diff --git "a/checkpoint-9000/trainer_state.json" "b/checkpoint-9000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9000/trainer_state.json"
@@ -0,0 +1,7534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.7778377961262444,
+  "eval_steps": 500,
+  "global_step": 9000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.8813431424399217,
+      "epoch": 0.003703989505363068,
+      "grad_norm": 0.0770227387547493,
+      "learning_rate": 4.4e-06,
+      "loss": 0.9689,
+      "mean_token_accuracy": 0.7800739134351412,
+      "num_tokens": 2121076.0,
+      "step": 12
+    },
+    {
+      "entropy": 0.8696938330928484,
+      "epoch": 0.007407979010726136,
+      "grad_norm": 0.07951901108026505,
+      "learning_rate": 9.2e-06,
+      "loss": 0.9476,
+      "mean_token_accuracy": 0.7851626686751842,
+      "num_tokens": 4244776.0,
+      "step": 24
+    },
+    {
+      "entropy": 0.9179414622485638,
+      "epoch": 0.011111968516089204,
+      "grad_norm": 0.07233385741710663,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.9521,
+      "mean_token_accuracy": 0.783088818192482,
+      "num_tokens": 6391196.0,
+      "step": 36
+    },
+    {
+      "entropy": 0.9946566758056482,
+      "epoch": 0.014815958021452273,
+      "grad_norm": 0.061557795852422714,
+      "learning_rate": 1.88e-05,
+      "loss": 0.9499,
+      "mean_token_accuracy": 0.7822788568834463,
+      "num_tokens": 8503157.0,
+      "step": 48
+    },
+    {
+      "entropy": 0.9443877513209978,
+      "epoch": 0.01851994752681534,
+      "grad_norm": 0.056875523179769516,
+      "learning_rate": 2.36e-05,
+      "loss": 0.9141,
+      "mean_token_accuracy": 0.7903395928442478,
+      "num_tokens": 10635230.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.9328317133088907,
+      "epoch": 0.02222393703217841,
+      "grad_norm": 0.06272337585687637,
+      "learning_rate": 2.84e-05,
+      "loss": 0.9189,
+      "mean_token_accuracy": 0.7886765040457249,
+      "num_tokens": 12765188.0,
+      "step": 72
+    },
+    {
+      "entropy": 0.9268471834560236,
+      "epoch": 0.025927926537541477,
+      "grad_norm": 0.06604354083538055,
+      "learning_rate": 3.32e-05,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7908310927450657,
+      "num_tokens": 14906939.0,
+      "step": 84
+    },
+    {
+      "entropy": 0.9278339172403017,
+      "epoch": 0.029631916042904546,
+      "grad_norm": 0.07385515421628952,
+      "learning_rate": 3.8e-05,
+      "loss": 0.9058,
+      "mean_token_accuracy": 0.7903257831931114,
+      "num_tokens": 17063752.0,
+      "step": 96
+    },
+    {
+      "entropy": 0.920132706562678,
+      "epoch": 0.03333590554826761,
+      "grad_norm": 0.08544895797967911,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 0.8985,
+      "mean_token_accuracy": 0.7918331262965997,
+      "num_tokens": 19212379.0,
+      "step": 108
+    },
+    {
+      "entropy": 0.9165268304447333,
+      "epoch": 0.03703989505363068,
+      "grad_norm": 0.07440265268087387,
+      "learning_rate": 4.76e-05,
+      "loss": 0.8934,
+      "mean_token_accuracy": 0.7929873218139013,
+      "num_tokens": 21338398.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.9263193979859352,
+      "epoch": 0.04074388455899375,
+      "grad_norm": 0.07524750381708145,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 0.9026,
+      "mean_token_accuracy": 0.7904796029130617,
+      "num_tokens": 23447525.0,
+      "step": 132
+    },
+    {
+      "entropy": 0.9026983591417471,
+      "epoch": 0.04444787406435682,
+      "grad_norm": 0.0853290930390358,
+      "learning_rate": 5.72e-05,
+      "loss": 0.8807,
+      "mean_token_accuracy": 0.7949702950815359,
+      "num_tokens": 25576292.0,
+      "step": 144
+    },
+    {
+      "entropy": 0.91398652891318,
+      "epoch": 0.048151863569719885,
+      "grad_norm": 0.08753567934036255,
+      "learning_rate": 6.2e-05,
+      "loss": 0.893,
+      "mean_token_accuracy": 0.7922145264844099,
+      "num_tokens": 27677943.0,
+      "step": 156
+    },
+    {
+      "entropy": 0.9047610275447369,
+      "epoch": 0.051855853075082954,
+      "grad_norm": 0.09057004004716873,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 0.8825,
+      "mean_token_accuracy": 0.7944387656946977,
+      "num_tokens": 29799957.0,
+      "step": 168
+    },
+    {
+      "entropy": 0.8962277062237263,
+      "epoch": 0.05555984258044602,
+      "grad_norm": 0.09516258537769318,
+      "learning_rate": 7.16e-05,
+      "loss": 0.8759,
+      "mean_token_accuracy": 0.7956550841530164,
+      "num_tokens": 31935987.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.8993425803879896,
+      "epoch": 0.05926383208580909,
+      "grad_norm": 0.09167130291461945,
+      "learning_rate": 7.64e-05,
+      "loss": 0.8781,
+      "mean_token_accuracy": 0.7949150291581949,
+      "num_tokens": 34073688.0,
+      "step": 192
+    },
+    {
+      "entropy": 0.8955355410774549,
+      "epoch": 0.06296782159117216,
+      "grad_norm": 0.10528211295604706,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 0.8769,
+      "mean_token_accuracy": 0.7950745013852915,
+      "num_tokens": 36211776.0,
+      "step": 204
+    },
+    {
+      "entropy": 0.890474279721578,
+      "epoch": 0.06667181109653522,
+      "grad_norm": 0.09239654988050461,
+      "learning_rate": 8.6e-05,
+      "loss": 0.8681,
+      "mean_token_accuracy": 0.7962082512676716,
+      "num_tokens": 38314176.0,
+      "step": 216
+    },
+    {
+      "entropy": 0.8846843503415585,
+      "epoch": 0.0703758006018983,
+      "grad_norm": 0.08734577894210815,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 0.864,
+      "mean_token_accuracy": 0.7977614911894003,
+      "num_tokens": 40427803.0,
+      "step": 228
+    },
+    {
+      "entropy": 0.8876174452404181,
+      "epoch": 0.07407979010726136,
+      "grad_norm": 0.09199592471122742,
+      "learning_rate": 9.56e-05,
+      "loss": 0.8663,
+      "mean_token_accuracy": 0.7968363290031751,
+      "num_tokens": 42601795.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.8842576717336973,
+      "epoch": 0.07778377961262443,
+      "grad_norm": 0.09066279232501984,
+      "learning_rate": 9.999999821283761e-05,
+      "loss": 0.8596,
+      "mean_token_accuracy": 0.7992542050778866,
+      "num_tokens": 44720639.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.8819859276215235,
+      "epoch": 0.0814877691179875,
+      "grad_norm": 0.09380369633436203,
+      "learning_rate": 9.999969796985704e-05,
+      "loss": 0.8606,
+      "mean_token_accuracy": 0.7987109025319418,
+      "num_tokens": 46822627.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.8764691551526388,
+      "epoch": 0.08519175862335057,
+      "grad_norm": 0.08952156454324722,
+      "learning_rate": 9.999888302765345e-05,
+      "loss": 0.8564,
+      "mean_token_accuracy": 0.7992806943754355,
+      "num_tokens": 48939905.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.8867798671126366,
+      "epoch": 0.08889574812871363,
+      "grad_norm": 0.0936591774225235,
+      "learning_rate": 9.999755339461591e-05,
+      "loss": 0.8668,
+      "mean_token_accuracy": 0.7967317899068197,
+      "num_tokens": 51041899.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.8787145999570688,
+      "epoch": 0.09259973763407671,
+      "grad_norm": 0.0926186814904213,
+      "learning_rate": 9.999570908443172e-05,
+      "loss": 0.8574,
+      "mean_token_accuracy": 0.7985943108797073,
+      "num_tokens": 53173319.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.8810293078422546,
+      "epoch": 0.09630372713943977,
+      "grad_norm": 0.09375683963298798,
+      "learning_rate": 9.99933501160863e-05,
+      "loss": 0.8602,
+      "mean_token_accuracy": 0.7984230530758699,
+      "num_tokens": 55298285.0,
+      "step": 312
+    },
+    {
+      "entropy": 0.8722913352151712,
+      "epoch": 0.10000771664480285,
+      "grad_norm": 0.09732625633478165,
+      "learning_rate": 9.999047651386295e-05,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7996316676338514,
+      "num_tokens": 57422076.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.8687134782473246,
+      "epoch": 0.10371170615016591,
+      "grad_norm": 0.08659056574106216,
+      "learning_rate": 9.99870883073427e-05,
+      "loss": 0.8453,
+      "mean_token_accuracy": 0.8012902028858662,
+      "num_tokens": 59585248.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.8830971730252107,
+      "epoch": 0.10741569565552897,
+      "grad_norm": 0.08845651149749756,
+      "learning_rate": 9.998318553140387e-05,
+      "loss": 0.8627,
+      "mean_token_accuracy": 0.7974087223410606,
+      "num_tokens": 61673684.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.8765653309722742,
+      "epoch": 0.11111968516089205,
+      "grad_norm": 0.08888445049524307,
+      "learning_rate": 9.997876822622186e-05,
+      "loss": 0.8556,
+      "mean_token_accuracy": 0.7995972596108913,
+      "num_tokens": 63792291.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.871281652400891,
+      "epoch": 0.11482367466625511,
+      "grad_norm": 0.09176748991012573,
+      "learning_rate": 9.99738364372686e-05,
+      "loss": 0.8534,
+      "mean_token_accuracy": 0.7989814169704914,
+      "num_tokens": 65901685.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.8688261434435844,
+      "epoch": 0.11852766417161818,
+      "grad_norm": 0.09121166169643402,
+      "learning_rate": 9.996839021531213e-05,
+      "loss": 0.8467,
+      "mean_token_accuracy": 0.8007690558830897,
+      "num_tokens": 68051060.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.8706431252261003,
+      "epoch": 0.12223165367698124,
+      "grad_norm": 0.09300525486469269,
+      "learning_rate": 9.996242961641615e-05,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7998151158293089,
+      "num_tokens": 70196769.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.8776354491710663,
+      "epoch": 0.12593564318234432,
+      "grad_norm": 0.08426107466220856,
+      "learning_rate": 9.995595470193933e-05,
+      "loss": 0.8545,
+      "mean_token_accuracy": 0.7993121594190598,
+      "num_tokens": 72307999.0,
+      "step": 408
+    },
+    {
+      "entropy": 0.8773620029290518,
+      "epoch": 0.1296396326877074,
+      "grad_norm": 0.09252817928791046,
+      "learning_rate": 9.994896553853472e-05,
+      "loss": 0.8581,
+      "mean_token_accuracy": 0.7982841742535433,
+      "num_tokens": 74422316.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.8797574390967687,
+      "epoch": 0.13334362219307044,
+      "grad_norm": 0.08833932876586914,
+      "learning_rate": 9.994146219814912e-05,
+      "loss": 0.8583,
+      "mean_token_accuracy": 0.7982658172647158,
+      "num_tokens": 76535665.0,
+      "step": 432
+    },
+    {
+      "entropy": 0.8648973529537519,
+      "epoch": 0.13704761169843352,
+      "grad_norm": 0.08894747495651245,
+      "learning_rate": 9.993344475802226e-05,
+      "loss": 0.8449,
+      "mean_token_accuracy": 0.8004408640166124,
+      "num_tokens": 78656429.0,
+      "step": 444
+    },
+    {
+      "entropy": 0.8582859995464484,
+      "epoch": 0.1407516012037966,
+      "grad_norm": 0.09439876675605774,
+      "learning_rate": 9.992491330068606e-05,
+      "loss": 0.8369,
+      "mean_token_accuracy": 0.8031645379960537,
+      "num_tokens": 80814117.0,
+      "step": 456
+    },
+    {
+      "entropy": 0.8811340468625227,
+      "epoch": 0.14445559070915967,
+      "grad_norm": 0.0849665030837059,
+      "learning_rate": 9.99158679139637e-05,
+      "loss": 0.858,
+      "mean_token_accuracy": 0.7984087901810805,
+      "num_tokens": 82925792.0,
+      "step": 468
+    },
+    {
+      "entropy": 0.8554788815478483,
+      "epoch": 0.14815958021452272,
+      "grad_norm": 0.0906689241528511,
+      "learning_rate": 9.990630869096883e-05,
+      "loss": 0.8383,
+      "mean_token_accuracy": 0.8024374966820081,
+      "num_tokens": 85082604.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8597272150218487,
+      "epoch": 0.1518635697198858,
+      "grad_norm": 0.08773977309465408,
+      "learning_rate": 9.989623573010455e-05,
+      "loss": 0.8391,
+      "mean_token_accuracy": 0.8020844186345736,
+      "num_tokens": 87207502.0,
+      "step": 492
+    },
+    {
+      "entropy": 0.8616875174144903,
+      "epoch": 0.15556755922524887,
+      "grad_norm": 0.08853704482316971,
+      "learning_rate": 9.988564913506238e-05,
+      "loss": 0.8414,
+      "mean_token_accuracy": 0.801638551056385,
+      "num_tokens": 89348327.0,
+      "step": 504
+    },
+    {
+      "entropy": 0.8478845059871674,
+      "epoch": 0.15927154873061192,
+      "grad_norm": 0.09445828199386597,
+      "learning_rate": 9.987454901482122e-05,
+      "loss": 0.8285,
+      "mean_token_accuracy": 0.8042828006048998,
+      "num_tokens": 91504049.0,
+      "step": 516
+    },
+    {
+      "entropy": 0.8598806957403818,
+      "epoch": 0.162975538235975,
+      "grad_norm": 0.09136994183063507,
+      "learning_rate": 9.986293548364622e-05,
+      "loss": 0.8411,
+      "mean_token_accuracy": 0.8020358892778555,
+      "num_tokens": 93626935.0,
+      "step": 528
+    },
+    {
+      "entropy": 0.8653969628115495,
+      "epoch": 0.16667952774133807,
+      "grad_norm": 0.09781411290168762,
+      "learning_rate": 9.985080866108762e-05,
+      "loss": 0.8444,
+      "mean_token_accuracy": 0.8006485402584076,
+      "num_tokens": 95703749.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8538096360862255,
+      "epoch": 0.17038351724670114,
+      "grad_norm": 0.08897630125284195,
+      "learning_rate": 9.983816867197953e-05,
+      "loss": 0.8359,
+      "mean_token_accuracy": 0.8030762349565824,
+      "num_tokens": 97836700.0,
+      "step": 552
+    },
+    {
+      "entropy": 0.8514401825765768,
+      "epoch": 0.1740875067520642,
+      "grad_norm": 0.09062942117452621,
+      "learning_rate": 9.982501564643852e-05,
+      "loss": 0.8295,
+      "mean_token_accuracy": 0.8046696037054062,
+      "num_tokens": 99924036.0,
+      "step": 564
+    },
+    {
+      "entropy": 0.8512975362439951,
+      "epoch": 0.17779149625742727,
+      "grad_norm": 0.0891498252749443,
+      "learning_rate": 9.98113497198625e-05,
+      "loss": 0.8304,
+      "mean_token_accuracy": 0.8033984725673994,
+      "num_tokens": 102057627.0,
+      "step": 576
+    },
+    {
+      "entropy": 0.8456763414045175,
+      "epoch": 0.18149548576279034,
+      "grad_norm": 0.0918072909116745,
+      "learning_rate": 9.979717103292912e-05,
+      "loss": 0.8262,
+      "mean_token_accuracy": 0.8045558805267016,
+      "num_tokens": 104162821.0,
+      "step": 588
+    },
+    {
+      "entropy": 0.8628777662913004,
+      "epoch": 0.18519947526815342,
+      "grad_norm": 0.09555666148662567,
+      "learning_rate": 9.978247973159448e-05,
+      "loss": 0.8427,
+      "mean_token_accuracy": 0.8011121414601803,
+      "num_tokens": 106260558.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8535663560032845,
+      "epoch": 0.18890346477351647,
+      "grad_norm": 0.09658853709697723,
+      "learning_rate": 9.97672759670915e-05,
+      "loss": 0.8311,
+      "mean_token_accuracy": 0.8037117148439089,
+      "num_tokens": 108347746.0,
+      "step": 612
+    },
+    {
+      "entropy": 0.850828155875206,
+      "epoch": 0.19260745427887954,
+      "grad_norm": 0.09168772399425507,
+      "learning_rate": 9.975155989592844e-05,
+      "loss": 0.8351,
+      "mean_token_accuracy": 0.8032047462960085,
+      "num_tokens": 110475744.0,
+      "step": 624
+    },
+    {
+      "entropy": 0.8446941214303175,
+      "epoch": 0.19631144378424262,
+      "grad_norm": 0.09660109132528305,
+      "learning_rate": 9.973533167988728e-05,
+      "loss": 0.8261,
+      "mean_token_accuracy": 0.805509191006422,
+      "num_tokens": 112562500.0,
+      "step": 636
+    },
+    {
+      "entropy": 0.8449327821532885,
+      "epoch": 0.2000154332896057,
+      "grad_norm": 0.0988537073135376,
+      "learning_rate": 9.971859148602202e-05,
+      "loss": 0.8235,
+      "mean_token_accuracy": 0.8050300491352876,
+      "num_tokens": 114663477.0,
+      "step": 648
+    },
+    {
+      "entropy": 0.8418795031805834,
+      "epoch": 0.20371942279496874,
+      "grad_norm": 0.10221028327941895,
+      "learning_rate": 9.970133948665702e-05,
+      "loss": 0.8259,
+      "mean_token_accuracy": 0.8041424031058947,
+      "num_tokens": 116824983.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8358608248333136,
+      "epoch": 0.20742341230033182,
+      "grad_norm": 0.09639787673950195,
+      "learning_rate": 9.968357585938515e-05,
+      "loss": 0.815,
+      "mean_token_accuracy": 0.8072937255104383,
+      "num_tokens": 118942813.0,
+      "step": 672
+    },
+    {
+      "entropy": 0.8451284418503443,
+      "epoch": 0.2111274018056949,
+      "grad_norm": 0.09338120371103287,
+      "learning_rate": 9.966530078706599e-05,
+      "loss": 0.8267,
+      "mean_token_accuracy": 0.8043443597853184,
+      "num_tokens": 121069569.0,
+      "step": 684
+    },
+    {
+      "entropy": 0.8439769372344017,
+      "epoch": 0.21483139131105794,
+      "grad_norm": 0.10154031217098236,
+      "learning_rate": 9.964651445782405e-05,
+      "loss": 0.8258,
+      "mean_token_accuracy": 0.8052287250757217,
+      "num_tokens": 123193984.0,
+      "step": 696
+    },
+    {
+      "entropy": 0.8514358488221964,
+      "epoch": 0.21853538081642102,
+      "grad_norm": 0.09599766135215759,
+      "learning_rate": 9.962721706504663e-05,
+      "loss": 0.8331,
+      "mean_token_accuracy": 0.8030676891406378,
+      "num_tokens": 125349706.0,
+      "step": 708
+    },
+    {
+      "entropy": 0.8406069638828436,
+      "epoch": 0.2222393703217841,
+      "grad_norm": 0.09738138318061829,
+      "learning_rate": 9.9607408807382e-05,
+      "loss": 0.8218,
+      "mean_token_accuracy": 0.8066203904648622,
+      "num_tokens": 127475845.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8436915278434753,
+      "epoch": 0.22594335982714717,
+      "grad_norm": 0.0990961417555809,
+      "learning_rate": 9.958708988873729e-05,
+      "loss": 0.8254,
+      "mean_token_accuracy": 0.8044916093349457,
+      "num_tokens": 129575388.0,
+      "step": 732
+    },
+    {
+      "entropy": 0.8330078596870104,
+      "epoch": 0.22964734933251021,
+      "grad_norm": 0.09744027256965637,
+      "learning_rate": 9.956626051827643e-05,
+      "loss": 0.8138,
+      "mean_token_accuracy": 0.8071556240320206,
+      "num_tokens": 131698547.0,
+      "step": 744
+    },
+    {
+      "entropy": 0.8432958399256071,
+      "epoch": 0.2333513388378733,
+      "grad_norm": 0.1007818877696991,
+      "learning_rate": 9.954492091041788e-05,
+      "loss": 0.8227,
+      "mean_token_accuracy": 0.804592452943325,
+      "num_tokens": 133841198.0,
+      "step": 756
+    },
+    {
+      "entropy": 0.836149275302887,
+      "epoch": 0.23705532834323637,
+      "grad_norm": 0.09519964456558228,
+      "learning_rate": 9.952307128483256e-05,
+      "loss": 0.8178,
+      "mean_token_accuracy": 0.8064361910025278,
+      "num_tokens": 135976771.0,
+      "step": 768
+    },
+    {
+      "entropy": 0.8368441437681516,
+      "epoch": 0.24075931784859944,
+      "grad_norm": 0.10342419147491455,
+      "learning_rate": 9.950071186644159e-05,
+      "loss": 0.8176,
+      "mean_token_accuracy": 0.8063248756031195,
+      "num_tokens": 138094553.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8424511601527532,
+      "epoch": 0.2444633073539625,
+      "grad_norm": 0.0969947949051857,
+      "learning_rate": 9.94778428854138e-05,
+      "loss": 0.8208,
+      "mean_token_accuracy": 0.8055132577816645,
+      "num_tokens": 140199892.0,
+      "step": 792
+    },
+    {
+      "entropy": 0.8211217597126961,
+      "epoch": 0.24816729685932556,
+      "grad_norm": 0.09834083914756775,
+      "learning_rate": 9.945446457716359e-05,
+      "loss": 0.8014,
+      "mean_token_accuracy": 0.8097951101760069,
+      "num_tokens": 142340717.0,
+      "step": 804
+    },
+    {
+      "entropy": 0.8515114995340506,
+      "epoch": 0.25187128636468864,
+      "grad_norm": 0.10160677880048752,
+      "learning_rate": 9.943057718234836e-05,
+      "loss": 0.8317,
+      "mean_token_accuracy": 0.802549467732509,
+      "num_tokens": 144490208.0,
+      "step": 816
+    },
+    {
+      "entropy": 0.8411018749078115,
+      "epoch": 0.2555752758700517,
+      "grad_norm": 0.10110847651958466,
+      "learning_rate": 9.940618094686603e-05,
+      "loss": 0.8243,
+      "mean_token_accuracy": 0.8038219797114531,
+      "num_tokens": 146598967.0,
+      "step": 828
+    },
+    {
+      "entropy": 0.8269097929199537,
+      "epoch": 0.2592792653754148,
+      "grad_norm": 0.10856305807828903,
+      "learning_rate": 9.938127612185261e-05,
+      "loss": 0.8078,
+      "mean_token_accuracy": 0.808723546564579,
+      "num_tokens": 148706602.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8350271495680014,
+      "epoch": 0.26298325488077784,
+      "grad_norm": 0.10275018215179443,
+      "learning_rate": 9.935586296367953e-05,
+      "loss": 0.8144,
+      "mean_token_accuracy": 0.8073840402066708,
+      "num_tokens": 150831294.0,
+      "step": 852
+    },
+    {
+      "entropy": 0.8402173941334089,
+      "epoch": 0.2666872443861409,
+      "grad_norm": 0.10895514488220215,
+      "learning_rate": 9.932994173395103e-05,
+      "loss": 0.823,
+      "mean_token_accuracy": 0.8050992513696352,
+      "num_tokens": 152914180.0,
+      "step": 864
+    },
+    {
+      "entropy": 0.835850744197766,
+      "epoch": 0.270391233891504,
+      "grad_norm": 0.1103854700922966,
+      "learning_rate": 9.930351269950143e-05,
+      "loss": 0.8149,
+      "mean_token_accuracy": 0.8063218059639136,
+      "num_tokens": 155043841.0,
+      "step": 876
+    },
+    {
+      "entropy": 0.8304483393828074,
+      "epoch": 0.27409522339686704,
+      "grad_norm": 0.09872893989086151,
+      "learning_rate": 9.927657613239247e-05,
+      "loss": 0.8124,
+      "mean_token_accuracy": 0.8073564060032368,
+      "num_tokens": 157194273.0,
+      "step": 888
+    },
+    {
+      "entropy": 0.8408761695027351,
+      "epoch": 0.2777992129022301,
+      "grad_norm": 0.10328029841184616,
+      "learning_rate": 9.924913230991044e-05,
+      "loss": 0.8201,
+      "mean_token_accuracy": 0.8051391566793124,
+      "num_tokens": 159325166.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.8261769798894724,
+      "epoch": 0.2815032024075932,
+      "grad_norm": 0.1065843477845192,
+      "learning_rate": 9.922118151456327e-05,
+      "loss": 0.8074,
+      "mean_token_accuracy": 0.8085715671380361,
+      "num_tokens": 161429666.0,
+      "step": 912
+    },
+    {
+      "entropy": 0.8249383146564165,
+      "epoch": 0.28520719191295624,
+      "grad_norm": 0.10588902980089188,
+      "learning_rate": 9.919272403407782e-05,
+      "loss": 0.8061,
+      "mean_token_accuracy": 0.8078173498312632,
+      "num_tokens": 163588436.0,
+      "step": 924
+    },
+    {
+      "entropy": 0.8257463040451208,
+      "epoch": 0.28891118141831934,
+      "grad_norm": 0.10355869680643082,
+      "learning_rate": 9.91637601613967e-05,
+      "loss": 0.8065,
+      "mean_token_accuracy": 0.8088524142901102,
+      "num_tokens": 165703594.0,
+      "step": 936
+    },
+    {
+      "entropy": 0.8323748372495174,
+      "epoch": 0.2926151709236824,
+      "grad_norm": 0.11059477180242538,
+      "learning_rate": 9.913429019467534e-05,
+      "loss": 0.8135,
+      "mean_token_accuracy": 0.8072844197352728,
+      "num_tokens": 167834747.0,
+      "step": 948
+    },
+    {
+      "entropy": 0.8198406957089901,
+      "epoch": 0.29631916042904544,
+      "grad_norm": 0.11167987436056137,
+      "learning_rate": 9.910431443727897e-05,
+      "loss": 0.8005,
+      "mean_token_accuracy": 0.8094635804494222,
+      "num_tokens": 169954318.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8273561559617519,
+      "epoch": 0.30002314993440854,
+      "grad_norm": 0.11263331025838852,
+      "learning_rate": 9.907383319777945e-05,
+      "loss": 0.8066,
+      "mean_token_accuracy": 0.8084728009998798,
+      "num_tokens": 172056219.0,
+      "step": 972
+    },
+    {
+      "entropy": 0.8382978811860085,
+      "epoch": 0.3037271394397716,
+      "grad_norm": 0.1031041070818901,
+      "learning_rate": 9.904284678995206e-05,
+      "loss": 0.8204,
+      "mean_token_accuracy": 0.8051142630477747,
+      "num_tokens": 174209607.0,
+      "step": 984
+    },
+    {
+      "entropy": 0.8209519870579243,
+      "epoch": 0.30743112894513464,
+      "grad_norm": 0.10300373286008835,
+      "learning_rate": 9.901135553277232e-05,
+      "loss": 0.798,
+      "mean_token_accuracy": 0.8102124035358429,
+      "num_tokens": 176340332.0,
+      "step": 996
+    },
+    {
+      "entropy": 0.8192687357465426,
+      "epoch": 0.31113511845049774,
+      "grad_norm": 0.11473007500171661,
+      "learning_rate": 9.897935975041273e-05,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.8089515815178553,
+      "num_tokens": 178465989.0,
+      "step": 1008
+    },
+    {
+      "entropy": 0.8185019778708617,
+      "epoch": 0.3148391079558608,
+      "grad_norm": 0.10434567183256149,
+      "learning_rate": 9.894685977223934e-05,
+      "loss": 0.7995,
+      "mean_token_accuracy": 0.8095376479128996,
+      "num_tokens": 180608399.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8181099407374859,
+      "epoch": 0.31854309746122383,
+      "grad_norm": 0.10841691493988037,
+      "learning_rate": 9.891385593280847e-05,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.8103865322967371,
+      "num_tokens": 182744332.0,
+      "step": 1032
+    },
+    {
+      "entropy": 0.8311556254824003,
+      "epoch": 0.32224708696658694,
+      "grad_norm": 0.11030473560094833,
+      "learning_rate": 9.888034857186315e-05,
+      "loss": 0.8107,
+      "mean_token_accuracy": 0.8080282248556614,
+      "num_tokens": 184878580.0,
+      "step": 1044
+    },
+    {
+      "entropy": 0.8324671102066835,
+      "epoch": 0.32595107647195,
+      "grad_norm": 0.11505813896656036,
+      "learning_rate": 9.884633803432972e-05,
+      "loss": 0.816,
+      "mean_token_accuracy": 0.8058654479682446,
+      "num_tokens": 186987179.0,
+      "step": 1056
+    },
+    {
+      "entropy": 0.8222536593675613,
+      "epoch": 0.3296550659773131,
+      "grad_norm": 0.11003649234771729,
+      "learning_rate": 9.881182467031427e-05,
+      "loss": 0.8024,
+      "mean_token_accuracy": 0.8091614345709482,
+      "num_tokens": 189146940.0,
+      "step": 1068
+    },
+    {
+      "entropy": 0.8215553897122542,
+      "epoch": 0.33335905548267614,
+      "grad_norm": 0.11053642630577087,
+      "learning_rate": 9.877680883509895e-05,
+      "loss": 0.8045,
+      "mean_token_accuracy": 0.8081031031906605,
+      "num_tokens": 191273140.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8301792852580547,
+      "epoch": 0.3370630449880392,
+      "grad_norm": 0.11618710309267044,
+      "learning_rate": 9.874129088913842e-05,
+      "loss": 0.8111,
+      "mean_token_accuracy": 0.8073916758100191,
+      "num_tokens": 193419995.0,
+      "step": 1092
+    },
+    {
+      "entropy": 0.8200336868564287,
+      "epoch": 0.3407670344934023,
+      "grad_norm": 0.11068252474069595,
+      "learning_rate": 9.870527119805603e-05,
+      "loss": 0.7987,
+      "mean_token_accuracy": 0.8100821475187937,
+      "num_tokens": 195570082.0,
+      "step": 1104
+    },
+    {
+      "entropy": 0.8287765470643839,
+      "epoch": 0.34447102399876534,
+      "grad_norm": 0.10776914656162262,
+      "learning_rate": 9.866875013264023e-05,
+      "loss": 0.8118,
+      "mean_token_accuracy": 0.8070496221383413,
+      "num_tokens": 197675630.0,
+      "step": 1116
+    },
+    {
+      "entropy": 0.8167796346048514,
+      "epoch": 0.3481750135041284,
+      "grad_norm": 0.10457509756088257,
+      "learning_rate": 9.863172806884051e-05,
+      "loss": 0.798,
+      "mean_token_accuracy": 0.8097474686801434,
