diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6511 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 924,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004329004329004329,
+      "grad_norm": 111.94686126708984,
+      "learning_rate": 0.0,
+      "loss": 5.7188,
+      "step": 1
+    },
+    {
+      "epoch": 0.008658008658008658,
+      "grad_norm": 32.00572967529297,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 5.7812,
+      "step": 2
+    },
+    {
+      "epoch": 0.012987012987012988,
+      "grad_norm": 32.85079574584961,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 5.7188,
+      "step": 3
+    },
+    {
+      "epoch": 0.017316017316017316,
+      "grad_norm": 95.10060119628906,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 5.5,
+      "step": 4
+    },
+    {
+      "epoch": 0.021645021645021644,
+      "grad_norm": 33.53082275390625,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 5.125,
+      "step": 5
+    },
+    {
+      "epoch": 0.025974025974025976,
+      "grad_norm": 46.92396926879883,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 4.5625,
+      "step": 6
+    },
+    {
+      "epoch": 0.030303030303030304,
+      "grad_norm": 52.68779754638672,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 4.0312,
+      "step": 7
+    },
+    {
+      "epoch": 0.03463203463203463,
+      "grad_norm": 46.13388442993164,
+      "learning_rate": 2.5e-05,
+      "loss": 3.6562,
+      "step": 8
+    },
+    {
+      "epoch": 0.03896103896103896,
+      "grad_norm": 81.87711334228516,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 3.4844,
+      "step": 9
+    },
+    {
+      "epoch": 0.04329004329004329,
+      "grad_norm": 58.840816497802734,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 3.0781,
+      "step": 10
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 55.59412384033203,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 2.7031,
+      "step": 11
+    },
+    {
+      "epoch": 0.05194805194805195,
+      "grad_norm": 34.48457717895508,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 2.7188,
+      "step": 12
+    },
+    {
+      "epoch": 0.05627705627705628,
+      "grad_norm": 39.92266082763672,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 2.5938,
+      "step": 13
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 39.542476654052734,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 2.4688,
+      "step": 14
+    },
+    {
+      "epoch": 0.06493506493506493,
+      "grad_norm": 32.79098129272461,
+      "learning_rate": 5e-05,
+      "loss": 2.4844,
+      "step": 15
+    },
+    {
+      "epoch": 0.06926406926406926,
+      "grad_norm": 32.08797073364258,
+      "learning_rate": 5.3571428571428575e-05,
+      "loss": 2.3438,
+      "step": 16
+    },
+    {
+      "epoch": 0.0735930735930736,
+      "grad_norm": 33.02616882324219,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 2.3594,
+      "step": 17
+    },
+    {
+      "epoch": 0.07792207792207792,
+      "grad_norm": 20.7552490234375,
+      "learning_rate": 6.0714285714285715e-05,
+      "loss": 2.1719,
+      "step": 18
+    },
+    {
+      "epoch": 0.08225108225108226,
+      "grad_norm": 79.99295043945312,
+      "learning_rate": 6.428571428571429e-05,
+      "loss": 2.2344,
+      "step": 19
+    },
+    {
+      "epoch": 0.08658008658008658,
+      "grad_norm": 31.502910614013672,
+      "learning_rate": 6.785714285714286e-05,
+      "loss": 1.9609,
+      "step": 20
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 11.919739723205566,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 1.9531,
+      "step": 21
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 21.967390060424805,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.9141,
+      "step": 22
+    },
+    {
+      "epoch": 0.09956709956709957,
+      "grad_norm": 13.574630737304688,
+      "learning_rate": 7.857142857142858e-05,
+      "loss": 1.8906,
+      "step": 23
+    },
+    {
+      "epoch": 0.1038961038961039,
+      "grad_norm": 15.675418853759766,
+      "learning_rate": 8.214285714285714e-05,
+      "loss": 1.7578,
+      "step": 24
+    },
+    {
+      "epoch": 0.10822510822510822,
+      "grad_norm": 12.390951156616211,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 1.7578,
+      "step": 25
+    },
+    {
+      "epoch": 0.11255411255411256,
+      "grad_norm": 14.530952453613281,
+      "learning_rate": 8.92857142857143e-05,
+      "loss": 1.6016,
+      "step": 26
+    },
+    {
+      "epoch": 0.11688311688311688,
+      "grad_norm": 7.964989185333252,
+      "learning_rate": 9.285714285714286e-05,
+      "loss": 1.4141,
+      "step": 27
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 9.956165313720703,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 1.4609,
+      "step": 28
+    },
+    {
+      "epoch": 0.12554112554112554,
+      "grad_norm": 7.215725421905518,
+      "learning_rate": 0.0001,
+      "loss": 1.5,
+      "step": 29
+    },
+    {
+      "epoch": 0.12987012987012986,
+      "grad_norm": 10.029833793640137,
+      "learning_rate": 9.99996926570257e-05,
+      "loss": 1.625,
+      "step": 30
+    },
+    {
+      "epoch": 0.1341991341991342,
+      "grad_norm": 9.986540794372559,
+      "learning_rate": 9.999877063188124e-05,
+      "loss": 1.5703,
+      "step": 31
+    },
+    {
+      "epoch": 0.13852813852813853,
+      "grad_norm": 7.54625129699707,
+      "learning_rate": 9.999723393590168e-05,
+      "loss": 1.3906,
+      "step": 32
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 10.881644248962402,
+      "learning_rate": 9.999508258797877e-05,
+      "loss": 1.4141,
+      "step": 33
+    },
+    {
+      "epoch": 0.1471861471861472,
+      "grad_norm": 7.18023681640625,
+      "learning_rate": 9.999231661456054e-05,
+      "loss": 1.3906,
+      "step": 34
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 7.544717311859131,
+      "learning_rate": 9.998893604965111e-05,
+      "loss": 1.3828,
+      "step": 35
+    },
+    {
+      "epoch": 0.15584415584415584,
+      "grad_norm": 19.247713088989258,
+      "learning_rate": 9.998494093481021e-05,
+      "loss": 1.3125,
+      "step": 36
+    },
+    {
+      "epoch": 0.16017316017316016,
+      "grad_norm": 30.700363159179688,
+      "learning_rate": 9.998033131915266e-05,
+      "loss": 1.4219,
+      "step": 37
+    },
+    {
+      "epoch": 0.1645021645021645,
+      "grad_norm": 16.199434280395508,
+      "learning_rate": 9.997510725934773e-05,
+      "loss": 1.2422,
+      "step": 38
+    },
+    {
+      "epoch": 0.16883116883116883,
+      "grad_norm": 5.880858898162842,
+      "learning_rate": 9.996926881961862e-05,
+      "loss": 1.3516,
+      "step": 39
+    },
+    {
+      "epoch": 0.17316017316017315,
+      "grad_norm": 7.811530113220215,
+      "learning_rate": 9.996281607174138e-05,
+      "loss": 1.2266,
+      "step": 40
+    },
+    {
+      "epoch": 0.1774891774891775,
+      "grad_norm": 10.326518058776855,
+      "learning_rate": 9.995574909504435e-05,
+      "loss": 1.3281,
+      "step": 41
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 20.549381256103516,
+      "learning_rate": 9.994806797640692e-05,
+      "loss": 1.2109,
+      "step": 42
+    },
+    {
+      "epoch": 0.18614718614718614,
+      "grad_norm": 5.703456878662109,
+      "learning_rate": 9.993977281025862e-05,
+      "loss": 1.3125,
+      "step": 43
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 20.923757553100586,
+      "learning_rate": 9.99308636985779e-05,
+      "loss": 1.3047,
+      "step": 44
+    },
+    {
+      "epoch": 0.19480519480519481,
+      "grad_norm": 14.915127754211426,
+      "learning_rate": 9.992134075089084e-05,
+      "loss": 1.1641,
+      "step": 45
+    },
+    {
+      "epoch": 0.19913419913419914,
+      "grad_norm": 6.1190290451049805,
+      "learning_rate": 9.991120408426991e-05,
+      "loss": 1.2188,
+      "step": 46
+    },
+    {
+      "epoch": 0.20346320346320346,
+      "grad_norm": 7.0726189613342285,
+      "learning_rate": 9.990045382333242e-05,
+      "loss": 1.2578,
+      "step": 47
+    },
+    {
+      "epoch": 0.2077922077922078,
+      "grad_norm": 7.223331928253174,
+      "learning_rate": 9.988909010023908e-05,
+      "loss": 1.1875,
+      "step": 48
+    },
+    {
+      "epoch": 0.21212121212121213,
+      "grad_norm": 7.597274303436279,
+      "learning_rate": 9.987711305469231e-05,
+      "loss": 1.0781,
+      "step": 49
+    },
+    {
+      "epoch": 0.21645021645021645,
+      "grad_norm": 5.60629415512085,
+      "learning_rate": 9.986452283393452e-05,
+      "loss": 1.1484,
+      "step": 50
+    },
+    {
+      "epoch": 0.22077922077922077,
+      "grad_norm": 15.291472434997559,
+      "learning_rate": 9.985131959274637e-05,
+      "loss": 0.9648,
+      "step": 51
+    },
+    {
+      "epoch": 0.22510822510822512,
+      "grad_norm": 5.605519771575928,
+      "learning_rate": 9.983750349344476e-05,
+      "loss": 1.1484,
+      "step": 52
+    },
+    {
+      "epoch": 0.22943722943722944,
+      "grad_norm": 6.039676666259766,
+      "learning_rate": 9.982307470588098e-05,
+      "loss": 1.2031,
+      "step": 53
+    },
+    {
+      "epoch": 0.23376623376623376,
+      "grad_norm": 5.730657577514648,
+      "learning_rate": 9.980803340743843e-05,
+      "loss": 1.1172,
+      "step": 54
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 4.605213642120361,
+      "learning_rate": 9.979237978303066e-05,
+      "loss": 1.0312,
+      "step": 55
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 9.674660682678223,
+      "learning_rate": 9.977611402509892e-05,
+      "loss": 1.1172,
+      "step": 56
+    },
+    {
+      "epoch": 0.24675324675324675,
+      "grad_norm": 7.661512851715088,
+      "learning_rate": 9.975923633360985e-05,
+      "loss": 1.1172,
+      "step": 57
+    },
+    {
+      "epoch": 0.2510822510822511,
+      "grad_norm": 5.796794414520264,
+      "learning_rate": 9.974174691605305e-05,
+      "loss": 1.1641,
+      "step": 58
+    },
+    {
+      "epoch": 0.2554112554112554,
+      "grad_norm": 5.243800163269043,
+      "learning_rate": 9.97236459874385e-05,
+      "loss": 0.9688,
+      "step": 59
+    },
+    {
+      "epoch": 0.2597402597402597,
+      "grad_norm": 5.435077667236328,
+      "learning_rate": 9.970493377029396e-05,
+      "loss": 1.0156,
+      "step": 60
+    },
+    {
+      "epoch": 0.26406926406926406,
+      "grad_norm": 4.097263813018799,
+      "learning_rate": 9.968561049466214e-05,
+      "loss": 1.1172,
+      "step": 61
+    },
+    {
+      "epoch": 0.2683982683982684,
+      "grad_norm": 4.291856288909912,
+      "learning_rate": 9.966567639809795e-05,
+      "loss": 1.1719,
+      "step": 62
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 4.458070755004883,
+      "learning_rate": 9.964513172566561e-05,
+      "loss": 1.0312,
+      "step": 63
+    },
+    {
+      "epoch": 0.27705627705627706,
+      "grad_norm": 5.127298355102539,
+      "learning_rate": 9.96239767299355e-05,
+      "loss": 0.8945,
+      "step": 64
+    },
+    {
+      "epoch": 0.2813852813852814,
+      "grad_norm": 5.836466312408447,
+      "learning_rate": 9.960221167098124e-05,
+      "loss": 1.0312,
+      "step": 65
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 4.4748053550720215,
+      "learning_rate": 9.957983681637631e-05,
+      "loss": 1.0312,
+      "step": 66
+    },
+    {
+      "epoch": 0.29004329004329005,
+      "grad_norm": 4.5356950759887695,
+      "learning_rate": 9.955685244119091e-05,
+      "loss": 1.0391,
+      "step": 67
+    },
+    {
+      "epoch": 0.2943722943722944,
+      "grad_norm": 10.683904647827148,
+      "learning_rate": 9.95332588279885e-05,
+      "loss": 1.1172,
+      "step": 68
+    },
+    {
+      "epoch": 0.2987012987012987,
+      "grad_norm": 4.90685510635376,
+      "learning_rate": 9.950905626682228e-05,
+      "loss": 1.0625,
+      "step": 69
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 5.8771891593933105,
+      "learning_rate": 9.948424505523179e-05,
+      "loss": 1.0859,
+      "step": 70
+    },
+    {
+      "epoch": 0.30735930735930733,
+      "grad_norm": 3.343855857849121,
+      "learning_rate": 9.945882549823906e-05,
+      "loss": 1.0156,
+      "step": 71
+    },
+    {
+      "epoch": 0.3116883116883117,
+      "grad_norm": 3.8755388259887695,
+      "learning_rate": 9.943279790834499e-05,
+      "loss": 0.9609,
+      "step": 72
+    },
+    {
+      "epoch": 0.31601731601731603,
+      "grad_norm": 4.613266944885254,
+      "learning_rate": 9.940616260552544e-05,
+      "loss": 1.0,
+      "step": 73
+    },
+    {
+      "epoch": 0.3203463203463203,
+      "grad_norm": 8.878193855285645,
+      "learning_rate": 9.937891991722736e-05,
+      "loss": 1.125,
+      "step": 74
+    },
+    {
+      "epoch": 0.3246753246753247,
+      "grad_norm": 3.524160861968994,
+      "learning_rate": 9.935107017836471e-05,
+      "loss": 0.9375,
+      "step": 75
+    },
+    {
+      "epoch": 0.329004329004329,
+      "grad_norm": 6.691828727722168,
+      "learning_rate": 9.932261373131433e-05,
+      "loss": 0.9219,
+      "step": 76
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 4.184720516204834,
+      "learning_rate": 9.92935509259118e-05,
+      "loss": 1.0859,
+      "step": 77
+    },
+    {
+      "epoch": 0.33766233766233766,
+      "grad_norm": 4.936676502227783,
+      "learning_rate": 9.926388211944706e-05,
+      "loss": 0.8711,
+      "step": 78
+    },
+    {
+      "epoch": 0.341991341991342,
+      "grad_norm": 5.211480617523193,
+      "learning_rate": 9.923360767666011e-05,
+      "loss": 0.9453,
+      "step": 79
+    },
+    {
+      "epoch": 0.3463203463203463,
+      "grad_norm": 4.066558361053467,
+      "learning_rate": 9.920272796973642e-05,
+      "loss": 1.1094,
+      "step": 80
+    },
+    {
+      "epoch": 0.35064935064935066,
+      "grad_norm": 4.604328155517578,
+      "learning_rate": 9.917124337830243e-05,
+      "loss": 1.0859,
+      "step": 81
+    },
+    {
+      "epoch": 0.354978354978355,
+      "grad_norm": 6.513655185699463,
+      "learning_rate": 9.913915428942087e-05,
+      "loss": 1.0547,
+      "step": 82
+    },
+    {
+      "epoch": 0.3593073593073593,
+      "grad_norm": 8.053498268127441,
+      "learning_rate": 9.910646109758596e-05,
+      "loss": 1.0234,
+      "step": 83
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 5.447996139526367,
+      "learning_rate": 9.907316420471863e-05,
+      "loss": 1.0859,
+      "step": 84
+    },
+    {
+      "epoch": 0.36796536796536794,
+      "grad_norm": 8.442547798156738,
+      "learning_rate": 9.903926402016153e-05,
+      "loss": 0.8828,
+      "step": 85
+    },
+    {
+      "epoch": 0.3722943722943723,
+      "grad_norm": 8.764233589172363,
+      "learning_rate": 9.900476096067398e-05,
+      "loss": 0.9688,
+      "step": 86
+    },
+    {
+      "epoch": 0.37662337662337664,
+      "grad_norm": 6.131944179534912,
+      "learning_rate": 9.89696554504269e-05,
+      "loss": 1.0,
+      "step": 87
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 5.368701934814453,
+      "learning_rate": 9.893394792099762e-05,
+      "loss": 0.9766,
+      "step": 88
+    },
+    {
+      "epoch": 0.3852813852813853,
+      "grad_norm": 7.1486711502075195,
+      "learning_rate": 9.889763881136439e-05,
+      "loss": 1.0938,
+      "step": 89
+    },
+    {
+      "epoch": 0.38961038961038963,
+      "grad_norm": 7.579170227050781,
+      "learning_rate": 9.886072856790125e-05,
+      "loss": 1.0078,
+      "step": 90
+    },
+    {
+      "epoch": 0.3939393939393939,
+      "grad_norm": 11.445550918579102,
+      "learning_rate": 9.882321764437234e-05,
+      "loss": 1.0859,
+      "step": 91
+    },
+    {
+      "epoch": 0.39826839826839827,
+      "grad_norm": 12.062833786010742,
+      "learning_rate": 9.878510650192643e-05,
+      "loss": 0.8164,
+      "step": 92
+    },
+    {
+      "epoch": 0.4025974025974026,
+      "grad_norm": 6.603275299072266,
+      "learning_rate": 9.874639560909117e-05,
+      "loss": 1.1016,
+      "step": 93
+    },
+    {
+      "epoch": 0.4069264069264069,
+      "grad_norm": 12.185580253601074,
+      "learning_rate": 9.870708544176745e-05,
+      "loss": 0.9453,
+      "step": 94
+    },
+    {
+      "epoch": 0.41125541125541126,
+      "grad_norm": 6.9988250732421875,
+      "learning_rate": 9.866717648322335e-05,
+      "loss": 0.9766,
+      "step": 95
+    },
+    {
+      "epoch": 0.4155844155844156,
+      "grad_norm": 5.269665241241455,
+      "learning_rate": 9.862666922408843e-05,
+      "loss": 0.9961,
+      "step": 96
+    },
+    {
+      "epoch": 0.4199134199134199,
+      "grad_norm": 4.827775478363037,
+      "learning_rate": 9.858556416234755e-05,
+      "loss": 0.918,
+      "step": 97
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 6.160229206085205,
+      "learning_rate": 9.854386180333479e-05,
+      "loss": 1.0,
+      "step": 98
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 7.695404052734375,
+      "learning_rate": 9.850156265972721e-05,
+      "loss": 0.9727,
+      "step": 99
+    },
+    {
+      "epoch": 0.4329004329004329,
+      "grad_norm": 7.208751201629639,
+      "learning_rate": 9.84586672515386e-05,
+      "loss": 0.9844,
+      "step": 100
+    },
+    {
+      "epoch": 0.43722943722943725,
+      "grad_norm": 4.856283664703369,
+      "learning_rate": 9.841517610611309e-05,
+      "loss": 0.9688,
+      "step": 101
+    },
+    {
+      "epoch": 0.44155844155844154,
+      "grad_norm": 4.092401504516602,
+      "learning_rate": 9.837108975811854e-05,
+      "loss": 0.9062,
+      "step": 102
+    },
+    {
+      "epoch": 0.4458874458874459,
+      "grad_norm": 4.5670623779296875,
+      "learning_rate": 9.832640874954016e-05,
+      "loss": 1.0156,
+      "step": 103
+    },
+    {
+      "epoch": 0.45021645021645024,
+      "grad_norm": 7.118216037750244,
+      "learning_rate": 9.828113362967372e-05,
+      "loss": 1.0938,
+      "step": 104
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 5.383735179901123,
+      "learning_rate": 9.82352649551188e-05,
+      "loss": 1.0547,
+      "step": 105
+    },
+    {
+      "epoch": 0.4588744588744589,
+      "grad_norm": 4.898381233215332,
+      "learning_rate": 9.8188803289772e-05,
+      "loss": 0.9062,
+      "step": 106
+    },
+    {
+      "epoch": 0.46320346320346323,
+      "grad_norm": 5.296220779418945,
+      "learning_rate": 9.814174920481998e-05,
+      "loss": 1.0156,
+      "step": 107
+    },
+    {
+      "epoch": 0.4675324675324675,
+      "grad_norm": 6.022840976715088,
+      "learning_rate": 9.809410327873244e-05,
+      "loss": 0.9883,
+      "step": 108
+    },
+    {
+      "epoch": 0.47186147186147187,
+      "grad_norm": 4.332733631134033,
+      "learning_rate": 9.804586609725499e-05,
+      "loss": 0.9297,
+      "step": 109
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 6.02630615234375,
+      "learning_rate": 9.799703825340198e-05,
+      "loss": 0.8789,
+      "step": 110
+    },
+    {
+      "epoch": 0.4805194805194805,
+      "grad_norm": 6.509529113769531,
+      "learning_rate": 9.794762034744925e-05,
+      "loss": 1.0703,
+      "step": 111
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 5.087254047393799,
+      "learning_rate": 9.789761298692657e-05,
+      "loss": 0.9141,
+      "step": 112
+    },
+    {
+      "epoch": 0.48917748917748916,
+      "grad_norm": 4.235925197601318,
+      "learning_rate": 9.784701678661045e-05,
+      "loss": 0.8672,
+      "step": 113
+    },
+    {
+      "epoch": 0.4935064935064935,
+      "grad_norm": 5.100911617279053,
+      "learning_rate": 9.779583236851631e-05,
+      "loss": 0.9336,
+      "step": 114
+    },
+    {
+      "epoch": 0.49783549783549785,
+      "grad_norm": 4.8926873207092285,
+      "learning_rate": 9.774406036189103e-05,
+      "loss": 0.9258,
+      "step": 115
+    },
+    {
+      "epoch": 0.5021645021645021,
+      "grad_norm": 4.586702823638916,
+      "learning_rate": 9.76917014032051e-05,
+      "loss": 0.9688,
+      "step": 116
+    },
+    {
+      "epoch": 0.5064935064935064,
+      "grad_norm": 4.065001487731934,
+      "learning_rate": 9.763875613614482e-05,
+      "loss": 0.918,
+      "step": 117
+    },
+    {
+      "epoch": 0.5108225108225108,
