diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7044 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.18666666666666668,
+  "eval_steps": 500,
+  "global_step": 7000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.1917937099933624,
+      "epoch": 2.6666666666666667e-05,
+      "grad_norm": 2.756018877029419,
+      "learning_rate": 0.0,
+      "loss": 1.2545,
+      "mean_token_accuracy": 0.696878045797348,
+      "num_tokens": 32768.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.225203620062934,
+      "epoch": 0.0002666666666666667,
+      "grad_norm": 1.0855209827423096,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 1.2994,
+      "mean_token_accuracy": 0.69396587047312,
+      "num_tokens": 327680.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.1899177700281143,
+      "epoch": 0.0005333333333333334,
+      "grad_norm": 0.6739585399627686,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.2378,
+      "mean_token_accuracy": 0.7016434505581856,
+      "num_tokens": 655360.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.1293971434235572,
+      "epoch": 0.0008,
+      "grad_norm": 0.592927098274231,
+      "learning_rate": 5.8e-06,
+      "loss": 1.1511,
+      "mean_token_accuracy": 0.7152706444263458,
+      "num_tokens": 983040.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.1152707174420358,
+      "epoch": 0.0010666666666666667,
+      "grad_norm": 0.5012957453727722,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 1.106,
+      "mean_token_accuracy": 0.7225225985050201,
+      "num_tokens": 1310720.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.0850900515913964,
+      "epoch": 0.0013333333333333333,
+      "grad_norm": 0.4958457052707672,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.0733,
+      "mean_token_accuracy": 0.7267992407083511,
+      "num_tokens": 1638400.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.0684167608618735,
+      "epoch": 0.0016,
+      "grad_norm": 0.46953684091567993,
+      "learning_rate": 1.18e-05,
+      "loss": 1.0543,
+      "mean_token_accuracy": 0.7275537639856339,
+      "num_tokens": 1966080.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.0187185630202293,
+      "epoch": 0.0018666666666666666,
+      "grad_norm": 0.4304210841655731,
+      "learning_rate": 1.38e-05,
+      "loss": 1.0117,
+      "mean_token_accuracy": 0.7367454767227173,
+      "num_tokens": 2293760.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.0244884178042413,
+      "epoch": 0.0021333333333333334,
+      "grad_norm": 0.4684528708457947,
+      "learning_rate": 1.58e-05,
+      "loss": 1.0233,
+      "mean_token_accuracy": 0.7338923513889313,
+      "num_tokens": 2621440.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.027450728416443,
+      "epoch": 0.0024,
+      "grad_norm": 0.473316490650177,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 1.0255,
+      "mean_token_accuracy": 0.7331347793340683,
+      "num_tokens": 2949120.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.033797836303711,
+      "epoch": 0.0026666666666666666,
+      "grad_norm": 0.46264511346817017,
+      "learning_rate": 1.98e-05,
+      "loss": 1.0247,
+      "mean_token_accuracy": 0.7324810594320297,
+      "num_tokens": 3276800.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.9973023891448974,
+      "epoch": 0.0029333333333333334,
+      "grad_norm": 0.4641575217247009,
+      "learning_rate": 1.9995187165775402e-05,
+      "loss": 0.9937,
+      "mean_token_accuracy": 0.7390884727239608,
+      "num_tokens": 3604480.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.9917041152715683,
+      "epoch": 0.0032,
+      "grad_norm": 0.549523115158081,
+      "learning_rate": 1.9989839572192514e-05,
+      "loss": 0.9887,
+      "mean_token_accuracy": 0.7388135403394699,
+      "num_tokens": 3932160.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.9998480260372162,
+      "epoch": 0.0034666666666666665,
+      "grad_norm": 0.4537445306777954,
+      "learning_rate": 1.998449197860963e-05,
+      "loss": 0.995,
+      "mean_token_accuracy": 0.7383186712861061,
+      "num_tokens": 4259840.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.9524103581905365,
+      "epoch": 0.0037333333333333333,
+      "grad_norm": 0.48012760281562805,
+      "learning_rate": 1.9979144385026738e-05,
+      "loss": 0.9467,
+      "mean_token_accuracy": 0.7473148837685585,
+      "num_tokens": 4587520.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.9688920095562935,
+      "epoch": 0.004,
+      "grad_norm": 0.47992178797721863,
+      "learning_rate": 1.9973796791443853e-05,
+      "loss": 0.9631,
+      "mean_token_accuracy": 0.7448161020874977,
+      "num_tokens": 4915200.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.9810811311006546,
+      "epoch": 0.004266666666666667,
+      "grad_norm": 0.4852803945541382,
+      "learning_rate": 1.9968449197860965e-05,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.742143203318119,
+      "num_tokens": 5242880.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.9711030036211014,
+      "epoch": 0.004533333333333334,
+      "grad_norm": 0.5114846229553223,
+      "learning_rate": 1.9963101604278076e-05,
+      "loss": 0.9694,
+      "mean_token_accuracy": 0.7441593334078789,
+      "num_tokens": 5570560.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.9730613678693771,
+      "epoch": 0.0048,
+      "grad_norm": 0.47124916315078735,
+      "learning_rate": 1.9957754010695188e-05,
+      "loss": 0.965,
+      "mean_token_accuracy": 0.743820258975029,
+      "num_tokens": 5898240.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.9529570192098618,
+      "epoch": 0.005066666666666666,
+      "grad_norm": 0.4511706233024597,
+      "learning_rate": 1.9952406417112303e-05,
+      "loss": 0.9462,
+      "mean_token_accuracy": 0.7474676162004471,
+      "num_tokens": 6225920.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.9540489405393601,
+      "epoch": 0.005333333333333333,
+      "grad_norm": 0.5116438865661621,
+      "learning_rate": 1.9947058823529412e-05,
+      "loss": 0.9515,
+      "mean_token_accuracy": 0.745259040594101,
+      "num_tokens": 6553600.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.9719951331615448,
+      "epoch": 0.0056,
+      "grad_norm": 0.4739775061607361,
+      "learning_rate": 1.9941711229946527e-05,
+      "loss": 0.9656,
+      "mean_token_accuracy": 0.7429618790745736,
+      "num_tokens": 6881280.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.9361593425273895,
+      "epoch": 0.005866666666666667,
+      "grad_norm": 0.46898579597473145,
+      "learning_rate": 1.993636363636364e-05,
+      "loss": 0.9309,
+      "mean_token_accuracy": 0.7507209226489067,
+      "num_tokens": 7208960.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.9282811805605888,
+      "epoch": 0.0061333333333333335,
+      "grad_norm": 0.48956814408302307,
+      "learning_rate": 1.993101604278075e-05,
+      "loss": 0.9205,
+      "mean_token_accuracy": 0.7508217230439186,
+      "num_tokens": 7536640.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.9626896977424622,
+      "epoch": 0.0064,
+      "grad_norm": 0.4619200825691223,
+      "learning_rate": 1.9925668449197863e-05,
+      "loss": 0.96,
+      "mean_token_accuracy": 0.7457844629883766,
+      "num_tokens": 7864320.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.9355827301740647,
+      "epoch": 0.006666666666666667,
+      "grad_norm": 0.4730663597583771,
+      "learning_rate": 1.9920320855614974e-05,
+      "loss": 0.9293,
+      "mean_token_accuracy": 0.7511669084429741,
+      "num_tokens": 8192000.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.9282837182283401,
+      "epoch": 0.006933333333333333,
+      "grad_norm": 0.4893971085548401,
+      "learning_rate": 1.9914973262032086e-05,
+      "loss": 0.9249,
+      "mean_token_accuracy": 0.7516342848539352,
+      "num_tokens": 8519680.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9069504201412201,
+      "epoch": 0.0072,
+      "grad_norm": 0.5181482434272766,
+      "learning_rate": 1.9909625668449198e-05,
+      "loss": 0.9026,
+      "mean_token_accuracy": 0.7560208901762963,
+      "num_tokens": 8847360.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9075347542762756,
+      "epoch": 0.007466666666666667,
+      "grad_norm": 0.5042241811752319,
+      "learning_rate": 1.9904278074866313e-05,
+      "loss": 0.904,
+      "mean_token_accuracy": 0.7554679840803147,
+      "num_tokens": 9175040.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9218019858002663,
+      "epoch": 0.007733333333333333,
+      "grad_norm": 0.4679202437400818,
+      "learning_rate": 1.9898930481283425e-05,
+      "loss": 0.9159,
+      "mean_token_accuracy": 0.7540017157793045,
+      "num_tokens": 9502720.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9411949530243874,
+      "epoch": 0.008,
+      "grad_norm": 0.44983187317848206,
+      "learning_rate": 1.9893582887700537e-05,
+      "loss": 0.9368,
+      "mean_token_accuracy": 0.7479838699102401,
+      "num_tokens": 9830400.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9111632362008095,
+      "epoch": 0.008266666666666667,
+      "grad_norm": 0.45098212361335754,
+      "learning_rate": 1.988823529411765e-05,
+      "loss": 0.9048,
+      "mean_token_accuracy": 0.7555535227060318,
+      "num_tokens": 10158080.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9299016907811165,
+      "epoch": 0.008533333333333334,
+      "grad_norm": 0.4565262794494629,
+      "learning_rate": 1.988288770053476e-05,
+      "loss": 0.9273,
+      "mean_token_accuracy": 0.7521291568875312,
+      "num_tokens": 10485760.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9363565370440483,
+      "epoch": 0.0088,
+      "grad_norm": 0.5131130218505859,
+      "learning_rate": 1.9877540106951872e-05,
+      "loss": 0.9336,
+      "mean_token_accuracy": 0.7517320349812507,
+      "num_tokens": 10813440.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9035725295543671,
+      "epoch": 0.009066666666666667,
+      "grad_norm": 0.4885678291320801,
+      "learning_rate": 1.9872192513368987e-05,
+      "loss": 0.8962,
+      "mean_token_accuracy": 0.7565249264240265,
+      "num_tokens": 11141120.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.931587140262127,
+      "epoch": 0.009333333333333334,
+      "grad_norm": 0.463820219039917,
+      "learning_rate": 1.9866844919786096e-05,
+      "loss": 0.9241,
+      "mean_token_accuracy": 0.7517564758658409,
+      "num_tokens": 11468800.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9012247174978256,
+      "epoch": 0.0096,
+      "grad_norm": 0.45758935809135437,
+      "learning_rate": 1.986149732620321e-05,
+      "loss": 0.8993,
+      "mean_token_accuracy": 0.7569189876317978,
+      "num_tokens": 11796480.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.8967527940869331,
+      "epoch": 0.009866666666666666,
+      "grad_norm": 0.4350438714027405,
+      "learning_rate": 1.9856149732620323e-05,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7589503869414329,
+      "num_tokens": 12124160.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.8967338830232621,
+      "epoch": 0.010133333333333333,
+      "grad_norm": 0.45170173048973083,
+      "learning_rate": 1.9850802139037435e-05,
+      "loss": 0.8963,
+      "mean_token_accuracy": 0.7554863154888153,
+      "num_tokens": 12451840.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9230557337403298,
+      "epoch": 0.0104,
+      "grad_norm": 0.4924105405807495,
+      "learning_rate": 1.9845454545454546e-05,
+      "loss": 0.9191,
+      "mean_token_accuracy": 0.754075026512146,
+      "num_tokens": 12779520.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.906305131316185,
+      "epoch": 0.010666666666666666,
+      "grad_norm": 0.45452919602394104,
+      "learning_rate": 1.984010695187166e-05,
+      "loss": 0.8964,
+      "mean_token_accuracy": 0.7582386374473572,
+      "num_tokens": 13107200.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8801364779472352,
+      "epoch": 0.010933333333333333,
+      "grad_norm": 0.45245060324668884,
+      "learning_rate": 1.983475935828877e-05,
+      "loss": 0.8784,
+      "mean_token_accuracy": 0.7602181121706962,
+      "num_tokens": 13434880.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8755475401878356,
+      "epoch": 0.0112,
+      "grad_norm": 0.5070144534111023,
+      "learning_rate": 1.9829411764705885e-05,
+      "loss": 0.8693,
+      "mean_token_accuracy": 0.7621395409107208,
+      "num_tokens": 13762560.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9023118302226066,
+      "epoch": 0.011466666666666667,
+      "grad_norm": 0.4301604628562927,
+      "learning_rate": 1.9824064171122997e-05,
+      "loss": 0.8984,
+      "mean_token_accuracy": 0.7581011667847634,
+      "num_tokens": 14090240.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.9035790309309959,
+      "epoch": 0.011733333333333333,
+      "grad_norm": 0.46501588821411133,
+      "learning_rate": 1.981871657754011e-05,
+      "loss": 0.8977,
+      "mean_token_accuracy": 0.7566990479826927,
+      "num_tokens": 14417920.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.8804532885551453,
+      "epoch": 0.012,
+      "grad_norm": 0.4364991784095764,
+      "learning_rate": 1.981336898395722e-05,
+      "loss": 0.8756,
+      "mean_token_accuracy": 0.7617485299706459,
+      "num_tokens": 14745600.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.9079241648316383,
+      "epoch": 0.012266666666666667,
+      "grad_norm": 0.4812929034233093,
+      "learning_rate": 1.9808021390374332e-05,
+      "loss": 0.906,
+      "mean_token_accuracy": 0.7552236095070839,
+      "num_tokens": 15073280.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.932107862830162,
+      "epoch": 0.012533333333333334,
+      "grad_norm": 0.5057600736618042,
+      "learning_rate": 1.9802673796791444e-05,
+      "loss": 0.9223,
+      "mean_token_accuracy": 0.7512493848800659,
+      "num_tokens": 15400960.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.9215750128030777,
+      "epoch": 0.0128,
+      "grad_norm": 0.5192552804946899,
+      "learning_rate": 1.979732620320856e-05,
+      "loss": 0.9236,
+      "mean_token_accuracy": 0.7527553722262382,
+      "num_tokens": 15728640.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8818250864744186,
+      "epoch": 0.013066666666666667,
+      "grad_norm": 0.4712938070297241,
+      "learning_rate": 1.979197860962567e-05,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7624266907572746,
+      "num_tokens": 16056320.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.855175918340683,
+      "epoch": 0.013333333333333334,
+      "grad_norm": 0.44197145104408264,
+      "learning_rate": 1.9786631016042783e-05,
+      "loss": 0.8508,
+      "mean_token_accuracy": 0.7657746762037277,
+      "num_tokens": 16384000.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8776819244027138,
+      "epoch": 0.0136,
+      "grad_norm": 0.4693980813026428,
+      "learning_rate": 1.9781283422459895e-05,
+      "loss": 0.8721,
+      "mean_token_accuracy": 0.7612261712551117,
+      "num_tokens": 16711680.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8772112146019936,
+      "epoch": 0.013866666666666666,
+      "grad_norm": 0.46298500895500183,
+      "learning_rate": 1.9775935828877007e-05,
+      "loss": 0.8662,
+      "mean_token_accuracy": 0.7619470894336701,
+      "num_tokens": 17039360.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.875455005466938,
+      "epoch": 0.014133333333333333,
+      "grad_norm": 0.4709574580192566,
+      "learning_rate": 1.977058823529412e-05,
+      "loss": 0.8718,
+      "mean_token_accuracy": 0.7629765406250953,
+      "num_tokens": 17367040.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.871045409142971,
+      "epoch": 0.0144,
+      "grad_norm": 0.46276578307151794,
+      "learning_rate": 1.976524064171123e-05,
+      "loss": 0.8644,
+      "mean_token_accuracy": 0.7631628751754761,
+      "num_tokens": 17694720.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8733671963214874,
+      "epoch": 0.014666666666666666,
+      "grad_norm": 0.4854150414466858,
+      "learning_rate": 1.9759893048128345e-05,
+      "loss": 0.8671,
+      "mean_token_accuracy": 0.7625336021184921,
+      "num_tokens": 18022400.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.8307457804679871,
+      "epoch": 0.014933333333333333,
+      "grad_norm": 0.5010337233543396,
+      "learning_rate": 1.9754545454545454e-05,
+      "loss": 0.8276,
+      "mean_token_accuracy": 0.7701735079288483,
+      "num_tokens": 18350080.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8965598121285439,
+      "epoch": 0.0152,
+      "grad_norm": 0.5053139925003052,
+      "learning_rate": 1.974919786096257e-05,
+      "loss": 0.89,
+      "mean_token_accuracy": 0.759066465497017,
+      "num_tokens": 18677760.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8558321312069893,
+      "epoch": 0.015466666666666667,
+      "grad_norm": 0.46031227707862854,
+      "learning_rate": 1.974385026737968e-05,
+      "loss": 0.8496,
+      "mean_token_accuracy": 0.766547529399395,
+      "num_tokens": 19005440.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8634923353791237,
+      "epoch": 0.015733333333333332,
+      "grad_norm": 0.45664703845977783,
+      "learning_rate": 1.9738502673796793e-05,
+      "loss": 0.8593,
+      "mean_token_accuracy": 0.7641953766345978,
+      "num_tokens": 19333120.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8536560088396072,
+      "epoch": 0.016,
+      "grad_norm": 0.5289744734764099,
+      "learning_rate": 1.9733155080213905e-05,
+      "loss": 0.8476,
+      "mean_token_accuracy": 0.7657227471470833,
+      "num_tokens": 19660800.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8970341011881828,
+      "epoch": 0.016266666666666665,
+      "grad_norm": 0.4675828218460083,
+      "learning_rate": 1.972780748663102e-05,
+      "loss": 0.8861,
+      "mean_token_accuracy": 0.7606579944491386,
+      "num_tokens": 19988480.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8475384518504143,
+      "epoch": 0.016533333333333334,
+      "grad_norm": 0.485074907541275,
+      "learning_rate": 1.9722459893048128e-05,
+      "loss": 0.8483,
+      "mean_token_accuracy": 0.7655914008617402,
+      "num_tokens": 20316160.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.859432689845562,
+      "epoch": 0.0168,
+      "grad_norm": 0.47643354535102844,
+      "learning_rate": 1.9717112299465243e-05,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.7655241936445236,
+      "num_tokens": 20643840.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8484608560800553,
+      "epoch": 0.017066666666666667,
+      "grad_norm": 0.5069352984428406,
+      "learning_rate": 1.9711764705882355e-05,
+      "loss": 0.8418,
+      "mean_token_accuracy": 0.7684536874294281,
+      "num_tokens": 20971520.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8371534153819085,
+      "epoch": 0.017333333333333333,
+      "grad_norm": 0.47192221879959106,
+      "learning_rate": 1.9706417112299467e-05,
+      "loss": 0.8288,
+      "mean_token_accuracy": 0.7725592643022537,
+      "num_tokens": 21299200.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.8368221953511238,
+      "epoch": 0.0176,
+      "grad_norm": 0.45516103506088257,
+      "learning_rate": 1.970106951871658e-05,
+      "loss": 0.831,
+      "mean_token_accuracy": 0.7701796174049378,
+      "num_tokens": 21626880.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8400416925549508,
+      "epoch": 0.017866666666666666,
+      "grad_norm": 0.5058351755142212,
+      "learning_rate": 1.969572192513369e-05,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.7710868760943412,
+      "num_tokens": 21954560.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8550396636128426,
+      "epoch": 0.018133333333333335,
+      "grad_norm": 0.48285484313964844,
+      "learning_rate": 1.9690374331550802e-05,
+      "loss": 0.8486,
+      "mean_token_accuracy": 0.7667980194091797,
+      "num_tokens": 22282240.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8442179337143898,
+      "epoch": 0.0184,
+      "grad_norm": 0.5167956948280334,
+      "learning_rate": 1.9685026737967918e-05,
+      "loss": 0.842,
+      "mean_token_accuracy": 0.7679282695055007,
+      "num_tokens": 22609920.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8682486891746521,
+      "epoch": 0.018666666666666668,
+      "grad_norm": 0.47264111042022705,
+      "learning_rate": 1.967967914438503e-05,
+      "loss": 0.8633,
+      "mean_token_accuracy": 0.7637371674180031,
+      "num_tokens": 22937600.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.8567087382078171,
+      "epoch": 0.018933333333333333,
+      "grad_norm": 0.47900059819221497,
+      "learning_rate": 1.967433155080214e-05,
+      "loss": 0.8541,
+      "mean_token_accuracy": 0.766712486743927,
+      "num_tokens": 23265280.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.849834431707859,
+      "epoch": 0.0192,
+      "grad_norm": 0.4607236385345459,
+      "learning_rate": 1.9668983957219253e-05,
+      "loss": 0.8401,
+      "mean_token_accuracy": 0.7682429119944573,
+      "num_tokens": 23592960.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8704495802521706,
+      "epoch": 0.019466666666666667,
+      "grad_norm": 0.48516854643821716,
+      "learning_rate": 1.9663636363636365e-05,
+      "loss": 0.8604,
+      "mean_token_accuracy": 0.7662176206707955,
+      "num_tokens": 23920640.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8398331791162491,
+      "epoch": 0.019733333333333332,
+      "grad_norm": 0.45983725786209106,
+      "learning_rate": 1.9658288770053477e-05,
+      "loss": 0.8316,
+      "mean_token_accuracy": 0.7716214582324028,
+      "num_tokens": 24248320.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.8403495743870735,
+      "epoch": 0.02,
+      "grad_norm": 0.546696662902832,
+      "learning_rate": 1.9652941176470592e-05,
+      "loss": 0.8378,
+      "mean_token_accuracy": 0.7692021027207374,
+      "num_tokens": 24576000.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.846625666320324,
+      "epoch": 0.020266666666666665,
+      "grad_norm": 0.4402243196964264,
+      "learning_rate": 1.9647593582887704e-05,
+      "loss": 0.8394,
+      "mean_token_accuracy": 0.7687958255410194,
+      "num_tokens": 24903680.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8273769333958626,
+      "epoch": 0.020533333333333334,
+      "grad_norm": 0.501061737537384,
+      "learning_rate": 1.9642245989304812e-05,
+      "loss": 0.8254,
+      "mean_token_accuracy": 0.7722537860274314,
+      "num_tokens": 25231360.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8746409237384796,
+      "epoch": 0.0208,
+      "grad_norm": 0.4820423126220703,
+      "learning_rate": 1.9636898395721927e-05,
+      "loss": 0.8702,
+      "mean_token_accuracy": 0.763709670305252,
+      "num_tokens": 25559040.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8554838418960571,
+      "epoch": 0.021066666666666668,
+      "grad_norm": 0.48708412051200867,
+      "learning_rate": 1.963155080213904e-05,
+      "loss": 0.8436,
+      "mean_token_accuracy": 0.7691623851656914,
+      "num_tokens": 25886720.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8408373609185219,
+      "epoch": 0.021333333333333333,
+      "grad_norm": 0.5047973990440369,
+      "learning_rate": 1.962620320855615e-05,
+      "loss": 0.8353,
+      "mean_token_accuracy": 0.7715023159980774,
+      "num_tokens": 26214400.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8513059750199318,
+      "epoch": 0.0216,
+      "grad_norm": 0.5162477493286133,
+      "learning_rate": 1.9620855614973263e-05,
+      "loss": 0.8452,
+      "mean_token_accuracy": 0.7680321365594864,
+      "num_tokens": 26542080.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8257234156131744,
+      "epoch": 0.021866666666666666,
+      "grad_norm": 0.48618701100349426,
+      "learning_rate": 1.9615508021390378e-05,
+      "loss": 0.8209,
+      "mean_token_accuracy": 0.774682305753231,
+      "num_tokens": 26869760.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8454226747155189,
+      "epoch": 0.022133333333333335,
+      "grad_norm": 0.4451475441455841,
+      "learning_rate": 1.9610160427807486e-05,
+      "loss": 0.8364,
+      "mean_token_accuracy": 0.7716367274522782,
+      "num_tokens": 27197440.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8426426142454148,
+      "epoch": 0.0224,
+      "grad_norm": 0.5013114213943481,
+      "learning_rate": 1.96048128342246e-05,
+      "loss": 0.8357,
+      "mean_token_accuracy": 0.770793616771698,
+      "num_tokens": 27525120.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8404464915394783,
+      "epoch": 0.02266666666666667,
+      "grad_norm": 0.45483893156051636,
+      "learning_rate": 1.9599465240641713e-05,
+      "loss": 0.8334,
+      "mean_token_accuracy": 0.7686675176024437,
+      "num_tokens": 27852800.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8530464544892311,
+      "epoch": 0.022933333333333333,
+      "grad_norm": 0.4812896251678467,
+      "learning_rate": 1.9594117647058825e-05,
+      "loss": 0.8464,
+      "mean_token_accuracy": 0.769425094127655,
+      "num_tokens": 28180480.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.7890083432197571,
+      "epoch": 0.0232,
+      "grad_norm": 0.51589035987854,
