diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,1058 +1,7847 @@
 {
-  "best_global_step": 1038,
-  "best_metric": 1.032158374786377,
-  "best_model_checkpoint": "C:\\unity_train\\output\\unity-coder-adapter\\checkpoint-1038",
-  "epoch": 1.0,
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 1038,
+  "global_step": 7839,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "entropy": 1.0811691626906395,
-      "epoch": 0.00963623223319682,
-      "grad_norm": 0.35738348960876465,
-      "learning_rate": 1.153846153846154e-05,
-      "loss": 2.42077693939209,
-      "mean_token_accuracy": 0.5671707078814506,
-      "num_tokens": 59463.0,
+      "entropy": 1.0787046998739243,
+      "epoch": 0.003827018752391887,
+      "grad_norm": 0.37200024724006653,
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 2.590205955505371,
+      "mean_token_accuracy": 0.5478626236319541,
+      "num_tokens": 43996.0,
       "step": 10
     },
     {
-      "entropy": 1.0596500962972641,
-      "epoch": 0.01927246446639364,
-      "grad_norm": 0.26245468854904175,
-      "learning_rate": 2.435897435897436e-05,
-      "loss": 2.3048450469970705,
-      "mean_token_accuracy": 0.5807798743247986,
-      "num_tokens": 122441.0,
+      "entropy": 1.1275236845016479,
+      "epoch": 0.007654037504783774,
+      "grad_norm": 0.4282406270503998,
+      "learning_rate": 9.693877551020408e-06,
+      "loss": 2.732739067077637,
+      "mean_token_accuracy": 0.532574575394392,
+      "num_tokens": 84448.0,
       "step": 20
     },
     {
-      "entropy": 1.1488549172878266,
-      "epoch": 0.02890869669959046,
-      "grad_norm": 0.21727126836776733,
-      "learning_rate": 3.717948717948718e-05,
-      "loss": 2.15008487701416,
-      "mean_token_accuracy": 0.5763025932013989,
-      "num_tokens": 178300.0,
+      "entropy": 1.1098140180110931,
+      "epoch": 0.011481056257175661,
+      "grad_norm": 0.45254817605018616,
+      "learning_rate": 1.479591836734694e-05,
+      "loss": 2.595915603637695,
+      "mean_token_accuracy": 0.5385765254497528,
+      "num_tokens": 127136.0,
       "step": 30
     },
     {
-      "entropy": 1.324630731344223,
-      "epoch": 0.03854492893278728,
-      "grad_norm": 0.09244390577077866,
-      "learning_rate": 5e-05,
-      "loss": 1.9429821014404296,
-      "mean_token_accuracy": 0.595233204215765,
-      "num_tokens": 240279.0,
+      "entropy": 1.1594089552760125,
+      "epoch": 0.015308075009567547,
+      "grad_norm": 0.26816287636756897,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 2.3643749237060545,
+      "mean_token_accuracy": 0.5597088657319546,
+      "num_tokens": 172549.0,
       "step": 40
     },
     {
-      "entropy": 1.54823177754879,
-      "epoch": 0.0481811611659841,
-      "grad_norm": 0.07658641040325165,
-      "learning_rate": 6.282051282051282e-05,
-      "loss": 1.8167322158813477,
-      "mean_token_accuracy": 0.6044709533452988,
-      "num_tokens": 302914.0,
+      "entropy": 1.2817068248987198,
+      "epoch": 0.019135093761959432,
+      "grad_norm": 0.19904343783855438,
+      "learning_rate": 2.5e-05,
+      "loss": 2.1694852828979494,
+      "mean_token_accuracy": 0.5605865910649299,
+      "num_tokens": 218317.0,
       "step": 50
     },
     {
-      "entropy": 1.6716329783201218,
-      "epoch": 0.05781739339918092,
-      "grad_norm": 0.0950954258441925,
-      "learning_rate": 7.564102564102564e-05,
-      "loss": 1.7055500030517579,
-      "mean_token_accuracy": 0.6348149433732033,
-      "num_tokens": 351888.0,
+      "entropy": 1.356394973397255,
+      "epoch": 0.022962112514351322,
+      "grad_norm": 0.21708081662654877,
+      "learning_rate": 3.0102040816326533e-05,
+      "loss": 2.0824514389038087,
+      "mean_token_accuracy": 0.584179612249136,
+      "num_tokens": 255107.0,
       "step": 60
     },
     {
-      "entropy": 1.519571453332901,
-      "epoch": 0.06745362563237774,
-      "grad_norm": 0.07054093480110168,
-      "learning_rate": 8.846153846153847e-05,
-      "loss": 1.5317837715148925,
-      "mean_token_accuracy": 0.6589120730757714,
-      "num_tokens": 414923.0,
+      "entropy": 1.4309053242206573,
+      "epoch": 0.026789131266743208,
+      "grad_norm": 0.09860006719827652,
+      "learning_rate": 3.520408163265306e-05,
+      "loss": 1.777943229675293,
+      "mean_token_accuracy": 0.6243460461497307,
+      "num_tokens": 298973.0,
       "step": 70
     },
     {
-      "entropy": 1.5119123458862305,
-      "epoch": 0.07708985786557455,
-      "grad_norm": 0.07167445123195648,
-      "learning_rate": 0.00010128205128205129,
-      "loss": 1.4823562622070312,
-      "mean_token_accuracy": 0.6631396897137165,
-      "num_tokens": 468593.0,
+      "entropy": 1.4572882741689681,
+      "epoch": 0.030616150019135095,
+      "grad_norm": 0.07813975214958191,
+      "learning_rate": 4.0306122448979596e-05,
+      "loss": 1.7085393905639648,
+      "mean_token_accuracy": 0.641383134573698,
+      "num_tokens": 343287.0,
       "step": 80
     },
     {
-      "entropy": 1.3805345997214318,
-      "epoch": 0.08672609009877139,
-      "grad_norm": 0.10775867104530334,
-      "learning_rate": 0.0001141025641025641,
-      "loss": 1.413505458831787,
-      "mean_token_accuracy": 0.6830234751105309,
-      "num_tokens": 525914.0,
+      "entropy": 1.5205285474658012,
+      "epoch": 0.03444316877152698,
+      "grad_norm": 0.08015387505292892,
+      "learning_rate": 4.5408163265306124e-05,
+      "loss": 1.680305290222168,
+      "mean_token_accuracy": 0.6431211873888969,
+      "num_tokens": 376882.0,
       "step": 90
     },
     {
-      "entropy": 1.4891180142760276,
-      "epoch": 0.0963623223319682,
-      "grad_norm": 0.05629459396004677,
-      "learning_rate": 0.00012692307692307693,
-      "loss": 1.5126714706420898,
-      "mean_token_accuracy": 0.6582696467638016,
-      "num_tokens": 589835.0,
+      "entropy": 1.529197846353054,
+      "epoch": 0.038270187523918864,
+      "grad_norm": 0.1533895879983902,
+      "learning_rate": 5.051020408163265e-05,
+      "loss": 1.538798999786377,
+      "mean_token_accuracy": 0.6604955434799195,
+      "num_tokens": 414084.0,
       "step": 100
     },
     {
-      "entropy": 1.4281011313199996,
-      "epoch": 0.10599855456516502,
-      "grad_norm": 0.05547759681940079,
-      "learning_rate": 0.00013974358974358974,
-      "loss": 1.44595947265625,
-      "mean_token_accuracy": 0.6682780802249908,
-      "num_tokens": 648138.0,
+      "entropy": 1.454574093222618,
+      "epoch": 0.04209720627631076,
+      "grad_norm": 0.08893708884716034,
+      "learning_rate": 5.561224489795919e-05,
+      "loss": 1.482753372192383,
+      "mean_token_accuracy": 0.6748222857713699,
+      "num_tokens": 451489.0,
       "step": 110
     },
     {
-      "entropy": 1.3978321179747581,
-      "epoch": 0.11563478679836184,
-      "grad_norm": 0.05473790690302849,
-      "learning_rate": 0.00015256410256410255,
-      "loss": 1.4433627128601074,
-      "mean_token_accuracy": 0.674263383448124,
-      "num_tokens": 702219.0,
+      "entropy": 1.4169663548469544,
+      "epoch": 0.045924225028702644,
+      "grad_norm": 0.10994797945022583,
+      "learning_rate": 6.0714285714285715e-05,
+      "loss": 1.4192767143249512,
+      "mean_token_accuracy": 0.6815833821892738,
+      "num_tokens": 492001.0,
       "step": 120
     },
     {
-      "entropy": 1.4054933816194535,
-      "epoch": 0.12527101903155866,
-      "grad_norm": 0.06133154034614563,
-      "learning_rate": 0.0001653846153846154,
-      "loss": 1.4284349441528321,
-      "mean_token_accuracy": 0.6733272626996041,
-      "num_tokens": 759042.0,
+      "entropy": 1.3825553365051746,
+      "epoch": 0.04975124378109453,
+      "grad_norm": 0.09565065056085587,
+      "learning_rate": 6.581632653061225e-05,
+      "loss": 1.4434242248535156,
+      "mean_token_accuracy": 0.6830388471484184,
+      "num_tokens": 534229.0,
       "step": 130
     },
     {
-      "entropy": 1.452716739475727,
-      "epoch": 0.13490725126475547,
-      "grad_norm": 0.05656077712774277,
-      "learning_rate": 0.00017820512820512823,
-      "loss": 1.4834243774414062,
-      "mean_token_accuracy": 0.6669867470860481,
-      "num_tokens": 822515.0,
+      "entropy": 1.3660246580839157,
+      "epoch": 0.053578262533486416,
+      "grad_norm": 0.09277962148189545,
+      "learning_rate": 7.091836734693877e-05,
+      "loss": 1.3881919860839844,
+      "mean_token_accuracy": 0.6916770502924919,
+      "num_tokens": 573333.0,
       "step": 140
     },
     {
-      "entropy": 1.2200356990098953,
-      "epoch": 0.1445434834979523,
-      "grad_norm": 0.06002284586429596,
-      "learning_rate": 0.00019102564102564104,
-      "loss": 1.2548653602600097,
-      "mean_token_accuracy": 0.7205198630690575,
-      "num_tokens": 876506.0,
+      "entropy": 1.3041961744427681,
+      "epoch": 0.0574052812858783,
+      "grad_norm": 0.14179331064224243,
+      "learning_rate": 7.60204081632653e-05,
+      "loss": 1.3051923751831054,
+      "mean_token_accuracy": 0.7079917460680007,
+      "num_tokens": 612198.0,
       "step": 150
     },
     {
-      "entropy": 1.3374179184436799,
-      "epoch": 0.1541797157311491,
-      "grad_norm": 0.07770856469869614,
-      "learning_rate": 0.00019979716024340772,
-      "loss": 1.383979606628418,
-      "mean_token_accuracy": 0.6905814677476882,
-      "num_tokens": 934413.0,
+      "entropy": 1.3157715648412704,
+      "epoch": 0.06123230003827019,
+      "grad_norm": 0.11061020940542221,
+      "learning_rate": 8.112244897959184e-05,
+      "loss": 1.3127019882202149,
+      "mean_token_accuracy": 0.6986684441566468,
+      "num_tokens": 654309.0,
       "step": 160
     },
     {
-      "entropy": 1.275234666466713,
-      "epoch": 0.16381594796434595,
-      "grad_norm": 0.06369830667972565,
-      "learning_rate": 0.0001991210277214334,
-      "loss": 1.2860990524291993,
-      "mean_token_accuracy": 0.7051964432001114,
-      "num_tokens": 992594.0,
+      "entropy": 1.2894421368837357,
+      "epoch": 0.06505931879066207,
+      "grad_norm": 0.12903185188770294,
+      "learning_rate": 8.622448979591838e-05,
+      "loss": 1.3279677391052247,
+      "mean_token_accuracy": 0.7028318449854851,
+      "num_tokens": 694076.0,
       "step": 170
     },
     {
-      "entropy": 1.3504198059439658,
-      "epoch": 0.17345218019754277,
-      "grad_norm": 0.07278598099946976,
-      "learning_rate": 0.0001984448951994591,
-      "loss": 1.3741103172302247,
-      "mean_token_accuracy": 0.6862095996737481,
-      "num_tokens": 1048524.0,
+      "entropy": 1.2676123276352882,
+      "epoch": 0.06888633754305395,
+      "grad_norm": 0.10816285014152527,
+      "learning_rate": 9.13265306122449e-05,
+      "loss": 1.3254461288452148,
+      "mean_token_accuracy": 0.7063754379749299,
+      "num_tokens": 733806.0,
       "step": 180
     },
     {
-      "entropy": 1.2644047453999518,
-      "epoch": 0.1830884124307396,
-      "grad_norm": 0.07138558477163315,
-      "learning_rate": 0.00019776876267748477,
-      "loss": 1.2995844841003419,
-      "mean_token_accuracy": 0.706048458814621,
-      "num_tokens": 1109157.0,
+      "entropy": 1.0988808318972587,
+      "epoch": 0.07271335629544584,
+      "grad_norm": 0.08472651243209839,
+      "learning_rate": 9.642857142857143e-05,
+      "loss": 1.1510659217834474,
+      "mean_token_accuracy": 0.7411173984408379,
+      "num_tokens": 771622.0,
       "step": 190
     },
     {
-      "entropy": 1.3248096346855163,
-      "epoch": 0.1927246446639364,
-      "grad_norm": 0.08208955079317093,
-      "learning_rate": 0.00019709263015551048,
-      "loss": 1.3252597808837892,
-      "mean_token_accuracy": 0.7000985190272331,
-      "num_tokens": 1170070.0,
+      "entropy": 1.2284281507134438,
+      "epoch": 0.07654037504783773,
+      "grad_norm": 0.10897475481033325,
+      "learning_rate": 0.00010153061224489797,
+      "loss": 1.2724005699157714,
+      "mean_token_accuracy": 0.7178638219833374,
+      "num_tokens": 813167.0,
       "step": 200
     },
     {
-      "entropy": 1.2682933300733565,
-      "epoch": 0.20236087689713322,
-      "grad_norm": 0.055171046406030655,
-      "learning_rate": 0.00019641649763353617,
-      "loss": 1.3352859497070313,
-      "mean_token_accuracy": 0.700250719487667,
-      "num_tokens": 1229436.0,
+      "entropy": 1.2159623876214027,
+      "epoch": 0.08036739380022963,
+      "grad_norm": 0.12170197069644928,
+      "learning_rate": 0.0001066326530612245,
+      "loss": 1.26397647857666,
+      "mean_token_accuracy": 0.7138236090540886,
+      "num_tokens": 856210.0,
       "step": 210
     },
     {
-      "entropy": 1.347843087464571,
-      "epoch": 0.21199710913033004,
-      "grad_norm": 0.06492508947849274,
-      "learning_rate": 0.00019574036511156188,
-      "loss": 1.3542288780212401,
-      "mean_token_accuracy": 0.6922637760639191,
-      "num_tokens": 1287532.0,
+      "entropy": 1.2309471271932124,
+      "epoch": 0.08419441255262151,
+      "grad_norm": 0.08406181633472443,
+      "learning_rate": 0.00011173469387755102,
+      "loss": 1.3110918998718262,
+      "mean_token_accuracy": 0.7173333883285522,
+      "num_tokens": 893432.0,
       "step": 220
     },
     {
-      "entropy": 1.218190498650074,
-      "epoch": 0.22163334136352686,
-      "grad_norm": 0.0640958771109581,
-      "learning_rate": 0.00019506423258958757,
-      "loss": 1.2281950950622558,
-      "mean_token_accuracy": 0.718992106616497,
-      "num_tokens": 1344357.0,
+      "entropy": 1.2228039711713792,
+      "epoch": 0.0880214313050134,
+      "grad_norm": 0.10588081181049347,
+      "learning_rate": 0.00011683673469387754,
+      "loss": 1.2445635795593262,
+      "mean_token_accuracy": 0.7160170584917068,
+      "num_tokens": 931919.0,
       "step": 230
     },
     {
-      "entropy": 1.3014725491404533,
-      "epoch": 0.23126957359672368,
-      "grad_norm": 0.05718906223773956,
-      "learning_rate": 0.00019438810006761325,
-      "loss": 1.3229006767272948,
-      "mean_token_accuracy": 0.6986020863056183,
-      "num_tokens": 1405209.0,
+      "entropy": 1.138296764343977,
+      "epoch": 0.09184845005740529,
+      "grad_norm": 0.110760398209095,
+      "learning_rate": 0.00012193877551020409,
+      "loss": 1.2083134651184082,
+      "mean_token_accuracy": 0.7306654810905456,
+      "num_tokens": 976039.0,
       "step": 240
     },
     {
-      "entropy": 1.227053464204073,
-      "epoch": 0.2409058058299205,
-      "grad_norm": 0.05763186141848564,
-      "learning_rate": 0.00019371196754563896,
-      "loss": 1.2807598114013672,
-      "mean_token_accuracy": 0.7116438299417496,
-      "num_tokens": 1457023.0,
+      "entropy": 1.1915819495916367,
+      "epoch": 0.09567546880979717,
+      "grad_norm": 0.15018120408058167,
+      "learning_rate": 0.00012704081632653063,
+      "loss": 1.2230369567871093,
+      "mean_token_accuracy": 0.719833716750145,
+      "num_tokens": 1019312.0,
       "step": 250
     },
     {
-      "entropy": 1.2856510818004607,
-      "epoch": 0.2505420380631173,
-      "grad_norm": 0.06267844140529633,
-      "learning_rate": 0.00019303583502366465,
-      "loss": 1.3038443565368651,
-      "mean_token_accuracy": 0.7054769903421402,
-      "num_tokens": 1513975.0,
+      "entropy": 1.2996815636754036,
+      "epoch": 0.09950248756218906,
+      "grad_norm": 0.10838313400745392,
+      "learning_rate": 0.00013214285714285715,
+      "loss": 1.3142367362976075,
+      "mean_token_accuracy": 0.7013067752122879,
+      "num_tokens": 1062901.0,
       "step": 260
     },
     {
-      "entropy": 1.228683941066265,
-      "epoch": 0.26017827029631413,
-      "grad_norm": 0.0609310083091259,
-      "learning_rate": 0.00019235970250169033,
-      "loss": 1.2614136695861817,
-      "mean_token_accuracy": 0.7095189347863198,
-      "num_tokens": 1579252.0,
+      "entropy": 1.149668525904417,
+      "epoch": 0.10332950631458095,
+      "grad_norm": 0.09911312907934189,
+      "learning_rate": 0.00013724489795918367,
+      "loss": 1.1573083877563477,
+      "mean_token_accuracy": 0.728775355219841,
+      "num_tokens": 1102630.0,
       "step": 270
     },
     {
-      "entropy": 1.2551469817757606,
-      "epoch": 0.26981450252951095,
-      "grad_norm": 0.06873613595962524,
-      "learning_rate": 0.00019168356997971604,
-      "loss": 1.2714256286621093,
-      "mean_token_accuracy": 0.712159389257431,
-      "num_tokens": 1637995.0,
+      "entropy": 1.271340447664261,
+      "epoch": 0.10715652506697283,
+      "grad_norm": 0.09665267169475555,
+      "learning_rate": 0.00014234693877551022,
+      "loss": 1.341374111175537,
+      "mean_token_accuracy": 0.7027333110570908,
+      "num_tokens": 1142262.0,
       "step": 280
     },
     {
-      "entropy": 1.2518196165561677,
-      "epoch": 0.27945073476270776,
-      "grad_norm": 0.07062622159719467,
-      "learning_rate": 0.00019100743745774173,
-      "loss": 1.302776050567627,
-      "mean_token_accuracy": 0.7063210532069206,
-      "num_tokens": 1703008.0,
+      "entropy": 1.225122657418251,
+      "epoch": 0.11098354381936472,
+      "grad_norm": 0.13240815699100494,
+      "learning_rate": 0.00014744897959183674,
+      "loss": 1.2614737510681153,
+      "mean_token_accuracy": 0.7198452442884445,
+      "num_tokens": 1182386.0,
       "step": 290
     },
     {
-      "entropy": 1.2446896508336067,
-      "epoch": 0.2890869669959046,
-      "grad_norm": 0.051750048995018005,
-      "learning_rate": 0.00019033130493576744,
-      "loss": 1.3015289306640625,
-      "mean_token_accuracy": 0.7073144510388374,
-      "num_tokens": 1766072.0,
+      "entropy": 1.2733432039618493,
+      "epoch": 0.1148105625717566,
+      "grad_norm": 0.10651895403862,
+      "learning_rate": 0.00015255102040816326,
+      "loss": 1.2933347702026368,
+      "mean_token_accuracy": 0.7025195896625519,
+      "num_tokens": 1222805.0,
       "step": 300
     },
     {
-      "entropy": 1.3015983402729034,
-      "epoch": 0.2987231992291014,
-      "grad_norm": 0.0565313994884491,
-      "learning_rate": 0.00018965517241379312,
-      "loss": 1.311717414855957,
-      "mean_token_accuracy": 0.7029094457626343,
-      "num_tokens": 1820923.0,
+      "entropy": 1.1543171763420106,
+      "epoch": 0.11863758132414849,
+      "grad_norm": 0.08577804267406464,
+      "learning_rate": 0.00015765306122448978,
+      "loss": 1.197078323364258,
+      "mean_token_accuracy": 0.7300360783934593,
+      "num_tokens": 1263121.0,
       "step": 310
     },
     {
-      "entropy": 1.1577203705906869,
-      "epoch": 0.3083594314622982,
-      "grad_norm": 0.05471700802445412,
-      "learning_rate": 0.0001889790398918188,
-      "loss": 1.2251465797424317,
-      "mean_token_accuracy": 0.7273583576083184,
-      "num_tokens": 1877116.0,
+      "entropy": 1.1908820882439612,
+      "epoch": 0.12246460007654038,
+      "grad_norm": 0.11925600469112396,
+      "learning_rate": 0.00016275510204081633,
+      "loss": 1.2366827964782714,
+      "mean_token_accuracy": 0.7277334719896317,
+      "num_tokens": 1296346.0,
       "step": 320
     },
     {
-      "entropy": 1.205883078277111,
-      "epoch": 0.31799566369549503,
-      "grad_norm": 0.05748973786830902,
-      "learning_rate": 0.00018830290736984452,
-      "loss": 1.244644546508789,
-      "mean_token_accuracy": 0.7220181196928024,
-      "num_tokens": 1935606.0,
+      "entropy": 1.1711702406406403,
+      "epoch": 0.12629161882893225,
+      "grad_norm": 0.12476309388875961,
+      "learning_rate": 0.00016785714285714288,
+      "loss": 1.2408350944519042,
+      "mean_token_accuracy": 0.7282937213778495,
+      "num_tokens": 1335547.0,
       "step": 330
     },
     {
-      "entropy": 1.258717157691717,
-      "epoch": 0.3276318959286919,
-      "grad_norm": 0.05800570175051689,
-      "learning_rate": 0.0001876267748478702,
-      "loss": 1.3071101188659668,
-      "mean_token_accuracy": 0.7034722596406937,
-      "num_tokens": 1997199.0,
+      "entropy": 1.1748667433857918,
+      "epoch": 0.13011863758132414,
+      "grad_norm": 0.08671289682388306,
+      "learning_rate": 0.0001729591836734694,
+      "loss": 1.199030303955078,
+      "mean_token_accuracy": 0.7312668621540069,
+      "num_tokens": 1377093.0,
       "step": 340
     },
     {
-      "entropy": 1.22360208183527,
-      "epoch": 0.3372681281618887,
-      "grad_norm": 0.09994573146104813,
-      "learning_rate": 0.00018695064232589589,
-      "loss": 1.2296308517456054,
-      "mean_token_accuracy": 0.7146901577711106,
-      "num_tokens": 2048172.0,
+      "entropy": 1.153764547407627,
+      "epoch": 0.13394565633371602,
+      "grad_norm": 0.10536976903676987,
+      "learning_rate": 0.00017806122448979592,
+      "loss": 1.201906967163086,
+      "mean_token_accuracy": 0.7266524419188499,
+      "num_tokens": 1417236.0,
       "step": 350
     },
     {
-      "entropy": 1.3035957127809525,
-      "epoch": 0.34690436039508554,
-      "grad_norm": 0.07320819795131683,
-      "learning_rate": 0.00018627450980392157,
-      "loss": 1.2939067840576173,
-      "mean_token_accuracy": 0.696925450861454,
-      "num_tokens": 2109948.0,
+      "entropy": 1.2303058430552483,
+      "epoch": 0.1377726750861079,
+      "grad_norm": 0.09069176018238068,
+      "learning_rate": 0.00018316326530612247,
+      "loss": 1.2867681503295898,
+      "mean_token_accuracy": 0.7198954582214355,
+      "num_tokens": 1460306.0,
       "step": 360
     },
     {
-      "entropy": 1.212366634607315,
-      "epoch": 0.35654059262828236,
-      "grad_norm": 0.06324014812707901,
-      "learning_rate": 0.00018559837728194725,
-      "loss": 1.2521446228027344,
-      "mean_token_accuracy": 0.717373288422823,
-      "num_tokens": 2171174.0,
+      "entropy": 1.1944106668233871,
+      "epoch": 0.1415996938384998,
+      "grad_norm": 0.08539925515651703,
+      "learning_rate": 0.000188265306122449,
+      "loss": 1.2451179504394532,
+      "mean_token_accuracy": 0.7186401098966598,
+      "num_tokens": 1505966.0,
       "step": 370
     },
     {
-      "entropy": 1.240752936899662,
-      "epoch": 0.3661768248614792,
-      "grad_norm": 0.07091817259788513,
-      "learning_rate": 0.00018492224475997297,
-      "loss": 1.2615043640136718,
-      "mean_token_accuracy": 0.7065044552087784,
-      "num_tokens": 2228430.0,
+      "entropy": 1.2239032357931137,
+      "epoch": 0.14542671259089168,
+      "grad_norm": 0.08434446156024933,
+      "learning_rate": 0.0001933673469387755,
+      "loss": 1.2803629875183105,
+      "mean_token_accuracy": 0.7178368359804154,
+      "num_tokens": 1544132.0,
       "step": 380
     },
     {
-      "entropy": 1.2120654836297036,
-      "epoch": 0.375813057094676,
-      "grad_norm": 0.05131813511252403,
-      "learning_rate": 0.00018424611223799865,
-      "loss": 1.2185423851013184,
-      "mean_token_accuracy": 0.7187462538480759,
-      "num_tokens": 2285248.0,
+      "entropy": 1.1901779979467393,
+      "epoch": 0.14925373134328357,
+      "grad_norm": 0.08662886172533035,
+      "learning_rate": 0.00019846938775510203,
+      "loss": 1.2282370567321776,
+      "mean_token_accuracy": 0.7221132159233093,
+      "num_tokens": 1587355.0,
       "step": 390
     },
     {
-      "entropy": 1.2625371024012566,
-      "epoch": 0.3854492893278728,
-      "grad_norm": 0.056896600872278214,
-      "learning_rate": 0.00018356997971602433,
-      "loss": 1.3041479110717773,
-      "mean_token_accuracy": 0.7014564648270607,
-      "num_tokens": 2345584.0,
+      "entropy": 1.106569343805313,
+      "epoch": 0.15308075009567546,
+      "grad_norm": 0.13149450719356537,
+      "learning_rate": 0.00019981200483416142,
+      "loss": 1.136556625366211,
+      "mean_token_accuracy": 0.7384186327457428,
+      "num_tokens": 1624638.0,
       "step": 400
     },
     {
-      "entropy": 1.229997194558382,
-      "epoch": 0.39508552156106963,
-      "grad_norm": 0.06733101606369019,
-      "learning_rate": 0.00018289384719405005,
-      "loss": 1.2783479690551758,
-      "mean_token_accuracy": 0.7130212724208832,
-      "num_tokens": 2402073.0,
+      "entropy": 1.0393452920019626,
+      "epoch": 0.15690776884806737,
+      "grad_norm": 0.13831719756126404,
+      "learning_rate": 0.00019954344031153484,
+      "loss": 1.074817180633545,
+      "mean_token_accuracy": 0.7567356958985328,
+      "num_tokens": 1665215.0,
       "step": 410
     },
     {
-      "entropy": 1.253336711972952,
-      "epoch": 0.40472175379426645,
-      "grad_norm": 0.1017632707953453,
-      "learning_rate": 0.00018221771467207573,
-      "loss": 1.2827540397644044,
-      "mean_token_accuracy": 0.703055490553379,
-      "num_tokens": 2461269.0,
+      "entropy": 1.1298492863774299,
+      "epoch": 0.16073478760045926,
+      "grad_norm": 0.10244159400463104,
+      "learning_rate": 0.0001992748757889083,
+      "loss": 1.1741769790649415,
+      "mean_token_accuracy": 0.7414689466357232,
+      "num_tokens": 1701543.0,
       "step": 420
     },
     {
-      "entropy": 1.1999207064509392,
-      "epoch": 0.41435798602746327,
-      "grad_norm": 0.0675441101193428,
-      "learning_rate": 0.00018154158215010142,
-      "loss": 1.1980154037475585,
-      "mean_token_accuracy": 0.7192850261926651,
-      "num_tokens": 2520272.0,
+      "entropy": 1.1646860882639885,
+      "epoch": 0.16456180635285114,
+      "grad_norm": 0.09356453269720078,
+      "learning_rate": 0.00019900631126628174,
+      "loss": 1.2229989051818848,
+      "mean_token_accuracy": 0.7278152450919151,
+      "num_tokens": 1744719.0,
       "step": 430
     },
     {
-      "entropy": 1.1966832533478737,
-      "epoch": 0.4239942182606601,
-      "grad_norm": 0.04782295227050781,
-      "learning_rate": 0.00018086544962812713,
-      "loss": 1.196799373626709,
-      "mean_token_accuracy": 0.7202263355255127,
-      "num_tokens": 2574916.0,
+      "entropy": 1.1580850452184677,
+      "epoch": 0.16838882510524303,
+      "grad_norm": 0.08699047565460205,
+      "learning_rate": 0.00019873774674365518,
+      "loss": 1.1999470710754394,
+      "mean_token_accuracy": 0.7270642057061195,
+      "num_tokens": 1787999.0,
       "step": 440
     },
     {
-      "entropy": 1.28217963129282,
-      "epoch": 0.4336304504938569,
-      "grad_norm": 0.055622052401304245,
-      "learning_rate": 0.0001801893171061528,
-      "loss": 1.2780956268310546,
-      "mean_token_accuracy": 0.6991938084363938,
-      "num_tokens": 2630803.0,
+      "entropy": 1.105088683962822,
+      "epoch": 0.17221584385763491,
+      "grad_norm": 0.10489863902330399,
+      "learning_rate": 0.0001984691822210286,
+      "loss": 1.123628807067871,
+      "mean_token_accuracy": 0.7375599846243859,
+      "num_tokens": 1825171.0,
       "step": 450
     },
     {
-      "entropy": 1.1060978904366494,
-      "epoch": 0.4432666827270537,
-      "grad_norm": 0.08565858751535416,
-      "learning_rate": 0.00017951318458417852,
-      "loss": 1.1542920112609862,
-      "mean_token_accuracy": 0.7370428621768952,
-      "num_tokens": 2686122.0,
+      "entropy": 1.0744315460324287,
+      "epoch": 0.1760428626100268,
+      "grad_norm": 0.10170256346464157,
+      "learning_rate": 0.00019820061769840205,
+      "loss": 1.1449885368347168,
+      "mean_token_accuracy": 0.7466330513358116,
+      "num_tokens": 1863245.0,
       "step": 460
     },
     {
-      "entropy": 1.3041150823235512,
-      "epoch": 0.45290291496025054,
-      "grad_norm": 0.05283431336283684,
-      "learning_rate": 0.0001788370520622042,
-      "loss": 1.2778042793273925,
-      "mean_token_accuracy": 0.700305138528347,
-      "num_tokens": 2751119.0,
+      "entropy": 1.021899376064539,
+      "epoch": 0.1798698813624187,
+      "grad_norm": 0.09046658873558044,
+      "learning_rate": 0.0001979320531757755,
+      "loss": 1.0087275505065918,
+      "mean_token_accuracy": 0.7654190301895142,
+      "num_tokens": 1902205.0,
       "step": 470
     },
     {
-      "entropy": 1.2362390145659448,
-      "epoch": 0.46253914719344735,
-      "grad_norm": 0.05927009880542755,
-      "learning_rate": 0.0001781609195402299,
-      "loss": 1.2784390449523926,
-      "mean_token_accuracy": 0.700747960805893,
-      "num_tokens": 2809393.0,
+      "entropy": 1.1595304682850838,
+      "epoch": 0.18369690011481057,
+      "grad_norm": 0.09361740201711655,
+      "learning_rate": 0.00019766348865314892,
+      "loss": 1.2238757133483886,
+      "mean_token_accuracy": 0.7275362908840179,
+      "num_tokens": 1943827.0,
       "step": 480
     },
     {
-      "entropy": 1.2329043842852117,
-      "epoch": 0.47217537942664417,
-      "grad_norm": 0.05709557607769966,
-      "learning_rate": 0.0001774847870182556,
-      "loss": 1.2346397399902345,
-      "mean_token_accuracy": 0.7168422609567642,
-      "num_tokens": 2870246.0,
+      "entropy": 1.1023890599608421,
+      "epoch": 0.18752391886720246,
+      "grad_norm": 0.08471602201461792,
+      "learning_rate": 0.00019739492413052236,
+      "loss": 1.1567111015319824,
+      "mean_token_accuracy": 0.742167092859745,
+      "num_tokens": 1982630.0,
       "step": 490
     },
     {
-      "entropy": 1.197921334207058,
-      "epoch": 0.481811611659841,
-      "grad_norm": 0.051337700337171555,
-      "learning_rate": 0.0001768086544962813,
-      "loss": 1.2222982406616212,
-      "mean_token_accuracy": 0.7162339583039283,
-      "num_tokens": 2934671.0,
+      "entropy": 1.1427962884306908,
+      "epoch": 0.19135093761959435,
+      "grad_norm": 0.1170063391327858,
+      "learning_rate": 0.0001971263596078958,
+      "loss": 1.208080005645752,
+      "mean_token_accuracy": 0.7308674260973931,
+      "num_tokens": 2021311.0,
       "step": 500
     },
     {
-      "entropy": 1.2013787925243378,
-      "epoch": 0.4914478438930378,
-      "grad_norm": 0.06907735019922256,
-      "learning_rate": 0.00017613252197430697,
-      "loss": 1.2295709609985352,
-      "mean_token_accuracy": 0.7151955872774124,
-      "num_tokens": 2989014.0,
+      "entropy": 1.0002792343497275,
+      "epoch": 0.19517795637198623,
+      "grad_norm": 0.10567828267812729,
+      "learning_rate": 0.00019685779508526926,
+      "loss": 1.026076889038086,
+      "mean_token_accuracy": 0.7641370877623558,
+      "num_tokens": 2055396.0,
       "step": 510
     },
     {
-      "entropy": 1.1978842347860337,
-      "epoch": 0.5010840761262346,
-      "grad_norm": 0.06152976304292679,
-      "learning_rate": 0.00017545638945233268,
-      "loss": 1.2030101776123048,
-      "mean_token_accuracy": 0.7203953012824058,
-      "num_tokens": 3045999.0,
+      "entropy": 1.1096693962812423,
+      "epoch": 0.19900497512437812,
+      "grad_norm": 0.08597096055746078,
+      "learning_rate": 0.00019658923056264268,
+      "loss": 1.1631651878356934,
+      "mean_token_accuracy": 0.7389342650771141,
+      "num_tokens": 2097931.0,
       "step": 520
     },
     {
-      "entropy": 1.3108120203018188,
-      "epoch": 0.5107203083594315,
-      "grad_norm": 0.05314662307500839,
-      "learning_rate": 0.00017478025693035837,
-      "loss": 1.3376919746398925,
-      "mean_token_accuracy": 0.6901254534721375,
-      "num_tokens": 3106021.0,
+      "entropy": 1.04201333373785,
+      "epoch": 0.20283199387677,
+      "grad_norm": 0.1260094940662384,
+      "learning_rate": 0.00019632066604001613,
+      "loss": 1.0765873908996582,
+      "mean_token_accuracy": 0.7539548426866531,
+      "num_tokens": 2137657.0,
       "step": 530
     },
     {
-      "entropy": 1.269498337060213,
-      "epoch": 0.5203565405926283,
-      "grad_norm": 0.052769869565963745,
-      "learning_rate": 0.00017410412440838405,
-      "loss": 1.3347968101501464,
-      "mean_token_accuracy": 0.7055287733674049,
-      "num_tokens": 3167001.0,
+      "entropy": 1.0748623803257942,
+      "epoch": 0.2066590126291619,
+      "grad_norm": 0.0845552608370781,
+      "learning_rate": 0.00019605210151738955,
+      "loss": 1.1419748306274413,
+      "mean_token_accuracy": 0.7481048628687859,
+      "num_tokens": 2177343.0,
       "step": 540
     },
     {
-      "entropy": 1.174779912084341,
-      "epoch": 0.5299927728258251,
-      "grad_norm": 0.057087887078523636,
-      "learning_rate": 0.00017342799188640974,
-      "loss": 1.204134464263916,
-      "mean_token_accuracy": 0.7242644309997559,
-      "num_tokens": 3222794.0,
+      "entropy": 1.0970528617501258,
+      "epoch": 0.21048603138155378,
+      "grad_norm": 0.07105763256549835,
+      "learning_rate": 0.000195783536994763,
+      "loss": 1.1284755706787108,
+      "mean_token_accuracy": 0.747196614742279,
+      "num_tokens": 2211423.0,
       "step": 550
     },
     {
-      "entropy": 1.1798104658722877,
-      "epoch": 0.5396290050590219,
-      "grad_norm": 0.06867323815822601,
-      "learning_rate": 0.00017275185936443542,
-      "loss": 1.218485164642334,
-      "mean_token_accuracy": 0.7220648050308227,
-      "num_tokens": 3282856.0,
+      "entropy": 1.0551751986145974,
+      "epoch": 0.21431305013394567,
+      "grad_norm": 0.12569685280323029,
+      "learning_rate": 0.00019551497247213644,
+      "loss": 1.1170102119445802,
+      "mean_token_accuracy": 0.7472440049052238,
+      "num_tokens": 2249350.0,
       "step": 560
     },
     {
-      "entropy": 1.2600811369717122,
-      "epoch": 0.5492652372922188,
-      "grad_norm": 0.06002269685268402,
-      "learning_rate": 0.00017207572684246113,
-      "loss": 1.2951341629028321,
-      "mean_token_accuracy": 0.7052147269248963,
-      "num_tokens": 3343572.0,
+      "entropy": 1.0562219873070717,
+      "epoch": 0.21814006888633755,
+      "grad_norm": 0.08452208340167999,
+      "learning_rate": 0.0001952464079495099,
+      "loss": 1.0921730995178223,
+      "mean_token_accuracy": 0.7492805704474449,
+      "num_tokens": 2289564.0,
       "step": 570
     },
     {
-      "entropy": 1.2493580430746078,
