diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.250214364460414,
+  "epoch": 0.500428728920828,
   "eval_steps": 766,
-  "global_step": 766,
+  "global_step": 1532,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -5385,6 +5385,5376 @@
       "eval_samples_per_second": 5.164,
       "eval_steps_per_second": 2.582,
       "step": 766
+    },
+    {
+      "epoch": 0.25054101506675924,
+      "grad_norm": 0.4362111985683441,
+      "learning_rate": 8.664921401487023e-05,
+      "loss": 1.0782,
+      "step": 767
+    },
+    {
+      "epoch": 0.2508676656731044,
+      "grad_norm": 0.4964749813079834,
+      "learning_rate": 8.661370661824e-05,
+      "loss": 1.1945,
+      "step": 768
+    },
+    {
+      "epoch": 0.2511943162794496,
+      "grad_norm": 0.5608397722244263,
+      "learning_rate": 8.657815936302337e-05,
+      "loss": 1.2661,
+      "step": 769
+    },
+    {
+      "epoch": 0.2515209668857948,
+      "grad_norm": 0.5862159132957458,
+      "learning_rate": 8.654257228791795e-05,
+      "loss": 1.0976,
+      "step": 770
+    },
+    {
+      "epoch": 0.25184761749213996,
+      "grad_norm": 0.7066771984100342,
+      "learning_rate": 8.650694543166475e-05,
+      "loss": 1.274,
+      "step": 771
+    },
+    {
+      "epoch": 0.25217426809848514,
+      "grad_norm": 0.7534899711608887,
+      "learning_rate": 8.647127883304799e-05,
+      "loss": 1.1669,
+      "step": 772
+    },
+    {
+      "epoch": 0.2525009187048303,
+      "grad_norm": 0.9842721819877625,
+      "learning_rate": 8.643557253089525e-05,
+      "loss": 1.3533,
+      "step": 773
+    },
+    {
+      "epoch": 0.25282756931117556,
+      "grad_norm": 1.294683814048767,
+      "learning_rate": 8.639982656407729e-05,
+      "loss": 1.7016,
+      "step": 774
+    },
+    {
+      "epoch": 0.25315421991752074,
+      "grad_norm": 1.7022143602371216,
+      "learning_rate": 8.636404097150802e-05,
+      "loss": 1.9253,
+      "step": 775
+    },
+    {
+      "epoch": 0.2534808705238659,
+      "grad_norm": 0.19106809794902802,
+      "learning_rate": 8.632821579214456e-05,
+      "loss": 0.7292,
+      "step": 776
+    },
+    {
+      "epoch": 0.2538075211302111,
+      "grad_norm": 0.2291499525308609,
+      "learning_rate": 8.629235106498708e-05,
+      "loss": 0.8147,
+      "step": 777
+    },
+    {
+      "epoch": 0.2541341717365563,
+      "grad_norm": 0.23071865737438202,
+      "learning_rate": 8.625644682907879e-05,
+      "loss": 0.7612,
+      "step": 778
+    },
+    {
+      "epoch": 0.25446082234290146,
+      "grad_norm": 0.24651503562927246,
+      "learning_rate": 8.622050312350594e-05,
+      "loss": 0.874,
+      "step": 779
+    },
+    {
+      "epoch": 0.25478747294924664,
+      "grad_norm": 0.2605278789997101,
+      "learning_rate": 8.618451998739774e-05,
+      "loss": 0.8279,
+      "step": 780
+    },
+    {
+      "epoch": 0.2551141235555919,
+      "grad_norm": 0.2556253969669342,
+      "learning_rate": 8.614849745992632e-05,
+      "loss": 0.8052,
+      "step": 781
+    },
+    {
+      "epoch": 0.25544077416193706,
+      "grad_norm": 0.28385311365127563,
+      "learning_rate": 8.611243558030668e-05,
+      "loss": 0.8039,
+      "step": 782
+    },
+    {
+      "epoch": 0.25576742476828224,
+      "grad_norm": 0.303835928440094,
+      "learning_rate": 8.60763343877967e-05,
+      "loss": 0.9079,
+      "step": 783
+    },
+    {
+      "epoch": 0.2560940753746274,
+      "grad_norm": 0.31251901388168335,
+      "learning_rate": 8.604019392169702e-05,
+      "loss": 0.8428,
+      "step": 784
+    },
+    {
+      "epoch": 0.2564207259809726,
+      "grad_norm": 0.335475891828537,
+      "learning_rate": 8.600401422135104e-05,
+      "loss": 0.9193,
+      "step": 785
+    },
+    {
+      "epoch": 0.2567473765873178,
+      "grad_norm": 0.3252756595611572,
+      "learning_rate": 8.596779532614488e-05,
+      "loss": 0.8837,
+      "step": 786
+    },
+    {
+      "epoch": 0.25707402719366296,
+      "grad_norm": 0.3637792766094208,
+      "learning_rate": 8.593153727550732e-05,
+      "loss": 0.9312,
+      "step": 787
+    },
+    {
+      "epoch": 0.2574006778000082,
+      "grad_norm": 0.4200628399848938,
+      "learning_rate": 8.589524010890977e-05,
+      "loss": 1.0147,
+      "step": 788
+    },
+    {
+      "epoch": 0.2577273284063534,
+      "grad_norm": 0.3719130754470825,
+      "learning_rate": 8.585890386586623e-05,
+      "loss": 0.9036,
+      "step": 789
+    },
+    {
+      "epoch": 0.25805397901269855,
+      "grad_norm": 0.4389430284500122,
+      "learning_rate": 8.582252858593324e-05,
+      "loss": 1.0327,
+      "step": 790
+    },
+    {
+      "epoch": 0.25838062961904373,
+      "grad_norm": 0.445206880569458,
+      "learning_rate": 8.578611430870979e-05,
+      "loss": 1.1391,
+      "step": 791
+    },
+    {
+      "epoch": 0.2587072802253889,
+      "grad_norm": 0.5291475057601929,
+      "learning_rate": 8.574966107383744e-05,
+      "loss": 1.2331,
+      "step": 792
+    },
+    {
+      "epoch": 0.2590339308317341,
+      "grad_norm": 0.5588846802711487,
+      "learning_rate": 8.5713168921e-05,
+      "loss": 1.0723,
+      "step": 793
+    },
+    {
+      "epoch": 0.2593605814380793,
+      "grad_norm": 0.6296544075012207,
+      "learning_rate": 8.567663788992377e-05,
+      "loss": 1.2545,
+      "step": 794
+    },
+    {
+      "epoch": 0.2596872320444245,
+      "grad_norm": 0.6767513155937195,
+      "learning_rate": 8.564006802037734e-05,
+      "loss": 1.1453,
+      "step": 795
+    },
+    {
+      "epoch": 0.2600138826507697,
+      "grad_norm": 0.7685205340385437,
+      "learning_rate": 8.560345935217155e-05,
+      "loss": 1.169,
+      "step": 796
+    },
+    {
+      "epoch": 0.26034053325711487,
+      "grad_norm": 0.975924015045166,
+      "learning_rate": 8.556681192515952e-05,
+      "loss": 1.3188,
+      "step": 797
+    },
+    {
+      "epoch": 0.26066718386346005,
+      "grad_norm": 1.1853704452514648,
+      "learning_rate": 8.553012577923653e-05,
+      "loss": 1.5078,
+      "step": 798
+    },
+    {
+      "epoch": 0.26099383446980523,
+      "grad_norm": 1.6453309059143066,
+      "learning_rate": 8.549340095434006e-05,
+      "loss": 1.5953,
+      "step": 799
+    },
+    {
+      "epoch": 0.2613204850761504,
+      "grad_norm": 1.7937952280044556,
+      "learning_rate": 8.54566374904496e-05,
+      "loss": 2.4296,
+      "step": 800
+    },
+    {
+      "epoch": 0.2616471356824956,
+      "grad_norm": 0.190406933426857,
+      "learning_rate": 8.541983542758685e-05,
+      "loss": 0.7342,
+      "step": 801
+    },
+    {
+      "epoch": 0.2619737862888408,
+      "grad_norm": 0.22838981449604034,
+      "learning_rate": 8.538299480581538e-05,
+      "loss": 0.7493,
+      "step": 802
+    },
+    {
+      "epoch": 0.262300436895186,
+      "grad_norm": 0.2559290826320648,
+      "learning_rate": 8.53461156652408e-05,
+      "loss": 0.8504,
+      "step": 803
+    },
+    {
+      "epoch": 0.2626270875015312,
+      "grad_norm": 0.2588362693786621,
+      "learning_rate": 8.53091980460107e-05,
+      "loss": 0.9,
+      "step": 804
+    },
+    {
+      "epoch": 0.26295373810787637,
+      "grad_norm": 0.2569828927516937,
+      "learning_rate": 8.527224198831447e-05,
+      "loss": 0.813,
+      "step": 805
+    },
+    {
+      "epoch": 0.26328038871422155,
+      "grad_norm": 0.2703791558742523,
+      "learning_rate": 8.523524753238342e-05,
+      "loss": 0.8616,
+      "step": 806
+    },
+    {
+      "epoch": 0.2636070393205667,
+      "grad_norm": 0.27567189931869507,
+      "learning_rate": 8.519821471849061e-05,
+      "loss": 0.9035,
+      "step": 807
+    },
+    {
+      "epoch": 0.2639336899269119,
+      "grad_norm": 0.2831965386867523,
+      "learning_rate": 8.516114358695089e-05,
+      "loss": 0.94,
+      "step": 808
+    },
+    {
+      "epoch": 0.2642603405332571,
+      "grad_norm": 0.28172117471694946,
+      "learning_rate": 8.51240341781208e-05,
+      "loss": 0.8232,
+      "step": 809
+    },
+    {
+      "epoch": 0.2645869911396023,
+      "grad_norm": 0.3024875521659851,
+      "learning_rate": 8.508688653239858e-05,
+      "loss": 0.8993,
+      "step": 810
+    },
+    {
+      "epoch": 0.2649136417459475,
+      "grad_norm": 0.3354199528694153,
+      "learning_rate": 8.504970069022404e-05,
+      "loss": 0.89,
+      "step": 811
+    },
+    {
+      "epoch": 0.2652402923522927,
+      "grad_norm": 0.34784260392189026,
+      "learning_rate": 8.501247669207864e-05,
+      "loss": 1.0391,
+      "step": 812
+    },
+    {
+      "epoch": 0.26556694295863786,
+      "grad_norm": 0.3697127103805542,
+      "learning_rate": 8.497521457848532e-05,
+      "loss": 1.0011,
+      "step": 813
+    },
+    {
+      "epoch": 0.26589359356498304,
+      "grad_norm": 0.37952542304992676,
+      "learning_rate": 8.493791439000855e-05,
+      "loss": 0.8449,
+      "step": 814
+    },
+    {
+      "epoch": 0.2662202441713282,
+      "grad_norm": 0.42014080286026,
+      "learning_rate": 8.490057616725424e-05,
+      "loss": 0.9275,
+      "step": 815
+    },
+    {
+      "epoch": 0.2665468947776734,
+      "grad_norm": 0.46746087074279785,
+      "learning_rate": 8.48631999508697e-05,
+      "loss": 1.2106,
+      "step": 816
+    },
+    {
+      "epoch": 0.26687354538401864,
+      "grad_norm": 0.5447984933853149,
+      "learning_rate": 8.482578578154361e-05,
+      "loss": 1.1964,
+      "step": 817
+    },
+    {
+      "epoch": 0.2672001959903638,
+      "grad_norm": 0.5636278390884399,
+      "learning_rate": 8.478833370000594e-05,
+      "loss": 1.1554,
+      "step": 818
+    },
+    {
+      "epoch": 0.267526846596709,
+      "grad_norm": 0.6589205861091614,
+      "learning_rate": 8.475084374702797e-05,
+      "loss": 1.2816,
+      "step": 819
+    },
+    {
+      "epoch": 0.2678534972030542,
+      "grad_norm": 0.716610312461853,
+      "learning_rate": 8.47133159634222e-05,
+      "loss": 1.2345,
+      "step": 820
+    },
+    {
+      "epoch": 0.26818014780939936,
+      "grad_norm": 0.866495668888092,
+      "learning_rate": 8.467575039004227e-05,
+      "loss": 1.2124,
+      "step": 821
+    },
+    {
+      "epoch": 0.26850679841574454,
+      "grad_norm": 1.089608073234558,
+      "learning_rate": 8.463814706778304e-05,
+      "loss": 1.69,
+      "step": 822
+    },
+    {
+      "epoch": 0.2688334490220897,
+      "grad_norm": 1.1608694791793823,
+      "learning_rate": 8.460050603758035e-05,
+      "loss": 1.732,
+      "step": 823
+    },
+    {
+      "epoch": 0.26916009962843496,
+      "grad_norm": 1.2007168531417847,
+      "learning_rate": 8.456282734041121e-05,
+      "loss": 1.83,
+      "step": 824
+    },
+    {
+      "epoch": 0.26948675023478014,
+      "grad_norm": 1.6301062107086182,
+      "learning_rate": 8.452511101729357e-05,
+      "loss": 1.664,
+      "step": 825
+    },
+    {
+      "epoch": 0.2698134008411253,
+      "grad_norm": 0.1741432547569275,
+      "learning_rate": 8.448735710928635e-05,
+      "loss": 0.6838,
+      "step": 826
+    },
+    {
+      "epoch": 0.2701400514474705,
+      "grad_norm": 0.21807898581027985,
+      "learning_rate": 8.444956565748937e-05,
+      "loss": 0.8771,
+      "step": 827
+    },
+    {
+      "epoch": 0.2704667020538157,
+      "grad_norm": 0.23946182429790497,
+      "learning_rate": 8.441173670304337e-05,
+      "loss": 0.7937,
+      "step": 828
+    },
+    {
+      "epoch": 0.27079335266016086,
+      "grad_norm": 0.24787762761116028,
+      "learning_rate": 8.437387028712984e-05,
+      "loss": 0.8956,
+      "step": 829
+    },
+    {
+      "epoch": 0.27112000326650604,
+      "grad_norm": 0.26188749074935913,
+      "learning_rate": 8.433596645097114e-05,
+      "loss": 0.8621,
+      "step": 830
+    },
+    {
+      "epoch": 0.2714466538728513,
+      "grad_norm": 0.28136762976646423,
+      "learning_rate": 8.429802523583032e-05,
+      "loss": 0.8664,
+      "step": 831
+    },
+    {
+      "epoch": 0.27177330447919645,
+      "grad_norm": 0.2663547992706299,
+      "learning_rate": 8.42600466830111e-05,
+      "loss": 0.8265,
+      "step": 832
+    },
+    {
+      "epoch": 0.27209995508554163,
+      "grad_norm": 0.2984652817249298,
+      "learning_rate": 8.422203083385791e-05,
+      "loss": 0.9242,
+      "step": 833
+    },
+    {
+      "epoch": 0.2724266056918868,
+      "grad_norm": 0.30643805861473083,
+      "learning_rate": 8.418397772975571e-05,
+      "loss": 0.914,
+      "step": 834
+    },
+    {
+      "epoch": 0.272753256298232,
+      "grad_norm": 0.33440685272216797,
+      "learning_rate": 8.414588741213004e-05,
+      "loss": 0.8974,
+      "step": 835
+    },
+    {
+      "epoch": 0.2730799069045772,
+      "grad_norm": 0.34194639325141907,
+      "learning_rate": 8.410775992244699e-05,
+      "loss": 0.9481,
+      "step": 836
+    },
+    {
+      "epoch": 0.27340655751092235,
+      "grad_norm": 0.3455904722213745,
+      "learning_rate": 8.406959530221308e-05,
+      "loss": 0.8916,
+      "step": 837
+    },
+    {
+      "epoch": 0.2737332081172676,
+      "grad_norm": 0.38675469160079956,
+      "learning_rate": 8.403139359297526e-05,
+      "loss": 0.981,
+      "step": 838
+    },
+    {
+      "epoch": 0.27405985872361277,
+      "grad_norm": 0.38662129640579224,
+      "learning_rate": 8.399315483632087e-05,
+      "loss": 0.9426,
+      "step": 839
+    },
+    {
+      "epoch": 0.27438650932995795,
+      "grad_norm": 0.4323444366455078,
+      "learning_rate": 8.395487907387751e-05,
+      "loss": 0.9693,
+      "step": 840
+    },
+    {
+      "epoch": 0.27471315993630313,
+      "grad_norm": 0.4435561001300812,
+      "learning_rate": 8.391656634731319e-05,
+      "loss": 0.9983,
+      "step": 841
+    },
+    {
+      "epoch": 0.2750398105426483,
+      "grad_norm": 0.5139126777648926,
+      "learning_rate": 8.387821669833606e-05,
+      "loss": 1.198,
+      "step": 842
+    },
+    {
+      "epoch": 0.2753664611489935,
+      "grad_norm": 0.5524410605430603,
+      "learning_rate": 8.383983016869448e-05,
+      "loss": 1.18,
+      "step": 843
+    },
+    {
+      "epoch": 0.27569311175533867,
+      "grad_norm": 0.6392261981964111,
+      "learning_rate": 8.380140680017703e-05,
+      "loss": 1.2321,
+      "step": 844
+    },
+    {
+      "epoch": 0.2760197623616839,
+      "grad_norm": 0.6585685610771179,
+      "learning_rate": 8.376294663461227e-05,
+      "loss": 1.2603,
+      "step": 845
+    },
+    {
+      "epoch": 0.2763464129680291,
+      "grad_norm": 0.8117931485176086,
+      "learning_rate": 8.372444971386894e-05,
+      "loss": 1.3741,
+      "step": 846
+    },
+    {
+      "epoch": 0.27667306357437427,
+      "grad_norm": 0.9951660633087158,
+      "learning_rate": 8.368591607985571e-05,
+      "loss": 1.3947,
+      "step": 847
+    },
+    {
+      "epoch": 0.27699971418071945,
+      "grad_norm": 1.1902111768722534,
+      "learning_rate": 8.364734577452127e-05,
+      "loss": 1.5964,
+      "step": 848
+    },
+    {
+      "epoch": 0.2773263647870646,
+      "grad_norm": 1.2848105430603027,
+      "learning_rate": 8.360873883985418e-05,
+      "loss": 1.3661,
+      "step": 849
+    },
+    {
+      "epoch": 0.2776530153934098,
+      "grad_norm": 1.5601998567581177,
+      "learning_rate": 8.357009531788293e-05,
+      "loss": 1.5122,
+      "step": 850
+    },
+    {
+      "epoch": 0.277979665999755,
+      "grad_norm": 0.17106448113918304,
+      "learning_rate": 8.353141525067579e-05,
+      "loss": 0.6374,
+      "step": 851
+    },
+    {
+      "epoch": 0.2783063166061002,
+      "grad_norm": 0.21366915106773376,
+      "learning_rate": 8.349269868034087e-05,
+      "loss": 0.8114,
+      "step": 852
+    },
+    {
+      "epoch": 0.2786329672124454,
+      "grad_norm": 0.2161540687084198,
+      "learning_rate": 8.345394564902594e-05,
+      "loss": 0.8004,
+      "step": 853
+    },
+    {
+      "epoch": 0.2789596178187906,
+      "grad_norm": 0.24288657307624817,
+      "learning_rate": 8.341515619891856e-05,
+      "loss": 0.8151,
+      "step": 854
+    },
+    {
+      "epoch": 0.27928626842513576,
+      "grad_norm": 0.25571781396865845,
+      "learning_rate": 8.337633037224583e-05,
+      "loss": 0.8574,
+      "step": 855
+    },
+    {
+      "epoch": 0.27961291903148094,
+      "grad_norm": 0.30678462982177734,
+      "learning_rate": 8.333746821127455e-05,
+      "loss": 0.8622,
+      "step": 856
+    },
+    {
+      "epoch": 0.2799395696378261,
+      "grad_norm": 0.2756674289703369,
+      "learning_rate": 8.329856975831103e-05,
+      "loss": 0.8822,
+      "step": 857
+    },
+    {
+      "epoch": 0.2802662202441713,
+      "grad_norm": 0.2979314923286438,
+      "learning_rate": 8.325963505570104e-05,
+      "loss": 0.8659,
+      "step": 858
+    },
+    {
+      "epoch": 0.28059287085051654,
+      "grad_norm": 0.2941073775291443,
+      "learning_rate": 8.322066414582992e-05,
+      "loss": 0.8474,
+      "step": 859
+    },
+    {
+      "epoch": 0.2809195214568617,
+      "grad_norm": 0.32220232486724854,
+      "learning_rate": 8.318165707112233e-05,
+      "loss": 0.865,
+      "step": 860
+    },
+    {
+      "epoch": 0.2812461720632069,
+      "grad_norm": 0.33228668570518494,
