diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,42 +3,10892 @@
   "best_model_checkpoint": null,
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 4,
+  "global_step": 7754,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.25,
-      "grad_norm": 0.5412713558869501,
-      "learning_rate": 3e-06,
-      "loss": 1.0726,
+      "epoch": 0.0,
+      "grad_norm": 0.735379659398005,
+      "learning_rate": 1.2886597938144328e-09,
+      "loss": 1.2789,
       "step": 1
     },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5235345835516636,
+      "learning_rate": 6.443298969072164e-09,
+      "loss": 1.192,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5442934452032187,
+      "learning_rate": 1.2886597938144328e-08,
+      "loss": 1.1924,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5884190600397569,
+      "learning_rate": 1.9329896907216496e-08,
+      "loss": 1.2645,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.3701312199921657,
+      "learning_rate": 2.5773195876288656e-08,
+      "loss": 1.2091,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.589331664948267,
+      "learning_rate": 3.221649484536082e-08,
+      "loss": 1.2322,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5885994565663679,
+      "learning_rate": 3.865979381443299e-08,
+      "loss": 1.2532,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6673446238983453,
+      "learning_rate": 4.510309278350515e-08,
+      "loss": 1.2047,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7606322375403376,
+      "learning_rate": 5.154639175257731e-08,
+      "loss": 1.2659,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6494497544141264,
+      "learning_rate": 5.7989690721649485e-08,
+      "loss": 1.2103,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4544625796772091,
+      "learning_rate": 6.443298969072164e-08,
+      "loss": 1.2018,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6304114122962914,
+      "learning_rate": 7.08762886597938e-08,
+      "loss": 1.2343,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4828602067421827,
+      "learning_rate": 7.731958762886598e-08,
+      "loss": 1.1482,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5331330554386856,
+      "learning_rate": 8.376288659793814e-08,
+      "loss": 1.2087,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5284807423190944,
+      "learning_rate": 9.02061855670103e-08,
+      "loss": 1.2048,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.59607006938996,
+      "learning_rate": 9.664948453608247e-08,
+      "loss": 1.2014,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5809334694259137,
+      "learning_rate": 1.0309278350515462e-07,
+      "loss": 1.201,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5882822416771716,
+      "learning_rate": 1.095360824742268e-07,
+      "loss": 1.1776,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.40901829894322184,
+      "learning_rate": 1.1597938144329897e-07,
+      "loss": 1.1638,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5846930601219722,
+      "learning_rate": 1.2242268041237112e-07,
+      "loss": 1.2056,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5080185144930495,
+      "learning_rate": 1.2886597938144328e-07,
+      "loss": 1.2088,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5417632636662292,
+      "learning_rate": 1.3530927835051546e-07,
+      "loss": 1.2077,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6087231199320666,
+      "learning_rate": 1.417525773195876e-07,
+      "loss": 1.2383,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6600360298182159,
+      "learning_rate": 1.481958762886598e-07,
+      "loss": 1.2235,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4720151441701456,
+      "learning_rate": 1.5463917525773197e-07,
+      "loss": 1.2094,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4520594302377389,
+      "learning_rate": 1.6108247422680412e-07,
+      "loss": 1.183,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4646007077422491,
+      "learning_rate": 1.6752577319587627e-07,
+      "loss": 1.2793,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7724348048785946,
+      "learning_rate": 1.7396907216494843e-07,
+      "loss": 1.2119,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5963766134260833,
+      "learning_rate": 1.804123711340206e-07,
+      "loss": 1.2124,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.44281136689198974,
+      "learning_rate": 1.868556701030928e-07,
+      "loss": 1.183,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5848813922825735,
+      "learning_rate": 1.9329896907216494e-07,
+      "loss": 1.1965,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6350181434204577,
+      "learning_rate": 1.9974226804123712e-07,
+      "loss": 1.2209,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4102105166564691,
+      "learning_rate": 2.0618556701030925e-07,
+      "loss": 1.233,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4547896713814094,
+      "learning_rate": 2.1262886597938143e-07,
+      "loss": 1.1977,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.48260421693181227,
+      "learning_rate": 2.190721649484536e-07,
+      "loss": 1.2407,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.47606213247480483,
+      "learning_rate": 2.2551546391752576e-07,
+      "loss": 1.2091,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4186261733793144,
+      "learning_rate": 2.3195876288659794e-07,
+      "loss": 1.1767,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5610682632154923,
+      "learning_rate": 2.3840206185567012e-07,
+      "loss": 1.195,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4991728132294886,
+      "learning_rate": 2.4484536082474224e-07,
+      "loss": 1.1884,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.377945615144518,
+      "learning_rate": 2.512886597938144e-07,
+      "loss": 1.1889,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6192794218394144,
+      "learning_rate": 2.5773195876288655e-07,
+      "loss": 1.2404,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46359803524900906,
+      "learning_rate": 2.6417525773195873e-07,
+      "loss": 1.1784,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6617405723758515,
+      "learning_rate": 2.706185567010309e-07,
+      "loss": 1.2117,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7958161334530682,
+      "learning_rate": 2.770618556701031e-07,
+      "loss": 1.1944,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8107074055312508,
+      "learning_rate": 2.835051546391752e-07,
+      "loss": 1.237,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46174371139366394,
+      "learning_rate": 2.8994845360824745e-07,
+      "loss": 1.2125,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5719944209888713,
+      "learning_rate": 2.963917525773196e-07,
+      "loss": 1.2475,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4351960107554103,
+      "learning_rate": 3.028350515463917e-07,
+      "loss": 1.1695,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44512933223413065,
+      "learning_rate": 3.0927835051546394e-07,
+      "loss": 1.2178,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5528660111247783,
+      "learning_rate": 3.1572164948453606e-07,
+      "loss": 1.2027,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5116091364221563,
+      "learning_rate": 3.2216494845360824e-07,
+      "loss": 1.2224,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.39343853393523637,
+      "learning_rate": 3.286082474226804e-07,
+      "loss": 1.218,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.571131594344541,
+      "learning_rate": 3.3505154639175255e-07,
+      "loss": 1.252,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4938530416662391,
+      "learning_rate": 3.4149484536082473e-07,
+      "loss": 1.162,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.520502953726534,
+      "learning_rate": 3.4793814432989685e-07,
+      "loss": 1.183,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.523597976170542,
+      "learning_rate": 3.543814432989691e-07,
+      "loss": 1.1383,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3815871429726415,
+      "learning_rate": 3.608247422680412e-07,
+      "loss": 1.1643,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4835696356079593,
+      "learning_rate": 3.672680412371134e-07,
+      "loss": 1.1912,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4403194396238357,
+      "learning_rate": 3.737113402061856e-07,
+      "loss": 1.1931,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5282589912864021,
+      "learning_rate": 3.801546391752577e-07,
+      "loss": 1.243,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3346525289367091,
+      "learning_rate": 3.865979381443299e-07,
+      "loss": 1.2049,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5207281874834205,
+      "learning_rate": 3.9304123711340206e-07,
+      "loss": 1.186,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4763510371696751,
+      "learning_rate": 3.9948453608247424e-07,
+      "loss": 1.2109,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5226573743165585,
+      "learning_rate": 4.0592783505154637e-07,
+      "loss": 1.1849,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6164257855641101,
+      "learning_rate": 4.123711340206185e-07,
+      "loss": 1.2765,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45640817162369646,
+      "learning_rate": 4.188144329896907e-07,
+      "loss": 1.1517,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.381252987434718,
+      "learning_rate": 4.2525773195876285e-07,
+      "loss": 1.1483,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.37716885012133083,
+      "learning_rate": 4.3170103092783503e-07,
+      "loss": 1.1802,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.27910258398784316,
+      "learning_rate": 4.381443298969072e-07,
+      "loss": 1.1466,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5177367856818617,
+      "learning_rate": 4.4458762886597934e-07,
+      "loss": 1.1812,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3887407344356738,
+      "learning_rate": 4.510309278350515e-07,
+      "loss": 1.2225,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48503162312739034,
+      "learning_rate": 4.574742268041237e-07,
+      "loss": 1.1639,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44237735879442,
+      "learning_rate": 4.639175257731959e-07,
+      "loss": 1.1754,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41609679946936035,
+      "learning_rate": 4.70360824742268e-07,
+      "loss": 1.1399,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.32556234251915184,
+      "learning_rate": 4.7680412371134024e-07,
+      "loss": 1.1638,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.27379067788041994,
+      "learning_rate": 4.832474226804123e-07,
+      "loss": 1.1508,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3796015001836942,
+      "learning_rate": 4.896907216494845e-07,
+      "loss": 1.1174,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4145408169029473,
+      "learning_rate": 4.961340206185567e-07,
+      "loss": 1.1459,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.32089801961474884,
+      "learning_rate": 5.025773195876288e-07,
+      "loss": 1.1574,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3829621826884815,
+      "learning_rate": 5.090206185567009e-07,
+      "loss": 1.2045,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4007285090776862,
+      "learning_rate": 5.154639175257731e-07,
+      "loss": 1.1678,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40052906809761024,
+      "learning_rate": 5.219072164948454e-07,
+      "loss": 1.1588,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38092344104849307,
+      "learning_rate": 5.283505154639175e-07,
+      "loss": 1.1386,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3320360466235796,
+      "learning_rate": 5.347938144329896e-07,
+      "loss": 1.1366,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4085094235588679,
+      "learning_rate": 5.412371134020618e-07,
+      "loss": 1.1709,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2703217201718773,
+      "learning_rate": 5.47680412371134e-07,
+      "loss": 1.1469,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5142835166070265,
+      "learning_rate": 5.541237113402062e-07,
+      "loss": 1.1582,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.32469088747549585,
+      "learning_rate": 5.605670103092784e-07,
+      "loss": 1.1627,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47227589008148696,
+      "learning_rate": 5.670103092783504e-07,
+      "loss": 1.1536,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.27737415293856066,
+      "learning_rate": 5.734536082474226e-07,
+      "loss": 1.1411,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47844441679101485,
+      "learning_rate": 5.798969072164949e-07,
+      "loss": 1.1376,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.24740711882829364,
+      "learning_rate": 5.86340206185567e-07,
+      "loss": 1.1428,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3596770690225834,
+      "learning_rate": 5.927835051546392e-07,
+      "loss": 1.1306,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3853999217330004,
+      "learning_rate": 5.992268041237113e-07,
+      "loss": 1.125,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.28203836839179436,
+      "learning_rate": 6.056701030927834e-07,
+      "loss": 1.0974,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3815782154177524,
+      "learning_rate": 6.121134020618557e-07,
+      "loss": 1.1435,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40660461351756055,
+      "learning_rate": 6.185567010309279e-07,
+      "loss": 1.1252,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4473570719883677,
+      "learning_rate": 6.249999999999999e-07,
+      "loss": 1.1534,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3016667814758067,
+      "learning_rate": 6.314432989690721e-07,
+      "loss": 1.1572,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.25046681744569643,
+      "learning_rate": 6.378865979381443e-07,
+      "loss": 1.1354,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.321101969151723,
+      "learning_rate": 6.443298969072165e-07,
+      "loss": 1.131,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3447078961627486,
+      "learning_rate": 6.507731958762887e-07,
+      "loss": 1.1675,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3058191384904833,
+      "learning_rate": 6.572164948453608e-07,
+      "loss": 1.1486,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4071749991366515,
+      "learning_rate": 6.636597938144329e-07,
+      "loss": 1.1214,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.27190774717782273,
+      "learning_rate": 6.701030927835051e-07,
+      "loss": 1.1129,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3173952397629306,
+      "learning_rate": 6.765463917525773e-07,
+      "loss": 1.1067,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3487331387664773,
+      "learning_rate": 6.829896907216495e-07,
+      "loss": 1.144,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3324048358789938,
+      "learning_rate": 6.894329896907216e-07,
+      "loss": 1.1353,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3006630286415635,
+      "learning_rate": 6.958762886597937e-07,
+      "loss": 1.1254,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2627056030538818,
+      "learning_rate": 7.023195876288659e-07,
+      "loss": 1.0809,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2903433078125946,
+      "learning_rate": 7.087628865979382e-07,
+      "loss": 1.1063,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4466761860876455,
+      "learning_rate": 7.152061855670102e-07,
+      "loss": 1.1151,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.23461073207386746,
+      "learning_rate": 7.216494845360824e-07,
+      "loss": 1.1166,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.27417488924150696,
+      "learning_rate": 7.280927835051546e-07,
+      "loss": 1.1462,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.27294376187262775,
+      "learning_rate": 7.345360824742268e-07,
+      "loss": 1.1061,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.34139898616900827,
+      "learning_rate": 7.40979381443299e-07,
+      "loss": 1.1079,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.31579050892163335,
+      "learning_rate": 7.474226804123711e-07,
+      "loss": 1.1162,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.259281614297188,
+      "learning_rate": 7.538659793814432e-07,
+      "loss": 1.1067,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3329073225220458,
+      "learning_rate": 7.603092783505154e-07,
+      "loss": 1.1235,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3110773744452709,
+      "learning_rate": 7.667525773195877e-07,
+      "loss": 1.1048,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.33810273741235664,
+      "learning_rate": 7.731958762886598e-07,
+      "loss": 1.104,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.20126338290056853,
+      "learning_rate": 7.796391752577319e-07,
+      "loss": 1.1078,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.25841417283721924,
+      "learning_rate": 7.860824742268041e-07,
+      "loss": 1.092,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2781143907781358,
+      "learning_rate": 7.925257731958762e-07,
+      "loss": 1.1189,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3207040738845204,
+      "learning_rate": 7.989690721649485e-07,
+      "loss": 1.074,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.25223586961135275,
+      "learning_rate": 8.054123711340207e-07,
+      "loss": 1.1419,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2669597345215262,
+      "learning_rate": 8.118556701030927e-07,
+      "loss": 1.0925,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.28871763373364795,
+      "learning_rate": 8.182989690721649e-07,
+      "loss": 1.1159,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2773152422736672,
+      "learning_rate": 8.24742268041237e-07,
+      "loss": 1.0917,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2775315009668929,
+      "learning_rate": 8.311855670103093e-07,
+      "loss": 1.0299,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.2961928411844274,
+      "learning_rate": 8.376288659793815e-07,
+      "loss": 1.0688,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.29673031901143204,
+      "learning_rate": 8.440721649484535e-07,
+      "loss": 1.0816,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2070863413789537,
+      "learning_rate": 8.505154639175257e-07,
+      "loss": 1.0594,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.21872035798434927,
+      "learning_rate": 8.569587628865979e-07,
+      "loss": 1.0542,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.21097167700938158,
+      "learning_rate": 8.634020618556701e-07,
+      "loss": 1.0686,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2570218696498895,
+      "learning_rate": 8.698453608247422e-07,
+      "loss": 1.0624,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2492663026420371,
+      "learning_rate": 8.762886597938144e-07,
+      "loss": 1.0779,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.25176442133697435,
+      "learning_rate": 8.827319587628865e-07,
+      "loss": 1.0618,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.26370754246965605,
+      "learning_rate": 8.891752577319587e-07,
+      "loss": 1.056,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2592920964960942,
+      "learning_rate": 8.95618556701031e-07,
+      "loss": 1.0621,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2563316335870609,
+      "learning_rate": 9.02061855670103e-07,
+      "loss": 1.0703,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2450187809012757,
+      "learning_rate": 9.085051546391752e-07,
+      "loss": 1.0742,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3274692712066339,
+      "learning_rate": 9.149484536082474e-07,
+      "loss": 1.0669,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.19325188989551953,
+      "learning_rate": 9.213917525773195e-07,
+      "loss": 1.043,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2628965042539924,
+      "learning_rate": 9.278350515463918e-07,
+      "loss": 1.0743,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.23958482705295195,
+      "learning_rate": 9.342783505154639e-07,
+      "loss": 1.0645,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3080341759711365,
+      "learning_rate": 9.40721649484536e-07,
+      "loss": 1.0434,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.23504320384291294,
+      "learning_rate": 9.471649484536082e-07,
+      "loss": 1.0259,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2678847914513063,
+      "learning_rate": 9.536082474226805e-07,
+      "loss": 1.0365,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.22144786075966805,
+      "learning_rate": 9.600515463917527e-07,
+      "loss": 1.0719,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.24556908642575295,
+      "learning_rate": 9.664948453608246e-07,
+      "loss": 1.0489,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.23231695185782006,
+      "learning_rate": 9.729381443298968e-07,
+      "loss": 1.0476,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.21134107850080222,
+      "learning_rate": 9.79381443298969e-07,
+      "loss": 1.0509,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.21488532511557354,
+      "learning_rate": 9.858247422680412e-07,
+      "loss": 1.0158,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.24143524806993258,
+      "learning_rate": 9.922680412371133e-07,
+      "loss": 1.0923,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2311111870194462,
+      "learning_rate": 9.987113402061855e-07,
+      "loss": 1.0076,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2195717729307349,
+      "learning_rate": 9.999991892299655e-07,
+      "loss": 1.0525,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2215554391894305,
+      "learning_rate": 9.999958954812064e-07,
+      "loss": 1.0331,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.22222011437465564,
+      "learning_rate": 9.999900680972736e-07,
+      "loss": 1.0282,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.23475602808234386,
+      "learning_rate": 9.99981707107696e-07,
+      "loss": 1.0503,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.18975538764394448,
+      "learning_rate": 9.999708125548415e-07,
+      "loss": 1.0327,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.19604732138331937,
+      "learning_rate": 9.999573844939157e-07,
+      "loss": 1.0428,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.25173731704055136,
+      "learning_rate": 9.999414229929635e-07,
+      "loss": 1.0237,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2272232189984373,
+      "learning_rate": 9.999229281328663e-07,
+      "loss": 1.0811,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.17980969681455283,
+      "learning_rate": 9.999019000073436e-07,
+      "loss": 1.0387,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.15126647869778184,
+      "learning_rate": 9.998783387229512e-07,
+      "loss": 1.006,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.17174505452180605,
+      "learning_rate": 9.998522443990818e-07,
+      "loss": 1.0253,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2511655236810563,
+      "learning_rate": 9.998236171679632e-07,
+      "loss": 1.0307,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.18636095621653437,
+      "learning_rate": 9.997924571746588e-07,
+      "loss": 1.043,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.21909056173060806,
+      "learning_rate": 9.997587645770658e-07,
+      "loss": 1.061,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.17291380095750897,
+      "learning_rate": 9.99722539545915e-07,
+      "loss": 1.0271,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2363460759161499,
+      "learning_rate": 9.996837822647702e-07,
+      "loss": 1.0206,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.22227402710302605,
+      "learning_rate": 9.996424929300262e-07,
+      "loss": 1.0574,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2741496290133062,
+      "learning_rate": 9.995986717509096e-07,
+      "loss": 0.9992,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.20826964829580846,
+      "learning_rate": 9.995523189494754e-07,
+      "loss": 1.0264,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2291084286120521,
+      "learning_rate": 9.995034347606082e-07,
+      "loss": 1.0209,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.22323279187922898,
+      "learning_rate": 9.994520194320188e-07,
+      "loss": 1.0187,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.16709823133230353,
+      "learning_rate": 9.993980732242452e-07,
+      "loss": 1.0394,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.19834685529225118,
+      "learning_rate": 9.993415964106496e-07,
+      "loss": 1.0156,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2321751605838502,
+      "learning_rate": 9.992825892774176e-07,
+      "loss": 1.0161,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.19805409529248438,
+      "learning_rate": 9.992210521235567e-07,
+      "loss": 1.0484,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1676051380713844,
+      "learning_rate": 9.991569852608949e-07,
+      "loss": 1.0293,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.24615936792801493,
+      "learning_rate": 9.990903890140788e-07,
+      "loss": 0.99,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2578091409628526,
+      "learning_rate": 9.990212637205727e-07,
+      "loss": 1.0223,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.18692480745384935,
+      "learning_rate": 9.989496097306556e-07,
+      "loss": 0.9981,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1652811495241546,
+      "learning_rate": 9.98875427407421e-07,
+      "loss": 0.9664,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.15662822597690493,
+      "learning_rate": 9.987987171267737e-07,
+      "loss": 1.0507,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.17056334334264195,
+      "learning_rate": 9.987194792774286e-07,
+      "loss": 0.9951,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.14148188737112052,
+      "learning_rate": 9.986377142609089e-07,
+      "loss": 1.023,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.19725317368255313,
+      "learning_rate": 9.985534224915428e-07,
+      "loss": 1.0182,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.14444259171535412,
+      "learning_rate": 9.984666043964635e-07,
+      "loss": 1.0378,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1502581445533813,
+      "learning_rate": 9.983772604156055e-07,
+      "loss": 0.9952,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.1780329553764791,
+      "learning_rate": 9.982853910017022e-07,
+      "loss": 1.0244,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.17727060881813794,
+      "learning_rate": 9.981909966202845e-07,
+      "loss": 0.9809,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.20589035306039621,
+      "learning_rate": 9.980940777496788e-07,
+      "loss": 1.0599,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.18545517599149922,
+      "learning_rate": 9.979946348810028e-07,
+      "loss": 1.0224,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.11588831779201968,
+      "learning_rate": 9.978926685181647e-07,
+      "loss": 1.0061,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.168924484846306,
+      "learning_rate": 9.977881791778596e-07,
+      "loss": 1.0152,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.18722526742768686,
+      "learning_rate": 9.976811673895678e-07,
+      "loss": 0.9997,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15424043401003767,
+      "learning_rate": 9.975716336955514e-07,
+      "loss": 1.0154,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15044745918361932,
+      "learning_rate": 9.974595786508517e-07,
+      "loss": 0.9882,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.14468834621260726,
+      "learning_rate": 9.973450028232867e-07,
+      "loss": 1.0199,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.19040015939009786,
+      "learning_rate": 9.97227906793448e-07,
+      "loss": 0.9649,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.16984918915025987,
+      "learning_rate": 9.971082911546973e-07,
+      "loss": 1.0095,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.12811611913997542,
+      "learning_rate": 9.969861565131647e-07,
+      "loss": 1.017,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.19000247404724804,
+      "learning_rate": 9.968615034877442e-07,
+      "loss": 0.9917,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15165176032649844,
+      "learning_rate": 9.967343327100918e-07,
+      "loss": 1.0281,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.1536174536291069,
+      "learning_rate": 9.966046448246218e-07,
+      "loss": 1.0183,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.13025500392160377,
+      "learning_rate": 9.964724404885024e-07,
+      "loss": 1.0048,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.15427238149979788,
+      "learning_rate": 9.963377203716547e-07,
+      "loss": 0.9857,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.17362206762762045,
+      "learning_rate": 9.962004851567477e-07,
+      "loss": 0.9868,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1515840740686245,
+      "learning_rate": 9.960607355391948e-07,
+      "loss": 1.0484,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.18078505990832724,
+      "learning_rate": 9.959184722271509e-07,
+      "loss": 0.989,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1400544442856198,