+      "grad_norm": 3.5030267238616943,
+      "learning_rate": 9.758522521160446e-05,
+      "loss": 0.7461,
+      "step": 118
+    },
+    {
+      "epoch": 0.5151515151515151,
+      "grad_norm": 4.101621150970459,
+      "learning_rate": 9.753110928767815e-05,
+      "loss": 0.8984,
+      "step": 119
+    },
+    {
+      "epoch": 0.5194805194805194,
+      "grad_norm": 5.357971668243408,
+      "learning_rate": 9.747640902965184e-05,
+      "loss": 0.9727,
+      "step": 120
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 3.756002426147461,
+      "learning_rate": 9.742112510999515e-05,
+      "loss": 0.9141,
+      "step": 121
+    },
+    {
+      "epoch": 0.5281385281385281,
+      "grad_norm": 4.6300272941589355,
+      "learning_rate": 9.736525820835303e-05,
+      "loss": 0.9414,
+      "step": 122
+    },
+    {
+      "epoch": 0.5324675324675324,
+      "grad_norm": 5.029032230377197,
+      "learning_rate": 9.730880901153747e-05,
+      "loss": 0.8594,
+      "step": 123
+    },
+    {
+      "epoch": 0.5367965367965368,
+      "grad_norm": 4.181064128875732,
+      "learning_rate": 9.725177821351907e-05,
+      "loss": 1.0156,
+      "step": 124
+    },
+    {
+      "epoch": 0.5411255411255411,
+      "grad_norm": 7.1881794929504395,
+      "learning_rate": 9.719416651541839e-05,
+      "loss": 0.9531,
+      "step": 125
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 5.557502269744873,
+      "learning_rate": 9.713597462549747e-05,
+      "loss": 0.8789,
+      "step": 126
+    },
+    {
+      "epoch": 0.5497835497835498,
+      "grad_norm": 5.64065408706665,
+      "learning_rate": 9.707720325915104e-05,
+      "loss": 0.9062,
+      "step": 127
+    },
+    {
+      "epoch": 0.5541125541125541,
+      "grad_norm": 5.925412178039551,
+      "learning_rate": 9.70178531388978e-05,
+      "loss": 0.8203,
+      "step": 128
+    },
+    {
+      "epoch": 0.5584415584415584,
+      "grad_norm": 4.839400768280029,
+      "learning_rate": 9.69579249943714e-05,
+      "loss": 0.9805,
+      "step": 129
+    },
+    {
+      "epoch": 0.5627705627705628,
+      "grad_norm": 4.0864739418029785,
+      "learning_rate": 9.689741956231163e-05,
+      "loss": 0.8516,
+      "step": 130
+    },
+    {
+      "epoch": 0.5670995670995671,
+      "grad_norm": 9.55315113067627,
+      "learning_rate": 9.683633758655529e-05,
+      "loss": 0.9258,
+      "step": 131
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 5.017940521240234,
+      "learning_rate": 9.677467981802696e-05,
+      "loss": 0.8125,
+      "step": 132
+    },
+    {
+      "epoch": 0.5757575757575758,
+      "grad_norm": 4.608086585998535,
+      "learning_rate": 9.671244701472999e-05,
+      "loss": 0.7656,
+      "step": 133
+    },
+    {
+      "epoch": 0.5800865800865801,
+      "grad_norm": 7.785727500915527,
+      "learning_rate": 9.664963994173695e-05,
+      "loss": 0.9375,
+      "step": 134
+    },
+    {
+      "epoch": 0.5844155844155844,
+      "grad_norm": 5.264962196350098,
+      "learning_rate": 9.658625937118034e-05,
+      "loss": 0.7969,
+      "step": 135
+    },
+    {
+      "epoch": 0.5887445887445888,
+      "grad_norm": 5.819239616394043,
+      "learning_rate": 9.652230608224307e-05,
+      "loss": 0.8242,
+      "step": 136
+    },
+    {
+      "epoch": 0.5930735930735931,
+      "grad_norm": 5.409610271453857,
+      "learning_rate": 9.645778086114892e-05,
+      "loss": 0.9297,
+      "step": 137
+    },
+    {
+      "epoch": 0.5974025974025974,
+      "grad_norm": 4.992769718170166,
+      "learning_rate": 9.639268450115282e-05,
+      "loss": 0.9648,
+      "step": 138
+    },
+    {
+      "epoch": 0.6017316017316018,
+      "grad_norm": 4.017268657684326,
+      "learning_rate": 9.632701780253111e-05,
+      "loss": 0.9141,
+      "step": 139
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 4.431084156036377,
+      "learning_rate": 9.626078157257176e-05,
+      "loss": 0.9375,
+      "step": 140
+    },
+    {
+      "epoch": 0.6103896103896104,
+      "grad_norm": 8.311599731445312,
+      "learning_rate": 9.619397662556435e-05,
+      "loss": 0.9336,
+      "step": 141
+    },
+    {
+      "epoch": 0.6147186147186147,
+      "grad_norm": 4.197081089019775,
+      "learning_rate": 9.612660378279012e-05,
+      "loss": 0.9375,
+      "step": 142
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 6.389054775238037,
+      "learning_rate": 9.605866387251186e-05,
+      "loss": 0.9062,
+      "step": 143
+    },
+    {
+      "epoch": 0.6233766233766234,
+      "grad_norm": 5.9323248863220215,
+      "learning_rate": 9.599015772996375e-05,
+      "loss": 0.8945,
+      "step": 144
+    },
+    {
+      "epoch": 0.6277056277056277,
+      "grad_norm": 3.7743008136749268,
+      "learning_rate": 9.592108619734106e-05,
+      "loss": 0.8555,
+      "step": 145
+    },
+    {
+      "epoch": 0.6320346320346321,
+      "grad_norm": 4.603461742401123,
+      "learning_rate": 9.585145012378978e-05,
+      "loss": 0.9141,
+      "step": 146
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 5.491341590881348,
+      "learning_rate": 9.578125036539624e-05,
+      "loss": 0.8633,
+      "step": 147
+    },
+    {
+      "epoch": 0.6406926406926406,
+      "grad_norm": 8.107928276062012,
+      "learning_rate": 9.571048778517654e-05,
+      "loss": 0.8672,
+      "step": 148
+    },
+    {
+      "epoch": 0.645021645021645,
+      "grad_norm": 5.107354164123535,
+      "learning_rate": 9.563916325306594e-05,
+      "loss": 0.9336,
+      "step": 149
+    },
+    {
+      "epoch": 0.6493506493506493,
+      "grad_norm": 4.870626926422119,
+      "learning_rate": 9.556727764590823e-05,
+      "loss": 0.9492,
+      "step": 150
+    },
+    {
+      "epoch": 0.6536796536796536,
+      "grad_norm": 4.701653003692627,
+      "learning_rate": 9.549483184744483e-05,
+      "loss": 1.0,
+      "step": 151
+    },
+    {
+      "epoch": 0.658008658008658,
+      "grad_norm": 5.646700859069824,
+      "learning_rate": 9.542182674830405e-05,
+      "loss": 0.875,
+      "step": 152
+    },
+    {
+      "epoch": 0.6623376623376623,
+      "grad_norm": 4.344261646270752,
+      "learning_rate": 9.534826324599003e-05,
+      "loss": 0.7695,
+      "step": 153
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 4.080512046813965,
+      "learning_rate": 9.527414224487182e-05,
+      "loss": 0.8711,
+      "step": 154
+    },
+    {
+      "epoch": 0.670995670995671,
+      "grad_norm": 4.235628604888916,
+      "learning_rate": 9.519946465617218e-05,
+      "loss": 0.8828,
+      "step": 155
+    },
+    {
+      "epoch": 0.6753246753246753,
+      "grad_norm": 5.588692665100098,
+      "learning_rate": 9.512423139795637e-05,
+      "loss": 0.918,
+      "step": 156
+    },
+    {
+      "epoch": 0.6796536796536796,
+      "grad_norm": 3.6299853324890137,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 0.8789,
+      "step": 157
+    },
+    {
+      "epoch": 0.683982683982684,
+      "grad_norm": 3.874685764312744,
+      "learning_rate": 9.497210157938234e-05,
+      "loss": 0.9297,
+      "step": 158
+    },
+    {
+      "epoch": 0.6883116883116883,
+      "grad_norm": 7.115217685699463,
+      "learning_rate": 9.489520688926534e-05,
+      "loss": 0.7695,
+      "step": 159
+    },
+    {
+      "epoch": 0.6926406926406926,
+      "grad_norm": 4.878574371337891,
+      "learning_rate": 9.481776027009167e-05,
+      "loss": 0.8008,
+      "step": 160
+    },
+    {
+      "epoch": 0.696969696969697,
+      "grad_norm": 6.1356682777404785,
+      "learning_rate": 9.473976267396831e-05,
+      "loss": 0.7461,
+      "step": 161
+    },
+    {
+      "epoch": 0.7012987012987013,
+      "grad_norm": 5.803933143615723,
+      "learning_rate": 9.466121505977577e-05,
+      "loss": 0.8633,
+      "step": 162
+    },
+    {
+      "epoch": 0.7056277056277056,
+      "grad_norm": 6.888575077056885,
+      "learning_rate": 9.458211839315636e-05,
+      "loss": 0.8086,
+      "step": 163
+    },
+    {
+      "epoch": 0.70995670995671,
+      "grad_norm": 10.345001220703125,
+      "learning_rate": 9.450247364650227e-05,
+      "loss": 0.8867,
+      "step": 164
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 8.069884300231934,
+      "learning_rate": 9.442228179894362e-05,
+      "loss": 0.8984,
+      "step": 165
+    },
+    {
+      "epoch": 0.7186147186147186,
+      "grad_norm": 5.542285442352295,
+      "learning_rate": 9.434154383633646e-05,
+      "loss": 0.8477,
+      "step": 166
+    },
+    {
+      "epoch": 0.7229437229437229,
+      "grad_norm": 8.194671630859375,
+      "learning_rate": 9.426026075125061e-05,
+      "loss": 0.8281,
+      "step": 167
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 8.291105270385742,
+      "learning_rate": 9.417843354295748e-05,
+      "loss": 0.9062,
+      "step": 168
+    },
+    {
+      "epoch": 0.7316017316017316,
+      "grad_norm": 14.5394868850708,
+      "learning_rate": 9.409606321741775e-05,
+      "loss": 0.9531,
+      "step": 169
+    },
+    {
+      "epoch": 0.7359307359307359,
+      "grad_norm": 6.165283679962158,
+      "learning_rate": 9.401315078726909e-05,
+      "loss": 0.832,
+      "step": 170
+    },
+    {
+      "epoch": 0.7402597402597403,
+      "grad_norm": 5.3768630027771,
+      "learning_rate": 9.392969727181359e-05,
+      "loss": 0.8555,
+      "step": 171
+    },
+    {
+      "epoch": 0.7445887445887446,
+      "grad_norm": 6.443775653839111,
+      "learning_rate": 9.384570369700531e-05,
+      "loss": 0.918,
+      "step": 172
+    },
+    {
+      "epoch": 0.7489177489177489,
+      "grad_norm": 7.081465721130371,
+      "learning_rate": 9.376117109543769e-05,
+      "loss": 0.9531,
+      "step": 173
+    },
+    {
+      "epoch": 0.7532467532467533,
+      "grad_norm": 6.45358419418335,
+      "learning_rate": 9.367610050633075e-05,
+      "loss": 0.8242,
+      "step": 174
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 8.647873878479004,
+      "learning_rate": 9.35904929755184e-05,
+      "loss": 0.9375,
+      "step": 175
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 11.034056663513184,
+      "learning_rate": 9.350434955543557e-05,
+      "loss": 0.832,
+      "step": 176
+    },
+    {
+      "epoch": 0.7662337662337663,
+      "grad_norm": 6.094070911407471,
+      "learning_rate": 9.341767130510528e-05,
+      "loss": 0.7656,
+      "step": 177
+    },
+    {
+      "epoch": 0.7705627705627706,
+      "grad_norm": 8.331881523132324,
+      "learning_rate": 9.333045929012556e-05,
+      "loss": 0.8789,
+      "step": 178
+    },
+    {
+      "epoch": 0.7748917748917749,
+      "grad_norm": 5.476624965667725,
+      "learning_rate": 9.32427145826564e-05,
+      "loss": 0.8203,
+      "step": 179
+    },
+    {
+      "epoch": 0.7792207792207793,
+      "grad_norm": 7.183038711547852,
+      "learning_rate": 9.315443826140661e-05,
+      "loss": 0.9141,
+      "step": 180
+    },
+    {
+      "epoch": 0.7835497835497836,
+      "grad_norm": 5.651857852935791,
+      "learning_rate": 9.306563141162046e-05,
+      "loss": 0.8086,
+      "step": 181
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 4.78335428237915,
+      "learning_rate": 9.297629512506439e-05,
+      "loss": 0.7539,
+      "step": 182
+    },
+    {
+      "epoch": 0.7922077922077922,
+      "grad_norm": 5.953134059906006,
+      "learning_rate": 9.288643050001361e-05,
+      "loss": 0.8359,
+      "step": 183
+    },
+    {
+      "epoch": 0.7965367965367965,
+      "grad_norm": 7.348382949829102,
+      "learning_rate": 9.279603864123858e-05,
+      "loss": 1.0078,
+      "step": 184
+    },
+    {
+      "epoch": 0.8008658008658008,
+      "grad_norm": 6.550226211547852,
+      "learning_rate": 9.270512065999137e-05,
+      "loss": 0.8242,
+      "step": 185
+    },
+    {
+      "epoch": 0.8051948051948052,
+      "grad_norm": 8.682317733764648,
+      "learning_rate": 9.261367767399213e-05,
+      "loss": 0.875,
+      "step": 186
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 12.234009742736816,
+      "learning_rate": 9.252171080741524e-05,
+      "loss": 0.875,
+      "step": 187
+    },
+    {
+      "epoch": 0.8138528138528138,
+      "grad_norm": 6.917994022369385,
+      "learning_rate": 9.242922119087548e-05,
+      "loss": 0.875,
+      "step": 188
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 10.946145057678223,
+      "learning_rate": 9.233620996141421e-05,
+      "loss": 0.8203,
+      "step": 189
+    },
+    {
+      "epoch": 0.8225108225108225,
+      "grad_norm": 15.607405662536621,
+      "learning_rate": 9.224267826248536e-05,
+      "loss": 0.9297,
+      "step": 190
+    },
+    {
+      "epoch": 0.8268398268398268,
+      "grad_norm": 7.996190547943115,
+      "learning_rate": 9.214862724394133e-05,
+      "loss": 0.7969,
+      "step": 191
+    },
+    {
+      "epoch": 0.8311688311688312,
+      "grad_norm": 8.081267356872559,
+      "learning_rate": 9.205405806201895e-05,
+      "loss": 0.9102,
+      "step": 192
+    },
+    {
+      "epoch": 0.8354978354978355,
+      "grad_norm": 6.424978256225586,
+      "learning_rate": 9.195897187932512e-05,
+      "loss": 0.8867,
+      "step": 193
+    },
+    {
+      "epoch": 0.8398268398268398,
+      "grad_norm": 10.088332176208496,
+      "learning_rate": 9.186336986482267e-05,
+      "loss": 0.8711,
+      "step": 194
+    },
+    {
+      "epoch": 0.8441558441558441,
+      "grad_norm": 6.9279093742370605,
+      "learning_rate": 9.176725319381589e-05,
+      "loss": 0.9648,
+      "step": 195
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 13.404962539672852,
+      "learning_rate": 9.167062304793614e-05,
+      "loss": 0.9102,
+      "step": 196
+    },
+    {
+      "epoch": 0.8528138528138528,
+      "grad_norm": 8.118035316467285,
+      "learning_rate": 9.157348061512727e-05,
+      "loss": 1.0234,
+      "step": 197
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 7.562191009521484,
+      "learning_rate": 9.147582708963104e-05,
+      "loss": 0.9375,
+      "step": 198
+    },
+    {
+      "epoch": 0.8614718614718615,
+      "grad_norm": 6.031793117523193,
+      "learning_rate": 9.137766367197246e-05,
+      "loss": 0.9375,
+      "step": 199
+    },
+    {
+      "epoch": 0.8658008658008658,
+      "grad_norm": 6.979994773864746,
+      "learning_rate": 9.1278991568945e-05,
+      "loss": 0.8867,
+      "step": 200
+    },
+    {
+      "epoch": 0.8701298701298701,
+      "grad_norm": 5.247851848602295,
+      "learning_rate": 9.117981199359574e-05,
+      "loss": 0.8125,
+      "step": 201
+    },
+    {
+      "epoch": 0.8744588744588745,
+      "grad_norm": 6.284152507781982,
+      "learning_rate": 9.108012616521055e-05,
+      "loss": 0.875,
+      "step": 202
+    },
+    {
+      "epoch": 0.8787878787878788,
+      "grad_norm": 18.2021427154541,
+      "learning_rate": 9.097993530929895e-05,
+      "loss": 0.9062,
+      "step": 203
+    },
+    {
+      "epoch": 0.8831168831168831,
+      "grad_norm": 21.63412857055664,
+      "learning_rate": 9.087924065757919e-05,
+      "loss": 0.8281,
+      "step": 204
+    },
+    {
+      "epoch": 0.8874458874458875,
+      "grad_norm": 4.992595195770264,
+      "learning_rate": 9.077804344796302e-05,
+      "loss": 0.8164,
+      "step": 205
+    },
+    {
+      "epoch": 0.8917748917748918,
+      "grad_norm": 5.149569988250732,
+      "learning_rate": 9.067634492454048e-05,
+      "loss": 0.918,
+      "step": 206
+    },
+    {
+      "epoch": 0.8961038961038961,
+      "grad_norm": 15.364137649536133,
+      "learning_rate": 9.057414633756464e-05,
+      "loss": 0.8477,
+      "step": 207
+    },
+    {
+      "epoch": 0.9004329004329005,
+      "grad_norm": 8.03525161743164,
+      "learning_rate": 9.047144894343623e-05,
+      "loss": 0.7383,
+      "step": 208
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 4.5894269943237305,
+      "learning_rate": 9.036825400468812e-05,
+      "loss": 0.8047,
+      "step": 209
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 5.035865306854248,
+      "learning_rate": 9.02645627899699e-05,
+      "loss": 0.8242,
+      "step": 210
+    },
+    {
+      "epoch": 0.9134199134199135,
+      "grad_norm": 11.406376838684082,
+      "learning_rate": 9.016037657403224e-05,
+      "loss": 0.7969,
+      "step": 211
+    },
+    {
+      "epoch": 0.9177489177489178,
+      "grad_norm": 6.819675922393799,
+      "learning_rate": 9.005569663771119e-05,
+      "loss": 0.7227,
+      "step": 212
+    },
+    {
+      "epoch": 0.922077922077922,
+      "grad_norm": 6.333429336547852,
+      "learning_rate": 8.995052426791247e-05,
+      "loss": 0.8164,
+      "step": 213
+    },
+    {
+      "epoch": 0.9264069264069265,
+      "grad_norm": 6.850126266479492,
+      "learning_rate": 8.98448607575956e-05,
+      "loss": 0.9023,
+      "step": 214
+    },
+    {
+      "epoch": 0.9307359307359307,
+      "grad_norm": 8.282530784606934,
+      "learning_rate": 8.973870740575814e-05,
+      "loss": 0.8633,
+      "step": 215
+    },
+    {
+      "epoch": 0.935064935064935,
+      "grad_norm": 7.20064640045166,
+      "learning_rate": 8.963206551741952e-05,
+      "loss": 0.9336,
+      "step": 216
+    },
+    {
+      "epoch": 0.9393939393939394,
+      "grad_norm": 6.47982931137085,
+      "learning_rate": 8.952493640360517e-05,
+      "loss": 0.8125,
+      "step": 217
+    },
+    {
+      "epoch": 0.9437229437229437,
+      "grad_norm": 7.856199741363525,
+      "learning_rate": 8.941732138133032e-05,
+      "loss": 0.9023,
+      "step": 218
+    },
+    {
+      "epoch": 0.948051948051948,
+      "grad_norm": 7.736241817474365,
+      "learning_rate": 8.930922177358379e-05,
+      "loss": 0.75,
+      "step": 219
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 7.214029788970947,
+      "learning_rate": 8.920063890931177e-05,
+      "loss": 0.7891,
+      "step": 220
+    },
+    {
+      "epoch": 0.9567099567099567,
+      "grad_norm": 6.125720977783203,
+      "learning_rate": 8.90915741234015e-05,
+      "loss": 0.7031,
+      "step": 221
+    },
+    {
+      "epoch": 0.961038961038961,
+      "grad_norm": 9.527080535888672,
+      "learning_rate": 8.89820287566648e-05,
+      "loss": 0.9102,
+      "step": 222
+    },
+    {
+      "epoch": 0.9653679653679653,
+      "grad_norm": 8.69723892211914,
+      "learning_rate": 8.887200415582161e-05,
+      "loss": 0.9336,
+      "step": 223
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 11.235180854797363,
+      "learning_rate": 8.876150167348348e-05,
+      "loss": 0.9062,
+      "step": 224
+    },
+    {
+      "epoch": 0.974025974025974,
+      "grad_norm": 6.430836200714111,
+      "learning_rate": 8.865052266813685e-05,
+      "loss": 0.8125,
+      "step": 225
+    },
+    {
+      "epoch": 0.9783549783549783,
+      "grad_norm": 5.01962947845459,
+      "learning_rate": 8.853906850412644e-05,
+      "loss": 0.8164,
+      "step": 226
+    },
+    {
+      "epoch": 0.9826839826839827,
+      "grad_norm": 6.740474700927734,
+      "learning_rate": 8.842714055163841e-05,
+      "loss": 0.7383,
+      "step": 227
+    },
+    {
+      "epoch": 0.987012987012987,
+      "grad_norm": 16.557838439941406,
+      "learning_rate": 8.831474018668355e-05,
+      "loss": 0.8945,
+      "step": 228
+    },
+    {
+      "epoch": 0.9913419913419913,
+      "grad_norm": 7.960574626922607,
+      "learning_rate": 8.820186879108038e-05,
+      "loss": 0.918,
+      "step": 229
+    },
+    {
+      "epoch": 0.9956709956709957,
+      "grad_norm": 6.437197685241699,
+      "learning_rate": 8.808852775243809e-05,
+      "loss": 0.8203,
+      "step": 230
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.2234907150268555,
+      "learning_rate": 8.797471846413958e-05,
+      "loss": 0.8906,
+      "step": 231
+    },
+    {
+      "epoch": 1.0043290043290043,
+      "grad_norm": 27.107839584350586,
+      "learning_rate": 8.786044232532424e-05,
+      "loss": 0.832,
+      "step": 232
+    },
+    {
+      "epoch": 1.0086580086580086,
+      "grad_norm": 11.702957153320312,
+      "learning_rate": 8.77457007408708e-05,
+      "loss": 0.625,
+      "step": 233
+    },
+    {
+      "epoch": 1.0129870129870129,
+      "grad_norm": 4.364512920379639,
+      "learning_rate": 8.763049512138008e-05,
+      "loss": 0.7578,
