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": 0.7508519500189322,
+  "epoch": 1.0,
   "eval_steps": 661,
-  "global_step": 1983,
+  "global_step": 2641,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -13920,6 +13920,4612 @@
       "eval_samples_per_second": 4.958,
       "eval_steps_per_second": 1.24,
       "step": 1983
+    },
+    {
+      "epoch": 0.751230594471791,
+      "grad_norm": 12.873611450195312,
+      "learning_rate": 2.943779944470404e-05,
+      "loss": 1.5529,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7516092389246497,
+      "grad_norm": 11.92525577545166,
+      "learning_rate": 2.9352917064632112e-05,
+      "loss": 0.7601,
+      "step": 1985
+    },
+    {
+      "epoch": 0.7519878833775085,
+      "grad_norm": 12.782543182373047,
+      "learning_rate": 2.926813618312134e-05,
+      "loss": 1.086,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7523665278303673,
+      "grad_norm": 10.654260635375977,
+      "learning_rate": 2.9183456921976304e-05,
+      "loss": 1.0646,
+      "step": 1987
+    },
+    {
+      "epoch": 0.752745172283226,
+      "grad_norm": 11.402060508728027,
+      "learning_rate": 2.909887940285554e-05,
+      "loss": 0.82,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7531238167360849,
+      "grad_norm": 15.131863594055176,
+      "learning_rate": 2.901440374727149e-05,
+      "loss": 1.0551,
+      "step": 1989
+    },
+    {
+      "epoch": 0.7535024611889436,
+      "grad_norm": 12.485123634338379,
+      "learning_rate": 2.8930030076590198e-05,
+      "loss": 0.7528,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7538811056418023,
+      "grad_norm": 7.999950885772705,
+      "learning_rate": 2.8845758512031186e-05,
+      "loss": 0.4973,
+      "step": 1991
+    },
+    {
+      "epoch": 0.7542597500946611,
+      "grad_norm": 13.8598051071167,
+      "learning_rate": 2.876158917466726e-05,
+      "loss": 0.8543,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7546383945475199,
+      "grad_norm": 13.514735221862793,
+      "learning_rate": 2.867752218542443e-05,
+      "loss": 0.9004,
+      "step": 1993
+    },
+    {
+      "epoch": 0.7550170390003786,
+      "grad_norm": 13.104480743408203,
+      "learning_rate": 2.8593557665081616e-05,
+      "loss": 0.8655,
+      "step": 1994
+    },
+    {
+      "epoch": 0.7553956834532374,
+      "grad_norm": 13.811482429504395,
+      "learning_rate": 2.8509695734270492e-05,
+      "loss": 0.5821,
+      "step": 1995
+    },
+    {
+      "epoch": 0.7557743279060962,
+      "grad_norm": 13.989906311035156,
+      "learning_rate": 2.8425936513475395e-05,
+      "loss": 0.6422,
+      "step": 1996
+    },
+    {
+      "epoch": 0.756152972358955,
+      "grad_norm": 10.727309226989746,
+      "learning_rate": 2.834228012303306e-05,
+      "loss": 0.4677,
+      "step": 1997
+    },
+    {
+      "epoch": 0.7565316168118137,
+      "grad_norm": 15.157230377197266,
+      "learning_rate": 2.8258726683132474e-05,
+      "loss": 0.8738,
+      "step": 1998
+    },
+    {
+      "epoch": 0.7569102612646724,
+      "grad_norm": 17.531665802001953,
+      "learning_rate": 2.8175276313814813e-05,
+      "loss": 0.7335,
+      "step": 1999
+    },
+    {
+      "epoch": 0.7572889057175313,
+      "grad_norm": 10.896133422851562,
+      "learning_rate": 2.809192913497306e-05,
+      "loss": 0.5936,
+      "step": 2000
+    },
+    {
+      "epoch": 0.75766755017039,
+      "grad_norm": 10.433149337768555,
+      "learning_rate": 2.8008685266351988e-05,
+      "loss": 2.4051,
+      "step": 2001
+    },
+    {
+      "epoch": 0.7580461946232487,
+      "grad_norm": 9.504281044006348,
+      "learning_rate": 2.7925544827547933e-05,
+      "loss": 1.7767,
+      "step": 2002
+    },
+    {
+      "epoch": 0.7584248390761076,
+      "grad_norm": 10.231252670288086,
+      "learning_rate": 2.7842507938008666e-05,
+      "loss": 1.8186,
+      "step": 2003
+    },
+    {
+      "epoch": 0.7588034835289663,
+      "grad_norm": 9.87753677368164,
+      "learning_rate": 2.775957471703311e-05,
+      "loss": 1.2146,
+      "step": 2004
+    },
+    {
+      "epoch": 0.759182127981825,
+      "grad_norm": 11.625948905944824,
+      "learning_rate": 2.7676745283771388e-05,
+      "loss": 1.5717,
+      "step": 2005
+    },
+    {
+      "epoch": 0.7595607724346838,
+      "grad_norm": 11.169037818908691,
+      "learning_rate": 2.7594019757224364e-05,
+      "loss": 1.3143,
+      "step": 2006
+    },
+    {
+      "epoch": 0.7599394168875426,
+      "grad_norm": 9.829212188720703,
+      "learning_rate": 2.7511398256243716e-05,
+      "loss": 0.9765,
+      "step": 2007
+    },
+    {
+      "epoch": 0.7603180613404014,
+      "grad_norm": 10.827011108398438,
+      "learning_rate": 2.7428880899531585e-05,
+      "loss": 0.7852,
+      "step": 2008
+    },
+    {
+      "epoch": 0.7606967057932601,
+      "grad_norm": 12.02198314666748,
+      "learning_rate": 2.7346467805640585e-05,
+      "loss": 1.0464,
+      "step": 2009
+    },
+    {
+      "epoch": 0.7610753502461189,
+      "grad_norm": 12.59335708618164,
+      "learning_rate": 2.7264159092973484e-05,
+      "loss": 0.8952,
+      "step": 2010
+    },
+    {
+      "epoch": 0.7614539946989777,
+      "grad_norm": 13.947036743164062,
+      "learning_rate": 2.718195487978308e-05,
+      "loss": 1.4524,
+      "step": 2011
+    },
+    {
+      "epoch": 0.7618326391518364,
+      "grad_norm": 12.672119140625,
+      "learning_rate": 2.7099855284172017e-05,
+      "loss": 1.3051,
+      "step": 2012
+    },
+    {
+      "epoch": 0.7622112836046951,
+      "grad_norm": 10.371209144592285,
+      "learning_rate": 2.7017860424092712e-05,
+      "loss": 0.7874,
+      "step": 2013
+    },
+    {
+      "epoch": 0.762589928057554,
+      "grad_norm": 12.310194969177246,
+      "learning_rate": 2.6935970417347057e-05,
+      "loss": 0.9649,
+      "step": 2014
+    },
+    {
+      "epoch": 0.7629685725104127,
+      "grad_norm": 10.61960506439209,
+      "learning_rate": 2.6854185381586273e-05,
+      "loss": 0.8684,
+      "step": 2015
+    },
+    {
+      "epoch": 0.7633472169632715,
+      "grad_norm": 14.664178848266602,
+      "learning_rate": 2.6772505434310803e-05,
+      "loss": 1.0722,
+      "step": 2016
+    },
+    {
+      "epoch": 0.7637258614161303,
+      "grad_norm": 8.189818382263184,
+      "learning_rate": 2.6690930692870143e-05,
+      "loss": 0.4872,
+      "step": 2017
+    },
+    {
+      "epoch": 0.764104505868989,
+      "grad_norm": 11.075437545776367,
+      "learning_rate": 2.6609461274462588e-05,
+      "loss": 0.7672,
+      "step": 2018
+    },
+    {
+      "epoch": 0.7644831503218478,
+      "grad_norm": 10.562186241149902,
+      "learning_rate": 2.6528097296135135e-05,
+      "loss": 0.727,
+      "step": 2019
+    },
+    {
+      "epoch": 0.7648617947747065,
+      "grad_norm": 13.31433391571045,
+      "learning_rate": 2.6446838874783254e-05,
+      "loss": 0.7146,
+      "step": 2020
+    },
+    {
+      "epoch": 0.7652404392275653,
+      "grad_norm": 10.966352462768555,
+      "learning_rate": 2.636568612715087e-05,
+      "loss": 0.3659,
+      "step": 2021
+    },
+    {
+      "epoch": 0.7656190836804241,
+      "grad_norm": 23.072904586791992,
+      "learning_rate": 2.6284639169829973e-05,
+      "loss": 0.6862,
+      "step": 2022
+    },
+    {
+      "epoch": 0.7659977281332828,
+      "grad_norm": 10.244979858398438,
+      "learning_rate": 2.6203698119260632e-05,
+      "loss": 0.3511,
+      "step": 2023
+    },
+    {
+      "epoch": 0.7663763725861417,
+      "grad_norm": 11.128731727600098,
+      "learning_rate": 2.6122863091730686e-05,
+      "loss": 0.329,
+      "step": 2024
+    },
+    {
+      "epoch": 0.7667550170390004,
+      "grad_norm": 34.793052673339844,
+      "learning_rate": 2.6042134203375767e-05,
+      "loss": 1.7021,
+      "step": 2025
+    },
+    {
+      "epoch": 0.7671336614918591,
+      "grad_norm": 10.744704246520996,
+      "learning_rate": 2.596151157017892e-05,
+      "loss": 2.497,
+      "step": 2026
+    },
+    {
+      "epoch": 0.7675123059447179,
+      "grad_norm": 11.219593048095703,
+      "learning_rate": 2.588099530797058e-05,
+      "loss": 2.192,
+      "step": 2027
+    },
+    {
+      "epoch": 0.7678909503975767,
+      "grad_norm": 10.790613174438477,
+      "learning_rate": 2.580058553242829e-05,
+      "loss": 1.1766,
+      "step": 2028
+    },
+    {
+      "epoch": 0.7682695948504354,
+      "grad_norm": 9.38663101196289,
+      "learning_rate": 2.572028235907673e-05,
+      "loss": 1.0091,
+      "step": 2029
+    },
+    {
+      "epoch": 0.7686482393032942,
+      "grad_norm": 12.048510551452637,
+      "learning_rate": 2.5640085903287313e-05,
+      "loss": 1.4117,
+      "step": 2030
+    },
+    {
+      "epoch": 0.769026883756153,
+      "grad_norm": 14.081696510314941,
+      "learning_rate": 2.5559996280278196e-05,
+      "loss": 1.3464,
+      "step": 2031
+    },
+    {
+      "epoch": 0.7694055282090118,
+      "grad_norm": 12.886058807373047,
+      "learning_rate": 2.548001360511396e-05,
+      "loss": 1.6561,
+      "step": 2032
+    },
+    {
+      "epoch": 0.7697841726618705,
+      "grad_norm": 10.942986488342285,
+      "learning_rate": 2.5400137992705686e-05,
+      "loss": 1.01,
+      "step": 2033
+    },
+    {
+      "epoch": 0.7701628171147292,
+      "grad_norm": 10.077105522155762,
+      "learning_rate": 2.5320369557810496e-05,
+      "loss": 0.8631,
+      "step": 2034
+    },
+    {
+      "epoch": 0.7705414615675881,
+      "grad_norm": 16.22908592224121,
+      "learning_rate": 2.52407084150316e-05,
+      "loss": 2.036,
+      "step": 2035
+    },
+    {
+      "epoch": 0.7709201060204468,
+      "grad_norm": 12.806578636169434,
+      "learning_rate": 2.516115467881801e-05,
+      "loss": 1.5009,
+      "step": 2036
+    },
+    {
+      "epoch": 0.7712987504733055,
+      "grad_norm": 8.642051696777344,
+      "learning_rate": 2.5081708463464525e-05,
+      "loss": 0.5311,
+      "step": 2037
+    },
+    {
+      "epoch": 0.7716773949261644,
+      "grad_norm": 9.904001235961914,
+      "learning_rate": 2.5002369883111375e-05,
+      "loss": 0.8588,
+      "step": 2038
+    },
+    {
+      "epoch": 0.7720560393790231,
+      "grad_norm": 11.078864097595215,
+      "learning_rate": 2.492313905174418e-05,
+      "loss": 0.6772,
+      "step": 2039
+    },
+    {
+      "epoch": 0.7724346838318819,
+      "grad_norm": 14.320399284362793,
+      "learning_rate": 2.4844016083193745e-05,
+      "loss": 1.6373,
+      "step": 2040
+    },
+    {
+      "epoch": 0.7728133282847406,
+      "grad_norm": 12.595783233642578,
+      "learning_rate": 2.4765001091135965e-05,
+      "loss": 0.7032,
+      "step": 2041
+    },
+    {
+      "epoch": 0.7731919727375994,
+      "grad_norm": 11.836488723754883,
+      "learning_rate": 2.4686094189091548e-05,
+      "loss": 0.4463,
+      "step": 2042
+    },
+    {
+      "epoch": 0.7735706171904582,
+      "grad_norm": 9.597789764404297,
+      "learning_rate": 2.460729549042592e-05,
+      "loss": 0.4927,
+      "step": 2043
+    },
+    {
+      "epoch": 0.7739492616433169,
+      "grad_norm": 8.733850479125977,
+      "learning_rate": 2.4528605108349044e-05,
+      "loss": 0.4759,
+      "step": 2044
+    },
+    {
+      "epoch": 0.7743279060961757,
+      "grad_norm": 11.219993591308594,
+      "learning_rate": 2.4450023155915304e-05,
+      "loss": 0.7328,
+      "step": 2045
+    },
+    {
+      "epoch": 0.7747065505490345,
+      "grad_norm": 27.845932006835938,
+      "learning_rate": 2.4371549746023214e-05,
+      "loss": 1.4339,
+      "step": 2046
+    },
+    {
+      "epoch": 0.7750851950018932,
+      "grad_norm": 10.190560340881348,
+      "learning_rate": 2.4293184991415496e-05,
+      "loss": 0.3212,
+      "step": 2047
+    },
+    {
+      "epoch": 0.775463839454752,
+      "grad_norm": 16.55055809020996,
+      "learning_rate": 2.4214929004678644e-05,
+      "loss": 0.726,
+      "step": 2048
+    },
+    {
+      "epoch": 0.7758424839076108,
+      "grad_norm": 19.725399017333984,
+      "learning_rate": 2.41367818982429e-05,
+      "loss": 0.6447,
+      "step": 2049
+    },
+    {
+      "epoch": 0.7762211283604695,
+      "grad_norm": 3.192269802093506,
+      "learning_rate": 2.405874378438212e-05,
+      "loss": 0.1195,
+      "step": 2050
+    },
+    {
+      "epoch": 0.7765997728133283,
+      "grad_norm": 9.078378677368164,
+      "learning_rate": 2.3980814775213546e-05,
+      "loss": 2.2485,
+      "step": 2051
+    },
+    {
+      "epoch": 0.7769784172661871,
+      "grad_norm": 12.071084976196289,
+      "learning_rate": 2.3902994982697625e-05,
+      "loss": 2.0394,
+      "step": 2052
+    },
+    {
+      "epoch": 0.7773570617190458,
+      "grad_norm": 11.375093460083008,
+      "learning_rate": 2.3825284518638026e-05,
+      "loss": 1.6176,
+      "step": 2053
+    },
+    {
+      "epoch": 0.7777357061719046,
+      "grad_norm": 10.902304649353027,
+      "learning_rate": 2.3747683494681193e-05,
+      "loss": 1.3598,
+      "step": 2054
+    },
+    {
+      "epoch": 0.7781143506247633,
+      "grad_norm": 9.987029075622559,
+      "learning_rate": 2.367019202231644e-05,
+      "loss": 0.9933,
+      "step": 2055
+    },
+    {
+      "epoch": 0.7784929950776222,
+      "grad_norm": 12.458052635192871,
+      "learning_rate": 2.3592810212875615e-05,
+      "loss": 1.0712,
+      "step": 2056
+    },
+    {
+      "epoch": 0.7788716395304809,
+      "grad_norm": 13.190988540649414,
+      "learning_rate": 2.351553817753309e-05,
+      "loss": 0.9407,
+      "step": 2057
+    },
+    {
+      "epoch": 0.7792502839833396,
+      "grad_norm": 13.746674537658691,
+      "learning_rate": 2.3438376027305486e-05,
+      "loss": 1.8128,
+      "step": 2058
+    },
+    {
+      "epoch": 0.7796289284361985,
+      "grad_norm": 12.625304222106934,
+      "learning_rate": 2.336132387305152e-05,
+      "loss": 1.0643,
+      "step": 2059
+    },
+    {
+      "epoch": 0.7800075728890572,
+      "grad_norm": 9.5991792678833,
+      "learning_rate": 2.32843818254719e-05,
+      "loss": 0.9507,
+      "step": 2060
+    },
+    {
+      "epoch": 0.7803862173419159,
+      "grad_norm": 11.39356517791748,
+      "learning_rate": 2.3207549995109213e-05,
+      "loss": 0.8554,
+      "step": 2061
+    },
+    {
+      "epoch": 0.7807648617947747,
+      "grad_norm": 10.822458267211914,
+      "learning_rate": 2.3130828492347613e-05,
+      "loss": 0.7323,
+      "step": 2062
+    },
+    {
+      "epoch": 0.7811435062476335,
+      "grad_norm": 14.810583114624023,
+      "learning_rate": 2.305421742741275e-05,
+      "loss": 1.0549,
+      "step": 2063
+    },
+    {
+      "epoch": 0.7815221507004922,
+      "grad_norm": 14.234593391418457,
