diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 3.68600129617628,
+  "epoch": 4.0,
   "eval_steps": 500,
-  "global_step": 45500,
+  "global_step": 49376,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -318507,6 +318507,27138 @@
       "learning_rate": 1.7444529456771233e-05,
       "loss": 0.2034,
       "step": 45500
+    },
+    {
+      "epoch": 3.6860823071937783,
+      "grad_norm": 0.08332670480012894,
+      "learning_rate": 1.7440028804176606e-05,
+      "loss": 0.2416,
+      "step": 45501
+    },
+    {
+      "epoch": 3.6861633182112765,
+      "grad_norm": 0.07239510118961334,
+      "learning_rate": 1.743552815158198e-05,
+      "loss": 0.1988,
+      "step": 45502
+    },
+    {
+      "epoch": 3.686244329228775,
+      "grad_norm": 0.08834824711084366,
+      "learning_rate": 1.7431027498987353e-05,
+      "loss": 0.234,
+      "step": 45503
+    },
+    {
+      "epoch": 3.6863253402462735,
+      "grad_norm": 0.06722290068864822,
+      "learning_rate": 1.7426526846392727e-05,
+      "loss": 0.236,
+      "step": 45504
+    },
+    {
+      "epoch": 3.6864063512637717,
+      "grad_norm": 0.07778163254261017,
+      "learning_rate": 1.74220261937981e-05,
+      "loss": 0.2244,
+      "step": 45505
+    },
+    {
+      "epoch": 3.6864873622812704,
+      "grad_norm": 0.06533099710941315,
+      "learning_rate": 1.7417525541203474e-05,
+      "loss": 0.181,
+      "step": 45506
+    },
+    {
+      "epoch": 3.6865683732987686,
+      "grad_norm": 0.07727663218975067,
+      "learning_rate": 1.741302488860885e-05,
+      "loss": 0.2273,
+      "step": 45507
+    },
+    {
+      "epoch": 3.686649384316267,
+      "grad_norm": 0.08147032558917999,
+      "learning_rate": 1.7408524236014225e-05,
+      "loss": 0.2455,
+      "step": 45508
+    },
+    {
+      "epoch": 3.6867303953337656,
+      "grad_norm": 0.07328832149505615,
+      "learning_rate": 1.7404023583419595e-05,
+      "loss": 0.2112,
+      "step": 45509
+    },
+    {
+      "epoch": 3.686811406351264,
+      "grad_norm": 0.08036161959171295,
+      "learning_rate": 1.7399522930824972e-05,
+      "loss": 0.2313,
+      "step": 45510
+    },
+    {
+      "epoch": 3.686892417368762,
+      "grad_norm": 0.060651037842035294,
+      "learning_rate": 1.7395022278230346e-05,
+      "loss": 0.181,
+      "step": 45511
+    },
+    {
+      "epoch": 3.6869734283862607,
+      "grad_norm": 0.06498783826828003,
+      "learning_rate": 1.7390521625635716e-05,
+      "loss": 0.2286,
+      "step": 45512
+    },
+    {
+      "epoch": 3.687054439403759,
+      "grad_norm": 0.07742474973201752,
+      "learning_rate": 1.7386020973041093e-05,
+      "loss": 0.2486,
+      "step": 45513
+    },
+    {
+      "epoch": 3.6871354504212572,
+      "grad_norm": 0.08565546572208405,
+      "learning_rate": 1.7381520320446466e-05,
+      "loss": 0.2706,
+      "step": 45514
+    },
+    {
+      "epoch": 3.687216461438756,
+      "grad_norm": 0.05786954611539841,
+      "learning_rate": 1.737701966785184e-05,
+      "loss": 0.2004,
+      "step": 45515
+    },
+    {
+      "epoch": 3.687297472456254,
+      "grad_norm": 0.07234077900648117,
+      "learning_rate": 1.7372519015257214e-05,
+      "loss": 0.2385,
+      "step": 45516
+    },
+    {
+      "epoch": 3.6873784834737524,
+      "grad_norm": 0.06998328864574432,
+      "learning_rate": 1.7368018362662587e-05,
+      "loss": 0.2232,
+      "step": 45517
+    },
+    {
+      "epoch": 3.6874594944912507,
+      "grad_norm": 0.08726705610752106,
+      "learning_rate": 1.736351771006796e-05,
+      "loss": 0.2643,
+      "step": 45518
+    },
+    {
+      "epoch": 3.6875405055087493,
+      "grad_norm": 0.06822450459003448,
+      "learning_rate": 1.7359017057473334e-05,
+      "loss": 0.1948,
+      "step": 45519
+    },
+    {
+      "epoch": 3.6876215165262476,
+      "grad_norm": 0.08558908849954605,
+      "learning_rate": 1.7354516404878708e-05,
+      "loss": 0.214,
+      "step": 45520
+    },
+    {
+      "epoch": 3.687702527543746,
+      "grad_norm": 0.0689425840973854,
+      "learning_rate": 1.735001575228408e-05,
+      "loss": 0.213,
+      "step": 45521
+    },
+    {
+      "epoch": 3.687783538561244,
+      "grad_norm": 0.06478188931941986,
+      "learning_rate": 1.7345515099689455e-05,
+      "loss": 0.2408,
+      "step": 45522
+    },
+    {
+      "epoch": 3.6878645495787428,
+      "grad_norm": 0.06854352355003357,
+      "learning_rate": 1.734101444709483e-05,
+      "loss": 0.2573,
+      "step": 45523
+    },
+    {
+      "epoch": 3.687945560596241,
+      "grad_norm": 0.09567511081695557,
+      "learning_rate": 1.7336513794500202e-05,
+      "loss": 0.2461,
+      "step": 45524
+    },
+    {
+      "epoch": 3.6880265716137393,
+      "grad_norm": 0.08633017539978027,
+      "learning_rate": 1.7332013141905576e-05,
+      "loss": 0.2201,
+      "step": 45525
+    },
+    {
+      "epoch": 3.688107582631238,
+      "grad_norm": 0.07063650339841843,
+      "learning_rate": 1.7327512489310953e-05,
+      "loss": 0.1685,
+      "step": 45526
+    },
+    {
+      "epoch": 3.688188593648736,
+      "grad_norm": 0.07147509604692459,
+      "learning_rate": 1.7323011836716323e-05,
+      "loss": 0.2507,
+      "step": 45527
+    },
+    {
+      "epoch": 3.6882696046662344,
+      "grad_norm": 0.08037501573562622,
+      "learning_rate": 1.73185111841217e-05,
+      "loss": 0.2337,
+      "step": 45528
+    },
+    {
+      "epoch": 3.688350615683733,
+      "grad_norm": 0.06710902601480484,
+      "learning_rate": 1.7314010531527074e-05,
+      "loss": 0.2117,
+      "step": 45529
+    },
+    {
+      "epoch": 3.6884316267012314,
+      "grad_norm": 0.07238079607486725,
+      "learning_rate": 1.7309509878932444e-05,
+      "loss": 0.2188,
+      "step": 45530
+    },
+    {
+      "epoch": 3.6885126377187296,
+      "grad_norm": 0.07218148559331894,
+      "learning_rate": 1.730500922633782e-05,
+      "loss": 0.227,
+      "step": 45531
+    },
+    {
+      "epoch": 3.6885936487362283,
+      "grad_norm": 0.09336385875940323,
+      "learning_rate": 1.7300508573743195e-05,
+      "loss": 0.2507,
+      "step": 45532
+    },
+    {
+      "epoch": 3.6886746597537265,
+      "grad_norm": 0.08338413387537003,
+      "learning_rate": 1.7296007921148568e-05,
+      "loss": 0.2168,
+      "step": 45533
+    },
+    {
+      "epoch": 3.688755670771225,
+      "grad_norm": 0.07480164617300034,
+      "learning_rate": 1.7291507268553942e-05,
+      "loss": 0.2406,
+      "step": 45534
+    },
+    {
+      "epoch": 3.6888366817887235,
+      "grad_norm": 0.061553705483675,
+      "learning_rate": 1.7287006615959315e-05,
+      "loss": 0.1878,
+      "step": 45535
+    },
+    {
+      "epoch": 3.6889176928062217,
+      "grad_norm": 0.08677572011947632,
+      "learning_rate": 1.728250596336469e-05,
+      "loss": 0.226,
+      "step": 45536
+    },
+    {
+      "epoch": 3.68899870382372,
+      "grad_norm": 0.096988245844841,
+      "learning_rate": 1.7278005310770063e-05,
+      "loss": 0.289,
+      "step": 45537
+    },
+    {
+      "epoch": 3.6890797148412187,
+      "grad_norm": 0.06231336668133736,
+      "learning_rate": 1.7273504658175436e-05,
+      "loss": 0.2039,
+      "step": 45538
+    },
+    {
+      "epoch": 3.689160725858717,
+      "grad_norm": 0.07273116707801819,
+      "learning_rate": 1.726900400558081e-05,
+      "loss": 0.2481,
+      "step": 45539
+    },
+    {
+      "epoch": 3.689241736876215,
+      "grad_norm": 0.06380755454301834,
+      "learning_rate": 1.7264503352986183e-05,
+      "loss": 0.2278,
+      "step": 45540
+    },
+    {
+      "epoch": 3.6893227478937134,
+      "grad_norm": 0.07865901291370392,
+      "learning_rate": 1.726000270039156e-05,
+      "loss": 0.237,
+      "step": 45541
+    },
+    {
+      "epoch": 3.689403758911212,
+      "grad_norm": 0.08549056202173233,
+      "learning_rate": 1.725550204779693e-05,
+      "loss": 0.196,
+      "step": 45542
+    },
+    {
+      "epoch": 3.6894847699287103,
+      "grad_norm": 0.07149454206228256,
+      "learning_rate": 1.7251001395202304e-05,
+      "loss": 0.2474,
+      "step": 45543
+    },
+    {
+      "epoch": 3.6895657809462086,
+      "grad_norm": 0.07113026827573776,
+      "learning_rate": 1.724650074260768e-05,
+      "loss": 0.2095,
+      "step": 45544
+    },
+    {
+      "epoch": 3.689646791963707,
+      "grad_norm": 0.07473193854093552,
+      "learning_rate": 1.724200009001305e-05,
+      "loss": 0.2236,
+      "step": 45545
+    },
+    {
+      "epoch": 3.6897278029812055,
+      "grad_norm": 0.08211532980203629,
+      "learning_rate": 1.723749943741843e-05,
+      "loss": 0.2621,
+      "step": 45546
+    },
+    {
+      "epoch": 3.6898088139987038,
+      "grad_norm": 0.07540342956781387,
+      "learning_rate": 1.7232998784823802e-05,
+      "loss": 0.246,
+      "step": 45547
+    },
+    {
+      "epoch": 3.689889825016202,
+      "grad_norm": 0.06144927814602852,
+      "learning_rate": 1.7228498132229172e-05,
+      "loss": 0.1974,
+      "step": 45548
+    },
+    {
+      "epoch": 3.6899708360337007,
+      "grad_norm": 0.06908820569515228,
+      "learning_rate": 1.722399747963455e-05,
+      "loss": 0.2061,
+      "step": 45549
+    },
+    {
+      "epoch": 3.690051847051199,
+      "grad_norm": 0.07357897609472275,
+      "learning_rate": 1.7219496827039923e-05,
+      "loss": 0.1908,
+      "step": 45550
+    },
+    {
+      "epoch": 3.690132858068697,
+      "grad_norm": 0.07140400260686874,
+      "learning_rate": 1.7214996174445293e-05,
+      "loss": 0.2132,
+      "step": 45551
+    },
+    {
+      "epoch": 3.690213869086196,
+      "grad_norm": 0.07593827694654465,
+      "learning_rate": 1.721049552185067e-05,
+      "loss": 0.2472,
+      "step": 45552
+    },
+    {
+      "epoch": 3.690294880103694,
+      "grad_norm": 0.07725334167480469,
+      "learning_rate": 1.7205994869256044e-05,
+      "loss": 0.2395,
+      "step": 45553
+    },
+    {
+      "epoch": 3.6903758911211924,
+      "grad_norm": 0.08335854858160019,
+      "learning_rate": 1.7201494216661417e-05,
+      "loss": 0.2146,
+      "step": 45554
+    },
+    {
+      "epoch": 3.690456902138691,
+      "grad_norm": 0.06782727688550949,
+      "learning_rate": 1.719699356406679e-05,
+      "loss": 0.2199,
+      "step": 45555
+    },
+    {
+      "epoch": 3.6905379131561893,
+      "grad_norm": 0.08147486299276352,
+      "learning_rate": 1.7192492911472164e-05,
+      "loss": 0.2419,
+      "step": 45556
+    },
+    {
+      "epoch": 3.6906189241736875,
+      "grad_norm": 0.08341187983751297,
+      "learning_rate": 1.7187992258877538e-05,
+      "loss": 0.2484,
+      "step": 45557
+    },
+    {
+      "epoch": 3.690699935191186,
+      "grad_norm": 0.07176518440246582,
+      "learning_rate": 1.718349160628291e-05,
+      "loss": 0.2306,
+      "step": 45558
+    },
+    {
+      "epoch": 3.6907809462086845,
+      "grad_norm": 0.071227066218853,
+      "learning_rate": 1.7178990953688285e-05,
+      "loss": 0.2252,
+      "step": 45559
+    },
+    {
+      "epoch": 3.6908619572261827,
+      "grad_norm": 0.06585358828306198,
+      "learning_rate": 1.717449030109366e-05,
+      "loss": 0.2561,
+      "step": 45560
+    },
+    {
+      "epoch": 3.6909429682436814,
+      "grad_norm": 0.10919830203056335,
+      "learning_rate": 1.7169989648499032e-05,
+      "loss": 0.2707,
+      "step": 45561
+    },
+    {
+      "epoch": 3.6910239792611796,
+      "grad_norm": 0.08778324723243713,
+      "learning_rate": 1.716548899590441e-05,
+      "loss": 0.2543,
+      "step": 45562
+    },
+    {
+      "epoch": 3.691104990278678,
+      "grad_norm": 0.06133211776614189,
+      "learning_rate": 1.716098834330978e-05,
+      "loss": 0.2007,
+      "step": 45563
+    },
+    {
+      "epoch": 3.691186001296176,
+      "grad_norm": 0.06192644312977791,
+      "learning_rate": 1.7156487690715153e-05,
+      "loss": 0.2036,
+      "step": 45564
+    },
+    {
+      "epoch": 3.691267012313675,
+      "grad_norm": 0.07539551705121994,
+      "learning_rate": 1.715198703812053e-05,
+      "loss": 0.2636,
+      "step": 45565
+    },
+    {
+      "epoch": 3.691348023331173,
+      "grad_norm": 0.06967008113861084,
+      "learning_rate": 1.71474863855259e-05,
+      "loss": 0.2439,
+      "step": 45566
+    },
+    {
+      "epoch": 3.6914290343486713,
+      "grad_norm": 0.08135908842086792,
+      "learning_rate": 1.7142985732931278e-05,
+      "loss": 0.2111,
+      "step": 45567
+    },
+    {
+      "epoch": 3.6915100453661696,
+      "grad_norm": 0.07239343971014023,
+      "learning_rate": 1.713848508033665e-05,
+      "loss": 0.235,
+      "step": 45568
+    },
+    {
+      "epoch": 3.6915910563836682,
+      "grad_norm": 0.07607700675725937,
+      "learning_rate": 1.713398442774202e-05,
+      "loss": 0.2198,
+      "step": 45569
+    },
+    {
+      "epoch": 3.6916720674011665,
+      "grad_norm": 0.0712643638253212,
+      "learning_rate": 1.71294837751474e-05,
+      "loss": 0.2369,
+      "step": 45570
+    },
+    {
+      "epoch": 3.6917530784186647,
+      "grad_norm": 0.06370777636766434,
+      "learning_rate": 1.7124983122552772e-05,
+      "loss": 0.1989,
+      "step": 45571
+    },
+    {
+      "epoch": 3.6918340894361634,
+      "grad_norm": 0.06736060231924057,
+      "learning_rate": 1.7120482469958142e-05,
+      "loss": 0.2148,
+      "step": 45572
+    },
+    {
+      "epoch": 3.6919151004536617,
+      "grad_norm": 0.05656532570719719,
+      "learning_rate": 1.711598181736352e-05,
+      "loss": 0.1881,
+      "step": 45573
+    },
+    {
+      "epoch": 3.69199611147116,
+      "grad_norm": 0.07122478634119034,
+      "learning_rate": 1.7111481164768893e-05,
+      "loss": 0.2198,
+      "step": 45574
+    },
+    {
+      "epoch": 3.6920771224886586,
+      "grad_norm": 0.08074306696653366,
+      "learning_rate": 1.7106980512174266e-05,
+      "loss": 0.2857,
+      "step": 45575
+    },
+    {
+      "epoch": 3.692158133506157,
+      "grad_norm": 0.07378386706113815,
+      "learning_rate": 1.710247985957964e-05,
+      "loss": 0.203,
+      "step": 45576
+    },
+    {
+      "epoch": 3.692239144523655,
+      "grad_norm": 0.06506312638521194,
+      "learning_rate": 1.7097979206985014e-05,
+      "loss": 0.2338,
+      "step": 45577
+    },
+    {
+      "epoch": 3.692320155541154,
+      "grad_norm": 0.0626230537891388,
+      "learning_rate": 1.7093478554390387e-05,
+      "loss": 0.2434,
+      "step": 45578
+    },
+    {
+      "epoch": 3.692401166558652,
+      "grad_norm": 0.09149368107318878,
+      "learning_rate": 1.708897790179576e-05,
+      "loss": 0.2593,
+      "step": 45579
+    },
+    {
+      "epoch": 3.6924821775761503,
+      "grad_norm": 0.07215756922960281,
+      "learning_rate": 1.7084477249201138e-05,
+      "loss": 0.219,
+      "step": 45580
+    },
+    {
+      "epoch": 3.692563188593649,
+      "grad_norm": 0.07592841237783432,
+      "learning_rate": 1.7079976596606508e-05,
+      "loss": 0.2666,
+      "step": 45581
+    },
+    {
+      "epoch": 3.692644199611147,
+      "grad_norm": 0.0961163192987442,
+      "learning_rate": 1.707547594401188e-05,
+      "loss": 0.2635,
+      "step": 45582
+    },
+    {
+      "epoch": 3.6927252106286454,
+      "grad_norm": 0.08747298270463943,
+      "learning_rate": 1.707097529141726e-05,
+      "loss": 0.233,
+      "step": 45583
+    },
+    {
+      "epoch": 3.692806221646144,
+      "grad_norm": 0.08068632334470749,
+      "learning_rate": 1.706647463882263e-05,
+      "loss": 0.2191,
+      "step": 45584
+    },
+    {
+      "epoch": 3.6928872326636424,
+      "grad_norm": 0.07764396071434021,
+      "learning_rate": 1.7061973986228002e-05,
+      "loss": 0.2235,
+      "step": 45585
+    },
+    {
+      "epoch": 3.6929682436811406,
+      "grad_norm": 0.07049860060214996,
+      "learning_rate": 1.705747333363338e-05,
+      "loss": 0.2229,
+      "step": 45586
+    },
+    {
+      "epoch": 3.693049254698639,
+      "grad_norm": 0.06508810073137283,
+      "learning_rate": 1.705297268103875e-05,
+      "loss": 0.2215,
+      "step": 45587
+    },
+    {
+      "epoch": 3.693130265716137,
+      "grad_norm": 0.06824822723865509,
+      "learning_rate": 1.7048472028444127e-05,
+      "loss": 0.1971,
+      "step": 45588
+    },
+    {
+      "epoch": 3.693211276733636,
+      "grad_norm": 0.07630736380815506,
+      "learning_rate": 1.70439713758495e-05,
+      "loss": 0.2508,
+      "step": 45589
+    },
+    {
+      "epoch": 3.693292287751134,
+      "grad_norm": 0.06600065529346466,
+      "learning_rate": 1.703947072325487e-05,
+      "loss": 0.2006,
+      "step": 45590
+    },
+    {
+      "epoch": 3.6933732987686323,
+      "grad_norm": 0.0757368728518486,
+      "learning_rate": 1.7034970070660247e-05,
+      "loss": 0.235,
+      "step": 45591
+    },
+    {
+      "epoch": 3.693454309786131,
+      "grad_norm": 0.0807269960641861,
+      "learning_rate": 1.703046941806562e-05,
+      "loss": 0.2203,
+      "step": 45592
+    },
+    {
+      "epoch": 3.6935353208036292,
+      "grad_norm": 0.0880458652973175,
+      "learning_rate": 1.7025968765470995e-05,
+      "loss": 0.1987,
+      "step": 45593
+    },
+    {
+      "epoch": 3.6936163318211275,
+      "grad_norm": 0.07349114865064621,
+      "learning_rate": 1.7021468112876368e-05,
+      "loss": 0.242,
+      "step": 45594
+    },
+    {
+      "epoch": 3.693697342838626,
+      "grad_norm": 0.07621750235557556,
+      "learning_rate": 1.7016967460281742e-05,
+      "loss": 0.2247,
+      "step": 45595
+    },
+    {
+      "epoch": 3.6937783538561244,
+      "grad_norm": 0.07071096450090408,
+      "learning_rate": 1.7012466807687115e-05,
+      "loss": 0.2282,
+      "step": 45596
+    },
+    {
+      "epoch": 3.6938593648736227,
+      "grad_norm": 0.05581112951040268,
+      "learning_rate": 1.700796615509249e-05,
+      "loss": 0.196,
+      "step": 45597
+    },
+    {
+      "epoch": 3.6939403758911213,
+      "grad_norm": 0.08093393594026566,
+      "learning_rate": 1.7003465502497863e-05,
+      "loss": 0.2417,
+      "step": 45598
+    },
+    {
+      "epoch": 3.6940213869086196,
+      "grad_norm": 0.07044923305511475,
+      "learning_rate": 1.6998964849903236e-05,
+      "loss": 0.2366,
+      "step": 45599
+    },
+    {
+      "epoch": 3.694102397926118,
+      "grad_norm": 0.0671868696808815,
+      "learning_rate": 1.699446419730861e-05,
+      "loss": 0.1992,
+      "step": 45600
+    },
+    {
+      "epoch": 3.6941834089436165,
+      "grad_norm": 0.06780609488487244,
+      "learning_rate": 1.6989963544713987e-05,
+      "loss": 0.1734,
+      "step": 45601
+    },
+    {
+      "epoch": 3.6942644199611148,
+      "grad_norm": 0.07801494002342224,
+      "learning_rate": 1.6985462892119357e-05,
+      "loss": 0.2503,
+      "step": 45602
+    },
+    {
+      "epoch": 3.694345430978613,
+      "grad_norm": 0.07000548392534256,
+      "learning_rate": 1.698096223952473e-05,
+      "loss": 0.2211,
+      "step": 45603
+    },
+    {
+      "epoch": 3.6944264419961117,
+      "grad_norm": 0.09492857754230499,
+      "learning_rate": 1.6976461586930108e-05,
+      "loss": 0.2736,
+      "step": 45604
+    },
+    {
+      "epoch": 3.69450745301361,
+      "grad_norm": 0.08605027943849564,
+      "learning_rate": 1.6971960934335478e-05,
+      "loss": 0.2609,
+      "step": 45605
+    },
+    {
+      "epoch": 3.694588464031108,
+      "grad_norm": 0.07421616464853287,
+      "learning_rate": 1.6967460281740855e-05,
+      "loss": 0.2275,
+      "step": 45606
+    },
+    {
+      "epoch": 3.694669475048607,
+      "grad_norm": 0.08524744212627411,
+      "learning_rate": 1.696295962914623e-05,
+      "loss": 0.2311,
+      "step": 45607
+    },
+    {
+      "epoch": 3.694750486066105,
+      "grad_norm": 0.06704798340797424,
+      "learning_rate": 1.69584589765516e-05,
+      "loss": 0.2533,
+      "step": 45608
+    },
+    {
+      "epoch": 3.6948314970836034,
+      "grad_norm": 0.049864400178194046,
+      "learning_rate": 1.6953958323956976e-05,
+      "loss": 0.2026,
+      "step": 45609
+    },
+    {
+      "epoch": 3.6949125081011016,
+      "grad_norm": 0.07128980755805969,
+      "learning_rate": 1.694945767136235e-05,
+      "loss": 0.2313,
+      "step": 45610
+    },
+    {
+      "epoch": 3.6949935191186,
+      "grad_norm": 0.06947165727615356,
+      "learning_rate": 1.694495701876772e-05,
+      "loss": 0.2192,
+      "step": 45611
+    },
+    {
+      "epoch": 3.6950745301360985,
+      "grad_norm": 0.06866627931594849,
+      "learning_rate": 1.6940456366173096e-05,
+      "loss": 0.2308,
+      "step": 45612
+    },
+    {
+      "epoch": 3.695155541153597,
+      "grad_norm": 0.07855135202407837,
+      "learning_rate": 1.693595571357847e-05,
+      "loss": 0.2266,
+      "step": 45613
+    },
+    {
+      "epoch": 3.695236552171095,
+      "grad_norm": 0.06514532119035721,
+      "learning_rate": 1.6931455060983844e-05,
+      "loss": 0.2095,
+      "step": 45614
+    },
+    {
+      "epoch": 3.6953175631885937,
+      "grad_norm": 0.058014001697301865,
+      "learning_rate": 1.6926954408389217e-05,
+      "loss": 0.2925,
+      "step": 45615
+    },
+    {
+      "epoch": 3.695398574206092,
+      "grad_norm": 0.06363331526517868,
+      "learning_rate": 1.692245375579459e-05,
+      "loss": 0.2324,
+      "step": 45616
+    },
+    {
+      "epoch": 3.69547958522359,
+      "grad_norm": 0.06826046854257584,
+      "learning_rate": 1.6917953103199964e-05,
+      "loss": 0.2338,
+      "step": 45617
+    },
+    {
+      "epoch": 3.695560596241089,
+      "grad_norm": 0.08055315911769867,
+      "learning_rate": 1.6913452450605338e-05,
+      "loss": 0.2187,
+      "step": 45618
+    },
+    {
+      "epoch": 3.695641607258587,
+      "grad_norm": 0.07960078120231628,
+      "learning_rate": 1.6908951798010715e-05,
+      "loss": 0.2421,
+      "step": 45619
+    },
+    {
+      "epoch": 3.6957226182760854,
+      "grad_norm": 0.05961255356669426,
+      "learning_rate": 1.6904451145416085e-05,
+      "loss": 0.2017,
+      "step": 45620
+    },
+    {
+      "epoch": 3.695803629293584,
+      "grad_norm": 0.07862067967653275,
+      "learning_rate": 1.689995049282146e-05,
+      "loss": 0.2251,
+      "step": 45621
+    },
+    {
+      "epoch": 3.6958846403110823,
+      "grad_norm": 0.07465961575508118,
+      "learning_rate": 1.6895449840226836e-05,
+      "loss": 0.222,
+      "step": 45622
+    },
+    {
+      "epoch": 3.6959656513285806,
+      "grad_norm": 0.07449892908334732,
+      "learning_rate": 1.6890949187632206e-05,
+      "loss": 0.2295,
+      "step": 45623
+    },
+    {
+      "epoch": 3.6960466623460793,
+      "grad_norm": 0.0704660639166832,
+      "learning_rate": 1.688644853503758e-05,
+      "loss": 0.2101,
+      "step": 45624
+    },
+    {
+      "epoch": 3.6961276733635775,
+      "grad_norm": 0.07495720684528351,
+      "learning_rate": 1.6881947882442957e-05,
+      "loss": 0.2571,
+      "step": 45625
+    },
+    {
+      "epoch": 3.6962086843810757,
+      "grad_norm": 0.06861800700426102,
+      "learning_rate": 1.6877447229848327e-05,
+      "loss": 0.2147,
+      "step": 45626
+    },
+    {
+      "epoch": 3.6962896953985744,
+      "grad_norm": 0.07803268730640411,
+      "learning_rate": 1.6872946577253704e-05,
+      "loss": 0.2491,
+      "step": 45627
+    },
+    {
+      "epoch": 3.6963707064160727,
+      "grad_norm": 0.0892467275261879,
+      "learning_rate": 1.6868445924659077e-05,
+      "loss": 0.2448,
+      "step": 45628
+    },
+    {
+      "epoch": 3.696451717433571,
+      "grad_norm": 0.07141023129224777,
+      "learning_rate": 1.6863945272064448e-05,
+      "loss": 0.215,
+      "step": 45629
+    },
+    {
+      "epoch": 3.6965327284510696,
+      "grad_norm": 0.06902121752500534,
+      "learning_rate": 1.6859444619469825e-05,
+      "loss": 0.2112,
+      "step": 45630
+    },
+    {
+      "epoch": 3.696613739468568,
+      "grad_norm": 0.058419931679964066,
+      "learning_rate": 1.6854943966875198e-05,
+      "loss": 0.196,
+      "step": 45631
+    },
+    {
+      "epoch": 3.696694750486066,
+      "grad_norm": 0.06349142640829086,
+      "learning_rate": 1.685044331428057e-05,
+      "loss": 0.2146,
+      "step": 45632
+    },
+    {
+      "epoch": 3.6967757615035644,
+      "grad_norm": 0.07252699136734009,
+      "learning_rate": 1.6845942661685945e-05,
+      "loss": 0.2223,
+      "step": 45633
+    },
+    {
+      "epoch": 3.6968567725210626,
+      "grad_norm": 0.09001344442367554,
+      "learning_rate": 1.684144200909132e-05,
+      "loss": 0.253,
+      "step": 45634
+    },
+    {
+      "epoch": 3.6969377835385613,
+      "grad_norm": 0.058856382966041565,
+      "learning_rate": 1.6836941356496693e-05,
+      "loss": 0.2569,
+      "step": 45635
+    },
+    {
+      "epoch": 3.6970187945560595,
+      "grad_norm": 0.0726538598537445,
+      "learning_rate": 1.6832440703902066e-05,
+      "loss": 0.216,
+      "step": 45636
+    },
+    {
+      "epoch": 3.6970998055735578,
+      "grad_norm": 0.06102930009365082,
+      "learning_rate": 1.682794005130744e-05,
+      "loss": 0.221,
+      "step": 45637
+    },
+    {
+      "epoch": 3.6971808165910565,
+      "grad_norm": 0.06977268308401108,
+      "learning_rate": 1.6823439398712814e-05,
+      "loss": 0.2021,
+      "step": 45638
+    },
+    {
+      "epoch": 3.6972618276085547,
+      "grad_norm": 0.06037342548370361,
+      "learning_rate": 1.6818938746118187e-05,
+      "loss": 0.2138,
+      "step": 45639
+    },
+    {
+      "epoch": 3.697342838626053,
+      "grad_norm": 0.07608211785554886,
+      "learning_rate": 1.6814438093523564e-05,
+      "loss": 0.2652,
+      "step": 45640
+    },
+    {
+      "epoch": 3.6974238496435516,
+      "grad_norm": 0.07579577714204788,
+      "learning_rate": 1.6809937440928934e-05,
+      "loss": 0.2378,
+      "step": 45641
+    },
+    {
+      "epoch": 3.69750486066105,
+      "grad_norm": 0.08454374969005585,
+      "learning_rate": 1.6805436788334308e-05,
+      "loss": 0.2475,
+      "step": 45642
+    },
+    {
+      "epoch": 3.697585871678548,
+      "grad_norm": 0.07013771682977676,
+      "learning_rate": 1.6800936135739685e-05,
+      "loss": 0.2294,
+      "step": 45643
+    },
+    {
+      "epoch": 3.697666882696047,
+      "grad_norm": 0.07883304357528687,
+      "learning_rate": 1.6796435483145055e-05,
+      "loss": 0.2483,
+      "step": 45644
+    },
+    {
+      "epoch": 3.697747893713545,
+      "grad_norm": 0.08239931613206863,
+      "learning_rate": 1.679193483055043e-05,
+      "loss": 0.2347,
+      "step": 45645
+    },
+    {
+      "epoch": 3.6978289047310433,
+      "grad_norm": 0.06723158061504364,
+      "learning_rate": 1.6787434177955806e-05,
+      "loss": 0.2109,
+      "step": 45646
+    },
+    {
+      "epoch": 3.697909915748542,
+      "grad_norm": 0.08971650153398514,
+      "learning_rate": 1.6782933525361176e-05,
+      "loss": 0.217,
+      "step": 45647
+    },
+    {
+      "epoch": 3.6979909267660402,
+      "grad_norm": 0.08268365263938904,
+      "learning_rate": 1.6778432872766553e-05,
+      "loss": 0.2135,
+      "step": 45648
+    },
+    {
+      "epoch": 3.6980719377835385,
+      "grad_norm": 0.06352110952138901,
+      "learning_rate": 1.6773932220171927e-05,
+      "loss": 0.1921,
+      "step": 45649
+    },
+    {
+      "epoch": 3.698152948801037,
+      "grad_norm": 0.08296002447605133,
+      "learning_rate": 1.6769431567577297e-05,
+      "loss": 0.2022,
+      "step": 45650
+    },
+    {
+      "epoch": 3.6982339598185354,
+      "grad_norm": 0.07762415707111359,
+      "learning_rate": 1.6764930914982674e-05,
+      "loss": 0.2247,
+      "step": 45651
+    },
+    {
+      "epoch": 3.6983149708360337,
+      "grad_norm": 0.06467720866203308,
+      "learning_rate": 1.6760430262388047e-05,
+      "loss": 0.2153,
+      "step": 45652
+    },
+    {
+      "epoch": 3.6983959818535324,
+      "grad_norm": 0.07439886033535004,
+      "learning_rate": 1.675592960979342e-05,
+      "loss": 0.2479,
+      "step": 45653
+    },
+    {
+      "epoch": 3.6984769928710306,
+      "grad_norm": 0.05738019570708275,
+      "learning_rate": 1.6751428957198795e-05,
+      "loss": 0.212,
+      "step": 45654
+    },
+    {
+      "epoch": 3.698558003888529,
+      "grad_norm": 0.07403714954853058,
+      "learning_rate": 1.6746928304604168e-05,
+      "loss": 0.2277,
+      "step": 45655
+    },
+    {
+      "epoch": 3.698639014906027,
+      "grad_norm": 0.07722627371549606,
+      "learning_rate": 1.6742427652009542e-05,
+      "loss": 0.2012,
+      "step": 45656
+    },
+    {
+      "epoch": 3.6987200259235253,
+      "grad_norm": 0.0807010680437088,
+      "learning_rate": 1.6737926999414915e-05,
+      "loss": 0.2054,
+      "step": 45657
+    },
+    {
+      "epoch": 3.698801036941024,
+      "grad_norm": 0.08320476114749908,
+      "learning_rate": 1.673342634682029e-05,
+      "loss": 0.2418,
+      "step": 45658
+    },
+    {
+      "epoch": 3.6988820479585223,
+      "grad_norm": 0.0703231692314148,
+      "learning_rate": 1.6728925694225663e-05,
+      "loss": 0.2332,
+      "step": 45659
+    },
+    {
+      "epoch": 3.6989630589760205,
+      "grad_norm": 0.06997893750667572,
+      "learning_rate": 1.6724425041631036e-05,
+      "loss": 0.2516,
+      "step": 45660
+    },
+    {
+      "epoch": 3.699044069993519,
+      "grad_norm": 0.06285632401704788,
+      "learning_rate": 1.6719924389036413e-05,
+      "loss": 0.2355,
+      "step": 45661
+    },
+    {
+      "epoch": 3.6991250810110174,
+      "grad_norm": 0.07425893098115921,
+      "learning_rate": 1.6715423736441783e-05,
+      "loss": 0.2049,
+      "step": 45662
+    },
+    {
+      "epoch": 3.6992060920285157,
+      "grad_norm": 0.07462365180253983,
+      "learning_rate": 1.6710923083847157e-05,
+      "loss": 0.217,
+      "step": 45663
+    },
+    {
+      "epoch": 3.6992871030460144,
+      "grad_norm": 0.07210385799407959,
+      "learning_rate": 1.6706422431252534e-05,
+      "loss": 0.2285,
+      "step": 45664
+    },
+    {
+      "epoch": 3.6993681140635126,
+      "grad_norm": 0.06601511687040329,
+      "learning_rate": 1.6701921778657904e-05,
+      "loss": 0.2469,
+      "step": 45665
+    },
+    {
+      "epoch": 3.699449125081011,
+      "grad_norm": 0.07676741480827332,
+      "learning_rate": 1.669742112606328e-05,
+      "loss": 0.2265,
+      "step": 45666
+    },
+    {
+      "epoch": 3.6995301360985096,
+      "grad_norm": 0.058619819581508636,
+      "learning_rate": 1.6692920473468655e-05,
+      "loss": 0.2046,
+      "step": 45667
+    },
+    {
+      "epoch": 3.699611147116008,
+      "grad_norm": 0.07764916867017746,
+      "learning_rate": 1.6688419820874025e-05,
+      "loss": 0.2205,
+      "step": 45668
+    },
+    {
+      "epoch": 3.699692158133506,
+      "grad_norm": 0.06930316984653473,
+      "learning_rate": 1.6683919168279402e-05,
+      "loss": 0.2196,
+      "step": 45669
+    },
+    {
+      "epoch": 3.6997731691510047,
+      "grad_norm": 0.08031480759382248,
+      "learning_rate": 1.6679418515684776e-05,
+      "loss": 0.2509,
+      "step": 45670
+    },
+    {
+      "epoch": 3.699854180168503,
+      "grad_norm": 0.07526817917823792,
+      "learning_rate": 1.667491786309015e-05,
+      "loss": 0.2521,
+      "step": 45671
+    },
+    {
+      "epoch": 3.6999351911860012,
+      "grad_norm": 0.0837547555565834,
+      "learning_rate": 1.6670417210495523e-05,
+      "loss": 0.2373,
+      "step": 45672
+    },
+    {
+      "epoch": 3.7000162022035,
+      "grad_norm": 0.11006888747215271,
+      "learning_rate": 1.6665916557900896e-05,
+      "loss": 0.2406,
+      "step": 45673
+    },
+    {
+      "epoch": 3.700097213220998,
+      "grad_norm": 0.060090772807598114,
+      "learning_rate": 1.666141590530627e-05,
+      "loss": 0.2311,
+      "step": 45674
+    },
+    {
+      "epoch": 3.7001782242384964,
+      "grad_norm": 0.07697679847478867,
+      "learning_rate": 1.6656915252711644e-05,
+      "loss": 0.2144,
+      "step": 45675
+    },
+    {
+      "epoch": 3.7002592352559946,
+      "grad_norm": 0.075130395591259,
+      "learning_rate": 1.6652414600117017e-05,
+      "loss": 0.2075,
+      "step": 45676
+    },
+    {
+      "epoch": 3.7003402462734933,
+      "grad_norm": 0.06719817966222763,
+      "learning_rate": 1.664791394752239e-05,
+      "loss": 0.262,
+      "step": 45677
+    },
+    {
+      "epoch": 3.7004212572909916,
+      "grad_norm": 0.07148056477308273,
+      "learning_rate": 1.6643413294927764e-05,
+      "loss": 0.2244,
+      "step": 45678
+    },
+    {
+      "epoch": 3.70050226830849,
+      "grad_norm": 0.07177882641553879,
+      "learning_rate": 1.663891264233314e-05,
+      "loss": 0.2164,
+      "step": 45679
+    },
+    {
+      "epoch": 3.700583279325988,
+      "grad_norm": 0.06076192483305931,
+      "learning_rate": 1.663441198973851e-05,
+      "loss": 0.2045,
+      "step": 45680
+    },
+    {
+      "epoch": 3.7006642903434868,
+      "grad_norm": 0.05700770765542984,
+      "learning_rate": 1.6629911337143885e-05,
+      "loss": 0.2023,
+      "step": 45681
+    },
+    {
+      "epoch": 3.700745301360985,
+      "grad_norm": 0.07019194215536118,
+      "learning_rate": 1.6625410684549262e-05,
+      "loss": 0.2374,
+      "step": 45682
+    },
+    {
+      "epoch": 3.7008263123784833,
+      "grad_norm": 0.059162236750125885,
+      "learning_rate": 1.6620910031954632e-05,
+      "loss": 0.2114,
+      "step": 45683
+    },
+    {
+      "epoch": 3.700907323395982,
+      "grad_norm": 0.07613490521907806,
+      "learning_rate": 1.6616409379360006e-05,
+      "loss": 0.2293,
+      "step": 45684
+    },
+    {
+      "epoch": 3.70098833441348,
+      "grad_norm": 0.08490029722452164,
+      "learning_rate": 1.6611908726765383e-05,
+      "loss": 0.1955,
+      "step": 45685
+    },
+    {
+      "epoch": 3.7010693454309784,
+      "grad_norm": 0.07354255020618439,
+      "learning_rate": 1.6607408074170757e-05,
+      "loss": 0.2268,
+      "step": 45686
+    },
+    {
+      "epoch": 3.701150356448477,
+      "grad_norm": 0.0803595781326294,
+      "learning_rate": 1.660290742157613e-05,
+      "loss": 0.2502,
+      "step": 45687
+    },
+    {
+      "epoch": 3.7012313674659754,
+      "grad_norm": 0.07173190265893936,
+      "learning_rate": 1.6598406768981504e-05,
+      "loss": 0.2447,
+      "step": 45688
+    },
+    {
+      "epoch": 3.7013123784834736,
+      "grad_norm": 0.08350685238838196,
+      "learning_rate": 1.6593906116386877e-05,
+      "loss": 0.2585,
+      "step": 45689
+    },
+    {
+      "epoch": 3.7013933895009723,
+      "grad_norm": 0.06672927737236023,
+      "learning_rate": 1.658940546379225e-05,
+      "loss": 0.21,
+      "step": 45690
+    },
+    {
+      "epoch": 3.7014744005184705,
+      "grad_norm": 0.07220529764890671,
+      "learning_rate": 1.6584904811197625e-05,
+      "loss": 0.2327,
+      "step": 45691
+    },
+    {
+      "epoch": 3.701555411535969,
+      "grad_norm": 0.07617419213056564,
+      "learning_rate": 1.6580404158602998e-05,
+      "loss": 0.2163,
+      "step": 45692
+    },
+    {
+      "epoch": 3.7016364225534675,
+      "grad_norm": 0.06999776512384415,
+      "learning_rate": 1.6575903506008372e-05,
+      "loss": 0.2125,
+      "step": 45693
+    },
+    {
+      "epoch": 3.7017174335709657,
+      "grad_norm": 0.07320782542228699,
+      "learning_rate": 1.6571402853413745e-05,
+      "loss": 0.2258,
+      "step": 45694
+    },
+    {
+      "epoch": 3.701798444588464,
+      "grad_norm": 0.07582980394363403,
+      "learning_rate": 1.656690220081912e-05,
+      "loss": 0.213,
+      "step": 45695
+    },
+    {
+      "epoch": 3.7018794556059627,
+      "grad_norm": 0.05617687106132507,
+      "learning_rate": 1.6562401548224493e-05,
+      "loss": 0.1939,
+      "step": 45696
+    },
+    {
+      "epoch": 3.701960466623461,
+      "grad_norm": 0.058806758373975754,
+      "learning_rate": 1.6557900895629866e-05,
+      "loss": 0.2449,
+      "step": 45697
+    },
+    {
+      "epoch": 3.702041477640959,
+      "grad_norm": 0.0760156661272049,
+      "learning_rate": 1.655340024303524e-05,
+      "loss": 0.238,
+      "step": 45698
+    },
+    {
+      "epoch": 3.7021224886584574,
+      "grad_norm": 0.06211948022246361,
+      "learning_rate": 1.6548899590440613e-05,
+      "loss": 0.2389,
+      "step": 45699
+    },
+    {
+      "epoch": 3.702203499675956,
+      "grad_norm": 0.0635630339384079,
+      "learning_rate": 1.654439893784599e-05,
+      "loss": 0.217,
+      "step": 45700
+    },
+    {
+      "epoch": 3.7022845106934543,
+      "grad_norm": 0.07854915410280228,
+      "learning_rate": 1.653989828525136e-05,
+      "loss": 0.2497,
+      "step": 45701
+    },
+    {
+      "epoch": 3.7023655217109526,
+      "grad_norm": 0.06777413934469223,
+      "learning_rate": 1.6535397632656734e-05,
+      "loss": 0.24,
+      "step": 45702
+    },
+    {
+      "epoch": 3.702446532728451,
+      "grad_norm": 0.06990412622690201,
+      "learning_rate": 1.653089698006211e-05,
+      "loss": 0.2094,
+      "step": 45703
+    },
+    {
+      "epoch": 3.7025275437459495,
+      "grad_norm": 0.06906888633966446,
+      "learning_rate": 1.6526396327467485e-05,
+      "loss": 0.2311,
+      "step": 45704
+    },
+    {
+      "epoch": 3.7026085547634477,
+      "grad_norm": 0.06097817420959473,
+      "learning_rate": 1.652189567487286e-05,
+      "loss": 0.1887,
+      "step": 45705
+    },
+    {
+      "epoch": 3.702689565780946,
+      "grad_norm": 0.07380296289920807,
+      "learning_rate": 1.6517395022278232e-05,
+      "loss": 0.2236,
+      "step": 45706
+    },
+    {
+      "epoch": 3.7027705767984447,
+      "grad_norm": 0.07449514418840408,
+      "learning_rate": 1.6512894369683606e-05,
+      "loss": 0.2055,
+      "step": 45707
+    },
+    {
+      "epoch": 3.702851587815943,
+      "grad_norm": 0.07819279283285141,
+      "learning_rate": 1.650839371708898e-05,
+      "loss": 0.2339,
+      "step": 45708
+    },
+    {
+      "epoch": 3.702932598833441,
+      "grad_norm": 0.06513992697000504,
+      "learning_rate": 1.6503893064494353e-05,
+      "loss": 0.216,
+      "step": 45709
+    },
+    {
+      "epoch": 3.70301360985094,
+      "grad_norm": 0.08240225911140442,
+      "learning_rate": 1.6499392411899726e-05,
+      "loss": 0.2141,
+      "step": 45710
+    },
+    {
+      "epoch": 3.703094620868438,
+      "grad_norm": 0.08612393587827682,
+      "learning_rate": 1.64948917593051e-05,
+      "loss": 0.2298,
+      "step": 45711
+    },
+    {
+      "epoch": 3.7031756318859363,
+      "grad_norm": 0.06854256242513657,
+      "learning_rate": 1.6490391106710474e-05,
+      "loss": 0.2128,
+      "step": 45712
+    },
+    {
+      "epoch": 3.703256642903435,
+      "grad_norm": 0.05733377858996391,
+      "learning_rate": 1.6485890454115847e-05,
+      "loss": 0.2351,
+      "step": 45713
+    },
+    {
+      "epoch": 3.7033376539209333,
+      "grad_norm": 0.06590426713228226,
+      "learning_rate": 1.648138980152122e-05,
+      "loss": 0.2482,
+      "step": 45714
+    },
+    {
+      "epoch": 3.7034186649384315,
+      "grad_norm": 0.06614598631858826,
+      "learning_rate": 1.6476889148926595e-05,
+      "loss": 0.2188,
+      "step": 45715
+    },
+    {
+      "epoch": 3.70349967595593,
+      "grad_norm": 0.06795570999383926,
+      "learning_rate": 1.6472388496331968e-05,
+      "loss": 0.2445,
+      "step": 45716
+    },
+    {
+      "epoch": 3.7035806869734285,
+      "grad_norm": 0.07279584556818008,
+      "learning_rate": 1.6467887843737342e-05,
+      "loss": 0.2135,
+      "step": 45717
+    },
+    {
+      "epoch": 3.7036616979909267,
+      "grad_norm": 0.0802987739443779,
+      "learning_rate": 1.6463387191142715e-05,
+      "loss": 0.2216,
+      "step": 45718
+    },
+    {
+      "epoch": 3.7037427090084254,
+      "grad_norm": 0.07141979783773422,
+      "learning_rate": 1.645888653854809e-05,
+      "loss": 0.2343,
+      "step": 45719
+    },
+    {
+      "epoch": 3.7038237200259236,
+      "grad_norm": 0.058926984667778015,
+      "learning_rate": 1.6454385885953463e-05,
+      "loss": 0.2207,
+      "step": 45720
+    },
+    {
+      "epoch": 3.703904731043422,
+      "grad_norm": 0.07551530748605728,
+      "learning_rate": 1.644988523335884e-05,
+      "loss": 0.2285,
+      "step": 45721
+    },
+    {
+      "epoch": 3.70398574206092,
+      "grad_norm": 0.08013749122619629,
+      "learning_rate": 1.6445384580764213e-05,
+      "loss": 0.2577,
+      "step": 45722
+    },
+    {
+      "epoch": 3.704066753078419,
+      "grad_norm": 0.0685802772641182,
+      "learning_rate": 1.6440883928169583e-05,
+      "loss": 0.2172,
+      "step": 45723
+    },
+    {
+      "epoch": 3.704147764095917,
+      "grad_norm": 0.07025080174207687,
+      "learning_rate": 1.643638327557496e-05,
+      "loss": 0.2394,
+      "step": 45724
+    },
+    {
+      "epoch": 3.7042287751134153,
+      "grad_norm": 0.07615825533866882,
+      "learning_rate": 1.6431882622980334e-05,
+      "loss": 0.2186,
+      "step": 45725
+    },
+    {
+      "epoch": 3.7043097861309136,
+      "grad_norm": 0.07653851807117462,
+      "learning_rate": 1.6427381970385708e-05,
+      "loss": 0.2106,
+      "step": 45726
+    },
+    {
+      "epoch": 3.7043907971484122,
+      "grad_norm": 0.07739897817373276,
+      "learning_rate": 1.642288131779108e-05,
+      "loss": 0.2066,
+      "step": 45727
+    },
+    {
+      "epoch": 3.7044718081659105,
+      "grad_norm": 0.07978235185146332,
+      "learning_rate": 1.6418380665196455e-05,
+      "loss": 0.252,
+      "step": 45728
+    },
+    {
+      "epoch": 3.7045528191834087,
+      "grad_norm": 0.07335931062698364,
+      "learning_rate": 1.641388001260183e-05,
+      "loss": 0.1746,
+      "step": 45729
+    },
+    {
+      "epoch": 3.7046338302009074,
+      "grad_norm": 0.08139511197805405,
+      "learning_rate": 1.6409379360007202e-05,
+      "loss": 0.2766,
+      "step": 45730
+    },
+    {
+      "epoch": 3.7047148412184057,
+      "grad_norm": 0.06404054909944534,
+      "learning_rate": 1.6404878707412576e-05,
+      "loss": 0.1958,
+      "step": 45731
+    },
+    {
+      "epoch": 3.704795852235904,
+      "grad_norm": 0.062324702739715576,
+      "learning_rate": 1.640037805481795e-05,
+      "loss": 0.2066,
+      "step": 45732
+    },
+    {
+      "epoch": 3.7048768632534026,
+      "grad_norm": 0.08526472002267838,
+      "learning_rate": 1.6395877402223323e-05,
+      "loss": 0.2312,
+      "step": 45733
+    },
+    {
+      "epoch": 3.704957874270901,
+      "grad_norm": 0.06586598604917526,
+      "learning_rate": 1.6391376749628696e-05,
+      "loss": 0.2201,
+      "step": 45734
+    },
+    {
+      "epoch": 3.705038885288399,
+      "grad_norm": 0.08027635514736176,
+      "learning_rate": 1.638687609703407e-05,
+      "loss": 0.2203,
+      "step": 45735
+    },
+    {
+      "epoch": 3.7051198963058978,
+      "grad_norm": 0.0849221721291542,
+      "learning_rate": 1.6382375444439444e-05,
+      "loss": 0.2481,
+      "step": 45736
+    },
+    {
+      "epoch": 3.705200907323396,
+      "grad_norm": 0.08749920129776001,
+      "learning_rate": 1.637787479184482e-05,
+      "loss": 0.2237,
+      "step": 45737
+    },
+    {
+      "epoch": 3.7052819183408943,
+      "grad_norm": 0.07888040691614151,
+      "learning_rate": 1.637337413925019e-05,
+      "loss": 0.224,
+      "step": 45738
+    },
+    {
+      "epoch": 3.705362929358393,
+      "grad_norm": 0.08087297528982162,
+      "learning_rate": 1.6368873486655568e-05,
+      "loss": 0.2175,
+      "step": 45739
+    },
+    {
+      "epoch": 3.705443940375891,
+      "grad_norm": 0.0801047757267952,
+      "learning_rate": 1.636437283406094e-05,
+      "loss": 0.2298,
+      "step": 45740
+    },
+    {
+      "epoch": 3.7055249513933894,
+      "grad_norm": 0.06400655955076218,
+      "learning_rate": 1.635987218146631e-05,
+      "loss": 0.2214,
+      "step": 45741
+    },
+    {
+      "epoch": 3.705605962410888,
+      "grad_norm": 0.06360850483179092,
+      "learning_rate": 1.635537152887169e-05,
+      "loss": 0.2173,
+      "step": 45742
+    },
+    {
+      "epoch": 3.7056869734283864,
+      "grad_norm": 0.06993993371725082,
+      "learning_rate": 1.6350870876277062e-05,
+      "loss": 0.2188,
+      "step": 45743
+    },
+    {
+      "epoch": 3.7057679844458846,
+      "grad_norm": 0.08145015686750412,
+      "learning_rate": 1.6346370223682432e-05,
+      "loss": 0.2529,
+      "step": 45744
+    },
+    {
+      "epoch": 3.705848995463383,
+      "grad_norm": 0.08533214777708054,
+      "learning_rate": 1.634186957108781e-05,
+      "loss": 0.2268,
+      "step": 45745
+    },
+    {
+      "epoch": 3.7059300064808816,
+      "grad_norm": 0.08632870018482208,
+      "learning_rate": 1.6337368918493183e-05,
+      "loss": 0.2359,
+      "step": 45746
+    },
+    {
+      "epoch": 3.70601101749838,
+      "grad_norm": 0.07463864982128143,
+      "learning_rate": 1.6332868265898557e-05,
+      "loss": 0.2247,
+      "step": 45747
+    },
+    {
+      "epoch": 3.706092028515878,
+      "grad_norm": 0.06898374110460281,
+      "learning_rate": 1.632836761330393e-05,
+      "loss": 0.2013,
+      "step": 45748
+    },
+    {
+      "epoch": 3.7061730395333763,
+      "grad_norm": 0.06634795665740967,
+      "learning_rate": 1.6323866960709304e-05,
+      "loss": 0.1984,
+      "step": 45749
+    },
+    {
+      "epoch": 3.706254050550875,
+      "grad_norm": 0.08424997329711914,
+      "learning_rate": 1.6319366308114677e-05,
+      "loss": 0.2292,
+      "step": 45750
+    },
+    {
+      "epoch": 3.7063350615683732,
+      "grad_norm": 0.08333491533994675,
+      "learning_rate": 1.631486565552005e-05,
+      "loss": 0.2562,
+      "step": 45751
+    },
+    {
+      "epoch": 3.7064160725858715,
+      "grad_norm": 0.07007382810115814,
+      "learning_rate": 1.6310365002925425e-05,
+      "loss": 0.2016,
+      "step": 45752
+    },
+    {
+      "epoch": 3.70649708360337,
+      "grad_norm": 0.07411225140094757,
+      "learning_rate": 1.6305864350330798e-05,
+      "loss": 0.2552,
+      "step": 45753
+    },
+    {
+      "epoch": 3.7065780946208684,
+      "grad_norm": 0.08287201076745987,
+      "learning_rate": 1.6301363697736172e-05,
+      "loss": 0.2521,
+      "step": 45754
+    },
+    {
+      "epoch": 3.7066591056383666,
+      "grad_norm": 0.07097062468528748,
+      "learning_rate": 1.629686304514155e-05,
+      "loss": 0.2275,
+      "step": 45755
+    },
+    {
+      "epoch": 3.7067401166558653,
+      "grad_norm": 0.07343895733356476,
+      "learning_rate": 1.629236239254692e-05,
+      "loss": 0.197,
+      "step": 45756
+    },
+    {
+      "epoch": 3.7068211276733636,
+      "grad_norm": 0.06564123928546906,
+      "learning_rate": 1.6287861739952293e-05,
+      "loss": 0.2087,
+      "step": 45757
+    },
+    {
+      "epoch": 3.706902138690862,
+      "grad_norm": 0.0649932399392128,
+      "learning_rate": 1.628336108735767e-05,
+      "loss": 0.2356,
+      "step": 45758
+    },
+    {
+      "epoch": 3.7069831497083605,
+      "grad_norm": 0.07513502985239029,
+      "learning_rate": 1.627886043476304e-05,
+      "loss": 0.2562,
+      "step": 45759
+    },
+    {
+      "epoch": 3.7070641607258588,
+      "grad_norm": 0.07147255539894104,
+      "learning_rate": 1.6274359782168417e-05,
+      "loss": 0.2592,
+      "step": 45760
+    },
+    {
+      "epoch": 3.707145171743357,
+      "grad_norm": 0.0936325266957283,
+      "learning_rate": 1.626985912957379e-05,
+      "loss": 0.2757,
+      "step": 45761
+    },
+    {
+      "epoch": 3.7072261827608557,
+      "grad_norm": 0.06833754479885101,
+      "learning_rate": 1.626535847697916e-05,
+      "loss": 0.2258,
+      "step": 45762
+    },
+    {
+      "epoch": 3.707307193778354,
+      "grad_norm": 0.08121831715106964,
+      "learning_rate": 1.6260857824384538e-05,
+      "loss": 0.211,
+      "step": 45763
+    },
+    {
+      "epoch": 3.707388204795852,
+      "grad_norm": 0.0541922003030777,
+      "learning_rate": 1.625635717178991e-05,
+      "loss": 0.1754,
+      "step": 45764
+    },
+    {
+      "epoch": 3.707469215813351,
+      "grad_norm": 0.07351001352071762,
+      "learning_rate": 1.6251856519195285e-05,
+      "loss": 0.2369,
+      "step": 45765
+    },
+    {
+      "epoch": 3.707550226830849,
+      "grad_norm": 0.1002897396683693,
+      "learning_rate": 1.624735586660066e-05,
+      "loss": 0.2499,
+      "step": 45766
+    },
+    {
+      "epoch": 3.7076312378483474,
+      "grad_norm": 0.053547605872154236,
+      "learning_rate": 1.6242855214006032e-05,
+      "loss": 0.2213,
+      "step": 45767
+    },
+    {
+      "epoch": 3.7077122488658456,
+      "grad_norm": 0.06925728917121887,
+      "learning_rate": 1.6238354561411406e-05,
+      "loss": 0.2017,
+      "step": 45768
+    },
+    {
+      "epoch": 3.7077932598833443,
+      "grad_norm": 0.06350000202655792,
+      "learning_rate": 1.623385390881678e-05,
+      "loss": 0.2297,
+      "step": 45769
+    },
+    {
+      "epoch": 3.7078742709008425,
+      "grad_norm": 0.06750939041376114,
+      "learning_rate": 1.6229353256222153e-05,
+      "loss": 0.1952,
+      "step": 45770
+    },
+    {
+      "epoch": 3.707955281918341,
+      "grad_norm": 0.0888800173997879,
+      "learning_rate": 1.6224852603627526e-05,
+      "loss": 0.2372,
+      "step": 45771
+    },
+    {
+      "epoch": 3.708036292935839,
+      "grad_norm": 0.06730051338672638,
+      "learning_rate": 1.62203519510329e-05,
+      "loss": 0.2262,
+      "step": 45772
+    },
+    {
+      "epoch": 3.7081173039533377,
+      "grad_norm": 0.09187749773263931,
+      "learning_rate": 1.6215851298438277e-05,
+      "loss": 0.2275,
+      "step": 45773
+    },
+    {
+      "epoch": 3.708198314970836,
+      "grad_norm": 0.06515898555517197,
+      "learning_rate": 1.6211350645843647e-05,
+      "loss": 0.199,
+      "step": 45774
+    },
+    {
+      "epoch": 3.708279325988334,
+      "grad_norm": 0.07909893989562988,
+      "learning_rate": 1.620684999324902e-05,
+      "loss": 0.2394,
+      "step": 45775
+    },
+    {
+      "epoch": 3.708360337005833,
+      "grad_norm": 0.0797453448176384,
+      "learning_rate": 1.6202349340654398e-05,
+      "loss": 0.2138,
+      "step": 45776
+    },
+    {
+      "epoch": 3.708441348023331,
+      "grad_norm": 0.06687083095312119,
+      "learning_rate": 1.6197848688059768e-05,
+      "loss": 0.228,
+      "step": 45777
+    },
+    {
+      "epoch": 3.7085223590408294,
+      "grad_norm": 0.0814642459154129,
+      "learning_rate": 1.6193348035465145e-05,
+      "loss": 0.1993,
+      "step": 45778
+    },
+    {
+      "epoch": 3.708603370058328,
+      "grad_norm": 0.07986731082201004,
+      "learning_rate": 1.618884738287052e-05,
+      "loss": 0.251,
+      "step": 45779
+    },
+    {
+      "epoch": 3.7086843810758263,
+      "grad_norm": 0.06904210150241852,
+      "learning_rate": 1.618434673027589e-05,
+      "loss": 0.2449,
+      "step": 45780
+    },
+    {
+      "epoch": 3.7087653920933246,
+      "grad_norm": 0.06334845721721649,
+      "learning_rate": 1.6179846077681266e-05,
+      "loss": 0.2306,
+      "step": 45781
+    },
+    {
+      "epoch": 3.7088464031108233,
+      "grad_norm": 0.06799927353858948,
+      "learning_rate": 1.617534542508664e-05,
+      "loss": 0.2138,
+      "step": 45782
+    },
+    {
+      "epoch": 3.7089274141283215,
+      "grad_norm": 0.07648967951536179,
+      "learning_rate": 1.617084477249201e-05,
+      "loss": 0.2229,
+      "step": 45783
+    },
+    {
+      "epoch": 3.7090084251458197,
+      "grad_norm": 0.05069386214017868,
+      "learning_rate": 1.6166344119897387e-05,
+      "loss": 0.1997,
+      "step": 45784
+    },
+    {
+      "epoch": 3.7090894361633184,
+      "grad_norm": 0.07770735770463943,
+      "learning_rate": 1.616184346730276e-05,
+      "loss": 0.2248,
+      "step": 45785
+    },
+    {
+      "epoch": 3.7091704471808167,
+      "grad_norm": 0.08397640287876129,
+      "learning_rate": 1.6157342814708134e-05,
+      "loss": 0.2367,
+      "step": 45786
+    },
+    {
+      "epoch": 3.709251458198315,
+      "grad_norm": 0.059977419674396515,
+      "learning_rate": 1.6152842162113508e-05,
+      "loss": 0.2107,
+      "step": 45787
+    },
+    {
+      "epoch": 3.7093324692158136,
+      "grad_norm": 0.0799662172794342,
+      "learning_rate": 1.614834150951888e-05,
+      "loss": 0.2496,
+      "step": 45788
+    },
+    {
+      "epoch": 3.709413480233312,
+      "grad_norm": 0.058951485902071,
+      "learning_rate": 1.6143840856924255e-05,
+      "loss": 0.211,
+      "step": 45789
+    },
+    {
+      "epoch": 3.70949449125081,
+      "grad_norm": 0.07071194797754288,
+      "learning_rate": 1.613934020432963e-05,
+      "loss": 0.2352,
+      "step": 45790
+    },
+    {
+      "epoch": 3.7095755022683083,
+      "grad_norm": 0.05729727819561958,
+      "learning_rate": 1.6134839551735005e-05,
+      "loss": 0.1978,
+      "step": 45791
+    },
+    {
+      "epoch": 3.709656513285807,
+      "grad_norm": 0.06328321993350983,
+      "learning_rate": 1.6130338899140376e-05,
+      "loss": 0.2145,
+      "step": 45792
+    },
+    {
+      "epoch": 3.7097375243033053,
+      "grad_norm": 0.07447624206542969,
+      "learning_rate": 1.612583824654575e-05,
+      "loss": 0.2312,
+      "step": 45793
+    },
+    {
+      "epoch": 3.7098185353208035,
+      "grad_norm": 0.06918629258871078,
+      "learning_rate": 1.6121337593951126e-05,
+      "loss": 0.2223,
+      "step": 45794
+    },
+    {
+      "epoch": 3.7098995463383018,
+      "grad_norm": 0.059608764946460724,
+      "learning_rate": 1.6116836941356496e-05,
+      "loss": 0.2501,
+      "step": 45795
+    },
+    {
+      "epoch": 3.7099805573558005,
+      "grad_norm": 0.07887037843465805,
+      "learning_rate": 1.611233628876187e-05,
+      "loss": 0.2245,
+      "step": 45796
+    },
+    {
+      "epoch": 3.7100615683732987,
+      "grad_norm": 0.06873272359371185,
+      "learning_rate": 1.6107835636167247e-05,
+      "loss": 0.2225,
+      "step": 45797
+    },
+    {
+      "epoch": 3.710142579390797,
+      "grad_norm": 0.09617675095796585,
+      "learning_rate": 1.6103334983572617e-05,
+      "loss": 0.2345,
+      "step": 45798
+    },
+    {
+      "epoch": 3.7102235904082956,
+      "grad_norm": 0.07761441171169281,
+      "learning_rate": 1.6098834330977994e-05,
+      "loss": 0.2673,
+      "step": 45799
+    },
+    {
+      "epoch": 3.710304601425794,
+      "grad_norm": 0.09091390669345856,
+      "learning_rate": 1.6094333678383368e-05,
+      "loss": 0.2659,
+      "step": 45800
+    },
+    {
+      "epoch": 3.710385612443292,
+      "grad_norm": 0.06801667809486389,
+      "learning_rate": 1.6089833025788738e-05,
+      "loss": 0.2206,
+      "step": 45801
+    },
+    {
+      "epoch": 3.710466623460791,
+      "grad_norm": 0.07345068454742432,
+      "learning_rate": 1.6085332373194115e-05,
+      "loss": 0.2409,
+      "step": 45802
+    },
+    {
+      "epoch": 3.710547634478289,
+      "grad_norm": 0.08711735904216766,
+      "learning_rate": 1.608083172059949e-05,
+      "loss": 0.2144,
+      "step": 45803
+    },
+    {
+      "epoch": 3.7106286454957873,
+      "grad_norm": 0.06707657873630524,
+      "learning_rate": 1.607633106800486e-05,
+      "loss": 0.2124,
+      "step": 45804
+    },
+    {
+      "epoch": 3.710709656513286,
+      "grad_norm": 0.08471059054136276,
+      "learning_rate": 1.6071830415410236e-05,
+      "loss": 0.2525,
+      "step": 45805
+    },
+    {
+      "epoch": 3.7107906675307842,
+      "grad_norm": 0.0858485996723175,
+      "learning_rate": 1.606732976281561e-05,
+      "loss": 0.2109,
+      "step": 45806
+    },
+    {
+      "epoch": 3.7108716785482825,
+      "grad_norm": 0.08600111305713654,
+      "learning_rate": 1.6062829110220983e-05,
+      "loss": 0.2317,
+      "step": 45807
+    },
+    {
+      "epoch": 3.710952689565781,
+      "grad_norm": 0.06662708520889282,
+      "learning_rate": 1.6058328457626357e-05,
+      "loss": 0.2328,
+      "step": 45808
+    },
+    {
+      "epoch": 3.7110337005832794,
+      "grad_norm": 0.05524429306387901,
+      "learning_rate": 1.605382780503173e-05,
+      "loss": 0.2432,
+      "step": 45809
+    },
+    {
+      "epoch": 3.7111147116007777,
+      "grad_norm": 0.07598711550235748,
+      "learning_rate": 1.6049327152437104e-05,
+      "loss": 0.2376,
+      "step": 45810
+    },
+    {
+      "epoch": 3.7111957226182763,
+      "grad_norm": 0.0858222246170044,
+      "learning_rate": 1.6044826499842477e-05,
+      "loss": 0.2052,
+      "step": 45811
+    },
+    {
+      "epoch": 3.7112767336357746,
+      "grad_norm": 0.06710031628608704,
+      "learning_rate": 1.6040325847247854e-05,
+      "loss": 0.2005,
+      "step": 45812
+    },
+    {
+      "epoch": 3.711357744653273,
+      "grad_norm": 0.07461856305599213,
+      "learning_rate": 1.6035825194653225e-05,
+      "loss": 0.2281,
+      "step": 45813
+    },
+    {
+      "epoch": 3.711438755670771,
+      "grad_norm": 0.06366579234600067,
+      "learning_rate": 1.6031324542058598e-05,
+      "loss": 0.2136,
+      "step": 45814
+    },
+    {
+      "epoch": 3.7115197666882693,
+      "grad_norm": 0.07957981526851654,
+      "learning_rate": 1.6026823889463975e-05,
+      "loss": 0.2297,
+      "step": 45815
+    },
+    {
+      "epoch": 3.711600777705768,
+      "grad_norm": 0.06554809212684631,
+      "learning_rate": 1.6022323236869345e-05,
+      "loss": 0.23,
+      "step": 45816
+    },
+    {
+      "epoch": 3.7116817887232663,
+      "grad_norm": 0.06822206825017929,
+      "learning_rate": 1.601782258427472e-05,
+      "loss": 0.204,
+      "step": 45817
+    },
+    {
+      "epoch": 3.7117627997407645,
+      "grad_norm": 0.07761096954345703,
+      "learning_rate": 1.6013321931680096e-05,
+      "loss": 0.2212,
+      "step": 45818
+    },
+    {
+      "epoch": 3.711843810758263,
+      "grad_norm": 0.05761200562119484,
+      "learning_rate": 1.6008821279085466e-05,
+      "loss": 0.2218,
+      "step": 45819
+    },
+    {
+      "epoch": 3.7119248217757614,
+      "grad_norm": 0.06800486892461777,
+      "learning_rate": 1.6004320626490843e-05,
+      "loss": 0.202,
+      "step": 45820
+    },
+    {
+      "epoch": 3.7120058327932597,
+      "grad_norm": 0.08612167090177536,
+      "learning_rate": 1.5999819973896217e-05,
+      "loss": 0.2504,
+      "step": 45821
+    },
+    {
+      "epoch": 3.7120868438107584,
+      "grad_norm": 0.0777377337217331,
+      "learning_rate": 1.5995319321301587e-05,
+      "loss": 0.2192,
+      "step": 45822
+    },
+    {
+      "epoch": 3.7121678548282566,
+      "grad_norm": 0.0771232396364212,
+      "learning_rate": 1.5990818668706964e-05,
+      "loss": 0.233,
+      "step": 45823
+    },
+    {
+      "epoch": 3.712248865845755,
+      "grad_norm": 0.0695451945066452,
+      "learning_rate": 1.5986318016112338e-05,
+      "loss": 0.2525,
+      "step": 45824
+    },
+    {
+      "epoch": 3.7123298768632536,
+      "grad_norm": 0.08157804608345032,
+      "learning_rate": 1.598181736351771e-05,
+      "loss": 0.2623,
+      "step": 45825
+    },
+    {
+      "epoch": 3.712410887880752,
+      "grad_norm": 0.07781726866960526,
+      "learning_rate": 1.5977316710923085e-05,
+      "loss": 0.2288,
+      "step": 45826
+    },
+    {
+      "epoch": 3.71249189889825,
+      "grad_norm": 0.08144915848970413,
+      "learning_rate": 1.597281605832846e-05,
+      "loss": 0.245,
+      "step": 45827
+    },
+    {
+      "epoch": 3.7125729099157487,
+      "grad_norm": 0.0604521706700325,
+      "learning_rate": 1.5968315405733832e-05,
+      "loss": 0.1962,
+      "step": 45828
+    },
+    {
+      "epoch": 3.712653920933247,
+      "grad_norm": 0.056658048182725906,
+      "learning_rate": 1.5963814753139206e-05,
+      "loss": 0.1805,
+      "step": 45829
+    },
+    {
+      "epoch": 3.712734931950745,
+      "grad_norm": 0.06701702624559402,
+      "learning_rate": 1.595931410054458e-05,
+      "loss": 0.2013,
+      "step": 45830
+    },
+    {
+      "epoch": 3.712815942968244,
+      "grad_norm": 0.07957711070775986,
+      "learning_rate": 1.5954813447949953e-05,
+      "loss": 0.2567,
+      "step": 45831
+    },
+    {
+      "epoch": 3.712896953985742,
+      "grad_norm": 0.07916118204593658,
+      "learning_rate": 1.5950312795355326e-05,
+      "loss": 0.2017,
+      "step": 45832
+    },
+    {
+      "epoch": 3.7129779650032404,
+      "grad_norm": 0.06471701711416245,
+      "learning_rate": 1.5945812142760703e-05,
+      "loss": 0.2304,
+      "step": 45833
+    },
+    {
+      "epoch": 3.713058976020739,
+      "grad_norm": 0.0795748233795166,
+      "learning_rate": 1.5941311490166074e-05,
+      "loss": 0.2586,
+      "step": 45834
+    },
+    {
+      "epoch": 3.7131399870382373,
+      "grad_norm": 0.07704074680805206,
+      "learning_rate": 1.5936810837571447e-05,
+      "loss": 0.2104,
+      "step": 45835
+    },
+    {
+      "epoch": 3.7132209980557356,
+      "grad_norm": 0.09900951385498047,
+      "learning_rate": 1.5932310184976824e-05,
+      "loss": 0.2398,
+      "step": 45836
+    },
+    {
+      "epoch": 3.713302009073234,
+      "grad_norm": 0.08793462067842484,
+      "learning_rate": 1.5927809532382194e-05,
+      "loss": 0.2122,
+      "step": 45837
+    },
+    {
+      "epoch": 3.713383020090732,
+      "grad_norm": 0.07224460691213608,
+      "learning_rate": 1.592330887978757e-05,
+      "loss": 0.2259,
+      "step": 45838
+    },
+    {
+      "epoch": 3.7134640311082308,
+      "grad_norm": 0.06400369852781296,
+      "learning_rate": 1.5918808227192945e-05,
+      "loss": 0.2213,
+      "step": 45839
+    },
+    {
+      "epoch": 3.713545042125729,
+      "grad_norm": 0.06662207841873169,
+      "learning_rate": 1.5914307574598315e-05,
+      "loss": 0.2238,
+      "step": 45840
+    },
+    {
+      "epoch": 3.7136260531432272,
+      "grad_norm": 0.07715484499931335,
+      "learning_rate": 1.5909806922003692e-05,
+      "loss": 0.2254,
+      "step": 45841
+    },
+    {
+      "epoch": 3.713707064160726,
+      "grad_norm": 0.08571289479732513,
+      "learning_rate": 1.5905306269409066e-05,
+      "loss": 0.227,
+      "step": 45842
+    },
+    {
+      "epoch": 3.713788075178224,
+      "grad_norm": 0.08624354749917984,
+      "learning_rate": 1.5900805616814436e-05,
+      "loss": 0.2444,
+      "step": 45843
+    },
+    {
+      "epoch": 3.7138690861957224,
+      "grad_norm": 0.07141070067882538,
+      "learning_rate": 1.5896304964219813e-05,
+      "loss": 0.2302,
+      "step": 45844
+    },
+    {
+      "epoch": 3.713950097213221,
+      "grad_norm": 0.07181796431541443,
+      "learning_rate": 1.5891804311625187e-05,
+      "loss": 0.2712,
+      "step": 45845
+    },
+    {
+      "epoch": 3.7140311082307194,
+      "grad_norm": 0.08645230531692505,
+      "learning_rate": 1.588730365903056e-05,
+      "loss": 0.2544,
+      "step": 45846
+    },
+    {
+      "epoch": 3.7141121192482176,
+      "grad_norm": 0.08005709946155548,
+      "learning_rate": 1.5882803006435934e-05,
+      "loss": 0.2271,
+      "step": 45847
+    },
+    {
+      "epoch": 3.7141931302657163,
+      "grad_norm": 0.06787635385990143,
+      "learning_rate": 1.5878302353841307e-05,
+      "loss": 0.2092,
+      "step": 45848
+    },
+    {
+      "epoch": 3.7142741412832145,
+      "grad_norm": 0.07872307300567627,
+      "learning_rate": 1.587380170124668e-05,
+      "loss": 0.1951,
+      "step": 45849
+    },
+    {
+      "epoch": 3.714355152300713,
+      "grad_norm": 0.06552909314632416,
+      "learning_rate": 1.5869301048652055e-05,
+      "loss": 0.2069,
+      "step": 45850
+    },
+    {
+      "epoch": 3.7144361633182115,
+      "grad_norm": 0.06676136702299118,
+      "learning_rate": 1.586480039605743e-05,
+      "loss": 0.2355,
+      "step": 45851
+    },
+    {
+      "epoch": 3.7145171743357097,
+      "grad_norm": 0.07451087981462479,
+      "learning_rate": 1.5860299743462802e-05,
+      "loss": 0.2289,
+      "step": 45852
+    },
+    {
+      "epoch": 3.714598185353208,
+      "grad_norm": 0.08217606693506241,
+      "learning_rate": 1.5855799090868175e-05,
+      "loss": 0.2305,
+      "step": 45853
+    },
+    {
+      "epoch": 3.7146791963707066,
+      "grad_norm": 0.08405117690563202,
+      "learning_rate": 1.5851298438273552e-05,
+      "loss": 0.2559,
+      "step": 45854
+    },
+    {
+      "epoch": 3.714760207388205,
+      "grad_norm": 0.049762554466724396,
+      "learning_rate": 1.5846797785678923e-05,
+      "loss": 0.229,
+      "step": 45855
+    },
+    {
+      "epoch": 3.714841218405703,
+      "grad_norm": 0.07197723537683487,
+      "learning_rate": 1.5842297133084296e-05,
+      "loss": 0.2394,
+      "step": 45856
+    },
+    {
+      "epoch": 3.714922229423202,
+      "grad_norm": 0.07479429244995117,
+      "learning_rate": 1.5837796480489673e-05,
+      "loss": 0.2368,
+      "step": 45857
+    },
+    {
+      "epoch": 3.7150032404407,
+      "grad_norm": 0.0639149472117424,
+      "learning_rate": 1.5833295827895044e-05,
+      "loss": 0.2262,
+      "step": 45858
+    },
+    {
+      "epoch": 3.7150842514581983,
+      "grad_norm": 0.06918316334486008,
+      "learning_rate": 1.582879517530042e-05,
+      "loss": 0.2128,
+      "step": 45859
+    },
+    {
+      "epoch": 3.7151652624756966,
+      "grad_norm": 0.08518219739198685,
+      "learning_rate": 1.5824294522705794e-05,
+      "loss": 0.265,
+      "step": 45860
+    },
+    {
+      "epoch": 3.715246273493195,
+      "grad_norm": 0.07202792167663574,
+      "learning_rate": 1.5819793870111164e-05,
+      "loss": 0.2217,
+      "step": 45861
+    },
+    {
+      "epoch": 3.7153272845106935,
+      "grad_norm": 0.06419474631547928,
+      "learning_rate": 1.581529321751654e-05,
+      "loss": 0.2155,
+      "step": 45862
+    },
+    {
+      "epoch": 3.7154082955281917,
+      "grad_norm": 0.07064595073461533,
+      "learning_rate": 1.5810792564921915e-05,
+      "loss": 0.1976,
+      "step": 45863
+    },
+    {
+      "epoch": 3.71548930654569,
+      "grad_norm": 0.0711565911769867,
+      "learning_rate": 1.580629191232729e-05,
+      "loss": 0.2549,
+      "step": 45864
+    },
+    {
+      "epoch": 3.7155703175631887,
+      "grad_norm": 0.06996343284845352,
+      "learning_rate": 1.5801791259732662e-05,
+      "loss": 0.1958,
+      "step": 45865
+    },
+    {
+      "epoch": 3.715651328580687,
+      "grad_norm": 0.06321109086275101,
+      "learning_rate": 1.5797290607138036e-05,
+      "loss": 0.2098,
+      "step": 45866
+    },
+    {
+      "epoch": 3.715732339598185,
+      "grad_norm": 0.07822613418102264,
+      "learning_rate": 1.579278995454341e-05,
+      "loss": 0.2079,
+      "step": 45867
+    },
+    {
+      "epoch": 3.715813350615684,
+      "grad_norm": 0.08399234712123871,
+      "learning_rate": 1.5788289301948783e-05,
+      "loss": 0.2388,
+      "step": 45868
+    },
+    {
+      "epoch": 3.715894361633182,
+      "grad_norm": 0.06259548664093018,
+      "learning_rate": 1.5783788649354157e-05,
+      "loss": 0.2342,
+      "step": 45869
+    },
+    {
+      "epoch": 3.7159753726506803,
+      "grad_norm": 0.08495364338159561,
+      "learning_rate": 1.577928799675953e-05,
+      "loss": 0.2148,
+      "step": 45870
+    },
+    {
+      "epoch": 3.716056383668179,
+      "grad_norm": 0.0643046498298645,
+      "learning_rate": 1.5774787344164904e-05,
+      "loss": 0.2099,
+      "step": 45871
+    },
+    {
+      "epoch": 3.7161373946856773,
+      "grad_norm": 0.06250204145908356,
+      "learning_rate": 1.577028669157028e-05,
+      "loss": 0.1988,
+      "step": 45872
+    },
+    {
+      "epoch": 3.7162184057031755,
+      "grad_norm": 0.0693424865603447,
+      "learning_rate": 1.576578603897565e-05,
+      "loss": 0.2287,
+      "step": 45873
+    },
+    {
+      "epoch": 3.716299416720674,
+      "grad_norm": 0.07121361792087555,
+      "learning_rate": 1.5761285386381025e-05,
+      "loss": 0.235,
+      "step": 45874
+    },
+    {
+      "epoch": 3.7163804277381725,
+      "grad_norm": 0.07043685764074326,
+      "learning_rate": 1.57567847337864e-05,
+      "loss": 0.2493,
+      "step": 45875
+    },
+    {
+      "epoch": 3.7164614387556707,
+      "grad_norm": 0.07397102564573288,
+      "learning_rate": 1.5752284081191772e-05,
+      "loss": 0.2426,
+      "step": 45876
+    },
+    {
+      "epoch": 3.7165424497731694,
+      "grad_norm": 0.0942964181303978,
+      "learning_rate": 1.574778342859715e-05,
+      "loss": 0.2565,
+      "step": 45877
+    },
+    {
+      "epoch": 3.7166234607906676,
+      "grad_norm": 0.08170833438634872,
+      "learning_rate": 1.5743282776002522e-05,
+      "loss": 0.2895,
+      "step": 45878
+    },
+    {
+      "epoch": 3.716704471808166,
+      "grad_norm": 0.07267139106988907,
+      "learning_rate": 1.5738782123407893e-05,
+      "loss": 0.22,
+      "step": 45879
+    },
+    {
+      "epoch": 3.7167854828256646,
+      "grad_norm": 0.06435095518827438,
+      "learning_rate": 1.573428147081327e-05,
+      "loss": 0.2089,
+      "step": 45880
+    },
+    {
+      "epoch": 3.716866493843163,
+      "grad_norm": 0.07378797233104706,
+      "learning_rate": 1.5729780818218643e-05,
+      "loss": 0.2211,
+      "step": 45881
+    },
+    {
+      "epoch": 3.716947504860661,
+      "grad_norm": 0.08473730087280273,
+      "learning_rate": 1.5725280165624017e-05,
+      "loss": 0.2114,
+      "step": 45882
+    },
+    {
+      "epoch": 3.7170285158781593,
+      "grad_norm": 0.06100822240114212,
+      "learning_rate": 1.572077951302939e-05,
+      "loss": 0.2234,
+      "step": 45883
+    },
+    {
+      "epoch": 3.7171095268956575,
+      "grad_norm": 0.08396106958389282,
+      "learning_rate": 1.5716278860434764e-05,
+      "loss": 0.2526,
+      "step": 45884
+    },
+    {
+      "epoch": 3.7171905379131562,
+      "grad_norm": 0.08896417915821075,
+      "learning_rate": 1.5711778207840138e-05,
+      "loss": 0.2871,
+      "step": 45885
+    },
+    {
+      "epoch": 3.7172715489306545,
+      "grad_norm": 0.06886778771877289,
+      "learning_rate": 1.570727755524551e-05,
+      "loss": 0.2309,
+      "step": 45886
+    },
+    {
+      "epoch": 3.7173525599481527,
+      "grad_norm": 0.06046053022146225,
+      "learning_rate": 1.5702776902650885e-05,
+      "loss": 0.2314,
+      "step": 45887
+    },
+    {
+      "epoch": 3.7174335709656514,
+      "grad_norm": 0.0786987841129303,
+      "learning_rate": 1.569827625005626e-05,
+      "loss": 0.2126,
+      "step": 45888
+    },
+    {
+      "epoch": 3.7175145819831497,
+      "grad_norm": 0.08876615762710571,
+      "learning_rate": 1.5693775597461632e-05,
+      "loss": 0.2403,
+      "step": 45889
+    },
+    {
+      "epoch": 3.717595593000648,
+      "grad_norm": 0.06113407388329506,
+      "learning_rate": 1.5689274944867006e-05,
+      "loss": 0.211,
+      "step": 45890
+    },
+    {
+      "epoch": 3.7176766040181466,
+      "grad_norm": 0.06730061769485474,
+      "learning_rate": 1.568477429227238e-05,
+      "loss": 0.2186,
+      "step": 45891
+    },
+    {
+      "epoch": 3.717757615035645,
+      "grad_norm": 0.07700332999229431,
+      "learning_rate": 1.5680273639677753e-05,
+      "loss": 0.2191,
+      "step": 45892
+    },
+    {
+      "epoch": 3.717838626053143,
+      "grad_norm": 0.06986545771360397,
+      "learning_rate": 1.567577298708313e-05,
+      "loss": 0.2744,
+      "step": 45893
+    },
+    {
+      "epoch": 3.7179196370706418,
+      "grad_norm": 0.07376525551080704,
+      "learning_rate": 1.56712723344885e-05,
+      "loss": 0.2353,
+      "step": 45894
+    },
+    {
+      "epoch": 3.71800064808814,
+      "grad_norm": 0.08193989098072052,
+      "learning_rate": 1.5666771681893874e-05,
+      "loss": 0.25,
+      "step": 45895
+    },
+    {
+      "epoch": 3.7180816591056383,
+      "grad_norm": 0.06663582473993301,
+      "learning_rate": 1.566227102929925e-05,
+      "loss": 0.2445,
+      "step": 45896
+    },
+    {
+      "epoch": 3.718162670123137,
+      "grad_norm": 0.08676223456859589,
+      "learning_rate": 1.565777037670462e-05,
+      "loss": 0.2503,
+      "step": 45897
+    },
+    {
+      "epoch": 3.718243681140635,
+      "grad_norm": 0.05338404327630997,
+      "learning_rate": 1.5653269724109998e-05,
+      "loss": 0.197,
+      "step": 45898
+    },
+    {
+      "epoch": 3.7183246921581334,
+      "grad_norm": 0.08339407294988632,
+      "learning_rate": 1.564876907151537e-05,
+      "loss": 0.2584,
+      "step": 45899
+    },
+    {
+      "epoch": 3.718405703175632,
+      "grad_norm": 0.08016695827245712,
+      "learning_rate": 1.5644268418920745e-05,
+      "loss": 0.2498,
+      "step": 45900
+    },
+    {
+      "epoch": 3.7184867141931304,
+      "grad_norm": 0.06563158333301544,
+      "learning_rate": 1.563976776632612e-05,
+      "loss": 0.2013,
+      "step": 45901
+    },
+    {
+      "epoch": 3.7185677252106286,
+      "grad_norm": 0.07572820782661438,
+      "learning_rate": 1.5635267113731492e-05,
+      "loss": 0.2516,
+      "step": 45902
+    },
+    {
+      "epoch": 3.718648736228127,
+      "grad_norm": 0.0666859969496727,
+      "learning_rate": 1.5630766461136866e-05,
+      "loss": 0.2378,
+      "step": 45903
+    },
+    {
+      "epoch": 3.7187297472456255,
+      "grad_norm": 0.0793486088514328,
+      "learning_rate": 1.562626580854224e-05,
+      "loss": 0.2311,
+      "step": 45904
+    },
+    {
+      "epoch": 3.718810758263124,
+      "grad_norm": 0.07341185212135315,
+      "learning_rate": 1.5621765155947613e-05,
+      "loss": 0.2338,
+      "step": 45905
+    },
+    {
+      "epoch": 3.718891769280622,
+      "grad_norm": 0.07319411635398865,
+      "learning_rate": 1.5617264503352987e-05,
+      "loss": 0.1935,
+      "step": 45906
+    },
+    {
+      "epoch": 3.7189727802981203,
+      "grad_norm": 0.0693352222442627,
+      "learning_rate": 1.561276385075836e-05,
+      "loss": 0.2155,
+      "step": 45907
+    },
+    {
+      "epoch": 3.719053791315619,
+      "grad_norm": 0.07195544987916946,
+      "learning_rate": 1.5608263198163734e-05,
+      "loss": 0.2404,
+      "step": 45908
+    },
+    {
+      "epoch": 3.719134802333117,
+      "grad_norm": 0.08030025660991669,
+      "learning_rate": 1.5603762545569107e-05,
+      "loss": 0.2222,
+      "step": 45909
+    },
+    {
+      "epoch": 3.7192158133506155,
+      "grad_norm": 0.0607793889939785,
+      "learning_rate": 1.559926189297448e-05,
+      "loss": 0.1801,
+      "step": 45910
+    },
+    {
+      "epoch": 3.719296824368114,
+      "grad_norm": 0.0874897688627243,
+      "learning_rate": 1.5594761240379858e-05,
+      "loss": 0.2549,
+      "step": 45911
+    },
+    {
+      "epoch": 3.7193778353856124,
+      "grad_norm": 0.08436858654022217,
+      "learning_rate": 1.5590260587785228e-05,
+      "loss": 0.23,
+      "step": 45912
+    },
+    {
+      "epoch": 3.7194588464031106,
+      "grad_norm": 0.06420475244522095,
+      "learning_rate": 1.5585759935190602e-05,
+      "loss": 0.216,
+      "step": 45913
+    },
+    {
+      "epoch": 3.7195398574206093,
+      "grad_norm": 0.0704539343714714,
+      "learning_rate": 1.558125928259598e-05,
+      "loss": 0.258,
+      "step": 45914
+    },
+    {
+      "epoch": 3.7196208684381076,
+      "grad_norm": 0.07884382456541061,
+      "learning_rate": 1.557675863000135e-05,
+      "loss": 0.2243,
+      "step": 45915
+    },
+    {
+      "epoch": 3.719701879455606,
+      "grad_norm": 0.07424849271774292,
+      "learning_rate": 1.5572257977406723e-05,
+      "loss": 0.2347,
+      "step": 45916
+    },
+    {
+      "epoch": 3.7197828904731045,
+      "grad_norm": 0.0690733939409256,
+      "learning_rate": 1.55677573248121e-05,
+      "loss": 0.2222,
+      "step": 45917
+    },
+    {
+      "epoch": 3.7198639014906028,
+      "grad_norm": 0.06833771616220474,
+      "learning_rate": 1.5563256672217473e-05,
+      "loss": 0.2025,
+      "step": 45918
+    },
+    {
+      "epoch": 3.719944912508101,
+      "grad_norm": 0.07035204023122787,
+      "learning_rate": 1.5558756019622847e-05,
+      "loss": 0.2337,
+      "step": 45919
+    },
+    {
+      "epoch": 3.7200259235255997,
+      "grad_norm": 0.07200058549642563,
+      "learning_rate": 1.555425536702822e-05,
+      "loss": 0.2247,
+      "step": 45920
+    },
+    {
+      "epoch": 3.720106934543098,
+      "grad_norm": 0.07681140303611755,
+      "learning_rate": 1.5549754714433594e-05,
+      "loss": 0.2033,
+      "step": 45921
+    },
+    {
+      "epoch": 3.720187945560596,
+      "grad_norm": 0.076214499771595,
+      "learning_rate": 1.5545254061838968e-05,
+      "loss": 0.237,
+      "step": 45922
+    },
+    {
+      "epoch": 3.720268956578095,
+      "grad_norm": 0.060587845742702484,
+      "learning_rate": 1.554075340924434e-05,
+      "loss": 0.2238,
+      "step": 45923
+    },
+    {
+      "epoch": 3.720349967595593,
+      "grad_norm": 0.05092794820666313,
+      "learning_rate": 1.5536252756649715e-05,
+      "loss": 0.2015,
+      "step": 45924
+    },
+    {
+      "epoch": 3.7204309786130914,
+      "grad_norm": 0.08442278951406479,
+      "learning_rate": 1.553175210405509e-05,
+      "loss": 0.2763,
+      "step": 45925
+    },
+    {
+      "epoch": 3.7205119896305896,
+      "grad_norm": 0.07512500137090683,
+      "learning_rate": 1.5527251451460462e-05,
+      "loss": 0.2318,
+      "step": 45926
+    },
+    {
+      "epoch": 3.7205930006480883,
+      "grad_norm": 0.07768537849187851,
+      "learning_rate": 1.5522750798865836e-05,
+      "loss": 0.2908,
+      "step": 45927
+    },
+    {
+      "epoch": 3.7206740116655865,
+      "grad_norm": 0.07198259234428406,
+      "learning_rate": 1.551825014627121e-05,
+      "loss": 0.2386,
+      "step": 45928
+    },
+    {
+      "epoch": 3.720755022683085,
+      "grad_norm": 0.06927481293678284,
+      "learning_rate": 1.5513749493676583e-05,
+      "loss": 0.221,
+      "step": 45929
+    },
+    {
+      "epoch": 3.720836033700583,
+      "grad_norm": 0.06755954027175903,
+      "learning_rate": 1.5509248841081956e-05,
+      "loss": 0.2161,
+      "step": 45930
+    },
+    {
+      "epoch": 3.7209170447180817,
+      "grad_norm": 0.08039725571870804,
+      "learning_rate": 1.550474818848733e-05,
+      "loss": 0.2205,
+      "step": 45931
+    },
+    {
+      "epoch": 3.72099805573558,
+      "grad_norm": 0.06722427904605865,
+      "learning_rate": 1.5500247535892707e-05,
+      "loss": 0.2134,
+      "step": 45932
+    },
+    {
+      "epoch": 3.721079066753078,
+      "grad_norm": 0.07862570881843567,
+      "learning_rate": 1.5495746883298077e-05,
+      "loss": 0.2321,
+      "step": 45933
+    },
+    {
+      "epoch": 3.721160077770577,
+      "grad_norm": 0.0632789134979248,
+      "learning_rate": 1.549124623070345e-05,
+      "loss": 0.2378,
+      "step": 45934
+    },
+    {
+      "epoch": 3.721241088788075,
+      "grad_norm": 0.06634281575679779,
+      "learning_rate": 1.5486745578108828e-05,
+      "loss": 0.2357,
+      "step": 45935
+    },
+    {
+      "epoch": 3.7213220998055734,
+      "grad_norm": 0.07613595575094223,
+      "learning_rate": 1.54822449255142e-05,
+      "loss": 0.2295,
+      "step": 45936
+    },
+    {
+      "epoch": 3.721403110823072,
+      "grad_norm": 0.0799480453133583,
+      "learning_rate": 1.5477744272919575e-05,
+      "loss": 0.2222,
+      "step": 45937
+    },
+    {
+      "epoch": 3.7214841218405703,
+      "grad_norm": 0.07500126212835312,
+      "learning_rate": 1.547324362032495e-05,
+      "loss": 0.2368,
+      "step": 45938
+    },
+    {
+      "epoch": 3.7215651328580686,
+      "grad_norm": 0.06687241792678833,
+      "learning_rate": 1.5468742967730322e-05,
+      "loss": 0.2251,
+      "step": 45939
+    },
+    {
+      "epoch": 3.7216461438755672,
+      "grad_norm": 0.06396705657243729,
+      "learning_rate": 1.5464242315135696e-05,
+      "loss": 0.2216,
+      "step": 45940
+    },
+    {
+      "epoch": 3.7217271548930655,
+      "grad_norm": 0.06962263584136963,
+      "learning_rate": 1.545974166254107e-05,
+      "loss": 0.2254,
+      "step": 45941
+    },
+    {
+      "epoch": 3.7218081659105637,
+      "grad_norm": 0.06181929260492325,
+      "learning_rate": 1.5455241009946443e-05,
+      "loss": 0.2235,
+      "step": 45942
+    },
+    {
+      "epoch": 3.7218891769280624,
+      "grad_norm": 0.07112901657819748,
+      "learning_rate": 1.5450740357351817e-05,
+      "loss": 0.2422,
+      "step": 45943
+    },
+    {
+      "epoch": 3.7219701879455607,
+      "grad_norm": 0.07527320086956024,
+      "learning_rate": 1.544623970475719e-05,
+      "loss": 0.217,
+      "step": 45944
+    },
+    {
+      "epoch": 3.722051198963059,
+      "grad_norm": 0.08394154906272888,
+      "learning_rate": 1.5441739052162564e-05,
+      "loss": 0.2236,
+      "step": 45945
+    },
+    {
+      "epoch": 3.7221322099805576,
+      "grad_norm": 0.07716701924800873,
+      "learning_rate": 1.5437238399567938e-05,
+      "loss": 0.2107,
+      "step": 45946
+    },
+    {
+      "epoch": 3.722213220998056,
+      "grad_norm": 0.06959621608257294,
+      "learning_rate": 1.543273774697331e-05,
+      "loss": 0.2397,
+      "step": 45947
+    },
+    {
+      "epoch": 3.722294232015554,
+      "grad_norm": 0.06327780336141586,
+      "learning_rate": 1.5428237094378685e-05,
+      "loss": 0.1967,
+      "step": 45948
+    },
+    {
+      "epoch": 3.7223752430330523,
+      "grad_norm": 0.0753185898065567,
+      "learning_rate": 1.542373644178406e-05,
+      "loss": 0.2441,
+      "step": 45949
+    },
+    {
+      "epoch": 3.722456254050551,
+      "grad_norm": 0.062304459512233734,
+      "learning_rate": 1.5419235789189435e-05,
+      "loss": 0.239,
+      "step": 45950
+    },
+    {
+      "epoch": 3.7225372650680493,
+      "grad_norm": 0.10284210741519928,
+      "learning_rate": 1.541473513659481e-05,
+      "loss": 0.2458,
+      "step": 45951
+    },
+    {
+      "epoch": 3.7226182760855475,
+      "grad_norm": 0.07348388433456421,
+      "learning_rate": 1.541023448400018e-05,
+      "loss": 0.2355,
+      "step": 45952
+    },
+    {
+      "epoch": 3.7226992871030458,
+      "grad_norm": 0.06129736825823784,
+      "learning_rate": 1.5405733831405556e-05,
+      "loss": 0.197,
+      "step": 45953
+    },
+    {
+      "epoch": 3.7227802981205445,
+      "grad_norm": 0.07383542507886887,
+      "learning_rate": 1.540123317881093e-05,
+      "loss": 0.2116,
+      "step": 45954
+    },
+    {
+      "epoch": 3.7228613091380427,
+      "grad_norm": 0.07537295669317245,
+      "learning_rate": 1.53967325262163e-05,
+      "loss": 0.2375,
+      "step": 45955
+    },
+    {
+      "epoch": 3.722942320155541,
+      "grad_norm": 0.06659074127674103,
+      "learning_rate": 1.5392231873621677e-05,
+      "loss": 0.2277,
+      "step": 45956
+    },
+    {
+      "epoch": 3.7230233311730396,
+      "grad_norm": 0.07395710796117783,
+      "learning_rate": 1.538773122102705e-05,
+      "loss": 0.2246,
+      "step": 45957
+    },
+    {
+      "epoch": 3.723104342190538,
+      "grad_norm": 0.07573916763067245,
+      "learning_rate": 1.5383230568432424e-05,
+      "loss": 0.2409,
+      "step": 45958
+    },
+    {
+      "epoch": 3.723185353208036,
+      "grad_norm": 0.06378524005413055,
+      "learning_rate": 1.5378729915837798e-05,
+      "loss": 0.2007,
+      "step": 45959
+    },
+    {
+      "epoch": 3.723266364225535,
+      "grad_norm": 0.0667334794998169,
+      "learning_rate": 1.537422926324317e-05,
+      "loss": 0.2533,
+      "step": 45960
+    },
+    {
+      "epoch": 3.723347375243033,
+      "grad_norm": 0.06946897506713867,
+      "learning_rate": 1.5369728610648545e-05,
+      "loss": 0.1876,
+      "step": 45961
+    },
+    {
+      "epoch": 3.7234283862605313,
+      "grad_norm": 0.06870533525943756,
+      "learning_rate": 1.536522795805392e-05,
+      "loss": 0.1848,
+      "step": 45962
+    },
+    {
+      "epoch": 3.72350939727803,
+      "grad_norm": 0.0735953152179718,
+      "learning_rate": 1.5360727305459292e-05,
+      "loss": 0.201,
+      "step": 45963
+    },
+    {
+      "epoch": 3.7235904082955282,
+      "grad_norm": 0.063729427754879,
+      "learning_rate": 1.5356226652864666e-05,
+      "loss": 0.2096,
+      "step": 45964
+    },
+    {
+      "epoch": 3.7236714193130265,
+      "grad_norm": 0.06849828362464905,
+      "learning_rate": 1.535172600027004e-05,
+      "loss": 0.2433,
+      "step": 45965
+    },
+    {
+      "epoch": 3.723752430330525,
+      "grad_norm": 0.08339733630418777,
+      "learning_rate": 1.5347225347675413e-05,
+      "loss": 0.2275,
+      "step": 45966
+    },
+    {
+      "epoch": 3.7238334413480234,
+      "grad_norm": 0.07081194967031479,
+      "learning_rate": 1.5342724695080787e-05,
+      "loss": 0.1976,
+      "step": 45967
+    },
+    {
+      "epoch": 3.7239144523655217,
+      "grad_norm": 0.08006429672241211,
+      "learning_rate": 1.533822404248616e-05,
+      "loss": 0.2303,
+      "step": 45968
+    },
+    {
+      "epoch": 3.7239954633830203,
+      "grad_norm": 0.0940973311662674,
+      "learning_rate": 1.5333723389891537e-05,
+      "loss": 0.2473,
+      "step": 45969
+    },
+    {
+      "epoch": 3.7240764744005186,
+      "grad_norm": 0.0769064798951149,
+      "learning_rate": 1.5329222737296907e-05,
+      "loss": 0.2101,
+      "step": 45970
+    },
+    {
+      "epoch": 3.724157485418017,
+      "grad_norm": 0.06387367844581604,
+      "learning_rate": 1.5324722084702284e-05,
+      "loss": 0.2787,
+      "step": 45971
+    },
+    {
+      "epoch": 3.724238496435515,
+      "grad_norm": 0.07406259328126907,
+      "learning_rate": 1.5320221432107658e-05,
+      "loss": 0.234,
+      "step": 45972
+    },
+    {
+      "epoch": 3.7243195074530138,
+      "grad_norm": 0.06665657460689545,
+      "learning_rate": 1.5315720779513028e-05,
+      "loss": 0.2568,
+      "step": 45973
+    },
+    {
+      "epoch": 3.724400518470512,
+      "grad_norm": 0.0778389424085617,
+      "learning_rate": 1.5311220126918405e-05,
+      "loss": 0.2208,
+      "step": 45974
+    },
+    {
+      "epoch": 3.7244815294880103,
+      "grad_norm": 0.06476572155952454,
+      "learning_rate": 1.530671947432378e-05,
+      "loss": 0.2047,
+      "step": 45975
+    },
+    {
+      "epoch": 3.7245625405055085,
+      "grad_norm": 0.07310286909341812,
+      "learning_rate": 1.530221882172915e-05,
+      "loss": 0.2068,
+      "step": 45976
+    },
+    {
+      "epoch": 3.724643551523007,
+      "grad_norm": 0.06835059821605682,
+      "learning_rate": 1.5297718169134526e-05,
+      "loss": 0.2077,
+      "step": 45977
+    },
+    {
+      "epoch": 3.7247245625405054,
+      "grad_norm": 0.08028055727481842,
+      "learning_rate": 1.52932175165399e-05,
+      "loss": 0.2292,
+      "step": 45978
+    },
+    {
+      "epoch": 3.7248055735580037,
+      "grad_norm": 0.06749919801950455,
+      "learning_rate": 1.5288716863945273e-05,
+      "loss": 0.1826,
+      "step": 45979
+    },
+    {
+      "epoch": 3.7248865845755024,
+      "grad_norm": 0.05876849219202995,
+      "learning_rate": 1.5284216211350647e-05,
+      "loss": 0.2335,
+      "step": 45980
+    },
+    {
+      "epoch": 3.7249675955930006,
+      "grad_norm": 0.07642512768507004,
+      "learning_rate": 1.527971555875602e-05,
+      "loss": 0.2121,
+      "step": 45981
+    },
+    {
+      "epoch": 3.725048606610499,
+      "grad_norm": 0.06255366653203964,
+      "learning_rate": 1.5275214906161394e-05,
+      "loss": 0.2031,
+      "step": 45982
+    },
+    {
+      "epoch": 3.7251296176279975,
+      "grad_norm": 0.06796018779277802,