-      "epoch": 0.5589014695254155,
-      "grad_norm": 0.06532762199640274,
-      "learning_rate": 0.00017139959432048682,
-      "loss": 1.2813149452209474,
-      "mean_token_accuracy": 0.7067271783947945,
-      "num_tokens": 3402401.0,
+      "entropy": 1.022915106266737,
+      "epoch": 0.22196708763872944,
+      "grad_norm": 0.08168510347604752,
+      "learning_rate": 0.00019497784342688333,
+      "loss": 1.064980697631836,
+      "mean_token_accuracy": 0.7562290355563164,
+      "num_tokens": 2332592.0,
       "step": 580
     },
     {
-      "entropy": 1.174679161608219,
-      "epoch": 0.5685377017586124,
-      "grad_norm": 0.0596516914665699,
-      "learning_rate": 0.0001707234617985125,
-      "loss": 1.2225926399230957,
-      "mean_token_accuracy": 0.7233842894434929,
-      "num_tokens": 3459042.0,
+      "entropy": 1.1041950330138206,
+      "epoch": 0.22579410639112132,
+      "grad_norm": 0.07596516609191895,
+      "learning_rate": 0.00019470927890425675,
+      "loss": 1.1410536766052246,
+      "mean_token_accuracy": 0.7350378915667534,
+      "num_tokens": 2380215.0,
       "step": 590
     },
     {
-      "entropy": 1.2576898023486138,
-      "epoch": 0.5781739339918092,
-      "grad_norm": 0.05735331028699875,
-      "learning_rate": 0.0001700473292765382,
-      "loss": 1.2835253715515136,
-      "mean_token_accuracy": 0.7045948460698128,
-      "num_tokens": 3527637.0,
+      "entropy": 1.1252056039869784,
+      "epoch": 0.2296211251435132,
+      "grad_norm": 0.07240597158670425,
+      "learning_rate": 0.0001944407143816302,
+      "loss": 1.1348044395446777,
+      "mean_token_accuracy": 0.7389790266752243,
+      "num_tokens": 2417819.0,
       "step": 600
     },
     {
-      "entropy": 1.2473611667752267,
-      "epoch": 0.587810166225006,
-      "grad_norm": 0.06076724827289581,
-      "learning_rate": 0.0001693711967545639,
-      "loss": 1.2636917114257813,
-      "mean_token_accuracy": 0.7045381426811218,
-      "num_tokens": 3589404.0,
+      "entropy": 1.0117394506931305,
+      "epoch": 0.2334481438959051,
+      "grad_norm": 0.08603253215551376,
+      "learning_rate": 0.00019417214985900362,
+      "loss": 1.0533303260803222,
+      "mean_token_accuracy": 0.7578112691640854,
+      "num_tokens": 2459072.0,
       "step": 610
     },
     {
-      "entropy": 1.2829708829522133,
-      "epoch": 0.5974463984582028,
-      "grad_norm": 0.05943462252616882,
-      "learning_rate": 0.0001686950642325896,
-      "loss": 1.2889452934265138,
-      "mean_token_accuracy": 0.7001010566949845,
-      "num_tokens": 3651124.0,
+      "entropy": 1.0193642653524875,
+      "epoch": 0.23727516264829698,
+      "grad_norm": 0.08400722593069077,
+      "learning_rate": 0.00019390358533637707,
+      "loss": 1.0955985069274903,
+      "mean_token_accuracy": 0.7598358646035195,
+      "num_tokens": 2497901.0,
       "step": 620
     },
     {
-      "entropy": 1.2208770334720611,
-      "epoch": 0.6070826306913997,
-      "grad_norm": 0.04695171117782593,
-      "learning_rate": 0.0001680189317106153,
-      "loss": 1.259203052520752,
-      "mean_token_accuracy": 0.7110297352075576,
-      "num_tokens": 3705293.0,
+      "entropy": 1.0488237984478475,
+      "epoch": 0.24110218140068887,
+      "grad_norm": 0.07221511751413345,
+      "learning_rate": 0.00019363502081375052,
+      "loss": 1.151495361328125,
+      "mean_token_accuracy": 0.7541669681668282,
+      "num_tokens": 2536530.0,
       "step": 630
     },
     {
-      "entropy": 1.301590697467327,
-      "epoch": 0.6167188629245964,
-      "grad_norm": 0.047615595161914825,
-      "learning_rate": 0.00016734279918864098,
-      "loss": 1.322781467437744,
-      "mean_token_accuracy": 0.6952215626835823,
-      "num_tokens": 3765429.0,
+      "entropy": 1.0595403373241425,
+      "epoch": 0.24492920015308076,
+      "grad_norm": 0.10258961468935013,
+      "learning_rate": 0.00019336645629112396,
+      "loss": 1.0888574600219727,
+      "mean_token_accuracy": 0.7474928990006446,
+      "num_tokens": 2571599.0,
       "step": 640
     },
     {
-      "entropy": 1.2000589437782765,
-      "epoch": 0.6263550951577933,
-      "grad_norm": 0.057772569358348846,
-      "learning_rate": 0.0001666666666666667,
-      "loss": 1.2448016166687013,
-      "mean_token_accuracy": 0.7192464172840118,
-      "num_tokens": 3826736.0,
+      "entropy": 1.0924376487731933,
+      "epoch": 0.24875621890547264,
+      "grad_norm": 0.0751282125711441,
+      "learning_rate": 0.0001930978917684974,
+      "loss": 1.1468082427978517,
+      "mean_token_accuracy": 0.7465573191642761,
+      "num_tokens": 2612401.0,
       "step": 650
     },
     {
-      "entropy": 1.187142127752304,
-      "epoch": 0.6359913273909901,
-      "grad_norm": 0.06310118734836578,
-      "learning_rate": 0.00016599053414469237,
-      "loss": 1.1989784240722656,
-      "mean_token_accuracy": 0.7233487412333488,
-      "num_tokens": 3886196.0,
+      "entropy": 0.9765479557216168,
+      "epoch": 0.2525832376578645,
+      "grad_norm": 0.09039046615362167,
+      "learning_rate": 0.00019282932724587083,
+      "loss": 1.054959201812744,
+      "mean_token_accuracy": 0.7649626806378365,
+      "num_tokens": 2652042.0,
       "step": 660
     },
     {
-      "entropy": 1.2190307170152663,
-      "epoch": 0.6456275596241869,
-      "grad_norm": 0.049476709216833115,
-      "learning_rate": 0.00016531440162271806,
-      "loss": 1.2834860801696777,
-      "mean_token_accuracy": 0.713593752682209,
-      "num_tokens": 3941611.0,
+      "entropy": 1.0833388939499855,
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.08905521035194397,
+      "learning_rate": 0.00019256076272324425,
+      "loss": 1.0937233924865724,
+      "mean_token_accuracy": 0.7510278865694999,
+      "num_tokens": 2692211.0,
       "step": 670
     },
     {
-      "entropy": 1.1955443389713765,
-      "epoch": 0.6552637918573838,
-      "grad_norm": 0.06877182424068451,
-      "learning_rate": 0.00016463826910074377,
-      "loss": 1.2530131340026855,
-      "mean_token_accuracy": 0.7211957752704621,
-      "num_tokens": 3998745.0,
+      "entropy": 1.0478161230683327,
+      "epoch": 0.2602372751626483,
+      "grad_norm": 0.09634676575660706,
+      "learning_rate": 0.0001922921982006177,
+      "loss": 1.1077991485595704,
+      "mean_token_accuracy": 0.7490576148033142,
+      "num_tokens": 2734562.0,
       "step": 680
     },
     {
-      "entropy": 1.255328443646431,
-      "epoch": 0.6649000240905806,
-      "grad_norm": 0.06499036401510239,
-      "learning_rate": 0.00016396213657876945,
-      "loss": 1.2712255477905274,
-      "mean_token_accuracy": 0.7075859263539315,
-      "num_tokens": 4057693.0,
+      "entropy": 1.13061283826828,
+      "epoch": 0.26406429391504016,
+      "grad_norm": 0.07757367938756943,
+      "learning_rate": 0.00019202363367799114,
+      "loss": 1.1330499649047852,
+      "mean_token_accuracy": 0.7350772902369499,
+      "num_tokens": 2778990.0,
       "step": 690
     },
     {
-      "entropy": 1.269506113231182,
-      "epoch": 0.6745362563237775,
-      "grad_norm": 0.07255974411964417,
-      "learning_rate": 0.00016328600405679514,
-      "loss": 1.294202709197998,
-      "mean_token_accuracy": 0.7018555745482444,
-      "num_tokens": 4113943.0,
+      "entropy": 1.0450415380299092,
+      "epoch": 0.26789131266743205,
+      "grad_norm": 0.06570328027009964,
+      "learning_rate": 0.0001917550691553646,
+      "loss": 1.1425201416015625,
+      "mean_token_accuracy": 0.7551864832639694,
+      "num_tokens": 2814475.0,
       "step": 700
     },
     {
-      "entropy": 1.24193774163723,
-      "epoch": 0.6841724885569742,
-      "grad_norm": 0.07056763768196106,
-      "learning_rate": 0.00016260987153482082,
-      "loss": 1.2850048065185546,
-      "mean_token_accuracy": 0.714261619746685,
-      "num_tokens": 4171377.0,
+      "entropy": 1.0114438571035862,
+      "epoch": 0.27171833141982393,
+      "grad_norm": 0.11020322889089584,
+      "learning_rate": 0.00019148650463273804,
+      "loss": 1.0655290603637695,
+      "mean_token_accuracy": 0.7622251763939858,
+      "num_tokens": 2847751.0,
       "step": 710
     },
     {
-      "entropy": 1.2012858629226684,
-      "epoch": 0.6938087207901711,
-      "grad_norm": 0.06263507157564163,
-      "learning_rate": 0.0001619337390128465,
-      "loss": 1.2244320869445802,
-      "mean_token_accuracy": 0.7127483233809471,
-      "num_tokens": 4224788.0,
+      "entropy": 1.1008106037974357,
+      "epoch": 0.2755453501722158,
+      "grad_norm": 0.07282241433858871,
+      "learning_rate": 0.00019121794011011146,
+      "loss": 1.139061450958252,
+      "mean_token_accuracy": 0.7368278667330742,
+      "num_tokens": 2889553.0,
       "step": 720
     },
     {
-      "entropy": 1.118230439722538,
-      "epoch": 0.7034449530233678,
-      "grad_norm": 0.0841294601559639,
-      "learning_rate": 0.00016125760649087222,
-      "loss": 1.1727646827697753,
-      "mean_token_accuracy": 0.7389020159840584,
-      "num_tokens": 4273260.0,
+      "entropy": 1.0155851803719997,
+      "epoch": 0.2793723689246077,
+      "grad_norm": 0.11799076199531555,
+      "learning_rate": 0.0001909493755874849,
+      "loss": 1.0634971618652345,
+      "mean_token_accuracy": 0.7578275159001351,
+      "num_tokens": 2926860.0,
       "step": 730
     },
     {
-      "entropy": 1.1662761889398099,
-      "epoch": 0.7130811852565647,
-      "grad_norm": 0.05674638971686363,
-      "learning_rate": 0.0001605814739688979,
-      "loss": 1.2040279388427735,
-      "mean_token_accuracy": 0.7247484371066093,
-      "num_tokens": 4330262.0,
+      "entropy": 1.0430648550391197,
+      "epoch": 0.2831993876769996,
+      "grad_norm": 0.08702066540718079,
+      "learning_rate": 0.00019068081106485832,
+      "loss": 1.0851760864257813,
+      "mean_token_accuracy": 0.7573095709085464,
+      "num_tokens": 2964029.0,
       "step": 740
     },
     {
-      "entropy": 1.1847384825348855,
-      "epoch": 0.7227174174897615,
-      "grad_norm": 0.0654911920428276,
-      "learning_rate": 0.00015990534144692359,
-      "loss": 1.245723056793213,
-      "mean_token_accuracy": 0.7201158210635186,
-      "num_tokens": 4387940.0,
+      "entropy": 1.0908455178141594,
+      "epoch": 0.2870264064293915,
+      "grad_norm": 0.06593967229127884,
+      "learning_rate": 0.00019041224654223177,
+      "loss": 1.0929256439208985,
+      "mean_token_accuracy": 0.7440102145075798,
+      "num_tokens": 3004528.0,
       "step": 750
     },
     {
-      "entropy": 1.2574817538261414,
-      "epoch": 0.7323536497229584,
-      "grad_norm": 0.05644860863685608,
-      "learning_rate": 0.0001592292089249493,
-      "loss": 1.2513961791992188,
-      "mean_token_accuracy": 0.7070513799786567,
-      "num_tokens": 4451167.0,
+      "entropy": 0.971546346694231,
+      "epoch": 0.29085342518178336,
+      "grad_norm": 0.08857332915067673,
+      "learning_rate": 0.00019014368201960522,
+      "loss": 1.0575197219848633,
+      "mean_token_accuracy": 0.7712547823786735,
+      "num_tokens": 3041203.0,
       "step": 760
     },
     {
-      "entropy": 1.1798933163285255,
-      "epoch": 0.7419898819561551,
-      "grad_norm": 0.04677167534828186,
-      "learning_rate": 0.00015855307640297498,
-      "loss": 1.2556601524353028,
-      "mean_token_accuracy": 0.7209865570068359,
-      "num_tokens": 4514662.0,
+      "entropy": 1.0496620319783687,
+      "epoch": 0.29468044393417525,
+      "grad_norm": 0.07172030210494995,
+      "learning_rate": 0.00018987511749697867,
+      "loss": 1.100113582611084,
+      "mean_token_accuracy": 0.747462597489357,
+      "num_tokens": 3086263.0,
       "step": 770
     },
     {
-      "entropy": 1.2833519145846366,
-      "epoch": 0.751626114189352,
-      "grad_norm": 0.05460638552904129,
-      "learning_rate": 0.0001578769438810007,
-      "loss": 1.3391889572143554,
-      "mean_token_accuracy": 0.699195285141468,
-      "num_tokens": 4577464.0,
+      "entropy": 1.0853942684829234,
+      "epoch": 0.29850746268656714,
+      "grad_norm": 0.0861373096704483,
+      "learning_rate": 0.0001896065529743521,
+      "loss": 1.1116449356079101,
+      "mean_token_accuracy": 0.7485424548387527,
+      "num_tokens": 3126741.0,
       "step": 780
     },
     {
-      "entropy": 1.301037323474884,
-      "epoch": 0.7612623464225488,
-      "grad_norm": 0.08067753911018372,
-      "learning_rate": 0.00015720081135902638,
-      "loss": 1.3280585289001465,
-      "mean_token_accuracy": 0.6965255320072175,
-      "num_tokens": 4640394.0,
+      "entropy": 1.039051755145192,
+      "epoch": 0.302334481438959,
+      "grad_norm": 0.07344193756580353,
+      "learning_rate": 0.00018933798845172553,
+      "loss": 1.092859935760498,
+      "mean_token_accuracy": 0.7524166733026505,
+      "num_tokens": 3164039.0,
       "step": 790
     },
     {
-      "entropy": 1.1654298216104508,
-      "epoch": 0.7708985786557456,
-      "grad_norm": 0.06938653439283371,
-      "learning_rate": 0.00015652467883705206,
-      "loss": 1.2016778945922852,
-      "mean_token_accuracy": 0.7259130507707596,
-      "num_tokens": 4694747.0,
+      "entropy": 1.021885236352682,
+      "epoch": 0.3061615001913509,
+      "grad_norm": 0.09843221306800842,
+      "learning_rate": 0.00018906942392909895,
+      "loss": 1.0813950538635253,
+      "mean_token_accuracy": 0.7612547591328621,
+      "num_tokens": 3202455.0,
       "step": 800
     },
     {
-      "entropy": 1.2972704201936722,
-      "epoch": 0.7805348108889424,
-      "grad_norm": 0.05404666066169739,
-      "learning_rate": 0.00015584854631507777,
-      "loss": 1.2715141296386718,
-      "mean_token_accuracy": 0.7007692798972129,
-      "num_tokens": 4754920.0,
+      "entropy": 1.0329275727272034,
+      "epoch": 0.3099885189437428,
+      "grad_norm": 0.07059452682733536,
+      "learning_rate": 0.0001888008594064724,
+      "loss": 1.0516587257385255,
+      "mean_token_accuracy": 0.7528441205620766,
+      "num_tokens": 3239911.0,
       "step": 810
     },
     {
-      "entropy": 1.191229759156704,
-      "epoch": 0.7901710431221393,
-      "grad_norm": 0.05804457888007164,
-      "learning_rate": 0.00015517241379310346,
-      "loss": 1.2225379943847656,
-      "mean_token_accuracy": 0.7155083760619163,
-      "num_tokens": 4813911.0,
+      "entropy": 1.0202949695289134,
+      "epoch": 0.31381553769613474,
+      "grad_norm": 0.07269048690795898,
+      "learning_rate": 0.00018853229488384585,
+      "loss": 1.0879244804382324,
+      "mean_token_accuracy": 0.7542849883437157,
+      "num_tokens": 3278682.0,
       "step": 820
     },
     {
-      "entropy": 1.1537679880857468,
-      "epoch": 0.799807275355336,
-      "grad_norm": 0.06406034529209137,
-      "learning_rate": 0.00015449628127112914,
-      "loss": 1.2089889526367188,
-      "mean_token_accuracy": 0.7305796332657337,
-      "num_tokens": 4867754.0,
+      "entropy": 1.0690899170935153,
+      "epoch": 0.3176425564485266,
+      "grad_norm": 0.14370054006576538,
+      "learning_rate": 0.0001882637303612193,
+      "loss": 1.1063778877258301,
+      "mean_token_accuracy": 0.75286915153265,
+      "num_tokens": 3325465.0,
       "step": 830
     },
     {
-      "entropy": 1.3313012823462487,
-      "epoch": 0.8094435075885329,
-      "grad_norm": 0.05331671983003616,
-      "learning_rate": 0.00015382014874915485,
-      "loss": 1.3202786445617676,
-      "mean_token_accuracy": 0.6896555438637734,
-      "num_tokens": 4932220.0,
+      "entropy": 1.0819261983036994,
+      "epoch": 0.3214695752009185,
+      "grad_norm": 0.0973975881934166,
+      "learning_rate": 0.00018799516583859274,
+      "loss": 1.0978353500366211,
+      "mean_token_accuracy": 0.749411192536354,
+      "num_tokens": 3363479.0,
       "step": 840
     },
     {
-      "entropy": 1.1428007125854491,
-      "epoch": 0.8190797398217297,
-      "grad_norm": 0.0543409064412117,
-      "learning_rate": 0.00015314401622718054,
-      "loss": 1.1924286842346192,
-      "mean_token_accuracy": 0.7276274234056472,
-      "num_tokens": 4991379.0,
+      "entropy": 1.0502549454569816,
+      "epoch": 0.3252965939533104,
+      "grad_norm": 0.11021706461906433,
+      "learning_rate": 0.0001877266013159662,
+      "loss": 1.1489330291748048,
+      "mean_token_accuracy": 0.7476440489292144,
+      "num_tokens": 3405863.0,
       "step": 850
     },
     {
-      "entropy": 1.1923618368804454,
-      "epoch": 0.8287159720549265,
-      "grad_norm": 0.04643765091896057,
-      "learning_rate": 0.00015246788370520625,
-      "loss": 1.2740966796875,
-      "mean_token_accuracy": 0.7193146347999573,
-      "num_tokens": 5050844.0,
+      "entropy": 1.1556663788855075,
+      "epoch": 0.3291236127057023,
+      "grad_norm": 0.06459799408912659,
+      "learning_rate": 0.0001874580367933396,
+      "loss": 1.1840539932250977,
+      "mean_token_accuracy": 0.7288095027208328,
+      "num_tokens": 3450829.0,
       "step": 860
     },
     {
-      "entropy": 1.2764008730649947,
-      "epoch": 0.8383522042881233,
-      "grad_norm": 0.06100849807262421,
-      "learning_rate": 0.00015179175118323193,
-      "loss": 1.2969582557678223,
-      "mean_token_accuracy": 0.7002251073718071,
-      "num_tokens": 5111937.0,
+      "entropy": 1.097336183488369,
+      "epoch": 0.33295063145809417,
+      "grad_norm": 0.06765513867139816,
+      "learning_rate": 0.00018718947227071303,
+      "loss": 1.1286226272583009,
+      "mean_token_accuracy": 0.7439226225018502,
+      "num_tokens": 3490640.0,
       "step": 870
     },
     {
-      "entropy": 1.1852023541927337,
-      "epoch": 0.8479884365213202,
-      "grad_norm": 0.08129674941301346,
-      "learning_rate": 0.00015111561866125762,
-      "loss": 1.215758991241455,
-      "mean_token_accuracy": 0.7221846386790276,
-      "num_tokens": 5173122.0,
+      "entropy": 1.0772622771561147,
+      "epoch": 0.33677765021048606,
+      "grad_norm": 0.08126482367515564,
+      "learning_rate": 0.00018692090774808648,
+      "loss": 1.1434885025024415,
+      "mean_token_accuracy": 0.7434220835566521,
+      "num_tokens": 3529438.0,
       "step": 880
     },
     {
-      "entropy": 1.2100131824612617,
-      "epoch": 0.857624668754517,
-      "grad_norm": 0.05021384358406067,
-      "learning_rate": 0.0001504394861392833,
-      "loss": 1.245238971710205,
-      "mean_token_accuracy": 0.7174671500921249,
-      "num_tokens": 5237440.0,
+      "entropy": 1.0091869838535785,
+      "epoch": 0.34060466896287794,
+      "grad_norm": 0.0654602199792862,
+      "learning_rate": 0.00018665234322545992,
+      "loss": 1.0767542839050293,
+      "mean_token_accuracy": 0.7644046351313591,
+      "num_tokens": 3565217.0,
       "step": 890
     },
     {
-      "entropy": 1.2959269508719444,
-      "epoch": 0.8672609009877138,
-      "grad_norm": 0.06317298859357834,
-      "learning_rate": 0.000149763353617309,
-      "loss": 1.298589324951172,
-      "mean_token_accuracy": 0.696705661714077,
-      "num_tokens": 5306311.0,
+      "entropy": 1.0432863399386405,
+      "epoch": 0.34443168771526983,
+      "grad_norm": 0.10025763511657715,
+      "learning_rate": 0.00018638377870283337,
+      "loss": 1.0826923370361328,
+      "mean_token_accuracy": 0.7591656729578972,
+      "num_tokens": 3603940.0,
       "step": 900
     },
     {
-      "entropy": 1.178437228500843,
-      "epoch": 0.8768971332209107,
-      "grad_norm": 0.0590016208589077,
-      "learning_rate": 0.00014908722109533467,
-      "loss": 1.2140485763549804,
-      "mean_token_accuracy": 0.7231360018253327,
-      "num_tokens": 5365204.0,
+      "entropy": 1.007722695171833,
+      "epoch": 0.3482587064676617,
+      "grad_norm": 0.06779270619153976,
+      "learning_rate": 0.00018611521418020682,
+      "loss": 1.0158637046813965,
+      "mean_token_accuracy": 0.763145099580288,
+      "num_tokens": 3644547.0,
       "step": 910
     },
     {
-      "entropy": 1.1817002773284913,
-      "epoch": 0.8865333654541074,
-      "grad_norm": 0.06499218195676804,
-      "learning_rate": 0.00014841108857336038,
-      "loss": 1.2365409851074218,
-      "mean_token_accuracy": 0.7207681879401207,
-      "num_tokens": 5425136.0,
+      "entropy": 1.0556264080107212,
+      "epoch": 0.3520857252200536,
+      "grad_norm": 0.07834554463624954,
+      "learning_rate": 0.00018584664965758026,
+      "loss": 1.091851806640625,
+      "mean_token_accuracy": 0.7483858004212379,
+      "num_tokens": 3691979.0,
       "step": 920
     },
     {
-      "entropy": 1.1871344536542892,
-      "epoch": 0.8961695976873043,
-      "grad_norm": 0.05251992866396904,
-      "learning_rate": 0.00014773495605138607,
-      "loss": 1.216776180267334,
-      "mean_token_accuracy": 0.7216181293129921,
-      "num_tokens": 5485557.0,
+      "entropy": 1.0730156242847442,
+      "epoch": 0.3559127439724455,
+      "grad_norm": 0.10772417485713959,
+      "learning_rate": 0.00018557808513495368,
+      "loss": 1.1370153427124023,
+      "mean_token_accuracy": 0.7466916054487228,
+      "num_tokens": 3728767.0,
       "step": 930
     },
     {
-      "entropy": 1.2011874541640282,
-      "epoch": 0.9058058299205011,
-      "grad_norm": 0.05413221940398216,
-      "learning_rate": 0.00014705882352941178,
-      "loss": 1.2213294982910157,
-      "mean_token_accuracy": 0.7150228247046471,
-      "num_tokens": 5545105.0,
+      "entropy": 1.081335111707449,
+      "epoch": 0.3597397627248374,
+      "grad_norm": 0.07669705897569656,
+      "learning_rate": 0.0001853095206123271,
+      "loss": 1.141366958618164,
+      "mean_token_accuracy": 0.744081811606884,
+      "num_tokens": 3772234.0,
       "step": 940
     },
     {
-      "entropy": 1.2181889459490776,
-      "epoch": 0.9154420621536979,
-      "grad_norm": 0.056055545806884766,
-      "learning_rate": 0.00014638269100743746,
-      "loss": 1.25078125,
-      "mean_token_accuracy": 0.7119832545518875,
-      "num_tokens": 5605134.0,
+      "entropy": 0.9984517656266689,
+      "epoch": 0.36356678147722926,
+      "grad_norm": 0.0695272758603096,
+      "learning_rate": 0.00018504095608970055,
+      "loss": 1.0501303672790527,
+      "mean_token_accuracy": 0.7590280339121819,
+      "num_tokens": 3816970.0,
       "step": 950
     },
     {
-      "entropy": 1.1678502529859542,
-      "epoch": 0.9250782943868947,
-      "grad_norm": 0.05342623591423035,
-      "learning_rate": 0.00014570655848546315,
-      "loss": 1.2045835494995116,
-      "mean_token_accuracy": 0.7228824377059937,
-      "num_tokens": 5663877.0,
+      "entropy": 0.910194194689393,
+      "epoch": 0.36739380022962115,
+      "grad_norm": 0.06411932408809662,
+      "learning_rate": 0.000184772391567074,
+      "loss": 0.9656248092651367,
+      "mean_token_accuracy": 0.7823562085628509,
+      "num_tokens": 3853816.0,
       "step": 960
     },
     {
-      "entropy": 1.1557280227541924,
-      "epoch": 0.9347145266200916,
-      "grad_norm": 0.056248344480991364,
-      "learning_rate": 0.00014503042596348886,
-      "loss": 1.175156021118164,
-      "mean_token_accuracy": 0.7239007547497749,
-      "num_tokens": 5726645.0,
+      "entropy": 0.9763211451470852,
+      "epoch": 0.37122081898201303,
+      "grad_norm": 0.08389662951231003,
+      "learning_rate": 0.00018450382704444744,
+      "loss": 1.0703671455383301,
+      "mean_token_accuracy": 0.76742093116045,
+      "num_tokens": 3896404.0,
       "step": 970
     },
     {
-      "entropy": 1.2256763622164726,
-      "epoch": 0.9443507588532883,
-      "grad_norm": 0.06137476861476898,
-      "learning_rate": 0.00014435429344151454,
-      "loss": 1.234114933013916,
-      "mean_token_accuracy": 0.7094191774725914,
-      "num_tokens": 5783468.0,
+      "entropy": 1.076941692829132,
+      "epoch": 0.3750478377344049,
+      "grad_norm": 0.13239043951034546,
+      "learning_rate": 0.0001842352625218209,
+      "loss": 1.1353830337524413,
+      "mean_token_accuracy": 0.7479456245899201,
+      "num_tokens": 3934187.0,
       "step": 980
     },
     {
-      "entropy": 1.1099035568535327,
-      "epoch": 0.9539869910864852,
-      "grad_norm": 0.052310869097709656,
-      "learning_rate": 0.00014367816091954023,
-      "loss": 1.158639907836914,
-      "mean_token_accuracy": 0.7366219267249108,
-      "num_tokens": 5840101.0,
+      "entropy": 1.077423833310604,
+      "epoch": 0.3788748564867968,
+      "grad_norm": 0.06203702092170715,
+      "learning_rate": 0.00018396669799919434,
+      "loss": 1.1363765716552734,
+      "mean_token_accuracy": 0.7437105163931846,
+      "num_tokens": 3975766.0,
       "step": 990
     },
     {
-      "entropy": 1.1962413311004638,
-      "epoch": 0.963623223319682,
-      "grad_norm": 0.055140018463134766,
-      "learning_rate": 0.00014300202839756594,
-      "loss": 1.210787010192871,
-      "mean_token_accuracy": 0.7173429980874062,
-      "num_tokens": 5903466.0,
+      "entropy": 1.009491826593876,
+      "epoch": 0.3827018752391887,
+      "grad_norm": 0.06740409135818481,
+      "learning_rate": 0.00018369813347656776,
+      "loss": 1.0752355575561523,
+      "mean_token_accuracy": 0.7598015293478966,
+      "num_tokens": 4018368.0,
       "step": 1000
     },
     {
-      "entropy": 1.0866067253053189,
-      "epoch": 0.9732594555528788,
-      "grad_norm": 0.07032942771911621,
-      "learning_rate": 0.00014232589587559162,
-      "loss": 1.1129253387451172,
-      "mean_token_accuracy": 0.7413682281970978,
-      "num_tokens": 5958697.0,
+      "entropy": 0.9744428530335426,
+      "epoch": 0.3865288939915806,
+      "grad_norm": 0.07750537246465683,
+      "learning_rate": 0.00018342956895394118,
+      "loss": 1.0554892539978027,
+      "mean_token_accuracy": 0.7682438552379608,
+      "num_tokens": 4057647.0,
       "step": 1010
     },
     {
-      "entropy": 1.1695112690329552,
-      "epoch": 0.9828956877860756,
-      "grad_norm": 0.06965778768062592,
-      "learning_rate": 0.00014164976335361734,
-      "loss": 1.2002062797546387,
-      "mean_token_accuracy": 0.7275112867355347,
-      "num_tokens": 6018820.0,
+      "entropy": 1.0335246473550797,
+      "epoch": 0.39035591274397247,
+      "grad_norm": 0.07627248764038086,
+      "learning_rate": 0.00018316100443131463,
+      "loss": 1.0600407600402832,
+      "mean_token_accuracy": 0.7552958622574806,
+      "num_tokens": 4098377.0,
       "step": 1020
     },
     {
-      "entropy": 1.129349136352539,
-      "epoch": 0.9925319200192725,
-      "grad_norm": 0.0639411062002182,
-      "learning_rate": 0.00014097363083164302,
-      "loss": 1.1613442420959472,
-      "mean_token_accuracy": 0.7342579141259193,
-      "num_tokens": 6074337.0,
+      "entropy": 1.0256185740232469,
+      "epoch": 0.39418293149636435,
+      "grad_norm": 0.10117889940738678,
+      "learning_rate": 0.00018289243990868807,
+      "loss": 1.0706727027893066,
+      "mean_token_accuracy": 0.75880047082901,
+      "num_tokens": 4141633.0,
       "step": 1030
     },
     {
-      "epoch": 1.0,
-      "eval_entropy": 1.0300818726266783,
-      "eval_loss": 1.032158374786377,
-      "eval_mean_token_accuracy": 0.7592671683222431,
-      "eval_num_tokens": 6115944.0,
-      "eval_runtime": 513.4443,
-      "eval_samples_per_second": 1.7,
-      "eval_steps_per_second": 1.7,
-      "step": 1038
+      "entropy": 0.9883378148078918,
+      "epoch": 0.39800995024875624,
+      "grad_norm": 0.064593605697155,
+      "learning_rate": 0.00018262387538606152,
+      "loss": 1.007016372680664,
+      "mean_token_accuracy": 0.764974731206894,
+      "num_tokens": 4181155.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0692595109343528,
+      "epoch": 0.4018369690011481,
+      "grad_norm": 0.07493151724338531,
+      "learning_rate": 0.00018235531086343497,
+      "loss": 1.123647975921631,
+      "mean_token_accuracy": 0.7452841177582741,
+      "num_tokens": 4218175.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.990117172151804,
+      "epoch": 0.40566398775354,
+      "grad_norm": 0.06332839280366898,
+      "learning_rate": 0.0001820867463408084,
+      "loss": 1.0538661003112793,
+      "mean_token_accuracy": 0.7637043848633767,
+      "num_tokens": 4262326.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.002436650544405,
+      "epoch": 0.4094910065059319,
+      "grad_norm": 0.07898294180631638,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 0.9973239898681641,
+      "mean_token_accuracy": 0.7644759714603424,
+      "num_tokens": 4300876.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9635432817041873,
+      "epoch": 0.4133180252583238,
+      "grad_norm": 0.09760674089193344,
+      "learning_rate": 0.00018154961729555525,
+      "loss": 1.0411369323730468,
+      "mean_token_accuracy": 0.7664693981409073,
+      "num_tokens": 4338887.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9780610945075751,
+      "epoch": 0.41714504401071567,
+      "grad_norm": 0.08076441287994385,
+      "learning_rate": 0.0001812810527729287,
+      "loss": 1.0544751167297364,
+      "mean_token_accuracy": 0.76624975502491,
+      "num_tokens": 4380678.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.0548741944134234,
+      "epoch": 0.42097206276310756,
+      "grad_norm": 0.0646439641714096,
+      "learning_rate": 0.00018101248825030215,
+      "loss": 1.128230667114258,
+      "mean_token_accuracy": 0.7504925444722176,
+      "num_tokens": 4422899.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.0767812803387642,
+      "epoch": 0.42479908151549944,
+      "grad_norm": 0.06994366645812988,
+      "learning_rate": 0.0001807439237276756,
+      "loss": 1.1209583282470703,
+      "mean_token_accuracy": 0.7477620646357537,
+      "num_tokens": 4461864.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.062944334745407,
+      "epoch": 0.42862610026789133,
+      "grad_norm": 0.11016593873500824,
+      "learning_rate": 0.00018047535920504904,
+      "loss": 1.0880105018615722,
+      "mean_token_accuracy": 0.7455767750740051,
+      "num_tokens": 4501378.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0511136516928672,
+      "epoch": 0.4324531190202832,
+      "grad_norm": 0.08707646280527115,
+      "learning_rate": 0.00018020679468242246,
+      "loss": 1.0764313697814942,
+      "mean_token_accuracy": 0.7519838035106658,
+      "num_tokens": 4541448.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9529998056590557,
+      "epoch": 0.4362801377726751,
+      "grad_norm": 0.07353853434324265,
+      "learning_rate": 0.00017993823015979588,
+      "loss": 1.0098756790161132,
+      "mean_token_accuracy": 0.7720077604055404,
+      "num_tokens": 4586147.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.143844011425972,
+      "epoch": 0.440107156525067,
+      "grad_norm": 0.06268489360809326,
+      "learning_rate": 0.00017966966563716933,
+      "loss": 1.1934361457824707,
+      "mean_token_accuracy": 0.7281116575002671,
+      "num_tokens": 4631906.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0761573910713196,
+      "epoch": 0.4439341752774589,
+      "grad_norm": 0.07078517228364944,
+      "learning_rate": 0.00017940110111454278,
+      "loss": 1.1359615325927734,
+      "mean_token_accuracy": 0.7409780561923981,
+      "num_tokens": 4674626.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9940036550164223,
+      "epoch": 0.44776119402985076,
+      "grad_norm": 0.08054502308368683,
+      "learning_rate": 0.00017913253659191622,
+      "loss": 1.033839225769043,
+      "mean_token_accuracy": 0.7682256817817688,
+      "num_tokens": 4715717.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9752239182591438,
+      "epoch": 0.45158821278224265,
+      "grad_norm": 0.08600450307130814,
+      "learning_rate": 0.00017886397206928967,
+      "loss": 1.0254844665527343,
+      "mean_token_accuracy": 0.7688430979847908,
+      "num_tokens": 4747316.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.064694558084011,
+      "epoch": 0.45541523153463453,
+      "grad_norm": 0.07270248234272003,
+      "learning_rate": 0.0001785954075466631,
+      "loss": 1.0806646347045898,
+      "mean_token_accuracy": 0.7534265503287315,
+      "num_tokens": 4788606.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.958549628406763,
+      "epoch": 0.4592422502870264,
+      "grad_norm": 0.0644846111536026,
+      "learning_rate": 0.00017832684302403654,
+      "loss": 1.0015847206115722,
+      "mean_token_accuracy": 0.7644492238759995,
+      "num_tokens": 4831371.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0885677203536033,
+      "epoch": 0.4630692690394183,
+      "grad_norm": 0.13487283885478973,
+      "learning_rate": 0.00017805827850140996,
+      "loss": 1.1495524406433106,
+      "mean_token_accuracy": 0.7414823487401009,
+      "num_tokens": 4871231.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.1117899976670742,
+      "epoch": 0.4668962877918102,
+      "grad_norm": 0.08015701174736023,
+      "learning_rate": 0.0001777897139787834,
+      "loss": 1.1366958618164062,
+      "mean_token_accuracy": 0.7351289570331574,
+      "num_tokens": 4911520.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9722193017601967,
+      "epoch": 0.4707233065442021,
+      "grad_norm": 0.06839531660079956,
+      "learning_rate": 0.00017752114945615685,
+      "loss": 1.0259140968322753,
+      "mean_token_accuracy": 0.7658233359456063,
+      "num_tokens": 4950296.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0021446757018566,
+      "epoch": 0.47455032529659397,
+      "grad_norm": 0.08231978863477707,
+      "learning_rate": 0.0001772525849335303,
+      "loss": 1.0437036514282227,
+      "mean_token_accuracy": 0.7644398525357247,
+      "num_tokens": 4989688.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9640353135764599,
+      "epoch": 0.47837734404898585,
+      "grad_norm": 0.11587074398994446,
