diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7414 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1053,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002849002849002849,
+      "grad_norm": 56.0523474942412,
+      "learning_rate": 0.0,
+      "loss": 11.2375,
+      "step": 1
+    },
+    {
+      "epoch": 0.005698005698005698,
+      "grad_norm": 56.692305048985084,
+      "learning_rate": 4.7169811320754717e-07,
+      "loss": 11.1498,
+      "step": 2
+    },
+    {
+      "epoch": 0.008547008547008548,
+      "grad_norm": 56.20671215302259,
+      "learning_rate": 9.433962264150943e-07,
+      "loss": 11.1995,
+      "step": 3
+    },
+    {
+      "epoch": 0.011396011396011397,
+      "grad_norm": 57.07016050341919,
+      "learning_rate": 1.4150943396226415e-06,
+      "loss": 11.17,
+      "step": 4
+    },
+    {
+      "epoch": 0.014245014245014245,
+      "grad_norm": 55.67070270531962,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 11.2141,
+      "step": 5
+    },
+    {
+      "epoch": 0.017094017094017096,
+      "grad_norm": 67.41256511207831,
+      "learning_rate": 2.358490566037736e-06,
+      "loss": 10.8094,
+      "step": 6
+    },
+    {
+      "epoch": 0.019943019943019943,
+      "grad_norm": 66.98704800790512,
+      "learning_rate": 2.830188679245283e-06,
+      "loss": 10.6191,
+      "step": 7
+    },
+    {
+      "epoch": 0.022792022792022793,
+      "grad_norm": 99.53214132413315,
+      "learning_rate": 3.30188679245283e-06,
+      "loss": 9.2224,
+      "step": 8
+    },
+    {
+      "epoch": 0.02564102564102564,
+      "grad_norm": 107.3638343426545,
+      "learning_rate": 3.7735849056603773e-06,
+      "loss": 8.9747,
+      "step": 9
+    },
+    {
+      "epoch": 0.02849002849002849,
+      "grad_norm": 65.50392578974085,
+      "learning_rate": 4.245283018867925e-06,
+      "loss": 3.7694,
+      "step": 10
+    },
+    {
+      "epoch": 0.03133903133903134,
+      "grad_norm": 52.753132297048275,
+      "learning_rate": 4.716981132075472e-06,
+      "loss": 3.1517,
+      "step": 11
+    },
+    {
+      "epoch": 0.03418803418803419,
+      "grad_norm": 34.22470483345867,
+      "learning_rate": 5.188679245283019e-06,
+      "loss": 2.3701,
+      "step": 12
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 8.289999698115077,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 1.5114,
+      "step": 13
+    },
+    {
+      "epoch": 0.039886039886039885,
+      "grad_norm": 4.907971366358463,
+      "learning_rate": 6.132075471698113e-06,
+      "loss": 1.3167,
+      "step": 14
+    },
+    {
+      "epoch": 0.042735042735042736,
+      "grad_norm": 3.878187177104665,
+      "learning_rate": 6.60377358490566e-06,
+      "loss": 1.1878,
+      "step": 15
+    },
+    {
+      "epoch": 0.045584045584045586,
+      "grad_norm": 2.614057256962741,
+      "learning_rate": 7.0754716981132075e-06,
+      "loss": 1.134,
+      "step": 16
+    },
+    {
+      "epoch": 0.04843304843304843,
+      "grad_norm": 2.1563070672973015,
+      "learning_rate": 7.547169811320755e-06,
+      "loss": 1.0503,
+      "step": 17
+    },
+    {
+      "epoch": 0.05128205128205128,
+      "grad_norm": 5.014628550733802,
+      "learning_rate": 8.018867924528302e-06,
+      "loss": 0.968,
+      "step": 18
+    },
+    {
+      "epoch": 0.05413105413105413,
+      "grad_norm": 5.938244849443368,
+      "learning_rate": 8.49056603773585e-06,
+      "loss": 0.919,
+      "step": 19
+    },
+    {
+      "epoch": 0.05698005698005698,
+      "grad_norm": 1.2998505975680303,
+      "learning_rate": 8.962264150943396e-06,
+      "loss": 0.9074,
+      "step": 20
+    },
+    {
+      "epoch": 0.05982905982905983,
+      "grad_norm": 1.0231800381234573,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 0.864,
+      "step": 21
+    },
+    {
+      "epoch": 0.06267806267806268,
+      "grad_norm": 0.791900948928139,
+      "learning_rate": 9.905660377358492e-06,
+      "loss": 0.8147,
+      "step": 22
+    },
+    {
+      "epoch": 0.06552706552706553,
+      "grad_norm": 0.7091561135310548,
+      "learning_rate": 1.0377358490566038e-05,
+      "loss": 0.8188,
+      "step": 23
+    },
+    {
+      "epoch": 0.06837606837606838,
+      "grad_norm": 0.7166203845304848,
+      "learning_rate": 1.0849056603773586e-05,
+      "loss": 0.7601,
+      "step": 24
+    },
+    {
+      "epoch": 0.07122507122507123,
+      "grad_norm": 0.8012651357824443,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 0.7598,
+      "step": 25
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 0.6383142924108046,
+      "learning_rate": 1.179245283018868e-05,
+      "loss": 0.7514,
+      "step": 26
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 0.5725581821662165,
+      "learning_rate": 1.2264150943396227e-05,
+      "loss": 0.7071,
+      "step": 27
+    },
+    {
+      "epoch": 0.07977207977207977,
+      "grad_norm": 0.6425822567656265,
+      "learning_rate": 1.2735849056603775e-05,
+      "loss": 0.7076,
+      "step": 28
+    },
+    {
+      "epoch": 0.08262108262108261,
+      "grad_norm": 0.6708098823752852,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 0.6887,
+      "step": 29
+    },
+    {
+      "epoch": 0.08547008547008547,
+      "grad_norm": 0.5685632358364251,
+      "learning_rate": 1.3679245283018869e-05,
+      "loss": 0.6724,
+      "step": 30
+    },
+    {
+      "epoch": 0.08831908831908832,
+      "grad_norm": 0.44124658499047564,
+      "learning_rate": 1.4150943396226415e-05,
+      "loss": 0.6856,
+      "step": 31
+    },
+    {
+      "epoch": 0.09116809116809117,
+      "grad_norm": 0.46800011010506615,
+      "learning_rate": 1.4622641509433963e-05,
+      "loss": 0.6644,
+      "step": 32
+    },
+    {
+      "epoch": 0.09401709401709402,
+      "grad_norm": 0.5235273680352841,
+      "learning_rate": 1.509433962264151e-05,
+      "loss": 0.6491,
+      "step": 33
+    },
+    {
+      "epoch": 0.09686609686609686,
+      "grad_norm": 0.46201236009771707,
+      "learning_rate": 1.5566037735849056e-05,
+      "loss": 0.6403,
+      "step": 34
+    },
+    {
+      "epoch": 0.09971509971509972,
+      "grad_norm": 0.38923522274855954,
+      "learning_rate": 1.6037735849056604e-05,
+      "loss": 0.6325,
+      "step": 35
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 0.38662678714338583,
+      "learning_rate": 1.650943396226415e-05,
+      "loss": 0.6432,
+      "step": 36
+    },
+    {
+      "epoch": 0.10541310541310542,
+      "grad_norm": 0.40192513620663517,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 0.6126,
+      "step": 37
+    },
+    {
+      "epoch": 0.10826210826210826,
+      "grad_norm": 0.3948475707755113,
+      "learning_rate": 1.7452830188679244e-05,
+      "loss": 0.647,
+      "step": 38
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.3356078989163562,
+      "learning_rate": 1.7924528301886792e-05,
+      "loss": 0.6174,
+      "step": 39
+    },
+    {
+      "epoch": 0.11396011396011396,
+      "grad_norm": 0.30046759356107233,
+      "learning_rate": 1.839622641509434e-05,
+      "loss": 0.6072,
+      "step": 40
+    },
+    {
+      "epoch": 0.1168091168091168,
+      "grad_norm": 0.30888114755625345,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 0.5935,
+      "step": 41
+    },
+    {
+      "epoch": 0.11965811965811966,
+      "grad_norm": 0.30715993049476403,
+      "learning_rate": 1.9339622641509436e-05,
+      "loss": 0.6021,
+      "step": 42
+    },
+    {
+      "epoch": 0.1225071225071225,
+      "grad_norm": 0.29429080617445386,
+      "learning_rate": 1.9811320754716984e-05,
+      "loss": 0.5897,
+      "step": 43
+    },
+    {
+      "epoch": 0.12535612535612536,
+      "grad_norm": 0.2806609748601335,
+      "learning_rate": 2.0283018867924532e-05,
+      "loss": 0.5712,
+      "step": 44
+    },
+    {
+      "epoch": 0.1282051282051282,
+      "grad_norm": 0.3035543416731401,
+      "learning_rate": 2.0754716981132076e-05,
+      "loss": 0.5879,
+      "step": 45
+    },
+    {
+      "epoch": 0.13105413105413105,
+      "grad_norm": 0.2757782348628876,
+      "learning_rate": 2.1226415094339624e-05,
+      "loss": 0.5777,
+      "step": 46
+    },
+    {
+      "epoch": 0.1339031339031339,
+      "grad_norm": 0.2639801457025701,
+      "learning_rate": 2.1698113207547172e-05,
+      "loss": 0.5605,
+      "step": 47
+    },
+    {
+      "epoch": 0.13675213675213677,
+      "grad_norm": 0.2712816187794645,
+      "learning_rate": 2.216981132075472e-05,
+      "loss": 0.5873,
+      "step": 48
+    },
+    {
+      "epoch": 0.1396011396011396,
+      "grad_norm": 0.28871963465497535,
+      "learning_rate": 2.2641509433962265e-05,
+      "loss": 0.5727,
+      "step": 49
+    },
+    {
+      "epoch": 0.14245014245014245,
+      "grad_norm": 0.28060397078353755,
+      "learning_rate": 2.3113207547169813e-05,
+      "loss": 0.5602,
+      "step": 50
+    },
+    {
+      "epoch": 0.1452991452991453,
+      "grad_norm": 0.2771839138643321,
+      "learning_rate": 2.358490566037736e-05,
+      "loss": 0.5543,
+      "step": 51
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.25817545404116393,
+      "learning_rate": 2.405660377358491e-05,
+      "loss": 0.5495,
+      "step": 52
+    },
+    {
+      "epoch": 0.150997150997151,
+      "grad_norm": 0.2715327298990826,
+      "learning_rate": 2.4528301886792453e-05,
+      "loss": 0.5386,
+      "step": 53
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.3009269107284128,
+      "learning_rate": 2.5e-05,
+      "loss": 0.54,
+      "step": 54
+    },
+    {
+      "epoch": 0.15669515669515668,
+      "grad_norm": 0.26603937399372396,
+      "learning_rate": 2.547169811320755e-05,
+      "loss": 0.5688,
+      "step": 55
+    },
+    {
+      "epoch": 0.15954415954415954,
+      "grad_norm": 0.23331692549766264,
+      "learning_rate": 2.5943396226415094e-05,
+      "loss": 0.5408,
+      "step": 56
+    },
+    {
+      "epoch": 0.1623931623931624,
+      "grad_norm": 0.257528511255327,
+      "learning_rate": 2.641509433962264e-05,
+      "loss": 0.5451,
+      "step": 57
+    },
+    {
+      "epoch": 0.16524216524216523,
+      "grad_norm": 0.24892299069207233,
+      "learning_rate": 2.688679245283019e-05,
+      "loss": 0.5488,
+      "step": 58
+    },
+    {
+      "epoch": 0.16809116809116809,
+      "grad_norm": 0.22893417370533484,
+      "learning_rate": 2.7358490566037738e-05,
+      "loss": 0.5427,
+      "step": 59
+    },
+    {
+      "epoch": 0.17094017094017094,
+      "grad_norm": 0.2529033488136545,
+      "learning_rate": 2.7830188679245282e-05,
+      "loss": 0.5667,
+      "step": 60
+    },
+    {
+      "epoch": 0.1737891737891738,
+      "grad_norm": 0.2363533954890101,
+      "learning_rate": 2.830188679245283e-05,
+      "loss": 0.5535,
+      "step": 61
+    },
+    {
+      "epoch": 0.17663817663817663,
+      "grad_norm": 0.22892978820463591,
+      "learning_rate": 2.8773584905660378e-05,
+      "loss": 0.5482,
+      "step": 62
+    },
+    {
+      "epoch": 0.1794871794871795,
+      "grad_norm": 0.23882495609332313,
+      "learning_rate": 2.9245283018867926e-05,
+      "loss": 0.535,
+      "step": 63
+    },
+    {
+      "epoch": 0.18233618233618235,
+      "grad_norm": 0.2506420779209896,
+      "learning_rate": 2.971698113207547e-05,
+      "loss": 0.5454,
+      "step": 64
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.3063275398024107,
+      "learning_rate": 3.018867924528302e-05,
+      "loss": 0.5444,
+      "step": 65
+    },
+    {
+      "epoch": 0.18803418803418803,
+      "grad_norm": 0.23091797227336122,
+      "learning_rate": 3.0660377358490567e-05,
+      "loss": 0.5524,
+      "step": 66
+    },
+    {
+      "epoch": 0.1908831908831909,
+      "grad_norm": 0.26395478944072415,
+      "learning_rate": 3.113207547169811e-05,
+      "loss": 0.5063,
+      "step": 67
+    },
+    {
+      "epoch": 0.19373219373219372,
+      "grad_norm": 0.24299930461895702,
+      "learning_rate": 3.160377358490566e-05,
+      "loss": 0.5369,
+      "step": 68
+    },
+    {
+      "epoch": 0.19658119658119658,
+      "grad_norm": 0.24673885752288957,
+      "learning_rate": 3.207547169811321e-05,
+      "loss": 0.51,
+      "step": 69
+    },
+    {
+      "epoch": 0.19943019943019943,
+      "grad_norm": 0.2706880306834431,
+      "learning_rate": 3.254716981132075e-05,
+      "loss": 0.5264,
+      "step": 70
+    },
+    {
+      "epoch": 0.2022792022792023,
+      "grad_norm": 0.26249806371875767,
+      "learning_rate": 3.30188679245283e-05,
+      "loss": 0.5451,
+      "step": 71
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 0.3079867132956617,
+      "learning_rate": 3.349056603773585e-05,
+      "loss": 0.5639,
+      "step": 72
+    },
+    {
+      "epoch": 0.20797720797720798,
+      "grad_norm": 0.2677449528243656,
+      "learning_rate": 3.39622641509434e-05,
+      "loss": 0.5236,
+      "step": 73
+    },
+    {
+      "epoch": 0.21082621082621084,
+      "grad_norm": 0.29313189476756446,
+      "learning_rate": 3.4433962264150943e-05,
+      "loss": 0.5238,
+      "step": 74
+    },
+    {
+      "epoch": 0.21367521367521367,
+      "grad_norm": 0.3077114867531552,
+      "learning_rate": 3.490566037735849e-05,
+      "loss": 0.5184,
+      "step": 75
+    },
+    {
+      "epoch": 0.21652421652421652,
+      "grad_norm": 0.24841288456262042,
+      "learning_rate": 3.537735849056604e-05,
+      "loss": 0.5164,
+      "step": 76
+    },
+    {
+      "epoch": 0.21937321937321938,
+      "grad_norm": 0.28963777073614116,
+      "learning_rate": 3.5849056603773584e-05,
+      "loss": 0.5073,
+      "step": 77
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.27823065639727734,
+      "learning_rate": 3.632075471698113e-05,
+      "loss": 0.5269,
+      "step": 78
+    },
+    {
+      "epoch": 0.22507122507122507,
+      "grad_norm": 0.32870802313426584,
+      "learning_rate": 3.679245283018868e-05,
+      "loss": 0.5043,
+      "step": 79
+    },
+    {
+      "epoch": 0.22792022792022792,
+      "grad_norm": 0.27025584380220774,
+      "learning_rate": 3.7264150943396224e-05,
+      "loss": 0.5024,
+      "step": 80
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.33251614434676685,
+      "learning_rate": 3.7735849056603776e-05,
+      "loss": 0.5305,
+      "step": 81
+    },
+    {
+      "epoch": 0.2336182336182336,
+      "grad_norm": 0.33641696003904814,
+      "learning_rate": 3.820754716981133e-05,
+      "loss": 0.5161,
+      "step": 82
+    },
+    {
+      "epoch": 0.23646723646723647,
+      "grad_norm": 0.27164595148241766,
+      "learning_rate": 3.867924528301887e-05,
+      "loss": 0.4933,
+      "step": 83
+    },
+    {
+      "epoch": 0.23931623931623933,
+      "grad_norm": 0.293519778964147,
+      "learning_rate": 3.9150943396226416e-05,
+      "loss": 0.5106,
+      "step": 84
+    },
+    {
+      "epoch": 0.24216524216524216,
+      "grad_norm": 0.2711786679654629,
+      "learning_rate": 3.962264150943397e-05,
+      "loss": 0.5056,
+      "step": 85
+    },
+    {
+      "epoch": 0.245014245014245,
+      "grad_norm": 0.2909911239118421,
+      "learning_rate": 4.009433962264151e-05,
+      "loss": 0.4906,
+      "step": 86
+    },
+    {
+      "epoch": 0.24786324786324787,
+      "grad_norm": 0.2749435561722525,
+      "learning_rate": 4.0566037735849064e-05,
+      "loss": 0.5239,
+      "step": 87
+    },
+    {
+      "epoch": 0.25071225071225073,
+      "grad_norm": 0.2966697733485334,
+      "learning_rate": 4.103773584905661e-05,
+      "loss": 0.4857,
+      "step": 88
+    },
+    {
+      "epoch": 0.2535612535612536,
+      "grad_norm": 0.2577003407624272,
+      "learning_rate": 4.150943396226415e-05,
+      "loss": 0.5017,
+      "step": 89
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.2509274192908608,
+      "learning_rate": 4.1981132075471704e-05,
+      "loss": 0.5094,
+      "step": 90
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 0.26220476456285075,
+      "learning_rate": 4.245283018867925e-05,
+      "loss": 0.521,
+      "step": 91
+    },
+    {
+      "epoch": 0.2621082621082621,
+      "grad_norm": 0.26233472271330177,
+      "learning_rate": 4.292452830188679e-05,
+      "loss": 0.5039,
+      "step": 92
+    },
+    {
+      "epoch": 0.26495726495726496,
+      "grad_norm": 0.24761878468553983,
+      "learning_rate": 4.3396226415094345e-05,
+      "loss": 0.4948,
+      "step": 93
+    },
+    {
+      "epoch": 0.2678062678062678,
+      "grad_norm": 0.2525575556034209,
+      "learning_rate": 4.386792452830189e-05,
+      "loss": 0.5044,
+      "step": 94
+    },
+    {
+      "epoch": 0.2706552706552707,
+      "grad_norm": 0.27089466017971453,
+      "learning_rate": 4.433962264150944e-05,
+      "loss": 0.5008,
+      "step": 95
+    },
+    {
+      "epoch": 0.27350427350427353,
+      "grad_norm": 0.25465583266476105,
+      "learning_rate": 4.4811320754716985e-05,
+      "loss": 0.4925,
+      "step": 96
+    },
+    {
+      "epoch": 0.27635327635327633,
+      "grad_norm": 0.2459248256786745,
+      "learning_rate": 4.528301886792453e-05,
+      "loss": 0.4716,
+      "step": 97
+    },
+    {
+      "epoch": 0.2792022792022792,
+      "grad_norm": 0.27690227321171584,
+      "learning_rate": 4.575471698113208e-05,
+      "loss": 0.5057,
+      "step": 98
+    },
+    {
+      "epoch": 0.28205128205128205,
+      "grad_norm": 0.31505919622587614,
+      "learning_rate": 4.6226415094339625e-05,
+      "loss": 0.5196,
+      "step": 99
+    },
+    {
+      "epoch": 0.2849002849002849,
+      "grad_norm": 0.2645004241178192,
+      "learning_rate": 4.669811320754717e-05,
+      "loss": 0.4741,
+      "step": 100
+    },
+    {
+      "epoch": 0.28774928774928776,
+      "grad_norm": 0.28420887135800427,
+      "learning_rate": 4.716981132075472e-05,
+      "loss": 0.489,
+      "step": 101
+    },
+    {
+      "epoch": 0.2905982905982906,
+      "grad_norm": 0.2979244733987184,
+      "learning_rate": 4.7641509433962266e-05,
+      "loss": 0.5049,
+      "step": 102
+    },
+    {
+      "epoch": 0.2934472934472934,
+      "grad_norm": 0.31699020934728167,
+      "learning_rate": 4.811320754716982e-05,
+      "loss": 0.4846,
+      "step": 103
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.2958211929457519,
+      "learning_rate": 4.858490566037736e-05,
+      "loss": 0.4904,
+      "step": 104
+    },
+    {
+      "epoch": 0.29914529914529914,
+      "grad_norm": 0.31430371098933885,
+      "learning_rate": 4.9056603773584906e-05,
+      "loss": 0.4864,
+      "step": 105
+    },
+    {
+      "epoch": 0.301994301994302,
+      "grad_norm": 0.31807223918157157,
+      "learning_rate": 4.952830188679246e-05,
+      "loss": 0.496,
+      "step": 106
+    },
+    {
+      "epoch": 0.30484330484330485,
+      "grad_norm": 0.3012030394747298,
+      "learning_rate": 5e-05,
+      "loss": 0.4889,
+      "step": 107
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.26530051400289184,
+      "learning_rate": 4.994720168954594e-05,
+      "loss": 0.4933,
+      "step": 108
+    },
+    {
+      "epoch": 0.31054131054131057,
+      "grad_norm": 0.3230552696656243,
+      "learning_rate": 4.989440337909187e-05,
+      "loss": 0.5068,
+      "step": 109
+    },
+    {
+      "epoch": 0.31339031339031337,
+      "grad_norm": 0.25705267366009066,
+      "learning_rate": 4.9841605068637805e-05,
+      "loss": 0.499,
+      "step": 110
+    },
+    {
+      "epoch": 0.3162393162393162,
+      "grad_norm": 0.3143313651222358,
+      "learning_rate": 4.978880675818374e-05,
+      "loss": 0.4784,
+      "step": 111
+    },
+    {
+      "epoch": 0.3190883190883191,
+      "grad_norm": 0.3376761893109501,
+      "learning_rate": 4.973600844772968e-05,
+      "loss": 0.482,
+      "step": 112
+    },
+    {
+      "epoch": 0.32193732193732194,
+      "grad_norm": 0.28882046798653027,
+      "learning_rate": 4.968321013727561e-05,
+      "loss": 0.4961,
+      "step": 113
+    },
+    {
+      "epoch": 0.3247863247863248,
+      "grad_norm": 0.34419030637706083,
+      "learning_rate": 4.9630411826821544e-05,
+      "loss": 0.4971,
+      "step": 114
+    },
+    {
+      "epoch": 0.32763532763532766,
+      "grad_norm": 0.29426005659261695,
+      "learning_rate": 4.957761351636748e-05,
+      "loss": 0.4903,
+      "step": 115
+    },
+    {
+      "epoch": 0.33048433048433046,
+      "grad_norm": 0.29019064666682737,
+      "learning_rate": 4.952481520591341e-05,
+      "loss": 0.4897,
+      "step": 116
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.33297662577690634,
+      "learning_rate": 4.947201689545935e-05,
+      "loss": 0.4792,
+      "step": 117
+    },
+    {
+      "epoch": 0.33618233618233617,
+      "grad_norm": 0.35409595270488886,
+      "learning_rate": 4.941921858500528e-05,
+      "loss": 0.4925,
+      "step": 118
+    },
+    {
+      "epoch": 0.33903133903133903,
+      "grad_norm": 0.292433952733655,
+      "learning_rate": 4.936642027455122e-05,
+      "loss": 0.5069,
+      "step": 119
+    },
+    {
+      "epoch": 0.3418803418803419,
+      "grad_norm": 0.3320923132157635,
+      "learning_rate": 4.931362196409715e-05,
+      "loss": 0.4946,
+      "step": 120
+    },
+    {
+      "epoch": 0.34472934472934474,
+      "grad_norm": 0.27067630597401476,
+      "learning_rate": 4.9260823653643085e-05,
+      "loss": 0.4871,
+      "step": 121
+    },
+    {
+      "epoch": 0.3475783475783476,
+      "grad_norm": 0.30304447173190324,
+      "learning_rate": 4.920802534318902e-05,
+      "loss": 0.487,
+      "step": 122
+    },
+    {
+      "epoch": 0.3504273504273504,
+      "grad_norm": 0.3039645539285113,
+      "learning_rate": 4.915522703273496e-05,
+      "loss": 0.4867,
+      "step": 123
+    },
+    {
+      "epoch": 0.35327635327635326,
+      "grad_norm": 0.2824982694185181,
+      "learning_rate": 4.9102428722280894e-05,
+      "loss": 0.4611,
+      "step": 124
+    },
+    {
+      "epoch": 0.3561253561253561,
+      "grad_norm": 0.28260131702992686,
+      "learning_rate": 4.9049630411826823e-05,
+      "loss": 0.4652,
+      "step": 125
+    },
+    {
+      "epoch": 0.358974358974359,
+      "grad_norm": 0.34703216534596504,
+      "learning_rate": 4.899683210137276e-05,
+      "loss": 0.4865,
+      "step": 126
+    },
+    {
+      "epoch": 0.36182336182336183,
+      "grad_norm": 0.3162837076477058,
+      "learning_rate": 4.894403379091869e-05,
+      "loss": 0.4954,
+      "step": 127
+    },
+    {
+      "epoch": 0.3646723646723647,
+      "grad_norm": 0.3293203697511355,
+      "learning_rate": 4.8891235480464626e-05,
+      "loss": 0.4995,
+      "step": 128
+    },
+    {
+      "epoch": 0.36752136752136755,
+      "grad_norm": 0.27295379955106025,
+      "learning_rate": 4.883843717001056e-05,
+      "loss": 0.4872,
+      "step": 129
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.3102376623883744,
+      "learning_rate": 4.87856388595565e-05,
+      "loss": 0.481,
+      "step": 130
+    },
+    {
+      "epoch": 0.3732193732193732,
+      "grad_norm": 0.31091911179304077,
+      "learning_rate": 4.8732840549102435e-05,
+      "loss": 0.4891,
+      "step": 131
+    },
+    {
+      "epoch": 0.37606837606837606,
+      "grad_norm": 0.29399727700411593,
+      "learning_rate": 4.8680042238648365e-05,
+      "loss": 0.4678,
+      "step": 132
+    },
+    {
+      "epoch": 0.3789173789173789,
