diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6142 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999713113578335,
+  "eval_steps": 500,
+  "global_step": 4357,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.3215898275375366,
+      "learning_rate": 4.587155963302753e-07,
+      "loss": 1.13,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.287673681974411,
+      "learning_rate": 2.2935779816513764e-06,
+      "loss": 1.0912,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.32439985871315,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 1.1711,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.36238783597946167,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 1.1451,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.28779423236846924,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 1.1446,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.27654480934143066,
+      "learning_rate": 1.1467889908256882e-05,
+      "loss": 1.1453,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.254067063331604,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 1.1415,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.22857211530208588,
+      "learning_rate": 1.6055045871559634e-05,
+      "loss": 1.0762,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.2028205841779709,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 1.0592,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.19469808042049408,
+      "learning_rate": 2.0642201834862388e-05,
+      "loss": 1.0423,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.1823880672454834,
+      "learning_rate": 2.2935779816513765e-05,
+      "loss": 1.0494,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.17865613102912903,
+      "learning_rate": 2.5229357798165138e-05,
+      "loss": 1.0494,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.1468791663646698,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 1.0315,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.15089406073093414,
+      "learning_rate": 2.9816513761467892e-05,
+      "loss": 1.0317,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.14719364047050476,
+      "learning_rate": 3.211009174311927e-05,
+      "loss": 1.0156,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1423943191766739,
+      "learning_rate": 3.4403669724770645e-05,
+      "loss": 1.0391,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.15449854731559753,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 1.0071,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.16721290349960327,
+      "learning_rate": 3.89908256880734e-05,
+      "loss": 1.0124,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.16245663166046143,
+      "learning_rate": 4.1284403669724776e-05,
+      "loss": 1.0091,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.16562700271606445,
+      "learning_rate": 4.3577981651376146e-05,
+      "loss": 1.0032,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.17218968272209167,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 1.0231,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.17193633317947388,
+      "learning_rate": 4.81651376146789e-05,
+      "loss": 1.0051,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.1750825196504593,
+      "learning_rate": 5.0458715596330276e-05,
+      "loss": 1.0053,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.17138659954071045,
+      "learning_rate": 5.2752293577981646e-05,
+      "loss": 0.9774,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.1962510198354721,
+      "learning_rate": 5.504587155963303e-05,
+      "loss": 1.0049,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.1920272558927536,
+      "learning_rate": 5.733944954128441e-05,
+      "loss": 0.9838,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2037298083305359,
+      "learning_rate": 5.9633027522935784e-05,
+      "loss": 0.9782,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.17946308851242065,
+      "learning_rate": 6.192660550458716e-05,
+      "loss": 1.0353,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.1685287058353424,
+      "learning_rate": 6.422018348623854e-05,
+      "loss": 1.0017,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.16375750303268433,
+      "learning_rate": 6.651376146788991e-05,
+      "loss": 0.9796,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.1683623045682907,
+      "learning_rate": 6.880733944954129e-05,
+      "loss": 0.9628,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.16674791276454926,
+      "learning_rate": 7.110091743119265e-05,
+      "loss": 0.9916,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.21644377708435059,
+      "learning_rate": 7.339449541284404e-05,
+      "loss": 0.9805,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.16961248219013214,
+      "learning_rate": 7.568807339449542e-05,
+      "loss": 0.9927,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.18196560442447662,
+      "learning_rate": 7.79816513761468e-05,
+      "loss": 0.9916,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.1680087149143219,
+      "learning_rate": 8.027522935779816e-05,
+      "loss": 1.0106,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.16869430243968964,
+      "learning_rate": 8.256880733944955e-05,
+      "loss": 0.9945,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.1714591681957245,
+      "learning_rate": 8.486238532110093e-05,
+      "loss": 0.9832,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.16921715438365936,
+      "learning_rate": 8.715596330275229e-05,
+      "loss": 0.9966,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.15646928548812866,
+      "learning_rate": 8.944954128440367e-05,
+      "loss": 0.9962,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.18113085627555847,
+      "learning_rate": 9.174311926605506e-05,
+      "loss": 1.01,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1637841761112213,
+      "learning_rate": 9.403669724770642e-05,
+      "loss": 0.9882,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1574498414993286,
+      "learning_rate": 9.63302752293578e-05,
+      "loss": 1.0153,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.14823563396930695,
+      "learning_rate": 9.862385321100918e-05,
+      "loss": 0.9807,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.15879377722740173,
+      "learning_rate": 0.00010091743119266055,
+      "loss": 0.9881,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.1585177183151245,
+      "learning_rate": 0.00010321100917431193,
+      "loss": 1.0072,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.15854527056217194,
+      "learning_rate": 0.00010550458715596329,
+      "loss": 0.986,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.15118667483329773,
+      "learning_rate": 0.0001077981651376147,
+      "loss": 1.0132,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1522296965122223,
+      "learning_rate": 0.00011009174311926606,
+      "loss": 0.9555,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.15178298950195312,
+      "learning_rate": 0.00011238532110091744,
+      "loss": 0.9811,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.15078498423099518,
+      "learning_rate": 0.00011467889908256881,
+      "loss": 0.9562,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.15718750655651093,
+      "learning_rate": 0.00011697247706422019,
+      "loss": 0.9807,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1401442289352417,
+      "learning_rate": 0.00011926605504587157,
+      "loss": 0.9782,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.15203869342803955,
+      "learning_rate": 0.00012155963302752293,
+      "loss": 0.9611,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.13769061863422394,
+      "learning_rate": 0.00012385321100917432,
+      "loss": 0.9753,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.16524429619312286,
+      "learning_rate": 0.0001261467889908257,
+      "loss": 0.9485,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.1456153839826584,
+      "learning_rate": 0.00012844036697247707,
+      "loss": 0.9725,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.16664983332157135,
+      "learning_rate": 0.00013073394495412844,
+      "loss": 1.0081,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1406298577785492,
+      "learning_rate": 0.00013302752293577983,
+      "loss": 0.9355,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13675642013549805,
+      "learning_rate": 0.0001353211009174312,
+      "loss": 0.9852,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.14151836931705475,
+      "learning_rate": 0.00013761467889908258,
+      "loss": 0.9922,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13591992855072021,
+      "learning_rate": 0.00013990825688073395,
+      "loss": 0.9792,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1306896060705185,
+      "learning_rate": 0.0001422018348623853,
+      "loss": 0.9558,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13921968638896942,
+      "learning_rate": 0.00014449541284403673,
+      "loss": 0.9618,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.1396915167570114,
+      "learning_rate": 0.0001467889908256881,
+      "loss": 0.9651,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.13790753483772278,
+      "learning_rate": 0.00014908256880733945,
+      "loss": 0.9513,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1389196813106537,
+      "learning_rate": 0.00015137614678899084,
+      "loss": 0.9594,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1367531716823578,
+      "learning_rate": 0.0001536697247706422,
+      "loss": 0.9743,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.15259866416454315,
+      "learning_rate": 0.0001559633027522936,
+      "loss": 0.9662,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.13314354419708252,
+      "learning_rate": 0.00015825688073394496,
+      "loss": 0.9856,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.13281571865081787,
+      "learning_rate": 0.00016055045871559632,
+      "loss": 0.9723,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.14012637734413147,
+      "learning_rate": 0.0001628440366972477,
+      "loss": 0.9839,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.1323760449886322,
+      "learning_rate": 0.0001651376146788991,
+      "loss": 1.0112,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.13485801219940186,
+      "learning_rate": 0.00016743119266055047,
+      "loss": 0.9696,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.13644790649414062,
+      "learning_rate": 0.00016972477064220186,
+      "loss": 0.9963,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.12859384715557098,
+      "learning_rate": 0.00017201834862385322,
+      "loss": 0.9648,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.14412426948547363,
+      "learning_rate": 0.00017431192660550458,
+      "loss": 0.958,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1310245394706726,
+      "learning_rate": 0.00017660550458715597,
+      "loss": 0.9784,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1349959522485733,
+      "learning_rate": 0.00017889908256880734,
+      "loss": 0.9882,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.1282072216272354,
+      "learning_rate": 0.00018119266055045873,
+      "loss": 0.9468,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.13157238066196442,
+      "learning_rate": 0.00018348623853211012,
+      "loss": 0.9695,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.12427080422639847,
+      "learning_rate": 0.00018577981651376148,
+      "loss": 0.9875,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.13389265537261963,
+      "learning_rate": 0.00018807339449541284,
+      "loss": 0.9578,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1387147307395935,
+      "learning_rate": 0.00019036697247706424,
+      "loss": 0.956,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.13312222063541412,
+      "learning_rate": 0.0001926605504587156,
+      "loss": 0.9751,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1331680715084076,
+      "learning_rate": 0.000194954128440367,
+      "loss": 0.9509,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.13201917707920074,
+      "learning_rate": 0.00019724770642201835,
+      "loss": 1.0068,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.12980668246746063,
+      "learning_rate": 0.00019954128440366972,
+      "loss": 0.9301,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.1294248253107071,
+      "learning_rate": 0.00019999948643469536,
+      "loss": 1.0185,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.12491504848003387,
+      "learning_rate": 0.00019999740008468594,
+      "loss": 0.9361,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.12944942712783813,
+      "learning_rate": 0.00019999370888559804,
+      "loss": 0.9744,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.13246968388557434,
+      "learning_rate": 0.0001999884128966714,
+      "loss": 0.9551,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1412133127450943,
+      "learning_rate": 0.00019998151220290082,
+      "loss": 0.9554,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13748963177204132,
+      "learning_rate": 0.00019997300691503497,
+      "loss": 0.9355,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.12698976695537567,
+      "learning_rate": 0.0001999628971695744,
+      "loss": 0.9429,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.12653863430023193,
+      "learning_rate": 0.00019995118312876944,
+      "loss": 0.9791,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13404478132724762,
+      "learning_rate": 0.0001999378649806177,
+      "loss": 0.9782,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.13883401453495026,
+      "learning_rate": 0.00019992294293886095,
+      "loss": 0.9706,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1233060210943222,
+      "learning_rate": 0.00019990641724298156,
+      "loss": 0.9651,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1447945535182953,
+      "learning_rate": 0.000199888288158199,
+      "loss": 0.9604,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.124223992228508,
+      "learning_rate": 0.00019986855597546526,
+      "loss": 0.9659,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12822726368904114,
+      "learning_rate": 0.00019984722101146029,
+      "loss": 0.964,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12251991778612137,
+      "learning_rate": 0.000199824283608587,
+      "loss": 0.9634,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.13198794424533844,
+      "learning_rate": 0.00019979974413496566,
+      "loss": 0.9545,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12475787103176117,
+      "learning_rate": 0.00019977360298442803,
+      "loss": 0.9553,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1245894804596901,
+      "learning_rate": 0.00019974586057651102,
+      "loss": 0.93,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.130756214261055,
+      "learning_rate": 0.00019971651735644995,
+      "loss": 0.9633,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.12664005160331726,
+      "learning_rate": 0.00019968557379517152,
+      "loss": 0.9301,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.13527999818325043,
+      "learning_rate": 0.00019965303038928608,
+      "loss": 0.9713,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1534983515739441,
+      "learning_rate": 0.00019961888766107972,
+      "loss": 0.9723,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12993352115154266,
+      "learning_rate": 0.00019958314615850598,
+      "loss": 0.9876,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13079367578029633,
+      "learning_rate": 0.00019954580645517697,
+      "loss": 0.9455,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13463374972343445,
+      "learning_rate": 0.0001995068691503541,
+      "loss": 0.9511,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12234938144683838,
+      "learning_rate": 0.00019946633486893865,