+      "step": 234
+    },
+    {
+      "epoch": 1.0173160173160174,
+      "grad_norm": 6.135038375854492,
+      "learning_rate": 8.751482688315757e-05,
+      "loss": 0.8281,
+      "step": 235
+    },
+    {
+      "epoch": 1.0216450216450217,
+      "grad_norm": 11.477568626403809,
+      "learning_rate": 8.739869744819609e-05,
+      "loss": 0.7656,
+      "step": 236
+    },
+    {
+      "epoch": 1.025974025974026,
+      "grad_norm": 6.294708251953125,
+      "learning_rate": 8.728210824415827e-05,
+      "loss": 0.6953,
+      "step": 237
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 6.623936176300049,
+      "learning_rate": 8.716506070435906e-05,
+      "loss": 0.8789,
+      "step": 238
+    },
+    {
+      "epoch": 1.0346320346320346,
+      "grad_norm": 5.694393634796143,
+      "learning_rate": 8.704755626774796e-05,
+      "loss": 0.6719,
+      "step": 239
+    },
+    {
+      "epoch": 1.0389610389610389,
+      "grad_norm": 7.346079349517822,
+      "learning_rate": 8.692959637889154e-05,
+      "loss": 0.6992,
+      "step": 240
+    },
+    {
+      "epoch": 1.0432900432900434,
+      "grad_norm": 5.832156181335449,
+      "learning_rate": 8.681118248795547e-05,
+      "loss": 0.6719,
+      "step": 241
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 5.225601673126221,
+      "learning_rate": 8.66923160506869e-05,
+      "loss": 0.7969,
+      "step": 242
+    },
+    {
+      "epoch": 1.051948051948052,
+      "grad_norm": 11.674131393432617,
+      "learning_rate": 8.657299852839639e-05,
+      "loss": 0.7734,
+      "step": 243
+    },
+    {
+      "epoch": 1.0562770562770563,
+      "grad_norm": 5.810673236846924,
+      "learning_rate": 8.645323138794001e-05,
+      "loss": 0.8047,
+      "step": 244
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 5.174658298492432,
+      "learning_rate": 8.633301610170135e-05,
+      "loss": 0.7109,
+      "step": 245
+    },
+    {
+      "epoch": 1.0649350649350648,
+      "grad_norm": 10.819430351257324,
+      "learning_rate": 8.621235414757336e-05,
+      "loss": 0.7656,
+      "step": 246
+    },
+    {
+      "epoch": 1.0692640692640694,
+      "grad_norm": 7.743246555328369,
+      "learning_rate": 8.609124700894016e-05,
+      "loss": 0.7578,
+      "step": 247
+    },
+    {
+      "epoch": 1.0735930735930737,
+      "grad_norm": 5.833282947540283,
+      "learning_rate": 8.596969617465889e-05,
+      "loss": 0.918,
+      "step": 248
+    },
+    {
+      "epoch": 1.077922077922078,
+      "grad_norm": 6.652664661407471,
+      "learning_rate": 8.584770313904137e-05,
+      "loss": 0.8281,
+      "step": 249
+    },
+    {
+      "epoch": 1.0822510822510822,
+      "grad_norm": 5.4417901039123535,
+      "learning_rate": 8.572526940183568e-05,
+      "loss": 0.7266,
+      "step": 250
+    },
+    {
+      "epoch": 1.0865800865800865,
+      "grad_norm": 5.821901798248291,
+      "learning_rate": 8.560239646820778e-05,
+      "loss": 0.6602,
+      "step": 251
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 5.3736371994018555,
+      "learning_rate": 8.547908584872297e-05,
+      "loss": 0.7695,
+      "step": 252
+    },
+    {
+      "epoch": 1.0952380952380953,
+      "grad_norm": 10.124473571777344,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 0.7812,
+      "step": 253
+    },
+    {
+      "epoch": 1.0995670995670996,
+      "grad_norm": 4.445034027099609,
+      "learning_rate": 8.523115762132925e-05,
+      "loss": 0.6406,
+      "step": 254
+    },
+    {
+      "epoch": 1.103896103896104,
+      "grad_norm": 7.271275520324707,
+      "learning_rate": 8.510654306138027e-05,
+      "loss": 0.7578,
+      "step": 255
+    },
+    {
+      "epoch": 1.1082251082251082,
+      "grad_norm": 4.753751754760742,
+      "learning_rate": 8.498149691145684e-05,
+      "loss": 0.6523,
+      "step": 256
+    },
+    {
+      "epoch": 1.1125541125541125,
+      "grad_norm": 4.876266002655029,
+      "learning_rate": 8.485602070884117e-05,
+      "loss": 0.8164,
+      "step": 257
+    },
+    {
+      "epoch": 1.1168831168831168,
+      "grad_norm": 6.6608805656433105,
+      "learning_rate": 8.473011599610244e-05,
+      "loss": 0.8047,
+      "step": 258
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 8.465462684631348,
+      "learning_rate": 8.460378432107779e-05,
+      "loss": 0.8164,
+      "step": 259
+    },
+    {
+      "epoch": 1.1255411255411256,
+      "grad_norm": 5.219013690948486,
+      "learning_rate": 8.447702723685335e-05,
+      "loss": 0.6836,
+      "step": 260
+    },
+    {
+      "epoch": 1.12987012987013,
+      "grad_norm": 4.994295120239258,
+      "learning_rate": 8.434984630174509e-05,
+      "loss": 0.8086,
+      "step": 261
+    },
+    {
+      "epoch": 1.1341991341991342,
+      "grad_norm": 5.287687301635742,
+      "learning_rate": 8.422224307927967e-05,
+      "loss": 0.793,
+      "step": 262
+    },
+    {
+      "epoch": 1.1385281385281385,
+      "grad_norm": 4.361443519592285,
+      "learning_rate": 8.409421913817526e-05,
+      "loss": 0.8359,
+      "step": 263
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 4.188826560974121,
+      "learning_rate": 8.39657760523222e-05,
+      "loss": 0.7344,
+      "step": 264
+    },
+    {
+      "epoch": 1.1471861471861473,
+      "grad_norm": 4.0929646492004395,
+      "learning_rate": 8.383691540076371e-05,
+      "loss": 0.6211,
+      "step": 265
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "grad_norm": 7.326121807098389,
+      "learning_rate": 8.370763876767642e-05,
+      "loss": 0.7422,
+      "step": 266
+    },
+    {
+      "epoch": 1.155844155844156,
+      "grad_norm": 7.618014812469482,
+      "learning_rate": 8.357794774235092e-05,
+      "loss": 0.7617,
+      "step": 267
+    },
+    {
+      "epoch": 1.1601731601731602,
+      "grad_norm": 4.826286792755127,
+      "learning_rate": 8.344784391917224e-05,
+      "loss": 0.7734,
+      "step": 268
+    },
+    {
+      "epoch": 1.1645021645021645,
+      "grad_norm": 5.376044273376465,
+      "learning_rate": 8.33173288976002e-05,
+      "loss": 0.7812,
+      "step": 269
+    },
+    {
+      "epoch": 1.1688311688311688,
+      "grad_norm": 5.67780876159668,
+      "learning_rate": 8.318640428214982e-05,
+      "loss": 0.7734,
+      "step": 270
+    },
+    {
+      "epoch": 1.173160173160173,
+      "grad_norm": 4.724506855010986,
+      "learning_rate": 8.305507168237152e-05,
+      "loss": 0.707,
+      "step": 271
+    },
+    {
+      "epoch": 1.1774891774891776,
+      "grad_norm": 3.9453792572021484,
+      "learning_rate": 8.292333271283134e-05,
+      "loss": 0.8086,
+      "step": 272
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 7.941428184509277,
+      "learning_rate": 8.279118899309122e-05,
+      "loss": 0.7617,
+      "step": 273
+    },
+    {
+      "epoch": 1.1861471861471862,
+      "grad_norm": 4.79378604888916,
+      "learning_rate": 8.265864214768883e-05,
+      "loss": 0.8516,
+      "step": 274
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 7.283705711364746,
+      "learning_rate": 8.252569380611792e-05,
+      "loss": 0.7539,
+      "step": 275
+    },
+    {
+      "epoch": 1.1948051948051948,
+      "grad_norm": 5.552491188049316,
+      "learning_rate": 8.2392345602808e-05,
+      "loss": 0.5859,
+      "step": 276
+    },
+    {
+      "epoch": 1.199134199134199,
+      "grad_norm": 4.452544689178467,
+      "learning_rate": 8.225859917710439e-05,
+      "loss": 0.6133,
+      "step": 277
+    },
+    {
+      "epoch": 1.2034632034632033,
+      "grad_norm": 4.441799640655518,
+      "learning_rate": 8.212445617324809e-05,
+      "loss": 0.7031,
+      "step": 278
+    },
+    {
+      "epoch": 1.2077922077922079,
+      "grad_norm": 4.305429935455322,
+      "learning_rate": 8.198991824035547e-05,
+      "loss": 0.7852,
+      "step": 279
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 10.642815589904785,
+      "learning_rate": 8.18549870323981e-05,
+      "loss": 0.6875,
+      "step": 280
+    },
+    {
+      "epoch": 1.2164502164502164,
+      "grad_norm": 5.574221134185791,
+      "learning_rate": 8.171966420818228e-05,
+      "loss": 0.6328,
+      "step": 281
+    },
+    {
+      "epoch": 1.2207792207792207,
+      "grad_norm": 7.952475070953369,
+      "learning_rate": 8.158395143132882e-05,
+      "loss": 0.7539,
+      "step": 282
+    },
+    {
+      "epoch": 1.225108225108225,
+      "grad_norm": 5.21942138671875,
+      "learning_rate": 8.144785037025247e-05,
+      "loss": 0.707,
+      "step": 283
+    },
+    {
+      "epoch": 1.2294372294372296,
+      "grad_norm": 8.209097862243652,
+      "learning_rate": 8.131136269814139e-05,
+      "loss": 0.7695,
+      "step": 284
+    },
+    {
+      "epoch": 1.2337662337662338,
+      "grad_norm": 5.223848342895508,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.7695,
+      "step": 285
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 13.468093872070312,
+      "learning_rate": 8.103723423731169e-05,
+      "loss": 0.6992,
+      "step": 286
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 4.496579170227051,
+      "learning_rate": 8.089959681865133e-05,
+      "loss": 0.7148,
+      "step": 287
+    },
+    {
+      "epoch": 1.2467532467532467,
+      "grad_norm": 4.73178768157959,
+      "learning_rate": 8.076157952903134e-05,
+      "loss": 0.6992,
+      "step": 288
+    },
+    {
+      "epoch": 1.251082251082251,
+      "grad_norm": 5.16973352432251,
+      "learning_rate": 8.062318406519751e-05,
+      "loss": 0.7461,
+      "step": 289
+    },
+    {
+      "epoch": 1.2554112554112553,
+      "grad_norm": 7.291965484619141,
+      "learning_rate": 8.048441212854476e-05,
+      "loss": 0.6719,
+      "step": 290
+    },
+    {
+      "epoch": 1.2597402597402598,
+      "grad_norm": 4.964224338531494,
+      "learning_rate": 8.034526542509628e-05,
+      "loss": 0.7148,
+      "step": 291
+    },
+    {
+      "epoch": 1.2640692640692641,
+      "grad_norm": 4.401447772979736,
+      "learning_rate": 8.020574566548256e-05,
+      "loss": 0.625,
+      "step": 292
+    },
+    {
+      "epoch": 1.2683982683982684,
+      "grad_norm": 14.532041549682617,
+      "learning_rate": 8.006585456492029e-05,
+      "loss": 0.7031,
+      "step": 293
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 11.620001792907715,
+      "learning_rate": 7.992559384319137e-05,
+      "loss": 0.6328,
+      "step": 294
+    },
+    {
+      "epoch": 1.277056277056277,
+      "grad_norm": 4.399411201477051,
+      "learning_rate": 7.978496522462167e-05,
+      "loss": 0.7266,
+      "step": 295
+    },
+    {
+      "epoch": 1.2813852813852815,
+      "grad_norm": 4.530557155609131,
+      "learning_rate": 7.964397043805993e-05,
+      "loss": 0.7734,
+      "step": 296
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 4.283398151397705,
+      "learning_rate": 7.950261121685641e-05,
+      "loss": 0.7344,
+      "step": 297
+    },
+    {
+      "epoch": 1.29004329004329,
+      "grad_norm": 4.7228288650512695,
+      "learning_rate": 7.936088929884166e-05,
+      "loss": 0.7344,
+      "step": 298
+    },
+    {
+      "epoch": 1.2943722943722944,
+      "grad_norm": 4.837503433227539,
+      "learning_rate": 7.921880642630514e-05,
+      "loss": 0.7031,
+      "step": 299
+    },
+    {
+      "epoch": 1.2987012987012987,
+      "grad_norm": 4.3210062980651855,
+      "learning_rate": 7.907636434597369e-05,
+      "loss": 0.75,
+      "step": 300
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 6.427566051483154,
+      "learning_rate": 7.89335648089903e-05,
+      "loss": 0.6836,
+      "step": 301
+    },
+    {
+      "epoch": 1.3073593073593073,
+      "grad_norm": 6.730422019958496,
+      "learning_rate": 7.879040957089228e-05,
+      "loss": 0.6523,
+      "step": 302
+    },
+    {
+      "epoch": 1.3116883116883118,
+      "grad_norm": 3.6803860664367676,
+      "learning_rate": 7.86469003915899e-05,
+      "loss": 0.6758,
+      "step": 303
+    },
+    {
+      "epoch": 1.316017316017316,
+      "grad_norm": 7.196141719818115,
+      "learning_rate": 7.850303903534473e-05,
+      "loss": 0.7656,
+      "step": 304
+    },
+    {
+      "epoch": 1.3203463203463204,
+      "grad_norm": 11.118663787841797,
+      "learning_rate": 7.835882727074779e-05,
+      "loss": 0.7109,
+      "step": 305
+    },
+    {
+      "epoch": 1.3246753246753247,
+      "grad_norm": 4.406792640686035,
+      "learning_rate": 7.821426687069803e-05,
+      "loss": 0.7773,
+      "step": 306
+    },
+    {
+      "epoch": 1.329004329004329,
+      "grad_norm": 4.833268165588379,
+      "learning_rate": 7.80693596123804e-05,
+      "loss": 0.7539,
+      "step": 307
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 4.882203102111816,
+      "learning_rate": 7.792410727724395e-05,
+      "loss": 0.7539,
+      "step": 308
+    },
+    {
+      "epoch": 1.3376623376623376,
+      "grad_norm": 4.0016865730285645,
+      "learning_rate": 7.777851165098012e-05,
+      "loss": 0.75,
+      "step": 309
+    },
+    {
+      "epoch": 1.341991341991342,
+      "grad_norm": 7.231997013092041,
+      "learning_rate": 7.763257452350058e-05,
+      "loss": 0.6562,
+      "step": 310
+    },
+    {
+      "epoch": 1.3463203463203464,
+      "grad_norm": 4.630609512329102,
+      "learning_rate": 7.748629768891543e-05,
+      "loss": 0.6914,
+      "step": 311
+    },
+    {
+      "epoch": 1.3506493506493507,
+      "grad_norm": 4.233306407928467,
+      "learning_rate": 7.733968294551088e-05,
+      "loss": 0.75,
+      "step": 312
+    },
+    {
+      "epoch": 1.354978354978355,
+      "grad_norm": 4.6222405433654785,
+      "learning_rate": 7.719273209572744e-05,
+      "loss": 0.7461,
+      "step": 313
+    },
+    {
+      "epoch": 1.3593073593073592,
+      "grad_norm": 6.420832633972168,
+      "learning_rate": 7.704544694613755e-05,
+      "loss": 0.707,
+      "step": 314
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 5.846024990081787,
+      "learning_rate": 7.689782930742344e-05,
+      "loss": 0.625,
+      "step": 315
+    },
+    {
+      "epoch": 1.3679653679653678,
+      "grad_norm": 7.410788536071777,
+      "learning_rate": 7.674988099435487e-05,
+      "loss": 0.6875,
+      "step": 316
+    },
+    {
+      "epoch": 1.3722943722943723,
+      "grad_norm": 13.918171882629395,
+      "learning_rate": 7.660160382576683e-05,
+      "loss": 0.7969,
+      "step": 317
+    },
+    {
+      "epoch": 1.3766233766233766,
+      "grad_norm": 5.576581954956055,
+      "learning_rate": 7.645299962453716e-05,
+      "loss": 0.6758,
+      "step": 318
+    },
+    {
+      "epoch": 1.380952380952381,
+      "grad_norm": 6.381285190582275,
+      "learning_rate": 7.630407021756418e-05,
+      "loss": 0.7305,
+      "step": 319
+    },
+    {
+      "epoch": 1.3852813852813852,
+      "grad_norm": 7.615292549133301,
+      "learning_rate": 7.615481743574411e-05,
+      "loss": 0.6719,
+      "step": 320
+    },
+    {
+      "epoch": 1.3896103896103895,
+      "grad_norm": 4.914462089538574,
+      "learning_rate": 7.600524311394873e-05,
+      "loss": 0.6914,
+      "step": 321
+    },
+    {
+      "epoch": 1.393939393939394,
+      "grad_norm": 6.371485710144043,
+      "learning_rate": 7.585534909100274e-05,
+      "loss": 0.7969,
+      "step": 322
+    },
+    {
+      "epoch": 1.3982683982683983,
+      "grad_norm": 4.094557285308838,
+      "learning_rate": 7.570513720966108e-05,
+      "loss": 0.6992,
+      "step": 323
+    },
+    {
+      "epoch": 1.4025974025974026,
+      "grad_norm": 4.090404510498047,
+      "learning_rate": 7.555460931658647e-05,
+      "loss": 0.7422,
+      "step": 324
+    },
+    {
+      "epoch": 1.406926406926407,
+      "grad_norm": 4.56276798248291,
+      "learning_rate": 7.540376726232648e-05,
+      "loss": 0.7734,
+      "step": 325
+    },
+    {
+      "epoch": 1.4112554112554112,
+      "grad_norm": 4.187833309173584,
+      "learning_rate": 7.525261290129095e-05,
+      "loss": 0.6992,
+      "step": 326
+    },
+    {
+      "epoch": 1.4155844155844157,
+      "grad_norm": 4.825344085693359,
+      "learning_rate": 7.51011480917291e-05,
+      "loss": 0.6719,
+      "step": 327
+    },
+    {
+      "epoch": 1.4199134199134198,
+      "grad_norm": 7.891578197479248,
+      "learning_rate": 7.494937469570676e-05,
+      "loss": 0.707,
+      "step": 328
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 6.907805442810059,
+      "learning_rate": 7.47972945790834e-05,
+      "loss": 0.6797,
+      "step": 329
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 5.062526702880859,
+      "learning_rate": 7.464490961148921e-05,
+      "loss": 0.6914,
+      "step": 330
+    },
+    {
+      "epoch": 1.432900432900433,
+      "grad_norm": 4.783566474914551,
+      "learning_rate": 7.449222166630218e-05,
+      "loss": 0.7109,
+      "step": 331
+    },
+    {
+      "epoch": 1.4372294372294372,
+      "grad_norm": 13.14342212677002,
+      "learning_rate": 7.4339232620625e-05,
+      "loss": 0.625,
+      "step": 332
+    },
+    {
+      "epoch": 1.4415584415584415,
+      "grad_norm": 13.73547077178955,
+      "learning_rate": 7.4185944355262e-05,
+      "loss": 0.7734,
+      "step": 333
+    },
+    {
+      "epoch": 1.445887445887446,
+      "grad_norm": 6.408390522003174,
+      "learning_rate": 7.403235875469603e-05,
+      "loss": 0.7383,
+      "step": 334
+    },
+    {
+      "epoch": 1.4502164502164503,
+      "grad_norm": 17.325075149536133,
+      "learning_rate": 7.387847770706531e-05,
+      "loss": 0.7109,
+      "step": 335
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 6.630992412567139,
+      "learning_rate": 7.372430310414016e-05,
+      "loss": 0.6602,
+      "step": 336
+    },
+    {
+      "epoch": 1.4588744588744589,
+      "grad_norm": 4.792052268981934,
+      "learning_rate": 7.35698368412999e-05,
+      "loss": 0.707,
+      "step": 337
+    },
+    {
+      "epoch": 1.4632034632034632,
+      "grad_norm": 5.60944938659668,
+      "learning_rate": 7.341508081750927e-05,
+      "loss": 0.7305,
+      "step": 338
+    },
+    {
+      "epoch": 1.4675324675324675,
+      "grad_norm": 6.859635829925537,
+      "learning_rate": 7.326003693529538e-05,
+      "loss": 0.6016,
+      "step": 339
+    },
+    {
+      "epoch": 1.4718614718614718,
+      "grad_norm": 4.734246730804443,
+      "learning_rate": 7.310470710072413e-05,
+      "loss": 0.668,
+      "step": 340
+    },
+    {
+      "epoch": 1.4761904761904763,
+      "grad_norm": 8.14387035369873,
+      "learning_rate": 7.294909322337689e-05,
+      "loss": 0.6875,
+      "step": 341
+    },
+    {
+      "epoch": 1.4805194805194806,
+      "grad_norm": 5.688750267028809,
+      "learning_rate": 7.27931972163269e-05,
+      "loss": 0.7891,
+      "step": 342
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 6.133918285369873,
+      "learning_rate": 7.263702099611586e-05,
+      "loss": 0.7656,
+      "step": 343
+    },
+    {
+      "epoch": 1.4891774891774892,
+      "grad_norm": 5.702477931976318,
+      "learning_rate": 7.248056648273034e-05,
+      "loss": 0.6914,
+      "step": 344
+    },
+    {
+      "epoch": 1.4935064935064934,
+      "grad_norm": 3.9108164310455322,
+      "learning_rate": 7.232383559957814e-05,
+      "loss": 0.6055,
+      "step": 345
+    },
+    {
+      "epoch": 1.497835497835498,
+      "grad_norm": 4.188071250915527,
+      "learning_rate": 7.216683027346475e-05,
+      "loss": 0.7148,
+      "step": 346
+    },
+    {
+      "epoch": 1.502164502164502,
+      "grad_norm": 5.449563503265381,
+      "learning_rate": 7.200955243456947e-05,
+      "loss": 0.7812,
+      "step": 347
+    },
+    {
+      "epoch": 1.5064935064935066,
+      "grad_norm": 4.995254039764404,
+      "learning_rate": 7.185200401642187e-05,
+      "loss": 0.6914,
+      "step": 348
+    },
+    {
+      "epoch": 1.5108225108225108,
+      "grad_norm": 10.685470581054688,
+      "learning_rate": 7.169418695587791e-05,
+      "loss": 0.6797,
+      "step": 349
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 12.334436416625977,
+      "learning_rate": 7.153610319309621e-05,
+      "loss": 0.8164,
+      "step": 350
+    },
+    {
+      "epoch": 1.5194805194805194,
+      "grad_norm": 9.344776153564453,
