diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 1350,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022185246810870773,
+      "grad_norm": 0.821755588054657,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 9.6991,
+      "step": 1
+    },
+    {
+      "epoch": 0.004437049362174155,
+      "grad_norm": 0.8529806137084961,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 9.5438,
+      "step": 2
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "grad_norm": 0.8350484371185303,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 9.5489,
+      "step": 3
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 0.8442343473434448,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 9.4286,
+      "step": 4
+    },
+    {
+      "epoch": 0.011092623405435386,
+      "grad_norm": 0.7834038734436035,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 9.4905,
+      "step": 5
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 0.8370130658149719,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 9.435,
+      "step": 6
+    },
+    {
+      "epoch": 0.01552967276760954,
+      "grad_norm": 0.7672736644744873,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 9.5043,
+      "step": 7
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 0.8143541812896729,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 9.8034,
+      "step": 8
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "grad_norm": 0.8909211754798889,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 9.3444,
+      "step": 9
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 0.7856597900390625,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 9.7278,
+      "step": 10
+    },
+    {
+      "epoch": 0.024403771491957847,
+      "grad_norm": 0.9116557836532593,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 9.4414,
+      "step": 11
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 0.84276282787323,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 9.5072,
+      "step": 12
+    },
+    {
+      "epoch": 0.028840820854132,
+      "grad_norm": 0.8671027421951294,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 9.644,
+      "step": 13
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 0.8132544755935669,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 9.5127,
+      "step": 14
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "grad_norm": 0.8500472903251648,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 9.6896,
+      "step": 15
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 0.7714488506317139,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 9.6935,
+      "step": 16
+    },
+    {
+      "epoch": 0.03771491957848031,
+      "grad_norm": 0.8332557082176208,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 9.5023,
+      "step": 17
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 0.7755995988845825,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 9.4974,
+      "step": 18
+    },
+    {
+      "epoch": 0.04215196894065446,
+      "grad_norm": 0.7929331660270691,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 9.5466,
+      "step": 19
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 0.7973807454109192,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 9.4291,
+      "step": 20
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "grad_norm": 0.8269110918045044,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 9.3955,
+      "step": 21
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 0.798378586769104,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 9.253,
+      "step": 22
+    },
+    {
+      "epoch": 0.051026067665002776,
+      "grad_norm": 0.7872747182846069,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 9.6441,
+      "step": 23
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 0.7602689862251282,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 9.6806,
+      "step": 24
+    },
+    {
+      "epoch": 0.05546311702717693,
+      "grad_norm": 0.826332151889801,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 9.2776,
+      "step": 25
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 0.8521766066551208,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 9.478,
+      "step": 26
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "grad_norm": 0.7788112163543701,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 9.474,
+      "step": 27
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 0.789384126663208,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 9.6497,
+      "step": 28
+    },
+    {
+      "epoch": 0.06433721575152523,
+      "grad_norm": 0.8079146146774292,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 9.5858,
+      "step": 29
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 0.7812151312828064,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 9.4212,
+      "step": 30
+    },
+    {
+      "epoch": 0.06877426511369938,
+      "grad_norm": 0.7668154835700989,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 9.4778,
+      "step": 31
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 0.7715058326721191,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 9.4226,
+      "step": 32
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "grad_norm": 0.785493791103363,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 9.5749,
+      "step": 33
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 0.7829617857933044,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 9.3278,
+      "step": 34
+    },
+    {
+      "epoch": 0.0776483638380477,
+      "grad_norm": 0.820789098739624,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 9.2442,
+      "step": 35
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 0.7322468757629395,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 9.4226,
+      "step": 36
+    },
+    {
+      "epoch": 0.08208541320022185,
+      "grad_norm": 0.7363814115524292,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 9.5617,
+      "step": 37
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 0.7910825610160828,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 9.1749,
+      "step": 38
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "grad_norm": 0.841255784034729,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 9.2142,
+      "step": 39
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 0.7606081962585449,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 9.2849,
+      "step": 40
+    },
+    {
+      "epoch": 0.09095951192457016,
+      "grad_norm": 0.6778098344802856,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 9.6956,
+      "step": 41
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 0.7811343669891357,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 9.2769,
+      "step": 42
+    },
+    {
+      "epoch": 0.09539656128674431,
+      "grad_norm": 0.7399757504463196,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 9.2201,
+      "step": 43
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 0.7313132286071777,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 9.2329,
+      "step": 44
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "grad_norm": 0.8048866391181946,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 9.2267,
+      "step": 45
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 0.720774233341217,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 9.2716,
+      "step": 46
+    },
+    {
+      "epoch": 0.10427066001109263,
+      "grad_norm": 0.7231023907661438,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 9.156,
+      "step": 47
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 0.7421720623970032,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 9.0762,
+      "step": 48
+    },
+    {
+      "epoch": 0.10870770937326678,
+      "grad_norm": 0.7229402661323547,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 9.3626,
+      "step": 49
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 0.7308664321899414,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 9.2076,
+      "step": 50
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "grad_norm": 0.7388616800308228,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 9.0501,
+      "step": 51
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 0.7040876150131226,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 9.1948,
+      "step": 52
+    },
+    {
+      "epoch": 0.11758180809761509,
+      "grad_norm": 0.694665253162384,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 9.4289,
+      "step": 53
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 0.7063983082771301,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 9.2478,
+      "step": 54
+    },
+    {
+      "epoch": 0.12201885745978924,
+      "grad_norm": 0.6970482468605042,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 9.1303,
+      "step": 55
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 0.7191748023033142,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 9.1062,
+      "step": 56
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "grad_norm": 0.6682525873184204,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 9.1811,
+      "step": 57
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 0.708154022693634,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 9.1545,
+      "step": 58
+    },
+    {
+      "epoch": 0.13089295618413754,
+      "grad_norm": 0.7285570502281189,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 9.0309,
+      "step": 59
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 0.6872223615646362,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 9.1293,
+      "step": 60
+    },
+    {
+      "epoch": 0.1353300055463117,
+      "grad_norm": 0.6554589867591858,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 9.0245,
+      "step": 61
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 0.6672403216362,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 8.9581,
+      "step": 62
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "grad_norm": 0.6655276417732239,
+      "learning_rate": 0.00014,
+      "loss": 9.2609,
+      "step": 63
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 0.6515478491783142,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 9.1138,
+      "step": 64
+    },
+    {
+      "epoch": 0.14420410427066002,
+      "grad_norm": 0.6779806613922119,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 8.9438,
+      "step": 65
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 0.6442580223083496,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 9.2061,
+      "step": 66
+    },
+    {
+      "epoch": 0.14864115363283417,
+      "grad_norm": 0.624996542930603,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 9.1503,
+      "step": 67
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 0.679361879825592,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 8.9717,
+      "step": 68
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "grad_norm": 0.6829831004142761,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 8.8656,
+      "step": 69
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 0.6774412989616394,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 8.8804,
+      "step": 70
+    },
+    {
+      "epoch": 0.15751525235718247,
+      "grad_norm": 0.6262904405593872,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 8.9934,
+      "step": 71
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 0.6383113265037537,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 8.87,
+      "step": 72
+    },
+    {
+      "epoch": 0.16195230171935662,
+      "grad_norm": 0.6287043690681458,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 8.9591,
+      "step": 73
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 0.6143686771392822,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 9.191,
+      "step": 74
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "grad_norm": 0.6699996590614319,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 8.8765,
+      "step": 75
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 0.5915237069129944,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 9.2072,
+      "step": 76
+    },
+    {
+      "epoch": 0.17082640044370492,
+      "grad_norm": 0.6287463903427124,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 8.9573,
+      "step": 77
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 0.601331889629364,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 8.7581,
+      "step": 78
+    },
+    {
+      "epoch": 0.1752634498058791,
+      "grad_norm": 0.589911937713623,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 9.0061,
+      "step": 79
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 0.6087385416030884,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 8.878,
+      "step": 80
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "grad_norm": 0.6036903262138367,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 9.0397,
+      "step": 81
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 0.5952363014221191,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 8.7567,
+      "step": 82
+    },
+    {
+      "epoch": 0.1841375485302274,
+      "grad_norm": 0.5887743830680847,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 8.6769,
+      "step": 83
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 0.5936828851699829,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 8.8882,
+      "step": 84
+    },
+    {
+      "epoch": 0.18857459789240155,
+      "grad_norm": 0.6422034502029419,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 8.8854,
+      "step": 85
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 0.6091381311416626,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 8.5834,
+      "step": 86
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "grad_norm": 0.5789250731468201,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 8.8168,
+      "step": 87
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 0.5961308479309082,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 8.721,
+      "step": 88
+    },
+    {
+      "epoch": 0.19744869661674985,
+      "grad_norm": 0.5762999653816223,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 8.7823,
+      "step": 89
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 0.5480250716209412,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 8.7113,
+      "step": 90
+    },
+    {
+      "epoch": 0.201885745978924,
+      "grad_norm": 0.5784609317779541,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 8.7829,
+      "step": 91
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 0.5829251408576965,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 8.653,
+      "step": 92
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "grad_norm": 0.5415821075439453,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 8.7245,
+      "step": 93
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 0.5465582013130188,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 8.833,
+      "step": 94
+    },
+    {
+      "epoch": 0.21075984470327233,
+      "grad_norm": 0.5495519042015076,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 8.6411,
+      "step": 95
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 0.5394369959831238,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 8.7427,
+      "step": 96
+    },
+    {
+      "epoch": 0.21519689406544648,
+      "grad_norm": 0.5436957478523254,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 8.4805,
+      "step": 97
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 0.5385772585868835,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 8.7552,
+      "step": 98
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "grad_norm": 0.5374452471733093,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 8.8496,
+      "step": 99
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 0.5541489720344543,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 8.3918,
+      "step": 100
+    },
+    {
+      "epoch": 0.22407099278979478,
+      "grad_norm": 0.5729667544364929,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 8.1485,
+      "step": 101
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 0.5473313927650452,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 8.4383,
+      "step": 102
+    },
+    {
+      "epoch": 0.22850804215196893,
+      "grad_norm": 0.5363402366638184,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 8.6733,
+      "step": 103
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 0.5916332602500916,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 8.2012,
+      "step": 104
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "grad_norm": 0.5329188108444214,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 8.4325,
+      "step": 105
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 0.5177275538444519,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 8.594,
+      "step": 106
+    },
+    {
+      "epoch": 0.23738214087631726,
+      "grad_norm": 0.5441237688064575,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 8.3921,
+      "step": 107
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 0.5013054013252258,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 8.6396,
+      "step": 108
+    },
+    {
+      "epoch": 0.2418191902384914,
+      "grad_norm": 0.5063790678977966,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 8.5116,
+      "step": 109
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 0.5051860213279724,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 8.5502,
+      "step": 110
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "grad_norm": 0.5111222267150879,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 8.5484,
+      "step": 111
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 0.5243113040924072,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 8.3099,
+      "step": 112
+    },
+    {
+      "epoch": 0.2506932889628397,
+      "grad_norm": 0.4713006317615509,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 8.1631,
+      "step": 113
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 0.4603446424007416,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 8.3347,
+      "step": 114
+    },
+    {
+      "epoch": 0.25513033832501386,
+      "grad_norm": 0.49237748980522156,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 8.4261,
+      "step": 115
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 0.49359843134880066,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 8.2622,
+      "step": 116
+    },
+    {
+      "epoch": 0.259567387687188,
+      "grad_norm": 0.49687451124191284,
+      "learning_rate": 0.00026,
+      "loss": 8.3877,
+      "step": 117
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 0.47837090492248535,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 8.3443,
+      "step": 118
+    },
+    {
+      "epoch": 0.26400443704936216,
+      "grad_norm": 0.4627683162689209,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 8.1771,
+      "step": 119
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 0.4726055860519409,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 8.3201,
+      "step": 120
+    },
+    {
+      "epoch": 0.2684414864115363,
+      "grad_norm": 0.47923341393470764,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 8.1676,
+      "step": 121
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 0.47404396533966064,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 8.1455,
+      "step": 122
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "grad_norm": 0.4649370610713959,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 8.0559,
+      "step": 123
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 0.473906546831131,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 8.2042,
+      "step": 124
+    },
+    {
+      "epoch": 0.2773155851358846,
+      "grad_norm": 0.42972394824028015,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 8.1928,
+      "step": 125
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 0.4697248935699463,
+      "learning_rate": 0.00028,
+      "loss": 8.262,
+      "step": 126
+    },
+    {
+      "epoch": 0.2817526344980588,
+      "grad_norm": 0.47600647807121277,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 8.237,
+      "step": 127
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 0.45681822299957275,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 8.3521,
+      "step": 128
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "grad_norm": 0.44468188285827637,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 8.1883,
+      "step": 129
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 0.44600602984428406,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 8.3341,
+      "step": 130
+    },
+    {
+      "epoch": 0.2906267332224071,
+      "grad_norm": 0.42914265394210815,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 8.2302,
+      "step": 131
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 0.47467923164367676,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 8.076,
+      "step": 132
+    },
+    {
+      "epoch": 0.29506378258458127,
+      "grad_norm": 0.42793264985084534,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 8.0827,
+      "step": 133
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 0.45945870876312256,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 8.2033,
+      "step": 134
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "grad_norm": 0.45133456587791443,
+      "learning_rate": 0.0003,
+      "loss": 7.827,
+      "step": 135
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 0.43471652269363403,
+      "learning_rate": 0.00029975308641975307,
+      "loss": 7.78,
+      "step": 136
+    },
+    {
+      "epoch": 0.30393788130892957,
+      "grad_norm": 0.42857375741004944,
+      "learning_rate": 0.00029950617283950616,
+      "loss": 8.0075,
+      "step": 137
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 0.4262228310108185,
+      "learning_rate": 0.00029925925925925926,
+      "loss": 8.0664,
+      "step": 138
+    },
+    {
+      "epoch": 0.3083749306711037,
+      "grad_norm": 0.4517154395580292,
+      "learning_rate": 0.0002990123456790123,
+      "loss": 8.0056,
+      "step": 139
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 0.4255889654159546,
+      "learning_rate": 0.0002987654320987654,
+      "loss": 7.8914,
+      "step": 140
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "grad_norm": 0.43145671486854553,
+      "learning_rate": 0.0002985185185185185,
+      "loss": 7.7376,
+      "step": 141
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 0.41400274634361267,
+      "learning_rate": 0.0002982716049382716,
+      "loss": 8.0241,
+      "step": 142
+    },
+    {
+      "epoch": 0.317249029395452,
+      "grad_norm": 0.4291524887084961,
+      "learning_rate": 0.0002980246913580247,
+      "loss": 7.771,
+      "step": 143
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 0.4104973077774048,
+      "learning_rate": 0.0002977777777777777,
+      "loss": 8.0867,
+      "step": 144
+    },
+    {
+      "epoch": 0.32168607875762617,
+      "grad_norm": 0.42850571870803833,
+      "learning_rate": 0.00029753086419753087,
+      "loss": 7.9236,
+      "step": 145
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 0.389720618724823,
+      "learning_rate": 0.0002972839506172839,
+      "loss": 7.7898,
+      "step": 146
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "grad_norm": 0.4229864776134491,
+      "learning_rate": 0.000297037037037037,
+      "loss": 7.6654,
+      "step": 147
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 0.4015505611896515,
+      "learning_rate": 0.0002967901234567901,
+      "loss": 7.9834,
+      "step": 148
+    },
+    {
+      "epoch": 0.33056017748197447,
+      "grad_norm": 0.4084751605987549,
+      "learning_rate": 0.0002965432098765432,
+      "loss": 7.9001,
+      "step": 149
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 0.4132940471172333,
+      "learning_rate": 0.00029629629629629624,
+      "loss": 7.8378,
+      "step": 150
+    },
+    {
+      "epoch": 0.3349972268441486,
+      "grad_norm": 0.4032317101955414,
+      "learning_rate": 0.00029604938271604933,
+      "loss": 8.0794,
+      "step": 151
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 0.39160147309303284,
+      "learning_rate": 0.0002958024691358025,
+      "loss": 7.8979,
+      "step": 152
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "grad_norm": 0.4023204743862152,
+      "learning_rate": 0.0002955555555555555,
+      "loss": 7.8445,
+      "step": 153
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 0.3891277611255646,
+      "learning_rate": 0.0002953086419753086,
+      "loss": 7.8758,
+      "step": 154
+    },
+    {
+      "epoch": 0.343871325568497,
+      "grad_norm": 0.3732168972492218,
+      "learning_rate": 0.0002950617283950617,
+      "loss": 7.8755,
+      "step": 155
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 0.3869277238845825,
+      "learning_rate": 0.0002948148148148148,
+      "loss": 7.7781,
+      "step": 156
+    },
+    {
+      "epoch": 0.3483083749306711,
+      "grad_norm": 0.38009315729141235,
+      "learning_rate": 0.00029456790123456785,
+      "loss": 7.9799,
+      "step": 157
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 0.3549535274505615,
+      "learning_rate": 0.00029432098765432095,
+      "loss": 7.9553,
+      "step": 158
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "grad_norm": 0.37830716371536255,
+      "learning_rate": 0.00029407407407407404,
+      "loss": 7.3761,
+      "step": 159
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 0.3759785294532776,
+      "learning_rate": 0.00029382716049382714,
+      "loss": 7.7633,
+      "step": 160
+    },
+    {
+      "epoch": 0.35718247365501943,
+      "grad_norm": 0.3714516758918762,
+      "learning_rate": 0.00029358024691358023,
+      "loss": 7.9676,
+      "step": 161
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 0.36620503664016724,
+      "learning_rate": 0.00029333333333333327,
+      "loss": 7.974,
+      "step": 162
+    },
+    {
+      "epoch": 0.3616195230171936,
+      "grad_norm": 0.3777652382850647,
+      "learning_rate": 0.0002930864197530864,
+      "loss": 7.624,
+      "step": 163
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 0.35747477412223816,
+      "learning_rate": 0.00029283950617283946,
+      "loss": 7.6377,
+      "step": 164
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "grad_norm": 0.3485727608203888,
+      "learning_rate": 0.00029259259259259256,
+      "loss": 7.6217,
+      "step": 165
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 0.354579895734787,
+      "learning_rate": 0.00029234567901234565,
+      "loss": 7.7828,
+      "step": 166
+    },
+    {
+      "epoch": 0.3704936217415419,
+      "grad_norm": 0.3804467022418976,
+      "learning_rate": 0.00029209876543209875,
+      "loss": 7.6122,
+      "step": 167
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 0.3737318217754364,
+      "learning_rate": 0.00029185185185185184,
+      "loss": 7.6699,
+      "step": 168
+    },
+    {
+      "epoch": 0.37493067110371603,
+      "grad_norm": 0.3586113452911377,
+      "learning_rate": 0.0002916049382716049,
+      "loss": 7.537,
+      "step": 169
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 0.3582717478275299,
+      "learning_rate": 0.000291358024691358,
+      "loss": 7.5965,
+      "step": 170
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "grad_norm": 0.35238462686538696,
+      "learning_rate": 0.0002911111111111111,
+      "loss": 7.7917,
+      "step": 171
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 0.3515949547290802,
+      "learning_rate": 0.00029086419753086417,
+      "loss": 7.3983,
+      "step": 172
+    },
+    {
+      "epoch": 0.38380476982806433,
+      "grad_norm": 0.38555556535720825,
+      "learning_rate": 0.00029061728395061726,
+      "loss": 7.2721,
+      "step": 173
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 0.3750784397125244,
+      "learning_rate": 0.00029037037037037036,
+      "loss": 7.5168,
+      "step": 174
+    },
+    {
+      "epoch": 0.3882418191902385,
+      "grad_norm": 0.33401113748550415,
+      "learning_rate": 0.00029012345679012345,
+      "loss": 7.5381,
+      "step": 175
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 0.3323742747306824,
+      "learning_rate": 0.0002898765432098765,
+      "loss": 7.4892,
+      "step": 176
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "grad_norm": 0.35648635029792786,
+      "learning_rate": 0.0002896296296296296,
+      "loss": 7.42,
+      "step": 177
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 0.3261418044567108,
+      "learning_rate": 0.0002893827160493827,
+      "loss": 7.2928,
+      "step": 178
+    },
+    {
+      "epoch": 0.3971159179145868,
+      "grad_norm": 0.33553892374038696,
+      "learning_rate": 0.0002891358024691358,
+      "loss": 7.6432,
