diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11815 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1308,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002296211251435132,
+      "grad_norm": 229.51694349053318,
+      "learning_rate": 0.0,
+      "loss": 6.4293,
+      "mean_token_accuracy": 0.24527385085821152,
+      "num_tokens": 8048.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.004592422502870264,
+      "grad_norm": 260.93733200634813,
+      "learning_rate": 7.633587786259542e-08,
+      "loss": 6.449,
+      "mean_token_accuracy": 0.23870357125997543,
+      "num_tokens": 16395.0,
+      "step": 2
+    },
+    {
+      "epoch": 0.006888633754305396,
+      "grad_norm": 212.89332558202028,
+      "learning_rate": 1.5267175572519085e-07,
+      "loss": 6.1925,
+      "mean_token_accuracy": 0.24638409912586212,
+      "num_tokens": 25367.0,
+      "step": 3
+    },
+    {
+      "epoch": 0.009184845005740528,
+      "grad_norm": 240.39254954257933,
+      "learning_rate": 2.2900763358778629e-07,
+      "loss": 6.2861,
+      "mean_token_accuracy": 0.2444230243563652,
+      "num_tokens": 34249.0,
+      "step": 4
+    },
+    {
+      "epoch": 0.011481056257175661,
+      "grad_norm": 238.37462984425684,
+      "learning_rate": 3.053435114503817e-07,
+      "loss": 6.1932,
+      "mean_token_accuracy": 0.25068463385105133,
+      "num_tokens": 43270.0,
+      "step": 5
+    },
+    {
+      "epoch": 0.013777267508610792,
+      "grad_norm": 199.92904972200893,
+      "learning_rate": 3.8167938931297716e-07,
+      "loss": 6.512,
+      "mean_token_accuracy": 0.22473817318677902,
+      "num_tokens": 50857.0,
+      "step": 6
+    },
+    {
+      "epoch": 0.016073478760045924,
+      "grad_norm": 160.89466541377257,
+      "learning_rate": 4.5801526717557257e-07,
+      "loss": 6.1286,
+      "mean_token_accuracy": 0.25135470926761627,
+      "num_tokens": 59529.0,
+      "step": 7
+    },
+    {
+      "epoch": 0.018369690011481057,
+      "grad_norm": 147.51987207820136,
+      "learning_rate": 5.34351145038168e-07,
+      "loss": 6.0412,
+      "mean_token_accuracy": 0.24442250281572342,
+      "num_tokens": 68052.0,
+      "step": 8
+    },
+    {
+      "epoch": 0.02066590126291619,
+      "grad_norm": 163.0592351843276,
+      "learning_rate": 6.106870229007634e-07,
+      "loss": 5.7259,
+      "mean_token_accuracy": 0.2685967832803726,
+      "num_tokens": 76641.0,
+      "step": 9
+    },
+    {
+      "epoch": 0.022962112514351322,
+      "grad_norm": 158.48342792022507,
+      "learning_rate": 6.870229007633589e-07,
+      "loss": 5.5303,
+      "mean_token_accuracy": 0.2884530574083328,
+      "num_tokens": 85359.0,
+      "step": 10
+    },
+    {
+      "epoch": 0.02525832376578645,
+      "grad_norm": 130.86793197807168,
+      "learning_rate": 7.633587786259543e-07,
+      "loss": 5.6226,
+      "mean_token_accuracy": 0.2677329182624817,
+      "num_tokens": 93891.0,
+      "step": 11
+    },
+    {
+      "epoch": 0.027554535017221583,
+      "grad_norm": 104.00956751526628,
+      "learning_rate": 8.396946564885497e-07,
+      "loss": 5.3732,
+      "mean_token_accuracy": 0.27050259709358215,
+      "num_tokens": 102370.0,
+      "step": 12
+    },
+    {
+      "epoch": 0.029850746268656716,
+      "grad_norm": 99.40028334950634,
+      "learning_rate": 9.160305343511451e-07,
+      "loss": 5.0133,
+      "mean_token_accuracy": 0.3034631311893463,
+      "num_tokens": 111166.0,
+      "step": 13
+    },
+    {
+      "epoch": 0.03214695752009185,
+      "grad_norm": 95.69115127634679,
+      "learning_rate": 9.923664122137404e-07,
+      "loss": 5.1532,
+      "mean_token_accuracy": 0.27586428821086884,
+      "num_tokens": 119984.0,
+      "step": 14
+    },
+    {
+      "epoch": 0.03444316877152698,
+      "grad_norm": 86.02965301796536,
+      "learning_rate": 1.068702290076336e-06,
+      "loss": 5.0176,
+      "mean_token_accuracy": 0.2830243855714798,
+      "num_tokens": 128118.0,
+      "step": 15
+    },
+    {
+      "epoch": 0.03673938002296211,
+      "grad_norm": 78.00146452059,
+      "learning_rate": 1.1450381679389313e-06,
+      "loss": 4.6278,
+      "mean_token_accuracy": 0.2986981123685837,
+      "num_tokens": 136224.0,
+      "step": 16
+    },
+    {
+      "epoch": 0.03903559127439724,
+      "grad_norm": 66.87154312014434,
+      "learning_rate": 1.2213740458015268e-06,
+      "loss": 4.3043,
+      "mean_token_accuracy": 0.31380923092365265,
+      "num_tokens": 145273.0,
+      "step": 17
+    },
+    {
+      "epoch": 0.04133180252583238,
+      "grad_norm": 59.42125943990684,
+      "learning_rate": 1.297709923664122e-06,
+      "loss": 4.1424,
+      "mean_token_accuracy": 0.3375231921672821,
+      "num_tokens": 154056.0,
+      "step": 18
+    },
+    {
+      "epoch": 0.04362801377726751,
+      "grad_norm": 66.81336544104899,
+      "learning_rate": 1.3740458015267178e-06,
+      "loss": 4.0962,
+      "mean_token_accuracy": 0.3393639922142029,
+      "num_tokens": 162085.0,
+      "step": 19
+    },
+    {
+      "epoch": 0.045924225028702644,
+      "grad_norm": 58.47016780554335,
+      "learning_rate": 1.450381679389313e-06,
+      "loss": 3.8631,
+      "mean_token_accuracy": 0.3567530959844589,
+      "num_tokens": 170777.0,
+      "step": 20
+    },
+    {
+      "epoch": 0.04822043628013777,
+      "grad_norm": 54.44999611454747,
+      "learning_rate": 1.5267175572519086e-06,
+      "loss": 3.6993,
+      "mean_token_accuracy": 0.370703861117363,
+      "num_tokens": 179346.0,
+      "step": 21
+    },
+    {
+      "epoch": 0.0505166475315729,
+      "grad_norm": 46.91496950434756,
+      "learning_rate": 1.603053435114504e-06,
+      "loss": 3.434,
+      "mean_token_accuracy": 0.38842637836933136,
+      "num_tokens": 187542.0,
+      "step": 22
+    },
+    {
+      "epoch": 0.05281285878300804,
+      "grad_norm": 37.64655098347395,
+      "learning_rate": 1.6793893129770995e-06,
+      "loss": 2.9801,
+      "mean_token_accuracy": 0.45116737484931946,
+      "num_tokens": 196292.0,
+      "step": 23
+    },
+    {
+      "epoch": 0.05510907003444317,
+      "grad_norm": 32.79577991421603,
+      "learning_rate": 1.7557251908396948e-06,
+      "loss": 2.8091,
+      "mean_token_accuracy": 0.4807087928056717,
+      "num_tokens": 205039.0,
+      "step": 24
+    },
+    {
+      "epoch": 0.0574052812858783,
+      "grad_norm": 31.280575984297307,
+      "learning_rate": 1.8320610687022903e-06,
+      "loss": 2.4441,
+      "mean_token_accuracy": 0.5409079194068909,
+      "num_tokens": 213476.0,
+      "step": 25
+    },
+    {
+      "epoch": 0.05970149253731343,
+      "grad_norm": 27.855232328041787,
+      "learning_rate": 1.908396946564886e-06,
+      "loss": 2.2419,
+      "mean_token_accuracy": 0.5574310421943665,
+      "num_tokens": 221924.0,
+      "step": 26
+    },
+    {
+      "epoch": 0.06199770378874857,
+      "grad_norm": 23.81164340578062,
+      "learning_rate": 1.984732824427481e-06,
+      "loss": 2.0357,
+      "mean_token_accuracy": 0.5961459577083588,
+      "num_tokens": 231062.0,
+      "step": 27
+    },
+    {
+      "epoch": 0.0642939150401837,
+      "grad_norm": 21.529452576593055,
+      "learning_rate": 2.0610687022900764e-06,
+      "loss": 1.9456,
+      "mean_token_accuracy": 0.5997802019119263,
+      "num_tokens": 240299.0,
+      "step": 28
+    },
+    {
+      "epoch": 0.06659012629161883,
+      "grad_norm": 18.903251841184556,
+      "learning_rate": 2.137404580152672e-06,
+      "loss": 1.6164,
+      "mean_token_accuracy": 0.6741090416908264,
+      "num_tokens": 249278.0,
+      "step": 29
+    },
+    {
+      "epoch": 0.06888633754305395,
+      "grad_norm": 17.193933621647524,
+      "learning_rate": 2.2137404580152674e-06,
+      "loss": 1.5233,
+      "mean_token_accuracy": 0.6750809252262115,
+      "num_tokens": 258115.0,
+      "step": 30
+    },
+    {
+      "epoch": 0.0711825487944891,
+      "grad_norm": 15.200943454586819,
+      "learning_rate": 2.2900763358778625e-06,
+      "loss": 1.3553,
+      "mean_token_accuracy": 0.7057067453861237,
+      "num_tokens": 267010.0,
+      "step": 31
+    },
+    {
+      "epoch": 0.07347876004592423,
+      "grad_norm": 14.722824921042143,
+      "learning_rate": 2.3664122137404585e-06,
+      "loss": 1.2616,
+      "mean_token_accuracy": 0.7307117283344269,
+      "num_tokens": 275768.0,
+      "step": 32
+    },
+    {
+      "epoch": 0.07577497129735936,
+      "grad_norm": 14.426329263536992,
+      "learning_rate": 2.4427480916030536e-06,
+      "loss": 1.0971,
+      "mean_token_accuracy": 0.7600332498550415,
+      "num_tokens": 284193.0,
+      "step": 33
+    },
+    {
+      "epoch": 0.07807118254879448,
+      "grad_norm": 12.451029205540483,
+      "learning_rate": 2.5190839694656487e-06,
+      "loss": 0.9751,
+      "mean_token_accuracy": 0.777172327041626,
+      "num_tokens": 292820.0,
+      "step": 34
+    },
+    {
+      "epoch": 0.08036739380022963,
+      "grad_norm": 9.396648181734896,
+      "learning_rate": 2.595419847328244e-06,
+      "loss": 0.872,
+      "mean_token_accuracy": 0.7961028814315796,
+      "num_tokens": 301428.0,
+      "step": 35
+    },
+    {
+      "epoch": 0.08266360505166476,
+      "grad_norm": 6.451306386893086,
+      "learning_rate": 2.67175572519084e-06,
+      "loss": 0.8211,
+      "mean_token_accuracy": 0.8031978607177734,
+      "num_tokens": 309761.0,
+      "step": 36
+    },
+    {
+      "epoch": 0.08495981630309989,
+      "grad_norm": 5.704733805994221,
+      "learning_rate": 2.7480916030534356e-06,
+      "loss": 0.7564,
+      "mean_token_accuracy": 0.8187287151813507,
+      "num_tokens": 317919.0,
+      "step": 37
+    },
+    {
+      "epoch": 0.08725602755453502,
+      "grad_norm": 6.426510690052956,
+      "learning_rate": 2.824427480916031e-06,
+      "loss": 0.8252,
+      "mean_token_accuracy": 0.8015343248844147,
+      "num_tokens": 327619.0,
+      "step": 38
+    },
+    {
+      "epoch": 0.08955223880597014,
+      "grad_norm": 5.972192595954285,
+      "learning_rate": 2.900763358778626e-06,
+      "loss": 0.6913,
+      "mean_token_accuracy": 0.8248153030872345,
+      "num_tokens": 335797.0,
+      "step": 39
+    },
+    {
+      "epoch": 0.09184845005740529,
+      "grad_norm": 4.797916468470806,
+      "learning_rate": 2.9770992366412218e-06,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.8103065192699432,
+      "num_tokens": 344344.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.09414466130884042,
+      "grad_norm": 4.373172001375282,
+      "learning_rate": 3.0534351145038173e-06,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.8204226493835449,
+      "num_tokens": 352720.0,
+      "step": 41
+    },
+    {
+      "epoch": 0.09644087256027555,
+      "grad_norm": 3.941830036098511,
+      "learning_rate": 3.129770992366413e-06,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.8231934309005737,
+      "num_tokens": 360689.0,
+      "step": 42
+    },
+    {
+      "epoch": 0.09873708381171067,
+      "grad_norm": 3.5966636042454296,
+      "learning_rate": 3.206106870229008e-06,
+      "loss": 0.7029,
+      "mean_token_accuracy": 0.8217072784900665,
+      "num_tokens": 369347.0,
+      "step": 43
+    },
+    {
+      "epoch": 0.1010332950631458,
+      "grad_norm": 3.435063805509848,
+      "learning_rate": 3.2824427480916034e-06,
+      "loss": 0.6775,
+      "mean_token_accuracy": 0.8245145678520203,
+      "num_tokens": 378057.0,
+      "step": 44
+    },
+    {
+      "epoch": 0.10332950631458095,
+      "grad_norm": 3.3391260623760695,
+      "learning_rate": 3.358778625954199e-06,
+      "loss": 0.6116,
+      "mean_token_accuracy": 0.8413296043872833,
+      "num_tokens": 386572.0,
+      "step": 45
+    },
+    {
+      "epoch": 0.10562571756601608,
+      "grad_norm": 3.2986324392797006,
+      "learning_rate": 3.4351145038167944e-06,
+      "loss": 0.6261,
+      "mean_token_accuracy": 0.8364266455173492,
+      "num_tokens": 395195.0,
+      "step": 46
+    },
+    {
+      "epoch": 0.1079219288174512,
+      "grad_norm": 3.121798831226759,
+      "learning_rate": 3.5114503816793895e-06,
+      "loss": 0.6178,
+      "mean_token_accuracy": 0.8419860899448395,
+      "num_tokens": 403885.0,
+      "step": 47
+    },
+    {
+      "epoch": 0.11021814006888633,
+      "grad_norm": 3.281277600444172,
+      "learning_rate": 3.587786259541985e-06,
+      "loss": 0.6022,
+      "mean_token_accuracy": 0.843781441450119,
+      "num_tokens": 412803.0,
+      "step": 48
+    },
+    {
+      "epoch": 0.11251435132032148,
+      "grad_norm": 3.4349397722984256,
+      "learning_rate": 3.6641221374045806e-06,
+      "loss": 0.6998,
+      "mean_token_accuracy": 0.8232371807098389,
+      "num_tokens": 421642.0,
+      "step": 49
+    },
+    {
+      "epoch": 0.1148105625717566,
+      "grad_norm": 3.744155192241371,
+      "learning_rate": 3.740458015267176e-06,
+      "loss": 0.6042,
+      "mean_token_accuracy": 0.8437739014625549,
+      "num_tokens": 429705.0,
+      "step": 50
+    },
+    {
+      "epoch": 0.11710677382319173,
+      "grad_norm": 3.0510174972572783,
+      "learning_rate": 3.816793893129772e-06,
+      "loss": 0.6075,
+      "mean_token_accuracy": 0.8399601876735687,
+      "num_tokens": 438041.0,
+      "step": 51
+    },
+    {
+      "epoch": 0.11940298507462686,
+      "grad_norm": 2.9575727304243857,
+      "learning_rate": 3.893129770992366e-06,
+      "loss": 0.6145,
+      "mean_token_accuracy": 0.8384945392608643,
+      "num_tokens": 446105.0,
+      "step": 52
+    },
+    {
+      "epoch": 0.12169919632606199,
+      "grad_norm": 3.0479823669864388,
+      "learning_rate": 3.969465648854962e-06,
+      "loss": 0.5908,
+      "mean_token_accuracy": 0.8420686721801758,
+      "num_tokens": 454798.0,
+      "step": 53
+    },
+    {
+      "epoch": 0.12399540757749714,
+      "grad_norm": 2.9022538443937638,
+      "learning_rate": 4.045801526717557e-06,
+      "loss": 0.564,
+      "mean_token_accuracy": 0.860763818025589,
+      "num_tokens": 462816.0,
+      "step": 54
+    },
+    {
+      "epoch": 0.12629161882893225,
+      "grad_norm": 2.8951440693242776,
+      "learning_rate": 4.122137404580153e-06,
+      "loss": 0.6396,
+      "mean_token_accuracy": 0.832654595375061,
+      "num_tokens": 471447.0,
+      "step": 55
+    },
+    {
+      "epoch": 0.1285878300803674,
+      "grad_norm": 3.178934352821412,
+      "learning_rate": 4.198473282442748e-06,
+      "loss": 0.6302,
+      "mean_token_accuracy": 0.8370863497257233,
+      "num_tokens": 479990.0,
+      "step": 56
+    },
+    {
+      "epoch": 0.13088404133180254,
+      "grad_norm": 2.9660810912432085,
+      "learning_rate": 4.274809160305344e-06,
+      "loss": 0.5902,
+      "mean_token_accuracy": 0.8445617258548737,
+      "num_tokens": 488437.0,
+      "step": 57
+    },
+    {
+      "epoch": 0.13318025258323765,
+      "grad_norm": 3.2418880861436588,
+      "learning_rate": 4.351145038167939e-06,
+      "loss": 0.599,
+      "mean_token_accuracy": 0.8445756733417511,
+      "num_tokens": 496772.0,
+      "step": 58
+    },
+    {
+      "epoch": 0.1354764638346728,
+      "grad_norm": 2.5744350059411474,
+      "learning_rate": 4.427480916030535e-06,
+      "loss": 0.5369,
+      "mean_token_accuracy": 0.8557581901550293,
+      "num_tokens": 505157.0,
+      "step": 59
+    },
+    {
+      "epoch": 0.1377726750861079,
+      "grad_norm": 2.958045102309166,
+      "learning_rate": 4.5038167938931296e-06,
+      "loss": 0.5729,
+      "mean_token_accuracy": 0.8472003936767578,
+      "num_tokens": 512703.0,
+      "step": 60
+    },
+    {
+      "epoch": 0.14006888633754305,
+      "grad_norm": 2.747681837443749,
+      "learning_rate": 4.580152671755725e-06,
+      "loss": 0.5506,
+      "mean_token_accuracy": 0.8555885851383209,
+      "num_tokens": 520736.0,
+      "step": 61
+    },
+    {
+      "epoch": 0.1423650975889782,
+      "grad_norm": 2.8784827513027262,
+      "learning_rate": 4.656488549618321e-06,
+      "loss": 0.5936,
+      "mean_token_accuracy": 0.8471256792545319,
+      "num_tokens": 528888.0,
+      "step": 62
+    },
+    {
+      "epoch": 0.1446613088404133,
+      "grad_norm": 2.5567962503787904,
+      "learning_rate": 4.732824427480917e-06,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.868518590927124,
+      "num_tokens": 537348.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.14695752009184845,
+      "grad_norm": 2.554783588236741,
+      "learning_rate": 4.8091603053435125e-06,
+      "loss": 0.562,
+      "mean_token_accuracy": 0.8473564088344574,
+      "num_tokens": 545665.0,
+      "step": 64
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 2.7349592098250914,
+      "learning_rate": 4.885496183206107e-06,
+      "loss": 0.6032,
+      "mean_token_accuracy": 0.8428749442100525,
+      "num_tokens": 554418.0,
+      "step": 65
+    },
+    {
+      "epoch": 0.1515499425947187,
+      "grad_norm": 2.667593100823692,
+      "learning_rate": 4.961832061068703e-06,
+      "loss": 0.555,
+      "mean_token_accuracy": 0.8565230667591095,
+      "num_tokens": 562822.0,
+      "step": 66
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 2.8153474231015765,
+      "learning_rate": 5.038167938931297e-06,
+      "loss": 0.5907,
+      "mean_token_accuracy": 0.8422298431396484,
+      "num_tokens": 571044.0,
+      "step": 67
+    },
+    {
+      "epoch": 0.15614236509758897,
+      "grad_norm": 2.689714383676648,
+      "learning_rate": 5.114503816793893e-06,
+      "loss": 0.5536,
+      "mean_token_accuracy": 0.8536189496517181,
+      "num_tokens": 579919.0,
+      "step": 68
+    },
+    {
+      "epoch": 0.1584385763490241,
+      "grad_norm": 2.8286865533809307,
+      "learning_rate": 5.190839694656488e-06,
+      "loss": 0.5475,
+      "mean_token_accuracy": 0.8530727624893188,
+      "num_tokens": 588187.0,
+      "step": 69
+    },
+    {
+      "epoch": 0.16073478760045926,
+      "grad_norm": 2.5732116098652824,
+      "learning_rate": 5.267175572519084e-06,
+      "loss": 0.5816,
+      "mean_token_accuracy": 0.8440060615539551,
+      "num_tokens": 597070.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.16303099885189437,
+      "grad_norm": 5.849662244774497,
+      "learning_rate": 5.34351145038168e-06,
+      "loss": 0.6158,
+      "mean_token_accuracy": 0.8365760147571564,
+      "num_tokens": 605974.0,
+      "step": 71
+    },
+    {
+      "epoch": 0.1653272101033295,
+      "grad_norm": 2.5861217316563847,
+      "learning_rate": 5.419847328244276e-06,
+      "loss": 0.5192,
+      "mean_token_accuracy": 0.8570695519447327,
+      "num_tokens": 614480.0,
+      "step": 72
+    },
+    {
+      "epoch": 0.16762342135476463,
+      "grad_norm": 7.858950865954448,
+      "learning_rate": 5.496183206106871e-06,
+      "loss": 0.5965,
+      "mean_token_accuracy": 0.8373670876026154,
+      "num_tokens": 623602.0,
+      "step": 73
+    },
+    {
+      "epoch": 0.16991963260619977,
+      "grad_norm": 2.6936755625283313,
+      "learning_rate": 5.572519083969467e-06,
+      "loss": 0.5896,
+      "mean_token_accuracy": 0.8433025479316711,
+      "num_tokens": 632394.0,
+      "step": 74
+    },
+    {
+      "epoch": 0.17221584385763491,
+      "grad_norm": 2.4981688926587515,
+      "learning_rate": 5.648854961832062e-06,
+      "loss": 0.5495,
+      "mean_token_accuracy": 0.8506866991519928,
+      "num_tokens": 641080.0,
+      "step": 75
+    },
+    {
+      "epoch": 0.17451205510907003,
+      "grad_norm": 2.3561163599059847,
+      "learning_rate": 5.725190839694656e-06,
+      "loss": 0.5381,
+      "mean_token_accuracy": 0.8588296175003052,
+      "num_tokens": 649932.0,
+      "step": 76
+    },
+    {
+      "epoch": 0.17680826636050517,
+      "grad_norm": 2.5320262536505345,
+      "learning_rate": 5.801526717557252e-06,
+      "loss": 0.5571,
+      "mean_token_accuracy": 0.856976330280304,
+      "num_tokens": 659106.0,
+      "step": 77
+    },
+    {
+      "epoch": 0.1791044776119403,
+      "grad_norm": 2.528882321221049,
+      "learning_rate": 5.877862595419848e-06,
+      "loss": 0.5177,
+      "mean_token_accuracy": 0.8589673340320587,
+      "num_tokens": 667671.0,
+      "step": 78
+    },
+    {
+      "epoch": 0.18140068886337543,
+      "grad_norm": 2.630114518607686,
+      "learning_rate": 5.9541984732824435e-06,
+      "loss": 0.5406,
+      "mean_token_accuracy": 0.854421466588974,
+      "num_tokens": 676107.0,
+      "step": 79
+    },
+    {
+      "epoch": 0.18369690011481057,
+      "grad_norm": 2.5274863895567816,
+      "learning_rate": 6.030534351145039e-06,
+      "loss": 0.5577,
+      "mean_token_accuracy": 0.8500943183898926,
+      "num_tokens": 684508.0,
+      "step": 80
+    },
+    {
+      "epoch": 0.1859931113662457,
+      "grad_norm": 2.6222083677705434,
+      "learning_rate": 6.1068702290076346e-06,
+      "loss": 0.5441,
+      "mean_token_accuracy": 0.8564147651195526,
+      "num_tokens": 692538.0,
+      "step": 81
+    },
+    {
+      "epoch": 0.18828932261768083,
+      "grad_norm": 2.5341941703229,
+      "learning_rate": 6.18320610687023e-06,
+      "loss": 0.6005,
+      "mean_token_accuracy": 0.8398896753787994,
+      "num_tokens": 701870.0,
+      "step": 82
+    },
+    {
+      "epoch": 0.19058553386911595,
+      "grad_norm": 2.19845227745982,
+      "learning_rate": 6.259541984732826e-06,
+      "loss": 0.5228,
+      "mean_token_accuracy": 0.8666427433490753,
+      "num_tokens": 710061.0,
+      "step": 83
+    },
+    {
+      "epoch": 0.1928817451205511,
+      "grad_norm": 2.4502405682898414,
+      "learning_rate": 6.335877862595419e-06,
+      "loss": 0.5663,
+      "mean_token_accuracy": 0.8515998423099518,
+      "num_tokens": 718456.0,
+      "step": 84
+    },
+    {
+      "epoch": 0.19517795637198623,
+      "grad_norm": 2.4221284279268005,
+      "learning_rate": 6.412213740458016e-06,
+      "loss": 0.529,
+      "mean_token_accuracy": 0.8607643842697144,
+      "num_tokens": 726983.0,
+      "step": 85
+    },
+    {
+      "epoch": 0.19747416762342135,
+      "grad_norm": 2.497913322964698,
+      "learning_rate": 6.488549618320611e-06,
+      "loss": 0.5486,
+      "mean_token_accuracy": 0.8596155345439911,
+      "num_tokens": 734786.0,
+      "step": 86
+    },
+    {
+      "epoch": 0.1997703788748565,
+      "grad_norm": 2.4533449424306175,
+      "learning_rate": 6.564885496183207e-06,
+      "loss": 0.5277,
+      "mean_token_accuracy": 0.8579114973545074,
+      "num_tokens": 743469.0,
+      "step": 87
+    },
+    {
+      "epoch": 0.2020665901262916,
+      "grad_norm": 2.5957962505302534,
+      "learning_rate": 6.641221374045802e-06,
+      "loss": 0.5586,
+      "mean_token_accuracy": 0.8558233380317688,
+      "num_tokens": 751212.0,
+      "step": 88
+    },
+    {
+      "epoch": 0.20436280137772675,
+      "grad_norm": 2.4850657467380572,
+      "learning_rate": 6.717557251908398e-06,
+      "loss": 0.5377,
+      "mean_token_accuracy": 0.8554850220680237,
+      "num_tokens": 759911.0,
+      "step": 89
+    },
+    {
+      "epoch": 0.2066590126291619,
+      "grad_norm": 2.3867682558423904,
+      "learning_rate": 6.793893129770993e-06,
+      "loss": 0.5316,
+      "mean_token_accuracy": 0.8591917753219604,
+      "num_tokens": 768497.0,
+      "step": 90
+    },
+    {
+      "epoch": 0.208955223880597,
+      "grad_norm": 2.4307238970274887,
+      "learning_rate": 6.870229007633589e-06,
+      "loss": 0.5468,
+      "mean_token_accuracy": 0.8605937361717224,
+      "num_tokens": 776742.0,
+      "step": 91
+    },
+    {
+      "epoch": 0.21125143513203215,
+      "grad_norm": 2.405141199872396,
+      "learning_rate": 6.946564885496184e-06,
+      "loss": 0.5774,
+      "mean_token_accuracy": 0.849651038646698,
+      "num_tokens": 784881.0,
+      "step": 92
+    },
+    {
+      "epoch": 0.21354764638346727,
+      "grad_norm": 2.4307188959204082,
+      "learning_rate": 7.022900763358779e-06,
+      "loss": 0.5604,
+      "mean_token_accuracy": 0.8517379760742188,
+      "num_tokens": 793391.0,
+      "step": 93
+    },
+    {
+      "epoch": 0.2158438576349024,
+      "grad_norm": 2.5682481827212165,
+      "learning_rate": 7.0992366412213746e-06,
+      "loss": 0.5989,
+      "mean_token_accuracy": 0.8400489985942841,
+      "num_tokens": 801922.0,
+      "step": 94
+    },
+    {
+      "epoch": 0.21814006888633755,
+      "grad_norm": 2.3868424445839085,
+      "learning_rate": 7.17557251908397e-06,
+      "loss": 0.5101,
+      "mean_token_accuracy": 0.8658720552921295,
+      "num_tokens": 810231.0,
+      "step": 95
+    },
+    {
+      "epoch": 0.22043628013777267,
+      "grad_norm": 2.5306315743707013,
+      "learning_rate": 7.251908396946566e-06,
+      "loss": 0.5782,
+      "mean_token_accuracy": 0.8491176962852478,
+      "num_tokens": 819238.0,
+      "step": 96
+    },
+    {
+      "epoch": 0.2227324913892078,
+      "grad_norm": 2.323818413984369,
+      "learning_rate": 7.328244274809161e-06,
+      "loss": 0.5055,
+      "mean_token_accuracy": 0.8615824282169342,
+      "num_tokens": 827788.0,
+      "step": 97
+    },
+    {
+      "epoch": 0.22502870264064295,
+      "grad_norm": 2.5698591864846794,
+      "learning_rate": 7.404580152671757e-06,
+      "loss": 0.5986,
+      "mean_token_accuracy": 0.8426684439182281,
+      "num_tokens": 836115.0,
+      "step": 98
+    },
+    {
+      "epoch": 0.22732491389207807,
+      "grad_norm": 2.1880670222795744,
+      "learning_rate": 7.480916030534352e-06,
+      "loss": 0.499,
+      "mean_token_accuracy": 0.8672885894775391,
+      "num_tokens": 844248.0,
+      "step": 99
+    },
+    {
+      "epoch": 0.2296211251435132,
+      "grad_norm": 2.5202733991533433,
+      "learning_rate": 7.557251908396948e-06,
+      "loss": 0.5552,
+      "mean_token_accuracy": 0.8480934798717499,
+      "num_tokens": 852654.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.23191733639494833,
+      "grad_norm": 2.4336080494083943,
+      "learning_rate": 7.633587786259543e-06,
+      "loss": 0.5644,
+      "mean_token_accuracy": 0.851811408996582,
+      "num_tokens": 860884.0,
+      "step": 101
+    },
+    {
+      "epoch": 0.23421354764638347,
+      "grad_norm": 2.350365215733331,
+      "learning_rate": 7.709923664122137e-06,
+      "loss": 0.5047,
+      "mean_token_accuracy": 0.8630471229553223,
+      "num_tokens": 869196.0,
+      "step": 102
+    },
+    {
+      "epoch": 0.2365097588978186,
+      "grad_norm": 2.485039268596074,
+      "learning_rate": 7.786259541984733e-06,
+      "loss": 0.5327,
+      "mean_token_accuracy": 0.8535143733024597,
+      "num_tokens": 878068.0,
+      "step": 103
+    },
+    {
+      "epoch": 0.23880597014925373,
+      "grad_norm": 2.4916200640810553,
+      "learning_rate": 7.862595419847328e-06,
+      "loss": 0.5126,
+      "mean_token_accuracy": 0.8594927787780762,
+      "num_tokens": 886056.0,
+      "step": 104
+    },
+    {
+      "epoch": 0.24110218140068887,
+      "grad_norm": 2.4312924658418322,
+      "learning_rate": 7.938931297709924e-06,
+      "loss": 0.5797,
+      "mean_token_accuracy": 0.8443798422813416,
+      "num_tokens": 894931.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.24339839265212398,
+      "grad_norm": 2.347741850201427,
+      "learning_rate": 8.015267175572519e-06,
+      "loss": 0.5672,
+      "mean_token_accuracy": 0.8537621200084686,
+      "num_tokens": 903283.0,
+      "step": 106
+    },
+    {
+      "epoch": 0.24569460390355913,
+      "grad_norm": 2.5127932895755154,
+      "learning_rate": 8.091603053435115e-06,
+      "loss": 0.5644,
+      "mean_token_accuracy": 0.8454359173774719,
+      "num_tokens": 911782.0,
+      "step": 107
+    },
+    {
+      "epoch": 0.24799081515499427,
+      "grad_norm": 2.4448237796636456,
+      "learning_rate": 8.16793893129771e-06,
+      "loss": 0.5323,
+      "mean_token_accuracy": 0.8550295531749725,
+      "num_tokens": 920421.0,
+      "step": 108
+    },
+    {
+      "epoch": 0.2502870264064294,
+      "grad_norm": 2.392726350588866,
+      "learning_rate": 8.244274809160306e-06,
+      "loss": 0.5379,
+      "mean_token_accuracy": 0.8547811210155487,
+      "num_tokens": 928935.0,
+      "step": 109
+    },
+    {
+      "epoch": 0.2525832376578645,
+      "grad_norm": 2.6525329862737506,
+      "learning_rate": 8.320610687022901e-06,
+      "loss": 0.5695,
+      "mean_token_accuracy": 0.8481451272964478,
+      "num_tokens": 937396.0,
+      "step": 110
+    },
+    {
+      "epoch": 0.25487944890929964,
+      "grad_norm": 2.2738462347905792,
+      "learning_rate": 8.396946564885497e-06,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8655006587505341,
+      "num_tokens": 945896.0,
+      "step": 111
+    },
+    {
+      "epoch": 0.2571756601607348,
+      "grad_norm": 2.5987367166075663,
+      "learning_rate": 8.473282442748092e-06,
+      "loss": 0.5504,
+      "mean_token_accuracy": 0.8562089204788208,
+      "num_tokens": 954093.0,
+      "step": 112
+    },
+    {
+      "epoch": 0.25947187141216993,
+      "grad_norm": 2.36155258245816,
+      "learning_rate": 8.549618320610688e-06,
+      "loss": 0.5616,
+      "mean_token_accuracy": 0.8483283519744873,
+      "num_tokens": 962476.0,
+      "step": 113
+    },
+    {
+      "epoch": 0.2617680826636051,
+      "grad_norm": 2.194907056839095,
+      "learning_rate": 8.625954198473283e-06,
+      "loss": 0.4965,
+      "mean_token_accuracy": 0.8635825514793396,
+      "num_tokens": 970713.0,
+      "step": 114
+    },
+    {
+      "epoch": 0.26406429391504016,
+      "grad_norm": 2.2797088830967955,
+      "learning_rate": 8.702290076335879e-06,
+      "loss": 0.5882,
+      "mean_token_accuracy": 0.8490095436573029,
+      "num_tokens": 978757.0,
+      "step": 115
+    },
+    {
+      "epoch": 0.2663605051664753,
+      "grad_norm": 2.179782854729792,
+      "learning_rate": 8.778625954198474e-06,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8600305914878845,
+      "num_tokens": 987362.0,
+      "step": 116
+    },
+    {
+      "epoch": 0.26865671641791045,
+      "grad_norm": 2.507994318085238,
+      "learning_rate": 8.85496183206107e-06,
+      "loss": 0.5925,
+      "mean_token_accuracy": 0.8439347445964813,
+      "num_tokens": 995328.0,
+      "step": 117
+    },
+    {
+      "epoch": 0.2709529276693456,
+      "grad_norm": 2.3533155243906405,
+      "learning_rate": 8.931297709923665e-06,
+      "loss": 0.5078,
+      "mean_token_accuracy": 0.8636853694915771,
+      "num_tokens": 1003723.0,
+      "step": 118
+    },
+    {
+      "epoch": 0.27324913892078073,
+      "grad_norm": 2.299727519924849,
+      "learning_rate": 9.007633587786259e-06,
+      "loss": 0.512,
+      "mean_token_accuracy": 0.8651121854782104,
+      "num_tokens": 1011550.0,
+      "step": 119
+    },
+    {
+      "epoch": 0.2755453501722158,
+      "grad_norm": 2.4569069368377594,
+      "learning_rate": 9.083969465648855e-06,
+      "loss": 0.546,
+      "mean_token_accuracy": 0.8517916202545166,
+      "num_tokens": 1020138.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.27784156142365096,
+      "grad_norm": 2.3409307229485283,
+      "learning_rate": 9.16030534351145e-06,
+      "loss": 0.5471,
+      "mean_token_accuracy": 0.859146237373352,
+      "num_tokens": 1028602.0,
+      "step": 121
+    },
+    {
+      "epoch": 0.2801377726750861,
+      "grad_norm": 2.379646332933809,
+      "learning_rate": 9.236641221374046e-06,
+      "loss": 0.5159,
+      "mean_token_accuracy": 0.8663808405399323,
+      "num_tokens": 1036690.0,
+      "step": 122
+    },
+    {
+      "epoch": 0.28243398392652125,
+      "grad_norm": 2.2073239575090406,
+      "learning_rate": 9.312977099236641e-06,
+      "loss": 0.4916,
+      "mean_token_accuracy": 0.8642817437648773,
+      "num_tokens": 1044943.0,
+      "step": 123
+    },
+    {
+      "epoch": 0.2847301951779564,
+      "grad_norm": 2.451649968262021,
+      "learning_rate": 9.389312977099237e-06,
+      "loss": 0.5364,
+      "mean_token_accuracy": 0.8598155379295349,
+      "num_tokens": 1053356.0,
+      "step": 124
+    },
+    {
+      "epoch": 0.2870264064293915,
+      "grad_norm": 2.2985242962579795,
+      "learning_rate": 9.465648854961834e-06,
+      "loss": 0.4997,
+      "mean_token_accuracy": 0.8676440715789795,
+      "num_tokens": 1061401.0,
+      "step": 125
+    },
+    {
+      "epoch": 0.2893226176808266,
+      "grad_norm": 2.393652411845452,
+      "learning_rate": 9.54198473282443e-06,
+      "loss": 0.5344,
+      "mean_token_accuracy": 0.8611836135387421,
+      "num_tokens": 1069993.0,
+      "step": 126
+    },
+    {
+      "epoch": 0.29161882893226176,
+      "grad_norm": 2.3695856814551415,
+      "learning_rate": 9.618320610687025e-06,
+      "loss": 0.5229,
+      "mean_token_accuracy": 0.858459860086441,
+      "num_tokens": 1078256.0,
+      "step": 127
+    },
+    {
+      "epoch": 0.2939150401836969,
+      "grad_norm": 2.576183944518372,
+      "learning_rate": 9.694656488549619e-06,
+      "loss": 0.5536,
+      "mean_token_accuracy": 0.85035640001297,
+      "num_tokens": 1087031.0,
+      "step": 128
+    },
+    {
+      "epoch": 0.29621125143513205,
+      "grad_norm": 2.1198193178784597,
+      "learning_rate": 9.770992366412214e-06,
+      "loss": 0.5305,
+      "mean_token_accuracy": 0.8602022230625153,
+      "num_tokens": 1096250.0,
+      "step": 129
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 2.23860990795247,
+      "learning_rate": 9.84732824427481e-06,
+      "loss": 0.5643,
+      "mean_token_accuracy": 0.8492049276828766,
+      "num_tokens": 1104790.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.3008036739380023,
+      "grad_norm": 2.154926263550849,
+      "learning_rate": 9.923664122137405e-06,
+      "loss": 0.513,
+      "mean_token_accuracy": 0.8637346625328064,
+      "num_tokens": 1113021.0,
+      "step": 131
+    },
+    {
+      "epoch": 0.3030998851894374,
+      "grad_norm": 2.163076376767292,
+      "learning_rate": 1e-05,
+      "loss": 0.5084,
+      "mean_token_accuracy": 0.8631474375724792,
+      "num_tokens": 1122313.0,
+      "step": 132
+    },
+    {
+      "epoch": 0.30539609644087257,
+      "grad_norm": 2.4978159436519296,
+      "learning_rate": 9.99998397016397e-06,
+      "loss": 0.5276,
+      "mean_token_accuracy": 0.8581156134605408,
+      "num_tokens": 1130232.0,
+      "step": 133
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 2.170014006004013,
+      "learning_rate": 9.999935880770081e-06,
+      "loss": 0.5242,
+      "mean_token_accuracy": 0.859887570142746,
+      "num_tokens": 1138793.0,
+      "step": 134
+    },
+    {
+      "epoch": 0.3099885189437428,
+      "grad_norm": 2.1951195758473037,
+      "learning_rate": 9.99985573216094e-06,
+      "loss": 0.5455,
+      "mean_token_accuracy": 0.8564448654651642,
+      "num_tokens": 1147456.0,
+      "step": 135
+    },
+    {
+      "epoch": 0.31228473019517794,
+      "grad_norm": 2.3116175804876464,
+      "learning_rate": 9.999743524907555e-06,
+      "loss": 0.5115,
+      "mean_token_accuracy": 0.8596856892108917,
+      "num_tokens": 1156362.0,
+      "step": 136
+    },
+    {
+      "epoch": 0.3145809414466131,
+      "grad_norm": 2.2627777768773276,
+      "learning_rate": 9.999599259809334e-06,
+      "loss": 0.5325,
+      "mean_token_accuracy": 0.8546680808067322,
+      "num_tokens": 1164775.0,
+      "step": 137
+    },
+    {
+      "epoch": 0.3168771526980482,
+      "grad_norm": 2.4317650374082453,
+      "learning_rate": 9.999422937894071e-06,
+      "loss": 0.5752,
+      "mean_token_accuracy": 0.847062736749649,
+      "num_tokens": 1173416.0,
+      "step": 138
+    },
+    {
+      "epoch": 0.31917336394948337,
+      "grad_norm": 2.063046040090074,
+      "learning_rate": 9.999214560417954e-06,
+      "loss": 0.4966,
+      "mean_token_accuracy": 0.8685108721256256,
+      "num_tokens": 1183393.0,
+      "step": 139
+    },
+    {
+      "epoch": 0.3214695752009185,
+      "grad_norm": 2.150627827419283,
+      "learning_rate": 9.998974128865539e-06,
+      "loss": 0.5295,
+      "mean_token_accuracy": 0.8605220317840576,
+      "num_tokens": 1192119.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.3237657864523536,
+      "grad_norm": 2.354864272278774,
+      "learning_rate": 9.998701644949749e-06,
+      "loss": 0.5651,
+      "mean_token_accuracy": 0.8480758965015411,
+      "num_tokens": 1200875.0,
+      "step": 141
+    },
+    {
+      "epoch": 0.32606199770378874,
+      "grad_norm": 2.2008553451668096,
+      "learning_rate": 9.998397110611862e-06,
+      "loss": 0.4888,
+      "mean_token_accuracy": 0.8692175149917603,
+      "num_tokens": 1209093.0,
+      "step": 142
+    },
+    {
+      "epoch": 0.3283582089552239,
+      "grad_norm": 2.1538255253627505,
+      "learning_rate": 9.998060528021493e-06,
+      "loss": 0.5174,
+      "mean_token_accuracy": 0.8634341061115265,
+      "num_tokens": 1217068.0,
+      "step": 143
+    },
+    {
+      "epoch": 0.330654420206659,
+      "grad_norm": 2.5696020084067923,
+      "learning_rate": 9.997691899576582e-06,
+      "loss": 0.5769,
+      "mean_token_accuracy": 0.8445126116275787,
+      "num_tokens": 1225355.0,
+      "step": 144
+    },
+    {
+      "epoch": 0.33295063145809417,
+      "grad_norm": 2.379182782604769,
+      "learning_rate": 9.997291227903376e-06,
+      "loss": 0.57,
+      "mean_token_accuracy": 0.8497141003608704,
+      "num_tokens": 1233546.0,
+      "step": 145
+    },
+    {
+      "epoch": 0.33524684270952926,
+      "grad_norm": 2.2267177706670447,
+      "learning_rate": 9.996858515856407e-06,
+      "loss": 0.5407,
+      "mean_token_accuracy": 0.8594411015510559,
+      "num_tokens": 1241830.0,
+      "step": 146
+    },
+    {
+      "epoch": 0.3375430539609644,
+      "grad_norm": 2.3221507176971197,
+      "learning_rate": 9.996393766518477e-06,
+      "loss": 0.5756,
+      "mean_token_accuracy": 0.8457295000553131,
+      "num_tokens": 1249706.0,
+      "step": 147
+    },
+    {
+      "epoch": 0.33983926521239954,
+      "grad_norm": 2.140923508085701,
+      "learning_rate": 9.995896983200633e-06,
+      "loss": 0.5434,
+      "mean_token_accuracy": 0.8575560450553894,
+      "num_tokens": 1257868.0,
+      "step": 148
+    },
+    {
+      "epoch": 0.3421354764638347,
+      "grad_norm": 2.309528822285661,
+      "learning_rate": 9.995368169442145e-06,
+      "loss": 0.5258,
+      "mean_token_accuracy": 0.8543792366981506,
+      "num_tokens": 1266632.0,
+      "step": 149
+    },
+    {
+      "epoch": 0.34443168771526983,
+      "grad_norm": 2.0445613964116958,
+      "learning_rate": 9.994807329010477e-06,
+      "loss": 0.4782,
+      "mean_token_accuracy": 0.8672169744968414,
+      "num_tokens": 1275484.0,
+      "step": 150
+    },
+    {
+      "epoch": 0.3467278989667049,
+      "grad_norm": 2.060779228214477,
+      "learning_rate": 9.994214465901264e-06,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8605171144008636,
+      "num_tokens": 1284250.0,
+      "step": 151
+    },
+    {
+      "epoch": 0.34902411021814006,
+      "grad_norm": 2.2087208906885833,
+      "learning_rate": 9.993589584338286e-06,
+      "loss": 0.4897,
+      "mean_token_accuracy": 0.8722191154956818,
+      "num_tokens": 1293063.0,
+      "step": 152
+    },
+    {
+      "epoch": 0.3513203214695752,
+      "grad_norm": 2.4873915365035737,
+      "learning_rate": 9.99293268877343e-06,
+      "loss": 0.5623,
+      "mean_token_accuracy": 0.8533986508846283,
+      "num_tokens": 1301260.0,
+      "step": 153
+    },
+    {
+      "epoch": 0.35361653272101035,
+      "grad_norm": 2.259024927731419,
+      "learning_rate": 9.992243783886663e-06,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8621634542942047,
+      "num_tokens": 1309919.0,
+      "step": 154
+    },
+    {
+      "epoch": 0.3559127439724455,
+      "grad_norm": 2.052948946762744,
+      "learning_rate": 9.991522874586e-06,
+      "loss": 0.5066,
+      "mean_token_accuracy": 0.8691699802875519,
+      "num_tokens": 1318153.0,
+      "step": 155
+    },
+    {
+      "epoch": 0.3582089552238806,
+      "grad_norm": 2.151498348544008,
+      "learning_rate": 9.990769966007469e-06,
+      "loss": 0.4744,
+      "mean_token_accuracy": 0.8722626268863678,
+      "num_tokens": 1326264.0,
+      "step": 156
+    },
+    {
+      "epoch": 0.3605051664753157,
+      "grad_norm": 2.3708171076279863,
+      "learning_rate": 9.989985063515069e-06,
+      "loss": 0.5829,
+      "mean_token_accuracy": 0.8453091382980347,
+      "num_tokens": 1335624.0,
+      "step": 157
+    },
+    {
+      "epoch": 0.36280137772675086,
+      "grad_norm": 2.194230368528433,
+      "learning_rate": 9.989168172700734e-06,
+      "loss": 0.5516,
+      "mean_token_accuracy": 0.851502001285553,
+      "num_tokens": 1344428.0,
+      "step": 158
+    },
+    {
+      "epoch": 0.365097588978186,
+      "grad_norm": 2.286807783209291,
+      "learning_rate": 9.988319299384301e-06,
+      "loss": 0.5486,
+      "mean_token_accuracy": 0.8558173179626465,
+      "num_tokens": 1352743.0,
+      "step": 159
+    },
+    {
+      "epoch": 0.36739380022962115,
+      "grad_norm": 2.1559663263056055,
+      "learning_rate": 9.98743844961346e-06,
+      "loss": 0.5196,
+      "mean_token_accuracy": 0.8616340458393097,
+      "num_tokens": 1361043.0,
+      "step": 160
+    },
+    {
+      "epoch": 0.36969001148105624,
+      "grad_norm": 2.227348880686468,
+      "learning_rate": 9.986525629663708e-06,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8611706495285034,
+      "num_tokens": 1369683.0,
+      "step": 161
+    },
+    {
+      "epoch": 0.3719862227324914,
+      "grad_norm": 2.3383408083917794,
+      "learning_rate": 9.985580846038315e-06,
+      "loss": 0.5438,
+      "mean_token_accuracy": 0.8501262068748474,
+      "num_tokens": 1378529.0,
+      "step": 162
+    },
+    {
+      "epoch": 0.3742824339839265,
+      "grad_norm": 2.264699690250417,
+      "learning_rate": 9.984604105468274e-06,
+      "loss": 0.5285,
+      "mean_token_accuracy": 0.857488214969635,
+      "num_tokens": 1386328.0,
+      "step": 163
+    },
+    {
+      "epoch": 0.37657864523536166,
+      "grad_norm": 2.234944558951655,
+      "learning_rate": 9.983595414912244e-06,
+      "loss": 0.5345,
+      "mean_token_accuracy": 0.8582321107387543,
+      "num_tokens": 1394462.0,