+      "learning_rate": 1.5270714253566768e-05,
+      "loss": 0.2456,
+      "step": 45983
+    },
+    {
+      "epoch": 3.725210628645496,
+      "grad_norm": 0.0734555795788765,
+      "learning_rate": 1.526621360097214e-05,
+      "loss": 0.2187,
+      "step": 45984
+    },
+    {
+      "epoch": 3.725291639662994,
+      "grad_norm": 0.06981045752763748,
+      "learning_rate": 1.5261712948377515e-05,
+      "loss": 0.2197,
+      "step": 45985
+    },
+    {
+      "epoch": 3.7253726506804927,
+      "grad_norm": 0.07305343449115753,
+      "learning_rate": 1.5257212295782888e-05,
+      "loss": 0.2002,
+      "step": 45986
+    },
+    {
+      "epoch": 3.725453661697991,
+      "grad_norm": 0.08334986120462418,
+      "learning_rate": 1.5252711643188264e-05,
+      "loss": 0.2253,
+      "step": 45987
+    },
+    {
+      "epoch": 3.725534672715489,
+      "grad_norm": 0.08052529394626617,
+      "learning_rate": 1.5248210990593637e-05,
+      "loss": 0.2338,
+      "step": 45988
+    },
+    {
+      "epoch": 3.725615683732988,
+      "grad_norm": 0.08272536844015121,
+      "learning_rate": 1.524371033799901e-05,
+      "loss": 0.2334,
+      "step": 45989
+    },
+    {
+      "epoch": 3.725696694750486,
+      "grad_norm": 0.07033362984657288,
+      "learning_rate": 1.5239209685404385e-05,
+      "loss": 0.2335,
+      "step": 45990
+    },
+    {
+      "epoch": 3.7257777057679844,
+      "grad_norm": 0.06358537077903748,
+      "learning_rate": 1.5234709032809758e-05,
+      "loss": 0.2542,
+      "step": 45991
+    },
+    {
+      "epoch": 3.725858716785483,
+      "grad_norm": 0.08354093134403229,
+      "learning_rate": 1.5230208380215133e-05,
+      "loss": 0.2444,
+      "step": 45992
+    },
+    {
+      "epoch": 3.7259397278029813,
+      "grad_norm": 0.07609385251998901,
+      "learning_rate": 1.5225707727620505e-05,
+      "loss": 0.2156,
+      "step": 45993
+    },
+    {
+      "epoch": 3.7260207388204796,
+      "grad_norm": 0.07945896685123444,
+      "learning_rate": 1.5221207075025879e-05,
+      "loss": 0.2156,
+      "step": 45994
+    },
+    {
+      "epoch": 3.726101749837978,
+      "grad_norm": 0.07363783568143845,
+      "learning_rate": 1.5216706422431254e-05,
+      "loss": 0.2019,
+      "step": 45995
+    },
+    {
+      "epoch": 3.7261827608554765,
+      "grad_norm": 0.07394561171531677,
+      "learning_rate": 1.5212205769836626e-05,
+      "loss": 0.1975,
+      "step": 45996
+    },
+    {
+      "epoch": 3.7262637718729748,
+      "grad_norm": 0.0713035985827446,
+      "learning_rate": 1.5207705117242001e-05,
+      "loss": 0.2476,
+      "step": 45997
+    },
+    {
+      "epoch": 3.726344782890473,
+      "grad_norm": 0.06830364465713501,
+      "learning_rate": 1.5203204464647375e-05,
+      "loss": 0.2395,
+      "step": 45998
+    },
+    {
+      "epoch": 3.7264257939079712,
+      "grad_norm": 0.07303628325462341,
+      "learning_rate": 1.5198703812052747e-05,
+      "loss": 0.2143,
+      "step": 45999
+    },
+    {
+      "epoch": 3.72650680492547,
+      "grad_norm": 0.07435938715934753,
+      "learning_rate": 1.5194203159458122e-05,
+      "loss": 0.2648,
+      "step": 46000
+    },
+    {
+      "epoch": 3.726587815942968,
+      "grad_norm": 0.0673922747373581,
+      "learning_rate": 1.5189702506863496e-05,
+      "loss": 0.2256,
+      "step": 46001
+    },
+    {
+      "epoch": 3.7266688269604664,
+      "grad_norm": 0.06162109971046448,
+      "learning_rate": 1.5185201854268868e-05,
+      "loss": 0.2254,
+      "step": 46002
+    },
+    {
+      "epoch": 3.726749837977965,
+      "grad_norm": 0.10011584311723709,
+      "learning_rate": 1.5180701201674243e-05,
+      "loss": 0.2665,
+      "step": 46003
+    },
+    {
+      "epoch": 3.7268308489954634,
+      "grad_norm": 0.07240945100784302,
+      "learning_rate": 1.5176200549079617e-05,
+      "loss": 0.2359,
+      "step": 46004
+    },
+    {
+      "epoch": 3.7269118600129616,
+      "grad_norm": 0.07601157575845718,
+      "learning_rate": 1.5171699896484992e-05,
+      "loss": 0.2338,
+      "step": 46005
+    },
+    {
+      "epoch": 3.7269928710304603,
+      "grad_norm": 0.08337153494358063,
+      "learning_rate": 1.5167199243890366e-05,
+      "loss": 0.2743,
+      "step": 46006
+    },
+    {
+      "epoch": 3.7270738820479585,
+      "grad_norm": 0.07271520048379898,
+      "learning_rate": 1.5162698591295738e-05,
+      "loss": 0.2244,
+      "step": 46007
+    },
+    {
+      "epoch": 3.7271548930654568,
+      "grad_norm": 0.07792726159095764,
+      "learning_rate": 1.5158197938701113e-05,
+      "loss": 0.247,
+      "step": 46008
+    },
+    {
+      "epoch": 3.7272359040829555,
+      "grad_norm": 0.07396295666694641,
+      "learning_rate": 1.5153697286106486e-05,
+      "loss": 0.265,
+      "step": 46009
+    },
+    {
+      "epoch": 3.7273169151004537,
+      "grad_norm": 0.06522729247808456,
+      "learning_rate": 1.5149196633511862e-05,
+      "loss": 0.2406,
+      "step": 46010
+    },
+    {
+      "epoch": 3.727397926117952,
+      "grad_norm": 0.07633328437805176,
+      "learning_rate": 1.5144695980917234e-05,
+      "loss": 0.2499,
+      "step": 46011
+    },
+    {
+      "epoch": 3.7274789371354506,
+      "grad_norm": 0.07195275276899338,
+      "learning_rate": 1.5140195328322607e-05,
+      "loss": 0.2147,
+      "step": 46012
+    },
+    {
+      "epoch": 3.727559948152949,
+      "grad_norm": 0.07411804050207138,
+      "learning_rate": 1.5135694675727983e-05,
+      "loss": 0.2065,
+      "step": 46013
+    },
+    {
+      "epoch": 3.727640959170447,
+      "grad_norm": 0.06602322310209274,
+      "learning_rate": 1.5131194023133354e-05,
+      "loss": 0.2006,
+      "step": 46014
+    },
+    {
+      "epoch": 3.727721970187946,
+      "grad_norm": 0.07067742198705673,
+      "learning_rate": 1.5126693370538728e-05,
+      "loss": 0.2145,
+      "step": 46015
+    },
+    {
+      "epoch": 3.727802981205444,
+      "grad_norm": 0.06092274561524391,
+      "learning_rate": 1.5122192717944103e-05,
+      "loss": 0.2087,
+      "step": 46016
+    },
+    {
+      "epoch": 3.7278839922229423,
+      "grad_norm": 0.07934263348579407,
+      "learning_rate": 1.5117692065349475e-05,
+      "loss": 0.2166,
+      "step": 46017
+    },
+    {
+      "epoch": 3.7279650032404406,
+      "grad_norm": 0.07236896455287933,
+      "learning_rate": 1.511319141275485e-05,
+      "loss": 0.2496,
+      "step": 46018
+    },
+    {
+      "epoch": 3.7280460142579392,
+      "grad_norm": 0.07020833343267441,
+      "learning_rate": 1.5108690760160224e-05,
+      "loss": 0.2294,
+      "step": 46019
+    },
+    {
+      "epoch": 3.7281270252754375,
+      "grad_norm": 0.0577675886452198,
+      "learning_rate": 1.5104190107565596e-05,
+      "loss": 0.2161,
+      "step": 46020
+    },
+    {
+      "epoch": 3.7282080362929357,
+      "grad_norm": 0.06877872347831726,
+      "learning_rate": 1.5099689454970973e-05,
+      "loss": 0.2433,
+      "step": 46021
+    },
+    {
+      "epoch": 3.728289047310434,
+      "grad_norm": 0.05976353585720062,
+      "learning_rate": 1.5095188802376345e-05,
+      "loss": 0.2133,
+      "step": 46022
+    },
+    {
+      "epoch": 3.7283700583279327,
+      "grad_norm": 0.07601095736026764,
+      "learning_rate": 1.509068814978172e-05,
+      "loss": 0.2678,
+      "step": 46023
+    },
+    {
+      "epoch": 3.728451069345431,
+      "grad_norm": 0.09052573144435883,
+      "learning_rate": 1.5086187497187094e-05,
+      "loss": 0.2468,
+      "step": 46024
+    },
+    {
+      "epoch": 3.728532080362929,
+      "grad_norm": 0.0643225833773613,
+      "learning_rate": 1.5081686844592466e-05,
+      "loss": 0.2259,
+      "step": 46025
+    },
+    {
+      "epoch": 3.728613091380428,
+      "grad_norm": 0.07924170792102814,
+      "learning_rate": 1.5077186191997841e-05,
+      "loss": 0.2399,
+      "step": 46026
+    },
+    {
+      "epoch": 3.728694102397926,
+      "grad_norm": 0.07029829919338226,
+      "learning_rate": 1.5072685539403215e-05,
+      "loss": 0.2177,
+      "step": 46027
+    },
+    {
+      "epoch": 3.7287751134154243,
+      "grad_norm": 0.07222612202167511,
+      "learning_rate": 1.5068184886808587e-05,
+      "loss": 0.2465,
+      "step": 46028
+    },
+    {
+      "epoch": 3.728856124432923,
+      "grad_norm": 0.08449197560548782,
+      "learning_rate": 1.5063684234213962e-05,
+      "loss": 0.2017,
+      "step": 46029
+    },
+    {
+      "epoch": 3.7289371354504213,
+      "grad_norm": 0.05938245728611946,
+      "learning_rate": 1.5059183581619335e-05,
+      "loss": 0.2076,
+      "step": 46030
+    },
+    {
+      "epoch": 3.7290181464679195,
+      "grad_norm": 0.07024046778678894,
+      "learning_rate": 1.505468292902471e-05,
+      "loss": 0.2041,
+      "step": 46031
+    },
+    {
+      "epoch": 3.729099157485418,
+      "grad_norm": 0.07162220031023026,
+      "learning_rate": 1.5050182276430083e-05,
+      "loss": 0.2144,
+      "step": 46032
+    },
+    {
+      "epoch": 3.7291801685029164,
+      "grad_norm": 0.06670884788036346,
+      "learning_rate": 1.5045681623835456e-05,
+      "loss": 0.2002,
+      "step": 46033
+    },
+    {
+      "epoch": 3.7292611795204147,
+      "grad_norm": 0.0706491693854332,
+      "learning_rate": 1.5041180971240832e-05,
+      "loss": 0.1878,
+      "step": 46034
+    },
+    {
+      "epoch": 3.7293421905379134,
+      "grad_norm": 0.07656066864728928,
+      "learning_rate": 1.5036680318646203e-05,
+      "loss": 0.1964,
+      "step": 46035
+    },
+    {
+      "epoch": 3.7294232015554116,
+      "grad_norm": 0.07184910029172897,
+      "learning_rate": 1.5032179666051579e-05,
+      "loss": 0.2506,
+      "step": 46036
+    },
+    {
+      "epoch": 3.72950421257291,
+      "grad_norm": 0.07077179104089737,
+      "learning_rate": 1.5027679013456952e-05,
+      "loss": 0.208,
+      "step": 46037
+    },
+    {
+      "epoch": 3.7295852235904086,
+      "grad_norm": 0.0829695537686348,
+      "learning_rate": 1.5023178360862324e-05,
+      "loss": 0.229,
+      "step": 46038
+    },
+    {
+      "epoch": 3.729666234607907,
+      "grad_norm": 0.06951703131198883,
+      "learning_rate": 1.5018677708267701e-05,
+      "loss": 0.2148,
+      "step": 46039
+    },
+    {
+      "epoch": 3.729747245625405,
+      "grad_norm": 0.0725255087018013,
+      "learning_rate": 1.5014177055673073e-05,
+      "loss": 0.2519,
+      "step": 46040
+    },
+    {
+      "epoch": 3.7298282566429033,
+      "grad_norm": 0.06751804798841476,
+      "learning_rate": 1.5009676403078445e-05,
+      "loss": 0.223,
+      "step": 46041
+    },
+    {
+      "epoch": 3.7299092676604015,
+      "grad_norm": 0.08554006367921829,
+      "learning_rate": 1.5005175750483822e-05,
+      "loss": 0.2673,
+      "step": 46042
+    },
+    {
+      "epoch": 3.7299902786779002,
+      "grad_norm": 0.06487467885017395,
+      "learning_rate": 1.5000675097889194e-05,
+      "loss": 0.2243,
+      "step": 46043
+    },
+    {
+      "epoch": 3.7300712896953985,
+      "grad_norm": 0.08155498653650284,
+      "learning_rate": 1.499617444529457e-05,
+      "loss": 0.2768,
+      "step": 46044
+    },
+    {
+      "epoch": 3.7301523007128967,
+      "grad_norm": 0.06040952354669571,
+      "learning_rate": 1.4991673792699943e-05,
+      "loss": 0.2139,
+      "step": 46045
+    },
+    {
+      "epoch": 3.7302333117303954,
+      "grad_norm": 0.07658139616250992,
+      "learning_rate": 1.4987173140105315e-05,
+      "loss": 0.2356,
+      "step": 46046
+    },
+    {
+      "epoch": 3.7303143227478937,
+      "grad_norm": 0.08683796226978302,
+      "learning_rate": 1.498267248751069e-05,
+      "loss": 0.2153,
+      "step": 46047
+    },
+    {
+      "epoch": 3.730395333765392,
+      "grad_norm": 0.08210854977369308,
+      "learning_rate": 1.4978171834916064e-05,
+      "loss": 0.2168,
+      "step": 46048
+    },
+    {
+      "epoch": 3.7304763447828906,
+      "grad_norm": 0.08197548985481262,
+      "learning_rate": 1.4973671182321436e-05,
+      "loss": 0.2389,
+      "step": 46049
+    },
+    {
+      "epoch": 3.730557355800389,
+      "grad_norm": 0.06574133038520813,
+      "learning_rate": 1.4969170529726811e-05,
+      "loss": 0.2428,
+      "step": 46050
+    },
+    {
+      "epoch": 3.730638366817887,
+      "grad_norm": 0.082312673330307,
+      "learning_rate": 1.4964669877132185e-05,
+      "loss": 0.2285,
+      "step": 46051
+    },
+    {
+      "epoch": 3.7307193778353858,
+      "grad_norm": 0.0676484927535057,
+      "learning_rate": 1.496016922453756e-05,
+      "loss": 0.2057,
+      "step": 46052
+    },
+    {
+      "epoch": 3.730800388852884,
+      "grad_norm": 0.06505704671144485,
+      "learning_rate": 1.4955668571942932e-05,
+      "loss": 0.2257,
+      "step": 46053
+    },
+    {
+      "epoch": 3.7308813998703823,
+      "grad_norm": 0.0626121386885643,
+      "learning_rate": 1.4951167919348305e-05,
+      "loss": 0.2512,
+      "step": 46054
+    },
+    {
+      "epoch": 3.730962410887881,
+      "grad_norm": 0.07555672526359558,
+      "learning_rate": 1.494666726675368e-05,
+      "loss": 0.1947,
+      "step": 46055
+    },
+    {
+      "epoch": 3.731043421905379,
+      "grad_norm": 0.05888738855719566,
+      "learning_rate": 1.4942166614159053e-05,
+      "loss": 0.208,
+      "step": 46056
+    },
+    {
+      "epoch": 3.7311244329228774,
+      "grad_norm": 0.0674942135810852,
+      "learning_rate": 1.493766596156443e-05,
+      "loss": 0.2275,
+      "step": 46057
+    },
+    {
+      "epoch": 3.731205443940376,
+      "grad_norm": 0.05916551873087883,
+      "learning_rate": 1.4933165308969801e-05,
+      "loss": 0.1929,
+      "step": 46058
+    },
+    {
+      "epoch": 3.7312864549578744,
+      "grad_norm": 0.08539033681154251,
+      "learning_rate": 1.4928664656375175e-05,
+      "loss": 0.2194,
+      "step": 46059
+    },
+    {
+      "epoch": 3.7313674659753726,
+      "grad_norm": 0.06798946112394333,
+      "learning_rate": 1.492416400378055e-05,
+      "loss": 0.2181,
+      "step": 46060
+    },
+    {
+      "epoch": 3.7314484769928713,
+      "grad_norm": 0.06484483182430267,
+      "learning_rate": 1.4919663351185922e-05,
+      "loss": 0.2043,
+      "step": 46061
+    },
+    {
+      "epoch": 3.7315294880103695,
+      "grad_norm": 0.0604698583483696,
+      "learning_rate": 1.4915162698591296e-05,
+      "loss": 0.1975,
+      "step": 46062
+    },
+    {
+      "epoch": 3.731610499027868,
+      "grad_norm": 0.0822327584028244,
+      "learning_rate": 1.4910662045996671e-05,
+      "loss": 0.2609,
+      "step": 46063
+    },
+    {
+      "epoch": 3.731691510045366,
+      "grad_norm": 0.09345957636833191,
+      "learning_rate": 1.4906161393402043e-05,
+      "loss": 0.2302,
+      "step": 46064
+    },
+    {
+      "epoch": 3.7317725210628643,
+      "grad_norm": 0.06493161618709564,
+      "learning_rate": 1.4901660740807418e-05,
+      "loss": 0.1933,
+      "step": 46065
+    },
+    {
+      "epoch": 3.731853532080363,
+      "grad_norm": 0.0702953115105629,
+      "learning_rate": 1.4897160088212792e-05,
+      "loss": 0.1897,
+      "step": 46066
+    },
+    {
+      "epoch": 3.731934543097861,
+      "grad_norm": 0.06461524218320847,
+      "learning_rate": 1.4892659435618164e-05,
+      "loss": 0.2407,
+      "step": 46067
+    },
+    {
+      "epoch": 3.7320155541153595,
+      "grad_norm": 0.08654788136482239,
+      "learning_rate": 1.488815878302354e-05,
+      "loss": 0.236,
+      "step": 46068
+    },
+    {
+      "epoch": 3.732096565132858,
+      "grad_norm": 0.0883837565779686,
+      "learning_rate": 1.4883658130428913e-05,
+      "loss": 0.2058,
+      "step": 46069
+    },
+    {
+      "epoch": 3.7321775761503564,
+      "grad_norm": 0.06732512265443802,
+      "learning_rate": 1.4879157477834288e-05,
+      "loss": 0.2224,
+      "step": 46070
+    },
+    {
+      "epoch": 3.7322585871678546,
+      "grad_norm": 0.06484067440032959,
+      "learning_rate": 1.487465682523966e-05,
+      "loss": 0.2419,
+      "step": 46071
+    },
+    {
+      "epoch": 3.7323395981853533,
+      "grad_norm": 0.0646497905254364,
+      "learning_rate": 1.4870156172645034e-05,
+      "loss": 0.1876,
+      "step": 46072
+    },
+    {
+      "epoch": 3.7324206092028516,
+      "grad_norm": 0.101152203977108,
+      "learning_rate": 1.4865655520050409e-05,
+      "loss": 0.2442,
+      "step": 46073
+    },
+    {
+      "epoch": 3.73250162022035,
+      "grad_norm": 0.08255450427532196,
+      "learning_rate": 1.486115486745578e-05,
+      "loss": 0.2307,
+      "step": 46074
+    },
+    {
+      "epoch": 3.7325826312378485,
+      "grad_norm": 0.08730298280715942,
+      "learning_rate": 1.4856654214861154e-05,
+      "loss": 0.2408,
+      "step": 46075
+    },
+    {
+      "epoch": 3.7326636422553467,
+      "grad_norm": 0.0731593444943428,
+      "learning_rate": 1.485215356226653e-05,
+      "loss": 0.1948,
+      "step": 46076
+    },
+    {
+      "epoch": 3.732744653272845,
+      "grad_norm": 0.07262025028467178,
+      "learning_rate": 1.4847652909671903e-05,
+      "loss": 0.2244,
+      "step": 46077
+    },
+    {
+      "epoch": 3.7328256642903437,
+      "grad_norm": 0.07813366502523422,
+      "learning_rate": 1.4843152257077279e-05,
+      "loss": 0.2623,
+      "step": 46078
+    },
+    {
+      "epoch": 3.732906675307842,
+      "grad_norm": 0.06445509940385818,
+      "learning_rate": 1.483865160448265e-05,
+      "loss": 0.2185,
+      "step": 46079
+    },
+    {
+      "epoch": 3.73298768632534,
+      "grad_norm": 0.07868556678295135,
+      "learning_rate": 1.4834150951888024e-05,
+      "loss": 0.212,
+      "step": 46080
+    },
+    {
+      "epoch": 3.733068697342839,
+      "grad_norm": 0.0866679698228836,
+      "learning_rate": 1.48296502992934e-05,
+      "loss": 0.2424,
+      "step": 46081
+    },
+    {
+      "epoch": 3.733149708360337,
+      "grad_norm": 0.06397400051355362,
+      "learning_rate": 1.4825149646698771e-05,
+      "loss": 0.2272,
+      "step": 46082
+    },
+    {
+      "epoch": 3.7332307193778353,
+      "grad_norm": 0.07194807380437851,
+      "learning_rate": 1.4820648994104147e-05,
+      "loss": 0.2261,
+      "step": 46083
+    },
+    {
+      "epoch": 3.733311730395334,
+      "grad_norm": 0.06789196282625198,
+      "learning_rate": 1.481614834150952e-05,
+      "loss": 0.2246,
+      "step": 46084
+    },
+    {
+      "epoch": 3.7333927414128323,
+      "grad_norm": 0.07250261306762695,
+      "learning_rate": 1.4811647688914892e-05,
+      "loss": 0.2306,
+      "step": 46085
+    },
+    {
+      "epoch": 3.7334737524303305,
+      "grad_norm": 0.10532908141613007,
+      "learning_rate": 1.4807147036320267e-05,
+      "loss": 0.2482,
+      "step": 46086
+    },
+    {
+      "epoch": 3.7335547634478288,
+      "grad_norm": 0.062470242381095886,
+      "learning_rate": 1.4802646383725641e-05,
+      "loss": 0.1862,
+      "step": 46087
+    },
+    {
+      "epoch": 3.733635774465327,
+      "grad_norm": 0.07458098232746124,
+      "learning_rate": 1.4798145731131013e-05,
+      "loss": 0.2172,
+      "step": 46088
+    },
+    {
+      "epoch": 3.7337167854828257,
+      "grad_norm": 0.07555191963911057,
+      "learning_rate": 1.4793645078536388e-05,
+      "loss": 0.2453,
+      "step": 46089
+    },
+    {
+      "epoch": 3.733797796500324,
+      "grad_norm": 0.08217816799879074,
+      "learning_rate": 1.4789144425941762e-05,
+      "loss": 0.2386,
+      "step": 46090
+    },
+    {
+      "epoch": 3.733878807517822,
+      "grad_norm": 0.07929562777280807,
+      "learning_rate": 1.4784643773347137e-05,
+      "loss": 0.2048,
+      "step": 46091
+    },
+    {
+      "epoch": 3.733959818535321,
+      "grad_norm": 0.09434565156698227,
+      "learning_rate": 1.4780143120752509e-05,
+      "loss": 0.238,
+      "step": 46092
+    },
+    {
+      "epoch": 3.734040829552819,
+      "grad_norm": 0.07045267522335052,
+      "learning_rate": 1.4775642468157883e-05,
+      "loss": 0.2308,
+      "step": 46093
+    },
+    {
+      "epoch": 3.7341218405703174,
+      "grad_norm": 0.0724666640162468,
+      "learning_rate": 1.4771141815563258e-05,
+      "loss": 0.1918,
+      "step": 46094
+    },
+    {
+      "epoch": 3.734202851587816,
+      "grad_norm": 0.08512706309556961,
+      "learning_rate": 1.4766641162968632e-05,
+      "loss": 0.2463,
+      "step": 46095
+    },
+    {
+      "epoch": 3.7342838626053143,
+      "grad_norm": 0.06601796299219131,
+      "learning_rate": 1.4762140510374007e-05,
+      "loss": 0.2152,
+      "step": 46096
+    },
+    {
+      "epoch": 3.7343648736228126,
+      "grad_norm": 0.0734667181968689,
+      "learning_rate": 1.4757639857779379e-05,
+      "loss": 0.2308,
+      "step": 46097
+    },
+    {
+      "epoch": 3.7344458846403112,
+      "grad_norm": 0.076157346367836,
+      "learning_rate": 1.4753139205184752e-05,
+      "loss": 0.1982,
+      "step": 46098
+    },
+    {
+      "epoch": 3.7345268956578095,
+      "grad_norm": 0.06739217787981033,
+      "learning_rate": 1.4748638552590128e-05,
+      "loss": 0.267,
+      "step": 46099
+    },
+    {
+      "epoch": 3.7346079066753077,
+      "grad_norm": 0.06588942557573318,
+      "learning_rate": 1.47441378999955e-05,
+      "loss": 0.2426,
+      "step": 46100
+    },
+    {
+      "epoch": 3.7346889176928064,
+      "grad_norm": 0.08635196089744568,
+      "learning_rate": 1.4739637247400873e-05,
+      "loss": 0.27,
+      "step": 46101
+    },
+    {
+      "epoch": 3.7347699287103047,
+      "grad_norm": 0.06514254957437515,
+      "learning_rate": 1.4735136594806248e-05,
+      "loss": 0.2243,
+      "step": 46102
+    },
+    {
+      "epoch": 3.734850939727803,
+      "grad_norm": 0.06313839554786682,
+      "learning_rate": 1.473063594221162e-05,
+      "loss": 0.2219,
+      "step": 46103
+    },
+    {
+      "epoch": 3.7349319507453016,
+      "grad_norm": 0.06508518010377884,
+      "learning_rate": 1.4726135289616996e-05,
+      "loss": 0.2105,
+      "step": 46104
+    },
+    {
+      "epoch": 3.7350129617628,
+      "grad_norm": 0.06646306067705154,
+      "learning_rate": 1.472163463702237e-05,
+      "loss": 0.2061,
+      "step": 46105
+    },
+    {
+      "epoch": 3.735093972780298,
+      "grad_norm": 0.07501812279224396,
+      "learning_rate": 1.4717133984427741e-05,
+      "loss": 0.2194,
+      "step": 46106
+    },
+    {
+      "epoch": 3.7351749837977968,
+      "grad_norm": 0.08135095983743668,
+      "learning_rate": 1.4712633331833116e-05,
+      "loss": 0.2341,
+      "step": 46107
+    },
+    {
+      "epoch": 3.735255994815295,
+      "grad_norm": 0.07120555639266968,
+      "learning_rate": 1.470813267923849e-05,
+      "loss": 0.23,
+      "step": 46108
+    },
+    {
+      "epoch": 3.7353370058327933,
+      "grad_norm": 0.05750159174203873,
+      "learning_rate": 1.4703632026643865e-05,
+      "loss": 0.2264,
+      "step": 46109
+    },
+    {
+      "epoch": 3.7354180168502915,
+      "grad_norm": 0.060754384845495224,
+      "learning_rate": 1.4699131374049237e-05,
+      "loss": 0.2099,
+      "step": 46110
+    },
+    {
+      "epoch": 3.7354990278677898,
+      "grad_norm": 0.08677901327610016,
+      "learning_rate": 1.4694630721454611e-05,
+      "loss": 0.2165,
+      "step": 46111
+    },
+    {
+      "epoch": 3.7355800388852884,
+      "grad_norm": 0.07808344811201096,
+      "learning_rate": 1.4690130068859986e-05,
+      "loss": 0.2626,
+      "step": 46112
+    },
+    {
+      "epoch": 3.7356610499027867,
+      "grad_norm": 0.08093435317277908,
+      "learning_rate": 1.468562941626536e-05,
+      "loss": 0.2381,
+      "step": 46113
+    },
+    {
+      "epoch": 3.735742060920285,
+      "grad_norm": 0.07375194877386093,
+      "learning_rate": 1.4681128763670732e-05,
+      "loss": 0.196,
+      "step": 46114
+    },
+    {
+      "epoch": 3.7358230719377836,
+      "grad_norm": 0.07342427223920822,
+      "learning_rate": 1.4676628111076107e-05,
+      "loss": 0.1924,
+      "step": 46115
+    },
+    {
+      "epoch": 3.735904082955282,
+      "grad_norm": 0.06829115003347397,
+      "learning_rate": 1.467212745848148e-05,
+      "loss": 0.2165,
+      "step": 46116
+    },
+    {
+      "epoch": 3.73598509397278,
+      "grad_norm": 0.07790973037481308,
+      "learning_rate": 1.4667626805886856e-05,
+      "loss": 0.2195,
+      "step": 46117
+    },
+    {
+      "epoch": 3.736066104990279,
+      "grad_norm": 0.06781540811061859,
+      "learning_rate": 1.4663126153292228e-05,
+      "loss": 0.1872,
+      "step": 46118
+    },
+    {
+      "epoch": 3.736147116007777,
+      "grad_norm": 0.08341260254383087,
+      "learning_rate": 1.4658625500697601e-05,
+      "loss": 0.2518,
+      "step": 46119
+    },
+    {
+      "epoch": 3.7362281270252753,
+      "grad_norm": 0.08755706995725632,
+      "learning_rate": 1.4654124848102977e-05,
+      "loss": 0.268,
+      "step": 46120
+    },
+    {
+      "epoch": 3.736309138042774,
+      "grad_norm": 0.08430002629756927,
+      "learning_rate": 1.4649624195508349e-05,
+      "loss": 0.2045,
+      "step": 46121
+    },
+    {
+      "epoch": 3.7363901490602722,
+      "grad_norm": 0.09913341701030731,
+      "learning_rate": 1.4645123542913724e-05,
+      "loss": 0.2489,
+      "step": 46122
+    },
+    {
+      "epoch": 3.7364711600777705,
+      "grad_norm": 0.08157473057508469,
+      "learning_rate": 1.4640622890319098e-05,
+      "loss": 0.2011,
+      "step": 46123
+    },
+    {
+      "epoch": 3.736552171095269,
+      "grad_norm": 0.06985532492399216,
+      "learning_rate": 1.463612223772447e-05,
+      "loss": 0.1987,
+      "step": 46124
+    },
+    {
+      "epoch": 3.7366331821127674,
+      "grad_norm": 0.0689232274889946,
+      "learning_rate": 1.4631621585129845e-05,
+      "loss": 0.1943,
+      "step": 46125
+    },
+    {
+      "epoch": 3.7367141931302656,
+      "grad_norm": 0.07660727947950363,
+      "learning_rate": 1.4627120932535218e-05,
+      "loss": 0.2355,
+      "step": 46126
+    },
+    {
+      "epoch": 3.7367952041477643,
+      "grad_norm": 0.07977120578289032,
+      "learning_rate": 1.462262027994059e-05,
+      "loss": 0.2234,
+      "step": 46127
+    },
+    {
+      "epoch": 3.7368762151652626,
+      "grad_norm": 0.09145716577768326,
+      "learning_rate": 1.4618119627345967e-05,
+      "loss": 0.2729,
+      "step": 46128
+    },
+    {
+      "epoch": 3.736957226182761,
+      "grad_norm": 0.06459056586027145,
+      "learning_rate": 1.4613618974751339e-05,
+      "loss": 0.2561,
+      "step": 46129
+    },
+    {
+      "epoch": 3.737038237200259,
+      "grad_norm": 0.07937449961900711,
+      "learning_rate": 1.4609118322156714e-05,
+      "loss": 0.2047,
+      "step": 46130
+    },
+    {
+      "epoch": 3.7371192482177578,
+      "grad_norm": 0.08180252462625504,
+      "learning_rate": 1.4604617669562088e-05,
+      "loss": 0.2194,
+      "step": 46131
+    },
+    {
+      "epoch": 3.737200259235256,
+      "grad_norm": 0.06420495361089706,
+      "learning_rate": 1.460011701696746e-05,
+      "loss": 0.1774,
+      "step": 46132
+    },
+    {
+      "epoch": 3.7372812702527543,
+      "grad_norm": 0.0610925629734993,
+      "learning_rate": 1.4595616364372835e-05,
+      "loss": 0.191,
+      "step": 46133
+    },
+    {
+      "epoch": 3.7373622812702525,
+      "grad_norm": 0.06337723135948181,
+      "learning_rate": 1.4591115711778209e-05,
+      "loss": 0.2092,
+      "step": 46134
+    },
+    {
+      "epoch": 3.737443292287751,
+      "grad_norm": 0.0656128078699112,
+      "learning_rate": 1.458661505918358e-05,
+      "loss": 0.2142,
+      "step": 46135
+    },
+    {
+      "epoch": 3.7375243033052494,
+      "grad_norm": 0.0756414532661438,
+      "learning_rate": 1.4582114406588956e-05,
+      "loss": 0.2215,
+      "step": 46136
+    },
+    {
+      "epoch": 3.7376053143227477,
+      "grad_norm": 0.06451795995235443,
+      "learning_rate": 1.457761375399433e-05,
+      "loss": 0.2521,
+      "step": 46137
+    },
+    {
+      "epoch": 3.7376863253402464,
+      "grad_norm": 0.07743828743696213,
+      "learning_rate": 1.4573113101399705e-05,
+      "loss": 0.2147,
+      "step": 46138
+    },
+    {
+      "epoch": 3.7377673363577446,
+      "grad_norm": 0.0701700821518898,
+      "learning_rate": 1.4568612448805077e-05,
+      "loss": 0.2088,
+      "step": 46139
+    },
+    {
+      "epoch": 3.737848347375243,
+      "grad_norm": 0.08126933127641678,
+      "learning_rate": 1.456411179621045e-05,
+      "loss": 0.251,
+      "step": 46140
+    },
+    {
+      "epoch": 3.7379293583927415,
+      "grad_norm": 0.07062458992004395,
+      "learning_rate": 1.4559611143615826e-05,
+      "loss": 0.2112,
+      "step": 46141
+    },
+    {
+      "epoch": 3.73801036941024,
+      "grad_norm": 0.08064697682857513,
+      "learning_rate": 1.4555110491021198e-05,
+      "loss": 0.2212,
+      "step": 46142
+    },
+    {
+      "epoch": 3.738091380427738,
+      "grad_norm": 0.06433752179145813,
+      "learning_rate": 1.4550609838426573e-05,
+      "loss": 0.2224,
+      "step": 46143
+    },
+    {
+      "epoch": 3.7381723914452367,
+      "grad_norm": 0.0687873363494873,
+      "learning_rate": 1.4546109185831947e-05,
+      "loss": 0.2246,
+      "step": 46144
+    },
+    {
+      "epoch": 3.738253402462735,
+      "grad_norm": 0.06918583065271378,
+      "learning_rate": 1.4541608533237318e-05,
+      "loss": 0.211,
+      "step": 46145
+    },
+    {
+      "epoch": 3.738334413480233,
+      "grad_norm": 0.07868772000074387,
+      "learning_rate": 1.4537107880642695e-05,
+      "loss": 0.2275,
+      "step": 46146
+    },
+    {
+      "epoch": 3.738415424497732,
+      "grad_norm": 0.06755036860704422,
+      "learning_rate": 1.4532607228048067e-05,
+      "loss": 0.195,
+      "step": 46147
+    },
+    {
+      "epoch": 3.73849643551523,
+      "grad_norm": 0.06591461598873138,
+      "learning_rate": 1.452810657545344e-05,
+      "loss": 0.1961,
+      "step": 46148
+    },
+    {
+      "epoch": 3.7385774465327284,
+      "grad_norm": 0.090169258415699,
+      "learning_rate": 1.4523605922858816e-05,
+      "loss": 0.243,
+      "step": 46149
+    },
+    {
+      "epoch": 3.738658457550227,
+      "grad_norm": 0.07051538676023483,
+      "learning_rate": 1.4519105270264188e-05,
+      "loss": 0.1963,
+      "step": 46150
+    },
+    {
+      "epoch": 3.7387394685677253,
+      "grad_norm": 0.08839746564626694,
+      "learning_rate": 1.4514604617669563e-05,
+      "loss": 0.2587,
+      "step": 46151
+    },
+    {
+      "epoch": 3.7388204795852236,
+      "grad_norm": 0.06953133642673492,
+      "learning_rate": 1.4510103965074937e-05,
+      "loss": 0.1979,
+      "step": 46152
+    },
+    {
+      "epoch": 3.738901490602722,
+      "grad_norm": 0.08543260395526886,
+      "learning_rate": 1.4505603312480309e-05,
+      "loss": 0.2256,
+      "step": 46153
+    },
+    {
+      "epoch": 3.7389825016202205,
+      "grad_norm": 0.06941500306129456,
+      "learning_rate": 1.4501102659885684e-05,
+      "loss": 0.2061,
+      "step": 46154
+    },
+    {
+      "epoch": 3.7390635126377187,
+      "grad_norm": 0.05090836063027382,
+      "learning_rate": 1.4496602007291058e-05,
+      "loss": 0.2119,
+      "step": 46155
+    },
+    {
+      "epoch": 3.739144523655217,
+      "grad_norm": 0.06198856979608536,
+      "learning_rate": 1.4492101354696433e-05,
+      "loss": 0.2057,
+      "step": 46156
+    },
+    {
+      "epoch": 3.7392255346727152,
+      "grad_norm": 0.07832913100719452,
+      "learning_rate": 1.4487600702101805e-05,
+      "loss": 0.1959,
+      "step": 46157
+    },
+    {
+      "epoch": 3.739306545690214,
+      "grad_norm": 0.07277149707078934,
+      "learning_rate": 1.4483100049507179e-05,
+      "loss": 0.2088,
+      "step": 46158
+    },
+    {
+      "epoch": 3.739387556707712,
+      "grad_norm": 0.08063270896673203,
+      "learning_rate": 1.4478599396912554e-05,
+      "loss": 0.2205,
+      "step": 46159
+    },
+    {
+      "epoch": 3.7394685677252104,
+      "grad_norm": 0.058787040412425995,
+      "learning_rate": 1.4474098744317926e-05,
+      "loss": 0.1775,
+      "step": 46160
+    },
+    {
+      "epoch": 3.739549578742709,
+      "grad_norm": 0.0893818661570549,
+      "learning_rate": 1.44695980917233e-05,
+      "loss": 0.2183,
+      "step": 46161
+    },
+    {
+      "epoch": 3.7396305897602073,
+      "grad_norm": 0.0777546837925911,
+      "learning_rate": 1.4465097439128675e-05,
+      "loss": 0.2185,
+      "step": 46162
+    },
+    {
+      "epoch": 3.7397116007777056,
+      "grad_norm": 0.0717715248465538,
+      "learning_rate": 1.4460596786534047e-05,
+      "loss": 0.2058,
+      "step": 46163
+    },
+    {
+      "epoch": 3.7397926117952043,
+      "grad_norm": 0.07632574439048767,
+      "learning_rate": 1.4456096133939424e-05,
+      "loss": 0.2519,
+      "step": 46164
+    },
+    {
+      "epoch": 3.7398736228127025,
+      "grad_norm": 0.07262930274009705,
+      "learning_rate": 1.4451595481344796e-05,
+      "loss": 0.2087,
+      "step": 46165
+    },
+    {
+      "epoch": 3.7399546338302008,
+      "grad_norm": 0.07834277302026749,
+      "learning_rate": 1.444709482875017e-05,
+      "loss": 0.2157,
+      "step": 46166
+    },
+    {
+      "epoch": 3.7400356448476995,
+      "grad_norm": 0.07401188462972641,
+      "learning_rate": 1.4442594176155545e-05,
+      "loss": 0.2323,
+      "step": 46167
+    },
+    {
+      "epoch": 3.7401166558651977,
+      "grad_norm": 0.06235014274716377,
+      "learning_rate": 1.4438093523560916e-05,
+      "loss": 0.2128,
+      "step": 46168
+    },
+    {
+      "epoch": 3.740197666882696,
+      "grad_norm": 0.09011727571487427,
+      "learning_rate": 1.4433592870966292e-05,
+      "loss": 0.251,
+      "step": 46169
+    },
+    {
+      "epoch": 3.7402786779001946,
+      "grad_norm": 0.07866238057613373,
+      "learning_rate": 1.4429092218371665e-05,
+      "loss": 0.2425,
+      "step": 46170
+    },
+    {
+      "epoch": 3.740359688917693,
+      "grad_norm": 0.06134997680783272,
+      "learning_rate": 1.4424591565777037e-05,
+      "loss": 0.191,
+      "step": 46171
+    },
+    {
+      "epoch": 3.740440699935191,
+      "grad_norm": 0.05997680500149727,
+      "learning_rate": 1.4420090913182413e-05,
+      "loss": 0.2054,
+      "step": 46172
+    },
+    {
+      "epoch": 3.74052171095269,
+      "grad_norm": 0.06824682652950287,
+      "learning_rate": 1.4415590260587786e-05,
+      "loss": 0.2271,
+      "step": 46173
+    },
+    {
+      "epoch": 3.740602721970188,
+      "grad_norm": 0.062017668038606644,
+      "learning_rate": 1.4411089607993158e-05,
+      "loss": 0.2037,
+      "step": 46174
+    },
+    {
+      "epoch": 3.7406837329876863,
+      "grad_norm": 0.06739827990531921,
+      "learning_rate": 1.4406588955398533e-05,
+      "loss": 0.2437,
+      "step": 46175
+    },
+    {
+      "epoch": 3.7407647440051845,
+      "grad_norm": 0.0841369777917862,
+      "learning_rate": 1.4402088302803907e-05,
+      "loss": 0.1901,
+      "step": 46176
+    },
+    {
+      "epoch": 3.7408457550226832,
+      "grad_norm": 0.06553324311971664,
+      "learning_rate": 1.4397587650209282e-05,
+      "loss": 0.211,
+      "step": 46177
+    },
+    {
+      "epoch": 3.7409267660401815,
+      "grad_norm": 0.07525978982448578,
+      "learning_rate": 1.4393086997614654e-05,
+      "loss": 0.2238,
+      "step": 46178
+    },
+    {
+      "epoch": 3.7410077770576797,
+      "grad_norm": 0.07539505511522293,
+      "learning_rate": 1.4388586345020028e-05,
+      "loss": 0.211,
+      "step": 46179
+    },
+    {
+      "epoch": 3.741088788075178,
+      "grad_norm": 0.063744455575943,
+      "learning_rate": 1.4384085692425403e-05,
+      "loss": 0.2007,
+      "step": 46180
+    },
+    {
+      "epoch": 3.7411697990926767,
+      "grad_norm": 0.07325825095176697,
+      "learning_rate": 1.4379585039830775e-05,
+      "loss": 0.2334,
+      "step": 46181
+    },
+    {
+      "epoch": 3.741250810110175,
+      "grad_norm": 0.07868387550115585,
+      "learning_rate": 1.4375084387236152e-05,
+      "loss": 0.2157,
+      "step": 46182
+    },
+    {
+      "epoch": 3.741331821127673,
+      "grad_norm": 0.07169875502586365,
+      "learning_rate": 1.4370583734641524e-05,
+      "loss": 0.2132,
+      "step": 46183
+    },
+    {
+      "epoch": 3.741412832145172,
+      "grad_norm": 0.07317327708005905,
+      "learning_rate": 1.4366083082046897e-05,
+      "loss": 0.253,
+      "step": 46184
+    },
+    {
+      "epoch": 3.74149384316267,
+      "grad_norm": 0.06858827918767929,
+      "learning_rate": 1.4361582429452273e-05,
+      "loss": 0.2038,
+      "step": 46185
+    },
+    {
+      "epoch": 3.7415748541801683,
+      "grad_norm": 0.05808888003230095,
+      "learning_rate": 1.4357081776857645e-05,
+      "loss": 0.1893,
+      "step": 46186
+    },
+    {
+      "epoch": 3.741655865197667,
+      "grad_norm": 0.06428666412830353,
+      "learning_rate": 1.4352581124263018e-05,
+      "loss": 0.2119,
+      "step": 46187
+    },
+    {
+      "epoch": 3.7417368762151653,
+      "grad_norm": 0.07074768841266632,
+      "learning_rate": 1.4348080471668394e-05,
+      "loss": 0.2288,
+      "step": 46188
+    },
+    {
+      "epoch": 3.7418178872326635,
+      "grad_norm": 0.08114055544137955,
+      "learning_rate": 1.4343579819073765e-05,
+      "loss": 0.2495,
+      "step": 46189
+    },
+    {
+      "epoch": 3.741898898250162,
+      "grad_norm": 0.07321310043334961,
+      "learning_rate": 1.433907916647914e-05,
+      "loss": 0.2476,
+      "step": 46190
+    },
+    {
+      "epoch": 3.7419799092676604,
+      "grad_norm": 0.08425623923540115,
+      "learning_rate": 1.4334578513884514e-05,
+      "loss": 0.2302,
+      "step": 46191
+    },
+    {
+      "epoch": 3.7420609202851587,
+      "grad_norm": 0.06360992789268494,
+      "learning_rate": 1.4330077861289886e-05,
+      "loss": 0.2127,
+      "step": 46192
+    },
+    {
+      "epoch": 3.7421419313026574,
+      "grad_norm": 0.07066339999437332,
+      "learning_rate": 1.4325577208695262e-05,
+      "loss": 0.2265,
+      "step": 46193
+    },
+    {
+      "epoch": 3.7422229423201556,
+      "grad_norm": 0.07735880464315414,
+      "learning_rate": 1.4321076556100635e-05,
+      "loss": 0.1989,
+      "step": 46194
+    },
+    {
+      "epoch": 3.742303953337654,
+      "grad_norm": 0.07494205236434937,
+      "learning_rate": 1.431657590350601e-05,
+      "loss": 0.2466,
+      "step": 46195
+    },
+    {
+      "epoch": 3.7423849643551526,
+      "grad_norm": 0.07548823207616806,
+      "learning_rate": 1.4312075250911382e-05,
+      "loss": 0.25,
+      "step": 46196
+    },
+    {
+      "epoch": 3.742465975372651,
+      "grad_norm": 0.0775391012430191,
+      "learning_rate": 1.4307574598316756e-05,
+      "loss": 0.2154,
+      "step": 46197
+    },
+    {
+      "epoch": 3.742546986390149,
+      "grad_norm": 0.06699874252080917,
+      "learning_rate": 1.4303073945722131e-05,
+      "loss": 0.2235,
+      "step": 46198
+    },
+    {
+      "epoch": 3.7426279974076473,
+      "grad_norm": 0.07778064161539078,
+      "learning_rate": 1.4298573293127503e-05,
+      "loss": 0.2105,
+      "step": 46199
+    },
+    {
+      "epoch": 3.742709008425146,
+      "grad_norm": 0.06393593549728394,
+      "learning_rate": 1.4294072640532877e-05,
+      "loss": 0.1742,
+      "step": 46200
+    },
+    {
+      "epoch": 3.7427900194426442,
+      "grad_norm": 0.0789911299943924,
+      "learning_rate": 1.4289571987938252e-05,
+      "loss": 0.2344,
+      "step": 46201
+    },
+    {
+      "epoch": 3.7428710304601425,
+      "grad_norm": 0.07193242013454437,
+      "learning_rate": 1.4285071335343626e-05,
+      "loss": 0.1964,
+      "step": 46202
+    },
+    {
+      "epoch": 3.7429520414776407,
+      "grad_norm": 0.04981725290417671,
+      "learning_rate": 1.4280570682749001e-05,
+      "loss": 0.1949,
+      "step": 46203
+    },
+    {
+      "epoch": 3.7430330524951394,
+      "grad_norm": 0.05438205972313881,
+      "learning_rate": 1.4276070030154373e-05,
+      "loss": 0.1834,
+      "step": 46204
+    },
+    {
+      "epoch": 3.7431140635126376,
+      "grad_norm": 0.07782084494829178,
+      "learning_rate": 1.4271569377559747e-05,
+      "loss": 0.2375,
+      "step": 46205
+    },
+    {
+      "epoch": 3.743195074530136,
+      "grad_norm": 0.08021578937768936,
+      "learning_rate": 1.4267068724965122e-05,
+      "loss": 0.2317,
+      "step": 46206
+    },
+    {
+      "epoch": 3.7432760855476346,
+      "grad_norm": 0.06510843336582184,
+      "learning_rate": 1.4262568072370494e-05,
+      "loss": 0.1998,
+      "step": 46207
+    },
+    {
+      "epoch": 3.743357096565133,
+      "grad_norm": 0.08877057582139969,
+      "learning_rate": 1.4258067419775867e-05,
+      "loss": 0.2367,
+      "step": 46208
+    },
+    {
+      "epoch": 3.743438107582631,
+      "grad_norm": 0.06314340233802795,
+      "learning_rate": 1.4253566767181243e-05,
+      "loss": 0.2043,
+      "step": 46209
+    },
+    {
+      "epoch": 3.7435191186001298,
+      "grad_norm": 0.06458821147680283,
+      "learning_rate": 1.4249066114586615e-05,
+      "loss": 0.1905,
+      "step": 46210
+    },
+    {
+      "epoch": 3.743600129617628,
+      "grad_norm": 0.054313499480485916,
+      "learning_rate": 1.424456546199199e-05,
+      "loss": 0.2063,
+      "step": 46211
+    },
+    {
+      "epoch": 3.7436811406351262,
+      "grad_norm": 0.059741366654634476,
+      "learning_rate": 1.4240064809397363e-05,
+      "loss": 0.1971,
+      "step": 46212
+    },
+    {
+      "epoch": 3.743762151652625,
+      "grad_norm": 0.07452462613582611,
+      "learning_rate": 1.4235564156802735e-05,
+      "loss": 0.2196,
+      "step": 46213
+    },
+    {
+      "epoch": 3.743843162670123,
+      "grad_norm": 0.07014549523591995,
+      "learning_rate": 1.423106350420811e-05,
+      "loss": 0.2512,
+      "step": 46214
+    },
+    {
+      "epoch": 3.7439241736876214,
+      "grad_norm": 0.08021154999732971,
+      "learning_rate": 1.4226562851613484e-05,
+      "loss": 0.2251,
+      "step": 46215
+    },
+    {
+      "epoch": 3.74400518470512,
+      "grad_norm": 0.06592398136854172,
+      "learning_rate": 1.422206219901886e-05,
+      "loss": 0.2422,
+      "step": 46216
+    },
+    {
+      "epoch": 3.7440861957226184,
+      "grad_norm": 0.0847722515463829,
+      "learning_rate": 1.4217561546424233e-05,
+      "loss": 0.2682,
+      "step": 46217
+    },
+    {
+      "epoch": 3.7441672067401166,
+      "grad_norm": 0.07960616797208786,
+      "learning_rate": 1.4213060893829605e-05,
+      "loss": 0.2091,
+      "step": 46218
+    },
+    {
+      "epoch": 3.7442482177576153,
+      "grad_norm": 0.06076245382428169,
+      "learning_rate": 1.420856024123498e-05,
+      "loss": 0.2446,
+      "step": 46219
+    },
+    {
+      "epoch": 3.7443292287751135,
+      "grad_norm": 0.07363170385360718,
+      "learning_rate": 1.4204059588640354e-05,
+      "loss": 0.218,
+      "step": 46220
+    },
+    {
+      "epoch": 3.744410239792612,
+      "grad_norm": 0.062311843037605286,
+      "learning_rate": 1.4199558936045726e-05,
+      "loss": 0.2067,
+      "step": 46221
+    },
+    {
+      "epoch": 3.74449125081011,
+      "grad_norm": 0.05965553969144821,
+      "learning_rate": 1.4195058283451101e-05,
+      "loss": 0.218,
+      "step": 46222
+    },
+    {
+      "epoch": 3.7445722618276087,
+      "grad_norm": 0.08739668875932693,
+      "learning_rate": 1.4190557630856475e-05,
+      "loss": 0.2213,
+      "step": 46223
+    },
+    {
+      "epoch": 3.744653272845107,
+      "grad_norm": 0.08150233328342438,
+      "learning_rate": 1.418605697826185e-05,
+      "loss": 0.2578,
+      "step": 46224
+    },
+    {
+      "epoch": 3.744734283862605,
+      "grad_norm": 0.09309550374746323,
+      "learning_rate": 1.4181556325667222e-05,
+      "loss": 0.2188,
+      "step": 46225
+    },
+    {
+      "epoch": 3.7448152948801035,
+      "grad_norm": 0.0847107321023941,
+      "learning_rate": 1.4177055673072596e-05,
+      "loss": 0.227,
+      "step": 46226
+    },
+    {
+      "epoch": 3.744896305897602,
+      "grad_norm": 0.06954275071620941,
+      "learning_rate": 1.4172555020477971e-05,
+      "loss": 0.2429,
+      "step": 46227
+    },
+    {
+      "epoch": 3.7449773169151004,
+      "grad_norm": 0.0696917250752449,
+      "learning_rate": 1.4168054367883343e-05,
+      "loss": 0.2489,
+      "step": 46228
+    },
+    {
+      "epoch": 3.7450583279325986,
+      "grad_norm": 0.08773817121982574,
+      "learning_rate": 1.4163553715288718e-05,
+      "loss": 0.2053,
+      "step": 46229
+    },
+    {
+      "epoch": 3.7451393389500973,
+      "grad_norm": 0.06404940038919449,
+      "learning_rate": 1.4159053062694092e-05,
+      "loss": 0.2722,
+      "step": 46230
+    },
+    {
+      "epoch": 3.7452203499675956,
+      "grad_norm": 0.06850332766771317,
+      "learning_rate": 1.4154552410099464e-05,
+      "loss": 0.2227,
+      "step": 46231
+    },
+    {
+      "epoch": 3.745301360985094,
+      "grad_norm": 0.06635698676109314,
+      "learning_rate": 1.4150051757504839e-05,
+      "loss": 0.1951,
+      "step": 46232
+    },
+    {
+      "epoch": 3.7453823720025925,
+      "grad_norm": 0.0643756315112114,
+      "learning_rate": 1.4145551104910213e-05,
+      "loss": 0.2088,
+      "step": 46233
+    },
+    {
+      "epoch": 3.7454633830200907,
+      "grad_norm": 0.08907685428857803,
+      "learning_rate": 1.4141050452315584e-05,
+      "loss": 0.2359,
+      "step": 46234
+    },
+    {
+      "epoch": 3.745544394037589,
+      "grad_norm": 0.07513679563999176,
+      "learning_rate": 1.4136549799720961e-05,
+      "loss": 0.1924,
+      "step": 46235
+    },
+    {
+      "epoch": 3.7456254050550877,
+      "grad_norm": 0.06546442955732346,
+      "learning_rate": 1.4132049147126333e-05,
+      "loss": 0.2136,
+      "step": 46236
+    },
+    {
+      "epoch": 3.745706416072586,
+      "grad_norm": 0.0697525292634964,
+      "learning_rate": 1.4127548494531709e-05,
+      "loss": 0.2123,
+      "step": 46237
+    },
+    {
+      "epoch": 3.745787427090084,
+      "grad_norm": 0.07150597870349884,
+      "learning_rate": 1.4123047841937082e-05,
+      "loss": 0.1814,
+      "step": 46238
+    },
+    {
+      "epoch": 3.745868438107583,
+      "grad_norm": 0.08265596628189087,
+      "learning_rate": 1.4118547189342454e-05,
+      "loss": 0.2069,
+      "step": 46239
+    },
+    {
+      "epoch": 3.745949449125081,
+      "grad_norm": 0.06267092376947403,
+      "learning_rate": 1.411404653674783e-05,
+      "loss": 0.2089,
+      "step": 46240
+    },
+    {
+      "epoch": 3.7460304601425793,
+      "grad_norm": 0.06582377105951309,
+      "learning_rate": 1.4109545884153203e-05,
+      "loss": 0.265,
+      "step": 46241
+    },
+    {
+      "epoch": 3.746111471160078,
+      "grad_norm": 0.061006009578704834,
+      "learning_rate": 1.4105045231558578e-05,
+      "loss": 0.2243,
+      "step": 46242
+    },
+    {
+      "epoch": 3.7461924821775763,
+      "grad_norm": 0.06815444678068161,
+      "learning_rate": 1.410054457896395e-05,
+      "loss": 0.2261,
+      "step": 46243
+    },
+    {
+      "epoch": 3.7462734931950745,
+      "grad_norm": 0.06948699802160263,
+      "learning_rate": 1.4096043926369324e-05,
+      "loss": 0.2476,
+      "step": 46244
+    },
+    {
+      "epoch": 3.7463545042125728,
+      "grad_norm": 0.06912975758314133,
+      "learning_rate": 1.4091543273774699e-05,
+      "loss": 0.2313,
+      "step": 46245
+    },
+    {
+      "epoch": 3.7464355152300715,
+      "grad_norm": 0.06383688002824783,
+      "learning_rate": 1.4087042621180071e-05,
+      "loss": 0.2016,
+      "step": 46246
+    },
+    {
+      "epoch": 3.7465165262475697,
+      "grad_norm": 0.08401155471801758,
+      "learning_rate": 1.4082541968585445e-05,
+      "loss": 0.2159,
+      "step": 46247
+    },
+    {
+      "epoch": 3.746597537265068,
+      "grad_norm": 0.06746657192707062,
+      "learning_rate": 1.407804131599082e-05,
+      "loss": 0.2241,
+      "step": 46248
+    },
+    {
+      "epoch": 3.746678548282566,
+      "grad_norm": 0.08890854567289352,
+      "learning_rate": 1.4073540663396192e-05,
+      "loss": 0.2303,
+      "step": 46249
+    },
+    {
+      "epoch": 3.746759559300065,
+      "grad_norm": 0.06370741128921509,
+      "learning_rate": 1.4069040010801567e-05,
+      "loss": 0.2286,
+      "step": 46250
+    },
+    {
+      "epoch": 3.746840570317563,
+      "grad_norm": 0.05881880968809128,
+      "learning_rate": 1.406453935820694e-05,
+      "loss": 0.195,
+      "step": 46251
+    },
+    {
+      "epoch": 3.7469215813350614,
+      "grad_norm": 0.06718897074460983,
+      "learning_rate": 1.4060038705612313e-05,
+      "loss": 0.2372,
+      "step": 46252
+    },
+    {
+      "epoch": 3.74700259235256,
+      "grad_norm": 0.06468194723129272,
+      "learning_rate": 1.405553805301769e-05,
+      "loss": 0.1888,
+      "step": 46253
+    },
+    {
+      "epoch": 3.7470836033700583,
+      "grad_norm": 0.06398289650678635,
+      "learning_rate": 1.4051037400423062e-05,
+      "loss": 0.1908,
+      "step": 46254
+    },
+    {
+      "epoch": 3.7471646143875565,
+      "grad_norm": 0.06329668313264847,
+      "learning_rate": 1.4046536747828437e-05,
+      "loss": 0.2213,
+      "step": 46255
+    },
+    {
+      "epoch": 3.7472456254050552,
+      "grad_norm": 0.07826821506023407,
+      "learning_rate": 1.404203609523381e-05,
+      "loss": 0.1977,
+      "step": 46256
+    },
+    {
+      "epoch": 3.7473266364225535,
+      "grad_norm": 0.0786462053656578,
+      "learning_rate": 1.4037535442639182e-05,
+      "loss": 0.2486,
+      "step": 46257
+    },
+    {
+      "epoch": 3.7474076474400517,
+      "grad_norm": 0.07066361606121063,
+      "learning_rate": 1.4033034790044558e-05,
+      "loss": 0.2057,
+      "step": 46258
+    },
+    {
+      "epoch": 3.7474886584575504,
+      "grad_norm": 0.07975254207849503,
+      "learning_rate": 1.4028534137449931e-05,
+      "loss": 0.2555,
+      "step": 46259
+    },
+    {
+      "epoch": 3.7475696694750487,
+      "grad_norm": 0.08178477734327316,
+      "learning_rate": 1.4024033484855303e-05,
+      "loss": 0.2112,
+      "step": 46260
+    },
+    {
+      "epoch": 3.747650680492547,
+      "grad_norm": 0.07174079120159149,
+      "learning_rate": 1.4019532832260678e-05,
+      "loss": 0.1987,
+      "step": 46261
+    },
+    {
+      "epoch": 3.7477316915100456,
+      "grad_norm": 0.06191396713256836,
+      "learning_rate": 1.4015032179666052e-05,
+      "loss": 0.2372,
+      "step": 46262
+    },
+    {
+      "epoch": 3.747812702527544,
+      "grad_norm": 0.0650116503238678,
+      "learning_rate": 1.4010531527071427e-05,
+      "loss": 0.2416,
+      "step": 46263
+    },
+    {
+      "epoch": 3.747893713545042,
+      "grad_norm": 0.07233452051877975,
+      "learning_rate": 1.40060308744768e-05,
+      "loss": 0.2519,
+      "step": 46264
+    },
+    {
+      "epoch": 3.7479747245625408,
+      "grad_norm": 0.06753069162368774,
+      "learning_rate": 1.4001530221882173e-05,
+      "loss": 0.1924,
+      "step": 46265
+    },
+    {
+      "epoch": 3.748055735580039,
+      "grad_norm": 0.08413957059383392,
+      "learning_rate": 1.3997029569287548e-05,
+      "loss": 0.2399,
+      "step": 46266
+    },
+    {
+      "epoch": 3.7481367465975373,
+      "grad_norm": 0.07612863928079605,
+      "learning_rate": 1.399252891669292e-05,
+      "loss": 0.1912,
+      "step": 46267
+    },
+    {
+      "epoch": 3.7482177576150355,
+      "grad_norm": 0.06556925177574158,
+      "learning_rate": 1.3988028264098297e-05,
+      "loss": 0.2088,
+      "step": 46268
+    },
+    {
+      "epoch": 3.7482987686325338,
+      "grad_norm": 0.07448197156190872,
+      "learning_rate": 1.3983527611503669e-05,
+      "loss": 0.2145,
+      "step": 46269
+    },
+    {
+      "epoch": 3.7483797796500324,
+      "grad_norm": 0.07443884760141373,
+      "learning_rate": 1.3979026958909041e-05,
+      "loss": 0.1992,
+      "step": 46270
+    },
+    {
+      "epoch": 3.7484607906675307,
+      "grad_norm": 0.07081007212400436,
+      "learning_rate": 1.3974526306314418e-05,
+      "loss": 0.2568,
+      "step": 46271
+    },
+    {
+      "epoch": 3.748541801685029,
+      "grad_norm": 0.06729474663734436,
+      "learning_rate": 1.397002565371979e-05,
+      "loss": 0.1957,
+      "step": 46272
+    },
+    {
+      "epoch": 3.7486228127025276,
+      "grad_norm": 0.06812798231840134,
+      "learning_rate": 1.3965525001125163e-05,
+      "loss": 0.2147,
+      "step": 46273
+    },
+    {
+      "epoch": 3.748703823720026,
+      "grad_norm": 0.07089778780937195,
+      "learning_rate": 1.3961024348530539e-05,
+      "loss": 0.2223,
+      "step": 46274
+    },
+    {
+      "epoch": 3.748784834737524,
+      "grad_norm": 0.06887844949960709,
+      "learning_rate": 1.395652369593591e-05,
+      "loss": 0.2128,
+      "step": 46275
+    },
+    {
+      "epoch": 3.748865845755023,
+      "grad_norm": 0.0692133754491806,
+      "learning_rate": 1.3952023043341286e-05,
+      "loss": 0.2063,
+      "step": 46276
+    },
+    {
+      "epoch": 3.748946856772521,
+      "grad_norm": 0.059787359088659286,
+      "learning_rate": 1.394752239074666e-05,
+      "loss": 0.2294,
+      "step": 46277
+    },
+    {
+      "epoch": 3.7490278677900193,
+      "grad_norm": 0.055009905248880386,
+      "learning_rate": 1.3943021738152031e-05,
+      "loss": 0.194,
+      "step": 46278
+    },
+    {
+      "epoch": 3.749108878807518,
+      "grad_norm": 0.07308180630207062,
+      "learning_rate": 1.3938521085557407e-05,
+      "loss": 0.2311,
+      "step": 46279
+    },
+    {
+      "epoch": 3.749189889825016,
+      "grad_norm": 0.0746115893125534,
+      "learning_rate": 1.393402043296278e-05,
+      "loss": 0.247,
+      "step": 46280
+    },
+    {
+      "epoch": 3.7492709008425145,
+      "grad_norm": 0.058564692735672,
+      "learning_rate": 1.3929519780368156e-05,
+      "loss": 0.1998,
+      "step": 46281
+    },
+    {
+      "epoch": 3.749351911860013,
+      "grad_norm": 0.06961791962385178,
+      "learning_rate": 1.3925019127773528e-05,
+      "loss": 0.2499,
+      "step": 46282
+    },
+    {
+      "epoch": 3.7494329228775114,
+      "grad_norm": 0.06499256938695908,
+      "learning_rate": 1.3920518475178901e-05,
+      "loss": 0.213,
+      "step": 46283
+    },
+    {
+      "epoch": 3.7495139338950096,
+      "grad_norm": 0.0894896611571312,
+      "learning_rate": 1.3916017822584276e-05,
+      "loss": 0.2478,
+      "step": 46284
+    },
+    {
+      "epoch": 3.7495949449125083,
+      "grad_norm": 0.08614595234394073,
+      "learning_rate": 1.3911517169989648e-05,
+      "loss": 0.209,
+      "step": 46285
+    },
+    {
+      "epoch": 3.7496759559300066,
+      "grad_norm": 0.06723782420158386,
+      "learning_rate": 1.3907016517395022e-05,
+      "loss": 0.2103,
+      "step": 46286
+    },
+    {
+      "epoch": 3.749756966947505,
+      "grad_norm": 0.07532931864261627,
+      "learning_rate": 1.3902515864800397e-05,
+      "loss": 0.2022,
+      "step": 46287
+    },
+    {
+      "epoch": 3.7498379779650035,
+      "grad_norm": 0.08021144568920135,
+      "learning_rate": 1.389801521220577e-05,
+      "loss": 0.2217,
+      "step": 46288
+    },
+    {
+      "epoch": 3.7499189889825018,
+      "grad_norm": 0.07499111443758011,
+      "learning_rate": 1.3893514559611146e-05,
+      "loss": 0.2636,
+      "step": 46289
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.10113779455423355,
+      "learning_rate": 1.3889013907016518e-05,
+      "loss": 0.2517,
+      "step": 46290
+    },
+    {
+      "epoch": 3.7500810110174982,
+      "grad_norm": 0.05415019765496254,
+      "learning_rate": 1.3884513254421892e-05,
+      "loss": 0.2003,
+      "step": 46291
+    },
+    {
+      "epoch": 3.7501620220349965,
+      "grad_norm": 0.06942694634199142,
+      "learning_rate": 1.3880012601827267e-05,
+      "loss": 0.2233,
+      "step": 46292
+    },
+    {
+      "epoch": 3.750243033052495,
+      "grad_norm": 0.07195769250392914,
+      "learning_rate": 1.3875511949232639e-05,
+      "loss": 0.1826,
+      "step": 46293
+    },
+    {
+      "epoch": 3.7503240440699934,
+      "grad_norm": 0.05428317189216614,
+      "learning_rate": 1.3871011296638012e-05,
+      "loss": 0.21,
+      "step": 46294
+    },
+    {
+      "epoch": 3.7504050550874917,
+      "grad_norm": 0.050274260342121124,
+      "learning_rate": 1.3866510644043388e-05,
+      "loss": 0.2175,
+      "step": 46295
+    },
+    {
+      "epoch": 3.7504860661049904,
+      "grad_norm": 0.06542641669511795,
+      "learning_rate": 1.386200999144876e-05,
+      "loss": 0.2478,
+      "step": 46296
+    },
+    {
+      "epoch": 3.7505670771224886,
+      "grad_norm": 0.07026327401399612,
+      "learning_rate": 1.3857509338854135e-05,
+      "loss": 0.2171,
+      "step": 46297
+    },
+    {
+      "epoch": 3.750648088139987,
+      "grad_norm": 0.07940451055765152,
+      "learning_rate": 1.3853008686259509e-05,
+      "loss": 0.2378,
+      "step": 46298
+    },
+    {
+      "epoch": 3.7507290991574855,
+      "grad_norm": 0.08603038638830185,
+      "learning_rate": 1.384850803366488e-05,
+      "loss": 0.2269,
+      "step": 46299
+    },
+    {
+      "epoch": 3.750810110174984,
+      "grad_norm": 0.07071874290704727,
+      "learning_rate": 1.3844007381070256e-05,
+      "loss": 0.2226,
+      "step": 46300
+    },
+    {
+      "epoch": 3.750891121192482,
+      "grad_norm": 0.0803099051117897,
+      "learning_rate": 1.383950672847563e-05,
+      "loss": 0.2084,
+      "step": 46301
+    },
+    {
+      "epoch": 3.7509721322099807,
+      "grad_norm": 0.07697241753339767,
+      "learning_rate": 1.3835006075881005e-05,
+      "loss": 0.214,
+      "step": 46302
+    },
+    {
+      "epoch": 3.751053143227479,
+      "grad_norm": 0.06945455074310303,
+      "learning_rate": 1.3830505423286377e-05,
+      "loss": 0.2187,
+      "step": 46303
+    },
+    {
+      "epoch": 3.751134154244977,
+      "grad_norm": 0.0705169215798378,
+      "learning_rate": 1.382600477069175e-05,
+      "loss": 0.2639,
+      "step": 46304
+    },
+    {
+      "epoch": 3.751215165262476,
+      "grad_norm": 0.0709058865904808,
+      "learning_rate": 1.3821504118097126e-05,
+      "loss": 0.2424,
+      "step": 46305
+    },
+    {
+      "epoch": 3.751296176279974,
+      "grad_norm": 0.0878700241446495,
+      "learning_rate": 1.3817003465502497e-05,
+      "loss": 0.2605,
+      "step": 46306
+    },
+    {
+      "epoch": 3.7513771872974724,
+      "grad_norm": 0.0711369663476944,
+      "learning_rate": 1.3812502812907871e-05,
+      "loss": 0.2516,
+      "step": 46307
+    },
+    {
+      "epoch": 3.751458198314971,
+      "grad_norm": 0.07304982095956802,
+      "learning_rate": 1.3808002160313246e-05,
+      "loss": 0.2469,
+      "step": 46308
+    },
+    {
+      "epoch": 3.7515392093324693,
+      "grad_norm": 0.0640924721956253,
+      "learning_rate": 1.380350150771862e-05,
+      "loss": 0.1904,
+      "step": 46309
+    },
+    {
+      "epoch": 3.7516202203499676,
+      "grad_norm": 0.0668100118637085,
+      "learning_rate": 1.3799000855123995e-05,
+      "loss": 0.2016,
+      "step": 46310
+    },
+    {
+      "epoch": 3.7517012313674662,
+      "grad_norm": 0.0843009278178215,
+      "learning_rate": 1.3794500202529367e-05,
+      "loss": 0.229,
+      "step": 46311
+    },
+    {
+      "epoch": 3.7517822423849645,
+      "grad_norm": 0.07822394371032715,
+      "learning_rate": 1.378999954993474e-05,
+      "loss": 0.2195,
+      "step": 46312
+    },
+    {
+      "epoch": 3.7518632534024627,
+      "grad_norm": 0.07780709862709045,
+      "learning_rate": 1.3785498897340116e-05,
+      "loss": 0.2342,
+      "step": 46313
+    },
+    {
+      "epoch": 3.751944264419961,
+      "grad_norm": 0.06893333792686462,
+      "learning_rate": 1.3780998244745488e-05,
+      "loss": 0.232,
+      "step": 46314
+    },
+    {
+      "epoch": 3.7520252754374592,
+      "grad_norm": 0.06503161042928696,
+      "learning_rate": 1.3776497592150863e-05,
+      "loss": 0.24,
+      "step": 46315
+    },
+    {
+      "epoch": 3.752106286454958,
+      "grad_norm": 0.06948009878396988,
+      "learning_rate": 1.3771996939556237e-05,
+      "loss": 0.2304,
+      "step": 46316
+    },
+    {
+      "epoch": 3.752187297472456,
+      "grad_norm": 0.07049506157636642,
+      "learning_rate": 1.3767496286961609e-05,
+      "loss": 0.1865,
+      "step": 46317
+    },
+    {
+      "epoch": 3.7522683084899544,
+      "grad_norm": 0.07123062759637833,
+      "learning_rate": 1.3762995634366984e-05,
+      "loss": 0.2311,
+      "step": 46318
+    },
+    {
+      "epoch": 3.752349319507453,
+      "grad_norm": 0.07314299792051315,
+      "learning_rate": 1.3758494981772358e-05,
+      "loss": 0.236,
+      "step": 46319
+    },
+    {
+      "epoch": 3.7524303305249513,
+      "grad_norm": 0.07087218761444092,
+      "learning_rate": 1.375399432917773e-05,
+      "loss": 0.2167,
+      "step": 46320
+    },
+    {
+      "epoch": 3.7525113415424496,
+      "grad_norm": 0.09059132635593414,
+      "learning_rate": 1.3749493676583105e-05,
+      "loss": 0.2234,
+      "step": 46321
+    },
+    {
+      "epoch": 3.7525923525599483,
+      "grad_norm": 0.06358359009027481,
+      "learning_rate": 1.3744993023988478e-05,
+      "loss": 0.2308,
+      "step": 46322
+    },
+    {
+      "epoch": 3.7526733635774465,
+      "grad_norm": 0.06344583630561829,
+      "learning_rate": 1.3740492371393854e-05,
+      "loss": 0.2081,
+      "step": 46323
+    },
+    {
+      "epoch": 3.7527543745949448,
+      "grad_norm": 0.0792200043797493,
+      "learning_rate": 1.3735991718799227e-05,
+      "loss": 0.2335,
+      "step": 46324
+    },
+    {
+      "epoch": 3.7528353856124435,
+      "grad_norm": 0.06948511302471161,
+      "learning_rate": 1.37314910662046e-05,
+      "loss": 0.2181,
+      "step": 46325
+    },
+    {
+      "epoch": 3.7529163966299417,
+      "grad_norm": 0.06803888082504272,
+      "learning_rate": 1.3726990413609975e-05,
+      "loss": 0.2079,
+      "step": 46326
+    },
+    {
+      "epoch": 3.75299740764744,
+      "grad_norm": 0.06263738125562668,
+      "learning_rate": 1.3722489761015348e-05,
+      "loss": 0.2381,
+      "step": 46327
+    },
+    {
+      "epoch": 3.7530784186649386,
+      "grad_norm": 0.06731061637401581,
+      "learning_rate": 1.3717989108420723e-05,
+      "loss": 0.2008,
+      "step": 46328
+    },
+    {
+      "epoch": 3.753159429682437,
+      "grad_norm": 0.07265542447566986,
+      "learning_rate": 1.3713488455826095e-05,
+      "loss": 0.2197,
+      "step": 46329
+    },
+    {
+      "epoch": 3.753240440699935,
+      "grad_norm": 0.08350956439971924,
+      "learning_rate": 1.3708987803231469e-05,
+      "loss": 0.2395,
+      "step": 46330
+    },
+    {
+      "epoch": 3.753321451717434,
+      "grad_norm": 0.07093897461891174,
+      "learning_rate": 1.3704487150636844e-05,
+      "loss": 0.2328,
+      "step": 46331
+    },
+    {
+      "epoch": 3.753402462734932,
+      "grad_norm": 0.10086548328399658,
+      "learning_rate": 1.3699986498042216e-05,
+      "loss": 0.2574,
+      "step": 46332
+    },
+    {
+      "epoch": 3.7534834737524303,
+      "grad_norm": 0.06807972490787506,
+      "learning_rate": 1.369548584544759e-05,
+      "loss": 0.2357,
+      "step": 46333
+    },
+    {
+      "epoch": 3.7535644847699285,
+      "grad_norm": 0.09186132252216339,
+      "learning_rate": 1.3690985192852965e-05,
+      "loss": 0.2512,
+      "step": 46334
+    },
+    {
+      "epoch": 3.7536454957874272,
+      "grad_norm": 0.06752340495586395,
+      "learning_rate": 1.3686484540258337e-05,
+      "loss": 0.212,
+      "step": 46335
+    },
+    {
+      "epoch": 3.7537265068049255,
+      "grad_norm": 0.07947950065135956,
+      "learning_rate": 1.3681983887663712e-05,
+      "loss": 0.2415,
+      "step": 46336
+    },
+    {
+      "epoch": 3.7538075178224237,
+      "grad_norm": 0.07616253942251205,
+      "learning_rate": 1.3677483235069086e-05,
+      "loss": 0.2219,
+      "step": 46337
+    },
+    {
+      "epoch": 3.753888528839922,
+      "grad_norm": 0.074013851583004,
+      "learning_rate": 1.3672982582474458e-05,
+      "loss": 0.2383,
+      "step": 46338
+    },
+    {
+      "epoch": 3.7539695398574207,
+      "grad_norm": 0.07347376644611359,
+      "learning_rate": 1.3668481929879833e-05,
+      "loss": 0.1882,
+      "step": 46339
+    },
+    {
+      "epoch": 3.754050550874919,
+      "grad_norm": 0.06440410017967224,
+      "learning_rate": 1.3663981277285207e-05,
+      "loss": 0.2126,
+      "step": 46340
+    },
+    {
+      "epoch": 3.754131561892417,
+      "grad_norm": 0.061819300055503845,
+      "learning_rate": 1.3659480624690582e-05,
+      "loss": 0.2163,
+      "step": 46341
+    },
+    {
+      "epoch": 3.754212572909916,
+      "grad_norm": 0.09030622243881226,
+      "learning_rate": 1.3654979972095956e-05,
+      "loss": 0.2193,
+      "step": 46342
+    },
+    {
+      "epoch": 3.754293583927414,
+      "grad_norm": 0.06111063435673714,
+      "learning_rate": 1.3650479319501328e-05,
+      "loss": 0.2516,
+      "step": 46343
+    },
+    {
+      "epoch": 3.7543745949449123,
+      "grad_norm": 0.07280724495649338,
+      "learning_rate": 1.3645978666906703e-05,
+      "loss": 0.2058,
+      "step": 46344
+    },
+    {
+      "epoch": 3.754455605962411,
+      "grad_norm": 0.0639389231801033,
+      "learning_rate": 1.3641478014312076e-05,
+      "loss": 0.2089,
+      "step": 46345
+    },
+    {
+      "epoch": 3.7545366169799093,
+      "grad_norm": 0.08013645559549332,
+      "learning_rate": 1.3636977361717448e-05,
+      "loss": 0.2546,
+      "step": 46346
+    },
+    {
+      "epoch": 3.7546176279974075,
+      "grad_norm": 0.08095403015613556,
+      "learning_rate": 1.3632476709122824e-05,
+      "loss": 0.218,
+      "step": 46347
+    },
+    {
+      "epoch": 3.754698639014906,
+      "grad_norm": 0.06311283260583878,
+      "learning_rate": 1.3627976056528197e-05,
+      "loss": 0.2036,
+      "step": 46348
+    },
+    {
+      "epoch": 3.7547796500324044,
+      "grad_norm": 0.07423945516347885,
+      "learning_rate": 1.3623475403933573e-05,
+      "loss": 0.2177,
+      "step": 46349
+    },
+    {
+      "epoch": 3.7548606610499027,
+      "grad_norm": 0.07765256613492966,
+      "learning_rate": 1.3618974751338944e-05,
+      "loss": 0.2528,
+      "step": 46350
+    },
+    {
+      "epoch": 3.7549416720674014,
+      "grad_norm": 0.05665789544582367,
+      "learning_rate": 1.3614474098744318e-05,
+      "loss": 0.2193,
+      "step": 46351
+    },
+    {
+      "epoch": 3.7550226830848996,
+      "grad_norm": 0.07820606231689453,
+      "learning_rate": 1.3609973446149693e-05,
+      "loss": 0.2629,
+      "step": 46352
+    },
+    {
+      "epoch": 3.755103694102398,
+      "grad_norm": 0.06132517755031586,
+      "learning_rate": 1.3605472793555065e-05,
+      "loss": 0.2099,
+      "step": 46353
+    },
+    {
+      "epoch": 3.7551847051198965,
+      "grad_norm": 0.07045713812112808,
+      "learning_rate": 1.360097214096044e-05,
+      "loss": 0.2298,
+      "step": 46354
+    },
+    {
+      "epoch": 3.755265716137395,
+      "grad_norm": 0.07467834651470184,
+      "learning_rate": 1.3596471488365814e-05,
+      "loss": 0.2029,
+      "step": 46355
+    },
+    {
+      "epoch": 3.755346727154893,
+      "grad_norm": 0.07477845996618271,
+      "learning_rate": 1.3591970835771186e-05,
+      "loss": 0.2221,
+      "step": 46356
+    },
+    {
+      "epoch": 3.7554277381723913,
+      "grad_norm": 0.06774298846721649,
+      "learning_rate": 1.3587470183176561e-05,
+      "loss": 0.2328,
+      "step": 46357
+    },
+    {
+      "epoch": 3.75550874918989,
+      "grad_norm": 0.06271317601203918,
+      "learning_rate": 1.3582969530581935e-05,
+      "loss": 0.2151,
+      "step": 46358
+    },
+    {
+      "epoch": 3.755589760207388,
+      "grad_norm": 0.07417146861553192,
+      "learning_rate": 1.3578468877987307e-05,
+      "loss": 0.2502,
+      "step": 46359
+    },
+    {
+      "epoch": 3.7556707712248865,
+      "grad_norm": 0.07773631066083908,
+      "learning_rate": 1.3573968225392684e-05,
+      "loss": 0.2285,
+      "step": 46360
+    },
+    {
+      "epoch": 3.7557517822423847,
+      "grad_norm": 0.07055094838142395,
+      "learning_rate": 1.3569467572798056e-05,
+      "loss": 0.2339,
+      "step": 46361
+    },
+    {
+      "epoch": 3.7558327932598834,
+      "grad_norm": 0.07049763202667236,
+      "learning_rate": 1.3564966920203431e-05,
+      "loss": 0.2265,
+      "step": 46362
+    },
+    {
+      "epoch": 3.7559138042773816,
+      "grad_norm": 0.07336530089378357,
+      "learning_rate": 1.3560466267608805e-05,
+      "loss": 0.1979,
+      "step": 46363
+    },
+    {
+      "epoch": 3.75599481529488,
+      "grad_norm": 0.0836474746465683,
+      "learning_rate": 1.3555965615014177e-05,
+      "loss": 0.223,
+      "step": 46364
+    },
+    {
+      "epoch": 3.7560758263123786,
+      "grad_norm": 0.076979860663414,
+      "learning_rate": 1.3551464962419552e-05,
+      "loss": 0.2199,
+      "step": 46365
+    },
+    {
+      "epoch": 3.756156837329877,
+      "grad_norm": 0.05993069335818291,
+      "learning_rate": 1.3546964309824925e-05,
+      "loss": 0.2014,
+      "step": 46366
+    },
+    {
+      "epoch": 3.756237848347375,
+      "grad_norm": 0.07212058454751968,
+      "learning_rate": 1.35424636572303e-05,
+      "loss": 0.2198,
+      "step": 46367
+    },
+    {
+      "epoch": 3.7563188593648738,
+      "grad_norm": 0.06068550795316696,
+      "learning_rate": 1.3537963004635673e-05,
+      "loss": 0.2511,
+      "step": 46368
+    },
+    {
+      "epoch": 3.756399870382372,
+      "grad_norm": 0.0721498504281044,
+      "learning_rate": 1.3533462352041046e-05,
+      "loss": 0.2439,
+      "step": 46369
+    },
+    {
+      "epoch": 3.7564808813998702,
+      "grad_norm": 0.07898570597171783,
+      "learning_rate": 1.3528961699446422e-05,
+      "loss": 0.2429,
+      "step": 46370
+    },
+    {
+      "epoch": 3.756561892417369,
+      "grad_norm": 0.06375134736299515,
+      "learning_rate": 1.3524461046851793e-05,
+      "loss": 0.2478,
+      "step": 46371
+    },
+    {
+      "epoch": 3.756642903434867,
+      "grad_norm": 0.08874838799238205,
+      "learning_rate": 1.3519960394257167e-05,
+      "loss": 0.2353,
+      "step": 46372
+    },
+    {
+      "epoch": 3.7567239144523654,
+      "grad_norm": 0.07910864055156708,
+      "learning_rate": 1.3515459741662542e-05,
+      "loss": 0.2272,
+      "step": 46373
+    },
+    {
+      "epoch": 3.756804925469864,
+      "grad_norm": 0.07340925931930542,
+      "learning_rate": 1.3510959089067914e-05,
+      "loss": 0.2493,
+      "step": 46374
+    },
+    {
+      "epoch": 3.7568859364873624,
+      "grad_norm": 0.0714777335524559,
+      "learning_rate": 1.3506458436473291e-05,
+      "loss": 0.2262,
+      "step": 46375
+    },
+    {
+      "epoch": 3.7569669475048606,
+      "grad_norm": 0.08851076662540436,
+      "learning_rate": 1.3501957783878663e-05,
+      "loss": 0.2639,
+      "step": 46376
+    },
+    {
+      "epoch": 3.7570479585223593,
+      "grad_norm": 0.06556161493062973,
+      "learning_rate": 1.3497457131284035e-05,
+      "loss": 0.2123,
+      "step": 46377
+    },
+    {
+      "epoch": 3.7571289695398575,
+      "grad_norm": 0.0682496726512909,
+      "learning_rate": 1.3492956478689412e-05,
+      "loss": 0.2372,
+      "step": 46378
+    },
+    {
+      "epoch": 3.7572099805573558,
+      "grad_norm": 0.07470902800559998,
+      "learning_rate": 1.3488455826094784e-05,
+      "loss": 0.2004,
+      "step": 46379
+    },
+    {
+      "epoch": 3.757290991574854,
+      "grad_norm": 0.0711580142378807,
+      "learning_rate": 1.3483955173500158e-05,
+      "loss": 0.2295,
+      "step": 46380
+    },
+    {
+      "epoch": 3.7573720025923527,
+      "grad_norm": 0.06948163360357285,
+      "learning_rate": 1.3479454520905533e-05,
+      "loss": 0.2627,
+      "step": 46381
+    },
+    {
+      "epoch": 3.757453013609851,
+      "grad_norm": 0.08253484219312668,
+      "learning_rate": 1.3474953868310905e-05,
+      "loss": 0.2362,
+      "step": 46382
+    },
+    {
+      "epoch": 3.757534024627349,
+      "grad_norm": 0.0675487071275711,
+      "learning_rate": 1.347045321571628e-05,
+      "loss": 0.2605,
+      "step": 46383
+    },
+    {
+      "epoch": 3.7576150356448474,
+      "grad_norm": 0.0675104409456253,
+      "learning_rate": 1.3465952563121654e-05,
+      "loss": 0.1905,
+      "step": 46384
+    },
+    {
+      "epoch": 3.757696046662346,
+      "grad_norm": 0.07248274981975555,
+      "learning_rate": 1.3461451910527026e-05,
+      "loss": 0.2062,
+      "step": 46385
+    },
+    {
+      "epoch": 3.7577770576798444,
+      "grad_norm": 0.06134706735610962,
+      "learning_rate": 1.3456951257932401e-05,
+      "loss": 0.1993,
+      "step": 46386
+    },
+    {
+      "epoch": 3.7578580686973426,
+      "grad_norm": 0.07231275737285614,
+      "learning_rate": 1.3452450605337775e-05,
+      "loss": 0.2197,
+      "step": 46387
+    },
+    {
+      "epoch": 3.7579390797148413,
+      "grad_norm": 0.07492779940366745,
+      "learning_rate": 1.344794995274315e-05,
+      "loss": 0.1983,
+      "step": 46388
+    },
+    {
+      "epoch": 3.7580200907323396,
+      "grad_norm": 0.05680622532963753,
+      "learning_rate": 1.3443449300148522e-05,
+      "loss": 0.2109,
+      "step": 46389
+    },
+    {
+      "epoch": 3.758101101749838,
+      "grad_norm": 0.06529416888952255,
+      "learning_rate": 1.3438948647553895e-05,
+      "loss": 0.2125,
+      "step": 46390
+    },
+    {
+      "epoch": 3.7581821127673365,
+      "grad_norm": 0.07710691541433334,
+      "learning_rate": 1.343444799495927e-05,
+      "loss": 0.2367,
+      "step": 46391
+    },
+    {
+      "epoch": 3.7582631237848347,
+      "grad_norm": 0.0891771912574768,
+      "learning_rate": 1.3429947342364643e-05,
+      "loss": 0.2411,
+      "step": 46392
+    },
+    {
+      "epoch": 3.758344134802333,
+      "grad_norm": 0.07291162014007568,
+      "learning_rate": 1.3425446689770016e-05,
+      "loss": 0.1957,
+      "step": 46393
+    },
+    {
+      "epoch": 3.7584251458198317,
+      "grad_norm": 0.07850871235132217,
+      "learning_rate": 1.3420946037175391e-05,
+      "loss": 0.2449,
+      "step": 46394
+    },
+    {
+      "epoch": 3.75850615683733,
+      "grad_norm": 0.07013242691755295,
+      "learning_rate": 1.3416445384580763e-05,
+      "loss": 0.256,
+      "step": 46395
+    },
+    {
+      "epoch": 3.758587167854828,
+      "grad_norm": 0.09972033649682999,
+      "learning_rate": 1.341194473198614e-05,
+      "loss": 0.1975,
+      "step": 46396
+    },
+    {
+      "epoch": 3.758668178872327,
+      "grad_norm": 0.06596890836954117,
+      "learning_rate": 1.3407444079391512e-05,
+      "loss": 0.1988,
+      "step": 46397
+    },
+    {
+      "epoch": 3.758749189889825,
+      "grad_norm": 0.07399401068687439,
+      "learning_rate": 1.3402943426796886e-05,
+      "loss": 0.1893,
+      "step": 46398
+    },
+    {
+      "epoch": 3.7588302009073233,
+      "grad_norm": 0.06135550141334534,
+      "learning_rate": 1.3398442774202261e-05,
+      "loss": 0.1996,
+      "step": 46399
+    },
+    {
+      "epoch": 3.758911211924822,
+      "grad_norm": 0.06000867113471031,
+      "learning_rate": 1.3393942121607633e-05,
+      "loss": 0.1857,
+      "step": 46400
+    },
+    {
+      "epoch": 3.7589922229423203,
+      "grad_norm": 0.06590692698955536,
+      "learning_rate": 1.3389441469013008e-05,
+      "loss": 0.2164,
+      "step": 46401
+    },
+    {
+      "epoch": 3.7590732339598185,
+      "grad_norm": 0.06405100971460342,
+      "learning_rate": 1.3384940816418382e-05,
+      "loss": 0.1724,
+      "step": 46402
+    },
+    {
+      "epoch": 3.7591542449773168,
+      "grad_norm": 0.06549198925495148,
+      "learning_rate": 1.3380440163823754e-05,
+      "loss": 0.2295,
+      "step": 46403
+    },
+    {
+      "epoch": 3.7592352559948155,
+      "grad_norm": 0.06221283972263336,
+      "learning_rate": 1.337593951122913e-05,
+      "loss": 0.2328,
+      "step": 46404
+    },
+    {
+      "epoch": 3.7593162670123137,
+      "grad_norm": 0.07838263362646103,
+      "learning_rate": 1.3371438858634503e-05,
+      "loss": 0.2247,
+      "step": 46405
+    },
+    {
+      "epoch": 3.759397278029812,
+      "grad_norm": 0.07541157305240631,
+      "learning_rate": 1.3366938206039875e-05,
+      "loss": 0.2444,
+      "step": 46406
+    },
+    {
+      "epoch": 3.75947828904731,
+      "grad_norm": 0.06589043140411377,
+      "learning_rate": 1.336243755344525e-05,
+      "loss": 0.1966,
+      "step": 46407
+    },
+    {
+      "epoch": 3.759559300064809,
+      "grad_norm": 0.06893175840377808,
+      "learning_rate": 1.3357936900850624e-05,
+      "loss": 0.1866,
+      "step": 46408
+    },
+    {
+      "epoch": 3.759640311082307,
+      "grad_norm": 0.07315077632665634,
+      "learning_rate": 1.3353436248255999e-05,
+      "loss": 0.2394,
+      "step": 46409
+    },
+    {
+      "epoch": 3.7597213220998054,
+      "grad_norm": 0.06642105430364609,
+      "learning_rate": 1.334893559566137e-05,
+      "loss": 0.2194,
+      "step": 46410
+    },
+    {
+      "epoch": 3.759802333117304,
+      "grad_norm": 0.06641194224357605,
+      "learning_rate": 1.3344434943066744e-05,
+      "loss": 0.2434,
+      "step": 46411
+    },
+    {
+      "epoch": 3.7598833441348023,
+      "grad_norm": 0.07174092531204224,
+      "learning_rate": 1.333993429047212e-05,
+      "loss": 0.2248,
+      "step": 46412
+    },
+    {
+      "epoch": 3.7599643551523005,
+      "grad_norm": 0.08386538177728653,
+      "learning_rate": 1.3335433637877492e-05,
+      "loss": 0.2254,
+      "step": 46413
+    },
+    {
+      "epoch": 3.7600453661697992,
+      "grad_norm": 0.07093953341245651,
+      "learning_rate": 1.3330932985282869e-05,
+      "loss": 0.2036,
+      "step": 46414
+    },
+    {
+      "epoch": 3.7601263771872975,
+      "grad_norm": 0.06664960831403732,
+      "learning_rate": 1.332643233268824e-05,
+      "loss": 0.1924,
+      "step": 46415
+    },
+    {
+      "epoch": 3.7602073882047957,
+      "grad_norm": 0.07157678157091141,
+      "learning_rate": 1.3321931680093614e-05,
+      "loss": 0.2119,
+      "step": 46416
+    },
+    {
+      "epoch": 3.7602883992222944,
+      "grad_norm": 0.07037477195262909,
+      "learning_rate": 1.331743102749899e-05,
+      "loss": 0.1769,
+      "step": 46417
+    },
+    {
+      "epoch": 3.7603694102397927,
+      "grad_norm": 0.07439891993999481,
+      "learning_rate": 1.3312930374904361e-05,
+      "loss": 0.2255,
+      "step": 46418
+    },
+    {
+      "epoch": 3.760450421257291,
+      "grad_norm": 0.0761137455701828,
+      "learning_rate": 1.3308429722309735e-05,
+      "loss": 0.2132,
+      "step": 46419
+    },
+    {
+      "epoch": 3.7605314322747896,
+      "grad_norm": 0.061975765973329544,
+      "learning_rate": 1.330392906971511e-05,
+      "loss": 0.1905,
+      "step": 46420
+    },
+    {
+      "epoch": 3.760612443292288,
+      "grad_norm": 0.06800936907529831,
+      "learning_rate": 1.3299428417120482e-05,
+      "loss": 0.2167,
+      "step": 46421
+    },
+    {
+      "epoch": 3.760693454309786,
+      "grad_norm": 0.06478226184844971,
+      "learning_rate": 1.3294927764525857e-05,
+      "loss": 0.2169,
+      "step": 46422
+    },
+    {
+      "epoch": 3.7607744653272848,
+      "grad_norm": 0.06802545487880707,
+      "learning_rate": 1.3290427111931231e-05,
+      "loss": 0.1845,
+      "step": 46423
+    },
+    {
+      "epoch": 3.760855476344783,
+      "grad_norm": 0.09708104282617569,
+      "learning_rate": 1.3285926459336603e-05,
+      "loss": 0.2462,
+      "step": 46424
+    },
+    {
+      "epoch": 3.7609364873622813,
+      "grad_norm": 0.06333111971616745,
+      "learning_rate": 1.3281425806741978e-05,
+      "loss": 0.209,
+      "step": 46425
+    },
+    {
+      "epoch": 3.7610174983797795,
+      "grad_norm": 0.07289381325244904,
+      "learning_rate": 1.3276925154147352e-05,
+      "loss": 0.1954,
+      "step": 46426
+    },
+    {
+      "epoch": 3.761098509397278,
+      "grad_norm": 0.06891093403100967,
+      "learning_rate": 1.3272424501552727e-05,
+      "loss": 0.216,
+      "step": 46427
+    },
+    {
+      "epoch": 3.7611795204147764,
+      "grad_norm": 0.06211341172456741,
+      "learning_rate": 1.3267923848958099e-05,
+      "loss": 0.217,
+      "step": 46428
+    },
+    {
+      "epoch": 3.7612605314322747,
+      "grad_norm": 0.07642439752817154,
+      "learning_rate": 1.3263423196363473e-05,
+      "loss": 0.2142,
+      "step": 46429
+    },
+    {
+      "epoch": 3.761341542449773,
+      "grad_norm": 0.06167442724108696,
+      "learning_rate": 1.3258922543768848e-05,
+      "loss": 0.2275,
+      "step": 46430
+    },
+    {
+      "epoch": 3.7614225534672716,
+      "grad_norm": 0.0749720111489296,
+      "learning_rate": 1.3254421891174222e-05,
+      "loss": 0.2434,
+      "step": 46431
+    },
+    {
+      "epoch": 3.76150356448477,
+      "grad_norm": 0.07655184715986252,
+      "learning_rate": 1.3249921238579593e-05,
+      "loss": 0.2593,
+      "step": 46432
+    },
+    {
+      "epoch": 3.761584575502268,
+      "grad_norm": 0.07166169583797455,
+      "learning_rate": 1.3245420585984969e-05,
+      "loss": 0.2194,
+      "step": 46433
+    },
+    {
+      "epoch": 3.761665586519767,
+      "grad_norm": 0.07488848268985748,
+      "learning_rate": 1.3240919933390342e-05,
+      "loss": 0.228,
+      "step": 46434
+    },
+    {
+      "epoch": 3.761746597537265,
+      "grad_norm": 0.07291141152381897,
+      "learning_rate": 1.3236419280795718e-05,
+      "loss": 0.2271,
+      "step": 46435
+    },
+    {
+      "epoch": 3.7618276085547633,
+      "grad_norm": 0.06775379180908203,
+      "learning_rate": 1.323191862820109e-05,
+      "loss": 0.2234,
+      "step": 46436
+    },
+    {
+      "epoch": 3.761908619572262,
+      "grad_norm": 0.09158001095056534,
+      "learning_rate": 1.3227417975606463e-05,
+      "loss": 0.2341,
+      "step": 46437
+    },
+    {
+      "epoch": 3.76198963058976,
+      "grad_norm": 0.07301585376262665,
+      "learning_rate": 1.3222917323011838e-05,
+      "loss": 0.2097,
+      "step": 46438
+    },
+    {
+      "epoch": 3.7620706416072585,
+      "grad_norm": 0.06108618527650833,
+      "learning_rate": 1.321841667041721e-05,
+      "loss": 0.2379,
+      "step": 46439
+    },
+    {
+      "epoch": 3.762151652624757,
+      "grad_norm": 0.08330275863409042,
+      "learning_rate": 1.3213916017822586e-05,
+      "loss": 0.2056,
+      "step": 46440
+    },
+    {
+      "epoch": 3.7622326636422554,
+      "grad_norm": 0.08911581337451935,
+      "learning_rate": 1.320941536522796e-05,
+      "loss": 0.2901,
+      "step": 46441
+    },
+    {
+      "epoch": 3.7623136746597536,
+      "grad_norm": 0.0730244368314743,
+      "learning_rate": 1.3204914712633331e-05,
+      "loss": 0.2179,
+      "step": 46442
+    },
+    {
+      "epoch": 3.7623946856772523,
+      "grad_norm": 0.08556057512760162,
+      "learning_rate": 1.3200414060038706e-05,
+      "loss": 0.2335,
+      "step": 46443
+    },
+    {
+      "epoch": 3.7624756966947506,
+      "grad_norm": 0.07071968913078308,
+      "learning_rate": 1.319591340744408e-05,
+      "loss": 0.2253,
+      "step": 46444
+    },
+    {
+      "epoch": 3.762556707712249,
+      "grad_norm": 0.07956349104642868,
+      "learning_rate": 1.3191412754849452e-05,
+      "loss": 0.2367,
+      "step": 46445
+    },
+    {
+      "epoch": 3.7626377187297475,
+      "grad_norm": 0.06337764859199524,
+      "learning_rate": 1.3186912102254827e-05,
+      "loss": 0.1947,
+      "step": 46446
+    },
+    {
+      "epoch": 3.7627187297472457,
+      "grad_norm": 0.060510363429784775,
+      "learning_rate": 1.3182411449660201e-05,
+      "loss": 0.2314,
+      "step": 46447
+    },
+    {
+      "epoch": 3.762799740764744,
+      "grad_norm": 0.062422338873147964,
+      "learning_rate": 1.3177910797065576e-05,
+      "loss": 0.1888,
+      "step": 46448
+    },
+    {
+      "epoch": 3.7628807517822422,
+      "grad_norm": 0.07294078171253204,
+      "learning_rate": 1.317341014447095e-05,
+      "loss": 0.2152,
+      "step": 46449
+    },
+    {
+      "epoch": 3.762961762799741,
+      "grad_norm": 0.0869693011045456,
+      "learning_rate": 1.3168909491876322e-05,
+      "loss": 0.227,
+      "step": 46450
+    },
+    {
+      "epoch": 3.763042773817239,
+      "grad_norm": 0.08141772449016571,
+      "learning_rate": 1.3164408839281697e-05,
+      "loss": 0.2268,
+      "step": 46451
+    },
+    {
+      "epoch": 3.7631237848347374,
+      "grad_norm": 0.07432594150304794,
+      "learning_rate": 1.315990818668707e-05,
+      "loss": 0.1958,
+      "step": 46452
+    },
+    {
+      "epoch": 3.7632047958522357,
+      "grad_norm": 0.07830096781253815,
+      "learning_rate": 1.3155407534092446e-05,
+      "loss": 0.2131,
+      "step": 46453
+    },
+    {
+      "epoch": 3.7632858068697344,
+      "grad_norm": 0.06278984993696213,
+      "learning_rate": 1.3150906881497818e-05,
+      "loss": 0.2175,
+      "step": 46454
+    },
+    {
+      "epoch": 3.7633668178872326,
+      "grad_norm": 0.06053311750292778,
+      "learning_rate": 1.3146406228903191e-05,
+      "loss": 0.2049,
+      "step": 46455
+    },
+    {
+      "epoch": 3.763447828904731,
+      "grad_norm": 0.0874246209859848,
+      "learning_rate": 1.3141905576308567e-05,
+      "loss": 0.2286,
+      "step": 46456
+    },
+    {
+      "epoch": 3.7635288399222295,
+      "grad_norm": 0.05780855566263199,
+      "learning_rate": 1.3137404923713939e-05,
+      "loss": 0.2368,
+      "step": 46457
+    },
+    {
+      "epoch": 3.7636098509397278,
+      "grad_norm": 0.057224925607442856,
+      "learning_rate": 1.3132904271119312e-05,
+      "loss": 0.1832,
+      "step": 46458
+    },
+    {
+      "epoch": 3.763690861957226,
+      "grad_norm": 0.08486709743738174,
+      "learning_rate": 1.3128403618524688e-05,
+      "loss": 0.1981,
+      "step": 46459
+    },
+    {
+      "epoch": 3.7637718729747247,
+      "grad_norm": 0.08651462197303772,
+      "learning_rate": 1.312390296593006e-05,
+      "loss": 0.2557,
+      "step": 46460
+    },
+    {
+      "epoch": 3.763852883992223,
+      "grad_norm": 0.05747639387845993,
+      "learning_rate": 1.3119402313335435e-05,
+      "loss": 0.1794,
+      "step": 46461
+    },
+    {
+      "epoch": 3.763933895009721,
+      "grad_norm": 0.04803988337516785,
+      "learning_rate": 1.3114901660740808e-05,
+      "loss": 0.2113,
+      "step": 46462
+    },
+    {
+      "epoch": 3.76401490602722,
+      "grad_norm": 0.0615316703915596,
+      "learning_rate": 1.311040100814618e-05,
+      "loss": 0.2208,
+      "step": 46463
+    },
+    {
+      "epoch": 3.764095917044718,
+      "grad_norm": 0.09276503324508667,
+      "learning_rate": 1.3105900355551556e-05,
+      "loss": 0.2528,
+      "step": 46464
+    },
+    {
+      "epoch": 3.7641769280622164,
+      "grad_norm": 0.07706604897975922,
+      "learning_rate": 1.3101399702956929e-05,
+      "loss": 0.2107,
+      "step": 46465
+    },
+    {
+      "epoch": 3.764257939079715,
+      "grad_norm": 0.0845157578587532,
+      "learning_rate": 1.3096899050362301e-05,
+      "loss": 0.2054,
+      "step": 46466
+    },
+    {
+      "epoch": 3.7643389500972133,
+      "grad_norm": 0.08171258866786957,