+      "step": 179
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 0.36701348423957825,
+      "learning_rate": 0.0002888888888888888,
+      "loss": 7.8191,
+      "step": 180
+    },
+    {
+      "epoch": 0.40155296727676093,
+      "grad_norm": 0.33658480644226074,
+      "learning_rate": 0.00028864197530864197,
+      "loss": 7.3822,
+      "step": 181
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 0.3158818185329437,
+      "learning_rate": 0.00028839506172839507,
+      "loss": 7.4079,
+      "step": 182
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "grad_norm": 0.36799246072769165,
+      "learning_rate": 0.0002881481481481481,
+      "loss": 7.5174,
+      "step": 183
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 0.34958580136299133,
+      "learning_rate": 0.0002879012345679012,
+      "loss": 7.7897,
+      "step": 184
+    },
+    {
+      "epoch": 0.4104270660011093,
+      "grad_norm": 0.32340824604034424,
+      "learning_rate": 0.0002876543209876543,
+      "loss": 7.5503,
+      "step": 185
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 0.3355458676815033,
+      "learning_rate": 0.0002874074074074074,
+      "loss": 7.3595,
+      "step": 186
+    },
+    {
+      "epoch": 0.41486411536328344,
+      "grad_norm": 0.3336404860019684,
+      "learning_rate": 0.00028716049382716043,
+      "loss": 7.5312,
+      "step": 187
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 0.36585137248039246,
+      "learning_rate": 0.00028691358024691353,
+      "loss": 7.7615,
+      "step": 188
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "grad_norm": 0.3408105969429016,
+      "learning_rate": 0.0002866666666666667,
+      "loss": 7.3813,
+      "step": 189
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 0.30925247073173523,
+      "learning_rate": 0.0002864197530864197,
+      "loss": 7.3238,
+      "step": 190
+    },
+    {
+      "epoch": 0.42373821408763174,
+      "grad_norm": 0.3370666801929474,
+      "learning_rate": 0.0002861728395061728,
+      "loss": 7.434,
+      "step": 191
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 0.326911062002182,
+      "learning_rate": 0.0002859259259259259,
+      "loss": 7.3604,
+      "step": 192
+    },
+    {
+      "epoch": 0.4281752634498059,
+      "grad_norm": 0.32766297459602356,
+      "learning_rate": 0.000285679012345679,
+      "loss": 7.6426,
+      "step": 193
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 0.31743839383125305,
+      "learning_rate": 0.00028543209876543204,
+      "loss": 7.3434,
+      "step": 194
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "grad_norm": 0.3063775300979614,
+      "learning_rate": 0.00028518518518518514,
+      "loss": 7.3935,
+      "step": 195
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 0.29284536838531494,
+      "learning_rate": 0.00028493827160493824,
+      "loss": 7.3607,
+      "step": 196
+    },
+    {
+      "epoch": 0.4370493621741542,
+      "grad_norm": 0.31224870681762695,
+      "learning_rate": 0.00028469135802469133,
+      "loss": 7.549,
+      "step": 197
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 0.3152666687965393,
+      "learning_rate": 0.0002844444444444444,
+      "loss": 7.5221,
+      "step": 198
+    },
+    {
+      "epoch": 0.44148641153632834,
+      "grad_norm": 0.3212338984012604,
+      "learning_rate": 0.0002841975308641975,
+      "loss": 7.537,
+      "step": 199
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 0.3412870764732361,
+      "learning_rate": 0.0002839506172839506,
+      "loss": 7.4701,
+      "step": 200
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "grad_norm": 0.3184579312801361,
+      "learning_rate": 0.00028370370370370366,
+      "loss": 7.2813,
+      "step": 201
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 0.31065618991851807,
+      "learning_rate": 0.00028345679012345675,
+      "loss": 7.5757,
+      "step": 202
+    },
+    {
+      "epoch": 0.45036051026067664,
+      "grad_norm": 0.291221022605896,
+      "learning_rate": 0.00028320987654320985,
+      "loss": 7.4003,
+      "step": 203
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 0.3027431070804596,
+      "learning_rate": 0.00028296296296296294,
+      "loss": 7.4765,
+      "step": 204
+    },
+    {
+      "epoch": 0.4547975596228508,
+      "grad_norm": 0.32293501496315,
+      "learning_rate": 0.00028271604938271604,
+      "loss": 7.4653,
+      "step": 205
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 0.30279305577278137,
+      "learning_rate": 0.0002824691358024691,
+      "loss": 7.2067,
+      "step": 206
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "grad_norm": 0.3169279098510742,
+      "learning_rate": 0.00028222222222222223,
+      "loss": 7.2822,
+      "step": 207
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 0.3232857584953308,
+      "learning_rate": 0.00028197530864197527,
+      "loss": 7.1141,
+      "step": 208
+    },
+    {
+      "epoch": 0.4636716583471991,
+      "grad_norm": 0.29470109939575195,
+      "learning_rate": 0.00028172839506172836,
+      "loss": 7.1161,
+      "step": 209
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 0.3023466169834137,
+      "learning_rate": 0.00028148148148148146,
+      "loss": 7.2033,
+      "step": 210
+    },
+    {
+      "epoch": 0.46810870770937324,
+      "grad_norm": 0.3022836148738861,
+      "learning_rate": 0.00028123456790123455,
+      "loss": 7.4983,
+      "step": 211
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 0.2998552918434143,
+      "learning_rate": 0.00028098765432098765,
+      "loss": 6.9715,
+      "step": 212
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "grad_norm": 0.28968268632888794,
+      "learning_rate": 0.0002807407407407407,
+      "loss": 7.407,
+      "step": 213
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 0.2811324894428253,
+      "learning_rate": 0.0002804938271604938,
+      "loss": 7.0647,
+      "step": 214
+    },
+    {
+      "epoch": 0.4769828064337216,
+      "grad_norm": 0.2970676124095917,
+      "learning_rate": 0.0002802469135802469,
+      "loss": 7.3096,
+      "step": 215
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 0.29108595848083496,
+      "learning_rate": 0.00028,
+      "loss": 7.0157,
+      "step": 216
+    },
+    {
+      "epoch": 0.48141985579589575,
+      "grad_norm": 0.31331101059913635,
+      "learning_rate": 0.00027975308641975307,
+      "loss": 7.3653,
+      "step": 217
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 0.2949845790863037,
+      "learning_rate": 0.00027950617283950617,
+      "loss": 7.016,
+      "step": 218
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "grad_norm": 0.30273276567459106,
+      "learning_rate": 0.00027925925925925926,
+      "loss": 7.1501,
+      "step": 219
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 0.2936885356903076,
+      "learning_rate": 0.0002790123456790123,
+      "loss": 7.3891,
+      "step": 220
+    },
+    {
+      "epoch": 0.49029395452024405,
+      "grad_norm": 0.2768365740776062,
+      "learning_rate": 0.0002787654320987654,
+      "loss": 7.4485,
+      "step": 221
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 0.2886095643043518,
+      "learning_rate": 0.0002785185185185185,
+      "loss": 7.1963,
+      "step": 222
+    },
+    {
+      "epoch": 0.4947310038824182,
+      "grad_norm": 0.2979474365711212,
+      "learning_rate": 0.0002782716049382716,
+      "loss": 7.4258,
+      "step": 223
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 0.2820936143398285,
+      "learning_rate": 0.00027802469135802463,
+      "loss": 7.1587,
+      "step": 224
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "grad_norm": 0.2848482131958008,
+      "learning_rate": 0.0002777777777777778,
+      "loss": 7.0982,
+      "step": 225
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 0.2747862935066223,
+      "learning_rate": 0.00027753086419753087,
+      "loss": 7.0616,
+      "step": 226
+    },
+    {
+      "epoch": 0.5036051026067665,
+      "grad_norm": 0.27541446685791016,
+      "learning_rate": 0.0002772839506172839,
+      "loss": 7.2486,
+      "step": 227
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 0.2819608449935913,
+      "learning_rate": 0.000277037037037037,
+      "loss": 7.1413,
+      "step": 228
+    },
+    {
+      "epoch": 0.5080421519689406,
+      "grad_norm": 0.2680410146713257,
+      "learning_rate": 0.0002767901234567901,
+      "loss": 7.0542,
+      "step": 229
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 0.2712319791316986,
+      "learning_rate": 0.0002765432098765432,
+      "loss": 7.2641,
+      "step": 230
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "grad_norm": 0.2802445590496063,
+      "learning_rate": 0.00027629629629629624,
+      "loss": 7.2343,
+      "step": 231
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 0.29296448826789856,
+      "learning_rate": 0.00027604938271604934,
+      "loss": 7.2466,
+      "step": 232
+    },
+    {
+      "epoch": 0.516916250693289,
+      "grad_norm": 0.2604992985725403,
+      "learning_rate": 0.00027580246913580243,
+      "loss": 7.0482,
+      "step": 233
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 0.29246747493743896,
+      "learning_rate": 0.0002755555555555555,
+      "loss": 7.0667,
+      "step": 234
+    },
+    {
+      "epoch": 0.5213533000554631,
+      "grad_norm": 0.30500155687332153,
+      "learning_rate": 0.0002753086419753086,
+      "loss": 6.9277,
+      "step": 235
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 0.2477540373802185,
+      "learning_rate": 0.0002750617283950617,
+      "loss": 7.1131,
+      "step": 236
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "grad_norm": 0.2841467559337616,
+      "learning_rate": 0.0002748148148148148,
+      "loss": 7.2877,
+      "step": 237
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 0.28137609362602234,
+      "learning_rate": 0.00027456790123456785,
+      "loss": 6.9902,
+      "step": 238
+    },
+    {
+      "epoch": 0.5302273987798114,
+      "grad_norm": 0.2635383605957031,
+      "learning_rate": 0.00027432098765432095,
+      "loss": 6.9635,
+      "step": 239
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 0.28244468569755554,
+      "learning_rate": 0.00027407407407407404,
+      "loss": 7.1117,
+      "step": 240
+    },
+    {
+      "epoch": 0.5346644481419855,
+      "grad_norm": 0.27359530329704285,
+      "learning_rate": 0.00027382716049382714,
+      "loss": 6.69,
+      "step": 241
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 0.25735512375831604,
+      "learning_rate": 0.00027358024691358023,
+      "loss": 6.9909,
+      "step": 242
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "grad_norm": 0.2677929103374481,
+      "learning_rate": 0.00027333333333333333,
+      "loss": 6.9976,
+      "step": 243
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 0.2864200174808502,
+      "learning_rate": 0.0002730864197530864,
+      "loss": 7.09,
+      "step": 244
+    },
+    {
+      "epoch": 0.5435385468663338,
+      "grad_norm": 0.26590779423713684,
+      "learning_rate": 0.00027283950617283946,
+      "loss": 6.7222,
+      "step": 245
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 0.2783195674419403,
+      "learning_rate": 0.00027259259259259256,
+      "loss": 7.0877,
+      "step": 246
+    },
+    {
+      "epoch": 0.547975596228508,
+      "grad_norm": 0.2596246004104614,
+      "learning_rate": 0.00027234567901234565,
+      "loss": 6.9997,
+      "step": 247
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 0.25567305088043213,
+      "learning_rate": 0.00027209876543209875,
+      "loss": 6.9712,
+      "step": 248
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "grad_norm": 0.266420841217041,
+      "learning_rate": 0.00027185185185185184,
+      "loss": 6.9589,
+      "step": 249
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 0.2516438364982605,
+      "learning_rate": 0.0002716049382716049,
+      "loss": 6.9321,
+      "step": 250
+    },
+    {
+      "epoch": 0.5568496949528563,
+      "grad_norm": 0.25865715742111206,
+      "learning_rate": 0.00027135802469135803,
+      "loss": 6.9016,
+      "step": 251
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 0.2775104343891144,
+      "learning_rate": 0.0002711111111111111,
+      "loss": 7.1388,
+      "step": 252
+    },
+    {
+      "epoch": 0.5612867443150305,
+      "grad_norm": 0.23623691499233246,
+      "learning_rate": 0.00027086419753086417,
+      "loss": 7.13,
+      "step": 253
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 0.2628632187843323,
+      "learning_rate": 0.00027061728395061727,
+      "loss": 6.945,
+      "step": 254
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "grad_norm": 0.25304099917411804,
+      "learning_rate": 0.00027037037037037036,
+      "loss": 6.9994,
+      "step": 255
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 0.2611977458000183,
+      "learning_rate": 0.00027012345679012346,
+      "loss": 6.9731,
+      "step": 256
+    },
+    {
+      "epoch": 0.5701608430393789,
+      "grad_norm": 0.25445589423179626,
+      "learning_rate": 0.0002698765432098765,
+      "loss": 6.9056,
+      "step": 257
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 0.24887242913246155,
+      "learning_rate": 0.0002696296296296296,
+      "loss": 7.0109,
+      "step": 258
+    },
+    {
+      "epoch": 0.574597892401553,
+      "grad_norm": 0.23668690025806427,
+      "learning_rate": 0.0002693827160493827,
+      "loss": 7.0443,
+      "step": 259
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 0.25199857354164124,
+      "learning_rate": 0.0002691358024691358,
+      "loss": 7.1156,
+      "step": 260
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "grad_norm": 0.2511538565158844,
+      "learning_rate": 0.0002688888888888889,
+      "loss": 7.0096,
+      "step": 261
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 0.2505515515804291,
+      "learning_rate": 0.00026864197530864197,
+      "loss": 6.474,
+      "step": 262
+    },
+    {
+      "epoch": 0.5834719911259013,
+      "grad_norm": 0.26783788204193115,
+      "learning_rate": 0.000268395061728395,
+      "loss": 6.7875,
+      "step": 263
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 0.2486739158630371,
+      "learning_rate": 0.0002681481481481481,
+      "loss": 6.84,
+      "step": 264
+    },
+    {
+      "epoch": 0.5879090404880755,
+      "grad_norm": 0.25090929865837097,
+      "learning_rate": 0.0002679012345679012,
+      "loss": 7.1033,
+      "step": 265
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 0.2413657307624817,
+      "learning_rate": 0.0002676543209876543,
+      "loss": 6.8682,
+      "step": 266
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "grad_norm": 0.2598755955696106,
+      "learning_rate": 0.0002674074074074074,
+      "loss": 7.1267,
+      "step": 267
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 0.22166892886161804,
+      "learning_rate": 0.00026716049382716043,
+      "loss": 7.0713,
+      "step": 268
+    },
+    {
+      "epoch": 0.5967831392124238,
+      "grad_norm": 0.22469143569469452,
+      "learning_rate": 0.0002669135802469136,
+      "loss": 7.0637,
+      "step": 269
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 0.2319795787334442,
+      "learning_rate": 0.0002666666666666666,
+      "loss": 6.8818,
+      "step": 270
+    },
+    {
+      "epoch": 0.6012201885745979,
+      "grad_norm": 0.25172027945518494,
+      "learning_rate": 0.0002664197530864197,
+      "loss": 6.8091,
+      "step": 271
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 0.24617278575897217,
+      "learning_rate": 0.0002661728395061728,
+      "loss": 6.7596,
+      "step": 272
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "grad_norm": 0.239463672041893,
+      "learning_rate": 0.0002659259259259259,
+      "loss": 6.831,
+      "step": 273
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 0.25568217039108276,
+      "learning_rate": 0.000265679012345679,
+      "loss": 6.9243,
+      "step": 274
+    },
+    {
+      "epoch": 0.6100942872989462,
+      "grad_norm": 0.244978129863739,
+      "learning_rate": 0.00026543209876543205,
+      "loss": 6.9591,
+      "step": 275
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 0.23129886388778687,
+      "learning_rate": 0.00026518518518518514,
+      "loss": 7.0333,
+      "step": 276
+    },
+    {
+      "epoch": 0.6145313366611204,
+      "grad_norm": 0.22391913831233978,
+      "learning_rate": 0.00026493827160493824,
+      "loss": 7.0782,
+      "step": 277
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 0.25436490774154663,
+      "learning_rate": 0.00026469135802469133,
+      "loss": 6.7715,
+      "step": 278
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "grad_norm": 0.27404317259788513,
+      "learning_rate": 0.00026444444444444443,
+      "loss": 7.0218,
+      "step": 279
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 0.2339683473110199,
+      "learning_rate": 0.0002641975308641975,
+      "loss": 6.7681,
+      "step": 280
+    },
+    {
+      "epoch": 0.6234054353854687,
+      "grad_norm": 0.2204495519399643,
+      "learning_rate": 0.0002639506172839506,
+      "loss": 6.8839,
+      "step": 281
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 0.2658957540988922,
+      "learning_rate": 0.00026370370370370366,
+      "loss": 6.3275,
+      "step": 282
+    },
+    {
+      "epoch": 0.6278424847476428,
+      "grad_norm": 0.24074450135231018,
+      "learning_rate": 0.00026345679012345675,
+      "loss": 7.0914,
+      "step": 283
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 0.2594529390335083,
+      "learning_rate": 0.00026320987654320985,
+      "loss": 6.9025,
+      "step": 284
+    },
+    {
+      "epoch": 0.632279534109817,
+      "grad_norm": 0.23277385532855988,
+      "learning_rate": 0.00026296296296296294,
+      "loss": 6.9584,
+      "step": 285
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 0.221964031457901,
+      "learning_rate": 0.00026271604938271604,
+      "loss": 6.9299,
+      "step": 286
+    },
+    {
+      "epoch": 0.6367165834719911,
+      "grad_norm": 0.21898798644542694,
+      "learning_rate": 0.00026246913580246913,
+      "loss": 6.8232,
+      "step": 287
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 0.3212040364742279,
+      "learning_rate": 0.00026222222222222223,
+      "loss": 6.5525,
+      "step": 288
+    },
+    {
+      "epoch": 0.6411536328341653,
+      "grad_norm": 0.21791699528694153,
+      "learning_rate": 0.00026197530864197527,
+      "loss": 6.8778,
+      "step": 289
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 0.3303765654563904,
+      "learning_rate": 0.00026172839506172837,
+      "loss": 6.4489,
+      "step": 290
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "grad_norm": 0.2186868041753769,
+      "learning_rate": 0.00026148148148148146,
+      "loss": 6.8181,
+      "step": 291
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 0.20761652290821075,
+      "learning_rate": 0.00026123456790123456,
+      "loss": 6.9191,
+      "step": 292
+    },
+    {
+      "epoch": 0.6500277315585136,
+      "grad_norm": 0.23982930183410645,
+      "learning_rate": 0.0002609876543209876,
+      "loss": 6.9956,
+      "step": 293
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 0.2362796515226364,
+      "learning_rate": 0.0002607407407407407,
+      "loss": 6.8129,
+      "step": 294
+    },
+    {
+      "epoch": 0.6544647809206877,
+      "grad_norm": 0.22984258830547333,
+      "learning_rate": 0.00026049382716049384,
+      "loss": 6.6444,
+      "step": 295
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 0.22558103501796722,
+      "learning_rate": 0.0002602469135802469,
+      "loss": 6.8448,
+      "step": 296
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "grad_norm": 0.2152688205242157,
+      "learning_rate": 0.00026,
+      "loss": 6.7468,
+      "step": 297
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 0.22881445288658142,
+      "learning_rate": 0.00025975308641975307,
+      "loss": 6.9136,
+      "step": 298
+    },
+    {
+      "epoch": 0.663338879645036,
+      "grad_norm": 0.22075128555297852,
+      "learning_rate": 0.00025950617283950617,
+      "loss": 7.006,
+      "step": 299
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 0.19859904050827026,
+      "learning_rate": 0.0002592592592592592,
+      "loss": 6.8462,
+      "step": 300
+    },
+    {
+      "epoch": 0.6677759290072102,
+      "grad_norm": 0.24378462135791779,
+      "learning_rate": 0.0002590123456790123,
+      "loss": 6.616,
+      "step": 301
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 0.21312250196933746,
+      "learning_rate": 0.0002587654320987654,
+      "loss": 6.8333,
+      "step": 302
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "grad_norm": 0.2282208800315857,
+      "learning_rate": 0.0002585185185185185,
+      "loss": 6.7884,
+      "step": 303
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 0.21728795766830444,
+      "learning_rate": 0.0002582716049382716,
+      "loss": 6.9997,
+      "step": 304
+    },
+    {
+      "epoch": 0.6766500277315585,
+      "grad_norm": 0.2376324087381363,
+      "learning_rate": 0.0002580246913580247,
+      "loss": 6.7874,
+      "step": 305
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 0.2543889582157135,
+      "learning_rate": 0.0002577777777777778,
+      "loss": 6.732,
+      "step": 306
+    },
+    {
+      "epoch": 0.6810870770937326,
+      "grad_norm": 0.19987723231315613,
+      "learning_rate": 0.0002575308641975308,
+      "loss": 6.8344,
+      "step": 307
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 0.20890119671821594,
+      "learning_rate": 0.0002572839506172839,
+      "loss": 6.6498,
+      "step": 308
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "grad_norm": 0.20659278333187103,
+      "learning_rate": 0.000257037037037037,
+      "loss": 6.8256,
+      "step": 309
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 0.2369864583015442,
+      "learning_rate": 0.0002567901234567901,
+      "loss": 6.6539,
+      "step": 310
+    },
+    {
+      "epoch": 0.689961175818081,
+      "grad_norm": 0.20262716710567474,
+      "learning_rate": 0.0002565432098765432,
+      "loss": 6.7868,
+      "step": 311
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 0.20186154544353485,
+      "learning_rate": 0.00025629629629629624,
+      "loss": 6.8413,
+      "step": 312
+    },
+    {
+      "epoch": 0.6943982251802552,
+      "grad_norm": 0.22297048568725586,
+      "learning_rate": 0.0002560493827160494,
+      "loss": 6.7506,
+      "step": 313
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 0.19991546869277954,
+      "learning_rate": 0.00025580246913580243,
+      "loss": 6.7112,
+      "step": 314
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "grad_norm": 0.2183772176504135,
+      "learning_rate": 0.00025555555555555553,
+      "loss": 6.5262,
+      "step": 315
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 0.23636843264102936,
+      "learning_rate": 0.0002553086419753086,
+      "loss": 6.8174,
+      "step": 316
+    },
+    {
+      "epoch": 0.7032723239046035,
+      "grad_norm": 0.19868767261505127,
+      "learning_rate": 0.0002550617283950617,
+      "loss": 6.6288,
+      "step": 317
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 0.2227199673652649,
+      "learning_rate": 0.0002548148148148148,
+      "loss": 6.5862,
+      "step": 318
+    },
+    {
+      "epoch": 0.7077093732667776,
+      "grad_norm": 0.2191808521747589,
+      "learning_rate": 0.00025456790123456785,
+      "loss": 6.7763,
+      "step": 319
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 0.22202734649181366,
+      "learning_rate": 0.00025432098765432095,
+      "loss": 6.7095,
+      "step": 320
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "grad_norm": 0.19778110086917877,
+      "learning_rate": 0.00025407407407407404,
+      "loss": 6.7363,
+      "step": 321
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 0.22813093662261963,
+      "learning_rate": 0.00025382716049382714,
+      "loss": 6.7854,
+      "step": 322
+    },
+    {
+      "epoch": 0.7165834719911259,
+      "grad_norm": 0.22549612820148468,
+      "learning_rate": 0.00025358024691358023,
+      "loss": 6.6854,
+      "step": 323
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 0.2016897052526474,
+      "learning_rate": 0.00025333333333333333,
+      "loss": 6.6915,
+      "step": 324
+    },
+    {
+      "epoch": 0.7210205213533001,
+      "grad_norm": 0.20843982696533203,
+      "learning_rate": 0.0002530864197530864,
+      "loss": 6.5311,
+      "step": 325
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 0.23362088203430176,
+      "learning_rate": 0.00025283950617283947,
+      "loss": 6.7923,
+      "step": 326
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "grad_norm": 0.20451270043849945,
+      "learning_rate": 0.00025259259259259256,
+      "loss": 6.8519,
+      "step": 327
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 0.22395287454128265,
+      "learning_rate": 0.00025234567901234566,
+      "loss": 6.7353,
+      "step": 328
+    },
+    {
+      "epoch": 0.7298946200776484,
+      "grad_norm": 0.23494094610214233,
+      "learning_rate": 0.00025209876543209875,
+      "loss": 6.763,
+      "step": 329
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 0.1996716856956482,
+      "learning_rate": 0.0002518518518518518,
+      "loss": 6.6067,
+      "step": 330
+    },
+    {
+      "epoch": 0.7343316694398225,
+      "grad_norm": 0.2379232943058014,
+      "learning_rate": 0.00025160493827160494,
+      "loss": 6.8842,
+      "step": 331
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 0.270158052444458,
+      "learning_rate": 0.00025135802469135804,
+      "loss": 7.0301,
+      "step": 332
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "grad_norm": 0.19227898120880127,
+      "learning_rate": 0.0002511111111111111,
+      "loss": 6.5412,
+      "step": 333
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 0.22434109449386597,
+      "learning_rate": 0.00025086419753086417,
+      "loss": 6.7027,
+      "step": 334
+    },
+    {
+      "epoch": 0.7432057681641708,
+      "grad_norm": 0.20972467958927155,
+      "learning_rate": 0.00025061728395061727,
+      "loss": 6.6081,
+      "step": 335
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 0.2025555521249771,
+      "learning_rate": 0.00025037037037037036,
+      "loss": 6.7364,
+      "step": 336
+    },
+    {
+      "epoch": 0.747642817526345,
+      "grad_norm": 0.22312957048416138,
+      "learning_rate": 0.0002501234567901234,
+      "loss": 6.9084,
+      "step": 337
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 0.21431638300418854,
+      "learning_rate": 0.0002498765432098765,
+      "loss": 6.8974,
+      "step": 338
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "grad_norm": 0.2023218423128128,
+      "learning_rate": 0.00024962962962962965,
+      "loss": 6.738,
+      "step": 339
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 0.20649074018001556,
+      "learning_rate": 0.0002493827160493827,
+      "loss": 6.928,
+      "step": 340
+    },
+    {
+      "epoch": 0.7565169162506933,
+      "grad_norm": 0.20891088247299194,
+      "learning_rate": 0.0002491358024691358,
+      "loss": 6.5171,
+      "step": 341
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 0.20974567532539368,
+      "learning_rate": 0.0002488888888888889,
+      "loss": 6.3653,
+      "step": 342
+    },
+    {
+      "epoch": 0.7609539656128674,
+      "grad_norm": 0.1922912448644638,
+      "learning_rate": 0.000248641975308642,
+      "loss": 6.8994,
+      "step": 343
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 0.21250607073307037,
+      "learning_rate": 0.000248395061728395,
+      "loss": 6.7234,
+      "step": 344
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "grad_norm": 0.22726888954639435,
+      "learning_rate": 0.0002481481481481481,
+      "loss": 6.892,
+      "step": 345
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 0.1928975135087967,
+      "learning_rate": 0.0002479012345679012,
+      "loss": 6.5797,
+      "step": 346
+    },
+    {
+      "epoch": 0.7698280643372157,
+      "grad_norm": 0.22753247618675232,
+      "learning_rate": 0.0002476543209876543,
+      "loss": 6.6032,
+      "step": 347
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 0.21166828274726868,
+      "learning_rate": 0.0002474074074074074,
+      "loss": 6.8044,
+      "step": 348
+    },
+    {
+      "epoch": 0.7742651136993899,
+      "grad_norm": 0.201234832406044,
+      "learning_rate": 0.0002471604938271605,
+      "loss": 6.8774,
+      "step": 349
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 0.20125554502010345,
+      "learning_rate": 0.0002469135802469136,
+      "loss": 6.7167,
+      "step": 350
+    },
+    {
+      "epoch": 0.778702163061564,