+      "learning_rate": 0.00017698402041090375,
+      "loss": 1.0072126388549805,
+      "mean_token_accuracy": 0.7740294471383095,
+      "num_tokens": 5029135.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0122342824935913,
+      "epoch": 0.48220436280137774,
+      "grad_norm": 0.07646426558494568,
+      "learning_rate": 0.00017671545588827717,
+      "loss": 1.0733034133911132,
+      "mean_token_accuracy": 0.7619047269225121,
+      "num_tokens": 5066488.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0465880073606968,
+      "epoch": 0.4860313815537696,
+      "grad_norm": 0.07594821602106094,
+      "learning_rate": 0.0001764468913656506,
+      "loss": 1.0953254699707031,
+      "mean_token_accuracy": 0.7511951208114624,
+      "num_tokens": 5102103.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0104024082422256,
+      "epoch": 0.4898584003061615,
+      "grad_norm": 0.07695835083723068,
+      "learning_rate": 0.00017617832684302403,
+      "loss": 1.1025714874267578,
+      "mean_token_accuracy": 0.7583977058529854,
+      "num_tokens": 5141113.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.044089037179947,
+      "epoch": 0.4936854190585534,
+      "grad_norm": 0.07186906039714813,
+      "learning_rate": 0.00017590976232039748,
+      "loss": 1.0713683128356934,
+      "mean_token_accuracy": 0.7546971932053566,
+      "num_tokens": 5181454.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9154780797660351,
+      "epoch": 0.4975124378109453,
+      "grad_norm": 0.08934911340475082,
+      "learning_rate": 0.00017564119779777093,
+      "loss": 0.9871469497680664,
+      "mean_token_accuracy": 0.7756836161017417,
+      "num_tokens": 5214143.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.05635926425457,
+      "epoch": 0.5013394565633371,
+      "grad_norm": 0.07880513370037079,
+      "learning_rate": 0.00017537263327514437,
+      "loss": 1.0985527038574219,
+      "mean_token_accuracy": 0.7524559125304222,
+      "num_tokens": 5258310.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.0448619149625302,
+      "epoch": 0.505166475315729,
+      "grad_norm": 0.10507462918758392,
+      "learning_rate": 0.0001751040687525178,
+      "loss": 1.1040778160095215,
+      "mean_token_accuracy": 0.7477249845862388,
+      "num_tokens": 5296865.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.0706947155296802,
+      "epoch": 0.5089934940681209,
+      "grad_norm": 0.09437765926122665,
+      "learning_rate": 0.00017483550422989124,
+      "loss": 1.1372867584228517,
+      "mean_token_accuracy": 0.7502224639058113,
+      "num_tokens": 5335301.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9736435614526272,
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.07162626087665558,
+      "learning_rate": 0.0001745669397072647,
+      "loss": 1.0273897171020507,
+      "mean_token_accuracy": 0.7711918234825135,
+      "num_tokens": 5372533.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9989161014556884,
+      "epoch": 0.5166475315729047,
+      "grad_norm": 0.08805254101753235,
+      "learning_rate": 0.0001742983751846381,
+      "loss": 1.0603778839111329,
+      "mean_token_accuracy": 0.7600028276443481,
+      "num_tokens": 5412651.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.063752220571041,
+      "epoch": 0.5204745503252965,
+      "grad_norm": 0.08056829869747162,
+      "learning_rate": 0.00017402981066201156,
+      "loss": 1.0876687049865723,
+      "mean_token_accuracy": 0.7487231969833374,
+      "num_tokens": 5454518.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.966426993906498,
+      "epoch": 0.5243015690776884,
+      "grad_norm": 0.06970727443695068,
+      "learning_rate": 0.000173761246139385,
+      "loss": 1.0302441596984864,
+      "mean_token_accuracy": 0.7640074551105499,
+      "num_tokens": 5495257.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9727898858487606,
+      "epoch": 0.5281285878300803,
+      "grad_norm": 0.09694326668977737,
+      "learning_rate": 0.00017349268161675842,
+      "loss": 1.0327792167663574,
+      "mean_token_accuracy": 0.7687779292464256,
+      "num_tokens": 5527573.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.043993879854679,
+      "epoch": 0.5319556065824722,
+      "grad_norm": 0.05676735192537308,
+      "learning_rate": 0.00017322411709413187,
+      "loss": 1.1139988899230957,
+      "mean_token_accuracy": 0.7597961351275444,
+      "num_tokens": 5566542.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9891408108174801,
+      "epoch": 0.5357826253348641,
+      "grad_norm": 0.08670998364686966,
+      "learning_rate": 0.00017295555257150532,
+      "loss": 1.0878351211547852,
+      "mean_token_accuracy": 0.7640718072652817,
+      "num_tokens": 5604986.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0097288101911546,
+      "epoch": 0.539609644087256,
+      "grad_norm": 0.09190856665372849,
+      "learning_rate": 0.00017268698804887876,
+      "loss": 1.079444408416748,
+      "mean_token_accuracy": 0.7590912491083145,
+      "num_tokens": 5642224.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9927844725549221,
+      "epoch": 0.5434366628396479,
+      "grad_norm": 0.08191007375717163,
+      "learning_rate": 0.00017241842352625218,
+      "loss": 1.0661033630371093,
+      "mean_token_accuracy": 0.7664914444088936,
+      "num_tokens": 5680912.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.973179691657424,
+      "epoch": 0.5472636815920398,
+      "grad_norm": 0.08161566406488419,
+      "learning_rate": 0.00017214985900362563,
+      "loss": 1.078667163848877,
+      "mean_token_accuracy": 0.7686992704868316,
+      "num_tokens": 5717171.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0467095457017421,
+      "epoch": 0.5510907003444316,
+      "grad_norm": 0.09403429925441742,
+      "learning_rate": 0.00017188129448099908,
+      "loss": 1.0912303924560547,
+      "mean_token_accuracy": 0.7550861686468124,
+      "num_tokens": 5755956.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0082954704761504,
+      "epoch": 0.5549177190968235,
+      "grad_norm": 0.09858231991529465,
+      "learning_rate": 0.0001716127299583725,
+      "loss": 1.0449023246765137,
+      "mean_token_accuracy": 0.7586705282330513,
+      "num_tokens": 5798765.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0528397418558597,
+      "epoch": 0.5587447378492154,
+      "grad_norm": 0.06697855144739151,
+      "learning_rate": 0.00017134416543574594,
+      "loss": 1.0901053428649903,
+      "mean_token_accuracy": 0.7517547190189362,
+      "num_tokens": 5839833.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.009619940817356,
+      "epoch": 0.5625717566016073,
+      "grad_norm": 0.07271189987659454,
+      "learning_rate": 0.0001710756009131194,
+      "loss": 1.0171070098876953,
+      "mean_token_accuracy": 0.7594649389386177,
+      "num_tokens": 5880613.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9699329622089863,
+      "epoch": 0.5663987753539992,
+      "grad_norm": 0.07800697535276413,
+      "learning_rate": 0.0001708070363904928,
+      "loss": 1.1077412605285644,
+      "mean_token_accuracy": 0.7667307928204536,
+      "num_tokens": 5918218.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.9389957278966904,
+      "epoch": 0.5702257941063911,
+      "grad_norm": 0.08150342851877213,
+      "learning_rate": 0.00017053847186786626,
+      "loss": 0.9634763717651367,
+      "mean_token_accuracy": 0.7806087970733643,
+      "num_tokens": 5960284.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0140163496136665,
+      "epoch": 0.574052812858783,
+      "grad_norm": 0.06430503726005554,
+      "learning_rate": 0.0001702699073452397,
+      "loss": 1.0751851081848145,
+      "mean_token_accuracy": 0.7564210310578346,
+      "num_tokens": 6000562.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.047791599482298,
+      "epoch": 0.5778798316111748,
+      "grad_norm": 0.07922326028347015,
+      "learning_rate": 0.00017000134282261313,
+      "loss": 1.1397698402404786,
+      "mean_token_accuracy": 0.7494736298918724,
+      "num_tokens": 6045192.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.1085532158613205,
+      "epoch": 0.5817068503635667,
+      "grad_norm": 0.1093953400850296,
+      "learning_rate": 0.00016973277829998657,
+      "loss": 1.142368698120117,
+      "mean_token_accuracy": 0.7414237394928932,
+      "num_tokens": 6090148.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.9427969709038735,
+      "epoch": 0.5855338691159586,
+      "grad_norm": 0.09579843282699585,
+      "learning_rate": 0.00016946421377736002,
+      "loss": 0.9980224609375,
+      "mean_token_accuracy": 0.7730853497982025,
+      "num_tokens": 6129845.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0571323171257974,
+      "epoch": 0.5893608878683505,
+      "grad_norm": 0.09482655674219131,
+      "learning_rate": 0.00016919564925473347,
+      "loss": 1.0665513038635255,
+      "mean_token_accuracy": 0.749831511080265,
+      "num_tokens": 6171961.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.978803563863039,
+      "epoch": 0.5931879066207424,
+      "grad_norm": 0.08609842509031296,
+      "learning_rate": 0.0001689270847321069,
+      "loss": 1.0541341781616211,
+      "mean_token_accuracy": 0.769312071800232,
+      "num_tokens": 6210126.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0714900024235248,
+      "epoch": 0.5970149253731343,
+      "grad_norm": 0.05390879884362221,
+      "learning_rate": 0.00016865852020948033,
+      "loss": 1.1057682037353516,
+      "mean_token_accuracy": 0.7402923837304115,
+      "num_tokens": 6262394.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.9234071888029576,
+      "epoch": 0.6008419441255262,
+      "grad_norm": 0.09692159295082092,
+      "learning_rate": 0.00016838995568685378,
+      "loss": 0.9622941017150879,
+      "mean_token_accuracy": 0.7832354381680489,
+      "num_tokens": 6292140.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.9591108359396457,
+      "epoch": 0.604668962877918,
+      "grad_norm": 0.0720645934343338,
+      "learning_rate": 0.0001681213911642272,
+      "loss": 1.019169235229492,
+      "mean_token_accuracy": 0.771478471159935,
+      "num_tokens": 6332134.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0945825845003128,
+      "epoch": 0.6084959816303099,
+      "grad_norm": 0.07380460202693939,
+      "learning_rate": 0.00016785282664160065,
+      "loss": 1.1463271141052247,
+      "mean_token_accuracy": 0.7419712334871292,
+      "num_tokens": 6373967.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.055589073896408,
+      "epoch": 0.6123230003827018,
+      "grad_norm": 0.07209772616624832,
+      "learning_rate": 0.0001675842621189741,
+      "loss": 1.1144783973693848,
+      "mean_token_accuracy": 0.7461996227502823,
+      "num_tokens": 6418236.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.1149650782346725,
+      "epoch": 0.6161500191350937,
+      "grad_norm": 0.07935164868831635,
+      "learning_rate": 0.00016731569759634754,
+      "loss": 1.1727729797363282,
+      "mean_token_accuracy": 0.7365738078951836,
+      "num_tokens": 6464589.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0068970195949078,
+      "epoch": 0.6199770378874856,
+      "grad_norm": 0.0804886594414711,
+      "learning_rate": 0.00016704713307372096,
+      "loss": 1.0483062744140625,
+      "mean_token_accuracy": 0.7632505163550377,
+      "num_tokens": 6504385.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.9996877416968346,
+      "epoch": 0.6238040566398775,
+      "grad_norm": 0.0723455473780632,
+      "learning_rate": 0.0001667785685510944,
+      "loss": 1.0592655181884765,
+      "mean_token_accuracy": 0.7597218692302704,
+      "num_tokens": 6545928.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.9159566629678011,
+      "epoch": 0.6276310753922695,
+      "grad_norm": 0.08028513193130493,
+      "learning_rate": 0.00016651000402846783,
+      "loss": 0.9434080123901367,
+      "mean_token_accuracy": 0.7826011970639228,
+      "num_tokens": 6586166.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0089009895920753,
+      "epoch": 0.6314580941446614,
+      "grad_norm": 0.09154181182384491,
+      "learning_rate": 0.00016624143950584128,
+      "loss": 1.0339744567871094,
+      "mean_token_accuracy": 0.7604109585285187,
+      "num_tokens": 6624706.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.0208431974053382,
+      "epoch": 0.6352851128970533,
+      "grad_norm": 0.08039630204439163,
+      "learning_rate": 0.00016597287498321472,
+      "loss": 1.0823830604553222,
+      "mean_token_accuracy": 0.7548153042793274,
+      "num_tokens": 6665626.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.9645413011312485,
+      "epoch": 0.6391121316494451,
+      "grad_norm": 0.08834270387887955,
+      "learning_rate": 0.00016570431046058817,
+      "loss": 1.0401340484619142,
+      "mean_token_accuracy": 0.7703565835952759,
+      "num_tokens": 6699532.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0028597339987755,
+      "epoch": 0.642939150401837,
+      "grad_norm": 0.08974612504243851,
+      "learning_rate": 0.00016543574593796162,
+      "loss": 1.0680928230285645,
+      "mean_token_accuracy": 0.7617413088679313,
+      "num_tokens": 6740574.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.8952145710587501,
+      "epoch": 0.6467661691542289,
+      "grad_norm": 0.09289242327213287,
+      "learning_rate": 0.00016516718141533504,
+      "loss": 0.9696210861206055,
+      "mean_token_accuracy": 0.7848611980676651,
+      "num_tokens": 6782208.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.9520700328052044,
+      "epoch": 0.6505931879066208,
+      "grad_norm": 0.07298107445240021,
+      "learning_rate": 0.00016489861689270848,
+      "loss": 0.9891908645629883,
+      "mean_token_accuracy": 0.7725306749343872,
+      "num_tokens": 6818937.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.9734554067254066,
+      "epoch": 0.6544202066590127,
+      "grad_norm": 0.08233233541250229,
+      "learning_rate": 0.0001646300523700819,
+      "loss": 1.0311893463134765,
+      "mean_token_accuracy": 0.7683428943157196,
+      "num_tokens": 6851548.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.9947349905967713,
+      "epoch": 0.6582472254114046,
+      "grad_norm": 0.08351403474807739,
+      "learning_rate": 0.00016436148784745535,
+      "loss": 1.0382192611694336,
+      "mean_token_accuracy": 0.7587384819984436,
+      "num_tokens": 6891553.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0383310310542584,
+      "epoch": 0.6620742441637965,
+      "grad_norm": 0.07240983843803406,
+      "learning_rate": 0.0001640929233248288,
+      "loss": 1.0978598594665527,
+      "mean_token_accuracy": 0.7515711337327957,
+      "num_tokens": 6930875.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.085533195734024,
+      "epoch": 0.6659012629161883,
+      "grad_norm": 0.06999973207712173,
+      "learning_rate": 0.00016382435880220225,
+      "loss": 1.1412755966186523,
+      "mean_token_accuracy": 0.7471029132604599,
+      "num_tokens": 6971721.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.978867219388485,
+      "epoch": 0.6697282816685802,
+      "grad_norm": 0.06091843172907829,
+      "learning_rate": 0.0001635557942795757,
+      "loss": 1.023170566558838,
+      "mean_token_accuracy": 0.7686347916722298,
+      "num_tokens": 7010340.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0823599390685559,
+      "epoch": 0.6735553004209721,
+      "grad_norm": 0.07732617110013962,
+      "learning_rate": 0.0001632872297569491,
+      "loss": 1.1036816596984864,
+      "mean_token_accuracy": 0.7404509574174881,
+      "num_tokens": 7061189.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.9984334908425808,
+      "epoch": 0.677382319173364,
+      "grad_norm": 0.11516186594963074,
+      "learning_rate": 0.00016301866523432253,
+      "loss": 1.1070829391479493,
+      "mean_token_accuracy": 0.7593477964401245,
+      "num_tokens": 7098345.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.0023914370685816,
+      "epoch": 0.6812093379257559,
+      "grad_norm": 0.08624757081270218,
+      "learning_rate": 0.00016275010071169598,
+      "loss": 1.043964958190918,
+      "mean_token_accuracy": 0.7635682225227356,
+      "num_tokens": 7136873.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0823404759168624,
+      "epoch": 0.6850363566781478,
+      "grad_norm": 0.0846925675868988,
+      "learning_rate": 0.00016248153618906943,
+      "loss": 1.1333115577697754,
+      "mean_token_accuracy": 0.7394865393638611,
+      "num_tokens": 7181553.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0224985226988792,
+      "epoch": 0.6888633754305397,
+      "grad_norm": 0.060152288526296616,
+      "learning_rate": 0.00016221297166644287,
+      "loss": 1.0782301902770997,
+      "mean_token_accuracy": 0.759938097000122,
+      "num_tokens": 7223076.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0880159534513951,
+      "epoch": 0.6926903941829315,
+      "grad_norm": 0.06577905267477036,
+      "learning_rate": 0.00016194440714381632,
+      "loss": 1.1103734016418456,
+      "mean_token_accuracy": 0.7457254812121391,
+      "num_tokens": 7265594.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0144932381808758,
+      "epoch": 0.6965174129353234,
+      "grad_norm": 0.07276095449924469,
+      "learning_rate": 0.00016167584262118974,
+      "loss": 1.0733102798461913,
+      "mean_token_accuracy": 0.759276558458805,
+      "num_tokens": 7305359.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.0186967477202415,
+      "epoch": 0.7003444316877153,
+      "grad_norm": 0.08775337040424347,
+      "learning_rate": 0.0001614072780985632,
+      "loss": 1.0672088623046876,
+      "mean_token_accuracy": 0.7567476496100426,
+      "num_tokens": 7348060.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.9723582908511161,
+      "epoch": 0.7041714504401072,
+      "grad_norm": 0.09250030666589737,
+      "learning_rate": 0.0001611387135759366,
+      "loss": 1.0032880783081055,
+      "mean_token_accuracy": 0.7693375036120415,
+      "num_tokens": 7387110.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.9738463938236237,
+      "epoch": 0.7079984691924991,
+      "grad_norm": 0.09884033352136612,
+      "learning_rate": 0.00016087014905331006,
+      "loss": 1.0514408111572267,
+      "mean_token_accuracy": 0.7625621780753136,
+      "num_tokens": 7428511.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.0910252556204796,
+      "epoch": 0.711825487944891,
+      "grad_norm": 0.09194686263799667,
+      "learning_rate": 0.0001606015845306835,
+      "loss": 1.1226488113403321,
+      "mean_token_accuracy": 0.7452121302485466,
+      "num_tokens": 7474736.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0159518368542195,
+      "epoch": 0.7156525066972829,
+      "grad_norm": 0.07921712845563889,
+      "learning_rate": 0.00016033302000805695,
+      "loss": 1.061795711517334,
+      "mean_token_accuracy": 0.761272345483303,
+      "num_tokens": 7516891.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8903906352818012,
+      "epoch": 0.7194795254496748,
+      "grad_norm": 0.10288332402706146,
+      "learning_rate": 0.0001600644554854304,
+      "loss": 0.9471863746643067,
+      "mean_token_accuracy": 0.7831206247210503,
+      "num_tokens": 7551597.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0759326584637166,
+      "epoch": 0.7233065442020666,
+      "grad_norm": 0.06488945335149765,
+      "learning_rate": 0.00015979589096280382,
+      "loss": 1.136262798309326,
+      "mean_token_accuracy": 0.742707334458828,
+      "num_tokens": 7597529.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.9951202683150768,
+      "epoch": 0.7271335629544585,
+      "grad_norm": 0.06628359109163284,
+      "learning_rate": 0.00015952732644017724,
+      "loss": 1.0448930740356446,
+      "mean_token_accuracy": 0.7615623638033867,
+      "num_tokens": 7634291.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.0593122780323028,
+      "epoch": 0.7309605817068504,
+      "grad_norm": 0.08212320506572723,
+      "learning_rate": 0.00015925876191755068,
+      "loss": 1.1099414825439453,
+      "mean_token_accuracy": 0.7455122962594032,
+      "num_tokens": 7677086.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0122868783771992,
+      "epoch": 0.7347876004592423,
+      "grad_norm": 0.06458455324172974,
+      "learning_rate": 0.00015899019739492413,
+      "loss": 1.0447346687316894,
+      "mean_token_accuracy": 0.7542453840374946,
+      "num_tokens": 7721785.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0801358975470066,
+      "epoch": 0.7386146192116342,
+      "grad_norm": 0.06971931457519531,
+      "learning_rate": 0.00015872163287229758,
+      "loss": 1.109630012512207,
+      "mean_token_accuracy": 0.7450100436806679,
+      "num_tokens": 7761887.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.9081138484179974,
+      "epoch": 0.7424416379640261,
+      "grad_norm": 0.06223156675696373,
+      "learning_rate": 0.00015845306834967102,
+      "loss": 1.0026588439941406,
+      "mean_token_accuracy": 0.7772255912423134,
+      "num_tokens": 7807072.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0170219503343105,
+      "epoch": 0.746268656716418,
+      "grad_norm": 0.0685853511095047,
+      "learning_rate": 0.00015818450382704447,
+      "loss": 1.055277442932129,
+      "mean_token_accuracy": 0.7580551549792289,
+      "num_tokens": 7845791.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.0753178864717483,
+      "epoch": 0.7500956754688098,
+      "grad_norm": 0.08306553959846497,
+      "learning_rate": 0.0001579159393044179,
+      "loss": 1.1332826614379883,
+      "mean_token_accuracy": 0.7421450033783913,
+      "num_tokens": 7891091.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.9297005102038384,
+      "epoch": 0.7539226942212017,
+      "grad_norm": 0.08018683642148972,
+      "learning_rate": 0.0001576473747817913,
+      "loss": 1.000318431854248,
+      "mean_token_accuracy": 0.7793557167053222,
+      "num_tokens": 7928252.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.0840253300964833,
+      "epoch": 0.7577497129735936,
+      "grad_norm": 0.06487595289945602,
+      "learning_rate": 0.00015737881025916476,
+      "loss": 1.1166275024414063,
+      "mean_token_accuracy": 0.7378593400120735,
+      "num_tokens": 7972071.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.0406386695802212,
+      "epoch": 0.7615767317259855,
+      "grad_norm": 0.0615115687251091,
+      "learning_rate": 0.0001571102457365382,
+      "loss": 1.0869349479675292,
+      "mean_token_accuracy": 0.7490768045186996,
+      "num_tokens": 8016865.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.9573215276002884,
+      "epoch": 0.7654037504783774,
+      "grad_norm": 0.0715412124991417,
+      "learning_rate": 0.00015684168121391165,
+      "loss": 1.0404720306396484,
+      "mean_token_accuracy": 0.7706617951393128,
+      "num_tokens": 8055917.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.9201878193765879,
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.07988248765468597,
+      "learning_rate": 0.0001565731166912851,
+      "loss": 0.9380558967590332,
+      "mean_token_accuracy": 0.782890722155571,
+      "num_tokens": 8093252.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.0045961767435074,
+      "epoch": 0.7730577879831612,
+      "grad_norm": 0.061089444905519485,
+      "learning_rate": 0.00015630455216865855,
+      "loss": 1.0528027534484863,
+      "mean_token_accuracy": 0.7598949059844017,
+      "num_tokens": 8135244.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.9942824639379978,
+      "epoch": 0.776884806735553,
+      "grad_norm": 0.06443686783313751,
+      "learning_rate": 0.00015603598764603197,
+      "loss": 1.0168493270874024,
+      "mean_token_accuracy": 0.7590687796473503,
+      "num_tokens": 8178961.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.9773981764912605,
+      "epoch": 0.7807118254879449,
+      "grad_norm": 0.0818348303437233,
+      "learning_rate": 0.0001557674231234054,
+      "loss": 1.0193141937255858,
+      "mean_token_accuracy": 0.7708378821611405,
+      "num_tokens": 8217139.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.9836540646851063,
+      "epoch": 0.7845388442403368,
+      "grad_norm": 0.06240411475300789,
+      "learning_rate": 0.00015549885860077883,
+      "loss": 1.0662775993347169,
+      "mean_token_accuracy": 0.7658124819397927,
+      "num_tokens": 8252825.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.036501456052065,
+      "epoch": 0.7883658629927287,
+      "grad_norm": 0.09231610596179962,
+      "learning_rate": 0.00015523029407815228,
+      "loss": 1.112645435333252,
+      "mean_token_accuracy": 0.7541953936219216,
+      "num_tokens": 8295113.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.9800528183579444,
+      "epoch": 0.7921928817451206,
+      "grad_norm": 0.08806589245796204,
+      "learning_rate": 0.00015496172955552573,
+      "loss": 1.0401280403137207,
+      "mean_token_accuracy": 0.7672899037599563,
+      "num_tokens": 8335977.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.9678378522396087,
+      "epoch": 0.7960199004975125,
+      "grad_norm": 0.08777868002653122,
+      "learning_rate": 0.00015469316503289918,
+      "loss": 1.0509014129638672,
+      "mean_token_accuracy": 0.7696513712406159,
+      "num_tokens": 8374917.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.042826947569847,
+      "epoch": 0.7998469192499044,
+      "grad_norm": 0.09018490463495255,
+      "learning_rate": 0.00015442460051027262,
+      "loss": 1.0869378089904784,
+      "mean_token_accuracy": 0.7507286682724953,
+      "num_tokens": 8415614.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.0548966623842717,
+      "epoch": 0.8036739380022963,
+      "grad_norm": 0.07267605513334274,
+      "learning_rate": 0.00015415603598764604,
+      "loss": 1.0960289001464845,
+      "mean_token_accuracy": 0.7545556098222732,
+      "num_tokens": 8455059.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.044345210492611,
+      "epoch": 0.8075009567546881,
+      "grad_norm": 0.08414279669523239,
+      "learning_rate": 0.00015388747146501946,
+      "loss": 1.1200661659240723,
+      "mean_token_accuracy": 0.7490431442856789,
+      "num_tokens": 8493866.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.0317029684782029,
+      "epoch": 0.81132797550708,
+      "grad_norm": 0.06549747288227081,
+      "learning_rate": 0.0001536189069423929,
+      "loss": 1.0583623886108398,
+      "mean_token_accuracy": 0.7555923700332642,
+      "num_tokens": 8536147.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.9694572634994983,
+      "epoch": 0.8151549942594719,
+      "grad_norm": 0.08112777769565582,
+      "learning_rate": 0.00015335034241976636,
+      "loss": 1.0503274917602539,
+      "mean_token_accuracy": 0.7646921187639236,
+      "num_tokens": 8578007.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.9358880028128624,
+      "epoch": 0.8189820130118638,
+      "grad_norm": 0.07176466286182404,
+      "learning_rate": 0.0001530817778971398,
+      "loss": 1.000410270690918,
+      "mean_token_accuracy": 0.773740467429161,
+      "num_tokens": 8620999.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.0444137938320637,
+      "epoch": 0.8228090317642557,
+      "grad_norm": 0.06355756521224976,
+      "learning_rate": 0.00015281321337451325,
+      "loss": 1.0860448837280274,
+      "mean_token_accuracy": 0.751850588619709,
+      "num_tokens": 8663354.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.9044980220496655,
+      "epoch": 0.8266360505166476,
+      "grad_norm": 0.080223448574543,
+      "learning_rate": 0.00015254464885188667,
+      "loss": 0.9434403419494629,
+      "mean_token_accuracy": 0.7828752338886261,
+      "num_tokens": 8699748.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.0172922544181346,
+      "epoch": 0.8304630692690395,
+      "grad_norm": 0.06971501559019089,
+      "learning_rate": 0.00015227608432926012,
+      "loss": 1.0325962066650392,
+      "mean_token_accuracy": 0.7651202365756035,
+      "num_tokens": 8739901.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.9639742732048034,
+      "epoch": 0.8342900880214313,
+      "grad_norm": 0.06396778672933578,
+      "learning_rate": 0.00015200751980663354,
+      "loss": 1.0435317039489747,
+      "mean_token_accuracy": 0.7667818054556846,
+      "num_tokens": 8778980.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8876220636069775,
+      "epoch": 0.8381171067738232,
+      "grad_norm": 0.09910868853330612,
+      "learning_rate": 0.00015173895528400698,
+      "loss": 0.9876300811767578,
+      "mean_token_accuracy": 0.7865215808153152,
+      "num_tokens": 8815525.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0369405087083579,
+      "epoch": 0.8419441255262151,
+      "grad_norm": 0.08775259554386139,
+      "learning_rate": 0.00015147039076138043,
+      "loss": 1.1244413375854492,
+      "mean_token_accuracy": 0.7550949841737747,
+      "num_tokens": 8857085.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.9762422502040863,
+      "epoch": 0.845771144278607,
+      "grad_norm": 0.08659302443265915,
+      "learning_rate": 0.00015120182623875388,
+      "loss": 1.0164811134338378,
+      "mean_token_accuracy": 0.771617329120636,
+      "num_tokens": 8894271.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.9543228000402451,
+      "epoch": 0.8495981630309989,
+      "grad_norm": 0.09588434547185898,
+      "learning_rate": 0.00015093326171612733,
+      "loss": 1.0303520202636718,
+      "mean_token_accuracy": 0.768992331624031,
+      "num_tokens": 8934095.0,
+      "step": 2220
+    },
+    {
+      "entropy": 1.1307236567139625,
+      "epoch": 0.8534251817833908,
+      "grad_norm": 0.07016360014677048,
+      "learning_rate": 0.00015066469719350075,
+      "loss": 1.1526556968688966,
+      "mean_token_accuracy": 0.7296861469745636,
+      "num_tokens": 8982341.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0867296956479549,
+      "epoch": 0.8572522005357827,
+      "grad_norm": 0.07838597148656845,
+      "learning_rate": 0.00015039613267087417,
+      "loss": 1.1031158447265625,
+      "mean_token_accuracy": 0.7445572927594185,
+      "num_tokens": 9027401.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.9492381684482097,
+      "epoch": 0.8610792192881745,
+      "grad_norm": 0.08416638523340225,
+      "learning_rate": 0.0001501275681482476,
+      "loss": 1.0079804420471192,
+      "mean_token_accuracy": 0.7709973976016045,
+      "num_tokens": 9069985.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.9767517909407616,
+      "epoch": 0.8649062380405664,
+      "grad_norm": 0.09798935055732727,
+      "learning_rate": 0.00014985900362562106,
+      "loss": 1.0394697189331055,
+      "mean_token_accuracy": 0.7647709026932716,
+      "num_tokens": 9108246.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.9779160171747208,
+      "epoch": 0.8687332567929583,
+      "grad_norm": 0.08669373393058777,
+      "learning_rate": 0.0001495904391029945,
+      "loss": 1.0398100852966308,
+      "mean_token_accuracy": 0.7669417649507523,
+      "num_tokens": 9147055.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.014696953445673,
+      "epoch": 0.8725602755453502,
+      "grad_norm": 0.07674991339445114,
+      "learning_rate": 0.00014932187458036795,
+      "loss": 1.0742408752441406,
+      "mean_token_accuracy": 0.7583330690860748,
+      "num_tokens": 9187727.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.9619584158062935,
+      "epoch": 0.8763872942977421,
+      "grad_norm": 0.09512930363416672,
+      "learning_rate": 0.00014905331005774137,
+      "loss": 1.01895112991333,
+      "mean_token_accuracy": 0.7718996241688728,
+      "num_tokens": 9228518.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8759313493967056,
+      "epoch": 0.880214313050134,
+      "grad_norm": 0.06927543133497238,
+      "learning_rate": 0.00014878474553511482,
+      "loss": 0.9590776443481446,
+      "mean_token_accuracy": 0.783099564909935,
+      "num_tokens": 9269392.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0930156745016575,
+      "epoch": 0.8840413318025259,
+      "grad_norm": 0.07149595022201538,
+      "learning_rate": 0.00014851618101248824,
+      "loss": 1.132398796081543,
+      "mean_token_accuracy": 0.7445679202675819,
+      "num_tokens": 9310993.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.9991384916007519,
+      "epoch": 0.8878683505549178,
+      "grad_norm": 0.100126251578331,
+      "learning_rate": 0.0001482476164898617,
+      "loss": 1.0395862579345703,
+      "mean_token_accuracy": 0.7618231356143952,
+      "num_tokens": 9349210.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.9891969002783298,
+      "epoch": 0.8916953693073096,
+      "grad_norm": 0.07942050695419312,
+      "learning_rate": 0.00014797905196723514,
+      "loss": 1.0403067588806152,
+      "mean_token_accuracy": 0.7636258214712143,
+      "num_tokens": 9386251.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.034816125780344,
+      "epoch": 0.8955223880597015,
+      "grad_norm": 0.07803855836391449,
+      "learning_rate": 0.00014771048744460858,
+      "loss": 1.088371467590332,
+      "mean_token_accuracy": 0.7585563778877258,
+      "num_tokens": 9425492.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.998091223090887,
+      "epoch": 0.8993494068120934,
+      "grad_norm": 0.06696243584156036,
+      "learning_rate": 0.00014744192292198203,