+      "learning_rate": 1.3092398397767678e-05,
+      "loss": 0.2411,
+      "step": 46467
+    },
+    {
+      "epoch": 3.7644199611147116,
+      "grad_norm": 0.08288677036762238,
+      "learning_rate": 1.308789774517305e-05,
+      "loss": 0.206,
+      "step": 46468
+    },
+    {
+      "epoch": 3.7645009721322102,
+      "grad_norm": 0.06667889654636383,
+      "learning_rate": 1.3083397092578425e-05,
+      "loss": 0.2449,
+      "step": 46469
+    },
+    {
+      "epoch": 3.7645819831497085,
+      "grad_norm": 0.060472164303064346,
+      "learning_rate": 1.3078896439983799e-05,
+      "loss": 0.2135,
+      "step": 46470
+    },
+    {
+      "epoch": 3.7646629941672067,
+      "grad_norm": 0.07328839600086212,
+      "learning_rate": 1.307439578738917e-05,
+      "loss": 0.2034,
+      "step": 46471
+    },
+    {
+      "epoch": 3.764744005184705,
+      "grad_norm": 0.06965433806180954,
+      "learning_rate": 1.3069895134794546e-05,
+      "loss": 0.1813,
+      "step": 46472
+    },
+    {
+      "epoch": 3.7648250162022032,
+      "grad_norm": 0.08212985843420029,
+      "learning_rate": 1.306539448219992e-05,
+      "loss": 0.2399,
+      "step": 46473
+    },
+    {
+      "epoch": 3.764906027219702,
+      "grad_norm": 0.06601189821958542,
+      "learning_rate": 1.3060893829605295e-05,
+      "loss": 0.202,
+      "step": 46474
+    },
+    {
+      "epoch": 3.7649870382372,
+      "grad_norm": 0.08131560683250427,
+      "learning_rate": 1.3056393177010667e-05,
+      "loss": 0.2358,
+      "step": 46475
+    },
+    {
+      "epoch": 3.7650680492546984,
+      "grad_norm": 0.07608696073293686,
+      "learning_rate": 1.305189252441604e-05,
+      "loss": 0.1951,
+      "step": 46476
+    },
+    {
+      "epoch": 3.765149060272197,
+      "grad_norm": 0.07890737056732178,
+      "learning_rate": 1.3047391871821416e-05,
+      "loss": 0.2582,
+      "step": 46477
+    },
+    {
+      "epoch": 3.7652300712896953,
+      "grad_norm": 0.09459563344717026,
+      "learning_rate": 1.3042891219226788e-05,
+      "loss": 0.2385,
+      "step": 46478
+    },
+    {
+      "epoch": 3.7653110823071936,
+      "grad_norm": 0.06472989171743393,
+      "learning_rate": 1.3038390566632161e-05,
+      "loss": 0.1975,
+      "step": 46479
+    },
+    {
+      "epoch": 3.7653920933246923,
+      "grad_norm": 0.10067077726125717,
+      "learning_rate": 1.3033889914037537e-05,
+      "loss": 0.2349,
+      "step": 46480
+    },
+    {
+      "epoch": 3.7654731043421905,
+      "grad_norm": 0.07255268096923828,
+      "learning_rate": 1.3029389261442908e-05,
+      "loss": 0.2272,
+      "step": 46481
+    },
+    {
+      "epoch": 3.7655541153596888,
+      "grad_norm": 0.06430191546678543,
+      "learning_rate": 1.3024888608848285e-05,
+      "loss": 0.1932,
+      "step": 46482
+    },
+    {
+      "epoch": 3.7656351263771874,
+      "grad_norm": 0.08290670067071915,
+      "learning_rate": 1.3020387956253657e-05,
+      "loss": 0.2751,
+      "step": 46483
+    },
+    {
+      "epoch": 3.7657161373946857,
+      "grad_norm": 0.0648224726319313,
+      "learning_rate": 1.301588730365903e-05,
+      "loss": 0.2181,
+      "step": 46484
+    },
+    {
+      "epoch": 3.765797148412184,
+      "grad_norm": 0.07298358529806137,
+      "learning_rate": 1.3011386651064406e-05,
+      "loss": 0.2469,
+      "step": 46485
+    },
+    {
+      "epoch": 3.7658781594296826,
+      "grad_norm": 0.07706646621227264,
+      "learning_rate": 1.3006885998469778e-05,
+      "loss": 0.2174,
+      "step": 46486
+    },
+    {
+      "epoch": 3.765959170447181,
+      "grad_norm": 0.07118535041809082,
+      "learning_rate": 1.3002385345875153e-05,
+      "loss": 0.223,
+      "step": 46487
+    },
+    {
+      "epoch": 3.766040181464679,
+      "grad_norm": 0.09811242669820786,
+      "learning_rate": 1.2997884693280527e-05,
+      "loss": 0.2232,
+      "step": 46488
+    },
+    {
+      "epoch": 3.766121192482178,
+      "grad_norm": 0.0811871886253357,
+      "learning_rate": 1.2993384040685899e-05,
+      "loss": 0.2519,
+      "step": 46489
+    },
+    {
+      "epoch": 3.766202203499676,
+      "grad_norm": 0.09767928719520569,
+      "learning_rate": 1.2988883388091274e-05,
+      "loss": 0.2349,
+      "step": 46490
+    },
+    {
+      "epoch": 3.7662832145171743,
+      "grad_norm": 0.07688478380441666,
+      "learning_rate": 1.2984382735496648e-05,
+      "loss": 0.2133,
+      "step": 46491
+    },
+    {
+      "epoch": 3.766364225534673,
+      "grad_norm": 0.0702710896730423,
+      "learning_rate": 1.297988208290202e-05,
+      "loss": 0.1993,
+      "step": 46492
+    },
+    {
+      "epoch": 3.7664452365521712,
+      "grad_norm": 0.0628383606672287,
+      "learning_rate": 1.2975381430307395e-05,
+      "loss": 0.197,
+      "step": 46493
+    },
+    {
+      "epoch": 3.7665262475696695,
+      "grad_norm": 0.06867960095405579,
+      "learning_rate": 1.2970880777712769e-05,
+      "loss": 0.2463,
+      "step": 46494
+    },
+    {
+      "epoch": 3.7666072585871677,
+      "grad_norm": 0.08608069270849228,
+      "learning_rate": 1.2966380125118144e-05,
+      "loss": 0.224,
+      "step": 46495
+    },
+    {
+      "epoch": 3.766688269604666,
+      "grad_norm": 0.06254792213439941,
+      "learning_rate": 1.2961879472523516e-05,
+      "loss": 0.2338,
+      "step": 46496
+    },
+    {
+      "epoch": 3.7667692806221647,
+      "grad_norm": 0.0764245092868805,
+      "learning_rate": 1.295737881992889e-05,
+      "loss": 0.1998,
+      "step": 46497
+    },
+    {
+      "epoch": 3.766850291639663,
+      "grad_norm": 0.08328287303447723,
+      "learning_rate": 1.2952878167334265e-05,
+      "loss": 0.226,
+      "step": 46498
+    },
+    {
+      "epoch": 3.766931302657161,
+      "grad_norm": 0.08136755973100662,
+      "learning_rate": 1.2948377514739637e-05,
+      "loss": 0.2651,
+      "step": 46499
+    },
+    {
+      "epoch": 3.76701231367466,
+      "grad_norm": 0.08298899978399277,
+      "learning_rate": 1.2943876862145014e-05,
+      "loss": 0.2291,
+      "step": 46500
+    },
+    {
+      "epoch": 3.767093324692158,
+      "grad_norm": 0.06559506058692932,
+      "learning_rate": 1.2939376209550386e-05,
+      "loss": 0.2238,
+      "step": 46501
+    },
+    {
+      "epoch": 3.7671743357096563,
+      "grad_norm": 0.08516070246696472,
+      "learning_rate": 1.2934875556955758e-05,
+      "loss": 0.2544,
+      "step": 46502
+    },
+    {
+      "epoch": 3.767255346727155,
+      "grad_norm": 0.06356891244649887,
+      "learning_rate": 1.2930374904361135e-05,
+      "loss": 0.2192,
+      "step": 46503
+    },
+    {
+      "epoch": 3.7673363577446533,
+      "grad_norm": 0.07620159536600113,
+      "learning_rate": 1.2925874251766506e-05,
+      "loss": 0.2559,
+      "step": 46504
+    },
+    {
+      "epoch": 3.7674173687621515,
+      "grad_norm": 0.06729257851839066,
+      "learning_rate": 1.292137359917188e-05,
+      "loss": 0.1718,
+      "step": 46505
+    },
+    {
+      "epoch": 3.76749837977965,
+      "grad_norm": 0.0670362114906311,
+      "learning_rate": 1.2916872946577255e-05,
+      "loss": 0.1767,
+      "step": 46506
+    },
+    {
+      "epoch": 3.7675793907971484,
+      "grad_norm": 0.07508490234613419,
+      "learning_rate": 1.2912372293982627e-05,
+      "loss": 0.2409,
+      "step": 46507
+    },
+    {
+      "epoch": 3.7676604018146467,
+      "grad_norm": 0.07815518975257874,
+      "learning_rate": 1.2907871641388003e-05,
+      "loss": 0.2148,
+      "step": 46508
+    },
+    {
+      "epoch": 3.7677414128321454,
+      "grad_norm": 0.07622785121202469,
+      "learning_rate": 1.2903370988793376e-05,
+      "loss": 0.2404,
+      "step": 46509
+    },
+    {
+      "epoch": 3.7678224238496436,
+      "grad_norm": 0.07081098109483719,
+      "learning_rate": 1.2898870336198748e-05,
+      "loss": 0.1935,
+      "step": 46510
+    },
+    {
+      "epoch": 3.767903434867142,
+      "grad_norm": 0.0799146220088005,
+      "learning_rate": 1.2894369683604123e-05,
+      "loss": 0.2275,
+      "step": 46511
+    },
+    {
+      "epoch": 3.7679844458846405,
+      "grad_norm": 0.08372996002435684,
+      "learning_rate": 1.2889869031009497e-05,
+      "loss": 0.2283,
+      "step": 46512
+    },
+    {
+      "epoch": 3.768065456902139,
+      "grad_norm": 0.07471054792404175,
+      "learning_rate": 1.2885368378414872e-05,
+      "loss": 0.2373,
+      "step": 46513
+    },
+    {
+      "epoch": 3.768146467919637,
+      "grad_norm": 0.07189594954252243,
+      "learning_rate": 1.2880867725820244e-05,
+      "loss": 0.1982,
+      "step": 46514
+    },
+    {
+      "epoch": 3.7682274789371357,
+      "grad_norm": 0.07205279171466827,
+      "learning_rate": 1.2876367073225618e-05,
+      "loss": 0.2298,
+      "step": 46515
+    },
+    {
+      "epoch": 3.768308489954634,
+      "grad_norm": 0.06825879216194153,
+      "learning_rate": 1.2871866420630993e-05,
+      "loss": 0.2154,
+      "step": 46516
+    },
+    {
+      "epoch": 3.768389500972132,
+      "grad_norm": 0.06254030764102936,
+      "learning_rate": 1.2867365768036365e-05,
+      "loss": 0.2073,
+      "step": 46517
+    },
+    {
+      "epoch": 3.7684705119896305,
+      "grad_norm": 0.06677734106779099,
+      "learning_rate": 1.2862865115441739e-05,
+      "loss": 0.1922,
+      "step": 46518
+    },
+    {
+      "epoch": 3.7685515230071287,
+      "grad_norm": 0.06440742313861847,
+      "learning_rate": 1.2858364462847114e-05,
+      "loss": 0.2021,
+      "step": 46519
+    },
+    {
+      "epoch": 3.7686325340246274,
+      "grad_norm": 0.08662337809801102,
+      "learning_rate": 1.2853863810252487e-05,
+      "loss": 0.2315,
+      "step": 46520
+    },
+    {
+      "epoch": 3.7687135450421256,
+      "grad_norm": 0.06901124119758606,
+      "learning_rate": 1.2849363157657863e-05,
+      "loss": 0.1859,
+      "step": 46521
+    },
+    {
+      "epoch": 3.768794556059624,
+      "grad_norm": 0.08281880617141724,
+      "learning_rate": 1.2844862505063235e-05,
+      "loss": 0.2549,
+      "step": 46522
+    },
+    {
+      "epoch": 3.7688755670771226,
+      "grad_norm": 0.07393230497837067,
+      "learning_rate": 1.2840361852468608e-05,
+      "loss": 0.2464,
+      "step": 46523
+    },
+    {
+      "epoch": 3.768956578094621,
+      "grad_norm": 0.07908966392278671,
+      "learning_rate": 1.2835861199873984e-05,
+      "loss": 0.2262,
+      "step": 46524
+    },
+    {
+      "epoch": 3.769037589112119,
+      "grad_norm": 0.08938897401094437,
+      "learning_rate": 1.2831360547279356e-05,
+      "loss": 0.2342,
+      "step": 46525
+    },
+    {
+      "epoch": 3.7691186001296177,
+      "grad_norm": 0.0999879539012909,
+      "learning_rate": 1.282685989468473e-05,
+      "loss": 0.2072,
+      "step": 46526
+    },
+    {
+      "epoch": 3.769199611147116,
+      "grad_norm": 0.05360986292362213,
+      "learning_rate": 1.2822359242090104e-05,
+      "loss": 0.1837,
+      "step": 46527
+    },
+    {
+      "epoch": 3.7692806221646142,
+      "grad_norm": 0.06904050707817078,
+      "learning_rate": 1.2817858589495476e-05,
+      "loss": 0.2043,
+      "step": 46528
+    },
+    {
+      "epoch": 3.769361633182113,
+      "grad_norm": 0.0860978439450264,
+      "learning_rate": 1.2813357936900852e-05,
+      "loss": 0.2197,
+      "step": 46529
+    },
+    {
+      "epoch": 3.769442644199611,
+      "grad_norm": 0.0657324492931366,
+      "learning_rate": 1.2808857284306225e-05,
+      "loss": 0.2159,
+      "step": 46530
+    },
+    {
+      "epoch": 3.7695236552171094,
+      "grad_norm": 0.06086688116192818,
+      "learning_rate": 1.2804356631711597e-05,
+      "loss": 0.214,
+      "step": 46531
+    },
+    {
+      "epoch": 3.769604666234608,
+      "grad_norm": 0.07376797497272491,
+      "learning_rate": 1.2799855979116972e-05,
+      "loss": 0.249,
+      "step": 46532
+    },
+    {
+      "epoch": 3.7696856772521063,
+      "grad_norm": 0.0695323571562767,
+      "learning_rate": 1.2795355326522346e-05,
+      "loss": 0.2095,
+      "step": 46533
+    },
+    {
+      "epoch": 3.7697666882696046,
+      "grad_norm": 0.0635128989815712,
+      "learning_rate": 1.2790854673927721e-05,
+      "loss": 0.1936,
+      "step": 46534
+    },
+    {
+      "epoch": 3.7698476992871033,
+      "grad_norm": 0.07343027740716934,
+      "learning_rate": 1.2786354021333093e-05,
+      "loss": 0.2146,
+      "step": 46535
+    },
+    {
+      "epoch": 3.7699287103046015,
+      "grad_norm": 0.06408240646123886,
+      "learning_rate": 1.2781853368738467e-05,
+      "loss": 0.2182,
+      "step": 46536
+    },
+    {
+      "epoch": 3.7700097213220998,
+      "grad_norm": 0.08215359598398209,
+      "learning_rate": 1.2777352716143842e-05,
+      "loss": 0.2187,
+      "step": 46537
+    },
+    {
+      "epoch": 3.7700907323395985,
+      "grad_norm": 0.07449215650558472,
+      "learning_rate": 1.2772852063549216e-05,
+      "loss": 0.2312,
+      "step": 46538
+    },
+    {
+      "epoch": 3.7701717433570967,
+      "grad_norm": 0.07269278913736343,
+      "learning_rate": 1.2768351410954588e-05,
+      "loss": 0.2118,
+      "step": 46539
+    },
+    {
+      "epoch": 3.770252754374595,
+      "grad_norm": 0.0680905431509018,
+      "learning_rate": 1.2763850758359963e-05,
+      "loss": 0.1955,
+      "step": 46540
+    },
+    {
+      "epoch": 3.770333765392093,
+      "grad_norm": 0.06702621281147003,
+      "learning_rate": 1.2759350105765337e-05,
+      "loss": 0.2244,
+      "step": 46541
+    },
+    {
+      "epoch": 3.7704147764095914,
+      "grad_norm": 0.09051982313394547,
+      "learning_rate": 1.2754849453170712e-05,
+      "loss": 0.2369,
+      "step": 46542
+    },
+    {
+      "epoch": 3.77049578742709,
+      "grad_norm": 0.07590014487504959,
+      "learning_rate": 1.2750348800576084e-05,
+      "loss": 0.2358,
+      "step": 46543
+    },
+    {
+      "epoch": 3.7705767984445884,
+      "grad_norm": 0.0708058774471283,
+      "learning_rate": 1.2745848147981457e-05,
+      "loss": 0.2117,
+      "step": 46544
+    },
+    {
+      "epoch": 3.7706578094620866,
+      "grad_norm": 0.0713592991232872,
+      "learning_rate": 1.2741347495386833e-05,
+      "loss": 0.2053,
+      "step": 46545
+    },
+    {
+      "epoch": 3.7707388204795853,
+      "grad_norm": 0.07061046361923218,
+      "learning_rate": 1.2736846842792205e-05,
+      "loss": 0.2271,
+      "step": 46546
+    },
+    {
+      "epoch": 3.7708198314970836,
+      "grad_norm": 0.08287011831998825,
+      "learning_rate": 1.273234619019758e-05,
+      "loss": 0.2432,
+      "step": 46547
+    },
+    {
+      "epoch": 3.770900842514582,
+      "grad_norm": 0.07770534604787827,
+      "learning_rate": 1.2727845537602953e-05,
+      "loss": 0.2462,
+      "step": 46548
+    },
+    {
+      "epoch": 3.7709818535320805,
+      "grad_norm": 0.07193398475646973,
+      "learning_rate": 1.2723344885008325e-05,
+      "loss": 0.218,
+      "step": 46549
+    },
+    {
+      "epoch": 3.7710628645495787,
+      "grad_norm": 0.06820261478424072,
+      "learning_rate": 1.27188442324137e-05,
+      "loss": 0.236,
+      "step": 46550
+    },
+    {
+      "epoch": 3.771143875567077,
+      "grad_norm": 0.07764853537082672,
+      "learning_rate": 1.2714343579819074e-05,
+      "loss": 0.2336,
+      "step": 46551
+    },
+    {
+      "epoch": 3.7712248865845757,
+      "grad_norm": 0.07523754239082336,
+      "learning_rate": 1.2709842927224446e-05,
+      "loss": 0.2062,
+      "step": 46552
+    },
+    {
+      "epoch": 3.771305897602074,
+      "grad_norm": 0.07026119530200958,
+      "learning_rate": 1.2705342274629821e-05,
+      "loss": 0.2321,
+      "step": 46553
+    },
+    {
+      "epoch": 3.771386908619572,
+      "grad_norm": 0.06876692920923233,
+      "learning_rate": 1.2700841622035195e-05,
+      "loss": 0.2054,
+      "step": 46554
+    },
+    {
+      "epoch": 3.771467919637071,
+      "grad_norm": 0.06335452198982239,
+      "learning_rate": 1.269634096944057e-05,
+      "loss": 0.1917,
+      "step": 46555
+    },
+    {
+      "epoch": 3.771548930654569,
+      "grad_norm": 0.07451608031988144,
+      "learning_rate": 1.2691840316845944e-05,
+      "loss": 0.255,
+      "step": 46556
+    },
+    {
+      "epoch": 3.7716299416720673,
+      "grad_norm": 0.07577665895223618,
+      "learning_rate": 1.2687339664251316e-05,
+      "loss": 0.2117,
+      "step": 46557
+    },
+    {
+      "epoch": 3.771710952689566,
+      "grad_norm": 0.07046057283878326,
+      "learning_rate": 1.2682839011656691e-05,
+      "loss": 0.2154,
+      "step": 46558
+    },
+    {
+      "epoch": 3.7717919637070643,
+      "grad_norm": 0.09273316711187363,
+      "learning_rate": 1.2678338359062065e-05,
+      "loss": 0.3072,
+      "step": 46559
+    },
+    {
+      "epoch": 3.7718729747245625,
+      "grad_norm": 0.07682998478412628,
+      "learning_rate": 1.267383770646744e-05,
+      "loss": 0.1781,
+      "step": 46560
+    },
+    {
+      "epoch": 3.7719539857420608,
+      "grad_norm": 0.0753307119011879,
+      "learning_rate": 1.2669337053872812e-05,
+      "loss": 0.2,
+      "step": 46561
+    },
+    {
+      "epoch": 3.7720349967595594,
+      "grad_norm": 0.06994988024234772,
+      "learning_rate": 1.2664836401278186e-05,
+      "loss": 0.2132,
+      "step": 46562
+    },
+    {
+      "epoch": 3.7721160077770577,
+      "grad_norm": 0.08743558824062347,
+      "learning_rate": 1.2660335748683561e-05,
+      "loss": 0.2153,
+      "step": 46563
+    },
+    {
+      "epoch": 3.772197018794556,
+      "grad_norm": 0.06563448905944824,
+      "learning_rate": 1.2655835096088933e-05,
+      "loss": 0.2165,
+      "step": 46564
+    },
+    {
+      "epoch": 3.772278029812054,
+      "grad_norm": 0.0646919459104538,
+      "learning_rate": 1.2651334443494306e-05,
+      "loss": 0.1949,
+      "step": 46565
+    },
+    {
+      "epoch": 3.772359040829553,
+      "grad_norm": 0.06894893944263458,
+      "learning_rate": 1.2646833790899682e-05,
+      "loss": 0.2411,
+      "step": 46566
+    },
+    {
+      "epoch": 3.772440051847051,
+      "grad_norm": 0.06888721883296967,
+      "learning_rate": 1.2642333138305054e-05,
+      "loss": 0.2475,
+      "step": 46567
+    },
+    {
+      "epoch": 3.7725210628645494,
+      "grad_norm": 0.0673423707485199,
+      "learning_rate": 1.2637832485710429e-05,
+      "loss": 0.2775,
+      "step": 46568
+    },
+    {
+      "epoch": 3.772602073882048,
+      "grad_norm": 0.06546924263238907,
+      "learning_rate": 1.2633331833115803e-05,
+      "loss": 0.248,
+      "step": 46569
+    },
+    {
+      "epoch": 3.7726830848995463,
+      "grad_norm": 0.06502770632505417,
+      "learning_rate": 1.2628831180521174e-05,
+      "loss": 0.2148,
+      "step": 46570
+    },
+    {
+      "epoch": 3.7727640959170445,
+      "grad_norm": 0.07278035581111908,
+      "learning_rate": 1.2624330527926551e-05,
+      "loss": 0.2191,
+      "step": 46571
+    },
+    {
+      "epoch": 3.7728451069345432,
+      "grad_norm": 0.0673990324139595,
+      "learning_rate": 1.2619829875331923e-05,
+      "loss": 0.2305,
+      "step": 46572
+    },
+    {
+      "epoch": 3.7729261179520415,
+      "grad_norm": 0.07896624505519867,
+      "learning_rate": 1.2615329222737299e-05,
+      "loss": 0.2684,
+      "step": 46573
+    },
+    {
+      "epoch": 3.7730071289695397,
+      "grad_norm": 0.06549516320228577,
+      "learning_rate": 1.2610828570142672e-05,
+      "loss": 0.2069,
+      "step": 46574
+    },
+    {
+      "epoch": 3.7730881399870384,
+      "grad_norm": 0.06927746534347534,
+      "learning_rate": 1.2606327917548044e-05,
+      "loss": 0.1988,
+      "step": 46575
+    },
+    {
+      "epoch": 3.7731691510045366,
+      "grad_norm": 0.07695776224136353,
+      "learning_rate": 1.260182726495342e-05,
+      "loss": 0.2593,
+      "step": 46576
+    },
+    {
+      "epoch": 3.773250162022035,
+      "grad_norm": 0.0668836161494255,
+      "learning_rate": 1.2597326612358793e-05,
+      "loss": 0.2381,
+      "step": 46577
+    },
+    {
+      "epoch": 3.7733311730395336,
+      "grad_norm": 0.0662456750869751,
+      "learning_rate": 1.2592825959764165e-05,
+      "loss": 0.2232,
+      "step": 46578
+    },
+    {
+      "epoch": 3.773412184057032,
+      "grad_norm": 0.07020401209592819,
+      "learning_rate": 1.258832530716954e-05,
+      "loss": 0.2331,
+      "step": 46579
+    },
+    {
+      "epoch": 3.77349319507453,
+      "grad_norm": 0.07420952618122101,
+      "learning_rate": 1.2583824654574914e-05,
+      "loss": 0.2642,
+      "step": 46580
+    },
+    {
+      "epoch": 3.7735742060920288,
+      "grad_norm": 0.0619262158870697,
+      "learning_rate": 1.2579324001980289e-05,
+      "loss": 0.2069,
+      "step": 46581
+    },
+    {
+      "epoch": 3.773655217109527,
+      "grad_norm": 0.06421462446451187,
+      "learning_rate": 1.2574823349385661e-05,
+      "loss": 0.1812,
+      "step": 46582
+    },
+    {
+      "epoch": 3.7737362281270252,
+      "grad_norm": 0.06833804398775101,
+      "learning_rate": 1.2570322696791035e-05,
+      "loss": 0.1944,
+      "step": 46583
+    },
+    {
+      "epoch": 3.7738172391445235,
+      "grad_norm": 0.08516917377710342,
+      "learning_rate": 1.256582204419641e-05,
+      "loss": 0.2427,
+      "step": 46584
+    },
+    {
+      "epoch": 3.773898250162022,
+      "grad_norm": 0.08085508644580841,
+      "learning_rate": 1.2561321391601782e-05,
+      "loss": 0.232,
+      "step": 46585
+    },
+    {
+      "epoch": 3.7739792611795204,
+      "grad_norm": 0.07463680952787399,
+      "learning_rate": 1.2556820739007157e-05,
+      "loss": 0.2403,
+      "step": 46586
+    },
+    {
+      "epoch": 3.7740602721970187,
+      "grad_norm": 0.07635892182588577,
+      "learning_rate": 1.255232008641253e-05,
+      "loss": 0.2527,
+      "step": 46587
+    },
+    {
+      "epoch": 3.774141283214517,
+      "grad_norm": 0.07656805962324142,
+      "learning_rate": 1.2547819433817903e-05,
+      "loss": 0.2201,
+      "step": 46588
+    },
+    {
+      "epoch": 3.7742222942320156,
+      "grad_norm": 0.08051590621471405,
+      "learning_rate": 1.254331878122328e-05,
+      "loss": 0.2406,
+      "step": 46589
+    },
+    {
+      "epoch": 3.774303305249514,
+      "grad_norm": 0.08109819889068604,
+      "learning_rate": 1.2538818128628652e-05,
+      "loss": 0.2175,
+      "step": 46590
+    },
+    {
+      "epoch": 3.774384316267012,
+      "grad_norm": 0.07000181823968887,
+      "learning_rate": 1.2534317476034023e-05,
+      "loss": 0.1995,
+      "step": 46591
+    },
+    {
+      "epoch": 3.774465327284511,
+      "grad_norm": 0.08554103225469589,
+      "learning_rate": 1.25298168234394e-05,
+      "loss": 0.2115,
+      "step": 46592
+    },
+    {
+      "epoch": 3.774546338302009,
+      "grad_norm": 0.06602445989847183,
+      "learning_rate": 1.2525316170844772e-05,
+      "loss": 0.2711,
+      "step": 46593
+    },
+    {
+      "epoch": 3.7746273493195073,
+      "grad_norm": 0.06864854693412781,
+      "learning_rate": 1.2520815518250148e-05,
+      "loss": 0.2129,
+      "step": 46594
+    },
+    {
+      "epoch": 3.774708360337006,
+      "grad_norm": 0.06876920908689499,
+      "learning_rate": 1.2516314865655521e-05,
+      "loss": 0.2075,
+      "step": 46595
+    },
+    {
+      "epoch": 3.774789371354504,
+      "grad_norm": 0.0704292431473732,
+      "learning_rate": 1.2511814213060893e-05,
+      "loss": 0.2022,
+      "step": 46596
+    },
+    {
+      "epoch": 3.7748703823720025,
+      "grad_norm": 0.06541324406862259,
+      "learning_rate": 1.2507313560466268e-05,
+      "loss": 0.2333,
+      "step": 46597
+    },
+    {
+      "epoch": 3.774951393389501,
+      "grad_norm": 0.08963797986507416,
+      "learning_rate": 1.2502812907871642e-05,
+      "loss": 0.2533,
+      "step": 46598
+    },
+    {
+      "epoch": 3.7750324044069994,
+      "grad_norm": 0.0644608810544014,
+      "learning_rate": 1.2498312255277016e-05,
+      "loss": 0.2091,
+      "step": 46599
+    },
+    {
+      "epoch": 3.7751134154244976,
+      "grad_norm": 0.07457215338945389,
+      "learning_rate": 1.249381160268239e-05,
+      "loss": 0.2618,
+      "step": 46600
+    },
+    {
+      "epoch": 3.7751944264419963,
+      "grad_norm": 0.06820184737443924,
+      "learning_rate": 1.2489310950087765e-05,
+      "loss": 0.2127,
+      "step": 46601
+    },
+    {
+      "epoch": 3.7752754374594946,
+      "grad_norm": 0.060140807181596756,
+      "learning_rate": 1.2484810297493137e-05,
+      "loss": 0.2029,
+      "step": 46602
+    },
+    {
+      "epoch": 3.775356448476993,
+      "grad_norm": 0.05878528580069542,
+      "learning_rate": 1.248030964489851e-05,
+      "loss": 0.2461,
+      "step": 46603
+    },
+    {
+      "epoch": 3.7754374594944915,
+      "grad_norm": 0.0732777863740921,
+      "learning_rate": 1.2475808992303885e-05,
+      "loss": 0.2469,
+      "step": 46604
+    },
+    {
+      "epoch": 3.7755184705119897,
+      "grad_norm": 0.07616322487592697,
+      "learning_rate": 1.2471308339709259e-05,
+      "loss": 0.2513,
+      "step": 46605
+    },
+    {
+      "epoch": 3.775599481529488,
+      "grad_norm": 0.07453019171953201,
+      "learning_rate": 1.2466807687114631e-05,
+      "loss": 0.2097,
+      "step": 46606
+    },
+    {
+      "epoch": 3.7756804925469862,
+      "grad_norm": 0.07242731750011444,
+      "learning_rate": 1.2462307034520006e-05,
+      "loss": 0.2427,
+      "step": 46607
+    },
+    {
+      "epoch": 3.775761503564485,
+      "grad_norm": 0.07155609875917435,
+      "learning_rate": 1.245780638192538e-05,
+      "loss": 0.236,
+      "step": 46608
+    },
+    {
+      "epoch": 3.775842514581983,
+      "grad_norm": 0.07473588734865189,
+      "learning_rate": 1.2453305729330753e-05,
+      "loss": 0.2572,
+      "step": 46609
+    },
+    {
+      "epoch": 3.7759235255994814,
+      "grad_norm": 0.06979592144489288,
+      "learning_rate": 1.2448805076736127e-05,
+      "loss": 0.2232,
+      "step": 46610
+    },
+    {
+      "epoch": 3.7760045366169797,
+      "grad_norm": 0.0813089907169342,
+      "learning_rate": 1.24443044241415e-05,
+      "loss": 0.2052,
+      "step": 46611
+    },
+    {
+      "epoch": 3.7760855476344783,
+      "grad_norm": 0.08803302049636841,
+      "learning_rate": 1.2439803771546874e-05,
+      "loss": 0.2069,
+      "step": 46612
+    },
+    {
+      "epoch": 3.7761665586519766,
+      "grad_norm": 0.06549622863531113,
+      "learning_rate": 1.243530311895225e-05,
+      "loss": 0.2477,
+      "step": 46613
+    },
+    {
+      "epoch": 3.776247569669475,
+      "grad_norm": 0.07154756784439087,
+      "learning_rate": 1.2430802466357623e-05,
+      "loss": 0.2317,
+      "step": 46614
+    },
+    {
+      "epoch": 3.7763285806869735,
+      "grad_norm": 0.06701797246932983,
+      "learning_rate": 1.2426301813762995e-05,
+      "loss": 0.211,
+      "step": 46615
+    },
+    {
+      "epoch": 3.7764095917044718,
+      "grad_norm": 0.06661625951528549,
+      "learning_rate": 1.242180116116837e-05,
+      "loss": 0.2281,
+      "step": 46616
+    },
+    {
+      "epoch": 3.77649060272197,
+      "grad_norm": 0.06580673903226852,
+      "learning_rate": 1.2417300508573744e-05,
+      "loss": 0.2164,
+      "step": 46617
+    },
+    {
+      "epoch": 3.7765716137394687,
+      "grad_norm": 0.0783298909664154,
+      "learning_rate": 1.2412799855979118e-05,
+      "loss": 0.2132,
+      "step": 46618
+    },
+    {
+      "epoch": 3.776652624756967,
+      "grad_norm": 0.10125415027141571,
+      "learning_rate": 1.2408299203384491e-05,
+      "loss": 0.2638,
+      "step": 46619
+    },
+    {
+      "epoch": 3.776733635774465,
+      "grad_norm": 0.06981062144041061,
+      "learning_rate": 1.2403798550789865e-05,
+      "loss": 0.1883,
+      "step": 46620
+    },
+    {
+      "epoch": 3.776814646791964,
+      "grad_norm": 0.06791463494300842,
+      "learning_rate": 1.2399297898195238e-05,
+      "loss": 0.2145,
+      "step": 46621
+    },
+    {
+      "epoch": 3.776895657809462,
+      "grad_norm": 0.06609434634447098,
+      "learning_rate": 1.2394797245600614e-05,
+      "loss": 0.2248,
+      "step": 46622
+    },
+    {
+      "epoch": 3.7769766688269604,
+      "grad_norm": 0.065769724547863,
+      "learning_rate": 1.2390296593005986e-05,
+      "loss": 0.2087,
+      "step": 46623
+    },
+    {
+      "epoch": 3.777057679844459,
+      "grad_norm": 0.07486658543348312,
+      "learning_rate": 1.238579594041136e-05,
+      "loss": 0.2623,
+      "step": 46624
+    },
+    {
+      "epoch": 3.7771386908619573,
+      "grad_norm": 0.06038758158683777,
+      "learning_rate": 1.2381295287816734e-05,
+      "loss": 0.1932,
+      "step": 46625
+    },
+    {
+      "epoch": 3.7772197018794555,
+      "grad_norm": 0.07746127247810364,
+      "learning_rate": 1.2376794635222108e-05,
+      "loss": 0.2262,
+      "step": 46626
+    },
+    {
+      "epoch": 3.7773007128969542,
+      "grad_norm": 0.07121626287698746,
+      "learning_rate": 1.2372293982627482e-05,
+      "loss": 0.2153,
+      "step": 46627
+    },
+    {
+      "epoch": 3.7773817239144525,
+      "grad_norm": 0.07847585529088974,
+      "learning_rate": 1.2367793330032855e-05,
+      "loss": 0.1917,
+      "step": 46628
+    },
+    {
+      "epoch": 3.7774627349319507,
+      "grad_norm": 0.0689721554517746,
+      "learning_rate": 1.2363292677438229e-05,
+      "loss": 0.2269,
+      "step": 46629
+    },
+    {
+      "epoch": 3.777543745949449,
+      "grad_norm": 0.08667279779911041,
+      "learning_rate": 1.2358792024843602e-05,
+      "loss": 0.222,
+      "step": 46630
+    },
+    {
+      "epoch": 3.7776247569669477,
+      "grad_norm": 0.06442601978778839,
+      "learning_rate": 1.2354291372248978e-05,
+      "loss": 0.2319,
+      "step": 46631
+    },
+    {
+      "epoch": 3.777705767984446,
+      "grad_norm": 0.0721411257982254,
+      "learning_rate": 1.234979071965435e-05,
+      "loss": 0.1926,
+      "step": 46632
+    },
+    {
+      "epoch": 3.777786779001944,
+      "grad_norm": 0.06399735063314438,
+      "learning_rate": 1.2345290067059723e-05,
+      "loss": 0.1966,
+      "step": 46633
+    },
+    {
+      "epoch": 3.7778677900194424,
+      "grad_norm": 0.0750143975019455,
+      "learning_rate": 1.2340789414465099e-05,
+      "loss": 0.2406,
+      "step": 46634
+    },
+    {
+      "epoch": 3.777948801036941,
+      "grad_norm": 0.08481192588806152,
+      "learning_rate": 1.2336288761870472e-05,
+      "loss": 0.2353,
+      "step": 46635
+    },
+    {
+      "epoch": 3.7780298120544393,
+      "grad_norm": 0.060849450528621674,
+      "learning_rate": 1.2331788109275846e-05,
+      "loss": 0.2453,
+      "step": 46636
+    },
+    {
+      "epoch": 3.7781108230719376,
+      "grad_norm": 0.06765329837799072,
+      "learning_rate": 1.232728745668122e-05,
+      "loss": 0.1992,
+      "step": 46637
+    },
+    {
+      "epoch": 3.7781918340894363,
+      "grad_norm": 0.07488850504159927,
+      "learning_rate": 1.2322786804086593e-05,
+      "loss": 0.2427,
+      "step": 46638
+    },
+    {
+      "epoch": 3.7782728451069345,
+      "grad_norm": 0.07653640955686569,
+      "learning_rate": 1.2318286151491967e-05,
+      "loss": 0.2297,
+      "step": 46639
+    },
+    {
+      "epoch": 3.7783538561244328,
+      "grad_norm": 0.07188255339860916,
+      "learning_rate": 1.231378549889734e-05,
+      "loss": 0.2335,
+      "step": 46640
+    },
+    {
+      "epoch": 3.7784348671419314,
+      "grad_norm": 0.07283948361873627,
+      "learning_rate": 1.2309284846302714e-05,
+      "loss": 0.2368,
+      "step": 46641
+    },
+    {
+      "epoch": 3.7785158781594297,
+      "grad_norm": 0.056375570595264435,
+      "learning_rate": 1.2304784193708087e-05,
+      "loss": 0.1894,
+      "step": 46642
+    },
+    {
+      "epoch": 3.778596889176928,
+      "grad_norm": 0.07126890867948532,
+      "learning_rate": 1.2300283541113463e-05,
+      "loss": 0.2171,
+      "step": 46643
+    },
+    {
+      "epoch": 3.7786779001944266,
+      "grad_norm": 0.07158375531435013,
+      "learning_rate": 1.2295782888518836e-05,
+      "loss": 0.1882,
+      "step": 46644
+    },
+    {
+      "epoch": 3.778758911211925,
+      "grad_norm": 0.072811059653759,
+      "learning_rate": 1.229128223592421e-05,
+      "loss": 0.2374,
+      "step": 46645
+    },
+    {
+      "epoch": 3.778839922229423,
+      "grad_norm": 0.06893569976091385,
+      "learning_rate": 1.2286781583329584e-05,
+      "loss": 0.2205,
+      "step": 46646
+    },
+    {
+      "epoch": 3.778920933246922,
+      "grad_norm": 0.07594628632068634,
+      "learning_rate": 1.2282280930734957e-05,
+      "loss": 0.2129,
+      "step": 46647
+    },
+    {
+      "epoch": 3.77900194426442,
+      "grad_norm": 0.06287289410829544,
+      "learning_rate": 1.227778027814033e-05,
+      "loss": 0.2281,
+      "step": 46648
+    },
+    {
+      "epoch": 3.7790829552819183,
+      "grad_norm": 0.07265198230743408,
+      "learning_rate": 1.2273279625545704e-05,
+      "loss": 0.2196,
+      "step": 46649
+    },
+    {
+      "epoch": 3.779163966299417,
+      "grad_norm": 0.05850711092352867,
+      "learning_rate": 1.2268778972951078e-05,
+      "loss": 0.2111,
+      "step": 46650
+    },
+    {
+      "epoch": 3.779244977316915,
+      "grad_norm": 0.06421723961830139,
+      "learning_rate": 1.2264278320356452e-05,
+      "loss": 0.2301,
+      "step": 46651
+    },
+    {
+      "epoch": 3.7793259883344135,
+      "grad_norm": 0.06834607571363449,
+      "learning_rate": 1.2259777667761827e-05,
+      "loss": 0.2464,
+      "step": 46652
+    },
+    {
+      "epoch": 3.7794069993519117,
+      "grad_norm": 0.06908293813467026,
+      "learning_rate": 1.2255277015167199e-05,
+      "loss": 0.1841,
+      "step": 46653
+    },
+    {
+      "epoch": 3.7794880103694104,
+      "grad_norm": 0.06474535912275314,
+      "learning_rate": 1.2250776362572574e-05,
+      "loss": 0.1876,
+      "step": 46654
+    },
+    {
+      "epoch": 3.7795690213869086,
+      "grad_norm": 0.07207901775836945,
+      "learning_rate": 1.2246275709977948e-05,
+      "loss": 0.2289,
+      "step": 46655
+    },
+    {
+      "epoch": 3.779650032404407,
+      "grad_norm": 0.05818149074912071,
+      "learning_rate": 1.2241775057383321e-05,
+      "loss": 0.1996,
+      "step": 46656
+    },
+    {
+      "epoch": 3.779731043421905,
+      "grad_norm": 0.0827532559633255,
+      "learning_rate": 1.2237274404788695e-05,
+      "loss": 0.2101,
+      "step": 46657
+    },
+    {
+      "epoch": 3.779812054439404,
+      "grad_norm": 0.09072255343198776,
+      "learning_rate": 1.2232773752194068e-05,
+      "loss": 0.256,
+      "step": 46658
+    },
+    {
+      "epoch": 3.779893065456902,
+      "grad_norm": 0.07674512267112732,
+      "learning_rate": 1.2228273099599442e-05,
+      "loss": 0.2505,
+      "step": 46659
+    },
+    {
+      "epoch": 3.7799740764744003,
+      "grad_norm": 0.060501959174871445,
+      "learning_rate": 1.2223772447004816e-05,
+      "loss": 0.2001,
+      "step": 46660
+    },
+    {
+      "epoch": 3.780055087491899,
+      "grad_norm": 0.06538248807191849,
+      "learning_rate": 1.2219271794410191e-05,
+      "loss": 0.2453,
+      "step": 46661
+    },
+    {
+      "epoch": 3.7801360985093972,
+      "grad_norm": 0.07560870796442032,
+      "learning_rate": 1.2214771141815563e-05,
+      "loss": 0.1887,
+      "step": 46662
+    },
+    {
+      "epoch": 3.7802171095268955,
+      "grad_norm": 0.08395927399396896,
+      "learning_rate": 1.2210270489220938e-05,
+      "loss": 0.1986,
+      "step": 46663
+    },
+    {
+      "epoch": 3.780298120544394,
+      "grad_norm": 0.05529021471738815,
+      "learning_rate": 1.2205769836626312e-05,
+      "loss": 0.1969,
+      "step": 46664
+    },
+    {
+      "epoch": 3.7803791315618924,
+      "grad_norm": 0.09511775523424149,
+      "learning_rate": 1.2201269184031685e-05,
+      "loss": 0.233,
+      "step": 46665
+    },
+    {
+      "epoch": 3.7804601425793907,
+      "grad_norm": 0.06025183945894241,
+      "learning_rate": 1.2196768531437059e-05,
+      "loss": 0.2091,
+      "step": 46666
+    },
+    {
+      "epoch": 3.7805411535968894,
+      "grad_norm": 0.07725583016872406,
+      "learning_rate": 1.2192267878842433e-05,
+      "loss": 0.2462,
+      "step": 46667
+    },
+    {
+      "epoch": 3.7806221646143876,
+      "grad_norm": 0.07531146705150604,
+      "learning_rate": 1.2187767226247806e-05,
+      "loss": 0.1891,
+      "step": 46668
+    },
+    {
+      "epoch": 3.780703175631886,
+      "grad_norm": 0.06599050760269165,
+      "learning_rate": 1.218326657365318e-05,
+      "loss": 0.2141,
+      "step": 46669
+    },
+    {
+      "epoch": 3.7807841866493845,
+      "grad_norm": 0.08106547594070435,
+      "learning_rate": 1.2178765921058555e-05,
+      "loss": 0.2459,
+      "step": 46670
+    },
+    {
+      "epoch": 3.780865197666883,
+      "grad_norm": 0.07727594673633575,
+      "learning_rate": 1.2174265268463927e-05,
+      "loss": 0.2672,
+      "step": 46671
+    },
+    {
+      "epoch": 3.780946208684381,
+      "grad_norm": 0.07024858891963959,
+      "learning_rate": 1.2169764615869302e-05,
+      "loss": 0.1984,
+      "step": 46672
+    },
+    {
+      "epoch": 3.7810272197018797,
+      "grad_norm": 0.05269385129213333,
+      "learning_rate": 1.2165263963274676e-05,
+      "loss": 0.1987,
+      "step": 46673
+    },
+    {
+      "epoch": 3.781108230719378,
+      "grad_norm": 0.07926183193922043,
+      "learning_rate": 1.216076331068005e-05,
+      "loss": 0.225,
+      "step": 46674
+    },
+    {
+      "epoch": 3.781189241736876,
+      "grad_norm": 0.08314008265733719,
+      "learning_rate": 1.2156262658085423e-05,
+      "loss": 0.2404,
+      "step": 46675
+    },
+    {
+      "epoch": 3.7812702527543745,
+      "grad_norm": 0.08210525661706924,
+      "learning_rate": 1.2151762005490797e-05,
+      "loss": 0.2526,
+      "step": 46676
+    },
+    {
+      "epoch": 3.781351263771873,
+      "grad_norm": 0.08011859655380249,
+      "learning_rate": 1.214726135289617e-05,
+      "loss": 0.2305,
+      "step": 46677
+    },
+    {
+      "epoch": 3.7814322747893714,
+      "grad_norm": 0.07893484830856323,
+      "learning_rate": 1.2142760700301546e-05,
+      "loss": 0.2348,
+      "step": 46678
+    },
+    {
+      "epoch": 3.7815132858068696,
+      "grad_norm": 0.06235817074775696,
+      "learning_rate": 1.2138260047706918e-05,
+      "loss": 0.1975,
+      "step": 46679
+    },
+    {
+      "epoch": 3.781594296824368,
+      "grad_norm": 0.0649525374174118,
+      "learning_rate": 1.2133759395112291e-05,
+      "loss": 0.2165,
+      "step": 46680
+    },
+    {
+      "epoch": 3.7816753078418666,
+      "grad_norm": 0.06788064539432526,
+      "learning_rate": 1.2129258742517666e-05,
+      "loss": 0.1885,
+      "step": 46681
+    },
+    {
+      "epoch": 3.781756318859365,
+      "grad_norm": 0.07292314618825912,
+      "learning_rate": 1.212475808992304e-05,
+      "loss": 0.2394,
+      "step": 46682
+    },
+    {
+      "epoch": 3.781837329876863,
+      "grad_norm": 0.06189664825797081,
+      "learning_rate": 1.2120257437328412e-05,
+      "loss": 0.2327,
+      "step": 46683
+    },
+    {
+      "epoch": 3.7819183408943617,
+      "grad_norm": 0.054467201232910156,
+      "learning_rate": 1.2115756784733787e-05,
+      "loss": 0.1965,
+      "step": 46684
+    },
+    {
+      "epoch": 3.78199935191186,
+      "grad_norm": 0.06842909753322601,
+      "learning_rate": 1.211125613213916e-05,
+      "loss": 0.2055,
+      "step": 46685
+    },
+    {
+      "epoch": 3.7820803629293582,
+      "grad_norm": 0.07629824429750443,
+      "learning_rate": 1.2106755479544534e-05,
+      "loss": 0.2288,
+      "step": 46686
+    },
+    {
+      "epoch": 3.782161373946857,
+      "grad_norm": 0.06854870170354843,
+      "learning_rate": 1.210225482694991e-05,
+      "loss": 0.2034,
+      "step": 46687
+    },
+    {
+      "epoch": 3.782242384964355,
+      "grad_norm": 0.08357995003461838,
+      "learning_rate": 1.2097754174355282e-05,
+      "loss": 0.2294,
+      "step": 46688
+    },
+    {
+      "epoch": 3.7823233959818534,
+      "grad_norm": 0.05101915821433067,
+      "learning_rate": 1.2093253521760655e-05,
+      "loss": 0.1895,
+      "step": 46689
+    },
+    {
+      "epoch": 3.782404406999352,
+      "grad_norm": 0.0792124941945076,
+      "learning_rate": 1.208875286916603e-05,
+      "loss": 0.216,
+      "step": 46690
+    },
+    {
+      "epoch": 3.7824854180168503,
+      "grad_norm": 0.07666429132223129,
+      "learning_rate": 1.2084252216571404e-05,
+      "loss": 0.2374,
+      "step": 46691
+    },
+    {
+      "epoch": 3.7825664290343486,
+      "grad_norm": 0.08312845975160599,
+      "learning_rate": 1.2079751563976776e-05,
+      "loss": 0.2384,
+      "step": 46692
+    },
+    {
+      "epoch": 3.7826474400518473,
+      "grad_norm": 0.08143067359924316,
+      "learning_rate": 1.2075250911382151e-05,
+      "loss": 0.2452,
+      "step": 46693
+    },
+    {
+      "epoch": 3.7827284510693455,
+      "grad_norm": 0.06815093010663986,
+      "learning_rate": 1.2070750258787525e-05,
+      "loss": 0.2508,
+      "step": 46694
+    },
+    {
+      "epoch": 3.7828094620868438,
+      "grad_norm": 0.07493395358324051,
+      "learning_rate": 1.2066249606192899e-05,
+      "loss": 0.2458,
+      "step": 46695
+    },
+    {
+      "epoch": 3.7828904731043425,
+      "grad_norm": 0.079828180372715,
+      "learning_rate": 1.2061748953598272e-05,
+      "loss": 0.2392,
+      "step": 46696
+    },
+    {
+      "epoch": 3.7829714841218407,
+      "grad_norm": 0.0668768361210823,
+      "learning_rate": 1.2057248301003646e-05,
+      "loss": 0.2474,
+      "step": 46697
+    },
+    {
+      "epoch": 3.783052495139339,
+      "grad_norm": 0.06913848966360092,
+      "learning_rate": 1.205274764840902e-05,
+      "loss": 0.2297,
+      "step": 46698
+    },
+    {
+      "epoch": 3.783133506156837,
+      "grad_norm": 0.07236789911985397,
+      "learning_rate": 1.2048246995814395e-05,
+      "loss": 0.2377,
+      "step": 46699
+    },
+    {
+      "epoch": 3.7832145171743354,
+      "grad_norm": 0.06536900997161865,
+      "learning_rate": 1.2043746343219768e-05,
+      "loss": 0.2262,
+      "step": 46700
+    },
+    {
+      "epoch": 3.783295528191834,
+      "grad_norm": 0.08164361864328384,
+      "learning_rate": 1.203924569062514e-05,
+      "loss": 0.2633,
+      "step": 46701
+    },
+    {
+      "epoch": 3.7833765392093324,
+      "grad_norm": 0.0710197389125824,
+      "learning_rate": 1.2034745038030515e-05,
+      "loss": 0.2106,
+      "step": 46702
+    },
+    {
+      "epoch": 3.7834575502268306,
+      "grad_norm": 0.08425655961036682,
+      "learning_rate": 1.2030244385435889e-05,
+      "loss": 0.2288,
+      "step": 46703
+    },
+    {
+      "epoch": 3.7835385612443293,
+      "grad_norm": 0.06478645652532578,
+      "learning_rate": 1.2025743732841263e-05,
+      "loss": 0.2321,
+      "step": 46704
+    },
+    {
+      "epoch": 3.7836195722618275,
+      "grad_norm": 0.06343191862106323,
+      "learning_rate": 1.2021243080246636e-05,
+      "loss": 0.2063,
+      "step": 46705
+    },
+    {
+      "epoch": 3.783700583279326,
+      "grad_norm": 0.08273705095052719,
+      "learning_rate": 1.201674242765201e-05,
+      "loss": 0.2683,
+      "step": 46706
+    },
+    {
+      "epoch": 3.7837815942968245,
+      "grad_norm": 0.05265715345740318,
+      "learning_rate": 1.2012241775057383e-05,
+      "loss": 0.2125,
+      "step": 46707
+    },
+    {
+      "epoch": 3.7838626053143227,
+      "grad_norm": 0.06474486738443375,
+      "learning_rate": 1.2007741122462759e-05,
+      "loss": 0.1809,
+      "step": 46708
+    },
+    {
+      "epoch": 3.783943616331821,
+      "grad_norm": 0.07659123092889786,
+      "learning_rate": 1.200324046986813e-05,
+      "loss": 0.2185,
+      "step": 46709
+    },
+    {
+      "epoch": 3.7840246273493197,
+      "grad_norm": 0.0784589946269989,
+      "learning_rate": 1.1998739817273504e-05,
+      "loss": 0.2195,
+      "step": 46710
+    },
+    {
+      "epoch": 3.784105638366818,
+      "grad_norm": 0.0916212946176529,
+      "learning_rate": 1.199423916467888e-05,
+      "loss": 0.2416,
+      "step": 46711
+    },
+    {
+      "epoch": 3.784186649384316,
+      "grad_norm": 0.0649249330163002,
+      "learning_rate": 1.1989738512084253e-05,
+      "loss": 0.2101,
+      "step": 46712
+    },
+    {
+      "epoch": 3.784267660401815,
+      "grad_norm": 0.0661170706152916,
+      "learning_rate": 1.1985237859489627e-05,
+      "loss": 0.2285,
+      "step": 46713
+    },
+    {
+      "epoch": 3.784348671419313,
+      "grad_norm": 0.08094004541635513,
+      "learning_rate": 1.1980737206895e-05,
+      "loss": 0.2624,
+      "step": 46714
+    },
+    {
+      "epoch": 3.7844296824368113,
+      "grad_norm": 0.06313817203044891,
+      "learning_rate": 1.1976236554300374e-05,
+      "loss": 0.1953,
+      "step": 46715
+    },
+    {
+      "epoch": 3.78451069345431,
+      "grad_norm": 0.07197370380163193,
+      "learning_rate": 1.1971735901705748e-05,
+      "loss": 0.2086,
+      "step": 46716
+    },
+    {
+      "epoch": 3.7845917044718083,
+      "grad_norm": 0.060122050344944,
+      "learning_rate": 1.1967235249111123e-05,
+      "loss": 0.2372,
+      "step": 46717
+    },
+    {
+      "epoch": 3.7846727154893065,
+      "grad_norm": 0.0652763694524765,
+      "learning_rate": 1.1962734596516495e-05,
+      "loss": 0.2288,
+      "step": 46718
+    },
+    {
+      "epoch": 3.784753726506805,
+      "grad_norm": 0.07435501366853714,
+      "learning_rate": 1.1958233943921868e-05,
+      "loss": 0.2219,
+      "step": 46719
+    },
+    {
+      "epoch": 3.7848347375243034,
+      "grad_norm": 0.06599126011133194,
+      "learning_rate": 1.1953733291327244e-05,
+      "loss": 0.221,
+      "step": 46720
+    },
+    {
+      "epoch": 3.7849157485418017,
+      "grad_norm": 0.08771032840013504,
+      "learning_rate": 1.1949232638732617e-05,
+      "loss": 0.2255,
+      "step": 46721
+    },
+    {
+      "epoch": 3.7849967595593,
+      "grad_norm": 0.08068722486495972,
+      "learning_rate": 1.194473198613799e-05,
+      "loss": 0.24,
+      "step": 46722
+    },
+    {
+      "epoch": 3.785077770576798,
+      "grad_norm": 0.06992881000041962,
+      "learning_rate": 1.1940231333543365e-05,
+      "loss": 0.2148,
+      "step": 46723
+    },
+    {
+      "epoch": 3.785158781594297,
+      "grad_norm": 0.06782852113246918,
+      "learning_rate": 1.1935730680948738e-05,
+      "loss": 0.2005,
+      "step": 46724
+    },
+    {
+      "epoch": 3.785239792611795,
+      "grad_norm": 0.07253803312778473,
+      "learning_rate": 1.1931230028354112e-05,
+      "loss": 0.2628,
+      "step": 46725
+    },
+    {
+      "epoch": 3.7853208036292934,
+      "grad_norm": 0.05372246354818344,
+      "learning_rate": 1.1926729375759485e-05,
+      "loss": 0.1846,
+      "step": 46726
+    },
+    {
+      "epoch": 3.785401814646792,
+      "grad_norm": 0.0695682018995285,
+      "learning_rate": 1.1922228723164859e-05,
+      "loss": 0.2172,
+      "step": 46727
+    },
+    {
+      "epoch": 3.7854828256642903,
+      "grad_norm": 0.07446936517953873,
+      "learning_rate": 1.1917728070570233e-05,
+      "loss": 0.2159,
+      "step": 46728
+    },
+    {
+      "epoch": 3.7855638366817885,
+      "grad_norm": 0.07652425020933151,
+      "learning_rate": 1.1913227417975608e-05,
+      "loss": 0.2145,
+      "step": 46729
+    },
+    {
+      "epoch": 3.785644847699287,
+      "grad_norm": 0.07526133209466934,
+      "learning_rate": 1.1908726765380981e-05,
+      "loss": 0.2344,
+      "step": 46730
+    },
+    {
+      "epoch": 3.7857258587167855,
+      "grad_norm": 0.06465907394886017,
+      "learning_rate": 1.1904226112786353e-05,
+      "loss": 0.2097,
+      "step": 46731
+    },
+    {
+      "epoch": 3.7858068697342837,
+      "grad_norm": 0.055496908724308014,
+      "learning_rate": 1.1899725460191729e-05,
+      "loss": 0.204,
+      "step": 46732
+    },
+    {
+      "epoch": 3.7858878807517824,
+      "grad_norm": 0.07140771299600601,
+      "learning_rate": 1.1895224807597102e-05,
+      "loss": 0.1999,
+      "step": 46733
+    },
+    {
+      "epoch": 3.7859688917692806,
+      "grad_norm": 0.07559964805841446,
+      "learning_rate": 1.1890724155002476e-05,
+      "loss": 0.2583,
+      "step": 46734
+    },
+    {
+      "epoch": 3.786049902786779,
+      "grad_norm": 0.0660320520401001,
+      "learning_rate": 1.188622350240785e-05,
+      "loss": 0.231,
+      "step": 46735
+    },
+    {
+      "epoch": 3.7861309138042776,
+      "grad_norm": 0.06079673767089844,
+      "learning_rate": 1.1881722849813223e-05,
+      "loss": 0.2112,
+      "step": 46736
+    },
+    {
+      "epoch": 3.786211924821776,
+      "grad_norm": 0.07198115438222885,
+      "learning_rate": 1.1877222197218597e-05,
+      "loss": 0.1936,
+      "step": 46737
+    },
+    {
+      "epoch": 3.786292935839274,
+      "grad_norm": 0.07653490453958511,
+      "learning_rate": 1.1872721544623972e-05,
+      "loss": 0.2059,
+      "step": 46738
+    },
+    {
+      "epoch": 3.7863739468567728,
+      "grad_norm": 0.07065163552761078,
+      "learning_rate": 1.1868220892029344e-05,
+      "loss": 0.197,
+      "step": 46739
+    },
+    {
+      "epoch": 3.786454957874271,
+      "grad_norm": 0.08107943832874298,
+      "learning_rate": 1.1863720239434717e-05,
+      "loss": 0.2431,
+      "step": 46740
+    },
+    {
+      "epoch": 3.7865359688917692,
+      "grad_norm": 0.056850139051675797,
+      "learning_rate": 1.1859219586840093e-05,
+      "loss": 0.2156,
+      "step": 46741
+    },
+    {
+      "epoch": 3.786616979909268,
+      "grad_norm": 0.10843037068843842,
+      "learning_rate": 1.1854718934245466e-05,
+      "loss": 0.2501,
+      "step": 46742
+    },
+    {
+      "epoch": 3.786697990926766,
+      "grad_norm": 0.0759415253996849,
+      "learning_rate": 1.185021828165084e-05,
+      "loss": 0.2182,
+      "step": 46743
+    },
+    {
+      "epoch": 3.7867790019442644,
+      "grad_norm": 0.07525965571403503,
+      "learning_rate": 1.1845717629056214e-05,
+      "loss": 0.237,
+      "step": 46744
+    },
+    {
+      "epoch": 3.7868600129617627,
+      "grad_norm": 0.07234998047351837,
+      "learning_rate": 1.1841216976461587e-05,
+      "loss": 0.2268,
+      "step": 46745
+    },
+    {
+      "epoch": 3.786941023979261,
+      "grad_norm": 0.05831919237971306,
+      "learning_rate": 1.183671632386696e-05,
+      "loss": 0.2266,
+      "step": 46746
+    },
+    {
+      "epoch": 3.7870220349967596,
+      "grad_norm": 0.07513531297445297,
+      "learning_rate": 1.1832215671272336e-05,
+      "loss": 0.2511,
+      "step": 46747
+    },
+    {
+      "epoch": 3.787103046014258,
+      "grad_norm": 0.06514651328325272,
+      "learning_rate": 1.1827715018677708e-05,
+      "loss": 0.1951,
+      "step": 46748
+    },
+    {
+      "epoch": 3.787184057031756,
+      "grad_norm": 0.07075642794370651,
+      "learning_rate": 1.1823214366083082e-05,
+      "loss": 0.2222,
+      "step": 46749
+    },
+    {
+      "epoch": 3.787265068049255,
+      "grad_norm": 0.09682905673980713,
+      "learning_rate": 1.1818713713488457e-05,
+      "loss": 0.2183,
+      "step": 46750
+    },
+    {
+      "epoch": 3.787346079066753,
+      "grad_norm": 0.07281651347875595,
+      "learning_rate": 1.181421306089383e-05,
+      "loss": 0.2481,
+      "step": 46751
+    },
+    {
+      "epoch": 3.7874270900842513,
+      "grad_norm": 0.055947430431842804,
+      "learning_rate": 1.1809712408299204e-05,
+      "loss": 0.2086,
+      "step": 46752
+    },
+    {
+      "epoch": 3.78750810110175,
+      "grad_norm": 0.07676409929990768,
+      "learning_rate": 1.1805211755704578e-05,
+      "loss": 0.2487,
+      "step": 46753
+    },
+    {
+      "epoch": 3.787589112119248,
+      "grad_norm": 0.07692111283540726,
+      "learning_rate": 1.1800711103109951e-05,
+      "loss": 0.2786,
+      "step": 46754
+    },
+    {
+      "epoch": 3.7876701231367464,
+      "grad_norm": 0.06674608588218689,
+      "learning_rate": 1.1796210450515325e-05,
+      "loss": 0.2249,
+      "step": 46755
+    },
+    {
+      "epoch": 3.787751134154245,
+      "grad_norm": 0.06333047896623611,
+      "learning_rate": 1.17917097979207e-05,
+      "loss": 0.2187,
+      "step": 46756
+    },
+    {
+      "epoch": 3.7878321451717434,
+      "grad_norm": 0.07365437597036362,
+      "learning_rate": 1.1787209145326072e-05,
+      "loss": 0.2089,
+      "step": 46757
+    },
+    {
+      "epoch": 3.7879131561892416,
+      "grad_norm": 0.09069455415010452,
+      "learning_rate": 1.1782708492731446e-05,
+      "loss": 0.2148,
+      "step": 46758
+    },
+    {
+      "epoch": 3.7879941672067403,
+      "grad_norm": 0.07776268571615219,
+      "learning_rate": 1.1778207840136821e-05,
+      "loss": 0.2311,
+      "step": 46759
+    },
+    {
+      "epoch": 3.7880751782242386,
+      "grad_norm": 0.08250845968723297,
+      "learning_rate": 1.1773707187542195e-05,
+      "loss": 0.2531,
+      "step": 46760
+    },
+    {
+      "epoch": 3.788156189241737,
+      "grad_norm": 0.08193209767341614,
+      "learning_rate": 1.1769206534947568e-05,
+      "loss": 0.2188,
+      "step": 46761
+    },
+    {
+      "epoch": 3.7882372002592355,
+      "grad_norm": 0.07312662899494171,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.2247,
+      "step": 46762
+    },
+    {
+      "epoch": 3.7883182112767337,
+      "grad_norm": 0.07228390127420425,
+      "learning_rate": 1.1760205229758315e-05,
+      "loss": 0.2473,
+      "step": 46763
+    },
+    {
+      "epoch": 3.788399222294232,
+      "grad_norm": 0.0762069895863533,
+      "learning_rate": 1.1755704577163689e-05,
+      "loss": 0.206,
+      "step": 46764
+    },
+    {
+      "epoch": 3.7884802333117307,
+      "grad_norm": 0.06151983514428139,
+      "learning_rate": 1.1751203924569063e-05,
+      "loss": 0.1886,
+      "step": 46765
+    },
+    {
+      "epoch": 3.788561244329229,
+      "grad_norm": 0.07053109258413315,
+      "learning_rate": 1.1746703271974436e-05,
+      "loss": 0.227,
+      "step": 46766
+    },
+    {
+      "epoch": 3.788642255346727,
+      "grad_norm": 0.06008979305624962,
+      "learning_rate": 1.174220261937981e-05,
+      "loss": 0.1776,
+      "step": 46767
+    },
+    {
+      "epoch": 3.7887232663642254,
+      "grad_norm": 0.06683298200368881,
+      "learning_rate": 1.1737701966785185e-05,
+      "loss": 0.2484,
+      "step": 46768
+    },
+    {
+      "epoch": 3.7888042773817237,
+      "grad_norm": 0.07074110209941864,
+      "learning_rate": 1.1733201314190557e-05,
+      "loss": 0.1979,
+      "step": 46769
+    },
+    {
+      "epoch": 3.7888852883992223,
+      "grad_norm": 0.07903765887022018,
+      "learning_rate": 1.1728700661595932e-05,
+      "loss": 0.2239,
+      "step": 46770
+    },
+    {
+      "epoch": 3.7889662994167206,
+      "grad_norm": 0.08197832852602005,
+      "learning_rate": 1.1724200009001306e-05,
+      "loss": 0.2449,
+      "step": 46771
+    },
+    {
+      "epoch": 3.789047310434219,
+      "grad_norm": 0.07146139442920685,
+      "learning_rate": 1.171969935640668e-05,
+      "loss": 0.2572,
+      "step": 46772
+    },
+    {
+      "epoch": 3.7891283214517175,
+      "grad_norm": 0.061843667179346085,
+      "learning_rate": 1.1715198703812053e-05,
+      "loss": 0.2238,
+      "step": 46773
+    },
+    {
+      "epoch": 3.7892093324692158,
+      "grad_norm": 0.07453931123018265,
+      "learning_rate": 1.1710698051217427e-05,
+      "loss": 0.1794,
+      "step": 46774
+    },
+    {
+      "epoch": 3.789290343486714,
+      "grad_norm": 0.06958021223545074,
+      "learning_rate": 1.17061973986228e-05,
+      "loss": 0.2162,
+      "step": 46775
+    },
+    {
+      "epoch": 3.7893713545042127,
+      "grad_norm": 0.09496792405843735,
+      "learning_rate": 1.1701696746028176e-05,
+      "loss": 0.2395,
+      "step": 46776
+    },
+    {
+      "epoch": 3.789452365521711,
+      "grad_norm": 0.06393705308437347,
+      "learning_rate": 1.169719609343355e-05,
+      "loss": 0.2149,
+      "step": 46777
+    },
+    {
+      "epoch": 3.789533376539209,
+      "grad_norm": 0.07385969907045364,
+      "learning_rate": 1.1692695440838921e-05,
+      "loss": 0.2408,
+      "step": 46778
+    },
+    {
+      "epoch": 3.789614387556708,
+      "grad_norm": 0.06838518381118774,
+      "learning_rate": 1.1688194788244296e-05,
+      "loss": 0.2357,
+      "step": 46779
+    },
+    {
+      "epoch": 3.789695398574206,
+      "grad_norm": 0.07647363841533661,
+      "learning_rate": 1.168369413564967e-05,
+      "loss": 0.2594,
+      "step": 46780
+    },
+    {
+      "epoch": 3.7897764095917044,
+      "grad_norm": 0.0688256323337555,
+      "learning_rate": 1.1679193483055044e-05,
+      "loss": 0.2013,
+      "step": 46781
+    },
+    {
+      "epoch": 3.789857420609203,
+      "grad_norm": 0.07052019238471985,
+      "learning_rate": 1.1674692830460417e-05,
+      "loss": 0.2156,
+      "step": 46782
+    },
+    {
+      "epoch": 3.7899384316267013,
+      "grad_norm": 0.07364077866077423,
+      "learning_rate": 1.1670192177865791e-05,
+      "loss": 0.2067,
+      "step": 46783
+    },
+    {
+      "epoch": 3.7900194426441995,
+      "grad_norm": 0.0811486765742302,
+      "learning_rate": 1.1665691525271165e-05,
+      "loss": 0.2418,
+      "step": 46784
+    },
+    {
+      "epoch": 3.7901004536616982,
+      "grad_norm": 0.08562948554754257,
+      "learning_rate": 1.166119087267654e-05,
+      "loss": 0.2343,
+      "step": 46785
+    },
+    {
+      "epoch": 3.7901814646791965,
+      "grad_norm": 0.08794140070676804,
+      "learning_rate": 1.1656690220081913e-05,
+      "loss": 0.2201,
+      "step": 46786
+    },
+    {
+      "epoch": 3.7902624756966947,
+      "grad_norm": 0.07163339853286743,
+      "learning_rate": 1.1652189567487285e-05,
+      "loss": 0.2319,
+      "step": 46787
+    },
+    {
+      "epoch": 3.790343486714193,
+      "grad_norm": 0.06453966349363327,
+      "learning_rate": 1.164768891489266e-05,
+      "loss": 0.2019,
+      "step": 46788
+    },
+    {
+      "epoch": 3.7904244977316917,
+      "grad_norm": 0.0652361512184143,
+      "learning_rate": 1.1643188262298034e-05,
+      "loss": 0.2053,
+      "step": 46789
+    },
+    {
+      "epoch": 3.79050550874919,
+      "grad_norm": 0.06884594261646271,
+      "learning_rate": 1.1638687609703408e-05,
+      "loss": 0.2362,
+      "step": 46790
+    },
+    {
+      "epoch": 3.790586519766688,
+      "grad_norm": 0.07537364959716797,
+      "learning_rate": 1.1634186957108781e-05,
+      "loss": 0.2326,
+      "step": 46791
+    },
+    {
+      "epoch": 3.7906675307841864,
+      "grad_norm": 0.06967596709728241,
+      "learning_rate": 1.1629686304514155e-05,
+      "loss": 0.1997,
+      "step": 46792
+    },
+    {
+      "epoch": 3.790748541801685,
+      "grad_norm": 0.08205866068601608,
+      "learning_rate": 1.1625185651919529e-05,
+      "loss": 0.2289,
+      "step": 46793
+    },
+    {
+      "epoch": 3.7908295528191833,
+      "grad_norm": 0.08043761551380157,
+      "learning_rate": 1.1620684999324904e-05,
+      "loss": 0.2037,
+      "step": 46794
+    },
+    {
+      "epoch": 3.7909105638366816,
+      "grad_norm": 0.07155504822731018,
+      "learning_rate": 1.1616184346730276e-05,
+      "loss": 0.2216,
+      "step": 46795
+    },
+    {
+      "epoch": 3.7909915748541803,
+      "grad_norm": 0.07402777671813965,
+      "learning_rate": 1.161168369413565e-05,
+      "loss": 0.237,
+      "step": 46796
+    },
+    {
+      "epoch": 3.7910725858716785,
+      "grad_norm": 0.07951734960079193,
+      "learning_rate": 1.1607183041541025e-05,
+      "loss": 0.2707,
+      "step": 46797
+    },
+    {
+      "epoch": 3.7911535968891767,
+      "grad_norm": 0.07128282636404037,
+      "learning_rate": 1.1602682388946398e-05,
+      "loss": 0.2531,
+      "step": 46798
+    },
+    {
+      "epoch": 3.7912346079066754,
+      "grad_norm": 0.07474765181541443,
+      "learning_rate": 1.1598181736351772e-05,
+      "loss": 0.2073,
+      "step": 46799
+    },
+    {
+      "epoch": 3.7913156189241737,
+      "grad_norm": 0.06123476102948189,
+      "learning_rate": 1.1593681083757146e-05,
+      "loss": 0.2372,
+      "step": 46800
+    },
+    {
+      "epoch": 3.791396629941672,
+      "grad_norm": 0.08566850423812866,
+      "learning_rate": 1.1589180431162519e-05,
+      "loss": 0.2129,
+      "step": 46801
+    },
+    {
+      "epoch": 3.7914776409591706,
+      "grad_norm": 0.0908622220158577,
+      "learning_rate": 1.1584679778567893e-05,
+      "loss": 0.2277,
+      "step": 46802
+    },
+    {
+      "epoch": 3.791558651976669,
+      "grad_norm": 0.05587979778647423,
+      "learning_rate": 1.1580179125973268e-05,
+      "loss": 0.2102,
+      "step": 46803
+    },
+    {
+      "epoch": 3.791639662994167,
+      "grad_norm": 0.08635761588811874,
+      "learning_rate": 1.157567847337864e-05,
+      "loss": 0.208,
+      "step": 46804
+    },
+    {
+      "epoch": 3.791720674011666,
+      "grad_norm": 0.07278425991535187,
+      "learning_rate": 1.1571177820784014e-05,
+      "loss": 0.2413,
+      "step": 46805
+    },
+    {
+      "epoch": 3.791801685029164,
+      "grad_norm": 0.08281299471855164,
+      "learning_rate": 1.1566677168189389e-05,
+      "loss": 0.2106,
+      "step": 46806
+    },
+    {
+      "epoch": 3.7918826960466623,
+      "grad_norm": 0.07331330329179764,
+      "learning_rate": 1.1562176515594762e-05,
+      "loss": 0.2529,
+      "step": 46807
+    },
+    {
+      "epoch": 3.791963707064161,
+      "grad_norm": 0.07193376123905182,
+      "learning_rate": 1.1557675863000134e-05,
+      "loss": 0.2057,
+      "step": 46808
+    },
+    {
+      "epoch": 3.792044718081659,
+      "grad_norm": 0.0806722342967987,
+      "learning_rate": 1.155317521040551e-05,
+      "loss": 0.2446,
+      "step": 46809
+    },
+    {
+      "epoch": 3.7921257290991575,
+      "grad_norm": 0.0638408288359642,
+      "learning_rate": 1.1548674557810883e-05,
+      "loss": 0.24,
+      "step": 46810
+    },
+    {
+      "epoch": 3.7922067401166557,
+      "grad_norm": 0.07036413997411728,
+      "learning_rate": 1.1544173905216257e-05,
+      "loss": 0.2169,
+      "step": 46811
+    },
+    {
+      "epoch": 3.7922877511341544,
+      "grad_norm": 0.06902635842561722,
+      "learning_rate": 1.153967325262163e-05,
+      "loss": 0.2231,
+      "step": 46812
+    },
+    {
+      "epoch": 3.7923687621516526,
+      "grad_norm": 0.06593969464302063,
+      "learning_rate": 1.1535172600027004e-05,
+      "loss": 0.2089,
+      "step": 46813
+    },
+    {
+      "epoch": 3.792449773169151,
+      "grad_norm": 0.07559620589017868,
+      "learning_rate": 1.1530671947432378e-05,
+      "loss": 0.2276,
+      "step": 46814
+    },
+    {
+      "epoch": 3.792530784186649,
+      "grad_norm": 0.06421814113855362,
+      "learning_rate": 1.1526171294837753e-05,
+      "loss": 0.2729,
+      "step": 46815
+    },
+    {
+      "epoch": 3.792611795204148,
+      "grad_norm": 0.06696971505880356,
+      "learning_rate": 1.1521670642243127e-05,
+      "loss": 0.2587,
+      "step": 46816
+    },
+    {
+      "epoch": 3.792692806221646,
+      "grad_norm": 0.08010052144527435,
+      "learning_rate": 1.1517169989648498e-05,
+      "loss": 0.1825,
+      "step": 46817
+    },
+    {
+      "epoch": 3.7927738172391443,
+      "grad_norm": 0.08656228333711624,
+      "learning_rate": 1.1512669337053874e-05,
+      "loss": 0.2341,
+      "step": 46818
+    },
+    {
+      "epoch": 3.792854828256643,
+      "grad_norm": 0.07407809793949127,
+      "learning_rate": 1.1508168684459247e-05,
+      "loss": 0.2599,
+      "step": 46819
+    },
+    {
+      "epoch": 3.7929358392741412,
+      "grad_norm": 0.07218343764543533,
+      "learning_rate": 1.1503668031864621e-05,
+      "loss": 0.2203,
+      "step": 46820
+    },
+    {
+      "epoch": 3.7930168502916395,
+      "grad_norm": 0.07398039847612381,
+      "learning_rate": 1.1499167379269995e-05,
+      "loss": 0.2177,
+      "step": 46821
+    },
+    {
+      "epoch": 3.793097861309138,
+      "grad_norm": 0.06317486613988876,
+      "learning_rate": 1.1494666726675368e-05,
+      "loss": 0.2141,
+      "step": 46822
+    },
+    {
+      "epoch": 3.7931788723266364,
+      "grad_norm": 0.07027360796928406,
+      "learning_rate": 1.1490166074080742e-05,
+      "loss": 0.2265,
+      "step": 46823
+    },
+    {
+      "epoch": 3.7932598833441347,
+      "grad_norm": 0.07729087769985199,
+      "learning_rate": 1.1485665421486117e-05,
+      "loss": 0.2256,
+      "step": 46824
+    },
+    {
+      "epoch": 3.7933408943616334,
+      "grad_norm": 0.07687903195619583,
+      "learning_rate": 1.1481164768891489e-05,
+      "loss": 0.2307,
+      "step": 46825
+    },
+    {
+      "epoch": 3.7934219053791316,
+      "grad_norm": 0.07208074629306793,
+      "learning_rate": 1.1476664116296863e-05,
+      "loss": 0.2155,
+      "step": 46826
+    },
+    {
+      "epoch": 3.79350291639663,
+      "grad_norm": 0.07030342519283295,
+      "learning_rate": 1.1472163463702238e-05,
+      "loss": 0.2055,
+      "step": 46827
+    },
+    {
+      "epoch": 3.7935839274141285,
+      "grad_norm": 0.06092243641614914,
+      "learning_rate": 1.1467662811107612e-05,
+      "loss": 0.2253,
+      "step": 46828
+    },
+    {
+      "epoch": 3.7936649384316268,
+      "grad_norm": 0.08080343902111053,
+      "learning_rate": 1.1463162158512985e-05,
+      "loss": 0.2315,
+      "step": 46829
+    },
+    {
+      "epoch": 3.793745949449125,
+      "grad_norm": 0.06889671087265015,
+      "learning_rate": 1.1458661505918359e-05,
+      "loss": 0.2071,
+      "step": 46830
+    },
+    {
+      "epoch": 3.7938269604666237,
+      "grad_norm": 0.061423543840646744,
+      "learning_rate": 1.1454160853323732e-05,
+      "loss": 0.1851,
+      "step": 46831
+    },
+    {
+      "epoch": 3.793907971484122,
+      "grad_norm": 0.06487643718719482,
+      "learning_rate": 1.1449660200729106e-05,
+      "loss": 0.225,
+      "step": 46832
+    },
+    {
+      "epoch": 3.79398898250162,
+      "grad_norm": 0.0830056294798851,
+      "learning_rate": 1.1445159548134481e-05,
+      "loss": 0.217,
+      "step": 46833
+    },
+    {
+      "epoch": 3.7940699935191184,
+      "grad_norm": 0.06530779600143433,
+      "learning_rate": 1.1440658895539853e-05,
+      "loss": 0.2175,
+      "step": 46834
+    },
+    {
+      "epoch": 3.794151004536617,
+      "grad_norm": 0.07887805998325348,
+      "learning_rate": 1.1436158242945227e-05,
+      "loss": 0.2445,
+      "step": 46835
+    },
+    {
+      "epoch": 3.7942320155541154,
+      "grad_norm": 0.06664882600307465,
+      "learning_rate": 1.1431657590350602e-05,
+      "loss": 0.2433,
+      "step": 46836
+    },
+    {
+      "epoch": 3.7943130265716136,
+      "grad_norm": 0.06435833871364594,
+      "learning_rate": 1.1427156937755976e-05,
+      "loss": 0.2001,
+      "step": 46837
+    },
+    {
+      "epoch": 3.794394037589112,
+      "grad_norm": 0.062327224761247635,
+      "learning_rate": 1.1422656285161348e-05,
+      "loss": 0.2213,
+      "step": 46838
+    },
+    {
+      "epoch": 3.7944750486066106,
+      "grad_norm": 0.07332039624452591,
+      "learning_rate": 1.1418155632566723e-05,
+      "loss": 0.1763,
+      "step": 46839
+    },
+    {
+      "epoch": 3.794556059624109,
+      "grad_norm": 0.09450053423643112,
+      "learning_rate": 1.1413654979972096e-05,
+      "loss": 0.2553,
+      "step": 46840
+    },
+    {
+      "epoch": 3.794637070641607,
+      "grad_norm": 0.06693083792924881,
+      "learning_rate": 1.140915432737747e-05,
+      "loss": 0.1962,
+      "step": 46841
+    },
+    {
+      "epoch": 3.7947180816591057,
+      "grad_norm": 0.07547756284475327,
+      "learning_rate": 1.1404653674782845e-05,
+      "loss": 0.217,
+      "step": 46842
+    },
+    {
+      "epoch": 3.794799092676604,
+      "grad_norm": 0.06735184788703918,
+      "learning_rate": 1.1400153022188217e-05,
+      "loss": 0.2216,
+      "step": 46843
+    },
+    {
+      "epoch": 3.7948801036941022,
+      "grad_norm": 0.06307090818881989,
+      "learning_rate": 1.1395652369593591e-05,
+      "loss": 0.2395,
+      "step": 46844
+    },
+    {
+      "epoch": 3.794961114711601,
+      "grad_norm": 0.06501543521881104,
+      "learning_rate": 1.1391151716998966e-05,
+      "loss": 0.2125,
+      "step": 46845
+    },
+    {
+      "epoch": 3.795042125729099,
+      "grad_norm": 0.09293830394744873,
+      "learning_rate": 1.138665106440434e-05,
+      "loss": 0.2492,
+      "step": 46846
+    },
+    {
+      "epoch": 3.7951231367465974,
+      "grad_norm": 0.06201693415641785,
+      "learning_rate": 1.1382150411809712e-05,
+      "loss": 0.2137,
+      "step": 46847
+    },
+    {
+      "epoch": 3.795204147764096,
+      "grad_norm": 0.08595079183578491,
+      "learning_rate": 1.1377649759215087e-05,
+      "loss": 0.2071,
+      "step": 46848
+    },
+    {
+      "epoch": 3.7952851587815943,
+      "grad_norm": 0.07711045444011688,
+      "learning_rate": 1.137314910662046e-05,
+      "loss": 0.2485,
+      "step": 46849
+    },
+    {
+      "epoch": 3.7953661697990926,
+      "grad_norm": 0.08922844380140305,
+      "learning_rate": 1.1368648454025834e-05,
+      "loss": 0.23,
+      "step": 46850
+    },
+    {
+      "epoch": 3.7954471808165913,
+      "grad_norm": 0.07273554801940918,
+      "learning_rate": 1.1364147801431208e-05,
+      "loss": 0.2015,
+      "step": 46851
+    },
+    {
+      "epoch": 3.7955281918340895,
+      "grad_norm": 0.06916754692792892,
+      "learning_rate": 1.1359647148836581e-05,
+      "loss": 0.2357,
+      "step": 46852
+    },
+    {
+      "epoch": 3.7956092028515878,
+      "grad_norm": 0.08604365587234497,
+      "learning_rate": 1.1355146496241955e-05,
+      "loss": 0.2403,
+      "step": 46853
+    },
+    {
+      "epoch": 3.7956902138690864,
+      "grad_norm": 0.07010132074356079,
+      "learning_rate": 1.135064584364733e-05,
+      "loss": 0.2159,
+      "step": 46854
+    },
+    {
+      "epoch": 3.7957712248865847,
+      "grad_norm": 0.08245091885328293,
+      "learning_rate": 1.1346145191052702e-05,
+      "loss": 0.2221,
+      "step": 46855
+    },
+    {
+      "epoch": 3.795852235904083,
+      "grad_norm": 0.07700005918741226,
+      "learning_rate": 1.1341644538458076e-05,
+      "loss": 0.2152,
+      "step": 46856
+    },
+    {
+      "epoch": 3.795933246921581,
+      "grad_norm": 0.08508516103029251,
+      "learning_rate": 1.1337143885863451e-05,
+      "loss": 0.2322,
+      "step": 46857
+    },
+    {
+      "epoch": 3.79601425793908,
+      "grad_norm": 0.08205549418926239,
+      "learning_rate": 1.1332643233268825e-05,
+      "loss": 0.1954,
+      "step": 46858
+    },
+    {
+      "epoch": 3.796095268956578,
+      "grad_norm": 0.06523613631725311,
+      "learning_rate": 1.1328142580674198e-05,
+      "loss": 0.2025,
+      "step": 46859
+    },
+    {
+      "epoch": 3.7961762799740764,
+      "grad_norm": 0.08023390918970108,
+      "learning_rate": 1.1323641928079572e-05,
+      "loss": 0.2702,
+      "step": 46860
+    },
+    {
+      "epoch": 3.7962572909915746,
+      "grad_norm": 0.08578156679868698,
+      "learning_rate": 1.1319141275484946e-05,
+      "loss": 0.2349,
+      "step": 46861
+    },
+    {
+      "epoch": 3.7963383020090733,
+      "grad_norm": 0.06631217896938324,
+      "learning_rate": 1.1314640622890319e-05,
+      "loss": 0.2388,
+      "step": 46862
+    },
+    {
+      "epoch": 3.7964193130265715,
+      "grad_norm": 0.07069762051105499,
+      "learning_rate": 1.1310139970295694e-05,
+      "loss": 0.2054,
+      "step": 46863
+    },
+    {
+      "epoch": 3.79650032404407,
+      "grad_norm": 0.07219357043504715,
+      "learning_rate": 1.1305639317701066e-05,
+      "loss": 0.2596,
+      "step": 46864
+    },
+    {
+      "epoch": 3.7965813350615685,
+      "grad_norm": 0.05612671747803688,
+      "learning_rate": 1.130113866510644e-05,
+      "loss": 0.2227,
+      "step": 46865
+    },
+    {
+      "epoch": 3.7966623460790667,
+      "grad_norm": 0.08102644979953766,
+      "learning_rate": 1.1296638012511815e-05,
+      "loss": 0.2556,
+      "step": 46866
+    },
+    {
+      "epoch": 3.796743357096565,
+      "grad_norm": 0.05597464740276337,
+      "learning_rate": 1.1292137359917189e-05,
+      "loss": 0.2073,
+      "step": 46867
+    },
+    {
+      "epoch": 3.7968243681140637,
+      "grad_norm": 0.060989245772361755,
+      "learning_rate": 1.1287636707322562e-05,
+      "loss": 0.2173,
+      "step": 46868
+    },
+    {
+      "epoch": 3.796905379131562,
+      "grad_norm": 0.07221823185682297,
+      "learning_rate": 1.1283136054727936e-05,
+      "loss": 0.2246,
+      "step": 46869
+    },
+    {
+      "epoch": 3.79698639014906,
+      "grad_norm": 0.07029426097869873,
+      "learning_rate": 1.127863540213331e-05,
+      "loss": 0.2219,
+      "step": 46870
+    },
+    {
+      "epoch": 3.797067401166559,
+      "grad_norm": 0.06306787580251694,
+      "learning_rate": 1.1274134749538683e-05,
+      "loss": 0.1918,
+      "step": 46871
+    },
+    {
+      "epoch": 3.797148412184057,
+      "grad_norm": 0.0684630498290062,
+      "learning_rate": 1.1269634096944059e-05,
+      "loss": 0.2177,
+      "step": 46872
+    },
+    {
+      "epoch": 3.7972294232015553,
+      "grad_norm": 0.06849277019500732,
+      "learning_rate": 1.126513344434943e-05,
+      "loss": 0.2131,
+      "step": 46873
+    },
+    {
+      "epoch": 3.797310434219054,
+      "grad_norm": 0.08130700141191483,
+      "learning_rate": 1.1260632791754806e-05,
+      "loss": 0.2531,
+      "step": 46874
+    },
+    {
+      "epoch": 3.7973914452365523,
+      "grad_norm": 0.0685175359249115,
+      "learning_rate": 1.125613213916018e-05,
+      "loss": 0.2292,
+      "step": 46875
+    },
+    {
+      "epoch": 3.7974724562540505,
+      "grad_norm": 0.054702240973711014,
+      "learning_rate": 1.1251631486565553e-05,
+      "loss": 0.1895,
+      "step": 46876
+    },
+    {
+      "epoch": 3.797553467271549,
+      "grad_norm": 0.07435936480760574,
+      "learning_rate": 1.1247130833970927e-05,
+      "loss": 0.2479,
+      "step": 46877
+    },
+    {
+      "epoch": 3.7976344782890474,
+      "grad_norm": 0.062439724802970886,
+      "learning_rate": 1.12426301813763e-05,
+      "loss": 0.2002,
+      "step": 46878
+    },
+    {
+      "epoch": 3.7977154893065457,
+      "grad_norm": 0.07218343019485474,
+      "learning_rate": 1.1238129528781674e-05,
+      "loss": 0.1966,
+      "step": 46879
+    },
+    {
+      "epoch": 3.797796500324044,
+      "grad_norm": 0.055278975516557693,
+      "learning_rate": 1.1233628876187047e-05,
+      "loss": 0.2051,
+      "step": 46880
+    },
+    {
+      "epoch": 3.7978775113415426,
+      "grad_norm": 0.07765598595142365,
+      "learning_rate": 1.1229128223592421e-05,
+      "loss": 0.2112,
+      "step": 46881
+    },
+    {
+      "epoch": 3.797958522359041,
+      "grad_norm": 0.07940090447664261,
+      "learning_rate": 1.1224627570997795e-05,
+      "loss": 0.2491,
+      "step": 46882
+    },
+    {
+      "epoch": 3.798039533376539,
+      "grad_norm": 0.06536324322223663,
+      "learning_rate": 1.122012691840317e-05,
+      "loss": 0.2599,
+      "step": 46883
+    },
+    {
+      "epoch": 3.7981205443940373,
+      "grad_norm": 0.06517624109983444,
+      "learning_rate": 1.1215626265808543e-05,
+      "loss": 0.1863,
+      "step": 46884
+    },
+    {
+      "epoch": 3.798201555411536,
+      "grad_norm": 0.06190231814980507,
+      "learning_rate": 1.1211125613213917e-05,
+      "loss": 0.2044,
+      "step": 46885
+    },
+    {
+      "epoch": 3.7982825664290343,
+      "grad_norm": 0.0682215765118599,
+      "learning_rate": 1.120662496061929e-05,
+      "loss": 0.2717,
+      "step": 46886
+    },
+    {
+      "epoch": 3.7983635774465325,
+      "grad_norm": 0.08895383030176163,
+      "learning_rate": 1.1202124308024664e-05,
+      "loss": 0.2608,
+      "step": 46887
+    },
+    {
+      "epoch": 3.798444588464031,
+      "grad_norm": 0.07362218201160431,
+      "learning_rate": 1.1197623655430038e-05,
+      "loss": 0.224,
+      "step": 46888
+    },
+    {
+      "epoch": 3.7985255994815295,
+      "grad_norm": 0.06116807460784912,
+      "learning_rate": 1.1193123002835411e-05,
+      "loss": 0.2436,
+      "step": 46889
+    },
+    {
+      "epoch": 3.7986066104990277,
+      "grad_norm": 0.06734014302492142,
+      "learning_rate": 1.1188622350240785e-05,
+      "loss": 0.2235,
+      "step": 46890
+    },
+    {
+      "epoch": 3.7986876215165264,
+      "grad_norm": 0.06932863593101501,
+      "learning_rate": 1.1184121697646159e-05,
+      "loss": 0.2126,
+      "step": 46891
+    },
+    {
+      "epoch": 3.7987686325340246,
+      "grad_norm": 0.06863842904567719,
+      "learning_rate": 1.1179621045051534e-05,
+      "loss": 0.2258,
+      "step": 46892
+    },
+    {
+      "epoch": 3.798849643551523,
+      "grad_norm": 0.06340306252241135,
+      "learning_rate": 1.1175120392456908e-05,
+      "loss": 0.2538,
+      "step": 46893
+    },
+    {
+      "epoch": 3.7989306545690216,
+      "grad_norm": 0.0711321234703064,
+      "learning_rate": 1.117061973986228e-05,
+      "loss": 0.2343,
+      "step": 46894
+    },
+    {
+      "epoch": 3.79901166558652,
+      "grad_norm": 0.08093009144067764,
+      "learning_rate": 1.1166119087267655e-05,
+      "loss": 0.2195,
+      "step": 46895
+    },
+    {
+      "epoch": 3.799092676604018,
+      "grad_norm": 0.08460341393947601,
+      "learning_rate": 1.1161618434673028e-05,
+      "loss": 0.2329,
+      "step": 46896
+    },
+    {
+      "epoch": 3.7991736876215167,
+      "grad_norm": 0.06339055299758911,
+      "learning_rate": 1.1157117782078402e-05,
+      "loss": 0.2479,
+      "step": 46897
+    },
+    {
+      "epoch": 3.799254698639015,
+      "grad_norm": 0.06539372354745865,
+      "learning_rate": 1.1152617129483776e-05,
+      "loss": 0.2171,
+      "step": 46898
+    },
+    {
+      "epoch": 3.7993357096565132,
+      "grad_norm": 0.0670798122882843,
+      "learning_rate": 1.114811647688915e-05,
+      "loss": 0.2258,
+      "step": 46899
+    },
+    {
+      "epoch": 3.799416720674012,
+      "grad_norm": 0.07918045669794083,
+      "learning_rate": 1.1143615824294523e-05,
+      "loss": 0.2393,
+      "step": 46900
+    },
+    {
+      "epoch": 3.79949773169151,
+      "grad_norm": 0.08320426940917969,
+      "learning_rate": 1.1139115171699898e-05,
+      "loss": 0.211,
+      "step": 46901
+    },
+    {
+      "epoch": 3.7995787427090084,
+      "grad_norm": 0.07436665892601013,
+      "learning_rate": 1.1134614519105272e-05,
+      "loss": 0.2237,
+      "step": 46902
+    },
+    {
+      "epoch": 3.7996597537265067,
+      "grad_norm": 0.0866570994257927,
+      "learning_rate": 1.1130113866510644e-05,
+      "loss": 0.2256,
+      "step": 46903
+    },
+    {
+      "epoch": 3.7997407647440054,
+      "grad_norm": 0.07788897305727005,
+      "learning_rate": 1.1125613213916019e-05,
+      "loss": 0.2382,
+      "step": 46904
+    },
+    {
+      "epoch": 3.7998217757615036,
+      "grad_norm": 0.06670193374156952,
+      "learning_rate": 1.1121112561321393e-05,
+      "loss": 0.2062,
+      "step": 46905
+    },
+    {
+      "epoch": 3.799902786779002,
+      "grad_norm": 0.0638703927397728,
+      "learning_rate": 1.1116611908726766e-05,
+      "loss": 0.2127,
+      "step": 46906
+    },
+    {
+      "epoch": 3.7999837977965,
+      "grad_norm": 0.06693035364151001,
+      "learning_rate": 1.111211125613214e-05,
+      "loss": 0.2352,
+      "step": 46907
+    },
+    {
+      "epoch": 3.8000648088139988,
+      "grad_norm": 0.08188080787658691,
+      "learning_rate": 1.1107610603537513e-05,
+      "loss": 0.2332,
+      "step": 46908
+    },
+    {
+      "epoch": 3.800145819831497,
+      "grad_norm": 0.06827425211668015,
+      "learning_rate": 1.1103109950942887e-05,
+      "loss": 0.2249,
+      "step": 46909
+    },
+    {
+      "epoch": 3.8002268308489953,
+      "grad_norm": 0.06886734068393707,
+      "learning_rate": 1.1098609298348262e-05,
+      "loss": 0.2149,
+      "step": 46910
+    },
+    {
+      "epoch": 3.800307841866494,
+      "grad_norm": 0.08088071644306183,
+      "learning_rate": 1.1094108645753634e-05,
+      "loss": 0.2399,
+      "step": 46911
+    },
+    {
+      "epoch": 3.800388852883992,
+      "grad_norm": 0.08014461398124695,
+      "learning_rate": 1.1089607993159008e-05,
+      "loss": 0.2109,
+      "step": 46912
+    },
+    {
+      "epoch": 3.8004698639014904,
+      "grad_norm": 0.08599234372377396,
+      "learning_rate": 1.1085107340564383e-05,
+      "loss": 0.2075,
+      "step": 46913
+    },
+    {
+      "epoch": 3.800550874918989,
+      "grad_norm": 0.07755023241043091,
+      "learning_rate": 1.1080606687969757e-05,
+      "loss": 0.2339,
+      "step": 46914
+    },
+    {
+      "epoch": 3.8006318859364874,
+      "grad_norm": 0.06985119730234146,
+      "learning_rate": 1.107610603537513e-05,
+      "loss": 0.2058,
+      "step": 46915
+    },
+    {
+      "epoch": 3.8007128969539856,
+      "grad_norm": 0.0760570541024208,
+      "learning_rate": 1.1071605382780504e-05,
+      "loss": 0.1955,
+      "step": 46916
+    },
+    {
+      "epoch": 3.8007939079714843,
+      "grad_norm": 0.06698282063007355,
+      "learning_rate": 1.1067104730185877e-05,
+      "loss": 0.1952,
+      "step": 46917
+    },
+    {
+      "epoch": 3.8008749189889826,
+      "grad_norm": 0.07477567344903946,
+      "learning_rate": 1.1062604077591251e-05,
+      "loss": 0.214,
+      "step": 46918
+    },
+    {
+      "epoch": 3.800955930006481,
+      "grad_norm": 0.07477416098117828,
+      "learning_rate": 1.1058103424996626e-05,
+      "loss": 0.2861,
+      "step": 46919
+    },
+    {
+      "epoch": 3.8010369410239795,
+      "grad_norm": 0.07379210740327835,
+      "learning_rate": 1.1053602772401998e-05,
+      "loss": 0.2168,
+      "step": 46920
+    },
+    {
+      "epoch": 3.8011179520414777,
+      "grad_norm": 0.07364281266927719,
+      "learning_rate": 1.1049102119807372e-05,
+      "loss": 0.2101,
+      "step": 46921
+    },
+    {
+      "epoch": 3.801198963058976,
+      "grad_norm": 0.06869223713874817,
+      "learning_rate": 1.1044601467212747e-05,
+      "loss": 0.1867,
+      "step": 46922
+    },
+    {
+      "epoch": 3.8012799740764747,
+      "grad_norm": 0.072836734354496,
+      "learning_rate": 1.104010081461812e-05,
+      "loss": 0.2036,
+      "step": 46923
+    },
+    {
+      "epoch": 3.801360985093973,
+      "grad_norm": 0.060940057039260864,
+      "learning_rate": 1.1035600162023493e-05,
+      "loss": 0.196,
+      "step": 46924
+    },
+    {
+      "epoch": 3.801441996111471,
+      "grad_norm": 0.07574928551912308,
+      "learning_rate": 1.1031099509428868e-05,
+      "loss": 0.2082,
+      "step": 46925
+    },
+    {
+      "epoch": 3.8015230071289694,
+      "grad_norm": 0.08537424355745316,
+      "learning_rate": 1.1026598856834242e-05,
+      "loss": 0.233,
+      "step": 46926
+    },
+    {
+      "epoch": 3.8016040181464676,
+      "grad_norm": 0.06580768525600433,
+      "learning_rate": 1.1022098204239615e-05,
+      "loss": 0.197,
+      "step": 46927
+    },
+    {
+      "epoch": 3.8016850291639663,
+      "grad_norm": 0.06638256460428238,
+      "learning_rate": 1.1017597551644989e-05,
+      "loss": 0.2215,
+      "step": 46928
+    },
+    {
+      "epoch": 3.8017660401814646,
+      "grad_norm": 0.09222010523080826,
+      "learning_rate": 1.1013096899050362e-05,
+      "loss": 0.2341,
+      "step": 46929
+    },
+    {
+      "epoch": 3.801847051198963,
+      "grad_norm": 0.07473092526197433,
+      "learning_rate": 1.1008596246455736e-05,
+      "loss": 0.2388,
+      "step": 46930
+    },
+    {
+      "epoch": 3.8019280622164615,
+      "grad_norm": 0.07252812385559082,
+      "learning_rate": 1.1004095593861111e-05,
+      "loss": 0.2267,
+      "step": 46931
+    },
+    {
+      "epoch": 3.8020090732339598,
+      "grad_norm": 0.077595055103302,
+      "learning_rate": 1.0999594941266485e-05,
+      "loss": 0.2154,
+      "step": 46932
+    },
+    {
+      "epoch": 3.802090084251458,
+      "grad_norm": 0.06827554106712341,
+      "learning_rate": 1.0995094288671857e-05,
+      "loss": 0.2222,
+      "step": 46933
+    },
+    {
+      "epoch": 3.8021710952689567,
+      "grad_norm": 0.07482986897230148,
+      "learning_rate": 1.0990593636077232e-05,
+      "loss": 0.1913,
+      "step": 46934
+    },
+    {
+      "epoch": 3.802252106286455,
+      "grad_norm": 0.07571776211261749,
+      "learning_rate": 1.0986092983482606e-05,
+      "loss": 0.2262,
+      "step": 46935
+    },
+    {
+      "epoch": 3.802333117303953,
+      "grad_norm": 0.08740594238042831,
+      "learning_rate": 1.098159233088798e-05,
+      "loss": 0.2704,
+      "step": 46936
+    },
+    {
+      "epoch": 3.802414128321452,
+      "grad_norm": 0.08148932456970215,
+      "learning_rate": 1.0977091678293353e-05,
+      "loss": 0.1973,
+      "step": 46937
+    },
+    {
+      "epoch": 3.80249513933895,
+      "grad_norm": 0.07424408197402954,
+      "learning_rate": 1.0972591025698727e-05,
+      "loss": 0.1896,
+      "step": 46938
+    },
+    {
+      "epoch": 3.8025761503564484,
+      "grad_norm": 0.06565883755683899,
+      "learning_rate": 1.09680903731041e-05,
+      "loss": 0.193,
+      "step": 46939
+    },
+    {
+      "epoch": 3.802657161373947,
+      "grad_norm": 0.0705159530043602,
+      "learning_rate": 1.0963589720509475e-05,
+      "loss": 0.218,
+      "step": 46940
+    },
+    {
+      "epoch": 3.8027381723914453,
+      "grad_norm": 0.07602539658546448,
+      "learning_rate": 1.0959089067914847e-05,
+      "loss": 0.2325,
+      "step": 46941
+    },
+    {
+      "epoch": 3.8028191834089435,
+      "grad_norm": 0.08233897387981415,
+      "learning_rate": 1.0954588415320221e-05,
+      "loss": 0.2499,
+      "step": 46942
+    },
+    {
+      "epoch": 3.8029001944264422,
+      "grad_norm": 0.07934436202049255,
+      "learning_rate": 1.0950087762725596e-05,
+      "loss": 0.2348,
+      "step": 46943
+    },
+    {
+      "epoch": 3.8029812054439405,
+      "grad_norm": 0.06771920621395111,
+      "learning_rate": 1.094558711013097e-05,
+      "loss": 0.2272,
+      "step": 46944
+    },
+    {
+      "epoch": 3.8030622164614387,
+      "grad_norm": 0.0798344835639,
+      "learning_rate": 1.0941086457536343e-05,
+      "loss": 0.2718,
+      "step": 46945
+    },
+    {
+      "epoch": 3.8031432274789374,
+      "grad_norm": 0.062396757304668427,
+      "learning_rate": 1.0936585804941717e-05,
+      "loss": 0.2419,
+      "step": 46946
+    },
+    {
+      "epoch": 3.8032242384964356,
+      "grad_norm": 0.0885952040553093,
+      "learning_rate": 1.093208515234709e-05,
+      "loss": 0.2712,
+      "step": 46947
+    },
+    {
+      "epoch": 3.803305249513934,
+      "grad_norm": 0.06853952258825302,
+      "learning_rate": 1.0927584499752464e-05,
+      "loss": 0.2351,
+      "step": 46948
+    },
+    {
+      "epoch": 3.803386260531432,
+      "grad_norm": 0.05572928115725517,
+      "learning_rate": 1.092308384715784e-05,
+      "loss": 0.181,
+      "step": 46949
+    },
+    {
+      "epoch": 3.8034672715489304,
+      "grad_norm": 0.07789760828018188,
+      "learning_rate": 1.0918583194563211e-05,
+      "loss": 0.2282,
+      "step": 46950
+    },
+    {
+      "epoch": 3.803548282566429,