+      "grad_norm": 0.18794187903404236,
+      "learning_rate": 0.0002466666666666666,
+      "loss": 6.7158,
+      "step": 351
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 0.18615470826625824,
+      "learning_rate": 0.0002464197530864197,
+      "loss": 6.5618,
+      "step": 352
+    },
+    {
+      "epoch": 0.7831392124237382,
+      "grad_norm": 0.20247696340084076,
+      "learning_rate": 0.0002461728395061728,
+      "loss": 6.8243,
+      "step": 353
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 0.23817774653434753,
+      "learning_rate": 0.0002459259259259259,
+      "loss": 6.7926,
+      "step": 354
+    },
+    {
+      "epoch": 0.7875762617859123,
+      "grad_norm": 0.2213386744260788,
+      "learning_rate": 0.000245679012345679,
+      "loss": 6.5369,
+      "step": 355
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 0.18856588006019592,
+      "learning_rate": 0.00024543209876543205,
+      "loss": 6.6819,
+      "step": 356
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "grad_norm": 0.22981244325637817,
+      "learning_rate": 0.0002451851851851852,
+      "loss": 6.6938,
+      "step": 357
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 0.19552747905254364,
+      "learning_rate": 0.00024493827160493824,
+      "loss": 6.5709,
+      "step": 358
+    },
+    {
+      "epoch": 0.7964503605102606,
+      "grad_norm": 0.19994227588176727,
+      "learning_rate": 0.00024469135802469133,
+      "loss": 6.4817,
+      "step": 359
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 0.2069280743598938,
+      "learning_rate": 0.00024444444444444443,
+      "loss": 6.6226,
+      "step": 360
+    },
+    {
+      "epoch": 0.8008874098724348,
+      "grad_norm": 0.2370084822177887,
+      "learning_rate": 0.0002441975308641975,
+      "loss": 6.4813,
+      "step": 361
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 0.20844882726669312,
+      "learning_rate": 0.00024395061728395062,
+      "loss": 6.6616,
+      "step": 362
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "grad_norm": 0.2083243876695633,
+      "learning_rate": 0.0002437037037037037,
+      "loss": 6.7965,
+      "step": 363
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 0.1990830898284912,
+      "learning_rate": 0.00024345679012345678,
+      "loss": 6.7275,
+      "step": 364
+    },
+    {
+      "epoch": 0.8097615085967831,
+      "grad_norm": 0.22553744912147522,
+      "learning_rate": 0.00024320987654320985,
+      "loss": 6.6698,
+      "step": 365
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 0.19122867286205292,
+      "learning_rate": 0.00024296296296296295,
+      "loss": 6.5848,
+      "step": 366
+    },
+    {
+      "epoch": 0.8141985579589573,
+      "grad_norm": 0.18741734325885773,
+      "learning_rate": 0.00024271604938271601,
+      "loss": 6.8291,
+      "step": 367
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 0.2022235244512558,
+      "learning_rate": 0.0002424691358024691,
+      "loss": 6.5923,
+      "step": 368
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "grad_norm": 0.19597862660884857,
+      "learning_rate": 0.0002422222222222222,
+      "loss": 6.8629,
+      "step": 369
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 0.22235015034675598,
+      "learning_rate": 0.00024197530864197527,
+      "loss": 6.6405,
+      "step": 370
+    },
+    {
+      "epoch": 0.8230726566833056,
+      "grad_norm": 0.19049280881881714,
+      "learning_rate": 0.0002417283950617284,
+      "loss": 6.5227,
+      "step": 371
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 0.22232288122177124,
+      "learning_rate": 0.00024148148148148146,
+      "loss": 6.7111,
+      "step": 372
+    },
+    {
+      "epoch": 0.8275097060454798,
+      "grad_norm": 0.20254629850387573,
+      "learning_rate": 0.00024123456790123456,
+      "loss": 6.6894,
+      "step": 373
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 0.20657235383987427,
+      "learning_rate": 0.00024098765432098763,
+      "loss": 6.4273,
+      "step": 374
+    },
+    {
+      "epoch": 0.831946755407654,
+      "grad_norm": 0.22631707787513733,
+      "learning_rate": 0.00024074074074074072,
+      "loss": 6.7432,
+      "step": 375
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 0.20757989585399628,
+      "learning_rate": 0.0002404938271604938,
+      "loss": 6.8315,
+      "step": 376
+    },
+    {
+      "epoch": 0.8363838047698281,
+      "grad_norm": 0.19854243099689484,
+      "learning_rate": 0.00024024691358024688,
+      "loss": 6.8156,
+      "step": 377
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 0.2011784315109253,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 6.7212,
+      "step": 378
+    },
+    {
+      "epoch": 0.8408208541320022,
+      "grad_norm": 0.1904282420873642,
+      "learning_rate": 0.00023975308641975305,
+      "loss": 6.6753,
+      "step": 379
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 0.20943498611450195,
+      "learning_rate": 0.00023950617283950617,
+      "loss": 6.7067,
+      "step": 380
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "grad_norm": 0.19518263638019562,
+      "learning_rate": 0.00023925925925925924,
+      "loss": 6.5098,
+      "step": 381
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 0.18988138437271118,
+      "learning_rate": 0.00023901234567901233,
+      "loss": 6.4081,
+      "step": 382
+    },
+    {
+      "epoch": 0.8496949528563505,
+      "grad_norm": 0.21128663420677185,
+      "learning_rate": 0.0002387654320987654,
+      "loss": 6.4415,
+      "step": 383
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 0.2018367350101471,
+      "learning_rate": 0.0002385185185185185,
+      "loss": 6.4838,
+      "step": 384
+    },
+    {
+      "epoch": 0.8541320022185247,
+      "grad_norm": 0.17642031610012054,
+      "learning_rate": 0.0002382716049382716,
+      "loss": 6.4359,
+      "step": 385
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 0.22816012799739838,
+      "learning_rate": 0.00023802469135802466,
+      "loss": 6.4939,
+      "step": 386
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "grad_norm": 0.19580289721488953,
+      "learning_rate": 0.00023777777777777775,
+      "loss": 6.8179,
+      "step": 387
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 0.24731579422950745,
+      "learning_rate": 0.00023753086419753082,
+      "loss": 6.7606,
+      "step": 388
+    },
+    {
+      "epoch": 0.863006100942873,
+      "grad_norm": 0.22626414895057678,
+      "learning_rate": 0.00023728395061728394,
+      "loss": 6.5185,
+      "step": 389
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 0.22785606980323792,
+      "learning_rate": 0.000237037037037037,
+      "loss": 6.7738,
+      "step": 390
+    },
+    {
+      "epoch": 0.8674431503050472,
+      "grad_norm": 0.21009711921215057,
+      "learning_rate": 0.0002367901234567901,
+      "loss": 6.6185,
+      "step": 391
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 0.22216084599494934,
+      "learning_rate": 0.0002365432098765432,
+      "loss": 6.5516,
+      "step": 392
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "grad_norm": 0.1848013997077942,
+      "learning_rate": 0.00023629629629629627,
+      "loss": 6.649,
+      "step": 393
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 0.2531176209449768,
+      "learning_rate": 0.00023604938271604937,
+      "loss": 7.0402,
+      "step": 394
+    },
+    {
+      "epoch": 0.8763172490293955,
+      "grad_norm": 0.18708011507987976,
+      "learning_rate": 0.00023580246913580243,
+      "loss": 6.4264,
+      "step": 395
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 0.2306434065103531,
+      "learning_rate": 0.00023555555555555553,
+      "loss": 6.5426,
+      "step": 396
+    },
+    {
+      "epoch": 0.8807542983915696,
+      "grad_norm": 0.20226739346981049,
+      "learning_rate": 0.0002353086419753086,
+      "loss": 6.7136,
+      "step": 397
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 0.23149196803569794,
+      "learning_rate": 0.00023506172839506172,
+      "loss": 6.6752,
+      "step": 398
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "grad_norm": 0.19331242144107819,
+      "learning_rate": 0.0002348148148148148,
+      "loss": 6.751,
+      "step": 399
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 0.20518051087856293,
+      "learning_rate": 0.00023456790123456788,
+      "loss": 6.7264,
+      "step": 400
+    },
+    {
+      "epoch": 0.8896283971159179,
+      "grad_norm": 0.24824081361293793,
+      "learning_rate": 0.00023432098765432098,
+      "loss": 6.1342,
+      "step": 401
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 0.1811339110136032,
+      "learning_rate": 0.00023407407407407405,
+      "loss": 6.6987,
+      "step": 402
+    },
+    {
+      "epoch": 0.894065446478092,
+      "grad_norm": 0.16965928673744202,
+      "learning_rate": 0.00023382716049382714,
+      "loss": 6.7013,
+      "step": 403
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 0.17786738276481628,
+      "learning_rate": 0.0002335802469135802,
+      "loss": 6.5635,
+      "step": 404
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "grad_norm": 0.22453954815864563,
+      "learning_rate": 0.0002333333333333333,
+      "loss": 6.8186,
+      "step": 405
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 0.20414142310619354,
+      "learning_rate": 0.00023308641975308637,
+      "loss": 6.5441,
+      "step": 406
+    },
+    {
+      "epoch": 0.9029395452024404,
+      "grad_norm": 0.201078400015831,
+      "learning_rate": 0.0002328395061728395,
+      "loss": 6.6341,
+      "step": 407
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 0.22197119891643524,
+      "learning_rate": 0.0002325925925925926,
+      "loss": 6.7618,
+      "step": 408
+    },
+    {
+      "epoch": 0.9073765945646145,
+      "grad_norm": 0.1963954120874405,
+      "learning_rate": 0.00023234567901234566,
+      "loss": 6.6847,
+      "step": 409
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 0.1784980744123459,
+      "learning_rate": 0.00023209876543209875,
+      "loss": 6.4572,
+      "step": 410
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "grad_norm": 0.22081004083156586,
+      "learning_rate": 0.00023185185185185182,
+      "loss": 6.3222,
+      "step": 411
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 0.2092115432024002,
+      "learning_rate": 0.00023160493827160492,
+      "loss": 6.4636,
+      "step": 412
+    },
+    {
+      "epoch": 0.9162506932889628,
+      "grad_norm": 0.21051687002182007,
+      "learning_rate": 0.00023135802469135798,
+      "loss": 6.4696,
+      "step": 413
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 0.22508850693702698,
+      "learning_rate": 0.00023111111111111108,
+      "loss": 6.6678,
+      "step": 414
+    },
+    {
+      "epoch": 0.920687742651137,
+      "grad_norm": 0.2016604244709015,
+      "learning_rate": 0.0002308641975308642,
+      "loss": 6.6161,
+      "step": 415
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 0.18506215512752533,
+      "learning_rate": 0.00023061728395061727,
+      "loss": 6.3852,
+      "step": 416
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "grad_norm": 0.19710281491279602,
+      "learning_rate": 0.00023037037037037036,
+      "loss": 6.3276,
+      "step": 417
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 0.18153361976146698,
+      "learning_rate": 0.00023012345679012343,
+      "loss": 6.4577,
+      "step": 418
+    },
+    {
+      "epoch": 0.9295618413754853,
+      "grad_norm": 0.19384880363941193,
+      "learning_rate": 0.00022987654320987653,
+      "loss": 6.5231,
+      "step": 419
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 0.17709828913211823,
+      "learning_rate": 0.0002296296296296296,
+      "loss": 6.5003,
+      "step": 420
+    },
+    {
+      "epoch": 0.9339988907376594,
+      "grad_norm": 0.19884225726127625,
+      "learning_rate": 0.0002293827160493827,
+      "loss": 6.5288,
+      "step": 421
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 0.21307963132858276,
+      "learning_rate": 0.00022913580246913579,
+      "loss": 6.7068,
+      "step": 422
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "grad_norm": 0.1869218349456787,
+      "learning_rate": 0.00022888888888888885,
+      "loss": 6.6843,
+      "step": 423
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 0.19299793243408203,
+      "learning_rate": 0.00022864197530864198,
+      "loss": 6.6051,
+      "step": 424
+    },
+    {
+      "epoch": 0.9428729894620078,
+      "grad_norm": 0.21579767763614655,
+      "learning_rate": 0.00022839506172839504,
+      "loss": 6.549,
+      "step": 425
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 0.22623635828495026,
+      "learning_rate": 0.00022814814814814814,
+      "loss": 6.6013,
+      "step": 426
+    },
+    {
+      "epoch": 0.947310038824182,
+      "grad_norm": 0.22867326438426971,
+      "learning_rate": 0.0002279012345679012,
+      "loss": 6.5156,
+      "step": 427
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 0.22814692556858063,
+      "learning_rate": 0.0002276543209876543,
+      "loss": 6.5319,
+      "step": 428
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "grad_norm": 0.16648977994918823,
+      "learning_rate": 0.00022740740740740737,
+      "loss": 6.37,
+      "step": 429
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 0.2465483844280243,
+      "learning_rate": 0.00022716049382716047,
+      "loss": 6.629,
+      "step": 430
+    },
+    {
+      "epoch": 0.9561841375485303,
+      "grad_norm": 0.17856603860855103,
+      "learning_rate": 0.00022691358024691356,
+      "loss": 6.4716,
+      "step": 431
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 0.1684563010931015,
+      "learning_rate": 0.00022666666666666663,
+      "loss": 6.5235,
+      "step": 432
+    },
+    {
+      "epoch": 0.9606211869107044,
+      "grad_norm": 0.19260463118553162,
+      "learning_rate": 0.00022641975308641975,
+      "loss": 6.654,
+      "step": 433
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 0.20749062299728394,
+      "learning_rate": 0.00022617283950617282,
+      "loss": 6.509,
+      "step": 434
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "grad_norm": 0.19797588884830475,
+      "learning_rate": 0.00022592592592592591,
+      "loss": 6.1826,
+      "step": 435
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 0.1886669397354126,
+      "learning_rate": 0.00022567901234567898,
+      "loss": 6.6416,
+      "step": 436
+    },
+    {
+      "epoch": 0.9694952856350527,
+      "grad_norm": 0.20788297057151794,
+      "learning_rate": 0.00022543209876543208,
+      "loss": 6.6273,
+      "step": 437
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 0.23814578354358673,
+      "learning_rate": 0.00022518518518518517,
+      "loss": 6.5548,
+      "step": 438
+    },
+    {
+      "epoch": 0.9739323349972269,
+      "grad_norm": 0.2105850726366043,
+      "learning_rate": 0.00022493827160493824,
+      "loss": 6.6328,
+      "step": 439
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 0.19008755683898926,
+      "learning_rate": 0.00022469135802469134,
+      "loss": 6.5152,
+      "step": 440
+    },
+    {
+      "epoch": 0.978369384359401,
+      "grad_norm": 0.19722850620746613,
+      "learning_rate": 0.0002244444444444444,
+      "loss": 6.3036,
+      "step": 441
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 0.20725344121456146,
+      "learning_rate": 0.00022419753086419753,
+      "loss": 6.3965,
+      "step": 442
+    },
+    {
+      "epoch": 0.9828064337215752,
+      "grad_norm": 0.18219025433063507,
+      "learning_rate": 0.0002239506172839506,
+      "loss": 6.8192,
+      "step": 443
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 0.18593676388263702,
+      "learning_rate": 0.0002237037037037037,
+      "loss": 6.443,
+      "step": 444
+    },
+    {
+      "epoch": 0.9872434830837493,
+      "grad_norm": 0.1842203438282013,
+      "learning_rate": 0.00022345679012345678,
+      "loss": 6.3543,
+      "step": 445
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 0.20712865889072418,
+      "learning_rate": 0.00022320987654320985,
+      "loss": 6.3589,
+      "step": 446
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "grad_norm": 0.22042232751846313,
+      "learning_rate": 0.00022296296296296295,
+      "loss": 6.1246,
+      "step": 447
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 0.20385918021202087,
+      "learning_rate": 0.00022271604938271602,
+      "loss": 6.6235,
+      "step": 448
+    },
+    {
+      "epoch": 0.9961175818080976,
+      "grad_norm": 0.19405104219913483,
+      "learning_rate": 0.0002224691358024691,
+      "loss": 6.3476,
+      "step": 449
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 0.21469907462596893,
+      "learning_rate": 0.00022222222222222218,
+      "loss": 6.4004,
+      "step": 450
+    },
+    {
+      "epoch": 1.0005546311702718,
+      "grad_norm": 0.19176262617111206,
+      "learning_rate": 0.0002219753086419753,
+      "loss": 6.6522,
+      "step": 451
+    },
+    {
+      "epoch": 1.0027731558513588,
+      "grad_norm": 0.2146284431219101,
+      "learning_rate": 0.0002217283950617284,
+      "loss": 6.5931,
+      "step": 452
+    },
+    {
+      "epoch": 1.004991680532446,
+      "grad_norm": 0.17713655531406403,
+      "learning_rate": 0.00022148148148148146,
+      "loss": 6.6465,
+      "step": 453
+    },
+    {
+      "epoch": 1.007210205213533,
+      "grad_norm": 0.2074706107378006,
+      "learning_rate": 0.00022123456790123456,
+      "loss": 6.3696,
+      "step": 454
+    },
+    {
+      "epoch": 1.00942872989462,
+      "grad_norm": 0.21191412210464478,
+      "learning_rate": 0.00022098765432098763,
+      "loss": 6.4346,
+      "step": 455
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 0.19734513759613037,
+      "learning_rate": 0.00022074074074074072,
+      "loss": 6.6394,
+      "step": 456
+    },
+    {
+      "epoch": 1.0138657792567942,
+      "grad_norm": 0.24888208508491516,
+      "learning_rate": 0.0002204938271604938,
+      "loss": 6.156,
+      "step": 457
+    },
+    {
+      "epoch": 1.0160843039378813,
+      "grad_norm": 0.1916578710079193,
+      "learning_rate": 0.00022024691358024689,
+      "loss": 6.4137,
+      "step": 458
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "grad_norm": 0.19581283628940582,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 6.4642,
+      "step": 459
+    },
+    {
+      "epoch": 1.0205213533000554,
+      "grad_norm": 0.1898852437734604,
+      "learning_rate": 0.00021975308641975308,
+      "loss": 6.4524,
+      "step": 460
+    },
+    {
+      "epoch": 1.0227398779811425,
+      "grad_norm": 0.17808125913143158,
+      "learning_rate": 0.00021950617283950617,
+      "loss": 6.2646,
+      "step": 461
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 0.18140365183353424,
+      "learning_rate": 0.00021925925925925924,
+      "loss": 6.5831,
+      "step": 462
+    },
+    {
+      "epoch": 1.0271769273433167,
+      "grad_norm": 0.2563443183898926,
+      "learning_rate": 0.00021901234567901233,
+      "loss": 6.5449,
+      "step": 463
+    },
+    {
+      "epoch": 1.0293954520244037,
+      "grad_norm": 0.1831277310848236,
+      "learning_rate": 0.0002187654320987654,
+      "loss": 6.5503,
+      "step": 464
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 0.19348184764385223,
+      "learning_rate": 0.0002185185185185185,
+      "loss": 6.5667,
+      "step": 465
+    },
+    {
+      "epoch": 1.033832501386578,
+      "grad_norm": 0.18697039783000946,
+      "learning_rate": 0.00021827160493827157,
+      "loss": 6.2686,
+      "step": 466
+    },
+    {
+      "epoch": 1.036051026067665,
+      "grad_norm": 0.16921605169773102,
+      "learning_rate": 0.00021802469135802466,
+      "loss": 6.4516,
+      "step": 467
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 0.1986284852027893,
+      "learning_rate": 0.00021777777777777778,
+      "loss": 6.2106,
+      "step": 468
+    },
+    {
+      "epoch": 1.0404880754298391,
+      "grad_norm": 0.18519148230552673,
+      "learning_rate": 0.00021753086419753085,
+      "loss": 6.4151,
+      "step": 469
+    },
+    {
+      "epoch": 1.0427066001109262,
+      "grad_norm": 0.1877288967370987,
+      "learning_rate": 0.00021728395061728395,
+      "loss": 6.3071,
+      "step": 470
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "grad_norm": 0.2320905178785324,
+      "learning_rate": 0.00021703703703703701,
+      "loss": 6.1191,
+      "step": 471
+    },
+    {
+      "epoch": 1.0471436494731003,
+      "grad_norm": 0.20258858799934387,
+      "learning_rate": 0.0002167901234567901,
+      "loss": 6.3574,
+      "step": 472
+    },
+    {
+      "epoch": 1.0493621741541874,
+      "grad_norm": 0.2265845686197281,
+      "learning_rate": 0.00021654320987654318,
+      "loss": 6.4436,
+      "step": 473
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 0.18015384674072266,
+      "learning_rate": 0.00021629629629629627,
+      "loss": 6.5596,
+      "step": 474
+    },
+    {
+      "epoch": 1.0537992235163616,
+      "grad_norm": 0.1810697615146637,
+      "learning_rate": 0.00021604938271604937,
+      "loss": 6.5247,
+      "step": 475
+    },
+    {
+      "epoch": 1.0560177481974486,
+      "grad_norm": 0.21950900554656982,
+      "learning_rate": 0.00021580246913580244,
+      "loss": 6.451,
+      "step": 476
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "grad_norm": 0.2353959083557129,
+      "learning_rate": 0.00021555555555555556,
+      "loss": 5.9724,
+      "step": 477
+    },
+    {
+      "epoch": 1.0604547975596228,
+      "grad_norm": 0.1910317838191986,
+      "learning_rate": 0.00021530864197530863,
+      "loss": 6.3963,
+      "step": 478
+    },
+    {
+      "epoch": 1.0626733222407099,
+      "grad_norm": 0.1783912032842636,
+      "learning_rate": 0.00021506172839506172,
+      "loss": 6.4344,
+      "step": 479
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 0.2117423713207245,
+      "learning_rate": 0.0002148148148148148,
+      "loss": 6.6353,
+      "step": 480
+    },
+    {
+      "epoch": 1.067110371602884,
+      "grad_norm": 0.1837896704673767,
+      "learning_rate": 0.00021456790123456788,
+      "loss": 6.2499,
+      "step": 481
+    },
+    {
+      "epoch": 1.069328896283971,
+      "grad_norm": 0.20292581617832184,
+      "learning_rate": 0.00021432098765432095,
+      "loss": 6.3629,
+      "step": 482
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "grad_norm": 0.2186204344034195,
+      "learning_rate": 0.00021407407407407405,
+      "loss": 6.4657,
+      "step": 483
+    },
+    {
+      "epoch": 1.0737659456461452,
+      "grad_norm": 0.22919540107250214,
+      "learning_rate": 0.00021382716049382714,
+      "loss": 6.6481,
+      "step": 484
+    },
+    {
+      "epoch": 1.0759844703272323,
+      "grad_norm": 0.19985345005989075,
+      "learning_rate": 0.0002135802469135802,
+      "loss": 6.3032,
+      "step": 485
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 0.21319784224033356,
+      "learning_rate": 0.00021333333333333333,
+      "loss": 6.7609,
+      "step": 486
+    },
+    {
+      "epoch": 1.0804215196894065,
+      "grad_norm": 0.21141400933265686,
+      "learning_rate": 0.0002130864197530864,
+      "loss": 6.4678,
+      "step": 487
+    },
+    {
+      "epoch": 1.0826400443704935,
+      "grad_norm": 0.2064743936061859,
+      "learning_rate": 0.0002128395061728395,
+      "loss": 6.4489,
+      "step": 488
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "grad_norm": 0.20920738577842712,
+      "learning_rate": 0.00021259259259259256,
+      "loss": 6.6139,
+      "step": 489
+    },
+    {
+      "epoch": 1.0870770937326677,
+      "grad_norm": 0.1768214851617813,
+      "learning_rate": 0.00021234567901234566,
+      "loss": 6.5732,
+      "step": 490
+    },
+    {
+      "epoch": 1.0892956184137548,
+      "grad_norm": 0.27665355801582336,
+      "learning_rate": 0.00021209876543209875,
+      "loss": 6.3108,
+      "step": 491
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 0.22936174273490906,
+      "learning_rate": 0.00021185185185185182,
+      "loss": 6.6017,
+      "step": 492
+    },
+    {
+      "epoch": 1.093732667775929,
+      "grad_norm": 0.22160974144935608,
+      "learning_rate": 0.00021160493827160492,
+      "loss": 6.6751,
+      "step": 493
+    },
+    {
+      "epoch": 1.095951192457016,
+      "grad_norm": 0.18380580842494965,
+      "learning_rate": 0.00021135802469135799,
+      "loss": 6.2872,
+      "step": 494
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 0.2047499120235443,
+      "learning_rate": 0.0002111111111111111,
+      "loss": 6.4333,
+      "step": 495
+    },
+    {
+      "epoch": 1.1003882418191901,
+      "grad_norm": 0.17706413567066193,
+      "learning_rate": 0.00021086419753086418,
+      "loss": 6.5059,
+      "step": 496
+    },
+    {
+      "epoch": 1.1026067665002772,
+      "grad_norm": 0.18459023535251617,
+      "learning_rate": 0.00021061728395061727,
+      "loss": 6.5914,
+      "step": 497
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 0.18384850025177002,
+      "learning_rate": 0.00021037037037037037,
+      "loss": 6.3995,
+      "step": 498
+    },
+    {
+      "epoch": 1.1070438158624514,
+      "grad_norm": 0.21881327033042908,
+      "learning_rate": 0.00021012345679012343,
+      "loss": 6.4929,
+      "step": 499
+    },
+    {
+      "epoch": 1.1092623405435384,
+      "grad_norm": 0.19585120677947998,
+      "learning_rate": 0.00020987654320987653,
+      "loss": 6.496,
+      "step": 500
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "grad_norm": 0.2033056616783142,
+      "learning_rate": 0.0002096296296296296,
+      "loss": 6.3576,
+      "step": 501
+    },
+    {
+      "epoch": 1.1136993899057126,
+      "grad_norm": 0.19977574050426483,
+      "learning_rate": 0.0002093827160493827,
+      "loss": 6.6284,
+      "step": 502
+    },
+    {
+      "epoch": 1.1159179145867997,
+      "grad_norm": 0.23055507242679596,
+      "learning_rate": 0.00020913580246913576,
+      "loss": 6.6794,
+      "step": 503
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 0.20923735201358795,
+      "learning_rate": 0.00020888888888888888,
+      "loss": 6.2955,
+      "step": 504
+    },
+    {
+      "epoch": 1.1203549639489738,
+      "grad_norm": 0.17264512181282043,
+      "learning_rate": 0.00020864197530864198,
+      "loss": 6.5346,
+      "step": 505
+    },
+    {
+      "epoch": 1.122573488630061,
+      "grad_norm": 0.175466388463974,
+      "learning_rate": 0.00020839506172839505,
+      "loss": 6.3655,
+      "step": 506
+    },
+    {
+      "epoch": 1.124792013311148,
+      "grad_norm": 0.1922537237405777,
+      "learning_rate": 0.00020814814814814814,
+      "loss": 6.6118,
+      "step": 507
+    },
+    {
+      "epoch": 1.127010537992235,
+      "grad_norm": 0.24795746803283691,
+      "learning_rate": 0.0002079012345679012,
+      "loss": 6.0227,
+      "step": 508
+    },
+    {
+      "epoch": 1.1292290626733221,
+      "grad_norm": 0.22703388333320618,
+      "learning_rate": 0.0002076543209876543,
+      "loss": 6.4168,
+      "step": 509
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 0.18241427838802338,
+      "learning_rate": 0.00020740740740740737,
+      "loss": 6.4846,
+      "step": 510
+    },
+    {
+      "epoch": 1.1336661120354963,
+      "grad_norm": 0.2413383424282074,
+      "learning_rate": 0.00020716049382716047,
+      "loss": 6.8129,
+      "step": 511
+    },
+    {
+      "epoch": 1.1358846367165834,
+      "grad_norm": 0.18262635171413422,
+      "learning_rate": 0.00020691358024691354,
+      "loss": 6.4401,
+      "step": 512
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "grad_norm": 0.19467784464359283,
+      "learning_rate": 0.00020666666666666666,
+      "loss": 6.3025,
+      "step": 513
+    },
+    {
+      "epoch": 1.1403216860787575,
+      "grad_norm": 0.24388191103935242,
+      "learning_rate": 0.00020641975308641975,
+      "loss": 6.0059,
+      "step": 514
+    },
+    {
+      "epoch": 1.1425402107598448,
+      "grad_norm": 0.2323257476091385,
+      "learning_rate": 0.00020617283950617282,
+      "loss": 6.3152,
+      "step": 515
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 0.2733853757381439,
+      "learning_rate": 0.00020592592592592592,
+      "loss": 6.8396,
+      "step": 516
+    },
+    {
+      "epoch": 1.146977260122019,
+      "grad_norm": 0.24921198189258575,
+      "learning_rate": 0.00020567901234567898,
+      "loss": 6.4865,
+      "step": 517
+    },
+    {
+      "epoch": 1.149195784803106,
+      "grad_norm": 0.21880708634853363,
+      "learning_rate": 0.00020543209876543208,
+      "loss": 6.3119,
+      "step": 518
+    },
+    {
+      "epoch": 1.151414309484193,
+      "grad_norm": 0.18126995861530304,
+      "learning_rate": 0.00020518518518518515,