+      "grad_norm": 0.3409889899938261,
+      "learning_rate": 4.86272439281943e-05,
+      "loss": 0.4881,
+      "step": 133
+    },
+    {
+      "epoch": 0.3817663817663818,
+      "grad_norm": 0.3046568772014337,
+      "learning_rate": 4.857444561774023e-05,
+      "loss": 0.4788,
+      "step": 134
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.24665708555682755,
+      "learning_rate": 4.852164730728617e-05,
+      "loss": 0.4572,
+      "step": 135
+    },
+    {
+      "epoch": 0.38746438746438744,
+      "grad_norm": 0.3016729598811574,
+      "learning_rate": 4.8468848996832103e-05,
+      "loss": 0.4804,
+      "step": 136
+    },
+    {
+      "epoch": 0.3903133903133903,
+      "grad_norm": 0.3530427507474638,
+      "learning_rate": 4.841605068637804e-05,
+      "loss": 0.4843,
+      "step": 137
+    },
+    {
+      "epoch": 0.39316239316239315,
+      "grad_norm": 0.2856167787643274,
+      "learning_rate": 4.8363252375923976e-05,
+      "loss": 0.4715,
+      "step": 138
+    },
+    {
+      "epoch": 0.396011396011396,
+      "grad_norm": 0.37847572959684284,
+      "learning_rate": 4.8310454065469906e-05,
+      "loss": 0.4968,
+      "step": 139
+    },
+    {
+      "epoch": 0.39886039886039887,
+      "grad_norm": 0.28801978777804754,
+      "learning_rate": 4.825765575501584e-05,
+      "loss": 0.4632,
+      "step": 140
+    },
+    {
+      "epoch": 0.4017094017094017,
+      "grad_norm": 0.3525074681127096,
+      "learning_rate": 4.820485744456177e-05,
+      "loss": 0.4827,
+      "step": 141
+    },
+    {
+      "epoch": 0.4045584045584046,
+      "grad_norm": 0.3526537528509632,
+      "learning_rate": 4.8152059134107715e-05,
+      "loss": 0.4732,
+      "step": 142
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 0.34857652018932117,
+      "learning_rate": 4.8099260823653645e-05,
+      "loss": 0.4867,
+      "step": 143
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 0.3727176334678898,
+      "learning_rate": 4.804646251319958e-05,
+      "loss": 0.4834,
+      "step": 144
+    },
+    {
+      "epoch": 0.4131054131054131,
+      "grad_norm": 0.3140976089416083,
+      "learning_rate": 4.799366420274552e-05,
+      "loss": 0.4675,
+      "step": 145
+    },
+    {
+      "epoch": 0.41595441595441596,
+      "grad_norm": 0.44474986509411835,
+      "learning_rate": 4.794086589229145e-05,
+      "loss": 0.4914,
+      "step": 146
+    },
+    {
+      "epoch": 0.4188034188034188,
+      "grad_norm": 0.3390407952306162,
+      "learning_rate": 4.788806758183738e-05,
+      "loss": 0.4731,
+      "step": 147
+    },
+    {
+      "epoch": 0.42165242165242167,
+      "grad_norm": 0.48836879750115086,
+      "learning_rate": 4.783526927138332e-05,
+      "loss": 0.4921,
+      "step": 148
+    },
+    {
+      "epoch": 0.42450142450142453,
+      "grad_norm": 0.33589080513403735,
+      "learning_rate": 4.7782470960929256e-05,
+      "loss": 0.4912,
+      "step": 149
+    },
+    {
+      "epoch": 0.42735042735042733,
+      "grad_norm": 0.36236282963696415,
+      "learning_rate": 4.7729672650475186e-05,
+      "loss": 0.4601,
+      "step": 150
+    },
+    {
+      "epoch": 0.4301994301994302,
+      "grad_norm": 0.2797897413436219,
+      "learning_rate": 4.767687434002112e-05,
+      "loss": 0.4672,
+      "step": 151
+    },
+    {
+      "epoch": 0.43304843304843305,
+      "grad_norm": 0.4148496937386019,
+      "learning_rate": 4.762407602956706e-05,
+      "loss": 0.4819,
+      "step": 152
+    },
+    {
+      "epoch": 0.4358974358974359,
+      "grad_norm": 0.3395999932409137,
+      "learning_rate": 4.757127771911299e-05,
+      "loss": 0.4701,
+      "step": 153
+    },
+    {
+      "epoch": 0.43874643874643876,
+      "grad_norm": 0.31394240245627125,
+      "learning_rate": 4.7518479408658925e-05,
+      "loss": 0.4702,
+      "step": 154
+    },
+    {
+      "epoch": 0.4415954415954416,
+      "grad_norm": 0.3976902607497981,
+      "learning_rate": 4.746568109820486e-05,
+      "loss": 0.4754,
+      "step": 155
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.26007701688937257,
+      "learning_rate": 4.74128827877508e-05,
+      "loss": 0.4719,
+      "step": 156
+    },
+    {
+      "epoch": 0.4472934472934473,
+      "grad_norm": 0.3615746631327949,
+      "learning_rate": 4.736008447729673e-05,
+      "loss": 0.483,
+      "step": 157
+    },
+    {
+      "epoch": 0.45014245014245013,
+      "grad_norm": 0.3262559663383385,
+      "learning_rate": 4.730728616684266e-05,
+      "loss": 0.4701,
+      "step": 158
+    },
+    {
+      "epoch": 0.452991452991453,
+      "grad_norm": 0.3054913734906384,
+      "learning_rate": 4.725448785638859e-05,
+      "loss": 0.4755,
+      "step": 159
+    },
+    {
+      "epoch": 0.45584045584045585,
+      "grad_norm": 0.28714899368154406,
+      "learning_rate": 4.720168954593453e-05,
+      "loss": 0.4715,
+      "step": 160
+    },
+    {
+      "epoch": 0.4586894586894587,
+      "grad_norm": 0.3094277794993974,
+      "learning_rate": 4.7148891235480466e-05,
+      "loss": 0.4621,
+      "step": 161
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.3440166736205401,
+      "learning_rate": 4.70960929250264e-05,
+      "loss": 0.485,
+      "step": 162
+    },
+    {
+      "epoch": 0.46438746438746437,
+      "grad_norm": 0.32050130614453004,
+      "learning_rate": 4.704329461457234e-05,
+      "loss": 0.4699,
+      "step": 163
+    },
+    {
+      "epoch": 0.4672364672364672,
+      "grad_norm": 0.3462845175709499,
+      "learning_rate": 4.699049630411827e-05,
+      "loss": 0.467,
+      "step": 164
+    },
+    {
+      "epoch": 0.4700854700854701,
+      "grad_norm": 0.25813062925512487,
+      "learning_rate": 4.6937697993664204e-05,
+      "loss": 0.4678,
+      "step": 165
+    },
+    {
+      "epoch": 0.47293447293447294,
+      "grad_norm": 0.2774578719211831,
+      "learning_rate": 4.6884899683210134e-05,
+      "loss": 0.4517,
+      "step": 166
+    },
+    {
+      "epoch": 0.4757834757834758,
+      "grad_norm": 0.27947314173878857,
+      "learning_rate": 4.683210137275608e-05,
+      "loss": 0.465,
+      "step": 167
+    },
+    {
+      "epoch": 0.47863247863247865,
+      "grad_norm": 0.22733723587768467,
+      "learning_rate": 4.677930306230201e-05,
+      "loss": 0.4537,
+      "step": 168
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 0.29661252845278596,
+      "learning_rate": 4.672650475184794e-05,
+      "loss": 0.4846,
+      "step": 169
+    },
+    {
+      "epoch": 0.4843304843304843,
+      "grad_norm": 0.3089830748079026,
+      "learning_rate": 4.667370644139388e-05,
+      "loss": 0.4557,
+      "step": 170
+    },
+    {
+      "epoch": 0.48717948717948717,
+      "grad_norm": 0.24997311130541303,
+      "learning_rate": 4.662090813093981e-05,
+      "loss": 0.45,
+      "step": 171
+    },
+    {
+      "epoch": 0.49002849002849,
+      "grad_norm": 0.33175156279149914,
+      "learning_rate": 4.6568109820485746e-05,
+      "loss": 0.4689,
+      "step": 172
+    },
+    {
+      "epoch": 0.4928774928774929,
+      "grad_norm": 0.2602987055448183,
+      "learning_rate": 4.651531151003168e-05,
+      "loss": 0.4659,
+      "step": 173
+    },
+    {
+      "epoch": 0.49572649572649574,
+      "grad_norm": 0.3460341458612977,
+      "learning_rate": 4.646251319957762e-05,
+      "loss": 0.4765,
+      "step": 174
+    },
+    {
+      "epoch": 0.4985754985754986,
+      "grad_norm": 0.2961420307282209,
+      "learning_rate": 4.640971488912355e-05,
+      "loss": 0.4924,
+      "step": 175
+    },
+    {
+      "epoch": 0.5014245014245015,
+      "grad_norm": 0.300448359282463,
+      "learning_rate": 4.6356916578669484e-05,
+      "loss": 0.4772,
+      "step": 176
+    },
+    {
+      "epoch": 0.5042735042735043,
+      "grad_norm": 0.27326740002063005,
+      "learning_rate": 4.630411826821542e-05,
+      "loss": 0.4481,
+      "step": 177
+    },
+    {
+      "epoch": 0.5071225071225072,
+      "grad_norm": 0.3260005365533567,
+      "learning_rate": 4.625131995776135e-05,
+      "loss": 0.4757,
+      "step": 178
+    },
+    {
+      "epoch": 0.50997150997151,
+      "grad_norm": 0.3072044287066991,
+      "learning_rate": 4.619852164730729e-05,
+      "loss": 0.4749,
+      "step": 179
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.29686452489417353,
+      "learning_rate": 4.614572333685322e-05,
+      "loss": 0.4599,
+      "step": 180
+    },
+    {
+      "epoch": 0.5156695156695157,
+      "grad_norm": 0.294950820175362,
+      "learning_rate": 4.609292502639916e-05,
+      "loss": 0.4821,
+      "step": 181
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.30491136720622375,
+      "learning_rate": 4.604012671594509e-05,
+      "loss": 0.4537,
+      "step": 182
+    },
+    {
+      "epoch": 0.5213675213675214,
+      "grad_norm": 0.30581018423213313,
+      "learning_rate": 4.5987328405491026e-05,
+      "loss": 0.4641,
+      "step": 183
+    },
+    {
+      "epoch": 0.5242165242165242,
+      "grad_norm": 0.35426832507345146,
+      "learning_rate": 4.593453009503696e-05,
+      "loss": 0.4671,
+      "step": 184
+    },
+    {
+      "epoch": 0.5270655270655271,
+      "grad_norm": 0.33464439271326646,
+      "learning_rate": 4.588173178458289e-05,
+      "loss": 0.4692,
+      "step": 185
+    },
+    {
+      "epoch": 0.5299145299145299,
+      "grad_norm": 0.30593521291345677,
+      "learning_rate": 4.5828933474128835e-05,
+      "loss": 0.4744,
+      "step": 186
+    },
+    {
+      "epoch": 0.5327635327635327,
+      "grad_norm": 0.27674703467670597,
+      "learning_rate": 4.5776135163674764e-05,
+      "loss": 0.4437,
+      "step": 187
+    },
+    {
+      "epoch": 0.5356125356125356,
+      "grad_norm": 0.280964484952,
+      "learning_rate": 4.57233368532207e-05,
+      "loss": 0.467,
+      "step": 188
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.26185851570092444,
+      "learning_rate": 4.567053854276663e-05,
+      "loss": 0.4426,
+      "step": 189
+    },
+    {
+      "epoch": 0.5413105413105413,
+      "grad_norm": 0.2584808882693542,
+      "learning_rate": 4.561774023231257e-05,
+      "loss": 0.4611,
+      "step": 190
+    },
+    {
+      "epoch": 0.5441595441595442,
+      "grad_norm": 0.26904667580690866,
+      "learning_rate": 4.55649419218585e-05,
+      "loss": 0.4575,
+      "step": 191
+    },
+    {
+      "epoch": 0.5470085470085471,
+      "grad_norm": 0.2487167605386707,
+      "learning_rate": 4.551214361140444e-05,
+      "loss": 0.457,
+      "step": 192
+    },
+    {
+      "epoch": 0.5498575498575499,
+      "grad_norm": 0.28142301387763136,
+      "learning_rate": 4.5459345300950376e-05,
+      "loss": 0.4586,
+      "step": 193
+    },
+    {
+      "epoch": 0.5527065527065527,
+      "grad_norm": 0.31342474806414733,
+      "learning_rate": 4.5406546990496306e-05,
+      "loss": 0.4866,
+      "step": 194
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.2847923191286588,
+      "learning_rate": 4.535374868004224e-05,
+      "loss": 0.4643,
+      "step": 195
+    },
+    {
+      "epoch": 0.5584045584045584,
+      "grad_norm": 0.3095635932106011,
+      "learning_rate": 4.530095036958817e-05,
+      "loss": 0.4645,
+      "step": 196
+    },
+    {
+      "epoch": 0.5612535612535613,
+      "grad_norm": 0.24639525285599803,
+      "learning_rate": 4.524815205913411e-05,
+      "loss": 0.4577,
+      "step": 197
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "grad_norm": 0.26955963685848316,
+      "learning_rate": 4.5195353748680044e-05,
+      "loss": 0.4626,
+      "step": 198
+    },
+    {
+      "epoch": 0.5669515669515669,
+      "grad_norm": 0.24214389870117736,
+      "learning_rate": 4.514255543822598e-05,
+      "loss": 0.4535,
+      "step": 199
+    },
+    {
+      "epoch": 0.5698005698005698,
+      "grad_norm": 0.2610341258337823,
+      "learning_rate": 4.508975712777192e-05,
+      "loss": 0.4583,
+      "step": 200
+    },
+    {
+      "epoch": 0.5726495726495726,
+      "grad_norm": 0.25907163228127456,
+      "learning_rate": 4.503695881731785e-05,
+      "loss": 0.4535,
+      "step": 201
+    },
+    {
+      "epoch": 0.5754985754985755,
+      "grad_norm": 0.2761897395293128,
+      "learning_rate": 4.498416050686378e-05,
+      "loss": 0.4543,
+      "step": 202
+    },
+    {
+      "epoch": 0.5783475783475783,
+      "grad_norm": 0.31059563404737606,
+      "learning_rate": 4.493136219640971e-05,
+      "loss": 0.4523,
+      "step": 203
+    },
+    {
+      "epoch": 0.5811965811965812,
+      "grad_norm": 0.286738470928712,
+      "learning_rate": 4.487856388595565e-05,
+      "loss": 0.4836,
+      "step": 204
+    },
+    {
+      "epoch": 0.584045584045584,
+      "grad_norm": 0.2967602486544982,
+      "learning_rate": 4.4825765575501585e-05,
+      "loss": 0.4637,
+      "step": 205
+    },
+    {
+      "epoch": 0.5868945868945868,
+      "grad_norm": 0.2708153047318558,
+      "learning_rate": 4.477296726504752e-05,
+      "loss": 0.4497,
+      "step": 206
+    },
+    {
+      "epoch": 0.5897435897435898,
+      "grad_norm": 0.2746307872171804,
+      "learning_rate": 4.472016895459346e-05,
+      "loss": 0.4512,
+      "step": 207
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.3328817061733048,
+      "learning_rate": 4.466737064413939e-05,
+      "loss": 0.4613,
+      "step": 208
+    },
+    {
+      "epoch": 0.5954415954415955,
+      "grad_norm": 0.2759545614945111,
+      "learning_rate": 4.4614572333685324e-05,
+      "loss": 0.4727,
+      "step": 209
+    },
+    {
+      "epoch": 0.5982905982905983,
+      "grad_norm": 0.318009089876618,
+      "learning_rate": 4.4561774023231254e-05,
+      "loss": 0.4747,
+      "step": 210
+    },
+    {
+      "epoch": 0.6011396011396012,
+      "grad_norm": 0.25640815318642596,
+      "learning_rate": 4.45089757127772e-05,
+      "loss": 0.4719,
+      "step": 211
+    },
+    {
+      "epoch": 0.603988603988604,
+      "grad_norm": 0.34436494976967913,
+      "learning_rate": 4.445617740232313e-05,
+      "loss": 0.4646,
+      "step": 212
+    },
+    {
+      "epoch": 0.6068376068376068,
+      "grad_norm": 1.7399032191303827,
+      "learning_rate": 4.440337909186906e-05,
+      "loss": 0.5284,
+      "step": 213
+    },
+    {
+      "epoch": 0.6096866096866097,
+      "grad_norm": 0.26447171131249575,
+      "learning_rate": 4.4350580781415e-05,
+      "loss": 0.4598,
+      "step": 214
+    },
+    {
+      "epoch": 0.6125356125356125,
+      "grad_norm": 0.2571865578633,
+      "learning_rate": 4.429778247096093e-05,
+      "loss": 0.4491,
+      "step": 215
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.28498607832783857,
+      "learning_rate": 4.4244984160506865e-05,
+      "loss": 0.4572,
+      "step": 216
+    },
+    {
+      "epoch": 0.6182336182336182,
+      "grad_norm": 0.23557920258321058,
+      "learning_rate": 4.41921858500528e-05,
+      "loss": 0.4489,
+      "step": 217
+    },
+    {
+      "epoch": 0.6210826210826211,
+      "grad_norm": 0.3049575913824341,
+      "learning_rate": 4.413938753959874e-05,
+      "loss": 0.4505,
+      "step": 218
+    },
+    {
+      "epoch": 0.6239316239316239,
+      "grad_norm": 0.2582271333688809,
+      "learning_rate": 4.408658922914467e-05,
+      "loss": 0.4608,
+      "step": 219
+    },
+    {
+      "epoch": 0.6267806267806267,
+      "grad_norm": 0.2823458887788208,
+      "learning_rate": 4.4033790918690604e-05,
+      "loss": 0.4533,
+      "step": 220
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 0.2386332818361518,
+      "learning_rate": 4.398099260823654e-05,
+      "loss": 0.4559,
+      "step": 221
+    },
+    {
+      "epoch": 0.6324786324786325,
+      "grad_norm": 0.28795366606672157,
+      "learning_rate": 4.392819429778247e-05,
+      "loss": 0.4566,
+      "step": 222
+    },
+    {
+      "epoch": 0.6353276353276354,
+      "grad_norm": 0.24919391725324347,
+      "learning_rate": 4.3875395987328407e-05,
+      "loss": 0.4716,
+      "step": 223
+    },
+    {
+      "epoch": 0.6381766381766382,
+      "grad_norm": 0.2818278431774406,
+      "learning_rate": 4.382259767687434e-05,
+      "loss": 0.4504,
+      "step": 224
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 0.2814493678841764,
+      "learning_rate": 4.376979936642028e-05,
+      "loss": 0.4434,
+      "step": 225
+    },
+    {
+      "epoch": 0.6438746438746439,
+      "grad_norm": 0.2666133240725091,
+      "learning_rate": 4.371700105596621e-05,
+      "loss": 0.4552,
+      "step": 226
+    },
+    {
+      "epoch": 0.6467236467236467,
+      "grad_norm": 0.2685467121594498,
+      "learning_rate": 4.3664202745512145e-05,
+      "loss": 0.4726,
+      "step": 227
+    },
+    {
+      "epoch": 0.6495726495726496,
+      "grad_norm": 0.299879211553589,
+      "learning_rate": 4.361140443505808e-05,
+      "loss": 0.4471,
+      "step": 228
+    },
+    {
+      "epoch": 0.6524216524216524,
+      "grad_norm": 0.23994886539791888,
+      "learning_rate": 4.355860612460401e-05,
+      "loss": 0.4588,
+      "step": 229
+    },
+    {
+      "epoch": 0.6552706552706553,
+      "grad_norm": 0.32262816469344513,
+      "learning_rate": 4.3505807814149955e-05,
+      "loss": 0.4518,
+      "step": 230
+    },
+    {
+      "epoch": 0.6581196581196581,
+      "grad_norm": 0.26072493194234536,
+      "learning_rate": 4.3453009503695884e-05,
+      "loss": 0.4659,
+      "step": 231
+    },
+    {
+      "epoch": 0.6609686609686609,
+      "grad_norm": 0.31586564589968463,
+      "learning_rate": 4.340021119324182e-05,
+      "loss": 0.456,
+      "step": 232
+    },
+    {
+      "epoch": 0.6638176638176638,
+      "grad_norm": 0.253201215778207,
+      "learning_rate": 4.334741288278775e-05,
+      "loss": 0.4519,
+      "step": 233
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.33113742268580176,
+      "learning_rate": 4.3294614572333687e-05,
+      "loss": 0.4533,
+      "step": 234
+    },
+    {
+      "epoch": 0.6695156695156695,
+      "grad_norm": 0.2760668240736258,
+      "learning_rate": 4.324181626187962e-05,
+      "loss": 0.4663,
+      "step": 235
+    },
+    {
+      "epoch": 0.6723646723646723,
+      "grad_norm": 0.344650634136037,
+      "learning_rate": 4.318901795142556e-05,
+      "loss": 0.4161,
+      "step": 236
+    },
+    {
+      "epoch": 0.6752136752136753,
+      "grad_norm": 0.26467354227622525,
+      "learning_rate": 4.3136219640971496e-05,
+      "loss": 0.4558,
+      "step": 237
+    },
+    {
+      "epoch": 0.6780626780626781,
+      "grad_norm": 0.310387292035022,
+      "learning_rate": 4.3083421330517425e-05,
+      "loss": 0.4507,
+      "step": 238
+    },
+    {
+      "epoch": 0.6809116809116809,
+      "grad_norm": 0.26104826670851183,
+      "learning_rate": 4.303062302006336e-05,
+      "loss": 0.4418,
+      "step": 239
+    },
+    {
+      "epoch": 0.6837606837606838,
+      "grad_norm": 0.33335191034296224,
+      "learning_rate": 4.297782470960929e-05,
+      "loss": 0.4696,
+      "step": 240
+    },
+    {
+      "epoch": 0.6866096866096866,
+      "grad_norm": 0.23349617986863688,
+      "learning_rate": 4.292502639915523e-05,
+      "loss": 0.4354,
+      "step": 241
+    },
+    {
+      "epoch": 0.6894586894586895,
+      "grad_norm": 0.3264954704665932,
+      "learning_rate": 4.2872228088701164e-05,
+      "loss": 0.4495,
+      "step": 242
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.23219348921049893,
+      "learning_rate": 4.28194297782471e-05,
+      "loss": 0.4368,
+      "step": 243
+    },
+    {
+      "epoch": 0.6951566951566952,
+      "grad_norm": 0.30581540788611133,
+      "learning_rate": 4.276663146779304e-05,
+      "loss": 0.4632,
+      "step": 244
+    },
+    {
+      "epoch": 0.698005698005698,
+      "grad_norm": 0.2544190208608847,
+      "learning_rate": 4.2713833157338966e-05,
+      "loss": 0.4289,
+      "step": 245
+    },
+    {
+      "epoch": 0.7008547008547008,
+      "grad_norm": 0.28161765276112943,
+      "learning_rate": 4.26610348468849e-05,
+      "loss": 0.4581,
+      "step": 246
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 0.22266448187906657,
+      "learning_rate": 4.260823653643083e-05,
+      "loss": 0.4442,
+      "step": 247
+    },
+    {
+      "epoch": 0.7065527065527065,
+      "grad_norm": 0.2845469466464988,
+      "learning_rate": 4.255543822597677e-05,
+      "loss": 0.4574,
+      "step": 248
+    },
+    {
+      "epoch": 0.7094017094017094,
+      "grad_norm": 0.27607235956300574,
+      "learning_rate": 4.2502639915522705e-05,
+      "loss": 0.4486,
+      "step": 249
+    },
+    {
+      "epoch": 0.7122507122507122,
+      "grad_norm": 0.2771911103653057,
+      "learning_rate": 4.244984160506864e-05,
+      "loss": 0.4502,
+      "step": 250
+    },
+    {
+      "epoch": 0.7150997150997151,
+      "grad_norm": 0.30336246911969705,
+      "learning_rate": 4.239704329461457e-05,
+      "loss": 0.4521,
+      "step": 251
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 0.2530206241923089,
+      "learning_rate": 4.234424498416051e-05,
+      "loss": 0.4548,
+      "step": 252
+    },
+    {
+      "epoch": 0.7207977207977208,
+      "grad_norm": 0.3101890920570867,
+      "learning_rate": 4.2291446673706444e-05,
+      "loss": 0.4477,
+      "step": 253
+    },
+    {
+      "epoch": 0.7236467236467237,
+      "grad_norm": 0.255737006138244,
+      "learning_rate": 4.2238648363252374e-05,
+      "loss": 0.4494,
+      "step": 254
+    },
+    {
+      "epoch": 0.7264957264957265,
+      "grad_norm": 0.3158292287331288,
+      "learning_rate": 4.218585005279832e-05,
+      "loss": 0.4486,
+      "step": 255
+    },
+    {
+      "epoch": 0.7293447293447294,
+      "grad_norm": 0.2874817334553929,
+      "learning_rate": 4.2133051742344246e-05,
+      "loss": 0.4492,
+      "step": 256
+    },
+    {
+      "epoch": 0.7321937321937322,
+      "grad_norm": 0.2993674086081719,
+      "learning_rate": 4.208025343189018e-05,
+      "loss": 0.446,
+      "step": 257
+    },
+    {
+      "epoch": 0.7350427350427351,
+      "grad_norm": 0.26020269412885194,
+      "learning_rate": 4.202745512143611e-05,
+      "loss": 0.4416,
+      "step": 258
+    },
+    {
+      "epoch": 0.7378917378917379,
+      "grad_norm": 0.29543250701797663,
+      "learning_rate": 4.197465681098205e-05,
+      "loss": 0.4481,
+      "step": 259
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.24853929821482904,
+      "learning_rate": 4.1921858500527985e-05,
+      "loss": 0.46,
+      "step": 260
+    },
+    {
+      "epoch": 0.7435897435897436,
+      "grad_norm": 0.3195779365953625,
+      "learning_rate": 4.186906019007392e-05,
+      "loss": 0.4508,
+      "step": 261
+    },
+    {
+      "epoch": 0.7464387464387464,
+      "grad_norm": 0.2326123884289747,
+      "learning_rate": 4.181626187961986e-05,
+      "loss": 0.446,
+      "step": 262
+    },
+    {
+      "epoch": 0.7492877492877493,
+      "grad_norm": 0.2848262848249219,
+      "learning_rate": 4.176346356916579e-05,
+      "loss": 0.4558,
+      "step": 263
+    },
+    {
+      "epoch": 0.7521367521367521,
+      "grad_norm": 0.2499757501835736,
+      "learning_rate": 4.1710665258711724e-05,
+      "loss": 0.4426,
+      "step": 264
+    },
+    {
+      "epoch": 0.7549857549857549,
+      "grad_norm": 0.2914547474394867,
+      "learning_rate": 4.1657866948257654e-05,
