diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24148 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9997083272568177,
+  "eval_steps": 200,
+  "global_step": 3428,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0005833454863642993,
+      "grad_norm": 17.78308868408203,
+      "learning_rate": 0.0,
+      "loss": 5.8378,
+      "step": 1
+    },
+    {
+      "epoch": 0.0011666909727285986,
+      "grad_norm": 5.597744464874268,
+      "learning_rate": 7.525749891599529e-06,
+      "loss": 2.3217,
+      "step": 2
+    },
+    {
+      "epoch": 0.0017500364590928978,
+      "grad_norm": 9.02523136138916,
+      "learning_rate": 1.192803136799156e-05,
+      "loss": 2.1816,
+      "step": 3
+    },
+    {
+      "epoch": 0.002333381945457197,
+      "grad_norm": 3.38098406791687,
+      "learning_rate": 1.5051499783199057e-05,
+      "loss": 1.7371,
+      "step": 4
+    },
+    {
+      "epoch": 0.002916727431821496,
+      "grad_norm": 3.2741100788116455,
+      "learning_rate": 1.7474250108400467e-05,
+      "loss": 1.6348,
+      "step": 5
+    },
+    {
+      "epoch": 0.0035000729181857955,
+      "grad_norm": 3.280085563659668,
+      "learning_rate": 1.945378125959109e-05,
+      "loss": 1.6431,
+      "step": 6
+    },
+    {
+      "epoch": 0.004083418404550095,
+      "grad_norm": 2.7186739444732666,
+      "learning_rate": 2.1127451000356418e-05,
+      "loss": 1.452,
+      "step": 7
+    },
+    {
+      "epoch": 0.004666763890914394,
+      "grad_norm": 2.5042686462402344,
+      "learning_rate": 2.2577249674798584e-05,
+      "loss": 1.3986,
+      "step": 8
+    },
+    {
+      "epoch": 0.005250109377278694,
+      "grad_norm": 2.6429643630981445,
+      "learning_rate": 2.385606273598312e-05,
+      "loss": 1.5973,
+      "step": 9
+    },
+    {
+      "epoch": 0.005833454863642992,
+      "grad_norm": 2.739990711212158,
+      "learning_rate": 2.4999999999999998e-05,
+      "loss": 1.4543,
+      "step": 10
+    },
+    {
+      "epoch": 0.006416800350007292,
+      "grad_norm": 2.085124969482422,
+      "learning_rate": 2.6034817128955623e-05,
+      "loss": 1.1983,
+      "step": 11
+    },
+    {
+      "epoch": 0.007000145836371591,
+      "grad_norm": 2.3114278316497803,
+      "learning_rate": 2.6979531151190617e-05,
+      "loss": 1.5163,
+      "step": 12
+    },
+    {
+      "epoch": 0.0075834913227358905,
+      "grad_norm": 2.1083810329437256,
+      "learning_rate": 2.7848583807670913e-05,
+      "loss": 1.3752,
+      "step": 13
+    },
+    {
+      "epoch": 0.00816683680910019,
+      "grad_norm": 2.667152166366577,
+      "learning_rate": 2.8653200891955945e-05,
+      "loss": 1.3118,
+      "step": 14
+    },
+    {
+      "epoch": 0.008750182295464488,
+      "grad_norm": 2.096151113510132,
+      "learning_rate": 2.940228147639203e-05,
+      "loss": 1.1005,
+      "step": 15
+    },
+    {
+      "epoch": 0.009333527781828789,
+      "grad_norm": 2.579360246658325,
+      "learning_rate": 3.0102999566398115e-05,
+      "loss": 1.5375,
+      "step": 16
+    },
+    {
+      "epoch": 0.009916873268193087,
+      "grad_norm": 2.0573806762695312,
+      "learning_rate": 3.076122303445685e-05,
+      "loss": 1.3534,
+      "step": 17
+    },
+    {
+      "epoch": 0.010500218754557387,
+      "grad_norm": 1.6891591548919678,
+      "learning_rate": 3.1381812627582646e-05,
+      "loss": 1.1602,
+      "step": 18
+    },
+    {
+      "epoch": 0.011083564240921686,
+      "grad_norm": 1.654665470123291,
+      "learning_rate": 3.1968840023820715e-05,
+      "loss": 1.2598,
+      "step": 19
+    },
+    {
+      "epoch": 0.011666909727285985,
+      "grad_norm": 1.51304030418396,
+      "learning_rate": 3.2525749891599525e-05,
+      "loss": 1.3738,
+      "step": 20
+    },
+    {
+      "epoch": 0.012250255213650285,
+      "grad_norm": 1.7645764350891113,
+      "learning_rate": 3.305548236834798e-05,
+      "loss": 1.3982,
+      "step": 21
+    },
+    {
+      "epoch": 0.012833600700014583,
+      "grad_norm": 2.0786468982696533,
+      "learning_rate": 3.3560567020555153e-05,
+      "loss": 1.2424,
+      "step": 22
+    },
+    {
+      "epoch": 0.013416946186378884,
+      "grad_norm": 1.5325442552566528,
+      "learning_rate": 3.404319590043982e-05,
+      "loss": 1.2803,
+      "step": 23
+    },
+    {
+      "epoch": 0.014000291672743182,
+      "grad_norm": 1.5579572916030884,
+      "learning_rate": 3.450528104279015e-05,
+      "loss": 1.4129,
+      "step": 24
+    },
+    {
+      "epoch": 0.01458363715910748,
+      "grad_norm": 1.6682405471801758,
+      "learning_rate": 3.4948500216800935e-05,
+      "loss": 1.1793,
+      "step": 25
+    },
+    {
+      "epoch": 0.015166982645471781,
+      "grad_norm": 1.558914303779602,
+      "learning_rate": 3.537433369927044e-05,
+      "loss": 1.2917,
+      "step": 26
+    },
+    {
+      "epoch": 0.01575032813183608,
+      "grad_norm": 2.367410182952881,
+      "learning_rate": 3.578409410397468e-05,
+      "loss": 1.3104,
+      "step": 27
+    },
+    {
+      "epoch": 0.01633367361820038,
+      "grad_norm": 1.7541717290878296,
+      "learning_rate": 3.6178950783555475e-05,
+      "loss": 1.1249,
+      "step": 28
+    },
+    {
+      "epoch": 0.016917019104564678,
+      "grad_norm": 1.3140240907669067,
+      "learning_rate": 3.65599499474739e-05,
+      "loss": 1.0676,
+      "step": 29
+    },
+    {
+      "epoch": 0.017500364590928977,
+      "grad_norm": 2.0365586280822754,
+      "learning_rate": 3.6928031367991554e-05,
+      "loss": 1.2787,
+      "step": 30
+    },
+    {
+      "epoch": 0.018083710077293275,
+      "grad_norm": 1.875240683555603,
+      "learning_rate": 3.728404234585681e-05,
+      "loss": 1.2876,
+      "step": 31
+    },
+    {
+      "epoch": 0.018667055563657577,
+      "grad_norm": 1.6128060817718506,
+      "learning_rate": 3.762874945799765e-05,
+      "loss": 1.1359,
+      "step": 32
+    },
+    {
+      "epoch": 0.019250401050021876,
+      "grad_norm": 1.761427640914917,
+      "learning_rate": 3.796284849694718e-05,
+      "loss": 1.2898,
+      "step": 33
+    },
+    {
+      "epoch": 0.019833746536386174,
+      "grad_norm": 1.8005716800689697,
+      "learning_rate": 3.8286972926056376e-05,
+      "loss": 1.0982,
+      "step": 34
+    },
+    {
+      "epoch": 0.020417092022750473,
+      "grad_norm": 1.7455655336380005,
+      "learning_rate": 3.8601701108756885e-05,
+      "loss": 1.1419,
+      "step": 35
+    },
+    {
+      "epoch": 0.021000437509114775,
+      "grad_norm": 1.207555890083313,
+      "learning_rate": 3.890756251918218e-05,
+      "loss": 1.3689,
+      "step": 36
+    },
+    {
+      "epoch": 0.021583782995479073,
+      "grad_norm": 1.4690238237380981,
+      "learning_rate": 3.920504310167487e-05,
+      "loss": 1.0896,
+      "step": 37
+    },
+    {
+      "epoch": 0.022167128481843372,
+      "grad_norm": 1.31458580493927,
+      "learning_rate": 3.949458991542025e-05,
+      "loss": 1.2124,
+      "step": 38
+    },
+    {
+      "epoch": 0.02275047396820767,
+      "grad_norm": 1.7508232593536377,
+      "learning_rate": 3.977661517566247e-05,
+      "loss": 1.0062,
+      "step": 39
+    },
+    {
+      "epoch": 0.02333381945457197,
+      "grad_norm": 1.7391034364700317,
+      "learning_rate": 4.005149978319905e-05,
+      "loss": 1.1858,
+      "step": 40
+    },
+    {
+      "epoch": 0.02391716494093627,
+      "grad_norm": 1.479203462600708,
+      "learning_rate": 4.031959641799338e-05,
+      "loss": 1.1886,
+      "step": 41
+    },
+    {
+      "epoch": 0.02450051042730057,
+      "grad_norm": 1.7898560762405396,
+      "learning_rate": 4.058123225994751e-05,
+      "loss": 1.204,
+      "step": 42
+    },
+    {
+      "epoch": 0.025083855913664868,
+      "grad_norm": 1.422057867050171,
+      "learning_rate": 4.0836711389489654e-05,
+      "loss": 1.1607,
+      "step": 43
+    },
+    {
+      "epoch": 0.025667201400029167,
+      "grad_norm": 2.161949872970581,
+      "learning_rate": 4.108631691215468e-05,
+      "loss": 1.1471,
+      "step": 44
+    },
+    {
+      "epoch": 0.026250546886393465,
+      "grad_norm": 1.9058781862258911,
+      "learning_rate": 4.133031284438358e-05,
+      "loss": 1.0656,
+      "step": 45
+    },
+    {
+      "epoch": 0.026833892372757767,
+      "grad_norm": 1.6354972124099731,
+      "learning_rate": 4.156894579203935e-05,
+      "loss": 1.0579,
+      "step": 46
+    },
+    {
+      "epoch": 0.027417237859122066,
+      "grad_norm": 1.6435108184814453,
+      "learning_rate": 4.180244644839293e-05,
+      "loss": 1.0281,
+      "step": 47
+    },
+    {
+      "epoch": 0.028000583345486364,
+      "grad_norm": 1.7230783700942993,
+      "learning_rate": 4.203103093438968e-05,
+      "loss": 1.086,
+      "step": 48
+    },
+    {
+      "epoch": 0.028583928831850663,
+      "grad_norm": 1.4264886379241943,
+      "learning_rate": 4.2254902000712836e-05,
+      "loss": 1.1384,
+      "step": 49
+    },
+    {
+      "epoch": 0.02916727431821496,
+      "grad_norm": 1.6517153978347778,
+      "learning_rate": 4.247425010840046e-05,
+      "loss": 1.2282,
+      "step": 50
+    },
+    {
+      "epoch": 0.029750619804579263,
+      "grad_norm": 2.1615817546844482,
+      "learning_rate": 4.2689254402448405e-05,
+      "loss": 1.2475,
+      "step": 51
+    },
+    {
+      "epoch": 0.030333965290943562,
+      "grad_norm": 1.5791531801223755,
+      "learning_rate": 4.290008359086998e-05,
+      "loss": 1.4099,
+      "step": 52
+    },
+    {
+      "epoch": 0.03091731077730786,
+      "grad_norm": 1.5546684265136719,
+      "learning_rate": 4.310689674001973e-05,
+      "loss": 1.2336,
+      "step": 53
+    },
+    {
+      "epoch": 0.03150065626367216,
+      "grad_norm": 1.5784398317337036,
+      "learning_rate": 4.330984399557421e-05,
+      "loss": 1.3104,
+      "step": 54
+    },
+    {
+      "epoch": 0.03208400175003646,
+      "grad_norm": 1.2873289585113525,
+      "learning_rate": 4.350906723735609e-05,
+      "loss": 1.3342,
+      "step": 55
+    },
+    {
+      "epoch": 0.03266734723640076,
+      "grad_norm": 1.6933963298797607,
+      "learning_rate": 4.370470067515501e-05,
+      "loss": 1.1348,
+      "step": 56
+    },
+    {
+      "epoch": 0.033250692722765054,
+      "grad_norm": 1.5604673624038696,
+      "learning_rate": 4.3896871391812285e-05,
+      "loss": 1.3968,
+      "step": 57
+    },
+    {
+      "epoch": 0.033834038209129357,
+      "grad_norm": 1.5108568668365479,
+      "learning_rate": 4.408569983907343e-05,
+      "loss": 1.1997,
+      "step": 58
+    },
+    {
+      "epoch": 0.03441738369549366,
+      "grad_norm": 3.0541892051696777,
+      "learning_rate": 4.42713002910536e-05,
+      "loss": 1.3484,
+      "step": 59
+    },
+    {
+      "epoch": 0.035000729181857954,
+      "grad_norm": 1.2510807514190674,
+      "learning_rate": 4.445378125959108e-05,
+      "loss": 1.3166,
+      "step": 60
+    },
+    {
+      "epoch": 0.035584074668222256,
+      "grad_norm": 1.6028800010681152,
+      "learning_rate": 4.463324587526917e-05,
+      "loss": 1.145,
+      "step": 61
+    },
+    {
+      "epoch": 0.03616742015458655,
+      "grad_norm": 1.5321674346923828,
+      "learning_rate": 4.4809792237456346e-05,
+      "loss": 1.2243,
+      "step": 62
+    },
+    {
+      "epoch": 0.03675076564095085,
+      "grad_norm": 1.4574953317642212,
+      "learning_rate": 4.498351373633954e-05,
+      "loss": 0.9716,
+      "step": 63
+    },
+    {
+      "epoch": 0.037334111127315155,
+      "grad_norm": 1.2741219997406006,
+      "learning_rate": 4.515449934959717e-05,
+      "loss": 1.083,
+      "step": 64
+    },
+    {
+      "epoch": 0.03791745661367945,
+      "grad_norm": 1.4046497344970703,
+      "learning_rate": 4.532283391607138e-05,
+      "loss": 1.1772,
+      "step": 65
+    },
+    {
+      "epoch": 0.03850080210004375,
+      "grad_norm": 1.3555320501327515,
+      "learning_rate": 4.548859838854671e-05,
+      "loss": 0.9133,
+      "step": 66
+    },
+    {
+      "epoch": 0.03908414758640805,
+      "grad_norm": 1.6886674165725708,
+      "learning_rate": 4.565187006752065e-05,
+      "loss": 1.1938,
+      "step": 67
+    },
+    {
+      "epoch": 0.03966749307277235,
+      "grad_norm": 1.430363416671753,
+      "learning_rate": 4.581272281765591e-05,
+      "loss": 1.1537,
+      "step": 68
+    },
+    {
+      "epoch": 0.04025083855913665,
+      "grad_norm": 1.6385464668273926,
+      "learning_rate": 4.597122726843138e-05,
+      "loss": 1.1556,
+      "step": 69
+    },
+    {
+      "epoch": 0.040834184045500946,
+      "grad_norm": 1.4001507759094238,
+      "learning_rate": 4.612745100035642e-05,
+      "loss": 1.1733,
+      "step": 70
+    },
+    {
+      "epoch": 0.04141752953186525,
+      "grad_norm": 1.5451918840408325,
+      "learning_rate": 4.628145871797688e-05,
+      "loss": 1.197,
+      "step": 71
+    },
+    {
+      "epoch": 0.04200087501822955,
+      "grad_norm": 1.5024776458740234,
+      "learning_rate": 4.643331241078171e-05,
+      "loss": 1.3481,
+      "step": 72
+    },
+    {
+      "epoch": 0.042584220504593845,
+      "grad_norm": 1.5428893566131592,
+      "learning_rate": 4.658307150301139e-05,
+      "loss": 1.1476,
+      "step": 73
+    },
+    {
+      "epoch": 0.04316756599095815,
+      "grad_norm": 1.2959415912628174,
+      "learning_rate": 4.67307929932744e-05,
+      "loss": 1.0119,
+      "step": 74
+    },
+    {
+      "epoch": 0.04375091147732244,
+      "grad_norm": 1.198490023612976,
+      "learning_rate": 4.687653158479249e-05,
+      "loss": 1.24,
+      "step": 75
+    },
+    {
+      "epoch": 0.044334256963686744,
+      "grad_norm": 1.3311015367507935,
+      "learning_rate": 4.702033980701978e-05,
+      "loss": 1.1441,
+      "step": 76
+    },
+    {
+      "epoch": 0.044917602450051046,
+      "grad_norm": 1.4103642702102661,
+      "learning_rate": 4.716226812931204e-05,
+      "loss": 1.0517,
+      "step": 77
+    },
+    {
+      "epoch": 0.04550094793641534,
+      "grad_norm": 1.260272741317749,
+      "learning_rate": 4.7302365067262006e-05,
+      "loss": 1.2173,
+      "step": 78
+    },
+    {
+      "epoch": 0.04608429342277964,
+      "grad_norm": 1.4968616962432861,
+      "learning_rate": 4.744067728226103e-05,
+      "loss": 1.0362,
+      "step": 79
+    },
+    {
+      "epoch": 0.04666763890914394,
+      "grad_norm": 1.2590322494506836,
+      "learning_rate": 4.757724967479858e-05,
+      "loss": 1.3106,
+      "step": 80
+    },
+    {
+      "epoch": 0.04725098439550824,
+      "grad_norm": 1.2166228294372559,
+      "learning_rate": 4.771212547196624e-05,
+      "loss": 1.0736,
+      "step": 81
+    },
+    {
+      "epoch": 0.04783432988187254,
+      "grad_norm": 1.19132399559021,
+      "learning_rate": 4.7845346309592914e-05,
+      "loss": 0.9754,
+      "step": 82
+    },
+    {
+      "epoch": 0.04841767536823684,
+      "grad_norm": 1.3279188871383667,
+      "learning_rate": 4.7976952309401844e-05,
+      "loss": 1.1815,
+      "step": 83
+    },
+    {
+      "epoch": 0.04900102085460114,
+      "grad_norm": 1.5890907049179077,
+      "learning_rate": 4.810698215154703e-05,
+      "loss": 1.0381,
+      "step": 84
+    },
+    {
+      "epoch": 0.049584366340965434,
+      "grad_norm": 1.2219558954238892,
+      "learning_rate": 4.823547314285732e-05,
+      "loss": 0.8901,
+      "step": 85
+    },
+    {
+      "epoch": 0.050167711827329736,
+      "grad_norm": 1.9673529863357544,
+      "learning_rate": 4.836246128108918e-05,
+      "loss": 1.087,
+      "step": 86
+    },
+    {
+      "epoch": 0.05075105731369404,
+      "grad_norm": 1.6241381168365479,
+      "learning_rate": 4.8487981315465456e-05,
+      "loss": 1.328,
+      "step": 87
+    },
+    {
+      "epoch": 0.05133440280005833,
+      "grad_norm": 1.289197325706482,
+      "learning_rate": 4.8612066803754214e-05,
+      "loss": 1.3474,
+      "step": 88
+    },
+    {
+      "epoch": 0.051917748286422635,
+      "grad_norm": 1.4262926578521729,
+      "learning_rate": 4.873475016612281e-05,
+      "loss": 1.0049,
+      "step": 89
+    },
+    {
+      "epoch": 0.05250109377278693,
+      "grad_norm": 1.382620930671692,
+      "learning_rate": 4.885606273598312e-05,
+      "loss": 0.94,
+      "step": 90
+    },
+    {
+      "epoch": 0.05308443925915123,
+      "grad_norm": 1.5160958766937256,
+      "learning_rate": 4.897603480802733e-05,
+      "loss": 1.1705,
+      "step": 91
+    },
+    {
+      "epoch": 0.053667784745515534,
+      "grad_norm": 2.5832581520080566,
+      "learning_rate": 4.909469568363888e-05,
+      "loss": 1.2074,
+      "step": 92
+    },
+    {
+      "epoch": 0.05425113023187983,
+      "grad_norm": 1.540708065032959,
+      "learning_rate": 4.9212073713848375e-05,
+      "loss": 1.2703,
+      "step": 93
+    },
+    {
+      "epoch": 0.05483447571824413,
+      "grad_norm": 1.8622777462005615,
+      "learning_rate": 4.932819633999246e-05,
+      "loss": 1.1169,
+      "step": 94
+    },
+    {
+      "epoch": 0.055417821204608426,
+      "grad_norm": 1.4035429954528809,
+      "learning_rate": 4.9443090132221186e-05,
+      "loss": 1.17,
+      "step": 95
+    },
+    {
+      "epoch": 0.05600116669097273,
+      "grad_norm": 1.4052140712738037,
+      "learning_rate": 4.9556780825989205e-05,
+      "loss": 0.9749,
+      "step": 96
+    },
+    {
+      "epoch": 0.05658451217733703,
+      "grad_norm": 1.3691339492797852,
+      "learning_rate": 4.9669293356656114e-05,
+      "loss": 0.9748,
+      "step": 97
+    },
+    {
+      "epoch": 0.057167857663701326,
+      "grad_norm": 1.2556352615356445,
+      "learning_rate": 4.978065189231237e-05,
+      "loss": 0.9405,
+      "step": 98
+    },
+    {
+      "epoch": 0.05775120315006563,
+      "grad_norm": 1.2853363752365112,
+      "learning_rate": 4.989087986493874e-05,
+      "loss": 1.3587,
+      "step": 99
+    },
+    {
+      "epoch": 0.05833454863642992,
+      "grad_norm": 2.179720640182495,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 1.181,
+      "step": 100
+    },
+    {
+      "epoch": 0.058917894122794225,
+      "grad_norm": 1.1849037408828735,
+      "learning_rate": 5e-05,
+      "loss": 1.0379,
+      "step": 101
+    },
+    {
+      "epoch": 0.05950123960915853,
+      "grad_norm": 1.335351586341858,
+      "learning_rate": 4.998497596153847e-05,
+      "loss": 1.2294,
+      "step": 102
+    },
+    {
+      "epoch": 0.06008458509552282,
+      "grad_norm": 1.5505980253219604,
+      "learning_rate": 4.9969951923076926e-05,
+      "loss": 0.9985,
+      "step": 103
+    },
+    {
+      "epoch": 0.060667930581887124,
+      "grad_norm": 1.2492345571517944,
+      "learning_rate": 4.9954927884615385e-05,
+      "loss": 1.009,
+      "step": 104
+    },
+    {
+      "epoch": 0.06125127606825142,
+      "grad_norm": 1.2737586498260498,
+      "learning_rate": 4.993990384615384e-05,
+      "loss": 1.075,
+      "step": 105
+    },
+    {
+      "epoch": 0.06183462155461572,
+      "grad_norm": 1.6333897113800049,
+      "learning_rate": 4.992487980769231e-05,
+      "loss": 1.2081,
+      "step": 106
+    },
+    {
+      "epoch": 0.06241796704098002,
+      "grad_norm": 1.5222417116165161,
+      "learning_rate": 4.9909855769230774e-05,
+      "loss": 1.2142,
+      "step": 107
+    },
+    {
+      "epoch": 0.06300131252734432,
+      "grad_norm": 1.683530569076538,
+      "learning_rate": 4.989483173076923e-05,
+      "loss": 0.9966,
+      "step": 108
+    },
+    {
+      "epoch": 0.06358465801370862,
+      "grad_norm": 1.8551238775253296,
+      "learning_rate": 4.98798076923077e-05,
+      "loss": 1.0807,
+      "step": 109
+    },
+    {
+      "epoch": 0.06416800350007291,
+      "grad_norm": 1.721418857574463,
+      "learning_rate": 4.9864783653846156e-05,
+      "loss": 0.907,
+      "step": 110
+    },
+    {
+      "epoch": 0.06475134898643722,
+      "grad_norm": 1.1599464416503906,
+      "learning_rate": 4.9849759615384615e-05,
+      "loss": 1.0121,
+      "step": 111
+    },
+    {
+      "epoch": 0.06533469447280152,
+      "grad_norm": 1.1434050798416138,
+      "learning_rate": 4.983473557692308e-05,
+      "loss": 1.0762,
+      "step": 112
+    },
+    {
+      "epoch": 0.06591803995916581,
+      "grad_norm": 1.3993638753890991,
+      "learning_rate": 4.981971153846154e-05,
+      "loss": 1.2264,
+      "step": 113
+    },
+    {
+      "epoch": 0.06650138544553011,
+      "grad_norm": 1.278794288635254,
+      "learning_rate": 4.9804687500000004e-05,
+      "loss": 1.1514,
+      "step": 114
+    },
+    {
+      "epoch": 0.06708473093189442,
+      "grad_norm": 4.100466728210449,
+      "learning_rate": 4.978966346153847e-05,
+      "loss": 1.1164,
+      "step": 115
+    },
+    {
+      "epoch": 0.06766807641825871,
+      "grad_norm": 1.2248213291168213,
+      "learning_rate": 4.977463942307693e-05,
+      "loss": 1.1503,
+      "step": 116
+    },
+    {
+      "epoch": 0.06825142190462301,
+      "grad_norm": 1.7549059391021729,
+      "learning_rate": 4.9759615384615386e-05,
+      "loss": 1.185,
+      "step": 117
+    },
+    {
+      "epoch": 0.06883476739098732,
+      "grad_norm": 1.747718334197998,
+      "learning_rate": 4.9744591346153844e-05,
+      "loss": 1.1494,
+      "step": 118
+    },
+    {
+      "epoch": 0.06941811287735161,
+      "grad_norm": 1.4090042114257812,
+      "learning_rate": 4.972956730769231e-05,
+      "loss": 1.0752,
+      "step": 119
+    },
+    {
+      "epoch": 0.07000145836371591,
+      "grad_norm": 1.3040906190872192,
+      "learning_rate": 4.9714543269230775e-05,
+      "loss": 0.9461,
+      "step": 120
+    },
+    {
+      "epoch": 0.07058480385008022,
+      "grad_norm": 1.625506043434143,
+      "learning_rate": 4.9699519230769233e-05,
+      "loss": 1.0583,
+      "step": 121
+    },
+    {
+      "epoch": 0.07116814933644451,
+      "grad_norm": 1.273437261581421,
+      "learning_rate": 4.968449519230769e-05,
+      "loss": 1.1448,
+      "step": 122
+    },
+    {
+      "epoch": 0.0717514948228088,
+      "grad_norm": 1.3089839220046997,
+      "learning_rate": 4.966947115384616e-05,
+      "loss": 1.0703,
+      "step": 123
+    },
+    {
+      "epoch": 0.0723348403091731,
+      "grad_norm": 1.1929200887680054,
+      "learning_rate": 4.9654447115384616e-05,
+      "loss": 0.9157,
+      "step": 124
+    },
+    {
+      "epoch": 0.07291818579553741,
+      "grad_norm": 1.1315945386886597,
+      "learning_rate": 4.963942307692308e-05,
+      "loss": 1.2748,
+      "step": 125
+    },
+    {
+      "epoch": 0.0735015312819017,
+      "grad_norm": 1.4779376983642578,
+      "learning_rate": 4.962439903846154e-05,
+      "loss": 1.0475,
+      "step": 126
+    },
+    {
+      "epoch": 0.074084876768266,
+      "grad_norm": 1.3037022352218628,
+      "learning_rate": 4.9609375000000005e-05,
+      "loss": 1.1542,
+      "step": 127
+    },
+    {
+      "epoch": 0.07466822225463031,
+      "grad_norm": 1.454216480255127,
+      "learning_rate": 4.959435096153846e-05,
+      "loss": 1.3286,
+      "step": 128
+    },
+    {
+      "epoch": 0.0752515677409946,
+      "grad_norm": 1.5513452291488647,
+      "learning_rate": 4.957932692307692e-05,
+      "loss": 1.2599,
+      "step": 129
+    },
+    {
+      "epoch": 0.0758349132273589,
+      "grad_norm": 1.7299277782440186,
+      "learning_rate": 4.956430288461539e-05,
+      "loss": 1.2156,
+      "step": 130
+    },
+    {
+      "epoch": 0.07641825871372321,
+      "grad_norm": 1.2468066215515137,
+      "learning_rate": 4.9549278846153846e-05,
+      "loss": 1.0193,
+      "step": 131
+    },
+    {
+      "epoch": 0.0770016042000875,
+      "grad_norm": 1.1203701496124268,
+      "learning_rate": 4.953425480769231e-05,
+      "loss": 1.0451,
+      "step": 132
+    },
+    {
+      "epoch": 0.0775849496864518,
+      "grad_norm": 1.4089607000350952,
+      "learning_rate": 4.9519230769230776e-05,
+      "loss": 1.0041,
+      "step": 133
+    },
+    {
+      "epoch": 0.0781682951728161,
+      "grad_norm": 1.40877103805542,
+      "learning_rate": 4.9504206730769235e-05,
+      "loss": 1.1952,
+      "step": 134
+    },
+    {
+      "epoch": 0.0787516406591804,
+      "grad_norm": 1.2676079273223877,
+      "learning_rate": 4.948918269230769e-05,
+      "loss": 0.9846,
+      "step": 135
+    },
+    {
+      "epoch": 0.0793349861455447,
+      "grad_norm": 1.3630707263946533,
+      "learning_rate": 4.947415865384616e-05,
+      "loss": 1.2925,
+      "step": 136
+    },
+    {
+      "epoch": 0.07991833163190899,
+      "grad_norm": 1.4193191528320312,
+      "learning_rate": 4.945913461538462e-05,
+      "loss": 0.9525,
+      "step": 137
+    },
+    {
+      "epoch": 0.0805016771182733,
+      "grad_norm": 1.4956103563308716,
+      "learning_rate": 4.944411057692308e-05,
+      "loss": 1.0733,
+      "step": 138
+    },
+    {
+      "epoch": 0.0810850226046376,
+      "grad_norm": 1.4280532598495483,
+      "learning_rate": 4.942908653846154e-05,
+      "loss": 0.9752,
+      "step": 139
+    },
+    {
+      "epoch": 0.08166836809100189,
+      "grad_norm": 1.4108835458755493,
+      "learning_rate": 4.94140625e-05,
+      "loss": 1.0615,
+      "step": 140
+    },
+    {
+      "epoch": 0.0822517135773662,
+      "grad_norm": 1.3212484121322632,
+      "learning_rate": 4.9399038461538464e-05,
+      "loss": 0.957,
+      "step": 141
+    },
+    {
+      "epoch": 0.0828350590637305,
+      "grad_norm": 1.3035906553268433,
+      "learning_rate": 4.938401442307692e-05,
+      "loss": 1.0273,
+      "step": 142
+    },
+    {
+      "epoch": 0.08341840455009479,
+      "grad_norm": 1.9610090255737305,
+      "learning_rate": 4.936899038461539e-05,
+      "loss": 1.2096,
+      "step": 143
+    },
+    {
+      "epoch": 0.0840017500364591,
+      "grad_norm": 1.5158385038375854,
+      "learning_rate": 4.935396634615385e-05,
+      "loss": 1.056,
+      "step": 144
+    },
+    {
+      "epoch": 0.0845850955228234,
+      "grad_norm": 1.9964501857757568,
+      "learning_rate": 4.933894230769231e-05,
+      "loss": 1.0332,
+      "step": 145
+    },
+    {
+      "epoch": 0.08516844100918769,
+      "grad_norm": 1.5921212434768677,
+      "learning_rate": 4.932391826923077e-05,
+      "loss": 1.0627,
+      "step": 146
+    },
+    {
+      "epoch": 0.08575178649555198,
+      "grad_norm": 1.6017435789108276,
+      "learning_rate": 4.930889423076923e-05,
+      "loss": 1.0619,
+      "step": 147
+    },
+    {
+      "epoch": 0.0863351319819163,
+      "grad_norm": 1.4326040744781494,
+      "learning_rate": 4.9293870192307694e-05,
+      "loss": 1.0111,
+      "step": 148
+    },
+    {
+      "epoch": 0.08691847746828059,
+      "grad_norm": 1.0582656860351562,
+      "learning_rate": 4.927884615384616e-05,
+      "loss": 0.978,
+      "step": 149
+    },
+    {
+      "epoch": 0.08750182295464488,
+      "grad_norm": 1.1291985511779785,
+      "learning_rate": 4.926382211538462e-05,
+      "loss": 1.2015,
+      "step": 150
+    },
+    {
+      "epoch": 0.08808516844100919,
+      "grad_norm": 1.6709433794021606,
+      "learning_rate": 4.924879807692308e-05,
+      "loss": 1.086,
+      "step": 151
+    },
+    {
+      "epoch": 0.08866851392737349,
+      "grad_norm": 1.3261191844940186,
+      "learning_rate": 4.923377403846154e-05,
+      "loss": 1.2622,
+      "step": 152
+    },
+    {
+      "epoch": 0.08925185941373778,
+      "grad_norm": 1.3250443935394287,
+      "learning_rate": 4.921875e-05,
+      "loss": 0.9489,
+      "step": 153
+    },
+    {
+      "epoch": 0.08983520490010209,
+      "grad_norm": 1.3647021055221558,
+      "learning_rate": 4.9203725961538466e-05,
+      "loss": 1.0411,
+      "step": 154
+    },
+    {
+      "epoch": 0.09041855038646639,
+      "grad_norm": 1.5882066488265991,
+      "learning_rate": 4.9188701923076924e-05,
+      "loss": 1.2391,
+      "step": 155
+    },
+    {
+      "epoch": 0.09100189587283068,
+      "grad_norm": 1.2136874198913574,
+      "learning_rate": 4.917367788461539e-05,
+      "loss": 1.0796,
+      "step": 156
+    },
+    {
+      "epoch": 0.09158524135919498,
+      "grad_norm": 1.3928202390670776,
+      "learning_rate": 4.915865384615385e-05,
+      "loss": 1.2077,
+      "step": 157
+    },
+    {
+      "epoch": 0.09216858684555929,
+      "grad_norm": 1.487825632095337,
+      "learning_rate": 4.9143629807692306e-05,
+      "loss": 1.1106,
+      "step": 158
+    },
+    {
+      "epoch": 0.09275193233192358,
+      "grad_norm": 1.7569891214370728,
+      "learning_rate": 4.912860576923077e-05,
+      "loss": 1.2142,
+      "step": 159
+    },
+    {
+      "epoch": 0.09333527781828788,
+      "grad_norm": 1.2210508584976196,
+      "learning_rate": 4.911358173076923e-05,
+      "loss": 1.1436,
+      "step": 160
+    },
+    {
+      "epoch": 0.09391862330465219,
+      "grad_norm": 1.2036561965942383,
+      "learning_rate": 4.9098557692307695e-05,
+      "loss": 1.227,
+      "step": 161
+    },
+    {
+      "epoch": 0.09450196879101648,
+      "grad_norm": 1.2647464275360107,
+      "learning_rate": 4.908353365384616e-05,
+      "loss": 0.9552,
+      "step": 162
+    },
+    {
+      "epoch": 0.09508531427738078,
+      "grad_norm": 1.2640854120254517,
+      "learning_rate": 4.906850961538462e-05,
+      "loss": 1.2703,
+      "step": 163
+    },
+    {
+      "epoch": 0.09566865976374508,
+      "grad_norm": 1.8966978788375854,
+      "learning_rate": 4.905348557692308e-05,
+      "loss": 1.2019,
+      "step": 164
+    },
+    {
+      "epoch": 0.09625200525010938,
+      "grad_norm": 1.3982689380645752,
+      "learning_rate": 4.9038461538461536e-05,
+      "loss": 1.0861,
+      "step": 165
+    },
+    {
+      "epoch": 0.09683535073647367,
+      "grad_norm": 1.3104490041732788,
+      "learning_rate": 4.90234375e-05,
+      "loss": 1.0978,
+      "step": 166
+    },
+    {
+      "epoch": 0.09741869622283797,
+      "grad_norm": 1.3252923488616943,
+      "learning_rate": 4.900841346153847e-05,
+      "loss": 0.984,
+      "step": 167
+    },
+    {
+      "epoch": 0.09800204170920228,
+      "grad_norm": 1.636171579360962,
+      "learning_rate": 4.8993389423076925e-05,
+      "loss": 1.0612,
+      "step": 168
+    },
+    {
+      "epoch": 0.09858538719556657,
+      "grad_norm": 1.3443262577056885,
+      "learning_rate": 4.897836538461539e-05,
+      "loss": 1.0664,
+      "step": 169
+    },
+    {
+      "epoch": 0.09916873268193087,
+      "grad_norm": 1.2440283298492432,
+      "learning_rate": 4.896334134615385e-05,
+      "loss": 1.083,
+      "step": 170
+    },
+    {
+      "epoch": 0.09975207816829518,
+      "grad_norm": 1.125069499015808,
+      "learning_rate": 4.894831730769231e-05,
+      "loss": 1.1206,
+      "step": 171
+    },
+    {
+      "epoch": 0.10033542365465947,
+      "grad_norm": 1.3264683485031128,
+      "learning_rate": 4.893329326923077e-05,
+      "loss": 0.9415,
+      "step": 172
+    },
+    {
+      "epoch": 0.10091876914102377,
+      "grad_norm": 1.1505907773971558,
+      "learning_rate": 4.891826923076923e-05,
+      "loss": 1.0793,
+      "step": 173
+    },
+    {
+      "epoch": 0.10150211462738808,
+      "grad_norm": 1.1637053489685059,
+      "learning_rate": 4.89032451923077e-05,
+      "loss": 0.9331,
+      "step": 174
+    },
+    {
+      "epoch": 0.10208546011375237,
+      "grad_norm": 1.6416714191436768,
+      "learning_rate": 4.888822115384616e-05,
+      "loss": 0.8152,
+      "step": 175
+    },
+    {
+      "epoch": 0.10266880560011667,
+      "grad_norm": 1.0731501579284668,
+      "learning_rate": 4.8873197115384614e-05,
+      "loss": 1.188,
+      "step": 176
+    },
+    {
+      "epoch": 0.10325215108648096,
+      "grad_norm": 1.5831621885299683,
+      "learning_rate": 4.885817307692308e-05,
+      "loss": 1.1591,
+      "step": 177
+    },
+    {
+      "epoch": 0.10383549657284527,
+      "grad_norm": 1.221384882926941,
+      "learning_rate": 4.884314903846154e-05,
+      "loss": 1.0392,
+      "step": 178
+    },
+    {
+      "epoch": 0.10441884205920957,
+      "grad_norm": 1.2436399459838867,
+      "learning_rate": 4.8828125e-05,
+      "loss": 1.0176,
+      "step": 179
+    },
+    {
+      "epoch": 0.10500218754557386,
+      "grad_norm": 1.6474889516830444,
+      "learning_rate": 4.881310096153847e-05,
+      "loss": 1.294,
+      "step": 180
+    },
+    {
+      "epoch": 0.10558553303193817,
+      "grad_norm": 1.7568507194519043,
+      "learning_rate": 4.8798076923076926e-05,
+      "loss": 1.1217,
+      "step": 181
+    },
+    {
+      "epoch": 0.10616887851830246,
+      "grad_norm": 1.6313364505767822,
+      "learning_rate": 4.8783052884615385e-05,
+      "loss": 1.0687,
+      "step": 182
+    },
+    {
+      "epoch": 0.10675222400466676,
+      "grad_norm": 1.3143326044082642,
+      "learning_rate": 4.8768028846153843e-05,
+      "loss": 1.001,
+      "step": 183
+    },
+    {
+      "epoch": 0.10733556949103107,
+      "grad_norm": 1.7886399030685425,
+      "learning_rate": 4.875300480769231e-05,
+      "loss": 1.2089,
+      "step": 184
+    },
+    {
+      "epoch": 0.10791891497739536,
+      "grad_norm": 1.494966983795166,
+      "learning_rate": 4.8737980769230774e-05,
+      "loss": 1.096,
+      "step": 185
+    },
+    {
+      "epoch": 0.10850226046375966,
+      "grad_norm": 1.311043620109558,
+      "learning_rate": 4.872295673076923e-05,
+      "loss": 1.0791,
+      "step": 186
+    },
+    {
+      "epoch": 0.10908560595012397,
+      "grad_norm": 1.4188566207885742,
+      "learning_rate": 4.87079326923077e-05,
+      "loss": 0.8453,
+      "step": 187
+    },
+    {
+      "epoch": 0.10966895143648826,
+      "grad_norm": 1.2535464763641357,
+      "learning_rate": 4.8692908653846156e-05,
+      "loss": 0.9653,
+      "step": 188
+    },
+    {
+      "epoch": 0.11025229692285256,
+      "grad_norm": 1.2611030340194702,
+      "learning_rate": 4.8677884615384615e-05,
+      "loss": 1.2116,
+      "step": 189
+    },
+    {
+      "epoch": 0.11083564240921685,
+      "grad_norm": 1.33787202835083,
+      "learning_rate": 4.866286057692308e-05,
+      "loss": 1.0238,
+      "step": 190
+    },
+    {
+      "epoch": 0.11141898789558116,
+      "grad_norm": 1.256029486656189,
+      "learning_rate": 4.864783653846154e-05,
+      "loss": 1.123,
+      "step": 191
+    },
+    {
+      "epoch": 0.11200233338194546,
+      "grad_norm": 1.2442421913146973,
+      "learning_rate": 4.8632812500000004e-05,
+      "loss": 1.0065,
+      "step": 192
+    },
+    {
+      "epoch": 0.11258567886830975,
+      "grad_norm": 1.540716290473938,
+      "learning_rate": 4.861778846153847e-05,
+      "loss": 1.1194,
+      "step": 193
+    },
+    {
+      "epoch": 0.11316902435467406,
+      "grad_norm": 1.2217652797698975,
+      "learning_rate": 4.860276442307692e-05,
+      "loss": 1.0814,
+      "step": 194
+    },
+    {
+      "epoch": 0.11375236984103836,
+      "grad_norm": 1.1543757915496826,
+      "learning_rate": 4.8587740384615386e-05,
+      "loss": 1.1429,
+      "step": 195
+    },
+    {
+      "epoch": 0.11433571532740265,
+      "grad_norm": 1.2116531133651733,
+      "learning_rate": 4.8572716346153845e-05,
+      "loss": 0.8905,
+      "step": 196
+    },
+    {
+      "epoch": 0.11491906081376696,
+      "grad_norm": 1.2289031744003296,
+      "learning_rate": 4.855769230769231e-05,
+      "loss": 0.9011,
+      "step": 197
+    },
+    {
+      "epoch": 0.11550240630013126,
+      "grad_norm": 1.6803086996078491,
+      "learning_rate": 4.8542668269230775e-05,
+      "loss": 1.0375,
+      "step": 198
+    },
+    {
+      "epoch": 0.11608575178649555,
+      "grad_norm": 1.5701960325241089,
+      "learning_rate": 4.8527644230769234e-05,
+      "loss": 1.0346,
+      "step": 199
+    },
+    {
+      "epoch": 0.11666909727285985,
+      "grad_norm": 1.2407475709915161,
+      "learning_rate": 4.851262019230769e-05,
+      "loss": 0.7745,
+      "step": 200
+    },
+    {
+      "epoch": 0.11666909727285985,
+      "eval_loss_squad": 1.1932364337146282,
+      "eval_perplexity": 7.5639171404401075,
+      "eval_perplexity_reconstruct": 1.8755490788116311,
+      "step": 200
+    },
+    {
+      "epoch": 0.11725244275922415,
+      "grad_norm": 1.604348063468933,
+      "learning_rate": 4.849759615384616e-05,
+      "loss": 1.1338,
+      "step": 201
+    },
+    {
+      "epoch": 0.11783578824558845,
+      "grad_norm": 1.2575210332870483,
+      "learning_rate": 4.8482572115384616e-05,
+      "loss": 1.1509,
+      "step": 202
+    },
+    {
+      "epoch": 0.11841913373195274,
+      "grad_norm": 1.662980318069458,
+      "learning_rate": 4.846754807692308e-05,
+      "loss": 1.1348,
+      "step": 203
+    },
+    {
+      "epoch": 0.11900247921831705,
+      "grad_norm": 1.3378013372421265,
+      "learning_rate": 4.845252403846154e-05,
+      "loss": 1.2898,
+      "step": 204
+    },
+    {
+      "epoch": 0.11958582470468135,
+      "grad_norm": 1.4811590909957886,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.9847,
+      "step": 205
+    },
+    {
+      "epoch": 0.12016917019104564,
+      "grad_norm": 1.452512502670288,
+      "learning_rate": 4.8422475961538464e-05,
+      "loss": 1.0591,
+      "step": 206
+    },
+    {
+      "epoch": 0.12075251567740995,
+      "grad_norm": 1.1843714714050293,
+      "learning_rate": 4.840745192307692e-05,
+      "loss": 0.7955,
+      "step": 207
+    },
+    {
+      "epoch": 0.12133586116377425,
+      "grad_norm": 1.3748295307159424,
+      "learning_rate": 4.839242788461539e-05,
+      "loss": 1.0937,
+      "step": 208
+    },
+    {
+      "epoch": 0.12191920665013854,
+      "grad_norm": 1.1598683595657349,
+      "learning_rate": 4.8377403846153846e-05,
+      "loss": 1.0785,
+      "step": 209
+    },
+    {
+      "epoch": 0.12250255213650284,
+      "grad_norm": 1.3770662546157837,
+      "learning_rate": 4.836237980769231e-05,
+      "loss": 1.1478,
+      "step": 210
+    },
+    {
+      "epoch": 0.12308589762286715,
+      "grad_norm": 1.3409414291381836,
+      "learning_rate": 4.8347355769230776e-05,
+      "loss": 1.2035,
+      "step": 211
+    },
+    {
+      "epoch": 0.12366924310923144,
+      "grad_norm": 1.0668489933013916,
+      "learning_rate": 4.833233173076923e-05,
+      "loss": 1.148,
+      "step": 212
+    },
+    {
+      "epoch": 0.12425258859559574,
+      "grad_norm": 1.2233036756515503,
+      "learning_rate": 4.8317307692307693e-05,
+      "loss": 0.9074,
+      "step": 213
+    },
+    {
+      "epoch": 0.12483593408196005,
+      "grad_norm": 1.2284631729125977,
+      "learning_rate": 4.830228365384616e-05,
+      "loss": 1.006,
+      "step": 214
+    },
+    {
+      "epoch": 0.12541927956832433,
+      "grad_norm": 1.0120066404342651,
+      "learning_rate": 4.828725961538462e-05,
+      "loss": 1.1474,
+      "step": 215
+    },
+    {
+      "epoch": 0.12600262505468865,
+      "grad_norm": 1.47971773147583,
+      "learning_rate": 4.827223557692308e-05,
+      "loss": 0.7954,
+      "step": 216
+    },
+    {
+      "epoch": 0.12658597054105294,
+      "grad_norm": 1.3422911167144775,
+      "learning_rate": 4.825721153846154e-05,
+      "loss": 1.0716,
+      "step": 217
+    },
+    {
+      "epoch": 0.12716931602741724,
+      "grad_norm": 1.1010000705718994,
+      "learning_rate": 4.82421875e-05,
+      "loss": 1.1334,
+      "step": 218
+    },
+    {
+      "epoch": 0.12775266151378153,
+      "grad_norm": 1.4309356212615967,
+      "learning_rate": 4.8227163461538465e-05,
+      "loss": 1.137,
+      "step": 219
+    },
+    {
+      "epoch": 0.12833600700014583,
+      "grad_norm": 1.333970308303833,
+      "learning_rate": 4.821213942307692e-05,
+      "loss": 1.2948,
+      "step": 220
+    },
+    {
+      "epoch": 0.12891935248651012,
+      "grad_norm": 1.3190792798995972,
+      "learning_rate": 4.819711538461539e-05,
+      "loss": 1.061,
+      "step": 221
+    },
+    {
+      "epoch": 0.12950269797287445,
+      "grad_norm": 1.3843986988067627,
+      "learning_rate": 4.818209134615385e-05,
+      "loss": 1.1526,
+      "step": 222
+    },
+    {
+      "epoch": 0.13008604345923874,
+      "grad_norm": 1.6656113862991333,
+      "learning_rate": 4.816706730769231e-05,
+      "loss": 1.2739,
+      "step": 223
+    },
+    {
+      "epoch": 0.13066938894560304,
+      "grad_norm": 1.3917752504348755,
+      "learning_rate": 4.815204326923077e-05,
+      "loss": 1.1356,
+      "step": 224
+    },
+    {
+      "epoch": 0.13125273443196733,
+      "grad_norm": 1.3735393285751343,
+      "learning_rate": 4.813701923076923e-05,
+      "loss": 0.9553,
+      "step": 225
+    },
+    {
+      "epoch": 0.13183607991833163,
+      "grad_norm": 1.253967523574829,
+      "learning_rate": 4.8121995192307695e-05,
+      "loss": 1.1367,
+      "step": 226
+    },
+    {
+      "epoch": 0.13241942540469592,
+      "grad_norm": 1.2027031183242798,
+      "learning_rate": 4.810697115384616e-05,
+      "loss": 1.129,
+      "step": 227
+    },
+    {
+      "epoch": 0.13300277089106022,
+      "grad_norm": 1.2082641124725342,
+      "learning_rate": 4.809194711538462e-05,
+      "loss": 1.0744,
+      "step": 228
+    },
+    {
+      "epoch": 0.13358611637742454,
+      "grad_norm": 1.4022235870361328,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 0.9777,
+      "step": 229
+    },
+    {
+      "epoch": 0.13416946186378884,
+      "grad_norm": 1.1199394464492798,
+      "learning_rate": 4.8061899038461535e-05,
+      "loss": 0.9213,
+      "step": 230
+    },
+    {
+      "epoch": 0.13475280735015313,
+      "grad_norm": 1.1612290143966675,
+      "learning_rate": 4.8046875e-05,
+      "loss": 1.1336,
+      "step": 231
+    },
+    {
+      "epoch": 0.13533615283651743,
+      "grad_norm": 1.43931245803833,
+      "learning_rate": 4.8031850961538466e-05,
+      "loss": 1.2295,
+      "step": 232
+    },
+    {
+      "epoch": 0.13591949832288172,
+      "grad_norm": 1.4430086612701416,
+      "learning_rate": 4.8016826923076924e-05,
+      "loss": 1.1819,
+      "step": 233
+    },
+    {
+      "epoch": 0.13650284380924602,
+      "grad_norm": 1.1566003561019897,
+      "learning_rate": 4.800180288461539e-05,
+      "loss": 0.9493,
+      "step": 234
+    },
+    {
+      "epoch": 0.13708618929561034,
+      "grad_norm": 1.1817492246627808,
+      "learning_rate": 4.798677884615385e-05,
+      "loss": 1.0723,
+      "step": 235
+    },
+    {
+      "epoch": 0.13766953478197463,
+      "grad_norm": 1.0918781757354736,
+      "learning_rate": 4.797175480769231e-05,
+      "loss": 1.1297,
+      "step": 236
+    },
+    {
+      "epoch": 0.13825288026833893,
+      "grad_norm": 1.2325400114059448,
+      "learning_rate": 4.795673076923077e-05,
+      "loss": 1.0465,
+      "step": 237
+    },
+    {
+      "epoch": 0.13883622575470322,
+      "grad_norm": 1.4473427534103394,
+      "learning_rate": 4.794170673076923e-05,
+      "loss": 1.1118,
+      "step": 238
+    },
+    {
+      "epoch": 0.13941957124106752,
+      "grad_norm": 1.2843005657196045,
+      "learning_rate": 4.7926682692307696e-05,
+      "loss": 0.7924,
+      "step": 239
+    },
+    {
+      "epoch": 0.14000291672743181,
+      "grad_norm": 1.225659728050232,
+      "learning_rate": 4.791165865384616e-05,
+      "loss": 0.9887,
+      "step": 240
+    },
+    {
+      "epoch": 0.1405862622137961,
+      "grad_norm": 1.3597815036773682,
+      "learning_rate": 4.789663461538462e-05,
+      "loss": 1.063,
+      "step": 241
+    },
+    {
+      "epoch": 0.14116960770016043,
+      "grad_norm": 1.2931313514709473,
+      "learning_rate": 4.788161057692308e-05,
+      "loss": 1.1008,
+      "step": 242
+    },
+    {
+      "epoch": 0.14175295318652473,
+      "grad_norm": 1.6043051481246948,
+      "learning_rate": 4.7866586538461537e-05,
+      "loss": 1.2153,
+      "step": 243
+    },
+    {
+      "epoch": 0.14233629867288902,
+      "grad_norm": 1.2612937688827515,
+      "learning_rate": 4.78515625e-05,
+      "loss": 1.05,
+      "step": 244
+    },
+    {
+      "epoch": 0.14291964415925332,
+      "grad_norm": 1.4583909511566162,
+      "learning_rate": 4.783653846153847e-05,
+      "loss": 1.0092,
+      "step": 245
+    },
+    {
+      "epoch": 0.1435029896456176,
+      "grad_norm": 1.4382933378219604,
+      "learning_rate": 4.7821514423076926e-05,
+      "loss": 0.8433,
+      "step": 246
+    },
+    {
+      "epoch": 0.1440863351319819,
+      "grad_norm": 1.1039222478866577,
+      "learning_rate": 4.780649038461539e-05,
+      "loss": 0.9546,
+      "step": 247
+    },
+    {
+      "epoch": 0.1446696806183462,
+      "grad_norm": 1.46884024143219,
+      "learning_rate": 4.779146634615384e-05,
+      "loss": 1.1191,
+      "step": 248
+    },
+    {
+      "epoch": 0.14525302610471053,
+      "grad_norm": 1.221130609512329,
+      "learning_rate": 4.777644230769231e-05,
+      "loss": 0.9995,
+      "step": 249
+    },
+    {
+      "epoch": 0.14583637159107482,
+      "grad_norm": 1.2932945489883423,
+      "learning_rate": 4.776141826923077e-05,
+      "loss": 1.0508,
+      "step": 250
+    },
+    {
+      "epoch": 0.14641971707743912,
+      "grad_norm": 1.174483060836792,
+      "learning_rate": 4.774639423076923e-05,
+      "loss": 1.1433,
+      "step": 251
+    },
+    {
+      "epoch": 0.1470030625638034,
+      "grad_norm": 1.366163969039917,
+      "learning_rate": 4.77313701923077e-05,
+      "loss": 1.1716,
+      "step": 252
+    },
+    {
+      "epoch": 0.1475864080501677,
+      "grad_norm": 1.9801900386810303,
+      "learning_rate": 4.7716346153846155e-05,
+      "loss": 0.9884,
+      "step": 253
+    },
+    {
+      "epoch": 0.148169753536532,
+      "grad_norm": 1.3597514629364014,
+      "learning_rate": 4.7701322115384614e-05,
+      "loss": 1.0706,
+      "step": 254
+    },
+    {
+      "epoch": 0.14875309902289632,
+      "grad_norm": 1.1929199695587158,
+      "learning_rate": 4.768629807692308e-05,
+      "loss": 0.996,
+      "step": 255
+    },
+    {
+      "epoch": 0.14933644450926062,
+      "grad_norm": 1.0896735191345215,
+      "learning_rate": 4.767127403846154e-05,
+      "loss": 1.2185,
+      "step": 256
+    },
+    {
+      "epoch": 0.1499197899956249,
+      "grad_norm": 1.0902819633483887,
+      "learning_rate": 4.765625e-05,
+      "loss": 1.3523,
+      "step": 257
+    },
+    {
+      "epoch": 0.1505031354819892,
+      "grad_norm": 1.1990022659301758,
+      "learning_rate": 4.764122596153847e-05,
+      "loss": 1.049,
+      "step": 258
+    },
+    {
+      "epoch": 0.1510864809683535,
+      "grad_norm": 1.2676620483398438,
+      "learning_rate": 4.762620192307693e-05,
+      "loss": 1.2471,
+      "step": 259
+    },
+    {
+      "epoch": 0.1516698264547178,
+      "grad_norm": 1.0942872762680054,
+      "learning_rate": 4.7611177884615385e-05,
+      "loss": 1.0276,
+      "step": 260
+    },
+    {
+      "epoch": 0.1522531719410821,
+      "grad_norm": 1.886791706085205,
+      "learning_rate": 4.7596153846153844e-05,
+      "loss": 0.9001,
+      "step": 261
+    },
+    {
+      "epoch": 0.15283651742744642,
+      "grad_norm": 1.3173785209655762,
+      "learning_rate": 4.758112980769231e-05,
+      "loss": 1.0866,
+      "step": 262
+    },
+    {
+      "epoch": 0.1534198629138107,
+      "grad_norm": 1.3886914253234863,
+      "learning_rate": 4.7566105769230774e-05,
+      "loss": 1.0645,
+      "step": 263
+    },
+    {
+      "epoch": 0.154003208400175,
+      "grad_norm": 1.5575705766677856,
+      "learning_rate": 4.755108173076923e-05,
+      "loss": 1.0723,
+      "step": 264
+    },
+    {
+      "epoch": 0.1545865538865393,
+      "grad_norm": 1.2715169191360474,
+      "learning_rate": 4.75360576923077e-05,
+      "loss": 0.7773,
+      "step": 265
+    },
+    {
+      "epoch": 0.1551698993729036,
+      "grad_norm": 1.5639898777008057,
+      "learning_rate": 4.752103365384616e-05,
+      "loss": 1.0814,
+      "step": 266
+    },
+    {
+      "epoch": 0.1557532448592679,
+      "grad_norm": 1.5546399354934692,
+      "learning_rate": 4.7506009615384615e-05,
+      "loss": 0.9091,
+      "step": 267
+    },
+    {
+      "epoch": 0.1563365903456322,
+      "grad_norm": 1.1600172519683838,
+      "learning_rate": 4.749098557692308e-05,
+      "loss": 1.2764,
+      "step": 268
+    },
+    {
+      "epoch": 0.1569199358319965,
+      "grad_norm": 1.4420632123947144,
+      "learning_rate": 4.747596153846154e-05,
+      "loss": 1.1393,
+      "step": 269
+    },
+    {
+      "epoch": 0.1575032813183608,
+      "grad_norm": 1.2289047241210938,
+      "learning_rate": 4.7460937500000004e-05,
+      "loss": 0.9656,
+      "step": 270
+    },
+    {
+      "epoch": 0.1580866268047251,
+      "grad_norm": 1.165598750114441,
+      "learning_rate": 4.744591346153846e-05,
+      "loss": 0.9472,
+      "step": 271
+    },
+    {
+      "epoch": 0.1586699722910894,
+      "grad_norm": 1.1388275623321533,
+      "learning_rate": 4.743088942307692e-05,
+      "loss": 1.0464,
+      "step": 272
+    },
+    {
+      "epoch": 0.1592533177774537,
+      "grad_norm": 1.1367133855819702,
+      "learning_rate": 4.7415865384615386e-05,
+      "loss": 0.9505,
+      "step": 273
+    },
+    {
+      "epoch": 0.15983666326381799,
+      "grad_norm": 1.086214542388916,
+      "learning_rate": 4.7400841346153845e-05,
+      "loss": 0.8583,
+      "step": 274
+    },
+    {
+      "epoch": 0.1604200087501823,
+      "grad_norm": 1.464219093322754,
+      "learning_rate": 4.738581730769231e-05,
+      "loss": 0.9827,
+      "step": 275
+    },
+    {
+      "epoch": 0.1610033542365466,
+      "grad_norm": 1.486207365989685,
+      "learning_rate": 4.7370793269230776e-05,
+      "loss": 1.0802,
+      "step": 276
+    },
+    {
+      "epoch": 0.1615866997229109,
+      "grad_norm": 1.265541434288025,
+      "learning_rate": 4.7355769230769234e-05,
+      "loss": 1.0812,
+      "step": 277
+    },
+    {
+      "epoch": 0.1621700452092752,
+      "grad_norm": 1.532962441444397,
+      "learning_rate": 4.734074519230769e-05,
+      "loss": 0.9444,
+      "step": 278
+    },
+    {
+      "epoch": 0.1627533906956395,
+      "grad_norm": 1.329508900642395,
+      "learning_rate": 4.732572115384616e-05,
+      "loss": 1.1593,
+      "step": 279
+    },
+    {
+      "epoch": 0.16333673618200378,
+      "grad_norm": 1.2211445569992065,
+      "learning_rate": 4.7310697115384616e-05,
+      "loss": 1.1621,
+      "step": 280
+    },
+    {
+      "epoch": 0.16392008166836808,
+      "grad_norm": 1.0366151332855225,
+      "learning_rate": 4.729567307692308e-05,
+      "loss": 1.0649,
+      "step": 281
+    },
+    {
+      "epoch": 0.1645034271547324,
+      "grad_norm": 1.4749221801757812,
+      "learning_rate": 4.728064903846154e-05,
+      "loss": 1.1227,
+      "step": 282
+    },
+    {
+      "epoch": 0.1650867726410967,
+      "grad_norm": 1.3342058658599854,
+      "learning_rate": 4.7265625000000005e-05,
+      "loss": 1.1781,
+      "step": 283
+    },
+    {
+      "epoch": 0.165670118127461,
+      "grad_norm": 1.0524542331695557,
+      "learning_rate": 4.7250600961538464e-05,
+      "loss": 1.0728,
+      "step": 284
+    },
+    {
+      "epoch": 0.1662534636138253,
+      "grad_norm": 1.2550618648529053,
+      "learning_rate": 4.723557692307692e-05,
+      "loss": 0.9786,
+      "step": 285
+    },
+    {
+      "epoch": 0.16683680910018958,
+      "grad_norm": 1.6661171913146973,
+      "learning_rate": 4.722055288461539e-05,
+      "loss": 1.1991,
+      "step": 286
+    },
+    {
+      "epoch": 0.16742015458655388,
+      "grad_norm": 1.480756163597107,
+      "learning_rate": 4.7205528846153846e-05,
+      "loss": 1.1048,
+      "step": 287
+    },
+    {
+      "epoch": 0.1680035000729182,
+      "grad_norm": 1.4135624170303345,
+      "learning_rate": 4.719050480769231e-05,
+      "loss": 1.1372,
+      "step": 288
+    },
+    {
+      "epoch": 0.1685868455592825,
+      "grad_norm": 1.5737360715866089,
+      "learning_rate": 4.717548076923077e-05,
+      "loss": 1.2462,
+      "step": 289
+    },
+    {
+      "epoch": 0.1691701910456468,
+      "grad_norm": 1.2390706539154053,
+      "learning_rate": 4.716045673076923e-05,
+      "loss": 0.9622,
+      "step": 290
+    },
+    {
+      "epoch": 0.16975353653201108,
+      "grad_norm": 1.193597435951233,
+      "learning_rate": 4.7145432692307694e-05,
+      "loss": 1.0312,
+      "step": 291
+    },
+    {
+      "epoch": 0.17033688201837538,
+      "grad_norm": 1.2671687602996826,
+      "learning_rate": 4.713040865384616e-05,
+      "loss": 1.0976,
+      "step": 292
+    },
+    {
+      "epoch": 0.17092022750473967,
+      "grad_norm": 1.220468521118164,
+      "learning_rate": 4.711538461538462e-05,
+      "loss": 0.8168,
+      "step": 293
+    },
+    {
+      "epoch": 0.17150357299110397,
+      "grad_norm": 1.7847779989242554,
+      "learning_rate": 4.710036057692308e-05,
+      "loss": 1.1623,
+      "step": 294
+    },
+    {
+      "epoch": 0.1720869184774683,
+      "grad_norm": 1.1684701442718506,
+      "learning_rate": 4.708533653846154e-05,
+      "loss": 1.0476,
+      "step": 295
+    },
+    {
+      "epoch": 0.1726702639638326,
+      "grad_norm": 1.2110151052474976,
+      "learning_rate": 4.70703125e-05,
+      "loss": 1.2198,
+      "step": 296
+    },
+    {
+      "epoch": 0.17325360945019688,
+      "grad_norm": 1.2417210340499878,
+      "learning_rate": 4.7055288461538465e-05,
+      "loss": 0.9928,
+      "step": 297
+    },
+    {
+      "epoch": 0.17383695493656118,
+      "grad_norm": 1.4387229681015015,
+      "learning_rate": 4.7040264423076924e-05,
+      "loss": 1.2669,
+      "step": 298
+    },
+    {
+      "epoch": 0.17442030042292547,
+      "grad_norm": 1.5898375511169434,
+      "learning_rate": 4.702524038461539e-05,
+      "loss": 1.055,
+      "step": 299
+    },
+    {
+      "epoch": 0.17500364590928977,
+      "grad_norm": 1.5764034986495972,
+      "learning_rate": 4.701021634615385e-05,
+      "loss": 1.0568,
+      "step": 300
+    },
+    {
+      "epoch": 0.17558699139565406,
+      "grad_norm": 1.1175717115402222,
+      "learning_rate": 4.699519230769231e-05,
+      "loss": 0.9626,
+      "step": 301
+    },
+    {
+      "epoch": 0.17617033688201839,
+      "grad_norm": 1.0562156438827515,
+      "learning_rate": 4.698016826923077e-05,
+      "loss": 1.0173,
+      "step": 302
+    },
+    {
+      "epoch": 0.17675368236838268,
+      "grad_norm": 1.1503074169158936,
+      "learning_rate": 4.696514423076923e-05,
+      "loss": 1.0513,
+      "step": 303
+    },
+    {
+      "epoch": 0.17733702785474698,
+      "grad_norm": 1.1285890340805054,
+      "learning_rate": 4.6950120192307695e-05,
+      "loss": 0.966,
+      "step": 304
+    },
+    {
+      "epoch": 0.17792037334111127,
+      "grad_norm": 1.0698230266571045,
+      "learning_rate": 4.693509615384616e-05,
+      "loss": 1.1029,
+      "step": 305
+    },
+    {
+      "epoch": 0.17850371882747557,
+      "grad_norm": 1.3292865753173828,
+      "learning_rate": 4.692007211538462e-05,
+      "loss": 1.19,
+      "step": 306
+    },
+    {
+      "epoch": 0.17908706431383986,
+      "grad_norm": 1.1927766799926758,
+      "learning_rate": 4.690504807692308e-05,
+      "loss": 1.1935,
+      "step": 307
+    },
+    {
+      "epoch": 0.17967040980020418,
+      "grad_norm": 2.0182723999023438,
+      "learning_rate": 4.6890024038461536e-05,
+      "loss": 1.2596,
+      "step": 308
+    },
+    {
+      "epoch": 0.18025375528656848,
+      "grad_norm": 1.8118665218353271,
+      "learning_rate": 4.6875e-05,
+      "loss": 1.0279,
+      "step": 309
+    },
+    {
+      "epoch": 0.18083710077293277,
+      "grad_norm": 1.3893784284591675,
+      "learning_rate": 4.6859975961538466e-05,
+      "loss": 1.0969,
+      "step": 310
+    },
+    {
+      "epoch": 0.18142044625929707,
+      "grad_norm": 4.505359172821045,
+      "learning_rate": 4.6844951923076925e-05,
+      "loss": 1.04,
+      "step": 311
+    },
+    {
+      "epoch": 0.18200379174566136,
+      "grad_norm": 1.258991003036499,
+      "learning_rate": 4.682992788461539e-05,
+      "loss": 0.9914,
+      "step": 312
+    },
+    {
+      "epoch": 0.18258713723202566,
+      "grad_norm": 1.0383793115615845,
+      "learning_rate": 4.681490384615385e-05,
+      "loss": 1.0399,
+      "step": 313
+    },
+    {
+      "epoch": 0.18317048271838995,
+      "grad_norm": 1.4799776077270508,
+      "learning_rate": 4.679987980769231e-05,
+      "loss": 0.9968,
+      "step": 314
+    },
+    {
+      "epoch": 0.18375382820475428,
+      "grad_norm": 1.3189499378204346,
+      "learning_rate": 4.678485576923077e-05,
+      "loss": 1.0315,
+      "step": 315
+    },
+    {
+      "epoch": 0.18433717369111857,
+      "grad_norm": 1.2589702606201172,
+      "learning_rate": 4.676983173076923e-05,
+      "loss": 0.8627,
+      "step": 316
+    },
+    {
+      "epoch": 0.18492051917748287,
+      "grad_norm": 1.2101918458938599,
+      "learning_rate": 4.6754807692307696e-05,
+      "loss": 1.2105,
+      "step": 317
+    },
+    {
+      "epoch": 0.18550386466384716,
+      "grad_norm": 1.3174864053726196,
+      "learning_rate": 4.673978365384616e-05,
+      "loss": 0.9905,
+      "step": 318
+    },
+    {
+      "epoch": 0.18608721015021146,
+      "grad_norm": 1.2675038576126099,
+      "learning_rate": 4.672475961538462e-05,
+      "loss": 1.0751,
+      "step": 319
+    },
+    {
+      "epoch": 0.18667055563657575,
+      "grad_norm": 1.8191653490066528,
+      "learning_rate": 4.670973557692308e-05,
+      "loss": 1.1342,
+      "step": 320
+    },
+    {
+      "epoch": 0.18725390112294005,
+      "grad_norm": 1.3822802305221558,
+      "learning_rate": 4.669471153846154e-05,
+      "loss": 0.9589,
+      "step": 321
+    },
+    {
+      "epoch": 0.18783724660930437,
+      "grad_norm": 1.386513590812683,
+      "learning_rate": 4.66796875e-05,
+      "loss": 1.1085,
+      "step": 322
+    },
+    {
+      "epoch": 0.18842059209566867,
+      "grad_norm": 1.1989296674728394,
+      "learning_rate": 4.666466346153847e-05,
+      "loss": 1.1038,
+      "step": 323
+    },
+    {
+      "epoch": 0.18900393758203296,
+      "grad_norm": 1.1725409030914307,
+      "learning_rate": 4.6649639423076926e-05,
+      "loss": 0.9671,
+      "step": 324
+    },
+    {
+      "epoch": 0.18958728306839726,
+      "grad_norm": 1.1871923208236694,
+      "learning_rate": 4.6634615384615384e-05,
+      "loss": 0.9863,
+      "step": 325
+    },
+    {
+      "epoch": 0.19017062855476155,
+      "grad_norm": 1.4059133529663086,
+      "learning_rate": 4.661959134615384e-05,
+      "loss": 0.858,
+      "step": 326
+    },
+    {
+      "epoch": 0.19075397404112585,
+      "grad_norm": 1.4571958780288696,
+      "learning_rate": 4.660456730769231e-05,
+      "loss": 1.1032,
+      "step": 327
+    },
+    {
+      "epoch": 0.19133731952749017,
+      "grad_norm": 1.4414441585540771,
+      "learning_rate": 4.6589543269230773e-05,
+      "loss": 1.1884,
+      "step": 328
+    },
+    {
+      "epoch": 0.19192066501385446,
+      "grad_norm": 1.9027329683303833,
+      "learning_rate": 4.657451923076923e-05,
+      "loss": 1.0415,
+      "step": 329
+    },
+    {
+      "epoch": 0.19250401050021876,
+      "grad_norm": 1.2300493717193604,
+      "learning_rate": 4.65594951923077e-05,
+      "loss": 0.8911,
+      "step": 330
+    },
+    {
+      "epoch": 0.19308735598658305,
+      "grad_norm": 3.4765067100524902,
+      "learning_rate": 4.6544471153846156e-05,
+      "loss": 0.9321,
+      "step": 331
+    },
+    {
+      "epoch": 0.19367070147294735,
+      "grad_norm": 1.4455662965774536,
+      "learning_rate": 4.6529447115384614e-05,
+      "loss": 0.9986,
+      "step": 332
+    },
+    {
+      "epoch": 0.19425404695931164,
+      "grad_norm": 1.371666669845581,
+      "learning_rate": 4.651442307692308e-05,
+      "loss": 1.1358,
+      "step": 333
+    },
+    {
+      "epoch": 0.19483739244567594,
+      "grad_norm": 1.2533602714538574,
+      "learning_rate": 4.649939903846154e-05,
+      "loss": 0.9561,
+      "step": 334
+    },
+    {
+      "epoch": 0.19542073793204026,
+      "grad_norm": 1.152901291847229,
+      "learning_rate": 4.6484375e-05,
+      "loss": 1.0428,
+      "step": 335
+    },
+    {
+      "epoch": 0.19600408341840456,
+      "grad_norm": 1.435378074645996,
+      "learning_rate": 4.646935096153847e-05,
+      "loss": 0.9433,
+      "step": 336
+    },
+    {
+      "epoch": 0.19658742890476885,
+      "grad_norm": 1.480175495147705,
+      "learning_rate": 4.645432692307693e-05,
+      "loss": 0.9606,
+      "step": 337
+    },
+    {
+      "epoch": 0.19717077439113315,
+      "grad_norm": 1.4078121185302734,
+      "learning_rate": 4.6439302884615386e-05,
+      "loss": 1.078,
+      "step": 338
+    },
+    {
+      "epoch": 0.19775411987749744,
+      "grad_norm": 1.3780559301376343,
+      "learning_rate": 4.6424278846153844e-05,
+      "loss": 1.1521,
+      "step": 339
+    },
+    {
+      "epoch": 0.19833746536386174,
+      "grad_norm": 1.5239243507385254,
+      "learning_rate": 4.640925480769231e-05,
+      "loss": 0.9855,
+      "step": 340
+    },
+    {
+      "epoch": 0.19892081085022606,
+      "grad_norm": 1.095953106880188,
+      "learning_rate": 4.6394230769230775e-05,
+      "loss": 1.1482,
+      "step": 341
+    },
+    {
+      "epoch": 0.19950415633659035,
+      "grad_norm": 1.4764552116394043,
+      "learning_rate": 4.637920673076923e-05,
+      "loss": 1.0308,
+      "step": 342
+    },
+    {
+      "epoch": 0.20008750182295465,
+      "grad_norm": 1.2678275108337402,
+      "learning_rate": 4.636418269230769e-05,
+      "loss": 1.0498,
+      "step": 343
+    },
+    {
+      "epoch": 0.20067084730931894,
+      "grad_norm": 1.3208849430084229,
+      "learning_rate": 4.634915865384616e-05,
+      "loss": 0.8625,
+      "step": 344
+    },
+    {
+      "epoch": 0.20125419279568324,
+      "grad_norm": 1.3627556562423706,
+      "learning_rate": 4.6334134615384615e-05,
+      "loss": 1.0739,
+      "step": 345
+    },
+    {
+      "epoch": 0.20183753828204753,
+      "grad_norm": 1.3360272645950317,
+      "learning_rate": 4.631911057692308e-05,
+      "loss": 0.9951,
+      "step": 346
+    },
+    {
+      "epoch": 0.20242088376841183,
+      "grad_norm": 1.2518396377563477,
+      "learning_rate": 4.630408653846154e-05,
+      "loss": 1.0037,
+      "step": 347
+    },
+    {
+      "epoch": 0.20300422925477615,
+      "grad_norm": 1.2477564811706543,
+      "learning_rate": 4.6289062500000005e-05,
+      "loss": 1.0325,
+      "step": 348
+    },
+    {
+      "epoch": 0.20358757474114045,
+      "grad_norm": 0.9390896558761597,
+      "learning_rate": 4.627403846153846e-05,
+      "loss": 0.9671,
+      "step": 349
+    },
+    {
+      "epoch": 0.20417092022750474,
+      "grad_norm": 1.2609106302261353,
+      "learning_rate": 4.625901442307692e-05,
+      "loss": 0.9002,
+      "step": 350
+    },
+    {
+      "epoch": 0.20475426571386904,
+      "grad_norm": 1.3142434358596802,
+      "learning_rate": 4.624399038461539e-05,
+      "loss": 1.0176,
+      "step": 351
+    },
+    {
+      "epoch": 0.20533761120023333,
+      "grad_norm": 2.33760142326355,
+      "learning_rate": 4.6228966346153845e-05,
+      "loss": 1.2718,
+      "step": 352
+    },
+    {
+      "epoch": 0.20592095668659763,
+      "grad_norm": 1.0962882041931152,
+      "learning_rate": 4.621394230769231e-05,
+      "loss": 1.1066,
+      "step": 353
+    },
+    {
+      "epoch": 0.20650430217296192,
+      "grad_norm": 1.1384212970733643,
+      "learning_rate": 4.6198918269230776e-05,
+      "loss": 0.986,
+      "step": 354
+    },
+    {
+      "epoch": 0.20708764765932625,
+      "grad_norm": 1.3480985164642334,
+      "learning_rate": 4.6183894230769234e-05,
+      "loss": 0.9396,
+      "step": 355
+    },
+    {
+      "epoch": 0.20767099314569054,
+      "grad_norm": 1.3062304258346558,
+      "learning_rate": 4.616887019230769e-05,
+      "loss": 1.1495,
+      "step": 356
+    },
+    {
+      "epoch": 0.20825433863205484,
+      "grad_norm": 1.1252844333648682,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 1.2231,
+      "step": 357
+    },
+    {
+      "epoch": 0.20883768411841913,
+      "grad_norm": 1.3172509670257568,
+      "learning_rate": 4.613882211538462e-05,
+      "loss": 1.181,
+      "step": 358
+    },
+    {
+      "epoch": 0.20942102960478343,
+      "grad_norm": 1.2648638486862183,
+      "learning_rate": 4.612379807692308e-05,
+      "loss": 1.2795,
+      "step": 359
+    },
+    {
+      "epoch": 0.21000437509114772,
+      "grad_norm": 1.266517162322998,
+      "learning_rate": 4.610877403846154e-05,
+      "loss": 1.1233,
+      "step": 360
+    },
+    {
+      "epoch": 0.21058772057751204,
+      "grad_norm": 1.5995312929153442,
+      "learning_rate": 4.609375e-05,
+      "loss": 0.9868,
+      "step": 361
+    },
+    {
+      "epoch": 0.21117106606387634,
+      "grad_norm": 1.338070034980774,
+      "learning_rate": 4.6078725961538464e-05,
+      "loss": 1.0835,
+      "step": 362
+    },
+    {
+      "epoch": 0.21175441155024063,
+      "grad_norm": 1.2133007049560547,
+      "learning_rate": 4.606370192307692e-05,
+      "loss": 1.0838,
+      "step": 363
+    },
+    {
+      "epoch": 0.21233775703660493,
+      "grad_norm": 1.2662042379379272,
+      "learning_rate": 4.604867788461539e-05,
+      "loss": 1.0508,
+      "step": 364
+    },
+    {
+      "epoch": 0.21292110252296922,
+      "grad_norm": 1.1545066833496094,
+      "learning_rate": 4.6033653846153846e-05,
+      "loss": 1.0582,
+      "step": 365
+    },
+    {
+      "epoch": 0.21350444800933352,
+      "grad_norm": 1.2459946870803833,
+      "learning_rate": 4.601862980769231e-05,
+      "loss": 1.0589,
+      "step": 366
+    },
+    {
+      "epoch": 0.21408779349569781,
+      "grad_norm": 1.1584820747375488,
+      "learning_rate": 4.600360576923077e-05,
+      "loss": 0.9987,
+      "step": 367
+    },
+    {
+      "epoch": 0.21467113898206214,
+      "grad_norm": 1.049363136291504,
+      "learning_rate": 4.598858173076923e-05,
+      "loss": 1.0067,
+      "step": 368
+    },
+    {
+      "epoch": 0.21525448446842643,
+      "grad_norm": 1.3197144269943237,
+      "learning_rate": 4.5973557692307694e-05,
+      "loss": 1.1545,
+      "step": 369
+    },
+    {
+      "epoch": 0.21583782995479073,
+      "grad_norm": 1.3010430335998535,
+      "learning_rate": 4.595853365384616e-05,
+      "loss": 1.0185,
+      "step": 370
+    },
+    {
+      "epoch": 0.21642117544115502,
+      "grad_norm": 1.2531121969223022,
+      "learning_rate": 4.594350961538462e-05,
+      "loss": 1.1825,
+      "step": 371
+    },
+    {
+      "epoch": 0.21700452092751932,
+      "grad_norm": 1.7184710502624512,
+      "learning_rate": 4.592848557692308e-05,
+      "loss": 1.0196,
+      "step": 372
+    },
+    {
+      "epoch": 0.2175878664138836,
+      "grad_norm": 1.173525094985962,
+      "learning_rate": 4.591346153846154e-05,
+      "loss": 1.2479,
+      "step": 373
+    },
+    {
+      "epoch": 0.21817121190024794,
+      "grad_norm": 1.450506567955017,
+      "learning_rate": 4.58984375e-05,
+      "loss": 1.1017,
+      "step": 374
+    },
+    {
+      "epoch": 0.21875455738661223,
+      "grad_norm": 1.1033124923706055,
+      "learning_rate": 4.5883413461538465e-05,
+      "loss": 0.8721,
+      "step": 375
+    },
+    {
+      "epoch": 0.21933790287297653,
+      "grad_norm": 1.3852168321609497,
+      "learning_rate": 4.5868389423076924e-05,
+      "loss": 1.0496,
+      "step": 376
+    },
+    {
+      "epoch": 0.21992124835934082,
+      "grad_norm": 1.2315133810043335,
+      "learning_rate": 4.585336538461539e-05,
+      "loss": 1.2742,
+      "step": 377
+    },
+    {
+      "epoch": 0.22050459384570512,
+      "grad_norm": 1.1665183305740356,
+      "learning_rate": 4.583834134615385e-05,
+      "loss": 0.8682,
+      "step": 378
+    },
+    {
+      "epoch": 0.2210879393320694,
+      "grad_norm": 1.1642358303070068,
+      "learning_rate": 4.5823317307692306e-05,
+      "loss": 1.0468,
+      "step": 379
+    },
+    {
+      "epoch": 0.2216712848184337,
+      "grad_norm": 1.1943094730377197,
+      "learning_rate": 4.580829326923077e-05,
+      "loss": 0.9771,
+      "step": 380
+    },
+    {
+      "epoch": 0.22225463030479803,
+      "grad_norm": 1.4384686946868896,
+      "learning_rate": 4.579326923076923e-05,
+      "loss": 1.1928,
+      "step": 381
+    },
+    {
+      "epoch": 0.22283797579116232,
+      "grad_norm": 2.045624256134033,
+      "learning_rate": 4.5778245192307695e-05,
+      "loss": 1.1249,
+      "step": 382
+    },
+    {
+      "epoch": 0.22342132127752662,
+      "grad_norm": 1.4147062301635742,
+      "learning_rate": 4.576322115384616e-05,
+      "loss": 1.0644,
+      "step": 383
+    },
+    {
+      "epoch": 0.22400466676389091,
+      "grad_norm": 1.4192973375320435,
+      "learning_rate": 4.574819711538462e-05,
+      "loss": 1.0954,
+      "step": 384
+    },
+    {
+      "epoch": 0.2245880122502552,
+      "grad_norm": 2.5359246730804443,
+      "learning_rate": 4.573317307692308e-05,
+      "loss": 0.9555,
+      "step": 385
+    },
+    {
+      "epoch": 0.2251713577366195,
+      "grad_norm": 1.3057085275650024,
+      "learning_rate": 4.5718149038461536e-05,
+      "loss": 1.2688,
+      "step": 386
+    },
+    {
+      "epoch": 0.2257547032229838,
+      "grad_norm": 1.5768152475357056,
+      "learning_rate": 4.5703125e-05,
+      "loss": 1.0067,
+      "step": 387
+    },
+    {
+      "epoch": 0.22633804870934812,
+      "grad_norm": 1.3447943925857544,
+      "learning_rate": 4.5688100961538467e-05,
+      "loss": 1.1379,
+      "step": 388
+    },
+    {
+      "epoch": 0.22692139419571242,
+      "grad_norm": 1.4754528999328613,
+      "learning_rate": 4.5673076923076925e-05,
+      "loss": 0.8801,
+      "step": 389
+    },
+    {
+      "epoch": 0.2275047396820767,
+      "grad_norm": 1.4515520334243774,
+      "learning_rate": 4.565805288461539e-05,
+      "loss": 1.0797,
+      "step": 390
+    },
+    {
+      "epoch": 0.228088085168441,
+      "grad_norm": 1.317018747329712,
+      "learning_rate": 4.564302884615385e-05,
+      "loss": 1.0002,
+      "step": 391
+    },
+    {
+      "epoch": 0.2286714306548053,
+      "grad_norm": 1.7866982221603394,
+      "learning_rate": 4.562800480769231e-05,
+      "loss": 1.0067,
+      "step": 392
+    },
+    {
+      "epoch": 0.2292547761411696,
+      "grad_norm": 1.3489786386489868,
+      "learning_rate": 4.561298076923077e-05,
+      "loss": 1.0879,
+      "step": 393
+    },
+    {
+      "epoch": 0.22983812162753392,
+      "grad_norm": 1.4815866947174072,
+      "learning_rate": 4.559795673076923e-05,
+      "loss": 1.1511,
+      "step": 394
+    },
+    {
+      "epoch": 0.23042146711389822,
+      "grad_norm": 1.099830985069275,
+      "learning_rate": 4.5582932692307696e-05,
+      "loss": 1.2723,
+      "step": 395
+    },
+    {
+      "epoch": 0.2310048126002625,
+      "grad_norm": 1.2879087924957275,
+      "learning_rate": 4.556790865384616e-05,
+      "loss": 1.0579,
+      "step": 396
+    },
+    {
+      "epoch": 0.2315881580866268,
+      "grad_norm": 1.2693564891815186,
+      "learning_rate": 4.5552884615384613e-05,
+      "loss": 1.2117,
+      "step": 397
+    },
+    {
+      "epoch": 0.2321715035729911,
+      "grad_norm": 1.5512031316757202,
+      "learning_rate": 4.553786057692308e-05,
+      "loss": 0.9932,
+      "step": 398
+    },
+    {
+      "epoch": 0.2327548490593554,
+      "grad_norm": 2.0312271118164062,
+      "learning_rate": 4.552283653846154e-05,
+      "loss": 0.8886,
+      "step": 399
+    },
+    {
+      "epoch": 0.2333381945457197,
+      "grad_norm": 1.1745553016662598,
+      "learning_rate": 4.55078125e-05,
+      "loss": 1.1362,
+      "step": 400
+    },
+    {
+      "epoch": 0.2333381945457197,
+      "eval_loss_squad": 1.0443613978661597,
+      "eval_perplexity": 7.789854227908129,
+      "eval_perplexity_reconstruct": 1.9117095009373144,
+      "step": 400
+    },
+    {
+      "epoch": 0.233921540032084,
+      "grad_norm": 1.2545403242111206,
+      "learning_rate": 4.549278846153847e-05,
+      "loss": 1.1109,
+      "step": 401
+    },
+    {
+      "epoch": 0.2345048855184483,
+      "grad_norm": 1.2915353775024414,
+      "learning_rate": 4.5477764423076926e-05,
+      "loss": 0.9874,
+      "step": 402
+    },
+    {
+      "epoch": 0.2350882310048126,
+      "grad_norm": 1.3028852939605713,
+      "learning_rate": 4.5462740384615385e-05,
+      "loss": 0.9613,
+      "step": 403
+    },
+    {
+      "epoch": 0.2356715764911769,
+      "grad_norm": 1.5871734619140625,
+      "learning_rate": 4.544771634615384e-05,
+      "loss": 1.0638,
+      "step": 404
+    },
+    {
+      "epoch": 0.2362549219775412,
+      "grad_norm": 1.4773112535476685,
+      "learning_rate": 4.543269230769231e-05,
+      "loss": 0.9079,
+      "step": 405
+    },
+    {
+      "epoch": 0.2368382674639055,
+      "grad_norm": 1.2807551622390747,
+      "learning_rate": 4.5417668269230774e-05,
+      "loss": 1.0703,
+      "step": 406
+    },
+    {
+      "epoch": 0.23742161295026978,
+      "grad_norm": 1.3610466718673706,
+      "learning_rate": 4.540264423076923e-05,
+      "loss": 0.962,
+      "step": 407
+    },
+    {
+      "epoch": 0.2380049584366341,
+      "grad_norm": 1.3228495121002197,
+      "learning_rate": 4.53876201923077e-05,
+      "loss": 1.0652,
+      "step": 408
+    },
+    {
+      "epoch": 0.2385883039229984,
+      "grad_norm": 1.153469443321228,
+      "learning_rate": 4.5372596153846156e-05,
+      "loss": 1.2528,
+      "step": 409
+    },
+    {
+      "epoch": 0.2391716494093627,
+      "grad_norm": 1.5087329149246216,
+      "learning_rate": 4.5357572115384615e-05,
+      "loss": 0.9128,
+      "step": 410
+    },
+    {
+      "epoch": 0.239754994895727,
+      "grad_norm": 1.523759365081787,
+      "learning_rate": 4.534254807692308e-05,
+      "loss": 1.2005,
+      "step": 411
+    },
+    {
+      "epoch": 0.2403383403820913,
+      "grad_norm": 1.2065186500549316,
+      "learning_rate": 4.532752403846154e-05,
+      "loss": 0.9159,
+      "step": 412
+    },
+    {
+      "epoch": 0.24092168586845558,
+      "grad_norm": 1.2266783714294434,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 1.1309,
+      "step": 413
+    },
+    {
+      "epoch": 0.2415050313548199,
+      "grad_norm": 1.5053402185440063,
+      "learning_rate": 4.529747596153847e-05,
+      "loss": 1.1071,
+      "step": 414
+    },
+    {
+      "epoch": 0.2420883768411842,
+      "grad_norm": 1.978948712348938,
+      "learning_rate": 4.528245192307692e-05,
+      "loss": 1.1895,
+      "step": 415
+    },
+    {
+      "epoch": 0.2426717223275485,
+      "grad_norm": 1.3853555917739868,
+      "learning_rate": 4.5267427884615386e-05,
+      "loss": 1.0861,
+      "step": 416
+    },
+    {
+      "epoch": 0.2432550678139128,
+      "grad_norm": 1.2731225490570068,
+      "learning_rate": 4.5252403846153844e-05,
+      "loss": 0.8939,
+      "step": 417
+    },
+    {
+      "epoch": 0.24383841330027708,
+      "grad_norm": 1.1784796714782715,
+      "learning_rate": 4.523737980769231e-05,
+      "loss": 1.1628,
+      "step": 418
+    },
+    {
+      "epoch": 0.24442175878664138,
+      "grad_norm": 1.4403506517410278,
+      "learning_rate": 4.5222355769230775e-05,
+      "loss": 1.2741,
+      "step": 419
+    },
+    {
+      "epoch": 0.24500510427300567,
+      "grad_norm": 1.3002524375915527,
+      "learning_rate": 4.5207331730769233e-05,
+      "loss": 1.0343,
+      "step": 420
+    },
+    {
+      "epoch": 0.24558844975937,
+      "grad_norm": 1.323982834815979,
+      "learning_rate": 4.519230769230769e-05,
+      "loss": 1.0944,
+      "step": 421
+    },
+    {
+      "epoch": 0.2461717952457343,
+      "grad_norm": 1.2002545595169067,
+      "learning_rate": 4.517728365384616e-05,
+      "loss": 1.0209,
+      "step": 422
+    },
+    {
+      "epoch": 0.2467551407320986,
+      "grad_norm": 1.1465034484863281,
+      "learning_rate": 4.5162259615384616e-05,
+      "loss": 1.0743,
+      "step": 423
+    },
+    {
+      "epoch": 0.24733848621846288,
+      "grad_norm": 1.2983150482177734,
+      "learning_rate": 4.514723557692308e-05,
+      "loss": 1.1408,
+      "step": 424
+    },
+    {
+      "epoch": 0.24792183170482718,
+      "grad_norm": 1.2283337116241455,
+      "learning_rate": 4.513221153846154e-05,
+      "loss": 0.8972,
+      "step": 425
+    },
+    {
+      "epoch": 0.24850517719119147,
+      "grad_norm": 1.3164033889770508,
+      "learning_rate": 4.5117187500000005e-05,
+      "loss": 1.2549,
+      "step": 426
+    },
+    {
+      "epoch": 0.2490885226775558,
+      "grad_norm": 1.2797776460647583,
+      "learning_rate": 4.510216346153846e-05,
+      "loss": 0.9427,
+      "step": 427
+    },
+    {
+      "epoch": 0.2496718681639201,
+      "grad_norm": 1.0905731916427612,
+      "learning_rate": 4.508713942307692e-05,
+      "loss": 1.127,
+      "step": 428