+      "learning_rate": 2.2977716910371617e-05,
+      "loss": 0.8004,
+      "step": 2064
+    },
+    {
+      "epoch": 0.781900795153351,
+      "grad_norm": 9.462289810180664,
+      "learning_rate": 2.2901327051132436e-05,
+      "loss": 0.6268,
+      "step": 2065
+    },
+    {
+      "epoch": 0.7822794396062097,
+      "grad_norm": 9.984630584716797,
+      "learning_rate": 2.2825047959444402e-05,
+      "loss": 0.5759,
+      "step": 2066
+    },
+    {
+      "epoch": 0.7826580840590686,
+      "grad_norm": 8.960870742797852,
+      "learning_rate": 2.2748879744897566e-05,
+      "loss": 0.478,
+      "step": 2067
+    },
+    {
+      "epoch": 0.7830367285119273,
+      "grad_norm": 12.594315528869629,
+      "learning_rate": 2.2672822516922664e-05,
+      "loss": 0.7478,
+      "step": 2068
+    },
+    {
+      "epoch": 0.783415372964786,
+      "grad_norm": 11.013339042663574,
+      "learning_rate": 2.2596876384791044e-05,
+      "loss": 0.4485,
+      "step": 2069
+    },
+    {
+      "epoch": 0.7837940174176449,
+      "grad_norm": 14.870506286621094,
+      "learning_rate": 2.25210414576144e-05,
+      "loss": 1.0851,
+      "step": 2070
+    },
+    {
+      "epoch": 0.7841726618705036,
+      "grad_norm": 11.4357271194458,
+      "learning_rate": 2.2445317844344648e-05,
+      "loss": 0.6047,
+      "step": 2071
+    },
+    {
+      "epoch": 0.7845513063233623,
+      "grad_norm": 14.853757858276367,
+      "learning_rate": 2.2369705653773765e-05,
+      "loss": 0.696,
+      "step": 2072
+    },
+    {
+      "epoch": 0.7849299507762211,
+      "grad_norm": 19.04551124572754,
+      "learning_rate": 2.2294204994533728e-05,
+      "loss": 0.5838,
+      "step": 2073
+    },
+    {
+      "epoch": 0.7853085952290799,
+      "grad_norm": 21.33306312561035,
+      "learning_rate": 2.2218815975096207e-05,
+      "loss": 0.232,
+      "step": 2074
+    },
+    {
+      "epoch": 0.7856872396819387,
+      "grad_norm": 7.215750694274902,
+      "learning_rate": 2.2143538703772493e-05,
+      "loss": 0.3659,
+      "step": 2075
+    },
+    {
+      "epoch": 0.7860658841347974,
+      "grad_norm": 10.023611068725586,
+      "learning_rate": 2.2068373288713294e-05,
+      "loss": 1.963,
+      "step": 2076
+    },
+    {
+      "epoch": 0.7864445285876562,
+      "grad_norm": 11.151053428649902,
+      "learning_rate": 2.1993319837908722e-05,
+      "loss": 1.8424,
+      "step": 2077
+    },
+    {
+      "epoch": 0.786823173040515,
+      "grad_norm": 10.198956489562988,
+      "learning_rate": 2.1918378459187928e-05,
+      "loss": 1.7327,
+      "step": 2078
+    },
+    {
+      "epoch": 0.7872018174933737,
+      "grad_norm": 13.809074401855469,
+      "learning_rate": 2.1843549260219075e-05,
+      "loss": 1.5646,
+      "step": 2079
+    },
+    {
+      "epoch": 0.7875804619462324,
+      "grad_norm": 10.254709243774414,
+      "learning_rate": 2.176883234850914e-05,
+      "loss": 0.9571,
+      "step": 2080
+    },
+    {
+      "epoch": 0.7879591063990913,
+      "grad_norm": 9.651457786560059,
+      "learning_rate": 2.1694227831403868e-05,
+      "loss": 1.0734,
+      "step": 2081
+    },
+    {
+      "epoch": 0.78833775085195,
+      "grad_norm": 12.739899635314941,
+      "learning_rate": 2.1619735816087417e-05,
+      "loss": 1.1163,
+      "step": 2082
+    },
+    {
+      "epoch": 0.7887163953048087,
+      "grad_norm": 9.453640937805176,
+      "learning_rate": 2.154535640958235e-05,
+      "loss": 0.7136,
+      "step": 2083
+    },
+    {
+      "epoch": 0.7890950397576676,
+      "grad_norm": 11.674582481384277,
+      "learning_rate": 2.147108971874946e-05,
+      "loss": 1.197,
+      "step": 2084
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 11.435443878173828,
+      "learning_rate": 2.1396935850287615e-05,
+      "loss": 0.8756,
+      "step": 2085
+    },
+    {
+      "epoch": 0.7898523286633851,
+      "grad_norm": 11.762517929077148,
+      "learning_rate": 2.1322894910733547e-05,
+      "loss": 0.8406,
+      "step": 2086
+    },
+    {
+      "epoch": 0.7902309731162438,
+      "grad_norm": 11.67284107208252,
+      "learning_rate": 2.1248967006461783e-05,
+      "loss": 1.0792,
+      "step": 2087
+    },
+    {
+      "epoch": 0.7906096175691026,
+      "grad_norm": 11.377511978149414,
+      "learning_rate": 2.117515224368438e-05,
+      "loss": 0.8553,
+      "step": 2088
+    },
+    {
+      "epoch": 0.7909882620219614,
+      "grad_norm": 13.04310131072998,
+      "learning_rate": 2.110145072845099e-05,
+      "loss": 0.9179,
+      "step": 2089
+    },
+    {
+      "epoch": 0.7913669064748201,
+      "grad_norm": 8.919587135314941,
+      "learning_rate": 2.1027862566648425e-05,
+      "loss": 0.415,
+      "step": 2090
+    },
+    {
+      "epoch": 0.791745550927679,
+      "grad_norm": 14.16692066192627,
+      "learning_rate": 2.095438786400068e-05,
+      "loss": 0.6586,
+      "step": 2091
+    },
+    {
+      "epoch": 0.7921241953805377,
+      "grad_norm": 9.801429748535156,
+      "learning_rate": 2.0881026726068775e-05,
+      "loss": 0.5952,
+      "step": 2092
+    },
+    {
+      "epoch": 0.7925028398333964,
+      "grad_norm": 14.237808227539062,
+      "learning_rate": 2.0807779258250537e-05,
+      "loss": 0.5729,
+      "step": 2093
+    },
+    {
+      "epoch": 0.7928814842862552,
+      "grad_norm": 10.763073921203613,
+      "learning_rate": 2.073464556578051e-05,
+      "loss": 0.634,
+      "step": 2094
+    },
+    {
+      "epoch": 0.793260128739114,
+      "grad_norm": 14.481758117675781,
+      "learning_rate": 2.0661625753729707e-05,
+      "loss": 0.6255,
+      "step": 2095
+    },
+    {
+      "epoch": 0.7936387731919727,
+      "grad_norm": 12.738697052001953,
+      "learning_rate": 2.058871992700567e-05,
+      "loss": 0.4656,
+      "step": 2096
+    },
+    {
+      "epoch": 0.7940174176448315,
+      "grad_norm": 9.8129243850708,
+      "learning_rate": 2.0515928190352052e-05,
+      "loss": 0.352,
+      "step": 2097
+    },
+    {
+      "epoch": 0.7943960620976903,
+      "grad_norm": 7.001706123352051,
+      "learning_rate": 2.0443250648348645e-05,
+      "loss": 0.1885,
+      "step": 2098
+    },
+    {
+      "epoch": 0.794774706550549,
+      "grad_norm": 10.005912780761719,
+      "learning_rate": 2.037068740541116e-05,
+      "loss": 0.4027,
+      "step": 2099
+    },
+    {
+      "epoch": 0.7951533510034078,
+      "grad_norm": 29.922603607177734,
+      "learning_rate": 2.0298238565791072e-05,
+      "loss": 0.9724,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7955319954562665,
+      "grad_norm": 9.894120216369629,
+      "learning_rate": 2.0225904233575586e-05,
+      "loss": 2.2812,
+      "step": 2101
+    },
+    {
+      "epoch": 0.7959106399091254,
+      "grad_norm": 10.518170356750488,
+      "learning_rate": 2.0153684512687297e-05,
+      "loss": 1.9124,
+      "step": 2102
+    },
+    {
+      "epoch": 0.7962892843619841,
+      "grad_norm": 11.77073860168457,
+      "learning_rate": 2.0081579506884184e-05,
+      "loss": 1.714,
+      "step": 2103
+    },
+    {
+      "epoch": 0.7966679288148428,
+      "grad_norm": 10.411770820617676,
+      "learning_rate": 2.0009589319759358e-05,
+      "loss": 1.084,
+      "step": 2104
+    },
+    {
+      "epoch": 0.7970465732677017,
+      "grad_norm": 10.7216796875,
+      "learning_rate": 1.9937714054741095e-05,
+      "loss": 1.2028,
+      "step": 2105
+    },
+    {
+      "epoch": 0.7974252177205604,
+      "grad_norm": 12.313328742980957,
+      "learning_rate": 1.9865953815092443e-05,
+      "loss": 1.3527,
+      "step": 2106
+    },
+    {
+      "epoch": 0.7978038621734191,
+      "grad_norm": 11.44083023071289,
+      "learning_rate": 1.9794308703911223e-05,
+      "loss": 1.308,
+      "step": 2107
+    },
+    {
+      "epoch": 0.7981825066262779,
+      "grad_norm": 7.195469856262207,
+      "learning_rate": 1.972277882412985e-05,
+      "loss": 0.6015,
+      "step": 2108
+    },
+    {
+      "epoch": 0.7985611510791367,
+      "grad_norm": 9.268226623535156,
+      "learning_rate": 1.965136427851525e-05,
+      "loss": 0.713,
+      "step": 2109
+    },
+    {
+      "epoch": 0.7989397955319955,
+      "grad_norm": 11.629197120666504,
+      "learning_rate": 1.958006516966857e-05,
+      "loss": 1.1342,
+      "step": 2110
+    },
+    {
+      "epoch": 0.7993184399848542,
+      "grad_norm": 9.284026145935059,
+      "learning_rate": 1.950888160002513e-05,
+      "loss": 0.652,
+      "step": 2111
+    },
+    {
+      "epoch": 0.799697084437713,
+      "grad_norm": 11.595834732055664,
+      "learning_rate": 1.9437813671854243e-05,
+      "loss": 0.7744,
+      "step": 2112
+    },
+    {
+      "epoch": 0.8000757288905718,
+      "grad_norm": 12.920877456665039,
+      "learning_rate": 1.9366861487259134e-05,
+      "loss": 0.7837,
+      "step": 2113
+    },
+    {
+      "epoch": 0.8004543733434305,
+      "grad_norm": 11.606154441833496,
+      "learning_rate": 1.92960251481767e-05,
+      "loss": 0.7631,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8008330177962892,
+      "grad_norm": 11.07241153717041,
+      "learning_rate": 1.9225304756377394e-05,
+      "loss": 0.6119,
+      "step": 2115
+    },
+    {
+      "epoch": 0.8012116622491481,
+      "grad_norm": 9.720746994018555,
+      "learning_rate": 1.9154700413465077e-05,
+      "loss": 0.5343,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8015903067020068,
+      "grad_norm": 13.459938049316406,
+      "learning_rate": 1.9084212220876942e-05,
+      "loss": 0.8311,
+      "step": 2117
+    },
+    {
+      "epoch": 0.8019689511548656,
+      "grad_norm": 19.92449188232422,
+      "learning_rate": 1.9013840279883267e-05,
+      "loss": 1.5106,
+      "step": 2118
+    },
+    {
+      "epoch": 0.8023475956077244,
+      "grad_norm": 12.470542907714844,
+      "learning_rate": 1.8943584691587313e-05,
+      "loss": 0.5453,
+      "step": 2119
+    },
+    {
+      "epoch": 0.8027262400605831,
+      "grad_norm": 11.011085510253906,
+      "learning_rate": 1.887344555692515e-05,
+      "loss": 0.9123,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8031048845134419,
+      "grad_norm": 9.609660148620605,
+      "learning_rate": 1.880342297666563e-05,
+      "loss": 0.3907,
+      "step": 2121
+    },
+    {
+      "epoch": 0.8034835289663006,
+      "grad_norm": 18.808673858642578,
+      "learning_rate": 1.8733517051410054e-05,
+      "loss": 0.7887,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8038621734191594,
+      "grad_norm": 11.743169784545898,
+      "learning_rate": 1.8663727881592176e-05,
+      "loss": 0.6305,
+      "step": 2123
+    },
+    {
+      "epoch": 0.8042408178720182,
+      "grad_norm": 15.70638656616211,
+      "learning_rate": 1.8594055567477965e-05,
+      "loss": 0.2316,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8046194623248769,
+      "grad_norm": 18.647750854492188,
+      "learning_rate": 1.8524500209165573e-05,
+      "loss": 0.2247,
+      "step": 2125
+    },
+    {
+      "epoch": 0.8049981067777358,
+      "grad_norm": 11.47064208984375,
+      "learning_rate": 1.8455061906585068e-05,
+      "loss": 2.2448,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8053767512305945,
+      "grad_norm": 11.74728775024414,
+      "learning_rate": 1.838574075949836e-05,
+      "loss": 1.9097,
+      "step": 2127
+    },
+    {
+      "epoch": 0.8057553956834532,
+      "grad_norm": 10.916732788085938,
+      "learning_rate": 1.8316536867499013e-05,
+      "loss": 1.5175,
+      "step": 2128
+    },
+    {
+      "epoch": 0.806134040136312,
+      "grad_norm": 12.84622573852539,
+      "learning_rate": 1.8247450330012206e-05,
+      "loss": 1.2506,
+      "step": 2129
+    },
+    {
+      "epoch": 0.8065126845891708,
+      "grad_norm": 11.161234855651855,
+      "learning_rate": 1.8178481246294433e-05,
+      "loss": 1.2847,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8068913290420295,
+      "grad_norm": 11.394064903259277,
+      "learning_rate": 1.8109629715433497e-05,
+      "loss": 1.2467,
+      "step": 2131
+    },
+    {
+      "epoch": 0.8072699734948883,
+      "grad_norm": 14.243141174316406,
+      "learning_rate": 1.804089583634825e-05,
+      "loss": 1.1874,
+      "step": 2132
+    },
+    {
+      "epoch": 0.8076486179477471,
+      "grad_norm": 9.28836727142334,
+      "learning_rate": 1.7972279707788608e-05,
+      "loss": 0.9387,
+      "step": 2133
+    },
+    {
+      "epoch": 0.8080272624006059,
+      "grad_norm": 8.111406326293945,
+      "learning_rate": 1.790378142833524e-05,
+      "loss": 0.538,
+      "step": 2134
+    },
+    {
+      "epoch": 0.8084059068534646,
+      "grad_norm": 12.397017478942871,
+      "learning_rate": 1.783540109639953e-05,
+      "loss": 0.8542,
+      "step": 2135
+    },
+    {
+      "epoch": 0.8087845513063233,
+      "grad_norm": 13.441441535949707,
+      "learning_rate": 1.776713881022337e-05,
+      "loss": 0.6562,
+      "step": 2136
+    },
+    {
+      "epoch": 0.8091631957591822,
+      "grad_norm": 15.309789657592773,
+      "learning_rate": 1.769899466787913e-05,
+      "loss": 1.038,
+      "step": 2137
+    },
+    {
+      "epoch": 0.8095418402120409,
+      "grad_norm": 9.063993453979492,
+      "learning_rate": 1.7630968767269396e-05,
+      "loss": 0.6075,
+      "step": 2138
+    },
+    {
+      "epoch": 0.8099204846648996,
+      "grad_norm": 13.524983406066895,
+      "learning_rate": 1.7563061206126875e-05,
+      "loss": 1.1014,
+      "step": 2139
+    },
+    {
+      "epoch": 0.8102991291177585,
+      "grad_norm": 10.966734886169434,
+      "learning_rate": 1.7495272082014235e-05,
+      "loss": 0.633,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8106777735706172,
+      "grad_norm": 13.029152870178223,
+      "learning_rate": 1.742760149232404e-05,
+      "loss": 0.7217,
+      "step": 2141
+    },
+    {
+      "epoch": 0.811056418023476,
+      "grad_norm": 10.429238319396973,
+      "learning_rate": 1.736004953427852e-05,
+      "loss": 0.7237,
+      "step": 2142
+    },
+    {
+      "epoch": 0.8114350624763347,
+      "grad_norm": 10.06002140045166,
+      "learning_rate": 1.7292616304929454e-05,
+      "loss": 0.5242,
+      "step": 2143
+    },
+    {
+      "epoch": 0.8118137069291935,
+      "grad_norm": 12.623034477233887,
+      "learning_rate": 1.7225301901158097e-05,
+      "loss": 0.679,
+      "step": 2144
+    },
+    {
+      "epoch": 0.8121923513820523,
+      "grad_norm": 11.331562995910645,
+      "learning_rate": 1.7158106419674956e-05,
+      "loss": 0.5334,
+      "step": 2145
+    },
+    {
+      "epoch": 0.812570995834911,
+      "grad_norm": 5.492001533508301,
+      "learning_rate": 1.7091029957019656e-05,
+      "loss": 0.2017,
+      "step": 2146
+    },
+    {