+      "loss": 0.975,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1227441057562828,
+      "learning_rate": 0.00019942420426146153,
+      "loss": 0.9519,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12618519365787506,
+      "learning_rate": 0.00019938047800407302,
+      "loss": 0.9649,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13038581609725952,
+      "learning_rate": 0.00019933515679853182,
+      "loss": 0.9584,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12579964101314545,
+      "learning_rate": 0.0001992882413721937,
+      "loss": 0.9871,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12814971804618835,
+      "learning_rate": 0.0001992397324780001,
+      "loss": 0.9475,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1240113154053688,
+      "learning_rate": 0.00019918963089446577,
+      "loss": 0.9585,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12410357594490051,
+      "learning_rate": 0.00019913793742566647,
+      "loss": 0.9534,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.13810954988002777,
+      "learning_rate": 0.00019908465290122585,
+      "loss": 0.9766,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12935467064380646,
+      "learning_rate": 0.00019902977817630243,
+      "loss": 0.9728,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12134500592947006,
+      "learning_rate": 0.00019897331413157548,
+      "loss": 0.9781,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1238410472869873,
+      "learning_rate": 0.00019891526167323145,
+      "loss": 1.0207,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12375107407569885,
+      "learning_rate": 0.0001988556217329488,
+      "loss": 0.9621,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12568603456020355,
+      "learning_rate": 0.00019879439526788341,
+      "loss": 0.9711,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1270875632762909,
+      "learning_rate": 0.00019873158326065327,
+      "loss": 0.9635,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12452532351016998,
+      "learning_rate": 0.00019866718671932249,
+      "loss": 0.9727,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12001644819974899,
+      "learning_rate": 0.00019860120667738516,
+      "loss": 0.9613,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12752273678779602,
+      "learning_rate": 0.00019853364419374902,
+      "loss": 0.9745,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12782256305217743,
+      "learning_rate": 0.00019846450035271808,
+      "loss": 0.922,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12796178460121155,
+      "learning_rate": 0.00019839377626397554,
+      "loss": 0.9801,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13660116493701935,
+      "learning_rate": 0.00019832147306256576,
+      "loss": 0.9628,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13185787200927734,
+      "learning_rate": 0.00019824759190887622,
+      "loss": 0.9181,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1231527104973793,
+      "learning_rate": 0.00019817213398861866,
+      "loss": 0.9493,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12743988633155823,
+      "learning_rate": 0.0001980951005128104,
+      "loss": 0.9535,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13102100789546967,
+      "learning_rate": 0.00019801649271775459,
+      "loss": 0.9983,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12197693437337875,
+      "learning_rate": 0.00019793631186502047,
+      "loss": 0.9879,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1295015811920166,
+      "learning_rate": 0.00019785455924142318,
+      "loss": 0.9838,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12491060048341751,
+      "learning_rate": 0.0001977712361590031,
+      "loss": 0.97,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12370824068784714,
+      "learning_rate": 0.00019768634395500465,
+      "loss": 0.9537,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13448503613471985,
+      "learning_rate": 0.00019759988399185505,
+      "loss": 0.9596,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12870986759662628,
+      "learning_rate": 0.00019751185765714234,
+      "loss": 0.9816,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12544594705104828,
+      "learning_rate": 0.00019742226636359296,
+      "loss": 0.9486,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12241456657648087,
+      "learning_rate": 0.00019733111154904943,
+      "loss": 0.93,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12719251215457916,
+      "learning_rate": 0.00019723839467644699,
+      "loss": 0.9478,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12441612035036087,
+      "learning_rate": 0.00019714411723379015,
+      "loss": 0.976,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1346851885318756,
+      "learning_rate": 0.0001970482807341289,
+      "loss": 0.9535,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13904596865177155,
+      "learning_rate": 0.0001969508867155345,
+      "loss": 0.9795,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12592722475528717,
+      "learning_rate": 0.00019685193674107452,
+      "loss": 0.9222,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12337442487478256,
+      "learning_rate": 0.00019675143239878805,
+      "loss": 0.9529,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1278764009475708,
+      "learning_rate": 0.00019664937530166002,
+      "loss": 0.9621,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13516350090503693,
+      "learning_rate": 0.00019654576708759538,
+      "loss": 0.9784,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13280658423900604,
+      "learning_rate": 0.00019644060941939286,
+      "loss": 0.946,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1284746676683426,
+      "learning_rate": 0.00019633390398471817,
+      "loss": 0.9616,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12750181555747986,
+      "learning_rate": 0.00019622565249607704,
+      "loss": 0.9827,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12647484242916107,
+      "learning_rate": 0.0001961158566907877,
+      "loss": 0.945,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.128961980342865,
+      "learning_rate": 0.00019600451833095287,
+      "loss": 0.9678,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12717647850513458,
+      "learning_rate": 0.00019589163920343163,
+      "loss": 0.986,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12336035072803497,
+      "learning_rate": 0.00019577722111981078,
+      "loss": 0.9589,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12848001718521118,
+      "learning_rate": 0.00019566126591637558,
+      "loss": 0.9722,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12276941537857056,
+      "learning_rate": 0.0001955437754540805,
+      "loss": 0.9318,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13224434852600098,
+      "learning_rate": 0.00019542475161851906,
+      "loss": 0.9481,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12429369986057281,
+      "learning_rate": 0.00019530419631989392,
+      "loss": 0.9991,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12761788070201874,
+      "learning_rate": 0.00019518211149298595,
+      "loss": 0.9426,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12386767566204071,
+      "learning_rate": 0.00019505849909712332,
+      "loss": 0.9413,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12389981746673584,
+      "learning_rate": 0.00019493336111615003,
+      "loss": 0.9703,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12416692823171616,
+      "learning_rate": 0.00019480669955839402,
+      "loss": 0.9741,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1296500414609909,
+      "learning_rate": 0.00019467851645663494,
+      "loss": 0.9691,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12780214846134186,
+      "learning_rate": 0.00019454881386807163,
+      "loss": 0.9478,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12381298094987869,
+      "learning_rate": 0.00019441759387428903,
+      "loss": 0.9819,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.13569380342960358,
+      "learning_rate": 0.00019428485858122472,
+      "loss": 0.9741,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12839724123477936,
+      "learning_rate": 0.00019415061011913523,
+      "loss": 0.9431,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12078074365854263,
+      "learning_rate": 0.00019401485064256176,
+      "loss": 0.9267,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12650105357170105,
+      "learning_rate": 0.0001938775823302957,
+      "loss": 1.0044,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12134157121181488,
+      "learning_rate": 0.00019373880738534358,
+      "loss": 0.9933,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12799027562141418,
+      "learning_rate": 0.00019359852803489168,
+      "loss": 0.9849,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.13634993135929108,
+      "learning_rate": 0.0001934567465302704,
+      "loss": 0.9605,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12403295189142227,
+      "learning_rate": 0.00019331346514691813,
+      "loss": 0.9726,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12981389462947845,
+      "learning_rate": 0.00019316868618434455,
+      "loss": 0.9581,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12720562517642975,
+      "learning_rate": 0.00019302241196609397,
+      "loss": 0.945,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1370326429605484,
+      "learning_rate": 0.0001928746448397078,
+      "loss": 0.95,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.131484255194664,
+      "learning_rate": 0.00019272538717668715,
+      "loss": 0.9586,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1213717982172966,
+      "learning_rate": 0.00019257464137245446,
+      "loss": 0.9828,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12702658772468567,
+      "learning_rate": 0.0001924224098463153,
+      "loss": 0.9434,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12242522090673447,
+      "learning_rate": 0.00019226869504141943,
+      "loss": 0.9532,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12427400797605515,
+      "learning_rate": 0.00019211349942472165,
+      "loss": 0.9277,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12835632264614105,
+      "learning_rate": 0.00019195682548694208,
+      "loss": 0.9738,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.15735691785812378,
+      "learning_rate": 0.00019179867574252638,
+      "loss": 0.9489,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12248051166534424,
+      "learning_rate": 0.00019163905272960528,
+      "loss": 0.9659,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1255878508090973,
+      "learning_rate": 0.0001914779590099538,
+      "loss": 0.9617,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12621420621871948,
+      "learning_rate": 0.00019131539716895024,
+      "loss": 0.9856,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.14763972163200378,
+      "learning_rate": 0.00019115136981553464,
+      "loss": 0.9914,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12763573229312897,
+      "learning_rate": 0.00019098587958216688,
+      "loss": 0.9576,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12668545544147491,
+      "learning_rate": 0.00019081892912478456,
+      "loss": 0.9812,
+      "step": 975
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.137617826461792,
+      "learning_rate": 0.00019065052112276018,
+      "loss": 0.9713,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12839971482753754,
+      "learning_rate": 0.00019048065827885827,
+      "loss": 0.9918,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.14924146234989166,
+      "learning_rate": 0.000190309343319192,
+      "loss": 0.9782,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12518009543418884,
+      "learning_rate": 0.00019013657899317942,
+      "loss": 0.9813,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12062101811170578,
+      "learning_rate": 0.0001899623680734993,
+      "loss": 0.9618,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12776224315166473,
+      "learning_rate": 0.00018978671335604665,
+      "loss": 0.9382,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12562039494514465,
+      "learning_rate": 0.00018960961765988792,
+      "loss": 0.9454,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.11983872205018997,
+      "learning_rate": 0.00018943108382721562,
+      "loss": 0.9559,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12643593549728394,
+      "learning_rate": 0.00018925111472330283,
+      "loss": 0.9788,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14982514083385468,
+      "learning_rate": 0.00018906971323645713,
+      "loss": 0.9438,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1314956545829773,
+      "learning_rate": 0.00018888688227797432,
+      "loss": 0.995,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1297961324453354,
+      "learning_rate": 0.00018870262478209163,
+      "loss": 0.9485,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12195736169815063,
+      "learning_rate": 0.00018851694370594069,
+      "loss": 0.9152,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.13816528022289276,
+      "learning_rate": 0.00018832984202949996,
+      "loss": 0.9788,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12382738292217255,
+      "learning_rate": 0.00018814132275554713,
+      "loss": 0.9577,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12794160842895508,
+      "learning_rate": 0.0001879513889096106,
+      "loss": 0.9426,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1255200058221817,
+      "learning_rate": 0.00018776004353992124,
+      "loss": 0.9696,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.14182856678962708,
+      "learning_rate": 0.00018756728971736327,
+      "loss": 0.9682,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13193269073963165,
+      "learning_rate": 0.00018737313053542512,
+      "loss": 0.9929,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12612929940223694,
+      "learning_rate": 0.0001871775691101496,
+      "loss": 0.9644,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12679129838943481,
+      "learning_rate": 0.00018698060858008403,
+      "loss": 0.9704,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12264223396778107,
+      "learning_rate": 0.00018678225210622986,
+      "loss": 0.9281,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12461719661951065,
+      "learning_rate": 0.00018658250287199196,
+      "loss": 0.9644,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13218529522418976,
+      "learning_rate": 0.00018638136408312728,
+      "loss": 0.9352,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1250574141740799,
+      "learning_rate": 0.0001861788389676939,
+      "loss": 0.9733,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1311146467924118,
+      "learning_rate": 0.00018597493077599867,
+      "loss": 0.9555,
+      "step": 1105
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12701354920864105,
+      "learning_rate": 0.00018576964278054544,
+      "loss": 1.0221,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12958282232284546,
+      "learning_rate": 0.00018556297827598242,
+      "loss": 0.9389,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13024652004241943,
+      "learning_rate": 0.00018535494057904915,
+      "loss": 0.9616,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12530991435050964,
+      "learning_rate": 0.00018514553302852356,
+      "loss": 0.9441,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.15070202946662903,
+      "learning_rate": 0.00018493475898516813,
+      "loss": 0.9751,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12876753509044647,
+      "learning_rate": 0.00018472262183167614,
+      "loss": 0.9923,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12746724486351013,
+      "learning_rate": 0.00018450912497261723,