+    },
+    {
+      "epoch": 0.25025521365028436,
+      "grad_norm": 1.4118478298187256,
+      "learning_rate": 4.507211538461539e-05,
+      "loss": 0.9869,
+      "step": 429
+    },
+    {
+      "epoch": 0.25083855913664865,
+      "grad_norm": 1.125809907913208,
+      "learning_rate": 4.5057091346153846e-05,
+      "loss": 1.0308,
+      "step": 430
+    },
+    {
+      "epoch": 0.251421904623013,
+      "grad_norm": 1.1028636693954468,
+      "learning_rate": 4.504206730769231e-05,
+      "loss": 1.0587,
+      "step": 431
+    },
+    {
+      "epoch": 0.2520052501093773,
+      "grad_norm": 1.59440279006958,
+      "learning_rate": 4.5027043269230776e-05,
+      "loss": 1.0767,
+      "step": 432
+    },
+    {
+      "epoch": 0.2525885955957416,
+      "grad_norm": 1.067054033279419,
+      "learning_rate": 4.501201923076923e-05,
+      "loss": 1.1787,
+      "step": 433
+    },
+    {
+      "epoch": 0.2531719410821059,
+      "grad_norm": 1.1260448694229126,
+      "learning_rate": 4.499699519230769e-05,
+      "loss": 1.0111,
+      "step": 434
+    },
+    {
+      "epoch": 0.2537552865684702,
+      "grad_norm": 1.079969048500061,
+      "learning_rate": 4.498197115384616e-05,
+      "loss": 1.0204,
+      "step": 435
+    },
+    {
+      "epoch": 0.2543386320548345,
+      "grad_norm": 1.1547199487686157,
+      "learning_rate": 4.496694711538462e-05,
+      "loss": 0.9295,
+      "step": 436
+    },
+    {
+      "epoch": 0.2549219775411988,
+      "grad_norm": 1.3175557851791382,
+      "learning_rate": 4.495192307692308e-05,
+      "loss": 0.9133,
+      "step": 437
+    },
+    {
+      "epoch": 0.25550532302756307,
+      "grad_norm": 1.323306918144226,
+      "learning_rate": 4.493689903846154e-05,
+      "loss": 0.9118,
+      "step": 438
+    },
+    {
+      "epoch": 0.25608866851392736,
+      "grad_norm": 1.2758287191390991,
+      "learning_rate": 4.4921875e-05,
+      "loss": 1.2725,
+      "step": 439
+    },
+    {
+      "epoch": 0.25667201400029166,
+      "grad_norm": 1.7028545141220093,
+      "learning_rate": 4.4906850961538465e-05,
+      "loss": 1.088,
+      "step": 440
+    },
+    {
+      "epoch": 0.25725535948665595,
+      "grad_norm": 1.8174322843551636,
+      "learning_rate": 4.489182692307692e-05,
+      "loss": 1.1273,
+      "step": 441
+    },
+    {
+      "epoch": 0.25783870497302025,
+      "grad_norm": 1.1969882249832153,
+      "learning_rate": 4.487680288461539e-05,
+      "loss": 1.2244,
+      "step": 442
+    },
+    {
+      "epoch": 0.25842205045938454,
+      "grad_norm": 0.9966511726379395,
+      "learning_rate": 4.486177884615385e-05,
+      "loss": 1.1073,
+      "step": 443
+    },
+    {
+      "epoch": 0.2590053959457489,
+      "grad_norm": 1.1165422201156616,
+      "learning_rate": 4.484675480769231e-05,
+      "loss": 0.9374,
+      "step": 444
+    },
+    {
+      "epoch": 0.2595887414321132,
+      "grad_norm": 1.374568223953247,
+      "learning_rate": 4.483173076923077e-05,
+      "loss": 1.0658,
+      "step": 445
+    },
+    {
+      "epoch": 0.2601720869184775,
+      "grad_norm": 1.2743726968765259,
+      "learning_rate": 4.481670673076923e-05,
+      "loss": 1.1376,
+      "step": 446
+    },
+    {
+      "epoch": 0.2607554324048418,
+      "grad_norm": 1.2559353113174438,
+      "learning_rate": 4.4801682692307694e-05,
+      "loss": 0.9594,
+      "step": 447
+    },
+    {
+      "epoch": 0.2613387778912061,
+      "grad_norm": 1.220955491065979,
+      "learning_rate": 4.478665865384616e-05,
+      "loss": 1.0246,
+      "step": 448
+    },
+    {
+      "epoch": 0.26192212337757037,
+      "grad_norm": 1.0389074087142944,
+      "learning_rate": 4.477163461538462e-05,
+      "loss": 0.973,
+      "step": 449
+    },
+    {
+      "epoch": 0.26250546886393467,
+      "grad_norm": 1.1613105535507202,
+      "learning_rate": 4.4756610576923083e-05,
+      "loss": 1.1086,
+      "step": 450
+    },
+    {
+      "epoch": 0.26308881435029896,
+      "grad_norm": 1.147077202796936,
+      "learning_rate": 4.4741586538461535e-05,
+      "loss": 1.2115,
+      "step": 451
+    },
+    {
+      "epoch": 0.26367215983666326,
+      "grad_norm": 1.1235055923461914,
+      "learning_rate": 4.47265625e-05,
+      "loss": 1.1007,
+      "step": 452
+    },
+    {
+      "epoch": 0.26425550532302755,
+      "grad_norm": 2.1269114017486572,
+      "learning_rate": 4.4711538461538466e-05,
+      "loss": 1.1473,
+      "step": 453
+    },
+    {
+      "epoch": 0.26483885080939185,
+      "grad_norm": 1.0030604600906372,
+      "learning_rate": 4.4696514423076924e-05,
+      "loss": 0.9971,
+      "step": 454
+    },
+    {
+      "epoch": 0.26542219629575614,
+      "grad_norm": 1.013385534286499,
+      "learning_rate": 4.468149038461539e-05,
+      "loss": 0.9949,
+      "step": 455
+    },
+    {
+      "epoch": 0.26600554178212044,
+      "grad_norm": 1.333153486251831,
+      "learning_rate": 4.466646634615385e-05,
+      "loss": 0.9749,
+      "step": 456
+    },
+    {
+      "epoch": 0.2665888872684848,
+      "grad_norm": 1.3025758266448975,
+      "learning_rate": 4.4651442307692306e-05,
+      "loss": 0.961,
+      "step": 457
+    },
+    {
+      "epoch": 0.2671722327548491,
+      "grad_norm": 1.1991068124771118,
+      "learning_rate": 4.463641826923077e-05,
+      "loss": 0.9674,
+      "step": 458
+    },
+    {
+      "epoch": 0.2677555782412134,
+      "grad_norm": 1.5575584173202515,
+      "learning_rate": 4.462139423076923e-05,
+      "loss": 0.9976,
+      "step": 459
+    },
+    {
+      "epoch": 0.26833892372757767,
+      "grad_norm": 1.234240174293518,
+      "learning_rate": 4.4606370192307696e-05,
+      "loss": 0.969,
+      "step": 460
+    },
+    {
+      "epoch": 0.26892226921394197,
+      "grad_norm": 1.2866865396499634,
+      "learning_rate": 4.459134615384616e-05,
+      "loss": 0.7539,
+      "step": 461
+    },
+    {
+      "epoch": 0.26950561470030626,
+      "grad_norm": 1.3814231157302856,
+      "learning_rate": 4.457632211538462e-05,
+      "loss": 1.014,
+      "step": 462
+    },
+    {
+      "epoch": 0.27008896018667056,
+      "grad_norm": 1.2410728931427002,
+      "learning_rate": 4.456129807692308e-05,
+      "loss": 1.1885,
+      "step": 463
+    },
+    {
+      "epoch": 0.27067230567303485,
+      "grad_norm": 1.1972429752349854,
+      "learning_rate": 4.4546274038461536e-05,
+      "loss": 1.1027,
+      "step": 464
+    },
+    {
+      "epoch": 0.27125565115939915,
+      "grad_norm": 1.2238330841064453,
+      "learning_rate": 4.453125e-05,
+      "loss": 0.9025,
+      "step": 465
+    },
+    {
+      "epoch": 0.27183899664576344,
+      "grad_norm": 1.2489745616912842,
+      "learning_rate": 4.451622596153847e-05,
+      "loss": 1.0432,
+      "step": 466
+    },
+    {
+      "epoch": 0.27242234213212774,
+      "grad_norm": 1.2359799146652222,
+      "learning_rate": 4.4501201923076925e-05,
+      "loss": 1.0075,
+      "step": 467
+    },
+    {
+      "epoch": 0.27300568761849203,
+      "grad_norm": 1.4064029455184937,
+      "learning_rate": 4.448617788461539e-05,
+      "loss": 1.1092,
+      "step": 468
+    },
+    {
+      "epoch": 0.2735890331048563,
+      "grad_norm": 1.2376477718353271,
+      "learning_rate": 4.447115384615384e-05,
+      "loss": 1.0766,
+      "step": 469
+    },
+    {
+      "epoch": 0.2741723785912207,
+      "grad_norm": 1.1279197931289673,
+      "learning_rate": 4.445612980769231e-05,
+      "loss": 1.1218,
+      "step": 470
+    },
+    {
+      "epoch": 0.274755724077585,
+      "grad_norm": 1.5163652896881104,
+      "learning_rate": 4.444110576923077e-05,
+      "loss": 0.9557,
+      "step": 471
+    },
+    {
+      "epoch": 0.27533906956394927,
+      "grad_norm": 1.198194146156311,
+      "learning_rate": 4.442608173076923e-05,
+      "loss": 0.8365,
+      "step": 472
+    },
+    {
+      "epoch": 0.27592241505031356,
+      "grad_norm": 1.205476999282837,
+      "learning_rate": 4.44110576923077e-05,
+      "loss": 0.9057,
+      "step": 473
+    },
+    {
+      "epoch": 0.27650576053667786,
+      "grad_norm": 1.0138362646102905,
+      "learning_rate": 4.4396033653846155e-05,
+      "loss": 1.0689,
+      "step": 474
+    },
+    {
+      "epoch": 0.27708910602304215,
+      "grad_norm": 1.353697657585144,
+      "learning_rate": 4.4381009615384614e-05,
+      "loss": 0.9925,
+      "step": 475
+    },
+    {
+      "epoch": 0.27767245150940645,
+      "grad_norm": 1.3365930318832397,
+      "learning_rate": 4.436598557692308e-05,
+      "loss": 0.8991,
+      "step": 476
+    },
+    {
+      "epoch": 0.27825579699577074,
+      "grad_norm": 1.3026283979415894,
+      "learning_rate": 4.435096153846154e-05,
+      "loss": 1.1419,
+      "step": 477
+    },
+    {
+      "epoch": 0.27883914248213504,
+      "grad_norm": 1.311883807182312,
+      "learning_rate": 4.43359375e-05,
+      "loss": 0.9451,
+      "step": 478
+    },
+    {
+      "epoch": 0.27942248796849933,
+      "grad_norm": 0.950835108757019,
+      "learning_rate": 4.432091346153847e-05,
+      "loss": 0.8928,
+      "step": 479
+    },
+    {
+      "epoch": 0.28000583345486363,
+      "grad_norm": 1.2371882200241089,
+      "learning_rate": 4.4305889423076927e-05,
+      "loss": 1.076,
+      "step": 480
+    },
+    {
+      "epoch": 0.2805891789412279,
+      "grad_norm": 1.6330941915512085,
+      "learning_rate": 4.4290865384615385e-05,
+      "loss": 1.2121,
+      "step": 481
+    },
+    {
+      "epoch": 0.2811725244275922,
+      "grad_norm": 1.3535494804382324,
+      "learning_rate": 4.4275841346153844e-05,
+      "loss": 1.1427,
+      "step": 482
+    },
+    {
+      "epoch": 0.2817558699139565,
+      "grad_norm": 1.4552642107009888,
+      "learning_rate": 4.426081730769231e-05,
+      "loss": 1.0305,
+      "step": 483
+    },
+    {
+      "epoch": 0.28233921540032086,
+      "grad_norm": 0.9442708492279053,
+      "learning_rate": 4.4245793269230774e-05,
+      "loss": 1.0597,
+      "step": 484
+    },
+    {
+      "epoch": 0.28292256088668516,
+      "grad_norm": 1.0428327322006226,
+      "learning_rate": 4.423076923076923e-05,
+      "loss": 0.9091,
+      "step": 485
+    },
+    {
+      "epoch": 0.28350590637304945,
+      "grad_norm": 1.2180997133255005,
+      "learning_rate": 4.42157451923077e-05,
+      "loss": 0.9809,
+      "step": 486
+    },
+    {
+      "epoch": 0.28408925185941375,
+      "grad_norm": 1.153455138206482,
+      "learning_rate": 4.4200721153846156e-05,
+      "loss": 1.0152,
+      "step": 487
+    },
+    {
+      "epoch": 0.28467259734577804,
+      "grad_norm": 1.2968738079071045,
+      "learning_rate": 4.4185697115384615e-05,
+      "loss": 1.0907,
+      "step": 488
+    },
+    {
+      "epoch": 0.28525594283214234,
+      "grad_norm": 1.2618736028671265,
+      "learning_rate": 4.417067307692308e-05,
+      "loss": 0.9007,
+      "step": 489
+    },
+    {
+      "epoch": 0.28583928831850663,
+      "grad_norm": 1.4131522178649902,
+      "learning_rate": 4.415564903846154e-05,
+      "loss": 1.0028,
+      "step": 490
+    },
+    {
+      "epoch": 0.28642263380487093,
+      "grad_norm": 1.2588627338409424,
+      "learning_rate": 4.4140625000000004e-05,
+      "loss": 1.1391,
+      "step": 491
+    },
+    {
+      "epoch": 0.2870059792912352,
+      "grad_norm": 1.4097051620483398,
+      "learning_rate": 4.412560096153846e-05,
+      "loss": 1.1168,
+      "step": 492
+    },
+    {
+      "epoch": 0.2875893247775995,
+      "grad_norm": 0.9398725032806396,
+      "learning_rate": 4.411057692307692e-05,
+      "loss": 0.961,
+      "step": 493
+    },
+    {
+      "epoch": 0.2881726702639638,
+      "grad_norm": 1.3721821308135986,
+      "learning_rate": 4.4095552884615386e-05,
+      "loss": 0.8813,
+      "step": 494
+    },
+    {
+      "epoch": 0.2887560157503281,
+      "grad_norm": 1.4227555990219116,
+      "learning_rate": 4.4080528846153845e-05,
+      "loss": 1.2802,
+      "step": 495
+    },
+    {
+      "epoch": 0.2893393612366924,
+      "grad_norm": 1.5191651582717896,
+      "learning_rate": 4.406550480769231e-05,
+      "loss": 1.1544,
+      "step": 496
+    },
+    {
+      "epoch": 0.28992270672305676,
+      "grad_norm": 1.350059986114502,
+      "learning_rate": 4.4050480769230775e-05,
+      "loss": 1.1791,
+      "step": 497
+    },
+    {
+      "epoch": 0.29050605220942105,
+      "grad_norm": 1.2704778909683228,
+      "learning_rate": 4.4035456730769234e-05,
+      "loss": 1.0238,
+      "step": 498
+    },
+    {
+      "epoch": 0.29108939769578535,
+      "grad_norm": 1.3052599430084229,
+      "learning_rate": 4.402043269230769e-05,
+      "loss": 0.8804,
+      "step": 499
+    },
+    {
+      "epoch": 0.29167274318214964,
+      "grad_norm": 1.4801050424575806,
+      "learning_rate": 4.400540865384616e-05,
+      "loss": 1.0356,
+      "step": 500
+    },
+    {
+      "epoch": 0.29225608866851394,
+      "grad_norm": 1.1770102977752686,
+      "learning_rate": 4.3990384615384616e-05,
+      "loss": 0.9618,
+      "step": 501
+    },
+    {
+      "epoch": 0.29283943415487823,
+      "grad_norm": 1.216208577156067,
+      "learning_rate": 4.397536057692308e-05,
+      "loss": 1.1352,
+      "step": 502
+    },
+    {
+      "epoch": 0.2934227796412425,
+      "grad_norm": 1.1072677373886108,
+      "learning_rate": 4.396033653846154e-05,
+      "loss": 1.2635,
+      "step": 503
+    },
+    {
+      "epoch": 0.2940061251276068,
+      "grad_norm": 1.2480254173278809,
+      "learning_rate": 4.3945312500000005e-05,
+      "loss": 1.0547,
+      "step": 504
+    },
+    {
+      "epoch": 0.2945894706139711,
+      "grad_norm": 1.227055549621582,
+      "learning_rate": 4.3930288461538464e-05,
+      "loss": 1.1527,
+      "step": 505
+    },
+    {
+      "epoch": 0.2951728161003354,
+      "grad_norm": 1.1896963119506836,
+      "learning_rate": 4.391526442307692e-05,
+      "loss": 1.0816,
+      "step": 506
+    },
+    {
+      "epoch": 0.2957561615866997,
+      "grad_norm": 1.480958342552185,
+      "learning_rate": 4.390024038461539e-05,
+      "loss": 1.0701,
+      "step": 507
+    },
+    {
+      "epoch": 0.296339507073064,
+      "grad_norm": 1.4093650579452515,
+      "learning_rate": 4.3885216346153846e-05,
+      "loss": 1.1059,
+      "step": 508
+    },
+    {
+      "epoch": 0.2969228525594283,
+      "grad_norm": 1.1637531518936157,
+      "learning_rate": 4.387019230769231e-05,
+      "loss": 1.1347,
+      "step": 509
+    },
+    {
+      "epoch": 0.29750619804579265,
+      "grad_norm": 1.185320496559143,
+      "learning_rate": 4.385516826923077e-05,
+      "loss": 1.0149,
+      "step": 510
+    },
+    {
+      "epoch": 0.29808954353215694,
+      "grad_norm": 1.3317254781723022,
+      "learning_rate": 4.384014423076923e-05,
+      "loss": 0.9834,
+      "step": 511
+    },
+    {
+      "epoch": 0.29867288901852124,
+      "grad_norm": 1.0526453256607056,
+      "learning_rate": 4.3825120192307693e-05,
+      "loss": 0.9779,
+      "step": 512
+    },
+    {
+      "epoch": 0.29925623450488553,
+      "grad_norm": 1.2846801280975342,
+      "learning_rate": 4.381009615384616e-05,
+      "loss": 1.1655,
+      "step": 513
+    },
+    {
+      "epoch": 0.2998395799912498,
+      "grad_norm": 1.4375282526016235,
+      "learning_rate": 4.379507211538462e-05,
+      "loss": 1.1173,
+      "step": 514
+    },
+    {
+      "epoch": 0.3004229254776141,
+      "grad_norm": 1.0759106874465942,
+      "learning_rate": 4.378004807692308e-05,
+      "loss": 0.909,
+      "step": 515
+    },
+    {
+      "epoch": 0.3010062709639784,
+      "grad_norm": 1.262014389038086,
+      "learning_rate": 4.376502403846154e-05,
+      "loss": 1.1499,
+      "step": 516
+    },
+    {
+      "epoch": 0.3015896164503427,
+      "grad_norm": 1.5526944398880005,
+      "learning_rate": 4.375e-05,
+      "loss": 0.975,
+      "step": 517
+    },
+    {
+      "epoch": 0.302172961936707,
+      "grad_norm": 1.2012171745300293,
+      "learning_rate": 4.3734975961538465e-05,
+      "loss": 0.791,
+      "step": 518
+    },
+    {
+      "epoch": 0.3027563074230713,
+      "grad_norm": 1.248779058456421,
+      "learning_rate": 4.371995192307692e-05,
+      "loss": 0.9285,
+      "step": 519
+    },
+    {
+      "epoch": 0.3033396529094356,
+      "grad_norm": 1.387321949005127,
+      "learning_rate": 4.370492788461539e-05,
+      "loss": 1.0779,
+      "step": 520
+    },
+    {
+      "epoch": 0.3039229983957999,
+      "grad_norm": 1.4328728914260864,
+      "learning_rate": 4.368990384615385e-05,
+      "loss": 1.1295,
+      "step": 521
+    },
+    {
+      "epoch": 0.3045063438821642,
+      "grad_norm": 1.231877326965332,
+      "learning_rate": 4.367487980769231e-05,
+      "loss": 0.9938,
+      "step": 522
+    },
+    {
+      "epoch": 0.30508968936852854,
+      "grad_norm": 1.1031498908996582,
+      "learning_rate": 4.365985576923077e-05,
+      "loss": 1.0653,
+      "step": 523
+    },
+    {
+      "epoch": 0.30567303485489283,
+      "grad_norm": 1.112518310546875,
+      "learning_rate": 4.364483173076923e-05,
+      "loss": 1.033,
+      "step": 524
+    },
+    {
+      "epoch": 0.30625638034125713,
+      "grad_norm": 1.2784297466278076,
+      "learning_rate": 4.3629807692307695e-05,
+      "loss": 1.026,
+      "step": 525
+    },
+    {
+      "epoch": 0.3068397258276214,
+      "grad_norm": 1.298776626586914,
+      "learning_rate": 4.361478365384616e-05,
+      "loss": 0.8997,
+      "step": 526
+    },
+    {
+      "epoch": 0.3074230713139857,
+      "grad_norm": 1.134751319885254,
+      "learning_rate": 4.359975961538462e-05,
+      "loss": 1.2168,
+      "step": 527
+    },
+    {
+      "epoch": 0.30800641680035,
+      "grad_norm": 1.0475013256072998,
+      "learning_rate": 4.358473557692308e-05,
+      "loss": 1.1632,
+      "step": 528
+    },
+    {
+      "epoch": 0.3085897622867143,
+      "grad_norm": 1.1225156784057617,
+      "learning_rate": 4.3569711538461535e-05,
+      "loss": 1.0383,
+      "step": 529
+    },
+    {
+      "epoch": 0.3091731077730786,
+      "grad_norm": 1.3873623609542847,
+      "learning_rate": 4.35546875e-05,
+      "loss": 1.0286,
+      "step": 530
+    },
+    {
+      "epoch": 0.3097564532594429,
+      "grad_norm": 1.9602346420288086,
+      "learning_rate": 4.3539663461538466e-05,
+      "loss": 1.1737,
+      "step": 531
+    },
+    {
+      "epoch": 0.3103397987458072,
+      "grad_norm": 1.5184245109558105,
+      "learning_rate": 4.3524639423076925e-05,
+      "loss": 1.0894,
+      "step": 532
+    },
+    {
+      "epoch": 0.3109231442321715,
+      "grad_norm": 1.3252931833267212,
+      "learning_rate": 4.350961538461539e-05,
+      "loss": 1.0596,
+      "step": 533
+    },
+    {
+      "epoch": 0.3115064897185358,
+      "grad_norm": 1.6967353820800781,
+      "learning_rate": 4.349459134615385e-05,
+      "loss": 1.0257,
+      "step": 534
+    },
+    {
+      "epoch": 0.3120898352049001,
+      "grad_norm": 1.2516505718231201,
+      "learning_rate": 4.347956730769231e-05,
+      "loss": 1.3018,
+      "step": 535
+    },
+    {
+      "epoch": 0.3126731806912644,
+      "grad_norm": 2.050544500350952,
+      "learning_rate": 4.346454326923077e-05,
+      "loss": 1.155,
+      "step": 536
+    },
+    {
+      "epoch": 0.3132565261776287,
+      "grad_norm": 1.6617178916931152,
+      "learning_rate": 4.344951923076923e-05,
+      "loss": 0.9574,
+      "step": 537
+    },
+    {
+      "epoch": 0.313839871663993,
+      "grad_norm": 1.0113744735717773,
+      "learning_rate": 4.3434495192307696e-05,
+      "loss": 1.0248,
+      "step": 538
+    },
+    {
+      "epoch": 0.3144232171503573,
+      "grad_norm": 1.1463007926940918,
+      "learning_rate": 4.341947115384616e-05,
+      "loss": 1.0335,
+      "step": 539
+    },
+    {
+      "epoch": 0.3150065626367216,
+      "grad_norm": 1.0310202836990356,
+      "learning_rate": 4.340444711538462e-05,
+      "loss": 0.8483,
+      "step": 540
+    },
+    {
+      "epoch": 0.3155899081230859,
+      "grad_norm": 1.2262356281280518,
+      "learning_rate": 4.338942307692308e-05,
+      "loss": 1.1332,
+      "step": 541
+    },
+    {
+      "epoch": 0.3161732536094502,
+      "grad_norm": 1.4348859786987305,
+      "learning_rate": 4.337439903846154e-05,
+      "loss": 0.9394,
+      "step": 542
+    },
+    {
+      "epoch": 0.3167565990958145,
+      "grad_norm": 1.4490423202514648,
+      "learning_rate": 4.3359375e-05,
+      "loss": 1.1187,
+      "step": 543
+    },
+    {
+      "epoch": 0.3173399445821788,
+      "grad_norm": 1.3444559574127197,
+      "learning_rate": 4.334435096153847e-05,
+      "loss": 1.0358,
+      "step": 544
+    },
+    {
+      "epoch": 0.3179232900685431,
+      "grad_norm": 1.1789164543151855,
+      "learning_rate": 4.3329326923076926e-05,
+      "loss": 0.9231,
+      "step": 545
+    },
+    {
+      "epoch": 0.3185066355549074,
+      "grad_norm": 1.231382966041565,
+      "learning_rate": 4.3314302884615384e-05,
+      "loss": 1.0525,
+      "step": 546
+    },
+    {
+      "epoch": 0.3190899810412717,
+      "grad_norm": 1.0804648399353027,
+      "learning_rate": 4.329927884615384e-05,
+      "loss": 0.9816,
+      "step": 547
+    },
+    {
+      "epoch": 0.31967332652763597,
+      "grad_norm": 1.2451800107955933,
+      "learning_rate": 4.328425480769231e-05,
+      "loss": 0.8803,
+      "step": 548
+    },
+    {
+      "epoch": 0.32025667201400027,
+      "grad_norm": 1.2045650482177734,
+      "learning_rate": 4.326923076923077e-05,
+      "loss": 1.1134,
+      "step": 549
+    },
+    {
+      "epoch": 0.3208400175003646,
+      "grad_norm": 1.3157379627227783,
+      "learning_rate": 4.325420673076923e-05,
+      "loss": 1.2101,
+      "step": 550
+    },
+    {
+      "epoch": 0.3214233629867289,
+      "grad_norm": 1.2755893468856812,
+      "learning_rate": 4.32391826923077e-05,
+      "loss": 0.9826,
+      "step": 551
+    },
+    {
+      "epoch": 0.3220067084730932,
+      "grad_norm": 1.2792096138000488,
+      "learning_rate": 4.3224158653846156e-05,
+      "loss": 0.8629,
+      "step": 552
+    },
+    {
+      "epoch": 0.3225900539594575,
+      "grad_norm": 1.1155221462249756,
+      "learning_rate": 4.3209134615384614e-05,
+      "loss": 1.1298,
+      "step": 553
+    },
+    {
+      "epoch": 0.3231733994458218,
+      "grad_norm": 1.3425167798995972,
+      "learning_rate": 4.319411057692308e-05,
+      "loss": 1.0914,
+      "step": 554
+    },
+    {
+      "epoch": 0.3237567449321861,
+      "grad_norm": 2.2145228385925293,
+      "learning_rate": 4.317908653846154e-05,
+      "loss": 0.9944,
+      "step": 555
+    },
+    {
+      "epoch": 0.3243400904185504,
+      "grad_norm": 1.3437769412994385,
+      "learning_rate": 4.31640625e-05,
+      "loss": 1.0431,
+      "step": 556
+    },
+    {
+      "epoch": 0.3249234359049147,
+      "grad_norm": 1.253940463066101,
+      "learning_rate": 4.314903846153847e-05,
+      "loss": 0.8568,
+      "step": 557
+    },
+    {
+      "epoch": 0.325506781391279,
+      "grad_norm": 1.6543011665344238,
+      "learning_rate": 4.313401442307693e-05,
+      "loss": 0.9353,
+      "step": 558
+    },
+    {
+      "epoch": 0.32609012687764327,
+      "grad_norm": 1.3182803392410278,
+      "learning_rate": 4.3118990384615385e-05,
+      "loss": 0.8821,
+      "step": 559
+    },
+    {
+      "epoch": 0.32667347236400757,
+      "grad_norm": 1.3994224071502686,
+      "learning_rate": 4.3103966346153844e-05,
+      "loss": 0.9704,
+      "step": 560
+    },
+    {
+      "epoch": 0.32725681785037186,
+      "grad_norm": 1.1596894264221191,
+      "learning_rate": 4.308894230769231e-05,
+      "loss": 1.1954,
+      "step": 561
+    },
+    {
+      "epoch": 0.32784016333673616,
+      "grad_norm": 1.1696090698242188,
+      "learning_rate": 4.3073918269230774e-05,
+      "loss": 0.994,
+      "step": 562
+    },
+    {
+      "epoch": 0.3284235088231005,
+      "grad_norm": 1.0677425861358643,
+      "learning_rate": 4.305889423076923e-05,
+      "loss": 0.9962,
+      "step": 563
+    },
+    {
+      "epoch": 0.3290068543094648,
+      "grad_norm": 1.27321457862854,
+      "learning_rate": 4.304387019230769e-05,
+      "loss": 1.1947,
+      "step": 564
+    },
+    {
+      "epoch": 0.3295901997958291,
+      "grad_norm": 1.2170658111572266,
+      "learning_rate": 4.302884615384616e-05,
+      "loss": 1.2042,
+      "step": 565
+    },
+    {
+      "epoch": 0.3301735452821934,
+      "grad_norm": 1.4280612468719482,
+      "learning_rate": 4.3013822115384615e-05,
+      "loss": 1.3016,
+      "step": 566
+    },
+    {
+      "epoch": 0.3307568907685577,
+      "grad_norm": 1.2479727268218994,
+      "learning_rate": 4.299879807692308e-05,
+      "loss": 1.172,
+      "step": 567
+    },
+    {
+      "epoch": 0.331340236254922,
+      "grad_norm": 1.192344069480896,
+      "learning_rate": 4.298377403846154e-05,
+      "loss": 1.0476,
+      "step": 568
+    },
+    {
+      "epoch": 0.3319235817412863,
+      "grad_norm": 1.4639259576797485,
+      "learning_rate": 4.2968750000000004e-05,
+      "loss": 0.8669,
+      "step": 569
+    },
+    {
+      "epoch": 0.3325069272276506,
+      "grad_norm": 1.4050641059875488,
+      "learning_rate": 4.295372596153846e-05,
+      "loss": 1.1723,
+      "step": 570
+    },
+    {
+      "epoch": 0.33309027271401487,
+      "grad_norm": 1.117199420928955,
+      "learning_rate": 4.293870192307692e-05,
+      "loss": 1.0075,
+      "step": 571
+    },
+    {
+      "epoch": 0.33367361820037916,
+      "grad_norm": 1.2617186307907104,
+      "learning_rate": 4.2923677884615387e-05,
+      "loss": 1.1577,
+      "step": 572
+    },
+    {
+      "epoch": 0.33425696368674346,
+      "grad_norm": 1.269479513168335,
+      "learning_rate": 4.2908653846153845e-05,
+      "loss": 1.1862,
+      "step": 573
+    },
+    {
+      "epoch": 0.33484030917310775,
+      "grad_norm": 1.3484537601470947,
+      "learning_rate": 4.289362980769231e-05,
+      "loss": 1.0458,
+      "step": 574
+    },
+    {
+      "epoch": 0.33542365465947205,
+      "grad_norm": 1.2798513174057007,
+      "learning_rate": 4.2878605769230776e-05,
+      "loss": 1.1412,
+      "step": 575
+    },
+    {
+      "epoch": 0.3360070001458364,
+      "grad_norm": 1.1217023134231567,
+      "learning_rate": 4.2863581730769234e-05,
+      "loss": 0.8785,
+      "step": 576
+    },
+    {
+      "epoch": 0.3365903456322007,
+      "grad_norm": 1.2047514915466309,
+      "learning_rate": 4.284855769230769e-05,
+      "loss": 1.0627,
+      "step": 577
+    },
+    {
+      "epoch": 0.337173691118565,
+      "grad_norm": 1.3945984840393066,
+      "learning_rate": 4.283353365384616e-05,
+      "loss": 1.122,
+      "step": 578
+    },
+    {
+      "epoch": 0.3377570366049293,
+      "grad_norm": 1.466818928718567,
+      "learning_rate": 4.2818509615384616e-05,
+      "loss": 1.0109,
+      "step": 579
+    },
+    {
+      "epoch": 0.3383403820912936,
+      "grad_norm": 1.4969720840454102,
+      "learning_rate": 4.280348557692308e-05,
+      "loss": 1.0855,
+      "step": 580
+    },
+    {
+      "epoch": 0.3389237275776579,
+      "grad_norm": 1.2596534490585327,
+      "learning_rate": 4.278846153846154e-05,
+      "loss": 1.0654,
+      "step": 581
+    },
+    {
+      "epoch": 0.33950707306402217,
+      "grad_norm": 1.1512525081634521,
+      "learning_rate": 4.27734375e-05,
+      "loss": 1.1286,
+      "step": 582
+    },
+    {
+      "epoch": 0.34009041855038646,
+      "grad_norm": 1.2515615224838257,
+      "learning_rate": 4.2758413461538464e-05,
+      "loss": 1.2436,
+      "step": 583
+    },
+    {
+      "epoch": 0.34067376403675076,
+      "grad_norm": 1.2093133926391602,
+      "learning_rate": 4.274338942307692e-05,
+      "loss": 1.0487,
+      "step": 584
+    },
+    {
+      "epoch": 0.34125710952311505,
+      "grad_norm": 1.1588521003723145,
+      "learning_rate": 4.272836538461539e-05,
+      "loss": 1.1469,
+      "step": 585
+    },
+    {
+      "epoch": 0.34184045500947935,
+      "grad_norm": 1.1132664680480957,
+      "learning_rate": 4.2713341346153846e-05,
+      "loss": 1.0547,
+      "step": 586
+    },
+    {
+      "epoch": 0.34242380049584364,
+      "grad_norm": 1.2172892093658447,
+      "learning_rate": 4.269831730769231e-05,
+      "loss": 1.2139,
+      "step": 587
+    },
+    {
+      "epoch": 0.34300714598220794,
+      "grad_norm": 1.605918526649475,
+      "learning_rate": 4.268329326923077e-05,
+      "loss": 0.9379,
+      "step": 588
+    },
+    {
+      "epoch": 0.34359049146857223,
+      "grad_norm": 1.560250163078308,
+      "learning_rate": 4.266826923076923e-05,
+      "loss": 0.992,
+      "step": 589
+    },
+    {
+      "epoch": 0.3441738369549366,
+      "grad_norm": 1.7285302877426147,
+      "learning_rate": 4.2653245192307694e-05,
+      "loss": 0.9653,
+      "step": 590
+    },
+    {
+      "epoch": 0.3447571824413009,
+      "grad_norm": 1.458788514137268,
+      "learning_rate": 4.263822115384616e-05,
+      "loss": 1.1627,
+      "step": 591
+    },
+    {
+      "epoch": 0.3453405279276652,
+      "grad_norm": 1.444248080253601,
+      "learning_rate": 4.262319711538462e-05,
+      "loss": 1.1464,
+      "step": 592
+    },
+    {
+      "epoch": 0.34592387341402947,
+      "grad_norm": 1.3933836221694946,
+      "learning_rate": 4.260817307692308e-05,
+      "loss": 1.0868,
+      "step": 593
+    },
+    {
+      "epoch": 0.34650721890039377,
+      "grad_norm": 1.6488311290740967,
+      "learning_rate": 4.259314903846154e-05,
+      "loss": 1.0009,
+      "step": 594
+    },
+    {
+      "epoch": 0.34709056438675806,
+      "grad_norm": 1.1838133335113525,
+      "learning_rate": 4.2578125e-05,
+      "loss": 0.884,
+      "step": 595
+    },
+    {
+      "epoch": 0.34767390987312236,
+      "grad_norm": 1.3585830926895142,
+      "learning_rate": 4.2563100961538465e-05,
+      "loss": 1.1594,
+      "step": 596
+    },
+    {
+      "epoch": 0.34825725535948665,
+      "grad_norm": 1.4692409038543701,
+      "learning_rate": 4.2548076923076924e-05,
+      "loss": 1.0424,
+      "step": 597
+    },
+    {
+      "epoch": 0.34884060084585095,
+      "grad_norm": 1.2625113725662231,
+      "learning_rate": 4.253305288461539e-05,
+      "loss": 1.171,
+      "step": 598
+    },
+    {
+      "epoch": 0.34942394633221524,
+      "grad_norm": 1.2111209630966187,
+      "learning_rate": 4.251802884615385e-05,
+      "loss": 0.9783,
+      "step": 599
+    },
+    {
+      "epoch": 0.35000729181857954,
+      "grad_norm": 1.224150538444519,
+      "learning_rate": 4.2503004807692306e-05,
+      "loss": 1.1189,
+      "step": 600
+    },
+    {
+      "epoch": 0.35000729181857954,
+      "eval_loss_squad": 1.0769534187577665,
+      "eval_perplexity": 7.957609155257721,
+      "eval_perplexity_reconstruct": 1.950415498039559,
+      "step": 600
+    },
+    {
+      "epoch": 0.35059063730494383,
+      "grad_norm": 2.5112569332122803,
+      "learning_rate": 4.248798076923077e-05,
+      "loss": 1.1328,
+      "step": 601
+    },
+    {
+      "epoch": 0.3511739827913081,
+      "grad_norm": 1.6178394556045532,
+      "learning_rate": 4.247295673076923e-05,
+      "loss": 1.1492,
+      "step": 602
+    },
+    {
+      "epoch": 0.3517573282776725,
+      "grad_norm": 0.9533804059028625,
+      "learning_rate": 4.2457932692307695e-05,
+      "loss": 1.1439,
+      "step": 603
+    },
+    {
+      "epoch": 0.35234067376403677,
+      "grad_norm": 1.2348041534423828,
+      "learning_rate": 4.244290865384616e-05,
+      "loss": 0.8496,
+      "step": 604
+    },
+    {
+      "epoch": 0.35292401925040107,
+      "grad_norm": 1.199015498161316,
+      "learning_rate": 4.242788461538462e-05,
+      "loss": 1.0841,
+      "step": 605
+    },
+    {
+      "epoch": 0.35350736473676536,
+      "grad_norm": 1.264140248298645,
+      "learning_rate": 4.241286057692308e-05,
+      "loss": 0.9882,
+      "step": 606
+    },
+    {
+      "epoch": 0.35409071022312966,
+      "grad_norm": 1.1264489889144897,
+      "learning_rate": 4.2397836538461536e-05,
+      "loss": 0.8837,
+      "step": 607
+    },
+    {
+      "epoch": 0.35467405570949395,
+      "grad_norm": 1.0938150882720947,
+      "learning_rate": 4.23828125e-05,
+      "loss": 0.9441,
+      "step": 608
+    },
+    {
+      "epoch": 0.35525740119585825,
+      "grad_norm": 1.190896987915039,
+      "learning_rate": 4.2367788461538466e-05,
+      "loss": 1.0634,
+      "step": 609
+    },
+    {
+      "epoch": 0.35584074668222254,
+      "grad_norm": 1.4452359676361084,
+      "learning_rate": 4.2352764423076925e-05,
+      "loss": 0.786,
+      "step": 610
+    },
+    {
+      "epoch": 0.35642409216858684,
+      "grad_norm": 1.373305082321167,
+      "learning_rate": 4.233774038461539e-05,
+      "loss": 0.9567,
+      "step": 611
+    },
+    {
+      "epoch": 0.35700743765495113,
+      "grad_norm": 1.3388804197311401,
+      "learning_rate": 4.232271634615385e-05,
+      "loss": 0.9216,
+      "step": 612
+    },
+    {
+      "epoch": 0.3575907831413154,
+      "grad_norm": 2.152236223220825,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 1.1881,
+      "step": 613
+    },
+    {
+      "epoch": 0.3581741286276797,
+      "grad_norm": 1.3865232467651367,
+      "learning_rate": 4.229266826923077e-05,
+      "loss": 1.0252,
+      "step": 614
+    },
+    {
+      "epoch": 0.358757474114044,
+      "grad_norm": 1.3287843465805054,
+      "learning_rate": 4.227764423076923e-05,
+      "loss": 1.1208,
+      "step": 615
+    },
+    {
+      "epoch": 0.35934081960040837,
+      "grad_norm": 1.2160487174987793,
+      "learning_rate": 4.2262620192307696e-05,
+      "loss": 0.7792,
+      "step": 616
+    },
+    {
+      "epoch": 0.35992416508677266,
+      "grad_norm": 1.3507564067840576,
+      "learning_rate": 4.224759615384616e-05,
+      "loss": 0.8473,
+      "step": 617
+    },
+    {
+      "epoch": 0.36050751057313696,
+      "grad_norm": 1.738625168800354,
+      "learning_rate": 4.223257211538461e-05,
+      "loss": 1.016,
+      "step": 618
+    },
+    {
+      "epoch": 0.36109085605950125,
+      "grad_norm": 1.3850781917572021,
+      "learning_rate": 4.221754807692308e-05,
+      "loss": 1.1247,
+      "step": 619
+    },
+    {
+      "epoch": 0.36167420154586555,
+      "grad_norm": 1.1078225374221802,
+      "learning_rate": 4.220252403846154e-05,
+      "loss": 1.0403,
+      "step": 620
+    },
+    {
+      "epoch": 0.36225754703222984,
+      "grad_norm": 1.5114189386367798,
+      "learning_rate": 4.21875e-05,
+      "loss": 1.136,
+      "step": 621
+    },
+    {
+      "epoch": 0.36284089251859414,
+      "grad_norm": 1.1798583269119263,
+      "learning_rate": 4.217247596153847e-05,
+      "loss": 1.0254,
+      "step": 622
+    },
+    {
+      "epoch": 0.36342423800495843,
+      "grad_norm": 1.6477906703948975,
+      "learning_rate": 4.2157451923076926e-05,
+      "loss": 1.0695,
+      "step": 623
+    },
+    {
+      "epoch": 0.36400758349132273,
+      "grad_norm": 1.1232982873916626,
+      "learning_rate": 4.2142427884615385e-05,
+      "loss": 0.7686,
+      "step": 624
+    },
+    {
+      "epoch": 0.364590928977687,
+      "grad_norm": 1.483515739440918,
+      "learning_rate": 4.212740384615384e-05,
+      "loss": 1.0878,
+      "step": 625
+    },
+    {
+      "epoch": 0.3651742744640513,
+      "grad_norm": 1.361569881439209,
+      "learning_rate": 4.211237980769231e-05,
+      "loss": 1.0778,
+      "step": 626
+    },
+    {
+      "epoch": 0.3657576199504156,
+      "grad_norm": 1.3304964303970337,
+      "learning_rate": 4.2097355769230774e-05,
+      "loss": 1.2297,
+      "step": 627
+    },
+    {
+      "epoch": 0.3663409654367799,
+      "grad_norm": 1.2258186340332031,
+      "learning_rate": 4.208233173076923e-05,
+      "loss": 1.0869,
+      "step": 628
+    },
+    {
+      "epoch": 0.36692431092314426,
+      "grad_norm": 2.0285935401916504,
+      "learning_rate": 4.20673076923077e-05,
+      "loss": 1.0043,
+      "step": 629
+    },
+    {
+      "epoch": 0.36750765640950855,
+      "grad_norm": 1.278552770614624,
+      "learning_rate": 4.2052283653846156e-05,
+      "loss": 0.9999,
+      "step": 630
+    },
+    {
+      "epoch": 0.36809100189587285,
+      "grad_norm": 1.9170531034469604,
+      "learning_rate": 4.2037259615384614e-05,
+      "loss": 0.8887,
+      "step": 631
+    },
+    {
+      "epoch": 0.36867434738223714,
+      "grad_norm": 1.070654034614563,
+      "learning_rate": 4.202223557692308e-05,
+      "loss": 0.8389,
+      "step": 632
+    },
+    {
+      "epoch": 0.36925769286860144,
+      "grad_norm": 1.3350582122802734,
+      "learning_rate": 4.200721153846154e-05,
+      "loss": 1.1057,
+      "step": 633
+    },
+    {
+      "epoch": 0.36984103835496573,
+      "grad_norm": 1.2793062925338745,
+      "learning_rate": 4.1992187500000003e-05,
+      "loss": 1.0407,
+      "step": 634
+    },
+    {
+      "epoch": 0.37042438384133003,
+      "grad_norm": 1.2050164937973022,
+      "learning_rate": 4.197716346153847e-05,
+      "loss": 1.0973,
+      "step": 635
+    },
+    {
+      "epoch": 0.3710077293276943,
+      "grad_norm": 1.1398264169692993,
+      "learning_rate": 4.196213942307692e-05,
+      "loss": 1.162,
+      "step": 636
+    },
+    {
+      "epoch": 0.3715910748140586,
+      "grad_norm": 1.2741928100585938,
+      "learning_rate": 4.1947115384615386e-05,
+      "loss": 0.9104,
+      "step": 637
+    },
+    {
+      "epoch": 0.3721744203004229,
+      "grad_norm": 1.6818182468414307,
+      "learning_rate": 4.1932091346153844e-05,
+      "loss": 0.9304,
+      "step": 638
+    },
+    {
+      "epoch": 0.3727577657867872,
+      "grad_norm": 1.0803583860397339,
+      "learning_rate": 4.191706730769231e-05,
+      "loss": 1.1518,
+      "step": 639
+    },
+    {
+      "epoch": 0.3733411112731515,
+      "grad_norm": 1.178480625152588,
+      "learning_rate": 4.1902043269230775e-05,
+      "loss": 0.9778,
+      "step": 640
+    },
+    {
+      "epoch": 0.3739244567595158,
+      "grad_norm": 1.3216358423233032,
+      "learning_rate": 4.188701923076923e-05,
+      "loss": 1.0803,
+      "step": 641
+    },
+    {
+      "epoch": 0.3745078022458801,
+      "grad_norm": 1.191552758216858,
+      "learning_rate": 4.187199519230769e-05,
+      "loss": 1.0794,
+      "step": 642
+    },
+    {
+      "epoch": 0.37509114773224445,
+      "grad_norm": 1.240075945854187,
+      "learning_rate": 4.185697115384616e-05,
+      "loss": 1.2029,
+      "step": 643
+    },
+    {
+      "epoch": 0.37567449321860874,
+      "grad_norm": 1.2938767671585083,
+      "learning_rate": 4.1841947115384616e-05,
+      "loss": 1.1531,
+      "step": 644
+    },
+    {
+      "epoch": 0.37625783870497304,
+      "grad_norm": 1.4226038455963135,
+      "learning_rate": 4.182692307692308e-05,
+      "loss": 1.0929,
+      "step": 645
+    },
+    {
+      "epoch": 0.37684118419133733,
+      "grad_norm": 1.0986605882644653,
+      "learning_rate": 4.181189903846154e-05,
+      "loss": 0.8105,
+      "step": 646
+    },
+    {
+      "epoch": 0.3774245296777016,
+      "grad_norm": 1.0812294483184814,
+      "learning_rate": 4.1796875000000005e-05,
+      "loss": 1.0476,
+      "step": 647
+    },
+    {
+      "epoch": 0.3780078751640659,
+      "grad_norm": 1.7585773468017578,
+      "learning_rate": 4.178185096153846e-05,
+      "loss": 1.2113,
+      "step": 648
+    },
+    {
+      "epoch": 0.3785912206504302,
+      "grad_norm": 1.1438621282577515,
+      "learning_rate": 4.176682692307692e-05,
+      "loss": 0.8594,
+      "step": 649
+    },
+    {
+      "epoch": 0.3791745661367945,
+      "grad_norm": 1.1411939859390259,
+      "learning_rate": 4.175180288461539e-05,
+      "loss": 0.8793,
+      "step": 650
+    },
+    {
+      "epoch": 0.3797579116231588,
+      "grad_norm": 1.1287474632263184,
+      "learning_rate": 4.1736778846153845e-05,
+      "loss": 0.8364,
+      "step": 651
+    },
+    {
+      "epoch": 0.3803412571095231,
+      "grad_norm": 1.285798192024231,
+      "learning_rate": 4.172175480769231e-05,
+      "loss": 1.1573,
+      "step": 652
+    },
+    {
+      "epoch": 0.3809246025958874,
+      "grad_norm": 1.1466621160507202,
+      "learning_rate": 4.1706730769230776e-05,
+      "loss": 0.9106,
+      "step": 653
+    },
+    {
+      "epoch": 0.3815079480822517,
+      "grad_norm": 1.2659651041030884,
+      "learning_rate": 4.1691706730769234e-05,
+      "loss": 0.9637,
+      "step": 654
+    },
+    {
+      "epoch": 0.382091293568616,
+      "grad_norm": 1.6592333316802979,
+      "learning_rate": 4.167668269230769e-05,
+      "loss": 0.9402,
+      "step": 655
+    },
+    {
+      "epoch": 0.38267463905498034,
+      "grad_norm": 0.9642985463142395,
+      "learning_rate": 4.166165865384616e-05,
+      "loss": 0.9339,
+      "step": 656
+    },
+    {
+      "epoch": 0.38325798454134463,
+      "grad_norm": 1.3237916231155396,
+      "learning_rate": 4.164663461538462e-05,
+      "loss": 1.1165,
+      "step": 657
+    },
+    {
+      "epoch": 0.3838413300277089,
+      "grad_norm": 1.2888929843902588,
+      "learning_rate": 4.163161057692308e-05,
+      "loss": 1.1668,
+      "step": 658
+    },
+    {
+      "epoch": 0.3844246755140732,
+      "grad_norm": 1.3504743576049805,
+      "learning_rate": 4.161658653846154e-05,
+      "loss": 1.2046,
+      "step": 659
+    },
+    {
+      "epoch": 0.3850080210004375,
+      "grad_norm": 1.2326107025146484,
+      "learning_rate": 4.16015625e-05,
+      "loss": 0.8843,
+      "step": 660
+    },
+    {
+      "epoch": 0.3855913664868018,
+      "grad_norm": 1.083722472190857,
+      "learning_rate": 4.1586538461538464e-05,
+      "loss": 0.8806,
+      "step": 661
+    },
+    {
+      "epoch": 0.3861747119731661,
+      "grad_norm": 1.2427271604537964,
+      "learning_rate": 4.157151442307692e-05,
+      "loss": 1.1388,
+      "step": 662
+    },
+    {
+      "epoch": 0.3867580574595304,
+      "grad_norm": 1.2181854248046875,
+      "learning_rate": 4.155649038461539e-05,
+      "loss": 0.979,
+      "step": 663
+    },
+    {
+      "epoch": 0.3873414029458947,
+      "grad_norm": 1.6642446517944336,
+      "learning_rate": 4.1541466346153847e-05,
+      "loss": 1.0273,
+      "step": 664
+    },
+    {
+      "epoch": 0.387924748432259,
+      "grad_norm": 3.092848539352417,
+      "learning_rate": 4.152644230769231e-05,
+      "loss": 0.9759,
+      "step": 665
+    },
+    {
+      "epoch": 0.3885080939186233,
+      "grad_norm": 1.4167457818984985,
+      "learning_rate": 4.151141826923077e-05,
+      "loss": 1.1041,
+      "step": 666
+    },
+    {
+      "epoch": 0.3890914394049876,
+      "grad_norm": 1.4154654741287231,
+      "learning_rate": 4.149639423076923e-05,
+      "loss": 1.0213,
+      "step": 667
+    },
+    {
+      "epoch": 0.3896747848913519,
+      "grad_norm": 1.112176775932312,
+      "learning_rate": 4.1481370192307694e-05,
+      "loss": 0.9382,
+      "step": 668
+    },
+    {
+      "epoch": 0.39025813037771623,
+      "grad_norm": 1.0001271963119507,
+      "learning_rate": 4.146634615384616e-05,
+      "loss": 1.0377,
+      "step": 669
+    },
+    {
+      "epoch": 0.3908414758640805,
+      "grad_norm": 1.0494935512542725,
+      "learning_rate": 4.145132211538462e-05,
+      "loss": 1.0897,
+      "step": 670
+    },
+    {
+      "epoch": 0.3914248213504448,
+      "grad_norm": 1.2128801345825195,
+      "learning_rate": 4.143629807692308e-05,
+      "loss": 0.9406,
+      "step": 671
+    },
+    {
+      "epoch": 0.3920081668368091,
+      "grad_norm": 1.1072816848754883,
+      "learning_rate": 4.142127403846154e-05,
+      "loss": 0.9358,
+      "step": 672
+    },
+    {
+      "epoch": 0.3925915123231734,
+      "grad_norm": 1.251513123512268,
+      "learning_rate": 4.140625e-05,
+      "loss": 1.0557,
+      "step": 673
+    },
+    {
+      "epoch": 0.3931748578095377,
+      "grad_norm": 1.124248743057251,
+      "learning_rate": 4.1391225961538465e-05,
+      "loss": 1.1116,
+      "step": 674
+    },
+    {
+      "epoch": 0.393758203295902,
+      "grad_norm": 1.1827398538589478,
+      "learning_rate": 4.1376201923076924e-05,
+      "loss": 1.1149,
+      "step": 675
+    },
+    {
+      "epoch": 0.3943415487822663,
+      "grad_norm": 1.1932939291000366,
+      "learning_rate": 4.136117788461539e-05,
+      "loss": 0.865,
+      "step": 676
+    },
+    {
+      "epoch": 0.3949248942686306,
+      "grad_norm": 1.1423388719558716,
+      "learning_rate": 4.134615384615385e-05,
+      "loss": 1.0161,
+      "step": 677
+    },
+    {
+      "epoch": 0.3955082397549949,
+      "grad_norm": 1.1362109184265137,
+      "learning_rate": 4.1331129807692306e-05,
+      "loss": 1.1078,
+      "step": 678
+    },
+    {
+      "epoch": 0.3960915852413592,
+      "grad_norm": 1.173803448677063,
+      "learning_rate": 4.131610576923077e-05,
+      "loss": 1.0828,
+      "step": 679
+    },
+    {
+      "epoch": 0.3966749307277235,
+      "grad_norm": 1.1851235628128052,
+      "learning_rate": 4.130108173076923e-05,
+      "loss": 1.0101,
+      "step": 680
+    },
+    {
+      "epoch": 0.39725827621408777,
+      "grad_norm": 1.1966065168380737,
+      "learning_rate": 4.1286057692307695e-05,
+      "loss": 1.0556,
+      "step": 681
+    },
+    {
+      "epoch": 0.3978416217004521,
+      "grad_norm": 1.042441725730896,
+      "learning_rate": 4.127103365384616e-05,
+      "loss": 0.9449,
+      "step": 682
+    },
+    {
+      "epoch": 0.3984249671868164,
+      "grad_norm": 1.3966326713562012,
+      "learning_rate": 4.125600961538462e-05,
+      "loss": 1.2354,
+      "step": 683
+    },
+    {
+      "epoch": 0.3990083126731807,
+      "grad_norm": 1.7100861072540283,
+      "learning_rate": 4.124098557692308e-05,
+      "loss": 1.0369,
+      "step": 684
+    },
+    {
+      "epoch": 0.399591658159545,
+      "grad_norm": 1.1547572612762451,
+      "learning_rate": 4.1225961538461536e-05,
+      "loss": 1.2298,
+      "step": 685
+    },
+    {
+      "epoch": 0.4001750036459093,
+      "grad_norm": 1.2655994892120361,
+      "learning_rate": 4.12109375e-05,
+      "loss": 1.0823,
+      "step": 686
+    },
+    {
+      "epoch": 0.4007583491322736,
+      "grad_norm": 1.1846497058868408,
+      "learning_rate": 4.119591346153847e-05,
+      "loss": 1.127,
+      "step": 687
+    },
+    {
+      "epoch": 0.4013416946186379,
+      "grad_norm": 1.3547903299331665,
+      "learning_rate": 4.1180889423076925e-05,
+      "loss": 0.9981,
+      "step": 688
+    },
+    {
+      "epoch": 0.4019250401050022,
+      "grad_norm": 1.16106379032135,
+      "learning_rate": 4.116586538461539e-05,
+      "loss": 1.0675,
+      "step": 689
+    },
+    {
+      "epoch": 0.4025083855913665,
+      "grad_norm": 1.3457812070846558,
+      "learning_rate": 4.115084134615385e-05,
+      "loss": 1.0359,
+      "step": 690
+    },
+    {
+      "epoch": 0.4030917310777308,
+      "grad_norm": 1.440121054649353,
+      "learning_rate": 4.113581730769231e-05,
+      "loss": 1.0681,
+      "step": 691
+    },
+    {
+      "epoch": 0.40367507656409507,
+      "grad_norm": 1.382415533065796,
+      "learning_rate": 4.112079326923077e-05,
+      "loss": 0.9819,
+      "step": 692
+    },
+    {
+      "epoch": 0.40425842205045937,
+      "grad_norm": 1.391350507736206,
+      "learning_rate": 4.110576923076923e-05,
+      "loss": 1.0845,
+      "step": 693
+    },
+    {
+      "epoch": 0.40484176753682366,
+      "grad_norm": 1.2903047800064087,
+      "learning_rate": 4.1090745192307696e-05,
+      "loss": 1.0631,
+      "step": 694
+    },
+    {
+      "epoch": 0.405425113023188,
+      "grad_norm": 1.1688337326049805,
+      "learning_rate": 4.107572115384616e-05,
+      "loss": 0.9691,
+      "step": 695
+    },
+    {
+      "epoch": 0.4060084585095523,
+      "grad_norm": 1.327972173690796,
+      "learning_rate": 4.1060697115384613e-05,
+      "loss": 1.0152,
+      "step": 696
+    },
+    {
+      "epoch": 0.4065918039959166,
+      "grad_norm": 1.1001508235931396,
+      "learning_rate": 4.104567307692308e-05,
+      "loss": 0.9932,
+      "step": 697
+    },
+    {
+      "epoch": 0.4071751494822809,
+      "grad_norm": 1.1654032468795776,
+      "learning_rate": 4.103064903846154e-05,
+      "loss": 1.0077,
+      "step": 698
+    },
+    {
+      "epoch": 0.4077584949686452,
+      "grad_norm": 1.3325926065444946,
+      "learning_rate": 4.1015625e-05,
+      "loss": 0.9754,
+      "step": 699
+    },
+    {
+      "epoch": 0.4083418404550095,
+      "grad_norm": 1.0101975202560425,
+      "learning_rate": 4.100060096153847e-05,
+      "loss": 0.835,
+      "step": 700
+    },
+    {
+      "epoch": 0.4089251859413738,
+      "grad_norm": 1.0377428531646729,
+      "learning_rate": 4.0985576923076926e-05,
+      "loss": 0.9193,
+      "step": 701
+    },
+    {
+      "epoch": 0.4095085314277381,
+      "grad_norm": 1.4344565868377686,
+      "learning_rate": 4.0970552884615385e-05,
+      "loss": 0.981,
+      "step": 702
+    },
+    {
+      "epoch": 0.41009187691410237,
+      "grad_norm": 1.0151726007461548,
+      "learning_rate": 4.095552884615384e-05,
+      "loss": 0.8954,
+      "step": 703
+    },
+    {
+      "epoch": 0.41067522240046667,
+      "grad_norm": 1.2651805877685547,
+      "learning_rate": 4.094050480769231e-05,
+      "loss": 0.8135,
+      "step": 704
+    },
+    {
+      "epoch": 0.41125856788683096,
+      "grad_norm": 1.1307834386825562,
+      "learning_rate": 4.0925480769230774e-05,
+      "loss": 1.0629,
+      "step": 705
+    },
+    {
+      "epoch": 0.41184191337319526,
+      "grad_norm": 1.5362266302108765,
+      "learning_rate": 4.091045673076923e-05,
+      "loss": 1.1536,
+      "step": 706
+    },
+    {
+      "epoch": 0.41242525885955955,
+      "grad_norm": 1.1772881746292114,
+      "learning_rate": 4.08954326923077e-05,
+      "loss": 1.3347,
+      "step": 707
+    },
+    {
+      "epoch": 0.41300860434592385,
+      "grad_norm": 1.2242248058319092,
+      "learning_rate": 4.0880408653846156e-05,
+      "loss": 0.9295,
+      "step": 708
+    },
+    {
+      "epoch": 0.4135919498322882,
+      "grad_norm": 1.2769187688827515,
+      "learning_rate": 4.0865384615384615e-05,
+      "loss": 1.0842,
+      "step": 709
+    },
+    {
+      "epoch": 0.4141752953186525,
+      "grad_norm": 0.9419771432876587,
+      "learning_rate": 4.085036057692308e-05,
+      "loss": 0.9203,
+      "step": 710
+    },
+    {
+      "epoch": 0.4147586408050168,
+      "grad_norm": 1.5197808742523193,
+      "learning_rate": 4.083533653846154e-05,
+      "loss": 0.8543,
+      "step": 711
+    },
+    {
+      "epoch": 0.4153419862913811,
+      "grad_norm": 1.1694600582122803,
+      "learning_rate": 4.0820312500000004e-05,
+      "loss": 1.262,
+      "step": 712
+    },
+    {
+      "epoch": 0.4159253317777454,
+      "grad_norm": 1.1101324558258057,
+      "learning_rate": 4.080528846153847e-05,
+      "loss": 0.8561,
+      "step": 713
+    },
+    {
+      "epoch": 0.4165086772641097,
+      "grad_norm": 1.1828705072402954,
+      "learning_rate": 4.079026442307692e-05,
+      "loss": 0.8932,
+      "step": 714
+    },
+    {
+      "epoch": 0.41709202275047397,
+      "grad_norm": 1.1211762428283691,
+      "learning_rate": 4.0775240384615386e-05,
+      "loss": 0.9816,
+      "step": 715
+    },
+    {
+      "epoch": 0.41767536823683826,
+      "grad_norm": 1.2857762575149536,
+      "learning_rate": 4.0760216346153845e-05,
+      "loss": 1.0257,
+      "step": 716
+    },
+    {
+      "epoch": 0.41825871372320256,
+      "grad_norm": 1.0974875688552856,
+      "learning_rate": 4.074519230769231e-05,
+      "loss": 0.6593,
+      "step": 717
+    },
+    {
+      "epoch": 0.41884205920956685,
+      "grad_norm": 1.2576038837432861,
+      "learning_rate": 4.0730168269230775e-05,
+      "loss": 0.9487,
+      "step": 718
+    },
+    {
+      "epoch": 0.41942540469593115,
+      "grad_norm": 1.1279481649398804,
+      "learning_rate": 4.0715144230769234e-05,
+      "loss": 1.0483,
+      "step": 719
+    },
+    {
+      "epoch": 0.42000875018229544,
+      "grad_norm": 1.191821575164795,
+      "learning_rate": 4.070012019230769e-05,
+      "loss": 1.0075,
+      "step": 720
+    },
+    {
+      "epoch": 0.42059209566865974,
+      "grad_norm": 0.9579320549964905,
+      "learning_rate": 4.068509615384616e-05,
+      "loss": 1.1948,
+      "step": 721
+    },
+    {
+      "epoch": 0.4211754411550241,
+      "grad_norm": 1.1844723224639893,
+      "learning_rate": 4.0670072115384616e-05,
+      "loss": 1.0395,
+      "step": 722
+    },
+    {
+      "epoch": 0.4217587866413884,
+      "grad_norm": 1.192607045173645,
+      "learning_rate": 4.065504807692308e-05,
+      "loss": 0.8447,
+      "step": 723
+    },
+    {
+      "epoch": 0.4223421321277527,
+      "grad_norm": 1.256598949432373,
+      "learning_rate": 4.064002403846154e-05,
+      "loss": 0.8741,
+      "step": 724
+    },
+    {
+      "epoch": 0.422925477614117,
+      "grad_norm": 1.1558480262756348,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 1.1007,
+      "step": 725
+    },
+    {
+      "epoch": 0.42350882310048127,
+      "grad_norm": 1.2486516237258911,
+      "learning_rate": 4.0609975961538463e-05,
+      "loss": 1.0104,
+      "step": 726
+    },
+    {
+      "epoch": 0.42409216858684556,
+      "grad_norm": 1.1530859470367432,
+      "learning_rate": 4.059495192307692e-05,
+      "loss": 0.9664,
+      "step": 727
+    },
+    {
+      "epoch": 0.42467551407320986,
+      "grad_norm": 1.1082444190979004,
+      "learning_rate": 4.057992788461539e-05,
+      "loss": 1.2137,
+      "step": 728
+    },
+    {
+      "epoch": 0.42525885955957415,
+      "grad_norm": 1.1947239637374878,
+      "learning_rate": 4.0564903846153846e-05,
+      "loss": 1.1242,
+      "step": 729
+    },
+    {
+      "epoch": 0.42584220504593845,
+      "grad_norm": 1.185736060142517,
+      "learning_rate": 4.054987980769231e-05,
+      "loss": 1.0753,
+      "step": 730
+    },
+    {
+      "epoch": 0.42642555053230274,
+      "grad_norm": 1.292222499847412,
+      "learning_rate": 4.0534855769230776e-05,
+      "loss": 1.2377,
+      "step": 731
+    },
+    {
+      "epoch": 0.42700889601866704,
+      "grad_norm": 1.3122601509094238,
+      "learning_rate": 4.051983173076923e-05,
+      "loss": 1.0001,
+      "step": 732
+    },
+    {
+      "epoch": 0.42759224150503133,
+      "grad_norm": 1.284077525138855,
+      "learning_rate": 4.050480769230769e-05,
+      "loss": 0.8811,
+      "step": 733
+    },
+    {
+      "epoch": 0.42817558699139563,
+      "grad_norm": 1.266852855682373,
+      "learning_rate": 4.048978365384616e-05,
+      "loss": 0.8545,
+      "step": 734
+    },
+    {
+      "epoch": 0.42875893247776,
+      "grad_norm": 1.1647552251815796,
+      "learning_rate": 4.047475961538462e-05,
+      "loss": 0.9484,
+      "step": 735
+    },
+    {
+      "epoch": 0.4293422779641243,
+      "grad_norm": 1.3035751581192017,
+      "learning_rate": 4.045973557692308e-05,
+      "loss": 1.0405,
+      "step": 736
+    },
+    {
+      "epoch": 0.42992562345048857,
+      "grad_norm": 1.1369545459747314,
+      "learning_rate": 4.044471153846154e-05,
+      "loss": 0.8982,
+      "step": 737
+    },
+    {
+      "epoch": 0.43050896893685286,
+      "grad_norm": 1.1648366451263428,
+      "learning_rate": 4.04296875e-05,
+      "loss": 1.0459,
+      "step": 738
+    },
+    {
+      "epoch": 0.43109231442321716,
+      "grad_norm": 1.3287396430969238,
+      "learning_rate": 4.0414663461538465e-05,
+      "loss": 1.0519,
+      "step": 739
+    },
+    {
+      "epoch": 0.43167565990958146,
+      "grad_norm": 1.1275254487991333,
+      "learning_rate": 4.039963942307692e-05,
+      "loss": 0.8556,
+      "step": 740
+    },
+    {
+      "epoch": 0.43225900539594575,
+      "grad_norm": 1.3335869312286377,
+      "learning_rate": 4.038461538461539e-05,
+      "loss": 0.9997,
+      "step": 741
+    },
+    {
+      "epoch": 0.43284235088231005,
+      "grad_norm": 1.073275089263916,
+      "learning_rate": 4.036959134615385e-05,
+      "loss": 1.1496,
+      "step": 742
+    },
+    {
+      "epoch": 0.43342569636867434,
+      "grad_norm": 1.194717288017273,
+      "learning_rate": 4.035456730769231e-05,
+      "loss": 0.958,
+      "step": 743
+    },
+    {
+      "epoch": 0.43400904185503864,
+      "grad_norm": 1.55983567237854,
+      "learning_rate": 4.033954326923077e-05,
+      "loss": 1.2247,
+      "step": 744
+    },
+    {
+      "epoch": 0.43459238734140293,
+      "grad_norm": 2.449402093887329,
+      "learning_rate": 4.032451923076923e-05,
+      "loss": 1.2115,
+      "step": 745
+    },
+    {
+      "epoch": 0.4351757328277672,
+      "grad_norm": 1.2797013521194458,
+      "learning_rate": 4.0309495192307694e-05,
+      "loss": 0.8518,
+      "step": 746
+    },
+    {
+      "epoch": 0.4357590783141315,
+      "grad_norm": 1.1281001567840576,
+      "learning_rate": 4.029447115384616e-05,
+      "loss": 1.0215,
+      "step": 747
+    },
+    {
+      "epoch": 0.43634242380049587,
+      "grad_norm": 1.1771413087844849,
+      "learning_rate": 4.027944711538462e-05,
+      "loss": 1.1645,
+      "step": 748
+    },
+    {
+      "epoch": 0.43692576928686017,
+      "grad_norm": 1.181060552597046,
+      "learning_rate": 4.0264423076923083e-05,
+      "loss": 0.8806,
+      "step": 749
+    },
+    {
+      "epoch": 0.43750911477322446,
+      "grad_norm": 1.2284411191940308,
+      "learning_rate": 4.0249399038461535e-05,
+      "loss": 1.1494,
+      "step": 750
+    },
+    {
+      "epoch": 0.43809246025958876,
+      "grad_norm": 1.0676560401916504,
+      "learning_rate": 4.0234375e-05,
+      "loss": 0.9234,
+      "step": 751
+    },
+    {
+      "epoch": 0.43867580574595305,
+      "grad_norm": 1.481394648551941,
+      "learning_rate": 4.0219350961538466e-05,
+      "loss": 0.904,
+      "step": 752
+    },
+    {
+      "epoch": 0.43925915123231735,
+      "grad_norm": 1.0527502298355103,
+      "learning_rate": 4.0204326923076924e-05,
+      "loss": 1.0413,
+      "step": 753
+    },
+    {
+      "epoch": 0.43984249671868164,
+      "grad_norm": 1.2585145235061646,
+      "learning_rate": 4.018930288461539e-05,
+      "loss": 0.9207,
+      "step": 754
+    },
+    {
+      "epoch": 0.44042584220504594,
+      "grad_norm": 1.5194056034088135,
+      "learning_rate": 4.017427884615385e-05,
+      "loss": 1.0445,
+      "step": 755
+    },
+    {
+      "epoch": 0.44100918769141023,
+      "grad_norm": 1.1574163436889648,
+      "learning_rate": 4.0159254807692307e-05,
+      "loss": 0.8065,
+      "step": 756
+    },
+    {
+      "epoch": 0.4415925331777745,
+      "grad_norm": 1.351405143737793,
+      "learning_rate": 4.014423076923077e-05,
+      "loss": 0.914,
+      "step": 757
+    },
+    {
+      "epoch": 0.4421758786641388,
+      "grad_norm": 1.1521351337432861,
+      "learning_rate": 4.012920673076923e-05,
+      "loss": 0.9017,
+      "step": 758
+    },
+    {
+      "epoch": 0.4427592241505031,
+      "grad_norm": 1.9487134218215942,
+      "learning_rate": 4.0114182692307696e-05,
+      "loss": 1.03,
+      "step": 759
+    },
+    {
+      "epoch": 0.4433425696368674,
+      "grad_norm": 2.560194969177246,
+      "learning_rate": 4.009915865384616e-05,
+      "loss": 1.0656,
+      "step": 760
+    },
+    {
+      "epoch": 0.4439259151232317,
+      "grad_norm": 1.140284538269043,
+      "learning_rate": 4.008413461538462e-05,
+      "loss": 1.0233,
+      "step": 761
+    },
+    {
+      "epoch": 0.44450926060959606,
+      "grad_norm": 1.3260776996612549,
+      "learning_rate": 4.006911057692308e-05,
+      "loss": 1.0017,
+      "step": 762
+    },
+    {
+      "epoch": 0.44509260609596035,
+      "grad_norm": 1.1119332313537598,
+      "learning_rate": 4.0054086538461536e-05,
+      "loss": 1.2699,
+      "step": 763
+    },
+    {
+      "epoch": 0.44567595158232465,
+      "grad_norm": 1.3355944156646729,
+      "learning_rate": 4.00390625e-05,
+      "loss": 0.9841,
+      "step": 764
+    },
+    {
+      "epoch": 0.44625929706868894,
+      "grad_norm": 1.3637195825576782,
+      "learning_rate": 4.002403846153847e-05,
+      "loss": 1.0625,
+      "step": 765
+    },
+    {
+      "epoch": 0.44684264255505324,
+      "grad_norm": 1.1988649368286133,
+      "learning_rate": 4.0009014423076925e-05,
+      "loss": 1.0488,
+      "step": 766
+    },
+    {
+      "epoch": 0.44742598804141753,
+      "grad_norm": 2.336710214614868,
+      "learning_rate": 3.999399038461539e-05,
+      "loss": 1.1126,
+      "step": 767
+    },
+    {
+      "epoch": 0.44800933352778183,
+      "grad_norm": 1.3706448078155518,
+      "learning_rate": 3.997896634615384e-05,
+      "loss": 1.0118,
+      "step": 768
+    },
+    {
+      "epoch": 0.4485926790141461,
+      "grad_norm": 1.171186089515686,
+      "learning_rate": 3.996394230769231e-05,
+      "loss": 1.024,
+      "step": 769
+    },
+    {
+      "epoch": 0.4491760245005104,
+      "grad_norm": 1.424375057220459,
+      "learning_rate": 3.994891826923077e-05,
+      "loss": 1.0885,
+      "step": 770
+    },
+    {
+      "epoch": 0.4497593699868747,
+      "grad_norm": 1.218055009841919,
+      "learning_rate": 3.993389423076923e-05,
+      "loss": 0.9143,
+      "step": 771
+    },
+    {
+      "epoch": 0.450342715473239,
+      "grad_norm": 1.1051896810531616,
+      "learning_rate": 3.99188701923077e-05,
+      "loss": 0.8874,
+      "step": 772
+    },
+    {
+      "epoch": 0.4509260609596033,
+      "grad_norm": 1.2627577781677246,
+      "learning_rate": 3.9903846153846155e-05,
+      "loss": 0.9562,
+      "step": 773
+    },
+    {
+      "epoch": 0.4515094064459676,
+      "grad_norm": 2.4132933616638184,
+      "learning_rate": 3.9888822115384614e-05,
+      "loss": 1.0477,
+      "step": 774
+    },
+    {
+      "epoch": 0.45209275193233195,
+      "grad_norm": 1.4425148963928223,
+      "learning_rate": 3.987379807692308e-05,
+      "loss": 0.8765,
+      "step": 775
+    },
+    {
+      "epoch": 0.45267609741869624,
+      "grad_norm": 1.0001459121704102,
+      "learning_rate": 3.985877403846154e-05,
+      "loss": 0.9992,
+      "step": 776
+    },
+    {
+      "epoch": 0.45325944290506054,
+      "grad_norm": 1.2499384880065918,
+      "learning_rate": 3.984375e-05,
+      "loss": 1.0017,
+      "step": 777
+    },
+    {
+      "epoch": 0.45384278839142483,
+      "grad_norm": 1.0234681367874146,
+      "learning_rate": 3.982872596153847e-05,
+      "loss": 0.9316,
+      "step": 778
+    },
+    {
+      "epoch": 0.45442613387778913,
+      "grad_norm": 1.1663625240325928,
+      "learning_rate": 3.981370192307693e-05,
+      "loss": 1.0178,
+      "step": 779
+    },
+    {
+      "epoch": 0.4550094793641534,
+      "grad_norm": 1.0986937284469604,
+      "learning_rate": 3.9798677884615385e-05,
+      "loss": 0.9092,
+      "step": 780
+    },
+    {
+      "epoch": 0.4555928248505177,
+      "grad_norm": 1.2490792274475098,
+      "learning_rate": 3.9783653846153844e-05,
+      "loss": 1.046,
+      "step": 781
+    },
+    {
+      "epoch": 0.456176170336882,
+      "grad_norm": 1.217383861541748,
+      "learning_rate": 3.976862980769231e-05,
+      "loss": 0.9513,
+      "step": 782
+    },
+    {
+      "epoch": 0.4567595158232463,
+      "grad_norm": 1.4201329946517944,
+      "learning_rate": 3.9753605769230774e-05,
+      "loss": 1.0895,
+      "step": 783
+    },
+    {
+      "epoch": 0.4573428613096106,
+      "grad_norm": 1.3387080430984497,
+      "learning_rate": 3.973858173076923e-05,
+      "loss": 0.9925,
+      "step": 784
+    },
+    {
+      "epoch": 0.4579262067959749,
+      "grad_norm": 1.2447290420532227,
+      "learning_rate": 3.97235576923077e-05,
+      "loss": 1.1017,
+      "step": 785
+    },
+    {
+      "epoch": 0.4585095522823392,
+      "grad_norm": 1.403903841972351,
+      "learning_rate": 3.9708533653846156e-05,
+      "loss": 0.8692,
+      "step": 786
+    },
+    {
+      "epoch": 0.4590928977687035,
+      "grad_norm": 1.3049734830856323,
+      "learning_rate": 3.9693509615384615e-05,
+      "loss": 1.1045,
+      "step": 787
+    },
+    {
+      "epoch": 0.45967624325506784,
+      "grad_norm": 2.3408448696136475,
+      "learning_rate": 3.967848557692308e-05,
+      "loss": 1.0606,
+      "step": 788
+    },
+    {
+      "epoch": 0.46025958874143214,
+      "grad_norm": 1.189640760421753,
+      "learning_rate": 3.966346153846154e-05,
+      "loss": 1.0287,
+      "step": 789
+    },
+    {
+      "epoch": 0.46084293422779643,
+      "grad_norm": 1.1157665252685547,
+      "learning_rate": 3.9648437500000004e-05,
+      "loss": 0.9822,
+      "step": 790
+    },
+    {
+      "epoch": 0.4614262797141607,
+      "grad_norm": 1.242864966392517,
+      "learning_rate": 3.963341346153846e-05,
+      "loss": 0.827,
+      "step": 791
+    },
+    {
+      "epoch": 0.462009625200525,
+      "grad_norm": 1.1176204681396484,
+      "learning_rate": 3.961838942307692e-05,
+      "loss": 1.2449,
+      "step": 792
+    },
+    {
+      "epoch": 0.4625929706868893,
+      "grad_norm": 1.0381675958633423,
+      "learning_rate": 3.9603365384615386e-05,
+      "loss": 1.1754,
+      "step": 793
+    },
+    {
+      "epoch": 0.4631763161732536,
+      "grad_norm": 1.2429289817810059,
+      "learning_rate": 3.9588341346153845e-05,
+      "loss": 1.0103,
+      "step": 794
+    },
+    {
+      "epoch": 0.4637596616596179,
+      "grad_norm": 1.214370608329773,
+      "learning_rate": 3.957331730769231e-05,
+      "loss": 0.9133,
+      "step": 795
+    },
+    {
+      "epoch": 0.4643430071459822,
+      "grad_norm": 1.249859094619751,
+      "learning_rate": 3.9558293269230775e-05,
+      "loss": 1.2252,
+      "step": 796
+    },
+    {
+      "epoch": 0.4649263526323465,
+      "grad_norm": 1.1522125005722046,
+      "learning_rate": 3.9543269230769234e-05,
+      "loss": 0.8547,
+      "step": 797
+    },
+    {
+      "epoch": 0.4655096981187108,
+      "grad_norm": 1.0914942026138306,
+      "learning_rate": 3.952824519230769e-05,
+      "loss": 1.1068,
+      "step": 798
+    },
+    {
+      "epoch": 0.4660930436050751,
+      "grad_norm": 1.2309699058532715,
+      "learning_rate": 3.951322115384616e-05,
+      "loss": 1.1301,
+      "step": 799
+    },
+    {
+      "epoch": 0.4666763890914394,
+      "grad_norm": 1.3135758638381958,
+      "learning_rate": 3.9498197115384616e-05,
+      "loss": 0.947,
+      "step": 800
+    },
+    {
+      "epoch": 0.4666763890914394,
+      "eval_loss_squad": 0.9123072922043503,
+      "eval_perplexity": 7.919284365740796,
+      "eval_perplexity_reconstruct": 1.9212373943429781,
+      "step": 800
+    },
+    {
+      "epoch": 0.46725973457780373,
+      "grad_norm": 1.0412598848342896,
+      "learning_rate": 3.948317307692308e-05,
+      "loss": 1.1682,
+      "step": 801
+    },
+    {
+      "epoch": 0.467843080064168,
+      "grad_norm": 0.986031711101532,
+      "learning_rate": 3.946814903846154e-05,
+      "loss": 0.7983,
+      "step": 802
+    },
+    {
+      "epoch": 0.4684264255505323,
+      "grad_norm": 1.149339199066162,
+      "learning_rate": 3.9453125000000005e-05,
+      "loss": 0.9834,
+      "step": 803
+    },
+    {
+      "epoch": 0.4690097710368966,
+      "grad_norm": 1.2143068313598633,
+      "learning_rate": 3.9438100961538464e-05,
+      "loss": 1.1033,
+      "step": 804
+    },
+    {
+      "epoch": 0.4695931165232609,
+      "grad_norm": 1.015660285949707,
+      "learning_rate": 3.942307692307692e-05,
+      "loss": 0.8502,
+      "step": 805
+    },
+    {
+      "epoch": 0.4701764620096252,
+      "grad_norm": 1.3838074207305908,
+      "learning_rate": 3.940805288461539e-05,
+      "loss": 0.9986,
+      "step": 806
+    },
+    {
+      "epoch": 0.4707598074959895,
+      "grad_norm": 1.1482564210891724,
+      "learning_rate": 3.9393028846153846e-05,
+      "loss": 1.0143,
+      "step": 807
+    },
+    {
+      "epoch": 0.4713431529823538,
+      "grad_norm": 1.0859732627868652,
+      "learning_rate": 3.937800480769231e-05,
+      "loss": 0.9108,
+      "step": 808
+    },
+    {
+      "epoch": 0.4719264984687181,
+      "grad_norm": 1.4561444520950317,
+      "learning_rate": 3.936298076923077e-05,
+      "loss": 0.7463,
+      "step": 809
+    },
+    {
+      "epoch": 0.4725098439550824,
+      "grad_norm": 1.909414529800415,
+      "learning_rate": 3.934795673076923e-05,
+      "loss": 1.0691,
+      "step": 810
+    },
+    {
+      "epoch": 0.4730931894414467,
+      "grad_norm": 2.1978914737701416,
+      "learning_rate": 3.9332932692307694e-05,
+      "loss": 1.263,
+      "step": 811
+    },
+    {
+      "epoch": 0.473676534927811,
+      "grad_norm": 1.4734632968902588,
+      "learning_rate": 3.931790865384616e-05,
+      "loss": 0.8494,
+      "step": 812
+    },
+    {
+      "epoch": 0.47425988041417527,
+      "grad_norm": 1.2567843198776245,
+      "learning_rate": 3.930288461538462e-05,
+      "loss": 1.0615,
+      "step": 813
+    },
+    {
+      "epoch": 0.47484322590053957,
+      "grad_norm": 1.2729908227920532,
+      "learning_rate": 3.928786057692308e-05,
+      "loss": 0.8998,
+      "step": 814
+    },
+    {
+      "epoch": 0.4754265713869039,
+      "grad_norm": 1.0162193775177002,
+      "learning_rate": 3.927283653846154e-05,
+      "loss": 1.0442,
+      "step": 815
+    },
+    {
+      "epoch": 0.4760099168732682,
+      "grad_norm": 1.0893824100494385,
+      "learning_rate": 3.92578125e-05,
+      "loss": 0.9158,
+      "step": 816
+    },
+    {
+      "epoch": 0.4765932623596325,
+      "grad_norm": 0.9300756454467773,
+      "learning_rate": 3.9242788461538465e-05,
+      "loss": 0.9869,
+      "step": 817
+    },
+    {
+      "epoch": 0.4771766078459968,
+      "grad_norm": 1.3809287548065186,
+      "learning_rate": 3.9227764423076923e-05,
+      "loss": 0.9937,
+      "step": 818
+    },
+    {
+      "epoch": 0.4777599533323611,
+      "grad_norm": 1.1738115549087524,
+      "learning_rate": 3.921274038461539e-05,
+      "loss": 1.0634,
+      "step": 819
+    },
+    {
+      "epoch": 0.4783432988187254,
+      "grad_norm": 1.2441519498825073,
+      "learning_rate": 3.919771634615385e-05,
+      "loss": 0.7852,
+      "step": 820
+    },
+    {
+      "epoch": 0.4789266443050897,
+      "grad_norm": 0.8443605303764343,
+      "learning_rate": 3.918269230769231e-05,
+      "loss": 1.0473,
+      "step": 821
+    },
+    {
+      "epoch": 0.479509989791454,
+      "grad_norm": 1.2155978679656982,
+      "learning_rate": 3.916766826923077e-05,
+      "loss": 0.9415,
+      "step": 822
+    },
+    {
+      "epoch": 0.4800933352778183,
+      "grad_norm": 1.1366358995437622,
+      "learning_rate": 3.915264423076923e-05,
+      "loss": 0.9134,
+      "step": 823
+    },
+    {
+      "epoch": 0.4806766807641826,
+      "grad_norm": 1.1710559129714966,
+      "learning_rate": 3.9137620192307695e-05,
+      "loss": 1.0786,
+      "step": 824
+    },
+    {
+      "epoch": 0.48126002625054687,
+      "grad_norm": 1.0255122184753418,
+      "learning_rate": 3.912259615384616e-05,
+      "loss": 1.0628,
+      "step": 825
+    },
+    {
+      "epoch": 0.48184337173691116,
+      "grad_norm": 0.9826768636703491,
+      "learning_rate": 3.910757211538462e-05,
+      "loss": 0.8179,
+      "step": 826
+    },
+    {
+      "epoch": 0.48242671722327546,
+      "grad_norm": 1.3819292783737183,
+      "learning_rate": 3.909254807692308e-05,
+      "loss": 1.0797,
+      "step": 827
+    },
+    {
+      "epoch": 0.4830100627096398,
+      "grad_norm": 1.038661003112793,
+      "learning_rate": 3.9077524038461536e-05,
+      "loss": 1.2463,
+      "step": 828
+    },
+    {
+      "epoch": 0.4835934081960041,
+      "grad_norm": 1.517325758934021,
+      "learning_rate": 3.90625e-05,
+      "loss": 1.0843,
+      "step": 829
+    },
+    {
+      "epoch": 0.4841767536823684,
+      "grad_norm": 1.1275871992111206,
+      "learning_rate": 3.9047475961538466e-05,
+      "loss": 1.0089,
+      "step": 830
+    },
+    {
+      "epoch": 0.4847600991687327,
+      "grad_norm": 2.5894081592559814,
+      "learning_rate": 3.9032451923076925e-05,
+      "loss": 1.0206,
+      "step": 831
+    },
+    {
+      "epoch": 0.485343444655097,
+      "grad_norm": 1.2647501230239868,
+      "learning_rate": 3.901742788461539e-05,
+      "loss": 1.0476,
+      "step": 832
+    },
+    {
+      "epoch": 0.4859267901414613,
+      "grad_norm": 1.1321070194244385,
+      "learning_rate": 3.900240384615385e-05,
+      "loss": 0.9772,
+      "step": 833
+    },
+    {
+      "epoch": 0.4865101356278256,
+      "grad_norm": 1.4880715608596802,
+      "learning_rate": 3.898737980769231e-05,
+      "loss": 0.7914,
+      "step": 834
+    },
+    {
+      "epoch": 0.4870934811141899,
+      "grad_norm": 1.07522714138031,
+      "learning_rate": 3.897235576923077e-05,
+      "loss": 0.9263,
+      "step": 835
+    },
+    {
+      "epoch": 0.48767682660055417,
+      "grad_norm": 1.0414209365844727,
+      "learning_rate": 3.895733173076923e-05,
+      "loss": 1.2107,
+      "step": 836
+    },
+    {
+      "epoch": 0.48826017208691846,
+      "grad_norm": 1.1117522716522217,
+      "learning_rate": 3.8942307692307696e-05,
+      "loss": 0.833,
+      "step": 837
+    },
+    {
+      "epoch": 0.48884351757328276,
+      "grad_norm": 1.225772500038147,
+      "learning_rate": 3.892728365384616e-05,
+      "loss": 1.0641,
+      "step": 838
+    },
+    {
+      "epoch": 0.48942686305964705,
+      "grad_norm": 1.2368497848510742,
+      "learning_rate": 3.891225961538462e-05,
+      "loss": 0.9278,
+      "step": 839
+    },
+    {
+      "epoch": 0.49001020854601135,
+      "grad_norm": 1.068732500076294,
+      "learning_rate": 3.889723557692308e-05,
+      "loss": 0.8247,
+      "step": 840
+    },
+    {
+      "epoch": 0.4905935540323757,
+      "grad_norm": 1.1534667015075684,
+      "learning_rate": 3.888221153846154e-05,
+      "loss": 0.9824,
+      "step": 841
+    },
+    {
+      "epoch": 0.49117689951874,
+      "grad_norm": 1.3575752973556519,
+      "learning_rate": 3.88671875e-05,
+      "loss": 1.0802,
+      "step": 842
+    },
+    {
+      "epoch": 0.4917602450051043,
+      "grad_norm": 1.2261594533920288,
+      "learning_rate": 3.885216346153847e-05,
+      "loss": 1.1202,
+      "step": 843
+    },
+    {
+      "epoch": 0.4923435904914686,
+      "grad_norm": 1.2742036581039429,
+      "learning_rate": 3.8837139423076926e-05,
+      "loss": 1.0831,
+      "step": 844
+    },
+    {
+      "epoch": 0.4929269359778329,
+      "grad_norm": 1.2143781185150146,
+      "learning_rate": 3.8822115384615384e-05,
+      "loss": 1.0475,
+      "step": 845
+    },
+    {
+      "epoch": 0.4935102814641972,
+      "grad_norm": 1.1177541017532349,
+      "learning_rate": 3.880709134615384e-05,
+      "loss": 1.0494,
+      "step": 846
+    },
+    {
+      "epoch": 0.49409362695056147,
+      "grad_norm": 1.1861687898635864,
+      "learning_rate": 3.879206730769231e-05,
+      "loss": 0.9165,
+      "step": 847
+    },
+    {
+      "epoch": 0.49467697243692577,
+      "grad_norm": 1.2765378952026367,
+      "learning_rate": 3.877704326923077e-05,
+      "loss": 0.891,
+      "step": 848
+    },
+    {
+      "epoch": 0.49526031792329006,
+      "grad_norm": 1.1258078813552856,
+      "learning_rate": 3.876201923076923e-05,
+      "loss": 0.9761,
+      "step": 849
+    },
+    {
+      "epoch": 0.49584366340965436,
+      "grad_norm": 1.1065664291381836,
+      "learning_rate": 3.87469951923077e-05,
+      "loss": 0.9383,
+      "step": 850
+    },
+    {
+      "epoch": 0.49642700889601865,
+      "grad_norm": 1.1673333644866943,
+      "learning_rate": 3.8731971153846156e-05,
+      "loss": 0.9632,
+      "step": 851
+    },
+    {
+      "epoch": 0.49701035438238295,
+      "grad_norm": 1.428477168083191,
+      "learning_rate": 3.8716947115384614e-05,
+      "loss": 0.9079,
+      "step": 852
+    },
+    {
+      "epoch": 0.49759369986874724,
+      "grad_norm": 1.1232894659042358,
+      "learning_rate": 3.870192307692308e-05,
+      "loss": 0.9956,
+      "step": 853
+    },
+    {
+      "epoch": 0.4981770453551116,
+      "grad_norm": 1.2544480562210083,
+      "learning_rate": 3.868689903846154e-05,
+      "loss": 1.0855,
+      "step": 854
+    },
+    {
+      "epoch": 0.4987603908414759,
+      "grad_norm": 1.152759313583374,
+      "learning_rate": 3.8671875e-05,
+      "loss": 1.0324,
+      "step": 855
+    },
+    {
+      "epoch": 0.4993437363278402,
+      "grad_norm": 1.1294784545898438,
+      "learning_rate": 3.865685096153847e-05,
+      "loss": 0.9175,
+      "step": 856
+    },
+    {
+      "epoch": 0.4999270818142045,
+      "grad_norm": 1.2857279777526855,
+      "learning_rate": 3.864182692307693e-05,
+      "loss": 1.2653,
+      "step": 857
+    },
+    {
+      "epoch": 0.5005104273005687,
+      "grad_norm": 1.9902052879333496,
+      "learning_rate": 3.8626802884615385e-05,