+      "loss": 1.0410521507263184,
+      "mean_token_accuracy": 0.7595112159848213,
+      "num_tokens": 9466862.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.9615898832678795,
+      "epoch": 0.9031764255644853,
+      "grad_norm": 0.07813845574855804,
+      "learning_rate": 0.00014717335839935545,
+      "loss": 1.0265610694885254,
+      "mean_token_accuracy": 0.7707905381917953,
+      "num_tokens": 9503827.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8776158876717091,
+      "epoch": 0.9070034443168772,
+      "grad_norm": 0.10287057608366013,
+      "learning_rate": 0.0001469047938767289,
+      "loss": 0.9231206893920898,
+      "mean_token_accuracy": 0.7909859612584114,
+      "num_tokens": 9536194.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.980732673406601,
+      "epoch": 0.9108304630692691,
+      "grad_norm": 0.06174289435148239,
+      "learning_rate": 0.00014663622935410232,
+      "loss": 1.0316704750061034,
+      "mean_token_accuracy": 0.7596900418400765,
+      "num_tokens": 9577621.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0083129487931728,
+      "epoch": 0.914657481821661,
+      "grad_norm": 0.08805451542139053,
+      "learning_rate": 0.00014636766483147576,
+      "loss": 1.0296180725097657,
+      "mean_token_accuracy": 0.7577597886323929,
+      "num_tokens": 9616522.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0002505116164684,
+      "epoch": 0.9184845005740528,
+      "grad_norm": 0.07697928696870804,
+      "learning_rate": 0.0001460991003088492,
+      "loss": 1.0411831855773925,
+      "mean_token_accuracy": 0.7589930936694145,
+      "num_tokens": 9659217.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.971958789229393,
+      "epoch": 0.9223115193264447,
+      "grad_norm": 0.08504882454872131,
+      "learning_rate": 0.00014583053578622266,
+      "loss": 1.015835952758789,
+      "mean_token_accuracy": 0.7664303690195083,
+      "num_tokens": 9694120.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.9250703640282154,
+      "epoch": 0.9261385380788366,
+      "grad_norm": 0.06279303133487701,
+      "learning_rate": 0.00014556197126359608,
+      "loss": 0.9673631668090821,
+      "mean_token_accuracy": 0.782692727446556,
+      "num_tokens": 9732460.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.0777716524899006,
+      "epoch": 0.9299655568312285,
+      "grad_norm": 0.06884833425283432,
+      "learning_rate": 0.00014529340674096952,
+      "loss": 1.1415311813354492,
+      "mean_token_accuracy": 0.7447684407234192,
+      "num_tokens": 9773760.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.0116477236151695,
+      "epoch": 0.9337925755836204,
+      "grad_norm": 0.06346814334392548,
+      "learning_rate": 0.00014502484221834297,
+      "loss": 1.0904932975769044,
+      "mean_token_accuracy": 0.7616935014724732,
+      "num_tokens": 9808910.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.9434679664671421,
+      "epoch": 0.9376195943360123,
+      "grad_norm": 0.09843038022518158,
+      "learning_rate": 0.0001447562776957164,
+      "loss": 1.0111047744750976,
+      "mean_token_accuracy": 0.774254959821701,
+      "num_tokens": 9846472.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.035598163306713,
+      "epoch": 0.9414466130884042,
+      "grad_norm": 0.08025770634412766,
+      "learning_rate": 0.00014448771317308984,
+      "loss": 1.1550275802612304,
+      "mean_token_accuracy": 0.7497850373387337,
+      "num_tokens": 9885082.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.057615876197815,
+      "epoch": 0.945273631840796,
+      "grad_norm": 0.07916443794965744,
+      "learning_rate": 0.00014421914865046329,
+      "loss": 1.114585781097412,
+      "mean_token_accuracy": 0.7495191320776939,
+      "num_tokens": 9924849.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.9576205931603908,
+      "epoch": 0.9491006505931879,
+      "grad_norm": 0.10745597630739212,
+      "learning_rate": 0.00014395058412783673,
+      "loss": 1.0471231460571289,
+      "mean_token_accuracy": 0.7697127804160118,
+      "num_tokens": 9969210.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.012363300472498,
+      "epoch": 0.9529276693455798,
+      "grad_norm": 0.09448845684528351,
+      "learning_rate": 0.00014368201960521015,
+      "loss": 1.0322566986083985,
+      "mean_token_accuracy": 0.7568502962589264,
+      "num_tokens": 10009532.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.9387446999549866,
+      "epoch": 0.9567546880979717,
+      "grad_norm": 0.08835543692111969,
+      "learning_rate": 0.0001434134550825836,
+      "loss": 0.9836790084838867,
+      "mean_token_accuracy": 0.7740270137786865,
+      "num_tokens": 10051767.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.043863268941641,
+      "epoch": 0.9605817068503636,
+      "grad_norm": 0.0590866394340992,
+      "learning_rate": 0.00014314489055995705,
+      "loss": 1.1286373138427734,
+      "mean_token_accuracy": 0.755294018983841,
+      "num_tokens": 10093518.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.068480123579502,
+      "epoch": 0.9644087256027555,
+      "grad_norm": 0.06240773946046829,
+      "learning_rate": 0.00014287632603733047,
+      "loss": 1.1243531227111816,
+      "mean_token_accuracy": 0.7457959160208703,
+      "num_tokens": 10137842.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.9648511357605457,
+      "epoch": 0.9682357443551474,
+      "grad_norm": 0.07577214390039444,
+      "learning_rate": 0.00014260776151470391,
+      "loss": 1.0646875381469727,
+      "mean_token_accuracy": 0.7689151406288147,
+      "num_tokens": 10177541.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0034234993159772,
+      "epoch": 0.9720627631075393,
+      "grad_norm": 0.06887607276439667,
+      "learning_rate": 0.00014233919699207736,
+      "loss": 1.0736650466918944,
+      "mean_token_accuracy": 0.7580653995275497,
+      "num_tokens": 10217056.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.9054977536201477,
+      "epoch": 0.9758897818599311,
+      "grad_norm": 0.12731540203094482,
+      "learning_rate": 0.00014207063246945078,
+      "loss": 0.9581779479980469,
+      "mean_token_accuracy": 0.7800818130373954,
+      "num_tokens": 10249622.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0892111197113992,
+      "epoch": 0.979716800612323,
+      "grad_norm": 0.08707671612501144,
+      "learning_rate": 0.00014180206794682423,
+      "loss": 1.1551457405090333,
+      "mean_token_accuracy": 0.7434241071343421,
+      "num_tokens": 10287483.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.9462251186370849,
+      "epoch": 0.9835438193647149,
+      "grad_norm": 0.10457631945610046,
+      "learning_rate": 0.00014153350342419768,
+      "loss": 0.9859563827514648,
+      "mean_token_accuracy": 0.7729493409395218,
+      "num_tokens": 10324562.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.9609014384448529,
+      "epoch": 0.9873708381171068,
+      "grad_norm": 0.1095169261097908,
+      "learning_rate": 0.0001412649389015711,
+      "loss": 1.00408992767334,
+      "mean_token_accuracy": 0.769461353123188,
+      "num_tokens": 10368482.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.9500531531870365,
+      "epoch": 0.9911978568694987,
+      "grad_norm": 0.12787973880767822,
+      "learning_rate": 0.00014099637437894454,
+      "loss": 1.0082733154296875,
+      "mean_token_accuracy": 0.7726384818553924,
+      "num_tokens": 10407666.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.9639500208199024,
+      "epoch": 0.9950248756218906,
+      "grad_norm": 0.08555731922388077,
+      "learning_rate": 0.000140727809856318,
+      "loss": 0.9910324096679688,
+      "mean_token_accuracy": 0.7700270056724549,
+      "num_tokens": 10445419.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.9984636768698693,
+      "epoch": 0.9988518943742825,
+      "grad_norm": 0.10294629633426666,
+      "learning_rate": 0.00014045924533369144,
+      "loss": 1.0837631225585938,
+      "mean_token_accuracy": 0.7655858203768731,
+      "num_tokens": 10483287.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.940229170024395,
+      "epoch": 1.0026789131266742,
+      "grad_norm": 0.10580310225486755,
+      "learning_rate": 0.00014019068081106486,
+      "loss": 0.9650541305541992,
+      "mean_token_accuracy": 0.7728109017014504,
+      "num_tokens": 10523841.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.9358184114098549,
+      "epoch": 1.0065059318790661,
+      "grad_norm": 0.12460961192846298,
+      "learning_rate": 0.0001399221162884383,
+      "loss": 0.9570166587829589,
+      "mean_token_accuracy": 0.7772100657224655,
+      "num_tokens": 10561636.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.010379894077778,
+      "epoch": 1.010332950631458,
+      "grad_norm": 0.0781383365392685,
+      "learning_rate": 0.00013965355176581175,
+      "loss": 1.0524909019470214,
+      "mean_token_accuracy": 0.7589353621006012,
+      "num_tokens": 10605899.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.977487600594759,
+      "epoch": 1.01415996938385,
+      "grad_norm": 0.0902724489569664,
+      "learning_rate": 0.00013938498724318517,
+      "loss": 1.0475889205932618,
+      "mean_token_accuracy": 0.7629667386412621,
+      "num_tokens": 10642372.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.9681369736790657,
+      "epoch": 1.0179869881362418,
+      "grad_norm": 0.06344746798276901,
+      "learning_rate": 0.00013911642272055862,
+      "loss": 1.0268775939941406,
+      "mean_token_accuracy": 0.7677509978413581,
+      "num_tokens": 10682308.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.9013996437191963,
+      "epoch": 1.0218140068886337,
+      "grad_norm": 0.09890369325876236,
+      "learning_rate": 0.00013884785819793206,
+      "loss": 0.969085693359375,
+      "mean_token_accuracy": 0.7815661624073982,
+      "num_tokens": 10720755.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.9415140472352505,
+      "epoch": 1.0256410256410255,
+      "grad_norm": 0.08691754937171936,
+      "learning_rate": 0.00013857929367530548,
+      "loss": 0.9783688545227051,
+      "mean_token_accuracy": 0.7722749456763267,
+      "num_tokens": 10759842.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.9437286920845509,
+      "epoch": 1.0294680443934174,
+      "grad_norm": 0.06577731668949127,
+      "learning_rate": 0.00013831072915267893,
+      "loss": 0.9904938697814941,
+      "mean_token_accuracy": 0.7716649904847145,
+      "num_tokens": 10803740.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.9657303221523762,
+      "epoch": 1.0332950631458093,
+      "grad_norm": 0.07847272604703903,
+      "learning_rate": 0.00013804216463005238,
+      "loss": 1.0073646545410155,
+      "mean_token_accuracy": 0.7678608119487762,
+      "num_tokens": 10841808.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.881027878075838,
+      "epoch": 1.0371220818982012,
+      "grad_norm": 0.12755495309829712,
+      "learning_rate": 0.00013777360010742583,
+      "loss": 0.955751895904541,
+      "mean_token_accuracy": 0.7835927039384842,
+      "num_tokens": 10880108.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8458237417042256,
+      "epoch": 1.040949100650593,
+      "grad_norm": 0.07641884684562683,
+      "learning_rate": 0.00013750503558479925,
+      "loss": 0.9140083312988281,
+      "mean_token_accuracy": 0.7939343526959419,
+      "num_tokens": 10916272.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8845301080495119,
+      "epoch": 1.044776119402985,
+      "grad_norm": 0.08896184712648392,
+      "learning_rate": 0.0001372364710621727,
+      "loss": 0.9332797050476074,
+      "mean_token_accuracy": 0.7884662911295891,
+      "num_tokens": 10951932.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.963884600251913,
+      "epoch": 1.0486031381553769,
+      "grad_norm": 0.10196536034345627,
+      "learning_rate": 0.00013696790653954614,
+      "loss": 1.0123867988586426,
+      "mean_token_accuracy": 0.7659088596701622,
+      "num_tokens": 10991548.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.9720129862427711,
+      "epoch": 1.0524301569077688,
+      "grad_norm": 0.07552212476730347,
+      "learning_rate": 0.00013669934201691956,
+      "loss": 1.015409564971924,
+      "mean_token_accuracy": 0.7689290955662728,
+      "num_tokens": 11028749.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.9871743015944958,
+      "epoch": 1.0562571756601606,
+      "grad_norm": 0.09255808591842651,
+      "learning_rate": 0.000136430777494293,
+      "loss": 1.0351217269897461,
+      "mean_token_accuracy": 0.7620491668581962,
+      "num_tokens": 11071336.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.809666246920824,
+      "epoch": 1.0600841944125525,
+      "grad_norm": 0.08891233056783676,
+      "learning_rate": 0.00013616221297166645,
+      "loss": 0.8595174789428711,
+      "mean_token_accuracy": 0.8053640425205231,
+      "num_tokens": 11107708.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.9220615286380053,
+      "epoch": 1.0639112131649444,
+      "grad_norm": 0.0731620192527771,
+      "learning_rate": 0.0001358936484490399,
+      "loss": 0.9694333076477051,
+      "mean_token_accuracy": 0.7767527863383293,
+      "num_tokens": 11149005.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8744502332061529,
+      "epoch": 1.0677382319173363,
+      "grad_norm": 0.0865791067481041,
+      "learning_rate": 0.00013562508392641332,
+      "loss": 0.9401009559631348,
+      "mean_token_accuracy": 0.7854847684502602,
+      "num_tokens": 11189214.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.989877526462078,
+      "epoch": 1.0715652506697282,
+      "grad_norm": 0.09394430369138718,
+      "learning_rate": 0.00013535651940378677,
+      "loss": 1.0487696647644043,
+      "mean_token_accuracy": 0.7607394486665726,
+      "num_tokens": 11225161.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.8656694941222668,
+      "epoch": 1.07539226942212,
+      "grad_norm": 0.10940351337194443,
+      "learning_rate": 0.0001350879548811602,
+      "loss": 0.9236039161682129,
+      "mean_token_accuracy": 0.7919901207089424,
+      "num_tokens": 11261274.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.063130483776331,
+      "epoch": 1.079219288174512,
+      "grad_norm": 0.06853083521127701,
+      "learning_rate": 0.00013481939035853364,
+      "loss": 1.0725152015686035,
+      "mean_token_accuracy": 0.7454188778996468,
+      "num_tokens": 11302522.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.92764787748456,
+      "epoch": 1.0830463069269038,
+      "grad_norm": 0.10344231128692627,
+      "learning_rate": 0.00013455082583590708,
+      "loss": 0.9725144386291504,
+      "mean_token_accuracy": 0.7810687303543091,
+      "num_tokens": 11339898.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.9415482886135578,
+      "epoch": 1.0868733256792957,
+      "grad_norm": 0.12117484956979752,
+      "learning_rate": 0.00013428226131328053,
+      "loss": 1.0216625213623047,
+      "mean_token_accuracy": 0.7713929772377014,
+      "num_tokens": 11380187.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.9300718136131764,
+      "epoch": 1.0907003444316876,
+      "grad_norm": 0.09950343519449234,
+      "learning_rate": 0.00013401369679065398,
+      "loss": 0.9862215042114257,
+      "mean_token_accuracy": 0.7748491272330285,
+      "num_tokens": 11417351.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.9016943011432886,
+      "epoch": 1.0945273631840795,
+      "grad_norm": 0.10104110836982727,
+      "learning_rate": 0.0001337451322680274,
+      "loss": 0.9565576553344727,
+      "mean_token_accuracy": 0.7823473244905472,
+      "num_tokens": 11455566.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.0184541821479798,
+      "epoch": 1.0983543819364714,
+      "grad_norm": 0.07055146247148514,
+      "learning_rate": 0.00013347656774540084,
+      "loss": 1.0644380569458007,
+      "mean_token_accuracy": 0.7551941126585007,
+      "num_tokens": 11499960.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.9143499568104744,
+      "epoch": 1.1021814006888633,
+      "grad_norm": 0.09798481315374374,
+      "learning_rate": 0.00013320800322277426,
+      "loss": 0.9477805137634278,
+      "mean_token_accuracy": 0.778240317106247,
+      "num_tokens": 11536434.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8803758375346661,
+      "epoch": 1.1060084194412552,
+      "grad_norm": 0.09720771014690399,
+      "learning_rate": 0.0001329394387001477,
+      "loss": 0.9369168281555176,
+      "mean_token_accuracy": 0.786097663640976,
+      "num_tokens": 11572420.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.9127089619636536,
+      "epoch": 1.109835438193647,
+      "grad_norm": 0.07493265718221664,
+      "learning_rate": 0.00013267087417752116,
+      "loss": 0.9610566139221192,
+      "mean_token_accuracy": 0.7780416712164879,
+      "num_tokens": 11607494.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.9359945230185985,
+      "epoch": 1.113662456946039,
+      "grad_norm": 0.09086300432682037,
+      "learning_rate": 0.0001324023096548946,
+      "loss": 0.9519670486450196,
+      "mean_token_accuracy": 0.7745376393198967,
+      "num_tokens": 11647057.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.9206651791930198,
+      "epoch": 1.1174894756984308,
+      "grad_norm": 0.10007902979850769,
+      "learning_rate": 0.00013213374513226805,
+      "loss": 0.9783179283142089,
+      "mean_token_accuracy": 0.778519794344902,
+      "num_tokens": 11685762.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.9937357418239117,
+      "epoch": 1.1213164944508227,
+      "grad_norm": 0.0993100181221962,
+      "learning_rate": 0.00013186518060964147,
+      "loss": 1.0440019607543944,
+      "mean_token_accuracy": 0.7590440228581429,
+      "num_tokens": 11727379.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.048055526614189,
+      "epoch": 1.1251435132032146,
+      "grad_norm": 0.11140380054712296,
+      "learning_rate": 0.0001315966160870149,
+      "loss": 1.1046284675598144,
+      "mean_token_accuracy": 0.7413847833871842,
+      "num_tokens": 11770734.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.9562077779322863,
+      "epoch": 1.1289705319556065,
+      "grad_norm": 0.11506770551204681,
+      "learning_rate": 0.00013132805156438834,
+      "loss": 0.9946146011352539,
+      "mean_token_accuracy": 0.7750585973262787,
+      "num_tokens": 11806270.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.9747304327785968,
+      "epoch": 1.1327975507079984,
+      "grad_norm": 0.1126897856593132,
+      "learning_rate": 0.00013105948704176179,
+      "loss": 1.061129093170166,
+      "mean_token_accuracy": 0.7613553464412689,
+      "num_tokens": 11852779.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0132145062088966,
+      "epoch": 1.1366245694603903,
+      "grad_norm": 0.08260762691497803,
+      "learning_rate": 0.00013079092251913523,
+      "loss": 1.0199948310852052,
+      "mean_token_accuracy": 0.7617463275790215,
+      "num_tokens": 11897084.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.9878915682435035,
+      "epoch": 1.1404515882127821,
+      "grad_norm": 0.08098926395177841,
+      "learning_rate": 0.00013052235799650868,
+      "loss": 1.0480783462524415,
+      "mean_token_accuracy": 0.763205036520958,
+      "num_tokens": 11938987.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0176467482000588,
+      "epoch": 1.144278606965174,
+      "grad_norm": 0.0966029092669487,
+      "learning_rate": 0.0001302537934738821,
+      "loss": 1.093599796295166,
+      "mean_token_accuracy": 0.7526282608509064,
+      "num_tokens": 11981156.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0054687768220902,
+      "epoch": 1.148105625717566,
+      "grad_norm": 0.09327300637960434,
+      "learning_rate": 0.00012998522895125555,
+      "loss": 1.039564609527588,
+      "mean_token_accuracy": 0.7592228040099144,
+      "num_tokens": 12025389.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.9626951858401298,
+      "epoch": 1.1519326444699578,
+      "grad_norm": 0.06154703348875046,
+      "learning_rate": 0.00012971666442862897,
+      "loss": 0.9993762016296387,
+      "mean_token_accuracy": 0.769777101278305,
+      "num_tokens": 12069545.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.9221224367618561,
+      "epoch": 1.1557596632223497,
+      "grad_norm": 0.1140643060207367,
+      "learning_rate": 0.00012944809990600241,
+      "loss": 0.9887493133544922,
+      "mean_token_accuracy": 0.7754134178161621,
+      "num_tokens": 12113892.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.011741641908884,
+      "epoch": 1.1595866819747416,
+      "grad_norm": 0.08721659332513809,
+      "learning_rate": 0.00012917953538337586,
+      "loss": 1.068478488922119,
+      "mean_token_accuracy": 0.7615607067942619,
+      "num_tokens": 12153746.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.9926261432468891,
+      "epoch": 1.1634137007271335,
+      "grad_norm": 0.07577186822891235,
+      "learning_rate": 0.0001289109708607493,
+      "loss": 1.047102451324463,
+      "mean_token_accuracy": 0.7669480383396149,
+      "num_tokens": 12199067.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.945004402846098,
+      "epoch": 1.1672407194795253,
+      "grad_norm": 0.08443465083837509,
+      "learning_rate": 0.00012864240633812276,
+      "loss": 0.9891506195068359,
+      "mean_token_accuracy": 0.7756656989455223,
+      "num_tokens": 12243766.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.9602406993508339,
+      "epoch": 1.1710677382319172,
+      "grad_norm": 0.07647141069173813,
+      "learning_rate": 0.00012837384181549618,
+      "loss": 1.0091946601867676,
+      "mean_token_accuracy": 0.7702717915177345,
+      "num_tokens": 12279555.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.9430582121014595,
+      "epoch": 1.1748947569843091,
+      "grad_norm": 0.10050038248300552,
+      "learning_rate": 0.0001281052772928696,
+      "loss": 1.0251899719238282,
+      "mean_token_accuracy": 0.7759435445070266,
+      "num_tokens": 12316974.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.0339640237390995,
+      "epoch": 1.178721775736701,
+      "grad_norm": 0.09026551991701126,
+      "learning_rate": 0.00012783671277024304,
+      "loss": 1.0652464866638183,
+      "mean_token_accuracy": 0.7533303231000901,
+      "num_tokens": 12358111.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.9808862328529357,
+      "epoch": 1.182548794489093,
+      "grad_norm": 0.08769362419843674,
+      "learning_rate": 0.0001275681482476165,
+      "loss": 1.0068347930908204,
+      "mean_token_accuracy": 0.7660810023546218,
+      "num_tokens": 12401669.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.9436531282961369,
+      "epoch": 1.1863758132414848,
+      "grad_norm": 0.09366963803768158,
+      "learning_rate": 0.00012729958372498994,
+      "loss": 1.0298351287841796,
+      "mean_token_accuracy": 0.7704201564192772,
+      "num_tokens": 12442005.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8712134130299092,
+      "epoch": 1.1902028319938767,
+      "grad_norm": 0.14041900634765625,
+      "learning_rate": 0.00012703101920236338,
+      "loss": 0.9094470977783203,
+      "mean_token_accuracy": 0.7861496224999428,
+      "num_tokens": 12484476.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.9474696554243565,
+      "epoch": 1.1940298507462686,
+      "grad_norm": 0.10449594259262085,
+      "learning_rate": 0.00012676245467973683,
+      "loss": 0.9729720115661621,
+      "mean_token_accuracy": 0.7746587276458741,
+      "num_tokens": 12521351.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.9215874671936035,
+      "epoch": 1.1978568694986604,
+      "grad_norm": 0.07733117789030075,
+      "learning_rate": 0.00012649389015711025,
+      "loss": 0.992548942565918,
+      "mean_token_accuracy": 0.7789316549897194,
+      "num_tokens": 12564603.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.9349980562925339,
+      "epoch": 1.2016838882510523,
+      "grad_norm": 0.06924714148044586,
+      "learning_rate": 0.00012622532563448367,
+      "loss": 1.010727596282959,
+      "mean_token_accuracy": 0.7728876963257789,
+      "num_tokens": 12606025.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.9719727545976639,
+      "epoch": 1.2055109070034442,
+      "grad_norm": 0.07646770775318146,
+      "learning_rate": 0.00012595676111185712,
+      "loss": 1.0482423782348633,
+      "mean_token_accuracy": 0.7659243881702423,
+      "num_tokens": 12647703.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0236301876604557,
+      "epoch": 1.209337925755836,
+      "grad_norm": 0.08547945320606232,
+      "learning_rate": 0.00012568819658923056,
+      "loss": 1.0771334648132325,
+      "mean_token_accuracy": 0.7551302567124367,
+      "num_tokens": 12692347.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.9277745552361012,
+      "epoch": 1.213164944508228,
+      "grad_norm": 0.10816850513219833,
+      "learning_rate": 0.000125419632066604,
+      "loss": 0.9680308341979981,
+      "mean_token_accuracy": 0.7722468450665474,
+      "num_tokens": 12729671.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.9760092988610267,
+      "epoch": 1.2169919632606199,
+      "grad_norm": 0.08950033783912659,
+      "learning_rate": 0.00012515106754397746,
+      "loss": 1.000643539428711,
+      "mean_token_accuracy": 0.7665232941508293,
+      "num_tokens": 12768100.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.9292771026492119,
+      "epoch": 1.2208189820130118,
+      "grad_norm": 0.08686704933643341,
+      "learning_rate": 0.0001248825030213509,
+      "loss": 1.019674015045166,
+      "mean_token_accuracy": 0.7758068069815636,
+      "num_tokens": 12801323.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8500060614198446,
+      "epoch": 1.2246460007654036,
+      "grad_norm": 0.07462778687477112,
+      "learning_rate": 0.00012461393849872433,
+      "loss": 0.9042973518371582,
+      "mean_token_accuracy": 0.7897424980998039,
+      "num_tokens": 12839880.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.9205234386026859,
+      "epoch": 1.2284730195177955,
+      "grad_norm": 0.07027672231197357,
+      "learning_rate": 0.00012434537397609775,
+      "loss": 0.9424190521240234,
+      "mean_token_accuracy": 0.7767854332923889,
+      "num_tokens": 12878349.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.9074239492416382,
+      "epoch": 1.2323000382701874,
+      "grad_norm": 0.09741132706403732,
+      "learning_rate": 0.0001240768094534712,
+      "loss": 0.9651589393615723,
+      "mean_token_accuracy": 0.7790584430098534,
+      "num_tokens": 12917588.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8874296098947525,
+      "epoch": 1.2361270570225793,
+      "grad_norm": 0.08608463406562805,
+      "learning_rate": 0.00012380824493084464,
+      "loss": 0.9437139511108399,
+      "mean_token_accuracy": 0.7854243695735932,
+      "num_tokens": 12956199.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.9470510125160218,
+      "epoch": 1.2399540757749712,
+      "grad_norm": 0.09247037768363953,
+      "learning_rate": 0.0001235396804082181,
+      "loss": 1.032781982421875,
+      "mean_token_accuracy": 0.7712572082877159,
+      "num_tokens": 13000822.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8850176699459553,
+      "epoch": 1.243781094527363,
+      "grad_norm": 0.08397585898637772,
+      "learning_rate": 0.00012327111588559153,
+      "loss": 0.9292671203613281,
+      "mean_token_accuracy": 0.787578609585762,
+      "num_tokens": 13043532.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8605544999241829,
+      "epoch": 1.247608113279755,
+      "grad_norm": 0.0952179804444313,
+      "learning_rate": 0.00012300255136296498,
+      "loss": 0.8990240097045898,
+      "mean_token_accuracy": 0.7919793605804444,
+      "num_tokens": 13081376.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.003395075351,
+      "epoch": 1.2514351320321468,
+      "grad_norm": 0.08914512395858765,
+      "learning_rate": 0.0001227339868403384,
+      "loss": 1.1446642875671387,
+      "mean_token_accuracy": 0.7565032340586185,
+      "num_tokens": 13119474.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.9566417217254639,
+      "epoch": 1.2552621507845387,
+      "grad_norm": 0.13220350444316864,
+      "learning_rate": 0.00012246542231771182,
+      "loss": 0.9976698875427246,
+      "mean_token_accuracy": 0.7722181305289268,
+      "num_tokens": 13162637.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.888442064449191,
+      "epoch": 1.2590891695369306,
+      "grad_norm": 0.10493922978639603,
+      "learning_rate": 0.00012219685779508527,
+      "loss": 0.916744613647461,
+      "mean_token_accuracy": 0.7896391779184342,
+      "num_tokens": 13199412.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.9262259535491466,
+      "epoch": 1.2629161882893225,
+      "grad_norm": 0.09022962301969528,
+      "learning_rate": 0.00012192829327245872,
+      "loss": 0.9885137557983399,
+      "mean_token_accuracy": 0.778158649802208,
+      "num_tokens": 13240292.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.9356066003441811,
+      "epoch": 1.2667432070417144,
+      "grad_norm": 0.09693239629268646,
+      "learning_rate": 0.00012165972874983216,
+      "loss": 0.9731400489807129,
+      "mean_token_accuracy": 0.7748182758688926,
+      "num_tokens": 13275876.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.868951104208827,
+      "epoch": 1.2705702257941063,
+      "grad_norm": 0.09237370640039444,
+      "learning_rate": 0.0001213911642272056,
+      "loss": 0.9127277374267578,
+      "mean_token_accuracy": 0.7890144631266593,
+      "num_tokens": 13314857.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.9311054348945618,
+      "epoch": 1.2743972445464982,
+      "grad_norm": 0.08701436221599579,
+      "learning_rate": 0.00012112259970457902,
+      "loss": 0.9666108131408692,
+      "mean_token_accuracy": 0.7752738267183303,
+      "num_tokens": 13357039.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.9256260149180889,
+      "epoch": 1.27822426329889,
+      "grad_norm": 0.08751461654901505,
+      "learning_rate": 0.00012085403518195246,
+      "loss": 0.9926286697387695,
+      "mean_token_accuracy": 0.7750931903719902,
+      "num_tokens": 13397058.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0074332721531392,
+      "epoch": 1.282051282051282,
+      "grad_norm": 0.07409587502479553,
+      "learning_rate": 0.00012058547065932591,
+      "loss": 1.062586498260498,
+      "mean_token_accuracy": 0.7546869352459907,
+      "num_tokens": 13441381.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.9596263833343983,
+      "epoch": 1.2858783008036738,
+      "grad_norm": 0.09343665838241577,
+      "learning_rate": 0.00012031690613669934,
+      "loss": 1.0023324012756347,
+      "mean_token_accuracy": 0.7719831839203835,
+      "num_tokens": 13481914.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.9313522674143314,
+      "epoch": 1.2897053195560657,
+      "grad_norm": 0.0879049226641655,
+      "learning_rate": 0.00012004834161407279,
+      "loss": 0.9833806991577149,
+      "mean_token_accuracy": 0.7737741976976394,
+      "num_tokens": 13519831.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8369917057454586,
+      "epoch": 1.2935323383084576,
+      "grad_norm": 0.14339204132556915,
+      "learning_rate": 0.00011977977709144624,
+      "loss": 0.9147489547729493,
+      "mean_token_accuracy": 0.7984762340784073,
+      "num_tokens": 13559768.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.9055653363466263,
+      "epoch": 1.2973593570608495,
+      "grad_norm": 0.1441742479801178,
+      "learning_rate": 0.00011951121256881967,
+      "loss": 0.9521515846252442,
+      "mean_token_accuracy": 0.7834478095173836,
+      "num_tokens": 13595966.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.9677796266973019,
+      "epoch": 1.3011863758132414,
+      "grad_norm": 0.11233013868331909,
+      "learning_rate": 0.00011924264804619309,
+      "loss": 1.0522055625915527,
+      "mean_token_accuracy": 0.7664702609181404,
+      "num_tokens": 13638463.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.9398517791181803,
+      "epoch": 1.3050133945656333,
+      "grad_norm": 0.088468998670578,
+      "learning_rate": 0.00011897408352356654,
+      "loss": 0.9618704795837403,
+      "mean_token_accuracy": 0.7755557060241699,
+      "num_tokens": 13677769.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8900398269295693,
+      "epoch": 1.3088404133180251,
+      "grad_norm": 0.09742283076047897,
+      "learning_rate": 0.00011870551900093999,
+      "loss": 0.9422917366027832,
+      "mean_token_accuracy": 0.7865706130862236,
+      "num_tokens": 13713374.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.9008657015860081,
+      "epoch": 1.312667432070417,
+      "grad_norm": 0.09111864864826202,
+      "learning_rate": 0.00011843695447831342,
+      "loss": 0.9726786613464355,
+      "mean_token_accuracy": 0.7835188135504723,
+      "num_tokens": 13753165.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.954158465564251,
+      "epoch": 1.316494450822809,
+      "grad_norm": 0.0949985608458519,
+      "learning_rate": 0.00011816838995568687,
+      "loss": 1.0072153091430665,
+      "mean_token_accuracy": 0.7668681025505066,
+      "num_tokens": 13790265.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.9259054005146027,
+      "epoch": 1.3203214695752008,
+      "grad_norm": 0.09144506603479385,
