Add files using upload-large-folder tool

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e8c248ec4666836589998771cc7adce4288e4ccf58543e0cebc494ebd216ed1
+size 6406264496

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105f4663b5fad5c3987dda6e7afa5851943aeadb9dde60562019277b15fcf67
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75dc74b7b4d33947ffc13a8119e1e3d408c30a0d42312a1193ba5e9a6212f08
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a3914ca45a9e6159eb8a9ca0f1fdcaafd61bd288875c2f103ea03b43c7123
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c182c63c0a0988d43d53f2652895f2bd9e70cdb1047fb10c7c033ab97b19624
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433d37b2c38d98436fb22da8657ac2765bd2b641d2e81ff0ce99abde525551de
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbbc1c06c68cb245deeceeebcfb5fc2cbeb555072a700f0f69cf58ec4b1a92f
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/trainer_state.json

@@ -0,0 +1,3534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.1859773107680863,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003719546215361726,
+      "grad_norm": 0.533352255821228,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 4.6966,
+      "step": 2
+    },
+    {
+      "epoch": 0.0007439092430723452,
+      "grad_norm": 0.5178775191307068,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 4.6745,
+      "step": 4
+    },
+    {
+      "epoch": 0.0011158638646085177,
+      "grad_norm": 0.5809087753295898,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.6424,
+      "step": 6
+    },
+    {
+      "epoch": 0.0014878184861446904,
+      "grad_norm": 0.5083269476890564,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 4.6588,
+      "step": 8
+    },
+    {
+      "epoch": 0.001859773107680863,
+      "grad_norm": 0.4586876630783081,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 4.6517,
+      "step": 10
+    },
+    {
+      "epoch": 0.0022317277292170355,
+      "grad_norm": 0.4595833718776703,
+      "learning_rate": 2.2e-06,
+      "loss": 4.6455,
+      "step": 12
+    },
+    {
+      "epoch": 0.002603682350753208,
+      "grad_norm": 0.384199321269989,
+      "learning_rate": 2.6e-06,
+      "loss": 4.6517,
+      "step": 14
+    },
+    {
+      "epoch": 0.0029756369722893808,
+      "grad_norm": 0.4929797947406769,
+      "learning_rate": 3e-06,
+      "loss": 4.6451,
+      "step": 16
+    },
+    {
+      "epoch": 0.0033475915938255534,
+      "grad_norm": 0.39049023389816284,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.6369,
+      "step": 18
+    },
+    {
+      "epoch": 0.003719546215361726,
+      "grad_norm": 0.43155384063720703,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 4.5906,
+      "step": 20
+    },
+    {
+      "epoch": 0.004091500836897899,
+      "grad_norm": 0.3580000698566437,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 4.6157,
+      "step": 22
+    },
+    {
+      "epoch": 0.004463455458434071,
+      "grad_norm": 0.41416099667549133,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 4.6162,
+      "step": 24
+    },
+    {
+      "epoch": 0.004835410079970244,
+      "grad_norm": 0.3172813653945923,
+      "learning_rate": 5e-06,
+      "loss": 4.5839,
+      "step": 26
+    },
+    {
+      "epoch": 0.005207364701506416,
+      "grad_norm": 0.4199622869491577,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 4.5856,
+      "step": 28
+    },
+    {
+      "epoch": 0.0055793193230425885,
+      "grad_norm": 0.34996122121810913,
+      "learning_rate": 5.8e-06,
+      "loss": 4.5934,
+      "step": 30
+    },
+    {
+      "epoch": 0.0059512739445787615,
+      "grad_norm": 0.3094606399536133,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 4.5658,
+      "step": 32
+    },
+    {
+      "epoch": 0.006323228566114934,
+      "grad_norm": 0.2902088761329651,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 4.5581,
+      "step": 34
+    },
+    {
+      "epoch": 0.006695183187651107,
+      "grad_norm": 0.2611074149608612,
+      "learning_rate": 7e-06,
+      "loss": 4.5329,
+      "step": 36
+    },
+    {
+      "epoch": 0.007067137809187279,
+      "grad_norm": 0.3137070834636688,
+      "learning_rate": 7.4e-06,
+      "loss": 4.5734,
+      "step": 38
+    },
+    {
+      "epoch": 0.007439092430723452,
+      "grad_norm": 0.23878967761993408,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 4.5578,
+      "step": 40
+    },
+    {
+      "epoch": 0.007811047052259624,
+      "grad_norm": 0.23213300108909607,
+      "learning_rate": 8.2e-06,
+      "loss": 4.5328,
+      "step": 42
+    },
+    {
+      "epoch": 0.008183001673795797,
+      "grad_norm": 0.2214154154062271,
+      "learning_rate": 8.6e-06,
+      "loss": 4.5215,
+      "step": 44
+    },
+    {
+      "epoch": 0.00855495629533197,
+      "grad_norm": 0.21772630512714386,
+      "learning_rate": 9e-06,
+      "loss": 4.5373,
+      "step": 46
+    },
+    {
+      "epoch": 0.008926910916868142,
+      "grad_norm": 0.20229972898960114,
+      "learning_rate": 9.4e-06,
+      "loss": 4.5051,
+      "step": 48
+    },
+    {
+      "epoch": 0.009298865538404314,
+      "grad_norm": 0.16475613415241241,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 4.4877,
+      "step": 50
+    },
+    {
+      "epoch": 0.009670820159940488,
+      "grad_norm": 0.18578968942165375,
+      "learning_rate": 1.02e-05,
+      "loss": 4.4884,
+      "step": 52
+    },
+    {
+      "epoch": 0.01004277478147666,
+      "grad_norm": 0.17791849374771118,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 4.5105,
+      "step": 54
+    },
+    {
+      "epoch": 0.010414729403012832,
+      "grad_norm": 0.23396077752113342,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 4.4926,
+      "step": 56
+    },
+    {
+      "epoch": 0.010786684024549005,
+      "grad_norm": 0.20297282934188843,
+      "learning_rate": 1.14e-05,
+      "loss": 4.4906,
+      "step": 58
+    },
+    {
+      "epoch": 0.011158638646085177,
+      "grad_norm": 0.20387418568134308,
+      "learning_rate": 1.18e-05,
+      "loss": 4.4872,
+      "step": 60
+    },
+    {
+      "epoch": 0.01153059326762135,
+      "grad_norm": 0.19193309545516968,
+      "learning_rate": 1.22e-05,
+      "loss": 4.5102,
+      "step": 62
+    },
+    {
+      "epoch": 0.011902547889157523,
+      "grad_norm": 0.1795346885919571,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 4.4742,
+      "step": 64
+    },
+    {
+      "epoch": 0.012274502510693695,
+      "grad_norm": 0.14911119639873505,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 4.4755,
+      "step": 66
+    },
+    {
+      "epoch": 0.012646457132229868,
+      "grad_norm": 0.14775897562503815,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 4.4817,
+      "step": 68
+    },
+    {
+      "epoch": 0.01301841175376604,
+      "grad_norm": 0.18325026333332062,
+      "learning_rate": 1.38e-05,
+      "loss": 4.4734,
+      "step": 70
+    },
+    {
+      "epoch": 0.013390366375302214,
+      "grad_norm": 0.17328917980194092,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 4.4442,
+      "step": 72
+    },
+    {
+      "epoch": 0.013762320996838386,
+      "grad_norm": 0.21017277240753174,
+      "learning_rate": 1.46e-05,
+      "loss": 4.4715,
+      "step": 74
+    },
+    {
+      "epoch": 0.014134275618374558,
+      "grad_norm": 0.1855219155550003,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 4.4133,
+      "step": 76
+    },
+    {
+      "epoch": 0.01450623023991073,
+      "grad_norm": 0.13497354090213776,
+      "learning_rate": 1.54e-05,
+      "loss": 4.469,
+      "step": 78
+    },
+    {
+      "epoch": 0.014878184861446904,
+      "grad_norm": 0.15246789157390594,
+      "learning_rate": 1.58e-05,
+      "loss": 4.4619,
+      "step": 80
+    },
+    {
+      "epoch": 0.015250139482983077,
+      "grad_norm": 0.1371084600687027,
+      "learning_rate": 1.62e-05,
+      "loss": 4.4557,
+      "step": 82
+    },
+    {
+      "epoch": 0.015622094104519249,
+      "grad_norm": 0.13912861049175262,
+      "learning_rate": 1.66e-05,
+      "loss": 4.448,
+      "step": 84
+    },
+    {
+      "epoch": 0.01599404872605542,
+      "grad_norm": 0.1682051420211792,
+      "learning_rate": 1.7e-05,
+      "loss": 4.4675,
+      "step": 86
+    },
+    {
+      "epoch": 0.016366003347591595,
+      "grad_norm": 0.1429281085729599,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 4.4653,
+      "step": 88
+    },
+    {
+      "epoch": 0.016737957969127765,
+      "grad_norm": 0.1489543914794922,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 4.4825,
+      "step": 90
+    },
+    {
+      "epoch": 0.01710991259066394,
+      "grad_norm": 0.1556553840637207,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 4.4475,
+      "step": 92
+    },
+    {
+      "epoch": 0.017481867212200113,
+      "grad_norm": 0.1656656563282013,
+      "learning_rate": 1.86e-05,
+      "loss": 4.4476,
+      "step": 94
+    },
+    {
+      "epoch": 0.017853821833736284,
+      "grad_norm": 0.1872718632221222,
+      "learning_rate": 1.9e-05,
+      "loss": 4.4654,
+      "step": 96
+    },
+    {
+      "epoch": 0.018225776455272458,
+      "grad_norm": 0.1351405382156372,
+      "learning_rate": 1.94e-05,
+      "loss": 4.4748,
+      "step": 98
+    },
+    {
+      "epoch": 0.018597731076808628,
+      "grad_norm": 0.16708168387413025,
+      "learning_rate": 1.98e-05,
+      "loss": 4.43,
+      "step": 100
+    },
+    {
+      "epoch": 0.018969685698344802,
+      "grad_norm": 0.14244522154331207,
+      "learning_rate": 1.9998122770790317e-05,
+      "loss": 4.4488,
+      "step": 102
+    },
+    {
+      "epoch": 0.019341640319880976,
+      "grad_norm": 0.15444527566432953,
+      "learning_rate": 1.9994368312370943e-05,
+      "loss": 4.4653,
+      "step": 104
+    },
+    {
+      "epoch": 0.019713594941417147,
+      "grad_norm": 0.15339770913124084,
+      "learning_rate": 1.9990613853951567e-05,
+      "loss": 4.4386,
+      "step": 106
+    },
+    {
+      "epoch": 0.02008554956295332,
+      "grad_norm": 0.14203651249408722,
+      "learning_rate": 1.9986859395532197e-05,
+      "loss": 4.4231,
+      "step": 108
+    },
+    {
+      "epoch": 0.02045750418448949,
+      "grad_norm": 0.13865697383880615,
+      "learning_rate": 1.9983104937112824e-05,
+      "loss": 4.4266,
+      "step": 110
+    },
+    {
+      "epoch": 0.020829458806025665,
+      "grad_norm": 0.18033641576766968,
+      "learning_rate": 1.997935047869345e-05,
+      "loss": 4.4272,
+      "step": 112
+    },
+    {
+      "epoch": 0.02120141342756184,
+      "grad_norm": 0.18702204525470734,
+      "learning_rate": 1.9975596020274077e-05,
+      "loss": 4.4445,
+      "step": 114
+    },
+    {
+      "epoch": 0.02157336804909801,
+      "grad_norm": 0.18074890971183777,
+      "learning_rate": 1.9971841561854704e-05,
+      "loss": 4.4448,
+      "step": 116
+    },
+    {
+      "epoch": 0.021945322670634183,
+      "grad_norm": 0.13851948082447052,
+      "learning_rate": 1.996808710343533e-05,
+      "loss": 4.4635,
+      "step": 118
+    },
+    {
+      "epoch": 0.022317277292170354,
+      "grad_norm": 0.17241060733795166,
+      "learning_rate": 1.9964332645015957e-05,
+      "loss": 4.4698,
+      "step": 120
+    },
+    {
+      "epoch": 0.022689231913706528,
+      "grad_norm": 0.15968725085258484,
+      "learning_rate": 1.9960578186596584e-05,
+      "loss": 4.4553,
+      "step": 122
+    },
+    {
+      "epoch": 0.0230611865352427,
+      "grad_norm": 0.19286251068115234,
+      "learning_rate": 1.995682372817721e-05,
+      "loss": 4.4491,
+      "step": 124
+    },
+    {
+      "epoch": 0.023433141156778872,
+      "grad_norm": 0.14362020790576935,
+      "learning_rate": 1.9953069269757838e-05,
+      "loss": 4.437,
+      "step": 126
+    },
+    {
+      "epoch": 0.023805095778315046,
+      "grad_norm": 0.1533016860485077,