+      "loss": 0.943,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12944746017456055,
+      "learning_rate": 0.00018429427183438288,
+      "loss": 0.9409,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.14462213218212128,
+      "learning_rate": 0.00018407806586513134,
+      "loss": 0.9705,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12713147699832916,
+      "learning_rate": 0.00018386051053473232,
+      "loss": 0.9669,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12527409195899963,
+      "learning_rate": 0.00018364160933471134,
+      "loss": 0.9475,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12756814062595367,
+      "learning_rate": 0.0001834213657781936,
+      "loss": 0.978,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1252342015504837,
+      "learning_rate": 0.00018319978339984767,
+      "loss": 0.9744,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13104258477687836,
+      "learning_rate": 0.0001829768657558288,
+      "loss": 0.9578,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12043920159339905,
+      "learning_rate": 0.00018275261642372175,
+      "loss": 0.9557,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12780900299549103,
+      "learning_rate": 0.0001825270390024834,
+      "loss": 0.9823,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13864734768867493,
+      "learning_rate": 0.00018230013711238513,
+      "loss": 0.9278,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1354120820760727,
+      "learning_rate": 0.00018207191439495438,
+      "loss": 0.9873,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1334461122751236,
+      "learning_rate": 0.00018184237451291665,
+      "loss": 0.9497,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1341424584388733,
+      "learning_rate": 0.00018161152115013637,
+      "loss": 0.9542,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12944547832012177,
+      "learning_rate": 0.00018137935801155794,
+      "loss": 0.9676,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.13224828243255615,
+      "learning_rate": 0.0001811458888231462,
+      "loss": 0.9856,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.13095225393772125,
+      "learning_rate": 0.0001809111173318267,
+      "loss": 0.9331,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.13331545889377594,
+      "learning_rate": 0.00018067504730542551,
+      "loss": 0.9716,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1268896907567978,
+      "learning_rate": 0.0001804376825326088,
+      "loss": 0.9503,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1275816410779953,
+      "learning_rate": 0.00018019902682282193,
+      "loss": 0.9823,
+      "step": 1235
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.127151757478714,
+      "learning_rate": 0.0001799590840062285,
+      "loss": 0.9418,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13317358493804932,
+      "learning_rate": 0.00017971785793364866,
+      "loss": 0.9655,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12957239151000977,
+      "learning_rate": 0.0001794753524764975,
+      "loss": 0.9794,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1317230761051178,
+      "learning_rate": 0.00017923157152672278,
+      "loss": 0.9699,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.137287437915802,
+      "learning_rate": 0.00017898651899674254,
+      "loss": 0.9397,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13417543470859528,
+      "learning_rate": 0.00017874019881938233,
+      "loss": 0.9702,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13501068949699402,
+      "learning_rate": 0.000178492614947812,
+      "loss": 0.9749,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12722639739513397,
+      "learning_rate": 0.00017824377135548236,
+      "loss": 0.9633,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12780694663524628,
+      "learning_rate": 0.00017799367203606128,
+      "loss": 0.9577,
+      "step": 1280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13101287186145782,
+      "learning_rate": 0.00017774232100336982,
+      "loss": 0.9659,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13307473063468933,
+      "learning_rate": 0.00017748972229131757,
+      "loss": 0.9558,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13980835676193237,
+      "learning_rate": 0.000177235879953838,
+      "loss": 0.9632,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.15319649875164032,
+      "learning_rate": 0.00017698079806482343,
+      "loss": 0.9658,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1296548694372177,
+      "learning_rate": 0.0001767244807180597,
+      "loss": 0.9732,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13174237310886383,
+      "learning_rate": 0.00017646693202716033,
+      "loss": 0.9144,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13287286460399628,
+      "learning_rate": 0.0001762081561255005,
+      "loss": 0.9289,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13563677668571472,
+      "learning_rate": 0.00017594815716615093,
+      "loss": 0.9701,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13716502487659454,
+      "learning_rate": 0.000175686939321811,
+      "loss": 0.9563,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12398369610309601,
+      "learning_rate": 0.00017542450678474184,
+      "loss": 0.97,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12686513364315033,
+      "learning_rate": 0.00017516086376669917,
+      "loss": 0.9641,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1259392350912094,
+      "learning_rate": 0.00017489601449886547,
+      "loss": 0.9427,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12787505984306335,
+      "learning_rate": 0.00017462996323178235,
+      "loss": 0.9471,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12886632978916168,
+      "learning_rate": 0.00017436271423528206,
+      "loss": 0.9382,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13796944916248322,
+      "learning_rate": 0.0001740942717984192,
+      "loss": 0.9863,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1262870579957962,
+      "learning_rate": 0.00017382464022940182,
+      "loss": 0.9718,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.15008293092250824,
+      "learning_rate": 0.00017355382385552206,
+      "loss": 0.9665,
+      "step": 1365
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13183139264583588,
+      "learning_rate": 0.0001732818270230871,
+      "loss": 0.9566,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13365882635116577,
+      "learning_rate": 0.000173008654097349,
+      "loss": 0.9446,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.15141069889068604,
+      "learning_rate": 0.000172734309462435,
+      "loss": 0.9607,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1343783438205719,
+      "learning_rate": 0.00017245879752127692,
+      "loss": 0.9667,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1255096197128296,
+      "learning_rate": 0.0001721821226955405,
+      "loss": 0.9842,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1343841850757599,
+      "learning_rate": 0.00017190428942555463,
+      "loss": 0.9488,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12529084086418152,
+      "learning_rate": 0.0001716253021702399,
+      "loss": 0.9698,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13085819780826569,
+      "learning_rate": 0.0001713451654070371,
+      "loss": 0.9665,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12959861755371094,
+      "learning_rate": 0.0001710638836318354,
+      "loss": 0.9964,
+      "step": 1410
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1374460756778717,
+      "learning_rate": 0.00017078146135890014,
+      "loss": 0.9788,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12908467650413513,
+      "learning_rate": 0.0001704979031208004,
+      "loss": 0.9613,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12516631186008453,
+      "learning_rate": 0.0001702132134683363,
+      "loss": 0.9415,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1285628229379654,
+      "learning_rate": 0.00016992739697046586,
+      "loss": 0.9635,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12479186803102493,
+      "learning_rate": 0.00016964045821423178,
+      "loss": 0.9473,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12942568957805634,
+      "learning_rate": 0.00016935240180468775,
+      "loss": 0.9828,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12682420015335083,
+      "learning_rate": 0.00016906323236482465,
+      "loss": 0.917,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12945665419101715,
+      "learning_rate": 0.00016877295453549614,
+      "loss": 0.947,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1369379162788391,
+      "learning_rate": 0.00016848157297534453,
+      "loss": 0.989,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12907510995864868,
+      "learning_rate": 0.0001681890923607256,
+      "loss": 0.9589,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1197751834988594,
+      "learning_rate": 0.00016789551738563384,
+      "loss": 0.9453,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.128217875957489,
+      "learning_rate": 0.00016760085276162708,
+      "loss": 0.9756,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1541936844587326,
+      "learning_rate": 0.00016730510321775075,
+      "loss": 0.9884,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13181723654270172,
+      "learning_rate": 0.00016700827350046206,
+      "loss": 0.9881,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1410752385854721,
+      "learning_rate": 0.00016671036837355386,
+      "loss": 0.9661,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1321258693933487,
+      "learning_rate": 0.00016641139261807818,
+      "loss": 0.9529,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12342269718647003,
+      "learning_rate": 0.00016611135103226937,
+      "loss": 0.9737,
+      "step": 1495
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13780297338962555,
+      "learning_rate": 0.00016581024843146725,
+      "loss": 0.9498,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12811343371868134,
+      "learning_rate": 0.00016550808964803978,
+      "loss": 0.9621,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12910670042037964,
+      "learning_rate": 0.00016520487953130552,
+      "loss": 0.9413,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1280854493379593,
+      "learning_rate": 0.00016490062294745571,
+      "loss": 0.961,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1290293186903,
+      "learning_rate": 0.00016459532477947634,
+      "loss": 0.9071,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12406402081251144,
+      "learning_rate": 0.00016428898992706955,
+      "loss": 0.9653,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12745514512062073,
+      "learning_rate": 0.00016398162330657533,
+      "loss": 0.9446,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12143076211214066,
+      "learning_rate": 0.0001636732298508922,
+      "loss": 0.9292,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12691473960876465,
+      "learning_rate": 0.0001633638145093984,
+      "loss": 0.9785,
+      "step": 1540
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.14538945257663727,
+      "learning_rate": 0.00016305338224787235,
+      "loss": 0.9603,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1339360773563385,
+      "learning_rate": 0.0001627419380484128,
+      "loss": 0.9829,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12584348022937775,
+      "learning_rate": 0.00016242948690935912,
+      "loss": 0.9141,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12381859123706818,
+      "learning_rate": 0.00016211603384521083,
+      "loss": 0.9765,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1408829391002655,
+      "learning_rate": 0.00016180158388654742,
+      "loss": 0.987,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.142933651804924,
+      "learning_rate": 0.00016148614207994735,
+      "loss": 0.9661,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12388478964567184,
+      "learning_rate": 0.00016116971348790712,
+      "loss": 0.9543,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.127301424741745,
+      "learning_rate": 0.0001608523031887601,
+      "loss": 0.95,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13419270515441895,
+      "learning_rate": 0.00016053391627659505,
+      "loss": 0.9256,
+      "step": 1585
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1313311904668808,
+      "learning_rate": 0.0001602145578611742,
+      "loss": 0.9337,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1303502321243286,
+      "learning_rate": 0.00015989423306785142,
+      "loss": 0.9813,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1231377124786377,
+      "learning_rate": 0.00015957294703748982,
+      "loss": 0.9679,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12922777235507965,
+      "learning_rate": 0.00015925070492637944,
+      "loss": 0.9313,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1293102353811264,
+      "learning_rate": 0.0001589275119061542,
+      "loss": 0.9735,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12279027700424194,
+      "learning_rate": 0.00015860337316370916,
+      "loss": 0.9211,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13127371668815613,
+      "learning_rate": 0.0001582782939011173,
+      "loss": 0.9729,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13137449324131012,
+      "learning_rate": 0.00015795227933554568,
+      "loss": 0.9706,
+      "step": 1625
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.127615287899971,
+      "learning_rate": 0.00015762533469917216,
+      "loss": 0.9568,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13084828853607178,
+      "learning_rate": 0.00015729746523910113,
+      "loss": 0.9678,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13544490933418274,
+      "learning_rate": 0.00015696867621727942,
+      "loss": 0.9684,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13566069304943085,
+      "learning_rate": 0.00015663897291041175,
+      "loss": 0.9455,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1366906464099884,
+      "learning_rate": 0.00015630836060987624,
+      "loss": 0.9278,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1368778944015503,
+      "learning_rate": 0.00015597684462163923,
+      "loss": 0.9132,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12608809769153595,
+      "learning_rate": 0.0001556444302661704,
+      "loss": 0.9732,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1579916626214981,
+      "learning_rate": 0.00015531112287835717,
+      "loss": 0.9355,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12954863905906677,
+      "learning_rate": 0.00015497692780741908,
+      "loss": 0.9523,
+      "step": 1670
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13201503455638885,
+      "learning_rate": 0.0001546418504168222,
+      "loss": 0.9702,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12748472392559052,
+      "learning_rate": 0.00015430589608419264,
+      "loss": 0.9128,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12701939046382904,
+      "learning_rate": 0.00015396907020123068,
+      "loss": 0.9864,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1267419010400772,
+      "learning_rate": 0.00015363137817362392,
+      "loss": 0.9913,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12519283592700958,
+      "learning_rate": 0.00015329282542096064,
+      "loss": 0.9631,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.14729751646518707,
+      "learning_rate": 0.00015295341737664285,
+      "loss": 0.9633,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12940853834152222,
+      "learning_rate": 0.000152613159487799,
+      "loss": 0.9634,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1306481808423996,
+      "learning_rate": 0.00015227205721519675,