+      "learning_rate": 0.00011789982543306031,
+      "loss": 1.0319811820983886,
+      "mean_token_accuracy": 0.77575224339962,
+      "num_tokens": 13830720.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.9554400585591794,
+      "epoch": 1.3241484883275927,
+      "grad_norm": 0.05986972153186798,
+      "learning_rate": 0.00011763126091043373,
+      "loss": 0.9840157508850098,
+      "mean_token_accuracy": 0.7714304268360138,
+      "num_tokens": 13874024.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.9618137650191784,
+      "epoch": 1.3279755070799846,
+      "grad_norm": 0.08746087551116943,
+      "learning_rate": 0.00011736269638780717,
+      "loss": 1.0280908584594726,
+      "mean_token_accuracy": 0.7679046332836151,
+      "num_tokens": 13916099.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.02601458132267,
+      "epoch": 1.3318025258323765,
+      "grad_norm": 0.09883694350719452,
+      "learning_rate": 0.00011709413186518061,
+      "loss": 1.0893220901489258,
+      "mean_token_accuracy": 0.7487106472253799,
+      "num_tokens": 13955163.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.025067638605833,
+      "epoch": 1.3356295445847683,
+      "grad_norm": 0.07656730711460114,
+      "learning_rate": 0.00011682556734255406,
+      "loss": 1.0527194023132325,
+      "mean_token_accuracy": 0.7569629296660423,
+      "num_tokens": 13996990.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8709930831566453,
+      "epoch": 1.3394565633371602,
+      "grad_norm": 0.1119026467204094,
+      "learning_rate": 0.0001165570028199275,
+      "loss": 0.9183405876159668,
+      "mean_token_accuracy": 0.784464044868946,
+      "num_tokens": 14040315.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.9783565014600754,
+      "epoch": 1.3432835820895521,
+      "grad_norm": 0.09997576475143433,
+      "learning_rate": 0.00011628843829730094,
+      "loss": 1.0318940162658692,
+      "mean_token_accuracy": 0.7614112690091133,
+      "num_tokens": 14083204.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.9975252889096737,
+      "epoch": 1.347110600841944,
+      "grad_norm": 0.10046812891960144,
+      "learning_rate": 0.00011601987377467437,
+      "loss": 1.0214290618896484,
+      "mean_token_accuracy": 0.7584437146782875,
+      "num_tokens": 14127039.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8959422588348389,
+      "epoch": 1.350937619594336,
+      "grad_norm": 0.09512703120708466,
+      "learning_rate": 0.0001157513092520478,
+      "loss": 0.9528075218200683,
+      "mean_token_accuracy": 0.7823959946632385,
+      "num_tokens": 14163989.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8903120748698712,
+      "epoch": 1.3547646383467278,
+      "grad_norm": 0.10500185191631317,
+      "learning_rate": 0.00011548274472942124,
+      "loss": 0.9784683227539063,
+      "mean_token_accuracy": 0.7854589730501175,
+      "num_tokens": 14198562.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.8580869071185588,
+      "epoch": 1.3585916570991197,
+      "grad_norm": 0.08716659992933273,
+      "learning_rate": 0.00011521418020679469,
+      "loss": 0.9078399658203125,
+      "mean_token_accuracy": 0.7894850671291351,
+      "num_tokens": 14236952.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.9841447554528713,
+      "epoch": 1.3624186758515116,
+      "grad_norm": 0.08638570457696915,
+      "learning_rate": 0.00011494561568416812,
+      "loss": 1.0438207626342773,
+      "mean_token_accuracy": 0.7629329964518548,
+      "num_tokens": 14278208.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9100395441055298,
+      "epoch": 1.3662456946039034,
+      "grad_norm": 0.09058145433664322,
+      "learning_rate": 0.00011467705116154157,
+      "loss": 0.9560261726379394,
+      "mean_token_accuracy": 0.7807327762246132,
+      "num_tokens": 14314076.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.8529263667762279,
+      "epoch": 1.3700727133562953,
+      "grad_norm": 0.08847236633300781,
+      "learning_rate": 0.00011440848663891502,
+      "loss": 0.9192025184631347,
+      "mean_token_accuracy": 0.7945622354745865,
+      "num_tokens": 14349740.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8977530397474766,
+      "epoch": 1.3738997321086872,
+      "grad_norm": 0.09535886347293854,
+      "learning_rate": 0.00011413992211628844,
+      "loss": 0.9331538200378418,
+      "mean_token_accuracy": 0.7803975984454155,
+      "num_tokens": 14392492.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0430821359157563,
+      "epoch": 1.377726750861079,
+      "grad_norm": 0.08564139902591705,
+      "learning_rate": 0.00011387135759366187,
+      "loss": 1.0767670631408692,
+      "mean_token_accuracy": 0.7479040876030922,
+      "num_tokens": 14436961.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8358541168272495,
+      "epoch": 1.381553769613471,
+      "grad_norm": 0.09847365319728851,
+      "learning_rate": 0.00011360279307103532,
+      "loss": 0.8758580207824707,
+      "mean_token_accuracy": 0.7964837267994881,
+      "num_tokens": 14472251.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8302674755454064,
+      "epoch": 1.3853807883658629,
+      "grad_norm": 0.08570406585931778,
+      "learning_rate": 0.00011333422854840876,
+      "loss": 0.9068514823913574,
+      "mean_token_accuracy": 0.7943103745579719,
+      "num_tokens": 14509818.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.9825982883572578,
+      "epoch": 1.3892078071182548,
+      "grad_norm": 0.10844281315803528,
+      "learning_rate": 0.0001130656640257822,
+      "loss": 1.0484787940979003,
+      "mean_token_accuracy": 0.7600376740097999,
+      "num_tokens": 14553567.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.0431513242423534,
+      "epoch": 1.3930348258706466,
+      "grad_norm": 0.0750717744231224,
+      "learning_rate": 0.00011279709950315564,
+      "loss": 1.0337225914001464,
+      "mean_token_accuracy": 0.7504511162638664,
+      "num_tokens": 14598239.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9319969929754734,
+      "epoch": 1.3968618446230385,
+      "grad_norm": 0.08307385444641113,
+      "learning_rate": 0.00011252853498052909,
+      "loss": 0.9771868705749511,
+      "mean_token_accuracy": 0.7778135031461716,
+      "num_tokens": 14638064.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.9992426164448261,
+      "epoch": 1.4006888633754304,
+      "grad_norm": 0.09222020208835602,
+      "learning_rate": 0.00011225997045790251,
+      "loss": 1.0516475677490233,
+      "mean_token_accuracy": 0.7587143570184708,
+      "num_tokens": 14682012.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.9670721650123596,
+      "epoch": 1.4045158821278223,
+      "grad_norm": 0.09432315081357956,
+      "learning_rate": 0.00011199140593527595,
+      "loss": 1.0164658546447753,
+      "mean_token_accuracy": 0.7670722231268883,
+      "num_tokens": 14722922.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.9808389253914356,
+      "epoch": 1.4083429008802142,
+      "grad_norm": 0.08502112329006195,
+      "learning_rate": 0.00011172284141264939,
+      "loss": 1.0553858757019043,
+      "mean_token_accuracy": 0.76065753698349,
+      "num_tokens": 14765083.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.011240091174841,
+      "epoch": 1.412169919632606,
+      "grad_norm": 0.07948844134807587,
+      "learning_rate": 0.00011145427689002284,
+      "loss": 1.0446209907531738,
+      "mean_token_accuracy": 0.75536377876997,
+      "num_tokens": 14806465.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.911352240294218,
+      "epoch": 1.415996938384998,
+      "grad_norm": 0.08382374793291092,
+      "learning_rate": 0.00011118571236739627,
+      "loss": 0.9388965606689453,
+      "mean_token_accuracy": 0.7807439729571343,
+      "num_tokens": 14850133.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.9055514119565486,
+      "epoch": 1.4198239571373898,
+      "grad_norm": 0.10713934898376465,
+      "learning_rate": 0.00011091714784476972,
+      "loss": 0.9727254867553711,
+      "mean_token_accuracy": 0.7801795959472656,
+      "num_tokens": 14887327.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.9338000696152449,
+      "epoch": 1.4236509758897817,
+      "grad_norm": 0.11418487876653671,
+      "learning_rate": 0.00011064858332214314,
+      "loss": 0.9989487648010253,
+      "mean_token_accuracy": 0.7747065275907516,
+      "num_tokens": 14927730.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.869029226526618,
+      "epoch": 1.4274779946421736,
+      "grad_norm": 0.10778038948774338,
+      "learning_rate": 0.00011038001879951659,
+      "loss": 0.9393071174621582,
+      "mean_token_accuracy": 0.7909289851784707,
+      "num_tokens": 14964847.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.8993408516049385,
+      "epoch": 1.4313050133945655,
+      "grad_norm": 0.08339972048997879,
+      "learning_rate": 0.00011011145427689002,
+      "loss": 0.9511364936828614,
+      "mean_token_accuracy": 0.7844893127679825,
+      "num_tokens": 15003449.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.9478372372686863,
+      "epoch": 1.4351320321469574,
+      "grad_norm": 0.07547847181558609,
+      "learning_rate": 0.00010984288975426347,
+      "loss": 0.9942925453186036,
+      "mean_token_accuracy": 0.772410535812378,
+      "num_tokens": 15046091.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.8367562972009182,
+      "epoch": 1.4389590508993493,
+      "grad_norm": 0.06902482360601425,
+      "learning_rate": 0.00010957432523163691,
+      "loss": 0.8951096534729004,
+      "mean_token_accuracy": 0.7985799089074135,
+      "num_tokens": 15091826.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.9437298484146595,
+      "epoch": 1.4427860696517412,
+      "grad_norm": 0.10231524705886841,
+      "learning_rate": 0.00010930576070901035,
+      "loss": 0.9919009208679199,
+      "mean_token_accuracy": 0.7663119360804558,
+      "num_tokens": 15133719.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.0057852260768414,
+      "epoch": 1.446613088404133,
+      "grad_norm": 0.09349844604730606,
+      "learning_rate": 0.0001090371961863838,
+      "loss": 1.0667811393737794,
+      "mean_token_accuracy": 0.757930365204811,
+      "num_tokens": 15173670.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.9152357578277588,
+      "epoch": 1.450440107156525,
+      "grad_norm": 0.09612533450126648,
+      "learning_rate": 0.00010876863166375722,
+      "loss": 0.9641363143920898,
+      "mean_token_accuracy": 0.7791497871279717,
+      "num_tokens": 15215154.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.849637558311224,
+      "epoch": 1.4542671259089168,
+      "grad_norm": 0.07079404592514038,
+      "learning_rate": 0.00010850006714113066,
+      "loss": 0.8924535751342774,
+      "mean_token_accuracy": 0.7958060145378113,
+      "num_tokens": 15261773.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.9689324770122767,
+      "epoch": 1.4580941446613087,
+      "grad_norm": 0.10107272863388062,
+      "learning_rate": 0.0001082315026185041,
+      "loss": 1.000623607635498,
+      "mean_token_accuracy": 0.7690365821123123,
+      "num_tokens": 15295693.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.8926774315536022,
+      "epoch": 1.4619211634137006,
+      "grad_norm": 0.0883372351527214,
+      "learning_rate": 0.00010796293809587754,
+      "loss": 0.9312380790710449,
+      "mean_token_accuracy": 0.7839185446500778,
+      "num_tokens": 15332324.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.9962236389517785,
+      "epoch": 1.4657481821660925,
+      "grad_norm": 0.09174945950508118,
+      "learning_rate": 0.00010769437357325099,
+      "loss": 1.0419865608215333,
+      "mean_token_accuracy": 0.7592507138848305,
+      "num_tokens": 15370812.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0249286435544491,
+      "epoch": 1.4695752009184844,
+      "grad_norm": 0.07152284681797028,
+      "learning_rate": 0.00010742580905062442,
+      "loss": 1.0437363624572753,
+      "mean_token_accuracy": 0.7567671984434128,
+      "num_tokens": 15417719.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.903605168312788,
+      "epoch": 1.4734022196708763,
+      "grad_norm": 0.09400783479213715,
+      "learning_rate": 0.00010715724452799784,
+      "loss": 0.9410040855407715,
+      "mean_token_accuracy": 0.7839412048459053,
+      "num_tokens": 15455856.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.0259956195950508,
+      "epoch": 1.4772292384232681,
+      "grad_norm": 0.08671914041042328,
+      "learning_rate": 0.00010688868000537129,
+      "loss": 1.1025453567504884,
+      "mean_token_accuracy": 0.7507242172956466,
+      "num_tokens": 15492109.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.9178053669631481,
+      "epoch": 1.48105625717566,
+      "grad_norm": 0.07717446982860565,
+      "learning_rate": 0.00010662011548274474,
+      "loss": 0.96353178024292,
+      "mean_token_accuracy": 0.7797438561916351,
+      "num_tokens": 15532130.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.9423278756439686,
+      "epoch": 1.484883275928052,
+      "grad_norm": 0.11039029061794281,
+      "learning_rate": 0.00010635155096011817,
+      "loss": 0.979669189453125,
+      "mean_token_accuracy": 0.7755513936281204,
+      "num_tokens": 15575609.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.8999218411743641,
+      "epoch": 1.4887102946804438,
+      "grad_norm": 0.08974706381559372,
+      "learning_rate": 0.00010608298643749162,
+      "loss": 0.9477033615112305,
+      "mean_token_accuracy": 0.7822227850556374,
+      "num_tokens": 15621264.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.8756623603403568,
+      "epoch": 1.4925373134328357,
+      "grad_norm": 0.10864510387182236,
+      "learning_rate": 0.00010581442191486505,
+      "loss": 0.9711783409118653,
+      "mean_token_accuracy": 0.7893951386213303,
+      "num_tokens": 15656959.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.951158057898283,
+      "epoch": 1.4963643321852276,
+      "grad_norm": 0.09398993104696274,
+      "learning_rate": 0.0001055458573922385,
+      "loss": 1.0387070655822754,
+      "mean_token_accuracy": 0.7698590591549873,
+      "num_tokens": 15700293.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.9240442231297493,
+      "epoch": 1.5001913509376195,
+      "grad_norm": 0.09761729091405869,
+      "learning_rate": 0.00010527729286961192,
+      "loss": 0.9758125305175781,
+      "mean_token_accuracy": 0.7737968236207962,
+      "num_tokens": 15739304.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.9025500647723674,
+      "epoch": 1.5040183696900113,
+      "grad_norm": 0.08816131204366684,
+      "learning_rate": 0.00010500872834698537,
+      "loss": 0.913144302368164,
+      "mean_token_accuracy": 0.7775477200746537,
+      "num_tokens": 15785086.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.8958883471786976,
+      "epoch": 1.5078453884424032,
+      "grad_norm": 0.09690563380718231,
+      "learning_rate": 0.0001047401638243588,
+      "loss": 0.9484706878662109,
+      "mean_token_accuracy": 0.7867727875709534,
+      "num_tokens": 15822631.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.8738761503249407,
+      "epoch": 1.5116724071947951,
+      "grad_norm": 0.08325833082199097,
+      "learning_rate": 0.00010447159930173225,
+      "loss": 0.9258977890014648,
+      "mean_token_accuracy": 0.7862061053514481,
+      "num_tokens": 15863533.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.952784775942564,
+      "epoch": 1.515499425947187,
+      "grad_norm": 0.09089304506778717,
+      "learning_rate": 0.0001042030347791057,
+      "loss": 0.9893428802490234,
+      "mean_token_accuracy": 0.769037912786007,
+      "num_tokens": 15903798.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.9974973328411579,
+      "epoch": 1.519326444699579,
+      "grad_norm": 0.06594393402338028,
+      "learning_rate": 0.00010393447025647913,
+      "loss": 0.9982621192932128,
+      "mean_token_accuracy": 0.7653156638145446,
+      "num_tokens": 15947894.0,
+      "step": 3970
+    },
+    {
+      "entropy": 1.042479208856821,
+      "epoch": 1.5231534634519708,
+      "grad_norm": 0.09250905364751816,
+      "learning_rate": 0.00010366590573385255,
+      "loss": 1.0862640380859374,
+      "mean_token_accuracy": 0.7515693128108978,
+      "num_tokens": 15985609.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.869631578028202,
+      "epoch": 1.5269804822043627,
+      "grad_norm": 0.10154584795236588,
+      "learning_rate": 0.000103397341211226,
+      "loss": 0.9275701522827149,
+      "mean_token_accuracy": 0.7910413116216659,
+      "num_tokens": 16022339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9228729590773582,
+      "epoch": 1.5308075009567546,
+      "grad_norm": 0.08860265463590622,
+      "learning_rate": 0.00010312877668859944,
+      "loss": 1.0074289321899415,
+      "mean_token_accuracy": 0.7770419105887413,
+      "num_tokens": 16063778.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9469372771680356,
+      "epoch": 1.5346345197091464,
+      "grad_norm": 0.08613952249288559,
+      "learning_rate": 0.00010286021216597287,
+      "loss": 1.0328418731689453,
+      "mean_token_accuracy": 0.7784339845180511,
+      "num_tokens": 16103389.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9240258730947971,
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.09255630522966385,
+      "learning_rate": 0.00010259164764334632,
+      "loss": 0.9813838958740234,
+      "mean_token_accuracy": 0.779928731918335,
+      "num_tokens": 16141739.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.8300335463136435,
+      "epoch": 1.5422885572139302,
+      "grad_norm": 0.11173315346240997,
+      "learning_rate": 0.00010232308312071977,
+      "loss": 0.8650222778320312,
+      "mean_token_accuracy": 0.8004546627402306,
+      "num_tokens": 16179442.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.970530441403389,
+      "epoch": 1.546115575966322,
+      "grad_norm": 0.08758437633514404,
+      "learning_rate": 0.0001020545185980932,
+      "loss": 1.029263401031494,
+      "mean_token_accuracy": 0.7669389978051185,
+      "num_tokens": 16220502.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.8929917253553867,
+      "epoch": 1.549942594718714,
+      "grad_norm": 0.0840209424495697,
+      "learning_rate": 0.00010178595407546662,
+      "loss": 0.9574555397033692,
+      "mean_token_accuracy": 0.7882895812392234,
+      "num_tokens": 16263944.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9571633011102676,
+      "epoch": 1.5537696134711059,
+      "grad_norm": 0.07731885462999344,
+      "learning_rate": 0.00010151738955284007,
+      "loss": 1.014600658416748,
+      "mean_token_accuracy": 0.7691228404641152,
+      "num_tokens": 16307508.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9627384431660175,
+      "epoch": 1.5575966322234978,
+      "grad_norm": 0.09968744218349457,
+      "learning_rate": 0.00010124882503021352,
+      "loss": 1.0220794677734375,
+      "mean_token_accuracy": 0.7685489565134048,
+      "num_tokens": 16349178.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.8696753971278668,
+      "epoch": 1.5614236509758896,
+      "grad_norm": 0.08411276340484619,
+      "learning_rate": 0.00010098026050758695,
+      "loss": 0.9325771331787109,
+      "mean_token_accuracy": 0.7903442814946174,
+      "num_tokens": 16390375.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.8790203854441643,
+      "epoch": 1.5652506697282815,
+      "grad_norm": 0.0969686210155487,
+      "learning_rate": 0.0001007116959849604,
+      "loss": 0.9325167655944824,
+      "mean_token_accuracy": 0.7890133559703827,
+      "num_tokens": 16429198.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9447548128664494,
+      "epoch": 1.5690776884806734,
+      "grad_norm": 0.07992373406887054,
+      "learning_rate": 0.00010044313146233384,
+      "loss": 0.9708291053771972,
+      "mean_token_accuracy": 0.7737105548381805,
+      "num_tokens": 16472336.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.974559249728918,
+      "epoch": 1.5729047072330653,
+      "grad_norm": 0.09685226529836655,
+      "learning_rate": 0.00010017456693970726,
+      "loss": 1.0289334297180175,
+      "mean_token_accuracy": 0.7674296617507934,
+      "num_tokens": 16511109.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.8575489681214095,
+      "epoch": 1.5767317259854572,
+      "grad_norm": 0.09298260509967804,
+      "learning_rate": 9.990600241708071e-05,
+      "loss": 0.8897696495056152,
+      "mean_token_accuracy": 0.7952411189675331,
+      "num_tokens": 16552802.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.869475956633687,
+      "epoch": 1.580558744737849,
+      "grad_norm": 0.129170760512352,
+      "learning_rate": 9.963743789445414e-05,
+      "loss": 0.9408356666564941,
+      "mean_token_accuracy": 0.7868246123194694,
+      "num_tokens": 16592603.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9167623318731785,
+      "epoch": 1.584385763490241,
+      "grad_norm": 0.08131655305624008,
+      "learning_rate": 9.936887337182759e-05,
+      "loss": 1.005775260925293,
+      "mean_token_accuracy": 0.7779423877596855,
+      "num_tokens": 16633674.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9069061763584614,
+      "epoch": 1.5882127822426328,
+      "grad_norm": 0.07485036551952362,
+      "learning_rate": 9.910030884920103e-05,
+      "loss": 0.9540878295898437,
+      "mean_token_accuracy": 0.7809736356139183,
+      "num_tokens": 16669966.0,
+      "step": 4150
+    },
+    {
+      "entropy": 1.0095594763755797,
+      "epoch": 1.5920398009950247,
+      "grad_norm": 0.11678522825241089,
+      "learning_rate": 9.883174432657446e-05,
+      "loss": 1.0742655754089356,
+      "mean_token_accuracy": 0.7636483564972878,
+      "num_tokens": 16711538.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.8342153321951628,
+      "epoch": 1.5958668197474166,
+      "grad_norm": 0.09654127061367035,
+      "learning_rate": 9.85631798039479e-05,
+      "loss": 0.8637946128845215,
+      "mean_token_accuracy": 0.7977021634578705,
+      "num_tokens": 16746947.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9147222273051738,
+      "epoch": 1.5996938384998085,
+      "grad_norm": 0.10032576322555542,
+      "learning_rate": 9.829461528132134e-05,
+      "loss": 0.9848580360412598,
+      "mean_token_accuracy": 0.7794791385531425,
+      "num_tokens": 16792089.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9350447114557028,
+      "epoch": 1.6035208572522004,
+      "grad_norm": 0.11322317272424698,
+      "learning_rate": 9.802605075869477e-05,
+      "loss": 0.9632351875305176,
+      "mean_token_accuracy": 0.7710213780403137,
+      "num_tokens": 16831782.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.8924577154219151,
+      "epoch": 1.6073478760045923,
+      "grad_norm": 0.08842343091964722,
+      "learning_rate": 9.775748623606822e-05,
+      "loss": 0.9661048889160156,
+      "mean_token_accuracy": 0.7863042891025543,
+      "num_tokens": 16867851.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9452814936637879,
+      "epoch": 1.6111748947569842,
+      "grad_norm": 0.10469862073659897,
+      "learning_rate": 9.748892171344167e-05,
+      "loss": 1.0315632820129395,
+      "mean_token_accuracy": 0.769272243976593,
+      "num_tokens": 16909819.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.8794655621051788,
+      "epoch": 1.615001913509376,
+      "grad_norm": 0.08528223633766174,
+      "learning_rate": 9.72203571908151e-05,
+      "loss": 0.9158189773559571,
+      "mean_token_accuracy": 0.791112196445465,
+      "num_tokens": 16945241.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.9216304633766412,
+      "epoch": 1.618828932261768,
+      "grad_norm": 0.07684458047151566,
+      "learning_rate": 9.695179266818853e-05,
+      "loss": 1.0047569274902344,
+      "mean_token_accuracy": 0.7764274105429649,
+      "num_tokens": 16986516.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.8806056842207909,
+      "epoch": 1.6226559510141598,
+      "grad_norm": 0.09925177693367004,
+      "learning_rate": 9.668322814556198e-05,
+      "loss": 0.9321705818176269,
+      "mean_token_accuracy": 0.7873435765504837,
+      "num_tokens": 17026974.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0260133132338525,
+      "epoch": 1.6264829697665517,
+      "grad_norm": 0.07781514525413513,
+      "learning_rate": 9.641466362293541e-05,
+      "loss": 1.0732348442077637,
+      "mean_token_accuracy": 0.755302457511425,
+      "num_tokens": 17063628.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.8771878894418478,
+      "epoch": 1.6303099885189436,
+      "grad_norm": 0.12377400696277618,
+      "learning_rate": 9.614609910030885e-05,
+      "loss": 0.9051324844360351,
+      "mean_token_accuracy": 0.7877693608403206,
+      "num_tokens": 17102243.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9575911372900009,
+      "epoch": 1.6341370072713355,
+      "grad_norm": 0.07953961193561554,
+      "learning_rate": 9.58775345776823e-05,
+      "loss": 1.0206258773803711,
+      "mean_token_accuracy": 0.770101509988308,
+      "num_tokens": 17143256.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9909125387668609,
+      "epoch": 1.6379640260237274,
+      "grad_norm": 0.09304741024971008,
+      "learning_rate": 9.560897005505573e-05,
+      "loss": 1.043109130859375,
+      "mean_token_accuracy": 0.7598145559430123,
+      "num_tokens": 17188878.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.8626054737716913,
+      "epoch": 1.6417910447761193,
+      "grad_norm": 0.08982561528682709,
+      "learning_rate": 9.534040553242916e-05,
+      "loss": 0.9062054634094239,
+      "mean_token_accuracy": 0.790125061571598,
+      "num_tokens": 17224537.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.919727610051632,
+      "epoch": 1.6456180635285111,
+      "grad_norm": 0.11226653307676315,
+      "learning_rate": 9.507184100980261e-05,
+      "loss": 0.970013427734375,
+      "mean_token_accuracy": 0.7747739493846894,
+      "num_tokens": 17262347.0,
+      "step": 4300
+    },
+    {
+      "entropy": 1.032866196334362,
+      "epoch": 1.649445082280903,
+      "grad_norm": 0.09440238773822784,
+      "learning_rate": 9.480327648717606e-05,
+      "loss": 1.0287545204162598,
+      "mean_token_accuracy": 0.7550160124897957,
+      "num_tokens": 17307578.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.907962580025196,
+      "epoch": 1.653272101033295,
+      "grad_norm": 0.11395370960235596,
+      "learning_rate": 9.453471196454948e-05,
+      "loss": 0.9705679893493653,
+      "mean_token_accuracy": 0.7807198286056518,
+      "num_tokens": 17342943.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.8495472550392151,
+      "epoch": 1.6570991197856868,
+      "grad_norm": 0.07685171812772751,
+      "learning_rate": 9.426614744192292e-05,
+      "loss": 0.9079866409301758,
+      "mean_token_accuracy": 0.7923004642128945,
+      "num_tokens": 17378158.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.8389323726296425,
+      "epoch": 1.6609261385380787,
+      "grad_norm": 0.09541229903697968,
+      "learning_rate": 9.399758291929637e-05,
+      "loss": 0.9092423439025878,
+      "mean_token_accuracy": 0.7946408927440644,
+      "num_tokens": 17412703.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.9035130314528942,
+      "epoch": 1.6647531572904706,
+      "grad_norm": 0.08291888236999512,
+      "learning_rate": 9.37290183966698e-05,
+      "loss": 0.9255120277404785,
+      "mean_token_accuracy": 0.7840688213706016,
+      "num_tokens": 17456250.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.8917031817138195,
+      "epoch": 1.6685801760428625,
+      "grad_norm": 0.08787538856267929,
+      "learning_rate": 9.346045387404324e-05,
+      "loss": 0.9318277359008789,
+      "mean_token_accuracy": 0.7854569494724274,
+      "num_tokens": 17492566.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.8860244527459145,
+      "epoch": 1.6724071947952543,
+      "grad_norm": 0.10287550836801529,
+      "learning_rate": 9.319188935141668e-05,
+      "loss": 0.9169553756713867,
+      "mean_token_accuracy": 0.7801365301012992,
+      "num_tokens": 17530267.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.8470614090561867,
+      "epoch": 1.6762342135476462,
+      "grad_norm": 0.13052308559417725,
+      "learning_rate": 9.292332482879013e-05,
+      "loss": 0.9004100799560547,
+      "mean_token_accuracy": 0.791596457362175,
+      "num_tokens": 17566336.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9627884522080421,
+      "epoch": 1.6800612323000381,
+      "grad_norm": 0.09305555373430252,
+      "learning_rate": 9.265476030616355e-05,
+      "loss": 0.9837147712707519,
+      "mean_token_accuracy": 0.7687505498528481,
+      "num_tokens": 17609294.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9614691123366356,
+      "epoch": 1.68388825105243,
+      "grad_norm": 0.08118042349815369,
+      "learning_rate": 9.2386195783537e-05,
+      "loss": 1.0093948364257812,
+      "mean_token_accuracy": 0.7686966329813003,
+      "num_tokens": 17653408.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.8255576498806476,
+      "epoch": 1.687715269804822,
+      "grad_norm": 0.07197146117687225,
+      "learning_rate": 9.211763126091045e-05,
+      "loss": 0.9013225555419921,
+      "mean_token_accuracy": 0.7994248151779175,
+      "num_tokens": 17693303.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9197361193597317,
+      "epoch": 1.6915422885572138,
+      "grad_norm": 0.10147208720445633,
+      "learning_rate": 9.184906673828388e-05,
+      "loss": 0.966912841796875,
+      "mean_token_accuracy": 0.774210800230503,
+      "num_tokens": 17734446.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.8828513637185097,
+      "epoch": 1.6953693073096057,
+      "grad_norm": 0.08126919716596603,
+      "learning_rate": 9.158050221565731e-05,
+      "loss": 0.9237348556518554,
+      "mean_token_accuracy": 0.788974218070507,
+      "num_tokens": 17776056.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.8538446951657533,
+      "epoch": 1.6991963260619976,
+      "grad_norm": 0.08602278679609299,
+      "learning_rate": 9.131193769303076e-05,
+      "loss": 0.9384878158569336,
+      "mean_token_accuracy": 0.7924654617905617,
+      "num_tokens": 17814560.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9160130321979523,
+      "epoch": 1.7030233448143894,
+      "grad_norm": 0.10127890110015869,
+      "learning_rate": 9.10433731704042e-05,
+      "loss": 0.9924029350280762,
+      "mean_token_accuracy": 0.7764815479516983,
+      "num_tokens": 17852872.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.8855723738670349,
+      "epoch": 1.7068503635667813,
+      "grad_norm": 0.09295201301574707,
+      "learning_rate": 9.077480864777763e-05,
+      "loss": 0.9131739616394043,
+      "mean_token_accuracy": 0.7876585990190506,
+      "num_tokens": 17893403.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.8825645297765732,
+      "epoch": 1.7106773823191732,
+      "grad_norm": 0.1038793995976448,
+      "learning_rate": 9.050624412515107e-05,
+      "loss": 0.9621119499206543,
+      "mean_token_accuracy": 0.7840767920017242,
+      "num_tokens": 17933069.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.9438045337796211,
+      "epoch": 1.714504401071565,
+      "grad_norm": 0.08998332172632217,
+      "learning_rate": 9.023767960252452e-05,
+      "loss": 1.0042546272277832,
+      "mean_token_accuracy": 0.7710625112056733,
+      "num_tokens": 17978081.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.9605814971029758,
+      "epoch": 1.718331419823957,
+      "grad_norm": 0.0936085507273674,
+      "learning_rate": 8.996911507989794e-05,
+      "loss": 1.0731863021850585,
+      "mean_token_accuracy": 0.7675610318779945,
+      "num_tokens": 18026355.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.9412197135388851,
+      "epoch": 1.7221584385763489,
+      "grad_norm": 0.11693151295185089,
+      "learning_rate": 8.970055055727139e-05,
+      "loss": 1.0271482467651367,
+      "mean_token_accuracy": 0.7749442532658577,
+      "num_tokens": 18064648.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.9840309470891953,
+      "epoch": 1.7259854573287408,
+      "grad_norm": 0.07721691578626633,
+      "learning_rate": 8.943198603464484e-05,
+      "loss": 0.9978925704956054,
+      "mean_token_accuracy": 0.7662706628441811,
+      "num_tokens": 18104352.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.9122109733521938,
+      "epoch": 1.7298124760811326,
+      "grad_norm": 0.10790548473596573,
+      "learning_rate": 8.916342151201827e-05,
+      "loss": 0.9912397384643554,
+      "mean_token_accuracy": 0.774566973745823,
+      "num_tokens": 18145632.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.8214024558663369,
+      "epoch": 1.7336394948335245,
+      "grad_norm": 0.0873790979385376,
+      "learning_rate": 8.88948569893917e-05,
+      "loss": 0.9174188613891602,
+      "mean_token_accuracy": 0.797317324578762,
+      "num_tokens": 18182216.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.8851194910705089,
+      "epoch": 1.7374665135859164,
+      "grad_norm": 0.08441472053527832,
+      "learning_rate": 8.862629246676515e-05,
+      "loss": 0.9345614433288574,
+      "mean_token_accuracy": 0.7909206628799439,
+      "num_tokens": 18220152.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.9045546390116215,
+      "epoch": 1.7412935323383083,
