diff --git "a/train_metrics.json" "b/train_metrics.json"
new file mode 100644--- /dev/null
+++ "b/train_metrics.json"
@@ -0,0 +1,6231 @@
+{
+  "final_eval": {
+    "eval_loss": 0.5153905153274536,
+    "eval_runtime": 2.7823,
+    "eval_samples_per_second": 646.94,
+    "eval_steps_per_second": 10.423,
+    "epoch": 5.0
+  },
+  "log_history": [
+    {
+      "epoch": 0.005804504295333178,
+      "grad_norm": 5.69194221496582,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 3.6452560424804688,
+      "step": 25
+    },
+    {
+      "epoch": 0.011609008590666357,
+      "grad_norm": 5.352695465087891,
+      "learning_rate": 7.35e-05,
+      "loss": 2.071630554199219,
+      "step": 50
+    },
+    {
+      "epoch": 0.017413512885999537,
+      "grad_norm": 2.875943183898926,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 1.2798121643066407,
+      "step": 75
+    },
+    {
+      "epoch": 0.023218017181332713,
+      "grad_norm": 2.5715341567993164,
+      "learning_rate": 0.00014849999999999998,
+      "loss": 0.957340087890625,
+      "step": 100
+    },
+    {
+      "epoch": 0.029022521476665893,
+      "grad_norm": 2.130218505859375,
+      "learning_rate": 0.000186,
+      "loss": 0.8733430480957032,
+      "step": 125
+    },
+    {
+      "epoch": 0.03482702577199907,
+      "grad_norm": 1.7817379236221313,
+      "learning_rate": 0.00022349999999999998,
+      "loss": 0.8688101959228516,
+      "step": 150
+    },
+    {
+      "epoch": 0.04063153006733225,
+      "grad_norm": 1.6357860565185547,
+      "learning_rate": 0.000261,
+      "loss": 0.7798442077636719,
+      "step": 175
+    },
+    {
+      "epoch": 0.046436034362665427,
+      "grad_norm": 1.8039973974227905,
+      "learning_rate": 0.0002985,
+      "loss": 0.7927973937988281,
+      "step": 200
+    },
+    {
+      "epoch": 0.05224053865799861,
+      "grad_norm": 1.6490068435668945,
+      "learning_rate": 0.0002999990633059839,
+      "loss": 0.8086864471435546,
+      "step": 225
+    },
+    {
+      "epoch": 0.05804504295333179,
+      "grad_norm": 1.3432488441467285,
+      "learning_rate": 0.00029999609549493696,
+      "loss": 0.7987381744384766,
+      "step": 250
+    },
+    {
+      "epoch": 0.06384954724866497,
+      "grad_norm": 1.5648763179779053,
+      "learning_rate": 0.0002999910949808714,
+      "loss": 0.7485366821289062,
+      "step": 275
+    },
+    {
+      "epoch": 0.06965405154399815,
+      "grad_norm": 1.4617934226989746,
+      "learning_rate": 0.00029998406183155284,
+      "loss": 0.6988004302978515,
+      "step": 300
+    },
+    {
+      "epoch": 0.07545855583933132,
+      "grad_norm": 1.2970765829086304,
+      "learning_rate": 0.00029997499614229235,
+      "loss": 0.6782588195800782,
+      "step": 325
+    },
+    {
+      "epoch": 0.0812630601346645,
+      "grad_norm": 1.2033019065856934,
+      "learning_rate": 0.00029996389803594545,
+      "loss": 0.6896741485595703,
+      "step": 350
+    },
+    {
+      "epoch": 0.08706756442999768,
+      "grad_norm": 1.0645822286605835,
+      "learning_rate": 0.00029995076766291056,
+      "loss": 0.6916038513183593,
+      "step": 375
+    },
+    {
+      "epoch": 0.09287206872533085,
+      "grad_norm": 1.1745402812957764,
+      "learning_rate": 0.00029993560520112655,
+      "loss": 0.667258529663086,
+      "step": 400
+    },
+    {
+      "epoch": 0.09867657302066403,
+      "grad_norm": 1.1060843467712402,
+      "learning_rate": 0.00029991841085607067,
+      "loss": 0.6681438446044922,
+      "step": 425
+    },
+    {
+      "epoch": 0.10448107731599722,
+      "grad_norm": 1.1345441341400146,
+      "learning_rate": 0.0002998991848607556,
+      "loss": 0.6844393157958985,
+      "step": 450
+    },
+    {
+      "epoch": 0.1102855816113304,
+      "grad_norm": 1.18587064743042,
+      "learning_rate": 0.00029987792747572647,
+      "loss": 0.6518809509277343,
+      "step": 475
+    },
+    {
+      "epoch": 0.11609008590666357,
+      "grad_norm": 1.1244792938232422,
+      "learning_rate": 0.00029985463898905693,
+      "loss": 0.6590523529052734,
+      "step": 500
+    },
+    {
+      "epoch": 0.12189459020199675,
+      "grad_norm": 1.1220368146896362,
+      "learning_rate": 0.0002998293197163459,
+      "loss": 0.6292624664306641,
+      "step": 525
+    },
+    {
+      "epoch": 0.12769909449732994,
+      "grad_norm": 1.1405223608016968,
+      "learning_rate": 0.0002998019700007127,
+      "loss": 0.6029394149780274,
+      "step": 550
+    },
+    {
+      "epoch": 0.1335035987926631,
+      "grad_norm": 1.0443934202194214,
+      "learning_rate": 0.0002997725902127925,
+      "loss": 0.6055988311767578,
+      "step": 575
+    },
+    {
+      "epoch": 0.1393081030879963,
+      "grad_norm": 1.188923716545105,
+      "learning_rate": 0.0002997411807507317,
+      "loss": 0.6280393981933594,
+      "step": 600
+    },
+    {
+      "epoch": 0.14511260738332946,
+      "grad_norm": 1.161245584487915,
+      "learning_rate": 0.0002997077420401821,
+      "loss": 0.597680549621582,
+      "step": 625
+    },
+    {
+      "epoch": 0.15091711167866265,
+      "grad_norm": 1.0529277324676514,
+      "learning_rate": 0.0002996722745342951,
+      "loss": 0.5948427963256836,
+      "step": 650
+    },
+    {
+      "epoch": 0.1567216159739958,
+      "grad_norm": 0.962946355342865,
+      "learning_rate": 0.0002996347787137162,
+      "loss": 0.5904410552978515,
+      "step": 675
+    },
+    {
+      "epoch": 0.162526120269329,
+      "grad_norm": 1.1480752229690552,
+      "learning_rate": 0.0002995952550865775,
+      "loss": 0.5858208847045898,
+      "step": 700
+    },
+    {
+      "epoch": 0.1683306245646622,
+      "grad_norm": 0.9963184595108032,
+      "learning_rate": 0.00029955370418849173,
+      "loss": 0.5757682800292969,
+      "step": 725
+    },
+    {
+      "epoch": 0.17413512885999535,
+      "grad_norm": 0.9630193710327148,
+      "learning_rate": 0.0002995101265825444,
+      "loss": 0.6166581726074218,
+      "step": 750
+    },
+    {
+      "epoch": 0.17993963315532854,
+      "grad_norm": 0.9596917033195496,
+      "learning_rate": 0.0002994645228592864,
+      "loss": 0.5472894287109376,
+      "step": 775
+    },
+    {
+      "epoch": 0.1857441374506617,
+      "grad_norm": 0.8414722681045532,
+      "learning_rate": 0.00029941689363672603,
+      "loss": 0.6030552673339844,
+      "step": 800
+    },
+    {
+      "epoch": 0.1915486417459949,
+      "grad_norm": 1.0176784992218018,
+      "learning_rate": 0.00029936723956032056,
+      "loss": 0.5507896423339844,
+      "step": 825
+    },
+    {
+      "epoch": 0.19735314604132806,
+      "grad_norm": 1.091508388519287,
+      "learning_rate": 0.0002993155613029673,
+      "loss": 0.5816512298583985,
+      "step": 850
+    },
+    {
+      "epoch": 0.20315765033666125,
+      "grad_norm": 1.3224419355392456,
+      "learning_rate": 0.00029926185956499497,
+      "loss": 0.5737775421142578,
+      "step": 875
+    },
+    {
+      "epoch": 0.20896215463199444,
+      "grad_norm": 1.048105001449585,
+      "learning_rate": 0.0002992061350741536,
+      "loss": 0.526537857055664,
+      "step": 900
+    },
+    {
+      "epoch": 0.2147666589273276,
+      "grad_norm": 0.893474280834198,
+      "learning_rate": 0.00029914838858560527,
+      "loss": 0.5547891235351563,
+      "step": 925
+    },
+    {
+      "epoch": 0.2205711632226608,
+      "grad_norm": 0.9299539923667908,
+      "learning_rate": 0.00029908862088191335,
+      "loss": 0.5534063339233398,
+      "step": 950
+    },
+    {
+      "epoch": 0.22637566751799396,
+      "grad_norm": 0.9638274908065796,
+      "learning_rate": 0.00029902683277303226,
+      "loss": 0.5526454544067383,
+      "step": 975
+    },
+    {
+      "epoch": 0.23218017181332715,
+      "grad_norm": 0.9239442348480225,
+      "learning_rate": 0.00029896302509629615,
+      "loss": 0.5293716430664063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23218017181332715,
+      "eval_loss": 0.6192298531532288,
+      "eval_runtime": 2.7502,
+      "eval_samples_per_second": 654.494,
+      "eval_steps_per_second": 10.545,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2379846761086603,
+      "grad_norm": 0.9017366766929626,
+      "learning_rate": 0.0002988971987164081,
+      "loss": 0.5201650619506836,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2437891804039935,
+      "grad_norm": 0.9971777200698853,
+      "learning_rate": 0.0002988293545254277,
+      "loss": 0.5148492050170899,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2495936846993267,
+      "grad_norm": 0.9764487743377686,
+      "learning_rate": 0.0002987594934427596,
+      "loss": 0.528057518005371,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2553981889946599,
+      "grad_norm": 4.4468607902526855,
+      "learning_rate": 0.0002986876164151406,
+      "loss": 0.5570476531982422,
+      "step": 1100
+    },
+    {
+      "epoch": 0.261202693289993,
+      "grad_norm": 0.9646501541137695,
+      "learning_rate": 0.00029861372441662714,
+      "loss": 0.504648551940918,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2670071975853262,
+      "grad_norm": 0.9292016625404358,
+      "learning_rate": 0.00029853781844858183,
+      "loss": 0.55253173828125,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2728117018806594,
+      "grad_norm": 0.8876047134399414,
+      "learning_rate": 0.00029845989953966004,
+      "loss": 0.5355271911621093,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2786162061759926,
+      "grad_norm": 0.7644872069358826,
+      "learning_rate": 0.00029837996874579597,
+      "loss": 0.5145393371582031,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2844207104713257,
+      "grad_norm": 0.8033650517463684,
+      "learning_rate": 0.0002982980271501882,
+      "loss": 0.526293716430664,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2902252147666589,
+      "grad_norm": 0.9155917763710022,
+      "learning_rate": 0.00029821407586328526,
+      "loss": 0.4953899383544922,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2960297190619921,
+      "grad_norm": 0.79842609167099,
+      "learning_rate": 0.00029812811602277016,
+      "loss": 0.48291160583496096,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3018342233573253,
+      "grad_norm": 0.7622478604316711,
+      "learning_rate": 0.00029804014879354553,
+      "loss": 0.48158973693847656,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3076387276526585,
+      "grad_norm": 1.1308367252349854,
+      "learning_rate": 0.0002979501753677174,
+      "loss": 0.5160054016113281,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3134432319479916,
+      "grad_norm": 0.9406518340110779,
+      "learning_rate": 0.00029785819696457927,
+      "loss": 0.5185168075561524,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3192477362433248,
+      "grad_norm": 0.9260241389274597,
+      "learning_rate": 0.0002977642148305952,
+      "loss": 0.5038124084472656,
+      "step": 1375
+    },
+    {
+      "epoch": 0.325052240538658,
+      "grad_norm": 1.093770980834961,
+      "learning_rate": 0.0002976682302393837,
+      "loss": 0.5050059509277344,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3308567448339912,
+      "grad_norm": 1.0194474458694458,
+      "learning_rate": 0.0002975702444916997,
+      "loss": 0.47533946990966797,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3366612491293244,
+      "grad_norm": 0.8335370421409607,
+      "learning_rate": 0.00029747025891541714,
+      "loss": 0.4969546508789062,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3424657534246575,
+      "grad_norm": 0.8942282199859619,
+      "learning_rate": 0.0002973682748655114,
+      "loss": 0.5012841033935547,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3482702577199907,
+      "grad_norm": 0.9489384889602661,
+      "learning_rate": 0.00029726429372404014,
+      "loss": 0.5426366806030274,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3540747620153239,
+      "grad_norm": 0.7177698016166687,
+      "learning_rate": 0.00029715831690012534,
+      "loss": 0.5021135711669922,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3598792663106571,
+      "grad_norm": 0.8748152256011963,
+      "learning_rate": 0.00029705034582993374,
+      "loss": 0.48067291259765627,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3656837706059902,
+      "grad_norm": 0.8138943314552307,
+      "learning_rate": 0.0002969403819766576,
+      "loss": 0.4815864181518555,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3714882749013234,
+      "grad_norm": 1.0092284679412842,
+      "learning_rate": 0.0002968284268304948,
+      "loss": 0.48710357666015625,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3772927791966566,
+      "grad_norm": 0.9536113142967224,
+      "learning_rate": 0.0002967144819086285,
+      "loss": 0.4791088104248047,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3830972834919898,
+      "grad_norm": 0.7592560052871704,
+      "learning_rate": 0.0002965985487552069,
+      "loss": 0.45892948150634766,
+      "step": 1650
+    },
+    {
+      "epoch": 0.388901787787323,
+      "grad_norm": 0.8118926286697388,
+      "learning_rate": 0.00029648062894132194,
+      "loss": 0.47702369689941404,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3947062920826561,
+      "grad_norm": 0.9567070603370667,
+      "learning_rate": 0.00029636072406498827,
+      "loss": 0.4639578628540039,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4005107963779893,
+      "grad_norm": 0.8154028058052063,
+      "learning_rate": 0.0002962388357511217,
+      "loss": 0.46378795623779295,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4063153006733225,
+      "grad_norm": 0.8768483996391296,
+      "learning_rate": 0.0002961149656515167,
+      "loss": 0.47179813385009767,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4121198049686557,
+      "grad_norm": 0.7809411883354187,
+      "learning_rate": 0.0002959891154448245,
+      "loss": 0.4563689041137695,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4179243092639889,
+      "grad_norm": 0.7244387865066528,
+      "learning_rate": 0.00029586128683653024,
+      "loss": 0.4537031936645508,
+      "step": 1800
+    },
+    {
+      "epoch": 0.423728813559322,
+      "grad_norm": 0.8164202570915222,
+      "learning_rate": 0.00029573148155892954,
+      "loss": 0.44888641357421877,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4295333178546552,
+      "grad_norm": 0.7967981696128845,
+      "learning_rate": 0.00029559970137110536,
+      "loss": 0.46922103881835936,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4353378221499884,
+      "grad_norm": 0.7919995784759521,
+      "learning_rate": 0.00029546594805890424,
+      "loss": 0.45260364532470704,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4411423264453216,
+      "grad_norm": 0.7547950744628906,
+      "learning_rate": 0.00029533022343491164,
+      "loss": 0.4570736312866211,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4469468307406547,
+      "grad_norm": 0.8537823557853699,
+      "learning_rate": 0.00029519252933842786,
+      "loss": 0.49619064331054685,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4527513350359879,
+      "grad_norm": 0.959092915058136,
+      "learning_rate": 0.00029505286763544284,
+      "loss": 0.4731414794921875,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4585558393313211,
+      "grad_norm": 0.8945581912994385,
+      "learning_rate": 0.00029491124021861087,
+      "loss": 0.4671721267700195,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4643603436266543,
+      "grad_norm": 0.8251673579216003,
+      "learning_rate": 0.00029476764900722516,
+      "loss": 0.45846458435058596,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4643603436266543,
+      "eval_loss": 0.5461503863334656,
+      "eval_runtime": 2.7522,
+      "eval_samples_per_second": 654.025,
+      "eval_steps_per_second": 10.537,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4701648479219875,
+      "grad_norm": 0.8458605408668518,
+      "learning_rate": 0.0002946220959471916,
+      "loss": 0.4619647216796875,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4759693522173206,
+      "grad_norm": 0.7998257875442505,
+      "learning_rate": 0.00029447458301100257,
+      "loss": 0.4405630493164063,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4817738565126538,
+      "grad_norm": 0.9507884383201599,
+      "learning_rate": 0.0002943251121977099,
+      "loss": 0.4459010314941406,
+      "step": 2075
+    },
+    {
+      "epoch": 0.487578360807987,
+      "grad_norm": 0.8649975061416626,
+      "learning_rate": 0.00029417368553289825,
+      "loss": 0.43765132904052734,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4933828651033202,
+      "grad_norm": 0.7733083963394165,
+      "learning_rate": 0.0002940203050686573,
+      "loss": 0.47525333404541015,
+      "step": 2125
+    },
+    {
+      "epoch": 0.4991873693986534,
+      "grad_norm": 0.734443724155426,
+      "learning_rate": 0.00029386497288355395,
+      "loss": 0.473653678894043,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5049918736939866,
+      "grad_norm": 0.812603235244751,
+      "learning_rate": 0.0002937076910826044,
+      "loss": 0.47013957977294923,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5107963779893198,
+      "grad_norm": 0.830108642578125,
+      "learning_rate": 0.0002935484617972454,
+      "loss": 0.4519306945800781,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5166008822846528,
+      "grad_norm": 0.8507217764854431,
+      "learning_rate": 0.0002933872871853054,
+      "loss": 0.43172496795654297,
+      "step": 2225
+    },
+    {
+      "epoch": 0.522405386579986,
+      "grad_norm": 0.7572310566902161,
+      "learning_rate": 0.00029322416943097546,
+      "loss": 0.45969573974609373,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5282098908753192,
+      "grad_norm": 0.7455559968948364,
+      "learning_rate": 0.0002930591107447795,
+      "loss": 0.4452248764038086,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5340143951706524,
+      "grad_norm": 0.8312911987304688,
+      "learning_rate": 0.0002928921133635442,
+      "loss": 0.45168830871582033,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5398188994659856,
+      "grad_norm": 0.7671687006950378,
+      "learning_rate": 0.0002927231795503691,
+      "loss": 0.465157585144043,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5456234037613188,
+      "grad_norm": 0.7312299013137817,
+      "learning_rate": 0.0002925523115945956,
+      "loss": 0.4251287841796875,
+      "step": 2350
+    },
+    {
+      "epoch": 0.551427908056652,
+      "grad_norm": 0.6426259875297546,
+      "learning_rate": 0.000292379511811776,
+      "loss": 0.43862083435058596,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5572324123519852,
+      "grad_norm": 0.8042207956314087,
+      "learning_rate": 0.00029220478254364225,
+      "loss": 0.4181458282470703,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5630369166473184,
+      "grad_norm": 0.9170446991920471,
+      "learning_rate": 0.000292028126158074,
+      "loss": 0.4312256622314453,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5688414209426514,
+      "grad_norm": 0.7042654752731323,
+      "learning_rate": 0.00029184954504906676,
+      "loss": 0.4229261016845703,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5746459252379846,
+      "grad_norm": 0.7955782413482666,
+      "learning_rate": 0.0002916690416366992,
+      "loss": 0.43401336669921875,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5804504295333178,
+      "grad_norm": 0.777469277381897,
+      "learning_rate": 0.00029148661836710045,
+      "loss": 0.4203028106689453,
+      "step": 2500
+    },
+    {
+      "epoch": 0.586254933828651,
+      "grad_norm": 0.764174222946167,
+      "learning_rate": 0.00029130227771241726,
+      "loss": 0.4425800704956055,
+      "step": 2525
+    },
+    {
+      "epoch": 0.5920594381239842,
+      "grad_norm": 0.8057777881622314,
+      "learning_rate": 0.00029111602217077987,
+      "loss": 0.4491798782348633,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5978639424193174,
+      "grad_norm": 0.7849870324134827,
+      "learning_rate": 0.00029092785426626874,
+      "loss": 0.4222267150878906,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6036684467146506,
+      "grad_norm": 0.8203422427177429,
+      "learning_rate": 0.00029073777654888,
+      "loss": 0.4202579116821289,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6094729510099838,
+      "grad_norm": 0.8008262515068054,
+      "learning_rate": 0.000290545791594491,
+      "loss": 0.4429645538330078,
+      "step": 2625
+    },
+    {
+      "epoch": 0.615277455305317,
+      "grad_norm": 0.6762766242027283,
+      "learning_rate": 0.0002903519020048255,
+      "loss": 0.41913318634033203,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6210819596006502,
+      "grad_norm": 0.8003602027893066,
+      "learning_rate": 0.0002901561104074182,
+      "loss": 0.42738876342773435,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6268864638959832,
+      "grad_norm": 0.8845598697662354,
+      "learning_rate": 0.0002899584194555793,
+      "loss": 0.41933670043945315,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6326909681913164,