+      "epoch": 0.8129496402877698,
+      "grad_norm": 15.369951248168945,
+      "learning_rate": 1.702407260956087e-05,
+      "loss": 0.6085,
+      "step": 2147
+    },
+    {
+      "epoch": 0.8133282847406286,
+      "grad_norm": 14.828356742858887,
+      "learning_rate": 1.6957234473496087e-05,
+      "loss": 0.5092,
+      "step": 2148
+    },
+    {
+      "epoch": 0.8137069291934873,
+      "grad_norm": 9.543940544128418,
+      "learning_rate": 1.6890515644851612e-05,
+      "loss": 0.283,
+      "step": 2149
+    },
+    {
+      "epoch": 0.814085573646346,
+      "grad_norm": 10.726631164550781,
+      "learning_rate": 1.6823916219482273e-05,
+      "loss": 0.5966,
+      "step": 2150
+    },
+    {
+      "epoch": 0.8144642180992049,
+      "grad_norm": 8.761448860168457,
+      "learning_rate": 1.6757436293071362e-05,
+      "loss": 1.9309,
+      "step": 2151
+    },
+    {
+      "epoch": 0.8148428625520636,
+      "grad_norm": 8.565874099731445,
+      "learning_rate": 1.6691075961130452e-05,
+      "loss": 1.2893,
+      "step": 2152
+    },
+    {
+      "epoch": 0.8152215070049224,
+      "grad_norm": 9.82201099395752,
+      "learning_rate": 1.662483531899941e-05,
+      "loss": 1.4838,
+      "step": 2153
+    },
+    {
+      "epoch": 0.8156001514577812,
+      "grad_norm": 14.917886734008789,
+      "learning_rate": 1.6558714461846025e-05,
+      "loss": 1.4838,
+      "step": 2154
+    },
+    {
+      "epoch": 0.8159787959106399,
+      "grad_norm": 10.092191696166992,
+      "learning_rate": 1.6492713484666057e-05,
+      "loss": 1.0222,
+      "step": 2155
+    },
+    {
+      "epoch": 0.8163574403634987,
+      "grad_norm": 10.887097358703613,
+      "learning_rate": 1.6426832482282973e-05,
+      "loss": 1.1809,
+      "step": 2156
+    },
+    {
+      "epoch": 0.8167360848163574,
+      "grad_norm": 9.894757270812988,
+      "learning_rate": 1.636107154934796e-05,
+      "loss": 1.2049,
+      "step": 2157
+    },
+    {
+      "epoch": 0.8171147292692162,
+      "grad_norm": 12.424516677856445,
+      "learning_rate": 1.629543078033964e-05,
+      "loss": 0.8421,
+      "step": 2158
+    },
+    {
+      "epoch": 0.817493373722075,
+      "grad_norm": 8.398805618286133,
+      "learning_rate": 1.622991026956401e-05,
+      "loss": 0.7292,
+      "step": 2159
+    },
+    {
+      "epoch": 0.8178720181749337,
+      "grad_norm": 7.698610305786133,
+      "learning_rate": 1.616451011115426e-05,
+      "loss": 0.6156,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8182506626277926,
+      "grad_norm": 12.355453491210938,
+      "learning_rate": 1.6099230399070763e-05,
+      "loss": 0.7348,
+      "step": 2161
+    },
+    {
+      "epoch": 0.8186293070806513,
+      "grad_norm": 9.122222900390625,
+      "learning_rate": 1.6034071227100755e-05,
+      "loss": 0.6662,
+      "step": 2162
+    },
+    {
+      "epoch": 0.81900795153351,
+      "grad_norm": 13.555700302124023,
+      "learning_rate": 1.596903268885832e-05,
+      "loss": 0.7399,
+      "step": 2163
+    },
+    {
+      "epoch": 0.8193865959863688,
+      "grad_norm": 10.907391548156738,
+      "learning_rate": 1.5904114877784205e-05,
+      "loss": 0.5784,
+      "step": 2164
+    },
+    {
+      "epoch": 0.8197652404392276,
+      "grad_norm": 10.284018516540527,
+      "learning_rate": 1.5839317887145798e-05,
+      "loss": 0.7392,
+      "step": 2165
+    },
+    {
+      "epoch": 0.8201438848920863,
+      "grad_norm": 11.968799591064453,
+      "learning_rate": 1.5774641810036793e-05,
+      "loss": 0.776,
+      "step": 2166
+    },
+    {
+      "epoch": 0.8205225293449451,
+      "grad_norm": 8.554612159729004,
+      "learning_rate": 1.571008673937724e-05,
+      "loss": 0.3263,
+      "step": 2167
+    },
+    {
+      "epoch": 0.8209011737978038,
+      "grad_norm": 12.886919021606445,
+      "learning_rate": 1.5645652767913287e-05,
+      "loss": 0.9284,
+      "step": 2168
+    },
+    {
+      "epoch": 0.8212798182506627,
+      "grad_norm": 13.11758041381836,
+      "learning_rate": 1.5581339988217157e-05,
+      "loss": 0.926,
+      "step": 2169
+    },
+    {
+      "epoch": 0.8216584627035214,
+      "grad_norm": 7.769589900970459,
+      "learning_rate": 1.5517148492686918e-05,
+      "loss": 0.4201,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8220371071563801,
+      "grad_norm": 11.231391906738281,
+      "learning_rate": 1.5453078373546405e-05,
+      "loss": 0.4707,
+      "step": 2171
+    },
+    {
+      "epoch": 0.822415751609239,
+      "grad_norm": 10.800396919250488,
+      "learning_rate": 1.538912972284502e-05,
+      "loss": 0.3822,
+      "step": 2172
+    },
+    {
+      "epoch": 0.8227943960620977,
+      "grad_norm": 4.893832683563232,
+      "learning_rate": 1.532530263245776e-05,
+      "loss": 0.2115,
+      "step": 2173
+    },
+    {
+      "epoch": 0.8231730405149564,
+      "grad_norm": 14.236945152282715,
+      "learning_rate": 1.5261597194084876e-05,
+      "loss": 0.3221,
+      "step": 2174
+    },
+    {
+      "epoch": 0.8235516849678152,
+      "grad_norm": 26.767309188842773,
+      "learning_rate": 1.5198013499251895e-05,
+      "loss": 1.1176,
+      "step": 2175
+    },
+    {
+      "epoch": 0.823930329420674,
+      "grad_norm": 10.102524757385254,
+      "learning_rate": 1.513455163930938e-05,
+      "loss": 2.0341,
+      "step": 2176
+    },
+    {
+      "epoch": 0.8243089738735327,
+      "grad_norm": 10.047316551208496,
+      "learning_rate": 1.5071211705432953e-05,
+      "loss": 1.598,
+      "step": 2177
+    },
+    {
+      "epoch": 0.8246876183263915,
+      "grad_norm": 11.753214836120605,
+      "learning_rate": 1.5007993788622977e-05,
+      "loss": 1.2507,
+      "step": 2178
+    },
+    {
+      "epoch": 0.8250662627792503,
+      "grad_norm": 9.349088668823242,
+      "learning_rate": 1.4944897979704531e-05,
+      "loss": 1.216,
+      "step": 2179
+    },
+    {
+      "epoch": 0.8254449072321091,
+      "grad_norm": 8.784585952758789,
+      "learning_rate": 1.4881924369327261e-05,
+      "loss": 0.7352,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8258235516849678,
+      "grad_norm": 11.812366485595703,
+      "learning_rate": 1.4819073047965304e-05,
+      "loss": 0.9288,
+      "step": 2181
+    },
+    {
+      "epoch": 0.8262021961378265,
+      "grad_norm": 8.885299682617188,
+      "learning_rate": 1.475634410591703e-05,
+      "loss": 0.8133,
+      "step": 2182
+    },
+    {
+      "epoch": 0.8265808405906854,
+      "grad_norm": 14.608964920043945,
+      "learning_rate": 1.4693737633305038e-05,
+      "loss": 1.073,
+      "step": 2183
+    },
+    {
+      "epoch": 0.8269594850435441,
+      "grad_norm": 12.234383583068848,
+      "learning_rate": 1.463125372007591e-05,
+      "loss": 1.1562,
+      "step": 2184
+    },
+    {
+      "epoch": 0.8273381294964028,
+      "grad_norm": 14.701723098754883,
+      "learning_rate": 1.456889245600026e-05,
+      "loss": 1.4302,
+      "step": 2185
+    },
+    {
+      "epoch": 0.8277167739492617,
+      "grad_norm": 8.882140159606934,
+      "learning_rate": 1.4506653930672387e-05,
+      "loss": 0.6384,
+      "step": 2186
+    },
+    {
+      "epoch": 0.8280954184021204,
+      "grad_norm": 9.586640357971191,
+      "learning_rate": 1.4444538233510296e-05,
+      "loss": 0.545,
+      "step": 2187
+    },
+    {
+      "epoch": 0.8284740628549792,
+      "grad_norm": 8.397727012634277,
+      "learning_rate": 1.4382545453755524e-05,
+      "loss": 0.4565,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8288527073078379,
+      "grad_norm": 13.309687614440918,
+      "learning_rate": 1.4320675680472995e-05,
+      "loss": 0.9385,
+      "step": 2189
+    },
+    {
+      "epoch": 0.8292313517606967,
+      "grad_norm": 12.719165802001953,
+      "learning_rate": 1.4258929002550925e-05,
+      "loss": 0.8029,
+      "step": 2190
+    },
+    {
+      "epoch": 0.8296099962135555,
+      "grad_norm": 11.410460472106934,
+      "learning_rate": 1.4197305508700665e-05,
+      "loss": 0.618,
+      "step": 2191
+    },
+    {
+      "epoch": 0.8299886406664142,
+      "grad_norm": 11.324352264404297,
+      "learning_rate": 1.4135805287456638e-05,
+      "loss": 0.543,
+      "step": 2192
+    },
+    {
+      "epoch": 0.830367285119273,
+      "grad_norm": 11.747859954833984,
+      "learning_rate": 1.407442842717609e-05,
+      "loss": 0.575,
+      "step": 2193
+    },
+    {
+      "epoch": 0.8307459295721318,
+      "grad_norm": 9.637734413146973,
+      "learning_rate": 1.4013175016039082e-05,
+      "loss": 0.3842,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8311245740249905,
+      "grad_norm": 10.062605857849121,
+      "learning_rate": 1.3952045142048287e-05,
+      "loss": 0.398,
+      "step": 2195
+    },
+    {
+      "epoch": 0.8315032184778492,
+      "grad_norm": 9.133951187133789,
+      "learning_rate": 1.3891038893028897e-05,
+      "loss": 0.3874,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8318818629307081,
+      "grad_norm": 11.680546760559082,
+      "learning_rate": 1.3830156356628531e-05,
+      "loss": 0.5123,
+      "step": 2197
+    },
+    {
+      "epoch": 0.8322605073835668,
+      "grad_norm": 9.764351844787598,
+      "learning_rate": 1.3769397620317038e-05,
+      "loss": 0.5171,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8326391518364256,
+      "grad_norm": 14.688189506530762,
+      "learning_rate": 1.3708762771386386e-05,
+      "loss": 0.2875,
+      "step": 2199
+    },
+    {
+      "epoch": 0.8330177962892844,
+      "grad_norm": 25.08920669555664,
+      "learning_rate": 1.364825189695056e-05,
+      "loss": 0.6495,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8333964407421431,
+      "grad_norm": 9.983552932739258,
+      "learning_rate": 1.3587865083945483e-05,
+      "loss": 1.9564,
+      "step": 2201
+    },
+    {
+      "epoch": 0.8337750851950019,
+      "grad_norm": 10.196476936340332,
+      "learning_rate": 1.3527602419128793e-05,
+      "loss": 1.8545,
+      "step": 2202
+    },
+    {
+      "epoch": 0.8341537296478606,
+      "grad_norm": 10.242213249206543,
+      "learning_rate": 1.3467463989079764e-05,
+      "loss": 1.3327,
+      "step": 2203
+    },
+    {
+      "epoch": 0.8345323741007195,
+      "grad_norm": 11.082919120788574,
+      "learning_rate": 1.3407449880199175e-05,
+      "loss": 1.2277,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8349110185535782,
+      "grad_norm": 14.372567176818848,
+      "learning_rate": 1.334756017870924e-05,
+      "loss": 1.2489,
+      "step": 2205
+    },
+    {
+      "epoch": 0.8352896630064369,
+      "grad_norm": 9.300726890563965,
+      "learning_rate": 1.328779497065339e-05,
+      "loss": 0.6838,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8356683074592958,
+      "grad_norm": 10.5520658493042,
+      "learning_rate": 1.3228154341896225e-05,
+      "loss": 0.9467,
+      "step": 2207
+    },
+    {
+      "epoch": 0.8360469519121545,
+      "grad_norm": 11.126348495483398,
+      "learning_rate": 1.316863837812331e-05,
+      "loss": 0.9089,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8364255963650132,
+      "grad_norm": 14.790655136108398,
+      "learning_rate": 1.3109247164841199e-05,
+      "loss": 0.6592,
+      "step": 2209
+    },
+    {
+      "epoch": 0.836804240817872,
+      "grad_norm": 10.294766426086426,
+      "learning_rate": 1.3049980787377126e-05,
+      "loss": 0.5883,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8371828852707308,
+      "grad_norm": 14.918768882751465,
+      "learning_rate": 1.2990839330879024e-05,
+      "loss": 0.8069,
+      "step": 2211
+    },
+    {
+      "epoch": 0.8375615297235895,
+      "grad_norm": 11.903068542480469,
+      "learning_rate": 1.2931822880315303e-05,
+      "loss": 1.0219,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8379401741764483,
+      "grad_norm": 11.20407772064209,
+      "learning_rate": 1.2872931520474873e-05,
+      "loss": 0.5456,
+      "step": 2213
+    },
+    {
+      "epoch": 0.8383188186293071,
+      "grad_norm": 11.456269264221191,
+      "learning_rate": 1.281416533596682e-05,
+      "loss": 0.5934,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8386974630821659,
+      "grad_norm": 13.103646278381348,
+      "learning_rate": 1.2755524411220455e-05,
+      "loss": 1.0718,
+      "step": 2215
+    },
+    {
+      "epoch": 0.8390761075350246,
+      "grad_norm": 11.291081428527832,
+      "learning_rate": 1.269700883048508e-05,
+      "loss": 0.7964,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8394547519878833,
+      "grad_norm": 9.637993812561035,
+      "learning_rate": 1.2638618677829983e-05,
+      "loss": 0.5809,
+      "step": 2217
+    },
+    {
+      "epoch": 0.8398333964407422,
+      "grad_norm": 17.41292953491211,
+      "learning_rate": 1.2580354037144194e-05,
+      "loss": 0.8842,
+      "step": 2218
+    },
+    {
+      "epoch": 0.8402120408936009,
+      "grad_norm": 7.992457866668701,
+      "learning_rate": 1.2522214992136449e-05,
+      "loss": 0.3386,
+      "step": 2219
+    },
+    {
+      "epoch": 0.8405906853464596,
+      "grad_norm": 8.05163288116455,
+      "learning_rate": 1.2464201626334982e-05,
+      "loss": 0.467,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8409693297993185,
+      "grad_norm": 7.449150562286377,
+      "learning_rate": 1.2406314023087584e-05,
+      "loss": 0.3405,
+      "step": 2221
+    },
+    {
+      "epoch": 0.8413479742521772,
+      "grad_norm": 8.64903736114502,
+      "learning_rate": 1.2348552265561242e-05,
+      "loss": 0.3131,
+      "step": 2222
+    },
+    {
+      "epoch": 0.841726618705036,
+      "grad_norm": 11.01096248626709,
+      "learning_rate": 1.2290916436742205e-05,
+      "loss": 0.377,
+      "step": 2223
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 9.92861270904541,
+      "learning_rate": 1.223340661943576e-05,
+      "loss": 0.2074,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8424839076107535,
+      "grad_norm": 8.516009330749512,
+      "learning_rate": 1.2176022896266214e-05,
+      "loss": 0.1966,
+      "step": 2225
+    },
+    {
+      "epoch": 0.8428625520636123,
+      "grad_norm": 13.25462818145752,
+      "learning_rate": 1.2118765349676664e-05,
+      "loss": 2.6535,
+      "step": 2226
+    },
+    {
+      "epoch": 0.843241196516471,
+      "grad_norm": 12.067037582397461,
+      "learning_rate": 1.206163406192895e-05,
+      "loss": 1.7434,
+      "step": 2227
+    },
+    {
+      "epoch": 0.8436198409693298,
+      "grad_norm": 11.304587364196777,
+      "learning_rate": 1.2004629115103471e-05,
+      "loss": 1.6441,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8439984854221886,
+      "grad_norm": 9.689388275146484,
+      "learning_rate": 1.1947750591099206e-05,