+      "learning_rate": 1.9949314811338468e-05,
+      "loss": 4.45,
+      "step": 128
+    },
+    {
+      "epoch": 0.024177050399851217,
+      "grad_norm": 0.1473267674446106,
+      "learning_rate": 1.9945560352919095e-05,
+      "loss": 4.4443,
+      "step": 130
+    },
+    {
+      "epoch": 0.02454900502138739,
+      "grad_norm": 0.22527185082435608,
+      "learning_rate": 1.9941805894499718e-05,
+      "loss": 4.4432,
+      "step": 132
+    },
+    {
+      "epoch": 0.024920959642923565,
+      "grad_norm": 0.17880135774612427,
+      "learning_rate": 1.9938051436080348e-05,
+      "loss": 4.4267,
+      "step": 134
+    },
+    {
+      "epoch": 0.025292914264459735,
+      "grad_norm": 0.15150447189807892,
+      "learning_rate": 1.9934296977660975e-05,
+      "loss": 4.4124,
+      "step": 136
+    },
+    {
+      "epoch": 0.02566486888599591,
+      "grad_norm": 0.15510566532611847,
+      "learning_rate": 1.99305425192416e-05,
+      "loss": 4.456,
+      "step": 138
+    },
+    {
+      "epoch": 0.02603682350753208,
+      "grad_norm": 0.15725497901439667,
+      "learning_rate": 1.9926788060822228e-05,
+      "loss": 4.4581,
+      "step": 140
+    },
+    {
+      "epoch": 0.026408778129068253,
+      "grad_norm": 0.1423843502998352,
+      "learning_rate": 1.9923033602402855e-05,
+      "loss": 4.4093,
+      "step": 142
+    },
+    {
+      "epoch": 0.026780732750604427,
+      "grad_norm": 0.13635650277137756,
+      "learning_rate": 1.9919279143983482e-05,
+      "loss": 4.4214,
+      "step": 144
+    },
+    {
+      "epoch": 0.027152687372140598,
+      "grad_norm": 0.13585899770259857,
+      "learning_rate": 1.991552468556411e-05,
+      "loss": 4.4419,
+      "step": 146
+    },
+    {
+      "epoch": 0.027524641993676772,
+      "grad_norm": 0.1389041543006897,
+      "learning_rate": 1.9911770227144735e-05,
+      "loss": 4.4246,
+      "step": 148
+    },
+    {
+      "epoch": 0.027896596615212942,
+      "grad_norm": 0.20044665038585663,
+      "learning_rate": 1.9908015768725365e-05,
+      "loss": 4.4067,
+      "step": 150
+    },
+    {
+      "epoch": 0.028268551236749116,
+      "grad_norm": 0.14172682166099548,
+      "learning_rate": 1.990426131030599e-05,
+      "loss": 4.4472,
+      "step": 152
+    },
+    {
+      "epoch": 0.02864050585828529,
+      "grad_norm": 0.150092214345932,
+      "learning_rate": 1.9900506851886615e-05,
+      "loss": 4.4632,
+      "step": 154
+    },
+    {
+      "epoch": 0.02901246047982146,
+      "grad_norm": 0.16723103821277618,
+      "learning_rate": 1.9896752393467246e-05,
+      "loss": 4.4103,
+      "step": 156
+    },
+    {
+      "epoch": 0.029384415101357635,
+      "grad_norm": 0.14167651534080505,
+      "learning_rate": 1.9892997935047872e-05,
+      "loss": 4.4015,
+      "step": 158
+    },
+    {
+      "epoch": 0.02975636972289381,
+      "grad_norm": 0.1327686309814453,
+      "learning_rate": 1.9889243476628496e-05,
+      "loss": 4.4184,
+      "step": 160
+    },
+    {
+      "epoch": 0.03012832434442998,
+      "grad_norm": 0.19752570986747742,
+      "learning_rate": 1.9885489018209126e-05,
+      "loss": 4.4612,
+      "step": 162
+    },
+    {
+      "epoch": 0.030500278965966153,
+      "grad_norm": 0.152671217918396,
+      "learning_rate": 1.9881734559789753e-05,
+      "loss": 4.435,
+      "step": 164
+    },
+    {
+      "epoch": 0.030872233587502323,
+      "grad_norm": 0.1559513658285141,
+      "learning_rate": 1.987798010137038e-05,
+      "loss": 4.4679,
+      "step": 166
+    },
+    {
+      "epoch": 0.031244188209038497,
+      "grad_norm": 0.147254079580307,
+      "learning_rate": 1.9874225642951006e-05,
+      "loss": 4.3905,
+      "step": 168
+    },
+    {
+      "epoch": 0.03161614283057467,
+      "grad_norm": 0.13503003120422363,
+      "learning_rate": 1.9870471184531633e-05,
+      "loss": 4.4504,
+      "step": 170
+    },
+    {
+      "epoch": 0.03198809745211084,
+      "grad_norm": 0.1407826691865921,
+      "learning_rate": 1.986671672611226e-05,
+      "loss": 4.4105,
+      "step": 172
+    },
+    {
+      "epoch": 0.03236005207364701,
+      "grad_norm": 0.16018295288085938,
+      "learning_rate": 1.9862962267692886e-05,
+      "loss": 4.4165,
+      "step": 174
+    },
+    {
+      "epoch": 0.03273200669518319,
+      "grad_norm": 0.1484498381614685,
+      "learning_rate": 1.9859207809273513e-05,
+      "loss": 4.4046,
+      "step": 176
+    },
+    {
+      "epoch": 0.03310396131671936,
+      "grad_norm": 0.1413147747516632,
+      "learning_rate": 1.985545335085414e-05,
+      "loss": 4.4246,
+      "step": 178
+    },
+    {
+      "epoch": 0.03347591593825553,
+      "grad_norm": 0.15720275044441223,
+      "learning_rate": 1.9851698892434767e-05,
+      "loss": 4.4034,
+      "step": 180
+    },
+    {
+      "epoch": 0.03384787055979171,
+      "grad_norm": 0.13905976712703705,
+      "learning_rate": 1.9847944434015397e-05,
+      "loss": 4.402,
+      "step": 182
+    },
+    {
+      "epoch": 0.03421982518132788,
+      "grad_norm": 0.1409996598958969,
+      "learning_rate": 1.9844189975596023e-05,
+      "loss": 4.4179,
+      "step": 184
+    },
+    {
+      "epoch": 0.03459177980286405,
+      "grad_norm": 0.1338507980108261,
+      "learning_rate": 1.9840435517176647e-05,
+      "loss": 4.4417,
+      "step": 186
+    },
+    {
+      "epoch": 0.03496373442440023,
+      "grad_norm": 0.13106590509414673,
+      "learning_rate": 1.9836681058757277e-05,
+      "loss": 4.4029,
+      "step": 188
+    },
+    {
+      "epoch": 0.0353356890459364,
+      "grad_norm": 0.1810423582792282,
+      "learning_rate": 1.9832926600337904e-05,
+      "loss": 4.4182,
+      "step": 190
+    },
+    {
+      "epoch": 0.03570764366747257,
+      "grad_norm": 0.13766156136989594,
+      "learning_rate": 1.982917214191853e-05,
+      "loss": 4.4252,
+      "step": 192
+    },
+    {
+      "epoch": 0.03607959828900874,
+      "grad_norm": 0.15079887211322784,
+      "learning_rate": 1.9825417683499157e-05,
+      "loss": 4.4083,
+      "step": 194
+    },
+    {
+      "epoch": 0.036451552910544915,
+      "grad_norm": 0.13945257663726807,
+      "learning_rate": 1.9821663225079784e-05,
+      "loss": 4.4515,
+      "step": 196
+    },
+    {
+      "epoch": 0.036823507532081086,
+      "grad_norm": 0.14432503283023834,
+      "learning_rate": 1.981790876666041e-05,
+      "loss": 4.4603,
+      "step": 198
+    },
+    {
+      "epoch": 0.037195462153617256,
+      "grad_norm": 0.13772405683994293,
+      "learning_rate": 1.9814154308241037e-05,
+      "loss": 4.4515,
+      "step": 200
+    },
+    {
+      "epoch": 0.037567416775153434,
+      "grad_norm": 0.1396275907754898,
+      "learning_rate": 1.9810399849821664e-05,
+      "loss": 4.4586,
+      "step": 202
+    },
+    {
+      "epoch": 0.037939371396689604,
+      "grad_norm": 0.13669896125793457,
+      "learning_rate": 1.9806645391402294e-05,
+      "loss": 4.4077,
+      "step": 204
+    },
+    {
+      "epoch": 0.038311326018225775,
+      "grad_norm": 0.18530793488025665,
+      "learning_rate": 1.9802890932982918e-05,
+      "loss": 4.4349,
+      "step": 206
+    },
+    {
+      "epoch": 0.03868328063976195,
+      "grad_norm": 0.14767539501190186,
+      "learning_rate": 1.9799136474563544e-05,
+      "loss": 4.4632,
+      "step": 208
+    },
+    {
+      "epoch": 0.03905523526129812,
+      "grad_norm": 0.12970390915870667,
+      "learning_rate": 1.9795382016144175e-05,
+      "loss": 4.4537,
+      "step": 210
+    },
+    {
+      "epoch": 0.03942718988283429,
+      "grad_norm": 0.1440563201904297,
+      "learning_rate": 1.9791627557724798e-05,
+      "loss": 4.4375,
+      "step": 212
+    },
+    {
+      "epoch": 0.039799144504370464,
+      "grad_norm": 0.1329580545425415,
+      "learning_rate": 1.9787873099305425e-05,
+      "loss": 4.445,
+      "step": 214
+    },
+    {
+      "epoch": 0.04017109912590664,
+      "grad_norm": 0.14992040395736694,
+      "learning_rate": 1.9784118640886055e-05,
+      "loss": 4.4401,
+      "step": 216
+    },
+    {
+      "epoch": 0.04054305374744281,
+      "grad_norm": 0.12870106101036072,
+      "learning_rate": 1.978036418246668e-05,
+      "loss": 4.4292,
+      "step": 218
+    },
+    {
+      "epoch": 0.04091500836897898,
+      "grad_norm": 0.12855996191501617,
+      "learning_rate": 1.9776609724047308e-05,
+      "loss": 4.4494,
+      "step": 220
+    },
+    {
+      "epoch": 0.04128696299051516,
+      "grad_norm": 0.16250288486480713,
+      "learning_rate": 1.9772855265627935e-05,
+      "loss": 4.3799,
+      "step": 222
+    },
+    {
+      "epoch": 0.04165891761205133,
+      "grad_norm": 0.14106884598731995,
+      "learning_rate": 1.9769100807208562e-05,
+      "loss": 4.4159,
+      "step": 224
+    },
+    {
+      "epoch": 0.0420308722335875,
+      "grad_norm": 0.13920925557613373,
+      "learning_rate": 1.976534634878919e-05,
+      "loss": 4.4049,
+      "step": 226
+    },
+    {
+      "epoch": 0.04240282685512368,
+      "grad_norm": 0.12977811694145203,
+      "learning_rate": 1.9761591890369815e-05,
+      "loss": 4.4282,
+      "step": 228
+    },
+    {
+      "epoch": 0.04277478147665985,
+      "grad_norm": 0.13805097341537476,
+      "learning_rate": 1.9757837431950442e-05,
+      "loss": 4.4297,
+      "step": 230
+    },
+    {
+      "epoch": 0.04314673609819602,
+      "grad_norm": 0.13858918845653534,
+      "learning_rate": 1.975408297353107e-05,
+      "loss": 4.3854,
+      "step": 232
+    },
+    {
+      "epoch": 0.04351869071973219,
+      "grad_norm": 0.13096466660499573,
+      "learning_rate": 1.9750328515111696e-05,
+      "loss": 4.4197,
+      "step": 234
+    },
+    {
+      "epoch": 0.04389064534126837,
+      "grad_norm": 0.13787825405597687,
+      "learning_rate": 1.9746574056692326e-05,
+      "loss": 4.3864,
+      "step": 236
+    },
+    {
+      "epoch": 0.04426259996280454,
+      "grad_norm": 0.14175890386104584,
+      "learning_rate": 1.9742819598272952e-05,
+      "loss": 4.3939,
+      "step": 238
+    },
+    {
+      "epoch": 0.04463455458434071,
+      "grad_norm": 0.14101317524909973,
+      "learning_rate": 1.9739065139853576e-05,
+      "loss": 4.4778,
+      "step": 240
+    },
+    {
+      "epoch": 0.045006509205876885,
+      "grad_norm": 0.13222607970237732,
+      "learning_rate": 1.9735310681434206e-05,
+      "loss": 4.3975,
+      "step": 242
+    },
+    {
+      "epoch": 0.045378463827413056,
+      "grad_norm": 0.1379491537809372,
+      "learning_rate": 1.9731556223014833e-05,
+      "loss": 4.396,
+      "step": 244
+    },
+    {
+      "epoch": 0.045750418448949226,
+      "grad_norm": 0.14780907332897186,
+      "learning_rate": 1.972780176459546e-05,
+      "loss": 4.4046,
+      "step": 246
+    },
+    {
+      "epoch": 0.0461223730704854,
+      "grad_norm": 0.13388580083847046,
+      "learning_rate": 1.9724047306176086e-05,
+      "loss": 4.4201,
+      "step": 248
+    },
+    {
+      "epoch": 0.046494327692021574,
+      "grad_norm": 0.12791724503040314,
+      "learning_rate": 1.9720292847756713e-05,
+      "loss": 4.401,
+      "step": 250
+    },
+    {
+      "epoch": 0.046866282313557744,
+      "grad_norm": 0.1948254257440567,
+      "learning_rate": 1.971653838933734e-05,
+      "loss": 4.4294,
+      "step": 252
+    },
+    {
+      "epoch": 0.04723823693509392,
+      "grad_norm": 0.14121216535568237,
+      "learning_rate": 1.9712783930917966e-05,
+      "loss": 4.3982,
+      "step": 254
+    },
+    {
+      "epoch": 0.04761019155663009,
+      "grad_norm": 0.13573412597179413,
+      "learning_rate": 1.9709029472498593e-05,
+      "loss": 4.4042,
+      "step": 256
+    },
+    {
+      "epoch": 0.04798214617816626,
+      "grad_norm": 0.1361589878797531,
+      "learning_rate": 1.970527501407922e-05,
+      "loss": 4.3416,
+      "step": 258
+    },
+    {
+      "epoch": 0.04835410079970243,
+      "grad_norm": 0.1423160284757614,