+      "learning_rate": 8.314261387404234e-05,
+      "loss": 0.9357,
+      "step": 861
+    },
+    {
+      "epoch": 0.2815728226695521,
+      "grad_norm": 0.3545335829257965,
+      "learning_rate": 8.310353459709333e-05,
+      "loss": 0.9402,
+      "step": 862
+    },
+    {
+      "epoch": 0.28189947327589726,
+      "grad_norm": 0.3884226679801941,
+      "learning_rate": 8.306441928281798e-05,
+      "loss": 1.0393,
+      "step": 863
+    },
+    {
+      "epoch": 0.28222612388224244,
+      "grad_norm": 0.40320566296577454,
+      "learning_rate": 8.302526797379822e-05,
+      "loss": 0.961,
+      "step": 864
+    },
+    {
+      "epoch": 0.2825527744885876,
+      "grad_norm": 0.4174101948738098,
+      "learning_rate": 8.298608071265507e-05,
+      "loss": 0.9835,
+      "step": 865
+    },
+    {
+      "epoch": 0.28287942509493286,
+      "grad_norm": 0.4447181820869446,
+      "learning_rate": 8.29468575420488e-05,
+      "loss": 1.0212,
+      "step": 866
+    },
+    {
+      "epoch": 0.28320607570127804,
+      "grad_norm": 0.46445220708847046,
+      "learning_rate": 8.29075985046787e-05,
+      "loss": 1.0336,
+      "step": 867
+    },
+    {
+      "epoch": 0.2835327263076232,
+      "grad_norm": 0.5036008358001709,
+      "learning_rate": 8.286830364328314e-05,
+      "loss": 1.0846,
+      "step": 868
+    },
+    {
+      "epoch": 0.2838593769139684,
+      "grad_norm": 0.5870608687400818,
+      "learning_rate": 8.282897300063946e-05,
+      "loss": 1.051,
+      "step": 869
+    },
+    {
+      "epoch": 0.2841860275203136,
+      "grad_norm": 0.6841760277748108,
+      "learning_rate": 8.278960661956401e-05,
+      "loss": 1.366,
+      "step": 870
+    },
+    {
+      "epoch": 0.28451267812665876,
+      "grad_norm": 0.799982488155365,
+      "learning_rate": 8.275020454291195e-05,
+      "loss": 1.2628,
+      "step": 871
+    },
+    {
+      "epoch": 0.28483932873300394,
+      "grad_norm": 1.000924825668335,
+      "learning_rate": 8.271076681357741e-05,
+      "loss": 1.4612,
+      "step": 872
+    },
+    {
+      "epoch": 0.2851659793393492,
+      "grad_norm": 1.1383004188537598,
+      "learning_rate": 8.267129347449322e-05,
+      "loss": 1.6271,
+      "step": 873
+    },
+    {
+      "epoch": 0.28549262994569435,
+      "grad_norm": 1.2377307415008545,
+      "learning_rate": 8.26317845686311e-05,
+      "loss": 1.2689,
+      "step": 874
+    },
+    {
+      "epoch": 0.28581928055203953,
+      "grad_norm": 1.6466963291168213,
+      "learning_rate": 8.259224013900137e-05,
+      "loss": 1.6594,
+      "step": 875
+    },
+    {
+      "epoch": 0.2861459311583847,
+      "grad_norm": 0.2105908840894699,
+      "learning_rate": 8.255266022865309e-05,
+      "loss": 0.7822,
+      "step": 876
+    },
+    {
+      "epoch": 0.2864725817647299,
+      "grad_norm": 0.22548696398735046,
+      "learning_rate": 8.251304488067393e-05,
+      "loss": 0.7938,
+      "step": 877
+    },
+    {
+      "epoch": 0.2867992323710751,
+      "grad_norm": 0.24404418468475342,
+      "learning_rate": 8.247339413819015e-05,
+      "loss": 0.7887,
+      "step": 878
+    },
+    {
+      "epoch": 0.28712588297742025,
+      "grad_norm": 0.25363683700561523,
+      "learning_rate": 8.243370804436649e-05,
+      "loss": 0.9139,
+      "step": 879
+    },
+    {
+      "epoch": 0.2874525335837655,
+      "grad_norm": 0.25927162170410156,
+      "learning_rate": 8.239398664240627e-05,
+      "loss": 0.8177,
+      "step": 880
+    },
+    {
+      "epoch": 0.28777918419011067,
+      "grad_norm": 0.2730761170387268,
+      "learning_rate": 8.235422997555114e-05,
+      "loss": 0.9055,
+      "step": 881
+    },
+    {
+      "epoch": 0.28810583479645585,
+      "grad_norm": 0.2785848081111908,
+      "learning_rate": 8.231443808708122e-05,
+      "loss": 0.9572,
+      "step": 882
+    },
+    {
+      "epoch": 0.28843248540280103,
+      "grad_norm": 0.27791181206703186,
+      "learning_rate": 8.227461102031493e-05,
+      "loss": 0.803,
+      "step": 883
+    },
+    {
+      "epoch": 0.2887591360091462,
+      "grad_norm": 0.28716158866882324,
+      "learning_rate": 8.2234748818609e-05,
+      "loss": 0.8582,
+      "step": 884
+    },
+    {
+      "epoch": 0.2890857866154914,
+      "grad_norm": 0.3149946630001068,
+      "learning_rate": 8.21948515253584e-05,
+      "loss": 0.9189,
+      "step": 885
+    },
+    {
+      "epoch": 0.28941243722183657,
+      "grad_norm": 0.3081023395061493,
+      "learning_rate": 8.215491918399633e-05,
+      "loss": 0.8939,
+      "step": 886
+    },
+    {
+      "epoch": 0.2897390878281818,
+      "grad_norm": 0.31537654995918274,
+      "learning_rate": 8.211495183799413e-05,
+      "loss": 0.8625,
+      "step": 887
+    },
+    {
+      "epoch": 0.290065738434527,
+      "grad_norm": 0.3406592905521393,
+      "learning_rate": 8.20749495308612e-05,
+      "loss": 0.9604,
+      "step": 888
+    },
+    {
+      "epoch": 0.29039238904087217,
+      "grad_norm": 0.3376745581626892,
+      "learning_rate": 8.20349123061451e-05,
+      "loss": 0.8982,
+      "step": 889
+    },
+    {
+      "epoch": 0.29071903964721735,
+      "grad_norm": 0.3791499435901642,
+      "learning_rate": 8.19948402074313e-05,
+      "loss": 0.9649,
+      "step": 890
+    },
+    {
+      "epoch": 0.29104569025356253,
+      "grad_norm": 0.4459807872772217,
+      "learning_rate": 8.195473327834329e-05,
+      "loss": 1.1565,
+      "step": 891
+    },
+    {
+      "epoch": 0.2913723408599077,
+      "grad_norm": 0.45107483863830566,
+      "learning_rate": 8.191459156254247e-05,
+      "loss": 0.9899,
+      "step": 892
+    },
+    {
+      "epoch": 0.2916989914662529,
+      "grad_norm": 0.50025475025177,
+      "learning_rate": 8.187441510372808e-05,
+      "loss": 1.1841,
+      "step": 893
+    },
+    {
+      "epoch": 0.2920256420725981,
+      "grad_norm": 0.5542461276054382,
+      "learning_rate": 8.183420394563724e-05,
+      "loss": 1.183,
+      "step": 894
+    },
+    {
+      "epoch": 0.2923522926789433,
+      "grad_norm": 0.6245063543319702,
+      "learning_rate": 8.179395813204477e-05,
+      "loss": 1.3295,
+      "step": 895
+    },
+    {
+      "epoch": 0.2926789432852885,
+      "grad_norm": 0.7166218757629395,
+      "learning_rate": 8.17536777067633e-05,
+      "loss": 1.3525,
+      "step": 896
+    },
+    {
+      "epoch": 0.29300559389163366,
+      "grad_norm": 0.9630147814750671,
+      "learning_rate": 8.171336271364308e-05,
+      "loss": 1.4074,
+      "step": 897
+    },
+    {
+      "epoch": 0.29333224449797884,
+      "grad_norm": 1.0214118957519531,
+      "learning_rate": 8.167301319657201e-05,
+      "loss": 1.4875,
+      "step": 898
+    },
+    {
+      "epoch": 0.293658895104324,
+      "grad_norm": 1.444045901298523,
+      "learning_rate": 8.163262919947557e-05,
+      "loss": 1.6154,
+      "step": 899
+    },
+    {
+      "epoch": 0.2939855457106692,
+      "grad_norm": 1.9283764362335205,
+      "learning_rate": 8.159221076631678e-05,
+      "loss": 2.093,
+      "step": 900
+    },
+    {
+      "epoch": 0.29431219631701444,
+      "grad_norm": 0.2009824812412262,
+      "learning_rate": 8.155175794109614e-05,
+      "loss": 0.8172,
+      "step": 901
+    },
+    {
+      "epoch": 0.2946388469233596,
+      "grad_norm": 0.21653543412685394,
+      "learning_rate": 8.15112707678516e-05,
+      "loss": 0.8646,
+      "step": 902
+    },
+    {
+      "epoch": 0.2949654975297048,
+      "grad_norm": 0.26212337613105774,
+      "learning_rate": 8.14707492906585e-05,
+      "loss": 0.9042,
+      "step": 903
+    },
+    {
+      "epoch": 0.29529214813605,
+      "grad_norm": 0.26977843046188354,
+      "learning_rate": 8.143019355362952e-05,
+      "loss": 0.8568,
+      "step": 904
+    },
+    {
+      "epoch": 0.29561879874239516,
+      "grad_norm": 0.2698366045951843,
+      "learning_rate": 8.138960360091463e-05,
+      "loss": 0.8116,
+      "step": 905
+    },
+    {
+      "epoch": 0.29594544934874034,
+      "grad_norm": 0.29477614164352417,
+      "learning_rate": 8.134897947670108e-05,
+      "loss": 0.8601,
+      "step": 906
+    },
+    {
+      "epoch": 0.2962720999550855,
+      "grad_norm": 0.2830169200897217,
+      "learning_rate": 8.130832122521327e-05,
+      "loss": 0.8487,
+      "step": 907
+    },
+    {
+      "epoch": 0.29659875056143076,
+      "grad_norm": 0.3099175691604614,
+      "learning_rate": 8.12676288907128e-05,
+      "loss": 0.9419,
+      "step": 908
+    },
+    {
+      "epoch": 0.29692540116777594,
+      "grad_norm": 0.3113427758216858,
+      "learning_rate": 8.122690251749834e-05,
+      "loss": 0.94,
+      "step": 909
+    },
+    {
+      "epoch": 0.2972520517741211,
+      "grad_norm": 0.317848265171051,
+      "learning_rate": 8.118614214990561e-05,
+      "loss": 0.9457,
+      "step": 910
+    },
+    {
+      "epoch": 0.2975787023804663,
+      "grad_norm": 0.32310110330581665,
+      "learning_rate": 8.114534783230739e-05,
+      "loss": 0.93,
+      "step": 911
+    },
+    {
+      "epoch": 0.2979053529868115,
+      "grad_norm": 0.3455216884613037,
+      "learning_rate": 8.110451960911333e-05,
+      "loss": 0.9539,
+      "step": 912
+    },
+    {
+      "epoch": 0.29823200359315666,
+      "grad_norm": 0.3817412853240967,
+      "learning_rate": 8.106365752477012e-05,
+      "loss": 1.0052,
+      "step": 913
+    },
+    {
+      "epoch": 0.29855865419950184,
+      "grad_norm": 0.39544934034347534,
+      "learning_rate": 8.102276162376117e-05,
+      "loss": 0.9988,
+      "step": 914
+    },
+    {
+      "epoch": 0.298885304805847,
+      "grad_norm": 0.43939855694770813,
+      "learning_rate": 8.09818319506068e-05,
+      "loss": 1.1415,
+      "step": 915
+    },
+    {
+      "epoch": 0.29921195541219225,
+      "grad_norm": 0.442859947681427,
+      "learning_rate": 8.094086854986405e-05,
+      "loss": 1.0522,
+      "step": 916
+    },
+    {
+      "epoch": 0.29953860601853743,
+      "grad_norm": 0.4725445508956909,
+      "learning_rate": 8.089987146612669e-05,
+      "loss": 1.0607,
+      "step": 917
+    },
+    {
+      "epoch": 0.2998652566248826,
+      "grad_norm": 0.5423445701599121,
+      "learning_rate": 8.085884074402518e-05,
+      "loss": 1.1992,
+      "step": 918
+    },
+    {
+      "epoch": 0.3001919072312278,
+      "grad_norm": 0.5503812432289124,
+      "learning_rate": 8.081777642822657e-05,
+      "loss": 1.1793,
+      "step": 919
+    },
+    {
+      "epoch": 0.300518557837573,
+      "grad_norm": 0.5893210172653198,
+      "learning_rate": 8.077667856343449e-05,
+      "loss": 1.0644,
+      "step": 920
+    },
+    {
+      "epoch": 0.30084520844391816,
+      "grad_norm": 0.7819690108299255,
+      "learning_rate": 8.073554719438908e-05,
+      "loss": 1.4015,
+      "step": 921
+    },
+    {
+      "epoch": 0.30117185905026334,
+      "grad_norm": 0.8868228793144226,
+      "learning_rate": 8.069438236586695e-05,
+      "loss": 1.3257,
+      "step": 922
+    },
+    {
+      "epoch": 0.30149850965660857,
+      "grad_norm": 1.201609492301941,
+      "learning_rate": 8.065318412268119e-05,
+      "loss": 1.3868,
+      "step": 923
+    },
+    {
+      "epoch": 0.30182516026295375,
+      "grad_norm": 1.1145105361938477,
+      "learning_rate": 8.061195250968121e-05,
+      "loss": 1.4812,
+      "step": 924
+    },
+    {
+      "epoch": 0.30215181086929893,
+      "grad_norm": 1.567140817642212,
+      "learning_rate": 8.057068757175276e-05,
+      "loss": 1.737,
+      "step": 925
+    },
+    {
+      "epoch": 0.3024784614756441,
+      "grad_norm": 0.21669495105743408,
+      "learning_rate": 8.052938935381786e-05,
+      "loss": 0.7445,
+      "step": 926
+    },
+    {
+      "epoch": 0.3028051120819893,
+      "grad_norm": 0.22615468502044678,
+      "learning_rate": 8.048805790083481e-05,
+      "loss": 0.8108,
+      "step": 927
+    },
+    {
+      "epoch": 0.30313176268833447,
+      "grad_norm": 0.2774372100830078,
+      "learning_rate": 8.0446693257798e-05,
+      "loss": 0.8657,
+      "step": 928
+    },
+    {
+      "epoch": 0.30345841329467965,
+      "grad_norm": 0.2653120756149292,
+      "learning_rate": 8.040529546973805e-05,
+      "loss": 0.8441,
+      "step": 929
+    },
+    {
+      "epoch": 0.3037850639010249,
+      "grad_norm": 0.303362637758255,
+      "learning_rate": 8.036386458172161e-05,
+      "loss": 0.869,
+      "step": 930
+    },
+    {
+      "epoch": 0.30411171450737007,
+      "grad_norm": 0.29433003067970276,
+      "learning_rate": 8.032240063885133e-05,
+      "loss": 0.8341,
+      "step": 931
+    },
+    {
+      "epoch": 0.30443836511371525,
+      "grad_norm": 0.301315575838089,
+      "learning_rate": 8.028090368626591e-05,
+      "loss": 0.8746,
+      "step": 932
+    },
+    {
+      "epoch": 0.30476501572006043,
+      "grad_norm": 0.30583563446998596,
+      "learning_rate": 8.023937376913996e-05,
+      "loss": 0.8948,
+      "step": 933
+    },
+    {
+      "epoch": 0.3050916663264056,
+      "grad_norm": 0.3299638330936432,
+      "learning_rate": 8.019781093268396e-05,
+      "loss": 0.8029,
+      "step": 934
+    },
+    {
+      "epoch": 0.3054183169327508,
+      "grad_norm": 0.3463474214076996,
+      "learning_rate": 8.015621522214429e-05,
+      "loss": 0.8575,
+      "step": 935
+    },
+    {
+      "epoch": 0.30574496753909597,
+      "grad_norm": 0.3461097776889801,
+      "learning_rate": 8.0114586682803e-05,
+      "loss": 0.9876,
+      "step": 936
+    },
+    {
+      "epoch": 0.3060716181454412,
+      "grad_norm": 0.354095458984375,
+      "learning_rate": 8.007292535997799e-05,
+      "loss": 0.9575,
+      "step": 937
+    },
+    {
+      "epoch": 0.3063982687517864,
+      "grad_norm": 0.3798564076423645,
+      "learning_rate": 8.00312312990228e-05,
+      "loss": 1.0164,
+      "step": 938
+    },
+    {
+      "epoch": 0.30672491935813156,
+      "grad_norm": 0.3784736692905426,
+      "learning_rate": 7.998950454532662e-05,
+      "loss": 0.97,
+      "step": 939
+    },
+    {
+      "epoch": 0.30705156996447674,
+      "grad_norm": 0.40412405133247375,
+      "learning_rate": 7.99477451443142e-05,
+      "loss": 0.9975,
+      "step": 940
+    },
+    {
+      "epoch": 0.3073782205708219,
+      "grad_norm": 0.4350655674934387,
+      "learning_rate": 7.990595314144587e-05,
+      "loss": 1.005,
+      "step": 941
+    },
+    {
+      "epoch": 0.3077048711771671,
+      "grad_norm": 0.4765323996543884,
+      "learning_rate": 7.986412858221746e-05,
+      "loss": 1.0305,
+      "step": 942
+    },
+    {
+      "epoch": 0.3080315217835123,
+      "grad_norm": 0.5018781423568726,
+      "learning_rate": 7.982227151216019e-05,
+      "loss": 1.0618,
+      "step": 943
+    },
+    {
+      "epoch": 0.3083581723898575,
+      "grad_norm": 0.574617862701416,
+      "learning_rate": 7.978038197684073e-05,
+      "loss": 1.1572,
+      "step": 944
+    },
+    {
+      "epoch": 0.3086848229962027,
+      "grad_norm": 0.6548877358436584,
+      "learning_rate": 7.973846002186103e-05,
+      "loss": 1.2569,
+      "step": 945
+    },
+    {
+      "epoch": 0.3090114736025479,
+      "grad_norm": 0.7577599883079529,
+      "learning_rate": 7.969650569285839e-05,
+      "loss": 1.3086,
+      "step": 946
+    },
+    {
+      "epoch": 0.30933812420889306,
+      "grad_norm": 0.8888007998466492,
+      "learning_rate": 7.965451903550531e-05,
+      "loss": 1.1316,
+      "step": 947
+    },
+    {
+      "epoch": 0.30966477481523824,
+      "grad_norm": 1.0013303756713867,
+      "learning_rate": 7.961250009550953e-05,
+      "loss": 1.3751,
+      "step": 948
+    },
+    {
+      "epoch": 0.3099914254215834,
+      "grad_norm": 1.2923164367675781,
+      "learning_rate": 7.95704489186139e-05,
+      "loss": 1.5979,
+      "step": 949
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "grad_norm": 1.940639853477478,
+      "learning_rate": 7.952836555059635e-05,
+      "loss": 2.0671,
+      "step": 950
+    },
+    {
+      "epoch": 0.31064472663427384,
+      "grad_norm": 0.20804435014724731,
+      "learning_rate": 7.94862500372699e-05,
+      "loss": 0.7342,
+      "step": 951
+    },
+    {
+      "epoch": 0.310971377240619,
+      "grad_norm": 0.22356468439102173,
+      "learning_rate": 7.944410242448253e-05,
+      "loss": 0.8373,
+      "step": 952
+    },
+    {
+      "epoch": 0.3112980278469642,
+      "grad_norm": 0.22745850682258606,
+      "learning_rate": 7.940192275811717e-05,
+      "loss": 0.8846,
+      "step": 953
+    },
+    {
+      "epoch": 0.3116246784533094,
+      "grad_norm": 0.22750817239284515,
+      "learning_rate": 7.935971108409166e-05,
+      "loss": 0.8154,
+      "step": 954
+    },
+    {
+      "epoch": 0.31195132905965456,
+      "grad_norm": 0.26259639859199524,
+      "learning_rate": 7.931746744835865e-05,
+      "loss": 0.8305,
+      "step": 955
+    },
+    {
+      "epoch": 0.31227797966599974,
+      "grad_norm": 0.26343998312950134,
+      "learning_rate": 7.927519189690562e-05,
+      "loss": 0.9104,
+      "step": 956
+    },
+    {
+      "epoch": 0.3126046302723449,
+      "grad_norm": 0.28506144881248474,
+      "learning_rate": 7.923288447575479e-05,
+      "loss": 0.8731,
+      "step": 957
+    },