+      "learning_rate": 7.137775467151411e-05,
+      "loss": 0.7188,
+      "step": 351
+    },
+    {
+      "epoch": 1.5238095238095237,
+      "grad_norm": 11.974285125732422,
+      "learning_rate": 7.121914333782384e-05,
+      "loss": 0.6445,
+      "step": 352
+    },
+    {
+      "epoch": 1.5281385281385282,
+      "grad_norm": 7.771387100219727,
+      "learning_rate": 7.106027114194855e-05,
+      "loss": 0.7031,
+      "step": 353
+    },
+    {
+      "epoch": 1.5324675324675323,
+      "grad_norm": 5.118148326873779,
+      "learning_rate": 7.090114003701838e-05,
+      "loss": 0.6094,
+      "step": 354
+    },
+    {
+      "epoch": 1.5367965367965368,
+      "grad_norm": 7.12678861618042,
+      "learning_rate": 7.074175197934639e-05,
+      "loss": 0.6523,
+      "step": 355
+    },
+    {
+      "epoch": 1.5411255411255411,
+      "grad_norm": 9.114030838012695,
+      "learning_rate": 7.058210892840461e-05,
+      "loss": 0.7266,
+      "step": 356
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 7.879207134246826,
+      "learning_rate": 7.042221284679982e-05,
+      "loss": 0.6641,
+      "step": 357
+    },
+    {
+      "epoch": 1.54978354978355,
+      "grad_norm": 4.762259006500244,
+      "learning_rate": 7.02620657002495e-05,
+      "loss": 0.6289,
+      "step": 358
+    },
+    {
+      "epoch": 1.554112554112554,
+      "grad_norm": 11.492424011230469,
+      "learning_rate": 7.010166945755768e-05,
+      "loss": 0.6953,
+      "step": 359
+    },
+    {
+      "epoch": 1.5584415584415585,
+      "grad_norm": 5.3279500007629395,
+      "learning_rate": 6.99410260905907e-05,
+      "loss": 0.6523,
+      "step": 360
+    },
+    {
+      "epoch": 1.5627705627705628,
+      "grad_norm": 6.404669284820557,
+      "learning_rate": 6.978013757425295e-05,
+      "loss": 0.7539,
+      "step": 361
+    },
+    {
+      "epoch": 1.567099567099567,
+      "grad_norm": 27.591524124145508,
+      "learning_rate": 6.961900588646263e-05,
+      "loss": 0.7227,
+      "step": 362
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 14.585436820983887,
+      "learning_rate": 6.945763300812745e-05,
+      "loss": 0.7227,
+      "step": 363
+    },
+    {
+      "epoch": 1.5757575757575757,
+      "grad_norm": 10.410850524902344,
+      "learning_rate": 6.929602092312023e-05,
+      "loss": 0.7188,
+      "step": 364
+    },
+    {
+      "epoch": 1.5800865800865802,
+      "grad_norm": 6.165812015533447,
+      "learning_rate": 6.91341716182545e-05,
+      "loss": 0.7695,
+      "step": 365
+    },
+    {
+      "epoch": 1.5844155844155843,
+      "grad_norm": 10.709397315979004,
+      "learning_rate": 6.897208708326013e-05,
+      "loss": 0.668,
+      "step": 366
+    },
+    {
+      "epoch": 1.5887445887445888,
+      "grad_norm": 7.277166843414307,
+      "learning_rate": 6.880976931075887e-05,
+      "loss": 0.6367,
+      "step": 367
+    },
+    {
+      "epoch": 1.593073593073593,
+      "grad_norm": 4.61423397064209,
+      "learning_rate": 6.864722029623977e-05,
+      "loss": 0.6406,
+      "step": 368
+    },
+    {
+      "epoch": 1.5974025974025974,
+      "grad_norm": 51.29463577270508,
+      "learning_rate": 6.848444203803476e-05,
+      "loss": 0.6641,
+      "step": 369
+    },
+    {
+      "epoch": 1.601731601731602,
+      "grad_norm": 5.557583332061768,
+      "learning_rate": 6.832143653729397e-05,
+      "loss": 0.7344,
+      "step": 370
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 6.214862823486328,
+      "learning_rate": 6.815820579796127e-05,
+      "loss": 0.6953,
+      "step": 371
+    },
+    {
+      "epoch": 1.6103896103896105,
+      "grad_norm": 7.193179130554199,
+      "learning_rate": 6.799475182674942e-05,
+      "loss": 0.8047,
+      "step": 372
+    },
+    {
+      "epoch": 1.6147186147186146,
+      "grad_norm": 9.319855690002441,
+      "learning_rate": 6.783107663311565e-05,
+      "loss": 0.7422,
+      "step": 373
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 5.943477153778076,
+      "learning_rate": 6.76671822292368e-05,
+      "loss": 0.75,
+      "step": 374
+    },
+    {
+      "epoch": 1.6233766233766234,
+      "grad_norm": 4.444772720336914,
+      "learning_rate": 6.750307062998462e-05,
+      "loss": 0.5859,
+      "step": 375
+    },
+    {
+      "epoch": 1.6277056277056277,
+      "grad_norm": 7.112820148468018,
+      "learning_rate": 6.733874385290097e-05,
+      "loss": 0.6992,
+      "step": 376
+    },
+    {
+      "epoch": 1.6320346320346322,
+      "grad_norm": 6.514984607696533,
+      "learning_rate": 6.717420391817306e-05,
+      "loss": 0.7266,
+      "step": 377
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 6.372430801391602,
+      "learning_rate": 6.700945284860862e-05,
+      "loss": 0.7305,
+      "step": 378
+    },
+    {
+      "epoch": 1.6406926406926408,
+      "grad_norm": 5.759349822998047,
+      "learning_rate": 6.6844492669611e-05,
+      "loss": 0.7422,
+      "step": 379
+    },
+    {
+      "epoch": 1.645021645021645,
+      "grad_norm": 6.78725528717041,
+      "learning_rate": 6.66793254091543e-05,
+      "loss": 0.6953,
+      "step": 380
+    },
+    {
+      "epoch": 1.6493506493506493,
+      "grad_norm": 5.6776885986328125,
+      "learning_rate": 6.651395309775837e-05,
+      "loss": 0.7344,
+      "step": 381
+    },
+    {
+      "epoch": 1.6536796536796536,
+      "grad_norm": 5.542745590209961,
+      "learning_rate": 6.634837776846393e-05,
+      "loss": 0.7148,
+      "step": 382
+    },
+    {
+      "epoch": 1.658008658008658,
+      "grad_norm": 4.985962390899658,
+      "learning_rate": 6.618260145680757e-05,
+      "loss": 0.6328,
+      "step": 383
+    },
+    {
+      "epoch": 1.6623376623376624,
+      "grad_norm": 4.856525897979736,
+      "learning_rate": 6.601662620079669e-05,
+      "loss": 0.668,
+      "step": 384
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 4.41532039642334,
+      "learning_rate": 6.585045404088441e-05,
+      "loss": 0.7617,
+      "step": 385
+    },
+    {
+      "epoch": 1.670995670995671,
+      "grad_norm": 4.6031999588012695,
+      "learning_rate": 6.568408701994458e-05,
+      "loss": 0.668,
+      "step": 386
+    },
+    {
+      "epoch": 1.6753246753246753,
+      "grad_norm": 4.298882007598877,
+      "learning_rate": 6.551752718324659e-05,
+      "loss": 0.7031,
+      "step": 387
+    },
+    {
+      "epoch": 1.6796536796536796,
+      "grad_norm": 3.378218173980713,
+      "learning_rate": 6.53507765784303e-05,
+      "loss": 0.6562,
+      "step": 388
+    },
+    {
+      "epoch": 1.6839826839826841,
+      "grad_norm": 4.7255353927612305,
+      "learning_rate": 6.518383725548074e-05,
+      "loss": 0.6367,
+      "step": 389
+    },
+    {
+      "epoch": 1.6883116883116882,
+      "grad_norm": 4.435005187988281,
+      "learning_rate": 6.501671126670306e-05,
+      "loss": 0.7109,
+      "step": 390
+    },
+    {
+      "epoch": 1.6926406926406927,
+      "grad_norm": 4.028753757476807,
+      "learning_rate": 6.484940066669718e-05,
+      "loss": 0.6992,
+      "step": 391
+    },
+    {
+      "epoch": 1.696969696969697,
+      "grad_norm": 4.245719909667969,
+      "learning_rate": 6.468190751233262e-05,
+      "loss": 0.707,
+      "step": 392
+    },
+    {
+      "epoch": 1.7012987012987013,
+      "grad_norm": 10.510866165161133,
+      "learning_rate": 6.451423386272312e-05,
+      "loss": 0.75,
+      "step": 393
+    },
+    {
+      "epoch": 1.7056277056277056,
+      "grad_norm": 6.969714641571045,
+      "learning_rate": 6.434638177920144e-05,
+      "loss": 0.6523,
+      "step": 394
+    },
+    {
+      "epoch": 1.70995670995671,
+      "grad_norm": 3.4226455688476562,
+      "learning_rate": 6.417835332529389e-05,
+      "loss": 0.7656,
+      "step": 395
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 4.687984466552734,
+      "learning_rate": 6.401015056669508e-05,
+      "loss": 0.7422,
+      "step": 396
+    },
+    {
+      "epoch": 1.7186147186147185,
+      "grad_norm": 8.397469520568848,
+      "learning_rate": 6.384177557124247e-05,
+      "loss": 0.6211,
+      "step": 397
+    },
+    {
+      "epoch": 1.722943722943723,
+      "grad_norm": 4.2439799308776855,
+      "learning_rate": 6.36732304088909e-05,
+      "loss": 0.6719,
+      "step": 398
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 5.938943386077881,
+      "learning_rate": 6.350451715168729e-05,
+      "loss": 0.6836,
+      "step": 399
+    },
+    {
+      "epoch": 1.7316017316017316,
+      "grad_norm": 5.608084678649902,
+      "learning_rate": 6.333563787374493e-05,
+      "loss": 0.7109,
+      "step": 400
+    },
+    {
+      "epoch": 1.7359307359307359,
+      "grad_norm": 9.191603660583496,
+      "learning_rate": 6.316659465121824e-05,
+      "loss": 0.7266,
+      "step": 401
+    },
+    {
+      "epoch": 1.7402597402597402,
+      "grad_norm": 4.028605937957764,
+      "learning_rate": 6.299738956227712e-05,
+      "loss": 0.7227,
+      "step": 402
+    },
+    {
+      "epoch": 1.7445887445887447,
+      "grad_norm": 3.698301315307617,
+      "learning_rate": 6.282802468708133e-05,
+      "loss": 0.6133,
+      "step": 403
+    },
+    {
+      "epoch": 1.7489177489177488,
+      "grad_norm": 6.0588202476501465,
+      "learning_rate": 6.26585021077551e-05,
+      "loss": 0.7148,
+      "step": 404
+    },
+    {
+      "epoch": 1.7532467532467533,
+      "grad_norm": 5.845741271972656,
+      "learning_rate": 6.248882390836135e-05,
+      "loss": 0.668,
+      "step": 405
+    },
+    {
+      "epoch": 1.7575757575757576,
+      "grad_norm": 4.519708156585693,
+      "learning_rate": 6.231899217487619e-05,
+      "loss": 0.6797,
+      "step": 406
+    },
+    {
+      "epoch": 1.7619047619047619,
+      "grad_norm": 6.589949131011963,
+      "learning_rate": 6.21490089951632e-05,
+      "loss": 0.6992,
+      "step": 407
+    },
+    {
+      "epoch": 1.7662337662337664,
+      "grad_norm": 5.7892537117004395,
+      "learning_rate": 6.197887645894784e-05,
+      "loss": 0.8086,
+      "step": 408
+    },
+    {
+      "epoch": 1.7705627705627704,
+      "grad_norm": 4.918306350708008,
+      "learning_rate": 6.180859665779172e-05,
+      "loss": 0.6953,
+      "step": 409
+    },
+    {
+      "epoch": 1.774891774891775,
+      "grad_norm": 6.09152364730835,
+      "learning_rate": 6.163817168506682e-05,
+      "loss": 0.6562,
+      "step": 410
+    },
+    {
+      "epoch": 1.7792207792207793,
+      "grad_norm": 6.416179656982422,
+      "learning_rate": 6.14676036359299e-05,
+      "loss": 0.7734,
+      "step": 411
+    },
+    {
+      "epoch": 1.7835497835497836,
+      "grad_norm": 8.586041450500488,
+      "learning_rate": 6.12968946072966e-05,
+      "loss": 0.7773,
+      "step": 412
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 4.886567115783691,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 0.6641,
+      "step": 413
+    },
+    {
+      "epoch": 1.7922077922077921,
+      "grad_norm": 19.161401748657227,
+      "learning_rate": 6.095506200784349e-05,
+      "loss": 0.7031,
+      "step": 414
+    },
+    {
+      "epoch": 1.7965367965367967,
+      "grad_norm": 6.034037113189697,
+      "learning_rate": 6.078394263941761e-05,
+      "loss": 0.6914,
+      "step": 415
+    },
+    {
+      "epoch": 1.8008658008658007,
+      "grad_norm": 9.692041397094727,
+      "learning_rate": 6.0612690696231534e-05,
+      "loss": 0.8438,
+      "step": 416
+    },
+    {
+      "epoch": 1.8051948051948052,
+      "grad_norm": 6.240139484405518,
+      "learning_rate": 6.04413082836085e-05,
+      "loss": 0.7031,
+      "step": 417
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 38.88730239868164,
+      "learning_rate": 6.0269797508475723e-05,
+      "loss": 0.7539,
+      "step": 418
+    },
+    {
+      "epoch": 1.8138528138528138,
+      "grad_norm": 4.775875091552734,
+      "learning_rate": 6.009816047933848e-05,
+      "loss": 0.7266,
+      "step": 419
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 8.797544479370117,
+      "learning_rate": 5.992639930625419e-05,
+      "loss": 0.7461,
+      "step": 420
+    },
+    {
+      "epoch": 1.8225108225108224,
+      "grad_norm": 4.717647552490234,
+      "learning_rate": 5.9754516100806423e-05,
+      "loss": 0.6562,
+      "step": 421
+    },
+    {
+      "epoch": 1.826839826839827,
+      "grad_norm": 4.935179233551025,
+      "learning_rate": 5.9582512976079e-05,
+      "loss": 0.6797,
+      "step": 422
+    },
+    {
+      "epoch": 1.8311688311688312,
+      "grad_norm": 4.978168487548828,
+      "learning_rate": 5.941039204663001e-05,
+      "loss": 0.6133,
+      "step": 423
+    },
+    {
+      "epoch": 1.8354978354978355,
+      "grad_norm": 4.850203514099121,
+      "learning_rate": 5.923815542846579e-05,
+      "loss": 0.6602,
+      "step": 424
+    },
+    {
+      "epoch": 1.8398268398268398,
+      "grad_norm": 4.365367889404297,
+      "learning_rate": 5.9065805239014923e-05,
+      "loss": 0.6172,
+      "step": 425
+    },
+    {
+      "epoch": 1.844155844155844,
+      "grad_norm": 11.571815490722656,
+      "learning_rate": 5.889334359710218e-05,
+      "loss": 0.7109,
+      "step": 426
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 9.26444149017334,
+      "learning_rate": 5.872077262292255e-05,
+      "loss": 0.6758,
+      "step": 427
+    },
+    {
+      "epoch": 1.8528138528138527,
+      "grad_norm": 5.720768928527832,
+      "learning_rate": 5.854809443801507e-05,
+      "loss": 0.7344,
+      "step": 428
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 4.822311878204346,
+      "learning_rate": 5.837531116523682e-05,
+      "loss": 0.707,
+      "step": 429
+    },
+    {
+      "epoch": 1.8614718614718615,
+      "grad_norm": 8.925905227661133,
+      "learning_rate": 5.820242492873682e-05,
+      "loss": 0.8008,
+      "step": 430
+    },
+    {
+      "epoch": 1.8658008658008658,
+      "grad_norm": 9.541458129882812,
+      "learning_rate": 5.802943785392986e-05,
+      "loss": 0.6992,
+      "step": 431
+    },
+    {
+      "epoch": 1.87012987012987,
+      "grad_norm": 11.834869384765625,
+      "learning_rate": 5.785635206747041e-05,
+      "loss": 0.7617,
+      "step": 432
+    },
+    {
+      "epoch": 1.8744588744588744,
+      "grad_norm": 5.379390239715576,
+      "learning_rate": 5.76831696972265e-05,
+      "loss": 0.6484,
+      "step": 433
+    },
+    {
+      "epoch": 1.878787878787879,
+      "grad_norm": 8.029460906982422,
+      "learning_rate": 5.7509892872253544e-05,
+      "loss": 0.6914,
+      "step": 434
+    },
+    {
+      "epoch": 1.883116883116883,
+      "grad_norm": 4.681676864624023,
+      "learning_rate": 5.733652372276809e-05,
+      "loss": 0.8008,
+      "step": 435
+    },
+    {
+      "epoch": 1.8874458874458875,
+      "grad_norm": 4.658121585845947,
+      "learning_rate": 5.716306438012176e-05,
+      "loss": 0.7695,
+      "step": 436
+    },
+    {
+      "epoch": 1.8917748917748918,
+      "grad_norm": 4.59970235824585,
+      "learning_rate": 5.698951697677498e-05,
+      "loss": 0.7812,
+      "step": 437
+    },
+    {
+      "epoch": 1.896103896103896,
+      "grad_norm": 4.782841682434082,
+      "learning_rate": 5.681588364627073e-05,
+      "loss": 0.6094,
+      "step": 438
+    },
+    {
+      "epoch": 1.9004329004329006,
+      "grad_norm": 3.726468086242676,
+      "learning_rate": 5.6642166523208394e-05,
+      "loss": 0.6367,
+      "step": 439
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 27.10385513305664,
+      "learning_rate": 5.646836774321745e-05,
+      "loss": 0.6523,
+      "step": 440
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 3.9916598796844482,
+      "learning_rate": 5.629448944293127e-05,
+      "loss": 0.7422,
+      "step": 441
+    },
+    {
+      "epoch": 1.9134199134199135,
+      "grad_norm": 3.4688949584960938,
+      "learning_rate": 5.6120533759960816e-05,
+      "loss": 0.5859,
+      "step": 442
+    },
+    {
+      "epoch": 1.9177489177489178,
+      "grad_norm": 5.159235000610352,
+      "learning_rate": 5.594650283286835e-05,
+      "loss": 0.7734,
+      "step": 443
+    },
+    {
+      "epoch": 1.922077922077922,
+      "grad_norm": 4.013790607452393,
+      "learning_rate": 5.577239880114121e-05,
+      "loss": 0.707,
+      "step": 444
+    },
+    {
+      "epoch": 1.9264069264069263,
+      "grad_norm": 5.714491844177246,
+      "learning_rate": 5.559822380516539e-05,
+      "loss": 0.6289,
+      "step": 445
+    },
+    {
+      "epoch": 1.9307359307359309,
+      "grad_norm": 3.6384077072143555,
+      "learning_rate": 5.54239799861994e-05,
+      "loss": 0.6992,
+      "step": 446
+    },
+    {
+      "epoch": 1.935064935064935,
+      "grad_norm": 3.4936230182647705,
+      "learning_rate": 5.5249669486347734e-05,
+      "loss": 0.7344,
+      "step": 447
+    },
+    {
+      "epoch": 1.9393939393939394,
+      "grad_norm": 7.805917263031006,
+      "learning_rate": 5.507529444853471e-05,
+      "loss": 0.6641,
+      "step": 448
+    },
+    {
+      "epoch": 1.9437229437229437,
+      "grad_norm": 7.636865615844727,
+      "learning_rate": 5.490085701647805e-05,
+      "loss": 0.7148,
+      "step": 449
+    },
+    {
+      "epoch": 1.948051948051948,
+      "grad_norm": 14.063284873962402,
+      "learning_rate": 5.472635933466248e-05,
+      "loss": 0.6094,
+      "step": 450
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 4.953294277191162,
+      "learning_rate": 5.455180354831351e-05,
+      "loss": 0.6953,
+      "step": 451
+    },
+    {
+      "epoch": 1.9567099567099566,
+      "grad_norm": 5.632043838500977,
+      "learning_rate": 5.43771918033709e-05,
+      "loss": 0.6914,
+      "step": 452
+    },
+    {
+      "epoch": 1.9610389610389611,
+      "grad_norm": 3.495882511138916,
+      "learning_rate": 5.420252624646238e-05,
+      "loss": 0.7539,
+      "step": 453
+    },
+    {
+      "epoch": 1.9653679653679652,
+      "grad_norm": 2.9653148651123047,
+      "learning_rate": 5.4027809024877204e-05,
+      "loss": 0.5781,
+      "step": 454
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 5.760696887969971,
+      "learning_rate": 5.3853042286539824e-05,
+      "loss": 0.7266,
+      "step": 455
+    },
+    {
+      "epoch": 1.974025974025974,
+      "grad_norm": 4.12476921081543,
+      "learning_rate": 5.367822817998338e-05,
+      "loss": 0.6953,
+      "step": 456
+    },
+    {
+      "epoch": 1.9783549783549783,
+      "grad_norm": 2.996063232421875,
+      "learning_rate": 5.3503368854323366e-05,
+      "loss": 0.6445,
+      "step": 457
+    },
+    {
+      "epoch": 1.9826839826839828,
+      "grad_norm": 3.6971347332000732,
+      "learning_rate": 5.332846645923122e-05,
+      "loss": 0.6641,
+      "step": 458
+    },
+    {
+      "epoch": 1.987012987012987,
+      "grad_norm": 3.8304529190063477,
+      "learning_rate": 5.31535231449078e-05,
+      "loss": 0.625,
+      "step": 459
+    },
+    {
+      "epoch": 1.9913419913419914,
+      "grad_norm": 3.9476420879364014,
+      "learning_rate": 5.297854106205708e-05,
+      "loss": 0.7227,
+      "step": 460
+    },
+    {
+      "epoch": 1.9956709956709957,
+      "grad_norm": 3.560429334640503,
+      "learning_rate": 5.2803522361859594e-05,
+      "loss": 0.7305,
+      "step": 461
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.96229887008667,
+      "learning_rate": 5.262846919594607e-05,
+      "loss": 0.5859,
+      "step": 462
+    },
+    {
+      "epoch": 2.0043290043290045,
+      "grad_norm": 3.0525240898132324,
+      "learning_rate": 5.245338371637091e-05,
+      "loss": 0.5391,
+      "step": 463
+    },
+    {
+      "epoch": 2.0086580086580086,
+      "grad_norm": 2.491299867630005,
+      "learning_rate": 5.2278268075585815e-05,
+      "loss": 0.5234,
+      "step": 464
+    },
+    {
+      "epoch": 2.012987012987013,
+      "grad_norm": 2.840453863143921,
+      "learning_rate": 5.2103124426413264e-05,
+      "loss": 0.6875,
+      "step": 465
+    },
+    {
+      "epoch": 2.017316017316017,
+      "grad_norm": 3.1786391735076904,
+      "learning_rate": 5.192795492202004e-05,
+      "loss": 0.5664,
+      "step": 466
+    },
+    {
+      "epoch": 2.0216450216450217,
+      "grad_norm": 3.5918378829956055,
+      "learning_rate": 5.175276171589082e-05,
+      "loss": 0.5664,
+      "step": 467
