
+      "learning_rate": 4.8333100569903365e-06,
+      "loss": 0.495,
+      "step": 237
+    },
+    {
+      "epoch": 1.1222570532915361,
+      "grad_norm": 0.48165130615234375,
+      "learning_rate": 4.8308956568130094e-06,
+      "loss": 0.5144,
+      "step": 238
+    },
+    {
+      "epoch": 1.1269592476489028,
+      "grad_norm": 0.5097485780715942,
+      "learning_rate": 4.828464508475934e-06,
+      "loss": 0.5081,
+      "step": 239
+    },
+    {
+      "epoch": 1.1316614420062696,
+      "grad_norm": 0.49503275752067566,
+      "learning_rate": 4.826016629447616e-06,
+      "loss": 0.5103,
+      "step": 240
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 0.5472146272659302,
+      "learning_rate": 4.823552037316775e-06,
+      "loss": 0.4857,
+      "step": 241
+    },
+    {
+      "epoch": 1.141065830721003,
+      "grad_norm": 0.5358877182006836,
+      "learning_rate": 4.821070749792218e-06,
+      "loss": 0.5418,
+      "step": 242
+    },
+    {
+      "epoch": 1.14576802507837,
+      "grad_norm": 0.4832223951816559,
+      "learning_rate": 4.818572784702713e-06,
+      "loss": 0.5121,
+      "step": 243
+    },
+    {
+      "epoch": 1.1504702194357366,
+      "grad_norm": 0.928979218006134,
+      "learning_rate": 4.816058159996863e-06,
+      "loss": 0.5218,
+      "step": 244
+    },
+    {
+      "epoch": 1.1551724137931034,
+      "grad_norm": 0.46563178300857544,
+      "learning_rate": 4.813526893742972e-06,
+      "loss": 0.5045,
+      "step": 245
+    },
+    {
+      "epoch": 1.1598746081504703,
+      "grad_norm": 0.5262567400932312,
+      "learning_rate": 4.810979004128924e-06,
+      "loss": 0.4984,
+      "step": 246
+    },
+    {
+      "epoch": 1.164576802507837,
+      "grad_norm": 0.4928685426712036,
+      "learning_rate": 4.808414509462042e-06,
+      "loss": 0.5183,
+      "step": 247
+    },
+    {
+      "epoch": 1.1692789968652038,
+      "grad_norm": 0.8799586892127991,
+      "learning_rate": 4.80583342816896e-06,
+      "loss": 0.4851,
+      "step": 248
+    },
+    {
+      "epoch": 1.1739811912225706,
+      "grad_norm": 0.5468968749046326,
+      "learning_rate": 4.803235778795496e-06,
+      "loss": 0.5242,
+      "step": 249
+    },
+    {
+      "epoch": 1.1786833855799372,
+      "grad_norm": 0.49233874678611755,
+      "learning_rate": 4.800621580006511e-06,
+      "loss": 0.4687,
+      "step": 250
+    },
+    {
+      "epoch": 1.183385579937304,
+      "grad_norm": 0.4705159366130829,
+      "learning_rate": 4.797990850585782e-06,
+      "loss": 0.5166,
+      "step": 251
+    },
+    {
+      "epoch": 1.188087774294671,
+      "grad_norm": 0.49235984683036804,
+      "learning_rate": 4.79534360943586e-06,
+      "loss": 0.4964,
+      "step": 252
+    },
+    {
+      "epoch": 1.1927899686520376,
+      "grad_norm": 0.5998479127883911,
+      "learning_rate": 4.792679875577937e-06,
+      "loss": 0.4816,
+      "step": 253
+    },
+    {
+      "epoch": 1.1974921630094044,
+      "grad_norm": 0.6658573746681213,
+      "learning_rate": 4.789999668151714e-06,
+      "loss": 0.5157,
+      "step": 254
+    },
+    {
+      "epoch": 1.2021943573667713,
+      "grad_norm": 0.47048208117485046,
+      "learning_rate": 4.7873030064152545e-06,
+      "loss": 0.4957,
+      "step": 255
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 0.47910359501838684,
+      "learning_rate": 4.784589909744856e-06,
+      "loss": 0.4906,
+      "step": 256
+    },
+    {
+      "epoch": 1.2115987460815048,
+      "grad_norm": 0.6204367876052856,
+      "learning_rate": 4.7818603976349005e-06,
+      "loss": 0.5032,
+      "step": 257
+    },
+    {
+      "epoch": 1.2163009404388714,
+      "grad_norm": 0.48487767577171326,
+      "learning_rate": 4.779114489697724e-06,
+      "loss": 0.4986,
+      "step": 258
+    },
+    {
+      "epoch": 1.2210031347962382,
+      "grad_norm": 0.5075157284736633,
+      "learning_rate": 4.776352205663469e-06,
+      "loss": 0.5021,
+      "step": 259
+    },
+    {
+      "epoch": 1.225705329153605,
+      "grad_norm": 0.4977814257144928,
+      "learning_rate": 4.773573565379947e-06,
+      "loss": 0.5132,
+      "step": 260
+    },
+    {
+      "epoch": 1.2304075235109717,
+      "grad_norm": 0.515428900718689,
+      "learning_rate": 4.770778588812489e-06,
+      "loss": 0.4783,
+      "step": 261
+    },
+    {
+      "epoch": 1.2351097178683386,
+      "grad_norm": 0.5740492939949036,
+      "learning_rate": 4.7679672960438135e-06,
+      "loss": 0.505,
+      "step": 262
+    },
+    {
+      "epoch": 1.2398119122257054,
+      "grad_norm": 0.46476778388023376,
+      "learning_rate": 4.765139707273872e-06,
+      "loss": 0.4925,
+      "step": 263
+    },
+    {
+      "epoch": 1.244514106583072,
+      "grad_norm": 0.4873107373714447,
+      "learning_rate": 4.762295842819707e-06,
+      "loss": 0.5057,
+      "step": 264
+    },
+    {
+      "epoch": 1.249216300940439,
+      "grad_norm": 0.5337681174278259,
+      "learning_rate": 4.759435723115308e-06,
+      "loss": 0.4709,
+      "step": 265
+    },
+    {
+      "epoch": 1.2539184952978055,
+      "grad_norm": 0.5014109015464783,
+      "learning_rate": 4.756559368711463e-06,
+      "loss": 0.5074,
+      "step": 266
+    },
+    {
+      "epoch": 1.2586206896551724,
+      "grad_norm": 0.4641067385673523,
+      "learning_rate": 4.75366680027561e-06,
+      "loss": 0.4882,
+      "step": 267
+    },
+    {
+      "epoch": 1.2633228840125392,
+      "grad_norm": 0.5014533996582031,
+      "learning_rate": 4.7507580385916906e-06,
+      "loss": 0.4911,
+      "step": 268
+    },
+    {
+      "epoch": 1.2680250783699059,
+      "grad_norm": 0.7282389402389526,
+      "learning_rate": 4.747833104559999e-06,
+      "loss": 0.4674,
+      "step": 269
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.44998112320899963,
+      "learning_rate": 4.744892019197033e-06,
+      "loss": 0.4818,
+      "step": 270
+    },
+    {
+      "epoch": 1.2774294670846396,
+      "grad_norm": 0.48868605494499207,
+      "learning_rate": 4.74193480363534e-06,
+      "loss": 0.4891,
+      "step": 271
+    },
+    {
+      "epoch": 1.2821316614420062,
+      "grad_norm": 0.49108532071113586,
+      "learning_rate": 4.738961479123373e-06,
+      "loss": 0.497,
+      "step": 272
+    },
+    {
+      "epoch": 1.286833855799373,
+      "grad_norm": 0.49707749485969543,
+      "learning_rate": 4.735972067025326e-06,
+      "loss": 0.5044,
+      "step": 273
+    },
+    {
+      "epoch": 1.29153605015674,
+      "grad_norm": 0.5466123223304749,
+      "learning_rate": 4.732966588820991e-06,
+      "loss": 0.4966,
+      "step": 274
+    },
+    {
+      "epoch": 1.2962382445141065,
+      "grad_norm": 0.4684564769268036,
+      "learning_rate": 4.729945066105599e-06,
+      "loss": 0.475,
+      "step": 275
+    },
+    {
+      "epoch": 1.3009404388714734,
+      "grad_norm": 0.5143007636070251,
+      "learning_rate": 4.726907520589664e-06,
+      "loss": 0.4669,
+      "step": 276
+    },
+    {
+      "epoch": 1.3056426332288402,
+      "grad_norm": 0.5112866163253784,
+      "learning_rate": 4.72385397409883e-06,
+      "loss": 0.5085,
+      "step": 277
+    },
+    {
+      "epoch": 1.3103448275862069,
+      "grad_norm": 0.5432227849960327,
+      "learning_rate": 4.720784448573712e-06,
+      "loss": 0.5003,
+      "step": 278
+    },
+    {
+      "epoch": 1.3150470219435737,
+      "grad_norm": 0.5111671686172485,
+      "learning_rate": 4.717698966069739e-06,
+      "loss": 0.5305,
+      "step": 279
+    },
+    {
+      "epoch": 1.3197492163009406,
+      "grad_norm": 0.5037922859191895,
+      "learning_rate": 4.7145975487569965e-06,
+      "loss": 0.5081,
+      "step": 280
+    },
+    {
+      "epoch": 1.3244514106583072,
+      "grad_norm": 0.7806075215339661,
+      "learning_rate": 4.711480218920064e-06,
+      "loss": 0.4698,
+      "step": 281
+    },
+    {
+      "epoch": 1.329153605015674,
+      "grad_norm": 0.5198766589164734,
+      "learning_rate": 4.708346998957859e-06,
+      "loss": 0.5188,
+      "step": 282
+    },
+    {
+      "epoch": 1.3338557993730409,
+      "grad_norm": 0.7277726531028748,
+      "learning_rate": 4.705197911383473e-06,
+      "loss": 0.4893,
+      "step": 283
+    },
+    {
+      "epoch": 1.3385579937304075,
+      "grad_norm": 0.4980437457561493,
+      "learning_rate": 4.7020329788240115e-06,
+      "loss": 0.476,
+      "step": 284
+    },
+    {
+      "epoch": 1.3432601880877744,
+      "grad_norm": 0.5318494439125061,
+      "learning_rate": 4.6988522240204325e-06,
+      "loss": 0.4853,
+      "step": 285
+    },
+    {
+      "epoch": 1.347962382445141,
+      "grad_norm": 0.5113736391067505,
+      "learning_rate": 4.695655669827377e-06,
+      "loss": 0.5005,
+      "step": 286
+    },
+    {
+      "epoch": 1.3526645768025078,
+      "grad_norm": 0.7522029876708984,
+      "learning_rate": 4.6924433392130135e-06,
+      "loss": 0.4924,
+      "step": 287
+    },
+    {
+      "epoch": 1.3573667711598745,
+      "grad_norm": 0.4861883521080017,
+      "learning_rate": 4.689215255258866e-06,
+      "loss": 0.5099,
+      "step": 288
+    },
+    {
+      "epoch": 1.3620689655172413,
+      "grad_norm": 0.46011000871658325,
+      "learning_rate": 4.685971441159653e-06,
+      "loss": 0.4792,
+      "step": 289
+    },
+    {
+      "epoch": 1.3667711598746082,
+      "grad_norm": 0.5154278874397278,
+      "learning_rate": 4.682711920223115e-06,
+      "loss": 0.4781,
+      "step": 290
+    },
+    {
+      "epoch": 1.3714733542319748,
+      "grad_norm": 0.45991218090057373,
+      "learning_rate": 4.679436715869856e-06,
+      "loss": 0.4989,
+      "step": 291
+    },
+    {
+      "epoch": 1.3761755485893417,
+      "grad_norm": 0.49241772294044495,
+      "learning_rate": 4.676145851633166e-06,
+      "loss": 0.5159,
+      "step": 292
+    },
+    {
+      "epoch": 1.3808777429467085,
+      "grad_norm": 0.523045003414154,
+      "learning_rate": 4.672839351158856e-06,
+      "loss": 0.5012,
+      "step": 293
+    },
+    {
+      "epoch": 1.3855799373040751,
+      "grad_norm": 0.5598923563957214,
+      "learning_rate": 4.669517238205089e-06,
+      "loss": 0.4855,
+      "step": 294
+    },
+    {
+      "epoch": 1.390282131661442,
+      "grad_norm": 0.507128894329071,
+      "learning_rate": 4.666179536642208e-06,
+      "loss": 0.4845,
+      "step": 295
+    },
+    {
+      "epoch": 1.3949843260188088,
+      "grad_norm": 0.721315860748291,
+      "learning_rate": 4.662826270452565e-06,
+      "loss": 0.4817,
+      "step": 296
+    },
+    {
+      "epoch": 1.3996865203761755,
+      "grad_norm": 0.560897171497345,
+      "learning_rate": 4.659457463730347e-06,
+      "loss": 0.4912,
+      "step": 297
+    },
+    {
+      "epoch": 1.4043887147335423,
+      "grad_norm": 0.5121816396713257,
+      "learning_rate": 4.6560731406814056e-06,
+      "loss": 0.5061,
+      "step": 298
+    },
+    {
+      "epoch": 1.4090909090909092,
+      "grad_norm": 0.4652915596961975,
+      "learning_rate": 4.65267332562308e-06,
+      "loss": 0.5088,
+      "step": 299
+    },
+    {
+      "epoch": 1.4137931034482758,
+      "grad_norm": 0.5149025321006775,
+      "learning_rate": 4.649258042984026e-06,
+      "loss": 0.5071,
+      "step": 300
+    },
+    {
+      "epoch": 1.4184952978056427,
+      "grad_norm": 0.4650140106678009,
+      "learning_rate": 4.6458273173040395e-06,
+      "loss": 0.4631,
+      "step": 301
+    },
+    {
+      "epoch": 1.4231974921630095,
+      "grad_norm": 0.9061777591705322,
+      "learning_rate": 4.642381173233874e-06,
+      "loss": 0.5008,
+      "step": 302
+    },
+    {
+      "epoch": 1.4278996865203761,
+      "grad_norm": 0.484729140996933,
+      "learning_rate": 4.638919635535073e-06,
+      "loss": 0.4564,
+      "step": 303
+    },
+    {
+      "epoch": 1.432601880877743,
+      "grad_norm": 1.1061334609985352,
+      "learning_rate": 4.635442729079788e-06,
+      "loss": 0.4843,
+      "step": 304
+    },
+    {
+      "epoch": 1.4373040752351098,
+      "grad_norm": 0.5028228163719177,
+      "learning_rate": 4.6319504788505956e-06,
+      "loss": 0.4794,
+      "step": 305
+    },
+    {
+      "epoch": 1.4420062695924765,
+      "grad_norm": 0.49323561787605286,
+      "learning_rate": 4.628442909940325e-06,
+      "loss": 0.4901,
+      "step": 306
+    },
+    {
+      "epoch": 1.4467084639498433,
+      "grad_norm": 0.5237039923667908,
+      "learning_rate": 4.624920047551874e-06,
+      "loss": 0.5079,
+      "step": 307
+    },
+    {
+      "epoch": 1.4514106583072102,
+      "grad_norm": 0.5407230854034424,
+      "learning_rate": 4.621381916998029e-06,
+      "loss": 0.4742,
+      "step": 308
+    },
+    {
+      "epoch": 1.4561128526645768,
+      "grad_norm": 0.4688400328159332,
+      "learning_rate": 4.6178285437012806e-06,
+      "loss": 0.5096,
+      "step": 309
+    },
+    {
+      "epoch": 1.4608150470219436,
+      "grad_norm": 0.6071098446846008,
+      "learning_rate": 4.6142599531936435e-06,
+      "loss": 0.4721,
+      "step": 310
+    },
+    {
+      "epoch": 1.4655172413793103,
+      "grad_norm": 0.47543954849243164,
+      "learning_rate": 4.610676171116475e-06,
+      "loss": 0.4924,
+      "step": 311
+    },
+    {
+      "epoch": 1.4702194357366771,
+      "grad_norm": 0.47207143902778625,
+      "learning_rate": 4.607077223220286e-06,
+      "loss": 0.4965,
+      "step": 312
+    },
+    {
+      "epoch": 1.4749216300940438,
+      "grad_norm": 0.5051419734954834,
+      "learning_rate": 4.603463135364556e-06,
+      "loss": 0.4671,
+      "step": 313
+    },
+    {
+      "epoch": 1.4796238244514106,
+      "grad_norm": 0.5427581667900085,
+      "learning_rate": 4.5998339335175555e-06,
+      "loss": 0.4896,
+      "step": 314
+    },
+    {
+      "epoch": 1.4843260188087775,
+      "grad_norm": 0.8648138642311096,
+      "learning_rate": 4.596189643756147e-06,
+      "loss": 0.4666,
+      "step": 315
+    },
+    {
+      "epoch": 1.489028213166144,
+      "grad_norm": 0.5034400224685669,
+      "learning_rate": 4.592530292265609e-06,
+      "loss": 0.4849,
+      "step": 316
+    },
+    {
+      "epoch": 1.493730407523511,
+      "grad_norm": 0.5227445960044861,
+      "learning_rate": 4.58885590533944e-06,
+      "loss": 0.4942,
+      "step": 317
+    },
+    {
+      "epoch": 1.4984326018808778,
+      "grad_norm": 0.4640982151031494,
+      "learning_rate": 4.585166509379173e-06,
+      "loss": 0.5165,
+      "step": 318
+    },
+    {
+      "epoch": 1.5031347962382444,
+      "grad_norm": 0.5094525814056396,
+      "learning_rate": 4.581462130894186e-06,
+      "loss": 0.4934,
+      "step": 319
+    },
+    {
+      "epoch": 1.5078369905956113,
+      "grad_norm": 0.5064478516578674,
+      "learning_rate": 4.57774279650151e-06,
+      "loss": 0.4847,
+      "step": 320
+    },
+    {
+      "epoch": 1.5125391849529781,
+      "grad_norm": 0.5131570100784302,
+      "learning_rate": 4.574008532925638e-06,
+      "loss": 0.5111,
+      "step": 321
+    },
+    {
+      "epoch": 1.5172413793103448,
+      "grad_norm": 0.5142849087715149,
+      "learning_rate": 4.570259366998336e-06,
+      "loss": 0.4953,
+      "step": 322
+    },
+    {
+      "epoch": 1.5219435736677116,
+      "grad_norm": 0.4943976402282715,
+      "learning_rate": 4.566495325658445e-06,
+      "loss": 0.5235,
+      "step": 323
+    },
+    {
+      "epoch": 1.5266457680250785,
+      "grad_norm": 0.46294504404067993,
+      "learning_rate": 4.5627164359516915e-06,
+      "loss": 0.505,
+      "step": 324
+    },
+    {
+      "epoch": 1.531347962382445,
+      "grad_norm": 0.6234234571456909,
+      "learning_rate": 4.558922725030491e-06,
+      "loss": 0.4776,
+      "step": 325
+    },
+    {
+      "epoch": 1.536050156739812,
+      "grad_norm": 0.5395891666412354,
+      "learning_rate": 4.555114220153755e-06,
+      "loss": 0.4313,
+      "step": 326
+    },
+    {
+      "epoch": 1.5407523510971788,
+      "grad_norm": 0.8332454562187195,
+      "learning_rate": 4.551290948686693e-06,
+      "loss": 0.5141,
+      "step": 327
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.4640675485134125,
+      "learning_rate": 4.547452938100615e-06,
+      "loss": 0.5178,
+      "step": 328
+    },
+    {
+      "epoch": 1.5501567398119123,
+      "grad_norm": 0.48269912600517273,
+      "learning_rate": 4.54360021597274e-06,
+      "loss": 0.4955,
+      "step": 329
+    },
+    {
+      "epoch": 1.5548589341692791,
+      "grad_norm": 0.5193467140197754,
+      "learning_rate": 4.539732809985989e-06,
+      "loss": 0.4864,
+      "step": 330
+    },
+    {
+      "epoch": 1.5595611285266457,
+      "grad_norm": 0.4845351576805115,
+      "learning_rate": 4.535850747928796e-06,
+      "loss": 0.5003,
+      "step": 331
+    },
+    {
+      "epoch": 1.5642633228840124,
+      "grad_norm": 0.5264540314674377,
+      "learning_rate": 4.531954057694897e-06,
+      "loss": 0.4828,
+      "step": 332
+    },
+    {
+      "epoch": 1.5689655172413794,
+      "grad_norm": 0.4795820415019989,
+      "learning_rate": 4.5280427672831414e-06,
+      "loss": 0.4904,
+      "step": 333
+    },
+    {
+      "epoch": 1.573667711598746,
+      "grad_norm": 0.5169357061386108,
+      "learning_rate": 4.524116904797281e-06,
+      "loss": 0.4661,
+      "step": 334
+    },
+    {
+      "epoch": 1.5783699059561127,
+      "grad_norm": 0.4765758514404297,
+      "learning_rate": 4.520176498445774e-06,
+      "loss": 0.4793,
+      "step": 335
+    },
+    {
+      "epoch": 1.5830721003134798,
+      "grad_norm": 0.5529272556304932,
+      "learning_rate": 4.516221576541581e-06,
+      "loss": 0.4793,
+      "step": 336
+    },
+    {
+      "epoch": 1.5877742946708464,
+      "grad_norm": 0.594944179058075,
+      "learning_rate": 4.512252167501959e-06,
+      "loss": 0.4786,
+      "step": 337
+    },
+    {
+      "epoch": 1.592476489028213,
+      "grad_norm": 0.4921863079071045,
+      "learning_rate": 4.508268299848262e-06,
+      "loss": 0.4854,
+      "step": 338
+    },
+    {
+      "epoch": 1.59717868338558,
+      "grad_norm": 0.4983494281768799,
+      "learning_rate": 4.50427000220573e-06,
+      "loss": 0.4999,
+      "step": 339
+    },
+    {
+      "epoch": 1.6018808777429467,
+      "grad_norm": 0.48530513048171997,
+      "learning_rate": 4.50025730330329e-06,