+      "grad_norm": 0.06651580333709717,
+      "learning_rate": 1.0914082541968585e-05,
+      "loss": 0.2261,
+      "step": 46951
+    },
+    {
+      "epoch": 3.8036292935839273,
+      "grad_norm": 0.07149095833301544,
+      "learning_rate": 1.090958188937396e-05,
+      "loss": 0.2143,
+      "step": 46952
+    },
+    {
+      "epoch": 3.8037103046014256,
+      "grad_norm": 0.07038377225399017,
+      "learning_rate": 1.0905081236779334e-05,
+      "loss": 0.2192,
+      "step": 46953
+    },
+    {
+      "epoch": 3.8037913156189243,
+      "grad_norm": 0.07646401226520538,
+      "learning_rate": 1.0900580584184706e-05,
+      "loss": 0.2406,
+      "step": 46954
+    },
+    {
+      "epoch": 3.8038723266364225,
+      "grad_norm": 0.09022786468267441,
+      "learning_rate": 1.0896079931590081e-05,
+      "loss": 0.2274,
+      "step": 46955
+    },
+    {
+      "epoch": 3.8039533376539207,
+      "grad_norm": 0.0737614631652832,
+      "learning_rate": 1.0891579278995455e-05,
+      "loss": 0.2387,
+      "step": 46956
+    },
+    {
+      "epoch": 3.8040343486714194,
+      "grad_norm": 0.06396791338920593,
+      "learning_rate": 1.0887078626400828e-05,
+      "loss": 0.2446,
+      "step": 46957
+    },
+    {
+      "epoch": 3.8041153596889177,
+      "grad_norm": 0.09200059622526169,
+      "learning_rate": 1.0882577973806204e-05,
+      "loss": 0.225,
+      "step": 46958
+    },
+    {
+      "epoch": 3.804196370706416,
+      "grad_norm": 0.07705798000097275,
+      "learning_rate": 1.0878077321211576e-05,
+      "loss": 0.2335,
+      "step": 46959
+    },
+    {
+      "epoch": 3.8042773817239146,
+      "grad_norm": 0.06533969193696976,
+      "learning_rate": 1.087357666861695e-05,
+      "loss": 0.2366,
+      "step": 46960
+    },
+    {
+      "epoch": 3.804358392741413,
+      "grad_norm": 0.06600935012102127,
+      "learning_rate": 1.0869076016022324e-05,
+      "loss": 0.2587,
+      "step": 46961
+    },
+    {
+      "epoch": 3.804439403758911,
+      "grad_norm": 0.06288919597864151,
+      "learning_rate": 1.0864575363427698e-05,
+      "loss": 0.2536,
+      "step": 46962
+    },
+    {
+      "epoch": 3.80452041477641,
+      "grad_norm": 0.09356596320867538,
+      "learning_rate": 1.086007471083307e-05,
+      "loss": 0.2503,
+      "step": 46963
+    },
+    {
+      "epoch": 3.804601425793908,
+      "grad_norm": 0.0647389367222786,
+      "learning_rate": 1.0855574058238445e-05,
+      "loss": 0.2205,
+      "step": 46964
+    },
+    {
+      "epoch": 3.8046824368114063,
+      "grad_norm": 0.06880980730056763,
+      "learning_rate": 1.0851073405643819e-05,
+      "loss": 0.2179,
+      "step": 46965
+    },
+    {
+      "epoch": 3.804763447828905,
+      "grad_norm": 0.06399885565042496,
+      "learning_rate": 1.0846572753049192e-05,
+      "loss": 0.2462,
+      "step": 46966
+    },
+    {
+      "epoch": 3.804844458846403,
+      "grad_norm": 0.07660163193941116,
+      "learning_rate": 1.0842072100454566e-05,
+      "loss": 0.2162,
+      "step": 46967
+    },
+    {
+      "epoch": 3.8049254698639015,
+      "grad_norm": 0.06323090195655823,
+      "learning_rate": 1.083757144785994e-05,
+      "loss": 0.2289,
+      "step": 46968
+    },
+    {
+      "epoch": 3.8050064808814,
+      "grad_norm": 0.05920364335179329,
+      "learning_rate": 1.0833070795265313e-05,
+      "loss": 0.2378,
+      "step": 46969
+    },
+    {
+      "epoch": 3.8050874918988984,
+      "grad_norm": 0.07590261846780777,
+      "learning_rate": 1.0828570142670689e-05,
+      "loss": 0.205,
+      "step": 46970
+    },
+    {
+      "epoch": 3.8051685029163966,
+      "grad_norm": 0.07777293771505356,
+      "learning_rate": 1.082406949007606e-05,
+      "loss": 0.2218,
+      "step": 46971
+    },
+    {
+      "epoch": 3.805249513933895,
+      "grad_norm": 0.09066983312368393,
+      "learning_rate": 1.0819568837481436e-05,
+      "loss": 0.2632,
+      "step": 46972
+    },
+    {
+      "epoch": 3.805330524951393,
+      "grad_norm": 0.07806780189275742,
+      "learning_rate": 1.081506818488681e-05,
+      "loss": 0.1912,
+      "step": 46973
+    },
+    {
+      "epoch": 3.805411535968892,
+      "grad_norm": 0.0826018676161766,
+      "learning_rate": 1.0810567532292183e-05,
+      "loss": 0.2424,
+      "step": 46974
+    },
+    {
+      "epoch": 3.80549254698639,
+      "grad_norm": 0.08856657892465591,
+      "learning_rate": 1.0806066879697557e-05,
+      "loss": 0.2551,
+      "step": 46975
+    },
+    {
+      "epoch": 3.8055735580038883,
+      "grad_norm": 0.06531494110822678,
+      "learning_rate": 1.080156622710293e-05,
+      "loss": 0.223,
+      "step": 46976
+    },
+    {
+      "epoch": 3.805654569021387,
+      "grad_norm": 0.07311249524354935,
+      "learning_rate": 1.0797065574508304e-05,
+      "loss": 0.2724,
+      "step": 46977
+    },
+    {
+      "epoch": 3.8057355800388852,
+      "grad_norm": 0.060811128467321396,
+      "learning_rate": 1.0792564921913677e-05,
+      "loss": 0.2253,
+      "step": 46978
+    },
+    {
+      "epoch": 3.8058165910563835,
+      "grad_norm": 0.06457604467868805,
+      "learning_rate": 1.0788064269319053e-05,
+      "loss": 0.2496,
+      "step": 46979
+    },
+    {
+      "epoch": 3.805897602073882,
+      "grad_norm": 0.08397595584392548,
+      "learning_rate": 1.0783563616724425e-05,
+      "loss": 0.2444,
+      "step": 46980
+    },
+    {
+      "epoch": 3.8059786130913804,
+      "grad_norm": 0.07146035879850388,
+      "learning_rate": 1.07790629641298e-05,
+      "loss": 0.2103,
+      "step": 46981
+    },
+    {
+      "epoch": 3.8060596241088787,
+      "grad_norm": 0.08192295581102371,
+      "learning_rate": 1.0774562311535174e-05,
+      "loss": 0.2084,
+      "step": 46982
+    },
+    {
+      "epoch": 3.8061406351263773,
+      "grad_norm": 0.057976480573415756,
+      "learning_rate": 1.0770061658940547e-05,
+      "loss": 0.2378,
+      "step": 46983
+    },
+    {
+      "epoch": 3.8062216461438756,
+      "grad_norm": 0.05698695778846741,
+      "learning_rate": 1.076556100634592e-05,
+      "loss": 0.1849,
+      "step": 46984
+    },
+    {
+      "epoch": 3.806302657161374,
+      "grad_norm": 0.06821150332689285,
+      "learning_rate": 1.0761060353751294e-05,
+      "loss": 0.2393,
+      "step": 46985
+    },
+    {
+      "epoch": 3.8063836681788725,
+      "grad_norm": 0.07215702533721924,
+      "learning_rate": 1.0756559701156668e-05,
+      "loss": 0.2485,
+      "step": 46986
+    },
+    {
+      "epoch": 3.8064646791963708,
+      "grad_norm": 0.08539849519729614,
+      "learning_rate": 1.0752059048562042e-05,
+      "loss": 0.2288,
+      "step": 46987
+    },
+    {
+      "epoch": 3.806545690213869,
+      "grad_norm": 0.08093149214982986,
+      "learning_rate": 1.0747558395967417e-05,
+      "loss": 0.2345,
+      "step": 46988
+    },
+    {
+      "epoch": 3.8066267012313677,
+      "grad_norm": 0.06490255147218704,
+      "learning_rate": 1.0743057743372789e-05,
+      "loss": 0.2143,
+      "step": 46989
+    },
+    {
+      "epoch": 3.806707712248866,
+      "grad_norm": 0.08042438328266144,
+      "learning_rate": 1.0738557090778164e-05,
+      "loss": 0.2459,
+      "step": 46990
+    },
+    {
+      "epoch": 3.806788723266364,
+      "grad_norm": 0.09989763796329498,
+      "learning_rate": 1.0734056438183538e-05,
+      "loss": 0.2358,
+      "step": 46991
+    },
+    {
+      "epoch": 3.806869734283863,
+      "grad_norm": 0.06737187504768372,
+      "learning_rate": 1.0729555785588911e-05,
+      "loss": 0.2002,
+      "step": 46992
+    },
+    {
+      "epoch": 3.806950745301361,
+      "grad_norm": 0.0732245147228241,
+      "learning_rate": 1.0725055132994285e-05,
+      "loss": 0.2308,
+      "step": 46993
+    },
+    {
+      "epoch": 3.8070317563188594,
+      "grad_norm": 0.08116856962442398,
+      "learning_rate": 1.0720554480399658e-05,
+      "loss": 0.2367,
+      "step": 46994
+    },
+    {
+      "epoch": 3.8071127673363576,
+      "grad_norm": 0.08011330664157867,
+      "learning_rate": 1.0716053827805032e-05,
+      "loss": 0.2478,
+      "step": 46995
+    },
+    {
+      "epoch": 3.807193778353856,
+      "grad_norm": 0.06650631129741669,
+      "learning_rate": 1.0711553175210406e-05,
+      "loss": 0.2343,
+      "step": 46996
+    },
+    {
+      "epoch": 3.8072747893713546,
+      "grad_norm": 0.075144462287426,
+      "learning_rate": 1.070705252261578e-05,
+      "loss": 0.2326,
+      "step": 46997
+    },
+    {
+      "epoch": 3.807355800388853,
+      "grad_norm": 0.05947421118617058,
+      "learning_rate": 1.0702551870021153e-05,
+      "loss": 0.1932,
+      "step": 46998
+    },
+    {
+      "epoch": 3.807436811406351,
+      "grad_norm": 0.08503783494234085,
+      "learning_rate": 1.0698051217426528e-05,
+      "loss": 0.2376,
+      "step": 46999
+    },
+    {
+      "epoch": 3.8075178224238497,
+      "grad_norm": 0.06805943697690964,
+      "learning_rate": 1.0693550564831902e-05,
+      "loss": 0.1997,
+      "step": 47000
+    },
+    {
+      "epoch": 3.807598833441348,
+      "grad_norm": 0.07300718128681183,
+      "learning_rate": 1.0689049912237275e-05,
+      "loss": 0.2294,
+      "step": 47001
+    },
+    {
+      "epoch": 3.807679844458846,
+      "grad_norm": 0.06472918391227722,
+      "learning_rate": 1.0684549259642649e-05,
+      "loss": 0.2222,
+      "step": 47002
+    },
+    {
+      "epoch": 3.807760855476345,
+      "grad_norm": 0.08821234107017517,
+      "learning_rate": 1.0680048607048023e-05,
+      "loss": 0.2618,
+      "step": 47003
+    },
+    {
+      "epoch": 3.807841866493843,
+      "grad_norm": 0.061438728123903275,
+      "learning_rate": 1.0675547954453396e-05,
+      "loss": 0.2295,
+      "step": 47004
+    },
+    {
+      "epoch": 3.8079228775113414,
+      "grad_norm": 0.0627550482749939,
+      "learning_rate": 1.067104730185877e-05,
+      "loss": 0.2232,
+      "step": 47005
+    },
+    {
+      "epoch": 3.80800388852884,
+      "grad_norm": 0.07976122945547104,
+      "learning_rate": 1.0666546649264143e-05,
+      "loss": 0.2091,
+      "step": 47006
+    },
+    {
+      "epoch": 3.8080848995463383,
+      "grad_norm": 0.06467067450284958,
+      "learning_rate": 1.0662045996669517e-05,
+      "loss": 0.2178,
+      "step": 47007
+    },
+    {
+      "epoch": 3.8081659105638366,
+      "grad_norm": 0.07180802524089813,
+      "learning_rate": 1.0657545344074892e-05,
+      "loss": 0.1963,
+      "step": 47008
+    },
+    {
+      "epoch": 3.8082469215813353,
+      "grad_norm": 0.06795401126146317,
+      "learning_rate": 1.0653044691480266e-05,
+      "loss": 0.2451,
+      "step": 47009
+    },
+    {
+      "epoch": 3.8083279325988335,
+      "grad_norm": 0.10496599227190018,
+      "learning_rate": 1.0648544038885638e-05,
+      "loss": 0.2441,
+      "step": 47010
+    },
+    {
+      "epoch": 3.8084089436163318,
+      "grad_norm": 0.0665077492594719,
+      "learning_rate": 1.0644043386291013e-05,
+      "loss": 0.2213,
+      "step": 47011
+    },
+    {
+      "epoch": 3.8084899546338304,
+      "grad_norm": 0.07035201042890549,
+      "learning_rate": 1.0639542733696387e-05,
+      "loss": 0.2269,
+      "step": 47012
+    },
+    {
+      "epoch": 3.8085709656513287,
+      "grad_norm": 0.07154831290245056,
+      "learning_rate": 1.063504208110176e-05,
+      "loss": 0.2242,
+      "step": 47013
+    },
+    {
+      "epoch": 3.808651976668827,
+      "grad_norm": 0.0710812658071518,
+      "learning_rate": 1.0630541428507134e-05,
+      "loss": 0.2438,
+      "step": 47014
+    },
+    {
+      "epoch": 3.808732987686325,
+      "grad_norm": 0.07493715733289719,
+      "learning_rate": 1.0626040775912508e-05,
+      "loss": 0.1937,
+      "step": 47015
+    },
+    {
+      "epoch": 3.808813998703824,
+      "grad_norm": 0.0741448625922203,
+      "learning_rate": 1.0621540123317881e-05,
+      "loss": 0.2626,
+      "step": 47016
+    },
+    {
+      "epoch": 3.808895009721322,
+      "grad_norm": 0.07966463267803192,
+      "learning_rate": 1.0617039470723256e-05,
+      "loss": 0.2063,
+      "step": 47017
+    },
+    {
+      "epoch": 3.8089760207388204,
+      "grad_norm": 0.0726398155093193,
+      "learning_rate": 1.061253881812863e-05,
+      "loss": 0.2499,
+      "step": 47018
+    },
+    {
+      "epoch": 3.8090570317563186,
+      "grad_norm": 0.07825640588998795,
+      "learning_rate": 1.0608038165534002e-05,
+      "loss": 0.2546,
+      "step": 47019
+    },
+    {
+      "epoch": 3.8091380427738173,
+      "grad_norm": 0.07793726027011871,
+      "learning_rate": 1.0603537512939377e-05,
+      "loss": 0.22,
+      "step": 47020
+    },
+    {
+      "epoch": 3.8092190537913155,
+      "grad_norm": 0.06841432303190231,
+      "learning_rate": 1.059903686034475e-05,
+      "loss": 0.186,
+      "step": 47021
+    },
+    {
+      "epoch": 3.809300064808814,
+      "grad_norm": 0.06918314844369888,
+      "learning_rate": 1.0594536207750124e-05,
+      "loss": 0.2519,
+      "step": 47022
+    },
+    {
+      "epoch": 3.8093810758263125,
+      "grad_norm": 0.07555932551622391,
+      "learning_rate": 1.0590035555155498e-05,
+      "loss": 0.2431,
+      "step": 47023
+    },
+    {
+      "epoch": 3.8094620868438107,
+      "grad_norm": 0.06252846866846085,
+      "learning_rate": 1.0585534902560872e-05,
+      "loss": 0.2124,
+      "step": 47024
+    },
+    {
+      "epoch": 3.809543097861309,
+      "grad_norm": 0.07834303379058838,
+      "learning_rate": 1.0581034249966245e-05,
+      "loss": 0.2625,
+      "step": 47025
+    },
+    {
+      "epoch": 3.8096241088788076,
+      "grad_norm": 0.08204419165849686,
+      "learning_rate": 1.057653359737162e-05,
+      "loss": 0.2187,
+      "step": 47026
+    },
+    {
+      "epoch": 3.809705119896306,
+      "grad_norm": 0.06510677188634872,
+      "learning_rate": 1.0572032944776992e-05,
+      "loss": 0.2681,
+      "step": 47027
+    },
+    {
+      "epoch": 3.809786130913804,
+      "grad_norm": 0.06332890689373016,
+      "learning_rate": 1.0567532292182366e-05,
+      "loss": 0.2501,
+      "step": 47028
+    },
+    {
+      "epoch": 3.809867141931303,
+      "grad_norm": 0.057162806391716,
+      "learning_rate": 1.0563031639587741e-05,
+      "loss": 0.2032,
+      "step": 47029
+    },
+    {
+      "epoch": 3.809948152948801,
+      "grad_norm": 0.06294887512922287,
+      "learning_rate": 1.0558530986993115e-05,
+      "loss": 0.2121,
+      "step": 47030
+    },
+    {
+      "epoch": 3.8100291639662993,
+      "grad_norm": 0.06234182044863701,
+      "learning_rate": 1.0554030334398489e-05,
+      "loss": 0.2268,
+      "step": 47031
+    },
+    {
+      "epoch": 3.810110174983798,
+      "grad_norm": 0.06718173623085022,
+      "learning_rate": 1.0549529681803862e-05,
+      "loss": 0.2252,
+      "step": 47032
+    },
+    {
+      "epoch": 3.8101911860012962,
+      "grad_norm": 0.07185987383127213,
+      "learning_rate": 1.0545029029209236e-05,
+      "loss": 0.24,
+      "step": 47033
+    },
+    {
+      "epoch": 3.8102721970187945,
+      "grad_norm": 0.07447812706232071,
+      "learning_rate": 1.054052837661461e-05,
+      "loss": 0.2539,
+      "step": 47034
+    },
+    {
+      "epoch": 3.810353208036293,
+      "grad_norm": 0.06620065122842789,
+      "learning_rate": 1.0536027724019985e-05,
+      "loss": 0.2141,
+      "step": 47035
+    },
+    {
+      "epoch": 3.8104342190537914,
+      "grad_norm": 0.07579522579908371,
+      "learning_rate": 1.0531527071425357e-05,
+      "loss": 0.227,
+      "step": 47036
+    },
+    {
+      "epoch": 3.8105152300712897,
+      "grad_norm": 0.07805155217647552,
+      "learning_rate": 1.052702641883073e-05,
+      "loss": 0.2272,
+      "step": 47037
+    },
+    {
+      "epoch": 3.810596241088788,
+      "grad_norm": 0.06929910182952881,
+      "learning_rate": 1.0522525766236105e-05,
+      "loss": 0.2378,
+      "step": 47038
+    },
+    {
+      "epoch": 3.8106772521062866,
+      "grad_norm": 0.0670161247253418,
+      "learning_rate": 1.0518025113641479e-05,
+      "loss": 0.1883,
+      "step": 47039
+    },
+    {
+      "epoch": 3.810758263123785,
+      "grad_norm": 0.06071063503623009,
+      "learning_rate": 1.0513524461046851e-05,
+      "loss": 0.2017,
+      "step": 47040
+    },
+    {
+      "epoch": 3.810839274141283,
+      "grad_norm": 0.07034777849912643,
+      "learning_rate": 1.0509023808452226e-05,
+      "loss": 0.2325,
+      "step": 47041
+    },
+    {
+      "epoch": 3.8109202851587813,
+      "grad_norm": 0.0668913945555687,
+      "learning_rate": 1.05045231558576e-05,
+      "loss": 0.2328,
+      "step": 47042
+    },
+    {
+      "epoch": 3.81100129617628,
+      "grad_norm": 0.07246463745832443,
+      "learning_rate": 1.0500022503262973e-05,
+      "loss": 0.2207,
+      "step": 47043
+    },
+    {
+      "epoch": 3.8110823071937783,
+      "grad_norm": 0.07312662899494171,
+      "learning_rate": 1.0495521850668349e-05,
+      "loss": 0.1934,
+      "step": 47044
+    },
+    {
+      "epoch": 3.8111633182112765,
+      "grad_norm": 0.08061020821332932,
+      "learning_rate": 1.049102119807372e-05,
+      "loss": 0.2648,
+      "step": 47045
+    },
+    {
+      "epoch": 3.811244329228775,
+      "grad_norm": 0.06351017206907272,
+      "learning_rate": 1.0486520545479094e-05,
+      "loss": 0.2229,
+      "step": 47046
+    },
+    {
+      "epoch": 3.8113253402462735,
+      "grad_norm": 0.06795711815357208,
+      "learning_rate": 1.048201989288447e-05,
+      "loss": 0.2055,
+      "step": 47047
+    },
+    {
+      "epoch": 3.8114063512637717,
+      "grad_norm": 0.06773922592401505,
+      "learning_rate": 1.0477519240289843e-05,
+      "loss": 0.2385,
+      "step": 47048
+    },
+    {
+      "epoch": 3.8114873622812704,
+      "grad_norm": 0.07266738265752792,
+      "learning_rate": 1.0473018587695215e-05,
+      "loss": 0.2618,
+      "step": 47049
+    },
+    {
+      "epoch": 3.8115683732987686,
+      "grad_norm": 0.08389297127723694,
+      "learning_rate": 1.046851793510059e-05,
+      "loss": 0.2324,
+      "step": 47050
+    },
+    {
+      "epoch": 3.811649384316267,
+      "grad_norm": 0.06552331149578094,
+      "learning_rate": 1.0464017282505964e-05,
+      "loss": 0.2226,
+      "step": 47051
+    },
+    {
+      "epoch": 3.8117303953337656,
+      "grad_norm": 0.07753307372331619,
+      "learning_rate": 1.0459516629911338e-05,
+      "loss": 0.2392,
+      "step": 47052
+    },
+    {
+      "epoch": 3.811811406351264,
+      "grad_norm": 0.07049240171909332,
+      "learning_rate": 1.0455015977316711e-05,
+      "loss": 0.1888,
+      "step": 47053
+    },
+    {
+      "epoch": 3.811892417368762,
+      "grad_norm": 0.07749372720718384,
+      "learning_rate": 1.0450515324722085e-05,
+      "loss": 0.2226,
+      "step": 47054
+    },
+    {
+      "epoch": 3.8119734283862607,
+      "grad_norm": 0.07106560468673706,
+      "learning_rate": 1.0446014672127458e-05,
+      "loss": 0.2248,
+      "step": 47055
+    },
+    {
+      "epoch": 3.812054439403759,
+      "grad_norm": 0.0729873925447464,
+      "learning_rate": 1.0441514019532834e-05,
+      "loss": 0.2439,
+      "step": 47056
+    },
+    {
+      "epoch": 3.8121354504212572,
+      "grad_norm": 0.06278953701257706,
+      "learning_rate": 1.0437013366938206e-05,
+      "loss": 0.1792,
+      "step": 47057
+    },
+    {
+      "epoch": 3.812216461438756,
+      "grad_norm": 0.08032325655221939,
+      "learning_rate": 1.043251271434358e-05,
+      "loss": 0.2306,
+      "step": 47058
+    },
+    {
+      "epoch": 3.812297472456254,
+      "grad_norm": 0.07312075048685074,
+      "learning_rate": 1.0428012061748955e-05,
+      "loss": 0.2361,
+      "step": 47059
+    },
+    {
+      "epoch": 3.8123784834737524,
+      "grad_norm": 0.09086312353610992,
+      "learning_rate": 1.0423511409154328e-05,
+      "loss": 0.2511,
+      "step": 47060
+    },
+    {
+      "epoch": 3.8124594944912507,
+      "grad_norm": 0.07702907919883728,
+      "learning_rate": 1.0419010756559702e-05,
+      "loss": 0.1923,
+      "step": 47061
+    },
+    {
+      "epoch": 3.8125405055087493,
+      "grad_norm": 0.09750400483608246,
+      "learning_rate": 1.0414510103965075e-05,
+      "loss": 0.2327,
+      "step": 47062
+    },
+    {
+      "epoch": 3.8126215165262476,
+      "grad_norm": 0.0724315345287323,
+      "learning_rate": 1.0410009451370449e-05,
+      "loss": 0.2207,
+      "step": 47063
+    },
+    {
+      "epoch": 3.812702527543746,
+      "grad_norm": 0.06693287193775177,
+      "learning_rate": 1.0405508798775823e-05,
+      "loss": 0.2127,
+      "step": 47064
+    },
+    {
+      "epoch": 3.812783538561244,
+      "grad_norm": 0.07102993130683899,
+      "learning_rate": 1.0401008146181198e-05,
+      "loss": 0.2034,
+      "step": 47065
+    },
+    {
+      "epoch": 3.8128645495787428,
+      "grad_norm": 0.07117815315723419,
+      "learning_rate": 1.039650749358657e-05,
+      "loss": 0.2773,
+      "step": 47066
+    },
+    {
+      "epoch": 3.812945560596241,
+      "grad_norm": 0.06364418566226959,
+      "learning_rate": 1.0392006840991943e-05,
+      "loss": 0.243,
+      "step": 47067
+    },
+    {
+      "epoch": 3.8130265716137393,
+      "grad_norm": 0.10085894912481308,
+      "learning_rate": 1.0387506188397319e-05,
+      "loss": 0.2299,
+      "step": 47068
+    },
+    {
+      "epoch": 3.813107582631238,
+      "grad_norm": 0.06981120258569717,
+      "learning_rate": 1.0383005535802692e-05,
+      "loss": 0.1809,
+      "step": 47069
+    },
+    {
+      "epoch": 3.813188593648736,
+      "grad_norm": 0.06516273319721222,
+      "learning_rate": 1.0378504883208064e-05,
+      "loss": 0.1991,
+      "step": 47070
+    },
+    {
+      "epoch": 3.8132696046662344,
+      "grad_norm": 0.08382715284824371,
+      "learning_rate": 1.037400423061344e-05,
+      "loss": 0.2271,
+      "step": 47071
+    },
+    {
+      "epoch": 3.813350615683733,
+      "grad_norm": 0.08625270426273346,
+      "learning_rate": 1.0369503578018813e-05,
+      "loss": 0.2135,
+      "step": 47072
+    },
+    {
+      "epoch": 3.8134316267012314,
+      "grad_norm": 0.07801330834627151,
+      "learning_rate": 1.0365002925424187e-05,
+      "loss": 0.2405,
+      "step": 47073
+    },
+    {
+      "epoch": 3.8135126377187296,
+      "grad_norm": 0.08635073155164719,
+      "learning_rate": 1.0360502272829562e-05,
+      "loss": 0.2164,
+      "step": 47074
+    },
+    {
+      "epoch": 3.8135936487362283,
+      "grad_norm": 0.07388366758823395,
+      "learning_rate": 1.0356001620234934e-05,
+      "loss": 0.2158,
+      "step": 47075
+    },
+    {
+      "epoch": 3.8136746597537265,
+      "grad_norm": 0.06860913336277008,
+      "learning_rate": 1.0351500967640307e-05,
+      "loss": 0.2161,
+      "step": 47076
+    },
+    {
+      "epoch": 3.813755670771225,
+      "grad_norm": 0.07081514596939087,
+      "learning_rate": 1.0347000315045683e-05,
+      "loss": 0.2519,
+      "step": 47077
+    },
+    {
+      "epoch": 3.8138366817887235,
+      "grad_norm": 0.07424487173557281,
+      "learning_rate": 1.0342499662451056e-05,
+      "loss": 0.2206,
+      "step": 47078
+    },
+    {
+      "epoch": 3.8139176928062217,
+      "grad_norm": 0.07972376048564911,
+      "learning_rate": 1.033799900985643e-05,
+      "loss": 0.2378,
+      "step": 47079
+    },
+    {
+      "epoch": 3.81399870382372,
+      "grad_norm": 0.07115171104669571,
+      "learning_rate": 1.0333498357261804e-05,
+      "loss": 0.2134,
+      "step": 47080
+    },
+    {
+      "epoch": 3.8140797148412187,
+      "grad_norm": 0.07627512514591217,
+      "learning_rate": 1.0328997704667177e-05,
+      "loss": 0.19,
+      "step": 47081
+    },
+    {
+      "epoch": 3.814160725858717,
+      "grad_norm": 0.07725601643323898,
+      "learning_rate": 1.032449705207255e-05,
+      "loss": 0.2713,
+      "step": 47082
+    },
+    {
+      "epoch": 3.814241736876215,
+      "grad_norm": 0.07265699654817581,
+      "learning_rate": 1.0319996399477924e-05,
+      "loss": 0.2425,
+      "step": 47083
+    },
+    {
+      "epoch": 3.8143227478937134,
+      "grad_norm": 0.07284402847290039,
+      "learning_rate": 1.0315495746883298e-05,
+      "loss": 0.2464,
+      "step": 47084
+    },
+    {
+      "epoch": 3.814403758911212,
+      "grad_norm": 0.07427023351192474,
+      "learning_rate": 1.0310995094288672e-05,
+      "loss": 0.2099,
+      "step": 47085
+    },
+    {
+      "epoch": 3.8144847699287103,
+      "grad_norm": 0.05913237854838371,
+      "learning_rate": 1.0306494441694047e-05,
+      "loss": 0.1989,
+      "step": 47086
+    },
+    {
+      "epoch": 3.8145657809462086,
+      "grad_norm": 0.07824961096048355,
+      "learning_rate": 1.030199378909942e-05,
+      "loss": 0.2294,
+      "step": 47087
+    },
+    {
+      "epoch": 3.814646791963707,
+      "grad_norm": 0.06919219344854355,
+      "learning_rate": 1.0297493136504794e-05,
+      "loss": 0.2452,
+      "step": 47088
+    },
+    {
+      "epoch": 3.8147278029812055,
+      "grad_norm": 0.09075573086738586,
+      "learning_rate": 1.0292992483910168e-05,
+      "loss": 0.2306,
+      "step": 47089
+    },
+    {
+      "epoch": 3.8148088139987038,
+      "grad_norm": 0.07499659061431885,
+      "learning_rate": 1.0288491831315541e-05,
+      "loss": 0.1971,
+      "step": 47090
+    },
+    {
+      "epoch": 3.814889825016202,
+      "grad_norm": 0.06980670988559723,
+      "learning_rate": 1.0283991178720915e-05,
+      "loss": 0.2373,
+      "step": 47091
+    },
+    {
+      "epoch": 3.8149708360337007,
+      "grad_norm": 0.06824301928281784,
+      "learning_rate": 1.0279490526126289e-05,
+      "loss": 0.2478,
+      "step": 47092
+    },
+    {
+      "epoch": 3.815051847051199,
+      "grad_norm": 0.06680431962013245,
+      "learning_rate": 1.0274989873531662e-05,
+      "loss": 0.2322,
+      "step": 47093
+    },
+    {
+      "epoch": 3.815132858068697,
+      "grad_norm": 0.06365301460027695,
+      "learning_rate": 1.0270489220937036e-05,
+      "loss": 0.2449,
+      "step": 47094
+    },
+    {
+      "epoch": 3.815213869086196,
+      "grad_norm": 0.07429978251457214,
+      "learning_rate": 1.0265988568342411e-05,
+      "loss": 0.2178,
+      "step": 47095
+    },
+    {
+      "epoch": 3.815294880103694,
+      "grad_norm": 0.07181497663259506,
+      "learning_rate": 1.0261487915747783e-05,
+      "loss": 0.2281,
+      "step": 47096
+    },
+    {
+      "epoch": 3.8153758911211924,
+      "grad_norm": 0.06791014224290848,
+      "learning_rate": 1.0256987263153158e-05,
+      "loss": 0.2259,
+      "step": 47097
+    },
+    {
+      "epoch": 3.815456902138691,
+      "grad_norm": 0.08312923461198807,
+      "learning_rate": 1.0252486610558532e-05,
+      "loss": 0.2231,
+      "step": 47098
+    },
+    {
+      "epoch": 3.8155379131561893,
+      "grad_norm": 0.07397427409887314,
+      "learning_rate": 1.0247985957963905e-05,
+      "loss": 0.2101,
+      "step": 47099
+    },
+    {
+      "epoch": 3.8156189241736875,
+      "grad_norm": 0.08106406778097153,
+      "learning_rate": 1.0243485305369279e-05,
+      "loss": 0.2218,
+      "step": 47100
+    },
+    {
+      "epoch": 3.815699935191186,
+      "grad_norm": 0.07498326152563095,
+      "learning_rate": 1.0238984652774653e-05,
+      "loss": 0.2559,
+      "step": 47101
+    },
+    {
+      "epoch": 3.8157809462086845,
+      "grad_norm": 0.07188771665096283,
+      "learning_rate": 1.0234484000180026e-05,
+      "loss": 0.2475,
+      "step": 47102
+    },
+    {
+      "epoch": 3.8158619572261827,
+      "grad_norm": 0.07121875882148743,
+      "learning_rate": 1.02299833475854e-05,
+      "loss": 0.238,
+      "step": 47103
+    },
+    {
+      "epoch": 3.8159429682436814,
+      "grad_norm": 0.067873515188694,
+      "learning_rate": 1.0225482694990775e-05,
+      "loss": 0.1926,
+      "step": 47104
+    },
+    {
+      "epoch": 3.8160239792611796,
+      "grad_norm": 0.07478199154138565,
+      "learning_rate": 1.0220982042396147e-05,
+      "loss": 0.245,
+      "step": 47105
+    },
+    {
+      "epoch": 3.816104990278678,
+      "grad_norm": 0.06381670385599136,
+      "learning_rate": 1.0216481389801522e-05,
+      "loss": 0.265,
+      "step": 47106
+    },
+    {
+      "epoch": 3.816186001296176,
+      "grad_norm": 0.07955259829759598,
+      "learning_rate": 1.0211980737206896e-05,
+      "loss": 0.2075,
+      "step": 47107
+    },
+    {
+      "epoch": 3.816267012313675,
+      "grad_norm": 0.09649759531021118,
+      "learning_rate": 1.020748008461227e-05,
+      "loss": 0.236,
+      "step": 47108
+    },
+    {
+      "epoch": 3.816348023331173,
+      "grad_norm": 0.06495945900678635,
+      "learning_rate": 1.0202979432017643e-05,
+      "loss": 0.2247,
+      "step": 47109
+    },
+    {
+      "epoch": 3.8164290343486713,
+      "grad_norm": 0.07418833673000336,
+      "learning_rate": 1.0198478779423017e-05,
+      "loss": 0.2183,
+      "step": 47110
+    },
+    {
+      "epoch": 3.8165100453661696,
+      "grad_norm": 0.0698823407292366,
+      "learning_rate": 1.019397812682839e-05,
+      "loss": 0.1966,
+      "step": 47111
+    },
+    {
+      "epoch": 3.8165910563836682,
+      "grad_norm": 0.08725552260875702,
+      "learning_rate": 1.0189477474233764e-05,
+      "loss": 0.2079,
+      "step": 47112
+    },
+    {
+      "epoch": 3.8166720674011665,
+      "grad_norm": 0.060063328593969345,
+      "learning_rate": 1.0184976821639138e-05,
+      "loss": 0.1972,
+      "step": 47113
+    },
+    {
+      "epoch": 3.8167530784186647,
+      "grad_norm": 0.06648138910531998,
+      "learning_rate": 1.0180476169044511e-05,
+      "loss": 0.2085,
+      "step": 47114
+    },
+    {
+      "epoch": 3.8168340894361634,
+      "grad_norm": 0.06786450743675232,
+      "learning_rate": 1.0175975516449886e-05,
+      "loss": 0.1979,
+      "step": 47115
+    },
+    {
+      "epoch": 3.8169151004536617,
+      "grad_norm": 0.06979463994503021,
+      "learning_rate": 1.017147486385526e-05,
+      "loss": 0.2417,
+      "step": 47116
+    },
+    {
+      "epoch": 3.81699611147116,
+      "grad_norm": 0.08767568320035934,
+      "learning_rate": 1.0166974211260634e-05,
+      "loss": 0.2051,
+      "step": 47117
+    },
+    {
+      "epoch": 3.8170771224886586,
+      "grad_norm": 0.061907850205898285,
+      "learning_rate": 1.0162473558666007e-05,
+      "loss": 0.2234,
+      "step": 47118
+    },
+    {
+      "epoch": 3.817158133506157,
+      "grad_norm": 0.06363208591938019,
+      "learning_rate": 1.0157972906071381e-05,
+      "loss": 0.1849,
+      "step": 47119
+    },
+    {
+      "epoch": 3.817239144523655,
+      "grad_norm": 0.07457547634840012,
+      "learning_rate": 1.0153472253476755e-05,
+      "loss": 0.2401,
+      "step": 47120
+    },
+    {
+      "epoch": 3.817320155541154,
+      "grad_norm": 0.07944779843091965,
+      "learning_rate": 1.0148971600882128e-05,
+      "loss": 0.2281,
+      "step": 47121
+    },
+    {
+      "epoch": 3.817401166558652,
+      "grad_norm": 0.07503432035446167,
+      "learning_rate": 1.0144470948287502e-05,
+      "loss": 0.2235,
+      "step": 47122
+    },
+    {
+      "epoch": 3.8174821775761503,
+      "grad_norm": 0.07481568306684494,
+      "learning_rate": 1.0139970295692875e-05,
+      "loss": 0.2295,
+      "step": 47123
+    },
+    {
+      "epoch": 3.817563188593649,
+      "grad_norm": 0.07879388332366943,
+      "learning_rate": 1.013546964309825e-05,
+      "loss": 0.2554,
+      "step": 47124
+    },
+    {
+      "epoch": 3.817644199611147,
+      "grad_norm": 0.0720246359705925,
+      "learning_rate": 1.0130968990503624e-05,
+      "loss": 0.1888,
+      "step": 47125
+    },
+    {
+      "epoch": 3.8177252106286454,
+      "grad_norm": 0.08811241388320923,
+      "learning_rate": 1.0126468337908996e-05,
+      "loss": 0.2323,
+      "step": 47126
+    },
+    {
+      "epoch": 3.817806221646144,
+      "grad_norm": 0.06210675463080406,
+      "learning_rate": 1.0121967685314371e-05,
+      "loss": 0.2095,
+      "step": 47127
+    },
+    {
+      "epoch": 3.8178872326636424,
+      "grad_norm": 0.0800977572798729,
+      "learning_rate": 1.0117467032719745e-05,
+      "loss": 0.2031,
+      "step": 47128
+    },
+    {
+      "epoch": 3.8179682436811406,
+      "grad_norm": 0.06897516548633575,
+      "learning_rate": 1.0112966380125119e-05,
+      "loss": 0.2186,
+      "step": 47129
+    },
+    {
+      "epoch": 3.818049254698639,
+      "grad_norm": 0.06213768571615219,
+      "learning_rate": 1.0108465727530494e-05,
+      "loss": 0.229,
+      "step": 47130
+    },
+    {
+      "epoch": 3.818130265716137,
+      "grad_norm": 0.07513062655925751,
+      "learning_rate": 1.0103965074935866e-05,
+      "loss": 0.2606,
+      "step": 47131
+    },
+    {
+      "epoch": 3.818211276733636,
+      "grad_norm": 0.0878511592745781,
+      "learning_rate": 1.009946442234124e-05,
+      "loss": 0.2577,
+      "step": 47132
+    },
+    {
+      "epoch": 3.818292287751134,
+      "grad_norm": 0.10174962133169174,
+      "learning_rate": 1.0094963769746615e-05,
+      "loss": 0.2628,
+      "step": 47133
+    },
+    {
+      "epoch": 3.8183732987686323,
+      "grad_norm": 0.08247354626655579,
+      "learning_rate": 1.0090463117151988e-05,
+      "loss": 0.223,
+      "step": 47134
+    },
+    {
+      "epoch": 3.818454309786131,
+      "grad_norm": 0.10168582201004028,
+      "learning_rate": 1.008596246455736e-05,
+      "loss": 0.2132,
+      "step": 47135
+    },
+    {
+      "epoch": 3.8185353208036292,
+      "grad_norm": 0.06533244997262955,
+      "learning_rate": 1.0081461811962736e-05,
+      "loss": 0.2634,
+      "step": 47136
+    },
+    {
+      "epoch": 3.8186163318211275,
+      "grad_norm": 0.05962993949651718,
+      "learning_rate": 1.0076961159368109e-05,
+      "loss": 0.2111,
+      "step": 47137
+    },
+    {
+      "epoch": 3.818697342838626,
+      "grad_norm": 0.06415306776762009,
+      "learning_rate": 1.0072460506773483e-05,
+      "loss": 0.2213,
+      "step": 47138
+    },
+    {
+      "epoch": 3.8187783538561244,
+      "grad_norm": 0.0562756322324276,
+      "learning_rate": 1.0067959854178856e-05,
+      "loss": 0.19,
+      "step": 47139
+    },
+    {
+      "epoch": 3.8188593648736227,
+      "grad_norm": 0.06657944619655609,
+      "learning_rate": 1.006345920158423e-05,
+      "loss": 0.2282,
+      "step": 47140
+    },
+    {
+      "epoch": 3.8189403758911213,
+      "grad_norm": 0.06836859881877899,
+      "learning_rate": 1.0058958548989604e-05,
+      "loss": 0.2284,
+      "step": 47141
+    },
+    {
+      "epoch": 3.8190213869086196,
+      "grad_norm": 0.07675183564424515,
+      "learning_rate": 1.0054457896394979e-05,
+      "loss": 0.2459,
+      "step": 47142
+    },
+    {
+      "epoch": 3.819102397926118,
+      "grad_norm": 0.08422303944826126,
+      "learning_rate": 1.004995724380035e-05,
+      "loss": 0.2848,
+      "step": 47143
+    },
+    {
+      "epoch": 3.8191834089436165,
+      "grad_norm": 0.07882612198591232,
+      "learning_rate": 1.0045456591205724e-05,
+      "loss": 0.2573,
+      "step": 47144
+    },
+    {
+      "epoch": 3.8192644199611148,
+      "grad_norm": 0.058982573449611664,
+      "learning_rate": 1.00409559386111e-05,
+      "loss": 0.2202,
+      "step": 47145
+    },
+    {
+      "epoch": 3.819345430978613,
+      "grad_norm": 0.07594690471887589,
+      "learning_rate": 1.0036455286016473e-05,
+      "loss": 0.2612,
+      "step": 47146
+    },
+    {
+      "epoch": 3.8194264419961117,
+      "grad_norm": 0.06958582997322083,
+      "learning_rate": 1.0031954633421847e-05,
+      "loss": 0.1971,
+      "step": 47147
+    },
+    {
+      "epoch": 3.81950745301361,
+      "grad_norm": 0.09242573380470276,
+      "learning_rate": 1.002745398082722e-05,
+      "loss": 0.2689,
+      "step": 47148
+    },
+    {
+      "epoch": 3.819588464031108,
+      "grad_norm": 0.08315322548151016,
+      "learning_rate": 1.0022953328232594e-05,
+      "loss": 0.2526,
+      "step": 47149
+    },
+    {
+      "epoch": 3.819669475048607,
+      "grad_norm": 0.06824847310781479,
+      "learning_rate": 1.0018452675637968e-05,
+      "loss": 0.2435,
+      "step": 47150
+    },
+    {
+      "epoch": 3.819750486066105,
+      "grad_norm": 0.07796567678451538,
+      "learning_rate": 1.0013952023043343e-05,
+      "loss": 0.221,
+      "step": 47151
+    },
+    {
+      "epoch": 3.8198314970836034,
+      "grad_norm": 0.07060608267784119,
+      "learning_rate": 1.0009451370448715e-05,
+      "loss": 0.2133,
+      "step": 47152
+    },
+    {
+      "epoch": 3.8199125081011016,
+      "grad_norm": 0.07457584887742996,
+      "learning_rate": 1.0004950717854088e-05,
+      "loss": 0.2156,
+      "step": 47153
+    },
+    {
+      "epoch": 3.8199935191186,
+      "grad_norm": 0.09381989389657974,
+      "learning_rate": 1.0000450065259464e-05,
+      "loss": 0.2183,
+      "step": 47154
+    },
+    {
+      "epoch": 3.8200745301360985,
+      "grad_norm": 0.0633532926440239,
+      "learning_rate": 9.995949412664837e-06,
+      "loss": 0.2166,
+      "step": 47155
+    },
+    {
+      "epoch": 3.820155541153597,
+      "grad_norm": 0.06335076689720154,
+      "learning_rate": 9.99144876007021e-06,
+      "loss": 0.1908,
+      "step": 47156
+    },
+    {
+      "epoch": 3.820236552171095,
+      "grad_norm": 0.0725879818201065,
+      "learning_rate": 9.986948107475585e-06,
+      "loss": 0.2631,
+      "step": 47157
+    },
+    {
+      "epoch": 3.8203175631885937,
+      "grad_norm": 0.07099777460098267,
+      "learning_rate": 9.982447454880958e-06,
+      "loss": 0.2009,
+      "step": 47158
+    },
+    {
+      "epoch": 3.820398574206092,
+      "grad_norm": 0.08141002804040909,
+      "learning_rate": 9.977946802286332e-06,
+      "loss": 0.2486,
+      "step": 47159
+    },
+    {
+      "epoch": 3.82047958522359,
+      "grad_norm": 0.06455133110284805,
+      "learning_rate": 9.973446149691707e-06,
+      "loss": 0.2197,
+      "step": 47160
+    },
+    {
+      "epoch": 3.820560596241089,
+      "grad_norm": 0.08412550389766693,
+      "learning_rate": 9.968945497097079e-06,
+      "loss": 0.2283,
+      "step": 47161
+    },
+    {
+      "epoch": 3.820641607258587,
+      "grad_norm": 0.08144482970237732,
+      "learning_rate": 9.964444844502453e-06,
+      "loss": 0.198,
+      "step": 47162
+    },
+    {
+      "epoch": 3.8207226182760854,
+      "grad_norm": 0.06888513267040253,
+      "learning_rate": 9.959944191907828e-06,
+      "loss": 0.2045,
+      "step": 47163
+    },
+    {
+      "epoch": 3.820803629293584,
+      "grad_norm": 0.06686069816350937,
+      "learning_rate": 9.955443539313202e-06,
+      "loss": 0.1933,
+      "step": 47164
+    },
+    {
+      "epoch": 3.8208846403110823,
+      "grad_norm": 0.1000877320766449,
+      "learning_rate": 9.950942886718573e-06,
+      "loss": 0.2362,
+      "step": 47165
+    },
+    {
+      "epoch": 3.8209656513285806,
+      "grad_norm": 0.07413871586322784,
+      "learning_rate": 9.946442234123949e-06,
+      "loss": 0.2607,
+      "step": 47166
+    },
+    {
+      "epoch": 3.8210466623460793,
+      "grad_norm": 0.07616769522428513,
+      "learning_rate": 9.941941581529322e-06,
+      "loss": 0.2294,
+      "step": 47167
+    },
+    {
+      "epoch": 3.8211276733635775,
+      "grad_norm": 0.08212704211473465,
+      "learning_rate": 9.937440928934696e-06,
+      "loss": 0.208,
+      "step": 47168
+    },
+    {
+      "epoch": 3.8212086843810757,
+      "grad_norm": 0.06919477134943008,
+      "learning_rate": 9.93294027634007e-06,
+      "loss": 0.2393,
+      "step": 47169
+    },
+    {
+      "epoch": 3.8212896953985744,
+      "grad_norm": 0.07358191162347794,
+      "learning_rate": 9.928439623745443e-06,
+      "loss": 0.2176,
+      "step": 47170
+    },
+    {
+      "epoch": 3.8213707064160727,
+      "grad_norm": 0.0823577418923378,
+      "learning_rate": 9.923938971150817e-06,
+      "loss": 0.2783,
+      "step": 47171
+    },
+    {
+      "epoch": 3.821451717433571,
+      "grad_norm": 0.06688446551561356,
+      "learning_rate": 9.919438318556192e-06,
+      "loss": 0.2079,
+      "step": 47172
+    },
+    {
+      "epoch": 3.8215327284510696,
+      "grad_norm": 0.0666053518652916,
+      "learning_rate": 9.914937665961566e-06,
+      "loss": 0.2031,
+      "step": 47173
+    },
+    {
+      "epoch": 3.821613739468568,
+      "grad_norm": 0.08584258705377579,
+      "learning_rate": 9.910437013366938e-06,
+      "loss": 0.2129,
+      "step": 47174
+    },
+    {
+      "epoch": 3.821694750486066,
+      "grad_norm": 0.07315590977668762,
+      "learning_rate": 9.905936360772313e-06,
+      "loss": 0.2045,
+      "step": 47175
+    },
+    {
+      "epoch": 3.8217757615035644,
+      "grad_norm": 0.05562111362814903,
+      "learning_rate": 9.901435708177686e-06,
+      "loss": 0.181,
+      "step": 47176
+    },
+    {
+      "epoch": 3.8218567725210626,
+      "grad_norm": 0.07396908849477768,
+      "learning_rate": 9.89693505558306e-06,
+      "loss": 0.2028,
+      "step": 47177
+    },
+    {
+      "epoch": 3.8219377835385613,
+      "grad_norm": 0.0769689679145813,
+      "learning_rate": 9.892434402988434e-06,
+      "loss": 0.2081,
+      "step": 47178
+    },
+    {
+      "epoch": 3.8220187945560595,
+      "grad_norm": 0.09181632101535797,
+      "learning_rate": 9.887933750393807e-06,
+      "loss": 0.2465,
+      "step": 47179
+    },
+    {
+      "epoch": 3.8220998055735578,
+      "grad_norm": 0.07666897028684616,
+      "learning_rate": 9.883433097799181e-06,
+      "loss": 0.2228,
+      "step": 47180
+    },
+    {
+      "epoch": 3.8221808165910565,
+      "grad_norm": 0.06831949949264526,
+      "learning_rate": 9.878932445204556e-06,
+      "loss": 0.2263,
+      "step": 47181
+    },
+    {
+      "epoch": 3.8222618276085547,
+      "grad_norm": 0.06521642953157425,
+      "learning_rate": 9.874431792609928e-06,
+      "loss": 0.2294,
+      "step": 47182
+    },
+    {
+      "epoch": 3.822342838626053,
+      "grad_norm": 0.07536672800779343,
+      "learning_rate": 9.869931140015302e-06,
+      "loss": 0.2143,
+      "step": 47183
+    },
+    {
+      "epoch": 3.8224238496435516,
+      "grad_norm": 0.08660756051540375,
+      "learning_rate": 9.865430487420677e-06,
+      "loss": 0.2777,
+      "step": 47184
+    },
+    {
+      "epoch": 3.82250486066105,
+      "grad_norm": 0.0728485956788063,
+      "learning_rate": 9.86092983482605e-06,
+      "loss": 0.2351,
+      "step": 47185
+    },
+    {
+      "epoch": 3.822585871678548,
+      "grad_norm": 0.06775187700986862,
+      "learning_rate": 9.856429182231424e-06,
+      "loss": 0.2358,
+      "step": 47186
+    },
+    {
+      "epoch": 3.822666882696047,
+      "grad_norm": 0.07535045593976974,
+      "learning_rate": 9.851928529636798e-06,
+      "loss": 0.2201,
+      "step": 47187
+    },
+    {
+      "epoch": 3.822747893713545,
+      "grad_norm": 0.07898474484682083,
+      "learning_rate": 9.847427877042171e-06,
+      "loss": 0.2667,
+      "step": 47188
+    },
+    {
+      "epoch": 3.8228289047310433,
+      "grad_norm": 0.08030825108289719,
+      "learning_rate": 9.842927224447545e-06,
+      "loss": 0.2676,
+      "step": 47189
+    },
+    {
+      "epoch": 3.822909915748542,
+      "grad_norm": 0.07572636753320694,
+      "learning_rate": 9.83842657185292e-06,
+      "loss": 0.2874,
+      "step": 47190
+    },
+    {
+      "epoch": 3.8229909267660402,
+      "grad_norm": 0.07902305573225021,
+      "learning_rate": 9.833925919258292e-06,
+      "loss": 0.2218,
+      "step": 47191
+    },
+    {
+      "epoch": 3.8230719377835385,
+      "grad_norm": 0.07862576842308044,
+      "learning_rate": 9.829425266663666e-06,
+      "loss": 0.2515,
+      "step": 47192
+    },
+    {
+      "epoch": 3.823152948801037,
+      "grad_norm": 0.07469526678323746,
+      "learning_rate": 9.824924614069041e-06,
+      "loss": 0.221,
+      "step": 47193
+    },
+    {
+      "epoch": 3.8232339598185354,
+      "grad_norm": 0.07735791057348251,
+      "learning_rate": 9.820423961474415e-06,
+      "loss": 0.2396,
+      "step": 47194
+    },
+    {
+      "epoch": 3.8233149708360337,
+      "grad_norm": 0.06498521566390991,
+      "learning_rate": 9.815923308879788e-06,
+      "loss": 0.2311,
+      "step": 47195
+    },
+    {
+      "epoch": 3.8233959818535324,
+      "grad_norm": 0.07805728167295456,
+      "learning_rate": 9.811422656285162e-06,
+      "loss": 0.2454,
+      "step": 47196
+    },
+    {
+      "epoch": 3.8234769928710306,
+      "grad_norm": 0.065180204808712,
+      "learning_rate": 9.806922003690536e-06,
+      "loss": 0.2308,
+      "step": 47197
+    },
+    {
+      "epoch": 3.823558003888529,
+      "grad_norm": 0.07017407566308975,
+      "learning_rate": 9.802421351095909e-06,
+      "loss": 0.2326,
+      "step": 47198
+    },
+    {
+      "epoch": 3.823639014906027,
+      "grad_norm": 0.08247072994709015,
+      "learning_rate": 9.797920698501283e-06,
+      "loss": 0.23,
+      "step": 47199
+    },
+    {
+      "epoch": 3.8237200259235253,
+      "grad_norm": 0.06262676417827606,
+      "learning_rate": 9.793420045906656e-06,
+      "loss": 0.2121,
+      "step": 47200
+    },
+    {
+      "epoch": 3.823801036941024,
+      "grad_norm": 0.10090301930904388,
+      "learning_rate": 9.78891939331203e-06,
+      "loss": 0.2373,
+      "step": 47201
+    },
+    {
+      "epoch": 3.8238820479585223,
+      "grad_norm": 0.08574521541595459,
+      "learning_rate": 9.784418740717405e-06,
+      "loss": 0.2366,
+      "step": 47202
+    },
+    {
+      "epoch": 3.8239630589760205,
+      "grad_norm": 0.0797303095459938,
+      "learning_rate": 9.779918088122779e-06,
+      "loss": 0.2047,
+      "step": 47203
+    },
+    {
+      "epoch": 3.824044069993519,
+      "grad_norm": 0.07990000396966934,
+      "learning_rate": 9.775417435528152e-06,
+      "loss": 0.194,
+      "step": 47204
+    },
+    {
+      "epoch": 3.8241250810110174,
+      "grad_norm": 0.06872112303972244,
+      "learning_rate": 9.770916782933526e-06,
+      "loss": 0.1983,
+      "step": 47205
+    },
+    {
+      "epoch": 3.8242060920285157,
+      "grad_norm": 0.06418676674365997,
+      "learning_rate": 9.7664161303389e-06,
+      "loss": 0.2198,
+      "step": 47206
+    },
+    {
+      "epoch": 3.8242871030460144,
+      "grad_norm": 0.0675126165151596,
+      "learning_rate": 9.761915477744273e-06,
+      "loss": 0.191,
+      "step": 47207
+    },
+    {
+      "epoch": 3.8243681140635126,
+      "grad_norm": 0.0721009373664856,
+      "learning_rate": 9.757414825149647e-06,
+      "loss": 0.2582,
+      "step": 47208
+    },
+    {
+      "epoch": 3.824449125081011,
+      "grad_norm": 0.05475969240069389,
+      "learning_rate": 9.75291417255502e-06,
+      "loss": 0.1922,
+      "step": 47209
+    },
+    {
+      "epoch": 3.8245301360985096,
+      "grad_norm": 0.07951171696186066,
+      "learning_rate": 9.748413519960394e-06,
+      "loss": 0.2298,
+      "step": 47210
+    },
+    {
+      "epoch": 3.824611147116008,
+      "grad_norm": 0.059227924793958664,
+      "learning_rate": 9.74391286736577e-06,
+      "loss": 0.2276,
+      "step": 47211
+    },
+    {
+      "epoch": 3.824692158133506,
+      "grad_norm": 0.06179160252213478,
+      "learning_rate": 9.739412214771141e-06,
+      "loss": 0.2209,
+      "step": 47212
+    },
+    {
+      "epoch": 3.8247731691510047,
+      "grad_norm": 0.06057712808251381,
+      "learning_rate": 9.734911562176517e-06,
+      "loss": 0.1778,
+      "step": 47213
+    },
+    {
+      "epoch": 3.824854180168503,
+      "grad_norm": 0.06524763256311417,
+      "learning_rate": 9.73041090958189e-06,
+      "loss": 0.2312,
+      "step": 47214
+    },
+    {
+      "epoch": 3.8249351911860012,
+      "grad_norm": 0.06625210493803024,
+      "learning_rate": 9.725910256987264e-06,
+      "loss": 0.2464,
+      "step": 47215
+    },
+    {
+      "epoch": 3.8250162022035,
+      "grad_norm": 0.05672033131122589,
+      "learning_rate": 9.721409604392637e-06,
+      "loss": 0.2107,
+      "step": 47216
+    },
+    {
+      "epoch": 3.825097213220998,
+      "grad_norm": 0.08716113865375519,
+      "learning_rate": 9.716908951798011e-06,
+      "loss": 0.2429,
+      "step": 47217
+    },
+    {
+      "epoch": 3.8251782242384964,
+      "grad_norm": 0.09075607359409332,
+      "learning_rate": 9.712408299203385e-06,
+      "loss": 0.2329,
+      "step": 47218
+    },
+    {
+      "epoch": 3.8252592352559946,
+      "grad_norm": 0.08668406307697296,
+      "learning_rate": 9.707907646608758e-06,
+      "loss": 0.2409,
+      "step": 47219
+    },
+    {
+      "epoch": 3.8253402462734933,
+      "grad_norm": 0.07025102525949478,
+      "learning_rate": 9.703406994014133e-06,
+      "loss": 0.1961,
+      "step": 47220
+    },
+    {
+      "epoch": 3.8254212572909916,
+      "grad_norm": 0.05928102135658264,
+      "learning_rate": 9.698906341419505e-06,
+      "loss": 0.2193,
+      "step": 47221
+    },
+    {
+      "epoch": 3.82550226830849,
+      "grad_norm": 0.0666092187166214,
+      "learning_rate": 9.69440568882488e-06,
+      "loss": 0.2267,
+      "step": 47222
+    },
+    {
+      "epoch": 3.825583279325988,
+      "grad_norm": 0.07980141043663025,
+      "learning_rate": 9.689905036230254e-06,
+      "loss": 0.232,
+      "step": 47223
+    },
+    {
+      "epoch": 3.8256642903434868,
+      "grad_norm": 0.09296827018260956,
+      "learning_rate": 9.685404383635628e-06,
+      "loss": 0.2557,
+      "step": 47224
+    },
+    {
+      "epoch": 3.825745301360985,
+      "grad_norm": 0.08211761713027954,
+      "learning_rate": 9.680903731041001e-06,
+      "loss": 0.2107,
+      "step": 47225
+    },
+    {
+      "epoch": 3.8258263123784833,
+      "grad_norm": 0.0850875973701477,
+      "learning_rate": 9.676403078446375e-06,
+      "loss": 0.2087,
+      "step": 47226
+    },
+    {
+      "epoch": 3.825907323395982,
+      "grad_norm": 0.0714445635676384,
+      "learning_rate": 9.671902425851749e-06,
+      "loss": 0.2302,
+      "step": 47227
+    },
+    {
+      "epoch": 3.82598833441348,
+      "grad_norm": 0.08113661408424377,
+      "learning_rate": 9.667401773257124e-06,
+      "loss": 0.2145,
+      "step": 47228
+    },
+    {
+      "epoch": 3.8260693454309784,
+      "grad_norm": 0.0769774541258812,
+      "learning_rate": 9.662901120662496e-06,
+      "loss": 0.2095,
+      "step": 47229
+    },
+    {
+      "epoch": 3.826150356448477,
+      "grad_norm": 0.0717720314860344,
+      "learning_rate": 9.65840046806787e-06,
+      "loss": 0.2377,
+      "step": 47230
+    },
+    {
+      "epoch": 3.8262313674659754,
+      "grad_norm": 0.06568815559148788,
+      "learning_rate": 9.653899815473245e-06,
+      "loss": 0.2288,
+      "step": 47231
+    },
+    {
+      "epoch": 3.8263123784834736,
+      "grad_norm": 0.0737098678946495,
+      "learning_rate": 9.649399162878618e-06,
+      "loss": 0.2059,
+      "step": 47232
+    },
+    {
+      "epoch": 3.8263933895009723,
+      "grad_norm": 0.06918147206306458,
+      "learning_rate": 9.644898510283992e-06,
+      "loss": 0.2238,
+      "step": 47233
+    },
+    {
+      "epoch": 3.8264744005184705,
+      "grad_norm": 0.0679457038640976,
+      "learning_rate": 9.640397857689366e-06,
+      "loss": 0.2414,
+      "step": 47234
+    },
+    {
+      "epoch": 3.826555411535969,
+      "grad_norm": 0.07485439628362656,
+      "learning_rate": 9.63589720509474e-06,
+      "loss": 0.1962,
+      "step": 47235
+    },
+    {
+      "epoch": 3.8266364225534675,
+      "grad_norm": 0.07543614506721497,
+      "learning_rate": 9.631396552500113e-06,
+      "loss": 0.2554,
+      "step": 47236
+    },
+    {
+      "epoch": 3.8267174335709657,
+      "grad_norm": 0.07579874992370605,
+      "learning_rate": 9.626895899905488e-06,
+      "loss": 0.1838,
+      "step": 47237
+    },
+    {
+      "epoch": 3.826798444588464,
+      "grad_norm": 0.06702306121587753,
+      "learning_rate": 9.62239524731086e-06,
+      "loss": 0.246,
+      "step": 47238
+    },
+    {
+      "epoch": 3.8268794556059627,
+      "grad_norm": 0.06871180236339569,
+      "learning_rate": 9.617894594716234e-06,
+      "loss": 0.232,
+      "step": 47239
+    },
+    {
+      "epoch": 3.826960466623461,
+      "grad_norm": 0.0695720836520195,
+      "learning_rate": 9.613393942121609e-06,
+      "loss": 0.2181,
+      "step": 47240
+    },
+    {
+      "epoch": 3.827041477640959,
+      "grad_norm": 0.07191573828458786,
+      "learning_rate": 9.608893289526983e-06,
+      "loss": 0.2541,
+      "step": 47241
+    },
+    {
+      "epoch": 3.8271224886584574,
+      "grad_norm": 0.07103823870420456,
+      "learning_rate": 9.604392636932354e-06,
+      "loss": 0.2188,
+      "step": 47242
+    },
+    {
+      "epoch": 3.827203499675956,
+      "grad_norm": 0.057614583522081375,
+      "learning_rate": 9.59989198433773e-06,
+      "loss": 0.247,
+      "step": 47243
+    },
+    {
+      "epoch": 3.8272845106934543,
+      "grad_norm": 0.08780107647180557,
+      "learning_rate": 9.595391331743103e-06,
+      "loss": 0.2499,
+      "step": 47244
+    },
+    {
+      "epoch": 3.8273655217109526,
+      "grad_norm": 0.07042431086301804,
+      "learning_rate": 9.590890679148477e-06,
+      "loss": 0.2408,
+      "step": 47245
+    },
+    {
+      "epoch": 3.827446532728451,
+      "grad_norm": 0.09787672013044357,
+      "learning_rate": 9.586390026553852e-06,
+      "loss": 0.233,
+      "step": 47246
+    },
+    {
+      "epoch": 3.8275275437459495,
+      "grad_norm": 0.05550922080874443,
+      "learning_rate": 9.581889373959224e-06,
+      "loss": 0.254,
+      "step": 47247
+    },
+    {
+      "epoch": 3.8276085547634477,
+      "grad_norm": 0.06615682691335678,
+      "learning_rate": 9.577388721364598e-06,
+      "loss": 0.2103,
+      "step": 47248
+    },
+    {
+      "epoch": 3.827689565780946,
+      "grad_norm": 0.0725463330745697,
+      "learning_rate": 9.572888068769973e-06,
+      "loss": 0.2129,
+      "step": 47249
+    },
+    {
+      "epoch": 3.8277705767984447,
+      "grad_norm": 0.08158998191356659,
+      "learning_rate": 9.568387416175347e-06,
+      "loss": 0.2423,
+      "step": 47250
+    },
+    {
+      "epoch": 3.827851587815943,
+      "grad_norm": 0.07956718653440475,
+      "learning_rate": 9.563886763580719e-06,
+      "loss": 0.2078,
+      "step": 47251
+    },
+    {
+      "epoch": 3.827932598833441,
+      "grad_norm": 0.07327364385128021,
+      "learning_rate": 9.559386110986094e-06,
+      "loss": 0.2675,
+      "step": 47252
+    },
+    {
+      "epoch": 3.82801360985094,
+      "grad_norm": 0.0754021480679512,
+      "learning_rate": 9.554885458391467e-06,
+      "loss": 0.2211,
+      "step": 47253
+    },
+    {
+      "epoch": 3.828094620868438,
+      "grad_norm": 0.07733095437288284,
+      "learning_rate": 9.550384805796841e-06,
+      "loss": 0.2136,
+      "step": 47254
+    },
+    {
+      "epoch": 3.8281756318859363,
+      "grad_norm": 0.0626547783613205,
+      "learning_rate": 9.545884153202215e-06,
+      "loss": 0.2049,
+      "step": 47255
+    },
+    {
+      "epoch": 3.828256642903435,
+      "grad_norm": 0.0786009356379509,
+      "learning_rate": 9.541383500607588e-06,
+      "loss": 0.226,
+      "step": 47256
+    },
+    {
+      "epoch": 3.8283376539209333,
+      "grad_norm": 0.07420649379491806,
+      "learning_rate": 9.536882848012962e-06,
+      "loss": 0.1857,
+      "step": 47257
+    },
+    {
+      "epoch": 3.8284186649384315,
+      "grad_norm": 0.08987807482481003,
+      "learning_rate": 9.532382195418337e-06,
+      "loss": 0.2385,
+      "step": 47258
+    },
+    {
+      "epoch": 3.82849967595593,
+      "grad_norm": 0.08396057784557343,
+      "learning_rate": 9.527881542823709e-06,
+      "loss": 0.227,
+      "step": 47259
+    },
+    {
+      "epoch": 3.8285806869734285,
+      "grad_norm": 0.08503457903862,
+      "learning_rate": 9.523380890229083e-06,
+      "loss": 0.2937,
+      "step": 47260
+    },
+    {
+      "epoch": 3.8286616979909267,
+      "grad_norm": 0.07922942191362381,
+      "learning_rate": 9.518880237634458e-06,
+      "loss": 0.2653,
+      "step": 47261
+    },
+    {
+      "epoch": 3.8287427090084254,
+      "grad_norm": 0.07605444639921188,
+      "learning_rate": 9.514379585039832e-06,
+      "loss": 0.2179,
+      "step": 47262
+    },
+    {
+      "epoch": 3.8288237200259236,
+      "grad_norm": 0.062154121696949005,
+      "learning_rate": 9.509878932445205e-06,
+      "loss": 0.2167,
+      "step": 47263
+    },
+    {
+      "epoch": 3.828904731043422,
+      "grad_norm": 0.059175748378038406,
+      "learning_rate": 9.505378279850579e-06,
+      "loss": 0.2376,
+      "step": 47264
+    },
+    {
+      "epoch": 3.82898574206092,
+      "grad_norm": 0.06250623613595963,
+      "learning_rate": 9.500877627255952e-06,
+      "loss": 0.2287,
+      "step": 47265
+    },
+    {
+      "epoch": 3.829066753078419,
+      "grad_norm": 0.06687180697917938,
+      "learning_rate": 9.496376974661326e-06,
+      "loss": 0.2141,
+      "step": 47266
+    },
+    {
+      "epoch": 3.829147764095917,
+      "grad_norm": 0.08554664254188538,
+      "learning_rate": 9.491876322066701e-06,
+      "loss": 0.2137,
+      "step": 47267
+    },
+    {
+      "epoch": 3.8292287751134153,
+      "grad_norm": 0.07720212638378143,
+      "learning_rate": 9.487375669472073e-06,
+      "loss": 0.2481,
+      "step": 47268
+    },
+    {
+      "epoch": 3.8293097861309136,
+      "grad_norm": 0.07368236035108566,
+      "learning_rate": 9.482875016877447e-06,
+      "loss": 0.1931,
+      "step": 47269
+    },
+    {
+      "epoch": 3.8293907971484122,
+      "grad_norm": 0.07523944973945618,
+      "learning_rate": 9.478374364282822e-06,
+      "loss": 0.198,
+      "step": 47270
+    },
+    {
+      "epoch": 3.8294718081659105,
+      "grad_norm": 0.07883183658123016,
+      "learning_rate": 9.473873711688196e-06,
+      "loss": 0.2222,
+      "step": 47271
+    },
+    {
+      "epoch": 3.8295528191834087,
+      "grad_norm": 0.06268235296010971,
+      "learning_rate": 9.469373059093568e-06,
+      "loss": 0.2399,
+      "step": 47272
+    },
+    {
+      "epoch": 3.8296338302009074,
+      "grad_norm": 0.06675557792186737,
+      "learning_rate": 9.464872406498943e-06,
+      "loss": 0.2021,
+      "step": 47273
+    },
+    {
+      "epoch": 3.8297148412184057,
+      "grad_norm": 0.07962408661842346,
+      "learning_rate": 9.460371753904317e-06,
+      "loss": 0.2065,
+      "step": 47274
+    },
+    {
+      "epoch": 3.829795852235904,
+      "grad_norm": 0.06950665265321732,
+      "learning_rate": 9.45587110130969e-06,
+      "loss": 0.2035,
+      "step": 47275
+    },
+    {
+      "epoch": 3.8298768632534026,
+      "grad_norm": 0.07848667353391647,
+      "learning_rate": 9.451370448715065e-06,
+      "loss": 0.2406,
+      "step": 47276
+    },
+    {
+      "epoch": 3.829957874270901,
+      "grad_norm": 0.07826760411262512,
+      "learning_rate": 9.446869796120437e-06,
+      "loss": 0.2433,
+      "step": 47277
+    },
+    {
+      "epoch": 3.830038885288399,
+      "grad_norm": 0.06589142233133316,
+      "learning_rate": 9.442369143525811e-06,
+      "loss": 0.2277,
+      "step": 47278
+    },
+    {
+      "epoch": 3.8301198963058978,
+      "grad_norm": 0.0657556802034378,
+      "learning_rate": 9.437868490931186e-06,
+      "loss": 0.2145,
+      "step": 47279
+    },
+    {
+      "epoch": 3.830200907323396,
+      "grad_norm": 0.08655353635549545,
+      "learning_rate": 9.43336783833656e-06,
+      "loss": 0.2212,
+      "step": 47280
+    },
+    {
+      "epoch": 3.8302819183408943,
+      "grad_norm": 0.06460024416446686,
+      "learning_rate": 9.428867185741932e-06,
+      "loss": 0.2437,
+      "step": 47281
+    },
+    {
+      "epoch": 3.830362929358393,
+      "grad_norm": 0.06946887820959091,
+      "learning_rate": 9.424366533147307e-06,
+      "loss": 0.2367,
+      "step": 47282
+    },
+    {
+      "epoch": 3.830443940375891,
+      "grad_norm": 0.07267460972070694,
+      "learning_rate": 9.41986588055268e-06,
+      "loss": 0.2402,
+      "step": 47283
+    },
+    {
+      "epoch": 3.8305249513933894,
+      "grad_norm": 0.06545892357826233,
+      "learning_rate": 9.415365227958054e-06,
+      "loss": 0.243,
+      "step": 47284
+    },
+    {
+      "epoch": 3.830605962410888,
+      "grad_norm": 0.07193044573068619,
+      "learning_rate": 9.410864575363428e-06,
+      "loss": 0.2309,
+      "step": 47285
+    },
+    {
+      "epoch": 3.8306869734283864,
+      "grad_norm": 0.07626271992921829,
+      "learning_rate": 9.406363922768801e-06,
+      "loss": 0.2003,
+      "step": 47286
+    },
+    {
+      "epoch": 3.8307679844458846,
+      "grad_norm": 0.06547614932060242,
+      "learning_rate": 9.401863270174175e-06,
+      "loss": 0.2529,
+      "step": 47287
+    },
+    {
+      "epoch": 3.830848995463383,
+      "grad_norm": 0.07183431833982468,
+      "learning_rate": 9.39736261757955e-06,
+      "loss": 0.1978,
+      "step": 47288
+    },
+    {
+      "epoch": 3.8309300064808816,
+      "grad_norm": 0.06502663344144821,
+      "learning_rate": 9.392861964984924e-06,
+      "loss": 0.2103,
+      "step": 47289
+    },
+    {
+      "epoch": 3.83101101749838,
+      "grad_norm": 0.0723065510392189,
+      "learning_rate": 9.388361312390296e-06,
+      "loss": 0.2106,
+      "step": 47290
+    },
+    {
+      "epoch": 3.831092028515878,
+      "grad_norm": 0.06067093834280968,
+      "learning_rate": 9.383860659795671e-06,
+      "loss": 0.2475,
+      "step": 47291
+    },
+    {
+      "epoch": 3.8311730395333763,
+      "grad_norm": 0.0675438940525055,
+      "learning_rate": 9.379360007201045e-06,
+      "loss": 0.2143,
+      "step": 47292
+    },
+    {
+      "epoch": 3.831254050550875,
+      "grad_norm": 0.08510125428438187,
+      "learning_rate": 9.374859354606418e-06,
+      "loss": 0.2459,
+      "step": 47293
+    },
+    {
+      "epoch": 3.8313350615683732,
+      "grad_norm": 0.08360667526721954,
+      "learning_rate": 9.370358702011792e-06,
+      "loss": 0.2058,
+      "step": 47294
+    },
+    {
+      "epoch": 3.8314160725858715,
+      "grad_norm": 0.06987543404102325,
+      "learning_rate": 9.365858049417166e-06,
+      "loss": 0.2325,
+      "step": 47295
+    },
+    {
+      "epoch": 3.83149708360337,
+      "grad_norm": 0.07574071735143661,
+      "learning_rate": 9.36135739682254e-06,
+      "loss": 0.2243,
+      "step": 47296
+    },
+    {
+      "epoch": 3.8315780946208684,
+      "grad_norm": 0.06666197627782822,
+      "learning_rate": 9.356856744227914e-06,
+      "loss": 0.2158,
+      "step": 47297
+    },
+    {
+      "epoch": 3.8316591056383666,
+      "grad_norm": 0.08488959819078445,
+      "learning_rate": 9.352356091633286e-06,
+      "loss": 0.2025,
+      "step": 47298
+    },
+    {
+      "epoch": 3.8317401166558653,
+      "grad_norm": 0.07799491286277771,
+      "learning_rate": 9.34785543903866e-06,
+      "loss": 0.209,
+      "step": 47299
+    },
+    {
+      "epoch": 3.8318211276733636,
+      "grad_norm": 0.07381635904312134,
+      "learning_rate": 9.343354786444035e-06,
+      "loss": 0.2521,
+      "step": 47300
+    },
+    {
+      "epoch": 3.831902138690862,
+      "grad_norm": 0.05815286561846733,
+      "learning_rate": 9.338854133849409e-06,
+      "loss": 0.2023,
+      "step": 47301
+    },
+    {
+      "epoch": 3.8319831497083605,
+      "grad_norm": 0.08299100399017334,
+      "learning_rate": 9.334353481254782e-06,
+      "loss": 0.1874,
+      "step": 47302
+    },
+    {
+      "epoch": 3.8320641607258588,
+      "grad_norm": 0.0680263414978981,
+      "learning_rate": 9.329852828660156e-06,
+      "loss": 0.2688,
+      "step": 47303
+    },
+    {
+      "epoch": 3.832145171743357,
+      "grad_norm": 0.07331136614084244,
+      "learning_rate": 9.32535217606553e-06,
+      "loss": 0.2756,
+      "step": 47304
+    },
+    {
+      "epoch": 3.8322261827608557,
+      "grad_norm": 0.07893449068069458,
+      "learning_rate": 9.320851523470903e-06,
+      "loss": 0.2511,
+      "step": 47305
+    },
+    {
+      "epoch": 3.832307193778354,
+      "grad_norm": 0.06692652404308319,
+      "learning_rate": 9.316350870876279e-06,
+      "loss": 0.1834,
+      "step": 47306
+    },
+    {
+      "epoch": 3.832388204795852,
+      "grad_norm": 0.10165359824895859,
+      "learning_rate": 9.31185021828165e-06,
+      "loss": 0.2666,
+      "step": 47307
+    },
+    {
+      "epoch": 3.832469215813351,
+      "grad_norm": 0.06731588393449783,
+      "learning_rate": 9.307349565687024e-06,
+      "loss": 0.2005,
+      "step": 47308
+    },
+    {
+      "epoch": 3.832550226830849,
+      "grad_norm": 0.09456529468297958,
+      "learning_rate": 9.3028489130924e-06,
+      "loss": 0.2468,
+      "step": 47309
+    },
+    {
+      "epoch": 3.8326312378483474,
+      "grad_norm": 0.07092951238155365,
+      "learning_rate": 9.298348260497773e-06,
+      "loss": 0.2123,
+      "step": 47310
+    },
+    {
+      "epoch": 3.8327122488658456,
+      "grad_norm": 0.07524535059928894,
+      "learning_rate": 9.293847607903147e-06,
+      "loss": 0.227,
+      "step": 47311
+    },
+    {
+      "epoch": 3.8327932598833443,
+      "grad_norm": 0.0925503671169281,
+      "learning_rate": 9.28934695530852e-06,
+      "loss": 0.2284,
+      "step": 47312
+    },
+    {
+      "epoch": 3.8328742709008425,
+      "grad_norm": 0.0901738703250885,
+      "learning_rate": 9.284846302713894e-06,
+      "loss": 0.2441,
+      "step": 47313
+    },
+    {
+      "epoch": 3.832955281918341,
+      "grad_norm": 0.06687404960393906,
+      "learning_rate": 9.280345650119267e-06,
+      "loss": 0.2427,
+      "step": 47314
+    },
+    {
+      "epoch": 3.833036292935839,
+      "grad_norm": 0.07051212340593338,
+      "learning_rate": 9.275844997524641e-06,
+      "loss": 0.2208,
+      "step": 47315
+    },
+    {
+      "epoch": 3.8331173039533377,
+      "grad_norm": 0.08045321702957153,
+      "learning_rate": 9.271344344930015e-06,
+      "loss": 0.2244,
+      "step": 47316
+    },
+    {
+      "epoch": 3.833198314970836,
+      "grad_norm": 0.06755134463310242,
+      "learning_rate": 9.266843692335388e-06,
+      "loss": 0.1857,
+      "step": 47317
+    },
+    {
+      "epoch": 3.833279325988334,
+      "grad_norm": 0.06752123683691025,
+      "learning_rate": 9.262343039740764e-06,
+      "loss": 0.2187,
+      "step": 47318
+    },
+    {
+      "epoch": 3.833360337005833,
+      "grad_norm": 0.07143384963274002,
+      "learning_rate": 9.257842387146137e-06,
+      "loss": 0.2296,
+      "step": 47319
+    },
+    {
+      "epoch": 3.833441348023331,
+      "grad_norm": 0.06815242767333984,
+      "learning_rate": 9.25334173455151e-06,
+      "loss": 0.1858,
+      "step": 47320
+    },
+    {
+      "epoch": 3.8335223590408294,
+      "grad_norm": 0.08003578335046768,
+      "learning_rate": 9.248841081956884e-06,
+      "loss": 0.2072,
+      "step": 47321
+    },
+    {
+      "epoch": 3.833603370058328,
+      "grad_norm": 0.07147976011037827,
+      "learning_rate": 9.244340429362258e-06,
+      "loss": 0.2295,
+      "step": 47322
+    },
+    {
+      "epoch": 3.8336843810758263,
+      "grad_norm": 0.07325585186481476,
+      "learning_rate": 9.239839776767632e-06,
+      "loss": 0.2423,
+      "step": 47323
+    },
+    {
+      "epoch": 3.8337653920933246,
+      "grad_norm": 0.06136981397867203,
+      "learning_rate": 9.235339124173005e-06,
+      "loss": 0.1829,
+      "step": 47324
+    },
+    {
+      "epoch": 3.8338464031108233,
+      "grad_norm": 0.07087627798318863,
+      "learning_rate": 9.230838471578379e-06,
+      "loss": 0.1777,
+      "step": 47325
+    },
+    {
+      "epoch": 3.8339274141283215,
+      "grad_norm": 0.07408089935779572,
+      "learning_rate": 9.226337818983752e-06,
+      "loss": 0.2293,
+      "step": 47326
+    },
+    {
+      "epoch": 3.8340084251458197,
+      "grad_norm": 0.08621295541524887,
+      "learning_rate": 9.221837166389128e-06,
+      "loss": 0.2657,
+      "step": 47327
+    },
+    {
+      "epoch": 3.8340894361633184,
+      "grad_norm": 0.06372913718223572,
+      "learning_rate": 9.2173365137945e-06,
+      "loss": 0.2012,
+      "step": 47328
+    },
+    {
+      "epoch": 3.8341704471808167,
+      "grad_norm": 0.07300638407468796,
+      "learning_rate": 9.212835861199875e-06,
+      "loss": 0.2104,
+      "step": 47329
+    },
+    {
+      "epoch": 3.834251458198315,
+      "grad_norm": 0.05497792363166809,
+      "learning_rate": 9.208335208605248e-06,
+      "loss": 0.2106,
+      "step": 47330
+    },
+    {
+      "epoch": 3.8343324692158136,
+      "grad_norm": 0.06223801150918007,
+      "learning_rate": 9.203834556010622e-06,
+      "loss": 0.1981,
+      "step": 47331
+    },
+    {
+      "epoch": 3.834413480233312,
+      "grad_norm": 0.08193052560091019,
+      "learning_rate": 9.199333903415996e-06,
+      "loss": 0.2798,
+      "step": 47332
+    },
+    {
+      "epoch": 3.83449449125081,
+      "grad_norm": 0.07334339618682861,
+      "learning_rate": 9.19483325082137e-06,
+      "loss": 0.2237,
+      "step": 47333
+    },
+    {
+      "epoch": 3.8345755022683083,
+      "grad_norm": 0.07783214747905731,
+      "learning_rate": 9.190332598226743e-06,
+      "loss": 0.2113,
+      "step": 47334
+    },
+    {
+      "epoch": 3.834656513285807,
+      "grad_norm": 0.07094062864780426,
+      "learning_rate": 9.185831945632118e-06,
+      "loss": 0.2396,
+      "step": 47335
+    },
+    {
+      "epoch": 3.8347375243033053,
+      "grad_norm": 0.07203572988510132,
+      "learning_rate": 9.181331293037492e-06,
+      "loss": 0.2329,
+      "step": 47336
+    },
+    {
+      "epoch": 3.8348185353208035,
+      "grad_norm": 0.06293873488903046,
+      "learning_rate": 9.176830640442864e-06,
+      "loss": 0.2344,
+      "step": 47337
+    },
+    {
+      "epoch": 3.8348995463383018,
+      "grad_norm": 0.07198001444339752,
+      "learning_rate": 9.172329987848239e-06,
+      "loss": 0.2002,
+      "step": 47338
+    },
+    {
+      "epoch": 3.8349805573558005,
+      "grad_norm": 0.07168415933847427,
+      "learning_rate": 9.167829335253613e-06,
+      "loss": 0.2593,
+      "step": 47339
+    },
+    {
+      "epoch": 3.8350615683732987,
+      "grad_norm": 0.06184476613998413,
+      "learning_rate": 9.163328682658986e-06,
+      "loss": 0.2313,
+      "step": 47340
+    },
+    {
+      "epoch": 3.835142579390797,
+      "grad_norm": 0.05653228610754013,
+      "learning_rate": 9.15882803006436e-06,
+      "loss": 0.1854,
+      "step": 47341
+    },
+    {
+      "epoch": 3.8352235904082956,
+      "grad_norm": 0.06754690408706665,
+      "learning_rate": 9.154327377469733e-06,
+      "loss": 0.2537,
+      "step": 47342
+    },
+    {
+      "epoch": 3.835304601425794,
+      "grad_norm": 0.07572982460260391,
+      "learning_rate": 9.149826724875107e-06,
+      "loss": 0.264,
+      "step": 47343
+    },
+    {
+      "epoch": 3.835385612443292,
+      "grad_norm": 0.07662025839090347,
+      "learning_rate": 9.145326072280482e-06,
+      "loss": 0.2123,
+      "step": 47344
+    },
+    {
+      "epoch": 3.835466623460791,
+      "grad_norm": 0.07293036580085754,
+      "learning_rate": 9.140825419685854e-06,
+      "loss": 0.231,
+      "step": 47345
+    },
+    {
+      "epoch": 3.835547634478289,
+      "grad_norm": 0.061868395656347275,
+      "learning_rate": 9.136324767091228e-06,
+      "loss": 0.2088,
+      "step": 47346
+    },
+    {
+      "epoch": 3.8356286454957873,
+      "grad_norm": 0.05718451365828514,
+      "learning_rate": 9.131824114496603e-06,
+      "loss": 0.2455,
+      "step": 47347
+    },
+    {
+      "epoch": 3.835709656513286,
+      "grad_norm": 0.06627894937992096,
+      "learning_rate": 9.127323461901977e-06,
+      "loss": 0.2288,
+      "step": 47348
+    },
+    {
+      "epoch": 3.8357906675307842,
+      "grad_norm": 0.08713270723819733,
+      "learning_rate": 9.12282280930735e-06,
+      "loss": 0.2371,
+      "step": 47349
+    },
+    {
+      "epoch": 3.8358716785482825,
+      "grad_norm": 0.05953427404165268,
+      "learning_rate": 9.118322156712724e-06,
+      "loss": 0.214,
+      "step": 47350
+    },
+    {
+      "epoch": 3.835952689565781,
+      "grad_norm": 0.06580721586942673,
+      "learning_rate": 9.113821504118098e-06,
+      "loss": 0.2079,
+      "step": 47351
+    },
+    {
+      "epoch": 3.8360337005832794,
+      "grad_norm": 0.08798111230134964,
+      "learning_rate": 9.109320851523471e-06,
+      "loss": 0.2701,
+      "step": 47352
+    },
+    {
+      "epoch": 3.8361147116007777,
+      "grad_norm": 0.0650232806801796,
+      "learning_rate": 9.104820198928846e-06,
+      "loss": 0.2124,
+      "step": 47353
+    },
+    {
+      "epoch": 3.8361957226182763,
+      "grad_norm": 0.0631076991558075,
+      "learning_rate": 9.100319546334218e-06,
+      "loss": 0.2093,
+      "step": 47354
+    },
+    {
+      "epoch": 3.8362767336357746,
+      "grad_norm": 0.0658016949892044,
+      "learning_rate": 9.095818893739592e-06,
+      "loss": 0.2163,
+      "step": 47355
+    },
+    {
+      "epoch": 3.836357744653273,
+      "grad_norm": 0.06750229001045227,
+      "learning_rate": 9.091318241144967e-06,
+      "loss": 0.2013,
+      "step": 47356
+    },
+    {
+      "epoch": 3.836438755670771,
+      "grad_norm": 0.07032305002212524,
+      "learning_rate": 9.08681758855034e-06,
+      "loss": 0.2533,
+      "step": 47357
+    },
+    {
+      "epoch": 3.8365197666882693,
+      "grad_norm": 0.0590241402387619,
+      "learning_rate": 9.082316935955713e-06,
+      "loss": 0.2212,
+      "step": 47358
+    },
+    {
+      "epoch": 3.836600777705768,
+      "grad_norm": 0.06706096976995468,
+      "learning_rate": 9.077816283361088e-06,
+      "loss": 0.2238,
+      "step": 47359
+    },
+    {
+      "epoch": 3.8366817887232663,
+      "grad_norm": 0.08542868494987488,
+      "learning_rate": 9.073315630766462e-06,
+      "loss": 0.2208,
+      "step": 47360
+    },
+    {
+      "epoch": 3.8367627997407645,
+      "grad_norm": 0.06941386312246323,
+      "learning_rate": 9.068814978171835e-06,
+      "loss": 0.2698,
+      "step": 47361
+    },
+    {
+      "epoch": 3.836843810758263,
+      "grad_norm": 0.060006238520145416,
+      "learning_rate": 9.06431432557721e-06,
+      "loss": 0.1835,
+      "step": 47362
+    },
+    {
+      "epoch": 3.8369248217757614,
+      "grad_norm": 0.06161088868975639,
+      "learning_rate": 9.059813672982582e-06,
+      "loss": 0.204,
+      "step": 47363
+    },
+    {
+      "epoch": 3.8370058327932597,
+      "grad_norm": 0.0850311890244484,
+      "learning_rate": 9.055313020387956e-06,
+      "loss": 0.2353,
+      "step": 47364
+    },
+    {
+      "epoch": 3.8370868438107584,
+      "grad_norm": 0.06477893143892288,
+      "learning_rate": 9.050812367793331e-06,
+      "loss": 0.2299,
+      "step": 47365
+    },
+    {
+      "epoch": 3.8371678548282566,
+      "grad_norm": 0.06894608587026596,
+      "learning_rate": 9.046311715198705e-06,
+      "loss": 0.232,
+      "step": 47366
+    },
+    {
+      "epoch": 3.837248865845755,
+      "grad_norm": 0.07398436218500137,
+      "learning_rate": 9.041811062604077e-06,
+      "loss": 0.2443,
+      "step": 47367
+    },
+    {
+      "epoch": 3.8373298768632536,
+      "grad_norm": 0.07592013478279114,
+      "learning_rate": 9.037310410009452e-06,
+      "loss": 0.2166,
+      "step": 47368
+    },
+    {
+      "epoch": 3.837410887880752,
+      "grad_norm": 0.07261329144239426,
+      "learning_rate": 9.032809757414826e-06,
+      "loss": 0.2163,
+      "step": 47369
+    },
+    {
+      "epoch": 3.83749189889825,
+      "grad_norm": 0.0708971843123436,
+      "learning_rate": 9.0283091048202e-06,
+      "loss": 0.256,
+      "step": 47370
+    },
+    {
+      "epoch": 3.8375729099157487,
+      "grad_norm": 0.07381059974431992,
+      "learning_rate": 9.023808452225573e-06,
+      "loss": 0.2307,
+      "step": 47371
+    },
+    {
+      "epoch": 3.837653920933247,
+      "grad_norm": 0.04990634694695473,
+      "learning_rate": 9.019307799630947e-06,
+      "loss": 0.2111,
+      "step": 47372
+    },
+    {
+      "epoch": 3.837734931950745,
+      "grad_norm": 0.06506974995136261,
+      "learning_rate": 9.01480714703632e-06,
+      "loss": 0.2267,
+      "step": 47373
+    },
+    {
+      "epoch": 3.837815942968244,
+      "grad_norm": 0.08760213106870651,
+      "learning_rate": 9.010306494441695e-06,
+      "loss": 0.2154,
+      "step": 47374
+    },
+    {
+      "epoch": 3.837896953985742,
+      "grad_norm": 0.07203999906778336,
+      "learning_rate": 9.005805841847069e-06,
+      "loss": 0.2392,
+      "step": 47375
+    },
+    {
+      "epoch": 3.8379779650032404,
+      "grad_norm": 0.0878889188170433,
+      "learning_rate": 9.001305189252441e-06,
+      "loss": 0.2725,
+      "step": 47376
+    },
+    {
+      "epoch": 3.838058976020739,
+      "grad_norm": 0.08162927627563477,
+      "learning_rate": 8.996804536657816e-06,
+      "loss": 0.217,
+      "step": 47377
+    },
+    {
+      "epoch": 3.8381399870382373,
+      "grad_norm": 0.07394520938396454,
+      "learning_rate": 8.99230388406319e-06,
+      "loss": 0.185,
+      "step": 47378
+    },
+    {
+      "epoch": 3.8382209980557356,
+      "grad_norm": 0.0606919601559639,
+      "learning_rate": 8.987803231468564e-06,
+      "loss": 0.2269,
+      "step": 47379
+    },
+    {
+      "epoch": 3.838302009073234,
+      "grad_norm": 0.07362279295921326,
+      "learning_rate": 8.983302578873937e-06,
+      "loss": 0.2205,
+      "step": 47380
+    },
+    {
+      "epoch": 3.838383020090732,
+      "grad_norm": 0.06929288804531097,
+      "learning_rate": 8.97880192627931e-06,
+      "loss": 0.1886,
+      "step": 47381
+    },
+    {
+      "epoch": 3.8384640311082308,
+      "grad_norm": 0.05175120756030083,
+      "learning_rate": 8.974301273684684e-06,
+      "loss": 0.1816,
+      "step": 47382
+    },
+    {
+      "epoch": 3.838545042125729,
+      "grad_norm": 0.06899210810661316,
+      "learning_rate": 8.96980062109006e-06,
+      "loss": 0.2392,
+      "step": 47383
+    },
+    {
+      "epoch": 3.8386260531432272,
+      "grad_norm": 0.07615984231233597,
+      "learning_rate": 8.965299968495432e-06,
+      "loss": 0.2399,
+      "step": 47384
+    },
+    {
+      "epoch": 3.838707064160726,
+      "grad_norm": 0.08293159306049347,
+      "learning_rate": 8.960799315900805e-06,
+      "loss": 0.2523,
+      "step": 47385
+    },
+    {
+      "epoch": 3.838788075178224,
+      "grad_norm": 0.06818405538797379,
+      "learning_rate": 8.95629866330618e-06,
+      "loss": 0.2465,
+      "step": 47386
+    },
+    {
+      "epoch": 3.8388690861957224,
+      "grad_norm": 0.07864660769701004,
+      "learning_rate": 8.951798010711554e-06,
+      "loss": 0.2487,
+      "step": 47387
+    },
+    {
+      "epoch": 3.838950097213221,
+      "grad_norm": 0.06878384947776794,
+      "learning_rate": 8.947297358116926e-06,
+      "loss": 0.2125,
+      "step": 47388
+    },
+    {
+      "epoch": 3.8390311082307194,
+      "grad_norm": 0.08059203624725342,
+      "learning_rate": 8.942796705522301e-06,
+      "loss": 0.1948,
+      "step": 47389
+    },
+    {
+      "epoch": 3.8391121192482176,
+      "grad_norm": 0.07346974313259125,
+      "learning_rate": 8.938296052927675e-06,
+      "loss": 0.2186,
+      "step": 47390
+    },
+    {
+      "epoch": 3.8391931302657163,
+      "grad_norm": 0.06083647534251213,
+      "learning_rate": 8.933795400333048e-06,
+      "loss": 0.2405,
+      "step": 47391
+    },
+    {
+      "epoch": 3.8392741412832145,
+      "grad_norm": 0.07009142637252808,
+      "learning_rate": 8.929294747738424e-06,
+      "loss": 0.2472,
+      "step": 47392
+    },
+    {
+      "epoch": 3.839355152300713,
+      "grad_norm": 0.07019902765750885,
+      "learning_rate": 8.924794095143796e-06,
+      "loss": 0.1912,
+      "step": 47393
+    },
+    {
+      "epoch": 3.8394361633182115,
+      "grad_norm": 0.07780317962169647,
+      "learning_rate": 8.92029344254917e-06,
+      "loss": 0.2321,
+      "step": 47394
+    },
+    {
+      "epoch": 3.8395171743357097,
+      "grad_norm": 0.061078991740942,
+      "learning_rate": 8.915792789954545e-06,
+      "loss": 0.1962,
+      "step": 47395
+    },
+    {
+      "epoch": 3.839598185353208,
+      "grad_norm": 0.07453559339046478,
+      "learning_rate": 8.911292137359918e-06,
+      "loss": 0.2545,
+      "step": 47396
+    },
+    {
+      "epoch": 3.8396791963707066,
+      "grad_norm": 0.0780182033777237,
+      "learning_rate": 8.90679148476529e-06,
+      "loss": 0.2747,
+      "step": 47397
+    },
+    {
+      "epoch": 3.839760207388205,
+      "grad_norm": 0.07100244611501694,
+      "learning_rate": 8.902290832170665e-06,
+      "loss": 0.2418,
+      "step": 47398
+    },
+    {
+      "epoch": 3.839841218405703,
+      "grad_norm": 0.053392693400382996,
+      "learning_rate": 8.897790179576039e-06,
+      "loss": 0.2013,
+      "step": 47399
+    },
+    {
+      "epoch": 3.839922229423202,
+      "grad_norm": 0.052919309586286545,
+      "learning_rate": 8.893289526981413e-06,
+      "loss": 0.2098,
+      "step": 47400
+    },
+    {
+      "epoch": 3.8400032404407,
+      "grad_norm": 0.0662708431482315,
+      "learning_rate": 8.888788874386786e-06,
+      "loss": 0.2121,
+      "step": 47401
+    },
+    {
+      "epoch": 3.8400842514581983,
+      "grad_norm": 0.07481697201728821,
+      "learning_rate": 8.88428822179216e-06,
+      "loss": 0.2062,
+      "step": 47402
+    },
+    {
+      "epoch": 3.8401652624756966,
+      "grad_norm": 0.07575065642595291,
+      "learning_rate": 8.879787569197533e-06,
+      "loss": 0.2033,
+      "step": 47403
+    },
+    {
+      "epoch": 3.840246273493195,
+      "grad_norm": 0.06485338509082794,
+      "learning_rate": 8.875286916602909e-06,
+      "loss": 0.2298,
+      "step": 47404
+    },
+    {
+      "epoch": 3.8403272845106935,
+      "grad_norm": 0.06933486461639404,
+      "learning_rate": 8.870786264008282e-06,
+      "loss": 0.2329,
+      "step": 47405
+    },
+    {
+      "epoch": 3.8404082955281917,
+      "grad_norm": 0.07605458796024323,
+      "learning_rate": 8.866285611413654e-06,
+      "loss": 0.2347,
+      "step": 47406
+    },
+    {
+      "epoch": 3.84048930654569,
+      "grad_norm": 0.07894845306873322,
+      "learning_rate": 8.86178495881903e-06,
+      "loss": 0.2167,
+      "step": 47407
+    },
+    {
+      "epoch": 3.8405703175631887,
+      "grad_norm": 0.0782691240310669,
+      "learning_rate": 8.857284306224403e-06,
+      "loss": 0.264,
+      "step": 47408
+    },
+    {
+      "epoch": 3.840651328580687,
+      "grad_norm": 0.07806120067834854,
+      "learning_rate": 8.852783653629777e-06,
+      "loss": 0.2047,
+      "step": 47409
+    },
+    {
+      "epoch": 3.840732339598185,
+      "grad_norm": 0.06680679321289062,
+      "learning_rate": 8.84828300103515e-06,
+      "loss": 0.1997,
+      "step": 47410
+    },
+    {
+      "epoch": 3.840813350615684,
+      "grad_norm": 0.09053415805101395,
+      "learning_rate": 8.843782348440524e-06,
+      "loss": 0.2392,
+      "step": 47411
+    },
+    {
+      "epoch": 3.840894361633182,
+      "grad_norm": 0.08757054805755615,
+      "learning_rate": 8.839281695845897e-06,
+      "loss": 0.2489,
+      "step": 47412
+    },
+    {
+      "epoch": 3.8409753726506803,
+      "grad_norm": 0.10022596269845963,
+      "learning_rate": 8.834781043251273e-06,
+      "loss": 0.2216,
+      "step": 47413
+    },
+    {
+      "epoch": 3.841056383668179,
+      "grad_norm": 0.06728453189134598,
+      "learning_rate": 8.830280390656645e-06,
+      "loss": 0.1942,
+      "step": 47414
+    },
+    {
+      "epoch": 3.8411373946856773,
+      "grad_norm": 0.07960285991430283,
+      "learning_rate": 8.825779738062018e-06,
+      "loss": 0.2327,
+      "step": 47415
+    },
+    {
+      "epoch": 3.8412184057031755,
+      "grad_norm": 0.0784974917769432,
+      "learning_rate": 8.821279085467394e-06,
+      "loss": 0.2411,
+      "step": 47416
+    },
+    {
+      "epoch": 3.841299416720674,
+      "grad_norm": 0.06307407468557358,
+      "learning_rate": 8.816778432872767e-06,
+      "loss": 0.198,
+      "step": 47417
+    },
+    {
+      "epoch": 3.8413804277381725,
+      "grad_norm": 0.07177886366844177,
+      "learning_rate": 8.81227778027814e-06,
+      "loss": 0.2221,
+      "step": 47418
+    },
+    {
+      "epoch": 3.8414614387556707,
+      "grad_norm": 0.07093466073274612,
+      "learning_rate": 8.807777127683514e-06,
+      "loss": 0.2285,
+      "step": 47419
+    },
+    {
+      "epoch": 3.8415424497731694,
+      "grad_norm": 0.07861575484275818,
+      "learning_rate": 8.803276475088888e-06,
+      "loss": 0.2226,
+      "step": 47420
+    },
+    {
+      "epoch": 3.8416234607906676,
+      "grad_norm": 0.08279623091220856,
+      "learning_rate": 8.798775822494262e-06,
+      "loss": 0.2354,
+      "step": 47421
+    },
+    {
+      "epoch": 3.841704471808166,
+      "grad_norm": 0.07097911089658737,
+      "learning_rate": 8.794275169899637e-06,
+      "loss": 0.1956,
+      "step": 47422
+    },
+    {
+      "epoch": 3.8417854828256646,
+      "grad_norm": 0.08817955106496811,
+      "learning_rate": 8.789774517305009e-06,
+      "loss": 0.2312,
+      "step": 47423
+    },
+    {
+      "epoch": 3.841866493843163,
+      "grad_norm": 0.06684723496437073,
+      "learning_rate": 8.785273864710382e-06,
+      "loss": 0.2451,
+      "step": 47424
+    },
+    {
+      "epoch": 3.841947504860661,
+      "grad_norm": 0.07929820567369461,
+      "learning_rate": 8.780773212115758e-06,
+      "loss": 0.2249,
+      "step": 47425
+    },
+    {
+      "epoch": 3.8420285158781593,
+      "grad_norm": 0.069803386926651,
+      "learning_rate": 8.776272559521131e-06,
+      "loss": 0.2111,
+      "step": 47426
+    },
+    {
+      "epoch": 3.8421095268956575,
+      "grad_norm": 0.07593082636594772,
+      "learning_rate": 8.771771906926505e-06,
+      "loss": 0.1928,
+      "step": 47427
+    },
+    {
+      "epoch": 3.8421905379131562,
+      "grad_norm": 0.0848335325717926,
+      "learning_rate": 8.767271254331879e-06,
+      "loss": 0.2202,
+      "step": 47428
+    },
+    {
+      "epoch": 3.8422715489306545,
+      "grad_norm": 0.07958927750587463,
+      "learning_rate": 8.762770601737252e-06,
+      "loss": 0.2404,
+      "step": 47429
+    },
+    {
+      "epoch": 3.8423525599481527,
+      "grad_norm": 0.08629477769136429,
+      "learning_rate": 8.758269949142626e-06,
+      "loss": 0.2137,
+      "step": 47430
+    },
+    {
+      "epoch": 3.8424335709656514,
+      "grad_norm": 0.0668138712644577,
+      "learning_rate": 8.753769296548e-06,
+      "loss": 0.2097,
+      "step": 47431
+    },
+    {
+      "epoch": 3.8425145819831497,
+      "grad_norm": 0.06344747543334961,
+      "learning_rate": 8.749268643953373e-06,
+      "loss": 0.2491,
+      "step": 47432
+    },
+    {
+      "epoch": 3.842595593000648,
+      "grad_norm": 0.05329788476228714,
+      "learning_rate": 8.744767991358748e-06,
+      "loss": 0.1815,
+      "step": 47433
+    },
+    {
+      "epoch": 3.8426766040181466,
+      "grad_norm": 0.06972850859165192,
+      "learning_rate": 8.740267338764122e-06,
+      "loss": 0.234,
+      "step": 47434
+    },
+    {
+      "epoch": 3.842757615035645,
+      "grad_norm": 0.06892770528793335,
+      "learning_rate": 8.735766686169495e-06,
+      "loss": 0.2041,
+      "step": 47435
+    },
+    {
+      "epoch": 3.842838626053143,
+      "grad_norm": 0.07953707873821259,
+      "learning_rate": 8.731266033574869e-06,