+    },
+    {
+      "epoch": 2.0259740259740258,
+      "grad_norm": 3.2464823722839355,
+      "learning_rate": 5.157754696180165e-05,
+      "loss": 0.6016,
+      "step": 468
+    },
+    {
+      "epoch": 2.0303030303030303,
+      "grad_norm": 5.0443501472473145,
+      "learning_rate": 5.140231281379345e-05,
+      "loss": 0.6445,
+      "step": 469
+    },
+    {
+      "epoch": 2.034632034632035,
+      "grad_norm": 3.5078518390655518,
+      "learning_rate": 5.122706142614562e-05,
+      "loss": 0.6367,
+      "step": 470
+    },
+    {
+      "epoch": 2.038961038961039,
+      "grad_norm": 5.425134181976318,
+      "learning_rate": 5.105179495334944e-05,
+      "loss": 0.6016,
+      "step": 471
+    },
+    {
+      "epoch": 2.0432900432900434,
+      "grad_norm": 5.983811855316162,
+      "learning_rate": 5.0876515550081716e-05,
+      "loss": 0.6523,
+      "step": 472
+    },
+    {
+      "epoch": 2.0476190476190474,
+      "grad_norm": 9.461071014404297,
+      "learning_rate": 5.070122537117812e-05,
+      "loss": 0.5664,
+      "step": 473
+    },
+    {
+      "epoch": 2.051948051948052,
+      "grad_norm": 3.252232789993286,
+      "learning_rate": 5.05259265716069e-05,
+      "loss": 0.5156,
+      "step": 474
+    },
+    {
+      "epoch": 2.0562770562770565,
+      "grad_norm": 4.877729415893555,
+      "learning_rate": 5.035062130644218e-05,
+      "loss": 0.5586,
+      "step": 475
+    },
+    {
+      "epoch": 2.0606060606060606,
+      "grad_norm": 27.354082107543945,
+      "learning_rate": 5.017531173083767e-05,
+      "loss": 0.6836,
+      "step": 476
+    },
+    {
+      "epoch": 2.064935064935065,
+      "grad_norm": 3.636356830596924,
+      "learning_rate": 5e-05,
+      "loss": 0.5469,
+      "step": 477
+    },
+    {
+      "epoch": 2.069264069264069,
+      "grad_norm": 17.058696746826172,
+      "learning_rate": 4.982468826916234e-05,
+      "loss": 0.6367,
+      "step": 478
+    },
+    {
+      "epoch": 2.0735930735930737,
+      "grad_norm": 6.493884563446045,
+      "learning_rate": 4.964937869355782e-05,
+      "loss": 0.5977,
+      "step": 479
+    },
+    {
+      "epoch": 2.0779220779220777,
+      "grad_norm": 7.743015766143799,
+      "learning_rate": 4.947407342839312e-05,
+      "loss": 0.5781,
+      "step": 480
+    },
+    {
+      "epoch": 2.0822510822510822,
+      "grad_norm": 4.716525077819824,
+      "learning_rate": 4.929877462882189e-05,
+      "loss": 0.5039,
+      "step": 481
+    },
+    {
+      "epoch": 2.0865800865800868,
+      "grad_norm": 6.498499393463135,
+      "learning_rate": 4.912348444991831e-05,
+      "loss": 0.5859,
+      "step": 482
+    },
+    {
+      "epoch": 2.090909090909091,
+      "grad_norm": 10.394189834594727,
+      "learning_rate": 4.894820504665056e-05,
+      "loss": 0.5273,
+      "step": 483
+    },
+    {
+      "epoch": 2.0952380952380953,
+      "grad_norm": 13.063650131225586,
+      "learning_rate": 4.877293857385439e-05,
+      "loss": 0.5625,
+      "step": 484
+    },
+    {
+      "epoch": 2.0995670995670994,
+      "grad_norm": 6.653393745422363,
+      "learning_rate": 4.859768718620656e-05,
+      "loss": 0.5703,
+      "step": 485
+    },
+    {
+      "epoch": 2.103896103896104,
+      "grad_norm": 5.993103981018066,
+      "learning_rate": 4.8422453038198364e-05,
+      "loss": 0.5859,
+      "step": 486
+    },
+    {
+      "epoch": 2.108225108225108,
+      "grad_norm": 5.752457141876221,
+      "learning_rate": 4.8247238284109194e-05,
+      "loss": 0.6406,
+      "step": 487
+    },
+    {
+      "epoch": 2.1125541125541125,
+      "grad_norm": 5.268364429473877,
+      "learning_rate": 4.8072045077979974e-05,
+      "loss": 0.6328,
+      "step": 488
+    },
+    {
+      "epoch": 2.116883116883117,
+      "grad_norm": 6.121761322021484,
+      "learning_rate": 4.7896875573586755e-05,
+      "loss": 0.6406,
+      "step": 489
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 5.07736349105835,
+      "learning_rate": 4.772173192441419e-05,
+      "loss": 0.5898,
+      "step": 490
+    },
+    {
+      "epoch": 2.1255411255411256,
+      "grad_norm": 9.910273551940918,
+      "learning_rate": 4.7546616283629105e-05,
+      "loss": 0.6094,
+      "step": 491
+    },
+    {
+      "epoch": 2.1298701298701297,
+      "grad_norm": 10.291518211364746,
+      "learning_rate": 4.7371530804053954e-05,
+      "loss": 0.625,
+      "step": 492
+    },
+    {
+      "epoch": 2.134199134199134,
+      "grad_norm": 8.513075828552246,
+      "learning_rate": 4.7196477638140404e-05,
+      "loss": 0.6016,
+      "step": 493
+    },
+    {
+      "epoch": 2.1385281385281387,
+      "grad_norm": 6.122551441192627,
+      "learning_rate": 4.702145893794292e-05,
+      "loss": 0.5547,
+      "step": 494
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 4.69831657409668,
+      "learning_rate": 4.684647685509221e-05,
+      "loss": 0.5938,
+      "step": 495
+    },
+    {
+      "epoch": 2.1471861471861473,
+      "grad_norm": 5.083981513977051,
+      "learning_rate": 4.6671533540768804e-05,
+      "loss": 0.5625,
+      "step": 496
+    },
+    {
+      "epoch": 2.1515151515151514,
+      "grad_norm": 5.303883075714111,
+      "learning_rate": 4.649663114567663e-05,
+      "loss": 0.5977,
+      "step": 497
+    },
+    {
+      "epoch": 2.155844155844156,
+      "grad_norm": 6.937793731689453,
+      "learning_rate": 4.6321771820016636e-05,
+      "loss": 0.5547,
+      "step": 498
+    },
+    {
+      "epoch": 2.16017316017316,
+      "grad_norm": 8.676101684570312,
+      "learning_rate": 4.6146957713460195e-05,
+      "loss": 0.4902,
+      "step": 499
+    },
+    {
+      "epoch": 2.1645021645021645,
+      "grad_norm": 5.4890217781066895,
+      "learning_rate": 4.597219097512279e-05,
+      "loss": 0.6094,
+      "step": 500
+    },
+    {
+      "epoch": 2.168831168831169,
+      "grad_norm": 5.979981422424316,
+      "learning_rate": 4.579747375353763e-05,
+      "loss": 0.4727,
+      "step": 501
+    },
+    {
+      "epoch": 2.173160173160173,
+      "grad_norm": 5.369205951690674,
+      "learning_rate": 4.562280819662911e-05,
+      "loss": 0.5664,
+      "step": 502
+    },
+    {
+      "epoch": 2.1774891774891776,
+      "grad_norm": 4.383188724517822,
+      "learning_rate": 4.54481964516865e-05,
+      "loss": 0.5312,
+      "step": 503
+    },
+    {
+      "epoch": 2.1818181818181817,
+      "grad_norm": 5.144348621368408,
+      "learning_rate": 4.5273640665337525e-05,
+      "loss": 0.5273,
+      "step": 504
+    },
+    {
+      "epoch": 2.186147186147186,
+      "grad_norm": 6.158354759216309,
+      "learning_rate": 4.509914298352197e-05,
+      "loss": 0.5703,
+      "step": 505
+    },
+    {
+      "epoch": 2.1904761904761907,
+      "grad_norm": 6.2422895431518555,
+      "learning_rate": 4.492470555146531e-05,
+      "loss": 0.5859,
+      "step": 506
+    },
+    {
+      "epoch": 2.1948051948051948,
+      "grad_norm": 6.819314956665039,
+      "learning_rate": 4.475033051365228e-05,
+      "loss": 0.6016,
+      "step": 507
+    },
+    {
+      "epoch": 2.1991341991341993,
+      "grad_norm": 8.764884948730469,
+      "learning_rate": 4.457602001380061e-05,
+      "loss": 0.5977,
+      "step": 508
+    },
+    {
+      "epoch": 2.2034632034632033,
+      "grad_norm": 8.94918155670166,
+      "learning_rate": 4.4401776194834613e-05,
+      "loss": 0.6055,
+      "step": 509
+    },
+    {
+      "epoch": 2.207792207792208,
+      "grad_norm": 6.284018516540527,
+      "learning_rate": 4.422760119885881e-05,
+      "loss": 0.4688,
+      "step": 510
+    },
+    {
+      "epoch": 2.212121212121212,
+      "grad_norm": 6.549793243408203,
+      "learning_rate": 4.405349716713165e-05,
+      "loss": 0.4824,
+      "step": 511
+    },
+    {
+      "epoch": 2.2164502164502164,
+      "grad_norm": 11.82210636138916,
+      "learning_rate": 4.3879466240039196e-05,
+      "loss": 0.668,
+      "step": 512
+    },
+    {
+      "epoch": 2.220779220779221,
+      "grad_norm": 7.305809497833252,
+      "learning_rate": 4.370551055706874e-05,
+      "loss": 0.5078,
+      "step": 513
+    },
+    {
+      "epoch": 2.225108225108225,
+      "grad_norm": 27.957542419433594,
+      "learning_rate": 4.3531632256782554e-05,
+      "loss": 0.6094,
+      "step": 514
+    },
+    {
+      "epoch": 2.2294372294372296,
+      "grad_norm": 6.99340295791626,
+      "learning_rate": 4.335783347679162e-05,
+      "loss": 0.6328,
+      "step": 515
+    },
+    {
+      "epoch": 2.2337662337662336,
+      "grad_norm": 75.38726806640625,
+      "learning_rate": 4.318411635372929e-05,
+      "loss": 0.5781,
+      "step": 516
+    },
+    {
+      "epoch": 2.238095238095238,
+      "grad_norm": 8.452545166015625,
+      "learning_rate": 4.3010483023225045e-05,
+      "loss": 0.4707,
+      "step": 517
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 9.213700294494629,
+      "learning_rate": 4.283693561987824e-05,
+      "loss": 0.5859,
+      "step": 518
+    },
+    {
+      "epoch": 2.2467532467532467,
+      "grad_norm": 4.827883243560791,
+      "learning_rate": 4.2663476277231915e-05,
+      "loss": 0.5117,
+      "step": 519
+    },
+    {
+      "epoch": 2.2510822510822512,
+      "grad_norm": 5.263610363006592,
+      "learning_rate": 4.249010712774648e-05,
+      "loss": 0.5312,
+      "step": 520
+    },
+    {
+      "epoch": 2.2554112554112553,
+      "grad_norm": 4.740518093109131,
+      "learning_rate": 4.231683030277349e-05,
+      "loss": 0.5195,
+      "step": 521
+    },
+    {
+      "epoch": 2.25974025974026,
+      "grad_norm": 4.616405963897705,
+      "learning_rate": 4.21436479325296e-05,
+      "loss": 0.5586,
+      "step": 522
+    },
+    {
+      "epoch": 2.264069264069264,
+      "grad_norm": 6.0123724937438965,
+      "learning_rate": 4.197056214607015e-05,
+      "loss": 0.5508,
+      "step": 523
+    },
+    {
+      "epoch": 2.2683982683982684,
+      "grad_norm": 22.863683700561523,
+      "learning_rate": 4.179757507126319e-05,
+      "loss": 0.5859,
+      "step": 524
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 7.452154636383057,
+      "learning_rate": 4.162468883476319e-05,
+      "loss": 0.6484,
+      "step": 525
+    },
+    {
+      "epoch": 2.277056277056277,
+      "grad_norm": 5.031999588012695,
+      "learning_rate": 4.1451905561984936e-05,
+      "loss": 0.6055,
+      "step": 526
+    },
+    {
+      "epoch": 2.2813852813852815,
+      "grad_norm": 4.893764495849609,
+      "learning_rate": 4.127922737707747e-05,
+      "loss": 0.5547,
+      "step": 527
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 4.312823295593262,
+      "learning_rate": 4.110665640289782e-05,
+      "loss": 0.5898,
+      "step": 528
+    },
+    {
+      "epoch": 2.29004329004329,
+      "grad_norm": 20.578651428222656,
+      "learning_rate": 4.093419476098509e-05,
+      "loss": 0.5586,
+      "step": 529
+    },
+    {
+      "epoch": 2.2943722943722946,
+      "grad_norm": 8.519278526306152,
+      "learning_rate": 4.076184457153423e-05,
+      "loss": 0.5234,
+      "step": 530
+    },
+    {
+      "epoch": 2.2987012987012987,
+      "grad_norm": 5.39682674407959,
+      "learning_rate": 4.0589607953370007e-05,
+      "loss": 0.5938,
+      "step": 531
+    },
+    {
+      "epoch": 2.303030303030303,
+      "grad_norm": 7.878849029541016,
+      "learning_rate": 4.0417487023921005e-05,
+      "loss": 0.6484,
+      "step": 532
+    },
+    {
+      "epoch": 2.3073593073593073,
+      "grad_norm": 7.999024868011475,
+      "learning_rate": 4.0245483899193595e-05,
+      "loss": 0.6016,
+      "step": 533
+    },
+    {
+      "epoch": 2.311688311688312,
+      "grad_norm": 5.9216790199279785,
+      "learning_rate": 4.007360069374583e-05,
+      "loss": 0.6133,
+      "step": 534
+    },
+    {
+      "epoch": 2.316017316017316,
+      "grad_norm": 5.330780029296875,
+      "learning_rate": 3.990183952066151e-05,
+      "loss": 0.5469,
+      "step": 535
+    },
+    {
+      "epoch": 2.3203463203463204,
+      "grad_norm": 5.472855091094971,
+      "learning_rate": 3.973020249152428e-05,
+      "loss": 0.5742,
+      "step": 536
+    },
+    {
+      "epoch": 2.324675324675325,
+      "grad_norm": 11.009156227111816,
+      "learning_rate": 3.955869171639152e-05,
+      "loss": 0.5781,
+      "step": 537
+    },
+    {
+      "epoch": 2.329004329004329,
+      "grad_norm": 4.404223918914795,
+      "learning_rate": 3.9387309303768484e-05,
+      "loss": 0.5195,
+      "step": 538
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 6.7264933586120605,
+      "learning_rate": 3.921605736058238e-05,
+      "loss": 0.625,
+      "step": 539
+    },
+    {
+      "epoch": 2.3376623376623376,
+      "grad_norm": 14.011961936950684,
+      "learning_rate": 3.9044937992156516e-05,
+      "loss": 0.5508,
+      "step": 540
+    },
+    {
+      "epoch": 2.341991341991342,
+      "grad_norm": 8.004422187805176,
+      "learning_rate": 3.887395330218429e-05,
+      "loss": 0.5742,
+      "step": 541
+    },
+    {
+      "epoch": 2.346320346320346,
+      "grad_norm": 7.568243026733398,
+      "learning_rate": 3.870310539270342e-05,
+      "loss": 0.5352,
+      "step": 542
+    },
+    {
+      "epoch": 2.3506493506493507,
+      "grad_norm": 5.435425758361816,
+      "learning_rate": 3.8532396364070115e-05,
+      "loss": 0.5352,
+      "step": 543
+    },
+    {
+      "epoch": 2.354978354978355,
+      "grad_norm": 5.02336311340332,
+      "learning_rate": 3.836182831493318e-05,
+      "loss": 0.6094,
+      "step": 544
+    },
+    {
+      "epoch": 2.3593073593073592,
+      "grad_norm": 7.283963203430176,
+      "learning_rate": 3.81914033422083e-05,
+      "loss": 0.5859,
+      "step": 545
+    },
+    {
+      "epoch": 2.3636363636363638,
+      "grad_norm": 6.607391834259033,
+      "learning_rate": 3.802112354105215e-05,
+      "loss": 0.5703,
+      "step": 546
+    },
+    {
+      "epoch": 2.367965367965368,
+      "grad_norm": 4.596790790557861,
+      "learning_rate": 3.785099100483681e-05,
+      "loss": 0.5977,
+      "step": 547
+    },
+    {
+      "epoch": 2.3722943722943723,
+      "grad_norm": 5.824214935302734,
+      "learning_rate": 3.768100782512384e-05,
+      "loss": 0.6016,
+      "step": 548
+    },
+    {
+      "epoch": 2.3766233766233764,
+      "grad_norm": 6.330157279968262,
+      "learning_rate": 3.7511176091638653e-05,
+      "loss": 0.5898,
+      "step": 549
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 9.937956809997559,
+      "learning_rate": 3.73414978922449e-05,
+      "loss": 0.543,
+      "step": 550
+    },
+    {
+      "epoch": 2.3852813852813854,
+      "grad_norm": 4.026739597320557,
+      "learning_rate": 3.717197531291867e-05,
+      "loss": 0.5078,
+      "step": 551
+    },
+    {
+      "epoch": 2.3896103896103895,
+      "grad_norm": 7.218325614929199,
+      "learning_rate": 3.700261043772291e-05,
+      "loss": 0.4863,
+      "step": 552
+    },
+    {
+      "epoch": 2.393939393939394,
+      "grad_norm": 4.606830596923828,
+      "learning_rate": 3.683340534878176e-05,
+      "loss": 0.5547,
+      "step": 553
+    },
+    {
+      "epoch": 2.398268398268398,
+      "grad_norm": 5.661990642547607,
+      "learning_rate": 3.666436212625509e-05,
+      "loss": 0.6289,
+      "step": 554
+    },
+    {
+      "epoch": 2.4025974025974026,
+      "grad_norm": 4.7327728271484375,
+      "learning_rate": 3.649548284831275e-05,
+      "loss": 0.4844,
+      "step": 555
+    },
+    {
+      "epoch": 2.4069264069264067,
+      "grad_norm": 4.367818355560303,
+      "learning_rate": 3.6326769591109086e-05,
+      "loss": 0.5781,
+      "step": 556
+    },
+    {
+      "epoch": 2.411255411255411,
+      "grad_norm": 4.332817554473877,
+      "learning_rate": 3.6158224428757535e-05,
+      "loss": 0.5664,
+      "step": 557
+    },
+    {
+      "epoch": 2.4155844155844157,
+      "grad_norm": 3.8692939281463623,
+      "learning_rate": 3.598984943330492e-05,
+      "loss": 0.5781,
+      "step": 558
+    },
+    {
+      "epoch": 2.41991341991342,
+      "grad_norm": 4.324326038360596,
+      "learning_rate": 3.582164667470612e-05,
+      "loss": 0.625,
+      "step": 559
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 6.796741008758545,
+      "learning_rate": 3.5653618220798575e-05,
+      "loss": 0.5742,
+      "step": 560
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 4.5253214836120605,
+      "learning_rate": 3.5485766137276894e-05,
+      "loss": 0.5312,
+      "step": 561
+    },
+    {
+      "epoch": 2.432900432900433,
+      "grad_norm": 4.087986946105957,
+      "learning_rate": 3.5318092487667396e-05,
+      "loss": 0.4941,
+      "step": 562
+    },
+    {
+      "epoch": 2.4372294372294374,
+      "grad_norm": 5.951512336730957,
+      "learning_rate": 3.515059933330282e-05,
+      "loss": 0.5742,
+      "step": 563
+    },
+    {
+      "epoch": 2.4415584415584415,
+      "grad_norm": 3.9411423206329346,
+      "learning_rate": 3.498328873329695e-05,
+      "loss": 0.5625,
+      "step": 564
+    },
+    {
+      "epoch": 2.445887445887446,
+      "grad_norm": 5.194113731384277,
+      "learning_rate": 3.4816162744519263e-05,
+      "loss": 0.6016,
+      "step": 565
+    },
+    {
+      "epoch": 2.45021645021645,
+      "grad_norm": 7.98148775100708,
+      "learning_rate": 3.4649223421569724e-05,
+      "loss": 0.6016,
+      "step": 566
+    },
+    {
+      "epoch": 2.4545454545454546,
+      "grad_norm": 3.878694534301758,
+      "learning_rate": 3.44824728167534e-05,
+      "loss": 0.5234,
+      "step": 567
+    },
+    {
+      "epoch": 2.458874458874459,
+      "grad_norm": 5.260382652282715,
+      "learning_rate": 3.4315912980055434e-05,
+      "loss": 0.6094,
+      "step": 568
+    },
+    {
+      "epoch": 2.463203463203463,
+      "grad_norm": 6.376096248626709,
+      "learning_rate": 3.4149545959115605e-05,
+      "loss": 0.5898,
+      "step": 569
+    },
+    {
+      "epoch": 2.4675324675324677,
+      "grad_norm": 4.26661491394043,
+      "learning_rate": 3.398337379920331e-05,
+      "loss": 0.5117,
+      "step": 570
+    },
+    {
+      "epoch": 2.4718614718614718,
+      "grad_norm": 5.104504585266113,
+      "learning_rate": 3.3817398543192424e-05,
+      "loss": 0.5781,
+      "step": 571
+    },
+    {
+      "epoch": 2.4761904761904763,
+      "grad_norm": 6.565718173980713,
+      "learning_rate": 3.3651622231536085e-05,
+      "loss": 0.5117,
+      "step": 572
+    },
+    {
+      "epoch": 2.4805194805194803,
+      "grad_norm": 3.7023003101348877,
+      "learning_rate": 3.3486046902241664e-05,
+      "loss": 0.6914,
+      "step": 573
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 7.503228187561035,
+      "learning_rate": 3.3320674590845715e-05,
+      "loss": 0.6406,
+      "step": 574
+    },
+    {
+      "epoch": 2.4891774891774894,
+      "grad_norm": 3.6865549087524414,
+      "learning_rate": 3.3155507330389e-05,
+      "loss": 0.5859,
+      "step": 575
+    },
+    {
+      "epoch": 2.4935064935064934,
+      "grad_norm": 5.931582450866699,
+      "learning_rate": 3.2990547151391395e-05,
+      "loss": 0.5195,
+      "step": 576
+    },
+    {
+      "epoch": 2.497835497835498,
+      "grad_norm": 5.7255072593688965,
+      "learning_rate": 3.282579608182694e-05,
+      "loss": 0.6797,
+      "step": 577
+    },
+    {
+      "epoch": 2.502164502164502,
+      "grad_norm": 3.8758506774902344,
+      "learning_rate": 3.266125614709905e-05,
+      "loss": 0.5664,
+      "step": 578
+    },
+    {
+      "epoch": 2.5064935064935066,
+      "grad_norm": 4.926198482513428,
+      "learning_rate": 3.249692937001538e-05,
+      "loss": 0.5547,
+      "step": 579
+    },
+    {
+      "epoch": 2.5108225108225106,
+      "grad_norm": 8.248186111450195,
+      "learning_rate": 3.2332817770763204e-05,
+      "loss": 0.5781,
+      "step": 580
+    },
+    {
+      "epoch": 2.515151515151515,
+      "grad_norm": 4.158359050750732,
+      "learning_rate": 3.216892336688435e-05,
+      "loss": 0.5078,
+      "step": 581
+    },
+    {
+      "epoch": 2.5194805194805197,
+      "grad_norm": 4.692091941833496,
+      "learning_rate": 3.200524817325059e-05,
+      "loss": 0.5039,
+      "step": 582
+    },
+    {
+      "epoch": 2.5238095238095237,
+      "grad_norm": 4.450368881225586,