+    {
+      "epoch": 0.31293128087869015,
+      "grad_norm": 0.2877998948097229,
+      "learning_rate": 7.919054523096306e-05,
+      "loss": 0.9688,
+      "step": 958
+    },
+    {
+      "epoch": 0.31325793148503533,
+      "grad_norm": 0.2984800636768341,
+      "learning_rate": 7.914817420862196e-05,
+      "loss": 0.8833,
+      "step": 959
+    },
+    {
+      "epoch": 0.3135845820913805,
+      "grad_norm": 0.3453596234321594,
+      "learning_rate": 7.910577145485765e-05,
+      "loss": 0.8775,
+      "step": 960
+    },
+    {
+      "epoch": 0.3139112326977257,
+      "grad_norm": 0.3432406187057495,
+      "learning_rate": 7.906333701583082e-05,
+      "loss": 0.8728,
+      "step": 961
+    },
+    {
+      "epoch": 0.3142378833040709,
+      "grad_norm": 0.3565903604030609,
+      "learning_rate": 7.902087093773663e-05,
+      "loss": 0.9814,
+      "step": 962
+    },
+    {
+      "epoch": 0.31456453391041606,
+      "grad_norm": 0.3908556401729584,
+      "learning_rate": 7.897837326680473e-05,
+      "loss": 1.0056,
+      "step": 963
+    },
+    {
+      "epoch": 0.31489118451676124,
+      "grad_norm": 0.392490029335022,
+      "learning_rate": 7.89358440492991e-05,
+      "loss": 1.0565,
+      "step": 964
+    },
+    {
+      "epoch": 0.31521783512310647,
+      "grad_norm": 0.4333183765411377,
+      "learning_rate": 7.889328333151814e-05,
+      "loss": 1.0835,
+      "step": 965
+    },
+    {
+      "epoch": 0.31554448572945165,
+      "grad_norm": 0.5302640199661255,
+      "learning_rate": 7.885069115979447e-05,
+      "loss": 1.112,
+      "step": 966
+    },
+    {
+      "epoch": 0.31587113633579683,
+      "grad_norm": 0.5419654846191406,
+      "learning_rate": 7.880806758049499e-05,
+      "loss": 1.166,
+      "step": 967
+    },
+    {
+      "epoch": 0.316197786942142,
+      "grad_norm": 0.5007505416870117,
+      "learning_rate": 7.876541264002078e-05,
+      "loss": 0.9182,
+      "step": 968
+    },
+    {
+      "epoch": 0.3165244375484872,
+      "grad_norm": 0.6056339144706726,
+      "learning_rate": 7.872272638480706e-05,
+      "loss": 1.2492,
+      "step": 969
+    },
+    {
+      "epoch": 0.31685108815483237,
+      "grad_norm": 0.7134028673171997,
+      "learning_rate": 7.868000886132316e-05,
+      "loss": 1.3831,
+      "step": 970
+    },
+    {
+      "epoch": 0.31717773876117755,
+      "grad_norm": 0.7521294355392456,
+      "learning_rate": 7.863726011607243e-05,
+      "loss": 1.4182,
+      "step": 971
+    },
+    {
+      "epoch": 0.3175043893675228,
+      "grad_norm": 0.9576146602630615,
+      "learning_rate": 7.859448019559217e-05,
+      "loss": 1.386,
+      "step": 972
+    },
+    {
+      "epoch": 0.31783103997386797,
+      "grad_norm": 1.1564505100250244,
+      "learning_rate": 7.855166914645372e-05,
+      "loss": 1.4906,
+      "step": 973
+    },
+    {
+      "epoch": 0.31815769058021315,
+      "grad_norm": 1.5212230682373047,
+      "learning_rate": 7.850882701526218e-05,
+      "loss": 1.6107,
+      "step": 974
+    },
+    {
+      "epoch": 0.31848434118655833,
+      "grad_norm": 2.16402268409729,
+      "learning_rate": 7.846595384865662e-05,
+      "loss": 2.8417,
+      "step": 975
+    },
+    {
+      "epoch": 0.3188109917929035,
+      "grad_norm": 0.20599240064620972,
+      "learning_rate": 7.84230496933098e-05,
+      "loss": 0.794,
+      "step": 976
+    },
+    {
+      "epoch": 0.3191376423992487,
+      "grad_norm": 0.21791720390319824,
+      "learning_rate": 7.838011459592824e-05,
+      "loss": 0.8069,
+      "step": 977
+    },
+    {
+      "epoch": 0.31946429300559387,
+      "grad_norm": 0.24189670383930206,
+      "learning_rate": 7.833714860325215e-05,
+      "loss": 0.8253,
+      "step": 978
+    },
+    {
+      "epoch": 0.3197909436119391,
+      "grad_norm": 0.23616375029087067,
+      "learning_rate": 7.829415176205539e-05,
+      "loss": 0.8416,
+      "step": 979
+    },
+    {
+      "epoch": 0.3201175942182843,
+      "grad_norm": 0.2511419951915741,
+      "learning_rate": 7.825112411914535e-05,
+      "loss": 0.8737,
+      "step": 980
+    },
+    {
+      "epoch": 0.32044424482462947,
+      "grad_norm": 0.24740338325500488,
+      "learning_rate": 7.820806572136301e-05,
+      "loss": 0.6864,
+      "step": 981
+    },
+    {
+      "epoch": 0.32077089543097465,
+      "grad_norm": 0.27360159158706665,
+      "learning_rate": 7.81649766155828e-05,
+      "loss": 0.9015,
+      "step": 982
+    },
+    {
+      "epoch": 0.3210975460373198,
+      "grad_norm": 0.27167263627052307,
+      "learning_rate": 7.812185684871261e-05,
+      "loss": 0.8641,
+      "step": 983
+    },
+    {
+      "epoch": 0.321424196643665,
+      "grad_norm": 0.27509331703186035,
+      "learning_rate": 7.807870646769364e-05,
+      "loss": 0.8864,
+      "step": 984
+    },
+    {
+      "epoch": 0.3217508472500102,
+      "grad_norm": 0.2927476167678833,
+      "learning_rate": 7.80355255195005e-05,
+      "loss": 0.8999,
+      "step": 985
+    },
+    {
+      "epoch": 0.3220774978563554,
+      "grad_norm": 0.30333781242370605,
+      "learning_rate": 7.799231405114102e-05,
+      "loss": 0.8984,
+      "step": 986
+    },
+    {
+      "epoch": 0.3224041484627006,
+      "grad_norm": 0.29891061782836914,
+      "learning_rate": 7.794907210965627e-05,
+      "loss": 0.8042,
+      "step": 987
+    },
+    {
+      "epoch": 0.3227307990690458,
+      "grad_norm": 0.31980669498443604,
+      "learning_rate": 7.790579974212052e-05,
+      "loss": 0.9125,
+      "step": 988
+    },
+    {
+      "epoch": 0.32305744967539096,
+      "grad_norm": 0.3497137427330017,
+      "learning_rate": 7.78624969956411e-05,
+      "loss": 1.0623,
+      "step": 989
+    },
+    {
+      "epoch": 0.32338410028173614,
+      "grad_norm": 0.3825185298919678,
+      "learning_rate": 7.781916391735847e-05,
+      "loss": 0.9571,
+      "step": 990
+    },
+    {
+      "epoch": 0.3237107508880813,
+      "grad_norm": 0.44200360774993896,
+      "learning_rate": 7.77758005544461e-05,
+      "loss": 0.9618,
+      "step": 991
+    },
+    {
+      "epoch": 0.3240374014944265,
+      "grad_norm": 0.4567243754863739,
+      "learning_rate": 7.773240695411042e-05,
+      "loss": 1.0149,
+      "step": 992
+    },
+    {
+      "epoch": 0.32436405210077174,
+      "grad_norm": 0.5161774158477783,
+      "learning_rate": 7.768898316359076e-05,
+      "loss": 1.1654,
+      "step": 993
+    },
+    {
+      "epoch": 0.3246907027071169,
+      "grad_norm": 0.5306985974311829,
+      "learning_rate": 7.764552923015935e-05,
+      "loss": 1.1412,
+      "step": 994
+    },
+    {
+      "epoch": 0.3250173533134621,
+      "grad_norm": 0.6168984174728394,
+      "learning_rate": 7.76020452011212e-05,
+      "loss": 1.158,
+      "step": 995
+    },
+    {
+      "epoch": 0.3253440039198073,
+      "grad_norm": 0.707940936088562,
+      "learning_rate": 7.755853112381411e-05,
+      "loss": 1.3226,
+      "step": 996
+    },
+    {
+      "epoch": 0.32567065452615246,
+      "grad_norm": 0.8782923817634583,
+      "learning_rate": 7.751498704560858e-05,
+      "loss": 1.3097,
+      "step": 997
+    },
+    {
+      "epoch": 0.32599730513249764,
+      "grad_norm": 1.1435363292694092,
+      "learning_rate": 7.747141301390777e-05,
+      "loss": 1.5753,
+      "step": 998
+    },
+    {
+      "epoch": 0.3263239557388428,
+      "grad_norm": 1.199203372001648,
+      "learning_rate": 7.742780907614742e-05,
+      "loss": 1.2035,
+      "step": 999
+    },
+    {
+      "epoch": 0.32665060634518805,
+      "grad_norm": 1.617131233215332,
+      "learning_rate": 7.73841752797959e-05,
+      "loss": 1.5258,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32697725695153323,
+      "grad_norm": 0.21024766564369202,
+      "learning_rate": 7.734051167235404e-05,
+      "loss": 0.8054,
+      "step": 1001
+    },
+    {
+      "epoch": 0.3273039075578784,
+      "grad_norm": 0.23412029445171356,
+      "learning_rate": 7.729681830135506e-05,
+      "loss": 0.8206,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3276305581642236,
+      "grad_norm": 0.26261237263679504,
+      "learning_rate": 7.725309521436473e-05,
+      "loss": 0.8588,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3279572087705688,
+      "grad_norm": 0.2516871690750122,
+      "learning_rate": 7.720934245898101e-05,
+      "loss": 0.842,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32828385937691396,
+      "grad_norm": 0.268623411655426,
+      "learning_rate": 7.716556008283428e-05,
+      "loss": 0.7826,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32861050998325914,
+      "grad_norm": 0.282501220703125,
+      "learning_rate": 7.712174813358709e-05,
+      "loss": 0.8606,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32893716058960437,
+      "grad_norm": 0.2929759621620178,
+      "learning_rate": 7.707790665893422e-05,
+      "loss": 0.9098,
+      "step": 1007
+    },
+    {
+      "epoch": 0.32926381119594955,
+      "grad_norm": 0.30565446615219116,
+      "learning_rate": 7.703403570660259e-05,
+      "loss": 0.905,
+      "step": 1008
+    },
+    {
+      "epoch": 0.32959046180229473,
+      "grad_norm": 0.3160889446735382,
+      "learning_rate": 7.699013532435119e-05,
+      "loss": 0.9065,
+      "step": 1009
+    },
+    {
+      "epoch": 0.3299171124086399,
+      "grad_norm": 0.34107425808906555,
+      "learning_rate": 7.694620555997107e-05,
+      "loss": 0.9131,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3302437630149851,
+      "grad_norm": 0.3465249538421631,
+      "learning_rate": 7.690224646128526e-05,
+      "loss": 0.8906,
+      "step": 1011
+    },
+    {
+      "epoch": 0.3305704136213303,
+      "grad_norm": 0.36077815294265747,
+      "learning_rate": 7.685825807614872e-05,
+      "loss": 0.9187,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33089706422767545,
+      "grad_norm": 0.37028247117996216,
+      "learning_rate": 7.681424045244829e-05,
+      "loss": 0.912,
+      "step": 1013
+    },
+    {
+      "epoch": 0.3312237148340207,
+      "grad_norm": 0.3931543827056885,
+      "learning_rate": 7.677019363810268e-05,
+      "loss": 1.0148,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33155036544036587,
+      "grad_norm": 0.4445292055606842,
+      "learning_rate": 7.672611768106227e-05,
+      "loss": 1.015,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33187701604671105,
+      "grad_norm": 0.49006712436676025,
+      "learning_rate": 7.668201262930927e-05,
+      "loss": 0.9695,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33220366665305623,
+      "grad_norm": 0.5494130253791809,
+      "learning_rate": 7.663787853085755e-05,
+      "loss": 1.2143,
+      "step": 1017
+    },
+    {
+      "epoch": 0.3325303172594014,
+      "grad_norm": 0.5494057536125183,
+      "learning_rate": 7.659371543375258e-05,
+      "loss": 1.1659,
+      "step": 1018
+    },
+    {
+      "epoch": 0.3328569678657466,
+      "grad_norm": 0.5971184372901917,
+      "learning_rate": 7.654952338607137e-05,
+      "loss": 1.1892,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33318361847209177,
+      "grad_norm": 0.7538458704948425,
+      "learning_rate": 7.650530243592248e-05,
+      "loss": 1.1604,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33351026907843695,
+      "grad_norm": 0.9583027362823486,
+      "learning_rate": 7.646105263144595e-05,
+      "loss": 1.3371,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3338369196847822,
+      "grad_norm": 1.0476845502853394,
+      "learning_rate": 7.64167740208132e-05,
+      "loss": 1.3216,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33416357029112737,
+      "grad_norm": 1.0443453788757324,
+      "learning_rate": 7.637246665222704e-05,
+      "loss": 1.225,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33449022089747255,
+      "grad_norm": 1.3026036024093628,
+      "learning_rate": 7.632813057392151e-05,
+      "loss": 1.4609,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3348168715038177,
+      "grad_norm": 1.555985450744629,
+      "learning_rate": 7.628376583416204e-05,
+      "loss": 1.7155,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3351435221101629,
+      "grad_norm": 0.19100616872310638,
+      "learning_rate": 7.623937248124513e-05,
+      "loss": 0.7413,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3354701727165081,
+      "grad_norm": 0.22931453585624695,
+      "learning_rate": 7.619495056349849e-05,
+      "loss": 0.894,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33579682332285327,
+      "grad_norm": 0.2343945950269699,
+      "learning_rate": 7.615050012928092e-05,
+      "loss": 0.8266,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3361234739291985,
+      "grad_norm": 0.27221325039863586,
+      "learning_rate": 7.610602122698227e-05,
+      "loss": 0.8433,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3364501245355437,
+      "grad_norm": 0.2684449553489685,
+      "learning_rate": 7.606151390502337e-05,
+      "loss": 0.7658,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33677677514188886,
+      "grad_norm": 0.2658660411834717,
+      "learning_rate": 7.6016978211856e-05,
+      "loss": 0.8912,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33710342574823404,
+      "grad_norm": 0.27399852871894836,
+      "learning_rate": 7.597241419596279e-05,
+      "loss": 0.8434,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3374300763545792,
+      "grad_norm": 0.27935051918029785,
+      "learning_rate": 7.592782190585725e-05,
+      "loss": 0.8468,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3377567269609244,
+      "grad_norm": 0.3147827684879303,
+      "learning_rate": 7.588320139008365e-05,
+      "loss": 0.8781,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3380833775672696,
+      "grad_norm": 0.31882667541503906,
+      "learning_rate": 7.583855269721697e-05,
+      "loss": 0.9354,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3384100281736148,
+      "grad_norm": 0.3448106348514557,
+      "learning_rate": 7.579387587586292e-05,
+      "loss": 0.8532,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33873667877996,
+      "grad_norm": 0.3815080225467682,
+      "learning_rate": 7.574917097465774e-05,
+      "loss": 0.945,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3390633293863052,
+      "grad_norm": 0.3913522958755493,
+      "learning_rate": 7.570443804226833e-05,
+      "loss": 0.8925,
+      "step": 1038
+    },
+    {
+      "epoch": 0.33938997999265036,
+      "grad_norm": 0.42671769857406616,
+      "learning_rate": 7.565967712739205e-05,
+      "loss": 1.0106,
+      "step": 1039
+    },
+    {
+      "epoch": 0.33971663059899554,
+      "grad_norm": 0.4863613247871399,
+      "learning_rate": 7.561488827875675e-05,
+      "loss": 1.0284,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3400432812053407,
+      "grad_norm": 0.507326066493988,
+      "learning_rate": 7.557007154512065e-05,
+      "loss": 1.0476,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3403699318116859,
+      "grad_norm": 0.5292708277702332,
+      "learning_rate": 7.55252269752724e-05,
+      "loss": 1.0151,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34069658241803114,
+      "grad_norm": 0.5429087281227112,
+      "learning_rate": 7.548035461803087e-05,
+      "loss": 1.0862,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3410232330243763,
+      "grad_norm": 0.6534260511398315,
+      "learning_rate": 7.543545452224523e-05,
+      "loss": 1.1288,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3413498836307215,
+      "grad_norm": 0.8366156816482544,
+      "learning_rate": 7.539052673679483e-05,
+      "loss": 1.5115,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3416765342370667,
+      "grad_norm": 0.8968381881713867,
+      "learning_rate": 7.534557131058917e-05,
+      "loss": 1.1607,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34200318484341186,
+      "grad_norm": 1.1521880626678467,
+      "learning_rate": 7.530058829256785e-05,
+      "loss": 1.2325,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34232983544975704,
+      "grad_norm": 1.234046459197998,
+      "learning_rate": 7.525557773170048e-05,
+      "loss": 1.6835,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3426564860561022,
+      "grad_norm": 2.0295400619506836,
+      "learning_rate": 7.521053967698669e-05,
+      "loss": 2.3934,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34298313666244745,
+      "grad_norm": 2.4439151287078857,
+      "learning_rate": 7.516547417745598e-05,
+      "loss": 2.534,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34330978726879263,
+      "grad_norm": 0.20437824726104736,
+      "learning_rate": 7.512038128216782e-05,
+      "loss": 0.8084,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3436364378751378,
+      "grad_norm": 0.22019585967063904,
+      "learning_rate": 7.507526104021141e-05,
+      "loss": 0.7823,
+      "step": 1052
+    },
+    {
+      "epoch": 0.343963088481483,
+      "grad_norm": 0.24627773463726044,
+      "learning_rate": 7.503011350070579e-05,
+      "loss": 0.87,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3442897390878282,