+      "loss": 0.4563,
+      "step": 265
+    },
+    {
+      "epoch": 0.7578347578347578,
+      "grad_norm": 0.2576885537277797,
+      "learning_rate": 4.160506863780359e-05,
+      "loss": 0.4396,
+      "step": 266
+    },
+    {
+      "epoch": 0.7606837606837606,
+      "grad_norm": 0.26204392829857903,
+      "learning_rate": 4.1552270327349526e-05,
+      "loss": 0.4472,
+      "step": 267
+    },
+    {
+      "epoch": 0.7635327635327636,
+      "grad_norm": 0.3265158503996716,
+      "learning_rate": 4.149947201689546e-05,
+      "loss": 0.451,
+      "step": 268
+    },
+    {
+      "epoch": 0.7663817663817664,
+      "grad_norm": 0.2609942596958091,
+      "learning_rate": 4.14466737064414e-05,
+      "loss": 0.4595,
+      "step": 269
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.268707951552568,
+      "learning_rate": 4.139387539598733e-05,
+      "loss": 0.4447,
+      "step": 270
+    },
+    {
+      "epoch": 0.7720797720797721,
+      "grad_norm": 0.28137031045693933,
+      "learning_rate": 4.1341077085533265e-05,
+      "loss": 0.4549,
+      "step": 271
+    },
+    {
+      "epoch": 0.7749287749287749,
+      "grad_norm": 0.29190078199597813,
+      "learning_rate": 4.1288278775079195e-05,
+      "loss": 0.4477,
+      "step": 272
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.2454715815920555,
+      "learning_rate": 4.123548046462513e-05,
+      "loss": 0.4324,
+      "step": 273
+    },
+    {
+      "epoch": 0.7806267806267806,
+      "grad_norm": 0.3094193045635567,
+      "learning_rate": 4.118268215417107e-05,
+      "loss": 0.427,
+      "step": 274
+    },
+    {
+      "epoch": 0.7834757834757835,
+      "grad_norm": 0.24434281879639438,
+      "learning_rate": 4.1129883843717004e-05,
+      "loss": 0.4446,
+      "step": 275
+    },
+    {
+      "epoch": 0.7863247863247863,
+      "grad_norm": 0.29785841511085437,
+      "learning_rate": 4.107708553326294e-05,
+      "loss": 0.4316,
+      "step": 276
+    },
+    {
+      "epoch": 0.7891737891737892,
+      "grad_norm": 0.24685554556974273,
+      "learning_rate": 4.102428722280887e-05,
+      "loss": 0.4376,
+      "step": 277
+    },
+    {
+      "epoch": 0.792022792022792,
+      "grad_norm": 0.24898713039939369,
+      "learning_rate": 4.0971488912354806e-05,
+      "loss": 0.4403,
+      "step": 278
+    },
+    {
+      "epoch": 0.7948717948717948,
+      "grad_norm": 0.30241270134797515,
+      "learning_rate": 4.0918690601900736e-05,
+      "loss": 0.4416,
+      "step": 279
+    },
+    {
+      "epoch": 0.7977207977207977,
+      "grad_norm": 0.24307748159009482,
+      "learning_rate": 4.086589229144668e-05,
+      "loss": 0.4398,
+      "step": 280
+    },
+    {
+      "epoch": 0.8005698005698005,
+      "grad_norm": 0.26394622572849324,
+      "learning_rate": 4.081309398099261e-05,
+      "loss": 0.4421,
+      "step": 281
+    },
+    {
+      "epoch": 0.8034188034188035,
+      "grad_norm": 0.24299948905266622,
+      "learning_rate": 4.0760295670538545e-05,
+      "loss": 0.456,
+      "step": 282
+    },
+    {
+      "epoch": 0.8062678062678063,
+      "grad_norm": 0.2538995235843933,
+      "learning_rate": 4.070749736008448e-05,
+      "loss": 0.4411,
+      "step": 283
+    },
+    {
+      "epoch": 0.8091168091168092,
+      "grad_norm": 0.24352793364283998,
+      "learning_rate": 4.065469904963041e-05,
+      "loss": 0.4457,
+      "step": 284
+    },
+    {
+      "epoch": 0.811965811965812,
+      "grad_norm": 0.754328353026504,
+      "learning_rate": 4.060190073917635e-05,
+      "loss": 0.4749,
+      "step": 285
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.2608557320949828,
+      "learning_rate": 4.0549102428722284e-05,
+      "loss": 0.4366,
+      "step": 286
+    },
+    {
+      "epoch": 0.8176638176638177,
+      "grad_norm": 0.25680855446240963,
+      "learning_rate": 4.049630411826822e-05,
+      "loss": 0.4435,
+      "step": 287
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.23235073760106573,
+      "learning_rate": 4.044350580781415e-05,
+      "loss": 0.4507,
+      "step": 288
+    },
+    {
+      "epoch": 0.8233618233618234,
+      "grad_norm": 0.2503349096849241,
+      "learning_rate": 4.0390707497360086e-05,
+      "loss": 0.4474,
+      "step": 289
+    },
+    {
+      "epoch": 0.8262108262108262,
+      "grad_norm": 0.24087457888823843,
+      "learning_rate": 4.033790918690602e-05,
+      "loss": 0.4312,
+      "step": 290
+    },
+    {
+      "epoch": 0.8290598290598291,
+      "grad_norm": 0.25448968083447765,
+      "learning_rate": 4.028511087645195e-05,
+      "loss": 0.4466,
+      "step": 291
+    },
+    {
+      "epoch": 0.8319088319088319,
+      "grad_norm": 0.24829756385015966,
+      "learning_rate": 4.023231256599789e-05,
+      "loss": 0.4545,
+      "step": 292
+    },
+    {
+      "epoch": 0.8347578347578347,
+      "grad_norm": 0.25607715347273097,
+      "learning_rate": 4.0179514255543825e-05,
+      "loss": 0.4565,
+      "step": 293
+    },
+    {
+      "epoch": 0.8376068376068376,
+      "grad_norm": 0.3091527511324264,
+      "learning_rate": 4.012671594508976e-05,
+      "loss": 0.4561,
+      "step": 294
+    },
+    {
+      "epoch": 0.8404558404558404,
+      "grad_norm": 0.24780329551924565,
+      "learning_rate": 4.007391763463569e-05,
+      "loss": 0.4442,
+      "step": 295
+    },
+    {
+      "epoch": 0.8433048433048433,
+      "grad_norm": 0.2931917392412753,
+      "learning_rate": 4.002111932418163e-05,
+      "loss": 0.4562,
+      "step": 296
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.23945446840779216,
+      "learning_rate": 3.9968321013727564e-05,
+      "loss": 0.4407,
+      "step": 297
+    },
+    {
+      "epoch": 0.8490028490028491,
+      "grad_norm": 0.27070479647529605,
+      "learning_rate": 3.991552270327349e-05,
+      "loss": 0.4371,
+      "step": 298
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 0.2692423182003942,
+      "learning_rate": 3.9862724392819437e-05,
+      "loss": 0.4461,
+      "step": 299
+    },
+    {
+      "epoch": 0.8547008547008547,
+      "grad_norm": 0.2707107693436041,
+      "learning_rate": 3.9809926082365366e-05,
+      "loss": 0.4435,
+      "step": 300
+    },
+    {
+      "epoch": 0.8575498575498576,
+      "grad_norm": 0.2500219756684885,
+      "learning_rate": 3.97571277719113e-05,
+      "loss": 0.4288,
+      "step": 301
+    },
+    {
+      "epoch": 0.8603988603988604,
+      "grad_norm": 0.2783790061607216,
+      "learning_rate": 3.970432946145723e-05,
+      "loss": 0.4486,
+      "step": 302
+    },
+    {
+      "epoch": 0.8632478632478633,
+      "grad_norm": 0.2407111058958332,
+      "learning_rate": 3.965153115100317e-05,
+      "loss": 0.4514,
+      "step": 303
+    },
+    {
+      "epoch": 0.8660968660968661,
+      "grad_norm": 0.30074552592413345,
+      "learning_rate": 3.9598732840549105e-05,
+      "loss": 0.4408,
+      "step": 304
+    },
+    {
+      "epoch": 0.8689458689458689,
+      "grad_norm": 0.21341465024781467,
+      "learning_rate": 3.954593453009504e-05,
+      "loss": 0.4216,
+      "step": 305
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "grad_norm": 0.2938618051639469,
+      "learning_rate": 3.949313621964098e-05,
+      "loss": 0.4416,
+      "step": 306
+    },
+    {
+      "epoch": 0.8746438746438746,
+      "grad_norm": 0.48732469986333266,
+      "learning_rate": 3.944033790918691e-05,
+      "loss": 0.4516,
+      "step": 307
+    },
+    {
+      "epoch": 0.8774928774928775,
+      "grad_norm": 0.2735554566382607,
+      "learning_rate": 3.9387539598732844e-05,
+      "loss": 0.4383,
+      "step": 308
+    },
+    {
+      "epoch": 0.8803418803418803,
+      "grad_norm": 0.28479664265954874,
+      "learning_rate": 3.933474128827877e-05,
+      "loss": 0.4458,
+      "step": 309
+    },
+    {
+      "epoch": 0.8831908831908832,
+      "grad_norm": 0.22798637141219172,
+      "learning_rate": 3.928194297782471e-05,
+      "loss": 0.442,
+      "step": 310
+    },
+    {
+      "epoch": 0.886039886039886,
+      "grad_norm": 0.24677969245732811,
+      "learning_rate": 3.9229144667370646e-05,
+      "loss": 0.4481,
+      "step": 311
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.29202684528549283,
+      "learning_rate": 3.917634635691658e-05,
+      "loss": 0.4412,
+      "step": 312
+    },
+    {
+      "epoch": 0.8917378917378918,
+      "grad_norm": 0.2606008473665173,
+      "learning_rate": 3.912354804646252e-05,
+      "loss": 0.435,
+      "step": 313
+    },
+    {
+      "epoch": 0.8945868945868946,
+      "grad_norm": 0.26749609172100086,
+      "learning_rate": 3.907074973600845e-05,
+      "loss": 0.4516,
+      "step": 314
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 0.22520943451800887,
+      "learning_rate": 3.9017951425554385e-05,
+      "loss": 0.4241,
+      "step": 315
+    },
+    {
+      "epoch": 0.9002849002849003,
+      "grad_norm": 0.22909468804333016,
+      "learning_rate": 3.8965153115100314e-05,
+      "loss": 0.4379,
+      "step": 316
+    },
+    {
+      "epoch": 0.9031339031339032,
+      "grad_norm": 0.2836733485621368,
+      "learning_rate": 3.891235480464625e-05,
+      "loss": 0.4476,
+      "step": 317
+    },
+    {
+      "epoch": 0.905982905982906,
+      "grad_norm": 0.244148163521808,
+      "learning_rate": 3.885955649419219e-05,
+      "loss": 0.4546,
+      "step": 318
+    },
+    {
+      "epoch": 0.9088319088319088,
+      "grad_norm": 0.24672915009941915,
+      "learning_rate": 3.8806758183738124e-05,
+      "loss": 0.4368,
+      "step": 319
+    },
+    {
+      "epoch": 0.9116809116809117,
+      "grad_norm": 0.4272615119652992,
+      "learning_rate": 3.875395987328406e-05,
+      "loss": 0.4334,
+      "step": 320
+    },
+    {
+      "epoch": 0.9145299145299145,
+      "grad_norm": 0.24895020848554578,
+      "learning_rate": 3.870116156282999e-05,
+      "loss": 0.4442,
+      "step": 321
+    },
+    {
+      "epoch": 0.9173789173789174,
+      "grad_norm": 0.2658053036431955,
+      "learning_rate": 3.8648363252375926e-05,
+      "loss": 0.4383,
+      "step": 322
+    },
+    {
+      "epoch": 0.9202279202279202,
+      "grad_norm": 0.2643219649864309,
+      "learning_rate": 3.8595564941921856e-05,
+      "loss": 0.4251,
+      "step": 323
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.2365954433037453,
+      "learning_rate": 3.85427666314678e-05,
+      "loss": 0.4499,
+      "step": 324
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.2633023483060389,
+      "learning_rate": 3.848996832101373e-05,
+      "loss": 0.4384,
+      "step": 325
+    },
+    {
+      "epoch": 0.9287749287749287,
+      "grad_norm": 0.2956130036845057,
+      "learning_rate": 3.8437170010559665e-05,
+      "loss": 0.4396,
+      "step": 326
+    },
+    {
+      "epoch": 0.9316239316239316,
+      "grad_norm": 0.24882987462496686,
+      "learning_rate": 3.83843717001056e-05,
+      "loss": 0.45,
+      "step": 327
+    },
+    {
+      "epoch": 0.9344729344729344,
+      "grad_norm": 1.5891449688358956,
+      "learning_rate": 3.833157338965153e-05,
+      "loss": 0.4415,
+      "step": 328
+    },
+    {
+      "epoch": 0.9373219373219374,
+      "grad_norm": 0.3349715325263592,
+      "learning_rate": 3.827877507919747e-05,
+      "loss": 0.4457,
+      "step": 329
+    },
+    {
+      "epoch": 0.9401709401709402,
+      "grad_norm": 0.2848178852794376,
+      "learning_rate": 3.8225976768743404e-05,
+      "loss": 0.456,
+      "step": 330
+    },
+    {
+      "epoch": 0.9430199430199431,
+      "grad_norm": 0.302751156256454,
+      "learning_rate": 3.817317845828934e-05,
+      "loss": 0.4387,
+      "step": 331
+    },
+    {
+      "epoch": 0.9458689458689459,
+      "grad_norm": 0.25955277850770286,
+      "learning_rate": 3.812038014783527e-05,
+      "loss": 0.44,
+      "step": 332
+    },
+    {
+      "epoch": 0.9487179487179487,
+      "grad_norm": 0.2502011385796011,
+      "learning_rate": 3.8067581837381206e-05,
+      "loss": 0.4515,
+      "step": 333
+    },
+    {
+      "epoch": 0.9515669515669516,
+      "grad_norm": 0.2650885248812146,
+      "learning_rate": 3.801478352692714e-05,
+      "loss": 0.4475,
+      "step": 334
+    },
+    {
+      "epoch": 0.9544159544159544,
+      "grad_norm": 0.24800530026732448,
+      "learning_rate": 3.796198521647307e-05,
+      "loss": 0.4589,
+      "step": 335
+    },
+    {
+      "epoch": 0.9572649572649573,
+      "grad_norm": 0.24826963064202007,
+      "learning_rate": 3.7909186906019015e-05,
+      "loss": 0.4338,
+      "step": 336
+    },
+    {
+      "epoch": 0.9601139601139601,
+      "grad_norm": 0.23027863612091976,
+      "learning_rate": 3.7856388595564945e-05,
+      "loss": 0.4301,
+      "step": 337
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.22579121856289947,
+      "learning_rate": 3.780359028511088e-05,
+      "loss": 0.4592,
+      "step": 338
+    },
+    {
+      "epoch": 0.9658119658119658,
+      "grad_norm": 0.2548134083345822,
+      "learning_rate": 3.775079197465681e-05,
+      "loss": 0.442,
+      "step": 339
+    },
+    {
+      "epoch": 0.9686609686609686,
+      "grad_norm": 0.24004323105561068,
+      "learning_rate": 3.769799366420275e-05,
+      "loss": 0.4379,
+      "step": 340
+    },
+    {
+      "epoch": 0.9715099715099715,
+      "grad_norm": 0.22500518319359325,
+      "learning_rate": 3.764519535374868e-05,
+      "loss": 0.4383,
+      "step": 341
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "grad_norm": 0.23655137638652632,
+      "learning_rate": 3.759239704329461e-05,
+      "loss": 0.4304,
+      "step": 342
+    },
+    {
+      "epoch": 0.9772079772079773,
+      "grad_norm": 0.23186063099612125,
+      "learning_rate": 3.7539598732840556e-05,
+      "loss": 0.457,
+      "step": 343
+    },
+    {
+      "epoch": 0.98005698005698,
+      "grad_norm": 0.22345800936056795,
+      "learning_rate": 3.7486800422386486e-05,
+      "loss": 0.4233,
+      "step": 344
+    },
+    {
+      "epoch": 0.9829059829059829,
+      "grad_norm": 0.22901659678419103,
+      "learning_rate": 3.743400211193242e-05,
+      "loss": 0.4575,
+      "step": 345
+    },
+    {
+      "epoch": 0.9857549857549858,
+      "grad_norm": 0.34541054663294773,
+      "learning_rate": 3.738120380147835e-05,
+      "loss": 0.4539,
+      "step": 346
+    },
+    {
+      "epoch": 0.9886039886039886,
+      "grad_norm": 0.24563352556354082,
+      "learning_rate": 3.732840549102429e-05,
+      "loss": 0.4326,
+      "step": 347
+    },
+    {
+      "epoch": 0.9914529914529915,
+      "grad_norm": 0.28378321755449876,
+      "learning_rate": 3.727560718057022e-05,
+      "loss": 0.4527,
+      "step": 348
+    },
+    {
+      "epoch": 0.9943019943019943,
+      "grad_norm": 0.2376023328816585,
+      "learning_rate": 3.722280887011616e-05,
+      "loss": 0.465,
+      "step": 349
+    },
+    {
+      "epoch": 0.9971509971509972,
+      "grad_norm": 0.23913998831345243,
+      "learning_rate": 3.717001055966209e-05,
+      "loss": 0.4502,
+      "step": 350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2446661739630635,
+      "learning_rate": 3.711721224920803e-05,
+      "loss": 0.436,
+      "step": 351
+    },
+    {
+      "epoch": 1.002849002849003,
+      "grad_norm": 0.30212623378795767,
+      "learning_rate": 3.7064413938753963e-05,
+      "loss": 0.3773,
+      "step": 352
+    },
+    {
+      "epoch": 1.0056980056980056,
+      "grad_norm": 0.2727102306693993,
+      "learning_rate": 3.701161562829989e-05,
+      "loss": 0.3721,
+      "step": 353
+    },
+    {
+      "epoch": 1.0085470085470085,
+      "grad_norm": 0.24519118257200442,
+      "learning_rate": 3.695881731784583e-05,
+      "loss": 0.3556,
+      "step": 354
+    },
+    {
+      "epoch": 1.0113960113960114,
+      "grad_norm": 0.2708532250345628,
+      "learning_rate": 3.6906019007391766e-05,
+      "loss": 0.3631,
+      "step": 355
+    },
+    {
+      "epoch": 1.0142450142450143,
+      "grad_norm": 0.2848868957990222,
+      "learning_rate": 3.68532206969377e-05,
+      "loss": 0.3647,
+      "step": 356
+    },
+    {
+      "epoch": 1.017094017094017,
+      "grad_norm": 0.2609789117933919,
+      "learning_rate": 3.680042238648363e-05,
+      "loss": 0.3598,
+      "step": 357
+    },
+    {
+      "epoch": 1.01994301994302,
+      "grad_norm": 0.30731469194417627,
+      "learning_rate": 3.674762407602957e-05,
+      "loss": 0.3905,
+      "step": 358
+    },
+    {
+      "epoch": 1.0227920227920229,
+      "grad_norm": 0.2751477526094167,
+      "learning_rate": 3.6694825765575505e-05,
+      "loss": 0.3766,
+      "step": 359
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 0.3041300782163382,
+      "learning_rate": 3.6642027455121434e-05,
+      "loss": 0.3598,
+      "step": 360
+    },
+    {
+      "epoch": 1.0284900284900285,
+      "grad_norm": 0.29025593237743824,
+      "learning_rate": 3.658922914466738e-05,
+      "loss": 0.3656,
+      "step": 361
+    },
+    {
+      "epoch": 1.0313390313390314,
+      "grad_norm": 0.26450155835177075,
+      "learning_rate": 3.653643083421331e-05,
+      "loss": 0.3765,
+      "step": 362
+    },
+    {
+      "epoch": 1.0341880341880343,
+      "grad_norm": 0.29763564022742284,
+      "learning_rate": 3.648363252375924e-05,
+      "loss": 0.3677,
+      "step": 363
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.2531649793647418,
+      "learning_rate": 3.643083421330517e-05,
+      "loss": 0.3687,
+      "step": 364
+    },
+    {
+      "epoch": 1.03988603988604,
+      "grad_norm": 0.278947460622146,
+      "learning_rate": 3.637803590285111e-05,
+      "loss": 0.3988,
+      "step": 365
+    },
+    {
+      "epoch": 1.0427350427350428,
+      "grad_norm": 0.28829274848081526,
+      "learning_rate": 3.6325237592397046e-05,
+      "loss": 0.3724,
+      "step": 366
+    },
+    {
+      "epoch": 1.0455840455840455,
+      "grad_norm": 0.2618749521918147,
+      "learning_rate": 3.6272439281942975e-05,
+      "loss": 0.3542,
+      "step": 367
+    },
+    {
+      "epoch": 1.0484330484330484,
+      "grad_norm": 2.0062858158052452,
+      "learning_rate": 3.621964097148892e-05,
+      "loss": 0.3987,
+      "step": 368
+    },
+    {
+      "epoch": 1.0512820512820513,
+      "grad_norm": 0.28445975801774975,
+      "learning_rate": 3.616684266103485e-05,
+      "loss": 0.3653,
+      "step": 369
+    },
+    {
+      "epoch": 1.0541310541310542,
+      "grad_norm": 0.2462050134870002,
+      "learning_rate": 3.6114044350580785e-05,
+      "loss": 0.3609,
+      "step": 370
+    },
+    {
+      "epoch": 1.056980056980057,
+      "grad_norm": 0.28062644181944496,
+      "learning_rate": 3.6061246040126714e-05,
+      "loss": 0.3865,
+      "step": 371
+    },
+    {
+      "epoch": 1.0598290598290598,
+      "grad_norm": 0.24606665304462752,
+      "learning_rate": 3.600844772967265e-05,
+      "loss": 0.3502,
+      "step": 372
+    },
+    {
+      "epoch": 1.0626780626780628,
+      "grad_norm": 0.3129397581417704,
+      "learning_rate": 3.595564941921859e-05,
+      "loss": 0.3373,
+      "step": 373
+    },
+    {
+      "epoch": 1.0655270655270654,
+      "grad_norm": 0.2957891434648334,
+      "learning_rate": 3.590285110876452e-05,
+      "loss": 0.3591,
+      "step": 374
+    },
+    {
+      "epoch": 1.0683760683760684,
+      "grad_norm": 0.2570065060278186,
+      "learning_rate": 3.585005279831046e-05,
+      "loss": 0.3861,
+      "step": 375
+    },
+    {
+      "epoch": 1.0712250712250713,
+      "grad_norm": 0.25605383657422265,
+      "learning_rate": 3.579725448785639e-05,
+      "loss": 0.3834,
+      "step": 376
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 0.26319733021304453,
+      "learning_rate": 3.5744456177402326e-05,
+      "loss": 0.3973,
+      "step": 377
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.26823999119784286,
+      "learning_rate": 3.5691657866948255e-05,
+      "loss": 0.3843,
+      "step": 378
+    },
+    {
+      "epoch": 1.0797720797720798,
+      "grad_norm": 0.2801739766417507,
+      "learning_rate": 3.563885955649419e-05,
+      "loss": 0.3613,
+      "step": 379
+    },
+    {
+      "epoch": 1.0826210826210827,
+      "grad_norm": 0.2752489664634236,
+      "learning_rate": 3.558606124604013e-05,
+      "loss": 0.3608,
+      "step": 380
+    },
+    {
+      "epoch": 1.0854700854700854,
+      "grad_norm": 0.253420517582108,
+      "learning_rate": 3.5533262935586064e-05,
+      "loss": 0.3654,
+      "step": 381
+    },
+    {
+      "epoch": 1.0883190883190883,
+      "grad_norm": 0.2352861368854798,
+      "learning_rate": 3.5480464625132e-05,
+      "loss": 0.3567,
+      "step": 382
+    },
+    {
+      "epoch": 1.0911680911680912,
+      "grad_norm": 0.24009458895765762,
+      "learning_rate": 3.542766631467793e-05,
+      "loss": 0.3622,
+      "step": 383
+    },
+    {
+      "epoch": 1.0940170940170941,
+      "grad_norm": 0.26857315547032595,
+      "learning_rate": 3.537486800422387e-05,
+      "loss": 0.3654,
+      "step": 384
+    },
+    {
+      "epoch": 1.0968660968660968,
+      "grad_norm": 0.3755228777720786,
+      "learning_rate": 3.5322069693769796e-05,
+      "loss": 0.3785,
+      "step": 385
+    },
+    {
+      "epoch": 1.0997150997150997,
+      "grad_norm": 0.25523099197348387,
+      "learning_rate": 3.526927138331573e-05,
+      "loss": 0.3703,
+      "step": 386
+    },
+    {
+      "epoch": 1.1025641025641026,
+      "grad_norm": 0.25215901520376743,
+      "learning_rate": 3.521647307286167e-05,
+      "loss": 0.3717,
+      "step": 387
+    },
+    {
+      "epoch": 1.1054131054131053,
+      "grad_norm": 0.21795121913627405,
+      "learning_rate": 3.5163674762407606e-05,
+      "loss": 0.3613,
+      "step": 388
+    },
+    {
+      "epoch": 1.1082621082621082,
+      "grad_norm": 0.21930209265394288,
+      "learning_rate": 3.511087645195354e-05,
+      "loss": 0.3656,
+      "step": 389
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.2443453451193516,
+      "learning_rate": 3.505807814149947e-05,
+      "loss": 0.3569,
+      "step": 390
+    },
+    {
+      "epoch": 1.1139601139601139,
+      "grad_norm": 0.23464081224607158,
+      "learning_rate": 3.500527983104541e-05,
+      "loss": 0.3733,
+      "step": 391
+    },
+    {
+      "epoch": 1.1168091168091168,
+      "grad_norm": 0.23312948395717237,
+      "learning_rate": 3.495248152059134e-05,
+      "loss": 0.3695,
+      "step": 392
+    },
+    {
+      "epoch": 1.1196581196581197,
+      "grad_norm": 0.20794997263096304,
+      "learning_rate": 3.489968321013728e-05,
+      "loss": 0.3573,
+      "step": 393
+    },
+    {
+      "epoch": 1.1225071225071226,
+      "grad_norm": 0.26627834231921077,
+      "learning_rate": 3.484688489968321e-05,
+      "loss": 0.3801,
+      "step": 394
+    },
+    {
+      "epoch": 1.1253561253561253,
+      "grad_norm": 0.210991273677547,
+      "learning_rate": 3.479408658922915e-05,
+      "loss": 0.3447,
+      "step": 395
+    },
+    {
+      "epoch": 1.1282051282051282,
+      "grad_norm": 0.23724562353300943,
+      "learning_rate": 3.474128827877508e-05,
+      "loss": 0.356,
+      "step": 396
+    },
+    {
+      "epoch": 1.131054131054131,
+      "grad_norm": 0.2629540478309352,
+      "learning_rate": 3.468848996832101e-05,
+      "loss": 0.3808,
+      "step": 397
+    },