+      "loss": 6.4942,
+      "step": 519
+    },
+    {
+      "epoch": 1.1536328341652802,
+      "grad_norm": 0.20584796369075775,
+      "learning_rate": 0.00020493827160493824,
+      "loss": 6.154,
+      "step": 520
+    },
+    {
+      "epoch": 1.1558513588463672,
+      "grad_norm": 0.20470276474952698,
+      "learning_rate": 0.00020469135802469136,
+      "loss": 6.3502,
+      "step": 521
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 0.19381025433540344,
+      "learning_rate": 0.00020444444444444443,
+      "loss": 6.2757,
+      "step": 522
+    },
+    {
+      "epoch": 1.1602884082085414,
+      "grad_norm": 0.19511373341083527,
+      "learning_rate": 0.00020419753086419753,
+      "loss": 6.4155,
+      "step": 523
+    },
+    {
+      "epoch": 1.1625069328896285,
+      "grad_norm": 0.19305947422981262,
+      "learning_rate": 0.0002039506172839506,
+      "loss": 6.4542,
+      "step": 524
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 0.21055199205875397,
+      "learning_rate": 0.0002037037037037037,
+      "loss": 6.4569,
+      "step": 525
+    },
+    {
+      "epoch": 1.1669439822518026,
+      "grad_norm": 0.21129964292049408,
+      "learning_rate": 0.00020345679012345676,
+      "loss": 6.3384,
+      "step": 526
+    },
+    {
+      "epoch": 1.1691625069328897,
+      "grad_norm": 0.20662035048007965,
+      "learning_rate": 0.00020320987654320985,
+      "loss": 6.1911,
+      "step": 527
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 0.23567277193069458,
+      "learning_rate": 0.00020296296296296295,
+      "loss": 6.6356,
+      "step": 528
+    },
+    {
+      "epoch": 1.1735995562950639,
+      "grad_norm": 0.20886799693107605,
+      "learning_rate": 0.00020271604938271602,
+      "loss": 6.3459,
+      "step": 529
+    },
+    {
+      "epoch": 1.175818080976151,
+      "grad_norm": 0.1925199031829834,
+      "learning_rate": 0.00020246913580246914,
+      "loss": 6.458,
+      "step": 530
+    },
+    {
+      "epoch": 1.178036605657238,
+      "grad_norm": 0.27434390783309937,
+      "learning_rate": 0.0002022222222222222,
+      "loss": 6.3863,
+      "step": 531
+    },
+    {
+      "epoch": 1.180255130338325,
+      "grad_norm": 0.2276240438222885,
+      "learning_rate": 0.0002019753086419753,
+      "loss": 6.4687,
+      "step": 532
+    },
+    {
+      "epoch": 1.1824736550194122,
+      "grad_norm": 0.23115848004817963,
+      "learning_rate": 0.00020172839506172837,
+      "loss": 5.9567,
+      "step": 533
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 0.1936827152967453,
+      "learning_rate": 0.00020148148148148147,
+      "loss": 6.1256,
+      "step": 534
+    },
+    {
+      "epoch": 1.1869107043815863,
+      "grad_norm": 0.23349805176258087,
+      "learning_rate": 0.00020123456790123456,
+      "loss": 6.5949,
+      "step": 535
+    },
+    {
+      "epoch": 1.1891292290626734,
+      "grad_norm": 0.20933206379413605,
+      "learning_rate": 0.00020098765432098763,
+      "loss": 6.0997,
+      "step": 536
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "grad_norm": 0.2081373780965805,
+      "learning_rate": 0.00020074074074074072,
+      "loss": 6.3459,
+      "step": 537
+    },
+    {
+      "epoch": 1.1935662784248475,
+      "grad_norm": 0.18655778467655182,
+      "learning_rate": 0.0002004938271604938,
+      "loss": 6.1833,
+      "step": 538
+    },
+    {
+      "epoch": 1.1957848031059346,
+      "grad_norm": 0.20673008263111115,
+      "learning_rate": 0.00020024691358024691,
+      "loss": 6.5906,
+      "step": 539
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 0.21831294894218445,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 6.1567,
+      "step": 540
+    },
+    {
+      "epoch": 1.2002218524681088,
+      "grad_norm": 0.2196969985961914,
+      "learning_rate": 0.00019975308641975308,
+      "loss": 6.377,
+      "step": 541
+    },
+    {
+      "epoch": 1.2024403771491958,
+      "grad_norm": 0.19153037667274475,
+      "learning_rate": 0.00019950617283950615,
+      "loss": 6.4193,
+      "step": 542
+    },
+    {
+      "epoch": 1.204658901830283,
+      "grad_norm": 0.18704770505428314,
+      "learning_rate": 0.00019925925925925924,
+      "loss": 6.4304,
+      "step": 543
+    },
+    {
+      "epoch": 1.20687742651137,
+      "grad_norm": 0.22730767726898193,
+      "learning_rate": 0.00019901234567901234,
+      "loss": 6.3007,
+      "step": 544
+    },
+    {
+      "epoch": 1.209095951192457,
+      "grad_norm": 0.23860228061676025,
+      "learning_rate": 0.0001987654320987654,
+      "loss": 6.2225,
+      "step": 545
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 0.1969393640756607,
+      "learning_rate": 0.0001985185185185185,
+      "loss": 6.1789,
+      "step": 546
+    },
+    {
+      "epoch": 1.2135330005546312,
+      "grad_norm": 0.20663700997829437,
+      "learning_rate": 0.00019827160493827157,
+      "loss": 6.1664,
+      "step": 547
+    },
+    {
+      "epoch": 1.2157515252357183,
+      "grad_norm": 0.1883891373872757,
+      "learning_rate": 0.0001980246913580247,
+      "loss": 6.2045,
+      "step": 548
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "grad_norm": 0.20720022916793823,
+      "learning_rate": 0.00019777777777777776,
+      "loss": 6.536,
+      "step": 549
+    },
+    {
+      "epoch": 1.2201885745978924,
+      "grad_norm": 0.20909571647644043,
+      "learning_rate": 0.00019753086419753085,
+      "loss": 6.2427,
+      "step": 550
+    },
+    {
+      "epoch": 1.2224070992789795,
+      "grad_norm": 0.19228670001029968,
+      "learning_rate": 0.00019728395061728395,
+      "loss": 6.3351,
+      "step": 551
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 0.22943055629730225,
+      "learning_rate": 0.00019703703703703702,
+      "loss": 6.3276,
+      "step": 552
+    },
+    {
+      "epoch": 1.2268441486411537,
+      "grad_norm": 0.1797686070203781,
+      "learning_rate": 0.0001967901234567901,
+      "loss": 6.2766,
+      "step": 553
+    },
+    {
+      "epoch": 1.2290626733222407,
+      "grad_norm": 0.1983470618724823,
+      "learning_rate": 0.00019654320987654318,
+      "loss": 6.3756,
+      "step": 554
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 0.17649473249912262,
+      "learning_rate": 0.00019629629629629627,
+      "loss": 6.3365,
+      "step": 555
+    },
+    {
+      "epoch": 1.2334997226844149,
+      "grad_norm": 0.23960424959659576,
+      "learning_rate": 0.00019604938271604934,
+      "loss": 6.4012,
+      "step": 556
+    },
+    {
+      "epoch": 1.235718247365502,
+      "grad_norm": 0.26655280590057373,
+      "learning_rate": 0.00019580246913580246,
+      "loss": 6.474,
+      "step": 557
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 0.18394790589809418,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 6.1055,
+      "step": 558
+    },
+    {
+      "epoch": 1.240155296727676,
+      "grad_norm": 0.21624158322811127,
+      "learning_rate": 0.00019530864197530863,
+      "loss": 6.3593,
+      "step": 559
+    },
+    {
+      "epoch": 1.2423738214087632,
+      "grad_norm": 0.20063906908035278,
+      "learning_rate": 0.00019506172839506172,
+      "loss": 6.2919,
+      "step": 560
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "grad_norm": 0.2094087451696396,
+      "learning_rate": 0.0001948148148148148,
+      "loss": 6.0082,
+      "step": 561
+    },
+    {
+      "epoch": 1.2468108707709373,
+      "grad_norm": 0.26124051213264465,
+      "learning_rate": 0.00019456790123456789,
+      "loss": 6.4618,
+      "step": 562
+    },
+    {
+      "epoch": 1.2490293954520244,
+      "grad_norm": 0.20131738483905792,
+      "learning_rate": 0.00019432098765432095,
+      "loss": 6.2538,
+      "step": 563
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 0.18895308673381805,
+      "learning_rate": 0.00019407407407407405,
+      "loss": 6.2691,
+      "step": 564
+    },
+    {
+      "epoch": 1.2534664448141986,
+      "grad_norm": 0.18961921334266663,
+      "learning_rate": 0.00019382716049382712,
+      "loss": 6.3447,
+      "step": 565
+    },
+    {
+      "epoch": 1.2556849694952856,
+      "grad_norm": 0.25451600551605225,
+      "learning_rate": 0.00019358024691358024,
+      "loss": 6.469,
+      "step": 566
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "grad_norm": 0.19923527538776398,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 6.4509,
+      "step": 567
+    },
+    {
+      "epoch": 1.2601220188574598,
+      "grad_norm": 0.22025370597839355,
+      "learning_rate": 0.0001930864197530864,
+      "loss": 6.1594,
+      "step": 568
+    },
+    {
+      "epoch": 1.2623405435385469,
+      "grad_norm": 0.18278685212135315,
+      "learning_rate": 0.0001928395061728395,
+      "loss": 6.2475,
+      "step": 569
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 0.18018651008605957,
+      "learning_rate": 0.00019259259259259257,
+      "loss": 6.0812,
+      "step": 570
+    },
+    {
+      "epoch": 1.266777592900721,
+      "grad_norm": 0.2012636512517929,
+      "learning_rate": 0.00019234567901234566,
+      "loss": 6.0023,
+      "step": 571
+    },
+    {
+      "epoch": 1.268996117581808,
+      "grad_norm": 0.19658373296260834,
+      "learning_rate": 0.00019209876543209873,
+      "loss": 6.3747,
+      "step": 572
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "grad_norm": 0.21594876050949097,
+      "learning_rate": 0.00019185185185185182,
+      "loss": 6.3939,
+      "step": 573
+    },
+    {
+      "epoch": 1.2734331669439822,
+      "grad_norm": 0.1881452053785324,
+      "learning_rate": 0.00019160493827160495,
+      "loss": 6.177,
+      "step": 574
+    },
+    {
+      "epoch": 1.2756516916250693,
+      "grad_norm": 0.19228719174861908,
+      "learning_rate": 0.00019135802469135801,
+      "loss": 6.5463,
+      "step": 575
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 0.23175892233848572,
+      "learning_rate": 0.0001911111111111111,
+      "loss": 5.8913,
+      "step": 576
+    },
+    {
+      "epoch": 1.2800887409872435,
+      "grad_norm": 0.2151862382888794,
+      "learning_rate": 0.00019086419753086418,
+      "loss": 6.4571,
+      "step": 577
+    },
+    {
+      "epoch": 1.2823072656683305,
+      "grad_norm": 0.19973792135715485,
+      "learning_rate": 0.00019061728395061727,
+      "loss": 6.4016,
+      "step": 578
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "grad_norm": 0.21149127185344696,
+      "learning_rate": 0.00019037037037037034,
+      "loss": 6.0528,
+      "step": 579
+    },
+    {
+      "epoch": 1.2867443150305047,
+      "grad_norm": 0.280665785074234,
+      "learning_rate": 0.00019012345679012344,
+      "loss": 6.065,
+      "step": 580
+    },
+    {
+      "epoch": 1.2889628397115918,
+      "grad_norm": 0.2355545610189438,
+      "learning_rate": 0.00018987654320987653,
+      "loss": 6.3773,
+      "step": 581
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 0.1583820879459381,
+      "learning_rate": 0.0001896296296296296,
+      "loss": 6.1666,
+      "step": 582
+    },
+    {
+      "epoch": 1.293399889073766,
+      "grad_norm": 0.20637574791908264,
+      "learning_rate": 0.00018938271604938272,
+      "loss": 5.8939,
+      "step": 583
+    },
+    {
+      "epoch": 1.295618413754853,
+      "grad_norm": 0.22117502987384796,
+      "learning_rate": 0.0001891358024691358,
+      "loss": 6.1958,
+      "step": 584
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 0.20912301540374756,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 6.3313,
+      "step": 585
+    },
+    {
+      "epoch": 1.3000554631170271,
+      "grad_norm": 0.2292405664920807,
+      "learning_rate": 0.00018864197530864195,
+      "loss": 6.3022,
+      "step": 586
+    },
+    {
+      "epoch": 1.3022739877981142,
+      "grad_norm": 0.19287514686584473,
+      "learning_rate": 0.00018839506172839505,
+      "loss": 6.3343,
+      "step": 587
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 0.20231373608112335,
+      "learning_rate": 0.00018814814814814814,
+      "loss": 6.4276,
+      "step": 588
+    },
+    {
+      "epoch": 1.3067110371602884,
+      "grad_norm": 0.19067882001399994,
+      "learning_rate": 0.0001879012345679012,
+      "loss": 6.3753,
+      "step": 589
+    },
+    {
+      "epoch": 1.3089295618413754,
+      "grad_norm": 0.19361427426338196,
+      "learning_rate": 0.0001876543209876543,
+      "loss": 6.1037,
+      "step": 590
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "grad_norm": 0.1861705332994461,
+      "learning_rate": 0.00018740740740740737,
+      "loss": 6.3874,
+      "step": 591
+    },
+    {
+      "epoch": 1.3133666112035496,
+      "grad_norm": 0.19755923748016357,
+      "learning_rate": 0.0001871604938271605,
+      "loss": 6.1248,
+      "step": 592
+    },
+    {
+      "epoch": 1.3155851358846367,
+      "grad_norm": 0.23093250393867493,
+      "learning_rate": 0.00018691358024691356,
+      "loss": 6.315,
+      "step": 593
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 0.16562411189079285,
+      "learning_rate": 0.00018666666666666666,
+      "loss": 6.4112,
+      "step": 594
+    },
+    {
+      "epoch": 1.3200221852468108,
+      "grad_norm": 0.20368768274784088,
+      "learning_rate": 0.00018641975308641973,
+      "loss": 6.4026,
+      "step": 595
+    },
+    {
+      "epoch": 1.3222407099278979,
+      "grad_norm": 0.20246431231498718,
+      "learning_rate": 0.00018617283950617282,
+      "loss": 6.3962,
+      "step": 596
+    },
+    {
+      "epoch": 1.324459234608985,
+      "grad_norm": 0.2142568975687027,
+      "learning_rate": 0.00018592592592592592,
+      "loss": 6.0198,
+      "step": 597
+    },
+    {
+      "epoch": 1.326677759290072,
+      "grad_norm": 0.23216456174850464,
+      "learning_rate": 0.00018567901234567899,
+      "loss": 6.1503,
+      "step": 598
+    },
+    {
+      "epoch": 1.328896283971159,
+      "grad_norm": 0.2031281292438507,
+      "learning_rate": 0.00018543209876543208,
+      "loss": 6.1189,
+      "step": 599
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 0.18922288715839386,
+      "learning_rate": 0.00018518518518518515,
+      "loss": 6.2074,
+      "step": 600
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.23726138472557068,
+      "learning_rate": 0.00018493827160493827,
+      "loss": 6.2106,
+      "step": 601
+    },
+    {
+      "epoch": 1.3355518580144203,
+      "grad_norm": 0.20407429337501526,
+      "learning_rate": 0.00018469135802469134,
+      "loss": 6.4764,
+      "step": 602
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "grad_norm": 0.1751442849636078,
+      "learning_rate": 0.00018444444444444443,
+      "loss": 6.1923,
+      "step": 603
+    },
+    {
+      "epoch": 1.3399889073765945,
+      "grad_norm": 0.23131608963012695,
+      "learning_rate": 0.00018419753086419753,
+      "loss": 6.7351,
+      "step": 604
+    },
+    {
+      "epoch": 1.3422074320576816,
+      "grad_norm": 0.19263847172260284,
+      "learning_rate": 0.0001839506172839506,
+      "loss": 6.4053,
+      "step": 605
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 0.21453282237052917,
+      "learning_rate": 0.0001837037037037037,
+      "loss": 6.4003,
+      "step": 606
+    },
+    {
+      "epoch": 1.346644481419856,
+      "grad_norm": 0.19269847869873047,
+      "learning_rate": 0.00018345679012345676,
+      "loss": 6.4137,
+      "step": 607
+    },
+    {
+      "epoch": 1.348863006100943,
+      "grad_norm": 0.1817368119955063,
+      "learning_rate": 0.00018320987654320986,
+      "loss": 6.0658,
+      "step": 608
+    },
+    {
+      "epoch": 1.35108153078203,
+      "grad_norm": 0.18056949973106384,
+      "learning_rate": 0.00018296296296296292,
+      "loss": 6.4385,
+      "step": 609
+    },
+    {
+      "epoch": 1.3533000554631172,
+      "grad_norm": 0.2026127278804779,
+      "learning_rate": 0.00018271604938271605,
+      "loss": 6.0226,
+      "step": 610
+    },
+    {
+      "epoch": 1.3555185801442042,
+      "grad_norm": 0.20608176290988922,
+      "learning_rate": 0.00018246913580246914,
+      "loss": 6.2652,
+      "step": 611
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 0.2840122580528259,
+      "learning_rate": 0.0001822222222222222,
+      "loss": 6.3847,
+      "step": 612
+    },
+    {
+      "epoch": 1.3599556295063784,
+      "grad_norm": 0.22608201205730438,
+      "learning_rate": 0.0001819753086419753,
+      "loss": 6.0664,
+      "step": 613
+    },
+    {
+      "epoch": 1.3621741541874655,
+      "grad_norm": 0.17816504836082458,
+      "learning_rate": 0.00018172839506172837,
+      "loss": 6.324,
+      "step": 614
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 0.27119243144989014,
+      "learning_rate": 0.00018148148148148147,
+      "loss": 6.3098,
+      "step": 615
+    },
+    {
+      "epoch": 1.3666112035496396,
+      "grad_norm": 0.19318264722824097,
+      "learning_rate": 0.00018123456790123454,
+      "loss": 6.3373,
+      "step": 616
+    },
+    {
+      "epoch": 1.3688297282307267,
+      "grad_norm": 0.2541736960411072,
+      "learning_rate": 0.00018098765432098763,
+      "loss": 6.211,
+      "step": 617
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 0.19523903727531433,
+      "learning_rate": 0.00018074074074074075,
+      "loss": 6.3743,
+      "step": 618
+    },
+    {
+      "epoch": 1.3732667775929008,
+      "grad_norm": 0.20722666382789612,
+      "learning_rate": 0.00018049382716049382,
+      "loss": 6.0889,
+      "step": 619
+    },
+    {
+      "epoch": 1.375485302273988,
+      "grad_norm": 0.18086417019367218,
+      "learning_rate": 0.00018024691358024692,
+      "loss": 6.2265,
+      "step": 620
+    },
+    {
+      "epoch": 1.377703826955075,
+      "grad_norm": 0.19405515491962433,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 6.4336,
+      "step": 621
+    },
+    {
+      "epoch": 1.379922351636162,
+      "grad_norm": 0.20033296942710876,
+      "learning_rate": 0.00017975308641975308,
+      "loss": 6.2734,
+      "step": 622
+    },
+    {
+      "epoch": 1.3821408763172491,
+      "grad_norm": 0.18684022128582,
+      "learning_rate": 0.00017950617283950615,
+      "loss": 6.2544,
+      "step": 623
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 0.23120370507240295,
+      "learning_rate": 0.00017925925925925924,
+      "loss": 6.2104,
+      "step": 624
+    },
+    {
+      "epoch": 1.3865779256794233,
+      "grad_norm": 0.22180043160915375,
+      "learning_rate": 0.0001790123456790123,
+      "loss": 6.3484,
+      "step": 625
+    },
+    {
+      "epoch": 1.3887964503605104,
+      "grad_norm": 0.22947464883327484,
+      "learning_rate": 0.0001787654320987654,
+      "loss": 5.9047,
+      "step": 626
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "grad_norm": 0.24498234689235687,
+      "learning_rate": 0.00017851851851851853,
+      "loss": 5.9456,
+      "step": 627
+    },
+    {
+      "epoch": 1.3932334997226845,
+      "grad_norm": 0.18429364264011383,
+      "learning_rate": 0.0001782716049382716,
+      "loss": 6.2798,
+      "step": 628
+    },
+    {
+      "epoch": 1.3954520244037716,
+      "grad_norm": 0.2579411268234253,
+      "learning_rate": 0.0001780246913580247,
+      "loss": 5.9011,
+      "step": 629
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 0.20957469940185547,
+      "learning_rate": 0.00017777777777777776,
+      "loss": 6.2853,
+      "step": 630
+    },
+    {
+      "epoch": 1.3998890737659457,
+      "grad_norm": 0.19234152138233185,
+      "learning_rate": 0.00017753086419753085,
+      "loss": 6.1349,
+      "step": 631
+    },
+    {
+      "epoch": 1.4021075984470328,
+      "grad_norm": 0.20171993970870972,
+      "learning_rate": 0.00017728395061728392,
+      "loss": 6.1418,
+      "step": 632
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "grad_norm": 0.20194658637046814,
+      "learning_rate": 0.00017703703703703702,
+      "loss": 6.1433,
+      "step": 633
+    },
+    {
+      "epoch": 1.406544647809207,
+      "grad_norm": 0.21205608546733856,
+      "learning_rate": 0.0001767901234567901,
+      "loss": 6.0333,
+      "step": 634
+    },
+    {
+      "epoch": 1.408763172490294,
+      "grad_norm": 0.2226116955280304,
+      "learning_rate": 0.00017654320987654318,
+      "loss": 6.2465,
+      "step": 635
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 0.22059370577335358,
+      "learning_rate": 0.0001762962962962963,
+      "loss": 6.0347,
+      "step": 636
+    },
+    {
+      "epoch": 1.4132002218524682,
+      "grad_norm": 0.2205394059419632,
+      "learning_rate": 0.00017604938271604937,
+      "loss": 6.0345,
+      "step": 637
+    },
+    {
+      "epoch": 1.4154187465335553,
+      "grad_norm": 0.24176302552223206,
+      "learning_rate": 0.00017580246913580247,
+      "loss": 6.4118,
+      "step": 638
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "grad_norm": 0.17885741591453552,
+      "learning_rate": 0.00017555555555555553,
+      "loss": 6.1874,
+      "step": 639
+    },
+    {
+      "epoch": 1.4198557958957294,
+      "grad_norm": 0.2823544442653656,
+      "learning_rate": 0.00017530864197530863,
+      "loss": 6.1418,
+      "step": 640
+    },
+    {
+      "epoch": 1.4220743205768165,
+      "grad_norm": 0.2030986249446869,
+      "learning_rate": 0.00017506172839506172,
+      "loss": 6.3094,
+      "step": 641
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 0.18841448426246643,
+      "learning_rate": 0.0001748148148148148,
+      "loss": 6.0259,
+      "step": 642
+    },
+    {
+      "epoch": 1.4265113699389906,
+      "grad_norm": 0.20291846990585327,
+      "learning_rate": 0.0001745679012345679,
+      "loss": 6.1485,
+      "step": 643
+    },
+    {
+      "epoch": 1.4287298946200777,
+      "grad_norm": 0.20315323770046234,
+      "learning_rate": 0.00017432098765432096,
+      "loss": 6.2558,
+      "step": 644
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "grad_norm": 0.24634063243865967,
+      "learning_rate": 0.00017407407407407408,
+      "loss": 6.3431,
+      "step": 645
+    },
+    {
+      "epoch": 1.4331669439822519,
+      "grad_norm": 0.2357073724269867,
+      "learning_rate": 0.00017382716049382715,
+      "loss": 6.1643,
+      "step": 646
+    },
+    {
+      "epoch": 1.435385468663339,
+      "grad_norm": 0.26205480098724365,
+      "learning_rate": 0.00017358024691358024,
+      "loss": 6.2731,
+      "step": 647
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 0.19826294481754303,
+      "learning_rate": 0.0001733333333333333,
+      "loss": 6.2272,
+      "step": 648
+    },
+    {
+      "epoch": 1.439822518025513,
+      "grad_norm": 0.223244309425354,
+      "learning_rate": 0.0001730864197530864,
+      "loss": 6.3727,
+      "step": 649
+    },
+    {
+      "epoch": 1.4420410427066002,
+      "grad_norm": 0.21920117735862732,
+      "learning_rate": 0.0001728395061728395,
+      "loss": 6.1392,
+      "step": 650
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "grad_norm": 0.2113947570323944,
+      "learning_rate": 0.00017259259259259257,
+      "loss": 6.4457,
+      "step": 651
+    },
+    {
+      "epoch": 1.4464780920687743,
+      "grad_norm": 0.2448064237833023,
+      "learning_rate": 0.00017234567901234566,
+      "loss": 5.9836,
+      "step": 652
+    },
+    {
+      "epoch": 1.4486966167498614,
+      "grad_norm": 0.22628512978553772,
+      "learning_rate": 0.00017209876543209873,
+      "loss": 6.0702,
+      "step": 653
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 0.20075459778308868,
+      "learning_rate": 0.00017185185185185185,
+      "loss": 6.0493,
+      "step": 654
+    },
+    {
+      "epoch": 1.4531336661120355,
+      "grad_norm": 0.2197025567293167,
+      "learning_rate": 0.00017160493827160492,
+      "loss": 6.0176,
+      "step": 655
+    },
+    {
+      "epoch": 1.4553521907931226,
+      "grad_norm": 0.21604877710342407,
+      "learning_rate": 0.00017135802469135802,
+      "loss": 6.0625,
+      "step": 656
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "grad_norm": 0.30104443430900574,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 5.913,
+      "step": 657
+    },
+    {
+      "epoch": 1.4597892401552968,
+      "grad_norm": 0.2126208245754242,
+      "learning_rate": 0.00017086419753086418,
+      "loss": 6.1396,
+      "step": 658
+    },
+    {
+      "epoch": 1.4620077648363838,
+      "grad_norm": 0.2932712137699127,
+      "learning_rate": 0.00017061728395061727,
+      "loss": 6.3671,
+      "step": 659
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 0.2532920241355896,
+      "learning_rate": 0.00017037037037037034,
+      "loss": 5.9704,
+      "step": 660
+    },
+    {
+      "epoch": 1.466444814198558,
+      "grad_norm": 0.2302643209695816,
+      "learning_rate": 0.00017012345679012344,
+      "loss": 6.2932,
+      "step": 661
+    },
+    {
+      "epoch": 1.468663338879645,
+      "grad_norm": 0.2035222202539444,
+      "learning_rate": 0.0001698765432098765,
+      "loss": 5.972,
+      "step": 662
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "grad_norm": 0.2598227858543396,
+      "learning_rate": 0.00016962962962962963,
+      "loss": 6.3461,
+      "step": 663
+    },
+    {
+      "epoch": 1.4731003882418192,
+      "grad_norm": 0.1860261708498001,
+      "learning_rate": 0.00016938271604938272,
+      "loss": 6.0669,
+      "step": 664
+    },
+    {
+      "epoch": 1.4753189129229063,
+      "grad_norm": 0.2792240381240845,
+      "learning_rate": 0.0001691358024691358,
+      "loss": 6.3579,
+      "step": 665
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 0.3398974537849426,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 6.3915,
+      "step": 666
+    },
+    {
+      "epoch": 1.4797559622850804,
+      "grad_norm": 0.24426370859146118,
+      "learning_rate": 0.00016864197530864195,
+      "loss": 6.3454,
+      "step": 667
+    },
+    {
+      "epoch": 1.4819744869661675,
+      "grad_norm": 0.21176092326641083,
+      "learning_rate": 0.00016839506172839505,
+      "loss": 6.1331,
+      "step": 668
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "grad_norm": 0.24312946200370789,
+      "learning_rate": 0.00016814814814814812,
+      "loss": 5.9543,
+      "step": 669
+    },
+    {
+      "epoch": 1.4864115363283417,
+      "grad_norm": 0.18067479133605957,
+      "learning_rate": 0.0001679012345679012,
+      "loss": 6.0353,
+      "step": 670
+    },
+    {
+      "epoch": 1.4886300610094287,
+      "grad_norm": 0.20809943974018097,
+      "learning_rate": 0.00016765432098765433,
+      "loss": 6.1867,
+      "step": 671
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 0.21670183539390564,
+      "learning_rate": 0.0001674074074074074,
+      "loss": 6.1889,
+      "step": 672
+    },
+    {
+      "epoch": 1.493067110371603,
+      "grad_norm": 0.26795223355293274,
+      "learning_rate": 0.0001671604938271605,
+      "loss": 6.187,
+      "step": 673
+    },
+    {
+      "epoch": 1.49528563505269,
+      "grad_norm": 0.4000786244869232,
+      "learning_rate": 0.00016691358024691357,
+      "loss": 6.2487,
+      "step": 674
+    },
+    {
+      "epoch": 1.497504159733777,
+      "grad_norm": 0.22324298322200775,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 6.1627,
+      "step": 675
+    },
+    {
+      "epoch": 1.4997226844148641,
+      "grad_norm": 0.23595760762691498,
+      "learning_rate": 0.00016641975308641973,
+      "loss": 6.3009,
+      "step": 676
+    },
+    {
+      "epoch": 1.5019412090959512,
+      "grad_norm": 0.22958624362945557,
+      "learning_rate": 0.00016617283950617282,
+      "loss": 6.2915,
+      "step": 677
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 0.2237258106470108,
+      "learning_rate": 0.0001659259259259259,
+      "loss": 5.8314,
+      "step": 678
+    },
+    {
+      "epoch": 1.5063782584581253,
+      "grad_norm": 0.27288156747817993,
+      "learning_rate": 0.000165679012345679,
+      "loss": 6.04,
+      "step": 679
+    },
+    {
+      "epoch": 1.5085967831392124,
+      "grad_norm": 0.24428753554821014,
+      "learning_rate": 0.0001654320987654321,
+      "loss": 6.2662,
+      "step": 680
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "grad_norm": 0.25249871611595154,
+      "learning_rate": 0.00016518518518518518,
+      "loss": 6.3509,
+      "step": 681
+    },
+    {
+      "epoch": 1.5130338325013866,
+      "grad_norm": 0.2816295921802521,
+      "learning_rate": 0.00016493827160493827,
+      "loss": 6.0099,
+      "step": 682
+    },
+    {
+      "epoch": 1.5152523571824736,
+      "grad_norm": 0.23814544081687927,
+      "learning_rate": 0.00016469135802469134,
+      "loss": 6.1403,
+      "step": 683
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 0.20159359276294708,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 6.1383,
+      "step": 684
+    },
+    {
+      "epoch": 1.5196894065446478,
+      "grad_norm": 0.23948323726654053,
+      "learning_rate": 0.0001641975308641975,