+      "loss": 0.2542,
+      "step": 47436
+    },
+    {
+      "epoch": 3.8429196370706418,
+      "grad_norm": 0.09926214069128036,
+      "learning_rate": 8.726765380980243e-06,
+      "loss": 0.2195,
+      "step": 47437
+    },
+    {
+      "epoch": 3.84300064808814,
+      "grad_norm": 0.08672218024730682,
+      "learning_rate": 8.722264728385616e-06,
+      "loss": 0.2131,
+      "step": 47438
+    },
+    {
+      "epoch": 3.8430816591056383,
+      "grad_norm": 0.07387741655111313,
+      "learning_rate": 8.71776407579099e-06,
+      "loss": 0.2321,
+      "step": 47439
+    },
+    {
+      "epoch": 3.843162670123137,
+      "grad_norm": 0.10262610018253326,
+      "learning_rate": 8.713263423196363e-06,
+      "loss": 0.2587,
+      "step": 47440
+    },
+    {
+      "epoch": 3.843243681140635,
+      "grad_norm": 0.0705871507525444,
+      "learning_rate": 8.708762770601737e-06,
+      "loss": 0.2413,
+      "step": 47441
+    },
+    {
+      "epoch": 3.8433246921581334,
+      "grad_norm": 0.0629960298538208,
+      "learning_rate": 8.704262118007112e-06,
+      "loss": 0.2274,
+      "step": 47442
+    },
+    {
+      "epoch": 3.843405703175632,
+      "grad_norm": 0.06705893576145172,
+      "learning_rate": 8.699761465412486e-06,
+      "loss": 0.2104,
+      "step": 47443
+    },
+    {
+      "epoch": 3.8434867141931304,
+      "grad_norm": 0.0768185630440712,
+      "learning_rate": 8.695260812817858e-06,
+      "loss": 0.2321,
+      "step": 47444
+    },
+    {
+      "epoch": 3.8435677252106286,
+      "grad_norm": 0.07738561928272247,
+      "learning_rate": 8.690760160223233e-06,
+      "loss": 0.2219,
+      "step": 47445
+    },
+    {
+      "epoch": 3.843648736228127,
+      "grad_norm": 0.087811179459095,
+      "learning_rate": 8.686259507628607e-06,
+      "loss": 0.2231,
+      "step": 47446
+    },
+    {
+      "epoch": 3.8437297472456255,
+      "grad_norm": 0.0710371658205986,
+      "learning_rate": 8.68175885503398e-06,
+      "loss": 0.2017,
+      "step": 47447
+    },
+    {
+      "epoch": 3.843810758263124,
+      "grad_norm": 0.06479393690824509,
+      "learning_rate": 8.677258202439354e-06,
+      "loss": 0.1882,
+      "step": 47448
+    },
+    {
+      "epoch": 3.843891769280622,
+      "grad_norm": 0.06025944650173187,
+      "learning_rate": 8.672757549844728e-06,
+      "loss": 0.2261,
+      "step": 47449
+    },
+    {
+      "epoch": 3.8439727802981203,
+      "grad_norm": 0.05862006917595863,
+      "learning_rate": 8.668256897250101e-06,
+      "loss": 0.213,
+      "step": 47450
+    },
+    {
+      "epoch": 3.844053791315619,
+      "grad_norm": 0.07549209147691727,
+      "learning_rate": 8.663756244655476e-06,
+      "loss": 0.2305,
+      "step": 47451
+    },
+    {
+      "epoch": 3.844134802333117,
+      "grad_norm": 0.07217980176210403,
+      "learning_rate": 8.65925559206085e-06,
+      "loss": 0.2431,
+      "step": 47452
+    },
+    {
+      "epoch": 3.8442158133506155,
+      "grad_norm": 0.07103247195482254,
+      "learning_rate": 8.654754939466222e-06,
+      "loss": 0.2087,
+      "step": 47453
+    },
+    {
+      "epoch": 3.844296824368114,
+      "grad_norm": 0.06943337619304657,
+      "learning_rate": 8.650254286871597e-06,
+      "loss": 0.2815,
+      "step": 47454
+    },
+    {
+      "epoch": 3.8443778353856124,
+      "grad_norm": 0.06522897630929947,
+      "learning_rate": 8.645753634276971e-06,
+      "loss": 0.1963,
+      "step": 47455
+    },
+    {
+      "epoch": 3.8444588464031106,
+      "grad_norm": 0.07871834933757782,
+      "learning_rate": 8.641252981682345e-06,
+      "loss": 0.2545,
+      "step": 47456
+    },
+    {
+      "epoch": 3.8445398574206093,
+      "grad_norm": 0.054104987531900406,
+      "learning_rate": 8.636752329087718e-06,
+      "loss": 0.2043,
+      "step": 47457
+    },
+    {
+      "epoch": 3.8446208684381076,
+      "grad_norm": 0.06654616445302963,
+      "learning_rate": 8.632251676493092e-06,
+      "loss": 0.2432,
+      "step": 47458
+    },
+    {
+      "epoch": 3.844701879455606,
+      "grad_norm": 0.09337121993303299,
+      "learning_rate": 8.627751023898465e-06,
+      "loss": 0.2573,
+      "step": 47459
+    },
+    {
+      "epoch": 3.8447828904731045,
+      "grad_norm": 0.0734034851193428,
+      "learning_rate": 8.62325037130384e-06,
+      "loss": 0.237,
+      "step": 47460
+    },
+    {
+      "epoch": 3.8448639014906028,
+      "grad_norm": 0.0719435065984726,
+      "learning_rate": 8.618749718709214e-06,
+      "loss": 0.1828,
+      "step": 47461
+    },
+    {
+      "epoch": 3.844944912508101,
+      "grad_norm": 0.0561227910220623,
+      "learning_rate": 8.614249066114586e-06,
+      "loss": 0.1634,
+      "step": 47462
+    },
+    {
+      "epoch": 3.8450259235255997,
+      "grad_norm": 0.06853722035884857,
+      "learning_rate": 8.609748413519961e-06,
+      "loss": 0.2296,
+      "step": 47463
+    },
+    {
+      "epoch": 3.845106934543098,
+      "grad_norm": 0.07261405140161514,
+      "learning_rate": 8.605247760925335e-06,
+      "loss": 0.2562,
+      "step": 47464
+    },
+    {
+      "epoch": 3.845187945560596,
+      "grad_norm": 0.06826271861791611,
+      "learning_rate": 8.600747108330709e-06,
+      "loss": 0.2318,
+      "step": 47465
+    },
+    {
+      "epoch": 3.845268956578095,
+      "grad_norm": 0.07757086306810379,
+      "learning_rate": 8.596246455736082e-06,
+      "loss": 0.1849,
+      "step": 47466
+    },
+    {
+      "epoch": 3.845349967595593,
+      "grad_norm": 0.07977145165205002,
+      "learning_rate": 8.591745803141456e-06,
+      "loss": 0.2338,
+      "step": 47467
+    },
+    {
+      "epoch": 3.8454309786130914,
+      "grad_norm": 0.06785174459218979,
+      "learning_rate": 8.58724515054683e-06,
+      "loss": 0.2292,
+      "step": 47468
+    },
+    {
+      "epoch": 3.8455119896305896,
+      "grad_norm": 0.07069353014230728,
+      "learning_rate": 8.582744497952205e-06,
+      "loss": 0.2281,
+      "step": 47469
+    },
+    {
+      "epoch": 3.8455930006480883,
+      "grad_norm": 0.0715603157877922,
+      "learning_rate": 8.578243845357577e-06,
+      "loss": 0.2052,
+      "step": 47470
+    },
+    {
+      "epoch": 3.8456740116655865,
+      "grad_norm": 0.08521236479282379,
+      "learning_rate": 8.57374319276295e-06,
+      "loss": 0.2068,
+      "step": 47471
+    },
+    {
+      "epoch": 3.845755022683085,
+      "grad_norm": 0.06399007141590118,
+      "learning_rate": 8.569242540168326e-06,
+      "loss": 0.2399,
+      "step": 47472
+    },
+    {
+      "epoch": 3.845836033700583,
+      "grad_norm": 0.06669558584690094,
+      "learning_rate": 8.5647418875737e-06,
+      "loss": 0.2103,
+      "step": 47473
+    },
+    {
+      "epoch": 3.8459170447180817,
+      "grad_norm": 0.06358367949724197,
+      "learning_rate": 8.560241234979071e-06,
+      "loss": 0.2175,
+      "step": 47474
+    },
+    {
+      "epoch": 3.84599805573558,
+      "grad_norm": 0.07232925295829773,
+      "learning_rate": 8.555740582384446e-06,
+      "loss": 0.264,
+      "step": 47475
+    },
+    {
+      "epoch": 3.846079066753078,
+      "grad_norm": 0.0711951032280922,
+      "learning_rate": 8.55123992978982e-06,
+      "loss": 0.2351,
+      "step": 47476
+    },
+    {
+      "epoch": 3.846160077770577,
+      "grad_norm": 0.07407108694314957,
+      "learning_rate": 8.546739277195194e-06,
+      "loss": 0.248,
+      "step": 47477
+    },
+    {
+      "epoch": 3.846241088788075,
+      "grad_norm": 0.06144161522388458,
+      "learning_rate": 8.542238624600569e-06,
+      "loss": 0.2372,
+      "step": 47478
+    },
+    {
+      "epoch": 3.8463220998055734,
+      "grad_norm": 0.07053658366203308,
+      "learning_rate": 8.53773797200594e-06,
+      "loss": 0.2325,
+      "step": 47479
+    },
+    {
+      "epoch": 3.846403110823072,
+      "grad_norm": 0.08187015354633331,
+      "learning_rate": 8.533237319411314e-06,
+      "loss": 0.2449,
+      "step": 47480
+    },
+    {
+      "epoch": 3.8464841218405703,
+      "grad_norm": 0.06435208022594452,
+      "learning_rate": 8.52873666681669e-06,
+      "loss": 0.2177,
+      "step": 47481
+    },
+    {
+      "epoch": 3.8465651328580686,
+      "grad_norm": 0.06776495277881622,
+      "learning_rate": 8.524236014222063e-06,
+      "loss": 0.197,
+      "step": 47482
+    },
+    {
+      "epoch": 3.8466461438755672,
+      "grad_norm": 0.07252959907054901,
+      "learning_rate": 8.519735361627435e-06,
+      "loss": 0.2276,
+      "step": 47483
+    },
+    {
+      "epoch": 3.8467271548930655,
+      "grad_norm": 0.08307579159736633,
+      "learning_rate": 8.51523470903281e-06,
+      "loss": 0.2332,
+      "step": 47484
+    },
+    {
+      "epoch": 3.8468081659105637,
+      "grad_norm": 0.07153511792421341,
+      "learning_rate": 8.510734056438184e-06,
+      "loss": 0.2307,
+      "step": 47485
+    },
+    {
+      "epoch": 3.8468891769280624,
+      "grad_norm": 0.07924754917621613,
+      "learning_rate": 8.506233403843558e-06,
+      "loss": 0.2227,
+      "step": 47486
+    },
+    {
+      "epoch": 3.8469701879455607,
+      "grad_norm": 0.08361861109733582,
+      "learning_rate": 8.501732751248931e-06,
+      "loss": 0.2129,
+      "step": 47487
+    },
+    {
+      "epoch": 3.847051198963059,
+      "grad_norm": 0.06627753376960754,
+      "learning_rate": 8.497232098654305e-06,
+      "loss": 0.2173,
+      "step": 47488
+    },
+    {
+      "epoch": 3.8471322099805576,
+      "grad_norm": 0.08294981718063354,
+      "learning_rate": 8.492731446059679e-06,
+      "loss": 0.2502,
+      "step": 47489
+    },
+    {
+      "epoch": 3.847213220998056,
+      "grad_norm": 0.060485485941171646,
+      "learning_rate": 8.488230793465054e-06,
+      "loss": 0.1837,
+      "step": 47490
+    },
+    {
+      "epoch": 3.847294232015554,
+      "grad_norm": 0.06141113489866257,
+      "learning_rate": 8.483730140870427e-06,
+      "loss": 0.2074,
+      "step": 47491
+    },
+    {
+      "epoch": 3.8473752430330523,
+      "grad_norm": 0.07824747264385223,
+      "learning_rate": 8.4792294882758e-06,
+      "loss": 0.2496,
+      "step": 47492
+    },
+    {
+      "epoch": 3.847456254050551,
+      "grad_norm": 0.058931589126586914,
+      "learning_rate": 8.474728835681175e-06,
+      "loss": 0.1947,
+      "step": 47493
+    },
+    {
+      "epoch": 3.8475372650680493,
+      "grad_norm": 0.08158572763204575,
+      "learning_rate": 8.470228183086548e-06,
+      "loss": 0.2319,
+      "step": 47494
+    },
+    {
+      "epoch": 3.8476182760855475,
+      "grad_norm": 0.06126769632101059,
+      "learning_rate": 8.465727530491922e-06,
+      "loss": 0.2286,
+      "step": 47495
+    },
+    {
+      "epoch": 3.8476992871030458,
+      "grad_norm": 0.05657956004142761,
+      "learning_rate": 8.461226877897295e-06,
+      "loss": 0.1731,
+      "step": 47496
+    },
+    {
+      "epoch": 3.8477802981205445,
+      "grad_norm": 0.07108405232429504,
+      "learning_rate": 8.456726225302669e-06,
+      "loss": 0.2344,
+      "step": 47497
+    },
+    {
+      "epoch": 3.8478613091380427,
+      "grad_norm": 0.07252726703882217,
+      "learning_rate": 8.452225572708043e-06,
+      "loss": 0.218,
+      "step": 47498
+    },
+    {
+      "epoch": 3.847942320155541,
+      "grad_norm": 0.07801450788974762,
+      "learning_rate": 8.447724920113418e-06,
+      "loss": 0.2097,
+      "step": 47499
+    },
+    {
+      "epoch": 3.8480233311730396,
+      "grad_norm": 0.06933198869228363,
+      "learning_rate": 8.44322426751879e-06,
+      "loss": 0.2117,
+      "step": 47500
+    },
+    {
+      "epoch": 3.848104342190538,
+      "grad_norm": 0.06913826614618301,
+      "learning_rate": 8.438723614924163e-06,
+      "loss": 0.2026,
+      "step": 47501
+    },
+    {
+      "epoch": 3.848185353208036,
+      "grad_norm": 0.08163614571094513,
+      "learning_rate": 8.434222962329539e-06,
+      "loss": 0.1878,
+      "step": 47502
+    },
+    {
+      "epoch": 3.848266364225535,
+      "grad_norm": 0.06842586398124695,
+      "learning_rate": 8.429722309734912e-06,
+      "loss": 0.2108,
+      "step": 47503
+    },
+    {
+      "epoch": 3.848347375243033,
+      "grad_norm": 0.07575143128633499,
+      "learning_rate": 8.425221657140284e-06,
+      "loss": 0.2302,
+      "step": 47504
+    },
+    {
+      "epoch": 3.8484283862605313,
+      "grad_norm": 0.06495438516139984,
+      "learning_rate": 8.42072100454566e-06,
+      "loss": 0.1875,
+      "step": 47505
+    },
+    {
+      "epoch": 3.84850939727803,
+      "grad_norm": 0.06851106882095337,
+      "learning_rate": 8.416220351951033e-06,
+      "loss": 0.2367,
+      "step": 47506
+    },
+    {
+      "epoch": 3.8485904082955282,
+      "grad_norm": 0.08247493207454681,
+      "learning_rate": 8.411719699356407e-06,
+      "loss": 0.2409,
+      "step": 47507
+    },
+    {
+      "epoch": 3.8486714193130265,
+      "grad_norm": 0.07875152677297592,
+      "learning_rate": 8.407219046761782e-06,
+      "loss": 0.246,
+      "step": 47508
+    },
+    {
+      "epoch": 3.848752430330525,
+      "grad_norm": 0.08067942410707474,
+      "learning_rate": 8.402718394167154e-06,
+      "loss": 0.2295,
+      "step": 47509
+    },
+    {
+      "epoch": 3.8488334413480234,
+      "grad_norm": 0.07155679911375046,
+      "learning_rate": 8.398217741572528e-06,
+      "loss": 0.2452,
+      "step": 47510
+    },
+    {
+      "epoch": 3.8489144523655217,
+      "grad_norm": 0.06344729661941528,
+      "learning_rate": 8.393717088977903e-06,
+      "loss": 0.2275,
+      "step": 47511
+    },
+    {
+      "epoch": 3.8489954633830203,
+      "grad_norm": 0.06181887909770012,
+      "learning_rate": 8.389216436383276e-06,
+      "loss": 0.1943,
+      "step": 47512
+    },
+    {
+      "epoch": 3.8490764744005186,
+      "grad_norm": 0.07348298281431198,
+      "learning_rate": 8.384715783788648e-06,
+      "loss": 0.1927,
+      "step": 47513
+    },
+    {
+      "epoch": 3.849157485418017,
+      "grad_norm": 0.06353019177913666,
+      "learning_rate": 8.380215131194024e-06,
+      "loss": 0.2485,
+      "step": 47514
+    },
+    {
+      "epoch": 3.849238496435515,
+      "grad_norm": 0.07545365393161774,
+      "learning_rate": 8.375714478599397e-06,
+      "loss": 0.2549,
+      "step": 47515
+    },
+    {
+      "epoch": 3.8493195074530138,
+      "grad_norm": 0.07624714821577072,
+      "learning_rate": 8.371213826004771e-06,
+      "loss": 0.2302,
+      "step": 47516
+    },
+    {
+      "epoch": 3.849400518470512,
+      "grad_norm": 0.06423209607601166,
+      "learning_rate": 8.366713173410144e-06,
+      "loss": 0.2249,
+      "step": 47517
+    },
+    {
+      "epoch": 3.8494815294880103,
+      "grad_norm": 0.08231019973754883,
+      "learning_rate": 8.362212520815518e-06,
+      "loss": 0.2052,
+      "step": 47518
+    },
+    {
+      "epoch": 3.8495625405055085,
+      "grad_norm": 0.06847333908081055,
+      "learning_rate": 8.357711868220892e-06,
+      "loss": 0.1864,
+      "step": 47519
+    },
+    {
+      "epoch": 3.849643551523007,
+      "grad_norm": 0.07198107242584229,
+      "learning_rate": 8.353211215626267e-06,
+      "loss": 0.2418,
+      "step": 47520
+    },
+    {
+      "epoch": 3.8497245625405054,
+      "grad_norm": 0.06648959964513779,
+      "learning_rate": 8.34871056303164e-06,
+      "loss": 0.2507,
+      "step": 47521
+    },
+    {
+      "epoch": 3.8498055735580037,
+      "grad_norm": 0.06837810575962067,
+      "learning_rate": 8.344209910437012e-06,
+      "loss": 0.2359,
+      "step": 47522
+    },
+    {
+      "epoch": 3.8498865845755024,
+      "grad_norm": 0.06529515236616135,
+      "learning_rate": 8.339709257842388e-06,
+      "loss": 0.2084,
+      "step": 47523
+    },
+    {
+      "epoch": 3.8499675955930006,
+      "grad_norm": 0.058206506073474884,
+      "learning_rate": 8.335208605247761e-06,
+      "loss": 0.186,
+      "step": 47524
+    },
+    {
+      "epoch": 3.850048606610499,
+      "grad_norm": 0.07718270272016525,
+      "learning_rate": 8.330707952653135e-06,
+      "loss": 0.2177,
+      "step": 47525
+    },
+    {
+      "epoch": 3.8501296176279975,
+      "grad_norm": 0.07879050821065903,
+      "learning_rate": 8.326207300058509e-06,
+      "loss": 0.2429,
+      "step": 47526
+    },
+    {
+      "epoch": 3.850210628645496,
+      "grad_norm": 0.0804138109087944,
+      "learning_rate": 8.321706647463882e-06,
+      "loss": 0.2276,
+      "step": 47527
+    },
+    {
+      "epoch": 3.850291639662994,
+      "grad_norm": 0.07796037197113037,
+      "learning_rate": 8.317205994869256e-06,
+      "loss": 0.2471,
+      "step": 47528
+    },
+    {
+      "epoch": 3.8503726506804927,
+      "grad_norm": 0.0824827328324318,
+      "learning_rate": 8.312705342274631e-06,
+      "loss": 0.2451,
+      "step": 47529
+    },
+    {
+      "epoch": 3.850453661697991,
+      "grad_norm": 0.06677177548408508,
+      "learning_rate": 8.308204689680003e-06,
+      "loss": 0.2119,
+      "step": 47530
+    },
+    {
+      "epoch": 3.850534672715489,
+      "grad_norm": 0.09421838819980621,
+      "learning_rate": 8.303704037085378e-06,
+      "loss": 0.2343,
+      "step": 47531
+    },
+    {
+      "epoch": 3.850615683732988,
+      "grad_norm": 0.06931822001934052,
+      "learning_rate": 8.299203384490752e-06,
+      "loss": 0.2017,
+      "step": 47532
+    },
+    {
+      "epoch": 3.850696694750486,
+      "grad_norm": 0.07877426594495773,
+      "learning_rate": 8.294702731896126e-06,
+      "loss": 0.2199,
+      "step": 47533
+    },
+    {
+      "epoch": 3.8507777057679844,
+      "grad_norm": 0.07271779328584671,
+      "learning_rate": 8.290202079301499e-06,
+      "loss": 0.2334,
+      "step": 47534
+    },
+    {
+      "epoch": 3.850858716785483,
+      "grad_norm": 0.0686742514371872,
+      "learning_rate": 8.285701426706873e-06,
+      "loss": 0.2049,
+      "step": 47535
+    },
+    {
+      "epoch": 3.8509397278029813,
+      "grad_norm": 0.08275677263736725,
+      "learning_rate": 8.281200774112246e-06,
+      "loss": 0.2126,
+      "step": 47536
+    },
+    {
+      "epoch": 3.8510207388204796,
+      "grad_norm": 0.059194862842559814,
+      "learning_rate": 8.27670012151762e-06,
+      "loss": 0.1964,
+      "step": 47537
+    },
+    {
+      "epoch": 3.851101749837978,
+      "grad_norm": 0.05792196840047836,
+      "learning_rate": 8.272199468922995e-06,
+      "loss": 0.1838,
+      "step": 47538
+    },
+    {
+      "epoch": 3.8511827608554765,
+      "grad_norm": 0.07841230928897858,
+      "learning_rate": 8.267698816328367e-06,
+      "loss": 0.2224,
+      "step": 47539
+    },
+    {
+      "epoch": 3.8512637718729748,
+      "grad_norm": 0.06587585806846619,
+      "learning_rate": 8.263198163733742e-06,
+      "loss": 0.2365,
+      "step": 47540
+    },
+    {
+      "epoch": 3.851344782890473,
+      "grad_norm": 0.06362347304821014,
+      "learning_rate": 8.258697511139116e-06,
+      "loss": 0.2163,
+      "step": 47541
+    },
+    {
+      "epoch": 3.8514257939079712,
+      "grad_norm": 0.06370340287685394,
+      "learning_rate": 8.25419685854449e-06,
+      "loss": 0.2243,
+      "step": 47542
+    },
+    {
+      "epoch": 3.85150680492547,
+      "grad_norm": 0.07428768277168274,
+      "learning_rate": 8.249696205949863e-06,
+      "loss": 0.2428,
+      "step": 47543
+    },
+    {
+      "epoch": 3.851587815942968,
+      "grad_norm": 0.08028055727481842,
+      "learning_rate": 8.245195553355237e-06,
+      "loss": 0.2172,
+      "step": 47544
+    },
+    {
+      "epoch": 3.8516688269604664,
+      "grad_norm": 0.06244838237762451,
+      "learning_rate": 8.24069490076061e-06,
+      "loss": 0.2239,
+      "step": 47545
+    },
+    {
+      "epoch": 3.851749837977965,
+      "grad_norm": 0.07794243097305298,
+      "learning_rate": 8.236194248165984e-06,
+      "loss": 0.2254,
+      "step": 47546
+    },
+    {
+      "epoch": 3.8518308489954634,
+      "grad_norm": 0.07461123168468475,
+      "learning_rate": 8.231693595571358e-06,
+      "loss": 0.2043,
+      "step": 47547
+    },
+    {
+      "epoch": 3.8519118600129616,
+      "grad_norm": 0.07418804615736008,
+      "learning_rate": 8.227192942976731e-06,
+      "loss": 0.2453,
+      "step": 47548
+    },
+    {
+      "epoch": 3.8519928710304603,
+      "grad_norm": 0.06374292820692062,
+      "learning_rate": 8.222692290382107e-06,
+      "loss": 0.205,
+      "step": 47549
+    },
+    {
+      "epoch": 3.8520738820479585,
+      "grad_norm": 0.061300892382860184,
+      "learning_rate": 8.21819163778748e-06,
+      "loss": 0.223,
+      "step": 47550
+    },
+    {
+      "epoch": 3.8521548930654568,
+      "grad_norm": 0.06799773126840591,
+      "learning_rate": 8.213690985192854e-06,
+      "loss": 0.2759,
+      "step": 47551
+    },
+    {
+      "epoch": 3.8522359040829555,
+      "grad_norm": 0.07855547964572906,
+      "learning_rate": 8.209190332598227e-06,
+      "loss": 0.2448,
+      "step": 47552
+    },
+    {
+      "epoch": 3.8523169151004537,
+      "grad_norm": 0.07066880166530609,
+      "learning_rate": 8.204689680003601e-06,
+      "loss": 0.2482,
+      "step": 47553
+    },
+    {
+      "epoch": 3.852397926117952,
+      "grad_norm": 0.08537422120571136,
+      "learning_rate": 8.200189027408975e-06,
+      "loss": 0.2267,
+      "step": 47554
+    },
+    {
+      "epoch": 3.8524789371354506,
+      "grad_norm": 0.07067983597517014,
+      "learning_rate": 8.195688374814348e-06,
+      "loss": 0.2373,
+      "step": 47555
+    },
+    {
+      "epoch": 3.852559948152949,
+      "grad_norm": 0.058485932648181915,
+      "learning_rate": 8.191187722219722e-06,
+      "loss": 0.1834,
+      "step": 47556
+    },
+    {
+      "epoch": 3.852640959170447,
+      "grad_norm": 0.06360162794589996,
+      "learning_rate": 8.186687069625095e-06,
+      "loss": 0.1954,
+      "step": 47557
+    },
+    {
+      "epoch": 3.852721970187946,
+      "grad_norm": 0.07332653552293777,
+      "learning_rate": 8.18218641703047e-06,
+      "loss": 0.2465,
+      "step": 47558
+    },
+    {
+      "epoch": 3.852802981205444,
+      "grad_norm": 0.07406601309776306,
+      "learning_rate": 8.177685764435844e-06,
+      "loss": 0.1827,
+      "step": 47559
+    },
+    {
+      "epoch": 3.8528839922229423,
+      "grad_norm": 0.06085878983139992,
+      "learning_rate": 8.173185111841216e-06,
+      "loss": 0.247,
+      "step": 47560
+    },
+    {
+      "epoch": 3.8529650032404406,
+      "grad_norm": 0.07216692715883255,
+      "learning_rate": 8.168684459246591e-06,
+      "loss": 0.2053,
+      "step": 47561
+    },
+    {
+      "epoch": 3.8530460142579392,
+      "grad_norm": 0.07375588268041611,
+      "learning_rate": 8.164183806651965e-06,
+      "loss": 0.2224,
+      "step": 47562
+    },
+    {
+      "epoch": 3.8531270252754375,
+      "grad_norm": 0.07635340839624405,
+      "learning_rate": 8.159683154057339e-06,
+      "loss": 0.2446,
+      "step": 47563
+    },
+    {
+      "epoch": 3.8532080362929357,
+      "grad_norm": 0.08759939670562744,
+      "learning_rate": 8.155182501462712e-06,
+      "loss": 0.2265,
+      "step": 47564
+    },
+    {
+      "epoch": 3.853289047310434,
+      "grad_norm": 0.0754440650343895,
+      "learning_rate": 8.150681848868086e-06,
+      "loss": 0.2195,
+      "step": 47565
+    },
+    {
+      "epoch": 3.8533700583279327,
+      "grad_norm": 0.08304356038570404,
+      "learning_rate": 8.14618119627346e-06,
+      "loss": 0.2309,
+      "step": 47566
+    },
+    {
+      "epoch": 3.853451069345431,
+      "grad_norm": 0.07643993943929672,
+      "learning_rate": 8.141680543678835e-06,
+      "loss": 0.2244,
+      "step": 47567
+    },
+    {
+      "epoch": 3.853532080362929,
+      "grad_norm": 0.08395697921514511,
+      "learning_rate": 8.137179891084208e-06,
+      "loss": 0.1951,
+      "step": 47568
+    },
+    {
+      "epoch": 3.853613091380428,
+      "grad_norm": 0.07898442447185516,
+      "learning_rate": 8.13267923848958e-06,
+      "loss": 0.2543,
+      "step": 47569
+    },
+    {
+      "epoch": 3.853694102397926,
+      "grad_norm": 0.0748404785990715,
+      "learning_rate": 8.128178585894956e-06,
+      "loss": 0.2381,
+      "step": 47570
+    },
+    {
+      "epoch": 3.8537751134154243,
+      "grad_norm": 0.05667143315076828,
+      "learning_rate": 8.12367793330033e-06,
+      "loss": 0.2349,
+      "step": 47571
+    },
+    {
+      "epoch": 3.853856124432923,
+      "grad_norm": 0.058684345334768295,
+      "learning_rate": 8.119177280705703e-06,
+      "loss": 0.2218,
+      "step": 47572
+    },
+    {
+      "epoch": 3.8539371354504213,
+      "grad_norm": 0.05510832741856575,
+      "learning_rate": 8.114676628111076e-06,
+      "loss": 0.1891,
+      "step": 47573
+    },
+    {
+      "epoch": 3.8540181464679195,
+      "grad_norm": 0.07431642711162567,
+      "learning_rate": 8.11017597551645e-06,
+      "loss": 0.2462,
+      "step": 47574
+    },
+    {
+      "epoch": 3.854099157485418,
+      "grad_norm": 0.07371494174003601,
+      "learning_rate": 8.105675322921824e-06,
+      "loss": 0.218,
+      "step": 47575
+    },
+    {
+      "epoch": 3.8541801685029164,
+      "grad_norm": 0.07297206670045853,
+      "learning_rate": 8.101174670327199e-06,
+      "loss": 0.23,
+      "step": 47576
+    },
+    {
+      "epoch": 3.8542611795204147,
+      "grad_norm": 0.07745885103940964,
+      "learning_rate": 8.096674017732573e-06,
+      "loss": 0.2658,
+      "step": 47577
+    },
+    {
+      "epoch": 3.8543421905379134,
+      "grad_norm": 0.07080569863319397,
+      "learning_rate": 8.092173365137944e-06,
+      "loss": 0.2377,
+      "step": 47578
+    },
+    {
+      "epoch": 3.8544232015554116,
+      "grad_norm": 0.07197723537683487,
+      "learning_rate": 8.08767271254332e-06,
+      "loss": 0.2297,
+      "step": 47579
+    },
+    {
+      "epoch": 3.85450421257291,
+      "grad_norm": 0.0773099884390831,
+      "learning_rate": 8.083172059948693e-06,
+      "loss": 0.2535,
+      "step": 47580
+    },
+    {
+      "epoch": 3.8545852235904086,
+      "grad_norm": 0.06363910436630249,
+      "learning_rate": 8.078671407354067e-06,
+      "loss": 0.2199,
+      "step": 47581
+    },
+    {
+      "epoch": 3.854666234607907,
+      "grad_norm": 0.06943506747484207,
+      "learning_rate": 8.07417075475944e-06,
+      "loss": 0.2324,
+      "step": 47582
+    },
+    {
+      "epoch": 3.854747245625405,
+      "grad_norm": 0.0709523931145668,
+      "learning_rate": 8.069670102164814e-06,
+      "loss": 0.2093,
+      "step": 47583
+    },
+    {
+      "epoch": 3.8548282566429033,
+      "grad_norm": 0.06644728779792786,
+      "learning_rate": 8.065169449570188e-06,
+      "loss": 0.1882,
+      "step": 47584
+    },
+    {
+      "epoch": 3.8549092676604015,
+      "grad_norm": 0.06858302652835846,
+      "learning_rate": 8.060668796975563e-06,
+      "loss": 0.2329,
+      "step": 47585
+    },
+    {
+      "epoch": 3.8549902786779002,
+      "grad_norm": 0.06228305771946907,
+      "learning_rate": 8.056168144380935e-06,
+      "loss": 0.2061,
+      "step": 47586
+    },
+    {
+      "epoch": 3.8550712896953985,
+      "grad_norm": 0.07189958542585373,
+      "learning_rate": 8.051667491786309e-06,
+      "loss": 0.2127,
+      "step": 47587
+    },
+    {
+      "epoch": 3.8551523007128967,
+      "grad_norm": 0.07673601806163788,
+      "learning_rate": 8.047166839191684e-06,
+      "loss": 0.181,
+      "step": 47588
+    },
+    {
+      "epoch": 3.8552333117303954,
+      "grad_norm": 0.07827714085578918,
+      "learning_rate": 8.042666186597057e-06,
+      "loss": 0.2282,
+      "step": 47589
+    },
+    {
+      "epoch": 3.8553143227478937,
+      "grad_norm": 0.08124902099370956,
+      "learning_rate": 8.03816553400243e-06,
+      "loss": 0.2034,
+      "step": 47590
+    },
+    {
+      "epoch": 3.855395333765392,
+      "grad_norm": 0.08442587405443192,
+      "learning_rate": 8.033664881407805e-06,
+      "loss": 0.2236,
+      "step": 47591
+    },
+    {
+      "epoch": 3.8554763447828906,
+      "grad_norm": 0.06765611469745636,
+      "learning_rate": 8.029164228813178e-06,
+      "loss": 0.2152,
+      "step": 47592
+    },
+    {
+      "epoch": 3.855557355800389,
+      "grad_norm": 0.06260386109352112,
+      "learning_rate": 8.024663576218552e-06,
+      "loss": 0.2243,
+      "step": 47593
+    },
+    {
+      "epoch": 3.855638366817887,
+      "grad_norm": 0.06030689924955368,
+      "learning_rate": 8.020162923623927e-06,
+      "loss": 0.2254,
+      "step": 47594
+    },
+    {
+      "epoch": 3.8557193778353858,
+      "grad_norm": 0.06486286222934723,
+      "learning_rate": 8.015662271029299e-06,
+      "loss": 0.2008,
+      "step": 47595
+    },
+    {
+      "epoch": 3.855800388852884,
+      "grad_norm": 0.06872467696666718,
+      "learning_rate": 8.011161618434673e-06,
+      "loss": 0.2458,
+      "step": 47596
+    },
+    {
+      "epoch": 3.8558813998703823,
+      "grad_norm": 0.06587346643209457,
+      "learning_rate": 8.006660965840048e-06,
+      "loss": 0.2315,
+      "step": 47597
+    },
+    {
+      "epoch": 3.855962410887881,
+      "grad_norm": 0.07045286148786545,
+      "learning_rate": 8.002160313245422e-06,
+      "loss": 0.1869,
+      "step": 47598
+    },
+    {
+      "epoch": 3.856043421905379,
+      "grad_norm": 0.09421037882566452,
+      "learning_rate": 7.997659660650794e-06,
+      "loss": 0.2215,
+      "step": 47599
+    },
+    {
+      "epoch": 3.8561244329228774,
+      "grad_norm": 0.09723097085952759,
+      "learning_rate": 7.993159008056169e-06,
+      "loss": 0.2381,
+      "step": 47600
+    },
+    {
+      "epoch": 3.856205443940376,
+      "grad_norm": 0.0791918933391571,
+      "learning_rate": 7.988658355461542e-06,
+      "loss": 0.2435,
+      "step": 47601
+    },
+    {
+      "epoch": 3.8562864549578744,
+      "grad_norm": 0.06977692991495132,
+      "learning_rate": 7.984157702866916e-06,
+      "loss": 0.2726,
+      "step": 47602
+    },
+    {
+      "epoch": 3.8563674659753726,
+      "grad_norm": 0.09157250821590424,
+      "learning_rate": 7.97965705027229e-06,
+      "loss": 0.234,
+      "step": 47603
+    },
+    {
+      "epoch": 3.8564484769928713,
+      "grad_norm": 0.07410857826471329,
+      "learning_rate": 7.975156397677663e-06,
+      "loss": 0.2527,
+      "step": 47604
+    },
+    {
+      "epoch": 3.8565294880103695,
+      "grad_norm": 0.05430734157562256,
+      "learning_rate": 7.970655745083037e-06,
+      "loss": 0.209,
+      "step": 47605
+    },
+    {
+      "epoch": 3.856610499027868,
+      "grad_norm": 0.08034774661064148,
+      "learning_rate": 7.966155092488412e-06,
+      "loss": 0.208,
+      "step": 47606
+    },
+    {
+      "epoch": 3.856691510045366,
+      "grad_norm": 0.07333406805992126,
+      "learning_rate": 7.961654439893786e-06,
+      "loss": 0.2328,
+      "step": 47607
+    },
+    {
+      "epoch": 3.8567725210628643,
+      "grad_norm": 0.06615491956472397,
+      "learning_rate": 7.957153787299158e-06,
+      "loss": 0.1903,
+      "step": 47608
+    },
+    {
+      "epoch": 3.856853532080363,
+      "grad_norm": 0.07822062820196152,
+      "learning_rate": 7.952653134704533e-06,
+      "loss": 0.2189,
+      "step": 47609
+    },
+    {
+      "epoch": 3.856934543097861,
+      "grad_norm": 0.06766904890537262,
+      "learning_rate": 7.948152482109907e-06,
+      "loss": 0.2208,
+      "step": 47610
+    },
+    {
+      "epoch": 3.8570155541153595,
+      "grad_norm": 0.08099085837602615,
+      "learning_rate": 7.94365182951528e-06,
+      "loss": 0.2364,
+      "step": 47611
+    },
+    {
+      "epoch": 3.857096565132858,
+      "grad_norm": 0.06338600814342499,
+      "learning_rate": 7.939151176920654e-06,
+      "loss": 0.2428,
+      "step": 47612
+    },
+    {
+      "epoch": 3.8571775761503564,
+      "grad_norm": 0.06889151781797409,
+      "learning_rate": 7.934650524326027e-06,
+      "loss": 0.232,
+      "step": 47613
+    },
+    {
+      "epoch": 3.8572585871678546,
+      "grad_norm": 0.08298153430223465,
+      "learning_rate": 7.930149871731401e-06,
+      "loss": 0.2129,
+      "step": 47614
+    },
+    {
+      "epoch": 3.8573395981853533,
+      "grad_norm": 0.09067533910274506,
+      "learning_rate": 7.925649219136776e-06,
+      "loss": 0.2504,
+      "step": 47615
+    },
+    {
+      "epoch": 3.8574206092028516,
+      "grad_norm": 0.07296990603208542,
+      "learning_rate": 7.921148566542148e-06,
+      "loss": 0.208,
+      "step": 47616
+    },
+    {
+      "epoch": 3.85750162022035,
+      "grad_norm": 0.09865029901266098,
+      "learning_rate": 7.916647913947522e-06,
+      "loss": 0.2474,
+      "step": 47617
+    },
+    {
+      "epoch": 3.8575826312378485,
+      "grad_norm": 0.06511934846639633,
+      "learning_rate": 7.912147261352897e-06,
+      "loss": 0.2175,
+      "step": 47618
+    },
+    {
+      "epoch": 3.8576636422553467,
+      "grad_norm": 0.08340641856193542,
+      "learning_rate": 7.90764660875827e-06,
+      "loss": 0.2267,
+      "step": 47619
+    },
+    {
+      "epoch": 3.857744653272845,
+      "grad_norm": 0.06559319794178009,
+      "learning_rate": 7.903145956163644e-06,
+      "loss": 0.2252,
+      "step": 47620
+    },
+    {
+      "epoch": 3.8578256642903437,
+      "grad_norm": 0.07104986160993576,
+      "learning_rate": 7.898645303569018e-06,
+      "loss": 0.2364,
+      "step": 47621
+    },
+    {
+      "epoch": 3.857906675307842,
+      "grad_norm": 0.06943206489086151,
+      "learning_rate": 7.894144650974391e-06,
+      "loss": 0.2119,
+      "step": 47622
+    },
+    {
+      "epoch": 3.85798768632534,
+      "grad_norm": 0.05858639255166054,
+      "learning_rate": 7.889643998379765e-06,
+      "loss": 0.2225,
+      "step": 47623
+    },
+    {
+      "epoch": 3.858068697342839,
+      "grad_norm": 0.0890335813164711,
+      "learning_rate": 7.88514334578514e-06,
+      "loss": 0.2403,
+      "step": 47624
+    },
+    {
+      "epoch": 3.858149708360337,
+      "grad_norm": 0.061061326414346695,
+      "learning_rate": 7.880642693190512e-06,
+      "loss": 0.2216,
+      "step": 47625
+    },
+    {
+      "epoch": 3.8582307193778353,
+      "grad_norm": 0.049848560243844986,
+      "learning_rate": 7.876142040595886e-06,
+      "loss": 0.2138,
+      "step": 47626
+    },
+    {
+      "epoch": 3.858311730395334,
+      "grad_norm": 0.08001705259084702,
+      "learning_rate": 7.871641388001261e-06,
+      "loss": 0.2296,
+      "step": 47627
+    },
+    {
+      "epoch": 3.8583927414128323,
+      "grad_norm": 0.09020736813545227,
+      "learning_rate": 7.867140735406635e-06,
+      "loss": 0.2792,
+      "step": 47628
+    },
+    {
+      "epoch": 3.8584737524303305,
+      "grad_norm": 0.09471940994262695,
+      "learning_rate": 7.862640082812008e-06,
+      "loss": 0.2496,
+      "step": 47629
+    },
+    {
+      "epoch": 3.8585547634478288,
+      "grad_norm": 0.08428974449634552,
+      "learning_rate": 7.858139430217382e-06,
+      "loss": 0.2456,
+      "step": 47630
+    },
+    {
+      "epoch": 3.858635774465327,
+      "grad_norm": 0.0708090215921402,
+      "learning_rate": 7.853638777622756e-06,
+      "loss": 0.2004,
+      "step": 47631
+    },
+    {
+      "epoch": 3.8587167854828257,
+      "grad_norm": 0.07518108189105988,
+      "learning_rate": 7.84913812502813e-06,
+      "loss": 0.2188,
+      "step": 47632
+    },
+    {
+      "epoch": 3.858797796500324,
+      "grad_norm": 0.07420327514410019,
+      "learning_rate": 7.844637472433503e-06,
+      "loss": 0.2057,
+      "step": 47633
+    },
+    {
+      "epoch": 3.858878807517822,
+      "grad_norm": 0.08655344694852829,
+      "learning_rate": 7.840136819838876e-06,
+      "loss": 0.2228,
+      "step": 47634
+    },
+    {
+      "epoch": 3.858959818535321,
+      "grad_norm": 0.0829836055636406,
+      "learning_rate": 7.83563616724425e-06,
+      "loss": 0.2225,
+      "step": 47635
+    },
+    {
+      "epoch": 3.859040829552819,
+      "grad_norm": 0.06697103381156921,
+      "learning_rate": 7.831135514649625e-06,
+      "loss": 0.1977,
+      "step": 47636
+    },
+    {
+      "epoch": 3.8591218405703174,
+      "grad_norm": 0.0661018043756485,
+      "learning_rate": 7.826634862054999e-06,
+      "loss": 0.2293,
+      "step": 47637
+    },
+    {
+      "epoch": 3.859202851587816,
+      "grad_norm": 0.07524219155311584,
+      "learning_rate": 7.822134209460373e-06,
+      "loss": 0.2315,
+      "step": 47638
+    },
+    {
+      "epoch": 3.8592838626053143,
+      "grad_norm": 0.06726586073637009,
+      "learning_rate": 7.817633556865746e-06,
+      "loss": 0.2015,
+      "step": 47639
+    },
+    {
+      "epoch": 3.8593648736228126,
+      "grad_norm": 0.06795717775821686,
+      "learning_rate": 7.81313290427112e-06,
+      "loss": 0.2322,
+      "step": 47640
+    },
+    {
+      "epoch": 3.8594458846403112,
+      "grad_norm": 0.07243220508098602,
+      "learning_rate": 7.808632251676493e-06,
+      "loss": 0.2447,
+      "step": 47641
+    },
+    {
+      "epoch": 3.8595268956578095,
+      "grad_norm": 0.06129208952188492,
+      "learning_rate": 7.804131599081867e-06,
+      "loss": 0.2262,
+      "step": 47642
+    },
+    {
+      "epoch": 3.8596079066753077,
+      "grad_norm": 0.06558024883270264,
+      "learning_rate": 7.79963094648724e-06,
+      "loss": 0.2312,
+      "step": 47643
+    },
+    {
+      "epoch": 3.8596889176928064,
+      "grad_norm": 0.07786918431520462,
+      "learning_rate": 7.795130293892614e-06,
+      "loss": 0.198,
+      "step": 47644
+    },
+    {
+      "epoch": 3.8597699287103047,
+      "grad_norm": 0.06113965064287186,
+      "learning_rate": 7.79062964129799e-06,
+      "loss": 0.2327,
+      "step": 47645
+    },
+    {
+      "epoch": 3.859850939727803,
+      "grad_norm": 0.06368988752365112,
+      "learning_rate": 7.786128988703361e-06,
+      "loss": 0.2153,
+      "step": 47646
+    },
+    {
+      "epoch": 3.8599319507453016,
+      "grad_norm": 0.06596409529447556,
+      "learning_rate": 7.781628336108737e-06,
+      "loss": 0.2012,
+      "step": 47647
+    },
+    {
+      "epoch": 3.8600129617628,
+      "grad_norm": 0.06855443865060806,
+      "learning_rate": 7.77712768351411e-06,
+      "loss": 0.1867,
+      "step": 47648
+    },
+    {
+      "epoch": 3.860093972780298,
+      "grad_norm": 0.08048706501722336,
+      "learning_rate": 7.772627030919484e-06,
+      "loss": 0.231,
+      "step": 47649
+    },
+    {
+      "epoch": 3.8601749837977968,
+      "grad_norm": 0.06914348155260086,
+      "learning_rate": 7.768126378324857e-06,
+      "loss": 0.1876,
+      "step": 47650
+    },
+    {
+      "epoch": 3.860255994815295,
+      "grad_norm": 0.059210408478975296,
+      "learning_rate": 7.763625725730231e-06,
+      "loss": 0.1888,
+      "step": 47651
+    },
+    {
+      "epoch": 3.8603370058327933,
+      "grad_norm": 0.07270028442144394,
+      "learning_rate": 7.759125073135605e-06,
+      "loss": 0.2167,
+      "step": 47652
+    },
+    {
+      "epoch": 3.8604180168502915,
+      "grad_norm": 0.07310686260461807,
+      "learning_rate": 7.754624420540978e-06,
+      "loss": 0.219,
+      "step": 47653
+    },
+    {
+      "epoch": 3.8604990278677898,
+      "grad_norm": 0.07835370302200317,
+      "learning_rate": 7.750123767946354e-06,
+      "loss": 0.2336,
+      "step": 47654
+    },
+    {
+      "epoch": 3.8605800388852884,
+      "grad_norm": 0.06291946768760681,
+      "learning_rate": 7.745623115351725e-06,
+      "loss": 0.2268,
+      "step": 47655
+    },
+    {
+      "epoch": 3.8606610499027867,
+      "grad_norm": 0.07243097573518753,
+      "learning_rate": 7.7411224627571e-06,
+      "loss": 0.2729,
+      "step": 47656
+    },
+    {
+      "epoch": 3.860742060920285,
+      "grad_norm": 0.07852333039045334,
+      "learning_rate": 7.736621810162474e-06,
+      "loss": 0.2527,
+      "step": 47657
+    },
+    {
+      "epoch": 3.8608230719377836,
+      "grad_norm": 0.0972912460565567,
+      "learning_rate": 7.732121157567848e-06,
+      "loss": 0.2353,
+      "step": 47658
+    },
+    {
+      "epoch": 3.860904082955282,
+      "grad_norm": 0.05944295972585678,
+      "learning_rate": 7.727620504973222e-06,
+      "loss": 0.1901,
+      "step": 47659
+    },
+    {
+      "epoch": 3.86098509397278,
+      "grad_norm": 0.06972935795783997,
+      "learning_rate": 7.723119852378595e-06,
+      "loss": 0.2063,
+      "step": 47660
+    },
+    {
+      "epoch": 3.861066104990279,
+      "grad_norm": 0.07527864724397659,
+      "learning_rate": 7.718619199783969e-06,
+      "loss": 0.2041,
+      "step": 47661
+    },
+    {
+      "epoch": 3.861147116007777,
+      "grad_norm": 0.06941043585538864,
+      "learning_rate": 7.714118547189342e-06,
+      "loss": 0.2499,
+      "step": 47662
+    },
+    {
+      "epoch": 3.8612281270252753,
+      "grad_norm": 0.06668348610401154,
+      "learning_rate": 7.709617894594718e-06,
+      "loss": 0.2021,
+      "step": 47663
+    },
+    {
+      "epoch": 3.861309138042774,
+      "grad_norm": 0.06672746688127518,
+      "learning_rate": 7.70511724200009e-06,
+      "loss": 0.2108,
+      "step": 47664
+    },
+    {
+      "epoch": 3.8613901490602722,
+      "grad_norm": 0.07876428961753845,
+      "learning_rate": 7.700616589405465e-06,
+      "loss": 0.2696,
+      "step": 47665
+    },
+    {
+      "epoch": 3.8614711600777705,
+      "grad_norm": 0.07343681901693344,
+      "learning_rate": 7.696115936810838e-06,
+      "loss": 0.2279,
+      "step": 47666
+    },
+    {
+      "epoch": 3.861552171095269,
+      "grad_norm": 0.06768275797367096,
+      "learning_rate": 7.691615284216212e-06,
+      "loss": 0.233,
+      "step": 47667
+    },
+    {
+      "epoch": 3.8616331821127674,
+      "grad_norm": 0.07021360844373703,
+      "learning_rate": 7.687114631621586e-06,
+      "loss": 0.2214,
+      "step": 47668
+    },
+    {
+      "epoch": 3.8617141931302656,
+      "grad_norm": 0.06176622211933136,
+      "learning_rate": 7.68261397902696e-06,
+      "loss": 0.2481,
+      "step": 47669
+    },
+    {
+      "epoch": 3.8617952041477643,
+      "grad_norm": 0.08668699860572815,
+      "learning_rate": 7.678113326432333e-06,
+      "loss": 0.2582,
+      "step": 47670
+    },
+    {
+      "epoch": 3.8618762151652626,
+      "grad_norm": 0.07205960154533386,
+      "learning_rate": 7.673612673837706e-06,
+      "loss": 0.2159,
+      "step": 47671
+    },
+    {
+      "epoch": 3.861957226182761,
+      "grad_norm": 0.0772663801908493,
+      "learning_rate": 7.66911202124308e-06,
+      "loss": 0.2653,
+      "step": 47672
+    },
+    {
+      "epoch": 3.862038237200259,
+      "grad_norm": 0.10105203837156296,
+      "learning_rate": 7.664611368648454e-06,
+      "loss": 0.2585,
+      "step": 47673
+    },
+    {
+      "epoch": 3.8621192482177578,
+      "grad_norm": 0.06276296079158783,
+      "learning_rate": 7.660110716053829e-06,
+      "loss": 0.1984,
+      "step": 47674
+    },
+    {
+      "epoch": 3.862200259235256,
+      "grad_norm": 0.06691402941942215,
+      "learning_rate": 7.655610063459203e-06,
+      "loss": 0.2076,
+      "step": 47675
+    },
+    {
+      "epoch": 3.8622812702527543,
+      "grad_norm": 0.06842536479234695,
+      "learning_rate": 7.651109410864575e-06,
+      "loss": 0.2234,
+      "step": 47676
+    },
+    {
+      "epoch": 3.8623622812702525,
+      "grad_norm": 0.08435767143964767,
+      "learning_rate": 7.64660875826995e-06,
+      "loss": 0.2745,
+      "step": 47677
+    },
+    {
+      "epoch": 3.862443292287751,
+      "grad_norm": 0.06818706542253494,
+      "learning_rate": 7.642108105675323e-06,
+      "loss": 0.2251,
+      "step": 47678
+    },
+    {
+      "epoch": 3.8625243033052494,
+      "grad_norm": 0.07116655260324478,
+      "learning_rate": 7.637607453080697e-06,
+      "loss": 0.2734,
+      "step": 47679
+    },
+    {
+      "epoch": 3.8626053143227477,
+      "grad_norm": 0.06385378539562225,
+      "learning_rate": 7.63310680048607e-06,
+      "loss": 0.2295,
+      "step": 47680
+    },
+    {
+      "epoch": 3.8626863253402464,
+      "grad_norm": 0.07078246772289276,
+      "learning_rate": 7.628606147891444e-06,
+      "loss": 0.2238,
+      "step": 47681
+    },
+    {
+      "epoch": 3.8627673363577446,
+      "grad_norm": 0.09647151082754135,
+      "learning_rate": 7.624105495296819e-06,
+      "loss": 0.2137,
+      "step": 47682
+    },
+    {
+      "epoch": 3.862848347375243,
+      "grad_norm": 0.07162981480360031,
+      "learning_rate": 7.619604842702192e-06,
+      "loss": 0.2156,
+      "step": 47683
+    },
+    {
+      "epoch": 3.8629293583927415,
+      "grad_norm": 0.06130341812968254,
+      "learning_rate": 7.615104190107567e-06,
+      "loss": 0.2207,
+      "step": 47684
+    },
+    {
+      "epoch": 3.86301036941024,
+      "grad_norm": 0.0788789615035057,
+      "learning_rate": 7.6106035375129395e-06,
+      "loss": 0.2426,
+      "step": 47685
+    },
+    {
+      "epoch": 3.863091380427738,
+      "grad_norm": 0.07394041866064072,
+      "learning_rate": 7.606102884918313e-06,
+      "loss": 0.2206,
+      "step": 47686
+    },
+    {
+      "epoch": 3.8631723914452367,
+      "grad_norm": 0.07278414815664291,
+      "learning_rate": 7.6016022323236875e-06,
+      "loss": 0.2395,
+      "step": 47687
+    },
+    {
+      "epoch": 3.863253402462735,
+      "grad_norm": 0.07134126871824265,
+      "learning_rate": 7.597101579729061e-06,
+      "loss": 0.2305,
+      "step": 47688
+    },
+    {
+      "epoch": 3.863334413480233,
+      "grad_norm": 0.07982486486434937,
+      "learning_rate": 7.592600927134434e-06,
+      "loss": 0.2611,
+      "step": 47689
+    },
+    {
+      "epoch": 3.863415424497732,
+      "grad_norm": 0.06409557163715363,
+      "learning_rate": 7.588100274539808e-06,
+      "loss": 0.2097,
+      "step": 47690
+    },
+    {
+      "epoch": 3.86349643551523,
+      "grad_norm": 0.0836891233921051,
+      "learning_rate": 7.583599621945183e-06,
+      "loss": 0.214,
+      "step": 47691
+    },
+    {
+      "epoch": 3.8635774465327284,
+      "grad_norm": 0.06255742907524109,
+      "learning_rate": 7.579098969350556e-06,
+      "loss": 0.2042,
+      "step": 47692
+    },
+    {
+      "epoch": 3.863658457550227,
+      "grad_norm": 0.08783721923828125,
+      "learning_rate": 7.574598316755931e-06,
+      "loss": 0.2644,
+      "step": 47693
+    },
+    {
+      "epoch": 3.8637394685677253,
+      "grad_norm": 0.05096682161092758,
+      "learning_rate": 7.570097664161304e-06,
+      "loss": 0.166,
+      "step": 47694
+    },
+    {
+      "epoch": 3.8638204795852236,
+      "grad_norm": 0.07620611041784286,
+      "learning_rate": 7.565597011566677e-06,
+      "loss": 0.2563,
+      "step": 47695
+    },
+    {
+      "epoch": 3.863901490602722,
+      "grad_norm": 0.0640488713979721,
+      "learning_rate": 7.561096358972052e-06,
+      "loss": 0.2226,
+      "step": 47696
+    },
+    {
+      "epoch": 3.8639825016202205,
+      "grad_norm": 0.05692597106099129,
+      "learning_rate": 7.556595706377425e-06,
+      "loss": 0.2158,
+      "step": 47697
+    },
+    {
+      "epoch": 3.8640635126377187,
+      "grad_norm": 0.08596763014793396,
+      "learning_rate": 7.552095053782798e-06,
+      "loss": 0.2586,
+      "step": 47698
+    },
+    {
+      "epoch": 3.864144523655217,
+      "grad_norm": 0.06715045869350433,
+      "learning_rate": 7.5475944011881725e-06,
+      "loss": 0.2337,
+      "step": 47699
+    },
+    {
+      "epoch": 3.8642255346727152,
+      "grad_norm": 0.06749710440635681,
+      "learning_rate": 7.543093748593547e-06,
+      "loss": 0.2108,
+      "step": 47700
+    },
+    {
+      "epoch": 3.864306545690214,
+      "grad_norm": 0.08679332584142685,
+      "learning_rate": 7.5385930959989205e-06,
+      "loss": 0.2405,
+      "step": 47701
+    },
+    {
+      "epoch": 3.864387556707712,
+      "grad_norm": 0.07425546646118164,
+      "learning_rate": 7.534092443404293e-06,
+      "loss": 0.2116,
+      "step": 47702
+    },
+    {
+      "epoch": 3.8644685677252104,
+      "grad_norm": 0.07318223267793655,
+      "learning_rate": 7.529591790809668e-06,
+      "loss": 0.2119,
+      "step": 47703
+    },
+    {
+      "epoch": 3.864549578742709,
+      "grad_norm": 0.0675434023141861,
+      "learning_rate": 7.525091138215041e-06,
+      "loss": 0.2143,
+      "step": 47704
+    },
+    {
+      "epoch": 3.8646305897602073,
+      "grad_norm": 0.07143059372901917,
+      "learning_rate": 7.520590485620416e-06,
+      "loss": 0.2349,
+      "step": 47705
+    },
+    {
+      "epoch": 3.8647116007777056,
+      "grad_norm": 0.07418014854192734,
+      "learning_rate": 7.516089833025789e-06,
+      "loss": 0.2372,
+      "step": 47706
+    },
+    {
+      "epoch": 3.8647926117952043,
+      "grad_norm": 0.0735107809305191,
+      "learning_rate": 7.511589180431162e-06,
+      "loss": 0.2023,
+      "step": 47707
+    },
+    {
+      "epoch": 3.8648736228127025,
+      "grad_norm": 0.06702899187803268,
+      "learning_rate": 7.507088527836537e-06,
+      "loss": 0.2527,
+      "step": 47708
+    },
+    {
+      "epoch": 3.8649546338302008,
+      "grad_norm": 0.0680052861571312,
+      "learning_rate": 7.502587875241911e-06,
+      "loss": 0.2181,
+      "step": 47709
+    },
+    {
+      "epoch": 3.8650356448476995,
+      "grad_norm": 0.07556014508008957,
+      "learning_rate": 7.498087222647285e-06,
+      "loss": 0.1964,
+      "step": 47710
+    },
+    {
+      "epoch": 3.8651166558651977,
+      "grad_norm": 0.06902261078357697,
+      "learning_rate": 7.493586570052657e-06,
+      "loss": 0.212,
+      "step": 47711
+    },
+    {
+      "epoch": 3.865197666882696,
+      "grad_norm": 0.07958406209945679,
+      "learning_rate": 7.489085917458032e-06,
+      "loss": 0.2346,
+      "step": 47712
+    },
+    {
+      "epoch": 3.8652786779001946,
+      "grad_norm": 0.076631560921669,
+      "learning_rate": 7.4845852648634055e-06,
+      "loss": 0.2123,
+      "step": 47713
+    },
+    {
+      "epoch": 3.865359688917693,
+      "grad_norm": 0.06917714327573776,
+      "learning_rate": 7.48008461226878e-06,
+      "loss": 0.2468,
+      "step": 47714
+    },
+    {
+      "epoch": 3.865440699935191,
+      "grad_norm": 0.08754683285951614,
+      "learning_rate": 7.475583959674153e-06,
+      "loss": 0.2226,
+      "step": 47715
+    },
+    {
+      "epoch": 3.86552171095269,
+      "grad_norm": 0.07786770164966583,
+      "learning_rate": 7.471083307079526e-06,
+      "loss": 0.2222,
+      "step": 47716
+    },
+    {
+      "epoch": 3.865602721970188,
+      "grad_norm": 0.06999102234840393,
+      "learning_rate": 7.466582654484901e-06,
+      "loss": 0.2052,
+      "step": 47717
+    },
+    {
+      "epoch": 3.8656837329876863,
+      "grad_norm": 0.0734010711312294,
+      "learning_rate": 7.462082001890275e-06,
+      "loss": 0.1994,
+      "step": 47718
+    },
+    {
+      "epoch": 3.8657647440051845,
+      "grad_norm": 0.07305295020341873,
+      "learning_rate": 7.457581349295648e-06,
+      "loss": 0.2284,
+      "step": 47719
+    },
+    {
+      "epoch": 3.8658457550226832,
+      "grad_norm": 0.07487478107213974,
+      "learning_rate": 7.4530806967010215e-06,
+      "loss": 0.2056,
+      "step": 47720
+    },
+    {
+      "epoch": 3.8659267660401815,
+      "grad_norm": 0.07713651657104492,
+      "learning_rate": 7.448580044106396e-06,
+      "loss": 0.2125,
+      "step": 47721
+    },
+    {
+      "epoch": 3.8660077770576797,
+      "grad_norm": 0.07192755490541458,
+      "learning_rate": 7.44407939151177e-06,
+      "loss": 0.2273,
+      "step": 47722
+    },
+    {
+      "epoch": 3.866088788075178,
+      "grad_norm": 0.07173226028680801,
+      "learning_rate": 7.439578738917144e-06,
+      "loss": 0.2011,
+      "step": 47723
+    },
+    {
+      "epoch": 3.8661697990926767,
+      "grad_norm": 0.09787497669458389,
+      "learning_rate": 7.435078086322517e-06,
+      "loss": 0.2243,
+      "step": 47724
+    },
+    {
+      "epoch": 3.866250810110175,
+      "grad_norm": 0.09983297437429428,
+      "learning_rate": 7.43057743372789e-06,
+      "loss": 0.277,
+      "step": 47725
+    },
+    {
+      "epoch": 3.866331821127673,
+      "grad_norm": 0.0823056623339653,
+      "learning_rate": 7.426076781133265e-06,
+      "loss": 0.2216,
+      "step": 47726
+    },
+    {
+      "epoch": 3.866412832145172,
+      "grad_norm": 0.059470660984516144,
+      "learning_rate": 7.421576128538639e-06,
+      "loss": 0.2341,
+      "step": 47727
+    },
+    {
+      "epoch": 3.86649384316267,
+      "grad_norm": 0.06996002048254013,
+      "learning_rate": 7.417075475944012e-06,
+      "loss": 0.2243,
+      "step": 47728
+    },
+    {
+      "epoch": 3.8665748541801683,
+      "grad_norm": 0.07670509815216064,
+      "learning_rate": 7.412574823349386e-06,
+      "loss": 0.1981,
+      "step": 47729
+    },
+    {
+      "epoch": 3.866655865197667,
+      "grad_norm": 0.07600407302379608,
+      "learning_rate": 7.40807417075476e-06,
+      "loss": 0.2125,
+      "step": 47730
+    },
+    {
+      "epoch": 3.8667368762151653,
+      "grad_norm": 0.0708894208073616,
+      "learning_rate": 7.403573518160134e-06,
+      "loss": 0.2284,
+      "step": 47731
+    },
+    {
+      "epoch": 3.8668178872326635,
+      "grad_norm": 0.07632531970739365,
+      "learning_rate": 7.3990728655655065e-06,
+      "loss": 0.2346,
+      "step": 47732
+    },
+    {
+      "epoch": 3.866898898250162,
+      "grad_norm": 0.06774915009737015,
+      "learning_rate": 7.394572212970881e-06,
+      "loss": 0.2466,
+      "step": 47733
+    },
+    {
+      "epoch": 3.8669799092676604,
+      "grad_norm": 0.07281273603439331,
+      "learning_rate": 7.3900715603762545e-06,
+      "loss": 0.2315,
+      "step": 47734
+    },
+    {
+      "epoch": 3.8670609202851587,
+      "grad_norm": 0.06819894909858704,
+      "learning_rate": 7.385570907781629e-06,
+      "loss": 0.2103,
+      "step": 47735
+    },
+    {
+      "epoch": 3.8671419313026574,
+      "grad_norm": 0.07330294698476791,
+      "learning_rate": 7.381070255187003e-06,
+      "loss": 0.2339,
+      "step": 47736
+    },
+    {
+      "epoch": 3.8672229423201556,
+      "grad_norm": 0.0813525915145874,
+      "learning_rate": 7.376569602592376e-06,
+      "loss": 0.2204,
+      "step": 47737
+    },
+    {
+      "epoch": 3.867303953337654,
+      "grad_norm": 0.05763068050146103,
+      "learning_rate": 7.37206894999775e-06,
+      "loss": 0.2022,
+      "step": 47738
+    },
+    {
+      "epoch": 3.8673849643551526,
+      "grad_norm": 0.07878169417381287,
+      "learning_rate": 7.367568297403124e-06,
+      "loss": 0.2099,
+      "step": 47739
+    },
+    {
+      "epoch": 3.867465975372651,
+      "grad_norm": 0.06456834822893143,
+      "learning_rate": 7.363067644808498e-06,
+      "loss": 0.2102,
+      "step": 47740
+    },
+    {
+      "epoch": 3.867546986390149,
+      "grad_norm": 0.06957262009382248,
+      "learning_rate": 7.358566992213871e-06,
+      "loss": 0.2178,
+      "step": 47741
+    },
+    {
+      "epoch": 3.8676279974076473,
+      "grad_norm": 0.06936120241880417,
+      "learning_rate": 7.354066339619245e-06,
+      "loss": 0.2231,
+      "step": 47742
+    },
+    {
+      "epoch": 3.867709008425146,
+      "grad_norm": 0.08154875040054321,
+      "learning_rate": 7.349565687024619e-06,
+      "loss": 0.2084,
+      "step": 47743
+    },
+    {
+      "epoch": 3.8677900194426442,
+      "grad_norm": 0.07597959786653519,
+      "learning_rate": 7.345065034429993e-06,
+      "loss": 0.2453,
+      "step": 47744
+    },
+    {
+      "epoch": 3.8678710304601425,
+      "grad_norm": 0.05550116300582886,
+      "learning_rate": 7.340564381835366e-06,
+      "loss": 0.2199,
+      "step": 47745
+    },
+    {
+      "epoch": 3.8679520414776407,
+      "grad_norm": 0.06986594945192337,
+      "learning_rate": 7.33606372924074e-06,
+      "loss": 0.2177,
+      "step": 47746
+    },
+    {
+      "epoch": 3.8680330524951394,
+      "grad_norm": 0.09499722719192505,
+      "learning_rate": 7.331563076646114e-06,
+      "loss": 0.2531,
+      "step": 47747
+    },
+    {
+      "epoch": 3.8681140635126376,
+      "grad_norm": 0.07436975091695786,
+      "learning_rate": 7.327062424051488e-06,
+      "loss": 0.2138,
+      "step": 47748
+    },
+    {
+      "epoch": 3.868195074530136,
+      "grad_norm": 0.08323274552822113,
+      "learning_rate": 7.322561771456862e-06,
+      "loss": 0.2353,
+      "step": 47749
+    },
+    {
+      "epoch": 3.8682760855476346,
+      "grad_norm": 0.06948281079530716,
+      "learning_rate": 7.318061118862235e-06,
+      "loss": 0.2377,
+      "step": 47750
+    },
+    {
+      "epoch": 3.868357096565133,
+      "grad_norm": 0.08530833572149277,
+      "learning_rate": 7.313560466267609e-06,
+      "loss": 0.2491,
+      "step": 47751
+    },
+    {
+      "epoch": 3.868438107582631,
+      "grad_norm": 0.07238127291202545,
+      "learning_rate": 7.309059813672984e-06,
+      "loss": 0.2442,
+      "step": 47752
+    },
+    {
+      "epoch": 3.8685191186001298,
+      "grad_norm": 0.06473894417285919,
+      "learning_rate": 7.304559161078357e-06,
+      "loss": 0.2375,
+      "step": 47753
+    },
+    {
+      "epoch": 3.868600129617628,
+      "grad_norm": 0.06814999133348465,
+      "learning_rate": 7.30005850848373e-06,
+      "loss": 0.2127,
+      "step": 47754
+    },
+    {
+      "epoch": 3.8686811406351262,
+      "grad_norm": 0.08416423946619034,
+      "learning_rate": 7.295557855889104e-06,
+      "loss": 0.2159,
+      "step": 47755
+    },
+    {
+      "epoch": 3.868762151652625,
+      "grad_norm": 0.06730145215988159,
+      "learning_rate": 7.291057203294478e-06,
+      "loss": 0.2514,
+      "step": 47756
+    },
+    {
+      "epoch": 3.868843162670123,
+      "grad_norm": 0.061186712235212326,
+      "learning_rate": 7.2865565506998525e-06,
+      "loss": 0.1958,
+      "step": 47757
+    },
+    {
+      "epoch": 3.8689241736876214,
+      "grad_norm": 0.07807473093271255,
+      "learning_rate": 7.282055898105225e-06,
+      "loss": 0.1886,
+      "step": 47758
+    },
+    {
+      "epoch": 3.86900518470512,
+      "grad_norm": 0.07652845978736877,
+      "learning_rate": 7.277555245510599e-06,
+      "loss": 0.2328,
+      "step": 47759
+    },
+    {
+      "epoch": 3.8690861957226184,
+      "grad_norm": 0.07402211427688599,
+      "learning_rate": 7.273054592915973e-06,
+      "loss": 0.2609,
+      "step": 47760
+    },
+    {
+      "epoch": 3.8691672067401166,
+      "grad_norm": 0.05133407190442085,
+      "learning_rate": 7.268553940321348e-06,
+      "loss": 0.2124,
+      "step": 47761
+    },
+    {
+      "epoch": 3.8692482177576153,
+      "grad_norm": 0.08077628910541534,
+      "learning_rate": 7.26405328772672e-06,
+      "loss": 0.2662,
+      "step": 47762
+    },
+    {
+      "epoch": 3.8693292287751135,
+      "grad_norm": 0.06384014338254929,
+      "learning_rate": 7.259552635132094e-06,
+      "loss": 0.2063,
+      "step": 47763
+    },
+    {
+      "epoch": 3.869410239792612,
+      "grad_norm": 0.06897179782390594,
+      "learning_rate": 7.2550519825374685e-06,
+      "loss": 0.227,
+      "step": 47764
+    },
+    {
+      "epoch": 3.86949125081011,
+      "grad_norm": 0.0741807222366333,
+      "learning_rate": 7.250551329942842e-06,
+      "loss": 0.1986,
+      "step": 47765
+    },
+    {
+      "epoch": 3.8695722618276087,
+      "grad_norm": 0.07519097626209259,
+      "learning_rate": 7.246050677348217e-06,
+      "loss": 0.2383,
+      "step": 47766
+    },
+    {
+      "epoch": 3.869653272845107,
+      "grad_norm": 0.07148484885692596,
+      "learning_rate": 7.241550024753589e-06,
+      "loss": 0.2481,
+      "step": 47767
+    },
+    {
+      "epoch": 3.869734283862605,
+      "grad_norm": 0.07364317774772644,
+      "learning_rate": 7.237049372158963e-06,
+      "loss": 0.2098,
+      "step": 47768
+    },
+    {
+      "epoch": 3.8698152948801035,
+      "grad_norm": 0.07559014856815338,
+      "learning_rate": 7.232548719564337e-06,
+      "loss": 0.2394,
+      "step": 47769
+    },
+    {
+      "epoch": 3.869896305897602,
+      "grad_norm": 0.07173273712396622,
+      "learning_rate": 7.228048066969712e-06,
+      "loss": 0.2245,
+      "step": 47770
+    },
+    {
+      "epoch": 3.8699773169151004,
+      "grad_norm": 0.07271765172481537,
+      "learning_rate": 7.223547414375085e-06,
+      "loss": 0.1872,
+      "step": 47771
+    },
+    {
+      "epoch": 3.8700583279325986,
+      "grad_norm": 0.06458864361047745,
+      "learning_rate": 7.219046761780458e-06,
+      "loss": 0.2289,
+      "step": 47772
+    },
+    {
+      "epoch": 3.8701393389500973,
+      "grad_norm": 0.07476481795310974,
+      "learning_rate": 7.214546109185833e-06,
+      "loss": 0.2548,
+      "step": 47773
+    },
+    {
+      "epoch": 3.8702203499675956,
+      "grad_norm": 0.07546652108430862,
+      "learning_rate": 7.210045456591206e-06,
+      "loss": 0.2514,
+      "step": 47774
+    },
+    {
+      "epoch": 3.870301360985094,
+      "grad_norm": 0.06332651525735855,
+      "learning_rate": 7.205544803996579e-06,
+      "loss": 0.1829,
+      "step": 47775
+    },
+    {
+      "epoch": 3.8703823720025925,
+      "grad_norm": 0.07313700765371323,
+      "learning_rate": 7.2010441514019535e-06,
+      "loss": 0.2082,
+      "step": 47776
+    },
+    {
+      "epoch": 3.8704633830200907,
+      "grad_norm": 0.06601651757955551,
+      "learning_rate": 7.196543498807327e-06,
+      "loss": 0.2213,
+      "step": 47777
+    },
+    {
+      "epoch": 3.870544394037589,
+      "grad_norm": 0.07746600359678268,
+      "learning_rate": 7.1920428462127015e-06,
+      "loss": 0.2392,
+      "step": 47778
+    },
+    {
+      "epoch": 3.8706254050550877,
+      "grad_norm": 0.08070232719182968,
+      "learning_rate": 7.187542193618076e-06,
+      "loss": 0.2264,
+      "step": 47779
+    },
+    {
+      "epoch": 3.870706416072586,
+      "grad_norm": 0.06326809525489807,
+      "learning_rate": 7.183041541023449e-06,
+      "loss": 0.2145,
+      "step": 47780
+    },
+    {
+      "epoch": 3.870787427090084,
+      "grad_norm": 0.058843906968832016,
+      "learning_rate": 7.178540888428822e-06,
+      "loss": 0.1969,
+      "step": 47781
+    },
+    {
+      "epoch": 3.870868438107583,
+      "grad_norm": 0.05918506532907486,
+      "learning_rate": 7.174040235834197e-06,
+      "loss": 0.2175,
+      "step": 47782
+    },
+    {
+      "epoch": 3.870949449125081,
+      "grad_norm": 0.071584552526474,
+      "learning_rate": 7.16953958323957e-06,
+      "loss": 0.1764,
+      "step": 47783
+    },
+    {
+      "epoch": 3.8710304601425793,
+      "grad_norm": 0.05277230218052864,
+      "learning_rate": 7.165038930644943e-06,
+      "loss": 0.2319,
+      "step": 47784
+    },
+    {
+      "epoch": 3.871111471160078,
+      "grad_norm": 0.07672587037086487,
+      "learning_rate": 7.160538278050318e-06,
+      "loss": 0.209,
+      "step": 47785
+    },
+    {
+      "epoch": 3.8711924821775763,
+      "grad_norm": 0.08189419656991959,
+      "learning_rate": 7.156037625455691e-06,
+      "loss": 0.2412,
+      "step": 47786
+    },
+    {
+      "epoch": 3.8712734931950745,
+      "grad_norm": 0.06885275989770889,
+      "learning_rate": 7.151536972861066e-06,
+      "loss": 0.2351,
+      "step": 47787
+    },
+    {
+      "epoch": 3.8713545042125728,
+      "grad_norm": 0.07223542034626007,
+      "learning_rate": 7.147036320266438e-06,
+      "loss": 0.2375,
+      "step": 47788
+    },
+    {
+      "epoch": 3.8714355152300715,
+      "grad_norm": 0.07468521595001221,
+      "learning_rate": 7.142535667671813e-06,
+      "loss": 0.2284,
+      "step": 47789
+    },
+    {
+      "epoch": 3.8715165262475697,
+      "grad_norm": 0.06014881283044815,
+      "learning_rate": 7.1380350150771865e-06,
+      "loss": 0.1972,
+      "step": 47790
+    },
+    {
+      "epoch": 3.871597537265068,
+      "grad_norm": 0.0729193314909935,
+      "learning_rate": 7.133534362482561e-06,
+      "loss": 0.2033,
+      "step": 47791
+    },
+    {
+      "epoch": 3.871678548282566,
+      "grad_norm": 0.0688919946551323,
+      "learning_rate": 7.129033709887934e-06,
+      "loss": 0.2597,
+      "step": 47792
+    },
+    {
+      "epoch": 3.871759559300065,
+      "grad_norm": 0.08578962832689285,
+      "learning_rate": 7.124533057293307e-06,
+      "loss": 0.2209,
+      "step": 47793
+    },
+    {
+      "epoch": 3.871840570317563,
+      "grad_norm": 0.06939221173524857,
+      "learning_rate": 7.120032404698682e-06,
+      "loss": 0.2107,
+      "step": 47794
+    },
+    {
+      "epoch": 3.8719215813350614,
+      "grad_norm": 0.07219339162111282,
+      "learning_rate": 7.115531752104055e-06,
+      "loss": 0.2238,
+      "step": 47795
+    },
+    {
+      "epoch": 3.87200259235256,
+      "grad_norm": 0.06591325253248215,
+      "learning_rate": 7.11103109950943e-06,
+      "loss": 0.2357,
+      "step": 47796
+    },
+    {
+      "epoch": 3.8720836033700583,
+      "grad_norm": 0.06282512098550797,
+      "learning_rate": 7.1065304469148025e-06,
+      "loss": 0.209,
+      "step": 47797
+    },
+    {
+      "epoch": 3.8721646143875565,
+      "grad_norm": 0.07997066527605057,
+      "learning_rate": 7.102029794320177e-06,
+      "loss": 0.2465,
+      "step": 47798
+    },
+    {
+      "epoch": 3.8722456254050552,
+      "grad_norm": 0.08144117146730423,
+      "learning_rate": 7.097529141725551e-06,
+      "loss": 0.2519,
+      "step": 47799
+    },
+    {
+      "epoch": 3.8723266364225535,
+      "grad_norm": 0.06622984260320663,
+      "learning_rate": 7.093028489130925e-06,
+      "loss": 0.1985,
+      "step": 47800
+    },
+    {
+      "epoch": 3.8724076474400517,
+      "grad_norm": 0.06841832399368286,
+      "learning_rate": 7.088527836536298e-06,
+      "loss": 0.2252,
+      "step": 47801
+    },
+    {
+      "epoch": 3.8724886584575504,
+      "grad_norm": 0.06852193176746368,
+      "learning_rate": 7.084027183941671e-06,
+      "loss": 0.2244,
+      "step": 47802
+    },
+    {
+      "epoch": 3.8725696694750487,
+      "grad_norm": 0.09205551445484161,
+      "learning_rate": 7.079526531347046e-06,
+      "loss": 0.2398,
+      "step": 47803
+    },
+    {
+      "epoch": 3.872650680492547,
+      "grad_norm": 0.07046569138765335,
+      "learning_rate": 7.0750258787524195e-06,
+      "loss": 0.2459,
+      "step": 47804
+    },
+    {
+      "epoch": 3.8727316915100456,
+      "grad_norm": 0.0705006867647171,
+      "learning_rate": 7.070525226157792e-06,
+      "loss": 0.2379,
+      "step": 47805
+    },
+    {
+      "epoch": 3.872812702527544,
+      "grad_norm": 0.06745926290750504,
+      "learning_rate": 7.066024573563167e-06,
+      "loss": 0.2444,
+      "step": 47806
+    },
+    {
+      "epoch": 3.872893713545042,
+      "grad_norm": 0.06479792296886444,
+      "learning_rate": 7.061523920968541e-06,
+      "loss": 0.1962,
+      "step": 47807
+    },
+    {
+      "epoch": 3.8729747245625408,
+      "grad_norm": 0.07828095555305481,
+      "learning_rate": 7.057023268373915e-06,
+      "loss": 0.2304,
+      "step": 47808
+    },
+    {
+      "epoch": 3.873055735580039,
+      "grad_norm": 0.05744514614343643,
+      "learning_rate": 7.052522615779289e-06,
+      "loss": 0.2172,
+      "step": 47809
+    },
+    {
+      "epoch": 3.8731367465975373,
+      "grad_norm": 0.06972257792949677,
+      "learning_rate": 7.048021963184662e-06,
+      "loss": 0.2432,
+      "step": 47810
+    },
+    {
+      "epoch": 3.8732177576150355,
+      "grad_norm": 0.058357443660497665,
+      "learning_rate": 7.0435213105900355e-06,
+      "loss": 0.1878,
+      "step": 47811
+    },
+    {
+      "epoch": 3.8732987686325338,
+      "grad_norm": 0.06073423847556114,
+      "learning_rate": 7.03902065799541e-06,
+      "loss": 0.2022,
+      "step": 47812
+    },
+    {
+      "epoch": 3.8733797796500324,
+      "grad_norm": 0.059745658189058304,
+      "learning_rate": 7.034520005400784e-06,
+      "loss": 0.1826,
+      "step": 47813
+    },
+    {
+      "epoch": 3.8734607906675307,
+      "grad_norm": 0.07808952778577805,
+      "learning_rate": 7.030019352806156e-06,
+      "loss": 0.2375,
+      "step": 47814
+    },
+    {
+      "epoch": 3.873541801685029,
+      "grad_norm": 0.07170235365629196,
+      "learning_rate": 7.025518700211531e-06,
+      "loss": 0.2244,
+      "step": 47815
+    },
+    {
+      "epoch": 3.8736228127025276,
+      "grad_norm": 0.07987401634454727,
+      "learning_rate": 7.021018047616905e-06,
+      "loss": 0.2086,
+      "step": 47816
+    },
+    {
+      "epoch": 3.873703823720026,
+      "grad_norm": 0.0680292397737503,
+      "learning_rate": 7.016517395022279e-06,
+      "loss": 0.2353,
+      "step": 47817
+    },
+    {
+      "epoch": 3.873784834737524,
+      "grad_norm": 0.06402581185102463,
+      "learning_rate": 7.012016742427652e-06,
+      "loss": 0.1892,
+      "step": 47818
+    },
+    {
+      "epoch": 3.873865845755023,
+      "grad_norm": 0.07108760625123978,
+      "learning_rate": 7.007516089833026e-06,
+      "loss": 0.2752,
+      "step": 47819
+    },
+    {
+      "epoch": 3.873946856772521,
+      "grad_norm": 0.07389482855796814,
+      "learning_rate": 7.0030154372384e-06,
+      "loss": 0.2215,
+      "step": 47820
+    },
+    {
+      "epoch": 3.8740278677900193,
+      "grad_norm": 0.07002807408571243,
+      "learning_rate": 6.998514784643774e-06,
+      "loss": 0.2196,
+      "step": 47821
+    },
+    {
+      "epoch": 3.874108878807518,
+      "grad_norm": 0.07865195721387863,
+      "learning_rate": 6.9940141320491485e-06,
+      "loss": 0.2383,
+      "step": 47822
+    },
+    {
+      "epoch": 3.874189889825016,
+      "grad_norm": 0.07489572465419769,
+      "learning_rate": 6.9895134794545205e-06,
+      "loss": 0.2034,
+      "step": 47823
+    },
+    {
+      "epoch": 3.8742709008425145,
+      "grad_norm": 0.0735786184668541,
+      "learning_rate": 6.985012826859895e-06,
+      "loss": 0.2558,
+      "step": 47824
+    },
+    {
+      "epoch": 3.874351911860013,
+      "grad_norm": 0.07100185006856918,
+      "learning_rate": 6.980512174265269e-06,
+      "loss": 0.2405,
+      "step": 47825
+    },
+    {
+      "epoch": 3.8744329228775114,
+      "grad_norm": 0.06581882387399673,
+      "learning_rate": 6.976011521670643e-06,
+      "loss": 0.1944,
+      "step": 47826
+    },
+    {
+      "epoch": 3.8745139338950096,
+      "grad_norm": 0.06488057225942612,
+      "learning_rate": 6.971510869076016e-06,
+      "loss": 0.224,
+      "step": 47827
+    },
+    {
+      "epoch": 3.8745949449125083,
+      "grad_norm": 0.07766662538051605,
+      "learning_rate": 6.96701021648139e-06,
+      "loss": 0.234,
+      "step": 47828
+    },
+    {
+      "epoch": 3.8746759559300066,
+      "grad_norm": 0.06239504739642143,
+      "learning_rate": 6.962509563886764e-06,
+      "loss": 0.1958,
+      "step": 47829
+    },
+    {
+      "epoch": 3.874756966947505,
+      "grad_norm": 0.06142323091626167,
+      "learning_rate": 6.958008911292138e-06,
+      "loss": 0.219,
+      "step": 47830
+    },
+    {
+      "epoch": 3.8748379779650035,
+      "grad_norm": 0.07786870002746582,
+      "learning_rate": 6.953508258697511e-06,
+      "loss": 0.239,
+      "step": 47831
+    },
+    {
+      "epoch": 3.8749189889825018,
+      "grad_norm": 0.062448933720588684,
+      "learning_rate": 6.949007606102885e-06,
+      "loss": 0.2096,
+      "step": 47832
+    },
+    {
+      "epoch": 3.875,
+      "grad_norm": 0.05940355360507965,
+      "learning_rate": 6.944506953508259e-06,
+      "loss": 0.1905,
+      "step": 47833
+    },
+    {
+      "epoch": 3.8750810110174982,
+      "grad_norm": 0.05543764308094978,
+      "learning_rate": 6.9400063009136335e-06,
+      "loss": 0.1915,
+      "step": 47834
+    },
+    {
+      "epoch": 3.8751620220349965,
+      "grad_norm": 0.08461619913578033,
+      "learning_rate": 6.935505648319006e-06,
+      "loss": 0.2633,
+      "step": 47835
+    },
+    {
+      "epoch": 3.875243033052495,
+      "grad_norm": 0.0797097235918045,
+      "learning_rate": 6.93100499572438e-06,
+      "loss": 0.2521,
+      "step": 47836
+    },
+    {
+      "epoch": 3.8753240440699934,
+      "grad_norm": 0.07515867799520493,
+      "learning_rate": 6.926504343129754e-06,
+      "loss": 0.2429,
+      "step": 47837
+    },
+    {
+      "epoch": 3.8754050550874917,
+      "grad_norm": 0.0636037066578865,
+      "learning_rate": 6.922003690535128e-06,
+      "loss": 0.1941,
+      "step": 47838
+    },
+    {
+      "epoch": 3.8754860661049904,
+      "grad_norm": 0.0706285685300827,
+      "learning_rate": 6.917503037940502e-06,
+      "loss": 0.2353,
+      "step": 47839
+    },
+    {
+      "epoch": 3.8755670771224886,
+      "grad_norm": 0.0567207969725132,
+      "learning_rate": 6.913002385345875e-06,
+      "loss": 0.1933,
+      "step": 47840
+    },
+    {
+      "epoch": 3.875648088139987,
+      "grad_norm": 0.07477942109107971,
+      "learning_rate": 6.908501732751249e-06,
+      "loss": 0.2183,
+      "step": 47841
+    },
+    {
+      "epoch": 3.8757290991574855,
+      "grad_norm": 0.057915329933166504,
+      "learning_rate": 6.904001080156623e-06,
+      "loss": 0.2068,
+      "step": 47842
+    },
+    {
+      "epoch": 3.875810110174984,
+      "grad_norm": 0.0781538337469101,
+      "learning_rate": 6.899500427561998e-06,
+      "loss": 0.1967,
+      "step": 47843
+    },
+    {
+      "epoch": 3.875891121192482,
+      "grad_norm": 0.07939945161342621,
+      "learning_rate": 6.89499977496737e-06,
+      "loss": 0.2356,
+      "step": 47844
+    },
+    {
+      "epoch": 3.8759721322099807,
+      "grad_norm": 0.05997350439429283,
+      "learning_rate": 6.890499122372744e-06,
+      "loss": 0.2141,
+      "step": 47845
+    },
+    {
+      "epoch": 3.876053143227479,
+      "grad_norm": 0.06055265665054321,
+      "learning_rate": 6.885998469778118e-06,
+      "loss": 0.2461,
+      "step": 47846
+    },
+    {
+      "epoch": 3.876134154244977,
+      "grad_norm": 0.07572788745164871,
+      "learning_rate": 6.881497817183492e-06,
+      "loss": 0.2126,
+      "step": 47847
+    },
+    {
+      "epoch": 3.876215165262476,
+      "grad_norm": 0.06472624838352203,
+      "learning_rate": 6.876997164588865e-06,
+      "loss": 0.189,
+      "step": 47848
+    },
+    {
+      "epoch": 3.876296176279974,
+      "grad_norm": 0.08194227516651154,
+      "learning_rate": 6.872496511994239e-06,
+      "loss": 0.1988,
+      "step": 47849
+    },
+    {
+      "epoch": 3.8763771872974724,
+      "grad_norm": 0.08095631748437881,
+      "learning_rate": 6.867995859399614e-06,
+      "loss": 0.2483,
+      "step": 47850
+    },
+    {
+      "epoch": 3.876458198314971,
+      "grad_norm": 0.075938880443573,
+      "learning_rate": 6.863495206804987e-06,
+      "loss": 0.2077,
+      "step": 47851
+    },
+    {
+      "epoch": 3.8765392093324693,
+      "grad_norm": 0.07606460154056549,
+      "learning_rate": 6.858994554210362e-06,
+      "loss": 0.2077,
+      "step": 47852
+    },
+    {
+      "epoch": 3.8766202203499676,
+      "grad_norm": 0.06377337127923965,
+      "learning_rate": 6.8544939016157345e-06,
+      "loss": 0.2375,
+      "step": 47853
+    },
+    {
+      "epoch": 3.8767012313674662,
+      "grad_norm": 0.06751275062561035,
+      "learning_rate": 6.849993249021108e-06,
+      "loss": 0.2163,
+      "step": 47854
+    },
+    {
+      "epoch": 3.8767822423849645,
+      "grad_norm": 0.06989561021327972,
+      "learning_rate": 6.8454925964264825e-06,
+      "loss": 0.2183,
+      "step": 47855
+    },
+    {
+      "epoch": 3.8768632534024627,
+      "grad_norm": 0.05857057496905327,
+      "learning_rate": 6.840991943831856e-06,
+      "loss": 0.2215,
+      "step": 47856
+    },
+    {
+      "epoch": 3.876944264419961,
+      "grad_norm": 0.09082645177841187,
+      "learning_rate": 6.836491291237229e-06,
+      "loss": 0.2536,
+      "step": 47857
+    },
+    {
+      "epoch": 3.8770252754374592,
+      "grad_norm": 0.0694134309887886,
+      "learning_rate": 6.831990638642603e-06,
+      "loss": 0.196,
+      "step": 47858
+    },
+    {
+      "epoch": 3.877106286454958,
+      "grad_norm": 0.07403349131345749,
+      "learning_rate": 6.827489986047978e-06,
+      "loss": 0.2341,
+      "step": 47859
+    },
+    {
+      "epoch": 3.877187297472456,
+      "grad_norm": 0.061810147017240524,
+      "learning_rate": 6.822989333453351e-06,
+      "loss": 0.2047,
+      "step": 47860
+    },
+    {
+      "epoch": 3.8772683084899544,
+      "grad_norm": 0.07295320928096771,
+      "learning_rate": 6.818488680858724e-06,
+      "loss": 0.257,
+      "step": 47861
+    },
+    {
+      "epoch": 3.877349319507453,
+      "grad_norm": 0.06367763131856918,
+      "learning_rate": 6.813988028264099e-06,
+      "loss": 0.2078,
+      "step": 47862
+    },
+    {
+      "epoch": 3.8774303305249513,
+      "grad_norm": 0.084407739341259,
+      "learning_rate": 6.809487375669472e-06,
+      "loss": 0.1852,
+      "step": 47863
+    },
+    {
+      "epoch": 3.8775113415424496,
+      "grad_norm": 0.07967525720596313,
+      "learning_rate": 6.804986723074847e-06,
+      "loss": 0.2155,
+      "step": 47864
+    },
+    {
+      "epoch": 3.8775923525599483,
+      "grad_norm": 0.0668228268623352,
+      "learning_rate": 6.80048607048022e-06,
+      "loss": 0.2117,
+      "step": 47865
+    },
+    {
+      "epoch": 3.8776733635774465,
+      "grad_norm": 0.06333678960800171,
+      "learning_rate": 6.795985417885593e-06,
+      "loss": 0.1996,
+      "step": 47866
+    },
+    {
+      "epoch": 3.8777543745949448,
+      "grad_norm": 0.07337368279695511,
+      "learning_rate": 6.7914847652909675e-06,
+      "loss": 0.2246,
+      "step": 47867
+    },
+    {
+      "epoch": 3.8778353856124435,
+      "grad_norm": 0.0809299424290657,
+      "learning_rate": 6.786984112696342e-06,
+      "loss": 0.2155,
+      "step": 47868
+    },
+    {
+      "epoch": 3.8779163966299417,
+      "grad_norm": 0.0768105685710907,
+      "learning_rate": 6.7824834601017155e-06,
+      "loss": 0.205,
+      "step": 47869
+    },
+    {
+      "epoch": 3.87799740764744,
+      "grad_norm": 0.07321664690971375,
+      "learning_rate": 6.777982807507088e-06,
+      "loss": 0.2521,
+      "step": 47870
+    },
+    {
+      "epoch": 3.8780784186649386,
+      "grad_norm": 0.06322470307350159,
+      "learning_rate": 6.773482154912463e-06,
+      "loss": 0.2193,
+      "step": 47871
+    },
+    {
+      "epoch": 3.878159429682437,
+      "grad_norm": 0.0774395614862442,
+      "learning_rate": 6.768981502317836e-06,
+      "loss": 0.2615,
+      "step": 47872
+    },
+    {
+      "epoch": 3.878240440699935,
+      "grad_norm": 0.07283959537744522,
+      "learning_rate": 6.764480849723211e-06,
+      "loss": 0.2156,
+      "step": 47873
+    },
+    {
+      "epoch": 3.878321451717434,
+      "grad_norm": 0.0624484121799469,
+      "learning_rate": 6.7599801971285835e-06,
+      "loss": 0.227,
+      "step": 47874
+    },
+    {
+      "epoch": 3.878402462734932,
+      "grad_norm": 0.07772690057754517,
+      "learning_rate": 6.755479544533957e-06,
+      "loss": 0.2378,
+      "step": 47875
+    },
+    {
+      "epoch": 3.8784834737524303,
+      "grad_norm": 0.06441731750965118,
+      "learning_rate": 6.750978891939332e-06,
+      "loss": 0.2066,
+      "step": 47876
+    },
+    {
+      "epoch": 3.8785644847699285,
+      "grad_norm": 0.07770302146673203,
+      "learning_rate": 6.746478239344706e-06,
+      "loss": 0.2372,
+      "step": 47877
+    },
+    {
+      "epoch": 3.8786454957874272,
+      "grad_norm": 0.07686355710029602,
+      "learning_rate": 6.741977586750079e-06,
+      "loss": 0.2351,
+      "step": 47878
+    },
+    {
+      "epoch": 3.8787265068049255,
+      "grad_norm": 0.09807416796684265,
+      "learning_rate": 6.737476934155452e-06,
+      "loss": 0.2255,
+      "step": 47879
+    },
+    {
+      "epoch": 3.8788075178224237,
+      "grad_norm": 0.06583904474973679,
+      "learning_rate": 6.732976281560827e-06,
+      "loss": 0.2109,
+      "step": 47880
+    },
+    {
+      "epoch": 3.878888528839922,
+      "grad_norm": 0.07584837824106216,
+      "learning_rate": 6.7284756289662005e-06,
+      "loss": 0.2419,
+      "step": 47881
+    },
+    {
+      "epoch": 3.8789695398574207,
+      "grad_norm": 0.05641576275229454,
+      "learning_rate": 6.723974976371575e-06,
+      "loss": 0.2139,
+      "step": 47882
+    },
+    {
+      "epoch": 3.879050550874919,
+      "grad_norm": 0.07774477452039719,
+      "learning_rate": 6.719474323776948e-06,
+      "loss": 0.2532,
+      "step": 47883
+    },
+    {
+      "epoch": 3.879131561892417,
+      "grad_norm": 0.06663135439157486,
+      "learning_rate": 6.714973671182321e-06,
+      "loss": 0.2139,
+      "step": 47884
+    },
+    {
+      "epoch": 3.879212572909916,
+      "grad_norm": 0.07237344980239868,
+      "learning_rate": 6.710473018587696e-06,
+      "loss": 0.188,
+      "step": 47885
+    },
+    {
+      "epoch": 3.879293583927414,
+      "grad_norm": 0.0650007426738739,
+      "learning_rate": 6.70597236599307e-06,
+      "loss": 0.2663,
+      "step": 47886
+    },
+    {
+      "epoch": 3.8793745949449123,
+      "grad_norm": 0.07538321614265442,
+      "learning_rate": 6.701471713398443e-06,
+      "loss": 0.2219,
+      "step": 47887
+    },
+    {
+      "epoch": 3.879455605962411,
+      "grad_norm": 0.0606030635535717,
+      "learning_rate": 6.6969710608038165e-06,
+      "loss": 0.2124,
+      "step": 47888
+    },
+    {
+      "epoch": 3.8795366169799093,
+      "grad_norm": 0.07034909725189209,
+      "learning_rate": 6.692470408209191e-06,
+      "loss": 0.2611,
+      "step": 47889
+    },
+    {
+      "epoch": 3.8796176279974075,
+      "grad_norm": 0.055674366652965546,
+      "learning_rate": 6.687969755614565e-06,
+      "loss": 0.1966,
+      "step": 47890
+    },
+    {
+      "epoch": 3.879698639014906,
+      "grad_norm": 0.06106676533818245,
+      "learning_rate": 6.683469103019937e-06,
+      "loss": 0.184,
+      "step": 47891
+    },
+    {
+      "epoch": 3.8797796500324044,
+      "grad_norm": 0.06368546187877655,
+      "learning_rate": 6.678968450425312e-06,
+      "loss": 0.2065,
+      "step": 47892
+    },
+    {
+      "epoch": 3.8798606610499027,
+      "grad_norm": 0.08897584676742554,
+      "learning_rate": 6.674467797830685e-06,
+      "loss": 0.2394,
+      "step": 47893
+    },
+    {
+      "epoch": 3.8799416720674014,
+      "grad_norm": 0.06789989769458771,
+      "learning_rate": 6.66996714523606e-06,
+      "loss": 0.2254,
+      "step": 47894
+    },
+    {
+      "epoch": 3.8800226830848996,
+      "grad_norm": 0.057918865233659744,
+      "learning_rate": 6.665466492641434e-06,
+      "loss": 0.2368,
+      "step": 47895
+    },
+    {
+      "epoch": 3.880103694102398,
+      "grad_norm": 0.06686865538358688,
+      "learning_rate": 6.660965840046807e-06,
+      "loss": 0.1985,
+      "step": 47896
+    },
+    {
+      "epoch": 3.8801847051198965,
+      "grad_norm": 0.07179523259401321,
+      "learning_rate": 6.656465187452181e-06,
+      "loss": 0.2398,
+      "step": 47897
+    },
+    {
+      "epoch": 3.880265716137395,
+      "grad_norm": 0.09124450385570526,
+      "learning_rate": 6.651964534857555e-06,
+      "loss": 0.2308,
+      "step": 47898
+    },
+    {
+      "epoch": 3.880346727154893,
+      "grad_norm": 0.06625550985336304,
+      "learning_rate": 6.647463882262929e-06,
+      "loss": 0.2009,
+      "step": 47899
+    },
+    {
+      "epoch": 3.8804277381723913,
+      "grad_norm": 0.08759491890668869,
+      "learning_rate": 6.6429632296683015e-06,
+      "loss": 0.2358,
+      "step": 47900
+    },
+    {
+      "epoch": 3.88050874918989,
+      "grad_norm": 0.08081474155187607,
+      "learning_rate": 6.638462577073676e-06,
+      "loss": 0.2259,
+      "step": 47901
+    },
+    {
+      "epoch": 3.880589760207388,
+      "grad_norm": 0.08028419315814972,
+      "learning_rate": 6.6339619244790495e-06,
+      "loss": 0.2476,
+      "step": 47902
+    },
+    {
+      "epoch": 3.8806707712248865,
+      "grad_norm": 0.0740567073225975,
+      "learning_rate": 6.629461271884424e-06,
+      "loss": 0.2456,
+      "step": 47903
+    },
+    {
+      "epoch": 3.8807517822423847,
+      "grad_norm": 0.06059883162379265,
+      "learning_rate": 6.624960619289797e-06,
+      "loss": 0.2061,
+      "step": 47904
+    },
+    {
+      "epoch": 3.8808327932598834,
+      "grad_norm": 0.06069719418883324,
+      "learning_rate": 6.620459966695171e-06,
+      "loss": 0.1905,
+      "step": 47905
+    },
+    {
+      "epoch": 3.8809138042773816,
+      "grad_norm": 0.07481228560209274,
+      "learning_rate": 6.615959314100545e-06,
+      "loss": 0.2472,
+      "step": 47906
+    },
+    {
+      "epoch": 3.88099481529488,
+      "grad_norm": 0.0695427656173706,
+      "learning_rate": 6.611458661505919e-06,
+      "loss": 0.2197,
+      "step": 47907
+    },
+    {
+      "epoch": 3.8810758263123786,
+      "grad_norm": 0.07964444160461426,
+      "learning_rate": 6.606958008911293e-06,
+      "loss": 0.2039,
+      "step": 47908
+    },
+    {
+      "epoch": 3.881156837329877,
+      "grad_norm": 0.06760581582784653,
+      "learning_rate": 6.602457356316666e-06,
+      "loss": 0.24,
+      "step": 47909
+    },
+    {
+      "epoch": 3.881237848347375,
+      "grad_norm": 0.08289235085248947,
+      "learning_rate": 6.59795670372204e-06,
+      "loss": 0.2075,
+      "step": 47910
+    },
+    {
+      "epoch": 3.8813188593648738,
+      "grad_norm": 0.05402611568570137,
+      "learning_rate": 6.593456051127414e-06,
+      "loss": 0.1988,
+      "step": 47911
+    },
+    {
+      "epoch": 3.881399870382372,
+      "grad_norm": 0.06173928454518318,
+      "learning_rate": 6.588955398532788e-06,
+      "loss": 0.2041,
+      "step": 47912
+    },
+    {
+      "epoch": 3.8814808813998702,
+      "grad_norm": 0.06798407435417175,
+      "learning_rate": 6.584454745938161e-06,
+      "loss": 0.2773,
+      "step": 47913
+    },
+    {
+      "epoch": 3.881561892417369,
+      "grad_norm": 0.07197803258895874,
+      "learning_rate": 6.579954093343535e-06,
+      "loss": 0.2325,
+      "step": 47914
+    },
+    {
+      "epoch": 3.881642903434867,
+      "grad_norm": 0.0872403159737587,
+      "learning_rate": 6.575453440748909e-06,
+      "loss": 0.2524,
+      "step": 47915
+    },
+    {
+      "epoch": 3.8817239144523654,
+      "grad_norm": 0.07195434719324112,
+      "learning_rate": 6.570952788154283e-06,
+      "loss": 0.2117,
+      "step": 47916
+    },
+    {
+      "epoch": 3.881804925469864,
+      "grad_norm": 0.05543294548988342,
+      "learning_rate": 6.566452135559656e-06,
+      "loss": 0.1988,
+      "step": 47917
+    },
+    {
+      "epoch": 3.8818859364873624,
+      "grad_norm": 0.0677102655172348,
+      "learning_rate": 6.56195148296503e-06,
+      "loss": 0.2404,
+      "step": 47918
+    },
+    {
+      "epoch": 3.8819669475048606,
+      "grad_norm": 0.07490658760070801,
+      "learning_rate": 6.557450830370404e-06,
+      "loss": 0.2111,
+      "step": 47919
+    },
+    {
+      "epoch": 3.8820479585223593,
+      "grad_norm": 0.06448684632778168,
+      "learning_rate": 6.552950177775778e-06,
+      "loss": 0.1931,
+      "step": 47920
+    },
+    {
+      "epoch": 3.8821289695398575,
+      "grad_norm": 0.06765627861022949,
+      "learning_rate": 6.5484495251811505e-06,
+      "loss": 0.2379,
+      "step": 47921
+    },
+    {
+      "epoch": 3.8822099805573558,
+      "grad_norm": 0.05987018346786499,
+      "learning_rate": 6.543948872586525e-06,
