diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4274 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 424,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.7230634093284607,
+      "epoch": 0.009478672985781991,
+      "grad_norm": 1.5876940488815308,
+      "learning_rate": 0.0,
+      "loss": 1.6636556386947632,
+      "mean_token_accuracy": 0.6497756391763687,
+      "num_tokens": 16375.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.7448452711105347,
+      "epoch": 0.018957345971563982,
+      "grad_norm": 1.609075665473938,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 1.7037649154663086,
+      "mean_token_accuracy": 0.6527450829744339,
+      "num_tokens": 31738.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.710326075553894,
+      "epoch": 0.02843601895734597,
+      "grad_norm": 1.7256412506103516,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.663683533668518,
+      "mean_token_accuracy": 0.6594014763832092,
+      "num_tokens": 48323.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.8404302299022675,
+      "epoch": 0.037914691943127965,
+      "grad_norm": 1.7094064950942993,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.8555431365966797,
+      "mean_token_accuracy": 0.6288548111915588,
+      "num_tokens": 63474.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.61773881316185,
+      "epoch": 0.04739336492890995,
+      "grad_norm": 1.4093204736709595,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.586500883102417,
+      "mean_token_accuracy": 0.6745017617940903,
+      "num_tokens": 81340.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.713468849658966,
+      "epoch": 0.05687203791469194,
+      "grad_norm": 1.4611302614212036,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.693520188331604,
+      "mean_token_accuracy": 0.6446037888526917,
+      "num_tokens": 98224.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.7964849770069122,
+      "epoch": 0.06635071090047394,
+      "grad_norm": 2.0188069343566895,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.7727513313293457,
+      "mean_token_accuracy": 0.637175589799881,
+      "num_tokens": 113456.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.8277963697910309,
+      "epoch": 0.07582938388625593,
+      "grad_norm": 2.102283000946045,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 1.74727463722229,
+      "mean_token_accuracy": 0.6403038948774338,
+      "num_tokens": 129166.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.7185819745063782,
+      "epoch": 0.08530805687203792,
+      "grad_norm": 1.5833488702774048,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.653383731842041,
+      "mean_token_accuracy": 0.6527138650417328,
+      "num_tokens": 145103.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.7715328633785248,
+      "epoch": 0.0947867298578199,
+      "grad_norm": 1.4368863105773926,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.6846437454223633,
+      "mean_token_accuracy": 0.6343469023704529,
+      "num_tokens": 160889.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.6594811379909515,
+      "epoch": 0.10426540284360189,
+      "grad_norm": 1.1473579406738281,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.5402858257293701,
+      "mean_token_accuracy": 0.663172259926796,
+      "num_tokens": 177508.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.6768087446689606,
+      "epoch": 0.11374407582938388,
+      "grad_norm": 1.0232007503509521,
+      "learning_rate": 1e-05,
+      "loss": 1.5721584558486938,
+      "mean_token_accuracy": 0.6572738438844681,
+      "num_tokens": 194325.0,
+      "step": 12
+    },
+    {
+      "entropy": 1.7810584902763367,
+      "epoch": 0.12322274881516587,
+      "grad_norm": 1.0588256120681763,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.6563284397125244,
+      "mean_token_accuracy": 0.6360882222652435,
+      "num_tokens": 209935.0,
+      "step": 13
+    },
+    {
+      "entropy": 1.6035055816173553,
+      "epoch": 0.13270142180094788,
+      "grad_norm": 0.9055944681167603,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 1.4942286014556885,
+      "mean_token_accuracy": 0.6743725687265396,
+      "num_tokens": 226137.0,
+      "step": 14
+    },
+    {
+      "entropy": 1.5955874025821686,
+      "epoch": 0.14218009478672985,
+      "grad_norm": 0.8588143587112427,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.4966249465942383,
+      "mean_token_accuracy": 0.66853067278862,
+      "num_tokens": 241952.0,
+      "step": 15
+    },
+    {
+      "entropy": 1.607325166463852,
+      "epoch": 0.15165876777251186,
+      "grad_norm": 0.9346417784690857,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.4750392436981201,
+      "mean_token_accuracy": 0.660357192158699,
+      "num_tokens": 257481.0,
+      "step": 16
+    },
+    {
+      "entropy": 1.5410204827785492,
+      "epoch": 0.16113744075829384,
+      "grad_norm": 0.8730961680412292,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.4322866201400757,
+      "mean_token_accuracy": 0.6814255267381668,
+      "num_tokens": 273287.0,
+      "step": 17
+    },
+    {
+      "entropy": 1.5345198810100555,
+      "epoch": 0.17061611374407584,
+      "grad_norm": 0.8538835048675537,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 1.3966718912124634,
+      "mean_token_accuracy": 0.6867176294326782,
+      "num_tokens": 288809.0,
+      "step": 18
+    },
+    {
+      "entropy": 1.4507653713226318,
+      "epoch": 0.18009478672985782,
+      "grad_norm": 0.7879626154899597,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.3630776405334473,
+      "mean_token_accuracy": 0.6779211610555649,
+      "num_tokens": 304801.0,
+      "step": 19
+    },
+    {
+      "entropy": 1.4710645079612732,
+      "epoch": 0.1895734597156398,
+      "grad_norm": 0.8410776257514954,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 1.4051768779754639,
+      "mean_token_accuracy": 0.6760597079992294,
+      "num_tokens": 320246.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3435666263103485,
+      "epoch": 0.1990521327014218,
+      "grad_norm": 0.759080708026886,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.267665147781372,
+      "mean_token_accuracy": 0.7003267407417297,
+      "num_tokens": 337053.0,
+      "step": 21
+    },
+    {
+      "entropy": 1.394632339477539,
+      "epoch": 0.20853080568720378,
+      "grad_norm": 0.8370321989059448,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 1.3486745357513428,
+      "mean_token_accuracy": 0.6893624067306519,
+      "num_tokens": 352268.0,
+      "step": 22
+    },
+    {
+      "entropy": 1.1503786146640778,
+      "epoch": 0.21800947867298578,
+      "grad_norm": 0.7440245747566223,
+      "learning_rate": 2e-05,
+      "loss": 1.1109555959701538,
+      "mean_token_accuracy": 0.7320543378591537,
+      "num_tokens": 370034.0,
+      "step": 23
+    },
+    {
+      "entropy": 1.1959196627140045,
+      "epoch": 0.22748815165876776,
+      "grad_norm": 0.8951184749603271,
+      "learning_rate": 1.9999694637689328e-05,
+      "loss": 1.170793056488037,
+      "mean_token_accuracy": 0.7196634411811829,
+      "num_tokens": 385616.0,
+      "step": 24
+    },
+    {
+      "entropy": 1.1576418280601501,
+      "epoch": 0.23696682464454977,
+      "grad_norm": 0.8570479154586792,
+      "learning_rate": 1.999877856940653e-05,
+      "loss": 1.1886144876480103,
+      "mean_token_accuracy": 0.7251218110322952,
+      "num_tokens": 401820.0,
+      "step": 25
+    },
+    {
+      "entropy": 1.1380924582481384,
+      "epoch": 0.24644549763033174,
+      "grad_norm": 0.8543166518211365,
+      "learning_rate": 1.999725185109816e-05,
+      "loss": 1.2082655429840088,
+      "mean_token_accuracy": 0.711901068687439,
+      "num_tokens": 417425.0,
+      "step": 26
+    },
+    {
+      "entropy": 0.9601393789052963,
+      "epoch": 0.2559241706161137,
+      "grad_norm": 0.863011360168457,
+      "learning_rate": 1.999511457600466e-05,
+      "loss": 1.0240533351898193,
+      "mean_token_accuracy": 0.7590129524469376,
+      "num_tokens": 433582.0,
+      "step": 27
+    },
+    {
+      "entropy": 1.1091893315315247,
+      "epoch": 0.26540284360189575,
+      "grad_norm": 0.8494651317596436,
+      "learning_rate": 1.9992366874654684e-05,
+      "loss": 1.2075942754745483,
+      "mean_token_accuracy": 0.7092257142066956,
+      "num_tokens": 448489.0,
+      "step": 28
+    },
+    {
+      "entropy": 1.0761875212192535,
+      "epoch": 0.27488151658767773,
+      "grad_norm": 0.7775688767433167,
+      "learning_rate": 1.9989008914857115e-05,
+      "loss": 1.1314418315887451,
+      "mean_token_accuracy": 0.7183841317892075,
+      "num_tokens": 465063.0,
+      "step": 29
+    },
+    {
+      "entropy": 1.0514411181211472,
+      "epoch": 0.2843601895734597,
+      "grad_norm": 0.7758051753044128,
+      "learning_rate": 1.998504090169083e-05,
+      "loss": 1.1378644704818726,
+      "mean_token_accuracy": 0.7295740246772766,
+      "num_tokens": 481645.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.0378742814064026,
+      "epoch": 0.2938388625592417,
+      "grad_norm": 0.7737782597541809,
+      "learning_rate": 1.998046307749216e-05,
+      "loss": 1.0405387878417969,
+      "mean_token_accuracy": 0.7435884028673172,
+      "num_tokens": 498020.0,
+      "step": 31
+    },
+    {
+      "entropy": 1.162422239780426,
+      "epoch": 0.3033175355450237,
+      "grad_norm": 0.8285279870033264,
+      "learning_rate": 1.9975275721840105e-05,
+      "loss": 1.2049440145492554,
+      "mean_token_accuracy": 0.7072447687387466,
+      "num_tokens": 514133.0,
+      "step": 32
+    },
+    {
+      "entropy": 1.0860552787780762,
+      "epoch": 0.3127962085308057,
+      "grad_norm": 0.7870988249778748,
+      "learning_rate": 1.9969479151539238e-05,
+      "loss": 1.088366985321045,
+      "mean_token_accuracy": 0.7330214083194733,
+      "num_tokens": 529902.0,
+      "step": 33
+    },
+    {
+      "entropy": 1.0763448625802994,
+      "epoch": 0.3222748815165877,
+      "grad_norm": 0.8604663610458374,
+      "learning_rate": 1.9963073720600383e-05,
+      "loss": 1.0562834739685059,
+      "mean_token_accuracy": 0.7405408322811127,
+      "num_tokens": 545365.0,
+      "step": 34
+    },
+    {
+      "entropy": 1.0641998052597046,
+      "epoch": 0.33175355450236965,
+      "grad_norm": 0.8490760922431946,
+      "learning_rate": 1.9956059820218982e-05,
+      "loss": 1.08937668800354,
+      "mean_token_accuracy": 0.7371959090232849,
+      "num_tokens": 560629.0,
+      "step": 35
+    },
+    {
+      "entropy": 1.024262085556984,
+      "epoch": 0.3412322274881517,
+      "grad_norm": 0.7459779381752014,
+      "learning_rate": 1.99484378787512e-05,
+      "loss": 1.0313022136688232,
+      "mean_token_accuracy": 0.7482277005910873,
+      "num_tokens": 575979.0,
+      "step": 36
+    },
+    {
+      "entropy": 1.0767400860786438,
+      "epoch": 0.35071090047393366,
+      "grad_norm": 0.6798542141914368,
+      "learning_rate": 1.9940208361687762e-05,
+      "loss": 1.1029484272003174,
+      "mean_token_accuracy": 0.7400920391082764,
+      "num_tokens": 593219.0,
+      "step": 37
+    },
+    {
+      "entropy": 1.0605872124433517,
+      "epoch": 0.36018957345971564,
+      "grad_norm": 0.775074303150177,
+      "learning_rate": 1.9931371771625545e-05,
+      "loss": 1.1065254211425781,
+      "mean_token_accuracy": 0.7417906671762466,
+      "num_tokens": 608173.0,
+      "step": 38
+    },
+    {
+      "entropy": 1.002692699432373,
+      "epoch": 0.3696682464454976,
+      "grad_norm": 0.7341852784156799,
+      "learning_rate": 1.9921928648236855e-05,
+      "loss": 1.013192892074585,
+      "mean_token_accuracy": 0.7493702918291092,
+      "num_tokens": 624455.0,
+      "step": 39
+    },
+    {
+      "entropy": 1.0264743566513062,
+      "epoch": 0.3791469194312796,
+      "grad_norm": 0.709384024143219,
+      "learning_rate": 1.9911879568236492e-05,
+      "loss": 1.050273060798645,
+      "mean_token_accuracy": 0.7467086464166641,
+      "num_tokens": 639997.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.9731618463993073,
+      "epoch": 0.3886255924170616,
+      "grad_norm": 0.7397781014442444,
+      "learning_rate": 1.990122514534651e-05,
+      "loss": 0.9792512655258179,
+      "mean_token_accuracy": 0.7615833282470703,
+      "num_tokens": 655244.0,
+      "step": 41
+    },
+    {
+      "entropy": 0.9393151998519897,
+      "epoch": 0.3981042654028436,
+      "grad_norm": 0.603070080280304,
+      "learning_rate": 1.9889966030258752e-05,
+      "loss": 0.9558578729629517,
+      "mean_token_accuracy": 0.7628461569547653,
+      "num_tokens": 672691.0,
+      "step": 42
+    },
+    {
+      "entropy": 0.9829478412866592,
+      "epoch": 0.4075829383886256,
+      "grad_norm": 0.736235499382019,
+      "learning_rate": 1.9878102910595097e-05,
+      "loss": 0.978938639163971,
+      "mean_token_accuracy": 0.7605015784502029,
+      "num_tokens": 688436.0,
+      "step": 43
+    },
+    {
+      "entropy": 0.9505813121795654,
+      "epoch": 0.41706161137440756,
+      "grad_norm": 0.6724395751953125,
+      "learning_rate": 1.9865636510865466e-05,
+      "loss": 0.978853702545166,
+      "mean_token_accuracy": 0.762572169303894,
+      "num_tokens": 705439.0,
+      "step": 44
+    },
+    {
+      "entropy": 0.9644436687231064,
+      "epoch": 0.4265402843601896,
+      "grad_norm": 0.7388243079185486,
+      "learning_rate": 1.985256759242359e-05,
+      "loss": 1.0140184164047241,
+      "mean_token_accuracy": 0.7558823525905609,
+      "num_tokens": 721153.0,
+      "step": 45
+    },
+    {
+      "entropy": 0.9982628226280212,
+      "epoch": 0.43601895734597157,
+      "grad_norm": 0.8184957504272461,
+      "learning_rate": 1.9838896953420495e-05,
+      "loss": 1.0390454530715942,
+      "mean_token_accuracy": 0.7469227313995361,
+      "num_tokens": 736548.0,
+      "step": 46
+    },
+    {
+      "entropy": 0.9748822003602982,
+      "epoch": 0.44549763033175355,
+      "grad_norm": 0.7347403764724731,
+      "learning_rate": 1.982462542875576e-05,
+      "loss": 0.9972015619277954,
+      "mean_token_accuracy": 0.7542161792516708,
+      "num_tokens": 753213.0,
+      "step": 47
+    },
+    {
+      "entropy": 0.9839194715023041,
+      "epoch": 0.4549763033175355,
+      "grad_norm": 0.7843495607376099,
+      "learning_rate": 1.9809753890026543e-05,
+      "loss": 0.9987256526947021,
+      "mean_token_accuracy": 0.7523878663778305,
+      "num_tokens": 768748.0,
+      "step": 48
+    },
+    {
+      "entropy": 1.0193238854408264,
+      "epoch": 0.46445497630331756,
+      "grad_norm": 0.7180641889572144,
+      "learning_rate": 1.979428324547432e-05,
+      "loss": 1.0118625164031982,
+      "mean_token_accuracy": 0.7445009052753448,
+      "num_tokens": 783698.0,
+      "step": 49
+    },
+    {
+      "entropy": 0.9688207358121872,
+      "epoch": 0.47393364928909953,
+      "grad_norm": 0.7978101372718811,
+      "learning_rate": 1.9778214439929453e-05,
+      "loss": 0.9732460379600525,
+      "mean_token_accuracy": 0.7674588412046432,
+      "num_tokens": 799825.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.000299409031868,
+      "epoch": 0.4834123222748815,
+      "grad_norm": 0.7578161358833313,