+      "loss": 0.9401,
+      "step": 2229
+    },
+    {
+      "epoch": 0.8443771298750473,
+      "grad_norm": 13.32154655456543,
+      "learning_rate": 1.1890998571633427e-05,
+      "loss": 1.0714,
+      "step": 2230
+    },
+    {
+      "epoch": 0.844755774327906,
+      "grad_norm": 11.89297103881836,
+      "learning_rate": 1.1834373138241672e-05,
+      "loss": 1.1308,
+      "step": 2231
+    },
+    {
+      "epoch": 0.8451344187807649,
+      "grad_norm": 11.245408058166504,
+      "learning_rate": 1.1777874372277597e-05,
+      "loss": 0.7964,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8455130632336236,
+      "grad_norm": 8.075695991516113,
+      "learning_rate": 1.1721502354912939e-05,
+      "loss": 0.5496,
+      "step": 2233
+    },
+    {
+      "epoch": 0.8458917076864824,
+      "grad_norm": 13.818008422851562,
+      "learning_rate": 1.1665257167137289e-05,
+      "loss": 0.8236,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8462703521393412,
+      "grad_norm": 10.466621398925781,
+      "learning_rate": 1.1609138889757998e-05,
+      "loss": 0.7633,
+      "step": 2235
+    },
+    {
+      "epoch": 0.8466489965921999,
+      "grad_norm": 10.655759811401367,
+      "learning_rate": 1.1553147603400139e-05,
+      "loss": 0.7656,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8470276410450587,
+      "grad_norm": 11.960763931274414,
+      "learning_rate": 1.1497283388506285e-05,
+      "loss": 0.7411,
+      "step": 2237
+    },
+    {
+      "epoch": 0.8474062854979174,
+      "grad_norm": 10.754295349121094,
+      "learning_rate": 1.1441546325336505e-05,
+      "loss": 0.6966,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8477849299507763,
+      "grad_norm": 10.447911262512207,
+      "learning_rate": 1.1385936493968108e-05,
+      "loss": 0.5975,
+      "step": 2239
+    },
+    {
+      "epoch": 0.848163574403635,
+      "grad_norm": 7.784918308258057,
+      "learning_rate": 1.1330453974295708e-05,
+      "loss": 0.5167,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8485422188564937,
+      "grad_norm": 7.070992946624756,
+      "learning_rate": 1.127509884603095e-05,
+      "loss": 0.4106,
+      "step": 2241
+    },
+    {
+      "epoch": 0.8489208633093526,
+      "grad_norm": 10.769533157348633,
+      "learning_rate": 1.1219871188702447e-05,
+      "loss": 0.6198,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8492995077622113,
+      "grad_norm": 11.375531196594238,
+      "learning_rate": 1.1164771081655712e-05,
+      "loss": 0.4514,
+      "step": 2243
+    },
+    {
+      "epoch": 0.84967815221507,
+      "grad_norm": 10.334220886230469,
+      "learning_rate": 1.1109798604052957e-05,
+      "loss": 0.239,
+      "step": 2244
+    },
+    {
+      "epoch": 0.8500567966679288,
+      "grad_norm": 5.8680315017700195,
+      "learning_rate": 1.1054953834873095e-05,
+      "loss": 0.1725,
+      "step": 2245
+    },
+    {
+      "epoch": 0.8504354411207876,
+      "grad_norm": 11.254388809204102,
+      "learning_rate": 1.1000236852911527e-05,
+      "loss": 0.4195,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8508140855736464,
+      "grad_norm": 10.101336479187012,
+      "learning_rate": 1.0945647736780052e-05,
+      "loss": 0.3015,
+      "step": 2247
+    },
+    {
+      "epoch": 0.8511927300265051,
+      "grad_norm": 10.285369873046875,
+      "learning_rate": 1.0891186564906742e-05,
+      "loss": 0.2216,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8515713744793639,
+      "grad_norm": 10.76842212677002,
+      "learning_rate": 1.083685341553593e-05,
+      "loss": 0.2561,
+      "step": 2249
+    },
+    {
+      "epoch": 0.8519500189322227,
+      "grad_norm": 24.11894989013672,
+      "learning_rate": 1.0782648366727965e-05,
+      "loss": 0.5513,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8523286633850814,
+      "grad_norm": 11.82591438293457,
+      "learning_rate": 1.072857149635914e-05,
+      "loss": 2.4543,
+      "step": 2251
+    },
+    {
+      "epoch": 0.8527073078379401,
+      "grad_norm": 9.538484573364258,
+      "learning_rate": 1.067462288212162e-05,
+      "loss": 1.1095,
+      "step": 2252
+    },
+    {
+      "epoch": 0.853085952290799,
+      "grad_norm": 11.511687278747559,
+      "learning_rate": 1.0620802601523316e-05,
+      "loss": 1.5651,
+      "step": 2253
+    },
+    {
+      "epoch": 0.8534645967436577,
+      "grad_norm": 11.207715034484863,
+      "learning_rate": 1.0567110731887742e-05,
+      "loss": 0.9555,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8538432411965164,
+      "grad_norm": 10.063400268554688,
+      "learning_rate": 1.0513547350353936e-05,
+      "loss": 1.0487,
+      "step": 2255
+    },
+    {
+      "epoch": 0.8542218856493753,
+      "grad_norm": 11.091261863708496,
+      "learning_rate": 1.0460112533876287e-05,
+      "loss": 1.0456,
+      "step": 2256
+    },
+    {
+      "epoch": 0.854600530102234,
+      "grad_norm": 11.866201400756836,
+      "learning_rate": 1.0406806359224574e-05,
+      "loss": 1.1276,
+      "step": 2257
+    },
+    {
+      "epoch": 0.8549791745550928,
+      "grad_norm": 12.267317771911621,
+      "learning_rate": 1.035362890298368e-05,
+      "loss": 0.8773,
+      "step": 2258
+    },
+    {
+      "epoch": 0.8553578190079515,
+      "grad_norm": 12.725680351257324,
+      "learning_rate": 1.030058024155357e-05,
+      "loss": 0.8035,
+      "step": 2259
+    },
+    {
+      "epoch": 0.8557364634608103,
+      "grad_norm": 11.86599063873291,
+      "learning_rate": 1.0247660451149166e-05,
+      "loss": 0.9379,
+      "step": 2260
+    },
+    {
+      "epoch": 0.8561151079136691,
+      "grad_norm": 13.769980430603027,
+      "learning_rate": 1.0194869607800305e-05,
+      "loss": 0.9552,
+      "step": 2261
+    },
+    {
+      "epoch": 0.8564937523665278,
+      "grad_norm": 10.183456420898438,
+      "learning_rate": 1.0142207787351465e-05,
+      "loss": 0.7875,
+      "step": 2262
+    },
+    {
+      "epoch": 0.8568723968193867,
+      "grad_norm": 13.380812644958496,
+      "learning_rate": 1.0089675065461834e-05,
+      "loss": 0.6164,
+      "step": 2263
+    },
+    {
+      "epoch": 0.8572510412722454,
+      "grad_norm": 9.049885749816895,
+      "learning_rate": 1.0037271517605063e-05,
+      "loss": 0.64,
+      "step": 2264
+    },
+    {
+      "epoch": 0.8576296857251041,
+      "grad_norm": 12.912697792053223,
+      "learning_rate": 9.984997219069304e-06,
+      "loss": 0.6789,
+      "step": 2265
+    },
+    {
+      "epoch": 0.8580083301779629,
+      "grad_norm": 9.801331520080566,
+      "learning_rate": 9.932852244956936e-06,
+      "loss": 0.4367,
+      "step": 2266
+    },
+    {
+      "epoch": 0.8583869746308217,
+      "grad_norm": 16.610326766967773,
+      "learning_rate": 9.880836670184567e-06,
+      "loss": 0.6217,
+      "step": 2267
+    },
+    {
+      "epoch": 0.8587656190836804,
+      "grad_norm": 10.286986351013184,
+      "learning_rate": 9.828950569482875e-06,
+      "loss": 0.4884,
+      "step": 2268
+    },
+    {
+      "epoch": 0.8591442635365392,
+      "grad_norm": 11.326078414916992,
+      "learning_rate": 9.777194017396595e-06,
+      "loss": 0.7175,
+      "step": 2269
+    },
+    {
+      "epoch": 0.859522907989398,
+      "grad_norm": 11.866266250610352,
+      "learning_rate": 9.72556708828427e-06,
+      "loss": 0.5466,
+      "step": 2270
+    },
+    {
+      "epoch": 0.8599015524422567,
+      "grad_norm": 13.79624080657959,
+      "learning_rate": 9.674069856318224e-06,
+      "loss": 0.5976,
+      "step": 2271
+    },
+    {
+      "epoch": 0.8602801968951155,
+      "grad_norm": 8.457572937011719,
+      "learning_rate": 9.622702395484451e-06,
+      "loss": 0.2856,
+      "step": 2272
+    },
+    {
+      "epoch": 0.8606588413479742,
+      "grad_norm": 17.572168350219727,
+      "learning_rate": 9.571464779582529e-06,
+      "loss": 0.2946,
+      "step": 2273
+    },
+    {
+      "epoch": 0.8610374858008331,
+      "grad_norm": 16.828079223632812,
+      "learning_rate": 9.52035708222545e-06,
+      "loss": 0.7045,
+      "step": 2274
+    },
+    {
+      "epoch": 0.8614161302536918,
+      "grad_norm": 19.846851348876953,
+      "learning_rate": 9.469379376839582e-06,
+      "loss": 0.1168,
+      "step": 2275
+    },
+    {
+      "epoch": 0.8617947747065505,
+      "grad_norm": 10.008796691894531,
+      "learning_rate": 9.418531736664483e-06,
+      "loss": 2.1273,
+      "step": 2276
+    },
+    {
+      "epoch": 0.8621734191594093,
+      "grad_norm": 10.289381980895996,
+      "learning_rate": 9.367814234752937e-06,
+      "loss": 1.7055,
+      "step": 2277
+    },
+    {
+      "epoch": 0.8625520636122681,
+      "grad_norm": 11.30695629119873,
+      "learning_rate": 9.31722694397067e-06,
+      "loss": 1.472,
+      "step": 2278
+    },
+    {
+      "epoch": 0.8629307080651268,
+      "grad_norm": 8.154560089111328,
+      "learning_rate": 9.266769936996389e-06,
+      "loss": 0.9404,
+      "step": 2279
+    },
+    {
+      "epoch": 0.8633093525179856,
+      "grad_norm": 10.423186302185059,
+      "learning_rate": 9.216443286321586e-06,
+      "loss": 1.3779,
+      "step": 2280
+    },
+    {
+      "epoch": 0.8636879969708444,
+      "grad_norm": 11.121103286743164,
+      "learning_rate": 9.166247064250477e-06,
+      "loss": 0.5472,
+      "step": 2281
+    },
+    {
+      "epoch": 0.8640666414237032,
+      "grad_norm": 9.69433307647705,
+      "learning_rate": 9.116181342899932e-06,
+      "loss": 0.7709,
+      "step": 2282
+    },
+    {
+      "epoch": 0.8644452858765619,
+      "grad_norm": 9.098278045654297,
+      "learning_rate": 9.06624619419928e-06,
+      "loss": 0.4906,
+      "step": 2283
+    },
+    {
+      "epoch": 0.8648239303294206,
+      "grad_norm": 14.784154891967773,
+      "learning_rate": 9.016441689890286e-06,
+      "loss": 0.8121,
+      "step": 2284
+    },
+    {
+      "epoch": 0.8652025747822795,
+      "grad_norm": 9.00545597076416,
+      "learning_rate": 8.966767901527007e-06,
+      "loss": 0.5983,
+      "step": 2285
+    },
+    {
+      "epoch": 0.8655812192351382,
+      "grad_norm": 10.185830116271973,
+      "learning_rate": 8.917224900475695e-06,
+      "loss": 0.8618,
+      "step": 2286
+    },
+    {
+      "epoch": 0.8659598636879969,
+      "grad_norm": 13.381327629089355,
+      "learning_rate": 8.867812757914694e-06,
+      "loss": 1.0504,
+      "step": 2287
+    },
+    {
+      "epoch": 0.8663385081408558,
+      "grad_norm": 14.24845027923584,
+      "learning_rate": 8.818531544834385e-06,
+      "loss": 0.7165,
+      "step": 2288
+    },
+    {
+      "epoch": 0.8667171525937145,
+      "grad_norm": 12.008301734924316,
+      "learning_rate": 8.76938133203702e-06,
+      "loss": 0.6329,
+      "step": 2289
+    },
+    {
+      "epoch": 0.8670957970465732,
+      "grad_norm": 8.582221031188965,
+      "learning_rate": 8.720362190136611e-06,
+      "loss": 0.5233,
+      "step": 2290
+    },
+    {
+      "epoch": 0.867474441499432,
+      "grad_norm": 14.074053764343262,
+      "learning_rate": 8.671474189558903e-06,
+      "loss": 0.8896,
+      "step": 2291
+    },
+    {
+      "epoch": 0.8678530859522908,
+      "grad_norm": 8.265854835510254,
+      "learning_rate": 8.622717400541192e-06,
+      "loss": 0.423,
+      "step": 2292
+    },
+    {
+      "epoch": 0.8682317304051496,
+      "grad_norm": 14.022979736328125,
+      "learning_rate": 8.57409189313233e-06,
+      "loss": 0.5092,
+      "step": 2293
+    },
+    {
+      "epoch": 0.8686103748580083,
+      "grad_norm": 7.6086106300354,
+      "learning_rate": 8.525597737192481e-06,
+      "loss": 0.3162,
+      "step": 2294
+    },
+    {
+      "epoch": 0.8689890193108671,
+      "grad_norm": 10.01663875579834,
+      "learning_rate": 8.477235002393147e-06,
+      "loss": 0.363,
+      "step": 2295
+    },
+    {
+      "epoch": 0.8693676637637259,
+      "grad_norm": 10.2891206741333,
+      "learning_rate": 8.429003758216959e-06,
+      "loss": 0.4843,
+      "step": 2296
+    },
+    {
+      "epoch": 0.8697463082165846,
+      "grad_norm": 10.89184284210205,
+      "learning_rate": 8.380904073957729e-06,
+      "loss": 0.6032,
+      "step": 2297
+    },
+    {
+      "epoch": 0.8701249526694433,
+      "grad_norm": 11.962915420532227,
+      "learning_rate": 8.332936018720171e-06,
+      "loss": 0.4728,
+      "step": 2298
+    },
+    {
+      "epoch": 0.8705035971223022,
+      "grad_norm": 11.58560562133789,
+      "learning_rate": 8.285099661419926e-06,
+      "loss": 0.4367,
+      "step": 2299
+    },
+    {
+      "epoch": 0.8708822415751609,
+      "grad_norm": 18.291141510009766,
+      "learning_rate": 8.237395070783404e-06,
+      "loss": 0.651,
+      "step": 2300
+    },
+    {
+      "epoch": 0.8712608860280197,
+      "grad_norm": 11.406949996948242,
+      "learning_rate": 8.189822315347762e-06,
+      "loss": 2.1475,
+      "step": 2301
+    },
+    {
+      "epoch": 0.8716395304808785,
+      "grad_norm": 10.053242683410645,
+      "learning_rate": 8.14238146346068e-06,
+      "loss": 1.4275,
+      "step": 2302
+    },
+    {
+      "epoch": 0.8720181749337372,
+      "grad_norm": 9.507181167602539,
+      "learning_rate": 8.09507258328036e-06,
+      "loss": 1.1072,
+      "step": 2303
+    },
+    {
+      "epoch": 0.872396819386596,
+      "grad_norm": 10.980565071105957,
+      "learning_rate": 8.04789574277538e-06,
+      "loss": 1.0862,
+      "step": 2304
+    },
+    {
+      "epoch": 0.8727754638394547,
+      "grad_norm": 10.014975547790527,
+      "learning_rate": 8.000851009724696e-06,
+      "loss": 0.8774,
+      "step": 2305
+    },
+    {
+      "epoch": 0.8731541082923135,
+      "grad_norm": 10.634843826293945,
+      "learning_rate": 7.95393845171737e-06,
+      "loss": 0.7953,
+      "step": 2306
+    },
+    {
+      "epoch": 0.8735327527451723,
+      "grad_norm": 12.149109840393066,
+      "learning_rate": 7.907158136152604e-06,
+      "loss": 0.8086,
+      "step": 2307
+    },
+    {
+      "epoch": 0.873911397198031,
+      "grad_norm": 9.334737777709961,
+      "learning_rate": 7.860510130239607e-06,
+      "loss": 0.5934,
+      "step": 2308
+    },
+    {
+      "epoch": 0.8742900416508899,
+      "grad_norm": 13.120880126953125,
+      "learning_rate": 7.813994500997524e-06,
+      "loss": 1.0779,
+      "step": 2309
+    },
+    {
+      "epoch": 0.8746686861037486,
+      "grad_norm": 10.691913604736328,
+      "learning_rate": 7.767611315255275e-06,
+      "loss": 0.9073,
+      "step": 2310
+    },
+    {
+      "epoch": 0.8750473305566073,
+      "grad_norm": 10.472341537475586,
+      "learning_rate": 7.72136063965152e-06,
+      "loss": 0.4952,
+      "step": 2311
+    },
+    {
+      "epoch": 0.8754259750094661,