+      "loss": 0.9604,
+      "step": 858
+    },
+    {
+      "epoch": 0.501093772786933,
+      "grad_norm": 1.4721367359161377,
+      "learning_rate": 3.8611778846153844e-05,
+      "loss": 1.066,
+      "step": 859
+    },
+    {
+      "epoch": 0.5016771182732973,
+      "grad_norm": 1.022174596786499,
+      "learning_rate": 3.859675480769231e-05,
+      "loss": 0.9604,
+      "step": 860
+    },
+    {
+      "epoch": 0.5022604637596617,
+      "grad_norm": 1.3919785022735596,
+      "learning_rate": 3.8581730769230775e-05,
+      "loss": 0.9068,
+      "step": 861
+    },
+    {
+      "epoch": 0.502843809246026,
+      "grad_norm": 1.2962651252746582,
+      "learning_rate": 3.856670673076923e-05,
+      "loss": 1.1166,
+      "step": 862
+    },
+    {
+      "epoch": 0.5034271547323903,
+      "grad_norm": 1.2550551891326904,
+      "learning_rate": 3.855168269230769e-05,
+      "loss": 0.9605,
+      "step": 863
+    },
+    {
+      "epoch": 0.5040105002187546,
+      "grad_norm": 1.1890307664871216,
+      "learning_rate": 3.853665865384616e-05,
+      "loss": 1.1311,
+      "step": 864
+    },
+    {
+      "epoch": 0.5045938457051189,
+      "grad_norm": 1.2987920045852661,
+      "learning_rate": 3.8521634615384615e-05,
+      "loss": 1.0043,
+      "step": 865
+    },
+    {
+      "epoch": 0.5051771911914832,
+      "grad_norm": 1.0546956062316895,
+      "learning_rate": 3.850661057692308e-05,
+      "loss": 1.1154,
+      "step": 866
+    },
+    {
+      "epoch": 0.5057605366778475,
+      "grad_norm": 1.1739052534103394,
+      "learning_rate": 3.849158653846154e-05,
+      "loss": 1.0501,
+      "step": 867
+    },
+    {
+      "epoch": 0.5063438821642118,
+      "grad_norm": 1.171311378479004,
+      "learning_rate": 3.8476562500000004e-05,
+      "loss": 1.109,
+      "step": 868
+    },
+    {
+      "epoch": 0.5069272276505761,
+      "grad_norm": 1.0606648921966553,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.9756,
+      "step": 869
+    },
+    {
+      "epoch": 0.5075105731369404,
+      "grad_norm": 1.1545205116271973,
+      "learning_rate": 3.844651442307692e-05,
+      "loss": 1.073,
+      "step": 870
+    },
+    {
+      "epoch": 0.5080939186233047,
+      "grad_norm": 1.2131757736206055,
+      "learning_rate": 3.843149038461539e-05,
+      "loss": 1.1398,
+      "step": 871
+    },
+    {
+      "epoch": 0.508677264109669,
+      "grad_norm": 1.140005111694336,
+      "learning_rate": 3.8416466346153845e-05,
+      "loss": 0.9707,
+      "step": 872
+    },
+    {
+      "epoch": 0.5092606095960333,
+      "grad_norm": 1.253605604171753,
+      "learning_rate": 3.840144230769231e-05,
+      "loss": 1.1389,
+      "step": 873
+    },
+    {
+      "epoch": 0.5098439550823975,
+      "grad_norm": 1.2395936250686646,
+      "learning_rate": 3.8386418269230776e-05,
+      "loss": 1.0647,
+      "step": 874
+    },
+    {
+      "epoch": 0.5104273005687618,
+      "grad_norm": 1.092179298400879,
+      "learning_rate": 3.8371394230769234e-05,
+      "loss": 1.0585,
+      "step": 875
+    },
+    {
+      "epoch": 0.5110106460551261,
+      "grad_norm": 1.6231690645217896,
+      "learning_rate": 3.835637019230769e-05,
+      "loss": 0.8506,
+      "step": 876
+    },
+    {
+      "epoch": 0.5115939915414904,
+      "grad_norm": 1.1169666051864624,
+      "learning_rate": 3.834134615384616e-05,
+      "loss": 0.8958,
+      "step": 877
+    },
+    {
+      "epoch": 0.5121773370278547,
+      "grad_norm": 1.2263548374176025,
+      "learning_rate": 3.8326322115384616e-05,
+      "loss": 0.9337,
+      "step": 878
+    },
+    {
+      "epoch": 0.512760682514219,
+      "grad_norm": 1.1182410717010498,
+      "learning_rate": 3.831129807692308e-05,
+      "loss": 0.8743,
+      "step": 879
+    },
+    {
+      "epoch": 0.5133440280005833,
+      "grad_norm": 1.1924457550048828,
+      "learning_rate": 3.829627403846154e-05,
+      "loss": 0.9632,
+      "step": 880
+    },
+    {
+      "epoch": 0.5139273734869476,
+      "grad_norm": 1.2248971462249756,
+      "learning_rate": 3.828125e-05,
+      "loss": 1.1191,
+      "step": 881
+    },
+    {
+      "epoch": 0.5145107189733119,
+      "grad_norm": 1.1758016347885132,
+      "learning_rate": 3.8266225961538464e-05,
+      "loss": 0.9399,
+      "step": 882
+    },
+    {
+      "epoch": 0.5150940644596762,
+      "grad_norm": 1.1751339435577393,
+      "learning_rate": 3.825120192307692e-05,
+      "loss": 0.8132,
+      "step": 883
+    },
+    {
+      "epoch": 0.5156774099460405,
+      "grad_norm": 1.093839168548584,
+      "learning_rate": 3.823617788461539e-05,
+      "loss": 1.1629,
+      "step": 884
+    },
+    {
+      "epoch": 0.5162607554324048,
+      "grad_norm": 3.200582265853882,
+      "learning_rate": 3.8221153846153846e-05,
+      "loss": 1.1263,
+      "step": 885
+    },
+    {
+      "epoch": 0.5168441009187691,
+      "grad_norm": 1.2929569482803345,
+      "learning_rate": 3.820612980769231e-05,
+      "loss": 1.1846,
+      "step": 886
+    },
+    {
+      "epoch": 0.5174274464051335,
+      "grad_norm": 1.3224855661392212,
+      "learning_rate": 3.819110576923077e-05,
+      "loss": 0.9277,
+      "step": 887
+    },
+    {
+      "epoch": 0.5180107918914978,
+      "grad_norm": 1.1607059240341187,
+      "learning_rate": 3.817608173076923e-05,
+      "loss": 0.9946,
+      "step": 888
+    },
+    {
+      "epoch": 0.5185941373778621,
+      "grad_norm": 2.7251431941986084,
+      "learning_rate": 3.8161057692307694e-05,
+      "loss": 1.1296,
+      "step": 889
+    },
+    {
+      "epoch": 0.5191774828642264,
+      "grad_norm": 1.1163549423217773,
+      "learning_rate": 3.814603365384616e-05,
+      "loss": 0.9401,
+      "step": 890
+    },
+    {
+      "epoch": 0.5197608283505907,
+      "grad_norm": 1.1019116640090942,
+      "learning_rate": 3.813100961538462e-05,
+      "loss": 0.8071,
+      "step": 891
+    },
+    {
+      "epoch": 0.520344173836955,
+      "grad_norm": 1.1490522623062134,
+      "learning_rate": 3.811598557692308e-05,
+      "loss": 0.9441,
+      "step": 892
+    },
+    {
+      "epoch": 0.5209275193233193,
+      "grad_norm": 1.1176124811172485,
+      "learning_rate": 3.810096153846154e-05,
+      "loss": 1.0805,
+      "step": 893
+    },
+    {
+      "epoch": 0.5215108648096836,
+      "grad_norm": 1.4119200706481934,
+      "learning_rate": 3.80859375e-05,
+      "loss": 1.3708,
+      "step": 894
+    },
+    {
+      "epoch": 0.5220942102960479,
+      "grad_norm": 1.2920783758163452,
+      "learning_rate": 3.8070913461538465e-05,
+      "loss": 1.2315,
+      "step": 895
+    },
+    {
+      "epoch": 0.5226775557824122,
+      "grad_norm": 1.2457376718521118,
+      "learning_rate": 3.8055889423076924e-05,
+      "loss": 1.0076,
+      "step": 896
+    },
+    {
+      "epoch": 0.5232609012687764,
+      "grad_norm": 1.5919691324234009,
+      "learning_rate": 3.804086538461539e-05,
+      "loss": 0.9201,
+      "step": 897
+    },
+    {
+      "epoch": 0.5238442467551407,
+      "grad_norm": 1.3675076961517334,
+      "learning_rate": 3.802584134615385e-05,
+      "loss": 0.9394,
+      "step": 898
+    },
+    {
+      "epoch": 0.524427592241505,
+      "grad_norm": 1.148927927017212,
+      "learning_rate": 3.8010817307692306e-05,
+      "loss": 0.8612,
+      "step": 899
+    },
+    {
+      "epoch": 0.5250109377278693,
+      "grad_norm": 1.258003830909729,
+      "learning_rate": 3.799579326923077e-05,
+      "loss": 1.1393,
+      "step": 900
+    },
+    {
+      "epoch": 0.5255942832142336,
+      "grad_norm": 1.0879266262054443,
+      "learning_rate": 3.798076923076923e-05,
+      "loss": 1.2231,
+      "step": 901
+    },
+    {
+      "epoch": 0.5261776287005979,
+      "grad_norm": 1.3922462463378906,
+      "learning_rate": 3.7965745192307695e-05,
+      "loss": 1.0593,
+      "step": 902
+    },
+    {
+      "epoch": 0.5267609741869622,
+      "grad_norm": 1.225117564201355,
+      "learning_rate": 3.795072115384616e-05,
+      "loss": 1.2312,
+      "step": 903
+    },
+    {
+      "epoch": 0.5273443196733265,
+      "grad_norm": 1.3208086490631104,
+      "learning_rate": 3.793569711538462e-05,
+      "loss": 1.3092,
+      "step": 904
+    },
+    {
+      "epoch": 0.5279276651596908,
+      "grad_norm": 1.0285180807113647,
+      "learning_rate": 3.792067307692308e-05,
+      "loss": 0.9734,
+      "step": 905
+    },
+    {
+      "epoch": 0.5285110106460551,
+      "grad_norm": 1.4396753311157227,
+      "learning_rate": 3.7905649038461536e-05,
+      "loss": 1.2026,
+      "step": 906
+    },
+    {
+      "epoch": 0.5290943561324194,
+      "grad_norm": 0.9777195453643799,
+      "learning_rate": 3.7890625e-05,
+      "loss": 0.9129,
+      "step": 907
+    },
+    {
+      "epoch": 0.5296777016187837,
+      "grad_norm": 1.0864167213439941,
+      "learning_rate": 3.7875600961538466e-05,
+      "loss": 0.8855,
+      "step": 908
+    },
+    {
+      "epoch": 0.530261047105148,
+      "grad_norm": 1.0939980745315552,
+      "learning_rate": 3.7860576923076925e-05,
+      "loss": 1.0853,
+      "step": 909
+    },
+    {
+      "epoch": 0.5308443925915123,
+      "grad_norm": 1.1439995765686035,
+      "learning_rate": 3.784555288461539e-05,
+      "loss": 1.1568,
+      "step": 910
+    },
+    {
+      "epoch": 0.5314277380778766,
+      "grad_norm": 1.2120246887207031,
+      "learning_rate": 3.783052884615385e-05,
+      "loss": 0.9977,
+      "step": 911
+    },
+    {
+      "epoch": 0.5320110835642409,
+      "grad_norm": 1.0372374057769775,
+      "learning_rate": 3.781550480769231e-05,
+      "loss": 0.8251,
+      "step": 912
+    },
+    {
+      "epoch": 0.5325944290506052,
+      "grad_norm": 1.2008637189865112,
+      "learning_rate": 3.780048076923077e-05,
+      "loss": 1.0291,
+      "step": 913
+    },
+    {
+      "epoch": 0.5331777745369696,
+      "grad_norm": 1.475831151008606,
+      "learning_rate": 3.778545673076923e-05,
+      "loss": 1.0554,
+      "step": 914
+    },
+    {
+      "epoch": 0.5337611200233339,
+      "grad_norm": 1.3396097421646118,
+      "learning_rate": 3.7770432692307696e-05,
+      "loss": 1.0433,
+      "step": 915
+    },
+    {
+      "epoch": 0.5343444655096982,
+      "grad_norm": 1.40950345993042,
+      "learning_rate": 3.775540865384616e-05,
+      "loss": 0.963,
+      "step": 916
+    },
+    {
+      "epoch": 0.5349278109960625,
+      "grad_norm": 1.8577003479003906,
+      "learning_rate": 3.774038461538461e-05,
+      "loss": 0.747,
+      "step": 917
+    },
+    {
+      "epoch": 0.5355111564824268,
+      "grad_norm": 1.0990678071975708,
+      "learning_rate": 3.772536057692308e-05,
+      "loss": 0.869,
+      "step": 918
+    },
+    {
+      "epoch": 0.536094501968791,
+      "grad_norm": 1.4199498891830444,
+      "learning_rate": 3.771033653846154e-05,
+      "loss": 0.8398,
+      "step": 919
+    },
+    {
+      "epoch": 0.5366778474551553,
+      "grad_norm": 1.5259379148483276,
+      "learning_rate": 3.76953125e-05,
+      "loss": 1.0978,
+      "step": 920
+    },
+    {
+      "epoch": 0.5372611929415196,
+      "grad_norm": 1.208513617515564,
+      "learning_rate": 3.768028846153847e-05,
+      "loss": 0.9836,
+      "step": 921
+    },
+    {
+      "epoch": 0.5378445384278839,
+      "grad_norm": 1.3118703365325928,
+      "learning_rate": 3.7665264423076926e-05,
+      "loss": 0.9274,
+      "step": 922
+    },
+    {
+      "epoch": 0.5384278839142482,
+      "grad_norm": 1.2613270282745361,
+      "learning_rate": 3.7650240384615385e-05,
+      "loss": 1.095,
+      "step": 923
+    },
+    {
+      "epoch": 0.5390112294006125,
+      "grad_norm": 1.056458592414856,
+      "learning_rate": 3.763521634615384e-05,
+      "loss": 1.0447,
+      "step": 924
+    },
+    {
+      "epoch": 0.5395945748869768,
+      "grad_norm": 1.104421854019165,
+      "learning_rate": 3.762019230769231e-05,
+      "loss": 0.9413,
+      "step": 925
+    },
+    {
+      "epoch": 0.5401779203733411,
+      "grad_norm": 1.1966310739517212,
+      "learning_rate": 3.7605168269230774e-05,
+      "loss": 1.0595,
+      "step": 926
+    },
+    {
+      "epoch": 0.5407612658597054,
+      "grad_norm": 1.1758238077163696,
+      "learning_rate": 3.759014423076923e-05,
+      "loss": 1.0223,
+      "step": 927
+    },
+    {
+      "epoch": 0.5413446113460697,
+      "grad_norm": 1.850199580192566,
+      "learning_rate": 3.75751201923077e-05,
+      "loss": 1.0916,
+      "step": 928
+    },
+    {
+      "epoch": 0.541927956832434,
+      "grad_norm": 1.0883837938308716,
+      "learning_rate": 3.7560096153846156e-05,
+      "loss": 1.1043,
+      "step": 929
+    },
+    {
+      "epoch": 0.5425113023187983,
+      "grad_norm": 1.1901664733886719,
+      "learning_rate": 3.7545072115384614e-05,
+      "loss": 1.0474,
+      "step": 930
+    },
+    {
+      "epoch": 0.5430946478051626,
+      "grad_norm": 1.5722588300704956,
+      "learning_rate": 3.753004807692308e-05,
+      "loss": 0.9751,
+      "step": 931
+    },
+    {
+      "epoch": 0.5436779932915269,
+      "grad_norm": 1.2307900190353394,
+      "learning_rate": 3.751502403846154e-05,
+      "loss": 1.0156,
+      "step": 932
+    },
+    {
+      "epoch": 0.5442613387778912,
+      "grad_norm": 1.954958200454712,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.0679,
+      "step": 933
+    },
+    {
+      "epoch": 0.5448446842642555,
+      "grad_norm": 1.6473888158798218,
+      "learning_rate": 3.748497596153847e-05,
+      "loss": 0.9303,
+      "step": 934
+    },
+    {
+      "epoch": 0.5454280297506198,
+      "grad_norm": 1.390716314315796,
+      "learning_rate": 3.746995192307692e-05,
+      "loss": 1.0829,
+      "step": 935
+    },
+    {
+      "epoch": 0.5460113752369841,
+      "grad_norm": 1.2111623287200928,
+      "learning_rate": 3.7454927884615386e-05,
+      "loss": 1.1691,
+      "step": 936
+    },
+    {
+      "epoch": 0.5465947207233484,
+      "grad_norm": 1.2147167921066284,
+      "learning_rate": 3.7439903846153844e-05,
+      "loss": 1.0071,
+      "step": 937
+    },
+    {
+      "epoch": 0.5471780662097127,
+      "grad_norm": 1.4748153686523438,
+      "learning_rate": 3.742487980769231e-05,
+      "loss": 0.945,
+      "step": 938
+    },
+    {
+      "epoch": 0.547761411696077,
+      "grad_norm": 1.2974998950958252,
+      "learning_rate": 3.7409855769230775e-05,
+      "loss": 0.9262,
+      "step": 939
+    },
+    {
+      "epoch": 0.5483447571824414,
+      "grad_norm": 1.0850262641906738,
+      "learning_rate": 3.739483173076923e-05,
+      "loss": 0.8514,
+      "step": 940
+    },
+    {
+      "epoch": 0.5489281026688057,
+      "grad_norm": 1.409722924232483,
+      "learning_rate": 3.737980769230769e-05,
+      "loss": 1.0453,
+      "step": 941
+    },
+    {
+      "epoch": 0.54951144815517,
+      "grad_norm": 1.253421664237976,
+      "learning_rate": 3.736478365384616e-05,
+      "loss": 0.8305,
+      "step": 942
+    },
+    {
+      "epoch": 0.5500947936415342,
+      "grad_norm": 1.1697018146514893,
+      "learning_rate": 3.7349759615384616e-05,
+      "loss": 1.0097,
+      "step": 943
+    },
+    {
+      "epoch": 0.5506781391278985,
+      "grad_norm": 1.3280630111694336,
+      "learning_rate": 3.733473557692308e-05,
+      "loss": 1.1064,
+      "step": 944
+    },
+    {
+      "epoch": 0.5512614846142628,
+      "grad_norm": 1.7148746252059937,
+      "learning_rate": 3.731971153846154e-05,
+      "loss": 1.152,
+      "step": 945
+    },
+    {
+      "epoch": 0.5518448301006271,
+      "grad_norm": 1.2414015531539917,
+      "learning_rate": 3.7304687500000005e-05,
+      "loss": 1.4578,
+      "step": 946
+    },
+    {
+      "epoch": 0.5524281755869914,
+      "grad_norm": 1.5570706129074097,
+      "learning_rate": 3.728966346153846e-05,
+      "loss": 1.1254,
+      "step": 947
+    },
+    {
+      "epoch": 0.5530115210733557,
+      "grad_norm": 1.2762681245803833,
+      "learning_rate": 3.727463942307692e-05,
+      "loss": 1.0296,
+      "step": 948
+    },
+    {
+      "epoch": 0.55359486655972,
+      "grad_norm": 1.1472910642623901,
+      "learning_rate": 3.725961538461539e-05,
+      "loss": 0.9723,
+      "step": 949
+    },
+    {
+      "epoch": 0.5541782120460843,
+      "grad_norm": 1.429591417312622,
+      "learning_rate": 3.7244591346153845e-05,
+      "loss": 1.0805,
+      "step": 950
+    },
+    {
+      "epoch": 0.5547615575324486,
+      "grad_norm": 1.136590600013733,
+      "learning_rate": 3.722956730769231e-05,
+      "loss": 1.2051,
+      "step": 951
+    },
+    {
+      "epoch": 0.5553449030188129,
+      "grad_norm": 1.1960887908935547,
+      "learning_rate": 3.7214543269230776e-05,
+      "loss": 0.9221,
+      "step": 952
+    },
+    {
+      "epoch": 0.5559282485051772,
+      "grad_norm": 1.2155873775482178,
+      "learning_rate": 3.719951923076923e-05,
+      "loss": 0.8845,
+      "step": 953
+    },
+    {
+      "epoch": 0.5565115939915415,
+      "grad_norm": 1.4684885740280151,
+      "learning_rate": 3.718449519230769e-05,
+      "loss": 1.1217,
+      "step": 954
+    },
+    {
+      "epoch": 0.5570949394779058,
+      "grad_norm": 1.1232365369796753,
+      "learning_rate": 3.716947115384616e-05,
+      "loss": 0.96,
+      "step": 955
+    },
+    {
+      "epoch": 0.5576782849642701,
+      "grad_norm": 1.3967763185501099,
+      "learning_rate": 3.715444711538462e-05,
+      "loss": 1.1756,
+      "step": 956
+    },
+    {
+      "epoch": 0.5582616304506344,
+      "grad_norm": 1.3869478702545166,
+      "learning_rate": 3.713942307692308e-05,
+      "loss": 0.904,
+      "step": 957
+    },
+    {
+      "epoch": 0.5588449759369987,
+      "grad_norm": 1.5184909105300903,
+      "learning_rate": 3.712439903846154e-05,
+      "loss": 1.065,
+      "step": 958
+    },
+    {
+      "epoch": 0.559428321423363,
+      "grad_norm": 1.1123069524765015,
+      "learning_rate": 3.7109375e-05,
+      "loss": 0.9782,
+      "step": 959
+    },
+    {
+      "epoch": 0.5600116669097273,
+      "grad_norm": 1.0042744874954224,
+      "learning_rate": 3.7094350961538464e-05,
+      "loss": 1.0655,
+      "step": 960
+    },
+    {
+      "epoch": 0.5605950123960916,
+      "grad_norm": 1.103111743927002,
+      "learning_rate": 3.707932692307692e-05,
+      "loss": 1.0633,
+      "step": 961
+    },
+    {
+      "epoch": 0.5611783578824558,
+      "grad_norm": 1.1853861808776855,
+      "learning_rate": 3.706430288461539e-05,
+      "loss": 1.1423,
+      "step": 962
+    },
+    {
+      "epoch": 0.5617617033688201,
+      "grad_norm": 1.1413124799728394,
+      "learning_rate": 3.704927884615385e-05,
+      "loss": 0.9878,
+      "step": 963
+    },
+    {
+      "epoch": 0.5623450488551844,
+      "grad_norm": 1.1289325952529907,
+      "learning_rate": 3.703425480769231e-05,
+      "loss": 0.9265,
+      "step": 964
+    },
+    {
+      "epoch": 0.5629283943415487,
+      "grad_norm": 1.21286141872406,
+      "learning_rate": 3.701923076923077e-05,
+      "loss": 0.9536,
+      "step": 965
+    },
+    {
+      "epoch": 0.563511739827913,
+      "grad_norm": 1.1824188232421875,
+      "learning_rate": 3.700420673076923e-05,
+      "loss": 0.8876,
+      "step": 966
+    },
+    {
+      "epoch": 0.5640950853142774,
+      "grad_norm": 1.1528284549713135,
+      "learning_rate": 3.6989182692307694e-05,
+      "loss": 0.9186,
+      "step": 967
+    },
+    {
+      "epoch": 0.5646784308006417,
+      "grad_norm": 0.9889248609542847,
+      "learning_rate": 3.697415865384616e-05,
+      "loss": 1.0231,
+      "step": 968
+    },
+    {
+      "epoch": 0.565261776287006,
+      "grad_norm": 1.1389588117599487,
+      "learning_rate": 3.695913461538462e-05,
+      "loss": 0.9333,
+      "step": 969
+    },
+    {
+      "epoch": 0.5658451217733703,
+      "grad_norm": 1.3849660158157349,
+      "learning_rate": 3.694411057692308e-05,
+      "loss": 1.016,
+      "step": 970
+    },
+    {
+      "epoch": 0.5664284672597346,
+      "grad_norm": 1.1598434448242188,
+      "learning_rate": 3.6929086538461535e-05,
+      "loss": 1.2011,
+      "step": 971
+    },
+    {
+      "epoch": 0.5670118127460989,
+      "grad_norm": 1.3359456062316895,
+      "learning_rate": 3.69140625e-05,
+      "loss": 1.0756,
+      "step": 972
+    },
+    {
+      "epoch": 0.5675951582324632,
+      "grad_norm": 1.4474626779556274,
+      "learning_rate": 3.6899038461538466e-05,
+      "loss": 1.0561,
+      "step": 973
+    },
+    {
+      "epoch": 0.5681785037188275,
+      "grad_norm": 1.1154292821884155,
+      "learning_rate": 3.6884014423076924e-05,
+      "loss": 0.8292,
+      "step": 974
+    },
+    {
+      "epoch": 0.5687618492051918,
+      "grad_norm": 1.0911437273025513,
+      "learning_rate": 3.686899038461539e-05,
+      "loss": 0.9813,
+      "step": 975
+    },
+    {
+      "epoch": 0.5693451946915561,
+      "grad_norm": 1.3380028009414673,
+      "learning_rate": 3.685396634615385e-05,
+      "loss": 1.2311,
+      "step": 976
+    },
+    {
+      "epoch": 0.5699285401779204,
+      "grad_norm": 1.0942449569702148,
+      "learning_rate": 3.6838942307692306e-05,
+      "loss": 0.8704,
+      "step": 977
+    },
+    {
+      "epoch": 0.5705118856642847,
+      "grad_norm": 1.0522291660308838,
+      "learning_rate": 3.682391826923077e-05,
+      "loss": 0.7929,
+      "step": 978
+    },
+    {
+      "epoch": 0.571095231150649,
+      "grad_norm": 1.7213867902755737,
+      "learning_rate": 3.680889423076923e-05,
+      "loss": 1.2793,
+      "step": 979
+    },
+    {
+      "epoch": 0.5716785766370133,
+      "grad_norm": 1.2406498193740845,
+      "learning_rate": 3.6793870192307695e-05,
+      "loss": 0.8329,
+      "step": 980
+    },
+    {
+      "epoch": 0.5722619221233776,
+      "grad_norm": 1.3200228214263916,
+      "learning_rate": 3.677884615384616e-05,
+      "loss": 1.0274,
+      "step": 981
+    },
+    {
+      "epoch": 0.5728452676097419,
+      "grad_norm": 1.1541839838027954,
+      "learning_rate": 3.676382211538462e-05,
+      "loss": 0.9626,
+      "step": 982
+    },
+    {
+      "epoch": 0.5734286130961062,
+      "grad_norm": 1.1976728439331055,
+      "learning_rate": 3.674879807692308e-05,
+      "loss": 0.8529,
+      "step": 983
+    },
+    {
+      "epoch": 0.5740119585824704,
+      "grad_norm": 1.458693027496338,
+      "learning_rate": 3.6733774038461536e-05,
+      "loss": 1.1267,
+      "step": 984
+    },
+    {
+      "epoch": 0.5745953040688347,
+      "grad_norm": 1.5604653358459473,
+      "learning_rate": 3.671875e-05,
+      "loss": 1.0798,
+      "step": 985
+    },
+    {
+      "epoch": 0.575178649555199,
+      "grad_norm": 1.269396424293518,
+      "learning_rate": 3.670372596153847e-05,
+      "loss": 1.0479,
+      "step": 986
+    },
+    {
+      "epoch": 0.5757619950415633,
+      "grad_norm": 1.206891655921936,
+      "learning_rate": 3.6688701923076925e-05,
+      "loss": 0.7532,
+      "step": 987
+    },
+    {
+      "epoch": 0.5763453405279276,
+      "grad_norm": 1.3230568170547485,
+      "learning_rate": 3.667367788461539e-05,
+      "loss": 0.9374,
+      "step": 988
+    },
+    {
+      "epoch": 0.5769286860142919,
+      "grad_norm": 1.2090986967086792,
+      "learning_rate": 3.665865384615384e-05,
+      "loss": 0.8378,
+      "step": 989
+    },
+    {
+      "epoch": 0.5775120315006562,
+      "grad_norm": 1.1172541379928589,
+      "learning_rate": 3.664362980769231e-05,
+      "loss": 0.9294,
+      "step": 990
+    },
+    {
+      "epoch": 0.5780953769870205,
+      "grad_norm": 1.2599154710769653,
+      "learning_rate": 3.662860576923077e-05,
+      "loss": 1.0339,
+      "step": 991
+    },
+    {
+      "epoch": 0.5786787224733848,
+      "grad_norm": 1.5349094867706299,
+      "learning_rate": 3.661358173076923e-05,
+      "loss": 1.2734,
+      "step": 992
+    },
+    {
+      "epoch": 0.5792620679597492,
+      "grad_norm": 1.4437059164047241,
+      "learning_rate": 3.6598557692307697e-05,
+      "loss": 1.0486,
+      "step": 993
+    },
+    {
+      "epoch": 0.5798454134461135,
+      "grad_norm": 1.0442750453948975,
+      "learning_rate": 3.6583533653846155e-05,
+      "loss": 0.9691,
+      "step": 994
+    },
+    {
+      "epoch": 0.5804287589324778,
+      "grad_norm": 1.3448185920715332,
+      "learning_rate": 3.6568509615384614e-05,
+      "loss": 1.2887,
+      "step": 995
+    },
+    {
+      "epoch": 0.5810121044188421,
+      "grad_norm": 1.0911777019500732,
+      "learning_rate": 3.655348557692308e-05,
+      "loss": 1.0297,
+      "step": 996
+    },
+    {
+      "epoch": 0.5815954499052064,
+      "grad_norm": 1.1703611612319946,
+      "learning_rate": 3.653846153846154e-05,
+      "loss": 0.8926,
+      "step": 997
+    },
+    {
+      "epoch": 0.5821787953915707,
+      "grad_norm": 1.7544491291046143,
+      "learning_rate": 3.65234375e-05,
+      "loss": 1.1923,
+      "step": 998
+    },
+    {
+      "epoch": 0.582762140877935,
+      "grad_norm": 1.3902952671051025,
+      "learning_rate": 3.650841346153847e-05,
+      "loss": 0.8508,
+      "step": 999
+    },
+    {
+      "epoch": 0.5833454863642993,
+      "grad_norm": 1.1169430017471313,
+      "learning_rate": 3.6493389423076926e-05,
+      "loss": 0.9376,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5833454863642993,
+      "eval_loss_squad": 0.8289175107888878,
+      "eval_perplexity": 8.038335329537642,
+      "eval_perplexity_reconstruct": 1.9354374343602856,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5839288318506636,
+      "grad_norm": 1.0589932203292847,
+      "learning_rate": 3.6478365384615385e-05,
+      "loss": 0.8848,
+      "step": 1001
+    },
+    {
+      "epoch": 0.5845121773370279,
+      "grad_norm": 1.3385180234909058,
+      "learning_rate": 3.6463341346153843e-05,
+      "loss": 1.005,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5850955228233922,
+      "grad_norm": 1.3170318603515625,
+      "learning_rate": 3.644831730769231e-05,
+      "loss": 1.0717,
+      "step": 1003
+    },
+    {
+      "epoch": 0.5856788683097565,
+      "grad_norm": 1.134464144706726,
+      "learning_rate": 3.6433293269230774e-05,
+      "loss": 1.1363,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5862622137961208,
+      "grad_norm": 1.2407094240188599,
+      "learning_rate": 3.641826923076923e-05,
+      "loss": 0.8989,
+      "step": 1005
+    },
+    {
+      "epoch": 0.586845559282485,
+      "grad_norm": 1.040368914604187,
+      "learning_rate": 3.64032451923077e-05,
+      "loss": 0.7677,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5874289047688493,
+      "grad_norm": 1.2355806827545166,
+      "learning_rate": 3.6388221153846156e-05,
+      "loss": 1.0903,
+      "step": 1007
+    },
+    {
+      "epoch": 0.5880122502552136,
+      "grad_norm": 2.0347537994384766,
+      "learning_rate": 3.6373197115384615e-05,
+      "loss": 1.0069,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5885955957415779,
+      "grad_norm": 1.3925468921661377,
+      "learning_rate": 3.635817307692308e-05,
+      "loss": 1.1185,
+      "step": 1009
+    },
+    {
+      "epoch": 0.5891789412279422,
+      "grad_norm": 1.4087902307510376,
+      "learning_rate": 3.634314903846154e-05,
+      "loss": 1.0334,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5897622867143065,
+      "grad_norm": 1.5874099731445312,
+      "learning_rate": 3.6328125000000004e-05,
+      "loss": 1.1728,
+      "step": 1011
+    },
+    {
+      "epoch": 0.5903456322006708,
+      "grad_norm": 1.1461565494537354,
+      "learning_rate": 3.631310096153846e-05,
+      "loss": 1.1395,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5909289776870351,
+      "grad_norm": 1.3284554481506348,
+      "learning_rate": 3.629807692307692e-05,
+      "loss": 0.9558,
+      "step": 1013
+    },
+    {
+      "epoch": 0.5915123231733994,
+      "grad_norm": 1.1890766620635986,
+      "learning_rate": 3.6283052884615386e-05,
+      "loss": 0.8911,
+      "step": 1014
+    },
+    {
+      "epoch": 0.5920956686597637,
+      "grad_norm": 1.15906822681427,
+      "learning_rate": 3.6268028846153845e-05,
+      "loss": 0.8283,
+      "step": 1015
+    },
+    {
+      "epoch": 0.592679014146128,
+      "grad_norm": 2.1104965209960938,
+      "learning_rate": 3.625300480769231e-05,
+      "loss": 0.8631,
+      "step": 1016
+    },
+    {
+      "epoch": 0.5932623596324923,
+      "grad_norm": 1.0914028882980347,
+      "learning_rate": 3.6237980769230775e-05,
+      "loss": 0.9244,
+      "step": 1017
+    },
+    {
+      "epoch": 0.5938457051188566,
+      "grad_norm": 1.1843571662902832,
+      "learning_rate": 3.6222956730769234e-05,
+      "loss": 0.9798,
+      "step": 1018
+    },
+    {
+      "epoch": 0.5944290506052209,
+      "grad_norm": 2.413841724395752,
+      "learning_rate": 3.620793269230769e-05,
+      "loss": 0.9308,
+      "step": 1019
+    },
+    {
+      "epoch": 0.5950123960915853,
+      "grad_norm": 2.1389975547790527,
+      "learning_rate": 3.619290865384616e-05,
+      "loss": 1.0376,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5955957415779496,
+      "grad_norm": 1.1918492317199707,
+      "learning_rate": 3.6177884615384616e-05,
+      "loss": 0.998,
+      "step": 1021
+    },
+    {
+      "epoch": 0.5961790870643139,
+      "grad_norm": 1.1064860820770264,
+      "learning_rate": 3.616286057692308e-05,
+      "loss": 1.0828,
+      "step": 1022
+    },
+    {
+      "epoch": 0.5967624325506782,
+      "grad_norm": 1.1806142330169678,
+      "learning_rate": 3.614783653846154e-05,
+      "loss": 1.1629,
+      "step": 1023
+    },
+    {
+      "epoch": 0.5973457780370425,
+      "grad_norm": 1.266573429107666,
+      "learning_rate": 3.6132812500000005e-05,
+      "loss": 0.9503,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5979291235234068,
+      "grad_norm": 1.1975195407867432,
+      "learning_rate": 3.6117788461538463e-05,
+      "loss": 0.9898,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5985124690097711,
+      "grad_norm": 1.3326911926269531,
+      "learning_rate": 3.610276442307692e-05,
+      "loss": 1.0076,
+      "step": 1026
+    },
+    {
+      "epoch": 0.5990958144961354,
+      "grad_norm": 1.2400132417678833,
+      "learning_rate": 3.608774038461539e-05,
+      "loss": 1.0404,
+      "step": 1027
+    },
+    {
+      "epoch": 0.5996791599824997,
+      "grad_norm": 1.5326324701309204,
+      "learning_rate": 3.6072716346153846e-05,
+      "loss": 0.9418,
+      "step": 1028
+    },
+    {
+      "epoch": 0.600262505468864,
+      "grad_norm": 1.4190855026245117,
+      "learning_rate": 3.605769230769231e-05,
+      "loss": 1.0581,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6008458509552282,
+      "grad_norm": 1.4408974647521973,
+      "learning_rate": 3.604266826923077e-05,
+      "loss": 0.8873,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6014291964415925,
+      "grad_norm": 1.0051006078720093,
+      "learning_rate": 3.602764423076923e-05,
+      "loss": 0.9694,
+      "step": 1031
+    },
+    {
+      "epoch": 0.6020125419279568,
+      "grad_norm": 1.304622769355774,
+      "learning_rate": 3.601262019230769e-05,
+      "loss": 0.894,
+      "step": 1032
+    },
+    {
+      "epoch": 0.6025958874143211,
+      "grad_norm": 1.6079553365707397,
+      "learning_rate": 3.599759615384616e-05,
+      "loss": 1.134,
+      "step": 1033
+    },
+    {
+      "epoch": 0.6031792329006854,
+      "grad_norm": 1.3666532039642334,
+      "learning_rate": 3.598257211538462e-05,
+      "loss": 1.171,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6037625783870497,
+      "grad_norm": 1.1331239938735962,
+      "learning_rate": 3.596754807692308e-05,
+      "loss": 1.014,
+      "step": 1035
+    },
+    {
+      "epoch": 0.604345923873414,
+      "grad_norm": 1.153573989868164,
+      "learning_rate": 3.595252403846154e-05,
+      "loss": 1.1011,
+      "step": 1036
+    },
+    {
+      "epoch": 0.6049292693597783,
+      "grad_norm": 1.182096004486084,
+      "learning_rate": 3.59375e-05,
+      "loss": 1.1392,
+      "step": 1037
+    },
+    {
+      "epoch": 0.6055126148461426,
+      "grad_norm": 1.1555769443511963,
+      "learning_rate": 3.5922475961538465e-05,
+      "loss": 0.8643,
+      "step": 1038
+    },
+    {
+      "epoch": 0.6060959603325069,
+      "grad_norm": 1.103007197380066,
+      "learning_rate": 3.590745192307692e-05,
+      "loss": 0.8785,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6066793058188712,
+      "grad_norm": 1.2092280387878418,
+      "learning_rate": 3.589242788461539e-05,
+      "loss": 1.0269,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6072626513052355,
+      "grad_norm": 1.2705990076065063,
+      "learning_rate": 3.587740384615385e-05,
+      "loss": 0.8287,
+      "step": 1041
+    },
+    {
+      "epoch": 0.6078459967915998,
+      "grad_norm": 1.1593817472457886,
+      "learning_rate": 3.586237980769231e-05,
+      "loss": 0.8295,
+      "step": 1042
+    },
+    {
+      "epoch": 0.6084293422779641,
+      "grad_norm": 1.1436785459518433,
+      "learning_rate": 3.584735576923077e-05,
+      "loss": 0.9481,
+      "step": 1043
+    },
+    {
+      "epoch": 0.6090126877643284,
+      "grad_norm": 1.4250012636184692,
+      "learning_rate": 3.583233173076923e-05,
+      "loss": 1.2019,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6095960332506927,
+      "grad_norm": 1.3171687126159668,
+      "learning_rate": 3.5817307692307695e-05,
+      "loss": 0.8549,
+      "step": 1045
+    },
+    {
+      "epoch": 0.6101793787370571,
+      "grad_norm": 1.4231929779052734,
+      "learning_rate": 3.580228365384616e-05,
+      "loss": 1.1054,
+      "step": 1046
+    },
+    {
+      "epoch": 0.6107627242234214,
+      "grad_norm": 1.2546910047531128,
+      "learning_rate": 3.578725961538462e-05,
+      "loss": 1.0624,
+      "step": 1047
+    },
+    {
+      "epoch": 0.6113460697097857,
+      "grad_norm": 1.1451596021652222,
+      "learning_rate": 3.577223557692308e-05,
+      "loss": 1.1191,
+      "step": 1048
+    },
+    {
+      "epoch": 0.61192941519615,
+      "grad_norm": 1.250524640083313,
+      "learning_rate": 3.5757211538461535e-05,
+      "loss": 1.0673,
+      "step": 1049
+    },
+    {
+      "epoch": 0.6125127606825143,
+      "grad_norm": 1.2106703519821167,
+      "learning_rate": 3.57421875e-05,
+      "loss": 0.9558,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6130961061688786,
+      "grad_norm": 1.4316116571426392,
+      "learning_rate": 3.5727163461538466e-05,
+      "loss": 1.146,
+      "step": 1051
+    },
+    {
+      "epoch": 0.6136794516552428,
+      "grad_norm": 1.1764512062072754,
+      "learning_rate": 3.5712139423076924e-05,
+      "loss": 0.8677,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6142627971416071,
+      "grad_norm": 1.0635274648666382,
+      "learning_rate": 3.569711538461539e-05,
+      "loss": 0.9785,
+      "step": 1053
+    },
+    {
+      "epoch": 0.6148461426279714,
+      "grad_norm": 1.1735188961029053,
+      "learning_rate": 3.568209134615385e-05,
+      "loss": 0.8116,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6154294881143357,
+      "grad_norm": 1.1931421756744385,
+      "learning_rate": 3.566706730769231e-05,
+      "loss": 1.0047,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6160128336007,
+      "grad_norm": 1.4057892560958862,
+      "learning_rate": 3.565204326923077e-05,
+      "loss": 0.9297,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6165961790870643,
+      "grad_norm": 1.0752261877059937,
+      "learning_rate": 3.563701923076923e-05,
+      "loss": 0.9164,
+      "step": 1057
+    },
+    {
+      "epoch": 0.6171795245734286,
+      "grad_norm": 1.052547574043274,
+      "learning_rate": 3.5621995192307696e-05,
+      "loss": 1.0254,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6177628700597929,
+      "grad_norm": 1.1009021997451782,
+      "learning_rate": 3.560697115384616e-05,
+      "loss": 0.888,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6183462155461572,
+      "grad_norm": 1.2356963157653809,
+      "learning_rate": 3.559194711538462e-05,
+      "loss": 1.1234,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6189295610325215,
+      "grad_norm": 1.765424370765686,
+      "learning_rate": 3.557692307692308e-05,
+      "loss": 1.2263,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6195129065188858,
+      "grad_norm": 1.3897560834884644,
+      "learning_rate": 3.5561899038461536e-05,
+      "loss": 1.0729,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6200962520052501,
+      "grad_norm": 1.1901566982269287,
+      "learning_rate": 3.5546875e-05,
+      "loss": 0.9953,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6206795974916144,
+      "grad_norm": 1.4633679389953613,
+      "learning_rate": 3.553185096153847e-05,
+      "loss": 0.9616,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6212629429779787,
+      "grad_norm": 1.3287920951843262,
+      "learning_rate": 3.5516826923076926e-05,
+      "loss": 1.0438,
+      "step": 1065
+    },
+    {
+      "epoch": 0.621846288464343,
+      "grad_norm": 1.7268593311309814,
+      "learning_rate": 3.5501802884615384e-05,
+      "loss": 0.8745,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6224296339507073,
+      "grad_norm": 1.0686813592910767,
+      "learning_rate": 3.548677884615384e-05,
+      "loss": 0.8533,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6230129794370716,
+      "grad_norm": 1.2637293338775635,
+      "learning_rate": 3.547175480769231e-05,
+      "loss": 0.8116,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6235963249234359,
+      "grad_norm": 1.3439655303955078,
+      "learning_rate": 3.545673076923077e-05,
+      "loss": 1.1462,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6241796704098002,
+      "grad_norm": 1.277295708656311,
+      "learning_rate": 3.544170673076923e-05,
+      "loss": 0.9344,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6247630158961645,
+      "grad_norm": 1.2058827877044678,
+      "learning_rate": 3.54266826923077e-05,
+      "loss": 1.0197,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6253463613825287,
+      "grad_norm": 1.288257360458374,
+      "learning_rate": 3.5411658653846155e-05,
+      "loss": 0.925,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6259297068688932,
+      "grad_norm": 1.1143733263015747,
+      "learning_rate": 3.5396634615384614e-05,
+      "loss": 1.0083,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6265130523552574,
+      "grad_norm": 1.0898163318634033,
+      "learning_rate": 3.538161057692308e-05,
+      "loss": 1.0865,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6270963978416217,
+      "grad_norm": 1.1873056888580322,
+      "learning_rate": 3.536658653846154e-05,
+      "loss": 1.0579,
+      "step": 1075
+    },
+    {
+      "epoch": 0.627679743327986,
+      "grad_norm": 1.0526707172393799,
+      "learning_rate": 3.53515625e-05,
+      "loss": 1.0129,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6282630888143503,
+      "grad_norm": 1.6970293521881104,
+      "learning_rate": 3.533653846153847e-05,
+      "loss": 1.1972,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6288464343007146,
+      "grad_norm": 1.2696473598480225,
+      "learning_rate": 3.532151442307693e-05,
+      "loss": 0.8886,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6294297797870789,
+      "grad_norm": 1.1362708806991577,
+      "learning_rate": 3.5306490384615385e-05,
+      "loss": 1.1162,
+      "step": 1079
+    },
+    {
+      "epoch": 0.6300131252734432,
+      "grad_norm": 1.0971933603286743,
+      "learning_rate": 3.5291466346153844e-05,
+      "loss": 0.9418,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6305964707598075,
+      "grad_norm": 1.5346184968948364,
+      "learning_rate": 3.527644230769231e-05,
+      "loss": 1.3067,
+      "step": 1081
+    },
+    {
+      "epoch": 0.6311798162461718,
+      "grad_norm": 1.10757315158844,
+      "learning_rate": 3.5261418269230774e-05,
+      "loss": 0.9629,
+      "step": 1082
+    },
+    {
+      "epoch": 0.6317631617325361,
+      "grad_norm": 1.1153584718704224,
+      "learning_rate": 3.524639423076923e-05,
+      "loss": 0.9693,
+      "step": 1083
+    },
+    {
+      "epoch": 0.6323465072189004,
+      "grad_norm": 1.2248104810714722,
+      "learning_rate": 3.523137019230769e-05,
+      "loss": 0.9687,
+      "step": 1084
+    },
+    {
+      "epoch": 0.6329298527052647,
+      "grad_norm": 1.2040361166000366,
+      "learning_rate": 3.5216346153846157e-05,
+      "loss": 0.8451,
+      "step": 1085
+    },
+    {
+      "epoch": 0.633513198191629,
+      "grad_norm": 1.2189350128173828,
+      "learning_rate": 3.5201322115384615e-05,
+      "loss": 0.8775,
+      "step": 1086
+    },
+    {
+      "epoch": 0.6340965436779933,
+      "grad_norm": 1.453861117362976,
+      "learning_rate": 3.518629807692308e-05,
+      "loss": 0.9802,
+      "step": 1087
+    },
+    {
+      "epoch": 0.6346798891643576,
+      "grad_norm": 1.6878211498260498,
+      "learning_rate": 3.517127403846154e-05,
+      "loss": 0.8514,
+      "step": 1088
+    },
+    {
+      "epoch": 0.6352632346507219,
+      "grad_norm": 1.2242681980133057,
+      "learning_rate": 3.5156250000000004e-05,
+      "loss": 0.7271,
+      "step": 1089
+    },
+    {
+      "epoch": 0.6358465801370862,
+      "grad_norm": 1.1206653118133545,
+      "learning_rate": 3.514122596153846e-05,
+      "loss": 1.0562,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6364299256234505,
+      "grad_norm": 1.2914477586746216,
+      "learning_rate": 3.512620192307692e-05,
+      "loss": 0.943,
+      "step": 1091
+    },
+    {
+      "epoch": 0.6370132711098148,
+      "grad_norm": 1.4054170846939087,
+      "learning_rate": 3.5111177884615386e-05,
+      "loss": 1.064,
+      "step": 1092
+    },
+    {
+      "epoch": 0.637596616596179,
+      "grad_norm": 1.1132476329803467,
+      "learning_rate": 3.5096153846153845e-05,
+      "loss": 1.0468,
+      "step": 1093
+    },
+    {
+      "epoch": 0.6381799620825434,
+      "grad_norm": 1.4797582626342773,
+      "learning_rate": 3.508112980769231e-05,
+      "loss": 1.0369,
+      "step": 1094
+    },
+    {
+      "epoch": 0.6387633075689076,
+      "grad_norm": 1.1891480684280396,
+      "learning_rate": 3.5066105769230775e-05,
+      "loss": 0.93,
+      "step": 1095
+    },
+    {
+      "epoch": 0.6393466530552719,
+      "grad_norm": 1.1823198795318604,
+      "learning_rate": 3.5051081730769234e-05,
+      "loss": 1.2398,
+      "step": 1096
+    },
+    {
+      "epoch": 0.6399299985416362,
+      "grad_norm": 1.0506694316864014,
+      "learning_rate": 3.503605769230769e-05,
+      "loss": 1.0028,
+      "step": 1097
+    },
+    {
+      "epoch": 0.6405133440280005,
+      "grad_norm": 1.1625219583511353,
+      "learning_rate": 3.502103365384616e-05,
+      "loss": 0.9668,
+      "step": 1098
+    },
+    {
+      "epoch": 0.6410966895143649,
+      "grad_norm": 1.3133008480072021,
+      "learning_rate": 3.5006009615384616e-05,
+      "loss": 0.8842,
+      "step": 1099
+    },
+    {
+      "epoch": 0.6416800350007292,
+      "grad_norm": 1.2925740480422974,
+      "learning_rate": 3.499098557692308e-05,
+      "loss": 0.9443,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6422633804870935,
+      "grad_norm": 1.1056509017944336,
+      "learning_rate": 3.497596153846154e-05,
+      "loss": 1.0339,
+      "step": 1101
+    },
+    {
+      "epoch": 0.6428467259734578,
+      "grad_norm": 1.4267328977584839,
+      "learning_rate": 3.49609375e-05,
+      "loss": 0.9475,
+      "step": 1102
+    },
+    {
+      "epoch": 0.6434300714598221,
+      "grad_norm": 1.0847243070602417,
+      "learning_rate": 3.4945913461538464e-05,
+      "loss": 1.0789,
+      "step": 1103
+    },
+    {
+      "epoch": 0.6440134169461864,
+      "grad_norm": 1.2231626510620117,
+      "learning_rate": 3.493088942307692e-05,
+      "loss": 0.9817,
+      "step": 1104
+    },
+    {
+      "epoch": 0.6445967624325507,
+      "grad_norm": 1.1734437942504883,
+      "learning_rate": 3.491586538461539e-05,
+      "loss": 1.0806,
+      "step": 1105
+    },
+    {
+      "epoch": 0.645180107918915,
+      "grad_norm": 1.1729824542999268,
+      "learning_rate": 3.4900841346153846e-05,
+      "loss": 1.0979,
+      "step": 1106
+    },
+    {
+      "epoch": 0.6457634534052793,
+      "grad_norm": 1.1877245903015137,
+      "learning_rate": 3.488581730769231e-05,
+      "loss": 1.0407,
+      "step": 1107
+    },
+    {
+      "epoch": 0.6463467988916436,
+      "grad_norm": 1.0097910165786743,
+      "learning_rate": 3.487079326923077e-05,
+      "loss": 0.9112,
+      "step": 1108
+    },
+    {
+      "epoch": 0.6469301443780079,
+      "grad_norm": 1.1973975896835327,
+      "learning_rate": 3.485576923076923e-05,
+      "loss": 0.873,
+      "step": 1109
+    },
+    {
+      "epoch": 0.6475134898643722,
+      "grad_norm": 1.1800600290298462,
+      "learning_rate": 3.4840745192307694e-05,
+      "loss": 0.9863,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6480968353507365,
+      "grad_norm": 1.5829706192016602,
+      "learning_rate": 3.482572115384616e-05,
+      "loss": 0.9469,
+      "step": 1111
+    },
+    {
+      "epoch": 0.6486801808371008,
+      "grad_norm": 1.5107544660568237,
+      "learning_rate": 3.481069711538462e-05,
+      "loss": 0.8981,
+      "step": 1112
+    },
+    {
+      "epoch": 0.6492635263234651,
+      "grad_norm": 1.1659681797027588,
+      "learning_rate": 3.479567307692308e-05,
+      "loss": 1.0151,
+      "step": 1113
+    },
+    {
+      "epoch": 0.6498468718098294,
+      "grad_norm": 1.2110625505447388,
+      "learning_rate": 3.478064903846154e-05,
+      "loss": 0.8399,
+      "step": 1114
+    },
+    {
+      "epoch": 0.6504302172961937,
+      "grad_norm": 1.2130272388458252,
+      "learning_rate": 3.4765625e-05,
+      "loss": 0.9638,
+      "step": 1115
+    },
+    {
+      "epoch": 0.651013562782558,
+      "grad_norm": 1.4531303644180298,
+      "learning_rate": 3.4750600961538465e-05,
+      "loss": 0.9451,
+      "step": 1116
+    },
+    {
+      "epoch": 0.6515969082689222,
+      "grad_norm": 1.2456012964248657,
+      "learning_rate": 3.4735576923076923e-05,
+      "loss": 0.9804,
+      "step": 1117
+    },
+    {
+      "epoch": 0.6521802537552865,
+      "grad_norm": 1.8596121072769165,
+      "learning_rate": 3.472055288461539e-05,
+      "loss": 1.0599,
+      "step": 1118
+    },
+    {
+      "epoch": 0.6527635992416508,
+      "grad_norm": 1.2399401664733887,
+      "learning_rate": 3.470552884615385e-05,
+      "loss": 0.978,
+      "step": 1119
+    },
+    {
+      "epoch": 0.6533469447280151,
+      "grad_norm": 1.25180983543396,
+      "learning_rate": 3.4690504807692306e-05,
+      "loss": 1.1725,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6539302902143794,
+      "grad_norm": 1.2084770202636719,
+      "learning_rate": 3.467548076923077e-05,
+      "loss": 0.8519,
+      "step": 1121
+    },
+    {
+      "epoch": 0.6545136357007437,
+      "grad_norm": 1.1235297918319702,
+      "learning_rate": 3.466045673076923e-05,
+      "loss": 0.9979,
+      "step": 1122
+    },
+    {
+      "epoch": 0.655096981187108,
+      "grad_norm": 1.1955933570861816,
+      "learning_rate": 3.4645432692307695e-05,
+      "loss": 1.2237,
+      "step": 1123
+    },
+    {
+      "epoch": 0.6556803266734723,
+      "grad_norm": 1.144810676574707,
+      "learning_rate": 3.463040865384616e-05,
+      "loss": 0.7198,
+      "step": 1124
+    },
+    {
+      "epoch": 0.6562636721598366,
+      "grad_norm": 1.1226835250854492,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 1.0139,
+      "step": 1125
+    },
+    {
+      "epoch": 0.656847017646201,
+      "grad_norm": 1.0557211637496948,
+      "learning_rate": 3.460036057692308e-05,
+      "loss": 1.2123,
+      "step": 1126
+    },
+    {
+      "epoch": 0.6574303631325653,
+      "grad_norm": 1.051958441734314,
+      "learning_rate": 3.4585336538461536e-05,
+      "loss": 1.1768,
+      "step": 1127
+    },
+    {
+      "epoch": 0.6580137086189296,
+      "grad_norm": 1.0615653991699219,
+      "learning_rate": 3.45703125e-05,
+      "loss": 0.9798,
+      "step": 1128
+    },
+    {
+      "epoch": 0.6585970541052939,
+      "grad_norm": 1.022555947303772,
+      "learning_rate": 3.4555288461538466e-05,
+      "loss": 0.7892,
+      "step": 1129
+    },
+    {
+      "epoch": 0.6591803995916582,
+      "grad_norm": 1.3960226774215698,
+      "learning_rate": 3.4540264423076925e-05,
+      "loss": 0.922,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6597637450780225,
+      "grad_norm": 1.0755395889282227,
+      "learning_rate": 3.452524038461539e-05,
+      "loss": 1.0775,
+      "step": 1131
+    },
+    {
+      "epoch": 0.6603470905643868,
+      "grad_norm": 1.075884222984314,
+      "learning_rate": 3.451021634615385e-05,
+      "loss": 1.149,
+      "step": 1132
+    },
+    {
+      "epoch": 0.6609304360507511,
+      "grad_norm": 1.4113826751708984,
+      "learning_rate": 3.449519230769231e-05,
+      "loss": 0.8611,
+      "step": 1133
+    },
+    {
+      "epoch": 0.6615137815371154,
+      "grad_norm": 1.0343314409255981,
+      "learning_rate": 3.448016826923077e-05,
+      "loss": 0.8762,
+      "step": 1134
+    },
+    {
+      "epoch": 0.6620971270234797,
+      "grad_norm": 1.0894453525543213,
+      "learning_rate": 3.446514423076923e-05,
+      "loss": 1.0281,
+      "step": 1135
+    },
+    {
+      "epoch": 0.662680472509844,
+      "grad_norm": 1.2878773212432861,
+      "learning_rate": 3.4450120192307696e-05,
+      "loss": 1.0719,
+      "step": 1136
+    },
+    {
+      "epoch": 0.6632638179962083,
+      "grad_norm": 1.1200342178344727,
+      "learning_rate": 3.443509615384616e-05,
+      "loss": 1.0834,
+      "step": 1137
+    },
+    {
+      "epoch": 0.6638471634825726,
+      "grad_norm": 1.2115342617034912,
+      "learning_rate": 3.442007211538461e-05,
+      "loss": 1.1008,
+      "step": 1138
+    },
+    {
+      "epoch": 0.6644305089689369,
+      "grad_norm": 1.214706301689148,
+      "learning_rate": 3.440504807692308e-05,
+      "loss": 1.2134,
+      "step": 1139
+    },
+    {
+      "epoch": 0.6650138544553011,
+      "grad_norm": 1.3579165935516357,
+      "learning_rate": 3.439002403846154e-05,
+      "loss": 0.8706,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6655971999416654,
+      "grad_norm": 1.3200846910476685,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.9311,
+      "step": 1141
+    },
+    {
+      "epoch": 0.6661805454280297,
+      "grad_norm": 1.4525578022003174,
+      "learning_rate": 3.435997596153847e-05,
+      "loss": 1.0345,
+      "step": 1142
+    },
+    {
+      "epoch": 0.666763890914394,
+      "grad_norm": 1.2601604461669922,
+      "learning_rate": 3.4344951923076926e-05,
+      "loss": 1.1084,
+      "step": 1143
+    },
+    {
+      "epoch": 0.6673472364007583,
+      "grad_norm": 1.278788447380066,
+      "learning_rate": 3.4329927884615384e-05,
+      "loss": 0.9541,
+      "step": 1144
+    },
+    {
+      "epoch": 0.6679305818871226,
+      "grad_norm": 0.9493159651756287,
+      "learning_rate": 3.431490384615384e-05,
+      "loss": 0.9375,
+      "step": 1145
+    },
+    {
+      "epoch": 0.6685139273734869,
+      "grad_norm": 1.0539159774780273,
+      "learning_rate": 3.429987980769231e-05,
+      "loss": 0.9884,
+      "step": 1146
+    },
+    {
+      "epoch": 0.6690972728598512,
+      "grad_norm": 1.1678050756454468,
+      "learning_rate": 3.4284855769230773e-05,
+      "loss": 1.0402,
+      "step": 1147
+    },
+    {
+      "epoch": 0.6696806183462155,
+      "grad_norm": 1.2369019985198975,
+      "learning_rate": 3.426983173076923e-05,
+      "loss": 0.9254,
+      "step": 1148
+    },
+    {
+      "epoch": 0.6702639638325798,
+      "grad_norm": 1.539076566696167,
+      "learning_rate": 3.42548076923077e-05,
+      "loss": 1.1155,
+      "step": 1149
+    },
+    {
+      "epoch": 0.6708473093189441,
+      "grad_norm": 1.074963092803955,
+      "learning_rate": 3.4239783653846156e-05,
+      "loss": 1.0213,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6714306548053084,
+      "grad_norm": 1.2291454076766968,
+      "learning_rate": 3.4224759615384614e-05,
+      "loss": 0.898,
+      "step": 1151
+    },
+    {
+      "epoch": 0.6720140002916728,
+      "grad_norm": 1.1478317975997925,
+      "learning_rate": 3.420973557692308e-05,
+      "loss": 0.8589,
+      "step": 1152
+    },
+    {
+      "epoch": 0.6725973457780371,
+      "grad_norm": 1.096078872680664,
+      "learning_rate": 3.419471153846154e-05,
+      "loss": 0.8751,
+      "step": 1153
+    },
+    {
+      "epoch": 0.6731806912644014,
+      "grad_norm": 1.4119384288787842,
+      "learning_rate": 3.41796875e-05,
+      "loss": 1.0693,
+      "step": 1154
+    },
+    {
+      "epoch": 0.6737640367507657,
+      "grad_norm": 1.2378814220428467,
+      "learning_rate": 3.416466346153847e-05,
+      "loss": 1.2076,
+      "step": 1155
+    },
+    {
+      "epoch": 0.67434738223713,
+      "grad_norm": 1.2882436513900757,
+      "learning_rate": 3.414963942307692e-05,
+      "loss": 1.1177,
+      "step": 1156
+    },
+    {
+      "epoch": 0.6749307277234943,
+      "grad_norm": 1.1691817045211792,
+      "learning_rate": 3.4134615384615386e-05,
+      "loss": 1.0826,
+      "step": 1157
+    },
+    {
+      "epoch": 0.6755140732098586,
+      "grad_norm": 1.168468713760376,
+      "learning_rate": 3.4119591346153844e-05,
+      "loss": 0.9258,
+      "step": 1158
+    },
+    {
+      "epoch": 0.6760974186962229,
+      "grad_norm": 1.3444772958755493,
+      "learning_rate": 3.410456730769231e-05,
+      "loss": 1.1413,
+      "step": 1159
+    },
+    {
+      "epoch": 0.6766807641825872,
+      "grad_norm": 1.2808760404586792,
+      "learning_rate": 3.4089543269230775e-05,
+      "loss": 0.8778,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6772641096689515,
+      "grad_norm": 1.1354130506515503,
+      "learning_rate": 3.407451923076923e-05,
+      "loss": 0.8865,
+      "step": 1161
+    },
+    {
+      "epoch": 0.6778474551553157,
+      "grad_norm": 1.0770645141601562,
+      "learning_rate": 3.405949519230769e-05,
+      "loss": 1.0861,
+      "step": 1162
+    },
+    {
+      "epoch": 0.67843080064168,
+      "grad_norm": 1.1578465700149536,
+      "learning_rate": 3.404447115384616e-05,
+      "loss": 1.0487,
+      "step": 1163
+    },
+    {
+      "epoch": 0.6790141461280443,
+      "grad_norm": 1.0803139209747314,
+      "learning_rate": 3.4029447115384615e-05,
+      "loss": 0.8303,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6795974916144086,
+      "grad_norm": 1.0990511178970337,
+      "learning_rate": 3.401442307692308e-05,
+      "loss": 1.1653,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6801808371007729,
+      "grad_norm": 0.9813050627708435,
+      "learning_rate": 3.399939903846154e-05,
+      "loss": 0.9875,
+      "step": 1166
+    },
+    {
+      "epoch": 0.6807641825871372,
+      "grad_norm": 1.0541377067565918,
+      "learning_rate": 3.3984375000000004e-05,
+      "loss": 0.9924,
+      "step": 1167
+    },
+    {
+      "epoch": 0.6813475280735015,
+      "grad_norm": 1.2727155685424805,
+      "learning_rate": 3.396935096153846e-05,
+      "loss": 0.7893,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6819308735598658,
+      "grad_norm": 1.1019082069396973,
+      "learning_rate": 3.395432692307692e-05,
+      "loss": 1.2714,
+      "step": 1169
+    },
+    {
+      "epoch": 0.6825142190462301,
+      "grad_norm": 0.9809292554855347,
+      "learning_rate": 3.393930288461539e-05,
+      "loss": 0.8502,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6830975645325944,
+      "grad_norm": 0.8613129258155823,
+      "learning_rate": 3.3924278846153845e-05,
+      "loss": 0.8276,
+      "step": 1171
+    },
+    {
+      "epoch": 0.6836809100189587,
+      "grad_norm": 1.049072265625,
+      "learning_rate": 3.390925480769231e-05,
+      "loss": 1.0627,
+      "step": 1172
+    },
+    {
+      "epoch": 0.684264255505323,
+      "grad_norm": 1.205977201461792,
+      "learning_rate": 3.3894230769230776e-05,
+      "loss": 0.9767,
+      "step": 1173
+    },
+    {
+      "epoch": 0.6848476009916873,
+      "grad_norm": 1.2076629400253296,
+      "learning_rate": 3.387920673076923e-05,
+      "loss": 0.8472,
+      "step": 1174
+    },
+    {
+      "epoch": 0.6854309464780516,
+      "grad_norm": 1.244746208190918,
+      "learning_rate": 3.386418269230769e-05,
+      "loss": 1.0404,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6860142919644159,
+      "grad_norm": 1.0318353176116943,
+      "learning_rate": 3.384915865384616e-05,
+      "loss": 1.2889,
+      "step": 1176
+    },
+    {
+      "epoch": 0.6865976374507802,
+      "grad_norm": 1.1536865234375,
+      "learning_rate": 3.3834134615384617e-05,
+      "loss": 1.0932,
+      "step": 1177
+    },
+    {
+      "epoch": 0.6871809829371445,
+      "grad_norm": 1.0421112775802612,
+      "learning_rate": 3.381911057692308e-05,
+      "loss": 1.1513,
+      "step": 1178
+    },
+    {
+      "epoch": 0.6877643284235089,
+      "grad_norm": 1.2524076700210571,
+      "learning_rate": 3.380408653846154e-05,
+      "loss": 1.0571,
+      "step": 1179
+    },
+    {
+      "epoch": 0.6883476739098732,
+      "grad_norm": 1.3088963031768799,
+      "learning_rate": 3.37890625e-05,
+      "loss": 1.0539,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6889310193962375,
+      "grad_norm": 1.4658859968185425,
+      "learning_rate": 3.3774038461538464e-05,
+      "loss": 0.9256,
+      "step": 1181
+    },
+    {
+      "epoch": 0.6895143648826018,
+      "grad_norm": 1.3050382137298584,
+      "learning_rate": 3.375901442307692e-05,
+      "loss": 1.1349,
+      "step": 1182
+    },
+    {
+      "epoch": 0.690097710368966,
+      "grad_norm": 1.318977952003479,
+      "learning_rate": 3.374399038461539e-05,
+      "loss": 1.0711,
+      "step": 1183
+    },
+    {
+      "epoch": 0.6906810558553304,
+      "grad_norm": 1.2855241298675537,
+      "learning_rate": 3.3728966346153846e-05,
+      "loss": 0.9612,
+      "step": 1184
+    },
+    {
+      "epoch": 0.6912644013416946,
+      "grad_norm": 1.3077634572982788,
+      "learning_rate": 3.371394230769231e-05,
+      "loss": 1.0041,
+      "step": 1185
+    },
+    {
+      "epoch": 0.6918477468280589,
+      "grad_norm": 1.1256729364395142,
+      "learning_rate": 3.369891826923077e-05,
+      "loss": 1.0035,
+      "step": 1186
+    },
+    {
+      "epoch": 0.6924310923144232,
+      "grad_norm": 1.3386636972427368,
+      "learning_rate": 3.368389423076923e-05,
+      "loss": 0.9681,
+      "step": 1187
+    },
+    {
+      "epoch": 0.6930144378007875,
+      "grad_norm": 1.4480712413787842,
+      "learning_rate": 3.3668870192307694e-05,
+      "loss": 1.1093,
+      "step": 1188
+    },
+    {
+      "epoch": 0.6935977832871518,
+      "grad_norm": 1.1406118869781494,
+      "learning_rate": 3.365384615384616e-05,
+      "loss": 1.0623,
+      "step": 1189
+    },
+    {
+      "epoch": 0.6941811287735161,
+      "grad_norm": 1.1809027194976807,
+      "learning_rate": 3.363882211538462e-05,
+      "loss": 0.9942,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6947644742598804,
+      "grad_norm": 1.3855853080749512,
+      "learning_rate": 3.362379807692308e-05,
+      "loss": 0.9911,
+      "step": 1191
+    },
+    {
+      "epoch": 0.6953478197462447,
+      "grad_norm": 1.2072291374206543,
+      "learning_rate": 3.3608774038461535e-05,
+      "loss": 0.9015,
+      "step": 1192
+    },
+    {
+      "epoch": 0.695931165232609,
+      "grad_norm": 1.3127961158752441,
+      "learning_rate": 3.359375e-05,
+      "loss": 0.8459,
+      "step": 1193
+    },
+    {
+      "epoch": 0.6965145107189733,
+      "grad_norm": 1.1735903024673462,
+      "learning_rate": 3.3578725961538465e-05,
+      "loss": 0.9313,
+      "step": 1194
+    },
+    {
+      "epoch": 0.6970978562053376,
+      "grad_norm": 1.1430635452270508,
+      "learning_rate": 3.3563701923076924e-05,
+      "loss": 1.1029,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6976812016917019,
+      "grad_norm": 1.0707919597625732,
+      "learning_rate": 3.354867788461539e-05,
+      "loss": 1.0674,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6982645471780662,
+      "grad_norm": 1.1463159322738647,
+      "learning_rate": 3.353365384615385e-05,
+      "loss": 1.0293,
+      "step": 1197
+    },
+    {
+      "epoch": 0.6988478926644305,
+      "grad_norm": 0.9569932818412781,
+      "learning_rate": 3.3518629807692306e-05,
+      "loss": 0.726,
+      "step": 1198
+    },
+    {
+      "epoch": 0.6994312381507948,
+      "grad_norm": 1.4707874059677124,
+      "learning_rate": 3.350360576923077e-05,
+      "loss": 0.9595,
+      "step": 1199
+    },
+    {
+      "epoch": 0.7000145836371591,
+      "grad_norm": 1.0669324398040771,
+      "learning_rate": 3.348858173076923e-05,
+      "loss": 1.1805,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7000145836371591,
+      "eval_loss_squad": 0.8398913412541151,
+      "eval_perplexity": 8.243036213620654,
+      "eval_perplexity_reconstruct": 1.9400207914704957,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7005979291235234,
+      "grad_norm": 1.3081518411636353,
+      "learning_rate": 3.3473557692307695e-05,
+      "loss": 1.2301,
+      "step": 1201
+    },
+    {
+      "epoch": 0.7011812746098877,
+      "grad_norm": 1.2794771194458008,
+      "learning_rate": 3.345853365384616e-05,
+      "loss": 0.7746,
+      "step": 1202
+    },
+    {
+      "epoch": 0.701764620096252,
+      "grad_norm": 1.2433689832687378,
+      "learning_rate": 3.344350961538462e-05,
+      "loss": 0.9256,
+      "step": 1203
+    },
+    {
+      "epoch": 0.7023479655826163,
+      "grad_norm": 1.1298249959945679,
+      "learning_rate": 3.342848557692308e-05,
+      "loss": 1.0316,
+      "step": 1204
+    },
+    {
+      "epoch": 0.7029313110689807,
+      "grad_norm": 1.1544654369354248,
+      "learning_rate": 3.3413461538461536e-05,
+      "loss": 0.9899,
+      "step": 1205
+    },
+    {
+      "epoch": 0.703514656555345,
+      "grad_norm": 1.4191893339157104,
+      "learning_rate": 3.33984375e-05,
+      "loss": 0.8943,
+      "step": 1206
+    },
+    {
+      "epoch": 0.7040980020417092,
+      "grad_norm": 1.1830098628997803,
+      "learning_rate": 3.3383413461538466e-05,
+      "loss": 0.9173,
+      "step": 1207
+    },
+    {
+      "epoch": 0.7046813475280735,
+      "grad_norm": 1.1737726926803589,
+      "learning_rate": 3.3368389423076925e-05,
+      "loss": 0.9999,
+      "step": 1208
+    },
+    {
+      "epoch": 0.7052646930144378,
+      "grad_norm": 1.3882325887680054,
+      "learning_rate": 3.335336538461539e-05,
+      "loss": 1.0172,
+      "step": 1209
+    },
+    {
+      "epoch": 0.7058480385008021,
+      "grad_norm": 1.1368088722229004,
+      "learning_rate": 3.333834134615384e-05,
+      "loss": 0.9639,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7064313839871664,
+      "grad_norm": 1.0348821878433228,
+      "learning_rate": 3.332331730769231e-05,
+      "loss": 1.0874,
+      "step": 1211
+    },
+    {
+      "epoch": 0.7070147294735307,
+      "grad_norm": 1.3544259071350098,
+      "learning_rate": 3.330829326923077e-05,
+      "loss": 1.0331,
+      "step": 1212
+    },
+    {
+      "epoch": 0.707598074959895,
+      "grad_norm": 0.9908936619758606,
+      "learning_rate": 3.329326923076923e-05,
+      "loss": 0.7225,
+      "step": 1213
+    },
+    {
+      "epoch": 0.7081814204462593,
+      "grad_norm": 1.2547880411148071,
+      "learning_rate": 3.3278245192307696e-05,
+      "loss": 1.0878,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7087647659326236,
+      "grad_norm": 1.617482304573059,
+      "learning_rate": 3.3263221153846155e-05,
+      "loss": 1.0804,
+      "step": 1215
+    },
+    {
+      "epoch": 0.7093481114189879,
+      "grad_norm": 1.039589285850525,
+      "learning_rate": 3.324819711538461e-05,
+      "loss": 0.9664,
+      "step": 1216
+    },
+    {
+      "epoch": 0.7099314569053522,
+      "grad_norm": 1.2409045696258545,
+      "learning_rate": 3.323317307692308e-05,
+      "loss": 0.9162,
+      "step": 1217
+    },
+    {
+      "epoch": 0.7105148023917165,
+      "grad_norm": 1.286959171295166,
+      "learning_rate": 3.321814903846154e-05,
+      "loss": 1.032,
+      "step": 1218
+    },
+    {
+      "epoch": 0.7110981478780808,
+      "grad_norm": 1.3693422079086304,
+      "learning_rate": 3.3203125e-05,
+      "loss": 1.0232,
+      "step": 1219
+    },
+    {
+      "epoch": 0.7116814933644451,
+      "grad_norm": 1.3210753202438354,
+      "learning_rate": 3.318810096153847e-05,
+      "loss": 0.9577,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7122648388508094,
+      "grad_norm": 1.1538783311843872,
+      "learning_rate": 3.3173076923076926e-05,
+      "loss": 1.0804,
+      "step": 1221
+    },
+    {
+      "epoch": 0.7128481843371737,
+      "grad_norm": 1.1558293104171753,
+      "learning_rate": 3.3158052884615385e-05,
+      "loss": 1.1509,
+      "step": 1222
+    },
+    {
+      "epoch": 0.713431529823538,
+      "grad_norm": 0.9820008277893066,
+      "learning_rate": 3.314302884615384e-05,
+      "loss": 0.814,
+      "step": 1223
+    },
+    {
+      "epoch": 0.7140148753099023,
+      "grad_norm": 0.9828884601593018,
+      "learning_rate": 3.312800480769231e-05,
+      "loss": 0.8456,
+      "step": 1224
+    },
+    {
+      "epoch": 0.7145982207962666,
+      "grad_norm": 1.3056386709213257,
+      "learning_rate": 3.3112980769230774e-05,
+      "loss": 1.0279,
+      "step": 1225
+    },
+    {
+      "epoch": 0.7151815662826309,
+      "grad_norm": 1.1356948614120483,
+      "learning_rate": 3.309795673076923e-05,
+      "loss": 1.0234,
+      "step": 1226
+    },
+    {
+      "epoch": 0.7157649117689951,
+      "grad_norm": 1.2792552709579468,
+      "learning_rate": 3.30829326923077e-05,
+      "loss": 1.1383,
+      "step": 1227
+    },
+    {
+      "epoch": 0.7163482572553594,
+      "grad_norm": 1.0980956554412842,
+      "learning_rate": 3.3067908653846156e-05,
+      "loss": 1.0534,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7169316027417237,
+      "grad_norm": 1.2338215112686157,
+      "learning_rate": 3.3052884615384615e-05,
+      "loss": 0.9844,
+      "step": 1229
+    },
+    {
+      "epoch": 0.717514948228088,
+      "grad_norm": 1.4420562982559204,
+      "learning_rate": 3.303786057692308e-05,
+      "loss": 0.8789,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7180982937144523,
+      "grad_norm": 1.26366126537323,
+      "learning_rate": 3.302283653846154e-05,
+      "loss": 0.8189,
+      "step": 1231
+    },
+    {
+      "epoch": 0.7186816392008167,
+      "grad_norm": 1.1623914241790771,
+      "learning_rate": 3.3007812500000004e-05,
+      "loss": 0.9789,
+      "step": 1232
+    },
+    {
+      "epoch": 0.719264984687181,
+      "grad_norm": 1.0107698440551758,
+      "learning_rate": 3.299278846153846e-05,
+      "loss": 0.7271,
+      "step": 1233
+    },
+    {
+      "epoch": 0.7198483301735453,
+      "grad_norm": 1.185608148574829,
+      "learning_rate": 3.297776442307692e-05,
+      "loss": 1.1248,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7204316756599096,
+      "grad_norm": 1.077970027923584,
+      "learning_rate": 3.2962740384615386e-05,
+      "loss": 1.1963,
+      "step": 1235
+    },
+    {
+      "epoch": 0.7210150211462739,
+      "grad_norm": 1.684244990348816,
+      "learning_rate": 3.2947716346153844e-05,
+      "loss": 0.8828,
+      "step": 1236
+    },
+    {
+      "epoch": 0.7215983666326382,
+      "grad_norm": 1.24001944065094,
+      "learning_rate": 3.293269230769231e-05,
+      "loss": 0.8627,
+      "step": 1237
+    },
+    {
+      "epoch": 0.7221817121190025,
+      "grad_norm": 1.0665417909622192,
+      "learning_rate": 3.2917668269230775e-05,
+      "loss": 0.7553,
+      "step": 1238
+    },
+    {
+      "epoch": 0.7227650576053668,
+      "grad_norm": 1.1986167430877686,
+      "learning_rate": 3.2902644230769233e-05,
+      "loss": 0.7785,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7233484030917311,
+      "grad_norm": 1.313407301902771,
+      "learning_rate": 3.288762019230769e-05,
+      "loss": 0.9125,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7239317485780954,
+      "grad_norm": 1.1136894226074219,
+      "learning_rate": 3.287259615384616e-05,
+      "loss": 0.975,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7245150940644597,
+      "grad_norm": 1.2097238302230835,
+      "learning_rate": 3.2857572115384616e-05,
+      "loss": 0.8688,
+      "step": 1242
+    },
+    {
+      "epoch": 0.725098439550824,
+      "grad_norm": 1.280327320098877,
+      "learning_rate": 3.284254807692308e-05,
+      "loss": 0.9758,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7256817850371883,
+      "grad_norm": 1.2413567304611206,
+      "learning_rate": 3.282752403846154e-05,
+      "loss": 0.9494,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7262651305235526,
+      "grad_norm": 1.1193758249282837,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 1.0908,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7268484760099169,
+      "grad_norm": 1.1849331855773926,
+      "learning_rate": 3.279747596153846e-05,
+      "loss": 1.0299,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7274318214962812,
+      "grad_norm": 2.731739044189453,
+      "learning_rate": 3.278245192307692e-05,
+      "loss": 0.8046,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7280151669826455,
+      "grad_norm": 1.145367980003357,
+      "learning_rate": 3.276742788461539e-05,
+      "loss": 0.9687,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7285985124690098,
+      "grad_norm": 1.2098665237426758,
+      "learning_rate": 3.2752403846153846e-05,
+      "loss": 0.8909,
+      "step": 1249
+    },
+    {
+      "epoch": 0.729181857955374,
+      "grad_norm": 1.4284601211547852,
+      "learning_rate": 3.273737980769231e-05,
+      "loss": 0.8545,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7297652034417383,
+      "grad_norm": 1.4550679922103882,
+      "learning_rate": 3.272235576923077e-05,
+      "loss": 0.9656,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7303485489281026,
+      "grad_norm": 1.2722722291946411,
+      "learning_rate": 3.270733173076923e-05,
+      "loss": 0.8022,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7309318944144669,
+      "grad_norm": 1.3001720905303955,
+      "learning_rate": 3.269230769230769e-05,
+      "loss": 0.9008,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7315152399008312,
+      "grad_norm": 1.610422968864441,
+      "learning_rate": 3.267728365384616e-05,
+      "loss": 0.7881,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7320985853871955,
+      "grad_norm": 1.0116015672683716,
+      "learning_rate": 3.266225961538462e-05,
+      "loss": 0.7952,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7326819308735598,
+      "grad_norm": 1.4856303930282593,
+      "learning_rate": 3.264723557692308e-05,
+      "loss": 1.0552,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7332652763599241,
+      "grad_norm": 1.7719351053237915,
+      "learning_rate": 3.263221153846154e-05,
+      "loss": 1.058,
+      "step": 1257
+    },
+    {
+      "epoch": 0.7338486218462885,
+      "grad_norm": 1.1480412483215332,
+      "learning_rate": 3.26171875e-05,
+      "loss": 0.8906,
+      "step": 1258
+    },
+    {
+      "epoch": 0.7344319673326528,
+      "grad_norm": 1.2761352062225342,
+      "learning_rate": 3.2602163461538464e-05,
+      "loss": 0.8908,
+      "step": 1259
+    },
+    {
+      "epoch": 0.7350153128190171,
+      "grad_norm": 1.1891727447509766,
+      "learning_rate": 3.258713942307692e-05,
+      "loss": 0.9932,
+      "step": 1260
+    },
+    {
+      "epoch": 0.7355986583053814,
+      "grad_norm": 1.0514845848083496,
+      "learning_rate": 3.257211538461539e-05,
+      "loss": 1.1528,
+      "step": 1261
+    },
+    {
+      "epoch": 0.7361820037917457,
+      "grad_norm": 1.4285988807678223,
+      "learning_rate": 3.255709134615385e-05,
+      "loss": 0.8171,
+      "step": 1262
+    },
+    {
+      "epoch": 0.73676534927811,
+      "grad_norm": 1.2109655141830444,
+      "learning_rate": 3.254206730769231e-05,
+      "loss": 1.1154,
+      "step": 1263
+    },
+    {
+      "epoch": 0.7373486947644743,
+      "grad_norm": 1.417160153388977,
+      "learning_rate": 3.252704326923077e-05,
+      "loss": 1.0018,
+      "step": 1264
+    },
+    {
+      "epoch": 0.7379320402508386,
+      "grad_norm": 1.045836091041565,
+      "learning_rate": 3.251201923076923e-05,
+      "loss": 0.9084,
+      "step": 1265
+    },
+    {
+      "epoch": 0.7385153857372029,
+      "grad_norm": 1.0985413789749146,
+      "learning_rate": 3.2496995192307694e-05,
+      "loss": 0.7832,
+      "step": 1266
+    },
+    {
+      "epoch": 0.7390987312235672,
+      "grad_norm": 1.1846632957458496,
+      "learning_rate": 3.248197115384616e-05,
+      "loss": 0.9933,
+      "step": 1267
+    },
+    {
+      "epoch": 0.7396820767099315,
+      "grad_norm": 1.1051980257034302,
+      "learning_rate": 3.246694711538462e-05,
+      "loss": 1.2966,
+      "step": 1268
+    },
+    {
+      "epoch": 0.7402654221962958,
+      "grad_norm": 1.1988706588745117,
+      "learning_rate": 3.2451923076923077e-05,
+      "loss": 1.052,
+      "step": 1269
+    },
+    {
+      "epoch": 0.7408487676826601,
+      "grad_norm": 1.1623855829238892,
+      "learning_rate": 3.2436899038461535e-05,
+      "loss": 0.9502,
+      "step": 1270
+    },
+    {
+      "epoch": 0.7414321131690244,
+      "grad_norm": 1.6816089153289795,
+      "learning_rate": 3.2421875e-05,
+      "loss": 1.1816,
+      "step": 1271
+    },
+    {
+      "epoch": 0.7420154586553886,
+      "grad_norm": 1.2776967287063599,
+      "learning_rate": 3.2406850961538466e-05,
+      "loss": 1.0167,
+      "step": 1272
+    },
+    {
+      "epoch": 0.7425988041417529,
+      "grad_norm": 1.7386460304260254,
+      "learning_rate": 3.2391826923076924e-05,
+      "loss": 1.0841,
+      "step": 1273
+    },
+    {
+      "epoch": 0.7431821496281172,
+      "grad_norm": 1.1177300214767456,
+      "learning_rate": 3.237680288461539e-05,
+      "loss": 0.8789,
+      "step": 1274
+    },
+    {
+      "epoch": 0.7437654951144815,
+      "grad_norm": 1.161293864250183,
+      "learning_rate": 3.236177884615385e-05,
+      "loss": 0.9454,
+      "step": 1275
+    },
+    {
+      "epoch": 0.7443488406008458,
+      "grad_norm": 1.609604001045227,
+      "learning_rate": 3.2346754807692306e-05,
+      "loss": 0.9307,
+      "step": 1276
+    },
+    {
+      "epoch": 0.7449321860872101,
+      "grad_norm": 0.9415054321289062,
+      "learning_rate": 3.233173076923077e-05,
+      "loss": 0.9462,
+      "step": 1277
+    },
+    {
+      "epoch": 0.7455155315735744,
+      "grad_norm": 1.4762784242630005,
+      "learning_rate": 3.231670673076923e-05,
+      "loss": 1.0188,
+      "step": 1278
+    },
+    {
+      "epoch": 0.7460988770599387,
+      "grad_norm": 1.2860867977142334,
+      "learning_rate": 3.2301682692307695e-05,
+      "loss": 0.8886,
+      "step": 1279
+    },
+    {
+      "epoch": 0.746682222546303,
+      "grad_norm": 1.2090308666229248,
+      "learning_rate": 3.228665865384616e-05,
+      "loss": 1.0265,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7472655680326673,
+      "grad_norm": 1.0692979097366333,
+      "learning_rate": 3.227163461538462e-05,
+      "loss": 1.014,
+      "step": 1281
+    },
+    {
+      "epoch": 0.7478489135190316,
+      "grad_norm": 1.2457678318023682,
+      "learning_rate": 3.225661057692308e-05,
+      "loss": 1.0442,
+      "step": 1282
+    },
+    {
+      "epoch": 0.7484322590053959,
+      "grad_norm": 1.1981621980667114,
+      "learning_rate": 3.2241586538461536e-05,
+      "loss": 0.924,
+      "step": 1283
+    },
+    {
+      "epoch": 0.7490156044917602,
+      "grad_norm": 1.448926568031311,
+      "learning_rate": 3.22265625e-05,
+      "loss": 1.2461,
+      "step": 1284
+    },
+    {
+      "epoch": 0.7495989499781246,
+      "grad_norm": 1.2373019456863403,
+      "learning_rate": 3.221153846153847e-05,
+      "loss": 1.1221,
+      "step": 1285
+    },
+    {
+      "epoch": 0.7501822954644889,
+      "grad_norm": 1.135291337966919,
+      "learning_rate": 3.2196514423076925e-05,
+      "loss": 0.9693,
+      "step": 1286
+    },
+    {
+      "epoch": 0.7507656409508532,
+      "grad_norm": 1.3530988693237305,
+      "learning_rate": 3.2181490384615384e-05,
+      "loss": 0.9934,
+      "step": 1287
+    },
+    {
+      "epoch": 0.7513489864372175,
+      "grad_norm": 1.216654658317566,