+      "loss": 0.477,
+      "step": 340
+    },
+    {
+      "epoch": 1.6065830721003134,
+      "grad_norm": 0.47819897532463074,
+      "learning_rate": 4.4962302319733445e-06,
+      "loss": 0.4933,
+      "step": 341
+    },
+    {
+      "epoch": 1.6112852664576802,
+      "grad_norm": 0.608711838722229,
+      "learning_rate": 4.492188817151565e-06,
+      "loss": 0.5285,
+      "step": 342
+    },
+    {
+      "epoch": 1.615987460815047,
+      "grad_norm": 0.5040555000305176,
+      "learning_rate": 4.488133087876688e-06,
+      "loss": 0.467,
+      "step": 343
+    },
+    {
+      "epoch": 1.6206896551724137,
+      "grad_norm": 0.4925503432750702,
+      "learning_rate": 4.484063073290301e-06,
+      "loss": 0.4651,
+      "step": 344
+    },
+    {
+      "epoch": 1.6253918495297806,
+      "grad_norm": 0.5364562273025513,
+      "learning_rate": 4.479978802636637e-06,
+      "loss": 0.5002,
+      "step": 345
+    },
+    {
+      "epoch": 1.6300940438871474,
+      "grad_norm": 0.5685595870018005,
+      "learning_rate": 4.475880305262362e-06,
+      "loss": 0.504,
+      "step": 346
+    },
+    {
+      "epoch": 1.634796238244514,
+      "grad_norm": 0.4894132912158966,
+      "learning_rate": 4.471767610616366e-06,
+      "loss": 0.4961,
+      "step": 347
+    },
+    {
+      "epoch": 1.6394984326018809,
+      "grad_norm": 0.5791006088256836,
+      "learning_rate": 4.467640748249549e-06,
+      "loss": 0.4706,
+      "step": 348
+    },
+    {
+      "epoch": 1.6442006269592477,
+      "grad_norm": 0.4814227521419525,
+      "learning_rate": 4.4634997478146125e-06,
+      "loss": 0.4895,
+      "step": 349
+    },
+    {
+      "epoch": 1.6489028213166144,
+      "grad_norm": 0.4842629134654999,
+      "learning_rate": 4.459344639065842e-06,
+      "loss": 0.482,
+      "step": 350
+    },
+    {
+      "epoch": 1.6536050156739812,
+      "grad_norm": 0.44065847992897034,
+      "learning_rate": 4.455175451858897e-06,
+      "loss": 0.4912,
+      "step": 351
+    },
+    {
+      "epoch": 1.658307210031348,
+      "grad_norm": 0.5212785601615906,
+      "learning_rate": 4.450992216150592e-06,
+      "loss": 0.4995,
+      "step": 352
+    },
+    {
+      "epoch": 1.6630094043887147,
+      "grad_norm": 0.6393409967422485,
+      "learning_rate": 4.446794961998689e-06,
+      "loss": 0.4675,
+      "step": 353
+    },
+    {
+      "epoch": 1.6677115987460815,
+      "grad_norm": 0.4786463975906372,
+      "learning_rate": 4.442583719561671e-06,
+      "loss": 0.4938,
+      "step": 354
+    },
+    {
+      "epoch": 1.6724137931034484,
+      "grad_norm": 0.5183210968971252,
+      "learning_rate": 4.438358519098536e-06,
+      "loss": 0.5032,
+      "step": 355
+    },
+    {
+      "epoch": 1.677115987460815,
+      "grad_norm": 0.7055342197418213,
+      "learning_rate": 4.4341193909685685e-06,
+      "loss": 0.4861,
+      "step": 356
+    },
+    {
+      "epoch": 1.6818181818181817,
+      "grad_norm": 0.4798656105995178,
+      "learning_rate": 4.429866365631134e-06,
+      "loss": 0.4917,
+      "step": 357
+    },
+    {
+      "epoch": 1.6865203761755487,
+      "grad_norm": 0.5035173296928406,
+      "learning_rate": 4.425599473645447e-06,
+      "loss": 0.4834,
+      "step": 358
+    },
+    {
+      "epoch": 1.6912225705329154,
+      "grad_norm": 0.5660416483879089,
+      "learning_rate": 4.421318745670364e-06,
+      "loss": 0.4829,
+      "step": 359
+    },
+    {
+      "epoch": 1.695924764890282,
+      "grad_norm": 0.7084828019142151,
+      "learning_rate": 4.4170242124641524e-06,
+      "loss": 0.4603,
+      "step": 360
+    },
+    {
+      "epoch": 1.700626959247649,
+      "grad_norm": 0.4513323903083801,
+      "learning_rate": 4.412715904884277e-06,
+      "loss": 0.4896,
+      "step": 361
+    },
+    {
+      "epoch": 1.7053291536050157,
+      "grad_norm": 0.48268529772758484,
+      "learning_rate": 4.4083938538871735e-06,
+      "loss": 0.4684,
+      "step": 362
+    },
+    {
+      "epoch": 1.7100313479623823,
+      "grad_norm": 0.4861423671245575,
+      "learning_rate": 4.4040580905280295e-06,
+      "loss": 0.4878,
+      "step": 363
+    },
+    {
+      "epoch": 1.7147335423197492,
+      "grad_norm": 0.48606055974960327,
+      "learning_rate": 4.3997086459605586e-06,
+      "loss": 0.4849,
+      "step": 364
+    },
+    {
+      "epoch": 1.719435736677116,
+      "grad_norm": 0.4912388026714325,
+      "learning_rate": 4.395345551436779e-06,
+      "loss": 0.509,
+      "step": 365
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 0.47986406087875366,
+      "learning_rate": 4.390968838306788e-06,
+      "loss": 0.4635,
+      "step": 366
+    },
+    {
+      "epoch": 1.7288401253918495,
+      "grad_norm": 0.5145572423934937,
+      "learning_rate": 4.386578538018535e-06,
+      "loss": 0.4631,
+      "step": 367
+    },
+    {
+      "epoch": 1.7335423197492164,
+      "grad_norm": 0.5538941621780396,
+      "learning_rate": 4.382174682117598e-06,
+      "loss": 0.5091,
+      "step": 368
+    },
+    {
+      "epoch": 1.738244514106583,
+      "grad_norm": 0.49918535351753235,
+      "learning_rate": 4.377757302246956e-06,
+      "loss": 0.4454,
+      "step": 369
+    },
+    {
+      "epoch": 1.7429467084639498,
+      "grad_norm": 0.48130425810813904,
+      "learning_rate": 4.373326430146762e-06,
+      "loss": 0.5011,
+      "step": 370
+    },
+    {
+      "epoch": 1.7476489028213167,
+      "grad_norm": 0.8833332657814026,
+      "learning_rate": 4.368882097654113e-06,
+      "loss": 0.4961,
+      "step": 371
+    },
+    {
+      "epoch": 1.7523510971786833,
+      "grad_norm": 0.45491841435432434,
+      "learning_rate": 4.364424336702825e-06,
+      "loss": 0.4708,
+      "step": 372
+    },
+    {
+      "epoch": 1.7570532915360502,
+      "grad_norm": 0.6433991193771362,
+      "learning_rate": 4.3599531793232e-06,
+      "loss": 0.488,
+      "step": 373
+    },
+    {
+      "epoch": 1.761755485893417,
+      "grad_norm": 0.49767547845840454,
+      "learning_rate": 4.355468657641797e-06,
+      "loss": 0.4836,
+      "step": 374
+    },
+    {
+      "epoch": 1.7664576802507836,
+      "grad_norm": 0.5577245950698853,
+      "learning_rate": 4.3509708038812035e-06,
+      "loss": 0.4879,
+      "step": 375
+    },
+    {
+      "epoch": 1.7711598746081505,
+      "grad_norm": 0.4875679314136505,
+      "learning_rate": 4.346459650359798e-06,
+      "loss": 0.4858,
+      "step": 376
+    },
+    {
+      "epoch": 1.7758620689655173,
+      "grad_norm": 0.47319531440734863,
+      "learning_rate": 4.341935229491525e-06,
+      "loss": 0.4554,
+      "step": 377
+    },
+    {
+      "epoch": 1.780564263322884,
+      "grad_norm": 0.7505528330802917,
+      "learning_rate": 4.337397573785659e-06,
+      "loss": 0.5043,
+      "step": 378
+    },
+    {
+      "epoch": 1.7852664576802508,
+      "grad_norm": 0.49199047684669495,
+      "learning_rate": 4.332846715846566e-06,
+      "loss": 0.4712,
+      "step": 379
+    },
+    {
+      "epoch": 1.7899686520376177,
+      "grad_norm": 0.5161461234092712,
+      "learning_rate": 4.328282688373479e-06,
+      "loss": 0.4927,
+      "step": 380
+    },
+    {
+      "epoch": 1.7946708463949843,
+      "grad_norm": 0.5262524485588074,
+      "learning_rate": 4.323705524160258e-06,
+      "loss": 0.4904,
+      "step": 381
+    },
+    {
+      "epoch": 1.799373040752351,
+      "grad_norm": 0.5441488027572632,
+      "learning_rate": 4.319115256095149e-06,
+      "loss": 0.4662,
+      "step": 382
+    },
+    {
+      "epoch": 1.804075235109718,
+      "grad_norm": 0.6795123815536499,
+      "learning_rate": 4.314511917160557e-06,
+      "loss": 0.4681,
+      "step": 383
+    },
+    {
+      "epoch": 1.8087774294670846,
+      "grad_norm": 0.47761398553848267,
+      "learning_rate": 4.3098955404328045e-06,
+      "loss": 0.4611,
+      "step": 384
+    },
+    {
+      "epoch": 1.8134796238244513,
+      "grad_norm": 0.7907284498214722,
+      "learning_rate": 4.305266159081895e-06,
+      "loss": 0.4826,
+      "step": 385
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.52130126953125,
+      "learning_rate": 4.3006238063712725e-06,
+      "loss": 0.4671,
+      "step": 386
+    },
+    {
+      "epoch": 1.822884012539185,
+      "grad_norm": 0.8447866439819336,
+      "learning_rate": 4.295968515657583e-06,
+      "loss": 0.5014,
+      "step": 387
+    },
+    {
+      "epoch": 1.8275862068965516,
+      "grad_norm": 0.4711958169937134,
+      "learning_rate": 4.29130032039044e-06,
+      "loss": 0.4844,
+      "step": 388
+    },
+    {
+      "epoch": 1.8322884012539185,
+      "grad_norm": 0.5012089610099792,
+      "learning_rate": 4.2866192541121755e-06,
+      "loss": 0.4747,
+      "step": 389
+    },
+    {
+      "epoch": 1.8369905956112853,
+      "grad_norm": 0.5260939598083496,
+      "learning_rate": 4.281925350457606e-06,
+      "loss": 0.4778,
+      "step": 390
+    },
+    {
+      "epoch": 1.841692789968652,
+      "grad_norm": 0.4744386672973633,
+      "learning_rate": 4.277218643153787e-06,
+      "loss": 0.4795,
+      "step": 391
+    },
+    {
+      "epoch": 1.8463949843260188,
+      "grad_norm": 0.4848107695579529,
+      "learning_rate": 4.272499166019771e-06,
+      "loss": 0.4783,
+      "step": 392
+    },
+    {
+      "epoch": 1.8510971786833856,
+      "grad_norm": 0.48980602622032166,
+      "learning_rate": 4.267766952966369e-06,
+      "loss": 0.4665,
+      "step": 393
+    },
+    {
+      "epoch": 1.8557993730407523,
+      "grad_norm": 0.49879398941993713,
+      "learning_rate": 4.2630220379959006e-06,
+      "loss": 0.444,
+      "step": 394
+    },
+    {
+      "epoch": 1.8605015673981191,
+      "grad_norm": 0.5382682681083679,
+      "learning_rate": 4.258264455201953e-06,
+      "loss": 0.466,
+      "step": 395
+    },
+    {
+      "epoch": 1.865203761755486,
+      "grad_norm": 0.47293514013290405,
+      "learning_rate": 4.2534942387691335e-06,
+      "loss": 0.4915,
+      "step": 396
+    },
+    {
+      "epoch": 1.8699059561128526,
+      "grad_norm": 0.4993828237056732,
+      "learning_rate": 4.248711422972829e-06,
+      "loss": 0.4771,
+      "step": 397
+    },
+    {
+      "epoch": 1.8746081504702194,
+      "grad_norm": 0.4950089156627655,
+      "learning_rate": 4.243916042178954e-06,
+      "loss": 0.4614,
+      "step": 398
+    },
+    {
+      "epoch": 1.8793103448275863,
+      "grad_norm": 0.48008930683135986,
+      "learning_rate": 4.239108130843709e-06,
+      "loss": 0.4684,
+      "step": 399
+    },
+    {
+      "epoch": 1.884012539184953,
+      "grad_norm": 0.5422408580780029,
+      "learning_rate": 4.234287723513326e-06,
+      "loss": 0.4933,
+      "step": 400
+    },
+    {
+      "epoch": 1.8887147335423198,
+      "grad_norm": 0.5405403971672058,
+      "learning_rate": 4.229454854823827e-06,
+      "loss": 0.4912,
+      "step": 401
+    },
+    {
+      "epoch": 1.8934169278996866,
+      "grad_norm": 0.6024321913719177,
+      "learning_rate": 4.224609559500772e-06,
+      "loss": 0.504,
+      "step": 402
+    },
+    {
+      "epoch": 1.8981191222570533,
+      "grad_norm": 0.4791870415210724,
+      "learning_rate": 4.21975187235901e-06,
+      "loss": 0.4547,
+      "step": 403
+    },
+    {
+      "epoch": 1.90282131661442,
+      "grad_norm": 0.9783840775489807,
+      "learning_rate": 4.21488182830243e-06,
+      "loss": 0.483,
+      "step": 404
+    },
+    {
+      "epoch": 1.907523510971787,
+      "grad_norm": 1.0899996757507324,
+      "learning_rate": 4.209999462323706e-06,
+      "loss": 0.4603,
+      "step": 405
+    },
+    {
+      "epoch": 1.9122257053291536,
+      "grad_norm": 0.6221780180931091,
+      "learning_rate": 4.20510480950405e-06,
+      "loss": 0.4926,
+      "step": 406
+    },
+    {
+      "epoch": 1.9169278996865202,
+      "grad_norm": 0.671617329120636,
+      "learning_rate": 4.200197905012961e-06,
+      "loss": 0.4536,
+      "step": 407
+    },
+    {
+      "epoch": 1.9216300940438873,
+      "grad_norm": 0.48824113607406616,
+      "learning_rate": 4.195278784107965e-06,
+      "loss": 0.4709,
+      "step": 408
+    },
+    {
+      "epoch": 1.926332288401254,
+      "grad_norm": 0.4700049161911011,
+      "learning_rate": 4.19034748213437e-06,
+      "loss": 0.4721,
+      "step": 409
+    },
+    {
+      "epoch": 1.9310344827586206,
+      "grad_norm": 0.5228457450866699,
+      "learning_rate": 4.185404034525008e-06,
+      "loss": 0.4644,
+      "step": 410
+    },
+    {
+      "epoch": 1.9357366771159876,
+      "grad_norm": 0.5192087888717651,
+      "learning_rate": 4.180448476799981e-06,
+      "loss": 0.5018,
+      "step": 411
+    },
+    {
+      "epoch": 1.9404388714733543,
+      "grad_norm": 0.49069511890411377,
+      "learning_rate": 4.175480844566404e-06,
+      "loss": 0.4747,
+      "step": 412
+    },
+    {
+      "epoch": 1.9451410658307209,
+      "grad_norm": 0.45442187786102295,
+      "learning_rate": 4.170501173518152e-06,
+      "loss": 0.4683,
+      "step": 413
+    },
+    {
+      "epoch": 1.9498432601880877,
+      "grad_norm": 0.5145336985588074,
+      "learning_rate": 4.165509499435604e-06,
+      "loss": 0.4677,
+      "step": 414
+    },
+    {
+      "epoch": 1.9545454545454546,
+      "grad_norm": 0.514231264591217,
+      "learning_rate": 4.16050585818538e-06,
+      "loss": 0.4812,
+      "step": 415
+    },
+    {
+      "epoch": 1.9592476489028212,
+      "grad_norm": 0.5052164793014526,
+      "learning_rate": 4.155490285720092e-06,
+      "loss": 0.5056,
+      "step": 416
+    },
+    {
+      "epoch": 1.963949843260188,
+      "grad_norm": 0.5336954593658447,
+      "learning_rate": 4.150462818078079e-06,
+      "loss": 0.4916,
+      "step": 417
+    },
+    {
+      "epoch": 1.968652037617555,
+      "grad_norm": 0.5100321769714355,
+      "learning_rate": 4.145423491383153e-06,
+      "loss": 0.4869,
+      "step": 418
+    },
+    {
+      "epoch": 1.9733542319749215,
+      "grad_norm": 0.5358601212501526,
+      "learning_rate": 4.14037234184433e-06,
+      "loss": 0.5041,
+      "step": 419
+    },
+    {
+      "epoch": 1.9780564263322884,
+      "grad_norm": 0.4950590133666992,
+      "learning_rate": 4.135309405755583e-06,
+      "loss": 0.4889,
+      "step": 420
+    },
+    {
+      "epoch": 1.9827586206896552,
+      "grad_norm": 0.5121294856071472,
+      "learning_rate": 4.130234719495574e-06,
+      "loss": 0.4736,
+      "step": 421
+    },
+    {
+      "epoch": 1.9874608150470219,
+      "grad_norm": 0.5014020800590515,
+      "learning_rate": 4.125148319527391e-06,
+      "loss": 0.4635,
+      "step": 422
+    },
+    {
+      "epoch": 1.9921630094043887,
+      "grad_norm": 0.5898067355155945,
+      "learning_rate": 4.1200502423982904e-06,
+      "loss": 0.4842,
+      "step": 423
+    },
+    {
+      "epoch": 1.9968652037617556,
+      "grad_norm": 0.535094141960144,
+      "learning_rate": 4.1149405247394295e-06,
+      "loss": 0.4885,
+      "step": 424
+    },
+    {
+      "epoch": 2.0047021943573666,
+      "grad_norm": 1.0523930788040161,
+      "learning_rate": 4.10981920326561e-06,
+      "loss": 0.9153,
+      "step": 425
+    },
+    {
+      "epoch": 2.0094043887147337,
+      "grad_norm": 0.46556714177131653,
+      "learning_rate": 4.104686314775009e-06,
+      "loss": 0.4418,
+      "step": 426
+    },
+    {
+      "epoch": 2.0141065830721003,
+      "grad_norm": 0.6543925404548645,
+      "learning_rate": 4.099541896148914e-06,
+      "loss": 0.4504,
+      "step": 427
+    },
+    {
+      "epoch": 2.018808777429467,
+      "grad_norm": 0.5143531560897827,
+      "learning_rate": 4.094385984351462e-06,
+      "loss": 0.4452,
+      "step": 428
+    },
+    {
+      "epoch": 2.023510971786834,
+      "grad_norm": 0.5021682977676392,
+      "learning_rate": 4.0892186164293715e-06,
+      "loss": 0.4655,
+      "step": 429
+    },
+    {
+      "epoch": 2.0282131661442007,
+      "grad_norm": 0.5309224128723145,
+      "learning_rate": 4.0840398295116745e-06,
+      "loss": 0.4531,
+      "step": 430
+    },
+    {
+      "epoch": 2.0329153605015673,
+      "grad_norm": 0.48278284072875977,
+      "learning_rate": 4.078849660809456e-06,
+      "loss": 0.4487,
+      "step": 431
+    },
+    {
+      "epoch": 2.0376175548589344,
+      "grad_norm": 0.6067396998405457,
+      "learning_rate": 4.073648147615579e-06,
+      "loss": 0.4315,
+      "step": 432
+    },
+    {
+      "epoch": 2.042319749216301,
+      "grad_norm": 0.48894888162612915,
+      "learning_rate": 4.068435327304421e-06,
+      "loss": 0.4573,
+      "step": 433
+    },
+    {
+      "epoch": 2.0470219435736676,
+      "grad_norm": 0.5110018253326416,
+      "learning_rate": 4.063211237331603e-06,
+      "loss": 0.4555,
+      "step": 434
+    },
+    {
+      "epoch": 2.0517241379310347,
+      "grad_norm": 0.6234620809555054,
+      "learning_rate": 4.057975915233725e-06,
+      "loss": 0.4391,
+      "step": 435
+    },
+    {
+      "epoch": 2.0564263322884013,
+      "grad_norm": 0.4931652843952179,
+      "learning_rate": 4.052729398628089e-06,
+      "loss": 0.4654,
+      "step": 436
+    },
+    {
+      "epoch": 2.061128526645768,
+      "grad_norm": 0.4912821650505066,
+      "learning_rate": 4.047471725212437e-06,
+      "loss": 0.4622,
+      "step": 437
+    },
+    {
+      "epoch": 2.0658307210031346,
+      "grad_norm": 0.575827956199646,
+      "learning_rate": 4.042202932764673e-06,
+      "loss": 0.4466,
+      "step": 438
+    },
+    {
+      "epoch": 2.0705329153605017,
+      "grad_norm": 0.4869770109653473,
+      "learning_rate": 4.036923059142595e-06,
+      "loss": 0.4487,
+      "step": 439
+    },
+    {
+      "epoch": 2.0752351097178683,
+      "grad_norm": 0.4875548779964447,
+      "learning_rate": 4.031632142283623e-06,
+      "loss": 0.4431,
+      "step": 440
+    },
+    {
+      "epoch": 2.079937304075235,
+      "grad_norm": 0.5250763893127441,
+      "learning_rate": 4.026330220204524e-06,
+      "loss": 0.4508,
+      "step": 441
+    },
+    {
+      "epoch": 2.084639498432602,
+      "grad_norm": 0.49760839343070984,
+      "learning_rate": 4.021017331001146e-06,
+      "loss": 0.4418,
+      "step": 442
+    },
+    {
+      "epoch": 2.0893416927899686,
+      "grad_norm": 0.5596722364425659,
+      "learning_rate": 4.015693512848131e-06,