+      "grad_norm": 0.2620931565761566,
+      "learning_rate": 7.498493871279967e-05,
+      "loss": 0.8776,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34461638969417335,
+      "grad_norm": 0.2705698609352112,
+      "learning_rate": 7.493973672567144e-05,
+      "loss": 0.8141,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34494304030051853,
+      "grad_norm": 0.2702180743217468,
+      "learning_rate": 7.48945075885291e-05,
+      "loss": 0.871,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34526969090686377,
+      "grad_norm": 0.29457515478134155,
+      "learning_rate": 7.484925135061022e-05,
+      "loss": 0.9953,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34559634151320895,
+      "grad_norm": 0.27578428387641907,
+      "learning_rate": 7.480396806118186e-05,
+      "loss": 0.8458,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34592299211955413,
+      "grad_norm": 0.3459113836288452,
+      "learning_rate": 7.475865776954051e-05,
+      "loss": 0.9171,
+      "step": 1059
+    },
+    {
+      "epoch": 0.3462496427258993,
+      "grad_norm": 0.300814151763916,
+      "learning_rate": 7.47133205250121e-05,
+      "loss": 0.9346,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3465762933322445,
+      "grad_norm": 0.31863105297088623,
+      "learning_rate": 7.466795637695184e-05,
+      "loss": 0.9133,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34690294393858967,
+      "grad_norm": 0.3595605790615082,
+      "learning_rate": 7.462256537474429e-05,
+      "loss": 0.9152,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34722959454493485,
+      "grad_norm": 0.3383408784866333,
+      "learning_rate": 7.457714756780322e-05,
+      "loss": 0.8437,
+      "step": 1063
+    },
+    {
+      "epoch": 0.3475562451512801,
+      "grad_norm": 0.3689708709716797,
+      "learning_rate": 7.453170300557156e-05,
+      "loss": 1.0964,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34788289575762527,
+      "grad_norm": 0.3869151175022125,
+      "learning_rate": 7.448623173752139e-05,
+      "loss": 1.0209,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34820954636397045,
+      "grad_norm": 0.4113345146179199,
+      "learning_rate": 7.444073381315388e-05,
+      "loss": 1.0003,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3485361969703156,
+      "grad_norm": 0.4470776915550232,
+      "learning_rate": 7.439520928199917e-05,
+      "loss": 1.0222,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3488628475766608,
+      "grad_norm": 0.4541364908218384,
+      "learning_rate": 7.434965819361638e-05,
+      "loss": 0.9672,
+      "step": 1068
+    },
+    {
+      "epoch": 0.349189498183006,
+      "grad_norm": 0.4776630103588104,
+      "learning_rate": 7.430408059759357e-05,
+      "loss": 1.1079,
+      "step": 1069
+    },
+    {
+      "epoch": 0.34951614878935117,
+      "grad_norm": 0.5205119252204895,
+      "learning_rate": 7.425847654354764e-05,
+      "loss": 0.9712,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3498427993956964,
+      "grad_norm": 0.58984375,
+      "learning_rate": 7.421284608112431e-05,
+      "loss": 1.2611,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3501694500020416,
+      "grad_norm": 0.8614017367362976,
+      "learning_rate": 7.416718925999797e-05,
+      "loss": 1.2577,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35049610060838676,
+      "grad_norm": 0.9795945286750793,
+      "learning_rate": 7.412150612987182e-05,
+      "loss": 1.2588,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35082275121473194,
+      "grad_norm": 1.1379746198654175,
+      "learning_rate": 7.407579674047763e-05,
+      "loss": 1.1972,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3511494018210771,
+      "grad_norm": 1.3634546995162964,
+      "learning_rate": 7.403006114157575e-05,
+      "loss": 1.9137,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3514760524274223,
+      "grad_norm": 0.19136269390583038,
+      "learning_rate": 7.398429938295511e-05,
+      "loss": 0.6746,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3518027030337675,
+      "grad_norm": 0.25908541679382324,
+      "learning_rate": 7.393851151443307e-05,
+      "loss": 0.8532,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3521293536401127,
+      "grad_norm": 0.2595674991607666,
+      "learning_rate": 7.389269758585546e-05,
+      "loss": 0.8023,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3524560042464579,
+      "grad_norm": 0.27125805616378784,
+      "learning_rate": 7.384685764709645e-05,
+      "loss": 0.8122,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3527826548528031,
+      "grad_norm": 0.282625675201416,
+      "learning_rate": 7.380099174805852e-05,
+      "loss": 0.878,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35310930545914826,
+      "grad_norm": 0.28404882550239563,
+      "learning_rate": 7.375509993867242e-05,
+      "loss": 0.7957,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35343595606549344,
+      "grad_norm": 0.3119284212589264,
+      "learning_rate": 7.370918226889713e-05,
+      "loss": 0.8864,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3537626066718386,
+      "grad_norm": 0.3108453154563904,
+      "learning_rate": 7.366323878871973e-05,
+      "loss": 0.9002,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3540892572781838,
+      "grad_norm": 0.3398928642272949,
+      "learning_rate": 7.361726954815547e-05,
+      "loss": 0.8979,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35441590788452904,
+      "grad_norm": 0.33085235953330994,
+      "learning_rate": 7.357127459724755e-05,
+      "loss": 0.9091,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3547425584908742,
+      "grad_norm": 0.39169177412986755,
+      "learning_rate": 7.352525398606724e-05,
+      "loss": 0.8724,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3550692090972194,
+      "grad_norm": 0.3578439950942993,
+      "learning_rate": 7.347920776471374e-05,
+      "loss": 0.8364,
+      "step": 1087
+    },
+    {
+      "epoch": 0.3553958597035646,
+      "grad_norm": 0.34978634119033813,
+      "learning_rate": 7.343313598331406e-05,
+      "loss": 0.8644,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35572251030990976,
+      "grad_norm": 0.40582436323165894,
+      "learning_rate": 7.33870386920231e-05,
+      "loss": 1.0534,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35604916091625494,
+      "grad_norm": 0.4203055202960968,
+      "learning_rate": 7.33409159410235e-05,
+      "loss": 0.9638,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3563758115226001,
+      "grad_norm": 0.45389387011528015,
+      "learning_rate": 7.329476778052565e-05,
+      "loss": 1.0034,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35670246212894535,
+      "grad_norm": 0.47722360491752625,
+      "learning_rate": 7.324859426076756e-05,
+      "loss": 0.9551,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35702911273529053,
+      "grad_norm": 0.4822474718093872,
+      "learning_rate": 7.320239543201489e-05,
+      "loss": 1.0724,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3573557633416357,
+      "grad_norm": 0.559536337852478,
+      "learning_rate": 7.315617134456079e-05,
+      "loss": 1.2312,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3576824139479809,
+      "grad_norm": 0.5810602307319641,
+      "learning_rate": 7.310992204872595e-05,
+      "loss": 0.9705,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3580090645543261,
+      "grad_norm": 0.7775752544403076,
+      "learning_rate": 7.306364759485853e-05,
+      "loss": 1.3561,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35833571516067125,
+      "grad_norm": 0.8379966020584106,
+      "learning_rate": 7.301734803333403e-05,
+      "loss": 1.2233,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35866236576701643,
+      "grad_norm": 0.927143931388855,
+      "learning_rate": 7.297102341455528e-05,
+      "loss": 1.3993,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35898901637336167,
+      "grad_norm": 1.2338497638702393,
+      "learning_rate": 7.292467378895243e-05,
+      "loss": 1.2431,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35931566697970685,
+      "grad_norm": 1.6832832098007202,
+      "learning_rate": 7.28782992069828e-05,
+      "loss": 1.5705,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35964231758605203,
+      "grad_norm": 0.2085774838924408,
+      "learning_rate": 7.283189971913094e-05,
+      "loss": 0.8405,
+      "step": 1101
+    },
+    {
+      "epoch": 0.3599689681923972,
+      "grad_norm": 0.23439809679985046,
+      "learning_rate": 7.278547537590845e-05,
+      "loss": 0.7714,
+      "step": 1102
+    },
+    {
+      "epoch": 0.3602956187987424,
+      "grad_norm": 0.25842559337615967,
+      "learning_rate": 7.273902622785405e-05,
+      "loss": 0.8394,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36062226940508757,
+      "grad_norm": 0.2644081115722656,
+      "learning_rate": 7.269255232553339e-05,
+      "loss": 0.8266,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36094892001143275,
+      "grad_norm": 0.26839661598205566,
+      "learning_rate": 7.264605371953915e-05,
+      "loss": 0.8877,
+      "step": 1105
+    },
+    {
+      "epoch": 0.361275570617778,
+      "grad_norm": 0.2822781503200531,
+      "learning_rate": 7.259953046049084e-05,
+      "loss": 0.9052,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36160222122412317,
+      "grad_norm": 0.30244433879852295,
+      "learning_rate": 7.255298259903482e-05,
+      "loss": 0.9289,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36192887183046835,
+      "grad_norm": 0.29979464411735535,
+      "learning_rate": 7.250641018584428e-05,
+      "loss": 0.9039,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3622555224368135,
+      "grad_norm": 0.31162193417549133,
+      "learning_rate": 7.245981327161905e-05,
+      "loss": 0.968,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3625821730431587,
+      "grad_norm": 0.32680559158325195,
+      "learning_rate": 7.241319190708575e-05,
+      "loss": 0.8592,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3629088236495039,
+      "grad_norm": 0.332279235124588,
+      "learning_rate": 7.236654614299748e-05,
+      "loss": 0.8753,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36323547425584907,
+      "grad_norm": 0.36247286200523376,
+      "learning_rate": 7.231987603013401e-05,
+      "loss": 0.9475,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3635621248621943,
+      "grad_norm": 0.3807973563671112,
+      "learning_rate": 7.227318161930157e-05,
+      "loss": 0.9499,
+      "step": 1113
+    },
+    {
+      "epoch": 0.3638887754685395,
+      "grad_norm": 0.40924564003944397,
+      "learning_rate": 7.222646296133287e-05,
+      "loss": 0.9879,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36421542607488466,
+      "grad_norm": 0.43880122900009155,
+      "learning_rate": 7.217972010708696e-05,
+      "loss": 0.9948,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36454207668122984,
+      "grad_norm": 0.4344307780265808,
+      "learning_rate": 7.213295310744928e-05,
+      "loss": 1.0174,
+      "step": 1116
+    },
+    {
+      "epoch": 0.364868727287575,
+      "grad_norm": 0.4451688528060913,
+      "learning_rate": 7.208616201333156e-05,
+      "loss": 0.9336,
+      "step": 1117
+    },
+    {
+      "epoch": 0.3651953778939202,
+      "grad_norm": 0.5209891200065613,
+      "learning_rate": 7.203934687567173e-05,
+      "loss": 1.0403,
+      "step": 1118
+    },
+    {
+      "epoch": 0.3655220285002654,
+      "grad_norm": 0.5560318231582642,
+      "learning_rate": 7.199250774543391e-05,
+      "loss": 1.0855,
+      "step": 1119
+    },
+    {
+      "epoch": 0.3658486791066106,
+      "grad_norm": 0.6434129476547241,
+      "learning_rate": 7.194564467360834e-05,
+      "loss": 1.1184,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3661753297129558,
+      "grad_norm": 0.8035168051719666,
+      "learning_rate": 7.189875771121129e-05,
+      "loss": 1.1999,
+      "step": 1121
+    },
+    {
+      "epoch": 0.366501980319301,
+      "grad_norm": 0.9151061177253723,
+      "learning_rate": 7.18518469092851e-05,
+      "loss": 1.2509,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36682863092564616,
+      "grad_norm": 1.119385004043579,
+      "learning_rate": 7.180491231889802e-05,
+      "loss": 1.6826,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36715528153199134,
+      "grad_norm": 1.2659269571304321,
+      "learning_rate": 7.17579539911442e-05,
+      "loss": 1.6055,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3674819321383365,
+      "grad_norm": 1.693357229232788,
+      "learning_rate": 7.171097197714363e-05,
+      "loss": 1.8833,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3678085827446817,
+      "grad_norm": 0.19237908720970154,
+      "learning_rate": 7.166396632804212e-05,
+      "loss": 0.646,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3681352333510269,
+      "grad_norm": 0.25181078910827637,
+      "learning_rate": 7.161693709501114e-05,
+      "loss": 0.8216,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3684618839573721,
+      "grad_norm": 0.26569420099258423,
+      "learning_rate": 7.156988432924791e-05,
+      "loss": 0.8657,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3687885345637173,
+      "grad_norm": 0.2794082462787628,
+      "learning_rate": 7.152280808197522e-05,
+      "loss": 0.7595,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3691151851700625,
+      "grad_norm": 0.2683536112308502,
+      "learning_rate": 7.147570840444145e-05,
+      "loss": 0.8327,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36944183577640766,
+      "grad_norm": 0.27695196866989136,
+      "learning_rate": 7.142858534792045e-05,
+      "loss": 0.8344,
+      "step": 1131
+    },
+    {
+      "epoch": 0.36976848638275284,
+      "grad_norm": 0.28606536984443665,
+      "learning_rate": 7.138143896371157e-05,
+      "loss": 0.8636,
+      "step": 1132
+    },
+    {
+      "epoch": 0.370095136989098,
+      "grad_norm": 0.2818695902824402,
+      "learning_rate": 7.133426930313951e-05,
+      "loss": 0.8219,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3704217875954432,
+      "grad_norm": 0.3215191960334778,
+      "learning_rate": 7.128707641755434e-05,
+      "loss": 0.9497,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37074843820178843,
+      "grad_norm": 0.3110451400279999,
+      "learning_rate": 7.123986035833141e-05,
+      "loss": 0.8942,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3710750888081336,
+      "grad_norm": 0.31207698583602905,
+      "learning_rate": 7.119262117687127e-05,
+      "loss": 0.9575,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3714017394144788,
+      "grad_norm": 0.314654678106308,
+      "learning_rate": 7.114535892459967e-05,
+      "loss": 0.8195,
+      "step": 1137
+    },
+    {
+      "epoch": 0.371728390020824,
+      "grad_norm": 0.34628891944885254,
+      "learning_rate": 7.109807365296748e-05,
+      "loss": 0.8786,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37205504062716915,
+      "grad_norm": 0.3640112578868866,
+      "learning_rate": 7.105076541345058e-05,
+      "loss": 0.9262,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37238169123351433,
+      "grad_norm": 0.4010256230831146,
+      "learning_rate": 7.100343425754993e-05,
+      "loss": 0.9402,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3727083418398595,
+      "grad_norm": 0.4014125466346741,
+      "learning_rate": 7.095608023679138e-05,
+      "loss": 1.0435,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37303499244620475,
+      "grad_norm": 0.46626025438308716,
+      "learning_rate": 7.090870340272568e-05,
+      "loss": 0.992,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37336164305254993,
+      "grad_norm": 0.5319213271141052,
+      "learning_rate": 7.086130380692841e-05,
+      "loss": 1.1192,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3736882936588951,
+      "grad_norm": 0.5756850838661194,
+      "learning_rate": 7.081388150099999e-05,
+      "loss": 1.1887,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3740149442652403,
+      "grad_norm": 0.6586177349090576,
+      "learning_rate": 7.076643653656549e-05,
+      "loss": 1.2207,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37434159487158547,
+      "grad_norm": 0.6644164323806763,
+      "learning_rate": 7.071896896527464e-05,
+      "loss": 1.194,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37466824547793065,
+      "grad_norm": 0.9249393939971924,
+      "learning_rate": 7.067147883880185e-05,
+      "loss": 1.3419,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37499489608427583,
+      "grad_norm": 1.1207913160324097,
+      "learning_rate": 7.062396620884605e-05,
+      "loss": 1.5119,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37532154669062107,
+      "grad_norm": 1.3481462001800537,
+      "learning_rate": 7.057643112713063e-05,
+      "loss": 1.2274,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37564819729696625,
+      "grad_norm": 1.6863646507263184,