+      "step": 164
+    },
+    {
+      "epoch": 0.3788748564867968,
+      "grad_norm": 2.2672897804036696,
+      "learning_rate": 9.982554781556512e-06,
+      "loss": 0.5265,
+      "mean_token_accuracy": 0.8563158214092255,
+      "num_tokens": 1402833.0,
+      "step": 165
+    },
+    {
+      "epoch": 0.3811710677382319,
+      "grad_norm": 2.2071965946363448,
+      "learning_rate": 9.981482212814938e-06,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.862550288438797,
+      "num_tokens": 1411486.0,
+      "step": 166
+    },
+    {
+      "epoch": 0.38346727898966704,
+      "grad_norm": 2.3296939281750624,
+      "learning_rate": 9.980377716328898e-06,
+      "loss": 0.5602,
+      "mean_token_accuracy": 0.8496019244194031,
+      "num_tokens": 1420514.0,
+      "step": 167
+    },
+    {
+      "epoch": 0.3857634902411022,
+      "grad_norm": 2.231037644360592,
+      "learning_rate": 9.979241299967238e-06,
+      "loss": 0.5253,
+      "mean_token_accuracy": 0.8627398610115051,
+      "num_tokens": 1429070.0,
+      "step": 168
+    },
+    {
+      "epoch": 0.3880597014925373,
+      "grad_norm": 2.2827962776242274,
+      "learning_rate": 9.978072971826208e-06,
+      "loss": 0.5248,
+      "mean_token_accuracy": 0.8555679619312286,
+      "num_tokens": 1437072.0,
+      "step": 169
+    },
+    {
+      "epoch": 0.39035591274397247,
+      "grad_norm": 2.097797632493463,
+      "learning_rate": 9.976872740229414e-06,
+      "loss": 0.5194,
+      "mean_token_accuracy": 0.8599036037921906,
+      "num_tokens": 1445371.0,
+      "step": 170
+    },
+    {
+      "epoch": 0.39265212399540755,
+      "grad_norm": 2.2550307093151862,
+      "learning_rate": 9.97564061372775e-06,
+      "loss": 0.5292,
+      "mean_token_accuracy": 0.8578495383262634,
+      "num_tokens": 1454016.0,
+      "step": 171
+    },
+    {
+      "epoch": 0.3949483352468427,
+      "grad_norm": 2.2308061221154483,
+      "learning_rate": 9.974376601099343e-06,
+      "loss": 0.5531,
+      "mean_token_accuracy": 0.8519489467144012,
+      "num_tokens": 1462173.0,
+      "step": 172
+    },
+    {
+      "epoch": 0.39724454649827784,
+      "grad_norm": 2.1198854833734075,
+      "learning_rate": 9.973080711349492e-06,
+      "loss": 0.5121,
+      "mean_token_accuracy": 0.863163560628891,
+      "num_tokens": 1470865.0,
+      "step": 173
+    },
+    {
+      "epoch": 0.399540757749713,
+      "grad_norm": 2.277866397169622,
+      "learning_rate": 9.971752953710594e-06,
+      "loss": 0.5144,
+      "mean_token_accuracy": 0.8589484691619873,
+      "num_tokens": 1479084.0,
+      "step": 174
+    },
+    {
+      "epoch": 0.4018369690011481,
+      "grad_norm": 2.1721150373857148,
+      "learning_rate": 9.970393337642088e-06,
+      "loss": 0.5235,
+      "mean_token_accuracy": 0.8603418469429016,
+      "num_tokens": 1487375.0,
+      "step": 175
+    },
+    {
+      "epoch": 0.4041331802525832,
+      "grad_norm": 2.229679048214231,
+      "learning_rate": 9.969001872830383e-06,
+      "loss": 0.5142,
+      "mean_token_accuracy": 0.8607938289642334,
+      "num_tokens": 1495703.0,
+      "step": 176
+    },
+    {
+      "epoch": 0.40642939150401836,
+      "grad_norm": 2.2645911733376862,
+      "learning_rate": 9.967578569188792e-06,
+      "loss": 0.4637,
+      "mean_token_accuracy": 0.8730162680149078,
+      "num_tokens": 1504075.0,
+      "step": 177
+    },
+    {
+      "epoch": 0.4087256027554535,
+      "grad_norm": 2.344506886693289,
+      "learning_rate": 9.966123436857462e-06,
+      "loss": 0.5362,
+      "mean_token_accuracy": 0.8511885404586792,
+      "num_tokens": 1512737.0,
+      "step": 178
+    },
+    {
+      "epoch": 0.41102181400688864,
+      "grad_norm": 2.123125532816247,
+      "learning_rate": 9.964636486203293e-06,
+      "loss": 0.4682,
+      "mean_token_accuracy": 0.8723355531692505,
+      "num_tokens": 1520384.0,
+      "step": 179
+    },
+    {
+      "epoch": 0.4133180252583238,
+      "grad_norm": 2.2860698786777145,
+      "learning_rate": 9.963117727819876e-06,
+      "loss": 0.5057,
+      "mean_token_accuracy": 0.8663891553878784,
+      "num_tokens": 1528537.0,
+      "step": 180
+    },
+    {
+      "epoch": 0.41561423650975887,
+      "grad_norm": 2.403466723782823,
+      "learning_rate": 9.961567172527409e-06,
+      "loss": 0.558,
+      "mean_token_accuracy": 0.8583913743495941,
+      "num_tokens": 1536660.0,
+      "step": 181
+    },
+    {
+      "epoch": 0.417910447761194,
+      "grad_norm": 2.419529844405388,
+      "learning_rate": 9.959984831372622e-06,
+      "loss": 0.5293,
+      "mean_token_accuracy": 0.8597600162029266,
+      "num_tokens": 1544695.0,
+      "step": 182
+    },
+    {
+      "epoch": 0.42020665901262916,
+      "grad_norm": 2.1682073892738396,
+      "learning_rate": 9.958370715628705e-06,
+      "loss": 0.5194,
+      "mean_token_accuracy": 0.8573304712772369,
+      "num_tokens": 1553611.0,
+      "step": 183
+    },
+    {
+      "epoch": 0.4225028702640643,
+      "grad_norm": 2.1774039459826673,
+      "learning_rate": 9.956724836795217e-06,
+      "loss": 0.5163,
+      "mean_token_accuracy": 0.8568669557571411,
+      "num_tokens": 1563061.0,
+      "step": 184
+    },
+    {
+      "epoch": 0.42479908151549944,
+      "grad_norm": 2.2548445417227763,
+      "learning_rate": 9.955047206598011e-06,
+      "loss": 0.5365,
+      "mean_token_accuracy": 0.8523006439208984,
+      "num_tokens": 1571817.0,
+      "step": 185
+    },
+    {
+      "epoch": 0.42709529276693453,
+      "grad_norm": 2.1317510854286383,
+      "learning_rate": 9.953337836989147e-06,
+      "loss": 0.5345,
+      "mean_token_accuracy": 0.8600687980651855,
+      "num_tokens": 1580317.0,
+      "step": 186
+    },
+    {
+      "epoch": 0.4293915040183697,
+      "grad_norm": 2.2331777865269022,
+      "learning_rate": 9.951596740146809e-06,
+      "loss": 0.5033,
+      "mean_token_accuracy": 0.8633032739162445,
+      "num_tokens": 1588797.0,
+      "step": 187
+    },
+    {
+      "epoch": 0.4316877152698048,
+      "grad_norm": 2.215118647248409,
+      "learning_rate": 9.949823928475218e-06,
+      "loss": 0.5266,
+      "mean_token_accuracy": 0.8598288595676422,
+      "num_tokens": 1597052.0,
+      "step": 188
+    },
+    {
+      "epoch": 0.43398392652123996,
+      "grad_norm": 2.2468328336081322,
+      "learning_rate": 9.948019414604542e-06,
+      "loss": 0.5067,
+      "mean_token_accuracy": 0.8644957542419434,
+      "num_tokens": 1605324.0,
+      "step": 189
+    },
+    {
+      "epoch": 0.4362801377726751,
+      "grad_norm": 2.0662245118865536,
+      "learning_rate": 9.946183211390809e-06,
+      "loss": 0.4723,
+      "mean_token_accuracy": 0.8702605366706848,
+      "num_tokens": 1614335.0,
+      "step": 190
+    },
+    {
+      "epoch": 0.43857634902411025,
+      "grad_norm": 1.9344865694839108,
+      "learning_rate": 9.944315331915817e-06,
+      "loss": 0.4322,
+      "mean_token_accuracy": 0.8826346695423126,
+      "num_tokens": 1622436.0,
+      "step": 191
+    },
+    {
+      "epoch": 0.44087256027554533,
+      "grad_norm": 2.236859385895564,
+      "learning_rate": 9.942415789487027e-06,
+      "loss": 0.5262,
+      "mean_token_accuracy": 0.8553067743778229,
+      "num_tokens": 1631082.0,
+      "step": 192
+    },
+    {
+      "epoch": 0.4431687715269805,
+      "grad_norm": 2.2838689355414217,
+      "learning_rate": 9.94048459763749e-06,
+      "loss": 0.5528,
+      "mean_token_accuracy": 0.8519115447998047,
+      "num_tokens": 1639316.0,
+      "step": 193
+    },
+    {
+      "epoch": 0.4454649827784156,
+      "grad_norm": 2.0754368783945516,
+      "learning_rate": 9.938521770125729e-06,
+      "loss": 0.4622,
+      "mean_token_accuracy": 0.874920517206192,
+      "num_tokens": 1647312.0,
+      "step": 194
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 2.1628922500942864,
+      "learning_rate": 9.936527320935664e-06,
+      "loss": 0.4833,
+      "mean_token_accuracy": 0.8715870082378387,
+      "num_tokens": 1655294.0,
+      "step": 195
+    },
+    {
+      "epoch": 0.4500574052812859,
+      "grad_norm": 2.0774307495618136,
+      "learning_rate": 9.934501264276484e-06,
+      "loss": 0.4757,
+      "mean_token_accuracy": 0.8658122420310974,
+      "num_tokens": 1664354.0,
+      "step": 196
+    },
+    {
+      "epoch": 0.452353616532721,
+      "grad_norm": 2.248072160765686,
+      "learning_rate": 9.932443614582576e-06,
+      "loss": 0.5502,
+      "mean_token_accuracy": 0.8551844656467438,
+      "num_tokens": 1672631.0,
+      "step": 197
+    },
+    {
+      "epoch": 0.45464982778415614,
+      "grad_norm": 2.130311257789971,
+      "learning_rate": 9.930354386513399e-06,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8619400858879089,
+      "num_tokens": 1680998.0,
+      "step": 198
+    },
+    {
+      "epoch": 0.4569460390355913,
+      "grad_norm": 2.165430303442314,
+      "learning_rate": 9.928233594953387e-06,
+      "loss": 0.4875,
+      "mean_token_accuracy": 0.8701637089252472,
+      "num_tokens": 1689422.0,
+      "step": 199
+    },
+    {
+      "epoch": 0.4592422502870264,
+      "grad_norm": 2.114427401963167,
+      "learning_rate": 9.92608125501185e-06,
+      "loss": 0.4964,
+      "mean_token_accuracy": 0.8703913986682892,
+      "num_tokens": 1697430.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 2.238022290513048,
+      "learning_rate": 9.923897382022859e-06,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8612644970417023,
+      "num_tokens": 1706504.0,
+      "step": 201
+    },
+    {
+      "epoch": 0.46383467278989665,
+      "grad_norm": 2.20492264496934,
+      "learning_rate": 9.921681991545133e-06,
+      "loss": 0.4588,
+      "mean_token_accuracy": 0.8717695474624634,
+      "num_tokens": 1714908.0,
+      "step": 202
+    },
+    {
+      "epoch": 0.4661308840413318,
+      "grad_norm": 2.2148447596843392,
+      "learning_rate": 9.91943509936194e-06,
+      "loss": 0.519,
+      "mean_token_accuracy": 0.8628759682178497,
+      "num_tokens": 1723546.0,
+      "step": 203
+    },
+    {
+      "epoch": 0.46842709529276694,
+      "grad_norm": 2.204588463348052,
+      "learning_rate": 9.917156721480972e-06,
+      "loss": 0.5057,
+      "mean_token_accuracy": 0.8603301346302032,
+      "num_tokens": 1733139.0,
+      "step": 204
+    },
+    {
+      "epoch": 0.4707233065442021,
+      "grad_norm": 2.137524435433877,
+      "learning_rate": 9.91484687413424e-06,
+      "loss": 0.4772,
+      "mean_token_accuracy": 0.8684304356575012,
+      "num_tokens": 1741722.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.4730195177956372,
+      "grad_norm": 2.0739018400656164,
+      "learning_rate": 9.91250557377796e-06,
+      "loss": 0.4613,
+      "mean_token_accuracy": 0.8703294992446899,
+      "num_tokens": 1750596.0,
+      "step": 206
+    },
+    {
+      "epoch": 0.4753157290470723,
+      "grad_norm": 2.182098942941018,
+      "learning_rate": 9.910132837092417e-06,
+      "loss": 0.483,
+      "mean_token_accuracy": 0.8632851541042328,
+      "num_tokens": 1759420.0,
+      "step": 207
+    },
+    {
+      "epoch": 0.47761194029850745,
+      "grad_norm": 2.1314066386032686,
+      "learning_rate": 9.907728680981875e-06,
+      "loss": 0.479,
+      "mean_token_accuracy": 0.8700879514217377,
+      "num_tokens": 1767498.0,
+      "step": 208
+    },
+    {
+      "epoch": 0.4799081515499426,
+      "grad_norm": 2.152209309286653,
+      "learning_rate": 9.905293122574433e-06,
+      "loss": 0.5212,
+      "mean_token_accuracy": 0.8582695722579956,
+      "num_tokens": 1775706.0,
+      "step": 209
+    },
+    {
+      "epoch": 0.48220436280137774,
+      "grad_norm": 2.1560424193684633,
+      "learning_rate": 9.902826179221916e-06,
+      "loss": 0.5027,
+      "mean_token_accuracy": 0.8670597672462463,
+      "num_tokens": 1783773.0,
+      "step": 210
+    },
+    {
+      "epoch": 0.4845005740528129,
+      "grad_norm": 2.055923926006456,
+      "learning_rate": 9.900327868499743e-06,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8610424995422363,
+      "num_tokens": 1792170.0,
+      "step": 211
+    },
+    {
+      "epoch": 0.48679678530424797,
+      "grad_norm": 2.285218869417989,
+      "learning_rate": 9.897798208206809e-06,
+      "loss": 0.5277,
+      "mean_token_accuracy": 0.8611573278903961,
+      "num_tokens": 1800148.0,
+      "step": 212
+    },
+    {
+      "epoch": 0.4890929965556831,
+      "grad_norm": 2.1946313025402095,
+      "learning_rate": 9.895237216365352e-06,
+      "loss": 0.5171,
+      "mean_token_accuracy": 0.8600347638130188,
+      "num_tokens": 1808864.0,
+      "step": 213
+    },
+    {
+      "epoch": 0.49138920780711826,
+      "grad_norm": 2.1516004774103523,
+      "learning_rate": 9.892644911220834e-06,
+      "loss": 0.5197,
+      "mean_token_accuracy": 0.8644483983516693,
+      "num_tokens": 1816993.0,
+      "step": 214
+    },
+    {
+      "epoch": 0.4936854190585534,
+      "grad_norm": 2.2234255940400964,
+      "learning_rate": 9.890021311241795e-06,
+      "loss": 0.5072,
+      "mean_token_accuracy": 0.8636603355407715,
+      "num_tokens": 1825886.0,
+      "step": 215
+    },
+    {
+      "epoch": 0.49598163030998854,
+      "grad_norm": 2.2789415430170132,
+      "learning_rate": 9.887366435119737e-06,
+      "loss": 0.5786,
+      "mean_token_accuracy": 0.8436568081378937,
+      "num_tokens": 1835004.0,
+      "step": 216
+    },
+    {
+      "epoch": 0.49827784156142363,
+      "grad_norm": 2.047102314131417,
+      "learning_rate": 9.884680301768988e-06,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8621849715709686,
+      "num_tokens": 1843347.0,
+      "step": 217
+    },
+    {
+      "epoch": 0.5005740528128588,
+      "grad_norm": 2.151745329621603,
+      "learning_rate": 9.881962930326554e-06,
+      "loss": 0.5112,
+      "mean_token_accuracy": 0.862118124961853,
+      "num_tokens": 1851377.0,
+      "step": 218
+    },
+    {
+      "epoch": 0.5028702640642939,
+      "grad_norm": 2.166998075887476,
+      "learning_rate": 9.879214340152006e-06,
+      "loss": 0.5327,
+      "mean_token_accuracy": 0.8560000360012054,
+      "num_tokens": 1861368.0,
+      "step": 219
+    },
+    {
+      "epoch": 0.505166475315729,
+      "grad_norm": 2.041967213809994,
+      "learning_rate": 9.876434550827315e-06,
+      "loss": 0.4986,
+      "mean_token_accuracy": 0.8631473183631897,
+      "num_tokens": 1870019.0,
+      "step": 220
+    },
+    {
+      "epoch": 0.5074626865671642,
+      "grad_norm": 2.196897407695158,
+      "learning_rate": 9.873623582156737e-06,
+      "loss": 0.4881,
+      "mean_token_accuracy": 0.864198625087738,
+      "num_tokens": 1879404.0,
+      "step": 221
+    },
+    {
+      "epoch": 0.5097588978185993,
+      "grad_norm": 2.214800788790173,
+      "learning_rate": 9.870781454166657e-06,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8669674098491669,
+      "num_tokens": 1887607.0,
+      "step": 222
+    },
+    {
+      "epoch": 0.5120551090700345,
+      "grad_norm": 2.0383025210584385,
+      "learning_rate": 9.867908187105449e-06,
+      "loss": 0.4714,
+      "mean_token_accuracy": 0.8715745806694031,
+      "num_tokens": 1895671.0,
+      "step": 223
+    },
+    {
+      "epoch": 0.5143513203214696,
+      "grad_norm": 2.02611163178601,
+      "learning_rate": 9.865003801443335e-06,
+      "loss": 0.4492,
+      "mean_token_accuracy": 0.8763221502304077,
+      "num_tokens": 1904302.0,
+      "step": 224
+    },
+    {
+      "epoch": 0.5166475315729047,
+      "grad_norm": 2.1881777288481947,
+      "learning_rate": 9.86206831787224e-06,
+      "loss": 0.4919,
+      "mean_token_accuracy": 0.8689846098423004,
+      "num_tokens": 1912914.0,
+      "step": 225
+    },
+    {
+      "epoch": 0.5189437428243399,
+      "grad_norm": 2.188182515922158,
+      "learning_rate": 9.859101757305639e-06,
+      "loss": 0.4892,
+      "mean_token_accuracy": 0.8691721558570862,
+      "num_tokens": 1921258.0,
+      "step": 226
+    },
+    {
+      "epoch": 0.521239954075775,
+      "grad_norm": 2.1488184551857064,
+      "learning_rate": 9.85610414087841e-06,
+      "loss": 0.5061,
+      "mean_token_accuracy": 0.858322024345398,
+      "num_tokens": 1930608.0,
+      "step": 227
+    },
+    {
+      "epoch": 0.5235361653272101,
+      "grad_norm": 2.2609195720261717,
+      "learning_rate": 9.853075489946686e-06,
+      "loss": 0.5151,
+      "mean_token_accuracy": 0.863874077796936,
+      "num_tokens": 1939054.0,
+      "step": 228
+    },
+    {
+      "epoch": 0.5258323765786452,
+      "grad_norm": 1.9926645825085283,
+      "learning_rate": 9.850015826087706e-06,
+      "loss": 0.4858,
+      "mean_token_accuracy": 0.8705992102622986,
+      "num_tokens": 1947709.0,
+      "step": 229
+    },
+    {
+      "epoch": 0.5281285878300803,
+      "grad_norm": 2.131696078400126,
+      "learning_rate": 9.846925171099648e-06,
+      "loss": 0.4659,
+      "mean_token_accuracy": 0.8738712668418884,
+      "num_tokens": 1956121.0,
+      "step": 230
+    },
+    {
+      "epoch": 0.5304247990815155,
+      "grad_norm": 2.251192057497694,
+      "learning_rate": 9.843803547001487e-06,
+      "loss": 0.537,
+      "mean_token_accuracy": 0.853398859500885,
+      "num_tokens": 1964623.0,
+      "step": 231
+    },
+    {
+      "epoch": 0.5327210103329506,
+      "grad_norm": 2.048408028122145,
+      "learning_rate": 9.840650976032835e-06,
+      "loss": 0.4689,
+      "mean_token_accuracy": 0.871528834104538,
+      "num_tokens": 1973218.0,
+      "step": 232
+    },
+    {
+      "epoch": 0.5350172215843858,
+      "grad_norm": 2.1976812478814307,
+      "learning_rate": 9.837467480653776e-06,
+      "loss": 0.5462,
+      "mean_token_accuracy": 0.8588831424713135,
+      "num_tokens": 1981580.0,
+      "step": 233
+    },
+    {
+      "epoch": 0.5373134328358209,
+      "grad_norm": 2.193559055476676,
+      "learning_rate": 9.834253083544718e-06,
+      "loss": 0.4982,
+      "mean_token_accuracy": 0.8636844456195831,
+      "num_tokens": 1990230.0,
+      "step": 234
+    },
+    {
+      "epoch": 0.539609644087256,
+      "grad_norm": 2.319399520500166,
+      "learning_rate": 9.831007807606217e-06,
+      "loss": 0.5187,
+      "mean_token_accuracy": 0.8605097234249115,
+      "num_tokens": 1999192.0,
+      "step": 235
+    },
+    {
+      "epoch": 0.5419058553386912,
+      "grad_norm": 2.1005570494059054,
+      "learning_rate": 9.827731675958825e-06,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.869504451751709,
+      "num_tokens": 2007926.0,
+      "step": 236
+    },
+    {
+      "epoch": 0.5442020665901263,
+      "grad_norm": 2.067306599261377,
+      "learning_rate": 9.824424711942923e-06,
+      "loss": 0.4765,
+      "mean_token_accuracy": 0.8713276386260986,
+      "num_tokens": 2016832.0,
+      "step": 237
+    },
+    {
+      "epoch": 0.5464982778415615,
+      "grad_norm": 2.0397419629446127,
+      "learning_rate": 9.821086939118549e-06,
+      "loss": 0.4763,
+      "mean_token_accuracy": 0.868956983089447,
+      "num_tokens": 2025496.0,
+      "step": 238
+    },
+    {
+      "epoch": 0.5487944890929966,
+      "grad_norm": 2.33765339992374,
+      "learning_rate": 9.81771838126524e-06,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.864446610212326,
+      "num_tokens": 2034385.0,
+      "step": 239
+    },
+    {
+      "epoch": 0.5510907003444316,
+      "grad_norm": 2.182354571726918,
+      "learning_rate": 9.814319062381851e-06,
+      "loss": 0.4716,
+      "mean_token_accuracy": 0.8674684464931488,
+      "num_tokens": 2042535.0,
+      "step": 240
+    },
+    {
+      "epoch": 0.5533869115958668,
+      "grad_norm": 2.140878493005251,
+      "learning_rate": 9.810889006686393e-06,
+      "loss": 0.4999,
+      "mean_token_accuracy": 0.8609205484390259,
+      "num_tokens": 2050527.0,
+      "step": 241
+    },
+    {
+      "epoch": 0.5556831228473019,
+      "grad_norm": 2.090879265805475,
+      "learning_rate": 9.807428238615858e-06,
+      "loss": 0.4814,
+      "mean_token_accuracy": 0.8661746978759766,
+      "num_tokens": 2058891.0,
+      "step": 242
+    },
+    {
+      "epoch": 0.5579793340987371,
+      "grad_norm": 2.064128140760521,
+      "learning_rate": 9.803936782826045e-06,
+      "loss": 0.4842,
+      "mean_token_accuracy": 0.8673856258392334,
+      "num_tokens": 2067598.0,
+      "step": 243
+    },
+    {
+      "epoch": 0.5602755453501722,
+      "grad_norm": 2.191999862669155,
+      "learning_rate": 9.800414664191379e-06,
+      "loss": 0.517,
+      "mean_token_accuracy": 0.862010270357132,
+      "num_tokens": 2076047.0,
+      "step": 244
+    },
+    {
+      "epoch": 0.5625717566016073,
+      "grad_norm": 2.2980446323667,
+      "learning_rate": 9.796861907804745e-06,
+      "loss": 0.5303,
+      "mean_token_accuracy": 0.8570604622364044,
+      "num_tokens": 2084288.0,
+      "step": 245
+    },
+    {
+      "epoch": 0.5648679678530425,
+      "grad_norm": 2.0928021739502047,
+      "learning_rate": 9.7932785389773e-06,
+      "loss": 0.5055,
+      "mean_token_accuracy": 0.8610836565494537,
+      "num_tokens": 2092450.0,
+      "step": 246
+    },
+    {
+      "epoch": 0.5671641791044776,
+      "grad_norm": 2.265011139075328,
+      "learning_rate": 9.78966458323829e-06,
+      "loss": 0.5149,
+      "mean_token_accuracy": 0.8571164011955261,
+      "num_tokens": 2101073.0,
+      "step": 247
+    },
+    {
+      "epoch": 0.5694603903559128,
+      "grad_norm": 2.1455275838240504,
+      "learning_rate": 9.786020066334884e-06,
+      "loss": 0.4849,
+      "mean_token_accuracy": 0.8698554635047913,
+      "num_tokens": 2109505.0,
+      "step": 248
+    },
+    {
+      "epoch": 0.5717566016073479,
+      "grad_norm": 2.1260232934624597,
+      "learning_rate": 9.782345014231972e-06,
+      "loss": 0.4734,
+      "mean_token_accuracy": 0.8671262264251709,
+      "num_tokens": 2117802.0,
+      "step": 249
+    },
+    {
+      "epoch": 0.574052812858783,
+      "grad_norm": 2.226749614978132,
+      "learning_rate": 9.778639453111993e-06,
+      "loss": 0.5231,
+      "mean_token_accuracy": 0.8561575412750244,
+      "num_tokens": 2126307.0,
+      "step": 250
+    },
+    {
+      "epoch": 0.5763490241102182,
+      "grad_norm": 2.0241107035782853,
+      "learning_rate": 9.774903409374737e-06,
+      "loss": 0.4469,
+      "mean_token_accuracy": 0.8723949491977692,
+      "num_tokens": 2134406.0,
+      "step": 251
+    },
+    {
+      "epoch": 0.5786452353616532,
+      "grad_norm": 2.2309141720885948,
+      "learning_rate": 9.771136909637172e-06,
+      "loss": 0.4991,
+      "mean_token_accuracy": 0.8653621971607208,
+      "num_tokens": 2141961.0,
+      "step": 252
+    },
+    {
+      "epoch": 0.5809414466130884,
+      "grad_norm": 2.2382868790645163,
+      "learning_rate": 9.76733998073324e-06,
+      "loss": 0.4969,
+      "mean_token_accuracy": 0.8631016612052917,
+      "num_tokens": 2150191.0,
+      "step": 253
+    },
+    {
+      "epoch": 0.5832376578645235,
+      "grad_norm": 2.2373122230849853,
+      "learning_rate": 9.763512649713673e-06,
+      "loss": 0.5006,
+      "mean_token_accuracy": 0.8603955805301666,
+      "num_tokens": 2159351.0,
+      "step": 254
+    },
+    {
+      "epoch": 0.5855338691159586,
+      "grad_norm": 2.113840881441921,
+      "learning_rate": 9.7596549438458e-06,
+      "loss": 0.4648,
+      "mean_token_accuracy": 0.8676906228065491,
+      "num_tokens": 2168249.0,
+      "step": 255
+    },
+    {
+      "epoch": 0.5878300803673938,
+      "grad_norm": 2.1293086871471107,
+      "learning_rate": 9.755766890613349e-06,
+      "loss": 0.5137,
+      "mean_token_accuracy": 0.8655804395675659,
+      "num_tokens": 2177310.0,
+      "step": 256
+    },
+    {
+      "epoch": 0.5901262916188289,
+      "grad_norm": 2.062105792980244,
+      "learning_rate": 9.75184851771626e-06,
+      "loss": 0.4934,
+      "mean_token_accuracy": 0.8667555749416351,
+      "num_tokens": 2186596.0,
+      "step": 257
+    },
+    {
+      "epoch": 0.5924225028702641,
+      "grad_norm": 2.0873019446348398,
+      "learning_rate": 9.747899853070474e-06,
+      "loss": 0.4775,
+      "mean_token_accuracy": 0.8633459508419037,
+      "num_tokens": 2195252.0,
+      "step": 258
+    },
+    {
+      "epoch": 0.5947187141216992,
+      "grad_norm": 2.1464083532595835,
+      "learning_rate": 9.743920924807744e-06,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8677262961864471,
+      "num_tokens": 2203443.0,
+      "step": 259
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 1.983203112119313,
+      "learning_rate": 9.739911761275439e-06,
+      "loss": 0.4467,
+      "mean_token_accuracy": 0.8782694935798645,
+      "num_tokens": 2211474.0,
+      "step": 260
+    },
+    {
+      "epoch": 0.5993111366245695,
+      "grad_norm": 2.213302254114781,
+      "learning_rate": 9.735872391036324e-06,
+      "loss": 0.4741,
+      "mean_token_accuracy": 0.8718343675136566,
+      "num_tokens": 2219694.0,
+      "step": 261
+    },
+    {
+      "epoch": 0.6016073478760046,
+      "grad_norm": 2.2360002997866455,
+      "learning_rate": 9.731802842868377e-06,
+      "loss": 0.5306,
+      "mean_token_accuracy": 0.8554394841194153,
+      "num_tokens": 2227790.0,
+      "step": 262
+    },
+    {
+      "epoch": 0.6039035591274398,
+      "grad_norm": 2.123374048544994,
+      "learning_rate": 9.72770314576457e-06,
+      "loss": 0.4791,
+      "mean_token_accuracy": 0.8704572319984436,
+      "num_tokens": 2236083.0,
+      "step": 263
+    },
+    {
+      "epoch": 0.6061997703788748,
+      "grad_norm": 2.189922262951786,
+      "learning_rate": 9.723573328932669e-06,
+      "loss": 0.5195,
+      "mean_token_accuracy": 0.859102874994278,
+      "num_tokens": 2244490.0,
+      "step": 264
+    },
+    {
+      "epoch": 0.6084959816303099,
+      "grad_norm": 1.9440644319947717,
+      "learning_rate": 9.71941342179502e-06,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.879481703042984,
+      "num_tokens": 2252898.0,
+      "step": 265
+    },
+    {
+      "epoch": 0.6107921928817451,
+      "grad_norm": 1.9918295680550606,
+      "learning_rate": 9.715223453988353e-06,
+      "loss": 0.4388,
+      "mean_token_accuracy": 0.8813164830207825,
+      "num_tokens": 2261163.0,
+      "step": 266
+    },
+    {
+      "epoch": 0.6130884041331802,
+      "grad_norm": 2.171141351167282,
+      "learning_rate": 9.711003455363553e-06,
+      "loss": 0.4732,
+      "mean_token_accuracy": 0.8696126937866211,
+      "num_tokens": 2269624.0,
+      "step": 267
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 2.369620901199698,
+      "learning_rate": 9.706753455985457e-06,
+      "loss": 0.5004,
+      "mean_token_accuracy": 0.8622152507305145,
+      "num_tokens": 2278763.0,
+      "step": 268
+    },
+    {
+      "epoch": 0.6176808266360505,
+      "grad_norm": 2.1219447199810535,
+      "learning_rate": 9.70247348613264e-06,
+      "loss": 0.4675,
+      "mean_token_accuracy": 0.8722095787525177,
+      "num_tokens": 2286949.0,
+      "step": 269
+    },
+    {
+      "epoch": 0.6199770378874856,
+      "grad_norm": 1.9993399099730869,
+      "learning_rate": 9.6981635762972e-06,
+      "loss": 0.4139,
+      "mean_token_accuracy": 0.8858119249343872,
+      "num_tokens": 2295132.0,
+      "step": 270
+    },
+    {
+      "epoch": 0.6222732491389208,
+      "grad_norm": 2.1626540788781723,
+      "learning_rate": 9.693823757184532e-06,
+      "loss": 0.4825,
+      "mean_token_accuracy": 0.8686717748641968,
+      "num_tokens": 2303156.0,
+      "step": 271
+    },
+    {
+      "epoch": 0.6245694603903559,
+      "grad_norm": 2.571393568996957,
+      "learning_rate": 9.689454059713124e-06,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.859822690486908,
+      "num_tokens": 2311779.0,
+      "step": 272
+    },
+    {
+      "epoch": 0.6268656716417911,
+      "grad_norm": 2.2117259367968303,
+      "learning_rate": 9.685054515014323e-06,
+      "loss": 0.4484,
+      "mean_token_accuracy": 0.8811606466770172,
+      "num_tokens": 2319747.0,
+      "step": 273
+    },
+    {
+      "epoch": 0.6291618828932262,
+      "grad_norm": 2.1548568168132807,
+      "learning_rate": 9.68062515443212e-06,
+      "loss": 0.4856,
+      "mean_token_accuracy": 0.8694153130054474,
+      "num_tokens": 2327916.0,
+      "step": 274
+    },
+    {
+      "epoch": 0.6314580941446614,
+      "grad_norm": 1.9032628380009569,
+      "learning_rate": 9.676166009522925e-06,
+      "loss": 0.4653,
+      "mean_token_accuracy": 0.8747443556785583,
+      "num_tokens": 2336529.0,
+      "step": 275
+    },
+    {
+      "epoch": 0.6337543053960965,
+      "grad_norm": 2.091201591219223,
+      "learning_rate": 9.671677112055344e-06,
+      "loss": 0.4591,
+      "mean_token_accuracy": 0.8777220249176025,
+      "num_tokens": 2344553.0,
+      "step": 276
+    },
+    {
+      "epoch": 0.6360505166475315,
+      "grad_norm": 1.972516625066699,
+      "learning_rate": 9.667158494009948e-06,
+      "loss": 0.4397,
+      "mean_token_accuracy": 0.8767642080783844,
+      "num_tokens": 2353063.0,
+      "step": 277
+    },
+    {
+      "epoch": 0.6383467278989667,
+      "grad_norm": 2.082665591881392,
+      "learning_rate": 9.662610187579055e-06,
+      "loss": 0.4832,
+      "mean_token_accuracy": 0.865018218755722,
+      "num_tokens": 2361844.0,
+      "step": 278
+    },
+    {
+      "epoch": 0.6406429391504018,
+      "grad_norm": 1.9883549039019055,
+      "learning_rate": 9.658032225166488e-06,
+      "loss": 0.4897,
+      "mean_token_accuracy": 0.8701239228248596,
+      "num_tokens": 2370348.0,
+      "step": 279
+    },
+    {
+      "epoch": 0.642939150401837,
+      "grad_norm": 2.139704380587931,
+      "learning_rate": 9.653424639387354e-06,
+      "loss": 0.4937,
+      "mean_token_accuracy": 0.8692125976085663,
+      "num_tokens": 2378475.0,
+      "step": 280
+    },
+    {
+      "epoch": 0.6452353616532721,
+      "grad_norm": 2.0644840427366327,
+      "learning_rate": 9.6487874630678e-06,
+      "loss": 0.4794,
+      "mean_token_accuracy": 0.8646616339683533,
+      "num_tokens": 2387191.0,
+      "step": 281
+    },
+    {
+      "epoch": 0.6475315729047072,
+      "grad_norm": 2.096331559450681,
+      "learning_rate": 9.644120729244803e-06,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8746865093708038,
+      "num_tokens": 2395305.0,
+      "step": 282
+    },
+    {
+      "epoch": 0.6498277841561424,
+      "grad_norm": 2.1830938713391634,
+      "learning_rate": 9.6394244711659e-06,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8599131405353546,
+      "num_tokens": 2404800.0,
+      "step": 283
+    },
+    {
+      "epoch": 0.6521239954075775,
+      "grad_norm": 2.1488241043227054,
+      "learning_rate": 9.634698722288983e-06,
+      "loss": 0.4779,
+      "mean_token_accuracy": 0.8726445436477661,
+      "num_tokens": 2412975.0,
+      "step": 284
+    },
+    {
+      "epoch": 0.6544202066590127,
+      "grad_norm": 2.2728683561268763,
+      "learning_rate": 9.629943516282041e-06,
+      "loss": 0.5286,
+      "mean_token_accuracy": 0.8544208109378815,
+      "num_tokens": 2421548.0,
+      "step": 285
+    },
+    {
+      "epoch": 0.6567164179104478,
+      "grad_norm": 2.1482545160405837,
+      "learning_rate": 9.62515888702293e-06,
+      "loss": 0.4536,
+      "mean_token_accuracy": 0.8680775165557861,
+      "num_tokens": 2430581.0,
+      "step": 286
+    },
+    {
+      "epoch": 0.6590126291618829,
+      "grad_norm": 2.086800709804692,
+      "learning_rate": 9.620344868599128e-06,
+      "loss": 0.4953,
+      "mean_token_accuracy": 0.8701208531856537,
+      "num_tokens": 2439234.0,
+      "step": 287
+    },
+    {
+      "epoch": 0.661308840413318,
+      "grad_norm": 2.142623104182288,
+      "learning_rate": 9.615501495307487e-06,
+      "loss": 0.4781,
+      "mean_token_accuracy": 0.865425318479538,
+      "num_tokens": 2447477.0,
+      "step": 288
+    },
+    {
+      "epoch": 0.6636050516647531,
+      "grad_norm": 2.161127060410454,
+      "learning_rate": 9.610628801654e-06,
+      "loss": 0.5359,
+      "mean_token_accuracy": 0.8608298003673553,
+      "num_tokens": 2455811.0,
+      "step": 289
+    },
+    {
+      "epoch": 0.6659012629161883,
+      "grad_norm": 2.1916747454696766,
+      "learning_rate": 9.605726822353548e-06,
+      "loss": 0.4756,
+      "mean_token_accuracy": 0.8768877983093262,
+      "num_tokens": 2463982.0,
+      "step": 290
+    },
+    {
+      "epoch": 0.6681974741676234,
+      "grad_norm": 2.0703307856855893,
+      "learning_rate": 9.60079559232965e-06,
+      "loss": 0.4618,
+      "mean_token_accuracy": 0.8752539157867432,
+      "num_tokens": 2472261.0,
+      "step": 291
+    },
+    {
+      "epoch": 0.6704936854190585,
+      "grad_norm": 2.1378404972476486,
+      "learning_rate": 9.595835146714226e-06,
+      "loss": 0.4853,
+      "mean_token_accuracy": 0.8636959791183472,
+      "num_tokens": 2480674.0,
+      "step": 292
+    },
+    {
+      "epoch": 0.6727898966704937,
+      "grad_norm": 2.0109068159543897,
+      "learning_rate": 9.590845520847328e-06,
+      "loss": 0.472,
+      "mean_token_accuracy": 0.8713710308074951,
+      "num_tokens": 2489555.0,
+      "step": 293
+    },
+    {
+      "epoch": 0.6750861079219288,
+      "grad_norm": 2.20775433836355,
+      "learning_rate": 9.585826750276909e-06,
+      "loss": 0.5313,
+      "mean_token_accuracy": 0.8548118770122528,
+      "num_tokens": 2497777.0,
+      "step": 294
+    },
+    {
+      "epoch": 0.677382319173364,
+      "grad_norm": 2.2391769153986747,
+      "learning_rate": 9.580778870758553e-06,
+      "loss": 0.46,
+      "mean_token_accuracy": 0.8769039809703827,
+      "num_tokens": 2506020.0,
+      "step": 295
+    },
+    {
+      "epoch": 0.6796785304247991,
+      "grad_norm": 1.939810112320941,
+      "learning_rate": 9.575701918255231e-06,
+      "loss": 0.4243,
+      "mean_token_accuracy": 0.8804407119750977,
+      "num_tokens": 2514378.0,
+      "step": 296
+    },
+    {
+      "epoch": 0.6819747416762342,
+      "grad_norm": 2.0887871860614147,
+      "learning_rate": 9.57059592893704e-06,
+      "loss": 0.4599,
+      "mean_token_accuracy": 0.8739171922206879,
+      "num_tokens": 2522881.0,
+      "step": 297
+    },
+    {
+      "epoch": 0.6842709529276694,
+      "grad_norm": 2.2769699295686645,
+      "learning_rate": 9.56546093918094e-06,
+      "loss": 0.4915,
+      "mean_token_accuracy": 0.863778293132782,
+      "num_tokens": 2531729.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.6865671641791045,
+      "grad_norm": 2.1037834496180885,
+      "learning_rate": 9.560296985570512e-06,
+      "loss": 0.4888,
+      "mean_token_accuracy": 0.8671713769435883,
+      "num_tokens": 2540326.0,
+      "step": 299
+    },
+    {
+      "epoch": 0.6888633754305397,
+      "grad_norm": 2.015882543674258,
+      "learning_rate": 9.555104104895675e-06,
+      "loss": 0.4659,
+      "mean_token_accuracy": 0.8747215270996094,
+      "num_tokens": 2549209.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.6911595866819747,
+      "grad_norm": 2.0807818179467072,
+      "learning_rate": 9.549882334152445e-06,
+      "loss": 0.4816,
+      "mean_token_accuracy": 0.8685511946678162,
+      "num_tokens": 2557972.0,
+      "step": 301
+    },
+    {
+      "epoch": 0.6934557979334098,
+      "grad_norm": 2.241600420816706,
+      "learning_rate": 9.544631710542654e-06,
+      "loss": 0.4637,
+      "mean_token_accuracy": 0.8750512003898621,
+      "num_tokens": 2566168.0,
+      "step": 302
+    },
+    {
+      "epoch": 0.695752009184845,
+      "grad_norm": 2.1224180103569594,
+      "learning_rate": 9.539352271473698e-06,
+      "loss": 0.4622,
+      "mean_token_accuracy": 0.8731256723403931,
+      "num_tokens": 2575051.0,
+      "step": 303
+    },
+    {
+      "epoch": 0.6980482204362801,
+      "grad_norm": 2.094687480482057,
+      "learning_rate": 9.534044054558262e-06,
+      "loss": 0.4591,
+      "mean_token_accuracy": 0.8775276243686676,
+      "num_tokens": 2583203.0,
+      "step": 304
+    },
+    {
+      "epoch": 0.7003444316877153,
+      "grad_norm": 2.01205602006792,
+      "learning_rate": 9.528707097614052e-06,
+      "loss": 0.4627,
+      "mean_token_accuracy": 0.8756323456764221,
+      "num_tokens": 2591964.0,
+      "step": 305
+    },
+    {
+      "epoch": 0.7026406429391504,
+      "grad_norm": 2.139892756921186,
+      "learning_rate": 9.523341438663538e-06,
+      "loss": 0.4829,
+      "mean_token_accuracy": 0.8637650907039642,
+      "num_tokens": 2600788.0,
+      "step": 306
+    },
+    {
+      "epoch": 0.7049368541905855,
+      "grad_norm": 2.1560674768393158,
+      "learning_rate": 9.517947115933665e-06,
+      "loss": 0.4658,
+      "mean_token_accuracy": 0.8747219145298004,
+      "num_tokens": 2609289.0,
+      "step": 307
+    },
+    {
+      "epoch": 0.7072330654420207,
+      "grad_norm": 2.035012514194049,
+      "learning_rate": 9.51252416785559e-06,
+      "loss": 0.4541,
+      "mean_token_accuracy": 0.8761809468269348,
+      "num_tokens": 2617410.0,
+      "step": 308
+    },
+    {
+      "epoch": 0.7095292766934558,
+      "grad_norm": 2.124488152080928,
+      "learning_rate": 9.507072633064416e-06,
+      "loss": 0.474,
+      "mean_token_accuracy": 0.8689852356910706,
+      "num_tokens": 2625970.0,
+      "step": 309
+    },
+    {
+      "epoch": 0.711825487944891,
+      "grad_norm": 2.0866316745371325,
+      "learning_rate": 9.501592550398901e-06,
+      "loss": 0.4444,
+      "mean_token_accuracy": 0.8802039325237274,
+      "num_tokens": 2634001.0,
+      "step": 310
+    },
+    {
+      "epoch": 0.7141216991963261,
+      "grad_norm": 2.191286334854948,
+      "learning_rate": 9.496083958901186e-06,
+      "loss": 0.4919,
+      "mean_token_accuracy": 0.8646764159202576,
+      "num_tokens": 2642579.0,
+      "step": 311
+    },
+    {
+      "epoch": 0.7164179104477612,
+      "grad_norm": 2.003178036562013,
+      "learning_rate": 9.490546897816526e-06,
+      "loss": 0.4815,
+      "mean_token_accuracy": 0.863891065120697,
+      "num_tokens": 2651960.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.7187141216991964,
+      "grad_norm": 1.8925330007837315,
+      "learning_rate": 9.484981406593002e-06,
+      "loss": 0.408,
+      "mean_token_accuracy": 0.8897265493869781,
+      "num_tokens": 2660565.0,
+      "step": 313
+    },
+    {
+      "epoch": 0.7210103329506314,
+      "grad_norm": 2.1195295413568034,
+      "learning_rate": 9.479387524881242e-06,
+      "loss": 0.4699,
+      "mean_token_accuracy": 0.8697757124900818,
+      "num_tokens": 2668704.0,
+      "step": 314
+    },
+    {
+      "epoch": 0.7233065442020666,
+      "grad_norm": 2.147687869621913,
+      "learning_rate": 9.473765292534136e-06,
+      "loss": 0.5027,
+      "mean_token_accuracy": 0.8629302084445953,
+      "num_tokens": 2677672.0,
+      "step": 315
+    },
+    {
+      "epoch": 0.7256027554535017,
+      "grad_norm": 2.0909916236560186,
+      "learning_rate": 9.46811474960656e-06,
+      "loss": 0.4693,
+      "mean_token_accuracy": 0.8679246604442596,
+      "num_tokens": 2686299.0,
+      "step": 316
+    },
+    {
+      "epoch": 0.7278989667049368,
+      "grad_norm": 2.130311340898713,
+      "learning_rate": 9.462435936355076e-06,
+      "loss": 0.4738,
+      "mean_token_accuracy": 0.8702786862850189,
+      "num_tokens": 2694515.0,
+      "step": 317
+    },
+    {
+      "epoch": 0.730195177956372,
+      "grad_norm": 2.243352243048058,
+      "learning_rate": 9.456728893237663e-06,
+      "loss": 0.4748,
+      "mean_token_accuracy": 0.8736391365528107,
+      "num_tokens": 2703036.0,
+      "step": 318
+    },
+    {
+      "epoch": 0.7324913892078071,
+      "grad_norm": 2.0901270191904326,
+      "learning_rate": 9.450993660913418e-06,
+      "loss": 0.4514,
+      "mean_token_accuracy": 0.8738319873809814,
+      "num_tokens": 2710923.0,
+      "step": 319
+    },
+    {
+      "epoch": 0.7347876004592423,
+      "grad_norm": 2.1498745913902444,
+      "learning_rate": 9.445230280242263e-06,
+      "loss": 0.4446,
+      "mean_token_accuracy": 0.87261563539505,
+      "num_tokens": 2719067.0,
+      "step": 320
+    },
+    {
+      "epoch": 0.7370838117106774,
+      "grad_norm": 2.2426940116091596,
+      "learning_rate": 9.439438792284665e-06,
+      "loss": 0.4567,
+      "mean_token_accuracy": 0.8763305246829987,
+      "num_tokens": 2727589.0,
+      "step": 321
+    },
+    {
+      "epoch": 0.7393800229621125,
+      "grad_norm": 2.3384417905434787,
+      "learning_rate": 9.43361923830134e-06,
+      "loss": 0.4846,
+      "mean_token_accuracy": 0.8693962693214417,
+      "num_tokens": 2736063.0,
+      "step": 322
+    },
+    {
+      "epoch": 0.7416762342135477,
+      "grad_norm": 2.1866762435414038,
+      "learning_rate": 9.427771659752947e-06,
+      "loss": 0.4815,
+      "mean_token_accuracy": 0.8684048056602478,
+      "num_tokens": 2744603.0,
+      "step": 323
+    },
+    {
+      "epoch": 0.7439724454649828,
+      "grad_norm": 2.188207816419705,
+      "learning_rate": 9.421896098299813e-06,
+      "loss": 0.46,
+      "mean_token_accuracy": 0.8795391321182251,
+      "num_tokens": 2752839.0,
+      "step": 324
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 2.034058466599538,
+      "learning_rate": 9.415992595801617e-06,
+      "loss": 0.4922,
+      "mean_token_accuracy": 0.8684838116168976,
+      "num_tokens": 2761340.0,
+      "step": 325
+    },
+    {
+      "epoch": 0.748564867967853,
+      "grad_norm": 2.2203971211583013,
+      "learning_rate": 9.410061194317108e-06,
+      "loss": 0.5506,
+      "mean_token_accuracy": 0.8482409715652466,
+      "num_tokens": 2769640.0,
+      "step": 326
+    },
+    {
+      "epoch": 0.7508610792192881,
+      "grad_norm": 2.1903365714685865,
+      "learning_rate": 9.404101936103793e-06,
+      "loss": 0.4916,
+      "mean_token_accuracy": 0.8678947985172272,
+      "num_tokens": 2777920.0,
+      "step": 327
+    },
+    {
+      "epoch": 0.7531572904707233,
+      "grad_norm": 1.993360860369345,
+      "learning_rate": 9.39811486361764e-06,
+      "loss": 0.4572,
+      "mean_token_accuracy": 0.8756724894046783,
+      "num_tokens": 2787362.0,
+      "step": 328
+    },
+    {
+      "epoch": 0.7554535017221584,