+      "learning_rate": 1.9701520555659847e-05,
+      "loss": 4.4366,
+      "step": 260
+    },
+    {
+      "epoch": 0.04872605542123861,
+      "grad_norm": 0.13465264439582825,
+      "learning_rate": 1.9697766097240473e-05,
+      "loss": 4.4494,
+      "step": 262
+    },
+    {
+      "epoch": 0.04909801004277478,
+      "grad_norm": 0.14051669836044312,
+      "learning_rate": 1.9694011638821103e-05,
+      "loss": 4.4369,
+      "step": 264
+    },
+    {
+      "epoch": 0.04946996466431095,
+      "grad_norm": 0.1428443044424057,
+      "learning_rate": 1.9690257180401727e-05,
+      "loss": 4.4154,
+      "step": 266
+    },
+    {
+      "epoch": 0.04984191928584713,
+      "grad_norm": 0.13515609502792358,
+      "learning_rate": 1.9686502721982354e-05,
+      "loss": 4.408,
+      "step": 268
+    },
+    {
+      "epoch": 0.0502138739073833,
+      "grad_norm": 0.12892986834049225,
+      "learning_rate": 1.9682748263562984e-05,
+      "loss": 4.3886,
+      "step": 270
+    },
+    {
+      "epoch": 0.05058582852891947,
+      "grad_norm": 0.13286785781383514,
+      "learning_rate": 1.967899380514361e-05,
+      "loss": 4.395,
+      "step": 272
+    },
+    {
+      "epoch": 0.05095778315045565,
+      "grad_norm": 0.13310199975967407,
+      "learning_rate": 1.9675239346724237e-05,
+      "loss": 4.4218,
+      "step": 274
+    },
+    {
+      "epoch": 0.05132973777199182,
+      "grad_norm": 0.13098660111427307,
+      "learning_rate": 1.9671484888304864e-05,
+      "loss": 4.3937,
+      "step": 276
+    },
+    {
+      "epoch": 0.05170169239352799,
+      "grad_norm": 0.13211186230182648,
+      "learning_rate": 1.966773042988549e-05,
+      "loss": 4.4387,
+      "step": 278
+    },
+    {
+      "epoch": 0.05207364701506416,
+      "grad_norm": 0.13477829098701477,
+      "learning_rate": 1.9663975971466117e-05,
+      "loss": 4.4251,
+      "step": 280
+    },
+    {
+      "epoch": 0.052445601636600336,
+      "grad_norm": 0.13602809607982635,
+      "learning_rate": 1.9660221513046744e-05,
+      "loss": 4.3771,
+      "step": 282
+    },
+    {
+      "epoch": 0.05281755625813651,
+      "grad_norm": 0.14181022346019745,
+      "learning_rate": 1.965646705462737e-05,
+      "loss": 4.4054,
+      "step": 284
+    },
+    {
+      "epoch": 0.05318951087967268,
+      "grad_norm": 0.16758321225643158,
+      "learning_rate": 1.9652712596207998e-05,
+      "loss": 4.4193,
+      "step": 286
+    },
+    {
+      "epoch": 0.053561465501208855,
+      "grad_norm": 0.15030401945114136,
+      "learning_rate": 1.9648958137788624e-05,
+      "loss": 4.4005,
+      "step": 288
+    },
+    {
+      "epoch": 0.053933420122745025,
+      "grad_norm": 0.14120005071163177,
+      "learning_rate": 1.9645203679369255e-05,
+      "loss": 4.4277,
+      "step": 290
+    },
+    {
+      "epoch": 0.054305374744281196,
+      "grad_norm": 0.14473338425159454,
+      "learning_rate": 1.964144922094988e-05,
+      "loss": 4.398,
+      "step": 292
+    },
+    {
+      "epoch": 0.05467732936581737,
+      "grad_norm": 0.14475566148757935,
+      "learning_rate": 1.9637694762530505e-05,
+      "loss": 4.4125,
+      "step": 294
+    },
+    {
+      "epoch": 0.055049283987353544,
+      "grad_norm": 0.14888055622577667,
+      "learning_rate": 1.9633940304111135e-05,
+      "loss": 4.3992,
+      "step": 296
+    },
+    {
+      "epoch": 0.055421238608889714,
+      "grad_norm": 0.13086941838264465,
+      "learning_rate": 1.963018584569176e-05,
+      "loss": 4.3975,
+      "step": 298
+    },
+    {
+      "epoch": 0.055793193230425885,
+      "grad_norm": 0.13955000042915344,
+      "learning_rate": 1.9626431387272388e-05,
+      "loss": 4.3917,
+      "step": 300
+    },
+    {
+      "epoch": 0.05616514785196206,
+      "grad_norm": 0.13799113035202026,
+      "learning_rate": 1.9622676928853015e-05,
+      "loss": 4.4204,
+      "step": 302
+    },
+    {
+      "epoch": 0.05653710247349823,
+      "grad_norm": 0.1439642608165741,
+      "learning_rate": 1.9618922470433642e-05,
+      "loss": 4.4221,
+      "step": 304
+    },
+    {
+      "epoch": 0.0569090570950344,
+      "grad_norm": 0.1410611867904663,
+      "learning_rate": 1.961516801201427e-05,
+      "loss": 4.4299,
+      "step": 306
+    },
+    {
+      "epoch": 0.05728101171657058,
+      "grad_norm": 0.14223700761795044,
+      "learning_rate": 1.9611413553594895e-05,
+      "loss": 4.3671,
+      "step": 308
+    },
+    {
+      "epoch": 0.05765296633810675,
+      "grad_norm": 0.12932097911834717,
+      "learning_rate": 1.9607659095175522e-05,
+      "loss": 4.4185,
+      "step": 310
+    },
+    {
+      "epoch": 0.05802492095964292,
+      "grad_norm": 0.1335015594959259,
+      "learning_rate": 1.960390463675615e-05,
+      "loss": 4.421,
+      "step": 312
+    },
+    {
+      "epoch": 0.0583968755811791,
+      "grad_norm": 0.14002254605293274,
+      "learning_rate": 1.9600150178336776e-05,
+      "loss": 4.4035,
+      "step": 314
+    },
+    {
+      "epoch": 0.05876883020271527,
+      "grad_norm": 0.14202599227428436,
+      "learning_rate": 1.9596395719917402e-05,
+      "loss": 4.3967,
+      "step": 316
+    },
+    {
+      "epoch": 0.05914078482425144,
+      "grad_norm": 0.14532427489757538,
+      "learning_rate": 1.9592641261498032e-05,
+      "loss": 4.4134,
+      "step": 318
+    },
+    {
+      "epoch": 0.05951273944578762,
+      "grad_norm": 0.1909485161304474,
+      "learning_rate": 1.9588886803078656e-05,
+      "loss": 4.4011,
+      "step": 320
+    },
+    {
+      "epoch": 0.05988469406732379,
+      "grad_norm": 0.13882078230381012,
+      "learning_rate": 1.9585132344659286e-05,
+      "loss": 4.4255,
+      "step": 322
+    },
+    {
+      "epoch": 0.06025664868885996,
+      "grad_norm": 0.14463871717453003,
+      "learning_rate": 1.9581377886239913e-05,
+      "loss": 4.3776,
+      "step": 324
+    },
+    {
+      "epoch": 0.06062860331039613,
+      "grad_norm": 0.13792262971401215,
+      "learning_rate": 1.957762342782054e-05,
+      "loss": 4.424,
+      "step": 326
+    },
+    {
+      "epoch": 0.061000557931932306,
+      "grad_norm": 0.13589176535606384,
+      "learning_rate": 1.9573868969401166e-05,
+      "loss": 4.3836,
+      "step": 328
+    },
+    {
+      "epoch": 0.061372512553468477,
+      "grad_norm": 0.13155587017536163,
+      "learning_rate": 1.9570114510981793e-05,
+      "loss": 4.379,
+      "step": 330
+    },
+    {
+      "epoch": 0.06174446717500465,
+      "grad_norm": 0.15130208432674408,
+      "learning_rate": 1.956636005256242e-05,
+      "loss": 4.4189,
+      "step": 332
+    },
+    {
+      "epoch": 0.062116421796540824,
+      "grad_norm": 0.1509840488433838,
+      "learning_rate": 1.9562605594143046e-05,
+      "loss": 4.4133,
+      "step": 334
+    },
+    {
+      "epoch": 0.062488376418076995,
+      "grad_norm": 0.1370144784450531,
+      "learning_rate": 1.9558851135723673e-05,
+      "loss": 4.4187,
+      "step": 336
+    },
+    {
+      "epoch": 0.06286033103961317,
+      "grad_norm": 0.13449043035507202,
+      "learning_rate": 1.95550966773043e-05,
+      "loss": 4.3976,
+      "step": 338
+    },
+    {
+      "epoch": 0.06323228566114934,
+      "grad_norm": 0.13795211911201477,
+      "learning_rate": 1.9551342218884927e-05,
+      "loss": 4.3746,
+      "step": 340
+    },
+    {
+      "epoch": 0.0636042402826855,
+      "grad_norm": 0.14288261532783508,
+      "learning_rate": 1.9547587760465553e-05,
+      "loss": 4.4115,
+      "step": 342
+    },
+    {
+      "epoch": 0.06397619490422168,
+      "grad_norm": 0.14026880264282227,
+      "learning_rate": 1.9543833302046183e-05,
+      "loss": 4.4228,
+      "step": 344
+    },
+    {
+      "epoch": 0.06434814952575786,
+      "grad_norm": 0.13257336616516113,
+      "learning_rate": 1.9540078843626807e-05,
+      "loss": 4.3813,
+      "step": 346
+    },
+    {
+      "epoch": 0.06472010414729402,
+      "grad_norm": 0.13401246070861816,
+      "learning_rate": 1.9536324385207434e-05,
+      "loss": 4.3923,
+      "step": 348
+    },
+    {
+      "epoch": 0.0650920587688302,
+      "grad_norm": 0.14204595983028412,
+      "learning_rate": 1.9532569926788064e-05,
+      "loss": 4.3696,
+      "step": 350
+    },
+    {
+      "epoch": 0.06546401339036638,
+      "grad_norm": 0.13598638772964478,
+      "learning_rate": 1.952881546836869e-05,
+      "loss": 4.4154,
+      "step": 352
+    },
+    {
+      "epoch": 0.06583596801190254,
+      "grad_norm": 0.13356125354766846,
+      "learning_rate": 1.9525061009949314e-05,
+      "loss": 4.3693,
+      "step": 354
+    },
+    {
+      "epoch": 0.06620792263343872,
+      "grad_norm": 0.13466878235340118,
+      "learning_rate": 1.9521306551529944e-05,
+      "loss": 4.4061,
+      "step": 356
+    },
+    {
+      "epoch": 0.0665798772549749,
+      "grad_norm": 0.13369284570217133,
+      "learning_rate": 1.951755209311057e-05,
+      "loss": 4.3761,
+      "step": 358
+    },
+    {
+      "epoch": 0.06695183187651106,
+      "grad_norm": 0.13528649508953094,
+      "learning_rate": 1.9513797634691197e-05,
+      "loss": 4.4264,
+      "step": 360
+    },
+    {
+      "epoch": 0.06732378649804724,
+      "grad_norm": 0.142612487077713,
+      "learning_rate": 1.9510043176271824e-05,
+      "loss": 4.3888,
+      "step": 362
+    },
+    {
+      "epoch": 0.06769574111958342,
+      "grad_norm": 0.13879339396953583,
+      "learning_rate": 1.950628871785245e-05,
+      "loss": 4.4022,
+      "step": 364
+    },
+    {
+      "epoch": 0.06806769574111958,
+      "grad_norm": 0.14401033520698547,
+      "learning_rate": 1.9502534259433078e-05,
+      "loss": 4.3991,
+      "step": 366
+    },
+    {
+      "epoch": 0.06843965036265576,
+      "grad_norm": 0.1392444521188736,
+      "learning_rate": 1.9498779801013704e-05,
+      "loss": 4.4214,
+      "step": 368
+    },
+    {
+      "epoch": 0.06881160498419193,
+      "grad_norm": 0.13341018557548523,
+      "learning_rate": 1.949502534259433e-05,
+      "loss": 4.4306,
+      "step": 370
+    },
+    {
+      "epoch": 0.0691835596057281,
+      "grad_norm": 0.16247300803661346,
+      "learning_rate": 1.949127088417496e-05,
+      "loss": 4.3782,
+      "step": 372
+    },
+    {
+      "epoch": 0.06955551422726428,
+      "grad_norm": 0.13398316502571106,
+      "learning_rate": 1.9487516425755585e-05,
+      "loss": 4.4159,
+      "step": 374
+    },
+    {
+      "epoch": 0.06992746884880045,
+      "grad_norm": 0.141139954328537,
+      "learning_rate": 1.9483761967336215e-05,
+      "loss": 4.3737,
+      "step": 376
+    },
+    {
+      "epoch": 0.07029942347033662,
+      "grad_norm": 0.13556070625782013,
+      "learning_rate": 1.948000750891684e-05,
+      "loss": 4.3945,
+      "step": 378
+    },
+    {
+      "epoch": 0.0706713780918728,
+      "grad_norm": 0.133789524435997,
+      "learning_rate": 1.947625305049747e-05,
+      "loss": 4.4378,
+      "step": 380
+    },
+    {
+      "epoch": 0.07104333271340896,
+      "grad_norm": 0.13298556208610535,
+      "learning_rate": 1.9472498592078095e-05,
+      "loss": 4.3831,
+      "step": 382
+    },
+    {
+      "epoch": 0.07141528733494514,
+      "grad_norm": 0.14020226895809174,
+      "learning_rate": 1.9468744133658722e-05,
+      "loss": 4.3762,
+      "step": 384
+    },
+    {
+      "epoch": 0.07178724195648131,
+      "grad_norm": 0.14437277615070343,
+      "learning_rate": 1.946498967523935e-05,
+      "loss": 4.391,
+      "step": 386
+    },
+    {
+      "epoch": 0.07215919657801748,
+      "grad_norm": 0.12593606114387512,
+      "learning_rate": 1.9461235216819975e-05,
+      "loss": 4.394,
+      "step": 388
+    },
+    {
+      "epoch": 0.07253115119955365,
+      "grad_norm": 0.13977281749248505,
+      "learning_rate": 1.9457480758400602e-05,
+      "loss": 4.3766,
+      "step": 390
+    },
+    {
+      "epoch": 0.07290310582108983,
+      "grad_norm": 0.12502682209014893,