+      "learning_rate": 7.05288736454035e-05,
+      "loss": 2.1351,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3759748479033114,
+      "grad_norm": 0.19038569927215576,
+      "learning_rate": 7.048129381543687e-05,
+      "loss": 0.6988,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3763014985096566,
+      "grad_norm": 0.2359238564968109,
+      "learning_rate": 7.043369168902732e-05,
+      "loss": 0.8121,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3766281491160018,
+      "grad_norm": 0.25924691557884216,
+      "learning_rate": 7.038606731799574e-05,
+      "loss": 0.8618,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37695479972234697,
+      "grad_norm": 0.24869892001152039,
+      "learning_rate": 7.033842075418718e-05,
+      "loss": 0.8703,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37728145032869215,
+      "grad_norm": 0.2840399742126465,
+      "learning_rate": 7.029075204947085e-05,
+      "loss": 0.7961,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3776081009350374,
+      "grad_norm": 0.2748403549194336,
+      "learning_rate": 7.024306125574009e-05,
+      "loss": 0.9051,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37793475154138256,
+      "grad_norm": 0.284496009349823,
+      "learning_rate": 7.019534842491228e-05,
+      "loss": 0.8884,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37826140214772774,
+      "grad_norm": 0.3085760176181793,
+      "learning_rate": 7.014761360892882e-05,
+      "loss": 0.9337,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3785880527540729,
+      "grad_norm": 0.3391752243041992,
+      "learning_rate": 7.009985685975495e-05,
+      "loss": 0.9858,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3789147033604181,
+      "grad_norm": 0.3305496871471405,
+      "learning_rate": 7.005207822937988e-05,
+      "loss": 0.8196,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3792413539667633,
+      "grad_norm": 0.34923458099365234,
+      "learning_rate": 7.00042777698166e-05,
+      "loss": 0.908,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37956800457310846,
+      "grad_norm": 0.41269630193710327,
+      "learning_rate": 6.99564555331019e-05,
+      "loss": 0.9519,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3798946551794537,
+      "grad_norm": 0.3687300682067871,
+      "learning_rate": 6.990861157129622e-05,
+      "loss": 1.0597,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3802213057857989,
+      "grad_norm": 0.40411537885665894,
+      "learning_rate": 6.986074593648367e-05,
+      "loss": 0.9359,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38054795639214406,
+      "grad_norm": 0.4321344792842865,
+      "learning_rate": 6.981285868077198e-05,
+      "loss": 1.0351,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38087460699848924,
+      "grad_norm": 0.4461156725883484,
+      "learning_rate": 6.976494985629242e-05,
+      "loss": 1.0586,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3812012576048344,
+      "grad_norm": 0.5063657164573669,
+      "learning_rate": 6.971701951519972e-05,
+      "loss": 1.163,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3815279082111796,
+      "grad_norm": 0.5494873523712158,
+      "learning_rate": 6.966906770967199e-05,
+      "loss": 1.1235,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3818545588175248,
+      "grad_norm": 0.6220507621765137,
+      "learning_rate": 6.962109449191077e-05,
+      "loss": 1.2436,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38218120942387,
+      "grad_norm": 0.6774858832359314,
+      "learning_rate": 6.957309991414092e-05,
+      "loss": 1.3258,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3825078600302152,
+      "grad_norm": 0.8799903392791748,
+      "learning_rate": 6.952508402861051e-05,
+      "loss": 1.4338,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3828345106365604,
+      "grad_norm": 1.1180598735809326,
+      "learning_rate": 6.94770468875908e-05,
+      "loss": 1.8223,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38316116124290556,
+      "grad_norm": 1.154464602470398,
+      "learning_rate": 6.942898854337621e-05,
+      "loss": 1.3532,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38348781184925074,
+      "grad_norm": 1.271112084388733,
+      "learning_rate": 6.938090904828428e-05,
+      "loss": 1.5536,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3838144624555959,
+      "grad_norm": 1.7882853746414185,
+      "learning_rate": 6.933280845465551e-05,
+      "loss": 1.9543,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3841411130619411,
+      "grad_norm": 0.1903381198644638,
+      "learning_rate": 6.92846868148534e-05,
+      "loss": 0.6825,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38446776366828633,
+      "grad_norm": 0.21171070635318756,
+      "learning_rate": 6.923654418126434e-05,
+      "loss": 0.7353,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3847944142746315,
+      "grad_norm": 0.22646358609199524,
+      "learning_rate": 6.918838060629762e-05,
+      "loss": 0.7659,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3851210648809767,
+      "grad_norm": 0.2556747496128082,
+      "learning_rate": 6.914019614238527e-05,
+      "loss": 0.8941,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3854477154873219,
+      "grad_norm": 0.278025358915329,
+      "learning_rate": 6.909199084198212e-05,
+      "loss": 0.8642,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38577436609366705,
+      "grad_norm": 0.2806093692779541,
+      "learning_rate": 6.904376475756563e-05,
+      "loss": 0.8426,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38610101670001223,
+      "grad_norm": 0.3061642646789551,
+      "learning_rate": 6.899551794163592e-05,
+      "loss": 0.9506,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3864276673063574,
+      "grad_norm": 0.31807222962379456,
+      "learning_rate": 6.894725044671566e-05,
+      "loss": 0.896,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38675431791270265,
+      "grad_norm": 0.330069899559021,
+      "learning_rate": 6.889896232535004e-05,
+      "loss": 0.9255,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38708096851904783,
+      "grad_norm": 0.3431348502635956,
+      "learning_rate": 6.885065363010671e-05,
+      "loss": 0.9511,
+      "step": 1185
+    },
+    {
+      "epoch": 0.387407619125393,
+      "grad_norm": 0.3761681020259857,
+      "learning_rate": 6.88023244135757e-05,
+      "loss": 0.9307,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3877342697317382,
+      "grad_norm": 0.352243572473526,
+      "learning_rate": 6.875397472836937e-05,
+      "loss": 0.9937,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38806092033808337,
+      "grad_norm": 0.38508570194244385,
+      "learning_rate": 6.870560462712243e-05,
+      "loss": 0.95,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38838757094442855,
+      "grad_norm": 0.41849932074546814,
+      "learning_rate": 6.865721416249175e-05,
+      "loss": 1.0807,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38871422155077373,
+      "grad_norm": 0.4687519669532776,
+      "learning_rate": 6.860880338715638e-05,
+      "loss": 1.0743,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38904087215711897,
+      "grad_norm": 0.4491303563117981,
+      "learning_rate": 6.856037235381746e-05,
+      "loss": 1.1039,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38936752276346415,
+      "grad_norm": 0.5182649493217468,
+      "learning_rate": 6.851192111519826e-05,
+      "loss": 0.9833,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3896941733698093,
+      "grad_norm": 0.5373132824897766,
+      "learning_rate": 6.846344972404399e-05,
+      "loss": 1.1743,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3900208239761545,
+      "grad_norm": 0.5523989796638489,
+      "learning_rate": 6.841495823312177e-05,
+      "loss": 1.1548,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3903474745824997,
+      "grad_norm": 0.6846272349357605,
+      "learning_rate": 6.836644669522065e-05,
+      "loss": 1.2027,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39067412518884487,
+      "grad_norm": 0.7508804202079773,
+      "learning_rate": 6.831791516315151e-05,
+      "loss": 1.2353,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39100077579519005,
+      "grad_norm": 1.0120075941085815,
+      "learning_rate": 6.826936368974696e-05,
+      "loss": 1.6322,
+      "step": 1197
+    },
+    {
+      "epoch": 0.3913274264015353,
+      "grad_norm": 1.1982024908065796,
+      "learning_rate": 6.822079232786134e-05,
+      "loss": 1.674,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39165407700788046,
+      "grad_norm": 1.4332988262176514,
+      "learning_rate": 6.817220113037062e-05,
+      "loss": 1.2769,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39198072761422564,
+      "grad_norm": 1.8049565553665161,
+      "learning_rate": 6.81235901501724e-05,
+      "loss": 1.7269,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3923073782205708,
+      "grad_norm": 0.20991112291812897,
+      "learning_rate": 6.807495944018577e-05,
+      "loss": 0.7704,
+      "step": 1201
+    },
+    {
+      "epoch": 0.392634028826916,
+      "grad_norm": 0.23148652911186218,
+      "learning_rate": 6.802630905335137e-05,
+      "loss": 0.7713,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3929606794332612,
+      "grad_norm": 0.24124076962471008,
+      "learning_rate": 6.797763904263115e-05,
+      "loss": 0.8331,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39328733003960636,
+      "grad_norm": 0.2620198130607605,
+      "learning_rate": 6.792894946100854e-05,
+      "loss": 0.8113,
+      "step": 1204
+    },
+    {
+      "epoch": 0.3936139806459516,
+      "grad_norm": 0.26299262046813965,
+      "learning_rate": 6.788024036148821e-05,
+      "loss": 0.8568,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3939406312522968,
+      "grad_norm": 0.27623122930526733,
+      "learning_rate": 6.783151179709609e-05,
+      "loss": 0.8808,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39426728185864196,
+      "grad_norm": 0.3048432171344757,
+      "learning_rate": 6.778276382087926e-05,
+      "loss": 0.9913,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39459393246498714,
+      "grad_norm": 0.3026992678642273,
+      "learning_rate": 6.773399648590602e-05,
+      "loss": 0.9271,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3949205830713323,
+      "grad_norm": 0.3055267632007599,
+      "learning_rate": 6.768520984526569e-05,
+      "loss": 0.8672,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3952472336776775,
+      "grad_norm": 0.2968289256095886,
+      "learning_rate": 6.76364039520686e-05,
+      "loss": 0.7734,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3955738842840227,
+      "grad_norm": 0.32888948917388916,
+      "learning_rate": 6.758757885944608e-05,
+      "loss": 0.9411,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3959005348903679,
+      "grad_norm": 0.334842711687088,
+      "learning_rate": 6.75387346205503e-05,
+      "loss": 0.9304,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3962271854967131,
+      "grad_norm": 0.3538110852241516,
+      "learning_rate": 6.74898712885543e-05,
+      "loss": 1.0136,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3965538361030583,
+      "grad_norm": 0.4119766652584076,
+      "learning_rate": 6.744098891665194e-05,
+      "loss": 0.9688,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39688048670940346,
+      "grad_norm": 0.40925925970077515,
+      "learning_rate": 6.739208755805778e-05,
+      "loss": 0.9954,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39720713731574864,
+      "grad_norm": 0.4552319049835205,
+      "learning_rate": 6.734316726600702e-05,
+      "loss": 0.96,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3975337879220938,
+      "grad_norm": 0.45739418268203735,
+      "learning_rate": 6.729422809375551e-05,
+      "loss": 0.9417,
+      "step": 1217
+    },
+    {
+      "epoch": 0.397860438528439,
+      "grad_norm": 0.5212276577949524,
+      "learning_rate": 6.724527009457966e-05,
+      "loss": 1.0401,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39818708913478423,
+      "grad_norm": 0.5565895438194275,
+      "learning_rate": 6.719629332177634e-05,
+      "loss": 1.1682,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3985137397411294,
+      "grad_norm": 0.6466187834739685,
+      "learning_rate": 6.714729782866291e-05,
+      "loss": 1.2577,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3988403903474746,
+      "grad_norm": 0.7221575975418091,
+      "learning_rate": 6.709828366857702e-05,
+      "loss": 1.2678,
+      "step": 1221
+    },
+    {
+      "epoch": 0.3991670409538198,
+      "grad_norm": 0.9043142199516296,
+      "learning_rate": 6.704925089487675e-05,
+      "loss": 1.3762,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39949369156016495,
+      "grad_norm": 1.0332460403442383,
+      "learning_rate": 6.700019956094035e-05,
+      "loss": 1.2477,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39982034216651013,
+      "grad_norm": 1.4478338956832886,
+      "learning_rate": 6.695112972016633e-05,
+      "loss": 1.4505,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4001469927728553,
+      "grad_norm": 1.8933912515640259,
+      "learning_rate": 6.690204142597333e-05,
+      "loss": 1.4914,
+      "step": 1225
+    },
+    {
+      "epoch": 0.40047364337920055,
+      "grad_norm": 0.21706634759902954,
+      "learning_rate": 6.68529347318001e-05,
+      "loss": 0.7785,
+      "step": 1226
+    },
+    {
+      "epoch": 0.40080029398554573,
+      "grad_norm": 0.2556523084640503,
+      "learning_rate": 6.680380969110537e-05,
+      "loss": 0.8661,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4011269445918909,
+      "grad_norm": 0.2508922815322876,
+      "learning_rate": 6.67546663573679e-05,
+      "loss": 0.86,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4014535951982361,
+      "grad_norm": 0.2551276385784149,
+      "learning_rate": 6.670550478408632e-05,
+      "loss": 0.8221,
+      "step": 1229
+    },
+    {
+      "epoch": 0.40178024580458127,
+      "grad_norm": 0.2717500627040863,
+      "learning_rate": 6.665632502477914e-05,
+      "loss": 0.8709,
+      "step": 1230
+    },
+    {
+      "epoch": 0.40210689641092645,
+      "grad_norm": 0.2777538299560547,
+      "learning_rate": 6.660712713298468e-05,
+      "loss": 0.8241,
+      "step": 1231
+    },
+    {
+      "epoch": 0.40243354701727163,
+      "grad_norm": 0.29997768998146057,
+      "learning_rate": 6.655791116226094e-05,
+      "loss": 0.8535,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4027601976236168,
+      "grad_norm": 0.3282143771648407,
+      "learning_rate": 6.650867716618567e-05,
+      "loss": 0.9073,
+      "step": 1233
+    },
+    {
+      "epoch": 0.40308684822996205,
+      "grad_norm": 0.3342377841472626,
+      "learning_rate": 6.645942519835623e-05,
+      "loss": 0.9987,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4034134988363072,
+      "grad_norm": 0.31825751066207886,
+      "learning_rate": 6.64101553123895e-05,
+      "loss": 0.7599,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4037401494426524,
+      "grad_norm": 0.3360239863395691,
+      "learning_rate": 6.636086756192193e-05,
+      "loss": 0.7941,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4040668000489976,
+      "grad_norm": 0.3323500156402588,
+      "learning_rate": 6.631156200060935e-05,
+      "loss": 0.945,
+      "step": 1237
+    },
+    {
+      "epoch": 0.40439345065534277,
+      "grad_norm": 0.4027165174484253,
+      "learning_rate": 6.626223868212702e-05,
+      "loss": 0.9231,
+      "step": 1238
+    },
+    {
+      "epoch": 0.40472010126168795,
+      "grad_norm": 0.42029932141304016,
+      "learning_rate": 6.621289766016955e-05,
+      "loss": 0.9594,
+      "step": 1239
+    },
+    {
+      "epoch": 0.40504675186803313,
+      "grad_norm": 0.4373820424079895,
+      "learning_rate": 6.616353898845076e-05,
+      "loss": 0.9466,
+      "step": 1240
+    },
+    {
+      "epoch": 0.40537340247437836,
+      "grad_norm": 0.48459959030151367,
+      "learning_rate": 6.611416272070377e-05,
+      "loss": 0.9867,
+      "step": 1241
+    },
+    {
+      "epoch": 0.40570005308072354,
+      "grad_norm": 0.4633182883262634,
+      "learning_rate": 6.606476891068074e-05,
+      "loss": 0.9491,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4060267036870687,
+      "grad_norm": 0.5320441722869873,
+      "learning_rate": 6.601535761215305e-05,
+      "loss": 1.1246,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4063533542934139,
+      "grad_norm": 0.5146458148956299,
+      "learning_rate": 6.596592887891103e-05,
+      "loss": 1.0492,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4066800048997591,
+      "grad_norm": 0.6374284029006958,
+      "learning_rate": 6.591648276476402e-05,
+      "loss": 1.1577,
+      "step": 1245
+    },
+    {
+      "epoch": 0.40700665550610426,
+      "grad_norm": 0.6750211715698242,
+      "learning_rate": 6.586701932354031e-05,
+      "loss": 1.2225,