+      "loss": 0.2281,
+      "step": 47922
+    },
+    {
+      "epoch": 3.882290991574854,
+      "grad_norm": 0.07392005622386932,
+      "learning_rate": 6.5394482199918994e-06,
+      "loss": 0.2199,
+      "step": 47923
+    },
+    {
+      "epoch": 3.8823720025923527,
+      "grad_norm": 0.07084795087575912,
+      "learning_rate": 6.534947567397273e-06,
+      "loss": 0.2391,
+      "step": 47924
+    },
+    {
+      "epoch": 3.882453013609851,
+      "grad_norm": 0.06273969262838364,
+      "learning_rate": 6.5304469148026475e-06,
+      "loss": 0.2309,
+      "step": 47925
+    },
+    {
+      "epoch": 3.882534024627349,
+      "grad_norm": 0.08438969403505325,
+      "learning_rate": 6.52594626220802e-06,
+      "loss": 0.2198,
+      "step": 47926
+    },
+    {
+      "epoch": 3.8826150356448474,
+      "grad_norm": 0.0724400132894516,
+      "learning_rate": 6.521445609613394e-06,
+      "loss": 0.2152,
+      "step": 47927
+    },
+    {
+      "epoch": 3.882696046662346,
+      "grad_norm": 0.060646384954452515,
+      "learning_rate": 6.516944957018768e-06,
+      "loss": 0.2088,
+      "step": 47928
+    },
+    {
+      "epoch": 3.8827770576798444,
+      "grad_norm": 0.07137048989534378,
+      "learning_rate": 6.512444304424143e-06,
+      "loss": 0.2723,
+      "step": 47929
+    },
+    {
+      "epoch": 3.8828580686973426,
+      "grad_norm": 0.07384997606277466,
+      "learning_rate": 6.507943651829515e-06,
+      "loss": 0.2015,
+      "step": 47930
+    },
+    {
+      "epoch": 3.8829390797148413,
+      "grad_norm": 0.082190603017807,
+      "learning_rate": 6.503442999234889e-06,
+      "loss": 0.2473,
+      "step": 47931
+    },
+    {
+      "epoch": 3.8830200907323396,
+      "grad_norm": 0.07313903421163559,
+      "learning_rate": 6.4989423466402635e-06,
+      "loss": 0.2301,
+      "step": 47932
+    },
+    {
+      "epoch": 3.883101101749838,
+      "grad_norm": 0.0644293874502182,
+      "learning_rate": 6.494441694045637e-06,
+      "loss": 0.2226,
+      "step": 47933
+    },
+    {
+      "epoch": 3.8831821127673365,
+      "grad_norm": 0.0785566046833992,
+      "learning_rate": 6.48994104145101e-06,
+      "loss": 0.2528,
+      "step": 47934
+    },
+    {
+      "epoch": 3.8832631237848347,
+      "grad_norm": 0.08307922631502151,
+      "learning_rate": 6.485440388856384e-06,
+      "loss": 0.207,
+      "step": 47935
+    },
+    {
+      "epoch": 3.883344134802333,
+      "grad_norm": 0.07435651123523712,
+      "learning_rate": 6.480939736261758e-06,
+      "loss": 0.2639,
+      "step": 47936
+    },
+    {
+      "epoch": 3.8834251458198317,
+      "grad_norm": 0.06681717187166214,
+      "learning_rate": 6.476439083667132e-06,
+      "loss": 0.1967,
+      "step": 47937
+    },
+    {
+      "epoch": 3.88350615683733,
+      "grad_norm": 0.06831416487693787,
+      "learning_rate": 6.471938431072507e-06,
+      "loss": 0.2202,
+      "step": 47938
+    },
+    {
+      "epoch": 3.883587167854828,
+      "grad_norm": 0.06847839057445526,
+      "learning_rate": 6.467437778477879e-06,
+      "loss": 0.1872,
+      "step": 47939
+    },
+    {
+      "epoch": 3.883668178872327,
+      "grad_norm": 0.07324326038360596,
+      "learning_rate": 6.462937125883253e-06,
+      "loss": 0.2338,
+      "step": 47940
+    },
+    {
+      "epoch": 3.883749189889825,
+      "grad_norm": 0.06632672995328903,
+      "learning_rate": 6.458436473288628e-06,
+      "loss": 0.2185,
+      "step": 47941
+    },
+    {
+      "epoch": 3.8838302009073233,
+      "grad_norm": 0.09587211906909943,
+      "learning_rate": 6.453935820694001e-06,
+      "loss": 0.2197,
+      "step": 47942
+    },
+    {
+      "epoch": 3.883911211924822,
+      "grad_norm": 0.07188690453767776,
+      "learning_rate": 6.449435168099374e-06,
+      "loss": 0.206,
+      "step": 47943
+    },
+    {
+      "epoch": 3.8839922229423203,
+      "grad_norm": 0.06359101831912994,
+      "learning_rate": 6.4449345155047485e-06,
+      "loss": 0.1849,
+      "step": 47944
+    },
+    {
+      "epoch": 3.8840732339598185,
+      "grad_norm": 0.08528807014226913,
+      "learning_rate": 6.440433862910122e-06,
+      "loss": 0.2048,
+      "step": 47945
+    },
+    {
+      "epoch": 3.8841542449773168,
+      "grad_norm": 0.09145567566156387,
+      "learning_rate": 6.4359332103154965e-06,
+      "loss": 0.2217,
+      "step": 47946
+    },
+    {
+      "epoch": 3.8842352559948155,
+      "grad_norm": 0.0667794719338417,
+      "learning_rate": 6.431432557720869e-06,
+      "loss": 0.2307,
+      "step": 47947
+    },
+    {
+      "epoch": 3.8843162670123137,
+      "grad_norm": 0.08978570997714996,
+      "learning_rate": 6.426931905126244e-06,
+      "loss": 0.2436,
+      "step": 47948
+    },
+    {
+      "epoch": 3.884397278029812,
+      "grad_norm": 0.09070317447185516,
+      "learning_rate": 6.422431252531617e-06,
+      "loss": 0.2234,
+      "step": 47949
+    },
+    {
+      "epoch": 3.88447828904731,
+      "grad_norm": 0.07344508171081543,
+      "learning_rate": 6.417930599936992e-06,
+      "loss": 0.1885,
+      "step": 47950
+    },
+    {
+      "epoch": 3.884559300064809,
+      "grad_norm": 0.0661882534623146,
+      "learning_rate": 6.413429947342365e-06,
+      "loss": 0.1988,
+      "step": 47951
+    },
+    {
+      "epoch": 3.884640311082307,
+      "grad_norm": 0.07575134187936783,
+      "learning_rate": 6.408929294747738e-06,
+      "loss": 0.2317,
+      "step": 47952
+    },
+    {
+      "epoch": 3.8847213220998054,
+      "grad_norm": 0.08309940248727798,
+      "learning_rate": 6.404428642153113e-06,
+      "loss": 0.2172,
+      "step": 47953
+    },
+    {
+      "epoch": 3.884802333117304,
+      "grad_norm": 0.06687768548727036,
+      "learning_rate": 6.399927989558486e-06,
+      "loss": 0.211,
+      "step": 47954
+    },
+    {
+      "epoch": 3.8848833441348023,
+      "grad_norm": 0.10296569019556046,
+      "learning_rate": 6.395427336963861e-06,
+      "loss": 0.2346,
+      "step": 47955
+    },
+    {
+      "epoch": 3.8849643551523005,
+      "grad_norm": 0.0639154314994812,
+      "learning_rate": 6.390926684369233e-06,
+      "loss": 0.2162,
+      "step": 47956
+    },
+    {
+      "epoch": 3.8850453661697992,
+      "grad_norm": 0.07917410880327225,
+      "learning_rate": 6.386426031774608e-06,
+      "loss": 0.2054,
+      "step": 47957
+    },
+    {
+      "epoch": 3.8851263771872975,
+      "grad_norm": 0.08458743989467621,
+      "learning_rate": 6.3819253791799815e-06,
+      "loss": 0.2153,
+      "step": 47958
+    },
+    {
+      "epoch": 3.8852073882047957,
+      "grad_norm": 0.07421503216028214,
+      "learning_rate": 6.377424726585356e-06,
+      "loss": 0.2597,
+      "step": 47959
+    },
+    {
+      "epoch": 3.8852883992222944,
+      "grad_norm": 0.08566386252641678,
+      "learning_rate": 6.372924073990729e-06,
+      "loss": 0.2534,
+      "step": 47960
+    },
+    {
+      "epoch": 3.8853694102397927,
+      "grad_norm": 0.07812653481960297,
+      "learning_rate": 6.368423421396102e-06,
+      "loss": 0.2537,
+      "step": 47961
+    },
+    {
+      "epoch": 3.885450421257291,
+      "grad_norm": 0.09011181443929672,
+      "learning_rate": 6.363922768801477e-06,
+      "loss": 0.2197,
+      "step": 47962
+    },
+    {
+      "epoch": 3.8855314322747896,
+      "grad_norm": 0.07989250868558884,
+      "learning_rate": 6.35942211620685e-06,
+      "loss": 0.2303,
+      "step": 47963
+    },
+    {
+      "epoch": 3.885612443292288,
+      "grad_norm": 0.08667084574699402,
+      "learning_rate": 6.354921463612223e-06,
+      "loss": 0.247,
+      "step": 47964
+    },
+    {
+      "epoch": 3.885693454309786,
+      "grad_norm": 0.07098482549190521,
+      "learning_rate": 6.3504208110175975e-06,
+      "loss": 0.2297,
+      "step": 47965
+    },
+    {
+      "epoch": 3.8857744653272848,
+      "grad_norm": 0.06345818936824799,
+      "learning_rate": 6.345920158422972e-06,
+      "loss": 0.1796,
+      "step": 47966
+    },
+    {
+      "epoch": 3.885855476344783,
+      "grad_norm": 0.06717762351036072,
+      "learning_rate": 6.341419505828346e-06,
+      "loss": 0.2043,
+      "step": 47967
+    },
+    {
+      "epoch": 3.8859364873622813,
+      "grad_norm": 0.06535407900810242,
+      "learning_rate": 6.33691885323372e-06,
+      "loss": 0.2326,
+      "step": 47968
+    },
+    {
+      "epoch": 3.8860174983797795,
+      "grad_norm": 0.07947888225317001,
+      "learning_rate": 6.332418200639093e-06,
+      "loss": 0.2643,
+      "step": 47969
+    },
+    {
+      "epoch": 3.886098509397278,
+      "grad_norm": 0.07099038362503052,
+      "learning_rate": 6.327917548044466e-06,
+      "loss": 0.2183,
+      "step": 47970
+    },
+    {
+      "epoch": 3.8861795204147764,
+      "grad_norm": 0.08765587210655212,
+      "learning_rate": 6.323416895449841e-06,
+      "loss": 0.2197,
+      "step": 47971
+    },
+    {
+      "epoch": 3.8862605314322747,
+      "grad_norm": 0.06536171585321426,
+      "learning_rate": 6.3189162428552145e-06,
+      "loss": 0.1928,
+      "step": 47972
+    },
+    {
+      "epoch": 3.886341542449773,
+      "grad_norm": 0.08230137079954147,
+      "learning_rate": 6.314415590260587e-06,
+      "loss": 0.231,
+      "step": 47973
+    },
+    {
+      "epoch": 3.8864225534672716,
+      "grad_norm": 0.06578972190618515,
+      "learning_rate": 6.309914937665962e-06,
+      "loss": 0.2168,
+      "step": 47974
+    },
+    {
+      "epoch": 3.88650356448477,
+      "grad_norm": 0.06204579770565033,
+      "learning_rate": 6.305414285071336e-06,
+      "loss": 0.2093,
+      "step": 47975
+    },
+    {
+      "epoch": 3.886584575502268,
+      "grad_norm": 0.06162213534116745,
+      "learning_rate": 6.30091363247671e-06,
+      "loss": 0.2351,
+      "step": 47976
+    },
+    {
+      "epoch": 3.886665586519767,
+      "grad_norm": 0.07075859606266022,
+      "learning_rate": 6.2964129798820825e-06,
+      "loss": 0.1968,
+      "step": 47977
+    },
+    {
+      "epoch": 3.886746597537265,
+      "grad_norm": 0.08070050179958344,
+      "learning_rate": 6.291912327287457e-06,
+      "loss": 0.2021,
+      "step": 47978
+    },
+    {
+      "epoch": 3.8868276085547633,
+      "grad_norm": 0.07428973913192749,
+      "learning_rate": 6.2874116746928305e-06,
+      "loss": 0.249,
+      "step": 47979
+    },
+    {
+      "epoch": 3.886908619572262,
+      "grad_norm": 0.08572898060083389,
+      "learning_rate": 6.282911022098205e-06,
+      "loss": 0.2257,
+      "step": 47980
+    },
+    {
+      "epoch": 3.88698963058976,
+      "grad_norm": 0.06448810547590256,
+      "learning_rate": 6.278410369503579e-06,
+      "loss": 0.1896,
+      "step": 47981
+    },
+    {
+      "epoch": 3.8870706416072585,
+      "grad_norm": 0.05051233246922493,
+      "learning_rate": 6.273909716908951e-06,
+      "loss": 0.2156,
+      "step": 47982
+    },
+    {
+      "epoch": 3.887151652624757,
+      "grad_norm": 0.08261623233556747,
+      "learning_rate": 6.269409064314326e-06,
+      "loss": 0.2675,
+      "step": 47983
+    },
+    {
+      "epoch": 3.8872326636422554,
+      "grad_norm": 0.05112019553780556,
+      "learning_rate": 6.2649084117197e-06,
+      "loss": 0.2252,
+      "step": 47984
+    },
+    {
+      "epoch": 3.8873136746597536,
+      "grad_norm": 0.09426064044237137,
+      "learning_rate": 6.260407759125074e-06,
+      "loss": 0.2285,
+      "step": 47985
+    },
+    {
+      "epoch": 3.8873946856772523,
+      "grad_norm": 0.08980914205312729,
+      "learning_rate": 6.255907106530447e-06,
+      "loss": 0.2585,
+      "step": 47986
+    },
+    {
+      "epoch": 3.8874756966947506,
+      "grad_norm": 0.0838213637471199,
+      "learning_rate": 6.251406453935821e-06,
+      "loss": 0.1998,
+      "step": 47987
+    },
+    {
+      "epoch": 3.887556707712249,
+      "grad_norm": 0.07811640948057175,
+      "learning_rate": 6.246905801341195e-06,
+      "loss": 0.2306,
+      "step": 47988
+    },
+    {
+      "epoch": 3.8876377187297475,
+      "grad_norm": 0.08223123103380203,
+      "learning_rate": 6.242405148746568e-06,
+      "loss": 0.2695,
+      "step": 47989
+    },
+    {
+      "epoch": 3.8877187297472457,
+      "grad_norm": 0.0751977488398552,
+      "learning_rate": 6.237904496151943e-06,
+      "loss": 0.2244,
+      "step": 47990
+    },
+    {
+      "epoch": 3.887799740764744,
+      "grad_norm": 0.06286336481571198,
+      "learning_rate": 6.2334038435573155e-06,
+      "loss": 0.202,
+      "step": 47991
+    },
+    {
+      "epoch": 3.8878807517822422,
+      "grad_norm": 0.08371991664171219,
+      "learning_rate": 6.22890319096269e-06,
+      "loss": 0.229,
+      "step": 47992
+    },
+    {
+      "epoch": 3.887961762799741,
+      "grad_norm": 0.054850783199071884,
+      "learning_rate": 6.2244025383680635e-06,
+      "loss": 0.2143,
+      "step": 47993
+    },
+    {
+      "epoch": 3.888042773817239,
+      "grad_norm": 0.056063272058963776,
+      "learning_rate": 6.219901885773437e-06,
+      "loss": 0.1868,
+      "step": 47994
+    },
+    {
+      "epoch": 3.8881237848347374,
+      "grad_norm": 0.07119081169366837,
+      "learning_rate": 6.2154012331788116e-06,
+      "loss": 0.2235,
+      "step": 47995
+    },
+    {
+      "epoch": 3.8882047958522357,
+      "grad_norm": 0.10805629193782806,
+      "learning_rate": 6.210900580584185e-06,
+      "loss": 0.1982,
+      "step": 47996
+    },
+    {
+      "epoch": 3.8882858068697344,
+      "grad_norm": 0.0731663703918457,
+      "learning_rate": 6.206399927989559e-06,
+      "loss": 0.2528,
+      "step": 47997
+    },
+    {
+      "epoch": 3.8883668178872326,
+      "grad_norm": 0.07942456007003784,
+      "learning_rate": 6.201899275394932e-06,
+      "loss": 0.2717,
+      "step": 47998
+    },
+    {
+      "epoch": 3.888447828904731,
+      "grad_norm": 0.0707758367061615,
+      "learning_rate": 6.197398622800307e-06,
+      "loss": 0.2002,
+      "step": 47999
+    },
+    {
+      "epoch": 3.8885288399222295,
+      "grad_norm": 0.06386274099349976,
+      "learning_rate": 6.19289797020568e-06,
+      "loss": 0.2274,
+      "step": 48000
+    },
+    {
+      "epoch": 3.8886098509397278,
+      "grad_norm": 0.07666601985692978,
+      "learning_rate": 6.188397317611054e-06,
+      "loss": 0.2049,
+      "step": 48001
+    },
+    {
+      "epoch": 3.888690861957226,
+      "grad_norm": 0.07881522923707962,
+      "learning_rate": 6.183896665016428e-06,
+      "loss": 0.2338,
+      "step": 48002
+    },
+    {
+      "epoch": 3.8887718729747247,
+      "grad_norm": 0.07827413827180862,
+      "learning_rate": 6.179396012421801e-06,
+      "loss": 0.2311,
+      "step": 48003
+    },
+    {
+      "epoch": 3.888852883992223,
+      "grad_norm": 0.08375756442546844,
+      "learning_rate": 6.174895359827175e-06,
+      "loss": 0.2089,
+      "step": 48004
+    },
+    {
+      "epoch": 3.888933895009721,
+      "grad_norm": 0.06420580297708511,
+      "learning_rate": 6.170394707232549e-06,
+      "loss": 0.2367,
+      "step": 48005
+    },
+    {
+      "epoch": 3.88901490602722,
+      "grad_norm": 0.08399081230163574,
+      "learning_rate": 6.165894054637923e-06,
+      "loss": 0.2573,
+      "step": 48006
+    },
+    {
+      "epoch": 3.889095917044718,
+      "grad_norm": 0.07225336134433746,
+      "learning_rate": 6.1613934020432965e-06,
+      "loss": 0.2114,
+      "step": 48007
+    },
+    {
+      "epoch": 3.8891769280622164,
+      "grad_norm": 0.07664166390895844,
+      "learning_rate": 6.15689274944867e-06,
+      "loss": 0.2658,
+      "step": 48008
+    },
+    {
+      "epoch": 3.889257939079715,
+      "grad_norm": 0.07305591553449631,
+      "learning_rate": 6.152392096854044e-06,
+      "loss": 0.2189,
+      "step": 48009
+    },
+    {
+      "epoch": 3.8893389500972133,
+      "grad_norm": 0.06890987604856491,
+      "learning_rate": 6.147891444259418e-06,
+      "loss": 0.2175,
+      "step": 48010
+    },
+    {
+      "epoch": 3.8894199611147116,
+      "grad_norm": 0.06536436080932617,
+      "learning_rate": 6.143390791664792e-06,
+      "loss": 0.251,
+      "step": 48011
+    },
+    {
+      "epoch": 3.8895009721322102,
+      "grad_norm": 0.07678571343421936,
+      "learning_rate": 6.138890139070165e-06,
+      "loss": 0.2241,
+      "step": 48012
+    },
+    {
+      "epoch": 3.8895819831497085,
+      "grad_norm": 0.06775952875614166,
+      "learning_rate": 6.134389486475539e-06,
+      "loss": 0.2281,
+      "step": 48013
+    },
+    {
+      "epoch": 3.8896629941672067,
+      "grad_norm": 0.09381411224603653,
+      "learning_rate": 6.129888833880913e-06,
+      "loss": 0.2554,
+      "step": 48014
+    },
+    {
+      "epoch": 3.889744005184705,
+      "grad_norm": 0.05685148388147354,
+      "learning_rate": 6.125388181286287e-06,
+      "loss": 0.239,
+      "step": 48015
+    },
+    {
+      "epoch": 3.8898250162022032,
+      "grad_norm": 0.07663068175315857,
+      "learning_rate": 6.120887528691661e-06,
+      "loss": 0.2633,
+      "step": 48016
+    },
+    {
+      "epoch": 3.889906027219702,
+      "grad_norm": 0.0696638897061348,
+      "learning_rate": 6.116386876097034e-06,
+      "loss": 0.2184,
+      "step": 48017
+    },
+    {
+      "epoch": 3.8899870382372,
+      "grad_norm": 0.06149803474545479,
+      "learning_rate": 6.111886223502408e-06,
+      "loss": 0.2162,
+      "step": 48018
+    },
+    {
+      "epoch": 3.8900680492546984,
+      "grad_norm": 0.07225798070430756,
+      "learning_rate": 6.1073855709077814e-06,
+      "loss": 0.2452,
+      "step": 48019
+    },
+    {
+      "epoch": 3.890149060272197,
+      "grad_norm": 0.06644891202449799,
+      "learning_rate": 6.102884918313156e-06,
+      "loss": 0.2403,
+      "step": 48020
+    },
+    {
+      "epoch": 3.8902300712896953,
+      "grad_norm": 0.07242048531770706,
+      "learning_rate": 6.0983842657185295e-06,
+      "loss": 0.2397,
+      "step": 48021
+    },
+    {
+      "epoch": 3.8903110823071936,
+      "grad_norm": 0.07406973838806152,
+      "learning_rate": 6.093883613123903e-06,
+      "loss": 0.223,
+      "step": 48022
+    },
+    {
+      "epoch": 3.8903920933246923,
+      "grad_norm": 0.06930666416883469,
+      "learning_rate": 6.0893829605292775e-06,
+      "loss": 0.2271,
+      "step": 48023
+    },
+    {
+      "epoch": 3.8904731043421905,
+      "grad_norm": 0.07608542591333389,
+      "learning_rate": 6.084882307934651e-06,
+      "loss": 0.2076,
+      "step": 48024
+    },
+    {
+      "epoch": 3.8905541153596888,
+      "grad_norm": 0.06247393414378166,
+      "learning_rate": 6.080381655340025e-06,
+      "loss": 0.2051,
+      "step": 48025
+    },
+    {
+      "epoch": 3.8906351263771874,
+      "grad_norm": 0.061167728155851364,
+      "learning_rate": 6.075881002745398e-06,
+      "loss": 0.2088,
+      "step": 48026
+    },
+    {
+      "epoch": 3.8907161373946857,
+      "grad_norm": 0.06517788022756577,
+      "learning_rate": 6.071380350150773e-06,
+      "loss": 0.241,
+      "step": 48027
+    },
+    {
+      "epoch": 3.890797148412184,
+      "grad_norm": 0.07304881513118744,
+      "learning_rate": 6.0668796975561456e-06,
+      "loss": 0.2088,
+      "step": 48028
+    },
+    {
+      "epoch": 3.8908781594296826,
+      "grad_norm": 0.06430026143789291,
+      "learning_rate": 6.06237904496152e-06,
+      "loss": 0.235,
+      "step": 48029
+    },
+    {
+      "epoch": 3.890959170447181,
+      "grad_norm": 0.06506291031837463,
+      "learning_rate": 6.057878392366894e-06,
+      "loss": 0.1662,
+      "step": 48030
+    },
+    {
+      "epoch": 3.891040181464679,
+      "grad_norm": 0.06876938790082932,
+      "learning_rate": 6.053377739772267e-06,
+      "loss": 0.2141,
+      "step": 48031
+    },
+    {
+      "epoch": 3.891121192482178,
+      "grad_norm": 0.08055271953344345,
+      "learning_rate": 6.048877087177641e-06,
+      "loss": 0.2293,
+      "step": 48032
+    },
+    {
+      "epoch": 3.891202203499676,
+      "grad_norm": 0.07058732956647873,
+      "learning_rate": 6.044376434583015e-06,
+      "loss": 0.2289,
+      "step": 48033
+    },
+    {
+      "epoch": 3.8912832145171743,
+      "grad_norm": 0.086977019906044,
+      "learning_rate": 6.039875781988388e-06,
+      "loss": 0.2359,
+      "step": 48034
+    },
+    {
+      "epoch": 3.891364225534673,
+      "grad_norm": 0.07203526794910431,
+      "learning_rate": 6.0353751293937625e-06,
+      "loss": 0.2086,
+      "step": 48035
+    },
+    {
+      "epoch": 3.8914452365521712,
+      "grad_norm": 0.06266345083713531,
+      "learning_rate": 6.030874476799136e-06,
+      "loss": 0.1983,
+      "step": 48036
+    },
+    {
+      "epoch": 3.8915262475696695,
+      "grad_norm": 0.07486934214830399,
+      "learning_rate": 6.02637382420451e-06,
+      "loss": 0.235,
+      "step": 48037
+    },
+    {
+      "epoch": 3.8916072585871677,
+      "grad_norm": 0.07059779763221741,
+      "learning_rate": 6.021873171609884e-06,
+      "loss": 0.2082,
+      "step": 48038
+    },
+    {
+      "epoch": 3.891688269604666,
+      "grad_norm": 0.06780023127794266,
+      "learning_rate": 6.017372519015258e-06,
+      "loss": 0.2632,
+      "step": 48039
+    },
+    {
+      "epoch": 3.8917692806221647,
+      "grad_norm": 0.0676514059305191,
+      "learning_rate": 6.012871866420631e-06,
+      "loss": 0.2406,
+      "step": 48040
+    },
+    {
+      "epoch": 3.891850291639663,
+      "grad_norm": 0.06825751811265945,
+      "learning_rate": 6.008371213826005e-06,
+      "loss": 0.2402,
+      "step": 48041
+    },
+    {
+      "epoch": 3.891931302657161,
+      "grad_norm": 0.06496651470661163,
+      "learning_rate": 6.003870561231379e-06,
+      "loss": 0.2294,
+      "step": 48042
+    },
+    {
+      "epoch": 3.89201231367466,
+      "grad_norm": 0.0673632025718689,
+      "learning_rate": 5.999369908636752e-06,
+      "loss": 0.206,
+      "step": 48043
+    },
+    {
+      "epoch": 3.892093324692158,
+      "grad_norm": 0.059508103877305984,
+      "learning_rate": 5.994869256042127e-06,
+      "loss": 0.2146,
+      "step": 48044
+    },
+    {
+      "epoch": 3.8921743357096563,
+      "grad_norm": 0.07362375408411026,
+      "learning_rate": 5.9903686034475e-06,
+      "loss": 0.2195,
+      "step": 48045
+    },
+    {
+      "epoch": 3.892255346727155,
+      "grad_norm": 0.07658713310956955,
+      "learning_rate": 5.985867950852874e-06,
+      "loss": 0.2371,
+      "step": 48046
+    },
+    {
+      "epoch": 3.8923363577446533,
+      "grad_norm": 0.06624037772417068,
+      "learning_rate": 5.981367298258247e-06,
+      "loss": 0.1921,
+      "step": 48047
+    },
+    {
+      "epoch": 3.8924173687621515,
+      "grad_norm": 0.08692865073680878,
+      "learning_rate": 5.976866645663622e-06,
+      "loss": 0.2362,
+      "step": 48048
+    },
+    {
+      "epoch": 3.89249837977965,
+      "grad_norm": 0.07544679939746857,
+      "learning_rate": 5.972365993068995e-06,
+      "loss": 0.221,
+      "step": 48049
+    },
+    {
+      "epoch": 3.8925793907971484,
+      "grad_norm": 0.05986184626817703,
+      "learning_rate": 5.967865340474369e-06,
+      "loss": 0.2001,
+      "step": 48050
+    },
+    {
+      "epoch": 3.8926604018146467,
+      "grad_norm": 0.08265157043933868,
+      "learning_rate": 5.963364687879743e-06,
+      "loss": 0.3066,
+      "step": 48051
+    },
+    {
+      "epoch": 3.8927414128321454,
+      "grad_norm": 0.07029622048139572,
+      "learning_rate": 5.958864035285116e-06,
+      "loss": 0.2507,
+      "step": 48052
+    },
+    {
+      "epoch": 3.8928224238496436,
+      "grad_norm": 0.07296372205018997,
+      "learning_rate": 5.954363382690491e-06,
+      "loss": 0.2628,
+      "step": 48053
+    },
+    {
+      "epoch": 3.892903434867142,
+      "grad_norm": 0.08202331513166428,
+      "learning_rate": 5.949862730095864e-06,
+      "loss": 0.236,
+      "step": 48054
+    },
+    {
+      "epoch": 3.8929844458846405,
+      "grad_norm": 0.07177530229091644,
+      "learning_rate": 5.945362077501238e-06,
+      "loss": 0.2239,
+      "step": 48055
+    },
+    {
+      "epoch": 3.893065456902139,
+      "grad_norm": 0.06391312927007675,
+      "learning_rate": 5.9408614249066115e-06,
+      "loss": 0.2415,
+      "step": 48056
+    },
+    {
+      "epoch": 3.893146467919637,
+      "grad_norm": 0.07368528842926025,
+      "learning_rate": 5.936360772311986e-06,
+      "loss": 0.2102,
+      "step": 48057
+    },
+    {
+      "epoch": 3.8932274789371357,
+      "grad_norm": 0.09229487180709839,
+      "learning_rate": 5.931860119717359e-06,
+      "loss": 0.2683,
+      "step": 48058
+    },
+    {
+      "epoch": 3.893308489954634,
+      "grad_norm": 0.07881072908639908,
+      "learning_rate": 5.927359467122733e-06,
+      "loss": 0.205,
+      "step": 48059
+    },
+    {
+      "epoch": 3.893389500972132,
+      "grad_norm": 0.08213283866643906,
+      "learning_rate": 5.922858814528107e-06,
+      "loss": 0.2216,
+      "step": 48060
+    },
+    {
+      "epoch": 3.8934705119896305,
+      "grad_norm": 0.06197618693113327,
+      "learning_rate": 5.91835816193348e-06,
+      "loss": 0.2131,
+      "step": 48061
+    },
+    {
+      "epoch": 3.8935515230071287,
+      "grad_norm": 0.06296448409557343,
+      "learning_rate": 5.913857509338854e-06,
+      "loss": 0.2414,
+      "step": 48062
+    },
+    {
+      "epoch": 3.8936325340246274,
+      "grad_norm": 0.057206038385629654,
+      "learning_rate": 5.9093568567442285e-06,
+      "loss": 0.2052,
+      "step": 48063
+    },
+    {
+      "epoch": 3.8937135450421256,
+      "grad_norm": 0.06184779480099678,
+      "learning_rate": 5.904856204149602e-06,
+      "loss": 0.1972,
+      "step": 48064
+    },
+    {
+      "epoch": 3.893794556059624,
+      "grad_norm": 0.06405945867300034,
+      "learning_rate": 5.900355551554976e-06,
+      "loss": 0.1939,
+      "step": 48065
+    },
+    {
+      "epoch": 3.8938755670771226,
+      "grad_norm": 0.08158735185861588,
+      "learning_rate": 5.89585489896035e-06,
+      "loss": 0.2208,
+      "step": 48066
+    },
+    {
+      "epoch": 3.893956578094621,
+      "grad_norm": 0.06656576693058014,
+      "learning_rate": 5.891354246365723e-06,
+      "loss": 0.2169,
+      "step": 48067
+    },
+    {
+      "epoch": 3.894037589112119,
+      "grad_norm": 0.06311673671007156,
+      "learning_rate": 5.886853593771097e-06,
+      "loss": 0.2384,
+      "step": 48068
+    },
+    {
+      "epoch": 3.8941186001296177,
+      "grad_norm": 0.0772138312458992,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.2311,
+      "step": 48069
+    },
+    {
+      "epoch": 3.894199611147116,
+      "grad_norm": 0.0871867686510086,
+      "learning_rate": 5.8778522885818445e-06,
+      "loss": 0.22,
+      "step": 48070
+    },
+    {
+      "epoch": 3.8942806221646142,
+      "grad_norm": 0.07590407133102417,
+      "learning_rate": 5.873351635987218e-06,
+      "loss": 0.2274,
+      "step": 48071
+    },
+    {
+      "epoch": 3.894361633182113,
+      "grad_norm": 0.0857917070388794,
+      "learning_rate": 5.868850983392593e-06,
+      "loss": 0.2184,
+      "step": 48072
+    },
+    {
+      "epoch": 3.894442644199611,
+      "grad_norm": 0.07525145262479782,
+      "learning_rate": 5.864350330797966e-06,
+      "loss": 0.1988,
+      "step": 48073
+    },
+    {
+      "epoch": 3.8945236552171094,
+      "grad_norm": 0.06835979968309402,
+      "learning_rate": 5.85984967820334e-06,
+      "loss": 0.2214,
+      "step": 48074
+    },
+    {
+      "epoch": 3.894604666234608,
+      "grad_norm": 0.06695519387722015,
+      "learning_rate": 5.855349025608713e-06,
+      "loss": 0.2288,
+      "step": 48075
+    },
+    {
+      "epoch": 3.8946856772521063,
+      "grad_norm": 0.08336454629898071,
+      "learning_rate": 5.850848373014088e-06,
+      "loss": 0.2016,
+      "step": 48076
+    },
+    {
+      "epoch": 3.8947666882696046,
+      "grad_norm": 0.06542129814624786,
+      "learning_rate": 5.846347720419461e-06,
+      "loss": 0.1926,
+      "step": 48077
+    },
+    {
+      "epoch": 3.8948476992871033,
+      "grad_norm": 0.07468671351671219,
+      "learning_rate": 5.841847067824835e-06,
+      "loss": 0.2031,
+      "step": 48078
+    },
+    {
+      "epoch": 3.8949287103046015,
+      "grad_norm": 0.08390598744153976,
+      "learning_rate": 5.837346415230209e-06,
+      "loss": 0.2465,
+      "step": 48079
+    },
+    {
+      "epoch": 3.8950097213220998,
+      "grad_norm": 0.0719221755862236,
+      "learning_rate": 5.832845762635582e-06,
+      "loss": 0.2255,
+      "step": 48080
+    },
+    {
+      "epoch": 3.8950907323395985,
+      "grad_norm": 0.07787550985813141,
+      "learning_rate": 5.828345110040957e-06,
+      "loss": 0.1899,
+      "step": 48081
+    },
+    {
+      "epoch": 3.8951717433570967,
+      "grad_norm": 0.06885991245508194,
+      "learning_rate": 5.82384445744633e-06,
+      "loss": 0.1858,
+      "step": 48082
+    },
+    {
+      "epoch": 3.895252754374595,
+      "grad_norm": 0.08002043515443802,
+      "learning_rate": 5.819343804851704e-06,
+      "loss": 0.2095,
+      "step": 48083
+    },
+    {
+      "epoch": 3.895333765392093,
+      "grad_norm": 0.06390658766031265,
+      "learning_rate": 5.8148431522570775e-06,
+      "loss": 0.2274,
+      "step": 48084
+    },
+    {
+      "epoch": 3.8954147764095914,
+      "grad_norm": 0.0804852768778801,
+      "learning_rate": 5.810342499662452e-06,
+      "loss": 0.2423,
+      "step": 48085
+    },
+    {
+      "epoch": 3.89549578742709,
+      "grad_norm": 0.06975232064723969,
+      "learning_rate": 5.805841847067825e-06,
+      "loss": 0.198,
+      "step": 48086
+    },
+    {
+      "epoch": 3.8955767984445884,
+      "grad_norm": 0.09129457175731659,
+      "learning_rate": 5.801341194473199e-06,
+      "loss": 0.1861,
+      "step": 48087
+    },
+    {
+      "epoch": 3.8956578094620866,
+      "grad_norm": 0.07378769665956497,
+      "learning_rate": 5.796840541878573e-06,
+      "loss": 0.2115,
+      "step": 48088
+    },
+    {
+      "epoch": 3.8957388204795853,
+      "grad_norm": 0.07973496615886688,
+      "learning_rate": 5.792339889283946e-06,
+      "loss": 0.2271,
+      "step": 48089
+    },
+    {
+      "epoch": 3.8958198314970836,
+      "grad_norm": 0.07622456550598145,
+      "learning_rate": 5.78783923668932e-06,
+      "loss": 0.2271,
+      "step": 48090
+    },
+    {
+      "epoch": 3.895900842514582,
+      "grad_norm": 0.06588424742221832,
+      "learning_rate": 5.7833385840946944e-06,
+      "loss": 0.2054,
+      "step": 48091
+    },
+    {
+      "epoch": 3.8959818535320805,
+      "grad_norm": 0.06296347826719284,
+      "learning_rate": 5.778837931500067e-06,
+      "loss": 0.1969,
+      "step": 48092
+    },
+    {
+      "epoch": 3.8960628645495787,
+      "grad_norm": 0.07207436114549637,
+      "learning_rate": 5.774337278905442e-06,
+      "loss": 0.2263,
+      "step": 48093
+    },
+    {
+      "epoch": 3.896143875567077,
+      "grad_norm": 0.07418610155582428,
+      "learning_rate": 5.769836626310815e-06,
+      "loss": 0.2296,
+      "step": 48094
+    },
+    {
+      "epoch": 3.8962248865845757,
+      "grad_norm": 0.08219721913337708,
+      "learning_rate": 5.765335973716189e-06,
+      "loss": 0.2525,
+      "step": 48095
+    },
+    {
+      "epoch": 3.896305897602074,
+      "grad_norm": 0.06975866854190826,
+      "learning_rate": 5.760835321121563e-06,
+      "loss": 0.2565,
+      "step": 48096
+    },
+    {
+      "epoch": 3.896386908619572,
+      "grad_norm": 0.06241089478135109,
+      "learning_rate": 5.756334668526937e-06,
+      "loss": 0.2418,
+      "step": 48097
+    },
+    {
+      "epoch": 3.896467919637071,
+      "grad_norm": 0.0809803307056427,
+      "learning_rate": 5.7518340159323105e-06,
+      "loss": 0.2403,
+      "step": 48098
+    },
+    {
+      "epoch": 3.896548930654569,
+      "grad_norm": 0.06909254193305969,
+      "learning_rate": 5.747333363337684e-06,
+      "loss": 0.2054,
+      "step": 48099
+    },
+    {
+      "epoch": 3.8966299416720673,
+      "grad_norm": 0.06723940372467041,
+      "learning_rate": 5.7428327107430586e-06,
+      "loss": 0.2459,
+      "step": 48100
+    },
+    {
+      "epoch": 3.896710952689566,
+      "grad_norm": 0.07118137925863266,
+      "learning_rate": 5.738332058148431e-06,
+      "loss": 0.2237,
+      "step": 48101
+    },
+    {
+      "epoch": 3.8967919637070643,
+      "grad_norm": 0.07935565710067749,
+      "learning_rate": 5.733831405553806e-06,
+      "loss": 0.2189,
+      "step": 48102
+    },
+    {
+      "epoch": 3.8968729747245625,
+      "grad_norm": 0.08424384146928787,
+      "learning_rate": 5.729330752959179e-06,
+      "loss": 0.2415,
+      "step": 48103
+    },
+    {
+      "epoch": 3.8969539857420608,
+      "grad_norm": 0.07037527859210968,
+      "learning_rate": 5.724830100364553e-06,
+      "loss": 0.1857,
+      "step": 48104
+    },
+    {
+      "epoch": 3.8970349967595594,
+      "grad_norm": 0.06845095008611679,
+      "learning_rate": 5.7203294477699266e-06,
+      "loss": 0.2233,
+      "step": 48105
+    },
+    {
+      "epoch": 3.8971160077770577,
+      "grad_norm": 0.08850236237049103,
+      "learning_rate": 5.715828795175301e-06,
+      "loss": 0.279,
+      "step": 48106
+    },
+    {
+      "epoch": 3.897197018794556,
+      "grad_norm": 0.06084459275007248,
+      "learning_rate": 5.711328142580674e-06,
+      "loss": 0.201,
+      "step": 48107
+    },
+    {
+      "epoch": 3.897278029812054,
+      "grad_norm": 0.07755912840366364,
+      "learning_rate": 5.706827489986048e-06,
+      "loss": 0.2315,
+      "step": 48108
+    },
+    {
+      "epoch": 3.897359040829553,
+      "grad_norm": 0.06639105826616287,
+      "learning_rate": 5.702326837391423e-06,
+      "loss": 0.2502,
+      "step": 48109
+    },
+    {
+      "epoch": 3.897440051847051,
+      "grad_norm": 0.07094542682170868,
+      "learning_rate": 5.6978261847967954e-06,
+      "loss": 0.1805,
+      "step": 48110
+    },
+    {
+      "epoch": 3.8975210628645494,
+      "grad_norm": 0.07018940150737762,
+      "learning_rate": 5.69332553220217e-06,
+      "loss": 0.1724,
+      "step": 48111
+    },
+    {
+      "epoch": 3.897602073882048,
+      "grad_norm": 0.06900765746831894,
+      "learning_rate": 5.6888248796075435e-06,
+      "loss": 0.2331,
+      "step": 48112
+    },
+    {
+      "epoch": 3.8976830848995463,
+      "grad_norm": 0.0696382075548172,
+      "learning_rate": 5.684324227012917e-06,
+      "loss": 0.2468,
+      "step": 48113
+    },
+    {
+      "epoch": 3.8977640959170445,
+      "grad_norm": 0.06245898827910423,
+      "learning_rate": 5.679823574418291e-06,
+      "loss": 0.2286,
+      "step": 48114
+    },
+    {
+      "epoch": 3.8978451069345432,
+      "grad_norm": 0.07008770108222961,
+      "learning_rate": 5.675322921823665e-06,
+      "loss": 0.2411,
+      "step": 48115
+    },
+    {
+      "epoch": 3.8979261179520415,
+      "grad_norm": 0.05966051667928696,
+      "learning_rate": 5.670822269229038e-06,
+      "loss": 0.1998,
+      "step": 48116
+    },
+    {
+      "epoch": 3.8980071289695397,
+      "grad_norm": 0.07350362837314606,
+      "learning_rate": 5.666321616634412e-06,
+      "loss": 0.2289,
+      "step": 48117
+    },
+    {
+      "epoch": 3.8980881399870384,
+      "grad_norm": 0.06739611178636551,
+      "learning_rate": 5.661820964039786e-06,
+      "loss": 0.2126,
+      "step": 48118
+    },
+    {
+      "epoch": 3.8981691510045366,
+      "grad_norm": 0.06899629533290863,
+      "learning_rate": 5.6573203114451596e-06,
+      "loss": 0.2236,
+      "step": 48119
+    },
+    {
+      "epoch": 3.898250162022035,
+      "grad_norm": 0.06305073946714401,
+      "learning_rate": 5.652819658850533e-06,
+      "loss": 0.2044,
+      "step": 48120
+    },
+    {
+      "epoch": 3.8983311730395336,
+      "grad_norm": 0.0846519023180008,
+      "learning_rate": 5.648319006255908e-06,
+      "loss": 0.2821,
+      "step": 48121
+    },
+    {
+      "epoch": 3.898412184057032,
+      "grad_norm": 0.06912169605493546,
+      "learning_rate": 5.643818353661281e-06,
+      "loss": 0.2427,
+      "step": 48122
+    },
+    {
+      "epoch": 3.89849319507453,
+      "grad_norm": 0.07149054855108261,
+      "learning_rate": 5.639317701066655e-06,
+      "loss": 0.2001,
+      "step": 48123
+    },
+    {
+      "epoch": 3.8985742060920288,
+      "grad_norm": 0.05685052648186684,
+      "learning_rate": 5.634817048472029e-06,
+      "loss": 0.2235,
+      "step": 48124
+    },
+    {
+      "epoch": 3.898655217109527,
+      "grad_norm": 0.0757313072681427,
+      "learning_rate": 5.630316395877403e-06,
+      "loss": 0.2373,
+      "step": 48125
+    },
+    {
+      "epoch": 3.8987362281270252,
+      "grad_norm": 0.08734872937202454,
+      "learning_rate": 5.6258157432827765e-06,
+      "loss": 0.2707,
+      "step": 48126
+    },
+    {
+      "epoch": 3.8988172391445235,
+      "grad_norm": 0.07042701542377472,
+      "learning_rate": 5.62131509068815e-06,
+      "loss": 0.2077,
+      "step": 48127
+    },
+    {
+      "epoch": 3.898898250162022,
+      "grad_norm": 0.06531260907649994,
+      "learning_rate": 5.616814438093524e-06,
+      "loss": 0.2404,
+      "step": 48128
+    },
+    {
+      "epoch": 3.8989792611795204,
+      "grad_norm": 0.06730636954307556,
+      "learning_rate": 5.612313785498897e-06,
+      "loss": 0.2318,
+      "step": 48129
+    },
+    {
+      "epoch": 3.8990602721970187,
+      "grad_norm": 0.08711560815572739,
+      "learning_rate": 5.607813132904272e-06,
+      "loss": 0.2223,
+      "step": 48130
+    },
+    {
+      "epoch": 3.899141283214517,
+      "grad_norm": 0.06874527782201767,
+      "learning_rate": 5.603312480309645e-06,
+      "loss": 0.2563,
+      "step": 48131
+    },
+    {
+      "epoch": 3.8992222942320156,
+      "grad_norm": 0.08103624731302261,
+      "learning_rate": 5.598811827715019e-06,
+      "loss": 0.2288,
+      "step": 48132
+    },
+    {
+      "epoch": 3.899303305249514,
+      "grad_norm": 0.07119531184434891,
+      "learning_rate": 5.5943111751203925e-06,
+      "loss": 0.2334,
+      "step": 48133
+    },
+    {
+      "epoch": 3.899384316267012,
+      "grad_norm": 0.06922445446252823,
+      "learning_rate": 5.589810522525767e-06,
+      "loss": 0.2301,
+      "step": 48134
+    },
+    {
+      "epoch": 3.899465327284511,
+      "grad_norm": 0.07962538301944733,
+      "learning_rate": 5.58530986993114e-06,
+      "loss": 0.2631,
+      "step": 48135
+    },
+    {
+      "epoch": 3.899546338302009,
+      "grad_norm": 0.07506944984197617,
+      "learning_rate": 5.580809217336514e-06,
+      "loss": 0.2203,
+      "step": 48136
+    },
+    {
+      "epoch": 3.8996273493195073,
+      "grad_norm": 0.08897983282804489,
+      "learning_rate": 5.576308564741888e-06,
+      "loss": 0.2219,
+      "step": 48137
+    },
+    {
+      "epoch": 3.899708360337006,
+      "grad_norm": 0.06361392885446548,
+      "learning_rate": 5.571807912147261e-06,
+      "loss": 0.2409,
+      "step": 48138
+    },
+    {
+      "epoch": 3.899789371354504,
+      "grad_norm": 0.07459505647420883,
+      "learning_rate": 5.567307259552636e-06,
+      "loss": 0.2043,
+      "step": 48139
+    },
+    {
+      "epoch": 3.8998703823720025,
+      "grad_norm": 0.059818338602781296,
+      "learning_rate": 5.5628066069580095e-06,
+      "loss": 0.2136,
+      "step": 48140
+    },
+    {
+      "epoch": 3.899951393389501,
+      "grad_norm": 0.08024513721466064,
+      "learning_rate": 5.558305954363383e-06,
+      "loss": 0.2156,
+      "step": 48141
+    },
+    {
+      "epoch": 3.9000324044069994,
+      "grad_norm": 0.06556698679924011,
+      "learning_rate": 5.553805301768757e-06,
+      "loss": 0.2035,
+      "step": 48142
+    },
+    {
+      "epoch": 3.9001134154244976,
+      "grad_norm": 0.08273103833198547,
+      "learning_rate": 5.549304649174131e-06,
+      "loss": 0.251,
+      "step": 48143
+    },
+    {
+      "epoch": 3.9001944264419963,
+      "grad_norm": 0.07010772824287415,
+      "learning_rate": 5.544803996579504e-06,
+      "loss": 0.2426,
+      "step": 48144
+    },
+    {
+      "epoch": 3.9002754374594946,
+      "grad_norm": 0.07389405369758606,
+      "learning_rate": 5.540303343984878e-06,
+      "loss": 0.2398,
+      "step": 48145
+    },
+    {
+      "epoch": 3.900356448476993,
+      "grad_norm": 0.051351238042116165,
+      "learning_rate": 5.535802691390252e-06,
+      "loss": 0.2364,
+      "step": 48146
+    },
+    {
+      "epoch": 3.9004374594944915,
+      "grad_norm": 0.0765320435166359,
+      "learning_rate": 5.5313020387956255e-06,
+      "loss": 0.2237,
+      "step": 48147
+    },
+    {
+      "epoch": 3.9005184705119897,
+      "grad_norm": 0.07203897833824158,
+      "learning_rate": 5.526801386200999e-06,
+      "loss": 0.2089,
+      "step": 48148
+    },
+    {
+      "epoch": 3.900599481529488,
+      "grad_norm": 0.06963177770376205,
+      "learning_rate": 5.522300733606374e-06,
+      "loss": 0.21,
+      "step": 48149
+    },
+    {
+      "epoch": 3.9006804925469862,
+      "grad_norm": 0.07618485391139984,
+      "learning_rate": 5.517800081011746e-06,
+      "loss": 0.2284,
+      "step": 48150
+    },
+    {
+      "epoch": 3.900761503564485,
+      "grad_norm": 0.08152638375759125,
+      "learning_rate": 5.513299428417121e-06,
+      "loss": 0.2354,
+      "step": 48151
+    },
+    {
+      "epoch": 3.900842514581983,
+      "grad_norm": 0.07824641466140747,
+      "learning_rate": 5.508798775822494e-06,
+      "loss": 0.2582,
+      "step": 48152
+    },
+    {
+      "epoch": 3.9009235255994814,
+      "grad_norm": 0.07195582240819931,
+      "learning_rate": 5.504298123227868e-06,
+      "loss": 0.1995,
+      "step": 48153
+    },
+    {
+      "epoch": 3.9010045366169797,
+      "grad_norm": 0.07551611214876175,
+      "learning_rate": 5.4997974706332424e-06,
+      "loss": 0.2055,
+      "step": 48154
+    },
+    {
+      "epoch": 3.9010855476344783,
+      "grad_norm": 0.06853777915239334,
+      "learning_rate": 5.495296818038616e-06,
+      "loss": 0.1924,
+      "step": 48155
+    },
+    {
+      "epoch": 3.9011665586519766,
+      "grad_norm": 0.09760008007287979,
+      "learning_rate": 5.49079616544399e-06,
+      "loss": 0.2279,
+      "step": 48156
+    },
+    {
+      "epoch": 3.901247569669475,
+      "grad_norm": 0.060199085623025894,
+      "learning_rate": 5.486295512849363e-06,
+      "loss": 0.1821,
+      "step": 48157
+    },
+    {
+      "epoch": 3.9013285806869735,
+      "grad_norm": 0.07289555668830872,
+      "learning_rate": 5.481794860254738e-06,
+      "loss": 0.241,
+      "step": 48158
+    },
+    {
+      "epoch": 3.9014095917044718,
+      "grad_norm": 0.08234255015850067,
+      "learning_rate": 5.4772942076601105e-06,
+      "loss": 0.2462,
+      "step": 48159
+    },
+    {
+      "epoch": 3.90149060272197,
+      "grad_norm": 0.07204477488994598,
+      "learning_rate": 5.472793555065485e-06,
+      "loss": 0.2149,
+      "step": 48160
+    },
+    {
+      "epoch": 3.9015716137394687,
+      "grad_norm": 0.0728304460644722,
+      "learning_rate": 5.4682929024708585e-06,
+      "loss": 0.2578,
+      "step": 48161
+    },
+    {
+      "epoch": 3.901652624756967,
+      "grad_norm": 0.08322672545909882,
+      "learning_rate": 5.463792249876232e-06,
+      "loss": 0.2468,
+      "step": 48162
+    },
+    {
+      "epoch": 3.901733635774465,
+      "grad_norm": 0.07119564712047577,
+      "learning_rate": 5.459291597281606e-06,
+      "loss": 0.2027,
+      "step": 48163
+    },
+    {
+      "epoch": 3.901814646791964,
+      "grad_norm": 0.06412103772163391,
+      "learning_rate": 5.45479094468698e-06,
+      "loss": 0.2256,
+      "step": 48164
+    },
+    {
+      "epoch": 3.901895657809462,
+      "grad_norm": 0.07036664336919785,
+      "learning_rate": 5.450290292092353e-06,
+      "loss": 0.2439,
+      "step": 48165
+    },
+    {
+      "epoch": 3.9019766688269604,
+      "grad_norm": 0.07236220687627792,
+      "learning_rate": 5.445789639497727e-06,
+      "loss": 0.232,
+      "step": 48166
+    },
+    {
+      "epoch": 3.902057679844459,
+      "grad_norm": 0.07063250243663788,
+      "learning_rate": 5.441288986903102e-06,
+      "loss": 0.1938,
+      "step": 48167
+    },
+    {
+      "epoch": 3.9021386908619573,
+      "grad_norm": 0.08155050873756409,
+      "learning_rate": 5.436788334308475e-06,
+      "loss": 0.207,
+      "step": 48168
+    },
+    {
+      "epoch": 3.9022197018794555,
+      "grad_norm": 0.09283901005983353,
+      "learning_rate": 5.432287681713849e-06,
+      "loss": 0.2062,
+      "step": 48169
+    },
+    {
+      "epoch": 3.9023007128969542,
+      "grad_norm": 0.07058636099100113,
+      "learning_rate": 5.427787029119223e-06,
+      "loss": 0.2262,
+      "step": 48170
+    },
+    {
+      "epoch": 3.9023817239144525,
+      "grad_norm": 0.07090901583433151,
+      "learning_rate": 5.423286376524596e-06,
+      "loss": 0.217,
+      "step": 48171
+    },
+    {
+      "epoch": 3.9024627349319507,
+      "grad_norm": 0.06685709208250046,
+      "learning_rate": 5.41878572392997e-06,
+      "loss": 0.2017,
+      "step": 48172
+    },
+    {
+      "epoch": 3.902543745949449,
+      "grad_norm": 0.0714755430817604,
+      "learning_rate": 5.414285071335344e-06,
+      "loss": 0.2207,
+      "step": 48173
+    },
+    {
+      "epoch": 3.9026247569669477,
+      "grad_norm": 0.06478474289178848,
+      "learning_rate": 5.409784418740718e-06,
+      "loss": 0.1963,
+      "step": 48174
+    },
+    {
+      "epoch": 3.902705767984446,
+      "grad_norm": 0.059301264584064484,
+      "learning_rate": 5.4052837661460915e-06,
+      "loss": 0.1878,
+      "step": 48175
+    },
+    {
+      "epoch": 3.902786779001944,
+      "grad_norm": 0.07554680854082108,
+      "learning_rate": 5.400783113551465e-06,
+      "loss": 0.196,
+      "step": 48176
+    },
+    {
+      "epoch": 3.9028677900194424,
+      "grad_norm": 0.071644127368927,
+      "learning_rate": 5.396282460956839e-06,
+      "loss": 0.2209,
+      "step": 48177
+    },
+    {
+      "epoch": 3.902948801036941,
+      "grad_norm": 0.07129331678152084,
+      "learning_rate": 5.391781808362212e-06,
+      "loss": 0.1822,
+      "step": 48178
+    },
+    {
+      "epoch": 3.9030298120544393,
+      "grad_norm": 0.08478628844022751,
+      "learning_rate": 5.387281155767587e-06,
+      "loss": 0.2269,
+      "step": 48179
+    },
+    {
+      "epoch": 3.9031108230719376,
+      "grad_norm": 0.06071711704134941,
+      "learning_rate": 5.38278050317296e-06,
+      "loss": 0.2276,
+      "step": 48180
+    },
+    {
+      "epoch": 3.9031918340894363,
+      "grad_norm": 0.0666237398982048,
+      "learning_rate": 5.378279850578334e-06,
+      "loss": 0.2171,
+      "step": 48181
+    },
+    {
+      "epoch": 3.9032728451069345,
+      "grad_norm": 0.07781444489955902,
+      "learning_rate": 5.373779197983708e-06,
+      "loss": 0.2382,
+      "step": 48182
+    },
+    {
+      "epoch": 3.9033538561244328,
+      "grad_norm": 0.07283270359039307,
+      "learning_rate": 5.369278545389082e-06,
+      "loss": 0.2168,
+      "step": 48183
+    },
+    {
+      "epoch": 3.9034348671419314,
+      "grad_norm": 0.07119312882423401,
+      "learning_rate": 5.364777892794456e-06,
+      "loss": 0.2312,
+      "step": 48184
+    },
+    {
+      "epoch": 3.9035158781594297,
+      "grad_norm": 0.06919389218091965,
+      "learning_rate": 5.360277240199829e-06,
+      "loss": 0.2035,
+      "step": 48185
+    },
+    {
+      "epoch": 3.903596889176928,
+      "grad_norm": 0.09564153850078583,
+      "learning_rate": 5.355776587605203e-06,
+      "loss": 0.2171,
+      "step": 48186
+    },
+    {
+      "epoch": 3.9036779001944266,
+      "grad_norm": 0.06871563196182251,
+      "learning_rate": 5.3512759350105764e-06,
+      "loss": 0.2065,
+      "step": 48187
+    },
+    {
+      "epoch": 3.903758911211925,
+      "grad_norm": 0.07969631999731064,
+      "learning_rate": 5.346775282415951e-06,
+      "loss": 0.2581,
+      "step": 48188
+    },
+    {
+      "epoch": 3.903839922229423,
+      "grad_norm": 0.07645699381828308,
+      "learning_rate": 5.3422746298213245e-06,
+      "loss": 0.2623,
+      "step": 48189
+    },
+    {
+      "epoch": 3.903920933246922,
+      "grad_norm": 0.07592182606458664,
+      "learning_rate": 5.337773977226698e-06,
+      "loss": 0.2024,
+      "step": 48190
+    },
+    {
+      "epoch": 3.90400194426442,
+      "grad_norm": 0.07032636553049088,
+      "learning_rate": 5.333273324632072e-06,
+      "loss": 0.2,
+      "step": 48191
+    },
+    {
+      "epoch": 3.9040829552819183,
+      "grad_norm": 0.06590494513511658,
+      "learning_rate": 5.328772672037446e-06,
+      "loss": 0.2019,
+      "step": 48192
+    },
+    {
+      "epoch": 3.904163966299417,
+      "grad_norm": 0.07423686981201172,
+      "learning_rate": 5.324272019442819e-06,
+      "loss": 0.2317,
+      "step": 48193
+    },
+    {
+      "epoch": 3.904244977316915,
+      "grad_norm": 0.0696929395198822,
+      "learning_rate": 5.319771366848193e-06,
+      "loss": 0.2315,
+      "step": 48194
+    },
+    {
+      "epoch": 3.9043259883344135,
+      "grad_norm": 0.08379188925027847,
+      "learning_rate": 5.315270714253567e-06,
+      "loss": 0.2312,
+      "step": 48195
+    },
+    {
+      "epoch": 3.9044069993519117,
+      "grad_norm": 0.07730215787887573,
+      "learning_rate": 5.3107700616589406e-06,
+      "loss": 0.2267,
+      "step": 48196
+    },
+    {
+      "epoch": 3.9044880103694104,
+      "grad_norm": 0.07866203784942627,
+      "learning_rate": 5.306269409064315e-06,
+      "loss": 0.2381,
+      "step": 48197
+    },
+    {
+      "epoch": 3.9045690213869086,
+      "grad_norm": 0.08509163558483124,
+      "learning_rate": 5.301768756469689e-06,
+      "loss": 0.2215,
+      "step": 48198
+    },
+    {
+      "epoch": 3.904650032404407,
+      "grad_norm": 0.07737905532121658,
+      "learning_rate": 5.297268103875062e-06,
+      "loss": 0.2104,
+      "step": 48199
+    },
+    {
+      "epoch": 3.904731043421905,
+      "grad_norm": 0.08024188876152039,
+      "learning_rate": 5.292767451280436e-06,
+      "loss": 0.2141,
+      "step": 48200
+    },
+    {
+      "epoch": 3.904812054439404,
+      "grad_norm": 0.0622677244246006,
+      "learning_rate": 5.28826679868581e-06,
+      "loss": 0.1994,
+      "step": 48201
+    },
+    {
+      "epoch": 3.904893065456902,
+      "grad_norm": 0.08390653878450394,
+      "learning_rate": 5.283766146091183e-06,
+      "loss": 0.2637,
+      "step": 48202
+    },
+    {
+      "epoch": 3.9049740764744003,
+      "grad_norm": 0.07863643020391464,
+      "learning_rate": 5.2792654934965575e-06,
+      "loss": 0.2527,
+      "step": 48203
+    },
+    {
+      "epoch": 3.905055087491899,
+      "grad_norm": 0.07339465618133545,
+      "learning_rate": 5.274764840901931e-06,
+      "loss": 0.2585,
+      "step": 48204
+    },
+    {
+      "epoch": 3.9051360985093972,
+      "grad_norm": 0.06816806644201279,
+      "learning_rate": 5.270264188307305e-06,
+      "loss": 0.1963,
+      "step": 48205
+    },
+    {
+      "epoch": 3.9052171095268955,
+      "grad_norm": 0.07438135147094727,
+      "learning_rate": 5.265763535712678e-06,
+      "loss": 0.2358,
+      "step": 48206
+    },
+    {
+      "epoch": 3.905298120544394,
+      "grad_norm": 0.09304346889257431,
+      "learning_rate": 5.261262883118053e-06,
+      "loss": 0.2418,
+      "step": 48207
+    },
+    {
+      "epoch": 3.9053791315618924,
+      "grad_norm": 0.07517294585704803,
+      "learning_rate": 5.2567622305234255e-06,
+      "loss": 0.2178,
+      "step": 48208
+    },
+    {
+      "epoch": 3.9054601425793907,
+      "grad_norm": 0.06006520986557007,
+      "learning_rate": 5.2522615779288e-06,
+      "loss": 0.2233,
+      "step": 48209
+    },
+    {
+      "epoch": 3.9055411535968894,
+      "grad_norm": 0.061688102781772614,
+      "learning_rate": 5.247760925334174e-06,
+      "loss": 0.1981,
+      "step": 48210
+    },
+    {
+      "epoch": 3.9056221646143876,
+      "grad_norm": 0.11346621811389923,
+      "learning_rate": 5.243260272739547e-06,
+      "loss": 0.2113,
+      "step": 48211
+    },
+    {
+      "epoch": 3.905703175631886,
+      "grad_norm": 0.06851883232593536,
+      "learning_rate": 5.238759620144922e-06,
+      "loss": 0.2678,
+      "step": 48212
+    },
+    {
+      "epoch": 3.9057841866493845,
+      "grad_norm": 0.07190129160881042,
+      "learning_rate": 5.234258967550295e-06,
+      "loss": 0.2227,
+      "step": 48213
+    },
+    {
+      "epoch": 3.905865197666883,
+      "grad_norm": 0.07402346283197403,
+      "learning_rate": 5.229758314955669e-06,
+      "loss": 0.23,
+      "step": 48214
+    },
+    {
+      "epoch": 3.905946208684381,
+      "grad_norm": 0.0806785300374031,
+      "learning_rate": 5.225257662361042e-06,
+      "loss": 0.2571,
+      "step": 48215
+    },
+    {
+      "epoch": 3.9060272197018797,
+      "grad_norm": 0.06920921057462692,
+      "learning_rate": 5.220757009766417e-06,
+      "loss": 0.2173,
+      "step": 48216
+    },
+    {
+      "epoch": 3.906108230719378,
+      "grad_norm": 0.06963902711868286,
+      "learning_rate": 5.21625635717179e-06,
+      "loss": 0.19,
+      "step": 48217
+    },
+    {
+      "epoch": 3.906189241736876,
+      "grad_norm": 0.06644149869680405,
+      "learning_rate": 5.211755704577164e-06,
+      "loss": 0.2332,
+      "step": 48218
+    },
+    {
+      "epoch": 3.9062702527543745,
+      "grad_norm": 0.06656293570995331,
+      "learning_rate": 5.207255051982538e-06,
+      "loss": 0.1953,
+      "step": 48219
+    },
+    {
+      "epoch": 3.906351263771873,
+      "grad_norm": 0.07155372947454453,
+      "learning_rate": 5.202754399387911e-06,
+      "loss": 0.2218,
+      "step": 48220
+    },
+    {
+      "epoch": 3.9064322747893714,
+      "grad_norm": 0.06269364804029465,
+      "learning_rate": 5.198253746793285e-06,
+      "loss": 0.1846,
+      "step": 48221
+    },
+    {
+      "epoch": 3.9065132858068696,
+      "grad_norm": 0.06916271895170212,
+      "learning_rate": 5.193753094198659e-06,
+      "loss": 0.2343,
+      "step": 48222
+    },
+    {
+      "epoch": 3.906594296824368,
+      "grad_norm": 0.07343260943889618,
+      "learning_rate": 5.189252441604032e-06,
+      "loss": 0.2274,
+      "step": 48223
+    },
+    {
+      "epoch": 3.9066753078418666,
+      "grad_norm": 0.08011267334222794,
+      "learning_rate": 5.1847517890094065e-06,
+      "loss": 0.2191,
+      "step": 48224
+    },
+    {
+      "epoch": 3.906756318859365,
+      "grad_norm": 0.07275541871786118,
+      "learning_rate": 5.180251136414781e-06,
+      "loss": 0.2005,
+      "step": 48225
+    },
+    {
+      "epoch": 3.906837329876863,
+      "grad_norm": 0.08077385276556015,
+      "learning_rate": 5.175750483820154e-06,
+      "loss": 0.2796,
+      "step": 48226
+    },
+    {
+      "epoch": 3.9069183408943617,
+      "grad_norm": 0.08260789513587952,
+      "learning_rate": 5.171249831225528e-06,
+      "loss": 0.2247,
+      "step": 48227
+    },
+    {
+      "epoch": 3.90699935191186,
+      "grad_norm": 0.05853225663304329,
+      "learning_rate": 5.166749178630902e-06,
+      "loss": 0.1863,
+      "step": 48228
+    },
+    {
+      "epoch": 3.9070803629293582,
+      "grad_norm": 0.08310955762863159,
+      "learning_rate": 5.162248526036275e-06,
+      "loss": 0.2158,
+      "step": 48229
+    },
+    {
+      "epoch": 3.907161373946857,
+      "grad_norm": 0.07807844877243042,
+      "learning_rate": 5.157747873441649e-06,
+      "loss": 0.2443,
+      "step": 48230
+    },
+    {
+      "epoch": 3.907242384964355,
+      "grad_norm": 0.07691746205091476,
+      "learning_rate": 5.1532472208470235e-06,
+      "loss": 0.2138,
+      "step": 48231
+    },
+    {
+      "epoch": 3.9073233959818534,
+      "grad_norm": 0.07122568041086197,
+      "learning_rate": 5.148746568252397e-06,
+      "loss": 0.2198,
+      "step": 48232
+    },
+    {
+      "epoch": 3.907404406999352,
+      "grad_norm": 0.06453979015350342,
+      "learning_rate": 5.144245915657771e-06,
+      "loss": 0.2408,
+      "step": 48233
+    },
+    {
+      "epoch": 3.9074854180168503,
+      "grad_norm": 0.07711661607027054,
+      "learning_rate": 5.139745263063144e-06,
+      "loss": 0.201,
+      "step": 48234
+    },
+    {
+      "epoch": 3.9075664290343486,
+      "grad_norm": 0.08223751932382584,
+      "learning_rate": 5.135244610468518e-06,
+      "loss": 0.2204,
+      "step": 48235
+    },
+    {
+      "epoch": 3.9076474400518473,
+      "grad_norm": 0.08006234467029572,
+      "learning_rate": 5.1307439578738915e-06,
+      "loss": 0.2498,
+      "step": 48236
+    },
+    {
+      "epoch": 3.9077284510693455,
+      "grad_norm": 0.06987843662500381,
+      "learning_rate": 5.126243305279266e-06,
+      "loss": 0.2437,
+      "step": 48237
+    },
+    {
+      "epoch": 3.9078094620868438,
+      "grad_norm": 0.06831565499305725,
+      "learning_rate": 5.1217426526846395e-06,
+      "loss": 0.1795,
+      "step": 48238
+    },
+    {
+      "epoch": 3.9078904731043425,
+      "grad_norm": 0.0886467844247818,
+      "learning_rate": 5.117242000090013e-06,
+      "loss": 0.2054,
+      "step": 48239
+    },
+    {
+      "epoch": 3.9079714841218407,
+      "grad_norm": 0.09053325653076172,
+      "learning_rate": 5.112741347495388e-06,
+      "loss": 0.2823,
+      "step": 48240
+    },
+    {
+      "epoch": 3.908052495139339,
+      "grad_norm": 0.07012274116277695,
+      "learning_rate": 5.108240694900761e-06,
+      "loss": 0.2037,
+      "step": 48241
+    },
+    {
+      "epoch": 3.908133506156837,
+      "grad_norm": 0.07728597521781921,
+      "learning_rate": 5.103740042306135e-06,
+      "loss": 0.2068,
+      "step": 48242
+    },
+    {
+      "epoch": 3.9082145171743354,
+      "grad_norm": 0.06844566017389297,
+      "learning_rate": 5.099239389711508e-06,
+      "loss": 0.1873,
+      "step": 48243
+    },
+    {
+      "epoch": 3.908295528191834,
+      "grad_norm": 0.0649128332734108,
+      "learning_rate": 5.094738737116882e-06,
+      "loss": 0.1839,
+      "step": 48244
+    },
+    {
+      "epoch": 3.9083765392093324,
+      "grad_norm": 0.06959527730941772,
+      "learning_rate": 5.090238084522256e-06,
+      "loss": 0.2126,
+      "step": 48245
+    },
+    {
+      "epoch": 3.9084575502268306,
+      "grad_norm": 0.0665079727768898,
+      "learning_rate": 5.08573743192763e-06,
+      "loss": 0.2057,
+      "step": 48246
+    },
+    {
+      "epoch": 3.9085385612443293,
+      "grad_norm": 0.09884722530841827,
+      "learning_rate": 5.081236779333004e-06,
+      "loss": 0.2282,
+      "step": 48247
+    },
+    {
+      "epoch": 3.9086195722618275,
+      "grad_norm": 0.07287062704563141,
+      "learning_rate": 5.076736126738377e-06,
+      "loss": 0.2146,
+      "step": 48248
+    },
+    {
+      "epoch": 3.908700583279326,
+      "grad_norm": 0.07356756180524826,
+      "learning_rate": 5.072235474143751e-06,
+      "loss": 0.2424,
+      "step": 48249
+    },
+    {
+      "epoch": 3.9087815942968245,
+      "grad_norm": 0.0727347582578659,
+      "learning_rate": 5.067734821549125e-06,
+      "loss": 0.1947,
+      "step": 48250
+    },
+    {
+      "epoch": 3.9088626053143227,
+      "grad_norm": 0.06179845333099365,
+      "learning_rate": 5.063234168954498e-06,
+      "loss": 0.2131,
+      "step": 48251
+    },
+    {
+      "epoch": 3.908943616331821,
+      "grad_norm": 0.08347577601671219,
+      "learning_rate": 5.0587335163598725e-06,
+      "loss": 0.2417,
+      "step": 48252
+    },
+    {
+      "epoch": 3.9090246273493197,
+      "grad_norm": 0.07791955024003983,
+      "learning_rate": 5.054232863765247e-06,
+      "loss": 0.2335,
+      "step": 48253
+    },
+    {
+      "epoch": 3.909105638366818,
+      "grad_norm": 0.07559989392757416,
+      "learning_rate": 5.04973221117062e-06,
+      "loss": 0.203,
+      "step": 48254
+    },
+    {
+      "epoch": 3.909186649384316,
+      "grad_norm": 0.06269510090351105,
+      "learning_rate": 5.045231558575994e-06,
+      "loss": 0.2261,
+      "step": 48255
+    },
+    {
+      "epoch": 3.909267660401815,
+      "grad_norm": 0.06181592866778374,
+      "learning_rate": 5.040730905981368e-06,
+      "loss": 0.2074,
+      "step": 48256
+    },
+    {
+      "epoch": 3.909348671419313,
+      "grad_norm": 0.07220128178596497,
+      "learning_rate": 5.036230253386741e-06,
+      "loss": 0.2157,
+      "step": 48257
+    },
+    {
+      "epoch": 3.9094296824368113,
+      "grad_norm": 0.07272609323263168,
+      "learning_rate": 5.031729600792115e-06,
+      "loss": 0.2441,
+      "step": 48258
+    },
+    {
+      "epoch": 3.90951069345431,
+      "grad_norm": 0.0771983340382576,
+      "learning_rate": 5.0272289481974894e-06,
+      "loss": 0.2157,
+      "step": 48259
+    },
+    {
+      "epoch": 3.9095917044718083,
+      "grad_norm": 0.08448202162981033,
+      "learning_rate": 5.022728295602862e-06,
+      "loss": 0.207,
+      "step": 48260
+    },
+    {
+      "epoch": 3.9096727154893065,
+      "grad_norm": 0.07362006604671478,
+      "learning_rate": 5.018227643008237e-06,
+      "loss": 0.2275,
+      "step": 48261
+    },
+    {
+      "epoch": 3.909753726506805,
+      "grad_norm": 0.05870373547077179,
+      "learning_rate": 5.01372699041361e-06,
+      "loss": 0.2229,
+      "step": 48262
+    },
+    {
+      "epoch": 3.9098347375243034,
+      "grad_norm": 0.07628266513347626,
+      "learning_rate": 5.009226337818984e-06,
+      "loss": 0.2208,
+      "step": 48263
+    },
+    {
+      "epoch": 3.9099157485418017,
+      "grad_norm": 0.08400886505842209,
+      "learning_rate": 5.0047256852243574e-06,
+      "loss": 0.2524,
+      "step": 48264
+    },
+    {
+      "epoch": 3.9099967595593,
+      "grad_norm": 0.0675138533115387,
+      "learning_rate": 5.000225032629732e-06,
+      "loss": 0.1857,
+      "step": 48265
+    },
+    {
+      "epoch": 3.910077770576798,
+      "grad_norm": 0.08384126424789429,
+      "learning_rate": 4.995724380035105e-06,
+      "loss": 0.2271,
+      "step": 48266
+    },
+    {
+      "epoch": 3.910158781594297,
+      "grad_norm": 0.0884663388133049,
+      "learning_rate": 4.991223727440479e-06,
+      "loss": 0.267,
+      "step": 48267
+    },
+    {
+      "epoch": 3.910239792611795,
+      "grad_norm": 0.06389939039945602,
+      "learning_rate": 4.9867230748458536e-06,
+      "loss": 0.2102,
+      "step": 48268
+    },
+    {
+      "epoch": 3.9103208036292934,
+      "grad_norm": 0.060799382627010345,
+      "learning_rate": 4.982222422251226e-06,
+      "loss": 0.2267,
+      "step": 48269
+    },
+    {
+      "epoch": 3.910401814646792,
+      "grad_norm": 0.07198642194271088,
+      "learning_rate": 4.977721769656601e-06,
+      "loss": 0.2249,
+      "step": 48270
+    },
+    {
+      "epoch": 3.9104828256642903,
+      "grad_norm": 0.0687641128897667,
+      "learning_rate": 4.973221117061974e-06,
+      "loss": 0.1983,
+      "step": 48271
+    },
+    {
+      "epoch": 3.9105638366817885,
+      "grad_norm": 0.07991275936365128,
+      "learning_rate": 4.968720464467348e-06,
+      "loss": 0.253,
+      "step": 48272
+    },
+    {
+      "epoch": 3.910644847699287,
+      "grad_norm": 0.07373076677322388,
+      "learning_rate": 4.9642198118727216e-06,
+      "loss": 0.235,
+      "step": 48273
+    },
+    {
+      "epoch": 3.9107258587167855,
+      "grad_norm": 0.07304935157299042,
+      "learning_rate": 4.959719159278096e-06,
+      "loss": 0.1938,
+      "step": 48274
+    },
+    {
+      "epoch": 3.9108068697342837,
+      "grad_norm": 0.06456219404935837,
+      "learning_rate": 4.955218506683469e-06,
+      "loss": 0.2212,
+      "step": 48275
+    },
+    {
+      "epoch": 3.9108878807517824,
+      "grad_norm": 0.06691355258226395,
+      "learning_rate": 4.950717854088843e-06,
+      "loss": 0.198,
+      "step": 48276
+    },
+    {
+      "epoch": 3.9109688917692806,
+      "grad_norm": 0.07013965398073196,
+      "learning_rate": 4.946217201494217e-06,
+      "loss": 0.2197,
+      "step": 48277
+    },
+    {
+      "epoch": 3.911049902786779,
+      "grad_norm": 0.0730832889676094,
+      "learning_rate": 4.9417165488995904e-06,
+      "loss": 0.1978,
+      "step": 48278
+    },
+    {
+      "epoch": 3.9111309138042776,
+      "grad_norm": 0.06756222993135452,
+      "learning_rate": 4.937215896304964e-06,
+      "loss": 0.2059,
+      "step": 48279
+    },
+    {
+      "epoch": 3.911211924821776,
+      "grad_norm": 0.05772652477025986,
+      "learning_rate": 4.9327152437103385e-06,
+      "loss": 0.2107,
+      "step": 48280
+    },
+    {
+      "epoch": 3.911292935839274,
+      "grad_norm": 0.07321856915950775,
+      "learning_rate": 4.928214591115712e-06,
+      "loss": 0.2468,
+      "step": 48281
+    },
+    {
+      "epoch": 3.9113739468567728,
+      "grad_norm": 0.07220222055912018,
+      "learning_rate": 4.923713938521086e-06,
+      "loss": 0.2097,
+      "step": 48282
+    },
+    {
+      "epoch": 3.911454957874271,
+      "grad_norm": 0.08239828050136566,
+      "learning_rate": 4.91921328592646e-06,
+      "loss": 0.2462,
+      "step": 48283
+    },
+    {
+      "epoch": 3.9115359688917692,
+      "grad_norm": 0.053629614412784576,
+      "learning_rate": 4.914712633331833e-06,
+      "loss": 0.1856,
+      "step": 48284
+    },
+    {
+      "epoch": 3.911616979909268,
+      "grad_norm": 0.07192451506853104,
+      "learning_rate": 4.910211980737207e-06,
+      "loss": 0.2489,
+      "step": 48285
+    },
+    {
+      "epoch": 3.911697990926766,
+      "grad_norm": 0.07000795751810074,
+      "learning_rate": 4.905711328142581e-06,
+      "loss": 0.2101,
+      "step": 48286
+    },
+    {
+      "epoch": 3.9117790019442644,
+      "grad_norm": 0.06989791244268417,
+      "learning_rate": 4.9012106755479546e-06,
+      "loss": 0.2376,
+      "step": 48287
+    },
+    {
+      "epoch": 3.9118600129617627,
+      "grad_norm": 0.08663799613714218,
+      "learning_rate": 4.896710022953328e-06,
+      "loss": 0.2485,
+      "step": 48288
+    },
+    {
+      "epoch": 3.911941023979261,
+      "grad_norm": 0.07052966952323914,
+      "learning_rate": 4.892209370358703e-06,
+      "loss": 0.1894,
+      "step": 48289
+    },
+    {
+      "epoch": 3.9120220349967596,
+      "grad_norm": 0.05605921521782875,
+      "learning_rate": 4.887708717764076e-06,
+      "loss": 0.1966,
+      "step": 48290
+    },
+    {
+      "epoch": 3.912103046014258,
+      "grad_norm": 0.06981483846902847,
+      "learning_rate": 4.88320806516945e-06,
+      "loss": 0.2424,
+      "step": 48291
+    },
+    {
+      "epoch": 3.912184057031756,
+      "grad_norm": 0.05360450968146324,
+      "learning_rate": 4.878707412574823e-06,
+      "loss": 0.1981,
+      "step": 48292
+    },
+    {
+      "epoch": 3.912265068049255,
+      "grad_norm": 0.07205212116241455,
+      "learning_rate": 4.874206759980197e-06,
+      "loss": 0.1913,
+      "step": 48293
+    },
+    {
+      "epoch": 3.912346079066753,
+      "grad_norm": 0.06900376826524734,
+      "learning_rate": 4.869706107385571e-06,
+      "loss": 0.2177,
+      "step": 48294
+    },
+    {
+      "epoch": 3.9124270900842513,
+      "grad_norm": 0.07426581531763077,
+      "learning_rate": 4.865205454790945e-06,
+      "loss": 0.2662,
+      "step": 48295
+    },
+    {
+      "epoch": 3.91250810110175,
+      "grad_norm": 0.0741383284330368,
+      "learning_rate": 4.860704802196319e-06,
+      "loss": 0.2323,
+      "step": 48296
+    },
+    {
+      "epoch": 3.912589112119248,
+      "grad_norm": 0.07219462841749191,
+      "learning_rate": 4.856204149601692e-06,
+      "loss": 0.2199,
+      "step": 48297
+    },
+    {
+      "epoch": 3.9126701231367464,
+      "grad_norm": 0.06240931153297424,
+      "learning_rate": 4.851703497007067e-06,
+      "loss": 0.2398,
+      "step": 48298
+    },
+    {
+      "epoch": 3.912751134154245,
+      "grad_norm": 0.0665903314948082,
+      "learning_rate": 4.84720284441244e-06,
+      "loss": 0.2174,
+      "step": 48299
+    },
+    {
+      "epoch": 3.9128321451717434,
+      "grad_norm": 0.06555341929197311,
+      "learning_rate": 4.842702191817814e-06,
+      "loss": 0.1886,
+      "step": 48300
+    },
+    {
+      "epoch": 3.9129131561892416,
+      "grad_norm": 0.06640680134296417,
+      "learning_rate": 4.8382015392231875e-06,
+      "loss": 0.1735,
+      "step": 48301
+    },
+    {
+      "epoch": 3.9129941672067403,
+      "grad_norm": 0.07242927700281143,
+      "learning_rate": 4.833700886628562e-06,
+      "loss": 0.2254,
+      "step": 48302
+    },
+    {
+      "epoch": 3.9130751782242386,
+      "grad_norm": 0.06851553171873093,
+      "learning_rate": 4.829200234033935e-06,
+      "loss": 0.1927,
+      "step": 48303
+    },
+    {
+      "epoch": 3.913156189241737,
+      "grad_norm": 0.07618321478366852,
+      "learning_rate": 4.824699581439309e-06,
+      "loss": 0.2333,
+      "step": 48304
+    },
+    {
+      "epoch": 3.9132372002592355,
+      "grad_norm": 0.07696599513292313,
+      "learning_rate": 4.820198928844683e-06,
+      "loss": 0.2151,
+      "step": 48305
+    },
+    {
+      "epoch": 3.9133182112767337,
+      "grad_norm": 0.07403657585382462,
+      "learning_rate": 4.815698276250056e-06,
+      "loss": 0.2144,
+      "step": 48306
+    },
+    {
+      "epoch": 3.913399222294232,
+      "grad_norm": 0.07694181799888611,
+      "learning_rate": 4.81119762365543e-06,
+      "loss": 0.2745,
+      "step": 48307
+    },
+    {
+      "epoch": 3.9134802333117307,
+      "grad_norm": 0.06381962448358536,
+      "learning_rate": 4.8066969710608045e-06,
+      "loss": 0.2746,
+      "step": 48308
+    },
+    {
+      "epoch": 3.913561244329229,
+      "grad_norm": 0.06298644095659256,
+      "learning_rate": 4.802196318466177e-06,
+      "loss": 0.1824,
+      "step": 48309
+    },
+    {
+      "epoch": 3.913642255346727,
+      "grad_norm": 0.07825368642807007,
+      "learning_rate": 4.797695665871552e-06,
+      "loss": 0.2237,
+      "step": 48310
+    },
+    {
+      "epoch": 3.9137232663642254,
+      "grad_norm": 0.05910641327500343,
+      "learning_rate": 4.793195013276926e-06,
+      "loss": 0.1973,
+      "step": 48311
+    },
+    {
+      "epoch": 3.9138042773817237,
+      "grad_norm": 0.08719029277563095,
+      "learning_rate": 4.788694360682299e-06,
+      "loss": 0.2168,
+      "step": 48312
+    },
+    {
+      "epoch": 3.9138852883992223,
+      "grad_norm": 0.07956347614526749,
+      "learning_rate": 4.784193708087673e-06,
+      "loss": 0.2315,
+      "step": 48313
+    },
+    {
+      "epoch": 3.9139662994167206,
+      "grad_norm": 0.07194239646196365,
+      "learning_rate": 4.779693055493047e-06,
+      "loss": 0.2449,
+      "step": 48314
+    },
+    {
+      "epoch": 3.914047310434219,
+      "grad_norm": 0.06650910526514053,
+      "learning_rate": 4.7751924028984205e-06,
+      "loss": 0.2138,
+      "step": 48315
+    },
+    {
+      "epoch": 3.9141283214517175,
+      "grad_norm": 0.07512077689170837,
+      "learning_rate": 4.770691750303794e-06,
+      "loss": 0.2245,
+      "step": 48316
+    },
+    {
+      "epoch": 3.9142093324692158,
+      "grad_norm": 0.09508345276117325,
+      "learning_rate": 4.766191097709169e-06,
+      "loss": 0.3053,
+      "step": 48317
+    },
+    {
+      "epoch": 3.914290343486714,
+      "grad_norm": 0.07228030264377594,
+      "learning_rate": 4.761690445114541e-06,
+      "loss": 0.2202,
+      "step": 48318
+    },
+    {
+      "epoch": 3.9143713545042127,
+      "grad_norm": 0.07119833678007126,
+      "learning_rate": 4.757189792519916e-06,
+      "loss": 0.1919,
+      "step": 48319
+    },
+    {
+      "epoch": 3.914452365521711,
+      "grad_norm": 0.059811804443597794,
+      "learning_rate": 4.752689139925289e-06,
+      "loss": 0.1986,
+      "step": 48320
+    },
+    {
+      "epoch": 3.914533376539209,
+      "grad_norm": 0.06910062581300735,
+      "learning_rate": 4.748188487330663e-06,
+      "loss": 0.2151,
+      "step": 48321
+    },
+    {
+      "epoch": 3.914614387556708,
+      "grad_norm": 0.06613852828741074,
+      "learning_rate": 4.743687834736037e-06,
+      "loss": 0.2022,
+      "step": 48322
+    },
+    {
+      "epoch": 3.914695398574206,
+      "grad_norm": 0.06570880860090256,
+      "learning_rate": 4.739187182141411e-06,
+      "loss": 0.1867,
+      "step": 48323
+    },
+    {
+      "epoch": 3.9147764095917044,
+      "grad_norm": 0.0788615420460701,
+      "learning_rate": 4.734686529546784e-06,
+      "loss": 0.2258,
+      "step": 48324
+    },
+    {
+      "epoch": 3.914857420609203,
+      "grad_norm": 0.0693746879696846,
+      "learning_rate": 4.730185876952158e-06,
+      "loss": 0.2217,
+      "step": 48325
+    },
+    {
+      "epoch": 3.9149384316267013,
+      "grad_norm": 0.08763092756271362,
+      "learning_rate": 4.725685224357533e-06,
+      "loss": 0.2442,
+      "step": 48326
+    },
+    {
+      "epoch": 3.9150194426441995,
+      "grad_norm": 0.06549570709466934,
+      "learning_rate": 4.7211845717629055e-06,
+      "loss": 0.2142,
+      "step": 48327
+    },
+    {
+      "epoch": 3.9151004536616982,
+      "grad_norm": 0.08535436540842056,
+      "learning_rate": 4.71668391916828e-06,
+      "loss": 0.249,
+      "step": 48328
+    },
+    {
+      "epoch": 3.9151814646791965,
+      "grad_norm": 0.0782715380191803,
+      "learning_rate": 4.7121832665736535e-06,
+      "loss": 0.2268,
+      "step": 48329
+    },
+    {
+      "epoch": 3.9152624756966947,
+      "grad_norm": 0.07017436623573303,
+      "learning_rate": 4.707682613979027e-06,
+      "loss": 0.231,
+      "step": 48330
+    },
+    {
+      "epoch": 3.915343486714193,
+      "grad_norm": 0.06549299508333206,
+      "learning_rate": 4.703181961384401e-06,
+      "loss": 0.2067,
+      "step": 48331
+    },
+    {
+      "epoch": 3.9154244977316917,
+      "grad_norm": 0.07573448866605759,
+      "learning_rate": 4.698681308789775e-06,
+      "loss": 0.2563,
+      "step": 48332
+    },
+    {
+      "epoch": 3.91550550874919,
+      "grad_norm": 0.07193144410848618,
+      "learning_rate": 4.694180656195148e-06,
+      "loss": 0.2246,
+      "step": 48333
+    },
+    {
+      "epoch": 3.915586519766688,
+      "grad_norm": 0.0836506187915802,
+      "learning_rate": 4.689680003600522e-06,
+      "loss": 0.2372,
+      "step": 48334
+    },
+    {
+      "epoch": 3.9156675307841864,
+      "grad_norm": 0.07173941284418106,
+      "learning_rate": 4.685179351005896e-06,
+      "loss": 0.2322,
+      "step": 48335
+    },
+    {
+      "epoch": 3.915748541801685,
+      "grad_norm": 0.0687011256814003,
+      "learning_rate": 4.68067869841127e-06,
+      "loss": 0.2385,
+      "step": 48336
+    },
+    {
+      "epoch": 3.9158295528191833,
+      "grad_norm": 0.06658374518156052,
+      "learning_rate": 4.676178045816643e-06,
+      "loss": 0.2497,
+      "step": 48337
+    },
+    {
+      "epoch": 3.9159105638366816,
+      "grad_norm": 0.0663018673658371,
+      "learning_rate": 4.671677393222018e-06,
+      "loss": 0.197,
+      "step": 48338
+    },
+    {
+      "epoch": 3.9159915748541803,
+      "grad_norm": 0.058599188923835754,
+      "learning_rate": 4.667176740627391e-06,
+      "loss": 0.2165,
+      "step": 48339
+    },
+    {
+      "epoch": 3.9160725858716785,
+      "grad_norm": 0.06904032081365585,
+      "learning_rate": 4.662676088032765e-06,
+      "loss": 0.2292,
+      "step": 48340
+    },
+    {
+      "epoch": 3.9161535968891767,
+      "grad_norm": 0.08278106898069382,
+      "learning_rate": 4.658175435438139e-06,
+      "loss": 0.2045,
+      "step": 48341
+    },
+    {
+      "epoch": 3.9162346079066754,
+      "grad_norm": 0.0761694386601448,
+      "learning_rate": 4.653674782843512e-06,
+      "loss": 0.2144,
+      "step": 48342
+    },
+    {
+      "epoch": 3.9163156189241737,
+      "grad_norm": 0.07530530542135239,
+      "learning_rate": 4.6491741302488865e-06,
+      "loss": 0.2537,
+      "step": 48343
+    },
+    {
+      "epoch": 3.916396629941672,
+      "grad_norm": 0.08102653175592422,
+      "learning_rate": 4.64467347765426e-06,
+      "loss": 0.2358,
+      "step": 48344
+    },
+    {
+      "epoch": 3.9164776409591706,
+      "grad_norm": 0.07261081039905548,
+      "learning_rate": 4.640172825059634e-06,
+      "loss": 0.249,
+      "step": 48345
+    },
+    {
+      "epoch": 3.916558651976669,
+      "grad_norm": 0.09279650449752808,
+      "learning_rate": 4.635672172465007e-06,
+      "loss": 0.2493,
+      "step": 48346
+    },
+    {
+      "epoch": 3.916639662994167,
+      "grad_norm": 0.08426755666732788,
+      "learning_rate": 4.631171519870382e-06,
+      "loss": 0.2518,
+      "step": 48347
+    },
+    {
+      "epoch": 3.916720674011666,
+      "grad_norm": 0.06982297450304031,
+      "learning_rate": 4.626670867275755e-06,
+      "loss": 0.2133,
+      "step": 48348
+    },
+    {
+      "epoch": 3.916801685029164,
+      "grad_norm": 0.062374137341976166,
+      "learning_rate": 4.622170214681129e-06,
+      "loss": 0.2058,
+      "step": 48349
+    },
+    {
+      "epoch": 3.9168826960466623,
+      "grad_norm": 0.06334874778985977,
+      "learning_rate": 4.617669562086503e-06,
+      "loss": 0.2218,
+      "step": 48350
+    },
+    {
+      "epoch": 3.916963707064161,
+      "grad_norm": 0.07470595091581345,
+      "learning_rate": 4.613168909491876e-06,
+      "loss": 0.2633,
+      "step": 48351
+    },
+    {
+      "epoch": 3.917044718081659,
+      "grad_norm": 0.06525980681180954,
+      "learning_rate": 4.60866825689725e-06,
+      "loss": 0.2238,
+      "step": 48352
+    },
+    {
+      "epoch": 3.9171257290991575,
+      "grad_norm": 0.0697428286075592,
+      "learning_rate": 4.604167604302624e-06,
+      "loss": 0.2647,
+      "step": 48353
+    },
+    {
+      "epoch": 3.9172067401166557,
+      "grad_norm": 0.07528168708086014,
+      "learning_rate": 4.599666951707998e-06,
+      "loss": 0.2266,
+      "step": 48354
+    },
+    {
+      "epoch": 3.9172877511341544,
+      "grad_norm": 0.07216835767030716,
+      "learning_rate": 4.5951662991133714e-06,
+      "loss": 0.2087,
+      "step": 48355
+    },
+    {
+      "epoch": 3.9173687621516526,
+      "grad_norm": 0.06816212087869644,
+      "learning_rate": 4.590665646518746e-06,
+      "loss": 0.1967,
+      "step": 48356
+    },
+    {
+      "epoch": 3.917449773169151,
+      "grad_norm": 0.07805240899324417,
+      "learning_rate": 4.5861649939241195e-06,
+      "loss": 0.2293,
+      "step": 48357
+    },
+    {
+      "epoch": 3.917530784186649,
+      "grad_norm": 0.06178323179483414,
+      "learning_rate": 4.581664341329493e-06,
+      "loss": 0.2204,
+      "step": 48358
+    },
+    {
+      "epoch": 3.917611795204148,
+      "grad_norm": 0.07420182973146439,
+      "learning_rate": 4.577163688734867e-06,
+      "loss": 0.2364,
+      "step": 48359
+    },
+    {
+      "epoch": 3.917692806221646,
+      "grad_norm": 0.06499535590410233,
+      "learning_rate": 4.572663036140241e-06,
+      "loss": 0.2054,
+      "step": 48360
+    },
+    {
+      "epoch": 3.9177738172391443,
+      "grad_norm": 0.07386469841003418,
+      "learning_rate": 4.568162383545614e-06,
+      "loss": 0.224,
+      "step": 48361
+    },
+    {
+      "epoch": 3.917854828256643,
+      "grad_norm": 0.07269761711359024,
+      "learning_rate": 4.563661730950988e-06,
+      "loss": 0.1961,
+      "step": 48362
+    },
+    {
+      "epoch": 3.9179358392741412,
+      "grad_norm": 0.07524976879358292,
+      "learning_rate": 4.559161078356362e-06,
+      "loss": 0.22,
+      "step": 48363
+    },
+    {
+      "epoch": 3.9180168502916395,
+      "grad_norm": 0.07413894683122635,
+      "learning_rate": 4.5546604257617356e-06,
+      "loss": 0.2123,
+      "step": 48364
+    },
+    {
+      "epoch": 3.918097861309138,
+      "grad_norm": 0.06486944854259491,
+      "learning_rate": 4.550159773167109e-06,
+      "loss": 0.2077,
+      "step": 48365
+    },
+    {
+      "epoch": 3.9181788723266364,
+      "grad_norm": 0.06430947780609131,
+      "learning_rate": 4.545659120572484e-06,
+      "loss": 0.2326,
+      "step": 48366
+    },
+    {
+      "epoch": 3.9182598833441347,
+      "grad_norm": 0.08823602646589279,
+      "learning_rate": 4.541158467977856e-06,
+      "loss": 0.2792,
+      "step": 48367
+    },
+    {
+      "epoch": 3.9183408943616334,
+      "grad_norm": 0.05794184282422066,
+      "learning_rate": 4.536657815383231e-06,
+      "loss": 0.2169,
+      "step": 48368
+    },
+    {
+      "epoch": 3.9184219053791316,
+      "grad_norm": 0.08605548739433289,
+      "learning_rate": 4.532157162788605e-06,
+      "loss": 0.2541,
+      "step": 48369
+    },
+    {
+      "epoch": 3.91850291639663,
+      "grad_norm": 0.08970436453819275,
+      "learning_rate": 4.527656510193978e-06,
+      "loss": 0.2634,
+      "step": 48370
+    },
+    {
+      "epoch": 3.9185839274141285,
+      "grad_norm": 0.07923775911331177,
+      "learning_rate": 4.5231558575993525e-06,
+      "loss": 0.2305,
+      "step": 48371
+    },
+    {
+      "epoch": 3.9186649384316268,
+      "grad_norm": 0.07092005014419556,
+      "learning_rate": 4.518655205004726e-06,
+      "loss": 0.2317,
+      "step": 48372
+    },
+    {
+      "epoch": 3.918745949449125,
+      "grad_norm": 0.08713727444410324,
+      "learning_rate": 4.5141545524101e-06,
+      "loss": 0.2262,
+      "step": 48373
+    },
+    {
+      "epoch": 3.9188269604666237,
+      "grad_norm": 0.06912411004304886,
+      "learning_rate": 4.509653899815473e-06,
+      "loss": 0.2173,
+      "step": 48374
+    },
+    {
+      "epoch": 3.918907971484122,
+      "grad_norm": 0.0842410996556282,
+      "learning_rate": 4.505153247220848e-06,
+      "loss": 0.2268,
+      "step": 48375
+    },
+    {
+      "epoch": 3.91898898250162,
+      "grad_norm": 0.06546398997306824,
+      "learning_rate": 4.5006525946262205e-06,
+      "loss": 0.2216,
+      "step": 48376
+    },
+    {
+      "epoch": 3.9190699935191184,
+      "grad_norm": 0.09463486075401306,
+      "learning_rate": 4.496151942031595e-06,
+      "loss": 0.3016,
+      "step": 48377
+    },
+    {
+      "epoch": 3.919151004536617,
+      "grad_norm": 0.0747930109500885,
+      "learning_rate": 4.4916512894369686e-06,
+      "loss": 0.2222,
+      "step": 48378
+    },
+    {
+      "epoch": 3.9192320155541154,
+      "grad_norm": 0.06386351585388184,
+      "learning_rate": 4.487150636842342e-06,
+      "loss": 0.204,
+      "step": 48379
+    },
+    {
+      "epoch": 3.9193130265716136,
+      "grad_norm": 0.06935272365808487,
+      "learning_rate": 4.482649984247716e-06,
+      "loss": 0.227,
+      "step": 48380
+    },
+    {
+      "epoch": 3.919394037589112,
+      "grad_norm": 0.0631873607635498,
+      "learning_rate": 4.47814933165309e-06,
+      "loss": 0.2332,
+      "step": 48381
+    },
+    {
+      "epoch": 3.9194750486066106,
+      "grad_norm": 0.07327888160943985,
+      "learning_rate": 4.473648679058463e-06,
+      "loss": 0.225,
+      "step": 48382
+    },
+    {
+      "epoch": 3.919556059624109,
+      "grad_norm": 0.07921938598155975,
+      "learning_rate": 4.469148026463837e-06,
+      "loss": 0.2254,
+      "step": 48383
+    },
+    {
+      "epoch": 3.919637070641607,
+      "grad_norm": 0.06906044483184814,
+      "learning_rate": 4.464647373869212e-06,
+      "loss": 0.2089,
+      "step": 48384
+    },
+    {
+      "epoch": 3.9197180816591057,
+      "grad_norm": 0.0595407597720623,
+      "learning_rate": 4.460146721274585e-06,
+      "loss": 0.1853,
+      "step": 48385
+    },
+    {
+      "epoch": 3.919799092676604,
+      "grad_norm": 0.08025841414928436,
+      "learning_rate": 4.455646068679959e-06,
+      "loss": 0.2379,
+      "step": 48386
+    },
+    {
+      "epoch": 3.9198801036941022,
+      "grad_norm": 0.06751351803541183,
+      "learning_rate": 4.451145416085333e-06,
+      "loss": 0.1824,
+      "step": 48387
+    },
+    {
+      "epoch": 3.919961114711601,
+      "grad_norm": 0.07128855586051941,
+      "learning_rate": 4.446644763490706e-06,
+      "loss": 0.2065,
+      "step": 48388
+    },
+    {
+      "epoch": 3.920042125729099,
+      "grad_norm": 0.07042013853788376,
+      "learning_rate": 4.44214411089608e-06,
+      "loss": 0.2286,
+      "step": 48389
+    },
+    {
+      "epoch": 3.9201231367465974,
+      "grad_norm": 0.07936536520719528,
+      "learning_rate": 4.437643458301454e-06,
+      "loss": 0.2273,
+      "step": 48390
+    },
+    {
+      "epoch": 3.920204147764096,
+      "grad_norm": 0.07564130425453186,
+      "learning_rate": 4.433142805706827e-06,
+      "loss": 0.2124,
+      "step": 48391
+    },
+    {
+      "epoch": 3.9202851587815943,
+      "grad_norm": 0.0812234953045845,
+      "learning_rate": 4.4286421531122015e-06,
+      "loss": 0.2354,
+      "step": 48392
+    },
+    {
+      "epoch": 3.9203661697990926,
+      "grad_norm": 0.07354864478111267,
+      "learning_rate": 4.424141500517575e-06,
+      "loss": 0.2063,
+      "step": 48393
+    },
+    {
+      "epoch": 3.9204471808165913,
+      "grad_norm": 0.060475897043943405,
+      "learning_rate": 4.419640847922949e-06,
+      "loss": 0.1771,
+      "step": 48394
+    },
+    {
+      "epoch": 3.9205281918340895,
+      "grad_norm": 0.06647646427154541,
+      "learning_rate": 4.415140195328322e-06,
+      "loss": 0.2066,
+      "step": 48395
+    },
+    {
+      "epoch": 3.9206092028515878,
+      "grad_norm": 0.07337382435798645,
+      "learning_rate": 4.410639542733697e-06,
+      "loss": 0.2287,
+      "step": 48396
+    },
+    {
+      "epoch": 3.9206902138690864,
+      "grad_norm": 0.07359199970960617,
+      "learning_rate": 4.40613889013907e-06,
+      "loss": 0.244,
+      "step": 48397
+    },
+    {
+      "epoch": 3.9207712248865847,
+      "grad_norm": 0.074060820043087,
+      "learning_rate": 4.401638237544444e-06,
+      "loss": 0.2418,
+      "step": 48398
+    },
+    {
+      "epoch": 3.920852235904083,
+      "grad_norm": 0.07838385552167892,
+      "learning_rate": 4.3971375849498185e-06,
+      "loss": 0.2612,
+      "step": 48399
+    },
+    {
+      "epoch": 3.920933246921581,
+      "grad_norm": 0.07551511377096176,
+      "learning_rate": 4.392636932355191e-06,
+      "loss": 0.2409,
+      "step": 48400
+    },
+    {
+      "epoch": 3.92101425793908,
+      "grad_norm": 0.08527684211730957,
+      "learning_rate": 4.388136279760566e-06,
+      "loss": 0.204,
+      "step": 48401
+    },
+    {
+      "epoch": 3.921095268956578,
+      "grad_norm": 0.06834513694047928,
+      "learning_rate": 4.383635627165939e-06,
+      "loss": 0.2157,
+      "step": 48402
+    },
+    {
+      "epoch": 3.9211762799740764,
+      "grad_norm": 0.06227676197886467,
+      "learning_rate": 4.379134974571313e-06,
+      "loss": 0.2034,
+      "step": 48403
+    },
+    {
+      "epoch": 3.9212572909915746,
+      "grad_norm": 0.07426196336746216,
+      "learning_rate": 4.3746343219766865e-06,
+      "loss": 0.2606,
+      "step": 48404
+    },
+    {
+      "epoch": 3.9213383020090733,
+      "grad_norm": 0.07575037330389023,
+      "learning_rate": 4.370133669382061e-06,
+      "loss": 0.22,
+      "step": 48405
+    },
+    {
+      "epoch": 3.9214193130265715,
+      "grad_norm": 0.06229977682232857,
+      "learning_rate": 4.3656330167874345e-06,
+      "loss": 0.2217,
+      "step": 48406
+    },
+    {
+      "epoch": 3.92150032404407,
+      "grad_norm": 0.07139471918344498,
+      "learning_rate": 4.361132364192808e-06,
+      "loss": 0.2437,
+      "step": 48407
+    },
+    {
+      "epoch": 3.9215813350615685,
+      "grad_norm": 0.06911724805831909,