+      "learning_rate": 1.945372629998123e-05,
+      "loss": 4.4014,
+      "step": 392
+    },
+    {
+      "epoch": 0.073275060442626,
+      "grad_norm": 0.12794789671897888,
+      "learning_rate": 1.9449971841561856e-05,
+      "loss": 4.4036,
+      "step": 394
+    },
+    {
+      "epoch": 0.07364701506416217,
+      "grad_norm": 0.13711746037006378,
+      "learning_rate": 1.9446217383142482e-05,
+      "loss": 4.3977,
+      "step": 396
+    },
+    {
+      "epoch": 0.07401896968569835,
+      "grad_norm": 0.13240709900856018,
+      "learning_rate": 1.9442462924723112e-05,
+      "loss": 4.4187,
+      "step": 398
+    },
+    {
+      "epoch": 0.07439092430723451,
+      "grad_norm": 0.1283445954322815,
+      "learning_rate": 1.9438708466303736e-05,
+      "loss": 4.4113,
+      "step": 400
+    },
+    {
+      "epoch": 0.07476287892877069,
+      "grad_norm": 0.14590546488761902,
+      "learning_rate": 1.9434954007884363e-05,
+      "loss": 4.4172,
+      "step": 402
+    },
+    {
+      "epoch": 0.07513483355030687,
+      "grad_norm": 0.14723080396652222,
+      "learning_rate": 1.9431199549464993e-05,
+      "loss": 4.3928,
+      "step": 404
+    },
+    {
+      "epoch": 0.07550678817184303,
+      "grad_norm": 0.13755865395069122,
+      "learning_rate": 1.942744509104562e-05,
+      "loss": 4.382,
+      "step": 406
+    },
+    {
+      "epoch": 0.07587874279337921,
+      "grad_norm": 0.13519679009914398,
+      "learning_rate": 1.9423690632626243e-05,
+      "loss": 4.3875,
+      "step": 408
+    },
+    {
+      "epoch": 0.07625069741491539,
+      "grad_norm": 0.1321786642074585,
+      "learning_rate": 1.9419936174206873e-05,
+      "loss": 4.37,
+      "step": 410
+    },
+    {
+      "epoch": 0.07662265203645155,
+      "grad_norm": 0.12842361629009247,
+      "learning_rate": 1.94161817157875e-05,
+      "loss": 4.3964,
+      "step": 412
+    },
+    {
+      "epoch": 0.07699460665798773,
+      "grad_norm": 0.1355700045824051,
+      "learning_rate": 1.9412427257368126e-05,
+      "loss": 4.4485,
+      "step": 414
+    },
+    {
+      "epoch": 0.0773665612795239,
+      "grad_norm": 0.13673913478851318,
+      "learning_rate": 1.9408672798948753e-05,
+      "loss": 4.4044,
+      "step": 416
+    },
+    {
+      "epoch": 0.07773851590106007,
+      "grad_norm": 0.1371925175189972,
+      "learning_rate": 1.940491834052938e-05,
+      "loss": 4.3994,
+      "step": 418
+    },
+    {
+      "epoch": 0.07811047052259625,
+      "grad_norm": 0.13546527922153473,
+      "learning_rate": 1.9401163882110007e-05,
+      "loss": 4.3984,
+      "step": 420
+    },
+    {
+      "epoch": 0.07848242514413242,
+      "grad_norm": 0.13531765341758728,
+      "learning_rate": 1.9397409423690633e-05,
+      "loss": 4.3669,
+      "step": 422
+    },
+    {
+      "epoch": 0.07885437976566859,
+      "grad_norm": 0.13071246445178986,
+      "learning_rate": 1.939365496527126e-05,
+      "loss": 4.4348,
+      "step": 424
+    },
+    {
+      "epoch": 0.07922633438720476,
+      "grad_norm": 0.13296689093112946,
+      "learning_rate": 1.938990050685189e-05,
+      "loss": 4.359,
+      "step": 426
+    },
+    {
+      "epoch": 0.07959828900874093,
+      "grad_norm": 0.13803820312023163,
+      "learning_rate": 1.9386146048432514e-05,
+      "loss": 4.3662,
+      "step": 428
+    },
+    {
+      "epoch": 0.0799702436302771,
+      "grad_norm": 0.14896219968795776,
+      "learning_rate": 1.9382391590013144e-05,
+      "loss": 4.3869,
+      "step": 430
+    },
+    {
+      "epoch": 0.08034219825181328,
+      "grad_norm": 0.13290515542030334,
+      "learning_rate": 1.937863713159377e-05,
+      "loss": 4.4208,
+      "step": 432
+    },
+    {
+      "epoch": 0.08071415287334945,
+      "grad_norm": 0.12930351495742798,
+      "learning_rate": 1.9374882673174394e-05,
+      "loss": 4.3861,
+      "step": 434
+    },
+    {
+      "epoch": 0.08108610749488562,
+      "grad_norm": 0.13772684335708618,
+      "learning_rate": 1.9371128214755024e-05,
+      "loss": 4.388,
+      "step": 436
+    },
+    {
+      "epoch": 0.0814580621164218,
+      "grad_norm": 0.13674704730510712,
+      "learning_rate": 1.936737375633565e-05,
+      "loss": 4.4175,
+      "step": 438
+    },
+    {
+      "epoch": 0.08183001673795796,
+      "grad_norm": 0.14121732115745544,
+      "learning_rate": 1.9363619297916277e-05,
+      "loss": 4.4174,
+      "step": 440
+    },
+    {
+      "epoch": 0.08220197135949414,
+      "grad_norm": 0.13311217725276947,
+      "learning_rate": 1.9359864839496904e-05,
+      "loss": 4.4044,
+      "step": 442
+    },
+    {
+      "epoch": 0.08257392598103032,
+      "grad_norm": 0.13370046019554138,
+      "learning_rate": 1.935611038107753e-05,
+      "loss": 4.3673,
+      "step": 444
+    },
+    {
+      "epoch": 0.08294588060256648,
+      "grad_norm": 0.14226962625980377,
+      "learning_rate": 1.9352355922658158e-05,
+      "loss": 4.3952,
+      "step": 446
+    },
+    {
+      "epoch": 0.08331783522410266,
+      "grad_norm": 0.13097019493579865,
+      "learning_rate": 1.9348601464238784e-05,
+      "loss": 4.3455,
+      "step": 448
+    },
+    {
+      "epoch": 0.08368978984563884,
+      "grad_norm": 0.14357155561447144,
+      "learning_rate": 1.934484700581941e-05,
+      "loss": 4.3871,
+      "step": 450
+    },
+    {
+      "epoch": 0.084061744467175,
+      "grad_norm": 0.12173137813806534,
+      "learning_rate": 1.934109254740004e-05,
+      "loss": 4.4448,
+      "step": 452
+    },
+    {
+      "epoch": 0.08443369908871118,
+      "grad_norm": 0.13161669671535492,
+      "learning_rate": 1.9337338088980665e-05,
+      "loss": 4.3652,
+      "step": 454
+    },
+    {
+      "epoch": 0.08480565371024736,
+      "grad_norm": 0.12711520493030548,
+      "learning_rate": 1.933358363056129e-05,
+      "loss": 4.4093,
+      "step": 456
+    },
+    {
+      "epoch": 0.08517760833178352,
+      "grad_norm": 0.1273198425769806,
+      "learning_rate": 1.932982917214192e-05,
+      "loss": 4.4018,
+      "step": 458
+    },
+    {
+      "epoch": 0.0855495629533197,
+      "grad_norm": 0.1377367079257965,
+      "learning_rate": 1.932607471372255e-05,
+      "loss": 4.4157,
+      "step": 460
+    },
+    {
+      "epoch": 0.08592151757485587,
+      "grad_norm": 0.1336887627840042,
+      "learning_rate": 1.932232025530317e-05,
+      "loss": 4.3873,
+      "step": 462
+    },
+    {
+      "epoch": 0.08629347219639204,
+      "grad_norm": 0.13426591455936432,
+      "learning_rate": 1.9318565796883802e-05,
+      "loss": 4.4107,
+      "step": 464
+    },
+    {
+      "epoch": 0.08666542681792822,
+      "grad_norm": 0.13059574365615845,
+      "learning_rate": 1.931481133846443e-05,
+      "loss": 4.3867,
+      "step": 466
+    },
+    {
+      "epoch": 0.08703738143946438,
+      "grad_norm": 0.13309559226036072,
+      "learning_rate": 1.9311056880045055e-05,
+      "loss": 4.3697,
+      "step": 468
+    },
+    {
+      "epoch": 0.08740933606100056,
+      "grad_norm": 0.14126601815223694,
+      "learning_rate": 1.9307302421625682e-05,
+      "loss": 4.3853,
+      "step": 470
+    },
+    {
+      "epoch": 0.08778129068253673,
+      "grad_norm": 0.13315638899803162,
+      "learning_rate": 1.930354796320631e-05,
+      "loss": 4.3723,
+      "step": 472
+    },
+    {
+      "epoch": 0.0881532453040729,
+      "grad_norm": 0.14206604659557343,
+      "learning_rate": 1.9299793504786936e-05,
+      "loss": 4.39,
+      "step": 474
+    },
+    {
+      "epoch": 0.08852519992560907,
+      "grad_norm": 0.13731342554092407,
+      "learning_rate": 1.9296039046367562e-05,
+      "loss": 4.3744,
+      "step": 476
+    },
+    {
+      "epoch": 0.08889715454714525,
+      "grad_norm": 0.1306009590625763,
+      "learning_rate": 1.929228458794819e-05,
+      "loss": 4.4182,
+      "step": 478
+    },
+    {
+      "epoch": 0.08926910916868142,
+      "grad_norm": 0.13948090374469757,
+      "learning_rate": 1.9288530129528816e-05,
+      "loss": 4.3894,
+      "step": 480
+    },
+    {
+      "epoch": 0.08964106379021759,
+      "grad_norm": 0.12682175636291504,
+      "learning_rate": 1.9284775671109443e-05,
+      "loss": 4.3898,
+      "step": 482
+    },
+    {
+      "epoch": 0.09001301841175377,
+      "grad_norm": 0.1436319649219513,
+      "learning_rate": 1.9281021212690073e-05,
+      "loss": 4.3757,
+      "step": 484
+    },
+    {
+      "epoch": 0.09038497303328993,
+      "grad_norm": 0.13116872310638428,
+      "learning_rate": 1.92772667542707e-05,
+      "loss": 4.3699,
+      "step": 486
+    },
+    {
+      "epoch": 0.09075692765482611,
+      "grad_norm": 0.1459694355726242,
+      "learning_rate": 1.9273512295851323e-05,
+      "loss": 4.3598,
+      "step": 488
+    },
+    {
+      "epoch": 0.09112888227636229,
+      "grad_norm": 0.14896368980407715,
+      "learning_rate": 1.9269757837431953e-05,
+      "loss": 4.3617,
+      "step": 490
+    },
+    {
+      "epoch": 0.09150083689789845,
+      "grad_norm": 0.13770367205142975,
+      "learning_rate": 1.926600337901258e-05,
+      "loss": 4.3511,
+      "step": 492
+    },
+    {
+      "epoch": 0.09187279151943463,
+      "grad_norm": 0.13081865012645721,
+      "learning_rate": 1.9262248920593206e-05,
+      "loss": 4.4267,
+      "step": 494
+    },
+    {
+      "epoch": 0.0922447461409708,
+      "grad_norm": 0.13160811364650726,
+      "learning_rate": 1.9258494462173833e-05,
+      "loss": 4.3666,
+      "step": 496
+    },
+    {
+      "epoch": 0.09261670076250697,
+      "grad_norm": 0.12845483422279358,
+      "learning_rate": 1.925474000375446e-05,
+      "loss": 4.4104,
+      "step": 498
+    },
+    {
+      "epoch": 0.09298865538404315,
+      "grad_norm": 0.13726617395877838,
+      "learning_rate": 1.9250985545335087e-05,
+      "loss": 4.3748,
+      "step": 500
+    },
+    {
+      "epoch": 0.09336061000557933,
+      "grad_norm": 0.13503549993038177,
+      "learning_rate": 1.9247231086915713e-05,
+      "loss": 4.3959,
+      "step": 502
+    },
+    {
+      "epoch": 0.09373256462711549,
+      "grad_norm": 0.13208508491516113,
+      "learning_rate": 1.924347662849634e-05,
+      "loss": 4.4004,
+      "step": 504
+    },
+    {
+      "epoch": 0.09410451924865167,
+      "grad_norm": 0.13900360465049744,
+      "learning_rate": 1.923972217007697e-05,
+      "loss": 4.4024,
+      "step": 506
+    },
+    {
+      "epoch": 0.09447647387018784,
+      "grad_norm": 0.13133364915847778,
+      "learning_rate": 1.9235967711657594e-05,
+      "loss": 4.381,
+      "step": 508
+    },
+    {
+      "epoch": 0.09484842849172401,
+      "grad_norm": 0.13647907972335815,
+      "learning_rate": 1.923221325323822e-05,
+      "loss": 4.3562,
+      "step": 510
+    },
+    {
+      "epoch": 0.09522038311326018,
+      "grad_norm": 0.14257870614528656,
+      "learning_rate": 1.922845879481885e-05,
+      "loss": 4.4036,
+      "step": 512
+    },
+    {
+      "epoch": 0.09559233773479635,
+      "grad_norm": 0.1274844855070114,
+      "learning_rate": 1.9224704336399477e-05,
+      "loss": 4.386,
+      "step": 514
+    },
+    {
+      "epoch": 0.09596429235633253,
+      "grad_norm": 0.12994174659252167,
+      "learning_rate": 1.9220949877980104e-05,
+      "loss": 4.386,
+      "step": 516
+    },
+    {
+      "epoch": 0.0963362469778687,
+      "grad_norm": 0.13224108517169952,
+      "learning_rate": 1.921719541956073e-05,
+      "loss": 4.4014,
+      "step": 518
+    },
+    {
+      "epoch": 0.09670820159940487,
+      "grad_norm": 0.12864114344120026,
+      "learning_rate": 1.9213440961141357e-05,
+      "loss": 4.4159,
+      "step": 520
+    },
+    {
+      "epoch": 0.09708015622094104,
+      "grad_norm": 0.13540008664131165,
+      "learning_rate": 1.9209686502721984e-05,
+      "loss": 4.3923,
+      "step": 522
+    },
+    {
+      "epoch": 0.09745211084247722,
+      "grad_norm": 0.14008626341819763,
+      "learning_rate": 1.920593204430261e-05,