+      "learning_rate": 1.9588770053475937e-05,
+      "loss": 0.7879,
+      "mean_token_accuracy": 0.7804404914379119,
+      "num_tokens": 28508160.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8365664348006249,
+      "epoch": 0.023466666666666667,
+      "grad_norm": 0.46631357073783875,
+      "learning_rate": 1.958342245989305e-05,
+      "loss": 0.8256,
+      "mean_token_accuracy": 0.7738453060388565,
+      "num_tokens": 28835840.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8198944807052613,
+      "epoch": 0.023733333333333332,
+      "grad_norm": 0.49191394448280334,
+      "learning_rate": 1.957807486631016e-05,
+      "loss": 0.8137,
+      "mean_token_accuracy": 0.7748839199543,
+      "num_tokens": 29163520.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8006534934043884,
+      "epoch": 0.024,
+      "grad_norm": 0.4927683174610138,
+      "learning_rate": 1.9572727272727276e-05,
+      "loss": 0.7934,
+      "mean_token_accuracy": 0.7795576736330986,
+      "num_tokens": 29491200.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.8472590684890747,
+      "epoch": 0.024266666666666666,
+      "grad_norm": 0.4960421323776245,
+      "learning_rate": 1.9567379679144387e-05,
+      "loss": 0.8408,
+      "mean_token_accuracy": 0.7685850456357002,
+      "num_tokens": 29818880.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8337089791893959,
+      "epoch": 0.024533333333333334,
+      "grad_norm": 0.45343852043151855,
+      "learning_rate": 1.95620320855615e-05,
+      "loss": 0.8265,
+      "mean_token_accuracy": 0.7731946513056756,
+      "num_tokens": 30146560.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8185087725520134,
+      "epoch": 0.0248,
+      "grad_norm": 0.527547299861908,
+      "learning_rate": 1.955668449197861e-05,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.7749144658446312,
+      "num_tokens": 30474240.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.826999232172966,
+      "epoch": 0.025066666666666668,
+      "grad_norm": 0.484590083360672,
+      "learning_rate": 1.9551336898395723e-05,
+      "loss": 0.8155,
+      "mean_token_accuracy": 0.7730846762657165,
+      "num_tokens": 30801920.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8298612013459206,
+      "epoch": 0.025333333333333333,
+      "grad_norm": 0.5021140575408936,
+      "learning_rate": 1.9545989304812835e-05,
+      "loss": 0.8271,
+      "mean_token_accuracy": 0.7713801324367523,
+      "num_tokens": 31129600.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8166377231478691,
+      "epoch": 0.0256,
+      "grad_norm": 0.45983028411865234,
+      "learning_rate": 1.954064171122995e-05,
+      "loss": 0.8117,
+      "mean_token_accuracy": 0.7748258739709855,
+      "num_tokens": 31457280.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8549644887447357,
+      "epoch": 0.025866666666666666,
+      "grad_norm": 0.47488975524902344,
+      "learning_rate": 1.9535294117647062e-05,
+      "loss": 0.851,
+      "mean_token_accuracy": 0.767937433719635,
+      "num_tokens": 31784960.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.7983048990368843,
+      "epoch": 0.026133333333333335,
+      "grad_norm": 0.5116825103759766,
+      "learning_rate": 1.952994652406417e-05,
+      "loss": 0.789,
+      "mean_token_accuracy": 0.7807368054986,
+      "num_tokens": 32112640.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8038114890456199,
+      "epoch": 0.0264,
+      "grad_norm": 0.46046605706214905,
+      "learning_rate": 1.9524598930481285e-05,
+      "loss": 0.7984,
+      "mean_token_accuracy": 0.7783907622098922,
+      "num_tokens": 32440320.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8306144312024116,
+      "epoch": 0.02666666666666667,
+      "grad_norm": 0.455465167760849,
+      "learning_rate": 1.9519251336898397e-05,
+      "loss": 0.819,
+      "mean_token_accuracy": 0.773463462293148,
+      "num_tokens": 32768000.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8105433046817779,
+      "epoch": 0.026933333333333333,
+      "grad_norm": 0.514305055141449,
+      "learning_rate": 1.951390374331551e-05,
+      "loss": 0.8075,
+      "mean_token_accuracy": 0.7765059873461724,
+      "num_tokens": 33095680.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.8218068391084671,
+      "epoch": 0.0272,
+      "grad_norm": 0.5273925065994263,
+      "learning_rate": 1.9508556149732624e-05,
+      "loss": 0.8143,
+      "mean_token_accuracy": 0.7753207445144653,
+      "num_tokens": 33423360.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8005619689822197,
+      "epoch": 0.027466666666666667,
+      "grad_norm": 0.49442076683044434,
+      "learning_rate": 1.9503208556149736e-05,
+      "loss": 0.7875,
+      "mean_token_accuracy": 0.7798081621527672,
+      "num_tokens": 33751040.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8247484147548676,
+      "epoch": 0.027733333333333332,
+      "grad_norm": 0.5219594836235046,
+      "learning_rate": 1.9497860962566844e-05,
+      "loss": 0.8213,
+      "mean_token_accuracy": 0.7736986815929413,
+      "num_tokens": 34078720.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.7826527014374733,
+      "epoch": 0.028,
+      "grad_norm": 0.44685232639312744,
+      "learning_rate": 1.949251336898396e-05,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.7804343834519386,
+      "num_tokens": 34406400.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.7927459478378296,
+      "epoch": 0.028266666666666666,
+      "grad_norm": 0.47030535340309143,
+      "learning_rate": 1.948716577540107e-05,
+      "loss": 0.7892,
+      "mean_token_accuracy": 0.7794110476970673,
+      "num_tokens": 34734080.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.7950563684105874,
+      "epoch": 0.028533333333333334,
+      "grad_norm": 0.46424150466918945,
+      "learning_rate": 1.9481818181818183e-05,
+      "loss": 0.7835,
+      "mean_token_accuracy": 0.7810972630977631,
+      "num_tokens": 35061760.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8059652045369148,
+      "epoch": 0.0288,
+      "grad_norm": 0.507229745388031,
+      "learning_rate": 1.94764705882353e-05,
+      "loss": 0.7985,
+      "mean_token_accuracy": 0.7788123145699501,
+      "num_tokens": 35389440.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8047100350260734,
+      "epoch": 0.029066666666666668,
+      "grad_norm": 0.4822532534599304,
+      "learning_rate": 1.9471122994652407e-05,
+      "loss": 0.7978,
+      "mean_token_accuracy": 0.7769672527909279,
+      "num_tokens": 35717120.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.784260918200016,
+      "epoch": 0.029333333333333333,
+      "grad_norm": 0.4959782063961029,
+      "learning_rate": 1.946577540106952e-05,
+      "loss": 0.7803,
+      "mean_token_accuracy": 0.7819984167814255,
+      "num_tokens": 36044800.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8240539342164993,
+      "epoch": 0.0296,
+      "grad_norm": 0.48158520460128784,
+      "learning_rate": 1.9460427807486634e-05,
+      "loss": 0.8129,
+      "mean_token_accuracy": 0.7740316450595855,
+      "num_tokens": 36372480.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.7918573170900345,
+      "epoch": 0.029866666666666666,
+      "grad_norm": 0.5358759760856628,
+      "learning_rate": 1.9455080213903746e-05,
+      "loss": 0.787,
+      "mean_token_accuracy": 0.7804343819618225,
+      "num_tokens": 36700160.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8246589750051498,
+      "epoch": 0.030133333333333335,
+      "grad_norm": 0.4864969253540039,
+      "learning_rate": 1.9449732620320857e-05,
+      "loss": 0.8202,
+      "mean_token_accuracy": 0.7726447984576226,
+      "num_tokens": 37027840.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.7894530639052391,
+      "epoch": 0.0304,
+      "grad_norm": 0.551214337348938,
+      "learning_rate": 1.944438502673797e-05,
+      "loss": 0.7828,
+      "mean_token_accuracy": 0.7802694290876389,
+      "num_tokens": 37355520.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.8148070260882377,
+      "epoch": 0.030666666666666665,
+      "grad_norm": 0.5003300309181213,
+      "learning_rate": 1.943903743315508e-05,
+      "loss": 0.8053,
+      "mean_token_accuracy": 0.7759316965937615,
+      "num_tokens": 37683200.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.7537903472781181,
+      "epoch": 0.030933333333333334,
+      "grad_norm": 0.522283673286438,
+      "learning_rate": 1.9433689839572193e-05,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.7881170570850372,
+      "num_tokens": 38010880.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.801523107290268,
+      "epoch": 0.0312,
+      "grad_norm": 0.5095292329788208,
+      "learning_rate": 1.9428342245989308e-05,
+      "loss": 0.7917,
+      "mean_token_accuracy": 0.7794079929590225,
+      "num_tokens": 38338560.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.7755785793066025,
+      "epoch": 0.031466666666666664,
+      "grad_norm": 0.48669058084487915,
+      "learning_rate": 1.942299465240642e-05,
+      "loss": 0.7686,
+      "mean_token_accuracy": 0.7830736801028252,
+      "num_tokens": 38666240.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7972671002149582,
+      "epoch": 0.031733333333333336,
+      "grad_norm": 0.5146931409835815,
+      "learning_rate": 1.941764705882353e-05,
+      "loss": 0.796,
+      "mean_token_accuracy": 0.780376347899437,
+      "num_tokens": 38993920.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.795520955324173,
+      "epoch": 0.032,
+      "grad_norm": 0.4943196475505829,
+      "learning_rate": 1.9412299465240643e-05,
+      "loss": 0.7778,
+      "mean_token_accuracy": 0.7812805503606797,
+      "num_tokens": 39321600.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7829627484083176,
+      "epoch": 0.032266666666666666,
+      "grad_norm": 0.4896015524864197,
+      "learning_rate": 1.9406951871657755e-05,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.7820869997143746,
+      "num_tokens": 39649280.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.818043765425682,
+      "epoch": 0.03253333333333333,
+      "grad_norm": 0.5356979966163635,
+      "learning_rate": 1.9401604278074867e-05,
+      "loss": 0.8154,
+      "mean_token_accuracy": 0.7758583828806878,
+      "num_tokens": 39976960.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7951482772827149,
+      "epoch": 0.0328,
+      "grad_norm": 0.4821169376373291,
+      "learning_rate": 1.9396256684491982e-05,
+      "loss": 0.7813,
+      "mean_token_accuracy": 0.7820075735449791,
+      "num_tokens": 40304640.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7990813001990318,
+      "epoch": 0.03306666666666667,
+      "grad_norm": 0.5052225589752197,
+      "learning_rate": 1.9390909090909094e-05,
+      "loss": 0.7988,
+      "mean_token_accuracy": 0.7780700147151947,
+      "num_tokens": 40632320.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7410405725240707,
+      "epoch": 0.03333333333333333,
+      "grad_norm": 0.49692273139953613,
+      "learning_rate": 1.9385561497326202e-05,
+      "loss": 0.732,
+      "mean_token_accuracy": 0.7925158813595772,
+      "num_tokens": 40960000.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7968778610229492,
+      "epoch": 0.0336,
+      "grad_norm": 0.4673176407814026,
+      "learning_rate": 1.9380213903743318e-05,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7777064964175224,
+      "num_tokens": 41287680.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.764365927875042,
+      "epoch": 0.03386666666666667,
+      "grad_norm": 0.5029505491256714,
+      "learning_rate": 1.937486631016043e-05,
+      "loss": 0.7572,
+      "mean_token_accuracy": 0.7866538390517235,
+      "num_tokens": 41615360.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7907108590006828,
+      "epoch": 0.034133333333333335,
+      "grad_norm": 0.5248188972473145,
+      "learning_rate": 1.936951871657754e-05,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.780819283425808,
+      "num_tokens": 41943040.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.788802333176136,
+      "epoch": 0.0344,
+      "grad_norm": 0.46518850326538086,
+      "learning_rate": 1.9364171122994656e-05,
+      "loss": 0.7827,
+      "mean_token_accuracy": 0.7826368525624275,
+      "num_tokens": 42270720.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7725998848676682,
+      "epoch": 0.034666666666666665,
+      "grad_norm": 0.5415055751800537,
+      "learning_rate": 1.9358823529411765e-05,
+      "loss": 0.7667,
+      "mean_token_accuracy": 0.7866630017757416,
+      "num_tokens": 42598400.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.8065502732992172,
+      "epoch": 0.03493333333333333,
+      "grad_norm": 0.46911802887916565,
+      "learning_rate": 1.9353475935828877e-05,
+      "loss": 0.7911,
+      "mean_token_accuracy": 0.7814271822571754,
+      "num_tokens": 42926080.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.7598569974303245,
+      "epoch": 0.0352,
+      "grad_norm": 0.5336980223655701,
+      "learning_rate": 1.9348128342245992e-05,
+      "loss": 0.7584,
+      "mean_token_accuracy": 0.7860154002904892,
+      "num_tokens": 43253760.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7780168011784554,
+      "epoch": 0.03546666666666667,
+      "grad_norm": 0.5248178243637085,
+      "learning_rate": 1.9342780748663104e-05,
+      "loss": 0.7716,
+      "mean_token_accuracy": 0.7840634182095527,
+      "num_tokens": 43581440.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.7802111297845841,
+      "epoch": 0.03573333333333333,
+      "grad_norm": 0.46950259804725647,
+      "learning_rate": 1.9337433155080216e-05,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.7839717730879784,
+      "num_tokens": 43909120.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7892262697219848,
+      "epoch": 0.036,
+      "grad_norm": 0.4483737051486969,
+      "learning_rate": 1.9332085561497327e-05,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.7805535197257996,
+      "num_tokens": 44236800.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7858917117118835,
+      "epoch": 0.03626666666666667,
+      "grad_norm": 0.43271368741989136,
+      "learning_rate": 1.932673796791444e-05,
+      "loss": 0.778,
+      "mean_token_accuracy": 0.7815004870295524,
+      "num_tokens": 44564480.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7805564686655998,
+      "epoch": 0.036533333333333334,
+      "grad_norm": 0.48232999444007874,
+      "learning_rate": 1.932139037433155e-05,
+      "loss": 0.7718,
+      "mean_token_accuracy": 0.7831622704863548,
+      "num_tokens": 44892160.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7665321186184884,
+      "epoch": 0.0368,
+      "grad_norm": 0.4853302836418152,
+      "learning_rate": 1.9316042780748666e-05,
+      "loss": 0.7661,
+      "mean_token_accuracy": 0.7845888301730156,
+      "num_tokens": 45219840.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7815655440092086,
+      "epoch": 0.037066666666666664,
+      "grad_norm": 0.5040138363838196,
+      "learning_rate": 1.9310695187165778e-05,
+      "loss": 0.7663,
+      "mean_token_accuracy": 0.7851325765252113,
+      "num_tokens": 45547520.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.7671736180782318,
+      "epoch": 0.037333333333333336,
+      "grad_norm": 0.46688610315322876,
+      "learning_rate": 1.930534759358289e-05,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.7883950427174569,
+      "num_tokens": 45875200.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7612995445728302,
+      "epoch": 0.0376,
+      "grad_norm": 0.4817868769168854,
+      "learning_rate": 1.93e-05,
+      "loss": 0.7611,
+      "mean_token_accuracy": 0.7856518849730492,
+      "num_tokens": 46202880.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8082423210144043,
+      "epoch": 0.037866666666666667,
+      "grad_norm": 0.5080907344818115,
+      "learning_rate": 1.9294652406417113e-05,
+      "loss": 0.7958,
+      "mean_token_accuracy": 0.7777889743447304,
+      "num_tokens": 46530560.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7901916623115539,
+      "epoch": 0.03813333333333333,
+      "grad_norm": 0.4505074918270111,
+      "learning_rate": 1.9289304812834225e-05,
+      "loss": 0.783,
+      "mean_token_accuracy": 0.7805962800979614,
+      "num_tokens": 46858240.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7650920733809471,
+      "epoch": 0.0384,
+      "grad_norm": 0.520160436630249,
+      "learning_rate": 1.928395721925134e-05,
+      "loss": 0.7554,
+      "mean_token_accuracy": 0.7868982166051864,
+      "num_tokens": 47185920.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7834387555718422,
+      "epoch": 0.03866666666666667,
+      "grad_norm": 0.4799220561981201,
+      "learning_rate": 1.9278609625668452e-05,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.78304313570261,
+      "num_tokens": 47513600.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.7920588865876198,
+      "epoch": 0.038933333333333334,
+      "grad_norm": 0.5058838725090027,
+      "learning_rate": 1.9273262032085564e-05,
+      "loss": 0.7891,
+      "mean_token_accuracy": 0.7809475794434547,
+      "num_tokens": 47841280.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7809412464499473,
+      "epoch": 0.0392,
+      "grad_norm": 0.5144160985946655,
+      "learning_rate": 1.9267914438502676e-05,
+      "loss": 0.773,
+      "mean_token_accuracy": 0.7836693584918976,
+      "num_tokens": 48168960.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.7659068420529366,
+      "epoch": 0.039466666666666664,
+      "grad_norm": 0.4488181471824646,
+      "learning_rate": 1.9262566844919788e-05,
+      "loss": 0.7524,
+      "mean_token_accuracy": 0.7890365302562714,
+      "num_tokens": 48496640.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.744123874604702,
+      "epoch": 0.039733333333333336,
+      "grad_norm": 0.48872411251068115,
+      "learning_rate": 1.92572192513369e-05,
+      "loss": 0.7411,
+      "mean_token_accuracy": 0.7905058667063714,
+      "num_tokens": 48824320.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7659653306007386,
+      "epoch": 0.04,
+      "grad_norm": 0.48394057154655457,
+      "learning_rate": 1.9251871657754015e-05,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7879429414868355,
+      "num_tokens": 49152000.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7519314661622047,
+      "epoch": 0.040266666666666666,
+      "grad_norm": 0.5193694829940796,
+      "learning_rate": 1.9246524064171123e-05,
+      "loss": 0.7455,
+      "mean_token_accuracy": 0.7914222911000252,
+      "num_tokens": 49479680.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7523004367947579,
+      "epoch": 0.04053333333333333,
+      "grad_norm": 0.4702441990375519,
+      "learning_rate": 1.9241176470588235e-05,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.787933774292469,
+      "num_tokens": 49807360.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7782173708081246,
+      "epoch": 0.0408,
+      "grad_norm": 0.560482382774353,
+      "learning_rate": 1.923582887700535e-05,
+      "loss": 0.7664,
+      "mean_token_accuracy": 0.7841245189309121,
+      "num_tokens": 50135040.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7670839697122573,
+      "epoch": 0.04106666666666667,
+      "grad_norm": 0.46597108244895935,
+      "learning_rate": 1.9230481283422462e-05,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.788737167418003,
+      "num_tokens": 50462720.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7554810360074043,
+      "epoch": 0.04133333333333333,
+      "grad_norm": 0.48917797207832336,
+      "learning_rate": 1.9225133689839574e-05,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7885966524481773,
+      "num_tokens": 50790400.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7538948372006417,
+      "epoch": 0.0416,
+      "grad_norm": 0.48391637206077576,
+      "learning_rate": 1.9219786096256685e-05,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7887738287448883,
+      "num_tokens": 51118080.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7791929230093956,
+      "epoch": 0.04186666666666667,
+      "grad_norm": 0.4691905379295349,
+      "learning_rate": 1.9214438502673797e-05,
+      "loss": 0.7722,
+      "mean_token_accuracy": 0.7833486109972,
+      "num_tokens": 51445760.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7446255341172219,
+      "epoch": 0.042133333333333335,
+      "grad_norm": 0.5347681045532227,
+      "learning_rate": 1.920909090909091e-05,
+      "loss": 0.7335,
+      "mean_token_accuracy": 0.7925830855965614,
+      "num_tokens": 51773440.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7673803761601448,
+      "epoch": 0.0424,
+      "grad_norm": 0.5345098376274109,
+      "learning_rate": 1.9203743315508024e-05,
+      "loss": 0.7613,
+      "mean_token_accuracy": 0.7856946498155594,
+      "num_tokens": 52101120.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7777209252119064,
+      "epoch": 0.042666666666666665,
+      "grad_norm": 0.4827149212360382,
+      "learning_rate": 1.9198395721925136e-05,
+      "loss": 0.7713,
+      "mean_token_accuracy": 0.7836663022637367,
+      "num_tokens": 52428800.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.7633250005543232,
+      "epoch": 0.04293333333333333,
+      "grad_norm": 0.5306099057197571,
+      "learning_rate": 1.9193048128342248e-05,
+      "loss": 0.7559,
+      "mean_token_accuracy": 0.7877260506153106,
+      "num_tokens": 52756480.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7480434983968735,
+      "epoch": 0.0432,
+      "grad_norm": 0.5288537740707397,
+      "learning_rate": 1.918770053475936e-05,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.7926105782389641,
+      "num_tokens": 53084160.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7811682939529419,
+      "epoch": 0.04346666666666667,
+      "grad_norm": 0.5437115430831909,
+      "learning_rate": 1.918235294117647e-05,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7843719482421875,
+      "num_tokens": 53411840.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.79571573138237,
+      "epoch": 0.04373333333333333,
+      "grad_norm": 0.47293466329574585,
+      "learning_rate": 1.9177005347593583e-05,
+      "loss": 0.7843,
+      "mean_token_accuracy": 0.7802144423127174,
+      "num_tokens": 53739520.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7749171122908592,
+      "epoch": 0.044,
+      "grad_norm": 0.5731510519981384,
+      "learning_rate": 1.91716577540107e-05,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.7833577677607536,
+      "num_tokens": 54067200.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7756189912557602,
+      "epoch": 0.04426666666666667,
+      "grad_norm": 0.5350993871688843,
+      "learning_rate": 1.916631016042781e-05,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.7844880238175392,
+      "num_tokens": 54394880.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.7374959006905556,
+      "epoch": 0.044533333333333334,
+      "grad_norm": 0.5313361883163452,
+      "learning_rate": 1.9160962566844922e-05,
+      "loss": 0.7267,
+      "mean_token_accuracy": 0.7932917892932891,
+      "num_tokens": 54722560.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.749508784711361,
+      "epoch": 0.0448,
+      "grad_norm": 0.4558972716331482,
+      "learning_rate": 1.9155614973262034e-05,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.7892137080430984,
+      "num_tokens": 55050240.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7335466608405113,
+      "epoch": 0.045066666666666665,
+      "grad_norm": 0.5259963870048523,
+      "learning_rate": 1.9150267379679146e-05,
+      "loss": 0.7288,
+      "mean_token_accuracy": 0.7918316274881363,
+      "num_tokens": 55377920.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7740557298064232,
+      "epoch": 0.04533333333333334,
+      "grad_norm": 0.48570960760116577,
+      "learning_rate": 1.9144919786096258e-05,
+      "loss": 0.7634,
+      "mean_token_accuracy": 0.7872067436575889,
+      "num_tokens": 55705600.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7654780521988869,
+      "epoch": 0.0456,
+      "grad_norm": 0.4950166642665863,
+      "learning_rate": 1.9139572192513373e-05,
+      "loss": 0.7701,
+      "mean_token_accuracy": 0.7850623235106469,
+      "num_tokens": 56033280.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.77603120803833,
+      "epoch": 0.04586666666666667,
+      "grad_norm": 0.4830263555049896,
+      "learning_rate": 1.913422459893048e-05,
+      "loss": 0.7609,
+      "mean_token_accuracy": 0.784258921444416,
+      "num_tokens": 56360960.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7609418869018555,
+      "epoch": 0.04613333333333333,
+      "grad_norm": 0.5481200814247131,
+      "learning_rate": 1.9128877005347596e-05,
+      "loss": 0.7484,
+      "mean_token_accuracy": 0.7893572762608528,
+      "num_tokens": 56688640.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7777589857578278,
+      "epoch": 0.0464,
+      "grad_norm": 0.5048011541366577,
+      "learning_rate": 1.9123529411764708e-05,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.7853005871176719,
+      "num_tokens": 57016320.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7774819910526276,
+      "epoch": 0.04666666666666667,
+      "grad_norm": 0.45590740442276,