+      "grad_norm": 1.9361661329568167,
+      "learning_rate": 9.39210001951278e-06,
+      "loss": 0.4477,
+      "mean_token_accuracy": 0.8757772147655487,
+      "num_tokens": 2795811.0,
+      "step": 329
+    },
+    {
+      "epoch": 0.7577497129735936,
+      "grad_norm": 1.9649185480893383,
+      "learning_rate": 9.386057446641195e-06,
+      "loss": 0.4637,
+      "mean_token_accuracy": 0.8730275928974152,
+      "num_tokens": 2804058.0,
+      "step": 330
+    },
+    {
+      "epoch": 0.7600459242250287,
+      "grad_norm": 1.968938141614438,
+      "learning_rate": 9.379987188052423e-06,
+      "loss": 0.4666,
+      "mean_token_accuracy": 0.8687047362327576,
+      "num_tokens": 2812684.0,
+      "step": 331
+    },
+    {
+      "epoch": 0.7623421354764638,
+      "grad_norm": 2.121361853270775,
+      "learning_rate": 9.37388928699324e-06,
+      "loss": 0.4779,
+      "mean_token_accuracy": 0.8685332536697388,
+      "num_tokens": 2821372.0,
+      "step": 332
+    },
+    {
+      "epoch": 0.764638346727899,
+      "grad_norm": 2.188578632518775,
+      "learning_rate": 9.367763786907358e-06,
+      "loss": 0.4715,
+      "mean_token_accuracy": 0.8692339956760406,
+      "num_tokens": 2829743.0,
+      "step": 333
+    },
+    {
+      "epoch": 0.7669345579793341,
+      "grad_norm": 2.2730322406850094,
+      "learning_rate": 9.361610731435116e-06,
+      "loss": 0.4811,
+      "mean_token_accuracy": 0.8684225678443909,
+      "num_tokens": 2838868.0,
+      "step": 334
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 2.1891900129718436,
+      "learning_rate": 9.355430164413168e-06,
+      "loss": 0.4883,
+      "mean_token_accuracy": 0.8674668669700623,
+      "num_tokens": 2847584.0,
+      "step": 335
+    },
+    {
+      "epoch": 0.7715269804822044,
+      "grad_norm": 2.200618379627659,
+      "learning_rate": 9.349222129874171e-06,
+      "loss": 0.4928,
+      "mean_token_accuracy": 0.8605104386806488,
+      "num_tokens": 2855661.0,
+      "step": 336
+    },
+    {
+      "epoch": 0.7738231917336394,
+      "grad_norm": 2.2250449378536747,
+      "learning_rate": 9.342986672046467e-06,
+      "loss": 0.498,
+      "mean_token_accuracy": 0.8653375804424286,
+      "num_tokens": 2864402.0,
+      "step": 337
+    },
+    {
+      "epoch": 0.7761194029850746,
+      "grad_norm": 2.32185072142467,
+      "learning_rate": 9.336723835353777e-06,
+      "loss": 0.5521,
+      "mean_token_accuracy": 0.8540611267089844,
+      "num_tokens": 2872951.0,
+      "step": 338
+    },
+    {
+      "epoch": 0.7784156142365097,
+      "grad_norm": 2.055506801136058,
+      "learning_rate": 9.330433664414875e-06,
+      "loss": 0.4185,
+      "mean_token_accuracy": 0.8842295408248901,
+      "num_tokens": 2881457.0,
+      "step": 339
+    },
+    {
+      "epoch": 0.7807118254879449,
+      "grad_norm": 2.165717304924899,
+      "learning_rate": 9.324116204043276e-06,
+      "loss": 0.4488,
+      "mean_token_accuracy": 0.8738264441490173,
+      "num_tokens": 2889775.0,
+      "step": 340
+    },
+    {
+      "epoch": 0.78300803673938,
+      "grad_norm": 2.053486634721317,
+      "learning_rate": 9.317771499246918e-06,
+      "loss": 0.4651,
+      "mean_token_accuracy": 0.8712034225463867,
+      "num_tokens": 2898353.0,
+      "step": 341
+    },
+    {
+      "epoch": 0.7853042479908151,
+      "grad_norm": 2.0130861313842647,
+      "learning_rate": 9.311399595227832e-06,
+      "loss": 0.4531,
+      "mean_token_accuracy": 0.876312255859375,
+      "num_tokens": 2906944.0,
+      "step": 342
+    },
+    {
+      "epoch": 0.7876004592422503,
+      "grad_norm": 1.9888787846013032,
+      "learning_rate": 9.30500053738183e-06,
+      "loss": 0.4363,
+      "mean_token_accuracy": 0.8803764879703522,
+      "num_tokens": 2914848.0,
+      "step": 343
+    },
+    {
+      "epoch": 0.7898966704936854,
+      "grad_norm": 2.1450895450259924,
+      "learning_rate": 9.298574371298182e-06,
+      "loss": 0.4811,
+      "mean_token_accuracy": 0.8657047748565674,
+      "num_tokens": 2923810.0,
+      "step": 344
+    },
+    {
+      "epoch": 0.7921928817451206,
+      "grad_norm": 2.1767640302925955,
+      "learning_rate": 9.292121142759278e-06,
+      "loss": 0.4394,
+      "mean_token_accuracy": 0.8757191002368927,
+      "num_tokens": 2932038.0,
+      "step": 345
+    },
+    {
+      "epoch": 0.7944890929965557,
+      "grad_norm": 2.1590700557546647,
+      "learning_rate": 9.285640897740316e-06,
+      "loss": 0.4301,
+      "mean_token_accuracy": 0.879925549030304,
+      "num_tokens": 2939932.0,
+      "step": 346
+    },
+    {
+      "epoch": 0.7967853042479908,
+      "grad_norm": 2.1113906021633113,
+      "learning_rate": 9.279133682408971e-06,
+      "loss": 0.4521,
+      "mean_token_accuracy": 0.8686864972114563,
+      "num_tokens": 2948964.0,
+      "step": 347
+    },
+    {
+      "epoch": 0.799081515499426,
+      "grad_norm": 2.096691654571855,
+      "learning_rate": 9.272599543125067e-06,
+      "loss": 0.4597,
+      "mean_token_accuracy": 0.8706798851490021,
+      "num_tokens": 2957439.0,
+      "step": 348
+    },
+    {
+      "epoch": 0.801377726750861,
+      "grad_norm": 2.0645969684605974,
+      "learning_rate": 9.266038526440233e-06,
+      "loss": 0.4596,
+      "mean_token_accuracy": 0.8719890117645264,
+      "num_tokens": 2965650.0,
+      "step": 349
+    },
+    {
+      "epoch": 0.8036739380022963,
+      "grad_norm": 2.206821279185941,
+      "learning_rate": 9.259450679097593e-06,
+      "loss": 0.4873,
+      "mean_token_accuracy": 0.867302417755127,
+      "num_tokens": 2974480.0,
+      "step": 350
+    },
+    {
+      "epoch": 0.8059701492537313,
+      "grad_norm": 1.978945278531648,
+      "learning_rate": 9.252836048031419e-06,
+      "loss": 0.4431,
+      "mean_token_accuracy": 0.8774751424789429,
+      "num_tokens": 2983214.0,
+      "step": 351
+    },
+    {
+      "epoch": 0.8082663605051664,
+      "grad_norm": 2.031599198737061,
+      "learning_rate": 9.246194680366802e-06,
+      "loss": 0.4578,
+      "mean_token_accuracy": 0.8689985871315002,
+      "num_tokens": 2991802.0,
+      "step": 352
+    },
+    {
+      "epoch": 0.8105625717566016,
+      "grad_norm": 2.0547611514047075,
+      "learning_rate": 9.239526623419312e-06,
+      "loss": 0.4388,
+      "mean_token_accuracy": 0.8751692771911621,
+      "num_tokens": 3000265.0,
+      "step": 353
+    },
+    {
+      "epoch": 0.8128587830080367,
+      "grad_norm": 1.9615782535846027,
+      "learning_rate": 9.232831924694663e-06,
+      "loss": 0.4234,
+      "mean_token_accuracy": 0.8776982128620148,
+      "num_tokens": 3008802.0,
+      "step": 354
+    },
+    {
+      "epoch": 0.8151549942594719,
+      "grad_norm": 2.1401985072521055,
+      "learning_rate": 9.226110631888378e-06,
+      "loss": 0.483,
+      "mean_token_accuracy": 0.86871337890625,
+      "num_tokens": 3017624.0,
+      "step": 355
+    },
+    {
+      "epoch": 0.817451205510907,
+      "grad_norm": 2.0973007575515568,
+      "learning_rate": 9.219362792885444e-06,
+      "loss": 0.4949,
+      "mean_token_accuracy": 0.8608673512935638,
+      "num_tokens": 3026572.0,
+      "step": 356
+    },
+    {
+      "epoch": 0.8197474167623421,
+      "grad_norm": 2.209735434869826,
+      "learning_rate": 9.21258845575997e-06,
+      "loss": 0.4621,
+      "mean_token_accuracy": 0.869554191827774,
+      "num_tokens": 3035743.0,
+      "step": 357
+    },
+    {
+      "epoch": 0.8220436280137773,
+      "grad_norm": 2.099831936213864,
+      "learning_rate": 9.205787668774857e-06,
+      "loss": 0.4254,
+      "mean_token_accuracy": 0.8812063932418823,
+      "num_tokens": 3044106.0,
+      "step": 358
+    },
+    {
+      "epoch": 0.8243398392652124,
+      "grad_norm": 2.238449519339984,
+      "learning_rate": 9.198960480381434e-06,
+      "loss": 0.4928,
+      "mean_token_accuracy": 0.8663815557956696,
+      "num_tokens": 3052580.0,
+      "step": 359
+    },
+    {
+      "epoch": 0.8266360505166476,
+      "grad_norm": 2.131395911868924,
+      "learning_rate": 9.192106939219129e-06,
+      "loss": 0.4693,
+      "mean_token_accuracy": 0.8705114424228668,
+      "num_tokens": 3061185.0,
+      "step": 360
+    },
+    {
+      "epoch": 0.8289322617680827,
+      "grad_norm": 2.131794087926971,
+      "learning_rate": 9.185227094115115e-06,
+      "loss": 0.4514,
+      "mean_token_accuracy": 0.8736622333526611,
+      "num_tokens": 3069356.0,
+      "step": 361
+    },
+    {
+      "epoch": 0.8312284730195177,
+      "grad_norm": 2.135446393353064,
+      "learning_rate": 9.178320994083968e-06,
+      "loss": 0.4753,
+      "mean_token_accuracy": 0.869425505399704,
+      "num_tokens": 3078053.0,
+      "step": 362
+    },
+    {
+      "epoch": 0.8335246842709529,
+      "grad_norm": 2.0665871523744594,
+      "learning_rate": 9.171388688327307e-06,
+      "loss": 0.4254,
+      "mean_token_accuracy": 0.878003716468811,
+      "num_tokens": 3086365.0,
+      "step": 363
+    },
+    {
+      "epoch": 0.835820895522388,
+      "grad_norm": 2.304034166099164,
+      "learning_rate": 9.164430226233458e-06,
+      "loss": 0.4905,
+      "mean_token_accuracy": 0.8681802153587341,
+      "num_tokens": 3094749.0,
+      "step": 364
+    },
+    {
+      "epoch": 0.8381171067738232,
+      "grad_norm": 2.26841997034153,
+      "learning_rate": 9.157445657377086e-06,
+      "loss": 0.4903,
+      "mean_token_accuracy": 0.8668373823165894,
+      "num_tokens": 3103278.0,
+      "step": 365
+    },
+    {
+      "epoch": 0.8404133180252583,
+      "grad_norm": 2.0403501601010285,
+      "learning_rate": 9.150435031518855e-06,
+      "loss": 0.4499,
+      "mean_token_accuracy": 0.8763203322887421,
+      "num_tokens": 3111400.0,
+      "step": 366
+    },
+    {
+      "epoch": 0.8427095292766934,
+      "grad_norm": 2.027138376569289,
+      "learning_rate": 9.143398398605073e-06,
+      "loss": 0.4402,
+      "mean_token_accuracy": 0.8767527639865875,
+      "num_tokens": 3120330.0,
+      "step": 367
+    },
+    {
+      "epoch": 0.8450057405281286,
+      "grad_norm": 1.9814607760723737,
+      "learning_rate": 9.136335808767324e-06,
+      "loss": 0.4451,
+      "mean_token_accuracy": 0.8783848881721497,
+      "num_tokens": 3128855.0,
+      "step": 368
+    },
+    {
+      "epoch": 0.8473019517795637,
+      "grad_norm": 2.0846317964257564,
+      "learning_rate": 9.129247312322126e-06,
+      "loss": 0.4306,
+      "mean_token_accuracy": 0.8776095807552338,
+      "num_tokens": 3137083.0,
+      "step": 369
+    },
+    {
+      "epoch": 0.8495981630309989,
+      "grad_norm": 2.3054197421685654,
+      "learning_rate": 9.122132959770554e-06,
+      "loss": 0.4866,
+      "mean_token_accuracy": 0.8668666183948517,
+      "num_tokens": 3145311.0,
+      "step": 370
+    },
+    {
+      "epoch": 0.851894374282434,
+      "grad_norm": 2.1439932867689357,
+      "learning_rate": 9.114992801797908e-06,
+      "loss": 0.4867,
+      "mean_token_accuracy": 0.8656428158283234,
+      "num_tokens": 3154170.0,
+      "step": 371
+    },
+    {
+      "epoch": 0.8541905855338691,
+      "grad_norm": 1.924391518675697,
+      "learning_rate": 9.107826889273319e-06,
+      "loss": 0.4086,
+      "mean_token_accuracy": 0.8895063102245331,
+      "num_tokens": 3162814.0,
+      "step": 372
+    },
+    {
+      "epoch": 0.8564867967853043,
+      "grad_norm": 2.124350377583813,
+      "learning_rate": 9.100635273249415e-06,
+      "loss": 0.4455,
+      "mean_token_accuracy": 0.8770100176334381,
+      "num_tokens": 3171035.0,
+      "step": 373
+    },
+    {
+      "epoch": 0.8587830080367393,
+      "grad_norm": 2.3163242994827304,
+      "learning_rate": 9.093418004961939e-06,
+      "loss": 0.4796,
+      "mean_token_accuracy": 0.8700608909130096,
+      "num_tokens": 3180111.0,
+      "step": 374
+    },
+    {
+      "epoch": 0.8610792192881745,
+      "grad_norm": 2.0359642533506075,
+      "learning_rate": 9.086175135829391e-06,
+      "loss": 0.4395,
+      "mean_token_accuracy": 0.8787518441677094,
+      "num_tokens": 3188846.0,
+      "step": 375
+    },
+    {
+      "epoch": 0.8633754305396096,
+      "grad_norm": 2.302463996238575,
+      "learning_rate": 9.078906717452666e-06,
+      "loss": 0.485,
+      "mean_token_accuracy": 0.8640980422496796,
+      "num_tokens": 3197956.0,
+      "step": 376
+    },
+    {
+      "epoch": 0.8656716417910447,
+      "grad_norm": 2.191948420504646,
+      "learning_rate": 9.071612801614672e-06,
+      "loss": 0.4894,
+      "mean_token_accuracy": 0.8683001399040222,
+      "num_tokens": 3206678.0,
+      "step": 377
+    },
+    {
+      "epoch": 0.8679678530424799,
+      "grad_norm": 2.3476316107952924,
+      "learning_rate": 9.06429344027998e-06,
+      "loss": 0.5289,
+      "mean_token_accuracy": 0.8573424518108368,
+      "num_tokens": 3215334.0,
+      "step": 378
+    },
+    {
+      "epoch": 0.870264064293915,
+      "grad_norm": 2.2031609784143606,
+      "learning_rate": 9.056948685594438e-06,
+      "loss": 0.4375,
+      "mean_token_accuracy": 0.8765662014484406,
+      "num_tokens": 3223329.0,
+      "step": 379
+    },
+    {
+      "epoch": 0.8725602755453502,
+      "grad_norm": 2.1479475250515314,
+      "learning_rate": 9.049578589884804e-06,
+      "loss": 0.4437,
+      "mean_token_accuracy": 0.877239465713501,
+      "num_tokens": 3231216.0,
+      "step": 380
+    },
+    {
+      "epoch": 0.8748564867967853,
+      "grad_norm": 2.081584648903112,
+      "learning_rate": 9.042183205658386e-06,
+      "loss": 0.4455,
+      "mean_token_accuracy": 0.8735238015651703,
+      "num_tokens": 3239360.0,
+      "step": 381
+    },
+    {
+      "epoch": 0.8771526980482205,
+      "grad_norm": 2.304560200054502,
+      "learning_rate": 9.034762585602643e-06,
+      "loss": 0.4713,
+      "mean_token_accuracy": 0.864691436290741,
+      "num_tokens": 3247234.0,
+      "step": 382
+    },
+    {
+      "epoch": 0.8794489092996556,
+      "grad_norm": 2.2202161147038106,
+      "learning_rate": 9.027316782584832e-06,
+      "loss": 0.4528,
+      "mean_token_accuracy": 0.8773845136165619,
+      "num_tokens": 3255473.0,
+      "step": 383
+    },
+    {
+      "epoch": 0.8817451205510907,
+      "grad_norm": 2.142993343589632,
+      "learning_rate": 9.01984584965162e-06,
+      "loss": 0.4597,
+      "mean_token_accuracy": 0.8736259043216705,
+      "num_tokens": 3264484.0,
+      "step": 384
+    },
+    {
+      "epoch": 0.8840413318025259,
+      "grad_norm": 2.1732501798792625,
+      "learning_rate": 9.012349840028707e-06,
+      "loss": 0.4481,
+      "mean_token_accuracy": 0.872738391160965,
+      "num_tokens": 3273195.0,
+      "step": 385
+    },
+    {
+      "epoch": 0.886337543053961,
+      "grad_norm": 2.2255690047859074,
+      "learning_rate": 9.004828807120455e-06,
+      "loss": 0.4601,
+      "mean_token_accuracy": 0.8765251040458679,
+      "num_tokens": 3281603.0,
+      "step": 386
+    },
+    {
+      "epoch": 0.8886337543053962,
+      "grad_norm": 2.1258927167945068,
+      "learning_rate": 8.997282804509493e-06,
+      "loss": 0.4346,
+      "mean_token_accuracy": 0.8746431767940521,
+      "num_tokens": 3289953.0,
+      "step": 387
+    },
+    {
+      "epoch": 0.8909299655568312,
+      "grad_norm": 2.1356728352054986,
+      "learning_rate": 8.98971188595635e-06,
+      "loss": 0.4216,
+      "mean_token_accuracy": 0.8815479278564453,
+      "num_tokens": 3297144.0,
+      "step": 388
+    },
+    {
+      "epoch": 0.8932261768082663,
+      "grad_norm": 2.184781661544893,
+      "learning_rate": 8.982116105399063e-06,
+      "loss": 0.485,
+      "mean_token_accuracy": 0.8615638911724091,
+      "num_tokens": 3305963.0,
+      "step": 389
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 2.0793601700320408,
+      "learning_rate": 8.974495516952793e-06,
+      "loss": 0.4221,
+      "mean_token_accuracy": 0.8774198889732361,
+      "num_tokens": 3314603.0,
+      "step": 390
+    },
+    {
+      "epoch": 0.8978185993111366,
+      "grad_norm": 2.0703833148101483,
+      "learning_rate": 8.966850174909446e-06,
+      "loss": 0.4387,
+      "mean_token_accuracy": 0.8746680915355682,
+      "num_tokens": 3323577.0,
+      "step": 391
+    },
+    {
+      "epoch": 0.9001148105625718,
+      "grad_norm": 2.2693011142199246,
+      "learning_rate": 8.959180133737277e-06,
+      "loss": 0.504,
+      "mean_token_accuracy": 0.8631562292575836,
+      "num_tokens": 3332767.0,
+      "step": 392
+    },
+    {
+      "epoch": 0.9024110218140069,
+      "grad_norm": 2.1742557727811564,
+      "learning_rate": 8.951485448080512e-06,
+      "loss": 0.4285,
+      "mean_token_accuracy": 0.8797023594379425,
+      "num_tokens": 3340847.0,
+      "step": 393
+    },
+    {
+      "epoch": 0.904707233065442,
+      "grad_norm": 2.136177012092341,
+      "learning_rate": 8.94376617275895e-06,
+      "loss": 0.4408,
+      "mean_token_accuracy": 0.8818738162517548,
+      "num_tokens": 3349159.0,
+      "step": 394
+    },
+    {
+      "epoch": 0.9070034443168772,
+      "grad_norm": 2.0682344352635527,
+      "learning_rate": 8.936022362767573e-06,
+      "loss": 0.4398,
+      "mean_token_accuracy": 0.8694310486316681,
+      "num_tokens": 3358087.0,
+      "step": 395
+    },
+    {
+      "epoch": 0.9092996555683123,
+      "grad_norm": 2.208646352560754,
+      "learning_rate": 8.928254073276166e-06,
+      "loss": 0.4246,
+      "mean_token_accuracy": 0.8833327293395996,
+      "num_tokens": 3366674.0,
+      "step": 396
+    },
+    {
+      "epoch": 0.9115958668197475,
+      "grad_norm": 2.429886869569641,
+      "learning_rate": 8.920461359628906e-06,
+      "loss": 0.4923,
+      "mean_token_accuracy": 0.8574588298797607,
+      "num_tokens": 3375188.0,
+      "step": 397
+    },
+    {
+      "epoch": 0.9138920780711826,
+      "grad_norm": 2.0838532786422235,
+      "learning_rate": 8.912644277343983e-06,
+      "loss": 0.4476,
+      "mean_token_accuracy": 0.8783973157405853,
+      "num_tokens": 3383208.0,
+      "step": 398
+    },
+    {
+      "epoch": 0.9161882893226176,
+      "grad_norm": 2.164213753880442,
+      "learning_rate": 8.904802882113195e-06,
+      "loss": 0.443,
+      "mean_token_accuracy": 0.875018835067749,
+      "num_tokens": 3391502.0,
+      "step": 399
+    },
+    {
+      "epoch": 0.9184845005740528,
+      "grad_norm": 2.2218555226901104,
+      "learning_rate": 8.896937229801554e-06,
+      "loss": 0.465,
+      "mean_token_accuracy": 0.8699540197849274,
+      "num_tokens": 3399797.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.9207807118254879,
+      "grad_norm": 2.147622790346399,
+      "learning_rate": 8.889047376446893e-06,
+      "loss": 0.4419,
+      "mean_token_accuracy": 0.8751032650470734,
+      "num_tokens": 3408170.0,
+      "step": 401
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 2.255811346095756,
+      "learning_rate": 8.881133378259455e-06,
+      "loss": 0.4689,
+      "mean_token_accuracy": 0.8697376251220703,
+      "num_tokens": 3416633.0,
+      "step": 402
+    },
+    {
+      "epoch": 0.9253731343283582,
+      "grad_norm": 2.2834797845830983,
+      "learning_rate": 8.873195291621505e-06,
+      "loss": 0.4537,
+      "mean_token_accuracy": 0.874300092458725,
+      "num_tokens": 3425193.0,
+      "step": 403
+    },
+    {
+      "epoch": 0.9276693455797933,
+      "grad_norm": 2.4018508328037536,
+      "learning_rate": 8.865233173086924e-06,
+      "loss": 0.4876,
+      "mean_token_accuracy": 0.8717392385005951,
+      "num_tokens": 3433301.0,
+      "step": 404
+    },
+    {
+      "epoch": 0.9299655568312285,
+      "grad_norm": 2.1535035916803973,
+      "learning_rate": 8.8572470793808e-06,
+      "loss": 0.4694,
+      "mean_token_accuracy": 0.8728501200675964,
+      "num_tokens": 3441889.0,
+      "step": 405
+    },
+    {
+      "epoch": 0.9322617680826636,
+      "grad_norm": 2.370300909735562,
+      "learning_rate": 8.849237067399033e-06,
+      "loss": 0.4978,
+      "mean_token_accuracy": 0.8638056814670563,
+      "num_tokens": 3449847.0,
+      "step": 406
+    },
+    {
+      "epoch": 0.9345579793340988,
+      "grad_norm": 2.1661421290280036,
+      "learning_rate": 8.841203194207925e-06,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.862493485212326,
+      "num_tokens": 3459080.0,
+      "step": 407
+    },
+    {
+      "epoch": 0.9368541905855339,
+      "grad_norm": 2.063163931217335,
+      "learning_rate": 8.833145517043773e-06,
+      "loss": 0.4149,
+      "mean_token_accuracy": 0.8839117288589478,
+      "num_tokens": 3466920.0,
+      "step": 408
+    },
+    {
+      "epoch": 0.939150401836969,
+      "grad_norm": 2.2263524441646467,
+      "learning_rate": 8.825064093312464e-06,
+      "loss": 0.4419,
+      "mean_token_accuracy": 0.8776047825813293,
+      "num_tokens": 3475294.0,
+      "step": 409
+    },
+    {
+      "epoch": 0.9414466130884042,
+      "grad_norm": 2.09009356371105,
+      "learning_rate": 8.816958980589063e-06,
+      "loss": 0.4226,
+      "mean_token_accuracy": 0.882329910993576,
+      "num_tokens": 3483641.0,
+      "step": 410
+    },
+    {
+      "epoch": 0.9437428243398392,
+      "grad_norm": 2.1947542455826907,
+      "learning_rate": 8.808830236617403e-06,
+      "loss": 0.4182,
+      "mean_token_accuracy": 0.8824563026428223,
+      "num_tokens": 3491389.0,
+      "step": 411
+    },
+    {
+      "epoch": 0.9460390355912744,
+      "grad_norm": 2.2450587114048575,
+      "learning_rate": 8.80067791930968e-06,
+      "loss": 0.4705,
+      "mean_token_accuracy": 0.8719064295291901,
+      "num_tokens": 3499996.0,
+      "step": 412
+    },
+    {
+      "epoch": 0.9483352468427095,
+      "grad_norm": 2.164069337475745,
+      "learning_rate": 8.79250208674603e-06,
+      "loss": 0.4009,
+      "mean_token_accuracy": 0.8835300207138062,
+      "num_tokens": 3508543.0,
+      "step": 413
+    },
+    {
+      "epoch": 0.9506314580941446,
+      "grad_norm": 2.19565518351803,
+      "learning_rate": 8.78430279717412e-06,
+      "loss": 0.4125,
+      "mean_token_accuracy": 0.8819408416748047,
+      "num_tokens": 3516679.0,
+      "step": 414
+    },
+    {
+      "epoch": 0.9529276693455798,
+      "grad_norm": 2.6404826225322613,
+      "learning_rate": 8.776080109008738e-06,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.8602061569690704,
+      "num_tokens": 3524891.0,
+      "step": 415
+    },
+    {
+      "epoch": 0.9552238805970149,
+      "grad_norm": 2.343586330912915,
+      "learning_rate": 8.76783408083137e-06,
+      "loss": 0.4625,
+      "mean_token_accuracy": 0.8713793456554413,
+      "num_tokens": 3533408.0,
+      "step": 416
+    },
+    {
+      "epoch": 0.9575200918484501,
+      "grad_norm": 2.2026102752802315,
+      "learning_rate": 8.759564771389784e-06,
+      "loss": 0.4472,
+      "mean_token_accuracy": 0.877887636423111,
+      "num_tokens": 3541865.0,
+      "step": 417
+    },
+    {
+      "epoch": 0.9598163030998852,
+      "grad_norm": 2.141056676509237,
+      "learning_rate": 8.751272239597612e-06,
+      "loss": 0.4369,
+      "mean_token_accuracy": 0.8801349401473999,
+      "num_tokens": 3550759.0,
+      "step": 418
+    },
+    {
+      "epoch": 0.9621125143513203,
+      "grad_norm": 2.1078714008541595,
+      "learning_rate": 8.742956544533931e-06,
+      "loss": 0.4292,
+      "mean_token_accuracy": 0.8815636038780212,
+      "num_tokens": 3559216.0,
+      "step": 419
+    },
+    {
+      "epoch": 0.9644087256027555,
+      "grad_norm": 2.2258651022680382,
+      "learning_rate": 8.734617745442845e-06,
+      "loss": 0.4366,
+      "mean_token_accuracy": 0.8792649209499359,
+      "num_tokens": 3567243.0,
+      "step": 420
+    },
+    {
+      "epoch": 0.9667049368541906,
+      "grad_norm": 2.1993571925087156,
+      "learning_rate": 8.726255901733058e-06,
+      "loss": 0.4695,
+      "mean_token_accuracy": 0.8719243407249451,
+      "num_tokens": 3575851.0,
+      "step": 421
+    },
+    {
+      "epoch": 0.9690011481056258,
+      "grad_norm": 2.1658299946417015,
+      "learning_rate": 8.717871072977449e-06,
+      "loss": 0.4368,
+      "mean_token_accuracy": 0.8799764215946198,
+      "num_tokens": 3583665.0,
+      "step": 422
+    },
+    {
+      "epoch": 0.9712973593570609,
+      "grad_norm": 2.173419058248531,
+      "learning_rate": 8.709463318912654e-06,
+      "loss": 0.4343,
+      "mean_token_accuracy": 0.8781344592571259,
+      "num_tokens": 3592166.0,
+      "step": 423
+    },
+    {
+      "epoch": 0.9735935706084959,
+      "grad_norm": 2.0259434304690775,
+      "learning_rate": 8.701032699438638e-06,
+      "loss": 0.4132,
+      "mean_token_accuracy": 0.8822500705718994,
+      "num_tokens": 3600795.0,
+      "step": 424
+    },
+    {
+      "epoch": 0.9758897818599311,
+      "grad_norm": 2.109015516478586,
+      "learning_rate": 8.692579274618268e-06,
+      "loss": 0.4144,
+      "mean_token_accuracy": 0.8797238171100616,
+      "num_tokens": 3609598.0,
+      "step": 425
+    },
+    {
+      "epoch": 0.9781859931113662,
+      "grad_norm": 2.251841052852613,
+      "learning_rate": 8.684103104676879e-06,
+      "loss": 0.407,
+      "mean_token_accuracy": 0.8866288661956787,
+      "num_tokens": 3617713.0,
+      "step": 426
+    },
+    {
+      "epoch": 0.9804822043628014,
+      "grad_norm": 2.3771517831573874,
+      "learning_rate": 8.675604250001857e-06,
+      "loss": 0.4567,
+      "mean_token_accuracy": 0.8709289133548737,
+      "num_tokens": 3627253.0,
+      "step": 427
+    },
+    {
+      "epoch": 0.9827784156142365,
+      "grad_norm": 2.6150547704751608,
+      "learning_rate": 8.667082771142203e-06,
+      "loss": 0.4364,
+      "mean_token_accuracy": 0.8812183439731598,
+      "num_tokens": 3635338.0,
+      "step": 428
+    },
+    {
+      "epoch": 0.9850746268656716,
+      "grad_norm": 2.5837518917935447,
+      "learning_rate": 8.658538728808097e-06,
+      "loss": 0.4553,
+      "mean_token_accuracy": 0.8745913803577423,
+      "num_tokens": 3643521.0,
+      "step": 429
+    },
+    {
+      "epoch": 0.9873708381171068,
+      "grad_norm": 2.4953559337033044,
+      "learning_rate": 8.649972183870472e-06,
+      "loss": 0.4565,
+      "mean_token_accuracy": 0.8725022077560425,
+      "num_tokens": 3652174.0,
+      "step": 430
+    },
+    {
+      "epoch": 0.9896670493685419,
+      "grad_norm": 2.151619938313446,
+      "learning_rate": 8.641383197360576e-06,
+      "loss": 0.4426,
+      "mean_token_accuracy": 0.878854513168335,
+      "num_tokens": 3661013.0,
+      "step": 431
+    },
+    {
+      "epoch": 0.9919632606199771,
+      "grad_norm": 2.381313949472344,
+      "learning_rate": 8.63277183046954e-06,
+      "loss": 0.491,
+      "mean_token_accuracy": 0.8669999241828918,
+      "num_tokens": 3669179.0,
+      "step": 432
+    },
+    {
+      "epoch": 0.9942594718714122,
+      "grad_norm": 2.217168623985446,
+      "learning_rate": 8.624138144547944e-06,
+      "loss": 0.4654,
+      "mean_token_accuracy": 0.8714777529239655,
+      "num_tokens": 3677501.0,
+      "step": 433
+    },
+    {
+      "epoch": 0.9965556831228473,
+      "grad_norm": 2.08077306425991,
+      "learning_rate": 8.615482201105373e-06,
+      "loss": 0.4314,
+      "mean_token_accuracy": 0.8810337781906128,
+      "num_tokens": 3686128.0,
+      "step": 434
+    },
+    {
+      "epoch": 0.9988518943742825,
+      "grad_norm": 2.310378192567032,
+      "learning_rate": 8.606804061809983e-06,
+      "loss": 0.4142,
+      "mean_token_accuracy": 0.8862297832965851,
+      "num_tokens": 3694270.0,
+      "step": 435
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.310378192567032,
+      "learning_rate": 8.598103788488064e-06,
+      "loss": 0.3905,
+      "mean_token_accuracy": 0.8739205598831177,
+      "num_tokens": 3694851.0,
+      "step": 436
+    },
+    {
+      "epoch": 1.002296211251435,
+      "grad_norm": 3.299265274245272,
+      "learning_rate": 8.589381443123597e-06,
+      "loss": 0.2908,
+      "mean_token_accuracy": 0.9172686636447906,
+      "num_tokens": 3704227.0,
+      "step": 437
+    },
+    {
+      "epoch": 1.0045924225028702,
+      "grad_norm": 1.7619661557984172,
+      "learning_rate": 8.58063708785781e-06,
+      "loss": 0.3068,
+      "mean_token_accuracy": 0.9116365909576416,
+      "num_tokens": 3713214.0,
+      "step": 438
+    },
+    {
+      "epoch": 1.0068886337543055,
+      "grad_norm": 1.8940175654015674,
+      "learning_rate": 8.571870784988739e-06,
+      "loss": 0.3213,
+      "mean_token_accuracy": 0.9102482199668884,
+      "num_tokens": 3721482.0,
+      "step": 439
+    },
+    {
+      "epoch": 1.0091848450057406,
+      "grad_norm": 2.096847833739993,
+      "learning_rate": 8.563082596970785e-06,
+      "loss": 0.2647,
+      "mean_token_accuracy": 0.9251429736614227,
+      "num_tokens": 3729929.0,
+      "step": 440
+    },
+    {
+      "epoch": 1.0114810562571757,
+      "grad_norm": 2.067677915190213,
+      "learning_rate": 8.554272586414259e-06,
+      "loss": 0.2609,
+      "mean_token_accuracy": 0.9298540949821472,
+      "num_tokens": 3738798.0,
+      "step": 441
+    },
+    {
+      "epoch": 1.0137772675086107,
+      "grad_norm": 3.519336492234775,
+      "learning_rate": 8.545440816084958e-06,
+      "loss": 0.2812,
+      "mean_token_accuracy": 0.9233300387859344,
+      "num_tokens": 3747583.0,
+      "step": 442
+    },
+    {
+      "epoch": 1.0160734787600458,
+      "grad_norm": 2.6798443828465826,
+      "learning_rate": 8.53658734890369e-06,
+      "loss": 0.2534,
+      "mean_token_accuracy": 0.9237805604934692,
+      "num_tokens": 3755736.0,
+      "step": 443
+    },
+    {
+      "epoch": 1.0183696900114811,
+      "grad_norm": 3.4748204708522366,
+      "learning_rate": 8.527712247945845e-06,
+      "loss": 0.3004,
+      "mean_token_accuracy": 0.9173048436641693,
+      "num_tokens": 3763925.0,
+      "step": 444
+    },
+    {
+      "epoch": 1.0206659012629162,
+      "grad_norm": 2.8829549758186115,
+      "learning_rate": 8.518815576440942e-06,
+      "loss": 0.2766,
+      "mean_token_accuracy": 0.9224980771541595,
+      "num_tokens": 3772458.0,
+      "step": 445
+    },
+    {
+      "epoch": 1.0229621125143513,
+      "grad_norm": 2.378888144472056,
+      "learning_rate": 8.509897397772174e-06,
+      "loss": 0.3092,
+      "mean_token_accuracy": 0.9134126007556915,
+      "num_tokens": 3780331.0,
+      "step": 446
+    },
+    {
+      "epoch": 1.0252583237657864,
+      "grad_norm": 2.222142359521808,
+      "learning_rate": 8.500957775475959e-06,
+      "loss": 0.3096,
+      "mean_token_accuracy": 0.9117675423622131,
+      "num_tokens": 3788989.0,
+      "step": 447
+    },
+    {
+      "epoch": 1.0275545350172215,
+      "grad_norm": 2.371632009666441,
+      "learning_rate": 8.491996773241488e-06,
+      "loss": 0.2791,
+      "mean_token_accuracy": 0.92186439037323,
+      "num_tokens": 3797293.0,
+      "step": 448
+    },
+    {
+      "epoch": 1.0298507462686568,
+      "grad_norm": 1.9642381666805062,
+      "learning_rate": 8.48301445491027e-06,
+      "loss": 0.2832,
+      "mean_token_accuracy": 0.9198655784130096,
+      "num_tokens": 3806013.0,
+      "step": 449
+    },
+    {
+      "epoch": 1.032146957520092,
+      "grad_norm": 1.9988735036016756,
+      "learning_rate": 8.474010884475682e-06,
+      "loss": 0.2921,
+      "mean_token_accuracy": 0.9186451137065887,
+      "num_tokens": 3814567.0,
+      "step": 450
+    },
+    {
+      "epoch": 1.034443168771527,
+      "grad_norm": 2.2942971786899036,
+      "learning_rate": 8.4649861260825e-06,
+      "loss": 0.2759,
+      "mean_token_accuracy": 0.9203961193561554,
+      "num_tokens": 3823491.0,
+      "step": 451
+    },
+    {
+      "epoch": 1.036739380022962,
+      "grad_norm": 1.9132993217507035,
+      "learning_rate": 8.45594024402646e-06,
+      "loss": 0.261,
+      "mean_token_accuracy": 0.9204106330871582,
+      "num_tokens": 3832634.0,
+      "step": 452
+    },
+    {
+      "epoch": 1.0390355912743972,
+      "grad_norm": 2.193075277867522,
+      "learning_rate": 8.446873302753783e-06,
+      "loss": 0.2847,
+      "mean_token_accuracy": 0.9202054142951965,
+      "num_tokens": 3840786.0,
+      "step": 453
+    },
+    {
+      "epoch": 1.0413318025258325,
+      "grad_norm": 2.064123984021009,
+      "learning_rate": 8.437785366860732e-06,
+      "loss": 0.2371,
+      "mean_token_accuracy": 0.9336122870445251,
+      "num_tokens": 3849096.0,
+      "step": 454
+    },
+    {
+      "epoch": 1.0436280137772675,
+      "grad_norm": 2.182973989376289,
+      "learning_rate": 8.428676501093135e-06,
+      "loss": 0.307,
+      "mean_token_accuracy": 0.9113952219486237,
+      "num_tokens": 3857935.0,
+      "step": 455
+    },
+    {
+      "epoch": 1.0459242250287026,
+      "grad_norm": 2.6441184548936656,
+      "learning_rate": 8.41954677034594e-06,
+      "loss": 0.3057,
+      "mean_token_accuracy": 0.9112354516983032,
+      "num_tokens": 3866436.0,
+      "step": 456
+    },
+    {
+      "epoch": 1.0482204362801377,
+      "grad_norm": 2.3684962920183676,
+      "learning_rate": 8.410396239662738e-06,
+      "loss": 0.2541,
+      "mean_token_accuracy": 0.9279707968235016,
+      "num_tokens": 3874571.0,
+      "step": 457
+    },
+    {
+      "epoch": 1.0505166475315728,
+      "grad_norm": 2.204714903244235,
+      "learning_rate": 8.401224974235313e-06,
+      "loss": 0.284,
+      "mean_token_accuracy": 0.9176671802997589,
+      "num_tokens": 3883251.0,
+      "step": 458
+    },
+    {
+      "epoch": 1.0528128587830081,
+      "grad_norm": 2.2121546466225595,
+      "learning_rate": 8.392033039403165e-06,
+      "loss": 0.2486,
+      "mean_token_accuracy": 0.9281324446201324,
+      "num_tokens": 3891960.0,
+      "step": 459
+    },
+    {
+      "epoch": 1.0551090700344432,
+      "grad_norm": 2.367344351524279,
+      "learning_rate": 8.382820500653055e-06,
+      "loss": 0.2905,
+      "mean_token_accuracy": 0.9154870808124542,
+      "num_tokens": 3901097.0,
+      "step": 460
+    },
+    {
+      "epoch": 1.0574052812858783,
+      "grad_norm": 2.062088931588969,
+      "learning_rate": 8.37358742361853e-06,
+      "loss": 0.2832,
+      "mean_token_accuracy": 0.9205671548843384,
+      "num_tokens": 3909514.0,
+      "step": 461
+    },
+    {
+      "epoch": 1.0597014925373134,
+      "grad_norm": 2.3356107061511335,
+      "learning_rate": 8.364333874079462e-06,
+      "loss": 0.2903,
+      "mean_token_accuracy": 0.9193686246871948,
+      "num_tokens": 3918243.0,
+      "step": 462
+    },
+    {
+      "epoch": 1.0619977037887485,
+      "grad_norm": 2.0182861514995776,
+      "learning_rate": 8.355059917961579e-06,
+      "loss": 0.2632,
+      "mean_token_accuracy": 0.9210003614425659,
+      "num_tokens": 3926932.0,
+      "step": 463
+    },
+    {
+      "epoch": 1.0642939150401838,
+      "grad_norm": 2.224080994795777,
+      "learning_rate": 8.345765621335987e-06,
+      "loss": 0.258,
+      "mean_token_accuracy": 0.9286116361618042,
+      "num_tokens": 3935440.0,
+      "step": 464
+    },
+    {
+      "epoch": 1.0665901262916189,
+      "grad_norm": 2.164135507057411,
+      "learning_rate": 8.33645105041871e-06,
+      "loss": 0.2734,
+      "mean_token_accuracy": 0.919291764497757,
+      "num_tokens": 3943868.0,
+      "step": 465
+    },
+    {
+      "epoch": 1.068886337543054,
+      "grad_norm": 2.4144950323392798,
+      "learning_rate": 8.327116271570212e-06,
+      "loss": 0.2621,
+      "mean_token_accuracy": 0.9243497848510742,
+      "num_tokens": 3952038.0,
+      "step": 466
+    },
+    {
+      "epoch": 1.071182548794489,
+      "grad_norm": 2.18938033610901,
+      "learning_rate": 8.317761351294924e-06,
+      "loss": 0.2757,
+      "mean_token_accuracy": 0.9234868884086609,
+      "num_tokens": 3960952.0,
+      "step": 467
+    },
+    {
+      "epoch": 1.0734787600459241,
+      "grad_norm": 2.3273577147819653,
+      "learning_rate": 8.30838635624078e-06,
+      "loss": 0.3044,
+      "mean_token_accuracy": 0.9116412699222565,
+      "num_tokens": 3968853.0,
+      "step": 468
+    },
+    {
+      "epoch": 1.0757749712973594,
+      "grad_norm": 2.4381898384265996,
+      "learning_rate": 8.298991353198722e-06,
+      "loss": 0.291,
+      "mean_token_accuracy": 0.9209294319152832,
+      "num_tokens": 3977726.0,
+      "step": 469
+    },
+    {
+      "epoch": 1.0780711825487945,
+      "grad_norm": 2.1387295846356182,
+      "learning_rate": 8.289576409102247e-06,
+      "loss": 0.2756,
+      "mean_token_accuracy": 0.9207946956157684,
+      "num_tokens": 3985853.0,
+      "step": 470
+    },
+    {
+      "epoch": 1.0803673938002296,
+      "grad_norm": 2.135059592857589,
+      "learning_rate": 8.280141591026911e-06,
+      "loss": 0.2395,
+      "mean_token_accuracy": 0.9277163445949554,
+      "num_tokens": 3994449.0,
+      "step": 471
+    },
+    {
+      "epoch": 1.0826636050516647,
+      "grad_norm": 2.2982414155760313,
+      "learning_rate": 8.270686966189866e-06,
+      "loss": 0.3543,
+      "mean_token_accuracy": 0.9008886218070984,
+      "num_tokens": 4003576.0,
+      "step": 472
+    },
+    {
+      "epoch": 1.0849598163030998,
+      "grad_norm": 2.5996058851819237,
+      "learning_rate": 8.261212601949374e-06,
+      "loss": 0.2796,
+      "mean_token_accuracy": 0.9217697381973267,
+      "num_tokens": 4012124.0,
+      "step": 473
+    },
+    {
+      "epoch": 1.087256027554535,
+      "grad_norm": 2.2083233552664185,
+      "learning_rate": 8.251718565804323e-06,
+      "loss": 0.2564,
+      "mean_token_accuracy": 0.92575803399086,
+      "num_tokens": 4020274.0,
+      "step": 474
+    },
+    {
+      "epoch": 1.0895522388059702,
+      "grad_norm": 2.2434262217633734,
+      "learning_rate": 8.242204925393754e-06,
+      "loss": 0.2654,
+      "mean_token_accuracy": 0.92165607213974,
+      "num_tokens": 4028969.0,
+      "step": 475
+    },
+    {
+      "epoch": 1.0918484500574053,
+      "grad_norm": 2.335912678513662,
+      "learning_rate": 8.232671748496379e-06,
+      "loss": 0.2771,
+      "mean_token_accuracy": 0.9201677739620209,
+      "num_tokens": 4037354.0,
+      "step": 476
+    },
+    {
+      "epoch": 1.0941446613088404,
+      "grad_norm": 2.3097257641985602,
+      "learning_rate": 8.223119103030092e-06,
+      "loss": 0.2746,
+      "mean_token_accuracy": 0.924422025680542,
+      "num_tokens": 4046154.0,
+      "step": 477
+    },
+    {
+      "epoch": 1.0964408725602754,
+      "grad_norm": 2.499713389274134,
+      "learning_rate": 8.213547057051484e-06,
+      "loss": 0.2834,
+      "mean_token_accuracy": 0.916504979133606,
+      "num_tokens": 4053803.0,
+      "step": 478
+    },
+    {
+      "epoch": 1.0987370838117108,
+      "grad_norm": 2.5635742576249,
+      "learning_rate": 8.203955678755376e-06,
+      "loss": 0.2808,
+      "mean_token_accuracy": 0.9188525974750519,
+      "num_tokens": 4062022.0,
+      "step": 479
+    },
+    {
+      "epoch": 1.1010332950631458,
+      "grad_norm": 2.2273839420452957,
+      "learning_rate": 8.194345036474305e-06,
+      "loss": 0.2634,
+      "mean_token_accuracy": 0.9245164692401886,
+      "num_tokens": 4070151.0,
+      "step": 480
+    },
+    {
+      "epoch": 1.103329506314581,
+      "grad_norm": 3.130170886945181,
+      "learning_rate": 8.184715198678057e-06,
+      "loss": 0.3195,
+      "mean_token_accuracy": 0.9131514728069305,
+      "num_tokens": 4079946.0,
+      "step": 481
+    },
+    {
+      "epoch": 1.105625717566016,
+      "grad_norm": 2.0802232488169894,
+      "learning_rate": 8.175066233973176e-06,
+      "loss": 0.251,
+      "mean_token_accuracy": 0.9243484139442444,
+      "num_tokens": 4088662.0,
+      "step": 482
+    },
+    {
+      "epoch": 1.107921928817451,
+      "grad_norm": 2.2169816244777114,
+      "learning_rate": 8.165398211102475e-06,
+      "loss": 0.2906,
+      "mean_token_accuracy": 0.9196761548519135,
+      "num_tokens": 4096830.0,
+      "step": 483
+    },
+    {
+      "epoch": 1.1102181400688864,
+      "grad_norm": 2.076482379674457,
+      "learning_rate": 8.155711198944536e-06,
+      "loss": 0.2601,
+      "mean_token_accuracy": 0.9225354194641113,
+      "num_tokens": 4106054.0,
+      "step": 484
+    },
+    {
+      "epoch": 1.1125143513203215,
+      "grad_norm": 2.2464173286250584,
+      "learning_rate": 8.146005266513237e-06,
+      "loss": 0.2814,
+      "mean_token_accuracy": 0.9188204109668732,
+      "num_tokens": 4114544.0,
+      "step": 485
+    },
+    {
+      "epoch": 1.1148105625717566,
+      "grad_norm": 2.3326269495320933,
+      "learning_rate": 8.136280482957243e-06,
+      "loss": 0.298,
+      "mean_token_accuracy": 0.9170420169830322,
+      "num_tokens": 4122853.0,
+      "step": 486
+    },
+    {
+      "epoch": 1.1171067738231917,
+      "grad_norm": 2.16212814303235,
+      "learning_rate": 8.126536917559529e-06,
+      "loss": 0.2541,
+      "mean_token_accuracy": 0.9287247061729431,
+      "num_tokens": 4130850.0,
+      "step": 487
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 2.264095578758906,
+      "learning_rate": 8.116774639736875e-06,
+      "loss": 0.2721,
+      "mean_token_accuracy": 0.9187359511852264,
+      "num_tokens": 4139564.0,
+      "step": 488
+    },
+    {
+      "epoch": 1.121699196326062,
+      "grad_norm": 2.2169444861630656,
+      "learning_rate": 8.106993719039374e-06,
+      "loss": 0.2609,
+      "mean_token_accuracy": 0.9231422245502472,
+      "num_tokens": 4148997.0,
+      "step": 489
+    },
+    {
+      "epoch": 1.1239954075774972,
+      "grad_norm": 2.6624306100805266,
+      "learning_rate": 8.09719422514994e-06,
+      "loss": 0.3027,
+      "mean_token_accuracy": 0.917034387588501,
+      "num_tokens": 4157672.0,
+      "step": 490
+    },
+    {
+      "epoch": 1.1262916188289323,
+      "grad_norm": 2.1602512668100857,
+      "learning_rate": 8.08737622788381e-06,
+      "loss": 0.2621,
+      "mean_token_accuracy": 0.9261443316936493,
+      "num_tokens": 4166385.0,
+      "step": 491
+    },
+    {
+      "epoch": 1.1285878300803673,
+      "grad_norm": 2.080583286307093,
+      "learning_rate": 8.077539797188042e-06,
+      "loss": 0.2866,
+      "mean_token_accuracy": 0.9170297086238861,
+      "num_tokens": 4174953.0,
+      "step": 492
+    },
+    {
+      "epoch": 1.1308840413318024,
+      "grad_norm": 2.5057095642268377,
