diff --git "a/checkpoint-802/trainer_state.json" "b/checkpoint-802/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-802/trainer_state.json"
@@ -0,0 +1,5635 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.12517558919931326,
+  "eval_steps": 500,
+  "global_step": 802,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9418012499809265,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.0495,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9342747330665588,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.1571,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9167969822883606,
+      "learning_rate": 6e-06,
+      "loss": 1.2975,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9712246656417847,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.7433,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1027685403823853,
+      "learning_rate": 1e-05,
+      "loss": 2.218,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.143164873123169,
+      "learning_rate": 1.2e-05,
+      "loss": 1.9224,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8237400650978088,
+      "learning_rate": 1.4e-05,
+      "loss": 1.767,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2908293008804321,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.6552,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9047379493713379,
+      "learning_rate": 1.8e-05,
+      "loss": 1.9778,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8817495107650757,
+      "learning_rate": 2e-05,
+      "loss": 1.9901,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2938435077667236,
+      "learning_rate": 1.9999998794084053e-05,
+      "loss": 1.7041,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2303466796875,
+      "learning_rate": 1.999999517633649e-05,
+      "loss": 1.6639,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.835405707359314,
+      "learning_rate": 1.999998914675819e-05,
+      "loss": 1.6789,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.340865135192871,
+      "learning_rate": 1.999998070535061e-05,
+      "loss": 1.7557,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8470378518104553,
+      "learning_rate": 1.9999969852115782e-05,
+      "loss": 1.592,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.643478274345398,
+      "learning_rate": 1.999995658705632e-05,
+      "loss": 2.1006,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2171984910964966,
+      "learning_rate": 1.9999940910175432e-05,
+      "loss": 1.9627,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8794788718223572,
+      "learning_rate": 1.9999922821476894e-05,
+      "loss": 2.3324,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7493646740913391,
+      "learning_rate": 1.9999902320965067e-05,
+      "loss": 1.6542,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8097922205924988,
+      "learning_rate": 1.99998794086449e-05,
+      "loss": 2.4314,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7926909327507019,
+      "learning_rate": 1.9999854084521913e-05,
+      "loss": 2.4093,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8623242378234863,
+      "learning_rate": 1.999982634860222e-05,
+      "loss": 2.2189,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8397309184074402,
+      "learning_rate": 1.999979620089251e-05,
+      "loss": 1.8798,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.703669548034668,
+      "learning_rate": 1.9999763641400046e-05,
+      "loss": 2.0517,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7364016771316528,
+      "learning_rate": 1.999972867013269e-05,
+      "loss": 2.0208,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7407152056694031,
+      "learning_rate": 1.9999691287098876e-05,
+      "loss": 1.0987,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8759409785270691,
+      "learning_rate": 1.999965149230761e-05,
+      "loss": 1.2877,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.769755482673645,
+      "learning_rate": 1.9999609285768505e-05,
+      "loss": 2.0839,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7315641641616821,
+      "learning_rate": 1.999956466749173e-05,
+      "loss": 1.8807,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.77801114320755,
+      "learning_rate": 1.9999517637488045e-05,
+      "loss": 2.0659,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8167088627815247,
+      "learning_rate": 1.9999468195768802e-05,
+      "loss": 1.5178,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7299829721450806,
+      "learning_rate": 1.9999416342345917e-05,
+      "loss": 2.1694,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.885439932346344,
+      "learning_rate": 1.9999362077231898e-05,
+      "loss": 1.6862,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8192880749702454,
+      "learning_rate": 1.999930540043984e-05,
+      "loss": 1.8014,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1161061525344849,
+      "learning_rate": 1.9999246311983403e-05,
+      "loss": 1.9917,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.687031090259552,
+      "learning_rate": 1.999918481187684e-05,
+      "loss": 1.7064,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9123612642288208,
+      "learning_rate": 1.999912090013499e-05,
+      "loss": 2.1708,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0114833116531372,
+      "learning_rate": 1.999905457677326e-05,
+      "loss": 1.8041,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9185965061187744,
+      "learning_rate": 1.999898584180765e-05,
+      "loss": 1.9668,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0017578601837158,
+      "learning_rate": 1.9998914695254737e-05,
+      "loss": 1.8251,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.718401312828064,
+      "learning_rate": 1.999884113713168e-05,
+      "loss": 1.9886,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7518029808998108,
+      "learning_rate": 1.999876516745622e-05,
+      "loss": 2.0368,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8515375852584839,
+      "learning_rate": 1.9998686786246682e-05,
+      "loss": 1.9362,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8468660712242126,
+      "learning_rate": 1.9998605993521965e-05,
+      "loss": 1.4887,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7643191814422607,
+      "learning_rate": 1.999852278930156e-05,
+      "loss": 1.7073,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1301417350769043,
+      "learning_rate": 1.999843717360553e-05,
+      "loss": 1.6808,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.728996217250824,
+      "learning_rate": 1.999834914645453e-05,
+      "loss": 1.8557,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8372696042060852,
+      "learning_rate": 1.9998258707869787e-05,
+      "loss": 1.8909,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.06674063205719,
+      "learning_rate": 1.9998165857873108e-05,
+      "loss": 1.7721,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.79304039478302,
+      "learning_rate": 1.99980705964869e-05,
+      "loss": 1.9567,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7896111607551575,
+      "learning_rate": 1.9997972923734123e-05,
+      "loss": 1.7498,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9860767126083374,
+      "learning_rate": 1.9997872839638347e-05,
+      "loss": 1.9903,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8512886166572571,
+      "learning_rate": 1.9997770344223705e-05,
+      "loss": 1.5083,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7645949721336365,
+      "learning_rate": 1.9997665437514914e-05,
+      "loss": 1.6338,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8389264941215515,
+      "learning_rate": 1.999755811953728e-05,
+      "loss": 1.6309,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9311525821685791,
+      "learning_rate": 1.9997448390316686e-05,
+      "loss": 1.4101,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6994078755378723,
+      "learning_rate": 1.9997336249879596e-05,
+      "loss": 1.4245,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7228380441665649,
+      "learning_rate": 1.999722169825306e-05,
+      "loss": 1.4485,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7117276787757874,
+      "learning_rate": 1.99971047354647e-05,
+      "loss": 1.3035,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6175885200500488,
+      "learning_rate": 1.999698536154273e-05,
+      "loss": 1.1573,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7536934018135071,
+      "learning_rate": 1.9996863576515937e-05,
+      "loss": 1.5875,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7840648889541626,
+      "learning_rate": 1.9996739380413696e-05,
+      "loss": 1.5524,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9118802547454834,
+      "learning_rate": 1.9996612773265963e-05,
+      "loss": 1.7814,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7639506459236145,
+      "learning_rate": 1.9996483755103266e-05,
+      "loss": 2.0718,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7495502233505249,
+      "learning_rate": 1.9996352325956732e-05,
+      "loss": 1.3045,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.83675217628479,
+      "learning_rate": 1.9996218485858054e-05,
+      "loss": 1.945,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9988659620285034,
+      "learning_rate": 1.999608223483951e-05,
+      "loss": 1.8062,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9340213537216187,
+      "learning_rate": 1.9995943572933963e-05,
+      "loss": 1.9097,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8286553621292114,
+      "learning_rate": 1.999580250017486e-05,
+      "loss": 1.7936,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8728215098381042,
+      "learning_rate": 1.999565901659622e-05,
+      "loss": 1.6033,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5670807361602783,
+      "learning_rate": 1.9995513122232656e-05,
+      "loss": 2.462,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.859460711479187,
+      "learning_rate": 1.999536481711935e-05,
+      "loss": 2.2517,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4409862756729126,
+      "learning_rate": 1.9995214101292065e-05,
+      "loss": 1.7278,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7657808661460876,
+      "learning_rate": 1.999506097478716e-05,
+      "loss": 1.7529,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7684109807014465,
+      "learning_rate": 1.999490543764157e-05,
+      "loss": 2.1585,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.765892744064331,
+      "learning_rate": 1.9994747489892793e-05,
+      "loss": 1.909,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8307526707649231,
+      "learning_rate": 1.9994587131578936e-05,
+      "loss": 1.5767,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.117607593536377,
+      "learning_rate": 1.999442436273867e-05,
+      "loss": 1.4128,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7918795347213745,
+      "learning_rate": 1.9994259183411257e-05,
+      "loss": 1.5989,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8969283699989319,
+      "learning_rate": 1.9994091593636527e-05,
+      "loss": 1.7017,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8903273940086365,
+      "learning_rate": 1.9993921593454904e-05,
+      "loss": 1.9551,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9065592885017395,
+      "learning_rate": 1.9993749182907392e-05,
+      "loss": 1.8349,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7999714016914368,
+      "learning_rate": 1.999357436203557e-05,
+      "loss": 1.39,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.914991557598114,
+      "learning_rate": 1.9993397130881602e-05,
+      "loss": 1.3705,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8639907836914062,
+      "learning_rate": 1.999321748948823e-05,
+      "loss": 1.694,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.80655437707901,
+      "learning_rate": 1.9993035437898794e-05,
+      "loss": 1.3385,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4766621589660645,
+      "learning_rate": 1.9992850976157187e-05,
+      "loss": 1.5354,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7555109262466431,
+      "learning_rate": 1.9992664104307904e-05,
+      "loss": 1.841,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1581100225448608,
+      "learning_rate": 1.9992474822396012e-05,
+      "loss": 1.8417,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9877253770828247,
+      "learning_rate": 1.999228313046717e-05,
+      "loss": 2.2302,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9449608325958252,
+      "learning_rate": 1.9992089028567604e-05,
+      "loss": 1.8298,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0140725374221802,
+      "learning_rate": 1.999189251674413e-05,
+      "loss": 1.705,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8601987957954407,
+      "learning_rate": 1.999169359504414e-05,
+      "loss": 2.1924,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1179931163787842,
+      "learning_rate": 1.999149226351562e-05,
+      "loss": 1.4521,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8562244772911072,
+      "learning_rate": 1.999128852220712e-05,
+      "loss": 1.8118,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.2716445922851562,
+      "learning_rate": 1.9991082371167784e-05,
+      "loss": 1.6766,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1563775539398193,
+      "learning_rate": 1.9990873810447326e-05,
+      "loss": 1.7789,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8975584506988525,
+      "learning_rate": 1.999066284009605e-05,
+      "loss": 1.935,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9221625924110413,
+      "learning_rate": 1.999044946016484e-05,
+      "loss": 1.5732,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8881098031997681,
+      "learning_rate": 1.9990233670705157e-05,