+      "learning_rate": 1.911818181818182e-05,
+      "loss": 0.7654,
+      "mean_token_accuracy": 0.7853494629263877,
+      "num_tokens": 57344000.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7336423471570015,
+      "epoch": 0.046933333333333334,
+      "grad_norm": 0.5403317809104919,
+      "learning_rate": 1.9112834224598932e-05,
+      "loss": 0.7266,
+      "mean_token_accuracy": 0.7936461389064788,
+      "num_tokens": 57671680.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7187768623232842,
+      "epoch": 0.0472,
+      "grad_norm": 0.5129840970039368,
+      "learning_rate": 1.9107486631016044e-05,
+      "loss": 0.7083,
+      "mean_token_accuracy": 0.7989980459213257,
+      "num_tokens": 57999360.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7413559228181839,
+      "epoch": 0.047466666666666664,
+      "grad_norm": 0.49861571192741394,
+      "learning_rate": 1.9102139037433155e-05,
+      "loss": 0.7359,
+      "mean_token_accuracy": 0.7919690847396851,
+      "num_tokens": 58327040.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7554563418030739,
+      "epoch": 0.047733333333333336,
+      "grad_norm": 0.5321679711341858,
+      "learning_rate": 1.9096791443850267e-05,
+      "loss": 0.7459,
+      "mean_token_accuracy": 0.7921706959605217,
+      "num_tokens": 58654720.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.74212187230587,
+      "epoch": 0.048,
+      "grad_norm": 0.45808517932891846,
+      "learning_rate": 1.9091443850267382e-05,
+      "loss": 0.738,
+      "mean_token_accuracy": 0.7917491480708122,
+      "num_tokens": 58982400.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7440304681658745,
+      "epoch": 0.048266666666666666,
+      "grad_norm": 0.5892611742019653,
+      "learning_rate": 1.9086096256684494e-05,
+      "loss": 0.7382,
+      "mean_token_accuracy": 0.7911229282617569,
+      "num_tokens": 59310080.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7768477439880371,
+      "epoch": 0.04853333333333333,
+      "grad_norm": 0.4835836589336395,
+      "learning_rate": 1.9080748663101606e-05,
+      "loss": 0.7631,
+      "mean_token_accuracy": 0.7863789096474647,
+      "num_tokens": 59637760.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7587182119488716,
+      "epoch": 0.0488,
+      "grad_norm": 0.4698182940483093,
+      "learning_rate": 1.9075401069518718e-05,
+      "loss": 0.7528,
+      "mean_token_accuracy": 0.7875672027468681,
+      "num_tokens": 59965440.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7687411174178124,
+      "epoch": 0.04906666666666667,
+      "grad_norm": 0.5421642065048218,
+      "learning_rate": 1.907005347593583e-05,
+      "loss": 0.7669,
+      "mean_token_accuracy": 0.7865133240818978,
+      "num_tokens": 60293120.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.7701867774128914,
+      "epoch": 0.04933333333333333,
+      "grad_norm": 0.5125485062599182,
+      "learning_rate": 1.906470588235294e-05,
+      "loss": 0.7596,
+      "mean_token_accuracy": 0.7859481945633888,
+      "num_tokens": 60620800.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7358735978603363,
+      "epoch": 0.0496,
+      "grad_norm": 0.5143376588821411,
+      "learning_rate": 1.9059358288770057e-05,
+      "loss": 0.7291,
+      "mean_token_accuracy": 0.7917277619242669,
+      "num_tokens": 60948480.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7245923519134522,
+      "epoch": 0.04986666666666666,
+      "grad_norm": 0.48511043190956116,
+      "learning_rate": 1.905401069518717e-05,
+      "loss": 0.7103,
+      "mean_token_accuracy": 0.7968475133180618,
+      "num_tokens": 61276160.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7460856273770332,
+      "epoch": 0.050133333333333335,
+      "grad_norm": 0.4994851052761078,
+      "learning_rate": 1.904866310160428e-05,
+      "loss": 0.7382,
+      "mean_token_accuracy": 0.7932642981410026,
+      "num_tokens": 61603840.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.7631218507885933,
+      "epoch": 0.0504,
+      "grad_norm": 0.4711359739303589,
+      "learning_rate": 1.9043315508021392e-05,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.7879765421152115,
+      "num_tokens": 61931520.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7486452057957649,
+      "epoch": 0.050666666666666665,
+      "grad_norm": 0.5361327528953552,
+      "learning_rate": 1.9037967914438504e-05,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7896108284592629,
+      "num_tokens": 62259200.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7843396544456482,
+      "epoch": 0.05093333333333333,
+      "grad_norm": 0.5279068350791931,
+      "learning_rate": 1.9032620320855616e-05,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.784051202237606,
+      "num_tokens": 62586880.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7031185373663902,
+      "epoch": 0.0512,
+      "grad_norm": 0.47015589475631714,
+      "learning_rate": 1.902727272727273e-05,
+      "loss": 0.708,
+      "mean_token_accuracy": 0.7975348263978959,
+      "num_tokens": 62914560.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7438496083021164,
+      "epoch": 0.05146666666666667,
+      "grad_norm": 0.5139464139938354,
+      "learning_rate": 1.902192513368984e-05,
+      "loss": 0.7305,
+      "mean_token_accuracy": 0.7940493702888489,
+      "num_tokens": 63242240.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7328460544347764,
+      "epoch": 0.05173333333333333,
+      "grad_norm": 0.5090652108192444,
+      "learning_rate": 1.9016577540106954e-05,
+      "loss": 0.7189,
+      "mean_token_accuracy": 0.7942265421152115,
+      "num_tokens": 63569920.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7520336791872978,
+      "epoch": 0.052,
+      "grad_norm": 0.4853484332561493,
+      "learning_rate": 1.9011229946524066e-05,
+      "loss": 0.7428,
+      "mean_token_accuracy": 0.7905761316418648,
+      "num_tokens": 63897600.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7502805277705192,
+      "epoch": 0.05226666666666667,
+      "grad_norm": 0.5005112886428833,
+      "learning_rate": 1.9005882352941178e-05,
+      "loss": 0.7448,
+      "mean_token_accuracy": 0.7892809137701988,
+      "num_tokens": 64225280.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7123803675174714,
+      "epoch": 0.052533333333333335,
+      "grad_norm": 0.4861668348312378,
+      "learning_rate": 1.900053475935829e-05,
+      "loss": 0.7081,
+      "mean_token_accuracy": 0.7956347778439522,
+      "num_tokens": 64552960.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7665349021553993,
+      "epoch": 0.0528,
+      "grad_norm": 0.5454739928245544,
+      "learning_rate": 1.89951871657754e-05,
+      "loss": 0.7588,
+      "mean_token_accuracy": 0.7865621954202652,
+      "num_tokens": 64880640.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7607908979058265,
+      "epoch": 0.053066666666666665,
+      "grad_norm": 0.5345908403396606,
+      "learning_rate": 1.8989839572192513e-05,
+      "loss": 0.7501,
+      "mean_token_accuracy": 0.7871731415390968,
+      "num_tokens": 65208320.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7199565961956977,
+      "epoch": 0.05333333333333334,
+      "grad_norm": 0.5356908440589905,
+      "learning_rate": 1.898449197860963e-05,
+      "loss": 0.7118,
+      "mean_token_accuracy": 0.798258800804615,
+      "num_tokens": 65536000.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7386236518621445,
+      "epoch": 0.0536,
+      "grad_norm": 0.47253668308258057,
+      "learning_rate": 1.897914438502674e-05,
+      "loss": 0.7284,
+      "mean_token_accuracy": 0.7946816995739937,
+      "num_tokens": 65863680.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7390207782387733,
+      "epoch": 0.05386666666666667,
+      "grad_norm": 0.5328691005706787,
+      "learning_rate": 1.8973796791443852e-05,
+      "loss": 0.7292,
+      "mean_token_accuracy": 0.792311216890812,
+      "num_tokens": 66191360.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7622766062617302,
+      "epoch": 0.05413333333333333,
+      "grad_norm": 0.4725175201892853,
+      "learning_rate": 1.8968449197860964e-05,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.7881476074457169,
+      "num_tokens": 66519040.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7201081782579422,
+      "epoch": 0.0544,
+      "grad_norm": 0.4898991882801056,
+      "learning_rate": 1.8963101604278076e-05,
+      "loss": 0.7115,
+      "mean_token_accuracy": 0.7967864155769349,
+      "num_tokens": 66846720.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7476230427622795,
+      "epoch": 0.05466666666666667,
+      "grad_norm": 0.49194473028182983,
+      "learning_rate": 1.8957754010695188e-05,
+      "loss": 0.7307,
+      "mean_token_accuracy": 0.7937072336673736,
+      "num_tokens": 67174400.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.730175219476223,
+      "epoch": 0.054933333333333334,
+      "grad_norm": 0.6022394895553589,
+      "learning_rate": 1.8952406417112303e-05,
+      "loss": 0.7273,
+      "mean_token_accuracy": 0.7936766847968102,
+      "num_tokens": 67502080.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.739497372508049,
+      "epoch": 0.0552,
+      "grad_norm": 0.5170940160751343,
+      "learning_rate": 1.8947058823529415e-05,
+      "loss": 0.7279,
+      "mean_token_accuracy": 0.7933528870344162,
+      "num_tokens": 67829760.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.706153379380703,
+      "epoch": 0.055466666666666664,
+      "grad_norm": 0.46231991052627563,
+      "learning_rate": 1.8941711229946527e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.8004734858870506,
+      "num_tokens": 68157440.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7078771367669106,
+      "epoch": 0.055733333333333336,
+      "grad_norm": 0.5030118823051453,
+      "learning_rate": 1.893636363636364e-05,
+      "loss": 0.6942,
+      "mean_token_accuracy": 0.801701495051384,
+      "num_tokens": 68485120.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7353975400328636,
+      "epoch": 0.056,
+      "grad_norm": 0.5152148604393005,
+      "learning_rate": 1.893101604278075e-05,
+      "loss": 0.728,
+      "mean_token_accuracy": 0.7925922527909279,
+      "num_tokens": 68812800.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7255185887217521,
+      "epoch": 0.056266666666666666,
+      "grad_norm": 0.527554988861084,
+      "learning_rate": 1.8925668449197862e-05,
+      "loss": 0.7227,
+      "mean_token_accuracy": 0.7935086742043496,
+      "num_tokens": 69140480.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7665341794490814,
+      "epoch": 0.05653333333333333,
+      "grad_norm": 0.5211319327354431,
+      "learning_rate": 1.8920320855614974e-05,
+      "loss": 0.7547,
+      "mean_token_accuracy": 0.7891953825950623,
+      "num_tokens": 69468160.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.695734278857708,
+      "epoch": 0.0568,
+      "grad_norm": 0.49992096424102783,
+      "learning_rate": 1.891497326203209e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.802168869972229,
+      "num_tokens": 69795840.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7313119895756245,
+      "epoch": 0.05706666666666667,
+      "grad_norm": 0.4708555340766907,
+      "learning_rate": 1.8909625668449197e-05,
+      "loss": 0.7146,
+      "mean_token_accuracy": 0.7958547204732895,
+      "num_tokens": 70123520.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.7185655139386654,
+      "epoch": 0.05733333333333333,
+      "grad_norm": 0.4482806921005249,
+      "learning_rate": 1.8904278074866313e-05,
+      "loss": 0.7129,
+      "mean_token_accuracy": 0.7958913758397103,
+      "num_tokens": 70451200.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.776476477086544,
+      "epoch": 0.0576,
+      "grad_norm": 0.518281102180481,
+      "learning_rate": 1.8898930481283424e-05,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7842864140868187,
+      "num_tokens": 70778880.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.7576399743556976,
+      "epoch": 0.057866666666666663,
+      "grad_norm": 0.49978503584861755,
+      "learning_rate": 1.8893582887700536e-05,
+      "loss": 0.7513,
+      "mean_token_accuracy": 0.7877107754349708,
+      "num_tokens": 71106560.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.6848648637533188,
+      "epoch": 0.058133333333333335,
+      "grad_norm": 0.4555778205394745,
+      "learning_rate": 1.8888235294117648e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.8048173308372497,
+      "num_tokens": 71434240.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7415667727589608,
+      "epoch": 0.0584,
+      "grad_norm": 0.46825316548347473,
+      "learning_rate": 1.888288770053476e-05,
+      "loss": 0.7323,
+      "mean_token_accuracy": 0.79247617572546,
+      "num_tokens": 71761920.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7620745912194252,
+      "epoch": 0.058666666666666666,
+      "grad_norm": 0.6024397611618042,
+      "learning_rate": 1.887754010695187e-05,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.7888288080692292,
+      "num_tokens": 72089600.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7299460664391517,
+      "epoch": 0.05893333333333333,
+      "grad_norm": 0.5377632975578308,
+      "learning_rate": 1.8872192513368987e-05,
+      "loss": 0.718,
+      "mean_token_accuracy": 0.7958425000309944,
+      "num_tokens": 72417280.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7022075489163399,
+      "epoch": 0.0592,
+      "grad_norm": 0.5637347102165222,
+      "learning_rate": 1.88668449197861e-05,
+      "loss": 0.6975,
+      "mean_token_accuracy": 0.8002566024661064,
+      "num_tokens": 72744960.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.7338330700993538,
+      "epoch": 0.05946666666666667,
+      "grad_norm": 0.5497167706489563,
+      "learning_rate": 1.886149732620321e-05,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.7961021512746811,
+      "num_tokens": 73072640.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7120828270912171,
+      "epoch": 0.05973333333333333,
+      "grad_norm": 0.49455058574676514,
+      "learning_rate": 1.8856149732620322e-05,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.7990713596343995,
+      "num_tokens": 73400320.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.6862977862358093,
+      "epoch": 0.06,
+      "grad_norm": 0.5340035557746887,
+      "learning_rate": 1.8850802139037434e-05,
+      "loss": 0.6758,
+      "mean_token_accuracy": 0.8064821600914002,
+      "num_tokens": 73728000.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7129529133439064,
+      "epoch": 0.06026666666666667,
+      "grad_norm": 0.5503765344619751,
+      "learning_rate": 1.8845454545454546e-05,
+      "loss": 0.7071,
+      "mean_token_accuracy": 0.8007697924971581,
+      "num_tokens": 74055680.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.7133767187595368,
+      "epoch": 0.060533333333333335,
+      "grad_norm": 0.5122734904289246,
+      "learning_rate": 1.884010695187166e-05,
+      "loss": 0.7041,
+      "mean_token_accuracy": 0.8009622424840928,
+      "num_tokens": 74383360.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7129033073782921,
+      "epoch": 0.0608,
+      "grad_norm": 0.49274197220802307,
+      "learning_rate": 1.8834759358288773e-05,
+      "loss": 0.6994,
+      "mean_token_accuracy": 0.7994990259408951,
+      "num_tokens": 74711040.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7428222849965096,
+      "epoch": 0.061066666666666665,
+      "grad_norm": 0.5870295166969299,
+      "learning_rate": 1.8829411764705885e-05,
+      "loss": 0.7365,
+      "mean_token_accuracy": 0.7908082813024521,
+      "num_tokens": 75038720.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7150219172239304,
+      "epoch": 0.06133333333333333,
+      "grad_norm": 0.5203629732131958,
+      "learning_rate": 1.8824064171122996e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.7989797174930573,
+      "num_tokens": 75366400.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.715831783413887,
+      "epoch": 0.0616,
+      "grad_norm": 0.5192694067955017,
+      "learning_rate": 1.8818716577540108e-05,
+      "loss": 0.7133,
+      "mean_token_accuracy": 0.7941257387399674,
+      "num_tokens": 75694080.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.6949592083692551,
+      "epoch": 0.06186666666666667,
+      "grad_norm": 0.4811273217201233,
+      "learning_rate": 1.881336898395722e-05,
+      "loss": 0.681,
+      "mean_token_accuracy": 0.8047592893242836,
+      "num_tokens": 76021760.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.730679227411747,
+      "epoch": 0.06213333333333333,
+      "grad_norm": 0.5096244215965271,
+      "learning_rate": 1.8808021390374335e-05,
+      "loss": 0.7253,
+      "mean_token_accuracy": 0.7963679134845734,
+      "num_tokens": 76349440.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.733339001238346,
+      "epoch": 0.0624,
+      "grad_norm": 0.6114660501480103,
+      "learning_rate": 1.8802673796791447e-05,
+      "loss": 0.7234,
+      "mean_token_accuracy": 0.7940829679369926,
+      "num_tokens": 76677120.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7035711303353309,
+      "epoch": 0.06266666666666666,
+      "grad_norm": 0.4455678164958954,
+      "learning_rate": 1.8797326203208555e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.802947823703289,
+      "num_tokens": 77004800.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.6731013461947442,
+      "epoch": 0.06293333333333333,
+      "grad_norm": 0.4920752942562103,
+      "learning_rate": 1.879197860962567e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8078354135155678,
+      "num_tokens": 77332480.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7298787772655487,
+      "epoch": 0.0632,
+      "grad_norm": 0.49286967515945435,
+      "learning_rate": 1.8786631016042782e-05,
+      "loss": 0.721,
+      "mean_token_accuracy": 0.7947244614362716,
+      "num_tokens": 77660160.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7234544113278389,
+      "epoch": 0.06346666666666667,
+      "grad_norm": 0.5942859649658203,
+      "learning_rate": 1.8781283422459894e-05,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.8001832857728004,
+      "num_tokens": 77987840.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7012406192719937,
+      "epoch": 0.06373333333333334,
+      "grad_norm": 0.686430811882019,
+      "learning_rate": 1.8775935828877006e-05,
+      "loss": 0.6978,
+      "mean_token_accuracy": 0.7987078458070755,
+      "num_tokens": 78315520.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7058230191469193,
+      "epoch": 0.064,
+      "grad_norm": 0.5323297381401062,
+      "learning_rate": 1.8770588235294118e-05,
+      "loss": 0.6899,
+      "mean_token_accuracy": 0.803005863726139,
+      "num_tokens": 78643200.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7069758854806423,
+      "epoch": 0.06426666666666667,
+      "grad_norm": 0.49161744117736816,
+      "learning_rate": 1.876524064171123e-05,
+      "loss": 0.6967,
+      "mean_token_accuracy": 0.801011124253273,
+      "num_tokens": 78970880.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7340013913810253,
+      "epoch": 0.06453333333333333,
+      "grad_norm": 0.5566353797912598,
+      "learning_rate": 1.8759893048128345e-05,
+      "loss": 0.727,
+      "mean_token_accuracy": 0.7926472336053848,
+      "num_tokens": 79298560.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.698576745390892,
+      "epoch": 0.0648,
+      "grad_norm": 0.5559256076812744,
+      "learning_rate": 1.8754545454545457e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.8034762978553772,
+      "num_tokens": 79626240.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.6885745882987976,
+      "epoch": 0.06506666666666666,
+      "grad_norm": 0.5433689951896667,
+      "learning_rate": 1.874919786096257e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8080125853419304,
+      "num_tokens": 79953920.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.6899311773478984,
+      "epoch": 0.06533333333333333,
+      "grad_norm": 0.5663343071937561,
+      "learning_rate": 1.874385026737968e-05,
+      "loss": 0.6851,
+      "mean_token_accuracy": 0.8039803296327591,
+      "num_tokens": 80281600.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7011104613542557,
+      "epoch": 0.0656,
+      "grad_norm": 0.455303817987442,
+      "learning_rate": 1.8738502673796792e-05,
+      "loss": 0.6851,
+      "mean_token_accuracy": 0.8048509284853935,
+      "num_tokens": 80609280.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.721408611536026,
+      "epoch": 0.06586666666666667,
+      "grad_norm": 0.5038881301879883,
+      "learning_rate": 1.8733155080213904e-05,
+      "loss": 0.7208,
+      "mean_token_accuracy": 0.7957477971911431,
+      "num_tokens": 80936960.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.7371288865804673,
+      "epoch": 0.06613333333333334,
+      "grad_norm": 0.600644588470459,
+      "learning_rate": 1.872780748663102e-05,
+      "loss": 0.7281,
+      "mean_token_accuracy": 0.7941990524530411,
+      "num_tokens": 81264640.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.6919610172510147,
+      "epoch": 0.0664,
+      "grad_norm": 0.5862867832183838,
+      "learning_rate": 1.872245989304813e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8039131224155426,
+      "num_tokens": 81592320.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.7386172406375409,
+      "epoch": 0.06666666666666667,
+      "grad_norm": 0.5530334711074829,
+      "learning_rate": 1.8717112299465243e-05,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.7973668143153191,
+      "num_tokens": 81920000.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7395651340484619,
+      "epoch": 0.06693333333333333,
+      "grad_norm": 0.5654215812683105,
+      "learning_rate": 1.8711764705882355e-05,
+      "loss": 0.7325,
+      "mean_token_accuracy": 0.791675828397274,
+      "num_tokens": 82247680.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7238991126418114,
+      "epoch": 0.0672,
+      "grad_norm": 0.5680257678031921,
+      "learning_rate": 1.8706417112299466e-05,
+      "loss": 0.7152,
+      "mean_token_accuracy": 0.7962212890386582,
+      "num_tokens": 82575360.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7331514492630958,
+      "epoch": 0.06746666666666666,
+      "grad_norm": 0.4596850275993347,
+      "learning_rate": 1.8701069518716578e-05,
+      "loss": 0.7219,
+      "mean_token_accuracy": 0.7966031298041344,
+      "num_tokens": 82903040.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7191907726228237,
+      "epoch": 0.06773333333333334,
+      "grad_norm": 0.5474050641059875,
+      "learning_rate": 1.8695721925133693e-05,
+      "loss": 0.709,
+      "mean_token_accuracy": 0.7983260005712509,
+      "num_tokens": 83230720.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.730270405113697,
+      "epoch": 0.068,
+      "grad_norm": 0.488977313041687,
+      "learning_rate": 1.8690374331550805e-05,
+      "loss": 0.7195,
+      "mean_token_accuracy": 0.7935361698269844,
+      "num_tokens": 83558400.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.6698543079197407,
+      "epoch": 0.06826666666666667,
+      "grad_norm": 0.4836066663265228,
+      "learning_rate": 1.8685026737967914e-05,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.8092833563685418,
+      "num_tokens": 83886080.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.6855506785213947,
+      "epoch": 0.06853333333333333,
+      "grad_norm": 0.5904099345207214,
+      "learning_rate": 1.867967914438503e-05,
+      "loss": 0.6773,
+      "mean_token_accuracy": 0.8049456238746643,
+      "num_tokens": 84213760.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.6905220828950405,
+      "epoch": 0.0688,
+      "grad_norm": 0.6067728996276855,
+      "learning_rate": 1.867433155080214e-05,
+      "loss": 0.6816,
+      "mean_token_accuracy": 0.8042919099330902,
+      "num_tokens": 84541440.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7021021544933319,
+      "epoch": 0.06906666666666667,
+      "grad_norm": 0.49781155586242676,
+      "learning_rate": 1.8668983957219252e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.8016923278570175,
+      "num_tokens": 84869120.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7121111631393433,
+      "epoch": 0.06933333333333333,
+      "grad_norm": 0.6203329563140869,
+      "learning_rate": 1.8663636363636368e-05,
+      "loss": 0.7095,
+      "mean_token_accuracy": 0.7976600632071496,
+      "num_tokens": 85196800.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.731182923913002,
+      "epoch": 0.0696,
+      "grad_norm": 0.5203900337219238,
+      "learning_rate": 1.8658288770053476e-05,
+      "loss": 0.7113,
+      "mean_token_accuracy": 0.7976172998547554,
+      "num_tokens": 85524480.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.6984265767037868,
+      "epoch": 0.06986666666666666,
+      "grad_norm": 0.49950000643730164,
+      "learning_rate": 1.8652941176470588e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.8012524425983429,
+      "num_tokens": 85852160.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7005846202373505,
+      "epoch": 0.07013333333333334,
+      "grad_norm": 0.543187141418457,