+      "grad_norm": 0.7606611251831055,
+      "learning_rate": 0.00028975883182835857,
+      "loss": 0.40116325378417966,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6384954724866496,
+      "grad_norm": 0.6977168321609497,
+      "learning_rate": 0.0002895573502305089,
+      "loss": 0.40608341217041016,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6442999767819828,
+      "grad_norm": 0.6315252184867859,
+      "learning_rate": 0.0002893539773924496,
+      "loss": 0.4157571029663086,
+      "step": 2775
+    },
+    {
+      "epoch": 0.650104481077316,
+      "grad_norm": 0.7427784204483032,
+      "learning_rate": 0.0002891487160702297,
+      "loss": 0.39236194610595704,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6559089853726492,
+      "grad_norm": 2.8248515129089355,
+      "learning_rate": 0.0002889415690454904,
+      "loss": 0.4149022674560547,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6617134896679824,
+      "grad_norm": 0.8709903359413147,
+      "learning_rate": 0.0002887325391254272,
+      "loss": 0.44318336486816406,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6675179939633156,
+      "grad_norm": 0.7474575042724609,
+      "learning_rate": 0.0002885216291427522,
+      "loss": 0.4298463439941406,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6733224982586488,
+      "grad_norm": 0.7034773230552673,
+      "learning_rate": 0.0002883088419556555,
+      "loss": 0.40021060943603515,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6791270025539818,
+      "grad_norm": 0.7190234661102295,
+      "learning_rate": 0.0002880941804477666,
+      "loss": 0.39487300872802733,
+      "step": 2925
+    },
+    {
+      "epoch": 0.684931506849315,
+      "grad_norm": 0.6730164885520935,
+      "learning_rate": 0.00028787764752811524,
+      "loss": 0.39985904693603513,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6907360111446482,
+      "grad_norm": 0.7853715419769287,
+      "learning_rate": 0.00028765924613109175,
+      "loss": 0.3991646194458008,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6965405154399814,
+      "grad_norm": 0.6262395977973938,
+      "learning_rate": 0.00028743897921640793,
+      "loss": 0.4213618469238281,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6965405154399814,
+      "eval_loss": 0.504619300365448,
+      "eval_runtime": 2.7434,
+      "eval_samples_per_second": 656.121,
+      "eval_steps_per_second": 10.571,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7023450197353146,
+      "grad_norm": 0.7541638612747192,
+      "learning_rate": 0.0002872168497690561,
+      "loss": 0.4166010665893555,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7081495240306478,
+      "grad_norm": 0.7970810532569885,
+      "learning_rate": 0.00028699286079926934,
+      "loss": 0.42811122894287107,
+      "step": 3050
+    },
+    {
+      "epoch": 0.713954028325981,
+      "grad_norm": 0.6460123062133789,
+      "learning_rate": 0.0002867670153424804,
+      "loss": 0.39784423828125,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7197585326213142,
+      "grad_norm": 0.7262775301933289,
+      "learning_rate": 0.0002865393164592805,
+      "loss": 0.41382804870605466,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7255630369166474,
+      "grad_norm": 0.806554913520813,
+      "learning_rate": 0.00028630976723537804,
+      "loss": 0.40353801727294925,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7313675412119804,
+      "grad_norm": 0.6862145662307739,
+      "learning_rate": 0.00028607837078155666,
+      "loss": 0.42254573822021485,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7371720455073136,
+      "grad_norm": 0.7457948923110962,
+      "learning_rate": 0.00028584513023363313,
+      "loss": 0.42627681732177736,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7429765498026468,
+      "grad_norm": 0.6706002354621887,
+      "learning_rate": 0.0002856100487524148,
+      "loss": 0.39841991424560547,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74878105409798,
+      "grad_norm": 0.6904786825180054,
+      "learning_rate": 0.00028537312952365684,
+      "loss": 0.4016173553466797,
+      "step": 3225
+    },
+    {
+      "epoch": 0.7545855583933132,
+      "grad_norm": 0.7361326217651367,
+      "learning_rate": 0.0002851343757580191,
+      "loss": 0.3995735549926758,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7603900626886464,
+      "grad_norm": 0.7154432535171509,
+      "learning_rate": 0.00028489379069102235,
+      "loss": 0.39771595001220705,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7661945669839796,
+      "grad_norm": 0.7316451072692871,
+      "learning_rate": 0.00028465137758300487,
+      "loss": 0.4077155685424805,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7719990712793128,
+      "grad_norm": 0.8750970363616943,
+      "learning_rate": 0.00028440713971907775,
+      "loss": 0.39318740844726563,
+      "step": 3325
+    },
+    {
+      "epoch": 0.777803575574646,
+      "grad_norm": 0.8618521690368652,
+      "learning_rate": 0.0002841610804090809,
+      "loss": 0.41357147216796875,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7836080798699792,
+      "grad_norm": 0.6657027006149292,
+      "learning_rate": 0.0002839132029875377,
+      "loss": 0.40673721313476563,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7894125841653122,
+      "grad_norm": 0.7076691389083862,
+      "learning_rate": 0.0002836635108136102,
+      "loss": 0.36732830047607423,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7952170884606454,
+      "grad_norm": 0.712277352809906,
+      "learning_rate": 0.0002834120072710533,
+      "loss": 0.3891130065917969,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8010215927559786,
+      "grad_norm": 0.8622549176216125,
+      "learning_rate": 0.0002831586957681692,
+      "loss": 0.3946627426147461,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8068260970513118,
+      "grad_norm": 0.654416024684906,
+      "learning_rate": 0.00028290357973776093,
+      "loss": 0.4170027160644531,
+      "step": 3475
+    },
+    {
+      "epoch": 0.812630601346645,
+      "grad_norm": 0.7020919919013977,
+      "learning_rate": 0.00028264666263708605,
+      "loss": 0.4049380111694336,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8184351056419782,
+      "grad_norm": 0.7606842517852783,
+      "learning_rate": 0.00028238794794780954,
+      "loss": 0.4042626190185547,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8242396099373114,
+      "grad_norm": 0.7094178199768066,
+      "learning_rate": 0.0002821274391759569,
+      "loss": 0.38492286682128907,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8300441142326446,
+      "grad_norm": 0.696834146976471,
+      "learning_rate": 0.0002818651398518665,
+      "loss": 0.39216320037841795,
+      "step": 3575
+    },
+    {
+      "epoch": 0.8358486185279778,
+      "grad_norm": 0.6950177550315857,
+      "learning_rate": 0.0002816010535301416,
+      "loss": 0.3956956481933594,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8416531228233108,
+      "grad_norm": 0.6812747716903687,
+      "learning_rate": 0.0002813351837896026,
+      "loss": 0.39050754547119143,
+      "step": 3625
+    },
+    {
+      "epoch": 0.847457627118644,
+      "grad_norm": 0.5986670851707458,
+      "learning_rate": 0.000281067534233238,
+      "loss": 0.36653255462646483,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8532621314139772,
+      "grad_norm": 0.754496157169342,
+      "learning_rate": 0.00028079810848815603,
+      "loss": 0.3782907867431641,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8590666357093104,
+      "grad_norm": 0.7311639785766602,
+      "learning_rate": 0.00028052691020553515,
+      "loss": 0.39001686096191407,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8648711400046436,
+      "grad_norm": 0.670073390007019,
+      "learning_rate": 0.00028025394306057495,
+      "loss": 0.3773397445678711,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8706756442999768,
+      "grad_norm": 0.7744466662406921,
+      "learning_rate": 0.00027997921075244585,
+      "loss": 0.35699127197265623,
+      "step": 3750
+    },
+    {
+      "epoch": 0.87648014859531,
+      "grad_norm": 0.5562940835952759,
+      "learning_rate": 0.0002797027170042394,
+      "loss": 0.3678006744384766,
+      "step": 3775
+    },
+    {
+      "epoch": 0.8822846528906432,
+      "grad_norm": 0.69828200340271,
+      "learning_rate": 0.0002794244655629177,
+      "loss": 0.3857844161987305,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8880891571859764,
+      "grad_norm": 0.6107988357543945,
+      "learning_rate": 0.0002791444601992625,
+      "loss": 0.36248313903808593,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8938936614813094,
+      "grad_norm": 0.9355084896087646,
+      "learning_rate": 0.00027886270470782416,
+      "loss": 0.38405906677246093,
+      "step": 3850
+    },
+    {
+      "epoch": 0.8996981657766426,
+      "grad_norm": 0.7552139759063721,
+      "learning_rate": 0.00027857920290687044,
+      "loss": 0.36699806213378905,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9055026700719758,
+      "grad_norm": 0.7360921502113342,
+      "learning_rate": 0.0002782939586383344,
+      "loss": 0.3859626007080078,
+      "step": 3900
+    },
+    {
+      "epoch": 0.911307174367309,
+      "grad_norm": 0.7035590410232544,
+      "learning_rate": 0.00027800697576776256,
+      "loss": 0.3939159393310547,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9171116786626422,
+      "grad_norm": 0.7978581190109253,
+      "learning_rate": 0.0002777182581842625,
+      "loss": 0.3993148040771484,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9229161829579754,
+      "grad_norm": 0.5821986198425293,
+      "learning_rate": 0.00027742780980045,
+      "loss": 0.39007686614990233,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9287206872533086,
+      "grad_norm": 0.7438158988952637,
+      "learning_rate": 0.0002771356345523964,
+      "loss": 0.38194828033447265,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9287206872533086,
+      "eval_loss": 0.48814767599105835,
+      "eval_runtime": 2.7823,
+      "eval_samples_per_second": 646.952,
+      "eval_steps_per_second": 10.423,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9345251915486418,
+      "grad_norm": 0.6110276579856873,
+      "learning_rate": 0.0002768417363995746,
+      "loss": 0.3739725494384766,
+      "step": 4025
+    },
+    {
+      "epoch": 0.940329695843975,
+      "grad_norm": 0.9904279112815857,
+      "learning_rate": 0.0002765461193248062,
+      "loss": 0.37986766815185546,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9461342001393082,
+      "grad_norm": 0.670891284942627,
+      "learning_rate": 0.0002762487873342069,
+      "loss": 0.3712030792236328,
+      "step": 4075
+    },
+    {
+      "epoch": 0.9519387044346412,
+      "grad_norm": 0.7104296088218689,
+      "learning_rate": 0.00027594974445713244,
+      "loss": 0.3993229293823242,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9577432087299744,
+      "grad_norm": 0.6775221824645996,
+      "learning_rate": 0.00027564899474612404,
+      "loss": 0.37310684204101563,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9635477130253076,
+      "grad_norm": 0.6754724979400635,
+      "learning_rate": 0.0002753465422768535,
+      "loss": 0.39303810119628907,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9693522173206408,
+      "grad_norm": 0.7326316237449646,
+      "learning_rate": 0.0002750423911480677,
+      "loss": 0.36520904541015625,
+      "step": 4175
+    },
+    {
+      "epoch": 0.975156721615974,
+      "grad_norm": 0.6391810178756714,
+      "learning_rate": 0.0002747365454815335,
+      "loss": 0.37214405059814454,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9809612259113072,
+      "grad_norm": 0.7489984631538391,
+      "learning_rate": 0.00027442900942198145,
+      "loss": 0.3918659210205078,
+      "step": 4225
+    },
+    {
+      "epoch": 0.9867657302066404,
+      "grad_norm": 0.6832382082939148,
+      "learning_rate": 0.00027411978713704995,
+      "loss": 0.3501068878173828,
+      "step": 4250
+    },
+    {
+      "epoch": 0.9925702345019736,
+      "grad_norm": 0.5727312564849854,
+      "learning_rate": 0.00027380888281722845,
+      "loss": 0.38026947021484375,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9983747387973068,
+      "grad_norm": 0.647875189781189,
+      "learning_rate": 0.00027349630067580107,
+      "loss": 0.3627927017211914,
+      "step": 4300
+    },
+    {
+      "epoch": 1.00417924309264,
+      "grad_norm": 0.7605851292610168,
+      "learning_rate": 0.00027318204494878906,
+      "loss": 0.3483236312866211,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0099837473879731,
+      "grad_norm": 0.6403733491897583,
+      "learning_rate": 0.0002728661198948937,
+      "loss": 0.32249099731445313,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0157882516833063,
+      "grad_norm": 0.6169142723083496,
+      "learning_rate": 0.0002725485297954386,
+      "loss": 0.3157264900207519,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0215927559786395,
+      "grad_norm": 0.6535440683364868,
+      "learning_rate": 0.0002722292789543114,
+      "loss": 0.31465414047241214,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0273972602739727,
+      "grad_norm": 0.6405704617500305,
+      "learning_rate": 0.0002719083716979057,
+      "loss": 0.3279232406616211,
+      "step": 4425
+    },
+    {
+      "epoch": 1.0332017645693057,
+      "grad_norm": 0.6021073460578918,
+      "learning_rate": 0.0002715858123750625,
+      "loss": 0.33366455078125,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0390062688646389,
+      "grad_norm": 0.6265737414360046,
+      "learning_rate": 0.0002712616053570108,
+      "loss": 0.320176887512207,
+      "step": 4475
+    },
+    {
+      "epoch": 1.044810773159972,
+      "grad_norm": 0.6770696043968201,
+      "learning_rate": 0.0002709357550373089,
+      "loss": 0.34064178466796874,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0506152774553053,
+      "grad_norm": 0.628902792930603,
+      "learning_rate": 0.0002706082658317847,
+      "loss": 0.32154853820800783,
+      "step": 4525
+    },
+    {
+      "epoch": 1.0564197817506384,
+      "grad_norm": 0.7100651860237122,
+      "learning_rate": 0.0002702791421784756,
+      "loss": 0.31644344329833984,
+      "step": 4550
+    },
+    {
+      "epoch": 1.0622242860459716,
+      "grad_norm": 0.6666051149368286,
+      "learning_rate": 0.0002699483885375687,
+      "loss": 0.3197053527832031,
+      "step": 4575
+    },
+    {
+      "epoch": 1.0680287903413048,
+      "grad_norm": 0.6767443418502808,
+      "learning_rate": 0.0002696160093913401,
+      "loss": 0.32004940032958984,
+      "step": 4600
+    },
+    {
+      "epoch": 1.073833294636638,
+      "grad_norm": 0.6652418971061707,
+      "learning_rate": 0.00026928200924409445,
+      "loss": 0.32636150360107424,
+      "step": 4625
+    },
+    {
+      "epoch": 1.0796377989319712,
+      "grad_norm": 0.802137553691864,
+      "learning_rate": 0.0002689463926221036,
+      "loss": 0.3170870018005371,
+      "step": 4650
+    },
+    {
+      "epoch": 1.0854423032273044,
+      "grad_norm": 0.6242552399635315,
+      "learning_rate": 0.00026860916407354533,
+      "loss": 0.32738426208496096,
+      "step": 4675
+    },
+    {
+      "epoch": 1.0912468075226376,
+      "grad_norm": 0.5225188136100769,
+      "learning_rate": 0.0002682703281684419,
+      "loss": 0.310830135345459,
+      "step": 4700
+    },
+    {
+      "epoch": 1.0970513118179708,
+      "grad_norm": 0.6679884791374207,
+      "learning_rate": 0.00026792988949859796,
+      "loss": 0.3206803894042969,
+      "step": 4725
+    },
+    {
+      "epoch": 1.102855816113304,
+      "grad_norm": 0.6550357341766357,
+      "learning_rate": 0.0002675878526775383,
+      "loss": 0.3191901397705078,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1086603204086372,
+      "grad_norm": 0.6053130626678467,
+      "learning_rate": 0.00026724422234044546,
+      "loss": 0.3172738838195801,
+      "step": 4775
+    },
+    {
+      "epoch": 1.1144648247039703,
+      "grad_norm": 0.7310574650764465,
+      "learning_rate": 0.0002668990031440969,
+      "loss": 0.3265264892578125,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1202693289993035,
+      "grad_norm": 0.7701820135116577,
+      "learning_rate": 0.0002665521997668016,
+      "loss": 0.32262577056884767,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1260738332946367,
+      "grad_norm": 0.8463571071624756,
+      "learning_rate": 0.0002662038169083371,
+      "loss": 0.3144503974914551,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1318783375899697,
+      "grad_norm": 0.5850328207015991,
+      "learning_rate": 0.00026585385928988553,
+      "loss": 0.3229551315307617,
+      "step": 4875
+    },
+    {
+      "epoch": 1.137682841885303,
+      "grad_norm": 0.7047256827354431,
+      "learning_rate": 0.00026550233165396984,
+      "loss": 0.32570037841796873,
+      "step": 4900
+    },
+    {
+      "epoch": 1.143487346180636,
+      "grad_norm": 0.62722247838974,
+      "learning_rate": 0.00026514923876438907,
+      "loss": 0.3462512969970703,
+      "step": 4925
+    },
+    {
+      "epoch": 1.1492918504759693,
+      "grad_norm": 0.6269782185554504,
+      "learning_rate": 0.0002647945854061545,
+      "loss": 0.32021629333496093,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1550963547713025,
+      "grad_norm": 0.7987203598022461,
+      "learning_rate": 0.00026443837638542416,
+      "loss": 0.31721311569213867,
+      "step": 4975
+    },
+    {
+      "epoch": 1.1609008590666356,
+      "grad_norm": 0.5987405776977539,
+      "learning_rate": 0.00026408061652943805,
+      "loss": 0.3331464385986328,
+      "step": 5000
+    },
+    {
+      "epoch": 1.1609008590666356,
+      "eval_loss": 0.48615628480911255,
+      "eval_runtime": 2.7568,
+      "eval_samples_per_second": 652.927,
+      "eval_steps_per_second": 10.519,
+      "step": 5000
+    },
+    {
+      "epoch": 1.1667053633619688,
+      "grad_norm": 0.636762797832489,
+      "learning_rate": 0.00026372131068645266,
+      "loss": 0.33049747467041013,
+      "step": 5025
+    },
+    {
+      "epoch": 1.172509867657302,
+      "grad_norm": 0.5950357913970947,
+      "learning_rate": 0.00026336046372567524,
+      "loss": 0.31558086395263674,
+      "step": 5050
+    },
+    {
+      "epoch": 1.1783143719526352,
+      "grad_norm": 0.7616565227508545,
+      "learning_rate": 0.0002629980805371977,
+      "loss": 0.34098865509033205,
+      "step": 5075
+    },
+    {
+      "epoch": 1.1841188762479684,
+      "grad_norm": 0.8171302676200867,
+      "learning_rate": 0.00026263416603193064,
+      "loss": 0.3206678009033203,
+      "step": 5100
+    },
+    {
+      "epoch": 1.1899233805433016,
+      "grad_norm": 0.6228875517845154,
+      "learning_rate": 0.00026226872514153646,
+      "loss": 0.3341557693481445,
+      "step": 5125
+    },
+    {
+      "epoch": 1.1957278848386348,
+      "grad_norm": 0.7468209266662598,
+      "learning_rate": 0.0002619017628183627,
+      "loss": 0.3183086585998535,
+      "step": 5150
+    },
+    {
+      "epoch": 1.201532389133968,
+      "grad_norm": 0.5977431535720825,
+      "learning_rate": 0.00026153328403537507,
+      "loss": 0.2975300788879395,
+      "step": 5175
+    },
+    {
+      "epoch": 1.2073368934293012,
+      "grad_norm": 0.7497389912605286,
+      "learning_rate": 0.0002611632937860897,
+      "loss": 0.3263800430297852,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2131413977246344,
+      "grad_norm": 0.6506994962692261,
+      "learning_rate": 0.00026079179708450584,
+      "loss": 0.31382217407226565,
+      "step": 5225
+    },
+    {
+      "epoch": 1.2189459020199676,
+      "grad_norm": 0.6796441078186035,
+      "learning_rate": 0.0002604187989650375,
+      "loss": 0.32802093505859375,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2247504063153007,
+      "grad_norm": 0.606259286403656,
+      "learning_rate": 0.00026004430448244573,
+      "loss": 0.31098365783691406,
+      "step": 5275
+    },
+    {
+      "epoch": 1.230554910610634,
+      "grad_norm": 0.6228957772254944,
+      "learning_rate": 0.00025966831871176966,
+      "loss": 0.3351668548583984,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2363594149059671,
+      "grad_norm": 0.6838172078132629,
+      "learning_rate": 0.00025929084674825806,
+      "loss": 0.30724201202392576,
+      "step": 5325
+    },
+    {
+      "epoch": 1.2421639192013,
+      "grad_norm": 0.6283859014511108,
+      "learning_rate": 0.0002589118937073,
+      "loss": 0.3236090850830078,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2479684234966335,
+      "grad_norm": 0.7445445656776428,
+      "learning_rate": 0.0002585314647243559,
+      "loss": 0.3089501953125,
+      "step": 5375
+    },
+    {
+      "epoch": 1.2537729277919665,
+      "grad_norm": 0.5628088712692261,
+      "learning_rate": 0.0002581495649548874,
+      "loss": 0.3196928215026855,
+      "step": 5400
+    },
+    {
+      "epoch": 1.2595774320872997,
+      "grad_norm": 0.6176928877830505,
+      "learning_rate": 0.0002577661995742881,
