diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.500428728920828,
+  "epoch": 0.7506430933812421,
   "eval_steps": 766,
-  "global_step": 1532,
+  "global_step": 2298,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10755,6 +10755,5376 @@
       "eval_samples_per_second": 5.155,
       "eval_steps_per_second": 2.578,
       "step": 1532
+    },
+    {
+      "epoch": 0.5007553795271732,
+      "grad_norm": 0.28496748208999634,
+      "learning_rate": 5.1173685523057366e-05,
+      "loss": 0.8344,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5010820301335185,
+      "grad_norm": 0.2947223484516144,
+      "learning_rate": 5.112153067576988e-05,
+      "loss": 0.9075,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5014086807398637,
+      "grad_norm": 0.3117678463459015,
+      "learning_rate": 5.106937460755636e-05,
+      "loss": 0.8458,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5017353313462088,
+      "grad_norm": 0.3499720096588135,
+      "learning_rate": 5.1017217375195206e-05,
+      "loss": 0.8336,
+      "step": 1536
+    },
+    {
+      "epoch": 0.502061981952554,
+      "grad_norm": 0.3558485209941864,
+      "learning_rate": 5.0965059035466054e-05,
+      "loss": 0.8773,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5023886325588992,
+      "grad_norm": 0.3849354386329651,
+      "learning_rate": 5.091289964514977e-05,
+      "loss": 0.853,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5027152831652444,
+      "grad_norm": 0.397390216588974,
+      "learning_rate": 5.086073926102837e-05,
+      "loss": 0.9433,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5030419337715896,
+      "grad_norm": 0.40508022904396057,
+      "learning_rate": 5.080857793988491e-05,
+      "loss": 0.9347,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5033685843779347,
+      "grad_norm": 0.477297842502594,
+      "learning_rate": 5.0756415738503526e-05,
+      "loss": 1.0416,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5036952349842799,
+      "grad_norm": 0.4862549901008606,
+      "learning_rate": 5.070425271366925e-05,
+      "loss": 1.0573,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5040218855906251,
+      "grad_norm": 0.5033325552940369,
+      "learning_rate": 5.0652088922168086e-05,
+      "loss": 0.9036,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5043485361969703,
+      "grad_norm": 0.5513638854026794,
+      "learning_rate": 5.059992442078677e-05,
+      "loss": 1.0243,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5046751868033155,
+      "grad_norm": 0.6637901663780212,
+      "learning_rate": 5.05477592663129e-05,
+      "loss": 1.191,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5050018374096606,
+      "grad_norm": 0.8518434166908264,
+      "learning_rate": 5.0495593515534744e-05,
+      "loss": 1.3829,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5053284880160058,
+      "grad_norm": 0.9972793459892273,
+      "learning_rate": 5.0443427225241246e-05,
+      "loss": 1.5652,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5056551386223511,
+      "grad_norm": 1.058459758758545,
+      "learning_rate": 5.039126045222189e-05,
+      "loss": 1.1677,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5059817892286963,
+      "grad_norm": 1.164061188697815,
+      "learning_rate": 5.0339093253266735e-05,
+      "loss": 1.4483,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5063084398350415,
+      "grad_norm": 2.050952911376953,
+      "learning_rate": 5.0286925685166275e-05,
+      "loss": 1.8747,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5066350904413867,
+      "grad_norm": 0.19119663536548615,
+      "learning_rate": 5.023475780471143e-05,
+      "loss": 0.7206,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5069617410477318,
+      "grad_norm": 0.24785088002681732,
+      "learning_rate": 5.018258966869343e-05,
+      "loss": 0.807,
+      "step": 1552
+    },
+    {
+      "epoch": 0.507288391654077,
+      "grad_norm": 0.23693665862083435,
+      "learning_rate": 5.01304213339038e-05,
+      "loss": 0.8574,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5076150422604222,
+      "grad_norm": 0.28269410133361816,
+      "learning_rate": 5.007825285713429e-05,
+      "loss": 0.8504,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5079416928667674,
+      "grad_norm": 0.27684375643730164,
+      "learning_rate": 5.002608429517677e-05,
+      "loss": 0.8391,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5082683434731126,
+      "grad_norm": 0.28662434220314026,
+      "learning_rate": 4.9973915704823244e-05,
+      "loss": 0.7969,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5085949940794577,
+      "grad_norm": 0.2877679169178009,
+      "learning_rate": 4.992174714286573e-05,
+      "loss": 0.8225,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5089216446858029,
+      "grad_norm": 0.3176509439945221,
+      "learning_rate": 4.9869578666096214e-05,
+      "loss": 0.9153,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5092482952921481,
+      "grad_norm": 0.328567773103714,
+      "learning_rate": 4.981741033130657e-05,
+      "loss": 0.8502,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5095749458984933,
+      "grad_norm": 0.3393353819847107,
+      "learning_rate": 4.976524219528858e-05,
+      "loss": 0.8563,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5099015965048385,
+      "grad_norm": 0.339493989944458,
+      "learning_rate": 4.9713074314833716e-05,
+      "loss": 0.8235,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5102282471111838,
+      "grad_norm": 0.3445899188518524,
+      "learning_rate": 4.9660906746733276e-05,
+      "loss": 0.9349,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5105548977175289,
+      "grad_norm": 0.386586993932724,
+      "learning_rate": 4.960873954777813e-05,
+      "loss": 0.9623,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5108815483238741,
+      "grad_norm": 0.411825567483902,
+      "learning_rate": 4.955657277475877e-05,
+      "loss": 0.9959,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5112081989302193,
+      "grad_norm": 0.4769602119922638,
+      "learning_rate": 4.950440648446527e-05,
+      "loss": 1.0555,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5115348495365645,
+      "grad_norm": 0.4826885163784027,
+      "learning_rate": 4.945224073368711e-05,
+      "loss": 1.0665,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5118615001429097,
+      "grad_norm": 0.5731224417686462,
+      "learning_rate": 4.9400075579213245e-05,
+      "loss": 1.1305,
+      "step": 1567
+    },
+    {
+      "epoch": 0.5121881507492548,
+      "grad_norm": 0.5455636978149414,
+      "learning_rate": 4.9347911077831946e-05,
+      "loss": 1.103,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5125148013556,
+      "grad_norm": 0.6146492958068848,
+      "learning_rate": 4.9295747286330745e-05,
+      "loss": 1.184,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5128414519619452,
+      "grad_norm": 0.7358331680297852,
+      "learning_rate": 4.9243584261496486e-05,
+      "loss": 1.3271,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5131681025682904,
+      "grad_norm": 0.9477192163467407,
+      "learning_rate": 4.919142206011509e-05,
+      "loss": 1.3478,
+      "step": 1571
+    },
+    {
+      "epoch": 0.5134947531746356,
+      "grad_norm": 1.0120941400527954,
+      "learning_rate": 4.9139260738971644e-05,
+      "loss": 1.4512,
+      "step": 1572
+    },
+    {
+      "epoch": 0.5138214037809807,
+      "grad_norm": 1.1496292352676392,
+      "learning_rate": 4.908710035485024e-05,
+      "loss": 1.097,
+      "step": 1573
+    },
+    {
+      "epoch": 0.5141480543873259,
+      "grad_norm": 1.183207631111145,
+      "learning_rate": 4.903494096453396e-05,
+      "loss": 1.5095,
+      "step": 1574
+    },
+    {
+      "epoch": 0.5144747049936711,
+      "grad_norm": 1.803774356842041,
+      "learning_rate": 4.8982782624804806e-05,
+      "loss": 2.2187,
+      "step": 1575
+    },
+    {
+      "epoch": 0.5148013556000164,
+      "grad_norm": 0.2078971266746521,
+      "learning_rate": 4.8930625392443657e-05,
+      "loss": 0.8109,
+      "step": 1576
+    },
+    {
+      "epoch": 0.5151280062063616,
+      "grad_norm": 0.2547096610069275,
+      "learning_rate": 4.887846932423012e-05,
+      "loss": 0.8169,
+      "step": 1577
+    },
+    {
+      "epoch": 0.5154546568127067,
+      "grad_norm": 0.2599959075450897,
+      "learning_rate": 4.882631447694264e-05,
+      "loss": 0.8768,
+      "step": 1578
+    },
+    {
+      "epoch": 0.5157813074190519,
+      "grad_norm": 0.28119954466819763,
+      "learning_rate": 4.877416090735822e-05,
+      "loss": 0.8372,
+      "step": 1579
+    },
+    {
+      "epoch": 0.5161079580253971,
+      "grad_norm": 0.28370240330696106,
+      "learning_rate": 4.872200867225257e-05,
+      "loss": 0.8628,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5164346086317423,
+      "grad_norm": 0.31476640701293945,
+      "learning_rate": 4.866985782839988e-05,
+      "loss": 0.9236,
+      "step": 1581
+    },
+    {
+      "epoch": 0.5167612592380875,
+      "grad_norm": 0.3151565492153168,
+      "learning_rate": 4.861770843257283e-05,
+      "loss": 0.8242,
+      "step": 1582
+    },
+    {
+      "epoch": 0.5170879098444326,
+      "grad_norm": 0.31610119342803955,
+      "learning_rate": 4.856556054154258e-05,
+      "loss": 0.7948,
+      "step": 1583
+    },
+    {
+      "epoch": 0.5174145604507778,
+      "grad_norm": 0.3681997060775757,
+      "learning_rate": 4.851341421207859e-05,
+      "loss": 0.9726,
+      "step": 1584
+    },
+    {
+      "epoch": 0.517741211057123,
+      "grad_norm": 0.3535841107368469,
+      "learning_rate": 4.846126950094864e-05,
+      "loss": 0.8712,
+      "step": 1585
+    },
+    {
+      "epoch": 0.5180678616634682,
+      "grad_norm": 0.378167986869812,
+      "learning_rate": 4.840912646491877e-05,
+      "loss": 1.0131,
+      "step": 1586
+    },
+    {
+      "epoch": 0.5183945122698134,
+      "grad_norm": 0.39976951479911804,
+      "learning_rate": 4.8356985160753155e-05,
+      "loss": 0.8891,
+      "step": 1587
+    },
+    {
+      "epoch": 0.5187211628761585,
+      "grad_norm": 0.39361557364463806,
+      "learning_rate": 4.830484564521414e-05,
+      "loss": 0.9145,
+      "step": 1588
+    },
+    {
+      "epoch": 0.5190478134825037,
+      "grad_norm": 0.44923731684684753,
+      "learning_rate": 4.8252707975062064e-05,
+      "loss": 0.9854,
+      "step": 1589
+    },
+    {
+      "epoch": 0.519374464088849,
+      "grad_norm": 0.4579111337661743,
+      "learning_rate": 4.820057220705529e-05,
+      "loss": 1.0045,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5197011146951942,
+      "grad_norm": 0.48675835132598877,
+      "learning_rate": 4.814843839795011e-05,
+      "loss": 1.024,
+      "step": 1591
+    },
+    {
+      "epoch": 0.5200277653015394,
+      "grad_norm": 0.5352386832237244,
+      "learning_rate": 4.8096306604500674e-05,
+      "loss": 1.0537,
+      "step": 1592
+    },
+    {
+      "epoch": 0.5203544159078846,
+      "grad_norm": 0.578682005405426,
+      "learning_rate": 4.8044176883458946e-05,
+      "loss": 1.0153,
+      "step": 1593
+    },
+    {
+      "epoch": 0.5206810665142297,
+      "grad_norm": 0.6972869634628296,
+      "learning_rate": 4.799204929157463e-05,
+      "loss": 1.1377,
+      "step": 1594
+    },
+    {
+      "epoch": 0.5210077171205749,
+      "grad_norm": 0.7344182133674622,
+      "learning_rate": 4.793992388559509e-05,
+      "loss": 1.28,
+      "step": 1595
+    },
+    {
+      "epoch": 0.5213343677269201,
+      "grad_norm": 0.9843777418136597,
+      "learning_rate": 4.788780072226537e-05,
+      "loss": 1.2008,
+      "step": 1596
+    },
+    {
+      "epoch": 0.5216610183332653,
+      "grad_norm": 1.1680152416229248,
+      "learning_rate": 4.783567985832799e-05,
+      "loss": 1.2584,
+      "step": 1597
+    },
+    {
+      "epoch": 0.5219876689396105,
+      "grad_norm": 1.1478536128997803,
+      "learning_rate": 4.7783561350523065e-05,
+      "loss": 1.2884,
+      "step": 1598
+    },
+    {
+      "epoch": 0.5223143195459556,
+      "grad_norm": 1.6914291381835938,
+      "learning_rate": 4.7731445255588056e-05,
+      "loss": 1.4843,
+      "step": 1599
+    },
+    {
+      "epoch": 0.5226409701523008,
+      "grad_norm": 2.582030773162842,
+      "learning_rate": 4.767933163025781e-05,
+      "loss": 2.0824,
+      "step": 1600
+    },
+    {
+      "epoch": 0.522967620758646,
+      "grad_norm": 0.21092741191387177,
+      "learning_rate": 4.762722053126457e-05,
+      "loss": 0.7676,
+      "step": 1601
+    },
+    {
+      "epoch": 0.5232942713649912,
+      "grad_norm": 0.22836419939994812,
+      "learning_rate": 4.7575112015337724e-05,
+      "loss": 0.7556,
+      "step": 1602
+    },
+    {
+      "epoch": 0.5236209219713364,
+      "grad_norm": 0.2456665188074112,
+      "learning_rate": 4.7523006139203883e-05,
+      "loss": 0.7892,
+      "step": 1603
+    },
+    {
+      "epoch": 0.5239475725776817,
+      "grad_norm": 0.27531933784484863,
+      "learning_rate": 4.747090295958683e-05,
+      "loss": 0.8169,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5242742231840268,
+      "grad_norm": 0.2720637619495392,
+      "learning_rate": 4.7418802533207335e-05,
+      "loss": 0.7968,
+      "step": 1605
+    },
+    {
+      "epoch": 0.524600873790372,
+      "grad_norm": 0.30058741569519043,
+      "learning_rate": 4.736670491678325e-05,
+      "loss": 0.872,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5249275243967172,
+      "grad_norm": 0.2886933982372284,
+      "learning_rate": 4.7314610167029297e-05,
+      "loss": 0.8462,
+      "step": 1607
+    },
+    {
+      "epoch": 0.5252541750030624,
+      "grad_norm": 0.27622488141059875,
+      "learning_rate": 4.7262518340657115e-05,
+      "loss": 0.8096,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5255808256094076,
+      "grad_norm": 0.31346070766448975,
+      "learning_rate": 4.721042949437516e-05,
+      "loss": 0.908,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5259074762157527,
+      "grad_norm": 0.338704913854599,
+      "learning_rate": 4.7158343684888645e-05,
+      "loss": 0.9125,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5262341268220979,
+      "grad_norm": 0.3269166648387909,
+      "learning_rate": 4.710626096889946e-05,
+      "loss": 0.8312,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5265607774284431,
+      "grad_norm": 0.3700317442417145,
+      "learning_rate": 4.7054181403106146e-05,
+      "loss": 0.8932,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5268874280347883,
+      "grad_norm": 0.3964787423610687,
+      "learning_rate": 4.700210504420378e-05,
+      "loss": 0.9223,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5272140786411335,
+      "grad_norm": 0.4625542163848877,
+      "learning_rate": 4.695003194888402e-05,
+      "loss": 0.9979,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5275407292474786,
+      "grad_norm": 0.46109652519226074,
+      "learning_rate": 4.68979621738349e-05,
+      "loss": 1.0069,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5278673798538238,
+      "grad_norm": 0.46963831782341003,
+      "learning_rate": 4.684589577574083e-05,
+      "loss": 0.9752,
+      "step": 1616
+    },
+    {
+      "epoch": 0.528194030460169,
+      "grad_norm": 0.49279284477233887,
+      "learning_rate": 4.679383281128265e-05,
+      "loss": 1.0698,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5285206810665142,
+      "grad_norm": 0.5191306471824646,
+      "learning_rate": 4.674177333713732e-05,
+      "loss": 1.0913,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5288473316728595,
+      "grad_norm": 0.587307333946228,
+      "learning_rate": 4.668971740997814e-05,
+      "loss": 1.0703,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5291739822792046,
+      "grad_norm": 0.6988803148269653,
+      "learning_rate": 4.663766508647443e-05,
+      "loss": 1.1106,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5295006328855498,
+      "grad_norm": 0.763789176940918,
+      "learning_rate": 4.658561642329163e-05,
+      "loss": 1.3298,
+      "step": 1621
+    },
+    {
+      "epoch": 0.529827283491895,
+      "grad_norm": 0.9712387323379517,
+      "learning_rate": 4.6533571477091245e-05,
+      "loss": 1.3375,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5301539340982402,
+      "grad_norm": 1.0462433099746704,
+      "learning_rate": 4.6481530304530646e-05,
+      "loss": 1.1795,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5304805847045854,
+      "grad_norm": 1.2691247463226318,
+      "learning_rate": 4.6429492962263185e-05,
+      "loss": 1.7746,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5308072353109305,
+      "grad_norm": 1.5794727802276611,
+      "learning_rate": 4.6377459506937966e-05,
+      "loss": 1.9366,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5311338859172757,
+      "grad_norm": 0.19865591824054718,
+      "learning_rate": 4.632542999519989e-05,
+      "loss": 0.6034,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5314605365236209,
+      "grad_norm": 0.2409842610359192,