+      "learning_rate": 1.8647593582887703e-05,
+      "loss": 0.6888,
+      "mean_token_accuracy": 0.8040811344981194,
+      "num_tokens": 86179840.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.6949969455599785,
+      "epoch": 0.0704,
+      "grad_norm": 0.5879446268081665,
+      "learning_rate": 1.8642245989304815e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.8033602148294449,
+      "num_tokens": 86507520.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7094046354293824,
+      "epoch": 0.07066666666666667,
+      "grad_norm": 0.4720987379550934,
+      "learning_rate": 1.8636898395721927e-05,
+      "loss": 0.698,
+      "mean_token_accuracy": 0.7995234608650208,
+      "num_tokens": 86835200.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.6734897688031196,
+      "epoch": 0.07093333333333333,
+      "grad_norm": 0.4727756679058075,
+      "learning_rate": 1.863155080213904e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8089717745780944,
+      "num_tokens": 87162880.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7182955905795098,
+      "epoch": 0.0712,
+      "grad_norm": 0.5930288434028625,
+      "learning_rate": 1.862620320855615e-05,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.7971926927566528,
+      "num_tokens": 87490560.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7393495932221412,
+      "epoch": 0.07146666666666666,
+      "grad_norm": 0.45827335119247437,
+      "learning_rate": 1.8620855614973262e-05,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.7936522513628006,
+      "num_tokens": 87818240.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7197345927357673,
+      "epoch": 0.07173333333333333,
+      "grad_norm": 0.49268782138824463,
+      "learning_rate": 1.8615508021390377e-05,
+      "loss": 0.7124,
+      "mean_token_accuracy": 0.7954484403133393,
+      "num_tokens": 88145920.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.6882727935910224,
+      "epoch": 0.072,
+      "grad_norm": 0.49876537919044495,
+      "learning_rate": 1.861016042780749e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8026851177215576,
+      "num_tokens": 88473600.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7437757819890976,
+      "epoch": 0.07226666666666667,
+      "grad_norm": 0.4466654062271118,
+      "learning_rate": 1.86048128342246e-05,
+      "loss": 0.7364,
+      "mean_token_accuracy": 0.7899499014019966,
+      "num_tokens": 88801280.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.7106141909956932,
+      "epoch": 0.07253333333333334,
+      "grad_norm": 0.6582798957824707,
+      "learning_rate": 1.8599465240641713e-05,
+      "loss": 0.7013,
+      "mean_token_accuracy": 0.8006231695413589,
+      "num_tokens": 89128960.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.6909909635782242,
+      "epoch": 0.0728,
+      "grad_norm": 0.5060355067253113,
+      "learning_rate": 1.8594117647058824e-05,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.8035312771797181,
+      "num_tokens": 89456640.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.6932483226060867,
+      "epoch": 0.07306666666666667,
+      "grad_norm": 0.5097729563713074,
+      "learning_rate": 1.8588770053475936e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.8053549632430077,
+      "num_tokens": 89784320.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.6751238405704498,
+      "epoch": 0.07333333333333333,
+      "grad_norm": 0.5460449457168579,
+      "learning_rate": 1.858342245989305e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.8071939155459404,
+      "num_tokens": 90112000.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7026051998138427,
+      "epoch": 0.0736,
+      "grad_norm": 0.5310487747192383,
+      "learning_rate": 1.8578074866310163e-05,
+      "loss": 0.6955,
+      "mean_token_accuracy": 0.8018206223845482,
+      "num_tokens": 90439680.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.690500158071518,
+      "epoch": 0.07386666666666666,
+      "grad_norm": 0.553963303565979,
+      "learning_rate": 1.8572727272727272e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.8035557180643081,
+      "num_tokens": 90767360.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.673224987089634,
+      "epoch": 0.07413333333333333,
+      "grad_norm": 0.5649405121803284,
+      "learning_rate": 1.8567379679144387e-05,
+      "loss": 0.6552,
+      "mean_token_accuracy": 0.8085166156291962,
+      "num_tokens": 91095040.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.6903910547494888,
+      "epoch": 0.0744,
+      "grad_norm": 0.42968282103538513,
+      "learning_rate": 1.85620320855615e-05,
+      "loss": 0.6833,
+      "mean_token_accuracy": 0.8032624661922455,
+      "num_tokens": 91422720.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7024453252553939,
+      "epoch": 0.07466666666666667,
+      "grad_norm": 0.508698046207428,
+      "learning_rate": 1.855668449197861e-05,
+      "loss": 0.6929,
+      "mean_token_accuracy": 0.8014418363571167,
+      "num_tokens": 91750400.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.6843134768307209,
+      "epoch": 0.07493333333333334,
+      "grad_norm": 0.6219918131828308,
+      "learning_rate": 1.8551336898395726e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.8044324308633805,
+      "num_tokens": 92078080.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.6933091588318347,
+      "epoch": 0.0752,
+      "grad_norm": 0.5639626383781433,
+      "learning_rate": 1.8545989304812834e-05,
+      "loss": 0.6849,
+      "mean_token_accuracy": 0.8041972160339356,
+      "num_tokens": 92405760.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.6970530793070793,
+      "epoch": 0.07546666666666667,
+      "grad_norm": 0.5608235001564026,
+      "learning_rate": 1.8540641711229946e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8021169364452362,
+      "num_tokens": 92733440.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7037189900875092,
+      "epoch": 0.07573333333333333,
+      "grad_norm": 0.4553059935569763,
+      "learning_rate": 1.853529411764706e-05,
+      "loss": 0.6974,
+      "mean_token_accuracy": 0.7995081886649131,
+      "num_tokens": 93061120.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7386654257774353,
+      "epoch": 0.076,
+      "grad_norm": 0.5346807837486267,
+      "learning_rate": 1.8529946524064173e-05,
+      "loss": 0.7261,
+      "mean_token_accuracy": 0.7940982401371002,
+      "num_tokens": 93388800.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.6811931923031807,
+      "epoch": 0.07626666666666666,
+      "grad_norm": 0.44799095392227173,
+      "learning_rate": 1.8524598930481285e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.8042552575469017,
+      "num_tokens": 93716480.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.6743794098496437,
+      "epoch": 0.07653333333333333,
+      "grad_norm": 0.47558820247650146,
+      "learning_rate": 1.8519251336898397e-05,
+      "loss": 0.6658,
+      "mean_token_accuracy": 0.8095766112208367,
+      "num_tokens": 94044160.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.6963659703731537,
+      "epoch": 0.0768,
+      "grad_norm": 0.44408637285232544,
+      "learning_rate": 1.851390374331551e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.8036198705434799,
+      "num_tokens": 94371840.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.668525793403387,
+      "epoch": 0.07706666666666667,
+      "grad_norm": 0.6240969300270081,
+      "learning_rate": 1.850855614973262e-05,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.808345553278923,
+      "num_tokens": 94699520.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.6763082042336463,
+      "epoch": 0.07733333333333334,
+      "grad_norm": 0.5965229868888855,
+      "learning_rate": 1.8503208556149735e-05,
+      "loss": 0.6642,
+      "mean_token_accuracy": 0.8080003693699837,
+      "num_tokens": 95027200.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7069110468029975,
+      "epoch": 0.0776,
+      "grad_norm": 0.502577543258667,
+      "learning_rate": 1.8497860962566847e-05,
+      "loss": 0.6967,
+      "mean_token_accuracy": 0.8004185035824776,
+      "num_tokens": 95354880.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.6777268260717392,
+      "epoch": 0.07786666666666667,
+      "grad_norm": 0.5277969837188721,
+      "learning_rate": 1.849251336898396e-05,
+      "loss": 0.6703,
+      "mean_token_accuracy": 0.8061033710837364,
+      "num_tokens": 95682560.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7034938246011734,
+      "epoch": 0.07813333333333333,
+      "grad_norm": 0.48426011204719543,
+      "learning_rate": 1.848716577540107e-05,
+      "loss": 0.6962,
+      "mean_token_accuracy": 0.7995967760682106,
+      "num_tokens": 96010240.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.70678820759058,
+      "epoch": 0.0784,
+      "grad_norm": 0.608834981918335,
+      "learning_rate": 1.8481818181818183e-05,
+      "loss": 0.6957,
+      "mean_token_accuracy": 0.8017839655280113,
+      "num_tokens": 96337920.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.6910052418708801,
+      "epoch": 0.07866666666666666,
+      "grad_norm": 0.6024418473243713,
+      "learning_rate": 1.8476470588235294e-05,
+      "loss": 0.6841,
+      "mean_token_accuracy": 0.8038795247673989,
+      "num_tokens": 96665600.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.6798450261354446,
+      "epoch": 0.07893333333333333,
+      "grad_norm": 0.548592209815979,
+      "learning_rate": 1.847112299465241e-05,
+      "loss": 0.6659,
+      "mean_token_accuracy": 0.8090145409107208,
+      "num_tokens": 96993280.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.679200055450201,
+      "epoch": 0.0792,
+      "grad_norm": 0.47433775663375854,
+      "learning_rate": 1.846577540106952e-05,
+      "loss": 0.6718,
+      "mean_token_accuracy": 0.8076796188950539,
+      "num_tokens": 97320960.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.6941119194030761,
+      "epoch": 0.07946666666666667,
+      "grad_norm": 0.6369531154632568,
+      "learning_rate": 1.8460427807486633e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8014357283711433,
+      "num_tokens": 97648640.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7004023909568786,
+      "epoch": 0.07973333333333334,
+      "grad_norm": 0.5593435764312744,
+      "learning_rate": 1.8455080213903745e-05,
+      "loss": 0.6882,
+      "mean_token_accuracy": 0.8030547395348548,
+      "num_tokens": 97976320.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7129723727703094,
+      "epoch": 0.08,
+      "grad_norm": 0.5572832226753235,
+      "learning_rate": 1.8449732620320857e-05,
+      "loss": 0.7114,
+      "mean_token_accuracy": 0.7975562065839767,
+      "num_tokens": 98304000.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.7065567284822464,
+      "epoch": 0.08026666666666667,
+      "grad_norm": 0.5433133840560913,
+      "learning_rate": 1.844438502673797e-05,
+      "loss": 0.6889,
+      "mean_token_accuracy": 0.8037970453500748,
+      "num_tokens": 98631680.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.710640449821949,
+      "epoch": 0.08053333333333333,
+      "grad_norm": 0.6438500881195068,
+      "learning_rate": 1.8439037433155084e-05,
+      "loss": 0.7056,
+      "mean_token_accuracy": 0.7980846777558327,
+      "num_tokens": 98959360.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.7123002260923386,
+      "epoch": 0.0808,
+      "grad_norm": 0.5158177018165588,
+      "learning_rate": 1.8433689839572192e-05,
+      "loss": 0.7038,
+      "mean_token_accuracy": 0.7997495114803315,
+      "num_tokens": 99287040.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.6933645069599151,
+      "epoch": 0.08106666666666666,
+      "grad_norm": 0.6604756712913513,
+      "learning_rate": 1.8428342245989307e-05,
+      "loss": 0.6727,
+      "mean_token_accuracy": 0.8053671807050705,
+      "num_tokens": 99614720.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6930786930024624,
+      "epoch": 0.08133333333333333,
+      "grad_norm": 0.5294865369796753,
+      "learning_rate": 1.842299465240642e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.8041758313775063,
+      "num_tokens": 99942400.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6785897724330425,
+      "epoch": 0.0816,
+      "grad_norm": 0.5470296740531921,
+      "learning_rate": 1.841764705882353e-05,
+      "loss": 0.6728,
+      "mean_token_accuracy": 0.8063447013497352,
+      "num_tokens": 100270080.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972546666860581,
+      "epoch": 0.08186666666666667,
+      "grad_norm": 0.5774398446083069,
+      "learning_rate": 1.8412299465240643e-05,
+      "loss": 0.6842,
+      "mean_token_accuracy": 0.8032013714313507,
+      "num_tokens": 100597760.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6924749717116356,
+      "epoch": 0.08213333333333334,
+      "grad_norm": 0.49980732798576355,
+      "learning_rate": 1.8406951871657755e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.801777857542038,
+      "num_tokens": 100925440.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6787292137742043,
+      "epoch": 0.0824,
+      "grad_norm": 0.5680364370346069,
+      "learning_rate": 1.8401604278074866e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.8057490229606629,
+      "num_tokens": 101253120.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6913809821009635,
+      "epoch": 0.08266666666666667,
+      "grad_norm": 0.5436722040176392,
+      "learning_rate": 1.8396256684491978e-05,
+      "loss": 0.6769,
+      "mean_token_accuracy": 0.804750120639801,
+      "num_tokens": 101580800.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6795099377632141,
+      "epoch": 0.08293333333333333,
+      "grad_norm": 0.4478617310523987,
+      "learning_rate": 1.8390909090909093e-05,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.8080370232462883,
+      "num_tokens": 101908480.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6731130488216877,
+      "epoch": 0.0832,
+      "grad_norm": 0.562049925327301,
+      "learning_rate": 1.8385561497326205e-05,
+      "loss": 0.66,
+      "mean_token_accuracy": 0.8096041113138199,
+      "num_tokens": 102236160.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.6568302936851979,
+      "epoch": 0.08346666666666666,
+      "grad_norm": 0.49974524974823,
+      "learning_rate": 1.8380213903743317e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.8136149808764458,
+      "num_tokens": 102563840.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.7086566746234894,
+      "epoch": 0.08373333333333334,
+      "grad_norm": 0.5515449643135071,
+      "learning_rate": 1.837486631016043e-05,
+      "loss": 0.6902,
+      "mean_token_accuracy": 0.8030486315488815,
+      "num_tokens": 102891520.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.6682414583861828,
+      "epoch": 0.084,
+      "grad_norm": 0.5746288895606995,
+      "learning_rate": 1.836951871657754e-05,
+      "loss": 0.6634,
+      "mean_token_accuracy": 0.8090542510151864,
+      "num_tokens": 103219200.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.709159791469574,
+      "epoch": 0.08426666666666667,
+      "grad_norm": 0.5816189646720886,
+      "learning_rate": 1.8364171122994653e-05,
+      "loss": 0.6953,
+      "mean_token_accuracy": 0.8001588493585586,
+      "num_tokens": 103546880.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.6834252879023552,
+      "epoch": 0.08453333333333334,
+      "grad_norm": 0.5478416681289673,
+      "learning_rate": 1.8358823529411768e-05,
+      "loss": 0.6745,
+      "mean_token_accuracy": 0.8062347248196602,
+      "num_tokens": 103874560.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.6626397490501403,
+      "epoch": 0.0848,
+      "grad_norm": 0.46254318952560425,
+      "learning_rate": 1.835347593582888e-05,
+      "loss": 0.6535,
+      "mean_token_accuracy": 0.8101050838828087,
+      "num_tokens": 104202240.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.6958767995238304,
+      "epoch": 0.08506666666666667,
+      "grad_norm": 0.5705205202102661,
+      "learning_rate": 1.834812834224599e-05,
+      "loss": 0.6807,
+      "mean_token_accuracy": 0.8055046394467353,
+      "num_tokens": 104529920.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.6469135418534279,
+      "epoch": 0.08533333333333333,
+      "grad_norm": 0.41772717237472534,
+      "learning_rate": 1.8342780748663103e-05,
+      "loss": 0.6429,
+      "mean_token_accuracy": 0.8131292730569839,
+      "num_tokens": 104857600.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.6653836384415627,
+      "epoch": 0.0856,
+      "grad_norm": 0.5370784401893616,
+      "learning_rate": 1.8337433155080215e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.8107251957058906,
+      "num_tokens": 105185280.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.6882105425000191,
+      "epoch": 0.08586666666666666,
+      "grad_norm": 0.5651310086250305,
+      "learning_rate": 1.8332085561497327e-05,
+      "loss": 0.6827,
+      "mean_token_accuracy": 0.8047653943300247,
+      "num_tokens": 105512960.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.685054711252451,
+      "epoch": 0.08613333333333334,
+      "grad_norm": 0.5525728464126587,
+      "learning_rate": 1.8326737967914442e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.8072519540786743,
+      "num_tokens": 105840640.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.6843914739787579,
+      "epoch": 0.0864,
+      "grad_norm": 0.5613085031509399,
+      "learning_rate": 1.832139037433155e-05,
+      "loss": 0.6764,
+      "mean_token_accuracy": 0.8058712184429169,
+      "num_tokens": 106168320.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.6781938016414643,
+      "epoch": 0.08666666666666667,
+      "grad_norm": 0.5632696151733398,
+      "learning_rate": 1.8316042780748666e-05,
+      "loss": 0.6704,
+      "mean_token_accuracy": 0.8056573808193207,
+      "num_tokens": 106496000.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.6432853825390339,
+      "epoch": 0.08693333333333333,
+      "grad_norm": 0.49366846680641174,
+      "learning_rate": 1.8310695187165777e-05,
+      "loss": 0.6296,
+      "mean_token_accuracy": 0.8155913963913918,
+      "num_tokens": 106823680.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.6580203235149383,
+      "epoch": 0.0872,
+      "grad_norm": 0.48049911856651306,
+      "learning_rate": 1.830534759358289e-05,
+      "loss": 0.6487,
+      "mean_token_accuracy": 0.8130865097045898,
+      "num_tokens": 107151360.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.6701589055359364,
+      "epoch": 0.08746666666666666,
+      "grad_norm": 0.5092346668243408,
+      "learning_rate": 1.83e-05,
+      "loss": 0.6629,
+      "mean_token_accuracy": 0.8087671101093292,
+      "num_tokens": 107479040.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.6702375099062919,
+      "epoch": 0.08773333333333333,
+      "grad_norm": 0.46218451857566833,
+      "learning_rate": 1.8294652406417113e-05,
+      "loss": 0.6648,
+      "mean_token_accuracy": 0.8095246806740761,
+      "num_tokens": 107806720.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.6846297360956669,
+      "epoch": 0.088,
+      "grad_norm": 0.5321958065032959,
+      "learning_rate": 1.8289304812834225e-05,
+      "loss": 0.673,
+      "mean_token_accuracy": 0.8060881033539772,
+      "num_tokens": 108134400.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.6802199646830559,
+      "epoch": 0.08826666666666666,
+      "grad_norm": 0.5790373682975769,
+      "learning_rate": 1.828395721925134e-05,
+      "loss": 0.6694,
+      "mean_token_accuracy": 0.8068029090762139,
+      "num_tokens": 108462080.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.6756490565836429,
+      "epoch": 0.08853333333333334,
+      "grad_norm": 0.4985637664794922,
+      "learning_rate": 1.827860962566845e-05,
+      "loss": 0.6626,
+      "mean_token_accuracy": 0.8084799617528915,
+      "num_tokens": 108789760.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.6908547803759575,
+      "epoch": 0.0888,
+      "grad_norm": 0.5945738554000854,
+      "learning_rate": 1.8273262032085563e-05,
+      "loss": 0.6881,
+      "mean_token_accuracy": 0.8015976309776306,
+      "num_tokens": 109117440.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.6680992089211941,
+      "epoch": 0.08906666666666667,
+      "grad_norm": 0.5361096858978271,
+      "learning_rate": 1.8267914438502675e-05,
+      "loss": 0.6576,
+      "mean_token_accuracy": 0.8096193805336952,
+      "num_tokens": 109445120.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.6517356902360916,
+      "epoch": 0.08933333333333333,
+      "grad_norm": 0.5997782945632935,
+      "learning_rate": 1.8262566844919787e-05,
+      "loss": 0.6393,
+      "mean_token_accuracy": 0.8154294982552528,
+      "num_tokens": 109772800.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.6735978633165359,
+      "epoch": 0.0896,
+      "grad_norm": 0.4732246398925781,
+      "learning_rate": 1.82572192513369e-05,
+      "loss": 0.6679,
+      "mean_token_accuracy": 0.8071022748947143,
+      "num_tokens": 110100480.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.673013449460268,
+      "epoch": 0.08986666666666666,
+      "grad_norm": 0.5807483792304993,
+      "learning_rate": 1.825187165775401e-05,
+      "loss": 0.6524,
+      "mean_token_accuracy": 0.8115774661302566,
+      "num_tokens": 110428160.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6996243745088577,
+      "epoch": 0.09013333333333333,
+      "grad_norm": 0.5526395440101624,
+      "learning_rate": 1.8246524064171126e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8016220644116402,
+      "num_tokens": 110755840.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.6826088994741439,
+      "epoch": 0.0904,
+      "grad_norm": 0.5243359208106995,
+      "learning_rate": 1.8241176470588238e-05,
+      "loss": 0.6734,
+      "mean_token_accuracy": 0.8074627310037613,
+      "num_tokens": 111083520.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.6364914231002331,
+      "epoch": 0.09066666666666667,
+      "grad_norm": 0.549103856086731,
+      "learning_rate": 1.823582887700535e-05,
+      "loss": 0.6313,
+      "mean_token_accuracy": 0.8167949721217156,
+      "num_tokens": 111411200.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7150142952799797,
+      "epoch": 0.09093333333333334,
+      "grad_norm": 0.5956588387489319,
+      "learning_rate": 1.823048128342246e-05,
+      "loss": 0.6999,
+      "mean_token_accuracy": 0.8005926236510277,
+      "num_tokens": 111738880.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.6763354152441025,
+      "epoch": 0.0912,
+      "grad_norm": 0.5426291823387146,
+      "learning_rate": 1.8225133689839573e-05,
+      "loss": 0.6687,
+      "mean_token_accuracy": 0.8070381224155426,
+      "num_tokens": 112066560.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7101250626146793,
+      "epoch": 0.09146666666666667,
+      "grad_norm": 0.8604161143302917,
+      "learning_rate": 1.8219786096256685e-05,
+      "loss": 0.7011,
+      "mean_token_accuracy": 0.8009255886077881,
+      "num_tokens": 112394240.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.6529824391007424,
+      "epoch": 0.09173333333333333,
+      "grad_norm": 0.45931896567344666,
+      "learning_rate": 1.82144385026738e-05,
+      "loss": 0.6454,
+      "mean_token_accuracy": 0.810450266301632,
+      "num_tokens": 112721920.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.6730405807495117,
+      "epoch": 0.092,
+      "grad_norm": 0.5787823796272278,
+      "learning_rate": 1.820909090909091e-05,
+      "loss": 0.6627,
+      "mean_token_accuracy": 0.808226415514946,
+      "num_tokens": 113049600.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.6584693357348442,
+      "epoch": 0.09226666666666666,
+      "grad_norm": 0.5029930472373962,
+      "learning_rate": 1.8203743315508024e-05,
+      "loss": 0.6439,
+      "mean_token_accuracy": 0.8129337728023529,
+      "num_tokens": 113377280.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.6774670705199242,
+      "epoch": 0.09253333333333333,
+      "grad_norm": 0.45803895592689514,
+      "learning_rate": 1.8198395721925135e-05,
+      "loss": 0.6716,
+      "mean_token_accuracy": 0.8076612904667855,
+      "num_tokens": 113704960.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.6801780223846435,
+      "epoch": 0.0928,
+      "grad_norm": 0.46918779611587524,
+      "learning_rate": 1.8193048128342247e-05,
+      "loss": 0.6704,
+      "mean_token_accuracy": 0.8068792790174484,
+      "num_tokens": 114032640.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.6732617512345314,
+      "epoch": 0.09306666666666667,
+      "grad_norm": 0.5564458966255188,
+      "learning_rate": 1.818770053475936e-05,
+      "loss": 0.6632,
+      "mean_token_accuracy": 0.8086021527647972,
+      "num_tokens": 114360320.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.6835436381399631,
+      "epoch": 0.09333333333333334,
+      "grad_norm": 0.6502693891525269,
+      "learning_rate": 1.818235294117647e-05,
+      "loss": 0.6709,
+      "mean_token_accuracy": 0.8071542054414749,
+      "num_tokens": 114688000.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.6898898124694824,
+      "epoch": 0.0936,
+      "grad_norm": 0.5657341480255127,