+      "grad_norm": 0.09491857141256332,
+      "learning_rate": 8.835772794413858e-05,
+      "loss": 1.0261774063110352,
+      "mean_token_accuracy": 0.7792002618312835,
+      "num_tokens": 18253615.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.8957971200346947,
+      "epoch": 1.7451205510907002,
+      "grad_norm": 0.07239943742752075,
+      "learning_rate": 8.808916342151202e-05,
+      "loss": 0.9220120429992675,
+      "mean_token_accuracy": 0.7812404081225395,
+      "num_tokens": 18292565.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.9733762003481388,
+      "epoch": 1.748947569843092,
+      "grad_norm": 0.07816951721906662,
+      "learning_rate": 8.782059889888546e-05,
+      "loss": 1.0176166534423827,
+      "mean_token_accuracy": 0.7636090680956841,
+      "num_tokens": 18337951.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.9952755816280842,
+      "epoch": 1.752774588595484,
+      "grad_norm": 0.09595679491758347,
+      "learning_rate": 8.75520343762589e-05,
+      "loss": 1.0484466552734375,
+      "mean_token_accuracy": 0.7624279737472535,
+      "num_tokens": 18378541.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.9325974151492119,
+      "epoch": 1.7566016073478758,
+      "grad_norm": 0.1425638496875763,
+      "learning_rate": 8.728346985363234e-05,
+      "loss": 1.007568645477295,
+      "mean_token_accuracy": 0.7753438904881478,
+      "num_tokens": 18416147.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.8879670143127442,
+      "epoch": 1.7604286261002677,
+      "grad_norm": 0.08936052024364471,
+      "learning_rate": 8.701490533100578e-05,
+      "loss": 0.9574133872985839,
+      "mean_token_accuracy": 0.7878236457705498,
+      "num_tokens": 18452658.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.9596087213605642,
+      "epoch": 1.7642556448526596,
+      "grad_norm": 0.08222804218530655,
+      "learning_rate": 8.674634080837921e-05,
+      "loss": 1.0134518623352051,
+      "mean_token_accuracy": 0.7686690568923951,
+      "num_tokens": 18493806.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.9412514306604862,
+      "epoch": 1.7680826636050515,
+      "grad_norm": 0.08482176810503006,
+      "learning_rate": 8.647777628575266e-05,
+      "loss": 0.9830768585205079,
+      "mean_token_accuracy": 0.7789766594767571,
+      "num_tokens": 18538740.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.8279416210949421,
+      "epoch": 1.7719096823574434,
+      "grad_norm": 0.12101086974143982,
+      "learning_rate": 8.620921176312609e-05,
+      "loss": 0.8422709465026855,
+      "mean_token_accuracy": 0.8009032368659973,
+      "num_tokens": 18579991.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.8889543637633324,
+      "epoch": 1.7757367011098353,
+      "grad_norm": 0.09586559236049652,
+      "learning_rate": 8.594064724049954e-05,
+      "loss": 0.9579720497131348,
+      "mean_token_accuracy": 0.7857530102133751,
+      "num_tokens": 18616195.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.9021936893463135,
+      "epoch": 1.7795637198622272,
+      "grad_norm": 0.0920713022351265,
+      "learning_rate": 8.567208271787297e-05,
+      "loss": 0.9568814277648926,
+      "mean_token_accuracy": 0.7835437625646591,
+      "num_tokens": 18650396.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.9605553701519967,
+      "epoch": 1.783390738614619,
+      "grad_norm": 0.0752284824848175,
+      "learning_rate": 8.54035181952464e-05,
+      "loss": 1.0107332229614259,
+      "mean_token_accuracy": 0.7712536633014679,
+      "num_tokens": 18692519.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.8929145928472281,
+      "epoch": 1.787217757367011,
+      "grad_norm": 0.08124406635761261,
+      "learning_rate": 8.513495367261985e-05,
+      "loss": 0.9392594337463379,
+      "mean_token_accuracy": 0.7837390914559365,
+      "num_tokens": 18730865.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.8995866551995277,
+      "epoch": 1.7910447761194028,
+      "grad_norm": 0.07306879013776779,
+      "learning_rate": 8.486638914999329e-05,
+      "loss": 0.9512563705444336,
+      "mean_token_accuracy": 0.7803288042545319,
+      "num_tokens": 18774851.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.9283428456634283,
+      "epoch": 1.7948717948717947,
+      "grad_norm": 0.06833672523498535,
+      "learning_rate": 8.459782462736673e-05,
+      "loss": 0.9614426612854003,
+      "mean_token_accuracy": 0.7776324123144149,
+      "num_tokens": 18815273.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.8980611331760884,
+      "epoch": 1.7986988136241866,
+      "grad_norm": 0.09426148980855942,
+      "learning_rate": 8.432926010474017e-05,
+      "loss": 0.9397372245788574,
+      "mean_token_accuracy": 0.7818324938416481,
+      "num_tokens": 18854806.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.9534067753702402,
+      "epoch": 1.8025258323765785,
+      "grad_norm": 0.11984719336032867,
+      "learning_rate": 8.40606955821136e-05,
+      "loss": 1.0058012008666992,
+      "mean_token_accuracy": 0.7710662186145782,
+      "num_tokens": 18893820.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.9396863542497158,
+      "epoch": 1.8063528511289704,
+      "grad_norm": 0.1126495823264122,
+      "learning_rate": 8.379213105948705e-05,
+      "loss": 0.9968315124511719,
+      "mean_token_accuracy": 0.7748499393463135,
+      "num_tokens": 18932078.0,
+      "step": 4720
+    },
+    {
+      "entropy": 1.0531147465109825,
+      "epoch": 1.8101798698813623,
+      "grad_norm": 0.0951380655169487,
+      "learning_rate": 8.352356653686048e-05,
+      "loss": 1.1058255195617677,
+      "mean_token_accuracy": 0.7495945364236831,
+      "num_tokens": 18974602.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.8520326256752014,
+      "epoch": 1.8140068886337541,
+      "grad_norm": 0.08623862266540527,
+      "learning_rate": 8.325500201423391e-05,
+      "loss": 0.8825064659118652,
+      "mean_token_accuracy": 0.7940022364258766,
+      "num_tokens": 19014261.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.979587784409523,
+      "epoch": 1.817833907386146,
+      "grad_norm": 0.11787699162960052,
+      "learning_rate": 8.298643749160736e-05,
+      "loss": 1.070664405822754,
+      "mean_token_accuracy": 0.7620177045464516,
+      "num_tokens": 19058223.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.8753061652183532,
+      "epoch": 1.821660926138538,
+      "grad_norm": 0.130862757563591,
+      "learning_rate": 8.271787296898081e-05,
+      "loss": 0.9366108894348144,
+      "mean_token_accuracy": 0.7874863654375076,
+      "num_tokens": 19100204.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.8777839131653309,
+      "epoch": 1.8254879448909298,
+      "grad_norm": 0.09261229634284973,
+      "learning_rate": 8.244930844635424e-05,
+      "loss": 0.9104420661926269,
+      "mean_token_accuracy": 0.7895808070898056,
+      "num_tokens": 19141588.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.9170226149260998,
+      "epoch": 1.8293149636433217,
+      "grad_norm": 0.06741383671760559,
+      "learning_rate": 8.218074392372768e-05,
+      "loss": 0.9543824195861816,
+      "mean_token_accuracy": 0.7765088111162186,
+      "num_tokens": 19182818.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.8602717489004135,
+      "epoch": 1.8331419823957136,
+      "grad_norm": 0.12861686944961548,
+      "learning_rate": 8.191217940110112e-05,
+      "loss": 0.926014518737793,
+      "mean_token_accuracy": 0.7917162731289864,
+      "num_tokens": 19217919.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.9471398882567883,
+      "epoch": 1.8369690011481055,
+      "grad_norm": 0.0744423121213913,
+      "learning_rate": 8.164361487847456e-05,
+      "loss": 0.9777775764465332,
+      "mean_token_accuracy": 0.7716371163725853,
+      "num_tokens": 19263991.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.8363759070634842,
+      "epoch": 1.8407960199004973,
+      "grad_norm": 0.08627327531576157,
+      "learning_rate": 8.137505035584799e-05,
+      "loss": 0.887846565246582,
+      "mean_token_accuracy": 0.7938979223370553,
+      "num_tokens": 19305441.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.9200571574270725,
+      "epoch": 1.8446230386528892,
+      "grad_norm": 0.08358518034219742,
+      "learning_rate": 8.110648583322144e-05,
+      "loss": 0.9703543663024903,
+      "mean_token_accuracy": 0.777827826142311,
+      "num_tokens": 19342309.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.9295372806489468,
+      "epoch": 1.8484500574052811,
+      "grad_norm": 0.0970570370554924,
+      "learning_rate": 8.083792131059487e-05,
+      "loss": 0.9934672355651856,
+      "mean_token_accuracy": 0.7739364430308342,
+      "num_tokens": 19387707.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.9003355488181114,
+      "epoch": 1.852277076157673,
+      "grad_norm": 0.09357219189405441,
+      "learning_rate": 8.05693567879683e-05,
+      "loss": 0.9544237136840821,
+      "mean_token_accuracy": 0.7824135825037957,
+      "num_tokens": 19428449.0,
+      "step": 4840
+    },
+    {
+      "entropy": 1.0107353992760182,
+      "epoch": 1.856104094910065,
+      "grad_norm": 0.08587910234928131,
+      "learning_rate": 8.030079226534175e-05,
+      "loss": 1.0694228172302247,
+      "mean_token_accuracy": 0.7585012704133988,
+      "num_tokens": 19469625.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.9866157718002796,
+      "epoch": 1.8599311136624568,
+      "grad_norm": 0.11663772910833359,
+      "learning_rate": 8.00322277427152e-05,
+      "loss": 1.0394890785217286,
+      "mean_token_accuracy": 0.7632519364356994,
+      "num_tokens": 19510480.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.881837759912014,
+      "epoch": 1.8637581324148487,
+      "grad_norm": 0.13599033653736115,
+      "learning_rate": 7.976366322008862e-05,
+      "loss": 0.9441938400268555,
+      "mean_token_accuracy": 0.7848336577415467,
+      "num_tokens": 19550305.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.9193019077181817,
+      "epoch": 1.8675851511672406,
+      "grad_norm": 0.09272989630699158,
+      "learning_rate": 7.949509869746207e-05,
+      "loss": 0.9862917900085449,
+      "mean_token_accuracy": 0.7736792579293251,
+      "num_tokens": 19588278.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.9219990812242032,
+      "epoch": 1.8714121699196324,
+      "grad_norm": 0.10006739944219589,
+      "learning_rate": 7.922653417483551e-05,
+      "loss": 0.9700265884399414,
+      "mean_token_accuracy": 0.780723437666893,
+      "num_tokens": 19625749.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.8564371943473816,
+      "epoch": 1.8752391886720245,
+      "grad_norm": 0.08216696232557297,
+      "learning_rate": 7.895796965220895e-05,
+      "loss": 0.9064787864685059,
+      "mean_token_accuracy": 0.7962334454059601,
+      "num_tokens": 19663935.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.9482992745935916,
+      "epoch": 1.8790662074244164,
+      "grad_norm": 0.06782303750514984,
+      "learning_rate": 7.868940512958238e-05,
+      "loss": 0.9920551300048828,
+      "mean_token_accuracy": 0.7696940049529075,
+      "num_tokens": 19704663.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.884397204965353,
+      "epoch": 1.8828932261768083,
+      "grad_norm": 0.06414399296045303,
+      "learning_rate": 7.842084060695583e-05,
+      "loss": 0.9083518981933594,
+      "mean_token_accuracy": 0.7858098462224007,
+      "num_tokens": 19753438.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.8019696604460478,
+      "epoch": 1.8867202449292002,
+      "grad_norm": 0.08456243574619293,
+      "learning_rate": 7.815227608432927e-05,
+      "loss": 0.8896969795227051,
+      "mean_token_accuracy": 0.8053153440356254,
+      "num_tokens": 19790404.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.8008564852178097,
+      "epoch": 1.890547263681592,
+      "grad_norm": 0.10543688386678696,
+      "learning_rate": 7.78837115617027e-05,
+      "loss": 0.8535223007202148,
+      "mean_token_accuracy": 0.8059684678912162,
+      "num_tokens": 19825645.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.8714719720184803,
+      "epoch": 1.894374282433984,
+      "grad_norm": 0.09498755633831024,
+      "learning_rate": 7.761514703907614e-05,
+      "loss": 0.9063860893249511,
+      "mean_token_accuracy": 0.7914781123399734,
+      "num_tokens": 19866092.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.9202240366488695,
+      "epoch": 1.8982013011863759,
+      "grad_norm": 0.07342597842216492,
+      "learning_rate": 7.734658251644959e-05,
+      "loss": 0.9767581939697265,
+      "mean_token_accuracy": 0.7755973920226097,
+      "num_tokens": 19907356.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.9477262906730175,
+      "epoch": 1.9020283199387678,
+      "grad_norm": 0.08742880076169968,
+      "learning_rate": 7.707801799382302e-05,
+      "loss": 1.0063783645629882,
+      "mean_token_accuracy": 0.7687567621469498,
+      "num_tokens": 19952869.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.977492806315422,
+      "epoch": 1.9058553386911596,
+      "grad_norm": 0.10321515798568726,
+      "learning_rate": 7.680945347119645e-05,
+      "loss": 1.0323823928833007,
+      "mean_token_accuracy": 0.7646988987922668,
+      "num_tokens": 19991372.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7999268680810928,
+      "epoch": 1.9096823574435515,
+      "grad_norm": 0.08925452828407288,
+      "learning_rate": 7.65408889485699e-05,
+      "loss": 0.8391226768493653,
+      "mean_token_accuracy": 0.8017501994967461,
+      "num_tokens": 20029189.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.8757653787732125,
+      "epoch": 1.9135093761959434,
+      "grad_norm": 0.1915360540151596,
+      "learning_rate": 7.627232442594334e-05,
+      "loss": 0.9243562698364258,
+      "mean_token_accuracy": 0.7841411307454109,
+      "num_tokens": 20070611.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.9357082359492779,
+      "epoch": 1.9173363949483353,
+      "grad_norm": 0.08219558745622635,
+      "learning_rate": 7.600375990331677e-05,
+      "loss": 0.9772232055664063,
+      "mean_token_accuracy": 0.7725088000297546,
+      "num_tokens": 20110392.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.9191611532121897,
+      "epoch": 1.9211634137007272,
+      "grad_norm": 0.07629676163196564,
+      "learning_rate": 7.573519538069022e-05,
+      "loss": 0.9754646301269532,
+      "mean_token_accuracy": 0.7830281540751457,
+      "num_tokens": 20150683.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.9279548175632953,
+      "epoch": 1.924990432453119,
+      "grad_norm": 0.09845773130655289,
+      "learning_rate": 7.546663085806366e-05,
+      "loss": 0.9818471908569336,
+      "mean_token_accuracy": 0.7738550245761872,
+      "num_tokens": 20190521.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.9281142316758633,
+      "epoch": 1.928817451205511,
+      "grad_norm": 0.10571245104074478,
+      "learning_rate": 7.519806633543708e-05,
+      "loss": 0.999634075164795,
+      "mean_token_accuracy": 0.7708285465836525,
+      "num_tokens": 20230615.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.8793018095195293,
+      "epoch": 1.9326444699579028,
+      "grad_norm": 0.11255183815956116,
+      "learning_rate": 7.492950181281053e-05,
+      "loss": 0.9399495124816895,
+      "mean_token_accuracy": 0.7893765285611153,
+      "num_tokens": 20269332.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.8188632413744926,
+      "epoch": 1.9364714887102947,
+      "grad_norm": 0.08683498203754425,
+      "learning_rate": 7.466093729018398e-05,
+      "loss": 0.8760917663574219,
+      "mean_token_accuracy": 0.800470444560051,
+      "num_tokens": 20316849.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.9165158126503229,
+      "epoch": 1.9402985074626866,
+      "grad_norm": 0.12123431265354156,
+      "learning_rate": 7.439237276755741e-05,
+      "loss": 0.9515151023864746,
+      "mean_token_accuracy": 0.7772148326039314,
+      "num_tokens": 20354641.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.8890400048345327,
+      "epoch": 1.9441255262150785,
+      "grad_norm": 0.09551843255758286,
+      "learning_rate": 7.412380824493084e-05,
+      "loss": 0.9720385551452637,
+      "mean_token_accuracy": 0.7855533555150032,
+      "num_tokens": 20400703.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.9226945102214813,
+      "epoch": 1.9479525449674704,
+      "grad_norm": 0.11462504416704178,
+      "learning_rate": 7.385524372230429e-05,
+      "loss": 0.9757321357727051,
+      "mean_token_accuracy": 0.7739654749631881,
+      "num_tokens": 20442145.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.8108384694904089,
+      "epoch": 1.9517795637198623,
+      "grad_norm": 0.13017524778842926,
+      "learning_rate": 7.358667919967772e-05,
+      "loss": 0.8620017051696778,
+      "mean_token_accuracy": 0.8028488114476204,
+      "num_tokens": 20472714.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.9563053950667382,
+      "epoch": 1.9556065824722542,
+      "grad_norm": 0.10588496923446655,
+      "learning_rate": 7.331811467705116e-05,
+      "loss": 0.9805202484130859,
+      "mean_token_accuracy": 0.7729632049798966,
+      "num_tokens": 20518593.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.9307407476007938,
+      "epoch": 1.959433601224646,
+      "grad_norm": 0.09899015724658966,
+      "learning_rate": 7.30495501544246e-05,
+      "loss": 0.998748779296875,
+      "mean_token_accuracy": 0.7733172833919525,
+      "num_tokens": 20558008.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.9505821786820888,
+      "epoch": 1.963260619977038,
+      "grad_norm": 0.0943673700094223,
+      "learning_rate": 7.278098563179804e-05,
+      "loss": 1.0047925949096679,
+      "mean_token_accuracy": 0.7691358909010887,
+      "num_tokens": 20603741.0,
+      "step": 5130
+    },
+    {
+      "entropy": 1.04148171544075,
+      "epoch": 1.9670876387294298,
+      "grad_norm": 0.08869694918394089,
+      "learning_rate": 7.251242110917149e-05,
+      "loss": 1.0801177024841309,
+      "mean_token_accuracy": 0.7499634683132171,
+      "num_tokens": 20645827.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.7822969853878021,
+      "epoch": 1.9709146574818217,
+      "grad_norm": 0.0994991883635521,
+      "learning_rate": 7.224385658654492e-05,
+      "loss": 0.8042619705200196,
+      "mean_token_accuracy": 0.8097834318876267,
+      "num_tokens": 20684019.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.918664800748229,
+      "epoch": 1.9747416762342136,
+      "grad_norm": 0.11157739907503128,
+      "learning_rate": 7.197529206391837e-05,
+      "loss": 0.983153247833252,
+      "mean_token_accuracy": 0.7776870116591453,
+      "num_tokens": 20726278.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.911195681989193,
+      "epoch": 1.9785686949866055,
+      "grad_norm": 0.13472694158554077,
+      "learning_rate": 7.17067275412918e-05,
+      "loss": 0.9662351608276367,
+      "mean_token_accuracy": 0.7743990138173104,
+      "num_tokens": 20759927.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.8238823972642422,
+      "epoch": 1.9823957137389974,
+      "grad_norm": 0.08864834159612656,
+      "learning_rate": 7.143816301866523e-05,
+      "loss": 0.8870213508605957,
+      "mean_token_accuracy": 0.7989589869976044,
+      "num_tokens": 20798325.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.9405660286545754,
+      "epoch": 1.9862227324913893,
+      "grad_norm": 0.08372621983289719,
+      "learning_rate": 7.116959849603868e-05,
+      "loss": 0.9449873924255371,
+      "mean_token_accuracy": 0.7792889401316643,
+      "num_tokens": 20837136.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.8287422813475132,
+      "epoch": 1.9900497512437811,
+      "grad_norm": 0.0968240275979042,
+      "learning_rate": 7.090103397341211e-05,
+      "loss": 0.8873905181884766,
+      "mean_token_accuracy": 0.7976622357964516,
+      "num_tokens": 20877693.0,
+      "step": 5200
+    },
+    {
+      "entropy": 0.9188660819083452,
+      "epoch": 1.993876769996173,
+      "grad_norm": 0.09275626391172409,
+      "learning_rate": 7.063246945078555e-05,
+      "loss": 0.989016342163086,
+      "mean_token_accuracy": 0.7755422025918961,
+      "num_tokens": 20924885.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.9058490604162216,
+      "epoch": 1.997703788748565,
+      "grad_norm": 0.08644875138998032,
+      "learning_rate": 7.0363904928159e-05,
+      "loss": 0.9660470008850097,
+      "mean_token_accuracy": 0.7761533245444298,
+      "num_tokens": 20966342.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.7741431064903737,
+      "epoch": 2.0015308075009566,
+      "grad_norm": 0.07492107152938843,
+      "learning_rate": 7.009534040553243e-05,
+      "loss": 0.8241374015808105,
+      "mean_token_accuracy": 0.8149536207318306,
+      "num_tokens": 21004798.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.8813200116157531,
+      "epoch": 2.0053578262533485,
+      "grad_norm": 0.07805436849594116,
+      "learning_rate": 6.982677588290588e-05,
+      "loss": 0.921663761138916,
+      "mean_token_accuracy": 0.7912002876400948,
+      "num_tokens": 21049021.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.8896506872028113,
+      "epoch": 2.0091848450057403,
+      "grad_norm": 0.13928763568401337,
+      "learning_rate": 6.955821136027931e-05,
+      "loss": 0.9278170585632324,
+      "mean_token_accuracy": 0.7765205070376396,
+      "num_tokens": 21086531.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.9149777121841908,
+      "epoch": 2.0130118637581322,
+      "grad_norm": 0.06992843002080917,
+      "learning_rate": 6.928964683765274e-05,
+      "loss": 0.9667098045349121,
+      "mean_token_accuracy": 0.7750229969620704,
+      "num_tokens": 21127453.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.8076952576637269,
+      "epoch": 2.016838882510524,
+      "grad_norm": 0.12632791697978973,
+      "learning_rate": 6.902108231502619e-05,
+      "loss": 0.8237466812133789,
+      "mean_token_accuracy": 0.804887568950653,
+      "num_tokens": 21165297.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.8818444184958935,
+      "epoch": 2.020665901262916,
+      "grad_norm": 0.08924616128206253,
+      "learning_rate": 6.875251779239962e-05,
+      "loss": 0.9049506187438965,
+      "mean_token_accuracy": 0.7822276562452316,
+      "num_tokens": 21206219.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.7953705489635468,
+      "epoch": 2.024492920015308,
+      "grad_norm": 0.1111336424946785,
+      "learning_rate": 6.848395326977307e-05,
+      "loss": 0.8433744430541992,
+      "mean_token_accuracy": 0.8049945279955864,
+      "num_tokens": 21249239.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.904665675573051,
+      "epoch": 2.0283199387677,
+      "grad_norm": 0.09494993835687637,
+      "learning_rate": 6.82153887471465e-05,
+      "loss": 0.9693451881408691,
+      "mean_token_accuracy": 0.779350683093071,
+      "num_tokens": 21289639.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.7958274722099304,
+      "epoch": 2.0321469575200917,
+      "grad_norm": 0.10396509617567062,
+      "learning_rate": 6.794682422451995e-05,
+      "loss": 0.8559811592102051,
+      "mean_token_accuracy": 0.8057383120059967,
+      "num_tokens": 21329136.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.9416906848549843,
+      "epoch": 2.0359739762724836,
+      "grad_norm": 0.08166563510894775,
+      "learning_rate": 6.767825970189338e-05,
+      "loss": 0.9891387939453125,
+      "mean_token_accuracy": 0.7737650781869888,
+      "num_tokens": 21371300.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.9342201549559832,
+      "epoch": 2.0398009950248754,
+      "grad_norm": 0.09459090232849121,
+      "learning_rate": 6.740969517926682e-05,
+      "loss": 0.9509946823120117,
+      "mean_token_accuracy": 0.7751364663243294,
+      "num_tokens": 21412268.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.8397190041840077,
+      "epoch": 2.0436280137772673,
+      "grad_norm": 0.10005268454551697,
+      "learning_rate": 6.714113065664026e-05,
+      "loss": 0.9056560516357421,
+      "mean_token_accuracy": 0.79336898624897,
+      "num_tokens": 21451975.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.9148454248905182,
+      "epoch": 2.047455032529659,
+      "grad_norm": 0.10257065296173096,
+      "learning_rate": 6.68725661340137e-05,
+      "loss": 0.9611604690551758,
+      "mean_token_accuracy": 0.7737416908144951,
+      "num_tokens": 21491818.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.9010646104812622,
+      "epoch": 2.051282051282051,
+      "grad_norm": 0.11826229095458984,
+      "learning_rate": 6.660400161138713e-05,
+      "loss": 0.9446893692016601,
+      "mean_token_accuracy": 0.7851994633674622,
+      "num_tokens": 21528066.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.8987722039222718,
+      "epoch": 2.055109070034443,
+      "grad_norm": 0.10371451824903488,
+      "learning_rate": 6.633543708876058e-05,
+      "loss": 0.9595455169677735,
+      "mean_token_accuracy": 0.7833559066057205,
+      "num_tokens": 21562883.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.8856854721903801,
+      "epoch": 2.058936088786835,
+      "grad_norm": 0.1089499220252037,
+      "learning_rate": 6.606687256613403e-05,
+      "loss": 0.9219722747802734,
+      "mean_token_accuracy": 0.7822227373719215,
+      "num_tokens": 21600910.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.8720096081495285,
+      "epoch": 2.0627631075392268,
+      "grad_norm": 0.09962328523397446,
+      "learning_rate": 6.579830804350745e-05,
+      "loss": 0.9654089927673339,
+      "mean_token_accuracy": 0.7856920391321183,
+      "num_tokens": 21640445.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.9440382812172174,
+      "epoch": 2.0665901262916186,
+      "grad_norm": 0.08670477569103241,
+      "learning_rate": 6.552974352088089e-05,
+      "loss": 0.9934238433837891,
+      "mean_token_accuracy": 0.7687147289514542,
+      "num_tokens": 21682432.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.774172055721283,
+      "epoch": 2.0704171450440105,
+      "grad_norm": 0.11862040311098099,
+      "learning_rate": 6.526117899825434e-05,
+      "loss": 0.8106603622436523,
+      "mean_token_accuracy": 0.8135839134454728,
+      "num_tokens": 21721359.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.9194908868521452,
+      "epoch": 2.0742441637964024,
+      "grad_norm": 0.10227365791797638,
+      "learning_rate": 6.499261447562777e-05,
+      "loss": 0.9410523414611817,
+      "mean_token_accuracy": 0.7788734346628189,
+      "num_tokens": 21763700.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.7955736435949803,
+      "epoch": 2.0780711825487943,
+      "grad_norm": 0.09657785296440125,
+      "learning_rate": 6.472404995300121e-05,
+      "loss": 0.8665301322937011,
+      "mean_token_accuracy": 0.8067882195115089,
+      "num_tokens": 21804190.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.8065498791635036,
+      "epoch": 2.081898201301186,
+      "grad_norm": 0.11568085849285126,
+      "learning_rate": 6.445548543037465e-05,
+      "loss": 0.8515932083129882,
+      "mean_token_accuracy": 0.8035058185458184,
+      "num_tokens": 21839801.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.9087674509733915,
+      "epoch": 2.085725220053578,
+      "grad_norm": 0.09318574517965317,
+      "learning_rate": 6.418692090774809e-05,
+      "loss": 0.9387861251831054,
+      "mean_token_accuracy": 0.77939523011446,
+      "num_tokens": 21877125.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.86418566852808,
+      "epoch": 2.08955223880597,
+      "grad_norm": 0.08796729892492294,
+      "learning_rate": 6.391835638512152e-05,
+      "loss": 0.9152085304260253,
+      "mean_token_accuracy": 0.7899368211627007,
+      "num_tokens": 21921493.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.8593201294541359,
+      "epoch": 2.093379257558362,
+      "grad_norm": 0.14465564489364624,
+      "learning_rate": 6.364979186249497e-05,
+      "loss": 0.8955412864685058,
+      "mean_token_accuracy": 0.7898772984743119,
+      "num_tokens": 21961188.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.8998314358294011,
+      "epoch": 2.0972062763107537,
+      "grad_norm": 0.11634784191846848,
+      "learning_rate": 6.338122733986842e-05,
+      "loss": 0.9114861488342285,
+      "mean_token_accuracy": 0.7838647082448006,
+      "num_tokens": 22001738.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.8693659231066704,
+      "epoch": 2.1010332950631456,
+      "grad_norm": 0.11536803841590881,
+      "learning_rate": 6.311266281724184e-05,
+      "loss": 0.9232154846191406,
+      "mean_token_accuracy": 0.7881089702248574,
+      "num_tokens": 22039626.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.9556272588670254,
+      "epoch": 2.1048603138155375,
+      "grad_norm": 0.09614596515893936,
+      "learning_rate": 6.284409829461528e-05,
+      "loss": 1.0266177177429199,
+      "mean_token_accuracy": 0.7646962344646454,
+      "num_tokens": 22081971.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.7735307298600673,
+      "epoch": 2.1086873325679294,
+      "grad_norm": 0.10002073645591736,
+      "learning_rate": 6.257553377198873e-05,
+      "loss": 0.8011887550354004,
+      "mean_token_accuracy": 0.8088100135326386,
+      "num_tokens": 22117897.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.8981072999536991,
+      "epoch": 2.1125143513203213,
+      "grad_norm": 0.10524707287549973,
+      "learning_rate": 6.230696924936216e-05,
+      "loss": 0.9659936904907227,
+      "mean_token_accuracy": 0.7843907788395882,
+      "num_tokens": 22161049.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.8891891561448574,
+      "epoch": 2.116341370072713,
+      "grad_norm": 0.10095740854740143,
+      "learning_rate": 6.20384047267356e-05,
+      "loss": 0.9199987411499023,
+      "mean_token_accuracy": 0.7833669245243072,
+      "num_tokens": 22201183.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.9359986830502749,
+      "epoch": 2.120168388825105,
+      "grad_norm": 0.08723930269479752,
+      "learning_rate": 6.176984020410904e-05,
+      "loss": 0.9635790824890137,
+      "mean_token_accuracy": 0.7724878415465355,
+      "num_tokens": 22240779.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.8017430886626243,
+      "epoch": 2.123995407577497,
+      "grad_norm": 0.10579924285411835,
+      "learning_rate": 6.150127568148249e-05,
+      "loss": 0.842125129699707,
+      "mean_token_accuracy": 0.8020379558205605,
+      "num_tokens": 22279289.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.7666160762310028,
+      "epoch": 2.127822426329889,
+      "grad_norm": 0.09871628880500793,
+      "learning_rate": 6.123271115885591e-05,
+      "loss": 0.8378163337707519,
+      "mean_token_accuracy": 0.8119754999876022,
+      "num_tokens": 22316715.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.9505756117403508,
+      "epoch": 2.1316494450822807,
+      "grad_norm": 0.11093632131814957,
+      "learning_rate": 6.096414663622936e-05,
+      "loss": 0.9677371025085449,
+      "mean_token_accuracy": 0.7698320209980011,
+      "num_tokens": 22360112.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.7982158973813057,
+      "epoch": 2.1354764638346726,
+      "grad_norm": 0.11260368674993515,
+      "learning_rate": 6.06955821136028e-05,
+      "loss": 0.8571239471435547,
+      "mean_token_accuracy": 0.804571321606636,
+      "num_tokens": 22399114.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.8869463637471199,
+      "epoch": 2.1393034825870645,
+      "grad_norm": 0.08550643920898438,
+      "learning_rate": 6.042701759097623e-05,
+      "loss": 0.9476675033569336,
+      "mean_token_accuracy": 0.7807673364877701,
+      "num_tokens": 22440187.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.9491269618272782,
+      "epoch": 2.1431305013394564,
+      "grad_norm": 0.09019884467124939,
+      "learning_rate": 6.015845306834967e-05,
+      "loss": 1.0232599258422852,
+      "mean_token_accuracy": 0.7681664958596229,
+      "num_tokens": 22479682.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.8861779697239399,
+      "epoch": 2.1469575200918483,
+      "grad_norm": 0.11756031215190887,
+      "learning_rate": 5.988988854572312e-05,
+      "loss": 0.9251557350158691,
+      "mean_token_accuracy": 0.7849425792694091,
+      "num_tokens": 22520352.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.8735060147941113,
+      "epoch": 2.15078453884424,
+      "grad_norm": 0.0996679812669754,
+      "learning_rate": 5.9621324023096546e-05,
+      "loss": 0.9677264213562011,
+      "mean_token_accuracy": 0.7881714150309562,
+      "num_tokens": 22561677.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.991636025160551,
+      "epoch": 2.154611557596632,
+      "grad_norm": 0.10682649165391922,
+      "learning_rate": 5.935275950046999e-05,
+      "loss": 1.050811195373535,
+      "mean_token_accuracy": 0.7574850931763649,
+      "num_tokens": 22609671.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.9028345100581646,
+      "epoch": 2.158438576349024,
+      "grad_norm": 0.11249802261590958,
+      "learning_rate": 5.908419497784343e-05,
+      "loss": 0.9876343727111816,
+      "mean_token_accuracy": 0.783162035048008,
+      "num_tokens": 22650924.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.868353420495987,
+      "epoch": 2.162265595101416,