+      "grad_norm": 10.533929824829102,
+      "learning_rate": 7.67524254063452e-06,
+      "loss": 0.7114,
+      "step": 2312
+    },
+    {
+      "epoch": 0.8758046194623249,
+      "grad_norm": 6.5337042808532715,
+      "learning_rate": 7.6292570844621045e-06,
+      "loss": 0.3483,
+      "step": 2313
+    },
+    {
+      "epoch": 0.8761832639151836,
+      "grad_norm": 8.839234352111816,
+      "learning_rate": 7.583404337201516e-06,
+      "loss": 0.5599,
+      "step": 2314
+    },
+    {
+      "epoch": 0.8765619083680424,
+      "grad_norm": 10.337873458862305,
+      "learning_rate": 7.5376843647293024e-06,
+      "loss": 0.5017,
+      "step": 2315
+    },
+    {
+      "epoch": 0.8769405528209012,
+      "grad_norm": 8.032730102539062,
+      "learning_rate": 7.4920972327312875e-06,
+      "loss": 0.4731,
+      "step": 2316
+    },
+    {
+      "epoch": 0.87731919727376,
+      "grad_norm": 8.474766731262207,
+      "learning_rate": 7.446643006702469e-06,
+      "loss": 0.3606,
+      "step": 2317
+    },
+    {
+      "epoch": 0.8776978417266187,
+      "grad_norm": 10.331032752990723,
+      "learning_rate": 7.4013217519468325e-06,
+      "loss": 0.21,
+      "step": 2318
+    },
+    {
+      "epoch": 0.8780764861794774,
+      "grad_norm": 15.5275297164917,
+      "learning_rate": 7.356133533577369e-06,
+      "loss": 0.4189,
+      "step": 2319
+    },
+    {
+      "epoch": 0.8784551306323363,
+      "grad_norm": 8.125926971435547,
+      "learning_rate": 7.311078416515926e-06,
+      "loss": 0.1247,
+      "step": 2320
+    },
+    {
+      "epoch": 0.878833775085195,
+      "grad_norm": 10.590972900390625,
+      "learning_rate": 7.266156465493124e-06,
+      "loss": 0.3929,
+      "step": 2321
+    },
+    {
+      "epoch": 0.8792124195380537,
+      "grad_norm": 6.178790092468262,
+      "learning_rate": 7.221367745048279e-06,
+      "loss": 0.2636,
+      "step": 2322
+    },
+    {
+      "epoch": 0.8795910639909126,
+      "grad_norm": 16.32522201538086,
+      "learning_rate": 7.1767123195292666e-06,
+      "loss": 0.4226,
+      "step": 2323
+    },
+    {
+      "epoch": 0.8799697084437713,
+      "grad_norm": 7.7940850257873535,
+      "learning_rate": 7.132190253092452e-06,
+      "loss": 0.1687,
+      "step": 2324
+    },
+    {
+      "epoch": 0.88034835289663,
+      "grad_norm": 10.466033935546875,
+      "learning_rate": 7.08780160970266e-06,
+      "loss": 0.4764,
+      "step": 2325
+    },
+    {
+      "epoch": 0.8807269973494888,
+      "grad_norm": 12.871282577514648,
+      "learning_rate": 7.043546453132977e-06,
+      "loss": 2.4303,
+      "step": 2326
+    },
+    {
+      "epoch": 0.8811056418023476,
+      "grad_norm": 11.891122817993164,
+      "learning_rate": 6.99942484696472e-06,
+      "loss": 1.6218,
+      "step": 2327
+    },
+    {
+      "epoch": 0.8814842862552064,
+      "grad_norm": 8.916067123413086,
+      "learning_rate": 6.955436854587327e-06,
+      "loss": 0.8128,
+      "step": 2328
+    },
+    {
+      "epoch": 0.8818629307080651,
+      "grad_norm": 8.550481796264648,
+      "learning_rate": 6.9115825391982806e-06,
+      "loss": 0.8219,
+      "step": 2329
+    },
+    {
+      "epoch": 0.8822415751609239,
+      "grad_norm": 10.76794719696045,
+      "learning_rate": 6.867861963803035e-06,
+      "loss": 0.7949,
+      "step": 2330
+    },
+    {
+      "epoch": 0.8826202196137827,
+      "grad_norm": 8.747878074645996,
+      "learning_rate": 6.824275191214868e-06,
+      "loss": 0.7019,
+      "step": 2331
+    },
+    {
+      "epoch": 0.8829988640666414,
+      "grad_norm": 11.711833953857422,
+      "learning_rate": 6.780822284054833e-06,
+      "loss": 0.7592,
+      "step": 2332
+    },
+    {
+      "epoch": 0.8833775085195001,
+      "grad_norm": 13.456823348999023,
+      "learning_rate": 6.7375033047516464e-06,
+      "loss": 0.9945,
+      "step": 2333
+    },
+    {
+      "epoch": 0.883756152972359,
+      "grad_norm": 8.14212703704834,
+      "learning_rate": 6.694318315541637e-06,
+      "loss": 0.6043,
+      "step": 2334
+    },
+    {
+      "epoch": 0.8841347974252177,
+      "grad_norm": 8.493631362915039,
+      "learning_rate": 6.651267378468584e-06,
+      "loss": 0.5508,
+      "step": 2335
+    },
+    {
+      "epoch": 0.8845134418780765,
+      "grad_norm": 10.33297061920166,
+      "learning_rate": 6.608350555383758e-06,
+      "loss": 0.6154,
+      "step": 2336
+    },
+    {
+      "epoch": 0.8848920863309353,
+      "grad_norm": 10.296257019042969,
+      "learning_rate": 6.565567907945658e-06,
+      "loss": 0.6082,
+      "step": 2337
+    },
+    {
+      "epoch": 0.885270730783794,
+      "grad_norm": 9.22463321685791,
+      "learning_rate": 6.522919497620073e-06,
+      "loss": 0.6438,
+      "step": 2338
+    },
+    {
+      "epoch": 0.8856493752366528,
+      "grad_norm": 11.198843002319336,
+      "learning_rate": 6.480405385679888e-06,
+      "loss": 0.7403,
+      "step": 2339
+    },
+    {
+      "epoch": 0.8860280196895115,
+      "grad_norm": 10.24124526977539,
+      "learning_rate": 6.43802563320508e-06,
+      "loss": 0.565,
+      "step": 2340
+    },
+    {
+      "epoch": 0.8864066641423703,
+      "grad_norm": 11.018882751464844,
+      "learning_rate": 6.395780301082577e-06,
+      "loss": 0.5413,
+      "step": 2341
+    },
+    {
+      "epoch": 0.8867853085952291,
+      "grad_norm": 10.295462608337402,
+      "learning_rate": 6.353669450006194e-06,
+      "loss": 0.4042,
+      "step": 2342
+    },
+    {
+      "epoch": 0.8871639530480878,
+      "grad_norm": 13.353784561157227,
+      "learning_rate": 6.3116931404765265e-06,
+      "loss": 0.5982,
+      "step": 2343
+    },
+    {
+      "epoch": 0.8875425975009467,
+      "grad_norm": 11.05883502960205,
+      "learning_rate": 6.269851432800855e-06,
+      "loss": 0.5592,
+      "step": 2344
+    },
+    {
+      "epoch": 0.8879212419538054,
+      "grad_norm": 14.049494743347168,
+      "learning_rate": 6.228144387093127e-06,
+      "loss": 0.4035,
+      "step": 2345
+    },
+    {
+      "epoch": 0.8882998864066641,
+      "grad_norm": 12.806116104125977,
+      "learning_rate": 6.1865720632737875e-06,
+      "loss": 0.514,
+      "step": 2346
+    },
+    {
+      "epoch": 0.8886785308595229,
+      "grad_norm": 9.267263412475586,
+      "learning_rate": 6.145134521069729e-06,
+      "loss": 0.5728,
+      "step": 2347
+    },
+    {
+      "epoch": 0.8890571753123817,
+      "grad_norm": 9.22109603881836,
+      "learning_rate": 6.103831820014194e-06,
+      "loss": 0.288,
+      "step": 2348
+    },
+    {
+      "epoch": 0.8894358197652404,
+      "grad_norm": 11.868034362792969,
+      "learning_rate": 6.062664019446751e-06,
+      "loss": 0.2524,
+      "step": 2349
+    },
+    {
+      "epoch": 0.8898144642180992,
+      "grad_norm": 9.9324951171875,
+      "learning_rate": 6.021631178513087e-06,
+      "loss": 0.4626,
+      "step": 2350
+    },
+    {
+      "epoch": 0.890193108670958,
+      "grad_norm": 11.608717918395996,
+      "learning_rate": 5.9807333561650355e-06,
+      "loss": 1.7653,
+      "step": 2351
+    },
+    {
+      "epoch": 0.8905717531238168,
+      "grad_norm": 10.697403907775879,
+      "learning_rate": 5.939970611160428e-06,
+      "loss": 1.5739,
+      "step": 2352
+    },
+    {
+      "epoch": 0.8909503975766755,
+      "grad_norm": 10.519599914550781,
+      "learning_rate": 5.899343002063063e-06,
+      "loss": 1.4995,
+      "step": 2353
+    },
+    {
+      "epoch": 0.8913290420295342,
+      "grad_norm": 12.994054794311523,
+      "learning_rate": 5.858850587242559e-06,
+      "loss": 1.0682,
+      "step": 2354
+    },
+    {
+      "epoch": 0.8917076864823931,
+      "grad_norm": 12.894742012023926,
+      "learning_rate": 5.818493424874294e-06,
+      "loss": 1.2836,
+      "step": 2355
+    },
+    {
+      "epoch": 0.8920863309352518,
+      "grad_norm": 12.090742111206055,
+      "learning_rate": 5.778271572939354e-06,
+      "loss": 1.4098,
+      "step": 2356
+    },
+    {
+      "epoch": 0.8924649753881105,
+      "grad_norm": 8.5830717086792,
+      "learning_rate": 5.738185089224424e-06,
+      "loss": 0.6956,
+      "step": 2357
+    },
+    {
+      "epoch": 0.8928436198409694,
+      "grad_norm": 9.2327880859375,
+      "learning_rate": 5.698234031321692e-06,
+      "loss": 0.4974,
+      "step": 2358
+    },
+    {
+      "epoch": 0.8932222642938281,
+      "grad_norm": 12.713973999023438,
+      "learning_rate": 5.658418456628778e-06,
+      "loss": 0.6973,
+      "step": 2359
+    },
+    {
+      "epoch": 0.8936009087466869,
+      "grad_norm": 8.313766479492188,
+      "learning_rate": 5.618738422348646e-06,
+      "loss": 0.4688,
+      "step": 2360
+    },
+    {
+      "epoch": 0.8939795531995456,
+      "grad_norm": 11.878710746765137,
+      "learning_rate": 5.579193985489584e-06,
+      "loss": 0.6091,
+      "step": 2361
+    },
+    {
+      "epoch": 0.8943581976524044,
+      "grad_norm": 9.968457221984863,
+      "learning_rate": 5.5397852028649996e-06,
+      "loss": 0.7507,
+      "step": 2362
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 9.140604972839355,
+      "learning_rate": 5.500512131093438e-06,
+      "loss": 0.4721,
+      "step": 2363
+    },
+    {
+      "epoch": 0.8951154865581219,
+      "grad_norm": 9.671671867370605,
+      "learning_rate": 5.461374826598453e-06,
+      "loss": 0.5719,
+      "step": 2364
+    },
+    {
+      "epoch": 0.8954941310109807,
+      "grad_norm": 14.937457084655762,
+      "learning_rate": 5.422373345608578e-06,
+      "loss": 0.5815,
+      "step": 2365
+    },
+    {
+      "epoch": 0.8958727754638395,
+      "grad_norm": 12.879511833190918,
+      "learning_rate": 5.383507744157179e-06,
+      "loss": 0.5853,
+      "step": 2366
+    },
+    {
+      "epoch": 0.8962514199166982,
+      "grad_norm": 10.898216247558594,
+      "learning_rate": 5.344778078082391e-06,
+      "loss": 0.4625,
+      "step": 2367
+    },
+    {
+      "epoch": 0.896630064369557,
+      "grad_norm": 14.941044807434082,
+      "learning_rate": 5.306184403027059e-06,
+      "loss": 0.6827,
+      "step": 2368
+    },
+    {
+      "epoch": 0.8970087088224158,
+      "grad_norm": 14.840867042541504,
+      "learning_rate": 5.267726774438697e-06,
+      "loss": 0.5103,
+      "step": 2369
+    },
+    {
+      "epoch": 0.8973873532752745,
+      "grad_norm": 4.7114577293396,
+      "learning_rate": 5.229405247569308e-06,
+      "loss": 0.1327,
+      "step": 2370
+    },
+    {
+      "epoch": 0.8977659977281333,
+      "grad_norm": 9.676827430725098,
+      "learning_rate": 5.191219877475373e-06,
+      "loss": 0.4447,
+      "step": 2371
+    },
+    {
+      "epoch": 0.8981446421809921,
+      "grad_norm": 7.863070964813232,
+      "learning_rate": 5.153170719017741e-06,
+      "loss": 0.2609,
+      "step": 2372
+    },
+    {
+      "epoch": 0.8985232866338508,
+      "grad_norm": 8.543060302734375,
+      "learning_rate": 5.115257826861619e-06,
+      "loss": 0.2568,
+      "step": 2373
+    },
+    {
+      "epoch": 0.8989019310867096,
+      "grad_norm": 11.265084266662598,
+      "learning_rate": 5.077481255476368e-06,
+      "loss": 0.2466,
+      "step": 2374
+    },
+    {
+      "epoch": 0.8992805755395683,
+      "grad_norm": 20.728683471679688,
+      "learning_rate": 5.039841059135553e-06,
+      "loss": 1.1415,
+      "step": 2375
+    },
+    {
+      "epoch": 0.8996592199924272,
+      "grad_norm": 10.630395889282227,
+      "learning_rate": 5.002337291916792e-06,
+      "loss": 1.858,
+      "step": 2376
+    },
+    {
+      "epoch": 0.9000378644452859,
+      "grad_norm": 10.02651596069336,
+      "learning_rate": 4.9649700077016635e-06,
+      "loss": 1.4943,
+      "step": 2377
+    },
+    {
+      "epoch": 0.9004165088981446,
+      "grad_norm": 11.16691780090332,
+      "learning_rate": 4.927739260175735e-06,
+      "loss": 1.3521,
+      "step": 2378
+    },
+    {
+      "epoch": 0.9007951533510034,
+      "grad_norm": 10.301544189453125,
+      "learning_rate": 4.8906451028283285e-06,
+      "loss": 0.8017,
+      "step": 2379
+    },
+    {
+      "epoch": 0.9011737978038622,
+      "grad_norm": 10.982889175415039,
+      "learning_rate": 4.853687588952594e-06,
+      "loss": 0.7324,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9015524422567209,
+      "grad_norm": 10.89743423461914,
+      "learning_rate": 4.816866771645323e-06,
+      "loss": 0.9049,
+      "step": 2381
+    },
+    {
+      "epoch": 0.9019310867095797,
+      "grad_norm": 10.184319496154785,
+      "learning_rate": 4.7801827038069234e-06,
+      "loss": 0.7181,
+      "step": 2382
+    },
+    {
+      "epoch": 0.9023097311624385,
+      "grad_norm": 7.678170680999756,
+      "learning_rate": 4.7436354381413476e-06,
+      "loss": 0.4933,
+      "step": 2383
+    },
+    {
+      "epoch": 0.9026883756152972,
+      "grad_norm": 8.026887893676758,
+      "learning_rate": 4.707225027156015e-06,
+      "loss": 0.5746,
+      "step": 2384
+    },
+    {
+      "epoch": 0.903067020068156,
+      "grad_norm": 11.845023155212402,
+      "learning_rate": 4.670951523161693e-06,
+      "loss": 0.537,
+      "step": 2385
+    },
+    {
+      "epoch": 0.9034456645210147,
+      "grad_norm": 12.805367469787598,
+      "learning_rate": 4.634814978272473e-06,
+      "loss": 0.6111,
+      "step": 2386
+    },
+    {
+      "epoch": 0.9038243089738736,
+      "grad_norm": 10.827343940734863,
+      "learning_rate": 4.598815444405691e-06,
+      "loss": 0.7474,
+      "step": 2387
+    },
+    {
+      "epoch": 0.9042029534267323,
+      "grad_norm": 12.786199569702148,
+      "learning_rate": 4.5629529732817864e-06,
+      "loss": 0.8605,
+      "step": 2388
+    },
+    {
+      "epoch": 0.904581597879591,
+      "grad_norm": 7.475493907928467,
+      "learning_rate": 4.527227616424368e-06,
+      "loss": 0.3846,
+      "step": 2389
+    },
+    {
+      "epoch": 0.9049602423324499,
+      "grad_norm": 16.920345306396484,
+      "learning_rate": 4.491639425159988e-06,
+      "loss": 0.8434,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9053388867853086,
+      "grad_norm": 13.1689453125,
+      "learning_rate": 4.4561884506181266e-06,
+      "loss": 0.5591,
+      "step": 2391
+    },
+    {
+      "epoch": 0.9057175312381673,
+      "grad_norm": 13.537805557250977,
+      "learning_rate": 4.420874743731163e-06,
+      "loss": 0.6376,
+      "step": 2392
+    },
+    {
+      "epoch": 0.9060961756910261,
+      "grad_norm": 12.794946670532227,
+      "learning_rate": 4.385698355234258e-06,
+      "loss": 0.6448,
+      "step": 2393
+    },
+    {
+      "epoch": 0.9064748201438849,
+      "grad_norm": 11.212878227233887,
+      "learning_rate": 4.350659335665275e-06,
+      "loss": 0.2639,
+      "step": 2394
+    },
+    {