+      "learning_rate": 8.067685003141025e-06,
+      "loss": 0.2868,
+      "mean_token_accuracy": 0.9153377413749695,
+      "num_tokens": 4183316.0,
+      "step": 493
+    },
+    {
+      "epoch": 1.1331802525832377,
+      "grad_norm": 2.2462956216043635,
+      "learning_rate": 8.057811915951972e-06,
+      "loss": 0.2337,
+      "mean_token_accuracy": 0.9339278936386108,
+      "num_tokens": 4191479.0,
+      "step": 494
+    },
+    {
+      "epoch": 1.1354764638346728,
+      "grad_norm": 2.120595984929579,
+      "learning_rate": 8.047920605960428e-06,
+      "loss": 0.2728,
+      "mean_token_accuracy": 0.9227598011493683,
+      "num_tokens": 4199254.0,
+      "step": 495
+    },
+    {
+      "epoch": 1.137772675086108,
+      "grad_norm": 2.5331130049681105,
+      "learning_rate": 8.038011143635756e-06,
+      "loss": 0.2776,
+      "mean_token_accuracy": 0.9207497239112854,
+      "num_tokens": 4208211.0,
+      "step": 496
+    },
+    {
+      "epoch": 1.140068886337543,
+      "grad_norm": 2.1680878335066742,
+      "learning_rate": 8.028083599576653e-06,
+      "loss": 0.2768,
+      "mean_token_accuracy": 0.920197993516922,
+      "num_tokens": 4216774.0,
+      "step": 497
+    },
+    {
+      "epoch": 1.142365097588978,
+      "grad_norm": 2.336792684906978,
+      "learning_rate": 8.018138044510625e-06,
+      "loss": 0.2722,
+      "mean_token_accuracy": 0.9219779968261719,
+      "num_tokens": 4225519.0,
+      "step": 498
+    },
+    {
+      "epoch": 1.1446613088404134,
+      "grad_norm": 2.1310173764330944,
+      "learning_rate": 8.008174549293509e-06,
+      "loss": 0.2751,
+      "mean_token_accuracy": 0.9219463169574738,
+      "num_tokens": 4233585.0,
+      "step": 499
+    },
+    {
+      "epoch": 1.1469575200918485,
+      "grad_norm": 2.395228805651906,
+      "learning_rate": 7.99819318490894e-06,
+      "loss": 0.262,
+      "mean_token_accuracy": 0.9220178425312042,
+      "num_tokens": 4242430.0,
+      "step": 500
+    },
+    {
+      "epoch": 1.1492537313432836,
+      "grad_norm": 2.1059788063077454,
+      "learning_rate": 7.988194022467874e-06,
+      "loss": 0.2728,
+      "mean_token_accuracy": 0.9255200028419495,
+      "num_tokens": 4250233.0,
+      "step": 501
+    },
+    {
+      "epoch": 1.1515499425947187,
+      "grad_norm": 2.3558172546528393,
+      "learning_rate": 7.978177133208053e-06,
+      "loss": 0.2769,
+      "mean_token_accuracy": 0.9202165007591248,
+      "num_tokens": 4258682.0,
+      "step": 502
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 2.2954117429997285,
+      "learning_rate": 7.968142588493522e-06,
+      "loss": 0.2669,
+      "mean_token_accuracy": 0.9231474995613098,
+      "num_tokens": 4267306.0,
+      "step": 503
+    },
+    {
+      "epoch": 1.156142365097589,
+      "grad_norm": 2.1205787236714686,
+      "learning_rate": 7.958090459814106e-06,
+      "loss": 0.2506,
+      "mean_token_accuracy": 0.9309555888175964,
+      "num_tokens": 4276434.0,
+      "step": 504
+    },
+    {
+      "epoch": 1.1584385763490241,
+      "grad_norm": 2.192468735638827,
+      "learning_rate": 7.948020818784902e-06,
+      "loss": 0.2667,
+      "mean_token_accuracy": 0.9251600503921509,
+      "num_tokens": 4284504.0,
+      "step": 505
+    },
+    {
+      "epoch": 1.1607347876004592,
+      "grad_norm": 2.1950002715270243,
+      "learning_rate": 7.937933737145777e-06,
+      "loss": 0.2753,
+      "mean_token_accuracy": 0.9212541878223419,
+      "num_tokens": 4293215.0,
+      "step": 506
+    },
+    {
+      "epoch": 1.1630309988518943,
+      "grad_norm": 2.2521953356611975,
+      "learning_rate": 7.927829286760846e-06,
+      "loss": 0.2669,
+      "mean_token_accuracy": 0.9232867658138275,
+      "num_tokens": 4301965.0,
+      "step": 507
+    },
+    {
+      "epoch": 1.1653272101033294,
+      "grad_norm": 2.30294380668812,
+      "learning_rate": 7.917707539617971e-06,
+      "loss": 0.2795,
+      "mean_token_accuracy": 0.9238947033882141,
+      "num_tokens": 4311056.0,
+      "step": 508
+    },
+    {
+      "epoch": 1.1676234213547647,
+      "grad_norm": 2.1098743692827666,
+      "learning_rate": 7.90756856782824e-06,
+      "loss": 0.2546,
+      "mean_token_accuracy": 0.9265170097351074,
+      "num_tokens": 4319612.0,
+      "step": 509
+    },
+    {
+      "epoch": 1.1699196326061998,
+      "grad_norm": 2.358042933780049,
+      "learning_rate": 7.89741244362545e-06,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.9085181355476379,
+      "num_tokens": 4328320.0,
+      "step": 510
+    },
+    {
+      "epoch": 1.1722158438576349,
+      "grad_norm": 2.532585626018824,
+      "learning_rate": 7.887239239365608e-06,
+      "loss": 0.284,
+      "mean_token_accuracy": 0.9212026596069336,
+      "num_tokens": 4336569.0,
+      "step": 511
+    },
+    {
+      "epoch": 1.17451205510907,
+      "grad_norm": 2.289511402503256,
+      "learning_rate": 7.8770490275264e-06,
+      "loss": 0.284,
+      "mean_token_accuracy": 0.9204729199409485,
+      "num_tokens": 4344882.0,
+      "step": 512
+    },
+    {
+      "epoch": 1.176808266360505,
+      "grad_norm": 2.3438682187657793,
+      "learning_rate": 7.866841880706682e-06,
+      "loss": 0.2674,
+      "mean_token_accuracy": 0.9246805906295776,
+      "num_tokens": 4353511.0,
+      "step": 513
+    },
+    {
+      "epoch": 1.1791044776119404,
+      "grad_norm": 2.3639472779533013,
+      "learning_rate": 7.856617871625963e-06,
+      "loss": 0.3203,
+      "mean_token_accuracy": 0.9086554050445557,
+      "num_tokens": 4361820.0,
+      "step": 514
+    },
+    {
+      "epoch": 1.1814006888633755,
+      "grad_norm": 2.2058901081811046,
+      "learning_rate": 7.846377073123877e-06,
+      "loss": 0.2379,
+      "mean_token_accuracy": 0.9300184547901154,
+      "num_tokens": 4369915.0,
+      "step": 515
+    },
+    {
+      "epoch": 1.1836969001148105,
+      "grad_norm": 2.2435536836459655,
+      "learning_rate": 7.836119558159682e-06,
+      "loss": 0.2754,
+      "mean_token_accuracy": 0.9213479459285736,
+      "num_tokens": 4378745.0,
+      "step": 516
+    },
+    {
+      "epoch": 1.1859931113662456,
+      "grad_norm": 2.370568949224425,
+      "learning_rate": 7.825845399811723e-06,
+      "loss": 0.2846,
+      "mean_token_accuracy": 0.9194954633712769,
+      "num_tokens": 4387114.0,
+      "step": 517
+    },
+    {
+      "epoch": 1.1882893226176807,
+      "grad_norm": 2.1464456852127025,
+      "learning_rate": 7.815554671276925e-06,
+      "loss": 0.248,
+      "mean_token_accuracy": 0.9261257648468018,
+      "num_tokens": 4395200.0,
+      "step": 518
+    },
+    {
+      "epoch": 1.190585533869116,
+      "grad_norm": 2.753657762219131,
+      "learning_rate": 7.80524744587026e-06,
+      "loss": 0.2677,
+      "mean_token_accuracy": 0.919915109872818,
+      "num_tokens": 4403809.0,
+      "step": 519
+    },
+    {
+      "epoch": 1.1928817451205511,
+      "grad_norm": 2.260384357482908,
+      "learning_rate": 7.794923797024232e-06,
+      "loss": 0.307,
+      "mean_token_accuracy": 0.9101102650165558,
+      "num_tokens": 4412313.0,
+      "step": 520
+    },
+    {
+      "epoch": 1.1951779563719862,
+      "grad_norm": 2.4218535140181965,
+      "learning_rate": 7.784583798288351e-06,
+      "loss": 0.2446,
+      "mean_token_accuracy": 0.9322145283222198,
+      "num_tokens": 4420752.0,
+      "step": 521
+    },
+    {
+      "epoch": 1.1974741676234213,
+      "grad_norm": 2.2735227067507062,
+      "learning_rate": 7.77422752332861e-06,
+      "loss": 0.279,
+      "mean_token_accuracy": 0.9220273196697235,
+      "num_tokens": 4428736.0,
+      "step": 522
+    },
+    {
+      "epoch": 1.1997703788748564,
+      "grad_norm": 2.3522542636226196,
+      "learning_rate": 7.76385504592696e-06,
+      "loss": 0.2566,
+      "mean_token_accuracy": 0.92378830909729,
+      "num_tokens": 4437281.0,
+      "step": 523
+    },
+    {
+      "epoch": 1.2020665901262917,
+      "grad_norm": 2.3557840615745818,
+      "learning_rate": 7.753466439980783e-06,
+      "loss": 0.2335,
+      "mean_token_accuracy": 0.9313016831874847,
+      "num_tokens": 4446227.0,
+      "step": 524
+    },
+    {
+      "epoch": 1.2043628013777268,
+      "grad_norm": 2.2754634166610934,
+      "learning_rate": 7.743061779502372e-06,
+      "loss": 0.2911,
+      "mean_token_accuracy": 0.9158861637115479,
+      "num_tokens": 4454699.0,
+      "step": 525
+    },
+    {
+      "epoch": 1.2066590126291619,
+      "grad_norm": 2.4726851812659913,
+      "learning_rate": 7.732641138618389e-06,
+      "loss": 0.2818,
+      "mean_token_accuracy": 0.9194687902927399,
+      "num_tokens": 4463021.0,
+      "step": 526
+    },
+    {
+      "epoch": 1.208955223880597,
+      "grad_norm": 2.5153382826137936,
+      "learning_rate": 7.722204591569353e-06,
+      "loss": 0.2601,
+      "mean_token_accuracy": 0.9225470125675201,
+      "num_tokens": 4472139.0,
+      "step": 527
+    },
+    {
+      "epoch": 1.211251435132032,
+      "grad_norm": 2.2235247624129197,
+      "learning_rate": 7.711752212709106e-06,
+      "loss": 0.2605,
+      "mean_token_accuracy": 0.9296205639839172,
+      "num_tokens": 4480268.0,
+      "step": 528
+    },
+    {
+      "epoch": 1.2135476463834673,
+      "grad_norm": 2.2895448657478914,
+      "learning_rate": 7.701284076504275e-06,
+      "loss": 0.2638,
+      "mean_token_accuracy": 0.9263229966163635,
+      "num_tokens": 4488846.0,
+      "step": 529
+    },
+    {
+      "epoch": 1.2158438576349024,
+      "grad_norm": 2.242861281977577,
+      "learning_rate": 7.690800257533755e-06,
+      "loss": 0.2535,
+      "mean_token_accuracy": 0.9259541630744934,
+      "num_tokens": 4497118.0,
+      "step": 530
+    },
+    {
+      "epoch": 1.2181400688863375,
+      "grad_norm": 2.1741283353590455,
+      "learning_rate": 7.680300830488166e-06,
+      "loss": 0.2844,
+      "mean_token_accuracy": 0.9187273681163788,
+      "num_tokens": 4505394.0,
+      "step": 531
+    },
+    {
+      "epoch": 1.2204362801377726,
+      "grad_norm": 2.4826430501517356,
+      "learning_rate": 7.669785870169328e-06,
+      "loss": 0.2821,
+      "mean_token_accuracy": 0.9220278561115265,
+      "num_tokens": 4513913.0,
+      "step": 532
+    },
+    {
+      "epoch": 1.2227324913892077,
+      "grad_norm": 2.3101598261266116,
+      "learning_rate": 7.659255451489726e-06,
+      "loss": 0.2679,
+      "mean_token_accuracy": 0.9224972128868103,
+      "num_tokens": 4521868.0,
+      "step": 533
+    },
+    {
+      "epoch": 1.225028702640643,
+      "grad_norm": 2.3624018970634117,
+      "learning_rate": 7.648709649471973e-06,
+      "loss": 0.2531,
+      "mean_token_accuracy": 0.927400678396225,
+      "num_tokens": 4530485.0,
+      "step": 534
+    },
+    {
+      "epoch": 1.227324913892078,
+      "grad_norm": 2.242642350498953,
+      "learning_rate": 7.638148539248286e-06,
+      "loss": 0.257,
+      "mean_token_accuracy": 0.9278309643268585,
+      "num_tokens": 4539040.0,
+      "step": 535
+    },
+    {
+      "epoch": 1.2296211251435132,
+      "grad_norm": 2.5339498984127005,
+      "learning_rate": 7.6275721960599335e-06,
+      "loss": 0.2695,
+      "mean_token_accuracy": 0.9212221503257751,
+      "num_tokens": 4547438.0,
+      "step": 536
+    },
+    {
+      "epoch": 1.2319173363949483,
+      "grad_norm": 2.380306111882361,
+      "learning_rate": 7.616980695256717e-06,
+      "loss": 0.2845,
+      "mean_token_accuracy": 0.9190695285797119,
+      "num_tokens": 4555876.0,
+      "step": 537
+    },
+    {
+      "epoch": 1.2342135476463834,
+      "grad_norm": 2.6203679928649626,
+      "learning_rate": 7.606374112296421e-06,
+      "loss": 0.284,
+      "mean_token_accuracy": 0.918077290058136,
+      "num_tokens": 4564325.0,
+      "step": 538
+    },
+    {
+      "epoch": 1.2365097588978187,
+      "grad_norm": 2.6122880567950113,
+      "learning_rate": 7.595752522744287e-06,
+      "loss": 0.2465,
+      "mean_token_accuracy": 0.9271360039710999,
+      "num_tokens": 4573006.0,
+      "step": 539
+    },
+    {
+      "epoch": 1.2388059701492538,
+      "grad_norm": 2.5958748623821273,
+      "learning_rate": 7.585116002272463e-06,
+      "loss": 0.3008,
+      "mean_token_accuracy": 0.9182252585887909,
+      "num_tokens": 4581073.0,
+      "step": 540
+    },
+    {
+      "epoch": 1.2411021814006888,
+      "grad_norm": 2.4236145358148633,
+      "learning_rate": 7.574464626659476e-06,
+      "loss": 0.2786,
+      "mean_token_accuracy": 0.9199983179569244,
+      "num_tokens": 4589115.0,
+      "step": 541
+    },
+    {
+      "epoch": 1.243398392652124,
+      "grad_norm": 2.364744515546965,
+      "learning_rate": 7.563798471789678e-06,
+      "loss": 0.2881,
+      "mean_token_accuracy": 0.9189049601554871,
+      "num_tokens": 4597267.0,
+      "step": 542
+    },
+    {
+      "epoch": 1.245694603903559,
+      "grad_norm": 2.218862420952384,
+      "learning_rate": 7.553117613652726e-06,
+      "loss": 0.2667,
+      "mean_token_accuracy": 0.9258249402046204,
+      "num_tokens": 4605932.0,
+      "step": 543
+    },
+    {
+      "epoch": 1.2479908151549943,
+      "grad_norm": 2.25369212455571,
+      "learning_rate": 7.542422128343018e-06,
+      "loss": 0.2647,
+      "mean_token_accuracy": 0.9258838295936584,
+      "num_tokens": 4614355.0,
+      "step": 544
+    },
+    {
+      "epoch": 1.2502870264064294,
+      "grad_norm": 2.3344716690933462,
+      "learning_rate": 7.531712092059167e-06,
+      "loss": 0.2967,
+      "mean_token_accuracy": 0.9121121466159821,
+      "num_tokens": 4623023.0,
+      "step": 545
+    },
+    {
+      "epoch": 1.2525832376578645,
+      "grad_norm": 2.0869702230261393,
+      "learning_rate": 7.520987581103451e-06,
+      "loss": 0.2268,
+      "mean_token_accuracy": 0.9309463798999786,
+      "num_tokens": 4631860.0,
+      "step": 546
+    },
+    {
+      "epoch": 1.2548794489092996,
+      "grad_norm": 1.9579900625778206,
+      "learning_rate": 7.51024867188127e-06,
+      "loss": 0.2349,
+      "mean_token_accuracy": 0.9305654168128967,
+      "num_tokens": 4640817.0,
+      "step": 547
+    },
+    {
+      "epoch": 1.2571756601607347,
+      "grad_norm": 2.311684560493296,
+      "learning_rate": 7.499495440900604e-06,
+      "loss": 0.2814,
+      "mean_token_accuracy": 0.9203662276268005,
+      "num_tokens": 4649728.0,
+      "step": 548
+    },
+    {
+      "epoch": 1.25947187141217,
+      "grad_norm": 2.3815843644907537,
+      "learning_rate": 7.488727964771466e-06,
+      "loss": 0.2667,
+      "mean_token_accuracy": 0.9252900779247284,
+      "num_tokens": 4657835.0,
+      "step": 549
+    },
+    {
+      "epoch": 1.261768082663605,
+      "grad_norm": 2.474198548935571,
+      "learning_rate": 7.477946320205358e-06,
+      "loss": 0.271,
+      "mean_token_accuracy": 0.9247826337814331,
+      "num_tokens": 4665931.0,
+      "step": 550
+    },
+    {
+      "epoch": 1.2640642939150402,
+      "grad_norm": 2.4418047070145237,
+      "learning_rate": 7.4671505840147195e-06,
+      "loss": 0.2791,
+      "mean_token_accuracy": 0.9177079796791077,
+      "num_tokens": 4674056.0,
+      "step": 551
+    },
+    {
+      "epoch": 1.2663605051664752,
+      "grad_norm": 2.260902231629871,
+      "learning_rate": 7.456340833112389e-06,
+      "loss": 0.2373,
+      "mean_token_accuracy": 0.9344957172870636,
+      "num_tokens": 4682351.0,
+      "step": 552
+    },
+    {
+      "epoch": 1.2686567164179103,
+      "grad_norm": 2.24729527974216,
+      "learning_rate": 7.445517144511048e-06,
+      "loss": 0.2258,
+      "mean_token_accuracy": 0.9328469038009644,
+      "num_tokens": 4691650.0,
+      "step": 553
+    },
+    {
+      "epoch": 1.2709529276693456,
+      "grad_norm": 2.2739517566396263,
+      "learning_rate": 7.434679595322675e-06,
+      "loss": 0.2557,
+      "mean_token_accuracy": 0.9269214868545532,
+      "num_tokens": 4700381.0,
+      "step": 554
+    },
+    {
+      "epoch": 1.2732491389207807,
+      "grad_norm": 2.56130865892528,
+      "learning_rate": 7.423828262757997e-06,
+      "loss": 0.2706,
+      "mean_token_accuracy": 0.9224008321762085,
+      "num_tokens": 4708971.0,
+      "step": 555
+    },
+    {
+      "epoch": 1.2755453501722158,
+      "grad_norm": 2.2118169316625553,
+      "learning_rate": 7.41296322412594e-06,
+      "loss": 0.2204,
+      "mean_token_accuracy": 0.937141478061676,
+      "num_tokens": 4717084.0,
+      "step": 556
+    },
+    {
+      "epoch": 1.277841561423651,
+      "grad_norm": 2.4777627324516405,
+      "learning_rate": 7.4020845568330765e-06,
+      "loss": 0.2647,
+      "mean_token_accuracy": 0.9250539541244507,
+      "num_tokens": 4725783.0,
+      "step": 557
+    },
+    {
+      "epoch": 1.280137772675086,
+      "grad_norm": 2.3278112367632766,
+      "learning_rate": 7.3911923383830745e-06,
+      "loss": 0.2368,
+      "mean_token_accuracy": 0.9306104779243469,
+      "num_tokens": 4734357.0,
+      "step": 558
+    },
+    {
+      "epoch": 1.2824339839265213,
+      "grad_norm": 2.307187757548464,
+      "learning_rate": 7.380286646376145e-06,
+      "loss": 0.2643,
+      "mean_token_accuracy": 0.9232437014579773,
+      "num_tokens": 4743382.0,
+      "step": 559
+    },
+    {
+      "epoch": 1.2847301951779564,
+      "grad_norm": 2.1856705807671912,
+      "learning_rate": 7.36936755850849e-06,
+      "loss": 0.2531,
+      "mean_token_accuracy": 0.9266568720340729,
+      "num_tokens": 4751941.0,
+      "step": 560
+    },
+    {
+      "epoch": 1.2870264064293915,
+      "grad_norm": 2.461167566283754,
+      "learning_rate": 7.358435152571749e-06,
+      "loss": 0.2888,
+      "mean_token_accuracy": 0.9186607599258423,
+      "num_tokens": 4760977.0,
+      "step": 561
+    },
+    {
+      "epoch": 1.2893226176808266,
+      "grad_norm": 2.5163871231722665,
+      "learning_rate": 7.347489506452443e-06,
+      "loss": 0.2768,
+      "mean_token_accuracy": 0.9201366603374481,
+      "num_tokens": 4769538.0,
+      "step": 562
+    },
+    {
+      "epoch": 1.2916188289322617,
+      "grad_norm": 2.0219929078717707,
+      "learning_rate": 7.336530698131424e-06,
+      "loss": 0.2499,
+      "mean_token_accuracy": 0.9269580245018005,
+      "num_tokens": 4778352.0,
+      "step": 563
+    },
+    {
+      "epoch": 1.293915040183697,
+      "grad_norm": 2.283628569162065,
+      "learning_rate": 7.325558805683313e-06,
+      "loss": 0.2718,
+      "mean_token_accuracy": 0.9196649491786957,
+      "num_tokens": 4786978.0,
+      "step": 564
+    },
+    {
+      "epoch": 1.296211251435132,
+      "grad_norm": 2.166863188951735,
+      "learning_rate": 7.314573907275948e-06,
+      "loss": 0.2299,
+      "mean_token_accuracy": 0.9321498572826385,
+      "num_tokens": 4795864.0,
+      "step": 565
+    },
+    {
+      "epoch": 1.2985074626865671,
+      "grad_norm": 2.0960664291737157,
+      "learning_rate": 7.30357608116983e-06,
+      "loss": 0.2367,
+      "mean_token_accuracy": 0.9318011105060577,
+      "num_tokens": 4804563.0,
+      "step": 566
+    },
+    {
+      "epoch": 1.3008036739380022,
+      "grad_norm": 2.2060613844042214,
+      "learning_rate": 7.2925654057175555e-06,
+      "loss": 0.2627,
+      "mean_token_accuracy": 0.9262889325618744,
+      "num_tokens": 4812876.0,
+      "step": 567
+    },
+    {
+      "epoch": 1.3030998851894373,
+      "grad_norm": 2.235273690912246,
+      "learning_rate": 7.281541959363268e-06,
+      "loss": 0.2664,
+      "mean_token_accuracy": 0.9233042895793915,
+      "num_tokens": 4821586.0,
+      "step": 568
+    },
+    {
+      "epoch": 1.3053960964408726,
+      "grad_norm": 2.4054044638462706,
+      "learning_rate": 7.270505820642096e-06,
+      "loss": 0.2628,
+      "mean_token_accuracy": 0.9242512583732605,
+      "num_tokens": 4830295.0,
+      "step": 569
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 2.2846567529706325,
+      "learning_rate": 7.259457068179593e-06,
+      "loss": 0.2212,
+      "mean_token_accuracy": 0.9308845102787018,
+      "num_tokens": 4839293.0,
+      "step": 570
+    },
+    {
+      "epoch": 1.3099885189437428,
+      "grad_norm": 2.0692779190690316,
+      "learning_rate": 7.248395780691177e-06,
+      "loss": 0.2717,
+      "mean_token_accuracy": 0.926284909248352,
+      "num_tokens": 4848326.0,
+      "step": 571
+    },
+    {
+      "epoch": 1.3122847301951779,
+      "grad_norm": 2.4699395391248355,
+      "learning_rate": 7.237322036981568e-06,
+      "loss": 0.2314,
+      "mean_token_accuracy": 0.9366464614868164,
+      "num_tokens": 4856869.0,
+      "step": 572
+    },
+    {
+      "epoch": 1.314580941446613,
+      "grad_norm": 2.3732860816756025,
+      "learning_rate": 7.226235915944233e-06,
+      "loss": 0.2624,
+      "mean_token_accuracy": 0.9259655773639679,
+      "num_tokens": 4864770.0,
+      "step": 573
+    },
+    {
+      "epoch": 1.3168771526980483,
+      "grad_norm": 2.2630254127394758,
+      "learning_rate": 7.215137496560816e-06,
+      "loss": 0.2488,
+      "mean_token_accuracy": 0.9286541044712067,
+      "num_tokens": 4872981.0,
+      "step": 574
+    },
+    {
+      "epoch": 1.3191733639494834,
+      "grad_norm": 2.8187093355188018,
+      "learning_rate": 7.204026857900583e-06,
+      "loss": 0.3022,
+      "mean_token_accuracy": 0.9133457243442535,
+      "num_tokens": 4881579.0,
+      "step": 575
+    },
+    {
+      "epoch": 1.3214695752009185,
+      "grad_norm": 2.66594660120616,
+      "learning_rate": 7.192904079119849e-06,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.9128538966178894,
+      "num_tokens": 4889658.0,
+      "step": 576
+    },
+    {
+      "epoch": 1.3237657864523535,
+      "grad_norm": 2.3860271671430793,
+      "learning_rate": 7.181769239461425e-06,
+      "loss": 0.2454,
+      "mean_token_accuracy": 0.9298608601093292,
+      "num_tokens": 4897572.0,
+      "step": 577
+    },
+    {
+      "epoch": 1.3260619977037886,
+      "grad_norm": 2.3675742002331903,
+      "learning_rate": 7.170622418254045e-06,
+      "loss": 0.2629,
+      "mean_token_accuracy": 0.9275126457214355,
+      "num_tokens": 4905640.0,
+      "step": 578
+    },
+    {
+      "epoch": 1.328358208955224,
+      "grad_norm": 1.9932281961422453,
+      "learning_rate": 7.159463694911806e-06,
+      "loss": 0.2217,
+      "mean_token_accuracy": 0.9336240887641907,
+      "num_tokens": 4914252.0,
+      "step": 579
+    },
+    {
+      "epoch": 1.330654420206659,
+      "grad_norm": 2.0477863894204833,
+      "learning_rate": 7.148293148933598e-06,
+      "loss": 0.2357,
+      "mean_token_accuracy": 0.9284582436084747,
+      "num_tokens": 4921932.0,
+      "step": 580
+    },
+    {
+      "epoch": 1.3329506314580941,
+      "grad_norm": 2.1778965754250676,
+      "learning_rate": 7.137110859902543e-06,
+      "loss": 0.2297,
+      "mean_token_accuracy": 0.9337016642093658,
+      "num_tokens": 4930338.0,
+      "step": 581
+    },
+    {
+      "epoch": 1.3352468427095292,
+      "grad_norm": 2.146947917797984,
+      "learning_rate": 7.125916907485423e-06,
+      "loss": 0.2588,
+      "mean_token_accuracy": 0.9280851185321808,
+      "num_tokens": 4937734.0,
+      "step": 582
+    },
+    {
+      "epoch": 1.3375430539609643,
+      "grad_norm": 2.3786668768356147,
+      "learning_rate": 7.114711371432113e-06,
+      "loss": 0.2536,
+      "mean_token_accuracy": 0.9297422170639038,
+      "num_tokens": 4946722.0,
+      "step": 583
+    },
+    {
+      "epoch": 1.3398392652123996,
+      "grad_norm": 2.4643726827768617,
+      "learning_rate": 7.103494331575015e-06,
+      "loss": 0.2837,
+      "mean_token_accuracy": 0.920375645160675,
+      "num_tokens": 4956040.0,
+      "step": 584
+    },
+    {
+      "epoch": 1.3421354764638347,
+      "grad_norm": 2.842572472055857,
+      "learning_rate": 7.09226586782849e-06,
+      "loss": 0.2723,
+      "mean_token_accuracy": 0.9209208786487579,
+      "num_tokens": 4964753.0,
+      "step": 585
+    },
+    {
+      "epoch": 1.3444316877152698,
+      "grad_norm": 2.668061540414086,
+      "learning_rate": 7.081026060188287e-06,
+      "loss": 0.2612,
+      "mean_token_accuracy": 0.9247652590274811,
+      "num_tokens": 4973305.0,
+      "step": 586
+    },
+    {
+      "epoch": 1.3467278989667049,
+      "grad_norm": 2.512632956290171,
+      "learning_rate": 7.069774988730968e-06,
+      "loss": 0.2772,
+      "mean_token_accuracy": 0.9193479716777802,
+      "num_tokens": 4981921.0,
+      "step": 587
+    },
+    {
+      "epoch": 1.34902411021814,
+      "grad_norm": 2.5945706782379174,
+      "learning_rate": 7.058512733613353e-06,
+      "loss": 0.2746,
+      "mean_token_accuracy": 0.9204272329807281,
+      "num_tokens": 4990157.0,
+      "step": 588
+    },
+    {
+      "epoch": 1.3513203214695753,
+      "grad_norm": 2.340016733605834,
+      "learning_rate": 7.047239375071926e-06,
+      "loss": 0.2587,
+      "mean_token_accuracy": 0.9230052828788757,
+      "num_tokens": 4999567.0,
+      "step": 589
+    },
+    {
+      "epoch": 1.3536165327210103,
+      "grad_norm": 2.380561472437187,
+      "learning_rate": 7.035954993422285e-06,
+      "loss": 0.275,
+      "mean_token_accuracy": 0.9247272908687592,
+      "num_tokens": 5007767.0,
+      "step": 590
+    },
+    {
+      "epoch": 1.3559127439724454,
+      "grad_norm": 2.216970633716535,
+      "learning_rate": 7.024659669058558e-06,
+      "loss": 0.2546,
+      "mean_token_accuracy": 0.9257120788097382,
+      "num_tokens": 5016650.0,
+      "step": 591
+    },
+    {
+      "epoch": 1.3582089552238805,
+      "grad_norm": 2.3390418939040143,
+      "learning_rate": 7.013353482452831e-06,
+      "loss": 0.2869,
+      "mean_token_accuracy": 0.9182985424995422,
+      "num_tokens": 5025358.0,
+      "step": 592
+    },
+    {
+      "epoch": 1.3605051664753156,
+      "grad_norm": 2.6823715050331827,
+      "learning_rate": 7.00203651415458e-06,
+      "loss": 0.2637,
+      "mean_token_accuracy": 0.9294079542160034,
+      "num_tokens": 5034214.0,
+      "step": 593
+    },
+    {
+      "epoch": 1.362801377726751,
+      "grad_norm": 2.2940109275808345,
+      "learning_rate": 6.990708844790093e-06,
+      "loss": 0.2761,
+      "mean_token_accuracy": 0.9192169308662415,
+      "num_tokens": 5042840.0,
+      "step": 594
+    },
+    {
+      "epoch": 1.365097588978186,
+      "grad_norm": 2.446739093200835,
+      "learning_rate": 6.979370555061893e-06,
+      "loss": 0.2577,
+      "mean_token_accuracy": 0.9250144362449646,
+      "num_tokens": 5051390.0,
+      "step": 595
+    },
+    {
+      "epoch": 1.367393800229621,
+      "grad_norm": 2.3976613682084715,
+      "learning_rate": 6.96802172574817e-06,
+      "loss": 0.2768,
+      "mean_token_accuracy": 0.9218591451644897,
+      "num_tokens": 5059975.0,
+      "step": 596
+    },
+    {
+      "epoch": 1.3696900114810562,
+      "grad_norm": 2.110675458081428,
+      "learning_rate": 6.956662437702201e-06,
+      "loss": 0.2445,
+      "mean_token_accuracy": 0.9325763583183289,
+      "num_tokens": 5068591.0,
+      "step": 597
+    },
+    {
+      "epoch": 1.3719862227324913,
+      "grad_norm": 2.1064686679243563,
+      "learning_rate": 6.945292771851777e-06,
+      "loss": 0.2778,
+      "mean_token_accuracy": 0.9207326471805573,
+      "num_tokens": 5076962.0,
+      "step": 598
+    },
+    {
+      "epoch": 1.3742824339839266,
+      "grad_norm": 2.1613698883837933,
+      "learning_rate": 6.933912809198618e-06,
+      "loss": 0.2667,
+      "mean_token_accuracy": 0.9268639385700226,
+      "num_tokens": 5084951.0,
+      "step": 599
+    },
+    {
+      "epoch": 1.3765786452353617,
+      "grad_norm": 2.3094277554975875,
+      "learning_rate": 6.922522630817808e-06,
+      "loss": 0.2633,
+      "mean_token_accuracy": 0.9271392226219177,
+      "num_tokens": 5093447.0,
+      "step": 600
+    },
+    {
+      "epoch": 1.3788748564867968,
+      "grad_norm": 2.2469564257917862,
+      "learning_rate": 6.911122317857213e-06,
+      "loss": 0.2453,
+      "mean_token_accuracy": 0.9309904277324677,
+      "num_tokens": 5102038.0,
+      "step": 601
+    },
+    {
+      "epoch": 1.3811710677382318,
+      "grad_norm": 2.5686029279896245,
+      "learning_rate": 6.899711951536895e-06,
+      "loss": 0.3073,
+      "mean_token_accuracy": 0.9159621298313141,
+      "num_tokens": 5110308.0,
+      "step": 602
+    },
+    {
+      "epoch": 1.383467278989667,
+      "grad_norm": 2.259550949807883,
+      "learning_rate": 6.888291613148543e-06,
+      "loss": 0.2466,
+      "mean_token_accuracy": 0.9273211061954498,
+      "num_tokens": 5118679.0,
+      "step": 603
+    },
+    {
+      "epoch": 1.3857634902411022,
+      "grad_norm": 2.0472993198576237,
+      "learning_rate": 6.876861384054894e-06,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.9334699511528015,
+      "num_tokens": 5126865.0,
+      "step": 604
+    },
+    {
+      "epoch": 1.3880597014925373,
+      "grad_norm": 2.3954190970105134,
+      "learning_rate": 6.865421345689147e-06,
+      "loss": 0.2477,
+      "mean_token_accuracy": 0.9291063249111176,
+      "num_tokens": 5134735.0,
+      "step": 605
+    },
+    {
+      "epoch": 1.3903559127439724,
+      "grad_norm": 2.412544289505079,
+      "learning_rate": 6.8539715795543825e-06,
+      "loss": 0.2603,
+      "mean_token_accuracy": 0.9260501563549042,
+      "num_tokens": 5143365.0,
+      "step": 606
+    },
+    {
+      "epoch": 1.3926521239954075,
+      "grad_norm": 2.499262423357595,
+      "learning_rate": 6.842512167222992e-06,
+      "loss": 0.2279,
+      "mean_token_accuracy": 0.9345058798789978,
+      "num_tokens": 5151510.0,
+      "step": 607
+    },
+    {
+      "epoch": 1.3949483352468426,
+      "grad_norm": 2.365509896034399,
+      "learning_rate": 6.831043190336086e-06,
+      "loss": 0.2662,
+      "mean_token_accuracy": 0.9227205216884613,
+      "num_tokens": 5160070.0,
+      "step": 608
+    },
+    {
+      "epoch": 1.397244546498278,
+      "grad_norm": 2.4193556090375536,
+      "learning_rate": 6.8195647306029185e-06,
+      "loss": 0.2581,
+      "mean_token_accuracy": 0.9251465797424316,
+      "num_tokens": 5167757.0,
+      "step": 609
+    },
+    {
+      "epoch": 1.399540757749713,
+      "grad_norm": 2.382259380307371,
+      "learning_rate": 6.808076869800302e-06,
+      "loss": 0.2561,
+      "mean_token_accuracy": 0.9263916611671448,
+      "num_tokens": 5176184.0,
+      "step": 610
+    },
+    {
+      "epoch": 1.401836969001148,
+      "grad_norm": 2.36399593565171,
+      "learning_rate": 6.79657968977202e-06,
+      "loss": 0.2652,
+      "mean_token_accuracy": 0.922259658575058,
+      "num_tokens": 5184756.0,
+      "step": 611
+    },
+    {
+      "epoch": 1.4041331802525832,
+      "grad_norm": 2.305147460999811,
+      "learning_rate": 6.785073272428262e-06,
+      "loss": 0.263,
+      "mean_token_accuracy": 0.9238495826721191,
+      "num_tokens": 5193074.0,
+      "step": 612
+    },
+    {
+      "epoch": 1.4064293915040182,
+      "grad_norm": 2.2295049315325817,
+      "learning_rate": 6.773557699745014e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.9340808689594269,
+      "num_tokens": 5202123.0,
+      "step": 613
+    },
+    {
+      "epoch": 1.4087256027554536,
+      "grad_norm": 2.1559288416306788,
+      "learning_rate": 6.762033053763501e-06,
+      "loss": 0.2588,
+      "mean_token_accuracy": 0.9293081164360046,
+      "num_tokens": 5210893.0,
+      "step": 614
+    },
+    {
+      "epoch": 1.4110218140068886,
+      "grad_norm": 2.3833984936854367,
+      "learning_rate": 6.750499416589581e-06,
+      "loss": 0.2482,
+      "mean_token_accuracy": 0.9323559403419495,
+      "num_tokens": 5219008.0,
+      "step": 615
+    },
+    {
+      "epoch": 1.4133180252583237,
+      "grad_norm": 2.325601659398918,
+      "learning_rate": 6.73895687039317e-06,
+      "loss": 0.2329,
+      "mean_token_accuracy": 0.9328633248806,
+      "num_tokens": 5227581.0,
+      "step": 616
+    },
+    {
+      "epoch": 1.4156142365097588,
+      "grad_norm": 2.2824731626539383,
+      "learning_rate": 6.727405497407655e-06,
+      "loss": 0.2612,
+      "mean_token_accuracy": 0.9266740679740906,
+      "num_tokens": 5236280.0,
+      "step": 617
+    },
+    {
+      "epoch": 1.417910447761194,
+      "grad_norm": 2.6450068406280267,
+      "learning_rate": 6.715845379929311e-06,
+      "loss": 0.2322,
+      "mean_token_accuracy": 0.9323435127735138,
+      "num_tokens": 5245145.0,
+      "step": 618
+    },
+    {
+      "epoch": 1.4202066590126292,
+      "grad_norm": 2.5094955705541473,
+      "learning_rate": 6.70427660031671e-06,
+      "loss": 0.2416,
+      "mean_token_accuracy": 0.9303381741046906,
+      "num_tokens": 5253870.0,
+      "step": 619
+    },
+    {
+      "epoch": 1.4225028702640643,
+      "grad_norm": 2.2859821865137673,
+      "learning_rate": 6.692699240990136e-06,
+      "loss": 0.2556,
+      "mean_token_accuracy": 0.9266341924667358,
+      "num_tokens": 5261975.0,
+      "step": 620
+    },
+    {
+      "epoch": 1.4247990815154994,
+      "grad_norm": 2.2027516015323627,
+      "learning_rate": 6.681113384431e-06,
+      "loss": 0.2584,
+      "mean_token_accuracy": 0.9267958104610443,
+      "num_tokens": 5270495.0,
+      "step": 621
+    },
+    {
+      "epoch": 1.4270952927669345,
+      "grad_norm": 2.376762327471697,
+      "learning_rate": 6.669519113181247e-06,
+      "loss": 0.2256,
+      "mean_token_accuracy": 0.9355348348617554,
+      "num_tokens": 5278565.0,
+      "step": 622
+    },
+    {
+      "epoch": 1.4293915040183696,
+      "grad_norm": 2.0693272625151002,
+      "learning_rate": 6.657916509842776e-06,
+      "loss": 0.2328,
+      "mean_token_accuracy": 0.9342992901802063,
+      "num_tokens": 5287196.0,
+      "step": 623
+    },
+    {
+      "epoch": 1.4316877152698049,
+      "grad_norm": 2.553985747809146,
+      "learning_rate": 6.646305657076839e-06,
+      "loss": 0.3034,
+      "mean_token_accuracy": 0.9176582396030426,
+      "num_tokens": 5295357.0,
+      "step": 624
+    },
+    {
+      "epoch": 1.43398392652124,
+      "grad_norm": 2.2269022830759604,
+      "learning_rate": 6.634686637603471e-06,
+      "loss": 0.2586,
+      "mean_token_accuracy": 0.9268090426921844,
+      "num_tokens": 5303232.0,
+      "step": 625
+    },
+    {
+      "epoch": 1.436280137772675,
+      "grad_norm": 2.3725290122595664,
+      "learning_rate": 6.623059534200883e-06,
+      "loss": 0.2372,
+      "mean_token_accuracy": 0.9320357739925385,
+      "num_tokens": 5311802.0,
+      "step": 626
+    },
+    {
+      "epoch": 1.4385763490241104,
+      "grad_norm": 2.320448965659554,
+      "learning_rate": 6.611424429704879e-06,
+      "loss": 0.2467,
+      "mean_token_accuracy": 0.9300343692302704,
+      "num_tokens": 5319522.0,
+      "step": 627
+    },
+    {
+      "epoch": 1.4408725602755452,
+      "grad_norm": 2.5580371972985216,
+      "learning_rate": 6.599781407008267e-06,
+      "loss": 0.275,
+      "mean_token_accuracy": 0.9213577210903168,
+      "num_tokens": 5327691.0,
+      "step": 628
+    },
+    {
+      "epoch": 1.4431687715269805,
+      "grad_norm": 2.1716284748303085,
+      "learning_rate": 6.588130549060267e-06,
+      "loss": 0.2138,
+      "mean_token_accuracy": 0.936636358499527,
+      "num_tokens": 5336003.0,
+      "step": 629
+    },
+    {
+      "epoch": 1.4454649827784156,
+      "grad_norm": 2.137393547522864,
+      "learning_rate": 6.576471938865917e-06,
+      "loss": 0.2466,
+      "mean_token_accuracy": 0.9275836050510406,
+      "num_tokens": 5344312.0,
+      "step": 630
+    },
+    {
+      "epoch": 1.4477611940298507,
+      "grad_norm": 2.1310891977757316,
+      "learning_rate": 6.564805659485493e-06,
+      "loss": 0.2158,
+      "mean_token_accuracy": 0.9363845884799957,
+      "num_tokens": 5352671.0,
+      "step": 631
+    },
+    {
+      "epoch": 1.450057405281286,
+      "grad_norm": 2.280651617216985,
+      "learning_rate": 6.553131794033902e-06,
+      "loss": 0.2228,
+      "mean_token_accuracy": 0.9336510896682739,
+      "num_tokens": 5360384.0,
+      "step": 632
+    },
+    {
+      "epoch": 1.4523536165327209,
+      "grad_norm": 2.4074422178337405,
+      "learning_rate": 6.541450425680098e-06,
+      "loss": 0.2819,
+      "mean_token_accuracy": 0.9174861311912537,
+      "num_tokens": 5369013.0,
+      "step": 633
+    },
+    {
+      "epoch": 1.4546498277841562,
+      "grad_norm": 2.3099331895625754,
+      "learning_rate": 6.529761637646492e-06,
+      "loss": 0.2395,
+      "mean_token_accuracy": 0.928705483675003,
+      "num_tokens": 5377302.0,
+      "step": 634
+    },
+    {
+      "epoch": 1.4569460390355913,
+      "grad_norm": 2.5185012084552163,
+      "learning_rate": 6.518065513208351e-06,
+      "loss": 0.2719,
+      "mean_token_accuracy": 0.9221943914890289,
+      "num_tokens": 5385252.0,
+      "step": 635
+    },
+    {
+      "epoch": 1.4592422502870264,
+      "grad_norm": 2.4038933930389046,
+      "learning_rate": 6.5063621356932105e-06,
+      "loss": 0.2584,
+      "mean_token_accuracy": 0.9285935163497925,
+      "num_tokens": 5394222.0,
+      "step": 636
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 2.622422471091549,
+      "learning_rate": 6.494651588480284e-06,
+      "loss": 0.2525,
+      "mean_token_accuracy": 0.9300331473350525,
+      "num_tokens": 5402650.0,
+      "step": 637
+    },
+    {
+      "epoch": 1.4638346727898965,
+      "grad_norm": 2.230919453942294,
+      "learning_rate": 6.482933954999858e-06,
+      "loss": 0.2634,
+      "mean_token_accuracy": 0.9266985654830933,
+      "num_tokens": 5411173.0,
+      "step": 638
+    },
+    {
+      "epoch": 1.4661308840413318,
+      "grad_norm": 2.4346384906419334,
+      "learning_rate": 6.471209318732711e-06,
+      "loss": 0.2328,
+      "mean_token_accuracy": 0.9308375716209412,
+      "num_tokens": 5419097.0,
+      "step": 639
+    },
+    {
+      "epoch": 1.468427095292767,
+      "grad_norm": 1.9419493108658765,
+      "learning_rate": 6.459477763209504e-06,
+      "loss": 0.2396,
+      "mean_token_accuracy": 0.930637925863266,
+      "num_tokens": 5428685.0,
+      "step": 640
+    },
+    {
+      "epoch": 1.470723306544202,
+      "grad_norm": 2.176359526087256,
+      "learning_rate": 6.447739372010201e-06,
+      "loss": 0.2307,
+      "mean_token_accuracy": 0.9319821000099182,
+      "num_tokens": 5437140.0,
+      "step": 641
+    },
+    {
+      "epoch": 1.4730195177956373,
+      "grad_norm": 2.173414309517703,
+      "learning_rate": 6.435994228763459e-06,
+      "loss": 0.2139,
+      "mean_token_accuracy": 0.9386887848377228,
+      "num_tokens": 5445013.0,
+      "step": 642
+    },
+    {
+      "epoch": 1.4753157290470722,
+      "grad_norm": 2.240558469773626,
+      "learning_rate": 6.424242417146047e-06,
+      "loss": 0.2439,
+      "mean_token_accuracy": 0.9305135905742645,
+      "num_tokens": 5453218.0,
+      "step": 643
+    },
+    {
+      "epoch": 1.4776119402985075,
+      "grad_norm": 2.429037208560937,
+      "learning_rate": 6.412484020882234e-06,
+      "loss": 0.2571,
+      "mean_token_accuracy": 0.9252912104129791,
+      "num_tokens": 5461470.0,
+      "step": 644
+    },
+    {
+      "epoch": 1.4799081515499426,
+      "grad_norm": 2.255649524768215,
+      "learning_rate": 6.400719123743206e-06,
+      "loss": 0.2698,
+      "mean_token_accuracy": 0.930851548910141,
+      "num_tokens": 5469562.0,
+      "step": 645
+    },
+    {
+      "epoch": 1.4822043628013777,
+      "grad_norm": 2.5868183194421692,
+      "learning_rate": 6.38894780954646e-06,
+      "loss": 0.2155,
+      "mean_token_accuracy": 0.9354539215564728,
+      "num_tokens": 5477497.0,
+      "step": 646
+    },
+    {
+      "epoch": 1.484500574052813,
+      "grad_norm": 2.3194595210124436,
+      "learning_rate": 6.3771701621552125e-06,
+      "loss": 0.234,
+      "mean_token_accuracy": 0.9306348562240601,
+      "num_tokens": 5486043.0,
+      "step": 647
+    },
+    {
+      "epoch": 1.4867967853042479,
+      "grad_norm": 2.1634530477000857,
+      "learning_rate": 6.365386265477803e-06,
+      "loss": 0.2312,
+      "mean_token_accuracy": 0.9336646795272827,
+      "num_tokens": 5493481.0,
+      "step": 648
+    },
+    {
+      "epoch": 1.4890929965556832,
+      "grad_norm": 2.6065356406892466,
+      "learning_rate": 6.353596203467085e-06,
+      "loss": 0.2499,
+      "mean_token_accuracy": 0.9247941076755524,
+      "num_tokens": 5502264.0,
+      "step": 649
+    },
+    {
+      "epoch": 1.4913892078071183,
+      "grad_norm": 2.3534040981862776,
+      "learning_rate": 6.341800060119845e-06,
+      "loss": 0.252,
+      "mean_token_accuracy": 0.9264822006225586,
+      "num_tokens": 5510096.0,
+      "step": 650
+    },
+    {
+      "epoch": 1.4936854190585533,
+      "grad_norm": 2.551702146631783,
+      "learning_rate": 6.329997919476189e-06,
+      "loss": 0.2521,
+      "mean_token_accuracy": 0.9294766485691071,
+      "num_tokens": 5519449.0,
+      "step": 651
+    },
+    {
+      "epoch": 1.4959816303099887,
+      "grad_norm": 2.127538367341916,
+      "learning_rate": 6.318189865618955e-06,
+      "loss": 0.278,
+      "mean_token_accuracy": 0.9192856848239899,
+      "num_tokens": 5527787.0,
+      "step": 652
+    },
+    {
+      "epoch": 1.4982778415614235,
+      "grad_norm": 2.1420860608514913,
+      "learning_rate": 6.3063759826731004e-06,
+      "loss": 0.2239,
+      "mean_token_accuracy": 0.9365537464618683,
+      "num_tokens": 5535707.0,
+      "step": 653
+    },
+    {
+      "epoch": 1.5005740528128588,
+      "grad_norm": 2.1670490724511606,
+      "learning_rate": 6.294556354805124e-06,
+      "loss": 0.2252,
+      "mean_token_accuracy": 0.9378991425037384,
+      "num_tokens": 5544081.0,
+      "step": 654
+    },
+    {
+      "epoch": 1.502870264064294,
+      "grad_norm": 1.9177509828358672,
+      "learning_rate": 6.282731066222442e-06,
+      "loss": 0.2254,
+      "mean_token_accuracy": 0.9325357377529144,
+      "num_tokens": 5553309.0,
+      "step": 655
+    },
+    {
+      "epoch": 1.505166475315729,
+      "grad_norm": 2.2148159463143706,
+      "learning_rate": 6.270900201172806e-06,
+      "loss": 0.2366,
+      "mean_token_accuracy": 0.9345337450504303,
+      "num_tokens": 5562044.0,
+      "step": 656
+    },
+    {
+      "epoch": 1.5074626865671643,