+      "learning_rate": 3.216646634615384e-05,
+      "loss": 1.1738,
+      "step": 1288
+    },
+    {
+      "epoch": 0.7519323319235818,
+      "grad_norm": 1.1719613075256348,
+      "learning_rate": 3.215144230769231e-05,
+      "loss": 0.8775,
+      "step": 1289
+    },
+    {
+      "epoch": 0.7525156774099461,
+      "grad_norm": 1.223272442817688,
+      "learning_rate": 3.213641826923077e-05,
+      "loss": 1.0273,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7530990228963104,
+      "grad_norm": 1.4900156259536743,
+      "learning_rate": 3.212139423076923e-05,
+      "loss": 1.0808,
+      "step": 1291
+    },
+    {
+      "epoch": 0.7536823683826747,
+      "grad_norm": 1.1983774900436401,
+      "learning_rate": 3.21063701923077e-05,
+      "loss": 1.1111,
+      "step": 1292
+    },
+    {
+      "epoch": 0.754265713869039,
+      "grad_norm": 1.4071288108825684,
+      "learning_rate": 3.2091346153846155e-05,
+      "loss": 0.8877,
+      "step": 1293
+    },
+    {
+      "epoch": 0.7548490593554033,
+      "grad_norm": 1.1557461023330688,
+      "learning_rate": 3.2076322115384614e-05,
+      "loss": 0.8536,
+      "step": 1294
+    },
+    {
+      "epoch": 0.7554324048417675,
+      "grad_norm": 1.1718254089355469,
+      "learning_rate": 3.206129807692308e-05,
+      "loss": 0.8273,
+      "step": 1295
+    },
+    {
+      "epoch": 0.7560157503281318,
+      "grad_norm": 1.2104593515396118,
+      "learning_rate": 3.204627403846154e-05,
+      "loss": 1.2501,
+      "step": 1296
+    },
+    {
+      "epoch": 0.7565990958144961,
+      "grad_norm": 2.3339264392852783,
+      "learning_rate": 3.203125e-05,
+      "loss": 0.8378,
+      "step": 1297
+    },
+    {
+      "epoch": 0.7571824413008604,
+      "grad_norm": 1.0478349924087524,
+      "learning_rate": 3.201622596153847e-05,
+      "loss": 1.0836,
+      "step": 1298
+    },
+    {
+      "epoch": 0.7577657867872247,
+      "grad_norm": 1.2753820419311523,
+      "learning_rate": 3.2001201923076926e-05,
+      "loss": 1.1445,
+      "step": 1299
+    },
+    {
+      "epoch": 0.758349132273589,
+      "grad_norm": 1.194672703742981,
+      "learning_rate": 3.1986177884615385e-05,
+      "loss": 1.0424,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7589324777599533,
+      "grad_norm": 1.2020694017410278,
+      "learning_rate": 3.1971153846153843e-05,
+      "loss": 1.0068,
+      "step": 1301
+    },
+    {
+      "epoch": 0.7595158232463176,
+      "grad_norm": 2.032259941101074,
+      "learning_rate": 3.195612980769231e-05,
+      "loss": 0.946,
+      "step": 1302
+    },
+    {
+      "epoch": 0.7600991687326819,
+      "grad_norm": 1.2049493789672852,
+      "learning_rate": 3.1941105769230774e-05,
+      "loss": 1.053,
+      "step": 1303
+    },
+    {
+      "epoch": 0.7606825142190462,
+      "grad_norm": 1.1551873683929443,
+      "learning_rate": 3.192608173076923e-05,
+      "loss": 1.0539,
+      "step": 1304
+    },
+    {
+      "epoch": 0.7612658597054105,
+      "grad_norm": 1.0860520601272583,
+      "learning_rate": 3.191105769230769e-05,
+      "loss": 1.1625,
+      "step": 1305
+    },
+    {
+      "epoch": 0.7618492051917748,
+      "grad_norm": 1.1426883935928345,
+      "learning_rate": 3.1896033653846156e-05,
+      "loss": 0.8321,
+      "step": 1306
+    },
+    {
+      "epoch": 0.7624325506781391,
+      "grad_norm": 1.2449581623077393,
+      "learning_rate": 3.1881009615384615e-05,
+      "loss": 1.0748,
+      "step": 1307
+    },
+    {
+      "epoch": 0.7630158961645034,
+      "grad_norm": 1.2475258111953735,
+      "learning_rate": 3.186598557692308e-05,
+      "loss": 0.9779,
+      "step": 1308
+    },
+    {
+      "epoch": 0.7635992416508677,
+      "grad_norm": 1.0726374387741089,
+      "learning_rate": 3.185096153846154e-05,
+      "loss": 0.9079,
+      "step": 1309
+    },
+    {
+      "epoch": 0.764182587137232,
+      "grad_norm": 1.169661521911621,
+      "learning_rate": 3.1835937500000004e-05,
+      "loss": 1.1156,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7647659326235964,
+      "grad_norm": 1.1962292194366455,
+      "learning_rate": 3.182091346153846e-05,
+      "loss": 0.9258,
+      "step": 1311
+    },
+    {
+      "epoch": 0.7653492781099607,
+      "grad_norm": 1.038885235786438,
+      "learning_rate": 3.180588942307692e-05,
+      "loss": 1.0377,
+      "step": 1312
+    },
+    {
+      "epoch": 0.765932623596325,
+      "grad_norm": 1.4127665758132935,
+      "learning_rate": 3.1790865384615386e-05,
+      "loss": 0.982,
+      "step": 1313
+    },
+    {
+      "epoch": 0.7665159690826893,
+      "grad_norm": 1.1078689098358154,
+      "learning_rate": 3.1775841346153845e-05,
+      "loss": 1.0159,
+      "step": 1314
+    },
+    {
+      "epoch": 0.7670993145690536,
+      "grad_norm": 1.047503113746643,
+      "learning_rate": 3.176081730769231e-05,
+      "loss": 0.7841,
+      "step": 1315
+    },
+    {
+      "epoch": 0.7676826600554179,
+      "grad_norm": 1.2618082761764526,
+      "learning_rate": 3.1745793269230775e-05,
+      "loss": 1.0228,
+      "step": 1316
+    },
+    {
+      "epoch": 0.7682660055417821,
+      "grad_norm": 2.2956039905548096,
+      "learning_rate": 3.1730769230769234e-05,
+      "loss": 0.8457,
+      "step": 1317
+    },
+    {
+      "epoch": 0.7688493510281464,
+      "grad_norm": 1.1614506244659424,
+      "learning_rate": 3.171574519230769e-05,
+      "loss": 1.0914,
+      "step": 1318
+    },
+    {
+      "epoch": 0.7694326965145107,
+      "grad_norm": 1.0440285205841064,
+      "learning_rate": 3.170072115384616e-05,
+      "loss": 0.9781,
+      "step": 1319
+    },
+    {
+      "epoch": 0.770016042000875,
+      "grad_norm": 1.2385691404342651,
+      "learning_rate": 3.1685697115384616e-05,
+      "loss": 0.904,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7705993874872393,
+      "grad_norm": 1.0237793922424316,
+      "learning_rate": 3.167067307692308e-05,
+      "loss": 0.8984,
+      "step": 1321
+    },
+    {
+      "epoch": 0.7711827329736036,
+      "grad_norm": 1.4518593549728394,
+      "learning_rate": 3.165564903846154e-05,
+      "loss": 1.1862,
+      "step": 1322
+    },
+    {
+      "epoch": 0.7717660784599679,
+      "grad_norm": 1.197102427482605,
+      "learning_rate": 3.1640625e-05,
+      "loss": 1.0267,
+      "step": 1323
+    },
+    {
+      "epoch": 0.7723494239463322,
+      "grad_norm": 1.3080718517303467,
+      "learning_rate": 3.1625600961538464e-05,
+      "loss": 1.0663,
+      "step": 1324
+    },
+    {
+      "epoch": 0.7729327694326965,
+      "grad_norm": 0.9802163243293762,
+      "learning_rate": 3.161057692307692e-05,
+      "loss": 1.0571,
+      "step": 1325
+    },
+    {
+      "epoch": 0.7735161149190608,
+      "grad_norm": 1.969308853149414,
+      "learning_rate": 3.159555288461539e-05,
+      "loss": 1.0338,
+      "step": 1326
+    },
+    {
+      "epoch": 0.7740994604054251,
+      "grad_norm": 1.0249823331832886,
+      "learning_rate": 3.1580528846153846e-05,
+      "loss": 0.9921,
+      "step": 1327
+    },
+    {
+      "epoch": 0.7746828058917894,
+      "grad_norm": 1.099127173423767,
+      "learning_rate": 3.156550480769231e-05,
+      "loss": 0.9504,
+      "step": 1328
+    },
+    {
+      "epoch": 0.7752661513781537,
+      "grad_norm": 1.2968648672103882,
+      "learning_rate": 3.155048076923077e-05,
+      "loss": 0.9594,
+      "step": 1329
+    },
+    {
+      "epoch": 0.775849496864518,
+      "grad_norm": 1.0648295879364014,
+      "learning_rate": 3.153545673076923e-05,
+      "loss": 0.8687,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7764328423508823,
+      "grad_norm": 1.802680253982544,
+      "learning_rate": 3.1520432692307693e-05,
+      "loss": 1.0598,
+      "step": 1331
+    },
+    {
+      "epoch": 0.7770161878372466,
+      "grad_norm": 1.080910563468933,
+      "learning_rate": 3.150540865384616e-05,
+      "loss": 0.9963,
+      "step": 1332
+    },
+    {
+      "epoch": 0.7775995333236109,
+      "grad_norm": 1.1521499156951904,
+      "learning_rate": 3.149038461538462e-05,
+      "loss": 1.0519,
+      "step": 1333
+    },
+    {
+      "epoch": 0.7781828788099752,
+      "grad_norm": 1.1837037801742554,
+      "learning_rate": 3.147536057692308e-05,
+      "loss": 0.9725,
+      "step": 1334
+    },
+    {
+      "epoch": 0.7787662242963395,
+      "grad_norm": 1.085605502128601,
+      "learning_rate": 3.146033653846154e-05,
+      "loss": 1.1865,
+      "step": 1335
+    },
+    {
+      "epoch": 0.7793495697827038,
+      "grad_norm": 1.0882564783096313,
+      "learning_rate": 3.14453125e-05,
+      "loss": 0.9782,
+      "step": 1336
+    },
+    {
+      "epoch": 0.7799329152690682,
+      "grad_norm": 1.1564704179763794,
+      "learning_rate": 3.1430288461538465e-05,
+      "loss": 0.8602,
+      "step": 1337
+    },
+    {
+      "epoch": 0.7805162607554325,
+      "grad_norm": 1.3103642463684082,
+      "learning_rate": 3.141526442307692e-05,
+      "loss": 1.144,
+      "step": 1338
+    },
+    {
+      "epoch": 0.7810996062417968,
+      "grad_norm": 1.2268692255020142,
+      "learning_rate": 3.140024038461539e-05,
+      "loss": 0.8534,
+      "step": 1339
+    },
+    {
+      "epoch": 0.781682951728161,
+      "grad_norm": 1.2421032190322876,
+      "learning_rate": 3.138521634615385e-05,
+      "loss": 1.1487,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7822662972145253,
+      "grad_norm": 1.0822752714157104,
+      "learning_rate": 3.1370192307692306e-05,
+      "loss": 0.963,
+      "step": 1341
+    },
+    {
+      "epoch": 0.7828496427008896,
+      "grad_norm": 1.2561531066894531,
+      "learning_rate": 3.135516826923077e-05,
+      "loss": 1.1393,
+      "step": 1342
+    },
+    {
+      "epoch": 0.7834329881872539,
+      "grad_norm": 0.9926168918609619,
+      "learning_rate": 3.134014423076923e-05,
+      "loss": 0.8662,
+      "step": 1343
+    },
+    {
+      "epoch": 0.7840163336736182,
+      "grad_norm": 1.2411295175552368,
+      "learning_rate": 3.1325120192307695e-05,
+      "loss": 1.0019,
+      "step": 1344
+    },
+    {
+      "epoch": 0.7845996791599825,
+      "grad_norm": 1.1900317668914795,
+      "learning_rate": 3.131009615384616e-05,
+      "loss": 1.0258,
+      "step": 1345
+    },
+    {
+      "epoch": 0.7851830246463468,
+      "grad_norm": 1.0790519714355469,
+      "learning_rate": 3.129507211538462e-05,
+      "loss": 1.1386,
+      "step": 1346
+    },
+    {
+      "epoch": 0.7857663701327111,
+      "grad_norm": 2.3410184383392334,
+      "learning_rate": 3.128004807692308e-05,
+      "loss": 0.8789,
+      "step": 1347
+    },
+    {
+      "epoch": 0.7863497156190754,
+      "grad_norm": 1.0258671045303345,
+      "learning_rate": 3.1265024038461535e-05,
+      "loss": 1.015,
+      "step": 1348
+    },
+    {
+      "epoch": 0.7869330611054397,
+      "grad_norm": 1.0533411502838135,
+      "learning_rate": 3.125e-05,
+      "loss": 0.8044,
+      "step": 1349
+    },
+    {
+      "epoch": 0.787516406591804,
+      "grad_norm": 1.1454368829727173,
+      "learning_rate": 3.1234975961538466e-05,
+      "loss": 1.0102,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7880997520781683,
+      "grad_norm": 1.3050988912582397,
+      "learning_rate": 3.1219951923076924e-05,
+      "loss": 1.1298,
+      "step": 1351
+    },
+    {
+      "epoch": 0.7886830975645326,
+      "grad_norm": 1.1478936672210693,
+      "learning_rate": 3.120492788461539e-05,
+      "loss": 0.9447,
+      "step": 1352
+    },
+    {
+      "epoch": 0.7892664430508969,
+      "grad_norm": 1.2745267152786255,
+      "learning_rate": 3.118990384615385e-05,
+      "loss": 0.9069,
+      "step": 1353
+    },
+    {
+      "epoch": 0.7898497885372612,
+      "grad_norm": 1.031055212020874,
+      "learning_rate": 3.117487980769231e-05,
+      "loss": 0.9162,
+      "step": 1354
+    },
+    {
+      "epoch": 0.7904331340236255,
+      "grad_norm": 1.1818459033966064,
+      "learning_rate": 3.115985576923077e-05,
+      "loss": 0.9005,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7910164795099898,
+      "grad_norm": 1.157064437866211,
+      "learning_rate": 3.114483173076923e-05,
+      "loss": 1.0922,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7915998249963541,
+      "grad_norm": 1.3568843603134155,
+      "learning_rate": 3.1129807692307696e-05,
+      "loss": 0.9815,
+      "step": 1357
+    },
+    {
+      "epoch": 0.7921831704827184,
+      "grad_norm": 0.9772933125495911,
+      "learning_rate": 3.111478365384616e-05,
+      "loss": 0.848,
+      "step": 1358
+    },
+    {
+      "epoch": 0.7927665159690827,
+      "grad_norm": 1.2958146333694458,
+      "learning_rate": 3.109975961538461e-05,
+      "loss": 0.903,
+      "step": 1359
+    },
+    {
+      "epoch": 0.793349861455447,
+      "grad_norm": 1.325095772743225,
+      "learning_rate": 3.108473557692308e-05,
+      "loss": 0.8335,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7939332069418112,
+      "grad_norm": 1.3909435272216797,
+      "learning_rate": 3.1069711538461537e-05,
+      "loss": 1.1868,
+      "step": 1361
+    },
+    {
+      "epoch": 0.7945165524281755,
+      "grad_norm": 1.1557323932647705,
+      "learning_rate": 3.10546875e-05,
+      "loss": 1.2876,
+      "step": 1362
+    },
+    {
+      "epoch": 0.7950998979145398,
+      "grad_norm": 1.0653504133224487,
+      "learning_rate": 3.103966346153847e-05,
+      "loss": 1.0039,
+      "step": 1363
+    },
+    {
+      "epoch": 0.7956832434009042,
+      "grad_norm": 1.1019622087478638,
+      "learning_rate": 3.1024639423076926e-05,
+      "loss": 0.8072,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7962665888872685,
+      "grad_norm": 1.1759988069534302,
+      "learning_rate": 3.1009615384615384e-05,
+      "loss": 1.038,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7968499343736328,
+      "grad_norm": 1.2272703647613525,
+      "learning_rate": 3.099459134615384e-05,
+      "loss": 1.0167,
+      "step": 1366
+    },
+    {
+      "epoch": 0.7974332798599971,
+      "grad_norm": 1.6301058530807495,
+      "learning_rate": 3.097956730769231e-05,
+      "loss": 0.9102,
+      "step": 1367
+    },
+    {
+      "epoch": 0.7980166253463614,
+      "grad_norm": 1.059002161026001,
+      "learning_rate": 3.096454326923077e-05,
+      "loss": 1.1956,
+      "step": 1368
+    },
+    {
+      "epoch": 0.7985999708327257,
+      "grad_norm": 1.0058438777923584,
+      "learning_rate": 3.094951923076923e-05,
+      "loss": 0.7865,
+      "step": 1369
+    },
+    {
+      "epoch": 0.79918331631909,
+      "grad_norm": 1.3153209686279297,
+      "learning_rate": 3.09344951923077e-05,
+      "loss": 0.995,
+      "step": 1370
+    },
+    {
+      "epoch": 0.7997666618054543,
+      "grad_norm": 1.1372942924499512,
+      "learning_rate": 3.0919471153846155e-05,
+      "loss": 0.901,
+      "step": 1371
+    },
+    {
+      "epoch": 0.8003500072918186,
+      "grad_norm": 1.0216442346572876,
+      "learning_rate": 3.0904447115384614e-05,
+      "loss": 1.1966,
+      "step": 1372
+    },
+    {
+      "epoch": 0.8009333527781829,
+      "grad_norm": 1.054608702659607,
+      "learning_rate": 3.088942307692308e-05,
+      "loss": 0.8792,
+      "step": 1373
+    },
+    {
+      "epoch": 0.8015166982645472,
+      "grad_norm": 1.2942436933517456,
+      "learning_rate": 3.087439903846154e-05,
+      "loss": 1.1201,
+      "step": 1374
+    },
+    {
+      "epoch": 0.8021000437509115,
+      "grad_norm": 1.2945632934570312,
+      "learning_rate": 3.0859375e-05,
+      "loss": 0.936,
+      "step": 1375
+    },
+    {
+      "epoch": 0.8026833892372758,
+      "grad_norm": 1.213426947593689,
+      "learning_rate": 3.084435096153847e-05,
+      "loss": 0.9309,
+      "step": 1376
+    },
+    {
+      "epoch": 0.8032667347236401,
+      "grad_norm": 1.0726510286331177,
+      "learning_rate": 3.082932692307692e-05,
+      "loss": 1.1424,
+      "step": 1377
+    },
+    {
+      "epoch": 0.8038500802100044,
+      "grad_norm": 1.3145674467086792,
+      "learning_rate": 3.0814302884615385e-05,
+      "loss": 1.0205,
+      "step": 1378
+    },
+    {
+      "epoch": 0.8044334256963687,
+      "grad_norm": 1.082029104232788,
+      "learning_rate": 3.0799278846153844e-05,
+      "loss": 0.8689,
+      "step": 1379
+    },
+    {
+      "epoch": 0.805016771182733,
+      "grad_norm": 1.0544030666351318,
+      "learning_rate": 3.078425480769231e-05,
+      "loss": 0.8943,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8056001166690973,
+      "grad_norm": 1.1208224296569824,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.9113,
+      "step": 1381
+    },
+    {
+      "epoch": 0.8061834621554615,
+      "grad_norm": 1.1705317497253418,
+      "learning_rate": 3.075420673076923e-05,
+      "loss": 0.7789,
+      "step": 1382
+    },
+    {
+      "epoch": 0.8067668076418258,
+      "grad_norm": 1.1692845821380615,
+      "learning_rate": 3.073918269230769e-05,
+      "loss": 1.1744,
+      "step": 1383
+    },
+    {
+      "epoch": 0.8073501531281901,
+      "grad_norm": 1.0401328802108765,
+      "learning_rate": 3.072415865384616e-05,
+      "loss": 0.9221,
+      "step": 1384
+    },
+    {
+      "epoch": 0.8079334986145544,
+      "grad_norm": 1.0775591135025024,
+      "learning_rate": 3.0709134615384615e-05,
+      "loss": 1.2103,
+      "step": 1385
+    },
+    {
+      "epoch": 0.8085168441009187,
+      "grad_norm": 1.3836114406585693,
+      "learning_rate": 3.069411057692308e-05,
+      "loss": 0.8438,
+      "step": 1386
+    },
+    {
+      "epoch": 0.809100189587283,
+      "grad_norm": 1.1026825904846191,
+      "learning_rate": 3.067908653846154e-05,
+      "loss": 0.8164,
+      "step": 1387
+    },
+    {
+      "epoch": 0.8096835350736473,
+      "grad_norm": 1.80268132686615,
+      "learning_rate": 3.0664062500000004e-05,
+      "loss": 1.0345,
+      "step": 1388
+    },
+    {
+      "epoch": 0.8102668805600116,
+      "grad_norm": 1.5059008598327637,
+      "learning_rate": 3.064903846153846e-05,
+      "loss": 0.9297,
+      "step": 1389
+    },
+    {
+      "epoch": 0.810850226046376,
+      "grad_norm": 1.0632637739181519,
+      "learning_rate": 3.063401442307692e-05,
+      "loss": 1.237,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8114335715327403,
+      "grad_norm": 1.1647098064422607,
+      "learning_rate": 3.0618990384615386e-05,
+      "loss": 1.0494,
+      "step": 1391
+    },
+    {
+      "epoch": 0.8120169170191046,
+      "grad_norm": 1.0377775430679321,
+      "learning_rate": 3.0603966346153845e-05,
+      "loss": 1.3689,
+      "step": 1392
+    },
+    {
+      "epoch": 0.8126002625054689,
+      "grad_norm": 1.713774561882019,
+      "learning_rate": 3.058894230769231e-05,
+      "loss": 1.1149,
+      "step": 1393
+    },
+    {
+      "epoch": 0.8131836079918332,
+      "grad_norm": 1.2139819860458374,
+      "learning_rate": 3.0573918269230776e-05,
+      "loss": 0.8279,
+      "step": 1394
+    },
+    {
+      "epoch": 0.8137669534781975,
+      "grad_norm": 1.3912312984466553,
+      "learning_rate": 3.055889423076923e-05,
+      "loss": 1.0654,
+      "step": 1395
+    },
+    {
+      "epoch": 0.8143502989645618,
+      "grad_norm": 1.276260256767273,
+      "learning_rate": 3.054387019230769e-05,
+      "loss": 0.8086,
+      "step": 1396
+    },
+    {
+      "epoch": 0.8149336444509261,
+      "grad_norm": 1.3147262334823608,
+      "learning_rate": 3.052884615384616e-05,
+      "loss": 1.1643,
+      "step": 1397
+    },
+    {
+      "epoch": 0.8155169899372904,
+      "grad_norm": 1.760912299156189,
+      "learning_rate": 3.0513822115384616e-05,
+      "loss": 1.2003,
+      "step": 1398
+    },
+    {
+      "epoch": 0.8161003354236547,
+      "grad_norm": 1.060642957687378,
+      "learning_rate": 3.0498798076923078e-05,
+      "loss": 0.8297,
+      "step": 1399
+    },
+    {
+      "epoch": 0.816683680910019,
+      "grad_norm": 1.1941111087799072,
+      "learning_rate": 3.0483774038461537e-05,
+      "loss": 0.8585,
+      "step": 1400
+    },
+    {
+      "epoch": 0.816683680910019,
+      "eval_loss_squad": 0.8743070242926478,
+      "eval_perplexity": 8.211708550571148,
+      "eval_perplexity_reconstruct": 1.9933093086287428,
+      "step": 1400
+    },
+    {
+      "epoch": 0.8172670263963833,
+      "grad_norm": 1.1372501850128174,
+      "learning_rate": 3.0468750000000002e-05,
+      "loss": 1.0104,
+      "step": 1401
+    },
+    {
+      "epoch": 0.8178503718827476,
+      "grad_norm": 1.203554391860962,
+      "learning_rate": 3.0453725961538464e-05,
+      "loss": 1.1503,
+      "step": 1402
+    },
+    {
+      "epoch": 0.8184337173691119,
+      "grad_norm": 1.5660467147827148,
+      "learning_rate": 3.0438701923076922e-05,
+      "loss": 0.9822,
+      "step": 1403
+    },
+    {
+      "epoch": 0.8190170628554762,
+      "grad_norm": 1.2876819372177124,
+      "learning_rate": 3.0423677884615388e-05,
+      "loss": 1.1155,
+      "step": 1404
+    },
+    {
+      "epoch": 0.8196004083418404,
+      "grad_norm": 1.0389353036880493,
+      "learning_rate": 3.0408653846153846e-05,
+      "loss": 0.8308,
+      "step": 1405
+    },
+    {
+      "epoch": 0.8201837538282047,
+      "grad_norm": 1.467089295387268,
+      "learning_rate": 3.0393629807692308e-05,
+      "loss": 1.308,
+      "step": 1406
+    },
+    {
+      "epoch": 0.820767099314569,
+      "grad_norm": 1.135607361793518,
+      "learning_rate": 3.0378605769230773e-05,
+      "loss": 0.9116,
+      "step": 1407
+    },
+    {
+      "epoch": 0.8213504448009333,
+      "grad_norm": 1.0728017091751099,
+      "learning_rate": 3.0363581730769232e-05,
+      "loss": 0.9576,
+      "step": 1408
+    },
+    {
+      "epoch": 0.8219337902872976,
+      "grad_norm": 1.2483171224594116,
+      "learning_rate": 3.0348557692307694e-05,
+      "loss": 1.0636,
+      "step": 1409
+    },
+    {
+      "epoch": 0.8225171357736619,
+      "grad_norm": 1.3540332317352295,
+      "learning_rate": 3.033353365384616e-05,
+      "loss": 0.8896,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8231004812600262,
+      "grad_norm": 1.4841009378433228,
+      "learning_rate": 3.0318509615384618e-05,
+      "loss": 0.9574,
+      "step": 1411
+    },
+    {
+      "epoch": 0.8236838267463905,
+      "grad_norm": 1.2146106958389282,
+      "learning_rate": 3.030348557692308e-05,
+      "loss": 1.0769,
+      "step": 1412
+    },
+    {
+      "epoch": 0.8242671722327548,
+      "grad_norm": 1.104175090789795,
+      "learning_rate": 3.0288461538461538e-05,
+      "loss": 0.9756,
+      "step": 1413
+    },
+    {
+      "epoch": 0.8248505177191191,
+      "grad_norm": 0.9717370271682739,
+      "learning_rate": 3.02734375e-05,
+      "loss": 0.9478,
+      "step": 1414
+    },
+    {
+      "epoch": 0.8254338632054834,
+      "grad_norm": 1.0630606412887573,
+      "learning_rate": 3.0258413461538465e-05,
+      "loss": 0.9061,
+      "step": 1415
+    },
+    {
+      "epoch": 0.8260172086918477,
+      "grad_norm": 1.1713072061538696,
+      "learning_rate": 3.0243389423076924e-05,
+      "loss": 1.1889,
+      "step": 1416
+    },
+    {
+      "epoch": 0.8266005541782121,
+      "grad_norm": 1.2068907022476196,
+      "learning_rate": 3.0228365384615385e-05,
+      "loss": 0.9438,
+      "step": 1417
+    },
+    {
+      "epoch": 0.8271838996645764,
+      "grad_norm": 0.9715260863304138,
+      "learning_rate": 3.0213341346153844e-05,
+      "loss": 0.7337,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8277672451509407,
+      "grad_norm": 1.0985807180404663,
+      "learning_rate": 3.019831730769231e-05,
+      "loss": 1.0818,
+      "step": 1419
+    },
+    {
+      "epoch": 0.828350590637305,
+      "grad_norm": 1.2472034692764282,
+      "learning_rate": 3.018329326923077e-05,
+      "loss": 0.9131,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8289339361236693,
+      "grad_norm": 1.1218814849853516,
+      "learning_rate": 3.016826923076923e-05,
+      "loss": 1.0787,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8295172816100336,
+      "grad_norm": 1.0618772506713867,
+      "learning_rate": 3.0153245192307695e-05,
+      "loss": 0.8397,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8301006270963979,
+      "grad_norm": 1.021227240562439,
+      "learning_rate": 3.0138221153846157e-05,
+      "loss": 0.7859,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8306839725827622,
+      "grad_norm": 1.3854846954345703,
+      "learning_rate": 3.0123197115384615e-05,
+      "loss": 0.9057,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8312673180691265,
+      "grad_norm": 1.0573982000350952,
+      "learning_rate": 3.010817307692308e-05,
+      "loss": 0.8431,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8318506635554908,
+      "grad_norm": 1.3047147989273071,
+      "learning_rate": 3.009314903846154e-05,
+      "loss": 1.1606,
+      "step": 1426
+    },
+    {
+      "epoch": 0.832434009041855,
+      "grad_norm": 1.6169391870498657,
+      "learning_rate": 3.0078125e-05,
+      "loss": 1.0701,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8330173545282193,
+      "grad_norm": 1.2799001932144165,
+      "learning_rate": 3.0063100961538466e-05,
+      "loss": 1.0659,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8336007000145836,
+      "grad_norm": 1.3066989183425903,
+      "learning_rate": 3.0048076923076925e-05,
+      "loss": 1.0707,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8341840455009479,
+      "grad_norm": 1.275020956993103,
+      "learning_rate": 3.0033052884615387e-05,
+      "loss": 1.0019,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8347673909873122,
+      "grad_norm": 1.084713101387024,
+      "learning_rate": 3.0018028846153845e-05,
+      "loss": 1.0027,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8353507364736765,
+      "grad_norm": 1.222685694694519,
+      "learning_rate": 3.0003004807692307e-05,
+      "loss": 0.8549,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8359340819600408,
+      "grad_norm": 1.1162559986114502,
+      "learning_rate": 2.9987980769230772e-05,
+      "loss": 0.9687,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8365174274464051,
+      "grad_norm": 1.1409958600997925,
+      "learning_rate": 2.997295673076923e-05,
+      "loss": 0.9169,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8371007729327694,
+      "grad_norm": 1.1237506866455078,
+      "learning_rate": 2.9957932692307693e-05,
+      "loss": 1.0596,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8376841184191337,
+      "grad_norm": 1.2333991527557373,
+      "learning_rate": 2.9942908653846158e-05,
+      "loss": 1.0413,
+      "step": 1436
+    },
+    {
+      "epoch": 0.838267463905498,
+      "grad_norm": 1.2132785320281982,
+      "learning_rate": 2.9927884615384616e-05,
+      "loss": 0.8801,
+      "step": 1437
+    },
+    {
+      "epoch": 0.8388508093918623,
+      "grad_norm": 2.4601593017578125,
+      "learning_rate": 2.991286057692308e-05,
+      "loss": 1.0372,
+      "step": 1438
+    },
+    {
+      "epoch": 0.8394341548782266,
+      "grad_norm": 1.1504793167114258,
+      "learning_rate": 2.9897836538461537e-05,
+      "loss": 1.0258,
+      "step": 1439
+    },
+    {
+      "epoch": 0.8400175003645909,
+      "grad_norm": 1.0757592916488647,
+      "learning_rate": 2.9882812500000002e-05,
+      "loss": 0.925,
+      "step": 1440
+    },
+    {
+      "epoch": 0.8406008458509552,
+      "grad_norm": 1.2550078630447388,
+      "learning_rate": 2.9867788461538464e-05,
+      "loss": 1.0328,
+      "step": 1441
+    },
+    {
+      "epoch": 0.8411841913373195,
+      "grad_norm": 1.108076810836792,
+      "learning_rate": 2.9852764423076923e-05,
+      "loss": 1.0529,
+      "step": 1442
+    },
+    {
+      "epoch": 0.8417675368236839,
+      "grad_norm": 1.1696008443832397,
+      "learning_rate": 2.9837740384615388e-05,
+      "loss": 0.8204,
+      "step": 1443
+    },
+    {
+      "epoch": 0.8423508823100482,
+      "grad_norm": 0.9712606072425842,
+      "learning_rate": 2.9822716346153846e-05,
+      "loss": 0.9446,
+      "step": 1444
+    },
+    {
+      "epoch": 0.8429342277964125,
+      "grad_norm": 1.253280520439148,
+      "learning_rate": 2.9807692307692308e-05,
+      "loss": 0.9655,
+      "step": 1445
+    },
+    {
+      "epoch": 0.8435175732827768,
+      "grad_norm": 1.2332576513290405,
+      "learning_rate": 2.9792668269230773e-05,
+      "loss": 0.9132,
+      "step": 1446
+    },
+    {
+      "epoch": 0.8441009187691411,
+      "grad_norm": 0.9151219129562378,
+      "learning_rate": 2.9777644230769232e-05,
+      "loss": 1.0091,
+      "step": 1447
+    },
+    {
+      "epoch": 0.8446842642555054,
+      "grad_norm": 1.0670338869094849,
+      "learning_rate": 2.9762620192307694e-05,
+      "loss": 0.9617,
+      "step": 1448
+    },
+    {
+      "epoch": 0.8452676097418697,
+      "grad_norm": 1.4182400703430176,
+      "learning_rate": 2.974759615384616e-05,
+      "loss": 0.9599,
+      "step": 1449
+    },
+    {
+      "epoch": 0.845850955228234,
+      "grad_norm": 1.4236427545547485,
+      "learning_rate": 2.9732572115384614e-05,
+      "loss": 1.0529,
+      "step": 1450
+    },
+    {
+      "epoch": 0.8464343007145982,
+      "grad_norm": 1.3941996097564697,
+      "learning_rate": 2.971754807692308e-05,
+      "loss": 0.8889,
+      "step": 1451
+    },
+    {
+      "epoch": 0.8470176462009625,
+      "grad_norm": 1.1831328868865967,
+      "learning_rate": 2.9702524038461538e-05,
+      "loss": 0.9023,
+      "step": 1452
+    },
+    {
+      "epoch": 0.8476009916873268,
+      "grad_norm": 1.4158825874328613,
+      "learning_rate": 2.96875e-05,
+      "loss": 1.1108,
+      "step": 1453
+    },
+    {
+      "epoch": 0.8481843371736911,
+      "grad_norm": 0.9731126427650452,
+      "learning_rate": 2.9672475961538465e-05,
+      "loss": 0.8187,
+      "step": 1454
+    },
+    {
+      "epoch": 0.8487676826600554,
+      "grad_norm": 1.0598933696746826,
+      "learning_rate": 2.9657451923076924e-05,
+      "loss": 0.8798,
+      "step": 1455
+    },
+    {
+      "epoch": 0.8493510281464197,
+      "grad_norm": 1.04989492893219,
+      "learning_rate": 2.9642427884615386e-05,
+      "loss": 0.9962,
+      "step": 1456
+    },
+    {
+      "epoch": 0.849934373632784,
+      "grad_norm": 1.1285686492919922,
+      "learning_rate": 2.9627403846153844e-05,
+      "loss": 0.8692,
+      "step": 1457
+    },
+    {
+      "epoch": 0.8505177191191483,
+      "grad_norm": 1.1516649723052979,
+      "learning_rate": 2.961237980769231e-05,
+      "loss": 0.8191,
+      "step": 1458
+    },
+    {
+      "epoch": 0.8511010646055126,
+      "grad_norm": 1.2343637943267822,
+      "learning_rate": 2.959735576923077e-05,
+      "loss": 0.8846,
+      "step": 1459
+    },
+    {
+      "epoch": 0.8516844100918769,
+      "grad_norm": 1.295943021774292,
+      "learning_rate": 2.958233173076923e-05,
+      "loss": 0.9154,
+      "step": 1460
+    },
+    {
+      "epoch": 0.8522677555782412,
+      "grad_norm": 1.3594167232513428,
+      "learning_rate": 2.9567307692307695e-05,
+      "loss": 0.9516,
+      "step": 1461
+    },
+    {
+      "epoch": 0.8528511010646055,
+      "grad_norm": 1.2709529399871826,
+      "learning_rate": 2.9552283653846157e-05,
+      "loss": 1.0265,
+      "step": 1462
+    },
+    {
+      "epoch": 0.8534344465509698,
+      "grad_norm": 1.2515851259231567,
+      "learning_rate": 2.9537259615384615e-05,
+      "loss": 0.9376,
+      "step": 1463
+    },
+    {
+      "epoch": 0.8540177920373341,
+      "grad_norm": 0.9908419251441956,
+      "learning_rate": 2.952223557692308e-05,
+      "loss": 1.0423,
+      "step": 1464
+    },
+    {
+      "epoch": 0.8546011375236984,
+      "grad_norm": 1.3674644231796265,
+      "learning_rate": 2.950721153846154e-05,
+      "loss": 1.2146,
+      "step": 1465
+    },
+    {
+      "epoch": 0.8551844830100627,
+      "grad_norm": 1.239865779876709,
+      "learning_rate": 2.94921875e-05,
+      "loss": 0.9979,
+      "step": 1466
+    },
+    {
+      "epoch": 0.855767828496427,
+      "grad_norm": 2.099808692932129,
+      "learning_rate": 2.9477163461538466e-05,
+      "loss": 1.1026,
+      "step": 1467
+    },
+    {
+      "epoch": 0.8563511739827913,
+      "grad_norm": 1.1289135217666626,
+      "learning_rate": 2.946213942307692e-05,
+      "loss": 1.0327,
+      "step": 1468
+    },
+    {
+      "epoch": 0.8569345194691556,
+      "grad_norm": 1.2344995737075806,
+      "learning_rate": 2.9447115384615387e-05,
+      "loss": 0.9183,
+      "step": 1469
+    },
+    {
+      "epoch": 0.85751786495552,
+      "grad_norm": 1.2652794122695923,
+      "learning_rate": 2.9432091346153845e-05,
+      "loss": 0.9498,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8581012104418843,
+      "grad_norm": 1.1644681692123413,
+      "learning_rate": 2.9417067307692307e-05,
+      "loss": 1.037,
+      "step": 1471
+    },
+    {
+      "epoch": 0.8586845559282485,
+      "grad_norm": 1.0714110136032104,
+      "learning_rate": 2.9402043269230772e-05,
+      "loss": 1.0802,
+      "step": 1472
+    },
+    {
+      "epoch": 0.8592679014146128,
+      "grad_norm": 1.1564422845840454,
+      "learning_rate": 2.938701923076923e-05,
+      "loss": 1.0889,
+      "step": 1473
+    },
+    {
+      "epoch": 0.8598512469009771,
+      "grad_norm": 1.3579368591308594,
+      "learning_rate": 2.9371995192307693e-05,
+      "loss": 0.8359,
+      "step": 1474
+    },
+    {
+      "epoch": 0.8604345923873414,
+      "grad_norm": 1.0801093578338623,
+      "learning_rate": 2.9356971153846158e-05,
+      "loss": 0.911,
+      "step": 1475
+    },
+    {
+      "epoch": 0.8610179378737057,
+      "grad_norm": 1.1530592441558838,
+      "learning_rate": 2.9341947115384617e-05,
+      "loss": 1.0432,
+      "step": 1476
+    },
+    {
+      "epoch": 0.86160128336007,
+      "grad_norm": 1.07007896900177,
+      "learning_rate": 2.932692307692308e-05,
+      "loss": 1.0042,
+      "step": 1477
+    },
+    {
+      "epoch": 0.8621846288464343,
+      "grad_norm": 1.128960371017456,
+      "learning_rate": 2.9311899038461537e-05,
+      "loss": 1.0481,
+      "step": 1478
+    },
+    {
+      "epoch": 0.8627679743327986,
+      "grad_norm": 1.7839515209197998,
+      "learning_rate": 2.9296875000000002e-05,
+      "loss": 1.1279,
+      "step": 1479
+    },
+    {
+      "epoch": 0.8633513198191629,
+      "grad_norm": 1.7222251892089844,
+      "learning_rate": 2.9281850961538464e-05,
+      "loss": 0.9325,
+      "step": 1480
+    },
+    {
+      "epoch": 0.8639346653055272,
+      "grad_norm": 0.8684887290000916,
+      "learning_rate": 2.9266826923076923e-05,
+      "loss": 1.1595,
+      "step": 1481
+    },
+    {
+      "epoch": 0.8645180107918915,
+      "grad_norm": 1.20137357711792,
+      "learning_rate": 2.9251802884615388e-05,
+      "loss": 1.0726,
+      "step": 1482
+    },
+    {
+      "epoch": 0.8651013562782558,
+      "grad_norm": 1.3149369955062866,
+      "learning_rate": 2.9236778846153846e-05,
+      "loss": 1.1417,
+      "step": 1483
+    },
+    {
+      "epoch": 0.8656847017646201,
+      "grad_norm": 1.2710528373718262,
+      "learning_rate": 2.922175480769231e-05,
+      "loss": 1.0942,
+      "step": 1484
+    },
+    {
+      "epoch": 0.8662680472509844,
+      "grad_norm": 1.145824909210205,
+      "learning_rate": 2.9206730769230774e-05,
+      "loss": 1.0122,
+      "step": 1485
+    },
+    {
+      "epoch": 0.8668513927373487,
+      "grad_norm": 1.2785214185714722,
+      "learning_rate": 2.9191706730769232e-05,
+      "loss": 1.1068,
+      "step": 1486
+    },
+    {
+      "epoch": 0.867434738223713,
+      "grad_norm": 1.3779847621917725,
+      "learning_rate": 2.9176682692307694e-05,
+      "loss": 1.1604,
+      "step": 1487
+    },
+    {
+      "epoch": 0.8680180837100773,
+      "grad_norm": 1.2660037279129028,
+      "learning_rate": 2.916165865384616e-05,
+      "loss": 0.9293,
+      "step": 1488
+    },
+    {
+      "epoch": 0.8686014291964416,
+      "grad_norm": 1.2929606437683105,
+      "learning_rate": 2.9146634615384614e-05,
+      "loss": 0.879,
+      "step": 1489
+    },
+    {
+      "epoch": 0.8691847746828059,
+      "grad_norm": 1.376483678817749,
+      "learning_rate": 2.913161057692308e-05,
+      "loss": 1.0383,
+      "step": 1490
+    },
+    {
+      "epoch": 0.8697681201691702,
+      "grad_norm": 1.1967178583145142,
+      "learning_rate": 2.9116586538461538e-05,
+      "loss": 0.981,
+      "step": 1491
+    },
+    {
+      "epoch": 0.8703514656555345,
+      "grad_norm": 1.1415852308273315,
+      "learning_rate": 2.91015625e-05,
+      "loss": 0.8465,
+      "step": 1492
+    },
+    {
+      "epoch": 0.8709348111418987,
+      "grad_norm": 1.2080132961273193,
+      "learning_rate": 2.9086538461538465e-05,
+      "loss": 1.0375,
+      "step": 1493
+    },
+    {
+      "epoch": 0.871518156628263,
+      "grad_norm": 1.1498886346817017,
+      "learning_rate": 2.9071514423076924e-05,
+      "loss": 1.1026,
+      "step": 1494
+    },
+    {
+      "epoch": 0.8721015021146273,
+      "grad_norm": 1.213456630706787,
+      "learning_rate": 2.9056490384615386e-05,
+      "loss": 1.1582,
+      "step": 1495
+    },
+    {
+      "epoch": 0.8726848476009917,
+      "grad_norm": 2.127002239227295,
+      "learning_rate": 2.9041466346153844e-05,
+      "loss": 0.9406,
+      "step": 1496
+    },
+    {
+      "epoch": 0.873268193087356,
+      "grad_norm": 1.454347848892212,
+      "learning_rate": 2.902644230769231e-05,
+      "loss": 0.9734,
+      "step": 1497
+    },
+    {
+      "epoch": 0.8738515385737203,
+      "grad_norm": 1.1562999486923218,
+      "learning_rate": 2.901141826923077e-05,
+      "loss": 1.1249,
+      "step": 1498
+    },
+    {
+      "epoch": 0.8744348840600846,
+      "grad_norm": 1.0088770389556885,
+      "learning_rate": 2.899639423076923e-05,
+      "loss": 1.1021,
+      "step": 1499
+    },
+    {
+      "epoch": 0.8750182295464489,
+      "grad_norm": 0.972812294960022,
+      "learning_rate": 2.8981370192307695e-05,
+      "loss": 0.8978,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8756015750328132,
+      "grad_norm": 1.3404576778411865,
+      "learning_rate": 2.8966346153846157e-05,
+      "loss": 1.0195,
+      "step": 1501
+    },
+    {
+      "epoch": 0.8761849205191775,
+      "grad_norm": 1.1055123805999756,
+      "learning_rate": 2.8951322115384616e-05,
+      "loss": 1.1829,
+      "step": 1502
+    },
+    {
+      "epoch": 0.8767682660055418,
+      "grad_norm": 0.9296578764915466,
+      "learning_rate": 2.893629807692308e-05,
+      "loss": 1.1158,
+      "step": 1503
+    },
+    {
+      "epoch": 0.8773516114919061,
+      "grad_norm": 1.197265625,
+      "learning_rate": 2.892127403846154e-05,
+      "loss": 1.0498,
+      "step": 1504
+    },
+    {
+      "epoch": 0.8779349569782704,
+      "grad_norm": 0.9996442794799805,
+      "learning_rate": 2.890625e-05,
+      "loss": 1.0372,
+      "step": 1505
+    },
+    {
+      "epoch": 0.8785183024646347,
+      "grad_norm": 1.2330306768417358,
+      "learning_rate": 2.8891225961538467e-05,
+      "loss": 1.0224,
+      "step": 1506
+    },
+    {
+      "epoch": 0.879101647950999,
+      "grad_norm": 0.902051568031311,
+      "learning_rate": 2.887620192307692e-05,
+      "loss": 0.6842,
+      "step": 1507
+    },
+    {
+      "epoch": 0.8796849934373633,
+      "grad_norm": 1.1468744277954102,
+      "learning_rate": 2.8861177884615387e-05,
+      "loss": 0.7994,
+      "step": 1508
+    },
+    {
+      "epoch": 0.8802683389237276,
+      "grad_norm": 1.3102710247039795,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 0.8041,
+      "step": 1509
+    },
+    {
+      "epoch": 0.8808516844100919,
+      "grad_norm": 1.234621286392212,
+      "learning_rate": 2.8831129807692307e-05,
+      "loss": 0.9069,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8814350298964562,
+      "grad_norm": 1.3230934143066406,
+      "learning_rate": 2.8816105769230773e-05,
+      "loss": 0.8703,
+      "step": 1511
+    },
+    {
+      "epoch": 0.8820183753828205,
+      "grad_norm": 1.1847695112228394,
+      "learning_rate": 2.880108173076923e-05,
+      "loss": 1.0461,
+      "step": 1512
+    },
+    {
+      "epoch": 0.8826017208691848,
+      "grad_norm": 1.535765528678894,
+      "learning_rate": 2.8786057692307693e-05,
+      "loss": 0.9742,
+      "step": 1513
+    },
+    {
+      "epoch": 0.883185066355549,
+      "grad_norm": 1.295670509338379,
+      "learning_rate": 2.8771033653846158e-05,
+      "loss": 0.9253,
+      "step": 1514
+    },
+    {
+      "epoch": 0.8837684118419133,
+      "grad_norm": 1.4603569507598877,
+      "learning_rate": 2.8756009615384617e-05,
+      "loss": 1.2047,
+      "step": 1515
+    },
+    {
+      "epoch": 0.8843517573282776,
+      "grad_norm": 1.117161512374878,
+      "learning_rate": 2.874098557692308e-05,
+      "loss": 1.1832,
+      "step": 1516
+    },
+    {
+      "epoch": 0.8849351028146419,
+      "grad_norm": 1.2172470092773438,
+      "learning_rate": 2.8725961538461537e-05,
+      "loss": 1.0312,
+      "step": 1517
+    },
+    {
+      "epoch": 0.8855184483010062,
+      "grad_norm": 1.279415249824524,
+      "learning_rate": 2.8710937500000002e-05,
+      "loss": 1.0095,
+      "step": 1518
+    },
+    {
+      "epoch": 0.8861017937873705,
+      "grad_norm": 1.3966619968414307,
+      "learning_rate": 2.8695913461538464e-05,
+      "loss": 0.921,
+      "step": 1519
+    },
+    {
+      "epoch": 0.8866851392737348,
+      "grad_norm": 1.2895619869232178,
+      "learning_rate": 2.8680889423076923e-05,
+      "loss": 0.9268,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8872684847600991,
+      "grad_norm": 1.188140869140625,
+      "learning_rate": 2.8665865384615388e-05,
+      "loss": 0.8283,
+      "step": 1521
+    },
+    {
+      "epoch": 0.8878518302464634,
+      "grad_norm": 1.200034499168396,
+      "learning_rate": 2.8650841346153847e-05,
+      "loss": 1.0621,
+      "step": 1522
+    },
+    {
+      "epoch": 0.8884351757328278,
+      "grad_norm": 1.3461898565292358,
+      "learning_rate": 2.863581730769231e-05,
+      "loss": 0.9668,
+      "step": 1523
+    },
+    {
+      "epoch": 0.8890185212191921,
+      "grad_norm": 1.2042393684387207,
+      "learning_rate": 2.8620793269230774e-05,
+      "loss": 1.145,
+      "step": 1524
+    },
+    {
+      "epoch": 0.8896018667055564,
+      "grad_norm": 1.2944340705871582,
+      "learning_rate": 2.860576923076923e-05,
+      "loss": 0.8788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.8901852121919207,
+      "grad_norm": 1.2656347751617432,
+      "learning_rate": 2.8590745192307694e-05,
+      "loss": 0.826,
+      "step": 1526
+    },
+    {
+      "epoch": 0.890768557678285,
+      "grad_norm": 1.079966425895691,
+      "learning_rate": 2.857572115384616e-05,
+      "loss": 0.9062,
+      "step": 1527
+    },
+    {
+      "epoch": 0.8913519031646493,
+      "grad_norm": 1.440693736076355,
+      "learning_rate": 2.8560697115384615e-05,
+      "loss": 1.0446,
+      "step": 1528
+    },
+    {
+      "epoch": 0.8919352486510136,
+      "grad_norm": 1.3850165605545044,
+      "learning_rate": 2.854567307692308e-05,
+      "loss": 0.9395,
+      "step": 1529
+    },
+    {
+      "epoch": 0.8925185941373779,
+      "grad_norm": 1.3351702690124512,
+      "learning_rate": 2.853064903846154e-05,
+      "loss": 1.1362,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8931019396237422,
+      "grad_norm": 0.9820516109466553,
+      "learning_rate": 2.8515625e-05,
+      "loss": 0.8793,
+      "step": 1531
+    },
+    {
+      "epoch": 0.8936852851101065,
+      "grad_norm": 1.218936562538147,
+      "learning_rate": 2.8500600961538466e-05,
+      "loss": 0.9328,
+      "step": 1532
+    },
+    {
+      "epoch": 0.8942686305964708,
+      "grad_norm": 1.2848559617996216,
+      "learning_rate": 2.8485576923076924e-05,
+      "loss": 0.955,
+      "step": 1533
+    },
+    {
+      "epoch": 0.8948519760828351,
+      "grad_norm": 1.3455287218093872,
+      "learning_rate": 2.8470552884615386e-05,
+      "loss": 0.9611,
+      "step": 1534
+    },
+    {
+      "epoch": 0.8954353215691994,
+      "grad_norm": 1.62530517578125,
+      "learning_rate": 2.8455528846153844e-05,
+      "loss": 0.9592,
+      "step": 1535
+    },
+    {
+      "epoch": 0.8960186670555637,
+      "grad_norm": 1.1084357500076294,
+      "learning_rate": 2.844050480769231e-05,
+      "loss": 1.1606,
+      "step": 1536
+    },
+    {
+      "epoch": 0.896602012541928,
+      "grad_norm": 1.11439847946167,
+      "learning_rate": 2.842548076923077e-05,
+      "loss": 0.9126,
+      "step": 1537
+    },
+    {
+      "epoch": 0.8971853580282922,
+      "grad_norm": 1.2643011808395386,
+      "learning_rate": 2.841045673076923e-05,
+      "loss": 1.1443,
+      "step": 1538
+    },
+    {
+      "epoch": 0.8977687035146565,
+      "grad_norm": 1.1699936389923096,
+      "learning_rate": 2.8395432692307695e-05,
+      "loss": 0.94,
+      "step": 1539
+    },
+    {
+      "epoch": 0.8983520490010208,
+      "grad_norm": 1.2582471370697021,
+      "learning_rate": 2.8380408653846157e-05,
+      "loss": 1.0579,
+      "step": 1540
+    },
+    {
+      "epoch": 0.8989353944873851,
+      "grad_norm": 1.1401822566986084,
+      "learning_rate": 2.8365384615384616e-05,
+      "loss": 0.8857,
+      "step": 1541
+    },
+    {
+      "epoch": 0.8995187399737494,
+      "grad_norm": 1.2778136730194092,
+      "learning_rate": 2.835036057692308e-05,
+      "loss": 0.7792,
+      "step": 1542
+    },
+    {
+      "epoch": 0.9001020854601137,
+      "grad_norm": 1.3046090602874756,
+      "learning_rate": 2.8335336538461536e-05,
+      "loss": 1.0387,
+      "step": 1543
+    },
+    {
+      "epoch": 0.900685430946478,
+      "grad_norm": 1.1095534563064575,
+      "learning_rate": 2.83203125e-05,
+      "loss": 0.8712,
+      "step": 1544
+    },
+    {
+      "epoch": 0.9012687764328423,
+      "grad_norm": 1.116352915763855,
+      "learning_rate": 2.8305288461538467e-05,
+      "loss": 0.9245,
+      "step": 1545
+    },
+    {
+      "epoch": 0.9018521219192066,
+      "grad_norm": 1.2703346014022827,
+      "learning_rate": 2.8290264423076922e-05,
+      "loss": 0.8139,
+      "step": 1546
+    },
+    {
+      "epoch": 0.9024354674055709,
+      "grad_norm": 1.2357991933822632,
+      "learning_rate": 2.8275240384615387e-05,
+      "loss": 1.1956,
+      "step": 1547
+    },
+    {
+      "epoch": 0.9030188128919352,
+      "grad_norm": 1.0954554080963135,
+      "learning_rate": 2.8260216346153846e-05,
+      "loss": 0.8301,
+      "step": 1548
+    },
+    {
+      "epoch": 0.9036021583782996,
+      "grad_norm": 1.6321262121200562,
+      "learning_rate": 2.8245192307692307e-05,
+      "loss": 0.9625,
+      "step": 1549
+    },
+    {
+      "epoch": 0.9041855038646639,
+      "grad_norm": 1.3000946044921875,
+      "learning_rate": 2.8230168269230773e-05,
+      "loss": 0.9551,
+      "step": 1550
+    },
+    {
+      "epoch": 0.9047688493510282,
+      "grad_norm": 1.1882951259613037,
+      "learning_rate": 2.821514423076923e-05,
+      "loss": 0.8661,
+      "step": 1551
+    },
+    {
+      "epoch": 0.9053521948373925,
+      "grad_norm": 1.136929988861084,
+      "learning_rate": 2.8200120192307693e-05,
+      "loss": 0.8345,
+      "step": 1552
+    },
+    {
+      "epoch": 0.9059355403237568,
+      "grad_norm": 1.2074054479599,
+      "learning_rate": 2.818509615384616e-05,
+      "loss": 1.0495,
+      "step": 1553
+    },
+    {
+      "epoch": 0.9065188858101211,
+      "grad_norm": 1.0739816427230835,
+      "learning_rate": 2.8170072115384617e-05,
+      "loss": 0.8297,
+      "step": 1554
+    },
+    {
+      "epoch": 0.9071022312964854,
+      "grad_norm": 1.0656182765960693,
+      "learning_rate": 2.815504807692308e-05,
+      "loss": 0.9902,
+      "step": 1555
+    },
+    {
+      "epoch": 0.9076855767828497,
+      "grad_norm": 1.072704553604126,
+      "learning_rate": 2.8140024038461537e-05,
+      "loss": 1.1225,
+      "step": 1556
+    },
+    {
+      "epoch": 0.908268922269214,
+      "grad_norm": 1.2448283433914185,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 1.2656,
+      "step": 1557
+    },
+    {
+      "epoch": 0.9088522677555783,
+      "grad_norm": 1.2833285331726074,
+      "learning_rate": 2.8109975961538465e-05,
+      "loss": 1.297,
+      "step": 1558
+    },
+    {
+      "epoch": 0.9094356132419426,
+      "grad_norm": 0.9518376588821411,
+      "learning_rate": 2.8094951923076923e-05,
+      "loss": 1.2577,
+      "step": 1559
+    },
+    {
+      "epoch": 0.9100189587283068,
+      "grad_norm": 1.1654349565505981,
+      "learning_rate": 2.8079927884615388e-05,
+      "loss": 0.9981,
+      "step": 1560
+    },
+    {
+      "epoch": 0.9106023042146711,
+      "grad_norm": 1.316017746925354,
+      "learning_rate": 2.8064903846153843e-05,
+      "loss": 0.8892,
+      "step": 1561
+    },
+    {
+      "epoch": 0.9111856497010354,
+      "grad_norm": 1.2900649309158325,
+      "learning_rate": 2.804987980769231e-05,
+      "loss": 0.9713,
+      "step": 1562
+    },
+    {
+      "epoch": 0.9117689951873997,
+      "grad_norm": 1.169692039489746,
+      "learning_rate": 2.8034855769230774e-05,
+      "loss": 1.0375,
+      "step": 1563
+    },
+    {
+      "epoch": 0.912352340673764,
+      "grad_norm": 1.1953729391098022,
+      "learning_rate": 2.801983173076923e-05,
+      "loss": 0.8052,
+      "step": 1564
+    },
+    {
+      "epoch": 0.9129356861601283,
+      "grad_norm": 1.3699668645858765,
+      "learning_rate": 2.8004807692307694e-05,
+      "loss": 0.968,
+      "step": 1565
+    },
+    {
+      "epoch": 0.9135190316464926,
+      "grad_norm": 1.2539699077606201,
+      "learning_rate": 2.7989783653846156e-05,
+      "loss": 1.0042,
+      "step": 1566
+    },
+    {
+      "epoch": 0.9141023771328569,
+      "grad_norm": 1.1665087938308716,
+      "learning_rate": 2.7974759615384615e-05,
+      "loss": 0.949,
+      "step": 1567
+    },
+    {
+      "epoch": 0.9146857226192212,
+      "grad_norm": 1.421539068222046,
+      "learning_rate": 2.795973557692308e-05,
+      "loss": 0.9799,
+      "step": 1568
+    },
+    {
+      "epoch": 0.9152690681055855,
+      "grad_norm": 1.0832866430282593,
+      "learning_rate": 2.794471153846154e-05,
+      "loss": 0.8219,
+      "step": 1569
+    },
+    {
+      "epoch": 0.9158524135919498,
+      "grad_norm": 1.2853337526321411,
+      "learning_rate": 2.79296875e-05,
+      "loss": 1.0887,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9164357590783141,
+      "grad_norm": 1.1525390148162842,
+      "learning_rate": 2.7914663461538466e-05,
+      "loss": 1.0594,
+      "step": 1571
+    },
+    {
+      "epoch": 0.9170191045646784,
+      "grad_norm": 1.3522123098373413,
+      "learning_rate": 2.7899639423076924e-05,
+      "loss": 1.012,
+      "step": 1572
+    },
+    {
+      "epoch": 0.9176024500510427,
+      "grad_norm": 1.1640657186508179,
+      "learning_rate": 2.7884615384615386e-05,
+      "loss": 0.9551,
+      "step": 1573
+    },
+    {
+      "epoch": 0.918185795537407,
+      "grad_norm": 1.099858045578003,
+      "learning_rate": 2.7869591346153845e-05,
+      "loss": 0.9387,
+      "step": 1574
+    },
+    {
+      "epoch": 0.9187691410237713,
+      "grad_norm": 1.295347809791565,
+      "learning_rate": 2.785456730769231e-05,
+      "loss": 0.9975,
+      "step": 1575
+    },
+    {
+      "epoch": 0.9193524865101357,
+      "grad_norm": 1.0438365936279297,
+      "learning_rate": 2.7839543269230772e-05,
+      "loss": 0.9287,
+      "step": 1576
+    },
+    {
+      "epoch": 0.9199358319965,
+      "grad_norm": 1.5085781812667847,
+      "learning_rate": 2.782451923076923e-05,
+      "loss": 0.8949,
+      "step": 1577
+    },
+    {
+      "epoch": 0.9205191774828643,
+      "grad_norm": 1.1125335693359375,
+      "learning_rate": 2.7809495192307696e-05,
+      "loss": 1.1754,
+      "step": 1578
+    },
+    {
+      "epoch": 0.9211025229692286,
+      "grad_norm": 1.4580250978469849,
+      "learning_rate": 2.7794471153846157e-05,
+      "loss": 1.188,
+      "step": 1579
+    },
+    {
+      "epoch": 0.9216858684555929,
+      "grad_norm": 1.5720735788345337,
+      "learning_rate": 2.7779447115384616e-05,
+      "loss": 0.97,
+      "step": 1580
+    },
+    {
+      "epoch": 0.9222692139419572,
+      "grad_norm": 1.5077130794525146,
+      "learning_rate": 2.776442307692308e-05,
+      "loss": 0.9886,
+      "step": 1581
+    },
+    {
+      "epoch": 0.9228525594283215,
+      "grad_norm": 0.9293258786201477,
+      "learning_rate": 2.7749399038461536e-05,
+      "loss": 0.9714,
+      "step": 1582
+    },
+    {
+      "epoch": 0.9234359049146857,
+      "grad_norm": 1.5474597215652466,
+      "learning_rate": 2.7734375e-05,
+      "loss": 0.9918,
+      "step": 1583
+    },
+    {
+      "epoch": 0.92401925040105,
+      "grad_norm": 1.3098106384277344,
+      "learning_rate": 2.7719350961538463e-05,
+      "loss": 1.1593,
+      "step": 1584
+    },
+    {
+      "epoch": 0.9246025958874143,
+      "grad_norm": 1.1443274021148682,
+      "learning_rate": 2.7704326923076922e-05,
+      "loss": 1.0293,
+      "step": 1585
+    },
+    {
+      "epoch": 0.9251859413737786,
+      "grad_norm": 1.362916111946106,
+      "learning_rate": 2.7689302884615387e-05,
+      "loss": 1.2039,
+      "step": 1586
+    },
+    {
+      "epoch": 0.9257692868601429,
+      "grad_norm": 1.2175136804580688,
+      "learning_rate": 2.7674278846153846e-05,
+      "loss": 0.9562,
+      "step": 1587
+    },
+    {
+      "epoch": 0.9263526323465072,
+      "grad_norm": 1.4667470455169678,
+      "learning_rate": 2.7659254807692308e-05,
+      "loss": 0.9903,
+      "step": 1588
+    },
+    {
+      "epoch": 0.9269359778328715,
+      "grad_norm": 1.317681908607483,
+      "learning_rate": 2.7644230769230773e-05,
+      "loss": 1.0178,
+      "step": 1589
+    },
+    {
+      "epoch": 0.9275193233192358,
+      "grad_norm": 1.2673649787902832,
+      "learning_rate": 2.762920673076923e-05,
+      "loss": 1.0001,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9281026688056001,
+      "grad_norm": 1.055626630783081,
+      "learning_rate": 2.7614182692307693e-05,
+      "loss": 1.0075,
+      "step": 1591
+    },
+    {
+      "epoch": 0.9286860142919644,
+      "grad_norm": 0.938339352607727,
+      "learning_rate": 2.759915865384616e-05,
+      "loss": 0.8492,
+      "step": 1592
+    },
+    {
+      "epoch": 0.9292693597783287,
+      "grad_norm": 1.1331336498260498,
+      "learning_rate": 2.7584134615384617e-05,
+      "loss": 1.1089,
+      "step": 1593
+    },
+    {
+      "epoch": 0.929852705264693,
+      "grad_norm": 0.9863076210021973,
+      "learning_rate": 2.756911057692308e-05,
+      "loss": 0.827,
+      "step": 1594
+    },
+    {
+      "epoch": 0.9304360507510573,
+      "grad_norm": 0.8568504452705383,
+      "learning_rate": 2.7554086538461537e-05,
+      "loss": 0.9642,
+      "step": 1595
+    },
+    {
+      "epoch": 0.9310193962374216,
+      "grad_norm": 1.3427821397781372,
+      "learning_rate": 2.7539062500000003e-05,
+      "loss": 1.0344,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9316027417237859,
+      "grad_norm": 1.1285762786865234,
+      "learning_rate": 2.7524038461538465e-05,
+      "loss": 0.9632,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9321860872101502,
+      "grad_norm": 1.079243779182434,
+      "learning_rate": 2.7509014423076923e-05,
+      "loss": 1.1732,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9327694326965145,
+      "grad_norm": 1.4731907844543457,
+      "learning_rate": 2.749399038461539e-05,
+      "loss": 1.0798,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9333527781828788,
+      "grad_norm": 1.8456404209136963,
+      "learning_rate": 2.7478966346153844e-05,
+      "loss": 0.8122,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9333527781828788,
+      "eval_loss_squad": 0.8185375917516649,
+      "eval_perplexity": 8.537540772600966,
+      "eval_perplexity_reconstruct": 1.9311449308939752,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9339361236692431,
+      "grad_norm": 1.5223437547683716,
+      "learning_rate": 2.746394230769231e-05,
+      "loss": 1.0494,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9345194691556075,
+      "grad_norm": 1.1983387470245361,
+      "learning_rate": 2.744891826923077e-05,
+      "loss": 1.1375,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9351028146419718,
+      "grad_norm": 1.253590703010559,
+      "learning_rate": 2.743389423076923e-05,
+      "loss": 0.86,
+      "step": 1603
+    },
+    {
+      "epoch": 0.935686160128336,
+      "grad_norm": 1.5768380165100098,
+      "learning_rate": 2.7418870192307695e-05,
+      "loss": 1.0707,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9362695056147003,
+      "grad_norm": 1.1141117811203003,
+      "learning_rate": 2.7403846153846156e-05,
+      "loss": 0.9249,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9368528511010646,
+      "grad_norm": 1.229311466217041,
+      "learning_rate": 2.7388822115384615e-05,
+      "loss": 1.1422,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9374361965874289,
+      "grad_norm": 1.0226390361785889,
+      "learning_rate": 2.737379807692308e-05,
+      "loss": 1.1093,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9380195420737932,
+      "grad_norm": 1.055548071861267,
+      "learning_rate": 2.735877403846154e-05,
+      "loss": 1.1264,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9386028875601575,
+      "grad_norm": 1.326737642288208,
+      "learning_rate": 2.734375e-05,
+      "loss": 0.7929,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9391862330465218,
+      "grad_norm": 1.2810485363006592,
+      "learning_rate": 2.7328725961538466e-05,
+      "loss": 0.9209,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9397695785328861,
+      "grad_norm": 1.1614972352981567,
+      "learning_rate": 2.7313701923076924e-05,
+      "loss": 0.8571,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9403529240192504,
+      "grad_norm": 1.063585638999939,
+      "learning_rate": 2.7298677884615386e-05,
+      "loss": 1.0951,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9409362695056147,
+      "grad_norm": 1.1287500858306885,
+      "learning_rate": 2.7283653846153845e-05,
+      "loss": 0.9155,
+      "step": 1613
+    },
+    {
+      "epoch": 0.941519614991979,
+      "grad_norm": 1.202637791633606,
+      "learning_rate": 2.726862980769231e-05,
+      "loss": 0.922,
+      "step": 1614
+    },
+    {
+      "epoch": 0.9421029604783433,
+      "grad_norm": 0.9116735458374023,
+      "learning_rate": 2.7253605769230772e-05,
+      "loss": 1.1287,
+      "step": 1615
+    },
+    {
+      "epoch": 0.9426863059647076,
+      "grad_norm": 1.2105416059494019,
+      "learning_rate": 2.723858173076923e-05,
+      "loss": 1.0168,
+      "step": 1616
+    },
+    {
+      "epoch": 0.9432696514510719,
+      "grad_norm": 1.4131770133972168,
+      "learning_rate": 2.7223557692307696e-05,
+      "loss": 0.8181,
+      "step": 1617
+    },
+    {
+      "epoch": 0.9438529969374362,
+      "grad_norm": 1.2585093975067139,
+      "learning_rate": 2.7208533653846158e-05,
+      "loss": 0.8861,
+      "step": 1618
+    },
+    {
+      "epoch": 0.9444363424238005,
+      "grad_norm": 1.1039953231811523,
+      "learning_rate": 2.7193509615384616e-05,
+      "loss": 0.9609,
+      "step": 1619
+    },
+    {
+      "epoch": 0.9450196879101648,
+      "grad_norm": 1.245587706565857,
+      "learning_rate": 2.7178485576923078e-05,
+      "loss": 0.8698,
+      "step": 1620
+    },
+    {
+      "epoch": 0.9456030333965291,
+      "grad_norm": 1.325048565864563,
+      "learning_rate": 2.7163461538461536e-05,
+      "loss": 0.8585,
+      "step": 1621
+    },
+    {
+      "epoch": 0.9461863788828934,
+      "grad_norm": 1.3289551734924316,
+      "learning_rate": 2.7148437500000002e-05,
+      "loss": 1.089,
+      "step": 1622
+    },
+    {
+      "epoch": 0.9467697243692577,
+      "grad_norm": 1.2364208698272705,
+      "learning_rate": 2.7133413461538464e-05,
+      "loss": 1.0367,
+      "step": 1623
+    },
+    {
+      "epoch": 0.947353069855622,
+      "grad_norm": 1.2870745658874512,
+      "learning_rate": 2.7118389423076922e-05,
+      "loss": 1.1767,
+      "step": 1624
+    },
+    {
+      "epoch": 0.9479364153419862,
+      "grad_norm": 1.1112834215164185,
+      "learning_rate": 2.7103365384615387e-05,
+      "loss": 0.9196,
+      "step": 1625
+    },
+    {
+      "epoch": 0.9485197608283505,
+      "grad_norm": 1.0553075075149536,
+      "learning_rate": 2.7088341346153846e-05,
+      "loss": 1.01,
+      "step": 1626
+    },
+    {
+      "epoch": 0.9491031063147148,
+      "grad_norm": 1.2069644927978516,
+      "learning_rate": 2.7073317307692308e-05,
+      "loss": 1.0086,
+      "step": 1627
+    },
+    {
+      "epoch": 0.9496864518010791,
+      "grad_norm": 1.0190929174423218,
+      "learning_rate": 2.7058293269230773e-05,
+      "loss": 0.9053,
+      "step": 1628
+    },
+    {
+      "epoch": 0.9502697972874435,
+      "grad_norm": 1.1159418821334839,
+      "learning_rate": 2.704326923076923e-05,
+      "loss": 0.9234,
+      "step": 1629
+    },
+    {
+      "epoch": 0.9508531427738078,
+      "grad_norm": 1.2140814065933228,
+      "learning_rate": 2.7028245192307693e-05,
+      "loss": 0.8535,
+      "step": 1630
+    },
+    {
+      "epoch": 0.9514364882601721,
+      "grad_norm": 1.3430871963500977,
+      "learning_rate": 2.701322115384616e-05,
+      "loss": 1.1157,
+      "step": 1631
+    },
+    {
+      "epoch": 0.9520198337465364,
+      "grad_norm": 1.2624151706695557,
+      "learning_rate": 2.6998197115384617e-05,
+      "loss": 0.9076,
+      "step": 1632
+    },
+    {
+      "epoch": 0.9526031792329007,
+      "grad_norm": 1.325024127960205,
+      "learning_rate": 2.698317307692308e-05,
+      "loss": 0.9707,
+      "step": 1633
+    },
+    {
+      "epoch": 0.953186524719265,
+      "grad_norm": 1.1497493982315063,
+      "learning_rate": 2.6968149038461538e-05,
+      "loss": 0.9365,
+      "step": 1634
+    },
+    {
+      "epoch": 0.9537698702056293,
+      "grad_norm": 1.139591097831726,
+      "learning_rate": 2.6953125000000003e-05,
+      "loss": 1.0114,
+      "step": 1635
+    },
+    {
+      "epoch": 0.9543532156919936,
+      "grad_norm": 1.202013373374939,
+      "learning_rate": 2.6938100961538465e-05,
+      "loss": 0.8738,
+      "step": 1636
+    },
+    {
+      "epoch": 0.9549365611783579,
+      "grad_norm": 1.1490192413330078,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.9644,
+      "step": 1637
+    },
+    {
+      "epoch": 0.9555199066647222,
+      "grad_norm": 1.1956679821014404,
+      "learning_rate": 2.6908052884615385e-05,
+      "loss": 0.8404,
+      "step": 1638
+    },
+    {
+      "epoch": 0.9561032521510865,
+      "grad_norm": 1.016502857208252,
+      "learning_rate": 2.6893028846153844e-05,
+      "loss": 0.9757,
+      "step": 1639
+    },
+    {
+      "epoch": 0.9566865976374508,
+      "grad_norm": 1.1245112419128418,
+      "learning_rate": 2.687800480769231e-05,
+      "loss": 0.8616,
+      "step": 1640
+    },
+    {
+      "epoch": 0.9572699431238151,
+      "grad_norm": 1.326154351234436,
+      "learning_rate": 2.686298076923077e-05,
+      "loss": 0.8661,
+      "step": 1641
+    },
+    {
+      "epoch": 0.9578532886101794,
+      "grad_norm": 1.2109782695770264,
+      "learning_rate": 2.684795673076923e-05,
+      "loss": 0.8722,
+      "step": 1642
+    },
+    {
+      "epoch": 0.9584366340965437,
+      "grad_norm": 1.2979744672775269,
+      "learning_rate": 2.6832932692307695e-05,
+      "loss": 0.9699,
+      "step": 1643
+    },
+    {
+      "epoch": 0.959019979582908,
+      "grad_norm": 1.4275164604187012,
+      "learning_rate": 2.6817908653846157e-05,
+      "loss": 1.3414,
+      "step": 1644
+    },
+    {
+      "epoch": 0.9596033250692723,
+      "grad_norm": 2.1155264377593994,
+      "learning_rate": 2.6802884615384615e-05,
+      "loss": 1.061,
+      "step": 1645
+    },
+    {
+      "epoch": 0.9601866705556366,
+      "grad_norm": 1.1777634620666504,
+      "learning_rate": 2.678786057692308e-05,
+      "loss": 0.8194,
+      "step": 1646
+    },
+    {
+      "epoch": 0.9607700160420009,
+      "grad_norm": 1.3087421655654907,
+      "learning_rate": 2.677283653846154e-05,
+      "loss": 1.084,
+      "step": 1647
+    },
+    {
+      "epoch": 0.9613533615283651,
+      "grad_norm": 1.2843700647354126,
+      "learning_rate": 2.67578125e-05,
+      "loss": 0.9156,
+      "step": 1648
+    },
+    {
+      "epoch": 0.9619367070147294,
+      "grad_norm": 1.2804768085479736,
+      "learning_rate": 2.6742788461538466e-05,
+      "loss": 1.1256,
+      "step": 1649
+    },
+    {
+      "epoch": 0.9625200525010937,
+      "grad_norm": 1.1783385276794434,
+      "learning_rate": 2.6727764423076925e-05,
+      "loss": 0.7652,
+      "step": 1650
+    },
+    {
+      "epoch": 0.963103397987458,
+      "grad_norm": 1.0474493503570557,
+      "learning_rate": 2.6712740384615386e-05,
+      "loss": 1.1225,
+      "step": 1651
+    },
+    {
+      "epoch": 0.9636867434738223,
+      "grad_norm": 1.003244161605835,
+      "learning_rate": 2.6697716346153845e-05,
+      "loss": 0.9582,
+      "step": 1652
+    },
+    {
+      "epoch": 0.9642700889601866,
+      "grad_norm": 1.3323856592178345,
+      "learning_rate": 2.668269230769231e-05,
+      "loss": 0.9056,
+      "step": 1653
+    },
+    {
+      "epoch": 0.9648534344465509,
+      "grad_norm": 0.8693028092384338,
+      "learning_rate": 2.6667668269230772e-05,
+      "loss": 1.2649,
+      "step": 1654
+    },
+    {
+      "epoch": 0.9654367799329153,
+      "grad_norm": 1.2013863325119019,
+      "learning_rate": 2.665264423076923e-05,
+      "loss": 1.0149,
+      "step": 1655
+    },
+    {
+      "epoch": 0.9660201254192796,
+      "grad_norm": 1.061509132385254,
+      "learning_rate": 2.6637620192307692e-05,
+      "loss": 0.9357,
+      "step": 1656
+    },
+    {
+      "epoch": 0.9666034709056439,
+      "grad_norm": 1.086358904838562,
+      "learning_rate": 2.6622596153846158e-05,
+      "loss": 0.7849,
+      "step": 1657
+    },
+    {
+      "epoch": 0.9671868163920082,
+      "grad_norm": 1.2358739376068115,
+      "learning_rate": 2.6607572115384616e-05,
+      "loss": 1.0885,
+      "step": 1658
+    },
+    {
+      "epoch": 0.9677701618783725,
+      "grad_norm": 1.0823692083358765,
+      "learning_rate": 2.6592548076923078e-05,
+      "loss": 0.8326,
+      "step": 1659
+    },
+    {
+      "epoch": 0.9683535073647368,
+      "grad_norm": 2.211810827255249,
+      "learning_rate": 2.6577524038461537e-05,
+      "loss": 1.0041,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9689368528511011,
+      "grad_norm": 1.0642420053482056,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 1.0824,
+      "step": 1661
+    },
+    {
+      "epoch": 0.9695201983374654,
+      "grad_norm": 1.0718188285827637,
+      "learning_rate": 2.6547475961538464e-05,
+      "loss": 1.024,
+      "step": 1662
+    },
+    {
+      "epoch": 0.9701035438238297,
+      "grad_norm": 0.972342848777771,
+      "learning_rate": 2.6532451923076922e-05,
+      "loss": 1.0009,
+      "step": 1663
+    },
+    {
+      "epoch": 0.970686889310194,
+      "grad_norm": 1.5624159574508667,
+      "learning_rate": 2.6517427884615388e-05,
+      "loss": 0.9796,
+      "step": 1664
+    },
+    {
+      "epoch": 0.9712702347965583,
+      "grad_norm": 1.251407504081726,
+      "learning_rate": 2.6502403846153846e-05,
+      "loss": 1.0386,
+      "step": 1665
+    },
+    {
+      "epoch": 0.9718535802829226,
+      "grad_norm": 1.3637851476669312,
+      "learning_rate": 2.6487379807692308e-05,
+      "loss": 1.0674,
+      "step": 1666
+    },
+    {
+      "epoch": 0.9724369257692869,
+      "grad_norm": 1.1037849187850952,
+      "learning_rate": 2.6472355769230773e-05,
+      "loss": 0.8173,
+      "step": 1667
+    },
+    {
+      "epoch": 0.9730202712556512,
+      "grad_norm": 1.0050426721572876,
+      "learning_rate": 2.6457331730769232e-05,
+      "loss": 0.9875,
+      "step": 1668
+    },
+    {
+      "epoch": 0.9736036167420155,
+      "grad_norm": 1.2693151235580444,
+      "learning_rate": 2.6442307692307694e-05,
+      "loss": 1.0777,
+      "step": 1669
+    },
+    {
+      "epoch": 0.9741869622283797,
+      "grad_norm": 1.139029622077942,
+      "learning_rate": 2.642728365384616e-05,
+      "loss": 1.1184,
+      "step": 1670
+    },
+    {
+      "epoch": 0.974770307714744,
+      "grad_norm": 1.156640887260437,
+      "learning_rate": 2.6412259615384617e-05,
+      "loss": 1.1217,
+      "step": 1671
+    },
+    {
+      "epoch": 0.9753536532011083,
+      "grad_norm": 1.2772996425628662,
+      "learning_rate": 2.639723557692308e-05,
+      "loss": 0.8868,
+      "step": 1672
+    },
+    {
+      "epoch": 0.9759369986874726,
+      "grad_norm": 1.1612448692321777,
+      "learning_rate": 2.6382211538461538e-05,
+      "loss": 0.912,
+      "step": 1673
+    },
+    {
+      "epoch": 0.9765203441738369,
+      "grad_norm": 1.0551241636276245,
+      "learning_rate": 2.63671875e-05,
+      "loss": 0.9191,
+      "step": 1674
+    },
+    {
+      "epoch": 0.9771036896602012,
+      "grad_norm": 1.092178225517273,
+      "learning_rate": 2.6352163461538465e-05,
+      "loss": 0.8177,
+      "step": 1675
+    },
+    {
+      "epoch": 0.9776870351465655,
+      "grad_norm": 1.261635422706604,
+      "learning_rate": 2.6337139423076923e-05,
+      "loss": 0.8914,
+      "step": 1676
+    },
+    {
+      "epoch": 0.9782703806329298,
+      "grad_norm": 1.3369121551513672,
+      "learning_rate": 2.6322115384615385e-05,
+      "loss": 0.7757,
+      "step": 1677
+    },
+    {
+      "epoch": 0.9788537261192941,
+      "grad_norm": 1.1056147813796997,
+      "learning_rate": 2.6307091346153844e-05,
+      "loss": 0.8988,
+      "step": 1678
+    },
+    {
+      "epoch": 0.9794370716056584,
+      "grad_norm": 1.2153946161270142,
+      "learning_rate": 2.629206730769231e-05,
+      "loss": 1.0653,
+      "step": 1679
+    },
+    {
+      "epoch": 0.9800204170920227,
+      "grad_norm": 1.085607886314392,
+      "learning_rate": 2.627704326923077e-05,
+      "loss": 1.0798,
+      "step": 1680
+    },
+    {
+      "epoch": 0.980603762578387,
+      "grad_norm": 1.152902603149414,
+      "learning_rate": 2.626201923076923e-05,
+      "loss": 1.193,
+      "step": 1681
+    },
+    {
+      "epoch": 0.9811871080647514,
+      "grad_norm": 1.100484848022461,
+      "learning_rate": 2.6246995192307695e-05,
+      "loss": 0.8515,
+      "step": 1682
+    },
+    {
+      "epoch": 0.9817704535511157,
+      "grad_norm": 0.9451040029525757,
+      "learning_rate": 2.6231971153846157e-05,
+      "loss": 1.015,
+      "step": 1683
+    },
+    {
+      "epoch": 0.98235379903748,
+      "grad_norm": 1.5376704931259155,
+      "learning_rate": 2.6216947115384615e-05,
+      "loss": 1.1115,
+      "step": 1684
+    },
+    {
+      "epoch": 0.9829371445238443,
+      "grad_norm": 1.2775416374206543,
+      "learning_rate": 2.620192307692308e-05,
+      "loss": 0.9463,
+      "step": 1685
+    },
+    {
+      "epoch": 0.9835204900102086,
+      "grad_norm": 1.0612167119979858,
+      "learning_rate": 2.618689903846154e-05,
+      "loss": 1.3035,
+      "step": 1686
+    },
+    {
+      "epoch": 0.9841038354965729,
+      "grad_norm": 1.1903184652328491,
+      "learning_rate": 2.6171875e-05,
+      "loss": 1.0081,
+      "step": 1687
+    },
+    {
+      "epoch": 0.9846871809829372,
+      "grad_norm": 1.1516382694244385,
+      "learning_rate": 2.6156850961538466e-05,
+      "loss": 0.8602,
+      "step": 1688
+    },
+    {
+      "epoch": 0.9852705264693015,
+      "grad_norm": 1.1271847486495972,
+      "learning_rate": 2.6141826923076925e-05,
+      "loss": 1.109,
+      "step": 1689
+    },
+    {
+      "epoch": 0.9858538719556658,
+      "grad_norm": 1.0397697687149048,
+      "learning_rate": 2.6126802884615387e-05,
+      "loss": 1.2407,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9864372174420301,
+      "grad_norm": 1.1553921699523926,
+      "learning_rate": 2.6111778846153845e-05,
+      "loss": 1.0275,
+      "step": 1691
+    },
+    {
+      "epoch": 0.9870205629283944,
+      "grad_norm": 1.3742332458496094,
+      "learning_rate": 2.6096754807692307e-05,
+      "loss": 1.0156,
+      "step": 1692
+    },
+    {
+      "epoch": 0.9876039084147586,
+      "grad_norm": 1.3807793855667114,
+      "learning_rate": 2.6081730769230772e-05,
+      "loss": 1.0403,
+      "step": 1693
+    },
+    {
+      "epoch": 0.9881872539011229,
+      "grad_norm": 2.3124213218688965,
+      "learning_rate": 2.606670673076923e-05,
+      "loss": 1.065,
+      "step": 1694
+    },
+    {
+      "epoch": 0.9887705993874872,
+      "grad_norm": 1.1907039880752563,
+      "learning_rate": 2.6051682692307693e-05,
+      "loss": 0.7911,
+      "step": 1695
+    },
+    {
+      "epoch": 0.9893539448738515,
+      "grad_norm": 1.244017481803894,
+      "learning_rate": 2.6036658653846158e-05,
+      "loss": 1.0102,
+      "step": 1696
+    },
+    {
+      "epoch": 0.9899372903602158,
+      "grad_norm": 1.2556577920913696,
+      "learning_rate": 2.6021634615384616e-05,
+      "loss": 0.949,
+      "step": 1697
+    },
+    {
+      "epoch": 0.9905206358465801,
+      "grad_norm": 1.2664211988449097,
+      "learning_rate": 2.6006610576923078e-05,
+      "loss": 0.9661,
+      "step": 1698
+    },
+    {
+      "epoch": 0.9911039813329444,
+      "grad_norm": 1.1367192268371582,
+      "learning_rate": 2.5991586538461537e-05,
+      "loss": 1.094,
+      "step": 1699
+    },
+    {
+      "epoch": 0.9916873268193087,
+      "grad_norm": 1.4298202991485596,
+      "learning_rate": 2.5976562500000002e-05,
+      "loss": 1.2788,
+      "step": 1700
+    },
+    {
+      "epoch": 0.992270672305673,
+      "grad_norm": 1.2524197101593018,
+      "learning_rate": 2.5961538461538464e-05,
+      "loss": 0.7913,
+      "step": 1701
+    },
+    {
+      "epoch": 0.9928540177920373,
+      "grad_norm": 1.3437864780426025,
+      "learning_rate": 2.5946514423076922e-05,
+      "loss": 1.1432,
+      "step": 1702
+    },
+    {
+      "epoch": 0.9934373632784016,
+      "grad_norm": 1.105908751487732,
+      "learning_rate": 2.5931490384615388e-05,
+      "loss": 0.8405,
+      "step": 1703
+    },
+    {
+      "epoch": 0.9940207087647659,
+      "grad_norm": 0.9571016430854797,
+      "learning_rate": 2.5916466346153846e-05,
+      "loss": 0.7865,
+      "step": 1704
+    },
+    {
+      "epoch": 0.9946040542511302,
+      "grad_norm": 1.0247722864151,
+      "learning_rate": 2.5901442307692308e-05,
+      "loss": 0.8726,
+      "step": 1705
+    },
+    {
+      "epoch": 0.9951873997374945,
+      "grad_norm": 1.160334825515747,
+      "learning_rate": 2.5886418269230773e-05,
+      "loss": 0.8283,
+      "step": 1706
+    },
+    {
+      "epoch": 0.9957707452238588,
+      "grad_norm": 1.4978736639022827,
+      "learning_rate": 2.5871394230769232e-05,
+      "loss": 1.101,
+      "step": 1707
+    },
+    {
+      "epoch": 0.9963540907102232,
+      "grad_norm": 1.0514538288116455,
+      "learning_rate": 2.5856370192307694e-05,
+      "loss": 0.9573,
+      "step": 1708
+    },
+    {
+      "epoch": 0.9969374361965875,
+      "grad_norm": 1.091744303703308,
+      "learning_rate": 2.584134615384616e-05,
+      "loss": 1.0072,
+      "step": 1709
+    },
+    {
+      "epoch": 0.9975207816829518,
+      "grad_norm": 1.261958122253418,
+      "learning_rate": 2.5826322115384614e-05,
+      "loss": 0.8659,
+      "step": 1710
+    },
+    {
+      "epoch": 0.9981041271693161,
+      "grad_norm": 1.4826487302780151,
+      "learning_rate": 2.581129807692308e-05,
+      "loss": 0.9641,
+      "step": 1711
+    },
+    {
+      "epoch": 0.9986874726556804,
+      "grad_norm": 1.4628450870513916,
+      "learning_rate": 2.5796274038461538e-05,
+      "loss": 0.7998,
+      "step": 1712
+    },
+    {
+      "epoch": 0.9992708181420447,
+      "grad_norm": 1.1515614986419678,
+      "learning_rate": 2.578125e-05,
+      "loss": 1.1375,
+      "step": 1713
+    },
+    {
+      "epoch": 0.999854163628409,
+      "grad_norm": 1.118147373199463,
+      "learning_rate": 2.5766225961538465e-05,
+      "loss": 0.7665,
+      "step": 1714
+    },
+    {
+      "epoch": 1.0004375091147732,
+      "grad_norm": 0.9037253260612488,
+      "learning_rate": 2.5751201923076924e-05,
+      "loss": 0.7602,
+      "step": 1715
+    },
+    {
+      "epoch": 1.0010208546011374,
+      "grad_norm": 1.0712151527404785,
+      "learning_rate": 2.5736177884615386e-05,
+      "loss": 0.7995,
+      "step": 1716
+    },