+      "epoch": 0.9068534645967437,
+      "grad_norm": 7.262661933898926,
+      "learning_rate": 4.315757735364712e-06,
+      "loss": 0.3209,
+      "step": 2395
+    },
+    {
+      "epoch": 0.9072321090496024,
+      "grad_norm": 8.17447566986084,
+      "learning_rate": 4.280993604475636e-06,
+      "loss": 0.3456,
+      "step": 2396
+    },
+    {
+      "epoch": 0.9076107535024612,
+      "grad_norm": 7.121233940124512,
+      "learning_rate": 4.246366992943662e-06,
+      "loss": 0.2374,
+      "step": 2397
+    },
+    {
+      "epoch": 0.90798939795532,
+      "grad_norm": 14.229004859924316,
+      "learning_rate": 4.211877950516763e-06,
+      "loss": 0.2939,
+      "step": 2398
+    },
+    {
+      "epoch": 0.9083680424081787,
+      "grad_norm": 55.41293716430664,
+      "learning_rate": 4.177526526745301e-06,
+      "loss": 0.5769,
+      "step": 2399
+    },
+    {
+      "epoch": 0.9087466868610374,
+      "grad_norm": 10.350135803222656,
+      "learning_rate": 4.143312770981911e-06,
+      "loss": 0.3735,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9091253313138963,
+      "grad_norm": 9.91360092163086,
+      "learning_rate": 4.109236732381461e-06,
+      "loss": 1.6225,
+      "step": 2401
+    },
+    {
+      "epoch": 0.909503975766755,
+      "grad_norm": 8.888208389282227,
+      "learning_rate": 4.075298459900933e-06,
+      "loss": 1.2812,
+      "step": 2402
+    },
+    {
+      "epoch": 0.9098826202196137,
+      "grad_norm": 14.481423377990723,
+      "learning_rate": 4.0414980022994045e-06,
+      "loss": 1.5647,
+      "step": 2403
+    },
+    {
+      "epoch": 0.9102612646724726,
+      "grad_norm": 9.676767349243164,
+      "learning_rate": 4.007835408137928e-06,
+      "loss": 0.696,
+      "step": 2404
+    },
+    {
+      "epoch": 0.9106399091253313,
+      "grad_norm": 8.741473197937012,
+      "learning_rate": 3.974310725779518e-06,
+      "loss": 0.7849,
+      "step": 2405
+    },
+    {
+      "epoch": 0.9110185535781901,
+      "grad_norm": 14.236115455627441,
+      "learning_rate": 3.940924003389046e-06,
+      "loss": 0.877,
+      "step": 2406
+    },
+    {
+      "epoch": 0.9113971980310488,
+      "grad_norm": 10.6664400100708,
+      "learning_rate": 3.907675288933144e-06,
+      "loss": 0.6007,
+      "step": 2407
+    },
+    {
+      "epoch": 0.9117758424839076,
+      "grad_norm": 9.50271987915039,
+      "learning_rate": 3.874564630180188e-06,
+      "loss": 0.7913,
+      "step": 2408
+    },
+    {
+      "epoch": 0.9121544869367664,
+      "grad_norm": 11.143498420715332,
+      "learning_rate": 3.84159207470024e-06,
+      "loss": 0.7398,
+      "step": 2409
+    },
+    {
+      "epoch": 0.9125331313896251,
+      "grad_norm": 9.227463722229004,
+      "learning_rate": 3.808757669864904e-06,
+      "loss": 0.7372,
+      "step": 2410
+    },
+    {
+      "epoch": 0.912911775842484,
+      "grad_norm": 12.129105567932129,
+      "learning_rate": 3.7760614628473357e-06,
+      "loss": 0.8929,
+      "step": 2411
+    },
+    {
+      "epoch": 0.9132904202953427,
+      "grad_norm": 10.593716621398926,
+      "learning_rate": 3.743503500622103e-06,
+      "loss": 0.7059,
+      "step": 2412
+    },
+    {
+      "epoch": 0.9136690647482014,
+      "grad_norm": 13.297348976135254,
+      "learning_rate": 3.711083829965212e-06,
+      "loss": 0.5883,
+      "step": 2413
+    },
+    {
+      "epoch": 0.9140477092010602,
+      "grad_norm": 8.455639839172363,
+      "learning_rate": 3.678802497453948e-06,
+      "loss": 0.4366,
+      "step": 2414
+    },
+    {
+      "epoch": 0.914426353653919,
+      "grad_norm": 8.234000205993652,
+      "learning_rate": 3.6466595494668353e-06,
+      "loss": 0.363,
+      "step": 2415
+    },
+    {
+      "epoch": 0.9148049981067777,
+      "grad_norm": 13.38591194152832,
+      "learning_rate": 3.6146550321836116e-06,
+      "loss": 0.5627,
+      "step": 2416
+    },
+    {
+      "epoch": 0.9151836425596365,
+      "grad_norm": 12.83345890045166,
+      "learning_rate": 3.58278899158514e-06,
+      "loss": 0.6422,
+      "step": 2417
+    },
+    {
+      "epoch": 0.9155622870124953,
+      "grad_norm": 13.449295997619629,
+      "learning_rate": 3.5510614734532876e-06,
+      "loss": 0.6531,
+      "step": 2418
+    },
+    {
+      "epoch": 0.915940931465354,
+      "grad_norm": 10.407699584960938,
+      "learning_rate": 3.519472523370948e-06,
+      "loss": 0.6168,
+      "step": 2419
+    },
+    {
+      "epoch": 0.9163195759182128,
+      "grad_norm": 17.99662208557129,
+      "learning_rate": 3.4880221867219064e-06,
+      "loss": 0.257,
+      "step": 2420
+    },
+    {
+      "epoch": 0.9166982203710715,
+      "grad_norm": 7.612576484680176,
+      "learning_rate": 3.45671050869083e-06,
+      "loss": 0.2634,
+      "step": 2421
+    },
+    {
+      "epoch": 0.9170768648239304,
+      "grad_norm": 6.6135358810424805,
+      "learning_rate": 3.425537534263168e-06,
+      "loss": 0.1463,
+      "step": 2422
+    },
+    {
+      "epoch": 0.9174555092767891,
+      "grad_norm": 6.440648078918457,
+      "learning_rate": 3.394503308225061e-06,
+      "loss": 0.1608,
+      "step": 2423
+    },
+    {
+      "epoch": 0.9178341537296478,
+      "grad_norm": 1.7489157915115356,
+      "learning_rate": 3.363607875163366e-06,
+      "loss": 0.047,
+      "step": 2424
+    },
+    {
+      "epoch": 0.9182127981825067,
+      "grad_norm": 30.02290153503418,
+      "learning_rate": 3.3328512794654652e-06,
+      "loss": 0.9788,
+      "step": 2425
+    },
+    {
+      "epoch": 0.9185914426353654,
+      "grad_norm": 11.564863204956055,
+      "learning_rate": 3.302233565319357e-06,
+      "loss": 2.0172,
+      "step": 2426
+    },
+    {
+      "epoch": 0.9189700870882241,
+      "grad_norm": 9.934062957763672,
+      "learning_rate": 3.2717547767134538e-06,
+      "loss": 1.326,
+      "step": 2427
+    },
+    {
+      "epoch": 0.9193487315410829,
+      "grad_norm": 10.045295715332031,
+      "learning_rate": 3.2414149574365836e-06,
+      "loss": 0.9745,
+      "step": 2428
+    },
+    {
+      "epoch": 0.9197273759939417,
+      "grad_norm": 11.001640319824219,
+      "learning_rate": 3.2112141510779127e-06,
+      "loss": 1.2666,
+      "step": 2429
+    },
+    {
+      "epoch": 0.9201060204468005,
+      "grad_norm": 10.749781608581543,
+      "learning_rate": 3.18115240102691e-06,
+      "loss": 1.0313,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9204846648996592,
+      "grad_norm": 11.977824211120605,
+      "learning_rate": 3.151229750473239e-06,
+      "loss": 0.8656,
+      "step": 2431
+    },
+    {
+      "epoch": 0.920863309352518,
+      "grad_norm": 11.430237770080566,
+      "learning_rate": 3.1214462424067335e-06,
+      "loss": 1.0509,
+      "step": 2432
+    },
+    {
+      "epoch": 0.9212419538053768,
+      "grad_norm": 10.933598518371582,
+      "learning_rate": 3.0918019196173096e-06,
+      "loss": 0.9562,
+      "step": 2433
+    },
+    {
+      "epoch": 0.9216205982582355,
+      "grad_norm": 7.044180393218994,
+      "learning_rate": 3.0622968246949213e-06,
+      "loss": 0.5424,
+      "step": 2434
+    },
+    {
+      "epoch": 0.9219992427110942,
+      "grad_norm": 14.46313762664795,
+      "learning_rate": 3.0329310000295153e-06,
+      "loss": 0.4784,
+      "step": 2435
+    },
+    {
+      "epoch": 0.9223778871639531,
+      "grad_norm": 9.44637393951416,
+      "learning_rate": 3.003704487810899e-06,
+      "loss": 0.649,
+      "step": 2436
+    },
+    {
+      "epoch": 0.9227565316168118,
+      "grad_norm": 8.902215003967285,
+      "learning_rate": 2.9746173300287837e-06,
+      "loss": 0.523,
+      "step": 2437
+    },
+    {
+      "epoch": 0.9231351760696705,
+      "grad_norm": 8.846084594726562,
+      "learning_rate": 2.945669568472631e-06,
+      "loss": 0.5283,
+      "step": 2438
+    },
+    {
+      "epoch": 0.9235138205225294,
+      "grad_norm": 9.392375946044922,
+      "learning_rate": 2.916861244731661e-06,
+      "loss": 0.5592,
+      "step": 2439
+    },
+    {
+      "epoch": 0.9238924649753881,
+      "grad_norm": 8.410120964050293,
+      "learning_rate": 2.888192400194745e-06,
+      "loss": 0.4376,
+      "step": 2440
+    },
+    {
+      "epoch": 0.9242711094282469,
+      "grad_norm": 9.047958374023438,
+      "learning_rate": 2.8596630760503673e-06,
+      "loss": 0.4578,
+      "step": 2441
+    },
+    {
+      "epoch": 0.9246497538811056,
+      "grad_norm": 9.126653671264648,
+      "learning_rate": 2.8312733132865754e-06,
+      "loss": 0.423,
+      "step": 2442
+    },
+    {
+      "epoch": 0.9250283983339644,
+      "grad_norm": 6.317657470703125,
+      "learning_rate": 2.803023152690887e-06,
+      "loss": 0.2639,
+      "step": 2443
+    },
+    {
+      "epoch": 0.9254070427868232,
+      "grad_norm": 10.31212043762207,
+      "learning_rate": 2.7749126348502684e-06,
+      "loss": 0.3573,
+      "step": 2444
+    },
+    {
+      "epoch": 0.9257856872396819,
+      "grad_norm": 15.265264511108398,
+      "learning_rate": 2.7469418001510704e-06,
+      "loss": 0.4853,
+      "step": 2445
+    },
+    {
+      "epoch": 0.9261643316925408,
+      "grad_norm": 7.364819526672363,
+      "learning_rate": 2.7191106887789473e-06,
+      "loss": 0.3669,
+      "step": 2446
+    },
+    {
+      "epoch": 0.9265429761453995,
+      "grad_norm": 9.149045944213867,
+      "learning_rate": 2.6914193407188146e-06,
+      "loss": 0.1694,
+      "step": 2447
+    },
+    {
+      "epoch": 0.9269216205982582,
+      "grad_norm": 6.317681789398193,
+      "learning_rate": 2.663867795754771e-06,
+      "loss": 0.34,
+      "step": 2448
+    },
+    {
+      "epoch": 0.927300265051117,
+      "grad_norm": 6.21943998336792,
+      "learning_rate": 2.636456093470119e-06,
+      "loss": 0.272,
+      "step": 2449
+    },
+    {
+      "epoch": 0.9276789095039758,
+      "grad_norm": 19.55099868774414,
+      "learning_rate": 2.6091842732472006e-06,
+      "loss": 0.6493,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9280575539568345,
+      "grad_norm": 9.162104606628418,
+      "learning_rate": 2.582052374267385e-06,
+      "loss": 1.5711,
+      "step": 2451
+    },
+    {
+      "epoch": 0.9284361984096933,
+      "grad_norm": 9.272793769836426,
+      "learning_rate": 2.555060435511025e-06,
+      "loss": 1.2648,
+      "step": 2452
+    },
+    {
+      "epoch": 0.9288148428625521,
+      "grad_norm": 12.99858283996582,
+      "learning_rate": 2.5282084957574226e-06,
+      "loss": 1.4216,
+      "step": 2453
+    },
+    {
+      "epoch": 0.9291934873154108,
+      "grad_norm": 11.940863609313965,
+      "learning_rate": 2.5014965935847178e-06,
+      "loss": 1.0989,
+      "step": 2454
+    },
+    {
+      "epoch": 0.9295721317682696,
+      "grad_norm": 10.320348739624023,
+      "learning_rate": 2.4749247673698573e-06,
+      "loss": 0.743,
+      "step": 2455
+    },
+    {
+      "epoch": 0.9299507762211283,
+      "grad_norm": 11.852913856506348,
+      "learning_rate": 2.4484930552885365e-06,
+      "loss": 0.9047,
+      "step": 2456
+    },
+    {
+      "epoch": 0.9303294206739872,
+      "grad_norm": 8.977890014648438,
+      "learning_rate": 2.4222014953151686e-06,
+      "loss": 0.7247,
+      "step": 2457
+    },
+    {
+      "epoch": 0.9307080651268459,
+      "grad_norm": 10.382013320922852,
+      "learning_rate": 2.396050125222793e-06,
+      "loss": 0.7898,
+      "step": 2458
+    },
+    {
+      "epoch": 0.9310867095797046,
+      "grad_norm": 11.049487113952637,
+      "learning_rate": 2.370038982583056e-06,
+      "loss": 0.8068,
+      "step": 2459
+    },
+    {
+      "epoch": 0.9314653540325635,
+      "grad_norm": 15.248514175415039,
+      "learning_rate": 2.344168104766109e-06,
+      "loss": 0.7894,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9318439984854222,
+      "grad_norm": 13.173613548278809,
+      "learning_rate": 2.3184375289406202e-06,
+      "loss": 1.3524,
+      "step": 2461
+    },
+    {
+      "epoch": 0.9322226429382809,
+      "grad_norm": 9.832080841064453,
+      "learning_rate": 2.2928472920736744e-06,
+      "loss": 0.3662,
+      "step": 2462
+    },
+    {
+      "epoch": 0.9326012873911397,
+      "grad_norm": 11.172598838806152,
+      "learning_rate": 2.2673974309307066e-06,
+      "loss": 0.5176,
+      "step": 2463
+    },
+    {
+      "epoch": 0.9329799318439985,
+      "grad_norm": 8.801765441894531,
+      "learning_rate": 2.2420879820755023e-06,
+      "loss": 0.4289,
+      "step": 2464
+    },
+    {
+      "epoch": 0.9333585762968573,
+      "grad_norm": 7.546141624450684,
+      "learning_rate": 2.2169189818701307e-06,
+      "loss": 0.4665,
+      "step": 2465
+    },
+    {
+      "epoch": 0.933737220749716,
+      "grad_norm": 11.858968734741211,
+      "learning_rate": 2.191890466474844e-06,
+      "loss": 0.5132,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9341158652025748,
+      "grad_norm": 11.472759246826172,
+      "learning_rate": 2.1670024718480675e-06,
+      "loss": 0.3973,
+      "step": 2467
+    },
+    {
+      "epoch": 0.9344945096554336,
+      "grad_norm": 11.55582332611084,
+      "learning_rate": 2.1422550337463322e-06,
+      "loss": 0.5675,
+      "step": 2468
+    },
+    {
+      "epoch": 0.9348731541082923,
+      "grad_norm": 8.275731086730957,
+      "learning_rate": 2.117648187724286e-06,
+      "loss": 0.3216,
+      "step": 2469
+    },
+    {
+      "epoch": 0.935251798561151,
+      "grad_norm": 17.31087875366211,
+      "learning_rate": 2.0931819691345277e-06,
+      "loss": 0.4359,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9356304430140099,
+      "grad_norm": 6.022243976593018,
+      "learning_rate": 2.06885641312764e-06,
+      "loss": 0.1946,
+      "step": 2471
+    },
+    {
+      "epoch": 0.9360090874668686,
+      "grad_norm": 7.993716239929199,
+      "learning_rate": 2.0446715546521112e-06,
+      "loss": 0.2927,
+      "step": 2472
+    },
+    {
+      "epoch": 0.9363877319197274,
+      "grad_norm": 5.820225238800049,
+      "learning_rate": 2.0206274284542804e-06,
+      "loss": 0.1208,
+      "step": 2473
+    },
+    {
+      "epoch": 0.9367663763725862,
+      "grad_norm": 10.578958511352539,
+      "learning_rate": 1.9967240690783262e-06,
+      "loss": 0.2527,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9371450208254449,
+      "grad_norm": 6.542937755584717,
+      "learning_rate": 1.972961510866178e-06,
+      "loss": 0.3776,
+      "step": 2475
+    },
+    {
+      "epoch": 0.9375236652783037,
+      "grad_norm": 9.507660865783691,
+      "learning_rate": 1.9493397879574493e-06,
+      "loss": 1.6838,
+      "step": 2476
+    },
+    {
+      "epoch": 0.9379023097311624,
+      "grad_norm": 11.040675163269043,
+      "learning_rate": 1.9258589342894485e-06,