+      "loss": 0.4491,
+      "step": 443
+    },
+    {
+      "epoch": 2.0940438871473352,
+      "grad_norm": 0.5050379037857056,
+      "learning_rate": 4.0103588039986556e-06,
+      "loss": 0.452,
+      "step": 444
+    },
+    {
+      "epoch": 2.0987460815047023,
+      "grad_norm": 0.5045352578163147,
+      "learning_rate": 4.005013242784146e-06,
+      "loss": 0.4606,
+      "step": 445
+    },
+    {
+      "epoch": 2.103448275862069,
+      "grad_norm": 0.568103015422821,
+      "learning_rate": 3.999656867614006e-06,
+      "loss": 0.4492,
+      "step": 446
+    },
+    {
+      "epoch": 2.1081504702194356,
+      "grad_norm": 0.48106372356414795,
+      "learning_rate": 3.994289716975341e-06,
+      "loss": 0.4634,
+      "step": 447
+    },
+    {
+      "epoch": 2.1128526645768027,
+      "grad_norm": 0.6244045495986938,
+      "learning_rate": 3.988911829432682e-06,
+      "loss": 0.4492,
+      "step": 448
+    },
+    {
+      "epoch": 2.1175548589341693,
+      "grad_norm": 0.514492392539978,
+      "learning_rate": 3.983523243627706e-06,
+      "loss": 0.4464,
+      "step": 449
+    },
+    {
+      "epoch": 2.122257053291536,
+      "grad_norm": 0.53345787525177,
+      "learning_rate": 3.978123998278962e-06,
+      "loss": 0.4363,
+      "step": 450
+    },
+    {
+      "epoch": 2.126959247648903,
+      "grad_norm": 0.675532877445221,
+      "learning_rate": 3.97271413218159e-06,
+      "loss": 0.4651,
+      "step": 451
+    },
+    {
+      "epoch": 2.1316614420062696,
+      "grad_norm": 0.473951518535614,
+      "learning_rate": 3.9672936842070425e-06,
+      "loss": 0.4627,
+      "step": 452
+    },
+    {
+      "epoch": 2.1363636363636362,
+      "grad_norm": 0.4946623742580414,
+      "learning_rate": 3.9618626933028086e-06,
+      "loss": 0.4664,
+      "step": 453
+    },
+    {
+      "epoch": 2.1410658307210033,
+      "grad_norm": 0.574018120765686,
+      "learning_rate": 3.956421198492128e-06,
+      "loss": 0.4482,
+      "step": 454
+    },
+    {
+      "epoch": 2.14576802507837,
+      "grad_norm": 0.5015276074409485,
+      "learning_rate": 3.950969238873714e-06,
+      "loss": 0.4473,
+      "step": 455
+    },
+    {
+      "epoch": 2.1504702194357366,
+      "grad_norm": 0.4798969030380249,
+      "learning_rate": 3.9455068536214765e-06,
+      "loss": 0.4763,
+      "step": 456
+    },
+    {
+      "epoch": 2.1551724137931036,
+      "grad_norm": 0.5117236375808716,
+      "learning_rate": 3.9400340819842335e-06,
+      "loss": 0.4562,
+      "step": 457
+    },
+    {
+      "epoch": 2.1598746081504703,
+      "grad_norm": 0.6655581593513489,
+      "learning_rate": 3.934550963285432e-06,
+      "loss": 0.4569,
+      "step": 458
+    },
+    {
+      "epoch": 2.164576802507837,
+      "grad_norm": 0.5741562843322754,
+      "learning_rate": 3.9290575369228664e-06,
+      "loss": 0.4292,
+      "step": 459
+    },
+    {
+      "epoch": 2.169278996865204,
+      "grad_norm": 0.5728350281715393,
+      "learning_rate": 3.923553842368396e-06,
+      "loss": 0.4686,
+      "step": 460
+    },
+    {
+      "epoch": 2.1739811912225706,
+      "grad_norm": 0.47696346044540405,
+      "learning_rate": 3.918039919167658e-06,
+      "loss": 0.4344,
+      "step": 461
+    },
+    {
+      "epoch": 2.1786833855799372,
+      "grad_norm": 0.5124625563621521,
+      "learning_rate": 3.912515806939786e-06,
+      "loss": 0.4483,
+      "step": 462
+    },
+    {
+      "epoch": 2.183385579937304,
+      "grad_norm": 1.2648731470108032,
+      "learning_rate": 3.906981545377124e-06,
+      "loss": 0.4613,
+      "step": 463
+    },
+    {
+      "epoch": 2.188087774294671,
+      "grad_norm": 0.5044319033622742,
+      "learning_rate": 3.901437174244943e-06,
+      "loss": 0.4298,
+      "step": 464
+    },
+    {
+      "epoch": 2.1927899686520376,
+      "grad_norm": 0.6312819123268127,
+      "learning_rate": 3.895882733381154e-06,
+      "loss": 0.4471,
+      "step": 465
+    },
+    {
+      "epoch": 2.197492163009404,
+      "grad_norm": 0.5056606531143188,
+      "learning_rate": 3.890318262696023e-06,
+      "loss": 0.4663,
+      "step": 466
+    },
+    {
+      "epoch": 2.2021943573667713,
+      "grad_norm": 0.4930380582809448,
+      "learning_rate": 3.8847438021718805e-06,
+      "loss": 0.4419,
+      "step": 467
+    },
+    {
+      "epoch": 2.206896551724138,
+      "grad_norm": 0.5264694094657898,
+      "learning_rate": 3.879159391862839e-06,
+      "loss": 0.4636,
+      "step": 468
+    },
+    {
+      "epoch": 2.2115987460815045,
+      "grad_norm": 0.517235279083252,
+      "learning_rate": 3.873565071894503e-06,
+      "loss": 0.4354,
+      "step": 469
+    },
+    {
+      "epoch": 2.2163009404388716,
+      "grad_norm": 0.5416566729545593,
+      "learning_rate": 3.86796088246368e-06,
+      "loss": 0.4317,
+      "step": 470
+    },
+    {
+      "epoch": 2.2210031347962382,
+      "grad_norm": 0.5619626045227051,
+      "learning_rate": 3.8623468638380905e-06,
+      "loss": 0.4155,
+      "step": 471
+    },
+    {
+      "epoch": 2.225705329153605,
+      "grad_norm": 0.5955201983451843,
+      "learning_rate": 3.856723056356085e-06,
+      "loss": 0.4586,
+      "step": 472
+    },
+    {
+      "epoch": 2.230407523510972,
+      "grad_norm": 0.6263668537139893,
+      "learning_rate": 3.851089500426346e-06,
+      "loss": 0.4459,
+      "step": 473
+    },
+    {
+      "epoch": 2.2351097178683386,
+      "grad_norm": 0.525097131729126,
+      "learning_rate": 3.845446236527605e-06,
+      "loss": 0.4451,
+      "step": 474
+    },
+    {
+      "epoch": 2.239811912225705,
+      "grad_norm": 0.48124319314956665,
+      "learning_rate": 3.8397933052083445e-06,
+      "loss": 0.4626,
+      "step": 475
+    },
+    {
+      "epoch": 2.2445141065830723,
+      "grad_norm": 0.5204625129699707,
+      "learning_rate": 3.834130747086512e-06,
+      "loss": 0.4481,
+      "step": 476
+    },
+    {
+      "epoch": 2.249216300940439,
+      "grad_norm": 0.5445339679718018,
+      "learning_rate": 3.828458602849226e-06,
+      "loss": 0.4484,
+      "step": 477
+    },
+    {
+      "epoch": 2.2539184952978055,
+      "grad_norm": 0.550906240940094,
+      "learning_rate": 3.822776913252485e-06,
+      "loss": 0.4505,
+      "step": 478
+    },
+    {
+      "epoch": 2.2586206896551726,
+      "grad_norm": 2.449854850769043,
+      "learning_rate": 3.817085719120872e-06,
+      "loss": 0.402,
+      "step": 479
+    },
+    {
+      "epoch": 2.2633228840125392,
+      "grad_norm": 0.5691458582878113,
+      "learning_rate": 3.811385061347263e-06,
+      "loss": 0.4457,
+      "step": 480
+    },
+    {
+      "epoch": 2.268025078369906,
+      "grad_norm": 0.6042170524597168,
+      "learning_rate": 3.805674980892535e-06,
+      "loss": 0.4703,
+      "step": 481
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.577621579170227,
+      "learning_rate": 3.7999555187852667e-06,
+      "loss": 0.458,
+      "step": 482
+    },
+    {
+      "epoch": 2.2774294670846396,
+      "grad_norm": 0.4807332754135132,
+      "learning_rate": 3.7942267161214497e-06,
+      "loss": 0.4454,
+      "step": 483
+    },
+    {
+      "epoch": 2.282131661442006,
+      "grad_norm": 0.5056536793708801,
+      "learning_rate": 3.7884886140641884e-06,
+      "loss": 0.4684,
+      "step": 484
+    },
+    {
+      "epoch": 2.2868338557993733,
+      "grad_norm": 0.9068124294281006,
+      "learning_rate": 3.7827412538434062e-06,
+      "loss": 0.4322,
+      "step": 485
+    },
+    {
+      "epoch": 2.29153605015674,
+      "grad_norm": 0.49843674898147583,
+      "learning_rate": 3.7769846767555495e-06,
+      "loss": 0.4594,
+      "step": 486
+    },
+    {
+      "epoch": 2.2962382445141065,
+      "grad_norm": 0.5128474235534668,
+      "learning_rate": 3.7712189241632898e-06,
+      "loss": 0.4549,
+      "step": 487
+    },
+    {
+      "epoch": 2.300940438871473,
+      "grad_norm": 0.5473759770393372,
+      "learning_rate": 3.7654440374952288e-06,
+      "loss": 0.4421,
+      "step": 488
+    },
+    {
+      "epoch": 2.30564263322884,
+      "grad_norm": 0.5089896321296692,
+      "learning_rate": 3.7596600582455976e-06,
+      "loss": 0.4191,
+      "step": 489
+    },
+    {
+      "epoch": 2.310344827586207,
+      "grad_norm": 0.5044212937355042,
+      "learning_rate": 3.75386702797396e-06,
+      "loss": 0.4206,
+      "step": 490
+    },
+    {
+      "epoch": 2.3150470219435735,
+      "grad_norm": 0.46690526604652405,
+      "learning_rate": 3.7480649883049164e-06,
+      "loss": 0.4257,
+      "step": 491
+    },
+    {
+      "epoch": 2.3197492163009406,
+      "grad_norm": 0.514756441116333,
+      "learning_rate": 3.7422539809277993e-06,
+      "loss": 0.4363,
+      "step": 492
+    },
+    {
+      "epoch": 2.324451410658307,
+      "grad_norm": 0.5000742077827454,
+      "learning_rate": 3.736434047596379e-06,
+      "loss": 0.4438,
+      "step": 493
+    },
+    {
+      "epoch": 2.329153605015674,
+      "grad_norm": 0.5082759857177734,
+      "learning_rate": 3.73060523012856e-06,
+      "loss": 0.4504,
+      "step": 494
+    },
+    {
+      "epoch": 2.333855799373041,
+      "grad_norm": 0.538390576839447,
+      "learning_rate": 3.724767570406082e-06,
+      "loss": 0.4685,
+      "step": 495
+    },
+    {
+      "epoch": 2.3385579937304075,
+      "grad_norm": 0.4855618476867676,
+      "learning_rate": 3.7189211103742206e-06,
+      "loss": 0.4257,
+      "step": 496
+    },
+    {
+      "epoch": 2.343260188087774,
+      "grad_norm": 0.5083416104316711,
+      "learning_rate": 3.7130658920414818e-06,
+      "loss": 0.4608,
+      "step": 497
+    },
+    {
+      "epoch": 2.347962382445141,
+      "grad_norm": 0.47180086374282837,
+      "learning_rate": 3.7072019574793034e-06,
+      "loss": 0.4726,
+      "step": 498
+    },
+    {
+      "epoch": 2.352664576802508,
+      "grad_norm": 0.5352308750152588,
+      "learning_rate": 3.701329348821752e-06,
+      "loss": 0.4525,
+      "step": 499
+    },
+    {
+      "epoch": 2.3573667711598745,
+      "grad_norm": 0.49689581990242004,
+      "learning_rate": 3.695448108265221e-06,
+      "loss": 0.4378,
+      "step": 500
+    },
+    {
+      "epoch": 2.3620689655172415,
+      "grad_norm": 0.4953297972679138,
+      "learning_rate": 3.6895582780681254e-06,
+      "loss": 0.4344,
+      "step": 501
+    },
+    {
+      "epoch": 2.366771159874608,
+      "grad_norm": 0.5224218368530273,
+      "learning_rate": 3.683659900550598e-06,
+      "loss": 0.4632,
+      "step": 502
+    },
+    {
+      "epoch": 2.371473354231975,
+      "grad_norm": 0.5154619812965393,
+      "learning_rate": 3.6777530180941894e-06,
+      "loss": 0.4449,
+      "step": 503
+    },
+    {
+      "epoch": 2.376175548589342,
+      "grad_norm": 0.48938247561454773,
+      "learning_rate": 3.671837673141559e-06,
+      "loss": 0.4314,
+      "step": 504
+    },
+    {
+      "epoch": 2.3808777429467085,
+      "grad_norm": 0.5059546828269958,
+      "learning_rate": 3.6659139081961707e-06,
+      "loss": 0.4475,
+      "step": 505
+    },
+    {
+      "epoch": 2.385579937304075,
+      "grad_norm": 0.6195068359375,
+      "learning_rate": 3.6599817658219916e-06,
+      "loss": 0.4492,
+      "step": 506
+    },
+    {
+      "epoch": 2.3902821316614418,
+      "grad_norm": 0.5169727206230164,
+      "learning_rate": 3.6540412886431796e-06,
+      "loss": 0.4612,
+      "step": 507
+    },
+    {
+      "epoch": 2.394984326018809,
+      "grad_norm": 1.0481178760528564,
+      "learning_rate": 3.648092519343783e-06,
+      "loss": 0.4446,
+      "step": 508
+    },
+    {
+      "epoch": 2.3996865203761755,
+      "grad_norm": 0.49299949407577515,
+      "learning_rate": 3.642135500667431e-06,
+      "loss": 0.4318,
+      "step": 509
+    },
+    {
+      "epoch": 2.4043887147335425,
+      "grad_norm": 0.508884072303772,
+      "learning_rate": 3.6361702754170247e-06,
+      "loss": 0.4476,
+      "step": 510
+    },
+    {
+      "epoch": 2.409090909090909,
+      "grad_norm": 0.4841139018535614,
+      "learning_rate": 3.630196886454435e-06,
+      "loss": 0.4294,
+      "step": 511
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 0.49162107706069946,
+      "learning_rate": 3.62421537670019e-06,
+      "loss": 0.4449,
+      "step": 512
+    },
+    {
+      "epoch": 2.4184952978056424,
+      "grad_norm": 0.5381420254707336,
+      "learning_rate": 3.618225789133167e-06,
+      "loss": 0.4472,
+      "step": 513
+    },
+    {
+      "epoch": 2.4231974921630095,
+      "grad_norm": 0.4767305552959442,
+      "learning_rate": 3.612228166790287e-06,
+      "loss": 0.4464,
+      "step": 514
+    },
+    {
+      "epoch": 2.427899686520376,
+      "grad_norm": 0.5070378184318542,
+      "learning_rate": 3.606222552766201e-06,
+      "loss": 0.4555,
+      "step": 515
+    },
+    {
+      "epoch": 2.4326018808777428,
+      "grad_norm": 0.503158688545227,
+      "learning_rate": 3.6002089902129844e-06,
+      "loss": 0.4484,
+      "step": 516
+    },
+    {
+      "epoch": 2.43730407523511,
+      "grad_norm": 0.7062323093414307,
+      "learning_rate": 3.5941875223398225e-06,
+      "loss": 0.4394,
+      "step": 517
+    },
+    {
+      "epoch": 2.4420062695924765,
+      "grad_norm": 0.48649832606315613,
+      "learning_rate": 3.588158192412707e-06,
+      "loss": 0.4354,
+      "step": 518
+    },
+    {
+      "epoch": 2.446708463949843,
+      "grad_norm": 0.49017950892448425,
+      "learning_rate": 3.582121043754116e-06,
+      "loss": 0.4387,
+      "step": 519
+    },
+    {
+      "epoch": 2.45141065830721,
+      "grad_norm": 0.4829757809638977,
+      "learning_rate": 3.5760761197427097e-06,
+      "loss": 0.4377,
+      "step": 520
+    },
+    {
+      "epoch": 2.456112852664577,
+      "grad_norm": 0.482098251581192,
+      "learning_rate": 3.570023463813017e-06,
+      "loss": 0.4297,
+      "step": 521
+    },
+    {
+      "epoch": 2.4608150470219434,
+      "grad_norm": 0.678365170955658,
+      "learning_rate": 3.5639631194551216e-06,
+      "loss": 0.4537,
+      "step": 522
+    },
+    {
+      "epoch": 2.4655172413793105,
+      "grad_norm": 0.5783978700637817,
+      "learning_rate": 3.557895130214352e-06,
+      "loss": 0.4443,
+      "step": 523
+    },
+    {
+      "epoch": 2.470219435736677,
+      "grad_norm": 0.47321340441703796,
+      "learning_rate": 3.5518195396909653e-06,
+      "loss": 0.4633,
+      "step": 524
+    },
+    {
+      "epoch": 2.4749216300940438,
+      "grad_norm": 0.5123101472854614,
+      "learning_rate": 3.5457363915398384e-06,
+      "loss": 0.4529,
+      "step": 525
+    },
+    {
+      "epoch": 2.479623824451411,
+      "grad_norm": 0.4978135824203491,
+      "learning_rate": 3.539645729470151e-06,
+      "loss": 0.4389,
+      "step": 526
+    },
+    {
+      "epoch": 2.4843260188087775,
+      "grad_norm": 0.5567356944084167,
+      "learning_rate": 3.5335475972450715e-06,
+      "loss": 0.4381,
+      "step": 527
+    },
+    {
+      "epoch": 2.489028213166144,
+      "grad_norm": 1.5688557624816895,
+      "learning_rate": 3.5274420386814458e-06,
+      "loss": 0.4652,
+      "step": 528
+    },
+    {
+      "epoch": 2.493730407523511,
+      "grad_norm": 0.6167457103729248,
+      "learning_rate": 3.521329097649478e-06,
+      "loss": 0.446,
+      "step": 529
+    },
+    {
+      "epoch": 2.498432601880878,
+      "grad_norm": 0.5396067500114441,
+      "learning_rate": 3.515208818072418e-06,
+      "loss": 0.443,
+      "step": 530
+    },
+    {
+      "epoch": 2.5031347962382444,
+      "grad_norm": 0.5053077936172485,
+      "learning_rate": 3.509081243926247e-06,
+      "loss": 0.4313,
+      "step": 531
+    },
+    {
+      "epoch": 2.507836990595611,
+      "grad_norm": 0.655543863773346,
+      "learning_rate": 3.5029464192393557e-06,
+      "loss": 0.4622,
+      "step": 532
+    },
+    {
+      "epoch": 2.512539184952978,
+      "grad_norm": 0.6995306611061096,
+      "learning_rate": 3.4968043880922363e-06,
+      "loss": 0.4152,
+      "step": 533
+    },
+    {
+      "epoch": 2.5172413793103448,
+      "grad_norm": 0.5307276248931885,
+      "learning_rate": 3.4906551946171603e-06,
+      "loss": 0.424,
+      "step": 534
+    },
+    {
+      "epoch": 2.521943573667712,
+      "grad_norm": 0.5830240249633789,
+      "learning_rate": 3.484498882997861e-06,
+      "loss": 0.4226,
+      "step": 535
+    },
+    {
+      "epoch": 2.5266457680250785,
+      "grad_norm": 0.7151938080787659,
+      "learning_rate": 3.478335497469219e-06,
+      "loss": 0.4514,
+      "step": 536
+    },
+    {
+      "epoch": 2.531347962382445,
+      "grad_norm": 0.5270131230354309,
+      "learning_rate": 3.472165082316943e-06,
+      "loss": 0.4493,
+      "step": 537
+    },
+    {
+      "epoch": 2.5360501567398117,
+      "grad_norm": 0.48484402894973755,
+      "learning_rate": 3.465987681877251e-06,
+      "loss": 0.4431,
+      "step": 538
+    },
+    {
+      "epoch": 2.540752351097179,
+      "grad_norm": 0.5078386664390564,
+      "learning_rate": 3.4598033405365527e-06,
+      "loss": 0.4559,
+      "step": 539
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 0.5258911848068237,
+      "learning_rate": 3.45361210273113e-06,
+      "loss": 0.448,
+      "step": 540
+    },
+    {
+      "epoch": 2.5501567398119125,
+      "grad_norm": 0.5508100986480713,
+      "learning_rate": 3.447414012946818e-06,
+      "loss": 0.4357,
+      "step": 541
+    },
+    {
+      "epoch": 2.554858934169279,
+      "grad_norm": 0.5883537530899048,
+      "learning_rate": 3.4412091157186853e-06,
+      "loss": 0.4504,
+      "step": 542
+    },
+    {
+      "epoch": 2.5595611285266457,
+      "grad_norm": 0.5134791731834412,
+      "learning_rate": 3.4349974556307146e-06,
+      "loss": 0.4409,
+      "step": 543
+    },
+    {
+      "epoch": 2.5642633228840124,
+      "grad_norm": 0.6710099577903748,
+      "learning_rate": 3.4287790773154807e-06,
+      "loss": 0.4655,
+      "step": 544
+    },
+    {
+      "epoch": 2.5689655172413794,
+      "grad_norm": 0.5207563638687134,
+      "learning_rate": 3.4225540254538297e-06,
+      "loss": 0.4626,
+      "step": 545
+    },
+    {
+      "epoch": 2.573667711598746,
+      "grad_norm": 0.5239883065223694,
+      "learning_rate": 3.416322344774562e-06,
+      "loss": 0.4413,
+      "step": 546
+    },