+      "grad_norm": 0.08846433460712433,
+      "learning_rate": 5.8815630455216867e-05,
+      "loss": 0.9271388053894043,
+      "mean_token_accuracy": 0.7898381799459457,
+      "num_tokens": 22691550.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.9247912406921387,
+      "epoch": 2.1660926138538077,
+      "grad_norm": 0.10013602674007416,
+      "learning_rate": 5.854706593259031e-05,
+      "loss": 1.0093653678894043,
+      "mean_token_accuracy": 0.7723490744829178,
+      "num_tokens": 22728956.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.82930968105793,
+      "epoch": 2.1699196326061996,
+      "grad_norm": 0.11004043370485306,
+      "learning_rate": 5.827850140996375e-05,
+      "loss": 0.8801467895507813,
+      "mean_token_accuracy": 0.798722094297409,
+      "num_tokens": 22765064.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.8950945638120175,
+      "epoch": 2.1737466513585915,
+      "grad_norm": 0.09994686394929886,
+      "learning_rate": 5.800993688733719e-05,
+      "loss": 0.9781051635742187,
+      "mean_token_accuracy": 0.7849533364176751,
+      "num_tokens": 22802213.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.8847132481634616,
+      "epoch": 2.1775736701109834,
+      "grad_norm": 0.09891512989997864,
+      "learning_rate": 5.774137236471062e-05,
+      "loss": 0.9338027954101562,
+      "mean_token_accuracy": 0.7867394030094147,
+      "num_tokens": 22839400.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.8212509788572788,
+      "epoch": 2.1814006888633752,
+      "grad_norm": 0.10451705008745193,
+      "learning_rate": 5.747280784208406e-05,
+      "loss": 0.8740688323974609,
+      "mean_token_accuracy": 0.7968196496367455,
+      "num_tokens": 22877771.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.7856742814183235,
+      "epoch": 2.185227707615767,
+      "grad_norm": 0.09351614862680435,
+      "learning_rate": 5.720424331945751e-05,
+      "loss": 0.8385543823242188,
+      "mean_token_accuracy": 0.8064358577132225,
+      "num_tokens": 22916159.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.9431014984846116,
+      "epoch": 2.189054726368159,
+      "grad_norm": 0.09432144463062286,
+      "learning_rate": 5.6935678796830935e-05,
+      "loss": 1.0021851539611817,
+      "mean_token_accuracy": 0.7693860113620759,
+      "num_tokens": 22958014.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.9080683786422015,
+      "epoch": 2.192881745120551,
+      "grad_norm": 0.08724278956651688,
+      "learning_rate": 5.666711427420438e-05,
+      "loss": 0.9878963470458985,
+      "mean_token_accuracy": 0.7802156403660774,
+      "num_tokens": 23003222.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.8772326201200485,
+      "epoch": 2.196708763872943,
+      "grad_norm": 0.1096489354968071,
+      "learning_rate": 5.639854975157782e-05,
+      "loss": 0.9326786041259766,
+      "mean_token_accuracy": 0.7881689593195915,
+      "num_tokens": 23039512.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.9084336057305336,
+      "epoch": 2.2005357826253347,
+      "grad_norm": 0.11137977987527847,
+      "learning_rate": 5.6129985228951256e-05,
+      "loss": 0.9574773788452149,
+      "mean_token_accuracy": 0.7860094889998436,
+      "num_tokens": 23078238.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.836103780195117,
+      "epoch": 2.2043628013777266,
+      "grad_norm": 0.11038387566804886,
+      "learning_rate": 5.5861420706324696e-05,
+      "loss": 0.88037109375,
+      "mean_token_accuracy": 0.7916925936937332,
+      "num_tokens": 23121089.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.9425606489181518,
+      "epoch": 2.2081898201301184,
+      "grad_norm": 0.10270453989505768,
+      "learning_rate": 5.5592856183698137e-05,
+      "loss": 0.983431339263916,
+      "mean_token_accuracy": 0.7715479463338852,
+      "num_tokens": 23158047.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.8212515480816365,
+      "epoch": 2.2120168388825103,
+      "grad_norm": 0.0880119651556015,
+      "learning_rate": 5.532429166107157e-05,
+      "loss": 0.887947940826416,
+      "mean_token_accuracy": 0.7997770145535469,
+      "num_tokens": 23204019.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.8668085850775242,
+      "epoch": 2.215843857634902,
+      "grad_norm": 0.11390146613121033,
+      "learning_rate": 5.505572713844501e-05,
+      "loss": 0.9010316848754882,
+      "mean_token_accuracy": 0.7880747586488723,
+      "num_tokens": 23241922.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.7907863073050976,
+      "epoch": 2.219670876387294,
+      "grad_norm": 0.11713080108165741,
+      "learning_rate": 5.478716261581846e-05,
+      "loss": 0.8595284461975098,
+      "mean_token_accuracy": 0.8068661123514176,
+      "num_tokens": 23280534.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.8358560226857662,
+      "epoch": 2.223497895139686,
+      "grad_norm": 0.11117064207792282,
+      "learning_rate": 5.45185980931919e-05,
+      "loss": 0.8745571136474609,
+      "mean_token_accuracy": 0.793362820148468,
+      "num_tokens": 23323119.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.8238232973963022,
+      "epoch": 2.227324913892078,
+      "grad_norm": 0.13185663521289825,
+      "learning_rate": 5.425003357056533e-05,
+      "loss": 0.8659845352172851,
+      "mean_token_accuracy": 0.8025152862071991,
+      "num_tokens": 23363749.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.8596846207976341,
+      "epoch": 2.2311519326444698,
+      "grad_norm": 0.09360291808843613,
+      "learning_rate": 5.398146904793877e-05,
+      "loss": 0.9118245124816895,
+      "mean_token_accuracy": 0.7882251426577568,
+      "num_tokens": 23402886.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.8035648860037327,
+      "epoch": 2.2349789513968616,
+      "grad_norm": 0.09347285330295563,
+      "learning_rate": 5.371290452531221e-05,
+      "loss": 0.8725827217102051,
+      "mean_token_accuracy": 0.8045972406864166,
+      "num_tokens": 23442339.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.9175308585166931,
+      "epoch": 2.2388059701492535,
+      "grad_norm": 0.12336985766887665,
+      "learning_rate": 5.3444340002685645e-05,
+      "loss": 0.9388077735900879,
+      "mean_token_accuracy": 0.7768721342086792,
+      "num_tokens": 23481344.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.868817687779665,
+      "epoch": 2.2426329889016454,
+      "grad_norm": 0.10311949998140335,
+      "learning_rate": 5.3175775480059086e-05,
+      "loss": 0.9337680816650391,
+      "mean_token_accuracy": 0.7877210825681686,
+      "num_tokens": 23520637.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.854228886961937,
+      "epoch": 2.2464600076540373,
+      "grad_norm": 0.10659918189048767,
+      "learning_rate": 5.2907210957432526e-05,
+      "loss": 0.9077530860900879,
+      "mean_token_accuracy": 0.7909654468297959,
+      "num_tokens": 23559877.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.8457217663526535,
+      "epoch": 2.250287026406429,
+      "grad_norm": 0.09633689373731613,
+      "learning_rate": 5.263864643480596e-05,
+      "loss": 0.8785475730895996,
+      "mean_token_accuracy": 0.7941769883036613,
+      "num_tokens": 23597033.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.8822055049240589,
+      "epoch": 2.254114045158821,
+      "grad_norm": 0.09562286734580994,
+      "learning_rate": 5.23700819121794e-05,
+      "loss": 0.8851138114929199,
+      "mean_token_accuracy": 0.7860250055789948,
+      "num_tokens": 23634788.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.8556318368762732,
+      "epoch": 2.257941063911213,
+      "grad_norm": 0.08814764767885208,
+      "learning_rate": 5.210151738955285e-05,
+      "loss": 0.8866415977478027,
+      "mean_token_accuracy": 0.7966004252433777,
+      "num_tokens": 23673283.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.7395530994981527,
+      "epoch": 2.261768082663605,
+      "grad_norm": 0.07671936601400375,
+      "learning_rate": 5.1832952866926274e-05,
+      "loss": 0.7680532455444335,
+      "mean_token_accuracy": 0.8190904691815376,
+      "num_tokens": 23711540.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.8898126773536206,
+      "epoch": 2.2655951014159967,
+      "grad_norm": 0.06960798799991608,
+      "learning_rate": 5.156438834429972e-05,
+      "loss": 1.026920700073242,
+      "mean_token_accuracy": 0.7816770374774933,
+      "num_tokens": 23756178.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.8902945756912232,
+      "epoch": 2.2694221201683886,
+      "grad_norm": 0.1114925891160965,
+      "learning_rate": 5.129582382167316e-05,
+      "loss": 0.9598423957824707,
+      "mean_token_accuracy": 0.784630736708641,
+      "num_tokens": 23792151.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.8439918398857117,
+      "epoch": 2.2732491389207805,
+      "grad_norm": 0.16730423271656036,
+      "learning_rate": 5.10272592990466e-05,
+      "loss": 0.851725959777832,
+      "mean_token_accuracy": 0.7940610617399215,
+      "num_tokens": 23830309.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.9178552135825158,
+      "epoch": 2.2770761576731724,
+      "grad_norm": 0.16359879076480865,
+      "learning_rate": 5.0758694776420035e-05,
+      "loss": 0.9417426109313964,
+      "mean_token_accuracy": 0.7781487166881561,
+      "num_tokens": 23874638.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.9053961969912052,
+      "epoch": 2.2809031764255643,
+      "grad_norm": 0.08877693116664886,
+      "learning_rate": 5.0490130253793475e-05,
+      "loss": 0.9975083351135254,
+      "mean_token_accuracy": 0.7837231978774071,
+      "num_tokens": 23918641.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.8590337552130223,
+      "epoch": 2.284730195177956,
+      "grad_norm": 0.1032002717256546,
+      "learning_rate": 5.022156573116692e-05,
+      "loss": 0.8895168304443359,
+      "mean_token_accuracy": 0.7937395930290222,
+      "num_tokens": 23964403.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.8678315542638302,
+      "epoch": 2.288557213930348,
+      "grad_norm": 0.12054577469825745,
+      "learning_rate": 4.9953001208540356e-05,
+      "loss": 0.9571179389953614,
+      "mean_token_accuracy": 0.7875312000513077,
+      "num_tokens": 24001736.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.8353918489068747,
+      "epoch": 2.29238423268274,
+      "grad_norm": 0.1126277968287468,
+      "learning_rate": 4.9684436685913796e-05,
+      "loss": 0.927174186706543,
+      "mean_token_accuracy": 0.7998543947935104,
+      "num_tokens": 24038494.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.7281714532524347,
+      "epoch": 2.296211251435132,
+      "grad_norm": 0.09404657036066055,
+      "learning_rate": 4.941587216328723e-05,
+      "loss": 0.7814407825469971,
+      "mean_token_accuracy": 0.8194777265191078,
+      "num_tokens": 24077404.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.8627386562526226,
+      "epoch": 2.3000382701875237,
+      "grad_norm": 0.07272294908761978,
+      "learning_rate": 4.914730764066067e-05,
+      "loss": 0.8920239448547364,
+      "mean_token_accuracy": 0.7905093863606453,
+      "num_tokens": 24123483.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.8679380901157856,
+      "epoch": 2.3038652889399156,
+      "grad_norm": 0.09443669021129608,
+      "learning_rate": 4.887874311803411e-05,
+      "loss": 0.874543571472168,
+      "mean_token_accuracy": 0.7891486629843711,
+      "num_tokens": 24165215.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.8942526787519455,
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.0953405573964119,
+      "learning_rate": 4.861017859540755e-05,
+      "loss": 0.9304584503173828,
+      "mean_token_accuracy": 0.7855148240923882,
+      "num_tokens": 24204454.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.7896301347762347,
+      "epoch": 2.3115193264446994,
+      "grad_norm": 0.11093971133232117,
+      "learning_rate": 4.834161407278099e-05,
+      "loss": 0.8957646369934082,
+      "mean_token_accuracy": 0.8066290900111198,
+      "num_tokens": 24245578.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.9012999664992094,
+      "epoch": 2.3153463451970913,
+      "grad_norm": 0.09953141212463379,
+      "learning_rate": 4.8073049550154424e-05,
+      "loss": 0.9699124336242676,
+      "mean_token_accuracy": 0.7792607560753823,
+      "num_tokens": 24286627.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.8553815156221389,
+      "epoch": 2.319173363949483,
+      "grad_norm": 0.09737669676542282,
+      "learning_rate": 4.7804485027527864e-05,
+      "loss": 0.9319831848144531,
+      "mean_token_accuracy": 0.7943563163280487,
+      "num_tokens": 24326050.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.8088245622813701,
+      "epoch": 2.323000382701875,
+      "grad_norm": 0.11754145473241806,
+      "learning_rate": 4.7535920504901305e-05,
+      "loss": 0.8612746238708496,
+      "mean_token_accuracy": 0.7998821645975113,
+      "num_tokens": 24365505.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.8720655493438244,
+      "epoch": 2.326827401454267,
+      "grad_norm": 0.10582665354013443,
+      "learning_rate": 4.726735598227474e-05,
+      "loss": 0.9663046836853028,
+      "mean_token_accuracy": 0.78773233294487,
+      "num_tokens": 24403619.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.814146314561367,
+      "epoch": 2.330654420206659,
+      "grad_norm": 0.10099766403436661,
+      "learning_rate": 4.6998791459648185e-05,
+      "loss": 0.8403602600097656,
+      "mean_token_accuracy": 0.8022790655493737,
+      "num_tokens": 24441133.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.8325122386217118,
+      "epoch": 2.3344814389590507,
+      "grad_norm": 0.0968555137515068,
+      "learning_rate": 4.673022693702162e-05,
+      "loss": 0.8952775955200195,
+      "mean_token_accuracy": 0.7972952157258988,
+      "num_tokens": 24487908.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.8313679326325655,
+      "epoch": 2.3383084577114426,
+      "grad_norm": 0.09856109321117401,
+      "learning_rate": 4.6461662414395066e-05,
+      "loss": 0.8740328788757324,
+      "mean_token_accuracy": 0.7973453208804131,
+      "num_tokens": 24528859.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.9734285809099674,
+      "epoch": 2.3421354764638345,
+      "grad_norm": 0.08564373850822449,
+      "learning_rate": 4.61930978917685e-05,
+      "loss": 1.0028407096862793,
+      "mean_token_accuracy": 0.761284664273262,
+      "num_tokens": 24574604.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.9015337243676186,
+      "epoch": 2.3459624952162264,
+      "grad_norm": 0.09626568853855133,
+      "learning_rate": 4.592453336914194e-05,
+      "loss": 0.9965445518493652,
+      "mean_token_accuracy": 0.7804829552769661,
+      "num_tokens": 24615926.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.8764280565083027,
+      "epoch": 2.3497895139686182,
+      "grad_norm": 0.09104456007480621,
+      "learning_rate": 4.565596884651538e-05,
+      "loss": 0.9158814430236817,
+      "mean_token_accuracy": 0.7859255224466324,
+      "num_tokens": 24656662.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.8626538865268231,
+      "epoch": 2.35361653272101,
+      "grad_norm": 0.10454346984624863,
+      "learning_rate": 4.5387404323888814e-05,
+      "loss": 0.9093445777893067,
+      "mean_token_accuracy": 0.7909897804260254,
+      "num_tokens": 24696048.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.9042750746011734,
+      "epoch": 2.357443551473402,
+      "grad_norm": 0.09976542741060257,
+      "learning_rate": 4.511883980126226e-05,
+      "loss": 0.9527711868286133,
+      "mean_token_accuracy": 0.7807446241378784,
+      "num_tokens": 24738856.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.892713101953268,
+      "epoch": 2.361270570225794,
+      "grad_norm": 0.09778838604688644,
+      "learning_rate": 4.4850275278635694e-05,
+      "loss": 0.9142132759094238,
+      "mean_token_accuracy": 0.7793798848986626,
+      "num_tokens": 24781940.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.8652282394468784,
+      "epoch": 2.365097588978186,
+      "grad_norm": 0.13737474381923676,
+      "learning_rate": 4.4581710756009134e-05,
+      "loss": 0.9030959129333496,
+      "mean_token_accuracy": 0.7882118329405785,
+      "num_tokens": 24818476.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.880942365527153,
+      "epoch": 2.3689246077305777,
+      "grad_norm": 0.09460416436195374,
+      "learning_rate": 4.4313146233382575e-05,
+      "loss": 0.9684123992919922,
+      "mean_token_accuracy": 0.7829654842615128,
+      "num_tokens": 24856950.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.9563789039850235,
+      "epoch": 2.3727516264829696,
+      "grad_norm": 0.10954713076353073,
+      "learning_rate": 4.404458171075601e-05,
+      "loss": 1.029030704498291,
+      "mean_token_accuracy": 0.7727080956101418,
+      "num_tokens": 24895606.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.827500730752945,
+      "epoch": 2.3765786452353614,
+      "grad_norm": 0.1212112084031105,
+      "learning_rate": 4.377601718812945e-05,
+      "loss": 0.8650990486145019,
+      "mean_token_accuracy": 0.7993797525763512,
+      "num_tokens": 24932482.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.8221234314143657,
+      "epoch": 2.3804056639877533,
+      "grad_norm": 0.10023710876703262,
+      "learning_rate": 4.350745266550289e-05,
+      "loss": 0.8777777671813964,
+      "mean_token_accuracy": 0.7987013593316078,
+      "num_tokens": 24975109.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.8734230428934098,
+      "epoch": 2.384232682740145,
+      "grad_norm": 0.09403553605079651,
+      "learning_rate": 4.323888814287633e-05,
+      "loss": 0.8978803634643555,
+      "mean_token_accuracy": 0.7872134670615196,
+      "num_tokens": 25020916.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.9003870271146297,
+      "epoch": 2.388059701492537,
+      "grad_norm": 0.09854581952095032,
+      "learning_rate": 4.297032362024977e-05,
+      "loss": 0.9225659370422363,
+      "mean_token_accuracy": 0.7807397484779358,
+      "num_tokens": 25061018.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.8118300527334213,
+      "epoch": 2.391886720244929,
+      "grad_norm": 0.11139514297246933,
+      "learning_rate": 4.27017590976232e-05,
+      "loss": 0.8876243591308594,
+      "mean_token_accuracy": 0.800039604306221,
+      "num_tokens": 25097954.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.8419897515326739,
+      "epoch": 2.395713738997321,
+      "grad_norm": 0.09123879671096802,
+      "learning_rate": 4.243319457499664e-05,
+      "loss": 0.86744384765625,
+      "mean_token_accuracy": 0.7919191718101501,
+      "num_tokens": 25134260.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.9123246632516384,
+      "epoch": 2.3995407577497128,
+      "grad_norm": 0.10300562530755997,
+      "learning_rate": 4.2164630052370084e-05,
+      "loss": 0.9368386268615723,
+      "mean_token_accuracy": 0.7797829449176789,
+      "num_tokens": 25176001.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.9066010326147079,
+      "epoch": 2.4033677765021046,
+      "grad_norm": 0.10231593996286392,
+      "learning_rate": 4.1896065529743524e-05,
+      "loss": 0.9637252807617187,
+      "mean_token_accuracy": 0.7807635113596916,
+      "num_tokens": 25214450.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.8680018067359925,
+      "epoch": 2.4071947952544965,
+      "grad_norm": 0.09813899546861649,
+      "learning_rate": 4.162750100711696e-05,
+      "loss": 0.9405930519104004,
+      "mean_token_accuracy": 0.7862071350216866,
+      "num_tokens": 25249019.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.8444254245609045,
+      "epoch": 2.4110218140068884,
+      "grad_norm": 0.09815159440040588,
+      "learning_rate": 4.1358936484490404e-05,
+      "loss": 0.9015726089477539,
+      "mean_token_accuracy": 0.7970604464411736,
+      "num_tokens": 25287466.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.9179269846528768,
+      "epoch": 2.4148488327592803,
+      "grad_norm": 0.1013285368680954,
+      "learning_rate": 4.109037196186384e-05,
+      "loss": 0.9629206657409668,
+      "mean_token_accuracy": 0.7756785362958908,
+      "num_tokens": 25325488.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.8627055402845144,
+      "epoch": 2.418675851511672,
+      "grad_norm": 0.09085863828659058,
+      "learning_rate": 4.082180743923728e-05,
+      "loss": 0.8825644493103028,
+      "mean_token_accuracy": 0.7927587017416954,
+      "num_tokens": 25362470.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.8909512132406234,
+      "epoch": 2.422502870264064,
+      "grad_norm": 0.12609654664993286,
+      "learning_rate": 4.055324291661072e-05,
+      "loss": 0.9005517959594727,
+      "mean_token_accuracy": 0.784729179739952,
+      "num_tokens": 25405399.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.8371693149209023,
+      "epoch": 2.426329889016456,
+      "grad_norm": 0.09511356055736542,
+      "learning_rate": 4.028467839398415e-05,
+      "loss": 0.8819235801696778,
+      "mean_token_accuracy": 0.7933985084295273,
+      "num_tokens": 25443537.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.8452706336975098,
+      "epoch": 2.430156907768848,
+      "grad_norm": 0.08440756797790527,
+      "learning_rate": 4.00161138713576e-05,
+      "loss": 0.9220956802368164,
+      "mean_token_accuracy": 0.791832709312439,
+      "num_tokens": 25482874.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.8533206440508365,
+      "epoch": 2.4339839265212397,
+      "grad_norm": 0.10529948770999908,
+      "learning_rate": 3.974754934873103e-05,
+      "loss": 0.8976041793823242,
+      "mean_token_accuracy": 0.7917203813791275,
+      "num_tokens": 25523091.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.8192368470132351,
+      "epoch": 2.4378109452736316,
+      "grad_norm": 0.08338342607021332,
+      "learning_rate": 3.947898482610447e-05,
+      "loss": 0.8657890319824219,
+      "mean_token_accuracy": 0.8002077579498291,
+      "num_tokens": 25566050.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.9303523369133473,
+      "epoch": 2.4416379640260235,
+      "grad_norm": 0.09010683745145798,
+      "learning_rate": 3.921042030347791e-05,
+      "loss": 0.9760264396667481,
+      "mean_token_accuracy": 0.7748634815216064,
+      "num_tokens": 25608936.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.7555282160639762,
+      "epoch": 2.4454649827784154,
+      "grad_norm": 0.11948851495981216,
+      "learning_rate": 3.894185578085135e-05,
+      "loss": 0.8005829811096191,
+      "mean_token_accuracy": 0.8136610746383667,
+      "num_tokens": 25647408.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.8959879912436008,
+      "epoch": 2.4492920015308073,
+      "grad_norm": 0.09189214557409286,
+      "learning_rate": 3.8673291258224794e-05,
+      "loss": 0.9070920944213867,
+      "mean_token_accuracy": 0.7838554188609124,
+      "num_tokens": 25690271.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.7601668298244476,
+      "epoch": 2.453119020283199,
+      "grad_norm": 0.11115460842847824,
+      "learning_rate": 3.840472673559823e-05,
+      "loss": 0.837701416015625,
+      "mean_token_accuracy": 0.8158529132604599,
+      "num_tokens": 25730098.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.9026189528405666,
+      "epoch": 2.456946039035591,
+      "grad_norm": 0.0951504036784172,
+      "learning_rate": 3.813616221297167e-05,
+      "loss": 0.9555998802185058,
+      "mean_token_accuracy": 0.7768774792551995,
+      "num_tokens": 25769649.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.8566267982125282,
+      "epoch": 2.460773057787983,
+      "grad_norm": 0.1477993279695511,
+      "learning_rate": 3.786759769034511e-05,
+      "loss": 0.901324462890625,
+      "mean_token_accuracy": 0.7918707326054573,
+      "num_tokens": 25805906.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.8576595298945904,
+      "epoch": 2.464600076540375,
+      "grad_norm": 0.08643563091754913,
+      "learning_rate": 3.759903316771854e-05,
+      "loss": 0.9027094841003418,
+      "mean_token_accuracy": 0.7925754263997078,
+      "num_tokens": 25847270.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.8848195761442185,
+      "epoch": 2.4684270952927667,
+      "grad_norm": 0.1148499846458435,
+      "learning_rate": 3.733046864509199e-05,
+      "loss": 0.9222222328186035,
+      "mean_token_accuracy": 0.7866752982139588,
+      "num_tokens": 25890454.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.8222585029900074,
+      "epoch": 2.4722541140451586,
+      "grad_norm": 0.1051439717411995,
+      "learning_rate": 3.706190412246542e-05,
+      "loss": 0.8674264907836914,
+      "mean_token_accuracy": 0.8014690011739731,
+      "num_tokens": 25927176.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.7895723138004541,
+      "epoch": 2.4760811327975505,
+      "grad_norm": 0.08904940634965897,
+      "learning_rate": 3.679333959983886e-05,
+      "loss": 0.8720718383789062,
+      "mean_token_accuracy": 0.8032544136047364,
+      "num_tokens": 25969008.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.8449521534144878,
+      "epoch": 2.4799081515499424,
+      "grad_norm": 0.09109736979007721,
+      "learning_rate": 3.65247750772123e-05,
+      "loss": 0.8994977951049805,
+      "mean_token_accuracy": 0.7939551532268524,
+      "num_tokens": 26008671.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.8769714809954167,
+      "epoch": 2.4837351703023343,
+      "grad_norm": 0.09221527725458145,
+      "learning_rate": 3.625621055458574e-05,
+      "loss": 0.9647493362426758,
+      "mean_token_accuracy": 0.7877351269125938,
+      "num_tokens": 26047583.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.840660959109664,
+      "epoch": 2.487562189054726,
+      "grad_norm": 0.0888860896229744,
+      "learning_rate": 3.598764603195918e-05,
+      "loss": 0.872824764251709,
+      "mean_token_accuracy": 0.7932088255882264,
+      "num_tokens": 26090690.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.9435165245085955,
+      "epoch": 2.491389207807118,
+      "grad_norm": 0.10055243968963623,
+      "learning_rate": 3.571908150933262e-05,
+      "loss": 1.008607769012451,
+      "mean_token_accuracy": 0.7684792190790176,
+      "num_tokens": 26134620.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.9596942149102687,
+      "epoch": 2.49521622655951,
+      "grad_norm": 0.11321604251861572,
+      "learning_rate": 3.545051698670606e-05,
+      "loss": 1.021597957611084,
+      "mean_token_accuracy": 0.7706323087215423,
+      "num_tokens": 26176850.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.9805667255073786,
+      "epoch": 2.499043245311902,
+      "grad_norm": 0.13084010779857635,
+      "learning_rate": 3.51819524640795e-05,
+      "loss": 1.0418537139892579,
+      "mean_token_accuracy": 0.763472905755043,
+      "num_tokens": 26220943.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.9104986634105444,
+      "epoch": 2.5028702640642937,
+      "grad_norm": 0.09176472574472427,
+      "learning_rate": 3.491338794145294e-05,
+      "loss": 0.972693920135498,
+      "mean_token_accuracy": 0.7809211134910583,
+      "num_tokens": 26262084.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.8316202580928802,
+      "epoch": 2.5066972828166856,
+      "grad_norm": 0.11009900271892548,
+      "learning_rate": 3.464482341882637e-05,
+      "loss": 0.8581557273864746,
+      "mean_token_accuracy": 0.7978575736284256,
+      "num_tokens": 26302790.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.9041007287800312,
+      "epoch": 2.5105243015690775,
+      "grad_norm": 0.12103740125894547,
+      "learning_rate": 3.437625889619981e-05,
+      "loss": 0.9546697616577149,
+      "mean_token_accuracy": 0.7800753250718117,
+      "num_tokens": 26347959.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.8139931574463845,
+      "epoch": 2.5143513203214694,
+      "grad_norm": 0.08679619431495667,
+      "learning_rate": 3.410769437357325e-05,
+      "loss": 0.8982272148132324,
+      "mean_token_accuracy": 0.8002956256270408,
+      "num_tokens": 26388946.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.838017127290368,
+      "epoch": 2.5181783390738612,
+      "grad_norm": 0.12066033482551575,
+      "learning_rate": 3.383912985094669e-05,
+      "loss": 0.8589006423950195,
+      "mean_token_accuracy": 0.7943052783608436,
+      "num_tokens": 26431191.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.8299121838063002,
+      "epoch": 2.522005357826253,
+      "grad_norm": 0.08988375216722488,
+      "learning_rate": 3.357056532832013e-05,
+      "loss": 0.9106943130493164,
+      "mean_token_accuracy": 0.7972570925951004,
+      "num_tokens": 26468346.0,
+      "step": 6590
+    },
+    {
+      "entropy": 1.0362544253468513,
+      "epoch": 2.525832376578645,
+      "grad_norm": 0.10034547746181488,
+      "learning_rate": 3.3302000805693566e-05,
+      "loss": 1.0991132736206055,
+      "mean_token_accuracy": 0.7502188056707382,
+      "num_tokens": 26508029.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.9098232574760914,
+      "epoch": 2.529659395331037,
+      "grad_norm": 0.12513861060142517,
+      "learning_rate": 3.303343628306701e-05,
+      "loss": 0.9807866096496582,
+      "mean_token_accuracy": 0.7815383434295654,
+      "num_tokens": 26549321.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.8234303712844848,
+      "epoch": 2.533486414083429,
+      "grad_norm": 0.08378947526216507,
+      "learning_rate": 3.2764871760440446e-05,
+      "loss": 0.8650754928588867,
+      "mean_token_accuracy": 0.7995569303631782,
+      "num_tokens": 26589472.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.769949347153306,
+      "epoch": 2.5373134328358207,
+      "grad_norm": 0.12056911736726761,
+      "learning_rate": 3.249630723781389e-05,
+      "loss": 0.8480927467346191,
+      "mean_token_accuracy": 0.818176555633545,
+      "num_tokens": 26627566.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.8099306054413319,
+      "epoch": 2.5411404515882126,
+      "grad_norm": 0.09869939833879471,
+      "learning_rate": 3.222774271518733e-05,
+      "loss": 0.8649662017822266,
+      "mean_token_accuracy": 0.7981634557247161,
+      "num_tokens": 26662566.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.8528701025992632,
+      "epoch": 2.5449674703406044,
+      "grad_norm": 0.10336704552173615,
+      "learning_rate": 3.195917819256076e-05,
+      "loss": 0.9127251625061035,
+      "mean_token_accuracy": 0.7928516089916229,
+      "num_tokens": 26705768.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.8498493686318398,
+      "epoch": 2.5487944890929963,
+      "grad_norm": 0.10704471170902252,
+      "learning_rate": 3.169061366993421e-05,
+      "loss": 0.863565731048584,
+      "mean_token_accuracy": 0.7932710304856301,
+      "num_tokens": 26743574.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.8566017836332321,
+      "epoch": 2.552621507845388,
+      "grad_norm": 0.12135261297225952,
+      "learning_rate": 3.142204914730764e-05,
+      "loss": 0.9187004089355468,
+      "mean_token_accuracy": 0.7913481816649437,
+      "num_tokens": 26784127.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.8302055161446333,
+      "epoch": 2.55644852659778,
+      "grad_norm": 0.1430647373199463,
+      "learning_rate": 3.115348462468108e-05,
+      "loss": 0.8857596397399903,
+      "mean_token_accuracy": 0.7965412393212319,
+      "num_tokens": 26823189.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.8327139757573605,
+      "epoch": 2.560275545350172,
+      "grad_norm": 0.09538804739713669,
+      "learning_rate": 3.088492010205452e-05,
+      "loss": 0.9255412101745606,
+      "mean_token_accuracy": 0.7939359977841377,
+      "num_tokens": 26861599.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.8530606523156166,
+      "epoch": 2.564102564102564,
+      "grad_norm": 0.09193538129329681,
+      "learning_rate": 3.0616355579427955e-05,
+      "loss": 0.9151040077209472,
+      "mean_token_accuracy": 0.7901859179139137,
+      "num_tokens": 26901064.0,
+      "step": 6700
+    },
+    {
+      "entropy": 0.794033832848072,
+      "epoch": 2.5679295828549558,
+      "grad_norm": 0.1283407062292099,
+      "learning_rate": 3.03477910568014e-05,
+      "loss": 0.8441056251525879,
+      "mean_token_accuracy": 0.8033816903829575,
+      "num_tokens": 26942161.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.9340717010200024,
+      "epoch": 2.5717566016073476,
+      "grad_norm": 0.09237734973430634,
+      "learning_rate": 3.0079226534174836e-05,
+      "loss": 0.9747485160827637,
+      "mean_token_accuracy": 0.7732965379953385,
+      "num_tokens": 26982759.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.8746799558401108,
+      "epoch": 2.5755836203597395,
+      "grad_norm": 0.1391710638999939,
+      "learning_rate": 2.9810662011548273e-05,
+      "loss": 0.9311764717102051,
+      "mean_token_accuracy": 0.7883311554789543,
+      "num_tokens": 27022926.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.8290158938616514,
+      "epoch": 2.5794106391121314,
+      "grad_norm": 0.10442391782999039,
+      "learning_rate": 2.9542097488921716e-05,
+      "loss": 0.8346040725708008,
+      "mean_token_accuracy": 0.7985544398427009,
+      "num_tokens": 27065028.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.8574424415826798,
+      "epoch": 2.5832376578645233,