+    {
+      "epoch": 1.0016042000875018,
+      "grad_norm": 0.8427433371543884,
+      "learning_rate": 2.5721153846153844e-05,
+      "loss": 0.9496,
+      "step": 1717
+    },
+    {
+      "epoch": 1.002187545573866,
+      "grad_norm": 0.9450975060462952,
+      "learning_rate": 2.570612980769231e-05,
+      "loss": 0.8904,
+      "step": 1718
+    },
+    {
+      "epoch": 1.0027708910602304,
+      "grad_norm": 1.037550926208496,
+      "learning_rate": 2.569110576923077e-05,
+      "loss": 0.8008,
+      "step": 1719
+    },
+    {
+      "epoch": 1.0033542365465946,
+      "grad_norm": 0.8638540506362915,
+      "learning_rate": 2.567608173076923e-05,
+      "loss": 0.639,
+      "step": 1720
+    },
+    {
+      "epoch": 1.003937582032959,
+      "grad_norm": 1.1321598291397095,
+      "learning_rate": 2.5661057692307695e-05,
+      "loss": 1.0667,
+      "step": 1721
+    },
+    {
+      "epoch": 1.0045209275193234,
+      "grad_norm": 1.1372548341751099,
+      "learning_rate": 2.5646033653846157e-05,
+      "loss": 0.8085,
+      "step": 1722
+    },
+    {
+      "epoch": 1.0051042730056876,
+      "grad_norm": 1.0561586618423462,
+      "learning_rate": 2.5631009615384615e-05,
+      "loss": 0.8129,
+      "step": 1723
+    },
+    {
+      "epoch": 1.005687618492052,
+      "grad_norm": 1.0580649375915527,
+      "learning_rate": 2.561598557692308e-05,
+      "loss": 0.6149,
+      "step": 1724
+    },
+    {
+      "epoch": 1.0062709639784162,
+      "grad_norm": 1.236765742301941,
+      "learning_rate": 2.560096153846154e-05,
+      "loss": 0.7003,
+      "step": 1725
+    },
+    {
+      "epoch": 1.0068543094647806,
+      "grad_norm": 1.0639208555221558,
+      "learning_rate": 2.55859375e-05,
+      "loss": 0.8107,
+      "step": 1726
+    },
+    {
+      "epoch": 1.0074376549511448,
+      "grad_norm": 1.3609492778778076,
+      "learning_rate": 2.5570913461538466e-05,
+      "loss": 0.9411,
+      "step": 1727
+    },
+    {
+      "epoch": 1.0080210004375092,
+      "grad_norm": 1.1644700765609741,
+      "learning_rate": 2.555588942307692e-05,
+      "loss": 0.701,
+      "step": 1728
+    },
+    {
+      "epoch": 1.0086043459238734,
+      "grad_norm": 1.2108243703842163,
+      "learning_rate": 2.5540865384615387e-05,
+      "loss": 0.9003,
+      "step": 1729
+    },
+    {
+      "epoch": 1.0091876914102378,
+      "grad_norm": 1.1662452220916748,
+      "learning_rate": 2.5525841346153845e-05,
+      "loss": 0.8709,
+      "step": 1730
+    },
+    {
+      "epoch": 1.009771036896602,
+      "grad_norm": 1.0945322513580322,
+      "learning_rate": 2.5510817307692307e-05,
+      "loss": 1.0255,
+      "step": 1731
+    },
+    {
+      "epoch": 1.0103543823829664,
+      "grad_norm": 1.0760470628738403,
+      "learning_rate": 2.5495793269230772e-05,
+      "loss": 0.7592,
+      "step": 1732
+    },
+    {
+      "epoch": 1.0109377278693306,
+      "grad_norm": 1.3232825994491577,
+      "learning_rate": 2.548076923076923e-05,
+      "loss": 0.7684,
+      "step": 1733
+    },
+    {
+      "epoch": 1.011521073355695,
+      "grad_norm": 1.0295486450195312,
+      "learning_rate": 2.5465745192307693e-05,
+      "loss": 0.741,
+      "step": 1734
+    },
+    {
+      "epoch": 1.0121044188420592,
+      "grad_norm": 1.247344970703125,
+      "learning_rate": 2.5450721153846158e-05,
+      "loss": 0.7828,
+      "step": 1735
+    },
+    {
+      "epoch": 1.0126877643284236,
+      "grad_norm": 1.0752184391021729,
+      "learning_rate": 2.5435697115384617e-05,
+      "loss": 0.9003,
+      "step": 1736
+    },
+    {
+      "epoch": 1.0132711098147877,
+      "grad_norm": 1.0817466974258423,
+      "learning_rate": 2.542067307692308e-05,
+      "loss": 0.8295,
+      "step": 1737
+    },
+    {
+      "epoch": 1.0138544553011521,
+      "grad_norm": 1.0235193967819214,
+      "learning_rate": 2.5405649038461537e-05,
+      "loss": 1.0862,
+      "step": 1738
+    },
+    {
+      "epoch": 1.0144378007875163,
+      "grad_norm": 1.1558364629745483,
+      "learning_rate": 2.5390625000000002e-05,
+      "loss": 0.7409,
+      "step": 1739
+    },
+    {
+      "epoch": 1.0150211462738807,
+      "grad_norm": 1.1982396841049194,
+      "learning_rate": 2.5375600961538464e-05,
+      "loss": 0.8249,
+      "step": 1740
+    },
+    {
+      "epoch": 1.015604491760245,
+      "grad_norm": 1.074081540107727,
+      "learning_rate": 2.5360576923076923e-05,
+      "loss": 0.7878,
+      "step": 1741
+    },
+    {
+      "epoch": 1.0161878372466093,
+      "grad_norm": 1.4815016984939575,
+      "learning_rate": 2.5345552884615388e-05,
+      "loss": 0.8043,
+      "step": 1742
+    },
+    {
+      "epoch": 1.0167711827329735,
+      "grad_norm": 1.1086952686309814,
+      "learning_rate": 2.5330528846153846e-05,
+      "loss": 0.7201,
+      "step": 1743
+    },
+    {
+      "epoch": 1.017354528219338,
+      "grad_norm": 1.3218934535980225,
+      "learning_rate": 2.5315504807692308e-05,
+      "loss": 0.6374,
+      "step": 1744
+    },
+    {
+      "epoch": 1.017937873705702,
+      "grad_norm": 1.0537618398666382,
+      "learning_rate": 2.5300480769230774e-05,
+      "loss": 0.7224,
+      "step": 1745
+    },
+    {
+      "epoch": 1.0185212191920665,
+      "grad_norm": 0.9964005947113037,
+      "learning_rate": 2.528545673076923e-05,
+      "loss": 0.7204,
+      "step": 1746
+    },
+    {
+      "epoch": 1.0191045646784307,
+      "grad_norm": 1.1964815855026245,
+      "learning_rate": 2.5270432692307694e-05,
+      "loss": 0.9753,
+      "step": 1747
+    },
+    {
+      "epoch": 1.019687910164795,
+      "grad_norm": 1.0601825714111328,
+      "learning_rate": 2.525540865384616e-05,
+      "loss": 0.9208,
+      "step": 1748
+    },
+    {
+      "epoch": 1.0202712556511595,
+      "grad_norm": 0.9419945478439331,
+      "learning_rate": 2.5240384615384614e-05,
+      "loss": 0.8482,
+      "step": 1749
+    },
+    {
+      "epoch": 1.0208546011375237,
+      "grad_norm": 1.0061907768249512,
+      "learning_rate": 2.522536057692308e-05,
+      "loss": 0.8908,
+      "step": 1750
+    },
+    {
+      "epoch": 1.021437946623888,
+      "grad_norm": 1.1203185319900513,
+      "learning_rate": 2.5210336538461538e-05,
+      "loss": 0.725,
+      "step": 1751
+    },
+    {
+      "epoch": 1.0220212921102523,
+      "grad_norm": 1.2346861362457275,
+      "learning_rate": 2.51953125e-05,
+      "loss": 0.8266,
+      "step": 1752
+    },
+    {
+      "epoch": 1.0226046375966167,
+      "grad_norm": 1.17784583568573,
+      "learning_rate": 2.5180288461538465e-05,
+      "loss": 0.8013,
+      "step": 1753
+    },
+    {
+      "epoch": 1.0231879830829809,
+      "grad_norm": 1.0398868322372437,
+      "learning_rate": 2.5165264423076924e-05,
+      "loss": 0.5803,
+      "step": 1754
+    },
+    {
+      "epoch": 1.0237713285693453,
+      "grad_norm": 1.372266173362732,
+      "learning_rate": 2.5150240384615386e-05,
+      "loss": 0.6542,
+      "step": 1755
+    },
+    {
+      "epoch": 1.0243546740557095,
+      "grad_norm": 1.2637945413589478,
+      "learning_rate": 2.5135216346153844e-05,
+      "loss": 1.0724,
+      "step": 1756
+    },
+    {
+      "epoch": 1.0249380195420739,
+      "grad_norm": 1.0931401252746582,
+      "learning_rate": 2.512019230769231e-05,
+      "loss": 0.734,
+      "step": 1757
+    },
+    {
+      "epoch": 1.025521365028438,
+      "grad_norm": 1.0705105066299438,
+      "learning_rate": 2.510516826923077e-05,
+      "loss": 0.8975,
+      "step": 1758
+    },
+    {
+      "epoch": 1.0261047105148025,
+      "grad_norm": 1.0411278009414673,
+      "learning_rate": 2.509014423076923e-05,
+      "loss": 0.829,
+      "step": 1759
+    },
+    {
+      "epoch": 1.0266880560011666,
+      "grad_norm": 1.1690353155136108,
+      "learning_rate": 2.5075120192307695e-05,
+      "loss": 0.8418,
+      "step": 1760
+    },
+    {
+      "epoch": 1.027271401487531,
+      "grad_norm": 1.1422929763793945,
+      "learning_rate": 2.5060096153846157e-05,
+      "loss": 1.003,
+      "step": 1761
+    },
+    {
+      "epoch": 1.0278547469738952,
+      "grad_norm": 1.2721543312072754,
+      "learning_rate": 2.5045072115384616e-05,
+      "loss": 0.8079,
+      "step": 1762
+    },
+    {
+      "epoch": 1.0284380924602596,
+      "grad_norm": 1.3369412422180176,
+      "learning_rate": 2.503004807692308e-05,
+      "loss": 0.9004,
+      "step": 1763
+    },
+    {
+      "epoch": 1.0290214379466238,
+      "grad_norm": 1.6825288534164429,
+      "learning_rate": 2.5015024038461536e-05,
+      "loss": 0.9225,
+      "step": 1764
+    },
+    {
+      "epoch": 1.0296047834329882,
+      "grad_norm": 1.25969660282135,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7035,
+      "step": 1765
+    },
+    {
+      "epoch": 1.0301881289193524,
+      "grad_norm": 1.099972128868103,
+      "learning_rate": 2.4984975961538463e-05,
+      "loss": 0.7718,
+      "step": 1766
+    },
+    {
+      "epoch": 1.0307714744057168,
+      "grad_norm": 1.1386865377426147,
+      "learning_rate": 2.496995192307692e-05,
+      "loss": 0.7239,
+      "step": 1767
+    },
+    {
+      "epoch": 1.031354819892081,
+      "grad_norm": 1.2388840913772583,
+      "learning_rate": 2.4954927884615387e-05,
+      "loss": 0.7997,
+      "step": 1768
+    },
+    {
+      "epoch": 1.0319381653784454,
+      "grad_norm": 1.159300446510315,
+      "learning_rate": 2.493990384615385e-05,
+      "loss": 1.0244,
+      "step": 1769
+    },
+    {
+      "epoch": 1.0325215108648096,
+      "grad_norm": 1.186693549156189,
+      "learning_rate": 2.4924879807692307e-05,
+      "loss": 0.7362,
+      "step": 1770
+    },
+    {
+      "epoch": 1.033104856351174,
+      "grad_norm": 1.0532599687576294,
+      "learning_rate": 2.490985576923077e-05,
+      "loss": 0.6688,
+      "step": 1771
+    },
+    {
+      "epoch": 1.0336882018375382,
+      "grad_norm": 1.267077088356018,
+      "learning_rate": 2.4894831730769234e-05,
+      "loss": 0.7482,
+      "step": 1772
+    },
+    {
+      "epoch": 1.0342715473239026,
+      "grad_norm": 1.1788893938064575,
+      "learning_rate": 2.4879807692307693e-05,
+      "loss": 0.6433,
+      "step": 1773
+    },
+    {
+      "epoch": 1.0348548928102668,
+      "grad_norm": 1.3695346117019653,
+      "learning_rate": 2.4864783653846155e-05,
+      "loss": 1.0139,
+      "step": 1774
+    },
+    {
+      "epoch": 1.0354382382966312,
+      "grad_norm": 1.364996075630188,
+      "learning_rate": 2.4849759615384617e-05,
+      "loss": 0.9402,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0360215837829956,
+      "grad_norm": 1.1488473415374756,
+      "learning_rate": 2.483473557692308e-05,
+      "loss": 0.6775,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0366049292693598,
+      "grad_norm": 1.1622604131698608,
+      "learning_rate": 2.481971153846154e-05,
+      "loss": 0.9029,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0371882747557242,
+      "grad_norm": 1.3263182640075684,
+      "learning_rate": 2.4804687500000002e-05,
+      "loss": 0.7835,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0377716202420884,
+      "grad_norm": 1.031337022781372,
+      "learning_rate": 2.478966346153846e-05,
+      "loss": 0.7646,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0383549657284528,
+      "grad_norm": 1.2709075212478638,
+      "learning_rate": 2.4774639423076923e-05,
+      "loss": 0.7746,
+      "step": 1780
+    },
+    {
+      "epoch": 1.038938311214817,
+      "grad_norm": 1.0792698860168457,
+      "learning_rate": 2.4759615384615388e-05,
+      "loss": 0.4939,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0395216567011814,
+      "grad_norm": 1.0642223358154297,
+      "learning_rate": 2.4744591346153847e-05,
+      "loss": 0.7588,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0401050021875455,
+      "grad_norm": 1.2650682926177979,
+      "learning_rate": 2.472956730769231e-05,
+      "loss": 0.866,
+      "step": 1783
+    },
+    {
+      "epoch": 1.04068834767391,
+      "grad_norm": 1.2211084365844727,
+      "learning_rate": 2.471454326923077e-05,
+      "loss": 0.8201,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0412716931602741,
+      "grad_norm": 1.2228623628616333,
+      "learning_rate": 2.4699519230769232e-05,
+      "loss": 0.625,
+      "step": 1785
+    },
+    {
+      "epoch": 1.0418550386466385,
+      "grad_norm": 1.2318840026855469,
+      "learning_rate": 2.4684495192307694e-05,
+      "loss": 0.796,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0424383841330027,
+      "grad_norm": 1.1144096851348877,
+      "learning_rate": 2.4669471153846156e-05,
+      "loss": 0.9259,
+      "step": 1787
+    },
+    {
+      "epoch": 1.0430217296193671,
+      "grad_norm": 1.1762030124664307,
+      "learning_rate": 2.4654447115384615e-05,
+      "loss": 0.8342,
+      "step": 1788
+    },
+    {
+      "epoch": 1.0436050751057313,
+      "grad_norm": 1.142924427986145,
+      "learning_rate": 2.463942307692308e-05,
+      "loss": 0.8825,
+      "step": 1789
+    },
+    {
+      "epoch": 1.0441884205920957,
+      "grad_norm": 1.3172521591186523,
+      "learning_rate": 2.462439903846154e-05,
+      "loss": 0.8078,
+      "step": 1790
+    },
+    {
+      "epoch": 1.04477176607846,
+      "grad_norm": 1.0718015432357788,
+      "learning_rate": 2.4609375e-05,
+      "loss": 0.8854,
+      "step": 1791
+    },
+    {
+      "epoch": 1.0453551115648243,
+      "grad_norm": 1.0188226699829102,
+      "learning_rate": 2.4594350961538462e-05,
+      "loss": 1.0314,
+      "step": 1792
+    },
+    {
+      "epoch": 1.0459384570511885,
+      "grad_norm": 1.0908758640289307,
+      "learning_rate": 2.4579326923076924e-05,
+      "loss": 0.6848,
+      "step": 1793
+    },
+    {
+      "epoch": 1.046521802537553,
+      "grad_norm": 1.072546362876892,
+      "learning_rate": 2.4564302884615386e-05,
+      "loss": 0.657,
+      "step": 1794
+    },
+    {
+      "epoch": 1.047105148023917,
+      "grad_norm": 1.0566478967666626,
+      "learning_rate": 2.4549278846153848e-05,
+      "loss": 0.9628,
+      "step": 1795
+    },
+    {
+      "epoch": 1.0476884935102815,
+      "grad_norm": 1.16886305809021,
+      "learning_rate": 2.453425480769231e-05,
+      "loss": 0.7434,
+      "step": 1796
+    },
+    {
+      "epoch": 1.0482718389966457,
+      "grad_norm": 1.1381522417068481,
+      "learning_rate": 2.4519230769230768e-05,
+      "loss": 0.559,
+      "step": 1797
+    },
+    {
+      "epoch": 1.04885518448301,
+      "grad_norm": 1.2783828973770142,
+      "learning_rate": 2.4504206730769233e-05,
+      "loss": 0.9,
+      "step": 1798
+    },
+    {
+      "epoch": 1.0494385299693743,
+      "grad_norm": 1.0993632078170776,
+      "learning_rate": 2.4489182692307695e-05,
+      "loss": 0.7602,
+      "step": 1799
+    },
+    {
+      "epoch": 1.0500218754557387,
+      "grad_norm": 0.9995693564414978,
+      "learning_rate": 2.4474158653846154e-05,
+      "loss": 0.8363,
+      "step": 1800
+    },
+    {
+      "epoch": 1.0500218754557387,
+      "eval_loss_squad": 0.8367310870531947,
+      "eval_perplexity": 8.296506422781007,
+      "eval_perplexity_reconstruct": 1.9039826490870362,
+      "step": 1800
+    },
+    {
+      "epoch": 1.0506052209421028,
+      "grad_norm": 1.3584226369857788,
+      "learning_rate": 2.4459134615384616e-05,
+      "loss": 0.7513,
+      "step": 1801
+    },
+    {
+      "epoch": 1.0511885664284673,
+      "grad_norm": 1.1026893854141235,
+      "learning_rate": 2.444411057692308e-05,
+      "loss": 0.8734,
+      "step": 1802
+    },
+    {
+      "epoch": 1.0517719119148317,
+      "grad_norm": 0.9750449061393738,
+      "learning_rate": 2.442908653846154e-05,
+      "loss": 0.7949,
+      "step": 1803
+    },
+    {
+      "epoch": 1.0523552574011958,
+      "grad_norm": 1.3138856887817383,
+      "learning_rate": 2.44140625e-05,
+      "loss": 0.7326,
+      "step": 1804
+    },
+    {
+      "epoch": 1.0529386028875602,
+      "grad_norm": 0.9921886920928955,
+      "learning_rate": 2.4399038461538463e-05,
+      "loss": 0.8876,
+      "step": 1805
+    },
+    {
+      "epoch": 1.0535219483739244,
+      "grad_norm": 1.2363988161087036,
+      "learning_rate": 2.4384014423076922e-05,
+      "loss": 0.8668,
+      "step": 1806
+    },
+    {
+      "epoch": 1.0541052938602888,
+      "grad_norm": 1.2714601755142212,
+      "learning_rate": 2.4368990384615387e-05,
+      "loss": 0.8749,
+      "step": 1807
+    },
+    {
+      "epoch": 1.054688639346653,
+      "grad_norm": 1.1158849000930786,
+      "learning_rate": 2.435396634615385e-05,
+      "loss": 1.065,
+      "step": 1808
+    },
+    {
+      "epoch": 1.0552719848330174,
+      "grad_norm": 1.125258207321167,
+      "learning_rate": 2.4338942307692307e-05,
+      "loss": 0.8081,
+      "step": 1809
+    },
+    {
+      "epoch": 1.0558553303193816,
+      "grad_norm": 1.039055585861206,
+      "learning_rate": 2.432391826923077e-05,
+      "loss": 0.8932,
+      "step": 1810
+    },
+    {
+      "epoch": 1.056438675805746,
+      "grad_norm": 1.0936514139175415,
+      "learning_rate": 2.4308894230769235e-05,
+      "loss": 0.9564,
+      "step": 1811
+    },
+    {
+      "epoch": 1.0570220212921102,
+      "grad_norm": 1.334757924079895,
+      "learning_rate": 2.4293870192307693e-05,
+      "loss": 0.8967,
+      "step": 1812
+    },
+    {
+      "epoch": 1.0576053667784746,
+      "grad_norm": 1.4477955102920532,
+      "learning_rate": 2.4278846153846155e-05,
+      "loss": 0.8336,
+      "step": 1813
+    },
+    {
+      "epoch": 1.0581887122648388,
+      "grad_norm": 1.319165825843811,
+      "learning_rate": 2.4263822115384617e-05,
+      "loss": 0.7005,
+      "step": 1814
+    },
+    {
+      "epoch": 1.0587720577512032,
+      "grad_norm": 1.1051411628723145,
+      "learning_rate": 2.424879807692308e-05,
+      "loss": 0.7516,
+      "step": 1815
+    },
+    {
+      "epoch": 1.0593554032375674,
+      "grad_norm": 1.2956829071044922,
+      "learning_rate": 2.423377403846154e-05,
+      "loss": 0.7716,
+      "step": 1816
+    },
+    {
+      "epoch": 1.0599387487239318,
+      "grad_norm": 1.199265956878662,
+      "learning_rate": 2.4218750000000003e-05,
+      "loss": 0.9904,
+      "step": 1817
+    },
+    {
+      "epoch": 1.060522094210296,
+      "grad_norm": 1.0964492559432983,
+      "learning_rate": 2.420372596153846e-05,
+      "loss": 0.7208,
+      "step": 1818
+    },
+    {
+      "epoch": 1.0611054396966604,
+      "grad_norm": 1.264791488647461,
+      "learning_rate": 2.4188701923076923e-05,
+      "loss": 0.7618,
+      "step": 1819
+    },
+    {
+      "epoch": 1.0616887851830246,
+      "grad_norm": 1.1390454769134521,
+      "learning_rate": 2.4173677884615388e-05,
+      "loss": 0.7634,
+      "step": 1820
+    },
+    {
+      "epoch": 1.062272130669389,
+      "grad_norm": 1.1044617891311646,
+      "learning_rate": 2.4158653846153847e-05,
+      "loss": 0.6295,
+      "step": 1821
+    },
+    {
+      "epoch": 1.0628554761557532,
+      "grad_norm": 1.4096314907073975,
+      "learning_rate": 2.414362980769231e-05,
+      "loss": 0.8327,
+      "step": 1822
+    },
+    {
+      "epoch": 1.0634388216421176,
+      "grad_norm": 1.2174628973007202,
+      "learning_rate": 2.412860576923077e-05,
+      "loss": 0.985,
+      "step": 1823
+    },
+    {
+      "epoch": 1.0640221671284817,
+      "grad_norm": 0.912973165512085,
+      "learning_rate": 2.4113581730769232e-05,
+      "loss": 1.0427,
+      "step": 1824
+    },
+    {
+      "epoch": 1.0646055126148462,
+      "grad_norm": 1.1908468008041382,
+      "learning_rate": 2.4098557692307694e-05,
+      "loss": 0.7616,
+      "step": 1825
+    },
+    {
+      "epoch": 1.0651888581012106,
+      "grad_norm": 1.260453462600708,
+      "learning_rate": 2.4083533653846156e-05,
+      "loss": 0.8113,
+      "step": 1826
+    },
+    {
+      "epoch": 1.0657722035875747,
+      "grad_norm": 1.2202224731445312,
+      "learning_rate": 2.4068509615384615e-05,
+      "loss": 0.7171,
+      "step": 1827
+    },
+    {
+      "epoch": 1.066355549073939,
+      "grad_norm": 0.8979166746139526,
+      "learning_rate": 2.405348557692308e-05,
+      "loss": 0.7942,
+      "step": 1828
+    },
+    {
+      "epoch": 1.0669388945603033,
+      "grad_norm": 1.1224358081817627,
+      "learning_rate": 2.4038461538461542e-05,
+      "loss": 0.9005,
+      "step": 1829
+    },
+    {
+      "epoch": 1.0675222400466677,
+      "grad_norm": 1.1409502029418945,
+      "learning_rate": 2.40234375e-05,
+      "loss": 0.7607,
+      "step": 1830
+    },
+    {
+      "epoch": 1.068105585533032,
+      "grad_norm": 1.1402530670166016,
+      "learning_rate": 2.4008413461538462e-05,
+      "loss": 1.0247,
+      "step": 1831
+    },
+    {
+      "epoch": 1.0686889310193963,
+      "grad_norm": 1.0991151332855225,
+      "learning_rate": 2.3993389423076924e-05,
+      "loss": 0.7754,
+      "step": 1832
+    },
+    {
+      "epoch": 1.0692722765057605,
+      "grad_norm": 1.229304552078247,
+      "learning_rate": 2.3978365384615386e-05,
+      "loss": 0.8271,
+      "step": 1833
+    },
+    {
+      "epoch": 1.069855621992125,
+      "grad_norm": 1.2119697332382202,
+      "learning_rate": 2.3963341346153848e-05,
+      "loss": 0.8626,
+      "step": 1834
+    },
+    {
+      "epoch": 1.070438967478489,
+      "grad_norm": 1.4418648481369019,
+      "learning_rate": 2.394831730769231e-05,
+      "loss": 0.8454,
+      "step": 1835
+    },
+    {
+      "epoch": 1.0710223129648535,
+      "grad_norm": 1.1836721897125244,
+      "learning_rate": 2.3933293269230768e-05,
+      "loss": 0.8298,
+      "step": 1836
+    },
+    {
+      "epoch": 1.0716056584512177,
+      "grad_norm": 1.2164535522460938,
+      "learning_rate": 2.3918269230769234e-05,
+      "loss": 0.8123,
+      "step": 1837
+    },
+    {
+      "epoch": 1.072189003937582,
+      "grad_norm": 1.3282859325408936,
+      "learning_rate": 2.3903245192307695e-05,
+      "loss": 1.004,
+      "step": 1838
+    },
+    {
+      "epoch": 1.0727723494239463,
+      "grad_norm": 1.1000369787216187,
+      "learning_rate": 2.3888221153846154e-05,
+      "loss": 0.7878,
+      "step": 1839
+    },
+    {
+      "epoch": 1.0733556949103107,
+      "grad_norm": 1.3156583309173584,
+      "learning_rate": 2.3873197115384616e-05,
+      "loss": 0.6691,
+      "step": 1840
+    },
+    {
+      "epoch": 1.0739390403966749,
+      "grad_norm": 1.264693021774292,
+      "learning_rate": 2.3858173076923078e-05,
+      "loss": 0.8582,
+      "step": 1841
+    },
+    {
+      "epoch": 1.0745223858830393,
+      "grad_norm": 1.1336017847061157,
+      "learning_rate": 2.384314903846154e-05,
+      "loss": 0.704,
+      "step": 1842
+    },
+    {
+      "epoch": 1.0751057313694035,
+      "grad_norm": 0.8796018362045288,
+      "learning_rate": 2.3828125e-05,
+      "loss": 0.7031,
+      "step": 1843
+    },
+    {
+      "epoch": 1.0756890768557679,
+      "grad_norm": 1.2418160438537598,
+      "learning_rate": 2.3813100961538463e-05,
+      "loss": 0.7327,
+      "step": 1844
+    },
+    {
+      "epoch": 1.076272422342132,
+      "grad_norm": 1.1050810813903809,
+      "learning_rate": 2.3798076923076922e-05,
+      "loss": 0.8006,
+      "step": 1845
+    },
+    {
+      "epoch": 1.0768557678284965,
+      "grad_norm": 1.2827439308166504,
+      "learning_rate": 2.3783052884615387e-05,
+      "loss": 0.777,
+      "step": 1846
+    },
+    {
+      "epoch": 1.0774391133148606,
+      "grad_norm": 1.3954315185546875,
+      "learning_rate": 2.376802884615385e-05,
+      "loss": 0.7289,
+      "step": 1847
+    },
+    {
+      "epoch": 1.078022458801225,
+      "grad_norm": 1.2399247884750366,
+      "learning_rate": 2.3753004807692308e-05,
+      "loss": 0.7638,
+      "step": 1848
+    },
+    {
+      "epoch": 1.0786058042875892,
+      "grad_norm": 1.351985216140747,
+      "learning_rate": 2.373798076923077e-05,
+      "loss": 0.7568,
+      "step": 1849
+    },
+    {
+      "epoch": 1.0791891497739536,
+      "grad_norm": 1.1881484985351562,
+      "learning_rate": 2.372295673076923e-05,
+      "loss": 0.8417,
+      "step": 1850
+    },
+    {
+      "epoch": 1.0797724952603178,
+      "grad_norm": 1.6460144519805908,
+      "learning_rate": 2.3707932692307693e-05,
+      "loss": 0.8467,
+      "step": 1851
+    },
+    {
+      "epoch": 1.0803558407466822,
+      "grad_norm": 1.1815186738967896,
+      "learning_rate": 2.3692908653846155e-05,
+      "loss": 0.7675,
+      "step": 1852
+    },
+    {
+      "epoch": 1.0809391862330466,
+      "grad_norm": 1.2811179161071777,
+      "learning_rate": 2.3677884615384617e-05,
+      "loss": 0.7745,
+      "step": 1853
+    },
+    {
+      "epoch": 1.0815225317194108,
+      "grad_norm": 1.1434037685394287,
+      "learning_rate": 2.366286057692308e-05,
+      "loss": 0.8752,
+      "step": 1854
+    },
+    {
+      "epoch": 1.0821058772057752,
+      "grad_norm": 1.2398650646209717,
+      "learning_rate": 2.364783653846154e-05,
+      "loss": 0.8398,
+      "step": 1855
+    },
+    {
+      "epoch": 1.0826892226921394,
+      "grad_norm": 1.2694295644760132,
+      "learning_rate": 2.3632812500000003e-05,
+      "loss": 0.8714,
+      "step": 1856
+    },
+    {
+      "epoch": 1.0832725681785038,
+      "grad_norm": 1.1594481468200684,
+      "learning_rate": 2.361778846153846e-05,
+      "loss": 0.6383,
+      "step": 1857
+    },
+    {
+      "epoch": 1.083855913664868,
+      "grad_norm": 1.2221319675445557,
+      "learning_rate": 2.3602764423076923e-05,
+      "loss": 0.9375,
+      "step": 1858
+    },
+    {
+      "epoch": 1.0844392591512324,
+      "grad_norm": 1.183512568473816,
+      "learning_rate": 2.3587740384615385e-05,
+      "loss": 0.8724,
+      "step": 1859
+    },
+    {
+      "epoch": 1.0850226046375966,
+      "grad_norm": 1.3200178146362305,
+      "learning_rate": 2.3572716346153847e-05,
+      "loss": 0.8969,
+      "step": 1860
+    },
+    {
+      "epoch": 1.085605950123961,
+      "grad_norm": 1.3450496196746826,
+      "learning_rate": 2.355769230769231e-05,
+      "loss": 0.822,
+      "step": 1861
+    },
+    {
+      "epoch": 1.0861892956103252,
+      "grad_norm": 1.2601107358932495,
+      "learning_rate": 2.354266826923077e-05,
+      "loss": 0.7901,
+      "step": 1862
+    },
+    {
+      "epoch": 1.0867726410966896,
+      "grad_norm": 1.089453101158142,
+      "learning_rate": 2.3527644230769233e-05,
+      "loss": 0.8132,
+      "step": 1863
+    },
+    {
+      "epoch": 1.0873559865830538,
+      "grad_norm": 0.8895475268363953,
+      "learning_rate": 2.3512620192307694e-05,
+      "loss": 0.6964,
+      "step": 1864
+    },
+    {
+      "epoch": 1.0879393320694182,
+      "grad_norm": 1.3046108484268188,
+      "learning_rate": 2.3497596153846156e-05,
+      "loss": 0.9721,
+      "step": 1865
+    },
+    {
+      "epoch": 1.0885226775557824,
+      "grad_norm": 1.102406620979309,
+      "learning_rate": 2.3482572115384615e-05,
+      "loss": 0.8803,
+      "step": 1866
+    },
+    {
+      "epoch": 1.0891060230421468,
+      "grad_norm": 1.1453616619110107,
+      "learning_rate": 2.346754807692308e-05,
+      "loss": 0.8508,
+      "step": 1867
+    },
+    {
+      "epoch": 1.089689368528511,
+      "grad_norm": 1.1119948625564575,
+      "learning_rate": 2.345252403846154e-05,
+      "loss": 0.6988,
+      "step": 1868
+    },
+    {
+      "epoch": 1.0902727140148754,
+      "grad_norm": 1.3853925466537476,
+      "learning_rate": 2.34375e-05,
+      "loss": 0.8331,
+      "step": 1869
+    },
+    {
+      "epoch": 1.0908560595012395,
+      "grad_norm": 1.315190315246582,
+      "learning_rate": 2.3422475961538462e-05,
+      "loss": 0.7391,
+      "step": 1870
+    },
+    {
+      "epoch": 1.091439404987604,
+      "grad_norm": 1.070307970046997,
+      "learning_rate": 2.3407451923076924e-05,
+      "loss": 0.8388,
+      "step": 1871
+    },
+    {
+      "epoch": 1.0920227504739681,
+      "grad_norm": 0.9866098165512085,
+      "learning_rate": 2.3392427884615386e-05,
+      "loss": 1.0789,
+      "step": 1872
+    },
+    {
+      "epoch": 1.0926060959603325,
+      "grad_norm": 1.2839645147323608,
+      "learning_rate": 2.3377403846153848e-05,
+      "loss": 0.7442,
+      "step": 1873
+    },
+    {
+      "epoch": 1.0931894414466967,
+      "grad_norm": 1.2288687229156494,
+      "learning_rate": 2.336237980769231e-05,
+      "loss": 0.7923,
+      "step": 1874
+    },
+    {
+      "epoch": 1.0937727869330611,
+      "grad_norm": 1.0700854063034058,
+      "learning_rate": 2.334735576923077e-05,
+      "loss": 0.9763,
+      "step": 1875
+    },
+    {
+      "epoch": 1.0943561324194253,
+      "grad_norm": 0.9807063937187195,
+      "learning_rate": 2.3332331730769234e-05,
+      "loss": 0.8784,
+      "step": 1876
+    },
+    {
+      "epoch": 1.0949394779057897,
+      "grad_norm": 1.1770750284194946,
+      "learning_rate": 2.3317307692307692e-05,
+      "loss": 0.7672,
+      "step": 1877
+    },
+    {
+      "epoch": 1.095522823392154,
+      "grad_norm": 1.1473841667175293,
+      "learning_rate": 2.3302283653846154e-05,
+      "loss": 0.7583,
+      "step": 1878
+    },
+    {
+      "epoch": 1.0961061688785183,
+      "grad_norm": 1.2041727304458618,
+      "learning_rate": 2.3287259615384616e-05,
+      "loss": 0.8327,
+      "step": 1879
+    },
+    {
+      "epoch": 1.0966895143648827,
+      "grad_norm": 1.1696784496307373,
+      "learning_rate": 2.3272235576923078e-05,
+      "loss": 0.853,
+      "step": 1880
+    },
+    {
+      "epoch": 1.097272859851247,
+      "grad_norm": 1.1404637098312378,
+      "learning_rate": 2.325721153846154e-05,
+      "loss": 0.8551,
+      "step": 1881
+    },
+    {
+      "epoch": 1.0978562053376113,
+      "grad_norm": 1.082005262374878,
+      "learning_rate": 2.32421875e-05,
+      "loss": 0.7462,
+      "step": 1882
+    },
+    {
+      "epoch": 1.0984395508239755,
+      "grad_norm": 1.3037984371185303,
+      "learning_rate": 2.3227163461538464e-05,
+      "loss": 1.0494,
+      "step": 1883
+    },
+    {
+      "epoch": 1.09902289631034,
+      "grad_norm": 1.3747504949569702,
+      "learning_rate": 2.3212139423076922e-05,
+      "loss": 0.9479,
+      "step": 1884
+    },
+    {
+      "epoch": 1.099606241796704,
+      "grad_norm": 1.2729436159133911,
+      "learning_rate": 2.3197115384615387e-05,
+      "loss": 0.8058,
+      "step": 1885
+    },
+    {
+      "epoch": 1.1001895872830685,
+      "grad_norm": 0.951274037361145,
+      "learning_rate": 2.3182091346153846e-05,
+      "loss": 0.7261,
+      "step": 1886
+    },
+    {
+      "epoch": 1.1007729327694327,
+      "grad_norm": 1.114943027496338,
+      "learning_rate": 2.3167067307692308e-05,
+      "loss": 0.8833,
+      "step": 1887
+    },
+    {
+      "epoch": 1.101356278255797,
+      "grad_norm": 1.2294602394104004,
+      "learning_rate": 2.315204326923077e-05,
+      "loss": 0.8286,
+      "step": 1888
+    },
+    {
+      "epoch": 1.1019396237421613,
+      "grad_norm": 1.0512102842330933,
+      "learning_rate": 2.313701923076923e-05,
+      "loss": 0.8782,
+      "step": 1889
+    },
+    {
+      "epoch": 1.1025229692285257,
+      "grad_norm": 1.2631170749664307,
+      "learning_rate": 2.3121995192307693e-05,
+      "loss": 0.7911,
+      "step": 1890
+    },
+    {
+      "epoch": 1.1031063147148898,
+      "grad_norm": 1.3144844770431519,
+      "learning_rate": 2.3106971153846155e-05,
+      "loss": 0.7812,
+      "step": 1891
+    },
+    {
+      "epoch": 1.1036896602012543,
+      "grad_norm": 1.0786577463150024,
+      "learning_rate": 2.3091947115384617e-05,
+      "loss": 0.7804,
+      "step": 1892
+    },
+    {
+      "epoch": 1.1042730056876184,
+      "grad_norm": 1.3957914113998413,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.9104,
+      "step": 1893
+    },
+    {
+      "epoch": 1.1048563511739828,
+      "grad_norm": 1.103511929512024,
+      "learning_rate": 2.306189903846154e-05,
+      "loss": 0.6835,
+      "step": 1894
+    },
+    {
+      "epoch": 1.105439696660347,
+      "grad_norm": 1.3668882846832275,
+      "learning_rate": 2.3046875e-05,
+      "loss": 0.8694,
+      "step": 1895
+    },
+    {
+      "epoch": 1.1060230421467114,
+      "grad_norm": 1.075162649154663,
+      "learning_rate": 2.303185096153846e-05,
+      "loss": 0.8763,
+      "step": 1896
+    },
+    {
+      "epoch": 1.1066063876330756,
+      "grad_norm": 1.2901614904403687,
+      "learning_rate": 2.3016826923076923e-05,
+      "loss": 0.8694,
+      "step": 1897
+    },
+    {
+      "epoch": 1.10718973311944,
+      "grad_norm": 1.0675395727157593,
+      "learning_rate": 2.3001802884615385e-05,
+      "loss": 0.7391,
+      "step": 1898
+    },
+    {
+      "epoch": 1.1077730786058042,
+      "grad_norm": 1.200480341911316,
+      "learning_rate": 2.2986778846153847e-05,
+      "loss": 0.868,
+      "step": 1899
+    },
+    {
+      "epoch": 1.1083564240921686,
+      "grad_norm": 1.254665732383728,
+      "learning_rate": 2.297175480769231e-05,
+      "loss": 0.7587,
+      "step": 1900
+    },
+    {
+      "epoch": 1.1089397695785328,
+      "grad_norm": 1.047951102256775,
+      "learning_rate": 2.295673076923077e-05,
+      "loss": 0.8484,
+      "step": 1901
+    },
+    {
+      "epoch": 1.1095231150648972,
+      "grad_norm": 1.0804102420806885,
+      "learning_rate": 2.2941706730769233e-05,
+      "loss": 0.7814,
+      "step": 1902
+    },
+    {
+      "epoch": 1.1101064605512614,
+      "grad_norm": 1.1867598295211792,
+      "learning_rate": 2.2926682692307695e-05,
+      "loss": 0.75,
+      "step": 1903
+    },
+    {
+      "epoch": 1.1106898060376258,
+      "grad_norm": 1.148829460144043,
+      "learning_rate": 2.2911658653846153e-05,
+      "loss": 0.8877,
+      "step": 1904
+    },
+    {
+      "epoch": 1.11127315152399,
+      "grad_norm": 1.0924962759017944,
+      "learning_rate": 2.2896634615384615e-05,
+      "loss": 0.7225,
+      "step": 1905
+    },
+    {
+      "epoch": 1.1118564970103544,
+      "grad_norm": 1.1306710243225098,
+      "learning_rate": 2.288161057692308e-05,
+      "loss": 0.8202,
+      "step": 1906
+    },
+    {
+      "epoch": 1.1124398424967188,
+      "grad_norm": 1.1604644060134888,
+      "learning_rate": 2.286658653846154e-05,
+      "loss": 0.8849,
+      "step": 1907
+    },
+    {
+      "epoch": 1.113023187983083,
+      "grad_norm": 1.2594935894012451,
+      "learning_rate": 2.28515625e-05,
+      "loss": 0.8069,
+      "step": 1908
+    },
+    {
+      "epoch": 1.1136065334694474,
+      "grad_norm": 1.1446539163589478,
+      "learning_rate": 2.2836538461538463e-05,
+      "loss": 0.6342,
+      "step": 1909
+    },
+    {
+      "epoch": 1.1141898789558116,
+      "grad_norm": 1.317840337753296,
+      "learning_rate": 2.2821514423076924e-05,
+      "loss": 0.8211,
+      "step": 1910
+    },
+    {
+      "epoch": 1.114773224442176,
+      "grad_norm": 1.0653162002563477,
+      "learning_rate": 2.2806490384615386e-05,
+      "loss": 0.8454,
+      "step": 1911
+    },
+    {
+      "epoch": 1.1153565699285402,
+      "grad_norm": 0.9791072607040405,
+      "learning_rate": 2.2791466346153848e-05,
+      "loss": 0.7433,
+      "step": 1912
+    },
+    {
+      "epoch": 1.1159399154149046,
+      "grad_norm": 1.1724233627319336,
+      "learning_rate": 2.2776442307692307e-05,
+      "loss": 0.7882,
+      "step": 1913
+    },
+    {
+      "epoch": 1.1165232609012687,
+      "grad_norm": 1.2540687322616577,
+      "learning_rate": 2.276141826923077e-05,
+      "loss": 0.7061,
+      "step": 1914
+    },
+    {
+      "epoch": 1.1171066063876331,
+      "grad_norm": 1.0810496807098389,
+      "learning_rate": 2.2746394230769234e-05,
+      "loss": 0.973,
+      "step": 1915
+    },
+    {
+      "epoch": 1.1176899518739973,
+      "grad_norm": 1.2553794384002686,
+      "learning_rate": 2.2731370192307692e-05,
+      "loss": 0.6909,
+      "step": 1916
+    },
+    {
+      "epoch": 1.1182732973603617,
+      "grad_norm": 1.2660142183303833,
+      "learning_rate": 2.2716346153846154e-05,
+      "loss": 0.8742,
+      "step": 1917
+    },
+    {
+      "epoch": 1.118856642846726,
+      "grad_norm": 1.4024698734283447,
+      "learning_rate": 2.2701322115384616e-05,
+      "loss": 0.7558,
+      "step": 1918
+    },
+    {
+      "epoch": 1.1194399883330903,
+      "grad_norm": 1.2419300079345703,
+      "learning_rate": 2.2686298076923078e-05,
+      "loss": 0.8231,
+      "step": 1919
+    },
+    {
+      "epoch": 1.1200233338194545,
+      "grad_norm": 1.2142603397369385,
+      "learning_rate": 2.267127403846154e-05,
+      "loss": 0.8333,
+      "step": 1920
+    },
+    {
+      "epoch": 1.120606679305819,
+      "grad_norm": 1.1328840255737305,
+      "learning_rate": 2.2656250000000002e-05,
+      "loss": 0.7284,
+      "step": 1921
+    },
+    {
+      "epoch": 1.121190024792183,
+      "grad_norm": 1.256993055343628,
+      "learning_rate": 2.264122596153846e-05,
+      "loss": 0.8874,
+      "step": 1922
+    },
+    {
+      "epoch": 1.1217733702785475,
+      "grad_norm": 0.9864529967308044,
+      "learning_rate": 2.2626201923076922e-05,
+      "loss": 0.7252,
+      "step": 1923
+    },
+    {
+      "epoch": 1.1223567157649117,
+      "grad_norm": 1.1391963958740234,
+      "learning_rate": 2.2611177884615387e-05,
+      "loss": 0.7308,
+      "step": 1924
+    },
+    {
+      "epoch": 1.122940061251276,
+      "grad_norm": 1.3632563352584839,
+      "learning_rate": 2.2596153846153846e-05,
+      "loss": 0.7668,
+      "step": 1925
+    },
+    {
+      "epoch": 1.1235234067376403,
+      "grad_norm": 0.9523732662200928,
+      "learning_rate": 2.2581129807692308e-05,
+      "loss": 0.6653,
+      "step": 1926
+    },
+    {
+      "epoch": 1.1241067522240047,
+      "grad_norm": 1.2235891819000244,
+      "learning_rate": 2.256610576923077e-05,
+      "loss": 0.7031,
+      "step": 1927
+    },
+    {
+      "epoch": 1.1246900977103689,
+      "grad_norm": 1.1225990056991577,
+      "learning_rate": 2.255108173076923e-05,
+      "loss": 0.7132,
+      "step": 1928
+    },
+    {
+      "epoch": 1.1252734431967333,
+      "grad_norm": 1.2185968160629272,
+      "learning_rate": 2.2536057692307694e-05,
+      "loss": 0.7747,
+      "step": 1929
+    },
+    {
+      "epoch": 1.1258567886830977,
+      "grad_norm": 1.2075011730194092,
+      "learning_rate": 2.2521033653846155e-05,
+      "loss": 0.7532,
+      "step": 1930
+    },
+    {
+      "epoch": 1.1264401341694619,
+      "grad_norm": 1.0588568449020386,
+      "learning_rate": 2.2506009615384614e-05,
+      "loss": 0.6433,
+      "step": 1931
+    },
+    {
+      "epoch": 1.127023479655826,
+      "grad_norm": 1.154813528060913,
+      "learning_rate": 2.249098557692308e-05,
+      "loss": 0.8853,
+      "step": 1932
+    },
+    {
+      "epoch": 1.1276068251421905,
+      "grad_norm": 1.280246376991272,
+      "learning_rate": 2.247596153846154e-05,
+      "loss": 0.8618,
+      "step": 1933
+    },
+    {
+      "epoch": 1.1281901706285549,
+      "grad_norm": 1.2844151258468628,
+      "learning_rate": 2.24609375e-05,
+      "loss": 0.7397,
+      "step": 1934
+    },
+    {
+      "epoch": 1.128773516114919,
+      "grad_norm": 1.1625947952270508,
+      "learning_rate": 2.244591346153846e-05,
+      "loss": 0.8301,
+      "step": 1935
+    },
+    {
+      "epoch": 1.1293568616012835,
+      "grad_norm": 1.3182225227355957,
+      "learning_rate": 2.2430889423076923e-05,
+      "loss": 0.7165,
+      "step": 1936
+    },
+    {
+      "epoch": 1.1299402070876476,
+      "grad_norm": 1.4218560457229614,
+      "learning_rate": 2.2415865384615385e-05,
+      "loss": 0.7256,
+      "step": 1937
+    },
+    {
+      "epoch": 1.130523552574012,
+      "grad_norm": 1.061953067779541,
+      "learning_rate": 2.2400841346153847e-05,
+      "loss": 0.8491,
+      "step": 1938
+    },
+    {
+      "epoch": 1.1311068980603762,
+      "grad_norm": 1.1568974256515503,
+      "learning_rate": 2.238581730769231e-05,
+      "loss": 0.7943,
+      "step": 1939
+    },
+    {
+      "epoch": 1.1316902435467406,
+      "grad_norm": 1.0611028671264648,
+      "learning_rate": 2.2370793269230768e-05,
+      "loss": 0.8708,
+      "step": 1940
+    },
+    {
+      "epoch": 1.1322735890331048,
+      "grad_norm": 1.1520451307296753,
+      "learning_rate": 2.2355769230769233e-05,
+      "loss": 0.7339,
+      "step": 1941
+    },
+    {
+      "epoch": 1.1328569345194692,
+      "grad_norm": 1.2494333982467651,
+      "learning_rate": 2.2340745192307695e-05,
+      "loss": 0.7952,
+      "step": 1942
+    },
+    {
+      "epoch": 1.1334402800058334,
+      "grad_norm": 1.1733430624008179,
+      "learning_rate": 2.2325721153846153e-05,
+      "loss": 0.7348,
+      "step": 1943
+    },
+    {
+      "epoch": 1.1340236254921978,
+      "grad_norm": 1.1535006761550903,
+      "learning_rate": 2.2310697115384615e-05,
+      "loss": 0.8333,
+      "step": 1944
+    },
+    {
+      "epoch": 1.134606970978562,
+      "grad_norm": 1.1618400812149048,
+      "learning_rate": 2.229567307692308e-05,
+      "loss": 0.8816,
+      "step": 1945
+    },
+    {
+      "epoch": 1.1351903164649264,
+      "grad_norm": 1.43650221824646,
+      "learning_rate": 2.228064903846154e-05,
+      "loss": 0.9497,
+      "step": 1946
+    },
+    {
+      "epoch": 1.1357736619512906,
+      "grad_norm": 1.2452301979064941,
+      "learning_rate": 2.2265625e-05,
+      "loss": 0.6446,
+      "step": 1947
+    },
+    {
+      "epoch": 1.136357007437655,
+      "grad_norm": 1.2092143297195435,
+      "learning_rate": 2.2250600961538463e-05,
+      "loss": 0.8083,
+      "step": 1948
+    },
+    {
+      "epoch": 1.1369403529240192,
+      "grad_norm": 1.0551011562347412,
+      "learning_rate": 2.223557692307692e-05,
+      "loss": 0.9274,
+      "step": 1949
+    },
+    {
+      "epoch": 1.1375236984103836,
+      "grad_norm": 1.204288125038147,
+      "learning_rate": 2.2220552884615386e-05,
+      "loss": 1.043,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1381070438967478,
+      "grad_norm": 1.4167194366455078,
+      "learning_rate": 2.220552884615385e-05,
+      "loss": 0.8985,
+      "step": 1951
+    },
+    {
+      "epoch": 1.1386903893831122,
+      "grad_norm": 1.2125877141952515,
+      "learning_rate": 2.2190504807692307e-05,
+      "loss": 0.7084,
+      "step": 1952
+    },
+    {
+      "epoch": 1.1392737348694764,
+      "grad_norm": 1.3489576578140259,
+      "learning_rate": 2.217548076923077e-05,
+      "loss": 0.7183,
+      "step": 1953
+    },
+    {
+      "epoch": 1.1398570803558408,
+      "grad_norm": 1.4352302551269531,
+      "learning_rate": 2.2160456730769234e-05,
+      "loss": 0.9226,
+      "step": 1954
+    },
+    {
+      "epoch": 1.140440425842205,
+      "grad_norm": 1.2473138570785522,
+      "learning_rate": 2.2145432692307693e-05,
+      "loss": 0.8703,
+      "step": 1955
+    },
+    {
+      "epoch": 1.1410237713285694,
+      "grad_norm": 1.1649906635284424,
+      "learning_rate": 2.2130408653846154e-05,
+      "loss": 0.7713,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1416071168149338,
+      "grad_norm": 1.1831884384155273,
+      "learning_rate": 2.2115384615384616e-05,
+      "loss": 0.9852,
+      "step": 1957
+    },
+    {
+      "epoch": 1.142190462301298,
+      "grad_norm": 1.1909598112106323,
+      "learning_rate": 2.2100360576923078e-05,
+      "loss": 0.7232,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1427738077876621,
+      "grad_norm": 1.0720024108886719,
+      "learning_rate": 2.208533653846154e-05,
+      "loss": 0.8207,
+      "step": 1959
+    },
+    {
+      "epoch": 1.1433571532740265,
+      "grad_norm": 1.0868524312973022,
+      "learning_rate": 2.2070312500000002e-05,
+      "loss": 0.7564,
+      "step": 1960
+    },
+    {
+      "epoch": 1.143940498760391,
+      "grad_norm": 1.2819868326187134,
+      "learning_rate": 2.205528846153846e-05,
+      "loss": 0.8171,
+      "step": 1961
+    },
+    {
+      "epoch": 1.1445238442467551,
+      "grad_norm": 1.2320795059204102,
+      "learning_rate": 2.2040264423076922e-05,
+      "loss": 0.6759,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1451071897331195,
+      "grad_norm": 1.2730618715286255,
+      "learning_rate": 2.2025240384615388e-05,
+      "loss": 0.8357,
+      "step": 1963
+    },
+    {
+      "epoch": 1.1456905352194837,
+      "grad_norm": 1.3447294235229492,
+      "learning_rate": 2.2010216346153846e-05,
+      "loss": 0.7348,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1462738807058481,
+      "grad_norm": 1.215040683746338,
+      "learning_rate": 2.1995192307692308e-05,
+      "loss": 0.9677,
+      "step": 1965
+    },
+    {
+      "epoch": 1.1468572261922123,
+      "grad_norm": 0.992956280708313,
+      "learning_rate": 2.198016826923077e-05,
+      "loss": 0.7393,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1474405716785767,
+      "grad_norm": 1.204768419265747,
+      "learning_rate": 2.1965144230769232e-05,
+      "loss": 0.7813,
+      "step": 1967
+    },
+    {
+      "epoch": 1.148023917164941,
+      "grad_norm": 1.2408292293548584,
+      "learning_rate": 2.1950120192307694e-05,
+      "loss": 0.9303,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1486072626513053,
+      "grad_norm": 1.1849360466003418,
+      "learning_rate": 2.1935096153846156e-05,
+      "loss": 0.836,
+      "step": 1969
+    },
+    {
+      "epoch": 1.1491906081376695,
+      "grad_norm": 1.2159719467163086,
+      "learning_rate": 2.1920072115384614e-05,
+      "loss": 0.6705,
+      "step": 1970
+    },
+    {
+      "epoch": 1.149773953624034,
+      "grad_norm": 1.2858052253723145,
+      "learning_rate": 2.190504807692308e-05,
+      "loss": 0.8785,
+      "step": 1971
+    },
+    {
+      "epoch": 1.150357299110398,
+      "grad_norm": 1.2040108442306519,
+      "learning_rate": 2.189002403846154e-05,
+      "loss": 0.7998,
+      "step": 1972
+    },
+    {
+      "epoch": 1.1509406445967625,
+      "grad_norm": 1.2969449758529663,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.7393,
+      "step": 1973
+    },
+    {
+      "epoch": 1.1515239900831267,
+      "grad_norm": 1.1521106958389282,
+      "learning_rate": 2.185997596153846e-05,
+      "loss": 0.9285,
+      "step": 1974
+    },
+    {
+      "epoch": 1.152107335569491,
+      "grad_norm": 1.139011025428772,
+      "learning_rate": 2.1844951923076924e-05,
+      "loss": 0.8499,
+      "step": 1975
+    },
+    {
+      "epoch": 1.1526906810558553,
+      "grad_norm": 1.3250030279159546,
+      "learning_rate": 2.1829927884615385e-05,
+      "loss": 0.7346,
+      "step": 1976
+    },
+    {
+      "epoch": 1.1532740265422197,
+      "grad_norm": 1.0421650409698486,
+      "learning_rate": 2.1814903846153847e-05,
+      "loss": 0.739,
+      "step": 1977
+    },
+    {
+      "epoch": 1.1538573720285838,
+      "grad_norm": 1.3398250341415405,
+      "learning_rate": 2.179987980769231e-05,
+      "loss": 0.644,
+      "step": 1978
+    },
+    {
+      "epoch": 1.1544407175149483,
+      "grad_norm": 1.0961638689041138,
+      "learning_rate": 2.1784855769230768e-05,
+      "loss": 0.8046,
+      "step": 1979
+    },
+    {
+      "epoch": 1.1550240630013124,
+      "grad_norm": 1.2048981189727783,
+      "learning_rate": 2.1769831730769233e-05,
+      "loss": 0.6226,
+      "step": 1980
+    },
+    {
+      "epoch": 1.1556074084876768,
+      "grad_norm": 1.395005702972412,
+      "learning_rate": 2.1754807692307695e-05,
+      "loss": 0.749,
+      "step": 1981
+    },
+    {
+      "epoch": 1.156190753974041,
+      "grad_norm": 1.2888487577438354,
+      "learning_rate": 2.1739783653846153e-05,
+      "loss": 0.8828,
+      "step": 1982
+    },
+    {
+      "epoch": 1.1567740994604054,
+      "grad_norm": 1.237052083015442,
+      "learning_rate": 2.1724759615384615e-05,
+      "loss": 0.742,
+      "step": 1983
+    },
+    {
+      "epoch": 1.1573574449467698,
+      "grad_norm": 1.3370431661605835,
+      "learning_rate": 2.170973557692308e-05,
+      "loss": 0.6737,
+      "step": 1984
+    },
+    {
+      "epoch": 1.157940790433134,
+      "grad_norm": 1.3525289297103882,
+      "learning_rate": 2.169471153846154e-05,
+      "loss": 0.6262,
+      "step": 1985
+    },
+    {
+      "epoch": 1.1585241359194982,
+      "grad_norm": 1.2294694185256958,
+      "learning_rate": 2.16796875e-05,
+      "loss": 0.6299,
+      "step": 1986
+    },
+    {
+      "epoch": 1.1591074814058626,
+      "grad_norm": 1.3560107946395874,
+      "learning_rate": 2.1664663461538463e-05,
+      "loss": 0.8025,
+      "step": 1987
+    },
+    {
+      "epoch": 1.159690826892227,
+      "grad_norm": 1.2413029670715332,
+      "learning_rate": 2.164963942307692e-05,
+      "loss": 0.7526,
+      "step": 1988
+    },
+    {
+      "epoch": 1.1602741723785912,
+      "grad_norm": 1.2291741371154785,
+      "learning_rate": 2.1634615384615387e-05,
+      "loss": 0.9964,
+      "step": 1989
+    },
+    {
+      "epoch": 1.1608575178649556,
+      "grad_norm": 1.1054651737213135,
+      "learning_rate": 2.161959134615385e-05,
+      "loss": 0.8606,
+      "step": 1990
+    },
+    {
+      "epoch": 1.1614408633513198,
+      "grad_norm": 1.2737064361572266,
+      "learning_rate": 2.1604567307692307e-05,
+      "loss": 0.7849,
+      "step": 1991
+    },
+    {
+      "epoch": 1.1620242088376842,
+      "grad_norm": 1.2561743259429932,
+      "learning_rate": 2.158954326923077e-05,
+      "loss": 0.914,
+      "step": 1992
+    },
+    {
+      "epoch": 1.1626075543240484,
+      "grad_norm": 1.4079277515411377,
+      "learning_rate": 2.1574519230769234e-05,
+      "loss": 0.8546,
+      "step": 1993
+    },
+    {
+      "epoch": 1.1631908998104128,
+      "grad_norm": 1.1819322109222412,
+      "learning_rate": 2.1559495192307693e-05,
+      "loss": 0.7467,
+      "step": 1994
+    },
+    {
+      "epoch": 1.163774245296777,
+      "grad_norm": 1.2287219762802124,
+      "learning_rate": 2.1544471153846155e-05,
+      "loss": 0.7146,
+      "step": 1995
+    },
+    {
+      "epoch": 1.1643575907831414,
+      "grad_norm": 1.2632535696029663,
+      "learning_rate": 2.1529447115384616e-05,
+      "loss": 0.7355,
+      "step": 1996
+    },
+    {
+      "epoch": 1.1649409362695056,
+      "grad_norm": 1.3165422677993774,
+      "learning_rate": 2.151442307692308e-05,
+      "loss": 0.7853,
+      "step": 1997
+    },
+    {
+      "epoch": 1.16552428175587,
+      "grad_norm": 1.056222915649414,
+      "learning_rate": 2.149939903846154e-05,
+      "loss": 0.7185,
+      "step": 1998
+    },
+    {
+      "epoch": 1.1661076272422342,
+      "grad_norm": 1.2883901596069336,
+      "learning_rate": 2.1484375000000002e-05,
+      "loss": 0.7979,
+      "step": 1999
+    },
+    {
+      "epoch": 1.1666909727285986,
+      "grad_norm": 1.367472767829895,
+      "learning_rate": 2.146935096153846e-05,
+      "loss": 0.8075,
+      "step": 2000
+    },
+    {
+      "epoch": 1.1666909727285986,
+      "eval_loss_squad": 0.833221665751189,
+      "eval_perplexity": 8.233223012549466,
+      "eval_perplexity_reconstruct": 1.9100281333491649,
+      "step": 2000
+    },
+    {
+      "epoch": 1.1672743182149627,
+      "grad_norm": 1.1449155807495117,
+      "learning_rate": 2.1454326923076923e-05,
+      "loss": 0.865,
+      "step": 2001
+    },
+    {
+      "epoch": 1.1678576637013272,
+      "grad_norm": 1.1930732727050781,
+      "learning_rate": 2.1439302884615388e-05,
+      "loss": 0.5906,
+      "step": 2002
+    },
+    {
+      "epoch": 1.1684410091876913,
+      "grad_norm": 1.4782718420028687,
+      "learning_rate": 2.1424278846153846e-05,
+      "loss": 0.9741,
+      "step": 2003
+    },
+    {
+      "epoch": 1.1690243546740557,
+      "grad_norm": 1.2961211204528809,
+      "learning_rate": 2.1409254807692308e-05,
+      "loss": 0.713,
+      "step": 2004
+    },
+    {
+      "epoch": 1.16960770016042,
+      "grad_norm": 1.250820279121399,
+      "learning_rate": 2.139423076923077e-05,
+      "loss": 0.8144,
+      "step": 2005
+    },
+    {
+      "epoch": 1.1701910456467843,
+      "grad_norm": 1.1511048078536987,
+      "learning_rate": 2.1379206730769232e-05,
+      "loss": 0.822,
+      "step": 2006
+    },
+    {
+      "epoch": 1.1707743911331485,
+      "grad_norm": 1.1328845024108887,
+      "learning_rate": 2.1364182692307694e-05,
+      "loss": 0.8251,
+      "step": 2007
+    },
+    {
+      "epoch": 1.171357736619513,
+      "grad_norm": 1.0141953229904175,
+      "learning_rate": 2.1349158653846156e-05,
+      "loss": 0.5794,
+      "step": 2008
+    },
+    {
+      "epoch": 1.171941082105877,
+      "grad_norm": 1.3566126823425293,
+      "learning_rate": 2.1334134615384614e-05,
+      "loss": 0.9965,
+      "step": 2009
+    },
+    {
+      "epoch": 1.1725244275922415,
+      "grad_norm": 1.246524691581726,
+      "learning_rate": 2.131911057692308e-05,
+      "loss": 0.7536,
+      "step": 2010
+    },
+    {
+      "epoch": 1.173107773078606,
+      "grad_norm": 1.3714301586151123,
+      "learning_rate": 2.130408653846154e-05,
+      "loss": 0.7692,
+      "step": 2011
+    },
+    {
+      "epoch": 1.17369111856497,
+      "grad_norm": 1.1998827457427979,
+      "learning_rate": 2.12890625e-05,
+      "loss": 0.8805,
+      "step": 2012
+    },
+    {
+      "epoch": 1.1742744640513343,
+      "grad_norm": 1.1431297063827515,
+      "learning_rate": 2.1274038461538462e-05,
+      "loss": 0.9395,
+      "step": 2013
+    },
+    {
+      "epoch": 1.1748578095376987,
+      "grad_norm": 1.1474422216415405,
+      "learning_rate": 2.1259014423076924e-05,
+      "loss": 0.5988,
+      "step": 2014
+    },
+    {
+      "epoch": 1.175441155024063,
+      "grad_norm": 1.1838253736495972,
+      "learning_rate": 2.1243990384615386e-05,
+      "loss": 1.0735,
+      "step": 2015
+    },
+    {
+      "epoch": 1.1760245005104273,
+      "grad_norm": 1.1664282083511353,
+      "learning_rate": 2.1228966346153847e-05,
+      "loss": 0.9661,
+      "step": 2016
+    },
+    {
+      "epoch": 1.1766078459967917,
+      "grad_norm": 1.2454452514648438,
+      "learning_rate": 2.121394230769231e-05,
+      "loss": 0.8219,
+      "step": 2017
+    },
+    {
+      "epoch": 1.1771911914831559,
+      "grad_norm": 1.3535860776901245,
+      "learning_rate": 2.1198918269230768e-05,
+      "loss": 0.855,
+      "step": 2018
+    },
+    {
+      "epoch": 1.1777745369695203,
+      "grad_norm": 1.3076273202896118,
+      "learning_rate": 2.1183894230769233e-05,
+      "loss": 0.9618,
+      "step": 2019
+    },
+    {
+      "epoch": 1.1783578824558845,
+      "grad_norm": 1.2071937322616577,
+      "learning_rate": 2.1168870192307695e-05,
+      "loss": 0.7426,
+      "step": 2020
+    },
+    {
+      "epoch": 1.1789412279422489,
+      "grad_norm": 1.273971676826477,
+      "learning_rate": 2.1153846153846154e-05,
+      "loss": 0.8335,
+      "step": 2021
+    },
+    {
+      "epoch": 1.179524573428613,
+      "grad_norm": 1.1485315561294556,
+      "learning_rate": 2.1138822115384615e-05,
+      "loss": 0.9572,
+      "step": 2022
+    },
+    {
+      "epoch": 1.1801079189149775,
+      "grad_norm": 1.3041000366210938,
+      "learning_rate": 2.112379807692308e-05,
+      "loss": 0.6271,
+      "step": 2023
+    },
+    {
+      "epoch": 1.1806912644013416,
+      "grad_norm": 1.3196794986724854,
+      "learning_rate": 2.110877403846154e-05,
+      "loss": 0.7458,
+      "step": 2024
+    },
+    {
+      "epoch": 1.181274609887706,
+      "grad_norm": 1.1841623783111572,
+      "learning_rate": 2.109375e-05,
+      "loss": 0.7578,
+      "step": 2025
+    },
+    {
+      "epoch": 1.1818579553740702,
+      "grad_norm": 1.098708152770996,
+      "learning_rate": 2.1078725961538463e-05,
+      "loss": 0.7583,
+      "step": 2026
+    },
+    {
+      "epoch": 1.1824413008604346,
+      "grad_norm": 1.4319161176681519,
+      "learning_rate": 2.106370192307692e-05,
+      "loss": 0.6168,
+      "step": 2027
+    },
+    {
+      "epoch": 1.1830246463467988,
+      "grad_norm": 1.0580462217330933,
+      "learning_rate": 2.1048677884615387e-05,
+      "loss": 0.9887,
+      "step": 2028
+    },
+    {
+      "epoch": 1.1836079918331632,
+      "grad_norm": 0.9840408563613892,
+      "learning_rate": 2.103365384615385e-05,
+      "loss": 0.8125,
+      "step": 2029
+    },
+    {
+      "epoch": 1.1841913373195274,
+      "grad_norm": 1.2513033151626587,
+      "learning_rate": 2.1018629807692307e-05,
+      "loss": 0.8072,
+      "step": 2030
+    },
+    {
+      "epoch": 1.1847746828058918,
+      "grad_norm": 1.3638144731521606,
+      "learning_rate": 2.100360576923077e-05,
+      "loss": 0.7152,
+      "step": 2031
+    },
+    {
+      "epoch": 1.185358028292256,
+      "grad_norm": 1.3127323389053345,
+      "learning_rate": 2.0988581730769234e-05,
+      "loss": 0.8506,
+      "step": 2032
+    },
+    {
+      "epoch": 1.1859413737786204,
+      "grad_norm": 1.028713345527649,
+      "learning_rate": 2.0973557692307693e-05,
+      "loss": 0.7184,
+      "step": 2033
+    },
+    {
+      "epoch": 1.1865247192649846,
+      "grad_norm": 1.3238139152526855,
+      "learning_rate": 2.0958533653846155e-05,
+      "loss": 0.9182,
+      "step": 2034
+    },
+    {
+      "epoch": 1.187108064751349,
+      "grad_norm": 1.1423969268798828,
+      "learning_rate": 2.0943509615384617e-05,
+      "loss": 0.6711,
+      "step": 2035
+    },
+    {
+      "epoch": 1.1876914102377132,
+      "grad_norm": 1.0575798749923706,
+      "learning_rate": 2.092848557692308e-05,
+      "loss": 0.6561,
+      "step": 2036
+    },
+    {
+      "epoch": 1.1882747557240776,
+      "grad_norm": 1.18056058883667,
+      "learning_rate": 2.091346153846154e-05,
+      "loss": 0.9018,
+      "step": 2037
+    },
+    {
+      "epoch": 1.188858101210442,
+      "grad_norm": 1.0494967699050903,
+      "learning_rate": 2.0898437500000002e-05,
+      "loss": 0.7882,
+      "step": 2038
+    },
+    {
+      "epoch": 1.1894414466968062,
+      "grad_norm": 1.2525426149368286,
+      "learning_rate": 2.088341346153846e-05,
+      "loss": 0.7677,
+      "step": 2039
+    },
+    {
+      "epoch": 1.1900247921831704,
+      "grad_norm": 1.2103064060211182,
+      "learning_rate": 2.0868389423076923e-05,
+      "loss": 0.7308,
+      "step": 2040
+    },
+    {
+      "epoch": 1.1906081376695348,
+      "grad_norm": 1.238573670387268,
+      "learning_rate": 2.0853365384615388e-05,
+      "loss": 0.8663,
+      "step": 2041
+    },
+    {
+      "epoch": 1.1911914831558992,
+      "grad_norm": 1.1449916362762451,
+      "learning_rate": 2.0838341346153846e-05,
+      "loss": 0.9087,
+      "step": 2042
+    },
+    {
+      "epoch": 1.1917748286422634,
+      "grad_norm": 0.9556616544723511,
+      "learning_rate": 2.082331730769231e-05,
+      "loss": 0.8262,
+      "step": 2043
+    },
+    {
+      "epoch": 1.1923581741286278,
+      "grad_norm": 1.0977288484573364,
+      "learning_rate": 2.080829326923077e-05,
+      "loss": 0.7063,
+      "step": 2044
+    },
+    {
+      "epoch": 1.192941519614992,
+      "grad_norm": 1.09841787815094,
+      "learning_rate": 2.0793269230769232e-05,
+      "loss": 0.8165,
+      "step": 2045
+    },
+    {
+      "epoch": 1.1935248651013564,
+      "grad_norm": 1.3817111253738403,
+      "learning_rate": 2.0778245192307694e-05,
+      "loss": 0.8761,
+      "step": 2046
+    },
+    {
+      "epoch": 1.1941082105877205,
+      "grad_norm": 1.6460163593292236,
+      "learning_rate": 2.0763221153846156e-05,
+      "loss": 0.9336,
+      "step": 2047
+    },
+    {
+      "epoch": 1.194691556074085,
+      "grad_norm": 1.1768983602523804,
+      "learning_rate": 2.0748197115384614e-05,
+      "loss": 0.7355,
+      "step": 2048
+    },
+    {
+      "epoch": 1.1952749015604491,
+      "grad_norm": 1.1434657573699951,
+      "learning_rate": 2.073317307692308e-05,
+      "loss": 0.8538,
+      "step": 2049
+    },
+    {
+      "epoch": 1.1958582470468135,
+      "grad_norm": 1.3496836423873901,
+      "learning_rate": 2.071814903846154e-05,
+      "loss": 0.6867,
+      "step": 2050
+    },
+    {
+      "epoch": 1.1964415925331777,
+      "grad_norm": 1.1143194437026978,
+      "learning_rate": 2.0703125e-05,
+      "loss": 0.4993,
+      "step": 2051
+    },
+    {
+      "epoch": 1.1970249380195421,
+      "grad_norm": 1.1078752279281616,
+      "learning_rate": 2.0688100961538462e-05,
+      "loss": 0.8276,
+      "step": 2052
+    },
+    {
+      "epoch": 1.1976082835059063,
+      "grad_norm": 1.3760533332824707,
+      "learning_rate": 2.0673076923076924e-05,
+      "loss": 0.7688,
+      "step": 2053
+    },
+    {
+      "epoch": 1.1981916289922707,
+      "grad_norm": 1.4191945791244507,
+      "learning_rate": 2.0658052884615386e-05,
+      "loss": 0.7183,
+      "step": 2054
+    },
+    {
+      "epoch": 1.198774974478635,
+      "grad_norm": 3.6387343406677246,
+      "learning_rate": 2.0643028846153848e-05,
+      "loss": 0.9276,
+      "step": 2055
+    },
+    {
+      "epoch": 1.1993583199649993,
+      "grad_norm": 1.3375924825668335,
+      "learning_rate": 2.062800480769231e-05,
+      "loss": 0.8397,
+      "step": 2056
+    },
+    {
+      "epoch": 1.1999416654513635,
+      "grad_norm": 1.2695611715316772,
+      "learning_rate": 2.0612980769230768e-05,
+      "loss": 1.0456,
+      "step": 2057
+    },
+    {
+      "epoch": 1.200525010937728,
+      "grad_norm": 1.278247594833374,
+      "learning_rate": 2.0597956730769233e-05,
+      "loss": 0.8718,
+      "step": 2058
+    },
+    {
+      "epoch": 1.201108356424092,
+      "grad_norm": 1.2946563959121704,
+      "learning_rate": 2.0582932692307695e-05,
+      "loss": 0.7343,
+      "step": 2059
+    },
+    {
+      "epoch": 1.2016917019104565,
+      "grad_norm": 1.5262877941131592,
+      "learning_rate": 2.0567908653846154e-05,
+      "loss": 0.8788,
+      "step": 2060
+    },
+    {
+      "epoch": 1.2022750473968207,
+      "grad_norm": 1.3010066747665405,
+      "learning_rate": 2.0552884615384616e-05,
+      "loss": 0.7022,
+      "step": 2061
+    },
+    {
+      "epoch": 1.202858392883185,
+      "grad_norm": 1.112301230430603,
+      "learning_rate": 2.053786057692308e-05,
+      "loss": 0.8348,
+      "step": 2062
+    },
+    {
+      "epoch": 1.2034417383695493,
+      "grad_norm": 1.1436960697174072,
+      "learning_rate": 2.052283653846154e-05,
+      "loss": 0.7332,
+      "step": 2063
+    },
+    {
+      "epoch": 1.2040250838559137,
+      "grad_norm": 1.1401253938674927,
+      "learning_rate": 2.05078125e-05,
+      "loss": 0.7297,
+      "step": 2064
+    },
+    {
+      "epoch": 1.204608429342278,
+      "grad_norm": 1.187334418296814,
+      "learning_rate": 2.0492788461538463e-05,
+      "loss": 0.8428,
+      "step": 2065
+    },
+    {
+      "epoch": 1.2051917748286423,
+      "grad_norm": 1.499940037727356,
+      "learning_rate": 2.047776442307692e-05,
+      "loss": 0.8032,
+      "step": 2066
+    },
+    {
+      "epoch": 1.2057751203150064,
+      "grad_norm": 1.4192659854888916,
+      "learning_rate": 2.0462740384615387e-05,
+      "loss": 0.7676,
+      "step": 2067
+    },
+    {
+      "epoch": 1.2063584658013708,
+      "grad_norm": 1.6800339221954346,
+      "learning_rate": 2.044771634615385e-05,
+      "loss": 0.7352,
+      "step": 2068
+    },
+    {
+      "epoch": 1.2069418112877353,
+      "grad_norm": 1.1487371921539307,
+      "learning_rate": 2.0432692307692307e-05,
+      "loss": 0.7481,
+      "step": 2069
+    },
+    {
+      "epoch": 1.2075251567740994,
+      "grad_norm": 1.2796707153320312,
+      "learning_rate": 2.041766826923077e-05,
+      "loss": 0.7011,
+      "step": 2070
+    },
+    {
+      "epoch": 1.2081085022604638,
+      "grad_norm": 2.365525722503662,
+      "learning_rate": 2.0402644230769235e-05,
+      "loss": 0.8039,
+      "step": 2071
+    },
+    {
+      "epoch": 1.208691847746828,
+      "grad_norm": 1.1125097274780273,
+      "learning_rate": 2.0387620192307693e-05,
+      "loss": 0.7027,
+      "step": 2072
+    },
+    {
+      "epoch": 1.2092751932331924,
+      "grad_norm": 1.3944780826568604,
+      "learning_rate": 2.0372596153846155e-05,
+      "loss": 0.8972,
+      "step": 2073
+    },
+    {
+      "epoch": 1.2098585387195566,
+      "grad_norm": 1.1701585054397583,
+      "learning_rate": 2.0357572115384617e-05,
+      "loss": 0.6956,
+      "step": 2074
+    },
+    {
+      "epoch": 1.210441884205921,
+      "grad_norm": 1.17959463596344,
+      "learning_rate": 2.034254807692308e-05,
+      "loss": 0.7278,
+      "step": 2075
+    },
+    {
+      "epoch": 1.2110252296922852,
+      "grad_norm": 1.3046574592590332,
+      "learning_rate": 2.032752403846154e-05,
+      "loss": 0.7054,
+      "step": 2076
+    },
+    {
+      "epoch": 1.2116085751786496,
+      "grad_norm": 1.1868085861206055,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.76,
+      "step": 2077
+    },
+    {
+      "epoch": 1.2121919206650138,
+      "grad_norm": 1.4176987409591675,
+      "learning_rate": 2.029747596153846e-05,
+      "loss": 0.8469,
+      "step": 2078
+    },
+    {
+      "epoch": 1.2127752661513782,
+      "grad_norm": 1.0783299207687378,
+      "learning_rate": 2.0282451923076923e-05,
+      "loss": 0.8074,
+      "step": 2079
+    },
+    {
+      "epoch": 1.2133586116377424,
+      "grad_norm": 1.113829493522644,
+      "learning_rate": 2.0267427884615388e-05,
+      "loss": 0.7359,
+      "step": 2080
+    },
+    {
+      "epoch": 1.2139419571241068,
+      "grad_norm": 1.3299684524536133,
+      "learning_rate": 2.0252403846153847e-05,
+      "loss": 0.8881,
+      "step": 2081
+    },
+    {
+      "epoch": 1.214525302610471,
+      "grad_norm": 1.1996777057647705,
+      "learning_rate": 2.023737980769231e-05,
+      "loss": 0.686,
+      "step": 2082
+    },
+    {
+      "epoch": 1.2151086480968354,
+      "grad_norm": 1.1687122583389282,
+      "learning_rate": 2.022235576923077e-05,
+      "loss": 0.8864,
+      "step": 2083
+    },
+    {
+      "epoch": 1.2156919935831996,
+      "grad_norm": 1.3415162563323975,
+      "learning_rate": 2.0207331730769232e-05,
+      "loss": 0.8052,
+      "step": 2084
+    },
+    {
+      "epoch": 1.216275339069564,
+      "grad_norm": 1.184653878211975,
+      "learning_rate": 2.0192307692307694e-05,
+      "loss": 0.7937,
+      "step": 2085
+    },
+    {
+      "epoch": 1.2168586845559282,
+      "grad_norm": 1.2197047472000122,
+      "learning_rate": 2.0177283653846156e-05,
+      "loss": 0.7549,
+      "step": 2086
+    },
+    {
+      "epoch": 1.2174420300422926,
+      "grad_norm": 1.2201251983642578,
+      "learning_rate": 2.0162259615384615e-05,
+      "loss": 0.818,
+      "step": 2087
+    },
+    {
+      "epoch": 1.2180253755286568,
+      "grad_norm": 1.0660450458526611,
+      "learning_rate": 2.014723557692308e-05,
+      "loss": 0.7244,
+      "step": 2088
+    },
+    {
+      "epoch": 1.2186087210150212,
+      "grad_norm": 1.112344741821289,
+      "learning_rate": 2.0132211538461542e-05,
+      "loss": 0.8623,
+      "step": 2089
+    },
+    {
+      "epoch": 1.2191920665013853,
+      "grad_norm": 1.066853404045105,
+      "learning_rate": 2.01171875e-05,
+      "loss": 0.7755,
+      "step": 2090
+    },
+    {
+      "epoch": 1.2197754119877497,
+      "grad_norm": 1.4612683057785034,
+      "learning_rate": 2.0102163461538462e-05,
+      "loss": 0.7503,
+      "step": 2091
+    },
+    {
+      "epoch": 1.2203587574741142,
+      "grad_norm": 1.3045361042022705,
+      "learning_rate": 2.0087139423076924e-05,
+      "loss": 1.054,
+      "step": 2092
+    },
+    {
+      "epoch": 1.2209421029604783,
+      "grad_norm": 1.891230821609497,
+      "learning_rate": 2.0072115384615386e-05,
+      "loss": 0.801,
+      "step": 2093
+    },
+    {
+      "epoch": 1.2215254484468425,
+      "grad_norm": 1.2401351928710938,
+      "learning_rate": 2.0057091346153848e-05,
+      "loss": 1.0192,
+      "step": 2094
+    },
+    {
+      "epoch": 1.222108793933207,
+      "grad_norm": 1.1924711465835571,
+      "learning_rate": 2.004206730769231e-05,
+      "loss": 0.7175,
+      "step": 2095
+    },
+    {
+      "epoch": 1.2226921394195713,
+      "grad_norm": 1.4150327444076538,
+      "learning_rate": 2.0027043269230768e-05,
+      "loss": 0.8763,
+      "step": 2096
+    },
+    {
+      "epoch": 1.2232754849059355,
+      "grad_norm": 1.2163945436477661,
+      "learning_rate": 2.0012019230769233e-05,
+      "loss": 0.678,
+      "step": 2097
+    },
+    {
+      "epoch": 1.2238588303923,
+      "grad_norm": 1.3845294713974,
+      "learning_rate": 1.9996995192307695e-05,
+      "loss": 0.658,
+      "step": 2098
+    },
+    {
+      "epoch": 1.224442175878664,
+      "grad_norm": 1.308927297592163,
+      "learning_rate": 1.9981971153846154e-05,
+      "loss": 0.6827,
+      "step": 2099
+    },
+    {
+      "epoch": 1.2250255213650285,
+      "grad_norm": 1.386710524559021,
+      "learning_rate": 1.9966947115384616e-05,
+      "loss": 0.6608,
+      "step": 2100
+    },
+    {
+      "epoch": 1.2256088668513927,
+      "grad_norm": 2.081791877746582,
+      "learning_rate": 1.9951923076923078e-05,
+      "loss": 0.8755,
+      "step": 2101
+    },
+    {
+      "epoch": 1.226192212337757,
+      "grad_norm": 1.240175724029541,
+      "learning_rate": 1.993689903846154e-05,
+      "loss": 0.7445,
+      "step": 2102
+    },
+    {
+      "epoch": 1.2267755578241213,
+      "grad_norm": 1.1266789436340332,
+      "learning_rate": 1.9921875e-05,
+      "loss": 0.816,
+      "step": 2103
+    },
+    {
+      "epoch": 1.2273589033104857,
+      "grad_norm": 1.2244000434875488,
+      "learning_rate": 1.9906850961538463e-05,
+      "loss": 0.8795,
+      "step": 2104
+    },
+    {
+      "epoch": 1.2279422487968499,
+      "grad_norm": 1.420829176902771,
+      "learning_rate": 1.9891826923076922e-05,
+      "loss": 0.909,
+      "step": 2105
+    },
+    {
+      "epoch": 1.2285255942832143,
+      "grad_norm": 1.031460165977478,
+      "learning_rate": 1.9876802884615387e-05,
+      "loss": 0.557,
+      "step": 2106
+    },
+    {
+      "epoch": 1.2291089397695785,
+      "grad_norm": 1.2154114246368408,
+      "learning_rate": 1.986177884615385e-05,
+      "loss": 0.7688,
+      "step": 2107
+    },
+    {
+      "epoch": 1.2296922852559429,
+      "grad_norm": 1.0314477682113647,
+      "learning_rate": 1.9846754807692307e-05,
+      "loss": 0.834,
+      "step": 2108
+    },
+    {
+      "epoch": 1.230275630742307,
+      "grad_norm": 1.227577805519104,
+      "learning_rate": 1.983173076923077e-05,
+      "loss": 0.7475,
+      "step": 2109
+    },
+    {
+      "epoch": 1.2308589762286715,
+      "grad_norm": 1.3141686916351318,
+      "learning_rate": 1.981670673076923e-05,
+      "loss": 0.8031,
+      "step": 2110
+    },
+    {
+      "epoch": 1.2314423217150356,
+      "grad_norm": 1.0540612936019897,
+      "learning_rate": 1.9801682692307693e-05,
+      "loss": 0.7732,
+      "step": 2111
+    },
+    {
+      "epoch": 1.2320256672014,
+      "grad_norm": 1.198327898979187,
+      "learning_rate": 1.9786658653846155e-05,
+      "loss": 0.8973,
+      "step": 2112
+    },
+    {
+      "epoch": 1.2326090126877642,
+      "grad_norm": 1.286871314048767,
+      "learning_rate": 1.9771634615384617e-05,
+      "loss": 0.6474,
+      "step": 2113
+    },
+    {
+      "epoch": 1.2331923581741286,
+      "grad_norm": 1.4305251836776733,
+      "learning_rate": 1.975661057692308e-05,
+      "loss": 1.0454,
+      "step": 2114
+    },
+    {
+      "epoch": 1.233775703660493,
+      "grad_norm": 1.2102510929107666,
+      "learning_rate": 1.974158653846154e-05,
+      "loss": 0.5195,
+      "step": 2115
+    },
+    {
+      "epoch": 1.2343590491468572,
+      "grad_norm": 1.3697700500488281,
+      "learning_rate": 1.9726562500000003e-05,
+      "loss": 0.7984,
+      "step": 2116
+    },
+    {
+      "epoch": 1.2349423946332214,
+      "grad_norm": 1.121392846107483,
+      "learning_rate": 1.971153846153846e-05,
+      "loss": 0.9984,
+      "step": 2117
+    },
+    {
+      "epoch": 1.2355257401195858,
+      "grad_norm": 1.2788634300231934,
+      "learning_rate": 1.9696514423076923e-05,
+      "loss": 0.7598,
+      "step": 2118
+    },
+    {
+      "epoch": 1.2361090856059502,
+      "grad_norm": 1.1477173566818237,
+      "learning_rate": 1.9681490384615385e-05,
+      "loss": 0.7552,
+      "step": 2119
+    },
+    {
+      "epoch": 1.2366924310923144,
+      "grad_norm": 1.1774203777313232,
+      "learning_rate": 1.9666466346153847e-05,
+      "loss": 0.8124,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2372757765786786,
+      "grad_norm": 1.0521777868270874,
+      "learning_rate": 1.965144230769231e-05,
+      "loss": 0.7731,
+      "step": 2121
+    },
+    {
+      "epoch": 1.237859122065043,
+      "grad_norm": 1.3230030536651611,
+      "learning_rate": 1.963641826923077e-05,
+      "loss": 0.8122,
+      "step": 2122
+    },
+    {
+      "epoch": 1.2384424675514074,
+      "grad_norm": 1.1762146949768066,
+      "learning_rate": 1.9621394230769232e-05,
+      "loss": 0.7926,
+      "step": 2123
+    },
+    {
+      "epoch": 1.2390258130377716,
+      "grad_norm": 1.3138278722763062,
+      "learning_rate": 1.9606370192307694e-05,
+      "loss": 0.8361,
+      "step": 2124
+    },
+    {
+      "epoch": 1.239609158524136,
+      "grad_norm": 1.3810380697250366,
+      "learning_rate": 1.9591346153846156e-05,
+      "loss": 0.8379,
+      "step": 2125
+    },
+    {
+      "epoch": 1.2401925040105002,
+      "grad_norm": 1.2080543041229248,
+      "learning_rate": 1.9576322115384615e-05,
+      "loss": 1.0409,
+      "step": 2126
+    },
+    {
+      "epoch": 1.2407758494968646,
+      "grad_norm": 1.35194730758667,
+      "learning_rate": 1.956129807692308e-05,
+      "loss": 0.5343,
+      "step": 2127
+    },
+    {
+      "epoch": 1.2413591949832288,
+      "grad_norm": 1.287524700164795,
+      "learning_rate": 1.954627403846154e-05,
+      "loss": 0.9522,
+      "step": 2128
+    },
+    {
+      "epoch": 1.2419425404695932,
+      "grad_norm": 1.3742008209228516,
+      "learning_rate": 1.953125e-05,
+      "loss": 0.6388,
+      "step": 2129
+    },
+    {
+      "epoch": 1.2425258859559574,
+      "grad_norm": 1.1241064071655273,
+      "learning_rate": 1.9516225961538462e-05,
+      "loss": 0.6185,
+      "step": 2130
+    },
+    {
+      "epoch": 1.2431092314423218,
+      "grad_norm": 1.279675006866455,
+      "learning_rate": 1.9501201923076924e-05,
+      "loss": 0.7721,
+      "step": 2131
+    },
+    {
+      "epoch": 1.243692576928686,
+      "grad_norm": 1.381134033203125,
+      "learning_rate": 1.9486177884615386e-05,
+      "loss": 0.7895,
+      "step": 2132
+    },
+    {
+      "epoch": 1.2442759224150504,
+      "grad_norm": 1.276642918586731,
+      "learning_rate": 1.9471153846153848e-05,
+      "loss": 0.855,
+      "step": 2133
+    },
+    {
+      "epoch": 1.2448592679014145,
+      "grad_norm": 1.0444273948669434,
+      "learning_rate": 1.945612980769231e-05,
+      "loss": 1.0147,
+      "step": 2134
+    },
+    {
+      "epoch": 1.245442613387779,
+      "grad_norm": 1.4284225702285767,
+      "learning_rate": 1.944110576923077e-05,
+      "loss": 0.8998,
+      "step": 2135
+    },
+    {
+      "epoch": 1.2460259588741431,
+      "grad_norm": 1.3835419416427612,
+      "learning_rate": 1.9426081730769234e-05,
+      "loss": 0.7556,
+      "step": 2136
+    },
+    {
+      "epoch": 1.2466093043605075,
+      "grad_norm": 1.3676860332489014,
+      "learning_rate": 1.9411057692307692e-05,
+      "loss": 0.7672,
+      "step": 2137
+    },
+    {
+      "epoch": 1.2471926498468717,
+      "grad_norm": 1.2197344303131104,
+      "learning_rate": 1.9396033653846154e-05,
+      "loss": 0.8391,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2477759953332361,
+      "grad_norm": 1.1958162784576416,
+      "learning_rate": 1.9381009615384616e-05,
+      "loss": 0.7733,
+      "step": 2139
+    },
+    {
+      "epoch": 1.2483593408196003,
+      "grad_norm": 1.3922827243804932,
+      "learning_rate": 1.9365985576923078e-05,
+      "loss": 0.6837,
+      "step": 2140
+    },
+    {
+      "epoch": 1.2489426863059647,
+      "grad_norm": 1.1373469829559326,
+      "learning_rate": 1.935096153846154e-05,
+      "loss": 0.8817,
+      "step": 2141
+    },
+    {
+      "epoch": 1.2495260317923291,
+      "grad_norm": 1.2826591730117798,
+      "learning_rate": 1.93359375e-05,
+      "loss": 0.8214,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2501093772786933,
+      "grad_norm": 1.4706931114196777,
+      "learning_rate": 1.9320913461538463e-05,
+      "loss": 0.8828,
+      "step": 2143
+    },
+    {
+      "epoch": 1.2506927227650575,
+      "grad_norm": 1.1995126008987427,
+      "learning_rate": 1.9305889423076922e-05,
+      "loss": 0.8192,
+      "step": 2144
+    },
+    {
+      "epoch": 1.251276068251422,