+      "loss": 1.5473,
+      "step": 2477
+    },
+    {
+      "epoch": 0.9382809541840212,
+      "grad_norm": 9.63912296295166,
+      "learning_rate": 1.902518983597068e-06,
+      "loss": 1.1266,
+      "step": 2478
+    },
+    {
+      "epoch": 0.93865959863688,
+      "grad_norm": 11.154900550842285,
+      "learning_rate": 1.879319969412796e-06,
+      "loss": 1.205,
+      "step": 2479
+    },
+    {
+      "epoch": 0.9390382430897387,
+      "grad_norm": 11.759103775024414,
+      "learning_rate": 1.8562619250666047e-06,
+      "loss": 1.2244,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9394168875425976,
+      "grad_norm": 10.546182632446289,
+      "learning_rate": 1.8333448836859723e-06,
+      "loss": 0.8522,
+      "step": 2481
+    },
+    {
+      "epoch": 0.9397955319954563,
+      "grad_norm": 11.785140991210938,
+      "learning_rate": 1.810568878195773e-06,
+      "loss": 0.8214,
+      "step": 2482
+    },
+    {
+      "epoch": 0.940174176448315,
+      "grad_norm": 13.62626838684082,
+      "learning_rate": 1.787933941318265e-06,
+      "loss": 0.9527,
+      "step": 2483
+    },
+    {
+      "epoch": 0.9405528209011738,
+      "grad_norm": 12.887011528015137,
+      "learning_rate": 1.7654401055730129e-06,
+      "loss": 1.0685,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9409314653540326,
+      "grad_norm": 9.904094696044922,
+      "learning_rate": 1.7430874032768885e-06,
+      "loss": 0.7109,
+      "step": 2485
+    },
+    {
+      "epoch": 0.9413101098068913,
+      "grad_norm": 7.248521327972412,
+      "learning_rate": 1.7208758665439917e-06,
+      "loss": 0.3924,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9416887542597501,
+      "grad_norm": 9.920915603637695,
+      "learning_rate": 1.6988055272855962e-06,
+      "loss": 0.3695,
+      "step": 2487
+    },
+    {
+      "epoch": 0.9420673987126088,
+      "grad_norm": 8.971296310424805,
+      "learning_rate": 1.676876417210127e-06,
+      "loss": 0.3631,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9424460431654677,
+      "grad_norm": 15.373361587524414,
+      "learning_rate": 1.6550885678231042e-06,
+      "loss": 0.9075,
+      "step": 2489
+    },
+    {
+      "epoch": 0.9428246876183264,
+      "grad_norm": 10.004647254943848,
+      "learning_rate": 1.6334420104271109e-06,
+      "loss": 0.6522,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9432033320711851,
+      "grad_norm": 10.934609413146973,
+      "learning_rate": 1.6119367761217142e-06,
+      "loss": 0.4777,
+      "step": 2491
+    },
+    {
+      "epoch": 0.943581976524044,
+      "grad_norm": 8.394415855407715,
+      "learning_rate": 1.590572895803455e-06,
+      "loss": 0.2631,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9439606209769027,
+      "grad_norm": 11.458495140075684,
+      "learning_rate": 1.569350400165781e-06,
+      "loss": 0.62,
+      "step": 2493
+    },
+    {
+      "epoch": 0.9443392654297614,
+      "grad_norm": 15.056721687316895,
+      "learning_rate": 1.548269319699036e-06,
+      "loss": 0.4027,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9447179098826202,
+      "grad_norm": 12.485556602478027,
+      "learning_rate": 1.5273296846903707e-06,
+      "loss": 0.3424,
+      "step": 2495
+    },
+    {
+      "epoch": 0.945096554335479,
+      "grad_norm": 10.782124519348145,
+      "learning_rate": 1.50653152522372e-06,
+      "loss": 0.4142,
+      "step": 2496
+    },
+    {
+      "epoch": 0.9454751987883377,
+      "grad_norm": 7.636911869049072,
+      "learning_rate": 1.4858748711797822e-06,
+      "loss": 0.2571,
+      "step": 2497
+    },
+    {
+      "epoch": 0.9458538432411965,
+      "grad_norm": 7.60603666305542,
+      "learning_rate": 1.4653597522359396e-06,
+      "loss": 0.2453,
+      "step": 2498
+    },
+    {
+      "epoch": 0.9462324876940553,
+      "grad_norm": 10.03763198852539,
+      "learning_rate": 1.444986197866227e-06,
+      "loss": 0.3472,
+      "step": 2499
+    },
+    {
+      "epoch": 0.9466111321469141,
+      "grad_norm": 3.0592634677886963,
+      "learning_rate": 1.424754237341297e-06,
+      "loss": 0.0667,
+      "step": 2500
+    },
+    {
+      "epoch": 0.9469897765997728,
+      "grad_norm": 10.650063514709473,
+      "learning_rate": 1.4046638997283978e-06,
+      "loss": 2.1954,
+      "step": 2501
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 8.665024757385254,
+      "learning_rate": 1.3847152138912744e-06,
+      "loss": 1.0272,
+      "step": 2502
+    },
+    {
+      "epoch": 0.9477470655054904,
+      "grad_norm": 9.950023651123047,
+      "learning_rate": 1.3649082084901676e-06,
+      "loss": 1.3072,
+      "step": 2503
+    },
+    {
+      "epoch": 0.9481257099583491,
+      "grad_norm": 9.413839340209961,
+      "learning_rate": 1.345242911981781e-06,
+      "loss": 0.8727,
+      "step": 2504
+    },
+    {
+      "epoch": 0.9485043544112078,
+      "grad_norm": 10.411212921142578,
+      "learning_rate": 1.3257193526192257e-06,
+      "loss": 0.6985,
+      "step": 2505
+    },
+    {
+      "epoch": 0.9488829988640667,
+      "grad_norm": 12.88664436340332,
+      "learning_rate": 1.3063375584519532e-06,
+      "loss": 0.6111,
+      "step": 2506
+    },
+    {
+      "epoch": 0.9492616433169254,
+      "grad_norm": 9.880784034729004,
+      "learning_rate": 1.2870975573257783e-06,
+      "loss": 0.6718,
+      "step": 2507
+    },
+    {
+      "epoch": 0.9496402877697842,
+      "grad_norm": 9.987343788146973,
+      "learning_rate": 1.267999376882767e-06,
+      "loss": 0.6956,
+      "step": 2508
+    },
+    {
+      "epoch": 0.9500189322226429,
+      "grad_norm": 10.966734886169434,
+      "learning_rate": 1.2490430445612488e-06,
+      "loss": 0.7292,
+      "step": 2509
+    },
+    {
+      "epoch": 0.9503975766755017,
+      "grad_norm": 13.379451751708984,
+      "learning_rate": 1.230228587595772e-06,
+      "loss": 0.527,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9507762211283605,
+      "grad_norm": 11.814391136169434,
+      "learning_rate": 1.2115560330170362e-06,
+      "loss": 0.6957,
+      "step": 2511
+    },
+    {
+      "epoch": 0.9511548655812192,
+      "grad_norm": 9.619377136230469,
+      "learning_rate": 1.1930254076518488e-06,
+      "loss": 0.753,
+      "step": 2512
+    },
+    {
+      "epoch": 0.951533510034078,
+      "grad_norm": 10.99007797241211,
+      "learning_rate": 1.1746367381231582e-06,
+      "loss": 0.463,
+      "step": 2513
+    },
+    {
+      "epoch": 0.9519121544869368,
+      "grad_norm": 13.722533226013184,
+      "learning_rate": 1.1563900508499425e-06,
+      "loss": 0.6937,
+      "step": 2514
+    },
+    {
+      "epoch": 0.9522907989397955,
+      "grad_norm": 11.671647071838379,
+      "learning_rate": 1.1382853720471764e-06,
+      "loss": 0.4535,
+      "step": 2515
+    },
+    {
+      "epoch": 0.9526694433926542,
+      "grad_norm": 7.917880535125732,
+      "learning_rate": 1.1203227277258198e-06,
+      "loss": 0.4509,
+      "step": 2516
+    },
+    {
+      "epoch": 0.9530480878455131,
+      "grad_norm": 12.8670072555542,
+      "learning_rate": 1.1025021436927962e-06,
+      "loss": 0.5246,
+      "step": 2517
+    },
+    {
+      "epoch": 0.9534267322983718,
+      "grad_norm": 13.827978134155273,
+      "learning_rate": 1.0848236455509031e-06,
+      "loss": 0.5806,
+      "step": 2518
+    },
+    {
+      "epoch": 0.9538053767512306,
+      "grad_norm": 12.70480728149414,
+      "learning_rate": 1.0672872586988237e-06,
+      "loss": 0.4097,
+      "step": 2519
+    },
+    {
+      "epoch": 0.9541840212040894,
+      "grad_norm": 10.580239295959473,
+      "learning_rate": 1.0498930083310376e-06,
+      "loss": 0.4366,
+      "step": 2520
+    },
+    {
+      "epoch": 0.9545626656569481,
+      "grad_norm": 8.138960838317871,
+      "learning_rate": 1.032640919437844e-06,
+      "loss": 0.2703,
+      "step": 2521
+    },
+    {
+      "epoch": 0.9549413101098069,
+      "grad_norm": 6.230162620544434,
+      "learning_rate": 1.0155310168053156e-06,
+      "loss": 0.2412,
+      "step": 2522
+    },
+    {
+      "epoch": 0.9553199545626656,
+      "grad_norm": 17.800689697265625,
+      "learning_rate": 9.985633250152116e-07,
+      "loss": 0.4268,
+      "step": 2523
+    },
+    {
+      "epoch": 0.9556985990155245,
+      "grad_norm": 29.00356674194336,
+      "learning_rate": 9.817378684449763e-07,
+      "loss": 0.2153,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9560772434683832,
+      "grad_norm": 14.519379615783691,
+      "learning_rate": 9.6505467126774e-07,
+      "loss": 0.1794,
+      "step": 2525
+    },
+    {
+      "epoch": 0.9564558879212419,
+      "grad_norm": 9.758588790893555,
+      "learning_rate": 9.485137574522185e-07,
+      "loss": 1.728,
+      "step": 2526
+    },
+    {
+      "epoch": 0.9568345323741008,
+      "grad_norm": 10.424653053283691,
+      "learning_rate": 9.321151507627135e-07,
+      "loss": 1.3501,
+      "step": 2527
+    },
+    {
+      "epoch": 0.9572131768269595,
+      "grad_norm": 10.270801544189453,
+      "learning_rate": 9.158588747590902e-07,
+      "loss": 1.3,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9575918212798182,
+      "grad_norm": 10.49842643737793,
+      "learning_rate": 8.997449527966994e-07,
+      "loss": 0.8599,
+      "step": 2529
+    },
+    {
+      "epoch": 0.957970465732677,
+      "grad_norm": 8.8501615524292,
+      "learning_rate": 8.837734080264116e-07,
+      "loss": 0.7401,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9583491101855358,
+      "grad_norm": 10.654547691345215,
+      "learning_rate": 8.679442633945156e-07,
+      "loss": 0.8812,
+      "step": 2531
+    },
+    {
+      "epoch": 0.9587277546383945,
+      "grad_norm": 10.941558837890625,
+      "learning_rate": 8.522575416426981e-07,
+      "loss": 1.1451,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9591063990912533,
+      "grad_norm": 11.180508613586426,
+      "learning_rate": 8.367132653080867e-07,
+      "loss": 0.774,
+      "step": 2533
+    },
+    {
+      "epoch": 0.9594850435441121,
+      "grad_norm": 11.109345436096191,
+      "learning_rate": 8.213114567230951e-07,
+      "loss": 0.5891,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9598636879969709,
+      "grad_norm": 10.302473068237305,
+      "learning_rate": 8.060521380154784e-07,
+      "loss": 0.6217,
+      "step": 2535
+    },
+    {
+      "epoch": 0.9602423324498296,
+      "grad_norm": 10.956271171569824,
+      "learning_rate": 7.90935331108289e-07,
+      "loss": 0.6269,
+      "step": 2536
+    },
+    {
+      "epoch": 0.9606209769026883,
+      "grad_norm": 11.260363578796387,
+      "learning_rate": 7.759610577198206e-07,
+      "loss": 0.5793,
+      "step": 2537
+    },
+    {
+      "epoch": 0.9609996213555472,
+      "grad_norm": 14.293161392211914,
+      "learning_rate": 7.611293393635755e-07,
+      "loss": 0.8875,
+      "step": 2538
+    },
+    {
+      "epoch": 0.9613782658084059,
+      "grad_norm": 11.707453727722168,
+      "learning_rate": 7.46440197348286e-07,
+      "loss": 0.5792,
+      "step": 2539
+    },
+    {
+      "epoch": 0.9617569102612646,
+      "grad_norm": 10.193665504455566,
+      "learning_rate": 7.318936527777931e-07,
+      "loss": 0.4881,
+      "step": 2540
+    },
+    {
+      "epoch": 0.9621355547141235,
+      "grad_norm": 9.130789756774902,
+      "learning_rate": 7.174897265511238e-07,
+      "loss": 0.4617,
+      "step": 2541
+    },
+    {
+      "epoch": 0.9625141991669822,
+      "grad_norm": 7.929288864135742,
+      "learning_rate": 7.032284393623579e-07,
+      "loss": 0.3009,
+      "step": 2542
+    },
+    {
+      "epoch": 0.962892843619841,
+      "grad_norm": 6.23085355758667,
+      "learning_rate": 6.891098117006833e-07,
+      "loss": 0.2789,
+      "step": 2543
+    },
+    {
+      "epoch": 0.9632714880726997,
+      "grad_norm": 16.519512176513672,
+      "learning_rate": 6.751338638502858e-07,
+      "loss": 0.558,
+      "step": 2544
+    },
+    {
+      "epoch": 0.9636501325255585,
+      "grad_norm": 9.027918815612793,
+      "learning_rate": 6.613006158904145e-07,
+      "loss": 0.3583,
+      "step": 2545
+    },
+    {
+      "epoch": 0.9640287769784173,
+      "grad_norm": 9.856430053710938,
+      "learning_rate": 6.476100876952718e-07,
+      "loss": 0.2975,
+      "step": 2546
+    },
+    {
+      "epoch": 0.964407421431276,
+      "grad_norm": 10.270342826843262,
+      "learning_rate": 6.340622989340128e-07,
+      "loss": 0.4139,
+      "step": 2547
+    },
+    {
+      "epoch": 0.9647860658841348,
+      "grad_norm": 10.692875862121582,
+      "learning_rate": 6.206572690707125e-07,
+      "loss": 0.2542,
+      "step": 2548
+    },
+    {
+      "epoch": 0.9651647103369936,
+      "grad_norm": 36.20051574707031,
+      "learning_rate": 6.073950173643873e-07,
+      "loss": 0.6115,
+      "step": 2549
+    },
+    {
+      "epoch": 0.9655433547898523,
+      "grad_norm": 26.375398635864258,
+      "learning_rate": 5.942755628688845e-07,
+      "loss": 0.4395,
+      "step": 2550
+    },
+    {
+      "epoch": 0.965921999242711,
+      "grad_norm": 10.20322322845459,
+      "learning_rate": 5.812989244328937e-07,
+      "loss": 1.5327,
+      "step": 2551
+    },
+    {
+      "epoch": 0.9663006436955699,
+      "grad_norm": 10.076367378234863,
+      "learning_rate": 5.684651206999347e-07,
+      "loss": 1.2404,
+      "step": 2552
+    },
+    {
+      "epoch": 0.9666792881484286,
+      "grad_norm": 9.947564125061035,
+      "learning_rate": 5.557741701083363e-07,
+      "loss": 0.924,
+      "step": 2553
+    },
+    {
+      "epoch": 0.9670579326012874,
+      "grad_norm": 10.444324493408203,
+      "learning_rate": 5.432260908911358e-07,
+      "loss": 1.1889,
+      "step": 2554
+    },
+    {
+      "epoch": 0.9674365770541462,
+      "grad_norm": 9.716720581054688,
+      "learning_rate": 5.308209010761678e-07,
+      "loss": 0.7316,
+      "step": 2555
+    },
+    {
+      "epoch": 0.9678152215070049,
+      "grad_norm": 14.292815208435059,
+      "learning_rate": 5.185586184859426e-07,
+      "loss": 0.8652,
+      "step": 2556
+    },
+    {
+      "epoch": 0.9681938659598637,
+      "grad_norm": 12.14730453491211,
+      "learning_rate": 5.064392607376567e-07,
+      "loss": 0.7933,
+      "step": 2557
+    },
+    {
+      "epoch": 0.9685725104127224,
+      "grad_norm": 9.1866455078125,
+      "learning_rate": 4.94462845243171e-07,
+      "loss": 0.501,
+      "step": 2558
+    },
+    {
+      "epoch": 0.9689511548655813,
+      "grad_norm": 12.76934814453125,
+      "learning_rate": 4.826293892089995e-07,
+      "loss": 0.6006,
+      "step": 2559
+    },
+    {
+      "epoch": 0.96932979931844,
+      "grad_norm": 13.298689842224121,