+      "loss": 2.4744,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6764086484909058,
+      "learning_rate": 1.9990015471769048e-05,
+      "loss": 0.7813,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7852362394332886,
+      "learning_rate": 1.998979486340914e-05,
+      "loss": 1.334,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0338146686553955,
+      "learning_rate": 1.9989571845678636e-05,
+      "loss": 1.6507,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.660638689994812,
+      "learning_rate": 1.998934641863133e-05,
+      "loss": 1.6307,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9326343536376953,
+      "learning_rate": 1.9989118582321586e-05,
+      "loss": 1.8679,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7414451241493225,
+      "learning_rate": 1.9988888336804354e-05,
+      "loss": 1.8633,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6003090143203735,
+      "learning_rate": 1.9988655682135173e-05,
+      "loss": 1.2029,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.214863657951355,
+      "learning_rate": 1.9988420618370143e-05,
+      "loss": 1.6799,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8017918467521667,
+      "learning_rate": 1.998818314556597e-05,
+      "loss": 1.7695,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7820467352867126,
+      "learning_rate": 1.9987943263779917e-05,
+      "loss": 2.1115,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.233542799949646,
+      "learning_rate": 1.998770097306985e-05,
+      "loss": 1.5579,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.075163722038269,
+      "learning_rate": 1.9987456273494197e-05,
+      "loss": 1.5154,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9855638742446899,
+      "learning_rate": 1.9987209165111982e-05,
+      "loss": 1.4482,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1536937952041626,
+      "learning_rate": 1.9986959647982798e-05,
+      "loss": 1.5155,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1218782663345337,
+      "learning_rate": 1.9986707722166825e-05,
+      "loss": 1.5752,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0412499904632568,
+      "learning_rate": 1.998645338772483e-05,
+      "loss": 1.4374,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0152816772460938,
+      "learning_rate": 1.9986196644718143e-05,
+      "loss": 1.9691,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.48252534866333,
+      "learning_rate": 1.9985937493208697e-05,
+      "loss": 1.9283,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9401952624320984,
+      "learning_rate": 1.998567593325899e-05,
+      "loss": 1.7933,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0944910049438477,
+      "learning_rate": 1.9985411964932104e-05,
+      "loss": 1.7656,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8697406053543091,
+      "learning_rate": 1.9985145588291706e-05,
+      "loss": 2.3612,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.024667739868164,
+      "learning_rate": 1.9984876803402043e-05,
+      "loss": 1.6229,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1907790899276733,
+      "learning_rate": 1.998460561032794e-05,
+      "loss": 1.8345,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9052940607070923,
+      "learning_rate": 1.99843320091348e-05,
+      "loss": 2.1532,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8829050660133362,
+      "learning_rate": 1.998405599988862e-05,
+      "loss": 1.0946,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8987332582473755,
+      "learning_rate": 1.9983777582655964e-05,
+      "loss": 1.7491,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8795452117919922,
+      "learning_rate": 1.998349675750398e-05,
+      "loss": 1.5047,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7775564789772034,
+      "learning_rate": 1.9983213524500396e-05,
+      "loss": 1.7593,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9722967147827148,
+      "learning_rate": 1.998292788371353e-05,
+      "loss": 1.4058,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8291569948196411,
+      "learning_rate": 1.998263983521227e-05,
+      "loss": 1.7716,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8410966992378235,
+      "learning_rate": 1.9982349379066086e-05,
+      "loss": 2.0283,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9493787884712219,
+      "learning_rate": 1.998205651534504e-05,
+      "loss": 1.8034,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8751788139343262,
+      "learning_rate": 1.9981761244119756e-05,
+      "loss": 1.3362,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8287643194198608,
+      "learning_rate": 1.9981463565461452e-05,
+      "loss": 1.8359,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9401163458824158,
+      "learning_rate": 1.9981163479441925e-05,
+      "loss": 2.0059,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7662250995635986,
+      "learning_rate": 1.998086098613355e-05,
+      "loss": 1.1899,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7782216668128967,
+      "learning_rate": 1.998055608560928e-05,
+      "loss": 1.6007,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9448866844177246,
+      "learning_rate": 1.9980248777942655e-05,
+      "loss": 2.0743,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.796358585357666,
+      "learning_rate": 1.9979939063207792e-05,
+      "loss": 2.2182,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7981236577033997,
+      "learning_rate": 1.9979626941479386e-05,
+      "loss": 2.0773,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8096224665641785,
+      "learning_rate": 1.997931241283272e-05,
+      "loss": 1.9287,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9485547542572021,
+      "learning_rate": 1.9978995477343655e-05,
+      "loss": 1.6551,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9155763387680054,
+      "learning_rate": 1.9978676135088623e-05,
+      "loss": 1.8185,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9059701561927795,
+      "learning_rate": 1.997835438614465e-05,
+      "loss": 1.8789,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8805462718009949,
+      "learning_rate": 1.997803023058933e-05,
+      "loss": 1.5399,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.797347903251648,
+      "learning_rate": 1.9977703668500853e-05,
+      "loss": 1.712,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9459641575813293,
+      "learning_rate": 1.9977374699957973e-05,
+      "loss": 1.5353,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4370256662368774,
+      "learning_rate": 1.9977043325040033e-05,
+      "loss": 1.4925,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3168740272521973,
+      "learning_rate": 1.997670954382696e-05,
+      "loss": 1.0978,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9876695275306702,
+      "learning_rate": 1.9976373356399247e-05,
+      "loss": 2.9874,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8283455967903137,
+      "learning_rate": 1.9976034762837987e-05,
+      "loss": 1.5847,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.004213571548462,
+      "learning_rate": 1.9975693763224834e-05,
+      "loss": 2.1154,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9788973331451416,
+      "learning_rate": 1.997535035764204e-05,
+      "loss": 1.7769,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8842994570732117,
+      "learning_rate": 1.9975004546172423e-05,
+      "loss": 1.4797,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9473729729652405,
+      "learning_rate": 1.9974656328899387e-05,
+      "loss": 1.0775,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.217759132385254,
+      "learning_rate": 1.9974305705906918e-05,
+      "loss": 1.665,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.258272886276245,
+      "learning_rate": 1.997395267727958e-05,
+      "loss": 1.3609,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9515790343284607,
+      "learning_rate": 1.9973597243102518e-05,
+      "loss": 1.9433,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8606133460998535,
+      "learning_rate": 1.9973239403461455e-05,
+      "loss": 1.4242,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6642533540725708,
+      "learning_rate": 1.9972879158442696e-05,
+      "loss": 1.8489,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7613729238510132,
+      "learning_rate": 1.997251650813313e-05,
+      "loss": 1.8288,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9006676077842712,
+      "learning_rate": 1.9972151452620215e-05,
+      "loss": 2.406,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5719473361968994,
+      "learning_rate": 1.9971783991992005e-05,
+      "loss": 1.2342,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9287020564079285,
+      "learning_rate": 1.9971414126337117e-05,
+      "loss": 2.0854,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0734272003173828,
+      "learning_rate": 1.9971041855744764e-05,
+      "loss": 1.89,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8500266671180725,
+      "learning_rate": 1.9970667180304727e-05,
+      "loss": 1.8576,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8321697115898132,
+      "learning_rate": 1.9970290100107372e-05,
+      "loss": 2.0847,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.739933967590332,
+      "learning_rate": 1.996991061524364e-05,
+      "loss": 1.9846,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9269589185714722,
+      "learning_rate": 1.9969528725805066e-05,
+      "loss": 1.2928,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9688999652862549,
+      "learning_rate": 1.996914443188375e-05,
+      "loss": 1.1562,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8473265171051025,
+      "learning_rate": 1.9968757733572372e-05,
+      "loss": 1.3455,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0271553993225098,
+      "learning_rate": 1.9968368630964206e-05,
+      "loss": 1.7439,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1127710342407227,
+      "learning_rate": 1.9967977124153095e-05,
+      "loss": 1.4497,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0394861698150635,
+      "learning_rate": 1.996758321323346e-05,
+      "loss": 1.8584,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2190815210342407,
+      "learning_rate": 1.9967186898300307e-05,
+      "loss": 1.9396,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9696542024612427,
+      "learning_rate": 1.9966788179449223e-05,
+      "loss": 2.2817,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0213602781295776,
+      "learning_rate": 1.996638705677637e-05,
+      "loss": 1.559,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.031296730041504,
+      "learning_rate": 1.9965983530378498e-05,
+      "loss": 1.6529,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9905886650085449,
+      "learning_rate": 1.996557760035292e-05,
+      "loss": 1.7384,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8480849266052246,
+      "learning_rate": 1.9965169266797545e-05,
+      "loss": 2.0596,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.219632387161255,
+      "learning_rate": 1.996475852981086e-05,
+      "loss": 1.2584,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9293888211250305,
+      "learning_rate": 1.996434538949192e-05,
+      "loss": 2.3078,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9126909971237183,
+      "learning_rate": 1.9963929845940376e-05,
+      "loss": 1.517,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9666692018508911,
+      "learning_rate": 1.996351189925644e-05,
+      "loss": 1.9152,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8901292085647583,
+      "learning_rate": 1.9963091549540926e-05,
+      "loss": 1.3486,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6530977487564087,
+      "learning_rate": 1.9962668796895207e-05,
+      "loss": 1.2948,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3919610977172852,
+      "learning_rate": 1.9962243641421243e-05,
+      "loss": 1.3113,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8002163767814636,
+      "learning_rate": 1.996181608322158e-05,
+      "loss": 1.4723,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7906051874160767,
+      "learning_rate": 1.9961386122399334e-05,
+      "loss": 1.1714,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6313331127166748,
+      "learning_rate": 1.9960953759058204e-05,
+      "loss": 1.0104,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.325727105140686,
+      "learning_rate": 1.9960518993302468e-05,
+      "loss": 1.7907,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8740698099136353,
+      "learning_rate": 1.996008182523699e-05,
+      "loss": 1.9081,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6601691246032715,
+      "learning_rate": 1.9959642254967203e-05,
+      "loss": 0.9907,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0757032632827759,
+      "learning_rate": 1.9959200282599123e-05,
+      "loss": 1.8676,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3735053539276123,
+      "learning_rate": 1.995875590823935e-05,
+      "loss": 1.5341,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8186277151107788,
+      "learning_rate": 1.9958309131995058e-05,
+      "loss": 1.4037,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7334154844284058,
+      "learning_rate": 1.9957859953974e-05,
+      "loss": 0.3896,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7954867482185364,
+      "learning_rate": 1.995740837428451e-05,
+      "loss": 2.3116,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6459976434707642,
+      "learning_rate": 1.9956954393035505e-05,
+      "loss": 1.2643,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8668067455291748,
+      "learning_rate": 1.9956498010336478e-05,
+      "loss": 1.106,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9993132948875427,