+      "learning_rate": 4.356631711598182e-06,
+      "loss": 0.1807,
+      "step": 48408
+    },
+    {
+      "epoch": 3.9216623460790667,
+      "grad_norm": 0.07534075528383255,
+      "learning_rate": 4.352131059003556e-06,
+      "loss": 0.2281,
+      "step": 48409
+    },
+    {
+      "epoch": 3.921743357096565,
+      "grad_norm": 0.07420405745506287,
+      "learning_rate": 4.347630406408929e-06,
+      "loss": 0.2289,
+      "step": 48410
+    },
+    {
+      "epoch": 3.9218243681140637,
+      "grad_norm": 0.07598921656608582,
+      "learning_rate": 4.343129753814303e-06,
+      "loss": 0.2104,
+      "step": 48411
+    },
+    {
+      "epoch": 3.921905379131562,
+      "grad_norm": 0.061629679054021835,
+      "learning_rate": 4.338629101219677e-06,
+      "loss": 0.2472,
+      "step": 48412
+    },
+    {
+      "epoch": 3.92198639014906,
+      "grad_norm": 0.07201918214559555,
+      "learning_rate": 4.334128448625051e-06,
+      "loss": 0.1946,
+      "step": 48413
+    },
+    {
+      "epoch": 3.922067401166559,
+      "grad_norm": 0.07076304405927658,
+      "learning_rate": 4.329627796030425e-06,
+      "loss": 0.221,
+      "step": 48414
+    },
+    {
+      "epoch": 3.922148412184057,
+      "grad_norm": 0.0704672783613205,
+      "learning_rate": 4.325127143435799e-06,
+      "loss": 0.2151,
+      "step": 48415
+    },
+    {
+      "epoch": 3.9222294232015553,
+      "grad_norm": 0.06875283271074295,
+      "learning_rate": 4.320626490841172e-06,
+      "loss": 0.1926,
+      "step": 48416
+    },
+    {
+      "epoch": 3.922310434219054,
+      "grad_norm": 0.05916054546833038,
+      "learning_rate": 4.316125838246546e-06,
+      "loss": 0.2194,
+      "step": 48417
+    },
+    {
+      "epoch": 3.9223914452365523,
+      "grad_norm": 0.060847945511341095,
+      "learning_rate": 4.31162518565192e-06,
+      "loss": 0.2302,
+      "step": 48418
+    },
+    {
+      "epoch": 3.9224724562540505,
+      "grad_norm": 0.07540497928857803,
+      "learning_rate": 4.307124533057293e-06,
+      "loss": 0.2381,
+      "step": 48419
+    },
+    {
+      "epoch": 3.922553467271549,
+      "grad_norm": 0.07262145727872849,
+      "learning_rate": 4.3026238804626675e-06,
+      "loss": 0.209,
+      "step": 48420
+    },
+    {
+      "epoch": 3.9226344782890474,
+      "grad_norm": 0.061988167464733124,
+      "learning_rate": 4.298123227868041e-06,
+      "loss": 0.1932,
+      "step": 48421
+    },
+    {
+      "epoch": 3.9227154893065457,
+      "grad_norm": 0.06330009549856186,
+      "learning_rate": 4.293622575273415e-06,
+      "loss": 0.222,
+      "step": 48422
+    },
+    {
+      "epoch": 3.922796500324044,
+      "grad_norm": 0.08140560984611511,
+      "learning_rate": 4.289121922678788e-06,
+      "loss": 0.2162,
+      "step": 48423
+    },
+    {
+      "epoch": 3.9228775113415426,
+      "grad_norm": 0.07807426899671555,
+      "learning_rate": 4.284621270084163e-06,
+      "loss": 0.2249,
+      "step": 48424
+    },
+    {
+      "epoch": 3.922958522359041,
+      "grad_norm": 0.07387779653072357,
+      "learning_rate": 4.2801206174895355e-06,
+      "loss": 0.2144,
+      "step": 48425
+    },
+    {
+      "epoch": 3.923039533376539,
+      "grad_norm": 0.06375554203987122,
+      "learning_rate": 4.27561996489491e-06,
+      "loss": 0.2085,
+      "step": 48426
+    },
+    {
+      "epoch": 3.9231205443940373,
+      "grad_norm": 0.07752867043018341,
+      "learning_rate": 4.2711193123002844e-06,
+      "loss": 0.2388,
+      "step": 48427
+    },
+    {
+      "epoch": 3.923201555411536,
+      "grad_norm": 0.07950585335493088,
+      "learning_rate": 4.266618659705657e-06,
+      "loss": 0.2315,
+      "step": 48428
+    },
+    {
+      "epoch": 3.9232825664290343,
+      "grad_norm": 0.0622391402721405,
+      "learning_rate": 4.262118007111032e-06,
+      "loss": 0.2157,
+      "step": 48429
+    },
+    {
+      "epoch": 3.9233635774465325,
+      "grad_norm": 0.0833074301481247,
+      "learning_rate": 4.257617354516405e-06,
+      "loss": 0.2555,
+      "step": 48430
+    },
+    {
+      "epoch": 3.923444588464031,
+      "grad_norm": 0.06539009511470795,
+      "learning_rate": 4.253116701921779e-06,
+      "loss": 0.2225,
+      "step": 48431
+    },
+    {
+      "epoch": 3.9235255994815295,
+      "grad_norm": 0.07489827275276184,
+      "learning_rate": 4.2486160493271525e-06,
+      "loss": 0.2581,
+      "step": 48432
+    },
+    {
+      "epoch": 3.9236066104990277,
+      "grad_norm": 0.06998512893915176,
+      "learning_rate": 4.244115396732527e-06,
+      "loss": 0.176,
+      "step": 48433
+    },
+    {
+      "epoch": 3.9236876215165264,
+      "grad_norm": 0.06104741990566254,
+      "learning_rate": 4.2396147441379e-06,
+      "loss": 0.2076,
+      "step": 48434
+    },
+    {
+      "epoch": 3.9237686325340246,
+      "grad_norm": 0.07176847755908966,
+      "learning_rate": 4.235114091543274e-06,
+      "loss": 0.2334,
+      "step": 48435
+    },
+    {
+      "epoch": 3.923849643551523,
+      "grad_norm": 0.05985512211918831,
+      "learning_rate": 4.230613438948648e-06,
+      "loss": 0.2031,
+      "step": 48436
+    },
+    {
+      "epoch": 3.9239306545690216,
+      "grad_norm": 0.06858471781015396,
+      "learning_rate": 4.226112786354021e-06,
+      "loss": 0.249,
+      "step": 48437
+    },
+    {
+      "epoch": 3.92401166558652,
+      "grad_norm": 0.07977214455604553,
+      "learning_rate": 4.221612133759395e-06,
+      "loss": 0.2112,
+      "step": 48438
+    },
+    {
+      "epoch": 3.924092676604018,
+      "grad_norm": 0.07993137091398239,
+      "learning_rate": 4.217111481164769e-06,
+      "loss": 0.1969,
+      "step": 48439
+    },
+    {
+      "epoch": 3.9241736876215167,
+      "grad_norm": 0.06767088174819946,
+      "learning_rate": 4.212610828570142e-06,
+      "loss": 0.2476,
+      "step": 48440
+    },
+    {
+      "epoch": 3.924254698639015,
+      "grad_norm": 0.0661626085639,
+      "learning_rate": 4.2081101759755166e-06,
+      "loss": 0.2613,
+      "step": 48441
+    },
+    {
+      "epoch": 3.9243357096565132,
+      "grad_norm": 0.07554125785827637,
+      "learning_rate": 4.203609523380891e-06,
+      "loss": 0.207,
+      "step": 48442
+    },
+    {
+      "epoch": 3.924416720674012,
+      "grad_norm": 0.06287429481744766,
+      "learning_rate": 4.199108870786264e-06,
+      "loss": 0.2114,
+      "step": 48443
+    },
+    {
+      "epoch": 3.92449773169151,
+      "grad_norm": 0.07063333690166473,
+      "learning_rate": 4.194608218191638e-06,
+      "loss": 0.2113,
+      "step": 48444
+    },
+    {
+      "epoch": 3.9245787427090084,
+      "grad_norm": 0.074994757771492,
+      "learning_rate": 4.190107565597012e-06,
+      "loss": 0.2693,
+      "step": 48445
+    },
+    {
+      "epoch": 3.9246597537265067,
+      "grad_norm": 0.07786433398723602,
+      "learning_rate": 4.1856069130023854e-06,
+      "loss": 0.2375,
+      "step": 48446
+    },
+    {
+      "epoch": 3.9247407647440054,
+      "grad_norm": 0.08209618926048279,
+      "learning_rate": 4.181106260407759e-06,
+      "loss": 0.2616,
+      "step": 48447
+    },
+    {
+      "epoch": 3.9248217757615036,
+      "grad_norm": 0.060023292899131775,
+      "learning_rate": 4.1766056078131335e-06,
+      "loss": 0.2174,
+      "step": 48448
+    },
+    {
+      "epoch": 3.924902786779002,
+      "grad_norm": 0.08991419523954391,
+      "learning_rate": 4.172104955218506e-06,
+      "loss": 0.2533,
+      "step": 48449
+    },
+    {
+      "epoch": 3.9249837977965,
+      "grad_norm": 0.07790505886077881,
+      "learning_rate": 4.167604302623881e-06,
+      "loss": 0.2452,
+      "step": 48450
+    },
+    {
+      "epoch": 3.9250648088139988,
+      "grad_norm": 0.07460498064756393,
+      "learning_rate": 4.163103650029254e-06,
+      "loss": 0.2162,
+      "step": 48451
+    },
+    {
+      "epoch": 3.925145819831497,
+      "grad_norm": 0.07503633946180344,
+      "learning_rate": 4.158602997434628e-06,
+      "loss": 0.1964,
+      "step": 48452
+    },
+    {
+      "epoch": 3.9252268308489953,
+      "grad_norm": 0.06638554483652115,
+      "learning_rate": 4.1541023448400015e-06,
+      "loss": 0.1769,
+      "step": 48453
+    },
+    {
+      "epoch": 3.925307841866494,
+      "grad_norm": 0.0684497132897377,
+      "learning_rate": 4.149601692245376e-06,
+      "loss": 0.1895,
+      "step": 48454
+    },
+    {
+      "epoch": 3.925388852883992,
+      "grad_norm": 0.06736957281827927,
+      "learning_rate": 4.1451010396507496e-06,
+      "loss": 0.2522,
+      "step": 48455
+    },
+    {
+      "epoch": 3.9254698639014904,
+      "grad_norm": 0.07692014425992966,
+      "learning_rate": 4.140600387056123e-06,
+      "loss": 0.2327,
+      "step": 48456
+    },
+    {
+      "epoch": 3.925550874918989,
+      "grad_norm": 0.07177066057920456,
+      "learning_rate": 4.136099734461498e-06,
+      "loss": 0.198,
+      "step": 48457
+    },
+    {
+      "epoch": 3.9256318859364874,
+      "grad_norm": 0.05650921165943146,
+      "learning_rate": 4.131599081866871e-06,
+      "loss": 0.2021,
+      "step": 48458
+    },
+    {
+      "epoch": 3.9257128969539856,
+      "grad_norm": 0.07662155479192734,
+      "learning_rate": 4.127098429272245e-06,
+      "loss": 0.1941,
+      "step": 48459
+    },
+    {
+      "epoch": 3.9257939079714843,
+      "grad_norm": 0.07912242412567139,
+      "learning_rate": 4.1225977766776184e-06,
+      "loss": 0.2419,
+      "step": 48460
+    },
+    {
+      "epoch": 3.9258749189889826,
+      "grad_norm": 0.09148503839969635,
+      "learning_rate": 4.118097124082992e-06,
+      "loss": 0.1989,
+      "step": 48461
+    },
+    {
+      "epoch": 3.925955930006481,
+      "grad_norm": 0.072085440158844,
+      "learning_rate": 4.113596471488366e-06,
+      "loss": 0.2192,
+      "step": 48462
+    },
+    {
+      "epoch": 3.9260369410239795,
+      "grad_norm": 0.0772864818572998,
+      "learning_rate": 4.10909581889374e-06,
+      "loss": 0.2057,
+      "step": 48463
+    },
+    {
+      "epoch": 3.9261179520414777,
+      "grad_norm": 0.07115405797958374,
+      "learning_rate": 4.104595166299114e-06,
+      "loss": 0.2298,
+      "step": 48464
+    },
+    {
+      "epoch": 3.926198963058976,
+      "grad_norm": 0.08247821033000946,
+      "learning_rate": 4.100094513704487e-06,
+      "loss": 0.2068,
+      "step": 48465
+    },
+    {
+      "epoch": 3.9262799740764747,
+      "grad_norm": 0.09296160936355591,
+      "learning_rate": 4.095593861109861e-06,
+      "loss": 0.2661,
+      "step": 48466
+    },
+    {
+      "epoch": 3.926360985093973,
+      "grad_norm": 0.0736137181520462,
+      "learning_rate": 4.091093208515235e-06,
+      "loss": 0.2183,
+      "step": 48467
+    },
+    {
+      "epoch": 3.926441996111471,
+      "grad_norm": 0.07801380753517151,
+      "learning_rate": 4.086592555920608e-06,
+      "loss": 0.2269,
+      "step": 48468
+    },
+    {
+      "epoch": 3.9265230071289694,
+      "grad_norm": 0.07622180134057999,
+      "learning_rate": 4.0820919033259826e-06,
+      "loss": 0.2498,
+      "step": 48469
+    },
+    {
+      "epoch": 3.9266040181464676,
+      "grad_norm": 0.08292534947395325,
+      "learning_rate": 4.077591250731356e-06,
+      "loss": 0.2366,
+      "step": 48470
+    },
+    {
+      "epoch": 3.9266850291639663,
+      "grad_norm": 0.06659669429063797,
+      "learning_rate": 4.07309059813673e-06,
+      "loss": 0.2112,
+      "step": 48471
+    },
+    {
+      "epoch": 3.9267660401814646,
+      "grad_norm": 0.07388713955879211,
+      "learning_rate": 4.068589945542104e-06,
+      "loss": 0.2077,
+      "step": 48472
+    },
+    {
+      "epoch": 3.926847051198963,
+      "grad_norm": 0.09441567212343216,
+      "learning_rate": 4.064089292947478e-06,
+      "loss": 0.2082,
+      "step": 48473
+    },
+    {
+      "epoch": 3.9269280622164615,
+      "grad_norm": 0.06913454085588455,
+      "learning_rate": 4.059588640352851e-06,
+      "loss": 0.2489,
+      "step": 48474
+    },
+    {
+      "epoch": 3.9270090732339598,
+      "grad_norm": 0.0766616016626358,
+      "learning_rate": 4.055087987758225e-06,
+      "loss": 0.2211,
+      "step": 48475
+    },
+    {
+      "epoch": 3.927090084251458,
+      "grad_norm": 0.07652173191308975,
+      "learning_rate": 4.0505873351635995e-06,
+      "loss": 0.1975,
+      "step": 48476
+    },
+    {
+      "epoch": 3.9271710952689567,
+      "grad_norm": 0.06538225710391998,
+      "learning_rate": 4.046086682568972e-06,
+      "loss": 0.2346,
+      "step": 48477
+    },
+    {
+      "epoch": 3.927252106286455,
+      "grad_norm": 0.06540494412183762,
+      "learning_rate": 4.041586029974347e-06,
+      "loss": 0.1988,
+      "step": 48478
+    },
+    {
+      "epoch": 3.927333117303953,
+      "grad_norm": 0.07058582454919815,
+      "learning_rate": 4.03708537737972e-06,
+      "loss": 0.2295,
+      "step": 48479
+    },
+    {
+      "epoch": 3.927414128321452,
+      "grad_norm": 0.06702663749456406,
+      "learning_rate": 4.032584724785094e-06,
+      "loss": 0.2176,
+      "step": 48480
+    },
+    {
+      "epoch": 3.92749513933895,
+      "grad_norm": 0.07759080082178116,
+      "learning_rate": 4.0280840721904675e-06,
+      "loss": 0.2394,
+      "step": 48481
+    },
+    {
+      "epoch": 3.9275761503564484,
+      "grad_norm": 0.07646369934082031,
+      "learning_rate": 4.023583419595842e-06,
+      "loss": 0.2168,
+      "step": 48482
+    },
+    {
+      "epoch": 3.927657161373947,
+      "grad_norm": 0.0533691830933094,
+      "learning_rate": 4.019082767001215e-06,
+      "loss": 0.2078,
+      "step": 48483
+    },
+    {
+      "epoch": 3.9277381723914453,
+      "grad_norm": 0.0704951360821724,
+      "learning_rate": 4.014582114406589e-06,
+      "loss": 0.2282,
+      "step": 48484
+    },
+    {
+      "epoch": 3.9278191834089435,
+      "grad_norm": 0.07538004964590073,
+      "learning_rate": 4.010081461811964e-06,
+      "loss": 0.2543,
+      "step": 48485
+    },
+    {
+      "epoch": 3.9279001944264422,
+      "grad_norm": 0.08612173050642014,
+      "learning_rate": 4.005580809217336e-06,
+      "loss": 0.2632,
+      "step": 48486
+    },
+    {
+      "epoch": 3.9279812054439405,
+      "grad_norm": 0.07190410047769547,
+      "learning_rate": 4.001080156622711e-06,
+      "loss": 0.216,
+      "step": 48487
+    },
+    {
+      "epoch": 3.9280622164614387,
+      "grad_norm": 0.07013415545225143,
+      "learning_rate": 3.996579504028084e-06,
+      "loss": 0.1973,
+      "step": 48488
+    },
+    {
+      "epoch": 3.9281432274789374,
+      "grad_norm": 0.06593007594347,
+      "learning_rate": 3.992078851433458e-06,
+      "loss": 0.2084,
+      "step": 48489
+    },
+    {
+      "epoch": 3.9282242384964356,
+      "grad_norm": 0.07043427973985672,
+      "learning_rate": 3.987578198838832e-06,
+      "loss": 0.2394,
+      "step": 48490
+    },
+    {
+      "epoch": 3.928305249513934,
+      "grad_norm": 0.06437504291534424,
+      "learning_rate": 3.983077546244206e-06,
+      "loss": 0.171,
+      "step": 48491
+    },
+    {
+      "epoch": 3.928386260531432,
+      "grad_norm": 0.07793059200048447,
+      "learning_rate": 3.978576893649579e-06,
+      "loss": 0.1901,
+      "step": 48492
+    },
+    {
+      "epoch": 3.9284672715489304,
+      "grad_norm": 0.0729973241686821,
+      "learning_rate": 3.974076241054953e-06,
+      "loss": 0.1898,
+      "step": 48493
+    },
+    {
+      "epoch": 3.928548282566429,
+      "grad_norm": 0.08081178367137909,
+      "learning_rate": 3.969575588460327e-06,
+      "loss": 0.2452,
+      "step": 48494
+    },
+    {
+      "epoch": 3.9286292935839273,
+      "grad_norm": 0.08382073789834976,
+      "learning_rate": 3.9650749358657005e-06,
+      "loss": 0.2119,
+      "step": 48495
+    },
+    {
+      "epoch": 3.9287103046014256,
+      "grad_norm": 0.07415829598903656,
+      "learning_rate": 3.960574283271074e-06,
+      "loss": 0.1941,
+      "step": 48496
+    },
+    {
+      "epoch": 3.9287913156189243,
+      "grad_norm": 0.06375744938850403,
+      "learning_rate": 3.9560736306764485e-06,
+      "loss": 0.2514,
+      "step": 48497
+    },
+    {
+      "epoch": 3.9288723266364225,
+      "grad_norm": 0.07081378996372223,
+      "learning_rate": 3.951572978081822e-06,
+      "loss": 0.2356,
+      "step": 48498
+    },
+    {
+      "epoch": 3.9289533376539207,
+      "grad_norm": 0.06521176546812057,
+      "learning_rate": 3.947072325487196e-06,
+      "loss": 0.2177,
+      "step": 48499
+    },
+    {
+      "epoch": 3.9290343486714194,
+      "grad_norm": 0.07624541968107224,
+      "learning_rate": 3.94257167289257e-06,
+      "loss": 0.1909,
+      "step": 48500
+    },
+    {
+      "epoch": 3.9291153596889177,
+      "grad_norm": 0.08679436892271042,
+      "learning_rate": 3.938071020297943e-06,
+      "loss": 0.2492,
+      "step": 48501
+    },
+    {
+      "epoch": 3.929196370706416,
+      "grad_norm": 0.06400018185377121,
+      "learning_rate": 3.933570367703317e-06,
+      "loss": 0.2234,
+      "step": 48502
+    },
+    {
+      "epoch": 3.9292773817239146,
+      "grad_norm": 0.07425360381603241,
+      "learning_rate": 3.929069715108691e-06,
+      "loss": 0.2425,
+      "step": 48503
+    },
+    {
+      "epoch": 3.929358392741413,
+      "grad_norm": 0.08528902381658554,
+      "learning_rate": 3.924569062514065e-06,
+      "loss": 0.2177,
+      "step": 48504
+    },
+    {
+      "epoch": 3.929439403758911,
+      "grad_norm": 0.05845775082707405,
+      "learning_rate": 3.920068409919438e-06,
+      "loss": 0.1795,
+      "step": 48505
+    },
+    {
+      "epoch": 3.92952041477641,
+      "grad_norm": 0.06745034456253052,
+      "learning_rate": 3.915567757324813e-06,
+      "loss": 0.2215,
+      "step": 48506
+    },
+    {
+      "epoch": 3.929601425793908,
+      "grad_norm": 0.07435566931962967,
+      "learning_rate": 3.911067104730186e-06,
+      "loss": 0.2253,
+      "step": 48507
+    },
+    {
+      "epoch": 3.9296824368114063,
+      "grad_norm": 0.08194363862276077,
+      "learning_rate": 3.90656645213556e-06,
+      "loss": 0.2381,
+      "step": 48508
+    },
+    {
+      "epoch": 3.929763447828905,
+      "grad_norm": 0.07673655450344086,
+      "learning_rate": 3.9020657995409335e-06,
+      "loss": 0.2508,
+      "step": 48509
+    },
+    {
+      "epoch": 3.929844458846403,
+      "grad_norm": 0.06954783946275711,
+      "learning_rate": 3.897565146946307e-06,
+      "loss": 0.2466,
+      "step": 48510
+    },
+    {
+      "epoch": 3.9299254698639015,
+      "grad_norm": 0.05643027275800705,
+      "learning_rate": 3.893064494351681e-06,
+      "loss": 0.2215,
+      "step": 48511
+    },
+    {
+      "epoch": 3.9300064808814,
+      "grad_norm": 0.06844652444124222,
+      "learning_rate": 3.888563841757055e-06,
+      "loss": 0.2043,
+      "step": 48512
+    },
+    {
+      "epoch": 3.9300874918988984,
+      "grad_norm": 0.07119164615869522,
+      "learning_rate": 3.884063189162429e-06,
+      "loss": 0.2427,
+      "step": 48513
+    },
+    {
+      "epoch": 3.9301685029163966,
+      "grad_norm": 0.07002280652523041,
+      "learning_rate": 3.879562536567802e-06,
+      "loss": 0.2559,
+      "step": 48514
+    },
+    {
+      "epoch": 3.930249513933895,
+      "grad_norm": 0.06633292883634567,
+      "learning_rate": 3.875061883973177e-06,
+      "loss": 0.2434,
+      "step": 48515
+    },
+    {
+      "epoch": 3.930330524951393,
+      "grad_norm": 0.06863868981599808,
+      "learning_rate": 3.87056123137855e-06,
+      "loss": 0.2181,
+      "step": 48516
+    },
+    {
+      "epoch": 3.930411535968892,
+      "grad_norm": 0.0700906440615654,
+      "learning_rate": 3.866060578783924e-06,
+      "loss": 0.2224,
+      "step": 48517
+    },
+    {
+      "epoch": 3.93049254698639,
+      "grad_norm": 0.07740800082683563,
+      "learning_rate": 3.861559926189298e-06,
+      "loss": 0.1978,
+      "step": 48518
+    },
+    {
+      "epoch": 3.9305735580038883,
+      "grad_norm": 0.07479584217071533,
+      "learning_rate": 3.857059273594671e-06,
+      "loss": 0.2067,
+      "step": 48519
+    },
+    {
+      "epoch": 3.930654569021387,
+      "grad_norm": 0.07812254130840302,
+      "learning_rate": 3.852558621000045e-06,
+      "loss": 0.2554,
+      "step": 48520
+    },
+    {
+      "epoch": 3.9307355800388852,
+      "grad_norm": 0.08209620416164398,
+      "learning_rate": 3.848057968405419e-06,
+      "loss": 0.2061,
+      "step": 48521
+    },
+    {
+      "epoch": 3.9308165910563835,
+      "grad_norm": 0.06698629260063171,
+      "learning_rate": 3.843557315810793e-06,
+      "loss": 0.2113,
+      "step": 48522
+    },
+    {
+      "epoch": 3.930897602073882,
+      "grad_norm": 0.0727633461356163,
+      "learning_rate": 3.8390566632161664e-06,
+      "loss": 0.2138,
+      "step": 48523
+    },
+    {
+      "epoch": 3.9309786130913804,
+      "grad_norm": 0.07083755731582642,
+      "learning_rate": 3.83455601062154e-06,
+      "loss": 0.2036,
+      "step": 48524
+    },
+    {
+      "epoch": 3.9310596241088787,
+      "grad_norm": 0.07174476236104965,
+      "learning_rate": 3.8300553580269145e-06,
+      "loss": 0.2266,
+      "step": 48525
+    },
+    {
+      "epoch": 3.9311406351263773,
+      "grad_norm": 0.07710791379213333,
+      "learning_rate": 3.825554705432287e-06,
+      "loss": 0.2302,
+      "step": 48526
+    },
+    {
+      "epoch": 3.9312216461438756,
+      "grad_norm": 0.07124129682779312,
+      "learning_rate": 3.821054052837662e-06,
+      "loss": 0.2003,
+      "step": 48527
+    },
+    {
+      "epoch": 3.931302657161374,
+      "grad_norm": 0.07960180193185806,
+      "learning_rate": 3.816553400243035e-06,
+      "loss": 0.2458,
+      "step": 48528
+    },
+    {
+      "epoch": 3.9313836681788725,
+      "grad_norm": 0.08520202338695526,
+      "learning_rate": 3.8120527476484093e-06,
+      "loss": 0.2287,
+      "step": 48529
+    },
+    {
+      "epoch": 3.9314646791963708,
+      "grad_norm": 0.06994429975748062,
+      "learning_rate": 3.8075520950537834e-06,
+      "loss": 0.2281,
+      "step": 48530
+    },
+    {
+      "epoch": 3.931545690213869,
+      "grad_norm": 0.07213662564754486,
+      "learning_rate": 3.8030514424591565e-06,
+      "loss": 0.1761,
+      "step": 48531
+    },
+    {
+      "epoch": 3.9316267012313677,
+      "grad_norm": 0.07031689584255219,
+      "learning_rate": 3.7985507898645306e-06,
+      "loss": 0.2011,
+      "step": 48532
+    },
+    {
+      "epoch": 3.931707712248866,
+      "grad_norm": 0.0783347338438034,
+      "learning_rate": 3.794050137269904e-06,
+      "loss": 0.2522,
+      "step": 48533
+    },
+    {
+      "epoch": 3.931788723266364,
+      "grad_norm": 0.07658717036247253,
+      "learning_rate": 3.789549484675278e-06,
+      "loss": 0.2342,
+      "step": 48534
+    },
+    {
+      "epoch": 3.931869734283863,
+      "grad_norm": 0.07086281478404999,
+      "learning_rate": 3.785048832080652e-06,
+      "loss": 0.2134,
+      "step": 48535
+    },
+    {
+      "epoch": 3.931950745301361,
+      "grad_norm": 0.06902775913476944,
+      "learning_rate": 3.780548179486026e-06,
+      "loss": 0.1919,
+      "step": 48536
+    },
+    {
+      "epoch": 3.9320317563188594,
+      "grad_norm": 0.06554694473743439,
+      "learning_rate": 3.776047526891399e-06,
+      "loss": 0.2314,
+      "step": 48537
+    },
+    {
+      "epoch": 3.9321127673363576,
+      "grad_norm": 0.08830972760915756,
+      "learning_rate": 3.7715468742967735e-06,
+      "loss": 0.2198,
+      "step": 48538
+    },
+    {
+      "epoch": 3.932193778353856,
+      "grad_norm": 0.07285527884960175,
+      "learning_rate": 3.7670462217021466e-06,
+      "loss": 0.2034,
+      "step": 48539
+    },
+    {
+      "epoch": 3.9322747893713546,
+      "grad_norm": 0.08034533262252808,
+      "learning_rate": 3.7625455691075207e-06,
+      "loss": 0.2121,
+      "step": 48540
+    },
+    {
+      "epoch": 3.932355800388853,
+      "grad_norm": 0.07474691420793533,
+      "learning_rate": 3.7580449165128947e-06,
+      "loss": 0.2456,
+      "step": 48541
+    },
+    {
+      "epoch": 3.932436811406351,
+      "grad_norm": 0.06668350100517273,
+      "learning_rate": 3.7535442639182683e-06,
+      "loss": 0.239,
+      "step": 48542
+    },
+    {
+      "epoch": 3.9325178224238497,
+      "grad_norm": 0.06283923983573914,
+      "learning_rate": 3.7490436113236423e-06,
+      "loss": 0.1962,
+      "step": 48543
+    },
+    {
+      "epoch": 3.932598833441348,
+      "grad_norm": 0.08116850256919861,
+      "learning_rate": 3.744542958729016e-06,
+      "loss": 0.256,
+      "step": 48544
+    },
+    {
+      "epoch": 3.932679844458846,
+      "grad_norm": 0.07628383487462997,
+      "learning_rate": 3.74004230613439e-06,
+      "loss": 0.2212,
+      "step": 48545
+    },
+    {
+      "epoch": 3.932760855476345,
+      "grad_norm": 0.07699400186538696,
+      "learning_rate": 3.735541653539763e-06,
+      "loss": 0.194,
+      "step": 48546
+    },
+    {
+      "epoch": 3.932841866493843,
+      "grad_norm": 0.08026984333992004,
+      "learning_rate": 3.7310410009451376e-06,
+      "loss": 0.1822,
+      "step": 48547
+    },
+    {
+      "epoch": 3.9329228775113414,
+      "grad_norm": 0.0703667625784874,
+      "learning_rate": 3.7265403483505108e-06,
+      "loss": 0.2552,
+      "step": 48548
+    },
+    {
+      "epoch": 3.93300388852884,
+      "grad_norm": 0.06922667473554611,
+      "learning_rate": 3.722039695755885e-06,
+      "loss": 0.2323,
+      "step": 48549
+    },
+    {
+      "epoch": 3.9330848995463383,
+      "grad_norm": 0.07332278788089752,
+      "learning_rate": 3.7175390431612584e-06,
+      "loss": 0.2274,
+      "step": 48550
+    },
+    {
+      "epoch": 3.9331659105638366,
+      "grad_norm": 0.07613363862037659,
+      "learning_rate": 3.7130383905666324e-06,
+      "loss": 0.2087,
+      "step": 48551
+    },
+    {
+      "epoch": 3.9332469215813353,
+      "grad_norm": 0.09742595255374908,
+      "learning_rate": 3.708537737972006e-06,
+      "loss": 0.2368,
+      "step": 48552
+    },
+    {
+      "epoch": 3.9333279325988335,
+      "grad_norm": 0.0734672024846077,
+      "learning_rate": 3.70403708537738e-06,
+      "loss": 0.2262,
+      "step": 48553
+    },
+    {
+      "epoch": 3.9334089436163318,
+      "grad_norm": 0.06810317933559418,
+      "learning_rate": 3.6995364327827532e-06,
+      "loss": 0.2113,
+      "step": 48554
+    },
+    {
+      "epoch": 3.9334899546338304,
+      "grad_norm": 0.07320161908864975,
+      "learning_rate": 3.6950357801881273e-06,
+      "loss": 0.2495,
+      "step": 48555
+    },
+    {
+      "epoch": 3.9335709656513287,
+      "grad_norm": 0.07830044627189636,
+      "learning_rate": 3.6905351275935017e-06,
+      "loss": 0.1991,
+      "step": 48556
+    },
+    {
+      "epoch": 3.933651976668827,
+      "grad_norm": 0.06398749351501465,
+      "learning_rate": 3.686034474998875e-06,
+      "loss": 0.2219,
+      "step": 48557
+    },
+    {
+      "epoch": 3.933732987686325,
+      "grad_norm": 0.07395961135625839,
+      "learning_rate": 3.681533822404249e-06,
+      "loss": 0.211,
+      "step": 48558
+    },
+    {
+      "epoch": 3.933813998703824,
+      "grad_norm": 0.0828605592250824,
+      "learning_rate": 3.6770331698096225e-06,
+      "loss": 0.2104,
+      "step": 48559
+    },
+    {
+      "epoch": 3.933895009721322,
+      "grad_norm": 0.06945844739675522,
+      "learning_rate": 3.6725325172149965e-06,
+      "loss": 0.2022,
+      "step": 48560
+    },
+    {
+      "epoch": 3.9339760207388204,
+      "grad_norm": 0.08338592201471329,
+      "learning_rate": 3.66803186462037e-06,
+      "loss": 0.2406,
+      "step": 48561
+    },
+    {
+      "epoch": 3.9340570317563186,
+      "grad_norm": 0.07173708081245422,
+      "learning_rate": 3.663531212025744e-06,
+      "loss": 0.2097,
+      "step": 48562
+    },
+    {
+      "epoch": 3.9341380427738173,
+      "grad_norm": 0.06706194579601288,
+      "learning_rate": 3.6590305594311174e-06,
+      "loss": 0.2191,
+      "step": 48563
+    },
+    {
+      "epoch": 3.9342190537913155,
+      "grad_norm": 0.06909432262182236,
+      "learning_rate": 3.654529906836492e-06,
+      "loss": 0.226,
+      "step": 48564
+    },
+    {
+      "epoch": 3.934300064808814,
+      "grad_norm": 0.07070121169090271,
+      "learning_rate": 3.650029254241865e-06,
+      "loss": 0.2187,
+      "step": 48565
+    },
+    {
+      "epoch": 3.9343810758263125,
+      "grad_norm": 0.07241905480623245,
+      "learning_rate": 3.645528601647239e-06,
+      "loss": 0.2526,
+      "step": 48566
+    },
+    {
+      "epoch": 3.9344620868438107,
+      "grad_norm": 0.059708014130592346,
+      "learning_rate": 3.6410279490526126e-06,
+      "loss": 0.2184,
+      "step": 48567
+    },
+    {
+      "epoch": 3.934543097861309,
+      "grad_norm": 0.08609946072101593,
+      "learning_rate": 3.6365272964579866e-06,
+      "loss": 0.2065,
+      "step": 48568
+    },
+    {
+      "epoch": 3.9346241088788076,
+      "grad_norm": 0.07386572659015656,
+      "learning_rate": 3.63202664386336e-06,
+      "loss": 0.2091,
+      "step": 48569
+    },
+    {
+      "epoch": 3.934705119896306,
+      "grad_norm": 0.06636470556259155,
+      "learning_rate": 3.6275259912687343e-06,
+      "loss": 0.2014,
+      "step": 48570
+    },
+    {
+      "epoch": 3.934786130913804,
+      "grad_norm": 0.07500702887773514,
+      "learning_rate": 3.6230253386741083e-06,
+      "loss": 0.2051,
+      "step": 48571
+    },
+    {
+      "epoch": 3.934867141931303,
+      "grad_norm": 0.06589839607477188,
+      "learning_rate": 3.6185246860794815e-06,
+      "loss": 0.2176,
+      "step": 48572
+    },
+    {
+      "epoch": 3.934948152948801,
+      "grad_norm": 0.06833402067422867,
+      "learning_rate": 3.614024033484856e-06,
+      "loss": 0.2436,
+      "step": 48573
+    },
+    {
+      "epoch": 3.9350291639662993,
+      "grad_norm": 0.07177960872650146,
+      "learning_rate": 3.609523380890229e-06,
+      "loss": 0.263,
+      "step": 48574
+    },
+    {
+      "epoch": 3.935110174983798,
+      "grad_norm": 0.06268391013145447,
+      "learning_rate": 3.605022728295603e-06,
+      "loss": 0.2019,
+      "step": 48575
+    },
+    {
+      "epoch": 3.9351911860012962,
+      "grad_norm": 0.0845642238855362,
+      "learning_rate": 3.6005220757009767e-06,
+      "loss": 0.2184,
+      "step": 48576
+    },
+    {
+      "epoch": 3.9352721970187945,
+      "grad_norm": 0.05564136430621147,
+      "learning_rate": 3.5960214231063508e-06,
+      "loss": 0.2024,
+      "step": 48577
+    },
+    {
+      "epoch": 3.935353208036293,
+      "grad_norm": 0.06291015446186066,
+      "learning_rate": 3.5915207705117244e-06,
+      "loss": 0.2304,
+      "step": 48578
+    },
+    {
+      "epoch": 3.9354342190537914,
+      "grad_norm": 0.07351517677307129,
+      "learning_rate": 3.5870201179170984e-06,
+      "loss": 0.234,
+      "step": 48579
+    },
+    {
+      "epoch": 3.9355152300712897,
+      "grad_norm": 0.07256698608398438,
+      "learning_rate": 3.5825194653224716e-06,
+      "loss": 0.2334,
+      "step": 48580
+    },
+    {
+      "epoch": 3.935596241088788,
+      "grad_norm": 0.08631699532270432,
+      "learning_rate": 3.5780188127278456e-06,
+      "loss": 0.2292,
+      "step": 48581
+    },
+    {
+      "epoch": 3.9356772521062866,
+      "grad_norm": 0.06245209649205208,
+      "learning_rate": 3.573518160133219e-06,
+      "loss": 0.2052,
+      "step": 48582
+    },
+    {
+      "epoch": 3.935758263123785,
+      "grad_norm": 0.06966308504343033,
+      "learning_rate": 3.5690175075385932e-06,
+      "loss": 0.2312,
+      "step": 48583
+    },
+    {
+      "epoch": 3.935839274141283,
+      "grad_norm": 0.057079412043094635,
+      "learning_rate": 3.564516854943967e-06,
+      "loss": 0.2015,
+      "step": 48584
+    },
+    {
+      "epoch": 3.9359202851587813,
+      "grad_norm": 0.0594225637614727,
+      "learning_rate": 3.560016202349341e-06,
+      "loss": 0.2156,
+      "step": 48585
+    },
+    {
+      "epoch": 3.93600129617628,
+      "grad_norm": 0.06722415238618851,
+      "learning_rate": 3.555515549754715e-06,
+      "loss": 0.1988,
+      "step": 48586
+    },
+    {
+      "epoch": 3.9360823071937783,
+      "grad_norm": 0.06993760913610458,
+      "learning_rate": 3.5510148971600885e-06,
+      "loss": 0.2112,
+      "step": 48587
+    },
+    {
+      "epoch": 3.9361633182112765,
+      "grad_norm": 0.08021251857280731,
+      "learning_rate": 3.5465142445654625e-06,
+      "loss": 0.2323,
+      "step": 48588
+    },
+    {
+      "epoch": 3.936244329228775,
+      "grad_norm": 0.07788848876953125,
+      "learning_rate": 3.5420135919708357e-06,
+      "loss": 0.2113,
+      "step": 48589
+    },
+    {
+      "epoch": 3.9363253402462735,
+      "grad_norm": 0.06252850592136383,
+      "learning_rate": 3.5375129393762097e-06,
+      "loss": 0.214,
+      "step": 48590
+    },
+    {
+      "epoch": 3.9364063512637717,
+      "grad_norm": 0.0875447690486908,
+      "learning_rate": 3.5330122867815833e-06,
+      "loss": 0.2773,
+      "step": 48591
+    },
+    {
+      "epoch": 3.9364873622812704,
+      "grad_norm": 0.067319855093956,
+      "learning_rate": 3.5285116341869574e-06,
+      "loss": 0.2524,
+      "step": 48592
+    },
+    {
+      "epoch": 3.9365683732987686,
+      "grad_norm": 0.06517969071865082,
+      "learning_rate": 3.524010981592331e-06,
+      "loss": 0.2369,
+      "step": 48593
+    },
+    {
+      "epoch": 3.936649384316267,
+      "grad_norm": 0.06585026532411575,
+      "learning_rate": 3.519510328997705e-06,
+      "loss": 0.2006,
+      "step": 48594
+    },
+    {
+      "epoch": 3.9367303953337656,
+      "grad_norm": 0.0712442547082901,
+      "learning_rate": 3.515009676403078e-06,
+      "loss": 0.1914,
+      "step": 48595
+    },
+    {
+      "epoch": 3.936811406351264,
+      "grad_norm": 0.07155847549438477,
+      "learning_rate": 3.5105090238084526e-06,
+      "loss": 0.2596,
+      "step": 48596
+    },
+    {
+      "epoch": 3.936892417368762,
+      "grad_norm": 0.0668448656797409,
+      "learning_rate": 3.506008371213826e-06,
+      "loss": 0.2167,
+      "step": 48597
+    },
+    {
+      "epoch": 3.9369734283862607,
+      "grad_norm": 0.06561889499425888,
+      "learning_rate": 3.5015077186192e-06,
+      "loss": 0.1859,
+      "step": 48598
+    },
+    {
+      "epoch": 3.937054439403759,
+      "grad_norm": 0.05717025324702263,
+      "learning_rate": 3.4970070660245743e-06,
+      "loss": 0.2039,
+      "step": 48599
+    },
+    {
+      "epoch": 3.9371354504212572,
+      "grad_norm": 0.0749245211482048,
+      "learning_rate": 3.4925064134299475e-06,
+      "loss": 0.2545,
+      "step": 48600
+    },
+    {
+      "epoch": 3.937216461438756,
+      "grad_norm": 0.0730648785829544,
+      "learning_rate": 3.4880057608353215e-06,
+      "loss": 0.2204,
+      "step": 48601
+    },
+    {
+      "epoch": 3.937297472456254,
+      "grad_norm": 0.06548863649368286,
+      "learning_rate": 3.483505108240695e-06,
+      "loss": 0.2004,
+      "step": 48602
+    },
+    {
+      "epoch": 3.9373784834737524,
+      "grad_norm": 0.06222781166434288,
+      "learning_rate": 3.479004455646069e-06,
+      "loss": 0.1914,
+      "step": 48603
+    },
+    {
+      "epoch": 3.9374594944912507,
+      "grad_norm": 0.061245959252119064,
+      "learning_rate": 3.4745038030514423e-06,
+      "loss": 0.2335,
+      "step": 48604
+    },
+    {
+      "epoch": 3.9375405055087493,
+      "grad_norm": 0.09785652905702591,
+      "learning_rate": 3.4700031504568167e-06,
+      "loss": 0.2717,
+      "step": 48605
+    },
+    {
+      "epoch": 3.9376215165262476,
+      "grad_norm": 0.07860812544822693,
+      "learning_rate": 3.46550249786219e-06,
+      "loss": 0.2264,
+      "step": 48606
+    },
+    {
+      "epoch": 3.937702527543746,
+      "grad_norm": 0.08016818761825562,
+      "learning_rate": 3.461001845267564e-06,
+      "loss": 0.2651,
+      "step": 48607
+    },
+    {
+      "epoch": 3.937783538561244,
+      "grad_norm": 0.07426834851503372,
+      "learning_rate": 3.4565011926729376e-06,
+      "loss": 0.2745,
+      "step": 48608
+    },
+    {
+      "epoch": 3.9378645495787428,
+      "grad_norm": 0.0777416005730629,
+      "learning_rate": 3.4520005400783116e-06,
+      "loss": 0.2511,
+      "step": 48609
+    },
+    {
+      "epoch": 3.937945560596241,
+      "grad_norm": 0.08751657605171204,
+      "learning_rate": 3.447499887483685e-06,
+      "loss": 0.2446,
+      "step": 48610
+    },
+    {
+      "epoch": 3.9380265716137393,
+      "grad_norm": 0.08031098544597626,
+      "learning_rate": 3.442999234889059e-06,
+      "loss": 0.2586,
+      "step": 48611
+    },
+    {
+      "epoch": 3.938107582631238,
+      "grad_norm": 0.0816061943769455,
+      "learning_rate": 3.4384985822944324e-06,
+      "loss": 0.2311,
+      "step": 48612
+    },
+    {
+      "epoch": 3.938188593648736,
+      "grad_norm": 0.07019905745983124,
+      "learning_rate": 3.433997929699807e-06,
+      "loss": 0.2114,
+      "step": 48613
+    },
+    {
+      "epoch": 3.9382696046662344,
+      "grad_norm": 0.061768822371959686,
+      "learning_rate": 3.429497277105181e-06,
+      "loss": 0.2184,
+      "step": 48614
+    },
+    {
+      "epoch": 3.938350615683733,
+      "grad_norm": 0.07346883416175842,
+      "learning_rate": 3.424996624510554e-06,
+      "loss": 0.2302,
+      "step": 48615
+    },
+    {
+      "epoch": 3.9384316267012314,
+      "grad_norm": 0.078464575111866,
+      "learning_rate": 3.420495971915928e-06,
+      "loss": 0.1974,
+      "step": 48616
+    },
+    {
+      "epoch": 3.9385126377187296,
+      "grad_norm": 0.08362342417240143,
+      "learning_rate": 3.4159953193213017e-06,
+      "loss": 0.2104,
+      "step": 48617
+    },
+    {
+      "epoch": 3.9385936487362283,
+      "grad_norm": 0.07425287365913391,
+      "learning_rate": 3.4114946667266757e-06,
+      "loss": 0.1987,
+      "step": 48618
+    },
+    {
+      "epoch": 3.9386746597537265,
+      "grad_norm": 0.05362372472882271,
+      "learning_rate": 3.4069940141320493e-06,
+      "loss": 0.1655,
+      "step": 48619
+    },
+    {
+      "epoch": 3.938755670771225,
+      "grad_norm": 0.0676586851477623,
+      "learning_rate": 3.4024933615374233e-06,
+      "loss": 0.1871,
+      "step": 48620
+    },
+    {
+      "epoch": 3.9388366817887235,
+      "grad_norm": 0.06317394226789474,
+      "learning_rate": 3.3979927089427965e-06,
+      "loss": 0.2735,
+      "step": 48621
+    },
+    {
+      "epoch": 3.9389176928062217,
+      "grad_norm": 0.06809334456920624,
+      "learning_rate": 3.393492056348171e-06,
+      "loss": 0.2171,
+      "step": 48622
+    },
+    {
+      "epoch": 3.93899870382372,
+      "grad_norm": 0.0660707876086235,
+      "learning_rate": 3.388991403753544e-06,
+      "loss": 0.2299,
+      "step": 48623
+    },
+    {
+      "epoch": 3.9390797148412187,
+      "grad_norm": 0.08089926093816757,
+      "learning_rate": 3.384490751158918e-06,
+      "loss": 0.2247,
+      "step": 48624
+    },
+    {
+      "epoch": 3.939160725858717,
+      "grad_norm": 0.06330505758523941,
+      "learning_rate": 3.3799900985642918e-06,
+      "loss": 0.2135,
+      "step": 48625
+    },
+    {
+      "epoch": 3.939241736876215,
+      "grad_norm": 0.06716379523277283,
+      "learning_rate": 3.375489445969666e-06,
+      "loss": 0.232,
+      "step": 48626
+    },
+    {
+      "epoch": 3.9393227478937134,
+      "grad_norm": 0.06354763358831406,
+      "learning_rate": 3.3709887933750394e-06,
+      "loss": 0.2079,
+      "step": 48627
+    },
+    {
+      "epoch": 3.939403758911212,
+      "grad_norm": 0.07505752146244049,
+      "learning_rate": 3.3664881407804134e-06,
+      "loss": 0.2218,
+      "step": 48628
+    },
+    {
+      "epoch": 3.9394847699287103,
+      "grad_norm": 0.08010073751211166,
+      "learning_rate": 3.3619874881857875e-06,
+      "loss": 0.2033,
+      "step": 48629
+    },
+    {
+      "epoch": 3.9395657809462086,
+      "grad_norm": 0.07689400762319565,
+      "learning_rate": 3.3574868355911606e-06,
+      "loss": 0.2303,
+      "step": 48630
+    },
+    {
+      "epoch": 3.939646791963707,
+      "grad_norm": 0.0748937800526619,
+      "learning_rate": 3.352986182996535e-06,
+      "loss": 0.2066,
+      "step": 48631
+    },
+    {
+      "epoch": 3.9397278029812055,
+      "grad_norm": 0.08342572301626205,
+      "learning_rate": 3.3484855304019083e-06,
+      "loss": 0.2457,
+      "step": 48632
+    },
+    {
+      "epoch": 3.9398088139987038,
+      "grad_norm": 0.08893360197544098,
+      "learning_rate": 3.3439848778072823e-06,
+      "loss": 0.2338,
+      "step": 48633
+    },
+    {
+      "epoch": 3.939889825016202,
+      "grad_norm": 0.06578105688095093,
+      "learning_rate": 3.339484225212656e-06,
+      "loss": 0.2092,
+      "step": 48634
+    },
+    {
+      "epoch": 3.9399708360337007,
+      "grad_norm": 0.06503915041685104,
+      "learning_rate": 3.33498357261803e-06,
+      "loss": 0.2476,
+      "step": 48635
+    },
+    {
+      "epoch": 3.940051847051199,
+      "grad_norm": 0.0719245970249176,
+      "learning_rate": 3.3304829200234035e-06,
+      "loss": 0.2677,
+      "step": 48636
+    },
+    {
+      "epoch": 3.940132858068697,
+      "grad_norm": 0.06504833698272705,
+      "learning_rate": 3.3259822674287776e-06,
+      "loss": 0.2252,
+      "step": 48637
+    },
+    {
+      "epoch": 3.940213869086196,
+      "grad_norm": 0.08234488219022751,
+      "learning_rate": 3.3214816148341507e-06,
+      "loss": 0.2319,
+      "step": 48638
+    },
+    {
+      "epoch": 3.940294880103694,
+      "grad_norm": 0.06820017844438553,
+      "learning_rate": 3.3169809622395248e-06,
+      "loss": 0.1929,
+      "step": 48639
+    },
+    {
+      "epoch": 3.9403758911211924,
+      "grad_norm": 0.05836334079504013,
+      "learning_rate": 3.3124803096448984e-06,
+      "loss": 0.2172,
+      "step": 48640
+    },
+    {
+      "epoch": 3.940456902138691,
+      "grad_norm": 0.06549455970525742,
+      "learning_rate": 3.3079796570502724e-06,
+      "loss": 0.2515,
+      "step": 48641
+    },
+    {
+      "epoch": 3.9405379131561893,
+      "grad_norm": 0.07402468472719193,
+      "learning_rate": 3.3034790044556464e-06,
+      "loss": 0.2103,
+      "step": 48642
+    },
+    {
+      "epoch": 3.9406189241736875,
+      "grad_norm": 0.06810709089040756,
+      "learning_rate": 3.29897835186102e-06,
+      "loss": 0.2017,
+      "step": 48643
+    },
+    {
+      "epoch": 3.940699935191186,
+      "grad_norm": 0.06563235074281693,
+      "learning_rate": 3.294477699266394e-06,
+      "loss": 0.2042,
+      "step": 48644
+    },
+    {
+      "epoch": 3.9407809462086845,
+      "grad_norm": 0.08226774632930756,
+      "learning_rate": 3.2899770466717677e-06,
+      "loss": 0.2507,
+      "step": 48645
+    },
+    {
+      "epoch": 3.9408619572261827,
+      "grad_norm": 0.06738846004009247,
+      "learning_rate": 3.2854763940771417e-06,
+      "loss": 0.2144,
+      "step": 48646
+    },
+    {
+      "epoch": 3.9409429682436814,
+      "grad_norm": 0.06427872180938721,
+      "learning_rate": 3.280975741482515e-06,
+      "loss": 0.2197,
+      "step": 48647
+    },
+    {
+      "epoch": 3.9410239792611796,
+      "grad_norm": 0.07773533463478088,
+      "learning_rate": 3.276475088887889e-06,
+      "loss": 0.2123,
+      "step": 48648
+    },
+    {
+      "epoch": 3.941104990278678,
+      "grad_norm": 0.08229377865791321,
+      "learning_rate": 3.2719744362932625e-06,
+      "loss": 0.2247,
+      "step": 48649
+    },
+    {
+      "epoch": 3.941186001296176,
+      "grad_norm": 0.0808233693242073,
+      "learning_rate": 3.2674737836986365e-06,
+      "loss": 0.2419,
+      "step": 48650
+    },
+    {
+      "epoch": 3.941267012313675,
+      "grad_norm": 0.08312023431062698,
+      "learning_rate": 3.26297313110401e-06,
+      "loss": 0.2376,
+      "step": 48651
+    },
+    {
+      "epoch": 3.941348023331173,
+      "grad_norm": 0.0641668364405632,
+      "learning_rate": 3.258472478509384e-06,
+      "loss": 0.2425,
+      "step": 48652
+    },
+    {
+      "epoch": 3.9414290343486713,
+      "grad_norm": 0.06582210958003998,
+      "learning_rate": 3.2539718259147573e-06,
+      "loss": 0.2453,
+      "step": 48653
+    },
+    {
+      "epoch": 3.9415100453661696,
+      "grad_norm": 0.07841324061155319,
+      "learning_rate": 3.2494711733201318e-06,
+      "loss": 0.2307,
+      "step": 48654
+    },
+    {
+      "epoch": 3.9415910563836682,
+      "grad_norm": 0.0673704668879509,
+      "learning_rate": 3.244970520725505e-06,
+      "loss": 0.2163,
+      "step": 48655
+    },
+    {
+      "epoch": 3.9416720674011665,
+      "grad_norm": 0.0752340778708458,
+      "learning_rate": 3.240469868130879e-06,
+      "loss": 0.2501,
+      "step": 48656
+    },
+    {
+      "epoch": 3.9417530784186647,
+      "grad_norm": 0.07765194028615952,
+      "learning_rate": 3.2359692155362534e-06,
+      "loss": 0.2192,
+      "step": 48657
+    },
+    {
+      "epoch": 3.9418340894361634,
+      "grad_norm": 0.06525509059429169,
+      "learning_rate": 3.2314685629416266e-06,
+      "loss": 0.223,
+      "step": 48658
+    },
+    {
+      "epoch": 3.9419151004536617,
+      "grad_norm": 0.06445495784282684,
+      "learning_rate": 3.2269679103470006e-06,
+      "loss": 0.2472,
+      "step": 48659
+    },
+    {
+      "epoch": 3.94199611147116,
+      "grad_norm": 0.06818129867315292,
+      "learning_rate": 3.2224672577523742e-06,
+      "loss": 0.2138,
+      "step": 48660
+    },
+    {
+      "epoch": 3.9420771224886586,
+      "grad_norm": 0.07735411822795868,
+      "learning_rate": 3.2179666051577483e-06,
+      "loss": 0.2252,
+      "step": 48661
+    },
+    {
+      "epoch": 3.942158133506157,
+      "grad_norm": 0.07455040514469147,
+      "learning_rate": 3.213465952563122e-06,
+      "loss": 0.2383,
+      "step": 48662
+    },
+    {
+      "epoch": 3.942239144523655,
+      "grad_norm": 0.07096820324659348,
+      "learning_rate": 3.208965299968496e-06,
+      "loss": 0.2368,
+      "step": 48663
+    },
+    {
+      "epoch": 3.942320155541154,
+      "grad_norm": 0.06698424369096756,
+      "learning_rate": 3.204464647373869e-06,
+      "loss": 0.2426,
+      "step": 48664
+    },
+    {
+      "epoch": 3.942401166558652,
+      "grad_norm": 0.05881981924176216,
+      "learning_rate": 3.199963994779243e-06,
+      "loss": 0.1789,
+      "step": 48665
+    },
+    {
+      "epoch": 3.9424821775761503,
+      "grad_norm": 0.07194176316261292,
+      "learning_rate": 3.1954633421846167e-06,
+      "loss": 0.2194,
+      "step": 48666
+    },
+    {
+      "epoch": 3.942563188593649,
+      "grad_norm": 0.06709557771682739,
+      "learning_rate": 3.1909626895899907e-06,
+      "loss": 0.2023,
+      "step": 48667
+    },
+    {
+      "epoch": 3.942644199611147,
+      "grad_norm": 0.08858643472194672,
+      "learning_rate": 3.1864620369953643e-06,
+      "loss": 0.2341,
+      "step": 48668
+    },
+    {
+      "epoch": 3.9427252106286454,
+      "grad_norm": 0.08386151492595673,
+      "learning_rate": 3.1819613844007384e-06,
+      "loss": 0.2279,
+      "step": 48669
+    },
+    {
+      "epoch": 3.942806221646144,
+      "grad_norm": 0.07175328582525253,
+      "learning_rate": 3.1774607318061115e-06,
+      "loss": 0.2233,
+      "step": 48670
+    },
+    {
+      "epoch": 3.9428872326636424,
+      "grad_norm": 0.0729939341545105,
+      "learning_rate": 3.172960079211486e-06,
+      "loss": 0.2228,
+      "step": 48671
+    },
+    {
+      "epoch": 3.9429682436811406,
+      "grad_norm": 0.07817017287015915,
+      "learning_rate": 3.16845942661686e-06,
+      "loss": 0.23,
+      "step": 48672
+    },
+    {
+      "epoch": 3.943049254698639,
+      "grad_norm": 0.06868911534547806,
+      "learning_rate": 3.163958774022233e-06,
+      "loss": 0.2288,
+      "step": 48673
+    },
+    {
+      "epoch": 3.943130265716137,
+      "grad_norm": 0.07988391816616058,
+      "learning_rate": 3.1594581214276072e-06,
+      "loss": 0.2186,
+      "step": 48674
+    },
+    {
+      "epoch": 3.943211276733636,
+      "grad_norm": 0.062225591391325,
+      "learning_rate": 3.154957468832981e-06,
+      "loss": 0.2169,
+      "step": 48675
+    },
+    {
+      "epoch": 3.943292287751134,
+      "grad_norm": 0.07657849788665771,
+      "learning_rate": 3.150456816238355e-06,
+      "loss": 0.2065,
+      "step": 48676
+    },
+    {
+      "epoch": 3.9433732987686323,
+      "grad_norm": 0.09984945505857468,
+      "learning_rate": 3.1459561636437285e-06,
+      "loss": 0.2521,
+      "step": 48677
+    },
+    {
+      "epoch": 3.943454309786131,
+      "grad_norm": 0.08694443851709366,
+      "learning_rate": 3.1414555110491025e-06,
+      "loss": 0.2002,
+      "step": 48678
+    },
+    {
+      "epoch": 3.9435353208036292,
+      "grad_norm": 0.06631725281476974,
+      "learning_rate": 3.1369548584544757e-06,
+      "loss": 0.1917,
+      "step": 48679
+    },
+    {
+      "epoch": 3.9436163318211275,
+      "grad_norm": 0.06001517176628113,
+      "learning_rate": 3.13245420585985e-06,
+      "loss": 0.203,
+      "step": 48680
+    },
+    {
+      "epoch": 3.943697342838626,
+      "grad_norm": 0.060536958277225494,
+      "learning_rate": 3.1279535532652233e-06,
+      "loss": 0.2354,
+      "step": 48681
+    },
+    {
+      "epoch": 3.9437783538561244,
+      "grad_norm": 0.07151622325181961,
+      "learning_rate": 3.1234529006705973e-06,
+      "loss": 0.2221,
+      "step": 48682
+    },
+    {
+      "epoch": 3.9438593648736227,
+      "grad_norm": 0.07806843519210815,
+      "learning_rate": 3.1189522480759714e-06,
+      "loss": 0.1761,
+      "step": 48683
+    },
+    {
+      "epoch": 3.9439403758911213,
+      "grad_norm": 0.061488077044487,
+      "learning_rate": 3.114451595481345e-06,
+      "loss": 0.1776,
+      "step": 48684
+    },
+    {
+      "epoch": 3.9440213869086196,
+      "grad_norm": 0.08169107884168625,
+      "learning_rate": 3.1099509428867186e-06,
+      "loss": 0.2266,
+      "step": 48685
+    },
+    {
+      "epoch": 3.944102397926118,
+      "grad_norm": 0.0697530210018158,
+      "learning_rate": 3.1054502902920926e-06,
+      "loss": 0.1994,
+      "step": 48686
+    },
+    {
+      "epoch": 3.9441834089436165,
+      "grad_norm": 0.06424905359745026,
+      "learning_rate": 3.100949637697466e-06,
+      "loss": 0.2624,
+      "step": 48687
+    },
+    {
+      "epoch": 3.9442644199611148,
+      "grad_norm": 0.0686437338590622,
+      "learning_rate": 3.09644898510284e-06,
+      "loss": 0.199,
+      "step": 48688
+    },
+    {
+      "epoch": 3.944345430978613,
+      "grad_norm": 0.07491966336965561,
+      "learning_rate": 3.091948332508214e-06,
+      "loss": 0.2548,
+      "step": 48689
+    },
+    {
+      "epoch": 3.9444264419961117,
+      "grad_norm": 0.08908192068338394,
+      "learning_rate": 3.0874476799135874e-06,
+      "loss": 0.2572,
+      "step": 48690
+    },
+    {
+      "epoch": 3.94450745301361,
+      "grad_norm": 0.07038780301809311,
+      "learning_rate": 3.0829470273189614e-06,
+      "loss": 0.2381,
+      "step": 48691
+    },
+    {
+      "epoch": 3.944588464031108,
+      "grad_norm": 0.06175041198730469,
+      "learning_rate": 3.078446374724335e-06,
+      "loss": 0.251,
+      "step": 48692
+    },
+    {
+      "epoch": 3.944669475048607,
+      "grad_norm": 0.07298500090837479,
+      "learning_rate": 3.073945722129709e-06,
+      "loss": 0.2434,
+      "step": 48693
+    },
+    {
+      "epoch": 3.944750486066105,
+      "grad_norm": 0.07891558855772018,
+      "learning_rate": 3.0694450695350827e-06,
+      "loss": 0.246,
+      "step": 48694
+    },
+    {
+      "epoch": 3.9448314970836034,
+      "grad_norm": 0.07210226356983185,
+      "learning_rate": 3.0649444169404567e-06,
+      "loss": 0.2258,
+      "step": 48695
+    },
+    {
+      "epoch": 3.9449125081011016,
+      "grad_norm": 0.07565312087535858,
+      "learning_rate": 3.0604437643458303e-06,
+      "loss": 0.2653,
+      "step": 48696
+    },
+    {
+      "epoch": 3.9449935191186,
+      "grad_norm": 0.06832912564277649,
+      "learning_rate": 3.055943111751204e-06,
+      "loss": 0.1948,
+      "step": 48697
+    },
+    {
+      "epoch": 3.9450745301360985,
+      "grad_norm": 0.07523760199546814,
+      "learning_rate": 3.051442459156578e-06,
+      "loss": 0.2343,
+      "step": 48698
+    },
+    {
+      "epoch": 3.945155541153597,
+      "grad_norm": 0.0894690528512001,
+      "learning_rate": 3.0469418065619515e-06,
+      "loss": 0.2323,
+      "step": 48699
+    },
+    {
+      "epoch": 3.945236552171095,
+      "grad_norm": 0.06247549131512642,
+      "learning_rate": 3.0424411539673256e-06,
+      "loss": 0.2019,
+      "step": 48700
+    },
+    {
+      "epoch": 3.9453175631885937,
+      "grad_norm": 0.06587101519107819,
+      "learning_rate": 3.037940501372699e-06,
+      "loss": 0.2405,
+      "step": 48701
+    },
+    {
+      "epoch": 3.945398574206092,
+      "grad_norm": 0.07532519102096558,
+      "learning_rate": 3.0334398487780728e-06,
+      "loss": 0.1945,
+      "step": 48702
+    },
+    {
+      "epoch": 3.94547958522359,
+      "grad_norm": 0.06462553888559341,
+      "learning_rate": 3.028939196183447e-06,
+      "loss": 0.2456,
+      "step": 48703
+    },
+    {
+      "epoch": 3.945560596241089,
+      "grad_norm": 0.05730045959353447,
+      "learning_rate": 3.0244385435888204e-06,
+      "loss": 0.1876,
+      "step": 48704
+    },
+    {
+      "epoch": 3.945641607258587,
+      "grad_norm": 0.08065466582775116,
+      "learning_rate": 3.019937890994194e-06,
+      "loss": 0.2045,
+      "step": 48705
+    },
+    {
+      "epoch": 3.9457226182760854,
+      "grad_norm": 0.08913881331682205,
+      "learning_rate": 3.015437238399568e-06,
+      "loss": 0.2335,
+      "step": 48706
+    },
+    {
+      "epoch": 3.945803629293584,
+      "grad_norm": 0.0686987042427063,
+      "learning_rate": 3.010936585804942e-06,
+      "loss": 0.2315,
+      "step": 48707
+    },
+    {
+      "epoch": 3.9458846403110823,
+      "grad_norm": 0.07149714231491089,
+      "learning_rate": 3.0064359332103157e-06,
+      "loss": 0.1983,
+      "step": 48708
+    },
+    {
+      "epoch": 3.9459656513285806,
+      "grad_norm": 0.09264829754829407,
+      "learning_rate": 3.0019352806156897e-06,
+      "loss": 0.2337,
+      "step": 48709
+    },
+    {
+      "epoch": 3.9460466623460793,
+      "grad_norm": 0.06271969527006149,
+      "learning_rate": 2.9974346280210633e-06,
+      "loss": 0.2145,
+      "step": 48710
+    },
+    {
+      "epoch": 3.9461276733635775,
+      "grad_norm": 0.06298379600048065,
+      "learning_rate": 2.992933975426437e-06,
+      "loss": 0.1905,
+      "step": 48711
+    },
+    {
+      "epoch": 3.9462086843810757,
+      "grad_norm": 0.09435836970806122,
+      "learning_rate": 2.988433322831811e-06,
+      "loss": 0.2418,
+      "step": 48712
+    },
+    {
+      "epoch": 3.9462896953985744,
+      "grad_norm": 0.06588099151849747,
+      "learning_rate": 2.9839326702371845e-06,
+      "loss": 0.2022,
+      "step": 48713
+    },
+    {
+      "epoch": 3.9463707064160727,
+      "grad_norm": 0.06596425175666809,
+      "learning_rate": 2.979432017642558e-06,
+      "loss": 0.1856,
+      "step": 48714
+    },
+    {
+      "epoch": 3.946451717433571,
+      "grad_norm": 0.06752095371484756,
+      "learning_rate": 2.974931365047932e-06,
+      "loss": 0.2004,
+      "step": 48715
+    },
+    {
+      "epoch": 3.9465327284510696,
+      "grad_norm": 0.07633611559867859,
+      "learning_rate": 2.9704307124533058e-06,
+      "loss": 0.2349,
+      "step": 48716
+    },
+    {
+      "epoch": 3.946613739468568,
+      "grad_norm": 0.08213532716035843,
+      "learning_rate": 2.9659300598586794e-06,
+      "loss": 0.2315,
+      "step": 48717
+    },
+    {
+      "epoch": 3.946694750486066,
+      "grad_norm": 0.06958873569965363,
+      "learning_rate": 2.9614294072640534e-06,
+      "loss": 0.2222,
+      "step": 48718
+    },
+    {
+      "epoch": 3.9467757615035644,
+      "grad_norm": 0.0933084487915039,
+      "learning_rate": 2.956928754669427e-06,
+      "loss": 0.2254,
+      "step": 48719
+    },
+    {
+      "epoch": 3.9468567725210626,
+      "grad_norm": 0.07122278958559036,
+      "learning_rate": 2.952428102074801e-06,
+      "loss": 0.2559,
+      "step": 48720
+    },
+    {
+      "epoch": 3.9469377835385613,
+      "grad_norm": 0.08493492752313614,
+      "learning_rate": 2.947927449480175e-06,
+      "loss": 0.2415,
+      "step": 48721
+    },
+    {
+      "epoch": 3.9470187945560595,
+      "grad_norm": 0.07494331151247025,
+      "learning_rate": 2.9434267968855487e-06,
+      "loss": 0.2334,
+      "step": 48722
+    },
+    {
+      "epoch": 3.9470998055735578,
+      "grad_norm": 0.07146378606557846,
+      "learning_rate": 2.9389261442909223e-06,
+      "loss": 0.217,
+      "step": 48723
+    },
+    {
+      "epoch": 3.9471808165910565,
+      "grad_norm": 0.05992691218852997,
+      "learning_rate": 2.9344254916962963e-06,
+      "loss": 0.2004,
+      "step": 48724
+    },
+    {
+      "epoch": 3.9472618276085547,
+      "grad_norm": 0.07522998005151749,
+      "learning_rate": 2.92992483910167e-06,
+      "loss": 0.1857,
+      "step": 48725
+    },
+    {
+      "epoch": 3.947342838626053,
+      "grad_norm": 0.07378930598497391,
+      "learning_rate": 2.925424186507044e-06,
+      "loss": 0.1981,
+      "step": 48726
+    },
+    {
+      "epoch": 3.9474238496435516,
+      "grad_norm": 0.08331488817930222,
+      "learning_rate": 2.9209235339124175e-06,
+      "loss": 0.2355,
+      "step": 48727
+    },
+    {
+      "epoch": 3.94750486066105,
+      "grad_norm": 0.08735651522874832,
+      "learning_rate": 2.916422881317791e-06,
+      "loss": 0.2182,
+      "step": 48728
+    },
+    {
+      "epoch": 3.947585871678548,
+      "grad_norm": 0.06788183748722076,
+      "learning_rate": 2.911922228723165e-06,
+      "loss": 0.2102,
+      "step": 48729
+    },
+    {
+      "epoch": 3.947666882696047,
+      "grad_norm": 0.0716099888086319,
+      "learning_rate": 2.9074215761285388e-06,
+      "loss": 0.2277,
+      "step": 48730
+    },
+    {
+      "epoch": 3.947747893713545,
+      "grad_norm": 0.08619321882724762,
+      "learning_rate": 2.9029209235339124e-06,
+      "loss": 0.2234,
+      "step": 48731
+    },
+    {
+      "epoch": 3.9478289047310433,
+      "grad_norm": 0.0653134360909462,
+      "learning_rate": 2.8984202709392864e-06,
+      "loss": 0.2295,
+      "step": 48732
+    },
+    {
+      "epoch": 3.947909915748542,
+      "grad_norm": 0.08110470324754715,
+      "learning_rate": 2.89391961834466e-06,
+      "loss": 0.2422,
+      "step": 48733
+    },
+    {
+      "epoch": 3.9479909267660402,
+      "grad_norm": 0.08325320482254028,
+      "learning_rate": 2.8894189657500336e-06,
+      "loss": 0.22,
+      "step": 48734
+    },
+    {
+      "epoch": 3.9480719377835385,
+      "grad_norm": 0.07577069103717804,
+      "learning_rate": 2.8849183131554076e-06,
+      "loss": 0.1979,
+      "step": 48735
+    },
+    {
+      "epoch": 3.948152948801037,
+      "grad_norm": 0.07868307828903198,
+      "learning_rate": 2.8804176605607816e-06,
+      "loss": 0.241,
+      "step": 48736
+    },
+    {
+      "epoch": 3.9482339598185354,
+      "grad_norm": 0.07431667298078537,
+      "learning_rate": 2.8759170079661552e-06,
+      "loss": 0.2376,
+      "step": 48737
+    },
+    {
+      "epoch": 3.9483149708360337,
+      "grad_norm": 0.08248012512922287,
+      "learning_rate": 2.8714163553715293e-06,
+      "loss": 0.2488,
+      "step": 48738
+    },
+    {
+      "epoch": 3.9483959818535324,
+      "grad_norm": 0.07398643344640732,
+      "learning_rate": 2.866915702776903e-06,
+      "loss": 0.1939,
+      "step": 48739
+    },
+    {
+      "epoch": 3.9484769928710306,
+      "grad_norm": 0.06579329073429108,
+      "learning_rate": 2.8624150501822765e-06,
+      "loss": 0.2379,
+      "step": 48740
+    },
+    {
+      "epoch": 3.948558003888529,
+      "grad_norm": 0.06077032908797264,
+      "learning_rate": 2.8579143975876505e-06,
+      "loss": 0.2155,
+      "step": 48741
+    },
+    {
+      "epoch": 3.948639014906027,
+      "grad_norm": 0.07503858208656311,
+      "learning_rate": 2.853413744993024e-06,
+      "loss": 0.1803,
+      "step": 48742
+    },
+    {
+      "epoch": 3.9487200259235253,
+      "grad_norm": 0.08480262011289597,
+      "learning_rate": 2.8489130923983977e-06,
+      "loss": 0.2527,
+      "step": 48743
+    },
+    {
+      "epoch": 3.948801036941024,
+      "grad_norm": 0.062442317605018616,
+      "learning_rate": 2.8444124398037717e-06,
+      "loss": 0.2152,
+      "step": 48744
+    },
+    {
+      "epoch": 3.9488820479585223,
+      "grad_norm": 0.07862494140863419,
+      "learning_rate": 2.8399117872091453e-06,
+      "loss": 0.2158,
+      "step": 48745
+    },
+    {
+      "epoch": 3.9489630589760205,
+      "grad_norm": 0.07681937515735626,
+      "learning_rate": 2.835411134614519e-06,
+      "loss": 0.2325,
+      "step": 48746
+    },
+    {
+      "epoch": 3.949044069993519,
+      "grad_norm": 0.0824529230594635,
+      "learning_rate": 2.830910482019893e-06,
+      "loss": 0.2331,
+      "step": 48747
+    },
+    {
+      "epoch": 3.9491250810110174,
+      "grad_norm": 0.06473097205162048,
+      "learning_rate": 2.8264098294252666e-06,
+      "loss": 0.2245,
+      "step": 48748
+    },
+    {
+      "epoch": 3.9492060920285157,
+      "grad_norm": 0.08356422930955887,
+      "learning_rate": 2.8219091768306406e-06,
+      "loss": 0.2531,
+      "step": 48749
+    },
+    {
+      "epoch": 3.9492871030460144,
+      "grad_norm": 0.0842425748705864,
+      "learning_rate": 2.8174085242360146e-06,
+      "loss": 0.2154,
+      "step": 48750
+    },
+    {
+      "epoch": 3.9493681140635126,
+      "grad_norm": 0.05836378410458565,
+      "learning_rate": 2.8129078716413882e-06,
+      "loss": 0.2178,
+      "step": 48751
+    },
+    {
+      "epoch": 3.949449125081011,
+      "grad_norm": 0.060801442712545395,
+      "learning_rate": 2.808407219046762e-06,
+      "loss": 0.2085,
+      "step": 48752
+    },
+    {
+      "epoch": 3.9495301360985096,
+      "grad_norm": 0.06836865842342377,
+      "learning_rate": 2.803906566452136e-06,
+      "loss": 0.2641,
+      "step": 48753
+    },
+    {
+      "epoch": 3.949611147116008,
+      "grad_norm": 0.06536520272493362,
+      "learning_rate": 2.7994059138575095e-06,
+      "loss": 0.228,
+      "step": 48754
+    },
+    {
+      "epoch": 3.949692158133506,
+      "grad_norm": 0.0712912380695343,
+      "learning_rate": 2.7949052612628835e-06,
+      "loss": 0.2276,
+      "step": 48755
+    },
+    {
+      "epoch": 3.9497731691510047,
+      "grad_norm": 0.060562681406736374,
+      "learning_rate": 2.790404608668257e-06,
+      "loss": 0.1713,
+      "step": 48756
+    },
+    {
+      "epoch": 3.949854180168503,
+      "grad_norm": 0.07603670656681061,
+      "learning_rate": 2.7859039560736307e-06,
+      "loss": 0.2296,
+      "step": 48757
+    },
+    {
+      "epoch": 3.9499351911860012,
+      "grad_norm": 0.08450186252593994,
+      "learning_rate": 2.7814033034790047e-06,
+      "loss": 0.2233,
+      "step": 48758
+    },
+    {
+      "epoch": 3.9500162022035,
+      "grad_norm": 0.07135901600122452,
+      "learning_rate": 2.7769026508843783e-06,
+      "loss": 0.2052,
+      "step": 48759
+    },
+    {
+      "epoch": 3.950097213220998,
+      "grad_norm": 0.07295495271682739,
+      "learning_rate": 2.772401998289752e-06,
+      "loss": 0.2177,
+      "step": 48760
+    },
+    {
+      "epoch": 3.9501782242384964,
+      "grad_norm": 0.07470090687274933,
+      "learning_rate": 2.767901345695126e-06,
+      "loss": 0.2413,
+      "step": 48761
+    },
+    {
+      "epoch": 3.9502592352559946,
+      "grad_norm": 0.08046706765890121,
+      "learning_rate": 2.7634006931004996e-06,
+      "loss": 0.2657,
+      "step": 48762
+    },
+    {
+      "epoch": 3.9503402462734933,
+      "grad_norm": 0.06358887255191803,
+      "learning_rate": 2.758900040505873e-06,
+      "loss": 0.2253,
+      "step": 48763
+    },
+    {
+      "epoch": 3.9504212572909916,
+      "grad_norm": 0.0710747092962265,
+      "learning_rate": 2.754399387911247e-06,
+      "loss": 0.2431,
+      "step": 48764
+    },
+    {
+      "epoch": 3.95050226830849,
+      "grad_norm": 0.08081281930208206,
+      "learning_rate": 2.7498987353166212e-06,
+      "loss": 0.2467,
+      "step": 48765
+    },
+    {
+      "epoch": 3.950583279325988,
+      "grad_norm": 0.07632328569889069,
+      "learning_rate": 2.745398082721995e-06,
+      "loss": 0.2273,
+      "step": 48766
+    },
+    {
+      "epoch": 3.9506642903434868,
+      "grad_norm": 0.09434718638658524,
+      "learning_rate": 2.740897430127369e-06,
+      "loss": 0.2381,
+      "step": 48767
+    },
+    {
+      "epoch": 3.950745301360985,
+      "grad_norm": 0.0825645849108696,
+      "learning_rate": 2.7363967775327425e-06,
+      "loss": 0.2702,
+      "step": 48768
+    },
+    {
+      "epoch": 3.9508263123784833,
+      "grad_norm": 0.07329137623310089,
+      "learning_rate": 2.731896124938116e-06,
+      "loss": 0.228,
+      "step": 48769
+    },
+    {
+      "epoch": 3.950907323395982,
+      "grad_norm": 0.06830742955207825,
+      "learning_rate": 2.72739547234349e-06,
+      "loss": 0.2123,
+      "step": 48770
+    },
+    {
+      "epoch": 3.95098833441348,
+      "grad_norm": 0.06959126144647598,
+      "learning_rate": 2.7228948197488637e-06,
+      "loss": 0.2137,
+      "step": 48771
+    },
+    {
+      "epoch": 3.9510693454309784,
+      "grad_norm": 0.07482346892356873,
+      "learning_rate": 2.7183941671542373e-06,
+      "loss": 0.2218,
+      "step": 48772
+    },
+    {
+      "epoch": 3.951150356448477,
+      "grad_norm": 0.0668899193406105,
+      "learning_rate": 2.7138935145596113e-06,
+      "loss": 0.2187,
+      "step": 48773
+    },
+    {
+      "epoch": 3.9512313674659754,
+      "grad_norm": 0.061617180705070496,
+      "learning_rate": 2.709392861964985e-06,
+      "loss": 0.2002,
+      "step": 48774
+    },
+    {
+      "epoch": 3.9513123784834736,
+      "grad_norm": 0.08560971915721893,
+      "learning_rate": 2.704892209370359e-06,
+      "loss": 0.2289,
+      "step": 48775
+    },
+    {
+      "epoch": 3.9513933895009723,
+      "grad_norm": 0.08538944274187088,
+      "learning_rate": 2.7003915567757326e-06,
+      "loss": 0.2263,
+      "step": 48776
+    },
+    {
+      "epoch": 3.9514744005184705,
+      "grad_norm": 0.07387255877256393,
+      "learning_rate": 2.695890904181106e-06,
+      "loss": 0.2137,
+      "step": 48777
+    },
+    {
+      "epoch": 3.951555411535969,
+      "grad_norm": 0.07042668759822845,
+      "learning_rate": 2.69139025158648e-06,
+      "loss": 0.1982,
+      "step": 48778
+    },
+    {
+      "epoch": 3.9516364225534675,
+      "grad_norm": 0.09200357645750046,
+      "learning_rate": 2.686889598991854e-06,
+      "loss": 0.21,
+      "step": 48779
+    },
+    {
+      "epoch": 3.9517174335709657,
+      "grad_norm": 0.11538971960544586,
+      "learning_rate": 2.682388946397228e-06,
+      "loss": 0.2395,
+      "step": 48780
+    },
+    {
+      "epoch": 3.951798444588464,
+      "grad_norm": 0.09120815247297287,
+      "learning_rate": 2.6778882938026014e-06,
+      "loss": 0.266,
+      "step": 48781
+    },
+    {
+      "epoch": 3.9518794556059627,
+      "grad_norm": 0.06618034094572067,
+      "learning_rate": 2.6733876412079754e-06,
+      "loss": 0.1992,
+      "step": 48782
+    },
+    {
+      "epoch": 3.951960466623461,
+      "grad_norm": 0.07401160895824432,
+      "learning_rate": 2.668886988613349e-06,
+      "loss": 0.207,
+      "step": 48783
+    },
+    {
+      "epoch": 3.952041477640959,
+      "grad_norm": 0.08141220360994339,
+      "learning_rate": 2.664386336018723e-06,
+      "loss": 0.2225,
+      "step": 48784
+    },
+    {
+      "epoch": 3.9521224886584574,
+      "grad_norm": 0.06158602237701416,
+      "learning_rate": 2.6598856834240967e-06,
+      "loss": 0.1807,
+      "step": 48785
+    },
+    {
+      "epoch": 3.952203499675956,
+      "grad_norm": 0.07168226689100266,
+      "learning_rate": 2.6553850308294703e-06,
+      "loss": 0.1947,
+      "step": 48786
+    },
+    {
+      "epoch": 3.9522845106934543,
+      "grad_norm": 0.10002150386571884,
+      "learning_rate": 2.6508843782348443e-06,
+      "loss": 0.227,
+      "step": 48787
+    },
+    {
+      "epoch": 3.9523655217109526,
+      "grad_norm": 0.060031428933143616,
+      "learning_rate": 2.646383725640218e-06,
+      "loss": 0.2223,
+      "step": 48788
+    },
+    {
+      "epoch": 3.952446532728451,
+      "grad_norm": 0.07038126140832901,
+      "learning_rate": 2.6418830730455915e-06,
+      "loss": 0.22,
+      "step": 48789
+    },
+    {
+      "epoch": 3.9525275437459495,
+      "grad_norm": 0.06693422794342041,
+      "learning_rate": 2.6373824204509655e-06,
+      "loss": 0.2072,
+      "step": 48790
+    },
+    {
+      "epoch": 3.9526085547634477,
+      "grad_norm": 0.07882367074489594,
+      "learning_rate": 2.632881767856339e-06,
+      "loss": 0.2548,
+      "step": 48791
+    },
+    {
+      "epoch": 3.952689565780946,
+      "grad_norm": 0.06965406239032745,
+      "learning_rate": 2.6283811152617127e-06,
+      "loss": 0.23,
+      "step": 48792
+    },
+    {
+      "epoch": 3.9527705767984447,
+      "grad_norm": 0.060692980885505676,
+      "learning_rate": 2.623880462667087e-06,
+      "loss": 0.2233,
+      "step": 48793
+    },
+    {
+      "epoch": 3.952851587815943,
+      "grad_norm": 0.06531994789838791,
+      "learning_rate": 2.619379810072461e-06,
+      "loss": 0.2353,
+      "step": 48794
+    },
+    {
+      "epoch": 3.952932598833441,
+      "grad_norm": 0.06411170959472656,
+      "learning_rate": 2.6148791574778344e-06,
+      "loss": 0.2196,
+      "step": 48795
+    },
+    {
+      "epoch": 3.95301360985094,
+      "grad_norm": 0.08259644359350204,
+      "learning_rate": 2.6103785048832084e-06,
+      "loss": 0.2309,
+      "step": 48796
+    },
+    {
+      "epoch": 3.953094620868438,
+      "grad_norm": 0.07093773037195206,
+      "learning_rate": 2.605877852288582e-06,
+      "loss": 0.2239,
+      "step": 48797
+    },
+    {
+      "epoch": 3.9531756318859363,
+      "grad_norm": 0.06640201807022095,
+      "learning_rate": 2.6013771996939556e-06,
+      "loss": 0.2004,
+      "step": 48798
+    },
+    {
+      "epoch": 3.953256642903435,
+      "grad_norm": 0.07526429742574692,
+      "learning_rate": 2.5968765470993297e-06,
+      "loss": 0.2639,
+      "step": 48799
+    },
+    {
+      "epoch": 3.9533376539209333,
+      "grad_norm": 0.06677767634391785,
+      "learning_rate": 2.5923758945047033e-06,
+      "loss": 0.2551,
+      "step": 48800
+    },
+    {
+      "epoch": 3.9534186649384315,
+      "grad_norm": 0.06592489778995514,
+      "learning_rate": 2.587875241910077e-06,
+      "loss": 0.2013,
+      "step": 48801
+    },
+    {
+      "epoch": 3.95349967595593,
+      "grad_norm": 0.07010385394096375,
+      "learning_rate": 2.583374589315451e-06,
+      "loss": 0.2542,
+      "step": 48802
+    },
+    {
+      "epoch": 3.9535806869734285,
+      "grad_norm": 0.05288232862949371,
+      "learning_rate": 2.5788739367208245e-06,
+      "loss": 0.2347,
+      "step": 48803
+    },
+    {
+      "epoch": 3.9536616979909267,
+      "grad_norm": 0.07920708507299423,
+      "learning_rate": 2.5743732841261985e-06,
+      "loss": 0.2599,
+      "step": 48804
+    },
+    {
+      "epoch": 3.9537427090084254,
+      "grad_norm": 0.07881376147270203,
+      "learning_rate": 2.569872631531572e-06,
+      "loss": 0.2526,
+      "step": 48805
+    },
+    {
+      "epoch": 3.9538237200259236,
+      "grad_norm": 0.07075386494398117,
+      "learning_rate": 2.5653719789369457e-06,
+      "loss": 0.2357,
+      "step": 48806
+    },
+    {
+      "epoch": 3.953904731043422,
+      "grad_norm": 0.0738983005285263,
+      "learning_rate": 2.5608713263423198e-06,
+      "loss": 0.2094,
+      "step": 48807
+    },
+    {
+      "epoch": 3.95398574206092,
+      "grad_norm": 0.06531957536935806,
+      "learning_rate": 2.556370673747694e-06,
+      "loss": 0.1814,
+      "step": 48808
+    },
+    {
+      "epoch": 3.954066753078419,
+      "grad_norm": 0.07314100116491318,
+      "learning_rate": 2.5518700211530674e-06,
+      "loss": 0.2491,
+      "step": 48809
+    },
+    {
+      "epoch": 3.954147764095917,
+      "grad_norm": 0.06974782794713974,
+      "learning_rate": 2.547369368558441e-06,
+      "loss": 0.2168,
+      "step": 48810
+    },
+    {
+      "epoch": 3.9542287751134153,
+      "grad_norm": 0.07167565077543259,
+      "learning_rate": 2.542868715963815e-06,
+      "loss": 0.2411,
+      "step": 48811
+    },
+    {
+      "epoch": 3.9543097861309136,
+      "grad_norm": 0.057915229350328445,
+      "learning_rate": 2.5383680633691886e-06,
+      "loss": 0.2,
+      "step": 48812
+    },
+    {
+      "epoch": 3.9543907971484122,
+      "grad_norm": 0.0739583894610405,
+      "learning_rate": 2.5338674107745627e-06,
+      "loss": 0.2263,
+      "step": 48813
+    },
+    {
+      "epoch": 3.9544718081659105,
+      "grad_norm": 0.08724256604909897,
+      "learning_rate": 2.5293667581799363e-06,
+      "loss": 0.2154,
+      "step": 48814
+    },
+    {
+      "epoch": 3.9545528191834087,
+      "grad_norm": 0.08466099947690964,
+      "learning_rate": 2.52486610558531e-06,
+      "loss": 0.2697,
+      "step": 48815
+    },
+    {
+      "epoch": 3.9546338302009074,
+      "grad_norm": 0.06930121034383774,
+      "learning_rate": 2.520365452990684e-06,
+      "loss": 0.1939,
+      "step": 48816
+    },
+    {
+      "epoch": 3.9547148412184057,
+      "grad_norm": 0.08779342472553253,
+      "learning_rate": 2.5158648003960575e-06,
+      "loss": 0.2267,
+      "step": 48817
+    },
+    {
+      "epoch": 3.954795852235904,
+      "grad_norm": 0.06339520961046219,
+      "learning_rate": 2.511364147801431e-06,
+      "loss": 0.2178,
+      "step": 48818
+    },
+    {
+      "epoch": 3.9548768632534026,
+      "grad_norm": 0.06660187989473343,
+      "learning_rate": 2.506863495206805e-06,
+      "loss": 0.2536,
+      "step": 48819
+    },
+    {
+      "epoch": 3.954957874270901,
+      "grad_norm": 0.06031389907002449,
+      "learning_rate": 2.5023628426121787e-06,
+      "loss": 0.1746,
+      "step": 48820
+    },
+    {
+      "epoch": 3.955038885288399,
+      "grad_norm": 0.07642403244972229,
+      "learning_rate": 2.4978621900175523e-06,
+      "loss": 0.2516,
+      "step": 48821
+    },
+    {
+      "epoch": 3.9551198963058978,
+      "grad_norm": 0.06653022021055222,
+      "learning_rate": 2.4933615374229268e-06,
+      "loss": 0.2368,
+      "step": 48822
+    },
+    {
+      "epoch": 3.955200907323396,
+      "grad_norm": 0.05918882414698601,
+      "learning_rate": 2.4888608848283004e-06,
+      "loss": 0.2194,
+      "step": 48823
+    },
+    {
+      "epoch": 3.9552819183408943,
+      "grad_norm": 0.090346559882164,
+      "learning_rate": 2.484360232233674e-06,
+      "loss": 0.226,
+      "step": 48824
+    },
+    {
+      "epoch": 3.955362929358393,
+      "grad_norm": 0.06755279004573822,
+      "learning_rate": 2.479859579639048e-06,
+      "loss": 0.2484,
+      "step": 48825
+    },
+    {
+      "epoch": 3.955443940375891,
+      "grad_norm": 0.06781566888093948,
+      "learning_rate": 2.4753589270444216e-06,
+      "loss": 0.1975,
+      "step": 48826
+    },
+    {
+      "epoch": 3.9555249513933894,
+      "grad_norm": 0.07741211354732513,
+      "learning_rate": 2.4708582744497952e-06,
+      "loss": 0.2321,
+      "step": 48827
+    },
+    {
+      "epoch": 3.955605962410888,
+      "grad_norm": 0.07529742270708084,
+      "learning_rate": 2.4663576218551692e-06,
+      "loss": 0.2329,
+      "step": 48828
+    },
+    {
+      "epoch": 3.9556869734283864,
+      "grad_norm": 0.06810367852449417,
+      "learning_rate": 2.461856969260543e-06,
+      "loss": 0.2081,
+      "step": 48829
+    },
+    {
+      "epoch": 3.9557679844458846,
+      "grad_norm": 0.06922873854637146,
+      "learning_rate": 2.4573563166659165e-06,
+      "loss": 0.2262,
+      "step": 48830
+    },
+    {
+      "epoch": 3.955848995463383,
+      "grad_norm": 0.07545153796672821,
+      "learning_rate": 2.4528556640712905e-06,
+      "loss": 0.2295,
+      "step": 48831
+    },
+    {
+      "epoch": 3.9559300064808816,
+      "grad_norm": 0.07698988914489746,
+      "learning_rate": 2.448355011476664e-06,
+      "loss": 0.2162,
+      "step": 48832
+    },
+    {
+      "epoch": 3.95601101749838,
+      "grad_norm": 0.06655178219079971,
+      "learning_rate": 2.443854358882038e-06,
+      "loss": 0.2242,
+      "step": 48833
+    },
+    {
+      "epoch": 3.956092028515878,
+      "grad_norm": 0.06900296360254288,
+      "learning_rate": 2.4393537062874117e-06,
+      "loss": 0.2319,
+      "step": 48834
+    },
+    {
+      "epoch": 3.9561730395333763,
+      "grad_norm": 0.0699462965130806,
+      "learning_rate": 2.4348530536927853e-06,
+      "loss": 0.2429,
+      "step": 48835
+    },
+    {
+      "epoch": 3.956254050550875,
+      "grad_norm": 0.07188187539577484,
+      "learning_rate": 2.4303524010981593e-06,
+      "loss": 0.2344,
+      "step": 48836
+    },
+    {
+      "epoch": 3.9563350615683732,
+      "grad_norm": 0.07619171589612961,
+      "learning_rate": 2.4258517485035334e-06,
+      "loss": 0.2031,
+      "step": 48837
+    },
+    {
+      "epoch": 3.9564160725858715,
+      "grad_norm": 0.0603308379650116,
+      "learning_rate": 2.421351095908907e-06,
+      "loss": 0.2033,
+      "step": 48838
+    },
+    {
+      "epoch": 3.95649708360337,
+      "grad_norm": 0.0679556280374527,
+      "learning_rate": 2.416850443314281e-06,
+      "loss": 0.2109,
+      "step": 48839
+    },
+    {
+      "epoch": 3.9565780946208684,
+      "grad_norm": 0.06686028093099594,
+      "learning_rate": 2.4123497907196546e-06,
+      "loss": 0.2583,
+      "step": 48840
+    },
+    {
+      "epoch": 3.9566591056383666,
+      "grad_norm": 0.07004310190677643,
+      "learning_rate": 2.407849138125028e-06,
+      "loss": 0.232,
+      "step": 48841
+    },
+    {
+      "epoch": 3.9567401166558653,
+      "grad_norm": 0.05624713748693466,
+      "learning_rate": 2.4033484855304022e-06,
+      "loss": 0.2123,
+      "step": 48842
+    },
+    {
+      "epoch": 3.9568211276733636,
+      "grad_norm": 0.0720772072672844,
+      "learning_rate": 2.398847832935776e-06,
+      "loss": 0.2112,
+      "step": 48843
+    },
+    {
+      "epoch": 3.956902138690862,
+      "grad_norm": 0.07022903114557266,
+      "learning_rate": 2.3943471803411494e-06,
+      "loss": 0.199,
+      "step": 48844
+    },
+    {
+      "epoch": 3.9569831497083605,
+      "grad_norm": 0.0740879699587822,
+      "learning_rate": 2.3898465277465235e-06,
+      "loss": 0.2469,
+      "step": 48845
+    },
+    {
+      "epoch": 3.9570641607258588,
+      "grad_norm": 0.07837864011526108,
+      "learning_rate": 2.385345875151897e-06,
+      "loss": 0.2414,
+      "step": 48846
+    },
+    {
+      "epoch": 3.957145171743357,
+      "grad_norm": 0.06510287523269653,
+      "learning_rate": 2.3808452225572707e-06,
+      "loss": 0.2119,
+      "step": 48847
+    },
+    {
+      "epoch": 3.9572261827608557,
+      "grad_norm": 0.06796301901340485,
+      "learning_rate": 2.3763445699626447e-06,
+      "loss": 0.1879,
+      "step": 48848
+    },
+    {
+      "epoch": 3.957307193778354,
+      "grad_norm": 0.0652165561914444,
+      "learning_rate": 2.3718439173680183e-06,
+      "loss": 0.2306,
+      "step": 48849
+    },
+    {
+      "epoch": 3.957388204795852,
+      "grad_norm": 0.08012350648641586,
+      "learning_rate": 2.367343264773392e-06,
+      "loss": 0.2511,
+      "step": 48850
+    },
+    {
+      "epoch": 3.957469215813351,
+      "grad_norm": 0.06359118223190308,
+      "learning_rate": 2.3628426121787664e-06,
+      "loss": 0.1956,
+      "step": 48851
+    },
+    {
+      "epoch": 3.957550226830849,
+      "grad_norm": 0.09778131544589996,
+      "learning_rate": 2.35834195958414e-06,
+      "loss": 0.2337,
+      "step": 48852
+    },
+    {
+      "epoch": 3.9576312378483474,
+      "grad_norm": 0.07835190743207932,
+      "learning_rate": 2.3538413069895136e-06,
+      "loss": 0.2037,
+      "step": 48853
+    },
+    {
+      "epoch": 3.9577122488658456,
+      "grad_norm": 0.07351531833410263,
+      "learning_rate": 2.3493406543948876e-06,
+      "loss": 0.276,
+      "step": 48854
+    },
+    {
+      "epoch": 3.9577932598833443,
+      "grad_norm": 0.0607091523706913,
+      "learning_rate": 2.344840001800261e-06,
+      "loss": 0.2262,
+      "step": 48855
+    },
+    {
+      "epoch": 3.9578742709008425,
+      "grad_norm": 0.06498251855373383,
+      "learning_rate": 2.340339349205635e-06,
+      "loss": 0.2159,
+      "step": 48856
+    },
+    {
+      "epoch": 3.957955281918341,
+      "grad_norm": 0.07029843330383301,
+      "learning_rate": 2.335838696611009e-06,
+      "loss": 0.1888,
+      "step": 48857
+    },
+    {
+      "epoch": 3.958036292935839,
+      "grad_norm": 0.07102137804031372,
+      "learning_rate": 2.3313380440163824e-06,
+      "loss": 0.2604,
+      "step": 48858
+    },
+    {
+      "epoch": 3.9581173039533377,
+      "grad_norm": 0.0782492458820343,
+      "learning_rate": 2.326837391421756e-06,
+      "loss": 0.2622,
+      "step": 48859
+    },
+    {
+      "epoch": 3.958198314970836,
+      "grad_norm": 0.06843463331460953,
+      "learning_rate": 2.32233673882713e-06,
+      "loss": 0.2179,
+      "step": 48860
+    },
+    {
+      "epoch": 3.958279325988334,
+      "grad_norm": 0.07084222882986069,
+      "learning_rate": 2.3178360862325037e-06,
+      "loss": 0.1941,
+      "step": 48861
+    },
+    {
+      "epoch": 3.958360337005833,
+      "grad_norm": 0.0704958513379097,
+      "learning_rate": 2.3133354336378777e-06,
+      "loss": 0.226,
+      "step": 48862
+    },
+    {
+      "epoch": 3.958441348023331,
+      "grad_norm": 0.06495588272809982,
+      "learning_rate": 2.3088347810432513e-06,
+      "loss": 0.2503,
+      "step": 48863
+    },
+    {
+      "epoch": 3.9585223590408294,
+      "grad_norm": 0.0666610449552536,
+      "learning_rate": 2.304334128448625e-06,
+      "loss": 0.2016,
+      "step": 48864
+    },
+    {
+      "epoch": 3.958603370058328,
+      "grad_norm": 0.0791105106472969,
+      "learning_rate": 2.299833475853999e-06,
+      "loss": 0.211,
+      "step": 48865
+    },
+    {
+      "epoch": 3.9586843810758263,
+      "grad_norm": 0.08585468679666519,
+      "learning_rate": 2.295332823259373e-06,
+      "loss": 0.2206,
+      "step": 48866
+    },
+    {
+      "epoch": 3.9587653920933246,
+      "grad_norm": 0.07571440190076828,
+      "learning_rate": 2.2908321706647466e-06,
+      "loss": 0.2165,
+      "step": 48867
+    },
+    {
+      "epoch": 3.9588464031108233,
+      "grad_norm": 0.07079493254423141,
+      "learning_rate": 2.2863315180701206e-06,
+      "loss": 0.2351,
+      "step": 48868
+    },
+    {
+      "epoch": 3.9589274141283215,
+      "grad_norm": 0.08157934993505478,
+      "learning_rate": 2.281830865475494e-06,
+      "loss": 0.209,
+      "step": 48869
+    },
+    {
+      "epoch": 3.9590084251458197,
+      "grad_norm": 0.058229394257068634,
+      "learning_rate": 2.2773302128808678e-06,
+      "loss": 0.216,
+      "step": 48870
+    },
+    {
+      "epoch": 3.9590894361633184,
+      "grad_norm": 0.06710825115442276,
+      "learning_rate": 2.272829560286242e-06,
+      "loss": 0.2343,
+      "step": 48871
+    },
+    {
+      "epoch": 3.9591704471808167,
+      "grad_norm": 0.07763312757015228,
+      "learning_rate": 2.2683289076916154e-06,
+      "loss": 0.2398,
+      "step": 48872
+    },
+    {
+      "epoch": 3.959251458198315,
+      "grad_norm": 0.07329755276441574,
+      "learning_rate": 2.263828255096989e-06,
+      "loss": 0.1929,
+      "step": 48873
+    },
+    {
+      "epoch": 3.9593324692158136,
+      "grad_norm": 0.052778083831071854,
+      "learning_rate": 2.259327602502363e-06,
+      "loss": 0.2198,
+      "step": 48874
+    },
+    {
+      "epoch": 3.959413480233312,
+      "grad_norm": 0.06382036954164505,
+      "learning_rate": 2.2548269499077366e-06,
+      "loss": 0.2334,
+      "step": 48875
+    },
+    {
+      "epoch": 3.95949449125081,
+      "grad_norm": 0.06736230105161667,
+      "learning_rate": 2.2503262973131103e-06,
+      "loss": 0.1772,
+      "step": 48876
+    },
+    {
+      "epoch": 3.9595755022683083,
+      "grad_norm": 0.09311016649007797,
+      "learning_rate": 2.2458256447184843e-06,
+      "loss": 0.2461,
+      "step": 48877
+    },
+    {
+      "epoch": 3.959656513285807,
+      "grad_norm": 0.067777618765831,
+      "learning_rate": 2.241324992123858e-06,
+      "loss": 0.2162,
+      "step": 48878
+    },
+    {
+      "epoch": 3.9597375243033053,
+      "grad_norm": 0.07997038960456848,
+      "learning_rate": 2.2368243395292315e-06,
+      "loss": 0.241,
+      "step": 48879
+    },
+    {
+      "epoch": 3.9598185353208035,
+      "grad_norm": 0.07150693982839584,
+      "learning_rate": 2.232323686934606e-06,
+      "loss": 0.2265,
+      "step": 48880
+    },
+    {
+      "epoch": 3.9598995463383018,
+      "grad_norm": 0.06763894110918045,
+      "learning_rate": 2.2278230343399795e-06,
+      "loss": 0.2165,
+      "step": 48881
+    },
+    {
+      "epoch": 3.9599805573558005,
+      "grad_norm": 0.07978539168834686,
+      "learning_rate": 2.223322381745353e-06,
+      "loss": 0.2443,
+      "step": 48882
+    },
+    {
+      "epoch": 3.9600615683732987,
+      "grad_norm": 0.06002137437462807,
+      "learning_rate": 2.218821729150727e-06,
+      "loss": 0.249,
+      "step": 48883
+    },
+    {
+      "epoch": 3.960142579390797,
+      "grad_norm": 0.06918959319591522,
+      "learning_rate": 2.2143210765561008e-06,
+      "loss": 0.2428,
+      "step": 48884
+    },
+    {
+      "epoch": 3.9602235904082956,
+      "grad_norm": 0.06803537160158157,
+      "learning_rate": 2.2098204239614744e-06,
+      "loss": 0.2001,
+      "step": 48885
+    },
+    {
+      "epoch": 3.960304601425794,
+      "grad_norm": 0.06053481996059418,
+      "learning_rate": 2.2053197713668484e-06,
+      "loss": 0.1956,
+      "step": 48886
+    },
+    {
+      "epoch": 3.960385612443292,
+      "grad_norm": 0.062240056693553925,
+      "learning_rate": 2.200819118772222e-06,
+      "loss": 0.2152,
+      "step": 48887
+    },
+    {
+      "epoch": 3.960466623460791,
+      "grad_norm": 0.06525016576051712,
+      "learning_rate": 2.1963184661775956e-06,
+      "loss": 0.1922,
+      "step": 48888
+    },
+    {
+      "epoch": 3.960547634478289,
+      "grad_norm": 0.07532958686351776,
+      "learning_rate": 2.1918178135829696e-06,
+      "loss": 0.2079,
+      "step": 48889
+    },
+    {
+      "epoch": 3.9606286454957873,
+      "grad_norm": 0.08174377679824829,
+      "learning_rate": 2.1873171609883432e-06,
+      "loss": 0.2211,
+      "step": 48890
+    },
+    {
+      "epoch": 3.960709656513286,
+      "grad_norm": 0.07705192267894745,
+      "learning_rate": 2.1828165083937173e-06,
+      "loss": 0.2116,
+      "step": 48891
+    },
+    {
+      "epoch": 3.9607906675307842,
+      "grad_norm": 0.08387063443660736,
+      "learning_rate": 2.178315855799091e-06,
+      "loss": 0.1979,
+      "step": 48892
+    },
+    {
+      "epoch": 3.9608716785482825,
+      "grad_norm": 0.07427896559238434,