+      "learning_rate": 4.627340448368961e-05,
+      "loss": 0.7811,
+      "step": 1627
+    },
+    {
+      "epoch": 0.5317871871299661,
+      "grad_norm": 0.26515185832977295,
+      "learning_rate": 4.6221383029043355e-05,
+      "loss": 0.8836,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5321138377363113,
+      "grad_norm": 0.2678782045841217,
+      "learning_rate": 4.616936568789295e-05,
+      "loss": 0.8236,
+      "step": 1629
+    },
+    {
+      "epoch": 0.5324404883426564,
+      "grad_norm": 0.27880924940109253,
+      "learning_rate": 4.611735251686581e-05,
+      "loss": 0.8432,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5327671389490016,
+      "grad_norm": 0.2928534150123596,
+      "learning_rate": 4.60653435725847e-05,
+      "loss": 0.8542,
+      "step": 1631
+    },
+    {
+      "epoch": 0.5330937895553468,
+      "grad_norm": 0.3031875789165497,
+      "learning_rate": 4.6013338911667896e-05,
+      "loss": 0.8876,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5334204401616921,
+      "grad_norm": 0.3030424118041992,
+      "learning_rate": 4.5961338590728937e-05,
+      "loss": 0.911,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5337470907680373,
+      "grad_norm": 0.33176282048225403,
+      "learning_rate": 4.5909342666376635e-05,
+      "loss": 0.8456,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5340737413743825,
+      "grad_norm": 0.3273261785507202,
+      "learning_rate": 4.585735119521508e-05,
+      "loss": 0.8488,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5344003919807276,
+      "grad_norm": 0.3342040181159973,
+      "learning_rate": 4.580536423384342e-05,
+      "loss": 0.8439,
+      "step": 1636
+    },
+    {
+      "epoch": 0.5347270425870728,
+      "grad_norm": 0.36872991919517517,
+      "learning_rate": 4.575338183885601e-05,
+      "loss": 1.0165,
+      "step": 1637
+    },
+    {
+      "epoch": 0.535053693193418,
+      "grad_norm": 0.38245999813079834,
+      "learning_rate": 4.570140406684214e-05,
+      "loss": 0.9385,
+      "step": 1638
+    },
+    {
+      "epoch": 0.5353803437997632,
+      "grad_norm": 0.39496734738349915,
+      "learning_rate": 4.564943097438608e-05,
+      "loss": 0.7809,
+      "step": 1639
+    },
+    {
+      "epoch": 0.5357069944061084,
+      "grad_norm": 0.4122662842273712,
+      "learning_rate": 4.559746261806704e-05,
+      "loss": 0.8743,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5360336450124535,
+      "grad_norm": 0.4760144054889679,
+      "learning_rate": 4.554549905445908e-05,
+      "loss": 1.0506,
+      "step": 1641
+    },
+    {
+      "epoch": 0.5363602956187987,
+      "grad_norm": 0.5267740488052368,
+      "learning_rate": 4.549354034013097e-05,
+      "loss": 1.0937,
+      "step": 1642
+    },
+    {
+      "epoch": 0.5366869462251439,
+      "grad_norm": 0.537794291973114,
+      "learning_rate": 4.5441586531646305e-05,
+      "loss": 1.0207,
+      "step": 1643
+    },
+    {
+      "epoch": 0.5370135968314891,
+      "grad_norm": 0.5599216222763062,
+      "learning_rate": 4.5389637685563244e-05,
+      "loss": 1.0285,
+      "step": 1644
+    },
+    {
+      "epoch": 0.5373402474378343,
+      "grad_norm": 0.6215500235557556,
+      "learning_rate": 4.533769385843462e-05,
+      "loss": 1.1371,
+      "step": 1645
+    },
+    {
+      "epoch": 0.5376668980441794,
+      "grad_norm": 0.7406798601150513,
+      "learning_rate": 4.528575510680775e-05,
+      "loss": 1.2815,
+      "step": 1646
+    },
+    {
+      "epoch": 0.5379935486505247,
+      "grad_norm": 0.7674838304519653,
+      "learning_rate": 4.523382148722443e-05,
+      "loss": 1.1173,
+      "step": 1647
+    },
+    {
+      "epoch": 0.5383201992568699,
+      "grad_norm": 0.8467667102813721,
+      "learning_rate": 4.51818930562209e-05,
+      "loss": 1.0583,
+      "step": 1648
+    },
+    {
+      "epoch": 0.5386468498632151,
+      "grad_norm": 1.289118766784668,
+      "learning_rate": 4.512996987032773e-05,
+      "loss": 1.6204,
+      "step": 1649
+    },
+    {
+      "epoch": 0.5389735004695603,
+      "grad_norm": 1.2637457847595215,
+      "learning_rate": 4.507805198606979e-05,
+      "loss": 1.373,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5393001510759055,
+      "grad_norm": 0.22539332509040833,
+      "learning_rate": 4.502613945996617e-05,
+      "loss": 0.7505,
+      "step": 1651
+    },
+    {
+      "epoch": 0.5396268016822506,
+      "grad_norm": 0.23382221162319183,
+      "learning_rate": 4.497423234853011e-05,
+      "loss": 0.7676,
+      "step": 1652
+    },
+    {
+      "epoch": 0.5399534522885958,
+      "grad_norm": 0.24569697678089142,
+      "learning_rate": 4.4922330708269005e-05,
+      "loss": 0.8458,
+      "step": 1653
+    },
+    {
+      "epoch": 0.540280102894941,
+      "grad_norm": 0.2564908564090729,
+      "learning_rate": 4.487043459568426e-05,
+      "loss": 0.8189,
+      "step": 1654
+    },
+    {
+      "epoch": 0.5406067535012862,
+      "grad_norm": 0.27906638383865356,
+      "learning_rate": 4.4818544067271226e-05,
+      "loss": 0.8337,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5409334041076314,
+      "grad_norm": 0.283110648393631,
+      "learning_rate": 4.476665917951927e-05,
+      "loss": 0.8424,
+      "step": 1656
+    },
+    {
+      "epoch": 0.5412600547139765,
+      "grad_norm": 0.2966436445713043,
+      "learning_rate": 4.4714779988911525e-05,
+      "loss": 0.8646,
+      "step": 1657
+    },
+    {
+      "epoch": 0.5415867053203217,
+      "grad_norm": 0.3120681643486023,
+      "learning_rate": 4.466290655192498e-05,
+      "loss": 0.8508,
+      "step": 1658
+    },
+    {
+      "epoch": 0.5419133559266669,
+      "grad_norm": 0.3354993164539337,
+      "learning_rate": 4.461103892503036e-05,
+      "loss": 0.8795,
+      "step": 1659
+    },
+    {
+      "epoch": 0.5422400065330121,
+      "grad_norm": 0.35160109400749207,
+      "learning_rate": 4.455917716469199e-05,
+      "loss": 0.8703,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5425666571393574,
+      "grad_norm": 0.34476974606513977,
+      "learning_rate": 4.450732132736792e-05,
+      "loss": 0.9471,
+      "step": 1661
+    },
+    {
+      "epoch": 0.5428933077457025,
+      "grad_norm": 0.37731653451919556,
+      "learning_rate": 4.445547146950966e-05,
+      "loss": 0.8646,
+      "step": 1662
+    },
+    {
+      "epoch": 0.5432199583520477,
+      "grad_norm": 0.41084280610084534,
+      "learning_rate": 4.4403627647562275e-05,
+      "loss": 0.9606,
+      "step": 1663
+    },
+    {
+      "epoch": 0.5435466089583929,
+      "grad_norm": 0.40829023718833923,
+      "learning_rate": 4.435178991796422e-05,
+      "loss": 1.0439,
+      "step": 1664
+    },
+    {
+      "epoch": 0.5438732595647381,
+      "grad_norm": 0.45158272981643677,
+      "learning_rate": 4.4299958337147314e-05,
+      "loss": 1.0997,
+      "step": 1665
+    },
+    {
+      "epoch": 0.5441999101710833,
+      "grad_norm": 0.4397522509098053,
+      "learning_rate": 4.4248132961536726e-05,
+      "loss": 0.9821,
+      "step": 1666
+    },
+    {
+      "epoch": 0.5445265607774284,
+      "grad_norm": 0.47100958228111267,
+      "learning_rate": 4.419631384755083e-05,
+      "loss": 1.0299,
+      "step": 1667
+    },
+    {
+      "epoch": 0.5448532113837736,
+      "grad_norm": 0.5272468328475952,
+      "learning_rate": 4.414450105160117e-05,
+      "loss": 1.045,
+      "step": 1668
+    },
+    {
+      "epoch": 0.5451798619901188,
+      "grad_norm": 0.5478541851043701,
+      "learning_rate": 4.409269463009248e-05,
+      "loss": 1.1235,
+      "step": 1669
+    },
+    {
+      "epoch": 0.545506512596464,
+      "grad_norm": 0.6685848236083984,
+      "learning_rate": 4.404089463942247e-05,
+      "loss": 1.2074,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5458331632028092,
+      "grad_norm": 0.7062515616416931,
+      "learning_rate": 4.398910113598193e-05,
+      "loss": 1.2954,
+      "step": 1671
+    },
+    {
+      "epoch": 0.5461598138091543,
+      "grad_norm": 0.9928080439567566,
+      "learning_rate": 4.3937314176154523e-05,
+      "loss": 1.404,
+      "step": 1672
+    },
+    {
+      "epoch": 0.5464864644154995,
+      "grad_norm": 0.9474202990531921,
+      "learning_rate": 4.388553381631682e-05,
+      "loss": 1.1694,
+      "step": 1673
+    },
+    {
+      "epoch": 0.5468131150218447,
+      "grad_norm": 1.3454349040985107,
+      "learning_rate": 4.3833760112838196e-05,
+      "loss": 1.5713,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54713976562819,
+      "grad_norm": 1.6091099977493286,
+      "learning_rate": 4.3781993122080795e-05,
+      "loss": 1.7261,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5474664162345352,
+      "grad_norm": 0.20016787946224213,
+      "learning_rate": 4.3730232900399436e-05,
+      "loss": 0.8039,
+      "step": 1676
+    },
+    {
+      "epoch": 0.5477930668408804,
+      "grad_norm": 0.23649336397647858,
+      "learning_rate": 4.36784795041416e-05,
+      "loss": 0.802,
+      "step": 1677
+    },
+    {
+      "epoch": 0.5481197174472255,
+      "grad_norm": 0.2561361491680145,
+      "learning_rate": 4.3626732989647256e-05,
+      "loss": 0.825,
+      "step": 1678
+    },
+    {
+      "epoch": 0.5484463680535707,
+      "grad_norm": 0.2604573965072632,
+      "learning_rate": 4.357499341324901e-05,
+      "loss": 0.8709,
+      "step": 1679
+    },
+    {
+      "epoch": 0.5487730186599159,
+      "grad_norm": 0.270620733499527,
+      "learning_rate": 4.3523260831271786e-05,
+      "loss": 0.8286,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5490996692662611,
+      "grad_norm": 0.2832171320915222,
+      "learning_rate": 4.3471535300032996e-05,
+      "loss": 0.8517,
+      "step": 1681
+    },
+    {
+      "epoch": 0.5494263198726063,
+      "grad_norm": 0.27463555335998535,
+      "learning_rate": 4.341981687584231e-05,
+      "loss": 0.7702,
+      "step": 1682
+    },
+    {
+      "epoch": 0.5497529704789514,
+      "grad_norm": 0.3110555112361908,
+      "learning_rate": 4.336810561500167e-05,
+      "loss": 0.8606,
+      "step": 1683
+    },
+    {
+      "epoch": 0.5500796210852966,
+      "grad_norm": 0.33201920986175537,
+      "learning_rate": 4.331640157380527e-05,
+      "loss": 0.9015,
+      "step": 1684
+    },
+    {
+      "epoch": 0.5504062716916418,
+      "grad_norm": 0.31452447175979614,
+      "learning_rate": 4.326470480853938e-05,
+      "loss": 0.8095,
+      "step": 1685
+    },
+    {
+      "epoch": 0.550732922297987,
+      "grad_norm": 0.35205140709877014,
+      "learning_rate": 4.3213015375482376e-05,
+      "loss": 0.9332,
+      "step": 1686
+    },
+    {
+      "epoch": 0.5510595729043322,
+      "grad_norm": 0.38866329193115234,
+      "learning_rate": 4.3161333330904676e-05,
+      "loss": 0.9075,
+      "step": 1687
+    },
+    {
+      "epoch": 0.5513862235106773,
+      "grad_norm": 0.4142824113368988,
+      "learning_rate": 4.31096587310686e-05,
+      "loss": 0.9783,
+      "step": 1688
+    },
+    {
+      "epoch": 0.5517128741170226,
+      "grad_norm": 0.41084322333335876,
+      "learning_rate": 4.305799163222843e-05,
+      "loss": 0.917,
+      "step": 1689
+    },
+    {
+      "epoch": 0.5520395247233678,
+      "grad_norm": 0.4371437728404999,
+      "learning_rate": 4.300633209063022e-05,
+      "loss": 0.99,
+      "step": 1690
+    },
+    {
+      "epoch": 0.552366175329713,
+      "grad_norm": 0.43960168957710266,
+      "learning_rate": 4.295468016251183e-05,
+      "loss": 0.9339,
+      "step": 1691
+    },
+    {
+      "epoch": 0.5526928259360582,
+      "grad_norm": 0.5299369096755981,
+      "learning_rate": 4.290303590410283e-05,
+      "loss": 0.9887,
+      "step": 1692
+    },
+    {
+      "epoch": 0.5530194765424034,
+      "grad_norm": 0.49330300092697144,
+      "learning_rate": 4.285139937162445e-05,
+      "loss": 0.9703,
+      "step": 1693
+    },
+    {
+      "epoch": 0.5533461271487485,
+      "grad_norm": 0.5797840356826782,
+      "learning_rate": 4.2799770621289466e-05,
+      "loss": 1.0255,
+      "step": 1694
+    },
+    {
+      "epoch": 0.5536727777550937,
+      "grad_norm": 0.7644694447517395,
+      "learning_rate": 4.274814970930225e-05,
+      "loss": 1.3008,
+      "step": 1695
+    },
+    {
+      "epoch": 0.5539994283614389,
+      "grad_norm": 0.8262007236480713,
+      "learning_rate": 4.2696536691858556e-05,
+      "loss": 1.4008,
+      "step": 1696
+    },
+    {
+      "epoch": 0.5543260789677841,
+      "grad_norm": 0.9084567427635193,
+      "learning_rate": 4.264493162514565e-05,
+      "loss": 1.3747,
+      "step": 1697
+    },
+    {
+      "epoch": 0.5546527295741293,
+      "grad_norm": 1.4628161191940308,
+      "learning_rate": 4.2593334565342034e-05,
+      "loss": 1.4698,
+      "step": 1698
+    },
+    {
+      "epoch": 0.5549793801804744,
+      "grad_norm": 1.404968500137329,
+      "learning_rate": 4.254174556861755e-05,
+      "loss": 1.9086,
+      "step": 1699
+    },
+    {
+      "epoch": 0.5553060307868196,
+      "grad_norm": 1.955612301826477,
+      "learning_rate": 4.249016469113329e-05,
+      "loss": 1.8205,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5556326813931648,
+      "grad_norm": 0.21000787615776062,
+      "learning_rate": 4.243859198904142e-05,
+      "loss": 0.7214,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55595933199951,
+      "grad_norm": 0.20873922109603882,
+      "learning_rate": 4.238702751848531e-05,
+      "loss": 0.7236,
+      "step": 1702
+    },
+    {
+      "epoch": 0.5562859826058553,
+      "grad_norm": 0.2349824458360672,
+      "learning_rate": 4.23354713355993e-05,
+      "loss": 0.8209,
+      "step": 1703
+    },
+    {
+      "epoch": 0.5566126332122004,
+      "grad_norm": 0.2445916384458542,
+      "learning_rate": 4.22839234965087e-05,
+      "loss": 0.8256,
+      "step": 1704
+    },
+    {
+      "epoch": 0.5569392838185456,
+      "grad_norm": 0.2502254545688629,
+      "learning_rate": 4.22323840573298e-05,
+      "loss": 0.7655,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5572659344248908,
+      "grad_norm": 0.28509315848350525,
+      "learning_rate": 4.2180853074169674e-05,
+      "loss": 0.8841,
+      "step": 1706
+    },
+    {
+      "epoch": 0.557592585031236,
+      "grad_norm": 0.28722479939460754,
+      "learning_rate": 4.212933060312626e-05,
+      "loss": 0.8389,
+      "step": 1707
+    },
+    {
+      "epoch": 0.5579192356375812,
+      "grad_norm": 0.29638615250587463,
+      "learning_rate": 4.2077816700288165e-05,
+      "loss": 0.7169,
+      "step": 1708
+    },
+    {
+      "epoch": 0.5582458862439263,
+      "grad_norm": 0.32606878876686096,
+      "learning_rate": 4.202631142173471e-05,
+      "loss": 0.8639,
+      "step": 1709
+    },
+    {
+      "epoch": 0.5585725368502715,
+      "grad_norm": 0.31630176305770874,
+      "learning_rate": 4.1974814823535815e-05,
+      "loss": 0.8409,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5588991874566167,
+      "grad_norm": 0.3516806364059448,
+      "learning_rate": 4.192332696175195e-05,
+      "loss": 0.9856,
+      "step": 1711
+    },
+    {
+      "epoch": 0.5592258380629619,
+      "grad_norm": 0.33567309379577637,
+      "learning_rate": 4.187184789243406e-05,
+      "loss": 0.8381,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5595524886693071,
+      "grad_norm": 0.3585726320743561,
+      "learning_rate": 4.1820377671623565e-05,
+      "loss": 0.8846,
+      "step": 1713
+    },
+    {
+      "epoch": 0.5598791392756522,
+      "grad_norm": 0.3721924424171448,
+      "learning_rate": 4.176891635535218e-05,
+      "loss": 1.0201,
+      "step": 1714
+    },
+    {
+      "epoch": 0.5602057898819974,
+      "grad_norm": 0.3866981267929077,
+      "learning_rate": 4.1717463999642006e-05,
+      "loss": 0.8751,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5605324404883426,
+      "grad_norm": 0.43328696489334106,
+      "learning_rate": 4.1666020660505334e-05,
+      "loss": 0.9957,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5608590910946878,
+      "grad_norm": 0.47395583987236023,
+      "learning_rate": 4.1614586393944624e-05,
+      "loss": 0.9973,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5611857417010331,
+      "grad_norm": 0.5074653029441833,
+      "learning_rate": 4.156316125595254e-05,
+      "loss": 1.0893,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5615123923073783,
+      "grad_norm": 0.5391126275062561,
+      "learning_rate": 4.1511745302511726e-05,
+      "loss": 0.957,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5618390429137234,
+      "grad_norm": 0.6273775696754456,
+      "learning_rate": 4.1460338589594885e-05,
+      "loss": 1.314,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5621656935200686,
+      "grad_norm": 0.7783089280128479,
+      "learning_rate": 4.1408941173164625e-05,
+      "loss": 1.2052,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5624923441264138,
+      "grad_norm": 0.9860546588897705,
+      "learning_rate": 4.1357553109173416e-05,
+      "loss": 1.4157,
+      "step": 1722
+    },
+    {
+      "epoch": 0.562818994732759,
+      "grad_norm": 1.039588451385498,
+      "learning_rate": 4.130617445356363e-05,