+      "grad_norm": 1.321846842765808,
+      "learning_rate": 1.9290865384615387e-05,
+      "loss": 0.7632,
+      "step": 2145
+    },
+    {
+      "epoch": 1.2518594137377863,
+      "grad_norm": 1.2701880931854248,
+      "learning_rate": 1.9275841346153846e-05,
+      "loss": 0.8689,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2524427592241505,
+      "grad_norm": 1.2398470640182495,
+      "learning_rate": 1.9260817307692308e-05,
+      "loss": 1.0235,
+      "step": 2147
+    },
+    {
+      "epoch": 1.2530261047105147,
+      "grad_norm": 1.2439467906951904,
+      "learning_rate": 1.924579326923077e-05,
+      "loss": 0.6753,
+      "step": 2148
+    },
+    {
+      "epoch": 1.253609450196879,
+      "grad_norm": 1.207457184791565,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.9007,
+      "step": 2149
+    },
+    {
+      "epoch": 1.2541927956832435,
+      "grad_norm": 1.1856311559677124,
+      "learning_rate": 1.9215745192307693e-05,
+      "loss": 0.6977,
+      "step": 2150
+    },
+    {
+      "epoch": 1.2547761411696077,
+      "grad_norm": 1.2477092742919922,
+      "learning_rate": 1.9200721153846155e-05,
+      "loss": 0.7035,
+      "step": 2151
+    },
+    {
+      "epoch": 1.255359486655972,
+      "grad_norm": 1.0819847583770752,
+      "learning_rate": 1.9185697115384617e-05,
+      "loss": 0.7668,
+      "step": 2152
+    },
+    {
+      "epoch": 1.2559428321423363,
+      "grad_norm": 1.4272301197052002,
+      "learning_rate": 1.917067307692308e-05,
+      "loss": 0.8473,
+      "step": 2153
+    },
+    {
+      "epoch": 1.2565261776287007,
+      "grad_norm": 1.478842854499817,
+      "learning_rate": 1.915564903846154e-05,
+      "loss": 0.8842,
+      "step": 2154
+    },
+    {
+      "epoch": 1.2571095231150649,
+      "grad_norm": 1.500758409500122,
+      "learning_rate": 1.9140625e-05,
+      "loss": 0.7502,
+      "step": 2155
+    },
+    {
+      "epoch": 1.2576928686014293,
+      "grad_norm": 1.173345685005188,
+      "learning_rate": 1.912560096153846e-05,
+      "loss": 0.7785,
+      "step": 2156
+    },
+    {
+      "epoch": 1.2582762140877934,
+      "grad_norm": 1.1746881008148193,
+      "learning_rate": 1.9110576923076923e-05,
+      "loss": 0.83,
+      "step": 2157
+    },
+    {
+      "epoch": 1.2588595595741578,
+      "grad_norm": 1.263342261314392,
+      "learning_rate": 1.9095552884615385e-05,
+      "loss": 0.896,
+      "step": 2158
+    },
+    {
+      "epoch": 1.259442905060522,
+      "grad_norm": 1.374733328819275,
+      "learning_rate": 1.9080528846153847e-05,
+      "loss": 0.8734,
+      "step": 2159
+    },
+    {
+      "epoch": 1.2600262505468864,
+      "grad_norm": 1.1060365438461304,
+      "learning_rate": 1.906550480769231e-05,
+      "loss": 0.684,
+      "step": 2160
+    },
+    {
+      "epoch": 1.2606095960332506,
+      "grad_norm": 1.289867639541626,
+      "learning_rate": 1.905048076923077e-05,
+      "loss": 0.8807,
+      "step": 2161
+    },
+    {
+      "epoch": 1.261192941519615,
+      "grad_norm": 0.9511016607284546,
+      "learning_rate": 1.9035456730769233e-05,
+      "loss": 0.9252,
+      "step": 2162
+    },
+    {
+      "epoch": 1.2617762870059792,
+      "grad_norm": 1.3257215023040771,
+      "learning_rate": 1.9020432692307695e-05,
+      "loss": 0.6726,
+      "step": 2163
+    },
+    {
+      "epoch": 1.2623596324923436,
+      "grad_norm": 1.180518627166748,
+      "learning_rate": 1.9005408653846153e-05,
+      "loss": 0.6201,
+      "step": 2164
+    },
+    {
+      "epoch": 1.262942977978708,
+      "grad_norm": 1.2338030338287354,
+      "learning_rate": 1.8990384615384615e-05,
+      "loss": 0.8228,
+      "step": 2165
+    },
+    {
+      "epoch": 1.2635263234650722,
+      "grad_norm": 0.9147897362709045,
+      "learning_rate": 1.897536057692308e-05,
+      "loss": 0.9485,
+      "step": 2166
+    },
+    {
+      "epoch": 1.2641096689514364,
+      "grad_norm": 1.209277629852295,
+      "learning_rate": 1.896033653846154e-05,
+      "loss": 0.9845,
+      "step": 2167
+    },
+    {
+      "epoch": 1.2646930144378008,
+      "grad_norm": 1.3521689176559448,
+      "learning_rate": 1.89453125e-05,
+      "loss": 0.8049,
+      "step": 2168
+    },
+    {
+      "epoch": 1.2652763599241652,
+      "grad_norm": 1.1422154903411865,
+      "learning_rate": 1.8930288461538462e-05,
+      "loss": 0.8085,
+      "step": 2169
+    },
+    {
+      "epoch": 1.2658597054105294,
+      "grad_norm": 1.3813396692276,
+      "learning_rate": 1.8915264423076924e-05,
+      "loss": 0.8315,
+      "step": 2170
+    },
+    {
+      "epoch": 1.2664430508968936,
+      "grad_norm": 1.1161137819290161,
+      "learning_rate": 1.8900240384615386e-05,
+      "loss": 0.8324,
+      "step": 2171
+    },
+    {
+      "epoch": 1.267026396383258,
+      "grad_norm": 1.3338303565979004,
+      "learning_rate": 1.8885216346153848e-05,
+      "loss": 0.8148,
+      "step": 2172
+    },
+    {
+      "epoch": 1.2676097418696224,
+      "grad_norm": 1.677724838256836,
+      "learning_rate": 1.8870192307692307e-05,
+      "loss": 0.8382,
+      "step": 2173
+    },
+    {
+      "epoch": 1.2681930873559866,
+      "grad_norm": 1.2323002815246582,
+      "learning_rate": 1.885516826923077e-05,
+      "loss": 0.8921,
+      "step": 2174
+    },
+    {
+      "epoch": 1.2687764328423508,
+      "grad_norm": 1.3112934827804565,
+      "learning_rate": 1.8840144230769234e-05,
+      "loss": 0.782,
+      "step": 2175
+    },
+    {
+      "epoch": 1.2693597783287152,
+      "grad_norm": 1.1889418363571167,
+      "learning_rate": 1.8825120192307692e-05,
+      "loss": 0.6976,
+      "step": 2176
+    },
+    {
+      "epoch": 1.2699431238150796,
+      "grad_norm": 1.2303264141082764,
+      "learning_rate": 1.8810096153846154e-05,
+      "loss": 0.7932,
+      "step": 2177
+    },
+    {
+      "epoch": 1.2705264693014438,
+      "grad_norm": 1.306663990020752,
+      "learning_rate": 1.8795072115384616e-05,
+      "loss": 0.8867,
+      "step": 2178
+    },
+    {
+      "epoch": 1.2711098147878082,
+      "grad_norm": 1.3073201179504395,
+      "learning_rate": 1.8780048076923078e-05,
+      "loss": 0.7185,
+      "step": 2179
+    },
+    {
+      "epoch": 1.2716931602741723,
+      "grad_norm": 1.2337119579315186,
+      "learning_rate": 1.876502403846154e-05,
+      "loss": 0.8584,
+      "step": 2180
+    },
+    {
+      "epoch": 1.2722765057605367,
+      "grad_norm": 1.0608634948730469,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.7698,
+      "step": 2181
+    },
+    {
+      "epoch": 1.272859851246901,
+      "grad_norm": 1.7613379955291748,
+      "learning_rate": 1.873497596153846e-05,
+      "loss": 0.8036,
+      "step": 2182
+    },
+    {
+      "epoch": 1.2734431967332653,
+      "grad_norm": 1.2020319700241089,
+      "learning_rate": 1.8719951923076922e-05,
+      "loss": 0.9223,
+      "step": 2183
+    },
+    {
+      "epoch": 1.2740265422196295,
+      "grad_norm": 1.1997487545013428,
+      "learning_rate": 1.8704927884615387e-05,
+      "loss": 0.8401,
+      "step": 2184
+    },
+    {
+      "epoch": 1.274609887705994,
+      "grad_norm": 1.2920103073120117,
+      "learning_rate": 1.8689903846153846e-05,
+      "loss": 0.7484,
+      "step": 2185
+    },
+    {
+      "epoch": 1.275193233192358,
+      "grad_norm": 1.2046105861663818,
+      "learning_rate": 1.8674879807692308e-05,
+      "loss": 0.6554,
+      "step": 2186
+    },
+    {
+      "epoch": 1.2757765786787225,
+      "grad_norm": 1.1587508916854858,
+      "learning_rate": 1.865985576923077e-05,
+      "loss": 0.9453,
+      "step": 2187
+    },
+    {
+      "epoch": 1.2763599241650867,
+      "grad_norm": 1.2004345655441284,
+      "learning_rate": 1.864483173076923e-05,
+      "loss": 0.7673,
+      "step": 2188
+    },
+    {
+      "epoch": 1.276943269651451,
+      "grad_norm": 1.2180094718933105,
+      "learning_rate": 1.8629807692307693e-05,
+      "loss": 0.7902,
+      "step": 2189
+    },
+    {
+      "epoch": 1.2775266151378153,
+      "grad_norm": 1.1839169263839722,
+      "learning_rate": 1.8614783653846155e-05,
+      "loss": 0.9672,
+      "step": 2190
+    },
+    {
+      "epoch": 1.2781099606241797,
+      "grad_norm": 1.2951889038085938,
+      "learning_rate": 1.8599759615384614e-05,
+      "loss": 0.8781,
+      "step": 2191
+    },
+    {
+      "epoch": 1.278693306110544,
+      "grad_norm": 1.199181318283081,
+      "learning_rate": 1.858473557692308e-05,
+      "loss": 0.8252,
+      "step": 2192
+    },
+    {
+      "epoch": 1.2792766515969083,
+      "grad_norm": 1.3422390222549438,
+      "learning_rate": 1.856971153846154e-05,
+      "loss": 0.8287,
+      "step": 2193
+    },
+    {
+      "epoch": 1.2798599970832725,
+      "grad_norm": 1.1554558277130127,
+      "learning_rate": 1.85546875e-05,
+      "loss": 0.8332,
+      "step": 2194
+    },
+    {
+      "epoch": 1.2804433425696369,
+      "grad_norm": 1.2082306146621704,
+      "learning_rate": 1.853966346153846e-05,
+      "loss": 0.8973,
+      "step": 2195
+    },
+    {
+      "epoch": 1.2810266880560013,
+      "grad_norm": 1.3980644941329956,
+      "learning_rate": 1.8524639423076923e-05,
+      "loss": 0.7318,
+      "step": 2196
+    },
+    {
+      "epoch": 1.2816100335423655,
+      "grad_norm": 1.290248155593872,
+      "learning_rate": 1.8509615384615385e-05,
+      "loss": 0.9601,
+      "step": 2197
+    },
+    {
+      "epoch": 1.2821933790287297,
+      "grad_norm": 1.2903095483779907,
+      "learning_rate": 1.8494591346153847e-05,
+      "loss": 0.853,
+      "step": 2198
+    },
+    {
+      "epoch": 1.282776724515094,
+      "grad_norm": 1.1215404272079468,
+      "learning_rate": 1.847956730769231e-05,
+      "loss": 0.8574,
+      "step": 2199
+    },
+    {
+      "epoch": 1.2833600700014585,
+      "grad_norm": 1.4814391136169434,
+      "learning_rate": 1.8464543269230767e-05,
+      "loss": 0.7185,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2833600700014585,
+      "eval_loss_squad": 0.8695367491571233,
+      "eval_perplexity": 8.314321659478429,
+      "eval_perplexity_reconstruct": 1.905092611405054,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2839434154878226,
+      "grad_norm": 1.5173953771591187,
+      "learning_rate": 1.8449519230769233e-05,
+      "loss": 0.7604,
+      "step": 2201
+    },
+    {
+      "epoch": 1.2845267609741868,
+      "grad_norm": 1.3316093683242798,
+      "learning_rate": 1.8434495192307695e-05,
+      "loss": 0.7515,
+      "step": 2202
+    },
+    {
+      "epoch": 1.2851101064605512,
+      "grad_norm": 1.444630742073059,
+      "learning_rate": 1.8419471153846153e-05,
+      "loss": 1.0531,
+      "step": 2203
+    },
+    {
+      "epoch": 1.2856934519469156,
+      "grad_norm": 1.2808470726013184,
+      "learning_rate": 1.8404447115384615e-05,
+      "loss": 0.6771,
+      "step": 2204
+    },
+    {
+      "epoch": 1.2862767974332798,
+      "grad_norm": 1.2151483297348022,
+      "learning_rate": 1.838942307692308e-05,
+      "loss": 0.8662,
+      "step": 2205
+    },
+    {
+      "epoch": 1.2868601429196442,
+      "grad_norm": 1.46257483959198,
+      "learning_rate": 1.837439903846154e-05,
+      "loss": 0.8319,
+      "step": 2206
+    },
+    {
+      "epoch": 1.2874434884060084,
+      "grad_norm": 1.3388731479644775,
+      "learning_rate": 1.8359375e-05,
+      "loss": 0.6788,
+      "step": 2207
+    },
+    {
+      "epoch": 1.2880268338923728,
+      "grad_norm": 1.563617467880249,
+      "learning_rate": 1.8344350961538463e-05,
+      "loss": 0.5617,
+      "step": 2208
+    },
+    {
+      "epoch": 1.288610179378737,
+      "grad_norm": 1.1539713144302368,
+      "learning_rate": 1.832932692307692e-05,
+      "loss": 0.7174,
+      "step": 2209
+    },
+    {
+      "epoch": 1.2891935248651014,
+      "grad_norm": 1.2138261795043945,
+      "learning_rate": 1.8314302884615386e-05,
+      "loss": 0.9121,
+      "step": 2210
+    },
+    {
+      "epoch": 1.2897768703514656,
+      "grad_norm": 1.2265892028808594,
+      "learning_rate": 1.8299278846153848e-05,
+      "loss": 0.7816,
+      "step": 2211
+    },
+    {
+      "epoch": 1.29036021583783,
+      "grad_norm": 1.1945728063583374,
+      "learning_rate": 1.8284254807692307e-05,
+      "loss": 0.8718,
+      "step": 2212
+    },
+    {
+      "epoch": 1.2909435613241942,
+      "grad_norm": 1.3123586177825928,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 0.7532,
+      "step": 2213
+    },
+    {
+      "epoch": 1.2915269068105586,
+      "grad_norm": 1.2278257608413696,
+      "learning_rate": 1.8254206730769234e-05,
+      "loss": 0.9726,
+      "step": 2214
+    },
+    {
+      "epoch": 1.2921102522969228,
+      "grad_norm": 1.100456714630127,
+      "learning_rate": 1.8239182692307692e-05,
+      "loss": 0.5921,
+      "step": 2215
+    },
+    {
+      "epoch": 1.2926935977832872,
+      "grad_norm": 1.1185805797576904,
+      "learning_rate": 1.8224158653846154e-05,
+      "loss": 0.7453,
+      "step": 2216
+    },
+    {
+      "epoch": 1.2932769432696514,
+      "grad_norm": 1.2159292697906494,
+      "learning_rate": 1.8209134615384616e-05,
+      "loss": 0.6872,
+      "step": 2217
+    },
+    {
+      "epoch": 1.2938602887560158,
+      "grad_norm": 1.1445262432098389,
+      "learning_rate": 1.8194110576923078e-05,
+      "loss": 0.7672,
+      "step": 2218
+    },
+    {
+      "epoch": 1.2944436342423802,
+      "grad_norm": 1.1859655380249023,
+      "learning_rate": 1.817908653846154e-05,
+      "loss": 0.6995,
+      "step": 2219
+    },
+    {
+      "epoch": 1.2950269797287444,
+      "grad_norm": 1.2108757495880127,
+      "learning_rate": 1.8164062500000002e-05,
+      "loss": 1.1017,
+      "step": 2220
+    },
+    {
+      "epoch": 1.2956103252151085,
+      "grad_norm": 1.0840204954147339,
+      "learning_rate": 1.814903846153846e-05,
+      "loss": 0.9441,
+      "step": 2221
+    },
+    {
+      "epoch": 1.296193670701473,
+      "grad_norm": 1.4668989181518555,
+      "learning_rate": 1.8134014423076922e-05,
+      "loss": 0.8163,
+      "step": 2222
+    },
+    {
+      "epoch": 1.2967770161878374,
+      "grad_norm": 1.3289495706558228,
+      "learning_rate": 1.8118990384615388e-05,
+      "loss": 0.7978,
+      "step": 2223
+    },
+    {
+      "epoch": 1.2973603616742015,
+      "grad_norm": 1.1220710277557373,
+      "learning_rate": 1.8103966346153846e-05,
+      "loss": 0.8322,
+      "step": 2224
+    },
+    {
+      "epoch": 1.2979437071605657,
+      "grad_norm": 1.1510266065597534,
+      "learning_rate": 1.8088942307692308e-05,
+      "loss": 0.735,
+      "step": 2225
+    },
+    {
+      "epoch": 1.2985270526469301,
+      "grad_norm": 1.145638108253479,
+      "learning_rate": 1.807391826923077e-05,
+      "loss": 0.8109,
+      "step": 2226
+    },
+    {
+      "epoch": 1.2991103981332945,
+      "grad_norm": 1.2273057699203491,
+      "learning_rate": 1.8058894230769232e-05,
+      "loss": 0.8697,
+      "step": 2227
+    },
+    {
+      "epoch": 1.2996937436196587,
+      "grad_norm": 0.9667252898216248,
+      "learning_rate": 1.8043870192307694e-05,
+      "loss": 0.7946,
+      "step": 2228
+    },
+    {
+      "epoch": 1.300277089106023,
+      "grad_norm": 1.183985710144043,
+      "learning_rate": 1.8028846153846156e-05,
+      "loss": 0.9245,
+      "step": 2229
+    },
+    {
+      "epoch": 1.3008604345923873,
+      "grad_norm": 1.4120750427246094,
+      "learning_rate": 1.8013822115384614e-05,
+      "loss": 0.899,
+      "step": 2230
+    },
+    {
+      "epoch": 1.3014437800787517,
+      "grad_norm": 1.9396076202392578,
+      "learning_rate": 1.799879807692308e-05,
+      "loss": 0.816,
+      "step": 2231
+    },
+    {
+      "epoch": 1.302027125565116,
+      "grad_norm": 1.4099103212356567,
+      "learning_rate": 1.798377403846154e-05,
+      "loss": 0.6821,
+      "step": 2232
+    },
+    {
+      "epoch": 1.3026104710514803,
+      "grad_norm": 1.2785017490386963,
+      "learning_rate": 1.796875e-05,
+      "loss": 0.9443,
+      "step": 2233
+    },
+    {
+      "epoch": 1.3031938165378445,
+      "grad_norm": 1.3573774099349976,
+      "learning_rate": 1.795372596153846e-05,
+      "loss": 0.7205,
+      "step": 2234
+    },
+    {
+      "epoch": 1.303777162024209,
+      "grad_norm": 1.1221401691436768,
+      "learning_rate": 1.7938701923076923e-05,
+      "loss": 0.7908,
+      "step": 2235
+    },
+    {
+      "epoch": 1.304360507510573,
+      "grad_norm": 1.1731091737747192,
+      "learning_rate": 1.7923677884615385e-05,
+      "loss": 0.6417,
+      "step": 2236
+    },
+    {
+      "epoch": 1.3049438529969375,
+      "grad_norm": 1.3627444505691528,
+      "learning_rate": 1.7908653846153847e-05,
+      "loss": 0.7929,
+      "step": 2237
+    },
+    {
+      "epoch": 1.3055271984833017,
+      "grad_norm": 1.4136035442352295,
+      "learning_rate": 1.789362980769231e-05,
+      "loss": 0.9224,
+      "step": 2238
+    },
+    {
+      "epoch": 1.306110543969666,
+      "grad_norm": 1.138240933418274,
+      "learning_rate": 1.7878605769230768e-05,
+      "loss": 0.8356,
+      "step": 2239
+    },
+    {
+      "epoch": 1.3066938894560303,
+      "grad_norm": 1.424336552619934,
+      "learning_rate": 1.7863581730769233e-05,
+      "loss": 0.7738,
+      "step": 2240
+    },
+    {
+      "epoch": 1.3072772349423947,
+      "grad_norm": 1.2529820203781128,
+      "learning_rate": 1.7848557692307695e-05,
+      "loss": 0.6549,
+      "step": 2241
+    },
+    {
+      "epoch": 1.3078605804287589,
+      "grad_norm": 1.7925548553466797,
+      "learning_rate": 1.7833533653846153e-05,
+      "loss": 0.648,
+      "step": 2242
+    },
+    {
+      "epoch": 1.3084439259151233,
+      "grad_norm": 1.3370999097824097,
+      "learning_rate": 1.7818509615384615e-05,
+      "loss": 0.8967,
+      "step": 2243
+    },
+    {
+      "epoch": 1.3090272714014874,
+      "grad_norm": 1.1147828102111816,
+      "learning_rate": 1.780348557692308e-05,
+      "loss": 0.8724,
+      "step": 2244
+    },
+    {
+      "epoch": 1.3096106168878519,
+      "grad_norm": 1.2542976140975952,
+      "learning_rate": 1.778846153846154e-05,
+      "loss": 0.8384,
+      "step": 2245
+    },
+    {
+      "epoch": 1.3101939623742163,
+      "grad_norm": 1.251609444618225,
+      "learning_rate": 1.77734375e-05,
+      "loss": 0.8426,
+      "step": 2246
+    },
+    {
+      "epoch": 1.3107773078605804,
+      "grad_norm": 1.2808489799499512,
+      "learning_rate": 1.7758413461538463e-05,
+      "loss": 0.7978,
+      "step": 2247
+    },
+    {
+      "epoch": 1.3113606533469446,
+      "grad_norm": 1.2761458158493042,
+      "learning_rate": 1.774338942307692e-05,
+      "loss": 0.9413,
+      "step": 2248
+    },
+    {
+      "epoch": 1.311943998833309,
+      "grad_norm": 1.263123631477356,
+      "learning_rate": 1.7728365384615387e-05,
+      "loss": 0.7363,
+      "step": 2249
+    },
+    {
+      "epoch": 1.3125273443196734,
+      "grad_norm": 1.2832683324813843,
+      "learning_rate": 1.771334134615385e-05,
+      "loss": 0.5699,
+      "step": 2250
+    },
+    {
+      "epoch": 1.3131106898060376,
+      "grad_norm": 1.2383707761764526,
+      "learning_rate": 1.7698317307692307e-05,
+      "loss": 0.9112,
+      "step": 2251
+    },
+    {
+      "epoch": 1.3136940352924018,
+      "grad_norm": 1.2074283361434937,
+      "learning_rate": 1.768329326923077e-05,
+      "loss": 0.7454,
+      "step": 2252
+    },
+    {
+      "epoch": 1.3142773807787662,
+      "grad_norm": 1.1046134233474731,
+      "learning_rate": 1.7668269230769234e-05,
+      "loss": 0.9412,
+      "step": 2253
+    },
+    {
+      "epoch": 1.3148607262651306,
+      "grad_norm": 1.1459115743637085,
+      "learning_rate": 1.7653245192307693e-05,
+      "loss": 0.7046,
+      "step": 2254
+    },
+    {
+      "epoch": 1.3154440717514948,
+      "grad_norm": 1.2635064125061035,
+      "learning_rate": 1.7638221153846155e-05,
+      "loss": 0.8815,
+      "step": 2255
+    },
+    {
+      "epoch": 1.316027417237859,
+      "grad_norm": 1.1063247919082642,
+      "learning_rate": 1.7623197115384616e-05,
+      "loss": 0.8305,
+      "step": 2256
+    },
+    {
+      "epoch": 1.3166107627242234,
+      "grad_norm": 2.1608684062957764,
+      "learning_rate": 1.7608173076923078e-05,
+      "loss": 0.9767,
+      "step": 2257
+    },
+    {
+      "epoch": 1.3171941082105878,
+      "grad_norm": 1.1712557077407837,
+      "learning_rate": 1.759314903846154e-05,
+      "loss": 1.0557,
+      "step": 2258
+    },
+    {
+      "epoch": 1.317777453696952,
+      "grad_norm": 1.1504344940185547,
+      "learning_rate": 1.7578125000000002e-05,
+      "loss": 1.0544,
+      "step": 2259
+    },
+    {
+      "epoch": 1.3183607991833164,
+      "grad_norm": 1.2168277502059937,
+      "learning_rate": 1.756310096153846e-05,
+      "loss": 0.8735,
+      "step": 2260
+    },
+    {
+      "epoch": 1.3189441446696806,
+      "grad_norm": 1.2982547283172607,
+      "learning_rate": 1.7548076923076922e-05,
+      "loss": 0.9633,
+      "step": 2261
+    },
+    {
+      "epoch": 1.319527490156045,
+      "grad_norm": 1.2747186422348022,
+      "learning_rate": 1.7533052884615388e-05,
+      "loss": 0.6536,
+      "step": 2262
+    },
+    {
+      "epoch": 1.3201108356424092,
+      "grad_norm": 1.1642917394638062,
+      "learning_rate": 1.7518028846153846e-05,
+      "loss": 0.7892,
+      "step": 2263
+    },
+    {
+      "epoch": 1.3206941811287736,
+      "grad_norm": 1.22393798828125,
+      "learning_rate": 1.7503004807692308e-05,
+      "loss": 0.8314,
+      "step": 2264
+    },
+    {
+      "epoch": 1.3212775266151378,
+      "grad_norm": 1.3126612901687622,
+      "learning_rate": 1.748798076923077e-05,
+      "loss": 0.695,
+      "step": 2265
+    },
+    {
+      "epoch": 1.3218608721015022,
+      "grad_norm": 1.1996201276779175,
+      "learning_rate": 1.7472956730769232e-05,
+      "loss": 0.7914,
+      "step": 2266
+    },
+    {
+      "epoch": 1.3224442175878663,
+      "grad_norm": 1.2702114582061768,
+      "learning_rate": 1.7457932692307694e-05,
+      "loss": 0.835,
+      "step": 2267
+    },
+    {
+      "epoch": 1.3230275630742308,
+      "grad_norm": 1.1210219860076904,
+      "learning_rate": 1.7442908653846156e-05,
+      "loss": 0.6297,
+      "step": 2268
+    },
+    {
+      "epoch": 1.323610908560595,
+      "grad_norm": 1.0922166109085083,
+      "learning_rate": 1.7427884615384614e-05,
+      "loss": 0.735,
+      "step": 2269
+    },
+    {
+      "epoch": 1.3241942540469593,
+      "grad_norm": 1.3747941255569458,
+      "learning_rate": 1.741286057692308e-05,
+      "loss": 0.9677,
+      "step": 2270
+    },
+    {
+      "epoch": 1.3247775995333235,
+      "grad_norm": 1.3291850090026855,
+      "learning_rate": 1.739783653846154e-05,
+      "loss": 0.9007,
+      "step": 2271
+    },
+    {
+      "epoch": 1.325360945019688,
+      "grad_norm": 1.1687302589416504,
+      "learning_rate": 1.73828125e-05,
+      "loss": 0.7095,
+      "step": 2272
+    },
+    {
+      "epoch": 1.3259442905060523,
+      "grad_norm": 1.1394537687301636,
+      "learning_rate": 1.7367788461538462e-05,
+      "loss": 0.6055,
+      "step": 2273
+    },
+    {
+      "epoch": 1.3265276359924165,
+      "grad_norm": 1.1962412595748901,
+      "learning_rate": 1.7352764423076924e-05,
+      "loss": 0.7964,
+      "step": 2274
+    },
+    {
+      "epoch": 1.3271109814787807,
+      "grad_norm": 1.5585336685180664,
+      "learning_rate": 1.7337740384615386e-05,
+      "loss": 0.725,
+      "step": 2275
+    },
+    {
+      "epoch": 1.327694326965145,
+      "grad_norm": 1.1137040853500366,
+      "learning_rate": 1.7322716346153847e-05,
+      "loss": 0.8106,
+      "step": 2276
+    },
+    {
+      "epoch": 1.3282776724515095,
+      "grad_norm": 1.2830206155776978,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.8546,
+      "step": 2277
+    },
+    {
+      "epoch": 1.3288610179378737,
+      "grad_norm": 1.1548091173171997,
+      "learning_rate": 1.7292668269230768e-05,
+      "loss": 0.8009,
+      "step": 2278
+    },
+    {
+      "epoch": 1.3294443634242379,
+      "grad_norm": 1.5664095878601074,
+      "learning_rate": 1.7277644230769233e-05,
+      "loss": 0.7502,
+      "step": 2279
+    },
+    {
+      "epoch": 1.3300277089106023,
+      "grad_norm": 1.4026076793670654,
+      "learning_rate": 1.7262620192307695e-05,
+      "loss": 0.886,
+      "step": 2280
+    },
+    {
+      "epoch": 1.3306110543969667,
+      "grad_norm": 1.2917288541793823,
+      "learning_rate": 1.7247596153846153e-05,
+      "loss": 0.7834,
+      "step": 2281
+    },
+    {
+      "epoch": 1.3311943998833309,
+      "grad_norm": 1.4439992904663086,
+      "learning_rate": 1.7232572115384615e-05,
+      "loss": 0.8932,
+      "step": 2282
+    },
+    {
+      "epoch": 1.331777745369695,
+      "grad_norm": 1.2439894676208496,
+      "learning_rate": 1.721754807692308e-05,
+      "loss": 0.7965,
+      "step": 2283
+    },
+    {
+      "epoch": 1.3323610908560595,
+      "grad_norm": 1.2655996084213257,
+      "learning_rate": 1.720252403846154e-05,
+      "loss": 0.7999,
+      "step": 2284
+    },
+    {
+      "epoch": 1.3329444363424239,
+      "grad_norm": 1.1132214069366455,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.8086,
+      "step": 2285
+    },
+    {
+      "epoch": 1.333527781828788,
+      "grad_norm": 1.3198105096817017,
+      "learning_rate": 1.7172475961538463e-05,
+      "loss": 0.824,
+      "step": 2286
+    },
+    {
+      "epoch": 1.3341111273151525,
+      "grad_norm": 1.1884894371032715,
+      "learning_rate": 1.715745192307692e-05,
+      "loss": 0.7186,
+      "step": 2287
+    },
+    {
+      "epoch": 1.3346944728015167,
+      "grad_norm": 1.179015040397644,
+      "learning_rate": 1.7142427884615387e-05,
+      "loss": 0.9886,
+      "step": 2288
+    },
+    {
+      "epoch": 1.335277818287881,
+      "grad_norm": 1.1946966648101807,
+      "learning_rate": 1.712740384615385e-05,
+      "loss": 0.7248,
+      "step": 2289
+    },
+    {
+      "epoch": 1.3358611637742452,
+      "grad_norm": 1.2960485219955444,
+      "learning_rate": 1.7112379807692307e-05,
+      "loss": 0.8273,
+      "step": 2290
+    },
+    {
+      "epoch": 1.3364445092606096,
+      "grad_norm": 1.2853033542633057,
+      "learning_rate": 1.709735576923077e-05,
+      "loss": 0.8596,
+      "step": 2291
+    },
+    {
+      "epoch": 1.3370278547469738,
+      "grad_norm": 1.0934782028198242,
+      "learning_rate": 1.7082331730769234e-05,
+      "loss": 0.71,
+      "step": 2292
+    },
+    {
+      "epoch": 1.3376112002333382,
+      "grad_norm": 1.2531418800354004,
+      "learning_rate": 1.7067307692307693e-05,
+      "loss": 0.8957,
+      "step": 2293
+    },
+    {
+      "epoch": 1.3381945457197024,
+      "grad_norm": 1.389156460762024,
+      "learning_rate": 1.7052283653846155e-05,
+      "loss": 0.8757,
+      "step": 2294
+    },
+    {
+      "epoch": 1.3387778912060668,
+      "grad_norm": 1.4648810625076294,
+      "learning_rate": 1.7037259615384617e-05,
+      "loss": 0.9267,
+      "step": 2295
+    },
+    {
+      "epoch": 1.339361236692431,
+      "grad_norm": 1.110659122467041,
+      "learning_rate": 1.702223557692308e-05,
+      "loss": 0.8775,
+      "step": 2296
+    },
+    {
+      "epoch": 1.3399445821787954,
+      "grad_norm": 1.2895351648330688,
+      "learning_rate": 1.700721153846154e-05,
+      "loss": 0.7839,
+      "step": 2297
+    },
+    {
+      "epoch": 1.3405279276651596,
+      "grad_norm": 1.207476258277893,
+      "learning_rate": 1.6992187500000002e-05,
+      "loss": 0.7952,
+      "step": 2298
+    },
+    {
+      "epoch": 1.341111273151524,
+      "grad_norm": 1.142844557762146,
+      "learning_rate": 1.697716346153846e-05,
+      "loss": 0.7482,
+      "step": 2299
+    },
+    {
+      "epoch": 1.3416946186378884,
+      "grad_norm": 1.2515149116516113,
+      "learning_rate": 1.6962139423076923e-05,
+      "loss": 0.9569,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3422779641242526,
+      "grad_norm": 1.090381145477295,
+      "learning_rate": 1.6947115384615388e-05,
+      "loss": 0.8499,
+      "step": 2301
+    },
+    {
+      "epoch": 1.3428613096106168,
+      "grad_norm": 1.2688581943511963,
+      "learning_rate": 1.6932091346153846e-05,
+      "loss": 0.6979,
+      "step": 2302
+    },
+    {
+      "epoch": 1.3434446550969812,
+      "grad_norm": 1.2520463466644287,
+      "learning_rate": 1.6917067307692308e-05,
+      "loss": 0.7364,
+      "step": 2303
+    },
+    {
+      "epoch": 1.3440280005833456,
+      "grad_norm": 1.1990320682525635,
+      "learning_rate": 1.690204326923077e-05,
+      "loss": 0.652,
+      "step": 2304
+    },
+    {
+      "epoch": 1.3446113460697098,
+      "grad_norm": 1.142889380455017,
+      "learning_rate": 1.6887019230769232e-05,
+      "loss": 0.9194,
+      "step": 2305
+    },
+    {
+      "epoch": 1.345194691556074,
+      "grad_norm": 1.1990599632263184,
+      "learning_rate": 1.6871995192307694e-05,
+      "loss": 0.7939,
+      "step": 2306
+    },
+    {
+      "epoch": 1.3457780370424384,
+      "grad_norm": 1.3533287048339844,
+      "learning_rate": 1.6856971153846156e-05,
+      "loss": 0.7524,
+      "step": 2307
+    },
+    {
+      "epoch": 1.3463613825288028,
+      "grad_norm": 1.157096028327942,
+      "learning_rate": 1.6841947115384614e-05,
+      "loss": 0.5834,
+      "step": 2308
+    },
+    {
+      "epoch": 1.346944728015167,
+      "grad_norm": 1.2179913520812988,
+      "learning_rate": 1.682692307692308e-05,
+      "loss": 0.9018,
+      "step": 2309
+    },
+    {
+      "epoch": 1.3475280735015314,
+      "grad_norm": 1.2479956150054932,
+      "learning_rate": 1.681189903846154e-05,
+      "loss": 0.7633,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3481114189878955,
+      "grad_norm": 1.1520054340362549,
+      "learning_rate": 1.6796875e-05,
+      "loss": 0.7101,
+      "step": 2311
+    },
+    {
+      "epoch": 1.34869476447426,
+      "grad_norm": 1.21807861328125,
+      "learning_rate": 1.6781850961538462e-05,
+      "loss": 0.7766,
+      "step": 2312
+    },
+    {
+      "epoch": 1.3492781099606241,
+      "grad_norm": 1.4313595294952393,
+      "learning_rate": 1.6766826923076924e-05,
+      "loss": 0.6487,
+      "step": 2313
+    },
+    {
+      "epoch": 1.3498614554469885,
+      "grad_norm": 1.3884174823760986,
+      "learning_rate": 1.6751802884615386e-05,
+      "loss": 0.7646,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3504448009333527,
+      "grad_norm": 1.1427439451217651,
+      "learning_rate": 1.6736778846153848e-05,
+      "loss": 0.8483,
+      "step": 2315
+    },
+    {
+      "epoch": 1.3510281464197171,
+      "grad_norm": 1.2759617567062378,
+      "learning_rate": 1.672175480769231e-05,
+      "loss": 0.7245,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3516114919060813,
+      "grad_norm": 1.3233177661895752,
+      "learning_rate": 1.6706730769230768e-05,
+      "loss": 1.0692,
+      "step": 2317
+    },
+    {
+      "epoch": 1.3521948373924457,
+      "grad_norm": 1.3573335409164429,
+      "learning_rate": 1.6691706730769233e-05,
+      "loss": 0.9137,
+      "step": 2318
+    },
+    {
+      "epoch": 1.35277818287881,
+      "grad_norm": 1.0537357330322266,
+      "learning_rate": 1.6676682692307695e-05,
+      "loss": 0.7244,
+      "step": 2319
+    },
+    {
+      "epoch": 1.3533615283651743,
+      "grad_norm": 1.126329779624939,
+      "learning_rate": 1.6661658653846154e-05,
+      "loss": 0.7374,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3539448738515385,
+      "grad_norm": 1.3769999742507935,
+      "learning_rate": 1.6646634615384616e-05,
+      "loss": 0.7223,
+      "step": 2321
+    },
+    {
+      "epoch": 1.354528219337903,
+      "grad_norm": 1.307424545288086,
+      "learning_rate": 1.6631610576923077e-05,
+      "loss": 0.7789,
+      "step": 2322
+    },
+    {
+      "epoch": 1.355111564824267,
+      "grad_norm": 1.2053377628326416,
+      "learning_rate": 1.661658653846154e-05,
+      "loss": 0.9478,
+      "step": 2323
+    },
+    {
+      "epoch": 1.3556949103106315,
+      "grad_norm": 1.478121280670166,
+      "learning_rate": 1.66015625e-05,
+      "loss": 0.7229,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3562782557969957,
+      "grad_norm": 1.2465181350708008,
+      "learning_rate": 1.6586538461538463e-05,
+      "loss": 0.8424,
+      "step": 2325
+    },
+    {
+      "epoch": 1.35686160128336,
+      "grad_norm": 1.366040825843811,
+      "learning_rate": 1.657151442307692e-05,
+      "loss": 0.8107,
+      "step": 2326
+    },
+    {
+      "epoch": 1.3574449467697245,
+      "grad_norm": 1.2248352766036987,
+      "learning_rate": 1.6556490384615387e-05,
+      "loss": 0.7464,
+      "step": 2327
+    },
+    {
+      "epoch": 1.3580282922560887,
+      "grad_norm": 1.4650728702545166,
+      "learning_rate": 1.654146634615385e-05,
+      "loss": 0.7499,
+      "step": 2328
+    },
+    {
+      "epoch": 1.3586116377424529,
+      "grad_norm": 1.5269290208816528,
+      "learning_rate": 1.6526442307692307e-05,
+      "loss": 0.8989,
+      "step": 2329
+    },
+    {
+      "epoch": 1.3591949832288173,
+      "grad_norm": 1.1337950229644775,
+      "learning_rate": 1.651141826923077e-05,
+      "loss": 0.6656,
+      "step": 2330
+    },
+    {
+      "epoch": 1.3597783287151817,
+      "grad_norm": 1.1645163297653198,
+      "learning_rate": 1.649639423076923e-05,
+      "loss": 0.7783,
+      "step": 2331
+    },
+    {
+      "epoch": 1.3603616742015459,
+      "grad_norm": 1.1996797323226929,
+      "learning_rate": 1.6481370192307693e-05,
+      "loss": 0.9066,
+      "step": 2332
+    },
+    {
+      "epoch": 1.36094501968791,
+      "grad_norm": 1.2484889030456543,
+      "learning_rate": 1.6466346153846155e-05,
+      "loss": 0.761,
+      "step": 2333
+    },
+    {
+      "epoch": 1.3615283651742744,
+      "grad_norm": 1.1876178979873657,
+      "learning_rate": 1.6451322115384617e-05,
+      "loss": 0.7001,
+      "step": 2334
+    },
+    {
+      "epoch": 1.3621117106606389,
+      "grad_norm": 1.1717162132263184,
+      "learning_rate": 1.643629807692308e-05,
+      "loss": 0.7374,
+      "step": 2335
+    },
+    {
+      "epoch": 1.362695056147003,
+      "grad_norm": 1.0891270637512207,
+      "learning_rate": 1.642127403846154e-05,
+      "loss": 0.9183,
+      "step": 2336
+    },
+    {
+      "epoch": 1.3632784016333674,
+      "grad_norm": 1.3006951808929443,
+      "learning_rate": 1.6406250000000002e-05,
+      "loss": 0.7246,
+      "step": 2337
+    },
+    {
+      "epoch": 1.3638617471197316,
+      "grad_norm": 1.3394348621368408,
+      "learning_rate": 1.639122596153846e-05,
+      "loss": 0.7199,
+      "step": 2338
+    },
+    {
+      "epoch": 1.364445092606096,
+      "grad_norm": 1.2449558973312378,
+      "learning_rate": 1.6376201923076923e-05,
+      "loss": 0.8916,
+      "step": 2339
+    },
+    {
+      "epoch": 1.3650284380924602,
+      "grad_norm": 1.0279120206832886,
+      "learning_rate": 1.6361177884615385e-05,
+      "loss": 0.9012,
+      "step": 2340
+    },
+    {
+      "epoch": 1.3656117835788246,
+      "grad_norm": 1.218778133392334,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 0.7377,
+      "step": 2341
+    },
+    {
+      "epoch": 1.3661951290651888,
+      "grad_norm": 1.2190182209014893,
+      "learning_rate": 1.633112980769231e-05,
+      "loss": 0.963,
+      "step": 2342
+    },
+    {
+      "epoch": 1.3667784745515532,
+      "grad_norm": 1.2658116817474365,
+      "learning_rate": 1.631610576923077e-05,
+      "loss": 0.8298,
+      "step": 2343
+    },
+    {
+      "epoch": 1.3673618200379174,
+      "grad_norm": 1.2883538007736206,
+      "learning_rate": 1.6301081730769232e-05,
+      "loss": 0.8406,
+      "step": 2344
+    },
+    {
+      "epoch": 1.3679451655242818,
+      "grad_norm": 1.2415398359298706,
+      "learning_rate": 1.6286057692307694e-05,
+      "loss": 0.7271,
+      "step": 2345
+    },
+    {
+      "epoch": 1.368528511010646,
+      "grad_norm": 1.2686741352081299,
+      "learning_rate": 1.6271033653846156e-05,
+      "loss": 0.8293,
+      "step": 2346
+    },
+    {
+      "epoch": 1.3691118564970104,
+      "grad_norm": 1.3187519311904907,
+      "learning_rate": 1.6256009615384614e-05,
+      "loss": 0.7959,
+      "step": 2347
+    },
+    {
+      "epoch": 1.3696952019833746,
+      "grad_norm": 1.2809123992919922,
+      "learning_rate": 1.624098557692308e-05,
+      "loss": 0.657,
+      "step": 2348
+    },
+    {
+      "epoch": 1.370278547469739,
+      "grad_norm": 1.2265044450759888,
+      "learning_rate": 1.6225961538461538e-05,
+      "loss": 0.8477,
+      "step": 2349
+    },
+    {
+      "epoch": 1.3708618929561034,
+      "grad_norm": 1.0081802606582642,
+      "learning_rate": 1.62109375e-05,
+      "loss": 0.8687,
+      "step": 2350
+    },
+    {
+      "epoch": 1.3714452384424676,
+      "grad_norm": 1.3187026977539062,
+      "learning_rate": 1.6195913461538462e-05,
+      "loss": 0.9389,
+      "step": 2351
+    },
+    {
+      "epoch": 1.3720285839288318,
+      "grad_norm": 1.283357858657837,
+      "learning_rate": 1.6180889423076924e-05,
+      "loss": 0.7913,
+      "step": 2352
+    },
+    {
+      "epoch": 1.3726119294151962,
+      "grad_norm": 1.126367211341858,
+      "learning_rate": 1.6165865384615386e-05,
+      "loss": 0.7502,
+      "step": 2353
+    },
+    {
+      "epoch": 1.3731952749015606,
+      "grad_norm": 1.1398643255233765,
+      "learning_rate": 1.6150841346153848e-05,
+      "loss": 0.8787,
+      "step": 2354
+    },
+    {
+      "epoch": 1.3737786203879248,
+      "grad_norm": 1.1188660860061646,
+      "learning_rate": 1.613581730769231e-05,
+      "loss": 0.5851,
+      "step": 2355
+    },
+    {
+      "epoch": 1.374361965874289,
+      "grad_norm": 1.200810194015503,
+      "learning_rate": 1.6120793269230768e-05,
+      "loss": 0.7128,
+      "step": 2356
+    },
+    {
+      "epoch": 1.3749453113606533,
+      "grad_norm": 1.1014255285263062,
+      "learning_rate": 1.6105769230769233e-05,
+      "loss": 0.8273,
+      "step": 2357
+    },
+    {
+      "epoch": 1.3755286568470177,
+      "grad_norm": 1.1966328620910645,
+      "learning_rate": 1.6090745192307692e-05,
+      "loss": 0.9342,
+      "step": 2358
+    },
+    {
+      "epoch": 1.376112002333382,
+      "grad_norm": 1.3458975553512573,
+      "learning_rate": 1.6075721153846154e-05,
+      "loss": 0.8317,
+      "step": 2359
+    },
+    {
+      "epoch": 1.3766953478197461,
+      "grad_norm": 1.364256501197815,
+      "learning_rate": 1.6060697115384616e-05,
+      "loss": 0.7107,
+      "step": 2360
+    },
+    {
+      "epoch": 1.3772786933061105,
+      "grad_norm": 1.2403984069824219,
+      "learning_rate": 1.6045673076923078e-05,
+      "loss": 0.6139,
+      "step": 2361
+    },
+    {
+      "epoch": 1.377862038792475,
+      "grad_norm": 1.3325952291488647,
+      "learning_rate": 1.603064903846154e-05,
+      "loss": 0.8053,
+      "step": 2362
+    },
+    {
+      "epoch": 1.3784453842788391,
+      "grad_norm": 1.0815925598144531,
+      "learning_rate": 1.6015625e-05,
+      "loss": 0.8323,
+      "step": 2363
+    },
+    {
+      "epoch": 1.3790287297652035,
+      "grad_norm": 1.0733479261398315,
+      "learning_rate": 1.6000600961538463e-05,
+      "loss": 0.7921,
+      "step": 2364
+    },
+    {
+      "epoch": 1.3796120752515677,
+      "grad_norm": 1.1240204572677612,
+      "learning_rate": 1.5985576923076922e-05,
+      "loss": 0.861,
+      "step": 2365
+    },
+    {
+      "epoch": 1.380195420737932,
+      "grad_norm": 1.2654609680175781,
+      "learning_rate": 1.5970552884615387e-05,
+      "loss": 0.6925,
+      "step": 2366
+    },
+    {
+      "epoch": 1.3807787662242963,
+      "grad_norm": 1.7558287382125854,
+      "learning_rate": 1.5955528846153846e-05,
+      "loss": 0.7423,
+      "step": 2367
+    },
+    {
+      "epoch": 1.3813621117106607,
+      "grad_norm": 1.3726961612701416,
+      "learning_rate": 1.5940504807692307e-05,
+      "loss": 0.7108,
+      "step": 2368
+    },
+    {
+      "epoch": 1.3819454571970249,
+      "grad_norm": 1.279613971710205,
+      "learning_rate": 1.592548076923077e-05,
+      "loss": 0.8721,
+      "step": 2369
+    },
+    {
+      "epoch": 1.3825288026833893,
+      "grad_norm": 1.324630856513977,
+      "learning_rate": 1.591045673076923e-05,
+      "loss": 0.6835,
+      "step": 2370
+    },
+    {
+      "epoch": 1.3831121481697535,
+      "grad_norm": 1.2332587242126465,
+      "learning_rate": 1.5895432692307693e-05,
+      "loss": 0.8593,
+      "step": 2371
+    },
+    {
+      "epoch": 1.3836954936561179,
+      "grad_norm": 1.0641486644744873,
+      "learning_rate": 1.5880408653846155e-05,
+      "loss": 0.8916,
+      "step": 2372
+    },
+    {
+      "epoch": 1.384278839142482,
+      "grad_norm": 1.2237672805786133,
+      "learning_rate": 1.5865384615384617e-05,
+      "loss": 0.8585,
+      "step": 2373
+    },
+    {
+      "epoch": 1.3848621846288465,
+      "grad_norm": 1.3848849534988403,
+      "learning_rate": 1.585036057692308e-05,
+      "loss": 0.8661,
+      "step": 2374
+    },
+    {
+      "epoch": 1.3854455301152107,
+      "grad_norm": 1.1622084379196167,
+      "learning_rate": 1.583533653846154e-05,
+      "loss": 0.8062,
+      "step": 2375
+    },
+    {
+      "epoch": 1.386028875601575,
+      "grad_norm": 1.1375772953033447,
+      "learning_rate": 1.58203125e-05,
+      "loss": 0.8455,
+      "step": 2376
+    },
+    {
+      "epoch": 1.3866122210879395,
+      "grad_norm": 1.1407866477966309,
+      "learning_rate": 1.580528846153846e-05,
+      "loss": 0.7599,
+      "step": 2377
+    },
+    {
+      "epoch": 1.3871955665743037,
+      "grad_norm": 1.3291012048721313,
+      "learning_rate": 1.5790264423076923e-05,
+      "loss": 0.7574,
+      "step": 2378
+    },
+    {
+      "epoch": 1.3877789120606678,
+      "grad_norm": 1.0793458223342896,
+      "learning_rate": 1.5775240384615385e-05,
+      "loss": 0.8483,
+      "step": 2379
+    },
+    {
+      "epoch": 1.3883622575470322,
+      "grad_norm": 1.419547438621521,
+      "learning_rate": 1.5760216346153847e-05,
+      "loss": 0.8905,
+      "step": 2380
+    },
+    {
+      "epoch": 1.3889456030333966,
+      "grad_norm": 1.1597005128860474,
+      "learning_rate": 1.574519230769231e-05,
+      "loss": 0.7328,
+      "step": 2381
+    },
+    {
+      "epoch": 1.3895289485197608,
+      "grad_norm": 1.3240809440612793,
+      "learning_rate": 1.573016826923077e-05,
+      "loss": 0.9296,
+      "step": 2382
+    },
+    {
+      "epoch": 1.390112294006125,
+      "grad_norm": 1.1506272554397583,
+      "learning_rate": 1.5715144230769232e-05,
+      "loss": 0.693,
+      "step": 2383
+    },
+    {
+      "epoch": 1.3906956394924894,
+      "grad_norm": 1.0295346975326538,
+      "learning_rate": 1.5700120192307694e-05,
+      "loss": 0.8175,
+      "step": 2384
+    },
+    {
+      "epoch": 1.3912789849788538,
+      "grad_norm": 1.1327847242355347,
+      "learning_rate": 1.5685096153846153e-05,
+      "loss": 0.8946,
+      "step": 2385
+    },
+    {
+      "epoch": 1.391862330465218,
+      "grad_norm": 1.3098478317260742,
+      "learning_rate": 1.5670072115384615e-05,
+      "loss": 0.7245,
+      "step": 2386
+    },
+    {
+      "epoch": 1.3924456759515822,
+      "grad_norm": 1.2984569072723389,
+      "learning_rate": 1.565504807692308e-05,
+      "loss": 1.0087,
+      "step": 2387
+    },
+    {
+      "epoch": 1.3930290214379466,
+      "grad_norm": 1.20249605178833,
+      "learning_rate": 1.564002403846154e-05,
+      "loss": 0.9404,
+      "step": 2388
+    },
+    {
+      "epoch": 1.393612366924311,
+      "grad_norm": 1.116599678993225,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.8229,
+      "step": 2389
+    },
+    {
+      "epoch": 1.3941957124106752,
+      "grad_norm": 1.4463697671890259,
+      "learning_rate": 1.5609975961538462e-05,
+      "loss": 0.8928,
+      "step": 2390
+    },
+    {
+      "epoch": 1.3947790578970396,
+      "grad_norm": 1.3660231828689575,
+      "learning_rate": 1.5594951923076924e-05,
+      "loss": 0.7133,
+      "step": 2391
+    },
+    {
+      "epoch": 1.3953624033834038,
+      "grad_norm": 1.2797685861587524,
+      "learning_rate": 1.5579927884615386e-05,
+      "loss": 0.8965,
+      "step": 2392
+    },
+    {
+      "epoch": 1.3959457488697682,
+      "grad_norm": 1.3361741304397583,
+      "learning_rate": 1.5564903846153848e-05,
+      "loss": 0.7193,
+      "step": 2393
+    },
+    {
+      "epoch": 1.3965290943561324,
+      "grad_norm": 1.1247235536575317,
+      "learning_rate": 1.5549879807692306e-05,
+      "loss": 0.7304,
+      "step": 2394
+    },
+    {
+      "epoch": 1.3971124398424968,
+      "grad_norm": 1.4382926225662231,
+      "learning_rate": 1.5534855769230768e-05,
+      "loss": 0.8633,
+      "step": 2395
+    },
+    {
+      "epoch": 1.397695785328861,
+      "grad_norm": 1.2390516996383667,
+      "learning_rate": 1.5519831730769234e-05,
+      "loss": 0.8239,
+      "step": 2396
+    },
+    {
+      "epoch": 1.3982791308152254,
+      "grad_norm": 1.3460718393325806,
+      "learning_rate": 1.5504807692307692e-05,
+      "loss": 1.0886,
+      "step": 2397
+    },
+    {
+      "epoch": 1.3988624763015896,
+      "grad_norm": 1.125791311264038,
+      "learning_rate": 1.5489783653846154e-05,
+      "loss": 0.8253,
+      "step": 2398
+    },
+    {
+      "epoch": 1.399445821787954,
+      "grad_norm": 1.2912181615829468,
+      "learning_rate": 1.5474759615384616e-05,
+      "loss": 0.7138,
+      "step": 2399
+    },
+    {
+      "epoch": 1.4000291672743181,
+      "grad_norm": 1.1510553359985352,
+      "learning_rate": 1.5459735576923078e-05,
+      "loss": 0.932,
+      "step": 2400
+    },
+    {
+      "epoch": 1.4000291672743181,
+      "eval_loss_squad": 0.8576014773617499,
+      "eval_perplexity": 8.10681335191388,
+      "eval_perplexity_reconstruct": 1.909424481734322,
+      "step": 2400
+    },
+    {
+      "epoch": 1.4006125127606825,
+      "grad_norm": 1.429721474647522,
+      "learning_rate": 1.544471153846154e-05,
+      "loss": 0.7134,
+      "step": 2401
+    },
+    {
+      "epoch": 1.4011958582470467,
+      "grad_norm": 1.3556239604949951,
+      "learning_rate": 1.54296875e-05,
+      "loss": 0.8164,
+      "step": 2402
+    },
+    {
+      "epoch": 1.4017792037334111,
+      "grad_norm": 1.2837475538253784,
+      "learning_rate": 1.541466346153846e-05,
+      "loss": 0.678,
+      "step": 2403
+    },
+    {
+      "epoch": 1.4023625492197755,
+      "grad_norm": 1.1232627630233765,
+      "learning_rate": 1.5399639423076922e-05,
+      "loss": 0.8262,
+      "step": 2404
+    },
+    {
+      "epoch": 1.4029458947061397,
+      "grad_norm": 1.3286436796188354,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.6756,
+      "step": 2405
+    },
+    {
+      "epoch": 1.403529240192504,
+      "grad_norm": 1.2372446060180664,
+      "learning_rate": 1.5369591346153846e-05,
+      "loss": 0.7997,
+      "step": 2406
+    },
+    {
+      "epoch": 1.4041125856788683,
+      "grad_norm": 1.1915537118911743,
+      "learning_rate": 1.5354567307692308e-05,
+      "loss": 0.7118,
+      "step": 2407
+    },
+    {
+      "epoch": 1.4046959311652327,
+      "grad_norm": 1.241876244544983,
+      "learning_rate": 1.533954326923077e-05,
+      "loss": 0.7073,
+      "step": 2408
+    },
+    {
+      "epoch": 1.405279276651597,
+      "grad_norm": 1.4120349884033203,
+      "learning_rate": 1.532451923076923e-05,
+      "loss": 0.8801,
+      "step": 2409
+    },
+    {
+      "epoch": 1.405862622137961,
+      "grad_norm": 1.0920681953430176,
+      "learning_rate": 1.5309495192307693e-05,
+      "loss": 0.8118,
+      "step": 2410
+    },
+    {
+      "epoch": 1.4064459676243255,
+      "grad_norm": 1.3178508281707764,
+      "learning_rate": 1.5294471153846155e-05,
+      "loss": 0.8296,
+      "step": 2411
+    },
+    {
+      "epoch": 1.40702931311069,
+      "grad_norm": 1.307503342628479,
+      "learning_rate": 1.5279447115384614e-05,
+      "loss": 0.8255,
+      "step": 2412
+    },
+    {
+      "epoch": 1.407612658597054,
+      "grad_norm": 1.4173907041549683,
+      "learning_rate": 1.526442307692308e-05,
+      "loss": 0.7455,
+      "step": 2413
+    },
+    {
+      "epoch": 1.4081960040834183,
+      "grad_norm": 1.2470276355743408,
+      "learning_rate": 1.5249399038461539e-05,
+      "loss": 0.7058,
+      "step": 2414
+    },
+    {
+      "epoch": 1.4087793495697827,
+      "grad_norm": 1.1090408563613892,
+      "learning_rate": 1.5234375000000001e-05,
+      "loss": 0.7262,
+      "step": 2415
+    },
+    {
+      "epoch": 1.409362695056147,
+      "grad_norm": 1.0376850366592407,
+      "learning_rate": 1.5219350961538461e-05,
+      "loss": 0.6755,
+      "step": 2416
+    },
+    {
+      "epoch": 1.4099460405425113,
+      "grad_norm": 1.1430914402008057,
+      "learning_rate": 1.5204326923076923e-05,
+      "loss": 0.7457,
+      "step": 2417
+    },
+    {
+      "epoch": 1.4105293860288757,
+      "grad_norm": 1.2482012510299683,
+      "learning_rate": 1.5189302884615387e-05,
+      "loss": 0.6143,
+      "step": 2418
+    },
+    {
+      "epoch": 1.4111127315152399,
+      "grad_norm": 1.3330813646316528,
+      "learning_rate": 1.5174278846153847e-05,
+      "loss": 0.6651,
+      "step": 2419
+    },
+    {
+      "epoch": 1.4116960770016043,
+      "grad_norm": 1.3887958526611328,
+      "learning_rate": 1.5159254807692309e-05,
+      "loss": 0.9114,
+      "step": 2420
+    },
+    {
+      "epoch": 1.4122794224879684,
+      "grad_norm": 1.328403353691101,
+      "learning_rate": 1.5144230769230769e-05,
+      "loss": 0.8242,
+      "step": 2421
+    },
+    {
+      "epoch": 1.4128627679743329,
+      "grad_norm": 1.185078740119934,
+      "learning_rate": 1.5129206730769233e-05,
+      "loss": 0.6619,
+      "step": 2422
+    },
+    {
+      "epoch": 1.413446113460697,
+      "grad_norm": 1.2361003160476685,
+      "learning_rate": 1.5114182692307693e-05,
+      "loss": 0.7416,
+      "step": 2423
+    },
+    {
+      "epoch": 1.4140294589470614,
+      "grad_norm": 1.1306205987930298,
+      "learning_rate": 1.5099158653846155e-05,
+      "loss": 0.8077,
+      "step": 2424
+    },
+    {
+      "epoch": 1.4146128044334256,
+      "grad_norm": 0.9908915162086487,
+      "learning_rate": 1.5084134615384615e-05,
+      "loss": 0.8779,
+      "step": 2425
+    },
+    {
+      "epoch": 1.41519614991979,
+      "grad_norm": 1.1640621423721313,
+      "learning_rate": 1.5069110576923078e-05,
+      "loss": 0.8623,
+      "step": 2426
+    },
+    {
+      "epoch": 1.4157794954061542,
+      "grad_norm": 2.4493496417999268,
+      "learning_rate": 1.505408653846154e-05,
+      "loss": 0.9976,
+      "step": 2427
+    },
+    {
+      "epoch": 1.4163628408925186,
+      "grad_norm": 1.1745401620864868,
+      "learning_rate": 1.50390625e-05,
+      "loss": 0.7252,
+      "step": 2428
+    },
+    {
+      "epoch": 1.4169461863788828,
+      "grad_norm": 1.2570523023605347,
+      "learning_rate": 1.5024038461538462e-05,
+      "loss": 0.7836,
+      "step": 2429
+    },
+    {
+      "epoch": 1.4175295318652472,
+      "grad_norm": 1.1713680028915405,
+      "learning_rate": 1.5009014423076923e-05,
+      "loss": 0.7318,
+      "step": 2430
+    },
+    {
+      "epoch": 1.4181128773516116,
+      "grad_norm": 1.224394679069519,
+      "learning_rate": 1.4993990384615386e-05,
+      "loss": 0.619,
+      "step": 2431
+    },
+    {
+      "epoch": 1.4186962228379758,
+      "grad_norm": 1.089404821395874,
+      "learning_rate": 1.4978966346153846e-05,
+      "loss": 0.712,
+      "step": 2432
+    },
+    {
+      "epoch": 1.41927956832434,
+      "grad_norm": 1.172507643699646,
+      "learning_rate": 1.4963942307692308e-05,
+      "loss": 0.6563,
+      "step": 2433
+    },
+    {
+      "epoch": 1.4198629138107044,
+      "grad_norm": 1.3817616701126099,
+      "learning_rate": 1.4948918269230768e-05,
+      "loss": 0.6572,
+      "step": 2434
+    },
+    {
+      "epoch": 1.4204462592970688,
+      "grad_norm": 1.1678885221481323,
+      "learning_rate": 1.4933894230769232e-05,
+      "loss": 0.9326,
+      "step": 2435
+    },
+    {
+      "epoch": 1.421029604783433,
+      "grad_norm": 1.2956814765930176,
+      "learning_rate": 1.4918870192307694e-05,
+      "loss": 0.8578,
+      "step": 2436
+    },
+    {
+      "epoch": 1.4216129502697972,
+      "grad_norm": 0.7640461325645447,
+      "learning_rate": 1.4903846153846154e-05,
+      "loss": 0.8301,
+      "step": 2437
+    },
+    {
+      "epoch": 1.4221962957561616,
+      "grad_norm": 1.1869699954986572,
+      "learning_rate": 1.4888822115384616e-05,
+      "loss": 0.8132,
+      "step": 2438
+    },
+    {
+      "epoch": 1.422779641242526,
+      "grad_norm": 1.1306724548339844,
+      "learning_rate": 1.487379807692308e-05,
+      "loss": 0.7028,
+      "step": 2439
+    },
+    {
+      "epoch": 1.4233629867288902,
+      "grad_norm": 1.2320770025253296,
+      "learning_rate": 1.485877403846154e-05,
+      "loss": 0.6797,
+      "step": 2440
+    },
+    {
+      "epoch": 1.4239463322152544,
+      "grad_norm": 1.3738675117492676,
+      "learning_rate": 1.484375e-05,
+      "loss": 0.752,
+      "step": 2441
+    },
+    {
+      "epoch": 1.4245296777016188,
+      "grad_norm": 1.2884392738342285,
+      "learning_rate": 1.4828725961538462e-05,
+      "loss": 0.7435,
+      "step": 2442
+    },
+    {
+      "epoch": 1.4251130231879832,
+      "grad_norm": 1.3447365760803223,
+      "learning_rate": 1.4813701923076922e-05,
+      "loss": 0.6959,
+      "step": 2443
+    },
+    {
+      "epoch": 1.4256963686743473,
+      "grad_norm": 1.270371913909912,
+      "learning_rate": 1.4798677884615386e-05,
+      "loss": 0.8662,
+      "step": 2444
+    },
+    {
+      "epoch": 1.4262797141607118,
+      "grad_norm": 1.3903402090072632,
+      "learning_rate": 1.4783653846153848e-05,
+      "loss": 0.7844,
+      "step": 2445
+    },
+    {
+      "epoch": 1.426863059647076,
+      "grad_norm": 1.2270903587341309,
+      "learning_rate": 1.4768629807692308e-05,
+      "loss": 0.6552,
+      "step": 2446
+    },
+    {
+      "epoch": 1.4274464051334403,
+      "grad_norm": 1.170554280281067,
+      "learning_rate": 1.475360576923077e-05,
+      "loss": 0.9123,
+      "step": 2447
+    },
+    {
+      "epoch": 1.4280297506198045,
+      "grad_norm": 0.9950125217437744,
+      "learning_rate": 1.4738581730769233e-05,
+      "loss": 0.7512,
+      "step": 2448
+    },
+    {
+      "epoch": 1.428613096106169,
+      "grad_norm": 1.3289663791656494,
+      "learning_rate": 1.4723557692307693e-05,
+      "loss": 0.7338,
+      "step": 2449
+    },
+    {
+      "epoch": 1.4291964415925331,
+      "grad_norm": 1.298034429550171,
+      "learning_rate": 1.4708533653846154e-05,
+      "loss": 0.7775,
+      "step": 2450
+    },
+    {
+      "epoch": 1.4297797870788975,
+      "grad_norm": 1.3974965810775757,
+      "learning_rate": 1.4693509615384615e-05,
+      "loss": 0.7119,
+      "step": 2451
+    },
+    {
+      "epoch": 1.4303631325652617,
+      "grad_norm": 1.3868650197982788,
+      "learning_rate": 1.4678485576923079e-05,
+      "loss": 0.8351,
+      "step": 2452
+    },
+    {
+      "epoch": 1.4309464780516261,
+      "grad_norm": 1.093425989151001,
+      "learning_rate": 1.466346153846154e-05,
+      "loss": 0.505,
+      "step": 2453
+    },
+    {
+      "epoch": 1.4315298235379903,
+      "grad_norm": 1.075796127319336,
+      "learning_rate": 1.4648437500000001e-05,
+      "loss": 1.0527,
+      "step": 2454
+    },
+    {
+      "epoch": 1.4321131690243547,
+      "grad_norm": 1.2513554096221924,
+      "learning_rate": 1.4633413461538461e-05,
+      "loss": 0.6941,
+      "step": 2455
+    },
+    {
+      "epoch": 1.4326965145107189,
+      "grad_norm": 1.2571288347244263,
+      "learning_rate": 1.4618389423076923e-05,
+      "loss": 0.8531,
+      "step": 2456
+    },
+    {
+      "epoch": 1.4332798599970833,
+      "grad_norm": 1.139385461807251,
+      "learning_rate": 1.4603365384615387e-05,
+      "loss": 0.648,
+      "step": 2457
+    },
+    {
+      "epoch": 1.4338632054834477,
+      "grad_norm": 1.1437668800354004,
+      "learning_rate": 1.4588341346153847e-05,
+      "loss": 0.7422,
+      "step": 2458
+    },
+    {
+      "epoch": 1.4344465509698119,
+      "grad_norm": 1.3354110717773438,
+      "learning_rate": 1.4573317307692307e-05,
+      "loss": 0.8899,
+      "step": 2459
+    },
+    {
+      "epoch": 1.435029896456176,
+      "grad_norm": 1.1407440900802612,
+      "learning_rate": 1.4558293269230769e-05,
+      "loss": 0.7084,
+      "step": 2460
+    },
+    {
+      "epoch": 1.4356132419425405,
+      "grad_norm": 1.2988560199737549,
+      "learning_rate": 1.4543269230769233e-05,
+      "loss": 0.803,
+      "step": 2461
+    },
+    {
+      "epoch": 1.4361965874289049,
+      "grad_norm": 1.2152442932128906,
+      "learning_rate": 1.4528245192307693e-05,
+      "loss": 0.5639,
+      "step": 2462
+    },
+    {
+      "epoch": 1.436779932915269,
+      "grad_norm": 1.1522806882858276,
+      "learning_rate": 1.4513221153846155e-05,
+      "loss": 0.8283,
+      "step": 2463
+    },
+    {
+      "epoch": 1.4373632784016332,
+      "grad_norm": 1.3547604084014893,
+      "learning_rate": 1.4498197115384615e-05,
+      "loss": 0.9728,
+      "step": 2464
+    },
+    {
+      "epoch": 1.4379466238879977,
+      "grad_norm": 1.0888975858688354,
+      "learning_rate": 1.4483173076923079e-05,
+      "loss": 0.9095,
+      "step": 2465
+    },
+    {
+      "epoch": 1.438529969374362,
+      "grad_norm": 1.2343162298202515,
+      "learning_rate": 1.446814903846154e-05,
+      "loss": 0.8971,
+      "step": 2466
+    },
+    {
+      "epoch": 1.4391133148607262,
+      "grad_norm": 1.2880823612213135,
+      "learning_rate": 1.4453125e-05,
+      "loss": 0.758,
+      "step": 2467
+    },
+    {
+      "epoch": 1.4396966603470904,
+      "grad_norm": 1.273984670639038,
+      "learning_rate": 1.443810096153846e-05,
+      "loss": 0.8011,
+      "step": 2468
+    },
+    {
+      "epoch": 1.4402800058334548,
+      "grad_norm": 1.221825122833252,
+      "learning_rate": 1.4423076923076923e-05,
+      "loss": 0.7709,
+      "step": 2469
+    },
+    {
+      "epoch": 1.4408633513198192,
+      "grad_norm": 1.1602380275726318,
+      "learning_rate": 1.4408052884615386e-05,
+      "loss": 0.8517,
+      "step": 2470
+    },
+    {
+      "epoch": 1.4414466968061834,
+      "grad_norm": 1.3913099765777588,
+      "learning_rate": 1.4393028846153847e-05,
+      "loss": 0.8339,
+      "step": 2471
+    },
+    {
+      "epoch": 1.4420300422925478,
+      "grad_norm": 1.3142465353012085,
+      "learning_rate": 1.4378004807692308e-05,
+      "loss": 0.7022,
+      "step": 2472
+    },
+    {
+      "epoch": 1.442613387778912,
+      "grad_norm": 1.2736009359359741,
+      "learning_rate": 1.4362980769230769e-05,
+      "loss": 0.8779,
+      "step": 2473
+    },
+    {
+      "epoch": 1.4431967332652764,
+      "grad_norm": 1.3056821823120117,
+      "learning_rate": 1.4347956730769232e-05,
+      "loss": 0.8236,
+      "step": 2474
+    },
+    {
+      "epoch": 1.4437800787516406,
+      "grad_norm": 1.3933846950531006,
+      "learning_rate": 1.4332932692307694e-05,
+      "loss": 0.9686,
+      "step": 2475
+    },
+    {
+      "epoch": 1.444363424238005,
+      "grad_norm": 1.3702456951141357,
+      "learning_rate": 1.4317908653846154e-05,
+      "loss": 0.8735,
+      "step": 2476
+    },
+    {
+      "epoch": 1.4449467697243692,
+      "grad_norm": 1.1643601655960083,
+      "learning_rate": 1.4302884615384614e-05,
+      "loss": 0.6383,
+      "step": 2477
+    },
+    {
+      "epoch": 1.4455301152107336,
+      "grad_norm": 1.2481948137283325,
+      "learning_rate": 1.428786057692308e-05,
+      "loss": 0.8129,
+      "step": 2478
+    },
+    {
+      "epoch": 1.4461134606970978,
+      "grad_norm": 1.904402494430542,
+      "learning_rate": 1.427283653846154e-05,
+      "loss": 0.6686,
+      "step": 2479
+    },
+    {
+      "epoch": 1.4466968061834622,
+      "grad_norm": 1.0331618785858154,
+      "learning_rate": 1.42578125e-05,
+      "loss": 0.6951,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4472801516698264,
+      "grad_norm": 1.3622822761535645,
+      "learning_rate": 1.4242788461538462e-05,
+      "loss": 0.8933,
+      "step": 2481
+    },
+    {
+      "epoch": 1.4478634971561908,
+      "grad_norm": 1.3097389936447144,
+      "learning_rate": 1.4227764423076922e-05,
+      "loss": 0.8643,
+      "step": 2482
+    },
+    {
+      "epoch": 1.448446842642555,
+      "grad_norm": 1.1944518089294434,
+      "learning_rate": 1.4212740384615386e-05,
+      "loss": 0.7313,
+      "step": 2483
+    },
+    {
+      "epoch": 1.4490301881289194,
+      "grad_norm": 1.163017749786377,
+      "learning_rate": 1.4197716346153848e-05,
+      "loss": 0.8936,
+      "step": 2484
+    },
+    {
+      "epoch": 1.4496135336152838,
+      "grad_norm": 1.350412130355835,
+      "learning_rate": 1.4182692307692308e-05,
+      "loss": 0.6471,
+      "step": 2485
+    },
+    {
+      "epoch": 1.450196879101648,
+      "grad_norm": 1.1196433305740356,
+      "learning_rate": 1.4167668269230768e-05,
+      "loss": 0.9139,
+      "step": 2486
+    },
+    {
+      "epoch": 1.4507802245880121,
+      "grad_norm": 1.0108141899108887,
+      "learning_rate": 1.4152644230769233e-05,
+      "loss": 0.7733,
+      "step": 2487
+    },
+    {
+      "epoch": 1.4513635700743766,
+      "grad_norm": 1.1067391633987427,
+      "learning_rate": 1.4137620192307694e-05,
+      "loss": 0.5762,
+      "step": 2488
+    },
+    {
+      "epoch": 1.451946915560741,
+      "grad_norm": 1.3588923215866089,
+      "learning_rate": 1.4122596153846154e-05,
+      "loss": 0.9662,
+      "step": 2489
+    },
+    {
+      "epoch": 1.4525302610471051,
+      "grad_norm": 1.15555739402771,
+      "learning_rate": 1.4107572115384616e-05,
+      "loss": 0.7992,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4531136065334693,
+      "grad_norm": 1.3854424953460693,
+      "learning_rate": 1.409254807692308e-05,
+      "loss": 0.7705,
+      "step": 2491
+    },
+    {
+      "epoch": 1.4536969520198337,
+      "grad_norm": 1.2606521844863892,
+      "learning_rate": 1.407752403846154e-05,
+      "loss": 0.7152,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4542802975061981,
+      "grad_norm": 1.3292980194091797,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.9197,
+      "step": 2493
+    },
+    {
+      "epoch": 1.4548636429925623,
+      "grad_norm": 1.5681556463241577,
+      "learning_rate": 1.4047475961538462e-05,
+      "loss": 0.811,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4554469884789265,
+      "grad_norm": 1.1809593439102173,
+      "learning_rate": 1.4032451923076922e-05,
+      "loss": 1.0052,
+      "step": 2495
+    },
+    {
+      "epoch": 1.456030333965291,
+      "grad_norm": 1.1254372596740723,
+      "learning_rate": 1.4017427884615387e-05,
+      "loss": 0.7758,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4566136794516553,
+      "grad_norm": 1.2921695709228516,
+      "learning_rate": 1.4002403846153847e-05,
+      "loss": 0.7628,
+      "step": 2497
+    },
+    {
+      "epoch": 1.4571970249380195,
+      "grad_norm": 1.3415536880493164,
+      "learning_rate": 1.3987379807692307e-05,
+      "loss": 0.6855,
+      "step": 2498
+    },
+    {
+      "epoch": 1.457780370424384,
+      "grad_norm": 1.3940538167953491,
+      "learning_rate": 1.397235576923077e-05,
+      "loss": 0.7203,
+      "step": 2499
+    },
+    {
+      "epoch": 1.458363715910748,
+      "grad_norm": 1.2635236978530884,
+      "learning_rate": 1.3957331730769233e-05,
+      "loss": 0.843,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4589470613971125,
+      "grad_norm": 1.1825194358825684,
+      "learning_rate": 1.3942307692307693e-05,
+      "loss": 0.6945,
+      "step": 2501
+    },
+    {
+      "epoch": 1.4595304068834767,
+      "grad_norm": 1.2193362712860107,
+      "learning_rate": 1.3927283653846155e-05,
+      "loss": 0.787,
+      "step": 2502
+    },
+    {
+      "epoch": 1.460113752369841,
+      "grad_norm": 1.249801754951477,
+      "learning_rate": 1.3912259615384615e-05,
+      "loss": 0.856,
+      "step": 2503
+    },
+    {
+      "epoch": 1.4606970978562053,
+      "grad_norm": 1.1366738080978394,
+      "learning_rate": 1.3897235576923079e-05,
+      "loss": 0.696,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4612804433425697,
+      "grad_norm": 1.3033206462860107,
+      "learning_rate": 1.388221153846154e-05,
+      "loss": 0.8147,
+      "step": 2505
+    },
+    {
+      "epoch": 1.4618637888289339,
+      "grad_norm": 1.3203999996185303,
+      "learning_rate": 1.38671875e-05,
+      "loss": 0.5789,
+      "step": 2506
+    },
+    {
+      "epoch": 1.4624471343152983,
+      "grad_norm": 1.3760713338851929,
+      "learning_rate": 1.3852163461538461e-05,
+      "loss": 0.7991,
+      "step": 2507
+    },
+    {
+      "epoch": 1.4630304798016625,
+      "grad_norm": 1.2778420448303223,
+      "learning_rate": 1.3837139423076923e-05,
+      "loss": 0.6555,
+      "step": 2508
+    },
+    {
+      "epoch": 1.4636138252880269,
+      "grad_norm": 1.1887472867965698,
+      "learning_rate": 1.3822115384615386e-05,
+      "loss": 0.8536,
+      "step": 2509
+    },
+    {
+      "epoch": 1.464197170774391,
+      "grad_norm": 1.0331052541732788,
+      "learning_rate": 1.3807091346153847e-05,
+      "loss": 0.6553,
+      "step": 2510
+    },
+    {
+      "epoch": 1.4647805162607554,
+      "grad_norm": 1.2358107566833496,
+      "learning_rate": 1.3792067307692309e-05,
+      "loss": 0.8954,
+      "step": 2511
+    },
+    {
+      "epoch": 1.4653638617471199,
+      "grad_norm": 1.4614425897598267,
+      "learning_rate": 1.3777043269230769e-05,
+      "loss": 0.7539,
+      "step": 2512
+    },
+    {
+      "epoch": 1.465947207233484,
+      "grad_norm": 1.2347095012664795,
+      "learning_rate": 1.3762019230769232e-05,
+      "loss": 0.8088,
+      "step": 2513
+    },
+    {
+      "epoch": 1.4665305527198482,
+      "grad_norm": 1.299397587776184,
+      "learning_rate": 1.3746995192307694e-05,
+      "loss": 0.5135,
+      "step": 2514
+    },
+    {
+      "epoch": 1.4671138982062126,
+      "grad_norm": 1.1845791339874268,
+      "learning_rate": 1.3731971153846154e-05,
+      "loss": 0.9701,
+      "step": 2515
+    },
+    {
+      "epoch": 1.467697243692577,
+      "grad_norm": 1.3026585578918457,
+      "learning_rate": 1.3716947115384615e-05,
+      "loss": 0.8863,
+      "step": 2516
+    },
+    {
+      "epoch": 1.4682805891789412,
+      "grad_norm": 1.1109139919281006,
+      "learning_rate": 1.3701923076923078e-05,
+      "loss": 0.703,
+      "step": 2517
+    },
+    {
+      "epoch": 1.4688639346653054,
+      "grad_norm": 1.1830486059188843,
+      "learning_rate": 1.368689903846154e-05,
+      "loss": 0.7957,
+      "step": 2518
+    },
+    {
+      "epoch": 1.4694472801516698,
+      "grad_norm": 1.1587508916854858,
+      "learning_rate": 1.3671875e-05,
+      "loss": 1.0695,
+      "step": 2519
+    },
+    {
+      "epoch": 1.4700306256380342,
+      "grad_norm": 1.0737591981887817,
+      "learning_rate": 1.3656850961538462e-05,
+      "loss": 0.7813,
+      "step": 2520
+    },
+    {
+      "epoch": 1.4706139711243984,
+      "grad_norm": 1.2657713890075684,
+      "learning_rate": 1.3641826923076922e-05,
+      "loss": 0.7455,
+      "step": 2521
+    },
+    {
+      "epoch": 1.4711973166107628,
+      "grad_norm": 1.3846890926361084,
+      "learning_rate": 1.3626802884615386e-05,
+      "loss": 0.7651,
+      "step": 2522
+    },
+    {
+      "epoch": 1.471780662097127,
+      "grad_norm": 1.0715590715408325,
+      "learning_rate": 1.3611778846153848e-05,
+      "loss": 0.8357,
+      "step": 2523
+    },
+    {
+      "epoch": 1.4723640075834914,
+      "grad_norm": 1.2787429094314575,
+      "learning_rate": 1.3596754807692308e-05,
+      "loss": 0.8817,
+      "step": 2524
+    },
+    {
+      "epoch": 1.4729473530698556,
+      "grad_norm": 1.324184536933899,
+      "learning_rate": 1.3581730769230768e-05,
+      "loss": 0.8724,
+      "step": 2525
+    },
+    {
+      "epoch": 1.47353069855622,
+      "grad_norm": 1.1958729028701782,
+      "learning_rate": 1.3566706730769232e-05,
+      "loss": 0.8143,
+      "step": 2526
+    },
+    {
+      "epoch": 1.4741140440425842,
+      "grad_norm": 0.9904715418815613,
+      "learning_rate": 1.3551682692307694e-05,
+      "loss": 0.8887,
+      "step": 2527
+    },
+    {
+      "epoch": 1.4746973895289486,
+      "grad_norm": 1.2931222915649414,
+      "learning_rate": 1.3536658653846154e-05,
+      "loss": 0.7175,
+      "step": 2528
+    },
+    {
+      "epoch": 1.4752807350153128,
+      "grad_norm": 1.3142303228378296,
+      "learning_rate": 1.3521634615384616e-05,
+      "loss": 0.7999,
+      "step": 2529
+    },
+    {
+      "epoch": 1.4758640805016772,
+      "grad_norm": 1.3524141311645508,
+      "learning_rate": 1.350661057692308e-05,
+      "loss": 0.7423,
+      "step": 2530
+    },
+    {
+      "epoch": 1.4764474259880414,
+      "grad_norm": 1.4455997943878174,
+      "learning_rate": 1.349158653846154e-05,
+      "loss": 0.5973,
+      "step": 2531
+    },
+    {
+      "epoch": 1.4770307714744058,
+      "grad_norm": 1.367997169494629,
+      "learning_rate": 1.3476562500000001e-05,
+      "loss": 0.7174,
+      "step": 2532
+    },
+    {
+      "epoch": 1.47761411696077,
+      "grad_norm": 1.2403534650802612,
+      "learning_rate": 1.3461538461538462e-05,
+      "loss": 0.8182,
+      "step": 2533
+    },
+    {
+      "epoch": 1.4781974624471343,
+      "grad_norm": 1.1000239849090576,
+      "learning_rate": 1.3446514423076922e-05,
+      "loss": 0.8016,
+      "step": 2534
+    },
+    {
+      "epoch": 1.4787808079334988,
+      "grad_norm": 1.4198569059371948,
+      "learning_rate": 1.3431490384615385e-05,
+      "loss": 0.6896,
+      "step": 2535
+    },
+    {
+      "epoch": 1.479364153419863,
+      "grad_norm": 1.2650169134140015,
+      "learning_rate": 1.3416466346153847e-05,
+      "loss": 0.7742,
+      "step": 2536
+    },
+    {
+      "epoch": 1.4799474989062271,
+      "grad_norm": 1.4280163049697876,
+      "learning_rate": 1.3401442307692308e-05,
+      "loss": 0.9482,
+      "step": 2537
+    },
+    {
+      "epoch": 1.4805308443925915,
+      "grad_norm": 1.2631523609161377,
+      "learning_rate": 1.338641826923077e-05,
+      "loss": 0.7142,
+      "step": 2538
+    },
+    {
+      "epoch": 1.481114189878956,
+      "grad_norm": 1.236832857131958,
+      "learning_rate": 1.3371394230769233e-05,
+      "loss": 0.6786,
+      "step": 2539
+    },
+    {
+      "epoch": 1.4816975353653201,
+      "grad_norm": 1.2481725215911865,
+      "learning_rate": 1.3356370192307693e-05,
+      "loss": 0.8536,
+      "step": 2540
+    },
+    {
+      "epoch": 1.4822808808516843,
+      "grad_norm": 1.3706049919128418,
+      "learning_rate": 1.3341346153846155e-05,
+      "loss": 0.725,
+      "step": 2541
+    },
+    {
+      "epoch": 1.4828642263380487,
+      "grad_norm": 1.2054919004440308,
+      "learning_rate": 1.3326322115384615e-05,
+      "loss": 0.9378,
+      "step": 2542
+    },
+    {
+      "epoch": 1.4834475718244131,
+      "grad_norm": 1.659885048866272,
+      "learning_rate": 1.3311298076923079e-05,
+      "loss": 0.7564,
+      "step": 2543
+    },
+    {
+      "epoch": 1.4840309173107773,
+      "grad_norm": 1.3692610263824463,
+      "learning_rate": 1.3296274038461539e-05,
+      "loss": 0.8413,
+      "step": 2544
+    },
+    {
+      "epoch": 1.4846142627971415,
+      "grad_norm": 1.287308931350708,
+      "learning_rate": 1.3281250000000001e-05,
+      "loss": 0.695,
+      "step": 2545
+    },
+    {
+      "epoch": 1.4851976082835059,
+      "grad_norm": 1.5530636310577393,
+      "learning_rate": 1.3266225961538461e-05,
+      "loss": 0.7356,
+      "step": 2546
+    },
+    {
+      "epoch": 1.4857809537698703,
+      "grad_norm": 1.2729878425598145,
+      "learning_rate": 1.3251201923076923e-05,
+      "loss": 0.7533,
+      "step": 2547
+    },
+    {
+      "epoch": 1.4863642992562345,
+      "grad_norm": 1.4389548301696777,
+      "learning_rate": 1.3236177884615387e-05,
+      "loss": 0.8441,
+      "step": 2548
+    },
+    {
+      "epoch": 1.4869476447425989,
+      "grad_norm": 1.0692722797393799,
+      "learning_rate": 1.3221153846153847e-05,
+      "loss": 0.8289,
+      "step": 2549
+    },
+    {
+      "epoch": 1.487530990228963,
+      "grad_norm": 1.1718000173568726,
+      "learning_rate": 1.3206129807692309e-05,
+      "loss": 0.7955,
+      "step": 2550
+    },
+    {
+      "epoch": 1.4881143357153275,
+      "grad_norm": 1.1499435901641846,
+      "learning_rate": 1.3191105769230769e-05,
+      "loss": 0.7256,
+      "step": 2551
+    },
+    {
+      "epoch": 1.4886976812016917,
+      "grad_norm": 1.250289797782898,
+      "learning_rate": 1.3176081730769232e-05,
+      "loss": 0.8081,
+      "step": 2552
+    },
+    {
+      "epoch": 1.489281026688056,
+      "grad_norm": 1.243996024131775,
+      "learning_rate": 1.3161057692307693e-05,
+      "loss": 0.6775,
+      "step": 2553
+    },
+    {
+      "epoch": 1.4898643721744202,
+      "grad_norm": 1.2744098901748657,
+      "learning_rate": 1.3146033653846155e-05,
+      "loss": 0.8471,
+      "step": 2554
+    },
+    {
+      "epoch": 1.4904477176607847,
+      "grad_norm": 1.2613190412521362,
+      "learning_rate": 1.3131009615384615e-05,
+      "loss": 0.7044,
+      "step": 2555
+    },
+    {
+      "epoch": 1.4910310631471488,
+      "grad_norm": 1.4642215967178345,
+      "learning_rate": 1.3115985576923078e-05,
+      "loss": 1.0189,
+      "step": 2556
+    },
+    {
+      "epoch": 1.4916144086335132,
+      "grad_norm": 1.1679459810256958,
+      "learning_rate": 1.310096153846154e-05,
+      "loss": 0.8816,
+      "step": 2557
+    },
+    {
+      "epoch": 1.4921977541198774,
+      "grad_norm": 1.1182615756988525,
+      "learning_rate": 1.30859375e-05,
+      "loss": 0.7353,
+      "step": 2558
+    },
+    {
+      "epoch": 1.4927810996062418,
+      "grad_norm": 1.2033658027648926,
+      "learning_rate": 1.3070913461538462e-05,
+      "loss": 0.7741,
+      "step": 2559
+    },
+    {
+      "epoch": 1.493364445092606,
+      "grad_norm": 1.1669368743896484,
+      "learning_rate": 1.3055889423076923e-05,
+      "loss": 0.8408,
+      "step": 2560
+    },
+    {
+      "epoch": 1.4939477905789704,
+      "grad_norm": 1.4733059406280518,
+      "learning_rate": 1.3040865384615386e-05,
+      "loss": 0.7797,
+      "step": 2561
+    },
+    {
+      "epoch": 1.4945311360653348,
+      "grad_norm": 1.431363582611084,
+      "learning_rate": 1.3025841346153846e-05,
+      "loss": 0.794,
+      "step": 2562
+    },
+    {
+      "epoch": 1.495114481551699,
+      "grad_norm": 0.9812197685241699,
+      "learning_rate": 1.3010817307692308e-05,
+      "loss": 0.6596,
+      "step": 2563
+    },
+    {
+      "epoch": 1.4956978270380632,
+      "grad_norm": 1.124457597732544,
+      "learning_rate": 1.2995793269230768e-05,
+      "loss": 0.906,
+      "step": 2564
+    },
+    {
+      "epoch": 1.4962811725244276,
+      "grad_norm": 1.3098465204238892,
+      "learning_rate": 1.2980769230769232e-05,
+      "loss": 1.0137,
+      "step": 2565
+    },
+    {
+      "epoch": 1.496864518010792,
+      "grad_norm": 1.0503084659576416,
+      "learning_rate": 1.2965745192307694e-05,
+      "loss": 0.7207,
+      "step": 2566
+    },
+    {
+      "epoch": 1.4974478634971562,
+      "grad_norm": 1.2493394613265991,
+      "learning_rate": 1.2950721153846154e-05,
+      "loss": 0.9324,
+      "step": 2567
+    },
+    {
+      "epoch": 1.4980312089835204,
+      "grad_norm": 1.1940001249313354,
+      "learning_rate": 1.2935697115384616e-05,
+      "loss": 0.7695,
+      "step": 2568
+    },
+    {
+      "epoch": 1.4986145544698848,
+      "grad_norm": 1.2989429235458374,
+      "learning_rate": 1.292067307692308e-05,
+      "loss": 0.5233,
+      "step": 2569
+    },
+    {
+      "epoch": 1.4991978999562492,
+      "grad_norm": 1.1896779537200928,
+      "learning_rate": 1.290564903846154e-05,
+      "loss": 0.8223,
+      "step": 2570
+    },
+    {
+      "epoch": 1.4997812454426134,
+      "grad_norm": 1.2201910018920898,
+      "learning_rate": 1.2890625e-05,
+      "loss": 0.7695,
+      "step": 2571
+    },
+    {
+      "epoch": 1.5003645909289776,
+      "grad_norm": 1.2268931865692139,
+      "learning_rate": 1.2875600961538462e-05,
+      "loss": 0.8114,
+      "step": 2572
+    },
+    {
+      "epoch": 1.500947936415342,
+      "grad_norm": 1.1898248195648193,