+    {
+      "epoch": 2.5783699059561127,
+      "grad_norm": 0.5160831809043884,
+      "learning_rate": 3.4100840800541055e-06,
+      "loss": 0.4632,
+      "step": 547
+    },
+    {
+      "epoch": 2.58307210031348,
+      "grad_norm": 0.4692513346672058,
+      "learning_rate": 3.4038392761161986e-06,
+      "loss": 0.4535,
+      "step": 548
+    },
+    {
+      "epoch": 2.5877742946708464,
+      "grad_norm": 0.5935313701629639,
+      "learning_rate": 3.3975879778315634e-06,
+      "loss": 0.423,
+      "step": 549
+    },
+    {
+      "epoch": 2.592476489028213,
+      "grad_norm": 0.868563711643219,
+      "learning_rate": 3.391330230117587e-06,
+      "loss": 0.3958,
+      "step": 550
+    },
+    {
+      "epoch": 2.5971786833855797,
+      "grad_norm": 0.49354490637779236,
+      "learning_rate": 3.385066077937997e-06,
+      "loss": 0.4658,
+      "step": 551
+    },
+    {
+      "epoch": 2.6018808777429467,
+      "grad_norm": 0.5423957109451294,
+      "learning_rate": 3.378795566302541e-06,
+      "loss": 0.425,
+      "step": 552
+    },
+    {
+      "epoch": 2.6065830721003134,
+      "grad_norm": 0.5455307364463806,
+      "learning_rate": 3.372518740266658e-06,
+      "loss": 0.4448,
+      "step": 553
+    },
+    {
+      "epoch": 2.6112852664576804,
+      "grad_norm": 0.5334805846214294,
+      "learning_rate": 3.36623564493116e-06,
+      "loss": 0.4563,
+      "step": 554
+    },
+    {
+      "epoch": 2.615987460815047,
+      "grad_norm": 0.49819517135620117,
+      "learning_rate": 3.3599463254419047e-06,
+      "loss": 0.4602,
+      "step": 555
+    },
+    {
+      "epoch": 2.6206896551724137,
+      "grad_norm": 0.5382740497589111,
+      "learning_rate": 3.3536508269894724e-06,
+      "loss": 0.4681,
+      "step": 556
+    },
+    {
+      "epoch": 2.6253918495297803,
+      "grad_norm": 0.4976508319377899,
+      "learning_rate": 3.347349194808842e-06,
+      "loss": 0.4549,
+      "step": 557
+    },
+    {
+      "epoch": 2.6300940438871474,
+      "grad_norm": 0.505546510219574,
+      "learning_rate": 3.3410414741790625e-06,
+      "loss": 0.4288,
+      "step": 558
+    },
+    {
+      "epoch": 2.634796238244514,
+      "grad_norm": 0.50031578540802,
+      "learning_rate": 3.3347277104229332e-06,
+      "loss": 0.4439,
+      "step": 559
+    },
+    {
+      "epoch": 2.639498432601881,
+      "grad_norm": 0.535987377166748,
+      "learning_rate": 3.3284079489066728e-06,
+      "loss": 0.4375,
+      "step": 560
+    },
+    {
+      "epoch": 2.6442006269592477,
+      "grad_norm": 0.5403039455413818,
+      "learning_rate": 3.3220822350395966e-06,
+      "loss": 0.431,
+      "step": 561
+    },
+    {
+      "epoch": 2.6489028213166144,
+      "grad_norm": 0.49778786301612854,
+      "learning_rate": 3.31575061427379e-06,
+      "loss": 0.4312,
+      "step": 562
+    },
+    {
+      "epoch": 2.653605015673981,
+      "grad_norm": 0.5499002933502197,
+      "learning_rate": 3.3094131321037783e-06,
+      "loss": 0.4388,
+      "step": 563
+    },
+    {
+      "epoch": 2.658307210031348,
+      "grad_norm": 0.4913281202316284,
+      "learning_rate": 3.303069834066206e-06,
+      "loss": 0.4334,
+      "step": 564
+    },
+    {
+      "epoch": 2.6630094043887147,
+      "grad_norm": 0.513525664806366,
+      "learning_rate": 3.2967207657395055e-06,
+      "loss": 0.475,
+      "step": 565
+    },
+    {
+      "epoch": 2.6677115987460818,
+      "grad_norm": 0.5432775020599365,
+      "learning_rate": 3.2903659727435692e-06,
+      "loss": 0.4629,
+      "step": 566
+    },
+    {
+      "epoch": 2.6724137931034484,
+      "grad_norm": 0.47809165716171265,
+      "learning_rate": 3.284005500739423e-06,
+      "loss": 0.4577,
+      "step": 567
+    },
+    {
+      "epoch": 2.677115987460815,
+      "grad_norm": 0.5284624695777893,
+      "learning_rate": 3.2776393954289e-06,
+      "loss": 0.4294,
+      "step": 568
+    },
+    {
+      "epoch": 2.6818181818181817,
+      "grad_norm": 0.49356934428215027,
+      "learning_rate": 3.271267702554307e-06,
+      "loss": 0.4208,
+      "step": 569
+    },
+    {
+      "epoch": 2.6865203761755487,
+      "grad_norm": 0.5608382821083069,
+      "learning_rate": 3.2648904678981032e-06,
+      "loss": 0.4646,
+      "step": 570
+    },
+    {
+      "epoch": 2.6912225705329154,
+      "grad_norm": 0.5191068649291992,
+      "learning_rate": 3.2585077372825636e-06,
+      "loss": 0.4132,
+      "step": 571
+    },
+    {
+      "epoch": 2.695924764890282,
+      "grad_norm": 0.5745784640312195,
+      "learning_rate": 3.2521195565694543e-06,
+      "loss": 0.4447,
+      "step": 572
+    },
+    {
+      "epoch": 2.700626959247649,
+      "grad_norm": 0.5302786827087402,
+      "learning_rate": 3.2457259716597023e-06,
+      "loss": 0.4459,
+      "step": 573
+    },
+    {
+      "epoch": 2.7053291536050157,
+      "grad_norm": 0.47739923000335693,
+      "learning_rate": 3.2393270284930658e-06,
+      "loss": 0.4542,
+      "step": 574
+    },
+    {
+      "epoch": 2.7100313479623823,
+      "grad_norm": 0.5023962259292603,
+      "learning_rate": 3.2329227730478026e-06,
+      "loss": 0.4305,
+      "step": 575
+    },
+    {
+      "epoch": 2.714733542319749,
+      "grad_norm": 0.48917847871780396,
+      "learning_rate": 3.2265132513403415e-06,
+      "loss": 0.4189,
+      "step": 576
+    },
+    {
+      "epoch": 2.719435736677116,
+      "grad_norm": 0.4881870448589325,
+      "learning_rate": 3.22009850942495e-06,
+      "loss": 0.4451,
+      "step": 577
+    },
+    {
+      "epoch": 2.7241379310344827,
+      "grad_norm": 0.550719141960144,
+      "learning_rate": 3.213678593393405e-06,
+      "loss": 0.4606,
+      "step": 578
+    },
+    {
+      "epoch": 2.7288401253918497,
+      "grad_norm": 0.5335834622383118,
+      "learning_rate": 3.207253549374662e-06,
+      "loss": 0.4567,
+      "step": 579
+    },
+    {
+      "epoch": 2.7335423197492164,
+      "grad_norm": 0.5054845809936523,
+      "learning_rate": 3.200823423534519e-06,
+      "loss": 0.4433,
+      "step": 580
+    },
+    {
+      "epoch": 2.738244514106583,
+      "grad_norm": 0.52922523021698,
+      "learning_rate": 3.194388262075293e-06,
+      "loss": 0.4485,
+      "step": 581
+    },
+    {
+      "epoch": 2.7429467084639496,
+      "grad_norm": 0.7209565043449402,
+      "learning_rate": 3.1879481112354804e-06,
+      "loss": 0.45,
+      "step": 582
+    },
+    {
+      "epoch": 2.7476489028213167,
+      "grad_norm": 0.5428304076194763,
+      "learning_rate": 3.181503017289428e-06,
+      "loss": 0.4109,
+      "step": 583
+    },
+    {
+      "epoch": 2.7523510971786833,
+      "grad_norm": 0.4802536964416504,
+      "learning_rate": 3.175053026547002e-06,
+      "loss": 0.4153,
+      "step": 584
+    },
+    {
+      "epoch": 2.7570532915360504,
+      "grad_norm": 0.4878472685813904,
+      "learning_rate": 3.16859818535325e-06,
+      "loss": 0.4581,
+      "step": 585
+    },
+    {
+      "epoch": 2.761755485893417,
+      "grad_norm": 0.48334580659866333,
+      "learning_rate": 3.1621385400880756e-06,
+      "loss": 0.4521,
+      "step": 586
+    },
+    {
+      "epoch": 2.7664576802507836,
+      "grad_norm": 0.5261298418045044,
+      "learning_rate": 3.1556741371658984e-06,
+      "loss": 0.4554,
+      "step": 587
+    },
+    {
+      "epoch": 2.7711598746081503,
+      "grad_norm": 0.5182640552520752,
+      "learning_rate": 3.1492050230353238e-06,
+      "loss": 0.447,
+      "step": 588
+    },
+    {
+      "epoch": 2.7758620689655173,
+      "grad_norm": 0.545038640499115,
+      "learning_rate": 3.142731244178809e-06,
+      "loss": 0.4206,
+      "step": 589
+    },
+    {
+      "epoch": 2.780564263322884,
+      "grad_norm": 0.5139180421829224,
+      "learning_rate": 3.1362528471123277e-06,
+      "loss": 0.4037,
+      "step": 590
+    },
+    {
+      "epoch": 2.785266457680251,
+      "grad_norm": 0.54038006067276,
+      "learning_rate": 3.129769878385039e-06,
+      "loss": 0.41,
+      "step": 591
+    },
+    {
+      "epoch": 2.7899686520376177,
+      "grad_norm": 0.4779907166957855,
+      "learning_rate": 3.1232823845789473e-06,
+      "loss": 0.451,
+      "step": 592
+    },
+    {
+      "epoch": 2.7946708463949843,
+      "grad_norm": 0.8652167320251465,
+      "learning_rate": 3.1167904123085736e-06,
+      "loss": 0.4557,
+      "step": 593
+    },
+    {
+      "epoch": 2.799373040752351,
+      "grad_norm": 0.5036542415618896,
+      "learning_rate": 3.110294008220617e-06,
+      "loss": 0.4442,
+      "step": 594
+    },
+    {
+      "epoch": 2.804075235109718,
+      "grad_norm": 0.5064799785614014,
+      "learning_rate": 3.1037932189936205e-06,
+      "loss": 0.4408,
+      "step": 595
+    },
+    {
+      "epoch": 2.8087774294670846,
+      "grad_norm": 0.5548039674758911,
+      "learning_rate": 3.097288091337635e-06,
+      "loss": 0.4359,
+      "step": 596
+    },
+    {
+      "epoch": 2.8134796238244513,
+      "grad_norm": 0.5237991809844971,
+      "learning_rate": 3.0907786719938876e-06,
+      "loss": 0.4258,
+      "step": 597
+    },
+    {
+      "epoch": 2.8181818181818183,
+      "grad_norm": 0.47527915239334106,
+      "learning_rate": 3.084265007734436e-06,
+      "loss": 0.4358,
+      "step": 598
+    },
+    {
+      "epoch": 2.822884012539185,
+      "grad_norm": 3.827120304107666,
+      "learning_rate": 3.0777471453618457e-06,
+      "loss": 0.4607,
+      "step": 599
+    },
+    {
+      "epoch": 2.8275862068965516,
+      "grad_norm": 0.44980505108833313,
+      "learning_rate": 3.0712251317088426e-06,
+      "loss": 0.4333,
+      "step": 600
+    },
+    {
+      "epoch": 2.8322884012539182,
+      "grad_norm": 0.5296145081520081,
+      "learning_rate": 3.064699013637983e-06,
+      "loss": 0.4517,
+      "step": 601
+    },
+    {
+      "epoch": 2.8369905956112853,
+      "grad_norm": 0.5016323924064636,
+      "learning_rate": 3.0581688380413115e-06,
+      "loss": 0.4365,
+      "step": 602
+    },
+    {
+      "epoch": 2.841692789968652,
+      "grad_norm": 0.5290690064430237,
+      "learning_rate": 3.0516346518400315e-06,
+      "loss": 0.4521,
+      "step": 603
+    },
+    {
+      "epoch": 2.846394984326019,
+      "grad_norm": 0.5085683465003967,
+      "learning_rate": 3.0450965019841593e-06,
+      "loss": 0.4518,
+      "step": 604
+    },
+    {
+      "epoch": 2.8510971786833856,
+      "grad_norm": 0.5380619764328003,
+      "learning_rate": 3.0385544354521957e-06,
+      "loss": 0.416,
+      "step": 605
+    },
+    {
+      "epoch": 2.8557993730407523,
+      "grad_norm": 0.48775514960289,
+      "learning_rate": 3.0320084992507814e-06,
+      "loss": 0.4421,
+      "step": 606
+    },
+    {
+      "epoch": 2.860501567398119,
+      "grad_norm": 0.5242875814437866,
+      "learning_rate": 3.0254587404143604e-06,
+      "loss": 0.479,
+      "step": 607
+    },
+    {
+      "epoch": 2.865203761755486,
+      "grad_norm": 0.4760294556617737,
+      "learning_rate": 3.0189052060048464e-06,
+      "loss": 0.4411,
+      "step": 608
+    },
+    {
+      "epoch": 2.8699059561128526,
+      "grad_norm": 0.5514405965805054,
+      "learning_rate": 3.01234794311128e-06,
+      "loss": 0.4384,
+      "step": 609
+    },
+    {
+      "epoch": 2.8746081504702197,
+      "grad_norm": 0.5242321491241455,
+      "learning_rate": 3.0057869988494925e-06,
+      "loss": 0.4629,
+      "step": 610
+    },
+    {
+      "epoch": 2.8793103448275863,
+      "grad_norm": 0.5526494383811951,
+      "learning_rate": 2.999222420361767e-06,
+      "loss": 0.4525,
+      "step": 611
+    },
+    {
+      "epoch": 2.884012539184953,
+      "grad_norm": 0.4866560101509094,
+      "learning_rate": 2.9926542548165e-06,
+      "loss": 0.4652,
+      "step": 612
+    },
+    {
+      "epoch": 2.8887147335423196,
+      "grad_norm": 0.5438888669013977,
+      "learning_rate": 2.9860825494078605e-06,
+      "loss": 0.4369,
+      "step": 613
+    },
+    {
+      "epoch": 2.8934169278996866,
+      "grad_norm": 0.5022105574607849,
+      "learning_rate": 2.979507351355454e-06,
+      "loss": 0.4529,
+      "step": 614
+    },
+    {
+      "epoch": 2.8981191222570533,
+      "grad_norm": 0.5202184915542603,
+      "learning_rate": 2.972928707903981e-06,
+      "loss": 0.4416,
+      "step": 615
+    },
+    {
+      "epoch": 2.9028213166144203,
+      "grad_norm": 0.5267678499221802,
+      "learning_rate": 2.966346666322898e-06,
+      "loss": 0.4392,
+      "step": 616
+    },
+    {
+      "epoch": 2.907523510971787,
+      "grad_norm": 0.6735110282897949,
+      "learning_rate": 2.9597612739060775e-06,
+      "loss": 0.4184,
+      "step": 617
+    },
+    {
+      "epoch": 2.9122257053291536,
+      "grad_norm": 1.0363622903823853,
+      "learning_rate": 2.9531725779714713e-06,
+      "loss": 0.4506,
+      "step": 618
+    },
+    {
+      "epoch": 2.91692789968652,
+      "grad_norm": 0.5508633852005005,
+      "learning_rate": 2.9465806258607653e-06,
+      "loss": 0.4562,
+      "step": 619
+    },
+    {
+      "epoch": 2.9216300940438873,
+      "grad_norm": 0.5185503363609314,
+      "learning_rate": 2.939985464939043e-06,
+      "loss": 0.459,
+      "step": 620
+    },
+    {
+      "epoch": 2.926332288401254,
+      "grad_norm": 0.557746410369873,
+      "learning_rate": 2.9333871425944434e-06,
+      "loss": 0.405,
+      "step": 621
+    },
+    {
+      "epoch": 2.9310344827586206,
+      "grad_norm": 0.7767590284347534,
+      "learning_rate": 2.926785706237822e-06,
+      "loss": 0.4344,
+      "step": 622
+    },
+    {
+      "epoch": 2.9357366771159876,
+      "grad_norm": 0.48968929052352905,
+      "learning_rate": 2.920181203302409e-06,
+      "loss": 0.4261,
+      "step": 623
+    },
+    {
+      "epoch": 2.9404388714733543,
+      "grad_norm": 0.521247148513794,
+      "learning_rate": 2.91357368124347e-06,
+      "loss": 0.4236,
+      "step": 624
+    },
+    {
+      "epoch": 2.945141065830721,
+      "grad_norm": 1.0396796464920044,
+      "learning_rate": 2.906963187537962e-06,
+      "loss": 0.4344,
+      "step": 625
+    },
+    {
+      "epoch": 2.9498432601880875,
+      "grad_norm": 0.6208511590957642,
+      "learning_rate": 2.9003497696841955e-06,
+      "loss": 0.4145,
+      "step": 626
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 0.5897473096847534,
+      "learning_rate": 2.8937334752014913e-06,
+      "loss": 0.466,
+      "step": 627
+    },
+    {
+      "epoch": 2.959247648902821,
+      "grad_norm": 0.5206231474876404,
+      "learning_rate": 2.887114351629839e-06,
+      "loss": 0.4431,
+      "step": 628
+    },
+    {
+      "epoch": 2.9639498432601883,
+      "grad_norm": 0.7496956586837769,
+      "learning_rate": 2.8804924465295575e-06,
+      "loss": 0.4658,
+      "step": 629
+    },
+    {
+      "epoch": 2.968652037617555,
+      "grad_norm": 0.5203924179077148,
+      "learning_rate": 2.873867807480951e-06,
+      "loss": 0.4734,
+      "step": 630
+    },
+    {
+      "epoch": 2.9733542319749215,
+      "grad_norm": 0.5175911784172058,
+      "learning_rate": 2.8672404820839676e-06,
+      "loss": 0.44,
+      "step": 631
+    },
+    {
+      "epoch": 2.978056426332288,
+      "grad_norm": 0.5651830434799194,
+      "learning_rate": 2.8606105179578584e-06,
+      "loss": 0.4466,
+      "step": 632
+    },
+    {
+      "epoch": 2.9827586206896552,
+      "grad_norm": 0.5229518413543701,
+      "learning_rate": 2.8539779627408332e-06,
+      "loss": 0.4259,
+      "step": 633
+    },
+    {
+      "epoch": 2.987460815047022,
+      "grad_norm": 0.5007745027542114,
+      "learning_rate": 2.847342864089721e-06,
+      "loss": 0.4436,
+      "step": 634
+    },
+    {
+      "epoch": 2.992163009404389,
+      "grad_norm": 0.5250042080879211,
+      "learning_rate": 2.8407052696796255e-06,
+      "loss": 0.4308,
+      "step": 635
+    },
+    {
+      "epoch": 2.9968652037617556,
+      "grad_norm": 0.5170105695724487,
+      "learning_rate": 2.834065227203584e-06,
+      "loss": 0.449,
+      "step": 636
+    },
+    {
+      "epoch": 3.0047021943573666,
+      "grad_norm": 0.49444133043289185,
+      "learning_rate": 2.8274227843722213e-06,
+      "loss": 0.8674,
+      "step": 637
+    },
+    {
+      "epoch": 3.0094043887147337,
+      "grad_norm": 0.5178474187850952,
+      "learning_rate": 2.820777988913412e-06,
+      "loss": 0.4192,
+      "step": 638
+    },
+    {
+      "epoch": 3.0141065830721003,
+      "grad_norm": 0.5056724548339844,
+      "learning_rate": 2.8141308885719337e-06,
+      "loss": 0.4166,
+      "step": 639
+    },
+    {
+      "epoch": 3.018808777429467,
+      "grad_norm": 0.48542484641075134,
+      "learning_rate": 2.8074815311091265e-06,
+      "loss": 0.3908,
+      "step": 640
+    },
+    {
+      "epoch": 3.023510971786834,
+      "grad_norm": 0.5061275362968445,
+      "learning_rate": 2.8008299643025477e-06,
+      "loss": 0.4326,
+      "step": 641
+    },
+    {
+      "epoch": 3.0282131661442007,
+      "grad_norm": 0.49305713176727295,
+      "learning_rate": 2.7941762359456294e-06,
+      "loss": 0.4259,
+      "step": 642
+    },
+    {
+      "epoch": 3.0329153605015673,
+      "grad_norm": 0.5171103477478027,
+      "learning_rate": 2.787520393847334e-06,
+      "loss": 0.4167,
+      "step": 643
+    },
+    {
+      "epoch": 3.0376175548589344,
+      "grad_norm": 0.5052148103713989,
+      "learning_rate": 2.780862485831814e-06,
+      "loss": 0.4282,
+      "step": 644
+    },
+    {
+      "epoch": 3.042319749216301,
+      "grad_norm": 0.5092865824699402,
+      "learning_rate": 2.7742025597380644e-06,
+      "loss": 0.4306,
+      "step": 645
+    },
+    {
+      "epoch": 3.0470219435736676,
+      "grad_norm": 0.5179376602172852,
+      "learning_rate": 2.7675406634195824e-06,
+      "loss": 0.423,
+      "step": 646
+    },
+    {
+      "epoch": 3.0517241379310347,
+      "grad_norm": 0.6232285499572754,
+      "learning_rate": 2.7608768447440193e-06,
+      "loss": 0.408,
+      "step": 647
+    },
+    {
+      "epoch": 3.0564263322884013,
+      "grad_norm": 0.5209566950798035,
+      "learning_rate": 2.754211151592841e-06,
+      "loss": 0.4124,
+      "step": 648
+    },
+    {
+      "epoch": 3.061128526645768,
+      "grad_norm": 0.5083103179931641,
+      "learning_rate": 2.7475436318609827e-06,
+      "loss": 0.4326,
+      "step": 649