+      "loss": 1.4652,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5631456453391042,
+      "grad_norm": 1.2351906299591064,
+      "learning_rate": 4.1254805262267296e-05,
+      "loss": 1.6218,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5634722959454493,
+      "grad_norm": 1.5483803749084473,
+      "learning_rate": 4.1203445591206204e-05,
+      "loss": 1.6209,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5637989465517945,
+      "grad_norm": 0.20997747778892517,
+      "learning_rate": 4.1152095496291764e-05,
+      "loss": 0.7082,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5641255971581397,
+      "grad_norm": 0.23678404092788696,
+      "learning_rate": 4.110075503342495e-05,
+      "loss": 0.7857,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5644522477644849,
+      "grad_norm": 0.2515248954296112,
+      "learning_rate": 4.1049424258496274e-05,
+      "loss": 0.8146,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5647788983708301,
+      "grad_norm": 0.27904850244522095,
+      "learning_rate": 4.099810322738568e-05,
+      "loss": 0.8893,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5651055489771752,
+      "grad_norm": 0.2811621427536011,
+      "learning_rate": 4.094679199596249e-05,
+      "loss": 0.8255,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5654321995835204,
+      "grad_norm": 0.2903529703617096,
+      "learning_rate": 4.0895490620085425e-05,
+      "loss": 0.8682,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5657588501898657,
+      "grad_norm": 0.29601550102233887,
+      "learning_rate": 4.084419915560238e-05,
+      "loss": 0.8585,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5660855007962109,
+      "grad_norm": 0.2924976944923401,
+      "learning_rate": 4.079291765835056e-05,
+      "loss": 0.8122,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5664121514025561,
+      "grad_norm": 0.3249928057193756,
+      "learning_rate": 4.074164618415625e-05,
+      "loss": 0.8835,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5667388020089013,
+      "grad_norm": 0.32690802216529846,
+      "learning_rate": 4.069038478883481e-05,
+      "loss": 0.893,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5670654526152464,
+      "grad_norm": 0.33490654826164246,
+      "learning_rate": 4.063913352819073e-05,
+      "loss": 0.9048,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5673921032215916,
+      "grad_norm": 0.36397576332092285,
+      "learning_rate": 4.058789245801736e-05,
+      "loss": 0.9796,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5677187538279368,
+      "grad_norm": 0.3783029615879059,
+      "learning_rate": 4.0536661634096977e-05,
+      "loss": 0.9378,
+      "step": 1738
+    },
+    {
+      "epoch": 0.568045404434282,
+      "grad_norm": 0.3783353567123413,
+      "learning_rate": 4.048544111220077e-05,
+      "loss": 0.8839,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5683720550406272,
+      "grad_norm": 0.4432784616947174,
+      "learning_rate": 4.043423094808863e-05,
+      "loss": 0.9956,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5686987056469723,
+      "grad_norm": 0.48436063528060913,
+      "learning_rate": 4.0383031197509226e-05,
+      "loss": 1.0038,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5690253562533175,
+      "grad_norm": 0.5452389717102051,
+      "learning_rate": 4.0331841916199867e-05,
+      "loss": 1.0593,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5693520068596627,
+      "grad_norm": 0.6084750890731812,
+      "learning_rate": 4.028066315988646e-05,
+      "loss": 1.1621,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5696786574660079,
+      "grad_norm": 0.6114654541015625,
+      "learning_rate": 4.0229494984283485e-05,
+      "loss": 1.1292,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5700053080723531,
+      "grad_norm": 0.7506347298622131,
+      "learning_rate": 4.017833744509387e-05,
+      "loss": 1.0845,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5703319586786983,
+      "grad_norm": 0.6660485863685608,
+      "learning_rate": 4.012719059800897e-05,
+      "loss": 1.0965,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5706586092850435,
+      "grad_norm": 0.8481147289276123,
+      "learning_rate": 4.0076054498708545e-05,
+      "loss": 1.34,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5709852598913887,
+      "grad_norm": 1.0453165769577026,
+      "learning_rate": 4.002492920286056e-05,
+      "loss": 1.3245,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5713119104977339,
+      "grad_norm": 1.374419093132019,
+      "learning_rate": 3.997381476612132e-05,
+      "loss": 1.4342,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5716385611040791,
+      "grad_norm": 1.6070213317871094,
+      "learning_rate": 3.992271124413527e-05,
+      "loss": 1.6983,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5719652117104242,
+      "grad_norm": 0.21011681854724884,
+      "learning_rate": 3.987161869253491e-05,
+      "loss": 0.7403,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5722918623167694,
+      "grad_norm": 0.25537699460983276,
+      "learning_rate": 3.982053716694092e-05,
+      "loss": 0.8284,
+      "step": 1752
+    },
+    {
+      "epoch": 0.5726185129231146,
+      "grad_norm": 0.2665100693702698,
+      "learning_rate": 3.976946672296187e-05,
+      "loss": 0.8535,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5729451635294598,
+      "grad_norm": 0.2679298520088196,
+      "learning_rate": 3.971840741619435e-05,
+      "loss": 0.8884,
+      "step": 1754
+    },
+    {
+      "epoch": 0.573271814135805,
+      "grad_norm": 0.27296388149261475,
+      "learning_rate": 3.9667359302222745e-05,
+      "loss": 0.8318,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5735984647421501,
+      "grad_norm": 0.2760670483112335,
+      "learning_rate": 3.961632243661929e-05,
+      "loss": 0.8232,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5739251153484953,
+      "grad_norm": 0.3013848662376404,
+      "learning_rate": 3.956529687494401e-05,
+      "loss": 0.9044,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5742517659548405,
+      "grad_norm": 0.2987593412399292,
+      "learning_rate": 3.9514282672744574e-05,
+      "loss": 0.9345,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5745784165611857,
+      "grad_norm": 0.3210752606391907,
+      "learning_rate": 3.946327988555632e-05,
+      "loss": 0.8186,
+      "step": 1759
+    },
+    {
+      "epoch": 0.574905067167531,
+      "grad_norm": 0.3270179033279419,
+      "learning_rate": 3.941228856890212e-05,
+      "loss": 0.8363,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5752317177738762,
+      "grad_norm": 0.3446122705936432,
+      "learning_rate": 3.93613087782924e-05,
+      "loss": 0.8473,
+      "step": 1761
+    },
+    {
+      "epoch": 0.5755583683802213,
+      "grad_norm": 0.37168315052986145,
+      "learning_rate": 3.931034056922501e-05,
+      "loss": 0.9515,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5758850189865665,
+      "grad_norm": 0.3891473710536957,
+      "learning_rate": 3.9259383997185216e-05,
+      "loss": 0.9106,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5762116695929117,
+      "grad_norm": 0.43029606342315674,
+      "learning_rate": 3.920843911764559e-05,
+      "loss": 1.0014,
+      "step": 1764
+    },
+    {
+      "epoch": 0.5765383201992569,
+      "grad_norm": 0.46320056915283203,
+      "learning_rate": 3.9157505986066e-05,
+      "loss": 1.0457,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5768649708056021,
+      "grad_norm": 0.47448238730430603,
+      "learning_rate": 3.9106584657893484e-05,
+      "loss": 1.0156,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5771916214119472,
+      "grad_norm": 0.48397818207740784,
+      "learning_rate": 3.9055675188562305e-05,
+      "loss": 0.9642,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5775182720182924,
+      "grad_norm": 0.5289010405540466,
+      "learning_rate": 3.900477763349374e-05,
+      "loss": 1.0498,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5778449226246376,
+      "grad_norm": 0.5463656187057495,
+      "learning_rate": 3.895389204809611e-05,
+      "loss": 1.0686,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5781715732309828,
+      "grad_norm": 0.641441822052002,
+      "learning_rate": 3.890301848776476e-05,
+      "loss": 1.1969,
+      "step": 1770
+    },
+    {
+      "epoch": 0.578498223837328,
+      "grad_norm": 0.7622916102409363,
+      "learning_rate": 3.885215700788188e-05,
+      "loss": 1.2451,
+      "step": 1771
+    },
+    {
+      "epoch": 0.5788248744436731,
+      "grad_norm": 0.8429807424545288,
+      "learning_rate": 3.880130766381655e-05,
+      "loss": 1.3678,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5791515250500183,
+      "grad_norm": 1.130521535873413,
+      "learning_rate": 3.875047051092462e-05,
+      "loss": 1.6483,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5794781756563636,
+      "grad_norm": 1.1499173641204834,
+      "learning_rate": 3.869964560454865e-05,
+      "loss": 1.4263,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5798048262627088,
+      "grad_norm": 1.3551626205444336,
+      "learning_rate": 3.864883300001793e-05,
+      "loss": 1.6207,
+      "step": 1775
+    },
+    {
+      "epoch": 0.580131476869054,
+      "grad_norm": 0.20396538078784943,
+      "learning_rate": 3.859803275264827e-05,
+      "loss": 0.8017,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5804581274753992,
+      "grad_norm": 0.23019051551818848,
+      "learning_rate": 3.854724491774212e-05,
+      "loss": 0.8158,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5807847780817443,
+      "grad_norm": 0.2596058249473572,
+      "learning_rate": 3.849646955058835e-05,
+      "loss": 0.7755,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5811114286880895,
+      "grad_norm": 0.26550063490867615,
+      "learning_rate": 3.844570670646227e-05,
+      "loss": 0.7717,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5814380792944347,
+      "grad_norm": 0.2764573097229004,
+      "learning_rate": 3.8394956440625605e-05,
+      "loss": 0.8305,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5817647299007799,
+      "grad_norm": 0.304564505815506,
+      "learning_rate": 3.834421880832632e-05,
+      "loss": 0.9052,
+      "step": 1781
+    },
+    {
+      "epoch": 0.5820913805071251,
+      "grad_norm": 0.31863516569137573,
+      "learning_rate": 3.8293493864798646e-05,
+      "loss": 0.8262,
+      "step": 1782
+    },
+    {
+      "epoch": 0.5824180311134702,
+      "grad_norm": 0.32435324788093567,
+      "learning_rate": 3.824278166526305e-05,
+      "loss": 0.8383,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5827446817198154,
+      "grad_norm": 0.32743144035339355,
+      "learning_rate": 3.8192082264926035e-05,
+      "loss": 0.8747,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5830713323261606,
+      "grad_norm": 0.3483351469039917,
+      "learning_rate": 3.8141395718980275e-05,
+      "loss": 0.8814,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5833979829325058,
+      "grad_norm": 0.37021541595458984,
+      "learning_rate": 3.809072208260437e-05,
+      "loss": 0.9519,
+      "step": 1786
+    },
+    {
+      "epoch": 0.583724633538851,
+      "grad_norm": 0.37525907158851624,
+      "learning_rate": 3.804006141096288e-05,
+      "loss": 0.8737,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5840512841451962,
+      "grad_norm": 0.43124061822891235,
+      "learning_rate": 3.79894137592063e-05,
+      "loss": 1.0225,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5843779347515414,
+      "grad_norm": 0.42931702733039856,
+      "learning_rate": 3.793877918247087e-05,
+      "loss": 0.9199,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5847045853578866,
+      "grad_norm": 0.4696066081523895,
+      "learning_rate": 3.78881577358787e-05,
+      "loss": 1.0533,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5850312359642318,
+      "grad_norm": 0.5039542317390442,
+      "learning_rate": 3.783754947453751e-05,
+      "loss": 1.0553,
+      "step": 1791
+    },
+    {
+      "epoch": 0.585357886570577,
+      "grad_norm": 0.5651768445968628,
+      "learning_rate": 3.77869544535407e-05,
+      "loss": 1.1349,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5856845371769221,
+      "grad_norm": 0.6366235017776489,
+      "learning_rate": 3.7736372727967284e-05,
+      "loss": 1.1709,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5860111877832673,
+      "grad_norm": 0.623162567615509,
+      "learning_rate": 3.768580435288177e-05,
+      "loss": 1.241,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5863378383896125,
+      "grad_norm": 0.7199271321296692,
+      "learning_rate": 3.7635249383334114e-05,
+      "loss": 1.1771,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5866644889959577,
+      "grad_norm": 0.8708171248435974,
+      "learning_rate": 3.758470787435977e-05,
+      "loss": 1.3232,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5869911396023029,
+      "grad_norm": 1.146207332611084,
+      "learning_rate": 3.7534179880979405e-05,
+      "loss": 1.5614,
+      "step": 1797
+    },
+    {
+      "epoch": 0.587317790208648,
+      "grad_norm": 1.1645411252975464,
+      "learning_rate": 3.748366545819909e-05,
+      "loss": 1.3229,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5876444408149932,
+      "grad_norm": 1.5172971487045288,
+      "learning_rate": 3.7433164661010066e-05,
+      "loss": 1.465,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5879710914213384,
+      "grad_norm": 1.9966884851455688,
+      "learning_rate": 3.738267754438872e-05,
+      "loss": 2.1441,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5882977420276836,
+      "grad_norm": 0.22755567729473114,
+      "learning_rate": 3.733220416329662e-05,
+      "loss": 0.8176,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5886243926340289,
+      "grad_norm": 0.22714000940322876,
+      "learning_rate": 3.7281744572680297e-05,
+      "loss": 0.7798,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5889510432403741,
+      "grad_norm": 0.2419450730085373,
+      "learning_rate": 3.723129882747135e-05,
+      "loss": 0.7938,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5892776938467192,
+      "grad_norm": 0.2522052228450775,
+      "learning_rate": 3.7180866982586234e-05,
+      "loss": 0.7725,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5896043444530644,
+      "grad_norm": 0.27660250663757324,
+      "learning_rate": 3.713044909292629e-05,
+      "loss": 0.8315,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5899309950594096,
+      "grad_norm": 0.283093124628067,
+      "learning_rate": 3.708004521337772e-05,
+      "loss": 0.8763,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5902576456657548,
+      "grad_norm": 0.288693368434906,
+      "learning_rate": 3.702965539881141e-05,
+      "loss": 0.9238,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5905842962721,
+      "grad_norm": 0.3175692558288574,
+      "learning_rate": 3.697927970408294e-05,
+      "loss": 0.8364,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5909109468784451,
+      "grad_norm": 0.31832441687583923,
+      "learning_rate": 3.6928918184032566e-05,
+      "loss": 0.8666,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5912375974847903,
+      "grad_norm": 0.32215848565101624,
+      "learning_rate": 3.687857089348509e-05,
+      "loss": 0.8798,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5915642480911355,
+      "grad_norm": 0.3508221209049225,
+      "learning_rate": 3.682823788724979e-05,
+      "loss": 0.9086,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5918908986974807,
+      "grad_norm": 0.3799271285533905,
+      "learning_rate": 3.677791922012045e-05,
+      "loss": 0.9276,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5922175493038259,
+      "grad_norm": 0.400530606508255,
+      "learning_rate": 3.6727614946875176e-05,
+      "loss": 0.9829,
+      "step": 1813
+    },
+    {
+      "epoch": 0.592544199910171,
+      "grad_norm": 0.4067695438861847,
+      "learning_rate": 3.667732512227649e-05,
+      "loss": 0.8977,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5928708505165162,
+      "grad_norm": 0.46635621786117554,
+      "learning_rate": 3.662704980107109e-05,
+      "loss": 1.0206,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5931975011228615,
+      "grad_norm": 0.47291892766952515,
+      "learning_rate": 3.657678903798999e-05,
+      "loss": 0.9792,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5935241517292067,
+      "grad_norm": 0.5139042735099792,
+      "learning_rate": 3.652654288774827e-05,
+      "loss": 0.9158,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5938508023355519,
+      "grad_norm": 0.5157402157783508,
+      "learning_rate": 3.64763114050451e-05,
+      "loss": 1.0019,
+      "step": 1818
+    },
+    {
+      "epoch": 0.594177452941897,
+      "grad_norm": 0.5718170404434204,
+      "learning_rate": 3.642609464456377e-05,
+      "loss": 0.9613,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5945041035482422,