+      "learning_rate": 1.2860576923076922e-05,
+      "loss": 0.8477,
+      "step": 2573
+    },
+    {
+      "epoch": 1.5015312819017064,
+      "grad_norm": 1.4626970291137695,
+      "learning_rate": 1.2845552884615386e-05,
+      "loss": 0.6557,
+      "step": 2574
+    },
+    {
+      "epoch": 1.5021146273880706,
+      "grad_norm": 1.2246737480163574,
+      "learning_rate": 1.2830528846153847e-05,
+      "loss": 0.9006,
+      "step": 2575
+    },
+    {
+      "epoch": 1.5026979728744347,
+      "grad_norm": 1.2037559747695923,
+      "learning_rate": 1.2815504807692308e-05,
+      "loss": 0.8817,
+      "step": 2576
+    },
+    {
+      "epoch": 1.5032813183607991,
+      "grad_norm": 1.3660948276519775,
+      "learning_rate": 1.280048076923077e-05,
+      "loss": 0.6706,
+      "step": 2577
+    },
+    {
+      "epoch": 1.5038646638471636,
+      "grad_norm": 1.359358787536621,
+      "learning_rate": 1.2785456730769233e-05,
+      "loss": 0.8003,
+      "step": 2578
+    },
+    {
+      "epoch": 1.5044480093335277,
+      "grad_norm": 1.2577052116394043,
+      "learning_rate": 1.2770432692307693e-05,
+      "loss": 0.9274,
+      "step": 2579
+    },
+    {
+      "epoch": 1.505031354819892,
+      "grad_norm": 1.2104783058166504,
+      "learning_rate": 1.2755408653846154e-05,
+      "loss": 0.787,
+      "step": 2580
+    },
+    {
+      "epoch": 1.5056147003062563,
+      "grad_norm": 1.3401927947998047,
+      "learning_rate": 1.2740384615384615e-05,
+      "loss": 0.7528,
+      "step": 2581
+    },
+    {
+      "epoch": 1.5061980457926207,
+      "grad_norm": 1.3545399904251099,
+      "learning_rate": 1.2725360576923079e-05,
+      "loss": 0.7889,
+      "step": 2582
+    },
+    {
+      "epoch": 1.506781391278985,
+      "grad_norm": 1.1138622760772705,
+      "learning_rate": 1.271033653846154e-05,
+      "loss": 0.9997,
+      "step": 2583
+    },
+    {
+      "epoch": 1.5073647367653493,
+      "grad_norm": 1.2229743003845215,
+      "learning_rate": 1.2695312500000001e-05,
+      "loss": 0.7661,
+      "step": 2584
+    },
+    {
+      "epoch": 1.5079480822517137,
+      "grad_norm": 1.1078226566314697,
+      "learning_rate": 1.2680288461538461e-05,
+      "loss": 0.6252,
+      "step": 2585
+    },
+    {
+      "epoch": 1.508531427738078,
+      "grad_norm": 1.1622321605682373,
+      "learning_rate": 1.2665264423076923e-05,
+      "loss": 0.8268,
+      "step": 2586
+    },
+    {
+      "epoch": 1.509114773224442,
+      "grad_norm": 1.3086098432540894,
+      "learning_rate": 1.2650240384615387e-05,
+      "loss": 0.8047,
+      "step": 2587
+    },
+    {
+      "epoch": 1.5096981187108065,
+      "grad_norm": 1.4337576627731323,
+      "learning_rate": 1.2635216346153847e-05,
+      "loss": 0.8714,
+      "step": 2588
+    },
+    {
+      "epoch": 1.510281464197171,
+      "grad_norm": 1.358109951019287,
+      "learning_rate": 1.2620192307692307e-05,
+      "loss": 0.7723,
+      "step": 2589
+    },
+    {
+      "epoch": 1.510864809683535,
+      "grad_norm": 1.1631211042404175,
+      "learning_rate": 1.2605168269230769e-05,
+      "loss": 0.788,
+      "step": 2590
+    },
+    {
+      "epoch": 1.5114481551698993,
+      "grad_norm": 1.421904444694519,
+      "learning_rate": 1.2590144230769233e-05,
+      "loss": 0.9168,
+      "step": 2591
+    },
+    {
+      "epoch": 1.5120315006562637,
+      "grad_norm": 1.1729179620742798,
+      "learning_rate": 1.2575120192307693e-05,
+      "loss": 0.9585,
+      "step": 2592
+    },
+    {
+      "epoch": 1.512614846142628,
+      "grad_norm": 1.3910799026489258,
+      "learning_rate": 1.2560096153846155e-05,
+      "loss": 0.6919,
+      "step": 2593
+    },
+    {
+      "epoch": 1.5131981916289923,
+      "grad_norm": 1.3072963953018188,
+      "learning_rate": 1.2545072115384615e-05,
+      "loss": 0.7613,
+      "step": 2594
+    },
+    {
+      "epoch": 1.5137815371153565,
+      "grad_norm": 1.4264005422592163,
+      "learning_rate": 1.2530048076923079e-05,
+      "loss": 0.9186,
+      "step": 2595
+    },
+    {
+      "epoch": 1.5143648826017209,
+      "grad_norm": 1.0973105430603027,
+      "learning_rate": 1.251502403846154e-05,
+      "loss": 0.6772,
+      "step": 2596
+    },
+    {
+      "epoch": 1.5149482280880853,
+      "grad_norm": 1.4456596374511719,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7738,
+      "step": 2597
+    },
+    {
+      "epoch": 1.5155315735744495,
+      "grad_norm": 1.246766209602356,
+      "learning_rate": 1.248497596153846e-05,
+      "loss": 0.9088,
+      "step": 2598
+    },
+    {
+      "epoch": 1.5161149190608136,
+      "grad_norm": 1.170255422592163,
+      "learning_rate": 1.2469951923076924e-05,
+      "loss": 0.8891,
+      "step": 2599
+    },
+    {
+      "epoch": 1.516698264547178,
+      "grad_norm": 1.2451422214508057,
+      "learning_rate": 1.2454927884615385e-05,
+      "loss": 0.8856,
+      "step": 2600
+    },
+    {
+      "epoch": 1.516698264547178,
+      "eval_loss_squad": 0.8741619079979137,
+      "eval_perplexity": 7.975801635804587,
+      "eval_perplexity_reconstruct": 1.8950837793713629,
+      "step": 2600
+    },
+    {
+      "epoch": 1.5172816100335424,
+      "grad_norm": 1.1058474779129028,
+      "learning_rate": 1.2439903846153846e-05,
+      "loss": 0.8623,
+      "step": 2601
+    },
+    {
+      "epoch": 1.5178649555199066,
+      "grad_norm": 1.1985957622528076,
+      "learning_rate": 1.2424879807692308e-05,
+      "loss": 0.5664,
+      "step": 2602
+    },
+    {
+      "epoch": 1.5184483010062708,
+      "grad_norm": 1.2489391565322876,
+      "learning_rate": 1.240985576923077e-05,
+      "loss": 0.9798,
+      "step": 2603
+    },
+    {
+      "epoch": 1.5190316464926352,
+      "grad_norm": 1.2178716659545898,
+      "learning_rate": 1.239483173076923e-05,
+      "loss": 0.8585,
+      "step": 2604
+    },
+    {
+      "epoch": 1.5196149919789996,
+      "grad_norm": 1.065341591835022,
+      "learning_rate": 1.2379807692307694e-05,
+      "loss": 0.6899,
+      "step": 2605
+    },
+    {
+      "epoch": 1.5201983374653638,
+      "grad_norm": 1.3119381666183472,
+      "learning_rate": 1.2364783653846154e-05,
+      "loss": 0.743,
+      "step": 2606
+    },
+    {
+      "epoch": 1.520781682951728,
+      "grad_norm": 1.1781443357467651,
+      "learning_rate": 1.2349759615384616e-05,
+      "loss": 0.7418,
+      "step": 2607
+    },
+    {
+      "epoch": 1.5213650284380926,
+      "grad_norm": 1.3264269828796387,
+      "learning_rate": 1.2334735576923078e-05,
+      "loss": 0.7401,
+      "step": 2608
+    },
+    {
+      "epoch": 1.5219483739244568,
+      "grad_norm": 1.2371203899383545,
+      "learning_rate": 1.231971153846154e-05,
+      "loss": 0.6457,
+      "step": 2609
+    },
+    {
+      "epoch": 1.522531719410821,
+      "grad_norm": 1.1963739395141602,
+      "learning_rate": 1.23046875e-05,
+      "loss": 0.871,
+      "step": 2610
+    },
+    {
+      "epoch": 1.5231150648971854,
+      "grad_norm": 1.4940956830978394,
+      "learning_rate": 1.2289663461538462e-05,
+      "loss": 0.6387,
+      "step": 2611
+    },
+    {
+      "epoch": 1.5236984103835498,
+      "grad_norm": 1.158118724822998,
+      "learning_rate": 1.2274639423076924e-05,
+      "loss": 0.8764,
+      "step": 2612
+    },
+    {
+      "epoch": 1.524281755869914,
+      "grad_norm": 1.2709519863128662,
+      "learning_rate": 1.2259615384615384e-05,
+      "loss": 0.795,
+      "step": 2613
+    },
+    {
+      "epoch": 1.5248651013562782,
+      "grad_norm": 1.1874806880950928,
+      "learning_rate": 1.2244591346153848e-05,
+      "loss": 0.6796,
+      "step": 2614
+    },
+    {
+      "epoch": 1.5254484468426426,
+      "grad_norm": 1.2645267248153687,
+      "learning_rate": 1.2229567307692308e-05,
+      "loss": 0.6134,
+      "step": 2615
+    },
+    {
+      "epoch": 1.526031792329007,
+      "grad_norm": 1.4414528608322144,
+      "learning_rate": 1.221454326923077e-05,
+      "loss": 0.661,
+      "step": 2616
+    },
+    {
+      "epoch": 1.5266151378153712,
+      "grad_norm": 1.1239551305770874,
+      "learning_rate": 1.2199519230769232e-05,
+      "loss": 0.6612,
+      "step": 2617
+    },
+    {
+      "epoch": 1.5271984833017354,
+      "grad_norm": 1.0646827220916748,
+      "learning_rate": 1.2184495192307694e-05,
+      "loss": 0.8923,
+      "step": 2618
+    },
+    {
+      "epoch": 1.5277818287880998,
+      "grad_norm": 1.2423840761184692,
+      "learning_rate": 1.2169471153846154e-05,
+      "loss": 1.1771,
+      "step": 2619
+    },
+    {
+      "epoch": 1.5283651742744642,
+      "grad_norm": 1.0845423936843872,
+      "learning_rate": 1.2154447115384617e-05,
+      "loss": 0.735,
+      "step": 2620
+    },
+    {
+      "epoch": 1.5289485197608284,
+      "grad_norm": 1.2939835786819458,
+      "learning_rate": 1.2139423076923077e-05,
+      "loss": 0.856,
+      "step": 2621
+    },
+    {
+      "epoch": 1.5295318652471925,
+      "grad_norm": 1.2328243255615234,
+      "learning_rate": 1.212439903846154e-05,
+      "loss": 0.7687,
+      "step": 2622
+    },
+    {
+      "epoch": 1.530115210733557,
+      "grad_norm": 1.2373398542404175,
+      "learning_rate": 1.2109375000000001e-05,
+      "loss": 0.7192,
+      "step": 2623
+    },
+    {
+      "epoch": 1.5306985562199213,
+      "grad_norm": 1.3078012466430664,
+      "learning_rate": 1.2094350961538461e-05,
+      "loss": 1.0764,
+      "step": 2624
+    },
+    {
+      "epoch": 1.5312819017062855,
+      "grad_norm": 1.34095299243927,
+      "learning_rate": 1.2079326923076923e-05,
+      "loss": 0.7552,
+      "step": 2625
+    },
+    {
+      "epoch": 1.5318652471926497,
+      "grad_norm": 1.2611044645309448,
+      "learning_rate": 1.2064302884615385e-05,
+      "loss": 0.5605,
+      "step": 2626
+    },
+    {
+      "epoch": 1.5324485926790141,
+      "grad_norm": 1.3855218887329102,
+      "learning_rate": 1.2049278846153847e-05,
+      "loss": 0.7463,
+      "step": 2627
+    },
+    {
+      "epoch": 1.5330319381653785,
+      "grad_norm": 1.3558884859085083,
+      "learning_rate": 1.2034254807692307e-05,
+      "loss": 0.8794,
+      "step": 2628
+    },
+    {
+      "epoch": 1.5336152836517427,
+      "grad_norm": 1.287604570388794,
+      "learning_rate": 1.2019230769230771e-05,
+      "loss": 0.8108,
+      "step": 2629
+    },
+    {
+      "epoch": 1.534198629138107,
+      "grad_norm": 1.35361909866333,
+      "learning_rate": 1.2004206730769231e-05,
+      "loss": 0.7766,
+      "step": 2630
+    },
+    {
+      "epoch": 1.5347819746244713,
+      "grad_norm": 1.1538270711898804,
+      "learning_rate": 1.1989182692307693e-05,
+      "loss": 0.8437,
+      "step": 2631
+    },
+    {
+      "epoch": 1.5353653201108357,
+      "grad_norm": 1.212172031402588,
+      "learning_rate": 1.1974158653846155e-05,
+      "loss": 0.9256,
+      "step": 2632
+    },
+    {
+      "epoch": 1.5359486655972,
+      "grad_norm": 1.2593048810958862,
+      "learning_rate": 1.1959134615384617e-05,
+      "loss": 0.8088,
+      "step": 2633
+    },
+    {
+      "epoch": 1.536532011083564,
+      "grad_norm": 1.1822506189346313,
+      "learning_rate": 1.1944110576923077e-05,
+      "loss": 1.1436,
+      "step": 2634
+    },
+    {
+      "epoch": 1.5371153565699287,
+      "grad_norm": 1.1885801553726196,
+      "learning_rate": 1.1929086538461539e-05,
+      "loss": 0.7486,
+      "step": 2635
+    },
+    {
+      "epoch": 1.5376987020562929,
+      "grad_norm": 1.276310920715332,
+      "learning_rate": 1.19140625e-05,
+      "loss": 0.8676,
+      "step": 2636
+    },
+    {
+      "epoch": 1.538282047542657,
+      "grad_norm": 1.1194807291030884,
+      "learning_rate": 1.1899038461538461e-05,
+      "loss": 0.8396,
+      "step": 2637
+    },
+    {
+      "epoch": 1.5388653930290215,
+      "grad_norm": 1.291012167930603,
+      "learning_rate": 1.1884014423076925e-05,
+      "loss": 0.9238,
+      "step": 2638
+    },
+    {
+      "epoch": 1.5394487385153859,
+      "grad_norm": 1.2062819004058838,
+      "learning_rate": 1.1868990384615385e-05,
+      "loss": 0.714,
+      "step": 2639
+    },
+    {
+      "epoch": 1.54003208400175,
+      "grad_norm": 1.2483253479003906,
+      "learning_rate": 1.1853966346153847e-05,
+      "loss": 0.7584,
+      "step": 2640
+    },
+    {
+      "epoch": 1.5406154294881143,
+      "grad_norm": 1.4191927909851074,
+      "learning_rate": 1.1838942307692309e-05,
+      "loss": 0.9147,
+      "step": 2641
+    },
+    {
+      "epoch": 1.5411987749744787,
+      "grad_norm": 1.292017936706543,
+      "learning_rate": 1.182391826923077e-05,
+      "loss": 0.783,
+      "step": 2642
+    },
+    {
+      "epoch": 1.541782120460843,
+      "grad_norm": 1.4348483085632324,
+      "learning_rate": 1.180889423076923e-05,
+      "loss": 0.8497,
+      "step": 2643
+    },
+    {
+      "epoch": 1.5423654659472072,
+      "grad_norm": 1.4889435768127441,
+      "learning_rate": 1.1793870192307692e-05,
+      "loss": 0.8401,
+      "step": 2644
+    },
+    {
+      "epoch": 1.5429488114335714,
+      "grad_norm": 1.2667125463485718,
+      "learning_rate": 1.1778846153846154e-05,
+      "loss": 0.8579,
+      "step": 2645
+    },
+    {
+      "epoch": 1.5435321569199358,
+      "grad_norm": 1.2864303588867188,
+      "learning_rate": 1.1763822115384616e-05,
+      "loss": 0.7385,
+      "step": 2646
+    },
+    {
+      "epoch": 1.5441155024063002,
+      "grad_norm": 1.079271912574768,
+      "learning_rate": 1.1748798076923078e-05,
+      "loss": 0.7407,
+      "step": 2647
+    },
+    {
+      "epoch": 1.5446988478926644,
+      "grad_norm": 1.0965354442596436,
+      "learning_rate": 1.173377403846154e-05,
+      "loss": 0.7776,
+      "step": 2648
+    },
+    {
+      "epoch": 1.5452821933790286,
+      "grad_norm": 1.260859489440918,
+      "learning_rate": 1.171875e-05,
+      "loss": 0.7992,
+      "step": 2649
+    },
+    {
+      "epoch": 1.545865538865393,
+      "grad_norm": 1.2462226152420044,
+      "learning_rate": 1.1703725961538462e-05,
+      "loss": 1.0434,
+      "step": 2650
+    },
+    {
+      "epoch": 1.5464488843517574,
+      "grad_norm": 1.1139503717422485,
+      "learning_rate": 1.1688701923076924e-05,
+      "loss": 0.6601,
+      "step": 2651
+    },
+    {
+      "epoch": 1.5470322298381216,
+      "grad_norm": 1.375958800315857,
+      "learning_rate": 1.1673677884615384e-05,
+      "loss": 0.8542,
+      "step": 2652
+    },
+    {
+      "epoch": 1.5476155753244858,
+      "grad_norm": 1.268680214881897,
+      "learning_rate": 1.1658653846153846e-05,
+      "loss": 0.9015,
+      "step": 2653
+    },
+    {
+      "epoch": 1.5481989208108502,
+      "grad_norm": 1.3285831212997437,
+      "learning_rate": 1.1643629807692308e-05,
+      "loss": 0.7092,
+      "step": 2654
+    },
+    {
+      "epoch": 1.5487822662972146,
+      "grad_norm": 1.4205631017684937,
+      "learning_rate": 1.162860576923077e-05,
+      "loss": 0.7721,
+      "step": 2655
+    },
+    {
+      "epoch": 1.5493656117835788,
+      "grad_norm": 1.3641010522842407,
+      "learning_rate": 1.1613581730769232e-05,
+      "loss": 0.6439,
+      "step": 2656
+    },
+    {
+      "epoch": 1.549948957269943,
+      "grad_norm": 1.1985429525375366,
+      "learning_rate": 1.1598557692307694e-05,
+      "loss": 0.7375,
+      "step": 2657
+    },
+    {
+      "epoch": 1.5505323027563074,
+      "grad_norm": 1.200697898864746,
+      "learning_rate": 1.1583533653846154e-05,
+      "loss": 0.7262,
+      "step": 2658
+    },
+    {
+      "epoch": 1.5511156482426718,
+      "grad_norm": 1.057250738143921,
+      "learning_rate": 1.1568509615384616e-05,
+      "loss": 0.9285,
+      "step": 2659
+    },
+    {
+      "epoch": 1.551698993729036,
+      "grad_norm": 1.2788975238800049,
+      "learning_rate": 1.1553485576923078e-05,
+      "loss": 0.7772,
+      "step": 2660
+    },
+    {
+      "epoch": 1.5522823392154004,
+      "grad_norm": 1.139838695526123,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 0.949,
+      "step": 2661
+    },
+    {
+      "epoch": 1.5528656847017648,
+      "grad_norm": 1.4383840560913086,
+      "learning_rate": 1.15234375e-05,
+      "loss": 0.7398,
+      "step": 2662
+    },
+    {
+      "epoch": 1.553449030188129,
+      "grad_norm": 1.2270270586013794,
+      "learning_rate": 1.1508413461538462e-05,
+      "loss": 0.8875,
+      "step": 2663
+    },
+    {
+      "epoch": 1.5540323756744931,
+      "grad_norm": 1.2396215200424194,
+      "learning_rate": 1.1493389423076924e-05,
+      "loss": 0.8724,
+      "step": 2664
+    },
+    {
+      "epoch": 1.5546157211608576,
+      "grad_norm": 1.183029294013977,
+      "learning_rate": 1.1478365384615385e-05,
+      "loss": 0.9306,
+      "step": 2665
+    },
+    {
+      "epoch": 1.555199066647222,
+      "grad_norm": 1.2103503942489624,
+      "learning_rate": 1.1463341346153847e-05,
+      "loss": 0.7269,
+      "step": 2666
+    },
+    {
+      "epoch": 1.5557824121335861,
+      "grad_norm": 1.5745123624801636,
+      "learning_rate": 1.1448317307692307e-05,
+      "loss": 0.5633,
+      "step": 2667
+    },
+    {
+      "epoch": 1.5563657576199503,
+      "grad_norm": 1.3425438404083252,
+      "learning_rate": 1.143329326923077e-05,
+      "loss": 0.6148,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5569491031063147,
+      "grad_norm": 1.2065016031265259,
+      "learning_rate": 1.1418269230769231e-05,
+      "loss": 0.6517,
+      "step": 2669
+    },
+    {
+      "epoch": 1.5575324485926791,
+      "grad_norm": 1.1521174907684326,
+      "learning_rate": 1.1403245192307693e-05,
+      "loss": 0.7774,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5581157940790433,
+      "grad_norm": 1.0685138702392578,
+      "learning_rate": 1.1388221153846153e-05,
+      "loss": 0.8086,
+      "step": 2671
+    },
+    {
+      "epoch": 1.5586991395654075,
+      "grad_norm": 1.131866216659546,
+      "learning_rate": 1.1373197115384617e-05,
+      "loss": 0.8954,
+      "step": 2672
+    },
+    {
+      "epoch": 1.559282485051772,
+      "grad_norm": 1.074537754058838,
+      "learning_rate": 1.1358173076923077e-05,
+      "loss": 0.8301,
+      "step": 2673
+    },
+    {
+      "epoch": 1.5598658305381363,
+      "grad_norm": 1.255470633506775,
+      "learning_rate": 1.1343149038461539e-05,
+      "loss": 0.7473,
+      "step": 2674
+    },
+    {
+      "epoch": 1.5604491760245005,
+      "grad_norm": 1.3070282936096191,
+      "learning_rate": 1.1328125000000001e-05,
+      "loss": 0.7232,
+      "step": 2675
+    },
+    {
+      "epoch": 1.5610325215108647,
+      "grad_norm": 1.4808948040008545,
+      "learning_rate": 1.1313100961538461e-05,
+      "loss": 1.0816,
+      "step": 2676
+    },
+    {
+      "epoch": 1.561615866997229,
+      "grad_norm": 1.2335243225097656,
+      "learning_rate": 1.1298076923076923e-05,
+      "loss": 0.9074,
+      "step": 2677
+    },
+    {
+      "epoch": 1.5621992124835935,
+      "grad_norm": 1.1924381256103516,
+      "learning_rate": 1.1283052884615385e-05,
+      "loss": 0.6063,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5627825579699577,
+      "grad_norm": 1.2754387855529785,
+      "learning_rate": 1.1268028846153847e-05,
+      "loss": 0.8983,
+      "step": 2679
+    },
+    {
+      "epoch": 1.5633659034563219,
+      "grad_norm": 1.326870083808899,
+      "learning_rate": 1.1253004807692307e-05,
+      "loss": 0.6428,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5639492489426863,
+      "grad_norm": 1.1194915771484375,
+      "learning_rate": 1.123798076923077e-05,
+      "loss": 0.6659,
+      "step": 2681
+    },
+    {
+      "epoch": 1.5645325944290507,
+      "grad_norm": 1.119391679763794,
+      "learning_rate": 1.122295673076923e-05,
+      "loss": 0.8714,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5651159399154149,
+      "grad_norm": 1.299011468887329,
+      "learning_rate": 1.1207932692307693e-05,
+      "loss": 0.8289,
+      "step": 2683
+    },
+    {
+      "epoch": 1.565699285401779,
+      "grad_norm": 1.0118248462677002,
+      "learning_rate": 1.1192908653846155e-05,
+      "loss": 0.6258,
+      "step": 2684
+    },
+    {
+      "epoch": 1.5662826308881435,
+      "grad_norm": 1.1196413040161133,
+      "learning_rate": 1.1177884615384616e-05,
+      "loss": 0.7186,
+      "step": 2685
+    },
+    {
+      "epoch": 1.5668659763745079,
+      "grad_norm": 1.2307488918304443,
+      "learning_rate": 1.1162860576923077e-05,
+      "loss": 0.7569,
+      "step": 2686
+    },
+    {
+      "epoch": 1.567449321860872,
+      "grad_norm": 1.2402465343475342,
+      "learning_rate": 1.114783653846154e-05,
+      "loss": 0.8829,
+      "step": 2687
+    },
+    {
+      "epoch": 1.5680326673472365,
+      "grad_norm": 1.262913465499878,
+      "learning_rate": 1.11328125e-05,
+      "loss": 0.8594,
+      "step": 2688
+    },
+    {
+      "epoch": 1.5686160128336009,
+      "grad_norm": 1.1202564239501953,
+      "learning_rate": 1.111778846153846e-05,
+      "loss": 0.8337,
+      "step": 2689
+    },
+    {
+      "epoch": 1.569199358319965,
+      "grad_norm": 1.263912320137024,
+      "learning_rate": 1.1102764423076924e-05,
+      "loss": 0.875,
+      "step": 2690
+    },
+    {
+      "epoch": 1.5697827038063292,
+      "grad_norm": 1.3599956035614014,
+      "learning_rate": 1.1087740384615384e-05,
+      "loss": 0.7736,
+      "step": 2691
+    },
+    {
+      "epoch": 1.5703660492926936,
+      "grad_norm": 1.435336947441101,
+      "learning_rate": 1.1072716346153846e-05,
+      "loss": 0.5259,
+      "step": 2692
+    },
+    {
+      "epoch": 1.570949394779058,
+      "grad_norm": 1.3584843873977661,
+      "learning_rate": 1.1057692307692308e-05,
+      "loss": 0.6383,
+      "step": 2693
+    },
+    {
+      "epoch": 1.5715327402654222,
+      "grad_norm": 1.44851815700531,
+      "learning_rate": 1.104266826923077e-05,
+      "loss": 0.8313,
+      "step": 2694
+    },
+    {
+      "epoch": 1.5721160857517864,
+      "grad_norm": 1.316826343536377,
+      "learning_rate": 1.102764423076923e-05,
+      "loss": 0.6728,
+      "step": 2695
+    },
+    {
+      "epoch": 1.5726994312381508,
+      "grad_norm": 1.5356096029281616,
+      "learning_rate": 1.1012620192307694e-05,
+      "loss": 0.7044,
+      "step": 2696
+    },
+    {
+      "epoch": 1.5732827767245152,
+      "grad_norm": 1.1947566270828247,
+      "learning_rate": 1.0997596153846154e-05,
+      "loss": 0.6776,
+      "step": 2697
+    },
+    {
+      "epoch": 1.5738661222108794,
+      "grad_norm": 1.3364490270614624,
+      "learning_rate": 1.0982572115384616e-05,
+      "loss": 0.8211,
+      "step": 2698
+    },
+    {
+      "epoch": 1.5744494676972436,
+      "grad_norm": 1.2481436729431152,
+      "learning_rate": 1.0967548076923078e-05,
+      "loss": 0.9184,
+      "step": 2699
+    },
+    {
+      "epoch": 1.575032813183608,
+      "grad_norm": 1.224586009979248,
+      "learning_rate": 1.095252403846154e-05,
+      "loss": 0.6932,
+      "step": 2700
+    },
+    {
+      "epoch": 1.5756161586699724,
+      "grad_norm": 1.1394731998443604,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.8501,
+      "step": 2701
+    },
+    {
+      "epoch": 1.5761995041563366,
+      "grad_norm": 1.1618777513504028,
+      "learning_rate": 1.0922475961538462e-05,
+      "loss": 0.9925,
+      "step": 2702
+    },
+    {
+      "epoch": 1.5767828496427008,
+      "grad_norm": 1.480178952217102,
+      "learning_rate": 1.0907451923076924e-05,
+      "loss": 0.8851,
+      "step": 2703
+    },
+    {
+      "epoch": 1.5773661951290652,
+      "grad_norm": 1.0637935400009155,
+      "learning_rate": 1.0892427884615384e-05,
+      "loss": 0.7666,
+      "step": 2704
+    },
+    {
+      "epoch": 1.5779495406154296,
+      "grad_norm": 1.2903245687484741,
+      "learning_rate": 1.0877403846153847e-05,
+      "loss": 0.7153,
+      "step": 2705
+    },
+    {
+      "epoch": 1.5785328861017938,
+      "grad_norm": 1.4703887701034546,
+      "learning_rate": 1.0862379807692308e-05,
+      "loss": 0.7709,
+      "step": 2706
+    },
+    {
+      "epoch": 1.579116231588158,
+      "grad_norm": 1.12290358543396,
+      "learning_rate": 1.084735576923077e-05,
+      "loss": 0.7594,
+      "step": 2707
+    },
+    {
+      "epoch": 1.5796995770745224,
+      "grad_norm": 1.511317253112793,
+      "learning_rate": 1.0832331730769231e-05,
+      "loss": 0.9694,
+      "step": 2708
+    },
+    {
+      "epoch": 1.5802829225608868,
+      "grad_norm": 1.799383282661438,
+      "learning_rate": 1.0817307692307693e-05,
+      "loss": 0.6799,
+      "step": 2709
+    },
+    {
+      "epoch": 1.580866268047251,
+      "grad_norm": 1.1799027919769287,
+      "learning_rate": 1.0802283653846154e-05,
+      "loss": 0.8357,
+      "step": 2710
+    },
+    {
+      "epoch": 1.5814496135336151,
+      "grad_norm": 1.2925686836242676,
+      "learning_rate": 1.0787259615384617e-05,
+      "loss": 0.6977,
+      "step": 2711
+    },
+    {
+      "epoch": 1.5820329590199795,
+      "grad_norm": 1.284294843673706,
+      "learning_rate": 1.0772235576923077e-05,
+      "loss": 0.8429,
+      "step": 2712
+    },
+    {
+      "epoch": 1.582616304506344,
+      "grad_norm": 1.1772273778915405,
+      "learning_rate": 1.075721153846154e-05,
+      "loss": 0.692,
+      "step": 2713
+    },
+    {
+      "epoch": 1.5831996499927081,
+      "grad_norm": 1.1449207067489624,
+      "learning_rate": 1.0742187500000001e-05,
+      "loss": 0.7098,
+      "step": 2714
+    },
+    {
+      "epoch": 1.5837829954790725,
+      "grad_norm": 1.4304105043411255,
+      "learning_rate": 1.0727163461538461e-05,
+      "loss": 0.8735,
+      "step": 2715
+    },
+    {
+      "epoch": 1.584366340965437,
+      "grad_norm": 1.024904727935791,
+      "learning_rate": 1.0712139423076923e-05,
+      "loss": 0.7593,
+      "step": 2716
+    },
+    {
+      "epoch": 1.5849496864518011,
+      "grad_norm": 1.2996480464935303,
+      "learning_rate": 1.0697115384615385e-05,
+      "loss": 0.7183,
+      "step": 2717
+    },
+    {
+      "epoch": 1.5855330319381653,
+      "grad_norm": 1.3686482906341553,
+      "learning_rate": 1.0682091346153847e-05,
+      "loss": 0.9928,
+      "step": 2718
+    },
+    {
+      "epoch": 1.5861163774245297,
+      "grad_norm": 1.2611160278320312,
+      "learning_rate": 1.0667067307692307e-05,
+      "loss": 0.937,
+      "step": 2719
+    },
+    {
+      "epoch": 1.5866997229108941,
+      "grad_norm": 1.0135339498519897,
+      "learning_rate": 1.065204326923077e-05,
+      "loss": 0.8119,
+      "step": 2720
+    },
+    {
+      "epoch": 1.5872830683972583,
+      "grad_norm": 1.1465262174606323,
+      "learning_rate": 1.0637019230769231e-05,
+      "loss": 0.788,
+      "step": 2721
+    },
+    {
+      "epoch": 1.5878664138836225,
+      "grad_norm": 1.3036527633666992,
+      "learning_rate": 1.0621995192307693e-05,
+      "loss": 0.746,
+      "step": 2722
+    },
+    {
+      "epoch": 1.588449759369987,
+      "grad_norm": 1.3011112213134766,
+      "learning_rate": 1.0606971153846155e-05,
+      "loss": 0.9224,
+      "step": 2723
+    },
+    {
+      "epoch": 1.5890331048563513,
+      "grad_norm": 1.1597706079483032,
+      "learning_rate": 1.0591947115384617e-05,
+      "loss": 0.9486,
+      "step": 2724
+    },
+    {
+      "epoch": 1.5896164503427155,
+      "grad_norm": 1.1415634155273438,
+      "learning_rate": 1.0576923076923077e-05,
+      "loss": 0.6527,
+      "step": 2725
+    },
+    {
+      "epoch": 1.5901997958290797,
+      "grad_norm": 1.1724604368209839,
+      "learning_rate": 1.056189903846154e-05,
+      "loss": 0.6936,
+      "step": 2726
+    },
+    {
+      "epoch": 1.590783141315444,
+      "grad_norm": 1.014355182647705,
+      "learning_rate": 1.0546875e-05,
+      "loss": 0.7157,
+      "step": 2727
+    },
+    {
+      "epoch": 1.5913664868018085,
+      "grad_norm": 1.2243266105651855,
+      "learning_rate": 1.053185096153846e-05,
+      "loss": 0.7737,
+      "step": 2728
+    },
+    {
+      "epoch": 1.5919498322881727,
+      "grad_norm": 1.347441554069519,
+      "learning_rate": 1.0516826923076924e-05,
+      "loss": 0.6544,
+      "step": 2729
+    },
+    {
+      "epoch": 1.5925331777745368,
+      "grad_norm": 1.1011338233947754,
+      "learning_rate": 1.0501802884615385e-05,
+      "loss": 0.6885,
+      "step": 2730
+    },
+    {
+      "epoch": 1.5931165232609013,
+      "grad_norm": 1.2773078680038452,
+      "learning_rate": 1.0486778846153846e-05,
+      "loss": 0.826,
+      "step": 2731
+    },
+    {
+      "epoch": 1.5936998687472657,
+      "grad_norm": 1.3734136819839478,
+      "learning_rate": 1.0471754807692308e-05,
+      "loss": 0.8921,
+      "step": 2732
+    },
+    {
+      "epoch": 1.5942832142336298,
+      "grad_norm": 1.4568568468093872,
+      "learning_rate": 1.045673076923077e-05,
+      "loss": 0.7519,
+      "step": 2733
+    },
+    {
+      "epoch": 1.594866559719994,
+      "grad_norm": 1.0718237161636353,
+      "learning_rate": 1.044170673076923e-05,
+      "loss": 0.7066,
+      "step": 2734
+    },
+    {
+      "epoch": 1.5954499052063584,
+      "grad_norm": 1.2433981895446777,
+      "learning_rate": 1.0426682692307694e-05,
+      "loss": 0.5966,
+      "step": 2735
+    },
+    {
+      "epoch": 1.5960332506927228,
+      "grad_norm": 1.313547134399414,
+      "learning_rate": 1.0411658653846154e-05,
+      "loss": 0.714,
+      "step": 2736
+    },
+    {
+      "epoch": 1.596616596179087,
+      "grad_norm": 1.2181950807571411,
+      "learning_rate": 1.0396634615384616e-05,
+      "loss": 0.8592,
+      "step": 2737
+    },
+    {
+      "epoch": 1.5971999416654512,
+      "grad_norm": 1.5490280389785767,
+      "learning_rate": 1.0381610576923078e-05,
+      "loss": 0.8276,
+      "step": 2738
+    },
+    {
+      "epoch": 1.5977832871518156,
+      "grad_norm": 1.3620082139968872,
+      "learning_rate": 1.036658653846154e-05,
+      "loss": 1.0517,
+      "step": 2739
+    },
+    {
+      "epoch": 1.59836663263818,
+      "grad_norm": 1.155208945274353,
+      "learning_rate": 1.03515625e-05,
+      "loss": 0.6895,
+      "step": 2740
+    },
+    {
+      "epoch": 1.5989499781245442,
+      "grad_norm": 1.2351773977279663,
+      "learning_rate": 1.0336538461538462e-05,
+      "loss": 0.5748,
+      "step": 2741
+    },
+    {
+      "epoch": 1.5995333236109086,
+      "grad_norm": 0.9607925415039062,
+      "learning_rate": 1.0321514423076924e-05,
+      "loss": 0.87,
+      "step": 2742
+    },
+    {
+      "epoch": 1.600116669097273,
+      "grad_norm": 1.342832326889038,
+      "learning_rate": 1.0306490384615384e-05,
+      "loss": 0.6786,
+      "step": 2743
+    },
+    {
+      "epoch": 1.6007000145836372,
+      "grad_norm": 1.3285024166107178,
+      "learning_rate": 1.0291466346153848e-05,
+      "loss": 0.6508,
+      "step": 2744
+    },
+    {
+      "epoch": 1.6012833600700014,
+      "grad_norm": 1.2747256755828857,
+      "learning_rate": 1.0276442307692308e-05,
+      "loss": 0.7721,
+      "step": 2745
+    },
+    {
+      "epoch": 1.6018667055563658,
+      "grad_norm": 1.2518932819366455,
+      "learning_rate": 1.026141826923077e-05,
+      "loss": 0.7242,
+      "step": 2746
+    },
+    {
+      "epoch": 1.6024500510427302,
+      "grad_norm": 1.4885451793670654,
+      "learning_rate": 1.0246394230769232e-05,
+      "loss": 0.831,
+      "step": 2747
+    },
+    {
+      "epoch": 1.6030333965290944,
+      "grad_norm": 1.324741005897522,
+      "learning_rate": 1.0231370192307693e-05,
+      "loss": 0.7975,
+      "step": 2748
+    },
+    {
+      "epoch": 1.6036167420154586,
+      "grad_norm": 1.2721138000488281,
+      "learning_rate": 1.0216346153846154e-05,
+      "loss": 0.8043,
+      "step": 2749
+    },
+    {
+      "epoch": 1.604200087501823,
+      "grad_norm": 1.3388422727584839,
+      "learning_rate": 1.0201322115384617e-05,
+      "loss": 0.8146,
+      "step": 2750
+    },
+    {
+      "epoch": 1.6047834329881874,
+      "grad_norm": 1.4431674480438232,
+      "learning_rate": 1.0186298076923077e-05,
+      "loss": 0.8857,
+      "step": 2751
+    },
+    {
+      "epoch": 1.6053667784745516,
+      "grad_norm": 1.3353809118270874,
+      "learning_rate": 1.017127403846154e-05,
+      "loss": 0.6763,
+      "step": 2752
+    },
+    {
+      "epoch": 1.6059501239609157,
+      "grad_norm": 1.4101769924163818,
+      "learning_rate": 1.0156250000000001e-05,
+      "loss": 0.7768,
+      "step": 2753
+    },
+    {
+      "epoch": 1.6065334694472801,
+      "grad_norm": 1.2769949436187744,
+      "learning_rate": 1.0141225961538461e-05,
+      "loss": 0.803,
+      "step": 2754
+    },
+    {
+      "epoch": 1.6071168149336446,
+      "grad_norm": 1.3187309503555298,
+      "learning_rate": 1.0126201923076923e-05,
+      "loss": 0.7019,
+      "step": 2755
+    },
+    {
+      "epoch": 1.6077001604200087,
+      "grad_norm": 1.083141565322876,
+      "learning_rate": 1.0111177884615385e-05,
+      "loss": 0.8864,
+      "step": 2756
+    },
+    {
+      "epoch": 1.608283505906373,
+      "grad_norm": 1.8760899305343628,
+      "learning_rate": 1.0096153846153847e-05,
+      "loss": 0.8891,
+      "step": 2757
+    },
+    {
+      "epoch": 1.6088668513927373,
+      "grad_norm": 1.2959517240524292,
+      "learning_rate": 1.0081129807692307e-05,
+      "loss": 0.8422,
+      "step": 2758
+    },
+    {
+      "epoch": 1.6094501968791017,
+      "grad_norm": 1.4058853387832642,
+      "learning_rate": 1.0066105769230771e-05,
+      "loss": 0.6244,
+      "step": 2759
+    },
+    {
+      "epoch": 1.610033542365466,
+      "grad_norm": 1.154642939567566,
+      "learning_rate": 1.0051081730769231e-05,
+      "loss": 0.859,
+      "step": 2760
+    },
+    {
+      "epoch": 1.61061688785183,
+      "grad_norm": 1.209807276725769,
+      "learning_rate": 1.0036057692307693e-05,
+      "loss": 0.6576,
+      "step": 2761
+    },
+    {
+      "epoch": 1.6112002333381945,
+      "grad_norm": 1.3753244876861572,
+      "learning_rate": 1.0021033653846155e-05,
+      "loss": 0.6883,
+      "step": 2762
+    },
+    {
+      "epoch": 1.611783578824559,
+      "grad_norm": 1.2860386371612549,
+      "learning_rate": 1.0006009615384617e-05,
+      "loss": 0.8279,
+      "step": 2763
+    },
+    {
+      "epoch": 1.612366924310923,
+      "grad_norm": 1.2552486658096313,
+      "learning_rate": 9.990985576923077e-06,
+      "loss": 0.8707,
+      "step": 2764
+    },
+    {
+      "epoch": 1.6129502697972873,
+      "grad_norm": 1.2394871711730957,
+      "learning_rate": 9.975961538461539e-06,
+      "loss": 0.9137,
+      "step": 2765
+    },
+    {
+      "epoch": 1.6135336152836517,
+      "grad_norm": 1.0615895986557007,
+      "learning_rate": 9.9609375e-06,
+      "loss": 0.9192,
+      "step": 2766
+    },
+    {
+      "epoch": 1.614116960770016,
+      "grad_norm": 1.2719849348068237,
+      "learning_rate": 9.945913461538461e-06,
+      "loss": 0.8623,
+      "step": 2767
+    },
+    {
+      "epoch": 1.6147003062563803,
+      "grad_norm": 1.2009553909301758,
+      "learning_rate": 9.930889423076924e-06,
+      "loss": 0.7557,
+      "step": 2768
+    },
+    {
+      "epoch": 1.6152836517427447,
+      "grad_norm": 1.2548385858535767,
+      "learning_rate": 9.915865384615385e-06,
+      "loss": 0.8356,
+      "step": 2769
+    },
+    {
+      "epoch": 1.615866997229109,
+      "grad_norm": 1.4359416961669922,
+      "learning_rate": 9.900841346153847e-06,
+      "loss": 0.954,
+      "step": 2770
+    },
+    {
+      "epoch": 1.6164503427154733,
+      "grad_norm": 1.3595879077911377,
+      "learning_rate": 9.885817307692308e-06,
+      "loss": 0.771,
+      "step": 2771
+    },
+    {
+      "epoch": 1.6170336882018375,
+      "grad_norm": 1.4051216840744019,
+      "learning_rate": 9.87079326923077e-06,
+      "loss": 0.7762,
+      "step": 2772
+    },
+    {
+      "epoch": 1.6176170336882019,
+      "grad_norm": 1.0524299144744873,
+      "learning_rate": 9.85576923076923e-06,
+      "loss": 0.7751,
+      "step": 2773
+    },
+    {
+      "epoch": 1.6182003791745663,
+      "grad_norm": 1.2120068073272705,
+      "learning_rate": 9.840745192307692e-06,
+      "loss": 0.7958,
+      "step": 2774
+    },
+    {
+      "epoch": 1.6187837246609305,
+      "grad_norm": 1.271024227142334,
+      "learning_rate": 9.825721153846154e-06,
+      "loss": 0.8882,
+      "step": 2775
+    },
+    {
+      "epoch": 1.6193670701472946,
+      "grad_norm": 1.1276352405548096,
+      "learning_rate": 9.810697115384616e-06,
+      "loss": 0.8336,
+      "step": 2776
+    },
+    {
+      "epoch": 1.619950415633659,
+      "grad_norm": 1.3857134580612183,
+      "learning_rate": 9.795673076923078e-06,
+      "loss": 0.766,
+      "step": 2777
+    },
+    {
+      "epoch": 1.6205337611200235,
+      "grad_norm": 1.1864246129989624,
+      "learning_rate": 9.78064903846154e-06,
+      "loss": 0.6705,
+      "step": 2778
+    },
+    {
+      "epoch": 1.6211171066063876,
+      "grad_norm": 1.193347692489624,
+      "learning_rate": 9.765625e-06,
+      "loss": 0.865,
+      "step": 2779
+    },
+    {
+      "epoch": 1.6217004520927518,
+      "grad_norm": 1.2671443223953247,
+      "learning_rate": 9.750600961538462e-06,
+      "loss": 0.7479,
+      "step": 2780
+    },
+    {
+      "epoch": 1.6222837975791162,
+      "grad_norm": 1.2168381214141846,
+      "learning_rate": 9.735576923076924e-06,
+      "loss": 0.8945,
+      "step": 2781
+    },
+    {
+      "epoch": 1.6228671430654806,
+      "grad_norm": 1.0259840488433838,
+      "learning_rate": 9.720552884615384e-06,
+      "loss": 1.031,
+      "step": 2782
+    },
+    {
+      "epoch": 1.6234504885518448,
+      "grad_norm": 1.3780772686004639,
+      "learning_rate": 9.705528846153846e-06,
+      "loss": 0.9144,
+      "step": 2783
+    },
+    {
+      "epoch": 1.624033834038209,
+      "grad_norm": 1.1934678554534912,
+      "learning_rate": 9.690504807692308e-06,
+      "loss": 0.8131,
+      "step": 2784
+    },
+    {
+      "epoch": 1.6246171795245734,
+      "grad_norm": 1.2687208652496338,
+      "learning_rate": 9.67548076923077e-06,
+      "loss": 0.7369,
+      "step": 2785
+    },
+    {
+      "epoch": 1.6252005250109378,
+      "grad_norm": 1.2904934883117676,
+      "learning_rate": 9.660456730769232e-06,
+      "loss": 0.6605,
+      "step": 2786
+    },
+    {
+      "epoch": 1.625783870497302,
+      "grad_norm": 1.6874489784240723,
+      "learning_rate": 9.645432692307694e-06,
+      "loss": 0.8216,
+      "step": 2787
+    },
+    {
+      "epoch": 1.6263672159836662,
+      "grad_norm": 1.156012773513794,
+      "learning_rate": 9.630408653846154e-06,
+      "loss": 0.7344,
+      "step": 2788
+    },
+    {
+      "epoch": 1.6269505614700306,
+      "grad_norm": 1.2649706602096558,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.7568,
+      "step": 2789
+    },
+    {
+      "epoch": 1.627533906956395,
+      "grad_norm": 1.2135905027389526,
+      "learning_rate": 9.600360576923078e-06,
+      "loss": 0.7164,
+      "step": 2790
+    },
+    {
+      "epoch": 1.6281172524427592,
+      "grad_norm": 1.3913686275482178,
+      "learning_rate": 9.58533653846154e-06,
+      "loss": 0.7437,
+      "step": 2791
+    },
+    {
+      "epoch": 1.6287005979291234,
+      "grad_norm": 1.119268536567688,
+      "learning_rate": 9.5703125e-06,
+      "loss": 0.8903,
+      "step": 2792
+    },
+    {
+      "epoch": 1.6292839434154878,
+      "grad_norm": 1.0997390747070312,
+      "learning_rate": 9.555288461538462e-06,
+      "loss": 0.8839,
+      "step": 2793
+    },
+    {
+      "epoch": 1.6298672889018522,
+      "grad_norm": 1.7211315631866455,
+      "learning_rate": 9.540264423076923e-06,
+      "loss": 0.7962,
+      "step": 2794
+    },
+    {
+      "epoch": 1.6304506343882164,
+      "grad_norm": 1.1959589719772339,
+      "learning_rate": 9.525240384615385e-06,
+      "loss": 0.7228,
+      "step": 2795
+    },
+    {
+      "epoch": 1.6310339798745808,
+      "grad_norm": 1.3861689567565918,
+      "learning_rate": 9.510216346153847e-06,
+      "loss": 0.8062,
+      "step": 2796
+    },
+    {
+      "epoch": 1.6316173253609452,
+      "grad_norm": 1.251761794090271,
+      "learning_rate": 9.495192307692307e-06,
+      "loss": 0.8465,
+      "step": 2797
+    },
+    {
+      "epoch": 1.6322006708473094,
+      "grad_norm": 1.3125969171524048,
+      "learning_rate": 9.48016826923077e-06,
+      "loss": 0.9648,
+      "step": 2798
+    },
+    {
+      "epoch": 1.6327840163336735,
+      "grad_norm": 1.0540010929107666,
+      "learning_rate": 9.465144230769231e-06,
+      "loss": 0.8933,
+      "step": 2799
+    },
+    {
+      "epoch": 1.633367361820038,
+      "grad_norm": 1.2731456756591797,
+      "learning_rate": 9.450120192307693e-06,
+      "loss": 0.6307,
+      "step": 2800
+    },
+    {
+      "epoch": 1.633367361820038,
+      "eval_loss_squad": 0.8574257939518429,
+      "eval_perplexity": 8.239154888947215,
+      "eval_perplexity_reconstruct": 1.9011622713541043,
+      "step": 2800
+    },
+    {
+      "epoch": 1.6339507073064024,
+      "grad_norm": 1.4683425426483154,
+      "learning_rate": 9.435096153846153e-06,
+      "loss": 0.8113,
+      "step": 2801
+    },
+    {
+      "epoch": 1.6345340527927665,
+      "grad_norm": 1.1839910745620728,
+      "learning_rate": 9.420072115384617e-06,
+      "loss": 0.6765,
+      "step": 2802
+    },
+    {
+      "epoch": 1.6351173982791307,
+      "grad_norm": 1.1331663131713867,
+      "learning_rate": 9.405048076923077e-06,
+      "loss": 0.6725,
+      "step": 2803
+    },
+    {
+      "epoch": 1.6357007437654951,
+      "grad_norm": 1.2307205200195312,
+      "learning_rate": 9.390024038461539e-06,
+      "loss": 0.728,
+      "step": 2804
+    },
+    {
+      "epoch": 1.6362840892518595,
+      "grad_norm": 1.178563117980957,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.8463,
+      "step": 2805
+    },
+    {
+      "epoch": 1.6368674347382237,
+      "grad_norm": 1.1501257419586182,
+      "learning_rate": 9.359975961538461e-06,
+      "loss": 0.9248,
+      "step": 2806
+    },
+    {
+      "epoch": 1.637450780224588,
+      "grad_norm": 1.2466018199920654,
+      "learning_rate": 9.344951923076923e-06,
+      "loss": 0.8406,
+      "step": 2807
+    },
+    {
+      "epoch": 1.6380341257109523,
+      "grad_norm": 1.2315740585327148,
+      "learning_rate": 9.329927884615385e-06,
+      "loss": 0.8917,
+      "step": 2808
+    },
+    {
+      "epoch": 1.6386174711973167,
+      "grad_norm": 1.3349418640136719,
+      "learning_rate": 9.314903846153847e-06,
+      "loss": 0.9475,
+      "step": 2809
+    },
+    {
+      "epoch": 1.639200816683681,
+      "grad_norm": 1.3950275182724,
+      "learning_rate": 9.299879807692307e-06,
+      "loss": 0.6757,
+      "step": 2810
+    },
+    {
+      "epoch": 1.639784162170045,
+      "grad_norm": 1.2387348413467407,
+      "learning_rate": 9.28485576923077e-06,
+      "loss": 0.7381,
+      "step": 2811
+    },
+    {
+      "epoch": 1.6403675076564095,
+      "grad_norm": 1.3129287958145142,
+      "learning_rate": 9.26983173076923e-06,
+      "loss": 0.6148,
+      "step": 2812
+    },
+    {
+      "epoch": 1.640950853142774,
+      "grad_norm": 1.1648530960083008,
+      "learning_rate": 9.254807692307693e-06,
+      "loss": 0.7276,
+      "step": 2813
+    },
+    {
+      "epoch": 1.641534198629138,
+      "grad_norm": 1.53678560256958,
+      "learning_rate": 9.239783653846154e-06,
+      "loss": 0.9696,
+      "step": 2814
+    },
+    {
+      "epoch": 1.6421175441155023,
+      "grad_norm": 1.1140925884246826,
+      "learning_rate": 9.224759615384616e-06,
+      "loss": 0.6841,
+      "step": 2815
+    },
+    {
+      "epoch": 1.6427008896018667,
+      "grad_norm": 1.36318838596344,
+      "learning_rate": 9.209735576923077e-06,
+      "loss": 0.8003,
+      "step": 2816
+    },
+    {
+      "epoch": 1.643284235088231,
+      "grad_norm": 1.1023435592651367,
+      "learning_rate": 9.19471153846154e-06,
+      "loss": 0.7749,
+      "step": 2817
+    },
+    {
+      "epoch": 1.6438675805745953,
+      "grad_norm": 1.7993230819702148,
+      "learning_rate": 9.1796875e-06,
+      "loss": 0.6514,
+      "step": 2818
+    },
+    {
+      "epoch": 1.6444509260609594,
+      "grad_norm": 1.1305932998657227,
+      "learning_rate": 9.16466346153846e-06,
+      "loss": 0.802,
+      "step": 2819
+    },
+    {
+      "epoch": 1.645034271547324,
+      "grad_norm": 1.2455143928527832,
+      "learning_rate": 9.149639423076924e-06,
+      "loss": 0.6783,
+      "step": 2820
+    },
+    {
+      "epoch": 1.6456176170336883,
+      "grad_norm": 1.2018240690231323,
+      "learning_rate": 9.134615384615384e-06,
+      "loss": 0.6678,
+      "step": 2821
+    },
+    {
+      "epoch": 1.6462009625200524,
+      "grad_norm": 1.5132325887680054,
+      "learning_rate": 9.119591346153846e-06,
+      "loss": 0.8185,
+      "step": 2822
+    },
+    {
+      "epoch": 1.6467843080064168,
+      "grad_norm": 1.1789501905441284,
+      "learning_rate": 9.104567307692308e-06,
+      "loss": 0.7072,
+      "step": 2823
+    },
+    {
+      "epoch": 1.6473676534927812,
+      "grad_norm": 1.2296137809753418,
+      "learning_rate": 9.08954326923077e-06,
+      "loss": 0.8957,
+      "step": 2824
+    },
+    {
+      "epoch": 1.6479509989791454,
+      "grad_norm": 1.2670032978057861,
+      "learning_rate": 9.07451923076923e-06,
+      "loss": 1.0086,
+      "step": 2825
+    },
+    {
+      "epoch": 1.6485343444655096,
+      "grad_norm": 1.2225745916366577,
+      "learning_rate": 9.059495192307694e-06,
+      "loss": 0.7689,
+      "step": 2826
+    },
+    {
+      "epoch": 1.649117689951874,
+      "grad_norm": 1.3422561883926392,
+      "learning_rate": 9.044471153846154e-06,
+      "loss": 0.7682,
+      "step": 2827
+    },
+    {
+      "epoch": 1.6497010354382384,
+      "grad_norm": 1.7664170265197754,
+      "learning_rate": 9.029447115384616e-06,
+      "loss": 0.9315,
+      "step": 2828
+    },
+    {
+      "epoch": 1.6502843809246026,
+      "grad_norm": 1.141822338104248,
+      "learning_rate": 9.014423076923078e-06,
+      "loss": 0.7782,
+      "step": 2829
+    },
+    {
+      "epoch": 1.6508677264109668,
+      "grad_norm": 1.266010046005249,
+      "learning_rate": 8.99939903846154e-06,
+      "loss": 0.7129,
+      "step": 2830
+    },
+    {
+      "epoch": 1.6514510718973312,
+      "grad_norm": 1.1951196193695068,
+      "learning_rate": 8.984375e-06,
+      "loss": 0.8082,
+      "step": 2831
+    },
+    {
+      "epoch": 1.6520344173836956,
+      "grad_norm": 1.1922125816345215,
+      "learning_rate": 8.969350961538462e-06,
+      "loss": 0.8675,
+      "step": 2832
+    },
+    {
+      "epoch": 1.6526177628700598,
+      "grad_norm": 1.4151067733764648,
+      "learning_rate": 8.954326923076924e-06,
+      "loss": 0.7596,
+      "step": 2833
+    },
+    {
+      "epoch": 1.653201108356424,
+      "grad_norm": 1.496769666671753,
+      "learning_rate": 8.939302884615384e-06,
+      "loss": 0.7633,
+      "step": 2834
+    },
+    {
+      "epoch": 1.6537844538427884,
+      "grad_norm": 1.1096853017807007,
+      "learning_rate": 8.924278846153847e-06,
+      "loss": 0.782,
+      "step": 2835
+    },
+    {
+      "epoch": 1.6543677993291528,
+      "grad_norm": 1.4348713159561157,
+      "learning_rate": 8.909254807692308e-06,
+      "loss": 0.7652,
+      "step": 2836
+    },
+    {
+      "epoch": 1.654951144815517,
+      "grad_norm": 1.3088988065719604,
+      "learning_rate": 8.89423076923077e-06,
+      "loss": 0.7302,
+      "step": 2837
+    },
+    {
+      "epoch": 1.6555344903018812,
+      "grad_norm": 1.1694996356964111,
+      "learning_rate": 8.879206730769231e-06,
+      "loss": 0.9367,
+      "step": 2838
+    },
+    {
+      "epoch": 1.6561178357882456,
+      "grad_norm": 1.274279236793518,
+      "learning_rate": 8.864182692307693e-06,
+      "loss": 0.7277,
+      "step": 2839
+    },
+    {
+      "epoch": 1.65670118127461,
+      "grad_norm": 1.4117878675460815,
+      "learning_rate": 8.849158653846153e-06,
+      "loss": 0.6493,
+      "step": 2840
+    },
+    {
+      "epoch": 1.6572845267609742,
+      "grad_norm": 1.1404685974121094,
+      "learning_rate": 8.834134615384617e-06,
+      "loss": 0.8478,
+      "step": 2841
+    },
+    {
+      "epoch": 1.6578678722473383,
+      "grad_norm": 1.1155657768249512,
+      "learning_rate": 8.819110576923077e-06,
+      "loss": 0.8945,
+      "step": 2842
+    },
+    {
+      "epoch": 1.6584512177337027,
+      "grad_norm": 1.2752269506454468,
+      "learning_rate": 8.804086538461539e-06,
+      "loss": 0.7728,
+      "step": 2843
+    },
+    {
+      "epoch": 1.6590345632200671,
+      "grad_norm": 1.6460522413253784,
+      "learning_rate": 8.789062500000001e-06,
+      "loss": 0.7511,
+      "step": 2844
+    },
+    {
+      "epoch": 1.6596179087064313,
+      "grad_norm": 1.4639447927474976,
+      "learning_rate": 8.774038461538461e-06,
+      "loss": 0.6837,
+      "step": 2845
+    },
+    {
+      "epoch": 1.6602012541927955,
+      "grad_norm": 1.4805471897125244,
+      "learning_rate": 8.759014423076923e-06,
+      "loss": 0.832,
+      "step": 2846
+    },
+    {
+      "epoch": 1.6607845996791601,
+      "grad_norm": 1.1659775972366333,
+      "learning_rate": 8.743990384615385e-06,
+      "loss": 0.7268,
+      "step": 2847
+    },
+    {
+      "epoch": 1.6613679451655243,
+      "grad_norm": 1.190755009651184,
+      "learning_rate": 8.728966346153847e-06,
+      "loss": 0.6614,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6619512906518885,
+      "grad_norm": 1.1789089441299438,
+      "learning_rate": 8.713942307692307e-06,
+      "loss": 0.8071,
+      "step": 2849
+    },
+    {
+      "epoch": 1.662534636138253,
+      "grad_norm": 1.2296158075332642,
+      "learning_rate": 8.69891826923077e-06,
+      "loss": 0.6635,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6631179816246173,
+      "grad_norm": 1.1168346405029297,
+      "learning_rate": 8.683894230769231e-06,
+      "loss": 0.714,
+      "step": 2851
+    },
+    {
+      "epoch": 1.6637013271109815,
+      "grad_norm": 1.1426185369491577,
+      "learning_rate": 8.668870192307693e-06,
+      "loss": 0.7065,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6642846725973457,
+      "grad_norm": 1.2550532817840576,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 0.6332,
+      "step": 2853
+    },
+    {
+      "epoch": 1.66486801808371,
+      "grad_norm": 1.2261812686920166,
+      "learning_rate": 8.638822115384617e-06,
+      "loss": 0.7257,
+      "step": 2854
+    },
+    {
+      "epoch": 1.6654513635700745,
+      "grad_norm": 1.1258302927017212,
+      "learning_rate": 8.623798076923077e-06,
+      "loss": 0.9171,
+      "step": 2855
+    },
+    {
+      "epoch": 1.6660347090564387,
+      "grad_norm": 1.2768467664718628,
+      "learning_rate": 8.60877403846154e-06,
+      "loss": 0.6836,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6666180545428029,
+      "grad_norm": 0.9513541460037231,
+      "learning_rate": 8.59375e-06,
+      "loss": 0.7402,
+      "step": 2857
+    },
+    {
+      "epoch": 1.6672014000291673,
+      "grad_norm": 1.2110565900802612,
+      "learning_rate": 8.57872596153846e-06,
+      "loss": 0.6598,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6677847455155317,
+      "grad_norm": 1.2651914358139038,
+      "learning_rate": 8.563701923076924e-06,
+      "loss": 0.8355,
+      "step": 2859
+    },
+    {
+      "epoch": 1.6683680910018959,
+      "grad_norm": 1.2185043096542358,
+      "learning_rate": 8.548677884615384e-06,
+      "loss": 0.6839,
+      "step": 2860
+    },
+    {
+      "epoch": 1.66895143648826,
+      "grad_norm": 1.2617533206939697,
+      "learning_rate": 8.533653846153846e-06,
+      "loss": 0.8926,
+      "step": 2861
+    },
+    {
+      "epoch": 1.6695347819746245,
+      "grad_norm": 1.3251715898513794,
+      "learning_rate": 8.518629807692308e-06,
+      "loss": 0.956,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6701181274609889,
+      "grad_norm": 1.2607977390289307,
+      "learning_rate": 8.50360576923077e-06,
+      "loss": 0.81,
+      "step": 2863
+    },
+    {
+      "epoch": 1.670701472947353,
+      "grad_norm": 1.0856292247772217,
+      "learning_rate": 8.48858173076923e-06,
+      "loss": 0.8838,
+      "step": 2864
+    },
+    {
+      "epoch": 1.6712848184337172,
+      "grad_norm": 1.4323976039886475,
+      "learning_rate": 8.473557692307694e-06,
+      "loss": 0.908,
+      "step": 2865
+    },
+    {
+      "epoch": 1.6718681639200816,
+      "grad_norm": 1.1892898082733154,
+      "learning_rate": 8.458533653846154e-06,
+      "loss": 0.8645,
+      "step": 2866
+    },
+    {
+      "epoch": 1.672451509406446,
+      "grad_norm": 1.328147530555725,
+      "learning_rate": 8.443509615384616e-06,
+      "loss": 0.7748,
+      "step": 2867
+    },
+    {
+      "epoch": 1.6730348548928102,
+      "grad_norm": 1.4214242696762085,
+      "learning_rate": 8.428485576923078e-06,
+      "loss": 0.8796,
+      "step": 2868
+    },
+    {
+      "epoch": 1.6736182003791744,
+      "grad_norm": 0.941706120967865,
+      "learning_rate": 8.41346153846154e-06,
+      "loss": 0.7776,
+      "step": 2869
+    },
+    {
+      "epoch": 1.6742015458655388,
+      "grad_norm": 1.255053997039795,
+      "learning_rate": 8.3984375e-06,
+      "loss": 0.6698,
+      "step": 2870
+    },
+    {
+      "epoch": 1.6747848913519032,
+      "grad_norm": 1.1494503021240234,
+      "learning_rate": 8.383413461538462e-06,
+      "loss": 0.581,
+      "step": 2871
+    },
+    {
+      "epoch": 1.6753682368382674,
+      "grad_norm": 1.3444379568099976,
+      "learning_rate": 8.368389423076924e-06,
+      "loss": 0.7983,
+      "step": 2872
+    },
+    {
+      "epoch": 1.6759515823246318,
+      "grad_norm": 1.1211199760437012,
+      "learning_rate": 8.353365384615384e-06,
+      "loss": 0.6528,
+      "step": 2873
+    },
+    {
+      "epoch": 1.6765349278109962,
+      "grad_norm": 1.30573570728302,
+      "learning_rate": 8.338341346153848e-06,
+      "loss": 0.7975,
+      "step": 2874
+    },
+    {
+      "epoch": 1.6771182732973604,
+      "grad_norm": 1.0336012840270996,
+      "learning_rate": 8.323317307692308e-06,
+      "loss": 0.5304,
+      "step": 2875
+    },
+    {
+      "epoch": 1.6777016187837246,
+      "grad_norm": 1.2422670125961304,
+      "learning_rate": 8.30829326923077e-06,
+      "loss": 0.6592,
+      "step": 2876
+    },
+    {
+      "epoch": 1.678284964270089,
+      "grad_norm": 1.3174580335617065,
+      "learning_rate": 8.293269230769232e-06,
+      "loss": 0.7867,
+      "step": 2877
+    },
+    {
+      "epoch": 1.6788683097564534,
+      "grad_norm": 1.1110163927078247,
+      "learning_rate": 8.278245192307693e-06,
+      "loss": 0.7349,
+      "step": 2878
+    },
+    {
+      "epoch": 1.6794516552428176,
+      "grad_norm": 1.2657557725906372,
+      "learning_rate": 8.263221153846154e-06,
+      "loss": 0.9434,
+      "step": 2879
+    },
+    {
+      "epoch": 1.6800350007291818,
+      "grad_norm": 1.080836296081543,
+      "learning_rate": 8.248197115384616e-06,
+      "loss": 0.7363,
+      "step": 2880
+    },
+    {
+      "epoch": 1.6806183462155462,
+      "grad_norm": 1.3697772026062012,
+      "learning_rate": 8.233173076923077e-06,
+      "loss": 0.7499,
+      "step": 2881
+    },
+    {
+      "epoch": 1.6812016917019106,
+      "grad_norm": 1.0886021852493286,
+      "learning_rate": 8.21814903846154e-06,
+      "loss": 0.9581,
+      "step": 2882
+    },
+    {
+      "epoch": 1.6817850371882748,
+      "grad_norm": 1.3300132751464844,
+      "learning_rate": 8.203125000000001e-06,
+      "loss": 0.7848,
+      "step": 2883
+    },
+    {
+      "epoch": 1.682368382674639,
+      "grad_norm": 1.4838621616363525,
+      "learning_rate": 8.188100961538461e-06,
+      "loss": 0.9271,
+      "step": 2884
+    },
+    {
+      "epoch": 1.6829517281610034,
+      "grad_norm": 1.3038266897201538,
+      "learning_rate": 8.173076923076923e-06,
+      "loss": 0.7117,
+      "step": 2885
+    },
+    {
+      "epoch": 1.6835350736473678,
+      "grad_norm": 1.1892757415771484,
+      "learning_rate": 8.158052884615385e-06,
+      "loss": 0.5654,
+      "step": 2886
+    },
+    {
+      "epoch": 1.684118419133732,
+      "grad_norm": 1.1837880611419678,
+      "learning_rate": 8.143028846153847e-06,
+      "loss": 0.7745,
+      "step": 2887
+    },
+    {
+      "epoch": 1.6847017646200961,
+      "grad_norm": 1.2857214212417603,
+      "learning_rate": 8.128004807692307e-06,
+      "loss": 0.7224,
+      "step": 2888
+    },
+    {
+      "epoch": 1.6852851101064605,
+      "grad_norm": 1.1604810953140259,
+      "learning_rate": 8.112980769230769e-06,
+      "loss": 0.8615,
+      "step": 2889
+    },
+    {
+      "epoch": 1.685868455592825,
+      "grad_norm": 1.1492291688919067,
+      "learning_rate": 8.097956730769231e-06,
+      "loss": 0.8235,
+      "step": 2890
+    },
+    {
+      "epoch": 1.6864518010791891,
+      "grad_norm": 1.4180827140808105,
+      "learning_rate": 8.082932692307693e-06,
+      "loss": 0.7844,
+      "step": 2891
+    },
+    {
+      "epoch": 1.6870351465655533,
+      "grad_norm": 1.3198400735855103,
+      "learning_rate": 8.067908653846155e-06,
+      "loss": 0.9021,
+      "step": 2892
+    },
+    {
+      "epoch": 1.6876184920519177,
+      "grad_norm": 1.2082802057266235,
+      "learning_rate": 8.052884615384617e-06,
+      "loss": 0.8417,
+      "step": 2893
+    },
+    {
+      "epoch": 1.6882018375382821,
+      "grad_norm": 1.4099825620651245,
+      "learning_rate": 8.037860576923077e-06,
+      "loss": 0.9223,
+      "step": 2894
+    },
+    {
+      "epoch": 1.6887851830246463,
+      "grad_norm": 1.4837796688079834,
+      "learning_rate": 8.022836538461539e-06,
+      "loss": 1.0981,
+      "step": 2895
+    },
+    {
+      "epoch": 1.6893685285110105,
+      "grad_norm": 1.2653106451034546,
+      "learning_rate": 8.0078125e-06,
+      "loss": 0.8805,
+      "step": 2896
+    },
+    {
+      "epoch": 1.689951873997375,
+      "grad_norm": 1.1850988864898682,
+      "learning_rate": 7.992788461538461e-06,
+      "loss": 0.6671,
+      "step": 2897
+    },
+    {
+      "epoch": 1.6905352194837393,
+      "grad_norm": 1.2096939086914062,
+      "learning_rate": 7.977764423076923e-06,
+      "loss": 0.615,
+      "step": 2898
+    },
+    {
+      "epoch": 1.6911185649701035,
+      "grad_norm": 1.1689413785934448,
+      "learning_rate": 7.962740384615385e-06,
+      "loss": 0.7358,
+      "step": 2899
+    },
+    {
+      "epoch": 1.691701910456468,
+      "grad_norm": 1.2683144807815552,
+      "learning_rate": 7.947716346153847e-06,
+      "loss": 0.8217,
+      "step": 2900
+    },
+    {
+      "epoch": 1.6922852559428323,
+      "grad_norm": 1.4966411590576172,
+      "learning_rate": 7.932692307692308e-06,
+      "loss": 0.8058,
+      "step": 2901
+    },
+    {
+      "epoch": 1.6928686014291965,
+      "grad_norm": 1.3604378700256348,
+      "learning_rate": 7.91766826923077e-06,
+      "loss": 0.8571,
+      "step": 2902
+    },
+    {
+      "epoch": 1.6934519469155607,
+      "grad_norm": 1.2549089193344116,
+      "learning_rate": 7.90264423076923e-06,
+      "loss": 0.8095,
+      "step": 2903
+    },
+    {
+      "epoch": 1.694035292401925,
+      "grad_norm": 1.293096899986267,
+      "learning_rate": 7.887620192307692e-06,
+      "loss": 0.888,
+      "step": 2904
+    },
+    {
+      "epoch": 1.6946186378882895,
+      "grad_norm": 1.2265640497207642,
+      "learning_rate": 7.872596153846154e-06,
+      "loss": 0.6778,
+      "step": 2905
+    },
+    {
+      "epoch": 1.6952019833746537,
+      "grad_norm": 1.3105190992355347,
+      "learning_rate": 7.857572115384616e-06,
+      "loss": 0.6253,
+      "step": 2906
+    },
+    {
+      "epoch": 1.6957853288610178,
+      "grad_norm": 1.0808024406433105,
+      "learning_rate": 7.842548076923076e-06,
+      "loss": 0.8923,
+      "step": 2907
+    },
+    {
+      "epoch": 1.6963686743473823,
+      "grad_norm": 1.20291268825531,
+      "learning_rate": 7.82752403846154e-06,
+      "loss": 0.5709,
+      "step": 2908
+    },
+    {
+      "epoch": 1.6969520198337467,
+      "grad_norm": 1.1902101039886475,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.7631,
+      "step": 2909
+    },
+    {
+      "epoch": 1.6975353653201108,
+      "grad_norm": 1.1930304765701294,
+      "learning_rate": 7.797475961538462e-06,
+      "loss": 0.8958,
+      "step": 2910
+    },
+    {
+      "epoch": 1.698118710806475,
+      "grad_norm": 1.1427156925201416,
+      "learning_rate": 7.782451923076924e-06,
+      "loss": 0.8889,
+      "step": 2911
+    },
+    {
+      "epoch": 1.6987020562928394,
+      "grad_norm": 1.172147512435913,
+      "learning_rate": 7.767427884615384e-06,
+      "loss": 0.733,
+      "step": 2912
+    },
+    {
+      "epoch": 1.6992854017792038,
+      "grad_norm": 1.1946724653244019,
+      "learning_rate": 7.752403846153846e-06,
+      "loss": 1.0244,
+      "step": 2913
+    },
+    {
+      "epoch": 1.699868747265568,
+      "grad_norm": 1.224003791809082,
+      "learning_rate": 7.737379807692308e-06,
+      "loss": 0.8716,
+      "step": 2914
+    },
+    {
+      "epoch": 1.7004520927519322,
+      "grad_norm": 1.2379382848739624,
+      "learning_rate": 7.72235576923077e-06,
+      "loss": 0.9171,
+      "step": 2915
+    },
+    {
+      "epoch": 1.7010354382382966,
+      "grad_norm": 1.3675721883773804,
+      "learning_rate": 7.70733173076923e-06,
+      "loss": 0.7572,
+      "step": 2916
+    },
+    {
+      "epoch": 1.701618783724661,
+      "grad_norm": 1.1662015914916992,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.726,
+      "step": 2917
+    },
+    {
+      "epoch": 1.7022021292110252,
+      "grad_norm": 0.9387063980102539,
+      "learning_rate": 7.677283653846154e-06,
+      "loss": 0.8387,
+      "step": 2918
+    },
+    {
+      "epoch": 1.7027854746973894,
+      "grad_norm": 1.1246999502182007,
+      "learning_rate": 7.662259615384616e-06,
+      "loss": 0.7998,
+      "step": 2919
+    },
+    {
+      "epoch": 1.7033688201837538,
+      "grad_norm": 1.011136770248413,
+      "learning_rate": 7.647235576923078e-06,
+      "loss": 0.6432,
+      "step": 2920
+    },
+    {
+      "epoch": 1.7039521656701182,
+      "grad_norm": 1.1951652765274048,
+      "learning_rate": 7.63221153846154e-06,
+      "loss": 0.9636,
+      "step": 2921
+    },
+    {
+      "epoch": 1.7045355111564824,
+      "grad_norm": 1.4355970621109009,
+      "learning_rate": 7.6171875000000005e-06,
+      "loss": 0.6864,
+      "step": 2922
+    },
+    {
+      "epoch": 1.7051188566428466,
+      "grad_norm": 1.2137242555618286,
+      "learning_rate": 7.6021634615384615e-06,
+      "loss": 0.9197,
+      "step": 2923
+    },
+    {
+      "epoch": 1.705702202129211,
+      "grad_norm": 1.3542622327804565,
+      "learning_rate": 7.5871394230769234e-06,
+      "loss": 0.5522,
+      "step": 2924
+    },
+    {
+      "epoch": 1.7062855476155754,
+      "grad_norm": 1.1288126707077026,
+      "learning_rate": 7.5721153846153845e-06,
+      "loss": 0.7477,
+      "step": 2925
+    },
+    {
+      "epoch": 1.7068688931019396,
+      "grad_norm": 1.3344682455062866,
+      "learning_rate": 7.557091346153846e-06,
+      "loss": 0.873,
+      "step": 2926
+    },
+    {
+      "epoch": 1.707452238588304,
+      "grad_norm": 1.3780889511108398,
+      "learning_rate": 7.542067307692307e-06,
+      "loss": 0.7851,
+      "step": 2927
+    },
+    {
+      "epoch": 1.7080355840746684,
+      "grad_norm": 1.1783818006515503,
+      "learning_rate": 7.52704326923077e-06,
+      "loss": 0.8652,
+      "step": 2928
+    },
+    {
+      "epoch": 1.7086189295610326,
+      "grad_norm": 1.1674573421478271,
+      "learning_rate": 7.512019230769231e-06,
+      "loss": 0.8358,
+      "step": 2929
+    },
+    {
+      "epoch": 1.7092022750473967,
+      "grad_norm": 1.1287046670913696,
+      "learning_rate": 7.496995192307693e-06,
+      "loss": 0.7869,
+      "step": 2930
+    },
+    {
+      "epoch": 1.7097856205337612,
+      "grad_norm": 1.240212082862854,
+      "learning_rate": 7.481971153846154e-06,
+      "loss": 0.9021,
+      "step": 2931
+    },
+    {
+      "epoch": 1.7103689660201256,
+      "grad_norm": 1.4593373537063599,
+      "learning_rate": 7.466947115384616e-06,
+      "loss": 0.8768,
+      "step": 2932
+    },
+    {
+      "epoch": 1.7109523115064897,
+      "grad_norm": 1.5059845447540283,
+      "learning_rate": 7.451923076923077e-06,
+      "loss": 0.7657,
+      "step": 2933
+    },
+    {
+      "epoch": 1.711535656992854,
+      "grad_norm": 1.0853300094604492,
+      "learning_rate": 7.43689903846154e-06,
+      "loss": 0.935,
+      "step": 2934
+    },
+    {
+      "epoch": 1.7121190024792183,
+      "grad_norm": 1.382460117340088,
+      "learning_rate": 7.421875e-06,
+      "loss": 0.644,
+      "step": 2935
+    },
+    {
+      "epoch": 1.7127023479655827,
+      "grad_norm": 1.4362154006958008,
+      "learning_rate": 7.406850961538461e-06,
+      "loss": 0.5482,
+      "step": 2936
+    },
+    {
+      "epoch": 1.713285693451947,
+      "grad_norm": 1.2484700679779053,
+      "learning_rate": 7.391826923076924e-06,
+      "loss": 0.799,
+      "step": 2937
+    },
+    {
+      "epoch": 1.713869038938311,
+      "grad_norm": 1.0450493097305298,
+      "learning_rate": 7.376802884615385e-06,
+      "loss": 0.7939,
+      "step": 2938
+    },
+    {
+      "epoch": 1.7144523844246755,
+      "grad_norm": 1.2090156078338623,
+      "learning_rate": 7.361778846153847e-06,
+      "loss": 0.7845,
+      "step": 2939
+    },
+    {
+      "epoch": 1.71503572991104,
+      "grad_norm": 1.0005606412887573,
+      "learning_rate": 7.346754807692308e-06,
+      "loss": 0.8881,
+      "step": 2940
+    },
+    {
+      "epoch": 1.715619075397404,
+      "grad_norm": 1.2529767751693726,
+      "learning_rate": 7.33173076923077e-06,
+      "loss": 0.9444,
+      "step": 2941
+    },
+    {
+      "epoch": 1.7162024208837683,
+      "grad_norm": 1.3556801080703735,
+      "learning_rate": 7.316706730769231e-06,
+      "loss": 0.8145,
+      "step": 2942
+    },
+    {
+      "epoch": 1.7167857663701327,
+      "grad_norm": 1.385504126548767,
+      "learning_rate": 7.301682692307693e-06,
+      "loss": 0.8717,
+      "step": 2943
+    },
+    {
+      "epoch": 1.717369111856497,
+      "grad_norm": 1.2711490392684937,
+      "learning_rate": 7.286658653846154e-06,
+      "loss": 0.6642,
+      "step": 2944
+    },
+    {
+      "epoch": 1.7179524573428613,
+      "grad_norm": 1.199277639389038,
+      "learning_rate": 7.271634615384616e-06,
+      "loss": 0.908,
+      "step": 2945
+    },
+    {
+      "epoch": 1.7185358028292255,
+      "grad_norm": 1.2609366178512573,
+      "learning_rate": 7.256610576923077e-06,
+      "loss": 0.9037,
+      "step": 2946
+    },
+    {
+      "epoch": 1.7191191483155899,
+      "grad_norm": 1.1617000102996826,
+      "learning_rate": 7.241586538461539e-06,
+      "loss": 0.8382,
+      "step": 2947
+    },
+    {
+      "epoch": 1.7197024938019543,
+      "grad_norm": 1.1657629013061523,
+      "learning_rate": 7.2265625e-06,
+      "loss": 1.1246,
+      "step": 2948
+    },
+    {
+      "epoch": 1.7202858392883185,
+      "grad_norm": 1.197189211845398,
+      "learning_rate": 7.211538461538461e-06,
+      "loss": 0.8029,
+      "step": 2949
+    },
+    {
+      "epoch": 1.7208691847746826,
+      "grad_norm": 1.611620306968689,
+      "learning_rate": 7.196514423076923e-06,
+      "loss": 0.9494,
+      "step": 2950
+    },
+    {
+      "epoch": 1.721452530261047,
+      "grad_norm": 1.2203035354614258,
+      "learning_rate": 7.181490384615384e-06,
+      "loss": 1.0667,
+      "step": 2951
+    },
+    {
+      "epoch": 1.7220358757474115,
+      "grad_norm": 1.2028281688690186,
+      "learning_rate": 7.166466346153847e-06,
+      "loss": 0.6483,
+      "step": 2952
+    },
+    {
+      "epoch": 1.7226192212337756,
+      "grad_norm": 1.251402735710144,
+      "learning_rate": 7.151442307692307e-06,
+      "loss": 0.8199,
+      "step": 2953
+    },
+    {
+      "epoch": 1.72320256672014,
+      "grad_norm": 1.2342396974563599,
+      "learning_rate": 7.13641826923077e-06,
+      "loss": 0.7685,
+      "step": 2954
+    },
+    {
+      "epoch": 1.7237859122065045,
+      "grad_norm": 1.1816489696502686,
+      "learning_rate": 7.121394230769231e-06,
+      "loss": 0.7799,
+      "step": 2955
+    },
+    {
+      "epoch": 1.7243692576928686,
+      "grad_norm": 1.373840093612671,
+      "learning_rate": 7.106370192307693e-06,
+      "loss": 0.7768,
+      "step": 2956
+    },
+    {
+      "epoch": 1.7249526031792328,
+      "grad_norm": 1.6701674461364746,
+      "learning_rate": 7.091346153846154e-06,
+      "loss": 0.953,
+      "step": 2957
+    },
+    {
+      "epoch": 1.7255359486655972,
+      "grad_norm": 1.3481673002243042,
+      "learning_rate": 7.076322115384617e-06,
+      "loss": 0.8016,
+      "step": 2958
+    },
+    {
+      "epoch": 1.7261192941519616,
+      "grad_norm": 1.2999407052993774,
+      "learning_rate": 7.061298076923077e-06,
+      "loss": 0.5779,
+      "step": 2959
+    },
+    {
+      "epoch": 1.7267026396383258,
+      "grad_norm": 1.1688950061798096,
+      "learning_rate": 7.04627403846154e-06,
+      "loss": 0.937,
+      "step": 2960
+    },
+    {
+      "epoch": 1.72728598512469,
+      "grad_norm": 1.5214632749557495,
+      "learning_rate": 7.031250000000001e-06,
+      "loss": 0.8384,
+      "step": 2961
+    },
+    {
+      "epoch": 1.7278693306110544,
+      "grad_norm": 1.1693167686462402,
+      "learning_rate": 7.016225961538461e-06,
+      "loss": 0.8144,
+      "step": 2962
+    },
+    {
+      "epoch": 1.7284526760974188,
+      "grad_norm": 1.2496626377105713,
+      "learning_rate": 7.001201923076924e-06,
+      "loss": 0.7753,
+      "step": 2963
+    },
+    {
+      "epoch": 1.729036021583783,
+      "grad_norm": 1.1495461463928223,
+      "learning_rate": 6.986177884615385e-06,
+      "loss": 0.7307,
+      "step": 2964
+    },
+    {
+      "epoch": 1.7296193670701472,
+      "grad_norm": 1.15946626663208,
+      "learning_rate": 6.9711538461538465e-06,
+      "loss": 0.8847,
+      "step": 2965
+    },
+    {
+      "epoch": 1.7302027125565116,
+      "grad_norm": 1.1948374509811401,
+      "learning_rate": 6.9561298076923076e-06,
+      "loss": 0.7088,
+      "step": 2966
+    },
+    {
+      "epoch": 1.730786058042876,
+      "grad_norm": 1.298287272453308,
+      "learning_rate": 6.94110576923077e-06,
+      "loss": 0.6594,
+      "step": 2967
+    },
+    {
+      "epoch": 1.7313694035292402,
+      "grad_norm": 1.1406173706054688,
+      "learning_rate": 6.9260817307692305e-06,
+      "loss": 0.9859,
+      "step": 2968
+    },
+    {
+      "epoch": 1.7319527490156044,
+      "grad_norm": 1.2161551713943481,
+      "learning_rate": 6.911057692307693e-06,
+      "loss": 0.7891,
+      "step": 2969
+    },
+    {
+      "epoch": 1.7325360945019688,
+      "grad_norm": 1.0298594236373901,
+      "learning_rate": 6.896033653846154e-06,
+      "loss": 0.7256,
+      "step": 2970
+    },
+    {
+      "epoch": 1.7331194399883332,
+      "grad_norm": 1.2798943519592285,
+      "learning_rate": 6.881009615384616e-06,
+      "loss": 0.8829,
+      "step": 2971
+    },
+    {
+      "epoch": 1.7337027854746974,
+      "grad_norm": 1.514907717704773,
+      "learning_rate": 6.865985576923077e-06,
+      "loss": 0.7678,
+      "step": 2972
+    },
+    {
+      "epoch": 1.7342861309610615,
+      "grad_norm": 1.2793967723846436,
+      "learning_rate": 6.850961538461539e-06,
+      "loss": 0.8854,
+      "step": 2973
+    },
+    {
+      "epoch": 1.734869476447426,
+      "grad_norm": 1.2280986309051514,
+      "learning_rate": 6.8359375e-06,
+      "loss": 0.9037,
+      "step": 2974
+    },
+    {
+      "epoch": 1.7354528219337904,
+      "grad_norm": 1.116003155708313,
+      "learning_rate": 6.820913461538461e-06,
+      "loss": 0.7507,
+      "step": 2975
+    },
+    {
+      "epoch": 1.7360361674201545,
+      "grad_norm": 1.0788702964782715,
+      "learning_rate": 6.805889423076924e-06,
+      "loss": 0.9569,
+      "step": 2976
+    },
+    {
+      "epoch": 1.7366195129065187,
+      "grad_norm": 1.329489827156067,
+      "learning_rate": 6.790865384615384e-06,
+      "loss": 0.7846,
+      "step": 2977
+    },
+    {
+      "epoch": 1.7372028583928831,
+      "grad_norm": 1.1528865098953247,
+      "learning_rate": 6.775841346153847e-06,
+      "loss": 0.7985,
+      "step": 2978
+    },
+    {
+      "epoch": 1.7377862038792475,
+      "grad_norm": 0.9944823980331421,
+      "learning_rate": 6.760817307692308e-06,
+      "loss": 0.7359,
+      "step": 2979
+    },
+    {
+      "epoch": 1.7383695493656117,
+      "grad_norm": 1.269169569015503,
+      "learning_rate": 6.74579326923077e-06,
+      "loss": 0.7273,
+      "step": 2980
+    },
+    {
+      "epoch": 1.7389528948519761,
+      "grad_norm": 1.0448265075683594,
+      "learning_rate": 6.730769230769231e-06,
+      "loss": 0.8073,
+      "step": 2981
+    },
+    {
+      "epoch": 1.7395362403383405,
+      "grad_norm": 1.3037874698638916,
+      "learning_rate": 6.715745192307693e-06,
+      "loss": 0.6997,
+      "step": 2982
+    },
+    {
+      "epoch": 1.7401195858247047,
+      "grad_norm": 1.2340432405471802,
+      "learning_rate": 6.700721153846154e-06,
+      "loss": 0.8573,
+      "step": 2983
+    },
+    {
+      "epoch": 1.740702931311069,
+      "grad_norm": 1.128672480583191,
+      "learning_rate": 6.6856971153846165e-06,
+      "loss": 0.8413,
+      "step": 2984
+    },
+    {
+      "epoch": 1.7412862767974333,
+      "grad_norm": 1.1064525842666626,
+      "learning_rate": 6.6706730769230775e-06,
+      "loss": 0.9186,
+      "step": 2985
+    },
+    {
+      "epoch": 1.7418696222837977,
+      "grad_norm": 1.9784815311431885,
+      "learning_rate": 6.6556490384615394e-06,
+      "loss": 0.8194,
+      "step": 2986
+    },
+    {
+      "epoch": 1.742452967770162,
+      "grad_norm": 1.0543274879455566,
+      "learning_rate": 6.6406250000000005e-06,
+      "loss": 0.6185,
+      "step": 2987
+    },
+    {
+      "epoch": 1.743036313256526,
+      "grad_norm": 1.2074421644210815,
+      "learning_rate": 6.6256009615384615e-06,
+      "loss": 0.5638,
+      "step": 2988
+    },
+    {
+      "epoch": 1.7436196587428905,
+      "grad_norm": 0.9913759231567383,
+      "learning_rate": 6.610576923076923e-06,
+      "loss": 0.7278,
+      "step": 2989
+    },
+    {
+      "epoch": 1.744203004229255,
+      "grad_norm": 1.276524543762207,
+      "learning_rate": 6.5955528846153845e-06,
+      "loss": 0.7934,
+      "step": 2990
+    },
+    {
+      "epoch": 1.744786349715619,
+      "grad_norm": 1.247419834136963,
+      "learning_rate": 6.580528846153846e-06,
+      "loss": 0.8914,
+      "step": 2991
+    },
+    {
+      "epoch": 1.7453696952019833,
+      "grad_norm": 1.3992359638214111,
+      "learning_rate": 6.565504807692307e-06,
+      "loss": 0.7806,
+      "step": 2992
+    },
+    {
+      "epoch": 1.7459530406883477,
+      "grad_norm": 1.1906291246414185,
+      "learning_rate": 6.55048076923077e-06,
+      "loss": 0.8607,
+      "step": 2993
+    },
+    {
+      "epoch": 1.746536386174712,
+      "grad_norm": 1.327694296836853,
+      "learning_rate": 6.535456730769231e-06,
+      "loss": 0.606,
+      "step": 2994
+    },
+    {
+      "epoch": 1.7471197316610763,
+      "grad_norm": 1.3064364194869995,
+      "learning_rate": 6.520432692307693e-06,
+      "loss": 0.8269,
+      "step": 2995
+    },
+    {
+      "epoch": 1.7477030771474404,
+      "grad_norm": 1.3761636018753052,
+      "learning_rate": 6.505408653846154e-06,
+      "loss": 0.8794,
+      "step": 2996
+    },
+    {
+      "epoch": 1.7482864226338048,
+      "grad_norm": 1.2299649715423584,
+      "learning_rate": 6.490384615384616e-06,
+      "loss": 0.9386,
+      "step": 2997
+    },
+    {
+      "epoch": 1.7488697681201693,
+      "grad_norm": 1.324949026107788,
+      "learning_rate": 6.475360576923077e-06,
+      "loss": 0.8732,
+      "step": 2998
+    },
+    {
+      "epoch": 1.7494531136065334,
+      "grad_norm": 1.2557692527770996,
+      "learning_rate": 6.46033653846154e-06,
+      "loss": 0.6888,
+      "step": 2999
+    },
+    {
+      "epoch": 1.7500364590928976,
+      "grad_norm": 1.2142000198364258,
+      "learning_rate": 6.4453125e-06,
+      "loss": 0.828,
+      "step": 3000
+    },
+    {
+      "epoch": 1.7500364590928976,
+      "eval_loss_squad": 0.852206126167439,
+      "eval_perplexity": 8.171628834408976,
+      "eval_perplexity_reconstruct": 1.8986877809494127,
+      "step": 3000
+    },
+    {
+      "epoch": 1.750619804579262,
+      "grad_norm": 1.2730783224105835,