+      "learning_rate": 2.1738152032044645e-06,
+      "loss": 0.2114,
+      "step": 48893
+    },
+    {
+      "epoch": 3.960952689565781,
+      "grad_norm": 0.0671217292547226,
+      "learning_rate": 2.1693145506098385e-06,
+      "loss": 0.2102,
+      "step": 48894
+    },
+    {
+      "epoch": 3.9610337005832794,
+      "grad_norm": 0.0731186717748642,
+      "learning_rate": 2.1648138980152125e-06,
+      "loss": 0.2358,
+      "step": 48895
+    },
+    {
+      "epoch": 3.9611147116007777,
+      "grad_norm": 0.08874718844890594,
+      "learning_rate": 2.160313245420586e-06,
+      "loss": 0.2466,
+      "step": 48896
+    },
+    {
+      "epoch": 3.9611957226182763,
+      "grad_norm": 0.08343454450368881,
+      "learning_rate": 2.15581259282596e-06,
+      "loss": 0.2146,
+      "step": 48897
+    },
+    {
+      "epoch": 3.9612767336357746,
+      "grad_norm": 0.07566836476325989,
+      "learning_rate": 2.1513119402313338e-06,
+      "loss": 0.2411,
+      "step": 48898
+    },
+    {
+      "epoch": 3.961357744653273,
+      "grad_norm": 0.08290202170610428,
+      "learning_rate": 2.1468112876367074e-06,
+      "loss": 0.2512,
+      "step": 48899
+    },
+    {
+      "epoch": 3.961438755670771,
+      "grad_norm": 0.07356373220682144,
+      "learning_rate": 2.1423106350420814e-06,
+      "loss": 0.2484,
+      "step": 48900
+    },
+    {
+      "epoch": 3.9615197666882693,
+      "grad_norm": 0.08584269881248474,
+      "learning_rate": 2.137809982447455e-06,
+      "loss": 0.2302,
+      "step": 48901
+    },
+    {
+      "epoch": 3.961600777705768,
+      "grad_norm": 0.06294574588537216,
+      "learning_rate": 2.1333093298528286e-06,
+      "loss": 0.2307,
+      "step": 48902
+    },
+    {
+      "epoch": 3.9616817887232663,
+      "grad_norm": 0.06678062677383423,
+      "learning_rate": 2.1288086772582026e-06,
+      "loss": 0.2059,
+      "step": 48903
+    },
+    {
+      "epoch": 3.9617627997407645,
+      "grad_norm": 0.06297358125448227,
+      "learning_rate": 2.1243080246635762e-06,
+      "loss": 0.2176,
+      "step": 48904
+    },
+    {
+      "epoch": 3.961843810758263,
+      "grad_norm": 0.07585348933935165,
+      "learning_rate": 2.11980737206895e-06,
+      "loss": 0.2497,
+      "step": 48905
+    },
+    {
+      "epoch": 3.9619248217757614,
+      "grad_norm": 0.07568811625242233,
+      "learning_rate": 2.115306719474324e-06,
+      "loss": 0.2114,
+      "step": 48906
+    },
+    {
+      "epoch": 3.9620058327932597,
+      "grad_norm": 0.07441581040620804,
+      "learning_rate": 2.1108060668796975e-06,
+      "loss": 0.287,
+      "step": 48907
+    },
+    {
+      "epoch": 3.9620868438107584,
+      "grad_norm": 0.06748269498348236,
+      "learning_rate": 2.106305414285071e-06,
+      "loss": 0.2269,
+      "step": 48908
+    },
+    {
+      "epoch": 3.9621678548282566,
+      "grad_norm": 0.07350584864616394,
+      "learning_rate": 2.1018047616904455e-06,
+      "loss": 0.2298,
+      "step": 48909
+    },
+    {
+      "epoch": 3.962248865845755,
+      "grad_norm": 0.07611532509326935,
+      "learning_rate": 2.097304109095819e-06,
+      "loss": 0.236,
+      "step": 48910
+    },
+    {
+      "epoch": 3.9623298768632536,
+      "grad_norm": 0.07814755290746689,
+      "learning_rate": 2.0928034565011927e-06,
+      "loss": 0.2699,
+      "step": 48911
+    },
+    {
+      "epoch": 3.962410887880752,
+      "grad_norm": 0.07571461796760559,
+      "learning_rate": 2.0883028039065667e-06,
+      "loss": 0.2033,
+      "step": 48912
+    },
+    {
+      "epoch": 3.96249189889825,
+      "grad_norm": 0.08205527067184448,
+      "learning_rate": 2.0838021513119403e-06,
+      "loss": 0.2449,
+      "step": 48913
+    },
+    {
+      "epoch": 3.9625729099157487,
+      "grad_norm": 0.06475473195314407,
+      "learning_rate": 2.079301498717314e-06,
+      "loss": 0.2245,
+      "step": 48914
+    },
+    {
+      "epoch": 3.962653920933247,
+      "grad_norm": 0.06566368043422699,
+      "learning_rate": 2.074800846122688e-06,
+      "loss": 0.1976,
+      "step": 48915
+    },
+    {
+      "epoch": 3.962734931950745,
+      "grad_norm": 0.08065775781869888,
+      "learning_rate": 2.0703001935280616e-06,
+      "loss": 0.2169,
+      "step": 48916
+    },
+    {
+      "epoch": 3.962815942968244,
+      "grad_norm": 0.06467075645923615,
+      "learning_rate": 2.0657995409334356e-06,
+      "loss": 0.2001,
+      "step": 48917
+    },
+    {
+      "epoch": 3.962896953985742,
+      "grad_norm": 0.06762147694826126,
+      "learning_rate": 2.0612988883388092e-06,
+      "loss": 0.2267,
+      "step": 48918
+    },
+    {
+      "epoch": 3.9629779650032404,
+      "grad_norm": 0.09147316217422485,
+      "learning_rate": 2.056798235744183e-06,
+      "loss": 0.2093,
+      "step": 48919
+    },
+    {
+      "epoch": 3.963058976020739,
+      "grad_norm": 0.050940994173288345,
+      "learning_rate": 2.052297583149557e-06,
+      "loss": 0.2315,
+      "step": 48920
+    },
+    {
+      "epoch": 3.9631399870382373,
+      "grad_norm": 0.06474053114652634,
+      "learning_rate": 2.0477969305549304e-06,
+      "loss": 0.2059,
+      "step": 48921
+    },
+    {
+      "epoch": 3.9632209980557356,
+      "grad_norm": 0.06484822183847427,
+      "learning_rate": 2.043296277960304e-06,
+      "loss": 0.2061,
+      "step": 48922
+    },
+    {
+      "epoch": 3.963302009073234,
+      "grad_norm": 0.0787152349948883,
+      "learning_rate": 2.038795625365678e-06,
+      "loss": 0.2333,
+      "step": 48923
+    },
+    {
+      "epoch": 3.963383020090732,
+      "grad_norm": 0.07503559440374374,
+      "learning_rate": 2.034294972771052e-06,
+      "loss": 0.2119,
+      "step": 48924
+    },
+    {
+      "epoch": 3.9634640311082308,
+      "grad_norm": 0.06612849235534668,
+      "learning_rate": 2.0297943201764257e-06,
+      "loss": 0.2236,
+      "step": 48925
+    },
+    {
+      "epoch": 3.963545042125729,
+      "grad_norm": 0.07943277806043625,
+      "learning_rate": 2.0252936675817997e-06,
+      "loss": 0.1927,
+      "step": 48926
+    },
+    {
+      "epoch": 3.9636260531432272,
+      "grad_norm": 0.08900720626115799,
+      "learning_rate": 2.0207930149871733e-06,
+      "loss": 0.2616,
+      "step": 48927
+    },
+    {
+      "epoch": 3.963707064160726,
+      "grad_norm": 0.07647895067930222,
+      "learning_rate": 2.016292362392547e-06,
+      "loss": 0.2124,
+      "step": 48928
+    },
+    {
+      "epoch": 3.963788075178224,
+      "grad_norm": 0.08729333430528641,
+      "learning_rate": 2.011791709797921e-06,
+      "loss": 0.225,
+      "step": 48929
+    },
+    {
+      "epoch": 3.9638690861957224,
+      "grad_norm": 0.06749184429645538,
+      "learning_rate": 2.0072910572032946e-06,
+      "loss": 0.2568,
+      "step": 48930
+    },
+    {
+      "epoch": 3.963950097213221,
+      "grad_norm": 0.07493914663791656,
+      "learning_rate": 2.002790404608668e-06,
+      "loss": 0.2483,
+      "step": 48931
+    },
+    {
+      "epoch": 3.9640311082307194,
+      "grad_norm": 0.07415442168712616,
+      "learning_rate": 1.998289752014042e-06,
+      "loss": 0.2709,
+      "step": 48932
+    },
+    {
+      "epoch": 3.9641121192482176,
+      "grad_norm": 0.055181387811899185,
+      "learning_rate": 1.993789099419416e-06,
+      "loss": 0.1915,
+      "step": 48933
+    },
+    {
+      "epoch": 3.9641931302657163,
+      "grad_norm": 0.0670875683426857,
+      "learning_rate": 1.9892884468247894e-06,
+      "loss": 0.2323,
+      "step": 48934
+    },
+    {
+      "epoch": 3.9642741412832145,
+      "grad_norm": 0.06657449901103973,
+      "learning_rate": 1.9847877942301634e-06,
+      "loss": 0.2334,
+      "step": 48935
+    },
+    {
+      "epoch": 3.964355152300713,
+      "grad_norm": 0.06159026175737381,
+      "learning_rate": 1.980287141635537e-06,
+      "loss": 0.1616,
+      "step": 48936
+    },
+    {
+      "epoch": 3.9644361633182115,
+      "grad_norm": 0.06844350695610046,
+      "learning_rate": 1.975786489040911e-06,
+      "loss": 0.2003,
+      "step": 48937
+    },
+    {
+      "epoch": 3.9645171743357097,
+      "grad_norm": 0.07265136390924454,
+      "learning_rate": 1.971285836446285e-06,
+      "loss": 0.2481,
+      "step": 48938
+    },
+    {
+      "epoch": 3.964598185353208,
+      "grad_norm": 0.07296200096607208,
+      "learning_rate": 1.9667851838516587e-06,
+      "loss": 0.2074,
+      "step": 48939
+    },
+    {
+      "epoch": 3.9646791963707066,
+      "grad_norm": 0.07084230333566666,
+      "learning_rate": 1.9622845312570323e-06,
+      "loss": 0.2267,
+      "step": 48940
+    },
+    {
+      "epoch": 3.964760207388205,
+      "grad_norm": 0.06154124438762665,
+      "learning_rate": 1.9577838786624063e-06,
+      "loss": 0.218,
+      "step": 48941
+    },
+    {
+      "epoch": 3.964841218405703,
+      "grad_norm": 0.05799073725938797,
+      "learning_rate": 1.95328322606778e-06,
+      "loss": 0.2076,
+      "step": 48942
+    },
+    {
+      "epoch": 3.964922229423202,
+      "grad_norm": 0.06497079133987427,
+      "learning_rate": 1.9487825734731535e-06,
+      "loss": 0.2096,
+      "step": 48943
+    },
+    {
+      "epoch": 3.9650032404407,
+      "grad_norm": 0.07975282520055771,
+      "learning_rate": 1.9442819208785276e-06,
+      "loss": 0.2206,
+      "step": 48944
+    },
+    {
+      "epoch": 3.9650842514581983,
+      "grad_norm": 0.06756958365440369,
+      "learning_rate": 1.939781268283901e-06,
+      "loss": 0.2282,
+      "step": 48945
+    },
+    {
+      "epoch": 3.9651652624756966,
+      "grad_norm": 0.08484717458486557,
+      "learning_rate": 1.935280615689275e-06,
+      "loss": 0.2343,
+      "step": 48946
+    },
+    {
+      "epoch": 3.965246273493195,
+      "grad_norm": 0.07329592853784561,
+      "learning_rate": 1.930779963094649e-06,
+      "loss": 0.2059,
+      "step": 48947
+    },
+    {
+      "epoch": 3.9653272845106935,
+      "grad_norm": 0.05684712901711464,
+      "learning_rate": 1.9262793105000224e-06,
+      "loss": 0.228,
+      "step": 48948
+    },
+    {
+      "epoch": 3.9654082955281917,
+      "grad_norm": 0.05936472862958908,
+      "learning_rate": 1.9217786579053964e-06,
+      "loss": 0.2148,
+      "step": 48949
+    },
+    {
+      "epoch": 3.96548930654569,
+      "grad_norm": 0.07850753515958786,
+      "learning_rate": 1.91727800531077e-06,
+      "loss": 0.2136,
+      "step": 48950
+    },
+    {
+      "epoch": 3.9655703175631887,
+      "grad_norm": 0.0721631646156311,
+      "learning_rate": 1.9127773527161436e-06,
+      "loss": 0.1994,
+      "step": 48951
+    },
+    {
+      "epoch": 3.965651328580687,
+      "grad_norm": 0.06391099095344543,
+      "learning_rate": 1.9082767001215177e-06,
+      "loss": 0.2079,
+      "step": 48952
+    },
+    {
+      "epoch": 3.965732339598185,
+      "grad_norm": 0.06663154065608978,
+      "learning_rate": 1.9037760475268917e-06,
+      "loss": 0.2159,
+      "step": 48953
+    },
+    {
+      "epoch": 3.965813350615684,
+      "grad_norm": 0.060114867985248566,
+      "learning_rate": 1.8992753949322653e-06,
+      "loss": 0.2138,
+      "step": 48954
+    },
+    {
+      "epoch": 3.965894361633182,
+      "grad_norm": 0.07615365833044052,
+      "learning_rate": 1.894774742337639e-06,
+      "loss": 0.2277,
+      "step": 48955
+    },
+    {
+      "epoch": 3.9659753726506803,
+      "grad_norm": 0.06525284796953201,
+      "learning_rate": 1.890274089743013e-06,
+      "loss": 0.2232,
+      "step": 48956
+    },
+    {
+      "epoch": 3.966056383668179,
+      "grad_norm": 0.06666583567857742,
+      "learning_rate": 1.8857734371483867e-06,
+      "loss": 0.2199,
+      "step": 48957
+    },
+    {
+      "epoch": 3.9661373946856773,
+      "grad_norm": 0.07888475805521011,
+      "learning_rate": 1.8812727845537603e-06,
+      "loss": 0.2368,
+      "step": 48958
+    },
+    {
+      "epoch": 3.9662184057031755,
+      "grad_norm": 0.05811379477381706,
+      "learning_rate": 1.8767721319591341e-06,
+      "loss": 0.2513,
+      "step": 48959
+    },
+    {
+      "epoch": 3.966299416720674,
+      "grad_norm": 0.07906852662563324,
+      "learning_rate": 1.872271479364508e-06,
+      "loss": 0.2457,
+      "step": 48960
+    },
+    {
+      "epoch": 3.9663804277381725,
+      "grad_norm": 0.07660603523254395,
+      "learning_rate": 1.8677708267698816e-06,
+      "loss": 0.2245,
+      "step": 48961
+    },
+    {
+      "epoch": 3.9664614387556707,
+      "grad_norm": 0.07632768154144287,
+      "learning_rate": 1.8632701741752554e-06,
+      "loss": 0.2333,
+      "step": 48962
+    },
+    {
+      "epoch": 3.9665424497731694,
+      "grad_norm": 0.07934503257274628,
+      "learning_rate": 1.8587695215806292e-06,
+      "loss": 0.2479,
+      "step": 48963
+    },
+    {
+      "epoch": 3.9666234607906676,
+      "grad_norm": 0.06583593785762787,
+      "learning_rate": 1.854268868986003e-06,
+      "loss": 0.1923,
+      "step": 48964
+    },
+    {
+      "epoch": 3.966704471808166,
+      "grad_norm": 0.07885053008794785,
+      "learning_rate": 1.8497682163913766e-06,
+      "loss": 0.2003,
+      "step": 48965
+    },
+    {
+      "epoch": 3.9667854828256646,
+      "grad_norm": 0.07885368168354034,
+      "learning_rate": 1.8452675637967509e-06,
+      "loss": 0.2529,
+      "step": 48966
+    },
+    {
+      "epoch": 3.966866493843163,
+      "grad_norm": 0.05907120555639267,
+      "learning_rate": 1.8407669112021245e-06,
+      "loss": 0.2194,
+      "step": 48967
+    },
+    {
+      "epoch": 3.966947504860661,
+      "grad_norm": 0.05766279622912407,
+      "learning_rate": 1.8362662586074983e-06,
+      "loss": 0.2244,
+      "step": 48968
+    },
+    {
+      "epoch": 3.9670285158781593,
+      "grad_norm": 0.08073402941226959,
+      "learning_rate": 1.831765606012872e-06,
+      "loss": 0.1892,
+      "step": 48969
+    },
+    {
+      "epoch": 3.9671095268956575,
+      "grad_norm": 0.08297807723283768,
+      "learning_rate": 1.827264953418246e-06,
+      "loss": 0.2275,
+      "step": 48970
+    },
+    {
+      "epoch": 3.9671905379131562,
+      "grad_norm": 0.07996219396591187,
+      "learning_rate": 1.8227643008236195e-06,
+      "loss": 0.2753,
+      "step": 48971
+    },
+    {
+      "epoch": 3.9672715489306545,
+      "grad_norm": 0.06728029996156693,
+      "learning_rate": 1.8182636482289933e-06,
+      "loss": 0.2084,
+      "step": 48972
+    },
+    {
+      "epoch": 3.9673525599481527,
+      "grad_norm": 0.07134328782558441,
+      "learning_rate": 1.8137629956343671e-06,
+      "loss": 0.2387,
+      "step": 48973
+    },
+    {
+      "epoch": 3.9674335709656514,
+      "grad_norm": 0.06936592608690262,
+      "learning_rate": 1.8092623430397407e-06,
+      "loss": 0.226,
+      "step": 48974
+    },
+    {
+      "epoch": 3.9675145819831497,
+      "grad_norm": 0.060569167137145996,
+      "learning_rate": 1.8047616904451146e-06,
+      "loss": 0.2323,
+      "step": 48975
+    },
+    {
+      "epoch": 3.967595593000648,
+      "grad_norm": 0.06395695358514786,
+      "learning_rate": 1.8002610378504884e-06,
+      "loss": 0.2362,
+      "step": 48976
+    },
+    {
+      "epoch": 3.9676766040181466,
+      "grad_norm": 0.0779598280787468,
+      "learning_rate": 1.7957603852558622e-06,
+      "loss": 0.215,
+      "step": 48977
+    },
+    {
+      "epoch": 3.967757615035645,
+      "grad_norm": 0.06363376975059509,
+      "learning_rate": 1.7912597326612358e-06,
+      "loss": 0.1997,
+      "step": 48978
+    },
+    {
+      "epoch": 3.967838626053143,
+      "grad_norm": 0.07200288027524948,
+      "learning_rate": 1.7867590800666096e-06,
+      "loss": 0.2048,
+      "step": 48979
+    },
+    {
+      "epoch": 3.9679196370706418,
+      "grad_norm": 0.07821229100227356,
+      "learning_rate": 1.7822584274719834e-06,
+      "loss": 0.2087,
+      "step": 48980
+    },
+    {
+      "epoch": 3.96800064808814,
+      "grad_norm": 0.062123917043209076,
+      "learning_rate": 1.7777577748773574e-06,
+      "loss": 0.2389,
+      "step": 48981
+    },
+    {
+      "epoch": 3.9680816591056383,
+      "grad_norm": 0.06906332820653915,
+      "learning_rate": 1.7732571222827313e-06,
+      "loss": 0.2279,
+      "step": 48982
+    },
+    {
+      "epoch": 3.968162670123137,
+      "grad_norm": 0.08044512569904327,
+      "learning_rate": 1.7687564696881049e-06,
+      "loss": 0.2301,
+      "step": 48983
+    },
+    {
+      "epoch": 3.968243681140635,
+      "grad_norm": 0.06454446911811829,
+      "learning_rate": 1.7642558170934787e-06,
+      "loss": 0.1969,
+      "step": 48984
+    },
+    {
+      "epoch": 3.9683246921581334,
+      "grad_norm": 0.08553464710712433,
+      "learning_rate": 1.7597551644988525e-06,
+      "loss": 0.2333,
+      "step": 48985
+    },
+    {
+      "epoch": 3.968405703175632,
+      "grad_norm": 0.06968870759010315,
+      "learning_rate": 1.7552545119042263e-06,
+      "loss": 0.2341,
+      "step": 48986
+    },
+    {
+      "epoch": 3.9684867141931304,
+      "grad_norm": 0.06505095958709717,
+      "learning_rate": 1.7507538593096e-06,
+      "loss": 0.2056,
+      "step": 48987
+    },
+    {
+      "epoch": 3.9685677252106286,
+      "grad_norm": 0.07731178402900696,
+      "learning_rate": 1.7462532067149737e-06,
+      "loss": 0.216,
+      "step": 48988
+    },
+    {
+      "epoch": 3.968648736228127,
+      "grad_norm": 0.05740072578191757,
+      "learning_rate": 1.7417525541203475e-06,
+      "loss": 0.2334,
+      "step": 48989
+    },
+    {
+      "epoch": 3.9687297472456255,
+      "grad_norm": 0.06639726459980011,
+      "learning_rate": 1.7372519015257211e-06,
+      "loss": 0.2386,
+      "step": 48990
+    },
+    {
+      "epoch": 3.968810758263124,
+      "grad_norm": 0.0703684538602829,
+      "learning_rate": 1.732751248931095e-06,
+      "loss": 0.2012,
+      "step": 48991
+    },
+    {
+      "epoch": 3.968891769280622,
+      "grad_norm": 0.07246779650449753,
+      "learning_rate": 1.7282505963364688e-06,
+      "loss": 0.2345,
+      "step": 48992
+    },
+    {
+      "epoch": 3.9689727802981203,
+      "grad_norm": 0.060479771345853806,
+      "learning_rate": 1.7237499437418426e-06,
+      "loss": 0.1812,
+      "step": 48993
+    },
+    {
+      "epoch": 3.969053791315619,
+      "grad_norm": 0.06786667555570602,
+      "learning_rate": 1.7192492911472162e-06,
+      "loss": 0.2288,
+      "step": 48994
+    },
+    {
+      "epoch": 3.969134802333117,
+      "grad_norm": 0.07820719480514526,
+      "learning_rate": 1.7147486385525904e-06,
+      "loss": 0.2097,
+      "step": 48995
+    },
+    {
+      "epoch": 3.9692158133506155,
+      "grad_norm": 0.06689919531345367,
+      "learning_rate": 1.710247985957964e-06,
+      "loss": 0.2155,
+      "step": 48996
+    },
+    {
+      "epoch": 3.969296824368114,
+      "grad_norm": 0.07324912399053574,
+      "learning_rate": 1.7057473333633379e-06,
+      "loss": 0.2158,
+      "step": 48997
+    },
+    {
+      "epoch": 3.9693778353856124,
+      "grad_norm": 0.07657838612794876,
+      "learning_rate": 1.7012466807687117e-06,
+      "loss": 0.2181,
+      "step": 48998
+    },
+    {
+      "epoch": 3.9694588464031106,
+      "grad_norm": 0.07740187644958496,
+      "learning_rate": 1.6967460281740855e-06,
+      "loss": 0.2351,
+      "step": 48999
+    },
+    {
+      "epoch": 3.9695398574206093,
+      "grad_norm": 0.048127420246601105,
+      "learning_rate": 1.692245375579459e-06,
+      "loss": 0.2236,
+      "step": 49000
+    },
+    {
+      "epoch": 3.9696208684381076,
+      "grad_norm": 0.07730501145124435,
+      "learning_rate": 1.687744722984833e-06,
+      "loss": 0.2137,
+      "step": 49001
+    },
+    {
+      "epoch": 3.969701879455606,
+      "grad_norm": 0.07064519822597504,
+      "learning_rate": 1.6832440703902067e-06,
+      "loss": 0.2536,
+      "step": 49002
+    },
+    {
+      "epoch": 3.9697828904731045,
+      "grad_norm": 0.07275062054395676,
+      "learning_rate": 1.6787434177955803e-06,
+      "loss": 0.2273,
+      "step": 49003
+    },
+    {
+      "epoch": 3.9698639014906028,
+      "grad_norm": 0.08206738531589508,
+      "learning_rate": 1.6742427652009541e-06,
+      "loss": 0.2448,
+      "step": 49004
+    },
+    {
+      "epoch": 3.969944912508101,
+      "grad_norm": 0.07399595528841019,
+      "learning_rate": 1.669742112606328e-06,
+      "loss": 0.2251,
+      "step": 49005
+    },
+    {
+      "epoch": 3.9700259235255997,
+      "grad_norm": 0.07586342096328735,
+      "learning_rate": 1.6652414600117018e-06,
+      "loss": 0.2214,
+      "step": 49006
+    },
+    {
+      "epoch": 3.970106934543098,
+      "grad_norm": 0.07879979908466339,
+      "learning_rate": 1.6607408074170754e-06,
+      "loss": 0.2474,
+      "step": 49007
+    },
+    {
+      "epoch": 3.970187945560596,
+      "grad_norm": 0.061993636190891266,
+      "learning_rate": 1.6562401548224492e-06,
+      "loss": 0.2171,
+      "step": 49008
+    },
+    {
+      "epoch": 3.970268956578095,
+      "grad_norm": 0.0825590267777443,
+      "learning_rate": 1.6517395022278232e-06,
+      "loss": 0.2091,
+      "step": 49009
+    },
+    {
+      "epoch": 3.970349967595593,
+      "grad_norm": 0.07057903707027435,
+      "learning_rate": 1.647238849633197e-06,
+      "loss": 0.2551,
+      "step": 49010
+    },
+    {
+      "epoch": 3.9704309786130914,
+      "grad_norm": 0.07474622130393982,
+      "learning_rate": 1.6427381970385708e-06,
+      "loss": 0.2076,
+      "step": 49011
+    },
+    {
+      "epoch": 3.9705119896305896,
+      "grad_norm": 0.06988409161567688,
+      "learning_rate": 1.6382375444439444e-06,
+      "loss": 0.2123,
+      "step": 49012
+    },
+    {
+      "epoch": 3.9705930006480883,
+      "grad_norm": 0.09014685451984406,
+      "learning_rate": 1.6337368918493183e-06,
+      "loss": 0.2558,
+      "step": 49013
+    },
+    {
+      "epoch": 3.9706740116655865,
+      "grad_norm": 0.05605039373040199,
+      "learning_rate": 1.629236239254692e-06,
+      "loss": 0.2196,
+      "step": 49014
+    },
+    {
+      "epoch": 3.970755022683085,
+      "grad_norm": 0.06118842214345932,
+      "learning_rate": 1.6247355866600659e-06,
+      "loss": 0.2252,
+      "step": 49015
+    },
+    {
+      "epoch": 3.970836033700583,
+      "grad_norm": 0.0719040036201477,
+      "learning_rate": 1.6202349340654395e-06,
+      "loss": 0.2198,
+      "step": 49016
+    },
+    {
+      "epoch": 3.9709170447180817,
+      "grad_norm": 0.07295656204223633,
+      "learning_rate": 1.6157342814708133e-06,
+      "loss": 0.2345,
+      "step": 49017
+    },
+    {
+      "epoch": 3.97099805573558,
+      "grad_norm": 0.07177040725946426,
+      "learning_rate": 1.6112336288761871e-06,
+      "loss": 0.2319,
+      "step": 49018
+    },
+    {
+      "epoch": 3.971079066753078,
+      "grad_norm": 0.06768198311328888,
+      "learning_rate": 1.606732976281561e-06,
+      "loss": 0.244,
+      "step": 49019
+    },
+    {
+      "epoch": 3.971160077770577,
+      "grad_norm": 0.07683272659778595,
+      "learning_rate": 1.6022323236869345e-06,
+      "loss": 0.2378,
+      "step": 49020
+    },
+    {
+      "epoch": 3.971241088788075,
+      "grad_norm": 0.06740285456180573,
+      "learning_rate": 1.5977316710923084e-06,
+      "loss": 0.2183,
+      "step": 49021
+    },
+    {
+      "epoch": 3.9713220998055734,
+      "grad_norm": 0.0736578181385994,
+      "learning_rate": 1.5932310184976822e-06,
+      "loss": 0.2054,
+      "step": 49022
+    },
+    {
+      "epoch": 3.971403110823072,
+      "grad_norm": 0.10093529522418976,
+      "learning_rate": 1.5887303659030558e-06,
+      "loss": 0.2277,
+      "step": 49023
+    },
+    {
+      "epoch": 3.9714841218405703,
+      "grad_norm": 0.08745209872722626,
+      "learning_rate": 1.58422971330843e-06,
+      "loss": 0.2632,
+      "step": 49024
+    },
+    {
+      "epoch": 3.9715651328580686,
+      "grad_norm": 0.07922547310590744,
+      "learning_rate": 1.5797290607138036e-06,
+      "loss": 0.2189,
+      "step": 49025
+    },
+    {
+      "epoch": 3.9716461438755672,
+      "grad_norm": 0.06714785844087601,
+      "learning_rate": 1.5752284081191774e-06,
+      "loss": 0.1954,
+      "step": 49026
+    },
+    {
+      "epoch": 3.9717271548930655,
+      "grad_norm": 0.07381433248519897,
+      "learning_rate": 1.5707277555245512e-06,
+      "loss": 0.2117,
+      "step": 49027
+    },
+    {
+      "epoch": 3.9718081659105637,
+      "grad_norm": 0.06595097482204437,
+      "learning_rate": 1.566227102929925e-06,
+      "loss": 0.2259,
+      "step": 49028
+    },
+    {
+      "epoch": 3.9718891769280624,
+      "grad_norm": 0.07303611189126968,
+      "learning_rate": 1.5617264503352987e-06,
+      "loss": 0.2157,
+      "step": 49029
+    },
+    {
+      "epoch": 3.9719701879455607,
+      "grad_norm": 0.09227094054222107,
+      "learning_rate": 1.5572257977406725e-06,
+      "loss": 0.2355,
+      "step": 49030
+    },
+    {
+      "epoch": 3.972051198963059,
+      "grad_norm": 0.07859620451927185,
+      "learning_rate": 1.5527251451460463e-06,
+      "loss": 0.2492,
+      "step": 49031
+    },
+    {
+      "epoch": 3.9721322099805576,
+      "grad_norm": 0.07130131870508194,
+      "learning_rate": 1.54822449255142e-06,
+      "loss": 0.2507,
+      "step": 49032
+    },
+    {
+      "epoch": 3.972213220998056,
+      "grad_norm": 0.08751437067985535,
+      "learning_rate": 1.5437238399567937e-06,
+      "loss": 0.22,
+      "step": 49033
+    },
+    {
+      "epoch": 3.972294232015554,
+      "grad_norm": 0.07616513967514038,
+      "learning_rate": 1.5392231873621675e-06,
+      "loss": 0.2547,
+      "step": 49034
+    },
+    {
+      "epoch": 3.9723752430330523,
+      "grad_norm": 0.07148608565330505,
+      "learning_rate": 1.5347225347675413e-06,
+      "loss": 0.2185,
+      "step": 49035
+    },
+    {
+      "epoch": 3.972456254050551,
+      "grad_norm": 0.07048953324556351,
+      "learning_rate": 1.5302218821729152e-06,
+      "loss": 0.2212,
+      "step": 49036
+    },
+    {
+      "epoch": 3.9725372650680493,
+      "grad_norm": 0.06672168523073196,
+      "learning_rate": 1.525721229578289e-06,
+      "loss": 0.1841,
+      "step": 49037
+    },
+    {
+      "epoch": 3.9726182760855475,
+      "grad_norm": 0.07018128037452698,
+      "learning_rate": 1.5212205769836628e-06,
+      "loss": 0.1995,
+      "step": 49038
+    },
+    {
+      "epoch": 3.9726992871030458,
+      "grad_norm": 0.07781115919351578,
+      "learning_rate": 1.5167199243890364e-06,
+      "loss": 0.2275,
+      "step": 49039
+    },
+    {
+      "epoch": 3.9727802981205445,
+      "grad_norm": 0.06993318349123001,
+      "learning_rate": 1.5122192717944102e-06,
+      "loss": 0.2015,
+      "step": 49040
+    },
+    {
+      "epoch": 3.9728613091380427,
+      "grad_norm": 0.0710805132985115,
+      "learning_rate": 1.507718619199784e-06,
+      "loss": 0.2146,
+      "step": 49041
+    },
+    {
+      "epoch": 3.972942320155541,
+      "grad_norm": 0.08362104743719101,
+      "learning_rate": 1.5032179666051578e-06,
+      "loss": 0.2333,
+      "step": 49042
+    },
+    {
+      "epoch": 3.9730233311730396,
+      "grad_norm": 0.07563608884811401,
+      "learning_rate": 1.4987173140105317e-06,
+      "loss": 0.214,
+      "step": 49043
+    },
+    {
+      "epoch": 3.973104342190538,
+      "grad_norm": 0.07376503944396973,
+      "learning_rate": 1.4942166614159055e-06,
+      "loss": 0.1837,
+      "step": 49044
+    },
+    {
+      "epoch": 3.973185353208036,
+      "grad_norm": 0.07147838920354843,
+      "learning_rate": 1.489716008821279e-06,
+      "loss": 0.2029,
+      "step": 49045
+    },
+    {
+      "epoch": 3.973266364225535,
+      "grad_norm": 0.08322165906429291,
+      "learning_rate": 1.4852153562266529e-06,
+      "loss": 0.2219,
+      "step": 49046
+    },
+    {
+      "epoch": 3.973347375243033,
+      "grad_norm": 0.07755421847105026,
+      "learning_rate": 1.4807147036320267e-06,
+      "loss": 0.2314,
+      "step": 49047
+    },
+    {
+      "epoch": 3.9734283862605313,
+      "grad_norm": 0.06550732254981995,
+      "learning_rate": 1.4762140510374005e-06,
+      "loss": 0.218,
+      "step": 49048
+    },
+    {
+      "epoch": 3.97350939727803,
+      "grad_norm": 0.062065187841653824,
+      "learning_rate": 1.4717133984427743e-06,
+      "loss": 0.2148,
+      "step": 49049
+    },
+    {
+      "epoch": 3.9735904082955282,
+      "grad_norm": 0.06424777954816818,
+      "learning_rate": 1.4672127458481481e-06,
+      "loss": 0.223,
+      "step": 49050
+    },
+    {
+      "epoch": 3.9736714193130265,
+      "grad_norm": 0.06383474171161652,
+      "learning_rate": 1.462712093253522e-06,
+      "loss": 0.216,
+      "step": 49051
+    },
+    {
+      "epoch": 3.973752430330525,
+      "grad_norm": 0.07289434969425201,
+      "learning_rate": 1.4582114406588956e-06,
+      "loss": 0.2535,
+      "step": 49052
+    },
+    {
+      "epoch": 3.9738334413480234,
+      "grad_norm": 0.08658458292484283,
+      "learning_rate": 1.4537107880642694e-06,
+      "loss": 0.2414,
+      "step": 49053
+    },
+    {
+      "epoch": 3.9739144523655217,
+      "grad_norm": 0.050192661583423615,
+      "learning_rate": 1.4492101354696432e-06,
+      "loss": 0.2052,
+      "step": 49054
+    },
+    {
+      "epoch": 3.9739954633830203,
+      "grad_norm": 0.061246853321790695,
+      "learning_rate": 1.4447094828750168e-06,
+      "loss": 0.2015,
+      "step": 49055
+    },
+    {
+      "epoch": 3.9740764744005186,
+      "grad_norm": 0.08989216387271881,
+      "learning_rate": 1.4402088302803908e-06,
+      "loss": 0.2483,
+      "step": 49056
+    },
+    {
+      "epoch": 3.974157485418017,
+      "grad_norm": 0.06548626720905304,
+      "learning_rate": 1.4357081776857646e-06,
+      "loss": 0.2245,
+      "step": 49057
+    },
+    {
+      "epoch": 3.974238496435515,
+      "grad_norm": 0.09296084940433502,
+      "learning_rate": 1.4312075250911382e-06,
+      "loss": 0.211,
+      "step": 49058
+    },
+    {
+      "epoch": 3.9743195074530138,
+      "grad_norm": 0.08216899633407593,
+      "learning_rate": 1.426706872496512e-06,
+      "loss": 0.2369,
+      "step": 49059
+    },
+    {
+      "epoch": 3.974400518470512,
+      "grad_norm": 0.06864972412586212,
+      "learning_rate": 1.4222062199018859e-06,
+      "loss": 0.2543,
+      "step": 49060
+    },
+    {
+      "epoch": 3.9744815294880103,
+      "grad_norm": 0.06113716587424278,
+      "learning_rate": 1.4177055673072595e-06,
+      "loss": 0.2291,
+      "step": 49061
+    },
+    {
+      "epoch": 3.9745625405055085,
+      "grad_norm": 0.07123496383428574,
+      "learning_rate": 1.4132049147126333e-06,
+      "loss": 0.192,
+      "step": 49062
+    },
+    {
+      "epoch": 3.974643551523007,
+      "grad_norm": 0.07263696938753128,
+      "learning_rate": 1.4087042621180073e-06,
+      "loss": 0.2148,
+      "step": 49063
+    },
+    {
+      "epoch": 3.9747245625405054,
+      "grad_norm": 0.07747448235750198,
+      "learning_rate": 1.404203609523381e-06,
+      "loss": 0.2068,
+      "step": 49064
+    },
+    {
+      "epoch": 3.9748055735580037,
+      "grad_norm": 0.06405054777860641,
+      "learning_rate": 1.3997029569287547e-06,
+      "loss": 0.2361,
+      "step": 49065
+    },
+    {
+      "epoch": 3.9748865845755024,
+      "grad_norm": 0.08656203746795654,
+      "learning_rate": 1.3952023043341286e-06,
+      "loss": 0.2595,
+      "step": 49066
+    },
+    {
+      "epoch": 3.9749675955930006,
+      "grad_norm": 0.06898082792758942,
+      "learning_rate": 1.3907016517395024e-06,
+      "loss": 0.1922,
+      "step": 49067
+    },
+    {
+      "epoch": 3.975048606610499,
+      "grad_norm": 0.08081252872943878,
+      "learning_rate": 1.386200999144876e-06,
+      "loss": 0.2472,
+      "step": 49068
+    },
+    {
+      "epoch": 3.9751296176279975,
+      "grad_norm": 0.1028057113289833,
+      "learning_rate": 1.3817003465502498e-06,
+      "loss": 0.2194,
+      "step": 49069
+    },
+    {
+      "epoch": 3.975210628645496,
+      "grad_norm": 0.0644211396574974,
+      "learning_rate": 1.3771996939556236e-06,
+      "loss": 0.2284,
+      "step": 49070
+    },
+    {
+      "epoch": 3.975291639662994,
+      "grad_norm": 0.07005336135625839,
+      "learning_rate": 1.3726990413609974e-06,
+      "loss": 0.2258,
+      "step": 49071
+    },
+    {
+      "epoch": 3.9753726506804927,
+      "grad_norm": 0.07396363466978073,
+      "learning_rate": 1.3681983887663712e-06,
+      "loss": 0.2059,
+      "step": 49072
+    },
+    {
+      "epoch": 3.975453661697991,
+      "grad_norm": 0.08447012305259705,
+      "learning_rate": 1.363697736171745e-06,
+      "loss": 0.2223,
+      "step": 49073
+    },
+    {
+      "epoch": 3.975534672715489,
+      "grad_norm": 0.07208585739135742,
+      "learning_rate": 1.3591970835771186e-06,
+      "loss": 0.2292,
+      "step": 49074
+    },
+    {
+      "epoch": 3.975615683732988,
+      "grad_norm": 0.07174715399742126,
+      "learning_rate": 1.3546964309824925e-06,
+      "loss": 0.2476,
+      "step": 49075
+    },
+    {
+      "epoch": 3.975696694750486,
+      "grad_norm": 0.06257397681474686,
+      "learning_rate": 1.3501957783878663e-06,
+      "loss": 0.2405,
+      "step": 49076
+    },
+    {
+      "epoch": 3.9757777057679844,
+      "grad_norm": 0.07049359381198883,
+      "learning_rate": 1.34569512579324e-06,
+      "loss": 0.2424,
+      "step": 49077
+    },
+    {
+      "epoch": 3.975858716785483,
+      "grad_norm": 0.06485053151845932,
+      "learning_rate": 1.341194473198614e-06,
+      "loss": 0.2209,
+      "step": 49078
+    },
+    {
+      "epoch": 3.9759397278029813,
+      "grad_norm": 0.07532617449760437,
+      "learning_rate": 1.3366938206039877e-06,
+      "loss": 0.2469,
+      "step": 49079
+    },
+    {
+      "epoch": 3.9760207388204796,
+      "grad_norm": 0.07897515594959259,
+      "learning_rate": 1.3321931680093615e-06,
+      "loss": 0.2256,
+      "step": 49080
+    },
+    {
+      "epoch": 3.976101749837978,
+      "grad_norm": 0.07902484387159348,
+      "learning_rate": 1.3276925154147351e-06,
+      "loss": 0.2179,
+      "step": 49081
+    },
+    {
+      "epoch": 3.9761827608554765,
+      "grad_norm": 0.07536326348781586,
+      "learning_rate": 1.323191862820109e-06,
+      "loss": 0.2311,
+      "step": 49082
+    },
+    {
+      "epoch": 3.9762637718729748,
+      "grad_norm": 0.1022820770740509,
+      "learning_rate": 1.3186912102254828e-06,
+      "loss": 0.2329,
+      "step": 49083
+    },
+    {
+      "epoch": 3.976344782890473,
+      "grad_norm": 0.08580672740936279,
+      "learning_rate": 1.3141905576308564e-06,
+      "loss": 0.2162,
+      "step": 49084
+    },
+    {
+      "epoch": 3.9764257939079712,
+      "grad_norm": 0.06963769346475601,
+      "learning_rate": 1.3096899050362304e-06,
+      "loss": 0.2087,
+      "step": 49085
+    },
+    {
+      "epoch": 3.97650680492547,
+      "grad_norm": 0.06818298995494843,
+      "learning_rate": 1.3051892524416042e-06,
+      "loss": 0.1951,
+      "step": 49086
+    },
+    {
+      "epoch": 3.976587815942968,
+      "grad_norm": 0.0778006762266159,
+      "learning_rate": 1.3006885998469778e-06,
+      "loss": 0.2266,
+      "step": 49087
+    },
+    {
+      "epoch": 3.9766688269604664,
+      "grad_norm": 0.06969862431287766,
+      "learning_rate": 1.2961879472523516e-06,
+      "loss": 0.2652,
+      "step": 49088
+    },
+    {
+      "epoch": 3.976749837977965,
+      "grad_norm": 0.07298214733600616,
+      "learning_rate": 1.2916872946577254e-06,
+      "loss": 0.2344,
+      "step": 49089
+    },
+    {
+      "epoch": 3.9768308489954634,
+      "grad_norm": 0.07190950959920883,
+      "learning_rate": 1.2871866420630993e-06,
+      "loss": 0.2142,
+      "step": 49090
+    },
+    {
+      "epoch": 3.9769118600129616,
+      "grad_norm": 0.07995373755693436,
+      "learning_rate": 1.2826859894684729e-06,
+      "loss": 0.1973,
+      "step": 49091
+    },
+    {
+      "epoch": 3.9769928710304603,
+      "grad_norm": 0.0613616481423378,
+      "learning_rate": 1.278185336873847e-06,
+      "loss": 0.2093,
+      "step": 49092
+    },
+    {
+      "epoch": 3.9770738820479585,
+      "grad_norm": 0.06585995107889175,
+      "learning_rate": 1.2736846842792205e-06,
+      "loss": 0.2344,
+      "step": 49093
+    },
+    {
+      "epoch": 3.9771548930654568,
+      "grad_norm": 0.07106611132621765,
+      "learning_rate": 1.2691840316845943e-06,
+      "loss": 0.2236,
+      "step": 49094
+    },
+    {
+      "epoch": 3.9772359040829555,
+      "grad_norm": 0.08742845058441162,
+      "learning_rate": 1.2646833790899681e-06,
+      "loss": 0.2795,
+      "step": 49095
+    },
+    {
+      "epoch": 3.9773169151004537,
+      "grad_norm": 0.06771603226661682,
+      "learning_rate": 1.260182726495342e-06,
+      "loss": 0.1974,
+      "step": 49096
+    },
+    {
+      "epoch": 3.977397926117952,
+      "grad_norm": 0.07327091693878174,
+      "learning_rate": 1.2556820739007155e-06,
+      "loss": 0.2038,
+      "step": 49097
+    },
+    {
+      "epoch": 3.9774789371354506,
+      "grad_norm": 0.08291606605052948,
+      "learning_rate": 1.2511814213060894e-06,
+      "loss": 0.2405,
+      "step": 49098
+    },
+    {
+      "epoch": 3.977559948152949,
+      "grad_norm": 0.07166020572185516,
+      "learning_rate": 1.2466807687114634e-06,
+      "loss": 0.2089,
+      "step": 49099
+    },
+    {
+      "epoch": 3.977640959170447,
+      "grad_norm": 0.0633031502366066,
+      "learning_rate": 1.242180116116837e-06,
+      "loss": 0.2083,
+      "step": 49100
+    },
+    {
+      "epoch": 3.977721970187946,
+      "grad_norm": 0.07655423879623413,
+      "learning_rate": 1.2376794635222108e-06,
+      "loss": 0.2237,
+      "step": 49101
+    },
+    {
+      "epoch": 3.977802981205444,
+      "grad_norm": 0.06801167130470276,
+      "learning_rate": 1.2331788109275846e-06,
+      "loss": 0.2008,
+      "step": 49102
+    },
+    {
+      "epoch": 3.9778839922229423,
+      "grad_norm": 0.09980975836515427,
+      "learning_rate": 1.2286781583329582e-06,
+      "loss": 0.2363,
+      "step": 49103
+    },
+    {
+      "epoch": 3.9779650032404406,
+      "grad_norm": 0.08504997938871384,
+      "learning_rate": 1.224177505738332e-06,
+      "loss": 0.241,
+      "step": 49104
+    },
+    {
+      "epoch": 3.9780460142579392,
+      "grad_norm": 0.07652637362480164,
+      "learning_rate": 1.2196768531437059e-06,
+      "loss": 0.2589,
+      "step": 49105
+    },
+    {
+      "epoch": 3.9781270252754375,
+      "grad_norm": 0.06277074664831161,
+      "learning_rate": 1.2151762005490797e-06,
+      "loss": 0.2218,
+      "step": 49106
+    },
+    {
+      "epoch": 3.9782080362929357,
+      "grad_norm": 0.06175466999411583,
+      "learning_rate": 1.2106755479544535e-06,
+      "loss": 0.2167,
+      "step": 49107
+    },
+    {
+      "epoch": 3.978289047310434,
+      "grad_norm": 0.06797228753566742,
+      "learning_rate": 1.2061748953598273e-06,
+      "loss": 0.1944,
+      "step": 49108
+    },
+    {
+      "epoch": 3.9783700583279327,
+      "grad_norm": 0.07684358209371567,
+      "learning_rate": 1.2016742427652011e-06,
+      "loss": 0.2701,
+      "step": 49109
+    },
+    {
+      "epoch": 3.978451069345431,
+      "grad_norm": 0.06503838300704956,
+      "learning_rate": 1.1971735901705747e-06,
+      "loss": 0.1913,
+      "step": 49110
+    },
+    {
+      "epoch": 3.978532080362929,
+      "grad_norm": 0.06464909762144089,
+      "learning_rate": 1.1926729375759485e-06,
+      "loss": 0.1783,
+      "step": 49111
+    },
+    {
+      "epoch": 3.978613091380428,
+      "grad_norm": 0.09183213114738464,
+      "learning_rate": 1.1881722849813223e-06,
+      "loss": 0.2243,
+      "step": 49112
+    },
+    {
+      "epoch": 3.978694102397926,
+      "grad_norm": 0.05835896357893944,
+      "learning_rate": 1.183671632386696e-06,
+      "loss": 0.1829,
+      "step": 49113
+    },
+    {
+      "epoch": 3.9787751134154243,
+      "grad_norm": 0.07257760316133499,
+      "learning_rate": 1.17917097979207e-06,
+      "loss": 0.2254,
+      "step": 49114
+    },
+    {
+      "epoch": 3.978856124432923,
+      "grad_norm": 0.06942275911569595,
+      "learning_rate": 1.1746703271974438e-06,
+      "loss": 0.2233,
+      "step": 49115
+    },
+    {
+      "epoch": 3.9789371354504213,
+      "grad_norm": 0.0803317055106163,
+      "learning_rate": 1.1701696746028174e-06,
+      "loss": 0.2426,
+      "step": 49116
+    },
+    {
+      "epoch": 3.9790181464679195,
+      "grad_norm": 0.08098768442869186,
+      "learning_rate": 1.1656690220081912e-06,
+      "loss": 0.2706,
+      "step": 49117
+    },
+    {
+      "epoch": 3.979099157485418,
+      "grad_norm": 0.07547670602798462,
+      "learning_rate": 1.161168369413565e-06,
+      "loss": 0.2932,
+      "step": 49118
+    },
+    {
+      "epoch": 3.9791801685029164,
+      "grad_norm": 0.060938771814107895,
+      "learning_rate": 1.1566677168189388e-06,
+      "loss": 0.2588,
+      "step": 49119
+    },
+    {
+      "epoch": 3.9792611795204147,
+      "grad_norm": 0.0760703831911087,
+      "learning_rate": 1.1521670642243124e-06,
+      "loss": 0.2226,
+      "step": 49120
+    },
+    {
+      "epoch": 3.9793421905379134,
+      "grad_norm": 0.0770978331565857,
+      "learning_rate": 1.1476664116296865e-06,
+      "loss": 0.2161,
+      "step": 49121
+    },
+    {
+      "epoch": 3.9794232015554116,
+      "grad_norm": 0.08614175021648407,
+      "learning_rate": 1.1431657590350603e-06,
+      "loss": 0.2087,
+      "step": 49122
+    },
+    {
+      "epoch": 3.97950421257291,
+      "grad_norm": 0.07408390194177628,
+      "learning_rate": 1.1386651064404339e-06,
+      "loss": 0.2756,
+      "step": 49123
+    },
+    {
+      "epoch": 3.9795852235904086,
+      "grad_norm": 0.07045788317918777,
+      "learning_rate": 1.1341644538458077e-06,
+      "loss": 0.25,
+      "step": 49124
+    },
+    {
+      "epoch": 3.979666234607907,
+      "grad_norm": 0.06843449175357819,
+      "learning_rate": 1.1296638012511815e-06,
+      "loss": 0.235,
+      "step": 49125
+    },
+    {
+      "epoch": 3.979747245625405,
+      "grad_norm": 0.06088000163435936,
+      "learning_rate": 1.1251631486565551e-06,
+      "loss": 0.1662,
+      "step": 49126
+    },
+    {
+      "epoch": 3.9798282566429033,
+      "grad_norm": 0.09108071774244308,
+      "learning_rate": 1.120662496061929e-06,
+      "loss": 0.1939,
+      "step": 49127
+    },
+    {
+      "epoch": 3.9799092676604015,
+      "grad_norm": 0.08113054931163788,
+      "learning_rate": 1.116161843467303e-06,
+      "loss": 0.2037,
+      "step": 49128
+    },
+    {
+      "epoch": 3.9799902786779002,
+      "grad_norm": 0.06638524681329727,
+      "learning_rate": 1.1116611908726766e-06,
+      "loss": 0.2,
+      "step": 49129
+    },
+    {
+      "epoch": 3.9800712896953985,
+      "grad_norm": 0.07080232352018356,
+      "learning_rate": 1.1071605382780504e-06,
+      "loss": 0.2203,
+      "step": 49130
+    },
+    {
+      "epoch": 3.9801523007128967,
+      "grad_norm": 0.06715063750743866,
+      "learning_rate": 1.1026598856834242e-06,
+      "loss": 0.2034,
+      "step": 49131
+    },
+    {
+      "epoch": 3.9802333117303954,
+      "grad_norm": 0.07261967658996582,
+      "learning_rate": 1.0981592330887978e-06,
+      "loss": 0.2407,
+      "step": 49132
+    },
+    {
+      "epoch": 3.9803143227478937,
+      "grad_norm": 0.08318926393985748,
+      "learning_rate": 1.0936585804941716e-06,
+      "loss": 0.1799,
+      "step": 49133
+    },
+    {
+      "epoch": 3.980395333765392,
+      "grad_norm": 0.07309187203645706,
+      "learning_rate": 1.0891579278995454e-06,
+      "loss": 0.2382,
+      "step": 49134
+    },
+    {
+      "epoch": 3.9804763447828906,
+      "grad_norm": 0.07083363831043243,
+      "learning_rate": 1.0846572753049192e-06,
+      "loss": 0.2248,
+      "step": 49135
+    },
+    {
+      "epoch": 3.980557355800389,
+      "grad_norm": 0.07788734883069992,
+      "learning_rate": 1.080156622710293e-06,
+      "loss": 0.2758,
+      "step": 49136
+    },
+    {
+      "epoch": 3.980638366817887,
+      "grad_norm": 0.07184889167547226,
+      "learning_rate": 1.0756559701156669e-06,
+      "loss": 0.2069,
+      "step": 49137
+    },
+    {
+      "epoch": 3.9807193778353858,
+      "grad_norm": 0.05995340272784233,
+      "learning_rate": 1.0711553175210407e-06,
+      "loss": 0.1908,
+      "step": 49138
+    },
+    {
+      "epoch": 3.980800388852884,
+      "grad_norm": 0.06793571263551712,
+      "learning_rate": 1.0666546649264143e-06,
+      "loss": 0.2183,
+      "step": 49139
+    },
+    {
+      "epoch": 3.9808813998703823,
+      "grad_norm": 0.0675991103053093,
+      "learning_rate": 1.0621540123317881e-06,
+      "loss": 0.2088,
+      "step": 49140
+    },
+    {
+      "epoch": 3.980962410887881,
+      "grad_norm": 0.06840286403894424,
+      "learning_rate": 1.057653359737162e-06,
+      "loss": 0.2176,
+      "step": 49141
+    },
+    {
+      "epoch": 3.981043421905379,
+      "grad_norm": 0.07902107387781143,
+      "learning_rate": 1.0531527071425355e-06,
+      "loss": 0.2281,
+      "step": 49142
+    },
+    {
+      "epoch": 3.9811244329228774,
+      "grad_norm": 0.06392716616392136,
+      "learning_rate": 1.0486520545479096e-06,
+      "loss": 0.2276,
+      "step": 49143
+    },
+    {
+      "epoch": 3.981205443940376,
+      "grad_norm": 0.07346402853727341,
+      "learning_rate": 1.0441514019532834e-06,
+      "loss": 0.2349,
+      "step": 49144
+    },
+    {
+      "epoch": 3.9812864549578744,
+      "grad_norm": 0.07689003646373749,
+      "learning_rate": 1.039650749358657e-06,
+      "loss": 0.2228,
+      "step": 49145
+    },
+    {
+      "epoch": 3.9813674659753726,
+      "grad_norm": 0.0874786451458931,
+      "learning_rate": 1.0351500967640308e-06,
+      "loss": 0.2419,
+      "step": 49146
+    },
+    {
+      "epoch": 3.9814484769928713,
+      "grad_norm": 0.07139569520950317,
+      "learning_rate": 1.0306494441694046e-06,
+      "loss": 0.252,
+      "step": 49147
+    },
+    {
+      "epoch": 3.9815294880103695,
+      "grad_norm": 0.08377675712108612,
+      "learning_rate": 1.0261487915747784e-06,
+      "loss": 0.2182,
+      "step": 49148
+    },
+    {
+      "epoch": 3.981610499027868,
+      "grad_norm": 0.0838305801153183,
+      "learning_rate": 1.021648138980152e-06,
+      "loss": 0.237,
+      "step": 49149
+    },
+    {
+      "epoch": 3.981691510045366,
+      "grad_norm": 0.07349993288516998,
+      "learning_rate": 1.017147486385526e-06,
+      "loss": 0.2277,
+      "step": 49150
+    },
+    {
+      "epoch": 3.9817725210628643,
+      "grad_norm": 0.07764997333288193,
+      "learning_rate": 1.0126468337908999e-06,
+      "loss": 0.2132,
+      "step": 49151
+    },
+    {
+      "epoch": 3.981853532080363,
+      "grad_norm": 0.07144745439291,
+      "learning_rate": 1.0081461811962735e-06,
+      "loss": 0.2223,
+      "step": 49152
+    },
+    {
+      "epoch": 3.981934543097861,
+      "grad_norm": 0.06616366654634476,
+      "learning_rate": 1.0036455286016473e-06,
+      "loss": 0.2418,
+      "step": 49153
+    },
+    {
+      "epoch": 3.9820155541153595,
+      "grad_norm": 0.0774175375699997,
+      "learning_rate": 9.99144876007021e-07,
+      "loss": 0.2486,
+      "step": 49154
+    },
+    {
+      "epoch": 3.982096565132858,
+      "grad_norm": 0.08736024051904678,
+      "learning_rate": 9.946442234123947e-07,
+      "loss": 0.2434,
+      "step": 49155
+    },
+    {
+      "epoch": 3.9821775761503564,
+      "grad_norm": 0.06919243931770325,
+      "learning_rate": 9.901435708177685e-07,
+      "loss": 0.2726,
+      "step": 49156
+    },
+    {
+      "epoch": 3.9822585871678546,
+      "grad_norm": 0.07379870116710663,
+      "learning_rate": 9.856429182231425e-07,
+      "loss": 0.2308,
+      "step": 49157
+    },
+    {
+      "epoch": 3.9823395981853533,
+      "grad_norm": 0.05552654340863228,
+      "learning_rate": 9.811422656285161e-07,
+      "loss": 0.2211,
+      "step": 49158
+    },
+    {
+      "epoch": 3.9824206092028516,
+      "grad_norm": 0.07735942304134369,
+      "learning_rate": 9.7664161303389e-07,
+      "loss": 0.2101,
+      "step": 49159
+    },
+    {
+      "epoch": 3.98250162022035,
+      "grad_norm": 0.07008510828018188,
+      "learning_rate": 9.721409604392638e-07,
+      "loss": 0.2402,
+      "step": 49160
+    },
+    {
+      "epoch": 3.9825826312378485,
+      "grad_norm": 0.06766650825738907,
+      "learning_rate": 9.676403078446376e-07,
+      "loss": 0.2189,
+      "step": 49161
+    },
+    {
+      "epoch": 3.9826636422553467,
+      "grad_norm": 0.06831780821084976,
+      "learning_rate": 9.631396552500112e-07,
+      "loss": 0.1818,
+      "step": 49162
+    },
+    {
+      "epoch": 3.982744653272845,
+      "grad_norm": 0.07361584156751633,
+      "learning_rate": 9.58639002655385e-07,
+      "loss": 0.2391,
+      "step": 49163
+    },
+    {
+      "epoch": 3.9828256642903437,
+      "grad_norm": 0.07479403913021088,
+      "learning_rate": 9.541383500607588e-07,
+      "loss": 0.2105,
+      "step": 49164
+    },
+    {
+      "epoch": 3.982906675307842,
+      "grad_norm": 0.07533962279558182,
+      "learning_rate": 9.496376974661326e-07,
+      "loss": 0.2128,
+      "step": 49165
+    },
+    {
+      "epoch": 3.98298768632534,
+      "grad_norm": 0.07377752661705017,
+      "learning_rate": 9.451370448715065e-07,
+      "loss": 0.2785,
+      "step": 49166
+    },
+    {
+      "epoch": 3.983068697342839,
+      "grad_norm": 0.06769118458032608,
+      "learning_rate": 9.406363922768802e-07,
+      "loss": 0.2386,
+      "step": 49167
+    },
+    {
+      "epoch": 3.983149708360337,
+      "grad_norm": 0.06710786372423172,
+      "learning_rate": 9.36135739682254e-07,
+      "loss": 0.2021,
+      "step": 49168
+    },
+    {
+      "epoch": 3.9832307193778353,
+      "grad_norm": 0.07649517804384232,
+      "learning_rate": 9.316350870876277e-07,
+      "loss": 0.2129,
+      "step": 49169
+    },
+    {
+      "epoch": 3.983311730395334,
+      "grad_norm": 0.07384105026721954,
+      "learning_rate": 9.271344344930015e-07,
+      "loss": 0.206,
+      "step": 49170
+    },
+    {
+      "epoch": 3.9833927414128323,
+      "grad_norm": 0.07594034075737,
+      "learning_rate": 9.226337818983754e-07,
+      "loss": 0.2499,
+      "step": 49171
+    },
+    {
+      "epoch": 3.9834737524303305,
+      "grad_norm": 0.07467283308506012,
+      "learning_rate": 9.181331293037491e-07,
+      "loss": 0.236,
+      "step": 49172
+    },
+    {
+      "epoch": 3.9835547634478288,
+      "grad_norm": 0.07646078616380692,
+      "learning_rate": 9.13632476709123e-07,
+      "loss": 0.2835,
+      "step": 49173
+    },
+    {
+      "epoch": 3.983635774465327,
+      "grad_norm": 0.079698346555233,
+      "learning_rate": 9.091318241144967e-07,
+      "loss": 0.2269,
+      "step": 49174
+    },
+    {
+      "epoch": 3.9837167854828257,
+      "grad_norm": 0.08032261580228806,
+      "learning_rate": 9.046311715198704e-07,
+      "loss": 0.2175,
+      "step": 49175
+    },
+    {
+      "epoch": 3.983797796500324,
+      "grad_norm": 0.08225135505199432,
+      "learning_rate": 9.001305189252442e-07,
+      "loss": 0.2507,
+      "step": 49176
+    },
+    {
+      "epoch": 3.983878807517822,
+      "grad_norm": 0.08566189557313919,
+      "learning_rate": 8.956298663306179e-07,
+      "loss": 0.2036,
+      "step": 49177
+    },
+    {
+      "epoch": 3.983959818535321,
+      "grad_norm": 0.07229893654584885,
+      "learning_rate": 8.911292137359917e-07,
+      "loss": 0.2109,
+      "step": 49178
+    },
+    {
+      "epoch": 3.984040829552819,
+      "grad_norm": 0.07611256837844849,
+      "learning_rate": 8.866285611413656e-07,
+      "loss": 0.2151,
+      "step": 49179
+    },
+    {
+      "epoch": 3.9841218405703174,
+      "grad_norm": 0.06633207947015762,
+      "learning_rate": 8.821279085467393e-07,
+      "loss": 0.2284,
+      "step": 49180
+    },
+    {
+      "epoch": 3.984202851587816,
+      "grad_norm": 0.0713014304637909,
+      "learning_rate": 8.776272559521132e-07,
+      "loss": 0.2267,
+      "step": 49181
+    },
+    {
+      "epoch": 3.9842838626053143,
+      "grad_norm": 0.05671551451086998,
+      "learning_rate": 8.731266033574869e-07,
+      "loss": 0.231,
+      "step": 49182
+    },
+    {
+      "epoch": 3.9843648736228126,
+      "grad_norm": 0.08908804506063461,
+      "learning_rate": 8.686259507628606e-07,
+      "loss": 0.243,
+      "step": 49183
+    },
+    {
+      "epoch": 3.9844458846403112,
+      "grad_norm": 0.08369207382202148,
+      "learning_rate": 8.641252981682344e-07,
+      "loss": 0.2124,
+      "step": 49184
+    },
+    {
+      "epoch": 3.9845268956578095,
+      "grad_norm": 0.075348399579525,
+      "learning_rate": 8.596246455736081e-07,
+      "loss": 0.2084,
+      "step": 49185
+    },
+    {
+      "epoch": 3.9846079066753077,
+      "grad_norm": 0.07731541991233826,
+      "learning_rate": 8.55123992978982e-07,
+      "loss": 0.2089,
+      "step": 49186
+    },
+    {
+      "epoch": 3.9846889176928064,
+      "grad_norm": 0.06858506798744202,
+      "learning_rate": 8.506233403843558e-07,
+      "loss": 0.1975,
+      "step": 49187
+    },
+    {
+      "epoch": 3.9847699287103047,
+      "grad_norm": 0.06854752451181412,
+      "learning_rate": 8.461226877897295e-07,
+      "loss": 0.2154,
+      "step": 49188
+    },
+    {
+      "epoch": 3.984850939727803,
+      "grad_norm": 0.0722111314535141,
+      "learning_rate": 8.416220351951034e-07,
+      "loss": 0.2499,
+      "step": 49189
+    },
+    {
+      "epoch": 3.9849319507453016,
+      "grad_norm": 0.07207582145929337,
+      "learning_rate": 8.371213826004771e-07,
+      "loss": 0.2138,
+      "step": 49190
+    },
+    {
+      "epoch": 3.9850129617628,
+      "grad_norm": 0.07529518753290176,
+      "learning_rate": 8.326207300058509e-07,
+      "loss": 0.2467,
+      "step": 49191
+    },
+    {
+      "epoch": 3.985093972780298,
+      "grad_norm": 0.0700589045882225,
+      "learning_rate": 8.281200774112246e-07,
+      "loss": 0.237,
+      "step": 49192
+    },
+    {
+      "epoch": 3.9851749837977968,
+      "grad_norm": 0.06596378237009048,
+      "learning_rate": 8.236194248165985e-07,
+      "loss": 0.19,
+      "step": 49193
+    },
+    {
+      "epoch": 3.985255994815295,
+      "grad_norm": 0.058588989078998566,
+      "learning_rate": 8.191187722219722e-07,
+      "loss": 0.2075,
+      "step": 49194
+    },
+    {
+      "epoch": 3.9853370058327933,
+      "grad_norm": 0.08515679091215134,
+      "learning_rate": 8.14618119627346e-07,
+      "loss": 0.2337,
+      "step": 49195
+    },
+    {
+      "epoch": 3.9854180168502915,
+      "grad_norm": 0.08090581744909286,
+      "learning_rate": 8.101174670327197e-07,
+      "loss": 0.2216,
+      "step": 49196
+    },
+    {
+      "epoch": 3.9854990278677898,
+      "grad_norm": 0.06988976150751114,
+      "learning_rate": 8.056168144380936e-07,
+      "loss": 0.2209,
+      "step": 49197
+    },
+    {
+      "epoch": 3.9855800388852884,
+      "grad_norm": 0.08122329413890839,
+      "learning_rate": 8.011161618434673e-07,
+      "loss": 0.2666,
+      "step": 49198
+    },
+    {
+      "epoch": 3.9856610499027867,
+      "grad_norm": 0.06371399760246277,
+      "learning_rate": 7.966155092488411e-07,
+      "loss": 0.2513,
+      "step": 49199
+    },
+    {
+      "epoch": 3.985742060920285,
+      "grad_norm": 0.07604499161243439,
+      "learning_rate": 7.92114856654215e-07,
+      "loss": 0.2035,
+      "step": 49200
+    },
+    {
+      "epoch": 3.9858230719377836,
+      "grad_norm": 0.07374300062656403,
+      "learning_rate": 7.876142040595887e-07,
+      "loss": 0.2181,
+      "step": 49201
+    },
+    {
+      "epoch": 3.985904082955282,
+      "grad_norm": 0.07626064866781235,
+      "learning_rate": 7.831135514649625e-07,
+      "loss": 0.239,
+      "step": 49202
+    },
+    {
+      "epoch": 3.98598509397278,
+      "grad_norm": 0.05903629586100578,
+      "learning_rate": 7.786128988703362e-07,
+      "loss": 0.2137,
+      "step": 49203
+    },
+    {
+      "epoch": 3.986066104990279,
+      "grad_norm": 0.07717857509851456,
+      "learning_rate": 7.7411224627571e-07,
+      "loss": 0.2144,
+      "step": 49204
+    },
+    {
+      "epoch": 3.986147116007777,
+      "grad_norm": 0.07685103267431259,
+      "learning_rate": 7.696115936810838e-07,
+      "loss": 0.2063,
+      "step": 49205
+    },
+    {
+      "epoch": 3.9862281270252753,
+      "grad_norm": 0.06778260320425034,
+      "learning_rate": 7.651109410864576e-07,
+      "loss": 0.2426,
+      "step": 49206
+    },
+    {
+      "epoch": 3.986309138042774,
+      "grad_norm": 0.06964470446109772,
+      "learning_rate": 7.606102884918314e-07,
+      "loss": 0.2167,
+      "step": 49207
+    },
+    {
+      "epoch": 3.9863901490602722,
+      "grad_norm": 0.0588604211807251,
+      "learning_rate": 7.561096358972051e-07,
+      "loss": 0.2095,
+      "step": 49208
+    },
+    {
+      "epoch": 3.9864711600777705,
+      "grad_norm": 0.06138395890593529,
+      "learning_rate": 7.516089833025789e-07,
+      "loss": 0.2115,
+      "step": 49209
+    },
+    {
+      "epoch": 3.986552171095269,
+      "grad_norm": 0.07522926479578018,
+      "learning_rate": 7.471083307079527e-07,
+      "loss": 0.228,
+      "step": 49210
+    },
+    {
+      "epoch": 3.9866331821127674,
+      "grad_norm": 0.06714662909507751,
+      "learning_rate": 7.426076781133264e-07,
+      "loss": 0.2457,
+      "step": 49211
+    },
+    {
+      "epoch": 3.9867141931302656,
+      "grad_norm": 0.06937988847494125,
+      "learning_rate": 7.381070255187003e-07,
+      "loss": 0.2112,
+      "step": 49212
+    },
+    {
+      "epoch": 3.9867952041477643,
+      "grad_norm": 0.06987298280000687,
+      "learning_rate": 7.336063729240741e-07,
+      "loss": 0.2207,
+      "step": 49213
+    },
+    {
+      "epoch": 3.9868762151652626,
+      "grad_norm": 0.07640345394611359,
+      "learning_rate": 7.291057203294478e-07,
+      "loss": 0.2241,
+      "step": 49214
+    },
+    {
+      "epoch": 3.986957226182761,
+      "grad_norm": 0.06703965365886688,
+      "learning_rate": 7.246050677348216e-07,
+      "loss": 0.2068,
+      "step": 49215
+    },
+    {
+      "epoch": 3.987038237200259,
+      "grad_norm": 0.06497196108102798,
+      "learning_rate": 7.201044151401954e-07,
+      "loss": 0.2513,
+      "step": 49216
+    },
+    {
+      "epoch": 3.9871192482177578,
+      "grad_norm": 0.08134781569242477,
+      "learning_rate": 7.156037625455691e-07,
+      "loss": 0.2198,
+      "step": 49217
+    },
+    {
+      "epoch": 3.987200259235256,
+      "grad_norm": 0.07386305928230286,
+      "learning_rate": 7.111031099509429e-07,
+      "loss": 0.2293,
+      "step": 49218
+    },
+    {
+      "epoch": 3.9872812702527543,
+      "grad_norm": 0.06610637903213501,
+      "learning_rate": 7.066024573563166e-07,
+      "loss": 0.2115,
+      "step": 49219
+    },
+    {
+      "epoch": 3.9873622812702525,
+      "grad_norm": 0.073157899081707,
+      "learning_rate": 7.021018047616905e-07,
+      "loss": 0.2124,
+      "step": 49220
+    },
+    {
+      "epoch": 3.987443292287751,
+      "grad_norm": 0.0644218772649765,
+      "learning_rate": 6.976011521670643e-07,
+      "loss": 0.2333,
+      "step": 49221
+    },
+    {
+      "epoch": 3.9875243033052494,
+      "grad_norm": 0.07077537477016449,
+      "learning_rate": 6.93100499572438e-07,
+      "loss": 0.1877,
+      "step": 49222
+    },
+    {
+      "epoch": 3.9876053143227477,
+      "grad_norm": 0.0648820549249649,
+      "learning_rate": 6.885998469778118e-07,
+      "loss": 0.2393,
+      "step": 49223
+    },
+    {
+      "epoch": 3.9876863253402464,
+      "grad_norm": 0.06722790747880936,
+      "learning_rate": 6.840991943831856e-07,
+      "loss": 0.2105,
+      "step": 49224
+    },
+    {
+      "epoch": 3.9877673363577446,
+      "grad_norm": 0.07644420117139816,
+      "learning_rate": 6.795985417885593e-07,
+      "loss": 0.2234,
+      "step": 49225
+    },
+    {
+      "epoch": 3.987848347375243,
+      "grad_norm": 0.07702749967575073,
+      "learning_rate": 6.750978891939331e-07,
+      "loss": 0.2136,
+      "step": 49226
+    },
+    {
+      "epoch": 3.9879293583927415,
+      "grad_norm": 0.06729872524738312,
+      "learning_rate": 6.70597236599307e-07,
+      "loss": 0.191,
+      "step": 49227
+    },
+    {
+      "epoch": 3.98801036941024,
+      "grad_norm": 0.07493147253990173,
+      "learning_rate": 6.660965840046808e-07,
+      "loss": 0.2039,
+      "step": 49228
+    },
+    {
+      "epoch": 3.988091380427738,
+      "grad_norm": 0.06625507026910782,
+      "learning_rate": 6.615959314100545e-07,
+      "loss": 0.2119,
+      "step": 49229
+    },
+    {
+      "epoch": 3.9881723914452367,
+      "grad_norm": 0.052606746554374695,
+      "learning_rate": 6.570952788154282e-07,
+      "loss": 0.2399,
+      "step": 49230
+    },
+    {
+      "epoch": 3.988253402462735,
+      "grad_norm": 0.06827935576438904,
+      "learning_rate": 6.525946262208021e-07,
+      "loss": 0.2288,
+      "step": 49231
+    },
+    {
+      "epoch": 3.988334413480233,
+      "grad_norm": 0.08367155492305756,
+      "learning_rate": 6.480939736261758e-07,
+      "loss": 0.2511,
+      "step": 49232
+    },
+    {
+      "epoch": 3.988415424497732,
+      "grad_norm": 0.08619492501020432,
+      "learning_rate": 6.435933210315496e-07,
+      "loss": 0.2279,
+      "step": 49233
+    },
+    {
+      "epoch": 3.98849643551523,
+      "grad_norm": 0.08508004993200302,
+      "learning_rate": 6.390926684369234e-07,
+      "loss": 0.2145,
+      "step": 49234
+    },
+    {
+      "epoch": 3.9885774465327284,
+      "grad_norm": 0.07652024179697037,
+      "learning_rate": 6.345920158422972e-07,
+      "loss": 0.3088,
+      "step": 49235
+    },
+    {
+      "epoch": 3.988658457550227,
+      "grad_norm": 0.09197687357664108,
+      "learning_rate": 6.30091363247671e-07,
+      "loss": 0.2427,
+      "step": 49236
+    },
+    {
+      "epoch": 3.9887394685677253,
+      "grad_norm": 0.05859651789069176,
+      "learning_rate": 6.255907106530447e-07,
+      "loss": 0.2136,
+      "step": 49237
+    },
+    {
+      "epoch": 3.9888204795852236,
+      "grad_norm": 0.08267036080360413,
+      "learning_rate": 6.210900580584185e-07,
+      "loss": 0.2678,
+      "step": 49238
+    },
+    {
+      "epoch": 3.988901490602722,
+      "grad_norm": 0.05907399207353592,
+      "learning_rate": 6.165894054637923e-07,
+      "loss": 0.2066,
+      "step": 49239
+    },
+    {
+      "epoch": 3.9889825016202205,
+      "grad_norm": 0.07600349932909012,
+      "learning_rate": 6.12088752869166e-07,
+      "loss": 0.2264,
+      "step": 49240
+    },
+    {
+      "epoch": 3.9890635126377187,
+      "grad_norm": 0.08025678247213364,
+      "learning_rate": 6.075881002745398e-07,
+      "loss": 0.2387,
+      "step": 49241
+    },
+    {
+      "epoch": 3.989144523655217,
+      "grad_norm": 0.06902442127466202,
+      "learning_rate": 6.030874476799137e-07,
+      "loss": 0.2211,
+      "step": 49242
+    },
+    {
+      "epoch": 3.9892255346727152,
+      "grad_norm": 0.08164741843938828,
+      "learning_rate": 5.985867950852874e-07,
+      "loss": 0.2501,
+      "step": 49243
+    },
+    {
+      "epoch": 3.989306545690214,
+      "grad_norm": 0.058244526386260986,
+      "learning_rate": 5.940861424906612e-07,
+      "loss": 0.2106,
+      "step": 49244
+    },
+    {
+      "epoch": 3.989387556707712,
+      "grad_norm": 0.06369130313396454,
+      "learning_rate": 5.89585489896035e-07,
+      "loss": 0.1867,
+      "step": 49245
+    },
+    {
+      "epoch": 3.9894685677252104,
+      "grad_norm": 0.05307641252875328,
+      "learning_rate": 5.850848373014087e-07,
+      "loss": 0.1732,
+      "step": 49246
+    },
+    {
+      "epoch": 3.989549578742709,
+      "grad_norm": 0.06525785475969315,
+      "learning_rate": 5.805841847067825e-07,
+      "loss": 0.2048,
+      "step": 49247
+    },
+    {
+      "epoch": 3.9896305897602073,
+      "grad_norm": 0.06935491412878036,
+      "learning_rate": 5.760835321121562e-07,
+      "loss": 0.1843,
+      "step": 49248
+    },
+    {
+      "epoch": 3.9897116007777056,
+      "grad_norm": 0.058017581701278687,
+      "learning_rate": 5.715828795175301e-07,
+      "loss": 0.21,
+      "step": 49249
+    },
+    {
+      "epoch": 3.9897926117952043,
+      "grad_norm": 0.0743686705827713,
+      "learning_rate": 5.670822269229039e-07,
+      "loss": 0.2,
+      "step": 49250
+    },
+    {
+      "epoch": 3.9898736228127025,
+      "grad_norm": 0.07183623313903809,
+      "learning_rate": 5.625815743282776e-07,
+      "loss": 0.2428,
+      "step": 49251
+    },
+    {
+      "epoch": 3.9899546338302008,
+      "grad_norm": 0.06011229380965233,
+      "learning_rate": 5.580809217336515e-07,
+      "loss": 0.2315,
+      "step": 49252
+    },
+    {
+      "epoch": 3.9900356448476995,
+      "grad_norm": 0.0707809180021286,
+      "learning_rate": 5.535802691390252e-07,
+      "loss": 0.2014,
+      "step": 49253
+    },
+    {
+      "epoch": 3.9901166558651977,
+      "grad_norm": 0.06794430315494537,
+      "learning_rate": 5.490796165443989e-07,
+      "loss": 0.2359,
+      "step": 49254
+    },
+    {
+      "epoch": 3.990197666882696,
+      "grad_norm": 0.07199983298778534,
+      "learning_rate": 5.445789639497727e-07,
+      "loss": 0.2156,
+      "step": 49255
+    },
+    {
+      "epoch": 3.9902786779001946,
+      "grad_norm": 0.10244887322187424,
+      "learning_rate": 5.400783113551465e-07,
+      "loss": 0.2666,
+      "step": 49256
+    },
+    {
+      "epoch": 3.990359688917693,
+      "grad_norm": 0.058576490730047226,
+      "learning_rate": 5.355776587605203e-07,
+      "loss": 0.197,
+      "step": 49257
+    },
+    {
+      "epoch": 3.990440699935191,
+      "grad_norm": 0.07232537120580673,
+      "learning_rate": 5.310770061658941e-07,
+      "loss": 0.203,
+      "step": 49258
+    },
+    {
+      "epoch": 3.99052171095269,
+      "grad_norm": 0.07895112782716751,
+      "learning_rate": 5.265763535712678e-07,
+      "loss": 0.1973,
+      "step": 49259
+    },
+    {
+      "epoch": 3.990602721970188,
+      "grad_norm": 0.0817318707704544,
+      "learning_rate": 5.220757009766417e-07,
+      "loss": 0.2765,
+      "step": 49260
+    },
+    {
+      "epoch": 3.9906837329876863,
+      "grad_norm": 0.06556425988674164,
+      "learning_rate": 5.175750483820154e-07,
+      "loss": 0.2576,
+      "step": 49261
+    },
+    {
+      "epoch": 3.9907647440051845,
+      "grad_norm": 0.07194989919662476,
+      "learning_rate": 5.130743957873892e-07,
+      "loss": 0.2051,
+      "step": 49262
+    },
+    {
+      "epoch": 3.9908457550226832,
+      "grad_norm": 0.07065384835004807,
+      "learning_rate": 5.08573743192763e-07,
+      "loss": 0.2483,
+      "step": 49263
+    },
+    {
+      "epoch": 3.9909267660401815,
+      "grad_norm": 0.07403761893510818,
+      "learning_rate": 5.040730905981367e-07,
+      "loss": 0.2408,
+      "step": 49264
+    },
+    {
+      "epoch": 3.9910077770576797,
+      "grad_norm": 0.06701704114675522,
+      "learning_rate": 4.995724380035106e-07,
+      "loss": 0.1916,
+      "step": 49265
+    },
+    {
+      "epoch": 3.991088788075178,
+      "grad_norm": 0.07365620136260986,
+      "learning_rate": 4.950717854088843e-07,
+      "loss": 0.2207,
+      "step": 49266
+    },
+    {
+      "epoch": 3.9911697990926767,
+      "grad_norm": 0.07243810594081879,
+      "learning_rate": 4.905711328142581e-07,
+      "loss": 0.1908,
+      "step": 49267
+    },
+    {
+      "epoch": 3.991250810110175,
+      "grad_norm": 0.06687340140342712,
+      "learning_rate": 4.860704802196319e-07,
+      "loss": 0.2387,
+      "step": 49268
+    },
+    {
+      "epoch": 3.991331821127673,
+      "grad_norm": 0.07321549952030182,
+      "learning_rate": 4.815698276250056e-07,
+      "loss": 0.2107,
+      "step": 49269
+    },
+    {
+      "epoch": 3.991412832145172,
+      "grad_norm": 0.06943608075380325,
+      "learning_rate": 4.770691750303794e-07,
+      "loss": 0.2445,
+      "step": 49270
+    },
+    {
+      "epoch": 3.99149384316267,
+      "grad_norm": 0.06473980844020844,
+      "learning_rate": 4.7256852243575323e-07,
+      "loss": 0.1774,
+      "step": 49271
+    },
+    {
+      "epoch": 3.9915748541801683,
+      "grad_norm": 0.08363772183656693,
+      "learning_rate": 4.68067869841127e-07,
+      "loss": 0.2475,
+      "step": 49272
+    },
+    {
+      "epoch": 3.991655865197667,
+      "grad_norm": 0.08146770298480988,
+      "learning_rate": 4.6356721724650075e-07,
+      "loss": 0.2079,
+      "step": 49273
+    },
+    {
+      "epoch": 3.9917368762151653,
+      "grad_norm": 0.07280329614877701,
+      "learning_rate": 4.5906656465187457e-07,
+      "loss": 0.2295,
+      "step": 49274
+    },
+    {
+      "epoch": 3.9918178872326635,
+      "grad_norm": 0.08392112702131271,
+      "learning_rate": 4.5456591205724833e-07,
+      "loss": 0.2488,
+      "step": 49275
+    },
+    {
+      "epoch": 3.991898898250162,
+      "grad_norm": 0.06567014008760452,
+      "learning_rate": 4.500652594626221e-07,
+      "loss": 0.2126,
+      "step": 49276
+    },
+    {
+      "epoch": 3.9919799092676604,
+      "grad_norm": 0.06484349817037582,
+      "learning_rate": 4.4556460686799585e-07,
+      "loss": 0.1919,
+      "step": 49277
+    },
+    {
+      "epoch": 3.9920609202851587,
+      "grad_norm": 0.07832783460617065,
+      "learning_rate": 4.4106395427336967e-07,
+      "loss": 0.2157,
+      "step": 49278
+    },
+    {
+      "epoch": 3.9921419313026574,
+      "grad_norm": 0.0768279954791069,
+      "learning_rate": 4.3656330167874343e-07,
+      "loss": 0.2463,
+      "step": 49279
+    },
+    {
+      "epoch": 3.9922229423201556,
+      "grad_norm": 0.06600246578454971,
+      "learning_rate": 4.320626490841172e-07,
+      "loss": 0.2531,
+      "step": 49280
+    },
+    {
+      "epoch": 3.992303953337654,
+      "grad_norm": 0.0632348507642746,
+      "learning_rate": 4.27561996489491e-07,
+      "loss": 0.2087,
+      "step": 49281
+    },
+    {
+      "epoch": 3.9923849643551526,
+      "grad_norm": 0.10359801352024078,
+      "learning_rate": 4.2306134389486477e-07,
+      "loss": 0.1958,
+      "step": 49282
+    },
+    {
+      "epoch": 3.992465975372651,
+      "grad_norm": 0.0757150650024414,
+      "learning_rate": 4.1856069130023853e-07,
+      "loss": 0.2425,
+      "step": 49283
+    },
+    {
+      "epoch": 3.992546986390149,
+      "grad_norm": 0.07178405672311783,
+      "learning_rate": 4.140600387056123e-07,
+      "loss": 0.2043,
+      "step": 49284
+    },
+    {
+      "epoch": 3.9926279974076473,
+      "grad_norm": 0.09844893217086792,
+      "learning_rate": 4.095593861109861e-07,
+      "loss": 0.2262,
+      "step": 49285
+    },
+    {
+      "epoch": 3.992709008425146,
+      "grad_norm": 0.07299178838729858,
+      "learning_rate": 4.0505873351635987e-07,
+      "loss": 0.2082,
+      "step": 49286
+    },
+    {
+      "epoch": 3.9927900194426442,
+      "grad_norm": 0.08776606619358063,
+      "learning_rate": 4.0055808092173363e-07,
+      "loss": 0.2229,
+      "step": 49287
+    },
+    {
+      "epoch": 3.9928710304601425,
+      "grad_norm": 0.06883752346038818,
+      "learning_rate": 3.960574283271075e-07,
+      "loss": 0.1953,
+      "step": 49288
+    },
+    {
+      "epoch": 3.9929520414776407,
+      "grad_norm": 0.06461341679096222,
+      "learning_rate": 3.9155677573248126e-07,
+      "loss": 0.2013,
+      "step": 49289
+    },
+    {
+      "epoch": 3.9930330524951394,
+      "grad_norm": 0.058432161808013916,
+      "learning_rate": 3.87056123137855e-07,
+      "loss": 0.2223,
+      "step": 49290
+    },
+    {
+      "epoch": 3.9931140635126376,
+      "grad_norm": 0.06838857382535934,
+      "learning_rate": 3.825554705432288e-07,
+      "loss": 0.2357,
+      "step": 49291
+    },
+    {
+      "epoch": 3.993195074530136,
+      "grad_norm": 0.07145708054304123,
+      "learning_rate": 3.7805481794860255e-07,
+      "loss": 0.2288,
+      "step": 49292
+    },
+    {
+      "epoch": 3.9932760855476346,
+      "grad_norm": 0.06667095422744751,
+      "learning_rate": 3.7355416535397637e-07,
+      "loss": 0.22,
+      "step": 49293
+    },
+    {
+      "epoch": 3.993357096565133,
+      "grad_norm": 0.06883816421031952,
+      "learning_rate": 3.6905351275935013e-07,
+      "loss": 0.2138,
+      "step": 49294
+    },
+    {
+      "epoch": 3.993438107582631,
+      "grad_norm": 0.07182978838682175,
+      "learning_rate": 3.645528601647239e-07,
+      "loss": 0.2382,
+      "step": 49295
+    },
+    {
+      "epoch": 3.9935191186001298,
+      "grad_norm": 0.08456697314977646,
+      "learning_rate": 3.600522075700977e-07,
+      "loss": 0.2418,
+      "step": 49296
+    },
+    {
+      "epoch": 3.993600129617628,
+      "grad_norm": 0.06518872827291489,
+      "learning_rate": 3.5555155497547147e-07,
+      "loss": 0.2274,
+      "step": 49297
+    },
+    {
+      "epoch": 3.9936811406351262,
+      "grad_norm": 0.07191832363605499,
+      "learning_rate": 3.5105090238084523e-07,
+      "loss": 0.2078,
+      "step": 49298
+    },
+    {
+      "epoch": 3.993762151652625,
+      "grad_norm": 0.07570493221282959,
+      "learning_rate": 3.46550249786219e-07,
+      "loss": 0.1972,
+      "step": 49299
+    },
+    {
+      "epoch": 3.993843162670123,
+      "grad_norm": 0.0672534927725792,
+      "learning_rate": 3.420495971915928e-07,
+      "loss": 0.2069,
+      "step": 49300
+    },
+    {
+      "epoch": 3.9939241736876214,
+      "grad_norm": 0.08297023177146912,
+      "learning_rate": 3.3754894459696657e-07,
+      "loss": 0.2224,
+      "step": 49301
+    },
+    {
+      "epoch": 3.99400518470512,
+      "grad_norm": 0.0740421712398529,
+      "learning_rate": 3.330482920023404e-07,
+      "loss": 0.2122,
+      "step": 49302
+    },
+    {
+      "epoch": 3.9940861957226184,
+      "grad_norm": 0.07864964753389359,
+      "learning_rate": 3.285476394077141e-07,
+      "loss": 0.2519,
+      "step": 49303
+    },
+    {
+      "epoch": 3.9941672067401166,
+      "grad_norm": 0.07417288422584534,
+      "learning_rate": 3.240469868130879e-07,
+      "loss": 0.2207,
+      "step": 49304
+    },
+    {
+      "epoch": 3.9942482177576153,
+      "grad_norm": 0.06941584497690201,
+      "learning_rate": 3.195463342184617e-07,
+      "loss": 0.2537,
+      "step": 49305
+    },
+    {
+      "epoch": 3.9943292287751135,
+      "grad_norm": 0.07486558705568314,
+      "learning_rate": 3.150456816238355e-07,
+      "loss": 0.2112,
+      "step": 49306
+    },
+    {
+      "epoch": 3.994410239792612,
+      "grad_norm": 0.06437824666500092,
+      "learning_rate": 3.1054502902920925e-07,
+      "loss": 0.2062,
+      "step": 49307
+    },
+    {
+      "epoch": 3.99449125081011,
+      "grad_norm": 0.07233487814664841,
+      "learning_rate": 3.06044376434583e-07,
+      "loss": 0.2241,
+      "step": 49308
+    },
+    {
+      "epoch": 3.9945722618276087,
+      "grad_norm": 0.06914579123258591,
+      "learning_rate": 3.015437238399568e-07,
+      "loss": 0.1937,
+      "step": 49309
+    },
+    {
+      "epoch": 3.994653272845107,
+      "grad_norm": 0.07031723856925964,
+      "learning_rate": 2.970430712453306e-07,
+      "loss": 0.2298,
+      "step": 49310
+    },
+    {
+      "epoch": 3.994734283862605,
+      "grad_norm": 0.0636644959449768,
+      "learning_rate": 2.9254241865070435e-07,
+      "loss": 0.2067,
+      "step": 49311
+    },
+    {
+      "epoch": 3.9948152948801035,
+      "grad_norm": 0.08062580972909927,
+      "learning_rate": 2.880417660560781e-07,
+      "loss": 0.2336,
+      "step": 49312
+    },
+    {
+      "epoch": 3.994896305897602,
+      "grad_norm": 0.07863558083772659,
+      "learning_rate": 2.8354111346145193e-07,
+      "loss": 0.202,
+      "step": 49313
+    },
+    {
+      "epoch": 3.9949773169151004,
+      "grad_norm": 0.0746251717209816,
+      "learning_rate": 2.7904046086682574e-07,
+      "loss": 0.2108,
+      "step": 49314
+    },
+    {
+      "epoch": 3.9950583279325986,
+      "grad_norm": 0.06599914282560349,
+      "learning_rate": 2.7453980827219945e-07,
+      "loss": 0.203,
+      "step": 49315
+    },
+    {
+      "epoch": 3.9951393389500973,
+      "grad_norm": 0.0633784607052803,
+      "learning_rate": 2.7003915567757327e-07,
+      "loss": 0.214,
+      "step": 49316
+    },
+    {
+      "epoch": 3.9952203499675956,
+      "grad_norm": 0.07074976712465286,
+      "learning_rate": 2.6553850308294703e-07,
+      "loss": 0.2186,
+      "step": 49317
+    },
+    {
+      "epoch": 3.995301360985094,
+      "grad_norm": 0.08301674574613571,
+      "learning_rate": 2.6103785048832084e-07,
+      "loss": 0.2604,
+      "step": 49318
+    },
+    {
+      "epoch": 3.9953823720025925,
+      "grad_norm": 0.07280449569225311,
+      "learning_rate": 2.565371978936946e-07,
+      "loss": 0.2365,
+      "step": 49319
+    },
+    {
+      "epoch": 3.9954633830200907,
+      "grad_norm": 0.06467054784297943,
+      "learning_rate": 2.5203654529906837e-07,
+      "loss": 0.2106,
+      "step": 49320
+    },
+    {
+      "epoch": 3.995544394037589,
+      "grad_norm": 0.07908143103122711,
+      "learning_rate": 2.4753589270444213e-07,
+      "loss": 0.2395,
+      "step": 49321
+    },
+    {
+      "epoch": 3.9956254050550877,
+      "grad_norm": 0.06821177899837494,
+      "learning_rate": 2.4303524010981594e-07,
+      "loss": 0.2398,
+      "step": 49322
+    },
+    {
+      "epoch": 3.995706416072586,
+      "grad_norm": 0.07683097571134567,
+      "learning_rate": 2.385345875151897e-07,
+      "loss": 0.2641,
+      "step": 49323
+    },
+    {
+      "epoch": 3.995787427090084,
+      "grad_norm": 0.0672578513622284,
+      "learning_rate": 2.340339349205635e-07,
+      "loss": 0.201,
+      "step": 49324
+    },
+    {
+      "epoch": 3.995868438107583,
+      "grad_norm": 0.10541965812444687,
+      "learning_rate": 2.2953328232593728e-07,
+      "loss": 0.1953,
+      "step": 49325
+    },
+    {
+      "epoch": 3.995949449125081,
+      "grad_norm": 0.07959267497062683,
+      "learning_rate": 2.2503262973131105e-07,
+      "loss": 0.2044,
+      "step": 49326
+    },
+    {
+      "epoch": 3.9960304601425793,
+      "grad_norm": 0.07636445015668869,
+      "learning_rate": 2.2053197713668483e-07,
+      "loss": 0.2205,
+      "step": 49327
+    },
+    {
+      "epoch": 3.996111471160078,
+      "grad_norm": 0.07331163436174393,
+      "learning_rate": 2.160313245420586e-07,
+      "loss": 0.2267,
+      "step": 49328
+    },
+    {
+      "epoch": 3.9961924821775763,
+      "grad_norm": 0.06154733523726463,
+      "learning_rate": 2.1153067194743239e-07,
+      "loss": 0.2145,
+      "step": 49329
+    },
+    {
+      "epoch": 3.9962734931950745,
+      "grad_norm": 0.07518443465232849,
+      "learning_rate": 2.0703001935280615e-07,
+      "loss": 0.2381,
+      "step": 49330
+    },
+    {
+      "epoch": 3.9963545042125728,
+      "grad_norm": 0.06782988458871841,
+      "learning_rate": 2.0252936675817994e-07,
+      "loss": 0.2327,
+      "step": 49331
+    },
+    {
+      "epoch": 3.9964355152300715,
+      "grad_norm": 0.08090611547231674,
+      "learning_rate": 1.9802871416355375e-07,
+      "loss": 0.2294,
+      "step": 49332
+    },
+    {
+      "epoch": 3.9965165262475697,
+      "grad_norm": 0.05896861106157303,
+      "learning_rate": 1.935280615689275e-07,
+      "loss": 0.2398,
+      "step": 49333
+    },
+    {
+      "epoch": 3.996597537265068,
+      "grad_norm": 0.06596443057060242,
+      "learning_rate": 1.8902740897430128e-07,
+      "loss": 0.2233,
+      "step": 49334
+    },
+    {
+      "epoch": 3.996678548282566,
+      "grad_norm": 0.0906447172164917,
+      "learning_rate": 1.8452675637967506e-07,
+      "loss": 0.25,
+      "step": 49335
+    },
+    {
+      "epoch": 3.996759559300065,
+      "grad_norm": 0.07093532383441925,
+      "learning_rate": 1.8002610378504885e-07,
+      "loss": 0.2599,
+      "step": 49336
+    },
+    {
+      "epoch": 3.996840570317563,
+      "grad_norm": 0.07621223479509354,
+      "learning_rate": 1.7552545119042261e-07,
+      "loss": 0.2288,
+      "step": 49337
+    },
+    {
+      "epoch": 3.9969215813350614,
+      "grad_norm": 0.06504885852336884,
+      "learning_rate": 1.710247985957964e-07,
+      "loss": 0.1997,
+      "step": 49338
+    },
+    {
+      "epoch": 3.99700259235256,
+      "grad_norm": 0.07198863476514816,
+      "learning_rate": 1.665241460011702e-07,
+      "loss": 0.2141,
+      "step": 49339
+    },
+    {
+      "epoch": 3.9970836033700583,
+      "grad_norm": 0.06278213858604431,
+      "learning_rate": 1.6202349340654395e-07,
+      "loss": 0.219,
+      "step": 49340
+    },
+    {
+      "epoch": 3.9971646143875565,
+      "grad_norm": 0.06734520196914673,
+      "learning_rate": 1.5752284081191774e-07,
+      "loss": 0.2033,
+      "step": 49341
+    },
+    {
+      "epoch": 3.9972456254050552,
+      "grad_norm": 0.05727878585457802,
+      "learning_rate": 1.530221882172915e-07,
+      "loss": 0.2046,
+      "step": 49342
+    },
+    {
+      "epoch": 3.9973266364225535,
+      "grad_norm": 0.08659632503986359,
+      "learning_rate": 1.485215356226653e-07,
+      "loss": 0.2383,
+      "step": 49343
+    },
+    {
+      "epoch": 3.9974076474400517,
+      "grad_norm": 0.06606918573379517,
+      "learning_rate": 1.4402088302803906e-07,
+      "loss": 0.1683,
+      "step": 49344
+    },
+    {
+      "epoch": 3.9974886584575504,
+      "grad_norm": 0.07601425051689148,
+      "learning_rate": 1.3952023043341287e-07,
+      "loss": 0.2367,
+      "step": 49345
+    },
+    {
+      "epoch": 3.9975696694750487,
+      "grad_norm": 0.07500436156988144,
+      "learning_rate": 1.3501957783878663e-07,
+      "loss": 0.2605,
+      "step": 49346
+    },
+    {
+      "epoch": 3.997650680492547,
+      "grad_norm": 0.07162158936262131,
+      "learning_rate": 1.3051892524416042e-07,
+      "loss": 0.2128,
+      "step": 49347
+    },
+    {
+      "epoch": 3.9977316915100456,
+      "grad_norm": 0.07516414672136307,
+      "learning_rate": 1.2601827264953418e-07,
+      "loss": 0.1984,
+      "step": 49348
+    },
+    {
+      "epoch": 3.997812702527544,
+      "grad_norm": 0.07357397675514221,
+      "learning_rate": 1.2151762005490797e-07,
+      "loss": 0.2305,
+      "step": 49349
+    },
+    {
+      "epoch": 3.997893713545042,
+      "grad_norm": 0.07578955590724945,
+      "learning_rate": 1.1701696746028175e-07,
+      "loss": 0.2729,
+      "step": 49350
+    },
+    {
+      "epoch": 3.9979747245625408,
+      "grad_norm": 0.08693145960569382,
+      "learning_rate": 1.1251631486565552e-07,
+      "loss": 0.2081,
+      "step": 49351
+    },
+    {
+      "epoch": 3.998055735580039,
+      "grad_norm": 0.06208210065960884,
+      "learning_rate": 1.080156622710293e-07,
+      "loss": 0.1848,
+      "step": 49352
+    },
+    {
+      "epoch": 3.9981367465975373,
+      "grad_norm": 0.06679573655128479,
+      "learning_rate": 1.0351500967640307e-07,
+      "loss": 0.2396,
+      "step": 49353
+    },
+    {
+      "epoch": 3.9982177576150355,
+      "grad_norm": 0.07043427228927612,
+      "learning_rate": 9.901435708177688e-08,
+      "loss": 0.2277,
+      "step": 49354
+    },
+    {
+      "epoch": 3.9982987686325338,
+      "grad_norm": 0.06167418509721756,
+      "learning_rate": 9.451370448715064e-08,
+      "loss": 0.2084,
+      "step": 49355
+    },
+    {
+      "epoch": 3.9983797796500324,
+      "grad_norm": 0.058673106133937836,
+      "learning_rate": 9.001305189252443e-08,
+      "loss": 0.2388,
+      "step": 49356
+    },
+    {
+      "epoch": 3.9984607906675307,
+      "grad_norm": 0.0721500888466835,
+      "learning_rate": 8.55123992978982e-08,
+      "loss": 0.2257,
+      "step": 49357
+    },
+    {
+      "epoch": 3.998541801685029,
+      "grad_norm": 0.07805679738521576,
+      "learning_rate": 8.101174670327198e-08,
+      "loss": 0.285,
+      "step": 49358
+    },
+    {
+      "epoch": 3.9986228127025276,
+      "grad_norm": 0.0756671130657196,
+      "learning_rate": 7.651109410864575e-08,
+      "loss": 0.2225,
+      "step": 49359
+    },
+    {
+      "epoch": 3.998703823720026,
+      "grad_norm": 0.0644574761390686,
+      "learning_rate": 7.201044151401953e-08,
+      "loss": 0.2034,
+      "step": 49360
+    },
+    {
+      "epoch": 3.998784834737524,
+      "grad_norm": 0.09972898662090302,
+      "learning_rate": 6.750978891939332e-08,
+      "loss": 0.2121,
+      "step": 49361
+    },
+    {
+      "epoch": 3.998865845755023,
+      "grad_norm": 0.07944905012845993,
+      "learning_rate": 6.300913632476709e-08,
+      "loss": 0.2141,
+      "step": 49362
+    },
+    {
+      "epoch": 3.998946856772521,
+      "grad_norm": 0.07724474370479584,
+      "learning_rate": 5.8508483730140874e-08,
+      "loss": 0.2399,
+      "step": 49363
+    },
+    {
+      "epoch": 3.9990278677900193,
+      "grad_norm": 0.06207426264882088,
+      "learning_rate": 5.400783113551465e-08,
+      "loss": 0.2202,
+      "step": 49364
+    },
+    {
+      "epoch": 3.999108878807518,
+      "grad_norm": 0.07973155379295349,
+      "learning_rate": 4.950717854088844e-08,
+      "loss": 0.2378,
+      "step": 49365
+    },
+    {
+      "epoch": 3.999189889825016,
+      "grad_norm": 0.07270248979330063,
+      "learning_rate": 4.500652594626221e-08,
+      "loss": 0.2186,
+      "step": 49366
+    },
+    {
+      "epoch": 3.9992709008425145,
+      "grad_norm": 0.06034340709447861,
+      "learning_rate": 4.050587335163599e-08,
+      "loss": 0.1782,
+      "step": 49367
+    },
+    {
+      "epoch": 3.999351911860013,
+      "grad_norm": 0.08102718740701675,
+      "learning_rate": 3.6005220757009764e-08,
+      "loss": 0.2278,
+      "step": 49368
+    },
+    {
+      "epoch": 3.9994329228775114,
+      "grad_norm": 0.060907378792762756,
+      "learning_rate": 3.1504568162383546e-08,
+      "loss": 0.2229,
+      "step": 49369
+    },
+    {
+      "epoch": 3.9995139338950096,
+      "grad_norm": 0.07012410461902618,
+      "learning_rate": 2.7003915567757325e-08,
+      "loss": 0.2241,
+      "step": 49370
+    },
+    {
+      "epoch": 3.9995949449125083,
+      "grad_norm": 0.0700206607580185,
+      "learning_rate": 2.2503262973131107e-08,
+      "loss": 0.2083,
+      "step": 49371
+    },
+    {
+      "epoch": 3.9996759559300066,
+      "grad_norm": 0.08111918717622757,
+      "learning_rate": 1.8002610378504882e-08,
+      "loss": 0.2562,
+      "step": 49372
+    },
+    {
+      "epoch": 3.999756966947505,
+      "grad_norm": 0.07391811162233353,
+      "learning_rate": 1.3501957783878662e-08,
+      "loss": 0.1937,
+      "step": 49373
+    },
+    {
+      "epoch": 3.9998379779650035,
+      "grad_norm": 0.05770977959036827,
+      "learning_rate": 9.001305189252441e-09,
+      "loss": 0.191,
+      "step": 49374
+    },
+    {
+      "epoch": 3.9999189889825018,
+      "grad_norm": 0.05946144461631775,
+      "learning_rate": 4.5006525946262205e-09,
+      "loss": 0.1792,
+      "step": 49375
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.07257899641990662,
+      "learning_rate": 0.0,
+      "loss": 0.2477,
+      "step": 49376
     }
   ],
   "logging_steps": 1,
@@ -318521,12 +345653,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 2.985306826203359e+19,
+  "total_flos": 3.2393504748092785e+19,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null