+      "learning_rate": 1.8177005347593583e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.8055382490158081,
+      "num_tokens": 115015680.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.6327576048672199,
+      "epoch": 0.09386666666666667,
+      "grad_norm": 0.5169603228569031,
+      "learning_rate": 1.8171657754010698e-05,
+      "loss": 0.6246,
+      "mean_token_accuracy": 0.8177663758397102,
+      "num_tokens": 115343360.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.6864586338400841,
+      "epoch": 0.09413333333333333,
+      "grad_norm": 0.5316112637519836,
+      "learning_rate": 1.816631016042781e-05,
+      "loss": 0.6717,
+      "mean_token_accuracy": 0.8051533445715904,
+      "num_tokens": 115671040.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.6481480613350868,
+      "epoch": 0.0944,
+      "grad_norm": 0.4560836851596832,
+      "learning_rate": 1.816096256684492e-05,
+      "loss": 0.639,
+      "mean_token_accuracy": 0.8150598749518394,
+      "num_tokens": 115998720.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.6558339282870292,
+      "epoch": 0.09466666666666666,
+      "grad_norm": 0.6353741884231567,
+      "learning_rate": 1.8155614973262033e-05,
+      "loss": 0.6459,
+      "mean_token_accuracy": 0.8130468010902405,
+      "num_tokens": 116326400.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.6471247412264347,
+      "epoch": 0.09493333333333333,
+      "grad_norm": 0.6068600416183472,
+      "learning_rate": 1.8150267379679145e-05,
+      "loss": 0.6318,
+      "mean_token_accuracy": 0.8173356547951698,
+      "num_tokens": 116654080.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.6869985245168209,
+      "epoch": 0.0952,
+      "grad_norm": 0.6087508797645569,
+      "learning_rate": 1.8144919786096257e-05,
+      "loss": 0.6814,
+      "mean_token_accuracy": 0.804798997938633,
+      "num_tokens": 116981760.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.6443760603666305,
+      "epoch": 0.09546666666666667,
+      "grad_norm": 0.5366358757019043,
+      "learning_rate": 1.8139572192513372e-05,
+      "loss": 0.6351,
+      "mean_token_accuracy": 0.8143267348408699,
+      "num_tokens": 117309440.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.6869163364171982,
+      "epoch": 0.09573333333333334,
+      "grad_norm": 0.5238228440284729,
+      "learning_rate": 1.8134224598930484e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.8057917848229408,
+      "num_tokens": 117637120.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.6869014196097851,
+      "epoch": 0.096,
+      "grad_norm": 0.5857342481613159,
+      "learning_rate": 1.8128877005347596e-05,
+      "loss": 0.6723,
+      "mean_token_accuracy": 0.8094147086143494,
+      "num_tokens": 117964800.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.6552226409316063,
+      "epoch": 0.09626666666666667,
+      "grad_norm": 0.6696491837501526,
+      "learning_rate": 1.8123529411764708e-05,
+      "loss": 0.6496,
+      "mean_token_accuracy": 0.810438048839569,
+      "num_tokens": 118292480.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.6580488629639148,
+      "epoch": 0.09653333333333333,
+      "grad_norm": 0.6682412028312683,
+      "learning_rate": 1.811818181818182e-05,
+      "loss": 0.6377,
+      "mean_token_accuracy": 0.815374507009983,
+      "num_tokens": 118620160.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.6649587631225586,
+      "epoch": 0.0968,
+      "grad_norm": 0.481740266084671,
+      "learning_rate": 1.811283422459893e-05,
+      "loss": 0.6611,
+      "mean_token_accuracy": 0.809032866358757,
+      "num_tokens": 118947840.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.6486426994204522,
+      "epoch": 0.09706666666666666,
+      "grad_norm": 0.4654616117477417,
+      "learning_rate": 1.8107486631016043e-05,
+      "loss": 0.6287,
+      "mean_token_accuracy": 0.8167491465806961,
+      "num_tokens": 119275520.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6471065364778041,
+      "epoch": 0.09733333333333333,
+      "grad_norm": 0.5740672945976257,
+      "learning_rate": 1.8102139037433158e-05,
+      "loss": 0.6398,
+      "mean_token_accuracy": 0.8137157917022705,
+      "num_tokens": 119603200.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6453163996338844,
+      "epoch": 0.0976,
+      "grad_norm": 0.5717991590499878,
+      "learning_rate": 1.8096791443850267e-05,
+      "loss": 0.6337,
+      "mean_token_accuracy": 0.814467254281044,
+      "num_tokens": 119930880.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6891070395708084,
+      "epoch": 0.09786666666666667,
+      "grad_norm": 0.4824608564376831,
+      "learning_rate": 1.8091443850267382e-05,
+      "loss": 0.6853,
+      "mean_token_accuracy": 0.804340785741806,
+      "num_tokens": 120258560.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6633891932666302,
+      "epoch": 0.09813333333333334,
+      "grad_norm": 0.6122506856918335,
+      "learning_rate": 1.8086096256684494e-05,
+      "loss": 0.6521,
+      "mean_token_accuracy": 0.8096041068434715,
+      "num_tokens": 120586240.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6793444007635117,
+      "epoch": 0.0984,
+      "grad_norm": 0.5250032544136047,
+      "learning_rate": 1.8080748663101605e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.809081745147705,
+      "num_tokens": 120913920.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.657922463864088,
+      "epoch": 0.09866666666666667,
+      "grad_norm": 0.5360099673271179,
+      "learning_rate": 1.8075401069518717e-05,
+      "loss": 0.6558,
+      "mean_token_accuracy": 0.8111375838518142,
+      "num_tokens": 121241600.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6484499901533127,
+      "epoch": 0.09893333333333333,
+      "grad_norm": 0.5630919933319092,
+      "learning_rate": 1.807005347593583e-05,
+      "loss": 0.6339,
+      "mean_token_accuracy": 0.8160160079598426,
+      "num_tokens": 121569280.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.6766409076750278,
+      "epoch": 0.0992,
+      "grad_norm": 0.4403109550476074,
+      "learning_rate": 1.806470588235294e-05,
+      "loss": 0.6667,
+      "mean_token_accuracy": 0.806356917321682,
+      "num_tokens": 121896960.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6551111288368702,
+      "epoch": 0.09946666666666666,
+      "grad_norm": 0.5420662760734558,
+      "learning_rate": 1.8059358288770056e-05,
+      "loss": 0.6484,
+      "mean_token_accuracy": 0.8131964832544327,
+      "num_tokens": 122224640.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6386684224009513,
+      "epoch": 0.09973333333333333,
+      "grad_norm": 0.4553351402282715,
+      "learning_rate": 1.8054010695187168e-05,
+      "loss": 0.6304,
+      "mean_token_accuracy": 0.8156769275665283,
+      "num_tokens": 122552320.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6490894019603729,
+      "epoch": 0.1,
+      "grad_norm": 0.5522193908691406,
+      "learning_rate": 1.804866310160428e-05,
+      "loss": 0.6355,
+      "mean_token_accuracy": 0.8151606768369675,
+      "num_tokens": 122880000.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6576007321476937,
+      "epoch": 0.10026666666666667,
+      "grad_norm": 0.5378384590148926,
+      "learning_rate": 1.804331550802139e-05,
+      "loss": 0.6487,
+      "mean_token_accuracy": 0.8118676722049714,
+      "num_tokens": 123207680.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.6551145255565644,
+      "epoch": 0.10053333333333334,
+      "grad_norm": 0.5451924204826355,
+      "learning_rate": 1.8037967914438503e-05,
+      "loss": 0.6471,
+      "mean_token_accuracy": 0.8133736565709114,
+      "num_tokens": 123535360.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.663975004851818,
+      "epoch": 0.1008,
+      "grad_norm": 0.5291251540184021,
+      "learning_rate": 1.8032620320855615e-05,
+      "loss": 0.6516,
+      "mean_token_accuracy": 0.8121242687106133,
+      "num_tokens": 123863040.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6401459172368049,
+      "epoch": 0.10106666666666667,
+      "grad_norm": 0.4927455484867096,
+      "learning_rate": 1.802727272727273e-05,
+      "loss": 0.6322,
+      "mean_token_accuracy": 0.8152798160910606,
+      "num_tokens": 124190720.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6521815165877343,
+      "epoch": 0.10133333333333333,
+      "grad_norm": 0.5538015961647034,
+      "learning_rate": 1.8021925133689842e-05,
+      "loss": 0.638,
+      "mean_token_accuracy": 0.8130498558282853,
+      "num_tokens": 124518400.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6528400540351867,
+      "epoch": 0.1016,
+      "grad_norm": 0.5210300087928772,
+      "learning_rate": 1.8016577540106954e-05,
+      "loss": 0.6449,
+      "mean_token_accuracy": 0.8122067421674728,
+      "num_tokens": 124846080.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6538536071777343,
+      "epoch": 0.10186666666666666,
+      "grad_norm": 0.49153628945350647,
+      "learning_rate": 1.8011229946524066e-05,
+      "loss": 0.642,
+      "mean_token_accuracy": 0.8124358490109443,
+      "num_tokens": 125173760.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6925169438123703,
+      "epoch": 0.10213333333333334,
+      "grad_norm": 0.4965587854385376,
+      "learning_rate": 1.8005882352941177e-05,
+      "loss": 0.6796,
+      "mean_token_accuracy": 0.8029356077313423,
+      "num_tokens": 125501440.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6405668064951897,
+      "epoch": 0.1024,
+      "grad_norm": 0.4650726020336151,
+      "learning_rate": 1.800053475935829e-05,
+      "loss": 0.6367,
+      "mean_token_accuracy": 0.8139846041798592,
+      "num_tokens": 125829120.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6471795573830604,
+      "epoch": 0.10266666666666667,
+      "grad_norm": 0.5829203128814697,
+      "learning_rate": 1.7995187165775404e-05,
+      "loss": 0.6382,
+      "mean_token_accuracy": 0.8144947439432144,
+      "num_tokens": 126156800.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6493782468140126,
+      "epoch": 0.10293333333333334,
+      "grad_norm": 0.4290027916431427,
+      "learning_rate": 1.7989839572192516e-05,
+      "loss": 0.6464,
+      "mean_token_accuracy": 0.8140120953321457,
+      "num_tokens": 126484480.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6586277067661286,
+      "epoch": 0.1032,
+      "grad_norm": 0.5769961476325989,
+      "learning_rate": 1.7984491978609625e-05,
+      "loss": 0.6463,
+      "mean_token_accuracy": 0.8136119276285172,
+      "num_tokens": 126812160.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.665315467864275,
+      "epoch": 0.10346666666666667,
+      "grad_norm": 0.5212368965148926,
+      "learning_rate": 1.797914438502674e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.8086540818214416,
+      "num_tokens": 127139840.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6647671014070511,
+      "epoch": 0.10373333333333333,
+      "grad_norm": 0.5242336988449097,
+      "learning_rate": 1.7973796791443852e-05,
+      "loss": 0.6473,
+      "mean_token_accuracy": 0.8119868040084839,
+      "num_tokens": 127467520.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6698678009212017,
+      "epoch": 0.104,
+      "grad_norm": 0.47622811794281006,
+      "learning_rate": 1.7968449197860964e-05,
+      "loss": 0.6691,
+      "mean_token_accuracy": 0.8059536918997765,
+      "num_tokens": 127795200.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6539292819797993,
+      "epoch": 0.10426666666666666,
+      "grad_norm": 0.5479148626327515,
+      "learning_rate": 1.796310160427808e-05,
+      "loss": 0.6418,
+      "mean_token_accuracy": 0.815692201256752,
+      "num_tokens": 128122880.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6711996026337147,
+      "epoch": 0.10453333333333334,
+      "grad_norm": 0.8144274353981018,
+      "learning_rate": 1.7957754010695187e-05,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.8106488212943077,
+      "num_tokens": 128450560.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6411418944597245,
+      "epoch": 0.1048,
+      "grad_norm": 0.5335118174552917,
+      "learning_rate": 1.79524064171123e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8152278810739517,
+      "num_tokens": 128778240.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6617153793573379,
+      "epoch": 0.10506666666666667,
+      "grad_norm": 0.5851503610610962,
+      "learning_rate": 1.7947058823529414e-05,
+      "loss": 0.6466,
+      "mean_token_accuracy": 0.8120478957891464,
+      "num_tokens": 129105920.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6449001625180244,
+      "epoch": 0.10533333333333333,
+      "grad_norm": 0.5159755945205688,
+      "learning_rate": 1.7941711229946526e-05,
+      "loss": 0.6356,
+      "mean_token_accuracy": 0.8163520261645317,
+      "num_tokens": 129433600.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6193127304315567,
+      "epoch": 0.1056,
+      "grad_norm": 0.48211485147476196,
+      "learning_rate": 1.7936363636363638e-05,
+      "loss": 0.6087,
+      "mean_token_accuracy": 0.820433160662651,
+      "num_tokens": 129761280.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6611559122800827,
+      "epoch": 0.10586666666666666,
+      "grad_norm": 0.5552424788475037,
+      "learning_rate": 1.793101604278075e-05,
+      "loss": 0.651,
+      "mean_token_accuracy": 0.8119440376758575,
+      "num_tokens": 130088960.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6432381875813007,
+      "epoch": 0.10613333333333333,
+      "grad_norm": 0.6014645099639893,
+      "learning_rate": 1.792566844919786e-05,
+      "loss": 0.6369,
+      "mean_token_accuracy": 0.8135630518198014,
+      "num_tokens": 130416640.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6687833905220032,
+      "epoch": 0.1064,
+      "grad_norm": 0.5963072180747986,
+      "learning_rate": 1.7920320855614973e-05,
+      "loss": 0.658,
+      "mean_token_accuracy": 0.809026762843132,
+      "num_tokens": 130744320.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6530608497560024,
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.7460877299308777,
+      "learning_rate": 1.791497326203209e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8133278340101242,
+      "num_tokens": 131072000.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6596694678068161,
+      "epoch": 0.10693333333333334,
+      "grad_norm": 0.5461241602897644,
+      "learning_rate": 1.79096256684492e-05,
+      "loss": 0.6431,
+      "mean_token_accuracy": 0.8132056474685669,
+      "num_tokens": 131399680.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.628665404766798,
+      "epoch": 0.1072,
+      "grad_norm": 0.47004592418670654,
+      "learning_rate": 1.7904278074866312e-05,
+      "loss": 0.6237,
+      "mean_token_accuracy": 0.8165689170360565,
+      "num_tokens": 131727360.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6237076640129089,
+      "epoch": 0.10746666666666667,
+      "grad_norm": 0.48614516854286194,
+      "learning_rate": 1.7898930481283424e-05,
+      "loss": 0.6118,
+      "mean_token_accuracy": 0.820772236585617,
+      "num_tokens": 132055040.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6402423575520515,
+      "epoch": 0.10773333333333333,
+      "grad_norm": 0.6164606809616089,
+      "learning_rate": 1.7893582887700536e-05,
+      "loss": 0.6331,
+      "mean_token_accuracy": 0.8167491465806961,
+      "num_tokens": 132382720.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6768997862935067,
+      "epoch": 0.108,
+      "grad_norm": 0.48716631531715393,
+      "learning_rate": 1.7888235294117647e-05,
+      "loss": 0.6601,
+      "mean_token_accuracy": 0.8081714317202568,
+      "num_tokens": 132710400.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.671170400083065,
+      "epoch": 0.10826666666666666,
+      "grad_norm": 0.6060397028923035,
+      "learning_rate": 1.7882887700534763e-05,
+      "loss": 0.6596,
+      "mean_token_accuracy": 0.8107618570327759,
+      "num_tokens": 133038080.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.655850450694561,
+      "epoch": 0.10853333333333333,
+      "grad_norm": 0.49084195494651794,
+      "learning_rate": 1.7877540106951874e-05,
+      "loss": 0.6499,
+      "mean_token_accuracy": 0.8101529002189636,
+      "num_tokens": 133365753.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6280888825654983,
+      "epoch": 0.1088,
+      "grad_norm": 0.5312396883964539,
+      "learning_rate": 1.7872192513368983e-05,
+      "loss": 0.6146,
+      "mean_token_accuracy": 0.8184811815619468,
+      "num_tokens": 133693433.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6363288000226021,
+      "epoch": 0.10906666666666667,
+      "grad_norm": 0.4588342607021332,
+      "learning_rate": 1.7866844919786098e-05,
+      "loss": 0.6273,
+      "mean_token_accuracy": 0.8149529561400414,
+      "num_tokens": 134021113.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6473316177725792,
+      "epoch": 0.10933333333333334,
+      "grad_norm": 0.4732758104801178,
+      "learning_rate": 1.786149732620321e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8144153237342835,
+      "num_tokens": 134348793.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6554132297635078,
+      "epoch": 0.1096,
+      "grad_norm": 0.5703415870666504,
+      "learning_rate": 1.785614973262032e-05,
+      "loss": 0.6442,
+      "mean_token_accuracy": 0.8124266847968101,
+      "num_tokens": 134676473.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6432550206780434,
+      "epoch": 0.10986666666666667,
+      "grad_norm": 0.5566362142562866,
+      "learning_rate": 1.7850802139037437e-05,
+      "loss": 0.6323,
+      "mean_token_accuracy": 0.8162054002285004,
+      "num_tokens": 135004153.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.5974282130599022,
+      "epoch": 0.11013333333333333,
+      "grad_norm": 0.49694034457206726,
+      "learning_rate": 1.7845454545454545e-05,
+      "loss": 0.5857,
+      "mean_token_accuracy": 0.8274865552783013,
+      "num_tokens": 135331833.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.663143989443779,
+      "epoch": 0.1104,
+      "grad_norm": 0.5567175149917603,
+      "learning_rate": 1.7840106951871657e-05,
+      "loss": 0.6579,
+      "mean_token_accuracy": 0.8079209432005883,
+      "num_tokens": 135659513.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6794201821088791,
+      "epoch": 0.11066666666666666,
+      "grad_norm": 0.5032562017440796,
+      "learning_rate": 1.7834759358288772e-05,
+      "loss": 0.6641,
+      "mean_token_accuracy": 0.8082783475518227,
+      "num_tokens": 135987193.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6614437021315098,
+      "epoch": 0.11093333333333333,
+      "grad_norm": 0.48676520586013794,
+      "learning_rate": 1.7829411764705884e-05,
+      "loss": 0.6506,
+      "mean_token_accuracy": 0.8112414509057999,
+      "num_tokens": 136314873.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6085740782320499,
+      "epoch": 0.1112,
+      "grad_norm": 0.492904931306839,
+      "learning_rate": 1.7824064171122996e-05,
+      "loss": 0.6027,
+      "mean_token_accuracy": 0.8242302000522613,
+      "num_tokens": 136642553.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6282576322555542,
+      "epoch": 0.11146666666666667,
+      "grad_norm": 0.5383780598640442,
+      "learning_rate": 1.781871657754011e-05,
+      "loss": 0.6213,
+      "mean_token_accuracy": 0.818753057718277,
+      "num_tokens": 136970233.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6601422399282455,
+      "epoch": 0.11173333333333334,
+      "grad_norm": 0.5213291645050049,
+      "learning_rate": 1.781336898395722e-05,
+      "loss": 0.6482,
+      "mean_token_accuracy": 0.811003178358078,
+      "num_tokens": 137297913.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.6488221094012261,
+      "epoch": 0.112,
+      "grad_norm": 0.6385365128517151,
+      "learning_rate": 1.780802139037433e-05,
+      "loss": 0.6424,
+      "mean_token_accuracy": 0.814769671857357,
+      "num_tokens": 137625593.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6658627107739449,
+      "epoch": 0.11226666666666667,
+      "grad_norm": 0.5553659200668335,
+      "learning_rate": 1.7802673796791446e-05,
+      "loss": 0.648,
+      "mean_token_accuracy": 0.8132178664207459,
+      "num_tokens": 137953273.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6293218336999417,
+      "epoch": 0.11253333333333333,
+      "grad_norm": 0.5905689597129822,
+      "learning_rate": 1.7797326203208558e-05,
+      "loss": 0.6203,
+      "mean_token_accuracy": 0.8189852133393287,
+      "num_tokens": 138280953.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6407099135220051,
+      "epoch": 0.1128,
+      "grad_norm": 0.5655410885810852,
+      "learning_rate": 1.779197860962567e-05,
+      "loss": 0.627,
+      "mean_token_accuracy": 0.8185942053794861,
+      "num_tokens": 138608633.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6537538126111031,
+      "epoch": 0.11306666666666666,
+      "grad_norm": 0.6247172951698303,
+      "learning_rate": 1.7786631016042782e-05,
+      "loss": 0.6459,
+      "mean_token_accuracy": 0.8115133136510849,
+      "num_tokens": 138936313.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6479903370141983,
+      "epoch": 0.11333333333333333,
+      "grad_norm": 0.4959455728530884,
+      "learning_rate": 1.7781283422459894e-05,
+      "loss": 0.639,
+      "mean_token_accuracy": 0.8130987301468849,
+      "num_tokens": 139263993.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6511910118162632,
+      "epoch": 0.1136,
+      "grad_norm": 0.5427896976470947,
+      "learning_rate": 1.7775935828877006e-05,
+      "loss": 0.6377,
+      "mean_token_accuracy": 0.8160068452358246,
+      "num_tokens": 139591673.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6399455040693283,
+      "epoch": 0.11386666666666667,
+      "grad_norm": 0.4341362714767456,
+      "learning_rate": 1.777058823529412e-05,
+      "loss": 0.6332,
+      "mean_token_accuracy": 0.8133431106805802,
+      "num_tokens": 139919353.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6267228953540325,
+      "epoch": 0.11413333333333334,
+      "grad_norm": 0.5249146819114685,
+      "learning_rate": 1.7765240641711233e-05,
+      "loss": 0.6139,
+      "mean_token_accuracy": 0.8193914964795113,
+      "num_tokens": 140247033.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6310237318277359,
+      "epoch": 0.1144,
+      "grad_norm": 0.483688622713089,
+      "learning_rate": 1.7759893048128344e-05,
+      "loss": 0.6216,
+      "mean_token_accuracy": 0.8173845320940017,
+      "num_tokens": 140574713.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.643636979162693,
+      "epoch": 0.11466666666666667,
+      "grad_norm": 0.5163200497627258,
+      "learning_rate": 1.7754545454545456e-05,
+      "loss": 0.636,
+      "mean_token_accuracy": 0.8143939390778542,
+      "num_tokens": 140902393.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.6258506819605827,
+      "epoch": 0.11493333333333333,
+      "grad_norm": 0.48647376894950867,
+      "learning_rate": 1.7749197860962568e-05,
+      "loss": 0.6146,
+      "mean_token_accuracy": 0.8206744849681854,
+      "num_tokens": 141230073.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6670894615352154,
+      "epoch": 0.1152,
+      "grad_norm": 0.5673144459724426,
+      "learning_rate": 1.774385026737968e-05,
+      "loss": 0.6567,
+      "mean_token_accuracy": 0.809934014081955,
+      "num_tokens": 141557753.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.627391516417265,
+      "epoch": 0.11546666666666666,
+      "grad_norm": 0.46794262528419495,
+      "learning_rate": 1.7738502673796795e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8202987447381019,
+      "num_tokens": 141885433.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6500027820467948,
+      "epoch": 0.11573333333333333,
+      "grad_norm": 0.49784207344055176,
+      "learning_rate": 1.7733155080213907e-05,
+      "loss": 0.6409,
+      "mean_token_accuracy": 0.8147238492965698,
+      "num_tokens": 142213113.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.6567294046282768,
+      "epoch": 0.116,
+      "grad_norm": 0.526779055595398,
+      "learning_rate": 1.7727807486631015e-05,
+      "loss": 0.6489,
+      "mean_token_accuracy": 0.813343109190464,
+      "num_tokens": 142540793.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.6548135101795196,
+      "epoch": 0.11626666666666667,
+      "grad_norm": 0.5357381105422974,
+      "learning_rate": 1.772245989304813e-05,
+      "loss": 0.6521,
+      "mean_token_accuracy": 0.8117088228464127,
+      "num_tokens": 142868473.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6462584212422371,
+      "epoch": 0.11653333333333334,
+      "grad_norm": 0.5073611736297607,
+      "learning_rate": 1.7717112299465242e-05,
+      "loss": 0.6407,
+      "mean_token_accuracy": 0.8138257563114166,
+      "num_tokens": 143196153.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6625854156911373,
+      "epoch": 0.1168,
+      "grad_norm": 0.499559611082077,
+      "learning_rate": 1.7711764705882354e-05,
+      "loss": 0.6461,
+      "mean_token_accuracy": 0.812066225707531,
+      "num_tokens": 143523833.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.5989975735545159,
+      "epoch": 0.11706666666666667,