+    {
+      "epoch": 1.133903133903134,
+      "grad_norm": 0.21190850006514703,
+      "learning_rate": 3.463569165786695e-05,
+      "loss": 0.3606,
+      "step": 398
+    },
+    {
+      "epoch": 1.1367521367521367,
+      "grad_norm": 0.22884348327135656,
+      "learning_rate": 3.4582893347412886e-05,
+      "loss": 0.3612,
+      "step": 399
+    },
+    {
+      "epoch": 1.1396011396011396,
+      "grad_norm": 0.2387204757429875,
+      "learning_rate": 3.453009503695882e-05,
+      "loss": 0.3663,
+      "step": 400
+    },
+    {
+      "epoch": 1.1424501424501425,
+      "grad_norm": 0.2531987312790015,
+      "learning_rate": 3.447729672650475e-05,
+      "loss": 0.3708,
+      "step": 401
+    },
+    {
+      "epoch": 1.1452991452991452,
+      "grad_norm": 0.23529785297450195,
+      "learning_rate": 3.442449841605069e-05,
+      "loss": 0.3908,
+      "step": 402
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.21187095337907066,
+      "learning_rate": 3.4371700105596624e-05,
+      "loss": 0.3746,
+      "step": 403
+    },
+    {
+      "epoch": 1.150997150997151,
+      "grad_norm": 0.20927824341881443,
+      "learning_rate": 3.4318901795142554e-05,
+      "loss": 0.3744,
+      "step": 404
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.23470612919324702,
+      "learning_rate": 3.42661034846885e-05,
+      "loss": 0.3546,
+      "step": 405
+    },
+    {
+      "epoch": 1.1566951566951567,
+      "grad_norm": 0.20792179111583126,
+      "learning_rate": 3.421330517423443e-05,
+      "loss": 0.3711,
+      "step": 406
+    },
+    {
+      "epoch": 1.1595441595441596,
+      "grad_norm": 0.2426450060857107,
+      "learning_rate": 3.416050686378036e-05,
+      "loss": 0.3446,
+      "step": 407
+    },
+    {
+      "epoch": 1.1623931623931625,
+      "grad_norm": 0.2583692183849906,
+      "learning_rate": 3.410770855332629e-05,
+      "loss": 0.3725,
+      "step": 408
+    },
+    {
+      "epoch": 1.1652421652421652,
+      "grad_norm": 0.24566054945662688,
+      "learning_rate": 3.405491024287223e-05,
+      "loss": 0.3802,
+      "step": 409
+    },
+    {
+      "epoch": 1.168091168091168,
+      "grad_norm": 0.26319005374456306,
+      "learning_rate": 3.4002111932418166e-05,
+      "loss": 0.3579,
+      "step": 410
+    },
+    {
+      "epoch": 1.170940170940171,
+      "grad_norm": 0.2371042478251275,
+      "learning_rate": 3.3949313621964095e-05,
+      "loss": 0.3713,
+      "step": 411
+    },
+    {
+      "epoch": 1.173789173789174,
+      "grad_norm": 0.21150424231515216,
+      "learning_rate": 3.389651531151004e-05,
+      "loss": 0.3617,
+      "step": 412
+    },
+    {
+      "epoch": 1.1766381766381766,
+      "grad_norm": 0.2660573468294059,
+      "learning_rate": 3.384371700105597e-05,
+      "loss": 0.3619,
+      "step": 413
+    },
+    {
+      "epoch": 1.1794871794871795,
+      "grad_norm": 0.2629889818118525,
+      "learning_rate": 3.3790918690601904e-05,
+      "loss": 0.3729,
+      "step": 414
+    },
+    {
+      "epoch": 1.1823361823361824,
+      "grad_norm": 0.21963192309042254,
+      "learning_rate": 3.3738120380147834e-05,
+      "loss": 0.3649,
+      "step": 415
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.2676340624788005,
+      "learning_rate": 3.368532206969377e-05,
+      "loss": 0.3557,
+      "step": 416
+    },
+    {
+      "epoch": 1.188034188034188,
+      "grad_norm": 0.21995558755670147,
+      "learning_rate": 3.363252375923971e-05,
+      "loss": 0.3645,
+      "step": 417
+    },
+    {
+      "epoch": 1.190883190883191,
+      "grad_norm": 0.21570771890834206,
+      "learning_rate": 3.357972544878564e-05,
+      "loss": 0.3576,
+      "step": 418
+    },
+    {
+      "epoch": 1.1937321937321936,
+      "grad_norm": 0.26463658898885334,
+      "learning_rate": 3.352692713833158e-05,
+      "loss": 0.3567,
+      "step": 419
+    },
+    {
+      "epoch": 1.1965811965811965,
+      "grad_norm": 0.21514849821934176,
+      "learning_rate": 3.347412882787751e-05,
+      "loss": 0.3745,
+      "step": 420
+    },
+    {
+      "epoch": 1.1994301994301995,
+      "grad_norm": 0.24477901943499678,
+      "learning_rate": 3.3421330517423445e-05,
+      "loss": 0.3733,
+      "step": 421
+    },
+    {
+      "epoch": 1.2022792022792024,
+      "grad_norm": 0.21967390180786867,
+      "learning_rate": 3.3368532206969375e-05,
+      "loss": 0.3848,
+      "step": 422
+    },
+    {
+      "epoch": 1.205128205128205,
+      "grad_norm": 0.22210804416610816,
+      "learning_rate": 3.331573389651531e-05,
+      "loss": 0.3502,
+      "step": 423
+    },
+    {
+      "epoch": 1.207977207977208,
+      "grad_norm": 0.23341712719478797,
+      "learning_rate": 3.326293558606125e-05,
+      "loss": 0.3788,
+      "step": 424
+    },
+    {
+      "epoch": 1.210826210826211,
+      "grad_norm": 0.22474183284157034,
+      "learning_rate": 3.3210137275607184e-05,
+      "loss": 0.3739,
+      "step": 425
+    },
+    {
+      "epoch": 1.2136752136752136,
+      "grad_norm": 0.20810663859008569,
+      "learning_rate": 3.315733896515312e-05,
+      "loss": 0.3575,
+      "step": 426
+    },
+    {
+      "epoch": 1.2165242165242165,
+      "grad_norm": 0.23332852349670008,
+      "learning_rate": 3.310454065469905e-05,
+      "loss": 0.3431,
+      "step": 427
+    },
+    {
+      "epoch": 1.2193732193732194,
+      "grad_norm": 0.2158486368866276,
+      "learning_rate": 3.305174234424499e-05,
+      "loss": 0.3579,
+      "step": 428
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.23876477976217,
+      "learning_rate": 3.2998944033790916e-05,
+      "loss": 0.3748,
+      "step": 429
+    },
+    {
+      "epoch": 1.225071225071225,
+      "grad_norm": 0.22671754097370178,
+      "learning_rate": 3.294614572333686e-05,
+      "loss": 0.3733,
+      "step": 430
+    },
+    {
+      "epoch": 1.227920227920228,
+      "grad_norm": 0.20882908615979884,
+      "learning_rate": 3.289334741288279e-05,
+      "loss": 0.3478,
+      "step": 431
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.22921490176990594,
+      "learning_rate": 3.2840549102428725e-05,
+      "loss": 0.3947,
+      "step": 432
+    },
+    {
+      "epoch": 1.2336182336182335,
+      "grad_norm": 0.24877033092993325,
+      "learning_rate": 3.278775079197466e-05,
+      "loss": 0.3587,
+      "step": 433
+    },
+    {
+      "epoch": 1.2364672364672364,
+      "grad_norm": 0.21983363052940397,
+      "learning_rate": 3.273495248152059e-05,
+      "loss": 0.3621,
+      "step": 434
+    },
+    {
+      "epoch": 1.2393162393162394,
+      "grad_norm": 0.23244050435128452,
+      "learning_rate": 3.268215417106653e-05,
+      "loss": 0.3414,
+      "step": 435
+    },
+    {
+      "epoch": 1.242165242165242,
+      "grad_norm": 0.3093091073113401,
+      "learning_rate": 3.262935586061246e-05,
+      "loss": 0.3708,
+      "step": 436
+    },
+    {
+      "epoch": 1.245014245014245,
+      "grad_norm": 0.21362305859740863,
+      "learning_rate": 3.25765575501584e-05,
+      "loss": 0.3516,
+      "step": 437
+    },
+    {
+      "epoch": 1.2478632478632479,
+      "grad_norm": 0.2454599928955601,
+      "learning_rate": 3.252375923970433e-05,
+      "loss": 0.3692,
+      "step": 438
+    },
+    {
+      "epoch": 1.2507122507122508,
+      "grad_norm": 0.25257783308579096,
+      "learning_rate": 3.2470960929250267e-05,
+      "loss": 0.3522,
+      "step": 439
+    },
+    {
+      "epoch": 1.2535612535612537,
+      "grad_norm": 0.24797928773992586,
+      "learning_rate": 3.2418162618796196e-05,
+      "loss": 0.3681,
+      "step": 440
+    },
+    {
+      "epoch": 1.2564102564102564,
+      "grad_norm": 0.2276797048697118,
+      "learning_rate": 3.236536430834213e-05,
+      "loss": 0.3419,
+      "step": 441
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 0.2646126048006256,
+      "learning_rate": 3.231256599788807e-05,
+      "loss": 0.3834,
+      "step": 442
+    },
+    {
+      "epoch": 1.2621082621082622,
+      "grad_norm": 0.23971069233365036,
+      "learning_rate": 3.2259767687434005e-05,
+      "loss": 0.3496,
+      "step": 443
+    },
+    {
+      "epoch": 1.264957264957265,
+      "grad_norm": 0.21565481165037637,
+      "learning_rate": 3.220696937697994e-05,
+      "loss": 0.3621,
+      "step": 444
+    },
+    {
+      "epoch": 1.2678062678062678,
+      "grad_norm": 0.2429214837642576,
+      "learning_rate": 3.215417106652587e-05,
+      "loss": 0.3653,
+      "step": 445
+    },
+    {
+      "epoch": 1.2706552706552707,
+      "grad_norm": 0.23114362643819203,
+      "learning_rate": 3.210137275607181e-05,
+      "loss": 0.3563,
+      "step": 446
+    },
+    {
+      "epoch": 1.2735042735042734,
+      "grad_norm": 0.22101693642272935,
+      "learning_rate": 3.204857444561774e-05,
+      "loss": 0.3485,
+      "step": 447
+    },
+    {
+      "epoch": 1.2763532763532763,
+      "grad_norm": 0.2483813722230642,
+      "learning_rate": 3.1995776135163674e-05,
+      "loss": 0.3679,
+      "step": 448
+    },
+    {
+      "epoch": 1.2792022792022792,
+      "grad_norm": 0.21814879797928116,
+      "learning_rate": 3.194297782470961e-05,
+      "loss": 0.3699,
+      "step": 449
+    },
+    {
+      "epoch": 1.282051282051282,
+      "grad_norm": 0.22150552594897266,
+      "learning_rate": 3.1890179514255547e-05,
+      "loss": 0.3595,
+      "step": 450
+    },
+    {
+      "epoch": 1.2849002849002849,
+      "grad_norm": 0.22366186353090892,
+      "learning_rate": 3.183738120380148e-05,
+      "loss": 0.3599,
+      "step": 451
+    },
+    {
+      "epoch": 1.2877492877492878,
+      "grad_norm": 0.23116335927041887,
+      "learning_rate": 3.178458289334741e-05,
+      "loss": 0.3676,
+      "step": 452
+    },
+    {
+      "epoch": 1.2905982905982907,
+      "grad_norm": 0.2033308734127439,
+      "learning_rate": 3.173178458289335e-05,
+      "loss": 0.3521,
+      "step": 453
+    },
+    {
+      "epoch": 1.2934472934472934,
+      "grad_norm": 0.2165082929329912,
+      "learning_rate": 3.167898627243928e-05,
+      "loss": 0.3542,
+      "step": 454
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.19988858957987832,
+      "learning_rate": 3.1626187961985215e-05,
+      "loss": 0.3498,
+      "step": 455
+    },
+    {
+      "epoch": 1.2991452991452992,
+      "grad_norm": 0.2292675192834246,
+      "learning_rate": 3.157338965153115e-05,
+      "loss": 0.3739,
+      "step": 456
+    },
+    {
+      "epoch": 1.301994301994302,
+      "grad_norm": 0.2110292978902313,
+      "learning_rate": 3.152059134107709e-05,
+      "loss": 0.3585,
+      "step": 457
+    },
+    {
+      "epoch": 1.3048433048433048,
+      "grad_norm": 0.22404470257022518,
+      "learning_rate": 3.1467793030623024e-05,
+      "loss": 0.3448,
+      "step": 458
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.244584760285099,
+      "learning_rate": 3.1414994720168954e-05,
+      "loss": 0.371,
+      "step": 459
+    },
+    {
+      "epoch": 1.3105413105413106,
+      "grad_norm": 0.2014664128466252,
+      "learning_rate": 3.136219640971489e-05,
+      "loss": 0.3568,
+      "step": 460
+    },
+    {
+      "epoch": 1.3133903133903133,
+      "grad_norm": 0.23086767835485536,
+      "learning_rate": 3.130939809926082e-05,
+      "loss": 0.359,
+      "step": 461
+    },
+    {
+      "epoch": 1.3162393162393162,
+      "grad_norm": 0.23970941434834442,
+      "learning_rate": 3.125659978880676e-05,
+      "loss": 0.3865,
+      "step": 462
+    },
+    {
+      "epoch": 1.3190883190883191,
+      "grad_norm": 0.22027324796314413,
+      "learning_rate": 3.120380147835269e-05,
+      "loss": 0.3814,
+      "step": 463
+    },
+    {
+      "epoch": 1.3219373219373218,
+      "grad_norm": 0.1997543044951977,
+      "learning_rate": 3.115100316789863e-05,
+      "loss": 0.3612,
+      "step": 464
+    },
+    {
+      "epoch": 1.3247863247863247,
+      "grad_norm": 0.2298484065968129,
+      "learning_rate": 3.1098204857444565e-05,
+      "loss": 0.3486,
+      "step": 465
+    },
+    {
+      "epoch": 1.3276353276353277,
+      "grad_norm": 0.23459882831026682,
+      "learning_rate": 3.1045406546990495e-05,
+      "loss": 0.3531,
+      "step": 466
+    },
+    {
+      "epoch": 1.3304843304843303,
+      "grad_norm": 0.19872573469259944,
+      "learning_rate": 3.099260823653643e-05,
+      "loss": 0.3572,
+      "step": 467
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.22072697065816743,
+      "learning_rate": 3.093980992608237e-05,
+      "loss": 0.3722,
+      "step": 468
+    },
+    {
+      "epoch": 1.3361823361823362,
+      "grad_norm": 0.24096874308813088,
+      "learning_rate": 3.0887011615628304e-05,
+      "loss": 0.3596,
+      "step": 469
+    },
+    {
+      "epoch": 1.339031339031339,
+      "grad_norm": 0.19914935516155746,
+      "learning_rate": 3.0834213305174234e-05,
+      "loss": 0.3515,
+      "step": 470
+    },
+    {
+      "epoch": 1.341880341880342,
+      "grad_norm": 0.2064034362417422,
+      "learning_rate": 3.078141499472017e-05,
+      "loss": 0.3683,
+      "step": 471
+    },
+    {
+      "epoch": 1.3447293447293447,
+      "grad_norm": 0.21371146142366845,
+      "learning_rate": 3.0728616684266106e-05,
+      "loss": 0.3517,
+      "step": 472
+    },
+    {
+      "epoch": 1.3475783475783476,
+      "grad_norm": 0.21765447741613048,
+      "learning_rate": 3.0675818373812036e-05,
+      "loss": 0.3602,
+      "step": 473
+    },
+    {
+      "epoch": 1.3504273504273505,
+      "grad_norm": 0.20977178211422376,
+      "learning_rate": 3.062302006335798e-05,
+      "loss": 0.356,
+      "step": 474
+    },
+    {
+      "epoch": 1.3532763532763532,
+      "grad_norm": 0.2204679501825524,
+      "learning_rate": 3.057022175290391e-05,
+      "loss": 0.3638,
+      "step": 475
+    },
+    {
+      "epoch": 1.3561253561253561,
+      "grad_norm": 0.2351375455226534,
+      "learning_rate": 3.0517423442449845e-05,
+      "loss": 0.3771,
+      "step": 476
+    },
+    {
+      "epoch": 1.358974358974359,
+      "grad_norm": 0.32620489900955313,
+      "learning_rate": 3.0464625131995778e-05,
+      "loss": 0.3753,
+      "step": 477
+    },
+    {
+      "epoch": 1.3618233618233617,
+      "grad_norm": 0.222755197443211,
+      "learning_rate": 3.041182682154171e-05,
+      "loss": 0.3674,
+      "step": 478
+    },
+    {
+      "epoch": 1.3646723646723646,
+      "grad_norm": 0.22289274030678055,
+      "learning_rate": 3.0359028511087644e-05,
+      "loss": 0.3573,
+      "step": 479
+    },
+    {
+      "epoch": 1.3675213675213675,
+      "grad_norm": 0.23379350169681865,
+      "learning_rate": 3.0306230200633577e-05,
+      "loss": 0.3653,
+      "step": 480
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 0.18992629897318228,
+      "learning_rate": 3.0253431890179517e-05,
+      "loss": 0.3364,
+      "step": 481
+    },
+    {
+      "epoch": 1.3732193732193732,
+      "grad_norm": 0.21790569721651318,
+      "learning_rate": 3.020063357972545e-05,
+      "loss": 0.3583,
+      "step": 482
+    },
+    {
+      "epoch": 1.376068376068376,
+      "grad_norm": 0.21174281018764105,
+      "learning_rate": 3.0147835269271386e-05,
+      "loss": 0.3525,
+      "step": 483
+    },
+    {
+      "epoch": 1.378917378917379,
+      "grad_norm": 0.23195477568030867,
+      "learning_rate": 3.009503695881732e-05,
+      "loss": 0.3603,
+      "step": 484
+    },
+    {
+      "epoch": 1.381766381766382,
+      "grad_norm": 0.2185865276436869,
+      "learning_rate": 3.0042238648363252e-05,
+      "loss": 0.3751,
+      "step": 485
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.21213207655408392,
+      "learning_rate": 2.9989440337909185e-05,
+      "loss": 0.3695,
+      "step": 486
+    },
+    {
+      "epoch": 1.3874643874643875,
+      "grad_norm": 0.21005915264245484,
+      "learning_rate": 2.9936642027455125e-05,
+      "loss": 0.356,
+      "step": 487
+    },
+    {
+      "epoch": 1.3903133903133904,
+      "grad_norm": 0.21163776861595562,
+      "learning_rate": 2.9883843717001058e-05,
+      "loss": 0.3709,
+      "step": 488
+    },
+    {
+      "epoch": 1.393162393162393,
+      "grad_norm": 0.20028548407646818,
+      "learning_rate": 2.983104540654699e-05,
+      "loss": 0.3677,
+      "step": 489
+    },
+    {
+      "epoch": 1.396011396011396,
+      "grad_norm": 0.18880981649360443,
+      "learning_rate": 2.9778247096092927e-05,
+      "loss": 0.349,
+      "step": 490
+    },
+    {
+      "epoch": 1.398860398860399,
+      "grad_norm": 0.2234668654648058,
+      "learning_rate": 2.972544878563886e-05,
+      "loss": 0.3663,
+      "step": 491
+    },
+    {
+      "epoch": 1.4017094017094016,
+      "grad_norm": 0.21853579314131166,
+      "learning_rate": 2.9672650475184793e-05,
+      "loss": 0.3613,
+      "step": 492
+    },
+    {
+      "epoch": 1.4045584045584045,
+      "grad_norm": 0.22246500288700016,
+      "learning_rate": 2.9619852164730733e-05,
+      "loss": 0.3803,
+      "step": 493
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 0.20429826743140633,
+      "learning_rate": 2.9567053854276666e-05,
+      "loss": 0.3658,
+      "step": 494
+    },
+    {
+      "epoch": 1.4102564102564101,
+      "grad_norm": 0.2539252714453901,
+      "learning_rate": 2.95142555438226e-05,
+      "loss": 0.3734,
+      "step": 495
+    },
+    {
+      "epoch": 1.413105413105413,
+      "grad_norm": 0.2176197906573368,
+      "learning_rate": 2.9461457233368532e-05,
+      "loss": 0.3697,
+      "step": 496
+    },
+    {
+      "epoch": 1.415954415954416,
+      "grad_norm": 0.20262652133163678,
+      "learning_rate": 2.940865892291447e-05,
+      "loss": 0.3552,
+      "step": 497
+    },
+    {
+      "epoch": 1.4188034188034189,
+      "grad_norm": 0.2307771718731254,
+      "learning_rate": 2.93558606124604e-05,
+      "loss": 0.3483,
+      "step": 498
+    },
+    {
+      "epoch": 1.4216524216524218,
+      "grad_norm": 0.19342225282003825,
+      "learning_rate": 2.930306230200634e-05,
+      "loss": 0.3689,
+      "step": 499
+    },
+    {
+      "epoch": 1.4245014245014245,
+      "grad_norm": 0.20142188241082173,
+      "learning_rate": 2.9250263991552274e-05,
+      "loss": 0.3515,
+      "step": 500
+    },
+    {
+      "epoch": 1.4273504273504274,
+      "grad_norm": 0.21412132531144384,
+      "learning_rate": 2.9197465681098207e-05,
+      "loss": 0.3507,
+      "step": 501
+    },
+    {
+      "epoch": 1.4301994301994303,
+      "grad_norm": 0.25094985245186724,
+      "learning_rate": 2.914466737064414e-05,
+      "loss": 0.383,
+      "step": 502
+    },
+    {
+      "epoch": 1.433048433048433,
+      "grad_norm": 0.21232864199916715,
+      "learning_rate": 2.9091869060190073e-05,
+      "loss": 0.3764,
+      "step": 503
+    },
+    {
+      "epoch": 1.435897435897436,
+      "grad_norm": 0.22049102509543148,
+      "learning_rate": 2.903907074973601e-05,
+      "loss": 0.3595,
+      "step": 504
+    },
+    {
+      "epoch": 1.4387464387464388,
+      "grad_norm": 0.21682319020739224,
+      "learning_rate": 2.8986272439281943e-05,
+      "loss": 0.3547,
+      "step": 505
+    },
+    {
+      "epoch": 1.4415954415954415,
+      "grad_norm": 0.23709929673053004,
+      "learning_rate": 2.8933474128827883e-05,
+      "loss": 0.3687,
+      "step": 506
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.2101951752079666,
+      "learning_rate": 2.8880675818373816e-05,
+      "loss": 0.359,
+      "step": 507
+    },
+    {
+      "epoch": 1.4472934472934473,
+      "grad_norm": 0.2474601153654613,
+      "learning_rate": 2.882787750791975e-05,
+      "loss": 0.3705,
+      "step": 508
+    },
+    {
+      "epoch": 1.45014245014245,
+      "grad_norm": 0.21254139978772235,
+      "learning_rate": 2.877507919746568e-05,
+      "loss": 0.3564,
+      "step": 509
+    },
+    {
+      "epoch": 1.452991452991453,
+      "grad_norm": 0.24070048810217534,
+      "learning_rate": 2.8722280887011615e-05,
+      "loss": 0.3706,
+      "step": 510
+    },
+    {
+      "epoch": 1.4558404558404558,
+      "grad_norm": 0.2987871246913325,
+      "learning_rate": 2.8669482576557548e-05,
+      "loss": 0.3687,
+      "step": 511
+    },
+    {
+      "epoch": 1.4586894586894588,
+      "grad_norm": 0.2300676722177236,
+      "learning_rate": 2.8616684266103487e-05,
+      "loss": 0.3647,
+      "step": 512
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.26581166943990137,
+      "learning_rate": 2.856388595564942e-05,
+      "loss": 0.372,
+      "step": 513
+    },
+    {
+      "epoch": 1.4643874643874644,
+      "grad_norm": 0.3655749618941395,
+      "learning_rate": 2.8511087645195357e-05,
+      "loss": 0.3776,
+      "step": 514
+    },
+    {
+      "epoch": 1.4672364672364673,
+      "grad_norm": 0.26115893033445464,
+      "learning_rate": 2.845828933474129e-05,
+      "loss": 0.3645,
+      "step": 515
+    },
+    {
+      "epoch": 1.4700854700854702,
+      "grad_norm": 0.21588849574942176,
+      "learning_rate": 2.8405491024287223e-05,
+      "loss": 0.3554,
+      "step": 516
+    },
+    {
+      "epoch": 1.4729344729344729,
+      "grad_norm": 0.22519123294647936,
+      "learning_rate": 2.8352692713833156e-05,
+      "loss": 0.3378,
+      "step": 517
+    },
+    {
+      "epoch": 1.4757834757834758,
+      "grad_norm": 0.24778294747275895,
+      "learning_rate": 2.8299894403379096e-05,
+      "loss": 0.3738,
+      "step": 518
+    },
+    {
+      "epoch": 1.4786324786324787,
+      "grad_norm": 0.2265365912202125,
+      "learning_rate": 2.824709609292503e-05,
+      "loss": 0.3691,
+      "step": 519
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.22460772397621678,
+      "learning_rate": 2.819429778247096e-05,
+      "loss": 0.3691,
+      "step": 520
+    },
+    {
+      "epoch": 1.4843304843304843,
+      "grad_norm": 0.21608141296269037,
+      "learning_rate": 2.8141499472016898e-05,
+      "loss": 0.3467,
+      "step": 521
+    },
+    {
+      "epoch": 1.4871794871794872,
+      "grad_norm": 0.2213366031457417,
+      "learning_rate": 2.808870116156283e-05,
+      "loss": 0.3622,
+      "step": 522
+    },
+    {
+      "epoch": 1.49002849002849,
+      "grad_norm": 0.21175641658464092,
+      "learning_rate": 2.8035902851108764e-05,
+      "loss": 0.3667,
+      "step": 523
+    },
+    {
+      "epoch": 1.4928774928774928,
+      "grad_norm": 0.2543379886177831,
+      "learning_rate": 2.7983104540654697e-05,
+      "loss": 0.3909,
+      "step": 524
+    },
+    {
+      "epoch": 1.4957264957264957,
+      "grad_norm": 0.2394839478000981,
+      "learning_rate": 2.7930306230200637e-05,
+      "loss": 0.3694,
+      "step": 525
+    },
+    {
+      "epoch": 1.4985754985754987,
+      "grad_norm": 0.2275501360298665,
+      "learning_rate": 2.787750791974657e-05,
+      "loss": 0.3559,
+      "step": 526
+    },
+    {
+      "epoch": 1.5014245014245016,
+      "grad_norm": 0.2763856292468043,
+      "learning_rate": 2.7824709609292503e-05,
+      "loss": 0.3508,
+      "step": 527
+    },
+    {
+      "epoch": 1.5042735042735043,
+      "grad_norm": 0.23957120121880873,
+      "learning_rate": 2.777191129883844e-05,
+      "loss": 0.3589,
+      "step": 528
+    },
+    {
+      "epoch": 1.5071225071225072,
+      "grad_norm": 0.23319500553009573,
+      "learning_rate": 2.7719112988384372e-05,
+      "loss": 0.3703,
+      "step": 529