+      "learning_rate": 1.9761548454753455e-05,
+      "loss": 1.0048537254333496,
+      "mean_token_accuracy": 0.7620798051357269,
+      "num_tokens": 814971.0,
+      "step": 51
+    },
+    {
+      "entropy": 0.9922938197851181,
+      "epoch": 0.4928909952606635,
+      "grad_norm": 0.7843224406242371,
+      "learning_rate": 1.9744286307779076e-05,
+      "loss": 0.9955688714981079,
+      "mean_token_accuracy": 0.7550618052482605,
+      "num_tokens": 830524.0,
+      "step": 52
+    },
+    {
+      "entropy": 0.9540591835975647,
+      "epoch": 0.5023696682464455,
+      "grad_norm": 0.7341641783714294,
+      "learning_rate": 1.972642905324813e-05,
+      "loss": 0.9120223522186279,
+      "mean_token_accuracy": 0.7745194137096405,
+      "num_tokens": 845863.0,
+      "step": 53
+    },
+    {
+      "entropy": 0.9302634298801422,
+      "epoch": 0.5118483412322274,
+      "grad_norm": 0.7071745991706848,
+      "learning_rate": 1.9707977781747126e-05,
+      "loss": 0.8850457668304443,
+      "mean_token_accuracy": 0.7787033319473267,
+      "num_tokens": 861849.0,
+      "step": 54
+    },
+    {
+      "entropy": 0.9491895437240601,
+      "epoch": 0.5213270142180095,
+      "grad_norm": 0.826166033744812,
+      "learning_rate": 1.9688933620140638e-05,
+      "loss": 0.9278963804244995,
+      "mean_token_accuracy": 0.7744901925325394,
+      "num_tokens": 877161.0,
+      "step": 55
+    },
+    {
+      "entropy": 0.990895539522171,
+      "epoch": 0.5308056872037915,
+      "grad_norm": 0.94631028175354,
+      "learning_rate": 1.966929773150251e-05,
+      "loss": 0.9650328159332275,
+      "mean_token_accuracy": 0.7605593204498291,
+      "num_tokens": 892160.0,
+      "step": 56
+    },
+    {
+      "entropy": 0.9609559625387192,
+      "epoch": 0.5402843601895735,
+      "grad_norm": 0.6704272031784058,
+      "learning_rate": 1.96490713150448e-05,
+      "loss": 0.9601552486419678,
+      "mean_token_accuracy": 0.7565764933824539,
+      "num_tokens": 908720.0,
+      "step": 57
+    },
+    {
+      "entropy": 0.9724608361721039,
+      "epoch": 0.5497630331753555,
+      "grad_norm": 0.7987351417541504,
+      "learning_rate": 1.9628255606044562e-05,
+      "loss": 0.9856890439987183,
+      "mean_token_accuracy": 0.7569890469312668,
+      "num_tokens": 924429.0,
+      "step": 58
+    },
+    {
+      "entropy": 0.9434510767459869,
+      "epoch": 0.5592417061611374,
+      "grad_norm": 0.7899335622787476,
+      "learning_rate": 1.9606851875768404e-05,
+      "loss": 0.9372917413711548,
+      "mean_token_accuracy": 0.7673724591732025,
+      "num_tokens": 940816.0,
+      "step": 59
+    },
+    {
+      "entropy": 0.8971584439277649,
+      "epoch": 0.5687203791469194,
+      "grad_norm": 0.7808126211166382,
+      "learning_rate": 1.9584861431394825e-05,
+      "loss": 0.8875389099121094,
+      "mean_token_accuracy": 0.7799597531557083,
+      "num_tokens": 956724.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.9393307864665985,
+      "epoch": 0.5781990521327014,
+      "grad_norm": 0.6976328492164612,
+      "learning_rate": 1.956228561593441e-05,
+      "loss": 0.9503510594367981,
+      "mean_token_accuracy": 0.7596384882926941,
+      "num_tokens": 973257.0,
+      "step": 61
+    },
+    {
+      "entropy": 1.021749958395958,
+      "epoch": 0.5876777251184834,
+      "grad_norm": 0.7356151342391968,
+      "learning_rate": 1.953912580814779e-05,
+      "loss": 1.0444157123565674,
+      "mean_token_accuracy": 0.7519654631614685,
+      "num_tokens": 990266.0,
+      "step": 62
+    },
+    {
+      "entropy": 0.8811668455600739,
+      "epoch": 0.5971563981042654,
+      "grad_norm": 0.8034309148788452,
+      "learning_rate": 1.9515383422461457e-05,
+      "loss": 0.888944685459137,
+      "mean_token_accuracy": 0.77887162566185,
+      "num_tokens": 1006065.0,
+      "step": 63
+    },
+    {
+      "entropy": 0.9695002734661102,
+      "epoch": 0.6066350710900474,
+      "grad_norm": 0.6834648251533508,
+      "learning_rate": 1.949105990888135e-05,
+      "loss": 0.9581116437911987,
+      "mean_token_accuracy": 0.7590216845273972,
+      "num_tokens": 1021785.0,
+      "step": 64
+    },
+    {
+      "entropy": 0.8993494361639023,
+      "epoch": 0.6161137440758294,
+      "grad_norm": 0.787246584892273,
+      "learning_rate": 1.9466156752904344e-05,
+      "loss": 0.874479353427887,
+      "mean_token_accuracy": 0.7846818715333939,
+      "num_tokens": 1037721.0,
+      "step": 65
+    },
+    {
+      "entropy": 0.9583495110273361,
+      "epoch": 0.6255924170616114,
+      "grad_norm": 0.7072925567626953,
+      "learning_rate": 1.944067547542748e-05,
+      "loss": 0.9677348136901855,
+      "mean_token_accuracy": 0.7663833945989609,
+      "num_tokens": 1054125.0,
+      "step": 66
+    },
+    {
+      "entropy": 0.926140233874321,
+      "epoch": 0.6350710900473934,
+      "grad_norm": 0.7638861536979675,
+      "learning_rate": 1.9414617632655114e-05,
+      "loss": 0.9518347382545471,
+      "mean_token_accuracy": 0.7670052349567413,
+      "num_tokens": 1070665.0,
+      "step": 67
+    },
+    {
+      "entropy": 0.9343065768480301,
+      "epoch": 0.6445497630331753,
+      "grad_norm": 0.8179787397384644,
+      "learning_rate": 1.9387984816003868e-05,
+      "loss": 0.9283621311187744,
+      "mean_token_accuracy": 0.7648810148239136,
+      "num_tokens": 1086022.0,
+      "step": 68
+    },
+    {
+      "entropy": 1.0115204006433487,
+      "epoch": 0.6540284360189573,
+      "grad_norm": 0.9361454844474792,
+      "learning_rate": 1.9360778652005416e-05,
+      "loss": 1.043932557106018,
+      "mean_token_accuracy": 0.7457028925418854,
+      "num_tokens": 1101344.0,
+      "step": 69
+    },
+    {
+      "entropy": 0.919638603925705,
+      "epoch": 0.6635071090047393,
+      "grad_norm": 0.7808894515037537,
+      "learning_rate": 1.933300080220719e-05,
+      "loss": 0.877983570098877,
+      "mean_token_accuracy": 0.7854871153831482,
+      "num_tokens": 1117702.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.9225042313337326,
+      "epoch": 0.6729857819905213,
+      "grad_norm": 0.7920427918434143,
+      "learning_rate": 1.9304652963070868e-05,
+      "loss": 0.8995954990386963,
+      "mean_token_accuracy": 0.7772288471460342,
+      "num_tokens": 1134538.0,
+      "step": 71
+    },
+    {
+      "entropy": 0.9348872601985931,
+      "epoch": 0.6824644549763034,
+      "grad_norm": 0.7429097294807434,
+      "learning_rate": 1.927573686586878e-05,
+      "loss": 0.9216326475143433,
+      "mean_token_accuracy": 0.7696940153837204,
+      "num_tokens": 1150205.0,
+      "step": 72
+    },
+    {
+      "entropy": 0.8841167837381363,
+      "epoch": 0.6919431279620853,
+      "grad_norm": 0.9018954634666443,
+      "learning_rate": 1.9246254276578175e-05,
+      "loss": 0.8575382232666016,
+      "mean_token_accuracy": 0.7889357209205627,
+      "num_tokens": 1166758.0,
+      "step": 73
+    },
+    {
+      "entropy": 0.8816715627908707,
+      "epoch": 0.7014218009478673,
+      "grad_norm": 0.8096207976341248,
+      "learning_rate": 1.9216206995773373e-05,
+      "loss": 0.8846977949142456,
+      "mean_token_accuracy": 0.7817238718271255,
+      "num_tokens": 1182364.0,
+      "step": 74
+    },
+    {
+      "entropy": 0.9715549945831299,
+      "epoch": 0.7109004739336493,
+      "grad_norm": 0.9433501958847046,
+      "learning_rate": 1.9185596858515797e-05,
+      "loss": 0.9738811254501343,
+      "mean_token_accuracy": 0.7634537816047668,
+      "num_tokens": 1196839.0,
+      "step": 75
+    },
+    {
+      "entropy": 0.8470079302787781,
+      "epoch": 0.7203791469194313,
+      "grad_norm": 0.7724857926368713,
+      "learning_rate": 1.9154425734241893e-05,
+      "loss": 0.8701345920562744,
+      "mean_token_accuracy": 0.78440722823143,
+      "num_tokens": 1213377.0,
+      "step": 76
+    },
+    {
+      "entropy": 0.9654579311609268,
+      "epoch": 0.7298578199052133,
+      "grad_norm": 0.8752750158309937,
+      "learning_rate": 1.9122695526648968e-05,
+      "loss": 0.9896650910377502,
+      "mean_token_accuracy": 0.752353623509407,
+      "num_tokens": 1229571.0,
+      "step": 77
+    },
+    {
+      "entropy": 0.8873108923435211,
+      "epoch": 0.7393364928909952,
+      "grad_norm": 0.7646530866622925,
+      "learning_rate": 1.9090408173578923e-05,
+      "loss": 0.8805227279663086,
+      "mean_token_accuracy": 0.7848268151283264,
+      "num_tokens": 1247252.0,
+      "step": 78
+    },
+    {
+      "entropy": 0.8589294850826263,
+      "epoch": 0.7488151658767772,
+      "grad_norm": 0.8699224591255188,
+      "learning_rate": 1.905756564689991e-05,
+      "loss": 0.8414448499679565,
+      "mean_token_accuracy": 0.7981384843587875,
+      "num_tokens": 1264824.0,
+      "step": 79
+    },
+    {
+      "entropy": 0.9470502883195877,
+      "epoch": 0.7582938388625592,
+      "grad_norm": 0.7720149159431458,
+      "learning_rate": 1.9024169952385887e-05,
+      "loss": 0.9226930141448975,
+      "mean_token_accuracy": 0.7769389748573303,
+      "num_tokens": 1279205.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.9329600483179092,
+      "epoch": 0.7677725118483413,
+      "grad_norm": 0.7501013875007629,
+      "learning_rate": 1.8990223129594146e-05,
+      "loss": 0.9248684644699097,
+      "mean_token_accuracy": 0.7771976292133331,
+      "num_tokens": 1295447.0,
+      "step": 81
+    },
+    {
+      "entropy": 0.9092026650905609,
+      "epoch": 0.7772511848341233,
+      "grad_norm": 0.7927464842796326,
+      "learning_rate": 1.8955727251740742e-05,
+      "loss": 0.9005250930786133,
+      "mean_token_accuracy": 0.7794855386018753,
+      "num_tokens": 1310786.0,
+      "step": 82
+    },
+    {
+      "entropy": 0.9609334319829941,
+      "epoch": 0.7867298578199052,
+      "grad_norm": 0.8200048804283142,
+      "learning_rate": 1.8920684425573865e-05,
+      "loss": 1.0015482902526855,
+      "mean_token_accuracy": 0.7510274350643158,
+      "num_tokens": 1326327.0,
+      "step": 83
+    },
+    {
+      "entropy": 0.9021405577659607,
+      "epoch": 0.7962085308056872,
+      "grad_norm": 0.7447288632392883,
+      "learning_rate": 1.888509679124519e-05,
+      "loss": 0.9190552234649658,
+      "mean_token_accuracy": 0.7851650565862656,
+      "num_tokens": 1342612.0,
+      "step": 84
+    },
+    {
+      "entropy": 0.8972304910421371,
+      "epoch": 0.8056872037914692,
+      "grad_norm": 0.8647413849830627,
+      "learning_rate": 1.884896652217917e-05,
+      "loss": 0.8718003034591675,
+      "mean_token_accuracy": 0.7822871953248978,
+      "num_tokens": 1358216.0,
+      "step": 85
+    },
+    {
+      "entropy": 0.8696665316820145,
+      "epoch": 0.8151658767772512,
+      "grad_norm": 0.8497427105903625,
+      "learning_rate": 1.8812295824940284e-05,
+      "loss": 0.8930065035820007,
+      "mean_token_accuracy": 0.7775844633579254,
+      "num_tokens": 1374136.0,
+      "step": 86
+    },
+    {
+      "entropy": 0.8540368229150772,
+      "epoch": 0.8246445497630331,
+      "grad_norm": 0.9805054068565369,
+      "learning_rate": 1.877508693909831e-05,
+      "loss": 0.8588962554931641,
+      "mean_token_accuracy": 0.7839875668287277,
+      "num_tokens": 1390236.0,
+      "step": 87
+    },
+    {
+      "entropy": 0.9288308620452881,
+      "epoch": 0.8341232227488151,
+      "grad_norm": 0.9268847107887268,
+      "learning_rate": 1.8737342137091523e-05,
+      "loss": 0.9457352161407471,
+      "mean_token_accuracy": 0.7653509080410004,
+      "num_tokens": 1405289.0,
+      "step": 88
+    },
+    {
+      "entropy": 0.8685550093650818,
+      "epoch": 0.8436018957345972,
+      "grad_norm": 0.7077257037162781,
+      "learning_rate": 1.8699063724087905e-05,
+      "loss": 0.8259266018867493,
+      "mean_token_accuracy": 0.7899300456047058,
+      "num_tokens": 1419954.0,
+      "step": 89
+    },
+    {
+      "entropy": 0.7893514037132263,
+      "epoch": 0.8530805687203792,
+      "grad_norm": 0.6737422943115234,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.7658373713493347,
+      "mean_token_accuracy": 0.8061323910951614,
+      "num_tokens": 1436833.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.874814048409462,
+      "epoch": 0.8625592417061612,
+      "grad_norm": 0.7337303757667542,
+      "learning_rate": 1.862091544856407e-05,
+      "loss": 0.881226658821106,
+      "mean_token_accuracy": 0.7787951081991196,
+      "num_tokens": 1452006.0,
+      "step": 91
+    },
+    {
+      "entropy": 0.8939783275127411,
+      "epoch": 0.8720379146919431,
+      "grad_norm": 0.8276046514511108,
+      "learning_rate": 1.8581050358751444e-05,
+      "loss": 0.8792139291763306,
+      "mean_token_accuracy": 0.7856588512659073,
+      "num_tokens": 1467177.0,
+      "step": 92
+    },
+    {
+      "entropy": 0.9284755736589432,
+      "epoch": 0.8815165876777251,
+      "grad_norm": 0.7748722434043884,
+      "learning_rate": 1.854066120306571e-05,
+      "loss": 0.9296675324440002,
+      "mean_token_accuracy": 0.7717541307210922,
+      "num_tokens": 1481900.0,
+      "step": 93
+    },
+    {
+      "entropy": 0.8540140986442566,
+      "epoch": 0.8909952606635071,
+      "grad_norm": 0.6757642030715942,
+      "learning_rate": 1.8499750448172046e-05,
+      "loss": 0.8415413498878479,
+      "mean_token_accuracy": 0.7882983386516571,
+      "num_tokens": 1498093.0,
+      "step": 94
+    },
+    {
+      "entropy": 0.8530154079198837,
+      "epoch": 0.9004739336492891,
+      "grad_norm": 0.7341352105140686,
+      "learning_rate": 1.8458320592590976e-05,
+      "loss": 0.8625093698501587,
+      "mean_token_accuracy": 0.7759335339069366,
+      "num_tokens": 1514287.0,
+      "step": 95
+    },
+    {
+      "entropy": 0.7864818871021271,
+      "epoch": 0.909952606635071,
+      "grad_norm": 0.7490381598472595,
+      "learning_rate": 1.841637416654579e-05,
+      "loss": 0.8058528900146484,
+      "mean_token_accuracy": 0.7964929491281509,
+      "num_tokens": 1530706.0,
+      "step": 96
+    },
+    {
+      "entropy": 0.8394529819488525,
+      "epoch": 0.919431279620853,
+      "grad_norm": 0.6680670380592346,
+      "learning_rate": 1.837391373180801e-05,
+      "loss": 0.8252418041229248,
+      "mean_token_accuracy": 0.7954028099775314,
+      "num_tokens": 1547595.0,
+      "step": 97
+    },
+    {
+      "entropy": 0.8288186490535736,
+      "epoch": 0.9289099526066351,
+      "grad_norm": 0.7968632578849792,
+      "learning_rate": 1.8330941881540917e-05,
+      "loss": 0.8349557518959045,
+      "mean_token_accuracy": 0.7844376415014267,
+      "num_tokens": 1563028.0,
+      "step": 98
+    },
+    {
+      "entropy": 0.8333224058151245,
+      "epoch": 0.9383886255924171,
+      "grad_norm": 0.691230297088623,
+      "learning_rate": 1.8287461240141217e-05,
+      "loss": 0.8373101353645325,
+      "mean_token_accuracy": 0.7931841313838959,
+      "num_tokens": 1578461.0,
+      "step": 99
+    },
+    {
+      "entropy": 0.9056068807840347,
+      "epoch": 0.9478672985781991,
+      "grad_norm": 0.81464022397995,
+      "learning_rate": 1.8243474463078738e-05,
+      "loss": 0.9318811297416687,
+      "mean_token_accuracy": 0.7732242494821548,
+      "num_tokens": 1594199.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.8589460551738739,
+      "epoch": 0.957345971563981,