+      "step": 1246
+    },
+    {
+      "epoch": 0.40733330611244944,
+      "grad_norm": 0.9183369278907776,
+      "learning_rate": 6.581753860908699e-05,
+      "loss": 1.301,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4076599567187947,
+      "grad_norm": 1.0853521823883057,
+      "learning_rate": 6.576804067527002e-05,
+      "loss": 1.5473,
+      "step": 1248
+    },
+    {
+      "epoch": 0.40798660732513986,
+      "grad_norm": 1.1827778816223145,
+      "learning_rate": 6.571852557597407e-05,
+      "loss": 1.4952,
+      "step": 1249
+    },
+    {
+      "epoch": 0.40831325793148504,
+      "grad_norm": 1.3170537948608398,
+      "learning_rate": 6.566899336510248e-05,
+      "loss": 1.6403,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4086399085378302,
+      "grad_norm": 0.2041776031255722,
+      "learning_rate": 6.561944409657726e-05,
+      "loss": 0.6767,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4089665591441754,
+      "grad_norm": 0.23141330480575562,
+      "learning_rate": 6.556987782433894e-05,
+      "loss": 0.8457,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4092932097505206,
+      "grad_norm": 0.25107938051223755,
+      "learning_rate": 6.552029460234664e-05,
+      "loss": 0.8766,
+      "step": 1253
+    },
+    {
+      "epoch": 0.40961986035686576,
+      "grad_norm": 0.26021620631217957,
+      "learning_rate": 6.547069448457785e-05,
+      "loss": 0.9075,
+      "step": 1254
+    },
+    {
+      "epoch": 0.409946510963211,
+      "grad_norm": 0.29366645216941833,
+      "learning_rate": 6.542107752502848e-05,
+      "loss": 0.8482,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4102731615695562,
+      "grad_norm": 0.2926657497882843,
+      "learning_rate": 6.537144377771279e-05,
+      "loss": 0.8924,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41059981217590136,
+      "grad_norm": 0.30195921659469604,
+      "learning_rate": 6.53217932966633e-05,
+      "loss": 0.9467,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41092646278224654,
+      "grad_norm": 0.29315873980522156,
+      "learning_rate": 6.527212613593074e-05,
+      "loss": 0.7876,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4112531133885917,
+      "grad_norm": 0.29667970538139343,
+      "learning_rate": 6.522244234958404e-05,
+      "loss": 0.8602,
+      "step": 1259
+    },
+    {
+      "epoch": 0.4115797639949369,
+      "grad_norm": 0.3073457181453705,
+      "learning_rate": 6.517274199171019e-05,
+      "loss": 0.9042,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4119064146012821,
+      "grad_norm": 0.3291796147823334,
+      "learning_rate": 6.512302511641419e-05,
+      "loss": 0.9297,
+      "step": 1261
+    },
+    {
+      "epoch": 0.4122330652076273,
+      "grad_norm": 0.36080867052078247,
+      "learning_rate": 6.507329177781911e-05,
+      "loss": 0.8909,
+      "step": 1262
+    },
+    {
+      "epoch": 0.4125597158139725,
+      "grad_norm": 0.39371705055236816,
+      "learning_rate": 6.502354203006588e-05,
+      "loss": 1.0463,
+      "step": 1263
+    },
+    {
+      "epoch": 0.4128863664203177,
+      "grad_norm": 0.3974571228027344,
+      "learning_rate": 6.497377592731329e-05,
+      "loss": 0.9081,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41321301702666285,
+      "grad_norm": 0.41697776317596436,
+      "learning_rate": 6.492399352373795e-05,
+      "loss": 0.9524,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41353966763300803,
+      "grad_norm": 0.4638747572898865,
+      "learning_rate": 6.487419487353421e-05,
+      "loss": 1.0998,
+      "step": 1266
+    },
+    {
+      "epoch": 0.4138663182393532,
+      "grad_norm": 0.5170689821243286,
+      "learning_rate": 6.482438003091414e-05,
+      "loss": 1.212,
+      "step": 1267
+    },
+    {
+      "epoch": 0.4141929688456984,
+      "grad_norm": 0.5564174652099609,
+      "learning_rate": 6.47745490501074e-05,
+      "loss": 1.0914,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41451961945204363,
+      "grad_norm": 0.7147950530052185,
+      "learning_rate": 6.47247019853612e-05,
+      "loss": 1.2383,
+      "step": 1269
+    },
+    {
+      "epoch": 0.4148462700583888,
+      "grad_norm": 0.6999964118003845,
+      "learning_rate": 6.467483889094033e-05,
+      "loss": 1.2716,
+      "step": 1270
+    },
+    {
+      "epoch": 0.415172920664734,
+      "grad_norm": 0.9136797785758972,
+      "learning_rate": 6.462495982112697e-05,
+      "loss": 1.1405,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41549957127107917,
+      "grad_norm": 0.9877931475639343,
+      "learning_rate": 6.457506483022068e-05,
+      "loss": 1.5912,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41582622187742435,
+      "grad_norm": 1.415334939956665,
+      "learning_rate": 6.452515397253844e-05,
+      "loss": 1.7894,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41615287248376953,
+      "grad_norm": 1.3017021417617798,
+      "learning_rate": 6.44752273024144e-05,
+      "loss": 1.4178,
+      "step": 1274
+    },
+    {
+      "epoch": 0.4164795230901147,
+      "grad_norm": 1.7668936252593994,
+      "learning_rate": 6.442528487419996e-05,
+      "loss": 1.9307,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41680617369645995,
+      "grad_norm": 0.20308293402194977,
+      "learning_rate": 6.437532674226372e-05,
+      "loss": 0.7709,
+      "step": 1276
+    },
+    {
+      "epoch": 0.4171328243028051,
+      "grad_norm": 0.21541514992713928,
+      "learning_rate": 6.432535296099132e-05,
+      "loss": 0.7705,
+      "step": 1277
+    },
+    {
+      "epoch": 0.4174594749091503,
+      "grad_norm": 0.24598878622055054,
+      "learning_rate": 6.427536358478542e-05,
+      "loss": 0.8412,
+      "step": 1278
+    },
+    {
+      "epoch": 0.4177861255154955,
+      "grad_norm": 0.2694220244884491,
+      "learning_rate": 6.422535866806576e-05,
+      "loss": 0.8584,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41811277612184067,
+      "grad_norm": 0.2715321183204651,
+      "learning_rate": 6.417533826526888e-05,
+      "loss": 0.8804,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41843942672818585,
+      "grad_norm": 0.26800206303596497,
+      "learning_rate": 6.412530243084824e-05,
+      "loss": 0.7912,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41876607733453103,
+      "grad_norm": 0.2986787259578705,
+      "learning_rate": 6.407525121927409e-05,
+      "loss": 0.8956,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41909272794087626,
+      "grad_norm": 0.3155290186405182,
+      "learning_rate": 6.40251846850334e-05,
+      "loss": 0.8807,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41941937854722144,
+      "grad_norm": 0.3152468502521515,
+      "learning_rate": 6.397510288262986e-05,
+      "loss": 0.8617,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4197460291535666,
+      "grad_norm": 0.33172619342803955,
+      "learning_rate": 6.392500586658376e-05,
+      "loss": 0.9051,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4200726797599118,
+      "grad_norm": 0.39893805980682373,
+      "learning_rate": 6.387489369143191e-05,
+      "loss": 1.0182,
+      "step": 1286
+    },
+    {
+      "epoch": 0.420399330366257,
+      "grad_norm": 0.34438183903694153,
+      "learning_rate": 6.38247664117277e-05,
+      "loss": 0.848,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42072598097260216,
+      "grad_norm": 0.37023016810417175,
+      "learning_rate": 6.377462408204093e-05,
+      "loss": 0.8406,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.41819244623184204,
+      "learning_rate": 6.372446675695778e-05,
+      "loss": 0.9544,
+      "step": 1289
+    },
+    {
+      "epoch": 0.4213792821852926,
+      "grad_norm": 0.4439786374568939,
+      "learning_rate": 6.367429449108072e-05,
+      "loss": 1.0086,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42170593279163776,
+      "grad_norm": 0.49234476685523987,
+      "learning_rate": 6.362410733902855e-05,
+      "loss": 1.0202,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42203258339798294,
+      "grad_norm": 0.5539708733558655,
+      "learning_rate": 6.357390535543623e-05,
+      "loss": 1.1179,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4223592340043281,
+      "grad_norm": 0.5921236276626587,
+      "learning_rate": 6.35236885949549e-05,
+      "loss": 1.3069,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4226858846106733,
+      "grad_norm": 0.5858140587806702,
+      "learning_rate": 6.347345711225176e-05,
+      "loss": 1.2039,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4230125352170185,
+      "grad_norm": 0.7424845695495605,
+      "learning_rate": 6.342321096201003e-05,
+      "loss": 1.4528,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42333918582336366,
+      "grad_norm": 0.8141841292381287,
+      "learning_rate": 6.33729501989289e-05,
+      "loss": 1.4032,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4236658364297089,
+      "grad_norm": 1.0641597509384155,
+      "learning_rate": 6.332267487772352e-05,
+      "loss": 1.4153,
+      "step": 1297
+    },
+    {
+      "epoch": 0.4239924870360541,
+      "grad_norm": 1.2870875597000122,
+      "learning_rate": 6.327238505312484e-05,
+      "loss": 1.6028,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42431913764239926,
+      "grad_norm": 1.4582892656326294,
+      "learning_rate": 6.322208077987958e-05,
+      "loss": 1.3119,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42464578824874444,
+      "grad_norm": 1.7859705686569214,
+      "learning_rate": 6.317176211275022e-05,
+      "loss": 1.7957,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4249724388550896,
+      "grad_norm": 0.19623303413391113,
+      "learning_rate": 6.312142910651492e-05,
+      "loss": 0.7682,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4252990894614348,
+      "grad_norm": 0.22996629774570465,
+      "learning_rate": 6.307108181596743e-05,
+      "loss": 0.8693,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42562574006778,
+      "grad_norm": 0.24149446189403534,
+      "learning_rate": 6.302072029591707e-05,
+      "loss": 0.8444,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4259523906741252,
+      "grad_norm": 0.25310564041137695,
+      "learning_rate": 6.297034460118861e-05,
+      "loss": 0.8612,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4262790412804704,
+      "grad_norm": 0.26213711500167847,
+      "learning_rate": 6.29199547866223e-05,
+      "loss": 0.8087,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4266056918868156,
+      "grad_norm": 0.28710299730300903,
+      "learning_rate": 6.286955090707371e-05,
+      "loss": 0.877,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42693234249316075,
+      "grad_norm": 0.3216702342033386,
+      "learning_rate": 6.281913301741378e-05,
+      "loss": 0.9834,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42725899309950593,
+      "grad_norm": 0.3331913948059082,
+      "learning_rate": 6.276870117252867e-05,
+      "loss": 1.0185,
+      "step": 1308
+    },
+    {
+      "epoch": 0.4275856437058511,
+      "grad_norm": 0.3292556405067444,
+      "learning_rate": 6.271825542731971e-05,
+      "loss": 0.9194,
+      "step": 1309
+    },
+    {
+      "epoch": 0.4279122943121963,
+      "grad_norm": 0.32852330803871155,
+      "learning_rate": 6.26677958367034e-05,
+      "loss": 0.9441,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42823894491854153,
+      "grad_norm": 0.33652690052986145,
+      "learning_rate": 6.261732245561129e-05,
+      "loss": 0.8488,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4285655955248867,
+      "grad_norm": 0.3745363652706146,
+      "learning_rate": 6.256683533898995e-05,
+      "loss": 0.8509,
+      "step": 1312
+    },
+    {
+      "epoch": 0.4288922461312319,
+      "grad_norm": 0.3785022795200348,
+      "learning_rate": 6.251633454180091e-05,
+      "loss": 0.8239,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42921889673757707,
+      "grad_norm": 0.4264315366744995,
+      "learning_rate": 6.24658201190206e-05,
+      "loss": 1.065,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42954554734392225,
+      "grad_norm": 0.43724411725997925,
+      "learning_rate": 6.241529212564025e-05,
+      "loss": 0.9565,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42987219795026743,
+      "grad_norm": 0.468350887298584,
+      "learning_rate": 6.236475061666588e-05,
+      "loss": 1.0929,
+      "step": 1316
+    },
+    {
+      "epoch": 0.4301988485566126,
+      "grad_norm": 0.5075478553771973,
+      "learning_rate": 6.231419564711826e-05,
+      "loss": 1.0679,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43052549916295785,
+      "grad_norm": 0.5075100660324097,
+      "learning_rate": 6.226362727203272e-05,
+      "loss": 1.0657,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43085214976930303,
+      "grad_norm": 0.5745466947555542,
+      "learning_rate": 6.22130455464593e-05,
+      "loss": 1.0917,
+      "step": 1319
+    },
+    {
+      "epoch": 0.4311788003756482,
+      "grad_norm": 0.6840534210205078,
+      "learning_rate": 6.216245052546251e-05,
+      "loss": 1.0422,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4315054509819934,
+      "grad_norm": 0.7854779958724976,
+      "learning_rate": 6.211184226412131e-05,
+      "loss": 1.2574,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43183210158833857,
+      "grad_norm": 0.9489834904670715,
+      "learning_rate": 6.206122081752913e-05,
+      "loss": 1.4515,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43215875219468375,
+      "grad_norm": 0.9874739050865173,
+      "learning_rate": 6.201058624079371e-05,
+      "loss": 1.2792,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43248540280102893,
+      "grad_norm": 1.4044889211654663,
+      "learning_rate": 6.195993858903713e-05,
+      "loss": 1.5486,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43281205340737416,
+      "grad_norm": 2.303447723388672,
+      "learning_rate": 6.190927791739565e-05,
+      "loss": 1.998,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43313870401371934,
+      "grad_norm": 0.1984492391347885,
+      "learning_rate": 6.185860428101974e-05,
+      "loss": 0.7388,
+      "step": 1326
+    },
+    {
+      "epoch": 0.4334653546200645,
+      "grad_norm": 0.24812500178813934,
+      "learning_rate": 6.180791773507396e-05,
+      "loss": 0.7755,
+      "step": 1327
+    },
+    {
+      "epoch": 0.4337920052264097,
+      "grad_norm": 0.2576265335083008,
+      "learning_rate": 6.175721833473697e-05,
+      "loss": 0.796,
+      "step": 1328
+    },
+    {
+      "epoch": 0.4341186558327549,
+      "grad_norm": 0.2764826714992523,
+      "learning_rate": 6.170650613520137e-05,
+      "loss": 0.8901,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43444530643910007,
+      "grad_norm": 0.28213411569595337,
+      "learning_rate": 6.16557811916737e-05,
+      "loss": 0.8744,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43477195704544525,
+      "grad_norm": 0.2741318345069885,
+      "learning_rate": 6.160504355937441e-05,
+      "loss": 0.7944,
+      "step": 1331
+    },
+    {
+      "epoch": 0.4350986076517905,
+      "grad_norm": 0.3129563331604004,
+      "learning_rate": 6.155429329353772e-05,
+      "loss": 0.9256,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43542525825813566,
+      "grad_norm": 0.2871616780757904,
+      "learning_rate": 6.150353044941166e-05,
+      "loss": 0.87,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43575190886448084,
+      "grad_norm": 0.2862168550491333,
+      "learning_rate": 6.145275508225789e-05,
+      "loss": 0.9194,
+      "step": 1334
+    },
+    {
+      "epoch": 0.436078559470826,
+      "grad_norm": 0.3074105381965637,
+      "learning_rate": 6.140196724735173e-05,
+      "loss": 0.8538,
+      "step": 1335
+    },
+    {
+      "epoch": 0.4364052100771712,
+      "grad_norm": 0.3276168704032898,
+      "learning_rate": 6.135116699998208e-05,
+      "loss": 0.903,
+      "step": 1336
+    },
+    {
+      "epoch": 0.4367318606835164,
+      "grad_norm": 0.3311191499233246,
+      "learning_rate": 6.130035439545137e-05,
+      "loss": 0.8529,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43705851128986156,
+      "grad_norm": 0.36968547105789185,
+      "learning_rate": 6.12495294890754e-05,
+      "loss": 1.0096,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43738516189620674,
+      "grad_norm": 0.3730202317237854,
+      "learning_rate": 6.119869233618347e-05,
+      "loss": 0.8173,
+      "step": 1339
+    },
+    {
+      "epoch": 0.437711812502552,
+      "grad_norm": 0.3859659433364868,
+      "learning_rate": 6.114784299211812e-05,
+      "loss": 0.9545,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43803846310889716,
+      "grad_norm": 0.4324837327003479,
+      "learning_rate": 6.109698151223524e-05,
+      "loss": 1.0696,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43836511371524234,
+      "grad_norm": 0.509543776512146,
+      "learning_rate": 6.10461079519039e-05,
+      "loss": 0.9784,
+      "step": 1342
+    },
+    {
+      "epoch": 0.4386917643215875,