+    },
+    {
+      "epoch": 1.50997150997151,
+      "grad_norm": 0.24921375732341508,
+      "learning_rate": 2.7666314677930305e-05,
+      "loss": 0.359,
+      "step": 530
+    },
+    {
+      "epoch": 1.5128205128205128,
+      "grad_norm": 0.21008771333534176,
+      "learning_rate": 2.7613516367476245e-05,
+      "loss": 0.3572,
+      "step": 531
+    },
+    {
+      "epoch": 1.5156695156695157,
+      "grad_norm": 0.21839097000403634,
+      "learning_rate": 2.7560718057022178e-05,
+      "loss": 0.3657,
+      "step": 532
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 0.22689335860669002,
+      "learning_rate": 2.750791974656811e-05,
+      "loss": 0.3707,
+      "step": 533
+    },
+    {
+      "epoch": 1.5213675213675213,
+      "grad_norm": 0.2454238635949199,
+      "learning_rate": 2.7455121436114044e-05,
+      "loss": 0.3665,
+      "step": 534
+    },
+    {
+      "epoch": 1.5242165242165242,
+      "grad_norm": 0.21117182341715873,
+      "learning_rate": 2.740232312565998e-05,
+      "loss": 0.3543,
+      "step": 535
+    },
+    {
+      "epoch": 1.5270655270655271,
+      "grad_norm": 0.2167859209575041,
+      "learning_rate": 2.7349524815205913e-05,
+      "loss": 0.3648,
+      "step": 536
+    },
+    {
+      "epoch": 1.5299145299145298,
+      "grad_norm": 0.25424621395457675,
+      "learning_rate": 2.7296726504751853e-05,
+      "loss": 0.3783,
+      "step": 537
+    },
+    {
+      "epoch": 1.5327635327635327,
+      "grad_norm": 0.21880596935336596,
+      "learning_rate": 2.7243928194297786e-05,
+      "loss": 0.3628,
+      "step": 538
+    },
+    {
+      "epoch": 1.5356125356125356,
+      "grad_norm": 0.21692887185646115,
+      "learning_rate": 2.719112988384372e-05,
+      "loss": 0.363,
+      "step": 539
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.2575849502562457,
+      "learning_rate": 2.7138331573389652e-05,
+      "loss": 0.369,
+      "step": 540
+    },
+    {
+      "epoch": 1.5413105413105415,
+      "grad_norm": 0.23885268332551002,
+      "learning_rate": 2.7085533262935585e-05,
+      "loss": 0.3501,
+      "step": 541
+    },
+    {
+      "epoch": 1.5441595441595442,
+      "grad_norm": 0.21686275578065198,
+      "learning_rate": 2.703273495248152e-05,
+      "loss": 0.3601,
+      "step": 542
+    },
+    {
+      "epoch": 1.547008547008547,
+      "grad_norm": 0.233651783121331,
+      "learning_rate": 2.6979936642027458e-05,
+      "loss": 0.3629,
+      "step": 543
+    },
+    {
+      "epoch": 1.54985754985755,
+      "grad_norm": 0.23142549801608311,
+      "learning_rate": 2.6927138331573394e-05,
+      "loss": 0.3756,
+      "step": 544
+    },
+    {
+      "epoch": 1.5527065527065527,
+      "grad_norm": 0.23654232452295837,
+      "learning_rate": 2.6874340021119327e-05,
+      "loss": 0.3836,
+      "step": 545
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.23955118714830956,
+      "learning_rate": 2.682154171066526e-05,
+      "loss": 0.3599,
+      "step": 546
+    },
+    {
+      "epoch": 1.5584045584045585,
+      "grad_norm": 0.22135467807774808,
+      "learning_rate": 2.6768743400211193e-05,
+      "loss": 0.3747,
+      "step": 547
+    },
+    {
+      "epoch": 1.5612535612535612,
+      "grad_norm": 0.21590468574896746,
+      "learning_rate": 2.6715945089757126e-05,
+      "loss": 0.3469,
+      "step": 548
+    },
+    {
+      "epoch": 1.564102564102564,
+      "grad_norm": 0.24081611853474838,
+      "learning_rate": 2.6663146779303063e-05,
+      "loss": 0.3448,
+      "step": 549
+    },
+    {
+      "epoch": 1.566951566951567,
+      "grad_norm": 0.22050273169428228,
+      "learning_rate": 2.6610348468849e-05,
+      "loss": 0.3665,
+      "step": 550
+    },
+    {
+      "epoch": 1.5698005698005697,
+      "grad_norm": 0.2358630867918145,
+      "learning_rate": 2.6557550158394935e-05,
+      "loss": 0.3693,
+      "step": 551
+    },
+    {
+      "epoch": 1.5726495726495726,
+      "grad_norm": 0.2256231077467112,
+      "learning_rate": 2.650475184794087e-05,
+      "loss": 0.3575,
+      "step": 552
+    },
+    {
+      "epoch": 1.5754985754985755,
+      "grad_norm": 0.2193287720903084,
+      "learning_rate": 2.64519535374868e-05,
+      "loss": 0.3602,
+      "step": 553
+    },
+    {
+      "epoch": 1.5783475783475782,
+      "grad_norm": 0.23352787405471498,
+      "learning_rate": 2.6399155227032734e-05,
+      "loss": 0.3697,
+      "step": 554
+    },
+    {
+      "epoch": 1.5811965811965814,
+      "grad_norm": 0.2395424874218671,
+      "learning_rate": 2.6346356916578667e-05,
+      "loss": 0.3635,
+      "step": 555
+    },
+    {
+      "epoch": 1.584045584045584,
+      "grad_norm": 0.43346665073783924,
+      "learning_rate": 2.6293558606124607e-05,
+      "loss": 0.3691,
+      "step": 556
+    },
+    {
+      "epoch": 1.5868945868945867,
+      "grad_norm": 0.23220070707086665,
+      "learning_rate": 2.624076029567054e-05,
+      "loss": 0.3731,
+      "step": 557
+    },
+    {
+      "epoch": 1.5897435897435899,
+      "grad_norm": 0.20943796833119865,
+      "learning_rate": 2.6187961985216473e-05,
+      "loss": 0.3698,
+      "step": 558
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.2257633484876588,
+      "learning_rate": 2.613516367476241e-05,
+      "loss": 0.3635,
+      "step": 559
+    },
+    {
+      "epoch": 1.5954415954415955,
+      "grad_norm": 0.22794583966236925,
+      "learning_rate": 2.6082365364308343e-05,
+      "loss": 0.3619,
+      "step": 560
+    },
+    {
+      "epoch": 1.5982905982905984,
+      "grad_norm": 0.21844307238312177,
+      "learning_rate": 2.6029567053854276e-05,
+      "loss": 0.3735,
+      "step": 561
+    },
+    {
+      "epoch": 1.601139601139601,
+      "grad_norm": 0.25182987504775967,
+      "learning_rate": 2.5976768743400215e-05,
+      "loss": 0.3657,
+      "step": 562
+    },
+    {
+      "epoch": 1.603988603988604,
+      "grad_norm": 0.2328339817261743,
+      "learning_rate": 2.5923970432946148e-05,
+      "loss": 0.3617,
+      "step": 563
+    },
+    {
+      "epoch": 1.606837606837607,
+      "grad_norm": 0.2198747745202425,
+      "learning_rate": 2.587117212249208e-05,
+      "loss": 0.3567,
+      "step": 564
+    },
+    {
+      "epoch": 1.6096866096866096,
+      "grad_norm": 0.22458365808076017,
+      "learning_rate": 2.5818373812038014e-05,
+      "loss": 0.3633,
+      "step": 565
+    },
+    {
+      "epoch": 1.6125356125356125,
+      "grad_norm": 0.2050810486479959,
+      "learning_rate": 2.576557550158395e-05,
+      "loss": 0.3662,
+      "step": 566
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.23109142394514665,
+      "learning_rate": 2.5712777191129884e-05,
+      "loss": 0.377,
+      "step": 567
+    },
+    {
+      "epoch": 1.618233618233618,
+      "grad_norm": 0.20488685566526976,
+      "learning_rate": 2.5659978880675823e-05,
+      "loss": 0.365,
+      "step": 568
+    },
+    {
+      "epoch": 1.6210826210826212,
+      "grad_norm": 0.22434853437066676,
+      "learning_rate": 2.5607180570221756e-05,
+      "loss": 0.359,
+      "step": 569
+    },
+    {
+      "epoch": 1.623931623931624,
+      "grad_norm": 0.22035628663691662,
+      "learning_rate": 2.555438225976769e-05,
+      "loss": 0.3613,
+      "step": 570
+    },
+    {
+      "epoch": 1.6267806267806266,
+      "grad_norm": 0.2296131512377323,
+      "learning_rate": 2.5501583949313622e-05,
+      "loss": 0.3583,
+      "step": 571
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.22699499325399286,
+      "learning_rate": 2.5448785638859555e-05,
+      "loss": 0.3615,
+      "step": 572
+    },
+    {
+      "epoch": 1.6324786324786325,
+      "grad_norm": 0.23137613302922347,
+      "learning_rate": 2.5395987328405492e-05,
+      "loss": 0.3493,
+      "step": 573
+    },
+    {
+      "epoch": 1.6353276353276354,
+      "grad_norm": 0.23513034060758498,
+      "learning_rate": 2.5343189017951425e-05,
+      "loss": 0.3536,
+      "step": 574
+    },
+    {
+      "epoch": 1.6381766381766383,
+      "grad_norm": 0.20556474209048517,
+      "learning_rate": 2.5290390707497365e-05,
+      "loss": 0.3494,
+      "step": 575
+    },
+    {
+      "epoch": 1.641025641025641,
+      "grad_norm": 0.20951535308942912,
+      "learning_rate": 2.5237592397043298e-05,
+      "loss": 0.3608,
+      "step": 576
+    },
+    {
+      "epoch": 1.6438746438746439,
+      "grad_norm": 0.21051061589323755,
+      "learning_rate": 2.518479408658923e-05,
+      "loss": 0.343,
+      "step": 577
+    },
+    {
+      "epoch": 1.6467236467236468,
+      "grad_norm": 0.20389495741565292,
+      "learning_rate": 2.5131995776135164e-05,
+      "loss": 0.3515,
+      "step": 578
+    },
+    {
+      "epoch": 1.6495726495726495,
+      "grad_norm": 0.20666357754178133,
+      "learning_rate": 2.5079197465681097e-05,
+      "loss": 0.3637,
+      "step": 579
+    },
+    {
+      "epoch": 1.6524216524216524,
+      "grad_norm": 0.2124678103193664,
+      "learning_rate": 2.5026399155227033e-05,
+      "loss": 0.3519,
+      "step": 580
+    },
+    {
+      "epoch": 1.6552706552706553,
+      "grad_norm": 0.21745017494419377,
+      "learning_rate": 2.497360084477297e-05,
+      "loss": 0.3594,
+      "step": 581
+    },
+    {
+      "epoch": 1.658119658119658,
+      "grad_norm": 0.1903592780911563,
+      "learning_rate": 2.4920802534318902e-05,
+      "loss": 0.336,
+      "step": 582
+    },
+    {
+      "epoch": 1.660968660968661,
+      "grad_norm": 0.23741085468942869,
+      "learning_rate": 2.486800422386484e-05,
+      "loss": 0.3543,
+      "step": 583
+    },
+    {
+      "epoch": 1.6638176638176638,
+      "grad_norm": 0.20157828540341033,
+      "learning_rate": 2.4815205913410772e-05,
+      "loss": 0.3342,
+      "step": 584
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.2204911302874124,
+      "learning_rate": 2.4762407602956705e-05,
+      "loss": 0.3586,
+      "step": 585
+    },
+    {
+      "epoch": 1.6695156695156697,
+      "grad_norm": 0.21109017199979024,
+      "learning_rate": 2.470960929250264e-05,
+      "loss": 0.3646,
+      "step": 586
+    },
+    {
+      "epoch": 1.6723646723646723,
+      "grad_norm": 0.20646510827582915,
+      "learning_rate": 2.4656810982048574e-05,
+      "loss": 0.3563,
+      "step": 587
+    },
+    {
+      "epoch": 1.6752136752136753,
+      "grad_norm": 0.21584768343148578,
+      "learning_rate": 2.460401267159451e-05,
+      "loss": 0.3535,
+      "step": 588
+    },
+    {
+      "epoch": 1.6780626780626782,
+      "grad_norm": 0.23185019886698816,
+      "learning_rate": 2.4551214361140447e-05,
+      "loss": 0.3697,
+      "step": 589
+    },
+    {
+      "epoch": 1.6809116809116809,
+      "grad_norm": 0.21268814828277405,
+      "learning_rate": 2.449841605068638e-05,
+      "loss": 0.3551,
+      "step": 590
+    },
+    {
+      "epoch": 1.6837606837606838,
+      "grad_norm": 0.21604754434232787,
+      "learning_rate": 2.4445617740232313e-05,
+      "loss": 0.3655,
+      "step": 591
+    },
+    {
+      "epoch": 1.6866096866096867,
+      "grad_norm": 0.22278622809811766,
+      "learning_rate": 2.439281942977825e-05,
+      "loss": 0.3538,
+      "step": 592
+    },
+    {
+      "epoch": 1.6894586894586894,
+      "grad_norm": 0.2052145744246248,
+      "learning_rate": 2.4340021119324182e-05,
+      "loss": 0.356,
+      "step": 593
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.19837577056881597,
+      "learning_rate": 2.4287222808870115e-05,
+      "loss": 0.356,
+      "step": 594
+    },
+    {
+      "epoch": 1.6951566951566952,
+      "grad_norm": 0.2605768702541718,
+      "learning_rate": 2.4234424498416052e-05,
+      "loss": 0.376,
+      "step": 595
+    },
+    {
+      "epoch": 1.698005698005698,
+      "grad_norm": 0.2150615912164544,
+      "learning_rate": 2.4181626187961988e-05,
+      "loss": 0.3442,
+      "step": 596
+    },
+    {
+      "epoch": 1.7008547008547008,
+      "grad_norm": 0.20499211512651233,
+      "learning_rate": 2.412882787750792e-05,
+      "loss": 0.3699,
+      "step": 597
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 0.21382275219015714,
+      "learning_rate": 2.4076029567053857e-05,
+      "loss": 0.3565,
+      "step": 598
+    },
+    {
+      "epoch": 1.7065527065527064,
+      "grad_norm": 0.20994705780047032,
+      "learning_rate": 2.402323125659979e-05,
+      "loss": 0.3598,
+      "step": 599
+    },
+    {
+      "epoch": 1.7094017094017095,
+      "grad_norm": 0.20135559643209863,
+      "learning_rate": 2.3970432946145723e-05,
+      "loss": 0.3596,
+      "step": 600
+    },
+    {
+      "epoch": 1.7122507122507122,
+      "grad_norm": 0.20036899067341168,
+      "learning_rate": 2.391763463569166e-05,
+      "loss": 0.3569,
+      "step": 601
+    },
+    {
+      "epoch": 1.7150997150997151,
+      "grad_norm": 0.21002141879928451,
+      "learning_rate": 2.3864836325237593e-05,
+      "loss": 0.3654,
+      "step": 602
+    },
+    {
+      "epoch": 1.717948717948718,
+      "grad_norm": 0.22764067995733198,
+      "learning_rate": 2.381203801478353e-05,
+      "loss": 0.3625,
+      "step": 603
+    },
+    {
+      "epoch": 1.7207977207977208,
+      "grad_norm": 0.19291409387702949,
+      "learning_rate": 2.3759239704329462e-05,
+      "loss": 0.3712,
+      "step": 604
+    },
+    {
+      "epoch": 1.7236467236467237,
+      "grad_norm": 0.19990638382401998,
+      "learning_rate": 2.37064413938754e-05,
+      "loss": 0.3466,
+      "step": 605
+    },
+    {
+      "epoch": 1.7264957264957266,
+      "grad_norm": 0.21172503078878085,
+      "learning_rate": 2.365364308342133e-05,
+      "loss": 0.373,
+      "step": 606
+    },
+    {
+      "epoch": 1.7293447293447293,
+      "grad_norm": 0.21340989500402005,
+      "learning_rate": 2.3600844772967265e-05,
+      "loss": 0.3607,
+      "step": 607
+    },
+    {
+      "epoch": 1.7321937321937322,
+      "grad_norm": 0.21348773962215495,
+      "learning_rate": 2.35480464625132e-05,
+      "loss": 0.3618,
+      "step": 608
+    },
+    {
+      "epoch": 1.735042735042735,
+      "grad_norm": 0.21866328753111697,
+      "learning_rate": 2.3495248152059134e-05,
+      "loss": 0.3556,
+      "step": 609
+    },
+    {
+      "epoch": 1.7378917378917378,
+      "grad_norm": 0.19763848284771224,
+      "learning_rate": 2.3442449841605067e-05,
+      "loss": 0.3607,
+      "step": 610
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.2152332174618033,
+      "learning_rate": 2.3389651531151003e-05,
+      "loss": 0.3704,
+      "step": 611
+    },
+    {
+      "epoch": 1.7435897435897436,
+      "grad_norm": 0.19269940169730435,
+      "learning_rate": 2.333685322069694e-05,
+      "loss": 0.3638,
+      "step": 612
+    },
+    {
+      "epoch": 1.7464387464387463,
+      "grad_norm": 0.1962642279399981,
+      "learning_rate": 2.3284054910242873e-05,
+      "loss": 0.3482,
+      "step": 613
+    },
+    {
+      "epoch": 1.7492877492877494,
+      "grad_norm": 0.2364856796721012,
+      "learning_rate": 2.323125659978881e-05,
+      "loss": 0.347,
+      "step": 614
+    },
+    {
+      "epoch": 1.7521367521367521,
+      "grad_norm": 0.22509925838340694,
+      "learning_rate": 2.3178458289334742e-05,
+      "loss": 0.3604,
+      "step": 615
+    },
+    {
+      "epoch": 1.7549857549857548,
+      "grad_norm": 0.2488709063431033,
+      "learning_rate": 2.3125659978880675e-05,
+      "loss": 0.3657,
+      "step": 616
+    },
+    {
+      "epoch": 1.757834757834758,
+      "grad_norm": 0.2398318070372559,
+      "learning_rate": 2.307286166842661e-05,
+      "loss": 0.3629,
+      "step": 617
+    },
+    {
+      "epoch": 1.7606837606837606,
+      "grad_norm": 0.22323781791119499,
+      "learning_rate": 2.3020063357972545e-05,
+      "loss": 0.359,
+      "step": 618
+    },
+    {
+      "epoch": 1.7635327635327636,
+      "grad_norm": 0.2022321694186785,
+      "learning_rate": 2.296726504751848e-05,
+      "loss": 0.3604,
+      "step": 619
+    },
+    {
+      "epoch": 1.7663817663817665,
+      "grad_norm": 0.21110134786679816,
+      "learning_rate": 2.2914466737064417e-05,
+      "loss": 0.3687,
+      "step": 620
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.19625112710435982,
+      "learning_rate": 2.286166842661035e-05,
+      "loss": 0.3599,
+      "step": 621
+    },
+    {
+      "epoch": 1.772079772079772,
+      "grad_norm": 0.20619851901285047,
+      "learning_rate": 2.2808870116156283e-05,
+      "loss": 0.3723,
+      "step": 622
+    },
+    {
+      "epoch": 1.774928774928775,
+      "grad_norm": 0.20469389194337959,
+      "learning_rate": 2.275607180570222e-05,
+      "loss": 0.3661,
+      "step": 623
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.20778072555494564,
+      "learning_rate": 2.2703273495248153e-05,
+      "loss": 0.3585,
+      "step": 624
+    },
+    {
+      "epoch": 1.7806267806267806,
+      "grad_norm": 0.19870626339902325,
+      "learning_rate": 2.2650475184794086e-05,
+      "loss": 0.3441,
+      "step": 625
+    },
+    {
+      "epoch": 1.7834757834757835,
+      "grad_norm": 0.19114151393977027,
+      "learning_rate": 2.2597676874340022e-05,
+      "loss": 0.3591,
+      "step": 626
+    },
+    {
+      "epoch": 1.7863247863247862,
+      "grad_norm": 0.1982172563967693,
+      "learning_rate": 2.254487856388596e-05,
+      "loss": 0.3854,
+      "step": 627
+    },
+    {
+      "epoch": 1.7891737891737893,
+      "grad_norm": 0.20351344941450986,
+      "learning_rate": 2.249208025343189e-05,
+      "loss": 0.3822,
+      "step": 628
+    },
+    {
+      "epoch": 1.792022792022792,
+      "grad_norm": 0.19819562909909946,
+      "learning_rate": 2.2439281942977825e-05,
+      "loss": 0.3645,
+      "step": 629
+    },
+    {
+      "epoch": 1.7948717948717947,
+      "grad_norm": 0.20300969334790384,
+      "learning_rate": 2.238648363252376e-05,
+      "loss": 0.3631,
+      "step": 630
+    },
+    {
+      "epoch": 1.7977207977207978,
+      "grad_norm": 0.20862867991682074,
+      "learning_rate": 2.2333685322069694e-05,
+      "loss": 0.3619,
+      "step": 631
+    },
+    {
+      "epoch": 1.8005698005698005,
+      "grad_norm": 0.21833179546935874,
+      "learning_rate": 2.2280887011615627e-05,
+      "loss": 0.3608,
+      "step": 632
+    },
+    {
+      "epoch": 1.8034188034188035,
+      "grad_norm": 0.6005070390326573,
+      "learning_rate": 2.2228088701161563e-05,
+      "loss": 0.3965,
+      "step": 633
+    },
+    {
+      "epoch": 1.8062678062678064,
+      "grad_norm": 0.35943818916514836,
+      "learning_rate": 2.21752903907075e-05,
+      "loss": 0.3594,
+      "step": 634
+    },
+    {
+      "epoch": 1.809116809116809,
+      "grad_norm": 0.2159038665602852,
+      "learning_rate": 2.2122492080253433e-05,
+      "loss": 0.3755,
+      "step": 635
+    },
+    {
+      "epoch": 1.811965811965812,
+      "grad_norm": 0.20549883442788655,
+      "learning_rate": 2.206969376979937e-05,
+      "loss": 0.3451,
+      "step": 636
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 0.20440135853564803,
+      "learning_rate": 2.2016895459345302e-05,
+      "loss": 0.3497,
+      "step": 637
+    },
+    {
+      "epoch": 1.8176638176638176,
+      "grad_norm": 0.25722335101398636,
+      "learning_rate": 2.1964097148891235e-05,
+      "loss": 0.3639,
+      "step": 638
+    },
+    {
+      "epoch": 1.8205128205128205,
+      "grad_norm": 0.19953052748781105,
+      "learning_rate": 2.191129883843717e-05,
+      "loss": 0.3481,
+      "step": 639
+    },
+    {
+      "epoch": 1.8233618233618234,
+      "grad_norm": 0.2341568009909009,
+      "learning_rate": 2.1858500527983104e-05,
+      "loss": 0.3593,
+      "step": 640
+    },
+    {
+      "epoch": 1.826210826210826,
+      "grad_norm": 0.21325737340547804,
+      "learning_rate": 2.180570221752904e-05,
+      "loss": 0.3636,
+      "step": 641
+    },
+    {
+      "epoch": 1.8290598290598292,
+      "grad_norm": 0.22284714887271365,
+      "learning_rate": 2.1752903907074977e-05,
+      "loss": 0.3797,
+      "step": 642
+    },
+    {
+      "epoch": 1.831908831908832,
+      "grad_norm": 0.20422527863661058,
+      "learning_rate": 2.170010559662091e-05,
+      "loss": 0.3527,
+      "step": 643
+    },
+    {
+      "epoch": 1.8347578347578346,
+      "grad_norm": 0.22652999934815843,
+      "learning_rate": 2.1647307286166843e-05,
+      "loss": 0.3717,
+      "step": 644
+    },
+    {
+      "epoch": 1.8376068376068377,
+      "grad_norm": 0.24779116320854475,
+      "learning_rate": 2.159450897571278e-05,
+      "loss": 0.3706,
+      "step": 645
+    },
+    {
+      "epoch": 1.8404558404558404,
+      "grad_norm": 0.21671225191720797,
+      "learning_rate": 2.1541710665258713e-05,
+      "loss": 0.3386,
+      "step": 646
+    },
+    {
+      "epoch": 1.8433048433048433,
+      "grad_norm": 0.22555743360580618,
+      "learning_rate": 2.1488912354804646e-05,
+      "loss": 0.3804,
+      "step": 647
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.23617998499869738,
+      "learning_rate": 2.1436114044350582e-05,
+      "loss": 0.3723,
+      "step": 648
+    },
+    {
+      "epoch": 1.849002849002849,
+      "grad_norm": 0.23014596171154433,
+      "learning_rate": 2.138331573389652e-05,
+      "loss": 0.3526,
+      "step": 649
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.22916111003439332,
+      "learning_rate": 2.133051742344245e-05,
+      "loss": 0.3451,
+      "step": 650
+    },
+    {
+      "epoch": 1.8547008547008548,
+      "grad_norm": 0.19810306429063848,
+      "learning_rate": 2.1277719112988384e-05,
+      "loss": 0.349,
+      "step": 651
+    },
+    {
+      "epoch": 1.8575498575498575,
+      "grad_norm": 0.2202335756589606,
+      "learning_rate": 2.122492080253432e-05,
+      "loss": 0.3653,
+      "step": 652
+    },
+    {
+      "epoch": 1.8603988603988604,
+      "grad_norm": 0.19827409095956613,
+      "learning_rate": 2.1172122492080254e-05,
+      "loss": 0.3457,
+      "step": 653
+    },
+    {
+      "epoch": 1.8632478632478633,
+      "grad_norm": 0.19935795905418485,
+      "learning_rate": 2.1119324181626187e-05,
+      "loss": 0.3554,
+      "step": 654
+    },
+    {
+      "epoch": 1.866096866096866,
+      "grad_norm": 0.20847219512744983,
+      "learning_rate": 2.1066525871172123e-05,
+      "loss": 0.351,
+      "step": 655
+    },
+    {
+      "epoch": 1.868945868945869,
+      "grad_norm": 0.2156361373347463,
+      "learning_rate": 2.1013727560718056e-05,
+      "loss": 0.3618,
+      "step": 656
+    },
+    {
+      "epoch": 1.8717948717948718,
+      "grad_norm": 0.20100877639931164,
+      "learning_rate": 2.0960929250263993e-05,
+      "loss": 0.3527,
+      "step": 657
+    },
+    {
+      "epoch": 1.8746438746438745,
+      "grad_norm": 0.21961338320882828,
+      "learning_rate": 2.090813093980993e-05,
+      "loss": 0.3655,
+      "step": 658
+    },
+    {
+      "epoch": 1.8774928774928776,
+      "grad_norm": 0.22970963785057988,
+      "learning_rate": 2.0855332629355862e-05,
+      "loss": 0.3578,
+      "step": 659
+    },
+    {
+      "epoch": 1.8803418803418803,
+      "grad_norm": 0.1985917394092499,
+      "learning_rate": 2.0802534318901795e-05,
+      "loss": 0.3555,
+      "step": 660
+    },
+    {
+      "epoch": 1.8831908831908832,
+      "grad_norm": 0.19782225715456844,
+      "learning_rate": 2.074973600844773e-05,