+      "grad_norm": 0.45968884229660034,
+      "learning_rate": 1.770641711229947e-05,
+      "loss": 0.5916,
+      "mean_token_accuracy": 0.8241752177476883,
+      "num_tokens": 143851513.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6358760423958302,
+      "epoch": 0.11733333333333333,
+      "grad_norm": 0.5765702724456787,
+      "learning_rate": 1.7701069518716578e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.8180382415652275,
+      "num_tokens": 144179193.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.640472250431776,
+      "epoch": 0.1176,
+      "grad_norm": 0.5118552446365356,
+      "learning_rate": 1.769572192513369e-05,
+      "loss": 0.6351,
+      "mean_token_accuracy": 0.8165017113089561,
+      "num_tokens": 144506873.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6135229803621769,
+      "epoch": 0.11786666666666666,
+      "grad_norm": 0.5522367358207703,
+      "learning_rate": 1.7690374331550805e-05,
+      "loss": 0.599,
+      "mean_token_accuracy": 0.8233565494418145,
+      "num_tokens": 144834553.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.621340736746788,
+      "epoch": 0.11813333333333334,
+      "grad_norm": 0.5372185111045837,
+      "learning_rate": 1.7685026737967916e-05,
+      "loss": 0.614,
+      "mean_token_accuracy": 0.8201948925852776,
+      "num_tokens": 145162233.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6485022202134132,
+      "epoch": 0.1184,
+      "grad_norm": 0.5407975316047668,
+      "learning_rate": 1.7679679144385028e-05,
+      "loss": 0.6383,
+      "mean_token_accuracy": 0.8138899087905884,
+      "num_tokens": 145489913.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6315679028630257,
+      "epoch": 0.11866666666666667,
+      "grad_norm": 0.6210809946060181,
+      "learning_rate": 1.767433155080214e-05,
+      "loss": 0.625,
+      "mean_token_accuracy": 0.8164986595511436,
+      "num_tokens": 145817593.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.6173768222332001,
+      "epoch": 0.11893333333333334,
+      "grad_norm": 0.5039273500442505,
+      "learning_rate": 1.7668983957219252e-05,
+      "loss": 0.6027,
+      "mean_token_accuracy": 0.8232099175453186,
+      "num_tokens": 146145273.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.6252644754946232,
+      "epoch": 0.1192,
+      "grad_norm": 0.4843173623085022,
+      "learning_rate": 1.7663636363636364e-05,
+      "loss": 0.609,
+      "mean_token_accuracy": 0.8224162280559539,
+      "num_tokens": 146472938.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.635653717070818,
+      "epoch": 0.11946666666666667,
+      "grad_norm": 0.4938509166240692,
+      "learning_rate": 1.765828877005348e-05,
+      "loss": 0.6314,
+      "mean_token_accuracy": 0.8175220012664794,
+      "num_tokens": 146800618.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.6297695375978947,
+      "epoch": 0.11973333333333333,
+      "grad_norm": 0.56241774559021,
+      "learning_rate": 1.765294117647059e-05,
+      "loss": 0.6144,
+      "mean_token_accuracy": 0.822266012430191,
+      "num_tokens": 147128298.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.6400511264801025,
+      "epoch": 0.12,
+      "grad_norm": 0.5157172679901123,
+      "learning_rate": 1.7647593582887702e-05,
+      "loss": 0.6337,
+      "mean_token_accuracy": 0.8142900750041008,
+      "num_tokens": 147455978.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.624137969315052,
+      "epoch": 0.12026666666666666,
+      "grad_norm": 0.7271736860275269,
+      "learning_rate": 1.7642245989304814e-05,
+      "loss": 0.6096,
+      "mean_token_accuracy": 0.8229013904929161,
+      "num_tokens": 147783658.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.6445712596178055,
+      "epoch": 0.12053333333333334,
+      "grad_norm": 0.5607998967170715,
+      "learning_rate": 1.7636898395721926e-05,
+      "loss": 0.6412,
+      "mean_token_accuracy": 0.8144122689962388,
+      "num_tokens": 148111338.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.631790180504322,
+      "epoch": 0.1208,
+      "grad_norm": 0.5451073050498962,
+      "learning_rate": 1.7631550802139038e-05,
+      "loss": 0.6142,
+      "mean_token_accuracy": 0.8202407076954842,
+      "num_tokens": 148439018.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.6397012956440449,
+      "epoch": 0.12106666666666667,
+      "grad_norm": 0.47844812273979187,
+      "learning_rate": 1.7626203208556153e-05,
+      "loss": 0.6322,
+      "mean_token_accuracy": 0.8165719702839851,
+      "num_tokens": 148766698.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.6353954397141933,
+      "epoch": 0.12133333333333333,
+      "grad_norm": 0.5860951542854309,
+      "learning_rate": 1.7620855614973265e-05,
+      "loss": 0.6225,
+      "mean_token_accuracy": 0.8201490685343742,
+      "num_tokens": 149094378.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.6271689295768738,
+      "epoch": 0.1216,
+      "grad_norm": 0.45326417684555054,
+      "learning_rate": 1.7615508021390377e-05,
+      "loss": 0.6189,
+      "mean_token_accuracy": 0.8166147366166114,
+      "num_tokens": 149422058.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.6260932311415672,
+      "epoch": 0.12186666666666666,
+      "grad_norm": 0.6371480226516724,
+      "learning_rate": 1.761016042780749e-05,
+      "loss": 0.6134,
+      "mean_token_accuracy": 0.8211388066411018,
+      "num_tokens": 149749738.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.6056594341993332,
+      "epoch": 0.12213333333333333,
+      "grad_norm": 0.48230496048927307,
+      "learning_rate": 1.76048128342246e-05,
+      "loss": 0.594,
+      "mean_token_accuracy": 0.8256689861416817,
+      "num_tokens": 150077418.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.6401201985776425,
+      "epoch": 0.1224,
+      "grad_norm": 0.6073299050331116,
+      "learning_rate": 1.7599465240641712e-05,
+      "loss": 0.6263,
+      "mean_token_accuracy": 0.816895766556263,
+      "num_tokens": 150405098.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.6221492014825344,
+      "epoch": 0.12266666666666666,
+      "grad_norm": 0.5682583451271057,
+      "learning_rate": 1.7594117647058827e-05,
+      "loss": 0.6155,
+      "mean_token_accuracy": 0.8207264199852944,
+      "num_tokens": 150732778.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.6160124816000462,
+      "epoch": 0.12293333333333334,
+      "grad_norm": 0.5284896492958069,
+      "learning_rate": 1.7588770053475936e-05,
+      "loss": 0.6061,
+      "mean_token_accuracy": 0.8231152191758155,
+      "num_tokens": 151060458.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.6351871989667416,
+      "epoch": 0.1232,
+      "grad_norm": 0.48556894063949585,
+      "learning_rate": 1.7583422459893047e-05,
+      "loss": 0.624,
+      "mean_token_accuracy": 0.8175983607769013,
+      "num_tokens": 151388138.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.6134949989616871,
+      "epoch": 0.12346666666666667,
+      "grad_norm": 0.6327890753746033,
+      "learning_rate": 1.7578074866310163e-05,
+      "loss": 0.6051,
+      "mean_token_accuracy": 0.820197942852974,
+      "num_tokens": 151715818.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.6176735311746597,
+      "epoch": 0.12373333333333333,
+      "grad_norm": 0.5667627453804016,
+      "learning_rate": 1.7572727272727275e-05,
+      "loss": 0.603,
+      "mean_token_accuracy": 0.8223118305206298,
+      "num_tokens": 152043498.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.6148377418518066,
+      "epoch": 0.124,
+      "grad_norm": 0.554008424282074,
+      "learning_rate": 1.7567379679144386e-05,
+      "loss": 0.6077,
+      "mean_token_accuracy": 0.8207141950726509,
+      "num_tokens": 152371178.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.5976212918758392,
+      "epoch": 0.12426666666666666,
+      "grad_norm": 0.5570823550224304,
+      "learning_rate": 1.7562032085561498e-05,
+      "loss": 0.5909,
+      "mean_token_accuracy": 0.8272207915782929,
+      "num_tokens": 152698858.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.6408797353506088,
+      "epoch": 0.12453333333333333,
+      "grad_norm": 0.5833103656768799,
+      "learning_rate": 1.755668449197861e-05,
+      "loss": 0.6317,
+      "mean_token_accuracy": 0.8147543981671334,
+      "num_tokens": 153026538.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.6273569658398628,
+      "epoch": 0.1248,
+      "grad_norm": 0.490336537361145,
+      "learning_rate": 1.7551336898395722e-05,
+      "loss": 0.6138,
+      "mean_token_accuracy": 0.8204973101615906,
+      "num_tokens": 153354218.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.6496482968330384,
+      "epoch": 0.12506666666666666,
+      "grad_norm": 0.7385754585266113,
+      "learning_rate": 1.7545989304812837e-05,
+      "loss": 0.6443,
+      "mean_token_accuracy": 0.8140395879745483,
+      "num_tokens": 153681898.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.6157389342784881,
+      "epoch": 0.12533333333333332,
+      "grad_norm": 0.57555091381073,
+      "learning_rate": 1.754064171122995e-05,
+      "loss": 0.601,
+      "mean_token_accuracy": 0.8214931607246398,
+      "num_tokens": 154009578.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.6385625988245011,
+      "epoch": 0.1256,
+      "grad_norm": 0.4716522693634033,
+      "learning_rate": 1.753529411764706e-05,
+      "loss": 0.6288,
+      "mean_token_accuracy": 0.8169904679059983,
+      "num_tokens": 154337258.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.6342476040124894,
+      "epoch": 0.12586666666666665,
+      "grad_norm": 0.5770987868309021,
+      "learning_rate": 1.7529946524064172e-05,
+      "loss": 0.6272,
+      "mean_token_accuracy": 0.8163306444883347,
+      "num_tokens": 154664938.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.6355664700269699,
+      "epoch": 0.12613333333333332,
+      "grad_norm": 0.4963732063770294,
+      "learning_rate": 1.7524598930481284e-05,
+      "loss": 0.6232,
+      "mean_token_accuracy": 0.8177999764680862,
+      "num_tokens": 154992618.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.6333813920617104,
+      "epoch": 0.1264,
+      "grad_norm": 0.5953047275543213,
+      "learning_rate": 1.7519251336898396e-05,
+      "loss": 0.6261,
+      "mean_token_accuracy": 0.8153317391872406,
+      "num_tokens": 155320298.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.6505439803004265,
+      "epoch": 0.12666666666666668,
+      "grad_norm": 0.4259377419948578,
+      "learning_rate": 1.751390374331551e-05,
+      "loss": 0.6409,
+      "mean_token_accuracy": 0.8139571115374565,
+      "num_tokens": 155647978.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.6410402148962021,
+      "epoch": 0.12693333333333334,
+      "grad_norm": 0.5407733917236328,
+      "learning_rate": 1.7508556149732623e-05,
+      "loss": 0.6286,
+      "mean_token_accuracy": 0.8189760476350785,
+      "num_tokens": 155975658.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.6278519466519356,
+      "epoch": 0.1272,
+      "grad_norm": 0.49069494009017944,
+      "learning_rate": 1.7503208556149735e-05,
+      "loss": 0.6223,
+      "mean_token_accuracy": 0.81726233959198,
+      "num_tokens": 156303338.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.6350674785673618,
+      "epoch": 0.12746666666666667,
+      "grad_norm": 0.5119713544845581,
+      "learning_rate": 1.7497860962566847e-05,
+      "loss": 0.6237,
+      "mean_token_accuracy": 0.8173631474375724,
+      "num_tokens": 156631018.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.6261625736951828,
+      "epoch": 0.12773333333333334,
+      "grad_norm": 0.6043604612350464,
+      "learning_rate": 1.749251336898396e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.819996340572834,
+      "num_tokens": 156958698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.6459453977644444,
+      "epoch": 0.128,
+      "grad_norm": 0.5381135940551758,
+      "learning_rate": 1.748716577540107e-05,
+      "loss": 0.6322,
+      "mean_token_accuracy": 0.8156738758087159,
+      "num_tokens": 157286378.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.6095151767134667,
+      "epoch": 0.12826666666666667,
+      "grad_norm": 0.5296955108642578,
+      "learning_rate": 1.7481818181818185e-05,
+      "loss": 0.6051,
+      "mean_token_accuracy": 0.8204759240150452,
+      "num_tokens": 157614058.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.6488602444529533,
+      "epoch": 0.12853333333333333,
+      "grad_norm": 0.5272141695022583,
+      "learning_rate": 1.7476470588235294e-05,
+      "loss": 0.6362,
+      "mean_token_accuracy": 0.8158266142010688,
+      "num_tokens": 157941738.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.6514977879822255,
+      "epoch": 0.1288,
+      "grad_norm": 0.6142001748085022,
+      "learning_rate": 1.747112299465241e-05,
+      "loss": 0.6417,
+      "mean_token_accuracy": 0.8125916421413422,
+      "num_tokens": 158269418.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.6048041738569736,
+      "epoch": 0.12906666666666666,
+      "grad_norm": 0.5288782715797424,
+      "learning_rate": 1.746577540106952e-05,
+      "loss": 0.5903,
+      "mean_token_accuracy": 0.8266465038061142,
+      "num_tokens": 158597098.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.599172817915678,
+      "epoch": 0.12933333333333333,
+      "grad_norm": 0.5724526643753052,
+      "learning_rate": 1.7460427807486633e-05,
+      "loss": 0.5942,
+      "mean_token_accuracy": 0.824966399371624,
+      "num_tokens": 158924778.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.6329713217914105,
+      "epoch": 0.1296,
+      "grad_norm": 0.533990740776062,
+      "learning_rate": 1.7455080213903744e-05,
+      "loss": 0.6144,
+      "mean_token_accuracy": 0.8196603089571,
+      "num_tokens": 159252458.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.6240798488259316,
+      "epoch": 0.12986666666666666,
+      "grad_norm": 0.4982832372188568,
+      "learning_rate": 1.7449732620320856e-05,
+      "loss": 0.6191,
+      "mean_token_accuracy": 0.8164528369903564,
+      "num_tokens": 159580138.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.6327751204371452,
+      "epoch": 0.13013333333333332,
+      "grad_norm": 0.574155867099762,
+      "learning_rate": 1.7444385026737968e-05,
+      "loss": 0.6161,
+      "mean_token_accuracy": 0.8202193319797516,
+      "num_tokens": 159907818.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.6164940394461155,
+      "epoch": 0.1304,
+      "grad_norm": 0.5245264172554016,
+      "learning_rate": 1.7439037433155083e-05,
+      "loss": 0.6082,
+      "mean_token_accuracy": 0.8210074499249458,
+      "num_tokens": 160235498.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.603496815264225,
+      "epoch": 0.13066666666666665,
+      "grad_norm": 0.4578472971916199,
+      "learning_rate": 1.7433689839572195e-05,
+      "loss": 0.5964,
+      "mean_token_accuracy": 0.8256231665611267,
+      "num_tokens": 160563178.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.6212443381547927,
+      "epoch": 0.13093333333333335,
+      "grad_norm": 0.5583697557449341,
+      "learning_rate": 1.7428342245989307e-05,
+      "loss": 0.6048,
+      "mean_token_accuracy": 0.8217803001403808,
+      "num_tokens": 160890858.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.6337346948683262,
+      "epoch": 0.1312,
+      "grad_norm": 0.5270321369171143,
+      "learning_rate": 1.742299465240642e-05,
+      "loss": 0.6259,
+      "mean_token_accuracy": 0.8176716729998589,
+      "num_tokens": 161218538.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.6687727048993111,
+      "epoch": 0.13146666666666668,
+      "grad_norm": 0.5283020734786987,
+      "learning_rate": 1.741764705882353e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.8100226044654846,
+      "num_tokens": 161546218.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.6009791202843189,
+      "epoch": 0.13173333333333334,
+      "grad_norm": 0.5109070539474487,
+      "learning_rate": 1.7412299465240642e-05,
+      "loss": 0.5955,
+      "mean_token_accuracy": 0.8242882415652275,
+      "num_tokens": 161873898.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.6141447208821773,
+      "epoch": 0.132,
+      "grad_norm": 0.6980870366096497,
+      "learning_rate": 1.7406951871657754e-05,
+      "loss": 0.6027,
+      "mean_token_accuracy": 0.8223698660731316,
+      "num_tokens": 162201578.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.6143863417208195,
+      "epoch": 0.13226666666666667,
+      "grad_norm": 0.5059211254119873,
+      "learning_rate": 1.740160427807487e-05,
+      "loss": 0.598,
+      "mean_token_accuracy": 0.823991933465004,
+      "num_tokens": 162529258.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.6049226514995099,
+      "epoch": 0.13253333333333334,
+      "grad_norm": 0.5974682569503784,
+      "learning_rate": 1.739625668449198e-05,
+      "loss": 0.6003,
+      "mean_token_accuracy": 0.8240163698792458,
+      "num_tokens": 162856938.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.6193537250161171,
+      "epoch": 0.1328,
+      "grad_norm": 0.45828455686569214,
+      "learning_rate": 1.7390909090909093e-05,
+      "loss": 0.6063,
+      "mean_token_accuracy": 0.8218566745519638,
+      "num_tokens": 163184618.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.6194451808929443,
+      "epoch": 0.13306666666666667,
+      "grad_norm": 0.458832323551178,
+      "learning_rate": 1.7385561497326205e-05,
+      "loss": 0.6093,
+      "mean_token_accuracy": 0.8211968436837196,
+      "num_tokens": 163512298.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.6164743147790432,
+      "epoch": 0.13333333333333333,
+      "grad_norm": 0.5728282332420349,
+      "learning_rate": 1.7380213903743316e-05,
+      "loss": 0.6089,
+      "mean_token_accuracy": 0.8214412242174148,
+      "num_tokens": 163839978.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.6621682949364185,
+      "epoch": 0.1336,
+      "grad_norm": 0.5097026228904724,
+      "learning_rate": 1.737486631016043e-05,
+      "loss": 0.6494,
+      "mean_token_accuracy": 0.8132300823926926,
+      "num_tokens": 164167658.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.6411984667181969,
+      "epoch": 0.13386666666666666,
+      "grad_norm": 0.6675585508346558,
+      "learning_rate": 1.7369518716577544e-05,
+      "loss": 0.6302,
+      "mean_token_accuracy": 0.8165383696556091,
+      "num_tokens": 164495338.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.6356221534311771,
+      "epoch": 0.13413333333333333,
+      "grad_norm": 0.5140653848648071,
+      "learning_rate": 1.7364171122994652e-05,
+      "loss": 0.6289,
+      "mean_token_accuracy": 0.8160221159458161,
+      "num_tokens": 164823018.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.6374156191945076,
+      "epoch": 0.1344,
+      "grad_norm": 0.5850067138671875,
+      "learning_rate": 1.7358823529411767e-05,
+      "loss": 0.6252,
+      "mean_token_accuracy": 0.8176350191235542,
+      "num_tokens": 165150698.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.6058228969573974,
+      "epoch": 0.13466666666666666,
+      "grad_norm": 0.675095796585083,
+      "learning_rate": 1.735347593582888e-05,
+      "loss": 0.5965,
+      "mean_token_accuracy": 0.8241843849420547,
+      "num_tokens": 165478378.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.6202493324875832,
+      "epoch": 0.13493333333333332,
+      "grad_norm": 0.5985092520713806,
+      "learning_rate": 1.734812834224599e-05,
+      "loss": 0.6058,
+      "mean_token_accuracy": 0.8212945997714997,
+      "num_tokens": 165806058.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.6014047771692276,
+      "epoch": 0.1352,
+      "grad_norm": 0.5780919790267944,
+      "learning_rate": 1.7342780748663103e-05,
+      "loss": 0.5905,
+      "mean_token_accuracy": 0.825891986489296,
+      "num_tokens": 166133738.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.5863475486636162,
+      "epoch": 0.13546666666666668,
+      "grad_norm": 0.6052114963531494,
+      "learning_rate": 1.7337433155080214e-05,
+      "loss": 0.5769,
+      "mean_token_accuracy": 0.8284671291708946,
+      "num_tokens": 166461418.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.6248861335217952,
+      "epoch": 0.13573333333333334,
+      "grad_norm": 0.48946908116340637,
+      "learning_rate": 1.7332085561497326e-05,
+      "loss": 0.6143,
+      "mean_token_accuracy": 0.8221651986241341,
+      "num_tokens": 166789098.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.6324261531233788,
+      "epoch": 0.136,
+      "grad_norm": 0.6792661547660828,
+      "learning_rate": 1.732673796791444e-05,
+      "loss": 0.6237,
+      "mean_token_accuracy": 0.817295940220356,
+      "num_tokens": 167116778.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.6304112918674946,
+      "epoch": 0.13626666666666667,
+      "grad_norm": 0.5585044026374817,
+      "learning_rate": 1.7321390374331553e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82105633020401,
+      "num_tokens": 167444458.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.6555210158228875,
+      "epoch": 0.13653333333333334,
+      "grad_norm": 0.5223856568336487,
+      "learning_rate": 1.7316042780748665e-05,
+      "loss": 0.6443,
+      "mean_token_accuracy": 0.8133614391088486,
+      "num_tokens": 167772138.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.6287809893488884,
+      "epoch": 0.1368,
+      "grad_norm": 0.5142949223518372,
+      "learning_rate": 1.7310695187165777e-05,
+      "loss": 0.6201,
+      "mean_token_accuracy": 0.8164772674441337,
+      "num_tokens": 168099818.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.6223357163369656,
+      "epoch": 0.13706666666666667,
+      "grad_norm": 0.4758523404598236,
+      "learning_rate": 1.730534759358289e-05,
+      "loss": 0.6133,
+      "mean_token_accuracy": 0.8215023264288902,
+      "num_tokens": 168427498.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.6123070672154427,
+      "epoch": 0.13733333333333334,
+      "grad_norm": 0.5273539423942566,
+      "learning_rate": 1.73e-05,
+      "loss": 0.6011,
+      "mean_token_accuracy": 0.8222324073314666,
+      "num_tokens": 168755178.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.6138663627207279,
+      "epoch": 0.1376,
+      "grad_norm": 0.6459339261054993,
+      "learning_rate": 1.7294652406417116e-05,
+      "loss": 0.5999,
+      "mean_token_accuracy": 0.8235214978456498,
+      "num_tokens": 169082858.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.6212375231087208,
+      "epoch": 0.13786666666666667,
+      "grad_norm": 0.7556997537612915,
+      "learning_rate": 1.7289304812834227e-05,
+      "loss": 0.6081,
+      "mean_token_accuracy": 0.8220979988574981,
+      "num_tokens": 169410538.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.5888078711926937,
+      "epoch": 0.13813333333333333,
+      "grad_norm": 0.4729660749435425,
+      "learning_rate": 1.728395721925134e-05,
+      "loss": 0.579,
+      "mean_token_accuracy": 0.8274529546499252,
+      "num_tokens": 169738218.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.6192933194339275,
+      "epoch": 0.1384,
+      "grad_norm": 0.6300689578056335,
+      "learning_rate": 1.727860962566845e-05,
+      "loss": 0.6123,
+      "mean_token_accuracy": 0.8203018084168434,
+      "num_tokens": 170065898.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.6196565598249435,
+      "epoch": 0.13866666666666666,
+      "grad_norm": 0.6588465571403503,
+      "learning_rate": 1.7273262032085563e-05,
+      "loss": 0.6061,
+      "mean_token_accuracy": 0.822409576177597,
+      "num_tokens": 170393578.0,
+      "step": 5200
+    },
+    {
+      "entropy": 0.6234577037394047,
+      "epoch": 0.13893333333333333,
+      "grad_norm": 0.5834973454475403,
+      "learning_rate": 1.7267914438502675e-05,
+      "loss": 0.6093,
+      "mean_token_accuracy": 0.8198069378733635,
+      "num_tokens": 170721258.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.6077614583075046,
+      "epoch": 0.1392,
+      "grad_norm": 0.4954999089241028,
+      "learning_rate": 1.7262566844919786e-05,
+      "loss": 0.6001,
+      "mean_token_accuracy": 0.8226722851395607,
+      "num_tokens": 171048938.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.613494723290205,
+      "epoch": 0.13946666666666666,
+      "grad_norm": 0.557677149772644,
+      "learning_rate": 1.72572192513369e-05,
+      "loss": 0.5975,
+      "mean_token_accuracy": 0.8248533740639686,
+      "num_tokens": 171376618.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.607421388477087,
+      "epoch": 0.13973333333333332,
+      "grad_norm": 0.5784938931465149,
+      "learning_rate": 1.725187165775401e-05,
+      "loss": 0.5972,
+      "mean_token_accuracy": 0.8232129752635956,
+      "num_tokens": 171704298.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.6144205823540687,
+      "epoch": 0.14,
+      "grad_norm": 0.5463047027587891,
+      "learning_rate": 1.7246524064171125e-05,
+      "loss": 0.603,
+      "mean_token_accuracy": 0.8228097483515739,
+      "num_tokens": 172031978.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.5937097623944283,
+      "epoch": 0.14026666666666668,
+      "grad_norm": 0.4457780420780182,
+      "learning_rate": 1.7241176470588237e-05,
+      "loss": 0.5863,
+      "mean_token_accuracy": 0.8276026368141174,