+      "num_tokens": 199804447.0,
+      "step": 1128
+    },
+    {
+      "entropy": 0.8168069496750832,
+      "epoch": 0.3518790030094915,
+      "grad_norm": 0.10983676463365555,
+      "learning_rate": 9.859420538776376e-05,
+      "loss": 0.7981,
+      "mean_token_accuracy": 0.8102098839978377,
+      "num_tokens": 201919269.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.808088593184948,
+      "epoch": 0.35558299251485453,
+      "grad_norm": 0.11442640423774719,
+      "learning_rate": 9.855618247567018e-05,
+      "loss": 0.7926,
+      "mean_token_accuracy": 0.8110264576971531,
+      "num_tokens": 204057926.0,
+      "step": 1152
+    },
+    {
+      "entropy": 0.8169751837849617,
+      "epoch": 0.3592869820202176,
+      "grad_norm": 0.10822667181491852,
+      "learning_rate": 9.851765972396943e-05,
+      "loss": 0.7967,
+      "mean_token_accuracy": 0.8103730641305447,
+      "num_tokens": 206175396.0,
+      "step": 1164
+    },
+    {
+      "entropy": 0.814443551003933,
+      "epoch": 0.3629909715255807,
+      "grad_norm": 0.11191259324550629,
+      "learning_rate": 9.847863752921649e-05,
+      "loss": 0.7933,
+      "mean_token_accuracy": 0.8108923596640428,
+      "num_tokens": 208298979.0,
+      "step": 1176
+    },
+    {
+      "entropy": 0.807010448227326,
+      "epoch": 0.36669496103094373,
+      "grad_norm": 0.11479055136442184,
+      "learning_rate": 9.843911629310764e-05,
+      "loss": 0.7892,
+      "mean_token_accuracy": 0.8113560838003954,
+      "num_tokens": 210441373.0,
+      "step": 1188
+    },
+    {
+      "entropy": 0.8038304224610329,
+      "epoch": 0.37039895053630684,
+      "grad_norm": 0.12469020485877991,
+      "learning_rate": 9.839909642247637e-05,
+      "loss": 0.7845,
+      "mean_token_accuracy": 0.8129827156662941,
+      "num_tokens": 212573549.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.8132020806272825,
+      "epoch": 0.3741029400416699,
+      "grad_norm": 0.11616237461566925,
+      "learning_rate": 9.835857832928908e-05,
+      "loss": 0.7946,
+      "mean_token_accuracy": 0.8103696592152119,
+      "num_tokens": 214646922.0,
+      "step": 1212
+    },
+    {
+      "entropy": 0.8121326218048731,
+      "epoch": 0.37780692954703293,
+      "grad_norm": 0.11775597929954529,
+      "learning_rate": 9.831756243064088e-05,
+      "loss": 0.7945,
+      "mean_token_accuracy": 0.8108564031620821,
+      "num_tokens": 216767641.0,
+      "step": 1224
+    },
+    {
+      "entropy": 0.8172868477801482,
+      "epoch": 0.38151091905239604,
+      "grad_norm": 0.11528316140174866,
+      "learning_rate": 9.827604914875139e-05,
+      "loss": 0.7979,
+      "mean_token_accuracy": 0.8101303689181805,
+      "num_tokens": 218905400.0,
+      "step": 1236
+    },
+    {
+      "entropy": 0.8001810808976492,
+      "epoch": 0.3852149085577591,
+      "grad_norm": 0.1265256702899933,
+      "learning_rate": 9.823403891096024e-05,
+      "loss": 0.7804,
+      "mean_token_accuracy": 0.813588964442412,
+      "num_tokens": 220993960.0,
+      "step": 1248
+    },
+    {
+      "entropy": 0.8007968428234259,
+      "epoch": 0.38891889806312213,
+      "grad_norm": 0.11904493719339371,
+      "learning_rate": 9.819153214972279e-05,
+      "loss": 0.7826,
+      "mean_token_accuracy": 0.8128631959358851,
+      "num_tokens": 223114091.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.8208761115868887,
+      "epoch": 0.39262288756848523,
+      "grad_norm": 0.12480930238962173,
+      "learning_rate": 9.814852930260561e-05,
+      "loss": 0.7983,
+      "mean_token_accuracy": 0.8095660296579202,
+      "num_tokens": 225247556.0,
+      "step": 1272
+    },
+    {
+      "entropy": 0.815931453059117,
+      "epoch": 0.3963268770738483,
+      "grad_norm": 0.11885613203048706,
+      "learning_rate": 9.810503081228202e-05,
+      "loss": 0.7996,
+      "mean_token_accuracy": 0.8100833334028721,
+      "num_tokens": 227351759.0,
+      "step": 1284
+    },
+    {
+      "entropy": 0.8205481581389904,
+      "epoch": 0.4000308665792114,
+      "grad_norm": 0.11844924092292786,
+      "learning_rate": 9.80610371265275e-05,
+      "loss": 0.8019,
+      "mean_token_accuracy": 0.808446753770113,
+      "num_tokens": 229441295.0,
+      "step": 1296
+    },
+    {
+      "entropy": 0.8117605733374754,
+      "epoch": 0.40373485608457443,
+      "grad_norm": 0.11241784691810608,
+      "learning_rate": 9.801654869821512e-05,
+      "loss": 0.7932,
+      "mean_token_accuracy": 0.81082005550464,
+      "num_tokens": 231566470.0,
+      "step": 1308
+    },
+    {
+      "entropy": 0.8152283951640129,
+      "epoch": 0.4074388455899375,
+      "grad_norm": 0.13755977153778076,
+      "learning_rate": 9.797156598531085e-05,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.8106731151541074,
+      "num_tokens": 233658940.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.8065759042898814,
+      "epoch": 0.4111428350953006,
+      "grad_norm": 0.11853731423616409,
+      "learning_rate": 9.79260894508688e-05,
+      "loss": 0.7894,
+      "mean_token_accuracy": 0.8116564750671387,
+      "num_tokens": 235792409.0,
+      "step": 1332
+    },
+    {
+      "entropy": 0.8189149846633276,
+      "epoch": 0.41484682460066363,
+      "grad_norm": 0.11743967235088348,
+      "learning_rate": 9.788011956302656e-05,
+      "loss": 0.798,
+      "mean_token_accuracy": 0.8094673914213976,
+      "num_tokens": 237882728.0,
+      "step": 1344
+    },
+    {
+      "entropy": 0.8161345782379309,
+      "epoch": 0.4185508141060267,
+      "grad_norm": 0.12426267564296722,
+      "learning_rate": 9.783365679500027e-05,
+      "loss": 0.7954,
+      "mean_token_accuracy": 0.8102164330581824,
+      "num_tokens": 240024867.0,
+      "step": 1356
+    },
+    {
+      "entropy": 0.7942858090003332,
+      "epoch": 0.4222548036113898,
+      "grad_norm": 0.11311879754066467,
+      "learning_rate": 9.778670162507986e-05,
+      "loss": 0.7759,
+      "mean_token_accuracy": 0.8148905138174692,
+      "num_tokens": 242149516.0,
+      "step": 1368
+    },
+    {
+      "entropy": 0.803426214804252,
+      "epoch": 0.42595879311675283,
+      "grad_norm": 0.11434963345527649,
+      "learning_rate": 9.773925453662403e-05,
+      "loss": 0.785,
+      "mean_token_accuracy": 0.8126038151482741,
+      "num_tokens": 244256116.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7926236540079117,
+      "epoch": 0.4296627826221159,
+      "grad_norm": 0.10715149343013763,
+      "learning_rate": 9.769131601805534e-05,
+      "loss": 0.7733,
+      "mean_token_accuracy": 0.8152074652413527,
+      "num_tokens": 246406475.0,
+      "step": 1392
+    },
+    {
+      "entropy": 0.8100047335028648,
+      "epoch": 0.433366772127479,
+      "grad_norm": 0.11607677489519119,
+      "learning_rate": 9.76428865628551e-05,
+      "loss": 0.7907,
+      "mean_token_accuracy": 0.8116505754490694,
+      "num_tokens": 248494590.0,
+      "step": 1404
+    },
+    {
+      "entropy": 0.8006227687001228,
+      "epoch": 0.43707076163284203,
+      "grad_norm": 0.11636164784431458,
+      "learning_rate": 9.75939666695584e-05,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.8131669908761978,
+      "num_tokens": 250624649.0,
+      "step": 1416
+    },
+    {
+      "entropy": 0.8028407072027525,
+      "epoch": 0.44077475113820513,
+      "grad_norm": 0.12113169580698013,
+      "learning_rate": 9.75445568417489e-05,
+      "loss": 0.7819,
+      "mean_token_accuracy": 0.812740029146274,
+      "num_tokens": 252753433.0,
+      "step": 1428
+    },
+    {
+      "entropy": 0.811029980580012,
+      "epoch": 0.4444787406435682,
+      "grad_norm": 0.1194261834025383,
+      "learning_rate": 9.74946575880537e-05,
+      "loss": 0.7925,
+      "mean_token_accuracy": 0.8101853169500828,
+      "num_tokens": 254919606.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8054500371217728,
+      "epoch": 0.44818273014893123,
+      "grad_norm": 0.12146873027086258,
+      "learning_rate": 9.744426942213799e-05,
+      "loss": 0.788,
+      "mean_token_accuracy": 0.8125789103408655,
+      "num_tokens": 257037052.0,
+      "step": 1452
+    },
+    {
+      "entropy": 0.7984961271286011,
+      "epoch": 0.45188671965429433,
+      "grad_norm": 0.12276361137628555,
+      "learning_rate": 9.739339286269995e-05,
+      "loss": 0.7787,
+      "mean_token_accuracy": 0.8136009263495604,
+      "num_tokens": 259200523.0,
+      "step": 1464
+    },
+    {
+      "entropy": 0.8060410469770432,
+      "epoch": 0.4555907091596574,
+      "grad_norm": 0.12075335532426834,
+      "learning_rate": 9.734202843346522e-05,
+      "loss": 0.7882,
+      "mean_token_accuracy": 0.8113779984414577,
+      "num_tokens": 261303525.0,
+      "step": 1476
+    },
+    {
+      "entropy": 0.8057870579262575,
+      "epoch": 0.45929469866502043,
+      "grad_norm": 0.12227274477481842,
+      "learning_rate": 9.729017666318165e-05,
+      "loss": 0.7868,
+      "mean_token_accuracy": 0.8118944627543291,
+      "num_tokens": 263437688.0,
+      "step": 1488
+    },
+    {
+      "entropy": 0.8005211018025875,
+      "epoch": 0.46299868817038353,
+      "grad_norm": 0.11424333602190018,
+      "learning_rate": 9.723783808561378e-05,
+      "loss": 0.7791,
+      "mean_token_accuracy": 0.8138281057278315,
+      "num_tokens": 265568368.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8023173411687216,
+      "epoch": 0.4667026776757466,
+      "grad_norm": 0.13906393945217133,
+      "learning_rate": 9.718501323953737e-05,
+      "loss": 0.7853,
+      "mean_token_accuracy": 0.811311274766922,
+      "num_tokens": 267706317.0,
+      "step": 1512
+    },
+    {
+      "entropy": 0.8110432215034962,
+      "epoch": 0.47040666718110963,
+      "grad_norm": 0.11983723938465118,
+      "learning_rate": 9.713170266873384e-05,
+      "loss": 0.7893,
+      "mean_token_accuracy": 0.8107657519479593,
+      "num_tokens": 269835542.0,
+      "step": 1524
+    },
+    {
+      "entropy": 0.7883443137009939,
+      "epoch": 0.47411065668647273,
+      "grad_norm": 0.13033325970172882,
+      "learning_rate": 9.70779069219847e-05,
+      "loss": 0.7734,
+      "mean_token_accuracy": 0.8156957700848579,
+      "num_tokens": 271983602.0,
+      "step": 1536
+    },
+    {
+      "entropy": 0.7950549945235252,
+      "epoch": 0.4778146461918358,
+      "grad_norm": 0.10897476971149445,
+      "learning_rate": 9.702362655306587e-05,
+      "loss": 0.775,
+      "mean_token_accuracy": 0.8149407207965851,
+      "num_tokens": 274107343.0,
+      "step": 1548
+    },
+    {
+      "entropy": 0.7990003265440464,
+      "epoch": 0.4815186356971989,
+      "grad_norm": 0.11902206391096115,
+      "learning_rate": 9.696886212074202e-05,
+      "loss": 0.7808,
+      "mean_token_accuracy": 0.8132348544895649,
+      "num_tokens": 276222435.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.801575344055891,
+      "epoch": 0.48522262520256193,
+      "grad_norm": 0.11533886194229126,
+      "learning_rate": 9.691361418876075e-05,
+      "loss": 0.7795,
+      "mean_token_accuracy": 0.8133863943318526,
+      "num_tokens": 278355706.0,
+      "step": 1572
+    },
+    {
+      "entropy": 0.7944580602149168,
+      "epoch": 0.488926614707925,
+      "grad_norm": 0.12622949481010437,
+      "learning_rate": 9.685788332584685e-05,
+      "loss": 0.7752,
+      "mean_token_accuracy": 0.814650778969129,
+      "num_tokens": 280460375.0,
+      "step": 1584
+    },
+    {
+      "entropy": 0.8005058703323206,
+      "epoch": 0.4926306042132881,
+      "grad_norm": 0.12395934015512466,
+      "learning_rate": 9.68016701056964e-05,
+      "loss": 0.7828,
+      "mean_token_accuracy": 0.8131459752718607,
+      "num_tokens": 282590228.0,
+      "step": 1596
+    },
+    {
+      "entropy": 0.8098582600553831,
+      "epoch": 0.49633459371865113,
+      "grad_norm": 0.13313600420951843,
+      "learning_rate": 9.674497510697097e-05,
+      "loss": 0.7903,
+      "mean_token_accuracy": 0.8115353050331274,
+      "num_tokens": 284669419.0,
+      "step": 1608
+    },
+    {
+      "entropy": 0.805266530563434,
+      "epoch": 0.5000385832240142,
+      "grad_norm": 0.11492042243480682,
+      "learning_rate": 9.668779891329147e-05,
+      "loss": 0.7851,
+      "mean_token_accuracy": 0.8125738625725111,
+      "num_tokens": 286824939.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7950548666218916,
+      "epoch": 0.5037425727293773,
+      "grad_norm": 0.12213092297315598,
+      "learning_rate": 9.663014211323233e-05,
+      "loss": 0.7738,
+      "mean_token_accuracy": 0.8150150341292223,
+      "num_tokens": 288958154.0,
+      "step": 1632
+    },
+    {
+      "entropy": 0.7952163182199001,
+      "epoch": 0.5074465622347404,
+      "grad_norm": 0.11614470928907394,
+      "learning_rate": 9.657200530031533e-05,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.8137414579590162,
+      "num_tokens": 291092060.0,
+      "step": 1644
+    },
+    {
+      "entropy": 0.8064338949819406,
+      "epoch": 0.5111505517401034,
+      "grad_norm": 0.1288743019104004,
+      "learning_rate": 9.651338907300354e-05,
+      "loss": 0.7836,
+      "mean_token_accuracy": 0.8120651505887508,
+      "num_tokens": 293214294.0,
+      "step": 1656
+    },
+    {
+      "entropy": 0.8109964442749819,
+      "epoch": 0.5148545412454665,
+      "grad_norm": 0.12403728067874908,
+      "learning_rate": 9.645429403469512e-05,
+      "loss": 0.794,
+      "mean_token_accuracy": 0.8106129181881746,
+      "num_tokens": 295328150.0,
+      "step": 1668
+    },
+    {
+      "entropy": 0.8038183848063151,
+      "epoch": 0.5185585307508296,
+      "grad_norm": 0.12052307277917862,
+      "learning_rate": 9.639472079371717e-05,
+      "loss": 0.7841,
+      "mean_token_accuracy": 0.8124721460044384,
+      "num_tokens": 297455756.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7984983747204145,
+      "epoch": 0.5222625202561926,
+      "grad_norm": 0.12845173478126526,
+      "learning_rate": 9.63346699633194e-05,
+      "loss": 0.7794,
+      "mean_token_accuracy": 0.8130777689317862,
+      "num_tokens": 299602538.0,
+      "step": 1692
+    },
+    {
+      "entropy": 0.7976608164608479,
+      "epoch": 0.5259665097615557,
+      "grad_norm": 0.1210513561964035,
+      "learning_rate": 9.627414216166787e-05,
+      "loss": 0.7789,
+      "mean_token_accuracy": 0.8136931844055653,
+      "num_tokens": 301711090.0,
+      "step": 1704
+    },
+    {
+      "entropy": 0.787033441166083,
+      "epoch": 0.5296704992669188,
+      "grad_norm": 0.11822398006916046,
+      "learning_rate": 9.621313801183858e-05,
+      "loss": 0.7655,
+      "mean_token_accuracy": 0.8165569653113683,
+      "num_tokens": 303823411.0,
+      "step": 1716
+    },
+    {
+      "entropy": 0.7807075269520283,
+      "epoch": 0.5333744887722818,
+      "grad_norm": 0.1327759176492691,
+      "learning_rate": 9.61516581418111e-05,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.8171720070143541,
+      "num_tokens": 305981408.0,
+      "step": 1728
+    },
+    {
+      "entropy": 0.7996973978976408,
+      "epoch": 0.5370784782776449,
+      "grad_norm": 0.12808099389076233,
+      "learning_rate": 9.608970318446208e-05,
+      "loss": 0.7815,
+      "mean_token_accuracy": 0.8129920872549216,
+      "num_tokens": 308103143.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7994385659694672,
+      "epoch": 0.540782467783008,
+      "grad_norm": 0.12494061887264252,
+      "learning_rate": 9.602727377755875e-05,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.8130986665685972,
+      "num_tokens": 310257572.0,
+      "step": 1752
+    },
+    {
+      "entropy": 0.7863332827885946,
+      "epoch": 0.544486457288371,
+      "grad_norm": 0.11875750869512558,
+      "learning_rate": 9.596437056375231e-05,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.8158110665778319,
+      "num_tokens": 312373153.0,
+      "step": 1764
+    },
+    {
+      "entropy": 0.7789370119571686,
+      "epoch": 0.5481904467937341,
+      "grad_norm": 0.12376753985881805,
+      "learning_rate": 9.590099419057141e-05,
+      "loss": 0.7592,
+      "mean_token_accuracy": 0.8181358501315117,
+      "num_tokens": 314480636.0,
+      "step": 1776
+    },
+    {
+      "entropy": 0.7844561214248339,
+      "epoch": 0.5518944362990972,
+      "grad_norm": 0.13164328038692474,
+      "learning_rate": 9.583714531041538e-05,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.8156558784345785,
+      "num_tokens": 316566703.0,
+      "step": 1788
+    },
+    {
+      "entropy": 0.7855853562553724,
+      "epoch": 0.5555984258044602,
+      "grad_norm": 0.13665804266929626,
+      "learning_rate": 9.577282458054755e-05,
+      "loss": 0.7639,
+      "mean_token_accuracy": 0.8170899252096812,
+      "num_tokens": 318671313.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7836192175745964,
+      "epoch": 0.5593024153098233,
+      "grad_norm": 0.12808938324451447,
+      "learning_rate": 9.570803266308854e-05,
+      "loss": 0.7644,
+      "mean_token_accuracy": 0.8164806043108305,
+      "num_tokens": 320785587.0,
+      "step": 1812
+    },
+    {
+      "entropy": 0.7717621214687824,
+      "epoch": 0.5630064048151864,
+      "grad_norm": 0.11582314968109131,
+      "learning_rate": 9.564277022500936e-05,
+      "loss": 0.753,
+      "mean_token_accuracy": 0.8190883584320545,
+      "num_tokens": 322901535.0,
+      "step": 1824
+    },
+    {
+      "entropy": 0.7969209514558315,
+      "epoch": 0.5667103943205494,
+      "grad_norm": 0.1297702044248581,
+      "learning_rate": 9.557703793812458e-05,
+      "loss": 0.7776,
+      "mean_token_accuracy": 0.8134247288107872,
+      "num_tokens": 325009488.0,
+      "step": 1836
+    },
+    {
+      "entropy": 0.7894351184368134,
+      "epoch": 0.5704143838259125,
+      "grad_norm": 0.13040970265865326,
+      "learning_rate": 9.551083647908546e-05,
+      "loss": 0.7707,
+      "mean_token_accuracy": 0.8147525365153948,
+      "num_tokens": 327151753.0,
+      "step": 1848
+    },
+    {
+      "entropy": 0.7986459036668142,
+      "epoch": 0.5741183733312756,
+      "grad_norm": 0.12837360799312592,
+      "learning_rate": 9.544416652937287e-05,
+      "loss": 0.7798,
+      "mean_token_accuracy": 0.8129371106624603,
+      "num_tokens": 329288710.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.782096286614736,
+      "epoch": 0.5778223628366387,
+      "grad_norm": 0.1294112354516983,
+      "learning_rate": 9.53770287752904e-05,
+      "loss": 0.7621,
+      "mean_token_accuracy": 0.8168822092314562,
+      "num_tokens": 331435486.0,
+      "step": 1872
+    },
+    {
+      "entropy": 0.802633331467708,
+      "epoch": 0.5815263523420017,
+      "grad_norm": 0.1259497106075287,
+      "learning_rate": 9.53094239079572e-05,
+      "loss": 0.7828,
+      "mean_token_accuracy": 0.8125941964487234,
+      "num_tokens": 333550517.0,
+      "step": 1884
+    },
+    {
+      "entropy": 0.7907603432734808,
+      "epoch": 0.5852303418473648,
+      "grad_norm": 0.1232752725481987,
+      "learning_rate": 9.524135262330098e-05,
+      "loss": 0.7692,
+      "mean_token_accuracy": 0.8152645404140154,
+      "num_tokens": 335680200.0,
+      "step": 1896
+    },
+    {
+      "entropy": 0.785350481669108,
+      "epoch": 0.5889343313527279,
+      "grad_norm": 0.12669149041175842,
+      "learning_rate": 9.517281562205067e-05,
+      "loss": 0.769,
+      "mean_token_accuracy": 0.8154158430794874,
+      "num_tokens": 337807097.0,
+      "step": 1908
+    },
+    {
+      "entropy": 0.7851421398421129,
+      "epoch": 0.5926383208580909,
+      "grad_norm": 0.1276937574148178,
+      "learning_rate": 9.510381360972938e-05,
+      "loss": 0.7641,
+      "mean_token_accuracy": 0.8159417261679968,
+      "num_tokens": 339954087.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7850158450504144,
+      "epoch": 0.596342310363454,
+      "grad_norm": 0.132017120718956,
+      "learning_rate": 9.503434729664705e-05,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.8161118167142073,
+      "num_tokens": 342092293.0,
+      "step": 1932
+    },
+    {
+      "entropy": 0.7878543138504028,
+      "epoch": 0.6000462998688171,
+      "grad_norm": 0.12451501935720444,
+      "learning_rate": 9.49644173978931e-05,
+      "loss": 0.7704,
+      "mean_token_accuracy": 0.8147625786562761,
+      "num_tokens": 344240873.0,
+      "step": 1944
+    },
+    {
+      "entropy": 0.7971215297778448,
+      "epoch": 0.6037502893741801,
+      "grad_norm": 0.12397521734237671,
+      "learning_rate": 9.489402463332923e-05,
+      "loss": 0.7757,
+      "mean_token_accuracy": 0.814307109763225,
+      "num_tokens": 346363936.0,
+      "step": 1956
+    },
+    {
+      "entropy": 0.7801007392505804,
+      "epoch": 0.6074542788795432,
+      "grad_norm": 0.12504300475120544,
+      "learning_rate": 9.482316972758181e-05,
+      "loss": 0.7628,
+      "mean_token_accuracy": 0.8172982657949129,
+      "num_tokens": 348481250.0,
+      "step": 1968
+    },
+    {
+      "entropy": 0.7930209897458553,
+      "epoch": 0.6111582683849063,
+      "grad_norm": 0.1284765601158142,
+      "learning_rate": 9.475185341003455e-05,
+      "loss": 0.7747,
+      "mean_token_accuracy": 0.8143445054690043,
+      "num_tokens": 350608362.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7802144425610701,
+      "epoch": 0.6148622578902693,
+      "grad_norm": 0.12513010203838348,
+      "learning_rate": 9.468007641482094e-05,
+      "loss": 0.7647,
+      "mean_token_accuracy": 0.8167619270582994,
+      "num_tokens": 352709516.0,
+      "step": 1992
+    },
+    {
+      "entropy": 0.7944958060979843,
+      "epoch": 0.6185662473956324,
+      "grad_norm": 0.13269025087356567,
+      "learning_rate": 9.460783948081675e-05,
+      "loss": 0.7739,
+      "mean_token_accuracy": 0.8146844121317068,
+      "num_tokens": 354822759.0,
+      "step": 2004
+    },
+    {
+      "entropy": 0.7854583573838075,
+      "epoch": 0.6222702369009955,
+      "grad_norm": 0.1287117898464203,
+      "learning_rate": 9.453514335163231e-05,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.8166800700128078,
+      "num_tokens": 356911949.0,
+      "step": 2016
+    },
+    {
+      "entropy": 0.7777937439580759,
+      "epoch": 0.6259742264063585,
+      "grad_norm": 0.12762552499771118,
+      "learning_rate": 9.446198877560497e-05,
+      "loss": 0.7571,
+      "mean_token_accuracy": 0.8182835541665554,
+      "num_tokens": 359049122.0,
+      "step": 2028
+    },
+    {
+      "entropy": 0.7750151492655277,
+      "epoch": 0.6296782159117216,
+      "grad_norm": 0.1277306228876114,
+      "learning_rate": 9.438837650579137e-05,
+      "loss": 0.7553,
+      "mean_token_accuracy": 0.8175727687776089,
+      "num_tokens": 361210090.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7802457685271899,
+      "epoch": 0.6333822054170847,
+      "grad_norm": 0.12481208890676498,
+      "learning_rate": 9.431430729995963e-05,
+      "loss": 0.7598,
+      "mean_token_accuracy": 0.8178850611050924,
+      "num_tokens": 363337396.0,
+      "step": 2052
+    },
+    {
+      "entropy": 0.7862026058137417,
+      "epoch": 0.6370861949224477,
+      "grad_norm": 0.13092230260372162,
+      "learning_rate": 9.42397819205816e-05,
+      "loss": 0.7688,
+      "mean_token_accuracy": 0.8156153832872709,
+      "num_tokens": 365477630.0,
+      "step": 2064
+    },
+    {
+      "entropy": 0.7690726555883884,
+      "epoch": 0.6407901844278108,
+      "grad_norm": 0.11987863481044769,
+      "learning_rate": 9.416480113482504e-05,
+      "loss": 0.7521,
+      "mean_token_accuracy": 0.8195419311523438,
+      "num_tokens": 367643528.0,
+      "step": 2076
+    },
+    {
+      "entropy": 0.7762465241054693,
+      "epoch": 0.6444941739331739,
+      "grad_norm": 0.13100093603134155,
+      "learning_rate": 9.408936571454566e-05,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.8186973209182421,
+      "num_tokens": 369782358.0,
+      "step": 2088
+    },
+    {
+      "entropy": 0.7813232329984506,
+      "epoch": 0.648198163438537,
+      "grad_norm": 0.13405582308769226,
+      "learning_rate": 9.401347643627915e-05,
+      "loss": 0.7632,
+      "mean_token_accuracy": 0.8167420464257399,
+      "num_tokens": 371928210.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7875241699318091,
+      "epoch": 0.6519021529439,
+      "grad_norm": 0.12580129504203796,
+      "learning_rate": 9.393713408123332e-05,
+      "loss": 0.7699,
+      "mean_token_accuracy": 0.8153914275268713,
+      "num_tokens": 374058726.0,
+      "step": 2112
+    },
+    {
+      "entropy": 0.7834992570181688,
+      "epoch": 0.6556061424492631,
+      "grad_norm": 0.1281077116727829,
+      "learning_rate": 9.38603394352799e-05,
+      "loss": 0.7638,
+      "mean_token_accuracy": 0.8171246275305748,
+      "num_tokens": 376203701.0,
+      "step": 2124
+    },
+    {
+      "entropy": 0.7805231350163618,
+      "epoch": 0.6593101319546262,
+      "grad_norm": 0.14219819009304047,
+      "learning_rate": 9.378309328894662e-05,
+      "loss": 0.7616,
+      "mean_token_accuracy": 0.8166681937873363,
+      "num_tokens": 378335272.0,
+      "step": 2136
+    },
+    {
+      "entropy": 0.7874154051144918,
+      "epoch": 0.6630141214599892,
+      "grad_norm": 0.1304338425397873,
+      "learning_rate": 9.370539643740883e-05,
+      "loss": 0.7637,
+      "mean_token_accuracy": 0.8164051709075769,
+      "num_tokens": 380467266.0,
+      "step": 2148
+    },
+    {
+      "entropy": 0.7733189848562082,
+      "epoch": 0.6667181109653523,
+      "grad_norm": 0.13076983392238617,
+      "learning_rate": 9.36272496804816e-05,
+      "loss": 0.7571,
+      "mean_token_accuracy": 0.8175699549416701,
+      "num_tokens": 382603869.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.776911374181509,
+      "epoch": 0.6704221004707154,
+      "grad_norm": 0.13455283641815186,
+      "learning_rate": 9.354865382261128e-05,
+      "loss": 0.7563,
+      "mean_token_accuracy": 0.8178863686819872,
+      "num_tokens": 384767350.0,
+      "step": 2172
+    },
+    {
+      "entropy": 0.7755288705229759,
+      "epoch": 0.6741260899760784,
+      "grad_norm": 0.1210244670510292,
+      "learning_rate": 9.346960967286728e-05,
+      "loss": 0.7556,
+      "mean_token_accuracy": 0.8183015001316866,
+      "num_tokens": 386934062.0,
+      "step": 2184
+    },
+    {
+      "entropy": 0.7806575360397497,
+      "epoch": 0.6778300794814415,
+      "grad_norm": 0.1282687783241272,
+      "learning_rate": 9.339011804493378e-05,
+      "loss": 0.7604,
+      "mean_token_accuracy": 0.8167587071657181,
+      "num_tokens": 389047648.0,
+      "step": 2196
+    },
+    {
+      "entropy": 0.7758863245447477,
+      "epoch": 0.6815340689868046,
+      "grad_norm": 0.13144470751285553,
+      "learning_rate": 9.331017975710132e-05,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.8180744908750057,
+      "num_tokens": 391155544.0,
+      "step": 2208
+    },
+    {
+      "entropy": 0.7667204054693381,
+      "epoch": 0.6852380584921676,
+      "grad_norm": 0.12816905975341797,
+      "learning_rate": 9.322979563225833e-05,
+      "loss": 0.7464,
+      "mean_token_accuracy": 0.8206619794170061,
+      "num_tokens": 393293375.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.7751789751152197,
+      "epoch": 0.6889420479975307,
+      "grad_norm": 0.14497670531272888,
+      "learning_rate": 9.314896649788277e-05,
+      "loss": 0.7561,
+      "mean_token_accuracy": 0.8184558848539988,
+      "num_tokens": 395411494.0,
+      "step": 2232
+    },
+    {
+      "entropy": 0.7844450324773788,
+      "epoch": 0.6926460375028938,
+      "grad_norm": 0.13543701171875,
+      "learning_rate": 9.306769318603348e-05,
+      "loss": 0.765,
+      "mean_token_accuracy": 0.8167361902693907,
+      "num_tokens": 397518224.0,
+      "step": 2244
+    },
+    {
+      "entropy": 0.7686401257912318,
+      "epoch": 0.6963500270082568,
+      "grad_norm": 0.13837109506130219,