+      "loss": 0.3544,
+      "step": 661
+    },
+    {
+      "epoch": 1.8860398860398861,
+      "grad_norm": 0.24471883420488305,
+      "learning_rate": 2.0696937697993664e-05,
+      "loss": 0.3689,
+      "step": 662
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.22271236910553516,
+      "learning_rate": 2.0644139387539597e-05,
+      "loss": 0.3871,
+      "step": 663
+    },
+    {
+      "epoch": 1.8917378917378918,
+      "grad_norm": 0.19560513729734122,
+      "learning_rate": 2.0591341077085534e-05,
+      "loss": 0.3561,
+      "step": 664
+    },
+    {
+      "epoch": 1.8945868945868947,
+      "grad_norm": 0.21617290509605866,
+      "learning_rate": 2.053854276663147e-05,
+      "loss": 0.3286,
+      "step": 665
+    },
+    {
+      "epoch": 1.8974358974358974,
+      "grad_norm": 0.22798617788173128,
+      "learning_rate": 2.0485744456177403e-05,
+      "loss": 0.3688,
+      "step": 666
+    },
+    {
+      "epoch": 1.9002849002849003,
+      "grad_norm": 0.21021348614356922,
+      "learning_rate": 2.043294614572334e-05,
+      "loss": 0.3562,
+      "step": 667
+    },
+    {
+      "epoch": 1.9031339031339032,
+      "grad_norm": 0.20996089592223216,
+      "learning_rate": 2.0380147835269273e-05,
+      "loss": 0.361,
+      "step": 668
+    },
+    {
+      "epoch": 1.9059829059829059,
+      "grad_norm": 0.23565900871607817,
+      "learning_rate": 2.0327349524815206e-05,
+      "loss": 0.3754,
+      "step": 669
+    },
+    {
+      "epoch": 1.9088319088319088,
+      "grad_norm": 0.18858686585228232,
+      "learning_rate": 2.0274551214361142e-05,
+      "loss": 0.3516,
+      "step": 670
+    },
+    {
+      "epoch": 1.9116809116809117,
+      "grad_norm": 0.1893165779352585,
+      "learning_rate": 2.0221752903907075e-05,
+      "loss": 0.3366,
+      "step": 671
+    },
+    {
+      "epoch": 1.9145299145299144,
+      "grad_norm": 0.21668943186182438,
+      "learning_rate": 2.016895459345301e-05,
+      "loss": 0.3502,
+      "step": 672
+    },
+    {
+      "epoch": 1.9173789173789175,
+      "grad_norm": 0.20149473922057556,
+      "learning_rate": 2.0116156282998944e-05,
+      "loss": 0.3615,
+      "step": 673
+    },
+    {
+      "epoch": 1.9202279202279202,
+      "grad_norm": 0.25869865465945596,
+      "learning_rate": 2.006335797254488e-05,
+      "loss": 0.3691,
+      "step": 674
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.2036066715728498,
+      "learning_rate": 2.0010559662090814e-05,
+      "loss": 0.3502,
+      "step": 675
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.2004788360097017,
+      "learning_rate": 1.9957761351636747e-05,
+      "loss": 0.3594,
+      "step": 676
+    },
+    {
+      "epoch": 1.9287749287749287,
+      "grad_norm": 0.21360843600832238,
+      "learning_rate": 1.9904963041182683e-05,
+      "loss": 0.3585,
+      "step": 677
+    },
+    {
+      "epoch": 1.9316239316239316,
+      "grad_norm": 0.19762168727225163,
+      "learning_rate": 1.9852164730728616e-05,
+      "loss": 0.3675,
+      "step": 678
+    },
+    {
+      "epoch": 1.9344729344729346,
+      "grad_norm": 0.20229203944486152,
+      "learning_rate": 1.9799366420274552e-05,
+      "loss": 0.355,
+      "step": 679
+    },
+    {
+      "epoch": 1.9373219373219372,
+      "grad_norm": 0.2022552654861909,
+      "learning_rate": 1.974656810982049e-05,
+      "loss": 0.3547,
+      "step": 680
+    },
+    {
+      "epoch": 1.9401709401709402,
+      "grad_norm": 0.2217881214201902,
+      "learning_rate": 1.9693769799366422e-05,
+      "loss": 0.3662,
+      "step": 681
+    },
+    {
+      "epoch": 1.943019943019943,
+      "grad_norm": 0.21089292714234747,
+      "learning_rate": 1.9640971488912355e-05,
+      "loss": 0.3632,
+      "step": 682
+    },
+    {
+      "epoch": 1.9458689458689458,
+      "grad_norm": 0.21183549408551403,
+      "learning_rate": 1.958817317845829e-05,
+      "loss": 0.3685,
+      "step": 683
+    },
+    {
+      "epoch": 1.9487179487179487,
+      "grad_norm": 0.22399399219360525,
+      "learning_rate": 1.9535374868004224e-05,
+      "loss": 0.389,
+      "step": 684
+    },
+    {
+      "epoch": 1.9515669515669516,
+      "grad_norm": 0.23478634085970435,
+      "learning_rate": 1.9482576557550157e-05,
+      "loss": 0.349,
+      "step": 685
+    },
+    {
+      "epoch": 1.9544159544159543,
+      "grad_norm": 0.18663083025931687,
+      "learning_rate": 1.9429778247096094e-05,
+      "loss": 0.3638,
+      "step": 686
+    },
+    {
+      "epoch": 1.9572649572649574,
+      "grad_norm": 0.20785309716673614,
+      "learning_rate": 1.937697993664203e-05,
+      "loss": 0.3529,
+      "step": 687
+    },
+    {
+      "epoch": 1.96011396011396,
+      "grad_norm": 0.22463804321134564,
+      "learning_rate": 1.9324181626187963e-05,
+      "loss": 0.3594,
+      "step": 688
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.1906326731349908,
+      "learning_rate": 1.92713833157339e-05,
+      "loss": 0.3502,
+      "step": 689
+    },
+    {
+      "epoch": 1.965811965811966,
+      "grad_norm": 0.2066700222331773,
+      "learning_rate": 1.9218585005279832e-05,
+      "loss": 0.3672,
+      "step": 690
+    },
+    {
+      "epoch": 1.9686609686609686,
+      "grad_norm": 0.1998642686060524,
+      "learning_rate": 1.9165786694825765e-05,
+      "loss": 0.364,
+      "step": 691
+    },
+    {
+      "epoch": 1.9715099715099715,
+      "grad_norm": 0.1887608555452163,
+      "learning_rate": 1.9112988384371702e-05,
+      "loss": 0.3509,
+      "step": 692
+    },
+    {
+      "epoch": 1.9743589743589745,
+      "grad_norm": 0.196768422003315,
+      "learning_rate": 1.9060190073917635e-05,
+      "loss": 0.342,
+      "step": 693
+    },
+    {
+      "epoch": 1.9772079772079771,
+      "grad_norm": 0.19163494415125434,
+      "learning_rate": 1.900739176346357e-05,
+      "loss": 0.3461,
+      "step": 694
+    },
+    {
+      "epoch": 1.98005698005698,
+      "grad_norm": 0.22754534461400272,
+      "learning_rate": 1.8954593453009508e-05,
+      "loss": 0.3764,
+      "step": 695
+    },
+    {
+      "epoch": 1.982905982905983,
+      "grad_norm": 0.23834935791341716,
+      "learning_rate": 1.890179514255544e-05,
+      "loss": 0.3484,
+      "step": 696
+    },
+    {
+      "epoch": 1.9857549857549857,
+      "grad_norm": 0.21264794974633236,
+      "learning_rate": 1.8848996832101374e-05,
+      "loss": 0.368,
+      "step": 697
+    },
+    {
+      "epoch": 1.9886039886039886,
+      "grad_norm": 0.21453842577290422,
+      "learning_rate": 1.8796198521647307e-05,
+      "loss": 0.3593,
+      "step": 698
+    },
+    {
+      "epoch": 1.9914529914529915,
+      "grad_norm": 0.20799231966171178,
+      "learning_rate": 1.8743400211193243e-05,
+      "loss": 0.3523,
+      "step": 699
+    },
+    {
+      "epoch": 1.9943019943019942,
+      "grad_norm": 0.2080913590301314,
+      "learning_rate": 1.8690601900739176e-05,
+      "loss": 0.3768,
+      "step": 700
+    },
+    {
+      "epoch": 1.9971509971509973,
+      "grad_norm": 0.20083753257614662,
+      "learning_rate": 1.863780359028511e-05,
+      "loss": 0.373,
+      "step": 701
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.25026072457519927,
+      "learning_rate": 1.8585005279831045e-05,
+      "loss": 0.3476,
+      "step": 702
+    },
+    {
+      "epoch": 2.0028490028490027,
+      "grad_norm": 0.31749430934031786,
+      "learning_rate": 1.8532206969376982e-05,
+      "loss": 0.2939,
+      "step": 703
+    },
+    {
+      "epoch": 2.005698005698006,
+      "grad_norm": 0.24903718426554183,
+      "learning_rate": 1.8479408658922915e-05,
+      "loss": 0.2786,
+      "step": 704
+    },
+    {
+      "epoch": 2.0085470085470085,
+      "grad_norm": 0.32095802283737634,
+      "learning_rate": 1.842661034846885e-05,
+      "loss": 0.2838,
+      "step": 705
+    },
+    {
+      "epoch": 2.011396011396011,
+      "grad_norm": 0.29328887995626335,
+      "learning_rate": 1.8373812038014784e-05,
+      "loss": 0.2801,
+      "step": 706
+    },
+    {
+      "epoch": 2.0142450142450143,
+      "grad_norm": 0.20925973424050445,
+      "learning_rate": 1.8321013727560717e-05,
+      "loss": 0.2631,
+      "step": 707
+    },
+    {
+      "epoch": 2.017094017094017,
+      "grad_norm": 0.28435426480239495,
+      "learning_rate": 1.8268215417106653e-05,
+      "loss": 0.2843,
+      "step": 708
+    },
+    {
+      "epoch": 2.0199430199430197,
+      "grad_norm": 0.2500899473941078,
+      "learning_rate": 1.8215417106652586e-05,
+      "loss": 0.2874,
+      "step": 709
+    },
+    {
+      "epoch": 2.022792022792023,
+      "grad_norm": 0.2642215470553276,
+      "learning_rate": 1.8162618796198523e-05,
+      "loss": 0.2812,
+      "step": 710
+    },
+    {
+      "epoch": 2.0256410256410255,
+      "grad_norm": 0.2787623192910837,
+      "learning_rate": 1.810982048574446e-05,
+      "loss": 0.2852,
+      "step": 711
+    },
+    {
+      "epoch": 2.0284900284900287,
+      "grad_norm": 0.2532708996249064,
+      "learning_rate": 1.8057022175290392e-05,
+      "loss": 0.2767,
+      "step": 712
+    },
+    {
+      "epoch": 2.0313390313390314,
+      "grad_norm": 0.24900158666379366,
+      "learning_rate": 1.8004223864836325e-05,
+      "loss": 0.2737,
+      "step": 713
+    },
+    {
+      "epoch": 2.034188034188034,
+      "grad_norm": 0.2635713749224828,
+      "learning_rate": 1.795142555438226e-05,
+      "loss": 0.2705,
+      "step": 714
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 0.20086792181553093,
+      "learning_rate": 1.7898627243928195e-05,
+      "loss": 0.2798,
+      "step": 715
+    },
+    {
+      "epoch": 2.03988603988604,
+      "grad_norm": 0.2432885568665745,
+      "learning_rate": 1.7845828933474128e-05,
+      "loss": 0.281,
+      "step": 716
+    },
+    {
+      "epoch": 2.0427350427350426,
+      "grad_norm": 0.21507703672726577,
+      "learning_rate": 1.7793030623020064e-05,
+      "loss": 0.2762,
+      "step": 717
+    },
+    {
+      "epoch": 2.0455840455840457,
+      "grad_norm": 0.21933858359248162,
+      "learning_rate": 1.7740232312566e-05,
+      "loss": 0.2729,
+      "step": 718
+    },
+    {
+      "epoch": 2.0484330484330484,
+      "grad_norm": 0.22017588537800314,
+      "learning_rate": 1.7687434002111933e-05,
+      "loss": 0.2855,
+      "step": 719
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 0.20337914754232117,
+      "learning_rate": 1.7634635691657866e-05,
+      "loss": 0.2715,
+      "step": 720
+    },
+    {
+      "epoch": 2.0541310541310542,
+      "grad_norm": 0.19719485220504646,
+      "learning_rate": 1.7581837381203803e-05,
+      "loss": 0.2611,
+      "step": 721
+    },
+    {
+      "epoch": 2.056980056980057,
+      "grad_norm": 0.2031620606156548,
+      "learning_rate": 1.7529039070749736e-05,
+      "loss": 0.2853,
+      "step": 722
+    },
+    {
+      "epoch": 2.0598290598290596,
+      "grad_norm": 0.22411987580889667,
+      "learning_rate": 1.747624076029567e-05,
+      "loss": 0.2807,
+      "step": 723
+    },
+    {
+      "epoch": 2.0626780626780628,
+      "grad_norm": 0.1997733419374099,
+      "learning_rate": 1.7423442449841605e-05,
+      "loss": 0.2749,
+      "step": 724
+    },
+    {
+      "epoch": 2.0655270655270654,
+      "grad_norm": 0.20442777603565418,
+      "learning_rate": 1.737064413938754e-05,
+      "loss": 0.2709,
+      "step": 725
+    },
+    {
+      "epoch": 2.0683760683760686,
+      "grad_norm": 0.2053651992461902,
+      "learning_rate": 1.7317845828933475e-05,
+      "loss": 0.2739,
+      "step": 726
+    },
+    {
+      "epoch": 2.0712250712250713,
+      "grad_norm": 0.2041941282028662,
+      "learning_rate": 1.726504751847941e-05,
+      "loss": 0.2787,
+      "step": 727
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.2289119309925019,
+      "learning_rate": 1.7212249208025344e-05,
+      "loss": 0.276,
+      "step": 728
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.20075289027919085,
+      "learning_rate": 1.7159450897571277e-05,
+      "loss": 0.2754,
+      "step": 729
+    },
+    {
+      "epoch": 2.07977207977208,
+      "grad_norm": 0.1910485513723653,
+      "learning_rate": 1.7106652587117213e-05,
+      "loss": 0.274,
+      "step": 730
+    },
+    {
+      "epoch": 2.0826210826210825,
+      "grad_norm": 0.2075379358650543,
+      "learning_rate": 1.7053854276663146e-05,
+      "loss": 0.2724,
+      "step": 731
+    },
+    {
+      "epoch": 2.0854700854700856,
+      "grad_norm": 0.21528112160170912,
+      "learning_rate": 1.7001055966209083e-05,
+      "loss": 0.2726,
+      "step": 732
+    },
+    {
+      "epoch": 2.0883190883190883,
+      "grad_norm": 0.2110995481791853,
+      "learning_rate": 1.694825765575502e-05,
+      "loss": 0.2772,
+      "step": 733
+    },
+    {
+      "epoch": 2.091168091168091,
+      "grad_norm": 0.1829351870654865,
+      "learning_rate": 1.6895459345300952e-05,
+      "loss": 0.2704,
+      "step": 734
+    },
+    {
+      "epoch": 2.094017094017094,
+      "grad_norm": 0.19540194440796962,
+      "learning_rate": 1.6842661034846885e-05,
+      "loss": 0.2658,
+      "step": 735
+    },
+    {
+      "epoch": 2.096866096866097,
+      "grad_norm": 0.21657896164501325,
+      "learning_rate": 1.678986272439282e-05,
+      "loss": 0.2746,
+      "step": 736
+    },
+    {
+      "epoch": 2.0997150997150995,
+      "grad_norm": 0.19563365784205655,
+      "learning_rate": 1.6737064413938755e-05,
+      "loss": 0.2652,
+      "step": 737
+    },
+    {
+      "epoch": 2.1025641025641026,
+      "grad_norm": 0.19212356110155093,
+      "learning_rate": 1.6684266103484688e-05,
+      "loss": 0.263,
+      "step": 738
+    },
+    {
+      "epoch": 2.1054131054131053,
+      "grad_norm": 0.20460727383155605,
+      "learning_rate": 1.6631467793030624e-05,
+      "loss": 0.2742,
+      "step": 739
+    },
+    {
+      "epoch": 2.1082621082621085,
+      "grad_norm": 0.21427650674256282,
+      "learning_rate": 1.657866948257656e-05,
+      "loss": 0.2825,
+      "step": 740
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.20672190281174416,
+      "learning_rate": 1.6525871172122493e-05,
+      "loss": 0.2839,
+      "step": 741
+    },
+    {
+      "epoch": 2.113960113960114,
+      "grad_norm": 0.20211237414090671,
+      "learning_rate": 1.647307286166843e-05,
+      "loss": 0.2826,
+      "step": 742
+    },
+    {
+      "epoch": 2.116809116809117,
+      "grad_norm": 0.19195099076895822,
+      "learning_rate": 1.6420274551214363e-05,
+      "loss": 0.2759,
+      "step": 743
+    },
+    {
+      "epoch": 2.1196581196581197,
+      "grad_norm": 0.20560314548089362,
+      "learning_rate": 1.6367476240760296e-05,
+      "loss": 0.2681,
+      "step": 744
+    },
+    {
+      "epoch": 2.1225071225071224,
+      "grad_norm": 0.21253013838923676,
+      "learning_rate": 1.631467793030623e-05,
+      "loss": 0.2691,
+      "step": 745
+    },
+    {
+      "epoch": 2.1253561253561255,
+      "grad_norm": 0.19213504270638337,
+      "learning_rate": 1.6261879619852165e-05,
+      "loss": 0.2709,
+      "step": 746
+    },
+    {
+      "epoch": 2.128205128205128,
+      "grad_norm": 0.19810360169754043,
+      "learning_rate": 1.6209081309398098e-05,
+      "loss": 0.2681,
+      "step": 747
+    },
+    {
+      "epoch": 2.131054131054131,
+      "grad_norm": 0.8925954469518941,
+      "learning_rate": 1.6156282998944034e-05,
+      "loss": 0.3358,
+      "step": 748
+    },
+    {
+      "epoch": 2.133903133903134,
+      "grad_norm": 0.21118384990812827,
+      "learning_rate": 1.610348468848997e-05,
+      "loss": 0.2834,
+      "step": 749
+    },
+    {
+      "epoch": 2.1367521367521367,
+      "grad_norm": 0.22061422429768562,
+      "learning_rate": 1.6050686378035904e-05,
+      "loss": 0.2782,
+      "step": 750
+    },
+    {
+      "epoch": 2.1396011396011394,
+      "grad_norm": 0.2124786481646926,
+      "learning_rate": 1.5997888067581837e-05,
+      "loss": 0.2998,
+      "step": 751
+    },
+    {
+      "epoch": 2.1424501424501425,
+      "grad_norm": 0.20503289997556293,
+      "learning_rate": 1.5945089757127773e-05,
+      "loss": 0.2807,
+      "step": 752
+    },
+    {
+      "epoch": 2.1452991452991452,
+      "grad_norm": 0.22522249962157764,
+      "learning_rate": 1.5892291446673706e-05,
+      "loss": 0.2806,
+      "step": 753
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 0.22270532746805854,
+      "learning_rate": 1.583949313621964e-05,
+      "loss": 0.273,
+      "step": 754
+    },
+    {
+      "epoch": 2.150997150997151,
+      "grad_norm": 0.19772734044545245,
+      "learning_rate": 1.5786694825765576e-05,
+      "loss": 0.272,
+      "step": 755
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.1967943644277297,
+      "learning_rate": 1.5733896515311512e-05,
+      "loss": 0.2753,
+      "step": 756
+    },
+    {
+      "epoch": 2.156695156695157,
+      "grad_norm": 0.19154676700453233,
+      "learning_rate": 1.5681098204857445e-05,
+      "loss": 0.2709,
+      "step": 757
+    },
+    {
+      "epoch": 2.1595441595441596,
+      "grad_norm": 0.2221016196227244,
+      "learning_rate": 1.562829989440338e-05,
+      "loss": 0.2823,
+      "step": 758
+    },
+    {
+      "epoch": 2.1623931623931623,
+      "grad_norm": 0.1937702666618624,
+      "learning_rate": 1.5575501583949314e-05,
+      "loss": 0.2645,
+      "step": 759
+    },
+    {
+      "epoch": 2.1652421652421654,
+      "grad_norm": 0.1969269538576539,
+      "learning_rate": 1.5522703273495247e-05,
+      "loss": 0.2761,
+      "step": 760
+    },
+    {
+      "epoch": 2.168091168091168,
+      "grad_norm": 0.20395502697309204,
+      "learning_rate": 1.5469904963041184e-05,
+      "loss": 0.2792,
+      "step": 761
+    },
+    {
+      "epoch": 2.1709401709401708,
+      "grad_norm": 0.20751492020213136,
+      "learning_rate": 1.5417106652587117e-05,
+      "loss": 0.2623,
+      "step": 762
+    },
+    {
+      "epoch": 2.173789173789174,
+      "grad_norm": 0.19379484649828319,
+      "learning_rate": 1.5364308342133053e-05,
+      "loss": 0.2606,
+      "step": 763
+    },
+    {
+      "epoch": 2.1766381766381766,
+      "grad_norm": 0.18718035792100948,
+      "learning_rate": 1.531151003167899e-05,
+      "loss": 0.2751,
+      "step": 764
+    },
+    {
+      "epoch": 2.1794871794871793,
+      "grad_norm": 0.19066401480212206,
+      "learning_rate": 1.5258711721224923e-05,
+      "loss": 0.2732,
+      "step": 765
+    },
+    {
+      "epoch": 2.1823361823361824,
+      "grad_norm": 0.20925353974218774,
+      "learning_rate": 1.5205913410770856e-05,
+      "loss": 0.2881,
+      "step": 766
+    },
+    {
+      "epoch": 2.185185185185185,
+      "grad_norm": 0.20547467764539895,
+      "learning_rate": 1.5153115100316789e-05,
+      "loss": 0.2638,
+      "step": 767
+    },
+    {
+      "epoch": 2.1880341880341883,
+      "grad_norm": 0.19527197604782076,
+      "learning_rate": 1.5100316789862725e-05,
+      "loss": 0.2783,
+      "step": 768
+    },
+    {
+      "epoch": 2.190883190883191,
+      "grad_norm": 0.1942190592627724,
+      "learning_rate": 1.504751847940866e-05,
+      "loss": 0.2709,
+      "step": 769
+    },
+    {
+      "epoch": 2.1937321937321936,
+      "grad_norm": 0.2103546914669498,
+      "learning_rate": 1.4994720168954593e-05,
+      "loss": 0.2703,
+      "step": 770
+    },
+    {
+      "epoch": 2.1965811965811968,
+      "grad_norm": 0.20158953401192725,
+      "learning_rate": 1.4941921858500529e-05,
+      "loss": 0.2786,
+      "step": 771
+    },
+    {
+      "epoch": 2.1994301994301995,
+      "grad_norm": 0.18874398829424063,
+      "learning_rate": 1.4889123548046464e-05,
+      "loss": 0.2625,
+      "step": 772
+    },
+    {
+      "epoch": 2.202279202279202,
+      "grad_norm": 0.1919940823226646,
+      "learning_rate": 1.4836325237592397e-05,
+      "loss": 0.2819,
+      "step": 773
+    },
+    {
+      "epoch": 2.2051282051282053,
+      "grad_norm": 0.20270992368788238,
+      "learning_rate": 1.4783526927138333e-05,
+      "loss": 0.2746,
+      "step": 774
+    },
+    {
+      "epoch": 2.207977207977208,
+      "grad_norm": 0.19972156431692997,
+      "learning_rate": 1.4730728616684266e-05,
+      "loss": 0.277,
+      "step": 775
+    },
+    {
+      "epoch": 2.2108262108262107,
+      "grad_norm": 0.2060428169408646,
+      "learning_rate": 1.46779303062302e-05,
+      "loss": 0.2866,
+      "step": 776
+    },
+    {
+      "epoch": 2.213675213675214,
+      "grad_norm": 0.2180940293343127,
+      "learning_rate": 1.4625131995776137e-05,
+      "loss": 0.2664,
+      "step": 777
+    },
+    {
+      "epoch": 2.2165242165242165,
+      "grad_norm": 0.19764083611553848,
+      "learning_rate": 1.457233368532207e-05,
+      "loss": 0.2698,
+      "step": 778
+    },
+    {
+      "epoch": 2.219373219373219,
+      "grad_norm": 0.18295950495897126,
+      "learning_rate": 1.4519535374868005e-05,
+      "loss": 0.266,
+      "step": 779
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.20138778009514838,
+      "learning_rate": 1.4466737064413941e-05,
+      "loss": 0.2732,
+      "step": 780
+    },
+    {
+      "epoch": 2.225071225071225,
+      "grad_norm": 0.21789436891588496,
+      "learning_rate": 1.4413938753959874e-05,
+      "loss": 0.2842,
+      "step": 781
+    },
+    {
+      "epoch": 2.2279202279202277,
+      "grad_norm": 0.19845024094048266,
+      "learning_rate": 1.4361140443505807e-05,
+      "loss": 0.2617,
+      "step": 782
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.20136450371041287,
+      "learning_rate": 1.4308342133051744e-05,
+      "loss": 0.2673,
+      "step": 783
+    },
+    {
+      "epoch": 2.2336182336182335,
+      "grad_norm": 0.20064270708774173,
+      "learning_rate": 1.4255543822597678e-05,
+      "loss": 0.2635,
+      "step": 784
+    },
+    {
+      "epoch": 2.2364672364672367,
+      "grad_norm": 0.20230154595496835,
+      "learning_rate": 1.4202745512143611e-05,
+      "loss": 0.2929,
+      "step": 785
+    },
+    {
+      "epoch": 2.2393162393162394,
+      "grad_norm": 0.201820653024005,
+      "learning_rate": 1.4149947201689548e-05,
+      "loss": 0.2834,
+      "step": 786
+    },
+    {
+      "epoch": 2.242165242165242,
+      "grad_norm": 0.19121793530935455,
+      "learning_rate": 1.409714889123548e-05,
+      "loss": 0.2625,
+      "step": 787
+    },
+    {
+      "epoch": 2.245014245014245,
+      "grad_norm": 0.19330950900844268,
+      "learning_rate": 1.4044350580781415e-05,
+      "loss": 0.2772,
+      "step": 788
+    },
+    {
+      "epoch": 2.247863247863248,
+      "grad_norm": 0.19306556715969658,
+      "learning_rate": 1.3991552270327348e-05,
+      "loss": 0.2627,
+      "step": 789
+    },
+    {
+      "epoch": 2.2507122507122506,
+      "grad_norm": 0.19901233397623183,
+      "learning_rate": 1.3938753959873285e-05,
+      "loss": 0.2696,
+      "step": 790
+    },
+    {
+      "epoch": 2.2535612535612537,
+      "grad_norm": 0.19925310664104037,
+      "learning_rate": 1.388595564941922e-05,
+      "loss": 0.2584,
+      "step": 791
+    },
+    {
+      "epoch": 2.2564102564102564,
+      "grad_norm": 0.2034999029639149,
+      "learning_rate": 1.3833157338965153e-05,
+      "loss": 0.2966,
+      "step": 792
+    },
+    {
+      "epoch": 2.259259259259259,