+      "learning_rate": 3.1841794202038765e-05,
+      "loss": 0.6094,
+      "step": 583
+    },
+    {
+      "epoch": 2.5281385281385282,
+      "grad_norm": 4.560518264770508,
+      "learning_rate": 3.167856346270602e-05,
+      "loss": 0.5664,
+      "step": 584
+    },
+    {
+      "epoch": 2.5324675324675323,
+      "grad_norm": 5.9654951095581055,
+      "learning_rate": 3.151555796196525e-05,
+      "loss": 0.5977,
+      "step": 585
+    },
+    {
+      "epoch": 2.536796536796537,
+      "grad_norm": 6.0134172439575195,
+      "learning_rate": 3.1352779703760236e-05,
+      "loss": 0.4863,
+      "step": 586
+    },
+    {
+      "epoch": 2.541125541125541,
+      "grad_norm": 13.266544342041016,
+      "learning_rate": 3.119023068924115e-05,
+      "loss": 0.5391,
+      "step": 587
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 3.8508310317993164,
+      "learning_rate": 3.102791291673987e-05,
+      "loss": 0.5078,
+      "step": 588
+    },
+    {
+      "epoch": 2.54978354978355,
+      "grad_norm": 3.2759432792663574,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 0.5469,
+      "step": 589
+    },
+    {
+      "epoch": 2.554112554112554,
+      "grad_norm": 6.6946539878845215,
+      "learning_rate": 3.070397907687979e-05,
+      "loss": 0.5234,
+      "step": 590
+    },
+    {
+      "epoch": 2.5584415584415585,
+      "grad_norm": 5.478668212890625,
+      "learning_rate": 3.054236699187255e-05,
+      "loss": 0.5312,
+      "step": 591
+    },
+    {
+      "epoch": 2.562770562770563,
+      "grad_norm": 5.531819820404053,
+      "learning_rate": 3.0380994113537377e-05,
+      "loss": 0.5742,
+      "step": 592
+    },
+    {
+      "epoch": 2.567099567099567,
+      "grad_norm": 5.906157970428467,
+      "learning_rate": 3.021986242574707e-05,
+      "loss": 0.5859,
+      "step": 593
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 3.8638484477996826,
+      "learning_rate": 3.005897390940933e-05,
+      "loss": 0.5312,
+      "step": 594
+    },
+    {
+      "epoch": 2.5757575757575757,
+      "grad_norm": 52.914833068847656,
+      "learning_rate": 2.9898330542442322e-05,
+      "loss": 0.582,
+      "step": 595
+    },
+    {
+      "epoch": 2.58008658008658,
+      "grad_norm": 6.1610798835754395,
+      "learning_rate": 2.9737934299750515e-05,
+      "loss": 0.5469,
+      "step": 596
+    },
+    {
+      "epoch": 2.5844155844155843,
+      "grad_norm": 4.666285514831543,
+      "learning_rate": 2.9577787153200197e-05,
+      "loss": 0.5742,
+      "step": 597
+    },
+    {
+      "epoch": 2.588744588744589,
+      "grad_norm": 5.322575569152832,
+      "learning_rate": 2.94178910715954e-05,
+      "loss": 0.625,
+      "step": 598
+    },
+    {
+      "epoch": 2.5930735930735933,
+      "grad_norm": 6.213116645812988,
+      "learning_rate": 2.9258248020653616e-05,
+      "loss": 0.6133,
+      "step": 599
+    },
+    {
+      "epoch": 2.5974025974025974,
+      "grad_norm": 7.3725128173828125,
+      "learning_rate": 2.909885996298165e-05,
+      "loss": 0.6367,
+      "step": 600
+    },
+    {
+      "epoch": 2.601731601731602,
+      "grad_norm": 4.934314250946045,
+      "learning_rate": 2.893972885805148e-05,
+      "loss": 0.6484,
+      "step": 601
+    },
+    {
+      "epoch": 2.606060606060606,
+      "grad_norm": 6.257793426513672,
+      "learning_rate": 2.8780856662176183e-05,
+      "loss": 0.5234,
+      "step": 602
+    },
+    {
+      "epoch": 2.6103896103896105,
+      "grad_norm": 4.658661365509033,
+      "learning_rate": 2.8622245328485907e-05,
+      "loss": 0.5859,
+      "step": 603
+    },
+    {
+      "epoch": 2.6147186147186146,
+      "grad_norm": 4.09123420715332,
+      "learning_rate": 2.8463896806903812e-05,
+      "loss": 0.5078,
+      "step": 604
+    },
+    {
+      "epoch": 2.619047619047619,
+      "grad_norm": 3.532109022140503,
+      "learning_rate": 2.8305813044122097e-05,
+      "loss": 0.582,
+      "step": 605
+    },
+    {
+      "epoch": 2.6233766233766236,
+      "grad_norm": 12.364994049072266,
+      "learning_rate": 2.8147995983578157e-05,
+      "loss": 0.4941,
+      "step": 606
+    },
+    {
+      "epoch": 2.6277056277056277,
+      "grad_norm": 4.307883262634277,
+      "learning_rate": 2.799044756543054e-05,
+      "loss": 0.6094,
+      "step": 607
+    },
+    {
+      "epoch": 2.632034632034632,
+      "grad_norm": 4.780121326446533,
+      "learning_rate": 2.783316972653526e-05,
+      "loss": 0.6094,
+      "step": 608
+    },
+    {
+      "epoch": 2.6363636363636362,
+      "grad_norm": 5.936129570007324,
+      "learning_rate": 2.7676164400421862e-05,
+      "loss": 0.5312,
+      "step": 609
+    },
+    {
+      "epoch": 2.6406926406926408,
+      "grad_norm": 3.988802671432495,
+      "learning_rate": 2.7519433517269665e-05,
+      "loss": 0.5781,
+      "step": 610
+    },
+    {
+      "epoch": 2.645021645021645,
+      "grad_norm": 5.583521842956543,
+      "learning_rate": 2.7362979003884152e-05,
+      "loss": 0.5039,
+      "step": 611
+    },
+    {
+      "epoch": 2.6493506493506493,
+      "grad_norm": 4.771373748779297,
+      "learning_rate": 2.72068027836731e-05,
+      "loss": 0.5664,
+      "step": 612
+    },
+    {
+      "epoch": 2.653679653679654,
+      "grad_norm": 4.412400722503662,
+      "learning_rate": 2.705090677662311e-05,
+      "loss": 0.4727,
+      "step": 613
+    },
+    {
+      "epoch": 2.658008658008658,
+      "grad_norm": 6.180281162261963,
+      "learning_rate": 2.689529289927586e-05,
+      "loss": 0.5859,
+      "step": 614
+    },
+    {
+      "epoch": 2.6623376623376624,
+      "grad_norm": 3.7497949600219727,
+      "learning_rate": 2.673996306470463e-05,
+      "loss": 0.5508,
+      "step": 615
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 4.117935657501221,
+      "learning_rate": 2.658491918249073e-05,
+      "loss": 0.5664,
+      "step": 616
+    },
+    {
+      "epoch": 2.670995670995671,
+      "grad_norm": 3.6996371746063232,
+      "learning_rate": 2.6430163158700115e-05,
+      "loss": 0.5195,
+      "step": 617
+    },
+    {
+      "epoch": 2.675324675324675,
+      "grad_norm": 4.7948832511901855,
+      "learning_rate": 2.6275696895859835e-05,
+      "loss": 0.5781,
+      "step": 618
+    },
+    {
+      "epoch": 2.6796536796536796,
+      "grad_norm": 6.667705059051514,
+      "learning_rate": 2.6121522292934697e-05,
+      "loss": 0.6211,
+      "step": 619
+    },
+    {
+      "epoch": 2.683982683982684,
+      "grad_norm": 5.253415584564209,
+      "learning_rate": 2.596764124530398e-05,
+      "loss": 0.5547,
+      "step": 620
+    },
+    {
+      "epoch": 2.688311688311688,
+      "grad_norm": 3.140655755996704,
+      "learning_rate": 2.581405564473801e-05,
+      "loss": 0.5195,
+      "step": 621
+    },
+    {
+      "epoch": 2.6926406926406927,
+      "grad_norm": 8.399757385253906,
+      "learning_rate": 2.5660767379375018e-05,
+      "loss": 0.5586,
+      "step": 622
+    },
+    {
+      "epoch": 2.6969696969696972,
+      "grad_norm": 8.054708480834961,
+      "learning_rate": 2.5507778333697818e-05,
+      "loss": 0.5898,
+      "step": 623
+    },
+    {
+      "epoch": 2.7012987012987013,
+      "grad_norm": 3.615949869155884,
+      "learning_rate": 2.5355090388510804e-05,
+      "loss": 0.6133,
+      "step": 624
+    },
+    {
+      "epoch": 2.7056277056277054,
+      "grad_norm": 4.037124156951904,
+      "learning_rate": 2.5202705420916627e-05,
+      "loss": 0.7031,
+      "step": 625
+    },
+    {
+      "epoch": 2.70995670995671,
+      "grad_norm": 4.980218887329102,
+      "learning_rate": 2.505062530429324e-05,
+      "loss": 0.5742,
+      "step": 626
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 3.3744056224823,
+      "learning_rate": 2.4898851908270904e-05,
+      "loss": 0.5703,
+      "step": 627
+    },
+    {
+      "epoch": 2.7186147186147185,
+      "grad_norm": 3.059828519821167,
+      "learning_rate": 2.474738709870907e-05,
+      "loss": 0.6133,
+      "step": 628
+    },
+    {
+      "epoch": 2.722943722943723,
+      "grad_norm": 4.344146251678467,
+      "learning_rate": 2.459623273767354e-05,
+      "loss": 0.6016,
+      "step": 629
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 4.098705768585205,
+      "learning_rate": 2.444539068341354e-05,
+      "loss": 0.5352,
+      "step": 630
+    },
+    {
+      "epoch": 2.7316017316017316,
+      "grad_norm": 3.279052495956421,
+      "learning_rate": 2.4294862790338917e-05,
+      "loss": 0.5273,
+      "step": 631
+    },
+    {
+      "epoch": 2.7359307359307357,
+      "grad_norm": 3.1682636737823486,
+      "learning_rate": 2.4144650908997286e-05,
+      "loss": 0.543,
+      "step": 632
+    },
+    {
+      "epoch": 2.74025974025974,
+      "grad_norm": 3.2705435752868652,
+      "learning_rate": 2.3994756886051268e-05,
+      "loss": 0.5273,
+      "step": 633
+    },
+    {
+      "epoch": 2.7445887445887447,
+      "grad_norm": 3.243323802947998,
+      "learning_rate": 2.38451825642559e-05,
+      "loss": 0.5352,
+      "step": 634
+    },
+    {
+      "epoch": 2.7489177489177488,
+      "grad_norm": 4.071381092071533,
+      "learning_rate": 2.3695929782435844e-05,
+      "loss": 0.6289,
+      "step": 635
+    },
+    {
+      "epoch": 2.7532467532467533,
+      "grad_norm": 5.3187994956970215,
+      "learning_rate": 2.354700037546285e-05,
+      "loss": 0.5312,
+      "step": 636
+    },
+    {
+      "epoch": 2.757575757575758,
+      "grad_norm": 3.202204942703247,
+      "learning_rate": 2.3398396174233178e-05,
+      "loss": 0.625,
+      "step": 637
+    },
+    {
+      "epoch": 2.761904761904762,
+      "grad_norm": 3.226416826248169,
+      "learning_rate": 2.3250119005645147e-05,
+      "loss": 0.5156,
+      "step": 638
+    },
+    {
+      "epoch": 2.7662337662337664,
+      "grad_norm": 2.6866841316223145,
+      "learning_rate": 2.3102170692576593e-05,
+      "loss": 0.5625,
+      "step": 639
+    },
+    {
+      "epoch": 2.7705627705627704,
+      "grad_norm": 3.2219960689544678,
+      "learning_rate": 2.2954553053862466e-05,
+      "loss": 0.5156,
+      "step": 640
+    },
+    {
+      "epoch": 2.774891774891775,
+      "grad_norm": 5.62014102935791,
+      "learning_rate": 2.280726790427258e-05,
+      "loss": 0.6211,
+      "step": 641
+    },
+    {
+      "epoch": 2.779220779220779,
+      "grad_norm": 3.118605613708496,
+      "learning_rate": 2.2660317054489145e-05,
+      "loss": 0.5078,
+      "step": 642
+    },
+    {
+      "epoch": 2.7835497835497836,
+      "grad_norm": 4.178648948669434,
+      "learning_rate": 2.2513702311084596e-05,
+      "loss": 0.5469,
+      "step": 643
+    },
+    {
+      "epoch": 2.787878787878788,
+      "grad_norm": 3.177177906036377,
+      "learning_rate": 2.236742547649942e-05,
+      "loss": 0.5391,
+      "step": 644
+    },
+    {
+      "epoch": 2.792207792207792,
+      "grad_norm": 2.7220258712768555,
+      "learning_rate": 2.2221488349019903e-05,
+      "loss": 0.5742,
+      "step": 645
+    },
+    {
+      "epoch": 2.7965367965367967,
+      "grad_norm": 3.851757526397705,
+      "learning_rate": 2.2075892722756054e-05,
+      "loss": 0.5039,
+      "step": 646
+    },
+    {
+      "epoch": 2.8008658008658007,
+      "grad_norm": 4.253355026245117,
+      "learning_rate": 2.1930640387619616e-05,
+      "loss": 0.6406,
+      "step": 647
+    },
+    {
+      "epoch": 2.8051948051948052,
+      "grad_norm": 2.6334195137023926,
+      "learning_rate": 2.1785733129301955e-05,
+      "loss": 0.5312,
+      "step": 648
+    },
+    {
+      "epoch": 2.8095238095238093,
+      "grad_norm": 2.6169304847717285,
+      "learning_rate": 2.164117272925221e-05,
+      "loss": 0.5234,
+      "step": 649
+    },
+    {
+      "epoch": 2.813852813852814,
+      "grad_norm": 3.360002040863037,
+      "learning_rate": 2.1496960964655288e-05,
+      "loss": 0.4668,
+      "step": 650
+    },
+    {
+      "epoch": 2.8181818181818183,
+      "grad_norm": 3.2552871704101562,
+      "learning_rate": 2.1353099608410093e-05,
+      "loss": 0.6211,
+      "step": 651
+    },
+    {
+      "epoch": 2.8225108225108224,
+      "grad_norm": 2.515920400619507,
+      "learning_rate": 2.1209590429107735e-05,
+      "loss": 0.4688,
+      "step": 652
+    },
+    {
+      "epoch": 2.826839826839827,
+      "grad_norm": 3.005284309387207,
+      "learning_rate": 2.1066435191009715e-05,
+      "loss": 0.5742,
+      "step": 653
+    },
+    {
+      "epoch": 2.8311688311688314,
+      "grad_norm": 2.7586934566497803,
+      "learning_rate": 2.0923635654026292e-05,
+      "loss": 0.5508,
+      "step": 654
+    },
+    {
+      "epoch": 2.8354978354978355,
+      "grad_norm": 4.918183326721191,
+      "learning_rate": 2.0781193573694867e-05,
+      "loss": 0.5742,
+      "step": 655
+    },
+    {
+      "epoch": 2.8398268398268396,
+      "grad_norm": 2.7554261684417725,
+      "learning_rate": 2.0639110701158337e-05,
+      "loss": 0.5469,
+      "step": 656
+    },
+    {
+      "epoch": 2.844155844155844,
+      "grad_norm": 3.0071940422058105,
+      "learning_rate": 2.0497388783143602e-05,
+      "loss": 0.6133,
+      "step": 657
+    },
+    {
+      "epoch": 2.8484848484848486,
+      "grad_norm": 3.413681745529175,
+      "learning_rate": 2.0356029561940077e-05,
+      "loss": 0.5625,
+      "step": 658
+    },
+    {
+      "epoch": 2.8528138528138527,
+      "grad_norm": 3.2392756938934326,
+      "learning_rate": 2.0215034775378332e-05,
+      "loss": 0.5781,
+      "step": 659
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 3.13970685005188,
+      "learning_rate": 2.0074406156808647e-05,
+      "loss": 0.4336,
+      "step": 660
+    },
+    {
+      "epoch": 2.8614718614718617,
+      "grad_norm": 3.217355251312256,
+      "learning_rate": 1.9934145435079702e-05,
+      "loss": 0.5508,
+      "step": 661
+    },
+    {
+      "epoch": 2.865800865800866,
+      "grad_norm": 5.194235324859619,
+      "learning_rate": 1.9794254334517443e-05,
+      "loss": 0.6758,
+      "step": 662
+    },
+    {
+      "epoch": 2.87012987012987,
+      "grad_norm": 2.9741697311401367,
+      "learning_rate": 1.965473457490372e-05,
+      "loss": 0.5234,
+      "step": 663
+    },
+    {
+      "epoch": 2.8744588744588744,
+      "grad_norm": 3.821227788925171,
+      "learning_rate": 1.9515587871455253e-05,
+      "loss": 0.6133,
+      "step": 664
+    },
+    {
+      "epoch": 2.878787878787879,
+      "grad_norm": 3.538374662399292,
+      "learning_rate": 1.9376815934802496e-05,
+      "loss": 0.6133,
+      "step": 665
+    },
+    {
+      "epoch": 2.883116883116883,
+      "grad_norm": 2.828070640563965,
+      "learning_rate": 1.9238420470968665e-05,
+      "loss": 0.5625,
+      "step": 666
+    },
+    {
+      "epoch": 2.8874458874458875,
+      "grad_norm": 2.487342357635498,
+      "learning_rate": 1.9100403181348686e-05,
+      "loss": 0.582,
+      "step": 667
+    },
+    {
+      "epoch": 2.891774891774892,
+      "grad_norm": 2.3850345611572266,
+      "learning_rate": 1.8962765762688312e-05,
+      "loss": 0.4824,
+      "step": 668
+    },
+    {
+      "epoch": 2.896103896103896,
+      "grad_norm": 2.993290662765503,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 0.5977,
+      "step": 669
+    },
+    {
+      "epoch": 2.9004329004329006,
+      "grad_norm": 3.269127130508423,
+      "learning_rate": 1.868863730185863e-05,
+      "loss": 0.5469,
+      "step": 670
+    },
+    {
+      "epoch": 2.9047619047619047,
+      "grad_norm": 2.541809320449829,
+      "learning_rate": 1.8552149629747555e-05,
+      "loss": 0.5898,
+      "step": 671
+    },
+    {
+      "epoch": 2.909090909090909,
+      "grad_norm": 3.9370598793029785,
+      "learning_rate": 1.8416048568671184e-05,
+      "loss": 0.6289,
+      "step": 672
+    },
+    {
+      "epoch": 2.9134199134199132,
+      "grad_norm": 3.6821393966674805,
+      "learning_rate": 1.8280335791817733e-05,
+      "loss": 0.5312,
+      "step": 673
+    },
+    {
+      "epoch": 2.9177489177489178,
+      "grad_norm": 3.546535015106201,
+      "learning_rate": 1.814501296760193e-05,
+      "loss": 0.543,
+      "step": 674
+    },
+    {
+      "epoch": 2.9220779220779223,
+      "grad_norm": 3.139725923538208,
+      "learning_rate": 1.8010081759644538e-05,
+      "loss": 0.5469,
+      "step": 675
+    },
+    {
+      "epoch": 2.9264069264069263,
+      "grad_norm": 3.3686389923095703,
+      "learning_rate": 1.787554382675193e-05,
+      "loss": 0.582,
+      "step": 676
+    },
+    {
+      "epoch": 2.930735930735931,
+      "grad_norm": 3.299114465713501,
+      "learning_rate": 1.774140082289563e-05,
+      "loss": 0.582,
+      "step": 677
+    },
+    {
+      "epoch": 2.935064935064935,
+      "grad_norm": 2.711329936981201,
+      "learning_rate": 1.7607654397192036e-05,
+      "loss": 0.6172,
+      "step": 678
+    },
+    {
+      "epoch": 2.9393939393939394,
+      "grad_norm": 3.074343204498291,
+      "learning_rate": 1.747430619388209e-05,
+      "loss": 0.6016,
+      "step": 679
+    },
+    {
+      "epoch": 2.9437229437229435,
+      "grad_norm": 2.7482223510742188,
+      "learning_rate": 1.7341357852311174e-05,
+      "loss": 0.6172,
+      "step": 680
+    },
+    {
+      "epoch": 2.948051948051948,
+      "grad_norm": 3.1472690105438232,
+      "learning_rate": 1.7208811006908798e-05,
+      "loss": 0.6211,
+      "step": 681
+    },
+    {
+      "epoch": 2.9523809523809526,
+      "grad_norm": 4.673397541046143,
+      "learning_rate": 1.7076667287168664e-05,
+      "loss": 0.5117,
+      "step": 682
+    },
+    {
+      "epoch": 2.9567099567099566,
+      "grad_norm": 4.071459770202637,
+      "learning_rate": 1.694492831762851e-05,
+      "loss": 0.6484,
+      "step": 683
+    },
+    {
+      "epoch": 2.961038961038961,
+      "grad_norm": 3.282602071762085,
+      "learning_rate": 1.6813595717850188e-05,
+      "loss": 0.6055,
+      "step": 684
+    },
+    {
+      "epoch": 2.965367965367965,
+      "grad_norm": 3.034823417663574,
+      "learning_rate": 1.6682671102399805e-05,
+      "loss": 0.5469,
+      "step": 685
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 4.059807777404785,
+      "learning_rate": 1.6552156080827757e-05,
+      "loss": 0.5938,
+      "step": 686
+    },
+    {
+      "epoch": 2.974025974025974,
+      "grad_norm": 5.692906856536865,
+      "learning_rate": 1.6422052257649078e-05,
+      "loss": 0.5391,
+      "step": 687
+    },
+    {
+      "epoch": 2.9783549783549783,
+      "grad_norm": 3.3726298809051514,
+      "learning_rate": 1.6292361232323584e-05,
+      "loss": 0.5742,
+      "step": 688
+    },
+    {
+      "epoch": 2.982683982683983,
+      "grad_norm": 3.0519375801086426,
+      "learning_rate": 1.6163084599236278e-05,
+      "loss": 0.6016,
+      "step": 689
+    },
+    {
+      "epoch": 2.987012987012987,
+      "grad_norm": 4.7093186378479,
+      "learning_rate": 1.603422394767779e-05,
+      "loss": 0.5508,
+      "step": 690
+    },
+    {
+      "epoch": 2.9913419913419914,
+      "grad_norm": 3.568066358566284,
+      "learning_rate": 1.590578086182475e-05,
+      "loss": 0.6094,
+      "step": 691
+    },
+    {
+      "epoch": 2.995670995670996,
+      "grad_norm": 2.9147307872772217,
+      "learning_rate": 1.5777756920720344e-05,
+      "loss": 0.6133,
+      "step": 692
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 2.635899066925049,
+      "learning_rate": 1.5650153698254916e-05,
+      "loss": 0.4414,
+      "step": 693
+    },
+    {
+      "epoch": 3.0043290043290045,
+      "grad_norm": 4.511540412902832,
+      "learning_rate": 1.5522972763146653e-05,
+      "loss": 0.4395,
+      "step": 694
+    },
+    {
+      "epoch": 3.0086580086580086,
+      "grad_norm": 3.0602030754089355,
+      "learning_rate": 1.539621567892222e-05,
+      "loss": 0.4336,
+      "step": 695
+    },
+    {
+      "epoch": 3.012987012987013,
+      "grad_norm": 3.3176186084747314,
+      "learning_rate": 1.5269884003897565e-05,
+      "loss": 0.5273,
+      "step": 696
+    },
+    {
+      "epoch": 3.017316017316017,
+      "grad_norm": 2.4772136211395264,
+      "learning_rate": 1.5143979291158838e-05,
+      "loss": 0.3672,
+      "step": 697
+    },
+    {
+      "epoch": 3.0216450216450217,
+      "grad_norm": 3.532236337661743,