+      "grad_norm": 2.197179606383364,
+      "learning_rate": 6.259063843943693e-06,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.9367739260196686,
+      "num_tokens": 5569865.0,
+      "step": 657
+    },
+    {
+      "epoch": 1.5097588978185992,
+      "grad_norm": 2.3754220741770684,
+      "learning_rate": 6.247222078861712e-06,
+      "loss": 0.2827,
+      "mean_token_accuracy": 0.9186223149299622,
+      "num_tokens": 5578503.0,
+      "step": 658
+    },
+    {
+      "epoch": 1.5120551090700345,
+      "grad_norm": 2.6795348461203736,
+      "learning_rate": 6.2353749902919966e-06,
+      "loss": 0.2474,
+      "mean_token_accuracy": 0.9284341931343079,
+      "num_tokens": 5587593.0,
+      "step": 659
+    },
+    {
+      "epoch": 1.5143513203214696,
+      "grad_norm": 2.693140600123538,
+      "learning_rate": 6.2235226626376075e-06,
+      "loss": 0.241,
+      "mean_token_accuracy": 0.9316118657588959,
+      "num_tokens": 5596578.0,
+      "step": 660
+    },
+    {
+      "epoch": 1.5166475315729047,
+      "grad_norm": 2.4403723895199096,
+      "learning_rate": 6.211665180338931e-06,
+      "loss": 0.2501,
+      "mean_token_accuracy": 0.9293335974216461,
+      "num_tokens": 5605089.0,
+      "step": 661
+    },
+    {
+      "epoch": 1.51894374282434,
+      "grad_norm": 2.3093534651018284,
+      "learning_rate": 6.199802627873077e-06,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.9408807754516602,
+      "num_tokens": 5613353.0,
+      "step": 662
+    },
+    {
+      "epoch": 1.5212399540757748,
+      "grad_norm": 2.1247442212648813,
+      "learning_rate": 6.187935089753278e-06,
+      "loss": 0.2264,
+      "mean_token_accuracy": 0.9375396072864532,
+      "num_tokens": 5621628.0,
+      "step": 663
+    },
+    {
+      "epoch": 1.5235361653272101,
+      "grad_norm": 2.146970591144253,
+      "learning_rate": 6.176062650528282e-06,
+      "loss": 0.2249,
+      "mean_token_accuracy": 0.9378609955310822,
+      "num_tokens": 5630255.0,
+      "step": 664
+    },
+    {
+      "epoch": 1.5258323765786452,
+      "grad_norm": 2.6381947997246007,
+      "learning_rate": 6.164185394781761e-06,
+      "loss": 0.2268,
+      "mean_token_accuracy": 0.9318024814128876,
+      "num_tokens": 5638142.0,
+      "step": 665
+    },
+    {
+      "epoch": 1.5281285878300803,
+      "grad_norm": 2.138289486002816,
+      "learning_rate": 6.1523034071316955e-06,
+      "loss": 0.2441,
+      "mean_token_accuracy": 0.92900151014328,
+      "num_tokens": 5646340.0,
+      "step": 666
+    },
+    {
+      "epoch": 1.5304247990815156,
+      "grad_norm": 2.7178947034554146,
+      "learning_rate": 6.140416772229785e-06,
+      "loss": 0.2871,
+      "mean_token_accuracy": 0.9212289750576019,
+      "num_tokens": 5655340.0,
+      "step": 667
+    },
+    {
+      "epoch": 1.5327210103329505,
+      "grad_norm": 2.0973582225030407,
+      "learning_rate": 6.128525574760827e-06,
+      "loss": 0.2236,
+      "mean_token_accuracy": 0.9344654083251953,
+      "num_tokens": 5664056.0,
+      "step": 668
+    },
+    {
+      "epoch": 1.5350172215843858,
+      "grad_norm": 2.4291607449995123,
+      "learning_rate": 6.116629899442134e-06,
+      "loss": 0.2436,
+      "mean_token_accuracy": 0.9329268038272858,
+      "num_tokens": 5671870.0,
+      "step": 669
+    },
+    {
+      "epoch": 1.537313432835821,
+      "grad_norm": 2.1800281231621734,
+      "learning_rate": 6.104729831022918e-06,
+      "loss": 0.2479,
+      "mean_token_accuracy": 0.929973840713501,
+      "num_tokens": 5681509.0,
+      "step": 670
+    },
+    {
+      "epoch": 1.539609644087256,
+      "grad_norm": 2.260234673847166,
+      "learning_rate": 6.0928254542836855e-06,
+      "loss": 0.2707,
+      "mean_token_accuracy": 0.9222416877746582,
+      "num_tokens": 5690062.0,
+      "step": 671
+    },
+    {
+      "epoch": 1.5419058553386913,
+      "grad_norm": 2.3881346936296524,
+      "learning_rate": 6.0809168540356415e-06,
+      "loss": 0.2785,
+      "mean_token_accuracy": 0.9203126728534698,
+      "num_tokens": 5698327.0,
+      "step": 672
+    },
+    {
+      "epoch": 1.5442020665901262,
+      "grad_norm": 2.391805066202114,
+      "learning_rate": 6.0690041151200775e-06,
+      "loss": 0.2347,
+      "mean_token_accuracy": 0.9306855201721191,
+      "num_tokens": 5706727.0,
+      "step": 673
+    },
+    {
+      "epoch": 1.5464982778415615,
+      "grad_norm": 2.3653958748256545,
+      "learning_rate": 6.0570873224077764e-06,
+      "loss": 0.2606,
+      "mean_token_accuracy": 0.925815224647522,
+      "num_tokens": 5714859.0,
+      "step": 674
+    },
+    {
+      "epoch": 1.5487944890929966,
+      "grad_norm": 2.514677353093876,
+      "learning_rate": 6.04516656079839e-06,
+      "loss": 0.2723,
+      "mean_token_accuracy": 0.9224299490451813,
+      "num_tokens": 5723027.0,
+      "step": 675
+    },
+    {
+      "epoch": 1.5510907003444316,
+      "grad_norm": 2.3539190883728907,
+      "learning_rate": 6.033241915219861e-06,
+      "loss": 0.2284,
+      "mean_token_accuracy": 0.9359452724456787,
+      "num_tokens": 5731562.0,
+      "step": 676
+    },
+    {
+      "epoch": 1.553386911595867,
+      "grad_norm": 2.222352021390498,
+      "learning_rate": 6.021313470627788e-06,
+      "loss": 0.2736,
+      "mean_token_accuracy": 0.9199343323707581,
+      "num_tokens": 5741405.0,
+      "step": 677
+    },
+    {
+      "epoch": 1.5556831228473018,
+      "grad_norm": 2.062204769011021,
+      "learning_rate": 6.00938131200485e-06,
+      "loss": 0.2145,
+      "mean_token_accuracy": 0.937533438205719,
+      "num_tokens": 5749455.0,
+      "step": 678
+    },
+    {
+      "epoch": 1.5579793340987371,
+      "grad_norm": 2.2225411410808524,
+      "learning_rate": 5.997445524360172e-06,
+      "loss": 0.2278,
+      "mean_token_accuracy": 0.9366957247257233,
+      "num_tokens": 5758071.0,
+      "step": 679
+    },
+    {
+      "epoch": 1.5602755453501722,
+      "grad_norm": 2.296150828284112,
+      "learning_rate": 5.985506192728744e-06,
+      "loss": 0.2514,
+      "mean_token_accuracy": 0.9286310970783234,
+      "num_tokens": 5766952.0,
+      "step": 680
+    },
+    {
+      "epoch": 1.5625717566016073,
+      "grad_norm": 2.4292559426792426,
+      "learning_rate": 5.973563402170799e-06,
+      "loss": 0.2441,
+      "mean_token_accuracy": 0.933300256729126,
+      "num_tokens": 5775228.0,
+      "step": 681
+    },
+    {
+      "epoch": 1.5648679678530426,
+      "grad_norm": 2.216591077475601,
+      "learning_rate": 5.961617237771217e-06,
+      "loss": 0.2206,
+      "mean_token_accuracy": 0.9310642778873444,
+      "num_tokens": 5784167.0,
+      "step": 682
+    },
+    {
+      "epoch": 1.5671641791044775,
+      "grad_norm": 2.3837512015665925,
+      "learning_rate": 5.949667784638909e-06,
+      "loss": 0.2451,
+      "mean_token_accuracy": 0.9270901083946228,
+      "num_tokens": 5792242.0,
+      "step": 683
+    },
+    {
+      "epoch": 1.5694603903559128,
+      "grad_norm": 2.430677800181167,
+      "learning_rate": 5.9377151279062205e-06,
+      "loss": 0.2282,
+      "mean_token_accuracy": 0.9373931288719177,
+      "num_tokens": 5800362.0,
+      "step": 684
+    },
+    {
+      "epoch": 1.5717566016073479,
+      "grad_norm": 2.2016307662162236,
+      "learning_rate": 5.925759352728322e-06,
+      "loss": 0.2602,
+      "mean_token_accuracy": 0.9265855550765991,
+      "num_tokens": 5809137.0,
+      "step": 685
+    },
+    {
+      "epoch": 1.574052812858783,
+      "grad_norm": 2.3915401376123278,
+      "learning_rate": 5.913800544282593e-06,
+      "loss": 0.2464,
+      "mean_token_accuracy": 0.9327437579631805,
+      "num_tokens": 5816997.0,
+      "step": 686
+    },
+    {
+      "epoch": 1.5763490241102183,
+      "grad_norm": 2.420796364903747,
+      "learning_rate": 5.901838787768033e-06,
+      "loss": 0.2322,
+      "mean_token_accuracy": 0.9331295192241669,
+      "num_tokens": 5826306.0,
+      "step": 687
+    },
+    {
+      "epoch": 1.5786452353616531,
+      "grad_norm": 2.336329520491462,
+      "learning_rate": 5.889874168404635e-06,
+      "loss": 0.2141,
+      "mean_token_accuracy": 0.9396253228187561,
+      "num_tokens": 5834659.0,
+      "step": 688
+    },
+    {
+      "epoch": 1.5809414466130884,
+      "grad_norm": 2.109666385002501,
+      "learning_rate": 5.8779067714327984e-06,
+      "loss": 0.2766,
+      "mean_token_accuracy": 0.9179757833480835,
+      "num_tokens": 5843104.0,
+      "step": 689
+    },
+    {
+      "epoch": 1.5832376578645235,
+      "grad_norm": 2.2267645534274876,
+      "learning_rate": 5.865936682112702e-06,
+      "loss": 0.2177,
+      "mean_token_accuracy": 0.9360098540782928,
+      "num_tokens": 5851456.0,
+      "step": 690
+    },
+    {
+      "epoch": 1.5855338691159586,
+      "grad_norm": 2.284530138119213,
+      "learning_rate": 5.853963985723713e-06,
+      "loss": 0.23,
+      "mean_token_accuracy": 0.9347943663597107,
+      "num_tokens": 5859686.0,
+      "step": 691
+    },
+    {
+      "epoch": 1.587830080367394,
+      "grad_norm": 2.460668903549717,
+      "learning_rate": 5.841988767563766e-06,
+      "loss": 0.2468,
+      "mean_token_accuracy": 0.9305404424667358,
+      "num_tokens": 5867717.0,
+      "step": 692
+    },
+    {
+      "epoch": 1.5901262916188288,
+      "grad_norm": 2.1628760690387496,
+      "learning_rate": 5.830011112948768e-06,
+      "loss": 0.2326,
+      "mean_token_accuracy": 0.9327261447906494,
+      "num_tokens": 5875847.0,
+      "step": 693
+    },
+    {
+      "epoch": 1.592422502870264,
+      "grad_norm": 2.2868832625220854,
+      "learning_rate": 5.81803110721198e-06,
+      "loss": 0.2325,
+      "mean_token_accuracy": 0.9338472187519073,
+      "num_tokens": 5884108.0,
+      "step": 694
+    },
+    {
+      "epoch": 1.5947187141216992,
+      "grad_norm": 2.3340846971919023,
+      "learning_rate": 5.806048835703412e-06,
+      "loss": 0.2723,
+      "mean_token_accuracy": 0.9252812266349792,
+      "num_tokens": 5892661.0,
+      "step": 695
+    },
+    {
+      "epoch": 1.5970149253731343,
+      "grad_norm": 2.649839523714754,
+      "learning_rate": 5.794064383789221e-06,
+      "loss": 0.2445,
+      "mean_token_accuracy": 0.9307736158370972,
+      "num_tokens": 5901007.0,
+      "step": 696
+    },
+    {
+      "epoch": 1.5993111366245696,
+      "grad_norm": 2.190417259997277,
+      "learning_rate": 5.782077836851094e-06,
+      "loss": 0.2167,
+      "mean_token_accuracy": 0.9389541745185852,
+      "num_tokens": 5909910.0,
+      "step": 697
+    },
+    {
+      "epoch": 1.6016073478760045,
+      "grad_norm": 2.1677256486102445,
+      "learning_rate": 5.770089280285647e-06,
+      "loss": 0.2369,
+      "mean_token_accuracy": 0.9284799695014954,
+      "num_tokens": 5918678.0,
+      "step": 698
+    },
+    {
+      "epoch": 1.6039035591274398,
+      "grad_norm": 2.2135208242005575,
+      "learning_rate": 5.758098799503809e-06,
+      "loss": 0.2248,
+      "mean_token_accuracy": 0.9350206255912781,
+      "num_tokens": 5926972.0,
+      "step": 699
+    },
+    {
+      "epoch": 1.6061997703788748,
+      "grad_norm": 2.862622514627209,
+      "learning_rate": 5.746106479930223e-06,
+      "loss": 0.2893,
+      "mean_token_accuracy": 0.9198987185955048,
+      "num_tokens": 5935118.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.60849598163031,
+      "grad_norm": 2.114266388477335,
+      "learning_rate": 5.734112407002626e-06,
+      "loss": 0.207,
+      "mean_token_accuracy": 0.9383877813816071,
+      "num_tokens": 5943913.0,
+      "step": 701
+    },
+    {
+      "epoch": 1.6107921928817452,
+      "grad_norm": 2.4190126165386063,
+      "learning_rate": 5.722116666171256e-06,
+      "loss": 0.2385,
+      "mean_token_accuracy": 0.9313597679138184,
+      "num_tokens": 5951827.0,
+      "step": 702
+    },
+    {
+      "epoch": 1.61308840413318,
+      "grad_norm": 2.5545889850085417,
+      "learning_rate": 5.7101193428982215e-06,
+      "loss": 0.2498,
+      "mean_token_accuracy": 0.926564484834671,
+      "num_tokens": 5960441.0,
+      "step": 703
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 2.0974717436935415,
+      "learning_rate": 5.698120522656916e-06,
+      "loss": 0.2375,
+      "mean_token_accuracy": 0.9322009086608887,
+      "num_tokens": 5968431.0,
+      "step": 704
+    },
+    {
+      "epoch": 1.6176808266360505,
+      "grad_norm": 2.4014139218026567,
+      "learning_rate": 5.686120290931392e-06,
+      "loss": 0.2349,
+      "mean_token_accuracy": 0.9332285225391388,
+      "num_tokens": 5976728.0,
+      "step": 705
+    },
+    {
+      "epoch": 1.6199770378874856,
+      "grad_norm": 2.1939210403974037,
+      "learning_rate": 5.674118733215757e-06,
+      "loss": 0.2383,
+      "mean_token_accuracy": 0.9336570501327515,
+      "num_tokens": 5985272.0,
+      "step": 706
+    },
+    {
+      "epoch": 1.622273249138921,
+      "grad_norm": 2.3685556352482386,
+      "learning_rate": 5.662115935013572e-06,
+      "loss": 0.2371,
+      "mean_token_accuracy": 0.9298917353153229,
+      "num_tokens": 5994724.0,
+      "step": 707
+    },
+    {
+      "epoch": 1.6245694603903558,
+      "grad_norm": 2.120532260181221,
+      "learning_rate": 5.6501119818372265e-06,
+      "loss": 0.205,
+      "mean_token_accuracy": 0.9388197660446167,
+      "num_tokens": 6003002.0,
+      "step": 708
+    },
+    {
+      "epoch": 1.626865671641791,
+      "grad_norm": 2.2009808663286776,
+      "learning_rate": 5.638106959207347e-06,
+      "loss": 0.2077,
+      "mean_token_accuracy": 0.9382572770118713,
+      "num_tokens": 6011637.0,
+      "step": 709
+    },
+    {
+      "epoch": 1.6291618828932262,
+      "grad_norm": 2.4039829896242755,
+      "learning_rate": 5.626100952652172e-06,
+      "loss": 0.2331,
+      "mean_token_accuracy": 0.935380220413208,
+      "num_tokens": 6020521.0,
+      "step": 710
+    },
+    {
+      "epoch": 1.6314580941446613,
+      "grad_norm": 2.2135500590512955,
+      "learning_rate": 5.614094047706956e-06,
+      "loss": 0.2572,
+      "mean_token_accuracy": 0.9282229840755463,
+      "num_tokens": 6028619.0,
+      "step": 711
+    },
+    {
+      "epoch": 1.6337543053960966,
+      "grad_norm": 2.4906751846229263,
+      "learning_rate": 5.60208632991335e-06,
+      "loss": 0.2245,
+      "mean_token_accuracy": 0.9348632991313934,
+      "num_tokens": 6037038.0,
+      "step": 712
+    },
+    {
+      "epoch": 1.6360505166475314,
+      "grad_norm": 2.1877988052188972,
+      "learning_rate": 5.590077884818798e-06,
+      "loss": 0.2317,
+      "mean_token_accuracy": 0.9365200996398926,
+      "num_tokens": 6045467.0,
+      "step": 713
+    },
+    {
+      "epoch": 1.6383467278989667,
+      "grad_norm": 2.2429864529311936,
+      "learning_rate": 5.578068797975924e-06,
+      "loss": 0.2345,
+      "mean_token_accuracy": 0.9305262565612793,
+      "num_tokens": 6054256.0,
+      "step": 714
+    },
+    {
+      "epoch": 1.6406429391504018,
+      "grad_norm": 2.2259834828031027,
+      "learning_rate": 5.566059154941925e-06,
+      "loss": 0.2266,
+      "mean_token_accuracy": 0.9362088739871979,
+      "num_tokens": 6062664.0,
+      "step": 715
+    },
+    {
+      "epoch": 1.642939150401837,
+      "grad_norm": 2.1651817839836975,
+      "learning_rate": 5.554049041277962e-06,
+      "loss": 0.2171,
+      "mean_token_accuracy": 0.9348065257072449,
+      "num_tokens": 6071257.0,
+      "step": 716
+    },
+    {
+      "epoch": 1.6452353616532722,
+      "grad_norm": 2.2679285834459355,
+      "learning_rate": 5.542038542548544e-06,
+      "loss": 0.2717,
+      "mean_token_accuracy": 0.9239723086357117,
+      "num_tokens": 6079433.0,
+      "step": 717
+    },
+    {
+      "epoch": 1.647531572904707,
+      "grad_norm": 2.559693253947525,
+      "learning_rate": 5.530027744320931e-06,
+      "loss": 0.2579,
+      "mean_token_accuracy": 0.9282408654689789,
+      "num_tokens": 6087621.0,
+      "step": 718
+    },
+    {
+      "epoch": 1.6498277841561424,
+      "grad_norm": 2.4906356767702875,
+      "learning_rate": 5.518016732164509e-06,
+      "loss": 0.1909,
+      "mean_token_accuracy": 0.941920816898346,
+      "num_tokens": 6096038.0,
+      "step": 719
+    },
+    {
+      "epoch": 1.6521239954075775,
+      "grad_norm": 2.0508616314639525,
+      "learning_rate": 5.5060055916501934e-06,
+      "loss": 0.2446,
+      "mean_token_accuracy": 0.9322466850280762,
+      "num_tokens": 6105425.0,
+      "step": 720
+    },
+    {
+      "epoch": 1.6544202066590126,
+      "grad_norm": 2.2785063893389084,
+      "learning_rate": 5.493994408349809e-06,
+      "loss": 0.2124,
+      "mean_token_accuracy": 0.9360540509223938,
+      "num_tokens": 6114302.0,
+      "step": 721
+    },
+    {
+      "epoch": 1.6567164179104479,
+      "grad_norm": 2.19781534237417,
+      "learning_rate": 5.481983267835492e-06,
+      "loss": 0.2154,
+      "mean_token_accuracy": 0.9386440515518188,
+      "num_tokens": 6122854.0,
+      "step": 722
+    },
+    {
+      "epoch": 1.6590126291618827,
+      "grad_norm": 2.1702752200859874,
+      "learning_rate": 5.469972255679071e-06,
+      "loss": 0.235,
+      "mean_token_accuracy": 0.93433478474617,
+      "num_tokens": 6131509.0,
+      "step": 723
+    },
+    {
+      "epoch": 1.661308840413318,
+      "grad_norm": 2.3595119568385825,
+      "learning_rate": 5.457961457451457e-06,
+      "loss": 0.2808,
+      "mean_token_accuracy": 0.9232488870620728,
+      "num_tokens": 6140441.0,
+      "step": 724
+    },
+    {
+      "epoch": 1.6636050516647531,
+      "grad_norm": 2.5503912344504243,
+      "learning_rate": 5.445950958722039e-06,
+      "loss": 0.2232,
+      "mean_token_accuracy": 0.9333747625350952,
+      "num_tokens": 6148438.0,
+      "step": 725
+    },
+    {
+      "epoch": 1.6659012629161882,
+      "grad_norm": 2.4080777038162076,
+      "learning_rate": 5.433940845058076e-06,
+      "loss": 0.2676,
+      "mean_token_accuracy": 0.9238480925559998,
+      "num_tokens": 6157068.0,
+      "step": 726
+    },
+    {
+      "epoch": 1.6681974741676235,
+      "grad_norm": 2.507446324249632,
+      "learning_rate": 5.421931202024078e-06,
+      "loss": 0.2176,
+      "mean_token_accuracy": 0.939238578081131,
+      "num_tokens": 6165358.0,
+      "step": 727
+    },
+    {
+      "epoch": 1.6704936854190584,
+      "grad_norm": 2.2583190315574355,
+      "learning_rate": 5.409922115181203e-06,
+      "loss": 0.2058,
+      "mean_token_accuracy": 0.9437562525272369,
+      "num_tokens": 6174108.0,
+      "step": 728
+    },
+    {
+      "epoch": 1.6727898966704937,
+      "grad_norm": 2.189721869532295,
+      "learning_rate": 5.397913670086651e-06,
+      "loss": 0.2491,
+      "mean_token_accuracy": 0.9319724440574646,
+      "num_tokens": 6182145.0,
+      "step": 729
+    },
+    {
+      "epoch": 1.6750861079219288,
+      "grad_norm": 2.475419387972065,
+      "learning_rate": 5.385905952293047e-06,
+      "loss": 0.2468,
+      "mean_token_accuracy": 0.9323441684246063,
+      "num_tokens": 6190647.0,
+      "step": 730
+    },
+    {
+      "epoch": 1.677382319173364,
+      "grad_norm": 2.2412699378808694,
+      "learning_rate": 5.373899047347831e-06,
+      "loss": 0.2122,
+      "mean_token_accuracy": 0.9362380802631378,
+      "num_tokens": 6200677.0,
+      "step": 731
+    },
+    {
+      "epoch": 1.6796785304247992,
+      "grad_norm": 2.0425012248129306,
+      "learning_rate": 5.361893040792655e-06,
+      "loss": 0.2128,
+      "mean_token_accuracy": 0.9367002546787262,
+      "num_tokens": 6208464.0,
+      "step": 732
+    },
+    {
+      "epoch": 1.681974741676234,
+      "grad_norm": 2.5352162577746746,
+      "learning_rate": 5.349888018162776e-06,
+      "loss": 0.2533,
+      "mean_token_accuracy": 0.9277266263961792,
+      "num_tokens": 6216854.0,
+      "step": 733
+    },
+    {
+      "epoch": 1.6842709529276694,
+      "grad_norm": 2.2007819761272143,
+      "learning_rate": 5.337884064986432e-06,
+      "loss": 0.2615,
+      "mean_token_accuracy": 0.9255687892436981,
+      "num_tokens": 6226248.0,
+      "step": 734
+    },
+    {
+      "epoch": 1.6865671641791045,
+      "grad_norm": 2.322532378731532,
+      "learning_rate": 5.325881266784244e-06,
+      "loss": 0.2085,
+      "mean_token_accuracy": 0.9423520267009735,
+      "num_tokens": 6234369.0,
+      "step": 735
+    },
+    {
+      "epoch": 1.6888633754305395,
+      "grad_norm": 2.160674719506299,
+      "learning_rate": 5.31387970906861e-06,
+      "loss": 0.232,
+      "mean_token_accuracy": 0.9313159584999084,
+      "num_tokens": 6243537.0,
+      "step": 736
+    },
+    {
+      "epoch": 1.6911595866819749,
+      "grad_norm": 2.507440453364873,
+      "learning_rate": 5.301879477343086e-06,
+      "loss": 0.2326,
+      "mean_token_accuracy": 0.9337201416492462,
+      "num_tokens": 6251542.0,
+      "step": 737
+    },
+    {
+      "epoch": 1.6934557979334097,
+      "grad_norm": 2.1729168140986364,
+      "learning_rate": 5.28988065710178e-06,
+      "loss": 0.2413,
+      "mean_token_accuracy": 0.9328451454639435,
+      "num_tokens": 6259766.0,
+      "step": 738
+    },
+    {
+      "epoch": 1.695752009184845,
+      "grad_norm": 2.5963673491099435,
+      "learning_rate": 5.277883333828746e-06,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.934342622756958,
+      "num_tokens": 6268162.0,
+      "step": 739
+    },
+    {
+      "epoch": 1.6980482204362801,
+      "grad_norm": 2.3210125978680636,
+      "learning_rate": 5.265887592997374e-06,
+      "loss": 0.2497,
+      "mean_token_accuracy": 0.9327953159809113,
+      "num_tokens": 6276282.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.7003444316877152,
+      "grad_norm": 2.5658394258161863,
+      "learning_rate": 5.25389352006978e-06,
+      "loss": 0.2417,
+      "mean_token_accuracy": 0.9310867190361023,
+      "num_tokens": 6284746.0,
+      "step": 741
+    },
+    {
+      "epoch": 1.7026406429391505,
+      "grad_norm": 2.3730652658413773,
+      "learning_rate": 5.2419012004961934e-06,
+      "loss": 0.1863,
+      "mean_token_accuracy": 0.9452896118164062,
+      "num_tokens": 6293312.0,
+      "step": 742
+    },
+    {
+      "epoch": 1.7049368541905854,
+      "grad_norm": 2.2746925597638907,
+      "learning_rate": 5.229910719714355e-06,
+      "loss": 0.2441,
+      "mean_token_accuracy": 0.9340402185916901,
+      "num_tokens": 6302395.0,
+      "step": 743
+    },
+    {
+      "epoch": 1.7072330654420207,
+      "grad_norm": 2.3794985429816338,
+      "learning_rate": 5.217922163148907e-06,
+      "loss": 0.2899,
+      "mean_token_accuracy": 0.917595386505127,
+      "num_tokens": 6311452.0,
+      "step": 744
+    },
+    {
+      "epoch": 1.7095292766934558,
+      "grad_norm": 2.4799352065771703,
+      "learning_rate": 5.205935616210781e-06,
+      "loss": 0.2035,
+      "mean_token_accuracy": 0.9438226521015167,
+      "num_tokens": 6320170.0,
+      "step": 745
+    },
+    {
+      "epoch": 1.7118254879448909,
+      "grad_norm": 2.2821146360369418,
+      "learning_rate": 5.19395116429659e-06,
+      "loss": 0.2422,
+      "mean_token_accuracy": 0.9317375719547272,
+      "num_tokens": 6328845.0,
+      "step": 746
+    },
+    {
+      "epoch": 1.7141216991963262,
+      "grad_norm": 2.4648820808432834,
+      "learning_rate": 5.181968892788022e-06,
+      "loss": 0.2122,
+      "mean_token_accuracy": 0.9389127194881439,
+      "num_tokens": 6337142.0,
+      "step": 747
+    },
+    {
+      "epoch": 1.716417910447761,
+      "grad_norm": 2.4096746760717735,
+      "learning_rate": 5.169988887051234e-06,
+      "loss": 0.2838,
+      "mean_token_accuracy": 0.9197812676429749,
+      "num_tokens": 6345742.0,
+      "step": 748
+    },
+    {
+      "epoch": 1.7187141216991964,
+      "grad_norm": 2.223124475136241,
+      "learning_rate": 5.158011232436235e-06,
+      "loss": 0.2291,
+      "mean_token_accuracy": 0.9339918494224548,
+      "num_tokens": 6353488.0,
+      "step": 749
+    },
+    {
+      "epoch": 1.7210103329506314,
+      "grad_norm": 2.210280464131307,
+      "learning_rate": 5.146036014276288e-06,
+      "loss": 0.2261,
+      "mean_token_accuracy": 0.9387891590595245,
+      "num_tokens": 6362328.0,
+      "step": 750
+    },
+    {
+      "epoch": 1.7233065442020665,
+      "grad_norm": 2.15542657277757,
+      "learning_rate": 5.134063317887299e-06,
+      "loss": 0.276,
+      "mean_token_accuracy": 0.9262830913066864,
+      "num_tokens": 6371245.0,
+      "step": 751
+    },
+    {
+      "epoch": 1.7256027554535018,
+      "grad_norm": 2.349265821801737,
+      "learning_rate": 5.122093228567205e-06,
+      "loss": 0.2318,
+      "mean_token_accuracy": 0.9352737367153168,
+      "num_tokens": 6379783.0,
+      "step": 752
+    },
+    {
+      "epoch": 1.7278989667049367,
+      "grad_norm": 2.1503287591924325,
+      "learning_rate": 5.110125831595367e-06,
+      "loss": 0.2229,
+      "mean_token_accuracy": 0.9350302815437317,
+      "num_tokens": 6388280.0,
+      "step": 753
+    },
+    {
+      "epoch": 1.730195177956372,
+      "grad_norm": 2.3093960352072926,
+      "learning_rate": 5.09816121223197e-06,
+      "loss": 0.1903,
+      "mean_token_accuracy": 0.9428891837596893,
+      "num_tokens": 6396931.0,
+      "step": 754
+    },
+    {
+      "epoch": 1.732491389207807,
+      "grad_norm": 2.020517469445219,
+      "learning_rate": 5.086199455717409e-06,
+      "loss": 0.2177,
+      "mean_token_accuracy": 0.9361197352409363,
+      "num_tokens": 6405471.0,
+      "step": 755
+    },
+    {
+      "epoch": 1.7347876004592422,
+      "grad_norm": 2.564031970693843,
+      "learning_rate": 5.074240647271681e-06,
+      "loss": 0.2393,
+      "mean_token_accuracy": 0.9344854354858398,
+      "num_tokens": 6413162.0,
+      "step": 756
+    },
+    {
+      "epoch": 1.7370838117106775,
+      "grad_norm": 2.4005072305139525,
+      "learning_rate": 5.06228487209378e-06,
+      "loss": 0.2362,
+      "mean_token_accuracy": 0.9323394000530243,
+      "num_tokens": 6421112.0,
+      "step": 757
+    },
+    {
+      "epoch": 1.7393800229621124,
+      "grad_norm": 2.6994683151203063,
+      "learning_rate": 5.050332215361092e-06,
+      "loss": 0.2351,
+      "mean_token_accuracy": 0.9336782097816467,
+      "num_tokens": 6429961.0,
+      "step": 758
+    },
+    {
+      "epoch": 1.7416762342135477,
+      "grad_norm": 2.299223671338463,
+      "learning_rate": 5.038382762228786e-06,
+      "loss": 0.2352,
+      "mean_token_accuracy": 0.9312534630298615,
+      "num_tokens": 6438510.0,
+      "step": 759
+    },
+    {
+      "epoch": 1.7439724454649828,
+      "grad_norm": 2.1687411276914683,
+      "learning_rate": 5.026436597829203e-06,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9482762813568115,
+      "num_tokens": 6448428.0,
+      "step": 760
+    },
+    {
+      "epoch": 1.7462686567164178,
+      "grad_norm": 2.351868033383212,
+      "learning_rate": 5.014493807271258e-06,
+      "loss": 0.2214,
+      "mean_token_accuracy": 0.9378836154937744,
+      "num_tokens": 6457377.0,
+      "step": 761
+    },
+    {
+      "epoch": 1.7485648679678532,
+      "grad_norm": 2.3469192423086325,
+      "learning_rate": 5.00255447563983e-06,
+      "loss": 0.2164,
+      "mean_token_accuracy": 0.9377240240573883,
+      "num_tokens": 6465011.0,
+      "step": 762
+    },
+    {
+      "epoch": 1.750861079219288,
+      "grad_norm": 2.386032235181974,
+      "learning_rate": 4.9906186879951534e-06,
+      "loss": 0.209,
+      "mean_token_accuracy": 0.937764048576355,
+      "num_tokens": 6474178.0,
+      "step": 763
+    },
+    {
+      "epoch": 1.7531572904707233,
+      "grad_norm": 2.4532547848529473,
+      "learning_rate": 4.978686529372213e-06,
+      "loss": 0.2324,
+      "mean_token_accuracy": 0.9330492913722992,
+      "num_tokens": 6482478.0,
+      "step": 764
+    },
+    {
+      "epoch": 1.7554535017221584,
+      "grad_norm": 2.254845320802166,
+      "learning_rate": 4.966758084780142e-06,
+      "loss": 0.1904,
+      "mean_token_accuracy": 0.9429523646831512,
+      "num_tokens": 6490529.0,
+      "step": 765
+    },
+    {
+      "epoch": 1.7577497129735935,
+      "grad_norm": 2.195279377916472,
+      "learning_rate": 4.954833439201611e-06,
+      "loss": 0.224,
+      "mean_token_accuracy": 0.9370987415313721,
+      "num_tokens": 6498775.0,
+      "step": 766
+    },
+    {
+      "epoch": 1.7600459242250288,
+      "grad_norm": 2.27356158810111,
+      "learning_rate": 4.942912677592227e-06,
+      "loss": 0.2393,
+      "mean_token_accuracy": 0.9322682917118073,
+      "num_tokens": 6506733.0,
+      "step": 767
+    },
+    {
+      "epoch": 1.7623421354764637,
+      "grad_norm": 2.5582359134764303,
+      "learning_rate": 4.930995884879923e-06,
+      "loss": 0.2486,
+      "mean_token_accuracy": 0.9312199950218201,
+      "num_tokens": 6515359.0,
+      "step": 768
+    },
+    {
+      "epoch": 1.764638346727899,
+      "grad_norm": 2.190935692070036,
+      "learning_rate": 4.91908314596436e-06,
+      "loss": 0.1989,
+      "mean_token_accuracy": 0.9429928958415985,
+      "num_tokens": 6523621.0,
+      "step": 769
+    },
+    {
+      "epoch": 1.766934557979334,
+      "grad_norm": 2.263241194839795,
+      "learning_rate": 4.907174545716317e-06,
+      "loss": 0.2134,
+      "mean_token_accuracy": 0.9392730295658112,
+      "num_tokens": 6531637.0,
+      "step": 770
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 2.621859717942934,
+      "learning_rate": 4.895270168977085e-06,
+      "loss": 0.2625,
+      "mean_token_accuracy": 0.925828754901886,
+      "num_tokens": 6539890.0,
+      "step": 771
+    },
+    {
+      "epoch": 1.7715269804822045,
+      "grad_norm": 2.185436482225333,
+      "learning_rate": 4.883370100557867e-06,
+      "loss": 0.2036,
+      "mean_token_accuracy": 0.9410973787307739,
+      "num_tokens": 6547865.0,
+      "step": 772
+    },
+    {
+      "epoch": 1.7738231917336393,
+      "grad_norm": 2.1482187970710847,
+      "learning_rate": 4.871474425239175e-06,
+      "loss": 0.2269,
+      "mean_token_accuracy": 0.9340025186538696,
+      "num_tokens": 6556631.0,
+      "step": 773
+    },
+    {
+      "epoch": 1.7761194029850746,
+      "grad_norm": 2.1955661449691415,
+      "learning_rate": 4.859583227770218e-06,
+      "loss": 0.1906,
+      "mean_token_accuracy": 0.9485273361206055,
+      "num_tokens": 6564958.0,
+      "step": 774
+    },
+    {
+      "epoch": 1.7784156142365097,
+      "grad_norm": 2.4043442117535556,
+      "learning_rate": 4.847696592868305e-06,
+      "loss": 0.2172,
+      "mean_token_accuracy": 0.940415620803833,
+      "num_tokens": 6572993.0,
+      "step": 775
+    },
+    {
+      "epoch": 1.7807118254879448,
+      "grad_norm": 2.1946954734481015,
+      "learning_rate": 4.83581460521824e-06,
+      "loss": 0.2214,
+      "mean_token_accuracy": 0.9348120093345642,
+      "num_tokens": 6581156.0,
+      "step": 776
+    },
+    {
+      "epoch": 1.7830080367393801,
+      "grad_norm": 2.4120721552185724,
+      "learning_rate": 4.823937349471719e-06,
+      "loss": 0.1921,
+      "mean_token_accuracy": 0.9444971084594727,
+      "num_tokens": 6590441.0,
+      "step": 777
+    },
+    {
+      "epoch": 1.785304247990815,
+      "grad_norm": 2.3721292255713737,
+      "learning_rate": 4.812064910246725e-06,
+      "loss": 0.179,
+      "mean_token_accuracy": 0.9430033564567566,
+      "num_tokens": 6598213.0,
+      "step": 778
+    },
+    {
+      "epoch": 1.7876004592422503,
+      "grad_norm": 2.2780602668386014,
+      "learning_rate": 4.800197372126925e-06,
+      "loss": 0.2087,
+      "mean_token_accuracy": 0.9378463625907898,
+      "num_tokens": 6606101.0,
+      "step": 779
+    },
+    {
+      "epoch": 1.7898966704936854,
+      "grad_norm": 2.4102535330791053,
+      "learning_rate": 4.78833481966107e-06,
+      "loss": 0.2074,
+      "mean_token_accuracy": 0.9403178691864014,
+      "num_tokens": 6614692.0,
+      "step": 780
+    },
+    {
+      "epoch": 1.7921928817451205,
+      "grad_norm": 2.2528895725947318,
+      "learning_rate": 4.776477337362394e-06,
+      "loss": 0.2041,
+      "mean_token_accuracy": 0.9398922324180603,
+      "num_tokens": 6622962.0,
+      "step": 781
+    },
+    {
+      "epoch": 1.7944890929965558,
+      "grad_norm": 2.762813115978806,
+      "learning_rate": 4.764625009708005e-06,
+      "loss": 0.2512,
+      "mean_token_accuracy": 0.9283100962638855,
+      "num_tokens": 6630869.0,
+      "step": 782
+    },
+    {
+      "epoch": 1.7967853042479907,
+      "grad_norm": 2.2806198835474847,
+      "learning_rate": 4.7527779211382895e-06,
+      "loss": 0.2309,
+      "mean_token_accuracy": 0.9319865703582764,
+      "num_tokens": 6639358.0,
+      "step": 783
+    },
+    {
+      "epoch": 1.799081515499426,
+      "grad_norm": 2.487206309074608,
+      "learning_rate": 4.7409361560563074e-06,
+      "loss": 0.2365,
+      "mean_token_accuracy": 0.9308879673480988,
+      "num_tokens": 6647782.0,
+      "step": 784
+    },
+    {
+      "epoch": 1.801377726750861,
+      "grad_norm": 2.7084257222262673,
+      "learning_rate": 4.729099798827196e-06,
+      "loss": 0.2235,
+      "mean_token_accuracy": 0.9362497329711914,
+      "num_tokens": 6656075.0,
+      "step": 785
+    },
+    {
+      "epoch": 1.8036739380022961,
+      "grad_norm": 1.9886557902757762,
+      "learning_rate": 4.71726893377756e-06,
+      "loss": 0.1954,
+      "mean_token_accuracy": 0.942346841096878,
+      "num_tokens": 6665191.0,
+      "step": 786
+    },
+    {
+      "epoch": 1.8059701492537314,
+      "grad_norm": 2.3444844971258165,
+      "learning_rate": 4.705443645194877e-06,
+      "loss": 0.2016,
+      "mean_token_accuracy": 0.942473441362381,
+      "num_tokens": 6673227.0,
+      "step": 787
+    },
+    {
+      "epoch": 1.8082663605051663,
+      "grad_norm": 2.2082597878187125,
+      "learning_rate": 4.6936240173269e-06,
+      "loss": 0.2133,
+      "mean_token_accuracy": 0.9384743571281433,
+      "num_tokens": 6681344.0,
+      "step": 788
+    },
+    {
+      "epoch": 1.8105625717566016,
+      "grad_norm": 1.9835899508362258,
+      "learning_rate": 4.681810134381048e-06,
+      "loss": 0.1865,
+      "mean_token_accuracy": 0.9445423185825348,
+      "num_tokens": 6689995.0,
+      "step": 789
+    },
+    {
+      "epoch": 1.8128587830080367,
+      "grad_norm": 2.116896958743143,
+      "learning_rate": 4.670002080523812e-06,
+      "loss": 0.2224,
+      "mean_token_accuracy": 0.9393620789051056,
+      "num_tokens": 6697809.0,
+      "step": 790
+    },
+    {
+      "epoch": 1.8151549942594718,
+      "grad_norm": 2.460310025725059,
+      "learning_rate": 4.658199939880156e-06,
+      "loss": 0.2004,
+      "mean_token_accuracy": 0.9412027597427368,
+      "num_tokens": 6706319.0,
+      "step": 791
+    },
+    {
+      "epoch": 1.817451205510907,
+      "grad_norm": 2.4180044193776604,
+      "learning_rate": 4.646403796532916e-06,
+      "loss": 0.2414,
+      "mean_token_accuracy": 0.9293927848339081,
+      "num_tokens": 6715236.0,
+      "step": 792
+    },
+    {
+      "epoch": 1.819747416762342,
+      "grad_norm": 2.378344102856168,
+      "learning_rate": 4.634613734522199e-06,
+      "loss": 0.2247,
+      "mean_token_accuracy": 0.9380782246589661,
+      "num_tokens": 6723509.0,
+      "step": 793
+    },
+    {
+      "epoch": 1.8220436280137773,
+      "grad_norm": 2.125538103444689,
+      "learning_rate": 4.622829837844788e-06,
+      "loss": 0.1848,
+      "mean_token_accuracy": 0.9444566667079926,
+      "num_tokens": 6731496.0,
+      "step": 794
+    },
+    {
+      "epoch": 1.8243398392652124,
+      "grad_norm": 2.3215181463516914,
+      "learning_rate": 4.611052190453541e-06,
+      "loss": 0.1802,
+      "mean_token_accuracy": 0.9499981999397278,
+      "num_tokens": 6739703.0,
+      "step": 795
+    },
+    {
+      "epoch": 1.8266360505166475,
+      "grad_norm": 2.2411122905427208,
+      "learning_rate": 4.5992808762567965e-06,
+      "loss": 0.2157,
+      "mean_token_accuracy": 0.9376723170280457,
+      "num_tokens": 6748184.0,
+      "step": 796
+    },
+    {
+      "epoch": 1.8289322617680828,
+      "grad_norm": 2.7382011800047095,
+      "learning_rate": 4.587515979117768e-06,
+      "loss": 0.2242,
+      "mean_token_accuracy": 0.9376608431339264,
+      "num_tokens": 6756518.0,
+      "step": 797
+    },
+    {
+      "epoch": 1.8312284730195176,
+      "grad_norm": 2.385663822321377,
+      "learning_rate": 4.575757582853955e-06,
+      "loss": 0.2433,
+      "mean_token_accuracy": 0.9314123094081879,
+      "num_tokens": 6765180.0,
+      "step": 798
+    },
+    {
+      "epoch": 1.833524684270953,
+      "grad_norm": 2.601940816436594,
+      "learning_rate": 4.564005771236541e-06,
+      "loss": 0.2045,
+      "mean_token_accuracy": 0.9398826062679291,
+      "num_tokens": 6773491.0,
+      "step": 799
+    },
+    {
+      "epoch": 1.835820895522388,
+      "grad_norm": 2.521847455331403,
+      "learning_rate": 4.552260627989802e-06,
+      "loss": 0.2033,
+      "mean_token_accuracy": 0.9426309168338776,
+      "num_tokens": 6782264.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.8381171067738231,
+      "grad_norm": 2.3933121616470587,
+      "learning_rate": 4.5405222367904975e-06,
+      "loss": 0.2183,
+      "mean_token_accuracy": 0.9412200748920441,
+      "num_tokens": 6790433.0,
+      "step": 801
+    },
+    {
+      "epoch": 1.8404133180252584,
+      "grad_norm": 2.36868692829037,
+      "learning_rate": 4.52879068126729e-06,
+      "loss": 0.2083,
+      "mean_token_accuracy": 0.9384964406490326,
+      "num_tokens": 6798827.0,
+      "step": 802
+    },
+    {
+      "epoch": 1.8427095292766933,
+      "grad_norm": 2.140815456385186,
+      "learning_rate": 4.517066045000142e-06,
+      "loss": 0.1929,
+      "mean_token_accuracy": 0.9435328245162964,
+      "num_tokens": 6806622.0,
+      "step": 803
+    },
+    {
+      "epoch": 1.8450057405281286,
+      "grad_norm": 2.030117137754188,
+      "learning_rate": 4.505348411519718e-06,
+      "loss": 0.2539,
+      "mean_token_accuracy": 0.9296467304229736,
+      "num_tokens": 6815667.0,
+      "step": 804
+    },
+    {
+      "epoch": 1.8473019517795637,
+      "grad_norm": 2.5439214459207014,
+      "learning_rate": 4.49363786430679e-06,
+      "loss": 0.2072,
+      "mean_token_accuracy": 0.9390068054199219,
+      "num_tokens": 6824279.0,
+      "step": 805
+    },
+    {
+      "epoch": 1.8495981630309988,
+      "grad_norm": 2.22732424486158,
+      "learning_rate": 4.481934486791651e-06,
+      "loss": 0.2034,
+      "mean_token_accuracy": 0.9410427212715149,
+      "num_tokens": 6832824.0,
+      "step": 806
+    },
+    {
+      "epoch": 1.851894374282434,
+      "grad_norm": 2.1834140463215865,
+      "learning_rate": 4.470238362353511e-06,
+      "loss": 0.2285,
+      "mean_token_accuracy": 0.9348617494106293,
+      "num_tokens": 6841358.0,
+      "step": 807
+    },
+    {
+      "epoch": 1.854190585533869,
+      "grad_norm": 2.309085082206576,
+      "learning_rate": 4.458549574319903e-06,
+      "loss": 0.187,
+      "mean_token_accuracy": 0.945079505443573,
+      "num_tokens": 6850456.0,
+      "step": 808
+    },
+    {
+      "epoch": 1.8564867967853043,
+      "grad_norm": 2.2996424197989955,
+      "learning_rate": 4.4468682059661e-06,
+      "loss": 0.2009,
+      "mean_token_accuracy": 0.943764716386795,
+      "num_tokens": 6858600.0,
+      "step": 809
+    },
+    {
+      "epoch": 1.8587830080367393,
+      "grad_norm": 2.918171320696195,
+      "learning_rate": 4.435194340514507e-06,
+      "loss": 0.2353,
+      "mean_token_accuracy": 0.9355550706386566,
+      "num_tokens": 6867383.0,
+      "step": 810
+    },
+    {
+      "epoch": 1.8610792192881744,
+      "grad_norm": 2.0199768056842067,
+      "learning_rate": 4.423528061134084e-06,
+      "loss": 0.2116,
+      "mean_token_accuracy": 0.9361621141433716,
+      "num_tokens": 6876054.0,
+      "step": 811
+    },
+    {
+      "epoch": 1.8633754305396097,
+      "grad_norm": 2.00150359948234,
+      "learning_rate": 4.4118694509397365e-06,
+      "loss": 0.2167,
+      "mean_token_accuracy": 0.9368508458137512,
+      "num_tokens": 6883982.0,
+      "step": 812
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 2.6609924089774752,
+      "learning_rate": 4.400218592991734e-06,
+      "loss": 0.2456,
+      "mean_token_accuracy": 0.9297665655612946,
+      "num_tokens": 6893183.0,
+      "step": 813
+    },
+    {
+      "epoch": 1.86796785304248,
+      "grad_norm": 2.3495169785375927,
+      "learning_rate": 4.388575570295123e-06,
+      "loss": 0.2041,
+      "mean_token_accuracy": 0.9421204924583435,
+      "num_tokens": 6901271.0,
+      "step": 814
+    },
+    {
+      "epoch": 1.870264064293915,
+      "grad_norm": 2.255465045282236,
+      "learning_rate": 4.376940465799118e-06,
+      "loss": 0.2136,
+      "mean_token_accuracy": 0.9405553042888641,
+      "num_tokens": 6909876.0,
+      "step": 815
+    },
+    {
+      "epoch": 1.87256027554535,
+      "grad_norm": 2.2784345404418227,
+      "learning_rate": 4.36531336239653e-06,
+      "loss": 0.2217,
+      "mean_token_accuracy": 0.9367011487483978,
+      "num_tokens": 6918670.0,
+      "step": 816
+    },
+    {
+      "epoch": 1.8748564867967854,
+      "grad_norm": 2.4677306325306128,
+      "learning_rate": 4.353694342923161e-06,
+      "loss": 0.2004,
+      "mean_token_accuracy": 0.943153589963913,
+      "num_tokens": 6927259.0,
+      "step": 817
+    },
+    {
+      "epoch": 1.8771526980482205,
+      "grad_norm": 2.151600803600609,
+      "learning_rate": 4.342083490157228e-06,
+      "loss": 0.225,
+      "mean_token_accuracy": 0.9363695979118347,
+      "num_tokens": 6936142.0,
+      "step": 818
+    },
+    {
+      "epoch": 1.8794489092996556,
+      "grad_norm": 2.2841669915718734,
+      "learning_rate": 4.330480886818755e-06,
+      "loss": 0.2056,
+      "mean_token_accuracy": 0.9428400099277496,
+      "num_tokens": 6944490.0,
+      "step": 819
+    },
+    {
+      "epoch": 1.8817451205510907,
+      "grad_norm": 2.548916249613302,
+      "learning_rate": 4.318886615569001e-06,
+      "loss": 0.2276,
+      "mean_token_accuracy": 0.9386773109436035,
+      "num_tokens": 6953294.0,
+      "step": 820
+    },
+    {
+      "epoch": 1.8840413318025258,