+      "loss": 0.3408711624145508,
+      "step": 5425
+    },
+    {
+      "epoch": 1.2653819363826329,
+      "grad_norm": 0.6318303942680359,
+      "learning_rate": 0.0002573813737778129,
+      "loss": 0.317446346282959,
+      "step": 5450
+    },
+    {
+      "epoch": 1.271186440677966,
+      "grad_norm": 0.6661486625671387,
+      "learning_rate": 0.00025699509278050804,
+      "loss": 0.34569507598876953,
+      "step": 5475
+    },
+    {
+      "epoch": 1.2769909449732992,
+      "grad_norm": 0.7848111987113953,
+      "learning_rate": 0.0002566073618171401,
+      "loss": 0.3127158546447754,
+      "step": 5500
+    },
+    {
+      "epoch": 1.2827954492686324,
+      "grad_norm": 0.6989896893501282,
+      "learning_rate": 0.0002562181861421251,
+      "loss": 0.31853891372680665,
+      "step": 5525
+    },
+    {
+      "epoch": 1.2885999535639656,
+      "grad_norm": 0.7261764407157898,
+      "learning_rate": 0.0002558275710294575,
+      "loss": 0.3159151840209961,
+      "step": 5550
+    },
+    {
+      "epoch": 1.2944044578592988,
+      "grad_norm": 0.6303324699401855,
+      "learning_rate": 0.00025543552177263864,
+      "loss": 0.31834716796875,
+      "step": 5575
+    },
+    {
+      "epoch": 1.300208962154632,
+      "grad_norm": 0.5532225966453552,
+      "learning_rate": 0.00025504204368460463,
+      "loss": 0.30585987091064454,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3060134664499652,
+      "grad_norm": 0.6741237640380859,
+      "learning_rate": 0.0002546471420976552,
+      "loss": 0.31569082260131837,
+      "step": 5625
+    },
+    {
+      "epoch": 1.3118179707452984,
+      "grad_norm": 0.6705361008644104,
+      "learning_rate": 0.00025425082236338044,
+      "loss": 0.3247222900390625,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3176224750406316,
+      "grad_norm": 0.7508572340011597,
+      "learning_rate": 0.000253853089852589,
+      "loss": 0.29854156494140627,
+      "step": 5675
+    },
+    {
+      "epoch": 1.3234269793359648,
+      "grad_norm": 0.5746467113494873,
+      "learning_rate": 0.0002534539499552351,
+      "loss": 0.32814361572265627,
+      "step": 5700
+    },
+    {
+      "epoch": 1.329231483631298,
+      "grad_norm": 0.6579439043998718,
+      "learning_rate": 0.0002530534080803452,
+      "loss": 0.3035394096374512,
+      "step": 5725
+    },
+    {
+      "epoch": 1.3350359879266311,
+      "grad_norm": 0.6473369002342224,
+      "learning_rate": 0.00025265146965594527,
+      "loss": 0.32496734619140627,
+      "step": 5750
+    },
+    {
+      "epoch": 1.340840492221964,
+      "grad_norm": 0.6616256833076477,
+      "learning_rate": 0.00025224814012898665,
+      "loss": 0.3252885437011719,
+      "step": 5775
+    },
+    {
+      "epoch": 1.3466449965172975,
+      "grad_norm": 0.7027599811553955,
+      "learning_rate": 0.0002518434249652727,
+      "loss": 0.3071519660949707,
+      "step": 5800
+    },
+    {
+      "epoch": 1.3524495008126305,
+      "grad_norm": 0.6210665106773376,
+      "learning_rate": 0.0002514373296493843,
+      "loss": 0.2968731307983398,
+      "step": 5825
+    },
+    {
+      "epoch": 1.358254005107964,
+      "grad_norm": 0.6813532710075378,
+      "learning_rate": 0.00025102985968460594,
+      "loss": 0.31244281768798826,
+      "step": 5850
+    },
+    {
+      "epoch": 1.3640585094032969,
+      "grad_norm": 0.5574554800987244,
+      "learning_rate": 0.00025062102059285085,
+      "loss": 0.32012893676757814,
+      "step": 5875
+    },
+    {
+      "epoch": 1.36986301369863,
+      "grad_norm": 0.6099461913108826,
+      "learning_rate": 0.00025021081791458627,
+      "loss": 0.31314926147460936,
+      "step": 5900
+    },
+    {
+      "epoch": 1.3756675179939633,
+      "grad_norm": 0.612482488155365,
+      "learning_rate": 0.00024979925720875845,
+      "loss": 0.30818965911865237,
+      "step": 5925
+    },
+    {
+      "epoch": 1.3814720222892964,
+      "grad_norm": 0.5875937342643738,
+      "learning_rate": 0.000249386344052717,
+      "loss": 0.30511083602905276,
+      "step": 5950
+    },
+    {
+      "epoch": 1.3872765265846296,
+      "grad_norm": 0.6854719519615173,
+      "learning_rate": 0.00024897208404214,
+      "loss": 0.3048601341247559,
+      "step": 5975
+    },
+    {
+      "epoch": 1.3930810308799628,
+      "grad_norm": 0.7227460741996765,
+      "learning_rate": 0.00024855648279095704,
+      "loss": 0.31671728134155275,
+      "step": 6000
+    },
+    {
+      "epoch": 1.3930810308799628,
+      "eval_loss": 0.4727562367916107,
+      "eval_runtime": 2.7654,
+      "eval_samples_per_second": 650.893,
+      "eval_steps_per_second": 10.487,
+      "step": 6000
+    },
+    {
+      "epoch": 1.398885535175296,
+      "grad_norm": 0.6788727641105652,
+      "learning_rate": 0.00024813954593127445,
+      "loss": 0.2970410919189453,
+      "step": 6025
+    },
+    {
+      "epoch": 1.4046900394706292,
+      "grad_norm": 0.6451123952865601,
+      "learning_rate": 0.0002477212791132979,
+      "loss": 0.2862021827697754,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4104945437659624,
+      "grad_norm": 0.6537806391716003,
+      "learning_rate": 0.0002473016880052565,
+      "loss": 0.30356426239013673,
+      "step": 6075
+    },
+    {
+      "epoch": 1.4162990480612956,
+      "grad_norm": 0.6873378753662109,
+      "learning_rate": 0.00024688077829332567,
+      "loss": 0.3050061225891113,
+      "step": 6100
+    },
+    {
+      "epoch": 1.4221035523566288,
+      "grad_norm": 0.8558492064476013,
+      "learning_rate": 0.0002464585556815502,
+      "loss": 0.30953935623168943,
+      "step": 6125
+    },
+    {
+      "epoch": 1.427908056651962,
+      "grad_norm": 0.6895678639411926,
+      "learning_rate": 0.00024603502589176684,
+      "loss": 0.315384521484375,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4337125609472952,
+      "grad_norm": 0.5806310772895813,
+      "learning_rate": 0.0002456101946635268,
+      "loss": 0.3132205009460449,
+      "step": 6175
+    },
+    {
+      "epoch": 1.4395170652426283,
+      "grad_norm": 0.5732521414756775,
+      "learning_rate": 0.0002451840677540182,
+      "loss": 0.31554349899291995,
+      "step": 6200
+    },
+    {
+      "epoch": 1.4453215695379615,
+      "grad_norm": 0.5964354276657104,
+      "learning_rate": 0.0002447566509379876,
+      "loss": 0.31711679458618164,
+      "step": 6225
+    },
+    {
+      "epoch": 1.4511260738332945,
+      "grad_norm": 0.6099135875701904,
+      "learning_rate": 0.00024432795000766224,
+      "loss": 0.3235733413696289,
+      "step": 6250
+    },
+    {
+      "epoch": 1.456930578128628,
+      "grad_norm": 0.6290714740753174,
+      "learning_rate": 0.00024389797077267112,
+      "loss": 0.30516586303710935,
+      "step": 6275
+    },
+    {
+      "epoch": 1.4627350824239609,
+      "grad_norm": 0.6703317165374756,
+      "learning_rate": 0.00024346671905996648,
+      "loss": 0.3056545829772949,
+      "step": 6300
+    },
+    {
+      "epoch": 1.4685395867192943,
+      "grad_norm": 0.6912330985069275,
+      "learning_rate": 0.00024303420071374493,
+      "loss": 0.27505996704101565,
+      "step": 6325
+    },
+    {
+      "epoch": 1.4743440910146273,
+      "grad_norm": 0.5378541946411133,
+      "learning_rate": 0.00024260042159536793,
+      "loss": 0.2972256278991699,
+      "step": 6350
+    },
+    {
+      "epoch": 1.4801485953099605,
+      "grad_norm": 0.7520794868469238,
+      "learning_rate": 0.0002421653875832827,
+      "loss": 0.32843109130859377,
+      "step": 6375
+    },
+    {
+      "epoch": 1.4859530996052936,
+      "grad_norm": 0.6301512718200684,
+      "learning_rate": 0.00024172910457294236,
+      "loss": 0.29020545959472654,
+      "step": 6400
+    },
+    {
+      "epoch": 1.4917576039006268,
+      "grad_norm": 0.743053138256073,
+      "learning_rate": 0.00024129157847672605,
+      "loss": 0.3057110404968262,
+      "step": 6425
+    },
+    {
+      "epoch": 1.49756210819596,
+      "grad_norm": 0.5653263926506042,
+      "learning_rate": 0.00024085281522385894,
+      "loss": 0.33673030853271485,
+      "step": 6450
+    },
+    {
+      "epoch": 1.5033666124912932,
+      "grad_norm": 0.5811220407485962,
+      "learning_rate": 0.00024041282076033165,
+      "loss": 0.30766809463500977,
+      "step": 6475
+    },
+    {
+      "epoch": 1.5091711167866264,
+      "grad_norm": 0.5101409554481506,
+      "learning_rate": 0.00023997160104882,
+      "loss": 0.3042826843261719,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5149756210819596,
+      "grad_norm": 0.5552115440368652,
+      "learning_rate": 0.00023952916206860378,
+      "loss": 0.311158447265625,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5207801253772928,
+      "grad_norm": 0.561336100101471,
+      "learning_rate": 0.00023908550981548612,
+      "loss": 0.29308698654174803,
+      "step": 6550
+    },
+    {
+      "epoch": 1.526584629672626,
+      "grad_norm": 0.543438196182251,
+      "learning_rate": 0.00023864065030171198,
+      "loss": 0.3029079627990723,
+      "step": 6575
+    },
+    {
+      "epoch": 1.5323891339679592,
+      "grad_norm": 0.6324707269668579,
+      "learning_rate": 0.00023819458955588686,
+      "loss": 0.2861163520812988,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5381936382632924,
+      "grad_norm": 0.6454148888587952,
+      "learning_rate": 0.0002377473336228949,
+      "loss": 0.29912094116210936,
+      "step": 6625
+    },
+    {
+      "epoch": 1.5439981425586256,
+      "grad_norm": 0.6285545229911804,
+      "learning_rate": 0.0002372988885638171,
+      "loss": 0.29781471252441405,
+      "step": 6650
+    },
+    {
+      "epoch": 1.5498026468539585,
+      "grad_norm": 0.5753786563873291,
+      "learning_rate": 0.00023684926045584916,
+      "loss": 0.2966350173950195,
+      "step": 6675
+    },
+    {
+      "epoch": 1.555607151149292,
+      "grad_norm": 0.6003463268280029,
+      "learning_rate": 0.00023639845539221907,
+      "loss": 0.28225135803222656,
+      "step": 6700
+    },
+    {
+      "epoch": 1.561411655444625,
+      "grad_norm": 0.6339415907859802,
+      "learning_rate": 0.00023594647948210464,
+      "loss": 0.32587059020996095,
+      "step": 6725
+    },
+    {
+      "epoch": 1.5672161597399583,
+      "grad_norm": 0.7381962537765503,
+      "learning_rate": 0.00023549333885055065,
+      "loss": 0.30733226776123046,
+      "step": 6750
+    },
+    {
+      "epoch": 1.5730206640352913,
+      "grad_norm": 0.6690865159034729,
+      "learning_rate": 0.00023503903963838574,
+      "loss": 0.30211912155151366,
+      "step": 6775
+    },
+    {
+      "epoch": 1.5788251683306247,
+      "grad_norm": 0.6530346870422363,
+      "learning_rate": 0.00023458358800213942,
+      "loss": 0.2939929008483887,
+      "step": 6800
+    },
+    {
+      "epoch": 1.5846296726259577,
+      "grad_norm": 0.6665822863578796,
+      "learning_rate": 0.00023412699011395843,
+      "loss": 0.305390510559082,
+      "step": 6825
+    },
+    {
+      "epoch": 1.590434176921291,
+      "grad_norm": 0.6341019868850708,
+      "learning_rate": 0.0002336692521615232,
+      "loss": 0.33179832458496095,
+      "step": 6850
+    },
+    {
+      "epoch": 1.596238681216624,
+      "grad_norm": 0.6109963059425354,
+      "learning_rate": 0.000233210380347964,
+      "loss": 0.3017910957336426,
+      "step": 6875
+    },
+    {
+      "epoch": 1.6020431855119572,
+      "grad_norm": 0.6072869300842285,
+      "learning_rate": 0.00023275038089177682,
+      "loss": 0.29215761184692385,
+      "step": 6900
+    },
+    {
+      "epoch": 1.6078476898072904,
+      "grad_norm": 0.6945996880531311,
+      "learning_rate": 0.00023228926002673915,
+      "loss": 0.31206249237060546,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6136521941026236,
+      "grad_norm": 0.6656129956245422,
+      "learning_rate": 0.00023182702400182547,
+      "loss": 0.300494441986084,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6194566983979568,
+      "grad_norm": 0.5853510499000549,
+      "learning_rate": 0.00023136367908112263,
+      "loss": 0.3122475814819336,
+      "step": 6975
+    },
+    {
+      "epoch": 1.62526120269329,
+      "grad_norm": 0.6586392521858215,
+      "learning_rate": 0.00023089923154374483,
+      "loss": 0.30273584365844725,
+      "step": 7000
+    },
+    {
+      "epoch": 1.62526120269329,
+      "eval_loss": 0.46578022837638855,
+      "eval_runtime": 2.7583,
+      "eval_samples_per_second": 652.587,
+      "eval_steps_per_second": 10.514,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6310657069886232,
+      "grad_norm": 0.6447516083717346,
+      "learning_rate": 0.00023043368768374864,
+      "loss": 0.2987114906311035,
+      "step": 7025
+    },
+    {
+      "epoch": 1.6368702112839564,
+      "grad_norm": 0.5993840098381042,
+      "learning_rate": 0.0002299670538100477,
+      "loss": 0.3093856620788574,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6426747155792896,
+      "grad_norm": 0.5721951723098755,
+      "learning_rate": 0.0002294993362463272,
+      "loss": 0.30892154693603513,
+      "step": 7075
+    },
+    {
+      "epoch": 1.6484792198746228,
+      "grad_norm": 0.5984885096549988,
+      "learning_rate": 0.00022903054133095805,
+      "loss": 0.31562023162841796,
+      "step": 7100
+    },
+    {
+      "epoch": 1.654283724169956,
+      "grad_norm": 0.6952109336853027,
+      "learning_rate": 0.00022856067541691136,
+      "loss": 0.2940500831604004,
+      "step": 7125
+    },
+    {
+      "epoch": 1.660088228465289,
+      "grad_norm": 0.6385327577590942,
+      "learning_rate": 0.0002280897448716719,
+      "loss": 0.2970525550842285,
+      "step": 7150
+    },
+    {
+      "epoch": 1.6658927327606223,
+      "grad_norm": 0.7174793481826782,
+      "learning_rate": 0.00022761775607715192,
+      "loss": 0.2998079872131348,
+      "step": 7175
+    },
+    {
+      "epoch": 1.6716972370559553,
+      "grad_norm": 0.6188244223594666,
+      "learning_rate": 0.00022714471542960513,
+      "loss": 0.2862828254699707,
+      "step": 7200
+    },
+    {
+      "epoch": 1.6775017413512887,
+      "grad_norm": 0.595618486404419,
+      "learning_rate": 0.00022667062933953922,
+      "loss": 0.29190004348754883,
+      "step": 7225
+    },
+    {
+      "epoch": 1.6833062456466217,
+      "grad_norm": 0.6140469908714294,
+      "learning_rate": 0.00022619550423162973,
+      "loss": 0.28848487854003907,
+      "step": 7250
+    },
+    {
+      "epoch": 1.689110749941955,
+      "grad_norm": 0.5486945509910583,
+      "learning_rate": 0.0002257193465446325,
+      "loss": 0.298336067199707,
+      "step": 7275
+    },
+    {
+      "epoch": 1.694915254237288,
+      "grad_norm": 0.6571700572967529,
+      "learning_rate": 0.00022524216273129655,
+      "loss": 0.3010477256774902,
+      "step": 7300
+    },
+    {
+      "epoch": 1.7007197585326215,
+      "grad_norm": 0.67644864320755,
+      "learning_rate": 0.00022476395925827682,
+      "loss": 0.2858085823059082,
+      "step": 7325
+    },
+    {
+      "epoch": 1.7065242628279544,
+      "grad_norm": 0.5826391577720642,
+      "learning_rate": 0.00022428474260604618,
+      "loss": 0.30058721542358396,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7123287671232876,
+      "grad_norm": 0.6315799355506897,
+      "learning_rate": 0.000223804519268808,
+      "loss": 0.2961753463745117,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7181332714186208,
+      "grad_norm": 0.5751924514770508,
+      "learning_rate": 0.00022332329575440782,
+      "loss": 0.2841574478149414,
+      "step": 7400
+    },
+    {
+      "epoch": 1.723937775713954,
+      "grad_norm": 0.666958749294281,
+      "learning_rate": 0.00022284107858424533,
+      "loss": 0.2895758056640625,
+      "step": 7425
+    },
+    {
+      "epoch": 1.7297422800092872,
+      "grad_norm": 0.6232280135154724,
+      "learning_rate": 0.00022235787429318604,
+      "loss": 0.27466938018798825,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7355467843046204,
+      "grad_norm": 0.6165049076080322,
+      "learning_rate": 0.00022187368942947231,
+      "loss": 0.2787031555175781,
+      "step": 7475
+    },
+    {
+      "epoch": 1.7413512885999536,
+      "grad_norm": 0.587227463722229,
+      "learning_rate": 0.00022138853055463545,
+      "loss": 0.2846013450622559,
+      "step": 7500
+    },
+    {
+      "epoch": 1.7471557928952868,
+      "grad_norm": 0.6277509927749634,
+      "learning_rate": 0.0002209024042434058,
+      "loss": 0.2921302032470703,
+      "step": 7525
+    },
+    {
+      "epoch": 1.75296029719062,
+      "grad_norm": 0.5310524702072144,
+      "learning_rate": 0.00022041531708362434,
+      "loss": 0.2808640670776367,
+      "step": 7550
+    },
+    {
+      "epoch": 1.758764801485953,
+      "grad_norm": 0.4926140606403351,
+      "learning_rate": 0.00021992727567615326,
+      "loss": 0.2831878662109375,
+      "step": 7575
+    },
+    {
+      "epoch": 1.7645693057812863,
+      "grad_norm": 0.596112847328186,
+      "learning_rate": 0.00021943828663478616,
+      "loss": 0.30501962661743165,
+      "step": 7600
+    },
+    {
+      "epoch": 1.7703738100766193,
+      "grad_norm": 0.5885482430458069,
+      "learning_rate": 0.00021894835658615904,
+      "loss": 0.27782888412475587,
+      "step": 7625
+    },
+    {
+      "epoch": 1.7761783143719527,
+      "grad_norm": 0.675792932510376,
+      "learning_rate": 0.00021845749216965984,
+      "loss": 0.29576391220092774,
+      "step": 7650
+    },
+    {
+      "epoch": 1.7819828186672857,
+      "grad_norm": 0.6968962550163269,
+      "learning_rate": 0.00021796570003733905,
+      "loss": 0.2900502586364746,
+      "step": 7675
+    },
+    {
+      "epoch": 1.787787322962619,
+      "grad_norm": 0.6951692700386047,
+      "learning_rate": 0.00021747298685381906,
+      "loss": 0.3032721710205078,
+      "step": 7700
+    },
+    {
+      "epoch": 1.793591827257952,
+      "grad_norm": 0.6293168663978577,
+      "learning_rate": 0.0002169793592962042,
+      "loss": 0.2854315376281738,
+      "step": 7725
+    },
+    {
+      "epoch": 1.7993963315532855,
+      "grad_norm": 0.6880147457122803,
+      "learning_rate": 0.0002164848240539902,
+      "loss": 0.2785432624816895,
+      "step": 7750
+    },
+    {
+      "epoch": 1.8052008358486185,
+      "grad_norm": 0.6682071685791016,
+      "learning_rate": 0.00021598938782897324,
+      "loss": 0.29008663177490235,
+      "step": 7775
+    },
+    {
+      "epoch": 1.8110053401439519,
+      "grad_norm": 0.5785953998565674,
+      "learning_rate": 0.00021549305733515955,
+      "loss": 0.2805482864379883,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8168098444392848,
+      "grad_norm": 0.5949696898460388,
+      "learning_rate": 0.00021499583929867415,
+      "loss": 0.2796248626708984,
+      "step": 7825
+    },
+    {
+      "epoch": 1.822614348734618,
+      "grad_norm": 0.6010411977767944,
+      "learning_rate": 0.00021449774045766978,
+      "loss": 0.2813014221191406,
+      "step": 7850
+    },
+    {
+      "epoch": 1.8284188530299512,
+      "grad_norm": 0.6029084920883179,
+      "learning_rate": 0.0002139987675622357,
+      "loss": 0.28535099029541017,
+      "step": 7875
+    },
+    {
+      "epoch": 1.8342233573252844,
+      "grad_norm": 0.5701190233230591,
+      "learning_rate": 0.00021349892737430577,
+      "loss": 0.2748454475402832,
+      "step": 7900
+    },
+    {
+      "epoch": 1.8400278616206176,
+      "grad_norm": 0.6483235359191895,
+      "learning_rate": 0.00021299822666756763,
+      "loss": 0.28274688720703123,
+      "step": 7925
+    },
+    {
+      "epoch": 1.8458323659159508,
+      "grad_norm": 0.538735568523407,
+      "learning_rate": 0.00021249667222737,
+      "loss": 0.27964962005615235,
+      "step": 7950
+    },
+    {
+      "epoch": 1.851636870211284,
+      "grad_norm": 0.6208842992782593,
+      "learning_rate": 0.00021199427085063133,
+      "loss": 0.28368295669555665,
+      "step": 7975
+    },
+    {
+      "epoch": 1.8574413745066172,
+      "grad_norm": 0.5354974865913391,
+      "learning_rate": 0.00021149102934574747,
+      "loss": 0.26251665115356443,
+      "step": 8000
+    },
+    {
+      "epoch": 1.8574413745066172,
+      "eval_loss": 0.45589447021484375,
+      "eval_runtime": 2.7517,
+      "eval_samples_per_second": 654.147,
+      "eval_steps_per_second": 10.539,
+      "step": 8000
+    },
+    {
+      "epoch": 1.8632458788019504,
+      "grad_norm": 0.5648078918457031,
+      "learning_rate": 0.00021098695453249949,
+      "loss": 0.28487659454345704,
+      "step": 8025
+    },
+    {
+      "epoch": 1.8690503830972833,
+      "grad_norm": 0.5411435961723328,
+      "learning_rate": 0.00021048205324196111,
+      "loss": 0.27927705764770505,
+      "step": 8050
+    },
+    {
+      "epoch": 1.8748548873926167,
+      "grad_norm": 0.5399224758148193,
+      "learning_rate": 0.0002099763323164063,
+      "loss": 0.2783413505554199,
+      "step": 8075
+    },
+    {
+      "epoch": 1.8806593916879497,
+      "grad_norm": 0.5484160780906677,
+      "learning_rate": 0.00020946979860921651,
+      "loss": 0.29580570220947267,
+      "step": 8100
+    },
+    {
+      "epoch": 1.8864638959832831,
+      "grad_norm": 0.7539390921592712,
+      "learning_rate": 0.00020896245898478769,
+      "loss": 0.28216598510742186,