+      "learning_rate": 9.298597653334178e-05,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.8197805831829706,
+      "num_tokens": 399648867.0,
+      "step": 2256
+    },
+    {
+      "entropy": 0.7626761943101883,
+      "epoch": 0.7000540165136199,
+      "grad_norm": 0.1293450891971588,
+      "learning_rate": 9.290381738100265e-05,
+      "loss": 0.7441,
+      "mean_token_accuracy": 0.8214877719680468,
+      "num_tokens": 401743610.0,
+      "step": 2268
+    },
+    {
+      "entropy": 0.7687733148535093,
+      "epoch": 0.703758006018983,
+      "grad_norm": 0.1483583152294159,
+      "learning_rate": 9.282121657476627e-05,
+      "loss": 0.7506,
+      "mean_token_accuracy": 0.8195670247077942,
+      "num_tokens": 403850723.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7698837158580621,
+      "epoch": 0.707461995524346,
+      "grad_norm": 0.13035227358341217,
+      "learning_rate": 9.273817496492917e-05,
+      "loss": 0.7503,
+      "mean_token_accuracy": 0.8193908805648485,
+      "num_tokens": 405973443.0,
+      "step": 2292
+    },
+    {
+      "entropy": 0.770644503335158,
+      "epoch": 0.7111659850297091,
+      "grad_norm": 0.13103176653385162,
+      "learning_rate": 9.265469340632557e-05,
+      "loss": 0.7538,
+      "mean_token_accuracy": 0.8188917512694994,
+      "num_tokens": 408105892.0,
+      "step": 2304
+    },
+    {
+      "entropy": 0.7735109639664491,
+      "epoch": 0.7148699745350722,
+      "grad_norm": 0.1398514360189438,
+      "learning_rate": 9.257077275831853e-05,
+      "loss": 0.7518,
+      "mean_token_accuracy": 0.819186095148325,
+      "num_tokens": 410296037.0,
+      "step": 2316
+    },
+    {
+      "entropy": 0.771765373647213,
+      "epoch": 0.7185739640404352,
+      "grad_norm": 0.14083071053028107,
+      "learning_rate": 9.248641388479111e-05,
+      "loss": 0.7526,
+      "mean_token_accuracy": 0.8195289969444275,
+      "num_tokens": 412380775.0,
+      "step": 2328
+    },
+    {
+      "entropy": 0.7654011559983095,
+      "epoch": 0.7222779535457983,
+      "grad_norm": 0.1325056552886963,
+      "learning_rate": 9.240161765413748e-05,
+      "loss": 0.7457,
+      "mean_token_accuracy": 0.8208736081918081,
+      "num_tokens": 414535472.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7590547104676565,
+      "epoch": 0.7259819430511614,
+      "grad_norm": 0.1285925805568695,
+      "learning_rate": 9.231638493925402e-05,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.8210619514187177,
+      "num_tokens": 416686381.0,
+      "step": 2352
+    },
+    {
+      "entropy": 0.7812760807573795,
+      "epoch": 0.7296859325565245,
+      "grad_norm": 0.12959104776382446,
+      "learning_rate": 9.223071661753024e-05,
+      "loss": 0.7615,
+      "mean_token_accuracy": 0.8172919216255347,
+      "num_tokens": 418808712.0,
+      "step": 2364
+    },
+    {
+      "entropy": 0.7873292689522108,
+      "epoch": 0.7333899220618875,
+      "grad_norm": 0.1371845006942749,
+      "learning_rate": 9.214461357083985e-05,
+      "loss": 0.7677,
+      "mean_token_accuracy": 0.8155596914390723,
+      "num_tokens": 420927036.0,
+      "step": 2376
+    },
+    {
+      "entropy": 0.7580399426321188,
+      "epoch": 0.7370939115672506,
+      "grad_norm": 0.14509984850883484,
+      "learning_rate": 9.205807668553164e-05,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.8222843247155348,
+      "num_tokens": 423036030.0,
+      "step": 2388
+    },
+    {
+      "entropy": 0.7687507805724939,
+      "epoch": 0.7407979010726137,
+      "grad_norm": 0.15484097599983215,
+      "learning_rate": 9.197110685242034e-05,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.8196417490641276,
+      "num_tokens": 425187345.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.764864676942428,
+      "epoch": 0.7445018905779767,
+      "grad_norm": 0.12323542684316635,
+      "learning_rate": 9.188370496677745e-05,
+      "loss": 0.7452,
+      "mean_token_accuracy": 0.8201479675869147,
+      "num_tokens": 427331828.0,
+      "step": 2412
+    },
+    {
+      "entropy": 0.7658702706297239,
+      "epoch": 0.7482058800833398,
+      "grad_norm": 0.12649372220039368,
+      "learning_rate": 9.179587192832209e-05,
+      "loss": 0.7474,
+      "mean_token_accuracy": 0.8195582069456577,
+      "num_tokens": 429448072.0,
+      "step": 2424
+    },
+    {
+      "entropy": 0.7575242506961027,
+      "epoch": 0.7519098695887029,
+      "grad_norm": 0.13158433139324188,
+      "learning_rate": 9.170760864121162e-05,
+      "loss": 0.7366,
+      "mean_token_accuracy": 0.8224081955850124,
+      "num_tokens": 431533949.0,
+      "step": 2436
+    },
+    {
+      "entropy": 0.7688459927837054,
+      "epoch": 0.7556138590940659,
+      "grad_norm": 0.12840636074543,
+      "learning_rate": 9.161891601403245e-05,
+      "loss": 0.7487,
+      "mean_token_accuracy": 0.8195722003777822,
+      "num_tokens": 433658430.0,
+      "step": 2448
+    },
+    {
+      "entropy": 0.7660383395850658,
+      "epoch": 0.759317848599429,
+      "grad_norm": 0.12874674797058105,
+      "learning_rate": 9.152979495979063e-05,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.8193290382623672,
+      "num_tokens": 435741465.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7668347917497158,
+      "epoch": 0.7630218381047921,
+      "grad_norm": 0.12673811614513397,
+      "learning_rate": 9.144024639590245e-05,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.8204079742232958,
+      "num_tokens": 437908304.0,
+      "step": 2472
+    },
+    {
+      "entropy": 0.7687832986315092,
+      "epoch": 0.7667258276101551,
+      "grad_norm": 0.12845873832702637,
+      "learning_rate": 9.135027124418499e-05,
+      "loss": 0.7536,
+      "mean_token_accuracy": 0.8192412157853445,
+      "num_tokens": 440028485.0,
+      "step": 2484
+    },
+    {
+      "entropy": 0.7862997514506181,
+      "epoch": 0.7704298171155182,
+      "grad_norm": 0.13068512082099915,
+      "learning_rate": 9.125987043084665e-05,
+      "loss": 0.7642,
+      "mean_token_accuracy": 0.8162732509275278,
+      "num_tokens": 442179533.0,
+      "step": 2496
+    },
+    {
+      "entropy": 0.7685239054262638,
+      "epoch": 0.7741338066208813,
+      "grad_norm": 0.13230957090854645,
+      "learning_rate": 9.116904488647764e-05,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.8193069696426392,
+      "num_tokens": 444264359.0,
+      "step": 2508
+    },
+    {
+      "entropy": 0.76015779748559,
+      "epoch": 0.7778377961262443,
+      "grad_norm": 0.13136623799800873,
+      "learning_rate": 9.107779554604035e-05,
+      "loss": 0.7418,
+      "mean_token_accuracy": 0.8217526078224182,
+      "num_tokens": 446400739.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7768492065370083,
+      "epoch": 0.7815417856316074,
+      "grad_norm": 0.13927006721496582,
+      "learning_rate": 9.098612334885972e-05,
+      "loss": 0.7581,
+      "mean_token_accuracy": 0.8174782631297907,
+      "num_tokens": 448488906.0,
+      "step": 2532
+    },
+    {
+      "entropy": 0.7547270730137825,
+      "epoch": 0.7852457751369705,
+      "grad_norm": 0.14124959707260132,
+      "learning_rate": 9.089402923861366e-05,
+      "loss": 0.7356,
+      "mean_token_accuracy": 0.8228224813938141,
+      "num_tokens": 450609731.0,
+      "step": 2544
+    },
+    {
+      "entropy": 0.7677095470329126,
+      "epoch": 0.7889497646423335,
+      "grad_norm": 0.14042387902736664,
+      "learning_rate": 9.080151416332319e-05,
+      "loss": 0.747,
+      "mean_token_accuracy": 0.8197631984949112,
+      "num_tokens": 452710661.0,
+      "step": 2556
+    },
+    {
+      "entropy": 0.768154501914978,
+      "epoch": 0.7926537541476966,
+      "grad_norm": 0.12782582640647888,
+      "learning_rate": 9.070857907534287e-05,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.8194541210929552,
+      "num_tokens": 454884003.0,
+      "step": 2568
+    },
+    {
+      "entropy": 0.7609333768486977,
+      "epoch": 0.7963577436530597,
+      "grad_norm": 0.141867533326149,
+      "learning_rate": 9.061522493135079e-05,
+      "loss": 0.7407,
+      "mean_token_accuracy": 0.82136203845342,
+      "num_tokens": 457030737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7599871419370174,
+      "epoch": 0.8000617331584228,
+      "grad_norm": 0.13808685541152954,
+      "learning_rate": 9.052145269233887e-05,
+      "loss": 0.7402,
+      "mean_token_accuracy": 0.8215218285719553,
+      "num_tokens": 459177274.0,
+      "step": 2592
+    },
+    {
+      "entropy": 0.7648821386198202,
+      "epoch": 0.8037657226637858,
+      "grad_norm": 0.15076588094234467,
+      "learning_rate": 9.042726332360292e-05,
+      "loss": 0.7454,
+      "mean_token_accuracy": 0.8200170012811819,
+      "num_tokens": 461306463.0,
+      "step": 2604
+    },
+    {
+      "entropy": 0.7687475743393103,
+      "epoch": 0.8074697121691489,
+      "grad_norm": 0.12965603172779083,
+      "learning_rate": 9.033265779473268e-05,
+      "loss": 0.7492,
+      "mean_token_accuracy": 0.819452028721571,
+      "num_tokens": 463455408.0,
+      "step": 2616
+    },
+    {
+      "entropy": 0.7646373882889748,
+      "epoch": 0.811173701674512,
+      "grad_norm": 0.1345282644033432,
+      "learning_rate": 9.023763707960188e-05,
+      "loss": 0.7434,
+      "mean_token_accuracy": 0.8209528836111227,
+      "num_tokens": 465550715.0,
+      "step": 2628
+    },
+    {
+      "entropy": 0.76267567401131,
+      "epoch": 0.814877691179875,
+      "grad_norm": 0.1315702497959137,
+      "learning_rate": 9.01422021563582e-05,
+      "loss": 0.7427,
+      "mean_token_accuracy": 0.820941454420487,
+      "num_tokens": 467676160.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7417604538301626,
+      "epoch": 0.8185816806852381,
+      "grad_norm": 0.12949636578559875,
+      "learning_rate": 9.00463540074132e-05,
+      "loss": 0.7243,
+      "mean_token_accuracy": 0.8254370614886284,
+      "num_tokens": 469811393.0,
+      "step": 2652
+    },
+    {
+      "entropy": 0.7587257462243239,
+      "epoch": 0.8222856701906012,
+      "grad_norm": 0.13278073072433472,
+      "learning_rate": 8.995009361943218e-05,
+      "loss": 0.7384,
+      "mean_token_accuracy": 0.8231843349834284,
+      "num_tokens": 471930105.0,
+      "step": 2664
+    },
+    {
+      "entropy": 0.7431893559793631,
+      "epoch": 0.8259896596959642,
+      "grad_norm": 0.14032749831676483,
+      "learning_rate": 8.985342198332407e-05,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.8248631035288175,
+      "num_tokens": 474075700.0,
+      "step": 2676
+    },
+    {
+      "entropy": 0.7719796399275461,
+      "epoch": 0.8296936492013273,
+      "grad_norm": 0.14027121663093567,
+      "learning_rate": 8.975634009423122e-05,
+      "loss": 0.752,
+      "mean_token_accuracy": 0.8185311630368233,
+      "num_tokens": 476204230.0,
+      "step": 2688
+    },
+    {
+      "entropy": 0.7616085906823477,
+      "epoch": 0.8333976387066904,
+      "grad_norm": 0.1393400877714157,
+      "learning_rate": 8.965884895151908e-05,
+      "loss": 0.7443,
+      "mean_token_accuracy": 0.8211442058285078,
+      "num_tokens": 478348559.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7600347759823004,
+      "epoch": 0.8371016282120534,
+      "grad_norm": 0.1410454660654068,
+      "learning_rate": 8.956094955876607e-05,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.8219936129947504,
+      "num_tokens": 480446587.0,
+      "step": 2712
+    },
+    {
+      "entropy": 0.7613047709067663,
+      "epoch": 0.8408056177174165,
+      "grad_norm": 0.140048548579216,
+      "learning_rate": 8.946264292375306e-05,
+      "loss": 0.7426,
+      "mean_token_accuracy": 0.8213416139284769,
+      "num_tokens": 482581346.0,
+      "step": 2724
+    },
+    {
+      "entropy": 0.7594624037543932,
+      "epoch": 0.8445096072227796,
+      "grad_norm": 0.1277952641248703,
+      "learning_rate": 8.936393005845316e-05,
+      "loss": 0.739,
+      "mean_token_accuracy": 0.8218294816712538,
+      "num_tokens": 484695228.0,
+      "step": 2736
+    },
+    {
+      "entropy": 0.7544240554173788,
+      "epoch": 0.8482135967281426,
+      "grad_norm": 0.154397651553154,
+      "learning_rate": 8.926481197902122e-05,
+      "loss": 0.7341,
+      "mean_token_accuracy": 0.8225894048810005,
+      "num_tokens": 486862761.0,
+      "step": 2748
+    },
+    {
+      "entropy": 0.7661119649807612,
+      "epoch": 0.8519175862335057,
+      "grad_norm": 0.14195363223552704,
+      "learning_rate": 8.916528970578333e-05,
+      "loss": 0.746,
+      "mean_token_accuracy": 0.8203906156122684,
+      "num_tokens": 489014996.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7686087116599083,
+      "epoch": 0.8556215757388688,
+      "grad_norm": 0.13611237704753876,
+      "learning_rate": 8.906536426322646e-05,
+      "loss": 0.7473,
+      "mean_token_accuracy": 0.8198287424941858,
+      "num_tokens": 491157432.0,
+      "step": 2772
+    },
+    {
+      "entropy": 0.7590742185711861,
+      "epoch": 0.8593255652442318,
+      "grad_norm": 0.12830516695976257,
+      "learning_rate": 8.896503667998777e-05,
+      "loss": 0.7414,
+      "mean_token_accuracy": 0.8214252628386021,
+      "num_tokens": 493327333.0,
+      "step": 2784
+    },
+    {
+      "entropy": 0.758670142541329,
+      "epoch": 0.8630295547495949,
+      "grad_norm": 0.13577738404273987,
+      "learning_rate": 8.886430798884406e-05,
+      "loss": 0.74,
+      "mean_token_accuracy": 0.8217101270953814,
+      "num_tokens": 495441259.0,
+      "step": 2796
+    },
+    {
+      "entropy": 0.7604162829617659,
+      "epoch": 0.866733544254958,
+      "grad_norm": 0.1389269232749939,
+      "learning_rate": 8.876317922670119e-05,
+      "loss": 0.7385,
+      "mean_token_accuracy": 0.8216200309495131,
+      "num_tokens": 497558061.0,
+      "step": 2808
+    },
+    {
+      "entropy": 0.7654056176543236,
+      "epoch": 0.870437533760321,
+      "grad_norm": 0.1383410394191742,
+      "learning_rate": 8.866165143458334e-05,
+      "loss": 0.7493,
+      "mean_token_accuracy": 0.819673765450716,
+      "num_tokens": 499689660.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7578516465922197,
+      "epoch": 0.8741415232656841,
+      "grad_norm": 0.144080251455307,
+      "learning_rate": 8.855972565762236e-05,
+      "loss": 0.7373,
+      "mean_token_accuracy": 0.822392825037241,
+      "num_tokens": 501811384.0,
+      "step": 2832
+    },
+    {
+      "entropy": 0.7520838553706805,
+      "epoch": 0.8778455127710472,
+      "grad_norm": 0.1445675939321518,
+      "learning_rate": 8.845740294504691e-05,
+      "loss": 0.7321,
+      "mean_token_accuracy": 0.8238844747344652,
+      "num_tokens": 503931258.0,
+      "step": 2844
+    },
+    {
+      "entropy": 0.7716477451225122,
+      "epoch": 0.8815495022764103,
+      "grad_norm": 0.13848181068897247,
+      "learning_rate": 8.835468435017183e-05,
+      "loss": 0.75,
+      "mean_token_accuracy": 0.8195607153077921,
+      "num_tokens": 506070704.0,
+      "step": 2856
+    },
+    {
+      "entropy": 0.7465110706786314,
+      "epoch": 0.8852534917817733,
+      "grad_norm": 0.1347406953573227,
+      "learning_rate": 8.825157093038708e-05,
+      "loss": 0.7295,
+      "mean_token_accuracy": 0.8236632930735747,
+      "num_tokens": 508203294.0,
+      "step": 2868
+    },
+    {
+      "entropy": 0.7651778521637121,
+      "epoch": 0.8889574812871364,
+      "grad_norm": 0.13069529831409454,
+      "learning_rate": 8.814806374714702e-05,
+      "loss": 0.7449,
+      "mean_token_accuracy": 0.8204664116104444,
+      "num_tokens": 510395477.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7543973810970783,
+      "epoch": 0.8926614707924995,
+      "grad_norm": 0.12907341122627258,
+      "learning_rate": 8.804416386595943e-05,
+      "loss": 0.733,
+      "mean_token_accuracy": 0.8230690496663252,
+      "num_tokens": 512509055.0,
+      "step": 2892
+    },
+    {
+      "entropy": 0.7565698722998301,
+      "epoch": 0.8963654602978625,
+      "grad_norm": 0.14323952794075012,
+      "learning_rate": 8.793987235637453e-05,
+      "loss": 0.7376,
+      "mean_token_accuracy": 0.8218328369160494,
+      "num_tokens": 514596287.0,
+      "step": 2904
+    },
+    {
+      "entropy": 0.7765912314256033,
+      "epoch": 0.9000694498032256,
+      "grad_norm": 0.13393579423427582,
+      "learning_rate": 8.783519029197398e-05,
+      "loss": 0.7567,
+      "mean_token_accuracy": 0.8180692394574484,
+      "num_tokens": 516697723.0,
+      "step": 2916
+    },
+    {
+      "entropy": 0.748100645840168,
+      "epoch": 0.9037734393085887,
+      "grad_norm": 0.1474265605211258,
+      "learning_rate": 8.773011875035983e-05,
+      "loss": 0.7294,
+      "mean_token_accuracy": 0.8240405097603798,
+      "num_tokens": 518833866.0,
+      "step": 2928
+    },
+    {
+      "entropy": 0.7694602260986964,
+      "epoch": 0.9074774288139517,
+      "grad_norm": 0.14337094128131866,
+      "learning_rate": 8.762465881314346e-05,
+      "loss": 0.7481,
+      "mean_token_accuracy": 0.8199893161654472,
+      "num_tokens": 520946745.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7642699517309666,
+      "epoch": 0.9111814183193148,
+      "grad_norm": 0.1440202295780182,
+      "learning_rate": 8.751881156593434e-05,
+      "loss": 0.7471,
+      "mean_token_accuracy": 0.8201769590377808,
+      "num_tokens": 523082474.0,
+      "step": 2952
+    },
+    {
+      "entropy": 0.7573205133279165,
+      "epoch": 0.9148854078246779,
+      "grad_norm": 0.14229246973991394,
+      "learning_rate": 8.7412578098329e-05,
+      "loss": 0.7388,
+      "mean_token_accuracy": 0.8218303211033344,
+      "num_tokens": 525230014.0,
+      "step": 2964
+    },
+    {
+      "entropy": 0.774663812170426,
+      "epoch": 0.9185893973300409,
+      "grad_norm": 0.1373312920331955,
+      "learning_rate": 8.730595950389968e-05,
+      "loss": 0.7537,
+      "mean_token_accuracy": 0.8188973863919576,
+      "num_tokens": 527358964.0,
+      "step": 2976
+    },
+    {
+      "entropy": 0.7440223582088947,
+      "epoch": 0.922293386835404,
+      "grad_norm": 0.13869309425354004,
+      "learning_rate": 8.71989568801832e-05,
+      "loss": 0.7243,
+      "mean_token_accuracy": 0.8259313181042671,
+      "num_tokens": 529446921.0,
+      "step": 2988
+    },
+    {
+      "entropy": 0.7558054054776827,
+      "epoch": 0.9259973763407671,
+      "grad_norm": 0.14386014640331268,
+      "learning_rate": 8.709157132866954e-05,
+      "loss": 0.7371,
+      "mean_token_accuracy": 0.8218563583989938,
+      "num_tokens": 531572822.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.7559870270391306,
+      "epoch": 0.9297013658461301,
+      "grad_norm": 0.14299409091472626,
+      "learning_rate": 8.698380395479058e-05,
+      "loss": 0.7346,
+      "mean_token_accuracy": 0.8229275730748972,
+      "num_tokens": 533671253.0,
+      "step": 3012
+    },
+    {
+      "entropy": 0.7501622214913368,
+      "epoch": 0.9334053553514932,
+      "grad_norm": 0.1446056067943573,
+      "learning_rate": 8.68756558679087e-05,
+      "loss": 0.7307,
+      "mean_token_accuracy": 0.8238628643254439,
+      "num_tokens": 535799339.0,
+      "step": 3024
+    },
+    {
+      "entropy": 0.7404041352371374,
+      "epoch": 0.9371093448568563,
+      "grad_norm": 0.14985917508602142,
+      "learning_rate": 8.676712818130534e-05,
+      "loss": 0.7227,
+      "mean_token_accuracy": 0.8250208596388499,
+      "num_tokens": 537936260.0,
+      "step": 3036
+    },
+    {
+      "entropy": 0.7513143469889959,
+      "epoch": 0.9408133343622193,
+      "grad_norm": 0.13818083703517914,
+      "learning_rate": 8.665822201216958e-05,
+      "loss": 0.7306,
+      "mean_token_accuracy": 0.8236184107760588,
+      "num_tokens": 540092247.0,
+      "step": 3048
+    },
+    {
+      "entropy": 0.7599398357172807,
+      "epoch": 0.9445173238675824,
+      "grad_norm": 0.14524659514427185,
+      "learning_rate": 8.654893848158658e-05,
+      "loss": 0.7398,
+      "mean_token_accuracy": 0.8216308690607548,
+      "num_tokens": 542253810.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.7556075366834799,
+      "epoch": 0.9482213133729455,
+      "grad_norm": 0.14592809975147247,
+      "learning_rate": 8.643927871452611e-05,
+      "loss": 0.7387,
+      "mean_token_accuracy": 0.8219700244565805,
+      "num_tokens": 544371064.0,
+      "step": 3072
+    },
+    {
+      "entropy": 0.7573518306016922,
+      "epoch": 0.9519253028783086,
+      "grad_norm": 0.136802539229393,
+      "learning_rate": 8.632924383983096e-05,
+      "loss": 0.7383,
+      "mean_token_accuracy": 0.8221282474696636,
+      "num_tokens": 546525688.0,
+      "step": 3084
+    },
+    {
+      "entropy": 0.760393563657999,
+      "epoch": 0.9556292923836716,
+      "grad_norm": 0.14470767974853516,
+      "learning_rate": 8.621883499020523e-05,
+      "loss": 0.7405,
+      "mean_token_accuracy": 0.8215649748841921,
+      "num_tokens": 548650119.0,
+      "step": 3096
+    },
+    {
+      "entropy": 0.7766042277216911,
+      "epoch": 0.9593332818890347,
+      "grad_norm": 0.15178759396076202,
+      "learning_rate": 8.610805330220275e-05,
+      "loss": 0.755,
+      "mean_token_accuracy": 0.818018895884355,
+      "num_tokens": 550763680.0,
+      "step": 3108
+    },
+    {
+      "entropy": 0.7451968391736349,
+      "epoch": 0.9630372713943978,
+      "grad_norm": 0.16293193399906158,
+      "learning_rate": 8.599689991621543e-05,
+      "loss": 0.7257,
+      "mean_token_accuracy": 0.8251891148587068,
+      "num_tokens": 552911674.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7404080505172411,
+      "epoch": 0.9667412608997608,
+      "grad_norm": 0.14965808391571045,
+      "learning_rate": 8.588537597646139e-05,
+      "loss": 0.7215,
+      "mean_token_accuracy": 0.8259551251928011,
+      "num_tokens": 555010287.0,
+      "step": 3132
+    },
+    {
+      "entropy": 0.7473993599414825,
+      "epoch": 0.9704452504051239,
+      "grad_norm": 0.13951192796230316,
+      "learning_rate": 8.577348263097324e-05,
+      "loss": 0.7281,
+      "mean_token_accuracy": 0.8238498754799366,
+      "num_tokens": 557140449.0,
+      "step": 3144
+    },
+    {
+      "entropy": 0.749595433473587,
+      "epoch": 0.974149239910487,
+      "grad_norm": 0.14272676408290863,
+      "learning_rate": 8.566122103158636e-05,
+      "loss": 0.7301,
+      "mean_token_accuracy": 0.8237165659666061,
+      "num_tokens": 559258810.0,
+      "step": 3156
+    },
+    {
+      "entropy": 0.7361926498512427,
+      "epoch": 0.97785322941585,
+      "grad_norm": 0.14749550819396973,
+      "learning_rate": 8.554859233392682e-05,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.8265771567821503,
+      "num_tokens": 561389026.0,
+      "step": 3168
+    },
+    {
+      "entropy": 0.7617857754230499,
+      "epoch": 0.9815572189212131,
+      "grad_norm": 0.14244970679283142,
+      "learning_rate": 8.543559769739974e-05,
+      "loss": 0.7436,
+      "mean_token_accuracy": 0.8205315048495928,
+      "num_tokens": 563506691.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7460702173411846,
+      "epoch": 0.9852612084265762,
+      "grad_norm": 0.15570423007011414,
+      "learning_rate": 8.532223828517716e-05,
+      "loss": 0.7265,
+      "mean_token_accuracy": 0.8244366881748041,
+      "num_tokens": 565591234.0,
+      "step": 3192
+    },
+    {
+      "entropy": 0.7432848749061426,
+      "epoch": 0.9889651979319392,
+      "grad_norm": 0.1433694213628769,
+      "learning_rate": 8.520851526418614e-05,
+      "loss": 0.7252,
+      "mean_token_accuracy": 0.8244002535939217,
+      "num_tokens": 567734464.0,
+      "step": 3204
+    },
+    {
+      "entropy": 0.7543482234080633,
+      "epoch": 0.9926691874373023,
+      "grad_norm": 0.1491028368473053,
+      "learning_rate": 8.509442980509678e-05,
+      "loss": 0.7326,
+      "mean_token_accuracy": 0.8236272583405176,
+      "num_tokens": 569859027.0,
+      "step": 3216
+    },
+    {
+      "entropy": 0.7251827741662661,
+      "epoch": 0.9963731769426654,
+      "grad_norm": 0.14663253724575043,
+      "learning_rate": 8.497998308231012e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.8289556242525578,
+      "num_tokens": 571982689.0,
+      "step": 3228
+    },
+    {
+      "entropy": 0.7527169965683146,
+      "epoch": 1.0,
+      "grad_norm": 0.1720859855413437,
+      "learning_rate": 8.486517627394606e-05,
+      "loss": 0.7331,
+      "mean_token_accuracy": 0.823117880111045,
+      "num_tokens": 574035358.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7543971252938112,
+      "epoch": 1.003703989505363,
+      "grad_norm": 0.14148341119289398,
+      "learning_rate": 8.475001056183124e-05,
+      "loss": 0.731,
+      "mean_token_accuracy": 0.8228112831711769,
+      "num_tokens": 576181278.0,
+      "step": 3252
+    },
+    {
+      "entropy": 0.7378036280473074,
+      "epoch": 1.0074079790107262,
+      "grad_norm": 0.13494503498077393,
+      "learning_rate": 8.463448713148687e-05,
+      "loss": 0.7178,
+      "mean_token_accuracy": 0.825627734263738,
+      "num_tokens": 578356052.0,
+      "step": 3264
+    },
+    {
+      "entropy": 0.7363200634717941,
+      "epoch": 1.0111119685160892,
+      "grad_norm": 0.14111706614494324,
+      "learning_rate": 8.451860717211653e-05,
+      "loss": 0.7176,
+      "mean_token_accuracy": 0.8264128789305687,
+      "num_tokens": 580483875.0,
+      "step": 3276
+    },
+    {
+      "entropy": 0.7324805557727814,
+      "epoch": 1.0148159580214522,
+      "grad_norm": 0.14493153989315033,
+      "learning_rate": 8.440237187659391e-05,
+      "loss": 0.7122,
+      "mean_token_accuracy": 0.8274283918241659,
+      "num_tokens": 582624687.0,
+      "step": 3288
+    },
+    {
+      "entropy": 0.7307334840297699,
+      "epoch": 1.0185199475268154,
+      "grad_norm": 0.14563046395778656,
+      "learning_rate": 8.42857824414506e-05,
+      "loss": 0.7103,
+      "mean_token_accuracy": 0.8277510106563568,
+      "num_tokens": 584748734.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7457285039126873,
+      "epoch": 1.0222239370321784,
+      "grad_norm": 0.15820999443531036,
+      "learning_rate": 8.416884006686366e-05,
+      "loss": 0.7266,
+      "mean_token_accuracy": 0.8236699538926283,
+      "num_tokens": 586852542.0,
+      "step": 3312
+    },
+    {
+      "entropy": 0.7318388596177101,
+      "epoch": 1.0259279265375414,
+      "grad_norm": 0.15957045555114746,
+      "learning_rate": 8.405154595664332e-05,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.8276907838881016,
+      "num_tokens": 588984878.0,
+      "step": 3324
+    },
+    {
+      "entropy": 0.7418795588115851,
+      "epoch": 1.0296319160429046,
+      "grad_norm": 0.14202064275741577,
+      "learning_rate": 8.39339013182207e-05,
+      "loss": 0.7207,
+      "mean_token_accuracy": 0.8251368477940559,
+      "num_tokens": 591095187.0,
+      "step": 3336
+    },
+    {
+      "entropy": 0.7447669468820095,
+      "epoch": 1.0333359055482676,
+      "grad_norm": 0.15226991474628448,
+      "learning_rate": 8.381590736263512e-05,
+      "loss": 0.724,
+      "mean_token_accuracy": 0.8251153019567331,
+      "num_tokens": 593222885.0,
+      "step": 3348
+    },
+    {
+      "entropy": 0.7374236173927784,
+      "epoch": 1.0370398950536306,
+      "grad_norm": 0.14633263647556305,
+      "learning_rate": 8.369756530452191e-05,
+      "loss": 0.7177,
+      "mean_token_accuracy": 0.8263744947810968,
+      "num_tokens": 595340453.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7381995966037115,
+      "epoch": 1.0407438845589938,
+      "grad_norm": 0.1505565196275711,
+      "learning_rate": 8.35788763620997e-05,
+      "loss": 0.7174,
+      "mean_token_accuracy": 0.8263110890984535,
+      "num_tokens": 597449422.0,
+      "step": 3372
+    },
+    {
+      "entropy": 0.7414120820661386,
+      "epoch": 1.0444478740643568,
+      "grad_norm": 0.143823504447937,
+      "learning_rate": 8.345984175715802e-05,