+      "grad_norm": 2.2903606555101104,
+      "learning_rate": 4.307300759009865e-06,
+      "loss": 0.2351,
+      "mean_token_accuracy": 0.9343721866607666,
+      "num_tokens": 6961888.0,
+      "step": 821
+    },
+    {
+      "epoch": 1.886337543053961,
+      "grad_norm": 2.3848352260955457,
+      "learning_rate": 4.295723399683291e-06,
+      "loss": 0.2299,
+      "mean_token_accuracy": 0.9372518062591553,
+      "num_tokens": 6970479.0,
+      "step": 822
+    },
+    {
+      "epoch": 1.8886337543053962,
+      "grad_norm": 2.282538333828109,
+      "learning_rate": 4.28415462007069e-06,
+      "loss": 0.227,
+      "mean_token_accuracy": 0.9358512759208679,
+      "num_tokens": 6978617.0,
+      "step": 823
+    },
+    {
+      "epoch": 1.8909299655568312,
+      "grad_norm": 2.28743453512881,
+      "learning_rate": 4.2725945025923455e-06,
+      "loss": 0.1974,
+      "mean_token_accuracy": 0.9430777132511139,
+      "num_tokens": 6986772.0,
+      "step": 824
+    },
+    {
+      "epoch": 1.8932261768082663,
+      "grad_norm": 2.1964403290083947,
+      "learning_rate": 4.261043129606832e-06,
+      "loss": 0.2227,
+      "mean_token_accuracy": 0.9344627559185028,
+      "num_tokens": 6995269.0,
+      "step": 825
+    },
+    {
+      "epoch": 1.8955223880597014,
+      "grad_norm": 2.3499004122561975,
+      "learning_rate": 4.2495005834104195e-06,
+      "loss": 0.2046,
+      "mean_token_accuracy": 0.9416947960853577,
+      "num_tokens": 7003809.0,
+      "step": 826
+    },
+    {
+      "epoch": 1.8978185993111367,
+      "grad_norm": 2.3493696810075035,
+      "learning_rate": 4.237966946236499e-06,
+      "loss": 0.2301,
+      "mean_token_accuracy": 0.9316303431987762,
+      "num_tokens": 7012830.0,
+      "step": 827
+    },
+    {
+      "epoch": 1.9001148105625718,
+      "grad_norm": 2.0752369578558243,
+      "learning_rate": 4.2264423002549845e-06,
+      "loss": 0.1936,
+      "mean_token_accuracy": 0.941378653049469,
+      "num_tokens": 7021980.0,
+      "step": 828
+    },
+    {
+      "epoch": 1.902411021814007,
+      "grad_norm": 2.6657211515838077,
+      "learning_rate": 4.214926727571741e-06,
+      "loss": 0.2461,
+      "mean_token_accuracy": 0.9303891658782959,
+      "num_tokens": 7030600.0,
+      "step": 829
+    },
+    {
+      "epoch": 1.904707233065442,
+      "grad_norm": 2.4450663533411,
+      "learning_rate": 4.203420310227981e-06,
+      "loss": 0.24,
+      "mean_token_accuracy": 0.9319871366024017,
+      "num_tokens": 7039439.0,
+      "step": 830
+    },
+    {
+      "epoch": 1.907003444316877,
+      "grad_norm": 2.2246640497717554,
+      "learning_rate": 4.191923130199701e-06,
+      "loss": 0.2368,
+      "mean_token_accuracy": 0.9315371215343475,
+      "num_tokens": 7048664.0,
+      "step": 831
+    },
+    {
+      "epoch": 1.9092996555683124,
+      "grad_norm": 2.1506633815822154,
+      "learning_rate": 4.180435269397084e-06,
+      "loss": 0.2135,
+      "mean_token_accuracy": 0.9400423765182495,
+      "num_tokens": 7056609.0,
+      "step": 832
+    },
+    {
+      "epoch": 1.9115958668197475,
+      "grad_norm": 2.0938117270619467,
+      "learning_rate": 4.168956809663914e-06,
+      "loss": 0.2364,
+      "mean_token_accuracy": 0.932383120059967,
+      "num_tokens": 7065156.0,
+      "step": 833
+    },
+    {
+      "epoch": 1.9138920780711826,
+      "grad_norm": 2.231377022156755,
+      "learning_rate": 4.157487832777008e-06,
+      "loss": 0.2128,
+      "mean_token_accuracy": 0.9373239576816559,
+      "num_tokens": 7073773.0,
+      "step": 834
+    },
+    {
+      "epoch": 1.9161882893226176,
+      "grad_norm": 2.249025437436194,
+      "learning_rate": 4.146028420445617e-06,
+      "loss": 0.2028,
+      "mean_token_accuracy": 0.940483957529068,
+      "num_tokens": 7082250.0,
+      "step": 835
+    },
+    {
+      "epoch": 1.9184845005740527,
+      "grad_norm": 1.9605041146618367,
+      "learning_rate": 4.134578654310854e-06,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9495405554771423,
+      "num_tokens": 7091079.0,
+      "step": 836
+    },
+    {
+      "epoch": 1.920780711825488,
+      "grad_norm": 1.8832628814147592,
+      "learning_rate": 4.123138615945107e-06,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9532471299171448,
+      "num_tokens": 7099218.0,
+      "step": 837
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 2.25437765077148,
+      "learning_rate": 4.1117083868514575e-06,
+      "loss": 0.2529,
+      "mean_token_accuracy": 0.9315299093723297,
+      "num_tokens": 7107421.0,
+      "step": 838
+    },
+    {
+      "epoch": 1.9253731343283582,
+      "grad_norm": 2.8741352624312344,
+      "learning_rate": 4.100288048463106e-06,
+      "loss": 0.1988,
+      "mean_token_accuracy": 0.9430383145809174,
+      "num_tokens": 7116109.0,
+      "step": 839
+    },
+    {
+      "epoch": 1.9276693455797933,
+      "grad_norm": 2.5597659036474947,
+      "learning_rate": 4.088877682142789e-06,
+      "loss": 0.2353,
+      "mean_token_accuracy": 0.9324654042720795,
+      "num_tokens": 7124901.0,
+      "step": 840
+    },
+    {
+      "epoch": 1.9299655568312284,
+      "grad_norm": 2.093491453936493,
+      "learning_rate": 4.0774773691821924e-06,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9492015838623047,
+      "num_tokens": 7132784.0,
+      "step": 841
+    },
+    {
+      "epoch": 1.9322617680826637,
+      "grad_norm": 2.332657332408011,
+      "learning_rate": 4.066087190801382e-06,
+      "loss": 0.2115,
+      "mean_token_accuracy": 0.9376582503318787,
+      "num_tokens": 7141202.0,
+      "step": 842
+    },
+    {
+      "epoch": 1.9345579793340988,
+      "grad_norm": 2.1692477934002747,
+      "learning_rate": 4.054707228148227e-06,
+      "loss": 0.1969,
+      "mean_token_accuracy": 0.9452978074550629,
+      "num_tokens": 7149895.0,
+      "step": 843
+    },
+    {
+      "epoch": 1.9368541905855339,
+      "grad_norm": 2.4158461184965025,
+      "learning_rate": 4.0433375622978e-06,
+      "loss": 0.1812,
+      "mean_token_accuracy": 0.947494626045227,
+      "num_tokens": 7158385.0,
+      "step": 844
+    },
+    {
+      "epoch": 1.939150401836969,
+      "grad_norm": 2.3952286430056446,
+      "learning_rate": 4.0319782742518325e-06,
+      "loss": 0.2163,
+      "mean_token_accuracy": 0.9374796748161316,
+      "num_tokens": 7166514.0,
+      "step": 845
+    },
+    {
+      "epoch": 1.941446613088404,
+      "grad_norm": 2.6788607198914933,
+      "learning_rate": 4.020629444938108e-06,
+      "loss": 0.2388,
+      "mean_token_accuracy": 0.9336684942245483,
+      "num_tokens": 7176129.0,
+      "step": 846
+    },
+    {
+      "epoch": 1.9437428243398394,
+      "grad_norm": 2.287803749155692,
+      "learning_rate": 4.009291155209909e-06,
+      "loss": 0.1951,
+      "mean_token_accuracy": 0.9437547624111176,
+      "num_tokens": 7184768.0,
+      "step": 847
+    },
+    {
+      "epoch": 1.9460390355912744,
+      "grad_norm": 2.357276592283698,
+      "learning_rate": 3.997963485845421e-06,
+      "loss": 0.1801,
+      "mean_token_accuracy": 0.9463637471199036,
+      "num_tokens": 7192979.0,
+      "step": 848
+    },
+    {
+      "epoch": 1.9483352468427095,
+      "grad_norm": 1.9952463225168529,
+      "learning_rate": 3.98664651754717e-06,
+      "loss": 0.1737,
+      "mean_token_accuracy": 0.9488488733768463,
+      "num_tokens": 7201314.0,
+      "step": 849
+    },
+    {
+      "epoch": 1.9506314580941446,
+      "grad_norm": 2.1563143434570193,
+      "learning_rate": 3.975340330941443e-06,
+      "loss": 0.2415,
+      "mean_token_accuracy": 0.9331545829772949,
+      "num_tokens": 7210182.0,
+      "step": 850
+    },
+    {
+      "epoch": 1.9529276693455797,
+      "grad_norm": 2.5652660810442063,
+      "learning_rate": 3.964045006577717e-06,
+      "loss": 0.2016,
+      "mean_token_accuracy": 0.9444333612918854,
+      "num_tokens": 7218457.0,
+      "step": 851
+    },
+    {
+      "epoch": 1.955223880597015,
+      "grad_norm": 2.4791025005519463,
+      "learning_rate": 3.952760624928076e-06,
+      "loss": 0.2203,
+      "mean_token_accuracy": 0.9352504014968872,
+      "num_tokens": 7226981.0,
+      "step": 852
+    },
+    {
+      "epoch": 1.95752009184845,
+      "grad_norm": 2.1824727841958893,
+      "learning_rate": 3.941487266386649e-06,
+      "loss": 0.199,
+      "mean_token_accuracy": 0.9426755905151367,
+      "num_tokens": 7235199.0,
+      "step": 853
+    },
+    {
+      "epoch": 1.9598163030998852,
+      "grad_norm": 2.6302674174312735,
+      "learning_rate": 3.930225011269033e-06,
+      "loss": 0.2246,
+      "mean_token_accuracy": 0.9328229129314423,
+      "num_tokens": 7244320.0,
+      "step": 854
+    },
+    {
+      "epoch": 1.9621125143513203,
+      "grad_norm": 2.221142714066129,
+      "learning_rate": 3.918973939811716e-06,
+      "loss": 0.1879,
+      "mean_token_accuracy": 0.945237398147583,
+      "num_tokens": 7253067.0,
+      "step": 855
+    },
+    {
+      "epoch": 1.9644087256027554,
+      "grad_norm": 2.1140814977507683,
+      "learning_rate": 3.907734132171511e-06,
+      "loss": 0.2117,
+      "mean_token_accuracy": 0.9411213397979736,
+      "num_tokens": 7261641.0,
+      "step": 856
+    },
+    {
+      "epoch": 1.9667049368541907,
+      "grad_norm": 2.3531428620506074,
+      "learning_rate": 3.896505668424986e-06,
+      "loss": 0.2133,
+      "mean_token_accuracy": 0.9388346076011658,
+      "num_tokens": 7269687.0,
+      "step": 857
+    },
+    {
+      "epoch": 1.9690011481056258,
+      "grad_norm": 2.584102685829503,
+      "learning_rate": 3.8852886285678896e-06,
+      "loss": 0.2178,
+      "mean_token_accuracy": 0.9339717626571655,
+      "num_tokens": 7278509.0,
+      "step": 858
+    },
+    {
+      "epoch": 1.9712973593570609,
+      "grad_norm": 2.152543152874329,
+      "learning_rate": 3.874083092514579e-06,
+      "loss": 0.1715,
+      "mean_token_accuracy": 0.9511209428310394,
+      "num_tokens": 7287050.0,
+      "step": 859
+    },
+    {
+      "epoch": 1.973593570608496,
+      "grad_norm": 2.084053846440779,
+      "learning_rate": 3.862889140097458e-06,
+      "loss": 0.1879,
+      "mean_token_accuracy": 0.9426831901073456,
+      "num_tokens": 7295564.0,
+      "step": 860
+    },
+    {
+      "epoch": 1.975889781859931,
+      "grad_norm": 2.0287717392599007,
+      "learning_rate": 3.851706851066403e-06,
+      "loss": 0.1705,
+      "mean_token_accuracy": 0.9504532217979431,
+      "num_tokens": 7304547.0,
+      "step": 861
+    },
+    {
+      "epoch": 1.9781859931113663,
+      "grad_norm": 2.146280623853515,
+      "learning_rate": 3.840536305088196e-06,
+      "loss": 0.2271,
+      "mean_token_accuracy": 0.9378258585929871,
+      "num_tokens": 7312680.0,
+      "step": 862
+    },
+    {
+      "epoch": 1.9804822043628014,
+      "grad_norm": 2.4344261292475617,
+      "learning_rate": 3.8293775817459565e-06,
+      "loss": 0.1916,
+      "mean_token_accuracy": 0.9476897716522217,
+      "num_tokens": 7320677.0,
+      "step": 863
+    },
+    {
+      "epoch": 1.9827784156142365,
+      "grad_norm": 2.112669733902935,
+      "learning_rate": 3.818230760538576e-06,
+      "loss": 0.2114,
+      "mean_token_accuracy": 0.9398373961448669,
+      "num_tokens": 7329157.0,
+      "step": 864
+    },
+    {
+      "epoch": 1.9850746268656716,
+      "grad_norm": 2.05019432890871,
+      "learning_rate": 3.807095920880154e-06,
+      "loss": 0.2255,
+      "mean_token_accuracy": 0.9353802800178528,
+      "num_tokens": 7338347.0,
+      "step": 865
+    },
+    {
+      "epoch": 1.9873708381171067,
+      "grad_norm": 2.970846411048413,
+      "learning_rate": 3.7959731420994196e-06,
+      "loss": 0.1948,
+      "mean_token_accuracy": 0.9437238872051239,
+      "num_tokens": 7346638.0,
+      "step": 866
+    },
+    {
+      "epoch": 1.989667049368542,
+      "grad_norm": 2.2085368994542067,
+      "learning_rate": 3.784862503439185e-06,
+      "loss": 0.1672,
+      "mean_token_accuracy": 0.9519528150558472,
+      "num_tokens": 7354911.0,
+      "step": 867
+    },
+    {
+      "epoch": 1.991963260619977,
+      "grad_norm": 2.1863848570417246,
+      "learning_rate": 3.7737640840557677e-06,
+      "loss": 0.1971,
+      "mean_token_accuracy": 0.9445248246192932,
+      "num_tokens": 7364208.0,
+      "step": 868
+    },
+    {
+      "epoch": 1.9942594718714122,
+      "grad_norm": 2.174306835113046,
+      "learning_rate": 3.762677963018433e-06,
+      "loss": 0.1925,
+      "mean_token_accuracy": 0.9417718648910522,
+      "num_tokens": 7372646.0,
+      "step": 869
+    },
+    {
+      "epoch": 1.9965556831228473,
+      "grad_norm": 2.493342930779866,
+      "learning_rate": 3.751604219308825e-06,
+      "loss": 0.164,
+      "mean_token_accuracy": 0.9525461494922638,
+      "num_tokens": 7380846.0,
+      "step": 870
+    },
+    {
+      "epoch": 1.9988518943742823,
+      "grad_norm": 2.1366920211925096,
+      "learning_rate": 3.7405429318204077e-06,
+      "loss": 0.2096,
+      "mean_token_accuracy": 0.939992219209671,
+      "num_tokens": 7389181.0,
+      "step": 871
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.1878191871554207,
+      "learning_rate": 3.7294941793579042e-06,
+      "loss": 0.1457,
+      "mean_token_accuracy": 0.944656491279602,
+      "num_tokens": 7389707.0,
+      "step": 872
+    },
+    {
+      "epoch": 2.0022962112514353,
+      "grad_norm": 1.405230295381346,
+      "learning_rate": 3.718458040636733e-06,
+      "loss": 0.1022,
+      "mean_token_accuracy": 0.9719491004943848,
+      "num_tokens": 7398662.0,
+      "step": 873
+    },
+    {
+      "epoch": 2.00459242250287,
+      "grad_norm": 1.5711012931197248,
+      "learning_rate": 3.7074345942824463e-06,
+      "loss": 0.0993,
+      "mean_token_accuracy": 0.9752732813358307,
+      "num_tokens": 7407157.0,
+      "step": 874
+    },
+    {
+      "epoch": 2.0068886337543055,
+      "grad_norm": 1.6002327948778614,
+      "learning_rate": 3.696423918830171e-06,
+      "loss": 0.1072,
+      "mean_token_accuracy": 0.9709347784519196,
+      "num_tokens": 7415652.0,
+      "step": 875
+    },
+    {
+      "epoch": 2.0091848450057403,
+      "grad_norm": 1.8785026628799746,
+      "learning_rate": 3.6854260927240536e-06,
+      "loss": 0.1131,
+      "mean_token_accuracy": 0.9700099527835846,
+      "num_tokens": 7423519.0,
+      "step": 876
+    },
+    {
+      "epoch": 2.0114810562571757,
+      "grad_norm": 1.7405850550428037,
+      "learning_rate": 3.674441194316689e-06,
+      "loss": 0.0921,
+      "mean_token_accuracy": 0.9760601818561554,
+      "num_tokens": 7431823.0,
+      "step": 877
+    },
+    {
+      "epoch": 2.013777267508611,
+      "grad_norm": 1.471158335966632,
+      "learning_rate": 3.6634693018685773e-06,
+      "loss": 0.0986,
+      "mean_token_accuracy": 0.9733099639415741,
+      "num_tokens": 7440258.0,
+      "step": 878
+    },
+    {
+      "epoch": 2.016073478760046,
+      "grad_norm": 1.5466261503056207,
+      "learning_rate": 3.6525104935475577e-06,
+      "loss": 0.1018,
+      "mean_token_accuracy": 0.9718340635299683,
+      "num_tokens": 7448729.0,
+      "step": 879
+    },
+    {
+      "epoch": 2.018369690011481,
+      "grad_norm": 1.8061385395909293,
+      "learning_rate": 3.641564847428254e-06,
+      "loss": 0.1124,
+      "mean_token_accuracy": 0.9707489609718323,
+      "num_tokens": 7456626.0,
+      "step": 880
+    },
+    {
+      "epoch": 2.020665901262916,
+      "grad_norm": 1.9660502229513621,
+      "learning_rate": 3.630632441491512e-06,
+      "loss": 0.1021,
+      "mean_token_accuracy": 0.9718666076660156,
+      "num_tokens": 7464601.0,
+      "step": 881
+    },
+    {
+      "epoch": 2.0229621125143513,
+      "grad_norm": 1.9142802980644513,
+      "learning_rate": 3.619713353623857e-06,
+      "loss": 0.0934,
+      "mean_token_accuracy": 0.9728955924510956,
+      "num_tokens": 7473097.0,
+      "step": 882
+    },
+    {
+      "epoch": 2.0252583237657866,
+      "grad_norm": 2.226924392534838,
+      "learning_rate": 3.6088076616169265e-06,
+      "loss": 0.1052,
+      "mean_token_accuracy": 0.9716982245445251,
+      "num_tokens": 7481500.0,
+      "step": 883
+    },
+    {
+      "epoch": 2.0275545350172215,
+      "grad_norm": 1.8593128416968028,
+      "learning_rate": 3.597915443166925e-06,
+      "loss": 0.1046,
+      "mean_token_accuracy": 0.9702856242656708,
+      "num_tokens": 7490035.0,
+      "step": 884
+    },
+    {
+      "epoch": 2.029850746268657,
+      "grad_norm": 2.1204017803276303,
+      "learning_rate": 3.58703677587406e-06,
+      "loss": 0.0985,
+      "mean_token_accuracy": 0.9728349447250366,
+      "num_tokens": 7498334.0,
+      "step": 885
+    },
+    {
+      "epoch": 2.0321469575200917,
+      "grad_norm": 2.2102608582461607,
+      "learning_rate": 3.5761717372420045e-06,
+      "loss": 0.1054,
+      "mean_token_accuracy": 0.970887154340744,
+      "num_tokens": 7506229.0,
+      "step": 886
+    },
+    {
+      "epoch": 2.034443168771527,
+      "grad_norm": 1.8282923240304922,
+      "learning_rate": 3.565320404677328e-06,
+      "loss": 0.0962,
+      "mean_token_accuracy": 0.9726462960243225,
+      "num_tokens": 7515067.0,
+      "step": 887
+    },
+    {
+      "epoch": 2.0367393800229623,
+      "grad_norm": 2.1777467862964524,
+      "learning_rate": 3.554482855488954e-06,
+      "loss": 0.1122,
+      "mean_token_accuracy": 0.9679645597934723,
+      "num_tokens": 7523467.0,
+      "step": 888
+    },
+    {
+      "epoch": 2.039035591274397,
+      "grad_norm": 2.055838585855398,
+      "learning_rate": 3.543659166887611e-06,
+      "loss": 0.107,
+      "mean_token_accuracy": 0.9688269793987274,
+      "num_tokens": 7531983.0,
+      "step": 889
+    },
+    {
+      "epoch": 2.0413318025258325,
+      "grad_norm": 2.1126569306041825,
+      "learning_rate": 3.532849415985281e-06,
+      "loss": 0.1064,
+      "mean_token_accuracy": 0.972036600112915,
+      "num_tokens": 7540204.0,
+      "step": 890
+    },
+    {
+      "epoch": 2.0436280137772673,
+      "grad_norm": 2.007137501473711,
+      "learning_rate": 3.5220536797946447e-06,
+      "loss": 0.1126,
+      "mean_token_accuracy": 0.9671331644058228,
+      "num_tokens": 7549502.0,
+      "step": 891
+    },
+    {
+      "epoch": 2.0459242250287026,
+      "grad_norm": 1.7482911924304798,
+      "learning_rate": 3.5112720352285355e-06,
+      "loss": 0.101,
+      "mean_token_accuracy": 0.9717611074447632,
+      "num_tokens": 7558207.0,
+      "step": 892
+    },
+    {
+      "epoch": 2.048220436280138,
+      "grad_norm": 1.8056124526589685,
+      "learning_rate": 3.500504559099397e-06,
+      "loss": 0.1068,
+      "mean_token_accuracy": 0.9714537560939789,
+      "num_tokens": 7566436.0,
+      "step": 893
+    },
+    {
+      "epoch": 2.050516647531573,
+      "grad_norm": 1.6263332556448826,
+      "learning_rate": 3.4897513281187317e-06,
+      "loss": 0.1001,
+      "mean_token_accuracy": 0.9707864224910736,
+      "num_tokens": 7575073.0,
+      "step": 894
+    },
+    {
+      "epoch": 2.052812858783008,
+      "grad_norm": 1.6072293435105862,
+      "learning_rate": 3.4790124188965518e-06,
+      "loss": 0.0951,
+      "mean_token_accuracy": 0.970915287733078,
+      "num_tokens": 7583667.0,
+      "step": 895
+    },
+    {
+      "epoch": 2.055109070034443,
+      "grad_norm": 1.8632414772987125,
+      "learning_rate": 3.468287907940835e-06,
+      "loss": 0.1057,
+      "mean_token_accuracy": 0.9700410664081573,
+      "num_tokens": 7592810.0,
+      "step": 896
+    },
+    {
+      "epoch": 2.0574052812858783,
+      "grad_norm": 1.764573267348085,
+      "learning_rate": 3.4575778716569825e-06,
+      "loss": 0.0972,
+      "mean_token_accuracy": 0.9723480939865112,
+      "num_tokens": 7600909.0,
+      "step": 897
+    },
+    {
+      "epoch": 2.0597014925373136,
+      "grad_norm": 1.802471363643832,
+      "learning_rate": 3.446882386347276e-06,
+      "loss": 0.0981,
+      "mean_token_accuracy": 0.9717832207679749,
+      "num_tokens": 7609597.0,
+      "step": 898
+    },
+    {
+      "epoch": 2.0619977037887485,
+      "grad_norm": 2.231664807030004,
+      "learning_rate": 3.4362015282103232e-06,
+      "loss": 0.1042,
+      "mean_token_accuracy": 0.9726667702198029,
+      "num_tokens": 7617935.0,
+      "step": 899
+    },
+    {
+      "epoch": 2.064293915040184,
+      "grad_norm": 1.6360657348053929,
+      "learning_rate": 3.4255353733405267e-06,
+      "loss": 0.0878,
+      "mean_token_accuracy": 0.9758914709091187,
+      "num_tokens": 7626178.0,
+      "step": 900
+    },
+    {
+      "epoch": 2.0665901262916186,
+      "grad_norm": 1.9874404589837515,
+      "learning_rate": 3.4148839977275373e-06,
+      "loss": 0.114,
+      "mean_token_accuracy": 0.9694769382476807,
+      "num_tokens": 7634375.0,
+      "step": 901
+    },
+    {
+      "epoch": 2.068886337543054,
+      "grad_norm": 1.6697766881877003,
+      "learning_rate": 3.4042474772557143e-06,
+      "loss": 0.0938,
+      "mean_token_accuracy": 0.9730155169963837,
+      "num_tokens": 7643001.0,
+      "step": 902
+    },
+    {
+      "epoch": 2.0711825487944893,
+      "grad_norm": 1.6585277078568317,
+      "learning_rate": 3.393625887703581e-06,
+      "loss": 0.0957,
+      "mean_token_accuracy": 0.9737425446510315,
+      "num_tokens": 7651868.0,
+      "step": 903
+    },
+    {
+      "epoch": 2.073478760045924,
+      "grad_norm": 1.857554399927938,
+      "learning_rate": 3.383019304743285e-06,
+      "loss": 0.0841,
+      "mean_token_accuracy": 0.9742405712604523,
+      "num_tokens": 7660588.0,
+      "step": 904
+    },
+    {
+      "epoch": 2.0757749712973594,
+      "grad_norm": 1.7096977053571714,
+      "learning_rate": 3.372427803940067e-06,
+      "loss": 0.1013,
+      "mean_token_accuracy": 0.970881462097168,
+      "num_tokens": 7669649.0,
+      "step": 905
+    },
+    {
+      "epoch": 2.0780711825487943,
+      "grad_norm": 2.0686630601195644,
+      "learning_rate": 3.361851460751715e-06,
+      "loss": 0.1148,
+      "mean_token_accuracy": 0.9686175882816315,
+      "num_tokens": 7677860.0,
+      "step": 906
+    },
+    {
+      "epoch": 2.0803673938002296,
+      "grad_norm": 2.290258808240003,
+      "learning_rate": 3.351290350528026e-06,
+      "loss": 0.1132,
+      "mean_token_accuracy": 0.9675470590591431,
+      "num_tokens": 7686159.0,
+      "step": 907
+    },
+    {
+      "epoch": 2.082663605051665,
+      "grad_norm": 1.9319648509022145,
+      "learning_rate": 3.340744548510276e-06,
+      "loss": 0.1027,
+      "mean_token_accuracy": 0.9745803475379944,
+      "num_tokens": 7694230.0,
+      "step": 908
+    },
+    {
+      "epoch": 2.0849598163031,
+      "grad_norm": 2.115890232825143,
+      "learning_rate": 3.3302141298306736e-06,
+      "loss": 0.1114,
+      "mean_token_accuracy": 0.9678735435009003,
+      "num_tokens": 7702982.0,
+      "step": 909
+    },
+    {
+      "epoch": 2.087256027554535,
+      "grad_norm": 2.0517725257637056,
+      "learning_rate": 3.3196991695118353e-06,
+      "loss": 0.1254,
+      "mean_token_accuracy": 0.9653653800487518,
+      "num_tokens": 7711730.0,
+      "step": 910
+    },
+    {
+      "epoch": 2.08955223880597,
+      "grad_norm": 1.6847590267627566,
+      "learning_rate": 3.309199742466245e-06,
+      "loss": 0.0907,
+      "mean_token_accuracy": 0.9746355712413788,
+      "num_tokens": 7719915.0,
+      "step": 911
+    },
+    {
+      "epoch": 2.0918484500574053,
+      "grad_norm": 2.0221392209472575,
+      "learning_rate": 3.2987159234957257e-06,
+      "loss": 0.1057,
+      "mean_token_accuracy": 0.9695805311203003,
+      "num_tokens": 7727877.0,
+      "step": 912
+    },
+    {
+      "epoch": 2.0941446613088406,
+      "grad_norm": 1.7819299331184606,
+      "learning_rate": 3.2882477872908965e-06,
+      "loss": 0.0965,
+      "mean_token_accuracy": 0.97397780418396,
+      "num_tokens": 7735934.0,
+      "step": 913
+    },
+    {
+      "epoch": 2.0964408725602754,
+      "grad_norm": 2.2553724952447043,
+      "learning_rate": 3.277795408430648e-06,
+      "loss": 0.1049,
+      "mean_token_accuracy": 0.9717682301998138,
+      "num_tokens": 7744988.0,
+      "step": 914
+    },
+    {
+      "epoch": 2.0987370838117108,
+      "grad_norm": 1.9620275879291489,
+      "learning_rate": 3.267358861381612e-06,
+      "loss": 0.0999,
+      "mean_token_accuracy": 0.9722248315811157,
+      "num_tokens": 7753734.0,
+      "step": 915
+    },
+    {
+      "epoch": 2.1010332950631456,
+      "grad_norm": 1.924734934545376,
+      "learning_rate": 3.256938220497631e-06,
+      "loss": 0.1077,
+      "mean_token_accuracy": 0.971040666103363,
+      "num_tokens": 7762850.0,
+      "step": 916
+    },
+    {
+      "epoch": 2.103329506314581,
+      "grad_norm": 1.521093092505037,
+      "learning_rate": 3.2465335600192183e-06,
+      "loss": 0.0937,
+      "mean_token_accuracy": 0.9730308353900909,
+      "num_tokens": 7771204.0,
+      "step": 917
+    },
+    {
+      "epoch": 2.1056257175660162,
+      "grad_norm": 1.6787247515570305,
+      "learning_rate": 3.236144954073042e-06,
+      "loss": 0.1022,
+      "mean_token_accuracy": 0.9723386466503143,
+      "num_tokens": 7779733.0,
+      "step": 918
+    },
+    {
+      "epoch": 2.107921928817451,
+      "grad_norm": 1.8017231061108854,
+      "learning_rate": 3.22577247667139e-06,
+      "loss": 0.104,
+      "mean_token_accuracy": 0.973284512758255,
+      "num_tokens": 7787479.0,
+      "step": 919
+    },
+    {
+      "epoch": 2.1102181400688864,
+      "grad_norm": 1.5989361004023193,
+      "learning_rate": 3.2154162017116495e-06,
+      "loss": 0.0938,
+      "mean_token_accuracy": 0.9749036133289337,
+      "num_tokens": 7796082.0,
+      "step": 920
+    },
+    {
+      "epoch": 2.1125143513203213,
+      "grad_norm": 1.675906154899016,
+      "learning_rate": 3.20507620297577e-06,
+      "loss": 0.0958,
+      "mean_token_accuracy": 0.9726968109607697,
+      "num_tokens": 7804351.0,
+      "step": 921
+    },
+    {
+      "epoch": 2.1148105625717566,
+      "grad_norm": 2.068928382513605,
+      "learning_rate": 3.1947525541297404e-06,
+      "loss": 0.1083,
+      "mean_token_accuracy": 0.9681409597396851,
+      "num_tokens": 7813549.0,
+      "step": 922
+    },
+    {
+      "epoch": 2.117106773823192,
+      "grad_norm": 1.798626963658433,
+      "learning_rate": 3.1844453287230744e-06,
+      "loss": 0.107,
+      "mean_token_accuracy": 0.9700435101985931,
+      "num_tokens": 7821564.0,
+      "step": 923
+    },
+    {
+      "epoch": 2.1194029850746268,
+      "grad_norm": 1.9160843927964275,
+      "learning_rate": 3.1741546001882773e-06,
+      "loss": 0.1058,
+      "mean_token_accuracy": 0.97170090675354,
+      "num_tokens": 7830416.0,
+      "step": 924
+    },
+    {
+      "epoch": 2.121699196326062,
+      "grad_norm": 1.9417409655141722,
+      "learning_rate": 3.16388044184032e-06,
+      "loss": 0.1025,
+      "mean_token_accuracy": 0.9696717858314514,
+      "num_tokens": 7838685.0,
+      "step": 925
+    },
+    {
+      "epoch": 2.123995407577497,
+      "grad_norm": 1.7324156711035859,
+      "learning_rate": 3.1536229268761244e-06,
+      "loss": 0.0977,
+      "mean_token_accuracy": 0.9723308682441711,
+      "num_tokens": 7847536.0,
+      "step": 926
+    },
+    {
+      "epoch": 2.1262916188289323,
+      "grad_norm": 1.5847351743478961,
+      "learning_rate": 3.1433821283740385e-06,
+      "loss": 0.0965,
+      "mean_token_accuracy": 0.9739912748336792,
+      "num_tokens": 7855500.0,
+      "step": 927
+    },
+    {
+      "epoch": 2.1285878300803676,
+      "grad_norm": 2.0972474043871725,
+      "learning_rate": 3.133158119293318e-06,
+      "loss": 0.0998,
+      "mean_token_accuracy": 0.9718705415725708,
+      "num_tokens": 7863673.0,
+      "step": 928
+    },
+    {
+      "epoch": 2.1308840413318024,
+      "grad_norm": 1.8340557473574228,
+      "learning_rate": 3.1229509724736003e-06,
+      "loss": 0.0899,
+      "mean_token_accuracy": 0.9734176099300385,
+      "num_tokens": 7872016.0,
+      "step": 929
+    },
+    {
+      "epoch": 2.1331802525832377,
+      "grad_norm": 1.671541734361259,
+      "learning_rate": 3.1127607606343934e-06,
+      "loss": 0.1006,
+      "mean_token_accuracy": 0.9730802774429321,
+      "num_tokens": 7879590.0,
+      "step": 930
+    },
+    {
+      "epoch": 2.1354764638346726,
+      "grad_norm": 2.1621040983950395,
+      "learning_rate": 3.102587556374553e-06,
+      "loss": 0.0998,
+      "mean_token_accuracy": 0.9699961543083191,
+      "num_tokens": 7888411.0,
+      "step": 931
+    },
+    {
+      "epoch": 2.137772675086108,
+      "grad_norm": 1.919391776170504,
+      "learning_rate": 3.0924314321717634e-06,
+      "loss": 0.1012,
+      "mean_token_accuracy": 0.9701557457447052,
+      "num_tokens": 7897090.0,
+      "step": 932
+    },
+    {
+      "epoch": 2.140068886337543,
+      "grad_norm": 1.8064798912877178,
+      "learning_rate": 3.082292460382029e-06,
+      "loss": 0.1039,
+      "mean_token_accuracy": 0.9721316397190094,
+      "num_tokens": 7905665.0,
+      "step": 933
+    },
+    {
+      "epoch": 2.142365097588978,
+      "grad_norm": 1.810938133007933,
+      "learning_rate": 3.0721707132391555e-06,
+      "loss": 0.0911,
+      "mean_token_accuracy": 0.973974734544754,
+      "num_tokens": 7914687.0,
+      "step": 934
+    },
+    {
+      "epoch": 2.1446613088404134,
+      "grad_norm": 1.9309802591900258,
+      "learning_rate": 3.0620662628542256e-06,
+      "loss": 0.102,
+      "mean_token_accuracy": 0.9695201516151428,
+      "num_tokens": 7923005.0,
+      "step": 935
+    },
+    {
+      "epoch": 2.1469575200918483,
+      "grad_norm": 2.0076541550750555,
+      "learning_rate": 3.0519791812151e-06,
+      "loss": 0.1179,
+      "mean_token_accuracy": 0.9667558670043945,
+      "num_tokens": 7931797.0,
+      "step": 936
+    },
+    {
+      "epoch": 2.1492537313432836,
+      "grad_norm": 1.7674162480188804,
+      "learning_rate": 3.0419095401858957e-06,
+      "loss": 0.0901,
+      "mean_token_accuracy": 0.9763481318950653,
+      "num_tokens": 7940070.0,
+      "step": 937
+    },
+    {
+      "epoch": 2.151549942594719,
+      "grad_norm": 1.9407510881961034,
+      "learning_rate": 3.0318574115064786e-06,
+      "loss": 0.1036,
+      "mean_token_accuracy": 0.9722597599029541,
+      "num_tokens": 7948395.0,
+      "step": 938
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 2.008582945377345,
+      "learning_rate": 3.021822866791948e-06,
+      "loss": 0.1025,
+      "mean_token_accuracy": 0.9712957739830017,
+      "num_tokens": 7956899.0,
+      "step": 939
+    },
+    {
+      "epoch": 2.156142365097589,
+      "grad_norm": 1.6707894297271144,
+      "learning_rate": 3.0118059775321274e-06,
+      "loss": 0.0856,
+      "mean_token_accuracy": 0.977104663848877,
+      "num_tokens": 7964924.0,
+      "step": 940
+    },
+    {
+      "epoch": 2.158438576349024,
+      "grad_norm": 1.9626361427823178,
+      "learning_rate": 3.001806815091059e-06,
+      "loss": 0.1028,
+      "mean_token_accuracy": 0.9724858105182648,
+      "num_tokens": 7973074.0,
+      "step": 941
+    },
+    {
+      "epoch": 2.1607347876004592,
+      "grad_norm": 2.089059117742878,
+      "learning_rate": 2.9918254507064925e-06,
+      "loss": 0.1147,
+      "mean_token_accuracy": 0.9691679775714874,
+      "num_tokens": 7981717.0,
+      "step": 942
+    },
+    {
+      "epoch": 2.1630309988518945,
+      "grad_norm": 1.858408344444402,
+      "learning_rate": 2.981861955489376e-06,
+      "loss": 0.1011,
+      "mean_token_accuracy": 0.973147988319397,
+      "num_tokens": 7989502.0,
+      "step": 943
+    },
+    {
+      "epoch": 2.1653272101033294,
+      "grad_norm": 3.5346117375637407,
+      "learning_rate": 2.9719164004233493e-06,
+      "loss": 0.1025,
+      "mean_token_accuracy": 0.9718505144119263,
+      "num_tokens": 7997968.0,
+      "step": 944
+    },
+    {
+      "epoch": 2.1676234213547647,
+      "grad_norm": 1.8149668424237893,
+      "learning_rate": 2.961988856364244e-06,
+      "loss": 0.1023,
+      "mean_token_accuracy": 0.9718449711799622,
+      "num_tokens": 8006505.0,
+      "step": 945
+    },
+    {
+      "epoch": 2.1699196326061996,
+      "grad_norm": 1.6549294650886006,
+      "learning_rate": 2.9520793940395735e-06,
+      "loss": 0.0885,
+      "mean_token_accuracy": 0.9767185747623444,
+      "num_tokens": 8014897.0,
+      "step": 946
+    },
+    {
+      "epoch": 2.172215843857635,
+      "grad_norm": 1.9254403944758416,
+      "learning_rate": 2.942188084048029e-06,
+      "loss": 0.105,
+      "mean_token_accuracy": 0.9713899195194244,
+      "num_tokens": 8023037.0,
+      "step": 947
+    },
+    {
+      "epoch": 2.17451205510907,
+      "grad_norm": 1.6701025291504077,
+      "learning_rate": 2.9323149968589766e-06,
+      "loss": 0.0927,
+      "mean_token_accuracy": 0.972253292798996,
+      "num_tokens": 8032418.0,
+      "step": 948
+    },
+    {
+      "epoch": 2.176808266360505,
+      "grad_norm": 1.7832938284762714,
+      "learning_rate": 2.9224602028119597e-06,
+      "loss": 0.1059,
+      "mean_token_accuracy": 0.9699230492115021,
+      "num_tokens": 8040552.0,
+      "step": 949
+    },
+    {
+      "epoch": 2.1791044776119404,
+      "grad_norm": 1.8512660507475882,
+      "learning_rate": 2.912623772116192e-06,
+      "loss": 0.1007,
+      "mean_token_accuracy": 0.972391813993454,
+      "num_tokens": 8048726.0,
+      "step": 950
+    },
+    {
+      "epoch": 2.1814006888633752,
+      "grad_norm": 1.8817893452337984,
+      "learning_rate": 2.9028057748500606e-06,
+      "loss": 0.0971,
+      "mean_token_accuracy": 0.9738054871559143,
+      "num_tokens": 8056764.0,
+      "step": 951
+    },
+    {
+      "epoch": 2.1836969001148105,
+      "grad_norm": 1.8940854367150515,
+      "learning_rate": 2.8930062809606273e-06,
+      "loss": 0.1018,
+      "mean_token_accuracy": 0.9719267785549164,
+      "num_tokens": 8065798.0,
+      "step": 952
+    },
+    {
+      "epoch": 2.185993111366246,
+      "grad_norm": 1.893973423071055,
+      "learning_rate": 2.8832253602631277e-06,
+      "loss": 0.0922,
+      "mean_token_accuracy": 0.9740723371505737,
+      "num_tokens": 8074045.0,
+      "step": 953
+    },
+    {
+      "epoch": 2.1882893226176807,
+      "grad_norm": 2.1851367125419636,
+      "learning_rate": 2.8734630824404718e-06,
+      "loss": 0.1145,
+      "mean_token_accuracy": 0.9680980443954468,
+      "num_tokens": 8082159.0,
+      "step": 954
+    },
+    {
+      "epoch": 2.190585533869116,
+      "grad_norm": 1.907882982566811,
+      "learning_rate": 2.8637195170427567e-06,
+      "loss": 0.09,
+      "mean_token_accuracy": 0.9757258296012878,
+      "num_tokens": 8090652.0,
+      "step": 955
+    },
+    {
+      "epoch": 2.192881745120551,
+      "grad_norm": 1.8964855407560484,
+      "learning_rate": 2.8539947334867657e-06,
+      "loss": 0.0959,
+      "mean_token_accuracy": 0.9719996750354767,
+      "num_tokens": 8099216.0,
+      "step": 956
+    },
+    {
+      "epoch": 2.195177956371986,
+      "grad_norm": 1.668141451413481,
+      "learning_rate": 2.8442888010554658e-06,
+      "loss": 0.0883,
+      "mean_token_accuracy": 0.9743695855140686,
+      "num_tokens": 8107901.0,
+      "step": 957
+    },
+    {
+      "epoch": 2.1974741676234215,
+      "grad_norm": 1.828600129600056,
+      "learning_rate": 2.8346017888975268e-06,
+      "loss": 0.0865,
+      "mean_token_accuracy": 0.9759150445461273,
+      "num_tokens": 8116669.0,
+      "step": 958
+    },
+    {
+      "epoch": 2.1997703788748564,
+      "grad_norm": 2.013249084129064,
+      "learning_rate": 2.824933766026823e-06,
+      "loss": 0.0996,
+      "mean_token_accuracy": 0.9717035889625549,
+      "num_tokens": 8125674.0,
+      "step": 959
+    },
+    {
+      "epoch": 2.2020665901262917,
+      "grad_norm": 1.88263929754765,
+      "learning_rate": 2.8152848013219443e-06,
+      "loss": 0.1126,
+      "mean_token_accuracy": 0.9650498330593109,
+      "num_tokens": 8134222.0,
+      "step": 960
+    },
+    {
+      "epoch": 2.2043628013777266,
+      "grad_norm": 1.8915178190534987,
+      "learning_rate": 2.805654963525698e-06,
+      "loss": 0.0895,
+      "mean_token_accuracy": 0.9738172590732574,
+      "num_tokens": 8142736.0,
+      "step": 961
+    },
+    {
+      "epoch": 2.206659012629162,
+      "grad_norm": 2.060145653363488,
+      "learning_rate": 2.7960443212446254e-06,
+      "loss": 0.111,
+      "mean_token_accuracy": 0.9683137834072113,
+      "num_tokens": 8150944.0,
+      "step": 962
+    },
+    {
+      "epoch": 2.208955223880597,
+      "grad_norm": 2.01023506957835,
+      "learning_rate": 2.7864529429485144e-06,
+      "loss": 0.1087,
+      "mean_token_accuracy": 0.9701187610626221,
+      "num_tokens": 8159587.0,
+      "step": 963
+    },
+    {
+      "epoch": 2.211251435132032,
+      "grad_norm": 1.760404120572593,
+      "learning_rate": 2.77688089696991e-06,
+      "loss": 0.0974,
+      "mean_token_accuracy": 0.9711872637271881,
+      "num_tokens": 8169095.0,
+      "step": 964
+    },
+    {
+      "epoch": 2.2135476463834673,
+      "grad_norm": 1.7563571619969711,
+      "learning_rate": 2.767328251503623e-06,
+      "loss": 0.0897,
+      "mean_token_accuracy": 0.9736784994602203,
+      "num_tokens": 8177561.0,
+      "step": 965
+    },
+    {
+      "epoch": 2.215843857634902,
+      "grad_norm": 1.6446540237824105,
+      "learning_rate": 2.7577950746062475e-06,
+      "loss": 0.0864,
+      "mean_token_accuracy": 0.9756155610084534,
+      "num_tokens": 8186086.0,
+      "step": 966
+    },
+    {
+      "epoch": 2.2181400688863375,
+      "grad_norm": 1.7464471783181186,
+      "learning_rate": 2.748281434195679e-06,
+      "loss": 0.0933,
+      "mean_token_accuracy": 0.9737519025802612,
+      "num_tokens": 8194159.0,
+      "step": 967
+    },
+    {
+      "epoch": 2.220436280137773,
+      "grad_norm": 2.037621984337446,
+      "learning_rate": 2.7387873980506286e-06,
+      "loss": 0.1129,
+      "mean_token_accuracy": 0.9695670902729034,
+      "num_tokens": 8202390.0,
+      "step": 968
+    },
+    {
+      "epoch": 2.2227324913892077,
+      "grad_norm": 1.812248319141627,
+      "learning_rate": 2.7293130338101358e-06,
+      "loss": 0.0987,
+      "mean_token_accuracy": 0.9726190567016602,
+      "num_tokens": 8211765.0,
+      "step": 969
+    },
+    {
+      "epoch": 2.225028702640643,
+      "grad_norm": 2.0340961005227554,
+      "learning_rate": 2.7198584089730896e-06,
+      "loss": 0.1006,
+      "mean_token_accuracy": 0.9732098281383514,
+      "num_tokens": 8220508.0,
+      "step": 970
+    },
+    {
+      "epoch": 2.227324913892078,
+      "grad_norm": 1.6960405359980284,
+      "learning_rate": 2.710423590897755e-06,
+      "loss": 0.0849,
+      "mean_token_accuracy": 0.9763515889644623,
+      "num_tokens": 8229308.0,
+      "step": 971
+    },
+    {
+      "epoch": 2.229621125143513,
+      "grad_norm": 1.7646795110851723,
+      "learning_rate": 2.7010086468012786e-06,
+      "loss": 0.097,
+      "mean_token_accuracy": 0.9726954102516174,
+      "num_tokens": 8237532.0,
+      "step": 972
+    },
+    {
+      "epoch": 2.2319173363949485,
+      "grad_norm": 1.642163333461019,
+      "learning_rate": 2.691613643759222e-06,
+      "loss": 0.097,
+      "mean_token_accuracy": 0.9722395241260529,
+      "num_tokens": 8246377.0,
+      "step": 973
+    },
+    {
+      "epoch": 2.2342135476463834,
+      "grad_norm": 1.8365417801103443,
+      "learning_rate": 2.6822386487050757e-06,
+      "loss": 0.0954,
+      "mean_token_accuracy": 0.9727992713451385,
+      "num_tokens": 8255195.0,
+      "step": 974
+    },
+    {
+      "epoch": 2.2365097588978187,
+      "grad_norm": 1.959870418463878,
+      "learning_rate": 2.6728837284297906e-06,
+      "loss": 0.0892,
+      "mean_token_accuracy": 0.9730575382709503,
+      "num_tokens": 8263166.0,
+      "step": 975
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 1.7846556148528456,
+      "learning_rate": 2.6635489495812916e-06,
+      "loss": 0.0987,
+      "mean_token_accuracy": 0.9716953039169312,
+      "num_tokens": 8271201.0,
+      "step": 976
+    },
+    {
+      "epoch": 2.241102181400689,
+      "grad_norm": 1.964891966329948,
+      "learning_rate": 2.654234378664014e-06,
+      "loss": 0.101,
+      "mean_token_accuracy": 0.9714516699314117,
+      "num_tokens": 8279957.0,
+      "step": 977
+    },
+    {
+      "epoch": 2.243398392652124,
+      "grad_norm": 2.1805713391306396,
+      "learning_rate": 2.644940082038423e-06,
+      "loss": 0.0967,
+      "mean_token_accuracy": 0.9732946753501892,
+      "num_tokens": 8288392.0,
+      "step": 978
+    },
+    {
+      "epoch": 2.245694603903559,
+      "grad_norm": 1.9980050894875911,
+      "learning_rate": 2.6356661259205396e-06,
+      "loss": 0.1015,
+      "mean_token_accuracy": 0.9697504341602325,
+      "num_tokens": 8296950.0,
+      "step": 979
+    },
+    {
+      "epoch": 2.2479908151549943,
+      "grad_norm": 1.7626563047349275,
+      "learning_rate": 2.6264125763814717e-06,
+      "loss": 0.0975,
+      "mean_token_accuracy": 0.9724236726760864,
+      "num_tokens": 8305722.0,
+      "step": 980
+    },
+    {
+      "epoch": 2.250287026406429,
+      "grad_norm": 1.740961055575534,
+      "learning_rate": 2.6171794993469455e-06,
+      "loss": 0.0968,
+      "mean_token_accuracy": 0.9731466770172119,
+      "num_tokens": 8313844.0,
+      "step": 981
+    },
+    {
+      "epoch": 2.2525832376578645,
+      "grad_norm": 1.8485390492629503,
+      "learning_rate": 2.607966960596836e-06,
+      "loss": 0.1037,
+      "mean_token_accuracy": 0.969839483499527,
+      "num_tokens": 8322714.0,
+      "step": 982
+    },
+    {
+      "epoch": 2.2548794489093,
+      "grad_norm": 2.0261345773187327,
+      "learning_rate": 2.598775025764688e-06,
+      "loss": 0.1082,
+      "mean_token_accuracy": 0.9705328047275543,
+      "num_tokens": 8331396.0,
+      "step": 983
+    },
+    {
+      "epoch": 2.2571756601607347,
+      "grad_norm": 1.6942556317589546,
+      "learning_rate": 2.589603760337262e-06,
+      "loss": 0.0935,
+      "mean_token_accuracy": 0.9747720956802368,
+      "num_tokens": 8339836.0,
+      "step": 984