+      "step": 8125
+    },
+    {
+      "epoch": 1.892268400278616,
+      "grad_norm": 0.7425509095191956,
+      "learning_rate": 0.0002084543203184373,
+      "loss": 0.28667551040649414,
+      "step": 8150
+    },
+    {
+      "epoch": 1.8980729045739495,
+      "grad_norm": 0.6277238130569458,
+      "learning_rate": 0.00020794538949631144,
+      "loss": 0.2975791358947754,
+      "step": 8175
+    },
+    {
+      "epoch": 1.9038774088692825,
+      "grad_norm": 0.5314567685127258,
+      "learning_rate": 0.0002074356734152909,
+      "loss": 0.2790555953979492,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9096819131646159,
+      "grad_norm": 0.6067845821380615,
+      "learning_rate": 0.00020692517898289844,
+      "loss": 0.3026649665832519,
+      "step": 8225
+    },
+    {
+      "epoch": 1.9154864174599489,
+      "grad_norm": 0.6647687554359436,
+      "learning_rate": 0.00020641391311720448,
+      "loss": 0.27243967056274415,
+      "step": 8250
+    },
+    {
+      "epoch": 1.921290921755282,
+      "grad_norm": 0.5201959609985352,
+      "learning_rate": 0.00020590188274673387,
+      "loss": 0.2778730773925781,
+      "step": 8275
+    },
+    {
+      "epoch": 1.9270954260506152,
+      "grad_norm": 0.5226133465766907,
+      "learning_rate": 0.0002053890948103718,
+      "loss": 0.27872365951538086,
+      "step": 8300
+    },
+    {
+      "epoch": 1.9328999303459484,
+      "grad_norm": 0.5825538635253906,
+      "learning_rate": 0.00020487555625726972,
+      "loss": 0.29387989044189455,
+      "step": 8325
+    },
+    {
+      "epoch": 1.9387044346412816,
+      "grad_norm": 0.5677186846733093,
+      "learning_rate": 0.00020436127404675116,
+      "loss": 0.2941434288024902,
+      "step": 8350
+    },
+    {
+      "epoch": 1.9445089389366148,
+      "grad_norm": 0.6852154731750488,
+      "learning_rate": 0.00020384625514821764,
+      "loss": 0.2821710205078125,
+      "step": 8375
+    },
+    {
+      "epoch": 1.950313443231948,
+      "grad_norm": 0.509062647819519,
+      "learning_rate": 0.0002033305065410539,
+      "loss": 0.2724331474304199,
+      "step": 8400
+    },
+    {
+      "epoch": 1.9561179475272812,
+      "grad_norm": 0.5549076795578003,
+      "learning_rate": 0.0002028140352145336,
+      "loss": 0.2702467346191406,
+      "step": 8425
+    },
+    {
+      "epoch": 1.9619224518226144,
+      "grad_norm": 0.6263270378112793,
+      "learning_rate": 0.00020229684816772437,
+      "loss": 0.2931397819519043,
+      "step": 8450
+    },
+    {
+      "epoch": 1.9677269561179476,
+      "grad_norm": 0.6422144770622253,
+      "learning_rate": 0.00020177895240939314,
+      "loss": 0.2782253074645996,
+      "step": 8475
+    },
+    {
+      "epoch": 1.9735314604132808,
+      "grad_norm": 0.5915998220443726,
+      "learning_rate": 0.0002012603549579111,
+      "loss": 0.2896983528137207,
+      "step": 8500
+    },
+    {
+      "epoch": 1.9793359647086137,
+      "grad_norm": 0.5663604140281677,
+      "learning_rate": 0.0002007410628411585,
+      "loss": 0.26976789474487306,
+      "step": 8525
+    },
+    {
+      "epoch": 1.9851404690039471,
+      "grad_norm": 0.6165004372596741,
+      "learning_rate": 0.00020022108309642955,
+      "loss": 0.278371696472168,
+      "step": 8550
+    },
+    {
+      "epoch": 1.99094497329928,
+      "grad_norm": 0.5176596641540527,
+      "learning_rate": 0.00019970042277033705,
+      "loss": 0.2713899612426758,
+      "step": 8575
+    },
+    {
+      "epoch": 1.9967494775946135,
+      "grad_norm": 0.5988627672195435,
+      "learning_rate": 0.00019917908891871667,
+      "loss": 0.30117774963378907,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0025539818899465,
+      "grad_norm": 0.5397821664810181,
+      "learning_rate": 0.0001986570886065317,
+      "loss": 0.2492702293395996,
+      "step": 8625
+    },
+    {
+      "epoch": 2.00835848618528,
+      "grad_norm": 0.4858548641204834,
+      "learning_rate": 0.00019813442890777695,
+      "loss": 0.2141697311401367,
+      "step": 8650
+    },
+    {
+      "epoch": 2.014162990480613,
+      "grad_norm": 0.5113976001739502,
+      "learning_rate": 0.00019761111690538318,
+      "loss": 0.21148900985717772,
+      "step": 8675
+    },
+    {
+      "epoch": 2.0199674947759463,
+      "grad_norm": 0.644262433052063,
+      "learning_rate": 0.0001970871596911208,
+      "loss": 0.2366372299194336,
+      "step": 8700
+    },
+    {
+      "epoch": 2.0257719990712793,
+      "grad_norm": 0.5584157705307007,
+      "learning_rate": 0.00019656256436550412,
+      "loss": 0.21513431549072265,
+      "step": 8725
+    },
+    {
+      "epoch": 2.0315765033666127,
+      "grad_norm": 0.49645885825157166,
+      "learning_rate": 0.00019603733803769482,
+      "loss": 0.22790033340454102,
+      "step": 8750
+    },
+    {
+      "epoch": 2.0373810076619456,
+      "grad_norm": 0.6204771399497986,
+      "learning_rate": 0.00019551148782540585,
+      "loss": 0.2140917205810547,
+      "step": 8775
+    },
+    {
+      "epoch": 2.043185511957279,
+      "grad_norm": 0.6201297044754028,
+      "learning_rate": 0.0001949850208548047,
+      "loss": 0.20648632049560547,
+      "step": 8800
+    },
+    {
+      "epoch": 2.048990016252612,
+      "grad_norm": 0.5970756411552429,
+      "learning_rate": 0.00019445794426041718,
+      "loss": 0.20710657119750978,
+      "step": 8825
+    },
+    {
+      "epoch": 2.0547945205479454,
+      "grad_norm": 0.5841864347457886,
+      "learning_rate": 0.00019393026518503044,
+      "loss": 0.20560750961303711,
+      "step": 8850
+    },
+    {
+      "epoch": 2.0605990248432784,
+      "grad_norm": 0.6393046379089355,
+      "learning_rate": 0.00019340199077959617,
+      "loss": 0.22906755447387694,
+      "step": 8875
+    },
+    {
+      "epoch": 2.0664035291386114,
+      "grad_norm": 0.6178106069564819,
+      "learning_rate": 0.0001928731282031341,
+      "loss": 0.21789752960205078,
+      "step": 8900
+    },
+    {
+      "epoch": 2.0722080334339448,
+      "grad_norm": 0.5272368788719177,
+      "learning_rate": 0.00019234368462263428,
+      "loss": 0.22274776458740234,
+      "step": 8925
+    },
+    {
+      "epoch": 2.0780125377292777,
+      "grad_norm": 0.6374201774597168,
+      "learning_rate": 0.0001918136672129607,
+      "loss": 0.22993766784667968,
+      "step": 8950
+    },
+    {
+      "epoch": 2.083817042024611,
+      "grad_norm": 0.6094024181365967,
+      "learning_rate": 0.00019128308315675358,
+      "loss": 0.21363552093505858,
+      "step": 8975
+    },
+    {
+      "epoch": 2.089621546319944,
+      "grad_norm": 0.46703869104385376,
+      "learning_rate": 0.00019075193964433211,
+      "loss": 0.2028689956665039,
+      "step": 9000
+    },
+    {
+      "epoch": 2.089621546319944,
+      "eval_loss": 0.4722580909729004,
+      "eval_runtime": 2.7862,
+      "eval_samples_per_second": 646.041,
+      "eval_steps_per_second": 10.408,
+      "step": 9000
+    },
+    {
+      "epoch": 2.0954260506152775,
+      "grad_norm": 0.563557505607605,
+      "learning_rate": 0.00019022024387359713,
+      "loss": 0.2152543830871582,
+      "step": 9025
+    },
+    {
+      "epoch": 2.1012305549106105,
+      "grad_norm": 0.6211534738540649,
+      "learning_rate": 0.0001896880030499336,
+      "loss": 0.22218914031982423,
+      "step": 9050
+    },
+    {
+      "epoch": 2.107035059205944,
+      "grad_norm": 0.6554946303367615,
+      "learning_rate": 0.0001891552243861127,
+      "loss": 0.21623214721679687,
+      "step": 9075
+    },
+    {
+      "epoch": 2.112839563501277,
+      "grad_norm": 0.6000050902366638,
+      "learning_rate": 0.00018862191510219442,
+      "loss": 0.21452957153320312,
+      "step": 9100
+    },
+    {
+      "epoch": 2.1186440677966103,
+      "grad_norm": 0.5918187499046326,
+      "learning_rate": 0.00018808808242542938,
+      "loss": 0.21045501708984374,
+      "step": 9125
+    },
+    {
+      "epoch": 2.1244485720919433,
+      "grad_norm": 0.522652804851532,
+      "learning_rate": 0.00018755373359016138,
+      "loss": 0.22067995071411134,
+      "step": 9150
+    },
+    {
+      "epoch": 2.1302530763872767,
+      "grad_norm": 0.574083685874939,
+      "learning_rate": 0.00018701887583772863,
+      "loss": 0.21983745574951172,
+      "step": 9175
+    },
+    {
+      "epoch": 2.1360575806826096,
+      "grad_norm": 0.4899635314941406,
+      "learning_rate": 0.00018648351641636644,
+      "loss": 0.22111549377441406,
+      "step": 9200
+    },
+    {
+      "epoch": 2.141862084977943,
+      "grad_norm": 0.6875515580177307,
+      "learning_rate": 0.0001859476625811083,
+      "loss": 0.22991109848022462,
+      "step": 9225
+    },
+    {
+      "epoch": 2.147666589273276,
+      "grad_norm": 0.6430684328079224,
+      "learning_rate": 0.00018541132159368798,
+      "loss": 0.22644443511962892,
+      "step": 9250
+    },
+    {
+      "epoch": 2.1534710935686094,
+      "grad_norm": 0.5825778245925903,
+      "learning_rate": 0.00018487450072244093,
+      "loss": 0.21741489410400391,
+      "step": 9275
+    },
+    {
+      "epoch": 2.1592755978639424,
+      "grad_norm": 0.5903428196907043,
+      "learning_rate": 0.00018433720724220597,
+      "loss": 0.20771631240844726,
+      "step": 9300
+    },
+    {
+      "epoch": 2.1650801021592754,
+      "grad_norm": 0.5719689130783081,
+      "learning_rate": 0.0001837994484342264,
+      "loss": 0.22163301467895508,
+      "step": 9325
+    },
+    {
+      "epoch": 2.170884606454609,
+      "grad_norm": 0.6366750597953796,
+      "learning_rate": 0.00018326123158605165,
+      "loss": 0.23154041290283203,
+      "step": 9350
+    },
+    {
+      "epoch": 2.1766891107499418,
+      "grad_norm": 0.552815854549408,
+      "learning_rate": 0.00018272256399143834,
+      "loss": 0.21899927139282227,
+      "step": 9375
+    },
+    {
+      "epoch": 2.182493615045275,
+      "grad_norm": 0.5769531726837158,
+      "learning_rate": 0.00018218345295025143,
+      "loss": 0.21298625946044922,
+      "step": 9400
+    },
+    {
+      "epoch": 2.188298119340608,
+      "grad_norm": 0.5878853797912598,
+      "learning_rate": 0.00018164390576836534,
+      "loss": 0.22514759063720702,
+      "step": 9425
+    },
+    {
+      "epoch": 2.1941026236359416,
+      "grad_norm": 0.480671763420105,
+      "learning_rate": 0.00018110392975756502,
+      "loss": 0.2228526496887207,
+      "step": 9450
+    },
+    {
+      "epoch": 2.1999071279312745,
+      "grad_norm": 0.5636632442474365,
+      "learning_rate": 0.00018056353223544665,
+      "loss": 0.21168306350708008,
+      "step": 9475
+    },
+    {
+      "epoch": 2.205711632226608,
+      "grad_norm": 0.48353374004364014,
+      "learning_rate": 0.00018002272052531878,
+      "loss": 0.2263408660888672,
+      "step": 9500
+    },
+    {
+      "epoch": 2.211516136521941,
+      "grad_norm": 0.5352545380592346,
+      "learning_rate": 0.0001794815019561027,
+      "loss": 0.22426290512084962,
+      "step": 9525
+    },
+    {
+      "epoch": 2.2173206408172743,
+      "grad_norm": 0.5699704885482788,
+      "learning_rate": 0.0001789398838622336,
+      "loss": 0.21549053192138673,
+      "step": 9550
+    },
+    {
+      "epoch": 2.2231251451126073,
+      "grad_norm": 0.5913279056549072,
+      "learning_rate": 0.00017839787358356067,
+      "loss": 0.22139385223388672,
+      "step": 9575
+    },
+    {
+      "epoch": 2.2289296494079407,
+      "grad_norm": 0.5757904648780823,
+      "learning_rate": 0.00017785547846524798,
+      "loss": 0.2186814308166504,
+      "step": 9600
+    },
+    {
+      "epoch": 2.2347341537032737,
+      "grad_norm": 0.5990129113197327,
+      "learning_rate": 0.0001773127058576748,
+      "loss": 0.2097479248046875,
+      "step": 9625
+    },
+    {
+      "epoch": 2.240538657998607,
+      "grad_norm": 0.5633386969566345,
+      "learning_rate": 0.00017676956311633613,
+      "loss": 0.22741294860839845,
+      "step": 9650
+    },
+    {
+      "epoch": 2.24634316229394,
+      "grad_norm": 0.6134929656982422,
+      "learning_rate": 0.00017622605760174271,
+      "loss": 0.21768487930297853,
+      "step": 9675
+    },
+    {
+      "epoch": 2.2521476665892735,
+      "grad_norm": 0.5604804754257202,
+      "learning_rate": 0.00017568219667932176,
+      "loss": 0.2174857711791992,
+      "step": 9700
+    },
+    {
+      "epoch": 2.2579521708846064,
+      "grad_norm": 0.5374922752380371,
+      "learning_rate": 0.00017513798771931653,
+      "loss": 0.21669696807861327,
+      "step": 9725
+    },
+    {
+      "epoch": 2.2637566751799394,
+      "grad_norm": 0.5428775548934937,
+      "learning_rate": 0.00017459343809668712,
+      "loss": 0.20766374588012695,
+      "step": 9750
+    },
+    {
+      "epoch": 2.269561179475273,
+      "grad_norm": 0.5475351810455322,
+      "learning_rate": 0.00017404855519100994,
+      "loss": 0.20494213104248046,
+      "step": 9775
+    },
+    {
+      "epoch": 2.275365683770606,
+      "grad_norm": 0.5451847910881042,
+      "learning_rate": 0.00017350334638637809,
+      "loss": 0.2123589324951172,
+      "step": 9800
+    },
+    {
+      "epoch": 2.281170188065939,
+      "grad_norm": 0.614493727684021,
+      "learning_rate": 0.00017295781907130108,
+      "loss": 0.23357988357543946,
+      "step": 9825
+    },
+    {
+      "epoch": 2.286974692361272,
+      "grad_norm": 0.6001668572425842,
+      "learning_rate": 0.00017241198063860485,
+      "loss": 0.22932859420776366,
+      "step": 9850
+    },
+    {
+      "epoch": 2.2927791966566056,
+      "grad_norm": 0.6225593686103821,
+      "learning_rate": 0.0001718658384853314,
+      "loss": 0.228519287109375,
+      "step": 9875
+    },
+    {
+      "epoch": 2.2985837009519385,
+      "grad_norm": 0.5498306751251221,
+      "learning_rate": 0.00017131940001263883,
+      "loss": 0.22619125366210938,
+      "step": 9900
+    },
+    {
+      "epoch": 2.304388205247272,
+      "grad_norm": 0.617794394493103,
+      "learning_rate": 0.00017077267262570077,
+      "loss": 0.24700286865234375,
+      "step": 9925
+    },
+    {
+      "epoch": 2.310192709542605,
+      "grad_norm": 0.5148032307624817,
+      "learning_rate": 0.0001702256637336061,
+      "loss": 0.21693771362304687,
+      "step": 9950
+    },
+    {
+      "epoch": 2.3159972138379383,
+      "grad_norm": 0.5845295190811157,
+      "learning_rate": 0.00016967838074925866,
+      "loss": 0.2280506134033203,
+      "step": 9975
+    },
+    {
+      "epoch": 2.3218017181332713,
+      "grad_norm": 0.6682146191596985,
+      "learning_rate": 0.0001691308310892766,
+      "loss": 0.2197197914123535,
+      "step": 10000
+    },
+    {
+      "epoch": 2.3218017181332713,
+      "eval_loss": 0.467913955450058,
+      "eval_runtime": 2.8534,
+      "eval_samples_per_second": 630.819,
+      "eval_steps_per_second": 10.163,
+      "step": 10000
+    },
+    {
+      "epoch": 2.3276062224286047,
+      "grad_norm": 0.5486340522766113,
+      "learning_rate": 0.00016858302217389213,
+      "loss": 0.21343572616577147,
+      "step": 10025
+    },
+    {
+      "epoch": 2.3334107267239377,
+      "grad_norm": 0.49393242597579956,
+      "learning_rate": 0.00016803496142685066,
+      "loss": 0.22367969512939453,
+      "step": 10050
+    },
+    {
+      "epoch": 2.339215231019271,
+      "grad_norm": 0.6150463819503784,
+      "learning_rate": 0.00016748665627531052,
+      "loss": 0.21864049911499023,
+      "step": 10075
+    },
+    {
+      "epoch": 2.345019735314604,
+      "grad_norm": 0.5299875736236572,
+      "learning_rate": 0.00016693811414974187,
+      "loss": 0.22662849426269532,
+      "step": 10100
+    },
+    {
+      "epoch": 2.3508242396099375,
+      "grad_norm": 0.5817348957061768,
+      "learning_rate": 0.00016638934248382664,
+      "loss": 0.22811294555664063,
+      "step": 10125
+    },
+    {
+      "epoch": 2.3566287439052704,
+      "grad_norm": 0.5914196372032166,
+      "learning_rate": 0.00016584034871435703,
+      "loss": 0.21525127410888673,
+      "step": 10150
+    },
+    {
+      "epoch": 2.362433248200604,
+      "grad_norm": 0.5707176923751831,
+      "learning_rate": 0.00016529114028113545,
+      "loss": 0.23242399215698242,
+      "step": 10175
+    },
+    {
+      "epoch": 2.368237752495937,
+      "grad_norm": 0.5395120978355408,
+      "learning_rate": 0.00016474172462687317,
+      "loss": 0.22955944061279296,
+      "step": 10200
+    },
+    {
+      "epoch": 2.3740422567912702,
+      "grad_norm": 0.6231363415718079,
+      "learning_rate": 0.00016419210919708973,
+      "loss": 0.22565114974975586,
+      "step": 10225
+    },
+    {
+      "epoch": 2.379846761086603,
+      "grad_norm": 0.5674353837966919,
+      "learning_rate": 0.000163642301440012,
+      "loss": 0.2098895263671875,
+      "step": 10250
+    },
+    {
+      "epoch": 2.385651265381936,
+      "grad_norm": 0.45566293597221375,
+      "learning_rate": 0.00016309230880647314,
+      "loss": 0.21583791732788085,
+      "step": 10275
+    },
+    {
+      "epoch": 2.3914557696772696,
+      "grad_norm": 0.5758122205734253,
+      "learning_rate": 0.00016254213874981175,
+      "loss": 0.22361513137817382,
+      "step": 10300
+    },
+    {
+      "epoch": 2.3972602739726026,
+      "grad_norm": 0.4571880102157593,
+      "learning_rate": 0.00016199179872577077,
+      "loss": 0.21916532516479492,
+      "step": 10325
+    },
+    {
+      "epoch": 2.403064778267936,
+      "grad_norm": 0.5546058416366577,
+      "learning_rate": 0.00016144129619239658,
+      "loss": 0.2137105369567871,
+      "step": 10350
+    },
+    {
+      "epoch": 2.408869282563269,
+      "grad_norm": 0.5631943345069885,
+      "learning_rate": 0.00016089063860993776,
+      "loss": 0.21274251937866212,
+      "step": 10375
+    },
+    {
+      "epoch": 2.4146737868586023,
+      "grad_norm": 0.5407511591911316,
+      "learning_rate": 0.000160339833440744,
+      "loss": 0.20957944869995118,
+      "step": 10400
+    },
+    {
+      "epoch": 2.4204782911539353,
+      "grad_norm": 0.4707408845424652,
+      "learning_rate": 0.00015978888814916528,
+      "loss": 0.21269603729248046,
+      "step": 10425
+    },
+    {
+      "epoch": 2.4262827954492687,
+      "grad_norm": 0.48145392537117004,
+      "learning_rate": 0.0001592378102014502,
+      "loss": 0.2084123420715332,
+      "step": 10450
+    },
+    {
+      "epoch": 2.4320872997446017,
+      "grad_norm": 0.5924553275108337,
+      "learning_rate": 0.0001586866070656453,
+      "loss": 0.2128961944580078,
+      "step": 10475
+    },
+    {
+      "epoch": 2.437891804039935,
+      "grad_norm": 0.572172999382019,
+      "learning_rate": 0.0001581352862114935,
+      "loss": 0.22029634475708007,
+      "step": 10500
+    },
+    {
+      "epoch": 2.443696308335268,
+      "grad_norm": 0.5437218546867371,
+      "learning_rate": 0.000157583855110333,
+      "loss": 0.21352359771728516,
+      "step": 10525
+    },
+    {
+      "epoch": 2.4495008126306015,
+      "grad_norm": 0.5065390467643738,
+      "learning_rate": 0.00015703232123499613,
+      "loss": 0.22642688751220702,
+      "step": 10550
+    },
+    {
+      "epoch": 2.4553053169259345,
+      "grad_norm": 0.6452273726463318,
+      "learning_rate": 0.00015648069205970795,
+      "loss": 0.21440177917480469,
+      "step": 10575
+    },
+    {
+      "epoch": 2.461109821221268,
+      "grad_norm": 0.5769652724266052,
+      "learning_rate": 0.0001559289750599849,
+      "loss": 0.22602521896362304,
+      "step": 10600
+    },
+    {
+      "epoch": 2.466914325516601,
+      "grad_norm": 0.47669705748558044,
+      "learning_rate": 0.0001553771777125338,
+      "loss": 0.22494300842285156,
+      "step": 10625
+    },
+    {
+      "epoch": 2.4727188298119342,
+      "grad_norm": 0.5278451442718506,
+      "learning_rate": 0.00015482530749514998,
+      "loss": 0.22209081649780274,
+      "step": 10650
+    },
+    {
+      "epoch": 2.478523334107267,
+      "grad_norm": 0.5236882567405701,
+      "learning_rate": 0.00015427337188661676,
+      "loss": 0.22514835357666016,
+      "step": 10675
+    },
+    {
+      "epoch": 2.4843278384026,
+      "grad_norm": 0.5578672289848328,
+      "learning_rate": 0.0001537213783666031,
+      "loss": 0.2285551071166992,
+      "step": 10700
+    },
+    {
+      "epoch": 2.4901323426979336,
+      "grad_norm": 0.6060502529144287,
+      "learning_rate": 0.0001531693344155632,
+      "loss": 0.20898460388183593,
+      "step": 10725
+    },
+    {
+      "epoch": 2.495936846993267,
+      "grad_norm": 0.5567417144775391,
+      "learning_rate": 0.0001526172475146344,
+      "loss": 0.21838787078857422,
+      "step": 10750
+    },
+    {
+      "epoch": 2.5017413512886,
+      "grad_norm": 0.5635561347007751,
+      "learning_rate": 0.00015206512514553622,
+      "loss": 0.2143534851074219,
+      "step": 10775
+    },
+    {
+      "epoch": 2.507545855583933,
+      "grad_norm": 0.5605906248092651,
+      "learning_rate": 0.00015151297479046874,
+      "loss": 0.22134967803955077,
+      "step": 10800
+    },
+    {
+      "epoch": 2.5133503598792664,
+      "grad_norm": 0.49493154883384705,
+      "learning_rate": 0.00015096080393201143,
+      "loss": 0.20288610458374023,
+      "step": 10825
+    },
+    {
+      "epoch": 2.5191548641745993,
+      "grad_norm": 0.5453201532363892,
+      "learning_rate": 0.00015040862005302153,