+      "learning_rate": 1.9956039226297496e-05,
+      "loss": 0.5023,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7291522026062012,
+      "learning_rate": 1.995557804102921e-05,
+      "loss": 0.7306,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0429105758666992,
+      "learning_rate": 1.9955114454642857e-05,
+      "loss": 1.491,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0801547765731812,
+      "learning_rate": 1.9954648467250243e-05,
+      "loss": 1.3588,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8252794742584229,
+      "learning_rate": 1.9954180078963753e-05,
+      "loss": 1.0519,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4998942613601685,
+      "learning_rate": 1.9953709289896355e-05,
+      "loss": 2.0603,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8373726606369019,
+      "learning_rate": 1.9953236100161598e-05,
+      "loss": 1.2329,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0468814373016357,
+      "learning_rate": 1.9952760509873607e-05,
+      "loss": 1.6357,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2933052778244019,
+      "learning_rate": 1.995228251914709e-05,
+      "loss": 0.9753,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8154112100601196,
+      "learning_rate": 1.995180212809732e-05,
+      "loss": 1.7323,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7924510836601257,
+      "learning_rate": 1.995131933684017e-05,
+      "loss": 1.3031,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.103473424911499,
+      "learning_rate": 1.9950834145492073e-05,
+      "loss": 1.7168,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7922141551971436,
+      "learning_rate": 1.9950346554170056e-05,
+      "loss": 1.3946,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7477668523788452,
+      "learning_rate": 1.994985656299171e-05,
+      "loss": 1.2364,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.860984206199646,
+      "learning_rate": 1.994936417207522e-05,
+      "loss": 1.6675,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8422707915306091,
+      "learning_rate": 1.9948869381539343e-05,
+      "loss": 1.7477,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.21062171459198,
+      "learning_rate": 1.9948372191503406e-05,
+      "loss": 1.0363,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8037005066871643,
+      "learning_rate": 1.994787260208733e-05,
+      "loss": 1.9047,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0549213886260986,
+      "learning_rate": 1.99473706134116e-05,
+      "loss": 1.8576,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9254148602485657,
+      "learning_rate": 1.99468662255973e-05,
+      "loss": 1.3331,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0943273305892944,
+      "learning_rate": 1.9946359438766066e-05,
+      "loss": 2.0064,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0273025035858154,
+      "learning_rate": 1.9945850253040134e-05,
+      "loss": 1.2028,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8689888715744019,
+      "learning_rate": 1.994533866854231e-05,
+      "loss": 1.7144,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8768132925033569,
+      "learning_rate": 1.994482468539598e-05,
+      "loss": 2.116,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.795387327671051,
+      "learning_rate": 1.9944308303725107e-05,
+      "loss": 1.3354,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7924376130104065,
+      "learning_rate": 1.9943789523654237e-05,
+      "loss": 1.027,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8999226689338684,
+      "learning_rate": 1.9943268345308483e-05,
+      "loss": 0.5476,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9620956778526306,
+      "learning_rate": 1.9942744768813553e-05,
+      "loss": 1.701,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7766193151473999,
+      "learning_rate": 1.994221879429572e-05,
+      "loss": 1.2552,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1024415493011475,
+      "learning_rate": 1.994169042188184e-05,
+      "loss": 1.8727,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9882749319076538,
+      "learning_rate": 1.994115965169935e-05,
+      "loss": 1.8009,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0686743259429932,
+      "learning_rate": 1.9940626483876263e-05,
+      "loss": 1.7133,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9760270118713379,
+      "learning_rate": 1.9940090918541168e-05,
+      "loss": 1.7522,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9487199187278748,
+      "learning_rate": 1.9939552955823237e-05,
+      "loss": 1.5481,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8242840766906738,
+      "learning_rate": 1.993901259585221e-05,
+      "loss": 1.4483,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.89918327331543,
+      "learning_rate": 1.993846983875843e-05,
+      "loss": 1.7293,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1786655187606812,
+      "learning_rate": 1.993792468467278e-05,
+      "loss": 1.8322,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9560681581497192,
+      "learning_rate": 1.9937377133726753e-05,
+      "loss": 1.6228,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0292149782180786,
+      "learning_rate": 1.993682718605241e-05,
+      "loss": 1.418,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1595954895019531,
+      "learning_rate": 1.9936274841782386e-05,
+      "loss": 1.768,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9066284894943237,
+      "learning_rate": 1.9935720101049897e-05,
+      "loss": 1.5641,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1178666353225708,
+      "learning_rate": 1.993516296398874e-05,
+      "loss": 1.8692,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1353600025177002,
+      "learning_rate": 1.9934603430733285e-05,
+      "loss": 2.6621,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8215340375900269,
+      "learning_rate": 1.9934041501418478e-05,
+      "loss": 1.6577,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9386087656021118,
+      "learning_rate": 1.9933477176179856e-05,
+      "loss": 1.6772,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2389590740203857,
+      "learning_rate": 1.9932910455153518e-05,
+      "loss": 1.9047,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8284682035446167,
+      "learning_rate": 1.993234133847615e-05,
+      "loss": 2.6177,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.399548053741455,
+      "learning_rate": 1.993176982628501e-05,
+      "loss": 1.9387,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.845256507396698,
+      "learning_rate": 1.993119591871794e-05,
+      "loss": 1.8301,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8806995153427124,
+      "learning_rate": 1.9930619615913364e-05,
+      "loss": 2.2612,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7267456650733948,
+      "learning_rate": 1.9930040918010263e-05,
+      "loss": 1.38,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8338924050331116,
+      "learning_rate": 1.9929459825148215e-05,
+      "loss": 2.1012,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8365648984909058,
+      "learning_rate": 1.9928876337467375e-05,
+      "loss": 1.541,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.187801718711853,
+      "learning_rate": 1.9928290455108464e-05,
+      "loss": 1.541,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.6386361122131348,
+      "learning_rate": 1.9927702178212786e-05,
+      "loss": 1.1252,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8757442831993103,
+      "learning_rate": 1.9927111506922227e-05,
+      "loss": 1.4014,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0362242460250854,
+      "learning_rate": 1.992651844137925e-05,
+      "loss": 2.171,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9764441847801208,
+      "learning_rate": 1.9925922981726885e-05,
+      "loss": 1.3608,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1725298166275024,
+      "learning_rate": 1.992532512810875e-05,
+      "loss": 1.8949,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0441815853118896,
+      "learning_rate": 1.992472488066904e-05,
+      "loss": 1.5673,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9130929112434387,
+      "learning_rate": 1.992412223955252e-05,
+      "loss": 1.5355,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9515172243118286,
+      "learning_rate": 1.9923517204904544e-05,
+      "loss": 1.855,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.7631518840789795,
+      "learning_rate": 1.9922909776871026e-05,
+      "loss": 1.6034,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1321780681610107,
+      "learning_rate": 1.9922299955598476e-05,
+      "loss": 1.4475,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9716091752052307,
+      "learning_rate": 1.992168774123397e-05,
+      "loss": 1.0144,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9520627856254578,
+      "learning_rate": 1.9921073133925163e-05,
+      "loss": 1.3794,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8643708229064941,
+      "learning_rate": 1.992045613382029e-05,
+      "loss": 1.7873,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8701338768005371,
+      "learning_rate": 1.9919836741068157e-05,
+      "loss": 1.4195,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.938417911529541,
+      "learning_rate": 1.9919214955818153e-05,
+      "loss": 1.917,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.946891188621521,
+      "learning_rate": 1.9918590778220243e-05,
+      "loss": 1.6889,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9969150424003601,
+      "learning_rate": 1.991796420842497e-05,
+      "loss": 1.6985,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7932005524635315,
+      "learning_rate": 1.9917335246583443e-05,
+      "loss": 1.8445,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6052041053771973,
+      "learning_rate": 1.991670389284737e-05,
+      "loss": 1.6405,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1532299518585205,
+      "learning_rate": 1.9916070147369017e-05,
+      "loss": 1.5678,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9578217267990112,
+      "learning_rate": 1.991543401030123e-05,
+      "loss": 0.9819,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9796022176742554,
+      "learning_rate": 1.991479548179744e-05,
+      "loss": 2.0374,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.816317617893219,
+      "learning_rate": 1.9914154562011643e-05,
+      "loss": 1.2706,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1207562685012817,
+      "learning_rate": 1.9913511251098426e-05,
+      "loss": 1.4879,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.951187789440155,
+      "learning_rate": 1.9912865549212935e-05,
+      "loss": 1.1507,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9572225213050842,
+      "learning_rate": 1.9912217456510913e-05,
+      "loss": 1.5829,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8443856239318848,
+      "learning_rate": 1.9911566973148663e-05,
+      "loss": 0.9778,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.02583909034729,
+      "learning_rate": 1.991091409928307e-05,
+      "loss": 2.2217,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6288923621177673,
+      "learning_rate": 1.9910258835071595e-05,
+      "loss": 1.0959,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.690157949924469,
+      "learning_rate": 1.9909601180672283e-05,
+      "loss": 0.5142,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7905344367027283,
+      "learning_rate": 1.9908941136243746e-05,
+      "loss": 0.5352,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8578112125396729,
+      "learning_rate": 1.9908278701945175e-05,
+      "loss": 1.7706,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6474728584289551,
+      "learning_rate": 1.990761387793634e-05,
+      "loss": 0.6515,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6532492637634277,
+      "learning_rate": 1.9906946664377577e-05,
+      "loss": 0.6143,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.868615448474884,
+      "learning_rate": 1.9906277061429818e-05,
+      "loss": 1.8124,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1038082838058472,
+      "learning_rate": 1.9905605069254554e-05,
+      "loss": 1.9446,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9355915188789368,
+      "learning_rate": 1.990493068801386e-05,
+      "loss": 1.9609,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1322965621948242,
+      "learning_rate": 1.9904253917870387e-05,
+      "loss": 1.4975,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7613868713378906,
+      "learning_rate": 1.9903574758987355e-05,
+      "loss": 1.5561,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9120030403137207,
+      "learning_rate": 1.9902893211528572e-05,
+      "loss": 1.3147,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.416876196861267,
+      "learning_rate": 1.990220927565841e-05,
+      "loss": 0.9732,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3078633546829224,
+      "learning_rate": 1.9901522951541827e-05,
+      "loss": 1.7904,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0230833292007446,
+      "learning_rate": 1.9900834239344352e-05,
+      "loss": 1.7119,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.742758572101593,
+      "learning_rate": 1.9900143139232095e-05,
+      "loss": 1.7543,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9155807495117188,
+      "learning_rate": 1.9899449651371727e-05,
+      "loss": 1.5954,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.844660222530365,
+      "learning_rate": 1.9898753775930513e-05,
+      "loss": 1.6544,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9266340136528015,
+      "learning_rate": 1.9898055513076288e-05,