+      "learning_rate": 9.957736959415087e-07,
+      "loss": 1.0248,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.14796181710930523,
+      "learning_rate": 9.956264074158947e-07,
+      "loss": 0.9962,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.16131319155245327,
+      "learning_rate": 9.954766073966657e-07,
+      "loss": 1.0091,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.11961473727194277,
+      "learning_rate": 9.953242966429055e-07,
+      "loss": 0.9676,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.19617377831011606,
+      "learning_rate": 9.951694759264198e-07,
+      "loss": 1.0401,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.15167856400502114,
+      "learning_rate": 9.950121460317337e-07,
+      "loss": 0.9875,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.15758120902216588,
+      "learning_rate": 9.94852307756087e-07,
+      "loss": 1.0142,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.1824885693101237,
+      "learning_rate": 9.946899619094295e-07,
+      "loss": 0.9885,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.14102156144718092,
+      "learning_rate": 9.945251093144187e-07,
+      "loss": 1.0167,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12522093088535893,
+      "learning_rate": 9.943577508064143e-07,
+      "loss": 1.0003,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.14673760387636478,
+      "learning_rate": 9.941878872334738e-07,
+      "loss": 1.0037,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.12909658867541232,
+      "learning_rate": 9.940155194563487e-07,
+      "loss": 0.998,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1318004222137836,
+      "learning_rate": 9.938406483484808e-07,
+      "loss": 0.9813,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1642253814198742,
+      "learning_rate": 9.936632747959961e-07,
+      "loss": 1.01,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.15273940217678547,
+      "learning_rate": 9.93483399697702e-07,
+      "loss": 1.0162,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.147954680503508,
+      "learning_rate": 9.933010239650817e-07,
+      "loss": 0.9754,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1285580753566918,
+      "learning_rate": 9.931161485222898e-07,
+      "loss": 0.9918,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1296590022014456,
+      "learning_rate": 9.92928774306148e-07,
+      "loss": 1.0132,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.14081183459860172,
+      "learning_rate": 9.927389022661398e-07,
+      "loss": 0.9739,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12747057419734933,
+      "learning_rate": 9.92546533364406e-07,
+      "loss": 1.0019,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.14594195188886933,
+      "learning_rate": 9.9235166857574e-07,
+      "loss": 0.994,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12053499310567327,
+      "learning_rate": 9.921543088875824e-07,
+      "loss": 0.9842,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.14811236379339104,
+      "learning_rate": 9.919544553000162e-07,
+      "loss": 0.9817,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1641024890846564,
+      "learning_rate": 9.91752108825762e-07,
+      "loss": 1.0132,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.12297856461945962,
+      "learning_rate": 9.915472704901724e-07,
+      "loss": 0.9964,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1231168753765529,
+      "learning_rate": 9.913399413312272e-07,
+      "loss": 0.9892,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.1547692025114942,
+      "learning_rate": 9.91130122399528e-07,
+      "loss": 0.9635,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.13208977967804597,
+      "learning_rate": 9.909178147582928e-07,
+      "loss": 0.9842,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1296374959772723,
+      "learning_rate": 9.907030194833506e-07,
+      "loss": 1.0207,
+      "step": 1205
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14047567894861018,
+      "learning_rate": 9.904857376631359e-07,
+      "loss": 0.9949,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12971642636167702,
+      "learning_rate": 9.902659703986835e-07,
+      "loss": 0.9987,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14741563386277243,
+      "learning_rate": 9.90043718803623e-07,
+      "loss": 0.9932,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.15198166013399284,
+      "learning_rate": 9.898189840041721e-07,
+      "loss": 0.9419,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12789127412963947,
+      "learning_rate": 9.895917671391325e-07,
+      "loss": 0.9723,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13331264765150608,
+      "learning_rate": 9.893620693598829e-07,
+      "loss": 0.9604,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1334119431045558,
+      "learning_rate": 9.891298918303732e-07,
+      "loss": 0.9638,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13939368536733832,
+      "learning_rate": 9.8889523572712e-07,
+      "loss": 0.9917,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.1531060103163608,
+      "learning_rate": 9.886581022391985e-07,
+      "loss": 0.9796,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14760001820883373,
+      "learning_rate": 9.88418492568238e-07,
+      "loss": 0.9889,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.12723064465221331,
+      "learning_rate": 9.881764079284159e-07,
+      "loss": 1.0023,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14728586609836652,
+      "learning_rate": 9.879318495464503e-07,
+      "loss": 0.9819,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.14555977172052836,
+      "learning_rate": 9.876848186615949e-07,
+      "loss": 0.9963,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.13819872477226244,
+      "learning_rate": 9.87435316525632e-07,
+      "loss": 0.9781,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12573391469885703,
+      "learning_rate": 9.871833444028673e-07,
+      "loss": 0.9794,
+      "step": 1280
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12077151555401149,
+      "learning_rate": 9.869289035701215e-07,
+      "loss": 1.006,
+      "step": 1285
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1467514469469745,
+      "learning_rate": 9.866719953167261e-07,
+      "loss": 0.9902,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.11708967635411316,
+      "learning_rate": 9.864126209445152e-07,
+      "loss": 0.9883,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.11360811182885677,
+      "learning_rate": 9.861507817678195e-07,
+      "loss": 0.9789,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13198590705652274,
+      "learning_rate": 9.858864791134604e-07,
+      "loss": 0.9688,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13575024467776314,
+      "learning_rate": 9.856197143207412e-07,
+      "loss": 0.9644,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13043038559635964,
+      "learning_rate": 9.853504887414428e-07,
+      "loss": 0.9992,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.11983900287466571,
+      "learning_rate": 9.850788037398146e-07,
+      "loss": 0.9772,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.13811512339323653,
+      "learning_rate": 9.8480466069257e-07,
+      "loss": 1.0026,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1311244619060084,
+      "learning_rate": 9.845280609888769e-07,
+      "loss": 0.983,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1414852776034834,
+      "learning_rate": 9.842490060303523e-07,
+      "loss": 0.9766,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12734569735165552,
+      "learning_rate": 9.839674972310547e-07,
+      "loss": 0.9885,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1357778005359003,
+      "learning_rate": 9.836835360174772e-07,
+      "loss": 0.9937,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.12592944796646743,
+      "learning_rate": 9.8339712382854e-07,
+      "loss": 0.9921,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.1457978128910316,
+      "learning_rate": 9.831082621155824e-07,
+      "loss": 0.947,
+      "step": 1355
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1325166381460097,
+      "learning_rate": 9.828169523423572e-07,
+      "loss": 0.9546,
+      "step": 1360
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13833214744042951,
+      "learning_rate": 9.82523195985022e-07,
+      "loss": 1.0091,
+      "step": 1365
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1146408721824335,
+      "learning_rate": 9.822269945321317e-07,
+      "loss": 0.9902,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.14160449889230736,
+      "learning_rate": 9.819283494846315e-07,
+      "loss": 0.957,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12248759233813743,
+      "learning_rate": 9.81627262355849e-07,
+      "loss": 0.9979,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12672187402686996,
+      "learning_rate": 9.813237346714866e-07,
+      "loss": 0.971,
+      "step": 1385
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12812244033197875,
+      "learning_rate": 9.810177679696138e-07,
+      "loss": 0.9794,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11747065298995094,
+      "learning_rate": 9.807093638006593e-07,
+      "loss": 0.9701,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13214800408242058,
+      "learning_rate": 9.803985237274033e-07,
+      "loss": 0.9943,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11405812793040136,
+      "learning_rate": 9.800852493249694e-07,
+      "loss": 0.9845,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.18105251260157323,
+      "learning_rate": 9.797695421808165e-07,
+      "loss": 0.9616,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.13316430762709344,
+      "learning_rate": 9.794514038947316e-07,
+      "loss": 0.997,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.1348504677245663,
+      "learning_rate": 9.791308360788204e-07,
+      "loss": 0.9695,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.12685419458554126,
+      "learning_rate": 9.788078403575e-07,
+      "loss": 0.9617,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.11840608790985199,
+      "learning_rate": 9.784824183674903e-07,
+      "loss": 0.9768,
+      "step": 1430
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13207408490130262,
+      "learning_rate": 9.781545717578064e-07,
+      "loss": 0.9936,
+      "step": 1435
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11304036704824351,
+      "learning_rate": 9.778243021897492e-07,
+      "loss": 0.9791,
+      "step": 1440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1122526120538181,
+      "learning_rate": 9.774916113368973e-07,
+      "loss": 0.9888,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1468262771140114,
+      "learning_rate": 9.771565008850998e-07,
+      "loss": 0.9777,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.13518121118171272,
+      "learning_rate": 9.768189725324651e-07,
+      "loss": 1.0079,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1209904755124303,
+      "learning_rate": 9.764790279893552e-07,
+      "loss": 0.9896,
+      "step": 1460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.19359250780157092,
+      "learning_rate": 9.76136668978375e-07,
+      "loss": 0.9495,
+      "step": 1465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12242033880749509,
+      "learning_rate": 9.757918972343643e-07,
+      "loss": 0.983,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12535113182416072,
+      "learning_rate": 9.754447145043889e-07,
+      "loss": 0.9463,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.1391622931596189,
+      "learning_rate": 9.750951225477322e-07,
+      "loss": 0.9917,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12193674639210303,
+      "learning_rate": 9.747431231358861e-07,
+      "loss": 0.9961,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12450683996290608,
+      "learning_rate": 9.743887180525407e-07,
+      "loss": 0.9679,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11445126212265636,
+      "learning_rate": 9.740319090935774e-07,
+      "loss": 0.9823,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11892931759502721,
+      "learning_rate": 9.736726980670587e-07,
+      "loss": 0.9511,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.12969159665700125,
+      "learning_rate": 9.733110867932189e-07,
+      "loss": 0.9557,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.11210576618142787,
+      "learning_rate": 9.729470771044549e-07,
+      "loss": 0.9783,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11347659350713385,
+      "learning_rate": 9.725806708453173e-07,
+      "loss": 0.9796,
+      "step": 1515
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1283052843940726,
+      "learning_rate": 9.72211869872501e-07,
+      "loss": 0.9741,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12296735167262948,
+      "learning_rate": 9.71840676054836e-07,
+      "loss": 0.9682,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.13085115071550882,
+      "learning_rate": 9.714670912732764e-07,
+      "loss": 0.9956,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12769176007630262,
+      "learning_rate": 9.710911174208936e-07,
+      "loss": 0.9657,
+      "step": 1535
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.10837307445722577,
+      "learning_rate": 9.707127564028639e-07,
+      "loss": 0.9728,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1217312829871794,
+      "learning_rate": 9.703320101364605e-07,
+      "loss": 1.0011,
+      "step": 1545
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11695765188963504,
+      "learning_rate": 9.699488805510442e-07,
+      "loss": 0.9879,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12917996114342756,
+      "learning_rate": 9.695633695880511e-07,
+      "loss": 0.9674,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12845898744262566,
+      "learning_rate": 9.691754792009864e-07,
+      "loss": 0.9806,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.10413846981546938,
+      "learning_rate": 9.687852113554114e-07,
+      "loss": 0.9675,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.12181336129845245,
+      "learning_rate": 9.683925680289347e-07,
+      "loss": 0.9415,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.1356758793892471,
+      "learning_rate": 9.67997551211203e-07,
+      "loss": 0.9607,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11905589158103425,
+      "learning_rate": 9.676001629038894e-07,
+      "loss": 0.9703,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.11587648523721791,
+      "learning_rate": 9.672004051206847e-07,
+      "loss": 0.9688,
+      "step": 1585
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.14376516716077894,
+      "learning_rate": 9.66798279887286e-07,
+      "loss": 0.981,
+      "step": 1590
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.129776505672388,
+      "learning_rate": 9.663937892413877e-07,
+      "loss": 1.0056,
+      "step": 1595
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12384072073184683,
+      "learning_rate": 9.659869352326695e-07,
+      "loss": 0.9893,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12119125322212589,
+      "learning_rate": 9.65577719922788e-07,
+      "loss": 0.9841,
+      "step": 1605
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12646358628409698,
+      "learning_rate": 9.651661453853648e-07,
+      "loss": 0.9734,
+      "step": 1610
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1475852617467925,
+      "learning_rate": 9.647522137059762e-07,
+      "loss": 0.9689,
+      "step": 1615
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.11856392949123994,
+      "learning_rate": 9.643359269821435e-07,
+      "loss": 0.983,
+      "step": 1620
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1367081155073447,
+      "learning_rate": 9.639172873233213e-07,
+      "loss": 0.9809,
+      "step": 1625
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.11040674904523864,
+      "learning_rate": 9.63496296850887e-07,
+      "loss": 0.9737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.11767440815015452,
+      "learning_rate": 9.630729576981311e-07,
+      "loss": 0.9426,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1293468383720689,
+      "learning_rate": 9.62647272010245e-07,
+      "loss": 0.9726,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.13291735985036215,
+      "learning_rate": 9.622192419443108e-07,
+      "loss": 0.98,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1076782094194722,
+      "learning_rate": 9.617888696692905e-07,
+      "loss": 1.0099,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12587437191905246,
+      "learning_rate": 9.613561573660143e-07,
+      "loss": 0.9509,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.1304868748703427,
+      "learning_rate": 9.609211072271708e-07,
+      "loss": 0.9498,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.12375523788413598,
+      "learning_rate": 9.604837214572948e-07,
+      "loss": 0.974,
+      "step": 1665
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12108928838694019,
+      "learning_rate": 9.600440022727563e-07,
+      "loss": 0.9805,
+      "step": 1670
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11987477249423575,
+      "learning_rate": 9.596019519017494e-07,
+      "loss": 0.979,
+      "step": 1675
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.122213780465657,
+      "learning_rate": 9.591575725842815e-07,
+      "loss": 0.9458,
+      "step": 1680
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11337442942828732,
+      "learning_rate": 9.58710866572161e-07,
+      "loss": 0.9901,
+      "step": 1685
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.1157208756675931,
+      "learning_rate": 9.582618361289869e-07,
+      "loss": 0.9729,
+      "step": 1690
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11427905380667161,
+      "learning_rate": 9.578104835301363e-07,
+      "loss": 0.9816,
+      "step": 1695
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.13096930408065172,
+      "learning_rate": 9.573568110627538e-07,
+      "loss": 0.978,
+      "step": 1700
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12486479832605693,
+      "learning_rate": 9.56900821025739e-07,
+      "loss": 0.9586,
+      "step": 1705
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12386202803994883,
+      "learning_rate": 9.56442515729736e-07,
+      "loss": 1.005,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.14065409142553154,
+      "learning_rate": 9.559818974971206e-07,
+      "loss": 0.9776,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.11322432292616115,
+      "learning_rate": 9.555189686619888e-07,
+      "loss": 0.9788,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12634031248565974,
+      "learning_rate": 9.550537315701459e-07,
+      "loss": 0.9969,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12309381631580273,
+      "learning_rate": 9.54586188579093e-07,
+      "loss": 0.9319,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.121835402954551,
+      "learning_rate": 9.541163420580163e-07,
+      "loss": 0.977,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.12935419984057203,
+      "learning_rate": 9.536441943877749e-07,
+      "loss": 0.9639,
+      "step": 1740
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12129980464103775,
+      "learning_rate": 9.531697479608883e-07,
+      "loss": 0.9373,
+      "step": 1745
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1326651035511269,
+      "learning_rate": 9.526930051815244e-07,
+      "loss": 0.9632,
+      "step": 1750
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12297750287210125,
+      "learning_rate": 9.522139684654876e-07,
+      "loss": 0.9559,
+      "step": 1755
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12067188452461271,
+      "learning_rate": 9.517326402402065e-07,
+      "loss": 0.9313,
+      "step": 1760
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1315894118442904,
+      "learning_rate": 9.512490229447212e-07,
+      "loss": 0.9697,
+      "step": 1765
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.11332588803910468,
+      "learning_rate": 9.507631190296716e-07,
+      "loss": 0.9568,
+      "step": 1770
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.1227083750915719,
+      "learning_rate": 9.502749309572842e-07,
+      "loss": 0.9719,
+      "step": 1775
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12814161346575262,
+      "learning_rate": 9.497844612013603e-07,
+      "loss": 0.9548,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.11751451859714467,
+      "learning_rate": 9.492917122472634e-07,
+      "loss": 0.9838,
+      "step": 1785
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.11587622441404007,
+      "learning_rate": 9.487966865919057e-07,
+      "loss": 0.9628,
+      "step": 1790
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12134140859929958,
+      "learning_rate": 9.482993867437372e-07,
+      "loss": 0.9523,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12849338023068144,
+      "learning_rate": 9.477998152227308e-07,
+      "loss": 0.9588,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12625891069355497,
+      "learning_rate": 9.472979745603716e-07,
+      "loss": 0.9798,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12253962694756051,
+      "learning_rate": 9.467938672996425e-07,
+      "loss": 0.9787,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.12260035298804736,
+      "learning_rate": 9.462874959950126e-07,
+      "loss": 0.9528,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.11736154982661531,
+      "learning_rate": 9.457788632124228e-07,
+      "loss": 0.9762,
+      "step": 1820
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1205989412299247,
+      "learning_rate": 9.452679715292746e-07,
+      "loss": 0.9818,
+      "step": 1825
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.13241922588449712,
+      "learning_rate": 9.447548235344152e-07,
+      "loss": 0.9767,
+      "step": 1830
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11642078256778576,
+      "learning_rate": 9.442394218281256e-07,
+      "loss": 1.0041,
+      "step": 1835
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11870531221632347,
+      "learning_rate": 9.437217690221071e-07,
+      "loss": 0.961,
+      "step": 1840
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11726042345167338,
+      "learning_rate": 9.432018677394679e-07,
+      "loss": 0.9765,
+      "step": 1845
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12613937220055743,
+      "learning_rate": 9.4267972061471e-07,
+      "loss": 0.9826,
+      "step": 1850
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.1284851302569357,
+      "learning_rate": 9.421553302937158e-07,
+      "loss": 0.9433,
+      "step": 1855
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11615001043895708,
+      "learning_rate": 9.416286994337345e-07,
+      "loss": 0.9665,
+      "step": 1860
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11875489889051563,
+      "learning_rate": 9.410998307033691e-07,
+      "loss": 0.9592,
+      "step": 1865
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11177135430794444,
+      "learning_rate": 9.405687267825622e-07,
+      "loss": 0.9468,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12101369757303639,
+      "learning_rate": 9.400353903625833e-07,
+      "loss": 0.9604,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.10224191697554577,
+      "learning_rate": 9.394998241460144e-07,
+      "loss": 0.9651,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11954311476860209,
+      "learning_rate": 9.389620308467365e-07,
+      "loss": 0.9927,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.11827360297873002,
+      "learning_rate": 9.384220131899159e-07,
+      "loss": 0.9861,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.12528598562533394,
+      "learning_rate": 9.378797739119908e-07,
+      "loss": 0.9539,
+      "step": 1895
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12077838103796736,
+      "learning_rate": 9.373353157606564e-07,
+      "loss": 1.008,
+      "step": 1900
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12157572608793726,
+      "learning_rate": 9.367886414948525e-07,
+      "loss": 0.9644,
+      "step": 1905
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11679023789316648,
+      "learning_rate": 9.362397538847476e-07,
+      "loss": 0.9727,
+      "step": 1910
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11779342566697402,
+      "learning_rate": 9.356886557117268e-07,
+      "loss": 0.9468,
+      "step": 1915
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11278724950455456,
+      "learning_rate": 9.351353497683763e-07,
+      "loss": 0.9698,
+      "step": 1920
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13113081100313076,
+      "learning_rate": 9.3457983885847e-07,
+      "loss": 0.9529,
+      "step": 1925
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12885728475683297,
+      "learning_rate": 9.34022125796955e-07,
+      "loss": 0.9792,
+      "step": 1930
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11775755061403537,
+      "learning_rate": 9.33462213409937e-07,
+      "loss": 0.9632,
+      "step": 1935
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12295005090776762,
+      "learning_rate": 9.329001045346673e-07,
+      "loss": 0.9725,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12535914174879317,
+      "learning_rate": 9.323358020195266e-07,
+      "loss": 0.9564,
+      "step": 1945
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12914283532793955,
+      "learning_rate": 9.317693087240118e-07,
+      "loss": 0.9876,
+      "step": 1950
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12911790555139133,
+      "learning_rate": 9.31200627518721e-07,
+      "loss": 0.9644,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.12341612797287908,
+      "learning_rate": 9.306297612853396e-07,
+      "loss": 0.968,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.1278443847633763,
+      "learning_rate": 9.300567129166245e-07,
+      "loss": 0.9782,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.11985544926407674,
+      "learning_rate": 9.294814853163911e-07,
+      "loss": 0.9717,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.13572222140252924,
+      "learning_rate": 9.289040813994964e-07,
+      "loss": 0.9346,
+      "step": 1975
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.11611937435652277,
+      "learning_rate": 9.283245040918267e-07,
+      "loss": 0.9177,
+      "step": 1980
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.11132255535564403,
+      "learning_rate": 9.277427563302806e-07,
+      "loss": 0.998,
+      "step": 1985
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12419662446899254,
+      "learning_rate": 9.271588410627559e-07,
+      "loss": 0.9611,
+      "step": 1990
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12038591990741773,
+      "learning_rate": 9.265727612481331e-07,
+      "loss": 0.957,
+      "step": 1995
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12480335025403838,
+      "learning_rate": 9.259845198562615e-07,
+      "loss": 0.9968,
+      "step": 2000
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12200287552574611,
+      "learning_rate": 9.25394119867944e-07,
+      "loss": 0.9979,
+      "step": 2005
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12350690343547352,
+      "learning_rate": 9.248015642749211e-07,
+      "loss": 0.9703,
+      "step": 2010
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12141019504963686,
+      "learning_rate": 9.242068560798572e-07,
+      "loss": 0.9635,
+      "step": 2015
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.11905567730302113,
+      "learning_rate": 9.236099982963239e-07,
+      "loss": 0.9419,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13032217117158715,
+      "learning_rate": 9.23010993948786e-07,
+      "loss": 0.9931,
+      "step": 2025
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12635482914205362,
+      "learning_rate": 9.224098460725855e-07,
+      "loss": 0.9651,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.1215769182297846,
+      "learning_rate": 9.218065577139259e-07,
+      "loss": 0.966,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13068384626096122,
+      "learning_rate": 9.212011319298577e-07,
+      "loss": 0.9436,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.12482024493118403,
+      "learning_rate": 9.205935717882621e-07,
+      "loss": 0.9498,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.11702857739266968,
+      "learning_rate": 9.199838803678361e-07,
+      "loss": 0.9983,
+      "step": 2050
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.11916950337238004,
+      "learning_rate": 9.19372060758076e-07,
+      "loss": 0.9628,
+      "step": 2055
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12996497328936654,
+      "learning_rate": 9.187581160592628e-07,
+      "loss": 0.9523,
+      "step": 2060
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.11495390283812988,
+      "learning_rate": 9.181420493824457e-07,
+      "loss": 0.9909,
+      "step": 2065
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1296523350381107,
+      "learning_rate": 9.175238638494269e-07,
+      "loss": 0.9797,
+      "step": 2070
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.13197855332382968,
+      "learning_rate": 9.169035625927451e-07,
+      "loss": 0.9677,
+      "step": 2075
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.11844146484544314,
+      "learning_rate": 9.162811487556606e-07,
+      "loss": 0.9712,
+      "step": 2080
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.11914757859089617,