+      "grad_norm": 0.8496865630149841,
+      "learning_rate": 1.8198984236734246e-05,
+      "loss": 0.8577089905738831,
+      "mean_token_accuracy": 0.7863648384809494,
+      "num_tokens": 1612009.0,
+      "step": 101
+    },
+    {
+      "entropy": 0.9072522073984146,
+      "epoch": 0.966824644549763,
+      "grad_norm": 0.6795774102210999,
+      "learning_rate": 1.8153993278235416e-05,
+      "loss": 0.9616482853889465,
+      "mean_token_accuracy": 0.7749254554510117,
+      "num_tokens": 1627621.0,
+      "step": 102
+    },
+    {
+      "entropy": 0.824530228972435,
+      "epoch": 0.976303317535545,
+      "grad_norm": 0.7762973308563232,
+      "learning_rate": 1.8108504335290852e-05,
+      "loss": 0.8160134553909302,
+      "mean_token_accuracy": 0.7989892214536667,
+      "num_tokens": 1642633.0,
+      "step": 103
+    },
+    {
+      "entropy": 0.8416652232408524,
+      "epoch": 0.985781990521327,
+      "grad_norm": 0.689736008644104,
+      "learning_rate": 1.80625201860223e-05,
+      "loss": 0.8192683458328247,
+      "mean_token_accuracy": 0.8011076003313065,
+      "num_tokens": 1658656.0,
+      "step": 104
+    },
+    {
+      "entropy": 0.8708516359329224,
+      "epoch": 0.995260663507109,
+      "grad_norm": 0.6898924708366394,
+      "learning_rate": 1.8016043638794975e-05,
+      "loss": 0.8683778047561646,
+      "mean_token_accuracy": 0.788833886384964,
+      "num_tokens": 1674845.0,
+      "step": 105
+    },
+    {
+      "entropy": 0.7872009575366974,
+      "epoch": 1.0,
+      "grad_norm": 0.8945826292037964,
+      "learning_rate": 1.7969077532046047e-05,
+      "loss": 0.7540068030357361,
+      "mean_token_accuracy": 0.8025133311748505,
+      "num_tokens": 1682713.0,
+      "step": 106
+    },
+    {
+      "entropy": 0.8132580518722534,
+      "epoch": 1.009478672985782,
+      "grad_norm": 0.7014424800872803,
+      "learning_rate": 1.7921624734111292e-05,
+      "loss": 0.7927298545837402,
+      "mean_token_accuracy": 0.7970700860023499,
+      "num_tokens": 1698838.0,
+      "step": 107
+    },
+    {
+      "entropy": 0.9552255123853683,
+      "epoch": 1.018957345971564,
+      "grad_norm": 0.7601428031921387,
+      "learning_rate": 1.787368814304992e-05,
+      "loss": 0.9353047609329224,
+      "mean_token_accuracy": 0.7636215835809708,
+      "num_tokens": 1714529.0,
+      "step": 108
+    },
+    {
+      "entropy": 0.8868769556283951,
+      "epoch": 1.028436018957346,
+      "grad_norm": 0.7216587066650391,
+      "learning_rate": 1.7825270686467567e-05,
+      "loss": 0.8972142934799194,
+      "mean_token_accuracy": 0.7766188830137253,
+      "num_tokens": 1730585.0,
+      "step": 109
+    },
+    {
+      "entropy": 0.7498472779989243,
+      "epoch": 1.037914691943128,
+      "grad_norm": 0.7536665201187134,
+      "learning_rate": 1.7776375321337523e-05,
+      "loss": 0.7597789764404297,
+      "mean_token_accuracy": 0.8086548745632172,
+      "num_tokens": 1747998.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.8525176346302032,
+      "epoch": 1.04739336492891,
+      "grad_norm": 0.6725131869316101,
+      "learning_rate": 1.7727005033820117e-05,
+      "loss": 0.8725385665893555,
+      "mean_token_accuracy": 0.7857676595449448,
+      "num_tokens": 1763858.0,
+      "step": 111
+    },
+    {
+      "entropy": 0.8737937361001968,
+      "epoch": 1.0568720379146919,
+      "grad_norm": 0.6200551390647888,
+      "learning_rate": 1.7677162839080365e-05,
+      "loss": 0.8674412965774536,
+      "mean_token_accuracy": 0.7846900969743729,
+      "num_tokens": 1779800.0,
+      "step": 112
+    },
+    {
+      "entropy": 0.8540749698877335,
+      "epoch": 1.066350710900474,
+      "grad_norm": 0.7130475640296936,
+      "learning_rate": 1.762685178110382e-05,
+      "loss": 0.8417346477508545,
+      "mean_token_accuracy": 0.7876130044460297,
+      "num_tokens": 1795033.0,
+      "step": 113
+    },
+    {
+      "entropy": 0.7300510853528976,
+      "epoch": 1.0758293838862558,
+      "grad_norm": 0.7067408561706543,
+      "learning_rate": 1.757607493251067e-05,
+      "loss": 0.7173537015914917,
+      "mean_token_accuracy": 0.8160942941904068,
+      "num_tokens": 1811709.0,
+      "step": 114
+    },
+    {
+      "entropy": 0.7862643301486969,
+      "epoch": 1.085308056872038,
+      "grad_norm": 0.6860957145690918,
+      "learning_rate": 1.752483539436807e-05,
+      "loss": 0.7681137323379517,
+      "mean_token_accuracy": 0.8050192445516586,
+      "num_tokens": 1828031.0,
+      "step": 115
+    },
+    {
+      "entropy": 0.8223290294408798,
+      "epoch": 1.09478672985782,
+      "grad_norm": 13.119089126586914,
+      "learning_rate": 1.747313629600077e-05,
+      "loss": 0.819756805896759,
+      "mean_token_accuracy": 0.7999143302440643,
+      "num_tokens": 1844448.0,
+      "step": 116
+    },
+    {
+      "entropy": 0.8199901729822159,
+      "epoch": 1.1042654028436019,
+      "grad_norm": 0.7526054978370667,
+      "learning_rate": 1.7420980794800013e-05,
+      "loss": 0.8049260377883911,
+      "mean_token_accuracy": 0.7947445511817932,
+      "num_tokens": 1860116.0,
+      "step": 117
+    },
+    {
+      "entropy": 0.9044909924268723,
+      "epoch": 1.113744075829384,
+      "grad_norm": 0.774227499961853,
+      "learning_rate": 1.7368372076030654e-05,
+      "loss": 0.8923323750495911,
+      "mean_token_accuracy": 0.7806955128908157,
+      "num_tokens": 1875432.0,
+      "step": 118
+    },
+    {
+      "entropy": 0.8348246514797211,
+      "epoch": 1.1232227488151658,
+      "grad_norm": 0.767121434211731,
+      "learning_rate": 1.731531335263669e-05,
+      "loss": 0.8475579023361206,
+      "mean_token_accuracy": 0.795254722237587,
+      "num_tokens": 1891887.0,
+      "step": 119
+    },
+    {
+      "entropy": 0.8337329626083374,
+      "epoch": 1.132701421800948,
+      "grad_norm": 0.6577139496803284,
+      "learning_rate": 1.726180786504499e-05,
+      "loss": 0.8073049783706665,
+      "mean_token_accuracy": 0.7889072597026825,
+      "num_tokens": 1908925.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.787194699048996,
+      "epoch": 1.1421800947867298,
+      "grad_norm": 0.7494071125984192,
+      "learning_rate": 1.720785888096743e-05,
+      "loss": 0.7753745913505554,
+      "mean_token_accuracy": 0.8019598126411438,
+      "num_tokens": 1924838.0,
+      "step": 121
+    },
+    {
+      "entropy": 0.7965055257081985,
+      "epoch": 1.1516587677725119,
+      "grad_norm": 0.7918292880058289,
+      "learning_rate": 1.7153469695201278e-05,
+      "loss": 0.7967703342437744,
+      "mean_token_accuracy": 0.7995534092187881,
+      "num_tokens": 1940925.0,
+      "step": 122
+    },
+    {
+      "entropy": 0.860944464802742,
+      "epoch": 1.161137440758294,
+      "grad_norm": 0.7927801609039307,
+      "learning_rate": 1.7098643629428035e-05,
+      "loss": 0.8686648607254028,
+      "mean_token_accuracy": 0.7804652601480484,
+      "num_tokens": 1956955.0,
+      "step": 123
+    },
+    {
+      "entropy": 0.7591796815395355,
+      "epoch": 1.1706161137440758,
+      "grad_norm": 0.7671242356300354,
+      "learning_rate": 1.7043384032010523e-05,
+      "loss": 0.8130097389221191,
+      "mean_token_accuracy": 0.8032640516757965,
+      "num_tokens": 1972680.0,
+      "step": 124
+    },
+    {
+      "entropy": 0.8104424327611923,
+      "epoch": 1.180094786729858,
+      "grad_norm": 0.8326568603515625,
+      "learning_rate": 1.698769427778842e-05,
+      "loss": 0.8218762874603271,
+      "mean_token_accuracy": 0.793466106057167,
+      "num_tokens": 1988526.0,
+      "step": 125
+    },
+    {
+      "entropy": 0.8348822742700577,
+      "epoch": 1.1895734597156398,
+      "grad_norm": 0.7889320850372314,
+      "learning_rate": 1.693157776787212e-05,
+      "loss": 0.8524150252342224,
+      "mean_token_accuracy": 0.7928779572248459,
+      "num_tokens": 2004119.0,
+      "step": 126
+    },
+    {
+      "entropy": 0.8161694556474686,
+      "epoch": 1.1990521327014219,
+      "grad_norm": 0.7603635191917419,
+      "learning_rate": 1.687503792943506e-05,
+      "loss": 0.8084812164306641,
+      "mean_token_accuracy": 0.7948593199253082,
+      "num_tokens": 2020391.0,
+      "step": 127
+    },
+    {
+      "entropy": 0.8825703114271164,
+      "epoch": 1.2085308056872037,
+      "grad_norm": 0.7401167750358582,
+      "learning_rate": 1.681807821550438e-05,
+      "loss": 0.8706303238868713,
+      "mean_token_accuracy": 0.7867066264152527,
+      "num_tokens": 2035638.0,
+      "step": 128
+    },
+    {
+      "entropy": 0.7736850678920746,
+      "epoch": 1.2180094786729858,
+      "grad_norm": 0.6535520553588867,
+      "learning_rate": 1.6760702104750046e-05,
+      "loss": 0.7504853010177612,
+      "mean_token_accuracy": 0.8084538877010345,
+      "num_tokens": 2052292.0,
+      "step": 129
+    },
+    {
+      "entropy": 0.9081571251153946,
+      "epoch": 1.2274881516587677,
+      "grad_norm": 0.6775054335594177,
+      "learning_rate": 1.670291310127242e-05,
+      "loss": 0.9248923659324646,
+      "mean_token_accuracy": 0.7756912559270859,
+      "num_tokens": 2068464.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.8087949007749557,
+      "epoch": 1.2369668246445498,
+      "grad_norm": 0.8837705850601196,
+      "learning_rate": 1.664471473438822e-05,
+      "loss": 0.7927989959716797,
+      "mean_token_accuracy": 0.7962196916341782,
+      "num_tokens": 2084023.0,
+      "step": 131
+    },
+    {
+      "entropy": 0.8239958882331848,
+      "epoch": 1.2464454976303316,
+      "grad_norm": 0.7673121690750122,
+      "learning_rate": 1.6586110558415006e-05,
+      "loss": 0.7868762016296387,
+      "mean_token_accuracy": 0.8091334104537964,
+      "num_tokens": 2100036.0,
+      "step": 132
+    },
+    {
+      "entropy": 0.7624151110649109,
+      "epoch": 1.2559241706161137,
+      "grad_norm": 0.6538701057434082,
+      "learning_rate": 1.6527104152454096e-05,
+      "loss": 0.7488104104995728,
+      "mean_token_accuracy": 0.8097258806228638,
+      "num_tokens": 2117206.0,
+      "step": 133
+    },
+    {
+      "entropy": 0.8748339861631393,
+      "epoch": 1.2654028436018958,
+      "grad_norm": 0.6970165967941284,
+      "learning_rate": 1.646769912017199e-05,
+      "loss": 0.8594735860824585,
+      "mean_token_accuracy": 0.7816884070634842,
+      "num_tokens": 2132466.0,
+      "step": 134
+    },
+    {
+      "entropy": 0.7696728706359863,
+      "epoch": 1.2748815165876777,
+      "grad_norm": 0.6646419763565063,
+      "learning_rate": 1.6407899089580263e-05,
+      "loss": 0.7375603914260864,
+      "mean_token_accuracy": 0.8088530749082565,
+      "num_tokens": 2148789.0,
+      "step": 135
+    },
+    {
+      "entropy": 0.768274188041687,
+      "epoch": 1.2843601895734598,
+      "grad_norm": 0.6656033992767334,
+      "learning_rate": 1.6347707712814023e-05,
+      "loss": 0.7362462878227234,
+      "mean_token_accuracy": 0.81095190346241,
+      "num_tokens": 2166975.0,
+      "step": 136
+    },
+    {
+      "entropy": 0.8369315713644028,
+      "epoch": 1.2938388625592416,
+      "grad_norm": 0.7545751929283142,
+      "learning_rate": 1.628712866590885e-05,
+      "loss": 0.8421776294708252,
+      "mean_token_accuracy": 0.789647251367569,
+      "num_tokens": 2182471.0,
+      "step": 137
+    },
+    {
+      "entropy": 0.9027784168720245,
+      "epoch": 1.3033175355450237,
+      "grad_norm": 0.8744738698005676,
+      "learning_rate": 1.622616564857629e-05,
+      "loss": 0.8971675038337708,
+      "mean_token_accuracy": 0.7683477252721786,
+      "num_tokens": 2197146.0,
+      "step": 138
+    },
+    {
+      "entropy": 0.7965396195650101,
+      "epoch": 1.3127962085308056,
+      "grad_norm": 0.7402711510658264,
+      "learning_rate": 1.6164822383977912e-05,
+      "loss": 0.8043927550315857,
+      "mean_token_accuracy": 0.7958309352397919,
+      "num_tokens": 2211761.0,
+      "step": 139
+    },
+    {
+      "entropy": 0.8152131289243698,
+      "epoch": 1.3222748815165877,
+      "grad_norm": 1.6655291318893433,
+      "learning_rate": 1.6103102618497922e-05,
+      "loss": 0.8047835230827332,
+      "mean_token_accuracy": 0.7941760718822479,
+      "num_tokens": 2229548.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.8045339435338974,
+      "epoch": 1.3317535545023698,
+      "grad_norm": 0.8306461572647095,
+      "learning_rate": 1.604101012151436e-05,
+      "loss": 0.8004187345504761,
+      "mean_token_accuracy": 0.7995999902486801,
+      "num_tokens": 2244710.0,
+      "step": 141
+    },
+    {
+      "entropy": 0.8147155493497849,
+      "epoch": 1.3412322274881516,
+      "grad_norm": 0.6874533891677856,
+      "learning_rate": 1.5978548685168892e-05,
+      "loss": 0.8141619563102722,
+      "mean_token_accuracy": 0.799629807472229,
+      "num_tokens": 2261252.0,
+      "step": 142
+    },
+    {
+      "entropy": 0.8661371916532516,
+      "epoch": 1.3507109004739337,
+      "grad_norm": 0.742089033126831,
+      "learning_rate": 1.5915722124135227e-05,
+      "loss": 0.8675738573074341,
+      "mean_token_accuracy": 0.7867332845926285,
+      "num_tokens": 2277329.0,
+      "step": 143
+    },
+    {
+      "entropy": 0.8389143347740173,
+      "epoch": 1.3601895734597156,
+      "grad_norm": 0.7694486379623413,
+      "learning_rate": 1.5852534275386133e-05,
+      "loss": 0.8360339403152466,
+      "mean_token_accuracy": 0.7909637689590454,
+      "num_tokens": 2291859.0,
+      "step": 144
+    },
+    {
+      "entropy": 0.8579956740140915,
+      "epoch": 1.3696682464454977,
+      "grad_norm": 0.6887325048446655,
+      "learning_rate": 1.5788988997959115e-05,
+      "loss": 0.8614661693572998,
+      "mean_token_accuracy": 0.7857953310012817,
+      "num_tokens": 2309035.0,
+      "step": 145
+    },
+    {
+      "entropy": 0.8863996863365173,
+      "epoch": 1.3791469194312795,
+      "grad_norm": 0.8338068127632141,
+      "learning_rate": 1.572509017272072e-05,
+      "loss": 0.8865163326263428,
+      "mean_token_accuracy": 0.7824681550264359,
+      "num_tokens": 2323401.0,
+      "step": 146
+    },
+    {
+      "entropy": 0.8635494112968445,
+      "epoch": 1.3886255924170616,
+      "grad_norm": 0.7118057012557983,
+      "learning_rate": 1.5660841702129533e-05,
+      "loss": 0.8939542770385742,
+      "mean_token_accuracy": 0.7877447158098221,
+      "num_tokens": 2340007.0,
+      "step": 147
+    },
+    {
+      "entropy": 0.8171515315771103,
+      "epoch": 1.3981042654028437,
+      "grad_norm": 0.7389315962791443,
+      "learning_rate": 1.5596247509997843e-05,
+      "loss": 0.8200218677520752,
+      "mean_token_accuracy": 0.7987903952598572,
+      "num_tokens": 2355339.0,
+      "step": 148
+    },
+    {
+      "entropy": 0.8425612896680832,
+      "epoch": 1.4075829383886256,
+      "grad_norm": 0.7023189067840576,
+      "learning_rate": 1.5531311541251995e-05,
+      "loss": 0.8327531814575195,
+      "mean_token_accuracy": 0.7908835858106613,
+      "num_tokens": 2371667.0,
+      "step": 149
+    },
+    {
+      "entropy": 0.7493451237678528,
+      "epoch": 1.4170616113744074,
+      "grad_norm": 0.778081476688385,
+      "learning_rate": 1.5466037761691493e-05,
+      "loss": 0.7227718830108643,
+      "mean_token_accuracy": 0.8122826516628265,