+      "learning_rate": 6.430288461538461e-06,
+      "loss": 0.8912,
+      "step": 3001
+    },
+    {
+      "epoch": 1.7512031500656264,
+      "grad_norm": 1.2231379747390747,
+      "learning_rate": 6.415264423076924e-06,
+      "loss": 0.8677,
+      "step": 3002
+    },
+    {
+      "epoch": 1.7517864955519906,
+      "grad_norm": 1.3539924621582031,
+      "learning_rate": 6.400240384615385e-06,
+      "loss": 0.8291,
+      "step": 3003
+    },
+    {
+      "epoch": 1.7523698410383548,
+      "grad_norm": 1.3997881412506104,
+      "learning_rate": 6.385216346153847e-06,
+      "loss": 0.9718,
+      "step": 3004
+    },
+    {
+      "epoch": 1.7529531865247194,
+      "grad_norm": 1.2460448741912842,
+      "learning_rate": 6.370192307692308e-06,
+      "loss": 0.9308,
+      "step": 3005
+    },
+    {
+      "epoch": 1.7535365320110836,
+      "grad_norm": 1.063209056854248,
+      "learning_rate": 6.35516826923077e-06,
+      "loss": 0.637,
+      "step": 3006
+    },
+    {
+      "epoch": 1.7541198774974478,
+      "grad_norm": 1.0733965635299683,
+      "learning_rate": 6.340144230769231e-06,
+      "loss": 0.7837,
+      "step": 3007
+    },
+    {
+      "epoch": 1.7547032229838122,
+      "grad_norm": 1.167663335800171,
+      "learning_rate": 6.325120192307693e-06,
+      "loss": 0.8921,
+      "step": 3008
+    },
+    {
+      "epoch": 1.7552865684701766,
+      "grad_norm": 1.3398858308792114,
+      "learning_rate": 6.310096153846154e-06,
+      "loss": 0.7743,
+      "step": 3009
+    },
+    {
+      "epoch": 1.7558699139565408,
+      "grad_norm": 1.180702805519104,
+      "learning_rate": 6.295072115384616e-06,
+      "loss": 0.6499,
+      "step": 3010
+    },
+    {
+      "epoch": 1.756453259442905,
+      "grad_norm": 1.4334017038345337,
+      "learning_rate": 6.280048076923077e-06,
+      "loss": 0.6341,
+      "step": 3011
+    },
+    {
+      "epoch": 1.7570366049292694,
+      "grad_norm": 1.1912078857421875,
+      "learning_rate": 6.265024038461539e-06,
+      "loss": 0.6193,
+      "step": 3012
+    },
+    {
+      "epoch": 1.7576199504156338,
+      "grad_norm": 1.267592430114746,
+      "learning_rate": 6.25e-06,
+      "loss": 0.8952,
+      "step": 3013
+    },
+    {
+      "epoch": 1.758203295901998,
+      "grad_norm": 1.6982357501983643,
+      "learning_rate": 6.234975961538462e-06,
+      "loss": 0.8871,
+      "step": 3014
+    },
+    {
+      "epoch": 1.7587866413883622,
+      "grad_norm": 1.41012704372406,
+      "learning_rate": 6.219951923076923e-06,
+      "loss": 0.7593,
+      "step": 3015
+    },
+    {
+      "epoch": 1.7593699868747266,
+      "grad_norm": 1.222355842590332,
+      "learning_rate": 6.204927884615385e-06,
+      "loss": 0.8029,
+      "step": 3016
+    },
+    {
+      "epoch": 1.759953332361091,
+      "grad_norm": 1.1510112285614014,
+      "learning_rate": 6.189903846153847e-06,
+      "loss": 0.7361,
+      "step": 3017
+    },
+    {
+      "epoch": 1.7605366778474552,
+      "grad_norm": 1.1464331150054932,
+      "learning_rate": 6.174879807692308e-06,
+      "loss": 0.6445,
+      "step": 3018
+    },
+    {
+      "epoch": 1.7611200233338193,
+      "grad_norm": 1.6973276138305664,
+      "learning_rate": 6.15985576923077e-06,
+      "loss": 0.7339,
+      "step": 3019
+    },
+    {
+      "epoch": 1.7617033688201837,
+      "grad_norm": 1.2533224821090698,
+      "learning_rate": 6.144831730769231e-06,
+      "loss": 0.735,
+      "step": 3020
+    },
+    {
+      "epoch": 1.7622867143065482,
+      "grad_norm": 0.871324360370636,
+      "learning_rate": 6.129807692307692e-06,
+      "loss": 0.7549,
+      "step": 3021
+    },
+    {
+      "epoch": 1.7628700597929123,
+      "grad_norm": 1.1804686784744263,
+      "learning_rate": 6.114783653846154e-06,
+      "loss": 0.9045,
+      "step": 3022
+    },
+    {
+      "epoch": 1.7634534052792765,
+      "grad_norm": 1.0688856840133667,
+      "learning_rate": 6.099759615384616e-06,
+      "loss": 0.965,
+      "step": 3023
+    },
+    {
+      "epoch": 1.764036750765641,
+      "grad_norm": 1.29346764087677,
+      "learning_rate": 6.084735576923077e-06,
+      "loss": 0.8251,
+      "step": 3024
+    },
+    {
+      "epoch": 1.7646200962520053,
+      "grad_norm": 1.247200846672058,
+      "learning_rate": 6.069711538461539e-06,
+      "loss": 0.8057,
+      "step": 3025
+    },
+    {
+      "epoch": 1.7652034417383695,
+      "grad_norm": 1.3537094593048096,
+      "learning_rate": 6.054687500000001e-06,
+      "loss": 0.7938,
+      "step": 3026
+    },
+    {
+      "epoch": 1.7657867872247337,
+      "grad_norm": 1.2823100090026855,
+      "learning_rate": 6.039663461538462e-06,
+      "loss": 0.9466,
+      "step": 3027
+    },
+    {
+      "epoch": 1.766370132711098,
+      "grad_norm": 1.4946316480636597,
+      "learning_rate": 6.0246394230769236e-06,
+      "loss": 0.7986,
+      "step": 3028
+    },
+    {
+      "epoch": 1.7669534781974625,
+      "grad_norm": 1.1807023286819458,
+      "learning_rate": 6.0096153846153855e-06,
+      "loss": 0.7605,
+      "step": 3029
+    },
+    {
+      "epoch": 1.7675368236838267,
+      "grad_norm": 1.2204418182373047,
+      "learning_rate": 5.9945913461538465e-06,
+      "loss": 0.8484,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7681201691701909,
+      "grad_norm": 0.9801056981086731,
+      "learning_rate": 5.979567307692308e-06,
+      "loss": 0.8015,
+      "step": 3031
+    },
+    {
+      "epoch": 1.7687035146565555,
+      "grad_norm": 1.2062498331069946,
+      "learning_rate": 5.9645432692307694e-06,
+      "loss": 0.7702,
+      "step": 3032
+    },
+    {
+      "epoch": 1.7692868601429197,
+      "grad_norm": 1.5518649816513062,
+      "learning_rate": 5.9495192307692305e-06,
+      "loss": 0.9168,
+      "step": 3033
+    },
+    {
+      "epoch": 1.7698702056292839,
+      "grad_norm": 1.196702480316162,
+      "learning_rate": 5.934495192307692e-06,
+      "loss": 0.7405,
+      "step": 3034
+    },
+    {
+      "epoch": 1.7704535511156483,
+      "grad_norm": 1.3550660610198975,
+      "learning_rate": 5.919471153846154e-06,
+      "loss": 0.8246,
+      "step": 3035
+    },
+    {
+      "epoch": 1.7710368966020127,
+      "grad_norm": 1.0379894971847534,
+      "learning_rate": 5.904447115384615e-06,
+      "loss": 0.7812,
+      "step": 3036
+    },
+    {
+      "epoch": 1.7716202420883769,
+      "grad_norm": 1.0726555585861206,
+      "learning_rate": 5.889423076923077e-06,
+      "loss": 1.0311,
+      "step": 3037
+    },
+    {
+      "epoch": 1.772203587574741,
+      "grad_norm": 1.0926289558410645,
+      "learning_rate": 5.874399038461539e-06,
+      "loss": 0.8649,
+      "step": 3038
+    },
+    {
+      "epoch": 1.7727869330611055,
+      "grad_norm": 1.2088254690170288,
+      "learning_rate": 5.859375e-06,
+      "loss": 0.6787,
+      "step": 3039
+    },
+    {
+      "epoch": 1.7733702785474699,
+      "grad_norm": 1.2711060047149658,
+      "learning_rate": 5.844350961538462e-06,
+      "loss": 0.6789,
+      "step": 3040
+    },
+    {
+      "epoch": 1.773953624033834,
+      "grad_norm": 1.2558777332305908,
+      "learning_rate": 5.829326923076923e-06,
+      "loss": 0.8561,
+      "step": 3041
+    },
+    {
+      "epoch": 1.7745369695201982,
+      "grad_norm": 1.1498215198516846,
+      "learning_rate": 5.814302884615385e-06,
+      "loss": 0.9378,
+      "step": 3042
+    },
+    {
+      "epoch": 1.7751203150065626,
+      "grad_norm": 1.2702573537826538,
+      "learning_rate": 5.799278846153847e-06,
+      "loss": 0.7744,
+      "step": 3043
+    },
+    {
+      "epoch": 1.775703660492927,
+      "grad_norm": 1.2708110809326172,
+      "learning_rate": 5.784254807692308e-06,
+      "loss": 0.8994,
+      "step": 3044
+    },
+    {
+      "epoch": 1.7762870059792912,
+      "grad_norm": 1.1213219165802002,
+      "learning_rate": 5.76923076923077e-06,
+      "loss": 0.5506,
+      "step": 3045
+    },
+    {
+      "epoch": 1.7768703514656554,
+      "grad_norm": 1.5344654321670532,
+      "learning_rate": 5.754206730769231e-06,
+      "loss": 0.8867,
+      "step": 3046
+    },
+    {
+      "epoch": 1.7774536969520198,
+      "grad_norm": 1.2419880628585815,
+      "learning_rate": 5.739182692307693e-06,
+      "loss": 0.7802,
+      "step": 3047
+    },
+    {
+      "epoch": 1.7780370424383842,
+      "grad_norm": 1.1512519121170044,
+      "learning_rate": 5.724158653846154e-06,
+      "loss": 0.6938,
+      "step": 3048
+    },
+    {
+      "epoch": 1.7786203879247484,
+      "grad_norm": 1.4187743663787842,
+      "learning_rate": 5.709134615384616e-06,
+      "loss": 0.7221,
+      "step": 3049
+    },
+    {
+      "epoch": 1.7792037334111126,
+      "grad_norm": 1.180336594581604,
+      "learning_rate": 5.694110576923077e-06,
+      "loss": 0.9465,
+      "step": 3050
+    },
+    {
+      "epoch": 1.779787078897477,
+      "grad_norm": 1.244598627090454,
+      "learning_rate": 5.6790865384615386e-06,
+      "loss": 0.7442,
+      "step": 3051
+    },
+    {
+      "epoch": 1.7803704243838414,
+      "grad_norm": 1.2341411113739014,
+      "learning_rate": 5.6640625000000005e-06,
+      "loss": 0.748,
+      "step": 3052
+    },
+    {
+      "epoch": 1.7809537698702056,
+      "grad_norm": 1.1170315742492676,
+      "learning_rate": 5.6490384615384615e-06,
+      "loss": 0.7429,
+      "step": 3053
+    },
+    {
+      "epoch": 1.7815371153565698,
+      "grad_norm": 1.3362447023391724,
+      "learning_rate": 5.634014423076923e-06,
+      "loss": 0.7197,
+      "step": 3054
+    },
+    {
+      "epoch": 1.7821204608429342,
+      "grad_norm": 1.4332809448242188,
+      "learning_rate": 5.618990384615385e-06,
+      "loss": 0.9911,
+      "step": 3055
+    },
+    {
+      "epoch": 1.7827038063292986,
+      "grad_norm": 1.2477706670761108,
+      "learning_rate": 5.603966346153846e-06,
+      "loss": 1.1378,
+      "step": 3056
+    },
+    {
+      "epoch": 1.7832871518156628,
+      "grad_norm": 1.2143217325210571,
+      "learning_rate": 5.588942307692308e-06,
+      "loss": 0.6478,
+      "step": 3057
+    },
+    {
+      "epoch": 1.7838704973020272,
+      "grad_norm": 1.2848975658416748,
+      "learning_rate": 5.57391826923077e-06,
+      "loss": 0.8657,
+      "step": 3058
+    },
+    {
+      "epoch": 1.7844538427883916,
+      "grad_norm": 1.3641600608825684,
+      "learning_rate": 5.55889423076923e-06,
+      "loss": 0.9904,
+      "step": 3059
+    },
+    {
+      "epoch": 1.7850371882747558,
+      "grad_norm": 1.3162271976470947,
+      "learning_rate": 5.543870192307692e-06,
+      "loss": 0.6504,
+      "step": 3060
+    },
+    {
+      "epoch": 1.78562053376112,
+      "grad_norm": 1.1452566385269165,
+      "learning_rate": 5.528846153846154e-06,
+      "loss": 0.8217,
+      "step": 3061
+    },
+    {
+      "epoch": 1.7862038792474844,
+      "grad_norm": 1.2015888690948486,
+      "learning_rate": 5.513822115384615e-06,
+      "loss": 0.9543,
+      "step": 3062
+    },
+    {
+      "epoch": 1.7867872247338488,
+      "grad_norm": 1.3449658155441284,
+      "learning_rate": 5.498798076923077e-06,
+      "loss": 0.8232,
+      "step": 3063
+    },
+    {
+      "epoch": 1.787370570220213,
+      "grad_norm": 1.3045237064361572,
+      "learning_rate": 5.483774038461539e-06,
+      "loss": 0.8037,
+      "step": 3064
+    },
+    {
+      "epoch": 1.7879539157065771,
+      "grad_norm": 1.1564418077468872,
+      "learning_rate": 5.46875e-06,
+      "loss": 0.7541,
+      "step": 3065
+    },
+    {
+      "epoch": 1.7885372611929415,
+      "grad_norm": 1.2231559753417969,
+      "learning_rate": 5.453725961538462e-06,
+      "loss": 0.7623,
+      "step": 3066
+    },
+    {
+      "epoch": 1.789120606679306,
+      "grad_norm": 1.238562822341919,
+      "learning_rate": 5.438701923076924e-06,
+      "loss": 0.8263,
+      "step": 3067
+    },
+    {
+      "epoch": 1.7897039521656701,
+      "grad_norm": 1.3414685726165771,
+      "learning_rate": 5.423677884615385e-06,
+      "loss": 0.8375,
+      "step": 3068
+    },
+    {
+      "epoch": 1.7902872976520343,
+      "grad_norm": 1.1576594114303589,
+      "learning_rate": 5.408653846153847e-06,
+      "loss": 0.8616,
+      "step": 3069
+    },
+    {
+      "epoch": 1.7908706431383987,
+      "grad_norm": 1.2848812341690063,
+      "learning_rate": 5.3936298076923085e-06,
+      "loss": 0.8936,
+      "step": 3070
+    },
+    {
+      "epoch": 1.7914539886247631,
+      "grad_norm": 1.2974978685379028,
+      "learning_rate": 5.37860576923077e-06,
+      "loss": 0.7183,
+      "step": 3071
+    },
+    {
+      "epoch": 1.7920373341111273,
+      "grad_norm": 1.0673803091049194,
+      "learning_rate": 5.363581730769231e-06,
+      "loss": 0.9361,
+      "step": 3072
+    },
+    {
+      "epoch": 1.7926206795974915,
+      "grad_norm": 1.156964659690857,
+      "learning_rate": 5.3485576923076925e-06,
+      "loss": 0.665,
+      "step": 3073
+    },
+    {
+      "epoch": 1.793204025083856,
+      "grad_norm": 1.2498539686203003,
+      "learning_rate": 5.3335336538461536e-06,
+      "loss": 0.683,
+      "step": 3074
+    },
+    {
+      "epoch": 1.7937873705702203,
+      "grad_norm": 1.0622308254241943,
+      "learning_rate": 5.3185096153846155e-06,
+      "loss": 0.7311,
+      "step": 3075
+    },
+    {
+      "epoch": 1.7943707160565845,
+      "grad_norm": 1.147762417793274,
+      "learning_rate": 5.303485576923077e-06,
+      "loss": 0.9415,
+      "step": 3076
+    },
+    {
+      "epoch": 1.7949540615429487,
+      "grad_norm": 1.6700465679168701,
+      "learning_rate": 5.288461538461538e-06,
+      "loss": 0.7959,
+      "step": 3077
+    },
+    {
+      "epoch": 1.795537407029313,
+      "grad_norm": 1.326983094215393,
+      "learning_rate": 5.2734375e-06,
+      "loss": 0.6907,
+      "step": 3078
+    },
+    {
+      "epoch": 1.7961207525156775,
+      "grad_norm": 1.2017778158187866,
+      "learning_rate": 5.258413461538462e-06,
+      "loss": 0.7195,
+      "step": 3079
+    },
+    {
+      "epoch": 1.7967040980020417,
+      "grad_norm": 1.3539180755615234,
+      "learning_rate": 5.243389423076923e-06,
+      "loss": 0.7981,
+      "step": 3080
+    },
+    {
+      "epoch": 1.7972874434884059,
+      "grad_norm": 1.202142357826233,
+      "learning_rate": 5.228365384615385e-06,
+      "loss": 0.8899,
+      "step": 3081
+    },
+    {
+      "epoch": 1.7978707889747703,
+      "grad_norm": 1.4579527378082275,
+      "learning_rate": 5.213341346153847e-06,
+      "loss": 0.7488,
+      "step": 3082
+    },
+    {
+      "epoch": 1.7984541344611347,
+      "grad_norm": 1.3240573406219482,
+      "learning_rate": 5.198317307692308e-06,
+      "loss": 0.6901,
+      "step": 3083
+    },
+    {
+      "epoch": 1.7990374799474989,
+      "grad_norm": 0.9925790429115295,
+      "learning_rate": 5.18329326923077e-06,
+      "loss": 0.7306,
+      "step": 3084
+    },
+    {
+      "epoch": 1.7996208254338633,
+      "grad_norm": 1.2076115608215332,
+      "learning_rate": 5.168269230769231e-06,
+      "loss": 0.7139,
+      "step": 3085
+    },
+    {
+      "epoch": 1.8002041709202277,
+      "grad_norm": 1.3439483642578125,
+      "learning_rate": 5.153245192307692e-06,
+      "loss": 0.7188,
+      "step": 3086
+    },
+    {
+      "epoch": 1.8007875164065918,
+      "grad_norm": 1.2917673587799072,
+      "learning_rate": 5.138221153846154e-06,
+      "loss": 0.8389,
+      "step": 3087
+    },
+    {
+      "epoch": 1.801370861892956,
+      "grad_norm": 1.1600747108459473,
+      "learning_rate": 5.123197115384616e-06,
+      "loss": 0.8402,
+      "step": 3088
+    },
+    {
+      "epoch": 1.8019542073793204,
+      "grad_norm": 1.4509084224700928,
+      "learning_rate": 5.108173076923077e-06,
+      "loss": 0.8959,
+      "step": 3089
+    },
+    {
+      "epoch": 1.8025375528656848,
+      "grad_norm": 1.1624780893325806,
+      "learning_rate": 5.093149038461539e-06,
+      "loss": 0.7454,
+      "step": 3090
+    },
+    {
+      "epoch": 1.803120898352049,
+      "grad_norm": 1.2352399826049805,
+      "learning_rate": 5.078125000000001e-06,
+      "loss": 0.8392,
+      "step": 3091
+    },
+    {
+      "epoch": 1.8037042438384132,
+      "grad_norm": 1.2602814435958862,
+      "learning_rate": 5.063100961538462e-06,
+      "loss": 0.6345,
+      "step": 3092
+    },
+    {
+      "epoch": 1.8042875893247776,
+      "grad_norm": 1.5061650276184082,
+      "learning_rate": 5.0480769230769235e-06,
+      "loss": 0.6893,
+      "step": 3093
+    },
+    {
+      "epoch": 1.804870934811142,
+      "grad_norm": 1.175838828086853,
+      "learning_rate": 5.0330528846153854e-06,
+      "loss": 0.7926,
+      "step": 3094
+    },
+    {
+      "epoch": 1.8054542802975062,
+      "grad_norm": 1.4317418336868286,
+      "learning_rate": 5.0180288461538465e-06,
+      "loss": 0.8662,
+      "step": 3095
+    },
+    {
+      "epoch": 1.8060376257838704,
+      "grad_norm": 1.0850125551223755,
+      "learning_rate": 5.003004807692308e-06,
+      "loss": 0.67,
+      "step": 3096
+    },
+    {
+      "epoch": 1.8066209712702348,
+      "grad_norm": 1.199704885482788,
+      "learning_rate": 4.987980769230769e-06,
+      "loss": 1.0953,
+      "step": 3097
+    },
+    {
+      "epoch": 1.8072043167565992,
+      "grad_norm": 1.1284786462783813,
+      "learning_rate": 4.9729567307692305e-06,
+      "loss": 0.8975,
+      "step": 3098
+    },
+    {
+      "epoch": 1.8077876622429634,
+      "grad_norm": 1.3013429641723633,
+      "learning_rate": 4.957932692307692e-06,
+      "loss": 0.7548,
+      "step": 3099
+    },
+    {
+      "epoch": 1.8083710077293276,
+      "grad_norm": 1.3402200937271118,
+      "learning_rate": 4.942908653846154e-06,
+      "loss": 0.7116,
+      "step": 3100
+    },
+    {
+      "epoch": 1.808954353215692,
+      "grad_norm": 1.1357074975967407,
+      "learning_rate": 4.927884615384615e-06,
+      "loss": 0.747,
+      "step": 3101
+    },
+    {
+      "epoch": 1.8095376987020564,
+      "grad_norm": 1.097558856010437,
+      "learning_rate": 4.912860576923077e-06,
+      "loss": 0.5583,
+      "step": 3102
+    },
+    {
+      "epoch": 1.8101210441884206,
+      "grad_norm": 1.726578712463379,
+      "learning_rate": 4.897836538461539e-06,
+      "loss": 0.9862,
+      "step": 3103
+    },
+    {
+      "epoch": 1.8107043896747848,
+      "grad_norm": 1.3363982439041138,
+      "learning_rate": 4.8828125e-06,
+      "loss": 0.7787,
+      "step": 3104
+    },
+    {
+      "epoch": 1.8112877351611492,
+      "grad_norm": 1.1213432550430298,
+      "learning_rate": 4.867788461538462e-06,
+      "loss": 0.9938,
+      "step": 3105
+    },
+    {
+      "epoch": 1.8118710806475136,
+      "grad_norm": 1.5810742378234863,
+      "learning_rate": 4.852764423076923e-06,
+      "loss": 0.8159,
+      "step": 3106
+    },
+    {
+      "epoch": 1.8124544261338777,
+      "grad_norm": 1.1917210817337036,
+      "learning_rate": 4.837740384615385e-06,
+      "loss": 0.8894,
+      "step": 3107
+    },
+    {
+      "epoch": 1.813037771620242,
+      "grad_norm": 1.112142562866211,
+      "learning_rate": 4.822716346153847e-06,
+      "loss": 0.7393,
+      "step": 3108
+    },
+    {
+      "epoch": 1.8136211171066063,
+      "grad_norm": 1.3024792671203613,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.7867,
+      "step": 3109
+    },
+    {
+      "epoch": 1.8142044625929707,
+      "grad_norm": 1.378227949142456,
+      "learning_rate": 4.79266826923077e-06,
+      "loss": 0.6839,
+      "step": 3110
+    },
+    {
+      "epoch": 1.814787808079335,
+      "grad_norm": 1.1621403694152832,
+      "learning_rate": 4.777644230769231e-06,
+      "loss": 0.7255,
+      "step": 3111
+    },
+    {
+      "epoch": 1.8153711535656993,
+      "grad_norm": 1.3222122192382812,
+      "learning_rate": 4.762620192307693e-06,
+      "loss": 0.7682,
+      "step": 3112
+    },
+    {
+      "epoch": 1.8159544990520637,
+      "grad_norm": 1.3286000490188599,
+      "learning_rate": 4.747596153846154e-06,
+      "loss": 0.7242,
+      "step": 3113
+    },
+    {
+      "epoch": 1.816537844538428,
+      "grad_norm": 1.1924617290496826,
+      "learning_rate": 4.732572115384616e-06,
+      "loss": 0.9124,
+      "step": 3114
+    },
+    {
+      "epoch": 1.817121190024792,
+      "grad_norm": 1.5701050758361816,
+      "learning_rate": 4.717548076923077e-06,
+      "loss": 0.9612,
+      "step": 3115
+    },
+    {
+      "epoch": 1.8177045355111565,
+      "grad_norm": 0.9868125319480896,
+      "learning_rate": 4.7025240384615385e-06,
+      "loss": 0.672,
+      "step": 3116
+    },
+    {
+      "epoch": 1.818287880997521,
+      "grad_norm": 1.411771535873413,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 0.7319,
+      "step": 3117
+    },
+    {
+      "epoch": 1.818871226483885,
+      "grad_norm": 1.2565867900848389,
+      "learning_rate": 4.6724759615384615e-06,
+      "loss": 0.6905,
+      "step": 3118
+    },
+    {
+      "epoch": 1.8194545719702493,
+      "grad_norm": 1.527400016784668,
+      "learning_rate": 4.657451923076923e-06,
+      "loss": 0.7905,
+      "step": 3119
+    },
+    {
+      "epoch": 1.8200379174566137,
+      "grad_norm": 1.2910518646240234,
+      "learning_rate": 4.642427884615385e-06,
+      "loss": 0.7373,
+      "step": 3120
+    },
+    {
+      "epoch": 1.820621262942978,
+      "grad_norm": 1.4282567501068115,
+      "learning_rate": 4.627403846153846e-06,
+      "loss": 0.8244,
+      "step": 3121
+    },
+    {
+      "epoch": 1.8212046084293423,
+      "grad_norm": 1.2436426877975464,
+      "learning_rate": 4.612379807692308e-06,
+      "loss": 0.8656,
+      "step": 3122
+    },
+    {
+      "epoch": 1.8217879539157065,
+      "grad_norm": 1.2247263193130493,
+      "learning_rate": 4.59735576923077e-06,
+      "loss": 0.8739,
+      "step": 3123
+    },
+    {
+      "epoch": 1.8223712994020709,
+      "grad_norm": 1.359684705734253,
+      "learning_rate": 4.58233173076923e-06,
+      "loss": 0.6765,
+      "step": 3124
+    },
+    {
+      "epoch": 1.8229546448884353,
+      "grad_norm": 1.2492992877960205,
+      "learning_rate": 4.567307692307692e-06,
+      "loss": 0.6232,
+      "step": 3125
+    },
+    {
+      "epoch": 1.8235379903747995,
+      "grad_norm": 1.314278244972229,
+      "learning_rate": 4.552283653846154e-06,
+      "loss": 0.7352,
+      "step": 3126
+    },
+    {
+      "epoch": 1.8241213358611637,
+      "grad_norm": 1.5234570503234863,
+      "learning_rate": 4.537259615384615e-06,
+      "loss": 0.9334,
+      "step": 3127
+    },
+    {
+      "epoch": 1.824704681347528,
+      "grad_norm": 1.2035748958587646,
+      "learning_rate": 4.522235576923077e-06,
+      "loss": 0.8481,
+      "step": 3128
+    },
+    {
+      "epoch": 1.8252880268338925,
+      "grad_norm": 1.1306325197219849,
+      "learning_rate": 4.507211538461539e-06,
+      "loss": 0.7214,
+      "step": 3129
+    },
+    {
+      "epoch": 1.8258713723202566,
+      "grad_norm": 1.265943169593811,
+      "learning_rate": 4.4921875e-06,
+      "loss": 0.7674,
+      "step": 3130
+    },
+    {
+      "epoch": 1.8264547178066208,
+      "grad_norm": 1.22818124294281,
+      "learning_rate": 4.477163461538462e-06,
+      "loss": 0.9115,
+      "step": 3131
+    },
+    {
+      "epoch": 1.8270380632929852,
+      "grad_norm": 1.2925962209701538,
+      "learning_rate": 4.462139423076924e-06,
+      "loss": 1.0034,
+      "step": 3132
+    },
+    {
+      "epoch": 1.8276214087793496,
+      "grad_norm": 1.3289649486541748,
+      "learning_rate": 4.447115384615385e-06,
+      "loss": 0.8526,
+      "step": 3133
+    },
+    {
+      "epoch": 1.8282047542657138,
+      "grad_norm": 1.2596584558486938,
+      "learning_rate": 4.432091346153847e-06,
+      "loss": 0.8599,
+      "step": 3134
+    },
+    {
+      "epoch": 1.828788099752078,
+      "grad_norm": 1.5771028995513916,
+      "learning_rate": 4.4170673076923085e-06,
+      "loss": 0.7632,
+      "step": 3135
+    },
+    {
+      "epoch": 1.8293714452384424,
+      "grad_norm": 1.3541091680526733,
+      "learning_rate": 4.4020432692307696e-06,
+      "loss": 0.7655,
+      "step": 3136
+    },
+    {
+      "epoch": 1.8299547907248068,
+      "grad_norm": 1.2546583414077759,
+      "learning_rate": 4.387019230769231e-06,
+      "loss": 0.7144,
+      "step": 3137
+    },
+    {
+      "epoch": 1.830538136211171,
+      "grad_norm": 1.216031789779663,
+      "learning_rate": 4.3719951923076925e-06,
+      "loss": 0.8208,
+      "step": 3138
+    },
+    {
+      "epoch": 1.8311214816975354,
+      "grad_norm": 1.219329595565796,
+      "learning_rate": 4.3569711538461535e-06,
+      "loss": 0.5568,
+      "step": 3139
+    },
+    {
+      "epoch": 1.8317048271838998,
+      "grad_norm": 1.1087514162063599,
+      "learning_rate": 4.3419471153846154e-06,
+      "loss": 0.9815,
+      "step": 3140
+    },
+    {
+      "epoch": 1.832288172670264,
+      "grad_norm": 1.2218693494796753,
+      "learning_rate": 4.326923076923077e-06,
+      "loss": 0.7445,
+      "step": 3141
+    },
+    {
+      "epoch": 1.8328715181566282,
+      "grad_norm": 1.487666130065918,
+      "learning_rate": 4.311899038461538e-06,
+      "loss": 0.8892,
+      "step": 3142
+    },
+    {
+      "epoch": 1.8334548636429926,
+      "grad_norm": 1.117554783821106,
+      "learning_rate": 4.296875e-06,
+      "loss": 0.6522,
+      "step": 3143
+    },
+    {
+      "epoch": 1.834038209129357,
+      "grad_norm": 1.2432799339294434,
+      "learning_rate": 4.281850961538462e-06,
+      "loss": 0.7858,
+      "step": 3144
+    },
+    {
+      "epoch": 1.8346215546157212,
+      "grad_norm": 1.2413891553878784,
+      "learning_rate": 4.266826923076923e-06,
+      "loss": 0.6435,
+      "step": 3145
+    },
+    {
+      "epoch": 1.8352049001020854,
+      "grad_norm": 1.193045735359192,
+      "learning_rate": 4.251802884615385e-06,
+      "loss": 0.7756,
+      "step": 3146
+    },
+    {
+      "epoch": 1.8357882455884498,
+      "grad_norm": 1.2209383249282837,
+      "learning_rate": 4.236778846153847e-06,
+      "loss": 0.7637,
+      "step": 3147
+    },
+    {
+      "epoch": 1.8363715910748142,
+      "grad_norm": 1.1595642566680908,
+      "learning_rate": 4.221754807692308e-06,
+      "loss": 0.7416,
+      "step": 3148
+    },
+    {
+      "epoch": 1.8369549365611784,
+      "grad_norm": 1.1345254182815552,
+      "learning_rate": 4.20673076923077e-06,
+      "loss": 0.4896,
+      "step": 3149
+    },
+    {
+      "epoch": 1.8375382820475425,
+      "grad_norm": 1.1374866962432861,
+      "learning_rate": 4.191706730769231e-06,
+      "loss": 0.8506,
+      "step": 3150
+    },
+    {
+      "epoch": 1.838121627533907,
+      "grad_norm": 1.126652717590332,
+      "learning_rate": 4.176682692307692e-06,
+      "loss": 0.7785,
+      "step": 3151
+    },
+    {
+      "epoch": 1.8387049730202714,
+      "grad_norm": 1.1542102098464966,
+      "learning_rate": 4.161658653846154e-06,
+      "loss": 0.6363,
+      "step": 3152
+    },
+    {
+      "epoch": 1.8392883185066355,
+      "grad_norm": 1.0646016597747803,
+      "learning_rate": 4.146634615384616e-06,
+      "loss": 0.6941,
+      "step": 3153
+    },
+    {
+      "epoch": 1.8398716639929997,
+      "grad_norm": 1.2552902698516846,
+      "learning_rate": 4.131610576923077e-06,
+      "loss": 0.7249,
+      "step": 3154
+    },
+    {
+      "epoch": 1.8404550094793641,
+      "grad_norm": 1.0623902082443237,
+      "learning_rate": 4.116586538461539e-06,
+      "loss": 0.7205,
+      "step": 3155
+    },
+    {
+      "epoch": 1.8410383549657285,
+      "grad_norm": 1.0725420713424683,
+      "learning_rate": 4.101562500000001e-06,
+      "loss": 0.7775,
+      "step": 3156
+    },
+    {
+      "epoch": 1.8416217004520927,
+      "grad_norm": 1.095056176185608,
+      "learning_rate": 4.086538461538462e-06,
+      "loss": 0.6987,
+      "step": 3157
+    },
+    {
+      "epoch": 1.842205045938457,
+      "grad_norm": 1.199759840965271,
+      "learning_rate": 4.0715144230769235e-06,
+      "loss": 0.7286,
+      "step": 3158
+    },
+    {
+      "epoch": 1.8427883914248213,
+      "grad_norm": 0.9884356260299683,
+      "learning_rate": 4.0564903846153846e-06,
+      "loss": 0.7607,
+      "step": 3159
+    },
+    {
+      "epoch": 1.8433717369111857,
+      "grad_norm": 1.2776890993118286,
+      "learning_rate": 4.0414663461538465e-06,
+      "loss": 0.6219,
+      "step": 3160
+    },
+    {
+      "epoch": 1.84395508239755,
+      "grad_norm": 1.3608481884002686,
+      "learning_rate": 4.026442307692308e-06,
+      "loss": 0.9211,
+      "step": 3161
+    },
+    {
+      "epoch": 1.844538427883914,
+      "grad_norm": 1.205276608467102,
+      "learning_rate": 4.011418269230769e-06,
+      "loss": 0.6703,
+      "step": 3162
+    },
+    {
+      "epoch": 1.8451217733702785,
+      "grad_norm": 1.4303468465805054,
+      "learning_rate": 3.9963942307692304e-06,
+      "loss": 0.7663,
+      "step": 3163
+    },
+    {
+      "epoch": 1.845705118856643,
+      "grad_norm": 1.1754591464996338,
+      "learning_rate": 3.981370192307692e-06,
+      "loss": 0.7512,
+      "step": 3164
+    },
+    {
+      "epoch": 1.846288464343007,
+      "grad_norm": 1.1947484016418457,
+      "learning_rate": 3.966346153846154e-06,
+      "loss": 0.715,
+      "step": 3165
+    },
+    {
+      "epoch": 1.8468718098293715,
+      "grad_norm": 1.4016227722167969,
+      "learning_rate": 3.951322115384615e-06,
+      "loss": 0.9545,
+      "step": 3166
+    },
+    {
+      "epoch": 1.847455155315736,
+      "grad_norm": 1.25657057762146,
+      "learning_rate": 3.936298076923077e-06,
+      "loss": 0.8419,
+      "step": 3167
+    },
+    {
+      "epoch": 1.8480385008021,
+      "grad_norm": 1.1799569129943848,
+      "learning_rate": 3.921274038461538e-06,
+      "loss": 0.9247,
+      "step": 3168
+    },
+    {
+      "epoch": 1.8486218462884643,
+      "grad_norm": 1.4153069257736206,
+      "learning_rate": 3.90625e-06,
+      "loss": 0.8692,
+      "step": 3169
+    },
+    {
+      "epoch": 1.8492051917748287,
+      "grad_norm": 1.103832483291626,
+      "learning_rate": 3.891225961538462e-06,
+      "loss": 0.7473,
+      "step": 3170
+    },
+    {
+      "epoch": 1.849788537261193,
+      "grad_norm": 1.3749737739562988,
+      "learning_rate": 3.876201923076923e-06,
+      "loss": 0.669,
+      "step": 3171
+    },
+    {
+      "epoch": 1.8503718827475573,
+      "grad_norm": 1.2396540641784668,
+      "learning_rate": 3.861177884615385e-06,
+      "loss": 0.7441,
+      "step": 3172
+    },
+    {
+      "epoch": 1.8509552282339214,
+      "grad_norm": 1.2196747064590454,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.7092,
+      "step": 3173
+    },
+    {
+      "epoch": 1.8515385737202859,
+      "grad_norm": 1.1058050394058228,
+      "learning_rate": 3.831129807692308e-06,
+      "loss": 0.6491,
+      "step": 3174
+    },
+    {
+      "epoch": 1.8521219192066503,
+      "grad_norm": 1.1973541975021362,
+      "learning_rate": 3.81610576923077e-06,
+      "loss": 0.8188,
+      "step": 3175
+    },
+    {
+      "epoch": 1.8527052646930144,
+      "grad_norm": 1.4277015924453735,
+      "learning_rate": 3.8010817307692308e-06,
+      "loss": 0.87,
+      "step": 3176
+    },
+    {
+      "epoch": 1.8532886101793786,
+      "grad_norm": 1.5026133060455322,
+      "learning_rate": 3.7860576923076922e-06,
+      "loss": 0.864,
+      "step": 3177
+    },
+    {
+      "epoch": 1.853871955665743,
+      "grad_norm": 1.1631428003311157,
+      "learning_rate": 3.7710336538461537e-06,
+      "loss": 0.8488,
+      "step": 3178
+    },
+    {
+      "epoch": 1.8544553011521074,
+      "grad_norm": 1.1146588325500488,
+      "learning_rate": 3.7560096153846156e-06,
+      "loss": 0.7239,
+      "step": 3179
+    },
+    {
+      "epoch": 1.8550386466384716,
+      "grad_norm": 1.4882702827453613,
+      "learning_rate": 3.740985576923077e-06,
+      "loss": 0.723,
+      "step": 3180
+    },
+    {
+      "epoch": 1.8556219921248358,
+      "grad_norm": 1.207821011543274,
+      "learning_rate": 3.7259615384615385e-06,
+      "loss": 0.6462,
+      "step": 3181
+    },
+    {
+      "epoch": 1.8562053376112002,
+      "grad_norm": 1.1613391637802124,
+      "learning_rate": 3.7109375e-06,
+      "loss": 0.9761,
+      "step": 3182
+    },
+    {
+      "epoch": 1.8567886830975646,
+      "grad_norm": 1.2488850355148315,
+      "learning_rate": 3.695913461538462e-06,
+      "loss": 0.9367,
+      "step": 3183
+    },
+    {
+      "epoch": 1.8573720285839288,
+      "grad_norm": 1.1730568408966064,
+      "learning_rate": 3.6808894230769233e-06,
+      "loss": 0.7909,
+      "step": 3184
+    },
+    {
+      "epoch": 1.857955374070293,
+      "grad_norm": 1.4373273849487305,
+      "learning_rate": 3.665865384615385e-06,
+      "loss": 0.824,
+      "step": 3185
+    },
+    {
+      "epoch": 1.8585387195566574,
+      "grad_norm": 1.7913013696670532,
+      "learning_rate": 3.6508413461538467e-06,
+      "loss": 0.8291,
+      "step": 3186
+    },
+    {
+      "epoch": 1.8591220650430218,
+      "grad_norm": 1.2750526666641235,
+      "learning_rate": 3.635817307692308e-06,
+      "loss": 0.9025,
+      "step": 3187
+    },
+    {
+      "epoch": 1.859705410529386,
+      "grad_norm": 1.5024226903915405,
+      "learning_rate": 3.6207932692307696e-06,
+      "loss": 0.643,
+      "step": 3188
+    },
+    {
+      "epoch": 1.8602887560157502,
+      "grad_norm": 1.1984363794326782,
+      "learning_rate": 3.6057692307692307e-06,
+      "loss": 0.8266,
+      "step": 3189
+    },
+    {
+      "epoch": 1.8608721015021146,
+      "grad_norm": 1.3658881187438965,
+      "learning_rate": 3.590745192307692e-06,
+      "loss": 0.8933,
+      "step": 3190
+    },
+    {
+      "epoch": 1.861455446988479,
+      "grad_norm": 1.5781782865524292,
+      "learning_rate": 3.5757211538461536e-06,
+      "loss": 0.9418,
+      "step": 3191
+    },
+    {
+      "epoch": 1.8620387924748432,
+      "grad_norm": 1.1617451906204224,
+      "learning_rate": 3.5606971153846155e-06,
+      "loss": 0.5586,
+      "step": 3192
+    },
+    {
+      "epoch": 1.8626221379612076,
+      "grad_norm": 1.3859959840774536,
+      "learning_rate": 3.545673076923077e-06,
+      "loss": 0.971,
+      "step": 3193
+    },
+    {
+      "epoch": 1.863205483447572,
+      "grad_norm": 1.154800534248352,
+      "learning_rate": 3.5306490384615384e-06,
+      "loss": 0.9459,
+      "step": 3194
+    },
+    {
+      "epoch": 1.8637888289339362,
+      "grad_norm": 1.2615742683410645,
+      "learning_rate": 3.5156250000000003e-06,
+      "loss": 0.8425,
+      "step": 3195
+    },
+    {
+      "epoch": 1.8643721744203003,
+      "grad_norm": 1.4125624895095825,
+      "learning_rate": 3.500600961538462e-06,
+      "loss": 0.8873,
+      "step": 3196
+    },
+    {
+      "epoch": 1.8649555199066647,
+      "grad_norm": 1.2564436197280884,
+      "learning_rate": 3.4855769230769233e-06,
+      "loss": 0.8137,
+      "step": 3197
+    },
+    {
+      "epoch": 1.8655388653930292,
+      "grad_norm": 1.1753637790679932,
+      "learning_rate": 3.470552884615385e-06,
+      "loss": 0.8264,
+      "step": 3198
+    },
+    {
+      "epoch": 1.8661222108793933,
+      "grad_norm": 1.0851452350616455,
+      "learning_rate": 3.4555288461538466e-06,
+      "loss": 0.7682,
+      "step": 3199
+    },
+    {
+      "epoch": 1.8667055563657575,
+      "grad_norm": 1.3882149457931519,
+      "learning_rate": 3.440504807692308e-06,
+      "loss": 0.723,
+      "step": 3200
+    },
+    {
+      "epoch": 1.8667055563657575,
+      "eval_loss_squad": 0.8567332937405445,
+      "eval_perplexity": 8.266156372908618,
+      "eval_perplexity_reconstruct": 1.8969144391866655,
+      "step": 3200
+    },
+    {
+      "epoch": 1.867288901852122,
+      "grad_norm": 1.3030450344085693,
+      "learning_rate": 3.4254807692307695e-06,
+      "loss": 0.8131,
+      "step": 3201
+    },
+    {
+      "epoch": 1.8678722473384863,
+      "grad_norm": 1.364791989326477,
+      "learning_rate": 3.4104567307692306e-06,
+      "loss": 0.8387,
+      "step": 3202
+    },
+    {
+      "epoch": 1.8684555928248505,
+      "grad_norm": 1.1007466316223145,
+      "learning_rate": 3.395432692307692e-06,
+      "loss": 0.745,
+      "step": 3203
+    },
+    {
+      "epoch": 1.8690389383112147,
+      "grad_norm": 1.2074437141418457,
+      "learning_rate": 3.380408653846154e-06,
+      "loss": 0.7523,
+      "step": 3204
+    },
+    {
+      "epoch": 1.869622283797579,
+      "grad_norm": 1.4134937524795532,
+      "learning_rate": 3.3653846153846154e-06,
+      "loss": 0.7514,
+      "step": 3205
+    },
+    {
+      "epoch": 1.8702056292839435,
+      "grad_norm": 1.2706308364868164,
+      "learning_rate": 3.350360576923077e-06,
+      "loss": 0.9497,
+      "step": 3206
+    },
+    {
+      "epoch": 1.8707889747703077,
+      "grad_norm": 1.3989231586456299,
+      "learning_rate": 3.3353365384615388e-06,
+      "loss": 0.8504,
+      "step": 3207
+    },
+    {
+      "epoch": 1.8713723202566719,
+      "grad_norm": 1.2219780683517456,
+      "learning_rate": 3.3203125000000002e-06,
+      "loss": 0.7592,
+      "step": 3208
+    },
+    {
+      "epoch": 1.8719556657430363,
+      "grad_norm": 1.3430885076522827,
+      "learning_rate": 3.3052884615384617e-06,
+      "loss": 0.8517,
+      "step": 3209
+    },
+    {
+      "epoch": 1.8725390112294007,
+      "grad_norm": 1.190152883529663,
+      "learning_rate": 3.290264423076923e-06,
+      "loss": 0.6381,
+      "step": 3210
+    },
+    {
+      "epoch": 1.8731223567157649,
+      "grad_norm": 1.4218403100967407,
+      "learning_rate": 3.275240384615385e-06,
+      "loss": 0.7845,
+      "step": 3211
+    },
+    {
+      "epoch": 1.873705702202129,
+      "grad_norm": 1.5346728563308716,
+      "learning_rate": 3.2602163461538465e-06,
+      "loss": 0.6876,
+      "step": 3212
+    },
+    {
+      "epoch": 1.8742890476884935,
+      "grad_norm": 1.3545806407928467,
+      "learning_rate": 3.245192307692308e-06,
+      "loss": 0.7361,
+      "step": 3213
+    },
+    {
+      "epoch": 1.8748723931748579,
+      "grad_norm": 1.127886176109314,
+      "learning_rate": 3.23016826923077e-06,
+      "loss": 1.1837,
+      "step": 3214
+    },
+    {
+      "epoch": 1.875455738661222,
+      "grad_norm": 1.2765008211135864,
+      "learning_rate": 3.2151442307692305e-06,
+      "loss": 0.7988,
+      "step": 3215
+    },
+    {
+      "epoch": 1.8760390841475862,
+      "grad_norm": 1.3014349937438965,
+      "learning_rate": 3.2001201923076924e-06,
+      "loss": 0.8047,
+      "step": 3216
+    },
+    {
+      "epoch": 1.8766224296339509,
+      "grad_norm": 1.284070372581482,
+      "learning_rate": 3.185096153846154e-06,
+      "loss": 0.9213,
+      "step": 3217
+    },
+    {
+      "epoch": 1.877205775120315,
+      "grad_norm": 1.1256341934204102,
+      "learning_rate": 3.1700721153846153e-06,
+      "loss": 0.9295,
+      "step": 3218
+    },
+    {
+      "epoch": 1.8777891206066792,
+      "grad_norm": 1.0452816486358643,
+      "learning_rate": 3.155048076923077e-06,
+      "loss": 0.7096,
+      "step": 3219
+    },
+    {
+      "epoch": 1.8783724660930436,
+      "grad_norm": 1.4434984922409058,
+      "learning_rate": 3.1400240384615387e-06,
+      "loss": 0.906,
+      "step": 3220
+    },
+    {
+      "epoch": 1.878955811579408,
+      "grad_norm": 1.309296727180481,
+      "learning_rate": 3.125e-06,
+      "loss": 0.8231,
+      "step": 3221
+    },
+    {
+      "epoch": 1.8795391570657722,
+      "grad_norm": 1.0300049781799316,
+      "learning_rate": 3.1099759615384616e-06,
+      "loss": 0.8218,
+      "step": 3222
+    },
+    {
+      "epoch": 1.8801225025521364,
+      "grad_norm": 1.3848848342895508,
+      "learning_rate": 3.0949519230769235e-06,
+      "loss": 0.6832,
+      "step": 3223
+    },
+    {
+      "epoch": 1.8807058480385008,
+      "grad_norm": 1.3134764432907104,
+      "learning_rate": 3.079927884615385e-06,
+      "loss": 0.7673,
+      "step": 3224
+    },
+    {
+      "epoch": 1.8812891935248652,
+      "grad_norm": 1.3645648956298828,
+      "learning_rate": 3.064903846153846e-06,
+      "loss": 0.8301,
+      "step": 3225
+    },
+    {
+      "epoch": 1.8818725390112294,
+      "grad_norm": 1.2801153659820557,
+      "learning_rate": 3.049879807692308e-06,
+      "loss": 0.8582,
+      "step": 3226
+    },
+    {
+      "epoch": 1.8824558844975936,
+      "grad_norm": 1.3583933115005493,
+      "learning_rate": 3.0348557692307694e-06,
+      "loss": 0.8457,
+      "step": 3227
+    },
+    {
+      "epoch": 1.883039229983958,
+      "grad_norm": 1.4235001802444458,
+      "learning_rate": 3.019831730769231e-06,
+      "loss": 0.8377,
+      "step": 3228
+    },
+    {
+      "epoch": 1.8836225754703224,
+      "grad_norm": 1.4059243202209473,
+      "learning_rate": 3.0048076923076927e-06,
+      "loss": 0.7966,
+      "step": 3229
+    },
+    {
+      "epoch": 1.8842059209566866,
+      "grad_norm": 1.2851868867874146,
+      "learning_rate": 2.989783653846154e-06,
+      "loss": 0.655,
+      "step": 3230
+    },
+    {
+      "epoch": 1.8847892664430508,
+      "grad_norm": 1.42172110080719,
+      "learning_rate": 2.9747596153846152e-06,
+      "loss": 0.774,
+      "step": 3231
+    },
+    {
+      "epoch": 1.8853726119294152,
+      "grad_norm": 1.1469957828521729,
+      "learning_rate": 2.959735576923077e-06,
+      "loss": 0.6521,
+      "step": 3232
+    },
+    {
+      "epoch": 1.8859559574157796,
+      "grad_norm": 1.7705121040344238,
+      "learning_rate": 2.9447115384615386e-06,
+      "loss": 0.9157,
+      "step": 3233
+    },
+    {
+      "epoch": 1.8865393029021438,
+      "grad_norm": 1.3892390727996826,
+      "learning_rate": 2.9296875e-06,
+      "loss": 0.846,
+      "step": 3234
+    },
+    {
+      "epoch": 1.887122648388508,
+      "grad_norm": 1.1506258249282837,
+      "learning_rate": 2.9146634615384615e-06,
+      "loss": 0.8809,
+      "step": 3235
+    },
+    {
+      "epoch": 1.8877059938748724,
+      "grad_norm": 1.0249956846237183,
+      "learning_rate": 2.8996394230769234e-06,
+      "loss": 0.7122,
+      "step": 3236
+    },
+    {
+      "epoch": 1.8882893393612368,
+      "grad_norm": 1.3840585947036743,
+      "learning_rate": 2.884615384615385e-06,
+      "loss": 0.8086,
+      "step": 3237
+    },
+    {
+      "epoch": 1.888872684847601,
+      "grad_norm": 1.3755419254302979,
+      "learning_rate": 2.8695913461538464e-06,
+      "loss": 0.9811,
+      "step": 3238
+    },
+    {
+      "epoch": 1.8894560303339651,
+      "grad_norm": 1.3516442775726318,
+      "learning_rate": 2.854567307692308e-06,
+      "loss": 0.7445,
+      "step": 3239
+    },
+    {
+      "epoch": 1.8900393758203295,
+      "grad_norm": 1.0328466892242432,
+      "learning_rate": 2.8395432692307693e-06,
+      "loss": 0.6279,
+      "step": 3240
+    },
+    {
+      "epoch": 1.890622721306694,
+      "grad_norm": 1.1680728197097778,
+      "learning_rate": 2.8245192307692307e-06,
+      "loss": 0.7424,
+      "step": 3241
+    },
+    {
+      "epoch": 1.8912060667930581,
+      "grad_norm": 1.1502512693405151,
+      "learning_rate": 2.8094951923076926e-06,
+      "loss": 0.8519,
+      "step": 3242
+    },
+    {
+      "epoch": 1.8917894122794223,
+      "grad_norm": 1.0622332096099854,
+      "learning_rate": 2.794471153846154e-06,
+      "loss": 0.6469,
+      "step": 3243
+    },
+    {
+      "epoch": 1.892372757765787,
+      "grad_norm": 1.0960564613342285,
+      "learning_rate": 2.779447115384615e-06,
+      "loss": 0.8083,
+      "step": 3244
+    },
+    {
+      "epoch": 1.8929561032521511,
+      "grad_norm": 1.226771593093872,
+      "learning_rate": 2.764423076923077e-06,
+      "loss": 0.6511,
+      "step": 3245
+    },
+    {
+      "epoch": 1.8935394487385153,
+      "grad_norm": 1.045430302619934,
+      "learning_rate": 2.7493990384615385e-06,
+      "loss": 0.9077,
+      "step": 3246
+    },
+    {
+      "epoch": 1.8941227942248797,
+      "grad_norm": 1.3713380098342896,
+      "learning_rate": 2.734375e-06,
+      "loss": 0.9468,
+      "step": 3247
+    },
+    {
+      "epoch": 1.8947061397112441,
+      "grad_norm": 1.352195143699646,
+      "learning_rate": 2.719350961538462e-06,
+      "loss": 1.0285,
+      "step": 3248
+    },
+    {
+      "epoch": 1.8952894851976083,
+      "grad_norm": 1.1400947570800781,
+      "learning_rate": 2.7043269230769233e-06,
+      "loss": 0.8711,
+      "step": 3249
+    },
+    {
+      "epoch": 1.8958728306839725,
+      "grad_norm": 1.3555407524108887,
+      "learning_rate": 2.689302884615385e-06,
+      "loss": 0.9186,
+      "step": 3250
+    },
+    {
+      "epoch": 1.896456176170337,
+      "grad_norm": 1.185897946357727,
+      "learning_rate": 2.6742788461538463e-06,
+      "loss": 0.7208,
+      "step": 3251
+    },
+    {
+      "epoch": 1.8970395216567013,
+      "grad_norm": 1.1496164798736572,
+      "learning_rate": 2.6592548076923077e-06,
+      "loss": 0.8495,
+      "step": 3252
+    },
+    {
+      "epoch": 1.8976228671430655,
+      "grad_norm": 1.4104948043823242,
+      "learning_rate": 2.644230769230769e-06,
+      "loss": 0.9402,
+      "step": 3253
+    },
+    {
+      "epoch": 1.8982062126294297,
+      "grad_norm": 1.2339621782302856,
+      "learning_rate": 2.629206730769231e-06,
+      "loss": 0.8283,
+      "step": 3254
+    },
+    {
+      "epoch": 1.898789558115794,
+      "grad_norm": 1.3827406167984009,
+      "learning_rate": 2.6141826923076926e-06,
+      "loss": 0.8392,
+      "step": 3255
+    },
+    {
+      "epoch": 1.8993729036021585,
+      "grad_norm": 1.5171111822128296,
+      "learning_rate": 2.599158653846154e-06,
+      "loss": 0.7742,
+      "step": 3256
+    },
+    {
+      "epoch": 1.8999562490885227,
+      "grad_norm": 1.340565800666809,
+      "learning_rate": 2.5841346153846155e-06,
+      "loss": 0.7106,
+      "step": 3257
+    },
+    {
+      "epoch": 1.9005395945748869,
+      "grad_norm": 1.0620558261871338,
+      "learning_rate": 2.569110576923077e-06,
+      "loss": 0.6781,
+      "step": 3258
+    },
+    {
+      "epoch": 1.9011229400612513,
+      "grad_norm": 1.4068678617477417,
+      "learning_rate": 2.5540865384615384e-06,
+      "loss": 0.6035,
+      "step": 3259
+    },
+    {
+      "epoch": 1.9017062855476157,
+      "grad_norm": 1.356840968132019,
+      "learning_rate": 2.5390625000000003e-06,
+      "loss": 0.8417,
+      "step": 3260
+    },
+    {
+      "epoch": 1.9022896310339799,
+      "grad_norm": 1.0584261417388916,
+      "learning_rate": 2.5240384615384618e-06,
+      "loss": 0.844,
+      "step": 3261
+    },
+    {
+      "epoch": 1.902872976520344,
+      "grad_norm": 1.1918290853500366,
+      "learning_rate": 2.5090144230769232e-06,
+      "loss": 0.8733,
+      "step": 3262
+    },
+    {
+      "epoch": 1.9034563220067084,
+      "grad_norm": 1.2692493200302124,
+      "learning_rate": 2.4939903846153847e-06,
+      "loss": 0.9443,
+      "step": 3263
+    },
+    {
+      "epoch": 1.9040396674930729,
+      "grad_norm": 1.1847847700119019,
+      "learning_rate": 2.478966346153846e-06,
+      "loss": 0.7985,
+      "step": 3264
+    },
+    {
+      "epoch": 1.904623012979437,
+      "grad_norm": 1.3458268642425537,
+      "learning_rate": 2.4639423076923076e-06,
+      "loss": 0.7175,
+      "step": 3265
+    },
+    {
+      "epoch": 1.9052063584658012,
+      "grad_norm": 1.0749139785766602,
+      "learning_rate": 2.4489182692307695e-06,
+      "loss": 0.7089,
+      "step": 3266
+    },
+    {
+      "epoch": 1.9057897039521656,
+      "grad_norm": 1.1422741413116455,
+      "learning_rate": 2.433894230769231e-06,
+      "loss": 0.7655,
+      "step": 3267
+    },
+    {
+      "epoch": 1.90637304943853,
+      "grad_norm": 1.0878541469573975,
+      "learning_rate": 2.4188701923076925e-06,
+      "loss": 0.7034,
+      "step": 3268
+    },
+    {
+      "epoch": 1.9069563949248942,
+      "grad_norm": 1.0475382804870605,
+      "learning_rate": 2.403846153846154e-06,
+      "loss": 0.8075,
+      "step": 3269
+    },
+    {
+      "epoch": 1.9075397404112586,
+      "grad_norm": 1.3869905471801758,
+      "learning_rate": 2.3888221153846154e-06,
+      "loss": 0.8407,
+      "step": 3270
+    },
+    {
+      "epoch": 1.908123085897623,
+      "grad_norm": 1.221571445465088,
+      "learning_rate": 2.373798076923077e-06,
+      "loss": 0.9829,
+      "step": 3271
+    },
+    {
+      "epoch": 1.9087064313839872,
+      "grad_norm": 1.1875793933868408,
+      "learning_rate": 2.3587740384615383e-06,
+      "loss": 0.7063,
+      "step": 3272
+    },
+    {
+      "epoch": 1.9092897768703514,
+      "grad_norm": 1.3426343202590942,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 0.8446,
+      "step": 3273
+    },
+    {
+      "epoch": 1.9098731223567158,
+      "grad_norm": 1.2949109077453613,
+      "learning_rate": 2.3287259615384617e-06,
+      "loss": 0.7361,
+      "step": 3274
+    },
+    {
+      "epoch": 1.9104564678430802,
+      "grad_norm": 1.5180912017822266,
+      "learning_rate": 2.313701923076923e-06,
+      "loss": 0.8648,
+      "step": 3275
+    },
+    {
+      "epoch": 1.9110398133294444,
+      "grad_norm": 1.1568496227264404,
+      "learning_rate": 2.298677884615385e-06,
+      "loss": 0.7235,
+      "step": 3276
+    },
+    {
+      "epoch": 1.9116231588158086,
+      "grad_norm": 1.4971147775650024,
+      "learning_rate": 2.283653846153846e-06,
+      "loss": 0.5042,
+      "step": 3277
+    },
+    {
+      "epoch": 1.912206504302173,
+      "grad_norm": 1.248701572418213,
+      "learning_rate": 2.2686298076923076e-06,
+      "loss": 0.6988,
+      "step": 3278
+    },
+    {
+      "epoch": 1.9127898497885374,
+      "grad_norm": 1.498721718788147,
+      "learning_rate": 2.2536057692307694e-06,
+      "loss": 0.7654,
+      "step": 3279
+    },
+    {
+      "epoch": 1.9133731952749016,
+      "grad_norm": 1.3017774820327759,
+      "learning_rate": 2.238581730769231e-06,
+      "loss": 0.8718,
+      "step": 3280
+    },
+    {
+      "epoch": 1.9139565407612658,
+      "grad_norm": 1.3298211097717285,
+      "learning_rate": 2.2235576923076924e-06,
+      "loss": 0.8205,
+      "step": 3281
+    },
+    {
+      "epoch": 1.9145398862476302,
+      "grad_norm": 1.2720110416412354,
+      "learning_rate": 2.2085336538461543e-06,
+      "loss": 0.7552,
+      "step": 3282
+    },
+    {
+      "epoch": 1.9151232317339946,
+      "grad_norm": 1.2517110109329224,
+      "learning_rate": 2.1935096153846153e-06,
+      "loss": 0.8019,
+      "step": 3283
+    },
+    {
+      "epoch": 1.9157065772203588,
+      "grad_norm": 1.2890255451202393,
+      "learning_rate": 2.1784855769230768e-06,
+      "loss": 0.7499,
+      "step": 3284
+    },
+    {
+      "epoch": 1.916289922706723,
+      "grad_norm": 1.0850515365600586,
+      "learning_rate": 2.1634615384615387e-06,
+      "loss": 0.8125,
+      "step": 3285
+    },
+    {
+      "epoch": 1.9168732681930873,
+      "grad_norm": 1.2969878911972046,
+      "learning_rate": 2.1484375e-06,
+      "loss": 0.7481,
+      "step": 3286
+    },
+    {
+      "epoch": 1.9174566136794517,
+      "grad_norm": 1.0845204591751099,
+      "learning_rate": 2.1334134615384616e-06,
+      "loss": 0.8015,
+      "step": 3287
+    },
+    {
+      "epoch": 1.918039959165816,
+      "grad_norm": 1.371109962463379,
+      "learning_rate": 2.1183894230769235e-06,
+      "loss": 0.7493,
+      "step": 3288
+    },
+    {
+      "epoch": 1.9186233046521801,
+      "grad_norm": 1.1624324321746826,
+      "learning_rate": 2.103365384615385e-06,
+      "loss": 0.7489,
+      "step": 3289
+    },
+    {
+      "epoch": 1.9192066501385445,
+      "grad_norm": 1.3240567445755005,
+      "learning_rate": 2.088341346153846e-06,
+      "loss": 0.847,
+      "step": 3290
+    },
+    {
+      "epoch": 1.919789995624909,
+      "grad_norm": 1.2875831127166748,
+      "learning_rate": 2.073317307692308e-06,
+      "loss": 1.0138,
+      "step": 3291
+    },
+    {
+      "epoch": 1.9203733411112731,
+      "grad_norm": 1.2602109909057617,
+      "learning_rate": 2.0582932692307694e-06,
+      "loss": 0.9535,
+      "step": 3292
+    },
+    {
+      "epoch": 1.9209566865976373,
+      "grad_norm": 1.1849132776260376,
+      "learning_rate": 2.043269230769231e-06,
+      "loss": 0.7271,
+      "step": 3293
+    },
+    {
+      "epoch": 1.9215400320840017,
+      "grad_norm": 1.2771323919296265,
+      "learning_rate": 2.0282451923076923e-06,
+      "loss": 0.9094,
+      "step": 3294
+    },
+    {
+      "epoch": 1.922123377570366,
+      "grad_norm": 1.2452751398086548,
+      "learning_rate": 2.013221153846154e-06,
+      "loss": 0.8227,
+      "step": 3295
+    },
+    {
+      "epoch": 1.9227067230567303,
+      "grad_norm": 1.2949745655059814,
+      "learning_rate": 1.9981971153846152e-06,
+      "loss": 0.776,
+      "step": 3296
+    },
+    {
+      "epoch": 1.9232900685430947,
+      "grad_norm": 1.4040955305099487,
+      "learning_rate": 1.983173076923077e-06,
+      "loss": 0.7911,
+      "step": 3297
+    },
+    {
+      "epoch": 1.923873414029459,
+      "grad_norm": 1.2795517444610596,
+      "learning_rate": 1.9681490384615386e-06,
+      "loss": 0.7292,
+      "step": 3298
+    },
+    {
+      "epoch": 1.9244567595158233,
+      "grad_norm": 1.1882306337356567,
+      "learning_rate": 1.953125e-06,
+      "loss": 0.8261,
+      "step": 3299
+    },
+    {
+      "epoch": 1.9250401050021875,
+      "grad_norm": 1.011581540107727,
+      "learning_rate": 1.9381009615384615e-06,
+      "loss": 0.5729,
+      "step": 3300
+    },
+    {
+      "epoch": 1.9256234504885519,
+      "grad_norm": 1.0311501026153564,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 0.8459,
+      "step": 3301
+    },
+    {
+      "epoch": 1.9262067959749163,
+      "grad_norm": 1.1920017004013062,
+      "learning_rate": 1.908052884615385e-06,
+      "loss": 0.9399,
+      "step": 3302
+    },
+    {
+      "epoch": 1.9267901414612805,
+      "grad_norm": 1.0624799728393555,
+      "learning_rate": 1.8930288461538461e-06,
+      "loss": 0.6593,
+      "step": 3303
+    },
+    {
+      "epoch": 1.9273734869476447,
+      "grad_norm": 1.4519715309143066,
+      "learning_rate": 1.8780048076923078e-06,
+      "loss": 0.7391,
+      "step": 3304
+    },
+    {
+      "epoch": 1.927956832434009,
+      "grad_norm": 1.2700550556182861,
+      "learning_rate": 1.8629807692307693e-06,
+      "loss": 0.9545,
+      "step": 3305
+    },
+    {
+      "epoch": 1.9285401779203735,
+      "grad_norm": 1.4977349042892456,
+      "learning_rate": 1.847956730769231e-06,
+      "loss": 0.7466,
+      "step": 3306
+    },
+    {
+      "epoch": 1.9291235234067377,
+      "grad_norm": 1.3640085458755493,
+      "learning_rate": 1.8329326923076924e-06,
+      "loss": 0.6418,
+      "step": 3307
+    },
+    {
+      "epoch": 1.9297068688931018,
+      "grad_norm": 1.5322325229644775,
+      "learning_rate": 1.817908653846154e-06,
+      "loss": 0.8667,
+      "step": 3308
+    },
+    {
+      "epoch": 1.9302902143794662,
+      "grad_norm": 1.1606767177581787,
+      "learning_rate": 1.8028846153846153e-06,
+      "loss": 0.618,
+      "step": 3309
+    },
+    {
+      "epoch": 1.9308735598658306,
+      "grad_norm": 1.2893586158752441,
+      "learning_rate": 1.7878605769230768e-06,
+      "loss": 0.6867,
+      "step": 3310
+    },
+    {
+      "epoch": 1.9314569053521948,
+      "grad_norm": 1.2085678577423096,
+      "learning_rate": 1.7728365384615385e-06,
+      "loss": 0.6753,
+      "step": 3311
+    },
+    {
+      "epoch": 1.932040250838559,
+      "grad_norm": 1.0935242176055908,
+      "learning_rate": 1.7578125000000002e-06,
+      "loss": 0.6407,
+      "step": 3312
+    },
+    {
+      "epoch": 1.9326235963249234,
+      "grad_norm": 1.3033195734024048,
+      "learning_rate": 1.7427884615384616e-06,
+      "loss": 0.7061,
+      "step": 3313
+    },
+    {
+      "epoch": 1.9332069418112878,
+      "grad_norm": 1.1596806049346924,
+      "learning_rate": 1.7277644230769233e-06,
+      "loss": 0.6813,
+      "step": 3314
+    },
+    {
+      "epoch": 1.933790287297652,
+      "grad_norm": 1.1415847539901733,
+      "learning_rate": 1.7127403846153848e-06,
+      "loss": 0.8781,
+      "step": 3315
+    },
+    {
+      "epoch": 1.9343736327840162,
+      "grad_norm": 1.3523577451705933,
+      "learning_rate": 1.697716346153846e-06,
+      "loss": 0.854,
+      "step": 3316
+    },
+    {
+      "epoch": 1.9349569782703806,
+      "grad_norm": 1.3228445053100586,
+      "learning_rate": 1.6826923076923077e-06,
+      "loss": 0.7102,
+      "step": 3317
+    },
+    {
+      "epoch": 1.935540323756745,
+      "grad_norm": 1.152653455734253,
+      "learning_rate": 1.6676682692307694e-06,
+      "loss": 0.8008,
+      "step": 3318
+    },
+    {
+      "epoch": 1.9361236692431092,
+      "grad_norm": 1.4329392910003662,
+      "learning_rate": 1.6526442307692309e-06,
+      "loss": 0.9733,
+      "step": 3319
+    },
+    {
+      "epoch": 1.9367070147294734,
+      "grad_norm": 1.1904407739639282,
+      "learning_rate": 1.6376201923076925e-06,
+      "loss": 0.7539,
+      "step": 3320
+    },
+    {
+      "epoch": 1.9372903602158378,
+      "grad_norm": 1.1539748907089233,
+      "learning_rate": 1.622596153846154e-06,
+      "loss": 0.8373,
+      "step": 3321
+    },
+    {
+      "epoch": 1.9378737057022022,
+      "grad_norm": 1.211431860923767,
+      "learning_rate": 1.6075721153846153e-06,
+      "loss": 0.7206,
+      "step": 3322
+    },
+    {
+      "epoch": 1.9384570511885664,
+      "grad_norm": 1.0680336952209473,
+      "learning_rate": 1.592548076923077e-06,
+      "loss": 0.8083,
+      "step": 3323
+    },
+    {
+      "epoch": 1.9390403966749308,
+      "grad_norm": 1.3347814083099365,
+      "learning_rate": 1.5775240384615384e-06,
+      "loss": 0.7336,
+      "step": 3324
+    },
+    {
+      "epoch": 1.9396237421612952,
+      "grad_norm": 1.1900486946105957,
+      "learning_rate": 1.5625e-06,
+      "loss": 0.8278,
+      "step": 3325
+    },
+    {
+      "epoch": 1.9402070876476594,
+      "grad_norm": 1.2089030742645264,
+      "learning_rate": 1.5474759615384618e-06,
+      "loss": 0.6548,
+      "step": 3326
+    },
+    {
+      "epoch": 1.9407904331340236,
+      "grad_norm": 1.1675139665603638,
+      "learning_rate": 1.532451923076923e-06,
+      "loss": 0.8378,
+      "step": 3327
+    },
+    {
+      "epoch": 1.941373778620388,
+      "grad_norm": 1.2819353342056274,
+      "learning_rate": 1.5174278846153847e-06,
+      "loss": 0.8911,
+      "step": 3328
+    },
+    {
+      "epoch": 1.9419571241067524,
+      "grad_norm": 1.1306663751602173,
+      "learning_rate": 1.5024038461538464e-06,
+      "loss": 0.8226,
+      "step": 3329
+    },
+    {
+      "epoch": 1.9425404695931165,
+      "grad_norm": 1.330399990081787,
+      "learning_rate": 1.4873798076923076e-06,
+      "loss": 0.7923,
+      "step": 3330
+    },
+    {
+      "epoch": 1.9431238150794807,
+      "grad_norm": 1.2952775955200195,
+      "learning_rate": 1.4723557692307693e-06,
+      "loss": 0.8788,
+      "step": 3331
+    },
+    {
+      "epoch": 1.9437071605658451,
+      "grad_norm": 1.1201375722885132,
+      "learning_rate": 1.4573317307692308e-06,
+      "loss": 0.5674,
+      "step": 3332
+    },
+    {
+      "epoch": 1.9442905060522095,
+      "grad_norm": 1.1273674964904785,
+      "learning_rate": 1.4423076923076924e-06,
+      "loss": 0.8559,
+      "step": 3333
+    },
+    {
+      "epoch": 1.9448738515385737,
+      "grad_norm": 1.462166428565979,
+      "learning_rate": 1.427283653846154e-06,
+      "loss": 0.774,
+      "step": 3334
+    },
+    {
+      "epoch": 1.945457197024938,
+      "grad_norm": 1.3099924325942993,
+      "learning_rate": 1.4122596153846154e-06,
+      "loss": 0.6735,
+      "step": 3335
+    },
+    {
+      "epoch": 1.9460405425113023,
+      "grad_norm": 1.3009566068649292,
+      "learning_rate": 1.397235576923077e-06,
+      "loss": 0.813,
+      "step": 3336
+    },
+    {
+      "epoch": 1.9466238879976667,
+      "grad_norm": 1.3090664148330688,
+      "learning_rate": 1.3822115384615385e-06,
+      "loss": 0.8125,
+      "step": 3337
+    },
+    {
+      "epoch": 1.947207233484031,
+      "grad_norm": 1.452247977256775,
+      "learning_rate": 1.3671875e-06,
+      "loss": 0.8553,
+      "step": 3338
+    },
+    {
+      "epoch": 1.947790578970395,
+      "grad_norm": 1.308634877204895,
+      "learning_rate": 1.3521634615384617e-06,
+      "loss": 0.7305,
+      "step": 3339
+    },
+    {
+      "epoch": 1.9483739244567595,
+      "grad_norm": 1.4182379245758057,
+      "learning_rate": 1.3371394230769231e-06,
+      "loss": 0.6694,
+      "step": 3340
+    },
+    {
+      "epoch": 1.948957269943124,
+      "grad_norm": 1.5073226690292358,
+      "learning_rate": 1.3221153846153846e-06,
+      "loss": 0.7478,
+      "step": 3341
+    },
+    {
+      "epoch": 1.949540615429488,
+      "grad_norm": 1.1056112051010132,
+      "learning_rate": 1.3070913461538463e-06,
+      "loss": 0.9675,
+      "step": 3342
+    },
+    {
+      "epoch": 1.9501239609158523,
+      "grad_norm": 1.3071699142456055,
+      "learning_rate": 1.2920673076923077e-06,
+      "loss": 0.9027,
+      "step": 3343
+    },
+    {
+      "epoch": 1.9507073064022167,
+      "grad_norm": 1.2924457788467407,
+      "learning_rate": 1.2770432692307692e-06,
+      "loss": 0.9082,
+      "step": 3344
+    },
+    {
+      "epoch": 1.951290651888581,
+      "grad_norm": 1.27849280834198,
+      "learning_rate": 1.2620192307692309e-06,
+      "loss": 0.8798,
+      "step": 3345
+    },
+    {
+      "epoch": 1.9518739973749453,
+      "grad_norm": 1.2095881700515747,
+      "learning_rate": 1.2469951923076924e-06,
+      "loss": 0.9073,
+      "step": 3346
+    },
+    {
+      "epoch": 1.9524573428613095,
+      "grad_norm": 1.2283076047897339,
+      "learning_rate": 1.2319711538461538e-06,
+      "loss": 0.6455,
+      "step": 3347
+    },
+    {
+      "epoch": 1.9530406883476739,
+      "grad_norm": 1.241592526435852,
+      "learning_rate": 1.2169471153846155e-06,
+      "loss": 0.7726,
+      "step": 3348
+    },
+    {
+      "epoch": 1.9536240338340383,
+      "grad_norm": 1.2830148935317993,
+      "learning_rate": 1.201923076923077e-06,
+      "loss": 0.7911,
+      "step": 3349
+    },
+    {
+      "epoch": 1.9542073793204024,
+      "grad_norm": 1.2363495826721191,
+      "learning_rate": 1.1868990384615384e-06,
+      "loss": 0.7075,
+      "step": 3350
+    },
+    {
+      "epoch": 1.9547907248067669,
+      "grad_norm": 1.2319027185440063,
+      "learning_rate": 1.1718750000000001e-06,
+      "loss": 0.8239,
+      "step": 3351
+    },
+    {
+      "epoch": 1.9553740702931313,
+      "grad_norm": 1.2770158052444458,
+      "learning_rate": 1.1568509615384616e-06,
+      "loss": 0.6991,
+      "step": 3352
+    },
+    {
+      "epoch": 1.9559574157794954,
+      "grad_norm": 1.1271159648895264,
+      "learning_rate": 1.141826923076923e-06,
+      "loss": 0.8684,
+      "step": 3353
+    },
+    {
+      "epoch": 1.9565407612658596,
+      "grad_norm": 1.2182567119598389,
+      "learning_rate": 1.1268028846153847e-06,
+      "loss": 0.8585,
+      "step": 3354
+    },
+    {
+      "epoch": 1.957124106752224,
+      "grad_norm": 1.2966188192367554,
+      "learning_rate": 1.1117788461538462e-06,
+      "loss": 0.7743,
+      "step": 3355
+    },
+    {
+      "epoch": 1.9577074522385884,
+      "grad_norm": 1.308058261871338,
+      "learning_rate": 1.0967548076923077e-06,
+      "loss": 0.7612,
+      "step": 3356
+    },
+    {
+      "epoch": 1.9582907977249526,
+      "grad_norm": 1.1872565746307373,
+      "learning_rate": 1.0817307692307693e-06,
+      "loss": 0.9376,
+      "step": 3357
+    },
+    {
+      "epoch": 1.9588741432113168,
+      "grad_norm": 1.1720060110092163,
+      "learning_rate": 1.0667067307692308e-06,
+      "loss": 0.689,
+      "step": 3358
+    },
+    {
+      "epoch": 1.9594574886976812,
+      "grad_norm": 1.202436923980713,
+      "learning_rate": 1.0516826923076925e-06,
+      "loss": 0.8309,
+      "step": 3359
+    },
+    {
+      "epoch": 1.9600408341840456,
+      "grad_norm": 1.271864414215088,
+      "learning_rate": 1.036658653846154e-06,
+      "loss": 0.8216,
+      "step": 3360
+    },
+    {
+      "epoch": 1.9606241796704098,
+      "grad_norm": 1.501631259918213,
+      "learning_rate": 1.0216346153846154e-06,
+      "loss": 0.9438,
+      "step": 3361
+    },
+    {
+      "epoch": 1.961207525156774,
+      "grad_norm": 1.3683483600616455,
+      "learning_rate": 1.006610576923077e-06,
+      "loss": 0.8276,
+      "step": 3362
+    },
+    {
+      "epoch": 1.9617908706431384,
+      "grad_norm": 1.202343225479126,
+      "learning_rate": 9.915865384615386e-07,
+      "loss": 0.682,
+      "step": 3363
+    },
+    {
+      "epoch": 1.9623742161295028,
+      "grad_norm": 1.0876096487045288,
+      "learning_rate": 9.765625e-07,
+      "loss": 0.7981,
+      "step": 3364
+    },
+    {
+      "epoch": 1.962957561615867,
+      "grad_norm": 1.5290825366973877,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 0.9378,
+      "step": 3365
+    },
+    {
+      "epoch": 1.9635409071022312,
+      "grad_norm": 1.267568588256836,
+      "learning_rate": 9.465144230769231e-07,
+      "loss": 0.7676,
+      "step": 3366
+    },
+    {
+      "epoch": 1.9641242525885956,
+      "grad_norm": 1.1771818399429321,
+      "learning_rate": 9.314903846153846e-07,
+      "loss": 0.6243,
+      "step": 3367
+    },
+    {
+      "epoch": 1.96470759807496,
+      "grad_norm": 1.4345792531967163,
+      "learning_rate": 9.164663461538462e-07,
+      "loss": 0.7595,
+      "step": 3368
+    },
+    {
+      "epoch": 1.9652909435613242,
+      "grad_norm": 2.9055094718933105,
+      "learning_rate": 9.014423076923077e-07,
+      "loss": 0.8694,
+      "step": 3369
+    },
+    {
+      "epoch": 1.9658742890476884,
+      "grad_norm": 1.2087029218673706,
+      "learning_rate": 8.864182692307692e-07,
+      "loss": 0.8892,
+      "step": 3370
+    },
+    {
+      "epoch": 1.9664576345340528,
+      "grad_norm": 1.1844966411590576,
+      "learning_rate": 8.713942307692308e-07,
+      "loss": 0.9044,
+      "step": 3371
+    },
+    {
+      "epoch": 1.9670409800204172,
+      "grad_norm": 1.4420499801635742,
+      "learning_rate": 8.563701923076924e-07,
+      "loss": 0.7416,
+      "step": 3372
+    },
+    {
+      "epoch": 1.9676243255067813,
+      "grad_norm": 1.2633183002471924,
+      "learning_rate": 8.413461538461539e-07,
+      "loss": 0.8739,
+      "step": 3373
+    },
+    {
+      "epoch": 1.9682076709931455,
+      "grad_norm": 1.2410845756530762,
+      "learning_rate": 8.263221153846154e-07,
+      "loss": 0.9586,
+      "step": 3374
+    },
+    {
+      "epoch": 1.96879101647951,
+      "grad_norm": 1.3219760656356812,
+      "learning_rate": 8.11298076923077e-07,
+      "loss": 0.8433,
+      "step": 3375
+    },
+    {
+      "epoch": 1.9693743619658743,
+      "grad_norm": 1.2453432083129883,
+      "learning_rate": 7.962740384615385e-07,
+      "loss": 0.6931,
+      "step": 3376
+    },
+    {
+      "epoch": 1.9699577074522385,
+      "grad_norm": 1.3527343273162842,
+      "learning_rate": 7.8125e-07,
+      "loss": 0.8611,
+      "step": 3377
+    },
+    {
+      "epoch": 1.970541052938603,
+      "grad_norm": 1.1058260202407837,
+      "learning_rate": 7.662259615384615e-07,
+      "loss": 0.9306,
+      "step": 3378
+    },
+    {
+      "epoch": 1.9711243984249673,
+      "grad_norm": 1.070041537284851,
+      "learning_rate": 7.512019230769232e-07,
+      "loss": 0.8113,
+      "step": 3379
+    },
+    {
+      "epoch": 1.9717077439113315,
+      "grad_norm": 2.1356186866760254,
+      "learning_rate": 7.361778846153846e-07,
+      "loss": 0.7269,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9722910893976957,
+      "grad_norm": 1.2367362976074219,
+      "learning_rate": 7.211538461538462e-07,
+      "loss": 0.8238,
+      "step": 3381
+    },
+    {
+      "epoch": 1.9728744348840601,
+      "grad_norm": 1.401431679725647,
+      "learning_rate": 7.061298076923077e-07,
+      "loss": 0.6869,
+      "step": 3382
+    },
+    {
+      "epoch": 1.9734577803704245,
+      "grad_norm": 1.330978512763977,
+      "learning_rate": 6.911057692307693e-07,
+      "loss": 0.7662,
+      "step": 3383
+    },
+    {
+      "epoch": 1.9740411258567887,
+      "grad_norm": 1.0215189456939697,
+      "learning_rate": 6.760817307692308e-07,
+      "loss": 0.7518,
+      "step": 3384
+    },
+    {
+      "epoch": 1.9746244713431529,
+      "grad_norm": 1.2366408109664917,
+      "learning_rate": 6.610576923076923e-07,
+      "loss": 0.9705,
+      "step": 3385
+    },
+    {
+      "epoch": 1.9752078168295173,
+      "grad_norm": 1.478739857673645,
+      "learning_rate": 6.460336538461539e-07,
+      "loss": 0.691,
+      "step": 3386
+    },
+    {
+      "epoch": 1.9757911623158817,
+      "grad_norm": 1.1755001544952393,
+      "learning_rate": 6.310096153846154e-07,
+      "loss": 0.589,
+      "step": 3387
+    },
+    {
+      "epoch": 1.9763745078022459,
+      "grad_norm": 1.2783706188201904,
+      "learning_rate": 6.159855769230769e-07,
+      "loss": 0.7802,
+      "step": 3388
+    },
+    {
+      "epoch": 1.97695785328861,
+      "grad_norm": 1.1876357793807983,
+      "learning_rate": 6.009615384615385e-07,
+      "loss": 0.7924,
+      "step": 3389
+    },
+    {
+      "epoch": 1.9775411987749745,
+      "grad_norm": 1.159026026725769,
+      "learning_rate": 5.859375000000001e-07,
+      "loss": 0.7157,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9781245442613389,
+      "grad_norm": 1.2553104162216187,
+      "learning_rate": 5.709134615384615e-07,
+      "loss": 0.7791,
+      "step": 3391
+    },
+    {
+      "epoch": 1.978707889747703,
+      "grad_norm": 1.2597992420196533,
+      "learning_rate": 5.558894230769231e-07,
+      "loss": 0.6958,
+      "step": 3392
+    },
+    {
+      "epoch": 1.9792912352340672,
+      "grad_norm": 1.3527934551239014,
+      "learning_rate": 5.408653846153847e-07,
+      "loss": 0.6869,
+      "step": 3393
+    },
+    {
+      "epoch": 1.9798745807204317,
+      "grad_norm": 1.4447914361953735,
+      "learning_rate": 5.258413461538462e-07,
+      "loss": 0.6232,
+      "step": 3394
+    },
+    {
+      "epoch": 1.980457926206796,
+      "grad_norm": 1.3879870176315308,
+      "learning_rate": 5.108173076923077e-07,
+      "loss": 0.7473,
+      "step": 3395
+    },
+    {
+      "epoch": 1.9810412716931602,
+      "grad_norm": 1.2205619812011719,
+      "learning_rate": 4.957932692307693e-07,
+      "loss": 0.8624,
+      "step": 3396
+    },
+    {
+      "epoch": 1.9816246171795244,
+      "grad_norm": 1.3446601629257202,
+      "learning_rate": 4.807692307692308e-07,
+      "loss": 0.6787,
+      "step": 3397
+    },
+    {
+      "epoch": 1.9822079626658888,
+      "grad_norm": 1.372126579284668,
+      "learning_rate": 4.657451923076923e-07,
+      "loss": 0.658,
+      "step": 3398
+    },
+    {
+      "epoch": 1.9827913081522532,
+      "grad_norm": 1.195761799812317,
+      "learning_rate": 4.5072115384615384e-07,
+      "loss": 0.8341,
+      "step": 3399
+    },
+    {
+      "epoch": 1.9833746536386174,
+      "grad_norm": 1.2088286876678467,
+      "learning_rate": 4.356971153846154e-07,
+      "loss": 0.6969,
+      "step": 3400
+    },
+    {
+      "epoch": 1.9833746536386174,
+      "eval_loss_squad": 0.8548139879200608,
+      "eval_perplexity": 8.228127154549277,
+      "eval_perplexity_reconstruct": 1.8982905574950977,
+      "step": 3400
+    },
+    {
+      "epoch": 1.9839579991249816,
+      "grad_norm": 1.5069819688796997,
+      "learning_rate": 4.2067307692307693e-07,
+      "loss": 0.9078,
+      "step": 3401
+    },
+    {
+      "epoch": 1.984541344611346,
+      "grad_norm": 1.4215034246444702,
+      "learning_rate": 4.056490384615385e-07,
+      "loss": 0.7893,
+      "step": 3402
+    },
+    {
+      "epoch": 1.9851246900977104,
+      "grad_norm": 1.417895793914795,
+      "learning_rate": 3.90625e-07,
+      "loss": 0.705,
+      "step": 3403
+    },
+    {
+      "epoch": 1.9857080355840746,
+      "grad_norm": 1.336230754852295,
+      "learning_rate": 3.756009615384616e-07,
+      "loss": 0.8535,
+      "step": 3404
+    },
+    {
+      "epoch": 1.986291381070439,
+      "grad_norm": 1.057672142982483,
+      "learning_rate": 3.605769230769231e-07,
+      "loss": 0.7968,
+      "step": 3405
+    },
+    {
+      "epoch": 1.9868747265568034,
+      "grad_norm": 1.924058198928833,
+      "learning_rate": 3.4555288461538463e-07,
+      "loss": 0.8941,
+      "step": 3406
+    },
+    {
+      "epoch": 1.9874580720431676,
+      "grad_norm": 1.2566702365875244,
+      "learning_rate": 3.3052884615384615e-07,
+      "loss": 0.9751,
+      "step": 3407
+    },
+    {
+      "epoch": 1.9880414175295318,
+      "grad_norm": 1.2347677946090698,
+      "learning_rate": 3.155048076923077e-07,
+      "loss": 0.9065,
+      "step": 3408
+    },
+    {
+      "epoch": 1.9886247630158962,
+      "grad_norm": 1.202021837234497,
+      "learning_rate": 3.0048076923076924e-07,
+      "loss": 1.108,
+      "step": 3409
+    },
+    {
+      "epoch": 1.9892081085022606,
+      "grad_norm": 1.1871834993362427,
+      "learning_rate": 2.8545673076923076e-07,
+      "loss": 0.7417,
+      "step": 3410
+    },
+    {
+      "epoch": 1.9897914539886248,
+      "grad_norm": 1.482557773590088,
+      "learning_rate": 2.7043269230769233e-07,
+      "loss": 0.8867,
+      "step": 3411
+    },
+    {
+      "epoch": 1.990374799474989,
+      "grad_norm": 1.1281570196151733,
+      "learning_rate": 2.5540865384615385e-07,
+      "loss": 0.7128,
+      "step": 3412
+    },
+    {
+      "epoch": 1.9909581449613534,
+      "grad_norm": 1.2373967170715332,
+      "learning_rate": 2.403846153846154e-07,
+      "loss": 0.9376,
+      "step": 3413
+    },
+    {
+      "epoch": 1.9915414904477178,
+      "grad_norm": 1.0291107892990112,
+      "learning_rate": 2.2536057692307692e-07,
+      "loss": 0.6016,
+      "step": 3414
+    },
+    {
+      "epoch": 1.992124835934082,
+      "grad_norm": 1.2457456588745117,
+      "learning_rate": 2.1033653846153846e-07,
+      "loss": 0.7507,
+      "step": 3415
+    },
+    {
+      "epoch": 1.9927081814204461,
+      "grad_norm": 1.7160253524780273,
+      "learning_rate": 1.953125e-07,
+      "loss": 0.6249,
+      "step": 3416
+    },
+    {
+      "epoch": 1.9932915269068106,
+      "grad_norm": 1.178257703781128,
+      "learning_rate": 1.8028846153846156e-07,
+      "loss": 0.9351,
+      "step": 3417
+    },
+    {
+      "epoch": 1.993874872393175,
+      "grad_norm": 1.1656147241592407,
+      "learning_rate": 1.6526442307692307e-07,
+      "loss": 0.7854,
+      "step": 3418
+    },
+    {
+      "epoch": 1.9944582178795391,
+      "grad_norm": 1.1849141120910645,
+      "learning_rate": 1.5024038461538462e-07,
+      "loss": 0.7907,
+      "step": 3419
+    },
+    {
+      "epoch": 1.9950415633659033,
+      "grad_norm": 1.764297366142273,
+      "learning_rate": 1.3521634615384617e-07,
+      "loss": 0.7752,
+      "step": 3420
+    },
+    {
+      "epoch": 1.9956249088522677,
+      "grad_norm": 1.2948668003082275,
+      "learning_rate": 1.201923076923077e-07,
+      "loss": 0.7394,
+      "step": 3421
+    },
+    {
+      "epoch": 1.9962082543386321,
+      "grad_norm": 1.251250982284546,
+      "learning_rate": 1.0516826923076923e-07,
+      "loss": 0.7703,
+      "step": 3422
+    },
+    {
+      "epoch": 1.9967915998249963,
+      "grad_norm": 1.1011062860488892,
+      "learning_rate": 9.014423076923078e-08,
+      "loss": 0.7541,
+      "step": 3423
+    },
+    {
+      "epoch": 1.9973749453113605,
+      "grad_norm": 1.1622389554977417,
+      "learning_rate": 7.512019230769231e-08,
+      "loss": 0.8255,
+      "step": 3424
+    },
+    {
+      "epoch": 1.997958290797725,
+      "grad_norm": 1.1566705703735352,
+      "learning_rate": 6.009615384615386e-08,
+      "loss": 0.8762,
+      "step": 3425
+    },
+    {
+      "epoch": 1.9985416362840893,
+      "grad_norm": 1.2555816173553467,
+      "learning_rate": 4.507211538461539e-08,
+      "loss": 0.7081,
+      "step": 3426
+    },
+    {
+      "epoch": 1.9991249817704535,
+      "grad_norm": 1.1134846210479736,
+      "learning_rate": 3.004807692307693e-08,
+      "loss": 0.7877,
+      "step": 3427
+    },
+    {
+      "epoch": 1.9997083272568177,
+      "grad_norm": 1.113712191581726,
+      "learning_rate": 1.5024038461538464e-08,
+      "loss": 0.9928,
+      "step": 3428
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3428,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 652,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6980351498544218e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}