+      "loss": 1.7524,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1642438173294067,
+      "learning_rate": 1.9897354862977453e-05,
+      "loss": 1.5596,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.894527018070221,
+      "learning_rate": 1.9896651825803003e-05,
+      "loss": 1.9041,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.207465648651123,
+      "learning_rate": 1.9895946401722495e-05,
+      "loss": 1.7203,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8541679978370667,
+      "learning_rate": 1.989523859090606e-05,
+      "loss": 1.3416,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9433109164237976,
+      "learning_rate": 1.989452839352442e-05,
+      "loss": 1.7141,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0357227325439453,
+      "learning_rate": 1.9893815809748854e-05,
+      "loss": 1.2034,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1258755922317505,
+      "learning_rate": 1.9893100839751233e-05,
+      "loss": 1.5666,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9345949292182922,
+      "learning_rate": 1.989238348370399e-05,
+      "loss": 1.4935,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.737742006778717,
+      "learning_rate": 1.9891663741780135e-05,
+      "loss": 1.4562,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9488717317581177,
+      "learning_rate": 1.9890941614153268e-05,
+      "loss": 1.8453,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.7368555068969727,
+      "learning_rate": 1.989021710099755e-05,
+      "loss": 1.4831,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8286874294281006,
+      "learning_rate": 1.988949020248772e-05,
+      "loss": 2.2786,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.999751627445221,
+      "learning_rate": 1.9888760918799093e-05,
+      "loss": 1.7505,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8931187391281128,
+      "learning_rate": 1.9888029250107565e-05,
+      "loss": 1.6505,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9812218546867371,
+      "learning_rate": 1.98872951965896e-05,
+      "loss": 1.5357,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9732285737991333,
+      "learning_rate": 1.988655875842223e-05,
+      "loss": 1.5624,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3475276231765747,
+      "learning_rate": 1.9885819935783085e-05,
+      "loss": 1.6918,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.884621262550354,
+      "learning_rate": 1.988507872885035e-05,
+      "loss": 1.6485,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8911191821098328,
+      "learning_rate": 1.9884335137802792e-05,
+      "loss": 1.1337,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1611531972885132,
+      "learning_rate": 1.9883589162819757e-05,
+      "loss": 1.7675,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5208086967468262,
+      "learning_rate": 1.9882840804081155e-05,
+      "loss": 1.7403,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.881212592124939,
+      "learning_rate": 1.9882090061767482e-05,
+      "loss": 1.2506,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6470460891723633,
+      "learning_rate": 1.9881336936059797e-05,
+      "loss": 1.0243,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.797415554523468,
+      "learning_rate": 1.9880581427139754e-05,
+      "loss": 0.6804,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0682158470153809,
+      "learning_rate": 1.987982353518956e-05,
+      "loss": 1.6586,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7121508121490479,
+      "learning_rate": 1.9879063260392012e-05,
+      "loss": 1.4205,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7593161463737488,
+      "learning_rate": 1.987830060293047e-05,
+      "loss": 0.566,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9202454686164856,
+      "learning_rate": 1.987753556298887e-05,
+      "loss": 1.4541,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8837355971336365,
+      "learning_rate": 1.9876768140751738e-05,
+      "loss": 1.3312,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2164474725723267,
+      "learning_rate": 1.9875998336404155e-05,
+      "loss": 1.8487,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8529666066169739,
+      "learning_rate": 1.987522615013179e-05,
+      "loss": 1.4369,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0183709859848022,
+      "learning_rate": 1.9874451582120883e-05,
+      "loss": 1.9902,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8488042950630188,
+      "learning_rate": 1.9873674632558237e-05,
+      "loss": 1.5772,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9346114993095398,
+      "learning_rate": 1.9872895301631245e-05,
+      "loss": 1.9671,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9772021174430847,
+      "learning_rate": 1.9872113589527872e-05,
+      "loss": 1.6645,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6140549182891846,
+      "learning_rate": 1.9871329496436646e-05,
+      "loss": 0.716,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8879808187484741,
+      "learning_rate": 1.9870543022546685e-05,
+      "loss": 1.7084,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8681232929229736,
+      "learning_rate": 1.986975416804767e-05,
+      "loss": 1.2311,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8040364980697632,
+      "learning_rate": 1.9868962933129857e-05,
+      "loss": 1.1052,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8864116668701172,
+      "learning_rate": 1.9868169317984082e-05,
+      "loss": 1.3987,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2038077116012573,
+      "learning_rate": 1.986737332280175e-05,
+      "loss": 1.9375,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7685527801513672,
+      "learning_rate": 1.9866574947774844e-05,
+      "loss": 1.8077,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9364449977874756,
+      "learning_rate": 1.9865774193095913e-05,
+      "loss": 1.7186,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1445951461791992,
+      "learning_rate": 1.9864971058958092e-05,
+      "loss": 1.788,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9366347193717957,
+      "learning_rate": 1.986416554555508e-05,
+      "loss": 1.984,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.359132170677185,
+      "learning_rate": 1.986335765308116e-05,
+      "loss": 0.2784,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.961164116859436,
+      "learning_rate": 1.9862547381731166e-05,
+      "loss": 2.5258,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5328058004379272,
+      "learning_rate": 1.9861734731700534e-05,
+      "loss": 1.5605,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8162689208984375,
+      "learning_rate": 1.9860919703185265e-05,
+      "loss": 1.4508,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7568703293800354,
+      "learning_rate": 1.986010229638192e-05,
+      "loss": 1.5941,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7346351146697998,
+      "learning_rate": 1.985928251148765e-05,
+      "loss": 1.6749,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1934709548950195,
+      "learning_rate": 1.985846034870017e-05,
+      "loss": 2.0963,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9638925194740295,
+      "learning_rate": 1.9857635808217778e-05,
+      "loss": 1.9696,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0379003286361694,
+      "learning_rate": 1.9856808890239333e-05,
+      "loss": 0.8772,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0562266111373901,
+      "learning_rate": 1.985597959496427e-05,
+      "loss": 1.7018,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0546568632125854,
+      "learning_rate": 1.9855147922592614e-05,
+      "loss": 0.9769,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7254536151885986,
+      "learning_rate": 1.985431387332494e-05,
+      "loss": 0.9393,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8547775149345398,
+      "learning_rate": 1.9853477447362412e-05,
+      "loss": 1.3414,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.802185595035553,
+      "learning_rate": 1.985263864490676e-05,
+      "loss": 0.9645,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8310496807098389,
+      "learning_rate": 1.9851797466160283e-05,
+      "loss": 1.3528,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5428261756896973,
+      "learning_rate": 1.985095391132587e-05,
+      "loss": 1.7086,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8697307109832764,
+      "learning_rate": 1.9850107980606964e-05,
+      "loss": 2.0801,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0970669984817505,
+      "learning_rate": 1.98492596742076e-05,
+      "loss": 1.7845,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1335629224777222,
+      "learning_rate": 1.984840899233236e-05,
+      "loss": 0.4997,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9874480366706848,
+      "learning_rate": 1.9847555935186424e-05,
+      "loss": 1.9856,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8648535013198853,
+      "learning_rate": 1.9846700502975527e-05,
+      "loss": 1.7317,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2264970541000366,
+      "learning_rate": 1.9845842695905997e-05,
+      "loss": 1.6865,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.876065194606781,
+      "learning_rate": 1.9844982514184718e-05,
+      "loss": 1.5677,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7052832841873169,
+      "learning_rate": 1.9844119958019146e-05,
+      "loss": 1.6304,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8926292657852173,
+      "learning_rate": 1.984325502761732e-05,
+      "loss": 1.4055,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9539690613746643,
+      "learning_rate": 1.9842387723187846e-05,
+      "loss": 1.4473,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9869019985198975,
+      "learning_rate": 1.98415180449399e-05,
+      "loss": 1.0927,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.443432331085205,
+      "learning_rate": 1.984064599308324e-05,
+      "loss": 1.1167,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8694020509719849,
+      "learning_rate": 1.9839771567828184e-05,
+      "loss": 1.1221,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9755352139472961,
+      "learning_rate": 1.9838894769385635e-05,
+      "loss": 0.8227,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9441053867340088,
+      "learning_rate": 1.9838015597967056e-05,
+      "loss": 0.8816,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7311797738075256,
+      "learning_rate": 1.983713405378449e-05,
+      "loss": 0.3256,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8313472867012024,
+      "learning_rate": 1.9836250137050553e-05,
+      "loss": 1.8942,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5323596000671387,
+      "learning_rate": 1.9835363847978428e-05,
+      "loss": 1.5672,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0517991781234741,
+      "learning_rate": 1.9834475186781877e-05,
+      "loss": 0.76,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.4064319133758545,
+      "learning_rate": 1.983358415367522e-05,
+      "loss": 1.5922,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9361559748649597,
+      "learning_rate": 1.9832690748873375e-05,
+      "loss": 1.2397,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6758370399475098,
+      "learning_rate": 1.9831794972591805e-05,
+      "loss": 0.5728,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9486380219459534,
+      "learning_rate": 1.9830896825046563e-05,
+      "loss": 1.5274,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9687122702598572,
+      "learning_rate": 1.9829996306454258e-05,
+      "loss": 0.9347,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1283320188522339,
+      "learning_rate": 1.9829093417032085e-05,
+      "loss": 1.8164,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7893568277359009,
+      "learning_rate": 1.982818815699781e-05,
+      "loss": 1.4499,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8257303237915039,
+      "learning_rate": 1.9827280526569756e-05,
+      "loss": 1.4763,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2423681020736694,
+      "learning_rate": 1.982637052596684e-05,
+      "loss": 1.757,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0532588958740234,
+      "learning_rate": 1.9825458155408533e-05,
+      "loss": 1.7136,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0051872730255127,
+      "learning_rate": 1.982454341511488e-05,
+      "loss": 1.6012,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8963203430175781,
+      "learning_rate": 1.9823626305306506e-05,
+      "loss": 1.072,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3821759223937988,
+      "learning_rate": 1.9822706826204606e-05,
+      "loss": 1.7187,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0313401222229004,
+      "learning_rate": 1.9821784978030933e-05,
+      "loss": 1.6027,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9623168706893921,
+      "learning_rate": 1.982086076100783e-05,
+      "loss": 1.4132,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9375492334365845,
+      "learning_rate": 1.9819934175358195e-05,
+      "loss": 1.7144,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7186566591262817,
+      "learning_rate": 1.9819005221305513e-05,
+      "loss": 1.3332,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7701579332351685,
+      "learning_rate": 1.9818073899073827e-05,
+      "loss": 1.1296,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8675704598426819,
+      "learning_rate": 1.9817140208887754e-05,
+      "loss": 1.7195,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9611474871635437,
+      "learning_rate": 1.9816204150972492e-05,
+      "loss": 1.4536,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.056624412536621,