+      "num_tokens": 2386902.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7571745365858078,
+      "epoch": 1.4265402843601895,
+      "grad_norm": 0.7302114963531494,
+      "learning_rate": 1.540043015774676e-05,
+      "loss": 0.7457150220870972,
+      "mean_token_accuracy": 0.8071363717317581,
+      "num_tokens": 2403068.0,
+      "step": 151
+    },
+    {
+      "entropy": 0.8944563120603561,
+      "epoch": 1.4360189573459716,
+      "grad_norm": 0.7927082180976868,
+      "learning_rate": 1.5334492736235703e-05,
+      "loss": 0.9174668192863464,
+      "mean_token_accuracy": 0.7727936506271362,
+      "num_tokens": 2418122.0,
+      "step": 152
+    },
+    {
+      "entropy": 0.8746097236871719,
+      "epoch": 1.4454976303317535,
+      "grad_norm": 0.673865556716919,
+      "learning_rate": 1.5268229524119007e-05,
+      "loss": 0.8767449855804443,
+      "mean_token_accuracy": 0.7917247861623764,
+      "num_tokens": 2434701.0,
+      "step": 153
+    },
+    {
+      "entropy": 0.8153167068958282,
+      "epoch": 1.4549763033175356,
+      "grad_norm": 0.7478446364402771,
+      "learning_rate": 1.5201644568254181e-05,
+      "loss": 0.79709392786026,
+      "mean_token_accuracy": 0.797386422753334,
+      "num_tokens": 2451672.0,
+      "step": 154
+    },
+    {
+      "entropy": 0.7992338985204697,
+      "epoch": 1.4644549763033177,
+      "grad_norm": 0.8660085201263428,
+      "learning_rate": 1.513474193514842e-05,
+      "loss": 0.7673428058624268,
+      "mean_token_accuracy": 0.8041483759880066,
+      "num_tokens": 2466644.0,
+      "step": 155
+    },
+    {
+      "entropy": 0.8519728779792786,
+      "epoch": 1.4739336492890995,
+      "grad_norm": 0.8062114119529724,
+      "learning_rate": 1.5067525710710253e-05,
+      "loss": 0.8258011341094971,
+      "mean_token_accuracy": 0.7964712530374527,
+      "num_tokens": 2481743.0,
+      "step": 156
+    },
+    {
+      "entropy": 0.7952524870634079,
+      "epoch": 1.4834123222748814,
+      "grad_norm": 0.7020323276519775,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7887269258499146,
+      "mean_token_accuracy": 0.8075706511735916,
+      "num_tokens": 2498442.0,
+      "step": 157
+    },
+    {
+      "entropy": 0.7479639947414398,
+      "epoch": 1.4928909952606635,
+      "grad_norm": 0.7170225977897644,
+      "learning_rate": 1.4932168926979074e-05,
+      "loss": 0.7105016708374023,
+      "mean_token_accuracy": 0.8205106854438782,
+      "num_tokens": 2514651.0,
+      "step": 158
+    },
+    {
+      "entropy": 0.8277395963668823,
+      "epoch": 1.5023696682464456,
+      "grad_norm": 0.7073144316673279,
+      "learning_rate": 1.4864036634258112e-05,
+      "loss": 0.8328120112419128,
+      "mean_token_accuracy": 0.7966006547212601,
+      "num_tokens": 2532309.0,
+      "step": 159
+    },
+    {
+      "entropy": 0.803072065114975,
+      "epoch": 1.5118483412322274,
+      "grad_norm": 0.7687756419181824,
+      "learning_rate": 1.479560728284398e-05,
+      "loss": 0.8329487442970276,
+      "mean_token_accuracy": 0.7969093024730682,
+      "num_tokens": 2548833.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.8335355520248413,
+      "epoch": 1.5213270142180095,
+      "grad_norm": 0.7092747092247009,
+      "learning_rate": 1.4726885051885654e-05,
+      "loss": 0.8255264759063721,
+      "mean_token_accuracy": 0.7908845394849777,
+      "num_tokens": 2563652.0,
+      "step": 161
+    },
+    {
+      "entropy": 0.7855267077684402,
+      "epoch": 1.5308056872037916,
+      "grad_norm": 0.8094468712806702,
+      "learning_rate": 1.465787413841898e-05,
+      "loss": 0.7922707200050354,
+      "mean_token_accuracy": 0.7990368455648422,
+      "num_tokens": 2579535.0,
+      "step": 162
+    },
+    {
+      "entropy": 0.815998300909996,
+      "epoch": 1.5402843601895735,
+      "grad_norm": 0.6849187612533569,
+      "learning_rate": 1.4588578757110359e-05,
+      "loss": 0.8124603629112244,
+      "mean_token_accuracy": 0.7991871684789658,
+      "num_tokens": 2596641.0,
+      "step": 163
+    },
+    {
+      "entropy": 0.8301991075277328,
+      "epoch": 1.5497630331753554,
+      "grad_norm": 0.7943881750106812,
+      "learning_rate": 1.451900313999934e-05,
+      "loss": 0.8316096067428589,
+      "mean_token_accuracy": 0.7986911088228226,
+      "num_tokens": 2611723.0,
+      "step": 164
+    },
+    {
+      "entropy": 0.8705786466598511,
+      "epoch": 1.5592417061611374,
+      "grad_norm": 0.8372196555137634,
+      "learning_rate": 1.4449151536240167e-05,
+      "loss": 0.8712157607078552,
+      "mean_token_accuracy": 0.7852582186460495,
+      "num_tokens": 2627866.0,
+      "step": 165
+    },
+    {
+      "entropy": 0.7600953280925751,
+      "epoch": 1.5687203791469195,
+      "grad_norm": 0.673876166343689,
+      "learning_rate": 1.4379028211842265e-05,
+      "loss": 0.7214274406433105,
+      "mean_token_accuracy": 0.8191453367471695,
+      "num_tokens": 2644305.0,
+      "step": 166
+    },
+    {
+      "entropy": 0.9016031622886658,
+      "epoch": 1.5781990521327014,
+      "grad_norm": 0.7573343515396118,
+      "learning_rate": 1.4308637449409705e-05,
+      "loss": 0.8627768158912659,
+      "mean_token_accuracy": 0.7764580994844437,
+      "num_tokens": 2659693.0,
+      "step": 167
+    },
+    {
+      "entropy": 0.7766188532114029,
+      "epoch": 1.5876777251184833,
+      "grad_norm": 0.6323472857475281,
+      "learning_rate": 1.4237983547879664e-05,
+      "loss": 0.7716467380523682,
+      "mean_token_accuracy": 0.8073431253433228,
+      "num_tokens": 2676499.0,
+      "step": 168
+    },
+    {
+      "entropy": 0.7772778868675232,
+      "epoch": 1.5971563981042654,
+      "grad_norm": 0.8210384845733643,
+      "learning_rate": 1.4167070822259868e-05,
+      "loss": 0.7849143147468567,
+      "mean_token_accuracy": 0.8017231673002243,
+      "num_tokens": 2692356.0,
+      "step": 169
+    },
+    {
+      "entropy": 0.8266542851924896,
+      "epoch": 1.6066350710900474,
+      "grad_norm": 0.8243103623390198,
+      "learning_rate": 1.4095903603365067e-05,
+      "loss": 0.8927056789398193,
+      "mean_token_accuracy": 0.7911294400691986,
+      "num_tokens": 2707486.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.8491052091121674,
+      "epoch": 1.6161137440758293,
+      "grad_norm": 0.8660029768943787,
+      "learning_rate": 1.402448623755254e-05,
+      "loss": 0.9026396870613098,
+      "mean_token_accuracy": 0.7884256094694138,
+      "num_tokens": 2722952.0,
+      "step": 171
+    },
+    {
+      "entropy": 0.7935148179531097,
+      "epoch": 1.6255924170616114,
+      "grad_norm": 0.7412772178649902,
+      "learning_rate": 1.3952823086456656e-05,
+      "loss": 0.8272484540939331,
+      "mean_token_accuracy": 0.8038238883018494,
+      "num_tokens": 2737748.0,
+      "step": 172
+    },
+    {
+      "entropy": 0.7457627803087234,
+      "epoch": 1.6350710900473935,
+      "grad_norm": 0.6538417935371399,
+      "learning_rate": 1.3880918526722497e-05,
+      "loss": 0.7617708444595337,
+      "mean_token_accuracy": 0.8141642510890961,
+      "num_tokens": 2754877.0,
+      "step": 173
+    },
+    {
+      "entropy": 0.8603504151105881,
+      "epoch": 1.6445497630331753,
+      "grad_norm": 0.7187417149543762,
+      "learning_rate": 1.3808776949738569e-05,
+      "loss": 0.8643186688423157,
+      "mean_token_accuracy": 0.7814531624317169,
+      "num_tokens": 2770322.0,
+      "step": 174
+    },
+    {
+      "entropy": 0.8017472177743912,
+      "epoch": 1.6540284360189572,
+      "grad_norm": 0.7812091708183289,
+      "learning_rate": 1.3736402761368597e-05,
+      "loss": 0.7945480346679688,
+      "mean_token_accuracy": 0.7994107306003571,
+      "num_tokens": 2785842.0,
+      "step": 175
+    },
+    {
+      "entropy": 0.839350700378418,
+      "epoch": 1.6635071090047393,
+      "grad_norm": 0.6516030430793762,
+      "learning_rate": 1.3663800381682465e-05,
+      "loss": 0.8218634724617004,
+      "mean_token_accuracy": 0.7927983403205872,
+      "num_tokens": 2802150.0,
+      "step": 176
+    },
+    {
+      "entropy": 0.8673035353422165,
+      "epoch": 1.6729857819905214,
+      "grad_norm": 0.7417457699775696,
+      "learning_rate": 1.3590974244686248e-05,
+      "loss": 0.8643198013305664,
+      "mean_token_accuracy": 0.7817965149879456,
+      "num_tokens": 2818036.0,
+      "step": 177
+    },
+    {
+      "entropy": 0.8689644038677216,
+      "epoch": 1.6824644549763033,
+      "grad_norm": 0.7800316214561462,
+      "learning_rate": 1.3517928798051442e-05,
+      "loss": 0.8703083992004395,
+      "mean_token_accuracy": 0.784311443567276,
+      "num_tokens": 2833551.0,
+      "step": 178
+    },
+    {
+      "entropy": 0.8547259271144867,
+      "epoch": 1.6919431279620853,
+      "grad_norm": 0.7157973647117615,
+      "learning_rate": 1.344466850284333e-05,
+      "loss": 0.8583805561065674,
+      "mean_token_accuracy": 0.7822477072477341,
+      "num_tokens": 2849148.0,
+      "step": 179
+    },
+    {
+      "entropy": 0.8259568810462952,
+      "epoch": 1.7014218009478674,
+      "grad_norm": 0.682313084602356,
+      "learning_rate": 1.3371197833248508e-05,
+      "loss": 0.8174458742141724,
+      "mean_token_accuracy": 0.7968255281448364,
+      "num_tokens": 2865308.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7793432623147964,
+      "epoch": 1.7109004739336493,
+      "grad_norm": 0.6921513080596924,
+      "learning_rate": 1.3297521276301666e-05,
+      "loss": 0.7710314393043518,
+      "mean_token_accuracy": 0.807743102312088,
+      "num_tokens": 2880734.0,
+      "step": 181
+    },
+    {
+      "entropy": 0.8243378251791,
+      "epoch": 1.7203791469194312,
+      "grad_norm": 0.8263442516326904,
+      "learning_rate": 1.3223643331611538e-05,
+      "loss": 0.8093314170837402,
+      "mean_token_accuracy": 0.8022452294826508,
+      "num_tokens": 2896023.0,
+      "step": 182
+    },
+    {
+      "entropy": 0.8052062392234802,
+      "epoch": 1.7298578199052133,
+      "grad_norm": 0.7710005640983582,
+      "learning_rate": 1.3149568511086104e-05,
+      "loss": 0.786173403263092,
+      "mean_token_accuracy": 0.8085711002349854,
+      "num_tokens": 2911649.0,
+      "step": 183
+    },
+    {
+      "entropy": 0.8499461859464645,
+      "epoch": 1.7393364928909953,
+      "grad_norm": 0.7619282603263855,
+      "learning_rate": 1.3075301338657036e-05,
+      "loss": 0.803632378578186,
+      "mean_token_accuracy": 0.7947988212108612,
+      "num_tokens": 2926955.0,
+      "step": 184
+    },
+    {
+      "entropy": 0.8158384412527084,
+      "epoch": 1.7488151658767772,
+      "grad_norm": 0.7673326134681702,
+      "learning_rate": 1.300084635000341e-05,
+      "loss": 0.7742278575897217,
+      "mean_token_accuracy": 0.8106317967176437,
+      "num_tokens": 2941760.0,
+      "step": 185
+    },
+    {
+      "entropy": 0.9341417849063873,
+      "epoch": 1.758293838862559,
+      "grad_norm": 0.7602595686912537,
+      "learning_rate": 1.2926208092274699e-05,
+      "loss": 0.9146112203598022,
+      "mean_token_accuracy": 0.7746770083904266,
+      "num_tokens": 2957629.0,
+      "step": 186
+    },
+    {
+      "entropy": 0.8594657629728317,
+      "epoch": 1.7677725118483414,
+      "grad_norm": 0.6417691111564636,
+      "learning_rate": 1.2851391123813075e-05,
+      "loss": 0.8513129949569702,
+      "mean_token_accuracy": 0.7823676913976669,
+      "num_tokens": 2974081.0,
+      "step": 187
+    },
+    {
+      "entropy": 0.7850838005542755,
+      "epoch": 1.7772511848341233,
+      "grad_norm": 0.7059190273284912,
+      "learning_rate": 1.2776400013875006e-05,
+      "loss": 0.7892745137214661,
+      "mean_token_accuracy": 0.8025059252977371,
+      "num_tokens": 2989825.0,
+      "step": 188
+    },
+    {
+      "entropy": 0.8054764866828918,
+      "epoch": 1.7867298578199051,
+      "grad_norm": 0.7688627243041992,
+      "learning_rate": 1.2701239342352223e-05,
+      "loss": 0.8145111799240112,
+      "mean_token_accuracy": 0.7917294800281525,
+      "num_tokens": 3005441.0,
+      "step": 189
+    },
+    {
+      "entropy": 0.7718167155981064,
+      "epoch": 1.7962085308056872,
+      "grad_norm": 0.7204188704490662,
+      "learning_rate": 1.2625913699491986e-05,
+      "loss": 0.7447738647460938,
+      "mean_token_accuracy": 0.8080335557460785,
+      "num_tokens": 3020864.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.769375741481781,
+      "epoch": 1.8056872037914693,
+      "grad_norm": 0.8016238212585449,
+      "learning_rate": 1.2550427685616767e-05,
+      "loss": 0.7878785133361816,
+      "mean_token_accuracy": 0.8049481511116028,
+      "num_tokens": 3036515.0,
+      "step": 191
+    },
+    {
+      "entropy": 0.7770668566226959,
+      "epoch": 1.8151658767772512,
+      "grad_norm": 0.7289695739746094,
+      "learning_rate": 1.2474785910843289e-05,
+      "loss": 0.7667026519775391,
+      "mean_token_accuracy": 0.808977484703064,
+      "num_tokens": 3052474.0,
+      "step": 192
+    },
+    {
+      "entropy": 0.7413365840911865,
+      "epoch": 1.824644549763033,
+      "grad_norm": 0.6931266784667969,
+      "learning_rate": 1.239899299480098e-05,
+      "loss": 0.7244135737419128,
+      "mean_token_accuracy": 0.8193869441747665,
+      "num_tokens": 3068429.0,
+      "step": 193
+    },
+    {
+      "entropy": 0.789655476808548,
+      "epoch": 1.8341232227488151,
+      "grad_norm": 0.6820263862609863,
+      "learning_rate": 1.2323053566349834e-05,
+      "loss": 0.772818386554718,
+      "mean_token_accuracy": 0.8061677515506744,
+      "num_tokens": 3084782.0,
+      "step": 194
+    },
+    {
+      "entropy": 0.8178127855062485,
+      "epoch": 1.8436018957345972,
+      "grad_norm": 0.6741772890090942,
+      "learning_rate": 1.2246972263297718e-05,
+      "loss": 0.8100478053092957,
+      "mean_token_accuracy": 0.789124608039856,
+      "num_tokens": 3101303.0,
+      "step": 195
+    },
+    {
+      "entropy": 0.7827744632959366,
+      "epoch": 1.853080568720379,
+      "grad_norm": 0.7667286992073059,
+      "learning_rate": 1.2170753732117138e-05,
+      "loss": 0.7773491144180298,
+      "mean_token_accuracy": 0.8108680844306946,
+      "num_tokens": 3116869.0,
+      "step": 196
+    },
+    {
+      "entropy": 0.7749081552028656,
+      "epoch": 1.8625592417061612,
+      "grad_norm": 0.663255512714386,
+      "learning_rate": 1.2094402627661447e-05,
+      "loss": 0.7651374340057373,
+      "mean_token_accuracy": 0.8063687533140182,
+      "num_tokens": 3133160.0,
+      "step": 197
+    },
+    {
+      "entropy": 0.7427408695220947,
+      "epoch": 1.8720379146919433,
+      "grad_norm": 0.8035187721252441,
+      "learning_rate": 1.2017923612880579e-05,
+      "loss": 0.7308788299560547,
+      "mean_token_accuracy": 0.8139677345752716,
+      "num_tokens": 3148591.0,
+      "step": 198
+    },
+    {
+      "entropy": 0.7870882898569107,
+      "epoch": 1.8815165876777251,
+      "grad_norm": 0.7356343269348145,
+      "learning_rate": 1.1941321358536278e-05,
+      "loss": 0.7814972400665283,
+      "mean_token_accuracy": 0.8018877655267715,
+      "num_tokens": 3164873.0,
+      "step": 199
+    },
+    {
+      "entropy": 0.7867356240749359,
+      "epoch": 1.890995260663507,
+      "grad_norm": 0.7502454519271851,
+      "learning_rate": 1.1864600542916813e-05,
+      "loss": 0.7805415391921997,