+      "loss": 0.9781,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1322164535522461,
+      "learning_rate": 0.00015193011603315503,
+      "loss": 0.9646,
+      "step": 1715
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13124747574329376,
+      "learning_rate": 0.00015158734142945644,
+      "loss": 0.9497,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12753304839134216,
+      "learning_rate": 0.0001512437389052591,
+      "loss": 0.959,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12572309374809265,
+      "learning_rate": 0.0001508993139750083,
+      "loss": 0.9349,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13170930743217468,
+      "learning_rate": 0.0001505540721663481,
+      "loss": 0.9745,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.14555582404136658,
+      "learning_rate": 0.0001502080190200325,
+      "loss": 0.964,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13863208889961243,
+      "learning_rate": 0.00014986116008983664,
+      "loss": 0.9549,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12952788174152374,
+      "learning_rate": 0.00014951350094246762,
+      "loss": 0.9378,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13103242218494415,
+      "learning_rate": 0.0001491650471574751,
+      "loss": 0.9607,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13382793962955475,
+      "learning_rate": 0.00014881580432716182,
+      "loss": 0.9295,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12682107090950012,
+      "learning_rate": 0.00014846577805649388,
+      "loss": 0.9358,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1278093308210373,
+      "learning_rate": 0.00014811497396301072,
+      "loss": 0.9637,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12879671156406403,
+      "learning_rate": 0.00014776339767673491,
+      "loss": 0.9454,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12679119408130646,
+      "learning_rate": 0.000147411054840082,
+      "loss": 0.9469,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12752725183963776,
+      "learning_rate": 0.00014705795110776974,
+      "loss": 0.9429,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1287354677915573,
+      "learning_rate": 0.0001467040921467275,
+      "loss": 0.9317,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1301649659872055,
+      "learning_rate": 0.00014634948363600518,
+      "loss": 0.9316,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13030129671096802,
+      "learning_rate": 0.00014599413126668213,
+      "loss": 0.9756,
+      "step": 1800
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.14094071090221405,
+      "learning_rate": 0.00014563804074177588,
+      "loss": 0.9487,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13078372180461884,
+      "learning_rate": 0.00014528121777615058,
+      "loss": 0.9518,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12776345014572144,
+      "learning_rate": 0.0001449236680964251,
+      "loss": 0.9694,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1331811100244522,
+      "learning_rate": 0.00014456539744088138,
+      "loss": 0.9961,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13004426658153534,
+      "learning_rate": 0.00014420641155937224,
+      "loss": 0.9921,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12411683052778244,
+      "learning_rate": 0.00014384671621322915,
+      "loss": 0.9211,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.14314796030521393,
+      "learning_rate": 0.00014348631717516953,
+      "loss": 0.9631,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1315309852361679,
+      "learning_rate": 0.00014312522022920444,
+      "loss": 0.9637,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1483338624238968,
+      "learning_rate": 0.00014276343117054563,
+      "loss": 0.9132,
+      "step": 1845
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13160067796707153,
+      "learning_rate": 0.00014240095580551234,
+      "loss": 0.9487,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13374783098697662,
+      "learning_rate": 0.0001420377999514384,
+      "loss": 0.9295,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13004569709300995,
+      "learning_rate": 0.00014167396943657874,
+      "loss": 0.9471,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13106559216976166,
+      "learning_rate": 0.0001413094701000158,
+      "loss": 0.972,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12326537072658539,
+      "learning_rate": 0.00014094430779156597,
+      "loss": 0.9176,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13554134964942932,
+      "learning_rate": 0.00014057848837168547,
+      "loss": 0.9511,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12836414575576782,
+      "learning_rate": 0.00014021201771137663,
+      "loss": 0.9777,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13130289316177368,
+      "learning_rate": 0.00013984490169209333,
+      "loss": 0.9525,
+      "step": 1885
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1261659413576126,
+      "learning_rate": 0.00013947714620564678,
+      "loss": 0.9591,
+      "step": 1890
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13300645351409912,
+      "learning_rate": 0.00013910875715411098,
+      "loss": 0.9552,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.124820277094841,
+      "learning_rate": 0.000138739740449728,
+      "loss": 0.9798,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1281551569700241,
+      "learning_rate": 0.0001383701020148129,
+      "loss": 0.9403,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13392414152622223,
+      "learning_rate": 0.000137999847781659,
+      "loss": 0.9414,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12957698106765747,
+      "learning_rate": 0.00013762898369244238,
+      "loss": 0.9525,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12338296324014664,
+      "learning_rate": 0.00013725751569912682,
+      "loss": 0.957,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13867761194705963,
+      "learning_rate": 0.00013688544976336783,
+      "loss": 0.9522,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12672200798988342,
+      "learning_rate": 0.00013651279185641752,
+      "loss": 0.9568,
+      "step": 1930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12760479748249054,
+      "learning_rate": 0.0001361395479590283,
+      "loss": 0.9249,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13211704790592194,
+      "learning_rate": 0.00013576572406135722,
+      "loss": 0.9545,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1382143646478653,
+      "learning_rate": 0.00013539132616286956,
+      "loss": 0.975,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.133601576089859,
+      "learning_rate": 0.00013501636027224282,
+      "loss": 0.9263,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12930597364902496,
+      "learning_rate": 0.0001346408324072701,
+      "loss": 0.9574,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1278691589832306,
+      "learning_rate": 0.0001342647485947635,
+      "loss": 0.9598,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12956111133098602,
+      "learning_rate": 0.00013388811487045766,
+      "loss": 0.93,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13355404138565063,
+      "learning_rate": 0.0001335109372789125,
+      "loss": 0.9629,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12528498470783234,
+      "learning_rate": 0.00013313322187341652,
+      "loss": 0.9524,
+      "step": 1975
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1293615698814392,
+      "learning_rate": 0.00013275497471588953,
+      "loss": 0.9645,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12838269770145416,
+      "learning_rate": 0.0001323762018767854,
+      "loss": 0.9566,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13092140853405,
+      "learning_rate": 0.00013199690943499457,
+      "loss": 0.9527,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13540823757648468,
+      "learning_rate": 0.0001316171034777466,
+      "loss": 0.9765,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12425151467323303,
+      "learning_rate": 0.00013123679010051232,
+      "loss": 0.9473,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12867219746112823,
+      "learning_rate": 0.00013085597540690618,
+      "loss": 0.9234,
+      "step": 2005
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1318441480398178,
+      "learning_rate": 0.00013047466550858812,
+      "loss": 0.9203,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13788054883480072,
+      "learning_rate": 0.00013009286652516575,
+      "loss": 1.0037,
+      "step": 2015
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12919679284095764,
+      "learning_rate": 0.00012971058458409576,
+      "loss": 0.9703,
+      "step": 2020
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12766031920909882,
+      "learning_rate": 0.00012932782582058584,
+      "loss": 0.9582,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13528689742088318,
+      "learning_rate": 0.00012894459637749627,
+      "loss": 0.9714,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12611286342144012,
+      "learning_rate": 0.0001285609024052411,
+      "loss": 0.9627,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13146758079528809,
+      "learning_rate": 0.00012817675006168963,
+      "loss": 0.9684,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1245497465133667,
+      "learning_rate": 0.00012779214551206746,
+      "loss": 0.9514,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13034938275814056,
+      "learning_rate": 0.0001274070949288577,
+      "loss": 0.9672,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12926582992076874,
+      "learning_rate": 0.00012702160449170165,
+      "loss": 0.9621,
+      "step": 2055
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13018296658992767,
+      "learning_rate": 0.0001266356803873,
+      "loss": 0.9479,
+      "step": 2060
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12351826578378677,
+      "learning_rate": 0.0001262493288093131,
+      "loss": 0.9696,
+      "step": 2065
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13164471089839935,
+      "learning_rate": 0.000125862555958262,
+      "loss": 0.9566,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12682011723518372,
+      "learning_rate": 0.00012547536804142858,
+      "loss": 0.9476,
+      "step": 2075
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13168583810329437,
+      "learning_rate": 0.0001250877712727561,
+      "loss": 0.9768,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12634898722171783,
+      "learning_rate": 0.00012469977187274945,
+      "loss": 0.9156,
+      "step": 2085
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13255025446414948,
+      "learning_rate": 0.00012431137606837538,
+      "loss": 0.916,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13105438649654388,
+      "learning_rate": 0.0001239225900929624,
+      "loss": 0.9673,
+      "step": 2095
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1347406804561615,
+      "learning_rate": 0.00012353342018610084,
+      "loss": 0.9505,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12653812766075134,
+      "learning_rate": 0.00012314387259354282,
+      "loss": 0.9328,
+      "step": 2105
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1317542940378189,
+      "learning_rate": 0.00012275395356710177,
+      "loss": 0.9495,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.12971192598342896,
+      "learning_rate": 0.0001223636693645523,
+      "loss": 0.9444,
+      "step": 2115
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13270126283168793,
+      "learning_rate": 0.00012197302624952971,
+      "loss": 0.9334,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.12628954648971558,
+      "learning_rate": 0.00012158203049142947,
+      "loss": 0.9732,
+      "step": 2125
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1267569661140442,
+      "learning_rate": 0.00012119068836530644,
+      "loss": 0.943,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13117669522762299,
+      "learning_rate": 0.00012079900615177449,
+      "loss": 0.9751,
+      "step": 2135
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13549381494522095,
+      "learning_rate": 0.00012040699013690543,
+      "loss": 0.9558,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.12888039648532867,
+      "learning_rate": 0.00012001464661212827,
+      "loss": 0.9196,
+      "step": 2145
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13966216146945953,
+      "learning_rate": 0.0001196219818741281,
+      "loss": 0.9387,
+      "step": 2150
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1288440227508545,
+      "learning_rate": 0.00011922900222474523,
+      "loss": 0.9315,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13282476365566254,
+      "learning_rate": 0.00011883571397087387,
+      "loss": 0.9475,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.12561114132404327,
+      "learning_rate": 0.00011844212342436112,
+      "loss": 0.9347,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13055801391601562,
+      "learning_rate": 0.00011804823690190538,
+      "loss": 0.9567,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1304216980934143,
+      "learning_rate": 0.00011765406072495528,
+      "loss": 0.9552,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13724645972251892,
+      "learning_rate": 0.00011725960121960806,
+      "loss": 0.9459,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1331196129322052,
+      "learning_rate": 0.00011686486471650798,
+      "loss": 0.9603,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13081231713294983,
+      "learning_rate": 0.0001164698575507449,
+      "loss": 0.9462,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13047103583812714,
+      "learning_rate": 0.00011607458606175253,
+      "loss": 0.9366,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.12709596753120422,
+      "learning_rate": 0.00011567905659320663,
+      "loss": 0.9535,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1364048272371292,
+      "learning_rate": 0.00011528327549292326,
+      "loss": 0.9205,
+      "step": 2205
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1281905621290207,
+      "learning_rate": 0.00011488724911275694,
+      "loss": 0.936,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13727106153964996,
+      "learning_rate": 0.00011449098380849858,
+      "loss": 1.0005,
+      "step": 2215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.12348493933677673,
+      "learning_rate": 0.00011409448593977363,
+      "loss": 0.9771,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.12625907361507416,
+      "learning_rate": 0.00011369776186993994,
+      "loss": 0.912,
+      "step": 2225
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1306263506412506,
+      "learning_rate": 0.0001133008179659856,
+      "loss": 0.9943,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.12563876807689667,
+      "learning_rate": 0.00011290366059842683,
+      "loss": 0.9227,
+      "step": 2235
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13259495794773102,
+      "learning_rate": 0.00011250629614120571,
+      "loss": 0.9656,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13005706667900085,
+      "learning_rate": 0.00011210873097158786,
+      "loss": 0.9405,
+      "step": 2245
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1286323219537735,
+      "learning_rate": 0.00011171097147006013,
+      "loss": 0.9348,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13030053675174713,
+      "learning_rate": 0.00011131302402022821,
+      "loss": 0.8958,
+      "step": 2255
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1308864951133728,
+      "learning_rate": 0.00011091489500871408,
+      "loss": 0.9692,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13217829167842865,
+      "learning_rate": 0.00011051659082505366,