+    },
+    {
+      "epoch": 3.0658307210031346,
+      "grad_norm": 0.5016148090362549,
+      "learning_rate": 2.7408743334565006e-06,
+      "loss": 0.4077,
+      "step": 650
+    },
+    {
+      "epoch": 3.0705329153605017,
+      "grad_norm": 0.5021759271621704,
+      "learning_rate": 2.734203304300235e-06,
+      "loss": 0.4003,
+      "step": 651
+    },
+    {
+      "epoch": 3.0752351097178683,
+      "grad_norm": 0.5263209939002991,
+      "learning_rate": 2.7275305923254607e-06,
+      "loss": 0.4127,
+      "step": 652
+    },
+    {
+      "epoch": 3.079937304075235,
+      "grad_norm": 0.5125823020935059,
+      "learning_rate": 2.720856245477544e-06,
+      "loss": 0.4028,
+      "step": 653
+    },
+    {
+      "epoch": 3.084639498432602,
+      "grad_norm": 0.5266130566596985,
+      "learning_rate": 2.7141803117135978e-06,
+      "loss": 0.397,
+      "step": 654
+    },
+    {
+      "epoch": 3.0893416927899686,
+      "grad_norm": 0.49692511558532715,
+      "learning_rate": 2.7075028390021385e-06,
+      "loss": 0.4,
+      "step": 655
+    },
+    {
+      "epoch": 3.0940438871473352,
+      "grad_norm": 0.5223091840744019,
+      "learning_rate": 2.7008238753227385e-06,
+      "loss": 0.4067,
+      "step": 656
+    },
+    {
+      "epoch": 3.0987460815047023,
+      "grad_norm": 0.4923226237297058,
+      "learning_rate": 2.694143468665685e-06,
+      "loss": 0.4271,
+      "step": 657
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 0.5270125269889832,
+      "learning_rate": 2.6874616670316338e-06,
+      "loss": 0.3946,
+      "step": 658
+    },
+    {
+      "epoch": 3.1081504702194356,
+      "grad_norm": 0.5231687426567078,
+      "learning_rate": 2.6807785184312618e-06,
+      "loss": 0.414,
+      "step": 659
+    },
+    {
+      "epoch": 3.1128526645768027,
+      "grad_norm": 0.506938099861145,
+      "learning_rate": 2.674094070884926e-06,
+      "loss": 0.4138,
+      "step": 660
+    },
+    {
+      "epoch": 3.1175548589341693,
+      "grad_norm": 0.5096138119697571,
+      "learning_rate": 2.6674083724223166e-06,
+      "loss": 0.433,
+      "step": 661
+    },
+    {
+      "epoch": 3.122257053291536,
+      "grad_norm": 0.5991004705429077,
+      "learning_rate": 2.6607214710821112e-06,
+      "loss": 0.4049,
+      "step": 662
+    },
+    {
+      "epoch": 3.126959247648903,
+      "grad_norm": 0.5364269018173218,
+      "learning_rate": 2.6540334149116304e-06,
+      "loss": 0.4182,
+      "step": 663
+    },
+    {
+      "epoch": 3.1316614420062696,
+      "grad_norm": 0.5317479968070984,
+      "learning_rate": 2.647344251966493e-06,
+      "loss": 0.415,
+      "step": 664
+    },
+    {
+      "epoch": 3.1363636363636362,
+      "grad_norm": 0.5510737895965576,
+      "learning_rate": 2.6406540303102714e-06,
+      "loss": 0.415,
+      "step": 665
+    },
+    {
+      "epoch": 3.1410658307210033,
+      "grad_norm": 0.5536981225013733,
+      "learning_rate": 2.6339627980141425e-06,
+      "loss": 0.4169,
+      "step": 666
+    },
+    {
+      "epoch": 3.14576802507837,
+      "grad_norm": 0.5343831777572632,
+      "learning_rate": 2.6272706031565482e-06,
+      "loss": 0.4018,
+      "step": 667
+    },
+    {
+      "epoch": 3.1504702194357366,
+      "grad_norm": 0.500717282295227,
+      "learning_rate": 2.6205774938228433e-06,
+      "loss": 0.3978,
+      "step": 668
+    },
+    {
+      "epoch": 3.1551724137931036,
+      "grad_norm": 0.6430208683013916,
+      "learning_rate": 2.6138835181049556e-06,
+      "loss": 0.4234,
+      "step": 669
+    },
+    {
+      "epoch": 3.1598746081504703,
+      "grad_norm": 0.5403954982757568,
+      "learning_rate": 2.6071887241010374e-06,
+      "loss": 0.4051,
+      "step": 670
+    },
+    {
+      "epoch": 3.164576802507837,
+      "grad_norm": 0.6121819615364075,
+      "learning_rate": 2.6004931599151223e-06,
+      "loss": 0.3929,
+      "step": 671
+    },
+    {
+      "epoch": 3.169278996865204,
+      "grad_norm": 0.519755482673645,
+      "learning_rate": 2.593796873656775e-06,
+      "loss": 0.4353,
+      "step": 672
+    },
+    {
+      "epoch": 3.1739811912225706,
+      "grad_norm": 0.4970746636390686,
+      "learning_rate": 2.587099913440749e-06,
+      "loss": 0.4155,
+      "step": 673
+    },
+    {
+      "epoch": 3.1786833855799372,
+      "grad_norm": 0.512956440448761,
+      "learning_rate": 2.580402327386643e-06,
+      "loss": 0.403,
+      "step": 674
+    },
+    {
+      "epoch": 3.183385579937304,
+      "grad_norm": 0.5539934635162354,
+      "learning_rate": 2.5737041636185496e-06,
+      "loss": 0.4095,
+      "step": 675
+    },
+    {
+      "epoch": 3.188087774294671,
+      "grad_norm": 0.7322019338607788,
+      "learning_rate": 2.5670054702647146e-06,
+      "loss": 0.3941,
+      "step": 676
+    },
+    {
+      "epoch": 3.1927899686520376,
+      "grad_norm": 1.063719391822815,
+      "learning_rate": 2.5603062954571872e-06,
+      "loss": 0.4184,
+      "step": 677
+    },
+    {
+      "epoch": 3.197492163009404,
+      "grad_norm": 0.5929884910583496,
+      "learning_rate": 2.553606687331477e-06,
+      "loss": 0.4378,
+      "step": 678
+    },
+    {
+      "epoch": 3.2021943573667713,
+      "grad_norm": 0.6147572994232178,
+      "learning_rate": 2.5469066940262073e-06,
+      "loss": 0.4078,
+      "step": 679
+    },
+    {
+      "epoch": 3.206896551724138,
+      "grad_norm": 0.5559990406036377,
+      "learning_rate": 2.540206363682768e-06,
+      "loss": 0.4004,
+      "step": 680
+    },
+    {
+      "epoch": 3.2115987460815045,
+      "grad_norm": 0.49420028924942017,
+      "learning_rate": 2.533505744444972e-06,
+      "loss": 0.4182,
+      "step": 681
+    },
+    {
+      "epoch": 3.2163009404388716,
+      "grad_norm": 0.4829191565513611,
+      "learning_rate": 2.526804884458707e-06,
+      "loss": 0.4093,
+      "step": 682
+    },
+    {
+      "epoch": 3.2210031347962382,
+      "grad_norm": 0.5211793780326843,
+      "learning_rate": 2.520103831871591e-06,
+      "loss": 0.4157,
+      "step": 683
+    },
+    {
+      "epoch": 3.225705329153605,
+      "grad_norm": 0.4876849055290222,
+      "learning_rate": 2.513402634832627e-06,
+      "loss": 0.3961,
+      "step": 684
+    },
+    {
+      "epoch": 3.230407523510972,
+      "grad_norm": 0.5313226580619812,
+      "learning_rate": 2.5067013414918523e-06,
+      "loss": 0.4005,
+      "step": 685
+    },
+    {
+      "epoch": 3.2351097178683386,
+      "grad_norm": 0.5361777544021606,
+      "learning_rate": 2.5e-06,
+      "loss": 0.414,
+      "step": 686
+    },
+    {
+      "epoch": 3.239811912225705,
+      "grad_norm": 0.5172957181930542,
+      "learning_rate": 2.493298658508149e-06,
+      "loss": 0.3953,
+      "step": 687
+    },
+    {
+      "epoch": 3.2445141065830723,
+      "grad_norm": 0.5602390170097351,
+      "learning_rate": 2.4865973651673743e-06,
+      "loss": 0.4137,
+      "step": 688
+    },
+    {
+      "epoch": 3.249216300940439,
+      "grad_norm": 0.5494715571403503,
+      "learning_rate": 2.4798961681284096e-06,
+      "loss": 0.4385,
+      "step": 689
+    },
+    {
+      "epoch": 3.2539184952978055,
+      "grad_norm": 0.521594762802124,
+      "learning_rate": 2.473195115541293e-06,
+      "loss": 0.3943,
+      "step": 690
+    },
+    {
+      "epoch": 3.2586206896551726,
+      "grad_norm": 0.5154503583908081,
+      "learning_rate": 2.466494255555029e-06,
+      "loss": 0.441,
+      "step": 691
+    },
+    {
+      "epoch": 3.2633228840125392,
+      "grad_norm": 0.6013544797897339,
+      "learning_rate": 2.459793636317233e-06,
+      "loss": 0.405,
+      "step": 692
+    },
+    {
+      "epoch": 3.268025078369906,
+      "grad_norm": 0.531454861164093,
+      "learning_rate": 2.4530933059737936e-06,
+      "loss": 0.4425,
+      "step": 693
+    },
+    {
+      "epoch": 3.2727272727272725,
+      "grad_norm": 0.5347344875335693,
+      "learning_rate": 2.4463933126685236e-06,
+      "loss": 0.4051,
+      "step": 694
+    },
+    {
+      "epoch": 3.2774294670846396,
+      "grad_norm": 0.5035227537155151,
+      "learning_rate": 2.439693704542814e-06,
+      "loss": 0.4021,
+      "step": 695
+    },
+    {
+      "epoch": 3.282131661442006,
+      "grad_norm": 0.4999369978904724,
+      "learning_rate": 2.432994529735286e-06,
+      "loss": 0.4097,
+      "step": 696
+    },
+    {
+      "epoch": 3.2868338557993733,
+      "grad_norm": 0.5022649765014648,
+      "learning_rate": 2.4262958363814512e-06,
+      "loss": 0.4189,
+      "step": 697
+    },
+    {
+      "epoch": 3.29153605015674,
+      "grad_norm": 0.5714133381843567,
+      "learning_rate": 2.4195976726133574e-06,
+      "loss": 0.4067,
+      "step": 698
+    },
+    {
+      "epoch": 3.2962382445141065,
+      "grad_norm": 0.5297165513038635,
+      "learning_rate": 2.4129000865592517e-06,
+      "loss": 0.4053,
+      "step": 699
+    },
+    {
+      "epoch": 3.300940438871473,
+      "grad_norm": 0.5833218693733215,
+      "learning_rate": 2.4062031263432267e-06,
+      "loss": 0.4269,
+      "step": 700
+    },
+    {
+      "epoch": 3.30564263322884,
+      "grad_norm": 0.48326241970062256,
+      "learning_rate": 2.3995068400848785e-06,
+      "loss": 0.404,
+      "step": 701
+    },
+    {
+      "epoch": 3.310344827586207,
+      "grad_norm": 0.5570845603942871,
+      "learning_rate": 2.392811275898963e-06,
+      "loss": 0.419,
+      "step": 702
+    },
+    {
+      "epoch": 3.3150470219435735,
+      "grad_norm": 0.5318753123283386,
+      "learning_rate": 2.3861164818950448e-06,
+      "loss": 0.387,
+      "step": 703
+    },
+    {
+      "epoch": 3.3197492163009406,
+      "grad_norm": 0.5461038947105408,
+      "learning_rate": 2.379422506177157e-06,
+      "loss": 0.4121,
+      "step": 704
+    },
+    {
+      "epoch": 3.324451410658307,
+      "grad_norm": 0.5123147368431091,
+      "learning_rate": 2.372729396843453e-06,
+      "loss": 0.4138,
+      "step": 705
+    },
+    {
+      "epoch": 3.329153605015674,
+      "grad_norm": 0.5471891760826111,
+      "learning_rate": 2.366037201985858e-06,
+      "loss": 0.4185,
+      "step": 706
+    },
+    {
+      "epoch": 3.333855799373041,
+      "grad_norm": 0.550929605960846,
+      "learning_rate": 2.3593459696897294e-06,
+      "loss": 0.3945,
+      "step": 707
+    },
+    {
+      "epoch": 3.3385579937304075,
+      "grad_norm": 0.5357216000556946,
+      "learning_rate": 2.352655748033508e-06,
+      "loss": 0.4136,
+      "step": 708
+    },
+    {
+      "epoch": 3.343260188087774,
+      "grad_norm": 0.5054169297218323,
+      "learning_rate": 2.3459665850883704e-06,
+      "loss": 0.4346,
+      "step": 709
+    },
+    {
+      "epoch": 3.347962382445141,
+      "grad_norm": 0.5135513544082642,
+      "learning_rate": 2.33927852891789e-06,
+      "loss": 0.4184,
+      "step": 710
+    },
+    {
+      "epoch": 3.352664576802508,
+      "grad_norm": 0.5367952585220337,
+      "learning_rate": 2.3325916275776834e-06,
+      "loss": 0.4398,
+      "step": 711
+    },
+    {
+      "epoch": 3.3573667711598745,
+      "grad_norm": 0.5645503997802734,
+      "learning_rate": 2.3259059291150744e-06,
+      "loss": 0.401,
+      "step": 712
+    },
+    {
+      "epoch": 3.3620689655172415,
+      "grad_norm": 0.5017855763435364,
+      "learning_rate": 2.319221481568739e-06,
+      "loss": 0.4206,
+      "step": 713
+    },
+    {
+      "epoch": 3.366771159874608,
+      "grad_norm": 0.5428129434585571,
+      "learning_rate": 2.3125383329683666e-06,
+      "loss": 0.4223,
+      "step": 714
+    },
+    {
+      "epoch": 3.371473354231975,
+      "grad_norm": 0.5728990435600281,
+      "learning_rate": 2.3058565313343152e-06,
+      "loss": 0.3896,
+      "step": 715
+    },
+    {
+      "epoch": 3.376175548589342,
+      "grad_norm": 0.5580390095710754,
+      "learning_rate": 2.2991761246772623e-06,
+      "loss": 0.405,
+      "step": 716
+    },
+    {
+      "epoch": 3.3808777429467085,
+      "grad_norm": 0.5162783861160278,
+      "learning_rate": 2.2924971609978623e-06,
+      "loss": 0.3961,
+      "step": 717
+    },
+    {
+      "epoch": 3.385579937304075,
+      "grad_norm": 0.5139627456665039,
+      "learning_rate": 2.285819688286403e-06,
+      "loss": 0.3877,
+      "step": 718
+    },
+    {
+      "epoch": 3.3902821316614418,
+      "grad_norm": 0.5245716571807861,
+      "learning_rate": 2.2791437545224563e-06,
+      "loss": 0.4341,
+      "step": 719
+    },
+    {
+      "epoch": 3.394984326018809,
+      "grad_norm": 0.5056993961334229,
+      "learning_rate": 2.2724694076745397e-06,
+      "loss": 0.4116,
+      "step": 720
+    },
+    {
+      "epoch": 3.3996865203761755,
+      "grad_norm": 0.5380951762199402,
+      "learning_rate": 2.265796695699766e-06,
+      "loss": 0.4154,
+      "step": 721
+    },
+    {
+      "epoch": 3.4043887147335425,
+      "grad_norm": 0.5752882361412048,
+      "learning_rate": 2.2591256665434998e-06,
+      "loss": 0.4177,
+      "step": 722
+    },
+    {
+      "epoch": 3.409090909090909,
+      "grad_norm": 0.5096545219421387,
+      "learning_rate": 2.252456368139019e-06,
+      "loss": 0.4139,
+      "step": 723
+    },
+    {
+      "epoch": 3.413793103448276,
+      "grad_norm": 0.5271265506744385,
+      "learning_rate": 2.245788848407159e-06,
+      "loss": 0.4223,
+      "step": 724
+    },
+    {
+      "epoch": 3.4184952978056424,
+      "grad_norm": 0.5337948203086853,
+      "learning_rate": 2.2391231552559815e-06,
+      "loss": 0.4208,
+      "step": 725
+    },
+    {
+      "epoch": 3.4231974921630095,
+      "grad_norm": 0.5831982493400574,
+      "learning_rate": 2.2324593365804184e-06,
+      "loss": 0.3903,
+      "step": 726
+    },
+    {
+      "epoch": 3.427899686520376,
+      "grad_norm": 0.5560181140899658,
+      "learning_rate": 2.225797440261936e-06,
+      "loss": 0.4321,
+      "step": 727
+    },
+    {
+      "epoch": 3.4326018808777428,
+      "grad_norm": 0.5280754566192627,
+      "learning_rate": 2.219137514168187e-06,
+      "loss": 0.3948,
+      "step": 728
+    },
+    {
+      "epoch": 3.43730407523511,
+      "grad_norm": 0.5112728476524353,
+      "learning_rate": 2.212479606152667e-06,
+      "loss": 0.426,
+      "step": 729
+    },
+    {
+      "epoch": 3.4420062695924765,
+      "grad_norm": 0.5265609622001648,
+      "learning_rate": 2.205823764054372e-06,
+      "loss": 0.4094,
+      "step": 730
+    },
+    {
+      "epoch": 3.446708463949843,
+      "grad_norm": 0.48872876167297363,
+      "learning_rate": 2.199170035697453e-06,
+      "loss": 0.4073,
+      "step": 731
+    },
+    {
+      "epoch": 3.45141065830721,
+      "grad_norm": 0.521526038646698,
+      "learning_rate": 2.1925184688908735e-06,
+      "loss": 0.4135,
+      "step": 732
+    },
+    {
+      "epoch": 3.456112852664577,
+      "grad_norm": 0.5472409725189209,
+      "learning_rate": 2.185869111428067e-06,
+      "loss": 0.4369,
+      "step": 733
+    },
+    {
+      "epoch": 3.4608150470219434,
+      "grad_norm": 0.5710989832878113,
+      "learning_rate": 2.1792220110865885e-06,
+      "loss": 0.4224,
+      "step": 734
+    },
+    {
+      "epoch": 3.4655172413793105,
+      "grad_norm": 0.6842237710952759,
+      "learning_rate": 2.1725772156277795e-06,
+      "loss": 0.4032,
+      "step": 735
+    },
+    {
+      "epoch": 3.470219435736677,
+      "grad_norm": 0.5267806053161621,
+      "learning_rate": 2.165934772796417e-06,
+      "loss": 0.412,
+      "step": 736
+    },
+    {
+      "epoch": 3.4749216300940438,
+      "grad_norm": 0.6110681295394897,
+      "learning_rate": 2.159294730320374e-06,
+      "loss": 0.4264,
+      "step": 737
+    },
+    {
+      "epoch": 3.479623824451411,
+      "grad_norm": 0.567146897315979,
+      "learning_rate": 2.15265713591028e-06,
+      "loss": 0.4145,
+      "step": 738
+    },
+    {
+      "epoch": 3.4843260188087775,
+      "grad_norm": 0.5455949306488037,
+      "learning_rate": 2.1460220372591676e-06,
+      "loss": 0.4338,
+      "step": 739
+    },
+    {
+      "epoch": 3.489028213166144,
+      "grad_norm": 0.5989809036254883,
+      "learning_rate": 2.139389482042142e-06,
+      "loss": 0.3985,
+      "step": 740
+    },
+    {
+      "epoch": 3.493730407523511,
+      "grad_norm": 0.5719276666641235,
+      "learning_rate": 2.1327595179160332e-06,
+      "loss": 0.4157,
+      "step": 741
+    },
+    {
+      "epoch": 3.498432601880878,
+      "grad_norm": 0.5467635989189148,
+      "learning_rate": 2.1261321925190492e-06,
+      "loss": 0.4055,
+      "step": 742
+    },
+    {
+      "epoch": 3.5031347962382444,
+      "grad_norm": 0.5934366583824158,
+      "learning_rate": 2.1195075534704433e-06,
+      "loss": 0.3826,
+      "step": 743
+    },
+    {
+      "epoch": 3.507836990595611,
+      "grad_norm": 0.5331693291664124,
+      "learning_rate": 2.1128856483701625e-06,
+      "loss": 0.4114,
+      "step": 744
+    },
+    {
+      "epoch": 3.512539184952978,
+      "grad_norm": 0.5111117959022522,
+      "learning_rate": 2.10626652479851e-06,
+      "loss": 0.3927,
+      "step": 745
+    },
+    {
+      "epoch": 3.5172413793103448,
+      "grad_norm": 0.5242960453033447,
+      "learning_rate": 2.0996502303158057e-06,
+      "loss": 0.4079,
+      "step": 746
+    },
+    {
+      "epoch": 3.521943573667712,
+      "grad_norm": 1.4825061559677124,
+      "learning_rate": 2.0930368124620385e-06,
+      "loss": 0.4089,
+      "step": 747
+    },
+    {
+      "epoch": 3.5266457680250785,
+      "grad_norm": 0.5873427391052246,
+      "learning_rate": 2.086426318756531e-06,
+      "loss": 0.4258,
+      "step": 748
+    },
+    {
+      "epoch": 3.531347962382445,
+      "grad_norm": 0.5490135550498962,
+      "learning_rate": 2.0798187966975917e-06,
+      "loss": 0.4313,
+      "step": 749
+    },
+    {
+      "epoch": 3.5360501567398117,
+      "grad_norm": 0.5256242752075195,
+      "learning_rate": 2.073214293762179e-06,
+      "loss": 0.4205,
+      "step": 750
+    },
+    {
+      "epoch": 3.540752351097179,
+      "grad_norm": 0.5382546782493591,
+      "learning_rate": 2.0666128574055575e-06,
+      "loss": 0.4182,
+      "step": 751
+    },
+    {
+      "epoch": 3.5454545454545454,
+      "grad_norm": 1.0182400941848755,
+      "learning_rate": 2.0600145350609585e-06,
+      "loss": 0.4176,
+      "step": 752
+    },
+    {