+    },
+    {
+      "epoch": 2.25947187141217,
+      "grad_norm": 1.7654694432632958,
+      "learning_rate": 2.5804532296540604e-06,
+      "loss": 0.0994,
+      "mean_token_accuracy": 0.9727143049240112,
+      "num_tokens": 8348181.0,
+      "step": 985
+    },
+    {
+      "epoch": 2.261768082663605,
+      "grad_norm": 1.8164424007406454,
+      "learning_rate": 2.5713234989068654e-06,
+      "loss": 0.1075,
+      "mean_token_accuracy": 0.9699047207832336,
+      "num_tokens": 8356717.0,
+      "step": 986
+    },
+    {
+      "epoch": 2.26406429391504,
+      "grad_norm": 1.5124866762285514,
+      "learning_rate": 2.5622146331392707e-06,
+      "loss": 0.0882,
+      "mean_token_accuracy": 0.9748025834560394,
+      "num_tokens": 8365955.0,
+      "step": 987
+    },
+    {
+      "epoch": 2.2663605051664755,
+      "grad_norm": 1.7687698326861936,
+      "learning_rate": 2.5531266972462176e-06,
+      "loss": 0.0921,
+      "mean_token_accuracy": 0.9736714959144592,
+      "num_tokens": 8374290.0,
+      "step": 988
+    },
+    {
+      "epoch": 2.2686567164179103,
+      "grad_norm": 1.8178253582840882,
+      "learning_rate": 2.544059755973541e-06,
+      "loss": 0.0977,
+      "mean_token_accuracy": 0.972365140914917,
+      "num_tokens": 8382542.0,
+      "step": 989
+    },
+    {
+      "epoch": 2.2709529276693456,
+      "grad_norm": 1.7300056823611158,
+      "learning_rate": 2.535013873917501e-06,
+      "loss": 0.1007,
+      "mean_token_accuracy": 0.9705684185028076,
+      "num_tokens": 8390747.0,
+      "step": 990
+    },
+    {
+      "epoch": 2.2732491389207805,
+      "grad_norm": 1.666620210779074,
+      "learning_rate": 2.52598911552432e-06,
+      "loss": 0.0855,
+      "mean_token_accuracy": 0.9765442609786987,
+      "num_tokens": 8399228.0,
+      "step": 991
+    },
+    {
+      "epoch": 2.275545350172216,
+      "grad_norm": 1.604114474258696,
+      "learning_rate": 2.5169855450897298e-06,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9765090346336365,
+      "num_tokens": 8408133.0,
+      "step": 992
+    },
+    {
+      "epoch": 2.277841561423651,
+      "grad_norm": 1.9251246408038933,
+      "learning_rate": 2.508003226758513e-06,
+      "loss": 0.1025,
+      "mean_token_accuracy": 0.9718685746192932,
+      "num_tokens": 8416424.0,
+      "step": 993
+    },
+    {
+      "epoch": 2.280137772675086,
+      "grad_norm": 2.073475871020677,
+      "learning_rate": 2.499042224524042e-06,
+      "loss": 0.102,
+      "mean_token_accuracy": 0.9719178080558777,
+      "num_tokens": 8424308.0,
+      "step": 994
+    },
+    {
+      "epoch": 2.2824339839265213,
+      "grad_norm": 1.686149329921028,
+      "learning_rate": 2.4901026022278273e-06,
+      "loss": 0.0931,
+      "mean_token_accuracy": 0.9728578925132751,
+      "num_tokens": 8433554.0,
+      "step": 995
+    },
+    {
+      "epoch": 2.284730195177956,
+      "grad_norm": 1.9473532749994775,
+      "learning_rate": 2.4811844235590587e-06,
+      "loss": 0.0988,
+      "mean_token_accuracy": 0.9713248014450073,
+      "num_tokens": 8442492.0,
+      "step": 996
+    },
+    {
+      "epoch": 2.2870264064293915,
+      "grad_norm": 1.8010678864721914,
+      "learning_rate": 2.472287752054157e-06,
+      "loss": 0.0916,
+      "mean_token_accuracy": 0.9755350649356842,
+      "num_tokens": 8450632.0,
+      "step": 997
+    },
+    {
+      "epoch": 2.289322617680827,
+      "grad_norm": 1.9219646538588298,
+      "learning_rate": 2.463412651096312e-06,
+      "loss": 0.0995,
+      "mean_token_accuracy": 0.9701071381568909,
+      "num_tokens": 8458888.0,
+      "step": 998
+    },
+    {
+      "epoch": 2.2916188289322617,
+      "grad_norm": 1.987215151863472,
+      "learning_rate": 2.454559183915043e-06,
+      "loss": 0.0999,
+      "mean_token_accuracy": 0.9710122048854828,
+      "num_tokens": 8467215.0,
+      "step": 999
+    },
+    {
+      "epoch": 2.293915040183697,
+      "grad_norm": 1.9019749297988546,
+      "learning_rate": 2.4457274135857415e-06,
+      "loss": 0.0969,
+      "mean_token_accuracy": 0.9713835418224335,
+      "num_tokens": 8475579.0,
+      "step": 1000
+    },
+    {
+      "epoch": 2.296211251435132,
+      "grad_norm": 1.9960437610279846,
+      "learning_rate": 2.436917403029219e-06,
+      "loss": 0.1062,
+      "mean_token_accuracy": 0.9702105820178986,
+      "num_tokens": 8483928.0,
+      "step": 1001
+    },
+    {
+      "epoch": 2.298507462686567,
+      "grad_norm": 2.0739291303161944,
+      "learning_rate": 2.428129215011263e-06,
+      "loss": 0.1068,
+      "mean_token_accuracy": 0.9693008661270142,
+      "num_tokens": 8492807.0,
+      "step": 1002
+    },
+    {
+      "epoch": 2.3008036739380024,
+      "grad_norm": 1.8651382336678106,
+      "learning_rate": 2.4193629121421912e-06,
+      "loss": 0.1052,
+      "mean_token_accuracy": 0.9696202278137207,
+      "num_tokens": 8501487.0,
+      "step": 1003
+    },
+    {
+      "epoch": 2.3030998851894373,
+      "grad_norm": 1.7659584094626088,
+      "learning_rate": 2.4106185568764045e-06,
+      "loss": 0.1004,
+      "mean_token_accuracy": 0.9726259708404541,
+      "num_tokens": 8509589.0,
+      "step": 1004
+    },
+    {
+      "epoch": 2.3053960964408726,
+      "grad_norm": 1.8563702370103488,
+      "learning_rate": 2.4018962115119376e-06,
+      "loss": 0.0932,
+      "mean_token_accuracy": 0.9722463190555573,
+      "num_tokens": 8518734.0,
+      "step": 1005
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 1.5025451872713913,
+      "learning_rate": 2.3931959381900184e-06,
+      "loss": 0.0896,
+      "mean_token_accuracy": 0.9753662049770355,
+      "num_tokens": 8526582.0,
+      "step": 1006
+    },
+    {
+      "epoch": 2.309988518943743,
+      "grad_norm": 1.8028779738233462,
+      "learning_rate": 2.384517798894628e-06,
+      "loss": 0.0976,
+      "mean_token_accuracy": 0.97175532579422,
+      "num_tokens": 8535385.0,
+      "step": 1007
+    },
+    {
+      "epoch": 2.312284730195178,
+      "grad_norm": 2.2492362561247194,
+      "learning_rate": 2.375861855452057e-06,
+      "loss": 0.1067,
+      "mean_token_accuracy": 0.9702053666114807,
+      "num_tokens": 8544244.0,
+      "step": 1008
+    },
+    {
+      "epoch": 2.314580941446613,
+      "grad_norm": 1.6631274747153038,
+      "learning_rate": 2.367228169530461e-06,
+      "loss": 0.0891,
+      "mean_token_accuracy": 0.9732209742069244,
+      "num_tokens": 8552977.0,
+      "step": 1009
+    },
+    {
+      "epoch": 2.3168771526980483,
+      "grad_norm": 1.9446406031632628,
+      "learning_rate": 2.358616802639425e-06,
+      "loss": 0.0984,
+      "mean_token_accuracy": 0.9696375131607056,
+      "num_tokens": 8561671.0,
+      "step": 1010
+    },
+    {
+      "epoch": 2.319173363949483,
+      "grad_norm": 1.8885382926993952,
+      "learning_rate": 2.3500278161295283e-06,
+      "loss": 0.0948,
+      "mean_token_accuracy": 0.9724015593528748,
+      "num_tokens": 8570420.0,
+      "step": 1011
+    },
+    {
+      "epoch": 2.3214695752009185,
+      "grad_norm": 1.7467685871148602,
+      "learning_rate": 2.341461271191903e-06,
+      "loss": 0.0891,
+      "mean_token_accuracy": 0.9735634922981262,
+      "num_tokens": 8578912.0,
+      "step": 1012
+    },
+    {
+      "epoch": 2.3237657864523538,
+      "grad_norm": 1.6616678854286182,
+      "learning_rate": 2.3329172288577983e-06,
+      "loss": 0.0946,
+      "mean_token_accuracy": 0.9727250635623932,
+      "num_tokens": 8587152.0,
+      "step": 1013
+    },
+    {
+      "epoch": 2.3260619977037886,
+      "grad_norm": 1.7069703606529893,
+      "learning_rate": 2.324395749998143e-06,
+      "loss": 0.0877,
+      "mean_token_accuracy": 0.9727701842784882,
+      "num_tokens": 8596437.0,
+      "step": 1014
+    },
+    {
+      "epoch": 2.328358208955224,
+      "grad_norm": 2.472509515821182,
+      "learning_rate": 2.3158968953231236e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.9738726913928986,
+      "num_tokens": 8605487.0,
+      "step": 1015
+    },
+    {
+      "epoch": 2.330654420206659,
+      "grad_norm": 2.087923035139329,
+      "learning_rate": 2.3074207253817347e-06,
+      "loss": 0.1046,
+      "mean_token_accuracy": 0.970540851354599,
+      "num_tokens": 8614005.0,
+      "step": 1016
+    },
+    {
+      "epoch": 2.332950631458094,
+      "grad_norm": 1.7113855793026547,
+      "learning_rate": 2.2989673005613634e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.9730663001537323,
+      "num_tokens": 8622962.0,
+      "step": 1017
+    },
+    {
+      "epoch": 2.3352468427095294,
+      "grad_norm": 1.874507919109776,
+      "learning_rate": 2.2905366810873464e-06,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9765822291374207,
+      "num_tokens": 8632269.0,
+      "step": 1018
+    },
+    {
+      "epoch": 2.3375430539609643,
+      "grad_norm": 1.7237428265663721,
+      "learning_rate": 2.282128927022553e-06,
+      "loss": 0.1002,
+      "mean_token_accuracy": 0.9698289334774017,
+      "num_tokens": 8640574.0,
+      "step": 1019
+    },
+    {
+      "epoch": 2.3398392652123996,
+      "grad_norm": 1.867189185744585,
+      "learning_rate": 2.273744098266943e-06,
+      "loss": 0.1117,
+      "mean_token_accuracy": 0.9711831212043762,
+      "num_tokens": 8649564.0,
+      "step": 1020
+    },
+    {
+      "epoch": 2.3421354764638345,
+      "grad_norm": 1.6581809541998571,
+      "learning_rate": 2.2653822545571543e-06,
+      "loss": 0.0898,
+      "mean_token_accuracy": 0.9738336205482483,
+      "num_tokens": 8657867.0,
+      "step": 1021
+    },
+    {
+      "epoch": 2.3444316877152698,
+      "grad_norm": 2.1933535119955594,
+      "learning_rate": 2.2570434554660696e-06,
+      "loss": 0.1062,
+      "mean_token_accuracy": 0.9711630642414093,
+      "num_tokens": 8666204.0,
+      "step": 1022
+    },
+    {
+      "epoch": 2.346727898966705,
+      "grad_norm": 1.80165383680953,
+      "learning_rate": 2.248727760402391e-06,
+      "loss": 0.0909,
+      "mean_token_accuracy": 0.974210262298584,
+      "num_tokens": 8675380.0,
+      "step": 1023
+    },
+    {
+      "epoch": 2.34902411021814,
+      "grad_norm": 1.9804995456686099,
+      "learning_rate": 2.2404352286102183e-06,
+      "loss": 0.1062,
+      "mean_token_accuracy": 0.9697749316692352,
+      "num_tokens": 8683584.0,
+      "step": 1024
+    },
+    {
+      "epoch": 2.3513203214695753,
+      "grad_norm": 1.670498265161908,
+      "learning_rate": 2.232165919168631e-06,
+      "loss": 0.0939,
+      "mean_token_accuracy": 0.9724120795726776,
+      "num_tokens": 8692060.0,
+      "step": 1025
+    },
+    {
+      "epoch": 2.35361653272101,
+      "grad_norm": 2.0041511673394585,
+      "learning_rate": 2.223919890991263e-06,
+      "loss": 0.0973,
+      "mean_token_accuracy": 0.972173273563385,
+      "num_tokens": 8700033.0,
+      "step": 1026
+    },
+    {
+      "epoch": 2.3559127439724454,
+      "grad_norm": 1.9766354409319344,
+      "learning_rate": 2.2156972028258817e-06,
+      "loss": 0.1004,
+      "mean_token_accuracy": 0.9731994271278381,
+      "num_tokens": 8708432.0,
+      "step": 1027
+    },
+    {
+      "epoch": 2.3582089552238807,
+      "grad_norm": 1.793242599636418,
+      "learning_rate": 2.207497913253972e-06,
+      "loss": 0.0957,
+      "mean_token_accuracy": 0.9744184911251068,
+      "num_tokens": 8716865.0,
+      "step": 1028
+    },
+    {
+      "epoch": 2.3605051664753156,
+      "grad_norm": 1.9462144308001539,
+      "learning_rate": 2.1993220806903203e-06,
+      "loss": 0.0985,
+      "mean_token_accuracy": 0.9713562428951263,
+      "num_tokens": 8725136.0,
+      "step": 1029
+    },
+    {
+      "epoch": 2.362801377726751,
+      "grad_norm": 1.6267070063318712,
+      "learning_rate": 2.191169763382597e-06,
+      "loss": 0.0885,
+      "mean_token_accuracy": 0.9753972589969635,
+      "num_tokens": 8733791.0,
+      "step": 1030
+    },
+    {
+      "epoch": 2.365097588978186,
+      "grad_norm": 1.9738635266281657,
+      "learning_rate": 2.1830410194109392e-06,
+      "loss": 0.1,
+      "mean_token_accuracy": 0.9721159934997559,
+      "num_tokens": 8741919.0,
+      "step": 1031
+    },
+    {
+      "epoch": 2.367393800229621,
+      "grad_norm": 1.8313531174406827,
+      "learning_rate": 2.1749359066875372e-06,
+      "loss": 0.0979,
+      "mean_token_accuracy": 0.9714204370975494,
+      "num_tokens": 8751171.0,
+      "step": 1032
+    },
+    {
+      "epoch": 2.3696900114810564,
+      "grad_norm": 1.7376655633420917,
+      "learning_rate": 2.1668544829562276e-06,
+      "loss": 0.0927,
+      "mean_token_accuracy": 0.9742630124092102,
+      "num_tokens": 8759651.0,
+      "step": 1033
+    },
+    {
+      "epoch": 2.3719862227324913,
+      "grad_norm": 1.8781102935052811,
+      "learning_rate": 2.158796805792076e-06,
+      "loss": 0.0962,
+      "mean_token_accuracy": 0.9730274975299835,
+      "num_tokens": 8768156.0,
+      "step": 1034
+    },
+    {
+      "epoch": 2.3742824339839266,
+      "grad_norm": 1.75792302072527,
+      "learning_rate": 2.1507629326009686e-06,
+      "loss": 0.1074,
+      "mean_token_accuracy": 0.9678171873092651,
+      "num_tokens": 8776667.0,
+      "step": 1035
+    },
+    {
+      "epoch": 2.3765786452353614,
+      "grad_norm": 1.707251512997097,
+      "learning_rate": 2.142752920619201e-06,
+      "loss": 0.0937,
+      "mean_token_accuracy": 0.9740056097507477,
+      "num_tokens": 8784878.0,
+      "step": 1036
+    },
+    {
+      "epoch": 2.3788748564867968,
+      "grad_norm": 1.680240330752541,
+      "learning_rate": 2.134766826913078e-06,
+      "loss": 0.0951,
+      "mean_token_accuracy": 0.9711031317710876,
+      "num_tokens": 8793073.0,
+      "step": 1037
+    },
+    {
+      "epoch": 2.381171067738232,
+      "grad_norm": 1.9936410682314054,
+      "learning_rate": 2.1268047083784952e-06,
+      "loss": 0.1054,
+      "mean_token_accuracy": 0.9698427617549896,
+      "num_tokens": 8801402.0,
+      "step": 1038
+    },
+    {
+      "epoch": 2.383467278989667,
+      "grad_norm": 1.7882435667057297,
+      "learning_rate": 2.118866621740546e-06,
+      "loss": 0.1053,
+      "mean_token_accuracy": 0.9717386662960052,
+      "num_tokens": 8809867.0,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3857634902411022,
+      "grad_norm": 1.6760822294305948,
+      "learning_rate": 2.1109526235531087e-06,
+      "loss": 0.0871,
+      "mean_token_accuracy": 0.9754285514354706,
+      "num_tokens": 8818282.0,
+      "step": 1040
+    },
+    {
+      "epoch": 2.388059701492537,
+      "grad_norm": 2.026061290741553,
+      "learning_rate": 2.1030627701984468e-06,
+      "loss": 0.0993,
+      "mean_token_accuracy": 0.9735712707042694,
+      "num_tokens": 8826207.0,
+      "step": 1041
+    },
+    {
+      "epoch": 2.3903559127439724,
+      "grad_norm": 1.9500932126404475,
+      "learning_rate": 2.0951971178868057e-06,
+      "loss": 0.0921,
+      "mean_token_accuracy": 0.9746260344982147,
+      "num_tokens": 8834530.0,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3926521239954077,
+      "grad_norm": 2.1063650510574727,
+      "learning_rate": 2.087355722656017e-06,
+      "loss": 0.1206,
+      "mean_token_accuracy": 0.9655192792415619,
+      "num_tokens": 8843569.0,
+      "step": 1043
+    },
+    {
+      "epoch": 2.3949483352468426,
+      "grad_norm": 1.8172536949905753,
+      "learning_rate": 2.079538640371096e-06,
+      "loss": 0.1026,
+      "mean_token_accuracy": 0.9716907739639282,
+      "num_tokens": 8852183.0,
+      "step": 1044
+    },
+    {
+      "epoch": 2.397244546498278,
+      "grad_norm": 1.77209403322718,
+      "learning_rate": 2.071745926723836e-06,
+      "loss": 0.097,
+      "mean_token_accuracy": 0.9736886620521545,
+      "num_tokens": 8860818.0,
+      "step": 1045
+    },
+    {
+      "epoch": 2.3995407577497128,
+      "grad_norm": 1.926336243788009,
+      "learning_rate": 2.0639776372324277e-06,
+      "loss": 0.1103,
+      "mean_token_accuracy": 0.9717413485050201,
+      "num_tokens": 8868984.0,
+      "step": 1046
+    },
+    {
+      "epoch": 2.401836969001148,
+      "grad_norm": 1.536703474980617,
+      "learning_rate": 2.056233827241052e-06,
+      "loss": 0.0887,
+      "mean_token_accuracy": 0.9743748605251312,
+      "num_tokens": 8877619.0,
+      "step": 1047
+    },
+    {
+      "epoch": 2.4041331802525834,
+      "grad_norm": 1.6743795774421835,
+      "learning_rate": 2.048514551919489e-06,
+      "loss": 0.0991,
+      "mean_token_accuracy": 0.9721062183380127,
+      "num_tokens": 8886879.0,
+      "step": 1048
+    },
+    {
+      "epoch": 2.4064293915040182,
+      "grad_norm": 1.7795958599592456,
+      "learning_rate": 2.040819866262724e-06,
+      "loss": 0.0989,
+      "mean_token_accuracy": 0.9716989398002625,
+      "num_tokens": 8895529.0,
+      "step": 1049
+    },
+    {
+      "epoch": 2.4087256027554536,
+      "grad_norm": 1.8683638945041219,
+      "learning_rate": 2.0331498250905553e-06,
+      "loss": 0.0981,
+      "mean_token_accuracy": 0.9717687666416168,
+      "num_tokens": 8903910.0,
+      "step": 1050
+    },
+    {
+      "epoch": 2.4110218140068884,
+      "grad_norm": 1.4759944667460212,
+      "learning_rate": 2.025504483047207e-06,
+      "loss": 0.0846,
+      "mean_token_accuracy": 0.9752329289913177,
+      "num_tokens": 8912230.0,
+      "step": 1051
+    },
+    {
+      "epoch": 2.4133180252583237,
+      "grad_norm": 1.9166225110804873,
+      "learning_rate": 2.017883894600938e-06,
+      "loss": 0.0961,
+      "mean_token_accuracy": 0.9726128578186035,
+      "num_tokens": 8920371.0,
+      "step": 1052
+    },
+    {
+      "epoch": 2.415614236509759,
+      "grad_norm": 1.7722524997705003,
+      "learning_rate": 2.010288114043651e-06,
+      "loss": 0.0973,
+      "mean_token_accuracy": 0.9724628627300262,
+      "num_tokens": 8928574.0,
+      "step": 1053
+    },
+    {
+      "epoch": 2.417910447761194,
+      "grad_norm": 1.5369659239686593,
+      "learning_rate": 2.002717195490507e-06,
+      "loss": 0.0893,
+      "mean_token_accuracy": 0.9749113321304321,
+      "num_tokens": 8936464.0,
+      "step": 1054
+    },
+    {
+      "epoch": 2.420206659012629,
+      "grad_norm": 2.195228423785846,
+      "learning_rate": 1.995171192879547e-06,
+      "loss": 0.1105,
+      "mean_token_accuracy": 0.9670826196670532,
+      "num_tokens": 8944336.0,
+      "step": 1055
+    },
+    {
+      "epoch": 2.422502870264064,
+      "grad_norm": 2.022017233167384,
+      "learning_rate": 1.9876501599712933e-06,
+      "loss": 0.0941,
+      "mean_token_accuracy": 0.9739137291908264,
+      "num_tokens": 8952840.0,
+      "step": 1056
+    },
+    {
+      "epoch": 2.4247990815154994,
+      "grad_norm": 1.8937487042090648,
+      "learning_rate": 1.980154150348382e-06,
+      "loss": 0.0952,
+      "mean_token_accuracy": 0.9731948375701904,
+      "num_tokens": 8961293.0,
+      "step": 1057
+    },
+    {
+      "epoch": 2.4270952927669347,
+      "grad_norm": 2.243021131946495,
+      "learning_rate": 1.9726832174151696e-06,
+      "loss": 0.1115,
+      "mean_token_accuracy": 0.9702674448490143,
+      "num_tokens": 8969580.0,
+      "step": 1058
+    },
+    {
+      "epoch": 2.4293915040183696,
+      "grad_norm": 1.8100199320013886,
+      "learning_rate": 1.9652374143973586e-06,
+      "loss": 0.0903,
+      "mean_token_accuracy": 0.9744397103786469,
+      "num_tokens": 8977554.0,
+      "step": 1059
+    },
+    {
+      "epoch": 2.431687715269805,
+      "grad_norm": 1.9906490965175785,
+      "learning_rate": 1.9578167943416154e-06,
+      "loss": 0.0935,
+      "mean_token_accuracy": 0.9754206538200378,
+      "num_tokens": 8986272.0,
+      "step": 1060
+    },
+    {
+      "epoch": 2.4339839265212397,
+      "grad_norm": 1.5777200620943583,
+      "learning_rate": 1.9504214101151964e-06,
+      "loss": 0.0816,
+      "mean_token_accuracy": 0.9764645993709564,
+      "num_tokens": 8994885.0,
+      "step": 1061
+    },
+    {
+      "epoch": 2.436280137772675,
+      "grad_norm": 1.7858803140896242,
+      "learning_rate": 1.943051314405564e-06,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9751078188419342,
+      "num_tokens": 9003896.0,
+      "step": 1062
+    },
+    {
+      "epoch": 2.4385763490241104,
+      "grad_norm": 1.5288533156224067,
+      "learning_rate": 1.935706559720022e-06,
+      "loss": 0.0875,
+      "mean_token_accuracy": 0.9750130474567413,
+      "num_tokens": 9012328.0,
+      "step": 1063
+    },
+    {
+      "epoch": 2.4408725602755452,
+      "grad_norm": 1.7680272105043533,
+      "learning_rate": 1.928387198385329e-06,
+      "loss": 0.0934,
+      "mean_token_accuracy": 0.9729612171649933,
+      "num_tokens": 9021082.0,
+      "step": 1064
+    },
+    {
+      "epoch": 2.4431687715269805,
+      "grad_norm": 1.9751950484876037,
+      "learning_rate": 1.9210932825473354e-06,
+      "loss": 0.0968,
+      "mean_token_accuracy": 0.973320871591568,
+      "num_tokens": 9029131.0,
+      "step": 1065
+    },
+    {
+      "epoch": 2.4454649827784154,
+      "grad_norm": 1.693874849630797,
+      "learning_rate": 1.9138248641706104e-06,
+      "loss": 0.0867,
+      "mean_token_accuracy": 0.9768656194210052,
+      "num_tokens": 9037047.0,
+      "step": 1066
+    },
+    {
+      "epoch": 2.4477611940298507,
+      "grad_norm": 2.115703250314276,
+      "learning_rate": 1.9065819950380634e-06,
+      "loss": 0.11,
+      "mean_token_accuracy": 0.96636763215065,
+      "num_tokens": 9045077.0,
+      "step": 1067
+    },
+    {
+      "epoch": 2.450057405281286,
+      "grad_norm": 2.023211890874884,
+      "learning_rate": 1.8993647267505865e-06,
+      "loss": 0.1036,
+      "mean_token_accuracy": 0.9724979996681213,
+      "num_tokens": 9053149.0,
+      "step": 1068
+    },
+    {
+      "epoch": 2.452353616532721,
+      "grad_norm": 1.610363103003913,
+      "learning_rate": 1.892173110726682e-06,
+      "loss": 0.0882,
+      "mean_token_accuracy": 0.9749874174594879,
+      "num_tokens": 9061376.0,
+      "step": 1069
+    },
+    {
+      "epoch": 2.454649827784156,
+      "grad_norm": 1.7879720640971457,
+      "learning_rate": 1.8850071982020945e-06,
+      "loss": 0.0875,
+      "mean_token_accuracy": 0.9755902588367462,
+      "num_tokens": 9069622.0,
+      "step": 1070
+    },
+    {
+      "epoch": 2.456946039035591,
+      "grad_norm": 1.7720545315905012,
+      "learning_rate": 1.877867040229447e-06,
+      "loss": 0.0841,
+      "mean_token_accuracy": 0.9766930043697357,
+      "num_tokens": 9078415.0,
+      "step": 1071
+    },
+    {
+      "epoch": 2.4592422502870264,
+      "grad_norm": 1.7194738129987883,
+      "learning_rate": 1.870752687677876e-06,
+      "loss": 0.0862,
+      "mean_token_accuracy": 0.9777931571006775,
+      "num_tokens": 9087562.0,
+      "step": 1072
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 1.9866693104965958,
+      "learning_rate": 1.8636641912326754e-06,
+      "loss": 0.1061,
+      "mean_token_accuracy": 0.9707072675228119,
+      "num_tokens": 9095528.0,
+      "step": 1073
+    },
+    {
+      "epoch": 2.4638346727898965,
+      "grad_norm": 1.9145159096107132,
+      "learning_rate": 1.8566016013949273e-06,
+      "loss": 0.1005,
+      "mean_token_accuracy": 0.972745269536972,
+      "num_tokens": 9103399.0,
+      "step": 1074
+    },
+    {
+      "epoch": 2.466130884041332,
+      "grad_norm": 1.7404262337759284,
+      "learning_rate": 1.8495649684811448e-06,
+      "loss": 0.0922,
+      "mean_token_accuracy": 0.9743312895298004,
+      "num_tokens": 9111456.0,
+      "step": 1075
+    },
+    {
+      "epoch": 2.4684270952927667,
+      "grad_norm": 1.8261107807260792,
+      "learning_rate": 1.8425543426229156e-06,
+      "loss": 0.097,
+      "mean_token_accuracy": 0.9734582304954529,
+      "num_tokens": 9119607.0,
+      "step": 1076
+    },
+    {
+      "epoch": 2.470723306544202,
+      "grad_norm": 1.9483540851692818,
+      "learning_rate": 1.835569773766544e-06,
+      "loss": 0.1014,
+      "mean_token_accuracy": 0.9732887446880341,
+      "num_tokens": 9127259.0,
+      "step": 1077
+    },
+    {
+      "epoch": 2.4730195177956373,
+      "grad_norm": 1.8794780658170742,
+      "learning_rate": 1.8286113116726928e-06,
+      "loss": 0.1001,
+      "mean_token_accuracy": 0.9706664085388184,
+      "num_tokens": 9135592.0,
+      "step": 1078
+    },
+    {
+      "epoch": 2.475315729047072,
+      "grad_norm": 1.750660320403182,
+      "learning_rate": 1.8216790059160327e-06,
+      "loss": 0.0989,
+      "mean_token_accuracy": 0.9726308286190033,
+      "num_tokens": 9143584.0,
+      "step": 1079
+    },
+    {
+      "epoch": 2.4776119402985075,
+      "grad_norm": 1.8345847915345292,
+      "learning_rate": 1.8147729058848846e-06,
+      "loss": 0.0962,
+      "mean_token_accuracy": 0.9721178114414215,
+      "num_tokens": 9152443.0,
+      "step": 1080
+    },
+    {
+      "epoch": 2.4799081515499424,
+      "grad_norm": 1.6937734103620645,
+      "learning_rate": 1.8078930607808726e-06,
+      "loss": 0.0948,
+      "mean_token_accuracy": 0.9740507900714874,
+      "num_tokens": 9160969.0,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4822043628013777,
+      "grad_norm": 1.9915055681644842,
+      "learning_rate": 1.801039519618567e-06,
+      "loss": 0.1027,
+      "mean_token_accuracy": 0.9717454612255096,
+      "num_tokens": 9169234.0,
+      "step": 1082
+    },
+    {
+      "epoch": 2.484500574052813,
+      "grad_norm": 2.1435959824494244,
+      "learning_rate": 1.7942123312251448e-06,
+      "loss": 0.1182,
+      "mean_token_accuracy": 0.9673901796340942,
+      "num_tokens": 9177970.0,
+      "step": 1083
+    },
+    {
+      "epoch": 2.486796785304248,
+      "grad_norm": 1.9209654624318502,
+      "learning_rate": 1.7874115442400303e-06,
+      "loss": 0.1067,
+      "mean_token_accuracy": 0.9688851833343506,
+      "num_tokens": 9186474.0,
+      "step": 1084
+    },
+    {
+      "epoch": 2.489092996555683,
+      "grad_norm": 1.9087655818616733,
+      "learning_rate": 1.7806372071145588e-06,
+      "loss": 0.1026,
+      "mean_token_accuracy": 0.9738331437110901,
+      "num_tokens": 9195098.0,
+      "step": 1085
+    },
+    {
+      "epoch": 2.491389207807118,
+      "grad_norm": 1.8124847645199362,
+      "learning_rate": 1.7738893681116237e-06,
+      "loss": 0.0951,
+      "mean_token_accuracy": 0.9721837341785431,
+      "num_tokens": 9204531.0,
+      "step": 1086
+    },
+    {
+      "epoch": 2.4936854190585533,
+      "grad_norm": 1.7918668013772585,
+      "learning_rate": 1.7671680753053384e-06,
+      "loss": 0.0921,
+      "mean_token_accuracy": 0.9745201170444489,
+      "num_tokens": 9213233.0,
+      "step": 1087
+    },
+    {
+      "epoch": 2.4959816303099887,
+      "grad_norm": 1.5872955781197515,
+      "learning_rate": 1.7604733765806903e-06,
+      "loss": 0.09,
+      "mean_token_accuracy": 0.9767709076404572,
+      "num_tokens": 9222517.0,
+      "step": 1088
+    },
+    {
+      "epoch": 2.4982778415614235,
+      "grad_norm": 1.7083526460132028,
+      "learning_rate": 1.7538053196331988e-06,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.975581169128418,
+      "num_tokens": 9231671.0,
+      "step": 1089
+    },
+    {
+      "epoch": 2.500574052812859,
+      "grad_norm": 2.012103802005448,
+      "learning_rate": 1.7471639519685807e-06,
+      "loss": 0.1108,
+      "mean_token_accuracy": 0.9691286981105804,
+      "num_tokens": 9240252.0,
+      "step": 1090
+    },
+    {
+      "epoch": 2.5028702640642937,
+      "grad_norm": 1.8875622251158413,
+      "learning_rate": 1.7405493209024072e-06,
+      "loss": 0.0913,
+      "mean_token_accuracy": 0.9746182560920715,
+      "num_tokens": 9248712.0,
+      "step": 1091
+    },
+    {
+      "epoch": 2.505166475315729,
+      "grad_norm": 1.7401326431500403,
+      "learning_rate": 1.7339614735597693e-06,
+      "loss": 0.0891,
+      "mean_token_accuracy": 0.9757702350616455,
+      "num_tokens": 9257002.0,
+      "step": 1092
+    },
+    {
+      "epoch": 2.5074626865671643,
+      "grad_norm": 1.7143781885494178,
+      "learning_rate": 1.7274004568749352e-06,
+      "loss": 0.0886,
+      "mean_token_accuracy": 0.975247323513031,
+      "num_tokens": 9265410.0,
+      "step": 1093
+    },
+    {
+      "epoch": 2.509758897818599,
+      "grad_norm": 1.5617851963378373,
+      "learning_rate": 1.7208663175910285e-06,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9773168563842773,
+      "num_tokens": 9273702.0,
+      "step": 1094
+    },
+    {
+      "epoch": 2.5120551090700345,
+      "grad_norm": 1.9319876811691554,
+      "learning_rate": 1.7143591022596846e-06,
+      "loss": 0.1058,
+      "mean_token_accuracy": 0.9697202444076538,
+      "num_tokens": 9282077.0,
+      "step": 1095
+    },
+    {
+      "epoch": 2.5143513203214694,
+      "grad_norm": 1.8096808405391662,
+      "learning_rate": 1.707878857240724e-06,
+      "loss": 0.097,
+      "mean_token_accuracy": 0.9726605117321014,
+      "num_tokens": 9291278.0,
+      "step": 1096
+    },
+    {
+      "epoch": 2.5166475315729047,
+      "grad_norm": 1.6987063680481191,
+      "learning_rate": 1.7014256287018205e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.973389059305191,
+      "num_tokens": 9299771.0,
+      "step": 1097
+    },
+    {
+      "epoch": 2.51894374282434,
+      "grad_norm": 1.5960946979798578,
+      "learning_rate": 1.6949994626181698e-06,
+      "loss": 0.0882,
+      "mean_token_accuracy": 0.9761817455291748,
+      "num_tokens": 9308062.0,
+      "step": 1098
+    },
+    {
+      "epoch": 2.521239954075775,
+      "grad_norm": 1.6012290037557055,
+      "learning_rate": 1.6886004047721698e-06,
+      "loss": 0.0932,
+      "mean_token_accuracy": 0.9723012745380402,
+      "num_tokens": 9316636.0,
+      "step": 1099
+    },
+    {
+      "epoch": 2.52353616532721,
+      "grad_norm": 1.815312261234719,
+      "learning_rate": 1.682228500753083e-06,
+      "loss": 0.0873,
+      "mean_token_accuracy": 0.9740700423717499,
+      "num_tokens": 9325040.0,
+      "step": 1100
+    },
+    {
+      "epoch": 2.525832376578645,
+      "grad_norm": 1.6448849556255118,
+      "learning_rate": 1.6758837959567245e-06,
+      "loss": 0.0968,
+      "mean_token_accuracy": 0.9712449610233307,
+      "num_tokens": 9333807.0,
+      "step": 1101
+    },
+    {
+      "epoch": 2.5281285878300803,
+      "grad_norm": 1.9342529931299182,
+      "learning_rate": 1.6695663355851255e-06,
+      "loss": 0.0888,
+      "mean_token_accuracy": 0.9742760062217712,
+      "num_tokens": 9342150.0,
+      "step": 1102
+    },
+    {
+      "epoch": 2.5304247990815156,
+      "grad_norm": 2.0800471027267946,
+      "learning_rate": 1.6632761646462247e-06,
+      "loss": 0.1063,
+      "mean_token_accuracy": 0.9715096354484558,
+      "num_tokens": 9350155.0,
+      "step": 1103
+    },
+    {
+      "epoch": 2.5327210103329505,
+      "grad_norm": 1.9725997425033863,
+      "learning_rate": 1.6570133279535335e-06,
+      "loss": 0.0951,
+      "mean_token_accuracy": 0.974311351776123,
+      "num_tokens": 9358787.0,
+      "step": 1104
+    },
+    {
+      "epoch": 2.535017221584386,
+      "grad_norm": 1.9183569786046375,
+      "learning_rate": 1.6507778701258309e-06,
+      "loss": 0.093,
+      "mean_token_accuracy": 0.9745320379734039,
+      "num_tokens": 9367451.0,
+      "step": 1105
+    },
+    {
+      "epoch": 2.5373134328358207,
+      "grad_norm": 1.8477001872438539,
+      "learning_rate": 1.6445698355868323e-06,
+      "loss": 0.0908,
+      "mean_token_accuracy": 0.9748466312885284,
+      "num_tokens": 9375998.0,
+      "step": 1106
+    },
+    {
+      "epoch": 2.539609644087256,
+      "grad_norm": 1.9487136762069042,
+      "learning_rate": 1.638389268564885e-06,
+      "loss": 0.101,
+      "mean_token_accuracy": 0.9727326035499573,
+      "num_tokens": 9383807.0,
+      "step": 1107
+    },
+    {
+      "epoch": 2.5419058553386913,
+      "grad_norm": 2.1523519301206107,
+      "learning_rate": 1.6322362130926427e-06,
+      "loss": 0.0931,
+      "mean_token_accuracy": 0.973209947347641,
+      "num_tokens": 9392507.0,
+      "step": 1108
+    },
+    {
+      "epoch": 2.544202066590126,
+      "grad_norm": 1.679860894523202,
+      "learning_rate": 1.6261107130067616e-06,
+      "loss": 0.0835,
+      "mean_token_accuracy": 0.9739972949028015,
+      "num_tokens": 9401485.0,
+      "step": 1109
+    },
+    {
+      "epoch": 2.5464982778415615,
+      "grad_norm": 2.128501039430456,
+      "learning_rate": 1.6200128119475787e-06,
+      "loss": 0.0893,
+      "mean_token_accuracy": 0.974321186542511,
+      "num_tokens": 9410385.0,
+      "step": 1110
+    },
+    {
+      "epoch": 2.5487944890929963,
+      "grad_norm": 1.7949374104618459,
+      "learning_rate": 1.6139425533588055e-06,
+      "loss": 0.0974,
+      "mean_token_accuracy": 0.9730968177318573,
+      "num_tokens": 9418226.0,
+      "step": 1111
+    },
+    {
+      "epoch": 2.5510907003444316,
+      "grad_norm": 1.6739510263526003,
+      "learning_rate": 1.6078999804872214e-06,
+      "loss": 0.0888,
+      "mean_token_accuracy": 0.9749470949172974,
+      "num_tokens": 9426054.0,
+      "step": 1112
+    },
+    {
+      "epoch": 2.553386911595867,
+      "grad_norm": 1.5394648191009563,
+      "learning_rate": 1.601885136382361e-06,
+      "loss": 0.0884,
+      "mean_token_accuracy": 0.9757075905799866,
+      "num_tokens": 9434853.0,
+      "step": 1113
+    },
+    {
+      "epoch": 2.555683122847302,
+      "grad_norm": 2.0130291153539654,
+      "learning_rate": 1.5958980638962099e-06,
+      "loss": 0.1018,
+      "mean_token_accuracy": 0.9710680544376373,
+      "num_tokens": 9443449.0,
+      "step": 1114
+    },
+    {
+      "epoch": 2.557979334098737,
+      "grad_norm": 1.7794400184921195,
+      "learning_rate": 1.5899388056828935e-06,
+      "loss": 0.0918,
+      "mean_token_accuracy": 0.9748867452144623,
+      "num_tokens": 9452425.0,
+      "step": 1115
+    },
+    {
+      "epoch": 2.560275545350172,
+      "grad_norm": 1.8956234892921622,
+      "learning_rate": 1.5840074041983835e-06,
+      "loss": 0.1054,
+      "mean_token_accuracy": 0.969905436038971,
+      "num_tokens": 9460869.0,
+      "step": 1116
+    },
+    {
+      "epoch": 2.5625717566016073,
+      "grad_norm": 1.8263719051623024,
+      "learning_rate": 1.5781039017001884e-06,
+      "loss": 0.1036,
+      "mean_token_accuracy": 0.971012532711029,
+      "num_tokens": 9469530.0,
+      "step": 1117
+    },
+    {
+      "epoch": 2.5648679678530426,
+      "grad_norm": 1.3605658314021898,
+      "learning_rate": 1.572228340247054e-06,
+      "loss": 0.0709,
+      "mean_token_accuracy": 0.980769544839859,
+      "num_tokens": 9478357.0,
+      "step": 1118
+    },
+    {
+      "epoch": 2.5671641791044775,
+      "grad_norm": 1.9334149811250738,
+      "learning_rate": 1.5663807616986628e-06,
+      "loss": 0.0958,
+      "mean_token_accuracy": 0.9725290536880493,
+      "num_tokens": 9486599.0,
+      "step": 1119
+    },
+    {
+      "epoch": 2.569460390355913,
+      "grad_norm": 2.0660108434663154,
+      "learning_rate": 1.5605612077153353e-06,
+      "loss": 0.1027,
+      "mean_token_accuracy": 0.9707440435886383,
+      "num_tokens": 9494774.0,
+      "step": 1120
+    },
+    {
+      "epoch": 2.5717566016073476,
+      "grad_norm": 1.8596422344033763,
+      "learning_rate": 1.554769719757739e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.974120020866394,
+      "num_tokens": 9503150.0,
+      "step": 1121
+    },
+    {
+      "epoch": 2.574052812858783,
+      "grad_norm": 1.7732313016725232,
+      "learning_rate": 1.5490063390865845e-06,
+      "loss": 0.0971,
+      "mean_token_accuracy": 0.9749862551689148,
+      "num_tokens": 9511065.0,
+      "step": 1122
+    },
+    {
+      "epoch": 2.5763490241102183,
+      "grad_norm": 1.6158215361998312,
+      "learning_rate": 1.5432711067623385e-06,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.9745330810546875,
+      "num_tokens": 9519345.0,
+      "step": 1123
+    },
+    {
+      "epoch": 2.578645235361653,
+      "grad_norm": 1.6368315621178922,
+      "learning_rate": 1.5375640636449253e-06,
+      "loss": 0.0891,
+      "mean_token_accuracy": 0.9747142493724823,
+      "num_tokens": 9527718.0,
+      "step": 1124
+    },
+    {
+      "epoch": 2.5809414466130884,
+      "grad_norm": 2.0402108566915045,
+      "learning_rate": 1.5318852503934425e-06,
+      "loss": 0.0837,
+      "mean_token_accuracy": 0.976736843585968,
+      "num_tokens": 9537234.0,
+      "step": 1125
+    },
+    {
+      "epoch": 2.5832376578645233,
+      "grad_norm": 1.7862147833624236,
+      "learning_rate": 1.526234707465864e-06,
+      "loss": 0.0904,
+      "mean_token_accuracy": 0.9734671413898468,
+      "num_tokens": 9545943.0,
+      "step": 1126
+    },
+    {
+      "epoch": 2.5855338691159586,
+      "grad_norm": 1.5337302608728125,
+      "learning_rate": 1.5206124751187595e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.9736779630184174,
+      "num_tokens": 9553891.0,
+      "step": 1127
+    },
+    {
+      "epoch": 2.587830080367394,
+      "grad_norm": 1.6608769277819104,
+      "learning_rate": 1.5150185934069988e-06,
+      "loss": 0.094,
+      "mean_token_accuracy": 0.9742108583450317,
+      "num_tokens": 9562679.0,
+      "step": 1128
+    },
+    {
+      "epoch": 2.590126291618829,
+      "grad_norm": 1.9931815779164552,
+      "learning_rate": 1.5094531021834754e-06,
+      "loss": 0.1048,
+      "mean_token_accuracy": 0.9705363214015961,
+      "num_tokens": 9571574.0,
+      "step": 1129
+    },
+    {
+      "epoch": 2.592422502870264,
+      "grad_norm": 2.2208511162965503,
+      "learning_rate": 1.5039160410988157e-06,
+      "loss": 0.0999,
+      "mean_token_accuracy": 0.9737870991230011,
+      "num_tokens": 9580075.0,
+      "step": 1130
+    },
+    {
+      "epoch": 2.594718714121699,
+      "grad_norm": 1.85320043070192,
+      "learning_rate": 1.4984074496011013e-06,
+      "loss": 0.0991,
+      "mean_token_accuracy": 0.9710549712181091,
+      "num_tokens": 9588510.0,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5970149253731343,
+      "grad_norm": 1.6747052242931202,
+      "learning_rate": 1.4929273669355846e-06,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9762501120567322,
+      "num_tokens": 9597716.0,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5993111366245696,
+      "grad_norm": 1.8529782872851952,
+      "learning_rate": 1.4874758321444091e-06,
+      "loss": 0.1037,
+      "mean_token_accuracy": 0.9718884229660034,
+      "num_tokens": 9606059.0,
+      "step": 1133
+    },
+    {
+      "epoch": 2.6016073478760045,
+      "grad_norm": 1.8079680632368718,
+      "learning_rate": 1.4820528840663364e-06,
+      "loss": 0.0949,
+      "mean_token_accuracy": 0.9715236127376556,
+      "num_tokens": 9614414.0,
+      "step": 1134
+    },
+    {
+      "epoch": 2.6039035591274398,
+      "grad_norm": 1.9651040429604154,
+      "learning_rate": 1.476658561336462e-06,
+      "loss": 0.0957,
+      "mean_token_accuracy": 0.97209033370018,
+      "num_tokens": 9622584.0,
+      "step": 1135
+    },
+    {
+      "epoch": 2.6061997703788746,
+      "grad_norm": 1.8630600424979433,
+      "learning_rate": 1.4712929023859487e-06,
+      "loss": 0.1105,
+      "mean_token_accuracy": 0.969320684671402,
+      "num_tokens": 9631585.0,
+      "step": 1136
+    },
+    {
+      "epoch": 2.60849598163031,
+      "grad_norm": 1.8534517143803337,
+      "learning_rate": 1.4659559454417402e-06,
+      "loss": 0.1009,
+      "mean_token_accuracy": 0.9724145531654358,
+      "num_tokens": 9639869.0,
+      "step": 1137
+    },
+    {
+      "epoch": 2.6107921928817452,
+      "grad_norm": 1.9567267692109327,
+      "learning_rate": 1.4606477285263024e-06,
+      "loss": 0.1066,
+      "mean_token_accuracy": 0.970019668340683,
+      "num_tokens": 9648995.0,
+      "step": 1138
+    },
+    {
+      "epoch": 2.61308840413318,
+      "grad_norm": 1.9064416403419044,
+      "learning_rate": 1.4553682894573454e-06,
+      "loss": 0.0939,
+      "mean_token_accuracy": 0.9729534685611725,
+      "num_tokens": 9657520.0,
+      "step": 1139
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 1.800587910797799,
+      "learning_rate": 1.4501176658475555e-06,
+      "loss": 0.0912,
+      "mean_token_accuracy": 0.9731960296630859,
+      "num_tokens": 9665465.0,
+      "step": 1140
+    },
+    {
+      "epoch": 2.6176808266360503,
+      "grad_norm": 1.7183510831691342,
+      "learning_rate": 1.4448958951043256e-06,
+      "loss": 0.085,
+      "mean_token_accuracy": 0.9752494692802429,
+      "num_tokens": 9673836.0,
+      "step": 1141
+    },
+    {
+      "epoch": 2.6199770378874856,
+      "grad_norm": 1.5922180313199095,
+      "learning_rate": 1.4397030144294896e-06,
+      "loss": 0.0882,
+      "mean_token_accuracy": 0.9733399152755737,
+      "num_tokens": 9682101.0,
+      "step": 1142
+    },
+    {
+      "epoch": 2.622273249138921,
+      "grad_norm": 1.964868345012733,
+      "learning_rate": 1.4345390608190608e-06,
+      "loss": 0.1014,
+      "mean_token_accuracy": 0.9714016318321228,
+      "num_tokens": 9690532.0,
+      "step": 1143
+    },
+    {
+      "epoch": 2.6245694603903558,
+      "grad_norm": 1.9408897229847113,
+      "learning_rate": 1.4294040710629617e-06,
+      "loss": 0.1028,
+      "mean_token_accuracy": 0.9705975949764252,
+      "num_tokens": 9698418.0,
+      "step": 1144
+    },
+    {
+      "epoch": 2.626865671641791,
+      "grad_norm": 1.5608063910378827,
+      "learning_rate": 1.4242980817447694e-06,
+      "loss": 0.0838,
+      "mean_token_accuracy": 0.9743890166282654,
+      "num_tokens": 9706783.0,
+      "step": 1145
+    },
+    {
+      "epoch": 2.629161882893226,
+      "grad_norm": 1.7389696515802169,
+      "learning_rate": 1.4192211292414465e-06,
+      "loss": 0.0865,
+      "mean_token_accuracy": 0.9760904312133789,
+      "num_tokens": 9714852.0,
+      "step": 1146
+    },
+    {
+      "epoch": 2.6314580941446613,
+      "grad_norm": 1.8409066135340655,
+      "learning_rate": 1.4141732497230917e-06,
+      "loss": 0.0899,
+      "mean_token_accuracy": 0.9742074906826019,
+      "num_tokens": 9723720.0,
+      "step": 1147
+    },
+    {