+      "loss": 0.7195,
+      "mean_token_accuracy": 0.8251637890934944,
+      "num_tokens": 599594017.0,
+      "step": 3384
+    },
+    {
+      "entropy": 0.7337635308504105,
+      "epoch": 1.0481518635697198,
+      "grad_norm": 0.14950762689113617,
+      "learning_rate": 8.334046271504465e-05,
+      "loss": 0.7151,
+      "mean_token_accuracy": 0.8268262160321077,
+      "num_tokens": 601723800.0,
+      "step": 3396
+    },
+    {
+      "entropy": 0.737778523315986,
+      "epoch": 1.051855853075083,
+      "grad_norm": 0.14158302545547485,
+      "learning_rate": 8.3220740464653e-05,
+      "loss": 0.7155,
+      "mean_token_accuracy": 0.8266446044047674,
+      "num_tokens": 603851467.0,
+      "step": 3408
+    },
+    {
+      "entropy": 0.7279616557061672,
+      "epoch": 1.055559842580446,
+      "grad_norm": 0.14092348515987396,
+      "learning_rate": 8.310067623840951e-05,
+      "loss": 0.7091,
+      "mean_token_accuracy": 0.8280529901385307,
+      "num_tokens": 605974553.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7397043655316035,
+      "epoch": 1.059263832085809,
+      "grad_norm": 0.1458451896905899,
+      "learning_rate": 8.298027127226093e-05,
+      "loss": 0.7186,
+      "mean_token_accuracy": 0.8254076987504959,
+      "num_tokens": 608111416.0,
+      "step": 3432
+    },
+    {
+      "entropy": 0.7327801200250784,
+      "epoch": 1.0629678215911722,
+      "grad_norm": 0.14519493281841278,
+      "learning_rate": 8.28595268056616e-05,
+      "loss": 0.7121,
+      "mean_token_accuracy": 0.8273726465801398,
+      "num_tokens": 610224373.0,
+      "step": 3444
+    },
+    {
+      "entropy": 0.7508813291788101,
+      "epoch": 1.0666718110965352,
+      "grad_norm": 0.1475428193807602,
+      "learning_rate": 8.273844408156066e-05,
+      "loss": 0.7283,
+      "mean_token_accuracy": 0.8238155034681162,
+      "num_tokens": 612367458.0,
+      "step": 3456
+    },
+    {
+      "entropy": 0.7356340115269026,
+      "epoch": 1.0703758006018984,
+      "grad_norm": 0.15480412542819977,
+      "learning_rate": 8.261702434638936e-05,
+      "loss": 0.7165,
+      "mean_token_accuracy": 0.8265450286368529,
+      "num_tokens": 614500562.0,
+      "step": 3468
+    },
+    {
+      "entropy": 0.7494183418651422,
+      "epoch": 1.0740797901072614,
+      "grad_norm": 0.14762075245380402,
+      "learning_rate": 8.249526885004809e-05,
+      "loss": 0.7274,
+      "mean_token_accuracy": 0.8240708733598391,
+      "num_tokens": 616628140.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7446132637560368,
+      "epoch": 1.0777837796126244,
+      "grad_norm": 0.15463441610336304,
+      "learning_rate": 8.237317884589361e-05,
+      "loss": 0.7247,
+      "mean_token_accuracy": 0.8243404676516851,
+      "num_tokens": 618719252.0,
+      "step": 3492
+    },
+    {
+      "entropy": 0.7410028763115406,
+      "epoch": 1.0814877691179876,
+      "grad_norm": 0.15182152390480042,
+      "learning_rate": 8.225075559072614e-05,
+      "loss": 0.7195,
+      "mean_token_accuracy": 0.8264857692023119,
+      "num_tokens": 620851273.0,
+      "step": 3504
+    },
+    {
+      "entropy": 0.7477393746376038,
+      "epoch": 1.0851917586233506,
+      "grad_norm": 0.14437253773212433,
+      "learning_rate": 8.212800034477637e-05,
+      "loss": 0.7256,
+      "mean_token_accuracy": 0.8241480415066084,
+      "num_tokens": 622956290.0,
+      "step": 3516
+    },
+    {
+      "entropy": 0.734037263939778,
+      "epoch": 1.0888957481287136,
+      "grad_norm": 0.15787097811698914,
+      "learning_rate": 8.200491437169251e-05,
+      "loss": 0.713,
+      "mean_token_accuracy": 0.8271950905521711,
+      "num_tokens": 625080917.0,
+      "step": 3528
+    },
+    {
+      "entropy": 0.7377768655618032,
+      "epoch": 1.0925997376340768,
+      "grad_norm": 0.14514772593975067,
+      "learning_rate": 8.188149893852732e-05,
+      "loss": 0.7162,
+      "mean_token_accuracy": 0.8267666287720203,
+      "num_tokens": 627213629.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7321221468349298,
+      "epoch": 1.0963037271394398,
+      "grad_norm": 0.15606361627578735,
+      "learning_rate": 8.175775531572501e-05,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.8280278158684572,
+      "num_tokens": 629338939.0,
+      "step": 3552
+    },
+    {
+      "entropy": 0.7345453649759293,
+      "epoch": 1.1000077166448028,
+      "grad_norm": 0.14273642003536224,
+      "learning_rate": 8.163368477710825e-05,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.8269894048571587,
+      "num_tokens": 631433430.0,
+      "step": 3564
+    },
+    {
+      "entropy": 0.7408470747371515,
+      "epoch": 1.103711706150166,
+      "grad_norm": 0.16014736890792847,
+      "learning_rate": 8.150928859986488e-05,
+      "loss": 0.7194,
+      "mean_token_accuracy": 0.8258067518472672,
+      "num_tokens": 633560482.0,
+      "step": 3576
+    },
+    {
+      "entropy": 0.7225839781264464,
+      "epoch": 1.107415695655529,
+      "grad_norm": 0.15194731950759888,
+      "learning_rate": 8.138456806453503e-05,
+      "loss": 0.7012,
+      "mean_token_accuracy": 0.8298800686995188,
+      "num_tokens": 635693516.0,
+      "step": 3588
+    },
+    {
+      "entropy": 0.7254953247805437,
+      "epoch": 1.111119685160892,
+      "grad_norm": 0.15114106237888336,
+      "learning_rate": 8.125952445499765e-05,
+      "loss": 0.7036,
+      "mean_token_accuracy": 0.8291740665833155,
+      "num_tokens": 637840171.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7438126876950264,
+      "epoch": 1.1148236746662552,
+      "grad_norm": 0.15659384429454803,
+      "learning_rate": 8.113415905845751e-05,
+      "loss": 0.7238,
+      "mean_token_accuracy": 0.8248664475977421,
+      "num_tokens": 639953063.0,
+      "step": 3612
+    },
+    {
+      "entropy": 0.7412439535061518,
+      "epoch": 1.1185276641716182,
+      "grad_norm": 0.14707545936107635,
+      "learning_rate": 8.100847316543185e-05,
+      "loss": 0.7207,
+      "mean_token_accuracy": 0.825270589441061,
+      "num_tokens": 642061202.0,
+      "step": 3624
+    },
+    {
+      "entropy": 0.7216433795789877,
+      "epoch": 1.1222316536769812,
+      "grad_norm": 0.14075824618339539,
+      "learning_rate": 8.088246806973712e-05,
+      "loss": 0.7012,
+      "mean_token_accuracy": 0.8303951819737753,
+      "num_tokens": 644170583.0,
+      "step": 3636
+    },
+    {
+      "entropy": 0.7379123046994209,
+      "epoch": 1.1259356431823444,
+      "grad_norm": 0.14188188314437866,
+      "learning_rate": 8.075614506847563e-05,
+      "loss": 0.7167,
+      "mean_token_accuracy": 0.8262317391733328,
+      "num_tokens": 646270524.0,
+      "step": 3648
+    },
+    {
+      "entropy": 0.7330505475401878,
+      "epoch": 1.1296396326877074,
+      "grad_norm": 0.14915235340595245,
+      "learning_rate": 8.062950546202228e-05,
+      "loss": 0.7143,
+      "mean_token_accuracy": 0.8266897139449915,
+      "num_tokens": 648408593.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.7359581738710403,
+      "epoch": 1.1333436221930704,
+      "grad_norm": 0.1471003293991089,
+      "learning_rate": 8.050255055401105e-05,
+      "loss": 0.7131,
+      "mean_token_accuracy": 0.8272727802395821,
+      "num_tokens": 650550976.0,
+      "step": 3672
+    },
+    {
+      "entropy": 0.7366929526130358,
+      "epoch": 1.1370476116984336,
+      "grad_norm": 0.14605937898159027,
+      "learning_rate": 8.03752816513217e-05,
+      "loss": 0.7169,
+      "mean_token_accuracy": 0.825990212460359,
+      "num_tokens": 652692582.0,
+      "step": 3684
+    },
+    {
+      "entropy": 0.7435482144355774,
+      "epoch": 1.1407516012037966,
+      "grad_norm": 0.15462058782577515,
+      "learning_rate": 8.024770006406628e-05,
+      "loss": 0.7251,
+      "mean_token_accuracy": 0.823987594495217,
+      "num_tokens": 654848825.0,
+      "step": 3696
+    },
+    {
+      "entropy": 0.7348724926511446,
+      "epoch": 1.1444555907091596,
+      "grad_norm": 0.1494790017604828,
+      "learning_rate": 8.011980710557554e-05,
+      "loss": 0.7139,
+      "mean_token_accuracy": 0.8266541995108128,
+      "num_tokens": 656948636.0,
+      "step": 3708
+    },
+    {
+      "entropy": 0.7282378102342287,
+      "epoch": 1.1481595802145228,
+      "grad_norm": 0.16433827579021454,
+      "learning_rate": 7.999160409238563e-05,
+      "loss": 0.7075,
+      "mean_token_accuracy": 0.8285117484629154,
+      "num_tokens": 659061066.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.738138652096192,
+      "epoch": 1.1518635697198858,
+      "grad_norm": 0.15777455270290375,
+      "learning_rate": 7.986309234422427e-05,
+      "loss": 0.7177,
+      "mean_token_accuracy": 0.8264009902874628,
+      "num_tokens": 661175184.0,
+      "step": 3732
+    },
+    {
+      "entropy": 0.730212123443683,
+      "epoch": 1.1555675592252488,
+      "grad_norm": 0.1504729837179184,
+      "learning_rate": 7.973427318399746e-05,
+      "loss": 0.7146,
+      "mean_token_accuracy": 0.827114554742972,
+      "num_tokens": 663279250.0,
+      "step": 3744
+    },
+    {
+      "entropy": 0.7331996709108353,
+      "epoch": 1.159271548730612,
+      "grad_norm": 0.15760678052902222,
+      "learning_rate": 7.960514793777559e-05,
+      "loss": 0.7107,
+      "mean_token_accuracy": 0.8281191426018873,
+      "num_tokens": 665384207.0,
+      "step": 3756
+    },
+    {
+      "entropy": 0.7236106917262077,
+      "epoch": 1.162975538235975,
+      "grad_norm": 0.15555773675441742,
+      "learning_rate": 7.947571793478e-05,
+      "loss": 0.7033,
+      "mean_token_accuracy": 0.8294017761945724,
+      "num_tokens": 667490801.0,
+      "step": 3768
+    },
+    {
+      "entropy": 0.7413870220383009,
+      "epoch": 1.166679527741338,
+      "grad_norm": 0.14847822487354279,
+      "learning_rate": 7.934598450736919e-05,
+      "loss": 0.7201,
+      "mean_token_accuracy": 0.8256005557874838,
+      "num_tokens": 669641877.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.7133117939035097,
+      "epoch": 1.1703835172467012,
+      "grad_norm": 0.14297045767307281,
+      "learning_rate": 7.921594899102505e-05,
+      "loss": 0.6898,
+      "mean_token_accuracy": 0.8321465166906515,
+      "num_tokens": 671760517.0,
+      "step": 3792
+    },
+    {
+      "entropy": 0.7260085244973501,
+      "epoch": 1.1740875067520642,
+      "grad_norm": 0.154536172747612,
+      "learning_rate": 7.908561272433932e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.828648411979278,
+      "num_tokens": 673879141.0,
+      "step": 3804
+    },
+    {
+      "entropy": 0.7265941066046556,
+      "epoch": 1.1777914962574272,
+      "grad_norm": 0.15274272859096527,
+      "learning_rate": 7.895497704899957e-05,
+      "loss": 0.7083,
+      "mean_token_accuracy": 0.8286111789445082,
+      "num_tokens": 676023807.0,
+      "step": 3816
+    },
+    {
+      "entropy": 0.7224672473967075,
+      "epoch": 1.1814954857627904,
+      "grad_norm": 0.16113263368606567,
+      "learning_rate": 7.882404330977556e-05,
+      "loss": 0.7002,
+      "mean_token_accuracy": 0.8300817757844925,
+      "num_tokens": 678114097.0,
+      "step": 3828
+    },
+    {
+      "entropy": 0.731824230402708,
+      "epoch": 1.1851994752681534,
+      "grad_norm": 0.16188302636146545,
+      "learning_rate": 7.869281285450527e-05,
+      "loss": 0.7136,
+      "mean_token_accuracy": 0.827064195026954,
+      "num_tokens": 680225565.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.745010394603014,
+      "epoch": 1.1889034647735164,
+      "grad_norm": 0.1467347890138626,
+      "learning_rate": 7.856128703408118e-05,
+      "loss": 0.7239,
+      "mean_token_accuracy": 0.8247304347654184,
+      "num_tokens": 682344124.0,
+      "step": 3852
+    },
+    {
+      "entropy": 0.7087207237879435,
+      "epoch": 1.1926074542788796,
+      "grad_norm": 0.14708933234214783,
+      "learning_rate": 7.842946720243617e-05,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.8324318118393421,
+      "num_tokens": 684423002.0,
+      "step": 3864
+    },
+    {
+      "entropy": 0.7357664232452711,
+      "epoch": 1.1963114437842426,
+      "grad_norm": 0.13852429389953613,
+      "learning_rate": 7.829735471652978e-05,
+      "loss": 0.7142,
+      "mean_token_accuracy": 0.8267920973400275,
+      "num_tokens": 686526635.0,
+      "step": 3876
+    },
+    {
+      "entropy": 0.717096570879221,
+      "epoch": 1.2000154332896056,
+      "grad_norm": 0.15155161917209625,
+      "learning_rate": 7.816495093633405e-05,
+      "loss": 0.6966,
+      "mean_token_accuracy": 0.8317411964138349,
+      "num_tokens": 688658046.0,
+      "step": 3888
+    },
+    {
+      "entropy": 0.7423039426406225,
+      "epoch": 1.2037194227949688,
+      "grad_norm": 0.16276352107524872,
+      "learning_rate": 7.80322572248197e-05,
+      "loss": 0.7213,
+      "mean_token_accuracy": 0.8252335165937742,
+      "num_tokens": 690786172.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.711784016340971,
+      "epoch": 1.2074234123003318,
+      "grad_norm": 0.1595115065574646,
+      "learning_rate": 7.7899274947942e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.8321591466665268,
+      "num_tokens": 692906988.0,
+      "step": 3912
+    },
+    {
+      "entropy": 0.7300878415505091,
+      "epoch": 1.2111274018056948,
+      "grad_norm": 0.14089138805866241,
+      "learning_rate": 7.77660054746267e-05,
+      "loss": 0.7072,
+      "mean_token_accuracy": 0.8286360974113146,
+      "num_tokens": 695033002.0,
+      "step": 3924
+    },
+    {
+      "entropy": 0.7247926443815231,
+      "epoch": 1.214831391311058,
+      "grad_norm": 0.1512719839811325,
+      "learning_rate": 7.763245017675596e-05,
+      "loss": 0.7032,
+      "mean_token_accuracy": 0.8293510377407074,
+      "num_tokens": 697153179.0,
+      "step": 3936
+    },
+    {
+      "entropy": 0.7274364518622557,
+      "epoch": 1.218535380816421,
+      "grad_norm": 0.14784738421440125,
+      "learning_rate": 7.749861042915424e-05,
+      "loss": 0.7094,
+      "mean_token_accuracy": 0.8280140981078148,
+      "num_tokens": 699267739.0,
+      "step": 3948
+    },
+    {
+      "entropy": 0.733709204941988,
+      "epoch": 1.222239370321784,
+      "grad_norm": 0.1493990421295166,
+      "learning_rate": 7.736448760957418e-05,
+      "loss": 0.7153,
+      "mean_token_accuracy": 0.826848104596138,
+      "num_tokens": 701376610.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7143348765869936,
+      "epoch": 1.2259433598271472,
+      "grad_norm": 0.16152378916740417,
+      "learning_rate": 7.72300830986823e-05,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.8318944051861763,
+      "num_tokens": 703516019.0,
+      "step": 3972
+    },
+    {
+      "entropy": 0.7241542227566242,
+      "epoch": 1.2296473493325102,
+      "grad_norm": 0.1497247815132141,
+      "learning_rate": 7.709539828004492e-05,
+      "loss": 0.7011,
+      "mean_token_accuracy": 0.8292028916378816,
+      "num_tokens": 705617753.0,
+      "step": 3984
+    },
+    {
+      "entropy": 0.7197034644583861,
+      "epoch": 1.2333513388378732,
+      "grad_norm": 0.1511349231004715,
+      "learning_rate": 7.696043454011387e-05,
+      "loss": 0.7001,
+      "mean_token_accuracy": 0.8302300423383713,
+      "num_tokens": 707732735.0,
+      "step": 3996
+    },
+    {
+      "entropy": 0.7363270024458567,
+      "epoch": 1.2370553283432364,
+      "grad_norm": 0.15632647275924683,
+      "learning_rate": 7.682519326821215e-05,
+      "loss": 0.716,
+      "mean_token_accuracy": 0.8261935313542684,
+      "num_tokens": 709835226.0,
+      "step": 4008
+    },
+    {
+      "entropy": 0.7174325250089169,
+      "epoch": 1.2407593178485994,
+      "grad_norm": 0.13901802897453308,
+      "learning_rate": 7.668967585651974e-05,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8306252273420492,
+      "num_tokens": 711988818.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7259048347671827,
+      "epoch": 1.2444633073539624,
+      "grad_norm": 0.1487317979335785,
+      "learning_rate": 7.65538837000592e-05,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8284634724259377,
+      "num_tokens": 714096477.0,
+      "step": 4032
+    },
+    {
+      "entropy": 0.7280755999187628,
+      "epoch": 1.2481672968593256,
+      "grad_norm": 0.14473356306552887,
+      "learning_rate": 7.64178181966813e-05,
+      "loss": 0.7081,
+      "mean_token_accuracy": 0.8283463989694914,
+      "num_tokens": 716245082.0,
+      "step": 4044
+    },
+    {
+      "entropy": 0.7249644311765829,
+      "epoch": 1.2518712863646886,
+      "grad_norm": 0.1691472977399826,
+      "learning_rate": 7.62814807470507e-05,
+      "loss": 0.7051,
+      "mean_token_accuracy": 0.8292301384111246,
+      "num_tokens": 718361640.0,
+      "step": 4056
+    },
+    {
+      "entropy": 0.720829289406538,
+      "epoch": 1.2555752758700516,
+      "grad_norm": 0.14902262389659882,
+      "learning_rate": 7.614487275463143e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.8298750383158525,
+      "num_tokens": 720489178.0,
+      "step": 4068
+    },
+    {
+      "entropy": 0.7293098631004492,
+      "epoch": 1.2592792653754148,
+      "grad_norm": 0.1404629945755005,
+      "learning_rate": 7.600799562567258e-05,
+      "loss": 0.7089,
+      "mean_token_accuracy": 0.8281568810343742,
+      "num_tokens": 722608223.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.722664033373197,
+      "epoch": 1.2629832548807778,
+      "grad_norm": 0.14373779296875,
+      "learning_rate": 7.587085076919369e-05,
+      "loss": 0.7024,
+      "mean_token_accuracy": 0.8302177861332893,
+      "num_tokens": 724757494.0,
+      "step": 4092
+    },
+    {
+      "entropy": 0.7234973087906837,
+      "epoch": 1.2666872443861408,
+      "grad_norm": 0.14617067575454712,
+      "learning_rate": 7.573343959697029e-05,
+      "loss": 0.7046,
+      "mean_token_accuracy": 0.8289206735789776,
+      "num_tokens": 726922696.0,
+      "step": 4104
+    },
+    {
+      "entropy": 0.7261980349818865,
+      "epoch": 1.270391233891504,
+      "grad_norm": 0.1495673805475235,
+      "learning_rate": 7.55957635235194e-05,
+      "loss": 0.706,
+      "mean_token_accuracy": 0.828493465979894,
+      "num_tokens": 729052352.0,
+      "step": 4116
+    },
+    {
+      "entropy": 0.7174485189219316,
+      "epoch": 1.274095223396867,
+      "grad_norm": 0.14732122421264648,
+      "learning_rate": 7.545782396608496e-05,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.8296906674901644,
+      "num_tokens": 731197978.0,
+      "step": 4128
+    },
+    {
+      "entropy": 0.7247343001266321,
+      "epoch": 1.27779921290223,
+      "grad_norm": 0.14333774149417877,
+      "learning_rate": 7.53196223446232e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.8288652760287126,
+      "num_tokens": 733333853.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7190474011003971,
+      "epoch": 1.2815032024075932,
+      "grad_norm": 0.15297874808311462,
+      "learning_rate": 7.518116008178805e-05,
+      "loss": 0.7012,
+      "mean_token_accuracy": 0.8297313017149767,
+      "num_tokens": 735431034.0,
+      "step": 4152
+    },
+    {
+      "entropy": 0.7267656040688356,
+      "epoch": 1.2852071919129562,
+      "grad_norm": 0.14727260172367096,
+      "learning_rate": 7.50424386029165e-05,
+      "loss": 0.7044,
+      "mean_token_accuracy": 0.8287480349342028,
+      "num_tokens": 737589354.0,
+      "step": 4164
+    },
+    {
+      "entropy": 0.7389783896505833,
+      "epoch": 1.2889111814183194,
+      "grad_norm": 0.15076899528503418,
+      "learning_rate": 7.490345933601395e-05,
+      "loss": 0.7211,
+      "mean_token_accuracy": 0.8257643903295199,
+      "num_tokens": 739720304.0,
+      "step": 4176
+    },
+    {
+      "entropy": 0.7206626770397028,
+      "epoch": 1.2926151709236824,
+      "grad_norm": 0.16268907487392426,
+      "learning_rate": 7.476422371173942e-05,
+      "loss": 0.6993,
+      "mean_token_accuracy": 0.8301717328528563,
+      "num_tokens": 741856208.0,
+      "step": 4188
+    },
+    {
+      "entropy": 0.7273670248687267,
+      "epoch": 1.2963191604290454,
+      "grad_norm": 0.158157080411911,
+      "learning_rate": 7.462473316339093e-05,
+      "loss": 0.7062,
+      "mean_token_accuracy": 0.8289822128911813,
+      "num_tokens": 743969037.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7194506376981735,
+      "epoch": 1.3000231499344086,
+      "grad_norm": 0.16528646647930145,
+      "learning_rate": 7.44849891268907e-05,
+      "loss": 0.699,
+      "mean_token_accuracy": 0.8304046653211117,
+      "num_tokens": 746092753.0,
+      "step": 4212
+    },
+    {
+      "entropy": 0.7224392394224802,
+      "epoch": 1.3037271394397716,
+      "grad_norm": 0.15017327666282654,
+      "learning_rate": 7.434499304077036e-05,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.8293930192788442,
+      "num_tokens": 748204673.0,
+      "step": 4224
+    },
+    {
+      "entropy": 0.7125812520583471,
+      "epoch": 1.3074311289451346,
+      "grad_norm": 0.15608710050582886,
+      "learning_rate": 7.420474634615617e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.8319937810301781,
+      "num_tokens": 750325027.0,
+      "step": 4236
+    },
+    {
+      "entropy": 0.7182938729723295,
+      "epoch": 1.3111351184504978,
+      "grad_norm": 0.15669798851013184,
+      "learning_rate": 7.406425048675409e-05,
+      "loss": 0.6978,
+      "mean_token_accuracy": 0.8300301494697729,
+      "num_tokens": 752470674.0,
+      "step": 4248
+    },
+    {
+      "entropy": 0.7105309578279654,
+      "epoch": 1.3148391079558608,
+      "grad_norm": 0.15128910541534424,
+      "learning_rate": 7.392350690883509e-05,
+      "loss": 0.6895,
+      "mean_token_accuracy": 0.8324328971405824,
+      "num_tokens": 754594955.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7113361358642578,
+      "epoch": 1.3185430974612238,
+      "grad_norm": 0.15378808975219727,
+      "learning_rate": 7.378251706122013e-05,
+      "loss": 0.6903,
+      "mean_token_accuracy": 0.8316868332525095,
+      "num_tokens": 756717367.0,
+      "step": 4272
+    },
+    {
+      "entropy": 0.7267830123504003,
+      "epoch": 1.322247086966587,
+      "grad_norm": 0.151032954454422,
+      "learning_rate": 7.364128239526525e-05,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8282621925075849,
+      "num_tokens": 758856430.0,
+      "step": 4284
+    },
+    {
+      "entropy": 0.721363440155983,
+      "epoch": 1.32595107647195,
+      "grad_norm": 0.16554951667785645,
+      "learning_rate": 7.349980436484672e-05,
+      "loss": 0.702,
+      "mean_token_accuracy": 0.8293261242409548,
+      "num_tokens": 761009855.0,
+      "step": 4296
+    },
+    {
+      "entropy": 0.717485940704743,
+      "epoch": 1.3296550659773132,
+      "grad_norm": 0.14937981963157654,
+      "learning_rate": 7.335808442634596e-05,
+      "loss": 0.6975,
+      "mean_token_accuracy": 0.8307504417995611,
+      "num_tokens": 763137834.0,
+      "step": 4308
+    },
+    {
+      "entropy": 0.7333954150478045,
+      "epoch": 1.3333590554826762,
+      "grad_norm": 0.14564256370067596,
+      "learning_rate": 7.321612403863465e-05,
+      "loss": 0.7109,
+      "mean_token_accuracy": 0.8272267108162245,
+      "num_tokens": 765243912.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.7177773999671141,
+      "epoch": 1.3370630449880392,
+      "grad_norm": 0.14997075498104095,
+      "learning_rate": 7.30739246630596e-05,
+      "loss": 0.6971,
+      "mean_token_accuracy": 0.8304961547255516,
+      "num_tokens": 767342231.0,
+      "step": 4332
+    },
+    {
+      "entropy": 0.7231075142820677,
+      "epoch": 1.3407670344934024,
+      "grad_norm": 0.15372461080551147,
+      "learning_rate": 7.293148776342787e-05,
+      "loss": 0.7022,
+      "mean_token_accuracy": 0.8295708782970905,
+      "num_tokens": 769459559.0,
+      "step": 4344
+    },
+    {
+      "entropy": 0.727438664684693,
+      "epoch": 1.3444710239987654,
+      "grad_norm": 0.1568097472190857,
+      "learning_rate": 7.278881480599151e-05,
+      "loss": 0.7063,
+      "mean_token_accuracy": 0.8285723961889744,
+      "num_tokens": 771586827.0,
+      "step": 4356
+    },
+    {
+      "entropy": 0.7256615745524565,
+      "epoch": 1.3481750135041284,
+      "grad_norm": 0.15446576476097107,
+      "learning_rate": 7.264590725943263e-05,
+      "loss": 0.7042,
+      "mean_token_accuracy": 0.8298846408724785,
+      "num_tokens": 773753359.0,
+      "step": 4368
+    },
+    {
+      "entropy": 0.716560627023379,
+      "epoch": 1.3518790030094916,
+      "grad_norm": 0.15649054944515228,
+      "learning_rate": 7.250276659484814e-05,
+      "loss": 0.6973,
+      "mean_token_accuracy": 0.8308561046918234,
+      "num_tokens": 775838605.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7301006155709425,
+      "epoch": 1.3555829925148546,
+      "grad_norm": 0.15302646160125732,
+      "learning_rate": 7.235939428573473e-05,
+      "loss": 0.7084,
+      "mean_token_accuracy": 0.8278401518861452,
+      "num_tokens": 777947001.0,
+      "step": 4392
+    },
+    {
+      "entropy": 0.7269327379763126,
+      "epoch": 1.3592869820202176,
+      "grad_norm": 0.15263816714286804,
+      "learning_rate": 7.221579180797365e-05,
+      "loss": 0.7051,
+      "mean_token_accuracy": 0.8292628613611063,
+      "num_tokens": 780073853.0,
+      "step": 4404
+    },
+    {
+      "entropy": 0.7197130558391412,
+      "epoch": 1.3629909715255808,
+      "grad_norm": 0.1532004326581955,
+      "learning_rate": 7.207196063981552e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.8306596241891384,
+      "num_tokens": 782197116.0,
+      "step": 4416
+    },
+    {
+      "entropy": 0.7153556197881699,
+      "epoch": 1.3666949610309438,
+      "grad_norm": 0.16322393715381622,
+      "learning_rate": 7.192790226186505e-05,
+      "loss": 0.694,
+      "mean_token_accuracy": 0.8311380048592886,
+      "num_tokens": 784336823.0,
+      "step": 4428
+    },
+    {
+      "entropy": 0.7153538229564825,
+      "epoch": 1.3703989505363068,
+      "grad_norm": 0.17276719212532043,
+      "learning_rate": 7.178361815706594e-05,
+      "loss": 0.6943,
+      "mean_token_accuracy": 0.830880576123794,
+      "num_tokens": 786476106.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7140049921969572,
+      "epoch": 1.37410294004167,
+      "grad_norm": 0.1465296596288681,
+      "learning_rate": 7.163910981068547e-05,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8320667843023936,
+      "num_tokens": 788592462.0,
+      "step": 4452
+    },
+    {
+      "entropy": 0.7188820218046507,
+      "epoch": 1.377806929547033,
+      "grad_norm": 0.1437097191810608,
+      "learning_rate": 7.14943787102993e-05,
+      "loss": 0.6992,
+      "mean_token_accuracy": 0.8301914831002554,
+      "num_tokens": 790710211.0,
+      "step": 4464
+    },
+    {
+      "entropy": 0.7231453433632851,
+      "epoch": 1.381510919052396,
+      "grad_norm": 0.15103840827941895,
+      "learning_rate": 7.134942634577614e-05,
+      "loss": 0.7028,
+      "mean_token_accuracy": 0.8289699579278628,
+      "num_tokens": 792853918.0,
+      "step": 4476
+    },
+    {
+      "entropy": 0.7152188581724962,
+      "epoch": 1.3852149085577592,
+      "grad_norm": 0.15957361459732056,
+      "learning_rate": 7.12042542092624e-05,
+      "loss": 0.6933,
+      "mean_token_accuracy": 0.8314299285411835,
+      "num_tokens": 794995486.0,
+      "step": 4488
+    },
+    {
+      "entropy": 0.7186589650809765,
+      "epoch": 1.3889188980631222,
+      "grad_norm": 0.15625914931297302,
+      "learning_rate": 7.105886379516679e-05,
+      "loss": 0.7,
+      "mean_token_accuracy": 0.8297496847808361,
+      "num_tokens": 797143590.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.7220030228296915,
+      "epoch": 1.3926228875684852,
+      "grad_norm": 0.15391391515731812,
+      "learning_rate": 7.091325660014505e-05,
+      "loss": 0.7001,