+      "mean_token_accuracy": 0.8045219480991364,
+      "num_tokens": 3180052.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.8339502215385437,
+      "epoch": 1.900473933649289,
+      "grad_norm": 0.7521458268165588,
+      "learning_rate": 1.1787765851551296e-05,
+      "loss": 0.8415899276733398,
+      "mean_token_accuracy": 0.7910822480916977,
+      "num_tokens": 3196566.0,
+      "step": 201
+    },
+    {
+      "entropy": 0.8327292948961258,
+      "epoch": 1.9099526066350712,
+      "grad_norm": 0.7168639898300171,
+      "learning_rate": 1.17108219769235e-05,
+      "loss": 0.8684878349304199,
+      "mean_token_accuracy": 0.7880802601575851,
+      "num_tokens": 3212078.0,
+      "step": 202
+    },
+    {
+      "entropy": 0.8257894515991211,
+      "epoch": 1.919431279620853,
+      "grad_norm": 0.7584367394447327,
+      "learning_rate": 1.1633773618185302e-05,
+      "loss": 0.8573964834213257,
+      "mean_token_accuracy": 0.7954303473234177,
+      "num_tokens": 3227754.0,
+      "step": 203
+    },
+    {
+      "entropy": 0.8100651800632477,
+      "epoch": 1.9289099526066351,
+      "grad_norm": 0.7311358451843262,
+      "learning_rate": 1.155662548086967e-05,
+      "loss": 0.7980862259864807,
+      "mean_token_accuracy": 0.8024295270442963,
+      "num_tokens": 3244342.0,
+      "step": 204
+    },
+    {
+      "entropy": 0.877414807677269,
+      "epoch": 1.9383886255924172,
+      "grad_norm": 0.8059820532798767,
+      "learning_rate": 1.14793822766033e-05,
+      "loss": 0.8878626823425293,
+      "mean_token_accuracy": 0.7836036533117294,
+      "num_tokens": 3260849.0,
+      "step": 205
+    },
+    {
+      "entropy": 0.8052125722169876,
+      "epoch": 1.947867298578199,
+      "grad_norm": 0.7401100397109985,
+      "learning_rate": 1.1402048722818862e-05,
+      "loss": 0.780508279800415,
+      "mean_token_accuracy": 0.8014171868562698,
+      "num_tokens": 3277712.0,
+      "step": 206
+    },
+    {
+      "entropy": 0.8691103160381317,
+      "epoch": 1.957345971563981,
+      "grad_norm": 0.717732310295105,
+      "learning_rate": 1.132462954246688e-05,
+      "loss": 0.8721846342086792,
+      "mean_token_accuracy": 0.781763345003128,
+      "num_tokens": 3293613.0,
+      "step": 207
+    },
+    {
+      "entropy": 0.8300742954015732,
+      "epoch": 1.966824644549763,
+      "grad_norm": 0.697457492351532,
+      "learning_rate": 1.1247129463727329e-05,
+      "loss": 0.8578647375106812,
+      "mean_token_accuracy": 0.7887941151857376,
+      "num_tokens": 3309463.0,
+      "step": 208
+    },
+    {
+      "entropy": 0.898152619600296,
+      "epoch": 1.9763033175355451,
+      "grad_norm": 0.6961869597434998,
+      "learning_rate": 1.1169553219720828e-05,
+      "loss": 0.9294345378875732,
+      "mean_token_accuracy": 0.7751467376947403,
+      "num_tokens": 3326245.0,
+      "step": 209
+    },
+    {
+      "entropy": 0.7650076150894165,
+      "epoch": 1.985781990521327,
+      "grad_norm": 0.7355565428733826,
+      "learning_rate": 1.1091905548219597e-05,
+      "loss": 0.7482494115829468,
+      "mean_token_accuracy": 0.8114674985408783,
+      "num_tokens": 3342296.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.8681088984012604,
+      "epoch": 1.9952606635071088,
+      "grad_norm": 0.6917135715484619,
+      "learning_rate": 1.1014191191358118e-05,
+      "loss": 0.850773811340332,
+      "mean_token_accuracy": 0.7946626245975494,
+      "num_tokens": 3357305.0,
+      "step": 211
+    },
+    {
+      "entropy": 0.7847772836685181,
+      "epoch": 2.0,
+      "grad_norm": 1.2537461519241333,
+      "learning_rate": 1.093641489534351e-05,
+      "loss": 0.8384180665016174,
+      "mean_token_accuracy": 0.8076447546482086,
+      "num_tokens": 3364885.0,
+      "step": 212
+    },
+    {
+      "entropy": 0.8841996192932129,
+      "epoch": 2.009478672985782,
+      "grad_norm": 0.8463948965072632,
+      "learning_rate": 1.085858141016566e-05,
+      "loss": 0.8430708646774292,
+      "mean_token_accuracy": 0.7934243977069855,
+      "num_tokens": 3379936.0,
+      "step": 213
+    },
+    {
+      "entropy": 0.8387620151042938,
+      "epoch": 2.018957345971564,
+      "grad_norm": 0.654754638671875,
+      "learning_rate": 1.0780695489307152e-05,
+      "loss": 0.8108071088790894,
+      "mean_token_accuracy": 0.7976965010166168,
+      "num_tokens": 3395599.0,
+      "step": 214
+    },
+    {
+      "entropy": 0.8131097704172134,
+      "epoch": 2.028436018957346,
+      "grad_norm": 0.7293988466262817,
+      "learning_rate": 1.070276188945293e-05,
+      "loss": 0.808356761932373,
+      "mean_token_accuracy": 0.7960507720708847,
+      "num_tokens": 3410351.0,
+      "step": 215
+    },
+    {
+      "entropy": 0.8139905482530594,
+      "epoch": 2.037914691943128,
+      "grad_norm": 0.8644604682922363,
+      "learning_rate": 1.062478537019983e-05,
+      "loss": 0.8001759052276611,
+      "mean_token_accuracy": 0.8025805652141571,
+      "num_tokens": 3425898.0,
+      "step": 216
+    },
+    {
+      "entropy": 0.7704232633113861,
+      "epoch": 2.0473933649289098,
+      "grad_norm": 0.6959671974182129,
+      "learning_rate": 1.0546770693765859e-05,
+      "loss": 0.7146846055984497,
+      "mean_token_accuracy": 0.820914089679718,
+      "num_tokens": 3441747.0,
+      "step": 217
+    },
+    {
+      "entropy": 0.774506226181984,
+      "epoch": 2.056872037914692,
+      "grad_norm": 0.7746772170066833,
+      "learning_rate": 1.0468722624699401e-05,
+      "loss": 0.7572252750396729,
+      "mean_token_accuracy": 0.8087823241949081,
+      "num_tokens": 3457430.0,
+      "step": 218
+    },
+    {
+      "entropy": 0.75509412586689,
+      "epoch": 2.066350710900474,
+      "grad_norm": 0.6639065742492676,
+      "learning_rate": 1.0390645929588197e-05,
+      "loss": 0.7484322786331177,
+      "mean_token_accuracy": 0.8096154481172562,
+      "num_tokens": 3473449.0,
+      "step": 219
+    },
+    {
+      "entropy": 0.8533463031053543,
+      "epoch": 2.075829383886256,
+      "grad_norm": 1.005789875984192,
+      "learning_rate": 1.0312545376768246e-05,
+      "loss": 0.8439828157424927,
+      "mean_token_accuracy": 0.785793200135231,
+      "num_tokens": 3488206.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.8072682321071625,
+      "epoch": 2.085308056872038,
+      "grad_norm": 0.6852388978004456,
+      "learning_rate": 1.0234425736032607e-05,
+      "loss": 0.8122374415397644,
+      "mean_token_accuracy": 0.8027433753013611,
+      "num_tokens": 3504731.0,
+      "step": 221
+    },
+    {
+      "entropy": 0.7405507862567902,
+      "epoch": 2.09478672985782,
+      "grad_norm": 0.6730477213859558,
+      "learning_rate": 1.015629177834008e-05,
+      "loss": 0.6996761560440063,
+      "mean_token_accuracy": 0.8150836676359177,
+      "num_tokens": 3520301.0,
+      "step": 222
+    },
+    {
+      "entropy": 0.8268642425537109,
+      "epoch": 2.104265402843602,
+      "grad_norm": 0.8005897402763367,
+      "learning_rate": 1.007814827552384e-05,
+      "loss": 0.8099018335342407,
+      "mean_token_accuracy": 0.7979452759027481,
+      "num_tokens": 3536522.0,
+      "step": 223
+    },
+    {
+      "entropy": 0.8399747163057327,
+      "epoch": 2.1137440758293837,
+      "grad_norm": 1.2597578763961792,
+      "learning_rate": 1e-05,
+      "loss": 0.8286103010177612,
+      "mean_token_accuracy": 0.7938411682844162,
+      "num_tokens": 3551654.0,
+      "step": 224
+    },
+    {
+      "entropy": 0.761796772480011,
+      "epoch": 2.123222748815166,
+      "grad_norm": 0.689793586730957,
+      "learning_rate": 9.92185172447616e-06,
+      "loss": 0.771800696849823,
+      "mean_token_accuracy": 0.8083701580762863,
+      "num_tokens": 3568275.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.8368710279464722,
+      "epoch": 2.132701421800948,
+      "grad_norm": 0.7853175401687622,
+      "learning_rate": 9.843708221659924e-06,
+      "loss": 0.8215783834457397,
+      "mean_token_accuracy": 0.7981610000133514,
+      "num_tokens": 3583250.0,
+      "step": 226
+    },
+    {
+      "entropy": 0.7556120902299881,
+      "epoch": 2.1421800947867298,
+      "grad_norm": 0.7487472295761108,
+      "learning_rate": 9.765574263967397e-06,
+      "loss": 0.7591477036476135,
+      "mean_token_accuracy": 0.8063594847917557,
+      "num_tokens": 3599684.0,
+      "step": 227
+    },
+    {
+      "entropy": 0.7687488049268723,
+      "epoch": 2.1516587677725116,
+      "grad_norm": 0.7922394275665283,
+      "learning_rate": 9.68745462323176e-06,
+      "loss": 0.7621670365333557,
+      "mean_token_accuracy": 0.8083523958921432,
+      "num_tokens": 3615776.0,
+      "step": 228
+    },
+    {
+      "entropy": 0.7205257564783096,
+      "epoch": 2.161137440758294,
+      "grad_norm": 0.8748376369476318,
+      "learning_rate": 9.609354070411807e-06,
+      "loss": 0.7108169794082642,
+      "mean_token_accuracy": 0.8180408775806427,
+      "num_tokens": 3631914.0,
+      "step": 229
+    },
+    {
+      "entropy": 0.7599902749061584,
+      "epoch": 2.170616113744076,
+      "grad_norm": 0.6921991109848022,
+      "learning_rate": 9.531277375300599e-06,
+      "loss": 0.7450228929519653,
+      "mean_token_accuracy": 0.815703496336937,
+      "num_tokens": 3648945.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.7611272037029266,
+      "epoch": 2.1800947867298577,
+      "grad_norm": 0.6716788411140442,
+      "learning_rate": 9.453229306234143e-06,
+      "loss": 0.769507646560669,
+      "mean_token_accuracy": 0.8149918764829636,
+      "num_tokens": 3664617.0,
+      "step": 231
+    },
+    {
+      "entropy": 0.8009829074144363,
+      "epoch": 2.18957345971564,
+      "grad_norm": 0.7548602223396301,
+      "learning_rate": 9.375214629800173e-06,
+      "loss": 0.796088695526123,
+      "mean_token_accuracy": 0.7964328676462173,
+      "num_tokens": 3679523.0,
+      "step": 232
+    },
+    {
+      "entropy": 0.7614959329366684,
+      "epoch": 2.199052132701422,
+      "grad_norm": 0.7149446606636047,
+      "learning_rate": 9.297238110547075e-06,
+      "loss": 0.7764595746994019,
+      "mean_token_accuracy": 0.7976703941822052,
+      "num_tokens": 3696512.0,
+      "step": 233
+    },
+    {
+      "entropy": 0.8016138821840286,
+      "epoch": 2.2085308056872037,
+      "grad_norm": 0.8237394690513611,
+      "learning_rate": 9.219304510692853e-06,
+      "loss": 0.7986416220664978,
+      "mean_token_accuracy": 0.7949725389480591,
+      "num_tokens": 3713141.0,
+      "step": 234
+    },
+    {
+      "entropy": 0.7871145308017731,
+      "epoch": 2.2180094786729856,
+      "grad_norm": 0.7732126116752625,
+      "learning_rate": 9.14141858983434e-06,
+      "loss": 0.8035563230514526,
+      "mean_token_accuracy": 0.8023014217615128,
+      "num_tokens": 3729153.0,
+      "step": 235
+    },
+    {
+      "entropy": 0.782372236251831,
+      "epoch": 2.227488151658768,
+      "grad_norm": 0.7351176738739014,
+      "learning_rate": 9.063585104656494e-06,
+      "loss": 0.765059769153595,
+      "mean_token_accuracy": 0.8090708404779434,
+      "num_tokens": 3744796.0,
+      "step": 236
+    },
+    {
+      "entropy": 0.8335847705602646,
+      "epoch": 2.2369668246445498,
+      "grad_norm": 0.6978676319122314,
+      "learning_rate": 8.985808808641883e-06,
+      "loss": 0.8426876664161682,
+      "mean_token_accuracy": 0.790928304195404,
+      "num_tokens": 3761370.0,
+      "step": 237
+    },
+    {
+      "entropy": 0.8038548976182938,
+      "epoch": 2.2464454976303316,
+      "grad_norm": 0.7371619939804077,
+      "learning_rate": 8.908094451780408e-06,
+      "loss": 0.8280337452888489,
+      "mean_token_accuracy": 0.7949777245521545,
+      "num_tokens": 3777078.0,
+      "step": 238
+    },
+    {
+      "entropy": 0.7301145642995834,
+      "epoch": 2.2559241706161135,
+      "grad_norm": 0.7276929020881653,
+      "learning_rate": 8.830446780279175e-06,
+      "loss": 0.7212296724319458,
+      "mean_token_accuracy": 0.8165305256843567,
+      "num_tokens": 3793426.0,
+      "step": 239
+    },
+    {
+      "entropy": 0.7998540550470352,
+      "epoch": 2.265402843601896,
+      "grad_norm": 0.6656012535095215,
+      "learning_rate": 8.752870536272673e-06,
+      "loss": 0.7798458337783813,
+      "mean_token_accuracy": 0.7948831617832184,
+      "num_tokens": 3809150.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.8189331293106079,
+      "epoch": 2.2748815165876777,
+      "grad_norm": 0.7528814673423767,
+      "learning_rate": 8.675370457533122e-06,
+      "loss": 0.8173599243164062,
+      "mean_token_accuracy": 0.7954176068305969,
+      "num_tokens": 3824693.0,
+      "step": 241
+    },
+    {
+      "entropy": 0.7802784144878387,
+      "epoch": 2.2843601895734595,
+      "grad_norm": 0.8117206692695618,
+      "learning_rate": 8.597951277181143e-06,
+      "loss": 0.8089747428894043,
+      "mean_token_accuracy": 0.8035465627908707,
+      "num_tokens": 3840276.0,
+      "step": 242
+    },
+    {
+      "entropy": 0.8260502219200134,
+      "epoch": 2.293838862559242,
+      "grad_norm": 0.7881060242652893,
+      "learning_rate": 8.520617723396702e-06,
+      "loss": 0.8358763456344604,
+      "mean_token_accuracy": 0.7957090586423874,
+      "num_tokens": 3856007.0,
+      "step": 243
+    },
+    {
+      "entropy": 0.7820662707090378,
+      "epoch": 2.3033175355450237,
+      "grad_norm": 0.7249373197555542,
+      "learning_rate": 8.443374519130332e-06,
+      "loss": 0.7823531031608582,
+      "mean_token_accuracy": 0.8001191914081573,
+      "num_tokens": 3872300.0,
+      "step": 244
+    },
+    {
+      "entropy": 0.8399253487586975,
+      "epoch": 2.3127962085308056,
+      "grad_norm": 0.8324182629585266,
+      "learning_rate": 8.366226381814698e-06,
+      "loss": 0.8358606100082397,
+      "mean_token_accuracy": 0.7946343570947647,
+      "num_tokens": 3888174.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.7229929268360138,
+      "epoch": 2.322274881516588,
+      "grad_norm": 0.8482832312583923,
+      "learning_rate": 8.289178023076501e-06,
+      "loss": 0.6760922074317932,
+      "mean_token_accuracy": 0.8288819640874863,
+      "num_tokens": 3904294.0,
+      "step": 246
+    },
+    {
+      "entropy": 0.756013348698616,
+      "epoch": 2.3317535545023698,
+      "grad_norm": 0.7549688816070557,
+      "learning_rate": 8.212234148448708e-06,
+      "loss": 0.7417251467704773,
+      "mean_token_accuracy": 0.8116127997636795,
+      "num_tokens": 3921037.0,
+      "step": 247
+    },
+    {
+      "entropy": 0.8636997491121292,
+      "epoch": 2.3412322274881516,
+      "grad_norm": 0.7435200810432434,
+      "learning_rate": 8.13539945708319e-06,
+      "loss": 0.8481851816177368,
+      "mean_token_accuracy": 0.7855764031410217,
+      "num_tokens": 3936399.0,
+      "step": 248
+    },
+    {
+      "entropy": 0.7722165882587433,
+      "epoch": 2.3507109004739335,
+      "grad_norm": 0.82920241355896,
+      "learning_rate": 8.058678641463724e-06,
+      "loss": 0.7838011980056763,
+      "mean_token_accuracy": 0.7997405827045441,
+      "num_tokens": 3951517.0,
+      "step": 249
+    },
+    {
+      "entropy": 0.7858306765556335,
+      "epoch": 2.360189573459716,
+      "grad_norm": 0.8119862675666809,
+      "learning_rate": 7.98207638711942e-06,