+      "epoch": 2.6337543053960966,
+      "grad_norm": 1.8442165310232357,
+      "learning_rate": 1.4091544791526725e-06,
+      "loss": 0.0954,
+      "mean_token_accuracy": 0.9739434719085693,
+      "num_tokens": 9732608.0,
+      "step": 1148
+    },
+    {
+      "epoch": 2.6360505166475314,
+      "grad_norm": 1.7935554575745583,
+      "learning_rate": 1.4041648532857763e-06,
+      "loss": 0.0918,
+      "mean_token_accuracy": 0.9730084538459778,
+      "num_tokens": 9740954.0,
+      "step": 1149
+    },
+    {
+      "epoch": 2.6383467278989667,
+      "grad_norm": 1.624803839752454,
+      "learning_rate": 1.3992044076703513e-06,
+      "loss": 0.0956,
+      "mean_token_accuracy": 0.9734368026256561,
+      "num_tokens": 9749624.0,
+      "step": 1150
+    },
+    {
+      "epoch": 2.6406429391504016,
+      "grad_norm": 1.7129234730832763,
+      "learning_rate": 1.394273177646454e-06,
+      "loss": 0.0929,
+      "mean_token_accuracy": 0.9734599888324738,
+      "num_tokens": 9757730.0,
+      "step": 1151
+    },
+    {
+      "epoch": 2.642939150401837,
+      "grad_norm": 2.3007304447434866,
+      "learning_rate": 1.389371198346001e-06,
+      "loss": 0.0989,
+      "mean_token_accuracy": 0.9725444614887238,
+      "num_tokens": 9765747.0,
+      "step": 1152
+    },
+    {
+      "epoch": 2.645235361653272,
+      "grad_norm": 1.7550149037993796,
+      "learning_rate": 1.3844985046925148e-06,
+      "loss": 0.0994,
+      "mean_token_accuracy": 0.9725069999694824,
+      "num_tokens": 9773992.0,
+      "step": 1153
+    },
+    {
+      "epoch": 2.647531572904707,
+      "grad_norm": 2.094446957532105,
+      "learning_rate": 1.3796551314008747e-06,
+      "loss": 0.1029,
+      "mean_token_accuracy": 0.969898134469986,
+      "num_tokens": 9782522.0,
+      "step": 1154
+    },
+    {
+      "epoch": 2.6498277841561424,
+      "grad_norm": 1.809089693301345,
+      "learning_rate": 1.3748411129770703e-06,
+      "loss": 0.0996,
+      "mean_token_accuracy": 0.9713851511478424,
+      "num_tokens": 9791012.0,
+      "step": 1155
+    },
+    {
+      "epoch": 2.6521239954075773,
+      "grad_norm": 1.5693532889510444,
+      "learning_rate": 1.370056483717959e-06,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9755423963069916,
+      "num_tokens": 9801265.0,
+      "step": 1156
+    },
+    {
+      "epoch": 2.6544202066590126,
+      "grad_norm": 1.7303846077711806,
+      "learning_rate": 1.3653012777110172e-06,
+      "loss": 0.0886,
+      "mean_token_accuracy": 0.9759488999843597,
+      "num_tokens": 9809828.0,
+      "step": 1157
+    },
+    {
+      "epoch": 2.656716417910448,
+      "grad_norm": 2.0012216399967704,
+      "learning_rate": 1.3605755288341014e-06,
+      "loss": 0.1044,
+      "mean_token_accuracy": 0.9692345857620239,
+      "num_tokens": 9817631.0,
+      "step": 1158
+    },
+    {
+      "epoch": 2.6590126291618827,
+      "grad_norm": 1.7280227739386638,
+      "learning_rate": 1.355879270755199e-06,
+      "loss": 0.0881,
+      "mean_token_accuracy": 0.9756309390068054,
+      "num_tokens": 9825900.0,
+      "step": 1159
+    },
+    {
+      "epoch": 2.661308840413318,
+      "grad_norm": 1.832227795048563,
+      "learning_rate": 1.3512125369321994e-06,
+      "loss": 0.0947,
+      "mean_token_accuracy": 0.9712029695510864,
+      "num_tokens": 9834518.0,
+      "step": 1160
+    },
+    {
+      "epoch": 2.663605051664753,
+      "grad_norm": 1.6431672918457534,
+      "learning_rate": 1.3465753606126483e-06,
+      "loss": 0.0944,
+      "mean_token_accuracy": 0.972135990858078,
+      "num_tokens": 9842953.0,
+      "step": 1161
+    },
+    {
+      "epoch": 2.6659012629161882,
+      "grad_norm": 1.9785590215310875,
+      "learning_rate": 1.3419677748335122e-06,
+      "loss": 0.0979,
+      "mean_token_accuracy": 0.9717593193054199,
+      "num_tokens": 9851478.0,
+      "step": 1162
+    },
+    {
+      "epoch": 2.6681974741676235,
+      "grad_norm": 1.939431811011092,
+      "learning_rate": 1.3373898124209455e-06,
+      "loss": 0.1022,
+      "mean_token_accuracy": 0.9725072383880615,
+      "num_tokens": 9860530.0,
+      "step": 1163
+    },
+    {
+      "epoch": 2.6704936854190584,
+      "grad_norm": 1.944867943696077,
+      "learning_rate": 1.332841505990052e-06,
+      "loss": 0.0884,
+      "mean_token_accuracy": 0.9749001562595367,
+      "num_tokens": 9868965.0,
+      "step": 1164
+    },
+    {
+      "epoch": 2.6727898966704937,
+      "grad_norm": 1.766965056593969,
+      "learning_rate": 1.3283228879446583e-06,
+      "loss": 0.0938,
+      "mean_token_accuracy": 0.9724652767181396,
+      "num_tokens": 9877727.0,
+      "step": 1165
+    },
+    {
+      "epoch": 2.6750861079219286,
+      "grad_norm": 1.828059360988063,
+      "learning_rate": 1.323833990477076e-06,
+      "loss": 0.0944,
+      "mean_token_accuracy": 0.9737249910831451,
+      "num_tokens": 9886261.0,
+      "step": 1166
+    },
+    {
+      "epoch": 2.677382319173364,
+      "grad_norm": 1.832847666848248,
+      "learning_rate": 1.3193748455678815e-06,
+      "loss": 0.0877,
+      "mean_token_accuracy": 0.9737261831760406,
+      "num_tokens": 9894444.0,
+      "step": 1167
+    },
+    {
+      "epoch": 2.679678530424799,
+      "grad_norm": 1.9551322547547938,
+      "learning_rate": 1.3149454849856771e-06,
+      "loss": 0.1004,
+      "mean_token_accuracy": 0.9722377359867096,
+      "num_tokens": 9903106.0,
+      "step": 1168
+    },
+    {
+      "epoch": 2.681974741676234,
+      "grad_norm": 1.6478796741865251,
+      "learning_rate": 1.310545940286876e-06,
+      "loss": 0.0812,
+      "mean_token_accuracy": 0.9772592186927795,
+      "num_tokens": 9911648.0,
+      "step": 1169
+    },
+    {
+      "epoch": 2.6842709529276694,
+      "grad_norm": 1.97565586672626,
+      "learning_rate": 1.3061762428154675e-06,
+      "loss": 0.1066,
+      "mean_token_accuracy": 0.9718965291976929,
+      "num_tokens": 9919679.0,
+      "step": 1170
+    },
+    {
+      "epoch": 2.6865671641791042,
+      "grad_norm": 1.871614743612357,
+      "learning_rate": 1.3018364237028013e-06,
+      "loss": 0.0846,
+      "mean_token_accuracy": 0.9769338071346283,
+      "num_tokens": 9928054.0,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6888633754305395,
+      "grad_norm": 1.84463266833113,
+      "learning_rate": 1.2975265138673609e-06,
+      "loss": 0.0972,
+      "mean_token_accuracy": 0.9719581305980682,
+      "num_tokens": 9936479.0,
+      "step": 1172
+    },
+    {
+      "epoch": 2.691159586681975,
+      "grad_norm": 1.862137388762856,
+      "learning_rate": 1.2932465440145444e-06,
+      "loss": 0.0924,
+      "mean_token_accuracy": 0.9735874831676483,
+      "num_tokens": 9944453.0,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6934557979334097,
+      "grad_norm": 1.754072605988632,
+      "learning_rate": 1.2889965446364483e-06,
+      "loss": 0.0947,
+      "mean_token_accuracy": 0.9703351259231567,
+      "num_tokens": 9952996.0,
+      "step": 1174
+    },
+    {
+      "epoch": 2.695752009184845,
+      "grad_norm": 1.6593807438074595,
+      "learning_rate": 1.284776546011648e-06,
+      "loss": 0.0976,
+      "mean_token_accuracy": 0.9721977412700653,
+      "num_tokens": 9961260.0,
+      "step": 1175
+    },
+    {
+      "epoch": 2.69804822043628,
+      "grad_norm": 1.572619752076202,
+      "learning_rate": 1.2805865782049806e-06,
+      "loss": 0.088,
+      "mean_token_accuracy": 0.974699467420578,
+      "num_tokens": 9969491.0,
+      "step": 1176
+    },
+    {
+      "epoch": 2.700344431687715,
+      "grad_norm": 1.6756736919780337,
+      "learning_rate": 1.2764266710673335e-06,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9780433177947998,
+      "num_tokens": 9977867.0,
+      "step": 1177
+    },
+    {
+      "epoch": 2.7026406429391505,
+      "grad_norm": 1.575346101257233,
+      "learning_rate": 1.272296854235431e-06,
+      "loss": 0.0924,
+      "mean_token_accuracy": 0.9762708246707916,
+      "num_tokens": 9985969.0,
+      "step": 1178
+    },
+    {
+      "epoch": 2.7049368541905854,
+      "grad_norm": 1.6873928459253107,
+      "learning_rate": 1.2681971571316234e-06,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.9752933084964752,
+      "num_tokens": 9994057.0,
+      "step": 1179
+    },
+    {
+      "epoch": 2.7072330654420207,
+      "grad_norm": 2.40334118404975,
+      "learning_rate": 1.2641276089636769e-06,
+      "loss": 0.1156,
+      "mean_token_accuracy": 0.9656489491462708,
+      "num_tokens": 10001996.0,
+      "step": 1180
+    },
+    {
+      "epoch": 2.7095292766934556,
+      "grad_norm": 2.231025984173633,
+      "learning_rate": 1.2600882387245625e-06,
+      "loss": 0.1164,
+      "mean_token_accuracy": 0.9670365750789642,
+      "num_tokens": 10010133.0,
+      "step": 1181
+    },
+    {
+      "epoch": 2.711825487944891,
+      "grad_norm": 1.6473298752721197,
+      "learning_rate": 1.2560790751922559e-06,
+      "loss": 0.0874,
+      "mean_token_accuracy": 0.9736159443855286,
+      "num_tokens": 10018466.0,
+      "step": 1182
+    },
+    {
+      "epoch": 2.714121699196326,
+      "grad_norm": 1.6152745019121313,
+      "learning_rate": 1.252100146929528e-06,
+      "loss": 0.088,
+      "mean_token_accuracy": 0.9741679728031158,
+      "num_tokens": 10026856.0,
+      "step": 1183
+    },
+    {
+      "epoch": 2.716417910447761,
+      "grad_norm": 1.8602485421080206,
+      "learning_rate": 1.2481514822837414e-06,
+      "loss": 0.0896,
+      "mean_token_accuracy": 0.9727961719036102,
+      "num_tokens": 10035240.0,
+      "step": 1184
+    },
+    {
+      "epoch": 2.7187141216991964,
+      "grad_norm": 2.003756730327132,
+      "learning_rate": 1.2442331093866517e-06,
+      "loss": 0.0996,
+      "mean_token_accuracy": 0.9727631509304047,
+      "num_tokens": 10043861.0,
+      "step": 1185
+    },
+    {
+      "epoch": 2.721010332950631,
+      "grad_norm": 1.6699885457808092,
+      "learning_rate": 1.2403450561542017e-06,
+      "loss": 0.0948,
+      "mean_token_accuracy": 0.9725164771080017,
+      "num_tokens": 10052220.0,
+      "step": 1186
+    },
+    {
+      "epoch": 2.7233065442020665,
+      "grad_norm": 1.7609723377928315,
+      "learning_rate": 1.2364873502863287e-06,
+      "loss": 0.0946,
+      "mean_token_accuracy": 0.9723201394081116,
+      "num_tokens": 10060917.0,
+      "step": 1187
+    },
+    {
+      "epoch": 2.725602755453502,
+      "grad_norm": 1.6275365880619772,
+      "learning_rate": 1.2326600192667612e-06,
+      "loss": 0.0881,
+      "mean_token_accuracy": 0.9740128517150879,
+      "num_tokens": 10068744.0,
+      "step": 1188
+    },
+    {
+      "epoch": 2.7278989667049367,
+      "grad_norm": 1.7899415237398262,
+      "learning_rate": 1.2288630903628293e-06,
+      "loss": 0.0939,
+      "mean_token_accuracy": 0.9739814400672913,
+      "num_tokens": 10076917.0,
+      "step": 1189
+    },
+    {
+      "epoch": 2.730195177956372,
+      "grad_norm": 1.523390324270752,
+      "learning_rate": 1.2250965906252635e-06,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9760325849056244,
+      "num_tokens": 10086251.0,
+      "step": 1190
+    },
+    {
+      "epoch": 2.732491389207807,
+      "grad_norm": 1.5699315963603524,
+      "learning_rate": 1.221360546888009e-06,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9749657809734344,
+      "num_tokens": 10094919.0,
+      "step": 1191
+    },
+    {
+      "epoch": 2.734787600459242,
+      "grad_norm": 1.7939795468706603,
+      "learning_rate": 1.2176549857680286e-06,
+      "loss": 0.0902,
+      "mean_token_accuracy": 0.9752355515956879,
+      "num_tokens": 10103295.0,
+      "step": 1192
+    },
+    {
+      "epoch": 2.7370838117106775,
+      "grad_norm": 2.0566004071652975,
+      "learning_rate": 1.213979933665117e-06,
+      "loss": 0.1028,
+      "mean_token_accuracy": 0.9716645181179047,
+      "num_tokens": 10111558.0,
+      "step": 1193
+    },
+    {
+      "epoch": 2.7393800229621124,
+      "grad_norm": 1.6359066168630447,
+      "learning_rate": 1.2103354167617113e-06,
+      "loss": 0.0851,
+      "mean_token_accuracy": 0.9766933917999268,
+      "num_tokens": 10120394.0,
+      "step": 1194
+    },
+    {
+      "epoch": 2.7416762342135477,
+      "grad_norm": 1.6326074540213966,
+      "learning_rate": 1.2067214610227029e-06,
+      "loss": 0.0912,
+      "mean_token_accuracy": 0.9756397604942322,
+      "num_tokens": 10128950.0,
+      "step": 1195
+    },
+    {
+      "epoch": 2.7439724454649825,
+      "grad_norm": 1.7932360018011757,
+      "learning_rate": 1.2031380921952558e-06,
+      "loss": 0.0983,
+      "mean_token_accuracy": 0.9708451628684998,
+      "num_tokens": 10137865.0,
+      "step": 1196
+    },
+    {
+      "epoch": 2.746268656716418,
+      "grad_norm": 1.744188442145005,
+      "learning_rate": 1.199585335808622e-06,
+      "loss": 0.0992,
+      "mean_token_accuracy": 0.9736141264438629,
+      "num_tokens": 10145921.0,
+      "step": 1197
+    },
+    {
+      "epoch": 2.748564867967853,
+      "grad_norm": 1.6980637475690457,
+      "learning_rate": 1.1960632171739574e-06,
+      "loss": 0.0997,
+      "mean_token_accuracy": 0.9711240530014038,
+      "num_tokens": 10154314.0,
+      "step": 1198
+    },
+    {
+      "epoch": 2.750861079219288,
+      "grad_norm": 1.677811372044132,
+      "learning_rate": 1.1925717613841432e-06,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.9752368330955505,
+      "num_tokens": 10162826.0,
+      "step": 1199
+    },
+    {
+      "epoch": 2.7531572904707233,
+      "grad_norm": 1.8250836281248486,
+      "learning_rate": 1.1891109933136084e-06,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.9772456288337708,
+      "num_tokens": 10171093.0,
+      "step": 1200
+    },
+    {
+      "epoch": 2.755453501722158,
+      "grad_norm": 2.00414285841927,
+      "learning_rate": 1.1856809376181512e-06,
+      "loss": 0.0997,
+      "mean_token_accuracy": 0.9703491628170013,
+      "num_tokens": 10179195.0,
+      "step": 1201
+    },
+    {
+      "epoch": 2.7577497129735935,
+      "grad_norm": 1.7816657587382503,
+      "learning_rate": 1.1822816187347625e-06,
+      "loss": 0.1044,
+      "mean_token_accuracy": 0.9707182347774506,
+      "num_tokens": 10187119.0,
+      "step": 1202
+    },
+    {
+      "epoch": 2.760045924225029,
+      "grad_norm": 1.6273236691022577,
+      "learning_rate": 1.178913060881452e-06,
+      "loss": 0.0759,
+      "mean_token_accuracy": 0.9784926176071167,
+      "num_tokens": 10195891.0,
+      "step": 1203
+    },
+    {
+      "epoch": 2.7623421354764637,
+      "grad_norm": 2.0162065143753054,
+      "learning_rate": 1.1755752880570785e-06,
+      "loss": 0.0899,
+      "mean_token_accuracy": 0.9741425216197968,
+      "num_tokens": 10203749.0,
+      "step": 1204
+    },
+    {
+      "epoch": 2.764638346727899,
+      "grad_norm": 1.9610381549473486,
+      "learning_rate": 1.1722683240411765e-06,
+      "loss": 0.095,
+      "mean_token_accuracy": 0.9720205962657928,
+      "num_tokens": 10212639.0,
+      "step": 1205
+    },
+    {
+      "epoch": 2.766934557979334,
+      "grad_norm": 1.801096197360057,
+      "learning_rate": 1.1689921923937852e-06,
+      "loss": 0.0905,
+      "mean_token_accuracy": 0.972496896982193,
+      "num_tokens": 10221365.0,
+      "step": 1206
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 1.7756096188788224,
+      "learning_rate": 1.1657469164552839e-06,
+      "loss": 0.0913,
+      "mean_token_accuracy": 0.9753103852272034,
+      "num_tokens": 10229734.0,
+      "step": 1207
+    },
+    {
+      "epoch": 2.7715269804822045,
+      "grad_norm": 1.7148291751796443,
+      "learning_rate": 1.1625325193462245e-06,
+      "loss": 0.0879,
+      "mean_token_accuracy": 0.9748569428920746,
+      "num_tokens": 10237972.0,
+      "step": 1208
+    },
+    {
+      "epoch": 2.7738231917336393,
+      "grad_norm": 2.384918621017701,
+      "learning_rate": 1.159349023967167e-06,
+      "loss": 0.1168,
+      "mean_token_accuracy": 0.9698925614356995,
+      "num_tokens": 10245822.0,
+      "step": 1209
+    },
+    {
+      "epoch": 2.7761194029850746,
+      "grad_norm": 1.602955363153048,
+      "learning_rate": 1.1561964529985143e-06,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.97429558634758,
+      "num_tokens": 10254124.0,
+      "step": 1210
+    },
+    {
+      "epoch": 2.7784156142365095,
+      "grad_norm": 1.6943540523861989,
+      "learning_rate": 1.1530748289003543e-06,
+      "loss": 0.0858,
+      "mean_token_accuracy": 0.9763672351837158,
+      "num_tokens": 10262376.0,
+      "step": 1211
+    },
+    {
+      "epoch": 2.780711825487945,
+      "grad_norm": 1.668980915808961,
+      "learning_rate": 1.1499841739122959e-06,
+      "loss": 0.0843,
+      "mean_token_accuracy": 0.9749711155891418,
+      "num_tokens": 10271184.0,
+      "step": 1212
+    },
+    {
+      "epoch": 2.78300803673938,
+      "grad_norm": 1.671742717825574,
+      "learning_rate": 1.1469245100533147e-06,
+      "loss": 0.0973,
+      "mean_token_accuracy": 0.9734241962432861,
+      "num_tokens": 10280809.0,
+      "step": 1213
+    },
+    {
+      "epoch": 2.785304247990815,
+      "grad_norm": 1.772672260726772,
+      "learning_rate": 1.1438958591215925e-06,
+      "loss": 0.0816,
+      "mean_token_accuracy": 0.9778502583503723,
+      "num_tokens": 10289022.0,
+      "step": 1214
+    },
+    {
+      "epoch": 2.7876004592422503,
+      "grad_norm": 1.7021181209681657,
+      "learning_rate": 1.1408982426943639e-06,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748720526695251,
+      "num_tokens": 10297500.0,
+      "step": 1215
+    },
+    {
+      "epoch": 2.789896670493685,
+      "grad_norm": 1.4811589393752829,
+      "learning_rate": 1.137931682127762e-06,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9772430658340454,
+      "num_tokens": 10305445.0,
+      "step": 1216
+    },
+    {
+      "epoch": 2.7921928817451205,
+      "grad_norm": 1.6355835778084726,
+      "learning_rate": 1.1349961985566668e-06,
+      "loss": 0.0861,
+      "mean_token_accuracy": 0.9754639267921448,
+      "num_tokens": 10313924.0,
+      "step": 1217
+    },
+    {
+      "epoch": 2.794489092996556,
+      "grad_norm": 1.8337097058777998,
+      "learning_rate": 1.1320918128945533e-06,
+      "loss": 0.0898,
+      "mean_token_accuracy": 0.9734431505203247,
+      "num_tokens": 10323909.0,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7967853042479907,
+      "grad_norm": 1.8767743696321686,
+      "learning_rate": 1.1292185458333453e-06,
+      "loss": 0.1026,
+      "mean_token_accuracy": 0.9695652723312378,
+      "num_tokens": 10332555.0,
+      "step": 1219
+    },
+    {
+      "epoch": 2.799081515499426,
+      "grad_norm": 1.7665921404591804,
+      "learning_rate": 1.126376417843264e-06,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9766588807106018,
+      "num_tokens": 10340541.0,
+      "step": 1220
+    },
+    {
+      "epoch": 2.801377726750861,
+      "grad_norm": 1.6721785827565807,
+      "learning_rate": 1.1235654491726853e-06,
+      "loss": 0.0904,
+      "mean_token_accuracy": 0.9759955704212189,
+      "num_tokens": 10349025.0,
+      "step": 1221
+    },
+    {
+      "epoch": 2.803673938002296,
+      "grad_norm": 1.9732992636926885,
+      "learning_rate": 1.1207856598479953e-06,
+      "loss": 0.0922,
+      "mean_token_accuracy": 0.9739929437637329,
+      "num_tokens": 10357088.0,
+      "step": 1222
+    },
+    {
+      "epoch": 2.8059701492537314,
+      "grad_norm": 1.6395652824656266,
+      "learning_rate": 1.1180370696734456e-06,
+      "loss": 0.0915,
+      "mean_token_accuracy": 0.975504994392395,
+      "num_tokens": 10365630.0,
+      "step": 1223
+    },
+    {
+      "epoch": 2.8082663605051663,
+      "grad_norm": 1.7930497036084363,
+      "learning_rate": 1.1153196982310139e-06,
+      "loss": 0.0937,
+      "mean_token_accuracy": 0.9721637070178986,
+      "num_tokens": 10374248.0,
+      "step": 1224
+    },
+    {
+      "epoch": 2.8105625717566016,
+      "grad_norm": 1.6856979558295928,
+      "learning_rate": 1.1126335648802637e-06,
+      "loss": 0.0903,
+      "mean_token_accuracy": 0.9748069047927856,
+      "num_tokens": 10383808.0,
+      "step": 1225
+    },
+    {
+      "epoch": 2.8128587830080365,
+      "grad_norm": 1.822416154718355,
+      "learning_rate": 1.1099786887582067e-06,
+      "loss": 0.0886,
+      "mean_token_accuracy": 0.9766135513782501,
+      "num_tokens": 10392100.0,
+      "step": 1226
+    },
+    {
+      "epoch": 2.815154994259472,
+      "grad_norm": 1.820775050119438,
+      "learning_rate": 1.1073550887791686e-06,
+      "loss": 0.0941,
+      "mean_token_accuracy": 0.9744702279567719,
+      "num_tokens": 10400557.0,
+      "step": 1227
+    },
+    {
+      "epoch": 2.817451205510907,
+      "grad_norm": 1.6893949344478256,
+      "learning_rate": 1.1047627836346487e-06,
+      "loss": 0.0852,
+      "mean_token_accuracy": 0.9763225317001343,
+      "num_tokens": 10408551.0,
+      "step": 1228
+    },
+    {
+      "epoch": 2.819747416762342,
+      "grad_norm": 1.9618813088128504,
+      "learning_rate": 1.102201791793193e-06,
+      "loss": 0.0963,
+      "mean_token_accuracy": 0.97347691655159,
+      "num_tokens": 10416804.0,
+      "step": 1229
+    },
+    {
+      "epoch": 2.8220436280137773,
+      "grad_norm": 1.7139934348707997,
+      "learning_rate": 1.0996721315002583e-06,
+      "loss": 0.0862,
+      "mean_token_accuracy": 0.976162850856781,
+      "num_tokens": 10425300.0,
+      "step": 1230
+    },
+    {
+      "epoch": 2.824339839265212,
+      "grad_norm": 1.8079474145455696,
+      "learning_rate": 1.0971738207780857e-06,
+      "loss": 0.0946,
+      "mean_token_accuracy": 0.9761255979537964,
+      "num_tokens": 10433889.0,
+      "step": 1231
+    },
+    {
+      "epoch": 2.8266360505166475,
+      "grad_norm": 1.7036594989993485,
+      "learning_rate": 1.0947068774255675e-06,
+      "loss": 0.0913,
+      "mean_token_accuracy": 0.9733274579048157,
+      "num_tokens": 10443096.0,
+      "step": 1232
+    },
+    {
+      "epoch": 2.8289322617680828,
+      "grad_norm": 1.7004788959139108,
+      "learning_rate": 1.0922713190181264e-06,
+      "loss": 0.0936,
+      "mean_token_accuracy": 0.9747470915317535,
+      "num_tokens": 10451622.0,
+      "step": 1233
+    },
+    {
+      "epoch": 2.8312284730195176,
+      "grad_norm": 1.9973882504840281,
+      "learning_rate": 1.0898671629075838e-06,
+      "loss": 0.0927,
+      "mean_token_accuracy": 0.9726047515869141,
+      "num_tokens": 10460078.0,
+      "step": 1234
+    },
+    {
+      "epoch": 2.833524684270953,
+      "grad_norm": 1.7325989336632206,
+      "learning_rate": 1.0874944262220428e-06,
+      "loss": 0.0876,
+      "mean_token_accuracy": 0.9761297106742859,
+      "num_tokens": 10468047.0,
+      "step": 1235
+    },
+    {
+      "epoch": 2.835820895522388,
+      "grad_norm": 1.674548135735757,
+      "learning_rate": 1.0851531258657597e-06,
+      "loss": 0.0855,
+      "mean_token_accuracy": 0.9756419062614441,
+      "num_tokens": 10476583.0,
+      "step": 1236
+    },
+    {
+      "epoch": 2.838117106773823,
+      "grad_norm": 1.7736784717840486,
+      "learning_rate": 1.082843278519029e-06,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.975772351026535,
+      "num_tokens": 10486085.0,
+      "step": 1237
+    },
+    {
+      "epoch": 2.8404133180252584,
+      "grad_norm": 1.747350989653511,
+      "learning_rate": 1.0805649006380617e-06,
+      "loss": 0.0888,
+      "mean_token_accuracy": 0.9747449457645416,
+      "num_tokens": 10493959.0,
+      "step": 1238
+    },
+    {
+      "epoch": 2.8427095292766933,
+      "grad_norm": 1.7023110237487589,
+      "learning_rate": 1.0783180084548677e-06,
+      "loss": 0.0941,
+      "mean_token_accuracy": 0.9730304181575775,
+      "num_tokens": 10502783.0,
+      "step": 1239
+    },
+    {
+      "epoch": 2.8450057405281286,
+      "grad_norm": 2.9597467945244915,
+      "learning_rate": 1.0761026179771423e-06,
+      "loss": 0.1351,
+      "mean_token_accuracy": 0.9664178788661957,
+      "num_tokens": 10512093.0,
+      "step": 1240
+    },
+    {
+      "epoch": 2.8473019517795635,
+      "grad_norm": 1.8218841934167085,
+      "learning_rate": 1.07391874498815e-06,
+      "loss": 0.094,
+      "mean_token_accuracy": 0.9733202457427979,
+      "num_tokens": 10521169.0,
+      "step": 1241
+    },
+    {
+      "epoch": 2.8495981630309988,
+      "grad_norm": 1.6690937494873967,
+      "learning_rate": 1.0717664050466138e-06,
+      "loss": 0.0772,
+      "mean_token_accuracy": 0.9777764678001404,
+      "num_tokens": 10529296.0,
+      "step": 1242
+    },
+    {
+      "epoch": 2.851894374282434,
+      "grad_norm": 1.798081304253357,
+      "learning_rate": 1.0696456134866027e-06,
+      "loss": 0.0903,
+      "mean_token_accuracy": 0.9758163392543793,
+      "num_tokens": 10537404.0,
+      "step": 1243
+    },
+    {
+      "epoch": 2.854190585533869,
+      "grad_norm": 1.6885902649159341,
+      "learning_rate": 1.0675563854174249e-06,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9756641089916229,
+      "num_tokens": 10546023.0,
+      "step": 1244
+    },
+    {
+      "epoch": 2.8564867967853043,
+      "grad_norm": 1.954362863475803,
+      "learning_rate": 1.065498735723516e-06,
+      "loss": 0.0963,
+      "mean_token_accuracy": 0.9731652736663818,
+      "num_tokens": 10554890.0,
+      "step": 1245
+    },
+    {
+      "epoch": 2.858783008036739,
+      "grad_norm": 1.8396736340750197,
+      "learning_rate": 1.0634726790643383e-06,
+      "loss": 0.0914,
+      "mean_token_accuracy": 0.9744046628475189,
+      "num_tokens": 10563214.0,
+      "step": 1246
+    },
+    {
+      "epoch": 2.8610792192881744,
+      "grad_norm": 2.082431595191121,
+      "learning_rate": 1.0614782298742712e-06,
+      "loss": 0.1125,
+      "mean_token_accuracy": 0.9673578143119812,
+      "num_tokens": 10571855.0,
+      "step": 1247
+    },
+    {
+      "epoch": 2.8633754305396097,
+      "grad_norm": 1.7739160987383606,
+      "learning_rate": 1.0595154023625123e-06,
+      "loss": 0.0957,
+      "mean_token_accuracy": 0.9706702530384064,
+      "num_tokens": 10580034.0,
+      "step": 1248
+    },
+    {
+      "epoch": 2.8656716417910446,
+      "grad_norm": 1.7889591443968462,
+      "learning_rate": 1.0575842105129743e-06,
+      "loss": 0.0912,
+      "mean_token_accuracy": 0.9731150269508362,
+      "num_tokens": 10588343.0,
+      "step": 1249
+    },
+    {
+      "epoch": 2.86796785304248,
+      "grad_norm": 1.7586376268205786,
+      "learning_rate": 1.0556846680841853e-06,
+      "loss": 0.0995,
+      "mean_token_accuracy": 0.9723497331142426,
+      "num_tokens": 10596445.0,
+      "step": 1250
+    },
+    {
+      "epoch": 2.870264064293915,
+      "grad_norm": 1.8087119868559267,
+      "learning_rate": 1.0538167886091913e-06,
+      "loss": 0.0905,
+      "mean_token_accuracy": 0.9779402315616608,
+      "num_tokens": 10604500.0,
+      "step": 1251
+    },
+    {
+      "epoch": 2.87256027554535,
+      "grad_norm": 1.846414541824757,
+      "learning_rate": 1.0519805853954596e-06,
+      "loss": 0.092,
+      "mean_token_accuracy": 0.9734601974487305,
+      "num_tokens": 10613299.0,
+      "step": 1252
+    },
+    {
+      "epoch": 2.8748564867967854,
+      "grad_norm": 1.979081923113397,
+      "learning_rate": 1.0501760715247846e-06,
+      "loss": 0.1046,
+      "mean_token_accuracy": 0.9701388478279114,
+      "num_tokens": 10622054.0,
+      "step": 1253
+    },
+    {
+      "epoch": 2.8771526980482207,
+      "grad_norm": 1.9927377429048534,
+      "learning_rate": 1.0484032598531933e-06,
+      "loss": 0.0932,
+      "mean_token_accuracy": 0.9749229252338409,
+      "num_tokens": 10630394.0,
+      "step": 1254
+    },
+    {
+      "epoch": 2.8794489092996556,
+      "grad_norm": 1.8193320286172596,
+      "learning_rate": 1.046662163010855e-06,
+      "loss": 0.1046,
+      "mean_token_accuracy": 0.9702285528182983,
+      "num_tokens": 10638476.0,
+      "step": 1255
+    },
+    {
+      "epoch": 2.8817451205510904,
+      "grad_norm": 1.7874168925321807,
+      "learning_rate": 1.04495279340199e-06,
+      "loss": 0.0949,
+      "mean_token_accuracy": 0.9733448922634125,
+      "num_tokens": 10647398.0,
+      "step": 1256
+    },
+    {
+      "epoch": 2.8840413318025258,
+      "grad_norm": 1.81830949393177,
+      "learning_rate": 1.043275163204783e-06,
+      "loss": 0.0887,
+      "mean_token_accuracy": 0.9746375381946564,
+      "num_tokens": 10656021.0,
+      "step": 1257
+    },
+    {
+      "epoch": 2.886337543053961,
+      "grad_norm": 1.6719296568221613,
+      "learning_rate": 1.0416292843712948e-06,
+      "loss": 0.0924,
+      "mean_token_accuracy": 0.9711298942565918,
+      "num_tokens": 10664648.0,
+      "step": 1258
+    },
+    {
+      "epoch": 2.8886337543053964,
+      "grad_norm": 1.838078110926522,
+      "learning_rate": 1.040015168627378e-06,
+      "loss": 0.0868,
+      "mean_token_accuracy": 0.97626131772995,
+      "num_tokens": 10673495.0,
+      "step": 1259
+    },
+    {
+      "epoch": 2.8909299655568312,
+      "grad_norm": 1.4374645072203425,
+      "learning_rate": 1.0384328274725932e-06,
+      "loss": 0.0768,
+      "mean_token_accuracy": 0.9784523546695709,
+      "num_tokens": 10682158.0,
+      "step": 1260
+    },
+    {
+      "epoch": 2.893226176808266,
+      "grad_norm": 1.5930876240673608,
+      "learning_rate": 1.0368822721801256e-06,
+      "loss": 0.0858,
+      "mean_token_accuracy": 0.9766460955142975,
+      "num_tokens": 10690348.0,
+      "step": 1261
+    },
+    {
+      "epoch": 2.8955223880597014,
+      "grad_norm": 2.1199563430658124,
+      "learning_rate": 1.0353635137967075e-06,
+      "loss": 0.1001,
+      "mean_token_accuracy": 0.9711456596851349,
+      "num_tokens": 10699428.0,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8978185993111367,
+      "grad_norm": 1.8186544491308645,
+      "learning_rate": 1.033876563142539e-06,
+      "loss": 0.0954,
+      "mean_token_accuracy": 0.9728469252586365,
+      "num_tokens": 10708012.0,
+      "step": 1263
+    },
+    {
+      "epoch": 2.900114810562572,
+      "grad_norm": 1.6592453492371337,
+      "learning_rate": 1.0324214308112082e-06,
+      "loss": 0.0947,
+      "mean_token_accuracy": 0.9729070961475372,
+      "num_tokens": 10716304.0,
+      "step": 1264
+    },
+    {
+      "epoch": 2.902411021814007,
+      "grad_norm": 1.8272403674072573,
+      "learning_rate": 1.0309981271696186e-06,
+      "loss": 0.0877,
+      "mean_token_accuracy": 0.9753663539886475,
+      "num_tokens": 10724864.0,
+      "step": 1265
+    },
+    {
+      "epoch": 2.9047072330654418,
+      "grad_norm": 1.6906213468521887,
+      "learning_rate": 1.029606662357914e-06,
+      "loss": 0.0942,
+      "mean_token_accuracy": 0.9716041684150696,
+      "num_tokens": 10733985.0,
+      "step": 1266
+    },
+    {
+      "epoch": 2.907003444316877,
+      "grad_norm": 1.7382113873648,
+      "learning_rate": 1.0282470462894078e-06,
+      "loss": 0.0927,
+      "mean_token_accuracy": 0.9735080301761627,
+      "num_tokens": 10742248.0,
+      "step": 1267
+    },
+    {
+      "epoch": 2.9092996555683124,
+      "grad_norm": 1.8579693254057839,
+      "learning_rate": 1.026919288650509e-06,
+      "loss": 0.1019,
+      "mean_token_accuracy": 0.9702234864234924,
+      "num_tokens": 10750537.0,
+      "step": 1268
+    },
+    {
+      "epoch": 2.9115958668197477,
+      "grad_norm": 1.7233765043962823,
+      "learning_rate": 1.0256233989006566e-06,
+      "loss": 0.0831,
+      "mean_token_accuracy": 0.977370411157608,
+      "num_tokens": 10759367.0,
+      "step": 1269
+    },
+    {
+      "epoch": 2.9138920780711826,
+      "grad_norm": 1.9071582937378728,
+      "learning_rate": 1.0243593862722503e-06,
+      "loss": 0.0954,
+      "mean_token_accuracy": 0.9740357100963593,
+      "num_tokens": 10768169.0,
+      "step": 1270
+    },
+    {
+      "epoch": 2.9161882893226174,
+      "grad_norm": 1.9493190224475765,
+      "learning_rate": 1.0231272597705868e-06,
+      "loss": 0.1008,
+      "mean_token_accuracy": 0.9729692041873932,
+      "num_tokens": 10776884.0,
+      "step": 1271
+    },
+    {
+      "epoch": 2.9184845005740527,
+      "grad_norm": 1.653638882155546,
+      "learning_rate": 1.0219270281737922e-06,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9771461486816406,
+      "num_tokens": 10785728.0,
+      "step": 1272
+    },
+    {
+      "epoch": 2.920780711825488,
+      "grad_norm": 2.072421381289451,
+      "learning_rate": 1.020758700032763e-06,
+      "loss": 0.0858,
+      "mean_token_accuracy": 0.9745146334171295,
+      "num_tokens": 10794020.0,
+      "step": 1273
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 1.673044153231585,
+      "learning_rate": 1.0196222836711025e-06,
+      "loss": 0.0858,
+      "mean_token_accuracy": 0.97358837723732,
+      "num_tokens": 10802535.0,
+      "step": 1274
+    },
+    {
+      "epoch": 2.925373134328358,
+      "grad_norm": 1.6370042892666163,
+      "learning_rate": 1.0185177871850633e-06,
+      "loss": 0.0911,
+      "mean_token_accuracy": 0.9754725694656372,
+      "num_tokens": 10811120.0,
+      "step": 1275
+    },
+    {
+      "epoch": 2.927669345579793,
+      "grad_norm": 1.7894277459696848,
+      "learning_rate": 1.0174452184434888e-06,
+      "loss": 0.0858,
+      "mean_token_accuracy": 0.9763053357601166,
+      "num_tokens": 10820489.0,
+      "step": 1276
+    },
+    {
+      "epoch": 2.9299655568312284,
+      "grad_norm": 1.9530911184851654,
+      "learning_rate": 1.0164045850877572e-06,
+      "loss": 0.0895,
+      "mean_token_accuracy": 0.9719912111759186,
+      "num_tokens": 10829356.0,
+      "step": 1277
+    },
+    {
+      "epoch": 2.9322617680826637,
+      "grad_norm": 2.074729867899164,
+      "learning_rate": 1.0153958945317272e-06,
+      "loss": 0.0917,
+      "mean_token_accuracy": 0.9750551581382751,
+      "num_tokens": 10837694.0,
+      "step": 1278
+    },
+    {
+      "epoch": 2.934557979334099,
+      "grad_norm": 1.9065414798422475,
+      "learning_rate": 1.0144191539616847e-06,
+      "loss": 0.0898,
+      "mean_token_accuracy": 0.9740335047245026,
+      "num_tokens": 10846467.0,
+      "step": 1279
+    },
+    {
+      "epoch": 2.936854190585534,
+      "grad_norm": 1.9289056714697697,
+      "learning_rate": 1.0134743703362934e-06,
+      "loss": 0.0957,
+      "mean_token_accuracy": 0.9711492657661438,
+      "num_tokens": 10854587.0,
+      "step": 1280
+    },
+    {
+      "epoch": 2.9391504018369687,
+      "grad_norm": 1.7368377967414343,
+      "learning_rate": 1.0125615503865421e-06,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9736557304859161,
+      "num_tokens": 10863138.0,
+      "step": 1281
+    },
+    {
+      "epoch": 2.941446613088404,
+      "grad_norm": 1.7173387961101678,
+      "learning_rate": 1.0116807006156996e-06,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9757134020328522,
+      "num_tokens": 10871528.0,
+      "step": 1282
+    },
+    {
+      "epoch": 2.9437428243398394,
+      "grad_norm": 1.798213674031941,
+      "learning_rate": 1.010831827299267e-06,
+      "loss": 0.0939,
+      "mean_token_accuracy": 0.9726366996765137,
+      "num_tokens": 10880127.0,
+      "step": 1283
+    },
+    {
+      "epoch": 2.9460390355912747,
+      "grad_norm": 1.648699123694954,
+      "learning_rate": 1.010014936484933e-06,
+      "loss": 0.0851,
+      "mean_token_accuracy": 0.9758202135562897,
+      "num_tokens": 10889004.0,
+      "step": 1284
+    },
+    {
+      "epoch": 2.9483352468427095,
+      "grad_norm": 1.8692209607374426,
+      "learning_rate": 1.0092300339925318e-06,
+      "loss": 0.1015,
+      "mean_token_accuracy": 0.9704893529415131,
+      "num_tokens": 10897169.0,
+      "step": 1285
+    },
+    {
+      "epoch": 2.9506314580941444,
+      "grad_norm": 1.8621910025629516,
+      "learning_rate": 1.008477125414e-06,
+      "loss": 0.0897,
+      "mean_token_accuracy": 0.9735351800918579,
+      "num_tokens": 10905307.0,
+      "step": 1286
+    },
+    {
+      "epoch": 2.9529276693455797,
+      "grad_norm": 1.7678578375489495,
+      "learning_rate": 1.0077562161133376e-06,
+      "loss": 0.0962,
+      "mean_token_accuracy": 0.9731249511241913,
+      "num_tokens": 10913487.0,
+      "step": 1287
+    },
+    {
+      "epoch": 2.955223880597015,
+      "grad_norm": 1.6104943544309167,
+      "learning_rate": 1.0070673112265711e-06,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9757913947105408,
+      "num_tokens": 10921704.0,
+      "step": 1288
+    },
+    {
+      "epoch": 2.9575200918484503,
+      "grad_norm": 1.7536147555736281,
+      "learning_rate": 1.0064104156617146e-06,
+      "loss": 0.0925,
+      "mean_token_accuracy": 0.9730434119701385,
+      "num_tokens": 10930516.0,
+      "step": 1289
+    },
+    {
+      "epoch": 2.959816303099885,
+      "grad_norm": 1.6468043601907105,
+      "learning_rate": 1.005785534098737e-06,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9742352962493896,
+      "num_tokens": 10939203.0,
+      "step": 1290
+    },
+    {
+      "epoch": 2.96211251435132,
+      "grad_norm": 1.6438861473924387,
+      "learning_rate": 1.0051926709895247e-06,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9770634174346924,
+      "num_tokens": 10947613.0,
+      "step": 1291
+    },
+    {
+      "epoch": 2.9644087256027554,
+      "grad_norm": 1.778175488188421,
+      "learning_rate": 1.0046318305578567e-06,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.974421501159668,
+      "num_tokens": 10956108.0,
+      "step": 1292
+    },
+    {
+      "epoch": 2.9667049368541907,
+      "grad_norm": 2.029985872014812,
+      "learning_rate": 1.0041030167993678e-06,
+      "loss": 0.0902,
+      "mean_token_accuracy": 0.9753710031509399,
+      "num_tokens": 10964276.0,
+      "step": 1293
+    },
+    {
+      "epoch": 2.969001148105626,
+      "grad_norm": 1.968733547748672,
+      "learning_rate": 1.0036062334815241e-06,
+      "loss": 0.0865,
+      "mean_token_accuracy": 0.975742906332016,
+      "num_tokens": 10972554.0,
+      "step": 1294
+    },
+    {
+      "epoch": 2.971297359357061,
+      "grad_norm": 1.8997499165333385,
+      "learning_rate": 1.0031414841435938e-06,
+      "loss": 0.0862,
+      "mean_token_accuracy": 0.9743078947067261,
+      "num_tokens": 10981148.0,
+      "step": 1295
+    },
+    {
+      "epoch": 2.9735935706084957,
+      "grad_norm": 1.8418114171304376,
+      "learning_rate": 1.002708772096625e-06,
+      "loss": 0.0946,
+      "mean_token_accuracy": 0.9721650183200836,
+      "num_tokens": 10989605.0,
+      "step": 1296
+    },
+    {
+      "epoch": 2.975889781859931,
+      "grad_norm": 1.8991753484871785,
+      "learning_rate": 1.0023081004234183e-06,
+      "loss": 0.0907,
+      "mean_token_accuracy": 0.9747517108917236,
+      "num_tokens": 10997620.0,
+      "step": 1297
+    },
+    {
+      "epoch": 2.9781859931113663,
+      "grad_norm": 1.8565305912915842,
+      "learning_rate": 1.0019394719785073e-06,
+      "loss": 0.0911,
+      "mean_token_accuracy": 0.9744541645050049,
+      "num_tokens": 11006031.0,
+      "step": 1298
+    },
+    {
+      "epoch": 2.9804822043628016,
+      "grad_norm": 1.618751990063323,
+      "learning_rate": 1.0016028893881387e-06,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9759952127933502,
+      "num_tokens": 11015084.0,
+      "step": 1299
+    },
+    {
+      "epoch": 2.9827784156142365,
+      "grad_norm": 1.6930158879050055,
+      "learning_rate": 1.0012983550502523e-06,
+      "loss": 0.0871,
+      "mean_token_accuracy": 0.9745848774909973,
+      "num_tokens": 11024092.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 1.4435819286481701,
+      "learning_rate": 1.0010258711344624e-06,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.978447824716568,
+      "num_tokens": 11033191.0,
+      "step": 1301
+    },
+    {
+      "epoch": 2.9873708381171067,
+      "grad_norm": 1.7887498192953641,
+      "learning_rate": 1.0007854395820472e-06,
+      "loss": 0.1035,
+      "mean_token_accuracy": 0.9726414084434509,
+      "num_tokens": 11041536.0,
+      "step": 1302
+    },
+    {
+      "epoch": 2.989667049368542,
+      "grad_norm": 2.2008612687952005,
+      "learning_rate": 1.0005770621059293e-06,
+      "loss": 0.1088,
+      "mean_token_accuracy": 0.9678494334220886,
+      "num_tokens": 11050032.0,
+      "step": 1303
+    },
+    {
+      "epoch": 2.9919632606199773,
+      "grad_norm": 1.7583102836136177,
+      "learning_rate": 1.000400740190668e-06,
+      "loss": 0.0959,
+      "mean_token_accuracy": 0.9743696749210358,
+      "num_tokens": 11058315.0,
+      "step": 1304
+    },
+    {
+      "epoch": 2.994259471871412,
+      "grad_norm": 1.6762632430975293,
+      "learning_rate": 1.000256475092446e-06,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9762474298477173,
+      "num_tokens": 11066486.0,
+      "step": 1305
+    },
+    {
+      "epoch": 2.996555683122847,
+      "grad_norm": 1.8938521329481015,
+      "learning_rate": 1.0001442678390612e-06,
+      "loss": 0.0991,
+      "mean_token_accuracy": 0.9710158407688141,
+      "num_tokens": 11075397.0,
+      "step": 1306
+    },
+    {
+      "epoch": 2.9988518943742823,
+      "grad_norm": 1.692336927353843,
+      "learning_rate": 1.00006411922992e-06,
+      "loss": 0.0873,
+      "mean_token_accuracy": 0.9752947986125946,
+      "num_tokens": 11083970.0,
+      "step": 1307
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.692336927353843,
+      "learning_rate": 1.0000160298360306e-06,
+      "loss": 0.0541,
+      "mean_token_accuracy": 0.9806763529777527,
+      "num_tokens": 11084593.0,
+      "step": 1308
+    },
+    {
+      "epoch": 3.0,
+      "step": 1308,
+      "total_flos": 358578844041216.0,
+      "train_loss": 0.37431958322313585,
+      "train_runtime": 18562.8676,
+      "train_samples_per_second": 2.25,
+      "train_steps_per_second": 0.07
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1308,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 800,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 358578844041216.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}