+      "num_tokens": 172359658.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.6227729670703411,
+      "epoch": 0.14053333333333334,
+      "grad_norm": 0.5007463097572327,
+      "learning_rate": 1.723582887700535e-05,
+      "loss": 0.611,
+      "mean_token_accuracy": 0.8213984563946723,
+      "num_tokens": 172687338.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.6367640405893326,
+      "epoch": 0.1408,
+      "grad_norm": 0.6137294173240662,
+      "learning_rate": 1.723048128342246e-05,
+      "loss": 0.6264,
+      "mean_token_accuracy": 0.8166116803884507,
+      "num_tokens": 173015018.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.5944077350199223,
+      "epoch": 0.14106666666666667,
+      "grad_norm": 0.4923824071884155,
+      "learning_rate": 1.7225133689839572e-05,
+      "loss": 0.5814,
+      "mean_token_accuracy": 0.8262463301420212,
+      "num_tokens": 173342698.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.6210280202329159,
+      "epoch": 0.14133333333333334,
+      "grad_norm": 0.5874246954917908,
+      "learning_rate": 1.7219786096256684e-05,
+      "loss": 0.6197,
+      "mean_token_accuracy": 0.8178060814738274,
+      "num_tokens": 173670378.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.6161349073052407,
+      "epoch": 0.1416,
+      "grad_norm": 0.542860746383667,
+      "learning_rate": 1.72144385026738e-05,
+      "loss": 0.594,
+      "mean_token_accuracy": 0.824835042655468,
+      "num_tokens": 173998058.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.60669986307621,
+      "epoch": 0.14186666666666667,
+      "grad_norm": 0.5829471945762634,
+      "learning_rate": 1.720909090909091e-05,
+      "loss": 0.604,
+      "mean_token_accuracy": 0.8207141935825348,
+      "num_tokens": 174325738.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.5983838602900505,
+      "epoch": 0.14213333333333333,
+      "grad_norm": 0.4586426913738251,
+      "learning_rate": 1.7203743315508023e-05,
+      "loss": 0.5854,
+      "mean_token_accuracy": 0.8274132445454597,
+      "num_tokens": 174653418.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.612214021384716,
+      "epoch": 0.1424,
+      "grad_norm": 0.6354243159294128,
+      "learning_rate": 1.7198395721925135e-05,
+      "loss": 0.6003,
+      "mean_token_accuracy": 0.825381837785244,
+      "num_tokens": 174981098.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.5991831764578819,
+      "epoch": 0.14266666666666666,
+      "grad_norm": 0.5400723814964294,
+      "learning_rate": 1.7193048128342247e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8272430807352066,
+      "num_tokens": 175308697.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.6009857349097729,
+      "epoch": 0.14293333333333333,
+      "grad_norm": 0.46825549006462097,
+      "learning_rate": 1.718770053475936e-05,
+      "loss": 0.5897,
+      "mean_token_accuracy": 0.8261974558234215,
+      "num_tokens": 175636377.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.6579540863633155,
+      "epoch": 0.1432,
+      "grad_norm": 0.5498165488243103,
+      "learning_rate": 1.7182352941176474e-05,
+      "loss": 0.6388,
+      "mean_token_accuracy": 0.8137463361024857,
+      "num_tokens": 175964057.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.6127194188535213,
+      "epoch": 0.14346666666666666,
+      "grad_norm": 0.66085284948349,
+      "learning_rate": 1.7177005347593585e-05,
+      "loss": 0.6048,
+      "mean_token_accuracy": 0.8231640979647636,
+      "num_tokens": 176291737.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.6140283323824406,
+      "epoch": 0.14373333333333332,
+      "grad_norm": 0.5286858677864075,
+      "learning_rate": 1.7171657754010697e-05,
+      "loss": 0.6008,
+      "mean_token_accuracy": 0.8218475058674812,
+      "num_tokens": 176619417.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.6195843808352948,
+      "epoch": 0.144,
+      "grad_norm": 0.4871726334095001,
+      "learning_rate": 1.716631016042781e-05,
+      "loss": 0.6077,
+      "mean_token_accuracy": 0.8201918363571167,
+      "num_tokens": 176947097.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.6193856991827488,
+      "epoch": 0.14426666666666665,
+      "grad_norm": 0.6905038356781006,
+      "learning_rate": 1.716096256684492e-05,
+      "loss": 0.6137,
+      "mean_token_accuracy": 0.8208119466900825,
+      "num_tokens": 177274777.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.6134863115847111,
+      "epoch": 0.14453333333333335,
+      "grad_norm": 0.4986840486526489,
+      "learning_rate": 1.7155614973262033e-05,
+      "loss": 0.5944,
+      "mean_token_accuracy": 0.8243615612387657,
+      "num_tokens": 177602457.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.5782571002840996,
+      "epoch": 0.1448,
+      "grad_norm": 0.5752607583999634,
+      "learning_rate": 1.7150267379679148e-05,
+      "loss": 0.5705,
+      "mean_token_accuracy": 0.8306207194924354,
+      "num_tokens": 177930137.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.5844120234251022,
+      "epoch": 0.14506666666666668,
+      "grad_norm": 0.5871946811676025,
+      "learning_rate": 1.714491978609626e-05,
+      "loss": 0.5712,
+      "mean_token_accuracy": 0.8312072321772576,
+      "num_tokens": 178257817.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.6243437945842742,
+      "epoch": 0.14533333333333334,
+      "grad_norm": 0.5043569803237915,
+      "learning_rate": 1.7139572192513368e-05,
+      "loss": 0.616,
+      "mean_token_accuracy": 0.8187591671943665,
+      "num_tokens": 178585497.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.6056512035429478,
+      "epoch": 0.1456,
+      "grad_norm": 0.7347299456596375,
+      "learning_rate": 1.7134224598930483e-05,
+      "loss": 0.5927,
+      "mean_token_accuracy": 0.8241996571421624,
+      "num_tokens": 178913177.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.6096108578145504,
+      "epoch": 0.14586666666666667,
+      "grad_norm": 0.5748003721237183,
+      "learning_rate": 1.7128877005347595e-05,
+      "loss": 0.594,
+      "mean_token_accuracy": 0.8256689876317977,
+      "num_tokens": 179240857.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.622131047397852,
+      "epoch": 0.14613333333333334,
+      "grad_norm": 0.630882203578949,
+      "learning_rate": 1.7123529411764707e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8198802530765533,
+      "num_tokens": 179568537.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.6202254898846149,
+      "epoch": 0.1464,
+      "grad_norm": 0.6485390663146973,
+      "learning_rate": 1.711818181818182e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8199718996882439,
+      "num_tokens": 179896217.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.6123473346233368,
+      "epoch": 0.14666666666666667,
+      "grad_norm": 0.5846719145774841,
+      "learning_rate": 1.711283422459893e-05,
+      "loss": 0.5948,
+      "mean_token_accuracy": 0.8253879517316818,
+      "num_tokens": 180223897.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.5946030661463737,
+      "epoch": 0.14693333333333333,
+      "grad_norm": 0.6383978128433228,
+      "learning_rate": 1.7107486631016042e-05,
+      "loss": 0.5918,
+      "mean_token_accuracy": 0.8256231665611267,
+      "num_tokens": 180551577.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.5912628397345543,
+      "epoch": 0.1472,
+      "grad_norm": 0.6081676483154297,
+      "learning_rate": 1.7102139037433158e-05,
+      "loss": 0.5806,
+      "mean_token_accuracy": 0.8290689155459404,
+      "num_tokens": 180879257.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.6031342394649982,
+      "epoch": 0.14746666666666666,
+      "grad_norm": 0.5201841592788696,
+      "learning_rate": 1.709679144385027e-05,
+      "loss": 0.589,
+      "mean_token_accuracy": 0.8267473116517067,
+      "num_tokens": 181206937.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.6211850836873054,
+      "epoch": 0.14773333333333333,
+      "grad_norm": 0.5524062514305115,
+      "learning_rate": 1.709144385026738e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8208669364452362,
+      "num_tokens": 181534617.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.587277277559042,
+      "epoch": 0.148,
+      "grad_norm": 0.5584370493888855,
+      "learning_rate": 1.7086096256684493e-05,
+      "loss": 0.5724,
+      "mean_token_accuracy": 0.8290414184331893,
+      "num_tokens": 181862297.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.5796420909464359,
+      "epoch": 0.14826666666666666,
+      "grad_norm": 0.5459908843040466,
+      "learning_rate": 1.7080748663101605e-05,
+      "loss": 0.5689,
+      "mean_token_accuracy": 0.8323955282568931,
+      "num_tokens": 182189977.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.5950359866023064,
+      "epoch": 0.14853333333333332,
+      "grad_norm": 0.7116152048110962,
+      "learning_rate": 1.7075401069518717e-05,
+      "loss": 0.5847,
+      "mean_token_accuracy": 0.8268389493227005,
+      "num_tokens": 182517657.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.6044580794870853,
+      "epoch": 0.1488,
+      "grad_norm": 0.5448054075241089,
+      "learning_rate": 1.7070053475935832e-05,
+      "loss": 0.5895,
+      "mean_token_accuracy": 0.8255620688199997,
+      "num_tokens": 182845337.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.5867747254669666,
+      "epoch": 0.14906666666666665,
+      "grad_norm": 0.5431827902793884,
+      "learning_rate": 1.7064705882352944e-05,
+      "loss": 0.5819,
+      "mean_token_accuracy": 0.8266770482063294,
+      "num_tokens": 183173017.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.6118120543658734,
+      "epoch": 0.14933333333333335,
+      "grad_norm": 0.5354688763618469,
+      "learning_rate": 1.7059358288770055e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8242485344409942,
+      "num_tokens": 183500697.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.588444821536541,
+      "epoch": 0.1496,
+      "grad_norm": 0.5365312099456787,
+      "learning_rate": 1.7054010695187167e-05,
+      "loss": 0.5796,
+      "mean_token_accuracy": 0.8284762904047966,
+      "num_tokens": 183828377.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.6157905429601669,
+      "epoch": 0.14986666666666668,
+      "grad_norm": 0.5922267436981201,
+      "learning_rate": 1.704866310160428e-05,
+      "loss": 0.5987,
+      "mean_token_accuracy": 0.8235215038061142,
+      "num_tokens": 184156057.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.6106136351823807,
+      "epoch": 0.15013333333333334,
+      "grad_norm": 0.4556030035018921,
+      "learning_rate": 1.704331550802139e-05,
+      "loss": 0.6013,
+      "mean_token_accuracy": 0.8222324013710022,
+      "num_tokens": 184483737.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.5784977562725544,
+      "epoch": 0.1504,
+      "grad_norm": 0.4618242084980011,
+      "learning_rate": 1.7037967914438506e-05,
+      "loss": 0.5708,
+      "mean_token_accuracy": 0.8316379517316819,
+      "num_tokens": 184811417.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.6016978219151496,
+      "epoch": 0.15066666666666667,
+      "grad_norm": 0.5729866027832031,
+      "learning_rate": 1.7032620320855618e-05,
+      "loss": 0.5959,
+      "mean_token_accuracy": 0.8241416156291962,
+      "num_tokens": 185139097.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.6180305503308773,
+      "epoch": 0.15093333333333334,
+      "grad_norm": 0.4492015540599823,
+      "learning_rate": 1.7027272727272726e-05,
+      "loss": 0.6123,
+      "mean_token_accuracy": 0.8206867098808288,
+      "num_tokens": 185466777.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.6110047794878483,
+      "epoch": 0.1512,
+      "grad_norm": 0.6685405373573303,
+      "learning_rate": 1.702192513368984e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8238758549094201,
+      "num_tokens": 185794457.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.6041754990816116,
+      "epoch": 0.15146666666666667,
+      "grad_norm": 0.4714350998401642,
+      "learning_rate": 1.7016577540106953e-05,
+      "loss": 0.5923,
+      "mean_token_accuracy": 0.822953325510025,
+      "num_tokens": 186122137.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.6022451117634773,
+      "epoch": 0.15173333333333333,
+      "grad_norm": 0.4793015420436859,
+      "learning_rate": 1.7011229946524065e-05,
+      "loss": 0.5933,
+      "mean_token_accuracy": 0.8254673719406128,
+      "num_tokens": 186449817.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.5991009667515754,
+      "epoch": 0.152,
+      "grad_norm": 0.5819358825683594,
+      "learning_rate": 1.700588235294118e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8254643216729164,
+      "num_tokens": 186777497.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.6146255634725094,
+      "epoch": 0.15226666666666666,
+      "grad_norm": 0.5260542035102844,
+      "learning_rate": 1.700053475935829e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.8216153457760811,
+      "num_tokens": 187105177.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.5842849433422088,
+      "epoch": 0.15253333333333333,
+      "grad_norm": 0.47430717945098877,
+      "learning_rate": 1.69951871657754e-05,
+      "loss": 0.5729,
+      "mean_token_accuracy": 0.8301991671323776,
+      "num_tokens": 187432857.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.5866012826561928,
+      "epoch": 0.1528,
+      "grad_norm": 0.47670090198516846,
+      "learning_rate": 1.6989839572192516e-05,
+      "loss": 0.5807,
+      "mean_token_accuracy": 0.8269122675061226,
+      "num_tokens": 187760537.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.6161247670650483,
+      "epoch": 0.15306666666666666,
+      "grad_norm": 0.5166377425193787,
+      "learning_rate": 1.6984491978609627e-05,
+      "loss": 0.6058,
+      "mean_token_accuracy": 0.8223240450024605,
+      "num_tokens": 188088217.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.6182661049067975,
+      "epoch": 0.15333333333333332,
+      "grad_norm": 0.5446411967277527,
+      "learning_rate": 1.697914438502674e-05,
+      "loss": 0.6089,
+      "mean_token_accuracy": 0.821279326081276,
+      "num_tokens": 188415897.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.5995870277285575,
+      "epoch": 0.1536,
+      "grad_norm": 0.6212722063064575,
+      "learning_rate": 1.697379679144385e-05,
+      "loss": 0.5864,
+      "mean_token_accuracy": 0.825137460231781,
+      "num_tokens": 188743577.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.60131761059165,
+      "epoch": 0.15386666666666668,
+      "grad_norm": 0.5571957230567932,
+      "learning_rate": 1.6968449197860963e-05,
+      "loss": 0.5914,
+      "mean_token_accuracy": 0.8264418333768845,
+      "num_tokens": 189071257.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.5689705647528172,
+      "epoch": 0.15413333333333334,
+      "grad_norm": 0.5237152576446533,
+      "learning_rate": 1.6963101604278075e-05,
+      "loss": 0.5627,
+      "mean_token_accuracy": 0.8333852604031563,
+      "num_tokens": 189398937.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.5922715999186039,
+      "epoch": 0.1544,
+      "grad_norm": 0.5055749416351318,
+      "learning_rate": 1.695775401069519e-05,
+      "loss": 0.581,
+      "mean_token_accuracy": 0.8264204517006875,
+      "num_tokens": 189726617.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.5955583088099956,
+      "epoch": 0.15466666666666667,
+      "grad_norm": 0.5033058524131775,
+      "learning_rate": 1.6952406417112302e-05,
+      "loss": 0.5791,
+      "mean_token_accuracy": 0.8278653472661972,
+      "num_tokens": 190054297.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.5790828920900821,
+      "epoch": 0.15493333333333334,
+      "grad_norm": 0.5274530649185181,
+      "learning_rate": 1.6947058823529414e-05,
+      "loss": 0.5712,
+      "mean_token_accuracy": 0.8309048131108284,
+      "num_tokens": 190381977.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.5889088690280915,
+      "epoch": 0.1552,
+      "grad_norm": 0.6046780347824097,
+      "learning_rate": 1.6941711229946525e-05,
+      "loss": 0.5774,
+      "mean_token_accuracy": 0.8278653442859649,
+      "num_tokens": 190709657.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.6081131473183632,
+      "epoch": 0.15546666666666667,
+      "grad_norm": 0.5690280795097351,
+      "learning_rate": 1.6936363636363637e-05,
+      "loss": 0.5977,
+      "mean_token_accuracy": 0.8226264640688896,
+      "num_tokens": 191037337.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.587796650081873,
+      "epoch": 0.15573333333333333,
+      "grad_norm": 0.5349929928779602,
+      "learning_rate": 1.693101604278075e-05,
+      "loss": 0.5784,
+      "mean_token_accuracy": 0.8295485109090805,
+      "num_tokens": 191365017.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.5915711678564548,
+      "epoch": 0.156,
+      "grad_norm": 0.4947117567062378,
+      "learning_rate": 1.6925668449197864e-05,
+      "loss": 0.5821,
+      "mean_token_accuracy": 0.8268175661563874,
+      "num_tokens": 191692697.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.6039751805365086,
+      "epoch": 0.15626666666666666,
+      "grad_norm": 0.5140764713287354,
+      "learning_rate": 1.6920320855614976e-05,
+      "loss": 0.5892,
+      "mean_token_accuracy": 0.8269794672727585,
+      "num_tokens": 192020377.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.6018185123801232,
+      "epoch": 0.15653333333333333,
+      "grad_norm": 0.5119272470474243,
+      "learning_rate": 1.6914973262032088e-05,
+      "loss": 0.5919,
+      "mean_token_accuracy": 0.8238697439432144,
+      "num_tokens": 192348057.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.5990505233407021,
+      "epoch": 0.1568,
+      "grad_norm": 0.5075955390930176,
+      "learning_rate": 1.69096256684492e-05,
+      "loss": 0.5929,
+      "mean_token_accuracy": 0.8244134932756424,
+      "num_tokens": 192675737.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.5888317279517651,
+      "epoch": 0.15706666666666666,
+      "grad_norm": 0.5657641291618347,
+      "learning_rate": 1.690427807486631e-05,
+      "loss": 0.5795,
+      "mean_token_accuracy": 0.8287054032087326,
+      "num_tokens": 193003417.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.6212611272931099,
+      "epoch": 0.15733333333333333,
+      "grad_norm": 0.5563691854476929,
+      "learning_rate": 1.6898930481283423e-05,
+      "loss": 0.6099,
+      "mean_token_accuracy": 0.8195136845111847,
+      "num_tokens": 193331097.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.6000641994178295,
+      "epoch": 0.1576,
+      "grad_norm": 0.5165938138961792,
+      "learning_rate": 1.689358288770054e-05,
+      "loss": 0.5856,
+      "mean_token_accuracy": 0.8289161771535873,
+      "num_tokens": 193658777.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.5825034789741039,
+      "epoch": 0.15786666666666666,
+      "grad_norm": 0.6044878959655762,
+      "learning_rate": 1.6888235294117647e-05,
+      "loss": 0.5738,
+      "mean_token_accuracy": 0.8299639493227005,
+      "num_tokens": 193986457.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.5889501094818115,
+      "epoch": 0.15813333333333332,
+      "grad_norm": 0.5247175097465515,
+      "learning_rate": 1.688288770053476e-05,
+      "loss": 0.5741,
+      "mean_token_accuracy": 0.8290872409939766,
+      "num_tokens": 194314137.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.6192642949521542,
+      "epoch": 0.1584,
+      "grad_norm": 0.6957064270973206,
+      "learning_rate": 1.6877540106951874e-05,
+      "loss": 0.6086,
+      "mean_token_accuracy": 0.8228280797600747,
+      "num_tokens": 194641817.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.6061706312000752,
+      "epoch": 0.15866666666666668,
+      "grad_norm": 0.49194401502609253,
+      "learning_rate": 1.6872192513368986e-05,
+      "loss": 0.5948,
+      "mean_token_accuracy": 0.8240530312061309,
+      "num_tokens": 194969497.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.5888864673674107,
+      "epoch": 0.15893333333333334,
+      "grad_norm": 0.5204460620880127,
+      "learning_rate": 1.6866844919786097e-05,
+      "loss": 0.5815,
+      "mean_token_accuracy": 0.8260477721691132,
+      "num_tokens": 195297177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.5943687185645103,
+      "epoch": 0.1592,
+      "grad_norm": 0.47262024879455566,
+      "learning_rate": 1.686149732620321e-05,
+      "loss": 0.5889,
+      "mean_token_accuracy": 0.8256995290517807,
+      "num_tokens": 195624857.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.6128756478428841,
+      "epoch": 0.15946666666666667,
+      "grad_norm": 0.6539174318313599,
+      "learning_rate": 1.685614973262032e-05,
+      "loss": 0.5983,
+      "mean_token_accuracy": 0.8248655945062637,
+      "num_tokens": 195952537.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.5854204565286636,
+      "epoch": 0.15973333333333334,
+      "grad_norm": 0.5238087773323059,
+      "learning_rate": 1.6850802139037433e-05,
+      "loss": 0.5732,
+      "mean_token_accuracy": 0.8302266627550126,
+      "num_tokens": 196280217.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.5909271024167537,
+      "epoch": 0.16,
+      "grad_norm": 0.5339305996894836,
+      "learning_rate": 1.6845454545454548e-05,
+      "loss": 0.5786,
+      "mean_token_accuracy": 0.8292766332626342,
+      "num_tokens": 196607897.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.5587038807570934,
+      "epoch": 0.16026666666666667,
+      "grad_norm": 0.6590559482574463,
+      "learning_rate": 1.684010695187166e-05,
+      "loss": 0.5477,
+      "mean_token_accuracy": 0.8365469172596931,
+      "num_tokens": 196935577.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.590944005548954,
+      "epoch": 0.16053333333333333,
+      "grad_norm": 0.6243144273757935,
+      "learning_rate": 1.683475935828877e-05,
+      "loss": 0.578,
+      "mean_token_accuracy": 0.8304802045226097,
+      "num_tokens": 197263257.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.5954900644719601,
+      "epoch": 0.1608,
+      "grad_norm": 0.5066710114479065,
+      "learning_rate": 1.6829411764705883e-05,
+      "loss": 0.5884,
+      "mean_token_accuracy": 0.8247159078717232,
+      "num_tokens": 197590937.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.5989583984017373,
+      "epoch": 0.16106666666666666,
+      "grad_norm": 0.609043300151825,
+      "learning_rate": 1.6824064171122995e-05,
+      "loss": 0.5893,
+      "mean_token_accuracy": 0.8264357253909111,
+      "num_tokens": 197918617.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.6119338430464267,
+      "epoch": 0.16133333333333333,
+      "grad_norm": 0.6996363997459412,
+      "learning_rate": 1.6818716577540107e-05,
+      "loss": 0.5994,
+      "mean_token_accuracy": 0.8232832416892052,
+      "num_tokens": 198246297.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.5968971669673919,
+      "epoch": 0.1616,
+      "grad_norm": 0.5081717371940613,
+      "learning_rate": 1.6813368983957222e-05,
+      "loss": 0.5843,
+      "mean_token_accuracy": 0.8268084034323693,
+      "num_tokens": 198573977.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.5901890307664871,
+      "epoch": 0.16186666666666666,
+      "grad_norm": 0.5805912613868713,
+      "learning_rate": 1.6808021390374334e-05,
+      "loss": 0.5812,
+      "mean_token_accuracy": 0.8283541023731231,
+      "num_tokens": 198901657.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.5717126734554767,
+      "epoch": 0.16213333333333332,
+      "grad_norm": 0.45765310525894165,
+      "learning_rate": 1.6802673796791446e-05,
+      "loss": 0.5632,
+      "mean_token_accuracy": 0.8317723602056504,
+      "num_tokens": 199229337.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.5864593662321568,
+      "epoch": 0.1624,
+      "grad_norm": 0.55170077085495,
+      "learning_rate": 1.6797326203208558e-05,
+      "loss": 0.572,
+      "mean_token_accuracy": 0.8300281047821045,
+      "num_tokens": 199557017.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.5870142191648483,
+      "epoch": 0.16266666666666665,
+      "grad_norm": 0.5163315534591675,
+      "learning_rate": 1.679197860962567e-05,
+      "loss": 0.5779,
+      "mean_token_accuracy": 0.8300861448049546,
+      "num_tokens": 199884697.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.5794356741011143,
+      "epoch": 0.16293333333333335,
+      "grad_norm": 0.5115768313407898,
+      "learning_rate": 1.678663101604278e-05,
+      "loss": 0.5704,
+      "mean_token_accuracy": 0.8301533490419388,
+      "num_tokens": 200212377.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.5924508973956109,
+      "epoch": 0.1632,
+      "grad_norm": 0.4875156879425049,
+      "learning_rate": 1.6781283422459896e-05,
+      "loss": 0.5772,
+      "mean_token_accuracy": 0.8296279311180115,
+      "num_tokens": 200540057.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.5937600165605545,
+      "epoch": 0.16346666666666668,
+      "grad_norm": 0.502435028553009,
+      "learning_rate": 1.6775935828877005e-05,
+      "loss": 0.5843,
+      "mean_token_accuracy": 0.8271933004260064,
+      "num_tokens": 200867737.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.5833610489964485,
+      "epoch": 0.16373333333333334,
+      "grad_norm": 0.5039084553718567,