+      "loss": 6.0528,
+      "step": 685
+    },
+    {
+      "epoch": 1.5219079312257349,
+      "grad_norm": 0.25047188997268677,
+      "learning_rate": 0.0001639506172839506,
+      "loss": 6.3751,
+      "step": 686
+    },
+    {
+      "epoch": 1.524126455906822,
+      "grad_norm": 0.322818398475647,
+      "learning_rate": 0.0001637037037037037,
+      "loss": 5.9922,
+      "step": 687
+    },
+    {
+      "epoch": 1.526344980587909,
+      "grad_norm": 0.26874974370002747,
+      "learning_rate": 0.00016345679012345676,
+      "loss": 6.3372,
+      "step": 688
+    },
+    {
+      "epoch": 1.528563505268996,
+      "grad_norm": 0.21626298129558563,
+      "learning_rate": 0.00016320987654320988,
+      "loss": 6.0653,
+      "step": 689
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 0.22277246415615082,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 6.191,
+      "step": 690
+    },
+    {
+      "epoch": 1.5330005546311702,
+      "grad_norm": 0.2507956624031067,
+      "learning_rate": 0.00016271604938271605,
+      "loss": 6.1485,
+      "step": 691
+    },
+    {
+      "epoch": 1.5352190793122573,
+      "grad_norm": 0.1885564923286438,
+      "learning_rate": 0.00016246913580246912,
+      "loss": 6.308,
+      "step": 692
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "grad_norm": 0.2940356731414795,
+      "learning_rate": 0.0001622222222222222,
+      "loss": 6.3243,
+      "step": 693
+    },
+    {
+      "epoch": 1.5396561286744315,
+      "grad_norm": 0.21090666949748993,
+      "learning_rate": 0.0001619753086419753,
+      "loss": 6.0212,
+      "step": 694
+    },
+    {
+      "epoch": 1.5418746533555185,
+      "grad_norm": 0.20208248496055603,
+      "learning_rate": 0.00016172839506172837,
+      "loss": 5.9366,
+      "step": 695
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 0.21864314377307892,
+      "learning_rate": 0.00016148148148148147,
+      "loss": 6.1852,
+      "step": 696
+    },
+    {
+      "epoch": 1.5463117027176927,
+      "grad_norm": 0.20108667016029358,
+      "learning_rate": 0.00016123456790123454,
+      "loss": 6.0767,
+      "step": 697
+    },
+    {
+      "epoch": 1.5485302273987798,
+      "grad_norm": 0.25402894616127014,
+      "learning_rate": 0.00016098765432098766,
+      "loss": 6.0971,
+      "step": 698
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "grad_norm": 0.2565523684024811,
+      "learning_rate": 0.00016074074074074073,
+      "loss": 5.9639,
+      "step": 699
+    },
+    {
+      "epoch": 1.552967276760954,
+      "grad_norm": 0.23282958567142487,
+      "learning_rate": 0.00016049382716049382,
+      "loss": 6.137,
+      "step": 700
+    },
+    {
+      "epoch": 1.555185801442041,
+      "grad_norm": 0.2591053545475006,
+      "learning_rate": 0.00016024691358024692,
+      "loss": 6.03,
+      "step": 701
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 0.21364083886146545,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 6.2765,
+      "step": 702
+    },
+    {
+      "epoch": 1.5596228508042151,
+      "grad_norm": 0.26426634192466736,
+      "learning_rate": 0.00015975308641975308,
+      "loss": 6.607,
+      "step": 703
+    },
+    {
+      "epoch": 1.5618413754853022,
+      "grad_norm": 0.2369103729724884,
+      "learning_rate": 0.00015950617283950615,
+      "loss": 6.1521,
+      "step": 704
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "grad_norm": 0.3579324185848236,
+      "learning_rate": 0.00015925925925925924,
+      "loss": 5.6616,
+      "step": 705
+    },
+    {
+      "epoch": 1.5662784248474764,
+      "grad_norm": 0.20192186534404755,
+      "learning_rate": 0.0001590123456790123,
+      "loss": 6.1991,
+      "step": 706
+    },
+    {
+      "epoch": 1.5684969495285634,
+      "grad_norm": 0.2008679211139679,
+      "learning_rate": 0.00015876543209876543,
+      "loss": 5.9809,
+      "step": 707
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 0.2082863450050354,
+      "learning_rate": 0.0001585185185185185,
+      "loss": 6.17,
+      "step": 708
+    },
+    {
+      "epoch": 1.5729339988907376,
+      "grad_norm": 0.23278185725212097,
+      "learning_rate": 0.0001582716049382716,
+      "loss": 6.2543,
+      "step": 709
+    },
+    {
+      "epoch": 1.5751525235718247,
+      "grad_norm": 0.22950628399848938,
+      "learning_rate": 0.0001580246913580247,
+      "loss": 6.2213,
+      "step": 710
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "grad_norm": 0.2823933959007263,
+      "learning_rate": 0.00015777777777777776,
+      "loss": 6.3554,
+      "step": 711
+    },
+    {
+      "epoch": 1.5795895729339988,
+      "grad_norm": 0.3217799961566925,
+      "learning_rate": 0.00015753086419753086,
+      "loss": 6.2396,
+      "step": 712
+    },
+    {
+      "epoch": 1.581808097615086,
+      "grad_norm": 0.2303139716386795,
+      "learning_rate": 0.00015728395061728392,
+      "loss": 6.0867,
+      "step": 713
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 0.1930440068244934,
+      "learning_rate": 0.00015703703703703702,
+      "loss": 6.1255,
+      "step": 714
+    },
+    {
+      "epoch": 1.58624514697726,
+      "grad_norm": 0.24806521832942963,
+      "learning_rate": 0.0001567901234567901,
+      "loss": 6.1359,
+      "step": 715
+    },
+    {
+      "epoch": 1.5884636716583471,
+      "grad_norm": 0.24836201965808868,
+      "learning_rate": 0.0001565432098765432,
+      "loss": 6.1833,
+      "step": 716
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "grad_norm": 0.24254190921783447,
+      "learning_rate": 0.0001562962962962963,
+      "loss": 6.1197,
+      "step": 717
+    },
+    {
+      "epoch": 1.5929007210205213,
+      "grad_norm": 0.23480485379695892,
+      "learning_rate": 0.00015604938271604937,
+      "loss": 6.09,
+      "step": 718
+    },
+    {
+      "epoch": 1.5951192457016083,
+      "grad_norm": 0.23228304088115692,
+      "learning_rate": 0.00015580246913580247,
+      "loss": 5.8635,
+      "step": 719
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 0.2540881335735321,
+      "learning_rate": 0.00015555555555555554,
+      "loss": 6.1969,
+      "step": 720
+    },
+    {
+      "epoch": 1.5995562950637825,
+      "grad_norm": 0.30461543798446655,
+      "learning_rate": 0.00015530864197530863,
+      "loss": 6.1072,
+      "step": 721
+    },
+    {
+      "epoch": 1.6017748197448696,
+      "grad_norm": 0.22198040783405304,
+      "learning_rate": 0.0001550617283950617,
+      "loss": 6.1692,
+      "step": 722
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "grad_norm": 0.3056945502758026,
+      "learning_rate": 0.0001548148148148148,
+      "loss": 5.7539,
+      "step": 723
+    },
+    {
+      "epoch": 1.6062118691070437,
+      "grad_norm": 0.2715371549129486,
+      "learning_rate": 0.00015456790123456792,
+      "loss": 6.3546,
+      "step": 724
+    },
+    {
+      "epoch": 1.6084303937881308,
+      "grad_norm": 0.2875449061393738,
+      "learning_rate": 0.00015432098765432098,
+      "loss": 5.8284,
+      "step": 725
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 0.263099730014801,
+      "learning_rate": 0.00015407407407407408,
+      "loss": 6.0792,
+      "step": 726
+    },
+    {
+      "epoch": 1.612867443150305,
+      "grad_norm": 0.20502090454101562,
+      "learning_rate": 0.00015382716049382715,
+      "loss": 5.9146,
+      "step": 727
+    },
+    {
+      "epoch": 1.615085967831392,
+      "grad_norm": 0.23513206839561462,
+      "learning_rate": 0.00015358024691358024,
+      "loss": 6.0989,
+      "step": 728
+    },
+    {
+      "epoch": 1.617304492512479,
+      "grad_norm": 0.2602792978286743,
+      "learning_rate": 0.0001533333333333333,
+      "loss": 6.2945,
+      "step": 729
+    },
+    {
+      "epoch": 1.6195230171935662,
+      "grad_norm": 0.19225040078163147,
+      "learning_rate": 0.0001530864197530864,
+      "loss": 5.937,
+      "step": 730
+    },
+    {
+      "epoch": 1.6217415418746532,
+      "grad_norm": 0.2802371382713318,
+      "learning_rate": 0.00015283950617283947,
+      "loss": 6.1373,
+      "step": 731
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 0.278146892786026,
+      "learning_rate": 0.00015259259259259257,
+      "loss": 5.8718,
+      "step": 732
+    },
+    {
+      "epoch": 1.6261785912368274,
+      "grad_norm": 0.26428931951522827,
+      "learning_rate": 0.0001523456790123457,
+      "loss": 5.9519,
+      "step": 733
+    },
+    {
+      "epoch": 1.6283971159179145,
+      "grad_norm": 0.2766032814979553,
+      "learning_rate": 0.00015209876543209873,
+      "loss": 5.9182,
+      "step": 734
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "grad_norm": 0.24515436589717865,
+      "learning_rate": 0.00015185185185185185,
+      "loss": 6.3677,
+      "step": 735
+    },
+    {
+      "epoch": 1.6328341652800886,
+      "grad_norm": 0.20896974205970764,
+      "learning_rate": 0.00015160493827160492,
+      "loss": 5.9981,
+      "step": 736
+    },
+    {
+      "epoch": 1.6350526899611757,
+      "grad_norm": 0.21279333531856537,
+      "learning_rate": 0.00015135802469135802,
+      "loss": 6.1013,
+      "step": 737
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 0.2345089316368103,
+      "learning_rate": 0.00015111111111111109,
+      "loss": 5.6381,
+      "step": 738
+    },
+    {
+      "epoch": 1.6394897393233498,
+      "grad_norm": 0.22571612894535065,
+      "learning_rate": 0.00015086419753086418,
+      "loss": 6.139,
+      "step": 739
+    },
+    {
+      "epoch": 1.641708264004437,
+      "grad_norm": 0.3228113353252411,
+      "learning_rate": 0.00015061728395061728,
+      "loss": 6.1779,
+      "step": 740
+    },
+    {
+      "epoch": 1.643926788685524,
+      "grad_norm": 0.24217116832733154,
+      "learning_rate": 0.00015037037037037034,
+      "loss": 6.2258,
+      "step": 741
+    },
+    {
+      "epoch": 1.646145313366611,
+      "grad_norm": 0.25396424531936646,
+      "learning_rate": 0.00015012345679012347,
+      "loss": 6.1989,
+      "step": 742
+    },
+    {
+      "epoch": 1.6483638380476981,
+      "grad_norm": 0.21835198998451233,
+      "learning_rate": 0.00014987654320987653,
+      "loss": 6.2196,
+      "step": 743
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 0.18969611823558807,
+      "learning_rate": 0.00014962962962962963,
+      "loss": 6.1037,
+      "step": 744
+    },
+    {
+      "epoch": 1.6528008874098723,
+      "grad_norm": 0.3678043484687805,
+      "learning_rate": 0.0001493827160493827,
+      "loss": 6.0256,
+      "step": 745
+    },
+    {
+      "epoch": 1.6550194120909594,
+      "grad_norm": 0.29937031865119934,
+      "learning_rate": 0.0001491358024691358,
+      "loss": 6.2805,
+      "step": 746
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "grad_norm": 0.22746650874614716,
+      "learning_rate": 0.00014888888888888886,
+      "loss": 5.9826,
+      "step": 747
+    },
+    {
+      "epoch": 1.6594564614531335,
+      "grad_norm": 0.3031845986843109,
+      "learning_rate": 0.00014864197530864196,
+      "loss": 6.2307,
+      "step": 748
+    },
+    {
+      "epoch": 1.6616749861342206,
+      "grad_norm": 0.25561073422431946,
+      "learning_rate": 0.00014839506172839505,
+      "loss": 6.1608,
+      "step": 749
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 0.31713274121284485,
+      "learning_rate": 0.00014814814814814812,
+      "loss": 6.0148,
+      "step": 750
+    },
+    {
+      "epoch": 1.6661120354963947,
+      "grad_norm": 0.23293517529964447,
+      "learning_rate": 0.00014790123456790124,
+      "loss": 5.924,
+      "step": 751
+    },
+    {
+      "epoch": 1.6683305601774818,
+      "grad_norm": 0.23364442586898804,
+      "learning_rate": 0.0001476543209876543,
+      "loss": 5.8867,
+      "step": 752
+    },
+    {
+      "epoch": 1.670549084858569,
+      "grad_norm": 0.26645639538764954,
+      "learning_rate": 0.0001474074074074074,
+      "loss": 5.9458,
+      "step": 753
+    },
+    {
+      "epoch": 1.672767609539656,
+      "grad_norm": 0.2720891833305359,
+      "learning_rate": 0.00014716049382716047,
+      "loss": 6.1426,
+      "step": 754
+    },
+    {
+      "epoch": 1.674986134220743,
+      "grad_norm": 0.26077571511268616,
+      "learning_rate": 0.00014691358024691357,
+      "loss": 6.198,
+      "step": 755
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 0.23528747260570526,
+      "learning_rate": 0.00014666666666666664,
+      "loss": 6.3057,
+      "step": 756
+    },
+    {
+      "epoch": 1.6794231835829172,
+      "grad_norm": 0.23686063289642334,
+      "learning_rate": 0.00014641975308641973,
+      "loss": 6.0669,
+      "step": 757
+    },
+    {
+      "epoch": 1.6816417082640043,
+      "grad_norm": 0.21433116495609283,
+      "learning_rate": 0.00014617283950617283,
+      "loss": 6.068,
+      "step": 758
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "grad_norm": 0.25886186957359314,
+      "learning_rate": 0.00014592592592592592,
+      "loss": 5.6088,
+      "step": 759
+    },
+    {
+      "epoch": 1.6860787576261784,
+      "grad_norm": 0.24864816665649414,
+      "learning_rate": 0.000145679012345679,
+      "loss": 5.9552,
+      "step": 760
+    },
+    {
+      "epoch": 1.6882972823072657,
+      "grad_norm": 0.20896707475185394,
+      "learning_rate": 0.00014543209876543208,
+      "loss": 5.9523,
+      "step": 761
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 0.22849605977535248,
+      "learning_rate": 0.00014518518518518518,
+      "loss": 5.997,
+      "step": 762
+    },
+    {
+      "epoch": 1.6927343316694399,
+      "grad_norm": 0.25133684277534485,
+      "learning_rate": 0.00014493827160493825,
+      "loss": 6.0499,
+      "step": 763
+    },
+    {
+      "epoch": 1.694952856350527,
+      "grad_norm": 0.2523936331272125,
+      "learning_rate": 0.00014469135802469134,
+      "loss": 5.9009,
+      "step": 764
+    },
+    {
+      "epoch": 1.697171381031614,
+      "grad_norm": 0.27618736028671265,
+      "learning_rate": 0.0001444444444444444,
+      "loss": 6.2321,
+      "step": 765
+    },
+    {
+      "epoch": 1.699389905712701,
+      "grad_norm": 0.24548856914043427,
+      "learning_rate": 0.00014419753086419753,
+      "loss": 5.844,
+      "step": 766
+    },
+    {
+      "epoch": 1.7016084303937882,
+      "grad_norm": 0.21411630511283875,
+      "learning_rate": 0.0001439506172839506,
+      "loss": 6.2182,
+      "step": 767
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 0.2824634611606598,
+      "learning_rate": 0.0001437037037037037,
+      "loss": 6.0211,
+      "step": 768
+    },
+    {
+      "epoch": 1.7060454797559623,
+      "grad_norm": 0.23638124763965607,
+      "learning_rate": 0.00014345679012345676,
+      "loss": 5.8827,
+      "step": 769
+    },
+    {
+      "epoch": 1.7082640044370494,
+      "grad_norm": 0.29876473546028137,
+      "learning_rate": 0.00014320987654320986,
+      "loss": 6.0262,
+      "step": 770
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "grad_norm": 0.19986480474472046,
+      "learning_rate": 0.00014296296296296295,
+      "loss": 6.1647,
+      "step": 771
+    },
+    {
+      "epoch": 1.7127010537992235,
+      "grad_norm": 0.22715692222118378,
+      "learning_rate": 0.00014271604938271602,
+      "loss": 5.8082,
+      "step": 772
+    },
+    {
+      "epoch": 1.7149195784803106,
+      "grad_norm": 0.2580621540546417,
+      "learning_rate": 0.00014246913580246912,
+      "loss": 6.1365,
+      "step": 773
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 0.23148922622203827,
+      "learning_rate": 0.0001422222222222222,
+      "loss": 5.9053,
+      "step": 774
+    },
+    {
+      "epoch": 1.7193566278424848,
+      "grad_norm": 0.37005814909935,
+      "learning_rate": 0.0001419753086419753,
+      "loss": 6.0033,
+      "step": 775
+    },
+    {
+      "epoch": 1.7215751525235718,
+      "grad_norm": 0.30081698298454285,
+      "learning_rate": 0.00014172839506172838,
+      "loss": 5.93,
+      "step": 776
+    },
+    {
+      "epoch": 1.723793677204659,
+      "grad_norm": 0.3135722577571869,
+      "learning_rate": 0.00014148148148148147,
+      "loss": 6.0762,
+      "step": 777
+    },
+    {
+      "epoch": 1.726012201885746,
+      "grad_norm": 0.2609674036502838,
+      "learning_rate": 0.00014123456790123454,
+      "loss": 6.063,
+      "step": 778
+    },
+    {
+      "epoch": 1.728230726566833,
+      "grad_norm": 0.22757306694984436,
+      "learning_rate": 0.00014098765432098763,
+      "loss": 6.2819,
+      "step": 779
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 0.31939446926116943,
+      "learning_rate": 0.00014074074074074073,
+      "loss": 6.2785,
+      "step": 780
+    },
+    {
+      "epoch": 1.7326677759290072,
+      "grad_norm": 0.27296581864356995,
+      "learning_rate": 0.00014049382716049382,
+      "loss": 5.7633,
+      "step": 781
+    },
+    {
+      "epoch": 1.7348863006100943,
+      "grad_norm": 0.19938498735427856,
+      "learning_rate": 0.0001402469135802469,
+      "loss": 5.8547,
+      "step": 782
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "grad_norm": 0.22074291110038757,
+      "learning_rate": 0.00014,
+      "loss": 6.0029,
+      "step": 783
+    },
+    {
+      "epoch": 1.7393233499722685,
+      "grad_norm": 0.31793221831321716,
+      "learning_rate": 0.00013975308641975308,
+      "loss": 5.9391,
+      "step": 784
+    },
+    {
+      "epoch": 1.7415418746533555,
+      "grad_norm": 0.2424820512533188,
+      "learning_rate": 0.00013950617283950615,
+      "loss": 6.0436,
+      "step": 785
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 0.2487836480140686,
+      "learning_rate": 0.00013925925925925925,
+      "loss": 5.9543,
+      "step": 786
+    },
+    {
+      "epoch": 1.7459789240155297,
+      "grad_norm": 0.25511640310287476,
+      "learning_rate": 0.00013901234567901231,
+      "loss": 5.7819,
+      "step": 787
+    },
+    {
+      "epoch": 1.7481974486966168,
+      "grad_norm": 0.25465914607048035,
+      "learning_rate": 0.00013876543209876544,
+      "loss": 5.962,
+      "step": 788
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "grad_norm": 0.2349788397550583,
+      "learning_rate": 0.0001385185185185185,
+      "loss": 5.8843,
+      "step": 789
+    },
+    {
+      "epoch": 1.752634498058791,
+      "grad_norm": 0.2511354684829712,
+      "learning_rate": 0.0001382716049382716,
+      "loss": 5.8841,
+      "step": 790
+    },
+    {
+      "epoch": 1.754853022739878,
+      "grad_norm": 0.2431793212890625,
+      "learning_rate": 0.00013802469135802467,
+      "loss": 6.0228,
+      "step": 791
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 0.3821371793746948,
+      "learning_rate": 0.00013777777777777776,
+      "loss": 6.0961,
+      "step": 792
+    },
+    {
+      "epoch": 1.7592900721020521,
+      "grad_norm": 0.2542405426502228,
+      "learning_rate": 0.00013753086419753086,
+      "loss": 5.8653,
+      "step": 793
+    },
+    {
+      "epoch": 1.7615085967831392,
+      "grad_norm": 0.246854767203331,
+      "learning_rate": 0.00013728395061728393,
+      "loss": 5.7234,
+      "step": 794
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "grad_norm": 0.25578078627586365,
+      "learning_rate": 0.00013703703703703702,
+      "loss": 5.8046,
+      "step": 795
+    },
+    {
+      "epoch": 1.7659456461453134,
+      "grad_norm": 0.31212514638900757,
+      "learning_rate": 0.00013679012345679012,
+      "loss": 6.2029,
+      "step": 796
+    },
+    {
+      "epoch": 1.7681641708264004,
+      "grad_norm": 0.2282676249742508,
+      "learning_rate": 0.0001365432098765432,
+      "loss": 6.0061,
+      "step": 797
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 0.25217947363853455,
+      "learning_rate": 0.00013629629629629628,
+      "loss": 6.0505,
+      "step": 798
+    },
+    {
+      "epoch": 1.7726012201885746,
+      "grad_norm": 0.23790553212165833,
+      "learning_rate": 0.00013604938271604937,
+      "loss": 5.7769,
+      "step": 799
+    },
+    {
+      "epoch": 1.7748197448696617,
+      "grad_norm": 0.2617262899875641,
+      "learning_rate": 0.00013580246913580244,
+      "loss": 5.7366,
+      "step": 800
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "grad_norm": 0.2397618293762207,
+      "learning_rate": 0.00013555555555555554,
+      "loss": 6.3369,
+      "step": 801
+    },
+    {
+      "epoch": 1.7792567942318358,
+      "grad_norm": 0.24851787090301514,
+      "learning_rate": 0.00013530864197530863,
+      "loss": 6.072,
+      "step": 802
+    },
+    {
+      "epoch": 1.7814753189129229,
+      "grad_norm": 0.2634022533893585,
+      "learning_rate": 0.00013506172839506173,
+      "loss": 5.9945,
+      "step": 803
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 0.4000261127948761,
+      "learning_rate": 0.0001348148148148148,
+      "loss": 5.8231,
+      "step": 804
+    },
+    {
+      "epoch": 1.785912368275097,
+      "grad_norm": 0.262912392616272,
+      "learning_rate": 0.0001345679012345679,
+      "loss": 6.2456,
+      "step": 805
+    },
+    {
+      "epoch": 1.788130892956184,
+      "grad_norm": 0.2732096016407013,
+      "learning_rate": 0.00013432098765432099,
+      "loss": 5.9438,
+      "step": 806
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "grad_norm": 0.30278339982032776,
+      "learning_rate": 0.00013407407407407405,
+      "loss": 5.7209,
+      "step": 807
+    },
+    {
+      "epoch": 1.7925679423183583,
+      "grad_norm": 0.2247074842453003,
+      "learning_rate": 0.00013382716049382715,
+      "loss": 5.9849,
+      "step": 808
+    },
+    {
+      "epoch": 1.7947864669994453,
+      "grad_norm": 0.23934341967105865,
+      "learning_rate": 0.00013358024691358022,
+      "loss": 5.966,
+      "step": 809
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 0.3007916212081909,
+      "learning_rate": 0.0001333333333333333,
+      "loss": 6.1602,
+      "step": 810
+    },
+    {
+      "epoch": 1.7992235163616195,
+      "grad_norm": 0.23580294847488403,
+      "learning_rate": 0.0001330864197530864,
+      "loss": 6.1169,
+      "step": 811
+    },
+    {
+      "epoch": 1.8014420410427066,
+      "grad_norm": 0.24809536337852478,
+      "learning_rate": 0.0001328395061728395,
+      "loss": 5.825,
+      "step": 812
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "grad_norm": 0.3738366663455963,
+      "learning_rate": 0.00013259259259259257,
+      "loss": 6.1161,
+      "step": 813
+    },
+    {
+      "epoch": 1.8058790904048807,
+      "grad_norm": 0.2683359980583191,
+      "learning_rate": 0.00013234567901234567,
+      "loss": 5.849,
+      "step": 814
+    },
+    {
+      "epoch": 1.8080976150859678,
+      "grad_norm": 0.3785717785358429,
+      "learning_rate": 0.00013209876543209876,
+      "loss": 5.7816,
+      "step": 815
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 0.2584047317504883,
+      "learning_rate": 0.00013185185185185183,
+      "loss": 6.157,
+      "step": 816
+    },
+    {
+      "epoch": 1.8125346644481422,
+      "grad_norm": 0.3109454810619354,
+      "learning_rate": 0.00013160493827160492,
+      "loss": 6.0591,
+      "step": 817
+    },
+    {
+      "epoch": 1.8147531891292292,
+      "grad_norm": 0.22494763135910034,
+      "learning_rate": 0.00013135802469135802,
+      "loss": 5.817,
+      "step": 818
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "grad_norm": 0.3236433267593384,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 5.7629,
+      "step": 819
+    },
+    {
+      "epoch": 1.8191902384914034,
+      "grad_norm": 0.283632755279541,
+      "learning_rate": 0.00013086419753086418,
+      "loss": 5.918,
+      "step": 820
+    },
+    {
+      "epoch": 1.8214087631724905,
+      "grad_norm": 0.23825879395008087,
+      "learning_rate": 0.00013061728395061728,
+      "loss": 5.9466,
+      "step": 821
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 0.23714566230773926,
+      "learning_rate": 0.00013037037037037035,
+      "loss": 6.0027,
+      "step": 822
+    },
+    {
+      "epoch": 1.8258458125346646,
+      "grad_norm": 0.2656237483024597,
+      "learning_rate": 0.00013012345679012344,
+      "loss": 5.7513,
+      "step": 823
+    },
+    {
+      "epoch": 1.8280643372157517,
+      "grad_norm": 0.23909664154052734,
+      "learning_rate": 0.00012987654320987654,
+      "loss": 5.724,
+      "step": 824
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "grad_norm": 0.2794102728366852,
+      "learning_rate": 0.0001296296296296296,
+      "loss": 6.1156,
+      "step": 825
+    },
+    {
+      "epoch": 1.8325013865779258,
+      "grad_norm": 0.31704336404800415,
+      "learning_rate": 0.0001293827160493827,
+      "loss": 5.7358,
+      "step": 826
+    },
+    {
+      "epoch": 1.834719911259013,
+      "grad_norm": 0.26242542266845703,
+      "learning_rate": 0.0001291358024691358,
+      "loss": 6.3808,
+      "step": 827
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 0.29361361265182495,
+      "learning_rate": 0.0001288888888888889,
+      "loss": 5.7511,
+      "step": 828
+    },
+    {
+      "epoch": 1.839156960621187,
+      "grad_norm": 0.3028067350387573,
+      "learning_rate": 0.00012864197530864196,
+      "loss": 5.8837,
+      "step": 829
+    },
+    {
+      "epoch": 1.8413754853022741,
+      "grad_norm": 0.2659502327442169,
+      "learning_rate": 0.00012839506172839505,
+      "loss": 5.7957,
+      "step": 830
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "grad_norm": 0.28728392720222473,
+      "learning_rate": 0.00012814814814814812,
+      "loss": 6.0276,
+      "step": 831
+    },
+    {
+      "epoch": 1.8458125346644483,
+      "grad_norm": 0.29493916034698486,
+      "learning_rate": 0.00012790123456790122,
+      "loss": 6.1133,
+      "step": 832
+    },
+    {
+      "epoch": 1.8480310593455354,
+      "grad_norm": 0.31482744216918945,
+      "learning_rate": 0.0001276543209876543,
+      "loss": 5.8787,
+      "step": 833
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 0.2839886546134949,
+      "learning_rate": 0.0001274074074074074,
+      "loss": 6.0281,
+      "step": 834
+    },
+    {
+      "epoch": 1.8524681087077095,
+      "grad_norm": 0.23971126973628998,
+      "learning_rate": 0.00012716049382716047,
+      "loss": 5.7007,
+      "step": 835
+    },
+    {
+      "epoch": 1.8546866333887966,
+      "grad_norm": 0.23984086513519287,
+      "learning_rate": 0.00012691358024691357,
+      "loss": 6.225,
+      "step": 836
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "grad_norm": 0.2594614028930664,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 5.9483,
+      "step": 837
+    },
+    {
+      "epoch": 1.8591236827509707,
+      "grad_norm": 0.264100044965744,
+      "learning_rate": 0.00012641975308641973,
+      "loss": 5.8773,
+      "step": 838
+    },
+    {
+      "epoch": 1.8613422074320578,
+      "grad_norm": 0.28265729546546936,
+      "learning_rate": 0.00012617283950617283,
+      "loss": 5.7805,
+      "step": 839
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 0.23254895210266113,
+      "learning_rate": 0.0001259259259259259,
+      "loss": 5.8838,
+      "step": 840
+    },
+    {
+      "epoch": 1.865779256794232,
+      "grad_norm": 0.32164058089256287,
+      "learning_rate": 0.00012567901234567902,
+      "loss": 6.0121,
+      "step": 841
+    },
+    {
+      "epoch": 1.867997781475319,
+      "grad_norm": 0.4063686430454254,
+      "learning_rate": 0.00012543209876543209,
+      "loss": 5.9025,
+      "step": 842
+    },
+    {
+      "epoch": 1.870216306156406,
+      "grad_norm": 0.27436476945877075,
+      "learning_rate": 0.00012518518518518518,
+      "loss": 5.8824,
+      "step": 843
+    },
+    {
+      "epoch": 1.8724348308374932,
+      "grad_norm": 0.328648179769516,
+      "learning_rate": 0.00012493827160493825,
+      "loss": 6.0596,
+      "step": 844
+    },
+    {
+      "epoch": 1.8746533555185803,
+      "grad_norm": 0.27509620785713196,
+      "learning_rate": 0.00012469135802469134,
+      "loss": 6.0802,
+      "step": 845
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 0.22732427716255188,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 5.8244,
+      "step": 846
+    },
+    {
+      "epoch": 1.8790904048807544,
+      "grad_norm": 0.2846013605594635,
+      "learning_rate": 0.0001241975308641975,
+      "loss": 5.9444,
+      "step": 847
+    },
+    {
+      "epoch": 1.8813089295618415,
+      "grad_norm": 0.29825639724731445,
+      "learning_rate": 0.0001239506172839506,
+      "loss": 5.6162,
+      "step": 848
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "grad_norm": 0.3013034164905548,
+      "learning_rate": 0.0001237037037037037,
+      "loss": 6.0064,
+      "step": 849
+    },
+    {
+      "epoch": 1.8857459789240156,
+      "grad_norm": 0.2674424350261688,
+      "learning_rate": 0.0001234567901234568,
+      "loss": 6.1452,
+      "step": 850
+    },
+    {
+      "epoch": 1.8879645036051027,
+      "grad_norm": 0.24443665146827698,