+      "grad_norm": 0.6026821732521057,
+      "learning_rate": 3.6375892660971464e-05,
+      "loss": 1.1645,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5948307541545874,
+      "grad_norm": 0.7391718626022339,
+      "learning_rate": 3.6325705508919286e-05,
+      "loss": 1.264,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5951574047609326,
+      "grad_norm": 0.936079740524292,
+      "learning_rate": 3.6275533243042246e-05,
+      "loss": 1.4989,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5954840553672778,
+      "grad_norm": 1.090334177017212,
+      "learning_rate": 3.622537591795907e-05,
+      "loss": 1.3695,
+      "step": 1823
+    },
+    {
+      "epoch": 0.595810705973623,
+      "grad_norm": 1.3231812715530396,
+      "learning_rate": 3.61752335882723e-05,
+      "loss": 1.6894,
+      "step": 1824
+    },
+    {
+      "epoch": 0.5961373565799681,
+      "grad_norm": 1.463555097579956,
+      "learning_rate": 3.61251063085681e-05,
+      "loss": 1.6237,
+      "step": 1825
+    },
+    {
+      "epoch": 0.5964640071863133,
+      "grad_norm": 0.21345098316669464,
+      "learning_rate": 3.607499413341626e-05,
+      "loss": 0.7963,
+      "step": 1826
+    },
+    {
+      "epoch": 0.5967906577926585,
+      "grad_norm": 0.2452787309885025,
+      "learning_rate": 3.602489711737015e-05,
+      "loss": 0.8304,
+      "step": 1827
+    },
+    {
+      "epoch": 0.5971173083990037,
+      "grad_norm": 0.2656680941581726,
+      "learning_rate": 3.59748153149666e-05,
+      "loss": 0.8257,
+      "step": 1828
+    },
+    {
+      "epoch": 0.5974439590053489,
+      "grad_norm": 0.25352826714515686,
+      "learning_rate": 3.592474878072592e-05,
+      "loss": 0.7629,
+      "step": 1829
+    },
+    {
+      "epoch": 0.597770609611694,
+      "grad_norm": 0.259874165058136,
+      "learning_rate": 3.587469756915177e-05,
+      "loss": 0.7529,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5980972602180393,
+      "grad_norm": 0.3154599666595459,
+      "learning_rate": 3.5824661734731126e-05,
+      "loss": 0.8974,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5984239108243845,
+      "grad_norm": 0.3104557693004608,
+      "learning_rate": 3.577464133193425e-05,
+      "loss": 0.9489,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5987505614307297,
+      "grad_norm": 0.29206931591033936,
+      "learning_rate": 3.572463641521456e-05,
+      "loss": 0.8588,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5990772120370749,
+      "grad_norm": 0.3030562996864319,
+      "learning_rate": 3.5674647039008694e-05,
+      "loss": 0.8096,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59940386264342,
+      "grad_norm": 0.353023499250412,
+      "learning_rate": 3.5624673257736294e-05,
+      "loss": 0.8584,
+      "step": 1835
+    },
+    {
+      "epoch": 0.5997305132497652,
+      "grad_norm": 0.35337188839912415,
+      "learning_rate": 3.557471512580004e-05,
+      "loss": 0.8645,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6000571638561104,
+      "grad_norm": 0.3872027099132538,
+      "learning_rate": 3.552477269758562e-05,
+      "loss": 0.849,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6003838144624556,
+      "grad_norm": 0.36967232823371887,
+      "learning_rate": 3.547484602746158e-05,
+      "loss": 0.8275,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6007104650688008,
+      "grad_norm": 0.3914700746536255,
+      "learning_rate": 3.5424935169779314e-05,
+      "loss": 0.9307,
+      "step": 1839
+    },
+    {
+      "epoch": 0.601037115675146,
+      "grad_norm": 0.4842604994773865,
+      "learning_rate": 3.537504017887305e-05,
+      "loss": 0.9365,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6013637662814911,
+      "grad_norm": 0.48530450463294983,
+      "learning_rate": 3.5325161109059665e-05,
+      "loss": 1.0595,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6016904168878363,
+      "grad_norm": 0.493562787771225,
+      "learning_rate": 3.52752980146388e-05,
+      "loss": 0.9919,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6020170674941815,
+      "grad_norm": 0.5630847811698914,
+      "learning_rate": 3.522545094989262e-05,
+      "loss": 1.1229,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6023437181005267,
+      "grad_norm": 0.6644653677940369,
+      "learning_rate": 3.517561996908587e-05,
+      "loss": 1.1268,
+      "step": 1844
+    },
+    {
+      "epoch": 0.602670368706872,
+      "grad_norm": 0.7322569489479065,
+      "learning_rate": 3.5125805126465795e-05,
+      "loss": 1.3862,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6029970193132171,
+      "grad_norm": 0.8114511966705322,
+      "learning_rate": 3.5076006476262066e-05,
+      "loss": 1.0942,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6033236699195623,
+      "grad_norm": 1.0138306617736816,
+      "learning_rate": 3.502622407268673e-05,
+      "loss": 1.1355,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6036503205259075,
+      "grad_norm": 1.3908495903015137,
+      "learning_rate": 3.497645796993414e-05,
+      "loss": 1.4335,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6039769711322527,
+      "grad_norm": 1.703803539276123,
+      "learning_rate": 3.492670822218089e-05,
+      "loss": 1.9552,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6043036217385979,
+      "grad_norm": 1.7327630519866943,
+      "learning_rate": 3.4876974883585814e-05,
+      "loss": 1.5885,
+      "step": 1850
+    },
+    {
+      "epoch": 0.604630272344943,
+      "grad_norm": 0.20497965812683105,
+      "learning_rate": 3.482725800828984e-05,
+      "loss": 0.6881,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6049569229512882,
+      "grad_norm": 0.24482519924640656,
+      "learning_rate": 3.477755765041596e-05,
+      "loss": 0.7687,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6052835735576334,
+      "grad_norm": 0.26200076937675476,
+      "learning_rate": 3.472787386406926e-05,
+      "loss": 0.8595,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6056102241639786,
+      "grad_norm": 0.2740213871002197,
+      "learning_rate": 3.467820670333671e-05,
+      "loss": 0.8254,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6059368747703238,
+      "grad_norm": 0.27905282378196716,
+      "learning_rate": 3.462855622228723e-05,
+      "loss": 0.8618,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6062635253766689,
+      "grad_norm": 0.2976115345954895,
+      "learning_rate": 3.4578922474971545e-05,
+      "loss": 0.8119,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6065901759830141,
+      "grad_norm": 0.3185284435749054,
+      "learning_rate": 3.452930551542216e-05,
+      "loss": 0.8803,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6069168265893593,
+      "grad_norm": 0.29103270173072815,
+      "learning_rate": 3.447970539765338e-05,
+      "loss": 0.7803,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6072434771957046,
+      "grad_norm": 0.34088513255119324,
+      "learning_rate": 3.4430122175661055e-05,
+      "loss": 0.8361,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6075701278020498,
+      "grad_norm": 0.33174625039100647,
+      "learning_rate": 3.438055590342276e-05,
+      "loss": 0.8778,
+      "step": 1860
+    },
+    {
+      "epoch": 0.607896778408395,
+      "grad_norm": 0.3289026916027069,
+      "learning_rate": 3.4331006634897534e-05,
+      "loss": 0.8508,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6082234290147401,
+      "grad_norm": 0.3580840229988098,
+      "learning_rate": 3.428147442402595e-05,
+      "loss": 0.9233,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6085500796210853,
+      "grad_norm": 0.38521304726600647,
+      "learning_rate": 3.423195932472999e-05,
+      "loss": 0.8952,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6088767302274305,
+      "grad_norm": 0.40166860818862915,
+      "learning_rate": 3.4182461390913024e-05,
+      "loss": 0.9741,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6092033808337757,
+      "grad_norm": 0.4904230535030365,
+      "learning_rate": 3.41329806764597e-05,
+      "loss": 1.0898,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6095300314401209,
+      "grad_norm": 0.47519782185554504,
+      "learning_rate": 3.408351723523599e-05,
+      "loss": 1.0139,
+      "step": 1866
+    },
+    {
+      "epoch": 0.609856682046466,
+      "grad_norm": 0.4924708902835846,
+      "learning_rate": 3.403407112108898e-05,
+      "loss": 0.9999,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6101833326528112,
+      "grad_norm": 0.5966106057167053,
+      "learning_rate": 3.398464238784696e-05,
+      "loss": 1.1533,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6105099832591564,
+      "grad_norm": 0.6565825939178467,
+      "learning_rate": 3.393523108931928e-05,
+      "loss": 1.0461,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6108366338655016,
+      "grad_norm": 0.7097316980361938,
+      "learning_rate": 3.388583727929625e-05,
+      "loss": 1.0233,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6111632844718468,
+      "grad_norm": 0.814643144607544,
+      "learning_rate": 3.383646101154924e-05,
+      "loss": 1.226,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6114899350781919,
+      "grad_norm": 1.0871217250823975,
+      "learning_rate": 3.3787102339830444e-05,
+      "loss": 1.2491,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6118165856845372,
+      "grad_norm": 1.5111687183380127,
+      "learning_rate": 3.373776131787298e-05,
+      "loss": 1.7033,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6121432362908824,
+      "grad_norm": 1.5856449604034424,
+      "learning_rate": 3.368843799939066e-05,
+      "loss": 1.7526,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6124698868972276,
+      "grad_norm": 1.6733304262161255,
+      "learning_rate": 3.363913243807808e-05,
+      "loss": 1.9496,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6127965375035728,
+      "grad_norm": 0.21695645153522491,
+      "learning_rate": 3.3589844687610506e-05,
+      "loss": 0.764,
+      "step": 1876
+    },
+    {
+      "epoch": 0.613123188109918,
+      "grad_norm": 0.267497181892395,
+      "learning_rate": 3.3540574801643776e-05,
+      "loss": 0.9128,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6134498387162631,
+      "grad_norm": 0.26708516478538513,
+      "learning_rate": 3.349132283381433e-05,
+      "loss": 0.9034,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6137764893226083,
+      "grad_norm": 0.27384889125823975,
+      "learning_rate": 3.3442088837739075e-05,
+      "loss": 0.8362,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6141031399289535,
+      "grad_norm": 0.29018422961235046,
+      "learning_rate": 3.339287286701534e-05,
+      "loss": 0.8634,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6144297905352987,
+      "grad_norm": 0.29773256182670593,
+      "learning_rate": 3.3343674975220867e-05,
+      "loss": 0.8958,
+      "step": 1881
+    },
+    {
+      "epoch": 0.6147564411416439,
+      "grad_norm": 0.30387818813323975,
+      "learning_rate": 3.32944952159137e-05,
+      "loss": 0.8797,
+      "step": 1882
+    },
+    {
+      "epoch": 0.615083091747989,
+      "grad_norm": 0.3108673691749573,
+      "learning_rate": 3.324533364263211e-05,
+      "loss": 0.7999,
+      "step": 1883
+    },
+    {
+      "epoch": 0.6154097423543342,
+      "grad_norm": 0.32888373732566833,
+      "learning_rate": 3.319619030889464e-05,
+      "loss": 0.8075,
+      "step": 1884
+    },
+    {
+      "epoch": 0.6157363929606794,
+      "grad_norm": 0.3333348333835602,
+      "learning_rate": 3.314706526819991e-05,
+      "loss": 0.8369,
+      "step": 1885
+    },
+    {
+      "epoch": 0.6160630435670246,
+      "grad_norm": 0.35576188564300537,
+      "learning_rate": 3.309795857402667e-05,
+      "loss": 0.9797,
+      "step": 1886
+    },
+    {
+      "epoch": 0.6163896941733699,
+      "grad_norm": 0.3945654034614563,
+      "learning_rate": 3.304887027983369e-05,
+      "loss": 1.0048,
+      "step": 1887
+    },
+    {
+      "epoch": 0.616716344779715,
+      "grad_norm": 0.3923867344856262,
+      "learning_rate": 3.299980043905966e-05,
+      "loss": 0.9386,
+      "step": 1888
+    },
+    {
+      "epoch": 0.6170429953860602,
+      "grad_norm": 0.3965389132499695,
+      "learning_rate": 3.295074910512327e-05,
+      "loss": 0.9106,
+      "step": 1889
+    },
+    {
+      "epoch": 0.6173696459924054,
+      "grad_norm": 0.45480525493621826,
+      "learning_rate": 3.2901716331422994e-05,
+      "loss": 1.0531,
+      "step": 1890
+    },
+    {
+      "epoch": 0.6176962965987506,
+      "grad_norm": 0.4836742579936981,
+      "learning_rate": 3.28527021713371e-05,
+      "loss": 1.1177,
+      "step": 1891
+    },
+    {
+      "epoch": 0.6180229472050958,
+      "grad_norm": 0.5407842397689819,
+      "learning_rate": 3.280370667822367e-05,
+      "loss": 1.1417,
+      "step": 1892
+    },
+    {
+      "epoch": 0.6183495978114409,
+      "grad_norm": 0.6356526017189026,
+      "learning_rate": 3.275472990542034e-05,
+      "loss": 1.0241,
+      "step": 1893
+    },
+    {
+      "epoch": 0.6186762484177861,
+      "grad_norm": 0.7225267291069031,
+      "learning_rate": 3.2705771906244496e-05,
+      "loss": 1.1239,
+      "step": 1894
+    },
+    {
+      "epoch": 0.6190028990241313,
+      "grad_norm": 0.7228987216949463,
+      "learning_rate": 3.2656832733993e-05,
+      "loss": 1.3345,
+      "step": 1895
+    },
+    {
+      "epoch": 0.6193295496304765,
+      "grad_norm": 0.7399430871009827,
+      "learning_rate": 3.260791244194225e-05,
+      "loss": 1.0652,
+      "step": 1896
+    },
+    {
+      "epoch": 0.6196562002368217,
+      "grad_norm": 1.0633045434951782,
+      "learning_rate": 3.255901108334807e-05,
+      "loss": 1.2763,
+      "step": 1897
+    },
+    {
+      "epoch": 0.6199828508431668,
+      "grad_norm": 1.297610878944397,
+      "learning_rate": 3.251012871144571e-05,
+      "loss": 1.2444,
+      "step": 1898
+    },
+    {
+      "epoch": 0.620309501449512,
+      "grad_norm": 1.577770471572876,
+      "learning_rate": 3.2461265379449725e-05,
+      "loss": 1.6499,
+      "step": 1899
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "grad_norm": 1.752236247062683,
+      "learning_rate": 3.241242114055394e-05,
+      "loss": 1.7408,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6209628026622025,
+      "grad_norm": 0.22202466428279877,
+      "learning_rate": 3.236359604793139e-05,
+      "loss": 0.7555,
+      "step": 1901
+    },
+    {
+      "epoch": 0.6212894532685477,
+      "grad_norm": 0.24626395106315613,
+      "learning_rate": 3.231479015473432e-05,
+      "loss": 0.817,
+      "step": 1902
+    },
+    {
+      "epoch": 0.6216161038748929,
+      "grad_norm": 0.2575581967830658,
+      "learning_rate": 3.226600351409399e-05,
+      "loss": 0.856,
+      "step": 1903
+    },
+    {
+      "epoch": 0.621942754481238,
+      "grad_norm": 0.29142552614212036,
+      "learning_rate": 3.221723617912075e-05,
+      "loss": 0.8771,
+      "step": 1904
+    },
+    {
+      "epoch": 0.6222694050875832,
+      "grad_norm": 0.27642571926116943,
+      "learning_rate": 3.2168488202903947e-05,
+      "loss": 0.8312,
+      "step": 1905
+    },
+    {
+      "epoch": 0.6225960556939284,
+      "grad_norm": 0.3057843744754791,
+      "learning_rate": 3.211975963851179e-05,
+      "loss": 0.9538,
+      "step": 1906
+    },
+    {
+      "epoch": 0.6229227063002736,
+      "grad_norm": 0.3055606186389923,
+      "learning_rate": 3.207105053899147e-05,
+      "loss": 0.8128,
+      "step": 1907
+    },
+    {
+      "epoch": 0.6232493569066188,
+      "grad_norm": 0.3037729561328888,
+      "learning_rate": 3.202236095736886e-05,
+      "loss": 0.8161,
+      "step": 1908
+    },
+    {
+      "epoch": 0.6235760075129639,
+      "grad_norm": 0.32354530692100525,
+      "learning_rate": 3.197369094664865e-05,
+      "loss": 0.8633,
+      "step": 1909
+    },
+    {
+      "epoch": 0.6239026581193091,
+      "grad_norm": 0.34619325399398804,
+      "learning_rate": 3.192504055981423e-05,
+      "loss": 0.8472,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6242293087256543,
+      "grad_norm": 0.3544459939002991,
+      "learning_rate": 3.187640984982761e-05,
+      "loss": 0.8922,
+      "step": 1911
+    },
+    {
+      "epoch": 0.6245559593319995,
+      "grad_norm": 0.36598286032676697,
+      "learning_rate": 3.1827798869629396e-05,
+      "loss": 0.8475,
+      "step": 1912
+    },
+    {
+      "epoch": 0.6248826099383447,
+      "grad_norm": 0.4017038643360138,
+      "learning_rate": 3.177920767213868e-05,
+      "loss": 0.8745,
+      "step": 1913
+    },
+    {
+      "epoch": 0.6252092605446898,
+      "grad_norm": 0.4290722608566284,
+      "learning_rate": 3.1730636310253047e-05,
+      "loss": 0.975,
+      "step": 1914
+    },
+    {
+      "epoch": 0.6255359111510351,
+      "grad_norm": 0.44876834750175476,
+      "learning_rate": 3.1682084836848497e-05,
+      "loss": 0.9702,
+      "step": 1915
+    },
+    {
+      "epoch": 0.6258625617573803,
+      "grad_norm": 0.48596665263175964,
+      "learning_rate": 3.1633553304779354e-05,