+      "learning_rate": 4.709389096362427e-07,
+      "loss": 0.654,
+      "step": 2560
+    },
+    {
+      "epoch": 0.9697084437712987,
+      "grad_norm": 12.402892112731934,
+      "learning_rate": 4.593914233205987e-07,
+      "loss": 1.0524,
+      "step": 2561
+    },
+    {
+      "epoch": 0.9700870882241576,
+      "grad_norm": 11.499163627624512,
+      "learning_rate": 4.4798694685231903e-07,
+      "loss": 0.5996,
+      "step": 2562
+    },
+    {
+      "epoch": 0.9704657326770163,
+      "grad_norm": 8.201879501342773,
+      "learning_rate": 4.367254966161971e-07,
+      "loss": 0.5142,
+      "step": 2563
+    },
+    {
+      "epoch": 0.970844377129875,
+      "grad_norm": 14.199493408203125,
+      "learning_rate": 4.2560708879154645e-07,
+      "loss": 0.5288,
+      "step": 2564
+    },
+    {
+      "epoch": 0.9712230215827338,
+      "grad_norm": 9.493123054504395,
+      "learning_rate": 4.1463173935216703e-07,
+      "loss": 0.4889,
+      "step": 2565
+    },
+    {
+      "epoch": 0.9716016660355926,
+      "grad_norm": 11.741094589233398,
+      "learning_rate": 4.037994640663345e-07,
+      "loss": 0.4841,
+      "step": 2566
+    },
+    {
+      "epoch": 0.9719803104884513,
+      "grad_norm": 10.63720417022705,
+      "learning_rate": 3.9311027849674444e-07,
+      "loss": 0.4452,
+      "step": 2567
+    },
+    {
+      "epoch": 0.9723589549413101,
+      "grad_norm": 10.38010025024414,
+      "learning_rate": 3.8256419800055675e-07,
+      "loss": 0.4737,
+      "step": 2568
+    },
+    {
+      "epoch": 0.9727375993941689,
+      "grad_norm": 12.463581085205078,
+      "learning_rate": 3.721612377292849e-07,
+      "loss": 0.2774,
+      "step": 2569
+    },
+    {
+      "epoch": 0.9731162438470277,
+      "grad_norm": 9.494747161865234,
+      "learning_rate": 3.6190141262887333e-07,
+      "loss": 0.2915,
+      "step": 2570
+    },
+    {
+      "epoch": 0.9734948882998864,
+      "grad_norm": 11.594443321228027,
+      "learning_rate": 3.517847374395755e-07,
+      "loss": 0.2961,
+      "step": 2571
+    },
+    {
+      "epoch": 0.9738735327527451,
+      "grad_norm": 7.852182388305664,
+      "learning_rate": 3.418112266960205e-07,
+      "loss": 0.2635,
+      "step": 2572
+    },
+    {
+      "epoch": 0.974252177205604,
+      "grad_norm": 4.673264026641846,
+      "learning_rate": 3.319808947271241e-07,
+      "loss": 0.0947,
+      "step": 2573
+    },
+    {
+      "epoch": 0.9746308216584627,
+      "grad_norm": 5.447509288787842,
+      "learning_rate": 3.222937556561223e-07,
+      "loss": 0.0963,
+      "step": 2574
+    },
+    {
+      "epoch": 0.9750094661113214,
+      "grad_norm": 4.423956394195557,
+      "learning_rate": 3.127498234005044e-07,
+      "loss": 0.1488,
+      "step": 2575
+    },
+    {
+      "epoch": 0.9753881105641803,
+      "grad_norm": 11.625571250915527,
+      "learning_rate": 3.033491116720244e-07,
+      "loss": 2.4858,
+      "step": 2576
+    },
+    {
+      "epoch": 0.975766755017039,
+      "grad_norm": 10.204331398010254,
+      "learning_rate": 2.940916339766675e-07,
+      "loss": 1.637,
+      "step": 2577
+    },
+    {
+      "epoch": 0.9761453994698978,
+      "grad_norm": 10.439549446105957,
+      "learning_rate": 2.849774036146502e-07,
+      "loss": 1.3357,
+      "step": 2578
+    },
+    {
+      "epoch": 0.9765240439227565,
+      "grad_norm": 10.696296691894531,
+      "learning_rate": 2.7600643368036473e-07,
+      "loss": 1.1023,
+      "step": 2579
+    },
+    {
+      "epoch": 0.9769026883756153,
+      "grad_norm": 12.158334732055664,
+      "learning_rate": 2.6717873706240125e-07,
+      "loss": 1.1039,
+      "step": 2580
+    },
+    {
+      "epoch": 0.9772813328284741,
+      "grad_norm": 11.918547630310059,
+      "learning_rate": 2.5849432644348136e-07,
+      "loss": 0.8639,
+      "step": 2581
+    },
+    {
+      "epoch": 0.9776599772813328,
+      "grad_norm": 9.715611457824707,
+      "learning_rate": 2.4995321430050235e-07,
+      "loss": 0.567,
+      "step": 2582
+    },
+    {
+      "epoch": 0.9780386217341916,
+      "grad_norm": 11.445382118225098,
+      "learning_rate": 2.415554129044595e-07,
+      "loss": 0.9997,
+      "step": 2583
+    },
+    {
+      "epoch": 0.9784172661870504,
+      "grad_norm": 10.35413932800293,
+      "learning_rate": 2.333009343204573e-07,
+      "loss": 0.7591,
+      "step": 2584
+    },
+    {
+      "epoch": 0.9787959106399091,
+      "grad_norm": 7.9862165451049805,
+      "learning_rate": 2.2518979040769827e-07,
+      "loss": 0.3323,
+      "step": 2585
+    },
+    {
+      "epoch": 0.9791745550927679,
+      "grad_norm": 12.614017486572266,
+      "learning_rate": 2.1722199281944967e-07,
+      "loss": 0.9701,
+      "step": 2586
+    },
+    {
+      "epoch": 0.9795531995456267,
+      "grad_norm": 11.604450225830078,
+      "learning_rate": 2.0939755300304342e-07,
+      "loss": 0.8196,
+      "step": 2587
+    },
+    {
+      "epoch": 0.9799318439984854,
+      "grad_norm": 12.22137451171875,
+      "learning_rate": 2.0171648219982074e-07,
+      "loss": 0.4567,
+      "step": 2588
+    },
+    {
+      "epoch": 0.9803104884513442,
+      "grad_norm": 11.08056926727295,
+      "learning_rate": 1.941787914451876e-07,
+      "loss": 0.7081,
+      "step": 2589
+    },
+    {
+      "epoch": 0.9806891329042029,
+      "grad_norm": 8.053107261657715,
+      "learning_rate": 1.8678449156852573e-07,
+      "loss": 0.5096,
+      "step": 2590
+    },
+    {
+      "epoch": 0.9810677773570617,
+      "grad_norm": 10.370348930358887,
+      "learning_rate": 1.7953359319320406e-07,
+      "loss": 0.5069,
+      "step": 2591
+    },
+    {
+      "epoch": 0.9814464218099205,
+      "grad_norm": 8.252673149108887,
+      "learning_rate": 1.7242610673658954e-07,
+      "loss": 0.2886,
+      "step": 2592
+    },
+    {
+      "epoch": 0.9818250662627792,
+      "grad_norm": 14.678871154785156,
+      "learning_rate": 1.6546204240999174e-07,
+      "loss": 0.3913,
+      "step": 2593
+    },
+    {
+      "epoch": 0.9822037107156381,
+      "grad_norm": 12.055886268615723,
+      "learning_rate": 1.5864141021868506e-07,
+      "loss": 0.4829,
+      "step": 2594
+    },
+    {
+      "epoch": 0.9825823551684968,
+      "grad_norm": 7.955121040344238,
+      "learning_rate": 1.5196421996184207e-07,
+      "loss": 0.3313,
+      "step": 2595
+    },
+    {
+      "epoch": 0.9829609996213555,
+      "grad_norm": 6.439930438995361,
+      "learning_rate": 1.4543048123257796e-07,
+      "loss": 0.1825,
+      "step": 2596
+    },
+    {
+      "epoch": 0.9833396440742143,
+      "grad_norm": 5.3940582275390625,
+      "learning_rate": 1.3904020341791724e-07,
+      "loss": 0.1774,
+      "step": 2597
+    },
+    {
+      "epoch": 0.9837182885270731,
+      "grad_norm": 23.00211524963379,
+      "learning_rate": 1.3279339569874926e-07,
+      "loss": 0.2979,
+      "step": 2598
+    },
+    {
+      "epoch": 0.9840969329799318,
+      "grad_norm": 10.200654983520508,
+      "learning_rate": 1.2669006704986164e-07,
+      "loss": 0.3489,
+      "step": 2599
+    },
+    {
+      "epoch": 0.9844755774327906,
+      "grad_norm": 12.444687843322754,
+      "learning_rate": 1.2073022623988462e-07,
+      "loss": 0.4883,
+      "step": 2600
+    },
+    {
+      "epoch": 0.9848542218856494,
+      "grad_norm": 8.539031982421875,
+      "learning_rate": 1.1491388183133556e-07,
+      "loss": 1.2458,
+      "step": 2601
+    },
+    {
+      "epoch": 0.9852328663385082,
+      "grad_norm": 12.276918411254883,
+      "learning_rate": 1.092410421805301e-07,
+      "loss": 1.1199,
+      "step": 2602
+    },
+    {
+      "epoch": 0.9856115107913669,
+      "grad_norm": 9.444790840148926,
+      "learning_rate": 1.0371171543763769e-07,
+      "loss": 1.1647,
+      "step": 2603
+    },
+    {
+      "epoch": 0.9859901552442256,
+      "grad_norm": 9.76896858215332,
+      "learning_rate": 9.832590954662602e-08,
+      "loss": 0.8912,
+      "step": 2604
+    },
+    {
+      "epoch": 0.9863687996970845,
+      "grad_norm": 9.74208927154541,
+      "learning_rate": 9.308363224528327e-08,
+      "loss": 0.8606,
+      "step": 2605
+    },
+    {
+      "epoch": 0.9867474441499432,
+      "grad_norm": 8.999712944030762,
+      "learning_rate": 8.798489106517371e-08,
+      "loss": 0.6109,
+      "step": 2606
+    },
+    {
+      "epoch": 0.9871260886028019,
+      "grad_norm": 7.842771530151367,
+      "learning_rate": 8.302969333165989e-08,
+      "loss": 0.5287,
+      "step": 2607
+    },
+    {
+      "epoch": 0.9875047330556608,
+      "grad_norm": 11.93226432800293,
+      "learning_rate": 7.821804616384709e-08,
+      "loss": 0.5248,
+      "step": 2608
+    },
+    {
+      "epoch": 0.9878833775085195,
+      "grad_norm": 14.00164794921875,
+      "learning_rate": 7.354995647465002e-08,
+      "loss": 0.996,
+      "step": 2609
+    },
+    {
+      "epoch": 0.9882620219613782,
+      "grad_norm": 9.637811660766602,
+      "learning_rate": 6.90254309706928e-08,
+      "loss": 0.5948,
+      "step": 2610
+    },
+    {
+      "epoch": 0.988640666414237,
+      "grad_norm": 10.103828430175781,
+      "learning_rate": 6.464447615235347e-08,
+      "loss": 0.667,
+      "step": 2611
+    },
+    {
+      "epoch": 0.9890193108670958,
+      "grad_norm": 8.21738338470459,
+      "learning_rate": 6.04070983137417e-08,
+      "loss": 0.6523,
+      "step": 2612
+    },
+    {
+      "epoch": 0.9893979553199546,
+      "grad_norm": 10.094844818115234,
+      "learning_rate": 5.631330354269882e-08,
+      "loss": 0.57,
+      "step": 2613
+    },
+    {
+      "epoch": 0.9897765997728133,
+      "grad_norm": 9.283138275146484,
+      "learning_rate": 5.236309772077563e-08,
+      "loss": 0.5518,
+      "step": 2614
+    },
+    {
+      "epoch": 0.9901552442256721,
+      "grad_norm": 8.835543632507324,
+      "learning_rate": 4.855648652321021e-08,
+      "loss": 0.3256,
+      "step": 2615
+    },
+    {
+      "epoch": 0.9905338886785309,
+      "grad_norm": 6.809988498687744,
+      "learning_rate": 4.4893475418983365e-08,
+      "loss": 0.1993,
+      "step": 2616
+    },
+    {
+      "epoch": 0.9909125331313896,
+      "grad_norm": 7.190274238586426,
+      "learning_rate": 4.137406967070767e-08,
+      "loss": 0.3838,
+      "step": 2617
+    },
+    {
+      "epoch": 0.9912911775842483,
+      "grad_norm": 13.799723625183105,
+      "learning_rate": 3.799827433472736e-08,
+      "loss": 0.5659,
+      "step": 2618
+    },
+    {
+      "epoch": 0.9916698220371072,
+      "grad_norm": 6.342631816864014,
+      "learning_rate": 3.47660942610295e-08,
+      "loss": 0.227,
+      "step": 2619
+    },
+    {
+      "epoch": 0.9920484664899659,
+      "grad_norm": 7.028225421905518,
+      "learning_rate": 3.1677534093299545e-08,
+      "loss": 0.2465,
+      "step": 2620
+    },
+    {
+      "epoch": 0.9924271109428247,
+      "grad_norm": 9.69013786315918,
+      "learning_rate": 2.873259826885466e-08,
+      "loss": 0.3781,
+      "step": 2621
+    },
+    {
+      "epoch": 0.9928057553956835,
+      "grad_norm": 9.710000038146973,
+      "learning_rate": 2.5931291018677086e-08,
+      "loss": 0.2296,
+      "step": 2622
+    },
+    {
+      "epoch": 0.9931843998485422,
+      "grad_norm": 11.306065559387207,
+      "learning_rate": 2.3273616367414097e-08,
+      "loss": 0.2997,
+      "step": 2623
+    },
+    {
+      "epoch": 0.993563044301401,
+      "grad_norm": 6.006664276123047,
+      "learning_rate": 2.0759578133333623e-08,
+      "loss": 0.1308,
+      "step": 2624
+    },
+    {
+      "epoch": 0.9939416887542597,
+      "grad_norm": 6.388942241668701,
+      "learning_rate": 1.8389179928357538e-08,
+      "loss": 0.1969,
+      "step": 2625
+    },
+    {
+      "epoch": 0.9943203332071185,
+      "grad_norm": 11.715986251831055,
+      "learning_rate": 1.616242515802835e-08,
+      "loss": 1.8615,
+      "step": 2626
+    },
+    {
+      "epoch": 0.9946989776599773,
+      "grad_norm": 9.530550003051758,
+      "learning_rate": 1.4079317021520321e-08,
+      "loss": 1.3649,
+      "step": 2627
+    },
+    {
+      "epoch": 0.995077622112836,
+      "grad_norm": 11.392560958862305,
+      "learning_rate": 1.2139858511628356e-08,
+      "loss": 0.9551,
+      "step": 2628
+    },
+    {
+      "epoch": 0.9954562665656949,
+      "grad_norm": 11.880497932434082,
+      "learning_rate": 1.0344052414779094e-08,
+      "loss": 0.8825,
+      "step": 2629
+    },
+    {
+      "epoch": 0.9958349110185536,
+      "grad_norm": 8.874764442443848,
+      "learning_rate": 8.691901310997619e-09,
+      "loss": 0.5746,
+      "step": 2630
+    },
+    {
+      "epoch": 0.9962135554714123,
+      "grad_norm": 10.284222602844238,
+      "learning_rate": 7.1834075739296566e-09,
+      "loss": 0.7843,
+      "step": 2631
+    },
+    {
+      "epoch": 0.9965921999242711,
+      "grad_norm": 10.232461929321289,
+      "learning_rate": 5.818573370830471e-09,
+      "loss": 0.7046,
+      "step": 2632
+    },
+    {
+      "epoch": 0.9969708443771299,
+      "grad_norm": 11.678064346313477,
+      "learning_rate": 4.597400662553764e-09,
+      "loss": 0.4164,
+      "step": 2633
+    },
+    {
+      "epoch": 0.9973494888299886,
+      "grad_norm": 9.27035903930664,
+      "learning_rate": 3.5198912035516727e-09,
+      "loss": 0.3791,
+      "step": 2634
+    },
+    {
+      "epoch": 0.9977281332828474,
+      "grad_norm": 12.330588340759277,
+      "learning_rate": 2.586046541874776e-09,
+      "loss": 0.5568,
+      "step": 2635
+    },
+    {
+      "epoch": 0.9981067777357062,
+      "grad_norm": 11.334096908569336,
+      "learning_rate": 1.7958680191942911e-09,
+      "loss": 0.5329,
+      "step": 2636
+    },
+    {
+      "epoch": 0.998485422188565,
+      "grad_norm": 12.605074882507324,
+      "learning_rate": 1.149356770746568e-09,
+      "loss": 0.4182,
+      "step": 2637
+    },
+    {
+      "epoch": 0.9988640666414237,
+      "grad_norm": 7.650578022003174,
+      "learning_rate": 6.465137253663934e-10,
+      "loss": 0.2754,
+      "step": 2638
+    },
+    {
+      "epoch": 0.9992427110942824,
+      "grad_norm": 8.941444396972656,
+      "learning_rate": 2.873396055091959e-10,
+      "loss": 0.3273,
+      "step": 2639
+    },
+    {
+      "epoch": 0.9996213555471413,
+      "grad_norm": 10.42584228515625,
+      "learning_rate": 7.183492717333096e-11,
+      "loss": 0.1813,
+      "step": 2640
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 7.037883281707764,
+      "learning_rate": 0.0,
+      "loss": 0.2703,
+      "step": 2641
     }
   ],
   "logging_steps": 1,
@@ -13934,12 +18540,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 1.1154459142461063e+19,
+  "total_flos": 1.485573706265238e+19,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null