+      "learning_rate": 9.156566254921382e-07,
+      "loss": 0.9695,
+      "step": 2085
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12621452494766983,
+      "learning_rate": 9.150299959668322e-07,
+      "loss": 0.9574,
+      "step": 2090
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12317065777299938,
+      "learning_rate": 9.144012633550698e-07,
+      "loss": 0.9606,
+      "step": 2095
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12010581474012526,
+      "learning_rate": 9.137704308428353e-07,
+      "loss": 0.9566,
+      "step": 2100
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12447883448147107,
+      "learning_rate": 9.131375016267538e-07,
+      "loss": 0.9476,
+      "step": 2105
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12274591175424188,
+      "learning_rate": 9.125024789140753e-07,
+      "loss": 0.9638,
+      "step": 2110
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12252334173704718,
+      "learning_rate": 9.118653659226576e-07,
+      "loss": 0.9798,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.1222767957152178,
+      "learning_rate": 9.112261658809513e-07,
+      "loss": 0.9685,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12268185431367663,
+      "learning_rate": 9.105848820279824e-07,
+      "loss": 0.9575,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.12300764959353264,
+      "learning_rate": 9.099415176133358e-07,
+      "loss": 0.9893,
+      "step": 2130
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1316049657846485,
+      "learning_rate": 9.092960758971402e-07,
+      "loss": 0.9494,
+      "step": 2135
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12593613376659166,
+      "learning_rate": 9.0864856015005e-07,
+      "loss": 0.9803,
+      "step": 2140
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12491280405268614,
+      "learning_rate": 9.079989736532292e-07,
+      "loss": 0.9526,
+      "step": 2145
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.13667972400904602,
+      "learning_rate": 9.073473196983355e-07,
+      "loss": 1.001,
+      "step": 2150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.120396037304972,
+      "learning_rate": 9.066936015875026e-07,
+      "loss": 0.9469,
+      "step": 2155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12023662183951149,
+      "learning_rate": 9.060378226333243e-07,
+      "loss": 0.9851,
+      "step": 2160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12187472737970531,
+      "learning_rate": 9.05379986158837e-07,
+      "loss": 0.9354,
+      "step": 2165
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12800207521977724,
+      "learning_rate": 9.047200954975032e-07,
+      "loss": 0.9381,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12166314579262971,
+      "learning_rate": 9.040581539931949e-07,
+      "loss": 0.9586,
+      "step": 2175
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12434303113647774,
+      "learning_rate": 9.033941650001762e-07,
+      "loss": 0.9365,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12516979145004994,
+      "learning_rate": 9.027281318830864e-07,
+      "loss": 0.9635,
+      "step": 2185
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1293824685918648,
+      "learning_rate": 9.02060058016923e-07,
+      "loss": 1.0137,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.1298085174968314,
+      "learning_rate": 9.013899467870248e-07,
+      "loss": 0.9423,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12429239024861925,
+      "learning_rate": 9.007178015890545e-07,
+      "loss": 0.9597,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.12397378241772136,
+      "learning_rate": 9.000436258289812e-07,
+      "loss": 0.9755,
+      "step": 2205
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12605131650079643,
+      "learning_rate": 8.993674229230641e-07,
+      "loss": 0.9811,
+      "step": 2210
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.11566383698614861,
+      "learning_rate": 8.98689196297834e-07,
+      "loss": 0.9424,
+      "step": 2215
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.11892600295495828,
+      "learning_rate": 8.980089493900771e-07,
+      "loss": 0.9574,
+      "step": 2220
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12113537776351631,
+      "learning_rate": 8.973266856468164e-07,
+      "loss": 0.9656,
+      "step": 2225
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1222139608324094,
+      "learning_rate": 8.966424085252952e-07,
+      "loss": 0.9775,
+      "step": 2230
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1297970669680985,
+      "learning_rate": 8.959561214929591e-07,
+      "loss": 0.9507,
+      "step": 2235
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1338252988966343,
+      "learning_rate": 8.952678280274387e-07,
+      "loss": 0.9547,
+      "step": 2240
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13517688877359477,
+      "learning_rate": 8.945775316165316e-07,
+      "loss": 0.9504,
+      "step": 2245
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12500641957192377,
+      "learning_rate": 8.938852357581851e-07,
+      "loss": 0.9434,
+      "step": 2250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1177837105657456,
+      "learning_rate": 8.931909439604781e-07,
+      "loss": 0.9252,
+      "step": 2255
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.1179877539798463,
+      "learning_rate": 8.924946597416041e-07,
+      "loss": 0.9721,
+      "step": 2260
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12723591406312473,
+      "learning_rate": 8.91796386629852e-07,
+      "loss": 0.9515,
+      "step": 2265
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12044456538597373,
+      "learning_rate": 8.910961281635898e-07,
+      "loss": 0.963,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.13552937162823855,
+      "learning_rate": 8.903938878912454e-07,
+      "loss": 0.9638,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12676999906732542,
+      "learning_rate": 8.896896693712894e-07,
+      "loss": 0.9461,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.12632053440927568,
+      "learning_rate": 8.889834761722168e-07,
+      "loss": 0.9541,
+      "step": 2285
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12896734155124848,
+      "learning_rate": 8.882753118725286e-07,
+      "loss": 0.9518,
+      "step": 2290
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.11150352419412583,
+      "learning_rate": 8.875651800607145e-07,
+      "loss": 0.9158,
+      "step": 2295
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13166934919759735,
+      "learning_rate": 8.868530843352336e-07,
+      "loss": 1.0062,
+      "step": 2300
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1319973844885429,
+      "learning_rate": 8.861390283044974e-07,
+      "loss": 0.9711,
+      "step": 2305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12441621558722263,
+      "learning_rate": 8.854230155868504e-07,
+      "loss": 0.9597,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12890914904233772,
+      "learning_rate": 8.847050498105524e-07,
+      "loss": 0.9656,
+      "step": 2315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1273617283902276,
+      "learning_rate": 8.839851346137599e-07,
+      "loss": 0.9934,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.11757718989612276,
+      "learning_rate": 8.832632736445078e-07,
+      "loss": 0.9601,
+      "step": 2325
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13076595044539988,
+      "learning_rate": 8.82539470560691e-07,
+      "loss": 0.9576,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12206246271024937,
+      "learning_rate": 8.818137290300452e-07,
+      "loss": 0.9298,
+      "step": 2335
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13535937307748616,
+      "learning_rate": 8.810860527301294e-07,
+      "loss": 0.9733,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.13713712744899167,
+      "learning_rate": 8.803564453483062e-07,
+      "loss": 0.9654,
+      "step": 2345
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12512145189174623,
+      "learning_rate": 8.79624910581724e-07,
+      "loss": 0.9699,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.1260724469087313,
+      "learning_rate": 8.788914521372975e-07,
+      "loss": 0.9549,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.12980048824126203,
+      "learning_rate": 8.781560737316895e-07,
+      "loss": 0.9772,
+      "step": 2360
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12049218660958246,
+      "learning_rate": 8.774187790912917e-07,
+      "loss": 0.9738,
+      "step": 2365
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13906663291546636,
+      "learning_rate": 8.766795719522062e-07,
+      "loss": 0.9535,
+      "step": 2370
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1149767822696225,
+      "learning_rate": 8.759384560602261e-07,
+      "loss": 0.9633,
+      "step": 2375
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12039678951317301,
+      "learning_rate": 8.751954351708168e-07,
+      "loss": 0.9586,
+      "step": 2380
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.11913613862640543,
+      "learning_rate": 8.744505130490969e-07,
+      "loss": 0.9669,
+      "step": 2385
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13007720862986635,
+      "learning_rate": 8.737036934698189e-07,
+      "loss": 0.9569,
+      "step": 2390
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.11959766565309056,
+      "learning_rate": 8.729549802173513e-07,
+      "loss": 0.9863,
+      "step": 2395
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12705965089741275,
+      "learning_rate": 8.72204377085657e-07,
+      "loss": 0.9456,
+      "step": 2400
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12279006581200576,
+      "learning_rate": 8.714518878782766e-07,
+      "loss": 0.9778,
+      "step": 2405
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.119533444273622,
+      "learning_rate": 8.706975164083074e-07,
+      "loss": 0.9421,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12268234161105362,
+      "learning_rate": 8.699412664983854e-07,
+      "loss": 0.967,
+      "step": 2415
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.13566599274266372,
+      "learning_rate": 8.691831419806643e-07,
+      "loss": 0.956,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1253320784302625,
+      "learning_rate": 8.684231466967977e-07,
+      "loss": 0.9463,
+      "step": 2425
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.12196615182740232,
+      "learning_rate": 8.676612844979187e-07,
+      "loss": 0.954,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.1252029993708072,
+      "learning_rate": 8.66897559244621e-07,
+      "loss": 0.9953,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.11997054377450589,
+      "learning_rate": 8.661319748069384e-07,
+      "loss": 0.9468,
+      "step": 2440
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12208675750883287,
+      "learning_rate": 8.653645350643264e-07,
+      "loss": 0.9733,
+      "step": 2445
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12070074886663928,
+      "learning_rate": 8.645952439056412e-07,
+      "loss": 0.9693,
+      "step": 2450
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13227346913782864,
+      "learning_rate": 8.638241052291212e-07,
+      "loss": 0.9535,
+      "step": 2455
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1274427599313127,
+      "learning_rate": 8.630511229423667e-07,
+      "loss": 0.9424,
+      "step": 2460
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13336060715091919,
+      "learning_rate": 8.622763009623202e-07,
+      "loss": 0.9608,
+      "step": 2465
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1388725575082033,
+      "learning_rate": 8.61499643215246e-07,
+      "loss": 0.9822,
+      "step": 2470
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12759910426866902,
+      "learning_rate": 8.607211536367116e-07,
+      "loss": 0.9519,
+      "step": 2475
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1186229590229591,
+      "learning_rate": 8.599408361715661e-07,
+      "loss": 0.9467,
+      "step": 2480
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.13149744238453148,
+      "learning_rate": 8.59158694773922e-07,
+      "loss": 0.9587,
+      "step": 2485
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12895156644202038,
+      "learning_rate": 8.583747334071334e-07,
+      "loss": 0.9562,
+      "step": 2490
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12857345006475474,
+      "learning_rate": 8.575889560437772e-07,
+      "loss": 0.9172,
+      "step": 2495
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.1232075425270835,
+      "learning_rate": 8.568013666656325e-07,
+      "loss": 0.9681,
+      "step": 2500
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.11636336633891423,
+      "learning_rate": 8.5601196926366e-07,
+      "loss": 0.9291,
+      "step": 2505
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12439182003994874,
+      "learning_rate": 8.552207678379832e-07,
+      "loss": 0.973,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12538810205254744,
+      "learning_rate": 8.54427766397866e-07,
+      "loss": 0.9877,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.12147536548689451,
+      "learning_rate": 8.536329689616943e-07,
+      "loss": 0.9593,
+      "step": 2520
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1335403459002424,
+      "learning_rate": 8.528363795569548e-07,
+      "loss": 0.9643,
+      "step": 2525
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12789604459057055,
+      "learning_rate": 8.520380022202141e-07,
+      "loss": 0.9694,
+      "step": 2530
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13179360034868276,
+      "learning_rate": 8.512378409970995e-07,
+      "loss": 0.9408,
+      "step": 2535
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1324972270894786,
+      "learning_rate": 8.504358999422776e-07,
+      "loss": 0.9589,
+      "step": 2540
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.11994295282967923,
+      "learning_rate": 8.496321831194336e-07,
+      "loss": 0.9402,
+      "step": 2545
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12029593183235637,
+      "learning_rate": 8.488266946012516e-07,
+      "loss": 0.9529,
+      "step": 2550
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.11917481891064877,
+      "learning_rate": 8.48019438469393e-07,
+      "loss": 0.9085,
+      "step": 2555
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1274216769254125,
+      "learning_rate": 8.472104188144766e-07,
+      "loss": 0.9611,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12866807928843538,
+      "learning_rate": 8.463996397360574e-07,
+      "loss": 0.9444,
+      "step": 2565
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13044453446529503,
+      "learning_rate": 8.455871053426056e-07,
+      "loss": 0.9491,
+      "step": 2570
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12233813334906042,
+      "learning_rate": 8.447728197514867e-07,
+      "loss": 0.9673,
+      "step": 2575
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1275771155355058,
+      "learning_rate": 8.439567870889399e-07,
+      "loss": 0.9635,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.12313878069336046,
+      "learning_rate": 8.431390114900569e-07,
+      "loss": 0.97,
+      "step": 2585
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.1327981257411731,
+      "learning_rate": 8.423194970987622e-07,
+      "loss": 0.9467,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.13514695117532505,
+      "learning_rate": 8.414982480677905e-07,
+      "loss": 0.9176,
+      "step": 2595
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12163202912591402,
+      "learning_rate": 8.40675268558667e-07,
+      "loss": 0.9762,
+      "step": 2600
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12286220242126533,
+      "learning_rate": 8.398505627416855e-07,
+      "loss": 0.9527,
+      "step": 2605
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1259083134108581,
+      "learning_rate": 8.390241347958877e-07,
+      "loss": 0.9645,
+      "step": 2610
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.11792059246211607,
+      "learning_rate": 8.381959889090417e-07,
+      "loss": 0.9534,
+      "step": 2615
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13136289068169052,
+      "learning_rate": 8.373661292776214e-07,
+      "loss": 0.9653,
+      "step": 2620
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1354661197279732,
+      "learning_rate": 8.365345601067841e-07,
+      "loss": 0.9683,
+      "step": 2625
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12852555122549478,
+      "learning_rate": 8.357012856103501e-07,
+      "loss": 0.9594,
+      "step": 2630
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12849992064342664,
+      "learning_rate": 8.348663100107816e-07,
+      "loss": 0.9981,
+      "step": 2635
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1279689300198508,
+      "learning_rate": 8.3402963753916e-07,
+      "loss": 0.9934,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.11816408569787522,
+      "learning_rate": 8.33191272435166e-07,
+      "loss": 0.9689,
+      "step": 2645
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1142131609144541,
+      "learning_rate": 8.32351218947057e-07,
+      "loss": 0.9989,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12216288478603315,
+      "learning_rate": 8.31509481331646e-07,
+      "loss": 0.9756,
+      "step": 2655
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.11589937923595327,
+      "learning_rate": 8.306660638542802e-07,
+      "loss": 0.9442,
+      "step": 2660
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.13166599879526164,
+      "learning_rate": 8.29820970788819e-07,
+      "loss": 0.9661,
+      "step": 2665
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.1268144052164066,
+      "learning_rate": 8.289742064176128e-07,
+      "loss": 0.9548,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.12788882934785634,
+      "learning_rate": 8.281257750314805e-07,
+      "loss": 0.9266,
+      "step": 2675
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13229414625103106,
+      "learning_rate": 8.272756809296887e-07,
+      "loss": 0.928,
+      "step": 2680
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.11747212647242553,
+      "learning_rate": 8.264239284199295e-07,
+      "loss": 0.95,
+      "step": 2685
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13379273157137722,
+      "learning_rate": 8.255705218182983e-07,
+      "loss": 0.9565,
+      "step": 2690
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1396695834143499,
+      "learning_rate": 8.24715465449273e-07,
+      "loss": 0.9602,
+      "step": 2695
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12245905752091622,
+      "learning_rate": 8.238587636456904e-07,
+      "loss": 0.9776,
+      "step": 2700
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12457129265827753,
+      "learning_rate": 8.230004207487263e-07,
+      "loss": 0.9404,
+      "step": 2705
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12650803624149431,
+      "learning_rate": 8.221404411078716e-07,
+      "loss": 0.9567,
+      "step": 2710
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1369001591120179,
+      "learning_rate": 8.212788290809116e-07,
+      "loss": 0.9613,
+      "step": 2715
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12932221492384852,
+      "learning_rate": 8.204155890339032e-07,
+      "loss": 0.9874,
+      "step": 2720
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12323953290234287,
+      "learning_rate": 8.195507253411527e-07,
+      "loss": 0.9607,
+      "step": 2725
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.1226432701531176,
+      "learning_rate": 8.186842423851946e-07,
+      "loss": 0.9652,
+      "step": 2730
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.13131615770228897,
+      "learning_rate": 8.178161445567681e-07,
+      "loss": 0.9461,
+      "step": 2735
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12776292090024383,
+      "learning_rate": 8.169464362547959e-07,
+      "loss": 0.9918,
+      "step": 2740
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12733009344534835,
+      "learning_rate": 8.160751218863611e-07,
+      "loss": 0.9095,
+      "step": 2745
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.12344217388603404,
+      "learning_rate": 8.152022058666854e-07,
+      "loss": 0.9653,
+      "step": 2750
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12036219918993828,
+      "learning_rate": 8.143276926191067e-07,
+      "loss": 0.9745,
+      "step": 2755
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12474505981797819,
+      "learning_rate": 8.134515865750563e-07,
+      "loss": 0.9253,
+      "step": 2760
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13110547921964624,
+      "learning_rate": 8.125738921740368e-07,
+      "loss": 0.9121,
+      "step": 2765
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13168309709339523,
+      "learning_rate": 8.116946138635997e-07,
+      "loss": 0.9259,
+      "step": 2770
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1379149989908605,
+      "learning_rate": 8.108137560993224e-07,
+      "loss": 0.941,
+      "step": 2775
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1264243641219311,
+      "learning_rate": 8.09931323344786e-07,
+      "loss": 0.9383,
+      "step": 2780
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13604437500700384,
+      "learning_rate": 8.090473200715524e-07,
+      "loss": 0.9706,
+      "step": 2785
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12011404997230099,
+      "learning_rate": 8.081617507591421e-07,
+      "loss": 0.9628,
+      "step": 2790
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13429788442946078,
+      "learning_rate": 8.072746198950109e-07,
+      "loss": 0.9821,
+      "step": 2795
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13518027656863413,
+      "learning_rate": 8.063859319745278e-07,
+      "loss": 0.9483,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.1209960307443581,
+      "learning_rate": 8.054956915009517e-07,
+      "loss": 0.9403,
+      "step": 2805
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12171245590852015,
+      "learning_rate": 8.046039029854085e-07,
+      "loss": 0.9747,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12699857176550366,
+      "learning_rate": 8.037105709468693e-07,
+      "loss": 0.9108,
+      "step": 2815
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.12368251260434174,
+      "learning_rate": 8.028156999121258e-07,
+      "loss": 0.9431,
+      "step": 2820
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.14111865299159115,
+      "learning_rate": 8.019192944157688e-07,
+      "loss": 0.9376,
+      "step": 2825
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.13068319360232591,
+      "learning_rate": 8.010213590001647e-07,
+      "loss": 0.9498,
+      "step": 2830
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13434787759306582,
+      "learning_rate": 8.001218982154319e-07,
+      "loss": 0.9722,
+      "step": 2835
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1233812847154311,
+      "learning_rate": 7.992209166194196e-07,
+      "loss": 0.9544,
+      "step": 2840
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12483817861211521,
+      "learning_rate": 7.98318418777682e-07,
+      "loss": 0.9401,
+      "step": 2845
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13009026512758384,
+      "learning_rate": 7.974144092634572e-07,
+      "loss": 0.951,
+      "step": 2850
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13697450210988552,
+      "learning_rate": 7.965088926576436e-07,
+      "loss": 0.973,
+      "step": 2855
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12585670143765343,
+      "learning_rate": 7.956018735487762e-07,
+      "loss": 0.9644,
+      "step": 2860
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.11488518133175933,
+      "learning_rate": 7.946933565330039e-07,
+      "loss": 0.9579,
+      "step": 2865
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1314584613462125,
+      "learning_rate": 7.937833462140657e-07,
+      "loss": 0.9384,
+      "step": 2870
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13574344595441162,
+      "learning_rate": 7.928718472032678e-07,
+      "loss": 0.9899,
+      "step": 2875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13080842686925537,
+      "learning_rate": 7.919588641194601e-07,
+      "loss": 0.914,
+      "step": 2880
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13034085175340068,
+      "learning_rate": 7.910444015890128e-07,
+      "loss": 0.9403,
+      "step": 2885
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13809170816721889,
+      "learning_rate": 7.901284642457921e-07,
+      "loss": 0.9408,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.12173600823417502,
+      "learning_rate": 7.892110567311392e-07,
+      "loss": 0.9285,
+      "step": 2895
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1464444063361684,
+      "learning_rate": 7.882921836938436e-07,
+      "loss": 0.9463,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.13147064300694333,
+      "learning_rate": 7.873718497901218e-07,
+      "loss": 0.9513,
+      "step": 2905
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1210225120610158,
+      "learning_rate": 7.864500596835925e-07,
+      "loss": 0.9688,
+      "step": 2910
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1326634527524828,
+      "learning_rate": 7.855268180452542e-07,
+      "loss": 0.9177,
+      "step": 2915
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13542168248570954,
+      "learning_rate": 7.8460212955346e-07,
+      "loss": 0.9521,
+      "step": 2920
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1323389287600922,
+      "learning_rate": 7.836759988938949e-07,
+      "loss": 0.9637,
+      "step": 2925
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13060800687982715,
+      "learning_rate": 7.82748430759552e-07,
+      "loss": 0.9786,
+      "step": 2930
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12657134041895668,
+      "learning_rate": 7.818194298507082e-07,
+      "loss": 0.9671,
+      "step": 2935
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1330575716919685,
+      "learning_rate": 7.80889000874901e-07,
+      "loss": 0.9912,
+      "step": 2940
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12059217256694675,
+      "learning_rate": 7.799571485469043e-07,
+      "loss": 0.9403,
+      "step": 2945
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13233889397056475,
+      "learning_rate": 7.790238775887046e-07,
+      "loss": 0.9688,
+      "step": 2950
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1316622948052838,
+      "learning_rate": 7.780891927294768e-07,
+      "loss": 0.9379,
+      "step": 2955
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.1272488711325029,
+      "learning_rate": 7.771530987055608e-07,
+      "loss": 0.9584,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13293517325189286,
+      "learning_rate": 7.762156002604369e-07,
+      "loss": 0.9665,
+      "step": 2965
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.13100693264694405,
+      "learning_rate": 7.752767021447022e-07,
+      "loss": 0.9581,
+      "step": 2970
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12423910087713855,
+      "learning_rate": 7.743364091160465e-07,
+      "loss": 0.9608,
+      "step": 2975
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12921295985843698,
+      "learning_rate": 7.733947259392277e-07,
+      "loss": 0.9578,
+      "step": 2980
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.12495255994471571,
+      "learning_rate": 7.724516573860483e-07,
+      "loss": 0.9726,
+      "step": 2985
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13114636136234115,
+      "learning_rate": 7.715072082353308e-07,
+      "loss": 0.9833,
+      "step": 2990
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.14363179538499532,
+      "learning_rate": 7.705613832728937e-07,
+      "loss": 0.9844,
+      "step": 2995
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.137970930067913,
+      "learning_rate": 7.696141872915269e-07,
+      "loss": 0.941,
+      "step": 3000
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13639060865228242,
+      "learning_rate": 7.686656250909686e-07,
+      "loss": 0.9404,
+      "step": 3005
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12106145306222679,
+      "learning_rate": 7.677157014778786e-07,
+      "loss": 0.9467,
+      "step": 3010
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13221511649231485,
+      "learning_rate": 7.667644212658164e-07,
+      "loss": 0.9253,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1248192690504065,
+      "learning_rate": 7.658117892752159e-07,
+      "loss": 1.0083,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13567821488062623,
+      "learning_rate": 7.648578103333602e-07,
+      "loss": 0.9582,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.12559276597971622,
+      "learning_rate": 7.639024892743581e-07,
+      "loss": 0.9397,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1260127186466067,
+      "learning_rate": 7.629458309391199e-07,
+      "loss": 0.9418,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1263444037067101,
+      "learning_rate": 7.619878401753315e-07,
+      "loss": 0.9634,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.1295773150815806,
+      "learning_rate": 7.61028521837431e-07,
+      "loss": 0.9602,
+      "step": 3045
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13812984343744542,