+      "learning_rate": 1.9815265725553797e-05,
+      "loss": 1.9253,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.585964560508728,
+      "learning_rate": 1.9814324932858005e-05,
+      "loss": 2.0826,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9207966327667236,
+      "learning_rate": 1.9813381773112016e-05,
+      "loss": 1.4885,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9804804921150208,
+      "learning_rate": 1.9812436246543304e-05,
+      "loss": 1.3071,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7058753371238708,
+      "learning_rate": 1.981148835337991e-05,
+      "loss": 0.8504,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.990956723690033,
+      "learning_rate": 1.9810538093850465e-05,
+      "loss": 2.1998,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9452444911003113,
+      "learning_rate": 1.980958546818414e-05,
+      "loss": 1.7364,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8717514276504517,
+      "learning_rate": 1.98086304766107e-05,
+      "loss": 0.8079,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1225824356079102,
+      "learning_rate": 1.980767311936047e-05,
+      "loss": 1.7555,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0479341745376587,
+      "learning_rate": 1.980671339666435e-05,
+      "loss": 1.6204,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9834145307540894,
+      "learning_rate": 1.9805751308753808e-05,
+      "loss": 1.5251,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.074107050895691,
+      "learning_rate": 1.9804786855860886e-05,
+      "loss": 1.8535,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1673707962036133,
+      "learning_rate": 1.980382003821819e-05,
+      "loss": 1.8278,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8728948831558228,
+      "learning_rate": 1.9802850856058903e-05,
+      "loss": 0.8509,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0313911437988281,
+      "learning_rate": 1.9801879309616777e-05,
+      "loss": 1.9013,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0420602560043335,
+      "learning_rate": 1.9800905399126125e-05,
+      "loss": 1.4502,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2369115352630615,
+      "learning_rate": 1.9799929124821844e-05,
+      "loss": 1.689,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1010773181915283,
+      "learning_rate": 1.97989504869394e-05,
+      "loss": 0.6274,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3504084348678589,
+      "learning_rate": 1.979796948571481e-05,
+      "loss": 1.2692,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8189222812652588,
+      "learning_rate": 1.9796986121384685e-05,
+      "loss": 1.5439,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9344761967658997,
+      "learning_rate": 1.9796000394186198e-05,
+      "loss": 1.5129,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0255612134933472,
+      "learning_rate": 1.979501230435708e-05,
+      "loss": 1.9559,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.449123501777649,
+      "learning_rate": 1.979402185213565e-05,
+      "loss": 1.7581,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0824624300003052,
+      "learning_rate": 1.9793029037760784e-05,
+      "loss": 2.1333,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0862467288970947,
+      "learning_rate": 1.9792033861471935e-05,
+      "loss": 1.6442,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3037303686141968,
+      "learning_rate": 1.9791036323509126e-05,
+      "loss": 1.4648,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.037331461906433,
+      "learning_rate": 1.9790036424112938e-05,
+      "loss": 1.017,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8827648162841797,
+      "learning_rate": 1.9789034163524535e-05,
+      "loss": 1.659,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8996952176094055,
+      "learning_rate": 1.9788029541985643e-05,
+      "loss": 1.3892,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0246918201446533,
+      "learning_rate": 1.978702255973856e-05,
+      "loss": 1.51,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9744868278503418,
+      "learning_rate": 1.9786013217026157e-05,
+      "loss": 1.3445,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9991737604141235,
+      "learning_rate": 1.9785001514091865e-05,
+      "loss": 1.6653,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4088870286941528,
+      "learning_rate": 1.9783987451179696e-05,
+      "loss": 1.57,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7513281106948853,
+      "learning_rate": 1.9782971028534223e-05,
+      "loss": 0.7615,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.09906005859375,
+      "learning_rate": 1.9781952246400587e-05,
+      "loss": 0.9732,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4216969013214111,
+      "learning_rate": 1.97809311050245e-05,
+      "loss": 1.1183,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9488307237625122,
+      "learning_rate": 1.977990760465225e-05,
+      "loss": 1.002,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9960178136825562,
+      "learning_rate": 1.977888174553068e-05,
+      "loss": 1.7178,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0002471208572388,
+      "learning_rate": 1.9777853527907224e-05,
+      "loss": 1.7894,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.886707603931427,
+      "learning_rate": 1.9776822952029854e-05,
+      "loss": 1.039,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7411413192749023,
+      "learning_rate": 1.977579001814714e-05,
+      "loss": 1.5746,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0032079219818115,
+      "learning_rate": 1.9774754726508202e-05,
+      "loss": 1.3436,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0437644720077515,
+      "learning_rate": 1.9773717077362734e-05,
+      "loss": 1.4815,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0581908226013184,
+      "learning_rate": 1.9772677070961007e-05,
+      "loss": 1.8118,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8130147457122803,
+      "learning_rate": 1.9771634707553846e-05,
+      "loss": 1.1538,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.009604573249817,
+      "learning_rate": 1.9770589987392653e-05,
+      "loss": 1.6894,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0172890424728394,
+      "learning_rate": 1.9769542910729397e-05,
+      "loss": 1.6646,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9655178189277649,
+      "learning_rate": 1.9768493477816618e-05,
+      "loss": 1.5982,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8684929013252258,
+      "learning_rate": 1.976744168890742e-05,
+      "loss": 2.0278,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0428229570388794,
+      "learning_rate": 1.9766387544255473e-05,
+      "loss": 0.4468,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1995832920074463,
+      "learning_rate": 1.9765331044115024e-05,
+      "loss": 1.671,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.851990282535553,
+      "learning_rate": 1.9764272188740885e-05,
+      "loss": 0.9449,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2639641761779785,
+      "learning_rate": 1.9763210978388426e-05,
+      "loss": 0.5276,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1194183826446533,
+      "learning_rate": 1.9762147413313596e-05,
+      "loss": 1.7057,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.371021747589111,
+      "learning_rate": 1.9761081493772914e-05,
+      "loss": 0.6986,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8994089365005493,
+      "learning_rate": 1.976001322002346e-05,
+      "loss": 2.5059,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8265968561172485,
+      "learning_rate": 1.975894259232288e-05,
+      "loss": 1.5496,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0377991199493408,
+      "learning_rate": 1.9757869610929396e-05,
+      "loss": 1.8287,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8921816945075989,
+      "learning_rate": 1.9756794276101784e-05,
+      "loss": 1.664,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.402585506439209,
+      "learning_rate": 1.975571658809941e-05,
+      "loss": 0.5236,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9020755290985107,
+      "learning_rate": 1.9754636547182183e-05,
+      "loss": 1.7119,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.136975884437561,
+      "learning_rate": 1.9753554153610598e-05,
+      "loss": 1.4358,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8028199672698975,
+      "learning_rate": 1.975246940764571e-05,
+      "loss": 1.1221,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9304431676864624,
+      "learning_rate": 1.9751382309549135e-05,
+      "loss": 1.1922,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1193586587905884,
+      "learning_rate": 1.9750292859583066e-05,
+      "loss": 0.5726,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8460230827331543,
+      "learning_rate": 1.974920105801026e-05,
+      "loss": 2.4396,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8110451698303223,
+      "learning_rate": 1.9748106905094045e-05,
+      "loss": 1.3223,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7961413860321045,
+      "learning_rate": 1.9747010401098307e-05,
+      "loss": 1.2879,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1107492446899414,
+      "learning_rate": 1.974591154628751e-05,
+      "loss": 1.6185,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.14578115940094,
+      "learning_rate": 1.9744810340926674e-05,
+      "loss": 1.4298,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3011332750320435,
+      "learning_rate": 1.974370678528139e-05,
+      "loss": 0.9556,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8401873111724854,
+      "learning_rate": 1.9742600879617823e-05,
+      "loss": 2.3746,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8258370757102966,
+      "learning_rate": 1.9741492624202698e-05,
+      "loss": 1.4742,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7155411243438721,
+      "learning_rate": 1.97403820193033e-05,
+      "loss": 0.9388,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8539145588874817,
+      "learning_rate": 1.9739269065187497e-05,
+      "loss": 1.4629,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8882036805152893,
+      "learning_rate": 1.9738153762123713e-05,
+      "loss": 1.889,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.810894250869751,
+      "learning_rate": 1.9737036110380935e-05,
+      "loss": 0.5582,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8205024600028992,
+      "learning_rate": 1.973591611022873e-05,
+      "loss": 1.3398,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0441452264785767,
+      "learning_rate": 1.9734793761937216e-05,
+      "loss": 1.4455,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1005403995513916,
+      "learning_rate": 1.973366906577709e-05,
+      "loss": 1.5484,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6423813700675964,
+      "learning_rate": 1.9732542022019603e-05,
+      "loss": 0.5292,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9499244689941406,
+      "learning_rate": 1.973141263093659e-05,
+      "loss": 1.3558,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9298450350761414,
+      "learning_rate": 1.973028089280043e-05,
+      "loss": 1.3631,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7698273062705994,
+      "learning_rate": 1.9729146807884084e-05,
+      "loss": 0.5124,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8734893798828125,
+      "learning_rate": 1.972801037646107e-05,
+      "loss": 1.637,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1014962196350098,
+      "learning_rate": 1.972687159880549e-05,
+      "loss": 1.3022,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2191619873046875,
+      "learning_rate": 1.972573047519198e-05,
+      "loss": 0.7652,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4204211235046387,
+      "learning_rate": 1.972458700589577e-05,
+      "loss": 1.8729,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8936397433280945,
+      "learning_rate": 1.9723441191192647e-05,
+      "loss": 2.155,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.3093836307525635,
+      "learning_rate": 1.9722293031358955e-05,
+      "loss": 1.8055,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.918250560760498,
+      "learning_rate": 1.9721142526671614e-05,
+      "loss": 1.1799,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.806936502456665,
+      "learning_rate": 1.971998967740811e-05,
+      "loss": 2.2243,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.198886752128601,
+      "learning_rate": 1.9718834483846485e-05,
+      "loss": 1.5366,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6140164136886597,
+      "learning_rate": 1.9717676946265358e-05,
+      "loss": 1.5059,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.061775803565979,
+      "learning_rate": 1.97165170649439e-05,
+      "loss": 1.5537,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.120010495185852,
+      "learning_rate": 1.9715354840161862e-05,
+      "loss": 1.5032,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4784014225006104,
+      "learning_rate": 1.9714190272199555e-05,
+      "loss": 2.0133,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0621649026870728,
+      "learning_rate": 1.971302336133784e-05,
+      "loss": 1.5967,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0540523529052734,
+      "learning_rate": 1.971185410785817e-05,
+      "loss": 1.447,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8714383840560913,
+      "learning_rate": 1.9710682512042548e-05,
+      "loss": 1.6458,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8371648192405701,
+      "learning_rate": 1.9709508574173535e-05,