+      "loss": 0.22211837768554688,
+      "step": 10850
+    },
+    {
+      "epoch": 2.5249593684699327,
+      "grad_norm": 0.6014910340309143,
+      "learning_rate": 0.0001498564306365327,
+      "loss": 0.22515666961669922,
+      "step": 10875
+    },
+    {
+      "epoch": 2.5307638727652657,
+      "grad_norm": 0.5902166962623596,
+      "learning_rate": 0.0001493042431656537,
+      "loss": 0.22335886001586913,
+      "step": 10900
+    },
+    {
+      "epoch": 2.536568377060599,
+      "grad_norm": 0.4785741865634918,
+      "learning_rate": 0.00014875206512346694,
+      "loss": 0.20825664520263673,
+      "step": 10925
+    },
+    {
+      "epoch": 2.542372881355932,
+      "grad_norm": 0.5128177404403687,
+      "learning_rate": 0.00014819990399292696,
+      "loss": 0.22325387954711914,
+      "step": 10950
+    },
+    {
+      "epoch": 2.5481773856512655,
+      "grad_norm": 0.6161273121833801,
+      "learning_rate": 0.00014764776725675918,
+      "loss": 0.21668289184570313,
+      "step": 10975
+    },
+    {
+      "epoch": 2.5539818899465985,
+      "grad_norm": 0.5592746138572693,
+      "learning_rate": 0.00014709566239735854,
+      "loss": 0.22518924713134766,
+      "step": 11000
+    },
+    {
+      "epoch": 2.5539818899465985,
+      "eval_loss": 0.46551966667175293,
+      "eval_runtime": 2.769,
+      "eval_samples_per_second": 650.047,
+      "eval_steps_per_second": 10.473,
+      "step": 11000
+    },
+    {
+      "epoch": 2.559786394241932,
+      "grad_norm": 0.5250012278556824,
+      "learning_rate": 0.0001465435968966878,
+      "loss": 0.22225908279418946,
+      "step": 11025
+    },
+    {
+      "epoch": 2.565590898537265,
+      "grad_norm": 0.548968493938446,
+      "learning_rate": 0.0001459915782361765,
+      "loss": 0.22030595779418946,
+      "step": 11050
+    },
+    {
+      "epoch": 2.5713954028325983,
+      "grad_norm": 0.5696165561676025,
+      "learning_rate": 0.00014543961389661917,
+      "loss": 0.2092125129699707,
+      "step": 11075
+    },
+    {
+      "epoch": 2.5771999071279312,
+      "grad_norm": 0.5397639870643616,
+      "learning_rate": 0.00014488771135807462,
+      "loss": 0.21414701461791993,
+      "step": 11100
+    },
+    {
+      "epoch": 2.583004411423264,
+      "grad_norm": 0.5502409934997559,
+      "learning_rate": 0.00014433587809976377,
+      "loss": 0.2023804473876953,
+      "step": 11125
+    },
+    {
+      "epoch": 2.5888089157185976,
+      "grad_norm": 0.5535282492637634,
+      "learning_rate": 0.00014378412159996885,
+      "loss": 0.2260698890686035,
+      "step": 11150
+    },
+    {
+      "epoch": 2.594613420013931,
+      "grad_norm": 0.5783640146255493,
+      "learning_rate": 0.00014323244933593185,
+      "loss": 0.21625955581665038,
+      "step": 11175
+    },
+    {
+      "epoch": 2.600417924309264,
+      "grad_norm": 0.5408916473388672,
+      "learning_rate": 0.00014268086878375315,
+      "loss": 0.21140981674194337,
+      "step": 11200
+    },
+    {
+      "epoch": 2.606222428604597,
+      "grad_norm": 0.4965577721595764,
+      "learning_rate": 0.00014212938741829044,
+      "loss": 0.20270973205566406,
+      "step": 11225
+    },
+    {
+      "epoch": 2.6120269328999304,
+      "grad_norm": 0.6295332312583923,
+      "learning_rate": 0.0001415780127130571,
+      "loss": 0.2084318733215332,
+      "step": 11250
+    },
+    {
+      "epoch": 2.617831437195264,
+      "grad_norm": 0.515028715133667,
+      "learning_rate": 0.0001410267521401212,
+      "loss": 0.20873857498168946,
+      "step": 11275
+    },
+    {
+      "epoch": 2.6236359414905968,
+      "grad_norm": 0.6074643731117249,
+      "learning_rate": 0.00014047561317000393,
+      "loss": 0.2222311782836914,
+      "step": 11300
+    },
+    {
+      "epoch": 2.6294404457859297,
+      "grad_norm": 0.5213595628738403,
+      "learning_rate": 0.00013992460327157886,
+      "loss": 0.2053067970275879,
+      "step": 11325
+    },
+    {
+      "epoch": 2.635244950081263,
+      "grad_norm": 0.5365227460861206,
+      "learning_rate": 0.00013937372991197023,
+      "loss": 0.204686279296875,
+      "step": 11350
+    },
+    {
+      "epoch": 2.641049454376596,
+      "grad_norm": 0.5262821912765503,
+      "learning_rate": 0.00013882300055645188,
+      "loss": 0.21241090774536134,
+      "step": 11375
+    },
+    {
+      "epoch": 2.6468539586719295,
+      "grad_norm": 0.5562365651130676,
+      "learning_rate": 0.00013827242266834625,
+      "loss": 0.22387901306152344,
+      "step": 11400
+    },
+    {
+      "epoch": 2.6526584629672625,
+      "grad_norm": 0.5751397013664246,
+      "learning_rate": 0.00013772200370892323,
+      "loss": 0.2154745101928711,
+      "step": 11425
+    },
+    {
+      "epoch": 2.658462967262596,
+      "grad_norm": 0.5250651836395264,
+      "learning_rate": 0.00013717175113729874,
+      "loss": 0.21518571853637694,
+      "step": 11450
+    },
+    {
+      "epoch": 2.664267471557929,
+      "grad_norm": 0.5995931625366211,
+      "learning_rate": 0.000136621672410334,
+      "loss": 0.21333885192871094,
+      "step": 11475
+    },
+    {
+      "epoch": 2.6700719758532623,
+      "grad_norm": 0.5499006509780884,
+      "learning_rate": 0.0001360717749825343,
+      "loss": 0.21046367645263672,
+      "step": 11500
+    },
+    {
+      "epoch": 2.6758764801485952,
+      "grad_norm": 0.518254280090332,
+      "learning_rate": 0.00013552206630594807,
+      "loss": 0.21456647872924806,
+      "step": 11525
+    },
+    {
+      "epoch": 2.681680984443928,
+      "grad_norm": 0.676092267036438,
+      "learning_rate": 0.0001349725538300658,
+      "loss": 0.222899169921875,
+      "step": 11550
+    },
+    {
+      "epoch": 2.6874854887392616,
+      "grad_norm": 0.5551995038986206,
+      "learning_rate": 0.00013442324500171905,
+      "loss": 0.21177555084228517,
+      "step": 11575
+    },
+    {
+      "epoch": 2.693289993034595,
+      "grad_norm": 0.6442759037017822,
+      "learning_rate": 0.0001338741472649797,
+      "loss": 0.20636798858642577,
+      "step": 11600
+    },
+    {
+      "epoch": 2.699094497329928,
+      "grad_norm": 0.5299642086029053,
+      "learning_rate": 0.000133325268061059,
+      "loss": 0.22177127838134766,
+      "step": 11625
+    },
+    {
+      "epoch": 2.704899001625261,
+      "grad_norm": 0.5842543244361877,
+      "learning_rate": 0.00013277661482820666,
+      "loss": 0.20731349945068359,
+      "step": 11650
+    },
+    {
+      "epoch": 2.7107035059205944,
+      "grad_norm": 0.5660545825958252,
+      "learning_rate": 0.00013222819500161007,
+      "loss": 0.1958559799194336,
+      "step": 11675
+    },
+    {
+      "epoch": 2.716508010215928,
+      "grad_norm": 0.5219541788101196,
+      "learning_rate": 0.00013168001601329357,
+      "loss": 0.1985594940185547,
+      "step": 11700
+    },
+    {
+      "epoch": 2.7223125145112608,
+      "grad_norm": 0.6481348872184753,
+      "learning_rate": 0.00013113208529201786,
+      "loss": 0.20331716537475586,
+      "step": 11725
+    },
+    {
+      "epoch": 2.7281170188065937,
+      "grad_norm": 0.5476353764533997,
+      "learning_rate": 0.00013058441026317903,
+      "loss": 0.2138274574279785,
+      "step": 11750
+    },
+    {
+      "epoch": 2.733921523101927,
+      "grad_norm": 0.5899817943572998,
+      "learning_rate": 0.0001300369983487082,
+      "loss": 0.21353500366210937,
+      "step": 11775
+    },
+    {
+      "epoch": 2.73972602739726,
+      "grad_norm": 0.5754044651985168,
+      "learning_rate": 0.00012948985696697066,
+      "loss": 0.2188318634033203,
+      "step": 11800
+    },
+    {
+      "epoch": 2.7455305316925935,
+      "grad_norm": 0.5576942563056946,
+      "learning_rate": 0.00012894299353266591,
+      "loss": 0.19502559661865235,
+      "step": 11825
+    },
+    {
+      "epoch": 2.7513350359879265,
+      "grad_norm": 0.5010995268821716,
+      "learning_rate": 0.00012839641545672644,
+      "loss": 0.21735916137695313,
+      "step": 11850
+    },
+    {
+      "epoch": 2.75713954028326,
+      "grad_norm": 0.5585890412330627,
+      "learning_rate": 0.00012785013014621774,
+      "loss": 0.20576717376708983,
+      "step": 11875
+    },
+    {
+      "epoch": 2.762944044578593,
+      "grad_norm": 0.5613226890563965,
+      "learning_rate": 0.00012730414500423782,
+      "loss": 0.20334775924682616,
+      "step": 11900
+    },
+    {
+      "epoch": 2.7687485488739263,
+      "grad_norm": 0.5809953808784485,
+      "learning_rate": 0.00012675846742981697,
+      "loss": 0.21360109329223634,
+      "step": 11925
+    },
+    {
+      "epoch": 2.7745530531692593,
+      "grad_norm": 0.6128183603286743,
+      "learning_rate": 0.0001262131048178174,
+      "loss": 0.19883115768432616,
+      "step": 11950
+    },
+    {
+      "epoch": 2.7803575574645927,
+      "grad_norm": 0.5348827242851257,
+      "learning_rate": 0.0001256680645588329,
+      "loss": 0.19767290115356445,
+      "step": 11975
+    },
+    {
+      "epoch": 2.7861620617599256,
+      "grad_norm": 0.6239649653434753,
+      "learning_rate": 0.00012512335403908894,
+      "loss": 0.2188077163696289,
+      "step": 12000
+    },
+    {
+      "epoch": 2.7861620617599256,
+      "eval_loss": 0.45799410343170166,
+      "eval_runtime": 2.7524,
+      "eval_samples_per_second": 653.973,
+      "eval_steps_per_second": 10.536,
+      "step": 12000
+    },
+    {
+      "epoch": 2.791966566055259,
+      "grad_norm": 0.4721246361732483,
+      "learning_rate": 0.00012457898064034258,
+      "loss": 0.21079427719116212,
+      "step": 12025
+    },
+    {
+      "epoch": 2.797771070350592,
+      "grad_norm": 0.5537365674972534,
+      "learning_rate": 0.00012403495173978206,
+      "loss": 0.21205425262451172,
+      "step": 12050
+    },
+    {
+      "epoch": 2.803575574645925,
+      "grad_norm": 0.44193127751350403,
+      "learning_rate": 0.0001234912747099273,
+      "loss": 0.20757282257080079,
+      "step": 12075
+    },
+    {
+      "epoch": 2.8093800789412584,
+      "grad_norm": 0.5894005298614502,
+      "learning_rate": 0.00012294795691852954,
+      "loss": 0.20790157318115235,
+      "step": 12100
+    },
+    {
+      "epoch": 2.815184583236592,
+      "grad_norm": 0.6244122385978699,
+      "learning_rate": 0.00012240500572847202,
+      "loss": 0.21242237091064453,
+      "step": 12125
+    },
+    {
+      "epoch": 2.820989087531925,
+      "grad_norm": 0.5062934160232544,
+      "learning_rate": 0.0001218624284976697,
+      "loss": 0.21416374206542968,
+      "step": 12150
+    },
+    {
+      "epoch": 2.8267935918272578,
+      "grad_norm": 0.4934367835521698,
+      "learning_rate": 0.00012132023257896978,
+      "loss": 0.198531551361084,
+      "step": 12175
+    },
+    {
+      "epoch": 2.832598096122591,
+      "grad_norm": 0.5566888451576233,
+      "learning_rate": 0.00012077842532005205,
+      "loss": 0.19445446014404297,
+      "step": 12200
+    },
+    {
+      "epoch": 2.8384026004179246,
+      "grad_norm": 0.5387080907821655,
+      "learning_rate": 0.00012023701406332935,
+      "loss": 0.2107783317565918,
+      "step": 12225
+    },
+    {
+      "epoch": 2.8442071047132576,
+      "grad_norm": 0.5526768565177917,
+      "learning_rate": 0.000119696006145848,
+      "loss": 0.20592342376708983,
+      "step": 12250
+    },
+    {
+      "epoch": 2.8500116090085905,
+      "grad_norm": 0.5601632595062256,
+      "learning_rate": 0.00011915540889918827,
+      "loss": 0.20973558425903321,
+      "step": 12275
+    },
+    {
+      "epoch": 2.855816113303924,
+      "grad_norm": 0.5619735717773438,
+      "learning_rate": 0.0001186152296493653,
+      "loss": 0.20025985717773437,
+      "step": 12300
+    },
+    {
+      "epoch": 2.861620617599257,
+      "grad_norm": 0.6121880412101746,
+      "learning_rate": 0.00011807547571672945,
+      "loss": 0.2027290916442871,
+      "step": 12325
+    },
+    {
+      "epoch": 2.8674251218945903,
+      "grad_norm": 0.5044817924499512,
+      "learning_rate": 0.0001175361544158676,
+      "loss": 0.20484689712524415,
+      "step": 12350
+    },
+    {
+      "epoch": 2.8732296261899233,
+      "grad_norm": 0.5082159042358398,
+      "learning_rate": 0.00011699727305550352,
+      "loss": 0.22294116973876954,
+      "step": 12375
+    },
+    {
+      "epoch": 2.8790341304852567,
+      "grad_norm": 0.5348130464553833,
+      "learning_rate": 0.00011645883893839902,
+      "loss": 0.18227848052978515,
+      "step": 12400
+    },
+    {
+      "epoch": 2.8848386347805897,
+      "grad_norm": 0.5175385475158691,
+      "learning_rate": 0.00011592085936125512,
+      "loss": 0.21235498428344726,
+      "step": 12425
+    },
+    {
+      "epoch": 2.890643139075923,
+      "grad_norm": 0.5243074297904968,
+      "learning_rate": 0.00011538334161461304,
+      "loss": 0.1997675895690918,
+      "step": 12450
+    },
+    {
+      "epoch": 2.896447643371256,
+      "grad_norm": 0.5306870937347412,
+      "learning_rate": 0.00011484629298275529,
+      "loss": 0.1953229331970215,
+      "step": 12475
+    },
+    {
+      "epoch": 2.902252147666589,
+      "grad_norm": 0.5092787146568298,
+      "learning_rate": 0.00011430972074360722,
+      "loss": 0.21192399978637697,
+      "step": 12500
+    },
+    {
+      "epoch": 2.9080566519619224,
+      "grad_norm": 0.5617599487304688,
+      "learning_rate": 0.00011377363216863806,
+      "loss": 0.1928584671020508,
+      "step": 12525
+    },
+    {
+      "epoch": 2.913861156257256,
+      "grad_norm": 0.5036365389823914,
+      "learning_rate": 0.00011323803452276277,
+      "loss": 0.21215980529785156,
+      "step": 12550
+    },
+    {
+      "epoch": 2.919665660552589,
+      "grad_norm": 0.5223519206047058,
+      "learning_rate": 0.00011270293506424331,
+      "loss": 0.2054371452331543,
+      "step": 12575
+    },
+    {
+      "epoch": 2.9254701648479218,
+      "grad_norm": 0.5945123434066772,
+      "learning_rate": 0.00011216834104459029,
+      "loss": 0.21232702255249022,
+      "step": 12600
+    },
+    {
+      "epoch": 2.931274669143255,
+      "grad_norm": 0.49228933453559875,
+      "learning_rate": 0.00011163425970846478,
+      "loss": 0.19937749862670898,
+      "step": 12625
+    },
+    {
+      "epoch": 2.9370791734385886,
+      "grad_norm": 0.5792744755744934,
+      "learning_rate": 0.00011110069829358018,
+      "loss": 0.19283594131469728,
+      "step": 12650
+    },
+    {
+      "epoch": 2.9428836777339216,
+      "grad_norm": 0.5588549375534058,
+      "learning_rate": 0.00011056766403060406,
+      "loss": 0.20069318771362304,
+      "step": 12675
+    },
+    {
+      "epoch": 2.9486881820292545,
+      "grad_norm": 0.4905478060245514,
+      "learning_rate": 0.00011003516414306009,
+      "loss": 0.20762886047363283,
+      "step": 12700
+    },
+    {
+      "epoch": 2.954492686324588,
+      "grad_norm": 0.6094077825546265,
+      "learning_rate": 0.00010950320584723029,
+      "loss": 0.20677431106567382,
+      "step": 12725
+    },
+    {
+      "epoch": 2.960297190619921,
+      "grad_norm": 0.5890378952026367,
+      "learning_rate": 0.00010897179635205728,
+      "loss": 0.20378313064575196,
+      "step": 12750
+    },
+    {
+      "epoch": 2.9661016949152543,
+      "grad_norm": 0.47871914505958557,
+      "learning_rate": 0.00010844094285904637,
+      "loss": 0.20004503250122072,
+      "step": 12775
+    },
+    {
+      "epoch": 2.9719061992105873,
+      "grad_norm": 0.6468496322631836,
+      "learning_rate": 0.00010791065256216815,
+      "loss": 0.21985828399658203,
+      "step": 12800
+    },
+    {
+      "epoch": 2.9777107035059207,
+      "grad_norm": 0.5449540019035339,
+      "learning_rate": 0.00010738093264776084,
+      "loss": 0.1990898895263672,
+      "step": 12825
+    },
+    {
+      "epoch": 2.9835152078012537,
+      "grad_norm": 0.516776442527771,
+      "learning_rate": 0.00010685179029443322,
+      "loss": 0.21026317596435548,
+      "step": 12850
+    },
+    {
+      "epoch": 2.989319712096587,
+      "grad_norm": 0.510644793510437,
+      "learning_rate": 0.00010632323267296699,
+      "loss": 0.21136085510253907,
+      "step": 12875
+    },
+    {
+      "epoch": 2.99512421639192,
+      "grad_norm": 0.4976428151130676,
+      "learning_rate": 0.00010579526694621966,
+      "loss": 0.18616481781005859,
+      "step": 12900
+    },
+    {
+      "epoch": 3.0009287206872535,
+      "grad_norm": 0.4483867287635803,
+      "learning_rate": 0.00010526790026902768,
+      "loss": 0.19157962799072265,
+      "step": 12925
+    },
+    {
+      "epoch": 3.0067332249825864,
+      "grad_norm": 0.5449016690254211,
+      "learning_rate": 0.00010474113978810941,
+      "loss": 0.13019381523132323,
+      "step": 12950
+    },
+    {
+      "epoch": 3.01253772927792,
+      "grad_norm": 0.4980277717113495,
+      "learning_rate": 0.00010421499264196801,
+      "loss": 0.1428825855255127,
+      "step": 12975
+    },
+    {
+      "epoch": 3.018342233573253,
+      "grad_norm": 0.5219682455062866,
+      "learning_rate": 0.0001036894659607951,
+      "loss": 0.137953519821167,
+      "step": 13000
+    },
+    {
+      "epoch": 3.018342233573253,
+      "eval_loss": 0.477819561958313,
+      "eval_runtime": 2.7724,
+      "eval_samples_per_second": 649.249,
+      "eval_steps_per_second": 10.46,
+      "step": 13000
+    },
+    {
+      "epoch": 3.0241467378685862,
+      "grad_norm": 0.49482089281082153,
+      "learning_rate": 0.00010316456686637382,
+      "loss": 0.1414840602874756,
+      "step": 13025
+    },
+    {
+      "epoch": 3.029951242163919,
+      "grad_norm": 0.5301785469055176,
+      "learning_rate": 0.00010264030247198257,
+      "loss": 0.14146761894226073,
+      "step": 13050
+    },
+    {
+      "epoch": 3.035755746459252,
+      "grad_norm": 0.4624095857143402,
+      "learning_rate": 0.0001021166798822983,
+      "loss": 0.13560797691345214,
+      "step": 13075
+    },
+    {
+      "epoch": 3.0415602507545856,
+      "grad_norm": 0.536002516746521,
+      "learning_rate": 0.00010159370619330059,
+      "loss": 0.13354150772094728,
+      "step": 13100
+    },
+    {
+      "epoch": 3.0473647550499185,
+      "grad_norm": 0.6252378821372986,
+      "learning_rate": 0.00010107138849217514,
+      "loss": 0.13856441497802735,
+      "step": 13125
+    },
+    {
+      "epoch": 3.053169259345252,
+      "grad_norm": 0.5481756329536438,
+      "learning_rate": 0.00010054973385721807,
+      "loss": 0.13322070121765137,
+      "step": 13150
+    },
+    {
+      "epoch": 3.058973763640585,
+      "grad_norm": 0.41996362805366516,
+      "learning_rate": 0.00010002874935773972,
+      "loss": 0.1397541046142578,
+      "step": 13175
+    },
+    {
+      "epoch": 3.0647782679359183,
+      "grad_norm": 0.4922993779182434,
+      "learning_rate": 9.95084420539689e-05,
+      "loss": 0.1373126697540283,
+      "step": 13200
+    },
+    {
+      "epoch": 3.0705827722312513,
+      "grad_norm": 0.5481604337692261,
+      "learning_rate": 9.898881899695735e-05,
+      "loss": 0.1354703998565674,
+      "step": 13225
+    },
+    {
+      "epoch": 3.0763872765265847,
+      "grad_norm": 0.38447105884552,
+      "learning_rate": 9.846988722848418e-05,
+      "loss": 0.13499070167541505,
+      "step": 13250
+    },
+    {
+      "epoch": 3.0821917808219177,
+      "grad_norm": 0.5143475532531738,
+      "learning_rate": 9.795165378096014e-05,
+      "loss": 0.13656951904296874,
+      "step": 13275
+    },
+    {
+      "epoch": 3.087996285117251,
+      "grad_norm": 0.48031729459762573,
+      "learning_rate": 9.743412567733272e-05,
+      "loss": 0.14267797470092775,
+      "step": 13300
+    },
+    {
+      "epoch": 3.093800789412584,
+      "grad_norm": 0.5093261003494263,
+      "learning_rate": 9.691730993099066e-05,
+      "loss": 0.14414706230163574,
+      "step": 13325
+    },
+    {
+      "epoch": 3.0996052937079175,
+      "grad_norm": 0.4485715329647064,
+      "learning_rate": 9.64012135456692e-05,
+      "loss": 0.12862277030944824,
+      "step": 13350
+    },
+    {
+      "epoch": 3.1054097980032505,
+      "grad_norm": 0.5383508801460266,
+      "learning_rate": 9.58858435153549e-05,
+      "loss": 0.13804192543029786,
+      "step": 13375
+    },
+    {
+      "epoch": 3.111214302298584,
+      "grad_norm": 0.5513857007026672,
+      "learning_rate": 9.537120682419104e-05,
+      "loss": 0.15050209045410157,
+      "step": 13400
+    },
+    {
+      "epoch": 3.117018806593917,
+      "grad_norm": 0.45335784554481506,
+      "learning_rate": 9.485731044638278e-05,
+      "loss": 0.13252761840820312,
+      "step": 13425
+    },
+    {
+      "epoch": 3.1228233108892502,
+      "grad_norm": 0.508185625076294,
+      "learning_rate": 9.43441613461029e-05,
+      "loss": 0.13793312072753905,
+      "step": 13450
+    },
+    {
+      "epoch": 3.128627815184583,
+      "grad_norm": 0.4407822787761688,
+      "learning_rate": 9.383176647739727e-05,
+      "loss": 0.1388423538208008,
+      "step": 13475
+    },
+    {
+      "epoch": 3.1344323194799166,
+      "grad_norm": 0.4198641777038574,
+      "learning_rate": 9.332013278409056e-05,
+      "loss": 0.12962688446044923,
+      "step": 13500
+    },
+    {
+      "epoch": 3.1402368237752496,
+      "grad_norm": 0.5023673176765442,
+      "learning_rate": 9.280926719969234e-05,
+      "loss": 0.13853322029113768,
+      "step": 13525
+    },
+    {