+      "learning_rate": 0.00012320987654320986,
+      "loss": 6.1815,
+      "step": 851
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 0.2643965184688568,
+      "learning_rate": 0.00012296296296296296,
+      "loss": 5.747,
+      "step": 852
+    },
+    {
+      "epoch": 1.8924015529672769,
+      "grad_norm": 0.21381598711013794,
+      "learning_rate": 0.00012271604938271602,
+      "loss": 5.9033,
+      "step": 853
+    },
+    {
+      "epoch": 1.894620077648364,
+      "grad_norm": 0.288196325302124,
+      "learning_rate": 0.00012246913580246912,
+      "loss": 6.0152,
+      "step": 854
+    },
+    {
+      "epoch": 1.896838602329451,
+      "grad_norm": 0.3348243236541748,
+      "learning_rate": 0.00012222222222222221,
+      "loss": 6.1525,
+      "step": 855
+    },
+    {
+      "epoch": 1.899057127010538,
+      "grad_norm": 0.22998231649398804,
+      "learning_rate": 0.00012197530864197531,
+      "loss": 5.9222,
+      "step": 856
+    },
+    {
+      "epoch": 1.9012756516916252,
+      "grad_norm": 0.38634994626045227,
+      "learning_rate": 0.00012172839506172839,
+      "loss": 5.6896,
+      "step": 857
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 0.25322917103767395,
+      "learning_rate": 0.00012148148148148147,
+      "loss": 5.9391,
+      "step": 858
+    },
+    {
+      "epoch": 1.9057127010537993,
+      "grad_norm": 0.40103426575660706,
+      "learning_rate": 0.00012123456790123455,
+      "loss": 6.0248,
+      "step": 859
+    },
+    {
+      "epoch": 1.9079312257348864,
+      "grad_norm": 0.2933688163757324,
+      "learning_rate": 0.00012098765432098764,
+      "loss": 6.0186,
+      "step": 860
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "grad_norm": 0.3911920189857483,
+      "learning_rate": 0.00012074074074074073,
+      "loss": 6.2543,
+      "step": 861
+    },
+    {
+      "epoch": 1.9123682750970605,
+      "grad_norm": 0.28124240040779114,
+      "learning_rate": 0.00012049382716049381,
+      "loss": 5.5445,
+      "step": 862
+    },
+    {
+      "epoch": 1.9145867997781476,
+      "grad_norm": 0.2626238167285919,
+      "learning_rate": 0.0001202469135802469,
+      "loss": 5.6441,
+      "step": 863
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 0.2588793635368347,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 5.663,
+      "step": 864
+    },
+    {
+      "epoch": 1.9190238491403218,
+      "grad_norm": 0.25592151284217834,
+      "learning_rate": 0.00011975308641975308,
+      "loss": 6.0373,
+      "step": 865
+    },
+    {
+      "epoch": 1.9212423738214088,
+      "grad_norm": 0.3990209698677063,
+      "learning_rate": 0.00011950617283950617,
+      "loss": 5.9661,
+      "step": 866
+    },
+    {
+      "epoch": 1.923460898502496,
+      "grad_norm": 0.32539787888526917,
+      "learning_rate": 0.00011925925925925925,
+      "loss": 5.6667,
+      "step": 867
+    },
+    {
+      "epoch": 1.925679423183583,
+      "grad_norm": 0.25194433331489563,
+      "learning_rate": 0.00011901234567901233,
+      "loss": 6.0549,
+      "step": 868
+    },
+    {
+      "epoch": 1.92789794786467,
+      "grad_norm": 0.2916874289512634,
+      "learning_rate": 0.00011876543209876541,
+      "loss": 5.9153,
+      "step": 869
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 0.2982962131500244,
+      "learning_rate": 0.0001185185185185185,
+      "loss": 6.0335,
+      "step": 870
+    },
+    {
+      "epoch": 1.9323349972268442,
+      "grad_norm": 0.21497942507266998,
+      "learning_rate": 0.0001182716049382716,
+      "loss": 5.8993,
+      "step": 871
+    },
+    {
+      "epoch": 1.9345535219079313,
+      "grad_norm": 0.24648873507976532,
+      "learning_rate": 0.00011802469135802468,
+      "loss": 5.6652,
+      "step": 872
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "grad_norm": 0.25967392325401306,
+      "learning_rate": 0.00011777777777777776,
+      "loss": 5.9299,
+      "step": 873
+    },
+    {
+      "epoch": 1.9389905712701054,
+      "grad_norm": 0.21153759956359863,
+      "learning_rate": 0.00011753086419753086,
+      "loss": 5.7651,
+      "step": 874
+    },
+    {
+      "epoch": 1.9412090959511925,
+      "grad_norm": 0.3023030459880829,
+      "learning_rate": 0.00011728395061728394,
+      "loss": 6.2341,
+      "step": 875
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 0.3058208227157593,
+      "learning_rate": 0.00011703703703703702,
+      "loss": 5.7102,
+      "step": 876
+    },
+    {
+      "epoch": 1.9456461453133667,
+      "grad_norm": 0.2829059958457947,
+      "learning_rate": 0.0001167901234567901,
+      "loss": 5.6879,
+      "step": 877
+    },
+    {
+      "epoch": 1.9478646699944537,
+      "grad_norm": 0.29789304733276367,
+      "learning_rate": 0.00011654320987654319,
+      "loss": 5.6981,
+      "step": 878
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "grad_norm": 0.40445083379745483,
+      "learning_rate": 0.0001162962962962963,
+      "loss": 6.0185,
+      "step": 879
+    },
+    {
+      "epoch": 1.9523017193566279,
+      "grad_norm": 0.31508758664131165,
+      "learning_rate": 0.00011604938271604938,
+      "loss": 5.865,
+      "step": 880
+    },
+    {
+      "epoch": 1.954520244037715,
+      "grad_norm": 0.27682992815971375,
+      "learning_rate": 0.00011580246913580246,
+      "loss": 5.8398,
+      "step": 881
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 0.25739553570747375,
+      "learning_rate": 0.00011555555555555554,
+      "loss": 6.0328,
+      "step": 882
+    },
+    {
+      "epoch": 1.958957293399889,
+      "grad_norm": 0.23963744938373566,
+      "learning_rate": 0.00011530864197530863,
+      "loss": 6.0462,
+      "step": 883
+    },
+    {
+      "epoch": 1.9611758180809762,
+      "grad_norm": 0.28837308287620544,
+      "learning_rate": 0.00011506172839506172,
+      "loss": 5.7485,
+      "step": 884
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "grad_norm": 0.39942774176597595,
+      "learning_rate": 0.0001148148148148148,
+      "loss": 5.5711,
+      "step": 885
+    },
+    {
+      "epoch": 1.9656128674431503,
+      "grad_norm": 0.3133490979671478,
+      "learning_rate": 0.00011456790123456789,
+      "loss": 5.7571,
+      "step": 886
+    },
+    {
+      "epoch": 1.9678313921242374,
+      "grad_norm": 0.39707231521606445,
+      "learning_rate": 0.00011432098765432099,
+      "loss": 6.2506,
+      "step": 887
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 0.31360530853271484,
+      "learning_rate": 0.00011407407407407407,
+      "loss": 5.9531,
+      "step": 888
+    },
+    {
+      "epoch": 1.9722684414864116,
+      "grad_norm": 0.37130114436149597,
+      "learning_rate": 0.00011382716049382715,
+      "loss": 5.965,
+      "step": 889
+    },
+    {
+      "epoch": 1.9744869661674986,
+      "grad_norm": 0.25294992327690125,
+      "learning_rate": 0.00011358024691358023,
+      "loss": 5.6356,
+      "step": 890
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "grad_norm": 0.35280200839042664,
+      "learning_rate": 0.00011333333333333331,
+      "loss": 5.89,
+      "step": 891
+    },
+    {
+      "epoch": 1.9789240155296728,
+      "grad_norm": 0.23285126686096191,
+      "learning_rate": 0.00011308641975308641,
+      "loss": 5.9129,
+      "step": 892
+    },
+    {
+      "epoch": 1.9811425402107599,
+      "grad_norm": 0.24178774654865265,
+      "learning_rate": 0.00011283950617283949,
+      "loss": 5.8143,
+      "step": 893
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 0.28542226552963257,
+      "learning_rate": 0.00011259259259259259,
+      "loss": 5.9507,
+      "step": 894
+    },
+    {
+      "epoch": 1.985579589572934,
+      "grad_norm": 0.2496650516986847,
+      "learning_rate": 0.00011234567901234567,
+      "loss": 6.0498,
+      "step": 895
+    },
+    {
+      "epoch": 1.987798114254021,
+      "grad_norm": 0.2457389384508133,
+      "learning_rate": 0.00011209876543209876,
+      "loss": 5.7837,
+      "step": 896
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "grad_norm": 0.2634519636631012,
+      "learning_rate": 0.00011185185185185184,
+      "loss": 5.66,
+      "step": 897
+    },
+    {
+      "epoch": 1.9922351636161952,
+      "grad_norm": 0.275718629360199,
+      "learning_rate": 0.00011160493827160493,
+      "loss": 5.4808,
+      "step": 898
+    },
+    {
+      "epoch": 1.9944536882972823,
+      "grad_norm": 0.264754980802536,
+      "learning_rate": 0.00011135802469135801,
+      "loss": 5.7911,
+      "step": 899
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 0.2596605122089386,
+      "learning_rate": 0.00011111111111111109,
+      "loss": 5.8167,
+      "step": 900
+    },
+    {
+      "epoch": 1.9988907376594565,
+      "grad_norm": 0.31804803013801575,
+      "learning_rate": 0.0001108641975308642,
+      "loss": 5.7689,
+      "step": 901
+    },
+    {
+      "epoch": 2.0011092623405435,
+      "grad_norm": 0.34040942788124084,
+      "learning_rate": 0.00011061728395061728,
+      "loss": 5.7754,
+      "step": 902
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "grad_norm": 0.271698534488678,
+      "learning_rate": 0.00011037037037037036,
+      "loss": 5.6283,
+      "step": 903
+    },
+    {
+      "epoch": 2.0055463117027177,
+      "grad_norm": 0.3146517276763916,
+      "learning_rate": 0.00011012345679012344,
+      "loss": 5.729,
+      "step": 904
+    },
+    {
+      "epoch": 2.0077648363838048,
+      "grad_norm": 0.27460524439811707,
+      "learning_rate": 0.00010987654320987654,
+      "loss": 6.0087,
+      "step": 905
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 0.4347095787525177,
+      "learning_rate": 0.00010962962962962962,
+      "loss": 5.8461,
+      "step": 906
+    },
+    {
+      "epoch": 2.012201885745979,
+      "grad_norm": 0.38177022337913513,
+      "learning_rate": 0.0001093827160493827,
+      "loss": 5.7319,
+      "step": 907
+    },
+    {
+      "epoch": 2.014420410427066,
+      "grad_norm": 0.31827589869499207,
+      "learning_rate": 0.00010913580246913578,
+      "loss": 5.8822,
+      "step": 908
+    },
+    {
+      "epoch": 2.016638935108153,
+      "grad_norm": 0.32661041617393494,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 5.8484,
+      "step": 909
+    },
+    {
+      "epoch": 2.01885745978924,
+      "grad_norm": 0.3018447160720825,
+      "learning_rate": 0.00010864197530864197,
+      "loss": 5.8465,
+      "step": 910
+    },
+    {
+      "epoch": 2.021075984470327,
+      "grad_norm": 0.2692470848560333,
+      "learning_rate": 0.00010839506172839505,
+      "loss": 5.8304,
+      "step": 911
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 0.3425772786140442,
+      "learning_rate": 0.00010814814814814814,
+      "loss": 5.5501,
+      "step": 912
+    },
+    {
+      "epoch": 2.0255130338325014,
+      "grad_norm": 0.25883182883262634,
+      "learning_rate": 0.00010790123456790122,
+      "loss": 5.6723,
+      "step": 913
+    },
+    {
+      "epoch": 2.0277315585135884,
+      "grad_norm": 0.24887698888778687,
+      "learning_rate": 0.00010765432098765431,
+      "loss": 5.9824,
+      "step": 914
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "grad_norm": 0.3898041248321533,
+      "learning_rate": 0.0001074074074074074,
+      "loss": 5.7357,
+      "step": 915
+    },
+    {
+      "epoch": 2.0321686078757626,
+      "grad_norm": 0.272840291261673,
+      "learning_rate": 0.00010716049382716048,
+      "loss": 6.0007,
+      "step": 916
+    },
+    {
+      "epoch": 2.0343871325568497,
+      "grad_norm": 0.3728686571121216,
+      "learning_rate": 0.00010691358024691357,
+      "loss": 5.6811,
+      "step": 917
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 0.2931363582611084,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 5.5942,
+      "step": 918
+    },
+    {
+      "epoch": 2.038824181919024,
+      "grad_norm": 0.32677826285362244,
+      "learning_rate": 0.00010641975308641975,
+      "loss": 5.8916,
+      "step": 919
+    },
+    {
+      "epoch": 2.041042706600111,
+      "grad_norm": 0.36013248562812805,
+      "learning_rate": 0.00010617283950617283,
+      "loss": 5.8869,
+      "step": 920
+    },
+    {
+      "epoch": 2.043261231281198,
+      "grad_norm": 0.22664707899093628,
+      "learning_rate": 0.00010592592592592591,
+      "loss": 5.8865,
+      "step": 921
+    },
+    {
+      "epoch": 2.045479755962285,
+      "grad_norm": 0.3170510232448578,
+      "learning_rate": 0.00010567901234567899,
+      "loss": 5.8611,
+      "step": 922
+    },
+    {
+      "epoch": 2.047698280643372,
+      "grad_norm": 0.34601473808288574,
+      "learning_rate": 0.00010543209876543209,
+      "loss": 5.9039,
+      "step": 923
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 0.3209538459777832,
+      "learning_rate": 0.00010518518518518518,
+      "loss": 5.6207,
+      "step": 924
+    },
+    {
+      "epoch": 2.0521353300055463,
+      "grad_norm": 0.2829282581806183,
+      "learning_rate": 0.00010493827160493826,
+      "loss": 5.8265,
+      "step": 925
+    },
+    {
+      "epoch": 2.0543538546866333,
+      "grad_norm": 0.2667284607887268,
+      "learning_rate": 0.00010469135802469135,
+      "loss": 5.907,
+      "step": 926
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "grad_norm": 0.315708190202713,
+      "learning_rate": 0.00010444444444444444,
+      "loss": 5.9075,
+      "step": 927
+    },
+    {
+      "epoch": 2.0587909040488075,
+      "grad_norm": 0.2909288704395294,
+      "learning_rate": 0.00010419753086419752,
+      "loss": 5.5482,
+      "step": 928
+    },
+    {
+      "epoch": 2.0610094287298946,
+      "grad_norm": 0.28952842950820923,
+      "learning_rate": 0.0001039506172839506,
+      "loss": 5.8228,
+      "step": 929
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 0.2282843291759491,
+      "learning_rate": 0.00010370370370370369,
+      "loss": 5.7522,
+      "step": 930
+    },
+    {
+      "epoch": 2.0654464780920687,
+      "grad_norm": 0.3299042880535126,
+      "learning_rate": 0.00010345679012345677,
+      "loss": 5.9082,
+      "step": 931
+    },
+    {
+      "epoch": 2.067665002773156,
+      "grad_norm": 0.27939003705978394,
+      "learning_rate": 0.00010320987654320988,
+      "loss": 5.4766,
+      "step": 932
+    },
+    {
+      "epoch": 2.069883527454243,
+      "grad_norm": 0.2615014612674713,
+      "learning_rate": 0.00010296296296296296,
+      "loss": 5.7367,
+      "step": 933
+    },
+    {
+      "epoch": 2.07210205213533,
+      "grad_norm": 0.2624553442001343,
+      "learning_rate": 0.00010271604938271604,
+      "loss": 6.0179,
+      "step": 934
+    },
+    {
+      "epoch": 2.074320576816417,
+      "grad_norm": 0.31764474511146545,
+      "learning_rate": 0.00010246913580246912,
+      "loss": 5.614,
+      "step": 935
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 0.31088414788246155,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 5.8404,
+      "step": 936
+    },
+    {
+      "epoch": 2.078757626178591,
+      "grad_norm": 0.24831072986125946,
+      "learning_rate": 0.0001019753086419753,
+      "loss": 5.5678,
+      "step": 937
+    },
+    {
+      "epoch": 2.0809761508596782,
+      "grad_norm": 0.2926657199859619,
+      "learning_rate": 0.00010172839506172838,
+      "loss": 5.7933,
+      "step": 938
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "grad_norm": 0.5182262063026428,
+      "learning_rate": 0.00010148148148148147,
+      "loss": 5.5909,
+      "step": 939
+    },
+    {
+      "epoch": 2.0854132002218524,
+      "grad_norm": 0.31408295035362244,
+      "learning_rate": 0.00010123456790123457,
+      "loss": 5.8363,
+      "step": 940
+    },
+    {
+      "epoch": 2.0876317249029395,
+      "grad_norm": 0.3134421110153198,
+      "learning_rate": 0.00010098765432098765,
+      "loss": 5.685,
+      "step": 941
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 0.2881038784980774,
+      "learning_rate": 0.00010074074074074073,
+      "loss": 5.7997,
+      "step": 942
+    },
+    {
+      "epoch": 2.0920687742651136,
+      "grad_norm": 0.3539988100528717,
+      "learning_rate": 0.00010049382716049381,
+      "loss": 5.4881,
+      "step": 943
+    },
+    {
+      "epoch": 2.0942872989462007,
+      "grad_norm": 0.32243612408638,
+      "learning_rate": 0.0001002469135802469,
+      "loss": 5.7568,
+      "step": 944
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "grad_norm": 0.31974735856056213,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 5.9,
+      "step": 945
+    },
+    {
+      "epoch": 2.098724348308375,
+      "grad_norm": 0.27942386269569397,
+      "learning_rate": 9.975308641975307e-05,
+      "loss": 5.7178,
+      "step": 946
+    },
+    {
+      "epoch": 2.100942872989462,
+      "grad_norm": 0.2576830983161926,
+      "learning_rate": 9.950617283950617e-05,
+      "loss": 5.6767,
+      "step": 947
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 0.23114994168281555,
+      "learning_rate": 9.925925925925925e-05,
+      "loss": 5.6844,
+      "step": 948
+    },
+    {
+      "epoch": 2.105379922351636,
+      "grad_norm": 0.2306007295846939,
+      "learning_rate": 9.901234567901234e-05,
+      "loss": 5.853,
+      "step": 949
+    },
+    {
+      "epoch": 2.107598447032723,
+      "grad_norm": 0.26609763503074646,
+      "learning_rate": 9.876543209876543e-05,
+      "loss": 5.6667,
+      "step": 950
+    },
+    {
+      "epoch": 2.10981697171381,
+      "grad_norm": 0.2440844029188156,
+      "learning_rate": 9.851851851851851e-05,
+      "loss": 5.7179,
+      "step": 951
+    },
+    {
+      "epoch": 2.1120354963948973,
+      "grad_norm": 0.23956617712974548,
+      "learning_rate": 9.827160493827159e-05,
+      "loss": 5.8038,
+      "step": 952
+    },
+    {
+      "epoch": 2.1142540210759844,
+      "grad_norm": 0.23151154816150665,
+      "learning_rate": 9.802469135802467e-05,
+      "loss": 5.7667,
+      "step": 953
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 0.26381441950798035,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 5.8737,
+      "step": 954
+    },
+    {
+      "epoch": 2.1186910704381585,
+      "grad_norm": 0.29417291283607483,
+      "learning_rate": 9.753086419753086e-05,
+      "loss": 5.9356,
+      "step": 955
+    },
+    {
+      "epoch": 2.1209095951192456,
+      "grad_norm": 0.29348111152648926,
+      "learning_rate": 9.728395061728394e-05,
+      "loss": 6.0369,
+      "step": 956
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "grad_norm": 0.3045520484447479,
+      "learning_rate": 9.703703703703702e-05,
+      "loss": 6.2882,
+      "step": 957
+    },
+    {
+      "epoch": 2.1253466444814197,
+      "grad_norm": 0.27828630805015564,
+      "learning_rate": 9.679012345679012e-05,
+      "loss": 5.8358,
+      "step": 958
+    },
+    {
+      "epoch": 2.127565169162507,
+      "grad_norm": 0.2621696889400482,
+      "learning_rate": 9.65432098765432e-05,
+      "loss": 5.8256,
+      "step": 959
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 0.3595537841320038,
+      "learning_rate": 9.629629629629628e-05,
+      "loss": 6.0909,
+      "step": 960
+    },
+    {
+      "epoch": 2.132002218524681,
+      "grad_norm": 0.3414286971092224,
+      "learning_rate": 9.604938271604936e-05,
+      "loss": 5.6904,
+      "step": 961
+    },
+    {
+      "epoch": 2.134220743205768,
+      "grad_norm": 0.2526043951511383,
+      "learning_rate": 9.580246913580247e-05,
+      "loss": 5.7732,
+      "step": 962
+    },
+    {
+      "epoch": 2.136439267886855,
+      "grad_norm": 0.39560428261756897,
+      "learning_rate": 9.555555555555555e-05,
+      "loss": 5.9921,
+      "step": 963
+    },
+    {
+      "epoch": 2.138657792567942,
+      "grad_norm": 0.2710109353065491,
+      "learning_rate": 9.530864197530864e-05,
+      "loss": 6.0475,
+      "step": 964
+    },
+    {
+      "epoch": 2.1408763172490293,
+      "grad_norm": 0.36910468339920044,
+      "learning_rate": 9.506172839506172e-05,
+      "loss": 5.9756,
+      "step": 965
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 0.3220480978488922,
+      "learning_rate": 9.48148148148148e-05,
+      "loss": 5.816,
+      "step": 966
+    },
+    {
+      "epoch": 2.1453133666112034,
+      "grad_norm": 0.31590473651885986,
+      "learning_rate": 9.45679012345679e-05,
+      "loss": 5.7443,
+      "step": 967
+    },
+    {
+      "epoch": 2.1475318912922905,
+      "grad_norm": 0.2871208190917969,
+      "learning_rate": 9.432098765432098e-05,
+      "loss": 6.0691,
+      "step": 968
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "grad_norm": 0.2640654146671295,
+      "learning_rate": 9.407407407407407e-05,
+      "loss": 5.7834,
+      "step": 969
+    },
+    {
+      "epoch": 2.1519689406544646,
+      "grad_norm": 0.31639233231544495,
+      "learning_rate": 9.382716049382715e-05,
+      "loss": 5.7328,
+      "step": 970
+    },
+    {
+      "epoch": 2.1541874653355517,
+      "grad_norm": 0.44726821780204773,
+      "learning_rate": 9.358024691358025e-05,
+      "loss": 5.7136,
+      "step": 971
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 0.28712353110313416,
+      "learning_rate": 9.333333333333333e-05,
+      "loss": 5.5569,
+      "step": 972
+    },
+    {
+      "epoch": 2.158624514697726,
+      "grad_norm": 0.3767266571521759,
+      "learning_rate": 9.308641975308641e-05,
+      "loss": 5.8708,
+      "step": 973
+    },
+    {
+      "epoch": 2.160843039378813,
+      "grad_norm": 0.28265538811683655,
+      "learning_rate": 9.283950617283949e-05,
+      "loss": 5.7513,
+      "step": 974
+    },
+    {
+      "epoch": 2.1630615640599,
+      "grad_norm": 0.24970827996730804,
+      "learning_rate": 9.259259259259257e-05,
+      "loss": 5.812,
+      "step": 975
+    },
+    {
+      "epoch": 2.165280088740987,
+      "grad_norm": 0.27550989389419556,
+      "learning_rate": 9.234567901234567e-05,
+      "loss": 5.8681,
+      "step": 976
+    },
+    {
+      "epoch": 2.167498613422074,
+      "grad_norm": 0.2920374572277069,
+      "learning_rate": 9.209876543209876e-05,
+      "loss": 5.7694,
+      "step": 977
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 0.2797470688819885,
+      "learning_rate": 9.185185185185185e-05,
+      "loss": 5.7208,
+      "step": 978
+    },
+    {
+      "epoch": 2.1719356627842483,
+      "grad_norm": 0.34863996505737305,
+      "learning_rate": 9.160493827160493e-05,
+      "loss": 6.0562,
+      "step": 979
+    },
+    {
+      "epoch": 2.1741541874653354,
+      "grad_norm": 0.24779745936393738,
+      "learning_rate": 9.135802469135802e-05,
+      "loss": 5.7886,
+      "step": 980
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "grad_norm": 0.3348560035228729,
+      "learning_rate": 9.11111111111111e-05,
+      "loss": 5.7503,
+      "step": 981
+    },
+    {
+      "epoch": 2.1785912368275095,
+      "grad_norm": 0.27655264735221863,
+      "learning_rate": 9.086419753086419e-05,
+      "loss": 5.5597,
+      "step": 982
+    },
+    {
+      "epoch": 2.1808097615085966,
+      "grad_norm": 0.44678738713264465,
+      "learning_rate": 9.061728395061727e-05,
+      "loss": 6.301,
+      "step": 983
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 0.25142818689346313,
+      "learning_rate": 9.037037037037038e-05,
+      "loss": 5.5336,
+      "step": 984
+    },
+    {
+      "epoch": 2.1852468108707708,
+      "grad_norm": 0.3007248044013977,
+      "learning_rate": 9.012345679012346e-05,
+      "loss": 6.0057,
+      "step": 985
+    },
+    {
+      "epoch": 2.187465335551858,
+      "grad_norm": 0.3076876401901245,
+      "learning_rate": 8.987654320987654e-05,
+      "loss": 5.9568,
+      "step": 986
+    },
+    {
+      "epoch": 2.189683860232945,
+      "grad_norm": 0.2927285432815552,
+      "learning_rate": 8.962962962962962e-05,
+      "loss": 5.5392,
+      "step": 987
+    },
+    {
+      "epoch": 2.191902384914032,
+      "grad_norm": 0.3983452618122101,
+      "learning_rate": 8.93827160493827e-05,
+      "loss": 5.3511,
+      "step": 988
+    },
+    {
+      "epoch": 2.194120909595119,
+      "grad_norm": 0.2616637945175171,
+      "learning_rate": 8.91358024691358e-05,
+      "loss": 5.8008,
+      "step": 989
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 0.2397451400756836,
+      "learning_rate": 8.888888888888888e-05,
+      "loss": 5.5456,
+      "step": 990
+    },
+    {
+      "epoch": 2.1985579589572932,
+      "grad_norm": 0.29482921957969666,
+      "learning_rate": 8.864197530864196e-05,
+      "loss": 5.6971,
+      "step": 991
+    },
+    {
+      "epoch": 2.2007764836383803,
+      "grad_norm": 0.2731839120388031,
+      "learning_rate": 8.839506172839506e-05,
+      "loss": 5.9089,
+      "step": 992
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "grad_norm": 0.35874083638191223,
+      "learning_rate": 8.814814814814815e-05,
+      "loss": 6.075,
+      "step": 993
+    },
+    {
+      "epoch": 2.2052135330005544,
+      "grad_norm": 0.26339277625083923,
+      "learning_rate": 8.790123456790123e-05,
+      "loss": 5.9288,
+      "step": 994
+    },
+    {
+      "epoch": 2.2074320576816415,
+      "grad_norm": 0.2823190987110138,
+      "learning_rate": 8.765432098765431e-05,
+      "loss": 5.71,
+      "step": 995
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 0.3332650661468506,
+      "learning_rate": 8.74074074074074e-05,
+      "loss": 5.5184,
+      "step": 996
+    },
+    {
+      "epoch": 2.2118691070438157,
+      "grad_norm": 0.24921450018882751,
+      "learning_rate": 8.716049382716048e-05,
+      "loss": 5.6956,
+      "step": 997
+    },
+    {
+      "epoch": 2.2140876317249027,
+      "grad_norm": 0.3111572265625,
+      "learning_rate": 8.691358024691357e-05,
+      "loss": 5.7561,
+      "step": 998
+    },
+    {
+      "epoch": 2.21630615640599,
+      "grad_norm": 0.4769589304924011,
+      "learning_rate": 8.666666666666665e-05,
+      "loss": 5.3652,
+      "step": 999
+    },
+    {
+      "epoch": 2.218524681087077,
+      "grad_norm": 0.3331756889820099,
+      "learning_rate": 8.641975308641975e-05,
+      "loss": 5.5479,
+      "step": 1000
+    },
+    {
+      "epoch": 2.220743205768164,
+      "grad_norm": 0.3365457355976105,
+      "learning_rate": 8.617283950617283e-05,
+      "loss": 5.6108,
+      "step": 1001
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 0.3461116850376129,
+      "learning_rate": 8.592592592592593e-05,
+      "loss": 5.9274,
+      "step": 1002
+    },
+    {
+      "epoch": 2.225180255130338,
+      "grad_norm": 0.2742120921611786,
+      "learning_rate": 8.567901234567901e-05,
+      "loss": 5.6285,
+      "step": 1003
+    },
+    {
+      "epoch": 2.227398779811425,
+      "grad_norm": 0.331744909286499,
+      "learning_rate": 8.543209876543209e-05,
+      "loss": 5.5965,
+      "step": 1004
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "grad_norm": 0.33731335401535034,
+      "learning_rate": 8.518518518518517e-05,
+      "loss": 5.6929,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2318358291735994,
+      "grad_norm": 0.27673766016960144,
+      "learning_rate": 8.493827160493825e-05,
+      "loss": 5.7867,
+      "step": 1006
+    },
+    {
+      "epoch": 2.2340543538546864,
+      "grad_norm": 0.41725805401802063,
+      "learning_rate": 8.469135802469136e-05,
+      "loss": 6.1468,
+      "step": 1007
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 0.2383277416229248,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 5.9339,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2384914032168606,
+      "grad_norm": 0.28230416774749756,
+      "learning_rate": 8.419753086419752e-05,
+      "loss": 6.0881,
+      "step": 1009
+    },
+    {
+      "epoch": 2.2407099278979477,
+      "grad_norm": 0.2738102972507477,
+      "learning_rate": 8.39506172839506e-05,
+      "loss": 5.884,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "grad_norm": 0.283938467502594,
+      "learning_rate": 8.37037037037037e-05,
+      "loss": 5.5652,
+      "step": 1011
+    },
+    {
+      "epoch": 2.245146977260122,
+      "grad_norm": 0.27542099356651306,
+      "learning_rate": 8.345679012345678e-05,
+      "loss": 5.5441,
+      "step": 1012
+    },
+    {
+      "epoch": 2.247365501941209,
+      "grad_norm": 0.32420995831489563,
+      "learning_rate": 8.320987654320986e-05,
+      "loss": 5.4843,
+      "step": 1013
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 0.26047033071517944,
+      "learning_rate": 8.296296296296295e-05,
+      "loss": 6.0151,
+      "step": 1014
+    },
+    {
+      "epoch": 2.251802551303383,
+      "grad_norm": 0.34595468640327454,
+      "learning_rate": 8.271604938271605e-05,
+      "loss": 5.9751,
+      "step": 1015
+    },
+    {
+      "epoch": 2.25402107598447,
+      "grad_norm": 0.3671579957008362,
+      "learning_rate": 8.246913580246914e-05,
+      "loss": 5.8779,
+      "step": 1016
+    },
+    {
+      "epoch": 2.256239600665557,