+      "loss": 0.9931,
+      "step": 2265
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13028296828269958,
+      "learning_rate": 0.00011011811786159416,
+      "loss": 0.9685,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.12667682766914368,
+      "learning_rate": 0.00010971948251339157,
+      "loss": 0.9149,
+      "step": 2275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1268107146024704,
+      "learning_rate": 0.00010932069117810787,
+      "loss": 0.9349,
+      "step": 2280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13369308412075043,
+      "learning_rate": 0.00010892175025590856,
+      "loss": 0.9638,
+      "step": 2285
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1276586800813675,
+      "learning_rate": 0.00010852266614935982,
+      "loss": 0.9797,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12555885314941406,
+      "learning_rate": 0.00010812344526332578,
+      "loss": 0.9271,
+      "step": 2295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12764300405979156,
+      "learning_rate": 0.00010772409400486571,
+      "loss": 0.9972,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13046477735042572,
+      "learning_rate": 0.00010732461878313125,
+      "loss": 0.9784,
+      "step": 2305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14327102899551392,
+      "learning_rate": 0.00010692502600926348,
+      "loss": 0.967,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13605019450187683,
+      "learning_rate": 0.00010652532209629011,
+      "loss": 0.9376,
+      "step": 2315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12832361459732056,
+      "learning_rate": 0.00010612551345902244,
+      "loss": 0.9816,
+      "step": 2320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13134075701236725,
+      "learning_rate": 0.00010572560651395258,
+      "loss": 0.9451,
+      "step": 2325
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12856796383857727,
+      "learning_rate": 0.0001053256076791503,
+      "loss": 0.9523,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13018175959587097,
+      "learning_rate": 0.00010492552337416007,
+      "loss": 0.9421,
+      "step": 2335
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13000188767910004,
+      "learning_rate": 0.00010452536001989815,
+      "loss": 0.9817,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14589110016822815,
+      "learning_rate": 0.00010412512403854942,
+      "loss": 0.9617,
+      "step": 2345
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1261352151632309,
+      "learning_rate": 0.00010372482185346435,
+      "loss": 0.96,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1398177444934845,
+      "learning_rate": 0.00010332445988905586,
+      "loss": 0.9918,
+      "step": 2355
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1412603110074997,
+      "learning_rate": 0.00010292404457069631,
+      "loss": 0.959,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.12706901133060455,
+      "learning_rate": 0.00010252358232461436,
+      "loss": 0.9433,
+      "step": 2365
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.12206675857305527,
+      "learning_rate": 0.00010212307957779173,
+      "loss": 0.9394,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14656612277030945,
+      "learning_rate": 0.00010172254275786017,
+      "loss": 0.9706,
+      "step": 2375
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13196878135204315,
+      "learning_rate": 0.0001013219782929983,
+      "loss": 0.9322,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13169744610786438,
+      "learning_rate": 0.0001009213926118284,
+      "loss": 0.979,
+      "step": 2385
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1299707442522049,
+      "learning_rate": 0.00010052079214331318,
+      "loss": 0.9561,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13158699870109558,
+      "learning_rate": 0.00010012018331665272,
+      "loss": 0.9773,
+      "step": 2395
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13001936674118042,
+      "learning_rate": 9.971957256118129e-05,
+      "loss": 0.9379,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13359415531158447,
+      "learning_rate": 9.931896630626402e-05,
+      "loss": 0.9628,
+      "step": 2405
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.130628764629364,
+      "learning_rate": 9.891837098119389e-05,
+      "loss": 0.9471,
+      "step": 2410
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.12780748307704926,
+      "learning_rate": 9.851779301508842e-05,
+      "loss": 0.9619,
+      "step": 2415
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14909674227237701,
+      "learning_rate": 9.811723883678654e-05,
+      "loss": 0.9465,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12833012640476227,
+      "learning_rate": 9.771671487474546e-05,
+      "loss": 0.9709,
+      "step": 2425
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12729236483573914,
+      "learning_rate": 9.731622755693737e-05,
+      "loss": 0.8963,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12885238230228424,
+      "learning_rate": 9.691578331074643e-05,
+      "loss": 0.9514,
+      "step": 2435
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.13713854551315308,
+      "learning_rate": 9.651538856286551e-05,
+      "loss": 0.9401,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1317557841539383,
+      "learning_rate": 9.611504973919311e-05,
+      "loss": 0.953,
+      "step": 2445
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12958717346191406,
+      "learning_rate": 9.571477326473021e-05,
+      "loss": 0.9694,
+      "step": 2450
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1365780085325241,
+      "learning_rate": 9.531456556347712e-05,
+      "loss": 0.9403,
+      "step": 2455
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12792883813381195,
+      "learning_rate": 9.491443305833043e-05,
+      "loss": 0.9407,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.12714137136936188,
+      "learning_rate": 9.451438217097994e-05,
+      "loss": 0.9335,
+      "step": 2465
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.12885482609272003,
+      "learning_rate": 9.411441932180554e-05,
+      "loss": 0.9826,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1269637644290924,
+      "learning_rate": 9.371455092977423e-05,
+      "loss": 0.943,
+      "step": 2475
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13940972089767456,
+      "learning_rate": 9.331478341233706e-05,
+      "loss": 0.962,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.12256721407175064,
+      "learning_rate": 9.291512318532614e-05,
+      "loss": 0.973,
+      "step": 2485
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13526390492916107,
+      "learning_rate": 9.251557666285174e-05,
+      "loss": 0.9415,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1359473466873169,
+      "learning_rate": 9.211615025719919e-05,
+      "loss": 0.9569,
+      "step": 2495
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1281965672969818,
+      "learning_rate": 9.17168503787262e-05,
+      "loss": 0.9524,
+      "step": 2500
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.12771692872047424,
+      "learning_rate": 9.131768343575979e-05,
+      "loss": 0.9524,
+      "step": 2505
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12928397953510284,
+      "learning_rate": 9.091865583449351e-05,
+      "loss": 0.9445,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13996970653533936,
+      "learning_rate": 9.051977397888464e-05,
+      "loss": 0.9847,
+      "step": 2515
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12119126319885254,
+      "learning_rate": 9.01210442705514e-05,
+      "loss": 0.9226,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1385476142168045,
+      "learning_rate": 8.972247310867027e-05,
+      "loss": 0.9396,
+      "step": 2525
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12217579782009125,
+      "learning_rate": 8.932406688987309e-05,
+      "loss": 0.937,
+      "step": 2530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12890292704105377,
+      "learning_rate": 8.892583200814466e-05,
+      "loss": 0.9368,
+      "step": 2535
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.12779349088668823,
+      "learning_rate": 8.852777485471997e-05,
+      "loss": 0.9438,
+      "step": 2540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13441909849643707,
+      "learning_rate": 8.81299018179817e-05,
+      "loss": 0.9768,
+      "step": 2545
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12224233895540237,
+      "learning_rate": 8.773221928335759e-05,
+      "loss": 0.9318,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12216504663228989,
+      "learning_rate": 8.73347336332181e-05,
+      "loss": 0.9533,
+      "step": 2555
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13171589374542236,
+      "learning_rate": 8.693745124677386e-05,
+      "loss": 0.9629,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12655037641525269,
+      "learning_rate": 8.654037849997342e-05,
+      "loss": 0.9703,
+      "step": 2565
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12626422941684723,
+      "learning_rate": 8.614352176540067e-05,
+      "loss": 0.9176,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12271951884031296,
+      "learning_rate": 8.57468874121729e-05,
+      "loss": 0.9791,
+      "step": 2575
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13416321575641632,
+      "learning_rate": 8.535048180583838e-05,
+      "loss": 0.9543,
+      "step": 2580
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.12512122094631195,
+      "learning_rate": 8.495431130827422e-05,
+      "loss": 0.9629,
+      "step": 2585
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.125463604927063,
+      "learning_rate": 8.455838227758432e-05,
+      "loss": 0.9783,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12435165047645569,
+      "learning_rate": 8.416270106799726e-05,
+      "loss": 0.9486,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13030491769313812,
+      "learning_rate": 8.376727402976447e-05,
+      "loss": 0.8885,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13024476170539856,
+      "learning_rate": 8.3372107509058e-05,
+      "loss": 1.0009,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12996867299079895,
+      "learning_rate": 8.297720784786906e-05,
+      "loss": 0.9507,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12702949345111847,
+      "learning_rate": 8.2582581383906e-05,
+      "loss": 0.9607,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12740886211395264,
+      "learning_rate": 8.218823445049265e-05,
+      "loss": 0.9319,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12772811949253082,
+      "learning_rate": 8.179417337646669e-05,
+      "loss": 0.9434,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1248592659831047,
+      "learning_rate": 8.140040448607804e-05,
+      "loss": 0.9582,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13648490607738495,
+      "learning_rate": 8.100693409888748e-05,
+      "loss": 0.9989,
+      "step": 2635
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13019031286239624,
+      "learning_rate": 8.061376852966495e-05,
+      "loss": 0.9547,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13454581797122955,
+      "learning_rate": 8.02209140882886e-05,
+      "loss": 0.9348,
+      "step": 2645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1266186535358429,
+      "learning_rate": 7.982837707964321e-05,
+      "loss": 0.9288,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1281275749206543,
+      "learning_rate": 7.943616380351913e-05,
+      "loss": 0.9542,
+      "step": 2655
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.12171490490436554,
+      "learning_rate": 7.904428055451118e-05,
+      "loss": 0.9445,
+      "step": 2660
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1287936419248581,
+      "learning_rate": 7.865273362191759e-05,
+      "loss": 0.9268,
+      "step": 2665
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1286572515964508,
+      "learning_rate": 7.826152928963904e-05,
+      "loss": 0.9482,
+      "step": 2670
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13047024607658386,
+      "learning_rate": 7.787067383607796e-05,
+      "loss": 0.9147,
+      "step": 2675
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12373257428407669,
+      "learning_rate": 7.748017353403748e-05,
+      "loss": 0.9487,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.13332262635231018,
+      "learning_rate": 7.70900346506211e-05,
+      "loss": 0.9821,
+      "step": 2685
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.129192054271698,
+      "learning_rate": 7.670026344713189e-05,
+      "loss": 0.9587,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1471257358789444,
+      "learning_rate": 7.631086617897203e-05,
+      "loss": 0.9698,
+      "step": 2695
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15062254667282104,
+      "learning_rate": 7.592184909554245e-05,
+      "loss": 0.9524,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12813173234462738,
+      "learning_rate": 7.553321844014258e-05,
+      "loss": 0.9401,
+      "step": 2705
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.13955838978290558,
+      "learning_rate": 7.514498044987009e-05,
+      "loss": 0.9648,
+      "step": 2710
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1307264268398285,
+      "learning_rate": 7.475714135552074e-05,
+      "loss": 0.9732,
+      "step": 2715
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12836076319217682,
+      "learning_rate": 7.43697073814885e-05,
+      "loss": 0.9547,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12638986110687256,
+      "learning_rate": 7.39826847456656e-05,
+      "loss": 0.9686,
+      "step": 2725
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1295274794101715,
+      "learning_rate": 7.359607965934274e-05,
+      "loss": 0.9335,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14086146652698517,
+      "learning_rate": 7.32098983271094e-05,
+      "loss": 0.9291,
+      "step": 2735
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1275961846113205,
+      "learning_rate": 7.282414694675426e-05,
+      "loss": 0.9479,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12459882348775864,
+      "learning_rate": 7.243883170916574e-05,
+      "loss": 0.9493,
+      "step": 2745
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13032321631908417,
+      "learning_rate": 7.205395879823271e-05,
+      "loss": 0.9763,
+      "step": 2750
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12591035664081573,
+      "learning_rate": 7.166953439074504e-05,
+      "loss": 0.9541,
+      "step": 2755
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13978464901447296,
+      "learning_rate": 7.128556465629475e-05,
+      "loss": 0.9651,
+      "step": 2760
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13441437482833862,
+      "learning_rate": 7.090205575717678e-05,
+      "loss": 0.9561,
+      "step": 2765
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.12430067360401154,
+      "learning_rate": 7.051901384829021e-05,
+      "loss": 0.9432,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.12496016919612885,
+      "learning_rate": 7.013644507703937e-05,
+      "loss": 0.9659,
+      "step": 2775
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1254076212644577,
+      "learning_rate": 6.975435558323532e-05,
+      "loss": 0.9626,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.12512649595737457,
+      "learning_rate": 6.937275149899725e-05,
+      "loss": 0.9804,
+      "step": 2785
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13425563275814056,
+      "learning_rate": 6.899163894865395e-05,
+      "loss": 0.9074,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13320429623126984,
+      "learning_rate": 6.86110240486457e-05,
+      "loss": 0.9121,
+      "step": 2795