+      "epoch": 3.1460413280705826,
+      "grad_norm": 0.44358789920806885,
+      "learning_rate": 9.229917664730283e-05,
+      "loss": 0.1367900276184082,
+      "step": 13550
+    },
+    {
+      "epoch": 3.151845832365916,
+      "grad_norm": 0.5299422144889832,
+      "learning_rate": 9.17898680395194e-05,
+      "loss": 0.15225461959838868,
+      "step": 13575
+    },
+    {
+      "epoch": 3.157650336661249,
+      "grad_norm": 0.5534155964851379,
+      "learning_rate": 9.128134827834267e-05,
+      "loss": 0.13373114585876464,
+      "step": 13600
+    },
+    {
+      "epoch": 3.1634548409565824,
+      "grad_norm": 0.5069191455841064,
+      "learning_rate": 9.077362425508303e-05,
+      "loss": 0.14482768058776854,
+      "step": 13625
+    },
+    {
+      "epoch": 3.1692593452519153,
+      "grad_norm": 0.5293791890144348,
+      "learning_rate": 9.026670285026724e-05,
+      "loss": 0.14137742042541504,
+      "step": 13650
+    },
+    {
+      "epoch": 3.1750638495472487,
+      "grad_norm": 0.5128117799758911,
+      "learning_rate": 8.976059093354532e-05,
+      "loss": 0.1423122501373291,
+      "step": 13675
+    },
+    {
+      "epoch": 3.1808683538425817,
+      "grad_norm": 0.48873060941696167,
+      "learning_rate": 8.925529536359725e-05,
+      "loss": 0.14092047691345214,
+      "step": 13700
+    },
+    {
+      "epoch": 3.186672858137915,
+      "grad_norm": 0.5173494219779968,
+      "learning_rate": 8.875082298804016e-05,
+      "loss": 0.14156684875488282,
+      "step": 13725
+    },
+    {
+      "epoch": 3.192477362433248,
+      "grad_norm": 0.5836243629455566,
+      "learning_rate": 8.824718064333533e-05,
+      "loss": 0.1422504997253418,
+      "step": 13750
+    },
+    {
+      "epoch": 3.1982818667285815,
+      "grad_norm": 0.48089680075645447,
+      "learning_rate": 8.774437515469609e-05,
+      "loss": 0.13057588577270507,
+      "step": 13775
+    },
+    {
+      "epoch": 3.2040863710239145,
+      "grad_norm": 0.5307944416999817,
+      "learning_rate": 8.724241333599456e-05,
+      "loss": 0.1337298583984375,
+      "step": 13800
+    },
+    {
+      "epoch": 3.209890875319248,
+      "grad_norm": 0.5847126245498657,
+      "learning_rate": 8.674130198966984e-05,
+      "loss": 0.13666083335876464,
+      "step": 13825
+    },
+    {
+      "epoch": 3.215695379614581,
+      "grad_norm": 0.574516236782074,
+      "learning_rate": 8.624104790663565e-05,
+      "loss": 0.1451045322418213,
+      "step": 13850
+    },
+    {
+      "epoch": 3.2214998839099143,
+      "grad_norm": 0.4958314001560211,
+      "learning_rate": 8.574165786618848e-05,
+      "loss": 0.13531614303588868,
+      "step": 13875
+    },
+    {
+      "epoch": 3.2273043882052472,
+      "grad_norm": 0.5250740051269531,
+      "learning_rate": 8.524313863591524e-05,
+      "loss": 0.1422373390197754,
+      "step": 13900
+    },
+    {
+      "epoch": 3.2331088925005806,
+      "grad_norm": 0.569054901599884,
+      "learning_rate": 8.474549697160229e-05,
+      "loss": 0.14293704986572264,
+      "step": 13925
+    },
+    {
+      "epoch": 3.2389133967959136,
+      "grad_norm": 0.5308824181556702,
+      "learning_rate": 8.42487396171431e-05,
+      "loss": 0.14238821983337402,
+      "step": 13950
+    },
+    {
+      "epoch": 3.2447179010912466,
+      "grad_norm": 0.48613181710243225,
+      "learning_rate": 8.37528733044475e-05,
+      "loss": 0.1415052318572998,
+      "step": 13975
+    },
+    {
+      "epoch": 3.25052240538658,
+      "grad_norm": 0.5107966661453247,
+      "learning_rate": 8.325790475335003e-05,
+      "loss": 0.14681419372558593,
+      "step": 14000
+    },
+    {
+      "epoch": 3.25052240538658,
+      "eval_loss": 0.4836580455303192,
+      "eval_runtime": 2.8729,
+      "eval_samples_per_second": 626.554,
+      "eval_steps_per_second": 10.094,
+      "step": 14000
+    },
+    {
+      "epoch": 3.2563269096819134,
+      "grad_norm": 0.525264322757721,
+      "learning_rate": 8.276384067151902e-05,
+      "loss": 0.1323159122467041,
+      "step": 14025
+    },
+    {
+      "epoch": 3.2621314139772464,
+      "grad_norm": 0.48356688022613525,
+      "learning_rate": 8.227068775436585e-05,
+      "loss": 0.13795961380004884,
+      "step": 14050
+    },
+    {
+      "epoch": 3.2679359182725793,
+      "grad_norm": 0.5610079169273376,
+      "learning_rate": 8.177845268495398e-05,
+      "loss": 0.14065306663513183,
+      "step": 14075
+    },
+    {
+      "epoch": 3.2737404225679128,
+      "grad_norm": 0.5489832162857056,
+      "learning_rate": 8.128714213390841e-05,
+      "loss": 0.13862586975097657,
+      "step": 14100
+    },
+    {
+      "epoch": 3.2795449268632457,
+      "grad_norm": 0.5396311283111572,
+      "learning_rate": 8.079676275932538e-05,
+      "loss": 0.1528916072845459,
+      "step": 14125
+    },
+    {
+      "epoch": 3.285349431158579,
+      "grad_norm": 0.6121686100959778,
+      "learning_rate": 8.030732120668201e-05,
+      "loss": 0.14702988624572755,
+      "step": 14150
+    },
+    {
+      "epoch": 3.291153935453912,
+      "grad_norm": 0.48160678148269653,
+      "learning_rate": 7.981882410874663e-05,
+      "loss": 0.1330477237701416,
+      "step": 14175
+    },
+    {
+      "epoch": 3.2969584397492455,
+      "grad_norm": 0.5235905647277832,
+      "learning_rate": 7.933127808548832e-05,
+      "loss": 0.14335980415344238,
+      "step": 14200
+    },
+    {
+      "epoch": 3.3027629440445785,
+      "grad_norm": 0.4550098776817322,
+      "learning_rate": 7.88446897439875e-05,
+      "loss": 0.14953779220581054,
+      "step": 14225
+    },
+    {
+      "epoch": 3.308567448339912,
+      "grad_norm": 0.49624404311180115,
+      "learning_rate": 7.835906567834641e-05,
+      "loss": 0.13442384719848632,
+      "step": 14250
+    },
+    {
+      "epoch": 3.314371952635245,
+      "grad_norm": 0.45998451113700867,
+      "learning_rate": 7.787441246959973e-05,
+      "loss": 0.13967049598693848,
+      "step": 14275
+    },
+    {
+      "epoch": 3.3201764569305783,
+      "grad_norm": 0.5856457948684692,
+      "learning_rate": 7.739073668562543e-05,
+      "loss": 0.15566299438476563,
+      "step": 14300
+    },
+    {
+      "epoch": 3.3259809612259112,
+      "grad_norm": 0.5796367526054382,
+      "learning_rate": 7.690804488105561e-05,
+      "loss": 0.14303340911865234,
+      "step": 14325
+    },
+    {
+      "epoch": 3.3317854655212447,
+      "grad_norm": 0.47587743401527405,
+      "learning_rate": 7.642634359718768e-05,
+      "loss": 0.12955387115478514,
+      "step": 14350
+    },
+    {
+      "epoch": 3.3375899698165776,
+      "grad_norm": 0.5801305770874023,
+      "learning_rate": 7.594563936189609e-05,
+      "loss": 0.14990788459777832,
+      "step": 14375
+    },
+    {
+      "epoch": 3.3433944741119106,
+      "grad_norm": 0.5009943842887878,
+      "learning_rate": 7.546593868954323e-05,
+      "loss": 0.14071702003479003,
+      "step": 14400
+    },
+    {
+      "epoch": 3.349198978407244,
+      "grad_norm": 0.5151167511940002,
+      "learning_rate": 7.498724808089187e-05,
+      "loss": 0.13961894035339356,
+      "step": 14425
+    },
+    {
+      "epoch": 3.3550034827025774,
+      "grad_norm": 0.46149569749832153,
+      "learning_rate": 7.450957402301635e-05,
+      "loss": 0.13761077880859374,
+      "step": 14450
+    },
+    {
+      "epoch": 3.3608079869979104,
+      "grad_norm": 0.47205618023872375,
+      "learning_rate": 7.403292298921528e-05,
+      "loss": 0.14032185554504395,
+      "step": 14475
+    },
+    {
+      "epoch": 3.3666124912932434,
+      "grad_norm": 0.5129955410957336,
+      "learning_rate": 7.355730143892336e-05,
+      "loss": 0.13600680351257324,
+      "step": 14500
+    },
+    {
+      "epoch": 3.3724169955885768,
+      "grad_norm": 0.4013858437538147,
+      "learning_rate": 7.308271581762408e-05,
+      "loss": 0.13269561767578125,
+      "step": 14525
+    },
+    {
+      "epoch": 3.3782214998839097,
+      "grad_norm": 0.5339265465736389,
+      "learning_rate": 7.26091725567624e-05,
+      "loss": 0.13466255187988282,
+      "step": 14550
+    },
+    {
+      "epoch": 3.384026004179243,
+      "grad_norm": 0.5491599440574646,
+      "learning_rate": 7.21366780736575e-05,
+      "loss": 0.15055075645446778,
+      "step": 14575
+    },
+    {
+      "epoch": 3.389830508474576,
+      "grad_norm": 0.47083404660224915,
+      "learning_rate": 7.166523877141578e-05,
+      "loss": 0.13845568656921386,
+      "step": 14600
+    },
+    {
+      "epoch": 3.3956350127699095,
+      "grad_norm": 0.6038816571235657,
+      "learning_rate": 7.119486103884408e-05,
+      "loss": 0.14179848670959472,
+      "step": 14625
+    },
+    {
+      "epoch": 3.4014395170652425,
+      "grad_norm": 0.45376163721084595,
+      "learning_rate": 7.072555125036337e-05,
+      "loss": 0.13612910270690917,
+      "step": 14650
+    },
+    {
+      "epoch": 3.407244021360576,
+      "grad_norm": 0.5730807185173035,
+      "learning_rate": 7.025731576592189e-05,
+      "loss": 0.13847175598144532,
+      "step": 14675
+    },
+    {
+      "epoch": 3.413048525655909,
+      "grad_norm": 0.5629493594169617,
+      "learning_rate": 6.979016093090949e-05,
+      "loss": 0.1329249095916748,
+      "step": 14700
+    },
+    {
+      "epoch": 3.4188530299512423,
+      "grad_norm": 0.5275311470031738,
+      "learning_rate": 6.932409307607119e-05,
+      "loss": 0.14288698196411131,
+      "step": 14725
+    },
+    {
+      "epoch": 3.4246575342465753,
+      "grad_norm": 0.6125319004058838,
+      "learning_rate": 6.885911851742154e-05,
+      "loss": 0.14283589363098145,
+      "step": 14750
+    },
+    {
+      "epoch": 3.4304620385419087,
+      "grad_norm": 0.4843354821205139,
+      "learning_rate": 6.839524355615926e-05,
+      "loss": 0.14155942916870118,
+      "step": 14775
+    },
+    {
+      "epoch": 3.4362665428372416,
+      "grad_norm": 0.5407298803329468,
+      "learning_rate": 6.79324744785816e-05,
+      "loss": 0.1371714973449707,
+      "step": 14800
+    },
+    {
+      "epoch": 3.442071047132575,
+      "grad_norm": 0.5010970234870911,
+      "learning_rate": 6.747081755599904e-05,
+      "loss": 0.13952279090881348,
+      "step": 14825
+    },
+    {
+      "epoch": 3.447875551427908,
+      "grad_norm": 0.4812754690647125,
+      "learning_rate": 6.70102790446506e-05,
+      "loss": 0.14485257148742675,
+      "step": 14850
+    },
+    {
+      "epoch": 3.4536800557232414,
+      "grad_norm": 0.5330603718757629,
+      "learning_rate": 6.65508651856188e-05,
+      "loss": 0.1368393325805664,
+      "step": 14875
+    },
+    {
+      "epoch": 3.4594845600185744,
+      "grad_norm": 0.5017295479774475,
+      "learning_rate": 6.609258220474537e-05,
+      "loss": 0.13505492210388184,
+      "step": 14900
+    },
+    {
+      "epoch": 3.4652890643139074,
+      "grad_norm": 0.4695931673049927,
+      "learning_rate": 6.563543631254667e-05,
+      "loss": 0.12914393424987794,
+      "step": 14925
+    },
+    {
+      "epoch": 3.471093568609241,
+      "grad_norm": 0.4295479655265808,
+      "learning_rate": 6.517943370412947e-05,
+      "loss": 0.13729445457458497,
+      "step": 14950
+    },
+    {
+      "epoch": 3.4768980729045738,
+      "grad_norm": 0.5042039752006531,
+      "learning_rate": 6.47245805591071e-05,
+      "loss": 0.13052743911743164,
+      "step": 14975
+    },
+    {
+      "epoch": 3.482702577199907,
+      "grad_norm": 0.5624018311500549,
+      "learning_rate": 6.42708830415158e-05,
+      "loss": 0.13101316452026368,
+      "step": 15000
+    },
+    {
+      "epoch": 3.482702577199907,
+      "eval_loss": 0.4849991798400879,
+      "eval_runtime": 2.759,
+      "eval_samples_per_second": 652.41,
+      "eval_steps_per_second": 10.511,
+      "step": 15000
+    },
+    {
+      "epoch": 3.48850708149524,
+      "grad_norm": 0.45914268493652344,
+      "learning_rate": 6.381834729973109e-05,
+      "loss": 0.13552204132080078,
+      "step": 15025
+    },
+    {
+      "epoch": 3.4943115857905735,
+      "grad_norm": 0.5392463207244873,
+      "learning_rate": 6.336697946638433e-05,
+      "loss": 0.1444261837005615,
+      "step": 15050
+    },
+    {
+      "epoch": 3.5001160900859065,
+      "grad_norm": 0.5812017321586609,
+      "learning_rate": 6.291678565827973e-05,
+      "loss": 0.13363587379455566,
+      "step": 15075
+    },
+    {
+      "epoch": 3.50592059438124,
+      "grad_norm": 0.6662575006484985,
+      "learning_rate": 6.246777197631166e-05,
+      "loss": 0.1365174674987793,
+      "step": 15100
+    },
+    {
+      "epoch": 3.511725098676573,
+      "grad_norm": 0.5669078230857849,
+      "learning_rate": 6.201994450538148e-05,
+      "loss": 0.12696731567382813,
+      "step": 15125
+    },
+    {
+      "epoch": 3.5175296029719063,
+      "grad_norm": 0.5152287483215332,
+      "learning_rate": 6.157330931431562e-05,
+      "loss": 0.1351464080810547,
+      "step": 15150
+    },
+    {
+      "epoch": 3.5233341072672393,
+      "grad_norm": 0.5453899502754211,
+      "learning_rate": 6.112787245578287e-05,
+      "loss": 0.12452445030212403,
+      "step": 15175
+    },
+    {
+      "epoch": 3.5291386115625727,
+      "grad_norm": 0.4677042067050934,
+      "learning_rate": 6.068363996621278e-05,
+      "loss": 0.1352981948852539,
+      "step": 15200
+    },
+    {
+      "epoch": 3.5349431158579057,
+      "grad_norm": 0.5128089189529419,
+      "learning_rate": 6.024061786571348e-05,
+      "loss": 0.12971482276916504,
+      "step": 15225
+    },
+    {
+      "epoch": 3.540747620153239,
+      "grad_norm": 0.6132077574729919,
+      "learning_rate": 5.9798812157990266e-05,
+      "loss": 0.1376301670074463,
+      "step": 15250
+    },
+    {
+      "epoch": 3.546552124448572,
+      "grad_norm": 0.49205633997917175,
+      "learning_rate": 5.9358228830264345e-05,
+      "loss": 0.14495808601379395,
+      "step": 15275
+    },
+    {
+      "epoch": 3.5523566287439055,
+      "grad_norm": 0.5420359373092651,
+      "learning_rate": 5.89188738531916e-05,
+      "loss": 0.1471403408050537,
+      "step": 15300
+    },
+    {
+      "epoch": 3.5581611330392384,
+      "grad_norm": 0.6124544143676758,
+      "learning_rate": 5.8480753180781506e-05,
+      "loss": 0.13519423484802245,
+      "step": 15325
+    },
+    {
+      "epoch": 3.5639656373345714,
+      "grad_norm": 0.442746639251709,
+      "learning_rate": 5.80438727503167e-05,
+      "loss": 0.1311307430267334,
+      "step": 15350
+    },
+    {
+      "epoch": 3.569770141629905,
+      "grad_norm": 0.5656249523162842,
+      "learning_rate": 5.7608238482272445e-05,
+      "loss": 0.13945990562438965,
+      "step": 15375
+    },
+    {
+      "epoch": 3.575574645925238,
+      "grad_norm": 0.5650652647018433,
+      "learning_rate": 5.717385628023641e-05,
+      "loss": 0.1330738067626953,
+      "step": 15400
+    },
+    {
+      "epoch": 3.581379150220571,
+      "grad_norm": 0.5396448373794556,
+      "learning_rate": 5.674073203082855e-05,
+      "loss": 0.13562232971191407,
+      "step": 15425
+    },
+    {
+      "epoch": 3.587183654515904,
+      "grad_norm": 0.5073605179786682,
+      "learning_rate": 5.630887160362142e-05,
+      "loss": 0.13419599533081056,
+      "step": 15450
+    },
+    {
+      "epoch": 3.5929881588112376,
+      "grad_norm": 0.6020980477333069,
+      "learning_rate": 5.587828085106061e-05,
+      "loss": 0.14006584167480468,
+      "step": 15475
+    },
+    {
+      "epoch": 3.5987926631065705,
+      "grad_norm": 0.45168739557266235,
+      "learning_rate": 5.5448965608385536e-05,
+      "loss": 0.14650278091430663,
+      "step": 15500
+    },
+    {
+      "epoch": 3.604597167401904,
+      "grad_norm": 0.5814923048019409,
+      "learning_rate": 5.5020931693550305e-05,
+      "loss": 0.13090405464172364,
+      "step": 15525
+    },
+    {
+      "epoch": 3.610401671697237,
+      "grad_norm": 0.48454540967941284,
+      "learning_rate": 5.459418490714473e-05,
+      "loss": 0.14684450149536132,
+      "step": 15550
+    },
+    {
+      "epoch": 3.6162061759925703,
+      "grad_norm": 0.47804901003837585,
+      "learning_rate": 5.416873103231581e-05,
+      "loss": 0.14159123420715333,
+      "step": 15575
+    },
+    {
+      "epoch": 3.6220106802879033,
+      "grad_norm": 0.47777873277664185,
+      "learning_rate": 5.3744575834689596e-05,
+      "loss": 0.13593565940856933,
+      "step": 15600
+    },
+    {
+      "epoch": 3.6278151845832367,
+      "grad_norm": 0.5319150686264038,
+      "learning_rate": 5.332172506229258e-05,
+      "loss": 0.13024145126342773,
+      "step": 15625
+    },
+    {
+      "epoch": 3.6336196888785697,
+      "grad_norm": 0.5830259919166565,
+      "learning_rate": 5.290018444547431e-05,
+      "loss": 0.13721506118774415,
+      "step": 15650
+    },
+    {
+      "epoch": 3.639424193173903,
+      "grad_norm": 0.5299803614616394,
+      "learning_rate": 5.247995969682929e-05,
+      "loss": 0.132626953125,
+      "step": 15675
+    },
+    {
+      "epoch": 3.645228697469236,
+      "grad_norm": 0.5016283392906189,
+      "learning_rate": 5.206105651111996e-05,
+      "loss": 0.14233938217163086,
+      "step": 15700
+    },
+    {
+      "epoch": 3.6510332017645695,
+      "grad_norm": 0.5307296514511108,
+      "learning_rate": 5.1643480565199105e-05,
+      "loss": 0.14133543968200685,
+      "step": 15725
+    },
+    {
+      "epoch": 3.6568377060599024,
+      "grad_norm": 0.5165666341781616,
+      "learning_rate": 5.122723751793337e-05,
+      "loss": 0.13104183197021485,
+      "step": 15750
+    },
+    {
+      "epoch": 3.6626422103552354,
+      "grad_norm": 0.5122671723365784,
+      "learning_rate": 5.081233301012618e-05,
+      "loss": 0.13036978721618653,
+      "step": 15775
+    },
+    {
+      "epoch": 3.668446714650569,
+      "grad_norm": 0.5937213897705078,
+      "learning_rate": 5.0398772664441524e-05,
+      "loss": 0.13120935440063478,
+      "step": 15800
+    },
+    {
+      "epoch": 3.6742512189459022,
+      "grad_norm": 0.5100651383399963,
+      "learning_rate": 4.998656208532776e-05,
+      "loss": 0.12783276557922363,
+      "step": 15825
+    },
+    {
+      "epoch": 3.680055723241235,
+      "grad_norm": 0.5779837369918823,
+      "learning_rate": 4.957570685894146e-05,
+      "loss": 0.13319705963134765,
+      "step": 15850
+    },
+    {
+      "epoch": 3.685860227536568,
+      "grad_norm": 0.4842281639575958,
+      "learning_rate": 4.916621255307206e-05,
+      "loss": 0.12948975563049317,
+      "step": 15875
+    },
+    {
+      "epoch": 3.6916647318319016,
+      "grad_norm": 0.6076027750968933,
+      "learning_rate": 4.875808471706597e-05,
+      "loss": 0.1366411018371582,
+      "step": 15900
+    },
+    {
+      "epoch": 3.697469236127235,
+      "grad_norm": 0.5423301458358765,
+      "learning_rate": 4.8351328881751854e-05,
+      "loss": 0.13042160034179687,
+      "step": 15925
+    },
+    {
+      "epoch": 3.703273740422568,
+      "grad_norm": 0.47805365920066833,
+      "learning_rate": 4.79459505593652e-05,
+      "loss": 0.12281619071960449,
+      "step": 15950
+    },
+    {
+      "epoch": 3.709078244717901,
+      "grad_norm": 0.5300124287605286,
+      "learning_rate": 4.75419552434739e-05,
+      "loss": 0.13224746704101562,
+      "step": 15975
+    },
+    {
+      "epoch": 3.7148827490132343,
+      "grad_norm": 0.5495495200157166,
+      "learning_rate": 4.713934840890386e-05,
+      "loss": 0.1343132209777832,
+      "step": 16000
+    },
+    {
+      "epoch": 3.7148827490132343,
+      "eval_loss": 0.4834541082382202,
+      "eval_runtime": 2.7826,
+      "eval_samples_per_second": 646.866,
+      "eval_steps_per_second": 10.422,
+      "step": 16000
+    },
+    {
+      "epoch": 3.7206872533085673,
+      "grad_norm": 0.5209447145462036,
+      "learning_rate": 4.673813551166462e-05,
+      "loss": 0.1316848564147949,
+      "step": 16025
+    },
+    {
+      "epoch": 3.7264917576039007,
+      "grad_norm": 0.5507779121398926,
+      "learning_rate": 4.6338321988875486e-05,
+      "loss": 0.12162506103515625,
+      "step": 16050
+    },
+    {
+      "epoch": 3.7322962618992337,
+      "grad_norm": 0.5463923811912537,
+      "learning_rate": 4.593991325869183e-05,
+      "loss": 0.1306621265411377,
+      "step": 16075
+    },
+    {
+      "epoch": 3.738100766194567,
+      "grad_norm": 0.6377285122871399,
+      "learning_rate": 4.554291472023163e-05,
+      "loss": 0.13987205505371095,
+      "step": 16100
+    },
+    {
+      "epoch": 3.7439052704899,
+      "grad_norm": 0.5184104442596436,
+      "learning_rate": 4.514733175350263e-05,
+      "loss": 0.1311191749572754,
+      "step": 16125
+    },
+    {
+      "epoch": 3.7497097747852335,
+      "grad_norm": 0.48756757378578186,
+      "learning_rate": 4.475316971932889e-05,
+      "loss": 0.1382326602935791,
+      "step": 16150
+    },
+    {
+      "epoch": 3.7555142790805665,
+      "grad_norm": 0.5549758672714233,
+      "learning_rate": 4.436043395927849e-05,
+      "loss": 0.13588230133056642,
+      "step": 16175
+    },
+    {
+      "epoch": 3.7613187833758994,
+      "grad_norm": 0.5385533571243286,
+      "learning_rate": 4.3969129795591015e-05,
+      "loss": 0.1300955295562744,
+      "step": 16200
+    },
+    {
+      "epoch": 3.767123287671233,
+      "grad_norm": 0.49518296122550964,
+      "learning_rate": 4.357926253110559e-05,