+      "epoch": 3.5501567398119125,
+      "grad_norm": 0.5244695544242859,
+      "learning_rate": 2.053419374139235e-06,
+      "loss": 0.4162,
+      "step": 753
+    },
+    {
+      "epoch": 3.554858934169279,
+      "grad_norm": 0.509304940700531,
+      "learning_rate": 2.0468274220285295e-06,
+      "loss": 0.4149,
+      "step": 754
+    },
+    {
+      "epoch": 3.5595611285266457,
+      "grad_norm": 0.5326141119003296,
+      "learning_rate": 2.0402387260939224e-06,
+      "loss": 0.4105,
+      "step": 755
+    },
+    {
+      "epoch": 3.5642633228840124,
+      "grad_norm": 0.5248592495918274,
+      "learning_rate": 2.033653333677103e-06,
+      "loss": 0.4214,
+      "step": 756
+    },
+    {
+      "epoch": 3.5689655172413794,
+      "grad_norm": 0.5611410737037659,
+      "learning_rate": 2.02707129209602e-06,
+      "loss": 0.4034,
+      "step": 757
+    },
+    {
+      "epoch": 3.573667711598746,
+      "grad_norm": 0.49410030245780945,
+      "learning_rate": 2.0204926486445463e-06,
+      "loss": 0.4229,
+      "step": 758
+    },
+    {
+      "epoch": 3.5783699059561127,
+      "grad_norm": 0.49964067339897156,
+      "learning_rate": 2.0139174505921403e-06,
+      "loss": 0.4433,
+      "step": 759
+    },
+    {
+      "epoch": 3.58307210031348,
+      "grad_norm": 0.5891857147216797,
+      "learning_rate": 2.0073457451835e-06,
+      "loss": 0.3995,
+      "step": 760
+    },
+    {
+      "epoch": 3.5877742946708464,
+      "grad_norm": 0.5028961896896362,
+      "learning_rate": 2.0007775796382335e-06,
+      "loss": 0.4242,
+      "step": 761
+    },
+    {
+      "epoch": 3.592476489028213,
+      "grad_norm": 0.5589643716812134,
+      "learning_rate": 1.994213001150508e-06,
+      "loss": 0.3929,
+      "step": 762
+    },
+    {
+      "epoch": 3.5971786833855797,
+      "grad_norm": 0.6315639615058899,
+      "learning_rate": 1.9876520568887207e-06,
+      "loss": 0.4196,
+      "step": 763
+    },
+    {
+      "epoch": 3.6018808777429467,
+      "grad_norm": 0.538480818271637,
+      "learning_rate": 1.981094793995155e-06,
+      "loss": 0.403,
+      "step": 764
+    },
+    {
+      "epoch": 3.6065830721003134,
+      "grad_norm": 0.5254942774772644,
+      "learning_rate": 1.974541259585641e-06,
+      "loss": 0.4313,
+      "step": 765
+    },
+    {
+      "epoch": 3.6112852664576804,
+      "grad_norm": 16.726457595825195,
+      "learning_rate": 1.9679915007492194e-06,
+      "loss": 0.4131,
+      "step": 766
+    },
+    {
+      "epoch": 3.615987460815047,
+      "grad_norm": 0.572813093662262,
+      "learning_rate": 1.9614455645478047e-06,
+      "loss": 0.4022,
+      "step": 767
+    },
+    {
+      "epoch": 3.6206896551724137,
+      "grad_norm": 0.5614109039306641,
+      "learning_rate": 1.9549034980158403e-06,
+      "loss": 0.4149,
+      "step": 768
+    },
+    {
+      "epoch": 3.6253918495297803,
+      "grad_norm": 0.5074586272239685,
+      "learning_rate": 1.9483653481599697e-06,
+      "loss": 0.42,
+      "step": 769
+    },
+    {
+      "epoch": 3.6300940438871474,
+      "grad_norm": 0.5631129145622253,
+      "learning_rate": 1.9418311619586897e-06,
+      "loss": 0.4257,
+      "step": 770
+    },
+    {
+      "epoch": 3.634796238244514,
+      "grad_norm": 0.5475037097930908,
+      "learning_rate": 1.935300986362018e-06,
+      "loss": 0.3986,
+      "step": 771
+    },
+    {
+      "epoch": 3.639498432601881,
+      "grad_norm": 0.5358164310455322,
+      "learning_rate": 1.9287748682911582e-06,
+      "loss": 0.4308,
+      "step": 772
+    },
+    {
+      "epoch": 3.6442006269592477,
+      "grad_norm": 0.612288773059845,
+      "learning_rate": 1.9222528546381543e-06,
+      "loss": 0.4216,
+      "step": 773
+    },
+    {
+      "epoch": 3.6489028213166144,
+      "grad_norm": 0.569261908531189,
+      "learning_rate": 1.9157349922655648e-06,
+      "loss": 0.3987,
+      "step": 774
+    },
+    {
+      "epoch": 3.653605015673981,
+      "grad_norm": 0.5050970911979675,
+      "learning_rate": 1.909221328006114e-06,
+      "loss": 0.4084,
+      "step": 775
+    },
+    {
+      "epoch": 3.658307210031348,
+      "grad_norm": 0.5210846066474915,
+      "learning_rate": 1.9027119086623647e-06,
+      "loss": 0.4222,
+      "step": 776
+    },
+    {
+      "epoch": 3.6630094043887147,
+      "grad_norm": 0.5151328444480896,
+      "learning_rate": 1.8962067810063806e-06,
+      "loss": 0.4373,
+      "step": 777
+    },
+    {
+      "epoch": 3.6677115987460818,
+      "grad_norm": 0.5783085823059082,
+      "learning_rate": 1.8897059917793844e-06,
+      "loss": 0.3932,
+      "step": 778
+    },
+    {
+      "epoch": 3.6724137931034484,
+      "grad_norm": 0.5122448205947876,
+      "learning_rate": 1.8832095876914268e-06,
+      "loss": 0.4111,
+      "step": 779
+    },
+    {
+      "epoch": 3.677115987460815,
+      "grad_norm": 0.5330267548561096,
+      "learning_rate": 1.8767176154210537e-06,
+      "loss": 0.4178,
+      "step": 780
+    },
+    {
+      "epoch": 3.6818181818181817,
+      "grad_norm": 0.543250560760498,
+      "learning_rate": 1.8702301216149616e-06,
+      "loss": 0.4066,
+      "step": 781
+    },
+    {
+      "epoch": 3.6865203761755487,
+      "grad_norm": 0.5525115728378296,
+      "learning_rate": 1.8637471528876727e-06,
+      "loss": 0.4079,
+      "step": 782
+    },
+    {
+      "epoch": 3.6912225705329154,
+      "grad_norm": 0.5233202576637268,
+      "learning_rate": 1.8572687558211923e-06,
+      "loss": 0.4194,
+      "step": 783
+    },
+    {
+      "epoch": 3.695924764890282,
+      "grad_norm": 0.6125171780586243,
+      "learning_rate": 1.850794976964677e-06,
+      "loss": 0.3814,
+      "step": 784
+    },
+    {
+      "epoch": 3.700626959247649,
+      "grad_norm": 0.5407599210739136,
+      "learning_rate": 1.8443258628341026e-06,
+      "loss": 0.4144,
+      "step": 785
+    },
+    {
+      "epoch": 3.7053291536050157,
+      "grad_norm": 0.5853657126426697,
+      "learning_rate": 1.837861459911925e-06,
+      "loss": 0.4249,
+      "step": 786
+    },
+    {
+      "epoch": 3.7100313479623823,
+      "grad_norm": 0.7031234502792358,
+      "learning_rate": 1.8314018146467505e-06,
+      "loss": 0.4052,
+      "step": 787
+    },
+    {
+      "epoch": 3.714733542319749,
+      "grad_norm": 1.9762593507766724,
+      "learning_rate": 1.8249469734529995e-06,
+      "loss": 0.3881,
+      "step": 788
+    },
+    {
+      "epoch": 3.719435736677116,
+      "grad_norm": 0.5328689217567444,
+      "learning_rate": 1.818496982710572e-06,
+      "loss": 0.4246,
+      "step": 789
+    },
+    {
+      "epoch": 3.7241379310344827,
+      "grad_norm": 0.5613765120506287,
+      "learning_rate": 1.81205188876452e-06,
+      "loss": 0.424,
+      "step": 790
+    },
+    {
+      "epoch": 3.7288401253918497,
+      "grad_norm": 0.6554239392280579,
+      "learning_rate": 1.8056117379247078e-06,
+      "loss": 0.4273,
+      "step": 791
+    },
+    {
+      "epoch": 3.7335423197492164,
+      "grad_norm": 1.2649489641189575,
+      "learning_rate": 1.7991765764654813e-06,
+      "loss": 0.4147,
+      "step": 792
+    },
+    {
+      "epoch": 3.738244514106583,
+      "grad_norm": 0.5478984713554382,
+      "learning_rate": 1.7927464506253394e-06,
+      "loss": 0.4252,
+      "step": 793
+    },
+    {
+      "epoch": 3.7429467084639496,
+      "grad_norm": 0.6170595288276672,
+      "learning_rate": 1.7863214066065951e-06,
+      "loss": 0.3919,
+      "step": 794
+    },
+    {
+      "epoch": 3.7476489028213167,
+      "grad_norm": 0.5435987710952759,
+      "learning_rate": 1.779901490575051e-06,
+      "loss": 0.4191,
+      "step": 795
+    },
+    {
+      "epoch": 3.7523510971786833,
+      "grad_norm": 0.49888044595718384,
+      "learning_rate": 1.7734867486596596e-06,
+      "loss": 0.4214,
+      "step": 796
+    },
+    {
+      "epoch": 3.7570532915360504,
+      "grad_norm": 0.6968030333518982,
+      "learning_rate": 1.767077226952198e-06,
+      "loss": 0.4273,
+      "step": 797
+    },
+    {
+      "epoch": 3.761755485893417,
+      "grad_norm": 0.5169183611869812,
+      "learning_rate": 1.7606729715069349e-06,
+      "loss": 0.4148,
+      "step": 798
+    },
+    {
+      "epoch": 3.7664576802507836,
+      "grad_norm": 0.5539206862449646,
+      "learning_rate": 1.7542740283402981e-06,
+      "loss": 0.4184,
+      "step": 799
+    },
+    {
+      "epoch": 3.7711598746081503,
+      "grad_norm": 0.6863934397697449,
+      "learning_rate": 1.7478804434305466e-06,
+      "loss": 0.4005,
+      "step": 800
+    },
+    {
+      "epoch": 3.7758620689655173,
+      "grad_norm": 0.6211932897567749,
+      "learning_rate": 1.741492262717438e-06,
+      "loss": 0.4177,
+      "step": 801
+    },
+    {
+      "epoch": 3.780564263322884,
+      "grad_norm": 0.6639383435249329,
+      "learning_rate": 1.7351095321018974e-06,
+      "loss": 0.3871,
+      "step": 802
+    },
+    {
+      "epoch": 3.785266457680251,
+      "grad_norm": 0.5569302439689636,
+      "learning_rate": 1.7287322974456933e-06,
+      "loss": 0.3931,
+      "step": 803
+    },
+    {
+      "epoch": 3.7899686520376177,
+      "grad_norm": 0.5468054413795471,
+      "learning_rate": 1.7223606045711006e-06,
+      "loss": 0.4165,
+      "step": 804
+    },
+    {
+      "epoch": 3.7946708463949843,
+      "grad_norm": 0.5721184015274048,
+      "learning_rate": 1.7159944992605774e-06,
+      "loss": 0.4209,
+      "step": 805
+    },
+    {
+      "epoch": 3.799373040752351,
+      "grad_norm": 0.5125740170478821,
+      "learning_rate": 1.7096340272564318e-06,
+      "loss": 0.4092,
+      "step": 806
+    },
+    {
+      "epoch": 3.804075235109718,
+      "grad_norm": 0.5384911298751831,
+      "learning_rate": 1.7032792342604947e-06,
+      "loss": 0.4004,
+      "step": 807
+    },
+    {
+      "epoch": 3.8087774294670846,
+      "grad_norm": 0.5560373067855835,
+      "learning_rate": 1.6969301659337944e-06,
+      "loss": 0.4157,
+      "step": 808
+    },
+    {
+      "epoch": 3.8134796238244513,
+      "grad_norm": 0.5534592866897583,
+      "learning_rate": 1.6905868678962225e-06,
+      "loss": 0.4208,
+      "step": 809
+    },
+    {
+      "epoch": 3.8181818181818183,
+      "grad_norm": 0.5902019739151001,
+      "learning_rate": 1.684249385726211e-06,
+      "loss": 0.4125,
+      "step": 810
+    },
+    {
+      "epoch": 3.822884012539185,
+      "grad_norm": 0.549802839756012,
+      "learning_rate": 1.677917764960404e-06,
+      "loss": 0.4017,
+      "step": 811
+    },
+    {
+      "epoch": 3.8275862068965516,
+      "grad_norm": 0.5346619486808777,
+      "learning_rate": 1.6715920510933277e-06,
+      "loss": 0.4317,
+      "step": 812
+    },
+    {
+      "epoch": 3.8322884012539182,
+      "grad_norm": 0.5298053622245789,
+      "learning_rate": 1.6652722895770676e-06,
+      "loss": 0.4281,
+      "step": 813
+    },
+    {
+      "epoch": 3.8369905956112853,
+      "grad_norm": 0.5283401608467102,
+      "learning_rate": 1.6589585258209383e-06,
+      "loss": 0.3778,
+      "step": 814
+    },
+    {
+      "epoch": 3.841692789968652,
+      "grad_norm": 0.5301955342292786,
+      "learning_rate": 1.6526508051911588e-06,
+      "loss": 0.4182,
+      "step": 815
+    },
+    {
+      "epoch": 3.846394984326019,
+      "grad_norm": 0.527755856513977,
+      "learning_rate": 1.6463491730105282e-06,
+      "loss": 0.4071,
+      "step": 816
+    },
+    {
+      "epoch": 3.8510971786833856,
+      "grad_norm": 0.5193698406219482,
+      "learning_rate": 1.6400536745580955e-06,
+      "loss": 0.3893,
+      "step": 817
+    },
+    {
+      "epoch": 3.8557993730407523,
+      "grad_norm": 0.5560773611068726,
+      "learning_rate": 1.6337643550688408e-06,
+      "loss": 0.4165,
+      "step": 818
+    },
+    {
+      "epoch": 3.860501567398119,
+      "grad_norm": 0.5100756883621216,
+      "learning_rate": 1.627481259733343e-06,
+      "loss": 0.3977,
+      "step": 819
+    },
+    {
+      "epoch": 3.865203761755486,
+      "grad_norm": 0.6014723181724548,
+      "learning_rate": 1.6212044336974598e-06,
+      "loss": 0.3951,
+      "step": 820
+    },
+    {
+      "epoch": 3.8699059561128526,
+      "grad_norm": 0.5284562706947327,
+      "learning_rate": 1.614933922062003e-06,
+      "loss": 0.4074,
+      "step": 821
+    },
+    {
+      "epoch": 3.8746081504702197,
+      "grad_norm": 0.563565731048584,
+      "learning_rate": 1.6086697698824144e-06,
+      "loss": 0.4023,
+      "step": 822
+    },
+    {
+      "epoch": 3.8793103448275863,
+      "grad_norm": 0.5110756158828735,
+      "learning_rate": 1.6024120221684373e-06,
+      "loss": 0.4003,
+      "step": 823
+    },
+    {
+      "epoch": 3.884012539184953,
+      "grad_norm": 0.5191101431846619,
+      "learning_rate": 1.5961607238838022e-06,
+      "loss": 0.4081,
+      "step": 824
+    },
+    {
+      "epoch": 3.8887147335423196,
+      "grad_norm": 0.5377500653266907,
+      "learning_rate": 1.589915919945894e-06,
+      "loss": 0.4184,
+      "step": 825
+    },
+    {
+      "epoch": 3.8934169278996866,
+      "grad_norm": 0.5440651774406433,
+      "learning_rate": 1.5836776552254386e-06,
+      "loss": 0.4355,
+      "step": 826
+    },
+    {
+      "epoch": 3.8981191222570533,
+      "grad_norm": 0.603829026222229,
+      "learning_rate": 1.5774459745461711e-06,
+      "loss": 0.4056,
+      "step": 827
+    },
+    {
+      "epoch": 3.9028213166144203,
+      "grad_norm": 0.5026826858520508,
+      "learning_rate": 1.5712209226845201e-06,
+      "loss": 0.3824,
+      "step": 828
+    },
+    {
+      "epoch": 3.907523510971787,
+      "grad_norm": 0.7123918533325195,
+      "learning_rate": 1.565002544369286e-06,
+      "loss": 0.4157,
+      "step": 829
+    },
+    {
+      "epoch": 3.9122257053291536,
+      "grad_norm": 0.541552722454071,
+      "learning_rate": 1.5587908842813142e-06,
+      "loss": 0.4057,
+      "step": 830
+    },
+    {
+      "epoch": 3.91692789968652,
+      "grad_norm": 0.5404677391052246,
+      "learning_rate": 1.5525859870531823e-06,
+      "loss": 0.4211,
+      "step": 831
+    },
+    {
+      "epoch": 3.9216300940438873,
+      "grad_norm": 0.6348632574081421,
+      "learning_rate": 1.5463878972688707e-06,
+      "loss": 0.4092,
+      "step": 832
+    },
+    {
+      "epoch": 3.926332288401254,
+      "grad_norm": 0.5390607714653015,
+      "learning_rate": 1.5401966594634483e-06,
+      "loss": 0.435,
+      "step": 833
+    },
+    {
+      "epoch": 3.9310344827586206,
+      "grad_norm": 0.5333203077316284,
+      "learning_rate": 1.5340123181227495e-06,
+      "loss": 0.4245,
+      "step": 834
+    },
+    {
+      "epoch": 3.9357366771159876,
+      "grad_norm": 0.5764657855033875,
+      "learning_rate": 1.527834917683058e-06,
+      "loss": 0.3909,
+      "step": 835
+    },
+    {
+      "epoch": 3.9404388714733543,
+      "grad_norm": 0.6087055802345276,
+      "learning_rate": 1.5216645025307813e-06,
+      "loss": 0.4059,
+      "step": 836
+    },
+    {
+      "epoch": 3.945141065830721,
+      "grad_norm": 0.558167576789856,
+      "learning_rate": 1.5155011170021399e-06,
+      "loss": 0.4142,
+      "step": 837
+    },
+    {
+      "epoch": 3.9498432601880875,
+      "grad_norm": 0.5065332055091858,
+      "learning_rate": 1.5093448053828402e-06,
+      "loss": 0.4237,
+      "step": 838
+    },
+    {
+      "epoch": 3.9545454545454546,
+      "grad_norm": 0.5446473360061646,
+      "learning_rate": 1.503195611907764e-06,
+      "loss": 0.4235,
+      "step": 839
+    },
+    {
+      "epoch": 3.959247648902821,
+      "grad_norm": 0.6033996939659119,
+      "learning_rate": 1.4970535807606453e-06,
+      "loss": 0.384,
+      "step": 840
+    },
+    {
+      "epoch": 3.9639498432601883,
+      "grad_norm": 0.5554048418998718,
+      "learning_rate": 1.4909187560737542e-06,
+      "loss": 0.3953,
+      "step": 841
+    },
+    {
+      "epoch": 3.968652037617555,
+      "grad_norm": 0.533640444278717,
+      "learning_rate": 1.4847911819275829e-06,
+      "loss": 0.4072,
+      "step": 842
+    },
+    {
+      "epoch": 3.9733542319749215,
+      "grad_norm": 0.5807821750640869,
+      "learning_rate": 1.4786709023505224e-06,
+      "loss": 0.3975,
+      "step": 843
+    },
+    {
+      "epoch": 3.978056426332288,
+      "grad_norm": 0.6907732486724854,
+      "learning_rate": 1.4725579613185549e-06,
+      "loss": 0.4232,
+      "step": 844
+    },
+    {
+      "epoch": 3.9827586206896552,
+      "grad_norm": 0.6130005717277527,
+      "learning_rate": 1.4664524027549291e-06,
+      "loss": 0.4094,
+      "step": 845
+    },
+    {
+      "epoch": 3.987460815047022,
+      "grad_norm": 0.6130813360214233,
+      "learning_rate": 1.4603542705298493e-06,
+      "loss": 0.3954,
+      "step": 846
+    },
+    {
+      "epoch": 3.992163009404389,
+      "grad_norm": 0.5968356132507324,
+      "learning_rate": 1.4542636084601624e-06,
+      "loss": 0.3675,
+      "step": 847
+    },
+    {
+      "epoch": 3.9968652037617556,
+      "grad_norm": 0.6450978517532349,
+      "learning_rate": 1.4481804603090358e-06,
+      "loss": 0.4113,
+      "step": 848
+    },
+    {
+      "epoch": 4.004702194357367,
+      "grad_norm": 1.4493160247802734,
+      "learning_rate": 1.4421048697856494e-06,
+      "loss": 0.7551,
+      "step": 849
+    },
+    {
+      "epoch": 4.009404388714733,
+      "grad_norm": 0.5255843997001648,
+      "learning_rate": 1.4360368805448788e-06,
+      "loss": 0.3801,
+      "step": 850
+    },
+    {
+      "epoch": 4.0141065830721,
+      "grad_norm": 0.5161401629447937,
+      "learning_rate": 1.4299765361869837e-06,
+      "loss": 0.3959,
+      "step": 851
+    },
+    {
+      "epoch": 4.018808777429467,
+      "grad_norm": 0.5007433295249939,
+      "learning_rate": 1.4239238802572908e-06,
+      "loss": 0.3553,
+      "step": 852
+    },
+    {
+      "epoch": 4.023510971786834,
+      "grad_norm": 0.5718136429786682,
+      "learning_rate": 1.4178789562458847e-06,
+      "loss": 0.3937,
+      "step": 853
+    },
+    {
+      "epoch": 4.028213166144201,
+      "grad_norm": 0.5353648066520691,
+      "learning_rate": 1.4118418075872936e-06,
+      "loss": 0.3816,
+      "step": 854
+    },
+    {
+      "epoch": 4.032915360501567,
+      "grad_norm": 0.5066031813621521,
+      "learning_rate": 1.405812477660178e-06,