+      "loss": 4.3772,
+      "step": 524
+    },
+    {
+      "epoch": 0.09782406546401338,
+      "grad_norm": 0.13843390345573425,
+      "learning_rate": 1.9202177585883238e-05,
+      "loss": 4.4156,
+      "step": 526
+    },
+    {
+      "epoch": 0.09819602008554956,
+      "grad_norm": 0.1354583203792572,
+      "learning_rate": 1.9198423127463864e-05,
+      "loss": 4.3688,
+      "step": 528
+    },
+    {
+      "epoch": 0.09856797470708574,
+      "grad_norm": 0.13531877100467682,
+      "learning_rate": 1.919466866904449e-05,
+      "loss": 4.3858,
+      "step": 530
+    },
+    {
+      "epoch": 0.0989399293286219,
+      "grad_norm": 0.14141754806041718,
+      "learning_rate": 1.9190914210625118e-05,
+      "loss": 4.3839,
+      "step": 532
+    },
+    {
+      "epoch": 0.09931188395015808,
+      "grad_norm": 0.13655562698841095,
+      "learning_rate": 1.9187159752205745e-05,
+      "loss": 4.3839,
+      "step": 534
+    },
+    {
+      "epoch": 0.09968383857169426,
+      "grad_norm": 0.12633736431598663,
+      "learning_rate": 1.918340529378637e-05,
+      "loss": 4.3472,
+      "step": 536
+    },
+    {
+      "epoch": 0.10005579319323042,
+      "grad_norm": 0.1333930641412735,
+      "learning_rate": 1.9179650835367e-05,
+      "loss": 4.3887,
+      "step": 538
+    },
+    {
+      "epoch": 0.1004277478147666,
+      "grad_norm": 0.12719005346298218,
+      "learning_rate": 1.917589637694763e-05,
+      "loss": 4.4159,
+      "step": 540
+    },
+    {
+      "epoch": 0.10079970243630278,
+      "grad_norm": 0.1315695345401764,
+      "learning_rate": 1.917214191852825e-05,
+      "loss": 4.3409,
+      "step": 542
+    },
+    {
+      "epoch": 0.10117165705783894,
+      "grad_norm": 0.12935490906238556,
+      "learning_rate": 1.9168387460108882e-05,
+      "loss": 4.3941,
+      "step": 544
+    },
+    {
+      "epoch": 0.10154361167937512,
+      "grad_norm": 0.13370758295059204,
+      "learning_rate": 1.916463300168951e-05,
+      "loss": 4.3935,
+      "step": 546
+    },
+    {
+      "epoch": 0.1019155663009113,
+      "grad_norm": 0.13255257904529572,
+      "learning_rate": 1.9160878543270135e-05,
+      "loss": 4.3811,
+      "step": 548
+    },
+    {
+      "epoch": 0.10228752092244746,
+      "grad_norm": 0.14214619994163513,
+      "learning_rate": 1.9157124084850762e-05,
+      "loss": 4.3744,
+      "step": 550
+    },
+    {
+      "epoch": 0.10265947554398364,
+      "grad_norm": 0.13551105558872223,
+      "learning_rate": 1.915336962643139e-05,
+      "loss": 4.3917,
+      "step": 552
+    },
+    {
+      "epoch": 0.10303143016551981,
+      "grad_norm": 0.15055108070373535,
+      "learning_rate": 1.9149615168012016e-05,
+      "loss": 4.3844,
+      "step": 554
+    },
+    {
+      "epoch": 0.10340338478705598,
+      "grad_norm": 0.1371210217475891,
+      "learning_rate": 1.9145860709592642e-05,
+      "loss": 4.3505,
+      "step": 556
+    },
+    {
+      "epoch": 0.10377533940859215,
+      "grad_norm": 0.13915640115737915,
+      "learning_rate": 1.914210625117327e-05,
+      "loss": 4.3148,
+      "step": 558
+    },
+    {
+      "epoch": 0.10414729403012832,
+      "grad_norm": 0.12923602759838104,
+      "learning_rate": 1.91383517927539e-05,
+      "loss": 4.3799,
+      "step": 560
+    },
+    {
+      "epoch": 0.1045192486516645,
+      "grad_norm": 0.13532975316047668,
+      "learning_rate": 1.9134597334334523e-05,
+      "loss": 4.3436,
+      "step": 562
+    },
+    {
+      "epoch": 0.10489120327320067,
+      "grad_norm": 0.13185806572437286,
+      "learning_rate": 1.913084287591515e-05,
+      "loss": 4.3523,
+      "step": 564
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.1326294094324112,
+      "learning_rate": 1.912708841749578e-05,
+      "loss": 4.372,
+      "step": 566
+    },
+    {
+      "epoch": 0.10563511251627301,
+      "grad_norm": 0.14001120626926422,
+      "learning_rate": 1.9123333959076403e-05,
+      "loss": 4.3564,
+      "step": 568
+    },
+    {
+      "epoch": 0.10600706713780919,
+      "grad_norm": 0.13127842545509338,
+      "learning_rate": 1.9119579500657033e-05,
+      "loss": 4.3919,
+      "step": 570
+    },
+    {
+      "epoch": 0.10637902175934535,
+      "grad_norm": 0.13883431255817413,
+      "learning_rate": 1.911582504223766e-05,
+      "loss": 4.3672,
+      "step": 572
+    },
+    {
+      "epoch": 0.10675097638088153,
+      "grad_norm": 0.13719244301319122,
+      "learning_rate": 1.9112070583818286e-05,
+      "loss": 4.3781,
+      "step": 574
+    },
+    {
+      "epoch": 0.10712293100241771,
+      "grad_norm": 0.12940996885299683,
+      "learning_rate": 1.9108316125398913e-05,
+      "loss": 4.3849,
+      "step": 576
+    },
+    {
+      "epoch": 0.10749488562395387,
+      "grad_norm": 0.13530011475086212,
+      "learning_rate": 1.910456166697954e-05,
+      "loss": 4.4296,
+      "step": 578
+    },
+    {
+      "epoch": 0.10786684024549005,
+      "grad_norm": 0.12660206854343414,
+      "learning_rate": 1.9100807208560167e-05,
+      "loss": 4.415,
+      "step": 580
+    },
+    {
+      "epoch": 0.10823879486702623,
+      "grad_norm": 0.13712631165981293,
+      "learning_rate": 1.9097052750140793e-05,
+      "loss": 4.3752,
+      "step": 582
+    },
+    {
+      "epoch": 0.10861074948856239,
+      "grad_norm": 0.12846918404102325,
+      "learning_rate": 1.909329829172142e-05,
+      "loss": 4.3982,
+      "step": 584
+    },
+    {
+      "epoch": 0.10898270411009857,
+      "grad_norm": 0.129754900932312,
+      "learning_rate": 1.9089543833302047e-05,
+      "loss": 4.3351,
+      "step": 586
+    },
+    {
+      "epoch": 0.10935465873163475,
+      "grad_norm": 0.12937118113040924,
+      "learning_rate": 1.9085789374882674e-05,
+      "loss": 4.3807,
+      "step": 588
+    },
+    {
+      "epoch": 0.10972661335317091,
+      "grad_norm": 0.12957482039928436,
+      "learning_rate": 1.90820349164633e-05,
+      "loss": 4.3663,
+      "step": 590
+    },
+    {
+      "epoch": 0.11009856797470709,
+      "grad_norm": 0.131433367729187,
+      "learning_rate": 1.907828045804393e-05,
+      "loss": 4.3798,
+      "step": 592
+    },
+    {
+      "epoch": 0.11047052259624326,
+      "grad_norm": 0.13091102242469788,
+      "learning_rate": 1.9074525999624557e-05,
+      "loss": 4.3949,
+      "step": 594
+    },
+    {
+      "epoch": 0.11084247721777943,
+      "grad_norm": 0.13071194291114807,
+      "learning_rate": 1.907077154120518e-05,
+      "loss": 4.4236,
+      "step": 596
+    },
+    {
+      "epoch": 0.1112144318393156,
+      "grad_norm": 0.1308487206697464,
+      "learning_rate": 1.906701708278581e-05,
+      "loss": 4.3599,
+      "step": 598
+    },
+    {
+      "epoch": 0.11158638646085177,
+      "grad_norm": 0.14426937699317932,
+      "learning_rate": 1.9063262624366437e-05,
+      "loss": 4.3439,
+      "step": 600
+    },
+    {
+      "epoch": 0.11195834108238795,
+      "grad_norm": 0.1397157460451126,
+      "learning_rate": 1.9059508165947064e-05,
+      "loss": 4.3874,
+      "step": 602
+    },
+    {
+      "epoch": 0.11233029570392412,
+      "grad_norm": 0.1336430460214615,
+      "learning_rate": 1.905575370752769e-05,
+      "loss": 4.3462,
+      "step": 604
+    },
+    {
+      "epoch": 0.11270225032546029,
+      "grad_norm": 0.14296448230743408,
+      "learning_rate": 1.9051999249108318e-05,
+      "loss": 4.3887,
+      "step": 606
+    },
+    {
+      "epoch": 0.11307420494699646,
+      "grad_norm": 0.1381000429391861,
+      "learning_rate": 1.9048244790688944e-05,
+      "loss": 4.3793,
+      "step": 608
+    },
+    {
+      "epoch": 0.11344615956853264,
+      "grad_norm": 0.12794850766658783,
+      "learning_rate": 1.904449033226957e-05,
+      "loss": 4.369,
+      "step": 610
+    },
+    {
+      "epoch": 0.1138181141900688,
+      "grad_norm": 0.13032430410385132,
+      "learning_rate": 1.9040735873850198e-05,
+      "loss": 4.3765,
+      "step": 612
+    },
+    {
+      "epoch": 0.11419006881160498,
+      "grad_norm": 0.13756553828716278,
+      "learning_rate": 1.9036981415430825e-05,
+      "loss": 4.3201,
+      "step": 614
+    },
+    {
+      "epoch": 0.11456202343314116,
+      "grad_norm": 0.1353667676448822,
+      "learning_rate": 1.903322695701145e-05,
+      "loss": 4.3625,
+      "step": 616
+    },
+    {
+      "epoch": 0.11493397805467732,
+      "grad_norm": 0.12717565894126892,
+      "learning_rate": 1.9029472498592078e-05,
+      "loss": 4.4,
+      "step": 618
+    },
+    {
+      "epoch": 0.1153059326762135,
+      "grad_norm": 0.1389060616493225,
+      "learning_rate": 1.902571804017271e-05,
+      "loss": 4.3831,
+      "step": 620
+    },
+    {
+      "epoch": 0.11567788729774968,
+      "grad_norm": 0.13268208503723145,
+      "learning_rate": 1.9021963581753332e-05,
+      "loss": 4.3828,
+      "step": 622
+    },
+    {
+      "epoch": 0.11604984191928584,
+      "grad_norm": 0.14441059529781342,
+      "learning_rate": 1.9018209123333962e-05,
+      "loss": 4.3624,
+      "step": 624
+    },
+    {
+      "epoch": 0.11642179654082202,
+      "grad_norm": 0.1376134306192398,
+      "learning_rate": 1.901445466491459e-05,
+      "loss": 4.3811,
+      "step": 626
+    },
+    {
+      "epoch": 0.1167937511623582,
+      "grad_norm": 0.12918676435947418,
+      "learning_rate": 1.9010700206495215e-05,
+      "loss": 4.3897,
+      "step": 628
+    },
+    {
+      "epoch": 0.11716570578389436,
+      "grad_norm": 0.13482637703418732,
+      "learning_rate": 1.9006945748075842e-05,
+      "loss": 4.3574,
+      "step": 630
+    },
+    {
+      "epoch": 0.11753766040543054,
+      "grad_norm": 0.13907648622989655,
+      "learning_rate": 1.900319128965647e-05,
+      "loss": 4.3398,
+      "step": 632
+    },
+    {
+      "epoch": 0.11790961502696672,
+      "grad_norm": 0.12552416324615479,
+      "learning_rate": 1.8999436831237096e-05,
+      "loss": 4.3875,
+      "step": 634
+    },
+    {
+      "epoch": 0.11828156964850288,
+      "grad_norm": 0.13106204569339752,
+      "learning_rate": 1.8995682372817722e-05,
+      "loss": 4.4173,
+      "step": 636
+    },
+    {
+      "epoch": 0.11865352427003906,
+      "grad_norm": 0.12973403930664062,
+      "learning_rate": 1.899192791439835e-05,
+      "loss": 4.3932,
+      "step": 638
+    },
+    {
+      "epoch": 0.11902547889157523,
+      "grad_norm": 0.13355118036270142,
+      "learning_rate": 1.898817345597898e-05,
+      "loss": 4.3635,
+      "step": 640
+    },
+    {
+      "epoch": 0.1193974335131114,
+      "grad_norm": 0.1307496279478073,
+      "learning_rate": 1.8984418997559603e-05,
+      "loss": 4.3589,
+      "step": 642
+    },
+    {
+      "epoch": 0.11976938813464758,
+      "grad_norm": 0.12556278705596924,
+      "learning_rate": 1.898066453914023e-05,
+      "loss": 4.3574,
+      "step": 644
+    },
+    {
+      "epoch": 0.12014134275618374,
+      "grad_norm": 0.1354147344827652,
+      "learning_rate": 1.897691008072086e-05,
+      "loss": 4.3809,
+      "step": 646
+    },
+    {
+      "epoch": 0.12051329737771992,
+      "grad_norm": 0.1342896968126297,
+      "learning_rate": 1.8973155622301486e-05,
+      "loss": 4.3703,
+      "step": 648
+    },
+    {
+      "epoch": 0.1208852519992561,
+      "grad_norm": 0.13371160626411438,
+      "learning_rate": 1.896940116388211e-05,
+      "loss": 4.3874,
+      "step": 650
+    },
+    {
+      "epoch": 0.12125720662079226,
+      "grad_norm": 0.13642026484012604,
+      "learning_rate": 1.896564670546274e-05,
+      "loss": 4.3789,
+      "step": 652
+    },
+    {
+      "epoch": 0.12162916124232843,
+      "grad_norm": 0.13028636574745178,
+      "learning_rate": 1.8961892247043366e-05,
+      "loss": 4.3535,
+      "step": 654
+    },
+    {
+      "epoch": 0.12200111586386461,
+      "grad_norm": 0.14043664932250977,
+      "learning_rate": 1.895813778862399e-05,
+      "loss": 4.3643,