+      "learning_rate": 7.600678807865835e-07,
+      "loss": 0.9721,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13263439713363698,
+      "learning_rate": 7.591059218906571e-07,
+      "loss": 0.9688,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.13148062912356961,
+      "learning_rate": 7.581426500241975e-07,
+      "loss": 0.9436,
+      "step": 3060
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12869119312285332,
+      "learning_rate": 7.571780700684038e-07,
+      "loss": 0.9726,
+      "step": 3065
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13378214633892332,
+      "learning_rate": 7.562121869111036e-07,
+      "loss": 0.9402,
+      "step": 3070
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13530973447558917,
+      "learning_rate": 7.552450054467279e-07,
+      "loss": 0.9666,
+      "step": 3075
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1343285161158426,
+      "learning_rate": 7.54276530576287e-07,
+      "loss": 0.9215,
+      "step": 3080
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.1289485361084503,
+      "learning_rate": 7.533067672073453e-07,
+      "loss": 0.9439,
+      "step": 3085
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12630366893375639,
+      "learning_rate": 7.523357202539961e-07,
+      "loss": 0.9863,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13397743413171725,
+      "learning_rate": 7.513633946368376e-07,
+      "loss": 0.9368,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13567942165764624,
+      "learning_rate": 7.503897952829467e-07,
+      "loss": 0.9436,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12755114849987834,
+      "learning_rate": 7.494149271258549e-07,
+      "loss": 0.9478,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12994334889464337,
+      "learning_rate": 7.484387951055237e-07,
+      "loss": 0.9283,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12412019789679693,
+      "learning_rate": 7.474614041683182e-07,
+      "loss": 0.9619,
+      "step": 3115
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.13799461076049485,
+      "learning_rate": 7.464827592669832e-07,
+      "loss": 0.9652,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.11947852095618566,
+      "learning_rate": 7.45502865360618e-07,
+      "loss": 0.9479,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12585173226324806,
+      "learning_rate": 7.445217274146502e-07,
+      "loss": 0.9649,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.14980148075275915,
+      "learning_rate": 7.435393504008124e-07,
+      "loss": 0.9252,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.12980724753796014,
+      "learning_rate": 7.425557392971149e-07,
+      "loss": 0.941,
+      "step": 3140
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.14040255281074032,
+      "learning_rate": 7.415708990878222e-07,
+      "loss": 0.9467,
+      "step": 3145
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13306755370627207,
+      "learning_rate": 7.405848347634268e-07,
+      "loss": 0.9398,
+      "step": 3150
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13650987173279677,
+      "learning_rate": 7.395975513206244e-07,
+      "loss": 0.9591,
+      "step": 3155
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13113039131383386,
+      "learning_rate": 7.38609053762288e-07,
+      "loss": 0.9392,
+      "step": 3160
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13325765862509942,
+      "learning_rate": 7.376193470974432e-07,
+      "loss": 0.9837,
+      "step": 3165
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.14225587837837686,
+      "learning_rate": 7.366284363412422e-07,
+      "loss": 0.9737,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1357989069941775,
+      "learning_rate": 7.356363265149391e-07,
+      "loss": 0.949,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13682740794381917,
+      "learning_rate": 7.346430226458639e-07,
+      "loss": 0.926,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12373039131723906,
+      "learning_rate": 7.33648529767397e-07,
+      "loss": 0.9533,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13583459356225527,
+      "learning_rate": 7.326528529189442e-07,
+      "loss": 0.9703,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13269131982818066,
+      "learning_rate": 7.316559971459106e-07,
+      "loss": 0.9491,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13561739013339438,
+      "learning_rate": 7.306579674996754e-07,
+      "loss": 0.9401,
+      "step": 3200
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.12689039560629126,
+      "learning_rate": 7.296587690375661e-07,
+      "loss": 0.9339,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.1509279629269088,
+      "learning_rate": 7.286584068228332e-07,
+      "loss": 0.9862,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.13215306021832438,
+      "learning_rate": 7.276568859246237e-07,
+      "loss": 0.9484,
+      "step": 3215
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13612738238898625,
+      "learning_rate": 7.266542114179567e-07,
+      "loss": 0.976,
+      "step": 3220
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12100484525886428,
+      "learning_rate": 7.256503883836968e-07,
+      "loss": 0.9328,
+      "step": 3225
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.11829929573438083,
+      "learning_rate": 7.246454219085283e-07,
+      "loss": 0.9515,
+      "step": 3230
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1372109053553825,
+      "learning_rate": 7.236393170849297e-07,
+      "loss": 0.9747,
+      "step": 3235
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13636098232877802,
+      "learning_rate": 7.226320790111482e-07,
+      "loss": 0.9619,
+      "step": 3240
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.14263069218636176,
+      "learning_rate": 7.216237127911733e-07,
+      "loss": 0.9645,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1267116547244488,
+      "learning_rate": 7.206142235347111e-07,
+      "loss": 0.9625,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.1721735725134802,
+      "learning_rate": 7.196036163571586e-07,
+      "loss": 0.942,
+      "step": 3255
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13662632560462493,
+      "learning_rate": 7.185918963795776e-07,
+      "loss": 0.9254,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13813446535464022,
+      "learning_rate": 7.175790687286689e-07,
+      "loss": 0.9292,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13234427891092346,
+      "learning_rate": 7.165651385367462e-07,
+      "loss": 0.9449,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13966175855812854,
+      "learning_rate": 7.155501109417098e-07,
+      "loss": 0.9226,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.14131851420104063,
+      "learning_rate": 7.145339910870214e-07,
+      "loss": 0.9407,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13669120694546202,
+      "learning_rate": 7.135167841216774e-07,
+      "loss": 0.9336,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.12658907757361437,
+      "learning_rate": 7.124984952001825e-07,
+      "loss": 0.9172,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.13387465379432154,
+      "learning_rate": 7.114791294825244e-07,
+      "loss": 0.9397,
+      "step": 3295
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13945877492400122,
+      "learning_rate": 7.104586921341472e-07,
+      "loss": 0.9448,
+      "step": 3300
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13167897739756673,
+      "learning_rate": 7.094371883259255e-07,
+      "loss": 0.9664,
+      "step": 3305
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13069413861865914,
+      "learning_rate": 7.084146232341373e-07,
+      "loss": 0.9702,
+      "step": 3310
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12013350872490329,
+      "learning_rate": 7.073910020404393e-07,
+      "loss": 0.9594,
+      "step": 3315
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13130889020177322,
+      "learning_rate": 7.063663299318392e-07,
+      "loss": 0.9658,
+      "step": 3320
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13647631612496072,
+      "learning_rate": 7.053406121006703e-07,
+      "loss": 0.9563,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13013799660423364,
+      "learning_rate": 7.043138537445648e-07,
+      "loss": 0.9571,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12822953513626817,
+      "learning_rate": 7.032860600664274e-07,
+      "loss": 0.9371,
+      "step": 3335
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13950345796181055,
+      "learning_rate": 7.022572362744096e-07,
+      "loss": 0.9106,
+      "step": 3340
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13268238587388465,
+      "learning_rate": 7.012273875818823e-07,
+      "loss": 0.9589,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.1381359099890364,
+      "learning_rate": 7.001965192074102e-07,
+      "loss": 0.9429,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12641967093193185,
+      "learning_rate": 6.991646363747248e-07,
+      "loss": 0.9503,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.12427125424661205,
+      "learning_rate": 6.981317443126983e-07,
+      "loss": 0.9256,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13059208215502052,
+      "learning_rate": 6.97097848255317e-07,
+      "loss": 0.9506,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.13764370334445172,
+      "learning_rate": 6.960629534416549e-07,
+      "loss": 0.9341,
+      "step": 3370
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13143989175588544,
+      "learning_rate": 6.950270651158466e-07,
+      "loss": 0.9043,
+      "step": 3375
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13670511139307948,
+      "learning_rate": 6.939901885270616e-07,
+      "loss": 0.9263,
+      "step": 3380
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13974854656434085,
+      "learning_rate": 6.929523289294769e-07,
+      "loss": 0.9289,
+      "step": 3385
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12755375697965024,
+      "learning_rate": 6.919134915822511e-07,
+      "loss": 0.9527,
+      "step": 3390
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.11921526210954565,
+      "learning_rate": 6.908736817494966e-07,
+      "loss": 0.9599,
+      "step": 3395
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.14308942393005877,
+      "learning_rate": 6.898329047002547e-07,
+      "loss": 0.9508,
+      "step": 3400
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1451110197685494,
+      "learning_rate": 6.887911657084669e-07,
+      "loss": 0.9394,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.1389288903877239,
+      "learning_rate": 6.8774847005295e-07,
+      "loss": 0.9415,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13821507736267624,
+      "learning_rate": 6.867048230173681e-07,
+      "loss": 0.9386,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13322887601669162,
+      "learning_rate": 6.856602298902061e-07,
+      "loss": 0.9436,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12556294877285898,
+      "learning_rate": 6.846146959647434e-07,
+      "loss": 0.9373,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.12770041296260315,
+      "learning_rate": 6.835682265390266e-07,
+      "loss": 0.9401,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13258742299068157,
+      "learning_rate": 6.825208269158425e-07,
+      "loss": 0.9441,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13476450384346994,
+      "learning_rate": 6.814725024026918e-07,
+      "loss": 0.9513,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13084876370035464,
+      "learning_rate": 6.80423258311762e-07,
+      "loss": 0.9318,
+      "step": 3445
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.13482812789178292,
+      "learning_rate": 6.793730999599e-07,
+      "loss": 0.9408,
+      "step": 3450
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13731488104864878,
+      "learning_rate": 6.783220326685859e-07,
+      "loss": 0.9387,
+      "step": 3455
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13343809673146842,
+      "learning_rate": 6.772700617639054e-07,
+      "loss": 0.9822,
+      "step": 3460
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13242290029347972,
+      "learning_rate": 6.762171925765234e-07,
+      "loss": 0.9354,
+      "step": 3465
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1253689199838533,
+      "learning_rate": 6.751634304416566e-07,
+      "loss": 0.9646,
+      "step": 3470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1342988575647191,
+      "learning_rate": 6.741087806990462e-07,
+      "loss": 0.9253,
+      "step": 3475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13424606309812645,
+      "learning_rate": 6.730532486929314e-07,
+      "loss": 0.9276,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12874324455997171,
+      "learning_rate": 6.719968397720223e-07,
+      "loss": 0.9262,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13603689734290478,
+      "learning_rate": 6.709395592894724e-07,
+      "loss": 0.9196,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13836561829500088,
+      "learning_rate": 6.698814126028518e-07,
+      "loss": 0.9899,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.14114382755478022,
+      "learning_rate": 6.688224050741198e-07,
+      "loss": 0.9428,
+      "step": 3500
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.1282470657903545,
+      "learning_rate": 6.677625420695978e-07,
+      "loss": 0.9579,
+      "step": 3505
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.12887683347152237,
+      "learning_rate": 6.667018289599424e-07,
+      "loss": 0.9389,
+      "step": 3510
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13626620843340756,
+      "learning_rate": 6.656402711201177e-07,
+      "loss": 0.9694,
+      "step": 3515
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13894117846209475,
+      "learning_rate": 6.645778739293686e-07,
+      "loss": 0.946,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.13432405155856159,
+      "learning_rate": 6.635146427711928e-07,
+      "loss": 0.9929,
+      "step": 3525
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.15040709672573466,
+      "learning_rate": 6.624505830333144e-07,
+      "loss": 0.93,
+      "step": 3530
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13932183398240652,
+      "learning_rate": 6.613857001076561e-07,
+      "loss": 0.9759,
+      "step": 3535
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12796432636317873,
+      "learning_rate": 6.603199993903119e-07,
+      "loss": 0.9559,
+      "step": 3540
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13237601573240831,
+      "learning_rate": 6.592534862815196e-07,
+      "loss": 0.9536,
+      "step": 3545
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13698037216563516,
+      "learning_rate": 6.581861661856338e-07,
+      "loss": 0.9378,
+      "step": 3550
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12335498745818481,
+      "learning_rate": 6.571180445110986e-07,
+      "loss": 0.9594,
+      "step": 3555
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13531056198332997,
+      "learning_rate": 6.560491266704194e-07,
+      "loss": 0.9522,
+      "step": 3560
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.14284854112075682,
+      "learning_rate": 6.549794180801365e-07,
+      "loss": 0.9388,
+      "step": 3565
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.12488046273890678,
+      "learning_rate": 6.539089241607972e-07,
+      "loss": 0.9682,
+      "step": 3570
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1375231463166149,
+      "learning_rate": 6.528376503369279e-07,
+      "loss": 0.9333,
+      "step": 3575
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.135362237831506,
+      "learning_rate": 6.517656020370074e-07,
+      "loss": 0.9299,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13529273905961453,
+      "learning_rate": 6.506927846934386e-07,
+      "loss": 0.9454,
+      "step": 3585
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13620178013016204,
+      "learning_rate": 6.49619203742522e-07,
+      "loss": 0.9214,
+      "step": 3590
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1289367370767514,
+      "learning_rate": 6.485448646244268e-07,
+      "loss": 0.9702,
+      "step": 3595
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.13151145282054438,
+      "learning_rate": 6.474697727831647e-07,
+      "loss": 0.9594,
+      "step": 3600
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.1354149074627231,
+      "learning_rate": 6.463939336665612e-07,
+      "loss": 0.9345,
+      "step": 3605
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1275620944261948,
+      "learning_rate": 6.453173527262291e-07,
+      "loss": 0.9384,
+      "step": 3610
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13732127946781197,
+      "learning_rate": 6.442400354175396e-07,
+      "loss": 0.9621,
+      "step": 3615
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.14254969349347904,
+      "learning_rate": 6.431619871995955e-07,
+      "loss": 0.9645,
+      "step": 3620
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1271792290680745,
+      "learning_rate": 6.420832135352035e-07,
+      "loss": 1.0083,
+      "step": 3625
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13634710315164617,
+      "learning_rate": 6.410037198908465e-07,
+      "loss": 0.9328,
+      "step": 3630
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12793099886724055,
+      "learning_rate": 6.399235117366551e-07,
+      "loss": 0.9372,
+      "step": 3635
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12399023297130307,
+      "learning_rate": 6.388425945463811e-07,
+      "loss": 0.9099,
+      "step": 3640
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13368900108332116,
+      "learning_rate": 6.377609737973693e-07,
+      "loss": 0.9279,
+      "step": 3645
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.12779154144482824,
+      "learning_rate": 6.366786549705296e-07,
+      "loss": 0.9569,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13468037542978326,
+      "learning_rate": 6.355956435503086e-07,
+      "loss": 0.9492,
+      "step": 3655
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13850631596265417,
+      "learning_rate": 6.345119450246632e-07,
+      "loss": 0.9903,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1340417045457032,
+      "learning_rate": 6.33427564885032e-07,
+      "loss": 0.9405,
+      "step": 3665
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13571904191102993,
+      "learning_rate": 6.323425086263077e-07,
+      "loss": 0.943,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.1397683934682841,
+      "learning_rate": 6.312567817468081e-07,
+      "loss": 0.9519,
+      "step": 3675
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.13257268547547574,
+      "learning_rate": 6.301703897482506e-07,
+      "loss": 0.9541,
+      "step": 3680
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12908145084066225,
+      "learning_rate": 6.29083338135722e-07,
+      "loss": 0.9517,
+      "step": 3685
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12815959139643793,
+      "learning_rate": 6.279956324176522e-07,
+      "loss": 0.9295,
+      "step": 3690
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1327573413234747,
+      "learning_rate": 6.269072781057854e-07,
+      "loss": 0.9604,
+      "step": 3695
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13518149044456693,
+      "learning_rate": 6.258182807151522e-07,
+      "loss": 0.9381,
+      "step": 3700
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.14047426063629007,
+      "learning_rate": 6.24728645764042e-07,
+      "loss": 0.9433,
+      "step": 3705
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13312121535562066,
+      "learning_rate": 6.236383787739754e-07,
+      "loss": 0.9502,
+      "step": 3710
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13492442787811681,
+      "learning_rate": 6.225474852696749e-07,
+      "loss": 0.9872,
+      "step": 3715
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1407600667641802,
+      "learning_rate": 6.214559707790385e-07,
+      "loss": 0.9577,
+      "step": 3720
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.14458729672487486,
+      "learning_rate": 6.203638408331104e-07,
+      "loss": 0.9067,
+      "step": 3725
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1342835496564921,
+      "learning_rate": 6.192711009660539e-07,
+      "loss": 0.9408,
+      "step": 3730
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13664873761412177,
+      "learning_rate": 6.181777567151228e-07,
+      "loss": 0.9383,
+      "step": 3735
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.15296020596616552,
+      "learning_rate": 6.170838136206332e-07,
+      "loss": 0.9482,
+      "step": 3740
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13159404228487726,
+      "learning_rate": 6.159892772259363e-07,
+      "loss": 0.9607,
+      "step": 3745
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.12984642148178688,
+      "learning_rate": 6.148941530773894e-07,
+      "loss": 0.9224,
+      "step": 3750
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.1316104262011331,
+      "learning_rate": 6.13798446724328e-07,
+      "loss": 0.9676,
+      "step": 3755
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.13107396614566502,
+      "learning_rate": 6.127021637190383e-07,
+      "loss": 0.9467,
+      "step": 3760
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.14577195732936218,
+      "learning_rate": 6.11605309616728e-07,
+      "loss": 0.9699,
+      "step": 3765
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13464014992920087,
+      "learning_rate": 6.105078899754991e-07,
+      "loss": 0.9277,
+      "step": 3770
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.12964735203321268,
+      "learning_rate": 6.094099103563191e-07,
+      "loss": 0.9125,
+      "step": 3775
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13567814037631482,
+      "learning_rate": 6.083113763229936e-07,
+      "loss": 0.9334,
+      "step": 3780
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.12985783697563005,
+      "learning_rate": 6.072122934421369e-07,
+      "loss": 0.9417,
+      "step": 3785
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1433244163139185,
+      "learning_rate": 6.061126672831449e-07,
+      "loss": 0.9522,
+      "step": 3790
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13388217668861677,
+      "learning_rate": 6.050125034181664e-07,
+      "loss": 0.9458,
+      "step": 3795
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13962762155324338,
+      "learning_rate": 6.039118074220749e-07,
+      "loss": 0.9306,
+      "step": 3800
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.13775535745056688,
+      "learning_rate": 6.028105848724404e-07,
+      "loss": 0.943,
+      "step": 3805
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1348344511038057,
+      "learning_rate": 6.017088413495009e-07,
+      "loss": 0.9383,
+      "step": 3810
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.14339659440983926,
+      "learning_rate": 6.006065824361349e-07,
+      "loss": 0.9373,
+      "step": 3815
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1336733837299364,
+      "learning_rate": 5.995038137178318e-07,
+      "loss": 0.9357,
+      "step": 3820
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.1396344951106276,
+      "learning_rate": 5.984005407826649e-07,
+      "loss": 0.9233,
+      "step": 3825
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.14178631935559058,
+      "learning_rate": 5.972967692212621e-07,
+      "loss": 0.9202,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.14078795292605772,
+      "learning_rate": 5.961925046267784e-07,
+      "loss": 0.9593,
+      "step": 3835
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.133807513361193,
+      "learning_rate": 5.950877525948669e-07,
+      "loss": 0.9705,
+      "step": 3840
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1441077869585748,
+      "learning_rate": 5.939825187236508e-07,
+      "loss": 0.9525,
+      "step": 3845
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1376659675460509,
+      "learning_rate": 5.928768086136949e-07,
+      "loss": 0.9681,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13087524389317895,
+      "learning_rate": 5.917706278679773e-07,
+      "loss": 0.9497,
+      "step": 3855
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1445351941638308,
+      "learning_rate": 5.906639820918607e-07,
+      "loss": 0.9475,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13563713054228066,
+      "learning_rate": 5.895568768930647e-07,
+      "loss": 0.9355,
+      "step": 3865
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.12392608402478489,
+      "learning_rate": 5.884493178816366e-07,
+      "loss": 0.9579,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14758308387309926,
+      "learning_rate": 5.873413106699236e-07,
+      "loss": 0.9215,
+      "step": 3875
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1443124044135253,
+      "learning_rate": 5.862328608725434e-07,
+      "loss": 0.9596,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14429982548565487,
+      "learning_rate": 5.851239741063574e-07,
+      "loss": 0.9757,
+      "step": 3885
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13931310141494072,
+      "learning_rate": 5.840146559904404e-07,
+      "loss": 0.9264,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.13883439197158579,
+      "learning_rate": 5.829049121460537e-07,
+      "loss": 0.9516,
+      "step": 3895
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.12907862124909195,
+      "learning_rate": 5.817947481966152e-07,
+      "loss": 0.918,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.15574290919549133,
+      "learning_rate": 5.806841697676721e-07,
+      "loss": 0.9588,
+      "step": 3905
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14166372520356102,
+      "learning_rate": 5.795731824868715e-07,
+      "loss": 0.922,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.1398585038868848,
+      "learning_rate": 5.78461791983933e-07,
+      "loss": 0.9183,
+      "step": 3915
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13715309799760672,
+      "learning_rate": 5.773500038906186e-07,
+      "loss": 0.9695,
+      "step": 3920
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1314419624457136,
+      "learning_rate": 5.762378238407055e-07,
+      "loss": 0.9336,
+      "step": 3925
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13742705154600574,
+      "learning_rate": 5.75125257469957e-07,
+      "loss": 0.9617,
+      "step": 3930
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1404427920437888,
+      "learning_rate": 5.74012310416094e-07,
+      "loss": 0.9317,
+      "step": 3935
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13437786429067455,
+      "learning_rate": 5.728989883187661e-07,
+      "loss": 0.9297,
+      "step": 3940
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.14135609824728507,
+      "learning_rate": 5.71785296819524e-07,
+      "loss": 0.9236,
+      "step": 3945
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.15196009953822173,
+      "learning_rate": 5.706712415617896e-07,
+      "loss": 0.9524,
+      "step": 3950
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13063431800367908,
+      "learning_rate": 5.695568281908289e-07,
+      "loss": 0.9226,
+      "step": 3955
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13219294595838113,
+      "learning_rate": 5.684420623537217e-07,
+      "loss": 0.9325,
+      "step": 3960
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.14257086537369668,
+      "learning_rate": 5.673269496993342e-07,
+      "loss": 0.9601,
+      "step": 3965
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1315676021445299,
+      "learning_rate": 5.662114958782904e-07,
+      "loss": 0.9287,
+      "step": 3970
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13103011334739176,
+      "learning_rate": 5.650957065429427e-07,
+      "loss": 0.9437,
+      "step": 3975
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.14320766682312397,
+      "learning_rate": 5.639795873473434e-07,
+      "loss": 0.949,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1361028352104983,
+      "learning_rate": 5.628631439472169e-07,
+      "loss": 0.9244,
+      "step": 3985
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.13799372110579064,
+      "learning_rate": 5.617463819999302e-07,
+      "loss": 0.9623,
+      "step": 3990
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.15190521516266728,
+      "learning_rate": 5.606293071644646e-07,
+      "loss": 0.9407,
+      "step": 3995
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.14601355573577898,
+      "learning_rate": 5.595119251013862e-07,
+      "loss": 0.9591,
+      "step": 4000
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.14983243939590704,
+      "learning_rate": 5.58394241472819e-07,
+      "loss": 0.9783,
+      "step": 4005
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1328863525735647,
+      "learning_rate": 5.572762619424144e-07,
+      "loss": 0.9357,
+      "step": 4010
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1363263559470564,
+      "learning_rate": 5.561579921753235e-07,
+      "loss": 0.923,
+      "step": 4015
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13036100443793086,
+      "learning_rate": 5.550394378381679e-07,
+      "loss": 0.9199,
+      "step": 4020
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.12970260477191972,
+      "learning_rate": 5.539206045990114e-07,
+      "loss": 0.9389,
+      "step": 4025