+      "loss": 0.12474268913269043,
+      "step": 16225
+    },
+    {
+      "epoch": 3.7729277919665662,
+      "grad_norm": 0.5287390947341919,
+      "learning_rate": 4.3190837449188864e-05,
+      "loss": 0.13263528823852538,
+      "step": 16250
+    },
+    {
+      "epoch": 3.778732296261899,
+      "grad_norm": 0.45688286423683167,
+      "learning_rate": 4.280385981366341e-05,
+      "loss": 0.12746773719787596,
+      "step": 16275
+    },
+    {
+      "epoch": 3.784536800557232,
+      "grad_norm": 0.5174150466918945,
+      "learning_rate": 4.241833486873641e-05,
+      "loss": 0.132682466506958,
+      "step": 16300
+    },
+    {
+      "epoch": 3.7903413048525656,
+      "grad_norm": 0.5170544981956482,
+      "learning_rate": 4.203426783892874e-05,
+      "loss": 0.1312546157836914,
+      "step": 16325
+    },
+    {
+      "epoch": 3.796145809147899,
+      "grad_norm": 0.6146380305290222,
+      "learning_rate": 4.1651663929003834e-05,
+      "loss": 0.12222254753112793,
+      "step": 16350
+    },
+    {
+      "epoch": 3.801950313443232,
+      "grad_norm": 0.5436851382255554,
+      "learning_rate": 4.1270528323897614e-05,
+      "loss": 0.12948195457458497,
+      "step": 16375
+    },
+    {
+      "epoch": 3.807754817738565,
+      "grad_norm": 0.4765893816947937,
+      "learning_rate": 4.08908661886477e-05,
+      "loss": 0.13217971801757813,
+      "step": 16400
+    },
+    {
+      "epoch": 3.8135593220338984,
+      "grad_norm": 0.5577421188354492,
+      "learning_rate": 4.0512682668323896e-05,
+      "loss": 0.12744462966918946,
+      "step": 16425
+    },
+    {
+      "epoch": 3.8193638263292313,
+      "grad_norm": 0.5351262092590332,
+      "learning_rate": 4.013598288795808e-05,
+      "loss": 0.13391121864318847,
+      "step": 16450
+    },
+    {
+      "epoch": 3.8251683306245647,
+      "grad_norm": 0.5594732761383057,
+      "learning_rate": 3.9760771952475106e-05,
+      "loss": 0.13245270729064942,
+      "step": 16475
+    },
+    {
+      "epoch": 3.8309728349198977,
+      "grad_norm": 0.5924484133720398,
+      "learning_rate": 3.9387054946623234e-05,
+      "loss": 0.12773351669311522,
+      "step": 16500
+    },
+    {
+      "epoch": 3.836777339215231,
+      "grad_norm": 0.5322756767272949,
+      "learning_rate": 3.901483693490563e-05,
+      "loss": 0.1348324680328369,
+      "step": 16525
+    },
+    {
+      "epoch": 3.842581843510564,
+      "grad_norm": 0.5091602802276611,
+      "learning_rate": 3.864412296151141e-05,
+      "loss": 0.12817475318908692,
+      "step": 16550
+    },
+    {
+      "epoch": 3.8483863478058975,
+      "grad_norm": 0.5416632294654846,
+      "learning_rate": 3.827491805024733e-05,
+      "loss": 0.12608692169189453,
+      "step": 16575
+    },
+    {
+      "epoch": 3.8541908521012305,
+      "grad_norm": 0.4963054060935974,
+      "learning_rate": 3.790722720446997e-05,
+      "loss": 0.13913899421691894,
+      "step": 16600
+    },
+    {
+      "epoch": 3.859995356396564,
+      "grad_norm": 0.5441203117370605,
+      "learning_rate": 3.7541055407017657e-05,
+      "loss": 0.12793876647949218,
+      "step": 16625
+    },
+    {
+      "epoch": 3.865799860691897,
+      "grad_norm": 0.5654854774475098,
+      "learning_rate": 3.717640762014299e-05,
+      "loss": 0.1283561897277832,
+      "step": 16650
+    },
+    {
+      "epoch": 3.8716043649872303,
+      "grad_norm": 0.5109981298446655,
+      "learning_rate": 3.6813288785445665e-05,
+      "loss": 0.1224953842163086,
+      "step": 16675
+    },
+    {
+      "epoch": 3.8774088692825632,
+      "grad_norm": 0.47984960675239563,
+      "learning_rate": 3.6451703823805356e-05,
+      "loss": 0.12291561126708984,
+      "step": 16700
+    },
+    {
+      "epoch": 3.883213373577896,
+      "grad_norm": 0.5661267042160034,
+      "learning_rate": 3.6091657635315315e-05,
+      "loss": 0.12208088874816894,
+      "step": 16725
+    },
+    {
+      "epoch": 3.8890178778732296,
+      "grad_norm": 0.5422415733337402,
+      "learning_rate": 3.573315509921575e-05,
+      "loss": 0.1347252082824707,
+      "step": 16750
+    },
+    {
+      "epoch": 3.894822382168563,
+      "grad_norm": 0.5453212857246399,
+      "learning_rate": 3.537620107382764e-05,
+      "loss": 0.13169493675231933,
+      "step": 16775
+    },
+    {
+      "epoch": 3.900626886463896,
+      "grad_norm": 0.47127565741539,
+      "learning_rate": 3.5020800396487044e-05,
+      "loss": 0.12600440979003907,
+      "step": 16800
+    },
+    {
+      "epoch": 3.906431390759229,
+      "grad_norm": 0.47693362832069397,
+      "learning_rate": 3.466695788347955e-05,
+      "loss": 0.12363935470581054,
+      "step": 16825
+    },
+    {
+      "epoch": 3.9122358950545624,
+      "grad_norm": 0.5626474022865295,
+      "learning_rate": 3.4314678329975e-05,
+      "loss": 0.12666434288024903,
+      "step": 16850
+    },
+    {
+      "epoch": 3.918040399349896,
+      "grad_norm": 0.49834829568862915,
+      "learning_rate": 3.3963966509962355e-05,
+      "loss": 0.1259426784515381,
+      "step": 16875
+    },
+    {
+      "epoch": 3.9238449036452288,
+      "grad_norm": 0.48170289397239685,
+      "learning_rate": 3.361482717618513e-05,
+      "loss": 0.13647454261779784,
+      "step": 16900
+    },
+    {
+      "epoch": 3.9296494079405617,
+      "grad_norm": 0.47936519980430603,
+      "learning_rate": 3.326726506007708e-05,
+      "loss": 0.12583857536315918,
+      "step": 16925
+    },
+    {
+      "epoch": 3.935453912235895,
+      "grad_norm": 0.5467544794082642,
+      "learning_rate": 3.2921284871697824e-05,
+      "loss": 0.13049958229064942,
+      "step": 16950
+    },
+    {
+      "epoch": 3.941258416531228,
+      "grad_norm": 0.5074909925460815,
+      "learning_rate": 3.2576891299669264e-05,
+      "loss": 0.13176217079162597,
+      "step": 16975
+    },
+    {
+      "epoch": 3.9470629208265615,
+      "grad_norm": 0.4521033763885498,
+      "learning_rate": 3.2234089011111934e-05,
+      "loss": 0.13002379417419432,
+      "step": 17000
+    },
+    {
+      "epoch": 3.9470629208265615,
+      "eval_loss": 0.48477375507354736,
+      "eval_runtime": 2.7685,
+      "eval_samples_per_second": 650.172,
+      "eval_steps_per_second": 10.475,
+      "step": 17000
+    },
+    {
+      "epoch": 3.9528674251218945,
+      "grad_norm": 0.5549446940422058,
+      "learning_rate": 3.189288265158164e-05,
+      "loss": 0.1334219169616699,
+      "step": 17025
+    },
+    {
+      "epoch": 3.958671929417228,
+      "grad_norm": 0.5062447786331177,
+      "learning_rate": 3.155327684500683e-05,
+      "loss": 0.12354134559631348,
+      "step": 17050
+    },
+    {
+      "epoch": 3.964476433712561,
+      "grad_norm": 0.5579578280448914,
+      "learning_rate": 3.1215276193625525e-05,
+      "loss": 0.12970406532287598,
+      "step": 17075
+    },
+    {
+      "epoch": 3.9702809380078943,
+      "grad_norm": 0.6180189251899719,
+      "learning_rate": 3.087888527792337e-05,
+      "loss": 0.1254385757446289,
+      "step": 17100
+    },
+    {
+      "epoch": 3.9760854423032272,
+      "grad_norm": 0.4233727753162384,
+      "learning_rate": 3.05441086565711e-05,
+      "loss": 0.1317064094543457,
+      "step": 17125
+    },
+    {
+      "epoch": 3.98188994659856,
+      "grad_norm": 0.5346817970275879,
+      "learning_rate": 3.0210950866363214e-05,
+      "loss": 0.12989388465881346,
+      "step": 17150
+    },
+    {
+      "epoch": 3.9876944508938936,
+      "grad_norm": 0.475293904542923,
+      "learning_rate": 2.9879416422156204e-05,
+      "loss": 0.12659363746643065,
+      "step": 17175
+    },
+    {
+      "epoch": 3.993498955189227,
+      "grad_norm": 0.5779196619987488,
+      "learning_rate": 2.9549509816807386e-05,
+      "loss": 0.12317920684814453,
+      "step": 17200
+    },
+    {
+      "epoch": 3.99930345948456,
+      "grad_norm": 0.5648007392883301,
+      "learning_rate": 2.922123552111421e-05,
+      "loss": 0.14431089401245117,
+      "step": 17225
+    },
+    {
+      "epoch": 4.005107963779893,
+      "grad_norm": 0.40254271030426025,
+      "learning_rate": 2.889459798375352e-05,
+      "loss": 0.0905971336364746,
+      "step": 17250
+    },
+    {
+      "epoch": 4.010912468075226,
+      "grad_norm": 0.41820237040519714,
+      "learning_rate": 2.85696016312212e-05,
+      "loss": 0.0847398567199707,
+      "step": 17275
+    },
+    {
+      "epoch": 4.01671697237056,
+      "grad_norm": 0.3613129258155823,
+      "learning_rate": 2.824625086777233e-05,
+      "loss": 0.08114529609680175,
+      "step": 17300
+    },
+    {
+      "epoch": 4.022521476665893,
+      "grad_norm": 0.45001861453056335,
+      "learning_rate": 2.7924550075361474e-05,
+      "loss": 0.08282732963562012,
+      "step": 17325
+    },
+    {
+      "epoch": 4.028325980961226,
+      "grad_norm": 0.4341457486152649,
+      "learning_rate": 2.7604503613583302e-05,
+      "loss": 0.08115843772888183,
+      "step": 17350
+    },
+    {
+      "epoch": 4.034130485256559,
+      "grad_norm": 0.4107372462749481,
+      "learning_rate": 2.7286115819613414e-05,
+      "loss": 0.08760406494140625,
+      "step": 17375
+    },
+    {
+      "epoch": 4.039934989551893,
+      "grad_norm": 0.47246018052101135,
+      "learning_rate": 2.6969391008149645e-05,
+      "loss": 0.08065567016601563,
+      "step": 17400
+    },
+    {
+      "epoch": 4.0457394938472255,
+      "grad_norm": 0.43137720227241516,
+      "learning_rate": 2.6654333471353534e-05,
+      "loss": 0.07961277484893799,
+      "step": 17425
+    },
+    {
+      "epoch": 4.0515439981425585,
+      "grad_norm": 0.4721987247467041,
+      "learning_rate": 2.6340947478792318e-05,
+      "loss": 0.08728896141052246,
+      "step": 17450
+    },
+    {
+      "epoch": 4.0573485024378915,
+      "grad_norm": 0.39357468485832214,
+      "learning_rate": 2.602923727738093e-05,
+      "loss": 0.07964069843292236,
+      "step": 17475
+    },
+    {
+      "epoch": 4.063153006733225,
+      "grad_norm": 0.4414040744304657,
+      "learning_rate": 2.5719207091324394e-05,
+      "loss": 0.0861448574066162,
+      "step": 17500
+    },
+    {
+      "epoch": 4.068957511028558,
+      "grad_norm": 0.5105574131011963,
+      "learning_rate": 2.5410861122060688e-05,
+      "loss": 0.08487778663635254,
+      "step": 17525
+    },
+    {
+      "epoch": 4.074762015323891,
+      "grad_norm": 0.4436439275741577,
+      "learning_rate": 2.5104203548203865e-05,
+      "loss": 0.08540387153625488,
+      "step": 17550
+    },
+    {
+      "epoch": 4.080566519619224,
+      "grad_norm": 0.46121156215667725,
+      "learning_rate": 2.4799238525487187e-05,
+      "loss": 0.07835439205169678,
+      "step": 17575
+    },
+    {
+      "epoch": 4.086371023914558,
+      "grad_norm": 0.4219311773777008,
+      "learning_rate": 2.4495970186707138e-05,
+      "loss": 0.07797367095947266,
+      "step": 17600
+    },
+    {
+      "epoch": 4.092175528209891,
+      "grad_norm": 0.35916611552238464,
+      "learning_rate": 2.4194402641667067e-05,
+      "loss": 0.0760021162033081,
+      "step": 17625
+    },
+    {
+      "epoch": 4.097980032505224,
+      "grad_norm": 0.41230475902557373,
+      "learning_rate": 2.389453997712181e-05,
+      "loss": 0.08162998199462891,
+      "step": 17650
+    },
+    {
+      "epoch": 4.103784536800557,
+      "grad_norm": 0.4652130901813507,
+      "learning_rate": 2.3596386256722027e-05,
+      "loss": 0.07778237342834472,
+      "step": 17675
+    },
+    {
+      "epoch": 4.109589041095891,
+      "grad_norm": 0.3833296597003937,
+      "learning_rate": 2.3299945520959435e-05,
+      "loss": 0.07606776237487793,
+      "step": 17700
+    },
+    {
+      "epoch": 4.115393545391224,
+      "grad_norm": 0.4073386788368225,
+      "learning_rate": 2.3005221787111704e-05,
+      "loss": 0.08484411239624023,
+      "step": 17725
+    },
+    {
+      "epoch": 4.121198049686557,
+      "grad_norm": 0.4941864311695099,
+      "learning_rate": 2.2712219049188363e-05,
+      "loss": 0.0828478717803955,
+      "step": 17750
+    },
+    {
+      "epoch": 4.12700255398189,
+      "grad_norm": 0.49691084027290344,
+      "learning_rate": 2.2420941277876402e-05,
+      "loss": 0.07821237564086914,
+      "step": 17775
+    },
+    {
+      "epoch": 4.132807058277223,
+      "grad_norm": 0.515224277973175,
+      "learning_rate": 2.2131392420486553e-05,
+      "loss": 0.07914109706878662,
+      "step": 17800
+    },
+    {
+      "epoch": 4.138611562572557,
+      "grad_norm": 0.5149050951004028,
+      "learning_rate": 2.184357640089992e-05,
+      "loss": 0.0818891716003418,
+      "step": 17825
+    },
+    {
+      "epoch": 4.1444160668678895,
+      "grad_norm": 0.4382188320159912,
+      "learning_rate": 2.1557497119514672e-05,
+      "loss": 0.0778397274017334,
+      "step": 17850
+    },
+    {
+      "epoch": 4.1502205711632225,
+      "grad_norm": 0.39906030893325806,
+      "learning_rate": 2.127315845319314e-05,
+      "loss": 0.08025979995727539,
+      "step": 17875
+    },
+    {
+      "epoch": 4.1560250754585555,
+      "grad_norm": 0.4684215486049652,
+      "learning_rate": 2.0990564255209408e-05,
+      "loss": 0.07968974590301514,
+      "step": 17900
+    },
+    {
+      "epoch": 4.161829579753889,
+      "grad_norm": 0.4352423846721649,
+      "learning_rate": 2.070971835519697e-05,
+      "loss": 0.07249928951263428,
+      "step": 17925
+    },
+    {
+      "epoch": 4.167634084049222,
+      "grad_norm": 0.44752639532089233,
+      "learning_rate": 2.0430624559097002e-05,
+      "loss": 0.07716537475585937,
+      "step": 17950
+    },
+    {
+      "epoch": 4.173438588344555,
+      "grad_norm": 0.4814079999923706,
+      "learning_rate": 2.0153286649106664e-05,
+      "loss": 0.08337730407714844,
+      "step": 17975
+    },
+    {
+      "epoch": 4.179243092639888,
+      "grad_norm": 0.38755491375923157,
+      "learning_rate": 1.9877708383627778e-05,
+      "loss": 0.07913094520568847,
+      "step": 18000
+    },
+    {
+      "epoch": 4.179243092639888,
+      "eval_loss": 0.5115543603897095,
+      "eval_runtime": 2.7419,
+      "eval_samples_per_second": 656.486,
+      "eval_steps_per_second": 10.577,
+      "step": 18000
+    },
+    {
+      "loss": 0.08247646331787109,
+      "grad_norm": 0.4921330213546753,
+      "learning_rate": 1.960389349721604e-05,
+      "epoch": 4.185047596935222,
+      "step": 18025
+    },
+    {
+      "loss": 0.08008021354675293,
+      "grad_norm": 0.48693451285362244,
+      "learning_rate": 1.9331845700530345e-05,
+      "epoch": 4.190852101230555,
+      "step": 18050
+    },
+    {
+      "loss": 0.07945438861846923,
+      "grad_norm": 0.4636516273021698,
+      "learning_rate": 1.9061568680282568e-05,
+      "epoch": 4.196656605525888,
+      "step": 18075
+    },
+    {
+      "loss": 0.08011796951293945,
+      "grad_norm": 0.4074261486530304,
+      "learning_rate": 1.8793066099187438e-05,
+      "epoch": 4.202461109821221,
+      "step": 18100
+    },
+    {
+      "loss": 0.08619665145874024,
+      "grad_norm": 0.512305498123169,
+      "learning_rate": 1.8526341595913053e-05,
+      "epoch": 4.208265614116555,
+      "step": 18125
+    },
+    {
+      "loss": 0.07600322723388672,
+      "grad_norm": 0.44227826595306396,
+      "learning_rate": 1.8261398785031473e-05,
+      "epoch": 4.214070118411888,
+      "step": 18150
+    },
+    {
+      "loss": 0.08032386779785156,
+      "grad_norm": 0.43220826983451843,
+      "learning_rate": 1.7998241256969886e-05,
+      "epoch": 4.219874622707221,
+      "step": 18175
+    },
+    {
+      "loss": 0.07907999515533447,
+      "grad_norm": 0.44413840770721436,
+      "learning_rate": 1.7736872577961818e-05,
+      "epoch": 4.225679127002554,
+      "step": 18200
+    },
+    {
+      "loss": 0.07748859882354736,
+      "grad_norm": 0.4341851770877838,
+      "learning_rate": 1.7477296289998787e-05,
+      "epoch": 4.231483631297887,
+      "step": 18225
+    },
+    {
+      "loss": 0.07818769931793212,
+      "grad_norm": 0.38617071509361267,
+      "learning_rate": 1.7219515910782373e-05,
+      "epoch": 4.237288135593221,
+      "step": 18250
+    },
+    {
+      "loss": 0.08131123542785644,
+      "grad_norm": 0.4250392019748688,
+      "learning_rate": 1.696353493367657e-05,
+      "epoch": 4.243092639888554,
+      "step": 18275
+    },
+    {
+      "loss": 0.07498374938964844,
+      "grad_norm": 0.42919921875,
+      "learning_rate": 1.6709356827660337e-05,
+      "epoch": 4.2488971441838865,
+      "step": 18300
+    },
+    {
+      "loss": 0.07327072620391846,
+      "grad_norm": 0.36674195528030396,
+      "learning_rate": 1.645698503728075e-05,
+      "epoch": 4.2547016484792195,
+      "step": 18325
+    },
+    {
+      "loss": 0.0801076602935791,
+      "grad_norm": 0.4721294939517975,
+      "learning_rate": 1.620642298260611e-05,
+      "epoch": 4.260506152774553,
+      "step": 18350
+    },
+    {
+      "loss": 0.07982881546020508,
+      "grad_norm": 0.47818973660469055,
+      "learning_rate": 1.5957674059179827e-05,
+      "epoch": 4.266310657069886,
+      "step": 18375
+    },
+    {
+      "loss": 0.07816235542297363,
+      "grad_norm": 0.49507322907447815,
+      "learning_rate": 1.5710741637974168e-05,
+      "epoch": 4.272115161365219,
+      "step": 18400
+    },
+    {
+      "loss": 0.07887996196746826,
+      "grad_norm": 0.49934637546539307,
+      "learning_rate": 1.546562906534481e-05,
+      "epoch": 4.277919665660552,
+      "step": 18425
+    },
+    {
+      "loss": 0.07781827926635743,
+      "grad_norm": 0.48375269770622253,
+      "learning_rate": 1.5222339662985273e-05,
+      "epoch": 4.283724169955886,
+      "step": 18450
+    },
+    {
+      "loss": 0.07988697052001953,
+      "grad_norm": 0.4781501889228821,
+      "learning_rate": 1.4980876727882124e-05,
+      "epoch": 4.289528674251219,
+      "step": 18475
+    },
+    {
+      "loss": 0.07793661117553712,
+      "grad_norm": 0.4536744952201843,
+      "learning_rate": 1.4741243532270098e-05,
+      "epoch": 4.295333178546552,
+      "step": 18500
+    },
+    {
+      "loss": 0.08092981338500976,
+      "grad_norm": 0.46479013562202454,
+      "learning_rate": 1.4503443323587811e-05,
+      "epoch": 4.301137682841885,
+      "step": 18525
+    },
+    {
+      "loss": 0.07952608585357666,
+      "grad_norm": 0.41237297654151917,
+      "learning_rate": 1.4267479324433867e-05,
+      "epoch": 4.306942187137219,
+      "step": 18550
+    },
+    {
+      "loss": 0.07980387210845948,
+      "grad_norm": 0.4342309236526489,
+      "learning_rate": 1.4033354732523077e-05,
+      "epoch": 4.312746691432552,
+      "step": 18575
+    },
+    {
+      "loss": 0.08078558921813965,
+      "grad_norm": 0.47663629055023193,
+      "learning_rate": 1.3801072720643125e-05,
+      "epoch": 4.318551195727885,
+      "step": 18600
+    },
+    {
+      "loss": 0.07813294887542725,
+      "grad_norm": 0.3211257755756378,
+      "learning_rate": 1.3570636436611564e-05,
+      "epoch": 4.324355700023218,
+      "step": 18625
+    },
+    {
+      "loss": 0.07818384647369385,
+      "grad_norm": 0.44311535358428955,
+      "learning_rate": 1.3342049003233212e-05,
+      "epoch": 4.330160204318551,
+      "step": 18650
+    },
+    {
+      "loss": 0.07564275741577148,
+      "grad_norm": 0.4093654155731201,
+      "learning_rate": 1.3115313518257809e-05,
+      "epoch": 4.335964708613885,
+      "step": 18675
+    },
+    {
+      "loss": 0.08086618423461914,
+      "grad_norm": 0.47484564781188965,
+      "learning_rate": 1.2890433054338085e-05,
+      "epoch": 4.341769212909218,
+      "step": 18700
+    },
+    {
+      "loss": 0.07786150455474854,
+      "grad_norm": 0.50920569896698,
+      "learning_rate": 1.2667410658987981e-05,
+      "epoch": 4.3475737172045505,
+      "step": 18725
+    },
+    {
+      "loss": 0.07455573081970215,
+      "grad_norm": 0.43661263585090637,
+      "learning_rate": 1.244624935454145e-05,
+      "epoch": 4.3533782214998835,
+      "step": 18750
+    },
+    {
+      "loss": 0.08102331161499024,
+      "grad_norm": 0.4164806604385376,
+      "learning_rate": 1.2226952138111545e-05,
+      "epoch": 4.359182725795217,
+      "step": 18775
+    },
+    {
+      "loss": 0.08130499839782715,
+      "grad_norm": 0.5355484485626221,
+      "learning_rate": 1.2009521981549736e-05,
+      "epoch": 4.36498723009055,
+      "step": 18800
+    },
+    {
+      "loss": 0.0766422700881958,
+      "grad_norm": 0.5275045037269592,
+      "learning_rate": 1.1793961831405607e-05,
+      "epoch": 4.370791734385883,
+      "step": 18825
+    },
+    {
+      "loss": 0.08015645027160645,
+      "grad_norm": 0.47182697057724,
+      "learning_rate": 1.1580274608886953e-05,
+      "epoch": 4.376596238681216,
+      "step": 18850
+    },
+    {
+      "loss": 0.0727626609802246,
+      "grad_norm": 0.34148523211479187,
+      "learning_rate": 1.1368463209820283e-05,
+      "epoch": 4.38240074297655,
+      "step": 18875
+    },
+    {
+      "loss": 0.08069220542907715,
+      "grad_norm": 0.41285601258277893,
+      "learning_rate": 1.1158530504611413e-05,
+      "epoch": 4.388205247271883,
+      "step": 18900
+    },
+    {
+      "loss": 0.07536788940429688,
+      "grad_norm": 0.41651299595832825,
+      "learning_rate": 1.0950479338206719e-05,
+      "epoch": 4.394009751567216,
+      "step": 18925
+    },
+    {