+      "loss": 1.0497,
+      "step": 1916
+    },
+    {
+      "epoch": 0.6261892123637255,
+      "grad_norm": 0.49547937512397766,
+      "learning_rate": 3.158504176687824e-05,
+      "loss": 1.0011,
+      "step": 1917
+    },
+    {
+      "epoch": 0.6265158629700707,
+      "grad_norm": 0.5421122908592224,
+      "learning_rate": 3.153655027595603e-05,
+      "loss": 0.9569,
+      "step": 1918
+    },
+    {
+      "epoch": 0.6268425135764158,
+      "grad_norm": 0.6027096509933472,
+      "learning_rate": 3.1488078884801734e-05,
+      "loss": 1.2515,
+      "step": 1919
+    },
+    {
+      "epoch": 0.627169164182761,
+      "grad_norm": 0.7056169509887695,
+      "learning_rate": 3.143962764618254e-05,
+      "loss": 1.1874,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6274958147891062,
+      "grad_norm": 0.791888415813446,
+      "learning_rate": 3.1391196612843646e-05,
+      "loss": 1.2552,
+      "step": 1921
+    },
+    {
+      "epoch": 0.6278224653954514,
+      "grad_norm": 0.998956024646759,
+      "learning_rate": 3.134278583750826e-05,
+      "loss": 1.3186,
+      "step": 1922
+    },
+    {
+      "epoch": 0.6281491160017966,
+      "grad_norm": 1.0044286251068115,
+      "learning_rate": 3.129439537287758e-05,
+      "loss": 1.3368,
+      "step": 1923
+    },
+    {
+      "epoch": 0.6284757666081418,
+      "grad_norm": 1.431918740272522,
+      "learning_rate": 3.124602527163062e-05,
+      "loss": 1.6185,
+      "step": 1924
+    },
+    {
+      "epoch": 0.6288024172144869,
+      "grad_norm": 1.8370330333709717,
+      "learning_rate": 3.119767558642432e-05,
+      "loss": 1.7886,
+      "step": 1925
+    },
+    {
+      "epoch": 0.6291290678208321,
+      "grad_norm": 0.2085886001586914,
+      "learning_rate": 3.1149346369893314e-05,
+      "loss": 0.6986,
+      "step": 1926
+    },
+    {
+      "epoch": 0.6294557184271773,
+      "grad_norm": 0.23808997869491577,
+      "learning_rate": 3.1101037674649966e-05,
+      "loss": 0.8573,
+      "step": 1927
+    },
+    {
+      "epoch": 0.6297823690335225,
+      "grad_norm": 0.24077343940734863,
+      "learning_rate": 3.1052749553284356e-05,
+      "loss": 0.7755,
+      "step": 1928
+    },
+    {
+      "epoch": 0.6301090196398677,
+      "grad_norm": 0.26138561964035034,
+      "learning_rate": 3.100448205836408e-05,
+      "loss": 0.8459,
+      "step": 1929
+    },
+    {
+      "epoch": 0.6304356702462129,
+      "grad_norm": 0.25806763768196106,
+      "learning_rate": 3.095623524243437e-05,
+      "loss": 0.8026,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6307623208525581,
+      "grad_norm": 0.2821289598941803,
+      "learning_rate": 3.0908009158017893e-05,
+      "loss": 0.8433,
+      "step": 1931
+    },
+    {
+      "epoch": 0.6310889714589033,
+      "grad_norm": 0.2813338339328766,
+      "learning_rate": 3.0859803857614734e-05,
+      "loss": 0.8486,
+      "step": 1932
+    },
+    {
+      "epoch": 0.6314156220652485,
+      "grad_norm": 0.29995447397232056,
+      "learning_rate": 3.08116193937024e-05,
+      "loss": 0.8243,
+      "step": 1933
+    },
+    {
+      "epoch": 0.6317422726715937,
+      "grad_norm": 0.31056755781173706,
+      "learning_rate": 3.076345581873568e-05,
+      "loss": 0.8101,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6320689232779388,
+      "grad_norm": 0.31584444642066956,
+      "learning_rate": 3.071531318514661e-05,
+      "loss": 0.8359,
+      "step": 1935
+    },
+    {
+      "epoch": 0.632395573884284,
+      "grad_norm": 0.33051276206970215,
+      "learning_rate": 3.066719154534451e-05,
+      "loss": 0.8212,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6327222244906292,
+      "grad_norm": 0.3518679141998291,
+      "learning_rate": 3.061909095171572e-05,
+      "loss": 1.0019,
+      "step": 1937
+    },
+    {
+      "epoch": 0.6330488750969744,
+      "grad_norm": 0.3903535008430481,
+      "learning_rate": 3.0571011456623786e-05,
+      "loss": 0.9155,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6333755257033196,
+      "grad_norm": 0.4148496091365814,
+      "learning_rate": 3.0522953112409224e-05,
+      "loss": 1.0204,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6337021763096647,
+      "grad_norm": 0.4300936460494995,
+      "learning_rate": 3.0474915971389506e-05,
+      "loss": 0.9795,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6340288269160099,
+      "grad_norm": 0.5103878974914551,
+      "learning_rate": 3.042690008585909e-05,
+      "loss": 1.0279,
+      "step": 1941
+    },
+    {
+      "epoch": 0.6343554775223551,
+      "grad_norm": 0.46250119805336,
+      "learning_rate": 3.037890550808923e-05,
+      "loss": 0.982,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6346821281287003,
+      "grad_norm": 0.5167394876480103,
+      "learning_rate": 3.033093229032803e-05,
+      "loss": 0.9676,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6350087787350456,
+      "grad_norm": 0.5678005218505859,
+      "learning_rate": 3.028298048480031e-05,
+      "loss": 0.9864,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6353354293413908,
+      "grad_norm": 0.6823439002037048,
+      "learning_rate": 3.0235050143707577e-05,
+      "loss": 1.2096,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6356620799477359,
+      "grad_norm": 0.8533816933631897,
+      "learning_rate": 3.018714131922802e-05,
+      "loss": 1.2273,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6359887305540811,
+      "grad_norm": 0.8832264542579651,
+      "learning_rate": 3.0139254063516343e-05,
+      "loss": 1.2583,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6363153811604263,
+      "grad_norm": 1.1121357679367065,
+      "learning_rate": 3.0091388428703793e-05,
+      "loss": 1.4092,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6366420317667715,
+      "grad_norm": 1.4677468538284302,
+      "learning_rate": 3.0043544466898115e-05,
+      "loss": 1.9588,
+      "step": 1949
+    },
+    {
+      "epoch": 0.6369686823731167,
+      "grad_norm": 1.5762861967086792,
+      "learning_rate": 2.9995722230183387e-05,
+      "loss": 1.1872,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6372953329794618,
+      "grad_norm": 0.217019721865654,
+      "learning_rate": 2.9947921770620126e-05,
+      "loss": 0.7654,
+      "step": 1951
+    },
+    {
+      "epoch": 0.637621983585807,
+      "grad_norm": 0.23723232746124268,
+      "learning_rate": 2.9900143140245067e-05,
+      "loss": 0.8003,
+      "step": 1952
+    },
+    {
+      "epoch": 0.6379486341921522,
+      "grad_norm": 0.26588693261146545,
+      "learning_rate": 2.9852386391071192e-05,
+      "loss": 0.856,
+      "step": 1953
+    },
+    {
+      "epoch": 0.6382752847984974,
+      "grad_norm": 0.25790339708328247,
+      "learning_rate": 2.9804651575087716e-05,
+      "loss": 0.8491,
+      "step": 1954
+    },
+    {
+      "epoch": 0.6386019354048426,
+      "grad_norm": 0.2517349421977997,
+      "learning_rate": 2.97569387442599e-05,
+      "loss": 0.8128,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6389285860111877,
+      "grad_norm": 0.27819013595581055,
+      "learning_rate": 2.9709247950529157e-05,
+      "loss": 0.8499,
+      "step": 1956
+    },
+    {
+      "epoch": 0.6392552366175329,
+      "grad_norm": 0.30752265453338623,
+      "learning_rate": 2.9661579245812844e-05,
+      "loss": 0.8082,
+      "step": 1957
+    },
+    {
+      "epoch": 0.6395818872238782,
+      "grad_norm": 0.31690144538879395,
+      "learning_rate": 2.961393268200426e-05,
+      "loss": 0.8634,
+      "step": 1958
+    },
+    {
+      "epoch": 0.6399085378302234,
+      "grad_norm": 0.33005377650260925,
+      "learning_rate": 2.956630831097268e-05,
+      "loss": 0.9546,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6402351884365686,
+      "grad_norm": 0.3165203630924225,
+      "learning_rate": 2.9518706184563162e-05,
+      "loss": 0.8628,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6405618390429137,
+      "grad_norm": 0.3594956398010254,
+      "learning_rate": 2.9471126354596517e-05,
+      "loss": 0.8658,
+      "step": 1961
+    },
+    {
+      "epoch": 0.6408884896492589,
+      "grad_norm": 0.3517896831035614,
+      "learning_rate": 2.9423568872869383e-05,
+      "loss": 0.9052,
+      "step": 1962
+    },
+    {
+      "epoch": 0.6412151402556041,
+      "grad_norm": 0.42132368683815,
+      "learning_rate": 2.9376033791153962e-05,
+      "loss": 0.8795,
+      "step": 1963
+    },
+    {
+      "epoch": 0.6415417908619493,
+      "grad_norm": 0.41029414534568787,
+      "learning_rate": 2.9328521161198152e-05,
+      "loss": 0.967,
+      "step": 1964
+    },
+    {
+      "epoch": 0.6418684414682945,
+      "grad_norm": 0.444322794675827,
+      "learning_rate": 2.9281031034725377e-05,
+      "loss": 0.9924,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6421950920746397,
+      "grad_norm": 0.5007185339927673,
+      "learning_rate": 2.9233563463434534e-05,
+      "loss": 1.0496,
+      "step": 1966
+    },
+    {
+      "epoch": 0.6425217426809848,
+      "grad_norm": 0.5201416015625,
+      "learning_rate": 2.9186118499000026e-05,
+      "loss": 0.9913,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64284839328733,
+      "grad_norm": 0.5279054641723633,
+      "learning_rate": 2.913869619307159e-05,
+      "loss": 0.9855,
+      "step": 1968
+    },
+    {
+      "epoch": 0.6431750438936752,
+      "grad_norm": 0.603148341178894,
+      "learning_rate": 2.9091296597274342e-05,
+      "loss": 1.0872,
+      "step": 1969
+    },
+    {
+      "epoch": 0.6435016945000204,
+      "grad_norm": 0.7442902326583862,
+      "learning_rate": 2.904391976320864e-05,
+      "loss": 1.2015,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6438283451063656,
+      "grad_norm": 0.9018768668174744,
+      "learning_rate": 2.8996565742450078e-05,
+      "loss": 1.2661,
+      "step": 1971
+    },
+    {
+      "epoch": 0.6441549957127108,
+      "grad_norm": 1.0673716068267822,
+      "learning_rate": 2.8949234586549418e-05,
+      "loss": 1.5745,
+      "step": 1972
+    },
+    {
+      "epoch": 0.644481646319056,
+      "grad_norm": 1.511343002319336,
+      "learning_rate": 2.890192634703255e-05,
+      "loss": 1.7574,
+      "step": 1973
+    },
+    {
+      "epoch": 0.6448082969254012,
+      "grad_norm": 1.3675079345703125,
+      "learning_rate": 2.8854641075400324e-05,
+      "loss": 1.4665,
+      "step": 1974
+    },
+    {
+      "epoch": 0.6451349475317464,
+      "grad_norm": 1.6643435955047607,
+      "learning_rate": 2.8807378823128744e-05,
+      "loss": 1.9878,
+      "step": 1975
+    },
+    {
+      "epoch": 0.6454615981380916,
+      "grad_norm": 0.23247745633125305,
+      "learning_rate": 2.8760139641668582e-05,
+      "loss": 0.7989,
+      "step": 1976
+    },
+    {
+      "epoch": 0.6457882487444367,
+      "grad_norm": 0.2454388439655304,
+      "learning_rate": 2.8712923582445662e-05,
+      "loss": 0.7655,
+      "step": 1977
+    },
+    {
+      "epoch": 0.6461148993507819,
+      "grad_norm": 0.257198691368103,
+      "learning_rate": 2.866573069686049e-05,
+      "loss": 0.8634,
+      "step": 1978
+    },
+    {
+      "epoch": 0.6464415499571271,
+      "grad_norm": 0.2427079677581787,
+      "learning_rate": 2.8618561036288437e-05,
+      "loss": 0.7737,
+      "step": 1979
+    },
+    {
+      "epoch": 0.6467682005634723,
+      "grad_norm": 0.2693164348602295,
+      "learning_rate": 2.8571414652079553e-05,
+      "loss": 0.8145,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6470948511698175,
+      "grad_norm": 0.2872451841831207,
+      "learning_rate": 2.852429159555856e-05,
+      "loss": 0.8418,
+      "step": 1981
+    },
+    {
+      "epoch": 0.6474215017761626,
+      "grad_norm": 0.2881939709186554,
+      "learning_rate": 2.8477191918024782e-05,
+      "loss": 0.9102,
+      "step": 1982
+    },
+    {
+      "epoch": 0.6477481523825078,
+      "grad_norm": 0.2885768711566925,
+      "learning_rate": 2.8430115670752118e-05,
+      "loss": 0.8406,
+      "step": 1983
+    },
+    {
+      "epoch": 0.648074802988853,
+      "grad_norm": 0.3151211738586426,
+      "learning_rate": 2.8383062904988865e-05,
+      "loss": 0.8316,
+      "step": 1984
+    },
+    {
+      "epoch": 0.6484014535951982,
+      "grad_norm": 0.3474157452583313,
+      "learning_rate": 2.8336033671957916e-05,
+      "loss": 0.8728,
+      "step": 1985
+    },
+    {
+      "epoch": 0.6487281042015435,
+      "grad_norm": 0.3520103096961975,
+      "learning_rate": 2.828902802285639e-05,
+      "loss": 0.8453,
+      "step": 1986
+    },
+    {
+      "epoch": 0.6490547548078887,
+      "grad_norm": 0.38095179200172424,
+      "learning_rate": 2.824204600885583e-05,
+      "loss": 1.0082,
+      "step": 1987
+    },
+    {
+      "epoch": 0.6493814054142338,
+      "grad_norm": 0.4117640554904938,
+      "learning_rate": 2.8195087681102e-05,
+      "loss": 0.9478,
+      "step": 1988
+    },
+    {
+      "epoch": 0.649708056020579,
+      "grad_norm": 0.41134095191955566,
+      "learning_rate": 2.8148153090714912e-05,
+      "loss": 1.0398,
+      "step": 1989
+    },
+    {
+      "epoch": 0.6500347066269242,
+      "grad_norm": 0.47175323963165283,
+      "learning_rate": 2.810124228878872e-05,
+      "loss": 0.9965,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6503613572332694,
+      "grad_norm": 0.5010828971862793,
+      "learning_rate": 2.805435532639168e-05,
+      "loss": 0.9616,
+      "step": 1991
+    },
+    {
+      "epoch": 0.6506880078396146,
+      "grad_norm": 0.5081592202186584,
+      "learning_rate": 2.8007492254566097e-05,
+      "loss": 0.966,
+      "step": 1992
+    },
+    {
+      "epoch": 0.6510146584459597,
+      "grad_norm": 0.5605658292770386,
+      "learning_rate": 2.7960653124328275e-05,
+      "loss": 1.1327,
+      "step": 1993
+    },
+    {
+      "epoch": 0.6513413090523049,
+      "grad_norm": 0.5658648610115051,
+      "learning_rate": 2.791383798666844e-05,
+      "loss": 0.9966,
+      "step": 1994
+    },
+    {
+      "epoch": 0.6516679596586501,
+      "grad_norm": 0.6892904043197632,
+      "learning_rate": 2.7867046892550718e-05,
+      "loss": 1.2015,
+      "step": 1995
+    },
+    {
+      "epoch": 0.6519946102649953,
+      "grad_norm": 0.7385665774345398,
+      "learning_rate": 2.782027989291307e-05,
+      "loss": 1.3089,
+      "step": 1996
+    },
+    {
+      "epoch": 0.6523212608713405,
+      "grad_norm": 0.9993577003479004,
+      "learning_rate": 2.777353703866714e-05,
+      "loss": 1.3718,
+      "step": 1997
+    },
+    {
+      "epoch": 0.6526479114776856,
+      "grad_norm": 1.1310564279556274,
+      "learning_rate": 2.772681838069845e-05,
+      "loss": 1.5095,
+      "step": 1998
+    },
+    {
+      "epoch": 0.6529745620840308,
+      "grad_norm": 1.478954553604126,
+      "learning_rate": 2.7680123969865988e-05,
+      "loss": 1.7489,
+      "step": 1999
+    },
+    {
+      "epoch": 0.6533012126903761,
+      "grad_norm": 1.603333592414856,
+      "learning_rate": 2.7633453857002546e-05,
+      "loss": 2.1632,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6536278632967213,
+      "grad_norm": 0.206992968916893,
+      "learning_rate": 2.7586808092914273e-05,
+      "loss": 0.7184,
+      "step": 2001
+    },
+    {
+      "epoch": 0.6539545139030665,
+      "grad_norm": 0.24995890259742737,
+      "learning_rate": 2.7540186728380958e-05,
+      "loss": 0.8657,
+      "step": 2002
+    },
+    {
+      "epoch": 0.6542811645094117,
+      "grad_norm": 0.2497495859861374,
+      "learning_rate": 2.749358981415574e-05,
+      "loss": 0.7396,
+      "step": 2003
+    },
+    {
+      "epoch": 0.6546078151157568,
+      "grad_norm": 0.2611045837402344,
+      "learning_rate": 2.744701740096518e-05,
+      "loss": 0.861,
+      "step": 2004
+    },
+    {
+      "epoch": 0.654934465722102,
+      "grad_norm": 0.279967725276947,
+      "learning_rate": 2.7400469539509177e-05,
+      "loss": 0.856,
+      "step": 2005
+    },
+    {
+      "epoch": 0.6552611163284472,
+      "grad_norm": 0.29646557569503784,
+      "learning_rate": 2.7353946280460857e-05,
+      "loss": 0.7972,
+      "step": 2006
+    },
+    {
+      "epoch": 0.6555877669347924,
+      "grad_norm": 0.2923833131790161,
+      "learning_rate": 2.7307447674466607e-05,
+      "loss": 0.8352,
+      "step": 2007
+    },
+    {
+      "epoch": 0.6559144175411376,
+      "grad_norm": 0.3097902238368988,
+      "learning_rate": 2.726097377214596e-05,
+      "loss": 0.7813,
+      "step": 2008
+    },
+    {
+      "epoch": 0.6562410681474827,
+      "grad_norm": 0.33259230852127075,
+      "learning_rate": 2.7214524624091566e-05,
+      "loss": 0.8932,
+      "step": 2009
+    },
+    {
+      "epoch": 0.6565677187538279,
+      "grad_norm": 0.3300251066684723,
+      "learning_rate": 2.716810028086906e-05,
+      "loss": 0.8678,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6568943693601731,
+      "grad_norm": 0.34975963830947876,
+      "learning_rate": 2.7121700793017214e-05,
+      "loss": 0.8711,
+      "step": 2011
+    },
+    {
+      "epoch": 0.6572210199665183,
+      "grad_norm": 0.3697607219219208,
+      "learning_rate": 2.707532621104757e-05,
+      "loss": 0.9565,
+      "step": 2012