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.12586569786071777,
+      "learning_rate": 6.823091290742602e-05,
+      "loss": 0.9652,
+      "step": 2800
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1295434683561325,
+      "learning_rate": 6.785131162536374e-05,
+      "loss": 0.958,
+      "step": 2805
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1287257969379425,
+      "learning_rate": 6.747222629464484e-05,
+      "loss": 0.9409,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13075041770935059,
+      "learning_rate": 6.709366299917497e-05,
+      "loss": 0.9443,
+      "step": 2815
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.12826195359230042,
+      "learning_rate": 6.671562781448166e-05,
+      "loss": 0.954,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1282183825969696,
+      "learning_rate": 6.633812680761684e-05,
+      "loss": 0.9551,
+      "step": 2825
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.12816640734672546,
+      "learning_rate": 6.59611660370594e-05,
+      "loss": 0.9435,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13234581053256989,
+      "learning_rate": 6.558475155261811e-05,
+      "loss": 0.9771,
+      "step": 2835
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1283937692642212,
+      "learning_rate": 6.52088893953344e-05,
+      "loss": 0.9637,
+      "step": 2840
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.12777334451675415,
+      "learning_rate": 6.48335855973855e-05,
+      "loss": 0.9349,
+      "step": 2845
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1314011514186859,
+      "learning_rate": 6.445884618198754e-05,
+      "loss": 0.9501,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1308794468641281,
+      "learning_rate": 6.408467716329894e-05,
+      "loss": 0.9638,
+      "step": 2855
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1292644888162613,
+      "learning_rate": 6.371108454632391e-05,
+      "loss": 0.9661,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13258324563503265,
+      "learning_rate": 6.33380743268159e-05,
+      "loss": 0.9533,
+      "step": 2865
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.12887369096279144,
+      "learning_rate": 6.29656524911817e-05,
+      "loss": 0.9611,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13668963313102722,
+      "learning_rate": 6.259382501638509e-05,
+      "loss": 0.9337,
+      "step": 2875
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1261463314294815,
+      "learning_rate": 6.222259786985101e-05,
+      "loss": 0.9274,
+      "step": 2880
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13323873281478882,
+      "learning_rate": 6.185197700936982e-05,
+      "loss": 0.9613,
+      "step": 2885
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13027900457382202,
+      "learning_rate": 6.14819683830016e-05,
+      "loss": 0.9459,
+      "step": 2890
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13354304432868958,
+      "learning_rate": 6.111257792898082e-05,
+      "loss": 0.9422,
+      "step": 2895
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.12853647768497467,
+      "learning_rate": 6.0743811575620846e-05,
+      "loss": 0.9602,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1265496462583542,
+      "learning_rate": 6.0375675241219e-05,
+      "loss": 0.9651,
+      "step": 2905
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.12821069359779358,
+      "learning_rate": 6.000817483396148e-05,
+      "loss": 0.9282,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.12187941372394562,
+      "learning_rate": 5.96413162518285e-05,
+      "loss": 0.941,
+      "step": 2915
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1419340819120407,
+      "learning_rate": 5.9275105382499694e-05,
+      "loss": 0.9675,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.12858334183692932,
+      "learning_rate": 5.890954810325966e-05,
+      "loss": 0.9482,
+      "step": 2925
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13023917376995087,
+      "learning_rate": 5.854465028090355e-05,
+      "loss": 0.9344,
+      "step": 2930
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13713504374027252,
+      "learning_rate": 5.8180417771643006e-05,
+      "loss": 0.9976,
+      "step": 2935
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13082273304462433,
+      "learning_rate": 5.781685642101196e-05,
+      "loss": 0.9281,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.13440482318401337,
+      "learning_rate": 5.7453972063773184e-05,
+      "loss": 0.9317,
+      "step": 2945
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12886522710323334,
+      "learning_rate": 5.7091770523824317e-05,
+      "loss": 0.9405,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12390702962875366,
+      "learning_rate": 5.673025761410462e-05,
+      "loss": 0.9355,
+      "step": 2955
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12815430760383606,
+      "learning_rate": 5.636943913650147e-05,
+      "loss": 0.9197,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.13874827325344086,
+      "learning_rate": 5.60093208817575e-05,
+      "loss": 0.9578,
+      "step": 2965
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1279187649488449,
+      "learning_rate": 5.564990862937744e-05,
+      "loss": 0.9499,
+      "step": 2970
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12600931525230408,
+      "learning_rate": 5.5291208147535466e-05,
+      "loss": 0.9487,
+      "step": 2975
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.13198550045490265,
+      "learning_rate": 5.4933225192982586e-05,
+      "loss": 0.9385,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12340636551380157,
+      "learning_rate": 5.457596551095441e-05,
+      "loss": 0.9059,
+      "step": 2985
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13031703233718872,
+      "learning_rate": 5.421943483507863e-05,
+      "loss": 0.9433,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12843509018421173,
+      "learning_rate": 5.3863638887283364e-05,
+      "loss": 0.9314,
+      "step": 2995
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1378251612186432,
+      "learning_rate": 5.350858337770498e-05,
+      "loss": 0.9402,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13093078136444092,
+      "learning_rate": 5.315427400459678e-05,
+      "loss": 0.9866,
+      "step": 3005
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13143950700759888,
+      "learning_rate": 5.280071645423726e-05,
+      "loss": 0.9115,
+      "step": 3010
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1258026361465454,
+      "learning_rate": 5.244791640083906e-05,
+      "loss": 0.9427,
+      "step": 3015
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.12516726553440094,
+      "learning_rate": 5.2095879506457736e-05,
+      "loss": 0.9441,
+      "step": 3020
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13248077034950256,
+      "learning_rate": 5.174461142090111e-05,
+      "loss": 0.8966,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12026099115610123,
+      "learning_rate": 5.139411778163832e-05,
+      "loss": 0.9186,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1348072737455368,
+      "learning_rate": 5.104440421370962e-05,
+      "loss": 0.9569,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12726454436779022,
+      "learning_rate": 5.0695476329635825e-05,
+      "loss": 0.9389,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12203843891620636,
+      "learning_rate": 5.034733972932855e-05,
+      "loss": 0.948,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12817218899726868,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.9747,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12777867913246155,
+      "learning_rate": 4.9653462716073594e-05,
+      "loss": 0.946,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12915916740894318,
+      "learning_rate": 4.930773343909434e-05,
+      "loss": 0.9466,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.131636381149292,
+      "learning_rate": 4.8962817717639555e-05,
+      "loss": 0.9619,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.12639881670475006,
+      "learning_rate": 4.8618721087230014e-05,
+      "loss": 0.9611,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1255439966917038,
+      "learning_rate": 4.8275449070240854e-05,
+      "loss": 0.9333,
+      "step": 3075
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1212364211678505,
+      "learning_rate": 4.793300717581308e-05,
+      "loss": 0.9347,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12620878219604492,
+      "learning_rate": 4.7591400899765234e-05,
+      "loss": 0.9247,
+      "step": 3085
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1264711171388626,
+      "learning_rate": 4.7250635724505e-05,
+      "loss": 0.9097,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12817326188087463,
+      "learning_rate": 4.6910717118941286e-05,
+      "loss": 0.95,
+      "step": 3095
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13274335861206055,
+      "learning_rate": 4.6571650538396615e-05,
+      "loss": 0.9784,
+      "step": 3100
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12643088400363922,
+      "learning_rate": 4.6233441424519295e-05,
+      "loss": 0.915,
+      "step": 3105
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12870340049266815,
+      "learning_rate": 4.5896095205196356e-05,
+      "loss": 0.9475,
+      "step": 3110
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12338779866695404,
+      "learning_rate": 4.5559617294466176e-05,
+      "loss": 0.8999,
+      "step": 3115
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12409630417823792,
+      "learning_rate": 4.522401309243193e-05,
+      "loss": 0.9669,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12497595697641373,
+      "learning_rate": 4.488928798517442e-05,
+      "loss": 0.906,
+      "step": 3125
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12633506953716278,
+      "learning_rate": 4.455544734466624e-05,
+      "loss": 0.9126,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12285150587558746,
+      "learning_rate": 4.422249652868506e-05,
+      "loss": 0.9588,
+      "step": 3135
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12945771217346191,
+      "learning_rate": 4.389044088072798e-05,
+      "loss": 0.9364,
+      "step": 3140
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12426690757274628,
+      "learning_rate": 4.355928572992547e-05,
+      "loss": 0.9243,
+      "step": 3145
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.13067524135112762,
+      "learning_rate": 4.322903639095619e-05,
+      "loss": 0.9503,
+      "step": 3150
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.12615260481834412,
+      "learning_rate": 4.289969816396132e-05,
+      "loss": 0.9486,
+      "step": 3155
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.12869814038276672,
+      "learning_rate": 4.2571276334459895e-05,
+      "loss": 0.9568,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1404549479484558,
+      "learning_rate": 4.224377617326353e-05,
+      "loss": 0.9591,
+      "step": 3165
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1280900239944458,
+      "learning_rate": 4.1917202936392265e-05,
+      "loss": 0.914,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.13070254027843475,
+      "learning_rate": 4.15915618649899e-05,
+      "loss": 0.9793,
+      "step": 3175
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.12568062543869019,
+      "learning_rate": 4.126685818524013e-05,
+      "loss": 0.941,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.13167956471443176,
+      "learning_rate": 4.094309710828236e-05,
+      "loss": 0.9277,
+      "step": 3185
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.12797915935516357,
+      "learning_rate": 4.0620283830128414e-05,
+      "loss": 0.8938,
+      "step": 3190
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.13595007359981537,
+      "learning_rate": 4.029842353157888e-05,
+      "loss": 0.949,
+      "step": 3195
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.12872837483882904,
+      "learning_rate": 3.9977521378140084e-05,
+      "loss": 0.9638,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1292254775762558,
+      "learning_rate": 3.965758251994115e-05,
+      "loss": 0.9986,
+      "step": 3205
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.13732695579528809,
+      "learning_rate": 3.933861209165146e-05,
+      "loss": 0.9521,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.13126568496227264,
+      "learning_rate": 3.9020615212398016e-05,
+      "loss": 0.93,
+      "step": 3215
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12328553199768066,
+      "learning_rate": 3.8703596985683556e-05,
+      "loss": 0.9443,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.13504023849964142,
+      "learning_rate": 3.838756249930439e-05,
+      "loss": 0.947,
+      "step": 3225
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12601937353610992,
+      "learning_rate": 3.807251682526902e-05,
+      "loss": 0.9328,
+      "step": 3230
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12721508741378784,
+      "learning_rate": 3.775846501971636e-05,
+      "loss": 0.9474,
+      "step": 3235
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12205570936203003,
+      "learning_rate": 3.7445412122835077e-05,
+      "loss": 0.9165,
+      "step": 3240
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.12871380150318146,
+      "learning_rate": 3.713336315878224e-05,
+      "loss": 0.9594,
+      "step": 3245
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.13344284892082214,
+      "learning_rate": 3.6822323135603054e-05,
+      "loss": 0.9186,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.12685367465019226,
+      "learning_rate": 3.651229704515018e-05,
+      "loss": 0.96,
+      "step": 3255
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.12866058945655823,
+      "learning_rate": 3.6203289863003905e-05,
+      "loss": 0.9266,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.12703388929367065,
+      "learning_rate": 3.5895306548392005e-05,
+      "loss": 0.9507,
+      "step": 3265
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.12836338579654694,
+      "learning_rate": 3.558835204411044e-05,
+      "loss": 0.9457,
+      "step": 3270
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.12500981986522675,
+      "learning_rate": 3.52824312764438e-05,
+      "loss": 0.9483,
+      "step": 3275
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1307361125946045,
+      "learning_rate": 3.497754915508632e-05,
+      "loss": 0.952,
+      "step": 3280
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1289496123790741,
+      "learning_rate": 3.467371057306318e-05,
+      "loss": 0.9533,
+      "step": 3285
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1315229833126068,
+      "learning_rate": 3.437092040665183e-05,
+      "loss": 0.9126,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.13625065982341766,
+      "learning_rate": 3.406918351530376e-05,
+      "loss": 0.9261,
+      "step": 3295
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.13835440576076508,
+      "learning_rate": 3.3768504741566664e-05,
+      "loss": 0.9564,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1260981410741806,
+      "learning_rate": 3.346888891100649e-05,
+      "loss": 0.9283,
+      "step": 3305
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1215910017490387,
+      "learning_rate": 3.3170340832130134e-05,
+      "loss": 0.9301,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1276848316192627,
+      "learning_rate": 3.287286529630832e-05,
+      "loss": 0.9261,
+      "step": 3315
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.11685376614332199,
+      "learning_rate": 3.2576467077698493e-05,
+      "loss": 0.9238,
+      "step": 3320
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.12497583776712418,
+      "learning_rate": 3.228115093316848e-05,
+      "loss": 0.99,