+      "grad_norm": 0.39769497513771057,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 5.7054,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2584581253466443,
+      "grad_norm": 0.33783620595932007,
+      "learning_rate": 8.19753086419753e-05,
+      "loss": 6.0444,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2606766500277313,
+      "grad_norm": 0.4008271098136902,
+      "learning_rate": 8.172839506172838e-05,
+      "loss": 5.7925,
+      "step": 1019
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 0.2604571580886841,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 5.8645,
+      "step": 1020
+    },
+    {
+      "epoch": 2.2651136993899055,
+      "grad_norm": 0.25824683904647827,
+      "learning_rate": 8.123456790123456e-05,
+      "loss": 5.8172,
+      "step": 1021
+    },
+    {
+      "epoch": 2.2673322240709926,
+      "grad_norm": 0.29511749744415283,
+      "learning_rate": 8.098765432098765e-05,
+      "loss": 5.3868,
+      "step": 1022
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "grad_norm": 0.2674168348312378,
+      "learning_rate": 8.074074074074073e-05,
+      "loss": 5.707,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2717692734331667,
+      "grad_norm": 0.5258546471595764,
+      "learning_rate": 8.049382716049383e-05,
+      "loss": 5.6617,
+      "step": 1024
+    },
+    {
+      "epoch": 2.2739877981142538,
+      "grad_norm": 0.29705774784088135,
+      "learning_rate": 8.024691358024691e-05,
+      "loss": 5.8703,
+      "step": 1025
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 0.2933792471885681,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 5.8164,
+      "step": 1026
+    },
+    {
+      "epoch": 2.278424847476428,
+      "grad_norm": 0.3253720998764038,
+      "learning_rate": 7.975308641975307e-05,
+      "loss": 5.4533,
+      "step": 1027
+    },
+    {
+      "epoch": 2.280643372157515,
+      "grad_norm": 0.28754323720932007,
+      "learning_rate": 7.950617283950616e-05,
+      "loss": 5.9328,
+      "step": 1028
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "grad_norm": 0.31493568420410156,
+      "learning_rate": 7.925925925925925e-05,
+      "loss": 5.5031,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2850804215196896,
+      "grad_norm": 0.42104557156562805,
+      "learning_rate": 7.901234567901235e-05,
+      "loss": 5.9094,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2872989462007767,
+      "grad_norm": 0.26261454820632935,
+      "learning_rate": 7.876543209876543e-05,
+      "loss": 5.7543,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 0.4576817452907562,
+      "learning_rate": 7.851851851851851e-05,
+      "loss": 5.9988,
+      "step": 1032
+    },
+    {
+      "epoch": 2.291735995562951,
+      "grad_norm": 0.2779466211795807,
+      "learning_rate": 7.82716049382716e-05,
+      "loss": 5.6808,
+      "step": 1033
+    },
+    {
+      "epoch": 2.293954520244038,
+      "grad_norm": 0.2949773967266083,
+      "learning_rate": 7.802469135802469e-05,
+      "loss": 5.7557,
+      "step": 1034
+    },
+    {
+      "epoch": 2.296173044925125,
+      "grad_norm": 0.3470287024974823,
+      "learning_rate": 7.777777777777777e-05,
+      "loss": 5.9226,
+      "step": 1035
+    },
+    {
+      "epoch": 2.298391569606212,
+      "grad_norm": 0.23799651861190796,
+      "learning_rate": 7.753086419753085e-05,
+      "loss": 5.8297,
+      "step": 1036
+    },
+    {
+      "epoch": 2.300610094287299,
+      "grad_norm": 0.32502299547195435,
+      "learning_rate": 7.728395061728396e-05,
+      "loss": 5.4988,
+      "step": 1037
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 0.3073773682117462,
+      "learning_rate": 7.703703703703704e-05,
+      "loss": 5.8565,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3050471436494733,
+      "grad_norm": 0.3692910075187683,
+      "learning_rate": 7.679012345679012e-05,
+      "loss": 5.3471,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3072656683305603,
+      "grad_norm": 0.2995283007621765,
+      "learning_rate": 7.65432098765432e-05,
+      "loss": 5.6778,
+      "step": 1040
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "grad_norm": 0.2855288088321686,
+      "learning_rate": 7.629629629629628e-05,
+      "loss": 5.9721,
+      "step": 1041
+    },
+    {
+      "epoch": 2.3117027176927345,
+      "grad_norm": 0.25702327489852905,
+      "learning_rate": 7.604938271604937e-05,
+      "loss": 5.8575,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3139212423738216,
+      "grad_norm": 0.35185128450393677,
+      "learning_rate": 7.580246913580246e-05,
+      "loss": 5.4278,
+      "step": 1043
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 0.22822247445583344,
+      "learning_rate": 7.555555555555554e-05,
+      "loss": 5.5199,
+      "step": 1044
+    },
+    {
+      "epoch": 2.3183582917359957,
+      "grad_norm": 0.3205663859844208,
+      "learning_rate": 7.530864197530864e-05,
+      "loss": 5.8534,
+      "step": 1045
+    },
+    {
+      "epoch": 2.320576816417083,
+      "grad_norm": 0.2613803744316101,
+      "learning_rate": 7.506172839506173e-05,
+      "loss": 5.7148,
+      "step": 1046
+    },
+    {
+      "epoch": 2.32279534109817,
+      "grad_norm": 0.2960833013057709,
+      "learning_rate": 7.481481481481481e-05,
+      "loss": 5.6873,
+      "step": 1047
+    },
+    {
+      "epoch": 2.325013865779257,
+      "grad_norm": 0.3325331211090088,
+      "learning_rate": 7.45679012345679e-05,
+      "loss": 5.5701,
+      "step": 1048
+    },
+    {
+      "epoch": 2.327232390460344,
+      "grad_norm": 0.3322075307369232,
+      "learning_rate": 7.432098765432098e-05,
+      "loss": 5.1744,
+      "step": 1049
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 0.33876657485961914,
+      "learning_rate": 7.407407407407406e-05,
+      "loss": 5.4963,
+      "step": 1050
+    },
+    {
+      "epoch": 2.331669439822518,
+      "grad_norm": 0.2897687554359436,
+      "learning_rate": 7.382716049382715e-05,
+      "loss": 5.6414,
+      "step": 1051
+    },
+    {
+      "epoch": 2.3338879645036053,
+      "grad_norm": 0.3089011609554291,
+      "learning_rate": 7.358024691358024e-05,
+      "loss": 5.5859,
+      "step": 1052
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "grad_norm": 0.3232138454914093,
+      "learning_rate": 7.333333333333332e-05,
+      "loss": 5.9171,
+      "step": 1053
+    },
+    {
+      "epoch": 2.3383250138657794,
+      "grad_norm": 0.30764898657798767,
+      "learning_rate": 7.308641975308641e-05,
+      "loss": 5.7594,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3405435385468665,
+      "grad_norm": 0.28956955671310425,
+      "learning_rate": 7.28395061728395e-05,
+      "loss": 5.4344,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 0.32743868231773376,
+      "learning_rate": 7.259259259259259e-05,
+      "loss": 5.771,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3449805879090406,
+      "grad_norm": 0.3195367455482483,
+      "learning_rate": 7.234567901234567e-05,
+      "loss": 5.5566,
+      "step": 1057
+    },
+    {
+      "epoch": 2.3471991125901277,
+      "grad_norm": 0.3113043010234833,
+      "learning_rate": 7.209876543209877e-05,
+      "loss": 5.767,
+      "step": 1058
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "grad_norm": 0.27252113819122314,
+      "learning_rate": 7.185185185185185e-05,
+      "loss": 5.6781,
+      "step": 1059
+    },
+    {
+      "epoch": 2.351636161952302,
+      "grad_norm": 0.3101687729358673,
+      "learning_rate": 7.160493827160493e-05,
+      "loss": 5.4751,
+      "step": 1060
+    },
+    {
+      "epoch": 2.353854686633389,
+      "grad_norm": 0.3405562937259674,
+      "learning_rate": 7.135802469135801e-05,
+      "loss": 5.6609,
+      "step": 1061
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 0.3748644292354584,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 5.7892,
+      "step": 1062
+    },
+    {
+      "epoch": 2.358291735995563,
+      "grad_norm": 0.34702062606811523,
+      "learning_rate": 7.086419753086419e-05,
+      "loss": 5.7523,
+      "step": 1063
+    },
+    {
+      "epoch": 2.36051026067665,
+      "grad_norm": 0.30943313241004944,
+      "learning_rate": 7.061728395061727e-05,
+      "loss": 5.5361,
+      "step": 1064
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "grad_norm": 0.37878015637397766,
+      "learning_rate": 7.037037037037036e-05,
+      "loss": 5.3051,
+      "step": 1065
+    },
+    {
+      "epoch": 2.3649473100388243,
+      "grad_norm": 0.2486504465341568,
+      "learning_rate": 7.012345679012345e-05,
+      "loss": 5.5709,
+      "step": 1066
+    },
+    {
+      "epoch": 2.3671658347199114,
+      "grad_norm": 0.33251139521598816,
+      "learning_rate": 6.987654320987654e-05,
+      "loss": 5.5883,
+      "step": 1067
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 0.2904629409313202,
+      "learning_rate": 6.962962962962962e-05,
+      "loss": 5.6096,
+      "step": 1068
+    },
+    {
+      "epoch": 2.3716028840820855,
+      "grad_norm": 0.26665249466896057,
+      "learning_rate": 6.938271604938272e-05,
+      "loss": 5.6798,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3738214087631726,
+      "grad_norm": 0.28194910287857056,
+      "learning_rate": 6.91358024691358e-05,
+      "loss": 5.5396,
+      "step": 1070
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "grad_norm": 0.3231011629104614,
+      "learning_rate": 6.888888888888888e-05,
+      "loss": 5.9443,
+      "step": 1071
+    },
+    {
+      "epoch": 2.3782584581253468,
+      "grad_norm": 0.3807702660560608,
+      "learning_rate": 6.864197530864196e-05,
+      "loss": 5.8895,
+      "step": 1072
+    },
+    {
+      "epoch": 2.380476982806434,
+      "grad_norm": 0.24254216253757477,
+      "learning_rate": 6.839506172839506e-05,
+      "loss": 5.7732,
+      "step": 1073
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 0.2645382881164551,
+      "learning_rate": 6.814814814814814e-05,
+      "loss": 5.658,
+      "step": 1074
+    },
+    {
+      "epoch": 2.384914032168608,
+      "grad_norm": 0.30094826221466064,
+      "learning_rate": 6.790123456790122e-05,
+      "loss": 5.659,
+      "step": 1075
+    },
+    {
+      "epoch": 2.387132556849695,
+      "grad_norm": 0.2918371260166168,
+      "learning_rate": 6.765432098765432e-05,
+      "loss": 5.5486,
+      "step": 1076
+    },
+    {
+      "epoch": 2.389351081530782,
+      "grad_norm": 0.30628493428230286,
+      "learning_rate": 6.74074074074074e-05,
+      "loss": 6.0358,
+      "step": 1077
+    },
+    {
+      "epoch": 2.391569606211869,
+      "grad_norm": 0.2610704302787781,
+      "learning_rate": 6.716049382716049e-05,
+      "loss": 5.8521,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3937881308929563,
+      "grad_norm": 0.3757915198802948,
+      "learning_rate": 6.691358024691357e-05,
+      "loss": 5.5844,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 0.35306328535079956,
+      "learning_rate": 6.666666666666666e-05,
+      "loss": 5.706,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3982251802551304,
+      "grad_norm": 0.4227609634399414,
+      "learning_rate": 6.641975308641975e-05,
+      "loss": 5.5648,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4004437049362175,
+      "grad_norm": 0.2674129605293274,
+      "learning_rate": 6.617283950617283e-05,
+      "loss": 5.6589,
+      "step": 1082
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "grad_norm": 0.27143964171409607,
+      "learning_rate": 6.592592592592591e-05,
+      "loss": 5.82,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4048807542983917,
+      "grad_norm": 0.2839408814907074,
+      "learning_rate": 6.567901234567901e-05,
+      "loss": 5.6178,
+      "step": 1084
+    },
+    {
+      "epoch": 2.4070992789794787,
+      "grad_norm": 0.25997573137283325,
+      "learning_rate": 6.543209876543209e-05,
+      "loss": 5.6461,
+      "step": 1085
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 0.27492618560791016,
+      "learning_rate": 6.518518518518517e-05,
+      "loss": 5.6736,
+      "step": 1086
+    },
+    {
+      "epoch": 2.411536328341653,
+      "grad_norm": 0.28796324133872986,
+      "learning_rate": 6.493827160493827e-05,
+      "loss": 5.5572,
+      "step": 1087
+    },
+    {
+      "epoch": 2.41375485302274,
+      "grad_norm": 0.2597390413284302,
+      "learning_rate": 6.469135802469135e-05,
+      "loss": 5.4974,
+      "step": 1088
+    },
+    {
+      "epoch": 2.415973377703827,
+      "grad_norm": 0.24780596792697906,
+      "learning_rate": 6.444444444444444e-05,
+      "loss": 5.7349,
+      "step": 1089
+    },
+    {
+      "epoch": 2.418191902384914,
+      "grad_norm": 0.360871821641922,
+      "learning_rate": 6.419753086419753e-05,
+      "loss": 5.6718,
+      "step": 1090
+    },
+    {
+      "epoch": 2.420410427066001,
+      "grad_norm": 0.3067885935306549,
+      "learning_rate": 6.395061728395061e-05,
+      "loss": 5.8697,
+      "step": 1091
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 0.2665919065475464,
+      "learning_rate": 6.37037037037037e-05,
+      "loss": 5.6249,
+      "step": 1092
+    },
+    {
+      "epoch": 2.4248474764281753,
+      "grad_norm": 0.27964499592781067,
+      "learning_rate": 6.345679012345678e-05,
+      "loss": 5.607,
+      "step": 1093
+    },
+    {
+      "epoch": 2.4270660011092624,
+      "grad_norm": 0.26800692081451416,
+      "learning_rate": 6.320987654320987e-05,
+      "loss": 5.598,
+      "step": 1094
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "grad_norm": 0.26584506034851074,
+      "learning_rate": 6.296296296296295e-05,
+      "loss": 5.7853,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4315030504714366,
+      "grad_norm": 0.24781134724617004,
+      "learning_rate": 6.271604938271604e-05,
+      "loss": 5.5888,
+      "step": 1096
+    },
+    {
+      "epoch": 2.4337215751525236,
+      "grad_norm": 0.33952412009239197,
+      "learning_rate": 6.246913580246912e-05,
+      "loss": 5.6871,
+      "step": 1097
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 0.3116956353187561,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 5.6901,
+      "step": 1098
+    },
+    {
+      "epoch": 2.438158624514698,
+      "grad_norm": 0.31081920862197876,
+      "learning_rate": 6.19753086419753e-05,
+      "loss": 5.503,
+      "step": 1099
+    },
+    {
+      "epoch": 2.440377149195785,
+      "grad_norm": 0.3148333430290222,
+      "learning_rate": 6.17283950617284e-05,
+      "loss": 5.5126,
+      "step": 1100
+    },
+    {
+      "epoch": 2.442595673876872,
+      "grad_norm": 0.32355841994285583,
+      "learning_rate": 6.148148148148148e-05,
+      "loss": 5.7286,
+      "step": 1101
+    },
+    {
+      "epoch": 2.444814198557959,
+      "grad_norm": 0.2812463939189911,
+      "learning_rate": 6.123456790123456e-05,
+      "loss": 5.5617,
+      "step": 1102
+    },
+    {
+      "epoch": 2.447032723239046,
+      "grad_norm": 0.2884272336959839,
+      "learning_rate": 6.0987654320987655e-05,
+      "loss": 5.6425,
+      "step": 1103
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 0.3218716084957123,
+      "learning_rate": 6.0740740740740736e-05,
+      "loss": 5.5098,
+      "step": 1104
+    },
+    {
+      "epoch": 2.4514697726012202,
+      "grad_norm": 0.33052629232406616,
+      "learning_rate": 6.049382716049382e-05,
+      "loss": 5.6182,
+      "step": 1105
+    },
+    {
+      "epoch": 2.4536882972823073,
+      "grad_norm": 0.3461867570877075,
+      "learning_rate": 6.0246913580246906e-05,
+      "loss": 5.5006,
+      "step": 1106
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "grad_norm": 0.42240285873413086,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 5.5731,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4581253466444815,
+      "grad_norm": 0.25891977548599243,
+      "learning_rate": 5.975308641975308e-05,
+      "loss": 5.5823,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4603438713255685,
+      "grad_norm": 0.3646548092365265,
+      "learning_rate": 5.9506172839506165e-05,
+      "loss": 5.1591,
+      "step": 1109
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 0.27875837683677673,
+      "learning_rate": 5.925925925925925e-05,
+      "loss": 5.6427,
+      "step": 1110
+    },
+    {
+      "epoch": 2.4647809206877427,
+      "grad_norm": 0.31725168228149414,
+      "learning_rate": 5.901234567901234e-05,
+      "loss": 5.6521,
+      "step": 1111
+    },
+    {
+      "epoch": 2.4669994453688298,
+      "grad_norm": 0.2551935911178589,
+      "learning_rate": 5.876543209876543e-05,
+      "loss": 5.5676,
+      "step": 1112
+    },
+    {
+      "epoch": 2.469217970049917,
+      "grad_norm": 0.5986329913139343,
+      "learning_rate": 5.851851851851851e-05,
+      "loss": 5.4685,
+      "step": 1113
+    },
+    {
+      "epoch": 2.471436494731004,
+      "grad_norm": 0.41441723704338074,
+      "learning_rate": 5.827160493827159e-05,
+      "loss": 5.8747,
+      "step": 1114
+    },
+    {
+      "epoch": 2.473655019412091,
+      "grad_norm": 0.30362680554389954,
+      "learning_rate": 5.802469135802469e-05,
+      "loss": 5.2807,
+      "step": 1115
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 0.30841830372810364,
+      "learning_rate": 5.777777777777777e-05,
+      "loss": 5.7462,
+      "step": 1116
+    },
+    {
+      "epoch": 2.478092068774265,
+      "grad_norm": 0.3365178406238556,
+      "learning_rate": 5.753086419753086e-05,
+      "loss": 5.7722,
+      "step": 1117
+    },
+    {
+      "epoch": 2.480310593455352,
+      "grad_norm": 0.28625693917274475,
+      "learning_rate": 5.7283950617283946e-05,
+      "loss": 5.4529,
+      "step": 1118
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "grad_norm": 0.3175484240055084,
+      "learning_rate": 5.7037037037037035e-05,
+      "loss": 5.7688,
+      "step": 1119
+    },
+    {
+      "epoch": 2.4847476428175264,
+      "grad_norm": 0.3571614921092987,
+      "learning_rate": 5.6790123456790116e-05,
+      "loss": 5.7018,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4869661674986134,
+      "grad_norm": 0.3747718930244446,
+      "learning_rate": 5.6543209876543205e-05,
+      "loss": 5.5873,
+      "step": 1121
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 0.2892332673072815,
+      "learning_rate": 5.629629629629629e-05,
+      "loss": 5.6774,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4914032168607876,
+      "grad_norm": 0.4634627401828766,
+      "learning_rate": 5.604938271604938e-05,
+      "loss": 5.2946,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4936217415418747,
+      "grad_norm": 0.26469242572784424,
+      "learning_rate": 5.580246913580246e-05,
+      "loss": 5.4704,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "grad_norm": 0.49113011360168457,
+      "learning_rate": 5.5555555555555545e-05,
+      "loss": 6.2237,
+      "step": 1125
+    },
+    {
+      "epoch": 2.498058790904049,
+      "grad_norm": 0.30069872736930847,
+      "learning_rate": 5.530864197530864e-05,
+      "loss": 6.0072,
+      "step": 1126
+    },
+    {
+      "epoch": 2.500277315585136,
+      "grad_norm": 0.3111431300640106,
+      "learning_rate": 5.506172839506172e-05,
+      "loss": 5.454,
+      "step": 1127
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 0.2739917039871216,
+      "learning_rate": 5.481481481481481e-05,
+      "loss": 5.7464,
+      "step": 1128
+    },
+    {
+      "epoch": 2.50471436494731,
+      "grad_norm": 0.3138275444507599,
+      "learning_rate": 5.456790123456789e-05,
+      "loss": 5.8816,
+      "step": 1129
+    },
+    {
+      "epoch": 2.506932889628397,
+      "grad_norm": 0.28242769837379456,
+      "learning_rate": 5.4320987654320986e-05,
+      "loss": 5.6062,
+      "step": 1130
+    },
+    {
+      "epoch": 2.509151414309484,
+      "grad_norm": 0.25297480821609497,
+      "learning_rate": 5.407407407407407e-05,
+      "loss": 5.5526,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5113699389905713,
+      "grad_norm": 0.34635624289512634,
+      "learning_rate": 5.3827160493827156e-05,
+      "loss": 5.6187,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5135884636716583,
+      "grad_norm": 0.3009355962276459,
+      "learning_rate": 5.358024691358024e-05,
+      "loss": 5.4882,
+      "step": 1133
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 0.42253124713897705,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 5.7271,
+      "step": 1134
+    },
+    {
+      "epoch": 2.5180255130338325,
+      "grad_norm": 0.26582252979278564,
+      "learning_rate": 5.3086419753086415e-05,
+      "loss": 5.5681,
+      "step": 1135
+    },
+    {
+      "epoch": 2.5202440377149196,
+      "grad_norm": 0.23530659079551697,
+      "learning_rate": 5.2839506172839496e-05,
+      "loss": 5.691,
+      "step": 1136
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "grad_norm": 0.4070867896080017,
+      "learning_rate": 5.259259259259259e-05,
+      "loss": 5.6099,
+      "step": 1137
+    },
+    {
+      "epoch": 2.5246810870770937,
+      "grad_norm": 0.26389339566230774,
+      "learning_rate": 5.234567901234567e-05,
+      "loss": 5.8424,
+      "step": 1138
+    },
+    {
+      "epoch": 2.526899611758181,
+      "grad_norm": 0.2637118697166443,
+      "learning_rate": 5.209876543209876e-05,
+      "loss": 5.8733,
+      "step": 1139
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 0.29476499557495117,
+      "learning_rate": 5.185185185185184e-05,
+      "loss": 5.5079,
+      "step": 1140
+    },
+    {
+      "epoch": 2.531336661120355,
+      "grad_norm": 0.40996670722961426,
+      "learning_rate": 5.160493827160494e-05,
+      "loss": 5.6281,
+      "step": 1141
+    },
+    {
+      "epoch": 2.533555185801442,
+      "grad_norm": 0.2837502360343933,
+      "learning_rate": 5.135802469135802e-05,
+      "loss": 5.5161,
+      "step": 1142
+    },
+    {
+      "epoch": 2.535773710482529,
+      "grad_norm": 0.3308456242084503,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 5.6226,
+      "step": 1143
+    },
+    {
+      "epoch": 2.537992235163616,
+      "grad_norm": 0.2901027202606201,
+      "learning_rate": 5.086419753086419e-05,
+      "loss": 5.9361,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5402107598447032,
+      "grad_norm": 0.3781372904777527,
+      "learning_rate": 5.0617283950617285e-05,
+      "loss": 5.6245,
+      "step": 1145
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 0.23980286717414856,
+      "learning_rate": 5.0370370370370366e-05,
+      "loss": 5.6528,
+      "step": 1146
+    },
+    {
+      "epoch": 2.5446478092068774,
+      "grad_norm": 0.370477557182312,
+      "learning_rate": 5.012345679012345e-05,
+      "loss": 5.7352,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5468663338879645,
+      "grad_norm": 0.30925801396369934,
+      "learning_rate": 4.9876543209876536e-05,
+      "loss": 5.629,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "grad_norm": 0.34497538208961487,
+      "learning_rate": 4.9629629629629625e-05,
+      "loss": 5.6815,
+      "step": 1149
+    },
+    {
+      "epoch": 2.5513033832501386,
+      "grad_norm": 0.40277189016342163,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 5.7416,
+      "step": 1150
+    },
+    {
+      "epoch": 2.5535219079312257,
+      "grad_norm": 0.3758951723575592,
+      "learning_rate": 4.9135802469135795e-05,
+      "loss": 5.6518,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 0.2651861011981964,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 5.5446,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5579589572934,
+      "grad_norm": 0.30919361114501953,
+      "learning_rate": 4.864197530864197e-05,
+      "loss": 5.7641,
+      "step": 1153
+    },
+    {
+      "epoch": 2.560177481974487,
+      "grad_norm": 0.3434813320636749,
+      "learning_rate": 4.839506172839506e-05,
+      "loss": 5.8382,
+      "step": 1154
+    },
+    {
+      "epoch": 2.562396006655574,
+      "grad_norm": 0.34428805112838745,
+      "learning_rate": 4.814814814814814e-05,
+      "loss": 5.548,
+      "step": 1155
+    },
+    {
+      "epoch": 2.564614531336661,
+      "grad_norm": 0.5566628575325012,
+      "learning_rate": 4.7901234567901237e-05,
+      "loss": 5.4863,
+      "step": 1156
+    },
+    {
+      "epoch": 2.566833056017748,
+      "grad_norm": 0.4321678578853607,
+      "learning_rate": 4.765432098765432e-05,
+      "loss": 5.5593,
+      "step": 1157
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 0.3943154811859131,
+      "learning_rate": 4.74074074074074e-05,
+      "loss": 5.9819,
+      "step": 1158
+    },
+    {
+      "epoch": 2.5712701053799223,
+      "grad_norm": 0.3628782331943512,
+      "learning_rate": 4.716049382716049e-05,
+      "loss": 5.5447,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5734886300610094,
+      "grad_norm": 0.2429974228143692,
+      "learning_rate": 4.6913580246913576e-05,
+      "loss": 5.4531,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "grad_norm": 0.2712736129760742,
+      "learning_rate": 4.6666666666666665e-05,
+      "loss": 5.7881,
+      "step": 1161
+    },
+    {
+      "epoch": 2.5779256794231835,
+      "grad_norm": 0.26362016797065735,
+      "learning_rate": 4.6419753086419746e-05,
+      "loss": 5.4896,
+      "step": 1162
+    },
+    {
+      "epoch": 2.5801442041042706,
+      "grad_norm": 0.27611616253852844,
+      "learning_rate": 4.6172839506172835e-05,
+      "loss": 5.4949,
+      "step": 1163
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 0.3090492784976959,
+      "learning_rate": 4.592592592592592e-05,
+      "loss": 5.6779,
+      "step": 1164
+    },
+    {
+      "epoch": 2.5845812534664447,
+      "grad_norm": 0.3601735532283783,
+      "learning_rate": 4.567901234567901e-05,
+      "loss": 5.8185,
+      "step": 1165
+    },
+    {
+      "epoch": 2.586799778147532,
+      "grad_norm": 0.3195878863334656,
+      "learning_rate": 4.543209876543209e-05,
+      "loss": 5.7804,
+      "step": 1166
+    },
+    {
+      "epoch": 2.589018302828619,
+      "grad_norm": 0.31010913848876953,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 5.7482,
+      "step": 1167
+    },
+    {
+      "epoch": 2.591236827509706,
+      "grad_norm": 0.285243958234787,
+      "learning_rate": 4.493827160493827e-05,
+      "loss": 5.5558,
+      "step": 1168
+    },
+    {
+      "epoch": 2.593455352190793,
+      "grad_norm": 0.357623815536499,
+      "learning_rate": 4.469135802469135e-05,
+      "loss": 5.5733,
+      "step": 1169
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 0.2810336947441101,
+      "learning_rate": 4.444444444444444e-05,
+      "loss": 5.6837,
+      "step": 1170
+    },
+    {
+      "epoch": 2.597892401552967,
+      "grad_norm": 0.3060472309589386,
+      "learning_rate": 4.419753086419753e-05,
+      "loss": 5.5904,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6001109262340543,
+      "grad_norm": 0.27810728549957275,
+      "learning_rate": 4.3950617283950617e-05,
+      "loss": 5.8101,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "grad_norm": 0.5223957300186157,
+      "learning_rate": 4.37037037037037e-05,
+      "loss": 5.57,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6045479755962284,
+      "grad_norm": 0.29377481341362,
+      "learning_rate": 4.3456790123456786e-05,
+      "loss": 5.6468,
+      "step": 1174
+    },
+    {
+      "epoch": 2.6067665002773155,
+      "grad_norm": 0.25924959778785706,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 5.5745,
+      "step": 1175
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 0.3703748285770416,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 5.9111,
+      "step": 1176
+    },
+    {
+      "epoch": 2.6112035496394896,
+      "grad_norm": 0.2434493601322174,
+      "learning_rate": 4.2716049382716045e-05,
+      "loss": 5.882,
+      "step": 1177
+    },
+    {
+      "epoch": 2.6134220743205767,
+      "grad_norm": 0.25615194439888,
+      "learning_rate": 4.2469135802469126e-05,
+      "loss": 5.4687,
+      "step": 1178
+    },
+    {
+      "epoch": 2.615640599001664,
+      "grad_norm": 0.3246670067310333,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 5.9336,
+      "step": 1179
+    },
+    {
+      "epoch": 2.617859123682751,
+      "grad_norm": 0.3051455020904541,
+      "learning_rate": 4.19753086419753e-05,
+      "loss": 5.6773,
+      "step": 1180
+    },
+    {
+      "epoch": 2.620077648363838,
+      "grad_norm": 0.46029454469680786,
+      "learning_rate": 4.172839506172839e-05,
+      "loss": 5.7999,
+      "step": 1181
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 0.28870654106140137,
+      "learning_rate": 4.148148148148147e-05,
+      "loss": 5.9969,
+      "step": 1182
+    },
+    {
+      "epoch": 2.624514697726012,
+      "grad_norm": 0.3221019506454468,
+      "learning_rate": 4.123456790123457e-05,
+      "loss": 5.3505,
+      "step": 1183
+    },
+    {
+      "epoch": 2.626733222407099,
+      "grad_norm": 0.29089629650115967,
+      "learning_rate": 4.098765432098765e-05,