+      "loss": 0.3774,
+      "step": 855
+    },
+    {
+      "epoch": 4.037617554858934,
+      "grad_norm": 0.5610586404800415,
+      "learning_rate": 1.3997910097870165e-06,
+      "loss": 0.4023,
+      "step": 856
+    },
+    {
+      "epoch": 4.0423197492163006,
+      "grad_norm": 0.5356720089912415,
+      "learning_rate": 1.3937774472337994e-06,
+      "loss": 0.4041,
+      "step": 857
+    },
+    {
+      "epoch": 4.047021943573668,
+      "grad_norm": 0.5321263670921326,
+      "learning_rate": 1.3877718332097146e-06,
+      "loss": 0.3905,
+      "step": 858
+    },
+    {
+      "epoch": 4.051724137931035,
+      "grad_norm": 0.5932567119598389,
+      "learning_rate": 1.3817742108668333e-06,
+      "loss": 0.3989,
+      "step": 859
+    },
+    {
+      "epoch": 4.056426332288401,
+      "grad_norm": 0.5155854225158691,
+      "learning_rate": 1.3757846232998118e-06,
+      "loss": 0.3369,
+      "step": 860
+    },
+    {
+      "epoch": 4.061128526645768,
+      "grad_norm": 0.9726412892341614,
+      "learning_rate": 1.369803113545566e-06,
+      "loss": 0.4116,
+      "step": 861
+    },
+    {
+      "epoch": 4.065830721003135,
+      "grad_norm": 0.5043352842330933,
+      "learning_rate": 1.3638297245829762e-06,
+      "loss": 0.4054,
+      "step": 862
+    },
+    {
+      "epoch": 4.070532915360501,
+      "grad_norm": 0.531753659248352,
+      "learning_rate": 1.3578644993325701e-06,
+      "loss": 0.4036,
+      "step": 863
+    },
+    {
+      "epoch": 4.075235109717869,
+      "grad_norm": 0.5960082411766052,
+      "learning_rate": 1.3519074806562165e-06,
+      "loss": 0.3946,
+      "step": 864
+    },
+    {
+      "epoch": 4.079937304075235,
+      "grad_norm": 0.5632782578468323,
+      "learning_rate": 1.3459587113568208e-06,
+      "loss": 0.3951,
+      "step": 865
+    },
+    {
+      "epoch": 4.084639498432602,
+      "grad_norm": 0.5298274755477905,
+      "learning_rate": 1.340018234178009e-06,
+      "loss": 0.3962,
+      "step": 866
+    },
+    {
+      "epoch": 4.089341692789969,
+      "grad_norm": 0.5057002902030945,
+      "learning_rate": 1.3340860918038295e-06,
+      "loss": 0.3635,
+      "step": 867
+    },
+    {
+      "epoch": 4.094043887147335,
+      "grad_norm": 0.5192775130271912,
+      "learning_rate": 1.328162326858442e-06,
+      "loss": 0.3871,
+      "step": 868
+    },
+    {
+      "epoch": 4.098746081504702,
+      "grad_norm": 0.5322587490081787,
+      "learning_rate": 1.3222469819058112e-06,
+      "loss": 0.3987,
+      "step": 869
+    },
+    {
+      "epoch": 4.103448275862069,
+      "grad_norm": 0.5546263456344604,
+      "learning_rate": 1.3163400994494025e-06,
+      "loss": 0.3965,
+      "step": 870
+    },
+    {
+      "epoch": 4.108150470219436,
+      "grad_norm": 0.5492004156112671,
+      "learning_rate": 1.3104417219318762e-06,
+      "loss": 0.3837,
+      "step": 871
+    },
+    {
+      "epoch": 4.112852664576803,
+      "grad_norm": 0.5725740790367126,
+      "learning_rate": 1.3045518917347791e-06,
+      "loss": 0.3935,
+      "step": 872
+    },
+    {
+      "epoch": 4.117554858934169,
+      "grad_norm": 0.6411507725715637,
+      "learning_rate": 1.2986706511782476e-06,
+      "loss": 0.3881,
+      "step": 873
+    },
+    {
+      "epoch": 4.122257053291536,
+      "grad_norm": 0.5744498372077942,
+      "learning_rate": 1.2927980425206968e-06,
+      "loss": 0.3913,
+      "step": 874
+    },
+    {
+      "epoch": 4.1269592476489025,
+      "grad_norm": 0.5495030283927917,
+      "learning_rate": 1.2869341079585184e-06,
+      "loss": 0.3919,
+      "step": 875
+    },
+    {
+      "epoch": 4.131661442006269,
+      "grad_norm": 0.5531148910522461,
+      "learning_rate": 1.2810788896257804e-06,
+      "loss": 0.3603,
+      "step": 876
+    },
+    {
+      "epoch": 4.136363636363637,
+      "grad_norm": 0.5476847290992737,
+      "learning_rate": 1.2752324295939178e-06,
+      "loss": 0.3956,
+      "step": 877
+    },
+    {
+      "epoch": 4.141065830721003,
+      "grad_norm": 0.5382871627807617,
+      "learning_rate": 1.2693947698714409e-06,
+      "loss": 0.4183,
+      "step": 878
+    },
+    {
+      "epoch": 4.14576802507837,
+      "grad_norm": 0.6912056803703308,
+      "learning_rate": 1.263565952403622e-06,
+      "loss": 0.383,
+      "step": 879
+    },
+    {
+      "epoch": 4.150470219435737,
+      "grad_norm": 0.5323772430419922,
+      "learning_rate": 1.2577460190722013e-06,
+      "loss": 0.3964,
+      "step": 880
+    },
+    {
+      "epoch": 4.155172413793103,
+      "grad_norm": 0.5875644683837891,
+      "learning_rate": 1.2519350116950842e-06,
+      "loss": 0.4001,
+      "step": 881
+    },
+    {
+      "epoch": 4.15987460815047,
+      "grad_norm": 0.5375142097473145,
+      "learning_rate": 1.2461329720260403e-06,
+      "loss": 0.3871,
+      "step": 882
+    },
+    {
+      "epoch": 4.164576802507837,
+      "grad_norm": 2.5649774074554443,
+      "learning_rate": 1.2403399417544033e-06,
+      "loss": 0.3983,
+      "step": 883
+    },
+    {
+      "epoch": 4.169278996865204,
+      "grad_norm": 0.5615472793579102,
+      "learning_rate": 1.2345559625047718e-06,
+      "loss": 0.404,
+      "step": 884
+    },
+    {
+      "epoch": 4.173981191222571,
+      "grad_norm": 0.567313015460968,
+      "learning_rate": 1.2287810758367104e-06,
+      "loss": 0.4089,
+      "step": 885
+    },
+    {
+      "epoch": 4.178683385579937,
+      "grad_norm": 0.718047559261322,
+      "learning_rate": 1.2230153232444511e-06,
+      "loss": 0.3899,
+      "step": 886
+    },
+    {
+      "epoch": 4.183385579937304,
+      "grad_norm": 0.5390134453773499,
+      "learning_rate": 1.217258746156594e-06,
+      "loss": 0.3802,
+      "step": 887
+    },
+    {
+      "epoch": 4.1880877742946705,
+      "grad_norm": 0.502026379108429,
+      "learning_rate": 1.2115113859358118e-06,
+      "loss": 0.3855,
+      "step": 888
+    },
+    {
+      "epoch": 4.192789968652038,
+      "grad_norm": 0.5020443797111511,
+      "learning_rate": 1.2057732838785514e-06,
+      "loss": 0.368,
+      "step": 889
+    },
+    {
+      "epoch": 4.197492163009405,
+      "grad_norm": 0.5389068126678467,
+      "learning_rate": 1.2000444812147333e-06,
+      "loss": 0.3808,
+      "step": 890
+    },
+    {
+      "epoch": 4.202194357366771,
+      "grad_norm": 0.5966452360153198,
+      "learning_rate": 1.1943250191074664e-06,
+      "loss": 0.3995,
+      "step": 891
+    },
+    {
+      "epoch": 4.206896551724138,
+      "grad_norm": 0.5827345848083496,
+      "learning_rate": 1.188614938652738e-06,
+      "loss": 0.4069,
+      "step": 892
+    },
+    {
+      "epoch": 4.2115987460815045,
+      "grad_norm": 0.6623309254646301,
+      "learning_rate": 1.1829142808791294e-06,
+      "loss": 0.3683,
+      "step": 893
+    },
+    {
+      "epoch": 4.216300940438871,
+      "grad_norm": 0.523257315158844,
+      "learning_rate": 1.177223086747516e-06,
+      "loss": 0.3805,
+      "step": 894
+    },
+    {
+      "epoch": 4.221003134796238,
+      "grad_norm": 0.5709425806999207,
+      "learning_rate": 1.1715413971507747e-06,
+      "loss": 0.3942,
+      "step": 895
+    },
+    {
+      "epoch": 4.225705329153605,
+      "grad_norm": 1.240659475326538,
+      "learning_rate": 1.1658692529134888e-06,
+      "loss": 0.3796,
+      "step": 896
+    },
+    {
+      "epoch": 4.230407523510972,
+      "grad_norm": 0.565609335899353,
+      "learning_rate": 1.1602066947916565e-06,
+      "loss": 0.4021,
+      "step": 897
+    },
+    {
+      "epoch": 4.235109717868339,
+      "grad_norm": 0.6735373139381409,
+      "learning_rate": 1.154553763472396e-06,
+      "loss": 0.3477,
+      "step": 898
+    },
+    {
+      "epoch": 4.239811912225705,
+      "grad_norm": 0.8588801026344299,
+      "learning_rate": 1.1489104995736543e-06,
+      "loss": 0.3809,
+      "step": 899
+    },
+    {
+      "epoch": 4.244514106583072,
+      "grad_norm": 0.5612583756446838,
+      "learning_rate": 1.1432769436439162e-06,
+      "loss": 0.3935,
+      "step": 900
+    },
+    {
+      "epoch": 4.2492163009404385,
+      "grad_norm": 0.49488264322280884,
+      "learning_rate": 1.1376531361619105e-06,
+      "loss": 0.4034,
+      "step": 901
+    },
+    {
+      "epoch": 4.253918495297806,
+      "grad_norm": 0.5561050176620483,
+      "learning_rate": 1.1320391175363225e-06,
+      "loss": 0.3779,
+      "step": 902
+    },
+    {
+      "epoch": 4.258620689655173,
+      "grad_norm": 0.5452584624290466,
+      "learning_rate": 1.126434928105497e-06,
+      "loss": 0.3842,
+      "step": 903
+    },
+    {
+      "epoch": 4.263322884012539,
+      "grad_norm": 0.6411417126655579,
+      "learning_rate": 1.1208406081371612e-06,
+      "loss": 0.3875,
+      "step": 904
+    },
+    {
+      "epoch": 4.268025078369906,
+      "grad_norm": 0.5172016024589539,
+      "learning_rate": 1.11525619782812e-06,
+      "loss": 0.3999,
+      "step": 905
+    },
+    {
+      "epoch": 4.2727272727272725,
+      "grad_norm": 0.5747371912002563,
+      "learning_rate": 1.1096817373039773e-06,
+      "loss": 0.4031,
+      "step": 906
+    },
+    {
+      "epoch": 4.277429467084639,
+      "grad_norm": 0.5466158390045166,
+      "learning_rate": 1.104117266618846e-06,
+      "loss": 0.3962,
+      "step": 907
+    },
+    {
+      "epoch": 4.282131661442007,
+      "grad_norm": 0.6000977158546448,
+      "learning_rate": 1.0985628257550575e-06,
+      "loss": 0.384,
+      "step": 908
+    },
+    {
+      "epoch": 4.286833855799373,
+      "grad_norm": 0.543651819229126,
+      "learning_rate": 1.0930184546228769e-06,
+      "loss": 0.3908,
+      "step": 909
+    },
+    {
+      "epoch": 4.29153605015674,
+      "grad_norm": 0.5661547183990479,
+      "learning_rate": 1.087484193060215e-06,
+      "loss": 0.3624,
+      "step": 910
+    },
+    {
+      "epoch": 4.2962382445141065,
+      "grad_norm": 0.5777263045310974,
+      "learning_rate": 1.0819600808323424e-06,
+      "loss": 0.3981,
+      "step": 911
+    },
+    {
+      "epoch": 4.300940438871473,
+      "grad_norm": 0.6490275859832764,
+      "learning_rate": 1.0764461576316041e-06,
+      "loss": 0.3801,
+      "step": 912
+    },
+    {
+      "epoch": 4.30564263322884,
+      "grad_norm": 0.8035992980003357,
+      "learning_rate": 1.0709424630771333e-06,
+      "loss": 0.3973,
+      "step": 913
+    },
+    {
+      "epoch": 4.310344827586207,
+      "grad_norm": 0.603229820728302,
+      "learning_rate": 1.0654490367145684e-06,
+      "loss": 0.3878,
+      "step": 914
+    },
+    {
+      "epoch": 4.315047021943574,
+      "grad_norm": 0.5470160841941833,
+      "learning_rate": 1.0599659180157678e-06,
+      "loss": 0.4047,
+      "step": 915
+    },
+    {
+      "epoch": 4.3197492163009406,
+      "grad_norm": 5.193161487579346,
+      "learning_rate": 1.0544931463785237e-06,
+      "loss": 0.4272,
+      "step": 916
+    },
+    {
+      "epoch": 4.324451410658307,
+      "grad_norm": 0.5719714760780334,
+      "learning_rate": 1.049030761126287e-06,
+      "loss": 0.4009,
+      "step": 917
+    },
+    {
+      "epoch": 4.329153605015674,
+      "grad_norm": 0.577610969543457,
+      "learning_rate": 1.043578801507874e-06,
+      "loss": 0.3806,
+      "step": 918
+    },
+    {
+      "epoch": 4.33385579937304,
+      "grad_norm": 0.5377278923988342,
+      "learning_rate": 1.038137306697193e-06,
+      "loss": 0.402,
+      "step": 919
+    },
+    {
+      "epoch": 4.338557993730408,
+      "grad_norm": 0.6189285516738892,
+      "learning_rate": 1.0327063157929582e-06,
+      "loss": 0.3915,
+      "step": 920
+    },
+    {
+      "epoch": 4.343260188087775,
+      "grad_norm": 0.5795995593070984,
+      "learning_rate": 1.027285867818411e-06,
+      "loss": 0.3957,
+      "step": 921
+    },
+    {
+      "epoch": 4.347962382445141,
+      "grad_norm": 0.49627596139907837,
+      "learning_rate": 1.021876001721039e-06,
+      "loss": 0.3395,
+      "step": 922
+    },
+    {
+      "epoch": 4.352664576802508,
+      "grad_norm": 0.6332801580429077,
+      "learning_rate": 1.016476756372295e-06,
+      "loss": 0.3891,
+      "step": 923
+    },
+    {
+      "epoch": 4.3573667711598745,
+      "grad_norm": 0.5847935676574707,
+      "learning_rate": 1.011088170567319e-06,
+      "loss": 0.3783,
+      "step": 924
+    },
+    {
+      "epoch": 4.362068965517241,
+      "grad_norm": 0.6362956166267395,
+      "learning_rate": 1.0057102830246596e-06,
+      "loss": 0.372,
+      "step": 925
+    },
+    {
+      "epoch": 4.366771159874608,
+      "grad_norm": 0.549068033695221,
+      "learning_rate": 1.0003431323859943e-06,
+      "loss": 0.4023,
+      "step": 926
+    },
+    {
+      "epoch": 4.371473354231975,
+      "grad_norm": 0.5768754482269287,
+      "learning_rate": 9.949867572158544e-07,
+      "loss": 0.3876,
+      "step": 927
+    },
+    {
+      "epoch": 4.376175548589342,
+      "grad_norm": 0.7501256465911865,
+      "learning_rate": 9.896411960013455e-07,
+      "loss": 0.4011,
+      "step": 928
+    },
+    {
+      "epoch": 4.3808777429467085,
+      "grad_norm": 0.5689511895179749,
+      "learning_rate": 9.843064871518694e-07,
+      "loss": 0.3816,
+      "step": 929
+    },
+    {
+      "epoch": 4.385579937304075,
+      "grad_norm": 0.6056525111198425,
+      "learning_rate": 9.78982668998856e-07,
+      "loss": 0.3764,
+      "step": 930
+    },
+    {
+      "epoch": 4.390282131661442,
+      "grad_norm": 0.5028645396232605,
+      "learning_rate": 9.736697797954766e-07,
+      "loss": 0.3996,
+      "step": 931
+    },
+    {
+      "epoch": 4.394984326018808,
+      "grad_norm": 0.5384755730628967,
+      "learning_rate": 9.683678577163788e-07,
+      "loss": 0.3924,
+      "step": 932
+    },
+    {
+      "epoch": 4.399686520376176,
+      "grad_norm": 0.5420899391174316,
+      "learning_rate": 9.630769408574065e-07,
+      "loss": 0.3664,
+      "step": 933
+    },
+    {
+      "epoch": 4.4043887147335425,
+      "grad_norm": 1.1258292198181152,
+      "learning_rate": 9.577970672353274e-07,
+      "loss": 0.372,
+      "step": 934
+    },
+    {
+      "epoch": 4.409090909090909,
+      "grad_norm": 0.5689091682434082,
+      "learning_rate": 9.525282747875636e-07,
+      "loss": 0.3684,
+      "step": 935
+    },
+    {
+      "epoch": 4.413793103448276,
+      "grad_norm": 0.5591897964477539,
+      "learning_rate": 9.472706013719113e-07,
+      "loss": 0.3958,
+      "step": 936
+    },
+    {
+      "epoch": 4.418495297805642,
+      "grad_norm": 0.5757090449333191,
+      "learning_rate": 9.420240847662759e-07,
+      "loss": 0.3796,
+      "step": 937
+    },
+    {
+      "epoch": 4.423197492163009,
+      "grad_norm": 0.5598184466362,
+      "learning_rate": 9.367887626683975e-07,
+      "loss": 0.4052,
+      "step": 938
+    },
+    {
+      "epoch": 4.427899686520377,
+      "grad_norm": 0.573148787021637,
+      "learning_rate": 9.315646726955798e-07,
+      "loss": 0.3831,
+      "step": 939
+    },
+    {
+      "epoch": 4.432601880877743,
+      "grad_norm": 0.5196001529693604,
+      "learning_rate": 9.263518523844211e-07,
+      "loss": 0.382,
+      "step": 940
+    },
+    {
+      "epoch": 4.43730407523511,
+      "grad_norm": 0.5644182562828064,
+      "learning_rate": 9.211503391905446e-07,
+      "loss": 0.3858,
+      "step": 941
+    },
+    {
+      "epoch": 4.4420062695924765,
+      "grad_norm": 0.5429992079734802,
+      "learning_rate": 9.159601704883253e-07,
+      "loss": 0.3919,
+      "step": 942
+    },
+    {
+      "epoch": 4.446708463949843,
+      "grad_norm": 0.5335330367088318,
+      "learning_rate": 9.107813835706303e-07,
+      "loss": 0.3594,
+      "step": 943
+    },
+    {
+      "epoch": 4.45141065830721,
+      "grad_norm": 0.5579376816749573,
+      "learning_rate": 9.056140156485385e-07,
+      "loss": 0.3768,
+      "step": 944
+    },
+    {
+      "epoch": 4.456112852664576,
+      "grad_norm": 1.2909058332443237,
+      "learning_rate": 9.004581038510865e-07,
+      "loss": 0.3877,
+      "step": 945
+    },
+    {
+      "epoch": 4.460815047021944,
+      "grad_norm": 0.5664709806442261,
+      "learning_rate": 8.953136852249922e-07,
+      "loss": 0.4057,
+      "step": 946
+    },
+    {
+      "epoch": 4.4655172413793105,
+      "grad_norm": 0.6413213610649109,
+      "learning_rate": 8.901807967343898e-07,
+      "loss": 0.4006,
+      "step": 947
+    },
+    {
+      "epoch": 4.470219435736677,
+      "grad_norm": 0.5393944978713989,
+      "learning_rate": 8.850594752605712e-07,
+      "loss": 0.3957,
+      "step": 948
+    },
+    {
+      "epoch": 4.474921630094044,
+      "grad_norm": 0.6381353139877319,
+      "learning_rate": 8.79949757601711e-07,
+      "loss": 0.3861,
+      "step": 949
+    },
+    {
+      "epoch": 4.47962382445141,
+      "grad_norm": 0.5642807483673096,
+      "learning_rate": 8.748516804726096e-07,
+      "loss": 0.3867,
+      "step": 950
+    },
+    {
+      "epoch": 4.484326018808777,
+      "grad_norm": 0.5574389696121216,
+      "learning_rate": 8.697652805044265e-07,
+      "loss": 0.3644,
+      "step": 951
+    },
+    {
+      "epoch": 4.4890282131661445,
+      "grad_norm": 0.5668476223945618,
+      "learning_rate": 8.646905942444172e-07,
+      "loss": 0.3735,
+      "step": 952
+    },
+    {
+      "epoch": 4.493730407523511,
+      "grad_norm": 0.5889971256256104,
+      "learning_rate": 8.59627658155671e-07,
+      "loss": 0.392,
+      "step": 953
+    },
+    {
+      "epoch": 4.498432601880878,
+      "grad_norm": 0.5788328647613525,
+      "learning_rate": 8.545765086168484e-07,
+      "loss": 0.3845,
+      "step": 954
+    },
+    {
+      "epoch": 4.503134796238244,
+      "grad_norm": 0.9370825886726379,
+      "learning_rate": 8.495371819219206e-07,
+      "loss": 0.3985,
+      "step": 955
+    },
+    {
+      "epoch": 4.507836990595611,
+      "grad_norm": 0.5701330900192261,
+      "learning_rate": 8.44509714279908e-07,
+      "loss": 0.4099,
+      "step": 956
+    },
+    {
+      "epoch": 4.512539184952978,
+      "grad_norm": 0.5952461361885071,
+      "learning_rate": 8.394941418146202e-07,
+      "loss": 0.4003,
+      "step": 957
+    },
+    {
+      "epoch": 4.517241379310345,
+      "grad_norm": 0.5201880931854248,
+      "learning_rate": 8.344905005643967e-07,
+      "loss": 0.4023,
+      "step": 958