+      "step": 656
+    },
+    {
+      "epoch": 0.12237307048540078,
+      "grad_norm": 0.13966239988803864,
+      "learning_rate": 1.895438333020462e-05,
+      "loss": 4.3726,
+      "step": 658
+    },
+    {
+      "epoch": 0.12274502510693695,
+      "grad_norm": 0.13146382570266724,
+      "learning_rate": 1.8950628871785247e-05,
+      "loss": 4.3731,
+      "step": 660
+    },
+    {
+      "epoch": 0.12311697972847313,
+      "grad_norm": 0.137840136885643,
+      "learning_rate": 1.8946874413365873e-05,
+      "loss": 4.3579,
+      "step": 662
+    },
+    {
+      "epoch": 0.1234889343500093,
+      "grad_norm": 0.13051000237464905,
+      "learning_rate": 1.89431199549465e-05,
+      "loss": 4.3703,
+      "step": 664
+    },
+    {
+      "epoch": 0.12386088897154547,
+      "grad_norm": 0.1382361799478531,
+      "learning_rate": 1.8939365496527127e-05,
+      "loss": 4.3825,
+      "step": 666
+    },
+    {
+      "epoch": 0.12423284359308165,
+      "grad_norm": 0.12843891978263855,
+      "learning_rate": 1.8935611038107754e-05,
+      "loss": 4.3887,
+      "step": 668
+    },
+    {
+      "epoch": 0.12460479821461781,
+      "grad_norm": 0.1376434564590454,
+      "learning_rate": 1.893185657968838e-05,
+      "loss": 4.3324,
+      "step": 670
+    },
+    {
+      "epoch": 0.12497675283615399,
+      "grad_norm": 0.13323715329170227,
+      "learning_rate": 1.8928102121269007e-05,
+      "loss": 4.383,
+      "step": 672
+    },
+    {
+      "epoch": 0.12534870745769017,
+      "grad_norm": 0.13153232634067535,
+      "learning_rate": 1.8924347662849637e-05,
+      "loss": 4.4038,
+      "step": 674
+    },
+    {
+      "epoch": 0.12572066207922633,
+      "grad_norm": 0.12972821295261383,
+      "learning_rate": 1.892059320443026e-05,
+      "loss": 4.3518,
+      "step": 676
+    },
+    {
+      "epoch": 0.1260926167007625,
+      "grad_norm": 0.13002286851406097,
+      "learning_rate": 1.891683874601089e-05,
+      "loss": 4.3653,
+      "step": 678
+    },
+    {
+      "epoch": 0.12646457132229869,
+      "grad_norm": 0.13510707020759583,
+      "learning_rate": 1.8913084287591518e-05,
+      "loss": 4.3415,
+      "step": 680
+    },
+    {
+      "epoch": 0.12683652594383485,
+      "grad_norm": 0.13580366969108582,
+      "learning_rate": 1.8909329829172144e-05,
+      "loss": 4.3992,
+      "step": 682
+    },
+    {
+      "epoch": 0.127208480565371,
+      "grad_norm": 0.13878294825553894,
+      "learning_rate": 1.890557537075277e-05,
+      "loss": 4.3479,
+      "step": 684
+    },
+    {
+      "epoch": 0.1275804351869072,
+      "grad_norm": 0.13697321712970734,
+      "learning_rate": 1.8901820912333398e-05,
+      "loss": 4.384,
+      "step": 686
+    },
+    {
+      "epoch": 0.12795238980844337,
+      "grad_norm": 0.1431383490562439,
+      "learning_rate": 1.8898066453914024e-05,
+      "loss": 4.318,
+      "step": 688
+    },
+    {
+      "epoch": 0.12832434442997953,
+      "grad_norm": 0.1371629387140274,
+      "learning_rate": 1.889431199549465e-05,
+      "loss": 4.3785,
+      "step": 690
+    },
+    {
+      "epoch": 0.12869629905151572,
+      "grad_norm": 0.12796024978160858,
+      "learning_rate": 1.8890557537075278e-05,
+      "loss": 4.3811,
+      "step": 692
+    },
+    {
+      "epoch": 0.12906825367305189,
+      "grad_norm": 0.1320897936820984,
+      "learning_rate": 1.8886803078655908e-05,
+      "loss": 4.4124,
+      "step": 694
+    },
+    {
+      "epoch": 0.12944020829458805,
+      "grad_norm": 0.13304924964904785,
+      "learning_rate": 1.888304862023653e-05,
+      "loss": 4.3056,
+      "step": 696
+    },
+    {
+      "epoch": 0.12981216291612424,
+      "grad_norm": 0.13780461251735687,
+      "learning_rate": 1.8879294161817158e-05,
+      "loss": 4.3879,
+      "step": 698
+    },
+    {
+      "epoch": 0.1301841175376604,
+      "grad_norm": 0.1410691887140274,
+      "learning_rate": 1.887553970339779e-05,
+      "loss": 4.3498,
+      "step": 700
+    },
+    {
+      "epoch": 0.13055607215919657,
+      "grad_norm": 0.13614703714847565,
+      "learning_rate": 1.8871785244978412e-05,
+      "loss": 4.3854,
+      "step": 702
+    },
+    {
+      "epoch": 0.13092802678073276,
+      "grad_norm": 0.13683539628982544,
+      "learning_rate": 1.886803078655904e-05,
+      "loss": 4.4051,
+      "step": 704
+    },
+    {
+      "epoch": 0.13129998140226892,
+      "grad_norm": 0.12316402792930603,
+      "learning_rate": 1.886427632813967e-05,
+      "loss": 4.3919,
+      "step": 706
+    },
+    {
+      "epoch": 0.1316719360238051,
+      "grad_norm": 0.1327832043170929,
+      "learning_rate": 1.8860521869720295e-05,
+      "loss": 4.3329,
+      "step": 708
+    },
+    {
+      "epoch": 0.13204389064534128,
+      "grad_norm": 0.13137994706630707,
+      "learning_rate": 1.8856767411300922e-05,
+      "loss": 4.3521,
+      "step": 710
+    },
+    {
+      "epoch": 0.13241584526687744,
+      "grad_norm": 0.14000530540943146,
+      "learning_rate": 1.885301295288155e-05,
+      "loss": 4.3785,
+      "step": 712
+    },
+    {
+      "epoch": 0.1327877998884136,
+      "grad_norm": 0.13646899163722992,
+      "learning_rate": 1.8849258494462176e-05,
+      "loss": 4.3712,
+      "step": 714
+    },
+    {
+      "epoch": 0.1331597545099498,
+      "grad_norm": 0.12759016454219818,
+      "learning_rate": 1.8845504036042802e-05,
+      "loss": 4.3448,
+      "step": 716
+    },
+    {
+      "epoch": 0.13353170913148596,
+      "grad_norm": 0.13022704422473907,
+      "learning_rate": 1.884174957762343e-05,
+      "loss": 4.382,
+      "step": 718
+    },
+    {
+      "epoch": 0.13390366375302212,
+      "grad_norm": 0.13437943160533905,
+      "learning_rate": 1.8837995119204056e-05,
+      "loss": 4.3742,
+      "step": 720
+    },
+    {
+      "epoch": 0.13427561837455831,
+      "grad_norm": 0.1388150453567505,
+      "learning_rate": 1.8834240660784683e-05,
+      "loss": 4.3729,
+      "step": 722
+    },
+    {
+      "epoch": 0.13464757299609448,
+      "grad_norm": 0.13424767553806305,
+      "learning_rate": 1.883048620236531e-05,
+      "loss": 4.4318,
+      "step": 724
+    },
+    {
+      "epoch": 0.13501952761763064,
+      "grad_norm": 0.13936109840869904,
+      "learning_rate": 1.8826731743945936e-05,
+      "loss": 4.3727,
+      "step": 726
+    },
+    {
+      "epoch": 0.13539148223916683,
+      "grad_norm": 0.12856172025203705,
+      "learning_rate": 1.8822977285526566e-05,
+      "loss": 4.3538,
+      "step": 728
+    },
+    {
+      "epoch": 0.135763436860703,
+      "grad_norm": 0.1381312608718872,
+      "learning_rate": 1.881922282710719e-05,
+      "loss": 4.37,
+      "step": 730
+    },
+    {
+      "epoch": 0.13613539148223916,
+      "grad_norm": 0.13347133994102478,
+      "learning_rate": 1.881546836868782e-05,
+      "loss": 4.3815,
+      "step": 732
+    },
+    {
+      "epoch": 0.13650734610377535,
+      "grad_norm": 0.12941986322402954,
+      "learning_rate": 1.8811713910268446e-05,
+      "loss": 4.3527,
+      "step": 734
+    },
+    {
+      "epoch": 0.13687930072531151,
+      "grad_norm": 0.13216722011566162,
+      "learning_rate": 1.8807959451849073e-05,
+      "loss": 4.3403,
+      "step": 736
+    },
+    {
+      "epoch": 0.13725125534684768,
+      "grad_norm": 0.13473846018314362,
+      "learning_rate": 1.88042049934297e-05,
+      "loss": 4.3244,
+      "step": 738
+    },
+    {
+      "epoch": 0.13762320996838387,
+      "grad_norm": 0.12531913816928864,
+      "learning_rate": 1.8800450535010327e-05,
+      "loss": 4.3608,
+      "step": 740
+    },
+    {
+      "epoch": 0.13799516458992003,
+      "grad_norm": 0.13019171357154846,
+      "learning_rate": 1.8796696076590953e-05,
+      "loss": 4.349,
+      "step": 742
+    },
+    {
+      "epoch": 0.1383671192114562,
+      "grad_norm": 0.1324409693479538,
+      "learning_rate": 1.879294161817158e-05,
+      "loss": 4.3391,
+      "step": 744
+    },
+    {
+      "epoch": 0.1387390738329924,
+      "grad_norm": 0.13451436161994934,
+      "learning_rate": 1.8789187159752207e-05,
+      "loss": 4.354,
+      "step": 746
+    },
+    {
+      "epoch": 0.13911102845452855,
+      "grad_norm": 0.13273227214813232,
+      "learning_rate": 1.8785432701332834e-05,
+      "loss": 4.3597,
+      "step": 748
+    },
+    {
+      "epoch": 0.13948298307606471,
+      "grad_norm": 0.12722504138946533,
+      "learning_rate": 1.878167824291346e-05,
+      "loss": 4.3928,
+      "step": 750
+    },
+    {
+      "epoch": 0.1398549376976009,
+      "grad_norm": 0.13655436038970947,
+      "learning_rate": 1.8777923784494087e-05,
+      "loss": 4.3545,
+      "step": 752
+    },
+    {
+      "epoch": 0.14022689231913707,
+      "grad_norm": 0.1300223469734192,
+      "learning_rate": 1.8774169326074717e-05,
+      "loss": 4.3818,
+      "step": 754
+    },
+    {
+      "epoch": 0.14059884694067323,
+      "grad_norm": 0.13225069642066956,
+      "learning_rate": 1.877041486765534e-05,
+      "loss": 4.3995,
+      "step": 756
+    },
+    {
+      "epoch": 0.1409708015622094,
+      "grad_norm": 0.13446737825870514,
+      "learning_rate": 1.8766660409235967e-05,
+      "loss": 4.3557,
+      "step": 758
+    },
+    {
+      "epoch": 0.1413427561837456,
+      "grad_norm": 0.12839320302009583,
+      "learning_rate": 1.8762905950816598e-05,
+      "loss": 4.3881,
+      "step": 760
+    },
+    {
+      "epoch": 0.14171471080528175,
+      "grad_norm": 0.13248784840106964,
+      "learning_rate": 1.8759151492397224e-05,
+      "loss": 4.3668,
+      "step": 762
+    },
+    {
+      "epoch": 0.14208666542681792,
+      "grad_norm": 0.13975068926811218,
+      "learning_rate": 1.875539703397785e-05,
+      "loss": 4.3976,
+      "step": 764
+    },
+    {
+      "epoch": 0.1424586200483541,
+      "grad_norm": 0.13105648756027222,
+      "learning_rate": 1.8751642575558478e-05,
+      "loss": 4.366,
+      "step": 766
+    },
+    {
+      "epoch": 0.14283057466989027,
+      "grad_norm": 0.1342775970697403,
+      "learning_rate": 1.8747888117139105e-05,
+      "loss": 4.3601,
+      "step": 768
+    },
+    {
+      "epoch": 0.14320252929142643,
+      "grad_norm": 0.13071490824222565,
+      "learning_rate": 1.874413365871973e-05,
+      "loss": 4.3627,
+      "step": 770
+    },
+    {
+      "epoch": 0.14357448391296262,
+      "grad_norm": 0.13421881198883057,
+      "learning_rate": 1.8740379200300358e-05,
+      "loss": 4.3559,
+      "step": 772
+    },
+    {
+      "epoch": 0.1439464385344988,
+      "grad_norm": 0.1397150754928589,
+      "learning_rate": 1.8736624741880985e-05,
+      "loss": 4.3344,
+      "step": 774
+    },
+    {
+      "epoch": 0.14431839315603495,
+      "grad_norm": 0.14625652134418488,
+      "learning_rate": 1.873287028346161e-05,
+      "loss": 4.3746,
+      "step": 776
+    },
+    {
+      "epoch": 0.14469034777757114,
+      "grad_norm": 0.1317351758480072,
+      "learning_rate": 1.8729115825042238e-05,
+      "loss": 4.3942,
+      "step": 778
+    },
+    {
+      "epoch": 0.1450623023991073,
+      "grad_norm": 0.1383458375930786,
+      "learning_rate": 1.8725361366622865e-05,
+      "loss": 4.4239,
+      "step": 780
+    },
+    {
+      "epoch": 0.14543425702064347,
+      "grad_norm": 0.12745338678359985,
+      "learning_rate": 1.8721606908203495e-05,
+      "loss": 4.3927,
+      "step": 782
+    },
+    {
+      "epoch": 0.14580621164217966,
+      "grad_norm": 0.14058049023151398,
+      "learning_rate": 1.871785244978412e-05,
+      "loss": 4.3699,
+      "step": 784
+    },
+    {
+      "epoch": 0.14617816626371583,
+      "grad_norm": 0.13223855197429657,
+      "learning_rate": 1.871409799136475e-05,
+      "loss": 4.3988,
+      "step": 786
+    },
+    {
+      "epoch": 0.146550120885252,
+      "grad_norm": 0.13360562920570374,
+      "learning_rate": 1.8710343532945375e-05,
+      "loss": 4.3629,
+      "step": 788
+    },
+    {