+      "grad_norm": 0.5357346534729004,
+      "learning_rate": 6.099522236650628e-05,
+      "loss": 1.109,
+      "step": 1343
+    },
+    {
+      "epoch": 0.4390184149279327,
+      "grad_norm": 0.5872611403465271,
+      "learning_rate": 6.09443248114377e-05,
+      "loss": 1.1932,
+      "step": 1344
+    },
+    {
+      "epoch": 0.4393450655342779,
+      "grad_norm": 0.6826693415641785,
+      "learning_rate": 6.089341534210652e-05,
+      "loss": 1.2509,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43967171614062306,
+      "grad_norm": 0.7309541702270508,
+      "learning_rate": 6.084249401393403e-05,
+      "loss": 1.2348,
+      "step": 1346
+    },
+    {
+      "epoch": 0.4399983667469683,
+      "grad_norm": 0.897671639919281,
+      "learning_rate": 6.0791560882354424e-05,
+      "loss": 1.3749,
+      "step": 1347
+    },
+    {
+      "epoch": 0.4403250173533135,
+      "grad_norm": 0.9450259804725647,
+      "learning_rate": 6.07406160028148e-05,
+      "loss": 1.3895,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44065166795965865,
+      "grad_norm": 1.1689409017562866,
+      "learning_rate": 6.0689659430775e-05,
+      "loss": 1.5146,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44097831856600384,
+      "grad_norm": 1.6155471801757812,
+      "learning_rate": 6.063869122170761e-05,
+      "loss": 1.6751,
+      "step": 1350
+    },
+    {
+      "epoch": 0.441304969172349,
+      "grad_norm": 0.227553129196167,
+      "learning_rate": 6.058771143109789e-05,
+      "loss": 0.8437,
+      "step": 1351
+    },
+    {
+      "epoch": 0.4416316197786942,
+      "grad_norm": 0.25323745608329773,
+      "learning_rate": 6.053672011444369e-05,
+      "loss": 0.8681,
+      "step": 1352
+    },
+    {
+      "epoch": 0.4419582703850394,
+      "grad_norm": 0.26084786653518677,
+      "learning_rate": 6.048571732725543e-05,
+      "loss": 0.8891,
+      "step": 1353
+    },
+    {
+      "epoch": 0.4422849209913846,
+      "grad_norm": 0.2727871239185333,
+      "learning_rate": 6.043470312505599e-05,
+      "loss": 0.7828,
+      "step": 1354
+    },
+    {
+      "epoch": 0.4426115715977298,
+      "grad_norm": 0.2814069390296936,
+      "learning_rate": 6.038367756338072e-05,
+      "loss": 0.8744,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44293822220407497,
+      "grad_norm": 0.2910081446170807,
+      "learning_rate": 6.0332640697777273e-05,
+      "loss": 0.7748,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44326487281042015,
+      "grad_norm": 0.2924244701862335,
+      "learning_rate": 6.028159258380567e-05,
+      "loss": 0.769,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44359152341676533,
+      "grad_norm": 0.30017054080963135,
+      "learning_rate": 6.0230533277038127e-05,
+      "loss": 0.8374,
+      "step": 1358
+    },
+    {
+      "epoch": 0.4439181740231105,
+      "grad_norm": 0.31334736943244934,
+      "learning_rate": 6.01794628330591e-05,
+      "loss": 0.8224,
+      "step": 1359
+    },
+    {
+      "epoch": 0.4442448246294557,
+      "grad_norm": 0.3215758800506592,
+      "learning_rate": 6.01283813074651e-05,
+      "loss": 0.8468,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44457147523580093,
+      "grad_norm": 0.3455059230327606,
+      "learning_rate": 6.007728875586476e-05,
+      "loss": 0.9048,
+      "step": 1361
+    },
+    {
+      "epoch": 0.4448981258421461,
+      "grad_norm": 0.36075106263160706,
+      "learning_rate": 6.002618523387868e-05,
+      "loss": 0.8895,
+      "step": 1362
+    },
+    {
+      "epoch": 0.4452247764484913,
+      "grad_norm": 0.37401801347732544,
+      "learning_rate": 5.9975070797139446e-05,
+      "loss": 0.8716,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44555142705483647,
+      "grad_norm": 0.40770915150642395,
+      "learning_rate": 5.992394550129148e-05,
+      "loss": 0.9879,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44587807766118165,
+      "grad_norm": 0.4294146001338959,
+      "learning_rate": 5.9872809401991034e-05,
+      "loss": 1.0252,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44620472826752683,
+      "grad_norm": 0.4494366943836212,
+      "learning_rate": 5.9821662554906144e-05,
+      "loss": 1.0504,
+      "step": 1366
+    },
+    {
+      "epoch": 0.446531378873872,
+      "grad_norm": 0.49997806549072266,
+      "learning_rate": 5.977050501571653e-05,
+      "loss": 0.9862,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44685802948021724,
+      "grad_norm": 0.5091699957847595,
+      "learning_rate": 5.971933684011355e-05,
+      "loss": 0.9708,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4471846800865624,
+      "grad_norm": 0.6185327172279358,
+      "learning_rate": 5.966815808380015e-05,
+      "loss": 1.1143,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4475113306929076,
+      "grad_norm": 0.7364172339439392,
+      "learning_rate": 5.961696880249079e-05,
+      "loss": 1.3154,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4478379812992528,
+      "grad_norm": 0.8025093674659729,
+      "learning_rate": 5.9565769051911376e-05,
+      "loss": 1.1813,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44816463190559797,
+      "grad_norm": 0.9280282258987427,
+      "learning_rate": 5.951455888779925e-05,
+      "loss": 1.0802,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44849128251194315,
+      "grad_norm": 1.183429479598999,
+      "learning_rate": 5.9463338365903035e-05,
+      "loss": 1.1206,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4488179331182883,
+      "grad_norm": 1.2044509649276733,
+      "learning_rate": 5.941210754198266e-05,
+      "loss": 1.4664,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44914458372463356,
+      "grad_norm": 1.775682806968689,
+      "learning_rate": 5.936086647180928e-05,
+      "loss": 2.0872,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44947123433097874,
+      "grad_norm": 0.19256451725959778,
+      "learning_rate": 5.9309615211165185e-05,
+      "loss": 0.7676,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4497978849373239,
+      "grad_norm": 0.22578822076320648,
+      "learning_rate": 5.925835381584377e-05,
+      "loss": 0.797,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4501245355436691,
+      "grad_norm": 0.24020101130008698,
+      "learning_rate": 5.9207082341649454e-05,
+      "loss": 0.7621,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4504511861500143,
+      "grad_norm": 0.262273371219635,
+      "learning_rate": 5.9155800844397625e-05,
+      "loss": 0.785,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45077783675635946,
+      "grad_norm": 0.2687849700450897,
+      "learning_rate": 5.9104509379914586e-05,
+      "loss": 0.847,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45110448736270464,
+      "grad_norm": 0.2742252051830292,
+      "learning_rate": 5.905320800403752e-05,
+      "loss": 0.798,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4514311379690499,
+      "grad_norm": 0.28020089864730835,
+      "learning_rate": 5.900189677261434e-05,
+      "loss": 0.7844,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45175778857539506,
+      "grad_norm": 0.3002532422542572,
+      "learning_rate": 5.8950575741503744e-05,
+      "loss": 0.9231,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45208443918174024,
+      "grad_norm": 0.2936133146286011,
+      "learning_rate": 5.889924496657506e-05,
+      "loss": 0.8288,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4524110897880854,
+      "grad_norm": 0.3212607502937317,
+      "learning_rate": 5.884790450370825e-05,
+      "loss": 0.8293,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4527377403944306,
+      "grad_norm": 0.3269292414188385,
+      "learning_rate": 5.87965544087938e-05,
+      "loss": 0.8997,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4530643910007758,
+      "grad_norm": 0.35607802867889404,
+      "learning_rate": 5.874519473773271e-05,
+      "loss": 0.9919,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45339104160712096,
+      "grad_norm": 0.3684079647064209,
+      "learning_rate": 5.869382554643639e-05,
+      "loss": 0.9916,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4537176922134662,
+      "grad_norm": 0.3907953202724457,
+      "learning_rate": 5.864244689082659e-05,
+      "loss": 1.0021,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4540443428198114,
+      "grad_norm": 0.3965863883495331,
+      "learning_rate": 5.85910588268354e-05,
+      "loss": 0.8898,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45437099342615656,
+      "grad_norm": 0.4447333514690399,
+      "learning_rate": 5.853966141040512e-05,
+      "loss": 0.9954,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45469764403250174,
+      "grad_norm": 0.4704182744026184,
+      "learning_rate": 5.848825469748828e-05,
+      "loss": 1.0979,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4550242946388469,
+      "grad_norm": 0.47324079275131226,
+      "learning_rate": 5.843683874404746e-05,
+      "loss": 0.9627,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4553509452451921,
+      "grad_norm": 0.618899941444397,
+      "learning_rate": 5.838541360605538e-05,
+      "loss": 1.104,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4556775958515373,
+      "grad_norm": 0.6069836616516113,
+      "learning_rate": 5.833397933949469e-05,
+      "loss": 1.2049,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4560042464578825,
+      "grad_norm": 0.7420676946640015,
+      "learning_rate": 5.8282536000358e-05,
+      "loss": 1.2061,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4563308970642277,
+      "grad_norm": 1.0195621252059937,
+      "learning_rate": 5.823108364464782e-05,
+      "loss": 1.4885,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45665754767057287,
+      "grad_norm": 1.0798938274383545,
+      "learning_rate": 5.817962232837645e-05,
+      "loss": 1.4734,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45698419827691805,
+      "grad_norm": 1.271763563156128,
+      "learning_rate": 5.8128152107565946e-05,
+      "loss": 1.8257,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45731084888326323,
+      "grad_norm": 1.607129693031311,
+      "learning_rate": 5.807667303824806e-05,
+      "loss": 1.8979,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4576374994896084,
+      "grad_norm": 0.2017160803079605,
+      "learning_rate": 5.8025185176464204e-05,
+      "loss": 0.6849,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4579641500959536,
+      "grad_norm": 0.21873755753040314,
+      "learning_rate": 5.7973688578265304e-05,
+      "loss": 0.7677,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45829080070229883,
+      "grad_norm": 0.25168201327323914,
+      "learning_rate": 5.792218329971184e-05,
+      "loss": 0.8155,
+      "step": 1403
+    },
+    {
+      "epoch": 0.458617451308644,
+      "grad_norm": 0.2593826651573181,
+      "learning_rate": 5.7870669396873754e-05,
+      "loss": 0.8939,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4589441019149892,
+      "grad_norm": 0.24560409784317017,
+      "learning_rate": 5.7819146925830324e-05,
+      "loss": 0.7275,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45927075252133437,
+      "grad_norm": 0.273952841758728,
+      "learning_rate": 5.7767615942670204e-05,
+      "loss": 0.8585,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45959740312767955,
+      "grad_norm": 0.2819421887397766,
+      "learning_rate": 5.7716076503491314e-05,
+      "loss": 0.8319,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45992405373402473,
+      "grad_norm": 0.3027994632720947,
+      "learning_rate": 5.766452866440072e-05,
+      "loss": 0.8609,
+      "step": 1408
+    },
+    {
+      "epoch": 0.4602507043403699,
+      "grad_norm": 0.3026481866836548,
+      "learning_rate": 5.761297248151469e-05,
+      "loss": 0.8992,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46057735494671515,
+      "grad_norm": 0.3143552839756012,
+      "learning_rate": 5.756140801095858e-05,
+      "loss": 0.9488,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4609040055530603,
+      "grad_norm": 0.3467939496040344,
+      "learning_rate": 5.750983530886672e-05,
+      "loss": 1.0314,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4612306561594055,
+      "grad_norm": 0.3449691832065582,
+      "learning_rate": 5.745825443138246e-05,
+      "loss": 0.9967,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4615573067657507,
+      "grad_norm": 0.35669663548469543,
+      "learning_rate": 5.740666543465798e-05,
+      "loss": 1.001,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46188395737209587,
+      "grad_norm": 0.40556496381759644,
+      "learning_rate": 5.735506837485437e-05,
+      "loss": 0.9929,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46221060797844105,
+      "grad_norm": 0.3812430500984192,
+      "learning_rate": 5.730346330814145e-05,
+      "loss": 1.0193,
+      "step": 1415
+    },
+    {
+      "epoch": 0.4625372585847862,
+      "grad_norm": 0.44601941108703613,
+      "learning_rate": 5.7251850290697774e-05,
+      "loss": 0.9902,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46286390919113146,
+      "grad_norm": 0.42666640877723694,
+      "learning_rate": 5.7200229378710546e-05,
+      "loss": 1.0404,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46319055979747664,
+      "grad_norm": 0.4538075625896454,
+      "learning_rate": 5.714860062837557e-05,
+      "loss": 1.0535,
+      "step": 1418
+    },
+    {
+      "epoch": 0.4635172104038218,
+      "grad_norm": 0.5554001331329346,
+      "learning_rate": 5.7096964095897174e-05,
+      "loss": 1.1326,
+      "step": 1419
+    },
+    {
+      "epoch": 0.463843861010167,
+      "grad_norm": 0.5905495882034302,
+      "learning_rate": 5.7045319837488186e-05,
+      "loss": 1.2233,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4641705116165122,
+      "grad_norm": 0.6906095147132874,
+      "learning_rate": 5.6993667909369794e-05,
+      "loss": 1.292,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46449716222285736,
+      "grad_norm": 0.793144941329956,
+      "learning_rate": 5.694200836777158e-05,
+      "loss": 1.2007,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46482381282920254,
+      "grad_norm": 1.0440678596496582,
+      "learning_rate": 5.68903412689314e-05,
+      "loss": 1.3643,
+      "step": 1423
+    },
+    {
+      "epoch": 0.4651504634355478,
+      "grad_norm": 1.2731975317001343,
+      "learning_rate": 5.683866666909533e-05,
+      "loss": 1.6552,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "grad_norm": 1.792528510093689,
+      "learning_rate": 5.6786984624517636e-05,
+      "loss": 1.6505,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46580376464823814,
+      "grad_norm": 0.20550623536109924,
+      "learning_rate": 5.6735295191460636e-05,
+      "loss": 0.7223,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4661304152545833,
+      "grad_norm": 0.24376057088375092,
+      "learning_rate": 5.668359842619474e-05,
+      "loss": 0.8327,
+      "step": 1427
+    },
+    {
+      "epoch": 0.4664570658609285,
+      "grad_norm": 0.23944585025310516,
+      "learning_rate": 5.663189438499833e-05,
+      "loss": 0.8034,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4667837164672737,
+      "grad_norm": 0.2565159499645233,
+      "learning_rate": 5.6580183124157714e-05,
+      "loss": 0.8629,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46711036707361886,
+      "grad_norm": 0.2701554298400879,
+      "learning_rate": 5.652846469996702e-05,
+      "loss": 0.8289,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4674370176799641,
+      "grad_norm": 0.30046141147613525,
+      "learning_rate": 5.647673916872822e-05,
+      "loss": 0.8658,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4677636682863093,
+      "grad_norm": 0.2801269590854645,
+      "learning_rate": 5.6425006586751004e-05,
+      "loss": 0.9112,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46809031889265446,
+      "grad_norm": 0.3160743713378906,
+      "learning_rate": 5.6373267010352736e-05,
+      "loss": 0.8969,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46841696949899964,
+      "grad_norm": 0.3159767687320709,
+      "learning_rate": 5.632152049585843e-05,
+      "loss": 0.9211,
+      "step": 1434
+    },
+    {
+      "epoch": 0.4687436201053448,
+      "grad_norm": 0.31142839789390564,
+      "learning_rate": 5.626976709960057e-05,
+      "loss": 0.8939,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46907027071169,
+      "grad_norm": 0.31656891107559204,
+      "learning_rate": 5.621800687791922e-05,
+      "loss": 0.8701,
+      "step": 1436
+    },
+    {
+      "epoch": 0.4693969213180352,
+      "grad_norm": 0.3449065387248993,
+      "learning_rate": 5.616623988716181e-05,
+      "loss": 0.9217,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4697235719243804,
+      "grad_norm": 0.39621448516845703,
+      "learning_rate": 5.611446618368319e-05,
+      "loss": 0.9877,
+      "step": 1438
+    },
+    {
+      "epoch": 0.4700502225307256,
+      "grad_norm": 0.39008018374443054,