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.14124743236054874,
+      "learning_rate": 5.528014981273312e-07,
+      "loss": 0.9401,
+      "step": 4030
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13632340511779245,
+      "learning_rate": 5.51682124093989e-07,
+      "loss": 0.9222,
+      "step": 4035
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13751946089563966,
+      "learning_rate": 5.505624881712017e-07,
+      "loss": 0.9422,
+      "step": 4040
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13566609380787936,
+      "learning_rate": 5.494425960325143e-07,
+      "loss": 0.9262,
+      "step": 4045
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1371124339134709,
+      "learning_rate": 5.483224533527692e-07,
+      "loss": 0.9365,
+      "step": 4050
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.14508754073949348,
+      "learning_rate": 5.472020658080792e-07,
+      "loss": 0.9373,
+      "step": 4055
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13791380947091894,
+      "learning_rate": 5.460814390757972e-07,
+      "loss": 0.9459,
+      "step": 4060
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.13134916192663565,
+      "learning_rate": 5.449605788344884e-07,
+      "loss": 0.9457,
+      "step": 4065
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1285721209041786,
+      "learning_rate": 5.438394907639014e-07,
+      "loss": 0.9706,
+      "step": 4070
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14090764371110798,
+      "learning_rate": 5.427181805449394e-07,
+      "loss": 0.9244,
+      "step": 4075
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13733261206100944,
+      "learning_rate": 5.415966538596306e-07,
+      "loss": 0.9404,
+      "step": 4080
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14006797053834766,
+      "learning_rate": 5.404749163911009e-07,
+      "loss": 0.9558,
+      "step": 4085
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13446255780633895,
+      "learning_rate": 5.393529738235437e-07,
+      "loss": 0.9585,
+      "step": 4090
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13311649674371925,
+      "learning_rate": 5.382308318421923e-07,
+      "loss": 0.9476,
+      "step": 4095
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.1347618737418472,
+      "learning_rate": 5.3710849613329e-07,
+      "loss": 0.9354,
+      "step": 4100
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.12961560012932238,
+      "learning_rate": 5.35985972384062e-07,
+      "loss": 0.9388,
+      "step": 4105
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14424819900607866,
+      "learning_rate": 5.348632662826862e-07,
+      "loss": 0.976,
+      "step": 4110
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14449135043738937,
+      "learning_rate": 5.337403835182648e-07,
+      "loss": 0.9547,
+      "step": 4115
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14449217241803017,
+      "learning_rate": 5.326173297807947e-07,
+      "loss": 0.9827,
+      "step": 4120
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13901470053756623,
+      "learning_rate": 5.3149411076114e-07,
+      "loss": 0.9574,
+      "step": 4125
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13419923786788326,
+      "learning_rate": 5.303707321510016e-07,
+      "loss": 0.9576,
+      "step": 4130
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.14069250877034248,
+      "learning_rate": 5.292471996428893e-07,
+      "loss": 0.9654,
+      "step": 4135
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13874172749757024,
+      "learning_rate": 5.281235189300928e-07,
+      "loss": 0.935,
+      "step": 4140
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.13769044617234658,
+      "learning_rate": 5.269996957066531e-07,
+      "loss": 0.934,
+      "step": 4145
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1401544191964316,
+      "learning_rate": 5.258757356673327e-07,
+      "loss": 0.966,
+      "step": 4150
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1545905419405542,
+      "learning_rate": 5.24751644507588e-07,
+      "loss": 0.9535,
+      "step": 4155
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14202184318177583,
+      "learning_rate": 5.236274279235391e-07,
+      "loss": 0.9649,
+      "step": 4160
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1386812529388826,
+      "learning_rate": 5.225030916119426e-07,
+      "loss": 0.9618,
+      "step": 4165
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14921224711344191,
+      "learning_rate": 5.213786412701612e-07,
+      "loss": 0.9514,
+      "step": 4170
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13917821745751763,
+      "learning_rate": 5.202540825961354e-07,
+      "loss": 0.9602,
+      "step": 4175
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1359045923337555,
+      "learning_rate": 5.19129421288355e-07,
+      "loss": 0.9564,
+      "step": 4180
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13701563166264225,
+      "learning_rate": 5.180046630458296e-07,
+      "loss": 0.9423,
+      "step": 4185
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13063447716648116,
+      "learning_rate": 5.1687981356806e-07,
+      "loss": 0.9484,
+      "step": 4190
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14364732423031132,
+      "learning_rate": 5.157548785550097e-07,
+      "loss": 0.9374,
+      "step": 4195
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13887408742862523,
+      "learning_rate": 5.146298637070749e-07,
+      "loss": 0.9143,
+      "step": 4200
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14506429292209078,
+      "learning_rate": 5.135047747250568e-07,
+      "loss": 0.9415,
+      "step": 4205
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14039111836690363,
+      "learning_rate": 5.123796173101327e-07,
+      "loss": 0.9772,
+      "step": 4210
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.14139428433060355,
+      "learning_rate": 5.112543971638257e-07,
+      "loss": 0.9074,
+      "step": 4215
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.13359420720687015,
+      "learning_rate": 5.101291199879777e-07,
+      "loss": 0.9288,
+      "step": 4220
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.12956818923822536,
+      "learning_rate": 5.09003791484719e-07,
+      "loss": 0.9593,
+      "step": 4225
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13440406469968824,
+      "learning_rate": 5.0787841735644e-07,
+      "loss": 0.9244,
+      "step": 4230
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14024326967442205,
+      "learning_rate": 5.067530033057627e-07,
+      "loss": 0.9216,
+      "step": 4235
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1453797060582809,
+      "learning_rate": 5.056275550355111e-07,
+      "loss": 0.959,
+      "step": 4240
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1344477563243494,
+      "learning_rate": 5.045020782486826e-07,
+      "loss": 0.9938,
+      "step": 4245
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.135306833491894,
+      "learning_rate": 5.03376578648419e-07,
+      "loss": 0.9635,
+      "step": 4250
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1313707819187669,
+      "learning_rate": 5.022510619379785e-07,
+      "loss": 0.9133,
+      "step": 4255
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.15531740853981216,
+      "learning_rate": 5.011255338207048e-07,
+      "loss": 0.9608,
+      "step": 4260
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14872241048619764,
+      "learning_rate": 5e-07,
+      "loss": 0.9516,
+      "step": 4265
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14395371169475257,
+      "learning_rate": 4.988744661792954e-07,
+      "loss": 0.9611,
+      "step": 4270
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14383641938913058,
+      "learning_rate": 4.977489380620215e-07,
+      "loss": 0.9562,
+      "step": 4275
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13676169288298823,
+      "learning_rate": 4.966234213515809e-07,
+      "loss": 0.9517,
+      "step": 4280
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.13082849026044158,
+      "learning_rate": 4.954979217513174e-07,
+      "loss": 0.9385,
+      "step": 4285
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.145634899221159,
+      "learning_rate": 4.94372444964489e-07,
+      "loss": 0.931,
+      "step": 4290
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.14989098040292567,
+      "learning_rate": 4.932469966942373e-07,
+      "loss": 0.9612,
+      "step": 4295
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.137226302131918,
+      "learning_rate": 4.921215826435599e-07,
+      "loss": 0.9095,
+      "step": 4300
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1461893206296984,
+      "learning_rate": 4.909962085152811e-07,
+      "loss": 0.9302,
+      "step": 4305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14353479153734497,
+      "learning_rate": 4.898708800120225e-07,
+      "loss": 0.9145,
+      "step": 4310
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.13562696311014116,
+      "learning_rate": 4.887456028361743e-07,
+      "loss": 0.9315,
+      "step": 4315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1439960756887905,
+      "learning_rate": 4.876203826898674e-07,
+      "loss": 0.9528,
+      "step": 4320
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1441504289820402,
+      "learning_rate": 4.864952252749431e-07,
+      "loss": 0.9416,
+      "step": 4325
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14771137094183798,
+      "learning_rate": 4.853701362929252e-07,
+      "loss": 0.9475,
+      "step": 4330
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.13322515728443313,
+      "learning_rate": 4.842451214449905e-07,
+      "loss": 0.9622,
+      "step": 4335
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1375484952573135,
+      "learning_rate": 4.831201864319399e-07,
+      "loss": 0.9184,
+      "step": 4340
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14887710613210275,
+      "learning_rate": 4.819953369541705e-07,
+      "loss": 0.9479,
+      "step": 4345
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1352345903750029,
+      "learning_rate": 4.808705787116451e-07,
+      "loss": 0.9223,
+      "step": 4350
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.12955627618554144,
+      "learning_rate": 4.797459174038646e-07,
+      "loss": 0.93,
+      "step": 4355
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.13838552160506265,
+      "learning_rate": 4.786213587298389e-07,
+      "loss": 0.9632,
+      "step": 4360
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14706771152988074,
+      "learning_rate": 4.774969083880574e-07,
+      "loss": 0.9807,
+      "step": 4365
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14197009629833574,
+      "learning_rate": 4.7637257207646085e-07,
+      "loss": 0.9242,
+      "step": 4370
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14281319185569147,
+      "learning_rate": 4.7524835549241227e-07,
+      "loss": 0.9357,
+      "step": 4375
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.14142000093076595,
+      "learning_rate": 4.741242643326673e-07,
+      "loss": 0.947,
+      "step": 4380
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14335727669952367,
+      "learning_rate": 4.73000304293347e-07,
+      "loss": 0.9265,
+      "step": 4385
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13851165933828438,
+      "learning_rate": 4.718764810699072e-07,
+      "loss": 0.9675,
+      "step": 4390
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14444348222595427,
+      "learning_rate": 4.7075280035711073e-07,
+      "loss": 0.9791,
+      "step": 4395
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.12368782508157358,
+      "learning_rate": 4.696292678489984e-07,
+      "loss": 0.8974,
+      "step": 4400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13488729071130998,
+      "learning_rate": 4.6850588923886e-07,
+      "loss": 0.9621,
+      "step": 4405
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1409059482513919,
+      "learning_rate": 4.673826702192052e-07,
+      "loss": 0.9539,
+      "step": 4410
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14758733921836875,
+      "learning_rate": 4.662596164817354e-07,
+      "loss": 0.9581,
+      "step": 4415
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14468429706703823,
+      "learning_rate": 4.6513673371731377e-07,
+      "loss": 0.9413,
+      "step": 4420
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14755643891513737,
+      "learning_rate": 4.6401402761593804e-07,
+      "loss": 0.9208,
+      "step": 4425
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13981241430039737,
+      "learning_rate": 4.6289150386671e-07,
+      "loss": 0.9632,
+      "step": 4430
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1494122587951464,
+      "learning_rate": 4.617691681578077e-07,
+      "loss": 0.9412,
+      "step": 4435
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.14530208469592773,
+      "learning_rate": 4.6064702617645623e-07,
+      "loss": 0.9314,
+      "step": 4440
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13355547335156492,
+      "learning_rate": 4.595250836088991e-07,
+      "loss": 0.9365,
+      "step": 4445
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1463801310470242,
+      "learning_rate": 4.584033461403694e-07,
+      "loss": 0.9445,
+      "step": 4450
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.13410275920436177,
+      "learning_rate": 4.572818194550608e-07,
+      "loss": 0.9429,
+      "step": 4455
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13655297950241482,
+      "learning_rate": 4.5616050923609844e-07,
+      "loss": 0.9208,
+      "step": 4460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13858841718077888,
+      "learning_rate": 4.5503942116551166e-07,
+      "loss": 0.9761,
+      "step": 4465
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.14893020947383356,
+      "learning_rate": 4.539185609242029e-07,
+      "loss": 0.9587,
+      "step": 4470
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.15200754780433523,
+      "learning_rate": 4.527979341919208e-07,
+      "loss": 0.938,
+      "step": 4475
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.14238609293971333,
+      "learning_rate": 4.516775466472307e-07,
+      "loss": 0.9343,
+      "step": 4480
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1391204297774506,
+      "learning_rate": 4.505574039674856e-07,
+      "loss": 0.9538,
+      "step": 4485
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.14496994840666477,
+      "learning_rate": 4.4943751182879824e-07,
+      "loss": 0.9458,
+      "step": 4490
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.15021111127846792,
+      "learning_rate": 4.483178759060111e-07,
+      "loss": 0.912,
+      "step": 4495
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.15610398732035918,
+      "learning_rate": 4.471985018726686e-07,
+      "loss": 0.9615,
+      "step": 4500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1361888595904314,
+      "learning_rate": 4.4607939540098853e-07,
+      "loss": 0.9327,
+      "step": 4505
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13762049558245606,
+      "learning_rate": 4.4496056216183217e-07,
+      "loss": 0.9413,
+      "step": 4510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1344487868542401,
+      "learning_rate": 4.4384200782467663e-07,
+      "loss": 0.949,
+      "step": 4515
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13683998349280319,
+      "learning_rate": 4.427237380575857e-07,
+      "loss": 0.9399,
+      "step": 4520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1513776321717849,
+      "learning_rate": 4.4160575852718093e-07,
+      "loss": 0.9467,
+      "step": 4525
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13814759985520336,
+      "learning_rate": 4.4048807489861376e-07,
+      "loss": 0.9405,
+      "step": 4530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.13704792504089905,
+      "learning_rate": 4.393706928355356e-07,
+      "loss": 0.9416,
+      "step": 4535
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.17484503869404533,
+      "learning_rate": 4.3825361800006966e-07,
+      "loss": 0.9397,
+      "step": 4540
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13022806064129525,
+      "learning_rate": 4.3713685605278306e-07,
+      "loss": 0.9378,
+      "step": 4545
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1348576989284782,
+      "learning_rate": 4.360204126526566e-07,
+      "loss": 0.9313,
+      "step": 4550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14234032836867147,
+      "learning_rate": 4.3490429345705725e-07,
+      "loss": 0.956,
+      "step": 4555
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14376320461696313,
+      "learning_rate": 4.337885041217095e-07,
+      "loss": 0.949,
+      "step": 4560
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14802533206059484,
+      "learning_rate": 4.326730503006656e-07,
+      "loss": 0.9599,
+      "step": 4565
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14390340243709662,
+      "learning_rate": 4.3155793764627837e-07,
+      "loss": 0.9738,
+      "step": 4570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13777145999732973,
+      "learning_rate": 4.304431718091712e-07,
+      "loss": 0.928,
+      "step": 4575
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14294635047404755,
+      "learning_rate": 4.2932875843821026e-07,
+      "loss": 0.9369,
+      "step": 4580
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.13563822521920618,
+      "learning_rate": 4.2821470318047615e-07,
+      "loss": 0.928,
+      "step": 4585
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.1373813746120795,
+      "learning_rate": 4.2710101168123394e-07,
+      "loss": 0.9196,
+      "step": 4590
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14877724321113303,
+      "learning_rate": 4.25987689583906e-07,
+      "loss": 0.9347,
+      "step": 4595
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14455194567541574,
+      "learning_rate": 4.2487474253004295e-07,
+      "loss": 0.9383,
+      "step": 4600
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.14891273651674156,
+      "learning_rate": 4.237621761592943e-07,
+      "loss": 0.9362,
+      "step": 4605
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.15143942839340951,
+      "learning_rate": 4.2264999610938136e-07,
+      "loss": 0.9339,
+      "step": 4610
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1409389182270421,
+      "learning_rate": 4.215382080160671e-07,
+      "loss": 0.9637,
+      "step": 4615
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14095542475122794,
+      "learning_rate": 4.2042681751312844e-07,
+      "loss": 0.9458,
+      "step": 4620
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.12644982225748203,
+      "learning_rate": 4.19315830232328e-07,
+      "loss": 0.9083,
+      "step": 4625
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14079396599753213,
+      "learning_rate": 4.1820525180338493e-07,
+      "loss": 0.9316,
+      "step": 4630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13274124330641743,
+      "learning_rate": 4.1709508785394634e-07,
+      "loss": 0.9553,
+      "step": 4635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14243695518154492,
+      "learning_rate": 4.1598534400955957e-07,
+      "loss": 0.9406,
+      "step": 4640
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13474497571879107,
+      "learning_rate": 4.148760258936426e-07,
+      "loss": 0.9012,
+      "step": 4645
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.15105241270076128,
+      "learning_rate": 4.1376713912745657e-07,
+      "loss": 0.9044,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1494320118851251,
+      "learning_rate": 4.126586893300766e-07,
+      "loss": 0.9595,
+      "step": 4655
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.13718789871080017,
+      "learning_rate": 4.115506821183633e-07,
+      "loss": 0.9625,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1463266585177864,
+      "learning_rate": 4.104431231069352e-07,
+      "loss": 0.9816,
+      "step": 4665
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14415807919200307,
+      "learning_rate": 4.093360179081393e-07,
+      "loss": 0.9231,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.14960626251189366,
+      "learning_rate": 4.082293721320227e-07,
+      "loss": 0.9289,
+      "step": 4675
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1336138052121237,
+      "learning_rate": 4.0712319138630514e-07,
+      "loss": 0.9369,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1345966886455586,
+      "learning_rate": 4.0601748127634916e-07,
+      "loss": 0.9376,
+      "step": 4685
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.15047743823981036,
+      "learning_rate": 4.049122474051332e-07,
+      "loss": 0.9538,
+      "step": 4690
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14223183881617826,
+      "learning_rate": 4.038074953732218e-07,
+      "loss": 0.9351,
+      "step": 4695
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14077365247559628,
+      "learning_rate": 4.02703230778738e-07,
+      "loss": 0.9833,
+      "step": 4700
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14469553955657283,
+      "learning_rate": 4.015994592173352e-07,
+      "loss": 0.9424,
+      "step": 4705
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13689605748742337,
+      "learning_rate": 4.0049618628216833e-07,
+      "loss": 0.9909,
+      "step": 4710
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1473975521854963,
+      "learning_rate": 3.9939341756386505e-07,
+      "loss": 0.9412,
+      "step": 4715
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14133881222005687,
+      "learning_rate": 3.98291158650499e-07,
+      "loss": 0.9435,
+      "step": 4720
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14157942615468969,
+      "learning_rate": 3.9718941512755953e-07,
+      "loss": 0.9116,
+      "step": 4725
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13823645427368084,
+      "learning_rate": 3.960881925779251e-07,
+      "loss": 0.9433,
+      "step": 4730
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1380429178122508,
+      "learning_rate": 3.949874965818336e-07,
+      "loss": 0.9597,
+      "step": 4735
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14745997022515944,
+      "learning_rate": 3.9388733271685515e-07,
+      "loss": 0.9374,
+      "step": 4740
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.13640740031612808,
+      "learning_rate": 3.9278770655786315e-07,
+      "loss": 0.9134,
+      "step": 4745
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.15331927389055094,
+      "learning_rate": 3.9168862367700635e-07,
+      "loss": 0.967,
+      "step": 4750
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.14455802258132033,
+      "learning_rate": 3.9059008964368083e-07,
+      "loss": 0.9352,
+      "step": 4755
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1357021725087507,
+      "learning_rate": 3.8949211002450107e-07,
+      "loss": 0.9592,
+      "step": 4760
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1380279312806291,
+      "learning_rate": 3.88394690383272e-07,
+      "loss": 0.9233,
+      "step": 4765
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14867486573811337,
+      "learning_rate": 3.872978362809618e-07,
+      "loss": 0.9812,
+      "step": 4770
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1485961679801426,
+      "learning_rate": 3.8620155327567193e-07,
+      "loss": 0.9379,
+      "step": 4775
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14200231172960304,
+      "learning_rate": 3.8510584692261063e-07,
+      "loss": 0.9241,
+      "step": 4780
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1336342654106056,
+      "learning_rate": 3.8401072277406363e-07,
+      "loss": 0.9285,
+      "step": 4785
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1418105975053479,
+      "learning_rate": 3.829161863793666e-07,
+      "loss": 0.9363,
+      "step": 4790
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14719004987049533,
+      "learning_rate": 3.8182224328487724e-07,
+      "loss": 0.9303,
+      "step": 4795
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14483228210557986,
+      "learning_rate": 3.8072889903394614e-07,
+      "loss": 0.9323,
+      "step": 4800
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14153232541162572,
+      "learning_rate": 3.796361591668895e-07,
+      "loss": 0.9345,
+      "step": 4805
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15373941821304266,
+      "learning_rate": 3.7854402922096157e-07,
+      "loss": 0.934,
+      "step": 4810
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1577553993524366,
+      "learning_rate": 3.7745251473032514e-07,
+      "loss": 0.9446,
+      "step": 4815
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.15009190987248588,
+      "learning_rate": 3.7636162122602457e-07,
+      "loss": 0.9691,
+      "step": 4820
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1392858358107291,
+      "learning_rate": 3.752713542359579e-07,
+      "loss": 0.9473,
+      "step": 4825
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.12872704651413325,
+      "learning_rate": 3.741817192848478e-07,
+      "loss": 0.9236,
+      "step": 4830
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1415907950891354,
+      "learning_rate": 3.7309272189421457e-07,
+      "loss": 0.9274,
+      "step": 4835
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1597059790363494,
+      "learning_rate": 3.7200436758234776e-07,
+      "loss": 0.9327,
+      "step": 4840
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.14195112592474582,
+      "learning_rate": 3.7091666186427785e-07,
+      "loss": 0.9477,
+      "step": 4845
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13841665997838887,
+      "learning_rate": 3.698296102517495e-07,
+      "loss": 0.9091,
+      "step": 4850
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14658674466631877,
+      "learning_rate": 3.68743218253192e-07,
+      "loss": 0.9542,
+      "step": 4855
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1374221383597684,
+      "learning_rate": 3.676574913736924e-07,
+      "loss": 0.8939,
+      "step": 4860
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.12789663128231268,
+      "learning_rate": 3.665724351149679e-07,
+      "loss": 0.9446,
+      "step": 4865
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.17249517082439667,
+      "learning_rate": 3.6548805497533664e-07,
+      "loss": 0.9579,
+      "step": 4870
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13735251666219012,
+      "learning_rate": 3.6440435644969147e-07,
+      "loss": 0.9561,
+      "step": 4875
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1303994841502591,
+      "learning_rate": 3.633213450294706e-07,
+      "loss": 0.9514,
+      "step": 4880
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.143225557897971,
+      "learning_rate": 3.6223902620263056e-07,
+      "loss": 0.928,
+      "step": 4885
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.1592095177645772,
+      "learning_rate": 3.611574054536189e-07,
+      "loss": 0.946,
+      "step": 4890
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14908204927455143,
+      "learning_rate": 3.6007648826334503e-07,
+      "loss": 0.9139,
+      "step": 4895
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.13952277034485105,
+      "learning_rate": 3.5899628010915357e-07,
+      "loss": 0.9084,
+      "step": 4900
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14474751163923796,
+      "learning_rate": 3.579167864647965e-07,
+      "loss": 0.942,
+      "step": 4905
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14873589370815438,
+      "learning_rate": 3.5683801280040437e-07,
+      "loss": 0.9065,
+      "step": 4910
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14219047553770567,
+      "learning_rate": 3.5575996458246044e-07,
+      "loss": 0.9544,
+      "step": 4915
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.14536914120761013,
+      "learning_rate": 3.5468264727377097e-07,
+      "loss": 0.9596,
+      "step": 4920
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.15351684813578242,
+      "learning_rate": 3.536060663334387e-07,
+      "loss": 1.0067,
+      "step": 4925
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13706606560071907,
+      "learning_rate": 3.525302272168353e-07,
+      "loss": 0.939,
+      "step": 4930
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14824834885535257,
+      "learning_rate": 3.5145513537557333e-07,
+      "loss": 0.9165,
+      "step": 4935
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13999285925462907,
+      "learning_rate": 3.503807962574781e-07,
+      "loss": 0.9418,
+      "step": 4940
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14112367259850458,
+      "learning_rate": 3.4930721530656145e-07,
+      "loss": 0.9318,
+      "step": 4945
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14385250272201497,
+      "learning_rate": 3.482343979629926e-07,
+      "loss": 0.9345,
+      "step": 4950
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14545016646947584,
+      "learning_rate": 3.471623496630721e-07,
+      "loss": 0.9457,
+      "step": 4955
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13018750794909917,
+      "learning_rate": 3.460910758392029e-07,
+      "loss": 0.9207,
+      "step": 4960
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14437018907158786,
+      "learning_rate": 3.4502058191986337e-07,
+      "loss": 0.9544,
+      "step": 4965