+      "learning_rate": 1.5018503088543173e-05,
+      "loss": 0.5664,
+      "step": 698
+    },
+    {
+      "epoch": 3.0259740259740258,
+      "grad_norm": 2.550928831100464,
+      "learning_rate": 1.4893456938619744e-05,
+      "loss": 0.4062,
+      "step": 699
+    },
+    {
+      "epoch": 3.0303030303030303,
+      "grad_norm": 4.138421058654785,
+      "learning_rate": 1.4768842378670755e-05,
+      "loss": 0.4531,
+      "step": 700
+    },
+    {
+      "epoch": 3.034632034632035,
+      "grad_norm": 3.217855930328369,
+      "learning_rate": 1.4644660940672627e-05,
+      "loss": 0.5039,
+      "step": 701
+    },
+    {
+      "epoch": 3.038961038961039,
+      "grad_norm": 5.836677551269531,
+      "learning_rate": 1.4520914151277037e-05,
+      "loss": 0.5312,
+      "step": 702
+    },
+    {
+      "epoch": 3.0432900432900434,
+      "grad_norm": 4.747433185577393,
+      "learning_rate": 1.439760353179223e-05,
+      "loss": 0.4277,
+      "step": 703
+    },
+    {
+      "epoch": 3.0476190476190474,
+      "grad_norm": 3.0238213539123535,
+      "learning_rate": 1.4274730598164331e-05,
+      "loss": 0.4844,
+      "step": 704
+    },
+    {
+      "epoch": 3.051948051948052,
+      "grad_norm": 3.510369300842285,
+      "learning_rate": 1.4152296860958642e-05,
+      "loss": 0.4609,
+      "step": 705
+    },
+    {
+      "epoch": 3.0562770562770565,
+      "grad_norm": 4.309097766876221,
+      "learning_rate": 1.4030303825341123e-05,
+      "loss": 0.457,
+      "step": 706
+    },
+    {
+      "epoch": 3.0606060606060606,
+      "grad_norm": 3.5565972328186035,
+      "learning_rate": 1.3908752991059853e-05,
+      "loss": 0.4414,
+      "step": 707
+    },
+    {
+      "epoch": 3.064935064935065,
+      "grad_norm": 4.466691493988037,
+      "learning_rate": 1.3787645852426661e-05,
+      "loss": 0.4531,
+      "step": 708
+    },
+    {
+      "epoch": 3.069264069264069,
+      "grad_norm": 3.082961320877075,
+      "learning_rate": 1.3666983898298657e-05,
+      "loss": 0.5234,
+      "step": 709
+    },
+    {
+      "epoch": 3.0735930735930737,
+      "grad_norm": 3.772768020629883,
+      "learning_rate": 1.3546768612059985e-05,
+      "loss": 0.457,
+      "step": 710
+    },
+    {
+      "epoch": 3.0779220779220777,
+      "grad_norm": 3.4244332313537598,
+      "learning_rate": 1.3427001471603623e-05,
+      "loss": 0.4844,
+      "step": 711
+    },
+    {
+      "epoch": 3.0822510822510822,
+      "grad_norm": 3.3153562545776367,
+      "learning_rate": 1.330768394931311e-05,
+      "loss": 0.4961,
+      "step": 712
+    },
+    {
+      "epoch": 3.0865800865800868,
+      "grad_norm": 4.436203479766846,
+      "learning_rate": 1.3188817512044544e-05,
+      "loss": 0.4258,
+      "step": 713
+    },
+    {
+      "epoch": 3.090909090909091,
+      "grad_norm": 2.813540458679199,
+      "learning_rate": 1.3070403621108485e-05,
+      "loss": 0.3926,
+      "step": 714
+    },
+    {
+      "epoch": 3.0952380952380953,
+      "grad_norm": 3.5020933151245117,
+      "learning_rate": 1.2952443732252057e-05,
+      "loss": 0.4688,
+      "step": 715
+    },
+    {
+      "epoch": 3.0995670995670994,
+      "grad_norm": 4.366274833679199,
+      "learning_rate": 1.2834939295640963e-05,
+      "loss": 0.377,
+      "step": 716
+    },
+    {
+      "epoch": 3.103896103896104,
+      "grad_norm": 2.8157784938812256,
+      "learning_rate": 1.2717891755841722e-05,
+      "loss": 0.5703,
+      "step": 717
+    },
+    {
+      "epoch": 3.108225108225108,
+      "grad_norm": 4.122098445892334,
+      "learning_rate": 1.2601302551803918e-05,
+      "loss": 0.4707,
+      "step": 718
+    },
+    {
+      "epoch": 3.1125541125541125,
+      "grad_norm": 3.7382733821868896,
+      "learning_rate": 1.248517311684243e-05,
+      "loss": 0.4668,
+      "step": 719
+    },
+    {
+      "epoch": 3.116883116883117,
+      "grad_norm": 2.8245136737823486,
+      "learning_rate": 1.2369504878619926e-05,
+      "loss": 0.5586,
+      "step": 720
+    },
+    {
+      "epoch": 3.121212121212121,
+      "grad_norm": 3.5146944522857666,
+      "learning_rate": 1.225429925912921e-05,
+      "loss": 0.5508,
+      "step": 721
+    },
+    {
+      "epoch": 3.1255411255411256,
+      "grad_norm": 3.9176745414733887,
+      "learning_rate": 1.2139557674675772e-05,
+      "loss": 0.4609,
+      "step": 722
+    },
+    {
+      "epoch": 3.1298701298701297,
+      "grad_norm": 3.906538486480713,
+      "learning_rate": 1.2025281535860433e-05,
+      "loss": 0.4512,
+      "step": 723
+    },
+    {
+      "epoch": 3.134199134199134,
+      "grad_norm": 3.5793933868408203,
+      "learning_rate": 1.1911472247561906e-05,
+      "loss": 0.4043,
+      "step": 724
+    },
+    {
+      "epoch": 3.1385281385281387,
+      "grad_norm": 3.5270469188690186,
+      "learning_rate": 1.1798131208919627e-05,
+      "loss": 0.5,
+      "step": 725
+    },
+    {
+      "epoch": 3.142857142857143,
+      "grad_norm": 3.6847422122955322,
+      "learning_rate": 1.1685259813316451e-05,
+      "loss": 0.4844,
+      "step": 726
+    },
+    {
+      "epoch": 3.1471861471861473,
+      "grad_norm": 3.4018149375915527,
+      "learning_rate": 1.15728594483616e-05,
+      "loss": 0.4727,
+      "step": 727
+    },
+    {
+      "epoch": 3.1515151515151514,
+      "grad_norm": 4.949450969696045,
+      "learning_rate": 1.146093149587356e-05,
+      "loss": 0.4883,
+      "step": 728
+    },
+    {
+      "epoch": 3.155844155844156,
+      "grad_norm": 3.1321475505828857,
+      "learning_rate": 1.134947733186315e-05,
+      "loss": 0.4883,
+      "step": 729
+    },
+    {
+      "epoch": 3.16017316017316,
+      "grad_norm": 2.945213794708252,
+      "learning_rate": 1.1238498326516533e-05,
+      "loss": 0.4434,
+      "step": 730
+    },
+    {
+      "epoch": 3.1645021645021645,
+      "grad_norm": 5.295434951782227,
+      "learning_rate": 1.1127995844178386e-05,
+      "loss": 0.4727,
+      "step": 731
+    },
+    {
+      "epoch": 3.168831168831169,
+      "grad_norm": 11.81176471710205,
+      "learning_rate": 1.1017971243335213e-05,
+      "loss": 0.457,
+      "step": 732
+    },
+    {
+      "epoch": 3.173160173160173,
+      "grad_norm": 3.3556666374206543,
+      "learning_rate": 1.090842587659851e-05,
+      "loss": 0.5352,
+      "step": 733
+    },
+    {
+      "epoch": 3.1774891774891776,
+      "grad_norm": 3.9434080123901367,
+      "learning_rate": 1.0799361090688249e-05,
+      "loss": 0.4785,
+      "step": 734
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 4.893848419189453,
+      "learning_rate": 1.069077822641622e-05,
+      "loss": 0.5117,
+      "step": 735
+    },
+    {
+      "epoch": 3.186147186147186,
+      "grad_norm": 6.217100143432617,
+      "learning_rate": 1.058267861866969e-05,
+      "loss": 0.457,
+      "step": 736
+    },
+    {
+      "epoch": 3.1904761904761907,
+      "grad_norm": 4.884580612182617,
+      "learning_rate": 1.047506359639483e-05,
+      "loss": 0.5391,
+      "step": 737
+    },
+    {
+      "epoch": 3.1948051948051948,
+      "grad_norm": 5.53536319732666,
+      "learning_rate": 1.0367934482580477e-05,
+      "loss": 0.5391,
+      "step": 738
+    },
+    {
+      "epoch": 3.1991341991341993,
+      "grad_norm": 2.760998249053955,
+      "learning_rate": 1.0261292594241872e-05,
+      "loss": 0.3906,
+      "step": 739
+    },
+    {
+      "epoch": 3.2034632034632033,
+      "grad_norm": 6.1857757568359375,
+      "learning_rate": 1.0155139242404404e-05,
+      "loss": 0.4863,
+      "step": 740
+    },
+    {
+      "epoch": 3.207792207792208,
+      "grad_norm": 2.888065814971924,
+      "learning_rate": 1.004947573208756e-05,
+      "loss": 0.4316,
+      "step": 741
+    },
+    {
+      "epoch": 3.212121212121212,
+      "grad_norm": 3.313202381134033,
+      "learning_rate": 9.944303362288814e-06,
+      "loss": 0.5391,
+      "step": 742
+    },
+    {
+      "epoch": 3.2164502164502164,
+      "grad_norm": 2.9406816959381104,
+      "learning_rate": 9.83962342596776e-06,
+      "loss": 0.4863,
+      "step": 743
+    },
+    {
+      "epoch": 3.220779220779221,
+      "grad_norm": 4.829558849334717,
+      "learning_rate": 9.735437210030101e-06,
+      "loss": 0.5195,
+      "step": 744
+    },
+    {
+      "epoch": 3.225108225108225,
+      "grad_norm": 3.786214828491211,
+      "learning_rate": 9.63174599531188e-06,
+      "loss": 0.5078,
+      "step": 745
+    },
+    {
+      "epoch": 3.2294372294372296,
+      "grad_norm": 3.5768957138061523,
+      "learning_rate": 9.528551056563784e-06,
+      "loss": 0.4922,
+      "step": 746
+    },
+    {
+      "epoch": 3.2337662337662336,
+      "grad_norm": 3.251338005065918,
+      "learning_rate": 9.42585366243537e-06,
+      "loss": 0.4531,
+      "step": 747
+    },
+    {
+      "epoch": 3.238095238095238,
+      "grad_norm": 8.76626205444336,
+      "learning_rate": 9.323655075459542e-06,
+      "loss": 0.4805,
+      "step": 748
+    },
+    {
+      "epoch": 3.242424242424242,
+      "grad_norm": 3.621119260787964,
+      "learning_rate": 9.221956552036992e-06,
+      "loss": 0.5273,
+      "step": 749
+    },
+    {
+      "epoch": 3.2467532467532467,
+      "grad_norm": 4.533940315246582,
+      "learning_rate": 9.12075934242082e-06,
+      "loss": 0.5586,
+      "step": 750
+    },
+    {
+      "epoch": 3.2510822510822512,
+      "grad_norm": 3.458012580871582,
+      "learning_rate": 9.020064690701069e-06,
+      "loss": 0.4023,
+      "step": 751
+    },
+    {
+      "epoch": 3.2554112554112553,
+      "grad_norm": 3.893784999847412,
+      "learning_rate": 8.919873834789466e-06,
+      "loss": 0.5625,
+      "step": 752
+    },
+    {
+      "epoch": 3.25974025974026,
+      "grad_norm": 3.6409213542938232,
+      "learning_rate": 8.820188006404268e-06,
+      "loss": 0.4785,
+      "step": 753
+    },
+    {
+      "epoch": 3.264069264069264,
+      "grad_norm": 4.683751583099365,
+      "learning_rate": 8.721008431055028e-06,
+      "loss": 0.459,
+      "step": 754
+    },
+    {
+      "epoch": 3.2683982683982684,
+      "grad_norm": 4.942468643188477,
+      "learning_rate": 8.622336328027553e-06,
+      "loss": 0.4238,
+      "step": 755
+    },
+    {
+      "epoch": 3.2727272727272725,
+      "grad_norm": 3.584601402282715,
+      "learning_rate": 8.52417291036897e-06,
+      "loss": 0.5156,
+      "step": 756
+    },
+    {
+      "epoch": 3.277056277056277,
+      "grad_norm": 3.85836124420166,
+      "learning_rate": 8.426519384872733e-06,
+      "loss": 0.4727,
+      "step": 757
+    },
+    {
+      "epoch": 3.2813852813852815,
+      "grad_norm": 3.6912739276885986,
+      "learning_rate": 8.329376952063867e-06,
+      "loss": 0.4805,
+      "step": 758
+    },
+    {
+      "epoch": 3.2857142857142856,
+      "grad_norm": 3.348270893096924,
+      "learning_rate": 8.232746806184117e-06,
+      "loss": 0.5039,
+      "step": 759
+    },
+    {
+      "epoch": 3.29004329004329,
+      "grad_norm": 3.8931868076324463,
+      "learning_rate": 8.136630135177337e-06,
+      "loss": 0.4961,
+      "step": 760
+    },
+    {
+      "epoch": 3.2943722943722946,
+      "grad_norm": 3.81941556930542,
+      "learning_rate": 8.041028120674893e-06,
+      "loss": 0.5586,
+      "step": 761
+    },
+    {
+      "epoch": 3.2987012987012987,
+      "grad_norm": 3.5875916481018066,
+      "learning_rate": 7.945941937981067e-06,
+      "loss": 0.4668,
+      "step": 762
+    },
+    {
+      "epoch": 3.303030303030303,
+      "grad_norm": 3.5765793323516846,
+      "learning_rate": 7.851372756058662e-06,
+      "loss": 0.4316,
+      "step": 763
+    },
+    {
+      "epoch": 3.3073593073593073,
+      "grad_norm": 3.3665988445281982,
+      "learning_rate": 7.757321737514645e-06,
+      "loss": 0.5391,
+      "step": 764
+    },
+    {
+      "epoch": 3.311688311688312,
+      "grad_norm": 4.428253173828125,
+      "learning_rate": 7.663790038585793e-06,
+      "loss": 0.459,
+      "step": 765
+    },
+    {
+      "epoch": 3.316017316017316,
+      "grad_norm": 5.810364723205566,
+      "learning_rate": 7.570778809124518e-06,
+      "loss": 0.4355,
+      "step": 766
+    },
+    {
+      "epoch": 3.3203463203463204,
+      "grad_norm": 2.7046561241149902,
+      "learning_rate": 7.478289192584764e-06,
+      "loss": 0.4551,
+      "step": 767
+    },
+    {
+      "epoch": 3.324675324675325,
+      "grad_norm": 3.418262004852295,
+      "learning_rate": 7.38632232600786e-06,
+      "loss": 0.4727,
+      "step": 768
+    },
+    {
+      "epoch": 3.329004329004329,
+      "grad_norm": 6.23099946975708,
+      "learning_rate": 7.2948793400086315e-06,
+      "loss": 0.4551,
+      "step": 769
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 4.131659507751465,
+      "learning_rate": 7.203961358761435e-06,
+      "loss": 0.5117,
+      "step": 770
+    },
+    {
+      "epoch": 3.3376623376623376,
+      "grad_norm": 4.813685894012451,
+      "learning_rate": 7.1135694999864e-06,
+      "loss": 0.4648,
+      "step": 771
+    },
+    {
+      "epoch": 3.341991341991342,
+      "grad_norm": 4.755728721618652,
+      "learning_rate": 7.023704874935627e-06,
+      "loss": 0.5898,
+      "step": 772
+    },
+    {
+      "epoch": 3.346320346320346,
+      "grad_norm": 13.759763717651367,
+      "learning_rate": 6.934368588379553e-06,
+      "loss": 0.457,
+      "step": 773
+    },
+    {
+      "epoch": 3.3506493506493507,
+      "grad_norm": 5.033351421356201,
+      "learning_rate": 6.845561738593392e-06,
+      "loss": 0.4375,
+      "step": 774
+    },
+    {
+      "epoch": 3.354978354978355,
+      "grad_norm": 5.149282455444336,
+      "learning_rate": 6.757285417343601e-06,
+      "loss": 0.418,
+      "step": 775
+    },
+    {
+      "epoch": 3.3593073593073592,
+      "grad_norm": 4.288097381591797,
+      "learning_rate": 6.669540709874461e-06,
+      "loss": 0.5742,
+      "step": 776
+    },
+    {
+      "epoch": 3.3636363636363638,
+      "grad_norm": 3.4480297565460205,
+      "learning_rate": 6.582328694894729e-06,
+      "loss": 0.4043,
+      "step": 777
+    },
+    {
+      "epoch": 3.367965367965368,
+      "grad_norm": 4.7171478271484375,
+      "learning_rate": 6.495650444564433e-06,
+      "loss": 0.4824,
+      "step": 778
+    },
+    {
+      "epoch": 3.3722943722943723,
+      "grad_norm": 3.703752040863037,
+      "learning_rate": 6.409507024481615e-06,
+      "loss": 0.5195,
+      "step": 779
+    },
+    {
+      "epoch": 3.3766233766233764,
+      "grad_norm": 3.4580540657043457,
+      "learning_rate": 6.3238994936692605e-06,
+      "loss": 0.4336,
+      "step": 780
+    },
+    {
+      "epoch": 3.380952380952381,
+      "grad_norm": 5.736301422119141,
+      "learning_rate": 6.238828904562316e-06,
+      "loss": 0.4707,
+      "step": 781
+    },
+    {
+      "epoch": 3.3852813852813854,
+      "grad_norm": 2.9964687824249268,
+      "learning_rate": 6.154296302994689e-06,
+      "loss": 0.4863,
+      "step": 782
+    },
+    {
+      "epoch": 3.3896103896103895,
+      "grad_norm": 4.596395492553711,
+      "learning_rate": 6.070302728186428e-06,
+      "loss": 0.416,
+      "step": 783
+    },
+    {
+      "epoch": 3.393939393939394,
+      "grad_norm": 4.620414733886719,
+      "learning_rate": 5.986849212730916e-06,
+      "loss": 0.4922,
+      "step": 784
+    },
+    {
+      "epoch": 3.398268398268398,
+      "grad_norm": 3.7630326747894287,
+      "learning_rate": 5.903936782582253e-06,
+      "loss": 0.5078,
+      "step": 785
+    },
+    {
+      "epoch": 3.4025974025974026,
+      "grad_norm": 3.636543035507202,
+      "learning_rate": 5.821566457042538e-06,
+      "loss": 0.5586,
+      "step": 786
+    },
+    {
+      "epoch": 3.4069264069264067,
+      "grad_norm": 3.620269775390625,
+      "learning_rate": 5.739739248749398e-06,
+      "loss": 0.3457,
+      "step": 787
+    },
+    {
+      "epoch": 3.411255411255411,
+      "grad_norm": 4.358973503112793,
+      "learning_rate": 5.65845616366355e-06,
+      "loss": 0.5352,
+      "step": 788
+    },
+    {
+      "epoch": 3.4155844155844157,
+      "grad_norm": 2.7407584190368652,
+      "learning_rate": 5.577718201056392e-06,
+      "loss": 0.3926,
+      "step": 789
+    },
+    {
+      "epoch": 3.41991341991342,
+      "grad_norm": 4.396733283996582,
+      "learning_rate": 5.497526353497751e-06,
+      "loss": 0.5117,
+      "step": 790
+    },
+    {
+      "epoch": 3.4242424242424243,
+      "grad_norm": 3.3989651203155518,
+      "learning_rate": 5.41788160684365e-06,
+      "loss": 0.418,
+      "step": 791
+    },
+    {
+      "epoch": 3.4285714285714284,
+      "grad_norm": 3.453202247619629,
+      "learning_rate": 5.33878494022424e-06,
+      "loss": 0.5039,
+      "step": 792
+    },
+    {
+      "epoch": 3.432900432900433,
+      "grad_norm": 3.6048474311828613,
+      "learning_rate": 5.260237326031697e-06,
+      "loss": 0.4648,
+      "step": 793
+    },
+    {
+      "epoch": 3.4372294372294374,
+      "grad_norm": 3.296344757080078,
+      "learning_rate": 5.182239729908333e-06,
+      "loss": 0.5312,
+      "step": 794
+    },
+    {
+      "epoch": 3.4415584415584415,
+      "grad_norm": 3.8676748275756836,
+      "learning_rate": 5.104793110734668e-06,
+      "loss": 0.4336,
+      "step": 795
+    },
+    {
+      "epoch": 3.445887445887446,
+      "grad_norm": 3.5097694396972656,
+      "learning_rate": 5.027898420617666e-06,
+      "loss": 0.498,
+      "step": 796
+    },
+    {
+      "epoch": 3.45021645021645,
+      "grad_norm": 3.448263168334961,
+      "learning_rate": 4.951556604879048e-06,
+      "loss": 0.4648,
+      "step": 797
+    },
+    {
+      "epoch": 3.4545454545454546,
+      "grad_norm": 4.15477180480957,
+      "learning_rate": 4.875768602043629e-06,
+      "loss": 0.5,
+      "step": 798
+    },
+    {
+      "epoch": 3.458874458874459,
+      "grad_norm": 2.8995680809020996,
+      "learning_rate": 4.800535343827833e-06,
+      "loss": 0.4453,
+      "step": 799
+    },
+    {
+      "epoch": 3.463203463203463,
+      "grad_norm": 3.3531198501586914,
+      "learning_rate": 4.725857755128188e-06,
+      "loss": 0.5078,
+      "step": 800
+    },
+    {
+      "epoch": 3.4675324675324677,
+      "grad_norm": 3.3565032482147217,
+      "learning_rate": 4.651736754009972e-06,
+      "loss": 0.4609,
+      "step": 801
+    },
+    {
+      "epoch": 3.4718614718614718,
+      "grad_norm": 3.5413527488708496,
+      "learning_rate": 4.5781732516959605e-06,
+      "loss": 0.4355,
+      "step": 802
+    },
+    {
+      "epoch": 3.4761904761904763,
+      "grad_norm": 2.8934743404388428,
+      "learning_rate": 4.505168152555173e-06,
+      "loss": 0.3906,
+      "step": 803
+    },
+    {
+      "epoch": 3.4805194805194803,
+      "grad_norm": 4.9387664794921875,
+      "learning_rate": 4.432722354091778e-06,
+      "loss": 0.4844,
+      "step": 804
+    },
+    {
+      "epoch": 3.484848484848485,
+      "grad_norm": 4.1749444007873535,
+      "learning_rate": 4.360836746934055e-06,
+      "loss": 0.5039,
+      "step": 805
+    },
+    {
+      "epoch": 3.4891774891774894,
+      "grad_norm": 3.1115996837615967,
+      "learning_rate": 4.289512214823466e-06,
+      "loss": 0.457,
+      "step": 806
+    },
+    {
+      "epoch": 3.4935064935064934,
+      "grad_norm": 3.8295235633850098,
+      "learning_rate": 4.2187496346037695e-06,
+      "loss": 0.4395,
+      "step": 807
+    },
+    {
+      "epoch": 3.497835497835498,
+      "grad_norm": 4.879920482635498,
+      "learning_rate": 4.148549876210223e-06,
+      "loss": 0.4766,
+      "step": 808
+    },
+    {
+      "epoch": 3.502164502164502,
+      "grad_norm": 3.5300912857055664,
+      "learning_rate": 4.078913802658946e-06,
+      "loss": 0.4648,
+      "step": 809
+    },
+    {
+      "epoch": 3.5064935064935066,
+      "grad_norm": 2.6126158237457275,
+      "learning_rate": 4.009842270036252e-06,
+      "loss": 0.4395,
+      "step": 810
+    },
+    {
+      "epoch": 3.5108225108225106,
+      "grad_norm": 4.793015003204346,
+      "learning_rate": 3.941336127488149e-06,
+      "loss": 0.5312,
+      "step": 811
+    },
+    {
+      "epoch": 3.515151515151515,
+      "grad_norm": 4.388156414031982,
+      "learning_rate": 3.873396217209891e-06,
+      "loss": 0.4922,
+      "step": 812
+    },
+    {