+      "grad_norm": 0.19015133424936778,
+      "learning_rate": 1.3780359028511089e-05,
+      "loss": 0.2865,
+      "step": 793
+    },
+    {
+      "epoch": 2.262108262108262,
+      "grad_norm": 0.22328712353887642,
+      "learning_rate": 1.3727560718057022e-05,
+      "loss": 0.2772,
+      "step": 794
+    },
+    {
+      "epoch": 2.264957264957265,
+      "grad_norm": 0.19798853546327766,
+      "learning_rate": 1.3674762407602957e-05,
+      "loss": 0.2632,
+      "step": 795
+    },
+    {
+      "epoch": 2.267806267806268,
+      "grad_norm": 0.18631225104292679,
+      "learning_rate": 1.3621964097148893e-05,
+      "loss": 0.2775,
+      "step": 796
+    },
+    {
+      "epoch": 2.2706552706552707,
+      "grad_norm": 0.1909878525568524,
+      "learning_rate": 1.3569165786694826e-05,
+      "loss": 0.2848,
+      "step": 797
+    },
+    {
+      "epoch": 2.2735042735042734,
+      "grad_norm": 0.2068674181031461,
+      "learning_rate": 1.351636747624076e-05,
+      "loss": 0.2754,
+      "step": 798
+    },
+    {
+      "epoch": 2.2763532763532766,
+      "grad_norm": 0.20247748116558922,
+      "learning_rate": 1.3463569165786697e-05,
+      "loss": 0.2681,
+      "step": 799
+    },
+    {
+      "epoch": 2.2792022792022792,
+      "grad_norm": 0.19577250906503782,
+      "learning_rate": 1.341077085533263e-05,
+      "loss": 0.2698,
+      "step": 800
+    },
+    {
+      "epoch": 2.282051282051282,
+      "grad_norm": 0.192461914460849,
+      "learning_rate": 1.3357972544878563e-05,
+      "loss": 0.2624,
+      "step": 801
+    },
+    {
+      "epoch": 2.284900284900285,
+      "grad_norm": 0.21973379890078096,
+      "learning_rate": 1.33051742344245e-05,
+      "loss": 0.2754,
+      "step": 802
+    },
+    {
+      "epoch": 2.2877492877492878,
+      "grad_norm": 0.19582967491065492,
+      "learning_rate": 1.3252375923970434e-05,
+      "loss": 0.277,
+      "step": 803
+    },
+    {
+      "epoch": 2.2905982905982905,
+      "grad_norm": 0.19580304220995717,
+      "learning_rate": 1.3199577613516367e-05,
+      "loss": 0.2716,
+      "step": 804
+    },
+    {
+      "epoch": 2.2934472934472936,
+      "grad_norm": 0.2588055907574961,
+      "learning_rate": 1.3146779303062304e-05,
+      "loss": 0.2772,
+      "step": 805
+    },
+    {
+      "epoch": 2.2962962962962963,
+      "grad_norm": 0.1985796106477907,
+      "learning_rate": 1.3093980992608237e-05,
+      "loss": 0.2725,
+      "step": 806
+    },
+    {
+      "epoch": 2.299145299145299,
+      "grad_norm": 0.21599405795203994,
+      "learning_rate": 1.3041182682154171e-05,
+      "loss": 0.2924,
+      "step": 807
+    },
+    {
+      "epoch": 2.301994301994302,
+      "grad_norm": 0.20158022481963558,
+      "learning_rate": 1.2988384371700108e-05,
+      "loss": 0.2885,
+      "step": 808
+    },
+    {
+      "epoch": 2.304843304843305,
+      "grad_norm": 0.18745667148509224,
+      "learning_rate": 1.293558606124604e-05,
+      "loss": 0.2943,
+      "step": 809
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.2065772682516858,
+      "learning_rate": 1.2882787750791975e-05,
+      "loss": 0.2778,
+      "step": 810
+    },
+    {
+      "epoch": 2.3105413105413106,
+      "grad_norm": 0.2170049193233547,
+      "learning_rate": 1.2829989440337912e-05,
+      "loss": 0.2842,
+      "step": 811
+    },
+    {
+      "epoch": 2.3133903133903133,
+      "grad_norm": 0.20199308409143069,
+      "learning_rate": 1.2777191129883845e-05,
+      "loss": 0.2717,
+      "step": 812
+    },
+    {
+      "epoch": 2.316239316239316,
+      "grad_norm": 0.20699440291292984,
+      "learning_rate": 1.2724392819429778e-05,
+      "loss": 0.2925,
+      "step": 813
+    },
+    {
+      "epoch": 2.319088319088319,
+      "grad_norm": 0.19508346223108827,
+      "learning_rate": 1.2671594508975712e-05,
+      "loss": 0.277,
+      "step": 814
+    },
+    {
+      "epoch": 2.321937321937322,
+      "grad_norm": 0.19079264766629775,
+      "learning_rate": 1.2618796198521649e-05,
+      "loss": 0.2778,
+      "step": 815
+    },
+    {
+      "epoch": 2.324786324786325,
+      "grad_norm": 0.18964264325936672,
+      "learning_rate": 1.2565997888067582e-05,
+      "loss": 0.2782,
+      "step": 816
+    },
+    {
+      "epoch": 2.3276353276353277,
+      "grad_norm": 0.20280215258566556,
+      "learning_rate": 1.2513199577613517e-05,
+      "loss": 0.2893,
+      "step": 817
+    },
+    {
+      "epoch": 2.3304843304843303,
+      "grad_norm": 0.2156167038301731,
+      "learning_rate": 1.2460401267159451e-05,
+      "loss": 0.2854,
+      "step": 818
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.1964934145520702,
+      "learning_rate": 1.2407602956705386e-05,
+      "loss": 0.2789,
+      "step": 819
+    },
+    {
+      "epoch": 2.336182336182336,
+      "grad_norm": 0.20506845724808467,
+      "learning_rate": 1.235480464625132e-05,
+      "loss": 0.2837,
+      "step": 820
+    },
+    {
+      "epoch": 2.339031339031339,
+      "grad_norm": 0.20608756180445,
+      "learning_rate": 1.2302006335797255e-05,
+      "loss": 0.2867,
+      "step": 821
+    },
+    {
+      "epoch": 2.341880341880342,
+      "grad_norm": 0.22728463316725367,
+      "learning_rate": 1.224920802534319e-05,
+      "loss": 0.2879,
+      "step": 822
+    },
+    {
+      "epoch": 2.3447293447293447,
+      "grad_norm": 0.21134582426808612,
+      "learning_rate": 1.2196409714889125e-05,
+      "loss": 0.2706,
+      "step": 823
+    },
+    {
+      "epoch": 2.347578347578348,
+      "grad_norm": 0.2117202491888567,
+      "learning_rate": 1.2143611404435058e-05,
+      "loss": 0.2901,
+      "step": 824
+    },
+    {
+      "epoch": 2.3504273504273505,
+      "grad_norm": 0.20417464138588973,
+      "learning_rate": 1.2090813093980994e-05,
+      "loss": 0.2838,
+      "step": 825
+    },
+    {
+      "epoch": 2.353276353276353,
+      "grad_norm": 0.1996078258358169,
+      "learning_rate": 1.2038014783526929e-05,
+      "loss": 0.2669,
+      "step": 826
+    },
+    {
+      "epoch": 2.3561253561253563,
+      "grad_norm": 0.22247317080848442,
+      "learning_rate": 1.1985216473072862e-05,
+      "loss": 0.2855,
+      "step": 827
+    },
+    {
+      "epoch": 2.358974358974359,
+      "grad_norm": 0.20677945710130288,
+      "learning_rate": 1.1932418162618796e-05,
+      "loss": 0.2617,
+      "step": 828
+    },
+    {
+      "epoch": 2.3618233618233617,
+      "grad_norm": 0.22160179441363054,
+      "learning_rate": 1.1879619852164731e-05,
+      "loss": 0.2756,
+      "step": 829
+    },
+    {
+      "epoch": 2.364672364672365,
+      "grad_norm": 0.22587987413436983,
+      "learning_rate": 1.1826821541710666e-05,
+      "loss": 0.2817,
+      "step": 830
+    },
+    {
+      "epoch": 2.3675213675213675,
+      "grad_norm": 0.20154794971028572,
+      "learning_rate": 1.17740232312566e-05,
+      "loss": 0.2651,
+      "step": 831
+    },
+    {
+      "epoch": 2.3703703703703702,
+      "grad_norm": 0.20403481294874842,
+      "learning_rate": 1.1721224920802534e-05,
+      "loss": 0.2775,
+      "step": 832
+    },
+    {
+      "epoch": 2.3732193732193734,
+      "grad_norm": 0.22073919271992984,
+      "learning_rate": 1.166842661034847e-05,
+      "loss": 0.2778,
+      "step": 833
+    },
+    {
+      "epoch": 2.376068376068376,
+      "grad_norm": 0.22085826451951224,
+      "learning_rate": 1.1615628299894405e-05,
+      "loss": 0.2716,
+      "step": 834
+    },
+    {
+      "epoch": 2.3789173789173788,
+      "grad_norm": 0.20286442068216853,
+      "learning_rate": 1.1562829989440338e-05,
+      "loss": 0.2725,
+      "step": 835
+    },
+    {
+      "epoch": 2.381766381766382,
+      "grad_norm": 0.20066512198655337,
+      "learning_rate": 1.1510031678986272e-05,
+      "loss": 0.2791,
+      "step": 836
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.2138581117365153,
+      "learning_rate": 1.1457233368532209e-05,
+      "loss": 0.2748,
+      "step": 837
+    },
+    {
+      "epoch": 2.3874643874643873,
+      "grad_norm": 0.21527795194888655,
+      "learning_rate": 1.1404435058078142e-05,
+      "loss": 0.2742,
+      "step": 838
+    },
+    {
+      "epoch": 2.3903133903133904,
+      "grad_norm": 0.20801036949744942,
+      "learning_rate": 1.1351636747624076e-05,
+      "loss": 0.2797,
+      "step": 839
+    },
+    {
+      "epoch": 2.393162393162393,
+      "grad_norm": 0.20792622413478573,
+      "learning_rate": 1.1298838437170011e-05,
+      "loss": 0.282,
+      "step": 840
+    },
+    {
+      "epoch": 2.396011396011396,
+      "grad_norm": 0.22033246310576216,
+      "learning_rate": 1.1246040126715946e-05,
+      "loss": 0.279,
+      "step": 841
+    },
+    {
+      "epoch": 2.398860398860399,
+      "grad_norm": 0.1967422885160002,
+      "learning_rate": 1.119324181626188e-05,
+      "loss": 0.2811,
+      "step": 842
+    },
+    {
+      "epoch": 2.4017094017094016,
+      "grad_norm": 0.1814858766804614,
+      "learning_rate": 1.1140443505807813e-05,
+      "loss": 0.2608,
+      "step": 843
+    },
+    {
+      "epoch": 2.4045584045584047,
+      "grad_norm": 0.19499452238873044,
+      "learning_rate": 1.108764519535375e-05,
+      "loss": 0.2678,
+      "step": 844
+    },
+    {
+      "epoch": 2.4074074074074074,
+      "grad_norm": 0.2102124248647648,
+      "learning_rate": 1.1034846884899685e-05,
+      "loss": 0.2818,
+      "step": 845
+    },
+    {
+      "epoch": 2.41025641025641,
+      "grad_norm": 0.19814574523643397,
+      "learning_rate": 1.0982048574445618e-05,
+      "loss": 0.2653,
+      "step": 846
+    },
+    {
+      "epoch": 2.4131054131054133,
+      "grad_norm": 0.18054162144741087,
+      "learning_rate": 1.0929250263991552e-05,
+      "loss": 0.2695,
+      "step": 847
+    },
+    {
+      "epoch": 2.415954415954416,
+      "grad_norm": 0.1808318357229325,
+      "learning_rate": 1.0876451953537489e-05,
+      "loss": 0.2801,
+      "step": 848
+    },
+    {
+      "epoch": 2.4188034188034186,
+      "grad_norm": 0.19987403185338987,
+      "learning_rate": 1.0823653643083422e-05,
+      "loss": 0.2721,
+      "step": 849
+    },
+    {
+      "epoch": 2.421652421652422,
+      "grad_norm": 0.2248142211042877,
+      "learning_rate": 1.0770855332629356e-05,
+      "loss": 0.2843,
+      "step": 850
+    },
+    {
+      "epoch": 2.4245014245014245,
+      "grad_norm": 0.1964174663784782,
+      "learning_rate": 1.0718057022175291e-05,
+      "loss": 0.2834,
+      "step": 851
+    },
+    {
+      "epoch": 2.427350427350427,
+      "grad_norm": 0.18315937428538173,
+      "learning_rate": 1.0665258711721226e-05,
+      "loss": 0.267,
+      "step": 852
+    },
+    {
+      "epoch": 2.4301994301994303,
+      "grad_norm": 0.1926077433772735,
+      "learning_rate": 1.061246040126716e-05,
+      "loss": 0.2748,
+      "step": 853
+    },
+    {
+      "epoch": 2.433048433048433,
+      "grad_norm": 0.20408583222643678,
+      "learning_rate": 1.0559662090813093e-05,
+      "loss": 0.2776,
+      "step": 854
+    },
+    {
+      "epoch": 2.435897435897436,
+      "grad_norm": 0.1973007081934127,
+      "learning_rate": 1.0506863780359028e-05,
+      "loss": 0.2748,
+      "step": 855
+    },
+    {
+      "epoch": 2.438746438746439,
+      "grad_norm": 0.18467705527546302,
+      "learning_rate": 1.0454065469904964e-05,
+      "loss": 0.2767,
+      "step": 856
+    },
+    {
+      "epoch": 2.4415954415954415,
+      "grad_norm": 0.18392804629969942,
+      "learning_rate": 1.0401267159450897e-05,
+      "loss": 0.2803,
+      "step": 857
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.19726145099036824,
+      "learning_rate": 1.0348468848996832e-05,
+      "loss": 0.2763,
+      "step": 858
+    },
+    {
+      "epoch": 2.4472934472934473,
+      "grad_norm": 0.19595567470607117,
+      "learning_rate": 1.0295670538542767e-05,
+      "loss": 0.288,
+      "step": 859
+    },
+    {
+      "epoch": 2.45014245014245,
+      "grad_norm": 0.19049627669313532,
+      "learning_rate": 1.0242872228088702e-05,
+      "loss": 0.2804,
+      "step": 860
+    },
+    {
+      "epoch": 2.452991452991453,
+      "grad_norm": 0.19465038042198368,
+      "learning_rate": 1.0190073917634636e-05,
+      "loss": 0.2853,
+      "step": 861
+    },
+    {
+      "epoch": 2.455840455840456,
+      "grad_norm": 0.18907018318647903,
+      "learning_rate": 1.0137275607180571e-05,
+      "loss": 0.262,
+      "step": 862
+    },
+    {
+      "epoch": 2.4586894586894585,
+      "grad_norm": 0.2117775134978185,
+      "learning_rate": 1.0084477296726506e-05,
+      "loss": 0.2832,
+      "step": 863
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.18369057037488082,
+      "learning_rate": 1.003167898627244e-05,
+      "loss": 0.2818,
+      "step": 864
+    },
+    {
+      "epoch": 2.4643874643874644,
+      "grad_norm": 0.1938139353246623,
+      "learning_rate": 9.978880675818373e-06,
+      "loss": 0.2687,
+      "step": 865
+    },
+    {
+      "epoch": 2.467236467236467,
+      "grad_norm": 0.21926317040236148,
+      "learning_rate": 9.926082365364308e-06,
+      "loss": 0.2762,
+      "step": 866
+    },
+    {
+      "epoch": 2.47008547008547,
+      "grad_norm": 0.20036724393686023,
+      "learning_rate": 9.873284054910244e-06,
+      "loss": 0.2788,
+      "step": 867
+    },
+    {
+      "epoch": 2.472934472934473,
+      "grad_norm": 0.195455632630333,
+      "learning_rate": 9.820485744456177e-06,
+      "loss": 0.2712,
+      "step": 868
+    },
+    {
+      "epoch": 2.4757834757834756,
+      "grad_norm": 0.19688632965075772,
+      "learning_rate": 9.767687434002112e-06,
+      "loss": 0.2928,
+      "step": 869
+    },
+    {
+      "epoch": 2.4786324786324787,
+      "grad_norm": 0.196362275701545,
+      "learning_rate": 9.714889123548047e-06,
+      "loss": 0.2856,
+      "step": 870
+    },
+    {
+      "epoch": 2.4814814814814814,
+      "grad_norm": 0.19446432785612933,
+      "learning_rate": 9.662090813093982e-06,
+      "loss": 0.279,
+      "step": 871
+    },
+    {
+      "epoch": 2.484330484330484,
+      "grad_norm": 0.1879688895780496,
+      "learning_rate": 9.609292502639916e-06,
+      "loss": 0.2854,
+      "step": 872
+    },
+    {
+      "epoch": 2.4871794871794872,
+      "grad_norm": 0.19437610153561674,
+      "learning_rate": 9.556494192185851e-06,
+      "loss": 0.2738,
+      "step": 873
+    },
+    {
+      "epoch": 2.49002849002849,
+      "grad_norm": 0.20143365503846905,
+      "learning_rate": 9.503695881731786e-06,
+      "loss": 0.2759,
+      "step": 874
+    },
+    {
+      "epoch": 2.492877492877493,
+      "grad_norm": 0.1935287651459583,
+      "learning_rate": 9.45089757127772e-06,
+      "loss": 0.2835,
+      "step": 875
+    },
+    {
+      "epoch": 2.4957264957264957,
+      "grad_norm": 0.19317522985073163,
+      "learning_rate": 9.398099260823653e-06,
+      "loss": 0.2858,
+      "step": 876
+    },
+    {
+      "epoch": 2.4985754985754984,
+      "grad_norm": 0.2012616552901727,
+      "learning_rate": 9.345300950369588e-06,
+      "loss": 0.2686,
+      "step": 877
+    },
+    {
+      "epoch": 2.5014245014245016,
+      "grad_norm": 0.18496934505004578,
+      "learning_rate": 9.292502639915523e-06,
+      "loss": 0.2649,
+      "step": 878
+    },
+    {
+      "epoch": 2.5042735042735043,
+      "grad_norm": 0.1952709023463477,
+      "learning_rate": 9.239704329461457e-06,
+      "loss": 0.2765,
+      "step": 879
+    },
+    {
+      "epoch": 2.5071225071225074,
+      "grad_norm": 0.1879189333509987,
+      "learning_rate": 9.186906019007392e-06,
+      "loss": 0.2832,
+      "step": 880
+    },
+    {
+      "epoch": 2.50997150997151,
+      "grad_norm": 0.18950447218698951,
+      "learning_rate": 9.134107708553327e-06,
+      "loss": 0.2843,
+      "step": 881
+    },
+    {
+      "epoch": 2.5128205128205128,
+      "grad_norm": 0.18653191496871754,
+      "learning_rate": 9.081309398099261e-06,
+      "loss": 0.277,
+      "step": 882
+    },
+    {
+      "epoch": 2.515669515669516,
+      "grad_norm": 0.17645739186127074,
+      "learning_rate": 9.028511087645196e-06,
+      "loss": 0.2686,
+      "step": 883
+    },
+    {
+      "epoch": 2.5185185185185186,
+      "grad_norm": 0.18764217708119302,
+      "learning_rate": 8.97571277719113e-06,
+      "loss": 0.284,
+      "step": 884
+    },
+    {
+      "epoch": 2.5213675213675213,
+      "grad_norm": 0.19467937140297095,
+      "learning_rate": 8.922914466737064e-06,
+      "loss": 0.2717,
+      "step": 885
+    },
+    {
+      "epoch": 2.5242165242165244,
+      "grad_norm": 0.19792747219286208,
+      "learning_rate": 8.870116156283e-06,
+      "loss": 0.2855,
+      "step": 886
+    },
+    {
+      "epoch": 2.527065527065527,
+      "grad_norm": 0.18656641235510737,
+      "learning_rate": 8.817317845828933e-06,
+      "loss": 0.274,
+      "step": 887
+    },
+    {
+      "epoch": 2.52991452991453,
+      "grad_norm": 0.19826113067880127,
+      "learning_rate": 8.764519535374868e-06,
+      "loss": 0.2973,
+      "step": 888
+    },
+    {
+      "epoch": 2.532763532763533,
+      "grad_norm": 0.19491004642921803,
+      "learning_rate": 8.711721224920803e-06,
+      "loss": 0.2755,
+      "step": 889
+    },
+    {
+      "epoch": 2.5356125356125356,
+      "grad_norm": 0.2012953347472746,
+      "learning_rate": 8.658922914466737e-06,
+      "loss": 0.2836,
+      "step": 890
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.19357544104347177,
+      "learning_rate": 8.606124604012672e-06,
+      "loss": 0.2671,
+      "step": 891
+    },
+    {
+      "epoch": 2.5413105413105415,
+      "grad_norm": 0.18892143464854252,
+      "learning_rate": 8.553326293558607e-06,
+      "loss": 0.2733,
+      "step": 892
+    },
+    {
+      "epoch": 2.544159544159544,
+      "grad_norm": 0.6265508514551001,
+      "learning_rate": 8.500527983104541e-06,
+      "loss": 0.3024,
+      "step": 893
+    },
+    {
+      "epoch": 2.547008547008547,
+      "grad_norm": 0.180843319457107,
+      "learning_rate": 8.447729672650476e-06,
+      "loss": 0.2758,
+      "step": 894
+    },
+    {
+      "epoch": 2.54985754985755,
+      "grad_norm": 0.18824519820336835,
+      "learning_rate": 8.39493136219641e-06,
+      "loss": 0.2777,
+      "step": 895
+    },
+    {
+      "epoch": 2.5527065527065527,
+      "grad_norm": 0.18679719864151892,
+      "learning_rate": 8.342133051742344e-06,
+      "loss": 0.273,
+      "step": 896
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 0.18954916069648725,
+      "learning_rate": 8.28933474128828e-06,
+      "loss": 0.271,
+      "step": 897
+    },
+    {
+      "epoch": 2.5584045584045585,
+      "grad_norm": 0.18840475356548164,
+      "learning_rate": 8.236536430834215e-06,
+      "loss": 0.2634,
+      "step": 898
+    },
+    {
+      "epoch": 2.561253561253561,
+      "grad_norm": 0.19100598305938737,
+      "learning_rate": 8.183738120380148e-06,
+      "loss": 0.281,
+      "step": 899
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 0.1759624339368046,
+      "learning_rate": 8.130939809926083e-06,
+      "loss": 0.2651,
+      "step": 900
+    },
+    {
+      "epoch": 2.566951566951567,
+      "grad_norm": 0.19734582647437332,
+      "learning_rate": 8.078141499472017e-06,
+      "loss": 0.2638,
+      "step": 901
+    },
+    {
+      "epoch": 2.5698005698005697,
+      "grad_norm": 0.19411668207145158,
+      "learning_rate": 8.025343189017952e-06,
+      "loss": 0.2736,
+      "step": 902
+    },
+    {
+      "epoch": 2.5726495726495724,
+      "grad_norm": 0.19879584464405567,
+      "learning_rate": 7.972544878563887e-06,
+      "loss": 0.2769,
+      "step": 903
+    },
+    {
+      "epoch": 2.5754985754985755,
+      "grad_norm": 0.1927813426522461,
+      "learning_rate": 7.91974656810982e-06,
+      "loss": 0.2764,
+      "step": 904
+    },
+    {
+      "epoch": 2.578347578347578,
+      "grad_norm": 0.19856911323488508,
+      "learning_rate": 7.866948257655756e-06,
+      "loss": 0.292,
+      "step": 905
+    },
+    {
+      "epoch": 2.5811965811965814,
+      "grad_norm": 0.17679245135236604,
+      "learning_rate": 7.81414994720169e-06,
+      "loss": 0.2785,
+      "step": 906
+    },
+    {
+      "epoch": 2.584045584045584,
+      "grad_norm": 0.19567448364102336,
+      "learning_rate": 7.761351636747624e-06,
+      "loss": 0.2867,
+      "step": 907
+    },
+    {
+      "epoch": 2.5868945868945867,
+      "grad_norm": 0.20115854063114447,
+      "learning_rate": 7.708553326293558e-06,
+      "loss": 0.2713,
+      "step": 908
+    },
+    {
+      "epoch": 2.58974358974359,
+      "grad_norm": 0.18545206529638672,
+      "learning_rate": 7.655755015839495e-06,
+      "loss": 0.2634,
+      "step": 909
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 0.19495758015732068,
+      "learning_rate": 7.602956705385428e-06,
+      "loss": 0.2723,
+      "step": 910
+    },
+    {
+      "epoch": 2.5954415954415957,
+      "grad_norm": 0.18613675067205288,
+      "learning_rate": 7.5501583949313625e-06,
+      "loss": 0.29,
+      "step": 911
+    },
+    {
+      "epoch": 2.5982905982905984,
+      "grad_norm": 0.19754545334787374,
+      "learning_rate": 7.497360084477296e-06,
+      "loss": 0.2716,
+      "step": 912
+    },
+    {
+      "epoch": 2.601139601139601,
+      "grad_norm": 0.1949333474464277,
+      "learning_rate": 7.444561774023232e-06,
+      "loss": 0.2786,
+      "step": 913
+    },
+    {
+      "epoch": 2.603988603988604,
+      "grad_norm": 0.1866184206464396,
+      "learning_rate": 7.3917634635691666e-06,
+      "loss": 0.2845,
+      "step": 914
+    },
+    {
+      "epoch": 2.606837606837607,
+      "grad_norm": 0.195676886907894,
+      "learning_rate": 7.3389651531151e-06,
+      "loss": 0.2769,
+      "step": 915
+    },
+    {
+      "epoch": 2.6096866096866096,
+      "grad_norm": 0.18429715351385098,
+      "learning_rate": 7.286166842661035e-06,
+      "loss": 0.2756,
+      "step": 916
+    },
+    {
+      "epoch": 2.6125356125356127,
+      "grad_norm": 0.18772973110071034,
+      "learning_rate": 7.233368532206971e-06,
+      "loss": 0.2559,
+      "step": 917
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.1836610271469343,
+      "learning_rate": 7.180570221752904e-06,
+      "loss": 0.2681,
+      "step": 918
+    },
+    {
+      "epoch": 2.618233618233618,
+      "grad_norm": 0.1754155052969751,
+      "learning_rate": 7.127771911298839e-06,
+      "loss": 0.272,
+      "step": 919
+    },
+    {
+      "epoch": 2.6210826210826212,
+      "grad_norm": 0.19652318272004426,
+      "learning_rate": 7.074973600844774e-06,
+      "loss": 0.2686,
+      "step": 920
+    },
+    {
+      "epoch": 2.623931623931624,
+      "grad_norm": 0.17933290326611778,
+      "learning_rate": 7.022175290390708e-06,
+      "loss": 0.2759,
+      "step": 921
+    },
+    {
+      "epoch": 2.6267806267806266,
+      "grad_norm": 0.18155759050319922,
+      "learning_rate": 6.9693769799366424e-06,
+      "loss": 0.2701,
+      "step": 922
+    },
+    {
+      "epoch": 2.6296296296296298,
+      "grad_norm": 0.18892897612236667,
+      "learning_rate": 6.916578669482576e-06,
+      "loss": 0.2769,
+      "step": 923
+    },
+    {
+      "epoch": 2.6324786324786325,
+      "grad_norm": 0.18477414784333626,
+      "learning_rate": 6.863780359028511e-06,
+      "loss": 0.2696,
+      "step": 924
+    },