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1404857495433008,
+      "learning_rate": 3.439508733295806e-07,
+      "loss": 0.9392,
+      "step": 4970
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13277736587760902,
+      "learning_rate": 3.4288195548890155e-07,
+      "loss": 0.9242,
+      "step": 4975
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.13801682472321009,
+      "learning_rate": 3.418138338143661e-07,
+      "loss": 0.9459,
+      "step": 4980
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1417857033242969,
+      "learning_rate": 3.407465137184805e-07,
+      "loss": 0.9477,
+      "step": 4985
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.14114057544363073,
+      "learning_rate": 3.3968000060968805e-07,
+      "loss": 0.9399,
+      "step": 4990
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1454108037427916,
+      "learning_rate": 3.3861429989234383e-07,
+      "loss": 0.9335,
+      "step": 4995
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1330886715685197,
+      "learning_rate": 3.375494169666856e-07,
+      "loss": 0.9631,
+      "step": 5000
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.14449411879854596,
+      "learning_rate": 3.3648535722880717e-07,
+      "loss": 0.9546,
+      "step": 5005
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.15695941733269012,
+      "learning_rate": 3.3542212607063144e-07,
+      "loss": 0.948,
+      "step": 5010
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.143769115506836,
+      "learning_rate": 3.343597288798823e-07,
+      "loss": 0.9197,
+      "step": 5015
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13276417580403352,
+      "learning_rate": 3.3329817104005755e-07,
+      "loss": 0.9446,
+      "step": 5020
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1359923615274355,
+      "learning_rate": 3.3223745793040227e-07,
+      "loss": 0.9438,
+      "step": 5025
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13705906977748641,
+      "learning_rate": 3.3117759492588016e-07,
+      "loss": 0.9305,
+      "step": 5030
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13985453928339647,
+      "learning_rate": 3.301185873971482e-07,
+      "loss": 0.9312,
+      "step": 5035
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13517797780792692,
+      "learning_rate": 3.2906044071052757e-07,
+      "loss": 0.9352,
+      "step": 5040
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.13207188344414794,
+      "learning_rate": 3.280031602279777e-07,
+      "loss": 0.919,
+      "step": 5045
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.14644444340430735,
+      "learning_rate": 3.2694675130706863e-07,
+      "loss": 0.9631,
+      "step": 5050
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1402621328131883,
+      "learning_rate": 3.2589121930095397e-07,
+      "loss": 0.9268,
+      "step": 5055
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.14725521781507542,
+      "learning_rate": 3.2483656955834335e-07,
+      "loss": 0.9366,
+      "step": 5060
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.16879467258873443,
+      "learning_rate": 3.2378280742347653e-07,
+      "loss": 0.9608,
+      "step": 5065
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.15098337048101163,
+      "learning_rate": 3.227299382360944e-07,
+      "loss": 0.967,
+      "step": 5070
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.143100261895118,
+      "learning_rate": 3.216779673314141e-07,
+      "loss": 0.9242,
+      "step": 5075
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1362708122904339,
+      "learning_rate": 3.206269000401e-07,
+      "loss": 0.9396,
+      "step": 5080
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1342951938616958,
+      "learning_rate": 3.1957674168823803e-07,
+      "loss": 0.9304,
+      "step": 5085
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1398639952404413,
+      "learning_rate": 3.185274975973081e-07,
+      "loss": 0.9431,
+      "step": 5090
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1515309189011652,
+      "learning_rate": 3.174791730841576e-07,
+      "loss": 0.9523,
+      "step": 5095
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1356078148032466,
+      "learning_rate": 3.1643177346097344e-07,
+      "loss": 0.9312,
+      "step": 5100
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.15368894096185542,
+      "learning_rate": 3.1538530403525667e-07,
+      "loss": 0.9465,
+      "step": 5105
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13520557169586825,
+      "learning_rate": 3.1433977010979385e-07,
+      "loss": 0.9425,
+      "step": 5110
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13640548619825407,
+      "learning_rate": 3.1329517698263197e-07,
+      "loss": 0.9639,
+      "step": 5115
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1427251972516278,
+      "learning_rate": 3.1225152994704996e-07,
+      "loss": 0.9532,
+      "step": 5120
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.15010846643161704,
+      "learning_rate": 3.11208834291533e-07,
+      "loss": 0.9381,
+      "step": 5125
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.14012989532534192,
+      "learning_rate": 3.101670952997454e-07,
+      "loss": 0.9241,
+      "step": 5130
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1481045171757905,
+      "learning_rate": 3.0912631825050343e-07,
+      "loss": 0.9485,
+      "step": 5135
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.13465165801277915,
+      "learning_rate": 3.0808650841774896e-07,
+      "loss": 0.9055,
+      "step": 5140
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1380149348947248,
+      "learning_rate": 3.0704767107052307e-07,
+      "loss": 0.8961,
+      "step": 5145
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.15035284770429122,
+      "learning_rate": 3.0600981147293833e-07,
+      "loss": 0.9113,
+      "step": 5150
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.15262884748665467,
+      "learning_rate": 3.0497293488415344e-07,
+      "loss": 0.9358,
+      "step": 5155
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1468915627762529,
+      "learning_rate": 3.0393704655834515e-07,
+      "loss": 0.9504,
+      "step": 5160
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13131077678912606,
+      "learning_rate": 3.029021517446829e-07,
+      "loss": 0.8981,
+      "step": 5165
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13507770132496325,
+      "learning_rate": 3.0186825568730176e-07,
+      "loss": 0.9386,
+      "step": 5170
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13682348828043034,
+      "learning_rate": 3.008353636252754e-07,
+      "loss": 0.9421,
+      "step": 5175
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1493448757363097,
+      "learning_rate": 2.998034807925899e-07,
+      "loss": 0.9801,
+      "step": 5180
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.14101649343908923,
+      "learning_rate": 2.987726124181178e-07,
+      "loss": 0.9279,
+      "step": 5185
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5353487699925709,
+      "learning_rate": 2.977427637255904e-07,
+      "loss": 0.9263,
+      "step": 5190
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1319298976109941,
+      "learning_rate": 2.967139399335726e-07,
+      "loss": 0.9051,
+      "step": 5195
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.15420951337334074,
+      "learning_rate": 2.956861462554353e-07,
+      "loss": 0.9396,
+      "step": 5200
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.14070639257331896,
+      "learning_rate": 2.9465938789932963e-07,
+      "loss": 0.9422,
+      "step": 5205
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1503847011446319,
+      "learning_rate": 2.9363367006816084e-07,
+      "loss": 0.9091,
+      "step": 5210
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.15131990596060924,
+      "learning_rate": 2.926089979595606e-07,
+      "loss": 0.9328,
+      "step": 5215
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.14525077268847594,
+      "learning_rate": 2.9158537676586267e-07,
+      "loss": 0.9219,
+      "step": 5220
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.14467729251248032,
+      "learning_rate": 2.9056281167407455e-07,
+      "loss": 0.9246,
+      "step": 5225
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.13856536700506328,
+      "learning_rate": 2.8954130786585266e-07,
+      "loss": 0.9428,
+      "step": 5230
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.13809798052984443,
+      "learning_rate": 2.8852087051747553e-07,
+      "loss": 0.9232,
+      "step": 5235
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.13922354642320361,
+      "learning_rate": 2.875015047998177e-07,
+      "loss": 0.9757,
+      "step": 5240
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1434265794323591,
+      "learning_rate": 2.864832158783226e-07,
+      "loss": 0.925,
+      "step": 5245
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14982498701052552,
+      "learning_rate": 2.854660089129787e-07,
+      "loss": 0.9581,
+      "step": 5250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1370163162797794,
+      "learning_rate": 2.844498890582901e-07,
+      "loss": 0.9331,
+      "step": 5255
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.15588729555811562,
+      "learning_rate": 2.834348614632538e-07,
+      "loss": 0.9123,
+      "step": 5260
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14527542736875546,
+      "learning_rate": 2.824209312713312e-07,
+      "loss": 0.9432,
+      "step": 5265
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1435553706240711,
+      "learning_rate": 2.814081036204223e-07,
+      "loss": 0.9279,
+      "step": 5270
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1452131557480785,
+      "learning_rate": 2.8039638364284147e-07,
+      "loss": 0.9666,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.15832823861548598,
+      "learning_rate": 2.79385776465289e-07,
+      "loss": 0.964,
+      "step": 5280
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14196480638934295,
+      "learning_rate": 2.783762872088268e-07,
+      "loss": 0.9438,
+      "step": 5285
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14295847328763944,
+      "learning_rate": 2.773679209888518e-07,
+      "loss": 0.9235,
+      "step": 5290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.12994812967281538,
+      "learning_rate": 2.763606829150703e-07,
+      "loss": 0.9296,
+      "step": 5295
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14888814798698385,
+      "learning_rate": 2.753545780914718e-07,
+      "loss": 0.9473,
+      "step": 5300
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.14780952889906895,
+      "learning_rate": 2.743496116163032e-07,
+      "loss": 0.9261,
+      "step": 5305
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1474839231655141,
+      "learning_rate": 2.733457885820432e-07,
+      "loss": 0.9377,
+      "step": 5310
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1411566401398317,
+      "learning_rate": 2.7234311407537623e-07,
+      "loss": 0.95,
+      "step": 5315
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15190340946596964,
+      "learning_rate": 2.713415931771671e-07,
+      "loss": 0.937,
+      "step": 5320
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1487943014848495,
+      "learning_rate": 2.703412309624338e-07,
+      "loss": 0.949,
+      "step": 5325
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15265482067955816,
+      "learning_rate": 2.6934203250032454e-07,
+      "loss": 0.959,
+      "step": 5330
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14493278321438574,
+      "learning_rate": 2.683440028540893e-07,
+      "loss": 0.9412,
+      "step": 5335
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1533526696374475,
+      "learning_rate": 2.6734714708105575e-07,
+      "loss": 0.9556,
+      "step": 5340
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14464538185393558,
+      "learning_rate": 2.6635147023260307e-07,
+      "loss": 0.9218,
+      "step": 5345
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13986665563585593,
+      "learning_rate": 2.6535697735413604e-07,
+      "loss": 0.9418,
+      "step": 5350
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1487921121642326,
+      "learning_rate": 2.6436367348506094e-07,
+      "loss": 0.948,
+      "step": 5355
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14914752970700781,
+      "learning_rate": 2.633715636587579e-07,
+      "loss": 0.9788,
+      "step": 5360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14859357811270174,
+      "learning_rate": 2.6238065290255694e-07,
+      "loss": 0.9523,
+      "step": 5365
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.13911386614831792,
+      "learning_rate": 2.613909462377121e-07,
+      "loss": 0.9356,
+      "step": 5370
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.15981724700918887,
+      "learning_rate": 2.6040244867937544e-07,
+      "loss": 0.9672,
+      "step": 5375
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.1354529366555852,
+      "learning_rate": 2.5941516523657314e-07,
+      "loss": 0.946,
+      "step": 5380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.14444109377970338,
+      "learning_rate": 2.5842910091217786e-07,
+      "loss": 0.9557,
+      "step": 5385
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1330565699513302,
+      "learning_rate": 2.5744426070288515e-07,
+      "loss": 0.9289,
+      "step": 5390
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1469298738190375,
+      "learning_rate": 2.564606495991877e-07,
+      "loss": 0.9125,
+      "step": 5395
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.163048910652909,
+      "learning_rate": 2.5547827258534973e-07,
+      "loss": 0.94,
+      "step": 5400
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15241582908323992,
+      "learning_rate": 2.5449713463938207e-07,
+      "loss": 0.9626,
+      "step": 5405
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1442720216165108,
+      "learning_rate": 2.535172407330167e-07,
+      "loss": 0.9546,
+      "step": 5410
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.14731902368379265,
+      "learning_rate": 2.525385958316818e-07,
+      "loss": 0.9383,
+      "step": 5415
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.14499075426412514,
+      "learning_rate": 2.515612048944763e-07,
+      "loss": 0.9379,
+      "step": 5420
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.13939663592336132,
+      "learning_rate": 2.5058507287414514e-07,
+      "loss": 0.9609,
+      "step": 5425
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.146174099748054,
+      "learning_rate": 2.496102047170533e-07,
+      "loss": 0.9384,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15142936177854155,
+      "learning_rate": 2.486366053631625e-07,
+      "loss": 0.9135,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2659056539312913,
+      "learning_rate": 2.4766427974600383e-07,
+      "loss": 0.937,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.13177417581127296,
+      "learning_rate": 2.466932327926548e-07,
+      "loss": 0.9598,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.15771363352906884,
+      "learning_rate": 2.4572346942371303e-07,
+      "loss": 0.9479,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.14818700516626399,
+      "learning_rate": 2.4475499455327205e-07,
+      "loss": 0.9133,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.14516785838102692,
+      "learning_rate": 2.4378781308889653e-07,
+      "loss": 0.8932,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1581359764786372,
+      "learning_rate": 2.4282192993159617e-07,
+      "loss": 0.9611,
+      "step": 5465
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1486376418861218,
+      "learning_rate": 2.418573499758025e-07,
+      "loss": 0.9384,
+      "step": 5470
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13325206770565617,
+      "learning_rate": 2.4089407810934285e-07,
+      "loss": 0.9449,
+      "step": 5475
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14038249032744746,
+      "learning_rate": 2.399321192134164e-07,
+      "loss": 0.9201,
+      "step": 5480
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14976357592011533,
+      "learning_rate": 2.38971478162569e-07,
+      "loss": 0.9667,
+      "step": 5485
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14040289159862654,
+      "learning_rate": 2.3801215982466844e-07,
+      "loss": 0.9401,
+      "step": 5490
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14519110278289096,
+      "learning_rate": 2.3705416906088e-07,
+      "loss": 0.9348,
+      "step": 5495
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14918246211599007,
+      "learning_rate": 2.3609751072564171e-07,
+      "loss": 0.9001,
+      "step": 5500
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14657168942267051,
+      "learning_rate": 2.3514218966664002e-07,
+      "loss": 0.9596,
+      "step": 5505
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13835626549781735,
+      "learning_rate": 2.3418821072478407e-07,
+      "loss": 0.9642,
+      "step": 5510
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.14920804750214364,
+      "learning_rate": 2.3323557873418364e-07,
+      "loss": 0.9582,
+      "step": 5515
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13759171797657285,
+      "learning_rate": 2.3228429852212155e-07,
+      "loss": 0.9626,
+      "step": 5520
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.13727533451162782,
+      "learning_rate": 2.313343749090314e-07,
+      "loss": 0.9554,
+      "step": 5525
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.15274281760998423,
+      "learning_rate": 2.30385812708473e-07,
+      "loss": 0.9818,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.12728594843230276,
+      "learning_rate": 2.2943861672710624e-07,
+      "loss": 0.8967,
+      "step": 5535
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.15525803427639978,
+      "learning_rate": 2.2849279176466928e-07,
+      "loss": 0.9487,
+      "step": 5540
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1430716661073758,
+      "learning_rate": 2.2754834261395178e-07,
+      "loss": 0.944,
+      "step": 5545
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14431962364082201,
+      "learning_rate": 2.2660527406077234e-07,
+      "loss": 0.9392,
+      "step": 5550
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14924640015197188,
+      "learning_rate": 2.2566359088395355e-07,
+      "loss": 0.9434,
+      "step": 5555
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.16316262310468632,
+      "learning_rate": 2.2472329785529775e-07,
+      "loss": 0.9032,
+      "step": 5560
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14635993393275185,
+      "learning_rate": 2.2378439973956308e-07,
+      "loss": 0.9523,
+      "step": 5565
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1352181903294165,
+      "learning_rate": 2.2284690129443924e-07,
+      "loss": 0.9656,
+      "step": 5570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14698125449483862,
+      "learning_rate": 2.2191080727052313e-07,
+      "loss": 0.9431,
+      "step": 5575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.15337865103341397,
+      "learning_rate": 2.2097612241129538e-07,
+      "loss": 0.9543,
+      "step": 5580
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1459808386520981,
+      "learning_rate": 2.2004285145309576e-07,
+      "loss": 0.9442,
+      "step": 5585
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14888783003458975,
+      "learning_rate": 2.1911099912509888e-07,
+      "loss": 0.902,
+      "step": 5590
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.13921484596613157,
+      "learning_rate": 2.1818057014929174e-07,
+      "loss": 0.9685,
+      "step": 5595
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14361928327104742,
+      "learning_rate": 2.1725156924044813e-07,
+      "loss": 0.9714,
+      "step": 5600
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.15223851515605843,
+      "learning_rate": 2.16324001106105e-07,
+      "loss": 0.9608,
+      "step": 5605
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1512486655585766,
+      "learning_rate": 2.1539787044654006e-07,
+      "loss": 0.9499,
+      "step": 5610
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1588495451012446,
+      "learning_rate": 2.144731819547456e-07,
+      "loss": 0.9549,
+      "step": 5615
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.14085073052726388,
+      "learning_rate": 2.1354994031640738e-07,
+      "loss": 0.9231,
+      "step": 5620
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1406469602158469,
+      "learning_rate": 2.1262815020987828e-07,
+      "loss": 0.9579,
+      "step": 5625
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14061668827299642,
+      "learning_rate": 2.117078163061564e-07,
+      "loss": 0.9589,
+      "step": 5630
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14825637611955333,
+      "learning_rate": 2.1078894326886083e-07,
+      "loss": 0.9424,
+      "step": 5635
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.15354522700007248,
+      "learning_rate": 2.098715357542078e-07,
+      "loss": 0.9465,
+      "step": 5640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14413749580427002,
+      "learning_rate": 2.0895559841098735e-07,
+      "loss": 0.9419,
+      "step": 5645
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14930908448614985,
+      "learning_rate": 2.0804113588053984e-07,
+      "loss": 0.9693,
+      "step": 5650
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.13826550345246552,
+      "learning_rate": 2.0712815279673213e-07,
+      "loss": 0.9473,
+      "step": 5655
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14246126528477193,
+      "learning_rate": 2.0621665378593428e-07,
+      "loss": 0.9362,
+      "step": 5660
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.17915124753582518,
+      "learning_rate": 2.0530664346699627e-07,
+      "loss": 0.9608,
+      "step": 5665
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14134623412007333,
+      "learning_rate": 2.0439812645122372e-07,
+      "loss": 0.9386,
+      "step": 5670
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1419819206839618,
+      "learning_rate": 2.0349110734235642e-07,
+      "loss": 0.9386,
+      "step": 5675
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.15503035746338864,
+      "learning_rate": 2.025855907365428e-07,
+      "loss": 0.9311,
+      "step": 5680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14037104599670938,
+      "learning_rate": 2.0168158122231805e-07,
+      "loss": 0.9685,
+      "step": 5685
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.14685629652964266,
+      "learning_rate": 2.007790833805806e-07,
+      "loss": 0.9079,
+      "step": 5690
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.15774423481324942,
+      "learning_rate": 1.9987810178456787e-07,
+      "loss": 0.9114,
+      "step": 5695
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.15647646513643051,
+      "learning_rate": 1.9897864099983554e-07,
+      "loss": 0.9447,
+      "step": 5700
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14678883153665453,
+      "learning_rate": 1.9808070558423134e-07,
+      "loss": 0.9361,
+      "step": 5705
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14320575942565272,
+      "learning_rate": 1.9718430008787435e-07,
+      "loss": 0.9507,
+      "step": 5710
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.15211145174335655,
+      "learning_rate": 1.9628942905313084e-07,
+      "loss": 0.8995,
+      "step": 5715
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14423770092393695,
+      "learning_rate": 1.953960970145913e-07,
+      "loss": 0.9518,
+      "step": 5720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14398550822926307,
+      "learning_rate": 1.9450430849904842e-07,
+      "loss": 0.9547,
+      "step": 5725
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.15383995676469528,
+      "learning_rate": 1.9361406802547225e-07,
+      "loss": 0.9345,
+      "step": 5730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.1514155541338021,
+      "learning_rate": 1.9272538010498913e-07,
+      "loss": 0.959,
+      "step": 5735
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.15583882912334796,
+      "learning_rate": 1.9183824924085795e-07,
+      "loss": 0.9632,
+      "step": 5740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.15499070363194284,
+      "learning_rate": 1.909526799284476e-07,
+      "loss": 0.9366,
+      "step": 5745
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14579783570871807,
+      "learning_rate": 1.9006867665521402e-07,
+      "loss": 0.9303,
+      "step": 5750
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.13751377865780262,
+      "learning_rate": 1.8918624390067755e-07,
+      "loss": 0.9769,
+      "step": 5755
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14559126333176414,
+      "learning_rate": 1.8830538613640025e-07,
+      "loss": 0.974,
+      "step": 5760
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14032785626384164,
+      "learning_rate": 1.8742610782596307e-07,
+      "loss": 0.9452,
+      "step": 5765
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14576887746784367,
+      "learning_rate": 1.8654841342494383e-07,
+      "loss": 0.9579,
+      "step": 5770
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.14408440121522134,
+      "learning_rate": 1.856723073808933e-07,
+      "loss": 0.9636,
+      "step": 5775
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1466373510776226,
+      "learning_rate": 1.847977941333147e-07,
+      "loss": 0.9909,
+      "step": 5780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1433141800305017,
+      "learning_rate": 1.8392487811363898e-07,
+      "loss": 0.9212,
+      "step": 5785
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.15047949362188842,
+      "learning_rate": 1.8305356374520397e-07,
+      "loss": 0.95,
+      "step": 5790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.15571730506640438,
+      "learning_rate": 1.821838554432319e-07,
+      "loss": 0.9106,
+      "step": 5795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1376305378573877,
+      "learning_rate": 1.8131575761480528e-07,
+      "loss": 0.9138,
+      "step": 5800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14524800820104464,
+      "learning_rate": 1.8044927465884728e-07,
+      "loss": 0.9288,
+      "step": 5805
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14224815693112722,
+      "learning_rate": 1.7958441096609688e-07,
+      "loss": 0.9558,
+      "step": 5810
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14858183062607463,
+      "learning_rate": 1.7872117091908838e-07,
+      "loss": 0.9275,
+      "step": 5815
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.13820390321197273,
+      "learning_rate": 1.7785955889212834e-07,
+      "loss": 0.9456,
+      "step": 5820
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14955775703093505,
+      "learning_rate": 1.7699957925127367e-07,
+      "loss": 0.9081,
+      "step": 5825
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14796694392618723,
+      "learning_rate": 1.7614123635430949e-07,
+      "loss": 0.951,
+      "step": 5830
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14095402079682437,
+      "learning_rate": 1.7528453455072706e-07,
+      "loss": 0.9431,
+      "step": 5835
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.14844431345946524,
+      "learning_rate": 1.744294781817016e-07,
+      "loss": 0.921,
+      "step": 5840
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.1562469956288134,
+      "learning_rate": 1.7357607158007054e-07,
+      "loss": 0.9448,
+      "step": 5845
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.13516727773011417,
+      "learning_rate": 1.7272431907031144e-07,
+      "loss": 0.9541,
+      "step": 5850
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1551154204634177,
+      "learning_rate": 1.7187422496851945e-07,
+      "loss": 0.9467,
+      "step": 5855
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14284664847460612,
+      "learning_rate": 1.7102579358238727e-07,
+      "loss": 0.9089,
+      "step": 5860
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14316784172746014,
+      "learning_rate": 1.7017902921118093e-07,
+      "loss": 0.9531,
+      "step": 5865
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14840597368256137,
+      "learning_rate": 1.6933393614571966e-07,
+      "loss": 0.9343,
+      "step": 5870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15999202542066532,
+      "learning_rate": 1.6849051866835407e-07,
+      "loss": 0.9367,
+      "step": 5875
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15453002806662577,
+      "learning_rate": 1.6764878105294294e-07,
+      "loss": 0.9388,
+      "step": 5880
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1449559351909655,
+      "learning_rate": 1.668087275648341e-07,
+      "loss": 0.9632,
+      "step": 5885
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1429923355440533,
+      "learning_rate": 1.6597036246084007e-07,
+      "loss": 0.9137,
+      "step": 5890
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15390095697756095,
+      "learning_rate": 1.651336899892185e-07,
+      "loss": 0.9156,
+      "step": 5895
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2929484244245197,
+      "learning_rate": 1.6429871438964988e-07,
+      "loss": 0.9367,
+      "step": 5900
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15835323281514943,
+      "learning_rate": 1.6346543989321593e-07,
+      "loss": 0.9518,
+      "step": 5905
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1544935535850286,