+    },
+    {
+      "epoch": 0.6575476705728635,
+      "grad_norm": 0.41733282804489136,
+      "learning_rate": 2.702897658544473e-05,
+      "loss": 0.8783,
+      "step": 2013
+    },
+    {
+      "epoch": 0.6578743211792087,
+      "grad_norm": 0.4162128269672394,
+      "learning_rate": 2.6982651966665984e-05,
+      "loss": 1.0363,
+      "step": 2014
+    },
+    {
+      "epoch": 0.6582009717855539,
+      "grad_norm": 0.4725761115550995,
+      "learning_rate": 2.6936352405141473e-05,
+      "loss": 0.9354,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6585276223918991,
+      "grad_norm": 0.46689558029174805,
+      "learning_rate": 2.6890077951274052e-05,
+      "loss": 1.0844,
+      "step": 2016
+    },
+    {
+      "epoch": 0.6588542729982443,
+      "grad_norm": 0.526522159576416,
+      "learning_rate": 2.684382865543924e-05,
+      "loss": 1.0829,
+      "step": 2017
+    },
+    {
+      "epoch": 0.6591809236045895,
+      "grad_norm": 0.5232835412025452,
+      "learning_rate": 2.6797604567985126e-05,
+      "loss": 0.9937,
+      "step": 2018
+    },
+    {
+      "epoch": 0.6595075742109346,
+      "grad_norm": 0.5499789118766785,
+      "learning_rate": 2.675140573923246e-05,
+      "loss": 0.9148,
+      "step": 2019
+    },
+    {
+      "epoch": 0.6598342248172798,
+      "grad_norm": 0.6198474764823914,
+      "learning_rate": 2.6705232219474352e-05,
+      "loss": 1.1502,
+      "step": 2020
+    },
+    {
+      "epoch": 0.660160875423625,
+      "grad_norm": 0.7691889405250549,
+      "learning_rate": 2.6659084058976512e-05,
+      "loss": 1.2477,
+      "step": 2021
+    },
+    {
+      "epoch": 0.6604875260299702,
+      "grad_norm": 1.0155854225158691,
+      "learning_rate": 2.6612961307976925e-05,
+      "loss": 1.6555,
+      "step": 2022
+    },
+    {
+      "epoch": 0.6608141766363154,
+      "grad_norm": 1.1848665475845337,
+      "learning_rate": 2.656686401668596e-05,
+      "loss": 1.6532,
+      "step": 2023
+    },
+    {
+      "epoch": 0.6611408272426605,
+      "grad_norm": 1.3584601879119873,
+      "learning_rate": 2.6520792235286278e-05,
+      "loss": 1.418,
+      "step": 2024
+    },
+    {
+      "epoch": 0.6614674778490057,
+      "grad_norm": 1.5498560667037964,
+      "learning_rate": 2.6474746013932762e-05,
+      "loss": 1.7049,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6617941284553509,
+      "grad_norm": 0.20128127932548523,
+      "learning_rate": 2.642872540275246e-05,
+      "loss": 0.7196,
+      "step": 2026
+    },
+    {
+      "epoch": 0.6621207790616961,
+      "grad_norm": 0.2390238493680954,
+      "learning_rate": 2.638273045184455e-05,
+      "loss": 0.8654,
+      "step": 2027
+    },
+    {
+      "epoch": 0.6624474296680414,
+      "grad_norm": 0.2611159384250641,
+      "learning_rate": 2.633676121128027e-05,
+      "loss": 0.9451,
+      "step": 2028
+    },
+    {
+      "epoch": 0.6627740802743866,
+      "grad_norm": 0.2779772877693176,
+      "learning_rate": 2.629081773110288e-05,
+      "loss": 0.8268,
+      "step": 2029
+    },
+    {
+      "epoch": 0.6631007308807317,
+      "grad_norm": 0.27132806181907654,
+      "learning_rate": 2.6244900061327594e-05,
+      "loss": 0.9116,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6634273814870769,
+      "grad_norm": 0.2833230197429657,
+      "learning_rate": 2.6199008251941483e-05,
+      "loss": 0.9296,
+      "step": 2031
+    },
+    {
+      "epoch": 0.6637540320934221,
+      "grad_norm": 0.281867653131485,
+      "learning_rate": 2.615314235290357e-05,
+      "loss": 0.8266,
+      "step": 2032
+    },
+    {
+      "epoch": 0.6640806826997673,
+      "grad_norm": 0.30270928144454956,
+      "learning_rate": 2.610730241414453e-05,
+      "loss": 0.8456,
+      "step": 2033
+    },
+    {
+      "epoch": 0.6644073333061125,
+      "grad_norm": 0.3150334060192108,
+      "learning_rate": 2.6061488485566944e-05,
+      "loss": 0.8777,
+      "step": 2034
+    },
+    {
+      "epoch": 0.6647339839124576,
+      "grad_norm": 0.3402860462665558,
+      "learning_rate": 2.6015700617044914e-05,
+      "loss": 0.9275,
+      "step": 2035
+    },
+    {
+      "epoch": 0.6650606345188028,
+      "grad_norm": 0.359759658575058,
+      "learning_rate": 2.596993885842427e-05,
+      "loss": 0.8404,
+      "step": 2036
+    },
+    {
+      "epoch": 0.665387285125148,
+      "grad_norm": 0.3418113887310028,
+      "learning_rate": 2.5924203259522394e-05,
+      "loss": 0.8434,
+      "step": 2037
+    },
+    {
+      "epoch": 0.6657139357314932,
+      "grad_norm": 0.38156867027282715,
+      "learning_rate": 2.587849387012819e-05,
+      "loss": 0.8724,
+      "step": 2038
+    },
+    {
+      "epoch": 0.6660405863378384,
+      "grad_norm": 0.40769273042678833,
+      "learning_rate": 2.5832810740002035e-05,
+      "loss": 0.8907,
+      "step": 2039
+    },
+    {
+      "epoch": 0.6663672369441835,
+      "grad_norm": 0.446984201669693,
+      "learning_rate": 2.5787153918875708e-05,
+      "loss": 1.0177,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6666938875505287,
+      "grad_norm": 0.47096148133277893,
+      "learning_rate": 2.5741523456452354e-05,
+      "loss": 1.0929,
+      "step": 2041
+    },
+    {
+      "epoch": 0.6670205381568739,
+      "grad_norm": 0.4894920587539673,
+      "learning_rate": 2.569591940240642e-05,
+      "loss": 0.9955,
+      "step": 2042
+    },
+    {
+      "epoch": 0.6673471887632192,
+      "grad_norm": 0.47973906993865967,
+      "learning_rate": 2.565034180638364e-05,
+      "loss": 0.9822,
+      "step": 2043
+    },
+    {
+      "epoch": 0.6676738393695644,
+      "grad_norm": 0.5622726082801819,
+      "learning_rate": 2.5604790718000836e-05,
+      "loss": 1.0153,
+      "step": 2044
+    },
+    {
+      "epoch": 0.6680004899759096,
+      "grad_norm": 0.6539722084999084,
+      "learning_rate": 2.5559266186846135e-05,
+      "loss": 1.0542,
+      "step": 2045
+    },
+    {
+      "epoch": 0.6683271405822547,
+      "grad_norm": 0.7086659073829651,
+      "learning_rate": 2.5513768262478592e-05,
+      "loss": 1.1257,
+      "step": 2046
+    },
+    {
+      "epoch": 0.6686537911885999,
+      "grad_norm": 0.8920603394508362,
+      "learning_rate": 2.5468296994428454e-05,
+      "loss": 1.3898,
+      "step": 2047
+    },
+    {
+      "epoch": 0.6689804417949451,
+      "grad_norm": 1.0379706621170044,
+      "learning_rate": 2.542285243219679e-05,
+      "loss": 1.2713,
+      "step": 2048
+    },
+    {
+      "epoch": 0.6693070924012903,
+      "grad_norm": 1.1866716146469116,
+      "learning_rate": 2.5377434625255715e-05,
+      "loss": 1.4265,
+      "step": 2049
+    },
+    {
+      "epoch": 0.6696337430076355,
+      "grad_norm": 1.54633629322052,
+      "learning_rate": 2.5332043623048168e-05,
+      "loss": 1.9121,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6699603936139806,
+      "grad_norm": 0.19281473755836487,
+      "learning_rate": 2.5286679474987917e-05,
+      "loss": 0.6382,
+      "step": 2051
+    },
+    {
+      "epoch": 0.6702870442203258,
+      "grad_norm": 0.2434716820716858,
+      "learning_rate": 2.5241342230459498e-05,
+      "loss": 0.8263,
+      "step": 2052
+    },
+    {
+      "epoch": 0.670613694826671,
+      "grad_norm": 0.2635152041912079,
+      "learning_rate": 2.5196031938818165e-05,
+      "loss": 0.8594,
+      "step": 2053
+    },
+    {
+      "epoch": 0.6709403454330162,
+      "grad_norm": 0.2647515833377838,
+      "learning_rate": 2.5150748649389777e-05,
+      "loss": 0.7964,
+      "step": 2054
+    },
+    {
+      "epoch": 0.6712669960393614,
+      "grad_norm": 0.28728336095809937,
+      "learning_rate": 2.510549241147091e-05,
+      "loss": 0.8066,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6715936466457065,
+      "grad_norm": 0.3012993633747101,
+      "learning_rate": 2.5060263274328578e-05,
+      "loss": 0.9016,
+      "step": 2056
+    },
+    {
+      "epoch": 0.6719202972520518,
+      "grad_norm": 0.3128151595592499,
+      "learning_rate": 2.5015061287200348e-05,
+      "loss": 0.8255,
+      "step": 2057
+    },
+    {
+      "epoch": 0.672246947858397,
+      "grad_norm": 0.3259854316711426,
+      "learning_rate": 2.496988649929422e-05,
+      "loss": 0.9146,
+      "step": 2058
+    },
+    {
+      "epoch": 0.6725735984647422,
+      "grad_norm": 0.3326560854911804,
+      "learning_rate": 2.492473895978859e-05,
+      "loss": 0.8752,
+      "step": 2059
+    },
+    {
+      "epoch": 0.6729002490710874,
+      "grad_norm": 0.35559192299842834,
+      "learning_rate": 2.4879618717832187e-05,
+      "loss": 0.9212,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6732268996774325,
+      "grad_norm": 0.355741947889328,
+      "learning_rate": 2.4834525822544018e-05,
+      "loss": 0.8778,
+      "step": 2061
+    },
+    {
+      "epoch": 0.6735535502837777,
+      "grad_norm": 0.4042257070541382,
+      "learning_rate": 2.4789460323013325e-05,
+      "loss": 0.8685,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6738802008901229,
+      "grad_norm": 0.44248056411743164,
+      "learning_rate": 2.4744422268299522e-05,
+      "loss": 0.9731,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6742068514964681,
+      "grad_norm": 0.4689655900001526,
+      "learning_rate": 2.4699411707432156e-05,
+      "loss": 1.0723,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6745335021028133,
+      "grad_norm": 0.4722408354282379,
+      "learning_rate": 2.465442868941083e-05,
+      "loss": 1.0131,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6748601527091584,
+      "grad_norm": 0.5208030343055725,
+      "learning_rate": 2.4609473263205197e-05,
+      "loss": 1.0655,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6751868033155036,
+      "grad_norm": 0.5427075028419495,
+      "learning_rate": 2.456454547775478e-05,
+      "loss": 1.0494,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6755134539218488,
+      "grad_norm": 0.5571867227554321,
+      "learning_rate": 2.4519645381969153e-05,
+      "loss": 0.954,
+      "step": 2068
+    },
+    {
+      "epoch": 0.675840104528194,
+      "grad_norm": 0.6750152111053467,
+      "learning_rate": 2.447477302472762e-05,
+      "loss": 1.1531,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6761667551345392,
+      "grad_norm": 0.7053940296173096,
+      "learning_rate": 2.4429928454879357e-05,
+      "loss": 1.2184,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6764934057408845,
+      "grad_norm": 0.9125105142593384,
+      "learning_rate": 2.4385111721243264e-05,
+      "loss": 1.2626,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6768200563472296,
+      "grad_norm": 0.9873994588851929,
+      "learning_rate": 2.4340322872607957e-05,
+      "loss": 1.3534,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6771467069535748,
+      "grad_norm": 1.1085621118545532,
+      "learning_rate": 2.4295561957731678e-05,
+      "loss": 1.0356,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67747335755992,
+      "grad_norm": 1.32113778591156,
+      "learning_rate": 2.4250829025342258e-05,
+      "loss": 1.2397,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6778000081662652,
+      "grad_norm": 1.7643086910247803,
+      "learning_rate": 2.4206124124137087e-05,
+      "loss": 1.8698,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6781266587726104,
+      "grad_norm": 0.22221608459949493,
+      "learning_rate": 2.416144730278302e-05,
+      "loss": 0.73,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6784533093789555,
+      "grad_norm": 0.2354651689529419,
+      "learning_rate": 2.4116798609916347e-05,
+      "loss": 0.7509,
+      "step": 2077
+    },
+    {
+      "epoch": 0.6787799599853007,
+      "grad_norm": 0.26706814765930176,
+      "learning_rate": 2.4072178094142745e-05,
+      "loss": 0.8493,
+      "step": 2078
+    },
+    {
+      "epoch": 0.6791066105916459,
+      "grad_norm": 0.27699151635169983,
+      "learning_rate": 2.4027585804037222e-05,
+      "loss": 0.779,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6794332611979911,
+      "grad_norm": 0.28536492586135864,
+      "learning_rate": 2.3983021788144e-05,
+      "loss": 0.8581,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6797599118043363,
+      "grad_norm": 0.29376420378685,
+      "learning_rate": 2.3938486094976635e-05,
+      "loss": 0.8732,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6800865624106814,
+      "grad_norm": 0.3210821747779846,
+      "learning_rate": 2.3893978773017716e-05,
+      "loss": 0.8949,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6804132130170266,
+      "grad_norm": 0.3225792348384857,
+      "learning_rate": 2.384949987071909e-05,
+      "loss": 0.8813,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6807398636233718,
+      "grad_norm": 0.3447169065475464,
+      "learning_rate": 2.380504943650152e-05,
+      "loss": 0.9181,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6810665142297171,
+      "grad_norm": 0.3425697088241577,
+      "learning_rate": 2.3760627518754884e-05,
+      "loss": 0.8702,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6813931648360623,
+      "grad_norm": 0.3333522379398346,
+      "learning_rate": 2.371623416583797e-05,
+      "loss": 0.8651,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6817198154424075,
+      "grad_norm": 0.36659929156303406,
+      "learning_rate": 2.3671869426078502e-05,
+      "loss": 0.8974,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6820464660487526,
+      "grad_norm": 0.3901701271533966,
+      "learning_rate": 2.362753334777298e-05,
+      "loss": 0.9199,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6823731166550978,
+      "grad_norm": 0.4033861458301544,
+      "learning_rate": 2.3583225979186817e-05,
+      "loss": 0.9545,
+      "step": 2089
+    },
+    {
+      "epoch": 0.682699767261443,
+      "grad_norm": 0.4583320617675781,
+      "learning_rate": 2.3538947368554055e-05,
+      "loss": 1.1257,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6830264178677882,
+      "grad_norm": 0.49666598439216614,
+      "learning_rate": 2.3494697564077538e-05,
+      "loss": 0.9947,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6833530684741334,
+      "grad_norm": 0.4935648441314697,
+      "learning_rate": 2.3450476613928656e-05,
+      "loss": 0.9831,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6836797190804785,
+      "grad_norm": 0.5261000394821167,
+      "learning_rate": 2.3406284566247444e-05,
+      "loss": 1.0624,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6840063696868237,
+      "grad_norm": 0.598313570022583,
+      "learning_rate": 2.3362121469142458e-05,
+      "loss": 1.0826,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6843330202931689,
+      "grad_norm": 0.6607427597045898,
+      "learning_rate": 2.3317987370690712e-05,
+      "loss": 1.131,
+      "step": 2095
+    },
+    {
+      "epoch": 0.6846596708995141,
+      "grad_norm": 0.8065424561500549,
+      "learning_rate": 2.3273882318937746e-05,
+      "loss": 1.0647,
+      "step": 2096
+    },
+    {
+      "epoch": 0.6849863215058593,
+      "grad_norm": 1.110991358757019,
+      "learning_rate": 2.3229806361897343e-05,
+      "loss": 1.1449,
+      "step": 2097
+    },
+    {
+      "epoch": 0.6853129721122044,
+      "grad_norm": 1.2615911960601807,
+      "learning_rate": 2.31857595475517e-05,
+      "loss": 1.7161,
+      "step": 2098
+    },
+    {
+      "epoch": 0.6856396227185497,
+      "grad_norm": 1.3655647039413452,
+      "learning_rate": 2.3141741923851274e-05,
+      "loss": 1.693,
+      "step": 2099
+    },
+    {
+      "epoch": 0.6859662733248949,
+      "grad_norm": 1.5947473049163818,
+      "learning_rate": 2.3097753538714756e-05,
+      "loss": 2.1393,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6862929239312401,
+      "grad_norm": 0.19925794005393982,
+      "learning_rate": 2.3053794440028927e-05,
+      "loss": 0.6479,
+      "step": 2101
+    },
+    {
+      "epoch": 0.6866195745375853,
+      "grad_norm": 0.24255867302417755,
+      "learning_rate": 2.300986467564883e-05,
+      "loss": 0.8837,
+      "step": 2102
+    },
+    {
+      "epoch": 0.6869462251439304,
+      "grad_norm": 0.26114514470100403,
+      "learning_rate": 2.2965964293397412e-05,
+      "loss": 0.7501,
+      "step": 2103
+    },
+    {
+      "epoch": 0.6872728757502756,
+      "grad_norm": 0.2674042880535126,
+      "learning_rate": 2.2922093341065798e-05,
+      "loss": 0.8345,
+      "step": 2104
+    },
+    {
+      "epoch": 0.6875995263566208,
+      "grad_norm": 0.27049896121025085,
+      "learning_rate": 2.2878251866412932e-05,
+      "loss": 0.7943,
+      "step": 2105
+    },
+    {
+      "epoch": 0.687926176962966,
+      "grad_norm": 0.2783470153808594,
+      "learning_rate": 2.283443991716574e-05,
+      "loss": 0.8056,
+      "step": 2106
+    },
+    {
+      "epoch": 0.6882528275693112,
+      "grad_norm": 0.3203645944595337,
+      "learning_rate": 2.2790657541019e-05,
+      "loss": 0.8878,
+      "step": 2107
+    },
+    {
+      "epoch": 0.6885794781756563,
+      "grad_norm": 0.3191487789154053,
+      "learning_rate": 2.274690478563529e-05,
+      "loss": 0.8506,
+      "step": 2108
+    },
+    {