+      "loss": 1.6041,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.397619605064392,
+      "learning_rate": 1.9708332294534274e-05,
+      "loss": 0.787,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9853465557098389,
+      "learning_rate": 1.9707153673408454e-05,
+      "loss": 1.5562,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.108499526977539,
+      "learning_rate": 1.970597271108035e-05,
+      "loss": 0.9748,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7067657709121704,
+      "learning_rate": 1.9704789407834778e-05,
+      "loss": 1.4679,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9193729162216187,
+      "learning_rate": 1.9703603763957142e-05,
+      "loss": 1.8292,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0290855169296265,
+      "learning_rate": 1.9702415779733392e-05,
+      "loss": 1.623,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1568219661712646,
+      "learning_rate": 1.970122545545006e-05,
+      "loss": 1.6309,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8143751621246338,
+      "learning_rate": 1.9700032791394215e-05,
+      "loss": 1.5616,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.996700644493103,
+      "learning_rate": 1.9698837787853523e-05,
+      "loss": 1.941,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7473616003990173,
+      "learning_rate": 1.9697640445116192e-05,
+      "loss": 1.6,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0816364288330078,
+      "learning_rate": 1.9696440763471e-05,
+      "loss": 1.764,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9372792840003967,
+      "learning_rate": 1.9695238743207292e-05,
+      "loss": 1.4362,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5626298189163208,
+      "learning_rate": 1.969403438461498e-05,
+      "loss": 0.6701,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8877453804016113,
+      "learning_rate": 1.9692827687984523e-05,
+      "loss": 1.6476,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8707518577575684,
+      "learning_rate": 1.969161865360697e-05,
+      "loss": 1.0084,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0139648914337158,
+      "learning_rate": 1.9690407281773905e-05,
+      "loss": 1.4887,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9187926054000854,
+      "learning_rate": 1.9689193572777505e-05,
+      "loss": 1.5581,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8604023456573486,
+      "learning_rate": 1.9687977526910488e-05,
+      "loss": 1.6908,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2782931327819824,
+      "learning_rate": 1.9686759144466142e-05,
+      "loss": 1.1137,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9516409635543823,
+      "learning_rate": 1.9685538425738326e-05,
+      "loss": 1.2642,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1259483098983765,
+      "learning_rate": 1.968431537102146e-05,
+      "loss": 1.555,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.298754334449768,
+      "learning_rate": 1.9683089980610512e-05,
+      "loss": 1.5994,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3935412168502808,
+      "learning_rate": 1.9681862254801034e-05,
+      "loss": 1.0852,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8852770924568176,
+      "learning_rate": 1.968063219388913e-05,
+      "loss": 1.9436,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.731374204158783,
+      "learning_rate": 1.9679399798171476e-05,
+      "loss": 0.7434,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.073830008506775,
+      "learning_rate": 1.9678165067945295e-05,
+      "loss": 2.2306,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2147220373153687,
+      "learning_rate": 1.967692800350839e-05,
+      "loss": 2.0442,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8961025476455688,
+      "learning_rate": 1.967568860515912e-05,
+      "loss": 1.6517,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1098088026046753,
+      "learning_rate": 1.9674446873196408e-05,
+      "loss": 1.5784,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8114632964134216,
+      "learning_rate": 1.9673202807919733e-05,
+      "loss": 0.5461,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8463197946548462,
+      "learning_rate": 1.967195640962915e-05,
+      "loss": 1.2396,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5498254299163818,
+      "learning_rate": 1.9670707678625264e-05,
+      "loss": 1.8367,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2922487258911133,
+      "learning_rate": 1.9669456615209247e-05,
+      "loss": 2.251,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8260951042175293,
+      "learning_rate": 1.966820321968284e-05,
+      "loss": 2.8105,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9870577454566956,
+      "learning_rate": 1.9666947492348334e-05,
+      "loss": 1.6524,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9321213960647583,
+      "learning_rate": 1.9665689433508596e-05,
+      "loss": 0.9929,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7950844764709473,
+      "learning_rate": 1.9664429043467044e-05,
+      "loss": 2.1832,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9330363273620605,
+      "learning_rate": 1.9663166322527668e-05,
+      "loss": 1.1176,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0382758378982544,
+      "learning_rate": 1.9661901270995013e-05,
+      "loss": 1.5673,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7401167154312134,
+      "learning_rate": 1.9660633889174185e-05,
+      "loss": 1.3154,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.299643874168396,
+      "learning_rate": 1.9659364177370856e-05,
+      "loss": 1.7258,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8082520961761475,
+      "learning_rate": 1.9658092135891256e-05,
+      "loss": 1.5656,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.076417088508606,
+      "learning_rate": 1.9656817765042192e-05,
+      "loss": 1.6585,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0347617864608765,
+      "learning_rate": 1.9655541065131013e-05,
+      "loss": 1.608,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.070332646369934,
+      "learning_rate": 1.965426203646563e-05,
+      "loss": 1.9664,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.838746190071106,
+      "learning_rate": 1.965298067935454e-05,
+      "loss": 2.1611,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9825697541236877,
+      "learning_rate": 1.965169699410677e-05,
+      "loss": 1.7168,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5898070335388184,
+      "learning_rate": 1.965041098103193e-05,
+      "loss": 1.3112,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9250902533531189,
+      "learning_rate": 1.9649122640440187e-05,
+      "loss": 0.9961,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.733625054359436,
+      "learning_rate": 1.964783197264226e-05,
+      "loss": 1.7441,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.052191972732544,
+      "learning_rate": 1.9646538977949444e-05,
+      "loss": 1.3966,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9679017663002014,
+      "learning_rate": 1.9645243656673582e-05,
+      "loss": 0.7808,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7678504586219788,
+      "learning_rate": 1.9643946009127088e-05,
+      "loss": 1.7577,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4840080738067627,
+      "learning_rate": 1.9642646035622926e-05,
+      "loss": 1.5794,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8202274441719055,
+      "learning_rate": 1.9641343736474635e-05,
+      "loss": 1.9119,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4255080223083496,
+      "learning_rate": 1.96400391119963e-05,
+      "loss": 1.1435,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1355290412902832,
+      "learning_rate": 1.9638732162502586e-05,
+      "loss": 1.6109,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0749764442443848,
+      "learning_rate": 1.9637422888308696e-05,
+      "loss": 1.1917,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9144006967544556,
+      "learning_rate": 1.963611128973041e-05,
+      "loss": 1.9197,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7404701113700867,
+      "learning_rate": 1.9634797367084065e-05,
+      "loss": 1.1256,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8343794345855713,
+      "learning_rate": 1.9633481120686554e-05,
+      "loss": 1.8625,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9903147220611572,
+      "learning_rate": 1.9632162550855335e-05,
+      "loss": 1.4753,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9988419413566589,
+      "learning_rate": 1.9630841657908424e-05,
+      "loss": 1.4715,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7479739189147949,
+      "learning_rate": 1.9629518442164397e-05,
+      "loss": 1.6179,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0459107160568237,
+      "learning_rate": 1.9628192903942395e-05,
+      "loss": 1.6056,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6827442646026611,
+      "learning_rate": 1.9626865043562113e-05,
+      "loss": 1.2805,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9262802004814148,
+      "learning_rate": 1.9625534861343806e-05,
+      "loss": 1.9221,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0962365865707397,
+      "learning_rate": 1.96242023576083e-05,
+      "loss": 1.1623,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8223599195480347,
+      "learning_rate": 1.9622867532676965e-05,
+      "loss": 1.4158,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8202844262123108,
+      "learning_rate": 1.962153038687174e-05,
+      "loss": 2.3395,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.769325315952301,
+      "learning_rate": 1.9620190920515124e-05,
+      "loss": 1.2261,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9759867787361145,
+      "learning_rate": 1.961884913393017e-05,
+      "loss": 1.4156,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8986252546310425,
+      "learning_rate": 1.96175050274405e-05,
+      "loss": 1.2528,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8722736239433289,
+      "learning_rate": 1.9616158601370284e-05,
+      "loss": 1.2981,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0854696035385132,
+      "learning_rate": 1.961480985604426e-05,
+      "loss": 1.6531,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9684545993804932,
+      "learning_rate": 1.9613458791787723e-05,
+      "loss": 1.149,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1259515285491943,
+      "learning_rate": 1.961210540892653e-05,
+      "loss": 1.6374,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6296025514602661,
+      "learning_rate": 1.9610749707787086e-05,
+      "loss": 0.1722,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9159917235374451,
+      "learning_rate": 1.960939168869637e-05,
+      "loss": 2.4642,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7028860449790955,
+      "learning_rate": 1.9608031351981913e-05,
+      "loss": 0.8297,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1244877576828003,
+      "learning_rate": 1.9606668697971804e-05,
+      "loss": 1.4262,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1643065214157104,
+      "learning_rate": 1.9605303726994692e-05,
+      "loss": 1.8621,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2138675451278687,
+      "learning_rate": 1.960393643937978e-05,
+      "loss": 0.7296,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5474352836608887,
+      "learning_rate": 1.9602566835456847e-05,
+      "loss": 1.2063,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1823064088821411,
+      "learning_rate": 1.9601194915556207e-05,
+      "loss": 0.1894,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6637358069419861,
+      "learning_rate": 1.959982068000875e-05,
+      "loss": 1.4456,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6221618056297302,
+      "learning_rate": 1.959844412914592e-05,
+      "loss": 0.5781,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9478740096092224,
+      "learning_rate": 1.959706526329971e-05,
+      "loss": 1.2688,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0652697086334229,
+      "learning_rate": 1.9595684082802687e-05,
+      "loss": 1.7915,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9758893847465515,
+      "learning_rate": 1.9594300587987964e-05,
+      "loss": 1.6521,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3758968114852905,
+      "learning_rate": 1.9592914779189218e-05,
+      "loss": 1.7467,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0427806377410889,
+      "learning_rate": 1.9591526656740685e-05,
+      "loss": 1.3578,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4715042114257812,
+      "learning_rate": 1.9590136220977156e-05,
+      "loss": 1.6052,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9868489503860474,
+      "learning_rate": 1.958874347223398e-05,
+      "loss": 1.7042,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8361290097236633,
+      "learning_rate": 1.958734841084706e-05,
+      "loss": 1.1737,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1467775106430054,
+      "learning_rate": 1.958595103715287e-05,
+      "loss": 1.135,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7978134751319885,
+      "learning_rate": 1.9584551351488425e-05,
+      "loss": 1.305,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9538130760192871,
+      "learning_rate": 1.9583149354191312e-05,
+      "loss": 1.8212,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.121382236480713,
+      "learning_rate": 1.9581745045599665e-05,
+      "loss": 0.3183,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7262074947357178,
+      "learning_rate": 1.9580338426052182e-05,
+      "loss": 2.0245,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7842314839363098,