+      "loss": 0.7850341796875,
+      "mean_token_accuracy": 0.8023760467767715,
+      "num_tokens": 3966658.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.713519737124443,
+      "epoch": 2.3696682464454977,
+      "grad_norm": 0.7162942886352539,
+      "learning_rate": 7.905597372338558e-06,
+      "loss": 0.6873078346252441,
+      "mean_token_accuracy": 0.8234204798936844,
+      "num_tokens": 3983390.0,
+      "step": 251
+    },
+    {
+      "entropy": 0.8907800167798996,
+      "epoch": 2.3791469194312795,
+      "grad_norm": 0.7015475034713745,
+      "learning_rate": 7.829246267882864e-06,
+      "loss": 0.9189035892486572,
+      "mean_token_accuracy": 0.7817337810993195,
+      "num_tokens": 3999794.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.797465056180954,
+      "epoch": 2.3886255924170614,
+      "grad_norm": 0.8406365513801575,
+      "learning_rate": 7.753027736702283e-06,
+      "loss": 0.7891885042190552,
+      "mean_token_accuracy": 0.8100517839193344,
+      "num_tokens": 4015172.0,
+      "step": 253
+    },
+    {
+      "entropy": 0.7660058289766312,
+      "epoch": 2.3981042654028437,
+      "grad_norm": 0.795070230960846,
+      "learning_rate": 7.67694643365017e-06,
+      "loss": 0.7752566933631897,
+      "mean_token_accuracy": 0.8050341606140137,
+      "num_tokens": 4031362.0,
+      "step": 254
+    },
+    {
+      "entropy": 0.750129446387291,
+      "epoch": 2.4075829383886256,
+      "grad_norm": 0.7366607189178467,
+      "learning_rate": 7.601007005199022e-06,
+      "loss": 0.7281040549278259,
+      "mean_token_accuracy": 0.8136068880558014,
+      "num_tokens": 4047478.0,
+      "step": 255
+    },
+    {
+      "entropy": 0.7423243522644043,
+      "epoch": 2.4170616113744074,
+      "grad_norm": 0.7090871930122375,
+      "learning_rate": 7.525214089156714e-06,
+      "loss": 0.7064968347549438,
+      "mean_token_accuracy": 0.824029728770256,
+      "num_tokens": 4062615.0,
+      "step": 256
+    },
+    {
+      "entropy": 0.8094542026519775,
+      "epoch": 2.4265402843601898,
+      "grad_norm": 0.7741425633430481,
+      "learning_rate": 7.449572314383237e-06,
+      "loss": 0.8382295370101929,
+      "mean_token_accuracy": 0.7949035614728928,
+      "num_tokens": 4079018.0,
+      "step": 257
+    },
+    {
+      "entropy": 0.762347549200058,
+      "epoch": 2.4360189573459716,
+      "grad_norm": 0.7608873844146729,
+      "learning_rate": 7.374086300508019e-06,
+      "loss": 0.747887134552002,
+      "mean_token_accuracy": 0.8116944581270218,
+      "num_tokens": 4094608.0,
+      "step": 258
+    },
+    {
+      "entropy": 0.8015813231468201,
+      "epoch": 2.4454976303317535,
+      "grad_norm": 0.8000101447105408,
+      "learning_rate": 7.298760657647779e-06,
+      "loss": 0.7769061923027039,
+      "mean_token_accuracy": 0.8080571591854095,
+      "num_tokens": 4110049.0,
+      "step": 259
+    },
+    {
+      "entropy": 0.7255821079015732,
+      "epoch": 2.4549763033175354,
+      "grad_norm": 0.7408568859100342,
+      "learning_rate": 7.223599986124994e-06,
+      "loss": 0.7272932529449463,
+      "mean_token_accuracy": 0.8140795826911926,
+      "num_tokens": 4126621.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.753772422671318,
+      "epoch": 2.4644549763033177,
+      "grad_norm": 0.7682618498802185,
+      "learning_rate": 7.148608876186931e-06,
+      "loss": 0.7212350368499756,
+      "mean_token_accuracy": 0.8186078071594238,
+      "num_tokens": 4142397.0,
+      "step": 261
+    },
+    {
+      "entropy": 0.855003148317337,
+      "epoch": 2.4739336492890995,
+      "grad_norm": 0.8246680498123169,
+      "learning_rate": 7.073791907725304e-06,
+      "loss": 0.861223578453064,
+      "mean_token_accuracy": 0.7906257957220078,
+      "num_tokens": 4157512.0,
+      "step": 262
+    },
+    {
+      "entropy": 0.7049210220575333,
+      "epoch": 2.4834123222748814,
+      "grad_norm": 0.7506635189056396,
+      "learning_rate": 6.999153649996595e-06,
+      "loss": 0.6862914562225342,
+      "mean_token_accuracy": 0.8244545608758926,
+      "num_tokens": 4173292.0,
+      "step": 263
+    },
+    {
+      "entropy": 0.7707203775644302,
+      "epoch": 2.4928909952606633,
+      "grad_norm": 0.7590407729148865,
+      "learning_rate": 6.924698661342968e-06,
+      "loss": 0.7646045684814453,
+      "mean_token_accuracy": 0.8014439344406128,
+      "num_tokens": 4189278.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.7228768467903137,
+      "epoch": 2.5023696682464456,
+      "grad_norm": 0.7332707047462463,
+      "learning_rate": 6.8504314889138956e-06,
+      "loss": 0.7292245626449585,
+      "mean_token_accuracy": 0.817297637462616,
+      "num_tokens": 4205312.0,
+      "step": 265
+    },
+    {
+      "entropy": 0.762321949005127,
+      "epoch": 2.5118483412322274,
+      "grad_norm": 0.763211190700531,
+      "learning_rate": 6.776356668388464e-06,
+      "loss": 0.7480930685997009,
+      "mean_token_accuracy": 0.8084921091794968,
+      "num_tokens": 4222070.0,
+      "step": 266
+    },
+    {
+      "entropy": 0.7291257232427597,
+      "epoch": 2.5213270142180093,
+      "grad_norm": 0.7732436060905457,
+      "learning_rate": 6.702478723698336e-06,
+      "loss": 0.7159913182258606,
+      "mean_token_accuracy": 0.8156870156526566,
+      "num_tokens": 4238086.0,
+      "step": 267
+    },
+    {
+      "entropy": 0.7663960009813309,
+      "epoch": 2.5308056872037916,
+      "grad_norm": 0.7954282760620117,
+      "learning_rate": 6.628802166751496e-06,
+      "loss": 0.7697724103927612,
+      "mean_token_accuracy": 0.8044588267803192,
+      "num_tokens": 4253062.0,
+      "step": 268
+    },
+    {
+      "entropy": 0.7694273293018341,
+      "epoch": 2.5402843601895735,
+      "grad_norm": 0.8077139854431152,
+      "learning_rate": 6.555331497156671e-06,
+      "loss": 0.7804722785949707,
+      "mean_token_accuracy": 0.8017380684614182,
+      "num_tokens": 4268074.0,
+      "step": 269
+    },
+    {
+      "entropy": 0.7905916273593903,
+      "epoch": 2.5497630331753554,
+      "grad_norm": 0.7760480642318726,
+      "learning_rate": 6.482071201948557e-06,
+      "loss": 0.8075841665267944,
+      "mean_token_accuracy": 0.7980497181415558,
+      "num_tokens": 4284203.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.797221377491951,
+      "epoch": 2.5592417061611377,
+      "grad_norm": 0.7789294123649597,
+      "learning_rate": 6.4090257553137566e-06,
+      "loss": 0.8040370345115662,
+      "mean_token_accuracy": 0.8009522259235382,
+      "num_tokens": 4298529.0,
+      "step": 271
+    },
+    {
+      "entropy": 0.7779994606971741,
+      "epoch": 2.5687203791469195,
+      "grad_norm": 0.7936727404594421,
+      "learning_rate": 6.336199618317538e-06,
+      "loss": 0.7707229256629944,
+      "mean_token_accuracy": 0.813504621386528,
+      "num_tokens": 4314557.0,
+      "step": 272
+    },
+    {
+      "entropy": 0.8157918900251389,
+      "epoch": 2.5781990521327014,
+      "grad_norm": 0.7624719142913818,
+      "learning_rate": 6.263597238631405e-06,
+      "loss": 0.8281031847000122,
+      "mean_token_accuracy": 0.7870993167161942,
+      "num_tokens": 4330159.0,
+      "step": 273
+    },
+    {
+      "entropy": 0.8026942908763885,
+      "epoch": 2.5876777251184833,
+      "grad_norm": 0.7207754254341125,
+      "learning_rate": 6.191223050261432e-06,
+      "loss": 0.7978195548057556,
+      "mean_token_accuracy": 0.7940388768911362,
+      "num_tokens": 4345645.0,
+      "step": 274
+    },
+    {
+      "entropy": 0.8023267984390259,
+      "epoch": 2.597156398104265,
+      "grad_norm": 0.764500081539154,
+      "learning_rate": 6.119081473277502e-06,
+      "loss": 0.7886009216308594,
+      "mean_token_accuracy": 0.8018842190504074,
+      "num_tokens": 4361883.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.7076465785503387,
+      "epoch": 2.6066350710900474,
+      "grad_norm": 0.8487837910652161,
+      "learning_rate": 6.047176913543348e-06,
+      "loss": 0.7041523456573486,
+      "mean_token_accuracy": 0.8193930238485336,
+      "num_tokens": 4377678.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.746178925037384,
+      "epoch": 2.6161137440758293,
+      "grad_norm": 0.7743781805038452,
+      "learning_rate": 5.975513762447465e-06,
+      "loss": 0.7622804641723633,
+      "mean_token_accuracy": 0.8078963905572891,
+      "num_tokens": 4393473.0,
+      "step": 277
+    },
+    {
+      "entropy": 0.7890691012144089,
+      "epoch": 2.625592417061611,
+      "grad_norm": 0.6980987191200256,
+      "learning_rate": 5.904096396634935e-06,
+      "loss": 0.7980008125305176,
+      "mean_token_accuracy": 0.8008007407188416,
+      "num_tokens": 4410182.0,
+      "step": 278
+    },
+    {
+      "entropy": 0.6923302114009857,
+      "epoch": 2.6350710900473935,
+      "grad_norm": 0.6928261518478394,
+      "learning_rate": 5.832929177740134e-06,
+      "loss": 0.7043225765228271,
+      "mean_token_accuracy": 0.827838584780693,
+      "num_tokens": 4427098.0,
+      "step": 279
+    },
+    {
+      "entropy": 0.7364738285541534,
+      "epoch": 2.6445497630331753,
+      "grad_norm": 0.742316722869873,
+      "learning_rate": 5.762016452120336e-06,
+      "loss": 0.7640200853347778,
+      "mean_token_accuracy": 0.8073264956474304,
+      "num_tokens": 4443351.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7357928454875946,
+      "epoch": 2.654028436018957,
+      "grad_norm": 0.7527856826782227,
+      "learning_rate": 5.6913625505902966e-06,
+      "loss": 0.7566514015197754,
+      "mean_token_accuracy": 0.8084013164043427,
+      "num_tokens": 4459636.0,
+      "step": 281
+    },
+    {
+      "entropy": 0.7626573443412781,
+      "epoch": 2.6635071090047395,
+      "grad_norm": 0.7092803120613098,
+      "learning_rate": 5.620971788157741e-06,
+      "loss": 0.7342270612716675,
+      "mean_token_accuracy": 0.8117237538099289,
+      "num_tokens": 4475796.0,
+      "step": 282
+    },
+    {
+      "entropy": 0.7978230267763138,
+      "epoch": 2.6729857819905214,
+      "grad_norm": 1.0863252878189087,
+      "learning_rate": 5.550848463759835e-06,
+      "loss": 0.7835026979446411,
+      "mean_token_accuracy": 0.8040365278720856,
+      "num_tokens": 4492101.0,
+      "step": 283
+    },
+    {
+      "entropy": 0.802836075425148,
+      "epoch": 2.6824644549763033,
+      "grad_norm": 0.777554988861084,
+      "learning_rate": 5.480996860000664e-06,
+      "loss": 0.824481725692749,
+      "mean_token_accuracy": 0.7915899306535721,
+      "num_tokens": 4509292.0,
+      "step": 284
+    },
+    {
+      "entropy": 0.7397547513246536,
+      "epoch": 2.6919431279620856,
+      "grad_norm": 0.7662098407745361,
+      "learning_rate": 5.411421242889643e-06,
+      "loss": 0.725763201713562,
+      "mean_token_accuracy": 0.8155921995639801,
+      "num_tokens": 4525300.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.8510775417089462,
+      "epoch": 2.7014218009478674,
+      "grad_norm": 0.7985575199127197,
+      "learning_rate": 5.342125861581022e-06,
+      "loss": 0.844895601272583,
+      "mean_token_accuracy": 0.7873697280883789,
+      "num_tokens": 4541512.0,
+      "step": 286
+    },
+    {
+      "entropy": 0.7734527587890625,
+      "epoch": 2.7109004739336493,
+      "grad_norm": 0.7651690244674683,
+      "learning_rate": 5.273114948114346e-06,
+      "loss": 0.7542927265167236,
+      "mean_token_accuracy": 0.8083190619945526,
+      "num_tokens": 4557742.0,
+      "step": 287
+    },
+    {
+      "entropy": 0.7724734246730804,
+      "epoch": 2.720379146919431,
+      "grad_norm": 0.8156644701957703,
+      "learning_rate": 5.204392717156021e-06,
+      "loss": 0.7651525139808655,
+      "mean_token_accuracy": 0.803424209356308,
+      "num_tokens": 4573502.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.7836635559797287,
+      "epoch": 2.729857819905213,
+      "grad_norm": 0.8487071990966797,
+      "learning_rate": 5.135963365741892e-06,
+      "loss": 0.7864376306533813,
+      "mean_token_accuracy": 0.8052554726600647,
+      "num_tokens": 4588151.0,
+      "step": 289
+    },
+    {
+      "entropy": 0.7748328298330307,
+      "epoch": 2.7393364928909953,
+      "grad_norm": 0.7466703057289124,
+      "learning_rate": 5.067831073020928e-06,
+      "loss": 0.7420310378074646,
+      "mean_token_accuracy": 0.8074294477701187,
+      "num_tokens": 4604464.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.8272679448127747,
+      "epoch": 2.748815165876777,
+      "grad_norm": 0.7622450590133667,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.7982164621353149,
+      "mean_token_accuracy": 0.7979923784732819,
+      "num_tokens": 4620322.0,
+      "step": 291
+    },
+    {
+      "entropy": 0.785077303647995,
+      "epoch": 2.758293838862559,
+      "grad_norm": 0.6796573996543884,
+      "learning_rate": 4.932474289289748e-06,
+      "loss": 0.7424709796905518,
+      "mean_token_accuracy": 0.8152509778738022,
+      "num_tokens": 4638091.0,
+      "step": 292
+    },
+    {
+      "entropy": 0.8112717866897583,
+      "epoch": 2.7677725118483414,
+      "grad_norm": 0.7888740301132202,
+      "learning_rate": 4.865258064851579e-06,
+      "loss": 0.8013285398483276,
+      "mean_token_accuracy": 0.8086142092943192,
+      "num_tokens": 4654194.0,
+      "step": 293
+    },
+    {
+      "entropy": 0.8527653962373734,
+      "epoch": 2.7772511848341233,
+      "grad_norm": 0.728568971157074,
+      "learning_rate": 4.7983554317458204e-06,
+      "loss": 0.8419825434684753,
+      "mean_token_accuracy": 0.7937730550765991,
+      "num_tokens": 4670285.0,
+      "step": 294
+    },
+    {
+      "entropy": 0.7469000071287155,
+      "epoch": 2.786729857819905,
+      "grad_norm": 0.7420099377632141,
+      "learning_rate": 4.731770475880995e-06,
+      "loss": 0.7262124419212341,
+      "mean_token_accuracy": 0.8157135248184204,
+      "num_tokens": 4686154.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.756457582116127,
+      "epoch": 2.7962085308056874,
+      "grad_norm": 0.7652328610420227,
+      "learning_rate": 4.665507263764299e-06,
+      "loss": 0.7771545648574829,
+      "mean_token_accuracy": 0.8060846775770187,
+      "num_tokens": 4702538.0,
+      "step": 296
+    },
+    {
+      "entropy": 0.8069571703672409,
+      "epoch": 2.8056872037914693,
+      "grad_norm": 0.8068948984146118,
+      "learning_rate": 4.599569842253244e-06,
+      "loss": 0.794147253036499,
+      "mean_token_accuracy": 0.7936616390943527,
+      "num_tokens": 4717366.0,
+      "step": 297
+    },
+    {
+      "entropy": 0.7628475576639175,
+      "epoch": 2.815165876777251,
+      "grad_norm": 0.7466553449630737,
+      "learning_rate": 4.5339622383085095e-06,
+      "loss": 0.7227836847305298,
+      "mean_token_accuracy": 0.8107776045799255,
+      "num_tokens": 4733915.0,
+      "step": 298
+    },
+    {
+      "entropy": 0.7967563271522522,
+      "epoch": 2.824644549763033,
+      "grad_norm": 0.7297375202178955,
+      "learning_rate": 4.468688458748006e-06,
+      "loss": 0.78502357006073,
+      "mean_token_accuracy": 0.7997068613767624,
+      "num_tokens": 4751223.0,
+      "step": 299
+    },
+    {
+      "entropy": 0.7497158050537109,
+      "epoch": 2.834123222748815,
+      "grad_norm": 0.7914976477622986,