+      "epoch": 0.6889061287820015,
+      "grad_norm": 0.35952916741371155,
+      "learning_rate": 2.2703181698644936e-05,
+      "loss": 0.8941,
+      "step": 2109
+    },
+    {
+      "epoch": 0.6892327793883467,
+      "grad_norm": 0.3359813094139099,
+      "learning_rate": 2.2659488327645984e-05,
+      "loss": 0.9174,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6895594299946919,
+      "grad_norm": 0.33734071254730225,
+      "learning_rate": 2.26158247202041e-05,
+      "loss": 0.9184,
+      "step": 2111
+    },
+    {
+      "epoch": 0.6898860806010371,
+      "grad_norm": 0.34754154086112976,
+      "learning_rate": 2.2572190923852577e-05,
+      "loss": 0.8007,
+      "step": 2112
+    },
+    {
+      "epoch": 0.6902127312073824,
+      "grad_norm": 0.36550602316856384,
+      "learning_rate": 2.2528586986092265e-05,
+      "loss": 0.9123,
+      "step": 2113
+    },
+    {
+      "epoch": 0.6905393818137275,
+      "grad_norm": 0.3971017599105835,
+      "learning_rate": 2.2485012954391428e-05,
+      "loss": 0.9644,
+      "step": 2114
+    },
+    {
+      "epoch": 0.6908660324200727,
+      "grad_norm": 0.42068859934806824,
+      "learning_rate": 2.2441468876185905e-05,
+      "loss": 0.8395,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6911926830264179,
+      "grad_norm": 0.42162227630615234,
+      "learning_rate": 2.2397954798878794e-05,
+      "loss": 0.9714,
+      "step": 2116
+    },
+    {
+      "epoch": 0.6915193336327631,
+      "grad_norm": 0.4547198414802551,
+      "learning_rate": 2.235447076984067e-05,
+      "loss": 0.9827,
+      "step": 2117
+    },
+    {
+      "epoch": 0.6918459842391083,
+      "grad_norm": 0.5274547338485718,
+      "learning_rate": 2.2311016836409247e-05,
+      "loss": 1.0212,
+      "step": 2118
+    },
+    {
+      "epoch": 0.6921726348454534,
+      "grad_norm": 0.604054868221283,
+      "learning_rate": 2.226759304588959e-05,
+      "loss": 1.0619,
+      "step": 2119
+    },
+    {
+      "epoch": 0.6924992854517986,
+      "grad_norm": 0.6196434497833252,
+      "learning_rate": 2.2224199445553906e-05,
+      "loss": 1.127,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6928259360581438,
+      "grad_norm": 0.8015429973602295,
+      "learning_rate": 2.2180836082641537e-05,
+      "loss": 1.2795,
+      "step": 2121
+    },
+    {
+      "epoch": 0.693152586664489,
+      "grad_norm": 0.9391926527023315,
+      "learning_rate": 2.2137503004358917e-05,
+      "loss": 1.1737,
+      "step": 2122
+    },
+    {
+      "epoch": 0.6934792372708342,
+      "grad_norm": 1.2050788402557373,
+      "learning_rate": 2.2094200257879522e-05,
+      "loss": 1.6547,
+      "step": 2123
+    },
+    {
+      "epoch": 0.6938058878771793,
+      "grad_norm": 1.375234842300415,
+      "learning_rate": 2.2050927890343737e-05,
+      "loss": 1.4072,
+      "step": 2124
+    },
+    {
+      "epoch": 0.6941325384835245,
+      "grad_norm": 1.6113313436508179,
+      "learning_rate": 2.2007685948858988e-05,
+      "loss": 1.9956,
+      "step": 2125
+    },
+    {
+      "epoch": 0.6944591890898697,
+      "grad_norm": 0.21915705502033234,
+      "learning_rate": 2.196447448049952e-05,
+      "loss": 0.6856,
+      "step": 2126
+    },
+    {
+      "epoch": 0.694785839696215,
+      "grad_norm": 0.2379796952009201,
+      "learning_rate": 2.1921293532306354e-05,
+      "loss": 0.7421,
+      "step": 2127
+    },
+    {
+      "epoch": 0.6951124903025602,
+      "grad_norm": 0.2671576142311096,
+      "learning_rate": 2.187814315128741e-05,
+      "loss": 0.7911,
+      "step": 2128
+    },
+    {
+      "epoch": 0.6954391409089054,
+      "grad_norm": 0.28185057640075684,
+      "learning_rate": 2.1835023384417185e-05,
+      "loss": 0.8545,
+      "step": 2129
+    },
+    {
+      "epoch": 0.6957657915152505,
+      "grad_norm": 0.28416943550109863,
+      "learning_rate": 2.1791934278637e-05,
+      "loss": 0.8578,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6960924421215957,
+      "grad_norm": 0.3110760748386383,
+      "learning_rate": 2.1748875880854664e-05,
+      "loss": 0.9482,
+      "step": 2131
+    },
+    {
+      "epoch": 0.6964190927279409,
+      "grad_norm": 0.3088090121746063,
+      "learning_rate": 2.1705848237944632e-05,
+      "loss": 0.8229,
+      "step": 2132
+    },
+    {
+      "epoch": 0.6967457433342861,
+      "grad_norm": 0.32084354758262634,
+      "learning_rate": 2.166285139674786e-05,
+      "loss": 0.9405,
+      "step": 2133
+    },
+    {
+      "epoch": 0.6970723939406313,
+      "grad_norm": 0.3231033384799957,
+      "learning_rate": 2.161988540407177e-05,
+      "loss": 0.8388,
+      "step": 2134
+    },
+    {
+      "epoch": 0.6973990445469764,
+      "grad_norm": 0.34001418948173523,
+      "learning_rate": 2.1576950306690207e-05,
+      "loss": 0.8904,
+      "step": 2135
+    },
+    {
+      "epoch": 0.6977256951533216,
+      "grad_norm": 0.36986929178237915,
+      "learning_rate": 2.1534046151343402e-05,
+      "loss": 0.9376,
+      "step": 2136
+    },
+    {
+      "epoch": 0.6980523457596668,
+      "grad_norm": 0.40653765201568604,
+      "learning_rate": 2.1491172984737816e-05,
+      "loss": 0.9017,
+      "step": 2137
+    },
+    {
+      "epoch": 0.698378996366012,
+      "grad_norm": 0.3974648416042328,
+      "learning_rate": 2.1448330853546317e-05,
+      "loss": 0.8849,
+      "step": 2138
+    },
+    {
+      "epoch": 0.6987056469723572,
+      "grad_norm": 0.4531319737434387,
+      "learning_rate": 2.1405519804407853e-05,
+      "loss": 0.9345,
+      "step": 2139
+    },
+    {
+      "epoch": 0.6990322975787023,
+      "grad_norm": 0.46302297711372375,
+      "learning_rate": 2.1362739883927607e-05,
+      "loss": 1.1282,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6993589481850475,
+      "grad_norm": 0.5013945698738098,
+      "learning_rate": 2.131999113867686e-05,
+      "loss": 1.0689,
+      "step": 2141
+    },
+    {
+      "epoch": 0.6996855987913928,
+      "grad_norm": 0.528487503528595,
+      "learning_rate": 2.1277273615192948e-05,
+      "loss": 0.9382,
+      "step": 2142
+    },
+    {
+      "epoch": 0.700012249397738,
+      "grad_norm": 0.572884202003479,
+      "learning_rate": 2.1234587359979234e-05,
+      "loss": 1.1927,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7003389000040832,
+      "grad_norm": 0.6027082204818726,
+      "learning_rate": 2.1191932419505024e-05,
+      "loss": 1.0561,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7006655506104283,
+      "grad_norm": 0.8764929175376892,
+      "learning_rate": 2.114930884020554e-05,
+      "loss": 1.3131,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7009922012167735,
+      "grad_norm": 1.0115678310394287,
+      "learning_rate": 2.1106716668481867e-05,
+      "loss": 1.4414,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7013188518231187,
+      "grad_norm": 1.0085214376449585,
+      "learning_rate": 2.1064155950700897e-05,
+      "loss": 1.3486,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7016455024294639,
+      "grad_norm": 1.156628966331482,
+      "learning_rate": 2.1021626733195283e-05,
+      "loss": 1.2661,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7019721530358091,
+      "grad_norm": 1.412764072418213,
+      "learning_rate": 2.0979129062263386e-05,
+      "loss": 1.7921,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7022988036421542,
+      "grad_norm": 1.8179196119308472,
+      "learning_rate": 2.0936662984169186e-05,
+      "loss": 2.0696,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7026254542484994,
+      "grad_norm": 0.22813886404037476,
+      "learning_rate": 2.0894228545142362e-05,
+      "loss": 0.7838,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7029521048548446,
+      "grad_norm": 0.25454890727996826,
+      "learning_rate": 2.085182579137804e-05,
+      "loss": 0.8303,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7032787554611898,
+      "grad_norm": 0.26593437790870667,
+      "learning_rate": 2.0809454769036956e-05,
+      "loss": 0.7726,
+      "step": 2153
+    },
+    {
+      "epoch": 0.703605406067535,
+      "grad_norm": 0.2822284400463104,
+      "learning_rate": 2.076711552424522e-05,
+      "loss": 0.8171,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7039320566738801,
+      "grad_norm": 0.2791661322116852,
+      "learning_rate": 2.0724808103094383e-05,
+      "loss": 0.8399,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7042587072802254,
+      "grad_norm": 0.2727159857749939,
+      "learning_rate": 2.068253255164136e-05,
+      "loss": 0.7566,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7045853578865706,
+      "grad_norm": 0.30942875146865845,
+      "learning_rate": 2.064028891590835e-05,
+      "loss": 0.8765,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7049120084929158,
+      "grad_norm": 0.3137757480144501,
+      "learning_rate": 2.0598077241882834e-05,
+      "loss": 0.8997,
+      "step": 2158
+    },
+    {
+      "epoch": 0.705238659099261,
+      "grad_norm": 0.33212578296661377,
+      "learning_rate": 2.0555897575517474e-05,
+      "loss": 0.8881,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7055653097056062,
+      "grad_norm": 0.36027947068214417,
+      "learning_rate": 2.05137499627301e-05,
+      "loss": 0.9479,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7058919603119513,
+      "grad_norm": 0.3743271827697754,
+      "learning_rate": 2.047163444940365e-05,
+      "loss": 1.0198,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7062186109182965,
+      "grad_norm": 0.3828783929347992,
+      "learning_rate": 2.042955108138612e-05,
+      "loss": 0.9335,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7065452615246417,
+      "grad_norm": 0.3867970108985901,
+      "learning_rate": 2.0387499904490463e-05,
+      "loss": 0.9367,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7068719121309869,
+      "grad_norm": 0.39452075958251953,
+      "learning_rate": 2.03454809644947e-05,
+      "loss": 0.9833,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7071985627373321,
+      "grad_norm": 0.4327092170715332,
+      "learning_rate": 2.0303494307141614e-05,
+      "loss": 0.9356,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7075252133436772,
+      "grad_norm": 0.469226598739624,
+      "learning_rate": 2.026153997813899e-05,
+      "loss": 0.9654,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7078518639500224,
+      "grad_norm": 0.5176733136177063,
+      "learning_rate": 2.021961802315929e-05,
+      "loss": 1.1734,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7081785145563676,
+      "grad_norm": 0.5132268667221069,
+      "learning_rate": 2.0177728487839815e-05,
+      "loss": 1.0919,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7085051651627128,
+      "grad_norm": 0.5670648217201233,
+      "learning_rate": 2.0135871417782547e-05,
+      "loss": 1.0412,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7088318157690581,
+      "grad_norm": 0.7028777003288269,
+      "learning_rate": 2.009404685855415e-05,
+      "loss": 1.1546,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7091584663754033,
+      "grad_norm": 0.8300199508666992,
+      "learning_rate": 2.0052254855685808e-05,
+      "loss": 1.5468,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7094851169817484,
+      "grad_norm": 0.9810918569564819,
+      "learning_rate": 2.0010495454673412e-05,
+      "loss": 1.2278,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7098117675880936,
+      "grad_norm": 1.0002435445785522,
+      "learning_rate": 1.9968768700977203e-05,
+      "loss": 1.1674,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7101384181944388,
+      "grad_norm": 1.225787878036499,
+      "learning_rate": 1.9927074640022025e-05,
+      "loss": 1.4316,
+      "step": 2174
+    },
+    {
+      "epoch": 0.710465068800784,
+      "grad_norm": 1.3633099794387817,
+      "learning_rate": 1.988541331719701e-05,
+      "loss": 1.6954,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7107917194071292,
+      "grad_norm": 0.2008531093597412,
+      "learning_rate": 1.984378477785573e-05,
+      "loss": 0.6916,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7111183700134743,
+      "grad_norm": 0.25324690341949463,
+      "learning_rate": 1.9802189067316034e-05,
+      "loss": 0.7937,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7114450206198195,
+      "grad_norm": 0.25503653287887573,
+      "learning_rate": 1.9760626230860046e-05,
+      "loss": 0.8058,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7117716712261647,
+      "grad_norm": 0.2745538651943207,
+      "learning_rate": 1.9719096313734097e-05,
+      "loss": 0.8408,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7120983218325099,
+      "grad_norm": 0.2929899990558624,
+      "learning_rate": 1.967759936114868e-05,
+      "loss": 0.8549,
+      "step": 2180
+    },
+    {
+      "epoch": 0.712424972438855,
+      "grad_norm": 0.29745033383369446,
+      "learning_rate": 1.9636135418278407e-05,
+      "loss": 0.8669,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7127516230452002,
+      "grad_norm": 0.3240223824977875,
+      "learning_rate": 1.959470453026195e-05,
+      "loss": 0.8879,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7130782736515454,
+      "grad_norm": 0.3056224286556244,
+      "learning_rate": 1.955330674220201e-05,
+      "loss": 0.8368,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7134049242578907,
+      "grad_norm": 0.3351876437664032,
+      "learning_rate": 1.9511942099165193e-05,
+      "loss": 0.7966,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7137315748642359,
+      "grad_norm": 0.33827874064445496,
+      "learning_rate": 1.9470610646182147e-05,
+      "loss": 0.8094,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7140582254705811,
+      "grad_norm": 0.38939139246940613,
+      "learning_rate": 1.9429312428247244e-05,
+      "loss": 0.9772,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7143848760769262,
+      "grad_norm": 0.39794233441352844,
+      "learning_rate": 1.9388047490318804e-05,
+      "loss": 0.8952,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7147115266832714,
+      "grad_norm": 0.3879866898059845,
+      "learning_rate": 1.9346815877318824e-05,
+      "loss": 0.8264,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7150381772896166,
+      "grad_norm": 0.43497055768966675,
+      "learning_rate": 1.930561763413306e-05,
+      "loss": 0.8812,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7153648278959618,
+      "grad_norm": 0.47120532393455505,
+      "learning_rate": 1.9264452805610943e-05,
+      "loss": 1.0343,
+      "step": 2190
+    },
+    {
+      "epoch": 0.715691478502307,
+      "grad_norm": 0.495140016078949,
+      "learning_rate": 1.9223321436565528e-05,
+      "loss": 0.9261,
+      "step": 2191
+    },
+    {
+      "epoch": 0.7160181291086521,
+      "grad_norm": 0.47505465149879456,
+      "learning_rate": 1.9182223571773433e-05,
+      "loss": 0.9779,
+      "step": 2192
+    },
+    {
+      "epoch": 0.7163447797149973,
+      "grad_norm": 0.5782510638237,
+      "learning_rate": 1.9141159255974817e-05,
+      "loss": 1.0439,
+      "step": 2193
+    },
+    {
+      "epoch": 0.7166714303213425,
+      "grad_norm": 0.6006953120231628,
+      "learning_rate": 1.9100128533873303e-05,
+      "loss": 1.0357,
+      "step": 2194
+    },
+    {
+      "epoch": 0.7169980809276877,
+      "grad_norm": 0.6175481081008911,
+      "learning_rate": 1.9059131450135948e-05,
+      "loss": 1.1558,
+      "step": 2195
+    },
+    {
+      "epoch": 0.7173247315340329,
+      "grad_norm": 0.8074755668640137,
+      "learning_rate": 1.901816804939322e-05,
+      "loss": 1.168,
+      "step": 2196
+    },
+    {
+      "epoch": 0.717651382140378,
+      "grad_norm": 0.9405984282493591,
+      "learning_rate": 1.8977238376238827e-05,
+      "loss": 1.2225,
+      "step": 2197
+    },
+    {
+      "epoch": 0.7179780327467233,
+      "grad_norm": 1.0690345764160156,
+      "learning_rate": 1.89363424752299e-05,
+      "loss": 1.5153,
+      "step": 2198
+    },
+    {
+      "epoch": 0.7183046833530685,
+      "grad_norm": 1.3357555866241455,
+      "learning_rate": 1.8895480390886655e-05,
+      "loss": 1.31,
+      "step": 2199
+    },
+    {
+      "epoch": 0.7186313339594137,
+      "grad_norm": 1.9466137886047363,
+      "learning_rate": 1.885465216769263e-05,
+      "loss": 2.1009,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7189579845657589,
+      "grad_norm": 0.20754139125347137,
+      "learning_rate": 1.88138578500944e-05,
+      "loss": 0.7188,
+      "step": 2201
+    },
+    {
+      "epoch": 0.7192846351721041,
+      "grad_norm": 0.24247871339321136,
+      "learning_rate": 1.8773097482501683e-05,
+      "loss": 0.8604,
+      "step": 2202
+    },
+    {
+      "epoch": 0.7196112857784492,
+      "grad_norm": 0.2585400938987732,
+      "learning_rate": 1.873237110928721e-05,
+      "loss": 0.8103,
+      "step": 2203
+    },
+    {
+      "epoch": 0.7199379363847944,
+      "grad_norm": 0.2706741690635681,
+      "learning_rate": 1.869167877478673e-05,
+      "loss": 0.7943,
+      "step": 2204
+    },
+    {
+      "epoch": 0.7202645869911396,