+      "learning_rate": 5.606268582384548e-05,
+      "loss": 0.889,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4703768731370708,
+      "grad_norm": 0.443905234336853,
+      "learning_rate": 5.601089886401808e-05,
+      "loss": 0.9621,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47070352374341595,
+      "grad_norm": 0.46601757407188416,
+      "learning_rate": 5.595910536057753e-05,
+      "loss": 0.9748,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47103017434976113,
+      "grad_norm": 0.4815382957458496,
+      "learning_rate": 5.5907305369907534e-05,
+      "loss": 0.9838,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4713568249561063,
+      "grad_norm": 0.5498834252357483,
+      "learning_rate": 5.5855498948398846e-05,
+      "loss": 1.069,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4716834755624515,
+      "grad_norm": 0.5492748022079468,
+      "learning_rate": 5.5803686152449184e-05,
+      "loss": 0.996,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4720101261687967,
+      "grad_norm": 0.6215171217918396,
+      "learning_rate": 5.575186703846328e-05,
+      "loss": 1.0071,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4723367767751419,
+      "grad_norm": 0.8084948658943176,
+      "learning_rate": 5.5700041662852684e-05,
+      "loss": 1.2216,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4726634273814871,
+      "grad_norm": 0.9253719449043274,
+      "learning_rate": 5.56482100820358e-05,
+      "loss": 1.1535,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47299007798783227,
+      "grad_norm": 1.2240797281265259,
+      "learning_rate": 5.559637235243773e-05,
+      "loss": 1.5206,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47331672859417745,
+      "grad_norm": 1.3156620264053345,
+      "learning_rate": 5.5544528530490345e-05,
+      "loss": 1.3771,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47364337920052263,
+      "grad_norm": 1.7129418849945068,
+      "learning_rate": 5.5492678672632094e-05,
+      "loss": 2.4055,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4739700298068678,
+      "grad_norm": 0.17648789286613464,
+      "learning_rate": 5.5440822835308026e-05,
+      "loss": 0.6887,
+      "step": 1451
+    },
+    {
+      "epoch": 0.474296680413213,
+      "grad_norm": 0.22524596750736237,
+      "learning_rate": 5.538896107496967e-05,
+      "loss": 0.7308,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4746233310195582,
+      "grad_norm": 0.23709015548229218,
+      "learning_rate": 5.5337093448075025e-05,
+      "loss": 0.8185,
+      "step": 1453
+    },
+    {
+      "epoch": 0.4749499816259034,
+      "grad_norm": 0.24522243440151215,
+      "learning_rate": 5.528522001108849e-05,
+      "loss": 0.7561,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4752766322322486,
+      "grad_norm": 0.2625182569026947,
+      "learning_rate": 5.523334082048075e-05,
+      "loss": 0.8076,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47560328283859377,
+      "grad_norm": 0.27700942754745483,
+      "learning_rate": 5.5181455932728785e-05,
+      "loss": 0.8427,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47592993344493895,
+      "grad_norm": 0.2912745475769043,
+      "learning_rate": 5.512956540431577e-05,
+      "loss": 0.8505,
+      "step": 1457
+    },
+    {
+      "epoch": 0.4762565840512841,
+      "grad_norm": 0.2912481129169464,
+      "learning_rate": 5.5077669291731006e-05,
+      "loss": 0.9061,
+      "step": 1458
+    },
+    {
+      "epoch": 0.4765832346576293,
+      "grad_norm": 0.36340487003326416,
+      "learning_rate": 5.502576765146989e-05,
+      "loss": 0.9754,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47690988526397454,
+      "grad_norm": 0.3103990852832794,
+      "learning_rate": 5.497386054003385e-05,
+      "loss": 0.9334,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4772365358703197,
+      "grad_norm": 0.33423370122909546,
+      "learning_rate": 5.492194801393023e-05,
+      "loss": 0.8043,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4775631864766649,
+      "grad_norm": 0.3521210253238678,
+      "learning_rate": 5.487003012967228e-05,
+      "loss": 0.9447,
+      "step": 1462
+    },
+    {
+      "epoch": 0.4778898370830101,
+      "grad_norm": 0.342441201210022,
+      "learning_rate": 5.4818106943779105e-05,
+      "loss": 0.9133,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47821648768935526,
+      "grad_norm": 0.39511221647262573,
+      "learning_rate": 5.476617851277559e-05,
+      "loss": 1.0284,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47854313829570044,
+      "grad_norm": 0.39670830965042114,
+      "learning_rate": 5.471424489319227e-05,
+      "loss": 0.9131,
+      "step": 1465
+    },
+    {
+      "epoch": 0.4788697889020456,
+      "grad_norm": 0.4468030333518982,
+      "learning_rate": 5.466230614156539e-05,
+      "loss": 0.93,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47919643950839086,
+      "grad_norm": 0.47740623354911804,
+      "learning_rate": 5.461036231443676e-05,
+      "loss": 1.0926,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47952309011473604,
+      "grad_norm": 0.5126772522926331,
+      "learning_rate": 5.455841346835371e-05,
+      "loss": 0.9977,
+      "step": 1468
+    },
+    {
+      "epoch": 0.4798497407210812,
+      "grad_norm": 0.5842729210853577,
+      "learning_rate": 5.4506459659869036e-05,
+      "loss": 1.0636,
+      "step": 1469
+    },
+    {
+      "epoch": 0.4801763913274264,
+      "grad_norm": 0.633167564868927,
+      "learning_rate": 5.445450094554094e-05,
+      "loss": 1.2103,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4805030419337716,
+      "grad_norm": 0.7382978200912476,
+      "learning_rate": 5.440253738193297e-05,
+      "loss": 1.3609,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48082969254011676,
+      "grad_norm": 1.0082590579986572,
+      "learning_rate": 5.435056902561393e-05,
+      "loss": 1.5858,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48115634314646194,
+      "grad_norm": 1.1187267303466797,
+      "learning_rate": 5.4298595933157884e-05,
+      "loss": 1.5626,
+      "step": 1473
+    },
+    {
+      "epoch": 0.4814829937528072,
+      "grad_norm": 1.2796682119369507,
+      "learning_rate": 5.4246618161144006e-05,
+      "loss": 1.774,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48180964435915236,
+      "grad_norm": 1.722538709640503,
+      "learning_rate": 5.4194635766156575e-05,
+      "loss": 2.2056,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48213629496549754,
+      "grad_norm": 0.22073139250278473,
+      "learning_rate": 5.414264880478493e-05,
+      "loss": 0.8193,
+      "step": 1476
+    },
+    {
+      "epoch": 0.4824629455718427,
+      "grad_norm": 0.23783589899539948,
+      "learning_rate": 5.409065733362337e-05,
+      "loss": 0.8528,
+      "step": 1477
+    },
+    {
+      "epoch": 0.4827895961781879,
+      "grad_norm": 0.2497948557138443,
+      "learning_rate": 5.403866140927109e-05,
+      "loss": 0.7723,
+      "step": 1478
+    },
+    {
+      "epoch": 0.4831162467845331,
+      "grad_norm": 0.25545740127563477,
+      "learning_rate": 5.3986661088332115e-05,
+      "loss": 0.8591,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48344289739087826,
+      "grad_norm": 0.2744065821170807,
+      "learning_rate": 5.3934656427415295e-05,
+      "loss": 0.8505,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4837695479972235,
+      "grad_norm": 0.2947370111942291,
+      "learning_rate": 5.3882647483134196e-05,
+      "loss": 0.8756,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4840961986035687,
+      "grad_norm": 0.3097544312477112,
+      "learning_rate": 5.3830634312107056e-05,
+      "loss": 0.8883,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48442284920991385,
+      "grad_norm": 0.31207266449928284,
+      "learning_rate": 5.3778616970956663e-05,
+      "loss": 0.8534,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48474949981625903,
+      "grad_norm": 0.38118842244148254,
+      "learning_rate": 5.3726595516310405e-05,
+      "loss": 0.9782,
+      "step": 1484
+    },
+    {
+      "epoch": 0.4850761504226042,
+      "grad_norm": 0.315085232257843,
+      "learning_rate": 5.367457000480011e-05,
+      "loss": 0.8068,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4854028010289494,
+      "grad_norm": 0.34587183594703674,
+      "learning_rate": 5.3622540493062046e-05,
+      "loss": 0.8694,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4857294516352946,
+      "grad_norm": 0.3551337718963623,
+      "learning_rate": 5.3570507037736826e-05,
+      "loss": 0.9294,
+      "step": 1487
+    },
+    {
+      "epoch": 0.4860561022416398,
+      "grad_norm": 0.3878238797187805,
+      "learning_rate": 5.351846969546935e-05,
+      "loss": 0.8701,
+      "step": 1488
+    },
+    {
+      "epoch": 0.486382752847985,
+      "grad_norm": 0.4408702254295349,
+      "learning_rate": 5.346642852290876e-05,
+      "loss": 1.2004,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48670940345433017,
+      "grad_norm": 0.5383159518241882,
+      "learning_rate": 5.341438357670838e-05,
+      "loss": 1.0944,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48703605406067535,
+      "grad_norm": 0.48991337418556213,
+      "learning_rate": 5.336233491352559e-05,
+      "loss": 1.0647,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48736270466702053,
+      "grad_norm": 0.5360439419746399,
+      "learning_rate": 5.3310282590021875e-05,
+      "loss": 1.0419,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4876893552733657,
+      "grad_norm": 0.5581495761871338,
+      "learning_rate": 5.325822666286268e-05,
+      "loss": 1.2277,
+      "step": 1493
+    },
+    {
+      "epoch": 0.4880160058797109,
+      "grad_norm": 0.6016497015953064,
+      "learning_rate": 5.320616718871736e-05,
+      "loss": 1.0748,
+      "step": 1494
+    },
+    {
+      "epoch": 0.4883426564860561,
+      "grad_norm": 0.6058472990989685,
+      "learning_rate": 5.315410422425917e-05,
+      "loss": 1.1947,
+      "step": 1495
+    },
+    {
+      "epoch": 0.4886693070924013,
+      "grad_norm": 0.7601789236068726,
+      "learning_rate": 5.310203782616513e-05,
+      "loss": 1.4049,
+      "step": 1496
+    },
+    {
+      "epoch": 0.4889959576987465,
+      "grad_norm": 0.9896534085273743,
+      "learning_rate": 5.304996805111599e-05,
+      "loss": 1.3961,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48932260830509167,
+      "grad_norm": 1.224902629852295,
+      "learning_rate": 5.299789495579621e-05,
+      "loss": 1.6296,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48964925891143685,
+      "grad_norm": 1.2904133796691895,
+      "learning_rate": 5.294581859689387e-05,
+      "loss": 1.5572,
+      "step": 1499
+    },
+    {
+      "epoch": 0.489975909517782,
+      "grad_norm": 1.9275236129760742,
+      "learning_rate": 5.2893739031100554e-05,
+      "loss": 2.2083,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4903025601241272,
+      "grad_norm": 0.18931369483470917,
+      "learning_rate": 5.2841656315111366e-05,
+      "loss": 0.69,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49062921073047244,
+      "grad_norm": 0.2273533195257187,
+      "learning_rate": 5.278957050562485e-05,
+      "loss": 0.8017,
+      "step": 1502
+    },
+    {
+      "epoch": 0.4909558613368176,
+      "grad_norm": 0.24772115051746368,
+      "learning_rate": 5.273748165934289e-05,
+      "loss": 0.8714,
+      "step": 1503
+    },
+    {
+      "epoch": 0.4912825119431628,
+      "grad_norm": 0.2568557858467102,
+      "learning_rate": 5.268538983297072e-05,
+      "loss": 0.827,
+      "step": 1504
+    },
+    {
+      "epoch": 0.491609162549508,
+      "grad_norm": 0.25856682658195496,
+      "learning_rate": 5.263329508321676e-05,
+      "loss": 0.8657,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49193581315585316,
+      "grad_norm": 0.2738785445690155,
+      "learning_rate": 5.258119746679266e-05,
+      "loss": 0.8616,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49226246376219834,
+      "grad_norm": 0.28627508878707886,
+      "learning_rate": 5.252909704041318e-05,
+      "loss": 0.7742,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4925891143685435,
+      "grad_norm": 0.3042372763156891,
+      "learning_rate": 5.247699386079613e-05,
+      "loss": 0.874,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49291576497488876,
+      "grad_norm": 0.328125536441803,
+      "learning_rate": 5.2424887984662294e-05,
+      "loss": 0.8562,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49324241558123394,
+      "grad_norm": 0.3250144124031067,
+      "learning_rate": 5.2372779468735446e-05,
+      "loss": 0.9107,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4935690661875791,
+      "grad_norm": 0.35796797275543213,
+      "learning_rate": 5.232066836974219e-05,
+      "loss": 0.98,
+      "step": 1511
+    },
+    {
+      "epoch": 0.4938957167939243,
+      "grad_norm": 0.3610784113407135,
+      "learning_rate": 5.226855474441197e-05,
+      "loss": 0.9441,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4942223674002695,
+      "grad_norm": 0.36936962604522705,
+      "learning_rate": 5.2216438649476954e-05,
+      "loss": 0.9559,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49454901800661466,
+      "grad_norm": 0.3782363533973694,
+      "learning_rate": 5.2164320141672006e-05,
+      "loss": 0.8496,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49487566861295984,
+      "grad_norm": 0.37794479727745056,
+      "learning_rate": 5.211219927773464e-05,
+      "loss": 0.8517,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4952023192193051,
+      "grad_norm": 0.48705726861953735,
+      "learning_rate": 5.206007611440491e-05,
+      "loss": 1.0162,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49552896982565026,
+      "grad_norm": 0.4541557729244232,
+      "learning_rate": 5.200795070842539e-05,
+      "loss": 1.0469,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49585562043199544,
+      "grad_norm": 0.4859015643596649,
+      "learning_rate": 5.195582311654107e-05,
+      "loss": 0.9468,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4961822710383406,
+      "grad_norm": 0.6146647334098816,
+      "learning_rate": 5.190369339549933e-05,
+      "loss": 1.0863,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4965089216446858,
+      "grad_norm": 0.6606687307357788,
+      "learning_rate": 5.18515616020499e-05,
+      "loss": 1.3782,
+      "step": 1520
+    },
+    {
+      "epoch": 0.496835572251031,
+      "grad_norm": 0.7691556215286255,
+      "learning_rate": 5.179942779294472e-05,
+      "loss": 1.1182,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49716222285737616,
+      "grad_norm": 0.9613664746284485,
+      "learning_rate": 5.174729202493794e-05,
+      "loss": 1.3084,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4974888734637214,
+      "grad_norm": 1.210338830947876,
+      "learning_rate": 5.169515435478587e-05,
+      "loss": 1.4691,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4978155240700666,
+      "grad_norm": 1.183440923690796,
+      "learning_rate": 5.164301483924685e-05,
+      "loss": 1.4419,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49814217467641175,
+      "grad_norm": 1.764573574066162,
+      "learning_rate": 5.159087353508125e-05,
+      "loss": 2.0149,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49846882528275693,
+      "grad_norm": 0.2246791124343872,
+      "learning_rate": 5.153873049905138e-05,
+      "loss": 0.7747,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4987954758891021,
+      "grad_norm": 0.24696335196495056,
+      "learning_rate": 5.1486585787921427e-05,
+      "loss": 0.7843,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4991221264954473,
+      "grad_norm": 0.24796541035175323,
+      "learning_rate": 5.1434439458457426e-05,
+      "loss": 0.8426,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4994487771017925,
+      "grad_norm": 0.2630245089530945,
+      "learning_rate": 5.1382291567427175e-05,
+      "loss": 0.9231,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4997754277081377,
+      "grad_norm": 0.2528044879436493,
+      "learning_rate": 5.133014217160014e-05,
+      "loss": 0.7866,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5001020783144828,
+      "grad_norm": 0.2634943425655365,
+      "learning_rate": 5.127799132774744e-05,
+      "loss": 0.7849,
+      "step": 1531
+    },
+    {
+      "epoch": 0.500428728920828,
+      "grad_norm": 0.27956002950668335,
+      "learning_rate": 5.122583909264178e-05,
+      "loss": 0.8601,
+      "step": 1532
+    },
+    {
+      "epoch": 0.500428728920828,
+      "eval_loss": 1.0430185794830322,
+      "eval_runtime": 500.0584,
+      "eval_samples_per_second": 5.155,
+      "eval_steps_per_second": 2.578,
+      "step": 1532
     }
   ],
   "logging_steps": 1,
@@ -5404,7 +10774,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.04481580677333e+17,
+  "total_flos": 1.0089220027501773e+18,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null