+      "mean_token_accuracy": 0.8300996559361616,
+      "num_tokens": 799290589.0,
+      "step": 4512
+    },
+    {
+      "entropy": 0.7087641693651676,
+      "epoch": 1.3963268770738484,
+      "grad_norm": 0.1555260866880417,
+      "learning_rate": 7.076743412308441e-05,
+      "loss": 0.6868,
+      "mean_token_accuracy": 0.8325341766079267,
+      "num_tokens": 801404554.0,
+      "step": 4524
+    },
+    {
+      "entropy": 0.734593483308951,
+      "epoch": 1.4000308665792114,
+      "grad_norm": 0.16026608645915985,
+      "learning_rate": 7.062139786508827e-05,
+      "loss": 0.7146,
+      "mean_token_accuracy": 0.8272804208099842,
+      "num_tokens": 803507633.0,
+      "step": 4536
+    },
+    {
+      "entropy": 0.7193813882768154,
+      "epoch": 1.4037348560845744,
+      "grad_norm": 0.15414464473724365,
+      "learning_rate": 7.047514932946068e-05,
+      "loss": 0.6982,
+      "mean_token_accuracy": 0.83012605458498,
+      "num_tokens": 805648024.0,
+      "step": 4548
+    },
+    {
+      "entropy": 0.7172121778130531,
+      "epoch": 1.4074388455899376,
+      "grad_norm": 0.15337833762168884,
+      "learning_rate": 7.032869002169088e-05,
+      "loss": 0.6981,
+      "mean_token_accuracy": 0.8307264360288779,
+      "num_tokens": 807789002.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.7184814922511578,
+      "epoch": 1.4111428350953006,
+      "grad_norm": 0.15992848575115204,
+      "learning_rate": 7.01820214494378e-05,
+      "loss": 0.6974,
+      "mean_token_accuracy": 0.8301586744685968,
+      "num_tokens": 809914470.0,
+      "step": 4572
+    },
+    {
+      "entropy": 0.7108126245439053,
+      "epoch": 1.4148468246006636,
+      "grad_norm": 0.15666691958904266,
+      "learning_rate": 7.00351451225146e-05,
+      "loss": 0.6883,
+      "mean_token_accuracy": 0.832117979725202,
+      "num_tokens": 812044112.0,
+      "step": 4584
+    },
+    {
+      "entropy": 0.7207119353115559,
+      "epoch": 1.4185508141060268,
+      "grad_norm": 0.162274569272995,
+      "learning_rate": 6.9888062552873e-05,
+      "loss": 0.6988,
+      "mean_token_accuracy": 0.830579354117314,
+      "num_tokens": 814207854.0,
+      "step": 4596
+    },
+    {
+      "entropy": 0.7099998084207376,
+      "epoch": 1.4222548036113898,
+      "grad_norm": 0.15512961149215698,
+      "learning_rate": 6.974077525458785e-05,
+      "loss": 0.6901,
+      "mean_token_accuracy": 0.8324882164597511,
+      "num_tokens": 816326311.0,
+      "step": 4608
+    },
+    {
+      "entropy": 0.7227374029656252,
+      "epoch": 1.4259587931167528,
+      "grad_norm": 0.15393690764904022,
+      "learning_rate": 6.95932847438415e-05,
+      "loss": 0.7023,
+      "mean_token_accuracy": 0.8297974628706773,
+      "num_tokens": 818475940.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.7139190497497717,
+      "epoch": 1.429662782622116,
+      "grad_norm": 0.14897596836090088,
+      "learning_rate": 6.944559253890809e-05,
+      "loss": 0.6923,
+      "mean_token_accuracy": 0.8322372958064079,
+      "num_tokens": 820570551.0,
+      "step": 4632
+    },
+    {
+      "entropy": 0.7113192056616148,
+      "epoch": 1.433366772127479,
+      "grad_norm": 0.16115012764930725,
+      "learning_rate": 6.92977001601381e-05,
+      "loss": 0.6919,
+      "mean_token_accuracy": 0.8324420315523943,
+      "num_tokens": 822687816.0,
+      "step": 4644
+    },
+    {
+      "entropy": 0.7261552785833677,
+      "epoch": 1.437070761632842,
+      "grad_norm": 0.17303551733493805,
+      "learning_rate": 6.914960912994257e-05,
+      "loss": 0.7047,
+      "mean_token_accuracy": 0.8293804277976354,
+      "num_tokens": 824846735.0,
+      "step": 4656
+    },
+    {
+      "entropy": 0.7077195967237154,
+      "epoch": 1.4407747511382052,
+      "grad_norm": 0.15039795637130737,
+      "learning_rate": 6.900132097277748e-05,
+      "loss": 0.687,
+      "mean_token_accuracy": 0.8330833079914252,
+      "num_tokens": 827013414.0,
+      "step": 4668
+    },
+    {
+      "entropy": 0.7014664622644583,
+      "epoch": 1.4444787406435682,
+      "grad_norm": 0.1548740714788437,
+      "learning_rate": 6.885283721512803e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.834053193529447,
+      "num_tokens": 829191903.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.7116682541867098,
+      "epoch": 1.4481827301489312,
+      "grad_norm": 0.15452542901039124,
+      "learning_rate": 6.870415938549292e-05,
+      "loss": 0.6917,
+      "mean_token_accuracy": 0.8317574722071489,
+      "num_tokens": 831331900.0,
+      "step": 4692
+    },
+    {
+      "entropy": 0.7132287646333376,
+      "epoch": 1.4518867196542944,
+      "grad_norm": 0.15533322095870972,
+      "learning_rate": 6.855528901436871e-05,
+      "loss": 0.6894,
+      "mean_token_accuracy": 0.8324517086148262,
+      "num_tokens": 833489694.0,
+      "step": 4704
+    },
+    {
+      "entropy": 0.7113682615260283,
+      "epoch": 1.4555907091596574,
+      "grad_norm": 0.1611347496509552,
+      "learning_rate": 6.840622763423391e-05,
+      "loss": 0.6906,
+      "mean_token_accuracy": 0.8323202182849249,
+      "num_tokens": 835603454.0,
+      "step": 4716
+    },
+    {
+      "entropy": 0.6990887063244978,
+      "epoch": 1.4592946986650204,
+      "grad_norm": 0.1729213446378708,
+      "learning_rate": 6.825697677953332e-05,
+      "loss": 0.6819,
+      "mean_token_accuracy": 0.8341849880913893,
+      "num_tokens": 837735389.0,
+      "step": 4728
+    },
+    {
+      "entropy": 0.7129435998698076,
+      "epoch": 1.4629986881703836,
+      "grad_norm": 0.1657145619392395,
+      "learning_rate": 6.810753798666223e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.8312515988945961,
+      "num_tokens": 839880792.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.7213788678248724,
+      "epoch": 1.4667026776757466,
+      "grad_norm": 0.1646701693534851,
+      "learning_rate": 6.795791279395052e-05,
+      "loss": 0.6996,
+      "mean_token_accuracy": 0.8302897252142429,
+      "num_tokens": 842005360.0,
+      "step": 4752
+    },
+    {
+      "entropy": 0.7203125320374966,
+      "epoch": 1.4704066671811096,
+      "grad_norm": 0.1686820685863495,
+      "learning_rate": 6.780810274164691e-05,
+      "loss": 0.6968,
+      "mean_token_accuracy": 0.8303221389651299,
+      "num_tokens": 844137345.0,
+      "step": 4764
+    },
+    {
+      "entropy": 0.7195424201587836,
+      "epoch": 1.4741106566864728,
+      "grad_norm": 0.1486455649137497,
+      "learning_rate": 6.765810937190306e-05,
+      "loss": 0.6999,
+      "mean_token_accuracy": 0.8301795323689779,
+      "num_tokens": 846276636.0,
+      "step": 4776
+    },
+    {
+      "entropy": 0.7059193042417368,
+      "epoch": 1.4778146461918358,
+      "grad_norm": 0.16239939630031586,
+      "learning_rate": 6.750793422875771e-05,
+      "loss": 0.6847,
+      "mean_token_accuracy": 0.833564005792141,
+      "num_tokens": 848408984.0,
+      "step": 4788
+    },
+    {
+      "entropy": 0.724933902422587,
+      "epoch": 1.4815186356971988,
+      "grad_norm": 0.16981275379657745,
+      "learning_rate": 6.73575788581208e-05,
+      "loss": 0.7054,
+      "mean_token_accuracy": 0.8286726363003254,
+      "num_tokens": 850546298.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.7115325927734375,
+      "epoch": 1.485222625202562,
+      "grad_norm": 0.17294248938560486,
+      "learning_rate": 6.720704480775753e-05,
+      "loss": 0.6899,
+      "mean_token_accuracy": 0.8321196387211481,
+      "num_tokens": 852661935.0,
+      "step": 4812
+    },
+    {
+      "entropy": 0.7177861680587133,
+      "epoch": 1.488926614707925,
+      "grad_norm": 0.160500168800354,
+      "learning_rate": 6.705633362727243e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.8320040429631869,
+      "num_tokens": 854775802.0,
+      "step": 4824
+    },
+    {
+      "entropy": 0.704534916828076,
+      "epoch": 1.492630604213288,
+      "grad_norm": 0.16029898822307587,
+      "learning_rate": 6.690544686809342e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.8342838796476523,
+      "num_tokens": 856867232.0,
+      "step": 4836
+    },
+    {
+      "entropy": 0.7108530278007189,
+      "epoch": 1.4963345937186512,
+      "grad_norm": 0.16785788536071777,
+      "learning_rate": 6.675438608345583e-05,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.832078884045283,
+      "num_tokens": 859022001.0,
+      "step": 4848
+    },
+    {
+      "entropy": 0.7115449421107769,
+      "epoch": 1.5000385832240142,
+      "grad_norm": 0.15710817277431488,
+      "learning_rate": 6.660315282838643e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.8314018162588278,
+      "num_tokens": 861145193.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.7248165036241213,
+      "epoch": 1.5037425727293772,
+      "grad_norm": 0.16734077036380768,
+      "learning_rate": 6.645174865968742e-05,
+      "loss": 0.7031,
+      "mean_token_accuracy": 0.8290742263197899,
+      "num_tokens": 863278162.0,
+      "step": 4872
+    },
+    {
+      "entropy": 0.7192536381383737,
+      "epoch": 1.5074465622347404,
+      "grad_norm": 0.15687040984630585,
+      "learning_rate": 6.630017513592035e-05,
+      "loss": 0.6977,
+      "mean_token_accuracy": 0.8306297038992246,
+      "num_tokens": 865417354.0,
+      "step": 4884
+    },
+    {
+      "entropy": 0.7058528165022532,
+      "epoch": 1.5111505517401034,
+      "grad_norm": 0.17756927013397217,
+      "learning_rate": 6.614843381739014e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.8327944378058115,
+      "num_tokens": 867563022.0,
+      "step": 4896
+    },
+    {
+      "entropy": 0.7131319008767605,
+      "epoch": 1.5148545412454664,
+      "grad_norm": 0.15551723539829254,
+      "learning_rate": 6.5996526266129e-05,
+      "loss": 0.6927,
+      "mean_token_accuracy": 0.8321846400698026,
+      "num_tokens": 869656511.0,
+      "step": 4908
+    },
+    {
+      "entropy": 0.7125682967404524,
+      "epoch": 1.5185585307508296,
+      "grad_norm": 0.15141348540782928,
+      "learning_rate": 6.584445404588038e-05,
+      "loss": 0.6929,
+      "mean_token_accuracy": 0.831573948264122,
+      "num_tokens": 871824051.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.7077928557991982,
+      "epoch": 1.5222625202561926,
+      "grad_norm": 0.16390904784202576,
+      "learning_rate": 6.569221872208277e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.8330778690675894,
+      "num_tokens": 873933807.0,
+      "step": 4932
+    },
+    {
+      "entropy": 0.7208421056469282,
+      "epoch": 1.5259665097615556,
+      "grad_norm": 0.1592690497636795,
+      "learning_rate": 6.553982186185374e-05,
+      "loss": 0.7006,
+      "mean_token_accuracy": 0.8300425770382086,
+      "num_tokens": 876062457.0,
+      "step": 4944
+    },
+    {
+      "entropy": 0.7219262520472208,
+      "epoch": 1.5296704992669188,
+      "grad_norm": 0.16316470503807068,
+      "learning_rate": 6.538726503397362e-05,
+      "loss": 0.6995,
+      "mean_token_accuracy": 0.8302759416401386,
+      "num_tokens": 878177430.0,
+      "step": 4956
+    },
+    {
+      "entropy": 0.7148293716212114,
+      "epoch": 1.5333744887722818,
+      "grad_norm": 0.16070803999900818,
+      "learning_rate": 6.523454980886957e-05,
+      "loss": 0.6935,
+      "mean_token_accuracy": 0.8319490551948547,
+      "num_tokens": 880334120.0,
+      "step": 4968
+    },
+    {
+      "entropy": 0.7131251158813635,
+      "epoch": 1.5370784782776448,
+      "grad_norm": 0.15670515596866608,
+      "learning_rate": 6.508167775859918e-05,
+      "loss": 0.692,
+      "mean_token_accuracy": 0.8319077553848425,
+      "num_tokens": 882447090.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7086505778133869,
+      "epoch": 1.540782467783008,
+      "grad_norm": 0.15983329713344574,
+      "learning_rate": 6.49286504568345e-05,
+      "loss": 0.6881,
+      "mean_token_accuracy": 0.8326895671586195,
+      "num_tokens": 884569531.0,
+      "step": 4992
+    },
+    {
+      "entropy": 0.7085077924033006,
+      "epoch": 1.544486457288371,
+      "grad_norm": 0.15563714504241943,
+      "learning_rate": 6.477546947884572e-05,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8332745619118214,
+      "num_tokens": 886683533.0,
+      "step": 5004
+    },
+    {
+      "entropy": 0.7201697329680125,
+      "epoch": 1.548190446793734,
+      "grad_norm": 0.15200090408325195,
+      "learning_rate": 6.462213640148495e-05,
+      "loss": 0.7011,
+      "mean_token_accuracy": 0.82986398289601,
+      "num_tokens": 888793750.0,
+      "step": 5016
+    },
+    {
+      "entropy": 0.7024264633655548,
+      "epoch": 1.5518944362990972,
+      "grad_norm": 0.17173556983470917,
+      "learning_rate": 6.446865280317005e-05,
+      "loss": 0.6812,
+      "mean_token_accuracy": 0.8346923552453518,
+      "num_tokens": 890929618.0,
+      "step": 5028
+    },
+    {
+      "entropy": 0.7268513465921084,
+      "epoch": 1.5555984258044602,
+      "grad_norm": 0.16845101118087769,
+      "learning_rate": 6.431502026386835e-05,
+      "loss": 0.7049,
+      "mean_token_accuracy": 0.8287280350923538,
+      "num_tokens": 893036881.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.7066880414883295,
+      "epoch": 1.5593024153098232,
+      "grad_norm": 0.15314914286136627,
+      "learning_rate": 6.416124036508035e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.8333163931965828,
+      "num_tokens": 895190885.0,
+      "step": 5052
+    },
+    {
+      "entropy": 0.7045964933931828,
+      "epoch": 1.5630064048151864,
+      "grad_norm": 0.16949310898780823,
+      "learning_rate": 6.400731468982353e-05,
+      "loss": 0.6842,
+      "mean_token_accuracy": 0.8338618675867716,
+      "num_tokens": 897319125.0,
+      "step": 5064
+    },
+    {
+      "entropy": 0.7159998863935471,
+      "epoch": 1.5667103943205494,
+      "grad_norm": 0.15537752211093903,
+      "learning_rate": 6.385324482261597e-05,
+      "loss": 0.6948,
+      "mean_token_accuracy": 0.8313898133734862,
+      "num_tokens": 899443900.0,
+      "step": 5076
+    },
+    {
+      "entropy": 0.7135510904093584,
+      "epoch": 1.5704143838259124,
+      "grad_norm": 0.15910863876342773,
+      "learning_rate": 6.369903234946003e-05,
+      "loss": 0.6941,
+      "mean_token_accuracy": 0.8314098492264748,
+      "num_tokens": 901571980.0,
+      "step": 5088
+    },
+    {
+      "entropy": 0.7235615576306978,
+      "epoch": 1.5741183733312756,
+      "grad_norm": 0.15910740196704865,
+      "learning_rate": 6.354467885782614e-05,
+      "loss": 0.7035,
+      "mean_token_accuracy": 0.8294711311658224,
+      "num_tokens": 903733303.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.725119836628437,
+      "epoch": 1.5778223628366388,
+      "grad_norm": 0.15643955767154694,
+      "learning_rate": 6.339018593663633e-05,
+      "loss": 0.7048,
+      "mean_token_accuracy": 0.8289666138589382,
+      "num_tokens": 905831645.0,
+      "step": 5112
+    },
+    {
+      "entropy": 0.7218106302122275,
+      "epoch": 1.5815263523420016,
+      "grad_norm": 0.15788429975509644,
+      "learning_rate": 6.323555517624792e-05,
+      "loss": 0.6997,
+      "mean_token_accuracy": 0.8302192588647207,
+      "num_tokens": 907975789.0,
+      "step": 5124
+    },
+    {
+      "entropy": 0.7207488107184569,
+      "epoch": 1.5852303418473648,
+      "grad_norm": 0.17031899094581604,
+      "learning_rate": 6.308078816843721e-05,
+      "loss": 0.6983,
+      "mean_token_accuracy": 0.830046666165193,
+      "num_tokens": 910106098.0,
+      "step": 5136
+    },
+    {
+      "entropy": 0.7064071434239546,
+      "epoch": 1.588934331352728,
+      "grad_norm": 0.16523528099060059,
+      "learning_rate": 6.292588650638298e-05,
+      "loss": 0.6825,
+      "mean_token_accuracy": 0.8339592938621839,
+      "num_tokens": 912222945.0,
+      "step": 5148
+    },
+    {
+      "entropy": 0.6991514513889948,
+      "epoch": 1.5926383208580908,
+      "grad_norm": 0.1478477269411087,
+      "learning_rate": 6.277085178465015e-05,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.8346445349355539,
+      "num_tokens": 914355424.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.7049112543463707,
+      "epoch": 1.596342310363454,
+      "grad_norm": 0.16001063585281372,
+      "learning_rate": 6.261568559917336e-05,
+      "loss": 0.6867,
+      "mean_token_accuracy": 0.8334415753682455,
+      "num_tokens": 916470693.0,
+      "step": 5172
+    },
+    {
+      "entropy": 0.7129873298108578,
+      "epoch": 1.6000462998688172,
+      "grad_norm": 0.15970784425735474,
+      "learning_rate": 6.24603895472406e-05,
+      "loss": 0.6907,
+      "mean_token_accuracy": 0.8320851363241673,
+      "num_tokens": 918571794.0,
+      "step": 5184
+    },
+    {
+      "entropy": 0.6842332072556019,
+      "epoch": 1.60375028937418,
+      "grad_norm": 0.15537609159946442,
+      "learning_rate": 6.230496522747666e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8387776849170526,
+      "num_tokens": 920723475.0,
+      "step": 5196
+    },
+    {
+      "entropy": 0.7234001358350118,
+      "epoch": 1.6074542788795432,
+      "grad_norm": 0.1752452850341797,
+      "learning_rate": 6.214941423982674e-05,
+      "loss": 0.7009,
+      "mean_token_accuracy": 0.829645175486803,
+      "num_tokens": 922804826.0,
+      "step": 5208
+    },
+    {
+      "entropy": 0.7220541623731455,
+      "epoch": 1.6111582683849064,
+      "grad_norm": 0.2073037028312683,
+      "learning_rate": 6.199373818553996e-05,
+      "loss": 0.6998,
+      "mean_token_accuracy": 0.8298226619760195,
+      "num_tokens": 924951733.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.7145507372915745,
+      "epoch": 1.6148622578902692,
+      "grad_norm": 0.15807439386844635,
+      "learning_rate": 6.183793866715285e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.8318213013311228,
+      "num_tokens": 927091486.0,
+      "step": 5232
+    },
+    {
+      "entropy": 0.7239471996823946,
+      "epoch": 1.6185662473956324,
+      "grad_norm": 0.14956313371658325,
+      "learning_rate": 6.168201728847298e-05,
+      "loss": 0.7061,
+      "mean_token_accuracy": 0.829018946737051,
+      "num_tokens": 929222367.0,
+      "step": 5244
+    },
+    {
+      "entropy": 0.7157949171960354,
+      "epoch": 1.6222702369009956,
+      "grad_norm": 0.15014518797397614,
+      "learning_rate": 6.152597565456225e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.831564124673605,
+      "num_tokens": 931355312.0,
+      "step": 5256
+    },
+    {
+      "entropy": 0.7077574829260508,
+      "epoch": 1.6259742264063584,
+      "grad_norm": 0.1514291763305664,
+      "learning_rate": 6.136981537172054e-05,
+      "loss": 0.6871,
+      "mean_token_accuracy": 0.8334880148371061,
+      "num_tokens": 933487638.0,
+      "step": 5268
+    },
+    {
+      "entropy": 0.7066128787895044,
+      "epoch": 1.6296782159117216,
+      "grad_norm": 0.1583339273929596,
+      "learning_rate": 6.121353804746907e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.8333603926002979,
+      "num_tokens": 935600348.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.716248465081056,
+      "epoch": 1.6333822054170848,
+      "grad_norm": 0.15541380643844604,
+      "learning_rate": 6.105714529053391e-05,
+      "loss": 0.6937,
+      "mean_token_accuracy": 0.8314491584897041,
+      "num_tokens": 937702410.0,
+      "step": 5292
+    },
+    {
+      "entropy": 0.7185319637258848,
+      "epoch": 1.6370861949224476,
+      "grad_norm": 0.1648872047662735,
+      "learning_rate": 6.090063871082941e-05,
+      "loss": 0.698,
+      "mean_token_accuracy": 0.8299818920592467,
+      "num_tokens": 939791729.0,
+      "step": 5304
+    },
+    {
+      "entropy": 0.7117224608858427,
+      "epoch": 1.6407901844278108,
+      "grad_norm": 0.16068147122859955,
+      "learning_rate": 6.0744019919441564e-05,
+      "loss": 0.6932,
+      "mean_token_accuracy": 0.8325706832110882,
+      "num_tokens": 941926316.0,
+      "step": 5316
+    },
+    {
+      "entropy": 0.703471340239048,
+      "epoch": 1.644494173933174,
+      "grad_norm": 0.15285624563694,
+      "learning_rate": 6.058729052861156e-05,
+      "loss": 0.6823,
+      "mean_token_accuracy": 0.8336320606370767,
+      "num_tokens": 944021627.0,
+      "step": 5328
+    },
+    {
+      "entropy": 0.7178343534469604,
+      "epoch": 1.648198163438537,
+      "grad_norm": 0.15492397546768188,
+      "learning_rate": 6.043045215171903e-05,
+      "loss": 0.6949,
+      "mean_token_accuracy": 0.8309007274607817,
+      "num_tokens": 946154913.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.710145698239406,
+      "epoch": 1.6519021529439,
+      "grad_norm": 0.15523739159107208,
+      "learning_rate": 6.0273506403265543e-05,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.832029789686203,
+      "num_tokens": 948306687.0,
+      "step": 5352
+    },
+    {
+      "entropy": 0.7194016513725122,
+      "epoch": 1.6556061424492632,
+      "grad_norm": 0.15659694373607635,
+      "learning_rate": 6.0116454898857974e-05,
+      "loss": 0.6978,
+      "mean_token_accuracy": 0.8309724852442741,
+      "num_tokens": 950424446.0,
+      "step": 5364
+    },
+    {
+      "entropy": 0.6990312064687411,
+      "epoch": 1.6593101319546262,
+      "grad_norm": 0.1696748435497284,
+      "learning_rate": 5.99592992551918e-05,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.835218321532011,
+      "num_tokens": 952534193.0,
+      "step": 5376
+    },
+    {
+      "entropy": 0.7059058484931787,
+      "epoch": 1.6630141214599892,
+      "grad_norm": 0.16559961438179016,
+      "learning_rate": 5.98020410900346e-05,
+      "loss": 0.6848,
+      "mean_token_accuracy": 0.8346103342870871,
+      "num_tokens": 954639431.0,
+      "step": 5388
+    },
+    {
+      "entropy": 0.7005784672995409,
+      "epoch": 1.6667181109653524,
+      "grad_norm": 0.1537712663412094,
+      "learning_rate": 5.964468202220919e-05,
+      "loss": 0.6797,
+      "mean_token_accuracy": 0.8347668411831061,
+      "num_tokens": 956782039.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.7165911011397839,
+      "epoch": 1.6704221004707154,
+      "grad_norm": 0.16163314878940582,
+      "learning_rate": 5.9487223671577206e-05,
+      "loss": 0.6932,
+      "mean_token_accuracy": 0.8314206625024477,
+      "num_tokens": 958932938.0,
+      "step": 5412
+    },
+    {
+      "entropy": 0.7195808986822764,
+      "epoch": 1.6741260899760784,
+      "grad_norm": 0.148748978972435,
+      "learning_rate": 5.932966765902225e-05,
+      "loss": 0.6988,
+      "mean_token_accuracy": 0.8302495119472345,
+      "num_tokens": 961062826.0,
+      "step": 5424
+    },
+    {
+      "entropy": 0.7168962607781092,
+      "epoch": 1.6778300794814416,
+      "grad_norm": 0.15100803971290588,
+      "learning_rate": 5.917201560643323e-05,
+      "loss": 0.6955,
+      "mean_token_accuracy": 0.8308056443929672,
+      "num_tokens": 963169661.0,
+      "step": 5436
+    },
+    {
+      "entropy": 0.7153752135733763,
+      "epoch": 1.6815340689868046,
+      "grad_norm": 0.17240118980407715,
+      "learning_rate": 5.901426913668777e-05,
+      "loss": 0.6941,
+      "mean_token_accuracy": 0.8311421684920788,
+      "num_tokens": 965320235.0,
+      "step": 5448
+    },
+    {
+      "entropy": 0.7057120179136595,
+      "epoch": 1.6852380584921676,
+      "grad_norm": 0.16611547768115997,
+      "learning_rate": 5.8856429873635366e-05,
+      "loss": 0.6831,
+      "mean_token_accuracy": 0.8337946385145187,
+      "num_tokens": 967435851.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.7108418879409631,
+      "epoch": 1.6889420479975308,
+      "grad_norm": 0.14767956733703613,
+      "learning_rate": 5.869849944208076e-05,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.8330086718002955,
+      "num_tokens": 969573643.0,
+      "step": 5472
+    },
+    {
+      "entropy": 0.7108013468484083,
+      "epoch": 1.6926460375028938,
+      "grad_norm": 0.17264387011528015,
+      "learning_rate": 5.854047946776717e-05,
+      "loss": 0.6897,
+      "mean_token_accuracy": 0.8324159942567348,
+      "num_tokens": 971716883.0,
+      "step": 5484
+    },
+    {
+      "entropy": 0.7093252340952555,
+      "epoch": 1.6963500270082568,
+      "grad_norm": 0.16304221749305725,
+      "learning_rate": 5.8382371577359584e-05,
+      "loss": 0.6887,
+      "mean_token_accuracy": 0.8329907159010569,
+      "num_tokens": 973827983.0,
+      "step": 5496
+    },
+    {
+      "entropy": 0.7014658078551292,
+      "epoch": 1.70005401651362,
+      "grad_norm": 0.15241508185863495,
+      "learning_rate": 5.8224177398428016e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.8354054316878319,
+      "num_tokens": 975953202.0,
+      "step": 5508
+    },
+    {
+      "entropy": 0.7008350156247616,
+      "epoch": 1.703758006018983,
+      "grad_norm": 0.1691211462020874,
+      "learning_rate": 5.8065898559430706e-05,
+      "loss": 0.6805,
+      "mean_token_accuracy": 0.834287978708744,
+      "num_tokens": 978114142.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.7018342142303785,
+      "epoch": 1.707461995524346,
+      "grad_norm": 0.1586218923330307,
+      "learning_rate": 5.790753668969742e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.8338803065319856,
+      "num_tokens": 980256061.0,
+      "step": 5532
+    },
+    {
+      "entropy": 0.7026222559312979,
+      "epoch": 1.7111659850297092,
+      "grad_norm": 0.15604816377162933,
+      "learning_rate": 5.7749093419412626e-05,
+      "loss": 0.6788,
+      "mean_token_accuracy": 0.8349755468467871,
+      "num_tokens": 982408166.0,
+      "step": 5544
+    },
+    {
+      "entropy": 0.6918922699987888,
+      "epoch": 1.7148699745350722,
+      "grad_norm": 0.1606925129890442,
+      "learning_rate": 5.759057037959872e-05,
+      "loss": 0.6716,
+      "mean_token_accuracy": 0.8362010270357132,
+      "num_tokens": 984543638.0,
+      "step": 5556
+    },
+    {
+      "entropy": 0.7039717622101307,
+      "epoch": 1.7185739640404352,
+      "grad_norm": 0.1536424607038498,
+      "learning_rate": 5.7431969202099287e-05,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8334637098014355,
+      "num_tokens": 986649156.0,
+      "step": 5568
+    },
+    {
+      "entropy": 0.7010701584319273,
+      "epoch": 1.7222779535457984,
+      "grad_norm": 0.1718849092721939,
+      "learning_rate": 5.727329151956225e-05,
+      "loss": 0.6789,
+      "mean_token_accuracy": 0.8351405846575896,
+      "num_tokens": 988730105.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.7192123532295227,
+      "epoch": 1.7259819430511614,
+      "grad_norm": 0.15781332552433014,
+      "learning_rate": 5.711453896542307e-05,
+      "loss": 0.6987,
+      "mean_token_accuracy": 0.8300843487183253,
+      "num_tokens": 990834861.0,
+      "step": 5592
+    },
+    {
+      "entropy": 0.7151969609161218,
+      "epoch": 1.7296859325565244,
+      "grad_norm": 0.15374763309955597,
+      "learning_rate": 5.695571317388794e-05,
+      "loss": 0.6951,
+      "mean_token_accuracy": 0.8305746080974737,
+      "num_tokens": 992957801.0,
+      "step": 5604
+    },
+    {
+      "entropy": 0.7098864167928696,
+      "epoch": 1.7333899220618876,
+      "grad_norm": 0.18953058123588562,
+      "learning_rate": 5.679681577991694e-05,
+      "loss": 0.6886,
+      "mean_token_accuracy": 0.8322906543811163,
+      "num_tokens": 995069535.0,
+      "step": 5616
+    },
+    {
+      "entropy": 0.6933085806667805,
+      "epoch": 1.7370939115672506,
+      "grad_norm": 0.16794349253177643,
+      "learning_rate": 5.6637848419207305e-05,
+      "loss": 0.674,
+      "mean_token_accuracy": 0.8359851129353046,
+      "num_tokens": 997197992.0,
+      "step": 5628
+    },
+    {
+      "entropy": 0.7013748064637184,
+      "epoch": 1.7407979010726136,
+      "grad_norm": 0.1656210571527481,
+      "learning_rate": 5.6478812728176435e-05,
+      "loss": 0.6782,
+      "mean_token_accuracy": 0.8345577555398146,