+      "grad_norm": 0.26471370458602905,
+      "learning_rate": 1.8651020523298923e-05,
+      "loss": 0.7537,
+      "step": 2205
+    },
+    {
+      "epoch": 0.7205912375974848,
+      "grad_norm": 0.28643709421157837,
+      "learning_rate": 1.8610396399085383e-05,
+      "loss": 0.829,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72091788820383,
+      "grad_norm": 0.2852267324924469,
+      "learning_rate": 1.8569806446370486e-05,
+      "loss": 0.8191,
+      "step": 2207
+    },
+    {
+      "epoch": 0.7212445388101751,
+      "grad_norm": 0.3136483430862427,
+      "learning_rate": 1.8529250709341516e-05,
+      "loss": 0.8443,
+      "step": 2208
+    },
+    {
+      "epoch": 0.7215711894165203,
+      "grad_norm": 0.31696316599845886,
+      "learning_rate": 1.8488729232148418e-05,
+      "loss": 0.8331,
+      "step": 2209
+    },
+    {
+      "epoch": 0.7218978400228655,
+      "grad_norm": 0.30480316281318665,
+      "learning_rate": 1.8448242058903876e-05,
+      "loss": 0.7433,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7222244906292107,
+      "grad_norm": 0.3624013364315033,
+      "learning_rate": 1.8407789233683238e-05,
+      "loss": 0.9242,
+      "step": 2211
+    },
+    {
+      "epoch": 0.722551141235556,
+      "grad_norm": 0.37554246187210083,
+      "learning_rate": 1.8367370800524443e-05,
+      "loss": 0.9267,
+      "step": 2212
+    },
+    {
+      "epoch": 0.7228777918419012,
+      "grad_norm": 0.3949258625507355,
+      "learning_rate": 1.8326986803428004e-05,
+      "loss": 0.8639,
+      "step": 2213
+    },
+    {
+      "epoch": 0.7232044424482463,
+      "grad_norm": 0.432748943567276,
+      "learning_rate": 1.8286637286356926e-05,
+      "loss": 0.9655,
+      "step": 2214
+    },
+    {
+      "epoch": 0.7235310930545915,
+      "grad_norm": 0.4370628595352173,
+      "learning_rate": 1.8246322293236706e-05,
+      "loss": 0.9346,
+      "step": 2215
+    },
+    {
+      "epoch": 0.7238577436609367,
+      "grad_norm": 0.463162362575531,
+      "learning_rate": 1.8206041867955237e-05,
+      "loss": 1.0436,
+      "step": 2216
+    },
+    {
+      "epoch": 0.7241843942672819,
+      "grad_norm": 0.48003801703453064,
+      "learning_rate": 1.8165796054362782e-05,
+      "loss": 0.9261,
+      "step": 2217
+    },
+    {
+      "epoch": 0.724511044873627,
+      "grad_norm": 0.5734748244285583,
+      "learning_rate": 1.812558489627193e-05,
+      "loss": 1.1375,
+      "step": 2218
+    },
+    {
+      "epoch": 0.7248376954799722,
+      "grad_norm": 0.6173572540283203,
+      "learning_rate": 1.8085408437457562e-05,
+      "loss": 1.044,
+      "step": 2219
+    },
+    {
+      "epoch": 0.7251643460863174,
+      "grad_norm": 0.7554861903190613,
+      "learning_rate": 1.804526672165671e-05,
+      "loss": 1.0502,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7254909966926626,
+      "grad_norm": 0.9487894177436829,
+      "learning_rate": 1.8005159792568715e-05,
+      "loss": 1.4518,
+      "step": 2221
+    },
+    {
+      "epoch": 0.7258176472990078,
+      "grad_norm": 0.9427092671394348,
+      "learning_rate": 1.796508769385491e-05,
+      "loss": 0.9175,
+      "step": 2222
+    },
+    {
+      "epoch": 0.726144297905353,
+      "grad_norm": 1.1542737483978271,
+      "learning_rate": 1.7925050469138792e-05,
+      "loss": 1.48,
+      "step": 2223
+    },
+    {
+      "epoch": 0.7264709485116981,
+      "grad_norm": 1.3235793113708496,
+      "learning_rate": 1.7885048162005887e-05,
+      "loss": 1.3694,
+      "step": 2224
+    },
+    {
+      "epoch": 0.7267975991180433,
+      "grad_norm": 1.7609529495239258,
+      "learning_rate": 1.7845080816003672e-05,
+      "loss": 1.9989,
+      "step": 2225
+    },
+    {
+      "epoch": 0.7271242497243886,
+      "grad_norm": 0.20694789290428162,
+      "learning_rate": 1.78051484746416e-05,
+      "loss": 0.6905,
+      "step": 2226
+    },
+    {
+      "epoch": 0.7274509003307338,
+      "grad_norm": 0.23232302069664001,
+      "learning_rate": 1.7765251181391018e-05,
+      "loss": 0.8029,
+      "step": 2227
+    },
+    {
+      "epoch": 0.727777550937079,
+      "grad_norm": 0.2653183341026306,
+      "learning_rate": 1.772538897968509e-05,
+      "loss": 0.8776,
+      "step": 2228
+    },
+    {
+      "epoch": 0.7281042015434241,
+      "grad_norm": 0.27905741333961487,
+      "learning_rate": 1.768556191291879e-05,
+      "loss": 0.7946,
+      "step": 2229
+    },
+    {
+      "epoch": 0.7284308521497693,
+      "grad_norm": 0.2898816764354706,
+      "learning_rate": 1.7645770024448865e-05,
+      "loss": 0.8578,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7287575027561145,
+      "grad_norm": 0.298490047454834,
+      "learning_rate": 1.760601335759374e-05,
+      "loss": 0.8191,
+      "step": 2231
+    },
+    {
+      "epoch": 0.7290841533624597,
+      "grad_norm": 0.30617550015449524,
+      "learning_rate": 1.756629195563352e-05,
+      "loss": 0.8365,
+      "step": 2232
+    },
+    {
+      "epoch": 0.7294108039688049,
+      "grad_norm": 0.3224206268787384,
+      "learning_rate": 1.7526605861809853e-05,
+      "loss": 0.8298,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72973745457515,
+      "grad_norm": 0.33401164412498474,
+      "learning_rate": 1.7486955119326075e-05,
+      "loss": 0.9214,
+      "step": 2234
+    },
+    {
+      "epoch": 0.7300641051814952,
+      "grad_norm": 0.34572505950927734,
+      "learning_rate": 1.74473397713469e-05,
+      "loss": 0.8883,
+      "step": 2235
+    },
+    {
+      "epoch": 0.7303907557878404,
+      "grad_norm": 0.35026663541793823,
+      "learning_rate": 1.740775986099864e-05,
+      "loss": 0.9183,
+      "step": 2236
+    },
+    {
+      "epoch": 0.7307174063941856,
+      "grad_norm": 0.38163483142852783,
+      "learning_rate": 1.7368215431368905e-05,
+      "loss": 0.9629,
+      "step": 2237
+    },
+    {
+      "epoch": 0.7310440570005308,
+      "grad_norm": 0.4158123731613159,
+      "learning_rate": 1.732870652550677e-05,
+      "loss": 0.9803,
+      "step": 2238
+    },
+    {
+      "epoch": 0.731370707606876,
+      "grad_norm": 0.4073357880115509,
+      "learning_rate": 1.7289233186422598e-05,
+      "loss": 0.8948,
+      "step": 2239
+    },
+    {
+      "epoch": 0.7316973582132212,
+      "grad_norm": 0.453538715839386,
+      "learning_rate": 1.7249795457088064e-05,
+      "loss": 0.939,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7320240088195664,
+      "grad_norm": 0.5066556334495544,
+      "learning_rate": 1.7210393380436003e-05,
+      "loss": 1.1209,
+      "step": 2241
+    },
+    {
+      "epoch": 0.7323506594259116,
+      "grad_norm": 0.5031617283821106,
+      "learning_rate": 1.717102699936055e-05,
+      "loss": 1.0776,
+      "step": 2242
+    },
+    {
+      "epoch": 0.7326773100322568,
+      "grad_norm": 0.5175497531890869,
+      "learning_rate": 1.7131696356716863e-05,
+      "loss": 0.9019,
+      "step": 2243
+    },
+    {
+      "epoch": 0.733003960638602,
+      "grad_norm": 0.5329145789146423,
+      "learning_rate": 1.7092401495321315e-05,
+      "loss": 1.0561,
+      "step": 2244
+    },
+    {
+      "epoch": 0.7333306112449471,
+      "grad_norm": 0.6562705039978027,
+      "learning_rate": 1.7053142457951215e-05,
+      "loss": 1.171,
+      "step": 2245
+    },
+    {
+      "epoch": 0.7336572618512923,
+      "grad_norm": 0.7525113224983215,
+      "learning_rate": 1.7013919287344938e-05,
+      "loss": 1.1676,
+      "step": 2246
+    },
+    {
+      "epoch": 0.7339839124576375,
+      "grad_norm": 0.8737092018127441,
+      "learning_rate": 1.69747320262018e-05,
+      "loss": 1.138,
+      "step": 2247
+    },
+    {
+      "epoch": 0.7343105630639827,
+      "grad_norm": 1.1151552200317383,
+      "learning_rate": 1.6935580717182003e-05,
+      "loss": 1.183,
+      "step": 2248
+    },
+    {
+      "epoch": 0.7346372136703279,
+      "grad_norm": 1.3022023439407349,
+      "learning_rate": 1.689646540290668e-05,
+      "loss": 1.7971,
+      "step": 2249
+    },
+    {
+      "epoch": 0.734963864276673,
+      "grad_norm": 1.8182604312896729,
+      "learning_rate": 1.6857386125957676e-05,
+      "loss": 1.5969,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7352905148830182,
+      "grad_norm": 0.21018333733081818,
+      "learning_rate": 1.6818342928877685e-05,
+      "loss": 0.7596,
+      "step": 2251
+    },
+    {
+      "epoch": 0.7356171654893634,
+      "grad_norm": 0.22652126848697662,
+      "learning_rate": 1.6779335854170087e-05,
+      "loss": 0.8035,
+      "step": 2252
+    },
+    {
+      "epoch": 0.7359438160957086,
+      "grad_norm": 0.2571515440940857,
+      "learning_rate": 1.674036494429897e-05,
+      "loss": 0.815,
+      "step": 2253
+    },
+    {
+      "epoch": 0.7362704667020538,
+      "grad_norm": 0.2594307065010071,
+      "learning_rate": 1.6701430241688977e-05,
+      "loss": 0.8435,
+      "step": 2254
+    },
+    {
+      "epoch": 0.736597117308399,
+      "grad_norm": 0.2772468328475952,
+      "learning_rate": 1.6662531788725456e-05,
+      "loss": 0.8371,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7369237679147442,
+      "grad_norm": 0.28283077478408813,
+      "learning_rate": 1.6623669627754162e-05,
+      "loss": 0.8683,
+      "step": 2256
+    },
+    {
+      "epoch": 0.7372504185210894,
+      "grad_norm": 0.2877868115901947,
+      "learning_rate": 1.6584843801081463e-05,
+      "loss": 0.8402,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7375770691274346,
+      "grad_norm": 0.29893478751182556,
+      "learning_rate": 1.6546054350974073e-05,
+      "loss": 0.854,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7379037197337798,
+      "grad_norm": 0.3104216158390045,
+      "learning_rate": 1.650730131965915e-05,
+      "loss": 0.8575,
+      "step": 2259
+    },
+    {
+      "epoch": 0.738230370340125,
+      "grad_norm": 0.31806516647338867,
+      "learning_rate": 1.6468584749324217e-05,
+      "loss": 0.8733,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7385570209464701,
+      "grad_norm": 0.3343586027622223,
+      "learning_rate": 1.6429904682117087e-05,
+      "loss": 0.8407,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7388836715528153,
+      "grad_norm": 0.35887300968170166,
+      "learning_rate": 1.6391261160145833e-05,
+      "loss": 0.9061,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7392103221591605,
+      "grad_norm": 0.3796674311161041,
+      "learning_rate": 1.635265422547875e-05,
+      "loss": 0.959,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7395369727655057,
+      "grad_norm": 0.39429622888565063,
+      "learning_rate": 1.6314083920144295e-05,
+      "loss": 0.9746,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7398636233718509,
+      "grad_norm": 0.4258107542991638,
+      "learning_rate": 1.6275550286131064e-05,
+      "loss": 0.9826,
+      "step": 2265
+    },
+    {
+      "epoch": 0.740190273978196,
+      "grad_norm": 0.4779113233089447,
+      "learning_rate": 1.6237053365387745e-05,
+      "loss": 1.0346,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7405169245845412,
+      "grad_norm": 0.48150232434272766,
+      "learning_rate": 1.6198593199822982e-05,
+      "loss": 0.9831,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7408435751908864,
+      "grad_norm": 0.5278285145759583,
+      "learning_rate": 1.616016983130552e-05,
+      "loss": 0.9729,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7411702257972317,
+      "grad_norm": 0.590967059135437,
+      "learning_rate": 1.612178330166394e-05,
+      "loss": 1.0734,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7414968764035769,
+      "grad_norm": 0.6668139696121216,
+      "learning_rate": 1.6083433652686824e-05,
+      "loss": 1.2927,
+      "step": 2270
+    },
+    {
+      "epoch": 0.741823527009922,
+      "grad_norm": 0.7391326427459717,
+      "learning_rate": 1.6045120926122492e-05,
+      "loss": 1.1458,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7421501776162672,
+      "grad_norm": 0.8988819122314453,
+      "learning_rate": 1.6006845163679156e-05,
+      "loss": 1.2814,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7424768282226124,
+      "grad_norm": 1.155909538269043,
+      "learning_rate": 1.5968606407024743e-05,
+      "loss": 1.3477,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7428034788289576,
+      "grad_norm": 1.253177285194397,
+      "learning_rate": 1.5930404697786928e-05,
+      "loss": 1.2069,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7431301294353028,
+      "grad_norm": 1.7872546911239624,
+      "learning_rate": 1.5892240077553018e-05,
+      "loss": 2.0905,
+      "step": 2275
+    },
+    {
+      "epoch": 0.743456780041648,
+      "grad_norm": 0.21168266236782074,
+      "learning_rate": 1.5854112587869984e-05,
+      "loss": 0.7841,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7437834306479931,
+      "grad_norm": 0.24086590111255646,
+      "learning_rate": 1.5816022270244306e-05,
+      "loss": 0.7966,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7441100812543383,
+      "grad_norm": 0.26869410276412964,
+      "learning_rate": 1.57779691661421e-05,
+      "loss": 0.8551,
+      "step": 2278
+    },
+    {
+      "epoch": 0.7444367318606835,
+      "grad_norm": 0.27102309465408325,
+      "learning_rate": 1.5739953316988904e-05,
+      "loss": 0.8208,
+      "step": 2279
+    },
+    {
+      "epoch": 0.7447633824670287,
+      "grad_norm": 0.2761625647544861,
+      "learning_rate": 1.5701974764169674e-05,
+      "loss": 0.7825,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7450900330733738,
+      "grad_norm": 0.2879788279533386,
+      "learning_rate": 1.5664033549028862e-05,
+      "loss": 0.8061,
+      "step": 2281
+    },
+    {
+      "epoch": 0.745416683679719,
+      "grad_norm": 0.3057396709918976,
+      "learning_rate": 1.562612971287015e-05,
+      "loss": 0.8626,
+      "step": 2282
+    },
+    {
+      "epoch": 0.7457433342860643,
+      "grad_norm": 0.30929502844810486,
+      "learning_rate": 1.5588263296956656e-05,
+      "loss": 0.8419,
+      "step": 2283
+    },
+    {
+      "epoch": 0.7460699848924095,
+      "grad_norm": 0.32596832513809204,
+      "learning_rate": 1.5550434342510646e-05,
+      "loss": 0.8941,
+      "step": 2284
+    },
+    {
+      "epoch": 0.7463966354987547,
+      "grad_norm": 0.3355480134487152,
+      "learning_rate": 1.551264289071367e-05,
+      "loss": 0.8169,
+      "step": 2285
+    },
+    {
+      "epoch": 0.7467232861050999,
+      "grad_norm": 0.38037872314453125,
+      "learning_rate": 1.5474888982706436e-05,
+      "loss": 0.8659,
+      "step": 2286
+    },
+    {
+      "epoch": 0.747049936711445,
+      "grad_norm": 0.3562227189540863,
+      "learning_rate": 1.5437172659588793e-05,
+      "loss": 0.8644,
+      "step": 2287
+    },
+    {
+      "epoch": 0.7473765873177902,
+      "grad_norm": 0.38439279794692993,
+      "learning_rate": 1.5399493962419653e-05,
+      "loss": 0.7793,
+      "step": 2288
+    },
+    {
+      "epoch": 0.7477032379241354,
+      "grad_norm": 0.4378209412097931,
+      "learning_rate": 1.5361852932216992e-05,
+      "loss": 0.9637,
+      "step": 2289
+    },
+    {
+      "epoch": 0.7480298885304806,
+      "grad_norm": 0.4555678367614746,
+      "learning_rate": 1.5324249609957725e-05,
+      "loss": 0.9449,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7483565391368258,
+      "grad_norm": 0.5037922263145447,
+      "learning_rate": 1.528668403657782e-05,
+      "loss": 0.9338,
+      "step": 2291
+    },
+    {
+      "epoch": 0.7486831897431709,
+      "grad_norm": 0.499163955450058,
+      "learning_rate": 1.5249156252972035e-05,
+      "loss": 0.9826,
+      "step": 2292
+    },
+    {
+      "epoch": 0.7490098403495161,
+      "grad_norm": 0.6524832844734192,
+      "learning_rate": 1.5211666299994071e-05,
+      "loss": 1.0665,
+      "step": 2293
+    },
+    {
+      "epoch": 0.7493364909558613,
+      "grad_norm": 0.6492843627929688,
+      "learning_rate": 1.5174214218456407e-05,
+      "loss": 1.0666,
+      "step": 2294
+    },
+    {
+      "epoch": 0.7496631415622065,
+      "grad_norm": 0.6795768737792969,
+      "learning_rate": 1.5136800049130312e-05,
+      "loss": 1.0677,
+      "step": 2295
+    },
+    {
+      "epoch": 0.7499897921685517,
+      "grad_norm": 0.876645028591156,
+      "learning_rate": 1.5099423832745774e-05,
+      "loss": 1.5102,
+      "step": 2296
+    },
+    {
+      "epoch": 0.750316442774897,
+      "grad_norm": 1.0358703136444092,
+      "learning_rate": 1.5062085609991461e-05,
+      "loss": 1.3524,
+      "step": 2297
+    },
+    {
+      "epoch": 0.7506430933812421,
+      "grad_norm": 1.2098932266235352,
+      "learning_rate": 1.5024785421514692e-05,
+      "loss": 1.4479,
+      "step": 2298
+    },
+    {
+      "epoch": 0.7506430933812421,
+      "eval_loss": 1.028357982635498,
+      "eval_runtime": 503.6236,
+      "eval_samples_per_second": 5.119,
+      "eval_steps_per_second": 2.559,
+      "step": 2298
     }
   ],
   "logging_steps": 1,
@@ -10774,7 +16144,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.0089220027501773e+18,
+  "total_flos": 1.5133624248230216e+18,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null