+      "epoch": 0.14692207550678818,
+      "grad_norm": 0.13008005917072296,
+      "learning_rate": 1.8706589074526e-05,
+      "loss": 4.3631,
+      "step": 790
+    },
+    {
+      "epoch": 0.14729403012832434,
+      "grad_norm": 0.13139700889587402,
+      "learning_rate": 1.870283461610663e-05,
+      "loss": 4.3553,
+      "step": 792
+    },
+    {
+      "epoch": 0.1476659847498605,
+      "grad_norm": 0.1296335905790329,
+      "learning_rate": 1.8699080157687256e-05,
+      "loss": 4.3994,
+      "step": 794
+    },
+    {
+      "epoch": 0.1480379393713967,
+      "grad_norm": 0.1294126957654953,
+      "learning_rate": 1.8695325699267882e-05,
+      "loss": 4.3772,
+      "step": 796
+    },
+    {
+      "epoch": 0.14840989399293286,
+      "grad_norm": 0.13823768496513367,
+      "learning_rate": 1.869157124084851e-05,
+      "loss": 4.3249,
+      "step": 798
+    },
+    {
+      "epoch": 0.14878184861446903,
+      "grad_norm": 0.13655458390712738,
+      "learning_rate": 1.8687816782429136e-05,
+      "loss": 4.3378,
+      "step": 800
+    },
+    {
+      "epoch": 0.14915380323600522,
+      "grad_norm": 0.13484789431095123,
+      "learning_rate": 1.8684062324009763e-05,
+      "loss": 4.359,
+      "step": 802
+    },
+    {
+      "epoch": 0.14952575785754138,
+      "grad_norm": 0.13672243058681488,
+      "learning_rate": 1.868030786559039e-05,
+      "loss": 4.3685,
+      "step": 804
+    },
+    {
+      "epoch": 0.14989771247907754,
+      "grad_norm": 0.12513422966003418,
+      "learning_rate": 1.8676553407171016e-05,
+      "loss": 4.3772,
+      "step": 806
+    },
+    {
+      "epoch": 0.15026966710061374,
+      "grad_norm": 0.12852630019187927,
+      "learning_rate": 1.8672798948751646e-05,
+      "loss": 4.38,
+      "step": 808
+    },
+    {
+      "epoch": 0.1506416217221499,
+      "grad_norm": 0.13008330762386322,
+      "learning_rate": 1.866904449033227e-05,
+      "loss": 4.3776,
+      "step": 810
+    },
+    {
+      "epoch": 0.15101357634368606,
+      "grad_norm": 0.13468845188617706,
+      "learning_rate": 1.8665290031912896e-05,
+      "loss": 4.3661,
+      "step": 812
+    },
+    {
+      "epoch": 0.15138553096522225,
+      "grad_norm": 0.12655708193778992,
+      "learning_rate": 1.8661535573493526e-05,
+      "loss": 4.3779,
+      "step": 814
+    },
+    {
+      "epoch": 0.15175748558675842,
+      "grad_norm": 0.13566717505455017,
+      "learning_rate": 1.8657781115074153e-05,
+      "loss": 4.3342,
+      "step": 816
+    },
+    {
+      "epoch": 0.15212944020829458,
+      "grad_norm": 0.13631191849708557,
+      "learning_rate": 1.865402665665478e-05,
+      "loss": 4.3552,
+      "step": 818
+    },
+    {
+      "epoch": 0.15250139482983077,
+      "grad_norm": 0.1328413039445877,
+      "learning_rate": 1.8650272198235407e-05,
+      "loss": 4.38,
+      "step": 820
+    },
+    {
+      "epoch": 0.15287334945136694,
+      "grad_norm": 0.1282702535390854,
+      "learning_rate": 1.8646517739816033e-05,
+      "loss": 4.3547,
+      "step": 822
+    },
+    {
+      "epoch": 0.1532453040729031,
+      "grad_norm": 0.13456816971302032,
+      "learning_rate": 1.864276328139666e-05,
+      "loss": 4.3662,
+      "step": 824
+    },
+    {
+      "epoch": 0.1536172586944393,
+      "grad_norm": 0.130121648311615,
+      "learning_rate": 1.8639008822977287e-05,
+      "loss": 4.3561,
+      "step": 826
+    },
+    {
+      "epoch": 0.15398921331597545,
+      "grad_norm": 0.1271282583475113,
+      "learning_rate": 1.8635254364557914e-05,
+      "loss": 4.355,
+      "step": 828
+    },
+    {
+      "epoch": 0.15436116793751162,
+      "grad_norm": 0.13474541902542114,
+      "learning_rate": 1.863149990613854e-05,
+      "loss": 4.3779,
+      "step": 830
+    },
+    {
+      "epoch": 0.1547331225590478,
+      "grad_norm": 0.12945345044136047,
+      "learning_rate": 1.8627745447719167e-05,
+      "loss": 4.3654,
+      "step": 832
+    },
+    {
+      "epoch": 0.15510507718058397,
+      "grad_norm": 0.14170309901237488,
+      "learning_rate": 1.8623990989299797e-05,
+      "loss": 4.3825,
+      "step": 834
+    },
+    {
+      "epoch": 0.15547703180212014,
+      "grad_norm": 0.13123078644275665,
+      "learning_rate": 1.862023653088042e-05,
+      "loss": 4.3409,
+      "step": 836
+    },
+    {
+      "epoch": 0.15584898642365633,
+      "grad_norm": 0.13768048584461212,
+      "learning_rate": 1.8616482072461047e-05,
+      "loss": 4.3732,
+      "step": 838
+    },
+    {
+      "epoch": 0.1562209410451925,
+      "grad_norm": 0.1368185579776764,
+      "learning_rate": 1.8612727614041678e-05,
+      "loss": 4.3548,
+      "step": 840
+    },
+    {
+      "epoch": 0.15659289566672865,
+      "grad_norm": 0.1346227526664734,
+      "learning_rate": 1.8608973155622304e-05,
+      "loss": 4.353,
+      "step": 842
+    },
+    {
+      "epoch": 0.15696485028826485,
+      "grad_norm": 0.13609376549720764,
+      "learning_rate": 1.8605218697202928e-05,
+      "loss": 4.3921,
+      "step": 844
+    },
+    {
+      "epoch": 0.157336804909801,
+      "grad_norm": 0.13070380687713623,
+      "learning_rate": 1.8601464238783558e-05,
+      "loss": 4.3627,
+      "step": 846
+    },
+    {
+      "epoch": 0.15770875953133717,
+      "grad_norm": 0.13254469633102417,
+      "learning_rate": 1.8597709780364185e-05,
+      "loss": 4.3838,
+      "step": 848
+    },
+    {
+      "epoch": 0.15808071415287334,
+      "grad_norm": 0.14121773838996887,
+      "learning_rate": 1.859395532194481e-05,
+      "loss": 4.3753,
+      "step": 850
+    },
+    {
+      "epoch": 0.15845266877440953,
+      "grad_norm": 0.13469132781028748,
+      "learning_rate": 1.8590200863525438e-05,
+      "loss": 4.4014,
+      "step": 852
+    },
+    {
+      "epoch": 0.1588246233959457,
+      "grad_norm": 0.1369878649711609,
+      "learning_rate": 1.8586446405106065e-05,
+      "loss": 4.346,
+      "step": 854
+    },
+    {
+      "epoch": 0.15919657801748185,
+      "grad_norm": 0.12302210927009583,
+      "learning_rate": 1.858269194668669e-05,
+      "loss": 4.3856,
+      "step": 856
+    },
+    {
+      "epoch": 0.15956853263901805,
+      "grad_norm": 0.13149301707744598,
+      "learning_rate": 1.8578937488267318e-05,
+      "loss": 4.3384,
+      "step": 858
+    },
+    {
+      "epoch": 0.1599404872605542,
+      "grad_norm": 0.13028648495674133,
+      "learning_rate": 1.8575183029847945e-05,
+      "loss": 4.3674,
+      "step": 860
+    },
+    {
+      "epoch": 0.16031244188209037,
+      "grad_norm": 0.13054941594600677,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 4.3697,
+      "step": 862
+    },
+    {
+      "epoch": 0.16068439650362656,
+      "grad_norm": 0.13475953042507172,
+      "learning_rate": 1.85676741130092e-05,
+      "loss": 4.3903,
+      "step": 864
+    },
+    {
+      "epoch": 0.16105635112516273,
+      "grad_norm": 0.1289464682340622,
+      "learning_rate": 1.8563919654589825e-05,
+      "loss": 4.4005,
+      "step": 866
+    },
+    {
+      "epoch": 0.1614283057466989,
+      "grad_norm": 0.1286720335483551,
+      "learning_rate": 1.8560165196170455e-05,
+      "loss": 4.407,
+      "step": 868
+    },
+    {
+      "epoch": 0.16180026036823508,
+      "grad_norm": 0.13745172321796417,
+      "learning_rate": 1.8556410737751082e-05,
+      "loss": 4.3069,
+      "step": 870
+    },
+    {
+      "epoch": 0.16217221498977125,
+      "grad_norm": 0.1290927678346634,
+      "learning_rate": 1.855265627933171e-05,
+      "loss": 4.3376,
+      "step": 872
+    },
+    {
+      "epoch": 0.1625441696113074,
+      "grad_norm": 0.1270480453968048,
+      "learning_rate": 1.8548901820912336e-05,
+      "loss": 4.3829,
+      "step": 874
+    },
+    {
+      "epoch": 0.1629161242328436,
+      "grad_norm": 0.1336260586977005,
+      "learning_rate": 1.8545147362492962e-05,
+      "loss": 4.3654,
+      "step": 876
+    },
+    {
+      "epoch": 0.16328807885437976,
+      "grad_norm": 0.13343611359596252,
+      "learning_rate": 1.854139290407359e-05,
+      "loss": 4.3649,
+      "step": 878
+    },
+    {
+      "epoch": 0.16366003347591593,
+      "grad_norm": 0.1356964409351349,
+      "learning_rate": 1.8537638445654216e-05,
+      "loss": 4.3508,
+      "step": 880
+    },
+    {
+      "epoch": 0.16403198809745212,
+      "grad_norm": 0.12377387285232544,
+      "learning_rate": 1.8533883987234843e-05,
+      "loss": 4.3883,
+      "step": 882
+    },
+    {
+      "epoch": 0.16440394271898828,
+      "grad_norm": 0.12946580350399017,
+      "learning_rate": 1.853012952881547e-05,
+      "loss": 4.3876,
+      "step": 884
+    },
+    {
+      "epoch": 0.16477589734052445,
+      "grad_norm": 0.14021383225917816,
+      "learning_rate": 1.8526375070396096e-05,
+      "loss": 4.355,
+      "step": 886
+    },
+    {
+      "epoch": 0.16514785196206064,
+      "grad_norm": 0.1321079134941101,
+      "learning_rate": 1.8522620611976726e-05,
+      "loss": 4.3401,
+      "step": 888
+    },
+    {
+      "epoch": 0.1655198065835968,
+      "grad_norm": 0.13128367066383362,
+      "learning_rate": 1.851886615355735e-05,
+      "loss": 4.3568,
+      "step": 890
+    },
+    {
+      "epoch": 0.16589176120513296,
+      "grad_norm": 0.13117848336696625,
+      "learning_rate": 1.8515111695137976e-05,
+      "loss": 4.3736,
+      "step": 892
+    },
+    {
+      "epoch": 0.16626371582666916,
+      "grad_norm": 0.13289998471736908,
+      "learning_rate": 1.8511357236718606e-05,
+      "loss": 4.3684,
+      "step": 894
+    },
+    {
+      "epoch": 0.16663567044820532,
+      "grad_norm": 0.1265881359577179,
+      "learning_rate": 1.8507602778299233e-05,
+      "loss": 4.3805,
+      "step": 896
+    },
+    {
+      "epoch": 0.16700762506974148,
+      "grad_norm": 0.13073360919952393,
+      "learning_rate": 1.8503848319879857e-05,
+      "loss": 4.376,
+      "step": 898
+    },
+    {
+      "epoch": 0.16737957969127767,
+      "grad_norm": 0.13863076269626617,
+      "learning_rate": 1.8500093861460487e-05,
+      "loss": 4.3789,
+      "step": 900
+    },
+    {
+      "epoch": 0.16775153431281384,
+      "grad_norm": 0.1364503800868988,
+      "learning_rate": 1.8496339403041113e-05,
+      "loss": 4.372,
+      "step": 902
+    },
+    {
+      "epoch": 0.16812348893435,
+      "grad_norm": 0.1466122567653656,
+      "learning_rate": 1.849258494462174e-05,
+      "loss": 4.3773,
+      "step": 904
+    },
+    {
+      "epoch": 0.1684954435558862,
+      "grad_norm": 0.14153607189655304,
+      "learning_rate": 1.8488830486202367e-05,
+      "loss": 4.3415,
+      "step": 906
+    },
+    {
+      "epoch": 0.16886739817742236,
+      "grad_norm": 0.140868678689003,
+      "learning_rate": 1.8485076027782994e-05,
+      "loss": 4.3837,
+      "step": 908
+    },
+    {
+      "epoch": 0.16923935279895852,
+      "grad_norm": 0.1419486254453659,
+      "learning_rate": 1.848132156936362e-05,
+      "loss": 4.3266,
+      "step": 910
+    },
+    {
+      "epoch": 0.1696113074204947,
+      "grad_norm": 0.13442707061767578,
+      "learning_rate": 1.8477567110944247e-05,
+      "loss": 4.3616,
+      "step": 912
+    },
+    {
+      "epoch": 0.16998326204203087,
+      "grad_norm": 0.13400700688362122,
+      "learning_rate": 1.8473812652524874e-05,
+      "loss": 4.3587,
+      "step": 914
+    },
+    {
+      "epoch": 0.17035521666356704,
+      "grad_norm": 0.14199835062026978,
+      "learning_rate": 1.8470058194105504e-05,
+      "loss": 4.3607,
+      "step": 916
+    },
+    {
+      "epoch": 0.17072717128510323,
+      "grad_norm": 0.13668681681156158,
+      "learning_rate": 1.8466303735686127e-05,
+      "loss": 4.3261,
+      "step": 918
+    },
+    {
+      "epoch": 0.1710991259066394,
+      "grad_norm": 0.13259084522724152,
+      "learning_rate": 1.8462549277266754e-05,
+      "loss": 4.3967,
+      "step": 920
+    },
+    {
+      "epoch": 0.17147108052817556,
+      "grad_norm": 0.13595031201839447,