+      "num_tokens": 999326233.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.716081328690052,
+      "epoch": 1.7445018905779768,
+      "grad_norm": 0.16797126829624176,
+      "learning_rate": 5.631971034394515e-05,
+      "loss": 0.6944,
+      "mean_token_accuracy": 0.8314446931083997,
+      "num_tokens": 1001429533.0,
+      "step": 5652
+    },
+    {
+      "entropy": 0.7122478174666563,
+      "epoch": 1.7482058800833398,
+      "grad_norm": 0.167856827378273,
+      "learning_rate": 5.616054290432082e-05,
+      "loss": 0.6916,
+      "mean_token_accuracy": 0.8324559306104978,
+      "num_tokens": 1003576389.0,
+      "step": 5664
+    },
+    {
+      "entropy": 0.7033972653249899,
+      "epoch": 1.7519098695887028,
+      "grad_norm": 0.15445369482040405,
+      "learning_rate": 5.6001312047780486e-05,
+      "loss": 0.682,
+      "mean_token_accuracy": 0.8342355067531267,
+      "num_tokens": 1005709667.0,
+      "step": 5676
+    },
+    {
+      "entropy": 0.6920518179734548,
+      "epoch": 1.755613859094066,
+      "grad_norm": 0.1603931039571762,
+      "learning_rate": 5.584201941345402e-05,
+      "loss": 0.6723,
+      "mean_token_accuracy": 0.8359791164596876,
+      "num_tokens": 1007876576.0,
+      "step": 5688
+    },
+    {
+      "entropy": 0.7032434691985449,
+      "epoch": 1.759317848599429,
+      "grad_norm": 0.15814517438411713,
+      "learning_rate": 5.568266664110722e-05,
+      "loss": 0.6818,
+      "mean_token_accuracy": 0.8343645632266998,
+      "num_tokens": 1010006821.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.710328691949447,
+      "epoch": 1.763021838104792,
+      "grad_norm": 0.17391884326934814,
+      "learning_rate": 5.552325537112497e-05,
+      "loss": 0.6875,
+      "mean_token_accuracy": 0.833385648826758,
+      "num_tokens": 1012148154.0,
+      "step": 5712
+    },
+    {
+      "entropy": 0.7074959563712279,
+      "epoch": 1.7667258276101552,
+      "grad_norm": 0.1595250517129898,
+      "learning_rate": 5.53637872444943e-05,
+      "loss": 0.688,
+      "mean_token_accuracy": 0.8327213364342848,
+      "num_tokens": 1014260818.0,
+      "step": 5724
+    },
+    {
+      "entropy": 0.6988821178674698,
+      "epoch": 1.7704298171155182,
+      "grad_norm": 0.16727086901664734,
+      "learning_rate": 5.5204263902787564e-05,
+      "loss": 0.6771,
+      "mean_token_accuracy": 0.8350557560722033,
+      "num_tokens": 1016337244.0,
+      "step": 5736
+    },
+    {
+      "entropy": 0.7048179631431898,
+      "epoch": 1.7741338066208812,
+      "grad_norm": 0.16889260709285736,
+      "learning_rate": 5.504468698814548e-05,
+      "loss": 0.6846,
+      "mean_token_accuracy": 0.8338587942222754,
+      "num_tokens": 1018453592.0,
+      "step": 5748
+    },
+    {
+      "entropy": 0.7078387662768364,
+      "epoch": 1.7778377961262444,
+      "grad_norm": 0.16895486414432526,
+      "learning_rate": 5.4885058143260227e-05,
+      "loss": 0.6869,
+      "mean_token_accuracy": 0.832654087493817,
+      "num_tokens": 1020585383.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.6954544944067796,
+      "epoch": 1.7815417856316074,
+      "grad_norm": 0.1606086641550064,
+      "learning_rate": 5.47253790113586e-05,
+      "loss": 0.6744,
+      "mean_token_accuracy": 0.8354161145786444,
+      "num_tokens": 1022716006.0,
+      "step": 5772
+    },
+    {
+      "entropy": 0.6906781146923701,
+      "epoch": 1.7852457751369704,
+      "grad_norm": 0.1552080512046814,
+      "learning_rate": 5.456565123618503e-05,
+      "loss": 0.6685,
+      "mean_token_accuracy": 0.8372346547742685,
+      "num_tokens": 1024830212.0,
+      "step": 5784
+    },
+    {
+      "entropy": 0.7113202723364035,
+      "epoch": 1.7889497646423336,
+      "grad_norm": 0.16079287230968475,
+      "learning_rate": 5.440587646198469e-05,
+      "loss": 0.6915,
+      "mean_token_accuracy": 0.8320710522433122,
+      "num_tokens": 1026944068.0,
+      "step": 5796
+    },
+    {
+      "entropy": 0.6925230113168558,
+      "epoch": 1.7926537541476966,
+      "grad_norm": 0.1665734201669693,
+      "learning_rate": 5.424605633348655e-05,
+      "loss": 0.671,
+      "mean_token_accuracy": 0.8368867188692093,
+      "num_tokens": 1029077844.0,
+      "step": 5808
+    },
+    {
+      "entropy": 0.7204868520299593,
+      "epoch": 1.7963577436530596,
+      "grad_norm": 0.1540994644165039,
+      "learning_rate": 5.408619249588644e-05,
+      "loss": 0.6991,
+      "mean_token_accuracy": 0.8302427244683107,
+      "num_tokens": 1031208806.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.6986957266926765,
+      "epoch": 1.8000617331584228,
+      "grad_norm": 0.15892110764980316,
+      "learning_rate": 5.392628659483021e-05,
+      "loss": 0.6767,
+      "mean_token_accuracy": 0.8347699269652367,
+      "num_tokens": 1033383899.0,
+      "step": 5832
+    },
+    {
+      "entropy": 0.7196601120134195,
+      "epoch": 1.8037657226637858,
+      "grad_norm": 0.16495993733406067,
+      "learning_rate": 5.3766340276396646e-05,
+      "loss": 0.7003,
+      "mean_token_accuracy": 0.8301363748808702,
+      "num_tokens": 1035511527.0,
+      "step": 5844
+    },
+    {
+      "entropy": 0.7001743813355764,
+      "epoch": 1.8074697121691488,
+      "grad_norm": 0.16940170526504517,
+      "learning_rate": 5.3606355187080595e-05,
+      "loss": 0.6812,
+      "mean_token_accuracy": 0.8345416759451231,
+      "num_tokens": 1037619457.0,
+      "step": 5856
+    },
+    {
+      "entropy": 0.7112437374889851,
+      "epoch": 1.811173701674512,
+      "grad_norm": 0.1692638248205185,
+      "learning_rate": 5.344633297377604e-05,
+      "loss": 0.6891,
+      "mean_token_accuracy": 0.832543725768725,
+      "num_tokens": 1039734230.0,
+      "step": 5868
+    },
+    {
+      "entropy": 0.7141083081563314,
+      "epoch": 1.814877691179875,
+      "grad_norm": 0.15631647408008575,
+      "learning_rate": 5.328627528375909e-05,
+      "loss": 0.6911,
+      "mean_token_accuracy": 0.832301444063584,
+      "num_tokens": 1041893850.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.7041296027600765,
+      "epoch": 1.818581680685238,
+      "grad_norm": 0.16098694503307343,
+      "learning_rate": 5.312618376467111e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.8333825121323267,
+      "num_tokens": 1044042558.0,
+      "step": 5892
+    },
+    {
+      "entropy": 0.691443515320619,
+      "epoch": 1.8222856701906012,
+      "grad_norm": 0.1699233502149582,
+      "learning_rate": 5.2966060064501645e-05,
+      "loss": 0.6704,
+      "mean_token_accuracy": 0.8364453675846258,
+      "num_tokens": 1046172232.0,
+      "step": 5904
+    },
+    {
+      "entropy": 0.7075358144938946,
+      "epoch": 1.8259896596959642,
+      "grad_norm": 0.1754993498325348,
+      "learning_rate": 5.280590583157152e-05,
+      "loss": 0.685,
+      "mean_token_accuracy": 0.8332440021137396,
+      "num_tokens": 1048284789.0,
+      "step": 5916
+    },
+    {
+      "entropy": 0.7079668007791042,
+      "epoch": 1.8296936492013272,
+      "grad_norm": 0.1584164947271347,
+      "learning_rate": 5.264572271451591e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.8330427743494511,
+      "num_tokens": 1050431870.0,
+      "step": 5928
+    },
+    {
+      "entropy": 0.7003171257674694,
+      "epoch": 1.8333976387066904,
+      "grad_norm": 0.15958057343959808,
+      "learning_rate": 5.248551236226724e-05,
+      "loss": 0.6776,
+      "mean_token_accuracy": 0.8346174930532774,
+      "num_tokens": 1052556525.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.7011785494784514,
+      "epoch": 1.8371016282120534,
+      "grad_norm": 0.15794256329536438,
+      "learning_rate": 5.232527642403841e-05,
+      "loss": 0.6802,
+      "mean_token_accuracy": 0.834118609627088,
+      "num_tokens": 1054656122.0,
+      "step": 5952
+    },
+    {
+      "entropy": 0.7017668796082338,
+      "epoch": 1.8408056177174164,
+      "grad_norm": 0.17382527887821198,
+      "learning_rate": 5.216501654930561e-05,
+      "loss": 0.6835,
+      "mean_token_accuracy": 0.8327813086410364,
+      "num_tokens": 1056796572.0,
+      "step": 5964
+    },
+    {
+      "entropy": 0.7102235816419125,
+      "epoch": 1.8445096072227796,
+      "grad_norm": 0.16711148619651794,
+      "learning_rate": 5.200473438779146e-05,
+      "loss": 0.6892,
+      "mean_token_accuracy": 0.8322862597803274,
+      "num_tokens": 1058920924.0,
+      "step": 5976
+    },
+    {
+      "entropy": 0.6888363808393478,
+      "epoch": 1.8482135967281426,
+      "grad_norm": 0.1576377898454666,
+      "learning_rate": 5.1844431589448025e-05,
+      "loss": 0.6666,
+      "mean_token_accuracy": 0.8380541230241457,
+      "num_tokens": 1061076284.0,
+      "step": 5988
+    },
+    {
+      "entropy": 0.7103967430690924,
+      "epoch": 1.8519175862335056,
+      "grad_norm": 0.16509290039539337,
+      "learning_rate": 5.1684109804439774e-05,
+      "loss": 0.6879,
+      "mean_token_accuracy": 0.8324746452271938,
+      "num_tokens": 1063152002.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.6895517694453398,
+      "epoch": 1.8556215757388688,
+      "grad_norm": 0.17304864525794983,
+      "learning_rate": 5.152377068312665e-05,
+      "loss": 0.6677,
+      "mean_token_accuracy": 0.8372837752103806,
+      "num_tokens": 1065304104.0,
+      "step": 6012
+    },
+    {
+      "entropy": 0.7109543569386005,
+      "epoch": 1.8593255652442318,
+      "grad_norm": 0.16532959043979645,
+      "learning_rate": 5.1363415876047036e-05,
+      "loss": 0.6926,
+      "mean_token_accuracy": 0.8319398574531078,
+      "num_tokens": 1067404193.0,
+      "step": 6024
+    },
+    {
+      "entropy": 0.7061434425413609,
+      "epoch": 1.8630295547495948,
+      "grad_norm": 0.1592184454202652,
+      "learning_rate": 5.1203047033900806e-05,
+      "loss": 0.6851,
+      "mean_token_accuracy": 0.8332473436991373,
+      "num_tokens": 1069527530.0,
+      "step": 6036
+    },
+    {
+      "entropy": 0.6931216232478619,
+      "epoch": 1.866733544254958,
+      "grad_norm": 0.15798227488994598,
+      "learning_rate": 5.10426658075323e-05,
+      "loss": 0.6708,
+      "mean_token_accuracy": 0.8361308053135872,
+      "num_tokens": 1071666406.0,
+      "step": 6048
+    },
+    {
+      "entropy": 0.6993834227323532,
+      "epoch": 1.870437533760321,
+      "grad_norm": 0.16762305796146393,
+      "learning_rate": 5.088227384791332e-05,
+      "loss": 0.6798,
+      "mean_token_accuracy": 0.8347605802118778,
+      "num_tokens": 1073823299.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.6961913978060087,
+      "epoch": 1.874141523265684,
+      "grad_norm": 0.1665721982717514,
+      "learning_rate": 5.072187280612621e-05,
+      "loss": 0.6726,
+      "mean_token_accuracy": 0.8361902994414171,
+      "num_tokens": 1075944213.0,
+      "step": 6072
+    },
+    {
+      "entropy": 0.702531552563111,
+      "epoch": 1.8778455127710472,
+      "grad_norm": 0.16553518176078796,
+      "learning_rate": 5.056146433334676e-05,
+      "loss": 0.6804,
+      "mean_token_accuracy": 0.8344750516116619,
+      "num_tokens": 1078033274.0,
+      "step": 6084
+    },
+    {
+      "entropy": 0.7059743218123913,
+      "epoch": 1.8815495022764104,
+      "grad_norm": 0.15529991686344147,
+      "learning_rate": 5.0401050080827297e-05,
+      "loss": 0.6855,
+      "mean_token_accuracy": 0.8335259656111399,
+      "num_tokens": 1080157323.0,
+      "step": 6096
+    },
+    {
+      "entropy": 0.7024221407870451,
+      "epoch": 1.8852534917817731,
+      "grad_norm": 0.15887407958507538,
+      "learning_rate": 5.024063169987958e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.8346465714275837,
+      "num_tokens": 1082289229.0,
+      "step": 6108
+    },
+    {
+      "entropy": 0.7151320551832517,
+      "epoch": 1.8889574812871364,
+      "grad_norm": 0.1628350168466568,
+      "learning_rate": 5.008021084185791e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.8322754862407843,
+      "num_tokens": 1084403486.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.7036687669654688,
+      "epoch": 1.8926614707924996,
+      "grad_norm": 0.16994571685791016,
+      "learning_rate": 4.99197891581421e-05,
+      "loss": 0.6836,
+      "mean_token_accuracy": 0.8344013380507628,
+      "num_tokens": 1086516462.0,
+      "step": 6132
+    },
+    {
+      "entropy": 0.7015910123785337,
+      "epoch": 1.8963654602978623,
+      "grad_norm": 0.15207666158676147,
+      "learning_rate": 4.975936830012043e-05,
+      "loss": 0.681,
+      "mean_token_accuracy": 0.8345483938852946,
+      "num_tokens": 1088624801.0,
+      "step": 6144
+    },
+    {
+      "entropy": 0.7084349157909552,
+      "epoch": 1.9000694498032256,
+      "grad_norm": 0.15862888097763062,
+      "learning_rate": 4.9598949919172715e-05,
+      "loss": 0.6858,
+      "mean_token_accuracy": 0.8337940263251463,
+      "num_tokens": 1090798611.0,
+      "step": 6156
+    },
+    {
+      "entropy": 0.6984652889271578,
+      "epoch": 1.9037734393085888,
+      "grad_norm": 0.16308139264583588,
+      "learning_rate": 4.9438535666653236e-05,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.8351361577709516,
+      "num_tokens": 1092940043.0,
+      "step": 6168
+    },
+    {
+      "entropy": 0.7053580147524675,
+      "epoch": 1.9074774288139515,
+      "grad_norm": 0.17202384769916534,
+      "learning_rate": 4.927812719387378e-05,
+      "loss": 0.6823,
+      "mean_token_accuracy": 0.8341307466228803,
+      "num_tokens": 1095068183.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.7106709815561771,
+      "epoch": 1.9111814183193148,
+      "grad_norm": 0.1545022875070572,
+      "learning_rate": 4.9117726152086694e-05,
+      "loss": 0.689,
+      "mean_token_accuracy": 0.8321913021306196,
+      "num_tokens": 1097208618.0,
+      "step": 6192
+    },
+    {
+      "entropy": 0.7134841966132323,
+      "epoch": 1.914885407824678,
+      "grad_norm": 0.17459839582443237,
+      "learning_rate": 4.895733419246772e-05,
+      "loss": 0.6918,
+      "mean_token_accuracy": 0.8316825156410536,
+      "num_tokens": 1099320273.0,
+      "step": 6204
+    },
+    {
+      "entropy": 0.7126207016408443,
+      "epoch": 1.9185893973300407,
+      "grad_norm": 0.1696072369813919,
+      "learning_rate": 4.879695296609921e-05,
+      "loss": 0.6921,
+      "mean_token_accuracy": 0.8322778431077799,
+      "num_tokens": 1101435574.0,
+      "step": 6216
+    },
+    {
+      "entropy": 0.6995201781392097,
+      "epoch": 1.922293386835404,
+      "grad_norm": 0.16167153418064117,
+      "learning_rate": 4.863658412395297e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.835229920844237,
+      "num_tokens": 1103574826.0,
+      "step": 6228
+    },
+    {
+      "entropy": 0.7126905384163061,
+      "epoch": 1.9259973763407672,
+      "grad_norm": 0.16883228719234467,
+      "learning_rate": 4.847622931687336e-05,
+      "loss": 0.6917,
+      "mean_token_accuracy": 0.8323102233310541,
+      "num_tokens": 1105701043.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.7085224725306034,
+      "epoch": 1.92970136584613,
+      "grad_norm": 0.1775846779346466,
+      "learning_rate": 4.831589019556024e-05,
+      "loss": 0.6842,
+      "mean_token_accuracy": 0.8335540195306143,
+      "num_tokens": 1107798646.0,
+      "step": 6252
+    },
+    {
+      "entropy": 0.7003039928774039,
+      "epoch": 1.9334053553514932,
+      "grad_norm": 0.15765723586082458,
+      "learning_rate": 4.815556841055198e-05,
+      "loss": 0.6806,
+      "mean_token_accuracy": 0.8348556930820147,
+      "num_tokens": 1109913551.0,
+      "step": 6264
+    },
+    {
+      "entropy": 0.7019528361658255,
+      "epoch": 1.9371093448568564,
+      "grad_norm": 0.15830156207084656,
+      "learning_rate": 4.799526561220855e-05,
+      "loss": 0.6835,
+      "mean_token_accuracy": 0.8335538630684217,
+      "num_tokens": 1112019237.0,
+      "step": 6276
+    },
+    {
+      "entropy": 0.7016400955617428,
+      "epoch": 1.9408133343622191,
+      "grad_norm": 0.18349109590053558,
+      "learning_rate": 4.7834983450694405e-05,
+      "loss": 0.6786,
+      "mean_token_accuracy": 0.8345780223608017,
+      "num_tokens": 1114144126.0,
+      "step": 6288
+    },
+    {
+      "entropy": 0.7051151817043623,
+      "epoch": 1.9445173238675824,
+      "grad_norm": 0.160055011510849,
+      "learning_rate": 4.767472357596159e-05,
+      "loss": 0.6865,
+      "mean_token_accuracy": 0.8335471215347449,
+      "num_tokens": 1116255204.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.6837548911571503,
+      "epoch": 1.9482213133729456,
+      "grad_norm": 0.1652652770280838,
+      "learning_rate": 4.751448763773275e-05,
+      "loss": 0.6624,
+      "mean_token_accuracy": 0.8393045626580715,
+      "num_tokens": 1118373731.0,
+      "step": 6312
+    },
+    {
+      "entropy": 0.6986660932501157,
+      "epoch": 1.9519253028783086,
+      "grad_norm": 0.16915340721607208,
+      "learning_rate": 4.735427728548412e-05,
+      "loss": 0.6762,
+      "mean_token_accuracy": 0.8349108224113783,
+      "num_tokens": 1120485114.0,
+      "step": 6324
+    },
+    {
+      "entropy": 0.7144566103816032,
+      "epoch": 1.9556292923836716,
+      "grad_norm": 0.16116300225257874,
+      "learning_rate": 4.7194094168428496e-05,
+      "loss": 0.6928,
+      "mean_token_accuracy": 0.8316774268945059,
+      "num_tokens": 1122622853.0,
+      "step": 6336
+    },
+    {
+      "entropy": 0.7057266781727473,
+      "epoch": 1.9593332818890348,
+      "grad_norm": 0.16374769806861877,
+      "learning_rate": 4.7033939935498366e-05,
+      "loss": 0.6834,
+      "mean_token_accuracy": 0.8338124553362528,
+      "num_tokens": 1124740405.0,
+      "step": 6348
+    },
+    {
+      "entropy": 0.7115621020396551,
+      "epoch": 1.9630372713943978,
+      "grad_norm": 0.17385482788085938,
+      "learning_rate": 4.6873816235328896e-05,
+      "loss": 0.6891,
+      "mean_token_accuracy": 0.8324099443852901,
+      "num_tokens": 1126901966.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.6988142629464468,
+      "epoch": 1.9667412608997608,
+      "grad_norm": 0.17653614282608032,
+      "learning_rate": 4.6713724716240915e-05,
+      "loss": 0.6779,
+      "mean_token_accuracy": 0.8349249089757601,
+      "num_tokens": 1128993855.0,
+      "step": 6372
+    },
+    {
+      "entropy": 0.7140787703295549,
+      "epoch": 1.970445250405124,
+      "grad_norm": 0.1632198989391327,
+      "learning_rate": 4.6553667026223975e-05,
+      "loss": 0.6932,
+      "mean_token_accuracy": 0.8312925559779009,
+      "num_tokens": 1131129546.0,
+      "step": 6384
+    },
+    {
+      "entropy": 0.7079446117083231,
+      "epoch": 1.974149239910487,
+      "grad_norm": 0.16609056293964386,
+      "learning_rate": 4.6393644812919416e-05,
+      "loss": 0.6874,
+      "mean_token_accuracy": 0.8327439626057943,
+      "num_tokens": 1133241525.0,
+      "step": 6396
+    },
+    {
+      "entropy": 0.6874182323614756,
+      "epoch": 1.97785322941585,
+      "grad_norm": 0.15895400941371918,
+      "learning_rate": 4.623365972360337e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.8374094466368357,
+      "num_tokens": 1135375918.0,
+      "step": 6408
+    },
+    {
+      "entropy": 0.7057366843024889,
+      "epoch": 1.9815572189212132,
+      "grad_norm": 0.17353583872318268,
+      "learning_rate": 4.607371340516979e-05,
+      "loss": 0.6828,
+      "mean_token_accuracy": 0.8338442370295525,
+      "num_tokens": 1137469495.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.6886672539015611,
+      "epoch": 1.9852612084265762,
+      "grad_norm": 0.1662573218345642,
+      "learning_rate": 4.5913807504113556e-05,
+      "loss": 0.6675,
+      "mean_token_accuracy": 0.8377264539400736,
+      "num_tokens": 1139593432.0,
+      "step": 6432
+    },
+    {
+      "entropy": 0.692137303451697,
+      "epoch": 1.9889651979319392,
+      "grad_norm": 0.18068836629390717,
+      "learning_rate": 4.5753943666513455e-05,
+      "loss": 0.6698,
+      "mean_token_accuracy": 0.8362124636769295,
+      "num_tokens": 1141753650.0,
+      "step": 6444
+    },
+    {
+      "entropy": 0.697237261881431,
+      "epoch": 1.9926691874373024,
+      "grad_norm": 0.1671893298625946,
+      "learning_rate": 4.559412353801533e-05,
+      "loss": 0.6768,
+      "mean_token_accuracy": 0.8347001535197099,
+      "num_tokens": 1143892293.0,
+      "step": 6456
+    },
+    {
+      "entropy": 0.704800813148419,
+      "epoch": 1.9963731769426654,
+      "grad_norm": 0.1634693741798401,
+      "learning_rate": 4.543434876381497e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.833663385361433,
+      "num_tokens": 1145995636.0,
+      "step": 6468
+    },
+    {
+      "entropy": 0.6973251129718537,
+      "epoch": 2.0,
+      "grad_norm": 0.1913071721792221,
+      "learning_rate": 4.52746209886414e-05,
+      "loss": 0.6774,
+      "mean_token_accuracy": 0.8352246829803954,
+      "num_tokens": 1148075160.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.6958606342474619,
+      "epoch": 2.003703989505363,
+      "grad_norm": 0.15933266282081604,
+      "learning_rate": 4.5114941856739785e-05,
+      "loss": 0.6727,
+      "mean_token_accuracy": 0.8364654754598936,
+      "num_tokens": 1150241776.0,
+      "step": 6492
+    },
+    {
+      "entropy": 0.6805713040133318,
+      "epoch": 2.007407979010726,
+      "grad_norm": 0.17767716944217682,
+      "learning_rate": 4.495531301185453e-05,
+      "loss": 0.6577,
+      "mean_token_accuracy": 0.8394694328308105,
+      "num_tokens": 1152398185.0,
+      "step": 6504
+    },
+    {
+      "entropy": 0.6998436215023199,
+      "epoch": 2.011111968516089,
+      "grad_norm": 0.176164448261261,
+      "learning_rate": 4.479573609721244e-05,
+      "loss": 0.678,
+      "mean_token_accuracy": 0.8348252177238464,
+      "num_tokens": 1154508126.0,
+      "step": 6516
+    },
+    {
+      "entropy": 0.6891536228358746,
+      "epoch": 2.0148159580214524,
+      "grad_norm": 0.16503383219242096,
+      "learning_rate": 4.463621275550571e-05,
+      "loss": 0.6645,
+      "mean_token_accuracy": 0.8373865385850271,
+      "num_tokens": 1156620964.0,
+      "step": 6528
+    },
+    {
+      "entropy": 0.6943193326393763,
+      "epoch": 2.018519947526815,
+      "grad_norm": 0.17573916912078857,
+      "learning_rate": 4.447674462887503e-05,
+      "loss": 0.6714,
+      "mean_token_accuracy": 0.836496909459432,
+      "num_tokens": 1158754680.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.692933322240909,
+      "epoch": 2.0222239370321784,
+      "grad_norm": 0.16031506657600403,
+      "learning_rate": 4.431733335889278e-05,
+      "loss": 0.6682,
+      "mean_token_accuracy": 0.8367396021882693,
+      "num_tokens": 1160874718.0,
+      "step": 6552
+    },
+    {
+      "entropy": 0.6829059620698293,
+      "epoch": 2.0259279265375416,
+      "grad_norm": 0.16216310858726501,
+      "learning_rate": 4.4157980586545974e-05,
+      "loss": 0.6582,
+      "mean_token_accuracy": 0.8389946781098843,
+      "num_tokens": 1163031354.0,
+      "step": 6564
+    },
+    {
+      "entropy": 0.6946393934388956,
+      "epoch": 2.0296319160429044,
+      "grad_norm": 0.18271124362945557,
+      "learning_rate": 4.399868795221951e-05,
+      "loss": 0.6707,
+      "mean_token_accuracy": 0.8366646692156792,
+      "num_tokens": 1165145296.0,
+      "step": 6576
+    },
+    {
+      "entropy": 0.6902418856819471,
+      "epoch": 2.0333359055482676,
+      "grad_norm": 0.15688392519950867,
+      "learning_rate": 4.383945709567919e-05,
+      "loss": 0.6682,
+      "mean_token_accuracy": 0.8372057154774666,
+      "num_tokens": 1167269330.0,
+      "step": 6588
+    },
+    {
+      "entropy": 0.6856662034988403,
+      "epoch": 2.037039895053631,
+      "grad_norm": 0.16447409987449646,
+      "learning_rate": 4.368028965605486e-05,
+      "loss": 0.6626,
+      "mean_token_accuracy": 0.8383788081506888,
+      "num_tokens": 1169406342.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.6875134222209454,
+      "epoch": 2.0407438845589936,
+      "grad_norm": 0.1646708846092224,
+      "learning_rate": 4.352118727182358e-05,
+      "loss": 0.663,
+      "mean_token_accuracy": 0.8381359900037447,
+      "num_tokens": 1171551058.0,
+      "step": 6612
+    },
+    {
+      "entropy": 0.6901310036579767,
+      "epoch": 2.044447874064357,
+      "grad_norm": 0.17784608900547028,
+      "learning_rate": 4.3362151580792707e-05,
+      "loss": 0.6689,
+      "mean_token_accuracy": 0.836744820078214,
+      "num_tokens": 1173700001.0,
+      "step": 6624
+    },
+    {
+      "entropy": 0.688694529235363,
+      "epoch": 2.04815186356972,
+      "grad_norm": 0.16807569563388824,
+      "learning_rate": 4.320318422008306e-05,
+      "loss": 0.6639,
+      "mean_token_accuracy": 0.838087435811758,
+      "num_tokens": 1175826029.0,
+      "step": 6636
+    },
+    {
+      "entropy": 0.6816125065088272,
+      "epoch": 2.051855853075083,
+      "grad_norm": 0.16541939973831177,
+      "learning_rate": 4.304428682611208e-05,
+      "loss": 0.659,
+      "mean_token_accuracy": 0.8384276715417703,
+      "num_tokens": 1177964197.0,
+      "step": 6648
+    },
+    {
+      "entropy": 0.6754038743674755,
+      "epoch": 2.055559842580446,
+      "grad_norm": 0.1831735372543335,
+      "learning_rate": 4.288546103457694e-05,
+      "loss": 0.6514,
+      "mean_token_accuracy": 0.8410420827567577,
+      "num_tokens": 1180073560.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.6953004685540994,
+      "epoch": 2.059263832085809,
+      "grad_norm": 0.16756285727024078,
+      "learning_rate": 4.272670848043776e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.836195650200049,
+      "num_tokens": 1182227884.0,
+      "step": 6672
+    },
+    {
+      "entropy": 0.6704849277933439,
+      "epoch": 2.062967821591172,
+      "grad_norm": 0.16364024579524994,
+      "learning_rate": 4.256803079790071e-05,
+      "loss": 0.6504,
+      "mean_token_accuracy": 0.8407212706903616,
+      "num_tokens": 1184338894.0,
+      "step": 6684
+    },
+    {
+      "entropy": 0.6869217567145824,
+      "epoch": 2.066671811096535,
+      "grad_norm": 0.17596305906772614,
+      "learning_rate": 4.240942962040128e-05,
+      "loss": 0.6662,
+      "mean_token_accuracy": 0.8376976748307546,
+      "num_tokens": 1186473192.0,
+      "step": 6696
+    },
+    {
+      "entropy": 0.6874274127185345,
+      "epoch": 2.0703758006018984,
+      "grad_norm": 0.168784037232399,
+      "learning_rate": 4.22509065805874e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8380283936858177,
+      "num_tokens": 1188591972.0,
+      "step": 6708
+    },
+    {
+      "entropy": 0.6954148449003696,
+      "epoch": 2.074079790107261,
+      "grad_norm": 0.18569393455982208,
+      "learning_rate": 4.20924633103026e-05,
+      "loss": 0.6734,
+      "mean_token_accuracy": 0.8357961600025495,
+      "num_tokens": 1190735606.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.6922590111692747,
+      "epoch": 2.0777837796126244,
+      "grad_norm": 0.15867288410663605,
+      "learning_rate": 4.19341014405693e-05,
+      "loss": 0.6663,
+      "mean_token_accuracy": 0.8377314632137617,
+      "num_tokens": 1192850698.0,
+      "step": 6732
+    },
+    {
+      "entropy": 0.6765066559116045,
+      "epoch": 2.0814877691179876,
+      "grad_norm": 0.17693735659122467,
+      "learning_rate": 4.1775822601572e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.8403165886799494,
+      "num_tokens": 1194969343.0,
+      "step": 6744
+    },
+    {
+      "entropy": 0.6881164262692133,
+      "epoch": 2.0851917586233504,
+      "grad_norm": 0.17512989044189453,
+      "learning_rate": 4.1617628422640434e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8381856555740038,
+      "num_tokens": 1197090640.0,
+      "step": 6756
+    },
+    {
+      "entropy": 0.6754906934996446,
+      "epoch": 2.0888957481287136,
+      "grad_norm": 0.17044971883296967,