+    {
+      "epoch": 2.635327635327635,
+      "grad_norm": 0.1797448137751472,
+      "learning_rate": 6.8109820485744465e-06,
+      "loss": 0.2721,
+      "step": 925
+    },
+    {
+      "epoch": 2.6381766381766383,
+      "grad_norm": 0.18868102200145584,
+      "learning_rate": 6.75818373812038e-06,
+      "loss": 0.2654,
+      "step": 926
+    },
+    {
+      "epoch": 2.641025641025641,
+      "grad_norm": 0.18760249991357172,
+      "learning_rate": 6.705385427666315e-06,
+      "loss": 0.2883,
+      "step": 927
+    },
+    {
+      "epoch": 2.6438746438746437,
+      "grad_norm": 0.1792089439057089,
+      "learning_rate": 6.65258711721225e-06,
+      "loss": 0.2631,
+      "step": 928
+    },
+    {
+      "epoch": 2.646723646723647,
+      "grad_norm": 0.18382428762667938,
+      "learning_rate": 6.599788806758184e-06,
+      "loss": 0.2832,
+      "step": 929
+    },
+    {
+      "epoch": 2.6495726495726495,
+      "grad_norm": 0.18732937470339528,
+      "learning_rate": 6.546990496304118e-06,
+      "loss": 0.2885,
+      "step": 930
+    },
+    {
+      "epoch": 2.652421652421652,
+      "grad_norm": 0.18238422692769168,
+      "learning_rate": 6.494192185850054e-06,
+      "loss": 0.2714,
+      "step": 931
+    },
+    {
+      "epoch": 2.6552706552706553,
+      "grad_norm": 0.6796049281047194,
+      "learning_rate": 6.441393875395988e-06,
+      "loss": 0.2736,
+      "step": 932
+    },
+    {
+      "epoch": 2.658119658119658,
+      "grad_norm": 0.1843964132149436,
+      "learning_rate": 6.388595564941922e-06,
+      "loss": 0.2812,
+      "step": 933
+    },
+    {
+      "epoch": 2.6609686609686607,
+      "grad_norm": 0.1883869870506764,
+      "learning_rate": 6.335797254487856e-06,
+      "loss": 0.2876,
+      "step": 934
+    },
+    {
+      "epoch": 2.663817663817664,
+      "grad_norm": 0.19372865429873587,
+      "learning_rate": 6.282998944033791e-06,
+      "loss": 0.2721,
+      "step": 935
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.17475184508734368,
+      "learning_rate": 6.230200633579726e-06,
+      "loss": 0.2678,
+      "step": 936
+    },
+    {
+      "epoch": 2.6695156695156697,
+      "grad_norm": 0.19140736568353386,
+      "learning_rate": 6.17740232312566e-06,
+      "loss": 0.2694,
+      "step": 937
+    },
+    {
+      "epoch": 2.6723646723646723,
+      "grad_norm": 0.19095212031796968,
+      "learning_rate": 6.124604012671595e-06,
+      "loss": 0.2714,
+      "step": 938
+    },
+    {
+      "epoch": 2.6752136752136755,
+      "grad_norm": 0.18688967708133333,
+      "learning_rate": 6.071805702217529e-06,
+      "loss": 0.2777,
+      "step": 939
+    },
+    {
+      "epoch": 2.678062678062678,
+      "grad_norm": 0.1808730509793612,
+      "learning_rate": 6.019007391763464e-06,
+      "loss": 0.2867,
+      "step": 940
+    },
+    {
+      "epoch": 2.680911680911681,
+      "grad_norm": 0.1914570882813849,
+      "learning_rate": 5.966209081309398e-06,
+      "loss": 0.2752,
+      "step": 941
+    },
+    {
+      "epoch": 2.683760683760684,
+      "grad_norm": 0.18796118324257888,
+      "learning_rate": 5.913410770855333e-06,
+      "loss": 0.282,
+      "step": 942
+    },
+    {
+      "epoch": 2.6866096866096867,
+      "grad_norm": 0.19242846471973257,
+      "learning_rate": 5.860612460401267e-06,
+      "loss": 0.2815,
+      "step": 943
+    },
+    {
+      "epoch": 2.6894586894586894,
+      "grad_norm": 0.19514299446898078,
+      "learning_rate": 5.807814149947202e-06,
+      "loss": 0.2655,
+      "step": 944
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.1879893908540662,
+      "learning_rate": 5.755015839493136e-06,
+      "loss": 0.2857,
+      "step": 945
+    },
+    {
+      "epoch": 2.695156695156695,
+      "grad_norm": 0.19017667637491129,
+      "learning_rate": 5.702217529039071e-06,
+      "loss": 0.2807,
+      "step": 946
+    },
+    {
+      "epoch": 2.698005698005698,
+      "grad_norm": 0.18886407972543068,
+      "learning_rate": 5.6494192185850055e-06,
+      "loss": 0.2822,
+      "step": 947
+    },
+    {
+      "epoch": 2.700854700854701,
+      "grad_norm": 0.2030341917370533,
+      "learning_rate": 5.59662090813094e-06,
+      "loss": 0.2979,
+      "step": 948
+    },
+    {
+      "epoch": 2.7037037037037037,
+      "grad_norm": 0.18801018868503175,
+      "learning_rate": 5.543822597676875e-06,
+      "loss": 0.2754,
+      "step": 949
+    },
+    {
+      "epoch": 2.7065527065527064,
+      "grad_norm": 0.18722591061825838,
+      "learning_rate": 5.491024287222809e-06,
+      "loss": 0.291,
+      "step": 950
+    },
+    {
+      "epoch": 2.7094017094017095,
+      "grad_norm": 0.18419313321990788,
+      "learning_rate": 5.438225976768744e-06,
+      "loss": 0.2786,
+      "step": 951
+    },
+    {
+      "epoch": 2.7122507122507122,
+      "grad_norm": 0.1795053639312283,
+      "learning_rate": 5.385427666314678e-06,
+      "loss": 0.2616,
+      "step": 952
+    },
+    {
+      "epoch": 2.715099715099715,
+      "grad_norm": 0.19092668654205353,
+      "learning_rate": 5.332629355860613e-06,
+      "loss": 0.266,
+      "step": 953
+    },
+    {
+      "epoch": 2.717948717948718,
+      "grad_norm": 0.18515878208885198,
+      "learning_rate": 5.279831045406547e-06,
+      "loss": 0.2683,
+      "step": 954
+    },
+    {
+      "epoch": 2.7207977207977208,
+      "grad_norm": 0.18512141190512552,
+      "learning_rate": 5.227032734952482e-06,
+      "loss": 0.2613,
+      "step": 955
+    },
+    {
+      "epoch": 2.7236467236467234,
+      "grad_norm": 0.17864087041783647,
+      "learning_rate": 5.174234424498416e-06,
+      "loss": 0.2728,
+      "step": 956
+    },
+    {
+      "epoch": 2.7264957264957266,
+      "grad_norm": 0.18397723270469116,
+      "learning_rate": 5.121436114044351e-06,
+      "loss": 0.2878,
+      "step": 957
+    },
+    {
+      "epoch": 2.7293447293447293,
+      "grad_norm": 0.19206914593194782,
+      "learning_rate": 5.0686378035902855e-06,
+      "loss": 0.2898,
+      "step": 958
+    },
+    {
+      "epoch": 2.732193732193732,
+      "grad_norm": 0.18145822655495156,
+      "learning_rate": 5.01583949313622e-06,
+      "loss": 0.2811,
+      "step": 959
+    },
+    {
+      "epoch": 2.735042735042735,
+      "grad_norm": 0.18983105326320673,
+      "learning_rate": 4.963041182682154e-06,
+      "loss": 0.2961,
+      "step": 960
+    },
+    {
+      "epoch": 2.737891737891738,
+      "grad_norm": 0.18514943684352905,
+      "learning_rate": 4.910242872228089e-06,
+      "loss": 0.2687,
+      "step": 961
+    },
+    {
+      "epoch": 2.7407407407407405,
+      "grad_norm": 0.19048747070950345,
+      "learning_rate": 4.857444561774023e-06,
+      "loss": 0.2742,
+      "step": 962
+    },
+    {
+      "epoch": 2.7435897435897436,
+      "grad_norm": 0.18667552629580247,
+      "learning_rate": 4.804646251319958e-06,
+      "loss": 0.2778,
+      "step": 963
+    },
+    {
+      "epoch": 2.7464387464387463,
+      "grad_norm": 0.34574568987357907,
+      "learning_rate": 4.751847940865893e-06,
+      "loss": 0.2956,
+      "step": 964
+    },
+    {
+      "epoch": 2.7492877492877494,
+      "grad_norm": 0.18370077080612704,
+      "learning_rate": 4.699049630411827e-06,
+      "loss": 0.2873,
+      "step": 965
+    },
+    {
+      "epoch": 2.752136752136752,
+      "grad_norm": 0.18353727774729006,
+      "learning_rate": 4.646251319957761e-06,
+      "loss": 0.2729,
+      "step": 966
+    },
+    {
+      "epoch": 2.754985754985755,
+      "grad_norm": 0.19726651604825912,
+      "learning_rate": 4.593453009503696e-06,
+      "loss": 0.2774,
+      "step": 967
+    },
+    {
+      "epoch": 2.757834757834758,
+      "grad_norm": 0.18831766661505295,
+      "learning_rate": 4.540654699049631e-06,
+      "loss": 0.2918,
+      "step": 968
+    },
+    {
+      "epoch": 2.7606837606837606,
+      "grad_norm": 0.1822011598025739,
+      "learning_rate": 4.487856388595565e-06,
+      "loss": 0.2686,
+      "step": 969
+    },
+    {
+      "epoch": 2.763532763532764,
+      "grad_norm": 0.1845843568198463,
+      "learning_rate": 4.4350580781415e-06,
+      "loss": 0.2835,
+      "step": 970
+    },
+    {
+      "epoch": 2.7663817663817665,
+      "grad_norm": 0.1801611398952323,
+      "learning_rate": 4.382259767687434e-06,
+      "loss": 0.2738,
+      "step": 971
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.1798470399085861,
+      "learning_rate": 4.329461457233369e-06,
+      "loss": 0.2668,
+      "step": 972
+    },
+    {
+      "epoch": 2.7720797720797723,
+      "grad_norm": 0.18404634330655054,
+      "learning_rate": 4.276663146779303e-06,
+      "loss": 0.2685,
+      "step": 973
+    },
+    {
+      "epoch": 2.774928774928775,
+      "grad_norm": 0.19542597110088794,
+      "learning_rate": 4.223864836325238e-06,
+      "loss": 0.2854,
+      "step": 974
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.17947114319794663,
+      "learning_rate": 4.171066525871172e-06,
+      "loss": 0.2757,
+      "step": 975
+    },
+    {
+      "epoch": 2.780626780626781,
+      "grad_norm": 0.17988682607436737,
+      "learning_rate": 4.1182682154171074e-06,
+      "loss": 0.2623,
+      "step": 976
+    },
+    {
+      "epoch": 2.7834757834757835,
+      "grad_norm": 0.18466891151924947,
+      "learning_rate": 4.065469904963041e-06,
+      "loss": 0.2673,
+      "step": 977
+    },
+    {
+      "epoch": 2.786324786324786,
+      "grad_norm": 0.1803744986096463,
+      "learning_rate": 4.012671594508976e-06,
+      "loss": 0.2769,
+      "step": 978
+    },
+    {
+      "epoch": 2.7891737891737893,
+      "grad_norm": 0.18375712425725002,
+      "learning_rate": 3.95987328405491e-06,
+      "loss": 0.2736,
+      "step": 979
+    },
+    {
+      "epoch": 2.792022792022792,
+      "grad_norm": 0.1823276203279524,
+      "learning_rate": 3.907074973600845e-06,
+      "loss": 0.2689,
+      "step": 980
+    },
+    {
+      "epoch": 2.7948717948717947,
+      "grad_norm": 0.1896182512304451,
+      "learning_rate": 3.854276663146779e-06,
+      "loss": 0.29,
+      "step": 981
+    },
+    {
+      "epoch": 2.797720797720798,
+      "grad_norm": 0.20688431780425465,
+      "learning_rate": 3.801478352692714e-06,
+      "loss": 0.2788,
+      "step": 982
+    },
+    {
+      "epoch": 2.8005698005698005,
+      "grad_norm": 0.1758754882132954,
+      "learning_rate": 3.748680042238648e-06,
+      "loss": 0.264,
+      "step": 983
+    },
+    {
+      "epoch": 2.8034188034188032,
+      "grad_norm": 0.17784331145573304,
+      "learning_rate": 3.6958817317845833e-06,
+      "loss": 0.266,
+      "step": 984
+    },
+    {
+      "epoch": 2.8062678062678064,
+      "grad_norm": 0.2198776065570171,
+      "learning_rate": 3.6430834213305176e-06,
+      "loss": 0.2898,
+      "step": 985
+    },
+    {
+      "epoch": 2.809116809116809,
+      "grad_norm": 0.21195194560926758,
+      "learning_rate": 3.590285110876452e-06,
+      "loss": 0.2743,
+      "step": 986
+    },
+    {
+      "epoch": 2.8119658119658117,
+      "grad_norm": 0.17653709190418312,
+      "learning_rate": 3.537486800422387e-06,
+      "loss": 0.2684,
+      "step": 987
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 0.17452382412070946,
+      "learning_rate": 3.4846884899683212e-06,
+      "loss": 0.2672,
+      "step": 988
+    },
+    {
+      "epoch": 2.8176638176638176,
+      "grad_norm": 0.1710375478613801,
+      "learning_rate": 3.4318901795142555e-06,
+      "loss": 0.2704,
+      "step": 989
+    },
+    {
+      "epoch": 2.8205128205128203,
+      "grad_norm": 0.17775988353031527,
+      "learning_rate": 3.37909186906019e-06,
+      "loss": 0.277,
+      "step": 990
+    },
+    {
+      "epoch": 2.8233618233618234,
+      "grad_norm": 0.17938796003091406,
+      "learning_rate": 3.326293558606125e-06,
+      "loss": 0.2844,
+      "step": 991
+    },
+    {
+      "epoch": 2.826210826210826,
+      "grad_norm": 0.1762865102421063,
+      "learning_rate": 3.273495248152059e-06,
+      "loss": 0.2761,
+      "step": 992
+    },
+    {
+      "epoch": 2.8290598290598292,
+      "grad_norm": 0.18155573837120068,
+      "learning_rate": 3.220696937697994e-06,
+      "loss": 0.2783,
+      "step": 993
+    },
+    {
+      "epoch": 2.831908831908832,
+      "grad_norm": 0.1815146018987946,
+      "learning_rate": 3.167898627243928e-06,
+      "loss": 0.264,
+      "step": 994
+    },
+    {
+      "epoch": 2.8347578347578346,
+      "grad_norm": 0.17387563307830614,
+      "learning_rate": 3.115100316789863e-06,
+      "loss": 0.2672,
+      "step": 995
+    },
+    {
+      "epoch": 2.8376068376068377,
+      "grad_norm": 0.17754589749967897,
+      "learning_rate": 3.0623020063357975e-06,
+      "loss": 0.2604,
+      "step": 996
+    },
+    {
+      "epoch": 2.8404558404558404,
+      "grad_norm": 0.1788982942306512,
+      "learning_rate": 3.009503695881732e-06,
+      "loss": 0.2779,
+      "step": 997
+    },
+    {
+      "epoch": 2.8433048433048436,
+      "grad_norm": 0.17594751406364015,
+      "learning_rate": 2.9567053854276665e-06,
+      "loss": 0.279,
+      "step": 998
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.17837671634444546,
+      "learning_rate": 2.903907074973601e-06,
+      "loss": 0.2701,
+      "step": 999
+    },
+    {
+      "epoch": 2.849002849002849,
+      "grad_norm": 0.17682311169747053,
+      "learning_rate": 2.8511087645195354e-06,
+      "loss": 0.2823,
+      "step": 1000
+    },
+    {
+      "epoch": 2.851851851851852,
+      "grad_norm": 0.18269585079770145,
+      "learning_rate": 2.79831045406547e-06,
+      "loss": 0.2813,
+      "step": 1001
+    },
+    {
+      "epoch": 2.8547008547008548,
+      "grad_norm": 0.17761877789216213,
+      "learning_rate": 2.7455121436114044e-06,
+      "loss": 0.2776,
+      "step": 1002
+    },
+    {
+      "epoch": 2.8575498575498575,
+      "grad_norm": 0.1827216995625995,
+      "learning_rate": 2.692713833157339e-06,
+      "loss": 0.2763,
+      "step": 1003
+    },
+    {
+      "epoch": 2.8603988603988606,
+      "grad_norm": 0.17763354260264558,
+      "learning_rate": 2.6399155227032734e-06,
+      "loss": 0.2717,
+      "step": 1004
+    },
+    {
+      "epoch": 2.8632478632478633,
+      "grad_norm": 0.17915576491895774,
+      "learning_rate": 2.587117212249208e-06,
+      "loss": 0.2806,
+      "step": 1005
+    },
+    {
+      "epoch": 2.866096866096866,
+      "grad_norm": 0.18114152761979074,
+      "learning_rate": 2.5343189017951427e-06,
+      "loss": 0.2666,
+      "step": 1006
+    },
+    {
+      "epoch": 2.868945868945869,
+      "grad_norm": 0.18225553283170726,
+      "learning_rate": 2.481520591341077e-06,
+      "loss": 0.2624,
+      "step": 1007
+    },
+    {
+      "epoch": 2.871794871794872,
+      "grad_norm": 0.1728604793697752,
+      "learning_rate": 2.4287222808870117e-06,
+      "loss": 0.2737,
+      "step": 1008
+    },
+    {
+      "epoch": 2.8746438746438745,
+      "grad_norm": 0.18389395414319457,
+      "learning_rate": 2.3759239704329464e-06,
+      "loss": 0.2775,
+      "step": 1009
+    },
+    {
+      "epoch": 2.8774928774928776,
+      "grad_norm": 0.1796113765481641,
+      "learning_rate": 2.3231256599788807e-06,
+      "loss": 0.2817,
+      "step": 1010
+    },
+    {
+      "epoch": 2.8803418803418803,
+      "grad_norm": 0.18478152892147015,
+      "learning_rate": 2.2703273495248154e-06,
+      "loss": 0.2618,
+      "step": 1011
+    },
+    {
+      "epoch": 2.883190883190883,
+      "grad_norm": 0.17453347268398672,
+      "learning_rate": 2.21752903907075e-06,
+      "loss": 0.2749,
+      "step": 1012
+    },
+    {
+      "epoch": 2.886039886039886,
+      "grad_norm": 0.1839452873289296,
+      "learning_rate": 2.1647307286166843e-06,
+      "loss": 0.2713,
+      "step": 1013
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.2010428027513303,
+      "learning_rate": 2.111932418162619e-06,
+      "loss": 0.2704,
+      "step": 1014
+    },
+    {
+      "epoch": 2.8917378917378915,
+      "grad_norm": 0.18767478898568274,
+      "learning_rate": 2.0591341077085537e-06,
+      "loss": 0.2806,
+      "step": 1015
+    },
+    {
+      "epoch": 2.8945868945868947,
+      "grad_norm": 0.1752592201926614,
+      "learning_rate": 2.006335797254488e-06,
+      "loss": 0.2781,
+      "step": 1016
+    },
+    {
+      "epoch": 2.8974358974358974,
+      "grad_norm": 0.18503198686977332,
+      "learning_rate": 1.9535374868004227e-06,
+      "loss": 0.2829,
+      "step": 1017
+    },
+    {
+      "epoch": 2.9002849002849,
+      "grad_norm": 0.17491938775507743,
+      "learning_rate": 1.900739176346357e-06,
+      "loss": 0.2735,
+      "step": 1018
+    },
+    {
+      "epoch": 2.903133903133903,
+      "grad_norm": 0.18604953590453358,
+      "learning_rate": 1.8479408658922916e-06,
+      "loss": 0.2822,
+      "step": 1019
+    },
+    {
+      "epoch": 2.905982905982906,
+      "grad_norm": 0.17680948230251936,
+      "learning_rate": 1.795142555438226e-06,
+      "loss": 0.2654,
+      "step": 1020
+    },
+    {
+      "epoch": 2.9088319088319086,
+      "grad_norm": 0.17779057298202855,
+      "learning_rate": 1.7423442449841606e-06,
+      "loss": 0.2809,
+      "step": 1021
+    },
+    {
+      "epoch": 2.9116809116809117,
+      "grad_norm": 0.17367317767184776,
+      "learning_rate": 1.689545934530095e-06,
+      "loss": 0.2553,
+      "step": 1022
+    },
+    {
+      "epoch": 2.9145299145299144,
+      "grad_norm": 0.1849086381558193,
+      "learning_rate": 1.6367476240760296e-06,
+      "loss": 0.2704,
+      "step": 1023
+    },
+    {
+      "epoch": 2.9173789173789175,
+      "grad_norm": 0.2504975572538142,
+      "learning_rate": 1.583949313621964e-06,
+      "loss": 0.2855,
+      "step": 1024
+    },
+    {
+      "epoch": 2.92022792022792,
+      "grad_norm": 0.17838279902935134,
+      "learning_rate": 1.5311510031678987e-06,
+      "loss": 0.2633,
+      "step": 1025
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.187133440325405,
+      "learning_rate": 1.4783526927138332e-06,
+      "loss": 0.2681,
+      "step": 1026
+    },
+    {
+      "epoch": 2.925925925925926,
+      "grad_norm": 0.17495304668665154,
+      "learning_rate": 1.4255543822597677e-06,
+      "loss": 0.2747,
+      "step": 1027
+    },
+    {
+      "epoch": 2.9287749287749287,
+      "grad_norm": 0.1768663958665078,
+      "learning_rate": 1.3727560718057022e-06,
+      "loss": 0.2646,
+      "step": 1028
+    },
+    {
+      "epoch": 2.931623931623932,
+      "grad_norm": 0.17508470828848635,
+      "learning_rate": 1.3199577613516367e-06,
+      "loss": 0.2664,
+      "step": 1029
+    },
+    {
+      "epoch": 2.9344729344729346,
+      "grad_norm": 0.17563477250519552,
+      "learning_rate": 1.2671594508975714e-06,
+      "loss": 0.276,
+      "step": 1030
+    },
+    {
+      "epoch": 2.9373219373219372,
+      "grad_norm": 0.18333810882390714,
+      "learning_rate": 1.2143611404435059e-06,
+      "loss": 0.2724,
+      "step": 1031
+    },
+    {
+      "epoch": 2.9401709401709404,
+      "grad_norm": 0.18258423976716023,
+      "learning_rate": 1.1615628299894403e-06,
+      "loss": 0.2688,
+      "step": 1032
+    },
+    {
+      "epoch": 2.943019943019943,
+      "grad_norm": 0.1740940942606421,
+      "learning_rate": 1.108764519535375e-06,
+      "loss": 0.2628,
+      "step": 1033
+    },
+    {
+      "epoch": 2.9458689458689458,
+      "grad_norm": 0.18677299556807853,
+      "learning_rate": 1.0559662090813095e-06,
+      "loss": 0.2727,
+      "step": 1034
+    },
+    {
+      "epoch": 2.948717948717949,
+      "grad_norm": 0.17102730623765852,
+      "learning_rate": 1.003167898627244e-06,
+      "loss": 0.2601,
+      "step": 1035
+    },
+    {
+      "epoch": 2.9515669515669516,
+      "grad_norm": 0.17386263200587151,
+      "learning_rate": 9.503695881731785e-07,
+      "loss": 0.2657,
+      "step": 1036
+    },
+    {
+      "epoch": 2.9544159544159543,
+      "grad_norm": 0.1693040202829608,
+      "learning_rate": 8.97571277719113e-07,
+      "loss": 0.2639,
+      "step": 1037
+    },
+    {
+      "epoch": 2.9572649572649574,
+      "grad_norm": 0.1823134515198666,
+      "learning_rate": 8.447729672650475e-07,
+      "loss": 0.2809,
+      "step": 1038
+    },
+    {
+      "epoch": 2.96011396011396,
+      "grad_norm": 0.18713660780573363,
+      "learning_rate": 7.91974656810982e-07,
+      "loss": 0.2985,
+      "step": 1039
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.17859164741168496,
+      "learning_rate": 7.391763463569166e-07,
+      "loss": 0.2728,
+      "step": 1040
+    },
+    {
+      "epoch": 2.965811965811966,
+      "grad_norm": 0.17765435111212308,
+      "learning_rate": 6.863780359028511e-07,
+      "loss": 0.273,
+      "step": 1041
+    },
+    {
+      "epoch": 2.9686609686609686,
+      "grad_norm": 0.172193961059641,
+      "learning_rate": 6.335797254487857e-07,
+      "loss": 0.2777,
+      "step": 1042
+    },
+    {
+      "epoch": 2.9715099715099713,
+      "grad_norm": 0.17703694074644172,
+      "learning_rate": 5.807814149947202e-07,
+      "loss": 0.2729,
+      "step": 1043
+    },
+    {
+      "epoch": 2.9743589743589745,
+      "grad_norm": 0.17094716083563843,
+      "learning_rate": 5.279831045406548e-07,
+      "loss": 0.2565,
+      "step": 1044
+    },
+    {
+      "epoch": 2.977207977207977,
+      "grad_norm": 0.17381058656887102,
+      "learning_rate": 4.7518479408658924e-07,
+      "loss": 0.2676,
+      "step": 1045
+    },
+    {
+      "epoch": 2.98005698005698,
+      "grad_norm": 0.17011520350387407,
+      "learning_rate": 4.2238648363252377e-07,
+      "loss": 0.2711,
+      "step": 1046
+    },
+    {
+      "epoch": 2.982905982905983,
+      "grad_norm": 0.1693608759086481,
+      "learning_rate": 3.695881731784583e-07,
+      "loss": 0.2777,
+      "step": 1047
+    },
+    {
+      "epoch": 2.9857549857549857,
+      "grad_norm": 0.1691539977152749,
+      "learning_rate": 3.1678986272439284e-07,
+      "loss": 0.258,
+      "step": 1048
+    },
+    {
+      "epoch": 2.9886039886039883,
+      "grad_norm": 0.1725795825235304,
+      "learning_rate": 2.639915522703274e-07,
+      "loss": 0.2698,
+      "step": 1049
+    },
+    {
+      "epoch": 2.9914529914529915,
+      "grad_norm": 0.17564106791013198,
+      "learning_rate": 2.1119324181626189e-07,
+      "loss": 0.2727,
+      "step": 1050
+    },
+    {
+      "epoch": 2.994301994301994,
+      "grad_norm": 0.17354288175962146,
+      "learning_rate": 1.5839493136219642e-07,
+      "loss": 0.2743,
+      "step": 1051
+    },
+    {
+      "epoch": 2.9971509971509973,
+      "grad_norm": 0.17083245194414953,
+      "learning_rate": 1.0559662090813094e-07,
+      "loss": 0.2757,
+      "step": 1052
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.17469201616735772,
+      "learning_rate": 5.279831045406547e-08,
+      "loss": 0.2663,
+      "step": 1053
+    },
+    {
+      "epoch": 3.0,
+      "step": 1053,
+      "total_flos": 1.1708723554781495e+19,
+      "train_loss": 0.4741822677856384,
+      "train_runtime": 33666.7653,
+      "train_samples_per_second": 0.5,
+      "train_steps_per_second": 0.031
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1053,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1708723554781495e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}