+      "learning_rate": 1.677058823529412e-05,
+      "loss": 0.5757,
+      "mean_token_accuracy": 0.8292399793863297,
+      "num_tokens": 201195417.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.5614327006042004,
+      "epoch": 0.164,
+      "grad_norm": 0.7296598553657532,
+      "learning_rate": 1.6765240641711232e-05,
+      "loss": 0.5494,
+      "mean_token_accuracy": 0.8348637595772743,
+      "num_tokens": 201523097.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.5644073359668255,
+      "epoch": 0.16426666666666667,
+      "grad_norm": 0.49619829654693604,
+      "learning_rate": 1.6759893048128344e-05,
+      "loss": 0.5521,
+      "mean_token_accuracy": 0.8348362669348717,
+      "num_tokens": 201850777.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.5702700860798359,
+      "epoch": 0.16453333333333334,
+      "grad_norm": 0.505085825920105,
+      "learning_rate": 1.6754545454545456e-05,
+      "loss": 0.5609,
+      "mean_token_accuracy": 0.8314424470067024,
+      "num_tokens": 202178457.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.5994322098791599,
+      "epoch": 0.1648,
+      "grad_norm": 0.5514662861824036,
+      "learning_rate": 1.6749197860962567e-05,
+      "loss": 0.5906,
+      "mean_token_accuracy": 0.8261974558234215,
+      "num_tokens": 202506137.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.6086098030209541,
+      "epoch": 0.16506666666666667,
+      "grad_norm": 0.48559609055519104,
+      "learning_rate": 1.674385026737968e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8223301500082016,
+      "num_tokens": 202833817.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.602690052986145,
+      "epoch": 0.16533333333333333,
+      "grad_norm": 0.5799172520637512,
+      "learning_rate": 1.673850267379679e-05,
+      "loss": 0.5925,
+      "mean_token_accuracy": 0.826533479988575,
+      "num_tokens": 203161497.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.611245246976614,
+      "epoch": 0.1656,
+      "grad_norm": 0.4895267188549042,
+      "learning_rate": 1.6733155080213906e-05,
+      "loss": 0.6012,
+      "mean_token_accuracy": 0.8238514184951782,
+      "num_tokens": 203489177.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.6056264981627464,
+      "epoch": 0.16586666666666666,
+      "grad_norm": 0.6508986949920654,
+      "learning_rate": 1.6727807486631018e-05,
+      "loss": 0.5971,
+      "mean_token_accuracy": 0.8228128015995025,
+      "num_tokens": 203816857.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.6069623030722141,
+      "epoch": 0.16613333333333333,
+      "grad_norm": 0.6548013091087341,
+      "learning_rate": 1.672245989304813e-05,
+      "loss": 0.5986,
+      "mean_token_accuracy": 0.8233718171715736,
+      "num_tokens": 204144537.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.5951053515076637,
+      "epoch": 0.1664,
+      "grad_norm": 0.4595625698566437,
+      "learning_rate": 1.671711229946524e-05,
+      "loss": 0.5801,
+      "mean_token_accuracy": 0.8285373896360397,
+      "num_tokens": 204472217.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.5886056572198868,
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.6029869914054871,
+      "learning_rate": 1.6711764705882353e-05,
+      "loss": 0.5854,
+      "mean_token_accuracy": 0.8261119231581688,
+      "num_tokens": 204799897.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.5991396136581898,
+      "epoch": 0.16693333333333332,
+      "grad_norm": 0.5357114672660828,
+      "learning_rate": 1.6706417112299465e-05,
+      "loss": 0.5838,
+      "mean_token_accuracy": 0.8271230399608612,
+      "num_tokens": 205127577.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.558577710390091,
+      "epoch": 0.1672,
+      "grad_norm": 0.4572122395038605,
+      "learning_rate": 1.670106951871658e-05,
+      "loss": 0.545,
+      "mean_token_accuracy": 0.8377565965056419,
+      "num_tokens": 205455257.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.5833405576646328,
+      "epoch": 0.16746666666666668,
+      "grad_norm": 0.5629472732543945,
+      "learning_rate": 1.6695721925133692e-05,
+      "loss": 0.5736,
+      "mean_token_accuracy": 0.829566840827465,
+      "num_tokens": 205782937.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.5613599091768264,
+      "epoch": 0.16773333333333335,
+      "grad_norm": 0.5343989729881287,
+      "learning_rate": 1.6690374331550804e-05,
+      "loss": 0.5518,
+      "mean_token_accuracy": 0.8349889993667603,
+      "num_tokens": 206110617.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.6083256095647812,
+      "epoch": 0.168,
+      "grad_norm": 0.5614194869995117,
+      "learning_rate": 1.6685026737967916e-05,
+      "loss": 0.6008,
+      "mean_token_accuracy": 0.8247342377901077,
+      "num_tokens": 206438297.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.602846211194992,
+      "epoch": 0.16826666666666668,
+      "grad_norm": 0.5676190853118896,
+      "learning_rate": 1.6679679144385028e-05,
+      "loss": 0.5859,
+      "mean_token_accuracy": 0.8254459947347641,
+      "num_tokens": 206765977.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.5713487163186073,
+      "epoch": 0.16853333333333334,
+      "grad_norm": 0.49047601222991943,
+      "learning_rate": 1.667433155080214e-05,
+      "loss": 0.5637,
+      "mean_token_accuracy": 0.8327193275094033,
+      "num_tokens": 207093657.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.5795545518398285,
+      "epoch": 0.1688,
+      "grad_norm": 0.5205618143081665,
+      "learning_rate": 1.6668983957219255e-05,
+      "loss": 0.5714,
+      "mean_token_accuracy": 0.8301594540476799,
+      "num_tokens": 207421337.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.5691454589366913,
+      "epoch": 0.16906666666666667,
+      "grad_norm": 0.5019081234931946,
+      "learning_rate": 1.6663636363636363e-05,
+      "loss": 0.5582,
+      "mean_token_accuracy": 0.8338709652423859,
+      "num_tokens": 207749017.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.5838815867900848,
+      "epoch": 0.16933333333333334,
+      "grad_norm": 0.5926812887191772,
+      "learning_rate": 1.6658288770053478e-05,
+      "loss": 0.5801,
+      "mean_token_accuracy": 0.8284396380186081,
+      "num_tokens": 208076697.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.5903220325708389,
+      "epoch": 0.1696,
+      "grad_norm": 0.5419896245002747,
+      "learning_rate": 1.665294117647059e-05,
+      "loss": 0.577,
+      "mean_token_accuracy": 0.8310789406299591,
+      "num_tokens": 208404377.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.5900990754365921,
+      "epoch": 0.16986666666666667,
+      "grad_norm": 0.5069217085838318,
+      "learning_rate": 1.6647593582887702e-05,
+      "loss": 0.5786,
+      "mean_token_accuracy": 0.8292613610625267,
+      "num_tokens": 208732057.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.5870651677250862,
+      "epoch": 0.17013333333333333,
+      "grad_norm": 0.6124886274337769,
+      "learning_rate": 1.6642245989304814e-05,
+      "loss": 0.5737,
+      "mean_token_accuracy": 0.8296309813857079,
+      "num_tokens": 209059737.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.6034312576055527,
+      "epoch": 0.1704,
+      "grad_norm": 0.48503589630126953,
+      "learning_rate": 1.6636898395721925e-05,
+      "loss": 0.5955,
+      "mean_token_accuracy": 0.8250946953892708,
+      "num_tokens": 209387417.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.5928628303110599,
+      "epoch": 0.17066666666666666,
+      "grad_norm": 0.6071211695671082,
+      "learning_rate": 1.6631550802139037e-05,
+      "loss": 0.5795,
+      "mean_token_accuracy": 0.8299303516745568,
+      "num_tokens": 209715097.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.6154975153505802,
+      "epoch": 0.17093333333333333,
+      "grad_norm": 0.5648673176765442,
+      "learning_rate": 1.6626203208556152e-05,
+      "loss": 0.6087,
+      "mean_token_accuracy": 0.8209341391921043,
+      "num_tokens": 210042777.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.5974450670182705,
+      "epoch": 0.1712,
+      "grad_norm": 0.6501736640930176,
+      "learning_rate": 1.6620855614973264e-05,
+      "loss": 0.5791,
+      "mean_token_accuracy": 0.8285709902644157,
+      "num_tokens": 210370457.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.6025378443300724,
+      "epoch": 0.17146666666666666,
+      "grad_norm": 0.4831034541130066,
+      "learning_rate": 1.6615508021390376e-05,
+      "loss": 0.5969,
+      "mean_token_accuracy": 0.8247403427958488,
+      "num_tokens": 210698137.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.6084349028766155,
+      "epoch": 0.17173333333333332,
+      "grad_norm": 0.5307326316833496,
+      "learning_rate": 1.6610160427807488e-05,
+      "loss": 0.5996,
+      "mean_token_accuracy": 0.8245876103639602,
+      "num_tokens": 211025817.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.5916773281991482,
+      "epoch": 0.172,
+      "grad_norm": 0.6217837929725647,
+      "learning_rate": 1.66048128342246e-05,
+      "loss": 0.5798,
+      "mean_token_accuracy": 0.8278867259621621,
+      "num_tokens": 211353497.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.5969949141144753,
+      "epoch": 0.17226666666666668,
+      "grad_norm": 0.6196262836456299,
+      "learning_rate": 1.659946524064171e-05,
+      "loss": 0.5814,
+      "mean_token_accuracy": 0.8283938154578209,
+      "num_tokens": 211681177.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.6075404763221741,
+      "epoch": 0.17253333333333334,
+      "grad_norm": 0.4899851083755493,
+      "learning_rate": 1.6594117647058827e-05,
+      "loss": 0.6022,
+      "mean_token_accuracy": 0.8226600661873817,
+      "num_tokens": 212008857.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.5535336039960385,
+      "epoch": 0.1728,
+      "grad_norm": 0.5444864630699158,
+      "learning_rate": 1.658877005347594e-05,
+      "loss": 0.5405,
+      "mean_token_accuracy": 0.8371945276856423,
+      "num_tokens": 212336537.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.5652446046471595,
+      "epoch": 0.17306666666666667,
+      "grad_norm": 0.6054065823554993,
+      "learning_rate": 1.658342245989305e-05,
+      "loss": 0.5549,
+      "mean_token_accuracy": 0.8349279016256332,
+      "num_tokens": 212664217.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.5821892686188221,
+      "epoch": 0.17333333333333334,
+      "grad_norm": 0.5747592449188232,
+      "learning_rate": 1.6578074866310162e-05,
+      "loss": 0.5716,
+      "mean_token_accuracy": 0.8304496541619301,
+      "num_tokens": 212991897.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.5729347303509712,
+      "epoch": 0.1736,
+      "grad_norm": 0.648250162601471,
+      "learning_rate": 1.6572727272727274e-05,
+      "loss": 0.5634,
+      "mean_token_accuracy": 0.833754888176918,
+      "num_tokens": 213319577.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.5992898546159268,
+      "epoch": 0.17386666666666667,
+      "grad_norm": 0.5348318815231323,
+      "learning_rate": 1.6567379679144386e-05,
+      "loss": 0.586,
+      "mean_token_accuracy": 0.8264143466949463,
+      "num_tokens": 213647257.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.5950483039021492,
+      "epoch": 0.17413333333333333,
+      "grad_norm": 0.564141035079956,
+      "learning_rate": 1.6562032085561498e-05,
+      "loss": 0.5829,
+      "mean_token_accuracy": 0.8273857519030571,
+      "num_tokens": 213974937.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.5864290438592434,
+      "epoch": 0.1744,
+      "grad_norm": 0.5282854437828064,
+      "learning_rate": 1.6556684491978613e-05,
+      "loss": 0.5734,
+      "mean_token_accuracy": 0.8302297130227089,
+      "num_tokens": 214302617.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.575074841082096,
+      "epoch": 0.17466666666666666,
+      "grad_norm": 0.7073062658309937,
+      "learning_rate": 1.655133689839572e-05,
+      "loss": 0.5675,
+      "mean_token_accuracy": 0.8316715508699417,
+      "num_tokens": 214630297.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.60054025426507,
+      "epoch": 0.17493333333333333,
+      "grad_norm": 0.542799174785614,
+      "learning_rate": 1.6545989304812836e-05,
+      "loss": 0.5877,
+      "mean_token_accuracy": 0.8250641465187073,
+      "num_tokens": 214957977.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.5946882419288159,
+      "epoch": 0.1752,
+      "grad_norm": 0.5276074409484863,
+      "learning_rate": 1.6540641711229948e-05,
+      "loss": 0.5819,
+      "mean_token_accuracy": 0.827055835723877,
+      "num_tokens": 215285657.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.559986025094986,
+      "epoch": 0.17546666666666666,
+      "grad_norm": 0.6103901267051697,
+      "learning_rate": 1.653529411764706e-05,
+      "loss": 0.5507,
+      "mean_token_accuracy": 0.8358565449714661,
+      "num_tokens": 215613337.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.5709345504641533,
+      "epoch": 0.17573333333333332,
+      "grad_norm": 0.5542101860046387,
+      "learning_rate": 1.6529946524064172e-05,
+      "loss": 0.5599,
+      "mean_token_accuracy": 0.8338892966508865,
+      "num_tokens": 215941017.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.6008732952177525,
+      "epoch": 0.176,
+      "grad_norm": 0.4829970896244049,
+      "learning_rate": 1.6524598930481284e-05,
+      "loss": 0.591,
+      "mean_token_accuracy": 0.823536778986454,
+      "num_tokens": 216268697.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.578897013515234,
+      "epoch": 0.17626666666666665,
+      "grad_norm": 0.4680696129798889,
+      "learning_rate": 1.6519251336898395e-05,
+      "loss": 0.571,
+      "mean_token_accuracy": 0.8300830811262131,
+      "num_tokens": 216596377.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.6060771636664868,
+      "epoch": 0.17653333333333332,
+      "grad_norm": 0.641050398349762,
+      "learning_rate": 1.651390374331551e-05,
+      "loss": 0.5989,
+      "mean_token_accuracy": 0.8213648572564125,
+      "num_tokens": 216924057.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.570600051432848,
+      "epoch": 0.1768,
+      "grad_norm": 0.5095393061637878,
+      "learning_rate": 1.6508556149732622e-05,
+      "loss": 0.5581,
+      "mean_token_accuracy": 0.8336479663848877,
+      "num_tokens": 217251737.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.5739085115492344,
+      "epoch": 0.17706666666666668,
+      "grad_norm": 0.46836841106414795,
+      "learning_rate": 1.6503208556149734e-05,
+      "loss": 0.5584,
+      "mean_token_accuracy": 0.8324046924710273,
+      "num_tokens": 217579417.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.5799085564911366,
+      "epoch": 0.17733333333333334,
+      "grad_norm": 0.6916756629943848,
+      "learning_rate": 1.6497860962566846e-05,
+      "loss": 0.5746,
+      "mean_token_accuracy": 0.8289345040917396,
+      "num_tokens": 217907097.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.5554201230406761,
+      "epoch": 0.1776,
+      "grad_norm": 0.5233904123306274,
+      "learning_rate": 1.6492513368983958e-05,
+      "loss": 0.5474,
+      "mean_token_accuracy": 0.8372525691986084,
+      "num_tokens": 218234777.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.6153477519750595,
+      "epoch": 0.17786666666666667,
+      "grad_norm": 0.49100571870803833,
+      "learning_rate": 1.648716577540107e-05,
+      "loss": 0.6032,
+      "mean_token_accuracy": 0.8213343068957328,
+      "num_tokens": 218562457.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.5932588912546635,
+      "epoch": 0.17813333333333334,
+      "grad_norm": 0.5693761706352234,
+      "learning_rate": 1.6481818181818185e-05,
+      "loss": 0.5838,
+      "mean_token_accuracy": 0.8263349220156669,
+      "num_tokens": 218890137.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.5758526347577572,
+      "epoch": 0.1784,
+      "grad_norm": 0.4948122501373291,
+      "learning_rate": 1.6476470588235297e-05,
+      "loss": 0.5647,
+      "mean_token_accuracy": 0.8308101162314415,
+      "num_tokens": 219217817.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.5738284692168236,
+      "epoch": 0.17866666666666667,
+      "grad_norm": 0.5143077373504639,
+      "learning_rate": 1.647112299465241e-05,
+      "loss": 0.5637,
+      "mean_token_accuracy": 0.8301075264811516,
+      "num_tokens": 219545497.0,
+      "step": 6700
+    },
+    {
+      "entropy": 0.5862717747688293,
+      "epoch": 0.17893333333333333,
+      "grad_norm": 0.5274414420127869,
+      "learning_rate": 1.646577540106952e-05,
+      "loss": 0.5743,
+      "mean_token_accuracy": 0.8307337448000908,
+      "num_tokens": 219873177.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.5796642459928989,
+      "epoch": 0.1792,
+      "grad_norm": 0.5390974283218384,
+      "learning_rate": 1.6460427807486632e-05,
+      "loss": 0.5651,
+      "mean_token_accuracy": 0.8310850381851196,
+      "num_tokens": 220200857.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.5697744220495224,
+      "epoch": 0.17946666666666666,
+      "grad_norm": 0.5789408087730408,
+      "learning_rate": 1.6455080213903744e-05,
+      "loss": 0.5583,
+      "mean_token_accuracy": 0.832731555402279,
+      "num_tokens": 220528537.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.5707231611013412,
+      "epoch": 0.17973333333333333,
+      "grad_norm": 0.6618065237998962,
+      "learning_rate": 1.644973262032086e-05,
+      "loss": 0.566,
+      "mean_token_accuracy": 0.8317754209041596,
+      "num_tokens": 220856217.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.5884986869990826,
+      "epoch": 0.18,
+      "grad_norm": 0.56649249792099,
+      "learning_rate": 1.644438502673797e-05,
+      "loss": 0.5703,
+      "mean_token_accuracy": 0.8303549617528916,
+      "num_tokens": 221183897.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.5946214132010936,
+      "epoch": 0.18026666666666666,
+      "grad_norm": 0.622614324092865,
+      "learning_rate": 1.643903743315508e-05,
+      "loss": 0.5848,
+      "mean_token_accuracy": 0.8271841391921043,
+      "num_tokens": 221511577.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.6006936706602574,
+      "epoch": 0.18053333333333332,
+      "grad_norm": 0.5833432078361511,
+      "learning_rate": 1.6433689839572194e-05,
+      "loss": 0.5924,
+      "mean_token_accuracy": 0.8243798896670341,
+      "num_tokens": 221839257.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.5920550875365734,
+      "epoch": 0.1808,
+      "grad_norm": 0.5601273775100708,
+      "learning_rate": 1.6428342245989306e-05,
+      "loss": 0.5823,
+      "mean_token_accuracy": 0.8285984843969345,
+      "num_tokens": 222166937.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.5985722951591015,
+      "epoch": 0.18106666666666665,
+      "grad_norm": 0.5148204565048218,
+      "learning_rate": 1.6422994652406418e-05,
+      "loss": 0.5853,
+      "mean_token_accuracy": 0.8262707710266113,
+      "num_tokens": 222494617.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.5800035849213601,
+      "epoch": 0.18133333333333335,
+      "grad_norm": 0.6548101902008057,
+      "learning_rate": 1.641764705882353e-05,
+      "loss": 0.5705,
+      "mean_token_accuracy": 0.8306207180023193,
+      "num_tokens": 222822297.0,
+      "step": 6800
+    },
+    {
+      "entropy": 0.5585036732256412,
+      "epoch": 0.1816,
+      "grad_norm": 0.7687767744064331,
+      "learning_rate": 1.641229946524064e-05,
+      "loss": 0.5542,
+      "mean_token_accuracy": 0.8359176456928253,
+      "num_tokens": 223149977.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.5878450192511082,
+      "epoch": 0.18186666666666668,
+      "grad_norm": 0.5204241275787354,
+      "learning_rate": 1.6406951871657753e-05,
+      "loss": 0.5737,
+      "mean_token_accuracy": 0.8306757047772407,
+      "num_tokens": 223477657.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.5614672973752022,
+      "epoch": 0.18213333333333334,
+      "grad_norm": 0.56757652759552,
+      "learning_rate": 1.640160427807487e-05,
+      "loss": 0.5491,
+      "mean_token_accuracy": 0.8362933769822121,
+      "num_tokens": 223805337.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.5568563647568225,
+      "epoch": 0.1824,
+      "grad_norm": 0.6357690095901489,
+      "learning_rate": 1.639625668449198e-05,
+      "loss": 0.5419,
+      "mean_token_accuracy": 0.8382422998547554,
+      "num_tokens": 224133017.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.5902631767094135,
+      "epoch": 0.18266666666666667,
+      "grad_norm": 0.5143821835517883,
+      "learning_rate": 1.6390909090909092e-05,
+      "loss": 0.5869,
+      "mean_token_accuracy": 0.8271871924400329,
+      "num_tokens": 224460697.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.5736261554062366,
+      "epoch": 0.18293333333333334,
+      "grad_norm": 0.5533968210220337,
+      "learning_rate": 1.6385561497326204e-05,
+      "loss": 0.565,
+      "mean_token_accuracy": 0.8309562891721726,
+      "num_tokens": 224788340.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.5782340429723263,
+      "epoch": 0.1832,
+      "grad_norm": 0.49603626132011414,
+      "learning_rate": 1.6380213903743316e-05,
+      "loss": 0.566,
+      "mean_token_accuracy": 0.8320259049534797,
+      "num_tokens": 225116020.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.5815359979867936,
+      "epoch": 0.18346666666666667,
+      "grad_norm": 0.6053258180618286,
+      "learning_rate": 1.6374866310160428e-05,
+      "loss": 0.5716,
+      "mean_token_accuracy": 0.8298509255051613,
+      "num_tokens": 225443700.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.5724253945052624,
+      "epoch": 0.18373333333333333,
+      "grad_norm": 0.5965639352798462,
+      "learning_rate": 1.6369518716577543e-05,
+      "loss": 0.5619,
+      "mean_token_accuracy": 0.8321144908666611,
+      "num_tokens": 225771380.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.5732321299612522,
+      "epoch": 0.184,
+      "grad_norm": 0.47802549600601196,
+      "learning_rate": 1.6364171122994655e-05,
+      "loss": 0.5594,
+      "mean_token_accuracy": 0.8322488993406296,
+      "num_tokens": 226099060.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.5756403647363186,
+      "epoch": 0.18426666666666666,
+      "grad_norm": 0.46938326954841614,
+      "learning_rate": 1.6358823529411767e-05,
+      "loss": 0.5635,
+      "mean_token_accuracy": 0.8330309122800827,
+      "num_tokens": 226426740.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.586973873525858,
+      "epoch": 0.18453333333333333,
+      "grad_norm": 0.7300186157226562,
+      "learning_rate": 1.635347593582888e-05,
+      "loss": 0.5772,
+      "mean_token_accuracy": 0.8305107519030571,
+      "num_tokens": 226754420.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.5723816439509392,
+      "epoch": 0.1848,
+      "grad_norm": 0.5234711170196533,
+      "learning_rate": 1.634812834224599e-05,
+      "loss": 0.5593,
+      "mean_token_accuracy": 0.834008426964283,
+      "num_tokens": 227082100.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.5828194707632065,
+      "epoch": 0.18506666666666666,
+      "grad_norm": 0.5104115009307861,
+      "learning_rate": 1.6342780748663102e-05,
+      "loss": 0.567,
+      "mean_token_accuracy": 0.8320075750350953,
+      "num_tokens": 227409780.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.5742134064435959,
+      "epoch": 0.18533333333333332,
+      "grad_norm": 0.519476056098938,
+      "learning_rate": 1.6337433155080217e-05,
+      "loss": 0.5641,
+      "mean_token_accuracy": 0.8308773189783096,
+      "num_tokens": 227737460.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.5768377736210824,
+      "epoch": 0.1856,
+      "grad_norm": 0.5409111976623535,
+      "learning_rate": 1.633208556149733e-05,
+      "loss": 0.5644,
+      "mean_token_accuracy": 0.8322916686534881,
+      "num_tokens": 228065140.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.619147316366434,
+      "epoch": 0.18586666666666668,
+      "grad_norm": 0.5893325209617615,
+      "learning_rate": 1.6326737967914437e-05,
+      "loss": 0.6102,
+      "mean_token_accuracy": 0.8220643892884254,
+      "num_tokens": 228392820.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.5809510044753552,
+      "epoch": 0.18613333333333335,
+      "grad_norm": 0.5766249895095825,
+      "learning_rate": 1.6321390374331553e-05,
+      "loss": 0.5694,
+      "mean_token_accuracy": 0.8289375618100167,
+      "num_tokens": 228720500.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.5793126672506332,
+      "epoch": 0.1864,
+      "grad_norm": 0.6064351201057434,
+      "learning_rate": 1.6316042780748664e-05,
+      "loss": 0.5727,
+      "mean_token_accuracy": 0.831185844540596,
+      "num_tokens": 229048180.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.5913950525224209,
+      "epoch": 0.18666666666666668,
+      "grad_norm": 0.5988383889198303,
+      "learning_rate": 1.6310695187165776e-05,
+      "loss": 0.5775,
+      "mean_token_accuracy": 0.8289558902382851,
+      "num_tokens": 229375860.0,
+      "step": 7000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 37500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 7000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.4215634285066977e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}