+      "learning_rate": 1.6263387072237855e-07,
+      "loss": 0.9408,
+      "step": 5910
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.16069535501518972,
+      "learning_rate": 1.6180401109095815e-07,
+      "loss": 0.9375,
+      "step": 5915
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.13795507520885678,
+      "learning_rate": 1.6097586520411227e-07,
+      "loss": 0.9344,
+      "step": 5920
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.14197746116097018,
+      "learning_rate": 1.601494372583145e-07,
+      "loss": 0.9612,
+      "step": 5925
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.15959293310402556,
+      "learning_rate": 1.5932473144133313e-07,
+      "loss": 0.9817,
+      "step": 5930
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.14277361438268601,
+      "learning_rate": 1.5850175193220944e-07,
+      "loss": 0.9232,
+      "step": 5935
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.13121174462080104,
+      "learning_rate": 1.5768050290123775e-07,
+      "loss": 0.9336,
+      "step": 5940
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1482878687428246,
+      "learning_rate": 1.568609885099431e-07,
+      "loss": 0.9452,
+      "step": 5945
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15183205837507238,
+      "learning_rate": 1.5604321291106005e-07,
+      "loss": 0.9264,
+      "step": 5950
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.14721301792138466,
+      "learning_rate": 1.5522718024851328e-07,
+      "loss": 0.9477,
+      "step": 5955
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15522328498415922,
+      "learning_rate": 1.5441289465739427e-07,
+      "loss": 0.9522,
+      "step": 5960
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1416299665189105,
+      "learning_rate": 1.5360036026394275e-07,
+      "loss": 0.9103,
+      "step": 5965
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1416788856503464,
+      "learning_rate": 1.5278958118552337e-07,
+      "loss": 0.9326,
+      "step": 5970
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.14849152816916514,
+      "learning_rate": 1.5198056153060702e-07,
+      "loss": 0.9383,
+      "step": 5975
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1564108020459229,
+      "learning_rate": 1.5117330539874846e-07,
+      "loss": 0.9382,
+      "step": 5980
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.14574262582008277,
+      "learning_rate": 1.5036781688056638e-07,
+      "loss": 0.9625,
+      "step": 5985
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.15291946687335647,
+      "learning_rate": 1.495641000577224e-07,
+      "loss": 0.8933,
+      "step": 5990
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1492672754438516,
+      "learning_rate": 1.487621590029004e-07,
+      "loss": 0.9364,
+      "step": 5995
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.13670124099244704,
+      "learning_rate": 1.4796199777978586e-07,
+      "loss": 0.9339,
+      "step": 6000
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1462187335815558,
+      "learning_rate": 1.471636204430452e-07,
+      "loss": 0.9445,
+      "step": 6005
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14936900265463246,
+      "learning_rate": 1.4636703103830557e-07,
+      "loss": 0.9337,
+      "step": 6010
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13732505168809617,
+      "learning_rate": 1.4557223360213394e-07,
+      "loss": 0.8995,
+      "step": 6015
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.15646750136284834,
+      "learning_rate": 1.4477923216201675e-07,
+      "loss": 0.9213,
+      "step": 6020
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.15130771262701795,
+      "learning_rate": 1.4398803073634002e-07,
+      "loss": 0.9921,
+      "step": 6025
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14728019116040417,
+      "learning_rate": 1.4319863333436756e-07,
+      "loss": 0.9312,
+      "step": 6030
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1438177846352029,
+      "learning_rate": 1.424110439562229e-07,
+      "loss": 0.9297,
+      "step": 6035
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13878474453369102,
+      "learning_rate": 1.4162526659286656e-07,
+      "loss": 0.9228,
+      "step": 6040
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.15123036493608685,
+      "learning_rate": 1.4084130522607807e-07,
+      "loss": 0.9248,
+      "step": 6045
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1400641525036828,
+      "learning_rate": 1.4005916382843386e-07,
+      "loss": 0.9508,
+      "step": 6050
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13769050883650436,
+      "learning_rate": 1.3927884636328852e-07,
+      "loss": 0.9137,
+      "step": 6055
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.15750716657347058,
+      "learning_rate": 1.3850035678475403e-07,
+      "loss": 0.9496,
+      "step": 6060
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14593052434192175,
+      "learning_rate": 1.3772369903767988e-07,
+      "loss": 0.911,
+      "step": 6065
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13744423962970861,
+      "learning_rate": 1.3694887705763325e-07,
+      "loss": 0.9227,
+      "step": 6070
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.13703879467574198,
+      "learning_rate": 1.361758947708788e-07,
+      "loss": 0.9288,
+      "step": 6075
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14566552186767442,
+      "learning_rate": 1.3540475609435886e-07,
+      "loss": 0.9384,
+      "step": 6080
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.14783289527795426,
+      "learning_rate": 1.3463546493567368e-07,
+      "loss": 0.9343,
+      "step": 6085
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1468037592323871,
+      "learning_rate": 1.338680251930615e-07,
+      "loss": 0.9321,
+      "step": 6090
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1434293335226695,
+      "learning_rate": 1.3310244075537896e-07,
+      "loss": 0.977,
+      "step": 6095
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.141833042835304,
+      "learning_rate": 1.323387155020812e-07,
+      "loss": 0.9396,
+      "step": 6100
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14014624762452657,
+      "learning_rate": 1.3157685330320244e-07,
+      "loss": 0.9118,
+      "step": 6105
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15303098125445588,
+      "learning_rate": 1.3081685801933574e-07,
+      "loss": 0.9406,
+      "step": 6110
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14226158715966636,
+      "learning_rate": 1.300587335016148e-07,
+      "loss": 0.9586,
+      "step": 6115
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14071667749428488,
+      "learning_rate": 1.2930248359169238e-07,
+      "loss": 0.9518,
+      "step": 6120
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15492332032970332,
+      "learning_rate": 1.2854811212172346e-07,
+      "loss": 0.9023,
+      "step": 6125
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.13966173330836318,
+      "learning_rate": 1.2779562291434297e-07,
+      "loss": 0.9662,
+      "step": 6130
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2875304493022803,
+      "learning_rate": 1.2704501978264865e-07,
+      "loss": 0.925,
+      "step": 6135
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1473755464184472,
+      "learning_rate": 1.2629630653018098e-07,
+      "loss": 0.9181,
+      "step": 6140
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.15749095692574885,
+      "learning_rate": 1.2554948695090322e-07,
+      "loss": 0.9383,
+      "step": 6145
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14534864166849096,
+      "learning_rate": 1.2480456482918328e-07,
+      "loss": 0.9606,
+      "step": 6150
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14564271918886731,
+      "learning_rate": 1.2406154393977392e-07,
+      "loss": 0.9151,
+      "step": 6155
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.14693769369349916,
+      "learning_rate": 1.2332042804779374e-07,
+      "loss": 0.9558,
+      "step": 6160
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.15348170071129194,
+      "learning_rate": 1.2258122090870826e-07,
+      "loss": 0.9478,
+      "step": 6165
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1379688405407708,
+      "learning_rate": 1.2184392626831054e-07,
+      "loss": 0.9432,
+      "step": 6170
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.13834654741254748,
+      "learning_rate": 1.2110854786270248e-07,
+      "loss": 0.9805,
+      "step": 6175
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14958310021779303,
+      "learning_rate": 1.20375089418276e-07,
+      "loss": 0.914,
+      "step": 6180
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14924105463228343,
+      "learning_rate": 1.1964355465169374e-07,
+      "loss": 0.9328,
+      "step": 6185
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14770097973850022,
+      "learning_rate": 1.1891394726987058e-07,
+      "loss": 0.9088,
+      "step": 6190
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14285912967198972,
+      "learning_rate": 1.1818627096995487e-07,
+      "loss": 0.9317,
+      "step": 6195
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1448926331764474,
+      "learning_rate": 1.1746052943930895e-07,
+      "loss": 0.9387,
+      "step": 6200
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.15554674436452814,
+      "learning_rate": 1.1673672635549209e-07,
+      "loss": 0.9302,
+      "step": 6205
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1485465558734535,
+      "learning_rate": 1.1601486538624017e-07,
+      "loss": 0.9283,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14987055955863104,
+      "learning_rate": 1.1529495018944757e-07,
+      "loss": 0.9086,
+      "step": 6215
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.16222432632497197,
+      "learning_rate": 1.1457698441314967e-07,
+      "loss": 0.9292,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14819700750848447,
+      "learning_rate": 1.1386097169550246e-07,
+      "loss": 0.9697,
+      "step": 6225
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14511350187116975,
+      "learning_rate": 1.1314691566476636e-07,
+      "loss": 0.9219,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.14808665186029626,
+      "learning_rate": 1.1243481993928555e-07,
+      "loss": 0.9309,
+      "step": 6235
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.15564170166578836,
+      "learning_rate": 1.1172468812747137e-07,
+      "loss": 0.928,
+      "step": 6240
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14721649203709825,
+      "learning_rate": 1.1101652382778326e-07,
+      "loss": 0.9556,
+      "step": 6245
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14250501930747192,
+      "learning_rate": 1.103103306287106e-07,
+      "loss": 0.9498,
+      "step": 6250
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.15075114497263598,
+      "learning_rate": 1.0960611210875459e-07,
+      "loss": 0.9514,
+      "step": 6255
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14712645790665124,
+      "learning_rate": 1.089038718364102e-07,
+      "loss": 0.9168,
+      "step": 6260
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.13675732744709934,
+      "learning_rate": 1.0820361337014794e-07,
+      "loss": 0.9261,
+      "step": 6265
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1514903468649154,
+      "learning_rate": 1.0750534025839592e-07,
+      "loss": 0.9112,
+      "step": 6270
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.16424702577189915,
+      "learning_rate": 1.0680905603952178e-07,
+      "loss": 0.9592,
+      "step": 6275
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14554388820710962,
+      "learning_rate": 1.0611476424181487e-07,
+      "loss": 0.9478,
+      "step": 6280
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14149588796524945,
+      "learning_rate": 1.0542246838346835e-07,
+      "loss": 0.956,
+      "step": 6285
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14763901967532567,
+      "learning_rate": 1.0473217197256135e-07,
+      "loss": 0.942,
+      "step": 6290
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.16807876526032808,
+      "learning_rate": 1.0404387850704077e-07,
+      "loss": 0.9429,
+      "step": 6295
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1531588371037586,
+      "learning_rate": 1.0335759147470485e-07,
+      "loss": 0.8944,
+      "step": 6300
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.1479278434562303,
+      "learning_rate": 1.0267331435318355e-07,
+      "loss": 0.9065,
+      "step": 6305
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.13988008260694076,
+      "learning_rate": 1.0199105060992297e-07,
+      "loss": 0.9049,
+      "step": 6310
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.14273683164353781,
+      "learning_rate": 1.0131080370216593e-07,
+      "loss": 0.9609,
+      "step": 6315
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15260327356915607,
+      "learning_rate": 1.0063257707693595e-07,
+      "loss": 0.972,
+      "step": 6320
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15136301027255647,
+      "learning_rate": 9.995637417101876e-08,
+      "loss": 0.9455,
+      "step": 6325
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1448565678398313,
+      "learning_rate": 9.928219841094554e-08,
+      "loss": 0.9625,
+      "step": 6330
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14892518095633353,
+      "learning_rate": 9.861005321297516e-08,
+      "loss": 0.9485,
+      "step": 6335
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14540904327814727,
+      "learning_rate": 9.793994198307692e-08,
+      "loss": 0.9162,
+      "step": 6340
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15134547829889822,
+      "learning_rate": 9.727186811691357e-08,
+      "loss": 0.925,
+      "step": 6345
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15276733257345612,
+      "learning_rate": 9.660583499982372e-08,
+      "loss": 0.9292,
+      "step": 6350
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15561812371164221,
+      "learning_rate": 9.5941846006805e-08,
+      "loss": 0.9537,
+      "step": 6355
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15076463881064173,
+      "learning_rate": 9.527990450249674e-08,
+      "loss": 0.9579,
+      "step": 6360
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14626561570792276,
+      "learning_rate": 9.462001384116297e-08,
+      "loss": 0.9507,
+      "step": 6365
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1429871425499177,
+      "learning_rate": 9.396217736667572e-08,
+      "loss": 0.9465,
+      "step": 6370
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15691587706423527,
+      "learning_rate": 9.330639841249727e-08,
+      "loss": 0.9513,
+      "step": 6375
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14169085108611337,
+      "learning_rate": 9.26526803016645e-08,
+      "loss": 0.9551,
+      "step": 6380
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.14354922978038434,
+      "learning_rate": 9.20010263467707e-08,
+      "loss": 0.9308,
+      "step": 6385
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1496004700267322,
+      "learning_rate": 9.135143984995009e-08,
+      "loss": 0.9733,
+      "step": 6390
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.15591297184300548,
+      "learning_rate": 9.070392410285977e-08,
+      "loss": 0.956,
+      "step": 6395
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1530411974995182,
+      "learning_rate": 9.005848238666403e-08,
+      "loss": 0.9372,
+      "step": 6400
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.15005095146745312,
+      "learning_rate": 8.941511797201767e-08,
+      "loss": 0.9269,
+      "step": 6405
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14209655586369008,
+      "learning_rate": 8.877383411904865e-08,
+      "loss": 0.9271,
+      "step": 6410
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.13938562848526928,
+      "learning_rate": 8.813463407734234e-08,
+      "loss": 0.946,
+      "step": 6415
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.13742946300577213,
+      "learning_rate": 8.749752108592473e-08,
+      "loss": 0.9239,
+      "step": 6420
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14799278887362524,
+      "learning_rate": 8.686249837324611e-08,
+      "loss": 0.9354,
+      "step": 6425
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14449635705426453,
+      "learning_rate": 8.622956915716472e-08,
+      "loss": 0.9178,
+      "step": 6430
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14573556226186718,
+      "learning_rate": 8.559873664493023e-08,
+      "loss": 0.9145,
+      "step": 6435
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14601478428609946,
+      "learning_rate": 8.497000403316784e-08,
+      "loss": 0.941,
+      "step": 6440
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.13309524069134993,
+      "learning_rate": 8.434337450786177e-08,
+      "loss": 0.9176,
+      "step": 6445
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14334005883614906,
+      "learning_rate": 8.371885124433948e-08,
+      "loss": 0.9555,
+      "step": 6450
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14490356209828278,
+      "learning_rate": 8.309643740725475e-08,
+      "loss": 0.9352,
+      "step": 6455
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1481792459959185,
+      "learning_rate": 8.247613615057319e-08,
+      "loss": 0.9208,
+      "step": 6460
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.14915919203000308,
+      "learning_rate": 8.185795061755418e-08,
+      "loss": 0.9294,
+      "step": 6465
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1419906942967439,
+      "learning_rate": 8.124188394073728e-08,
+      "loss": 0.9253,
+      "step": 6470
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14141336857407205,
+      "learning_rate": 8.062793924192407e-08,
+      "loss": 0.9378,
+      "step": 6475
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14824992332348141,
+      "learning_rate": 8.001611963216386e-08,
+      "loss": 0.9204,
+      "step": 6480
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14643896598578376,
+      "learning_rate": 7.940642821173788e-08,
+      "loss": 0.96,
+      "step": 6485
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14433637548920009,
+      "learning_rate": 7.87988680701423e-08,
+      "loss": 0.9336,
+      "step": 6490
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.13705874464500314,
+      "learning_rate": 7.819344228607411e-08,
+      "loss": 0.9545,
+      "step": 6495
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14978616507380743,
+      "learning_rate": 7.759015392741453e-08,
+      "loss": 0.8976,
+      "step": 6500
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1556316332003196,
+      "learning_rate": 7.698900605121389e-08,
+      "loss": 0.9356,
+      "step": 6505
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.16005341278555965,
+      "learning_rate": 7.639000170367604e-08,
+      "loss": 0.9402,
+      "step": 6510
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1421370350768468,
+      "learning_rate": 7.57931439201428e-08,
+      "loss": 0.9354,
+      "step": 6515
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14859386159076712,
+      "learning_rate": 7.519843572507884e-08,
+      "loss": 0.9273,
+      "step": 6520
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14663123744767753,
+      "learning_rate": 7.460588013205598e-08,
+      "loss": 0.9545,
+      "step": 6525
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14695116492562266,
+      "learning_rate": 7.401548014373849e-08,
+      "loss": 0.9608,
+      "step": 6530
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15020068266098566,
+      "learning_rate": 7.342723875186685e-08,
+      "loss": 0.9089,
+      "step": 6535
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14508787018359748,
+      "learning_rate": 7.284115893724418e-08,
+      "loss": 0.9338,
+      "step": 6540
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.15388773915800702,
+      "learning_rate": 7.225724366971931e-08,
+      "loss": 0.9215,
+      "step": 6545
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.14539527518424947,
+      "learning_rate": 7.167549590817334e-08,
+      "loss": 0.9383,
+      "step": 6550
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1497313263273385,
+      "learning_rate": 7.109591860050367e-08,
+      "loss": 0.9801,
+      "step": 6555
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.14347260175134707,
+      "learning_rate": 7.051851468360897e-08,
+      "loss": 0.9268,
+      "step": 6560
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15459751132159824,
+      "learning_rate": 6.994328708337544e-08,
+      "loss": 0.9403,
+      "step": 6565
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1477090001161445,
+      "learning_rate": 6.937023871466047e-08,
+      "loss": 0.9484,
+      "step": 6570
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.14556900121938973,
+      "learning_rate": 6.879937248127898e-08,
+      "loss": 0.9413,
+      "step": 6575
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15140699807268207,
+      "learning_rate": 6.82306912759883e-08,
+      "loss": 0.941,
+      "step": 6580
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15703069505911108,
+      "learning_rate": 6.766419798047346e-08,
+      "loss": 0.9309,
+      "step": 6585
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.13917722670578686,
+      "learning_rate": 6.709989546533268e-08,
+      "loss": 0.9297,
+      "step": 6590
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.14828126113931187,
+      "learning_rate": 6.653778659006293e-08,
+      "loss": 0.943,
+      "step": 6595
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15466694674121959,
+      "learning_rate": 6.597787420304506e-08,
+      "loss": 0.9629,
+      "step": 6600
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15133062353205806,
+      "learning_rate": 6.542016114152998e-08,
+      "loss": 0.9344,
+      "step": 6605
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15658226729272964,
+      "learning_rate": 6.48646502316238e-08,
+      "loss": 0.9327,
+      "step": 6610
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.14060715935195342,
+      "learning_rate": 6.431134428827317e-08,
+      "loss": 0.9491,
+      "step": 6615
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.15038725461623328,
+      "learning_rate": 6.376024611525232e-08,
+      "loss": 0.9089,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.14519563980791136,
+      "learning_rate": 6.321135850514752e-08,
+      "loss": 0.9511,
+      "step": 6625
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14477249355138036,
+      "learning_rate": 6.266468423934346e-08,
+      "loss": 0.9528,
+      "step": 6630
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1502829009510528,
+      "learning_rate": 6.212022608800932e-08,
+      "loss": 0.9737,
+      "step": 6635
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14574985095527662,
+      "learning_rate": 6.157798681008402e-08,
+      "loss": 0.9495,
+      "step": 6640
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14872466096663917,
+      "learning_rate": 6.103796915326358e-08,
+      "loss": 0.926,
+      "step": 6645
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14362370889716344,
+      "learning_rate": 6.050017585398553e-08,
+      "loss": 0.9412,
+      "step": 6650
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14596267490965656,
+      "learning_rate": 5.996460963741668e-08,
+      "loss": 0.9529,
+      "step": 6655
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1478188334948812,
+      "learning_rate": 5.9431273217437794e-08,
+      "loss": 0.9425,
+      "step": 6660
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14351580251878562,
+      "learning_rate": 5.8900169296631e-08,
+      "loss": 0.9495,
+      "step": 6665
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1388055416701122,
+      "learning_rate": 5.837130056626555e-08,
+      "loss": 0.9562,
+      "step": 6670
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14113833234422388,
+      "learning_rate": 5.7844669706284265e-08,
+      "loss": 0.9378,
+      "step": 6675
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14907144609927325,
+      "learning_rate": 5.7320279385289995e-08,
+      "loss": 0.9966,
+      "step": 6680
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.15052805532546973,
+      "learning_rate": 5.679813226053204e-08,
+      "loss": 0.9107,
+      "step": 6685
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14025278889376477,
+      "learning_rate": 5.6278230977892804e-08,
+      "loss": 0.9805,
+      "step": 6690
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14041980282569524,
+      "learning_rate": 5.57605781718743e-08,
+      "loss": 0.9791,
+      "step": 6695
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.14154522895253088,
+      "learning_rate": 5.524517646558474e-08,
+      "loss": 0.9055,
+      "step": 6700
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.1435054556881103,
+      "learning_rate": 5.473202847072533e-08,
+      "loss": 0.9314,
+      "step": 6705
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14401115871443515,
+      "learning_rate": 5.4221136787577e-08,
+      "loss": 0.9203,
+      "step": 6710
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14117097888779312,
+      "learning_rate": 5.37125040049875e-08,
+      "loss": 0.9232,
+      "step": 6715
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15180425528736383,
+      "learning_rate": 5.3206132700357396e-08,
+      "loss": 0.9414,
+      "step": 6720
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16134287442193163,
+      "learning_rate": 5.270202543962854e-08,
+      "loss": 0.9208,
+      "step": 6725
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14765650374075312,
+      "learning_rate": 5.220018477726917e-08,
+      "loss": 0.9525,
+      "step": 6730
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15176487048568363,
+      "learning_rate": 5.17006132562629e-08,
+      "loss": 0.9139,
+      "step": 6735
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14291952917443312,
+      "learning_rate": 5.120331340809425e-08,
+      "loss": 0.9573,
+      "step": 6740
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.13862797874942046,
+      "learning_rate": 5.0708287752736556e-08,
+      "loss": 0.9435,
+      "step": 6745
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1586911290767296,
+      "learning_rate": 5.0215538798639655e-08,
+      "loss": 0.9595,
+      "step": 6750
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.15419805702600275,
+      "learning_rate": 4.972506904271584e-08,
+      "loss": 0.9405,
+      "step": 6755
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1451750560642017,
+      "learning_rate": 4.923688097032841e-08,
+      "loss": 0.9267,
+      "step": 6760
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14212278869309322,
+      "learning_rate": 4.8750977055278753e-08,
+      "loss": 0.9231,
+      "step": 6765
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16376169710394883,
+      "learning_rate": 4.8267359759793513e-08,
+      "loss": 0.9252,
+      "step": 6770
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.16088542695532626,
+      "learning_rate": 4.778603153451233e-08,
+      "loss": 0.9466,
+      "step": 6775
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.14485477261849478,
+      "learning_rate": 4.7306994818475555e-08,
+      "loss": 0.9721,
+      "step": 6780
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.15045594890803865,
+      "learning_rate": 4.683025203911167e-08,
+      "loss": 0.9229,
+      "step": 6785
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14642940230023818,
+      "learning_rate": 4.635580561222496e-08,
+      "loss": 0.939,
+      "step": 6790
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14208825000598843,
+      "learning_rate": 4.588365794198362e-08,
+      "loss": 0.9431,
+      "step": 6795
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14670521863074298,
+      "learning_rate": 4.5413811420906945e-08,
+      "loss": 0.9647,
+      "step": 6800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.15656846772154634,
+      "learning_rate": 4.494626842985416e-08,
+      "loss": 0.9135,
+      "step": 6805
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14407135385186193,
+      "learning_rate": 4.448103133801101e-08,
+      "loss": 0.9251,
+      "step": 6810
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1516797797487741,
+      "learning_rate": 4.401810250287935e-08,
+      "loss": 0.9429,
+      "step": 6815
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16442999164686467,
+      "learning_rate": 4.3557484270263954e-08,
+      "loss": 0.9093,
+      "step": 6820
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14500358655031642,
+      "learning_rate": 4.3099178974260795e-08,
+      "loss": 0.9033,
+      "step": 6825
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14177799284958195,
+      "learning_rate": 4.2643188937246165e-08,
+      "loss": 0.9528,
+      "step": 6830
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1541158964382166,
+      "learning_rate": 4.2189516469863576e-08,
+      "loss": 0.9552,
+      "step": 6835
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14114621137283356,
+      "learning_rate": 4.1738163871013034e-08,
+      "loss": 0.917,
+      "step": 6840
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.16156195005267762,
+      "learning_rate": 4.128913342783885e-08,
+      "loss": 0.9131,
+      "step": 6845