+      "learning_rate": 4.145952053223284e-05,
+      "loss": 0.6526,
+      "mean_token_accuracy": 0.8403135687112808,
+      "num_tokens": 1199225773.0,
+      "step": 6768
+    },
+    {
+      "entropy": 0.6851841695606709,
+      "epoch": 2.092599737634077,
+      "grad_norm": 0.16464471817016602,
+      "learning_rate": 4.130150055791926e-05,
+      "loss": 0.6623,
+      "mean_token_accuracy": 0.8387955824534098,
+      "num_tokens": 1201348698.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.6839038406809171,
+      "epoch": 2.0963037271394396,
+      "grad_norm": 0.17445310950279236,
+      "learning_rate": 4.114357012636465e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.8389583652218183,
+      "num_tokens": 1203501839.0,
+      "step": 6792
+    },
+    {
+      "entropy": 0.6837594596048197,
+      "epoch": 2.100007716644803,
+      "grad_norm": 0.16313707828521729,
+      "learning_rate": 4.0985730863312234e-05,
+      "loss": 0.6668,
+      "mean_token_accuracy": 0.8375067251423994,
+      "num_tokens": 1205659039.0,
+      "step": 6804
+    },
+    {
+      "entropy": 0.6823192238807678,
+      "epoch": 2.103711706150166,
+      "grad_norm": 0.18319004774093628,
+      "learning_rate": 4.082798439356677e-05,
+      "loss": 0.6571,
+      "mean_token_accuracy": 0.8394167373577753,
+      "num_tokens": 1207800655.0,
+      "step": 6816
+    },
+    {
+      "entropy": 0.6888291624685129,
+      "epoch": 2.1074156956555288,
+      "grad_norm": 0.16149847209453583,
+      "learning_rate": 4.0670332340977765e-05,
+      "loss": 0.6669,
+      "mean_token_accuracy": 0.8372188744445642,
+      "num_tokens": 1209942817.0,
+      "step": 6828
+    },
+    {
+      "entropy": 0.683031198879083,
+      "epoch": 2.111119685160892,
+      "grad_norm": 0.182169571518898,
+      "learning_rate": 4.051277632842281e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.8390312877794107,
+      "num_tokens": 1212050353.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.6798936774333318,
+      "epoch": 2.114823674666255,
+      "grad_norm": 0.16732676327228546,
+      "learning_rate": 4.0355317977790815e-05,
+      "loss": 0.6555,
+      "mean_token_accuracy": 0.8397704648474852,
+      "num_tokens": 1214180248.0,
+      "step": 6852
+    },
+    {
+      "entropy": 0.6978944800794125,
+      "epoch": 2.118527664171618,
+      "grad_norm": 0.16686294972896576,
+      "learning_rate": 4.019795890996543e-05,
+      "loss": 0.6753,
+      "mean_token_accuracy": 0.8350128419697285,
+      "num_tokens": 1216254353.0,
+      "step": 6864
+    },
+    {
+      "entropy": 0.6932212188839912,
+      "epoch": 2.122231653676981,
+      "grad_norm": 0.1676548272371292,
+      "learning_rate": 4.0040700744808204e-05,
+      "loss": 0.6702,
+      "mean_token_accuracy": 0.8364811092615128,
+      "num_tokens": 1218387247.0,
+      "step": 6876
+    },
+    {
+      "entropy": 0.6828142193456491,
+      "epoch": 2.1259356431823444,
+      "grad_norm": 0.16696545481681824,
+      "learning_rate": 3.988354510114203e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.8394288209577402,
+      "num_tokens": 1220553394.0,
+      "step": 6888
+    },
+    {
+      "entropy": 0.6806088127195835,
+      "epoch": 2.1296396326877076,
+      "grad_norm": 0.1975679099559784,
+      "learning_rate": 3.972649359673446e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.8398889203866323,
+      "num_tokens": 1222724624.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.6907341443002224,
+      "epoch": 2.1333436221930704,
+      "grad_norm": 0.16510289907455444,
+      "learning_rate": 3.956954784828098e-05,
+      "loss": 0.6697,
+      "mean_token_accuracy": 0.836446632941564,
+      "num_tokens": 1224844469.0,
+      "step": 6912
+    },
+    {
+      "entropy": 0.6722525469958782,
+      "epoch": 2.1370476116984336,
+      "grad_norm": 0.17981542646884918,
+      "learning_rate": 3.9412709471388446e-05,
+      "loss": 0.6504,
+      "mean_token_accuracy": 0.8412780488530794,
+      "num_tokens": 1226985482.0,
+      "step": 6924
+    },
+    {
+      "entropy": 0.6808725881079832,
+      "epoch": 2.140751601203797,
+      "grad_norm": 0.17366532981395721,
+      "learning_rate": 3.925598008055844e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.8392424657940865,
+      "num_tokens": 1229092079.0,
+      "step": 6936
+    },
+    {
+      "entropy": 0.6670869911710421,
+      "epoch": 2.1444555907091596,
+      "grad_norm": 0.1617341935634613,
+      "learning_rate": 3.90993612891706e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.842083066701889,
+      "num_tokens": 1231212442.0,
+      "step": 6948
+    },
+    {
+      "entropy": 0.6818137591083845,
+      "epoch": 2.148159580214523,
+      "grad_norm": 0.16834302246570587,
+      "learning_rate": 3.894285470946609e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.8394963393608729,
+      "num_tokens": 1233347325.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.6732950036724409,
+      "epoch": 2.151863569719886,
+      "grad_norm": 0.1831921488046646,
+      "learning_rate": 3.878646195253095e-05,
+      "loss": 0.649,
+      "mean_token_accuracy": 0.8414178465803465,
+      "num_tokens": 1235473249.0,
+      "step": 6972
+    },
+    {
+      "entropy": 0.6829912650088469,
+      "epoch": 2.155567559225249,
+      "grad_norm": 0.17452828586101532,
+      "learning_rate": 3.863018462827948e-05,
+      "loss": 0.6606,
+      "mean_token_accuracy": 0.8385308894018332,
+      "num_tokens": 1237606878.0,
+      "step": 6984
+    },
+    {
+      "entropy": 0.6933281992872556,
+      "epoch": 2.159271548730612,
+      "grad_norm": 0.1628619283437729,
+      "learning_rate": 3.847402434543777e-05,
+      "loss": 0.6721,
+      "mean_token_accuracy": 0.8357371079425017,
+      "num_tokens": 1239737890.0,
+      "step": 6996
+    },
+    {
+      "entropy": 0.6939899697899818,
+      "epoch": 2.162975538235975,
+      "grad_norm": 0.1830848604440689,
+      "learning_rate": 3.831798271152704e-05,
+      "loss": 0.6715,
+      "mean_token_accuracy": 0.8364391488333544,
+      "num_tokens": 1241820691.0,
+      "step": 7008
+    },
+    {
+      "entropy": 0.6830517227451006,
+      "epoch": 2.166679527741338,
+      "grad_norm": 0.1746508777141571,
+      "learning_rate": 3.816206133284716e-05,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.8387770056724548,
+      "num_tokens": 1243913932.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.6800741304953893,
+      "epoch": 2.170383517246701,
+      "grad_norm": 0.17165830731391907,
+      "learning_rate": 3.800626181446007e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.8394532613456249,
+      "num_tokens": 1246054747.0,
+      "step": 7032
+    },
+    {
+      "entropy": 0.6898703599969546,
+      "epoch": 2.1740875067520644,
+      "grad_norm": 0.17083138227462769,
+      "learning_rate": 3.7850585760173275e-05,
+      "loss": 0.6656,
+      "mean_token_accuracy": 0.837770772476991,
+      "num_tokens": 1248153486.0,
+      "step": 7044
+    },
+    {
+      "entropy": 0.6814120809237162,
+      "epoch": 2.177791496257427,
+      "grad_norm": 0.16952723264694214,
+      "learning_rate": 3.769503477252335e-05,
+      "loss": 0.6586,
+      "mean_token_accuracy": 0.8393961998323599,
+      "num_tokens": 1250281780.0,
+      "step": 7056
+    },
+    {
+      "entropy": 0.6871961392462254,
+      "epoch": 2.1814954857627904,
+      "grad_norm": 0.16225868463516235,
+      "learning_rate": 3.75396104527594e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8381969705224037,
+      "num_tokens": 1252424662.0,
+      "step": 7068
+    },
+    {
+      "entropy": 0.6821385646859804,
+      "epoch": 2.1851994752681536,
+      "grad_norm": 0.16517527401447296,
+      "learning_rate": 3.738431440082665e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.839236855506897,
+      "num_tokens": 1254557921.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.6806115495661894,
+      "epoch": 2.1889034647735164,
+      "grad_norm": 0.1751754730939865,
+      "learning_rate": 3.7229148215349884e-05,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.8389156634608904,
+      "num_tokens": 1256677173.0,
+      "step": 7092
+    },
+    {
+      "entropy": 0.6796896619101366,
+      "epoch": 2.1926074542788796,
+      "grad_norm": 0.18694189190864563,
+      "learning_rate": 3.707411349361705e-05,
+      "loss": 0.6551,
+      "mean_token_accuracy": 0.8396575612326463,
+      "num_tokens": 1258777752.0,
+      "step": 7104
+    },
+    {
+      "entropy": 0.6812881293396155,
+      "epoch": 2.196311443784243,
+      "grad_norm": 0.1656647026538849,
+      "learning_rate": 3.6919211831562806e-05,
+      "loss": 0.6573,
+      "mean_token_accuracy": 0.8394532203674316,
+      "num_tokens": 1260896416.0,
+      "step": 7116
+    },
+    {
+      "entropy": 0.6926145292818546,
+      "epoch": 2.2000154332896056,
+      "grad_norm": 0.1721578985452652,
+      "learning_rate": 3.676444482375208e-05,
+      "loss": 0.6692,
+      "mean_token_accuracy": 0.836610042800506,
+      "num_tokens": 1263001893.0,
+      "step": 7128
+    },
+    {
+      "entropy": 0.698531299829483,
+      "epoch": 2.203719422794969,
+      "grad_norm": 0.16790412366390228,
+      "learning_rate": 3.660981406336369e-05,
+      "loss": 0.6751,
+      "mean_token_accuracy": 0.8358333731691042,
+      "num_tokens": 1265170858.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.6893511265516281,
+      "epoch": 2.207423412300332,
+      "grad_norm": 0.17855829000473022,
+      "learning_rate": 3.645532114217387e-05,
+      "loss": 0.667,
+      "mean_token_accuracy": 0.837383933365345,
+      "num_tokens": 1267306591.0,
+      "step": 7152
+    },
+    {
+      "entropy": 0.6948994646469752,
+      "epoch": 2.211127401805695,
+      "grad_norm": 0.17504854500293732,
+      "learning_rate": 3.630096765053997e-05,
+      "loss": 0.6735,
+      "mean_token_accuracy": 0.8362486995756626,
+      "num_tokens": 1269448580.0,
+      "step": 7164
+    },
+    {
+      "entropy": 0.6904158741235733,
+      "epoch": 2.214831391311058,
+      "grad_norm": 0.16738073527812958,
+      "learning_rate": 3.614675517738405e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.8369380322595438,
+      "num_tokens": 1271571840.0,
+      "step": 7176
+    },
+    {
+      "entropy": 0.6801648748417696,
+      "epoch": 2.218535380816421,
+      "grad_norm": 0.18572407960891724,
+      "learning_rate": 3.599268531017646e-05,
+      "loss": 0.658,
+      "mean_token_accuracy": 0.8388838407893976,
+      "num_tokens": 1273701179.0,
+      "step": 7188
+    },
+    {
+      "entropy": 0.6940938420593739,
+      "epoch": 2.222239370321784,
+      "grad_norm": 0.1790064424276352,
+      "learning_rate": 3.583875963491964e-05,
+      "loss": 0.6714,
+      "mean_token_accuracy": 0.8359104084471861,
+      "num_tokens": 1275846470.0,
+      "step": 7200
+    },
+    {
+      "entropy": 0.685243288675944,
+      "epoch": 2.225943359827147,
+      "grad_norm": 0.17031192779541016,
+      "learning_rate": 3.568497973613166e-05,
+      "loss": 0.6625,
+      "mean_token_accuracy": 0.8384867298106352,
+      "num_tokens": 1277981074.0,
+      "step": 7212
+    },
+    {
+      "entropy": 0.6832674543062845,
+      "epoch": 2.2296473493325104,
+      "grad_norm": 0.17609767615795135,
+      "learning_rate": 3.5531347196829964e-05,
+      "loss": 0.6602,
+      "mean_token_accuracy": 0.8384842636684576,
+      "num_tokens": 1280123692.0,
+      "step": 7224
+    },
+    {
+      "entropy": 0.6736332066357136,
+      "epoch": 2.233351338837873,
+      "grad_norm": 0.18072204291820526,
+      "learning_rate": 3.537786359851506e-05,
+      "loss": 0.6482,
+      "mean_token_accuracy": 0.8411118065317472,
+      "num_tokens": 1282280311.0,
+      "step": 7236
+    },
+    {
+      "entropy": 0.6906834666927656,
+      "epoch": 2.2370553283432364,
+      "grad_norm": 0.16962793469429016,
+      "learning_rate": 3.52245305211543e-05,
+      "loss": 0.669,
+      "mean_token_accuracy": 0.8369547414282957,
+      "num_tokens": 1284394136.0,
+      "step": 7248
+    },
+    {
+      "entropy": 0.681410993138949,
+      "epoch": 2.2407593178485996,
+      "grad_norm": 0.1630384773015976,
+      "learning_rate": 3.5071349543165513e-05,
+      "loss": 0.6591,
+      "mean_token_accuracy": 0.8389986592034498,
+      "num_tokens": 1286525608.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.6817056524256865,
+      "epoch": 2.2444633073539624,
+      "grad_norm": 0.17839214205741882,
+      "learning_rate": 3.491832224140083e-05,
+      "loss": 0.6592,
+      "mean_token_accuracy": 0.8388963937759399,
+      "num_tokens": 1288627106.0,
+      "step": 7272
+    },
+    {
+      "entropy": 0.6757811804612478,
+      "epoch": 2.2481672968593256,
+      "grad_norm": 0.17169944941997528,
+      "learning_rate": 3.476545019113046e-05,
+      "loss": 0.6562,
+      "mean_token_accuracy": 0.8402783883114656,
+      "num_tokens": 1290730620.0,
+      "step": 7284
+    },
+    {
+      "entropy": 0.6782712750136852,
+      "epoch": 2.251871286364689,
+      "grad_norm": 0.17680226266384125,
+      "learning_rate": 3.4612734966026387e-05,
+      "loss": 0.6531,
+      "mean_token_accuracy": 0.8399842207630476,
+      "num_tokens": 1292839997.0,
+      "step": 7296
+    },
+    {
+      "entropy": 0.6741428785026073,
+      "epoch": 2.2555752758700516,
+      "grad_norm": 0.1740843504667282,
+      "learning_rate": 3.446017813814627e-05,
+      "loss": 0.6539,
+      "mean_token_accuracy": 0.8401829078793526,
+      "num_tokens": 1294968879.0,
+      "step": 7308
+    },
+    {
+      "entropy": 0.6835886935393015,
+      "epoch": 2.259279265375415,
+      "grad_norm": 0.16604800522327423,
+      "learning_rate": 3.430778127791723e-05,
+      "loss": 0.6604,
+      "mean_token_accuracy": 0.8385746528704962,
+      "num_tokens": 1297112365.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.6774425928791364,
+      "epoch": 2.262983254880778,
+      "grad_norm": 0.16330775618553162,
+      "learning_rate": 3.415554595411963e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.8393272707859675,
+      "num_tokens": 1299263272.0,
+      "step": 7332
+    },
+    {
+      "entropy": 0.6806386026243368,
+      "epoch": 2.2666872443861408,
+      "grad_norm": 0.17268003523349762,
+      "learning_rate": 3.400347373387099e-05,
+      "loss": 0.6591,
+      "mean_token_accuracy": 0.839379072189331,
+      "num_tokens": 1301389980.0,
+      "step": 7344
+    },
+    {
+      "entropy": 0.6829134225845337,
+      "epoch": 2.270391233891504,
+      "grad_norm": 0.17173625528812408,
+      "learning_rate": 3.385156618260988e-05,
+      "loss": 0.6615,
+      "mean_token_accuracy": 0.8388151476780573,
+      "num_tokens": 1303525932.0,
+      "step": 7356
+    },
+    {
+      "entropy": 0.6792828490336736,
+      "epoch": 2.274095223396867,
+      "grad_norm": 0.1721930056810379,
+      "learning_rate": 3.369982486407968e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.8391289735833803,
+      "num_tokens": 1305664388.0,
+      "step": 7368
+    },
+    {
+      "entropy": 0.6927899805208048,
+      "epoch": 2.27779921290223,
+      "grad_norm": 0.17168404161930084,
+      "learning_rate": 3.354825134031261e-05,
+      "loss": 0.6682,
+      "mean_token_accuracy": 0.8373626172542572,
+      "num_tokens": 1307780556.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.6774941322704157,
+      "epoch": 2.281503202407593,
+      "grad_norm": 0.1669553965330124,
+      "learning_rate": 3.339684717161358e-05,
+      "loss": 0.6543,
+      "mean_token_accuracy": 0.8401380938788255,
+      "num_tokens": 1309926320.0,
+      "step": 7392
+    },
+    {
+      "entropy": 0.6856278690199057,
+      "epoch": 2.2852071919129564,
+      "grad_norm": 0.18108035624027252,
+      "learning_rate": 3.3245613916544186e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8376220539212227,
+      "num_tokens": 1312077047.0,
+      "step": 7404
+    },
+    {
+      "entropy": 0.6788114123046398,
+      "epoch": 2.288911181418319,
+      "grad_norm": 0.1684533655643463,
+      "learning_rate": 3.30945531319066e-05,
+      "loss": 0.6553,
+      "mean_token_accuracy": 0.8395294894774755,
+      "num_tokens": 1314210244.0,
+      "step": 7416
+    },
+    {
+      "entropy": 0.6847534601887068,
+      "epoch": 2.2926151709236824,
+      "grad_norm": 0.17420773208141327,
+      "learning_rate": 3.294366637272758e-05,
+      "loss": 0.664,
+      "mean_token_accuracy": 0.8375212388734022,
+      "num_tokens": 1316332813.0,
+      "step": 7428
+    },
+    {
+      "entropy": 0.67829688017567,
+      "epoch": 2.2963191604290456,
+      "grad_norm": 0.17259854078292847,
+      "learning_rate": 3.279295519224248e-05,
+      "loss": 0.6552,
+      "mean_token_accuracy": 0.840360164642334,
+      "num_tokens": 1318446966.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.682950338969628,
+      "epoch": 2.3000231499344084,
+      "grad_norm": 0.17080390453338623,
+      "learning_rate": 3.26424211418792e-05,
+      "loss": 0.6622,
+      "mean_token_accuracy": 0.838694820801417,
+      "num_tokens": 1320578326.0,
+      "step": 7452
+    },
+    {
+      "entropy": 0.6896775799492995,
+      "epoch": 2.3037271394397716,
+      "grad_norm": 0.17524172365665436,
+      "learning_rate": 3.2492065771242294e-05,
+      "loss": 0.6665,
+      "mean_token_accuracy": 0.8372247604032358,
+      "num_tokens": 1322730718.0,
+      "step": 7464
+    },
+    {
+      "entropy": 0.6799197805424532,
+      "epoch": 2.307431128945135,
+      "grad_norm": 0.17451880872249603,
+      "learning_rate": 3.234189062809695e-05,
+      "loss": 0.6573,
+      "mean_token_accuracy": 0.8392599709331989,
+      "num_tokens": 1324847293.0,
+      "step": 7476
+    },
+    {
+      "entropy": 0.667460153500239,
+      "epoch": 2.3111351184504976,
+      "grad_norm": 0.18068212270736694,
+      "learning_rate": 3.219189725835311e-05,
+      "loss": 0.6471,
+      "mean_token_accuracy": 0.841596515228351,
+      "num_tokens": 1326958062.0,
+      "step": 7488
+    },
+    {
+      "entropy": 0.6783041432499886,
+      "epoch": 2.314839107955861,
+      "grad_norm": 0.18286266922950745,
+      "learning_rate": 3.20420872060495e-05,
+      "loss": 0.6564,
+      "mean_token_accuracy": 0.8399129547178745,
+      "num_tokens": 1329086141.0,
+      "step": 7500
+    },
+    {
+      "entropy": 0.6802773574988047,
+      "epoch": 2.318543097461224,
+      "grad_norm": 0.17960713803768158,
+      "learning_rate": 3.189246201333778e-05,
+      "loss": 0.6568,
+      "mean_token_accuracy": 0.8396291807293892,
+      "num_tokens": 1331208721.0,
+      "step": 7512
+    },
+    {
+      "entropy": 0.6929339257379373,
+      "epoch": 2.3222470869665868,
+      "grad_norm": 0.1749773770570755,
+      "learning_rate": 3.174302322046669e-05,
+      "loss": 0.6688,
+      "mean_token_accuracy": 0.8368431739509106,
+      "num_tokens": 1333333100.0,
+      "step": 7524
+    },
+    {
+      "entropy": 0.6909980537990729,
+      "epoch": 2.32595107647195,
+      "grad_norm": 0.16345363855361938,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 0.668,
+      "mean_token_accuracy": 0.8363487422466278,
+      "num_tokens": 1335449123.0,
+      "step": 7536
+    },
+    {
+      "entropy": 0.6702785243590673,
+      "epoch": 2.329655065977313,
+      "grad_norm": 0.17578427493572235,
+      "learning_rate": 3.144471098563131e-05,
+      "loss": 0.6476,
+      "mean_token_accuracy": 0.8414431102573872,
+      "num_tokens": 1337603503.0,
+      "step": 7548
+    },
+    {
+      "entropy": 0.6757986781497797,
+      "epoch": 2.333359055482676,
+      "grad_norm": 0.17358356714248657,
+      "learning_rate": 3.129584061450709e-05,
+      "loss": 0.6545,
+      "mean_token_accuracy": 0.8400241335233053,
+      "num_tokens": 1339719225.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.681422288219134,
+      "epoch": 2.337063044988039,
+      "grad_norm": 0.18540477752685547,
+      "learning_rate": 3.1147162784871986e-05,
+      "loss": 0.6578,
+      "mean_token_accuracy": 0.8393968902528286,
+      "num_tokens": 1341867232.0,
+      "step": 7572
+    },
+    {
+      "entropy": 0.6797411205867926,
+      "epoch": 2.3407670344934024,
+      "grad_norm": 0.17587818205356598,
+      "learning_rate": 3.0998679027222535e-05,
+      "loss": 0.6608,
+      "mean_token_accuracy": 0.8387786311407884,
+      "num_tokens": 1343972966.0,
+      "step": 7584
+    },
+    {
+      "entropy": 0.676243698845307,
+      "epoch": 2.344471023998765,
+      "grad_norm": 0.1766098141670227,
+      "learning_rate": 3.0850390870057435e-05,
+      "loss": 0.6539,
+      "mean_token_accuracy": 0.8406179261704286,
+      "num_tokens": 1346124725.0,
+      "step": 7596
+    },
+    {
+      "entropy": 0.683370524396499,
+      "epoch": 2.3481750135041284,
+      "grad_norm": 0.18542231619358063,
+      "learning_rate": 3.0702299839861915e-05,
+      "loss": 0.6598,
+      "mean_token_accuracy": 0.8393504520257314,
+      "num_tokens": 1348237486.0,
+      "step": 7608
+    },
+    {
+      "entropy": 0.6867342926561832,
+      "epoch": 2.3518790030094916,
+      "grad_norm": 0.17579790949821472,
+      "learning_rate": 3.0554407461091925e-05,
+      "loss": 0.6611,
+      "mean_token_accuracy": 0.8385164514183998,
+      "num_tokens": 1350337121.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.682259933402141,
+      "epoch": 2.3555829925148544,
+      "grad_norm": 0.1693212389945984,
+      "learning_rate": 3.0406715256158524e-05,
+      "loss": 0.66,
+      "mean_token_accuracy": 0.8392416251202425,
+      "num_tokens": 1352466879.0,
+      "step": 7632
+    },
+    {
+      "entropy": 0.6840335975090662,
+      "epoch": 2.3592869820202176,
+      "grad_norm": 0.16809053719043732,
+      "learning_rate": 3.0259224745412157e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.8384669708708922,
+      "num_tokens": 1354609035.0,
+      "step": 7644
+    },
+    {
+      "entropy": 0.6911391988396645,
+      "epoch": 2.362990971525581,
+      "grad_norm": 0.16995146870613098,
+      "learning_rate": 3.011193744712701e-05,
+      "loss": 0.6697,
+      "mean_token_accuracy": 0.8367390433947245,
+      "num_tokens": 1356726626.0,
+      "step": 7656
+    },
+    {
+      "entropy": 0.6708775386214256,
+      "epoch": 2.3666949610309436,
+      "grad_norm": 0.18156154453754425,
+      "learning_rate": 2.996485487748542e-05,
+      "loss": 0.6465,
+      "mean_token_accuracy": 0.841506606588761,
+      "num_tokens": 1358868598.0,
+      "step": 7668
+    },
+    {
+      "entropy": 0.6810227843622366,
+      "epoch": 2.370398950536307,
+      "grad_norm": 0.18769192695617676,
+      "learning_rate": 2.9817978550562208e-05,
+      "loss": 0.6577,
+      "mean_token_accuracy": 0.8396956846117973,
+      "num_tokens": 1360988797.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.6622078480819861,
+      "epoch": 2.37410294004167,
+      "grad_norm": 0.18112178146839142,
+      "learning_rate": 2.9671309978309132e-05,
+      "loss": 0.6407,
+      "mean_token_accuracy": 0.8430601954460144,
+      "num_tokens": 1363110692.0,
+      "step": 7692
+    },
+    {
+      "entropy": 0.6774277985095978,
+      "epoch": 2.3778069295470328,
+      "grad_norm": 0.17796562612056732,
+      "learning_rate": 2.9524850670539327e-05,
+      "loss": 0.6551,
+      "mean_token_accuracy": 0.8401034958660603,
+      "num_tokens": 1365200120.0,
+      "step": 7704
+    },
+    {
+      "entropy": 0.6842349829773108,
+      "epoch": 2.381510919052396,
+      "grad_norm": 0.1720336526632309,
+      "learning_rate": 2.937860213491173e-05,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.8388624712824821,
+      "num_tokens": 1367347094.0,
+      "step": 7716
+    },
+    {
+      "entropy": 0.6832546703517437,
+      "epoch": 2.385214908557759,
+      "grad_norm": 0.1650596261024475,
+      "learning_rate": 2.9232565876915585e-05,
+      "loss": 0.6605,
+      "mean_token_accuracy": 0.8394694936772188,
+      "num_tokens": 1369448080.0,
+      "step": 7728
+    },
+    {
+      "entropy": 0.6817479953169823,
+      "epoch": 2.388918898063122,
+      "grad_norm": 0.17462390661239624,
+      "learning_rate": 2.9086743399854972e-05,
+      "loss": 0.6581,
+      "mean_token_accuracy": 0.839547095199426,
+      "num_tokens": 1371582172.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.6731888850529989,
+      "epoch": 2.392622887568485,
+      "grad_norm": 0.18051554262638092,
+      "learning_rate": 2.894113620483323e-05,
+      "loss": 0.6517,
+      "mean_token_accuracy": 0.8407465231915315,
+      "num_tokens": 1373684475.0,
+      "step": 7752
+    },
+    {
+      "entropy": 0.6890167209009329,
+      "epoch": 2.3963268770738484,
+      "grad_norm": 0.1616596132516861,
+      "learning_rate": 2.8795745790737612e-05,
+      "loss": 0.6663,
+      "mean_token_accuracy": 0.8371898656090101,
+      "num_tokens": 1375839368.0,
+      "step": 7764
+    },
+    {
+      "entropy": 0.6708348778386911,
+      "epoch": 2.400030866579211,
+      "grad_norm": 0.1778324991464615,
+      "learning_rate": 2.865057365422386e-05,
+      "loss": 0.6503,
+      "mean_token_accuracy": 0.8408366913596789,
+      "num_tokens": 1377951641.0,
+      "step": 7776
+    },
+    {
+      "entropy": 0.6847917512059212,
+      "epoch": 2.4037348560845744,
+      "grad_norm": 0.17857876420021057,
+      "learning_rate": 2.8505621289700705e-05,
+      "loss": 0.6618,
+      "mean_token_accuracy": 0.8389340154826641,
+      "num_tokens": 1380072849.0,
+      "step": 7788
+    },
+    {
+      "entropy": 0.6858414982755979,
+      "epoch": 2.4074388455899376,
+      "grad_norm": 0.17925503849983215,
+      "learning_rate": 2.8360890189314548e-05,
+      "loss": 0.6617,
+      "mean_token_accuracy": 0.8384112877150377,
+      "num_tokens": 1382150591.0,
+      "step": 7800
+    },
+    {
+      "entropy": 0.6786315503219763,
+      "epoch": 2.4111428350953004,
+      "grad_norm": 0.16545531153678894,
+      "learning_rate": 2.821638184293408e-05,
+      "loss": 0.6566,
+      "mean_token_accuracy": 0.8399387697378794,
+      "num_tokens": 1384273874.0,
+      "step": 7812
+    },
+    {
+      "entropy": 0.6717840371032556,
+      "epoch": 2.4148468246006636,
+      "grad_norm": 0.18790796399116516,
+      "learning_rate": 2.8072097738134946e-05,
+      "loss": 0.6487,
+      "mean_token_accuracy": 0.8416854639848074,
+      "num_tokens": 1386421067.0,
+      "step": 7824
+    },
+    {
+      "entropy": 0.6912858026723067,
+      "epoch": 2.418550814106027,
+      "grad_norm": 0.19037222862243652,
+      "learning_rate": 2.792803936018449e-05,
+      "loss": 0.6669,
+      "mean_token_accuracy": 0.8376653293768564,
+      "num_tokens": 1388555640.0,
+      "step": 7836
+    },
+    {
+      "entropy": 0.6990839106341203,
+      "epoch": 2.4222548036113896,
+      "grad_norm": 0.177109956741333,
+      "learning_rate": 2.778420819202634e-05,
+      "loss": 0.6791,
+      "mean_token_accuracy": 0.8342719053228697,
+      "num_tokens": 1390701540.0,
+      "step": 7848
+    },
+    {
+      "entropy": 0.6801580414175987,
+      "epoch": 2.4259587931167528,
+      "grad_norm": 0.18449810147285461,
+      "learning_rate": 2.764060571426527e-05,
+      "loss": 0.6581,
+      "mean_token_accuracy": 0.839683381219705,
+      "num_tokens": 1392847168.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.677167497575283,
+      "epoch": 2.429662782622116,
+      "grad_norm": 0.17673510313034058,
+      "learning_rate": 2.7497233405151873e-05,
+      "loss": 0.6546,
+      "mean_token_accuracy": 0.8402298440535864,
+      "num_tokens": 1394982282.0,
+      "step": 7872
+    },
+    {
+      "entropy": 0.6832474445303282,
+      "epoch": 2.4333667721274788,
+      "grad_norm": 0.1732633411884308,
+      "learning_rate": 2.7354092740567395e-05,
+      "loss": 0.6585,
+      "mean_token_accuracy": 0.8391941326359907,
+      "num_tokens": 1397146217.0,
+      "step": 7884
+    },
+    {
+      "entropy": 0.6827088569601377,
+      "epoch": 2.437070761632842,