+    },
+    {
+      "epoch": 4.521943573667712,
+      "grad_norm": 0.5678507089614868,
+      "learning_rate": 8.294988264818488e-07,
+      "loss": 0.3913,
+      "step": 959
+    },
+    {
+      "epoch": 4.5266457680250785,
+      "grad_norm": 0.5855143070220947,
+      "learning_rate": 8.245191554335963e-07,
+      "loss": 0.3852,
+      "step": 960
+    },
+    {
+      "epoch": 4.531347962382445,
+      "grad_norm": 0.5465763807296753,
+      "learning_rate": 8.1955152320002e-07,
+      "loss": 0.387,
+      "step": 961
+    },
+    {
+      "epoch": 4.536050156739812,
+      "grad_norm": 0.5709990859031677,
+      "learning_rate": 8.145959654749924e-07,
+      "loss": 0.3997,
+      "step": 962
+    },
+    {
+      "epoch": 4.540752351097178,
+      "grad_norm": 0.8095040321350098,
+      "learning_rate": 8.096525178656306e-07,
+      "loss": 0.3686,
+      "step": 963
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 0.5552055239677429,
+      "learning_rate": 8.047212158920362e-07,
+      "loss": 0.3958,
+      "step": 964
+    },
+    {
+      "epoch": 4.5501567398119125,
+      "grad_norm": 0.5294213891029358,
+      "learning_rate": 7.998020949870402e-07,
+      "loss": 0.4134,
+      "step": 965
+    },
+    {
+      "epoch": 4.554858934169279,
+      "grad_norm": 0.5582411289215088,
+      "learning_rate": 7.948951904959504e-07,
+      "loss": 0.3799,
+      "step": 966
+    },
+    {
+      "epoch": 4.559561128526646,
+      "grad_norm": 0.5506302714347839,
+      "learning_rate": 7.900005376762948e-07,
+      "loss": 0.3899,
+      "step": 967
+    },
+    {
+      "epoch": 4.564263322884012,
+      "grad_norm": 0.5499927997589111,
+      "learning_rate": 7.851181716975703e-07,
+      "loss": 0.398,
+      "step": 968
+    },
+    {
+      "epoch": 4.568965517241379,
+      "grad_norm": 0.530657172203064,
+      "learning_rate": 7.802481276409896e-07,
+      "loss": 0.3622,
+      "step": 969
+    },
+    {
+      "epoch": 4.5736677115987465,
+      "grad_norm": 0.9928011894226074,
+      "learning_rate": 7.75390440499228e-07,
+      "loss": 0.3735,
+      "step": 970
+    },
+    {
+      "epoch": 4.578369905956113,
+      "grad_norm": 0.6327497363090515,
+      "learning_rate": 7.705451451761734e-07,
+      "loss": 0.3686,
+      "step": 971
+    },
+    {
+      "epoch": 4.58307210031348,
+      "grad_norm": 1.5083693265914917,
+      "learning_rate": 7.657122764866754e-07,
+      "loss": 0.3706,
+      "step": 972
+    },
+    {
+      "epoch": 4.587774294670846,
+      "grad_norm": 0.5787728428840637,
+      "learning_rate": 7.608918691562914e-07,
+      "loss": 0.4063,
+      "step": 973
+    },
+    {
+      "epoch": 4.592476489028213,
+      "grad_norm": 0.580129861831665,
+      "learning_rate": 7.560839578210466e-07,
+      "loss": 0.3693,
+      "step": 974
+    },
+    {
+      "epoch": 4.59717868338558,
+      "grad_norm": 0.5664162635803223,
+      "learning_rate": 7.512885770271722e-07,
+      "loss": 0.393,
+      "step": 975
+    },
+    {
+      "epoch": 4.601880877742946,
+      "grad_norm": 0.6353042125701904,
+      "learning_rate": 7.465057612308676e-07,
+      "loss": 0.3902,
+      "step": 976
+    },
+    {
+      "epoch": 4.606583072100314,
+      "grad_norm": 0.5410131812095642,
+      "learning_rate": 7.417355447980484e-07,
+      "loss": 0.3948,
+      "step": 977
+    },
+    {
+      "epoch": 4.61128526645768,
+      "grad_norm": 0.5826568603515625,
+      "learning_rate": 7.369779620041001e-07,
+      "loss": 0.3828,
+      "step": 978
+    },
+    {
+      "epoch": 4.615987460815047,
+      "grad_norm": 0.5563565492630005,
+      "learning_rate": 7.322330470336314e-07,
+      "loss": 0.4095,
+      "step": 979
+    },
+    {
+      "epoch": 4.620689655172414,
+      "grad_norm": 0.5592769980430603,
+      "learning_rate": 7.275008339802295e-07,
+      "loss": 0.3924,
+      "step": 980
+    },
+    {
+      "epoch": 4.62539184952978,
+      "grad_norm": 0.744817316532135,
+      "learning_rate": 7.227813568462141e-07,
+      "loss": 0.3737,
+      "step": 981
+    },
+    {
+      "epoch": 4.630094043887147,
+      "grad_norm": 0.5768243074417114,
+      "learning_rate": 7.180746495423946e-07,
+      "loss": 0.3906,
+      "step": 982
+    },
+    {
+      "epoch": 4.6347962382445145,
+      "grad_norm": 0.5390825867652893,
+      "learning_rate": 7.133807458878247e-07,
+      "loss": 0.3872,
+      "step": 983
+    },
+    {
+      "epoch": 4.639498432601881,
+      "grad_norm": 0.5862735509872437,
+      "learning_rate": 7.086996796095599e-07,
+      "loss": 0.3839,
+      "step": 984
+    },
+    {
+      "epoch": 4.644200626959248,
+      "grad_norm": 0.5880054235458374,
+      "learning_rate": 7.040314843424173e-07,
+      "loss": 0.3936,
+      "step": 985
+    },
+    {
+      "epoch": 4.648902821316614,
+      "grad_norm": 0.522564172744751,
+      "learning_rate": 6.99376193628728e-07,
+      "loss": 0.3827,
+      "step": 986
+    },
+    {
+      "epoch": 4.653605015673981,
+      "grad_norm": 0.5806529521942139,
+      "learning_rate": 6.947338409181056e-07,
+      "loss": 0.387,
+      "step": 987
+    },
+    {
+      "epoch": 4.658307210031348,
+      "grad_norm": 0.5591483116149902,
+      "learning_rate": 6.90104459567196e-07,
+      "loss": 0.3896,
+      "step": 988
+    },
+    {
+      "epoch": 4.663009404388715,
+      "grad_norm": 0.8104233145713806,
+      "learning_rate": 6.854880828394442e-07,
+      "loss": 0.3795,
+      "step": 989
+    },
+    {
+      "epoch": 4.667711598746082,
+      "grad_norm": 0.5946338772773743,
+      "learning_rate": 6.808847439048524e-07,
+      "loss": 0.4052,
+      "step": 990
+    },
+    {
+      "epoch": 4.672413793103448,
+      "grad_norm": 0.6517843008041382,
+      "learning_rate": 6.762944758397432e-07,
+      "loss": 0.3922,
+      "step": 991
+    },
+    {
+      "epoch": 4.677115987460815,
+      "grad_norm": 0.526491641998291,
+      "learning_rate": 6.717173116265208e-07,
+      "loss": 0.3853,
+      "step": 992
+    },
+    {
+      "epoch": 4.681818181818182,
+      "grad_norm": 0.5598382353782654,
+      "learning_rate": 6.671532841534345e-07,
+      "loss": 0.3931,
+      "step": 993
+    },
+    {
+      "epoch": 4.686520376175548,
+      "grad_norm": 0.6231300830841064,
+      "learning_rate": 6.626024262143421e-07,
+      "loss": 0.401,
+      "step": 994
+    },
+    {
+      "epoch": 4.691222570532915,
+      "grad_norm": 0.5638094544410706,
+      "learning_rate": 6.58064770508475e-07,
+      "loss": 0.3838,
+      "step": 995
+    },
+    {
+      "epoch": 4.695924764890282,
+      "grad_norm": 0.5840059518814087,
+      "learning_rate": 6.535403496402023e-07,
+      "loss": 0.3726,
+      "step": 996
+    },
+    {
+      "epoch": 4.700626959247649,
+      "grad_norm": 0.5452519655227661,
+      "learning_rate": 6.490291961187975e-07,
+      "loss": 0.3735,
+      "step": 997
+    },
+    {
+      "epoch": 4.705329153605016,
+      "grad_norm": 0.6175218224525452,
+      "learning_rate": 6.445313423582039e-07,
+      "loss": 0.4116,
+      "step": 998
+    },
+    {
+      "epoch": 4.710031347962382,
+      "grad_norm": 0.5606091022491455,
+      "learning_rate": 6.400468206768004e-07,
+      "loss": 0.3907,
+      "step": 999
+    },
+    {
+      "epoch": 4.714733542319749,
+      "grad_norm": 0.5126330256462097,
+      "learning_rate": 6.35575663297176e-07,
+      "loss": 0.3978,
+      "step": 1000
+    },
+    {
+      "epoch": 4.7194357366771165,
+      "grad_norm": 0.542105495929718,
+      "learning_rate": 6.31117902345888e-07,
+      "loss": 0.3408,
+      "step": 1001
+    },
+    {
+      "epoch": 4.724137931034483,
+      "grad_norm": 0.7690749168395996,
+      "learning_rate": 6.266735698532392e-07,
+      "loss": 0.3865,
+      "step": 1002
+    },
+    {
+      "epoch": 4.72884012539185,
+      "grad_norm": 0.5448989868164062,
+      "learning_rate": 6.222426977530449e-07,
+      "loss": 0.4078,
+      "step": 1003
+    },
+    {
+      "epoch": 4.733542319749216,
+      "grad_norm": 0.5502550601959229,
+      "learning_rate": 6.178253178824029e-07,
+      "loss": 0.3987,
+      "step": 1004
+    },
+    {
+      "epoch": 4.738244514106583,
+      "grad_norm": 0.5256698131561279,
+      "learning_rate": 6.134214619814657e-07,
+      "loss": 0.3797,
+      "step": 1005
+    },
+    {
+      "epoch": 4.74294670846395,
+      "grad_norm": 0.5733087062835693,
+      "learning_rate": 6.090311616932127e-07,
+      "loss": 0.3729,
+      "step": 1006
+    },
+    {
+      "epoch": 4.747648902821316,
+      "grad_norm": 0.5204405188560486,
+      "learning_rate": 6.04654448563221e-07,
+      "loss": 0.3798,
+      "step": 1007
+    },
+    {
+      "epoch": 4.752351097178684,
+      "grad_norm": 0.5861319899559021,
+      "learning_rate": 6.002913540394417e-07,
+      "loss": 0.3605,
+      "step": 1008
+    },
+    {
+      "epoch": 4.75705329153605,
+      "grad_norm": 0.5983589887619019,
+      "learning_rate": 5.959419094719713e-07,
+      "loss": 0.406,
+      "step": 1009
+    },
+    {
+      "epoch": 4.761755485893417,
+      "grad_norm": 0.5816555023193359,
+      "learning_rate": 5.916061461128269e-07,
+      "loss": 0.3813,
+      "step": 1010
+    },
+    {
+      "epoch": 4.766457680250784,
+      "grad_norm": 0.6101461052894592,
+      "learning_rate": 5.872840951157241e-07,
+      "loss": 0.3656,
+      "step": 1011
+    },
+    {
+      "epoch": 4.77115987460815,
+      "grad_norm": 0.5770082473754883,
+      "learning_rate": 5.829757875358477e-07,
+      "loss": 0.3824,
+      "step": 1012
+    },
+    {
+      "epoch": 4.775862068965517,
+      "grad_norm": 0.5549518465995789,
+      "learning_rate": 5.786812543296372e-07,
+      "loss": 0.3919,
+      "step": 1013
+    },
+    {
+      "epoch": 4.7805642633228835,
+      "grad_norm": 0.5969197750091553,
+      "learning_rate": 5.744005263545538e-07,
+      "loss": 0.4094,
+      "step": 1014
+    },
+    {
+      "epoch": 4.785266457680251,
+      "grad_norm": 0.6006177663803101,
+      "learning_rate": 5.701336343688671e-07,
+      "loss": 0.4076,
+      "step": 1015
+    },
+    {
+      "epoch": 4.789968652037618,
+      "grad_norm": 0.5562496185302734,
+      "learning_rate": 5.658806090314322e-07,
+      "loss": 0.3756,
+      "step": 1016
+    },
+    {
+      "epoch": 4.794670846394984,
+      "grad_norm": 0.619269609451294,
+      "learning_rate": 5.616414809014647e-07,
+      "loss": 0.3803,
+      "step": 1017
+    },
+    {
+      "epoch": 4.799373040752351,
+      "grad_norm": 0.5273744463920593,
+      "learning_rate": 5.574162804383293e-07,
+      "loss": 0.3903,
+      "step": 1018
+    },
+    {
+      "epoch": 4.804075235109718,
+      "grad_norm": 0.5756296515464783,
+      "learning_rate": 5.532050380013115e-07,
+      "loss": 0.3834,
+      "step": 1019
+    },
+    {
+      "epoch": 4.808777429467085,
+      "grad_norm": 0.6292474269866943,
+      "learning_rate": 5.490077838494079e-07,
+      "loss": 0.4114,
+      "step": 1020
+    },
+    {
+      "epoch": 4.813479623824452,
+      "grad_norm": 0.5505243539810181,
+      "learning_rate": 5.448245481411041e-07,
+      "loss": 0.3919,
+      "step": 1021
+    },
+    {
+      "epoch": 4.818181818181818,
+      "grad_norm": 0.5497763156890869,
+      "learning_rate": 5.406553609341586e-07,
+      "loss": 0.3913,
+      "step": 1022
+    },
+    {
+      "epoch": 4.822884012539185,
+      "grad_norm": 0.5497800707817078,
+      "learning_rate": 5.365002521853882e-07,
+      "loss": 0.3753,
+      "step": 1023
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 0.5281540155410767,
+      "learning_rate": 5.32359251750452e-07,
+      "loss": 0.3828,
+      "step": 1024
+    },
+    {
+      "epoch": 4.832288401253918,
+      "grad_norm": 0.5442911982536316,
+      "learning_rate": 5.282323893836347e-07,
+      "loss": 0.4074,
+      "step": 1025
+    },
+    {
+      "epoch": 4.836990595611285,
+      "grad_norm": 0.5267000794410706,
+      "learning_rate": 5.241196947376382e-07,
+      "loss": 0.3844,
+      "step": 1026
+    },
+    {
+      "epoch": 4.841692789968652,
+      "grad_norm": 0.5562646985054016,
+      "learning_rate": 5.200211973633632e-07,
+      "loss": 0.4098,
+      "step": 1027
+    },
+    {
+      "epoch": 4.846394984326019,
+      "grad_norm": 0.53923499584198,
+      "learning_rate": 5.15936926709699e-07,
+      "loss": 0.3987,
+      "step": 1028
+    },
+    {
+      "epoch": 4.851097178683386,
+      "grad_norm": 0.5662857890129089,
+      "learning_rate": 5.118669121233127e-07,
+      "loss": 0.3928,
+      "step": 1029
+    },
+    {
+      "epoch": 4.855799373040752,
+      "grad_norm": 0.5717837810516357,
+      "learning_rate": 5.078111828484347e-07,
+      "loss": 0.3914,
+      "step": 1030
+    },
+    {
+      "epoch": 4.860501567398119,
+      "grad_norm": 0.6961056590080261,
+      "learning_rate": 5.037697680266565e-07,
+      "loss": 0.394,
+      "step": 1031
+    },
+    {
+      "epoch": 4.8652037617554855,
+      "grad_norm": 0.6069244742393494,
+      "learning_rate": 4.997426966967106e-07,
+      "loss": 0.3949,
+      "step": 1032
+    },
+    {
+      "epoch": 4.869905956112853,
+      "grad_norm": 0.51577228307724,
+      "learning_rate": 4.957299977942704e-07,
+      "loss": 0.3792,
+      "step": 1033
+    },
+    {
+      "epoch": 4.87460815047022,
+      "grad_norm": 0.5260924696922302,
+      "learning_rate": 4.917317001517389e-07,
+      "loss": 0.3851,
+      "step": 1034
+    },
+    {
+      "epoch": 4.879310344827586,
+      "grad_norm": 0.5808457732200623,
+      "learning_rate": 4.877478324980412e-07,
+      "loss": 0.4056,
+      "step": 1035
+    },
+    {
+      "epoch": 4.884012539184953,
+      "grad_norm": 0.5486719012260437,
+      "learning_rate": 4.837784234584194e-07,
+      "loss": 0.3748,
+      "step": 1036
+    },
+    {
+      "epoch": 4.88871473354232,
+      "grad_norm": 0.5316773653030396,
+      "learning_rate": 4.79823501554226e-07,
+      "loss": 0.3973,
+      "step": 1037
+    },
+    {
+      "epoch": 4.893416927899686,
+      "grad_norm": 0.5671694874763489,
+      "learning_rate": 4.7588309520271934e-07,
+      "loss": 0.3817,
+      "step": 1038
+    },
+    {
+      "epoch": 4.898119122257054,
+      "grad_norm": 0.5756440758705139,
+      "learning_rate": 4.7195723271685893e-07,
+      "loss": 0.4129,
+      "step": 1039
+    },
+    {
+      "epoch": 4.90282131661442,
+      "grad_norm": 0.5772919058799744,
+      "learning_rate": 4.6804594230510286e-07,
+      "loss": 0.4097,
+      "step": 1040
+    },
+    {
+      "epoch": 4.907523510971787,
+      "grad_norm": 0.5606109499931335,
+      "learning_rate": 4.641492520712043e-07,
+      "loss": 0.3859,
+      "step": 1041
+    },
+    {
+      "epoch": 4.912225705329154,
+      "grad_norm": 0.6109121441841125,
+      "learning_rate": 4.60267190014011e-07,
+      "loss": 0.3974,
+      "step": 1042
+    },
+    {
+      "epoch": 4.91692789968652,
+      "grad_norm": 0.5530210137367249,
+      "learning_rate": 4.563997840272602e-07,
+      "loss": 0.3813,
+      "step": 1043
+    },
+    {
+      "epoch": 4.921630094043887,
+      "grad_norm": 0.527423083782196,
+      "learning_rate": 4.5254706189938545e-07,
+      "loss": 0.3682,
+      "step": 1044
+    },
+    {
+      "epoch": 4.9263322884012535,
+      "grad_norm": 0.8316900134086609,
+      "learning_rate": 4.4870905131330827e-07,
+      "loss": 0.4091,
+      "step": 1045
+    },
+    {
+      "epoch": 4.931034482758621,
+      "grad_norm": 0.5423528552055359,
+      "learning_rate": 4.448857798462455e-07,
+      "loss": 0.4069,
+      "step": 1046
+    },
+    {
+      "epoch": 4.935736677115988,
+      "grad_norm": 0.590660572052002,
+      "learning_rate": 4.4107727496950913e-07,
+      "loss": 0.3789,
+      "step": 1047
+    },
+    {
+      "epoch": 4.940438871473354,
+      "grad_norm": 0.6766872406005859,
+      "learning_rate": 4.372835640483089e-07,
+      "loss": 0.4012,
+      "step": 1048
+    },
+    {
+      "epoch": 4.945141065830721,
+      "grad_norm": 0.6196874380111694,
+      "learning_rate": 4.3350467434155526e-07,
+      "loss": 0.3943,
+      "step": 1049
+    },
+    {
+      "epoch": 4.9498432601880875,
+      "grad_norm": 0.6170504689216614,
+      "learning_rate": 4.297406330016643e-07,
+      "loss": 0.3849,
+      "step": 1050
+    },
+    {
+      "epoch": 4.954545454545455,
+      "grad_norm": 0.6729781627655029,
+      "learning_rate": 4.25991467074362e-07,
+      "loss": 0.3746,
+      "step": 1051
+    },
+    {
+      "epoch": 4.959247648902822,
+      "grad_norm": 0.7265579700469971,
+      "learning_rate": 4.2225720349849063e-07,
+      "loss": 0.3881,
+      "step": 1052
+    },
+    {
+      "epoch": 4.963949843260188,
+      "grad_norm": 0.5406692624092102,
+      "learning_rate": 4.185378691058145e-07,
+      "loss": 0.3832,
+      "step": 1053
+    },
+    {
+      "epoch": 4.968652037617555,
+      "grad_norm": 0.6976690292358398,
+      "learning_rate": 4.148334906208273e-07,
+      "loss": 0.4138,
+      "step": 1054
+    },
+    {
+      "epoch": 4.9733542319749215,
+      "grad_norm": 0.5301910042762756,
+      "learning_rate": 4.1114409466056107e-07,
+      "loss": 0.3883,
+      "step": 1055
+    },
+    {
+      "epoch": 4.978056426332288,
+      "grad_norm": 0.5670992136001587,
+      "learning_rate": 4.0746970773439115e-07,
+      "loss": 0.4175,
+      "step": 1056
+    },
+    {
+      "epoch": 4.982758620689655,
+      "grad_norm": 0.5474900007247925,
+      "learning_rate": 4.0381035624385336e-07,
+      "loss": 0.4017,
+      "step": 1057
+    },
+    {
+      "epoch": 4.987460815047022,
+      "grad_norm": 0.5067923069000244,
+      "learning_rate": 4.0016606648244555e-07,
+      "loss": 0.3591,
+      "step": 1058
+    },
+    {
+      "epoch": 4.992163009404389,
+      "grad_norm": 0.6014583706855774,
+      "learning_rate": 3.9653686463544447e-07,
+      "loss": 0.4094,
+      "step": 1059
+    },
+    {
+      "epoch": 4.996865203761756,
+      "grad_norm": 0.549430251121521,
+      "learning_rate": 3.929227767797153e-07,
+      "loss": 0.4017,
+      "step": 1060
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1272,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 212,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.827573142662572e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}