+      "step": 3325
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.12684114277362823,
+      "learning_rate": 3.198692160221987e-05,
+      "loss": 0.9526,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.13080139458179474,
+      "learning_rate": 3.169378380691218e-05,
+      "loss": 0.9411,
+      "step": 3335
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.12943178415298462,
+      "learning_rate": 3.140174225178692e-05,
+      "loss": 0.9641,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1278689205646515,
+      "learning_rate": 3.111080162379215e-05,
+      "loss": 0.9164,
+      "step": 3345
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.14064881205558777,
+      "learning_rate": 3.082096659220722e-05,
+      "loss": 0.9376,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.13648462295532227,
+      "learning_rate": 3.0532241808567966e-05,
+      "loss": 0.9487,
+      "step": 3355
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.13532975316047668,
+      "learning_rate": 3.0244631906591825e-05,
+      "loss": 0.9688,
+      "step": 3360
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1282080113887787,
+      "learning_rate": 2.9958141502103722e-05,
+      "loss": 0.9622,
+      "step": 3365
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.12577365338802338,
+      "learning_rate": 2.9672775192961756e-05,
+      "loss": 0.9407,
+      "step": 3370
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1276453137397766,
+      "learning_rate": 2.938853755898364e-05,
+      "loss": 0.9449,
+      "step": 3375
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1280127912759781,
+      "learning_rate": 2.910543316187301e-05,
+      "loss": 0.9252,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12652656435966492,
+      "learning_rate": 2.882346654514627e-05,
+      "loss": 0.9314,
+      "step": 3385
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12612468004226685,
+      "learning_rate": 2.8542642234059725e-05,
+      "loss": 0.9518,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12408517301082611,
+      "learning_rate": 2.826296473553697e-05,
+      "loss": 0.9313,
+      "step": 3395
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1247304305434227,
+      "learning_rate": 2.7984438538096392e-05,
+      "loss": 0.909,
+      "step": 3400
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12919586896896362,
+      "learning_rate": 2.7707068111779377e-05,
+      "loss": 0.9589,
+      "step": 3405
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13544316589832306,
+      "learning_rate": 2.7430857908078345e-05,
+      "loss": 0.9415,
+      "step": 3410
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.12835930287837982,
+      "learning_rate": 2.7155812359865517e-05,
+      "loss": 0.9334,
+      "step": 3415
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1330425888299942,
+      "learning_rate": 2.6881935881321563e-05,
+      "loss": 0.9921,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13000650703907013,
+      "learning_rate": 2.6609232867864896e-05,
+      "loss": 0.9392,
+      "step": 3425
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.11919603496789932,
+      "learning_rate": 2.6337707696081094e-05,
+      "loss": 0.9161,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.12460532784461975,
+      "learning_rate": 2.606736472365272e-05,
+      "loss": 0.931,
+      "step": 3435
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13255023956298828,
+      "learning_rate": 2.5798208289289204e-05,
+      "loss": 0.9396,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13449080288410187,
+      "learning_rate": 2.5530242712657492e-05,
+      "loss": 0.9563,
+      "step": 3445
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.12719547748565674,
+      "learning_rate": 2.526347229431242e-05,
+      "loss": 0.979,
+      "step": 3450
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13245166838169098,
+      "learning_rate": 2.499790131562797e-05,
+      "loss": 0.9698,
+      "step": 3455
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1335286647081375,
+      "learning_rate": 2.4733534038728257e-05,
+      "loss": 0.9413,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1279202252626419,
+      "learning_rate": 2.4470374706419485e-05,
+      "loss": 0.9579,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12700194120407104,
+      "learning_rate": 2.4208427542121504e-05,
+      "loss": 0.9458,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.13555917143821716,
+      "learning_rate": 2.394769674980035e-05,
+      "loss": 0.948,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.13046029210090637,
+      "learning_rate": 2.3688186513900455e-05,
+      "loss": 0.9529,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12478239834308624,
+      "learning_rate": 2.34299009992778e-05,
+      "loss": 0.9384,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12927207350730896,
+      "learning_rate": 2.317284435113278e-05,
+      "loss": 0.9683,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1260586529970169,
+      "learning_rate": 2.2917020694944023e-05,
+      "loss": 0.9227,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.12429028004407883,
+      "learning_rate": 2.2662434136401722e-05,
+      "loss": 0.929,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1263958215713501,
+      "learning_rate": 2.2409088761342235e-05,
+      "loss": 0.9532,
+      "step": 3505
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12389031797647476,
+      "learning_rate": 2.215698863568213e-05,
+      "loss": 0.9447,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12953083217144012,
+      "learning_rate": 2.1906137805353212e-05,
+      "loss": 0.9569,
+      "step": 3515
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1324577033519745,
+      "learning_rate": 2.1656540296237316e-05,
+      "loss": 0.9382,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12997481226921082,
+      "learning_rate": 2.1408200114101985e-05,
+      "loss": 0.9382,
+      "step": 3525
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.13136348128318787,
+      "learning_rate": 2.116112124453592e-05,
+      "loss": 0.9658,
+      "step": 3530
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12820036709308624,
+      "learning_rate": 2.0915307652885164e-05,
+      "loss": 0.927,
+      "step": 3535
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.13575664162635803,
+      "learning_rate": 2.067076328418949e-05,
+      "loss": 0.9297,
+      "step": 3540
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12840761244297028,
+      "learning_rate": 2.0427492063118935e-05,
+      "loss": 0.9609,
+      "step": 3545
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.12105870246887207,
+      "learning_rate": 2.018549789391102e-05,
+      "loss": 0.9455,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.13013601303100586,
+      "learning_rate": 1.994478466030787e-05,
+      "loss": 0.9279,
+      "step": 3555
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.13018429279327393,
+      "learning_rate": 1.970535622549401e-05,
+      "loss": 0.8892,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.13495218753814697,
+      "learning_rate": 1.946721643203443e-05,
+      "loss": 0.9426,
+      "step": 3565
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.13938775658607483,
+      "learning_rate": 1.923036910181275e-05,
+      "loss": 0.9811,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12356399744749069,
+      "learning_rate": 1.8994818035969975e-05,
+      "loss": 0.9452,
+      "step": 3575
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12685617804527283,
+      "learning_rate": 1.8760567014843545e-05,
+      "loss": 0.9323,
+      "step": 3580
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.12429524958133698,
+      "learning_rate": 1.8527619797906494e-05,
+      "loss": 0.9221,
+      "step": 3585
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1301840841770172,
+      "learning_rate": 1.8295980123707357e-05,
+      "loss": 0.9549,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1298561841249466,
+      "learning_rate": 1.8065651709809905e-05,
+      "loss": 0.9414,
+      "step": 3595
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1277400255203247,
+      "learning_rate": 1.783663825273372e-05,
+      "loss": 0.9421,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12424223124980927,
+      "learning_rate": 1.7608943427894686e-05,
+      "loss": 0.9324,
+      "step": 3605
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12309490144252777,
+      "learning_rate": 1.7382570889546124e-05,
+      "loss": 0.9295,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12646757066249847,
+      "learning_rate": 1.7157524270720036e-05,
+      "loss": 0.9224,
+      "step": 3615
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12803521752357483,
+      "learning_rate": 1.6933807183168994e-05,
+      "loss": 0.9822,
+      "step": 3620
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12189158797264099,
+      "learning_rate": 1.6711423217307885e-05,
+      "loss": 0.9606,
+      "step": 3625
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12300534546375275,
+      "learning_rate": 1.64903759421566e-05,
+      "loss": 0.9426,
+      "step": 3630
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.12959307432174683,
+      "learning_rate": 1.627066890528247e-05,
+      "loss": 0.9251,
+      "step": 3635
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1331903040409088,
+      "learning_rate": 1.6052305632743592e-05,
+      "loss": 0.9666,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.12735921144485474,
+      "learning_rate": 1.583528962903197e-05,
+      "loss": 0.9214,
+      "step": 3645
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.13565155863761902,
+      "learning_rate": 1.5619624377017537e-05,
+      "loss": 0.9456,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.13129490613937378,
+      "learning_rate": 1.540531333789207e-05,
+      "loss": 0.9639,
+      "step": 3655
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.12856322526931763,
+      "learning_rate": 1.5192359951113755e-05,
+      "loss": 0.9212,
+      "step": 3660
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.12934236228466034,
+      "learning_rate": 1.4980767634351877e-05,
+      "loss": 0.9577,
+      "step": 3665
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.12420625239610672,
+      "learning_rate": 1.4770539783432113e-05,
+      "loss": 0.9268,
+      "step": 3670
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1385476291179657,
+      "learning_rate": 1.4561679772281877e-05,
+      "loss": 0.9372,
+      "step": 3675
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1316072940826416,
+      "learning_rate": 1.4354190952876334e-05,
+      "loss": 0.9541,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1349179893732071,
+      "learning_rate": 1.4148076655184373e-05,
+      "loss": 0.9286,
+      "step": 3685
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12385948747396469,
+      "learning_rate": 1.3943340187115494e-05,
+      "loss": 0.9141,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12740720808506012,
+      "learning_rate": 1.373998483446638e-05,
+      "loss": 0.9472,
+      "step": 3695
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12889592349529266,
+      "learning_rate": 1.3538013860868436e-05,
+      "loss": 0.9319,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.13519205152988434,
+      "learning_rate": 1.3337430507735205e-05,
+      "loss": 0.9521,
+      "step": 3705
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.12681438028812408,
+      "learning_rate": 1.313823799421051e-05,
+      "loss": 0.9559,
+      "step": 3710
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.13137029111385345,
+      "learning_rate": 1.2940439517116676e-05,
+      "loss": 0.926,
+      "step": 3715
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1269741803407669,
+      "learning_rate": 1.2744038250903267e-05,
+      "loss": 0.9795,
+      "step": 3720
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1284601241350174,
+      "learning_rate": 1.2549037347596115e-05,
+      "loss": 0.9277,
+      "step": 3725
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.12921680510044098,
+      "learning_rate": 1.2355439936746827e-05,
+      "loss": 0.9528,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.13634294271469116,
+      "learning_rate": 1.2163249125382426e-05,
+      "loss": 0.9505,
+      "step": 3735
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.12901267409324646,
+      "learning_rate": 1.1972467997955595e-05,
+      "loss": 0.9421,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.12938185036182404,
+      "learning_rate": 1.1783099616295056e-05,
+      "loss": 0.9163,
+      "step": 3745
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.12726463377475739,
+      "learning_rate": 1.1595147019556607e-05,
+      "loss": 0.9362,
+      "step": 3750
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1396276354789734,
+      "learning_rate": 1.140861322417417e-05,
+      "loss": 0.9136,
+      "step": 3755
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.13083621859550476,
+      "learning_rate": 1.1223501223811451e-05,
+      "loss": 0.9711,
+      "step": 3760
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1297810822725296,
+      "learning_rate": 1.1039813989313951e-05,
+      "loss": 0.9431,
+      "step": 3765
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1269226372241974,
+      "learning_rate": 1.085755446866119e-05,
+      "loss": 0.9394,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1303691416978836,
+      "learning_rate": 1.0676725586919457e-05,
+      "loss": 0.9404,
+      "step": 3775
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12751063704490662,
+      "learning_rate": 1.0497330246194848e-05,
+      "loss": 0.9302,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1254378855228424,
+      "learning_rate": 1.031937132558668e-05,
+      "loss": 0.9261,
+      "step": 3785
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1306389421224594,
+      "learning_rate": 1.014285168114133e-05,
+      "loss": 0.9094,
+      "step": 3790
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12737561762332916,
+      "learning_rate": 9.96777414580633e-06,
+      "loss": 0.9409,
+      "step": 3795
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1371740698814392,
+      "learning_rate": 9.794141529384915e-06,
+      "loss": 0.9636,
+      "step": 3800
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12934444844722748,
+      "learning_rate": 9.621956618491024e-06,
+      "loss": 0.9872,
+      "step": 3805
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.12252593040466309,
+      "learning_rate": 9.451222176504414e-06,
+      "loss": 0.9168,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12601937353610992,
+      "learning_rate": 9.281940943526491e-06,
+      "loss": 0.9354,
+      "step": 3815
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.13008485734462738,
+      "learning_rate": 9.114115636336152e-06,
+      "loss": 0.9214,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1278260201215744,
+      "learning_rate": 8.947748948346357e-06,
+      "loss": 0.9354,
+      "step": 3825
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12461239099502563,
+      "learning_rate": 8.782843549560771e-06,
+      "loss": 0.9272,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1232757493853569,
+      "learning_rate": 8.61940208653097e-06,
+      "loss": 0.9259,
+      "step": 3835
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12535670399665833,
+      "learning_rate": 8.457427182313937e-06,
+      "loss": 0.9669,
+      "step": 3840
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.13315121829509735,
+      "learning_rate": 8.296921436430071e-06,
+      "loss": 0.9486,
+      "step": 3845
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12598228454589844,
+      "learning_rate": 8.137887424821277e-06,