+      "loss": 5.5693,
+      "step": 1184
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "grad_norm": 0.3627305328845978,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 5.7762,
+      "step": 1185
+    },
+    {
+      "epoch": 2.6311702717692733,
+      "grad_norm": 0.2954031229019165,
+      "learning_rate": 4.0493827160493827e-05,
+      "loss": 5.6733,
+      "step": 1186
+    },
+    {
+      "epoch": 2.6333887964503604,
+      "grad_norm": 0.24689443409442902,
+      "learning_rate": 4.0246913580246915e-05,
+      "loss": 5.7755,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 0.2822924554347992,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 5.6837,
+      "step": 1188
+    },
+    {
+      "epoch": 2.6378258458125345,
+      "grad_norm": 0.25982946157455444,
+      "learning_rate": 3.975308641975308e-05,
+      "loss": 5.5118,
+      "step": 1189
+    },
+    {
+      "epoch": 2.6400443704936216,
+      "grad_norm": 0.37345850467681885,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 5.7398,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "grad_norm": 0.41383078694343567,
+      "learning_rate": 3.9259259259259255e-05,
+      "loss": 5.4673,
+      "step": 1191
+    },
+    {
+      "epoch": 2.6444814198557958,
+      "grad_norm": 0.2415016144514084,
+      "learning_rate": 3.901234567901234e-05,
+      "loss": 5.5872,
+      "step": 1192
+    },
+    {
+      "epoch": 2.646699944536883,
+      "grad_norm": 0.36533409357070923,
+      "learning_rate": 3.8765432098765425e-05,
+      "loss": 6.0727,
+      "step": 1193
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 0.38330957293510437,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 5.7172,
+      "step": 1194
+    },
+    {
+      "epoch": 2.651136993899057,
+      "grad_norm": 0.33583369851112366,
+      "learning_rate": 3.82716049382716e-05,
+      "loss": 5.4306,
+      "step": 1195
+    },
+    {
+      "epoch": 2.653355518580144,
+      "grad_norm": 0.35398778319358826,
+      "learning_rate": 3.802469135802468e-05,
+      "loss": 5.2899,
+      "step": 1196
+    },
+    {
+      "epoch": 2.655574043261231,
+      "grad_norm": 0.28680330514907837,
+      "learning_rate": 3.777777777777777e-05,
+      "loss": 5.5226,
+      "step": 1197
+    },
+    {
+      "epoch": 2.657792567942318,
+      "grad_norm": 0.2895892560482025,
+      "learning_rate": 3.7530864197530867e-05,
+      "loss": 5.677,
+      "step": 1198
+    },
+    {
+      "epoch": 2.6600110926234053,
+      "grad_norm": 0.2150576263666153,
+      "learning_rate": 3.728395061728395e-05,
+      "loss": 5.6135,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 0.341113805770874,
+      "learning_rate": 3.703703703703703e-05,
+      "loss": 5.4525,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6644481419855794,
+      "grad_norm": 0.3083145022392273,
+      "learning_rate": 3.679012345679012e-05,
+      "loss": 5.6202,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.2874879539012909,
+      "learning_rate": 3.6543209876543207e-05,
+      "loss": 5.8789,
+      "step": 1202
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "grad_norm": 0.312677264213562,
+      "learning_rate": 3.6296296296296295e-05,
+      "loss": 5.5538,
+      "step": 1203
+    },
+    {
+      "epoch": 2.6711037160288407,
+      "grad_norm": 0.40666621923446655,
+      "learning_rate": 3.604938271604938e-05,
+      "loss": 5.8201,
+      "step": 1204
+    },
+    {
+      "epoch": 2.6733222407099277,
+      "grad_norm": 0.27613338828086853,
+      "learning_rate": 3.5802469135802465e-05,
+      "loss": 5.612,
+      "step": 1205
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 0.2720215320587158,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 5.6038,
+      "step": 1206
+    },
+    {
+      "epoch": 2.677759290072102,
+      "grad_norm": 0.31409865617752075,
+      "learning_rate": 3.5308641975308635e-05,
+      "loss": 5.7534,
+      "step": 1207
+    },
+    {
+      "epoch": 2.679977814753189,
+      "grad_norm": 0.3614393174648285,
+      "learning_rate": 3.506172839506172e-05,
+      "loss": 5.7097,
+      "step": 1208
+    },
+    {
+      "epoch": 2.682196339434276,
+      "grad_norm": 0.2858182191848755,
+      "learning_rate": 3.481481481481481e-05,
+      "loss": 5.5965,
+      "step": 1209
+    },
+    {
+      "epoch": 2.684414864115363,
+      "grad_norm": 0.4262728989124298,
+      "learning_rate": 3.45679012345679e-05,
+      "loss": 5.5067,
+      "step": 1210
+    },
+    {
+      "epoch": 2.68663338879645,
+      "grad_norm": 0.30125781893730164,
+      "learning_rate": 3.432098765432098e-05,
+      "loss": 5.5796,
+      "step": 1211
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 0.296257883310318,
+      "learning_rate": 3.407407407407407e-05,
+      "loss": 5.8722,
+      "step": 1212
+    },
+    {
+      "epoch": 2.691070438158625,
+      "grad_norm": 0.33235904574394226,
+      "learning_rate": 3.382716049382716e-05,
+      "loss": 5.8703,
+      "step": 1213
+    },
+    {
+      "epoch": 2.693288962839712,
+      "grad_norm": 0.24581898748874664,
+      "learning_rate": 3.3580246913580247e-05,
+      "loss": 5.4259,
+      "step": 1214
+    },
+    {
+      "epoch": 2.695507487520799,
+      "grad_norm": 0.3715357184410095,
+      "learning_rate": 3.333333333333333e-05,
+      "loss": 5.6583,
+      "step": 1215
+    },
+    {
+      "epoch": 2.697726012201886,
+      "grad_norm": 0.30542024970054626,
+      "learning_rate": 3.3086419753086417e-05,
+      "loss": 5.4771,
+      "step": 1216
+    },
+    {
+      "epoch": 2.699944536882973,
+      "grad_norm": 0.3289650082588196,
+      "learning_rate": 3.2839506172839505e-05,
+      "loss": 5.6341,
+      "step": 1217
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 0.3260744512081146,
+      "learning_rate": 3.2592592592592586e-05,
+      "loss": 5.8712,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7043815862451472,
+      "grad_norm": 0.30479201674461365,
+      "learning_rate": 3.2345679012345675e-05,
+      "loss": 5.6594,
+      "step": 1219
+    },
+    {
+      "epoch": 2.7066001109262343,
+      "grad_norm": 0.4105900228023529,
+      "learning_rate": 3.209876543209876e-05,
+      "loss": 5.7686,
+      "step": 1220
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "grad_norm": 0.3681042492389679,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 5.6466,
+      "step": 1221
+    },
+    {
+      "epoch": 2.7110371602884085,
+      "grad_norm": 0.2787615656852722,
+      "learning_rate": 3.160493827160493e-05,
+      "loss": 5.6676,
+      "step": 1222
+    },
+    {
+      "epoch": 2.7132556849694955,
+      "grad_norm": 0.36532846093177795,
+      "learning_rate": 3.135802469135802e-05,
+      "loss": 5.7602,
+      "step": 1223
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 0.39529645442962646,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 5.8207,
+      "step": 1224
+    },
+    {
+      "epoch": 2.7176927343316697,
+      "grad_norm": 0.2880403995513916,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 5.6029,
+      "step": 1225
+    },
+    {
+      "epoch": 2.7199112590127568,
+      "grad_norm": 0.30469566583633423,
+      "learning_rate": 3.061728395061728e-05,
+      "loss": 5.4041,
+      "step": 1226
+    },
+    {
+      "epoch": 2.722129783693844,
+      "grad_norm": 0.26085764169692993,
+      "learning_rate": 3.0370370370370368e-05,
+      "loss": 5.5399,
+      "step": 1227
+    },
+    {
+      "epoch": 2.724348308374931,
+      "grad_norm": 0.24870218336582184,
+      "learning_rate": 3.0123456790123453e-05,
+      "loss": 5.8019,
+      "step": 1228
+    },
+    {
+      "epoch": 2.726566833056018,
+      "grad_norm": 0.3325478136539459,
+      "learning_rate": 2.987654320987654e-05,
+      "loss": 5.6605,
+      "step": 1229
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 0.23484086990356445,
+      "learning_rate": 2.9629629629629627e-05,
+      "loss": 5.6749,
+      "step": 1230
+    },
+    {
+      "epoch": 2.731003882418192,
+      "grad_norm": 0.24568010866641998,
+      "learning_rate": 2.9382716049382715e-05,
+      "loss": 5.9053,
+      "step": 1231
+    },
+    {
+      "epoch": 2.733222407099279,
+      "grad_norm": 0.436434268951416,
+      "learning_rate": 2.9135802469135796e-05,
+      "loss": 5.4359,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "grad_norm": 0.3233731687068939,
+      "learning_rate": 2.8888888888888885e-05,
+      "loss": 5.438,
+      "step": 1233
+    },
+    {
+      "epoch": 2.7376594564614534,
+      "grad_norm": 0.2769094705581665,
+      "learning_rate": 2.8641975308641973e-05,
+      "loss": 5.7864,
+      "step": 1234
+    },
+    {
+      "epoch": 2.7398779811425404,
+      "grad_norm": 0.3109165132045746,
+      "learning_rate": 2.8395061728395058e-05,
+      "loss": 5.913,
+      "step": 1235
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 0.3026023209095001,
+      "learning_rate": 2.8148148148148147e-05,
+      "loss": 5.6368,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7443150305047146,
+      "grad_norm": 0.2862052321434021,
+      "learning_rate": 2.790123456790123e-05,
+      "loss": 5.6706,
+      "step": 1237
+    },
+    {
+      "epoch": 2.7465335551858017,
+      "grad_norm": 0.28596755862236023,
+      "learning_rate": 2.765432098765432e-05,
+      "loss": 5.5556,
+      "step": 1238
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "grad_norm": 0.33789557218551636,
+      "learning_rate": 2.7407407407407405e-05,
+      "loss": 5.6659,
+      "step": 1239
+    },
+    {
+      "epoch": 2.750970604547976,
+      "grad_norm": 0.4839741289615631,
+      "learning_rate": 2.7160493827160493e-05,
+      "loss": 5.5593,
+      "step": 1240
+    },
+    {
+      "epoch": 2.753189129229063,
+      "grad_norm": 0.43794745206832886,
+      "learning_rate": 2.6913580246913578e-05,
+      "loss": 5.5943,
+      "step": 1241
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 0.3052460253238678,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 5.8761,
+      "step": 1242
+    },
+    {
+      "epoch": 2.757626178591237,
+      "grad_norm": 0.3472137749195099,
+      "learning_rate": 2.6419753086419748e-05,
+      "loss": 5.5366,
+      "step": 1243
+    },
+    {
+      "epoch": 2.759844703272324,
+      "grad_norm": 0.29986676573753357,
+      "learning_rate": 2.6172839506172837e-05,
+      "loss": 5.7103,
+      "step": 1244
+    },
+    {
+      "epoch": 2.762063227953411,
+      "grad_norm": 0.2676811218261719,
+      "learning_rate": 2.592592592592592e-05,
+      "loss": 5.8252,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7642817526344983,
+      "grad_norm": 0.34701329469680786,
+      "learning_rate": 2.567901234567901e-05,
+      "loss": 5.8085,
+      "step": 1246
+    },
+    {
+      "epoch": 2.7665002773155853,
+      "grad_norm": 0.3022293150424957,
+      "learning_rate": 2.5432098765432095e-05,
+      "loss": 5.6928,
+      "step": 1247
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 0.33803585171699524,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 5.3223,
+      "step": 1248
+    },
+    {
+      "epoch": 2.7709373266777595,
+      "grad_norm": 0.31117570400238037,
+      "learning_rate": 2.4938271604938268e-05,
+      "loss": 5.5491,
+      "step": 1249
+    },
+    {
+      "epoch": 2.7731558513588466,
+      "grad_norm": 0.4825194776058197,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 5.494,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "grad_norm": 0.36083611845970154,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 5.8684,
+      "step": 1251
+    },
+    {
+      "epoch": 2.7775929007210207,
+      "grad_norm": 0.2890376150608063,
+      "learning_rate": 2.419753086419753e-05,
+      "loss": 5.373,
+      "step": 1252
+    },
+    {
+      "epoch": 2.779811425402108,
+      "grad_norm": 0.39233866333961487,
+      "learning_rate": 2.3950617283950618e-05,
+      "loss": 5.7392,
+      "step": 1253
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 0.3447974622249603,
+      "learning_rate": 2.37037037037037e-05,
+      "loss": 5.6871,
+      "step": 1254
+    },
+    {
+      "epoch": 2.784248474764282,
+      "grad_norm": 0.25832536816596985,
+      "learning_rate": 2.3456790123456788e-05,
+      "loss": 5.8838,
+      "step": 1255
+    },
+    {
+      "epoch": 2.786466999445369,
+      "grad_norm": 0.29684215784072876,
+      "learning_rate": 2.3209876543209873e-05,
+      "loss": 5.5467,
+      "step": 1256
+    },
+    {
+      "epoch": 2.788685524126456,
+      "grad_norm": 0.29306212067604065,
+      "learning_rate": 2.296296296296296e-05,
+      "loss": 5.8789,
+      "step": 1257
+    },
+    {
+      "epoch": 2.790904048807543,
+      "grad_norm": 0.27690747380256653,
+      "learning_rate": 2.2716049382716047e-05,
+      "loss": 5.5281,
+      "step": 1258
+    },
+    {
+      "epoch": 2.7931225734886302,
+      "grad_norm": 0.32010582089424133,
+      "learning_rate": 2.2469135802469135e-05,
+      "loss": 5.7625,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 0.2817310094833374,
+      "learning_rate": 2.222222222222222e-05,
+      "loss": 5.7958,
+      "step": 1260
+    },
+    {
+      "epoch": 2.7975596228508044,
+      "grad_norm": 0.3532772958278656,
+      "learning_rate": 2.1975308641975308e-05,
+      "loss": 5.542,
+      "step": 1261
+    },
+    {
+      "epoch": 2.7997781475318915,
+      "grad_norm": 0.40652695298194885,
+      "learning_rate": 2.1728395061728393e-05,
+      "loss": 5.5105,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "grad_norm": 0.3250524699687958,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 5.469,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8042151968940656,
+      "grad_norm": 0.2543196380138397,
+      "learning_rate": 2.1234567901234563e-05,
+      "loss": 5.541,
+      "step": 1264
+    },
+    {
+      "epoch": 2.8064337215751527,
+      "grad_norm": 0.31647297739982605,
+      "learning_rate": 2.098765432098765e-05,
+      "loss": 5.7287,
+      "step": 1265
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 0.3016076683998108,
+      "learning_rate": 2.0740740740740737e-05,
+      "loss": 5.4418,
+      "step": 1266
+    },
+    {
+      "epoch": 2.810870770937327,
+      "grad_norm": 0.37522077560424805,
+      "learning_rate": 2.0493827160493825e-05,
+      "loss": 5.7464,
+      "step": 1267
+    },
+    {
+      "epoch": 2.813089295618414,
+      "grad_norm": 0.26741987466812134,
+      "learning_rate": 2.0246913580246913e-05,
+      "loss": 5.3691,
+      "step": 1268
+    },
+    {
+      "epoch": 2.815307820299501,
+      "grad_norm": 0.3142679035663605,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 5.852,
+      "step": 1269
+    },
+    {
+      "epoch": 2.817526344980588,
+      "grad_norm": 0.29070940613746643,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 5.5619,
+      "step": 1270
+    },
+    {
+      "epoch": 2.819744869661675,
+      "grad_norm": 0.3353137969970703,
+      "learning_rate": 1.950617283950617e-05,
+      "loss": 5.6332,
+      "step": 1271
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 0.2878232002258301,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 5.423,
+      "step": 1272
+    },
+    {
+      "epoch": 2.8241819190238493,
+      "grad_norm": 0.28252172470092773,
+      "learning_rate": 1.901234567901234e-05,
+      "loss": 5.6515,
+      "step": 1273
+    },
+    {
+      "epoch": 2.8264004437049364,
+      "grad_norm": 0.3346900939941406,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 5.4817,
+      "step": 1274
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "grad_norm": 0.38916054368019104,
+      "learning_rate": 1.8518518518518515e-05,
+      "loss": 5.564,
+      "step": 1275
+    },
+    {
+      "epoch": 2.8308374930671105,
+      "grad_norm": 0.4908605217933655,
+      "learning_rate": 1.8271604938271603e-05,
+      "loss": 5.6351,
+      "step": 1276
+    },
+    {
+      "epoch": 2.8330560177481976,
+      "grad_norm": 0.27528101205825806,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 5.6483,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 0.3419778048992157,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 5.8491,
+      "step": 1278
+    },
+    {
+      "epoch": 2.8374930671103717,
+      "grad_norm": 0.3032209277153015,
+      "learning_rate": 1.753086419753086e-05,
+      "loss": 5.4059,
+      "step": 1279
+    },
+    {
+      "epoch": 2.839711591791459,
+      "grad_norm": 0.3024456799030304,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 5.7314,
+      "step": 1280
+    },
+    {
+      "epoch": 2.841930116472546,
+      "grad_norm": 0.28747376799583435,
+      "learning_rate": 1.7037037037037035e-05,
+      "loss": 5.3625,
+      "step": 1281
+    },
+    {
+      "epoch": 2.844148641153633,
+      "grad_norm": 0.32421809434890747,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 5.8115,
+      "step": 1282
+    },
+    {
+      "epoch": 2.84636716583472,
+      "grad_norm": 0.4033227264881134,
+      "learning_rate": 1.6543209876543208e-05,
+      "loss": 5.5321,
+      "step": 1283
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 0.3036157488822937,
+      "learning_rate": 1.6296296296296293e-05,
+      "loss": 5.7194,
+      "step": 1284
+    },
+    {
+      "epoch": 2.850804215196894,
+      "grad_norm": 0.3595792055130005,
+      "learning_rate": 1.604938271604938e-05,
+      "loss": 5.4342,
+      "step": 1285
+    },
+    {
+      "epoch": 2.8530227398779813,
+      "grad_norm": 0.4414348900318146,
+      "learning_rate": 1.5802469135802467e-05,
+      "loss": 5.5598,
+      "step": 1286
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "grad_norm": 0.4223410487174988,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 5.0597,
+      "step": 1287
+    },
+    {
+      "epoch": 2.8574597892401554,
+      "grad_norm": 0.32326826453208923,
+      "learning_rate": 1.530864197530864e-05,
+      "loss": 5.7042,
+      "step": 1288
+    },
+    {
+      "epoch": 2.8596783139212425,
+      "grad_norm": 0.3161419928073883,
+      "learning_rate": 1.5061728395061727e-05,
+      "loss": 5.4573,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 0.33762913942337036,
+      "learning_rate": 1.4814814814814813e-05,
+      "loss": 5.5068,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8641153632834166,
+      "grad_norm": 0.27829432487487793,
+      "learning_rate": 1.4567901234567898e-05,
+      "loss": 5.607,
+      "step": 1291
+    },
+    {
+      "epoch": 2.8663338879645037,
+      "grad_norm": 0.3077280819416046,
+      "learning_rate": 1.4320987654320987e-05,
+      "loss": 5.4475,
+      "step": 1292
+    },
+    {
+      "epoch": 2.868552412645591,
+      "grad_norm": 0.30675008893013,
+      "learning_rate": 1.4074074074074073e-05,
+      "loss": 5.5355,
+      "step": 1293
+    },
+    {
+      "epoch": 2.870770937326678,
+      "grad_norm": 0.444680392742157,
+      "learning_rate": 1.382716049382716e-05,
+      "loss": 5.685,
+      "step": 1294
+    },
+    {
+      "epoch": 2.872989462007765,
+      "grad_norm": 0.2679378390312195,
+      "learning_rate": 1.3580246913580247e-05,
+      "loss": 5.6203,
+      "step": 1295
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 0.2623269259929657,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 5.6765,
+      "step": 1296
+    },
+    {
+      "epoch": 2.877426511369939,
+      "grad_norm": 0.45007357001304626,
+      "learning_rate": 1.3086419753086418e-05,
+      "loss": 5.4125,
+      "step": 1297
+    },
+    {
+      "epoch": 2.879645036051026,
+      "grad_norm": 0.3131972551345825,
+      "learning_rate": 1.2839506172839505e-05,
+      "loss": 5.5816,
+      "step": 1298
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "grad_norm": 0.2932264804840088,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 5.8004,
+      "step": 1299
+    },
+    {
+      "epoch": 2.8840820854132003,
+      "grad_norm": 0.36489757895469666,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 5.6153,
+      "step": 1300
+    },
+    {
+      "epoch": 2.8863006100942874,
+      "grad_norm": 0.29600924253463745,
+      "learning_rate": 1.2098765432098765e-05,
+      "loss": 5.4705,
+      "step": 1301
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 0.33899274468421936,
+      "learning_rate": 1.185185185185185e-05,
+      "loss": 5.6694,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8907376594564616,
+      "grad_norm": 0.2075345516204834,
+      "learning_rate": 1.1604938271604937e-05,
+      "loss": 5.3638,
+      "step": 1303
+    },
+    {
+      "epoch": 2.8929561841375486,
+      "grad_norm": 0.30049288272857666,
+      "learning_rate": 1.1358024691358023e-05,
+      "loss": 5.4466,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "grad_norm": 0.371520459651947,
+      "learning_rate": 1.111111111111111e-05,
+      "loss": 5.5593,
+      "step": 1305
+    },
+    {
+      "epoch": 2.8973932334997228,
+      "grad_norm": 0.2759955823421478,
+      "learning_rate": 1.0864197530864197e-05,
+      "loss": 5.6186,
+      "step": 1306
+    },
+    {
+      "epoch": 2.89961175818081,
+      "grad_norm": 0.2611229121685028,
+      "learning_rate": 1.0617283950617282e-05,
+      "loss": 5.6353,
+      "step": 1307
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 0.24202197790145874,
+      "learning_rate": 1.0370370370370368e-05,
+      "loss": 5.7467,
+      "step": 1308
+    },
+    {
+      "epoch": 2.904048807542984,
+      "grad_norm": 0.33226239681243896,
+      "learning_rate": 1.0123456790123457e-05,
+      "loss": 5.4022,
+      "step": 1309
+    },
+    {
+      "epoch": 2.906267332224071,
+      "grad_norm": 0.47180378437042236,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 5.4317,
+      "step": 1310
+    },
+    {
+      "epoch": 2.908485856905158,
+      "grad_norm": 0.3268556296825409,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 5.7161,
+      "step": 1311
+    },
+    {
+      "epoch": 2.9107043815862452,
+      "grad_norm": 0.42263200879096985,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 5.573,
+      "step": 1312
+    },
+    {
+      "epoch": 2.9129229062673323,
+      "grad_norm": 0.35489287972450256,
+      "learning_rate": 9.135802469135802e-06,
+      "loss": 5.3497,
+      "step": 1313
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 0.27711161971092224,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 5.3444,
+      "step": 1314
+    },
+    {
+      "epoch": 2.9173599556295065,
+      "grad_norm": 0.2760302722454071,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 5.5586,
+      "step": 1315
+    },
+    {
+      "epoch": 2.9195784803105935,
+      "grad_norm": 0.3397156000137329,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 5.8604,
+      "step": 1316
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "grad_norm": 0.4675195813179016,
+      "learning_rate": 8.148148148148147e-06,
+      "loss": 5.476,
+      "step": 1317
+    },
+    {
+      "epoch": 2.9240155296727677,
+      "grad_norm": 0.3612266182899475,
+      "learning_rate": 7.901234567901233e-06,
+      "loss": 5.6963,
+      "step": 1318
+    },
+    {
+      "epoch": 2.9262340543538548,
+      "grad_norm": 0.24482806026935577,
+      "learning_rate": 7.65432098765432e-06,
+      "loss": 5.7758,
+      "step": 1319
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 0.2885003685951233,
+      "learning_rate": 7.407407407407407e-06,
+      "loss": 5.8167,
+      "step": 1320
+    },
+    {
+      "epoch": 2.930671103716029,
+      "grad_norm": 0.2528499662876129,
+      "learning_rate": 7.160493827160493e-06,
+      "loss": 5.634,
+      "step": 1321
+    },
+    {
+      "epoch": 2.932889628397116,
+      "grad_norm": 0.3025006949901581,
+      "learning_rate": 6.91358024691358e-06,
+      "loss": 5.6498,
+      "step": 1322
+    },
+    {
+      "epoch": 2.935108153078203,
+      "grad_norm": 0.3571071922779083,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 5.4741,
+      "step": 1323
+    },
+    {
+      "epoch": 2.93732667775929,
+      "grad_norm": 0.2797204852104187,
+      "learning_rate": 6.4197530864197525e-06,
+      "loss": 5.5532,
+      "step": 1324
+    },
+    {
+      "epoch": 2.939545202440377,
+      "grad_norm": 0.3282558023929596,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 5.7961,
+      "step": 1325
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 0.2982020378112793,
+      "learning_rate": 5.925925925925925e-06,
+      "loss": 5.7198,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9439822518025514,
+      "grad_norm": 0.33854547142982483,
+      "learning_rate": 5.679012345679012e-06,
+      "loss": 5.6293,
+      "step": 1327
+    },
+    {
+      "epoch": 2.9462007764836384,
+      "grad_norm": 0.30710917711257935,
+      "learning_rate": 5.432098765432098e-06,
+      "loss": 5.8276,
+      "step": 1328
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "grad_norm": 0.28382933139801025,
+      "learning_rate": 5.185185185185184e-06,
+      "loss": 5.461,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9506378258458126,
+      "grad_norm": 0.3094613552093506,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 5.7462,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9528563505268997,
+      "grad_norm": 0.2754240334033966,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 5.9546,
+      "step": 1331
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 0.30117082595825195,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 5.7989,
+      "step": 1332
+    },
+    {
+      "epoch": 2.957293399889074,
+      "grad_norm": 0.3703874349594116,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 5.502,
+      "step": 1333
+    },
+    {
+      "epoch": 2.959511924570161,
+      "grad_norm": 0.2984669506549835,
+      "learning_rate": 3.950617283950617e-06,
+      "loss": 5.5666,
+      "step": 1334
+    },
+    {
+      "epoch": 2.961730449251248,
+      "grad_norm": 0.2734062075614929,
+      "learning_rate": 3.7037037037037033e-06,
+      "loss": 5.5416,
+      "step": 1335
+    },
+    {
+      "epoch": 2.963948973932335,
+      "grad_norm": 0.2687673568725586,
+      "learning_rate": 3.45679012345679e-06,
+      "loss": 5.5845,
+      "step": 1336
+    },
+    {
+      "epoch": 2.966167498613422,
+      "grad_norm": 0.29137223958969116,
+      "learning_rate": 3.2098765432098762e-06,
+      "loss": 5.43,
+      "step": 1337
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 0.31121063232421875,
+      "learning_rate": 2.9629629629629625e-06,
+      "loss": 5.7341,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9706045479755963,
+      "grad_norm": 0.3627428412437439,
+      "learning_rate": 2.716049382716049e-06,
+      "loss": 5.8749,
+      "step": 1339
+    },
+    {
+      "epoch": 2.9728230726566833,
+      "grad_norm": 0.29423636198043823,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 5.4453,
+      "step": 1340
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "grad_norm": 0.4342155158519745,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 5.4307,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9772601220188575,
+      "grad_norm": 0.38601741194725037,
+      "learning_rate": 1.9753086419753083e-06,
+      "loss": 5.6417,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9794786466999446,
+      "grad_norm": 0.23738257586956024,
+      "learning_rate": 1.728395061728395e-06,
+      "loss": 5.7321,
+      "step": 1343
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 0.2786853015422821,
+      "learning_rate": 1.4814814814814812e-06,
+      "loss": 5.6749,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9839156960621187,
+      "grad_norm": 0.3341871500015259,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 5.8431,
+      "step": 1345
+    },
+    {
+      "epoch": 2.986134220743206,
+      "grad_norm": 0.5152813792228699,
+      "learning_rate": 9.876543209876542e-07,
+      "loss": 5.8223,
+      "step": 1346
+    },
+    {
+      "epoch": 2.988352745424293,
+      "grad_norm": 0.24548043310642242,
+      "learning_rate": 7.407407407407406e-07,
+      "loss": 5.4762,
+      "step": 1347
+    },
+    {
+      "epoch": 2.99057127010538,
+      "grad_norm": 0.27913621068000793,
+      "learning_rate": 4.938271604938271e-07,
+      "loss": 5.7996,
+      "step": 1348
+    },
+    {
+      "epoch": 2.992789794786467,
+      "grad_norm": 0.2954283654689789,
+      "learning_rate": 2.4691358024691354e-07,
+      "loss": 5.6713,
+      "step": 1349
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 0.3164568841457367,
+      "learning_rate": 0.0,
+      "loss": 5.6531,
+      "step": 1350
+    },
+    {
+      "epoch": 2.995008319467554,
+      "step": 1350,
+      "total_flos": 6.956724734918656e+16,
+      "train_loss": 6.468862925282231,
+      "train_runtime": 1569.439,
+      "train_samples_per_second": 13.784,
+      "train_steps_per_second": 0.86
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.956724734918656e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}