+      "loss": 0.0890130615234375,
+      "grad_norm": 0.49972784519195557,
+      "learning_rate": 1.0744312530054444e-05,
+      "epoch": 4.399814255862549,
+      "step": 18950
+    },
+    {
+      "loss": 0.0806863021850586,
+      "grad_norm": 0.382648229598999,
+      "learning_rate": 1.0540032874066639e-05,
+      "epoch": 4.405618760157883,
+      "step": 18975
+    },
+    {
+      "loss": 0.0785952615737915,
+      "grad_norm": 0.4752176105976105,
+      "learning_rate": 1.033764313858118e-05,
+      "epoch": 4.411423264453216,
+      "step": 19000
+    },
+    {
+      "eval_loss": 0.5139902830123901,
+      "eval_runtime": 2.742,
+      "eval_samples_per_second": 656.467,
+      "eval_steps_per_second": 10.576,
+      "epoch": 4.411423264453216,
+      "step": 19000
+    },
+    {
+      "loss": 0.07371102333068848,
+      "grad_norm": 0.3996540904045105,
+      "learning_rate": 1.0137146066324264e-05,
+      "epoch": 4.417227768748549,
+      "step": 19025
+    },
+    {
+      "loss": 0.07812005996704102,
+      "grad_norm": 0.4201669991016388,
+      "learning_rate": 9.938544374373354e-06,
+      "epoch": 4.423032273043882,
+      "step": 19050
+    },
+    {
+      "loss": 0.0782503080368042,
+      "grad_norm": 0.3941466808319092,
+      "learning_rate": 9.741840754120223e-06,
+      "epoch": 4.428836777339216,
+      "step": 19075
+    },
+    {
+      "loss": 0.07590310096740722,
+      "grad_norm": 0.4408641755580902,
+      "learning_rate": 9.547037871234542e-06,
+      "epoch": 4.434641281634549,
+      "step": 19100
+    },
+    {
+      "loss": 0.07825098037719727,
+      "grad_norm": 0.4821837544441223,
+      "learning_rate": 9.35413836562775e-06,
+      "epoch": 4.440445785929882,
+      "step": 19125
+    },
+    {
+      "loss": 0.07823121547698975,
+      "grad_norm": 0.4407540261745453,
+      "learning_rate": 9.16314485141722e-06,
+      "epoch": 4.446250290225215,
+      "step": 19150
+    },
+    {
+      "loss": 0.07731770515441895,
+      "grad_norm": 0.46196237206459045,
+      "learning_rate": 8.974059916891035e-06,
+      "epoch": 4.4520547945205475,
+      "step": 19175
+    },
+    {
+      "loss": 0.07914790153503418,
+      "grad_norm": 0.48064059019088745,
+      "learning_rate": 8.786886124472636e-06,
+      "epoch": 4.457859298815881,
+      "step": 19200
+    },
+    {
+      "loss": 0.08182929039001464,
+      "grad_norm": 0.470699280500412,
+      "learning_rate": 8.601626010686263e-06,
+      "epoch": 4.463663803111214,
+      "step": 19225
+    },
+    {
+      "loss": 0.08012688636779786,
+      "grad_norm": 0.5592325329780579,
+      "learning_rate": 8.418282086122535e-06,
+      "epoch": 4.469468307406547,
+      "step": 19250
+    },
+    {
+      "loss": 0.08872594833374023,
+      "grad_norm": 0.526476263999939,
+      "learning_rate": 8.236856835404476e-06,
+      "epoch": 4.47527281170188,
+      "step": 19275
+    },
+    {
+      "loss": 0.0844990348815918,
+      "grad_norm": 0.38734254240989685,
+      "learning_rate": 8.057352717153793e-06,
+      "epoch": 4.481077315997214,
+      "step": 19300
+    },
+    {
+      "loss": 0.08157651901245117,
+      "grad_norm": 0.4687935411930084,
+      "learning_rate": 7.879772163957532e-06,
+      "epoch": 4.486881820292547,
+      "step": 19325
+    },
+    {
+      "loss": 0.07577196598052978,
+      "grad_norm": 0.5619013905525208,
+      "learning_rate": 7.704117582335212e-06,
+      "epoch": 4.49268632458788,
+      "step": 19350
+    },
+    {
+      "loss": 0.07518729209899902,
+      "grad_norm": 0.4339287579059601,
+      "learning_rate": 7.530391352706061e-06,
+      "epoch": 4.498490828883213,
+      "step": 19375
+    },
+    {
+      "loss": 0.0839167022705078,
+      "grad_norm": 0.4794290363788605,
+      "learning_rate": 7.358595829356945e-06,
+      "epoch": 4.504295333178547,
+      "step": 19400
+    },
+    {
+      "loss": 0.07811093330383301,
+      "grad_norm": 0.42739877104759216,
+      "learning_rate": 7.1887333404103245e-06,
+      "epoch": 4.51009983747388,
+      "step": 19425
+    },
+    {
+      "loss": 0.07879485130310059,
+      "grad_norm": 0.4844427704811096,
+      "learning_rate": 7.0208061877927494e-06,
+      "epoch": 4.515904341769213,
+      "step": 19450
+    },
+    {
+      "loss": 0.08143006324768066,
+      "grad_norm": 0.5435075759887695,
+      "learning_rate": 6.854816647203615e-06,
+      "epoch": 4.521708846064546,
+      "step": 19475
+    },
+    {
+      "loss": 0.08163432121276855,
+      "grad_norm": 0.4060407876968384,
+      "learning_rate": 6.690766968084471e-06,
+      "epoch": 4.527513350359879,
+      "step": 19500
+    },
+    {
+      "loss": 0.07742088317871093,
+      "grad_norm": 0.45898979902267456,
+      "learning_rate": 6.528659373588329e-06,
+      "epoch": 4.533317854655213,
+      "step": 19525
+    },
+    {
+      "loss": 0.0791277027130127,
+      "grad_norm": 0.4365909993648529,
+      "learning_rate": 6.368496060549705e-06,
+      "epoch": 4.539122358950546,
+      "step": 19550
+    },
+    {
+      "loss": 0.07588704109191895,
+      "grad_norm": 0.41925954818725586,
+      "learning_rate": 6.2102791994547204e-06,
+      "epoch": 4.544926863245879,
+      "step": 19575
+    },
+    {
+      "loss": 0.08074780464172364,
+      "grad_norm": 0.4266278147697449,
+      "learning_rate": 6.054010934411868e-06,
+      "epoch": 4.550731367541212,
+      "step": 19600
+    },
+    {
+      "loss": 0.07541372776031494,
+      "grad_norm": 0.45072221755981445,
+      "learning_rate": 5.899693383122694e-06,
+      "epoch": 4.556535871836545,
+      "step": 19625
+    },
+    {
+      "loss": 0.08534663200378417,
+      "grad_norm": 0.4925159811973572,
+      "learning_rate": 5.7473286368533596e-06,
+      "epoch": 4.562340376131878,
+      "step": 19650
+    },
+    {
+      "loss": 0.08116405487060546,
+      "grad_norm": 0.48844000697135925,
+      "learning_rate": 5.596918760406127e-06,
+      "epoch": 4.568144880427211,
+      "step": 19675
+    },
+    {
+      "loss": 0.08000840187072754,
+      "grad_norm": 0.4892956614494324,
+      "learning_rate": 5.448465792091467e-06,
+      "epoch": 4.573949384722544,
+      "step": 19700
+    },
+    {
+      "loss": 0.08057500839233399,
+      "grad_norm": 0.4181472659111023,
+      "learning_rate": 5.301971743700378e-06,
+      "epoch": 4.579753889017878,
+      "step": 19725
+    },
+    {
+      "loss": 0.08139314651489257,
+      "grad_norm": 0.40270179510116577,
+      "learning_rate": 5.157438600477148e-06,
+      "epoch": 4.585558393313211,
+      "step": 19750
+    },
+    {
+      "loss": 0.07780666351318359,
+      "grad_norm": 0.45924490690231323,
+      "learning_rate": 5.0148683210925026e-06,
+      "epoch": 4.591362897608544,
+      "step": 19775
+    },
+    {
+      "loss": 0.0774199914932251,
+      "grad_norm": 0.4562431871891022,
+      "learning_rate": 4.874262837616977e-06,
+      "epoch": 4.597167401903877,
+      "step": 19800
+    },
+    {
+      "loss": 0.07871145725250245,
+      "grad_norm": 0.5946608185768127,
+      "learning_rate": 4.7356240554947745e-06,
+      "epoch": 4.602971906199211,
+      "step": 19825
+    },
+    {
+      "loss": 0.07936087131500244,
+      "grad_norm": 0.5524415969848633,
+      "learning_rate": 4.598953853517934e-06,
+      "epoch": 4.608776410494544,
+      "step": 19850
+    },
+    {
+      "loss": 0.07612611770629883,
+      "grad_norm": 0.48315760493278503,
+      "learning_rate": 4.46425408380085e-06,
+      "epoch": 4.614580914789877,
+      "step": 19875
+    },
+    {
+      "loss": 0.07877078056335449,
+      "grad_norm": 0.5121182799339294,
+      "learning_rate": 4.331526571755278e-06,
+      "epoch": 4.62038541908521,
+      "step": 19900
+    },
+    {
+      "loss": 0.07609409332275391,
+      "grad_norm": 0.4489760994911194,
+      "learning_rate": 4.200773116065437e-06,
+      "epoch": 4.626189923380544,
+      "step": 19925
+    },
+    {
+      "loss": 0.07558590412139893,
+      "grad_norm": 0.4567054808139801,
+      "learning_rate": 4.071995488663776e-06,
+      "epoch": 4.631994427675877,
+      "step": 19950
+    },
+    {
+      "loss": 0.07805031776428223,
+      "grad_norm": 0.5078436732292175,
+      "learning_rate": 3.945195434706816e-06,
+      "epoch": 4.63779893197121,
+      "step": 19975
+    },
+    {
+      "loss": 0.07685207843780517,
+      "grad_norm": 0.43479177355766296,
+      "learning_rate": 3.820374672551663e-06,
+      "epoch": 4.643603436266543,
+      "step": 20000
+    },
+    {
+      "eval_loss": 0.5150907635688782,
+      "eval_runtime": 2.7319,
+      "eval_samples_per_second": 658.893,
+      "eval_steps_per_second": 10.616,
+      "epoch": 4.643603436266543,
+      "step": 20000
+    },
+    {
+      "loss": 0.0795011568069458,
+      "grad_norm": 0.476205438375473,
+      "learning_rate": 3.6975348937326454e-06,
+      "epoch": 4.649407940561876,
+      "step": 20025
+    },
+    {
+      "loss": 0.08185251235961914,
+      "grad_norm": 0.4407123327255249,
+      "learning_rate": 3.576677762938318e-06,
+      "epoch": 4.655212444857209,
+      "step": 20050
+    },
+    {
+      "loss": 0.08033641815185547,
+      "grad_norm": 0.4826345145702362,
+      "learning_rate": 3.4578049179890098e-06,
+      "epoch": 4.661016949152542,
+      "step": 20075
+    },
+    {
+      "loss": 0.0772846508026123,
+      "grad_norm": 0.4939717650413513,
+      "learning_rate": 3.3409179698146106e-06,
+      "epoch": 4.666821453447875,
+      "step": 20100
+    },
+    {
+      "loss": 0.07670364379882813,
+      "grad_norm": 0.4766358435153961,
+      "learning_rate": 3.226018502432687e-06,
+      "epoch": 4.672625957743209,
+      "step": 20125
+    },
+    {
+      "loss": 0.07934382438659668,
+      "grad_norm": 0.41519227623939514,
+      "learning_rate": 3.1131080729270673e-06,
+      "epoch": 4.678430462038542,
+      "step": 20150
+    },
+    {
+      "loss": 0.07627038955688477,
+      "grad_norm": 0.4669382572174072,
+      "learning_rate": 3.002188211426726e-06,
+      "epoch": 4.684234966333875,
+      "step": 20175
+    },
+    {
+      "loss": 0.07136449337005615,
+      "grad_norm": 0.389437735080719,
+      "learning_rate": 2.8932604210850474e-06,
+      "epoch": 4.690039470629208,
+      "step": 20200
+    },
+    {
+      "loss": 0.07769662857055665,
+      "grad_norm": 0.4156358540058136,
+      "learning_rate": 2.786326178059428e-06,
+      "epoch": 4.695843974924541,
+      "step": 20225
+    },
+    {
+      "loss": 0.07760199546813965,
+      "grad_norm": 0.5581144690513611,
+      "learning_rate": 2.681386931491292e-06,
+      "epoch": 4.701648479219875,
+      "step": 20250
+    },
+    {
+      "loss": 0.07693316459655762,
+      "grad_norm": 0.4294784665107727,
+      "learning_rate": 2.578444103486521e-06,
+      "epoch": 4.707452983515208,
+      "step": 20275
+    },
+    {
+      "loss": 0.08216179847717285,
+      "grad_norm": 0.5029132962226868,
+      "learning_rate": 2.4774990890960743e-06,
+      "epoch": 4.713257487810541,
+      "step": 20300
+    },
+    {
+      "loss": 0.0754340934753418,
+      "grad_norm": 0.4067029356956482,
+      "learning_rate": 2.3785532562971e-06,
+      "epoch": 4.719061992105874,
+      "step": 20325
+    },
+    {
+      "loss": 0.07610186100006104,
+      "grad_norm": 0.42405033111572266,
+      "learning_rate": 2.2816079459744676e-06,
+      "epoch": 4.724866496401208,
+      "step": 20350
+    },
+    {
+      "loss": 0.0798477840423584,
+      "grad_norm": 0.4587237536907196,
+      "learning_rate": 2.1866644719025493e-06,
+      "epoch": 4.730671000696541,
+      "step": 20375
+    },
+    {
+      "loss": 0.07270728588104249,
+      "grad_norm": 0.4628477096557617,
+      "learning_rate": 2.0937241207274025e-06,
+      "epoch": 4.736475504991874,
+      "step": 20400
+    },
+    {
+      "loss": 0.07922406196594238,
+      "grad_norm": 0.5066894888877869,
+      "learning_rate": 2.0027881519493804e-06,
+      "epoch": 4.742280009287207,
+      "step": 20425
+    },
+    {
+      "loss": 0.07608590602874755,
+      "grad_norm": 0.40147361159324646,
+      "learning_rate": 1.913857797905999e-06,
+      "epoch": 4.7480845135825405,
+      "step": 20450
+    },
+    {
+      "loss": 0.07383220672607421,
+      "grad_norm": 0.3422751724720001,
+      "learning_rate": 1.8269342637552643e-06,
+      "epoch": 4.753889017877873,
+      "step": 20475
+    },
+    {
+      "loss": 0.0794384765625,
+      "grad_norm": 0.5053935050964355,
+      "learning_rate": 1.742018727459421e-06,
+      "epoch": 4.759693522173206,
+      "step": 20500
+    },
+    {
+      "loss": 0.07553751945495606,
+      "grad_norm": 0.4747840166091919,
+      "learning_rate": 1.6591123397688477e-06,
+      "epoch": 4.765498026468539,
+      "step": 20525
+    },
+    {
+      "loss": 0.07691112995147704,
+      "grad_norm": 0.4658749997615814,
+      "learning_rate": 1.5782162242065521e-06,
+      "epoch": 4.771302530763872,
+      "step": 20550
+    },
+    {
+      "loss": 0.0758073377609253,
+      "grad_norm": 0.4638631343841553,
+      "learning_rate": 1.4993314770529342e-06,
+      "epoch": 4.777107035059206,
+      "step": 20575
+    },
+    {
+      "loss": 0.07755953788757325,
+      "grad_norm": 0.4154756963253021,
+      "learning_rate": 1.4224591673308804e-06,
+      "epoch": 4.782911539354539,
+      "step": 20600
+    },
+    {
+      "loss": 0.07841875076293946,
+      "grad_norm": 0.5114679932594299,
+      "learning_rate": 1.3476003367913934e-06,
+      "epoch": 4.788716043649872,
+      "step": 20625
+    },
+    {
+      "loss": 0.08010381698608399,
+      "grad_norm": 0.4619429111480713,
+      "learning_rate": 1.2747559998993184e-06,
+      "epoch": 4.794520547945205,
+      "step": 20650
+    },
+    {
+      "loss": 0.08214847564697265,
+      "grad_norm": 0.4827802777290344,
+      "learning_rate": 1.2039271438197385e-06,
+      "epoch": 4.800325052240539,
+      "step": 20675
+    },
+    {
+      "loss": 0.07512083530426025,
+      "grad_norm": 0.4229465425014496,
+      "learning_rate": 1.1351147284044682e-06,
+      "epoch": 4.806129556535872,
+      "step": 20700
+    },
+    {
+      "loss": 0.07927950382232667,
+      "grad_norm": 0.4403354525566101,
+      "learning_rate": 1.0683196861791642e-06,
+      "epoch": 4.811934060831205,
+      "step": 20725
+    },
+    {
+      "loss": 0.07306070804595947,
+      "grad_norm": 0.4678843021392822,
+      "learning_rate": 1.0035429223306023e-06,
+      "epoch": 4.817738565126538,
+      "step": 20750
+    },
+    {
+      "loss": 0.08375411987304687,
+      "grad_norm": 0.5157707929611206,
+      "learning_rate": 9.407853146944866e-07,
+      "epoch": 4.823543069421872,
+      "step": 20775
+    },
+    {
+      "loss": 0.07699784278869629,
+      "grad_norm": 0.4578299820423126,
+      "learning_rate": 8.800477137434425e-07,
+      "epoch": 4.829347573717205,
+      "step": 20800
+    },
+    {
+      "loss": 0.08535214424133301,
+      "grad_norm": 0.46220582723617554,
+      "learning_rate": 8.213309425756098e-07,
+      "epoch": 4.835152078012538,
+      "step": 20825
+    },
+    {
+      "loss": 0.07462476253509522,
+      "grad_norm": 0.389831006526947,
+      "learning_rate": 7.646357969034178e-07,
+      "epoch": 4.840956582307871,
+      "step": 20850
+    },
+    {
+      "loss": 0.07763785362243653,
+      "grad_norm": 0.45414626598358154,
+      "learning_rate": 7.099630450428273e-07,
+      "epoch": 4.846761086603204,
+      "step": 20875
+    },
+    {
+      "loss": 0.08067751884460449,
+      "grad_norm": 0.5022978782653809,
+      "learning_rate": 6.573134279028891e-07,
+      "epoch": 4.8525655908985375,
+      "step": 20900
+    },
+    {
+      "loss": 0.07359580039978027,
+      "grad_norm": 0.4772122800350189,
+      "learning_rate": 6.066876589757852e-07,
+      "epoch": 4.85837009519387,
+      "step": 20925
+    },
+    {
+      "loss": 0.07639682292938232,
+      "grad_norm": 0.39041006565093994,
+      "learning_rate": 5.580864243270367e-07,
+      "epoch": 4.864174599489203,
+      "step": 20950
+    },
+    {
+      "loss": 0.07517142772674561,
+      "grad_norm": 0.4305885434150696,
+      "learning_rate": 5.115103825862943e-07,
+      "epoch": 4.869979103784537,
+      "step": 20975
+    },
+    {
+      "loss": 0.07514492988586426,
+      "grad_norm": 0.4646373987197876,
+      "learning_rate": 4.6696016493839583e-07,
+      "epoch": 4.87578360807987,
+      "step": 21000
+    },
+    {
+      "eval_loss": 0.5153464674949646,
+      "eval_runtime": 2.7486,
+      "eval_samples_per_second": 654.882,
+      "eval_steps_per_second": 10.551,
+      "epoch": 4.87578360807987,
+      "step": 21000
+    },
+    {
+      "loss": 0.07574666023254395,
+      "grad_norm": 0.4699035584926605,
+      "learning_rate": 4.2443637511482277e-07,
+      "epoch": 4.881588112375203,
+      "step": 21025
+    },
+    {
+      "loss": 0.0750859546661377,
+      "grad_norm": 0.4289429783821106,
+      "learning_rate": 3.839395893854569e-07,
+      "epoch": 4.887392616670536,
+      "step": 21050
+    },
+    {
+      "loss": 0.08001879692077636,
+      "grad_norm": 0.44778937101364136,
+      "learning_rate": 3.454703565508532e-07,
+      "epoch": 4.893197120965869,
+      "step": 21075
+    },
+    {
+      "loss": 0.07488130569458008,
+      "grad_norm": 0.43965455889701843,
+      "learning_rate": 3.090291979347459e-07,
+      "epoch": 4.899001625261203,
+      "step": 21100
+    },
+    {
+      "loss": 0.07950814247131348,
+      "grad_norm": 0.48464298248291016,
+      "learning_rate": 2.7461660737700397e-07,
+      "epoch": 4.904806129556536,
+      "step": 21125
+    },
+    {
+      "loss": 0.07310739994049072,
+      "grad_norm": 0.4172207713127136,
+      "learning_rate": 2.422330512269699e-07,
+      "epoch": 4.910610633851869,
+      "step": 21150
+    },
+    {
+      "loss": 0.07573895931243896,
+      "grad_norm": 0.4198155999183655,
+      "learning_rate": 2.1187896833706475e-07,
+      "epoch": 4.916415138147202,
+      "step": 21175
+    },
+    {
+      "loss": 0.07721238613128661,
+      "grad_norm": 0.42362892627716064,
+      "learning_rate": 1.835547700568929e-07,
+      "epoch": 4.922219642442536,
+      "step": 21200
+    },
+    {
+      "loss": 0.07186219692230225,
+      "grad_norm": 0.4425536096096039,
+      "learning_rate": 1.572608402276798e-07,
+      "epoch": 4.928024146737869,
+      "step": 21225
+    },
+    {
+      "loss": 0.07345245361328125,
+      "grad_norm": 0.40979814529418945,
+      "learning_rate": 1.3299753517704293e-07,
+      "epoch": 4.933828651033202,
+      "step": 21250
+    },
+    {
+      "loss": 0.07667166709899903,
+      "grad_norm": 0.45127779245376587,
+      "learning_rate": 1.1076518371416209e-07,
+      "epoch": 4.939633155328535,
+      "step": 21275
+    },
+    {
+      "loss": 0.07520007610321044,
+      "grad_norm": 0.42959147691726685,
+      "learning_rate": 9.056408712531659e-08,
+      "epoch": 4.9454376596238685,
+      "step": 21300
+    },
+    {
+      "loss": 0.0765018892288208,
+      "grad_norm": 0.4762932360172272,
+      "learning_rate": 7.239451916985494e-08,
+      "epoch": 4.9512421639192015,
+      "step": 21325
+    },
+    {
+      "loss": 0.08274571418762207,
+      "grad_norm": 0.49068811535835266,
+      "learning_rate": 5.6256726076397975e-08,
+      "epoch": 4.957046668214534,
+      "step": 21350
+    },
+    {
+      "loss": 0.07985799312591553,
+      "grad_norm": 0.5473991632461548,
+      "learning_rate": 4.215092653960806e-08,
+      "epoch": 4.962851172509867,
+      "step": 21375
+    },
+    {
+      "loss": 0.0765773344039917,
+      "grad_norm": 0.4471685290336609,
+      "learning_rate": 3.007731171712491e-08,
+      "epoch": 4.9686556768052,
+      "step": 21400
+    },
+    {
+      "loss": 0.07661473751068115,
+      "grad_norm": 0.43240880966186523,
+      "learning_rate": 2.0036045227000975e-08,
+      "epoch": 4.974460181100534,
+      "step": 21425
+    },
+    {
+      "loss": 0.08009207725524903,
+      "grad_norm": 0.47322556376457214,
+      "learning_rate": 1.2027263145569787e-08,
+      "epoch": 4.980264685395867,
+      "step": 21450
+    },
+    {
+      "loss": 0.08029013633728027,
+      "grad_norm": 0.519874095916748,
+      "learning_rate": 6.0510740054808825e-09,
+      "epoch": 4.9860691896912,
+      "step": 21475
+    },
+    {
+      "loss": 0.07674485683441162,
+      "grad_norm": 0.4265105724334717,
+      "learning_rate": 2.1075587942842586e-09,
+      "epoch": 4.991873693986534,
+      "step": 21500
+    },
+    {
+      "loss": 0.08361578941345214,
+      "grad_norm": 0.564140796661377,
+      "learning_rate": 1.9677095334791336e-10,
+      "epoch": 4.997678198281867,
+      "step": 21525
+    },
+    {
+      "eval_loss": 0.5153905153274536,
+      "eval_runtime": 2.7051,
+      "eval_samples_per_second": 665.408,
+      "eval_steps_per_second": 10.72,
+      "epoch": 5.0,
+      "step": 21535
+    },
+    {
+      "train_runtime": 840.6046,
+      "train_samples_per_second": 1639.528,
+      "train_steps_per_second": 25.618,
+      "total_flos": 8.05529052384768e+16,
+      "train_loss": 0.012863880613161844,
+      "epoch": 5.0,
+      "step": 21535
+    },
+    {
+      "eval_loss": 0.5153905153274536,
+      "eval_runtime": 2.7823,
+      "eval_samples_per_second": 646.94,
+      "eval_steps_per_second": 10.423,
+      "epoch": 5.0,
+      "step": 21535
+    }
+  ]
+}
\ No newline at end of file