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.14932706316025446,
+      "learning_rate": 4.084242741571847e-08,
+      "loss": 0.9404,
+      "step": 6850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.15102473096191327,
+      "learning_rate": 4.039804809825054e-08,
+      "loss": 0.9206,
+      "step": 6855
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.1379254786483874,
+      "learning_rate": 3.995599772724373e-08,
+      "loss": 0.949,
+      "step": 6860
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1453156781480565,
+      "learning_rate": 3.951627854270517e-08,
+      "loss": 0.9292,
+      "step": 6865
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.14095594847966553,
+      "learning_rate": 3.907889277282905e-08,
+      "loss": 0.895,
+      "step": 6870
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1548469459290545,
+      "learning_rate": 3.864384263398573e-08,
+      "loss": 0.9294,
+      "step": 6875
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.13961139300851014,
+      "learning_rate": 3.821113033070961e-08,
+      "loss": 0.9385,
+      "step": 6880
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.13866650476140935,
+      "learning_rate": 3.778075805568914e-08,
+      "loss": 0.9435,
+      "step": 6885
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1428594362868762,
+      "learning_rate": 3.7352727989754974e-08,
+      "loss": 0.8997,
+      "step": 6890
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.14514646927832345,
+      "learning_rate": 3.692704230186877e-08,
+      "loss": 0.8933,
+      "step": 6895
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.15592365206172723,
+      "learning_rate": 3.650370314911294e-08,
+      "loss": 0.9314,
+      "step": 6900
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1485979116033507,
+      "learning_rate": 3.608271267667873e-08,
+      "loss": 0.9478,
+      "step": 6905
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.14186995089350088,
+      "learning_rate": 3.566407301785651e-08,
+      "loss": 0.9457,
+      "step": 6910
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1602585547940211,
+      "learning_rate": 3.5247786294023754e-08,
+      "loss": 0.9341,
+      "step": 6915
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1476117822606847,
+      "learning_rate": 3.483385461463528e-08,
+      "loss": 0.9268,
+      "step": 6920
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1617691162675761,
+      "learning_rate": 3.442228007721198e-08,
+      "loss": 0.9251,
+      "step": 6925
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1535926461605713,
+      "learning_rate": 3.401306476733051e-08,
+      "loss": 0.9373,
+      "step": 6930
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.14605892070363205,
+      "learning_rate": 3.360621075861242e-08,
+      "loss": 0.9124,
+      "step": 6935
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15410746680806106,
+      "learning_rate": 3.3201720112713885e-08,
+      "loss": 0.9821,
+      "step": 6940
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.16008110182542223,
+      "learning_rate": 3.279959487931527e-08,
+      "loss": 0.9468,
+      "step": 6945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14459203174613822,
+      "learning_rate": 3.2399837096110496e-08,
+      "loss": 0.9397,
+      "step": 6950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14740638732942146,
+      "learning_rate": 3.2002448788797045e-08,
+      "loss": 0.9305,
+      "step": 6955
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14477527865257664,
+      "learning_rate": 3.160743197106519e-08,
+      "loss": 0.945,
+      "step": 6960
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14154958018345654,
+      "learning_rate": 3.121478864458865e-08,
+      "loss": 0.946,
+      "step": 6965
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.15389201547810336,
+      "learning_rate": 3.082452079901354e-08,
+      "loss": 0.977,
+      "step": 6970
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.16096467949477414,
+      "learning_rate": 3.0436630411948774e-08,
+      "loss": 0.9533,
+      "step": 6975
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1488973964826823,
+      "learning_rate": 3.0051119448956e-08,
+      "loss": 0.9458,
+      "step": 6980
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13595041143319034,
+      "learning_rate": 2.96679898635393e-08,
+      "loss": 0.9191,
+      "step": 6985
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.13771958237832188,
+      "learning_rate": 2.928724359713619e-08,
+      "loss": 0.9221,
+      "step": 6990
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14925799977264773,
+      "learning_rate": 2.8908882579106397e-08,
+      "loss": 0.9441,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14579008834856663,
+      "learning_rate": 2.8532908726723427e-08,
+      "loss": 0.8936,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14755857160085628,
+      "learning_rate": 2.8159323945164015e-08,
+      "loss": 0.9368,
+      "step": 7005
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.14931861035152827,
+      "learning_rate": 2.7788130127498753e-08,
+      "loss": 0.9357,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1422469899840142,
+      "learning_rate": 2.7419329154682703e-08,
+      "loss": 0.9228,
+      "step": 7015
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.16036637195866657,
+      "learning_rate": 2.7052922895545226e-08,
+      "loss": 0.9516,
+      "step": 7020
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15360788103708464,
+      "learning_rate": 2.6688913206781184e-08,
+      "loss": 0.9427,
+      "step": 7025
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.15209386368160918,
+      "learning_rate": 2.6327301932941194e-08,
+      "loss": 0.9616,
+      "step": 7030
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14940332262984665,
+      "learning_rate": 2.596809090642249e-08,
+      "loss": 0.9377,
+      "step": 7035
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14680650958916713,
+      "learning_rate": 2.5611281947459318e-08,
+      "loss": 0.9132,
+      "step": 7040
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14430427737323326,
+      "learning_rate": 2.525687686411404e-08,
+      "loss": 0.9267,
+      "step": 7045
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1464791194560844,
+      "learning_rate": 2.4904877452267658e-08,
+      "loss": 0.9343,
+      "step": 7050
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1458450110613775,
+      "learning_rate": 2.455528549561109e-08,
+      "loss": 0.9385,
+      "step": 7055
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14580397045338558,
+      "learning_rate": 2.4208102765635897e-08,
+      "loss": 0.9164,
+      "step": 7060
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1453341116160097,
+      "learning_rate": 2.3863331021625022e-08,
+      "loss": 0.9436,
+      "step": 7065
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1453772432105652,
+      "learning_rate": 2.352097201064479e-08,
+      "loss": 0.945,
+      "step": 7070
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1475735243211823,
+      "learning_rate": 2.3181027467534807e-08,
+      "loss": 0.9426,
+      "step": 7075
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.1537065689493565,
+      "learning_rate": 2.284349911490019e-08,
+      "loss": 0.9571,
+      "step": 7080
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.14857164596238134,
+      "learning_rate": 2.2508388663102562e-08,
+      "loss": 0.9547,
+      "step": 7085
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.13601486658656098,
+      "learning_rate": 2.2175697810250805e-08,
+      "loss": 0.9761,
+      "step": 7090
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1541408749793138,
+      "learning_rate": 2.184542824219365e-08,
+      "loss": 0.9371,
+      "step": 7095
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.14994347755395715,
+      "learning_rate": 2.1517581632509717e-08,
+      "loss": 0.9403,
+      "step": 7100
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.14898778065724533,
+      "learning_rate": 2.1192159642500153e-08,
+      "loss": 0.9302,
+      "step": 7105
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.14109422429935103,
+      "learning_rate": 2.0869163921179656e-08,
+      "loss": 0.9535,
+      "step": 7110
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15342117015714668,
+      "learning_rate": 2.054859610526838e-08,
+      "loss": 0.9213,
+      "step": 7115
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.13455135140020671,
+      "learning_rate": 2.0230457819183367e-08,
+      "loss": 0.9069,
+      "step": 7120
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1421116860217116,
+      "learning_rate": 1.991475067503057e-08,
+      "loss": 0.9343,
+      "step": 7125
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15067170420364728,
+      "learning_rate": 1.9601476272596617e-08,
+      "loss": 0.9166,
+      "step": 7130
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1498555961927461,
+      "learning_rate": 1.9290636199340625e-08,
+      "loss": 0.9217,
+      "step": 7135
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.14144813336872367,
+      "learning_rate": 1.8982232030386237e-08,
+      "loss": 0.935,
+      "step": 7140
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15362836020907553,
+      "learning_rate": 1.867626532851335e-08,
+      "loss": 0.9342,
+      "step": 7145
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1593688707382894,
+      "learning_rate": 1.8372737644151094e-08,
+      "loss": 0.9469,
+      "step": 7150
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15826169338318843,
+      "learning_rate": 1.807165051536852e-08,
+      "loss": 0.951,
+      "step": 7155
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.15529080960970387,
+      "learning_rate": 1.7773005467868308e-08,
+      "loss": 0.934,
+      "step": 7160
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.14808986389455056,
+      "learning_rate": 1.747680401497803e-08,
+      "loss": 0.9645,
+      "step": 7165
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.1398305917908475,
+      "learning_rate": 1.7183047657642668e-08,
+      "loss": 0.9191,
+      "step": 7170
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14912454905006825,
+      "learning_rate": 1.6891737884417612e-08,
+      "loss": 0.9425,
+      "step": 7175
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.13947656404275413,
+      "learning_rate": 1.6602876171460112e-08,
+      "loss": 0.888,
+      "step": 7180
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1580943870386505,
+      "learning_rate": 1.6316463982522677e-08,
+      "loss": 0.9141,
+      "step": 7185
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14277902950319307,
+      "learning_rate": 1.6032502768945187e-08,
+      "loss": 0.9617,
+      "step": 7190
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.15423089132360732,
+      "learning_rate": 1.5750993969647684e-08,
+      "loss": 0.9764,
+      "step": 7195
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1563954411271587,
+      "learning_rate": 1.547193901112315e-08,
+      "loss": 0.9307,
+      "step": 7200
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1376852483457104,
+      "learning_rate": 1.519533930743e-08,
+      "loss": 0.918,
+      "step": 7205
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1399902887724196,
+      "learning_rate": 1.4921196260185286e-08,
+      "loss": 0.9197,
+      "step": 7210
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14325213646636392,
+      "learning_rate": 1.464951125855729e-08,
+      "loss": 0.96,
+      "step": 7215
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14495122048010833,
+      "learning_rate": 1.4380285679258807e-08,
+      "loss": 0.9236,
+      "step": 7220
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.13850527791914688,
+      "learning_rate": 1.411352088653961e-08,
+      "loss": 0.9163,
+      "step": 7225
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1516574090520955,
+      "learning_rate": 1.3849218232180271e-08,
+      "loss": 0.9173,
+      "step": 7230
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14850370564098128,
+      "learning_rate": 1.3587379055484794e-08,
+      "loss": 0.9657,
+      "step": 7235
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.14781037598904162,
+      "learning_rate": 1.3328004683273885e-08,
+      "loss": 0.952,
+      "step": 7240
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1482715651557921,
+      "learning_rate": 1.307109642987847e-08,
+      "loss": 0.9507,
+      "step": 7245
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.14117868089463456,
+      "learning_rate": 1.2816655597132741e-08,
+      "loss": 0.9977,
+      "step": 7250
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15259420530695983,
+      "learning_rate": 1.2564683474367898e-08,
+      "loss": 0.9365,
+      "step": 7255
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.14690109737766083,
+      "learning_rate": 1.2315181338405089e-08,
+      "loss": 0.9269,
+      "step": 7260
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1442520456136398,
+      "learning_rate": 1.2068150453549642e-08,
+      "loss": 0.9267,
+      "step": 7265
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15614889958789588,
+      "learning_rate": 1.182359207158401e-08,
+      "loss": 0.9063,
+      "step": 7270
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.14155349041673665,
+      "learning_rate": 1.1581507431761839e-08,
+      "loss": 0.9052,
+      "step": 7275
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15331957056564893,
+      "learning_rate": 1.1341897760801577e-08,
+      "loss": 0.9478,
+      "step": 7280
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1449080566245621,
+      "learning_rate": 1.110476427288004e-08,
+      "loss": 0.961,
+      "step": 7285
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15363661791040062,
+      "learning_rate": 1.0870108169626691e-08,
+      "loss": 0.9474,
+      "step": 7290
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1482497331098839,
+      "learning_rate": 1.06379306401172e-08,
+      "loss": 0.9071,
+      "step": 7295
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15718376657557753,
+      "learning_rate": 1.040823286086745e-08,
+      "loss": 0.9379,
+      "step": 7300
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15751569816070118,
+      "learning_rate": 1.0181015995827824e-08,
+      "loss": 0.9708,
+      "step": 7305
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.15107845744459764,
+      "learning_rate": 9.956281196377037e-09,
+      "loss": 0.9187,
+      "step": 7310
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.13823530270431622,
+      "learning_rate": 9.734029601316363e-09,
+      "loss": 0.9538,
+      "step": 7315
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.1379236877233237,
+      "learning_rate": 9.514262336864087e-09,
+      "loss": 0.9271,
+      "step": 7320
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.14118759001503478,
+      "learning_rate": 9.296980516649456e-09,
+      "loss": 0.9312,
+      "step": 7325
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.13930705947755528,
+      "learning_rate": 9.082185241707063e-09,
+      "loss": 0.9003,
+      "step": 7330
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.140127918712579,
+      "learning_rate": 8.869877600471865e-09,
+      "loss": 0.9382,
+      "step": 7335
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14938811801669172,
+      "learning_rate": 8.660058668772675e-09,
+      "loss": 0.9601,
+      "step": 7340
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14408204104128175,
+      "learning_rate": 8.45272950982756e-09,
+      "loss": 0.9064,
+      "step": 7345
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.15047086155150724,
+      "learning_rate": 8.24789117423802e-09,
+      "loss": 0.9471,
+      "step": 7350
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14584734687802067,
+      "learning_rate": 8.045544699983753e-09,
+      "loss": 0.9313,
+      "step": 7355
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1446048773437771,
+      "learning_rate": 7.84569111241762e-09,
+      "loss": 0.9249,
+      "step": 7360
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14942869129542674,
+      "learning_rate": 7.648331424259968e-09,
+      "loss": 0.9465,
+      "step": 7365
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.13719069475176607,
+      "learning_rate": 7.45346663559393e-09,
+      "loss": 0.8949,
+      "step": 7370
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.15653544055409904,
+      "learning_rate": 7.2610977338601886e-09,
+      "loss": 0.936,
+      "step": 7375
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.17291892169116618,
+      "learning_rate": 7.07122569385199e-09,
+      "loss": 0.9537,
+      "step": 7380
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.136655901132649,
+      "learning_rate": 6.883851477710146e-09,
+      "loss": 0.9414,
+      "step": 7385
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1455088529773193,
+      "learning_rate": 6.698976034918258e-09,
+      "loss": 0.9137,
+      "step": 7390
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1463970210162346,
+      "learning_rate": 6.516600302297948e-09,
+      "loss": 0.9401,
+      "step": 7395
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.14642661142513488,
+      "learning_rate": 6.336725204003857e-09,
+      "loss": 0.9146,
+      "step": 7400
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1526574840462531,
+      "learning_rate": 6.159351651519262e-09,
+      "loss": 0.9122,
+      "step": 7405
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15126511554520636,
+      "learning_rate": 5.984480543651249e-09,
+      "loss": 0.9308,
+      "step": 7410
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15422937195320308,
+      "learning_rate": 5.812112766526322e-09,
+      "loss": 0.9489,
+      "step": 7415
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15273241831668485,
+      "learning_rate": 5.6422491935857464e-09,
+      "loss": 0.9333,
+      "step": 7420
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15290915195544652,
+      "learning_rate": 5.474890685581157e-09,
+      "loss": 0.9598,
+      "step": 7425
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.13926317777592223,
+      "learning_rate": 5.310038090570512e-09,
+      "loss": 0.9239,
+      "step": 7430
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.14427428408626766,
+      "learning_rate": 5.147692243913149e-09,
+      "loss": 0.9199,
+      "step": 7435
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1546283344078747,
+      "learning_rate": 4.9878539682662865e-09,
+      "loss": 0.9246,
+      "step": 7440
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.14921223355979293,
+      "learning_rate": 4.830524073580144e-09,
+      "loss": 0.9113,
+      "step": 7445
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15069094141152864,
+      "learning_rate": 4.675703357094496e-09,
+      "loss": 0.9551,
+      "step": 7450
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1392308339727246,
+      "learning_rate": 4.523392603334175e-09,
+      "loss": 0.9534,
+      "step": 7455
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1478557506150279,
+      "learning_rate": 4.373592584105301e-09,
+      "loss": 0.9641,
+      "step": 7460
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.14707122573264736,
+      "learning_rate": 4.226304058491281e-09,
+      "loss": 0.9473,
+      "step": 7465
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1529307039024127,
+      "learning_rate": 4.081527772849036e-09,
+      "loss": 0.9308,
+      "step": 7470
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.1484757993225338,
+      "learning_rate": 3.93926446080517e-09,
+      "loss": 0.9066,
+      "step": 7475
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.15884901449960445,
+      "learning_rate": 3.799514843252194e-09,
+      "loss": 0.9207,
+      "step": 7480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.16322511661932754,
+      "learning_rate": 3.6622796283451442e-09,
+      "loss": 0.9598,
+      "step": 7485
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1499514473791366,
+      "learning_rate": 3.5275595114975243e-09,
+      "loss": 0.9256,
+      "step": 7490
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14447013304340453,
+      "learning_rate": 3.3953551753783117e-09,
+      "loss": 0.9288,
+      "step": 7495
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15033957728876848,
+      "learning_rate": 3.2656672899080693e-09,
+      "loss": 0.9492,
+      "step": 7500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15156539968975968,
+      "learning_rate": 3.1384965122557284e-09,
+      "loss": 0.9497,
+      "step": 7505
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15383668461696767,
+      "learning_rate": 3.013843486835366e-09,
+      "loss": 0.9355,
+      "step": 7510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14050741672798409,
+      "learning_rate": 2.891708845302654e-09,
+      "loss": 0.9517,
+      "step": 7515
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15193093061711,
+      "learning_rate": 2.772093206551973e-09,
+      "loss": 0.9506,
+      "step": 7520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.142649475661762,
+      "learning_rate": 2.654997176713136e-09,
+      "loss": 0.9623,
+      "step": 7525
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14775132238275115,
+      "learning_rate": 2.5404213491481696e-09,
+      "loss": 0.9317,
+      "step": 7530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14524940131234637,
+      "learning_rate": 2.428366304448537e-09,
+      "loss": 0.9199,
+      "step": 7535
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1419261556486215,
+      "learning_rate": 2.318832610432142e-09,
+      "loss": 0.9469,
+      "step": 7540
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.15542135039117955,
+      "learning_rate": 2.2118208221403868e-09,
+      "loss": 0.9273,
+      "step": 7545
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.14884292747415864,
+      "learning_rate": 2.1073314818353392e-09,
+      "loss": 0.9158,
+      "step": 7550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.16433309531163842,
+      "learning_rate": 2.0053651189971265e-09,
+      "loss": 0.9595,
+      "step": 7555
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.1608047056562379,
+      "learning_rate": 1.9059222503211014e-09,
+      "loss": 0.9071,
+      "step": 7560
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1481062932242135,
+      "learning_rate": 1.8090033797152905e-09,
+      "loss": 0.9451,
+      "step": 7565
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15977731543366752,
+      "learning_rate": 1.7146089982978396e-09,
+      "loss": 0.955,
+      "step": 7570
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14166877924261584,
+      "learning_rate": 1.6227395843945169e-09,
+      "loss": 0.9229,
+      "step": 7575
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14816812797725118,
+      "learning_rate": 1.5333956035363249e-09,
+      "loss": 0.9338,
+      "step": 7580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1570920954974337,
+      "learning_rate": 1.446577508457114e-09,
+      "loss": 0.8897,
+      "step": 7585
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1410974050472185,
+      "learning_rate": 1.3622857390912512e-09,
+      "loss": 0.9348,
+      "step": 7590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15064424041010413,
+      "learning_rate": 1.2805207225713432e-09,
+      "loss": 0.9542,
+      "step": 7595
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14474532868560921,
+      "learning_rate": 1.2012828732262947e-09,
+      "loss": 0.904,
+      "step": 7600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15156204315283767,
+      "learning_rate": 1.1245725925789762e-09,
+      "loss": 0.98,
+      "step": 7605
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1341421119803407,
+      "learning_rate": 1.0503902693443923e-09,
+      "loss": 0.9225,
+      "step": 7610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1513902937089699,
+      "learning_rate": 9.787362794274056e-10,
+      "loss": 0.8951,
+      "step": 7615
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14479590415987184,
+      "learning_rate": 9.096109859211831e-10,
+      "loss": 0.9181,
+      "step": 7620
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14267405232958089,
+      "learning_rate": 8.430147391051967e-10,
+      "loss": 0.9274,
+      "step": 7625
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.15122488038687992,
+      "learning_rate": 7.789478764432811e-10,
+      "loss": 0.8939,
+      "step": 7630
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.14094944830250608,
+      "learning_rate": 7.174107225823567e-10,
+      "loss": 0.9221,
+      "step": 7635
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15176791458444605,
+      "learning_rate": 6.584035893503203e-10,
+      "loss": 0.9557,
+      "step": 7640
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15278131399009062,
+      "learning_rate": 6.019267757547131e-10,
+      "loss": 0.9597,
+      "step": 7645
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1525905051128073,
+      "learning_rate": 5.479805679811101e-10,
+      "loss": 0.9427,
+      "step": 7650
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.15564705276851742,
+      "learning_rate": 4.965652393918996e-10,
+      "loss": 0.9121,
+      "step": 7655
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14450593104481904,
+      "learning_rate": 4.4768105052450656e-10,
+      "loss": 0.9168,
+      "step": 7660
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.16352089895487387,
+      "learning_rate": 4.0132824909039354e-10,
+      "loss": 0.9412,
+      "step": 7665
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.13655421313515312,
+      "learning_rate": 3.575070699736726e-10,
+      "loss": 0.919,
+      "step": 7670
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14332597532453037,
+      "learning_rate": 3.1621773522988446e-10,
+      "loss": 0.9376,
+      "step": 7675
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14351436422074912,
+      "learning_rate": 2.7746045408499897e-10,
+      "loss": 0.9682,
+      "step": 7680
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1445493727801075,
+      "learning_rate": 2.4123542293424947e-10,
+      "loss": 0.9182,
+      "step": 7685
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14172530359376276,
+      "learning_rate": 2.0754282534113377e-10,
+      "loss": 0.9331,
+      "step": 7690
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14336488389245616,
+      "learning_rate": 1.7638283203663673e-10,
+      "loss": 0.9665,
+      "step": 7695
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1580353039492496,
+      "learning_rate": 1.477556009180647e-10,
+      "loss": 0.8873,
+      "step": 7700
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1628976225865982,
+      "learning_rate": 1.216612770486569e-10,
+      "loss": 0.9339,
+      "step": 7705
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14850661402040227,
+      "learning_rate": 9.809999265636416e-11,
+      "loss": 0.9293,
+      "step": 7710
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.14544412310554528,
+      "learning_rate": 7.707186713357127e-11,
+      "loss": 0.9969,
+      "step": 7715
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.14763772393490937,
+      "learning_rate": 5.857700703643109e-11,
+      "loss": 0.9527,
+      "step": 7720
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.13590000625498125,
+      "learning_rate": 4.261550608414266e-11,
+      "loss": 0.9408,
+      "step": 7725
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15631513020595628,
+      "learning_rate": 2.9187445158618264e-11,
+      "loss": 0.9406,
+      "step": 7730
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.16641878770004168,
+      "learning_rate": 1.8292892303983787e-11,
+      "loss": 0.9383,
+      "step": 7735
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.15451260400151193,
+      "learning_rate": 9.93190272635669e-12,
+      "loss": 0.8986,
+      "step": 7740
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1537998952423965,
+      "learning_rate": 4.104518793512923e-12,
+      "loss": 0.9428,
+      "step": 7745
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1502803362427983,
+      "learning_rate": 8.107700345538759e-13,
+      "loss": 0.9321,
+      "step": 7750
+    },
     {
       "epoch": 1.0,
-      "eval_loss": 1.0769915580749512,
-      "eval_runtime": 144.6355,
-      "eval_samples_per_second": 4.245,
-      "eval_steps_per_second": 0.138,
-      "step": 4
+      "eval_loss": 0.9774965643882751,
+      "eval_runtime": 10236.3981,
+      "eval_samples_per_second": 3.028,
+      "eval_steps_per_second": 0.095,
+      "step": 7754
     },
     {
       "epoch": 1.0,
-      "step": 4,
-      "total_flos": 7015829078016.0,
-      "train_loss": 1.0347997546195984,
-      "train_runtime": 278.4186,
-      "train_samples_per_second": 0.431,
-      "train_steps_per_second": 0.014
+      "step": 7754,
+      "total_flos": 1.3600184667734016e+16,
+      "train_loss": 0.9098342092251821,
+      "train_runtime": 184655.0174,
+      "train_samples_per_second": 1.344,
+      "train_steps_per_second": 0.042
     }
   ],
   "logging_steps": 5,
-  "max_steps": 4,
+  "max_steps": 7754,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
-  "total_flos": 7015829078016.0,
+  "total_flos": 1.3600184667734016e+16,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null