+      "grad_norm": 0.1737031638622284,
+      "learning_rate": 2.7211185194008492e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8383902385830879,
+      "num_tokens": 1399291383.0,
+      "step": 7896
+    },
+    {
+      "entropy": 0.68471717958649,
+      "epoch": 2.440774751138205,
+      "grad_norm": 0.2073935717344284,
+      "learning_rate": 2.706851223657214e-05,
+      "loss": 0.6628,
+      "mean_token_accuracy": 0.8384527899324894,
+      "num_tokens": 1401422348.0,
+      "step": 7908
+    },
+    {
+      "entropy": 0.6754190859695276,
+      "epoch": 2.444478740643568,
+      "grad_norm": 0.18220025300979614,
+      "learning_rate": 2.6926075336940404e-05,
+      "loss": 0.651,
+      "mean_token_accuracy": 0.8403765348096689,
+      "num_tokens": 1403549979.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.6639910079538822,
+      "epoch": 2.448182730148931,
+      "grad_norm": 0.1840001493692398,
+      "learning_rate": 2.6783875961365378e-05,
+      "loss": 0.6427,
+      "mean_token_accuracy": 0.842850154886643,
+      "num_tokens": 1405631673.0,
+      "step": 7932
+    },
+    {
+      "entropy": 0.6794233421484629,
+      "epoch": 2.4518867196542944,
+      "grad_norm": 0.1842009425163269,
+      "learning_rate": 2.664191557365404e-05,
+      "loss": 0.6593,
+      "mean_token_accuracy": 0.8393058615426222,
+      "num_tokens": 1407746726.0,
+      "step": 7944
+    },
+    {
+      "entropy": 0.696602446337541,
+      "epoch": 2.455590709159657,
+      "grad_norm": 0.20143558084964752,
+      "learning_rate": 2.6500195635153287e-05,
+      "loss": 0.6741,
+      "mean_token_accuracy": 0.835796115299066,
+      "num_tokens": 1409892372.0,
+      "step": 7956
+    },
+    {
+      "entropy": 0.6946029762427012,
+      "epoch": 2.4592946986650204,
+      "grad_norm": 0.17518438398838043,
+      "learning_rate": 2.635871760473475e-05,
+      "loss": 0.6691,
+      "mean_token_accuracy": 0.8366143380602201,
+      "num_tokens": 1412008656.0,
+      "step": 7968
+    },
+    {
+      "entropy": 0.6815679098169009,
+      "epoch": 2.4629986881703836,
+      "grad_norm": 0.18056343495845795,
+      "learning_rate": 2.6217482938779882e-05,
+      "loss": 0.6573,
+      "mean_token_accuracy": 0.8400058262050152,
+      "num_tokens": 1414116887.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.6889658160507679,
+      "epoch": 2.4667026776757464,
+      "grad_norm": 0.184426948428154,
+      "learning_rate": 2.607649309116492e-05,
+      "loss": 0.6678,
+      "mean_token_accuracy": 0.8373049336175123,
+      "num_tokens": 1416227606.0,
+      "step": 7992
+    },
+    {
+      "entropy": 0.6798962478836378,
+      "epoch": 2.4704066671811096,
+      "grad_norm": 0.1737222820520401,
+      "learning_rate": 2.5935749513245927e-05,
+      "loss": 0.655,
+      "mean_token_accuracy": 0.8400146377583345,
+      "num_tokens": 1418343060.0,
+      "step": 8004
+    },
+    {
+      "entropy": 0.6849433220922947,
+      "epoch": 2.474110656686473,
+      "grad_norm": 0.17856653034687042,
+      "learning_rate": 2.579525365384386e-05,
+      "loss": 0.6641,
+      "mean_token_accuracy": 0.8383715438346068,
+      "num_tokens": 1420448289.0,
+      "step": 8016
+    },
+    {
+      "entropy": 0.6673534649113814,
+      "epoch": 2.4778146461918356,
+      "grad_norm": 0.18288525938987732,
+      "learning_rate": 2.5655006959229655e-05,
+      "loss": 0.6439,
+      "mean_token_accuracy": 0.8429509215056896,
+      "num_tokens": 1422567600.0,
+      "step": 8028
+    },
+    {
+      "entropy": 0.6816601393123468,
+      "epoch": 2.4815186356971988,
+      "grad_norm": 0.17290998995304108,
+      "learning_rate": 2.5515010873109313e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.8389557711780071,
+      "num_tokens": 1424653554.0,
+      "step": 8040
+    },
+    {
+      "entropy": 0.6713767026861509,
+      "epoch": 2.485222625202562,
+      "grad_norm": 0.16905297338962555,
+      "learning_rate": 2.537526683660909e-05,
+      "loss": 0.6482,
+      "mean_token_accuracy": 0.841906680415074,
+      "num_tokens": 1426786823.0,
+      "step": 8052
+    },
+    {
+      "entropy": 0.6828158932427565,
+      "epoch": 2.4889266147079248,
+      "grad_norm": 0.17191940546035767,
+      "learning_rate": 2.5235776288260593e-05,
+      "loss": 0.6614,
+      "mean_token_accuracy": 0.8386979786058267,
+      "num_tokens": 1428924792.0,
+      "step": 8064
+    },
+    {
+      "entropy": 0.6837918112675349,
+      "epoch": 2.492630604213288,
+      "grad_norm": 0.16740551590919495,
+      "learning_rate": 2.5096540663986067e-05,
+      "loss": 0.6631,
+      "mean_token_accuracy": 0.8389793510238329,
+      "num_tokens": 1431082569.0,
+      "step": 8076
+    },
+    {
+      "entropy": 0.6886960230767727,
+      "epoch": 2.496334593718651,
+      "grad_norm": 0.17835856974124908,
+      "learning_rate": 2.4957561397083506e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8381286015113195,
+      "num_tokens": 1433227615.0,
+      "step": 8088
+    },
+    {
+      "entropy": 0.6757631773749987,
+      "epoch": 2.500038583224014,
+      "grad_norm": 0.1775360405445099,
+      "learning_rate": 2.4818839918211962e-05,
+      "loss": 0.6531,
+      "mean_token_accuracy": 0.8405404339234034,
+      "num_tokens": 1435347410.0,
+      "step": 8100
+    },
+    {
+      "entropy": 0.6918977871537209,
+      "epoch": 2.503742572729377,
+      "grad_norm": 0.17831064760684967,
+      "learning_rate": 2.4680377655376814e-05,
+      "loss": 0.6699,
+      "mean_token_accuracy": 0.8364169212679068,
+      "num_tokens": 1437513213.0,
+      "step": 8112
+    },
+    {
+      "entropy": 0.6668169523278872,
+      "epoch": 2.5074465622347404,
+      "grad_norm": 0.1669531911611557,
+      "learning_rate": 2.4542176033915048e-05,
+      "loss": 0.6448,
+      "mean_token_accuracy": 0.8414871692657471,
+      "num_tokens": 1439633767.0,
+      "step": 8124
+    },
+    {
+      "entropy": 0.6661192377408346,
+      "epoch": 2.511150551740103,
+      "grad_norm": 0.18062232434749603,
+      "learning_rate": 2.4404236476480608e-05,
+      "loss": 0.6454,
+      "mean_token_accuracy": 0.8420434022943178,
+      "num_tokens": 1441737922.0,
+      "step": 8136
+    },
+    {
+      "entropy": 0.678036684791247,
+      "epoch": 2.5148545412454664,
+      "grad_norm": 0.2435832917690277,
+      "learning_rate": 2.426656040302973e-05,
+      "loss": 0.654,
+      "mean_token_accuracy": 0.8405960587163767,
+      "num_tokens": 1443852135.0,
+      "step": 8148
+    },
+    {
+      "entropy": 0.6822113630672296,
+      "epoch": 2.5185585307508296,
+      "grad_norm": 0.16967307031154633,
+      "learning_rate": 2.4129149230806337e-05,
+      "loss": 0.6602,
+      "mean_token_accuracy": 0.8390215809146563,
+      "num_tokens": 1445946633.0,
+      "step": 8160
+    },
+    {
+      "entropy": 0.6625789310783148,
+      "epoch": 2.5222625202561924,
+      "grad_norm": 0.16657769680023193,
+      "learning_rate": 2.399200437432744e-05,
+      "loss": 0.642,
+      "mean_token_accuracy": 0.8431124811371168,
+      "num_tokens": 1448087716.0,
+      "step": 8172
+    },
+    {
+      "entropy": 0.6771047189831734,
+      "epoch": 2.5259665097615556,
+      "grad_norm": 0.1971009075641632,
+      "learning_rate": 2.385512724536857e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.8407648056745529,
+      "num_tokens": 1450243227.0,
+      "step": 8184
+    },
+    {
+      "entropy": 0.6659577091534933,
+      "epoch": 2.529670499266919,
+      "grad_norm": 0.17768594622612,
+      "learning_rate": 2.3718519252949316e-05,
+      "loss": 0.6443,
+      "mean_token_accuracy": 0.8426598384976387,
+      "num_tokens": 1452367356.0,
+      "step": 8196
+    },
+    {
+      "entropy": 0.6778051033616066,
+      "epoch": 2.5333744887722816,
+      "grad_norm": 0.1780218631029129,
+      "learning_rate": 2.358218180331871e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.840303952495257,
+      "num_tokens": 1454523799.0,
+      "step": 8208
+    },
+    {
+      "entropy": 0.6757925574978193,
+      "epoch": 2.5370784782776448,
+      "grad_norm": 0.175310879945755,
+      "learning_rate": 2.3446116299940814e-05,
+      "loss": 0.6527,
+      "mean_token_accuracy": 0.840245266755422,
+      "num_tokens": 1456682468.0,
+      "step": 8220
+    },
+    {
+      "entropy": 0.6883031961818536,
+      "epoch": 2.540782467783008,
+      "grad_norm": 0.17527994513511658,
+      "learning_rate": 2.331032414348026e-05,
+      "loss": 0.668,
+      "mean_token_accuracy": 0.837483424693346,
+      "num_tokens": 1458745387.0,
+      "step": 8232
+    },
+    {
+      "entropy": 0.6782544503609339,
+      "epoch": 2.5444864572883708,
+      "grad_norm": 0.18310600519180298,
+      "learning_rate": 2.3174806731787833e-05,
+      "loss": 0.6579,
+      "mean_token_accuracy": 0.8392157728473345,
+      "num_tokens": 1460852728.0,
+      "step": 8244
+    },
+    {
+      "entropy": 0.678227295478185,
+      "epoch": 2.548190446793734,
+      "grad_norm": 0.16748745739459991,
+      "learning_rate": 2.3039565459886144e-05,
+      "loss": 0.6544,
+      "mean_token_accuracy": 0.8402854837477207,
+      "num_tokens": 1462984834.0,
+      "step": 8256
+    },
+    {
+      "entropy": 0.6784488918880621,
+      "epoch": 2.551894436299097,
+      "grad_norm": 0.17468595504760742,
+      "learning_rate": 2.290460171995508e-05,
+      "loss": 0.6579,
+      "mean_token_accuracy": 0.8394933181504408,
+      "num_tokens": 1465111628.0,
+      "step": 8268
+    },
+    {
+      "entropy": 0.68077094728748,
+      "epoch": 2.55559842580446,
+      "grad_norm": 0.17412585020065308,
+      "learning_rate": 2.2769916901317717e-05,
+      "loss": 0.6588,
+      "mean_token_accuracy": 0.8396103344857693,
+      "num_tokens": 1467237410.0,
+      "step": 8280
+    },
+    {
+      "entropy": 0.6686192614336809,
+      "epoch": 2.559302415309823,
+      "grad_norm": 0.18562890589237213,
+      "learning_rate": 2.2635512390425834e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.8421800620853901,
+      "num_tokens": 1469391192.0,
+      "step": 8292
+    },
+    {
+      "entropy": 0.6803597112496694,
+      "epoch": 2.5630064048151864,
+      "grad_norm": 0.17133885622024536,
+      "learning_rate": 2.2501389570845767e-05,
+      "loss": 0.6557,
+      "mean_token_accuracy": 0.8401055956880251,
+      "num_tokens": 1471504796.0,
+      "step": 8304
+    },
+    {
+      "entropy": 0.67287752404809,
+      "epoch": 2.566710394320549,
+      "grad_norm": 0.17606225609779358,
+      "learning_rate": 2.236754982324406e-05,
+      "loss": 0.6517,
+      "mean_token_accuracy": 0.8406522087752819,
+      "num_tokens": 1473637509.0,
+      "step": 8316
+    },
+    {
+      "entropy": 0.6758027486503124,
+      "epoch": 2.5704143838259124,
+      "grad_norm": 0.16274742782115936,
+      "learning_rate": 2.223399452537332e-05,
+      "loss": 0.6514,
+      "mean_token_accuracy": 0.8412458946307501,
+      "num_tokens": 1475756831.0,
+      "step": 8328
+    },
+    {
+      "entropy": 0.6769342236220837,
+      "epoch": 2.5741183733312756,
+      "grad_norm": 0.173589825630188,
+      "learning_rate": 2.2100725052058006e-05,
+      "loss": 0.6535,
+      "mean_token_accuracy": 0.8408135734498501,
+      "num_tokens": 1477899347.0,
+      "step": 8340
+    },
+    {
+      "entropy": 0.6732165577510992,
+      "epoch": 2.577822362836639,
+      "grad_norm": 0.18250826001167297,
+      "learning_rate": 2.1967742775180306e-05,
+      "loss": 0.6503,
+      "mean_token_accuracy": 0.8415469999114672,
+      "num_tokens": 1480035263.0,
+      "step": 8352
+    },
+    {
+      "entropy": 0.6719749377419552,
+      "epoch": 2.5815263523420016,
+      "grad_norm": 0.174708753824234,
+      "learning_rate": 2.183504906366595e-05,
+      "loss": 0.6495,
+      "mean_token_accuracy": 0.8410700398186842,
+      "num_tokens": 1482156466.0,
+      "step": 8364
+    },
+    {
+      "entropy": 0.6818638009329637,
+      "epoch": 2.5852303418473648,
+      "grad_norm": 0.18486975133419037,
+      "learning_rate": 2.1702645283470236e-05,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.8392534951368967,
+      "num_tokens": 1484290285.0,
+      "step": 8376
+    },
+    {
+      "entropy": 0.683278722067674,
+      "epoch": 2.588934331352728,
+      "grad_norm": 0.17580677568912506,
+      "learning_rate": 2.1570532797563852e-05,
+      "loss": 0.6613,
+      "mean_token_accuracy": 0.8384759289522966,
+      "num_tokens": 1486400150.0,
+      "step": 8388
+    },
+    {
+      "entropy": 0.6725541402896246,
+      "epoch": 2.5926383208580908,
+      "grad_norm": 0.19462420046329498,
+      "learning_rate": 2.143871296591885e-05,
+      "loss": 0.6482,
+      "mean_token_accuracy": 0.8413905973235766,
+      "num_tokens": 1488514217.0,
+      "step": 8400
+    },
+    {
+      "entropy": 0.6712416261434555,
+      "epoch": 2.596342310363454,
+      "grad_norm": 0.17859038710594177,
+      "learning_rate": 2.1307187145494727e-05,
+      "loss": 0.6476,
+      "mean_token_accuracy": 0.8417484226326147,
+      "num_tokens": 1490640601.0,
+      "step": 8412
+    },
+    {
+      "entropy": 0.6808934770524502,
+      "epoch": 2.600046299868817,
+      "grad_norm": 0.16942450404167175,
+      "learning_rate": 2.1175956690224453e-05,
+      "loss": 0.6565,
+      "mean_token_accuracy": 0.8403363898396492,
+      "num_tokens": 1492760780.0,
+      "step": 8424
+    },
+    {
+      "entropy": 0.688660574456056,
+      "epoch": 2.60375028937418,
+      "grad_norm": 0.17530061304569244,
+      "learning_rate": 2.1045022951000437e-05,
+      "loss": 0.6658,
+      "mean_token_accuracy": 0.8372594018777212,
+      "num_tokens": 1494877831.0,
+      "step": 8436
+    },
+    {
+      "entropy": 0.6738361803193887,
+      "epoch": 2.607454278879543,
+      "grad_norm": 0.169802725315094,
+      "learning_rate": 2.091438727566069e-05,
+      "loss": 0.6533,
+      "mean_token_accuracy": 0.8404905224839846,
+      "num_tokens": 1497011614.0,
+      "step": 8448
+    },
+    {
+      "entropy": 0.6933514513075352,
+      "epoch": 2.6111582683849064,
+      "grad_norm": 0.16949047148227692,
+      "learning_rate": 2.0784051008974957e-05,
+      "loss": 0.6672,
+      "mean_token_accuracy": 0.8373734119037787,
+      "num_tokens": 1499130158.0,
+      "step": 8460
+    },
+    {
+      "entropy": 0.676812877257665,
+      "epoch": 2.614862257890269,
+      "grad_norm": 0.16732563078403473,
+      "learning_rate": 2.0654015492630842e-05,
+      "loss": 0.6561,
+      "mean_token_accuracy": 0.8405400899549326,
+      "num_tokens": 1501274725.0,
+      "step": 8472
+    },
+    {
+      "entropy": 0.6772327373425165,
+      "epoch": 2.6185662473956324,
+      "grad_norm": 0.1803075224161148,
+      "learning_rate": 2.0524282065219995e-05,
+      "loss": 0.6551,
+      "mean_token_accuracy": 0.839802881081899,
+      "num_tokens": 1503401542.0,
+      "step": 8484
+    },
+    {
+      "entropy": 0.6690971752007803,
+      "epoch": 2.6222702369009956,
+      "grad_norm": 0.182891383767128,
+      "learning_rate": 2.0394852062224408e-05,
+      "loss": 0.6458,
+      "mean_token_accuracy": 0.84250757843256,
+      "num_tokens": 1505486844.0,
+      "step": 8496
+    },
+    {
+      "entropy": 0.6797614445288976,
+      "epoch": 2.6259742264063584,
+      "grad_norm": 0.1719876229763031,
+      "learning_rate": 2.026572681600257e-05,
+      "loss": 0.6556,
+      "mean_token_accuracy": 0.8399543985724449,
+      "num_tokens": 1507602524.0,
+      "step": 8508
+    },
+    {
+      "entropy": 0.6750110884507498,
+      "epoch": 2.6296782159117216,
+      "grad_norm": 0.16174942255020142,
+      "learning_rate": 2.0136907655775743e-05,
+      "loss": 0.6544,
+      "mean_token_accuracy": 0.8410836247106394,
+      "num_tokens": 1509738493.0,
+      "step": 8520
+    },
+    {
+      "entropy": 0.6792822231849035,
+      "epoch": 2.633382205417085,
+      "grad_norm": 0.18377965688705444,
+      "learning_rate": 2.000839590761439e-05,
+      "loss": 0.6583,
+      "mean_token_accuracy": 0.8397029526531696,
+      "num_tokens": 1511827824.0,
+      "step": 8532
+    },
+    {
+      "entropy": 0.6866883126397928,
+      "epoch": 2.6370861949224476,
+      "grad_norm": 0.17742934823036194,
+      "learning_rate": 1.988019289442446e-05,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.8386820207039515,
+      "num_tokens": 1513923734.0,
+      "step": 8544
+    },
+    {
+      "entropy": 0.6748844794929028,
+      "epoch": 2.6407901844278108,
+      "grad_norm": 0.1755974441766739,
+      "learning_rate": 1.9752299935933732e-05,
+      "loss": 0.6528,
+      "mean_token_accuracy": 0.8406555938223997,
+      "num_tokens": 1516031350.0,
+      "step": 8556
+    },
+    {
+      "entropy": 0.6851431752244631,
+      "epoch": 2.644494173933174,
+      "grad_norm": 0.17667120695114136,
+      "learning_rate": 1.9624718348678302e-05,
+      "loss": 0.659,
+      "mean_token_accuracy": 0.8388123251497746,
+      "num_tokens": 1518147388.0,
+      "step": 8568
+    },
+    {
+      "entropy": 0.6720710135996342,
+      "epoch": 2.648198163438537,
+      "grad_norm": 0.1714988797903061,
+      "learning_rate": 1.9497449445988964e-05,
+      "loss": 0.6477,
+      "mean_token_accuracy": 0.8416066728532314,
+      "num_tokens": 1520240737.0,
+      "step": 8580
+    },
+    {
+      "entropy": 0.6758585870265961,
+      "epoch": 2.6519021529439,
+      "grad_norm": 0.2017345130443573,
+      "learning_rate": 1.9370494537977724e-05,
+      "loss": 0.6538,
+      "mean_token_accuracy": 0.8405953471859297,
+      "num_tokens": 1522358544.0,
+      "step": 8592
+    },
+    {
+      "entropy": 0.6837990420560042,
+      "epoch": 2.655606142449263,
+      "grad_norm": 0.18071149289608002,
+      "learning_rate": 1.9243854931524363e-05,
+      "loss": 0.6622,
+      "mean_token_accuracy": 0.8380974618097147,
+      "num_tokens": 1524491885.0,
+      "step": 8604
+    },
+    {
+      "entropy": 0.6755645809074243,
+      "epoch": 2.6593101319546264,
+      "grad_norm": 0.1797918975353241,
+      "learning_rate": 1.911753193026288e-05,
+      "loss": 0.6517,
+      "mean_token_accuracy": 0.8409853937725226,
+      "num_tokens": 1526599435.0,
+      "step": 8616
+    },
+    {
+      "entropy": 0.6866617997487386,
+      "epoch": 2.663014121459989,
+      "grad_norm": 0.1666131615638733,
+      "learning_rate": 1.8991526834568146e-05,
+      "loss": 0.6633,
+      "mean_token_accuracy": 0.8389130185047785,
+      "num_tokens": 1528740068.0,
+      "step": 8628
+    },
+    {
+      "entropy": 0.6782820535202821,
+      "epoch": 2.6667181109653524,
+      "grad_norm": 0.1680057942867279,
+      "learning_rate": 1.8865840941542506e-05,
+      "loss": 0.6564,
+      "mean_token_accuracy": 0.8396158503989378,
+      "num_tokens": 1530887262.0,
+      "step": 8640
+    },
+    {
+      "entropy": 0.6818838343024254,
+      "epoch": 2.6704221004707156,
+      "grad_norm": 0.18536606431007385,
+      "learning_rate": 1.8740475545002357e-05,
+      "loss": 0.6589,
+      "mean_token_accuracy": 0.8392697448531786,
+      "num_tokens": 1533019334.0,
+      "step": 8652
+    },
+    {
+      "entropy": 0.6834930268426737,
+      "epoch": 2.6741260899760784,
+      "grad_norm": 0.16942651569843292,
+      "learning_rate": 1.8615431935464982e-05,
+      "loss": 0.6633,
+      "mean_token_accuracy": 0.8381645083427429,
+      "num_tokens": 1535141514.0,
+      "step": 8664
+    },
+    {
+      "entropy": 0.6799823269248009,
+      "epoch": 2.6778300794814416,
+      "grad_norm": 0.18408891558647156,
+      "learning_rate": 1.8490711400135118e-05,
+      "loss": 0.6584,
+      "mean_token_accuracy": 0.8390690212448438,
+      "num_tokens": 1537222924.0,
+      "step": 8676
+    },
+    {
+      "entropy": 0.6704868152737617,
+      "epoch": 2.681534068986805,
+      "grad_norm": 0.17771634459495544,
+      "learning_rate": 1.8366315222891772e-05,
+      "loss": 0.649,
+      "mean_token_accuracy": 0.8417287928362688,
+      "num_tokens": 1539361332.0,
+      "step": 8688
+    },
+    {
+      "entropy": 0.6824560165405273,
+      "epoch": 2.6852380584921676,
+      "grad_norm": 0.1805276870727539,
+      "learning_rate": 1.8242244684274994e-05,
+      "loss": 0.662,
+      "mean_token_accuracy": 0.8386474835375944,
+      "num_tokens": 1541489251.0,
+      "step": 8700
+    },
+    {
+      "entropy": 0.6674566405514876,
+      "epoch": 2.688942047997531,
+      "grad_norm": 0.17611855268478394,
+      "learning_rate": 1.8118501061472686e-05,
+      "loss": 0.6428,
+      "mean_token_accuracy": 0.8432766410211722,
+      "num_tokens": 1543623822.0,
+      "step": 8712
+    },
+    {
+      "entropy": 0.6744134798645973,
+      "epoch": 2.692646037502894,
+      "grad_norm": 0.17920047044754028,
+      "learning_rate": 1.7995085628307494e-05,
+      "loss": 0.6498,
+      "mean_token_accuracy": 0.8411453490455946,
+      "num_tokens": 1545739068.0,
+      "step": 8724
+    },
+    {
+      "entropy": 0.6672421346108118,
+      "epoch": 2.6963500270082568,
+      "grad_norm": 0.17886677384376526,
+      "learning_rate": 1.7871999655223636e-05,
+      "loss": 0.6447,
+      "mean_token_accuracy": 0.8418544096251329,
+      "num_tokens": 1547875527.0,
+      "step": 8736
+    },
+    {
+      "entropy": 0.6673632053037485,
+      "epoch": 2.70005401651362,
+      "grad_norm": 0.18474158644676208,
+      "learning_rate": 1.774924440927386e-05,
+      "loss": 0.6458,
+      "mean_token_accuracy": 0.841834536443154,
+      "num_tokens": 1549985851.0,
+      "step": 8748
+    },
+    {
+      "entropy": 0.6826143587629,
+      "epoch": 2.703758006018983,
+      "grad_norm": 0.1684853732585907,
+      "learning_rate": 1.7626821154106382e-05,
+      "loss": 0.6594,
+      "mean_token_accuracy": 0.8393335652848085,
+      "num_tokens": 1552149497.0,
+      "step": 8760
+    },
+    {
+      "entropy": 0.6779943940540155,
+      "epoch": 2.707461995524346,
+      "grad_norm": 0.16713795065879822,
+      "learning_rate": 1.7504731149951913e-05,
+      "loss": 0.6564,
+      "mean_token_accuracy": 0.8403484423955282,
+      "num_tokens": 1554311306.0,
+      "step": 8772
+    },
+    {
+      "entropy": 0.6786763022343317,
+      "epoch": 2.711165985029709,
+      "grad_norm": 0.18719340860843658,
+      "learning_rate": 1.738297565361065e-05,
+      "loss": 0.6533,
+      "mean_token_accuracy": 0.8405405295391878,
+      "num_tokens": 1556410114.0,
+      "step": 8784
+    },
+    {
+      "entropy": 0.6622005340953668,
+      "epoch": 2.7148699745350724,
+      "grad_norm": 0.1840929239988327,
+      "learning_rate": 1.7261555918439347e-05,
+      "loss": 0.6393,
+      "mean_token_accuracy": 0.8439254661401113,
+      "num_tokens": 1558582974.0,
+      "step": 8796
+    },
+    {
+      "entropy": 0.6725914205114046,
+      "epoch": 2.718573964040435,
+      "grad_norm": 0.17707188427448273,
+      "learning_rate": 1.714047319433842e-05,
+      "loss": 0.6485,
+      "mean_token_accuracy": 0.8418245787421862,
+      "num_tokens": 1560714505.0,
+      "step": 8808
+    },
+    {
+      "entropy": 0.6861369709173838,
+      "epoch": 2.7222779535457984,
+      "grad_norm": 0.18081851303577423,
+      "learning_rate": 1.7019728727739082e-05,
+      "loss": 0.6616,
+      "mean_token_accuracy": 0.8398788695534071,
+      "num_tokens": 1562841818.0,
+      "step": 8820
+    },
+    {
+      "entropy": 0.676196767638127,
+      "epoch": 2.7259819430511616,
+      "grad_norm": 0.17770478129386902,
+      "learning_rate": 1.6899323761590486e-05,
+      "loss": 0.6529,
+      "mean_token_accuracy": 0.8412167988717556,
+      "num_tokens": 1564958639.0,
+      "step": 8832
+    },
+    {
+      "entropy": 0.6725533629457155,
+      "epoch": 2.7296859325565244,
+      "grad_norm": 0.18487904965877533,
+      "learning_rate": 1.6779259535346996e-05,
+      "loss": 0.6497,
+      "mean_token_accuracy": 0.8410086408257484,
+      "num_tokens": 1567108439.0,
+      "step": 8844
+    },
+    {
+      "entropy": 0.6586884421606859,
+      "epoch": 2.7333899220618876,
+      "grad_norm": 0.16857771575450897,
+      "learning_rate": 1.6659537284955358e-05,
+      "loss": 0.6388,
+      "mean_token_accuracy": 0.8442123333613077,
+      "num_tokens": 1569262392.0,
+      "step": 8856
+    },
+    {
+      "entropy": 0.6725297843416532,
+      "epoch": 2.737093911567251,
+      "grad_norm": 0.17279498279094696,
+      "learning_rate": 1.654015824284198e-05,
+      "loss": 0.6484,
+      "mean_token_accuracy": 0.8418673065801462,
+      "num_tokens": 1571400293.0,
+      "step": 8868
+    },
+    {
+      "entropy": 0.665387824177742,
+      "epoch": 2.7407979010726136,
+      "grad_norm": 0.19026906788349152,
+      "learning_rate": 1.642112363790031e-05,
+      "loss": 0.6424,
+      "mean_token_accuracy": 0.8429265518983206,
+      "num_tokens": 1573528403.0,
+      "step": 8880
+    },
+    {
+      "entropy": 0.6819708024462064,
+      "epoch": 2.7445018905779768,
+      "grad_norm": 0.17325517535209656,
+      "learning_rate": 1.6302434695478107e-05,
+      "loss": 0.6607,
+      "mean_token_accuracy": 0.8388937674462795,
+      "num_tokens": 1575627080.0,
+      "step": 8892
+    },
+    {
+      "entropy": 0.6820539807279905,
+      "epoch": 2.74820588008334,
+      "grad_norm": 0.1792973279953003,
+      "learning_rate": 1.618409263736489e-05,
+      "loss": 0.658,
+      "mean_token_accuracy": 0.8399195012946924,
+      "num_tokens": 1577728236.0,
+      "step": 8904
+    },
+    {
+      "entropy": 0.6730578330655893,
+      "epoch": 2.7519098695887028,
+      "grad_norm": 0.18479055166244507,
+      "learning_rate": 1.606609868177932e-05,
+      "loss": 0.6522,
+      "mean_token_accuracy": 0.8412285819649696,
+      "num_tokens": 1579821668.0,
+      "step": 8916
+    },
+    {
+      "entropy": 0.6686800830066204,
+      "epoch": 2.755613859094066,
+      "grad_norm": 0.17226146161556244,
+      "learning_rate": 1.594845404335668e-05,
+      "loss": 0.644,
+      "mean_token_accuracy": 0.842714703331391,
+      "num_tokens": 1581941943.0,
+      "step": 8928
+    },
+    {
+      "entropy": 0.6718707966307799,
+      "epoch": 2.759317848599429,
+      "grad_norm": 0.17666469514369965,
+      "learning_rate": 1.583115993313637e-05,
+      "loss": 0.6497,
+      "mean_token_accuracy": 0.8410539838174979,
+      "num_tokens": 1584072385.0,
+      "step": 8940
+    },
+    {
+      "entropy": 0.6763085102041563,
+      "epoch": 2.763021838104792,
+      "grad_norm": 0.17432096600532532,
+      "learning_rate": 1.571421755854941e-05,
+      "loss": 0.6526,
+      "mean_token_accuracy": 0.8408862004677454,
+      "num_tokens": 1586203502.0,
+      "step": 8952
+    },
+    {
+      "entropy": 0.6714817471802235,
+      "epoch": 2.766725827610155,
+      "grad_norm": 0.1722986400127411,
+      "learning_rate": 1.5597628123406095e-05,
+      "loss": 0.6489,
+      "mean_token_accuracy": 0.8418647820750872,
+      "num_tokens": 1588322897.0,
+      "step": 8964
+    },
+    {
+      "entropy": 0.6764586915572485,
+      "epoch": 2.7704298171155184,
+      "grad_norm": 0.16954948008060455,
+      "learning_rate": 1.548139282788349e-05,
+      "loss": 0.6549,
+      "mean_token_accuracy": 0.8404773225386938,
+      "num_tokens": 1590459142.0,
+      "step": 8976
+    },
+    {
+      "entropy": 0.6723885672787825,
+      "epoch": 2.774133806620881,
+      "grad_norm": 0.18865634500980377,
+      "learning_rate": 1.5365512868513144e-05,
+      "loss": 0.6509,
+      "mean_token_accuracy": 0.8412475138902664,
+      "num_tokens": 1592577106.0,
+      "step": 8988
+    },
+    {
+      "entropy": 0.669227180381616,
+      "epoch": 2.7778377961262444,
+      "grad_norm": 0.17359550297260284,
+      "learning_rate": 1.5249989438168771e-05,
+      "loss": 0.6457,
+      "mean_token_accuracy": 0.8424140512943268,
+      "num_tokens": 1594699366.0,
+      "step": 9000
+    }
+  ],
+  "logging_steps": 12,
+  "max_steps": 12000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0682543308592801e+20,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}