+      "epoch": 3.5194805194805197,
+      "grad_norm": 2.92972731590271,
+      "learning_rate": 3.8060233744356633e-06,
+      "loss": 0.4609,
+      "step": 813
+    },
+    {
+      "epoch": 3.5238095238095237,
+      "grad_norm": 10.127378463745117,
+      "learning_rate": 3.7392184274282537e-06,
+      "loss": 0.4766,
+      "step": 814
+    },
+    {
+      "epoch": 3.5281385281385282,
+      "grad_norm": 3.8105361461639404,
+      "learning_rate": 3.6729821974688937e-06,
+      "loss": 0.4258,
+      "step": 815
+    },
+    {
+      "epoch": 3.5324675324675323,
+      "grad_norm": 4.038154602050781,
+      "learning_rate": 3.6073154988471967e-06,
+      "loss": 0.4609,
+      "step": 816
+    },
+    {
+      "epoch": 3.536796536796537,
+      "grad_norm": 2.8943018913269043,
+      "learning_rate": 3.542219138851094e-06,
+      "loss": 0.4492,
+      "step": 817
+    },
+    {
+      "epoch": 3.541125541125541,
+      "grad_norm": 3.5218379497528076,
+      "learning_rate": 3.4776939177569468e-06,
+      "loss": 0.4453,
+      "step": 818
+    },
+    {
+      "epoch": 3.5454545454545454,
+      "grad_norm": 2.75854229927063,
+      "learning_rate": 3.4137406288196725e-06,
+      "loss": 0.4258,
+      "step": 819
+    },
+    {
+      "epoch": 3.54978354978355,
+      "grad_norm": 2.9081122875213623,
+      "learning_rate": 3.350360058263058e-06,
+      "loss": 0.4082,
+      "step": 820
+    },
+    {
+      "epoch": 3.554112554112554,
+      "grad_norm": 3.7427148818969727,
+      "learning_rate": 3.2875529852700147e-06,
+      "loss": 0.5078,
+      "step": 821
+    },
+    {
+      "epoch": 3.5584415584415585,
+      "grad_norm": 4.302583694458008,
+      "learning_rate": 3.225320181973035e-06,
+      "loss": 0.4844,
+      "step": 822
+    },
+    {
+      "epoch": 3.562770562770563,
+      "grad_norm": 3.51001238822937,
+      "learning_rate": 3.1636624134447344e-06,
+      "loss": 0.4199,
+      "step": 823
+    },
+    {
+      "epoch": 3.567099567099567,
+      "grad_norm": 2.6731579303741455,
+      "learning_rate": 3.1025804376883793e-06,
+      "loss": 0.3555,
+      "step": 824
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 3.562731981277466,
+      "learning_rate": 3.0420750056286195e-06,
+      "loss": 0.4023,
+      "step": 825
+    },
+    {
+      "epoch": 3.5757575757575757,
+      "grad_norm": 3.564955711364746,
+      "learning_rate": 2.9821468611022198e-06,
+      "loss": 0.4922,
+      "step": 826
+    },
+    {
+      "epoch": 3.58008658008658,
+      "grad_norm": 3.912365674972534,
+      "learning_rate": 2.9227967408489653e-06,
+      "loss": 0.5117,
+      "step": 827
+    },
+    {
+      "epoch": 3.5844155844155843,
+      "grad_norm": 3.809464931488037,
+      "learning_rate": 2.8640253745025493e-06,
+      "loss": 0.5039,
+      "step": 828
+    },
+    {
+      "epoch": 3.588744588744589,
+      "grad_norm": 3.265876531600952,
+      "learning_rate": 2.8058334845816213e-06,
+      "loss": 0.4766,
+      "step": 829
+    },
+    {
+      "epoch": 3.5930735930735933,
+      "grad_norm": 5.604315757751465,
+      "learning_rate": 2.7482217864809477e-06,
+      "loss": 0.5977,
+      "step": 830
+    },
+    {
+      "epoch": 3.5974025974025974,
+      "grad_norm": 5.586942195892334,
+      "learning_rate": 2.691190988462522e-06,
+      "loss": 0.5156,
+      "step": 831
+    },
+    {
+      "epoch": 3.601731601731602,
+      "grad_norm": 11.440213203430176,
+      "learning_rate": 2.634741791646983e-06,
+      "loss": 0.5625,
+      "step": 832
+    },
+    {
+      "epoch": 3.606060606060606,
+      "grad_norm": 3.830442190170288,
+      "learning_rate": 2.5788748900048676e-06,
+      "loss": 0.4395,
+      "step": 833
+    },
+    {
+      "epoch": 3.6103896103896105,
+      "grad_norm": 4.997652530670166,
+      "learning_rate": 2.5235909703481665e-06,
+      "loss": 0.4258,
+      "step": 834
+    },
+    {
+      "epoch": 3.6147186147186146,
+      "grad_norm": 3.331897020339966,
+      "learning_rate": 2.468890712321864e-06,
+      "loss": 0.373,
+      "step": 835
+    },
+    {
+      "epoch": 3.619047619047619,
+      "grad_norm": 4.876908302307129,
+      "learning_rate": 2.4147747883955464e-06,
+      "loss": 0.4805,
+      "step": 836
+    },
+    {
+      "epoch": 3.6233766233766236,
+      "grad_norm": 3.1443872451782227,
+      "learning_rate": 2.361243863855184e-06,
+      "loss": 0.5039,
+      "step": 837
+    },
+    {
+      "epoch": 3.6277056277056277,
+      "grad_norm": 3.7315304279327393,
+      "learning_rate": 2.3082985967949167e-06,
+      "loss": 0.4375,
+      "step": 838
+    },
+    {
+      "epoch": 3.632034632034632,
+      "grad_norm": 4.784104824066162,
+      "learning_rate": 2.2559396381089838e-06,
+      "loss": 0.4062,
+      "step": 839
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 3.6795904636383057,
+      "learning_rate": 2.204167631483689e-06,
+      "loss": 0.459,
+      "step": 840
+    },
+    {
+      "epoch": 3.6406926406926408,
+      "grad_norm": 3.126923084259033,
+      "learning_rate": 2.152983213389559e-06,
+      "loss": 0.4648,
+      "step": 841
+    },
+    {
+      "epoch": 3.645021645021645,
+      "grad_norm": 3.659628391265869,
+      "learning_rate": 2.1023870130734257e-06,
+      "loss": 0.5312,
+      "step": 842
+    },
+    {
+      "epoch": 3.6493506493506493,
+      "grad_norm": 3.094866991043091,
+      "learning_rate": 2.0523796525507623e-06,
+      "loss": 0.4473,
+      "step": 843
+    },
+    {
+      "epoch": 3.653679653679654,
+      "grad_norm": 3.4081637859344482,
+      "learning_rate": 2.0029617465980142e-06,
+      "loss": 0.4961,
+      "step": 844
+    },
+    {
+      "epoch": 3.658008658008658,
+      "grad_norm": 3.06144380569458,
+      "learning_rate": 1.9541339027450256e-06,
+      "loss": 0.5,
+      "step": 845
+    },
+    {
+      "epoch": 3.6623376623376624,
+      "grad_norm": 3.2039878368377686,
+      "learning_rate": 1.9058967212675804e-06,
+      "loss": 0.4473,
+      "step": 846
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 13.117010116577148,
+      "learning_rate": 1.8582507951800277e-06,
+      "loss": 0.582,
+      "step": 847
+    },
+    {
+      "epoch": 3.670995670995671,
+      "grad_norm": 22.922861099243164,
+      "learning_rate": 1.811196710228008e-06,
+      "loss": 0.4824,
+      "step": 848
+    },
+    {
+      "epoch": 3.675324675324675,
+      "grad_norm": 2.8894295692443848,
+      "learning_rate": 1.7647350448812106e-06,
+      "loss": 0.4883,
+      "step": 849
+    },
+    {
+      "epoch": 3.6796536796536796,
+      "grad_norm": 3.592529058456421,
+      "learning_rate": 1.718866370326283e-06,
+      "loss": 0.4883,
+      "step": 850
+    },
+    {
+      "epoch": 3.683982683982684,
+      "grad_norm": 4.659455299377441,
+      "learning_rate": 1.6735912504598383e-06,
+      "loss": 0.4883,
+      "step": 851
+    },
+    {
+      "epoch": 3.688311688311688,
+      "grad_norm": 4.902064323425293,
+      "learning_rate": 1.6289102418814705e-06,
+      "loss": 0.5117,
+      "step": 852
+    },
+    {
+      "epoch": 3.6926406926406927,
+      "grad_norm": 3.393808603286743,
+      "learning_rate": 1.584823893886933e-06,
+      "loss": 0.4668,
+      "step": 853
+    },
+    {
+      "epoch": 3.6969696969696972,
+      "grad_norm": 3.582292079925537,
+      "learning_rate": 1.5413327484613993e-06,
+      "loss": 0.4668,
+      "step": 854
+    },
+    {
+      "epoch": 3.7012987012987013,
+      "grad_norm": 3.2243218421936035,
+      "learning_rate": 1.4984373402728014e-06,
+      "loss": 0.4492,
+      "step": 855
+    },
+    {
+      "epoch": 3.7056277056277054,
+      "grad_norm": 3.1523277759552,
+      "learning_rate": 1.4561381966652243e-06,
+      "loss": 0.457,
+      "step": 856
+    },
+    {
+      "epoch": 3.70995670995671,
+      "grad_norm": 4.405923366546631,
+      "learning_rate": 1.4144358376524503e-06,
+      "loss": 0.498,
+      "step": 857
+    },
+    {
+      "epoch": 3.7142857142857144,
+      "grad_norm": 2.567735195159912,
+      "learning_rate": 1.3733307759115743e-06,
+      "loss": 0.4707,
+      "step": 858
+    },
+    {
+      "epoch": 3.7186147186147185,
+      "grad_norm": 3.0225648880004883,
+      "learning_rate": 1.3328235167766546e-06,
+      "loss": 0.4844,
+      "step": 859
+    },
+    {
+      "epoch": 3.722943722943723,
+      "grad_norm": 8.063156127929688,
+      "learning_rate": 1.2929145582325663e-06,
+      "loss": 0.4922,
+      "step": 860
+    },
+    {
+      "epoch": 3.7272727272727275,
+      "grad_norm": 2.602567195892334,
+      "learning_rate": 1.2536043909088191e-06,
+      "loss": 0.4922,
+      "step": 861
+    },
+    {
+      "epoch": 3.7316017316017316,
+      "grad_norm": 5.210265159606934,
+      "learning_rate": 1.214893498073577e-06,
+      "loss": 0.4434,
+      "step": 862
+    },
+    {
+      "epoch": 3.7359307359307357,
+      "grad_norm": 2.8473033905029297,
+      "learning_rate": 1.1767823556276647e-06,
+      "loss": 0.4512,
+      "step": 863
+    },
+    {
+      "epoch": 3.74025974025974,
+      "grad_norm": 7.142506122589111,
+      "learning_rate": 1.1392714320987596e-06,
+      "loss": 0.4727,
+      "step": 864
+    },
+    {
+      "epoch": 3.7445887445887447,
+      "grad_norm": 3.187635898590088,
+      "learning_rate": 1.10236118863562e-06,
+      "loss": 0.4355,
+      "step": 865
+    },
+    {
+      "epoch": 3.7489177489177488,
+      "grad_norm": 3.71706485748291,
+      "learning_rate": 1.0660520790023998e-06,
+      "loss": 0.5352,
+      "step": 866
+    },
+    {
+      "epoch": 3.7532467532467533,
+      "grad_norm": 4.7947564125061035,
+      "learning_rate": 1.0303445495730869e-06,
+      "loss": 0.4512,
+      "step": 867
+    },
+    {
+      "epoch": 3.757575757575758,
+      "grad_norm": 4.167108058929443,
+      "learning_rate": 9.952390393260235e-07,
+      "loss": 0.4609,
+      "step": 868
+    },
+    {
+      "epoch": 3.761904761904762,
+      "grad_norm": 3.4184553623199463,
+      "learning_rate": 9.607359798384785e-07,
+      "loss": 0.5195,
+      "step": 869
+    },
+    {
+      "epoch": 3.7662337662337664,
+      "grad_norm": 3.1368608474731445,
+      "learning_rate": 9.268357952813722e-07,
+      "loss": 0.4648,
+      "step": 870
+    },
+    {
+      "epoch": 3.7705627705627704,
+      "grad_norm": 2.8823840618133545,
+      "learning_rate": 8.935389024140483e-07,
+      "loss": 0.4648,
+      "step": 871
+    },
+    {
+      "epoch": 3.774891774891775,
+      "grad_norm": 4.08203125,
+      "learning_rate": 8.608457105791446e-07,
+      "loss": 0.4902,
+      "step": 872
+    },
+    {
+      "epoch": 3.779220779220779,
+      "grad_norm": 3.601361036300659,
+      "learning_rate": 8.287566216975795e-07,
+      "loss": 0.582,
+      "step": 873
+    },
+    {
+      "epoch": 3.7835497835497836,
+      "grad_norm": 9.086109161376953,
+      "learning_rate": 7.972720302635906e-07,
+      "loss": 0.4258,
+      "step": 874
+    },
+    {
+      "epoch": 3.787878787878788,
+      "grad_norm": 2.7145943641662598,
+      "learning_rate": 7.663923233398928e-07,
+      "loss": 0.3828,
+      "step": 875
+    },
+    {
+      "epoch": 3.792207792207792,
+      "grad_norm": 3.3459510803222656,
+      "learning_rate": 7.36117880552939e-07,
+      "loss": 0.4258,
+      "step": 876
+    },
+    {
+      "epoch": 3.7965367965367967,
+      "grad_norm": 5.249660015106201,
+      "learning_rate": 7.064490740882057e-07,
+      "loss": 0.5039,
+      "step": 877
+    },
+    {
+      "epoch": 3.8008658008658007,
+      "grad_norm": 3.8521053791046143,
+      "learning_rate": 6.773862686856702e-07,
+      "loss": 0.4102,
+      "step": 878
+    },
+    {
+      "epoch": 3.8051948051948052,
+      "grad_norm": 3.913069725036621,
+      "learning_rate": 6.489298216352912e-07,
+      "loss": 0.4375,
+      "step": 879
+    },
+    {
+      "epoch": 3.8095238095238093,
+      "grad_norm": 2.9759268760681152,
+      "learning_rate": 6.210800827726349e-07,
+      "loss": 0.4473,
+      "step": 880
+    },
+    {
+      "epoch": 3.813852813852814,
+      "grad_norm": 3.4686899185180664,
+      "learning_rate": 5.938373944745612e-07,
+      "loss": 0.4824,
+      "step": 881
+    },
+    {
+      "epoch": 3.8181818181818183,
+      "grad_norm": 2.9231326580047607,
+      "learning_rate": 5.672020916550224e-07,
+      "loss": 0.4863,
+      "step": 882
+    },
+    {
+      "epoch": 3.8225108225108224,
+      "grad_norm": 3.0240039825439453,
+      "learning_rate": 5.411745017609493e-07,
+      "loss": 0.498,
+      "step": 883
+    },
+    {
+      "epoch": 3.826839826839827,
+      "grad_norm": 2.8907124996185303,
+      "learning_rate": 5.157549447682208e-07,
+      "loss": 0.4199,
+      "step": 884
+    },
+    {
+      "epoch": 3.8311688311688314,
+      "grad_norm": 3.0946829319000244,
+      "learning_rate": 4.909437331777179e-07,
+      "loss": 0.3789,
+      "step": 885
+    },
+    {
+      "epoch": 3.8354978354978355,
+      "grad_norm": 5.231670379638672,
+      "learning_rate": 4.667411720115145e-07,
+      "loss": 0.5195,
+      "step": 886
+    },
+    {
+      "epoch": 3.8398268398268396,
+      "grad_norm": 4.364688873291016,
+      "learning_rate": 4.431475588090872e-07,
+      "loss": 0.5273,
+      "step": 887
+    },
+    {
+      "epoch": 3.844155844155844,
+      "grad_norm": 3.3589601516723633,
+      "learning_rate": 4.2016318362368966e-07,
+      "loss": 0.4395,
+      "step": 888
+    },
+    {
+      "epoch": 3.8484848484848486,
+      "grad_norm": 4.5765485763549805,
+      "learning_rate": 3.9778832901876675e-07,
+      "loss": 0.5234,
+      "step": 889
+    },
+    {
+      "epoch": 3.8528138528138527,
+      "grad_norm": 3.577456474304199,
+      "learning_rate": 3.7602327006450167e-07,
+      "loss": 0.5391,
+      "step": 890
+    },
+    {
+      "epoch": 3.857142857142857,
+      "grad_norm": 3.7017719745635986,
+      "learning_rate": 3.5486827433440785e-07,
+      "loss": 0.4375,
+      "step": 891
+    },
+    {
+      "epoch": 3.8614718614718617,
+      "grad_norm": 3.819176197052002,
+      "learning_rate": 3.343236019020535e-07,
+      "loss": 0.5195,
+      "step": 892
+    },
+    {
+      "epoch": 3.865800865800866,
+      "grad_norm": 2.649247169494629,
+      "learning_rate": 3.143895053378698e-07,
+      "loss": 0.4238,
+      "step": 893
+    },
+    {
+      "epoch": 3.87012987012987,
+      "grad_norm": 3.46985125541687,
+      "learning_rate": 2.95066229706048e-07,
+      "loss": 0.4316,
+      "step": 894
+    },
+    {
+      "epoch": 3.8744588744588744,
+      "grad_norm": 3.742621660232544,
+      "learning_rate": 2.7635401256149163e-07,
+      "loss": 0.4688,
+      "step": 895
+    },
+    {
+      "epoch": 3.878787878787879,
+      "grad_norm": 3.2430999279022217,
+      "learning_rate": 2.582530839469521e-07,
+      "loss": 0.3789,
+      "step": 896
+    },
+    {
+      "epoch": 3.883116883116883,
+      "grad_norm": 3.4116854667663574,
+      "learning_rate": 2.407636663901591e-07,
+      "loss": 0.4453,
+      "step": 897
+    },
+    {
+      "epoch": 3.8874458874458875,
+      "grad_norm": 3.6003456115722656,
+      "learning_rate": 2.238859749010891e-07,
+      "loss": 0.4668,
+      "step": 898
+    },
+    {
+      "epoch": 3.891774891774892,
+      "grad_norm": 2.811554193496704,
+      "learning_rate": 2.0762021696933976e-07,
+      "loss": 0.4277,
+      "step": 899
+    },
+    {
+      "epoch": 3.896103896103896,
+      "grad_norm": 3.481433868408203,
+      "learning_rate": 1.91966592561571e-07,
+      "loss": 0.4824,
+      "step": 900
+    },
+    {
+      "epoch": 3.9004329004329006,
+      "grad_norm": 3.300210475921631,
+      "learning_rate": 1.7692529411904578e-07,
+      "loss": 0.4258,
+      "step": 901
+    },
+    {
+      "epoch": 3.9047619047619047,
+      "grad_norm": 3.654827833175659,
+      "learning_rate": 1.6249650655525417e-07,
+      "loss": 0.5859,
+      "step": 902
+    },
+    {
+      "epoch": 3.909090909090909,
+      "grad_norm": 4.8551788330078125,
+      "learning_rate": 1.4868040725365407e-07,
+      "loss": 0.4395,
+      "step": 903
+    },
+    {
+      "epoch": 3.9134199134199132,
+      "grad_norm": 3.247840642929077,
+      "learning_rate": 1.3547716606548966e-07,
+      "loss": 0.4258,
+      "step": 904
+    },
+    {
+      "epoch": 3.9177489177489178,
+      "grad_norm": 5.86729621887207,
+      "learning_rate": 1.228869453076986e-07,
+      "loss": 0.5508,
+      "step": 905
+    },
+    {
+      "epoch": 3.9220779220779223,
+      "grad_norm": 3.8882832527160645,
+      "learning_rate": 1.1090989976091926e-07,
+      "loss": 0.5039,
+      "step": 906
+    },
+    {
+      "epoch": 3.9264069264069263,
+      "grad_norm": 3.6708812713623047,
+      "learning_rate": 9.954617666758648e-08,
+      "loss": 0.4082,
+      "step": 907
+    },
+    {
+      "epoch": 3.930735930735931,
+      "grad_norm": 44.70121765136719,
+      "learning_rate": 8.879591573010548e-08,
+      "loss": 0.4902,
+      "step": 908
+    },
+    {
+      "epoch": 3.935064935064935,
+      "grad_norm": 3.3498098850250244,
+      "learning_rate": 7.865924910916977e-08,
+      "loss": 0.5312,
+      "step": 909
+    },
+    {
+      "epoch": 3.9393939393939394,
+      "grad_norm": 3.8387317657470703,
+      "learning_rate": 6.913630142211802e-08,
+      "loss": 0.4824,
+      "step": 910
+    },
+    {
+      "epoch": 3.9437229437229435,
+      "grad_norm": 2.8696885108947754,
+      "learning_rate": 6.022718974137975e-08,
+      "loss": 0.459,
+      "step": 911
+    },
+    {
+      "epoch": 3.948051948051948,
+      "grad_norm": 3.543076753616333,
+      "learning_rate": 5.193202359308202e-08,
+      "loss": 0.5117,
+      "step": 912
+    },
+    {
+      "epoch": 3.9523809523809526,
+      "grad_norm": 8.554274559020996,
+      "learning_rate": 4.4250904955656095e-08,
+      "loss": 0.4863,
+      "step": 913
+    },
+    {
+      "epoch": 3.9567099567099566,
+      "grad_norm": 7.712956428527832,
+      "learning_rate": 3.718392825862171e-08,
+      "loss": 0.4219,
+      "step": 914
+    },
+    {
+      "epoch": 3.961038961038961,
+      "grad_norm": 3.2742745876312256,
+      "learning_rate": 3.073118038139921e-08,
+      "loss": 0.4551,
+      "step": 915
+    },
+    {
+      "epoch": 3.965367965367965,
+      "grad_norm": 3.382878303527832,
+      "learning_rate": 2.4892740652265877e-08,
+      "loss": 0.5352,
+      "step": 916
+    },
+    {
+      "epoch": 3.9696969696969697,
+      "grad_norm": 3.246838092803955,
+      "learning_rate": 1.9668680847356735e-08,
+      "loss": 0.457,
+      "step": 917
+    },
+    {
+      "epoch": 3.974025974025974,
+      "grad_norm": 3.6596264839172363,
+      "learning_rate": 1.5059065189787503e-08,
+      "loss": 0.459,
+      "step": 918
+    },
+    {
+      "epoch": 3.9783549783549783,
+      "grad_norm": 3.897545337677002,
+      "learning_rate": 1.1063950348888518e-08,
+      "loss": 0.5352,
+      "step": 919
+    },
+    {
+      "epoch": 3.982683982683983,
+      "grad_norm": 3.9618396759033203,
+      "learning_rate": 7.68338543946645e-09,
+      "loss": 0.4414,
+      "step": 920
+    },
+    {
+      "epoch": 3.987012987012987,
+      "grad_norm": 3.2661845684051514,
+      "learning_rate": 4.917412021249179e-09,
+      "loss": 0.4863,
+      "step": 921
+    },
+    {
+      "epoch": 3.9913419913419914,
+      "grad_norm": 3.596438407897949,
+      "learning_rate": 2.7660640983251383e-09,
+      "loss": 0.5,
+      "step": 922
+    },
+    {
+      "epoch": 3.995670995670996,
+      "grad_norm": 3.350722074508667,
+      "learning_rate": 1.22936811877139e-09,
+      "loss": 0.4805,
+      "step": 923
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 4.776801109313965,
+      "learning_rate": 3.073429742872502e-10,
+      "loss": 0.457,
+      "step": 924
+    },
+    {
+      "epoch": 4.0,
+      "step": 924,
+      "total_flos": 9.183655652856168e+18,
+      "train_loss": 0.7426292782738095,
+      "train_runtime": 23451.5766,
+      "train_samples_per_second": 20.16,
+      "train_steps_per_second": 0.039
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 924,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "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": 9.183655652856168e+18,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}