+      "grad_norm": 0.13001689314842224,
+      "learning_rate": 2.9273532966295153e-05,
+      "loss": 0.906099510192871,
+      "mean_token_accuracy": 0.790866918861866,
+      "num_tokens": 27100867.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.840974472463131,
+      "epoch": 2.587064676616915,
+      "grad_norm": 0.1224556565284729,
+      "learning_rate": 2.9004968443668594e-05,
+      "loss": 0.8969048500061035,
+      "mean_token_accuracy": 0.7953185483813285,
+      "num_tokens": 27137338.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.8477607406675816,
+      "epoch": 2.590891695369307,
+      "grad_norm": 0.09641005098819733,
+      "learning_rate": 2.873640392104203e-05,
+      "loss": 0.9569526672363281,
+      "mean_token_accuracy": 0.7941199511289596,
+      "num_tokens": 27178308.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.8317056275904179,
+      "epoch": 2.594718714121699,
+      "grad_norm": 0.11853990703821182,
+      "learning_rate": 2.8467839398415468e-05,
+      "loss": 0.9125295639038086,
+      "mean_token_accuracy": 0.7960822626948356,
+      "num_tokens": 27216898.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.8558823302388191,
+      "epoch": 2.598545732874091,
+      "grad_norm": 0.10477570444345474,
+      "learning_rate": 2.819927487578891e-05,
+      "loss": 0.8844131469726563,
+      "mean_token_accuracy": 0.7940610870718956,
+      "num_tokens": 27254443.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.8210954669862985,
+      "epoch": 2.6023727516264827,
+      "grad_norm": 0.14100609719753265,
+      "learning_rate": 2.7930710353162348e-05,
+      "loss": 0.8684535980224609,
+      "mean_token_accuracy": 0.7988820597529411,
+      "num_tokens": 27290079.0,
+      "step": 6800
+    },
+    {
+      "entropy": 0.8657392464578152,
+      "epoch": 2.6061997703788746,
+      "grad_norm": 0.09813658148050308,
+      "learning_rate": 2.7662145830535785e-05,
+      "loss": 0.9158803939819335,
+      "mean_token_accuracy": 0.7908033922314643,
+      "num_tokens": 27328190.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.8866597019135952,
+      "epoch": 2.6100267891312665,
+      "grad_norm": 0.11115613579750061,
+      "learning_rate": 2.739358130790923e-05,
+      "loss": 0.9120420455932617,
+      "mean_token_accuracy": 0.7854148596525192,
+      "num_tokens": 27369945.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.7982962183654309,
+      "epoch": 2.6138538078836584,
+      "grad_norm": 0.1377696692943573,
+      "learning_rate": 2.7125016785282666e-05,
+      "loss": 0.8332090377807617,
+      "mean_token_accuracy": 0.8022376418113708,
+      "num_tokens": 27406302.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.8424798093736172,
+      "epoch": 2.6176808266360503,
+      "grad_norm": 0.11442425101995468,
+      "learning_rate": 2.6856452262656106e-05,
+      "loss": 0.8876424789428711,
+      "mean_token_accuracy": 0.7893706291913987,
+      "num_tokens": 27449733.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.9239407800137996,
+      "epoch": 2.621507845388442,
+      "grad_norm": 0.0799759030342102,
+      "learning_rate": 2.6587887740029543e-05,
+      "loss": 0.9658034324645997,
+      "mean_token_accuracy": 0.7757296651601792,
+      "num_tokens": 27492884.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.8720928959548473,
+      "epoch": 2.625334864140834,
+      "grad_norm": 0.11632338911294937,
+      "learning_rate": 2.631932321740298e-05,
+      "loss": 0.9089359283447266,
+      "mean_token_accuracy": 0.7913818553090095,
+      "num_tokens": 27531878.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.9302754916250706,
+      "epoch": 2.629161882893226,
+      "grad_norm": 0.11215951293706894,
+      "learning_rate": 2.6050758694776423e-05,
+      "loss": 1.0027677536010742,
+      "mean_token_accuracy": 0.7739164605736732,
+      "num_tokens": 27567970.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.9016003269702196,
+      "epoch": 2.632988901645618,
+      "grad_norm": 0.11951353400945663,
+      "learning_rate": 2.578219417214986e-05,
+      "loss": 0.9493217468261719,
+      "mean_token_accuracy": 0.7779877439141274,
+      "num_tokens": 27609840.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.8870487026870251,
+      "epoch": 2.6368159203980097,
+      "grad_norm": 0.1124744564294815,
+      "learning_rate": 2.55136296495233e-05,
+      "loss": 1.0031387329101562,
+      "mean_token_accuracy": 0.7866110280156136,
+      "num_tokens": 27649655.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.9296976864337921,
+      "epoch": 2.6406429391504016,
+      "grad_norm": 0.1161704882979393,
+      "learning_rate": 2.5245065126896738e-05,
+      "loss": 1.012251853942871,
+      "mean_token_accuracy": 0.7726465791463852,
+      "num_tokens": 27694105.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.8415393102914095,
+      "epoch": 2.6444699579027935,
+      "grad_norm": 0.0987096056342125,
+      "learning_rate": 2.4976500604270178e-05,
+      "loss": 0.9147520065307617,
+      "mean_token_accuracy": 0.7973951831459999,
+      "num_tokens": 27730663.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.8274203538894653,
+      "epoch": 2.6482969766551854,
+      "grad_norm": 0.1101188212633133,
+      "learning_rate": 2.4707936081643615e-05,
+      "loss": 0.8873770713806153,
+      "mean_token_accuracy": 0.7974746853113175,
+      "num_tokens": 27772881.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.7984559834003448,
+      "epoch": 2.6521239954075773,
+      "grad_norm": 0.10185439884662628,
+      "learning_rate": 2.4439371559017055e-05,
+      "loss": 0.8775921821594238,
+      "mean_token_accuracy": 0.807880648970604,
+      "num_tokens": 27809534.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.887981615960598,
+      "epoch": 2.655951014159969,
+      "grad_norm": 0.08309295773506165,
+      "learning_rate": 2.4170807036390495e-05,
+      "loss": 0.9466443061828613,
+      "mean_token_accuracy": 0.7859978228807449,
+      "num_tokens": 27852591.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.9378888584673405,
+      "epoch": 2.659778032912361,
+      "grad_norm": 0.136076882481575,
+      "learning_rate": 2.3902242513763932e-05,
+      "loss": 1.0269956588745117,
+      "mean_token_accuracy": 0.7709244459867477,
+      "num_tokens": 27892120.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.9220107842236758,
+      "epoch": 2.663605051664753,
+      "grad_norm": 0.08248933404684067,
+      "learning_rate": 2.363367799113737e-05,
+      "loss": 0.9726594924926758,
+      "mean_token_accuracy": 0.7753236919641495,
+      "num_tokens": 27935380.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.7793348811566829,
+      "epoch": 2.667432070417145,
+      "grad_norm": 0.08308061957359314,
+      "learning_rate": 2.336511346851081e-05,
+      "loss": 0.7947993278503418,
+      "mean_token_accuracy": 0.8088447406888009,
+      "num_tokens": 27973020.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.9587450519204139,
+      "epoch": 2.6712590891695367,
+      "grad_norm": 0.10263237357139587,
+      "learning_rate": 2.309654894588425e-05,
+      "loss": 0.9791707038879395,
+      "mean_token_accuracy": 0.7663016110658646,
+      "num_tokens": 28016389.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.8766636185348033,
+      "epoch": 2.6750861079219286,
+      "grad_norm": 0.09917714446783066,
+      "learning_rate": 2.282798442325769e-05,
+      "loss": 0.9187355041503906,
+      "mean_token_accuracy": 0.7864622801542283,
+      "num_tokens": 28058100.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.8623256701976061,
+      "epoch": 2.6789131266743205,
+      "grad_norm": 0.08802894502878189,
+      "learning_rate": 2.255941990063113e-05,
+      "loss": 0.9108509063720703,
+      "mean_token_accuracy": 0.7891170993447304,
+      "num_tokens": 28095166.0,
+      "step": 7000
+    },
+    {
+      "entropy": 0.919238954409957,
+      "epoch": 2.6827401454267124,
+      "grad_norm": 0.11916540563106537,
+      "learning_rate": 2.2290855378004567e-05,
+      "loss": 0.9972674369812011,
+      "mean_token_accuracy": 0.7765705808997154,
+      "num_tokens": 28137533.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.918128065392375,
+      "epoch": 2.6865671641791042,
+      "grad_norm": 0.09536208212375641,
+      "learning_rate": 2.2022290855378004e-05,
+      "loss": 0.9865476608276367,
+      "mean_token_accuracy": 0.7736267536878586,
+      "num_tokens": 28179301.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.8265572734177112,
+      "epoch": 2.690394182931496,
+      "grad_norm": 0.09432680904865265,
+      "learning_rate": 2.1753726332751444e-05,
+      "loss": 0.8995939254760742,
+      "mean_token_accuracy": 0.7947996065020562,
+      "num_tokens": 28223849.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.8321899034082889,
+      "epoch": 2.694221201683888,
+      "grad_norm": 0.1223755031824112,
+      "learning_rate": 2.1485161810124885e-05,
+      "loss": 0.9003139495849609,
+      "mean_token_accuracy": 0.7975824415683747,
+      "num_tokens": 28268485.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.9064472205936909,
+      "epoch": 2.69804822043628,
+      "grad_norm": 0.13409113883972168,
+      "learning_rate": 2.121659728749832e-05,
+      "loss": 0.9323970794677734,
+      "mean_token_accuracy": 0.7808707699179649,
+      "num_tokens": 28307792.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.9527742668986321,
+      "epoch": 2.701875239188672,
+      "grad_norm": 0.09863030910491943,
+      "learning_rate": 2.0948032764871762e-05,
+      "loss": 1.0056820869445802,
+      "mean_token_accuracy": 0.7673134744167328,
+      "num_tokens": 28355447.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.8202732041478157,
+      "epoch": 2.7057022579410637,
+      "grad_norm": 0.10251973569393158,
+      "learning_rate": 2.0679468242245202e-05,
+      "loss": 0.8743599891662598,
+      "mean_token_accuracy": 0.7957186102867126,
+      "num_tokens": 28397195.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.9328485410660505,
+      "epoch": 2.7095292766934556,
+      "grad_norm": 0.09044504910707474,
+      "learning_rate": 2.041090371961864e-05,
+      "loss": 0.9707870483398438,
+      "mean_token_accuracy": 0.7739486545324326,
+      "num_tokens": 28440070.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.9110265091061592,
+      "epoch": 2.7133562954458474,
+      "grad_norm": 0.10417858511209488,
+      "learning_rate": 2.0142339196992076e-05,
+      "loss": 0.9495024681091309,
+      "mean_token_accuracy": 0.7784481555223465,
+      "num_tokens": 28483039.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.907703897356987,
+      "epoch": 2.7171833141982393,
+      "grad_norm": 0.10365665704011917,
+      "learning_rate": 1.9873774674365516e-05,
+      "loss": 0.9539920806884765,
+      "mean_token_accuracy": 0.7803053423762322,
+      "num_tokens": 28524922.0,
+      "step": 7100
+    },
+    {
+      "entropy": 0.8090648584067821,
+      "epoch": 2.721010332950631,
+      "grad_norm": 0.13015250861644745,
+      "learning_rate": 1.9605210151738957e-05,
+      "loss": 0.8559967994689941,
+      "mean_token_accuracy": 0.7999090999364853,
+      "num_tokens": 28565638.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.832624789327383,
+      "epoch": 2.724837351703023,
+      "grad_norm": 0.12992241978645325,
+      "learning_rate": 1.9336645629112397e-05,
+      "loss": 0.886108112335205,
+      "mean_token_accuracy": 0.7986625626683235,
+      "num_tokens": 28603666.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.8167526118457318,
+      "epoch": 2.728664370455415,
+      "grad_norm": 0.0879233330488205,
+      "learning_rate": 1.9068081106485834e-05,
+      "loss": 0.8744274139404297,
+      "mean_token_accuracy": 0.8013173520565033,
+      "num_tokens": 28647331.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.8693740144371986,
+      "epoch": 2.732491389207807,
+      "grad_norm": 0.11505398899316788,
+      "learning_rate": 1.879951658385927e-05,
+      "loss": 0.9142866134643555,
+      "mean_token_accuracy": 0.7936322972178459,
+      "num_tokens": 28683073.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.7896613411605358,
+      "epoch": 2.7363184079601988,
+      "grad_norm": 0.10490158945322037,
+      "learning_rate": 1.853095206123271e-05,
+      "loss": 0.8762624740600586,
+      "mean_token_accuracy": 0.8044975116848946,
+      "num_tokens": 28722340.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.8261051677167416,
+      "epoch": 2.7401454267125906,
+      "grad_norm": 0.10280875116586685,
+      "learning_rate": 1.826238753860615e-05,
+      "loss": 0.888590145111084,
+      "mean_token_accuracy": 0.7989666223526001,
+      "num_tokens": 28757940.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.8630577899515629,
+      "epoch": 2.7439724454649825,
+      "grad_norm": 0.12757791578769684,
+      "learning_rate": 1.799382301597959e-05,
+      "loss": 0.9082697868347168,
+      "mean_token_accuracy": 0.7890564352273941,
+      "num_tokens": 28796985.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.8979216992855072,
+      "epoch": 2.7477994642173744,
+      "grad_norm": 0.13048897683620453,
+      "learning_rate": 1.772525849335303e-05,
+      "loss": 0.9468406677246094,
+      "mean_token_accuracy": 0.7829687342047691,
+      "num_tokens": 28838091.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.9002114910632372,
+      "epoch": 2.7516264829697663,
+      "grad_norm": 0.130500927567482,
+      "learning_rate": 1.745669397072647e-05,
+      "loss": 0.9897032737731933,
+      "mean_token_accuracy": 0.7817048847675323,
+      "num_tokens": 28879084.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.861878028512001,
+      "epoch": 2.755453501722158,
+      "grad_norm": 0.10523588210344315,
+      "learning_rate": 1.7188129448099906e-05,
+      "loss": 0.9628341674804688,
+      "mean_token_accuracy": 0.7882343173027039,
+      "num_tokens": 28918018.0,
+      "step": 7200
+    },
+    {
+      "entropy": 0.7814029835164547,
+      "epoch": 2.75928052047455,
+      "grad_norm": 0.14345957338809967,
+      "learning_rate": 1.6919564925473346e-05,
+      "loss": 0.8377615928649902,
+      "mean_token_accuracy": 0.8100636526942253,
+      "num_tokens": 28953674.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.8798072785139084,
+      "epoch": 2.763107539226942,
+      "grad_norm": 0.10911094397306442,
+      "learning_rate": 1.6651000402846783e-05,
+      "loss": 0.9405971527099609,
+      "mean_token_accuracy": 0.7843327835202217,
+      "num_tokens": 28995212.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.7432700909674168,
+      "epoch": 2.766934557979334,
+      "grad_norm": 0.09271088242530823,
+      "learning_rate": 1.6382435880220223e-05,
+      "loss": 0.7987990856170655,
+      "mean_token_accuracy": 0.8185402989387512,
+      "num_tokens": 29034878.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.7937459200620651,
+      "epoch": 2.7707615767317257,
+      "grad_norm": 0.11122163385152817,
+      "learning_rate": 1.6113871357593664e-05,
+      "loss": 0.8469036102294922,
+      "mean_token_accuracy": 0.8031805381178856,
+      "num_tokens": 29074372.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.8456454008817673,
+      "epoch": 2.7745885954841176,
+      "grad_norm": 0.11189702153205872,
+      "learning_rate": 1.5845306834967104e-05,
+      "loss": 0.8942484855651855,
+      "mean_token_accuracy": 0.7923400938510895,
+      "num_tokens": 29117619.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.885396859049797,
+      "epoch": 2.7784156142365095,
+      "grad_norm": 0.10170719027519226,
+      "learning_rate": 1.557674231234054e-05,
+      "loss": 0.9175837516784668,
+      "mean_token_accuracy": 0.7860854491591454,
+      "num_tokens": 29156601.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.8742636401206255,
+      "epoch": 2.7822426329889014,
+      "grad_norm": 0.11130956560373306,
+      "learning_rate": 1.5308177789713978e-05,
+      "loss": 0.9322646141052247,
+      "mean_token_accuracy": 0.7902692511677742,
+      "num_tokens": 29200295.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.8523757141083479,
+      "epoch": 2.7860696517412933,
+      "grad_norm": 0.08611233532428741,
+      "learning_rate": 1.5039613267087418e-05,
+      "loss": 0.9210372924804687,
+      "mean_token_accuracy": 0.7912763133645058,
+      "num_tokens": 29235323.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.7804547689855099,
+      "epoch": 2.789896670493685,
+      "grad_norm": 0.08091949671506882,
+      "learning_rate": 1.4771048744460858e-05,
+      "loss": 0.8202395439147949,
+      "mean_token_accuracy": 0.8117679923772811,
+      "num_tokens": 29270182.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.8199648998677731,
+      "epoch": 2.793723689246077,
+      "grad_norm": 0.07486634701490402,
+      "learning_rate": 1.4502484221834297e-05,
+      "loss": 0.8396285057067872,
+      "mean_token_accuracy": 0.8032143607735633,
+      "num_tokens": 29311588.0,
+      "step": 7300
+    },
+    {
+      "entropy": 0.9650515951216221,
+      "epoch": 2.797550707998469,
+      "grad_norm": 0.10391585528850555,
+      "learning_rate": 1.4233919699207734e-05,
+      "loss": 1.047046184539795,
+      "mean_token_accuracy": 0.7648886650800705,
+      "num_tokens": 29353979.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.7674700990319252,
+      "epoch": 2.801377726750861,
+      "grad_norm": 0.09043332189321518,
+      "learning_rate": 1.3965355176581174e-05,
+      "loss": 0.8154891014099122,
+      "mean_token_accuracy": 0.8105725541710853,
+      "num_tokens": 29393298.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.7795201197266579,
+      "epoch": 2.8052047455032527,
+      "grad_norm": 0.14624197781085968,
+      "learning_rate": 1.3696790653954614e-05,
+      "loss": 0.7968831062316895,
+      "mean_token_accuracy": 0.808569261431694,
+      "num_tokens": 29423547.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.9187626458704472,
+      "epoch": 2.8090317642556446,
+      "grad_norm": 0.1368781179189682,
+      "learning_rate": 1.3428226131328053e-05,
+      "loss": 0.9583258628845215,
+      "mean_token_accuracy": 0.7731027945876121,
+      "num_tokens": 29465593.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.9403511643409729,
+      "epoch": 2.8128587830080365,
+      "grad_norm": 0.10892713069915771,
+      "learning_rate": 1.315966160870149e-05,
+      "loss": 0.9621626853942871,
+      "mean_token_accuracy": 0.767315211892128,
+      "num_tokens": 29506888.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.842640140466392,
+      "epoch": 2.8166858017604284,
+      "grad_norm": 0.08862321823835373,
+      "learning_rate": 1.289109708607493e-05,
+      "loss": 0.9031145095825195,
+      "mean_token_accuracy": 0.7967306047677993,
+      "num_tokens": 29550811.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.8931968793272972,
+      "epoch": 2.8205128205128203,
+      "grad_norm": 0.0979296937584877,
+      "learning_rate": 1.2622532563448369e-05,
+      "loss": 0.9369117736816406,
+      "mean_token_accuracy": 0.785995215177536,
+      "num_tokens": 29587036.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.8621913805603981,
+      "epoch": 2.824339839265212,
+      "grad_norm": 0.08778136223554611,
+      "learning_rate": 1.2353968040821807e-05,
+      "loss": 0.884724235534668,
+      "mean_token_accuracy": 0.790358729660511,
+      "num_tokens": 29627992.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.8695362661033869,
+      "epoch": 2.828166858017604,
+      "grad_norm": 0.09141552448272705,
+      "learning_rate": 1.2085403518195248e-05,
+      "loss": 0.9539263725280762,
+      "mean_token_accuracy": 0.78631162494421,
+      "num_tokens": 29668509.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.8454725466668606,
+      "epoch": 2.831993876769996,
+      "grad_norm": 0.10090988874435425,
+      "learning_rate": 1.1816838995568685e-05,
+      "loss": 0.9256816864013672,
+      "mean_token_accuracy": 0.7941092774271965,
+      "num_tokens": 29706794.0,
+      "step": 7400
+    },
+    {
+      "entropy": 0.8406473740935325,
+      "epoch": 2.835820895522388,
+      "grad_norm": 0.12991519272327423,
+      "learning_rate": 1.1548274472942125e-05,
+      "loss": 0.8969921112060547,
+      "mean_token_accuracy": 0.7950825378298759,
+      "num_tokens": 29745883.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.8951507560908795,
+      "epoch": 2.8396479142747797,
+      "grad_norm": 0.14208164811134338,
+      "learning_rate": 1.1279709950315565e-05,
+      "loss": 0.9443653106689454,
+      "mean_token_accuracy": 0.7820898026227951,
+      "num_tokens": 29788428.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.859702505543828,
+      "epoch": 2.8434749330271716,
+      "grad_norm": 0.10485101491212845,
+      "learning_rate": 1.1011145427689002e-05,
+      "loss": 0.9106481552124024,
+      "mean_token_accuracy": 0.7910059571266175,
+      "num_tokens": 29829552.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.838575328886509,
+      "epoch": 2.8473019517795635,
+      "grad_norm": 0.09105801582336426,
+      "learning_rate": 1.0742580905062442e-05,
+      "loss": 0.9367799758911133,
+      "mean_token_accuracy": 0.7953649654984474,
+      "num_tokens": 29869380.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.9112126015126705,
+      "epoch": 2.8511289705319554,
+      "grad_norm": 0.09724974632263184,
+      "learning_rate": 1.0474016382435881e-05,
+      "loss": 0.9621581077575684,
+      "mean_token_accuracy": 0.7795066565275193,
+      "num_tokens": 29913977.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.7964273016899824,
+      "epoch": 2.8549559892843472,
+      "grad_norm": 0.09481512755155563,
+      "learning_rate": 1.020545185980932e-05,
+      "loss": 0.8208577156066894,
+      "mean_token_accuracy": 0.8045729547739029,
+      "num_tokens": 29949229.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.9103045649826527,
+      "epoch": 2.858783008036739,
+      "grad_norm": 0.08678591996431351,
+      "learning_rate": 9.936887337182758e-06,
+      "loss": 0.9599167823791503,
+      "mean_token_accuracy": 0.7792657531797886,
+      "num_tokens": 29999070.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.8333844318985939,
+      "epoch": 2.862610026789131,
+      "grad_norm": 0.07823742181062698,
+      "learning_rate": 9.668322814556198e-06,
+      "loss": 0.8832645416259766,
+      "mean_token_accuracy": 0.7986885383725166,
+      "num_tokens": 30041797.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.8970901295542717,
+      "epoch": 2.866437045541523,
+      "grad_norm": 0.11852974444627762,
+      "learning_rate": 9.399758291929635e-06,
+      "loss": 0.9755334854125977,
+      "mean_token_accuracy": 0.7814395651221275,
+      "num_tokens": 30080534.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.8733609687536955,
+      "epoch": 2.870264064293915,
+      "grad_norm": 0.08307944238185883,
+      "learning_rate": 9.131193769303076e-06,
+      "loss": 0.9116435050964355,
+      "mean_token_accuracy": 0.786429825425148,
+      "num_tokens": 30123488.0,
+      "step": 7500
+    },
+    {
+      "entropy": 0.7967244807630778,
+      "epoch": 2.8740910830463067,
+      "grad_norm": 0.121941938996315,
+      "learning_rate": 8.862629246676514e-06,
+      "loss": 0.8209601402282715,
+      "mean_token_accuracy": 0.8040247783064842,
+      "num_tokens": 30158076.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.8655086796730757,
+      "epoch": 2.8779181017986986,
+      "grad_norm": 0.10017320513725281,
+      "learning_rate": 8.594064724049953e-06,
+      "loss": 0.9246477127075196,
+      "mean_token_accuracy": 0.7905631095170975,
+      "num_tokens": 30198329.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.7916971929371357,
+      "epoch": 2.8817451205510904,
+      "grad_norm": 0.08822990953922272,
+      "learning_rate": 8.325500201423391e-06,
+      "loss": 0.8695680618286132,
+      "mean_token_accuracy": 0.8063599601387977,
+      "num_tokens": 30239990.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.7693583916872739,
+      "epoch": 2.8855721393034823,
+      "grad_norm": 0.1178632378578186,
+      "learning_rate": 8.056935678796832e-06,
+      "loss": 0.8029808044433594,
+      "mean_token_accuracy": 0.808713173866272,
+      "num_tokens": 30272583.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.9072235215455293,
+      "epoch": 2.889399158055874,
+      "grad_norm": 0.11368006467819214,
+      "learning_rate": 7.78837115617027e-06,
+      "loss": 0.9859001159667968,
+      "mean_token_accuracy": 0.7825366839766502,
+      "num_tokens": 30314370.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.909162075817585,
+      "epoch": 2.893226176808266,
+      "grad_norm": 0.10643935948610306,
+      "learning_rate": 7.519806633543709e-06,
+      "loss": 0.9263824462890625,
+      "mean_token_accuracy": 0.7813168540596962,
+      "num_tokens": 30362103.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.8779693342745304,
+      "epoch": 2.897053195560658,
+      "grad_norm": 0.12511365115642548,
+      "learning_rate": 7.2512421109171484e-06,
+      "loss": 0.9283166885375976,
+      "mean_token_accuracy": 0.7876154363155365,
+      "num_tokens": 30400468.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.9308112382888794,
+      "epoch": 2.90088021431305,
+      "grad_norm": 0.08942066878080368,
+      "learning_rate": 6.982677588290587e-06,
+      "loss": 0.9894198417663574,
+      "mean_token_accuracy": 0.7739586725831031,
+      "num_tokens": 30444628.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.8830183774232865,
+      "epoch": 2.9047072330654418,
+      "grad_norm": 0.08949998021125793,
+      "learning_rate": 6.7141130656640265e-06,
+      "loss": 0.9515928268432617,
+      "mean_token_accuracy": 0.7846902176737786,
+      "num_tokens": 30485845.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.8058773010969162,
+      "epoch": 2.9085342518178336,
+      "grad_norm": 0.1035229042172432,
+      "learning_rate": 6.445548543037465e-06,
+      "loss": 0.846186637878418,
+      "mean_token_accuracy": 0.8066700398921967,
+      "num_tokens": 30523979.0,
+      "step": 7600
+    },
+    {
+      "entropy": 0.9146121144294739,
+      "epoch": 2.9123612705702255,
+      "grad_norm": 0.09379884600639343,
+      "learning_rate": 6.176984020410904e-06,
+      "loss": 0.9735233306884765,
+      "mean_token_accuracy": 0.7774886921048164,
+      "num_tokens": 30564775.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.8396586284041405,
+      "epoch": 2.9161882893226174,
+      "grad_norm": 0.11920839548110962,
+      "learning_rate": 5.908419497784342e-06,
+      "loss": 0.9061779022216797,
+      "mean_token_accuracy": 0.7974281132221221,
+      "num_tokens": 30609113.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.8665836162865161,
+      "epoch": 2.9200153080750093,
+      "grad_norm": 0.10214731842279434,
+      "learning_rate": 5.639854975157783e-06,
+      "loss": 0.9333956718444825,
+      "mean_token_accuracy": 0.7912585958838463,
+      "num_tokens": 30652409.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.8082432024180889,
+      "epoch": 2.923842326827401,
+      "grad_norm": 0.09191566705703735,
+      "learning_rate": 5.371290452531221e-06,
+      "loss": 0.8443769454956055,
+      "mean_token_accuracy": 0.797667445242405,
+      "num_tokens": 30689299.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.8395522754639387,
+      "epoch": 2.927669345579793,
+      "grad_norm": 0.08281564712524414,
+      "learning_rate": 5.10272592990466e-06,
+      "loss": 0.8710539817810059,
+      "mean_token_accuracy": 0.7973509266972542,
+      "num_tokens": 30724619.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.8130493897944688,
+      "epoch": 2.931496364332185,
+      "grad_norm": 0.0996284931898117,
+      "learning_rate": 4.834161407278099e-06,
+      "loss": 0.8514342308044434,
+      "mean_token_accuracy": 0.800888329744339,
+      "num_tokens": 30764224.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.7793916609138251,
+      "epoch": 2.935323383084577,
+      "grad_norm": 0.09503267705440521,
+      "learning_rate": 4.565596884651538e-06,
+      "loss": 0.8305204391479493,
+      "mean_token_accuracy": 0.8106261268258095,
+      "num_tokens": 30800800.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.817446855083108,
+      "epoch": 2.9391504018369687,
+      "grad_norm": 0.13637053966522217,
+      "learning_rate": 4.2970323620249764e-06,
+      "loss": 0.839473819732666,
+      "mean_token_accuracy": 0.8018909886479377,
+      "num_tokens": 30841481.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.8140060313045978,
+      "epoch": 2.9429774205893606,
+      "grad_norm": 0.13390128314495087,
+      "learning_rate": 4.028467839398416e-06,
+      "loss": 0.8653444290161133,
+      "mean_token_accuracy": 0.8000675857067108,
+      "num_tokens": 30880001.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.7898532018065453,
+      "epoch": 2.9468044393417525,
+      "grad_norm": 0.11585478484630585,
+      "learning_rate": 3.7599033167718545e-06,
+      "loss": 0.8074365615844726,
+      "mean_token_accuracy": 0.8053972944617271,
+      "num_tokens": 30915563.0,
+      "step": 7700
+    },
+    {
+      "entropy": 0.8091453645378351,
+      "epoch": 2.9506314580941444,
+      "grad_norm": 0.09755035489797592,
+      "learning_rate": 3.4913387941452935e-06,
+      "loss": 0.8457134246826172,
+      "mean_token_accuracy": 0.8031114682555198,
+      "num_tokens": 30955410.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.8444364190101623,
+      "epoch": 2.9544584768465363,
+      "grad_norm": 0.1297679990530014,
+      "learning_rate": 3.2227742715187325e-06,
+      "loss": 0.910922622680664,
+      "mean_token_accuracy": 0.7976488128304482,
+      "num_tokens": 30997246.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.8454434804618358,
+      "epoch": 2.958285495598928,
+      "grad_norm": 0.15091662108898163,
+      "learning_rate": 2.954209748892171e-06,
+      "loss": 0.8977128982543945,
+      "mean_token_accuracy": 0.7951600447297096,
+      "num_tokens": 31042192.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.838621474429965,
+      "epoch": 2.96211251435132,
+      "grad_norm": 0.10101021081209183,
+      "learning_rate": 2.6856452262656106e-06,
+      "loss": 0.9142851829528809,
+      "mean_token_accuracy": 0.7966463148593903,
+      "num_tokens": 31082777.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.8021124713122845,
+      "epoch": 2.965939533103712,
+      "grad_norm": 0.11373798549175262,
+      "learning_rate": 2.4170807036390496e-06,
+      "loss": 0.845030403137207,
+      "mean_token_accuracy": 0.8039181783795357,
+      "num_tokens": 31122973.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.8570070005953312,
+      "epoch": 2.969766551856104,
+      "grad_norm": 0.0995812863111496,
+      "learning_rate": 2.1485161810124882e-06,
+      "loss": 0.8876262664794922,
+      "mean_token_accuracy": 0.7908932328224182,
+      "num_tokens": 31166313.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.9019658699631691,
+      "epoch": 2.9735935706084957,
+      "grad_norm": 0.10546575486660004,
+      "learning_rate": 1.8799516583859272e-06,
+      "loss": 0.9777070999145507,
+      "mean_token_accuracy": 0.7821963891386986,
+      "num_tokens": 31202060.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.9346055820584297,
+      "epoch": 2.9774205893608876,
+      "grad_norm": 0.11632298678159714,
+      "learning_rate": 1.6113871357593663e-06,
+      "loss": 1.017040729522705,
+      "mean_token_accuracy": 0.7751505836844444,
+      "num_tokens": 31241536.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.8882534563541412,
+      "epoch": 2.9812476081132795,
+      "grad_norm": 0.13064302504062653,
+      "learning_rate": 1.3428226131328053e-06,
+      "loss": 0.9505605697631836,
+      "mean_token_accuracy": 0.7848831593990326,
+      "num_tokens": 31278060.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.8854026839137077,
+      "epoch": 2.9850746268656714,
+      "grad_norm": 0.0977831557393074,
+      "learning_rate": 1.0742580905062441e-06,
+      "loss": 0.9311306953430176,
+      "mean_token_accuracy": 0.7847100362181664,
+      "num_tokens": 31325802.0,
+      "step": 7800
+    },
+    {
+      "entropy": 0.9448695838451385,
+      "epoch": 2.9889016456180633,
+      "grad_norm": 0.11724492162466049,
+      "learning_rate": 8.056935678796831e-07,
+      "loss": 0.983949089050293,
+      "mean_token_accuracy": 0.7636413291096688,
+      "num_tokens": 31367954.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.8787743166089058,
+      "epoch": 2.992728664370455,
+      "grad_norm": 0.09530383348464966,
+      "learning_rate": 5.371290452531221e-07,
+      "loss": 0.9605165481567383,
+      "mean_token_accuracy": 0.7847816556692123,
+      "num_tokens": 31410151.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.810061177611351,
+      "epoch": 2.996555683122847,
+      "grad_norm": 0.09042539447546005,
+      "learning_rate": 2.6856452262656103e-07,
+      "loss": 0.8766719818115234,
+      "mean_token_accuracy": 0.8047587737441063,
+      "num_tokens": 31451314.0,
+      "step": 7830
     }
   ],
   "logging_steps": 10,
-  "max_steps": 3114,
+  "max_steps": 7839,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
@@ -1063,12 +7852,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 2.1498282716531098e+18,
+  "total_flos": 1.17346463002948e+19,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null