+      "learning_rate": 4.40375249000216e-06,
+      "loss": 0.7375324964523315,
+      "mean_token_accuracy": 0.81224524974823,
+      "num_tokens": 4767616.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.763348177075386,
+      "epoch": 2.843601895734597,
+      "grad_norm": 0.7949478030204773,
+      "learning_rate": 4.339158297870469e-06,
+      "loss": 0.7617372870445251,
+      "mean_token_accuracy": 0.8068404793739319,
+      "num_tokens": 4783264.0,
+      "step": 301
+    },
+    {
+      "entropy": 0.796427458524704,
+      "epoch": 2.853080568720379,
+      "grad_norm": 0.8715385794639587,
+      "learning_rate": 4.274909827279283e-06,
+      "loss": 0.7832701206207275,
+      "mean_token_accuracy": 0.8009211868047714,
+      "num_tokens": 4798500.0,
+      "step": 302
+    },
+    {
+      "entropy": 0.8474353551864624,
+      "epoch": 2.862559241706161,
+      "grad_norm": 0.6873095035552979,
+      "learning_rate": 4.211011002040885e-06,
+      "loss": 0.8359758853912354,
+      "mean_token_accuracy": 0.7899835258722305,
+      "num_tokens": 4815731.0,
+      "step": 303
+    },
+    {
+      "entropy": 0.7330864071846008,
+      "epoch": 2.8720379146919433,
+      "grad_norm": 0.771339476108551,
+      "learning_rate": 4.14746572461387e-06,
+      "loss": 0.7329787611961365,
+      "mean_token_accuracy": 0.8161643743515015,
+      "num_tokens": 4831321.0,
+      "step": 304
+    },
+    {
+      "entropy": 0.7069019228219986,
+      "epoch": 2.881516587677725,
+      "grad_norm": 0.7667499780654907,
+      "learning_rate": 4.084277875864776e-06,
+      "loss": 0.6972311735153198,
+      "mean_token_accuracy": 0.8227463960647583,
+      "num_tokens": 4847145.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.7471538782119751,
+      "epoch": 2.890995260663507,
+      "grad_norm": 0.7328512668609619,
+      "learning_rate": 4.021451314831113e-06,
+      "loss": 0.7734975814819336,
+      "mean_token_accuracy": 0.8095995932817459,
+      "num_tokens": 4863550.0,
+      "step": 306
+    },
+    {
+      "entropy": 0.7084924876689911,
+      "epoch": 2.9004739336492893,
+      "grad_norm": 0.707899808883667,
+      "learning_rate": 3.958989878485644e-06,
+      "loss": 0.6905066967010498,
+      "mean_token_accuracy": 0.8189463019371033,
+      "num_tokens": 4879708.0,
+      "step": 307
+    },
+    {
+      "entropy": 0.8038249611854553,
+      "epoch": 2.909952606635071,
+      "grad_norm": 0.7571305632591248,
+      "learning_rate": 3.896897381502081e-06,
+      "loss": 0.8163067102432251,
+      "mean_token_accuracy": 0.797750398516655,
+      "num_tokens": 4896533.0,
+      "step": 308
+    },
+    {
+      "entropy": 0.8064074218273163,
+      "epoch": 2.919431279620853,
+      "grad_norm": 0.7309732437133789,
+      "learning_rate": 3.83517761602209e-06,
+      "loss": 0.792744517326355,
+      "mean_token_accuracy": 0.7964637726545334,
+      "num_tokens": 4912287.0,
+      "step": 309
+    },
+    {
+      "entropy": 0.7508046329021454,
+      "epoch": 2.9289099526066353,
+      "grad_norm": 0.8369373679161072,
+      "learning_rate": 3.773834351423711e-06,
+      "loss": 0.7411284446716309,
+      "mean_token_accuracy": 0.8095607906579971,
+      "num_tokens": 4927555.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7564988732337952,
+      "epoch": 2.938388625592417,
+      "grad_norm": 0.859829843044281,
+      "learning_rate": 3.712871334091154e-06,
+      "loss": 0.7624916434288025,
+      "mean_token_accuracy": 0.8062203228473663,
+      "num_tokens": 4942708.0,
+      "step": 311
+    },
+    {
+      "entropy": 0.7336708754301071,
+      "epoch": 2.947867298578199,
+      "grad_norm": 0.7614916563034058,
+      "learning_rate": 3.652292287185979e-06,
+      "loss": 0.737271785736084,
+      "mean_token_accuracy": 0.8161173611879349,
+      "num_tokens": 4958673.0,
+      "step": 312
+    },
+    {
+      "entropy": 0.7636863887310028,
+      "epoch": 2.957345971563981,
+      "grad_norm": 0.7981767654418945,
+      "learning_rate": 3.592100910419738e-06,
+      "loss": 0.7631924748420715,
+      "mean_token_accuracy": 0.8060728013515472,
+      "num_tokens": 4974424.0,
+      "step": 313
+    },
+    {
+      "entropy": 0.7589444667100906,
+      "epoch": 2.966824644549763,
+      "grad_norm": 0.7552541494369507,
+      "learning_rate": 3.532300879828013e-06,
+      "loss": 0.7564074993133545,
+      "mean_token_accuracy": 0.8108653426170349,
+      "num_tokens": 4990180.0,
+      "step": 314
+    },
+    {
+      "entropy": 0.7788964807987213,
+      "epoch": 2.976303317535545,
+      "grad_norm": 0.7552350163459778,
+      "learning_rate": 3.4728958475459052e-06,
+      "loss": 0.785426139831543,
+      "mean_token_accuracy": 0.7967838943004608,
+      "num_tokens": 5006243.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.7222247868776321,
+      "epoch": 2.985781990521327,
+      "grad_norm": 0.8034160733222961,
+      "learning_rate": 3.413889441584999e-06,
+      "loss": 0.723943829536438,
+      "mean_token_accuracy": 0.8158971667289734,
+      "num_tokens": 5022940.0,
+      "step": 316
+    },
+    {
+      "entropy": 0.7745039016008377,
+      "epoch": 2.995260663507109,
+      "grad_norm": 0.7772722840309143,
+      "learning_rate": 3.355285265611784e-06,
+      "loss": 0.7676458954811096,
+      "mean_token_accuracy": 0.8127903640270233,
+      "num_tokens": 5039298.0,
+      "step": 317
+    },
+    {
+      "entropy": 0.8236450850963593,
+      "epoch": 3.0,
+      "grad_norm": 1.20962655544281,
+      "learning_rate": 3.297086898727583e-06,
+      "loss": 0.8019519448280334,
+      "mean_token_accuracy": 0.8037053942680359,
+      "num_tokens": 5046658.0,
+      "step": 318
+    },
+    {
+      "entropy": 0.7478779852390289,
+      "epoch": 3.009478672985782,
+      "grad_norm": 0.7697426080703735,
+      "learning_rate": 3.2392978952499553e-06,
+      "loss": 0.7716673612594604,
+      "mean_token_accuracy": 0.8018212914466858,
+      "num_tokens": 5061883.0,
+      "step": 319
+    },
+    {
+      "entropy": 0.785962700843811,
+      "epoch": 3.018957345971564,
+      "grad_norm": 0.81953364610672,
+      "learning_rate": 3.1819217844956216e-06,
+      "loss": 0.7965259552001953,
+      "mean_token_accuracy": 0.8047526180744171,
+      "num_tokens": 5077195.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7776083499193192,
+      "epoch": 3.028436018957346,
+      "grad_norm": 0.6892860531806946,
+      "learning_rate": 3.1249620705649417e-06,
+      "loss": 0.7372241020202637,
+      "mean_token_accuracy": 0.8053242713212967,
+      "num_tokens": 5092842.0,
+      "step": 321
+    },
+    {
+      "entropy": 0.7553633451461792,
+      "epoch": 3.037914691943128,
+      "grad_norm": 0.7420003414154053,
+      "learning_rate": 3.0684222321278824e-06,
+      "loss": 0.7217973470687866,
+      "mean_token_accuracy": 0.8165572732686996,
+      "num_tokens": 5107920.0,
+      "step": 322
+    },
+    {
+      "entropy": 0.8048459887504578,
+      "epoch": 3.0473933649289098,
+      "grad_norm": 0.7054750919342041,
+      "learning_rate": 3.0123057222115835e-06,
+      "loss": 0.8139917850494385,
+      "mean_token_accuracy": 0.7946743965148926,
+      "num_tokens": 5123998.0,
+      "step": 323
+    },
+    {
+      "entropy": 0.6766252219676971,
+      "epoch": 3.056872037914692,
+      "grad_norm": 0.6615550518035889,
+      "learning_rate": 2.956615967989479e-06,
+      "loss": 0.6536415219306946,
+      "mean_token_accuracy": 0.8297200500965118,
+      "num_tokens": 5140358.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.7243975102901459,
+      "epoch": 3.066350710900474,
+      "grad_norm": 0.7182250618934631,
+      "learning_rate": 2.9013563705719673e-06,
+      "loss": 0.712523341178894,
+      "mean_token_accuracy": 0.819568395614624,
+      "num_tokens": 5156707.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.743099570274353,
+      "epoch": 3.075829383886256,
+      "grad_norm": 0.765571653842926,
+      "learning_rate": 2.846530304798727e-06,
+      "loss": 0.7095881700515747,
+      "mean_token_accuracy": 0.8208495527505875,
+      "num_tokens": 5172592.0,
+      "step": 326
+    },
+    {
+      "entropy": 0.7437466382980347,
+      "epoch": 3.085308056872038,
+      "grad_norm": 0.7301930785179138,
+      "learning_rate": 2.7921411190325753e-06,
+      "loss": 0.741960883140564,
+      "mean_token_accuracy": 0.8145260065793991,
+      "num_tokens": 5188198.0,
+      "step": 327
+    },
+    {
+      "entropy": 0.7824875563383102,
+      "epoch": 3.09478672985782,
+      "grad_norm": 0.7977142930030823,
+      "learning_rate": 2.73819213495501e-06,
+      "loss": 0.7875393629074097,
+      "mean_token_accuracy": 0.8107055574655533,
+      "num_tokens": 5202724.0,
+      "step": 328
+    },
+    {
+      "entropy": 0.7904134392738342,
+      "epoch": 3.104265402843602,
+      "grad_norm": 0.7295860648155212,
+      "learning_rate": 2.6846866473633126e-06,
+      "loss": 0.7971727848052979,
+      "mean_token_accuracy": 0.7985574156045914,
+      "num_tokens": 5218061.0,
+      "step": 329
+    },
+    {
+      "entropy": 0.7173660099506378,
+      "epoch": 3.1137440758293837,
+      "grad_norm": 0.7091414928436279,
+      "learning_rate": 2.6316279239693467e-06,
+      "loss": 0.7103780508041382,
+      "mean_token_accuracy": 0.8185557723045349,
+      "num_tokens": 5234608.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.6958528906106949,
+      "epoch": 3.123222748815166,
+      "grad_norm": 0.7575947642326355,
+      "learning_rate": 2.579019205199992e-06,
+      "loss": 0.6730659604072571,
+      "mean_token_accuracy": 0.8267861753702164,
+      "num_tokens": 5250548.0,
+      "step": 331
+    },
+    {
+      "entropy": 0.7260988652706146,
+      "epoch": 3.132701421800948,
+      "grad_norm": 0.6860673427581787,
+      "learning_rate": 2.5268637039992296e-06,
+      "loss": 0.7105354070663452,
+      "mean_token_accuracy": 0.8181567341089249,
+      "num_tokens": 5266449.0,
+      "step": 332
+    },
+    {
+      "entropy": 0.6991846412420273,
+      "epoch": 3.1421800947867298,
+      "grad_norm": 0.8046501874923706,
+      "learning_rate": 2.4751646056319334e-06,
+      "loss": 0.6967523097991943,
+      "mean_token_accuracy": 0.8243842422962189,
+      "num_tokens": 5282201.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.8095178604125977,
+      "epoch": 3.1516587677725116,
+      "grad_norm": 0.743904173374176,
+      "learning_rate": 2.4239250674893345e-06,
+      "loss": 0.8167105913162231,
+      "mean_token_accuracy": 0.7961531728506088,
+      "num_tokens": 5298517.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.8006535023450851,
+      "epoch": 3.161137440758294,
+      "grad_norm": 0.79604572057724,
+      "learning_rate": 2.373148218896182e-06,
+      "loss": 0.8157421350479126,
+      "mean_token_accuracy": 0.7973491996526718,
+      "num_tokens": 5313538.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.760479748249054,
+      "epoch": 3.170616113744076,
+      "grad_norm": 0.7916611433029175,
+      "learning_rate": 2.32283716091964e-06,
+      "loss": 0.761253833770752,
+      "mean_token_accuracy": 0.8064158409833908,
+      "num_tokens": 5329238.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.7230188250541687,
+      "epoch": 3.1800947867298577,
+      "grad_norm": 0.7429656982421875,
+      "learning_rate": 2.2729949661798876e-06,
+      "loss": 0.7148363590240479,
+      "mean_token_accuracy": 0.8210389465093613,
+      "num_tokens": 5344858.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.7517809122800827,
+      "epoch": 3.18957345971564,
+      "grad_norm": 0.7186969518661499,
+      "learning_rate": 2.2236246786624794e-06,
+      "loss": 0.7122472524642944,
+      "mean_token_accuracy": 0.817431777715683,
+      "num_tokens": 5361465.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.7896288186311722,
+      "epoch": 3.199052132701422,
+      "grad_norm": 0.7846527695655823,
+      "learning_rate": 2.174729313532433e-06,
+      "loss": 0.7668826580047607,
+      "mean_token_accuracy": 0.8099963366985321,
+      "num_tokens": 5376610.0,
+      "step": 339
+    },
+    {
+      "entropy": 0.7389908730983734,
+      "epoch": 3.2085308056872037,
+      "grad_norm": 0.7456527352333069,
+      "learning_rate": 2.1263118569500797e-06,
+      "loss": 0.7495481967926025,
+      "mean_token_accuracy": 0.8134677708148956,
+      "num_tokens": 5393335.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7424813508987427,
+      "epoch": 3.2180094786729856,
+      "grad_norm": 0.6882187724113464,
+      "learning_rate": 2.078375265888707e-06,
+      "loss": 0.7607920169830322,
+      "mean_token_accuracy": 0.8078367114067078,
+      "num_tokens": 5410633.0,
+      "step": 341
+    },
+    {
+      "entropy": 0.6920035630464554,
+      "epoch": 3.227488151658768,
+      "grad_norm": 0.7300297021865845,
+      "learning_rate": 2.0309224679539554e-06,
+      "loss": 0.6803273558616638,
+      "mean_token_accuracy": 0.8286775499582291,
+      "num_tokens": 5426722.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.763126015663147,
+      "epoch": 3.2369668246445498,
+      "grad_norm": 0.7958213090896606,
+      "learning_rate": 1.9839563612050273e-06,
+      "loss": 0.7651641368865967,
+      "mean_token_accuracy": 0.8074810951948166,
+      "num_tokens": 5442247.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.7819797247648239,
+      "epoch": 3.2464454976303316,
+      "grad_norm": 0.7771494388580322,
+      "learning_rate": 1.937479813977703e-06,
+      "loss": 0.7724591493606567,
+      "mean_token_accuracy": 0.8058855086565018,
+      "num_tokens": 5458691.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.7670372873544693,
+      "epoch": 3.2559241706161135,
+      "grad_norm": 0.7733996510505676,
+      "learning_rate": 1.8914956647091497e-06,
+      "loss": 0.77671217918396,
+      "mean_token_accuracy": 0.8054281175136566,
+      "num_tokens": 5474026.0,
+      "step": 345
+    },
+    {
+      "entropy": 0.816511332988739,
+      "epoch": 3.265402843601896,
+      "grad_norm": 0.835043728351593,
+      "learning_rate": 1.846006721764586e-06,
+      "loss": 0.820414125919342,
+      "mean_token_accuracy": 0.7958302795886993,
+      "num_tokens": 5489864.0,
+      "step": 346
+    },
+    {
+      "entropy": 0.7935537993907928,
+      "epoch": 3.2748815165876777,
+      "grad_norm": 0.7735744118690491,
+      "learning_rate": 1.8010157632657544e-06,
+      "loss": 0.7576620578765869,
+      "mean_token_accuracy": 0.8064090013504028,
+      "num_tokens": 5505997.0,
+      "step": 347
+    },
+    {
+      "entropy": 0.808742880821228,
+      "epoch": 3.2843601895734595,
+      "grad_norm": 0.7906506657600403,
+      "learning_rate": 1.7565255369212664e-06,
+      "loss": 0.7947472333908081,
+      "mean_token_accuracy": 0.7945470958948135,
+      "num_tokens": 5521795.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.7717082351446152,
+      "epoch": 3.293838862559242,
+      "grad_norm": 0.72123783826828,
+      "learning_rate": 1.7125387598587862e-06,
+      "loss": 0.785935640335083,
+      "mean_token_accuracy": 0.805345818400383,
+      "num_tokens": 5538348.0,
+      "step": 349
+    },
+    {
+      "entropy": 0.7952146828174591,
+      "epoch": 3.3033175355450237,
+      "grad_norm": 0.821018397808075,
+      "learning_rate": 1.6690581184590859e-06,