+      "learning_rate": 1.9578929495888116e-05,
+      "loss": 0.9158,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7360401153564453,
+      "learning_rate": 1.9577518255447274e-05,
+      "loss": 0.8348,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8156128525733948,
+      "learning_rate": 1.9576104705070027e-05,
+      "loss": 1.3912,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7814883589744568,
+      "learning_rate": 1.9574688845097302e-05,
+      "loss": 1.0388,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.038300633430481,
+      "learning_rate": 1.9573270675870573e-05,
+      "loss": 0.751,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9030879735946655,
+      "learning_rate": 1.9571850197731884e-05,
+      "loss": 1.3984,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8930461406707764,
+      "learning_rate": 1.9570427411023827e-05,
+      "loss": 1.8717,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5750823020935059,
+      "learning_rate": 1.956900231608956e-05,
+      "loss": 2.1555,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8204957246780396,
+      "learning_rate": 1.9567574913272788e-05,
+      "loss": 1.3371,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9168577790260315,
+      "learning_rate": 1.9566145202917773e-05,
+      "loss": 1.132,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9453327059745789,
+      "learning_rate": 1.9564713185369344e-05,
+      "loss": 0.852,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.316981554031372,
+      "learning_rate": 1.9563278860972876e-05,
+      "loss": 1.8398,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0372123718261719,
+      "learning_rate": 1.9561842230074302e-05,
+      "loss": 1.653,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1009215116500854,
+      "learning_rate": 1.956040329302012e-05,
+      "loss": 1.6638,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9586663246154785,
+      "learning_rate": 1.955896205015737e-05,
+      "loss": 1.0886,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3998528718948364,
+      "learning_rate": 1.9557518501833658e-05,
+      "loss": 1.9539,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9007513523101807,
+      "learning_rate": 1.955607264839715e-05,
+      "loss": 0.4139,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7559785842895508,
+      "learning_rate": 1.955462449019655e-05,
+      "loss": 1.8494,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0630534887313843,
+      "learning_rate": 1.9553174027581138e-05,
+      "loss": 1.6309,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9597139358520508,
+      "learning_rate": 1.955172126090074e-05,
+      "loss": 0.854,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.003679871559143,
+      "learning_rate": 1.9550266190505735e-05,
+      "loss": 1.1375,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2234400510787964,
+      "learning_rate": 1.954880881674706e-05,
+      "loss": 1.0621,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8920372128486633,
+      "learning_rate": 1.954734913997622e-05,
+      "loss": 0.8149,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0686150789260864,
+      "learning_rate": 1.9545887160545252e-05,
+      "loss": 1.0463,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2131983041763306,
+      "learning_rate": 1.9544422878806768e-05,
+      "loss": 1.4061,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1307955980300903,
+      "learning_rate": 1.9542956295113925e-05,
+      "loss": 1.6915,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9545600414276123,
+      "learning_rate": 1.9541487409820442e-05,
+      "loss": 1.9788,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9386622905731201,
+      "learning_rate": 1.954001622328059e-05,
+      "loss": 2.3962,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0973879098892212,
+      "learning_rate": 1.9538542735849188e-05,
+      "loss": 1.9024,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0093085765838623,
+      "learning_rate": 1.953706694788162e-05,
+      "loss": 1.6269,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.729065477848053,
+      "learning_rate": 1.953558885973382e-05,
+      "loss": 0.7673,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9341952204704285,
+      "learning_rate": 1.953410847176228e-05,
+      "loss": 1.3542,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6524180769920349,
+      "learning_rate": 1.9532625784324047e-05,
+      "loss": 1.4862,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0662424564361572,
+      "learning_rate": 1.9531140797776714e-05,
+      "loss": 1.8857,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7807791233062744,
+      "learning_rate": 1.952965351247844e-05,
+      "loss": 1.6716,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0078587532043457,
+      "learning_rate": 1.952816392878793e-05,
+      "loss": 1.9718,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2724988460540771,
+      "learning_rate": 1.9526672047064448e-05,
+      "loss": 1.348,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6617332696914673,
+      "learning_rate": 1.952517786766781e-05,
+      "loss": 1.6268,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9120773077011108,
+      "learning_rate": 1.9523681390958393e-05,
+      "loss": 1.9604,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9997332096099854,
+      "learning_rate": 1.952218261729711e-05,
+      "loss": 1.6832,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9092682003974915,
+      "learning_rate": 1.952068154704545e-05,
+      "loss": 1.9769,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9539963006973267,
+      "learning_rate": 1.951917818056544e-05,
+      "loss": 0.9615,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8873115181922913,
+      "learning_rate": 1.951767251821967e-05,
+      "loss": 1.4593,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8866798281669617,
+      "learning_rate": 1.951616456037128e-05,
+      "loss": 1.491,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.098756194114685,
+      "learning_rate": 1.9514654307383966e-05,
+      "loss": 0.8209,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7631105184555054,
+      "learning_rate": 1.9513141759621973e-05,
+      "loss": 1.7128,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 12.292449951171875,
+      "learning_rate": 1.95116269174501e-05,
+      "loss": 3.2222,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9233523607254028,
+      "learning_rate": 1.9510109781233706e-05,
+      "loss": 1.2803,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8416237235069275,
+      "learning_rate": 1.9508590351338693e-05,
+      "loss": 1.1823,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.337841033935547,
+      "learning_rate": 1.950706862813153e-05,
+      "loss": 1.2577,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9062979221343994,
+      "learning_rate": 1.9505544611979226e-05,
+      "loss": 2.1593,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9518592953681946,
+      "learning_rate": 1.950401830324935e-05,
+      "loss": 1.1195,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.020942211151123,
+      "learning_rate": 1.9502489702310018e-05,
+      "loss": 1.5837,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.879868745803833,
+      "learning_rate": 1.9500958809529903e-05,
+      "loss": 2.0357,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6751201152801514,
+      "learning_rate": 1.949942562527824e-05,
+      "loss": 1.4446,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0675795078277588,
+      "learning_rate": 1.94978901499248e-05,
+      "loss": 1.4454,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8025611639022827,
+      "learning_rate": 1.949635238383991e-05,
+      "loss": 1.212,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0021713972091675,
+      "learning_rate": 1.9494812327394458e-05,
+      "loss": 1.3265,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0278031826019287,
+      "learning_rate": 1.9493269980959885e-05,
+      "loss": 1.0983,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.649834394454956,
+      "learning_rate": 1.9491725344908167e-05,
+      "loss": 1.5833,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8162665963172913,
+      "learning_rate": 1.9490178419611853e-05,
+      "loss": 1.6149,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1506603956222534,
+      "learning_rate": 1.9488629205444034e-05,
+      "loss": 1.0756,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4848748445510864,
+      "learning_rate": 1.9487077702778353e-05,
+      "loss": 1.7649,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9423287510871887,
+      "learning_rate": 1.948552391198901e-05,
+      "loss": 1.5612,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.312781572341919,
+      "learning_rate": 1.9483967833450745e-05,
+      "loss": 1.8895,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3580660820007324,
+      "learning_rate": 1.9482409467538863e-05,
+      "loss": 1.7204,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9630597233772278,
+      "learning_rate": 1.9480848814629217e-05,
+      "loss": 1.9857,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7201079726219177,
+      "learning_rate": 1.947928587509821e-05,
+      "loss": 1.1478,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9898256063461304,
+      "learning_rate": 1.94777206493228e-05,
+      "loss": 1.5353,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0257985591888428,
+      "learning_rate": 1.9476153137680482e-05,
+      "loss": 1.4052,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8972013592720032,
+      "learning_rate": 1.9474583340549324e-05,
+      "loss": 1.8627,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9348580837249756,
+      "learning_rate": 1.947301125830793e-05,
+      "loss": 1.8107,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8059454560279846,
+      "learning_rate": 1.947143689133546e-05,
+      "loss": 1.7113,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1312888860702515,
+      "learning_rate": 1.9469860240011626e-05,
+      "loss": 1.8456,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2592461109161377,
+      "learning_rate": 1.946828130471669e-05,
+      "loss": 1.7084,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8490139842033386,
+      "learning_rate": 1.9466700085831466e-05,
+      "loss": 1.7365,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3371502161026,
+      "learning_rate": 1.9465116583737314e-05,
+      "loss": 1.7185,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1870206594467163,
+      "learning_rate": 1.946353079881615e-05,
+      "loss": 1.5285,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8405884504318237,
+      "learning_rate": 1.9461942731450435e-05,
+      "loss": 1.6512,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3664089441299438,
+      "learning_rate": 1.9460352382023192e-05,
+      "loss": 1.649,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7940143346786499,
+      "learning_rate": 1.945875975091798e-05,
+      "loss": 1.7897,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7846907377243042,
+      "learning_rate": 1.9457164838518916e-05,
+      "loss": 1.5796,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7551419734954834,
+      "learning_rate": 1.9455567645210668e-05,
+      "loss": 1.4008,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0351200103759766,
+      "learning_rate": 1.945396817137845e-05,
+      "loss": 1.5813,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0557612180709839,
+      "learning_rate": 1.9452366417408032e-05,
+      "loss": 1.3311,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1308883428573608,
+      "learning_rate": 1.9450762383685725e-05,
+      "loss": 0.6483,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9998852610588074,
+      "learning_rate": 1.9449156070598397e-05,
+      "loss": 1.7178,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8182191848754883,
+      "learning_rate": 1.9447547478533467e-05,
+      "loss": 1.9668,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0509693622589111,
+      "learning_rate": 1.9445936607878898e-05,
+      "loss": 0.9926,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0919716358184814,
+      "learning_rate": 1.94443234590232e-05,
+      "loss": 1.2947,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7361246943473816,
+      "learning_rate": 1.9442708032355444e-05,
+      "loss": 1.4167,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.714868426322937,
+      "learning_rate": 1.9441090328265242e-05,
+      "loss": 1.5424,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8188365697860718,
+      "learning_rate": 1.9439470347142756e-05,
+      "loss": 1.7068,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9500496983528137,
+      "learning_rate": 1.94378480893787e-05,
+      "loss": 1.1715,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.612969994544983,
+      "learning_rate": 1.9436223555364333e-05,
+      "loss": 1.8178,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6630369424819946,
+      "learning_rate": 1.943459674549147e-05,
+      "loss": 1.5356,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4056222438812256,
+      "learning_rate": 1.9432967660152456e-05,
+      "loss": 1.0915,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9931894540786743,
+      "learning_rate": 1.943133629974022e-05,
+      "loss": 1.3489,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.721828281879425,
+      "learning_rate": 1.9429702664648204e-05,
+      "loss": 0.9029,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1730399131774902,
+      "learning_rate": 1.942806675527042e-05,
+      "loss": 1.7768,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9179857969284058,
+      "learning_rate": 1.9426428572001418e-05,