+      "learning_rate": 1.8458794818847384e-05,
+      "loss": 4.3546,
+      "step": 922
+    },
+    {
+      "epoch": 0.17184303514971175,
+      "grad_norm": 0.13679197430610657,
+      "learning_rate": 1.8455040360428008e-05,
+      "loss": 4.3584,
+      "step": 924
+    },
+    {
+      "epoch": 0.1722149897712479,
+      "grad_norm": 0.13515466451644897,
+      "learning_rate": 1.8451285902008638e-05,
+      "loss": 4.3386,
+      "step": 926
+    },
+    {
+      "epoch": 0.17258694439278408,
+      "grad_norm": 0.13040442764759064,
+      "learning_rate": 1.8447531443589265e-05,
+      "loss": 4.3632,
+      "step": 928
+    },
+    {
+      "epoch": 0.17295889901432027,
+      "grad_norm": 0.13664096593856812,
+      "learning_rate": 1.844377698516989e-05,
+      "loss": 4.3245,
+      "step": 930
+    },
+    {
+      "epoch": 0.17333085363585643,
+      "grad_norm": 0.13531778752803802,
+      "learning_rate": 1.8440022526750518e-05,
+      "loss": 4.3332,
+      "step": 932
+    },
+    {
+      "epoch": 0.1737028082573926,
+      "grad_norm": 0.14314793050289154,
+      "learning_rate": 1.8436268068331145e-05,
+      "loss": 4.3801,
+      "step": 934
+    },
+    {
+      "epoch": 0.17407476287892876,
+      "grad_norm": 0.1312064826488495,
+      "learning_rate": 1.843251360991177e-05,
+      "loss": 4.3444,
+      "step": 936
+    },
+    {
+      "epoch": 0.17444671750046495,
+      "grad_norm": 0.132525235414505,
+      "learning_rate": 1.8428759151492398e-05,
+      "loss": 4.3874,
+      "step": 938
+    },
+    {
+      "epoch": 0.1748186721220011,
+      "grad_norm": 0.132554292678833,
+      "learning_rate": 1.8425004693073025e-05,
+      "loss": 4.3402,
+      "step": 940
+    },
+    {
+      "epoch": 0.17519062674353728,
+      "grad_norm": 0.13214528560638428,
+      "learning_rate": 1.8421250234653655e-05,
+      "loss": 4.3761,
+      "step": 942
+    },
+    {
+      "epoch": 0.17556258136507347,
+      "grad_norm": 0.13260036706924438,
+      "learning_rate": 1.841749577623428e-05,
+      "loss": 4.3866,
+      "step": 944
+    },
+    {
+      "epoch": 0.17593453598660963,
+      "grad_norm": 0.13670088350772858,
+      "learning_rate": 1.8413741317814905e-05,
+      "loss": 4.3804,
+      "step": 946
+    },
+    {
+      "epoch": 0.1763064906081458,
+      "grad_norm": 0.13371911644935608,
+      "learning_rate": 1.8409986859395535e-05,
+      "loss": 4.3643,
+      "step": 948
+    },
+    {
+      "epoch": 0.17667844522968199,
+      "grad_norm": 0.13325561583042145,
+      "learning_rate": 1.8406232400976162e-05,
+      "loss": 4.4,
+      "step": 950
+    },
+    {
+      "epoch": 0.17705039985121815,
+      "grad_norm": 0.13221578299999237,
+      "learning_rate": 1.8402477942556785e-05,
+      "loss": 4.3571,
+      "step": 952
+    },
+    {
+      "epoch": 0.1774223544727543,
+      "grad_norm": 0.1316671222448349,
+      "learning_rate": 1.8398723484137416e-05,
+      "loss": 4.3978,
+      "step": 954
+    },
+    {
+      "epoch": 0.1777943090942905,
+      "grad_norm": 0.1374165564775467,
+      "learning_rate": 1.8394969025718042e-05,
+      "loss": 4.3726,
+      "step": 956
+    },
+    {
+      "epoch": 0.17816626371582667,
+      "grad_norm": 0.13154198229312897,
+      "learning_rate": 1.839121456729867e-05,
+      "loss": 4.3231,
+      "step": 958
+    },
+    {
+      "epoch": 0.17853821833736283,
+      "grad_norm": 0.13324327766895294,
+      "learning_rate": 1.8387460108879296e-05,
+      "loss": 4.3587,
+      "step": 960
+    },
+    {
+      "epoch": 0.17891017295889902,
+      "grad_norm": 0.13346631824970245,
+      "learning_rate": 1.8383705650459923e-05,
+      "loss": 4.3598,
+      "step": 962
+    },
+    {
+      "epoch": 0.17928212758043519,
+      "grad_norm": 0.13723908364772797,
+      "learning_rate": 1.837995119204055e-05,
+      "loss": 4.3738,
+      "step": 964
+    },
+    {
+      "epoch": 0.17965408220197135,
+      "grad_norm": 0.13951462507247925,
+      "learning_rate": 1.8376196733621176e-05,
+      "loss": 4.3818,
+      "step": 966
+    },
+    {
+      "epoch": 0.18002603682350754,
+      "grad_norm": 0.13636671006679535,
+      "learning_rate": 1.8372442275201803e-05,
+      "loss": 4.3292,
+      "step": 968
+    },
+    {
+      "epoch": 0.1803979914450437,
+      "grad_norm": 0.14734932780265808,
+      "learning_rate": 1.836868781678243e-05,
+      "loss": 4.3438,
+      "step": 970
+    },
+    {
+      "epoch": 0.18076994606657987,
+      "grad_norm": 0.13459952175617218,
+      "learning_rate": 1.8364933358363056e-05,
+      "loss": 4.3803,
+      "step": 972
+    },
+    {
+      "epoch": 0.18114190068811606,
+      "grad_norm": 0.1226210668683052,
+      "learning_rate": 1.8361178899943683e-05,
+      "loss": 4.3606,
+      "step": 974
+    },
+    {
+      "epoch": 0.18151385530965222,
+      "grad_norm": 0.12903852760791779,
+      "learning_rate": 1.8357424441524313e-05,
+      "loss": 4.3537,
+      "step": 976
+    },
+    {
+      "epoch": 0.18188580993118839,
+      "grad_norm": 0.133047953248024,
+      "learning_rate": 1.8353669983104937e-05,
+      "loss": 4.3541,
+      "step": 978
+    },
+    {
+      "epoch": 0.18225776455272458,
+      "grad_norm": 0.13539519906044006,
+      "learning_rate": 1.8349915524685567e-05,
+      "loss": 4.3654,
+      "step": 980
+    },
+    {
+      "epoch": 0.18262971917426074,
+      "grad_norm": 0.14255937933921814,
+      "learning_rate": 1.8346161066266193e-05,
+      "loss": 4.3222,
+      "step": 982
+    },
+    {
+      "epoch": 0.1830016737957969,
+      "grad_norm": 0.12916487455368042,
+      "learning_rate": 1.834240660784682e-05,
+      "loss": 4.3154,
+      "step": 984
+    },
+    {
+      "epoch": 0.1833736284173331,
+      "grad_norm": 0.13622789084911346,
+      "learning_rate": 1.8338652149427447e-05,
+      "loss": 4.3643,
+      "step": 986
+    },
+    {
+      "epoch": 0.18374558303886926,
+      "grad_norm": 0.13157576322555542,
+      "learning_rate": 1.8334897691008074e-05,
+      "loss": 4.3579,
+      "step": 988
+    },
+    {
+      "epoch": 0.18411753766040542,
+      "grad_norm": 0.12925177812576294,
+      "learning_rate": 1.83311432325887e-05,
+      "loss": 4.381,
+      "step": 990
+    },
+    {
+      "epoch": 0.1844894922819416,
+      "grad_norm": 0.12902005016803741,
+      "learning_rate": 1.8327388774169327e-05,
+      "loss": 4.3486,
+      "step": 992
+    },
+    {
+      "epoch": 0.18486144690347778,
+      "grad_norm": 0.12998996675014496,
+      "learning_rate": 1.8323634315749954e-05,
+      "loss": 4.3611,
+      "step": 994
+    },
+    {
+      "epoch": 0.18523340152501394,
+      "grad_norm": 0.13314658403396606,
+      "learning_rate": 1.8319879857330584e-05,
+      "loss": 4.3792,
+      "step": 996
+    },
+    {
+      "epoch": 0.18560535614655013,
+      "grad_norm": 0.1358042061328888,
+      "learning_rate": 1.8316125398911207e-05,
+      "loss": 4.3937,
+      "step": 998
+    },
+    {
+      "epoch": 0.1859773107680863,
+      "grad_norm": 0.13492144644260406,
+      "learning_rate": 1.8312370940491834e-05,
+      "loss": 4.3734,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 10754,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294afaafe149f9aa6ccbc4dc1b0ce1bc77cf36fac4c1eb6be3924f3ea1b0062
+size 5368

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25b0fca0f2d41a58e98af909542ac12a73c77837d9d4437d07d0c671a9a73d71
+size 6406264496

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105f4663b5fad5c3987dda6e7afa5851943aeadb9dde60562019277b15fcf67
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75dc74b7b4d33947ffc13a8119e1e3d408c30a0d42312a1193ba5e9a6212f08
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a3914ca45a9e6159eb8a9ca0f1fdcaafd61bd288875c2f103ea03b43c7123
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c182c63c0a0988d43d53f2652895f2bd9e70cdb1047fb10c7c033ab97b19624
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433d37b2c38d98436fb22da8657ac2765bd2b641d2e81ff0ce99abde525551de
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbbc1c06c68cb245deeceeebcfb5fc2cbeb555072a700f0f69cf58ec4b1a92f
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-2000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294afaafe149f9aa6ccbc4dc1b0ce1bc77cf36fac4c1eb6be3924f3ea1b0062
+size 5368

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:746f8b946180a4ae643c04c9092aeeddf8b883c99e18c1456feac31c7612539b
+size 6406264496

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105f4663b5fad5c3987dda6e7afa5851943aeadb9dde60562019277b15fcf67
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75dc74b7b4d33947ffc13a8119e1e3d408c30a0d42312a1193ba5e9a6212f08
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a3914ca45a9e6159eb8a9ca0f1fdcaafd61bd288875c2f103ea03b43c7123
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c182c63c0a0988d43d53f2652895f2bd9e70cdb1047fb10c7c033ab97b19624
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433d37b2c38d98436fb22da8657ac2765bd2b641d2e81ff0ce99abde525551de
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbbc1c06c68cb245deeceeebcfb5fc2cbeb555072a700f0f69cf58ec4b1a92f
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-3000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294afaafe149f9aa6ccbc4dc1b0ce1bc77cf36fac4c1eb6be3924f3ea1b0062
+size 5368

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a07523d2849a81e024977bdac9187f02023d4d40bc503f858cabf05f5192138a
+size 6406264496

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105f4663b5fad5c3987dda6e7afa5851943aeadb9dde60562019277b15fcf67
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75dc74b7b4d33947ffc13a8119e1e3d408c30a0d42312a1193ba5e9a6212f08
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a3914ca45a9e6159eb8a9ca0f1fdcaafd61bd288875c2f103ea03b43c7123
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c182c63c0a0988d43d53f2652895f2bd9e70cdb1047fb10c7c033ab97b19624
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433d37b2c38d98436fb22da8657ac2765bd2b641d2e81ff0ce99abde525551de
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbbc1c06c68cb245deeceeebcfb5fc2cbeb555072a700f0f69cf58ec4b1a92f
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-4000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294afaafe149f9aa6ccbc4dc1b0ce1bc77cf36fac4c1eb6be3924f3ea1b0062
+size 5368

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5db3351fa1e9a0e25954bc86d585ba457ccc8ca535d23a8006cac77c3a4917bb
+size 6406264496

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f105f4663b5fad5c3987dda6e7afa5851943aeadb9dde60562019277b15fcf67
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b75dc74b7b4d33947ffc13a8119e1e3d408c30a0d42312a1193ba5e9a6212f08
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a86a3914ca45a9e6159eb8a9ca0f1fdcaafd61bd288875c2f103ea03b43c7123
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c182c63c0a0988d43d53f2652895f2bd9e70cdb1047fb10c7c033ab97b19624
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:433d37b2c38d98436fb22da8657ac2765bd2b641d2e81ff0ce99abde525551de
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bbbc1c06c68cb245deeceeebcfb5fc2cbeb555072a700f0f69cf58ec4b1a92f
+size 15472

dia-tts-malaysian-emilia-full-mixed-precision-podcast/checkpoint-5000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294afaafe149f9aa6ccbc4dc1b0ce1bc77cf36fac4c1eb6be3924f3ea1b0062
+size 5368

dia-tts-malaysian-emilia-full-mixed-precision-podcast/runs/May18_21-57-02_xe9680-cn-4/events.out.tfevents.1747576646.xe9680-cn-4.2678614.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42533ae0a578d42c41378078af8d78498fad459a374848a50e89c4e4e20239d9
+size 19057

dia-tts-malaysian-emilia-full-mixed-precision-podcast/runs/May18_22-17-30_xe9680-cn-4/events.out.tfevents.1747577872.xe9680-cn-4.2691405.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95e11439a6b6dcfe74c5bd0bda98fd82fd841d75334993174c2813a9759875d0
+size 587069