+      "loss": 0.9627,
+      "step": 3850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.12394389510154724,
+      "learning_rate": 7.980327699809832e-06,
+      "loss": 0.9579,
+      "step": 3855
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1374145895242691,
+      "learning_rate": 7.824244790057223e-06,
+      "loss": 0.978,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.13007713854312897,
+      "learning_rate": 7.66964120052377e-06,
+      "loss": 0.9271,
+      "step": 3865
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.12833282351493835,
+      "learning_rate": 7.516519412428203e-06,
+      "loss": 0.9547,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.12410393357276917,
+      "learning_rate": 7.3648818832080745e-06,
+      "loss": 0.9518,
+      "step": 3875
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1256374567747116,
+      "learning_rate": 7.214731046480094e-06,
+      "loss": 0.944,
+      "step": 3880
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.12737049162387848,
+      "learning_rate": 7.066069312001289e-06,
+      "loss": 0.9339,
+      "step": 3885
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1293206810951233,
+      "learning_rate": 6.918899065630113e-06,
+      "loss": 0.933,
+      "step": 3890
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.12668545544147491,
+      "learning_rate": 6.773222669288359e-06,
+      "loss": 0.9464,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.12287843972444534,
+      "learning_rate": 6.629042460923096e-06,
+      "loss": 0.9365,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1297907680273056,
+      "learning_rate": 6.486360754469234e-06,
+      "loss": 0.9396,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13317760825157166,
+      "learning_rate": 6.345179839812343e-06,
+      "loss": 0.9352,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.12976738810539246,
+      "learning_rate": 6.205501982751971e-06,
+      "loss": 0.9679,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.12888483703136444,
+      "learning_rate": 6.067329424965162e-06,
+      "loss": 0.9222,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.12933102250099182,
+      "learning_rate": 5.930664383970641e-06,
+      "loss": 0.9749,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13075292110443115,
+      "learning_rate": 5.795509053093029e-06,
+      "loss": 0.943,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1283501237630844,
+      "learning_rate": 5.6618656014278406e-06,
+      "loss": 0.9358,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13466443121433258,
+      "learning_rate": 5.5297361738065325e-06,
+      "loss": 0.954,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1338685303926468,
+      "learning_rate": 5.399122890762143e-06,
+      "loss": 0.9317,
+      "step": 3945
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12677209079265594,
+      "learning_rate": 5.270027848495207e-06,
+      "loss": 0.9487,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12455341219902039,
+      "learning_rate": 5.1424531188402405e-06,
+      "loss": 0.9772,
+      "step": 3955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12164168804883957,
+      "learning_rate": 5.016400749232297e-06,
+      "loss": 0.9512,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12740525603294373,
+      "learning_rate": 4.89187276267431e-06,
+      "loss": 0.9445,
+      "step": 3965
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12866896390914917,
+      "learning_rate": 4.7688711577044354e-06,
+      "loss": 0.9574,
+      "step": 3970
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12449478358030319,
+      "learning_rate": 4.647397908364182e-06,
+      "loss": 0.9412,
+      "step": 3975
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.12847180664539337,
+      "learning_rate": 4.5274549641665105e-06,
+      "loss": 0.958,
+      "step": 3980
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.13862472772598267,
+      "learning_rate": 4.40904425006472e-06,
+      "loss": 0.9125,
+      "step": 3985
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1288689523935318,
+      "learning_rate": 4.2921676664214535e-06,
+      "loss": 0.9262,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1387001872062683,
+      "learning_rate": 4.176827088978297e-06,
+      "loss": 0.9519,
+      "step": 3995
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.126756489276886,
+      "learning_rate": 4.0630243688255185e-06,
+      "loss": 0.9626,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.12510916590690613,
+      "learning_rate": 3.950761332372543e-06,
+      "loss": 0.9687,
+      "step": 4005
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.12398547679185867,
+      "learning_rate": 3.8400397813185054e-06,
+      "loss": 0.9472,
+      "step": 4010
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.12644776701927185,
+      "learning_rate": 3.7308614926234165e-06,
+      "loss": 0.9361,
+      "step": 4015
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.11997824162244797,
+      "learning_rate": 3.6232282184795794e-06,
+      "loss": 0.913,
+      "step": 4020
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.12514357268810272,
+      "learning_rate": 3.517141686283498e-06,
+      "loss": 0.9873,
+      "step": 4025
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.12552134692668915,
+      "learning_rate": 3.412603598608188e-06,
+      "loss": 0.9346,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.12378017604351044,
+      "learning_rate": 3.3096156331758e-06,
+      "loss": 0.9771,
+      "step": 4035
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.12916335463523865,
+      "learning_rate": 3.2081794428307278e-06,
+      "loss": 0.9136,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1269010454416275,
+      "learning_rate": 3.1082966555130654e-06,
+      "loss": 0.934,
+      "step": 4045
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.12663115561008453,
+      "learning_rate": 3.0099688742324715e-06,
+      "loss": 0.9525,
+      "step": 4050
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.13114877045154572,
+      "learning_rate": 2.913197677042456e-06,
+      "loss": 0.9158,
+      "step": 4055
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.13205552101135254,
+      "learning_rate": 2.8179846170150903e-06,
+      "loss": 0.9717,
+      "step": 4060
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1268589347600937,
+      "learning_rate": 2.7243312222159924e-06,
+      "loss": 0.9592,
+      "step": 4065
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1290740966796875,
+      "learning_rate": 2.6322389956799143e-06,
+      "loss": 0.9339,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12721258401870728,
+      "learning_rate": 2.541709415386495e-06,
+      "loss": 0.9651,
+      "step": 4075
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.13566437363624573,
+      "learning_rate": 2.4527439342366785e-06,
+      "loss": 0.9476,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12669555842876434,
+      "learning_rate": 2.3653439800292556e-06,
+      "loss": 0.9584,
+      "step": 4085
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12517812848091125,
+      "learning_rate": 2.2795109554381024e-06,
+      "loss": 0.9109,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12395232915878296,
+      "learning_rate": 2.195246237989479e-06,
+      "loss": 0.9318,
+      "step": 4095
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1283487230539322,
+      "learning_rate": 2.1125511800401234e-06,
+      "loss": 0.9392,
+      "step": 4100
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12367790192365646,
+      "learning_rate": 2.0314271087554126e-06,
+      "loss": 0.9216,
+      "step": 4105
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12328365445137024,
+      "learning_rate": 1.951875326088104e-06,
+      "loss": 0.94,
+      "step": 4110
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.12500682473182678,
+      "learning_rate": 1.8738971087574275e-06,
+      "loss": 0.9206,
+      "step": 4115
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.13211338222026825,
+      "learning_rate": 1.797493708228659e-06,
+      "loss": 0.9582,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.13043932616710663,
+      "learning_rate": 1.7226663506929142e-06,
+      "loss": 0.9842,
+      "step": 4125
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.12930817902088165,
+      "learning_rate": 1.6494162370475852e-06,
+      "loss": 0.965,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1328224539756775,
+      "learning_rate": 1.5777445428770022e-06,
+      "loss": 0.9102,
+      "step": 4135
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14062470197677612,
+      "learning_rate": 1.5076524184336027e-06,
+      "loss": 0.9142,
+      "step": 4140
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.12664052844047546,
+      "learning_rate": 1.4391409886194474e-06,
+      "loss": 0.9458,
+      "step": 4145
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.12660756707191467,
+      "learning_rate": 1.3722113529681668e-06,
+      "loss": 0.937,
+      "step": 4150
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1241002082824707,
+      "learning_rate": 1.306864585627332e-06,
+      "loss": 0.9567,
+      "step": 4155
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.12647829949855804,
+      "learning_rate": 1.2431017353412233e-06,
+      "loss": 0.9531,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.11990875005722046,
+      "learning_rate": 1.1809238254339105e-06,
+      "loss": 0.9209,
+      "step": 4165
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.12788672745227814,
+      "learning_rate": 1.1203318537929996e-06,
+      "loss": 0.9554,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.13351422548294067,
+      "learning_rate": 1.0613267928534453e-06,
+      "loss": 0.9684,
+      "step": 4175
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1278548538684845,
+      "learning_rate": 1.0039095895820639e-06,
+      "loss": 0.9765,
+      "step": 4180
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1261758953332901,
+      "learning_rate": 9.480811654622557e-07,
+      "loss": 0.9328,
+      "step": 4185
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1272057145833969,
+      "learning_rate": 8.938424164792736e-07,
+      "loss": 0.9002,
+      "step": 4190
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.12137813121080399,
+      "learning_rate": 8.411942131058115e-07,
+      "loss": 0.935,
+      "step": 4195
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.128991961479187,
+      "learning_rate": 7.90137400288049e-07,
+      "loss": 0.9596,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13175086677074432,
+      "learning_rate": 7.406727974320627e-07,
+      "loss": 0.9432,
+      "step": 4205
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13171623647212982,
+      "learning_rate": 6.928011983907245e-07,
+      "loss": 0.925,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13396240770816803,
+      "learning_rate": 6.465233714509245e-07,
+      "loss": 0.9722,
+      "step": 4215
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.12811309099197388,
+      "learning_rate": 6.01840059321257e-07,
+      "loss": 0.9378,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.12516991794109344,
+      "learning_rate": 5.587519791200869e-07,
+      "loss": 0.9292,
+      "step": 4225
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.13231581449508667,
+      "learning_rate": 5.172598223640468e-07,
+      "loss": 0.9458,
+      "step": 4230
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.126018688082695,
+      "learning_rate": 4.773642549569579e-07,
+      "loss": 0.9302,
+      "step": 4235
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.12554921209812164,
+      "learning_rate": 4.390659171790934e-07,
+      "loss": 0.9,
+      "step": 4240
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.12394853681325912,
+      "learning_rate": 4.023654236769647e-07,
+      "loss": 0.9385,
+      "step": 4245
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12245740741491318,
+      "learning_rate": 3.672633634534295e-07,
+      "loss": 0.9453,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1226479634642601,
+      "learning_rate": 3.3376029985819903e-07,
+      "loss": 0.9563,
+      "step": 4255
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1348881870508194,
+      "learning_rate": 3.0185677057887885e-07,
+      "loss": 0.9995,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12937822937965393,
+      "learning_rate": 2.715532876322646e-07,
+      "loss": 0.9252,
+      "step": 4265
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12891468405723572,
+      "learning_rate": 2.428503373561708e-07,
+      "loss": 0.9411,
+      "step": 4270
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12466026842594147,
+      "learning_rate": 2.1574838040161473e-07,
+      "loss": 0.9325,
+      "step": 4275
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12525266408920288,
+      "learning_rate": 1.9024785172541136e-07,
+      "loss": 0.9509,
+      "step": 4280
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.12894925475120544,
+      "learning_rate": 1.6634916058319018e-07,
+      "loss": 0.9448,
+      "step": 4285
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.126129150390625,
+      "learning_rate": 1.4405269052284455e-07,
+      "loss": 0.9721,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.13286429643630981,
+      "learning_rate": 1.2335879937839246e-07,
+      "loss": 0.9722,
+      "step": 4295
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1288774460554123,
+      "learning_rate": 1.0426781926416996e-07,
+      "loss": 0.9527,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12878000736236572,
+      "learning_rate": 8.678005656957977e-08,
+      "loss": 0.9344,
+      "step": 4305
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12277129292488098,
+      "learning_rate": 7.089579195409534e-08,
+      "loss": 0.9675,
+      "step": 4310
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1256117969751358,
+      "learning_rate": 5.6615280342842135e-08,
+      "loss": 0.92,
+      "step": 4315
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12705624103546143,
+      "learning_rate": 4.3938750922412064e-08,
+      "loss": 0.9532,
+      "step": 4320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12469390034675598,
+      "learning_rate": 3.286640713727751e-08,
+      "loss": 0.9212,
+      "step": 4325
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12652264535427094,
+      "learning_rate": 2.3398426686471743e-08,
+      "loss": 0.9675,
+      "step": 4330
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.12550239264965057,
+      "learning_rate": 1.5534961520724533e-08,
+      "loss": 0.9402,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.12904945015907288,
+      "learning_rate": 9.276137840075194e-09,
+      "loss": 0.954,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1253081113100052,
+      "learning_rate": 4.6220560918075474e-09,
+      "loss": 0.946,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.13169953227043152,
+      "learning_rate": 1.572790968851212e-09,
+      "loss": 1.0039,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.12590309977531433,
+      "learning_rate": 1.2839140858256215e-10,
+      "loss": 0.9491,
+      "step": 4355
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.9484747648239136,
+      "eval_runtime": 3746.9119,
+      "eval_samples_per_second": 4.118,
+      "eval_steps_per_second": 0.515,
+      "step": 4357
+    },
+    {
+      "epoch": 1.0,
+      "step": 4357,
+      "total_flos": 1.2254141370096157e+19,
+      "train_loss": 0.9568730254279437,
+      "train_runtime": 82594.0234,
+      "train_samples_per_second": 1.688,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4357,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 1.2254141370096157e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}