+      "loss": 2.3638,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0844107866287231,
+      "learning_rate": 1.9424788115236303e-05,
+      "loss": 1.5415,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0590695142745972,
+      "learning_rate": 1.942314538537072e-05,
+      "loss": 1.6702,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7648866772651672,
+      "learning_rate": 1.9421500382800877e-05,
+      "loss": 1.8638,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0730507373809814,
+      "learning_rate": 1.9419853107923515e-05,
+      "loss": 1.5271,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8001253604888916,
+      "learning_rate": 1.9418203561135927e-05,
+      "loss": 2.9047,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8725544214248657,
+      "learning_rate": 1.9416551742835964e-05,
+      "loss": 1.4623,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1127315759658813,
+      "learning_rate": 1.941489765342201e-05,
+      "loss": 1.7306,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1793948411941528,
+      "learning_rate": 1.9413241293293006e-05,
+      "loss": 0.3196,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8964352011680603,
+      "learning_rate": 1.9411582662848435e-05,
+      "loss": 1.053,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3153489828109741,
+      "learning_rate": 1.9409921762488335e-05,
+      "loss": 1.3615,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9132351279258728,
+      "learning_rate": 1.9408258592613286e-05,
+      "loss": 1.5991,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7754179239273071,
+      "learning_rate": 1.9406593153624416e-05,
+      "loss": 0.8552,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7985204458236694,
+      "learning_rate": 1.94049254459234e-05,
+      "loss": 0.6634,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9285178780555725,
+      "learning_rate": 1.9403255469912467e-05,
+      "loss": 0.345,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.291887640953064,
+      "learning_rate": 1.9401583225994378e-05,
+      "loss": 2.0273,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5887540578842163,
+      "learning_rate": 1.9399908714572457e-05,
+      "loss": 1.6046,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.165919542312622,
+      "learning_rate": 1.9398231936050562e-05,
+      "loss": 1.8375,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0102756023406982,
+      "learning_rate": 1.939655289083311e-05,
+      "loss": 1.3056,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0535539388656616,
+      "learning_rate": 1.9394871579325057e-05,
+      "loss": 0.9489,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4022819995880127,
+      "learning_rate": 1.9393188001931906e-05,
+      "loss": 0.8971,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8239139318466187,
+      "learning_rate": 1.9391502159059704e-05,
+      "loss": 1.7716,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7645108103752136,
+      "learning_rate": 1.9389814051115054e-05,
+      "loss": 1.6232,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.692162036895752,
+      "learning_rate": 1.9388123678505095e-05,
+      "loss": 1.0834,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2660837173461914,
+      "learning_rate": 1.938643104163752e-05,
+      "loss": 1.2877,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6457018256187439,
+      "learning_rate": 1.9384736140920563e-05,
+      "loss": 0.912,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6962850093841553,
+      "learning_rate": 1.9383038976763004e-05,
+      "loss": 1.4546,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.3156652450561523,
+      "learning_rate": 1.938133954957417e-05,
+      "loss": 0.3989,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8944481611251831,
+      "learning_rate": 1.937963785976394e-05,
+      "loss": 1.1755,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6729207038879395,
+      "learning_rate": 1.9377933907742722e-05,
+      "loss": 1.5848,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8458144068717957,
+      "learning_rate": 1.9376227693921493e-05,
+      "loss": 0.6623,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8411895632743835,
+      "learning_rate": 1.9374519218711758e-05,
+      "loss": 1.0984,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1506187915802002,
+      "learning_rate": 1.937280848252557e-05,
+      "loss": 1.6137,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9453433156013489,
+      "learning_rate": 1.9371095485775534e-05,
+      "loss": 0.9001,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8604451417922974,
+      "learning_rate": 1.9369380228874794e-05,
+      "loss": 1.6193,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1416794061660767,
+      "learning_rate": 1.9367662712237043e-05,
+      "loss": 1.8017,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0399991273880005,
+      "learning_rate": 1.9365942936276513e-05,
+      "loss": 1.7701,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7908676862716675,
+      "learning_rate": 1.9364220901407985e-05,
+      "loss": 1.7858,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8512731790542603,
+      "learning_rate": 1.9362496608046788e-05,
+      "loss": 1.0902,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.198036789894104,
+      "learning_rate": 1.936077005660879e-05,
+      "loss": 1.3758,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9642990231513977,
+      "learning_rate": 1.9359041247510414e-05,
+      "loss": 1.1954,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0655434131622314,
+      "learning_rate": 1.935731018116861e-05,
+      "loss": 1.5582,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4649953842163086,
+      "learning_rate": 1.9355576858000885e-05,
+      "loss": 1.4024,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3037927150726318,
+      "learning_rate": 1.9353841278425293e-05,
+      "loss": 1.5681,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0754226446151733,
+      "learning_rate": 1.9352103442860415e-05,
+      "loss": 1.432,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2181533575057983,
+      "learning_rate": 1.93503633517254e-05,
+      "loss": 1.4216,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7769268751144409,
+      "learning_rate": 1.9348621005439917e-05,
+      "loss": 2.0223,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0469359159469604,
+      "learning_rate": 1.9346876404424202e-05,
+      "loss": 2.1334,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.877581834793091,
+      "learning_rate": 1.9345129549099016e-05,
+      "loss": 1.2296,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8810202479362488,
+      "learning_rate": 1.9343380439885674e-05,
+      "loss": 1.4057,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2003124952316284,
+      "learning_rate": 1.934162907720603e-05,
+      "loss": 0.7902,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8182734251022339,
+      "learning_rate": 1.933987546148249e-05,
+      "loss": 0.7547,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8032336235046387,
+      "learning_rate": 1.9338119593137985e-05,
+      "loss": 1.6296,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7411457300186157,
+      "learning_rate": 1.933636147259601e-05,
+      "loss": 1.8137,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.225765347480774,
+      "learning_rate": 1.933460110028059e-05,
+      "loss": 1.3601,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3987914323806763,
+      "learning_rate": 1.9332838476616298e-05,
+      "loss": 1.4657,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8765820264816284,
+      "learning_rate": 1.933107360202825e-05,
+      "loss": 1.5071,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7324680685997009,
+      "learning_rate": 1.9329306476942102e-05,
+      "loss": 0.7867,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.091023325920105,
+      "learning_rate": 1.932753710178406e-05,
+      "loss": 1.6422,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.734923243522644,
+      "learning_rate": 1.932576547698086e-05,
+      "loss": 0.3846,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9723742008209229,
+      "learning_rate": 1.9323991602959796e-05,
+      "loss": 0.6102,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8249735236167908,
+      "learning_rate": 1.932221548014869e-05,
+      "loss": 1.4364,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7934770584106445,
+      "learning_rate": 1.9320437108975915e-05,
+      "loss": 1.431,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0751607418060303,
+      "learning_rate": 1.9318656489870387e-05,
+      "loss": 1.5596,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2927584648132324,
+      "learning_rate": 1.9316873623261555e-05,
+      "loss": 2.008,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2063711881637573,
+      "learning_rate": 1.9315088509579427e-05,
+      "loss": 1.9501,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7476193308830261,
+      "learning_rate": 1.9313301149254533e-05,
+      "loss": 1.7395,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9957605004310608,
+      "learning_rate": 1.9311511542717956e-05,
+      "loss": 1.3785,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9378927946090698,
+      "learning_rate": 1.9309719690401323e-05,
+      "loss": 1.189,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8630698919296265,
+      "learning_rate": 1.9307925592736797e-05,
+      "loss": 1.6241,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8105760216712952,
+      "learning_rate": 1.930612925015708e-05,
+      "loss": 1.3733,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.809337854385376,
+      "learning_rate": 1.9304330663095426e-05,
+      "loss": 2.8394,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.237175464630127,
+      "learning_rate": 1.930252983198562e-05,
+      "loss": 1.5124,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9589737057685852,
+      "learning_rate": 1.930072675726199e-05,
+      "loss": 1.9335,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1641108989715576,
+      "learning_rate": 1.929892143935942e-05,
+      "loss": 1.5772,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.968217670917511,
+      "learning_rate": 1.9297113878713302e-05,
+      "loss": 1.6706,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.723006546497345,
+      "learning_rate": 1.9295304075759605e-05,
+      "loss": 1.3032,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7680689692497253,
+      "learning_rate": 1.9293492030934814e-05,
+      "loss": 1.2522,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8543024063110352,
+      "learning_rate": 1.9291677744675976e-05,
+      "loss": 1.2821,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7917808890342712,
+      "learning_rate": 1.9289861217420652e-05,
+      "loss": 1.6358,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8850444555282593,
+      "learning_rate": 1.9288042449606963e-05,
+      "loss": 1.5017,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8327540755271912,
+      "learning_rate": 1.9286221441673566e-05,
+      "loss": 0.6834,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8329907655715942,
+      "learning_rate": 1.9284398194059662e-05,
+      "loss": 1.3125,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2092479467391968,
+      "learning_rate": 1.928257270720498e-05,
+      "loss": 0.3374,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6056079864501953,
+      "learning_rate": 1.9280744981549803e-05,
+      "loss": 1.4002,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9556896090507507,
+      "learning_rate": 1.927891501753494e-05,
+      "loss": 1.4812,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7827396392822266,
+      "learning_rate": 1.9277082815601756e-05,
+      "loss": 0.5404,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7754768133163452,
+      "learning_rate": 1.9275248376192146e-05,
+      "loss": 1.2932,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1375806331634521,
+      "learning_rate": 1.9273411699748542e-05,
+      "loss": 2.6472,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.649649977684021,
+      "learning_rate": 1.9271572786713915e-05,
+      "loss": 1.7296,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3229631185531616,
+      "learning_rate": 1.9269731637531793e-05,
+      "loss": 0.9259,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8701810836791992,
+      "learning_rate": 1.9267888252646224e-05,
+      "loss": 1.4846,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2067527770996094,
+      "learning_rate": 1.9266042632501795e-05,
+      "loss": 1.5756,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9912710785865784,
+      "learning_rate": 1.9264194777543646e-05,
+      "loss": 1.3123,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2506612539291382,
+      "learning_rate": 1.926234468821745e-05,
+      "loss": 1.7554,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2464430332183838,
+      "learning_rate": 1.9260492364969412e-05,
+      "loss": 0.4939,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.230602741241455,
+      "learning_rate": 1.9258637808246282e-05,
+      "loss": 1.2158,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0682477951049805,
+      "learning_rate": 1.925678101849535e-05,
+      "loss": 1.8281,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7103515267372131,
+      "learning_rate": 1.9254921996164446e-05,
+      "loss": 1.6215,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.81144118309021,
+      "learning_rate": 1.9253060741701928e-05,
+      "loss": 1.1984,
+      "step": 802
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 6407,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 401,
+  "total_flos": 3.024569725385441e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}