+      "loss": 0.8067296743392944,
+      "mean_token_accuracy": 0.7967140823602676,
+      "num_tokens": 5553641.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7633357793092728,
+      "epoch": 3.3127962085308056,
+      "grad_norm": 0.7896275520324707,
+      "learning_rate": 1.6260862681919965e-06,
+      "loss": 0.7645949721336365,
+      "mean_token_accuracy": 0.8082640618085861,
+      "num_tokens": 5569753.0,
+      "step": 351
+    },
+    {
+      "entropy": 0.749266117811203,
+      "epoch": 3.322274881516588,
+      "grad_norm": 0.7245807647705078,
+      "learning_rate": 1.583625833454211e-06,
+      "loss": 0.7372440099716187,
+      "mean_token_accuracy": 0.8141664415597916,
+      "num_tokens": 5586188.0,
+      "step": 352
+    },
+    {
+      "entropy": 0.7516562193632126,
+      "epoch": 3.3317535545023698,
+      "grad_norm": 0.7951227426528931,
+      "learning_rate": 1.5416794074090258e-06,
+      "loss": 0.7702144384384155,
+      "mean_token_accuracy": 0.804230198264122,
+      "num_tokens": 5601649.0,
+      "step": 353
+    },
+    {
+      "entropy": 0.7971460670232773,
+      "epoch": 3.3412322274881516,
+      "grad_norm": 0.8076654076576233,
+      "learning_rate": 1.500249551827958e-06,
+      "loss": 0.7622384428977966,
+      "mean_token_accuracy": 0.8001376986503601,
+      "num_tokens": 5616188.0,
+      "step": 354
+    },
+    {
+      "entropy": 0.7576956003904343,
+      "epoch": 3.3507109004739335,
+      "grad_norm": 0.7258307933807373,
+      "learning_rate": 1.459338796934293e-06,
+      "loss": 0.7539409399032593,
+      "mean_token_accuracy": 0.8105089962482452,
+      "num_tokens": 5635082.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.7855661660432816,
+      "epoch": 3.360189573459716,
+      "grad_norm": 0.7409619688987732,
+      "learning_rate": 1.4189496412485593e-06,
+      "loss": 0.7873318195343018,
+      "mean_token_accuracy": 0.8049680441617966,
+      "num_tokens": 5652264.0,
+      "step": 356
+    },
+    {
+      "entropy": 0.8110678791999817,
+      "epoch": 3.3696682464454977,
+      "grad_norm": 0.7768043279647827,
+      "learning_rate": 1.3790845514359363e-06,
+      "loss": 0.7789173126220703,
+      "mean_token_accuracy": 0.8034784644842148,
+      "num_tokens": 5668104.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.7338491827249527,
+      "epoch": 3.3791469194312795,
+      "grad_norm": 0.7567062973976135,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.742594838142395,
+      "mean_token_accuracy": 0.8137776404619217,
+      "num_tokens": 5684769.0,
+      "step": 358
+    },
+    {
+      "entropy": 0.6945641934871674,
+      "epoch": 3.3886255924170614,
+      "grad_norm": 0.6800957918167114,
+      "learning_rate": 1.300936275912098e-06,
+      "loss": 0.7036481499671936,
+      "mean_token_accuracy": 0.8183564692735672,
+      "num_tokens": 5702754.0,
+      "step": 359
+    },
+    {
+      "entropy": 0.7719693928956985,
+      "epoch": 3.3981042654028437,
+      "grad_norm": 0.8576886653900146,
+      "learning_rate": 1.2626578629084784e-06,
+      "loss": 0.77418053150177,
+      "mean_token_accuracy": 0.8084691315889359,
+      "num_tokens": 5718742.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7500364035367966,
+      "epoch": 3.4075829383886256,
+      "grad_norm": 0.7127568125724792,
+      "learning_rate": 1.224913060901688e-06,
+      "loss": 0.7334154844284058,
+      "mean_token_accuracy": 0.8134698867797852,
+      "num_tokens": 5734815.0,
+      "step": 361
+    },
+    {
+      "entropy": 0.7798032909631729,
+      "epoch": 3.4170616113744074,
+      "grad_norm": 0.7596098184585571,
+      "learning_rate": 1.1877041750597174e-06,
+      "loss": 0.7881962060928345,
+      "mean_token_accuracy": 0.7973756939172745,
+      "num_tokens": 5751466.0,
+      "step": 362
+    },
+    {
+      "entropy": 0.7736654728651047,
+      "epoch": 3.4265402843601898,
+      "grad_norm": 0.8830217719078064,
+      "learning_rate": 1.1510334778208332e-06,
+      "loss": 0.7824318408966064,
+      "mean_token_accuracy": 0.8039457201957703,
+      "num_tokens": 5766604.0,
+      "step": 363
+    },
+    {
+      "entropy": 0.7304582744836807,
+      "epoch": 3.4360189573459716,
+      "grad_norm": 0.7475463151931763,
+      "learning_rate": 1.1149032087548117e-06,
+      "loss": 0.6829269528388977,
+      "mean_token_accuracy": 0.8150535821914673,
+      "num_tokens": 5782424.0,
+      "step": 364
+    },
+    {
+      "entropy": 0.7141664177179337,
+      "epoch": 3.4454976303317535,
+      "grad_norm": 0.8083080053329468,
+      "learning_rate": 1.0793155744261352e-06,
+      "loss": 0.7171058058738708,
+      "mean_token_accuracy": 0.8167122006416321,
+      "num_tokens": 5798061.0,
+      "step": 365
+    },
+    {
+      "entropy": 0.7113132625818253,
+      "epoch": 3.4549763033175354,
+      "grad_norm": 0.7315077185630798,
+      "learning_rate": 1.0442727482592596e-06,
+      "loss": 0.7110543847084045,
+      "mean_token_accuracy": 0.827058881521225,
+      "num_tokens": 5813964.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.794712170958519,
+      "epoch": 3.4644549763033177,
+      "grad_norm": 0.6843957901000977,
+      "learning_rate": 1.0097768704058542e-06,
+      "loss": 0.7619876265525818,
+      "mean_token_accuracy": 0.8037076890468597,
+      "num_tokens": 5830167.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.7343897819519043,
+      "epoch": 3.4739336492890995,
+      "grad_norm": 0.7758681178092957,
+      "learning_rate": 9.75830047614117e-07,
+      "loss": 0.7308058738708496,
+      "mean_token_accuracy": 0.8147377222776413,
+      "num_tokens": 5847113.0,
+      "step": 368
+    },
+    {
+      "entropy": 0.7304037660360336,
+      "epoch": 3.4834123222748814,
+      "grad_norm": 0.8322577476501465,
+      "learning_rate": 9.424343531000968e-07,
+      "loss": 0.7368799448013306,
+      "mean_token_accuracy": 0.8109440207481384,
+      "num_tokens": 5863085.0,
+      "step": 369
+    },
+    {
+      "entropy": 0.6853264719247818,
+      "epoch": 3.4928909952606633,
+      "grad_norm": 0.8498042821884155,
+      "learning_rate": 9.095918264210779e-07,
+      "loss": 0.67824387550354,
+      "mean_token_accuracy": 0.822025716304779,
+      "num_tokens": 5879230.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7402042001485825,
+      "epoch": 3.5023696682464456,
+      "grad_norm": 0.7984394431114197,
+      "learning_rate": 8.773044733510338e-07,
+      "loss": 0.7216671705245972,
+      "mean_token_accuracy": 0.8155932128429413,
+      "num_tokens": 5893567.0,
+      "step": 371
+    },
+    {
+      "entropy": 0.7720407843589783,
+      "epoch": 3.5118483412322274,
+      "grad_norm": 0.7581180334091187,
+      "learning_rate": 8.455742657581067e-07,
+      "loss": 0.7469707727432251,
+      "mean_token_accuracy": 0.8138709366321564,
+      "num_tokens": 5908988.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.7801290154457092,
+      "epoch": 3.5213270142180093,
+      "grad_norm": 0.8222272992134094,
+      "learning_rate": 8.144031414842012e-07,
+      "loss": 0.7962188124656677,
+      "mean_token_accuracy": 0.7990644872188568,
+      "num_tokens": 5924822.0,
+      "step": 373
+    },
+    {
+      "entropy": 0.8463993817567825,
+      "epoch": 3.5308056872037916,
+      "grad_norm": 0.827546238899231,
+      "learning_rate": 7.837930042266262e-07,
+      "loss": 0.8375414609909058,
+      "mean_token_accuracy": 0.7954217940568924,
+      "num_tokens": 5940568.0,
+      "step": 374
+    },
+    {
+      "entropy": 0.792610689997673,
+      "epoch": 3.5402843601895735,
+      "grad_norm": 0.7309672832489014,
+      "learning_rate": 7.537457234218271e-07,
+      "loss": 0.8083318471908569,
+      "mean_token_accuracy": 0.8017033785581589,
+      "num_tokens": 5956348.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.7301171571016312,
+      "epoch": 3.5497630331753554,
+      "grad_norm": 0.6857864260673523,
+      "learning_rate": 7.242631341312234e-07,
+      "loss": 0.691540002822876,
+      "mean_token_accuracy": 0.8231353312730789,
+      "num_tokens": 5971747.0,
+      "step": 376
+    },
+    {
+      "entropy": 0.8102651387453079,
+      "epoch": 3.5592417061611377,
+      "grad_norm": 0.6648951768875122,
+      "learning_rate": 6.953470369291349e-07,
+      "loss": 0.7997016906738281,
+      "mean_token_accuracy": 0.8020312041044235,
+      "num_tokens": 5988831.0,
+      "step": 377
+    },
+    {
+      "entropy": 0.7594100683927536,
+      "epoch": 3.5687203791469195,
+      "grad_norm": 0.7718584537506104,
+      "learning_rate": 6.669991977928103e-07,
+      "loss": 0.7339326739311218,
+      "mean_token_accuracy": 0.8102892637252808,
+      "num_tokens": 6004610.0,
+      "step": 378
+    },
+    {
+      "entropy": 0.7296763807535172,
+      "epoch": 3.5781990521327014,
+      "grad_norm": 0.7523729801177979,
+      "learning_rate": 6.392213479945852e-07,
+      "loss": 0.7240911722183228,
+      "mean_token_accuracy": 0.8162074387073517,
+      "num_tokens": 6019991.0,
+      "step": 379
+    },
+    {
+      "entropy": 0.8269750326871872,
+      "epoch": 3.5876777251184833,
+      "grad_norm": 0.7713115811347961,
+      "learning_rate": 6.120151839961363e-07,
+      "loss": 0.8135149478912354,
+      "mean_token_accuracy": 0.7890425622463226,
+      "num_tokens": 6036037.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7625905424356461,
+      "epoch": 3.597156398104265,
+      "grad_norm": 0.8032811880111694,
+      "learning_rate": 5.853823673448877e-07,
+      "loss": 0.7643755078315735,
+      "mean_token_accuracy": 0.8126851618289948,
+      "num_tokens": 6051369.0,
+      "step": 381
+    },
+    {
+      "entropy": 0.7515154629945755,
+      "epoch": 3.6066350710900474,
+      "grad_norm": 0.693193793296814,
+      "learning_rate": 5.593245245725231e-07,
+      "loss": 0.7446131706237793,
+      "mean_token_accuracy": 0.8108726441860199,
+      "num_tokens": 6068292.0,
+      "step": 382
+    },
+    {
+      "entropy": 0.8014443516731262,
+      "epoch": 3.6161137440758293,
+      "grad_norm": 0.8282812833786011,
+      "learning_rate": 5.33843247095659e-07,
+      "loss": 0.7868412137031555,
+      "mean_token_accuracy": 0.8031998574733734,
+      "num_tokens": 6083746.0,
+      "step": 383
+    },
+    {
+      "entropy": 0.7436373382806778,
+      "epoch": 3.625592417061611,
+      "grad_norm": 0.7708470225334167,
+      "learning_rate": 5.089400911186504e-07,
+      "loss": 0.7162569165229797,
+      "mean_token_accuracy": 0.8107694387435913,
+      "num_tokens": 6099838.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.7486245483160019,
+      "epoch": 3.6350710900473935,
+      "grad_norm": 0.7626857161521912,
+      "learning_rate": 4.846165775385459e-07,
+      "loss": 0.7347097396850586,
+      "mean_token_accuracy": 0.8151373565196991,
+      "num_tokens": 6115229.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.7286288738250732,
+      "epoch": 3.6445497630331753,
+      "grad_norm": 0.8355617523193359,
+      "learning_rate": 4.6087419185220973e-07,
+      "loss": 0.7086689472198486,
+      "mean_token_accuracy": 0.811622828245163,
+      "num_tokens": 6131495.0,
+      "step": 386
+    },
+    {
+      "entropy": 0.7331616282463074,
+      "epoch": 3.654028436018957,
+      "grad_norm": 0.8304053544998169,
+      "learning_rate": 4.3771438406559173e-07,
+      "loss": 0.6994441747665405,
+      "mean_token_accuracy": 0.8185713589191437,
+      "num_tokens": 6147229.0,
+      "step": 387
+    },
+    {
+      "entropy": 0.8535281419754028,
+      "epoch": 3.6635071090047395,
+      "grad_norm": 0.8453361988067627,
+      "learning_rate": 4.1513856860517676e-07,
+      "loss": 0.8577812314033508,
+      "mean_token_accuracy": 0.7872657924890518,
+      "num_tokens": 6162417.0,
+      "step": 388
+    },
+    {
+      "entropy": 0.7746417671442032,
+      "epoch": 3.6729857819905214,
+      "grad_norm": 0.7792283892631531,
+      "learning_rate": 3.931481242315993e-07,
+      "loss": 0.7408183217048645,
+      "mean_token_accuracy": 0.8043603003025055,
+      "num_tokens": 6177688.0,
+      "step": 389
+    },
+    {
+      "entropy": 0.8118592947721481,
+      "epoch": 3.6824644549763033,
+      "grad_norm": 0.7837164998054504,
+      "learning_rate": 3.7174439395543884e-07,
+      "loss": 0.8173757791519165,
+      "mean_token_accuracy": 0.7939213812351227,
+      "num_tokens": 6193823.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.744156688451767,
+      "epoch": 3.6919431279620856,
+      "grad_norm": 0.7805715799331665,
+      "learning_rate": 3.5092868495520294e-07,
+      "loss": 0.7480678558349609,
+      "mean_token_accuracy": 0.8105067759752274,
+      "num_tokens": 6209504.0,
+      "step": 391
+    },
+    {
+      "entropy": 0.7389348596334457,
+      "epoch": 3.7014218009478674,
+      "grad_norm": 0.7441927790641785,
+      "learning_rate": 3.3070226849749367e-07,
+      "loss": 0.7176021337509155,
+      "mean_token_accuracy": 0.8149614036083221,
+      "num_tokens": 6226077.0,
+      "step": 392
+    },
+    {
+      "entropy": 0.8332685679197311,
+      "epoch": 3.7109004739336493,
+      "grad_norm": 0.7937362194061279,
+      "learning_rate": 3.110663798593616e-07,
+      "loss": 0.8166630864143372,
+      "mean_token_accuracy": 0.7967136949300766,
+      "num_tokens": 6241257.0,
+      "step": 393
+    },
+    {
+      "entropy": 0.8039902150630951,
+      "epoch": 3.720379146919431,
+      "grad_norm": 0.7978947162628174,
+      "learning_rate": 2.920222182528754e-07,
+      "loss": 0.7984297871589661,
+      "mean_token_accuracy": 0.798541709780693,
+      "num_tokens": 6257243.0,
+      "step": 394
+    },
+    {
+      "entropy": 0.8189259767532349,
+      "epoch": 3.729857819905213,
+      "grad_norm": 0.8544439673423767,
+      "learning_rate": 2.735709467518699e-07,
+      "loss": 0.8043655753135681,
+      "mean_token_accuracy": 0.8035803735256195,
+      "num_tokens": 6271493.0,
+      "step": 395
+    },
+    {
+      "entropy": 0.7036288529634476,
+      "epoch": 3.7393364928909953,
+      "grad_norm": 0.7824522852897644,
+      "learning_rate": 2.557136922209269e-07,
+      "loss": 0.7129417061805725,
+      "mean_token_accuracy": 0.818129375576973,
+      "num_tokens": 6287353.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.7242246717214584,
+      "epoch": 3.748815165876777,
+      "grad_norm": 0.8705534934997559,
+      "learning_rate": 2.384515452465475e-07,
+      "loss": 0.7420156002044678,
+      "mean_token_accuracy": 0.8138059079647064,
+      "num_tokens": 6303511.0,
+      "step": 397
+    },
+    {
+      "entropy": 0.7527247965335846,
+      "epoch": 3.758293838862559,
+      "grad_norm": 0.756104588508606,
+      "learning_rate": 2.2178556007054876e-07,
+      "loss": 0.7536898851394653,
+      "mean_token_accuracy": 0.8014432936906815,
+      "num_tokens": 6319766.0,
+      "step": 398
+    },
+    {
+      "entropy": 0.7517576813697815,
+      "epoch": 3.7677725118483414,
+      "grad_norm": 0.7568046450614929,
+      "learning_rate": 2.0571675452567997e-07,
+      "loss": 0.7543550729751587,
+      "mean_token_accuracy": 0.8108285665512085,
+      "num_tokens": 6336445.0,
+      "step": 399
+    },
+    {
+      "entropy": 0.7451558709144592,
+      "epoch": 3.7772511848341233,
+      "grad_norm": 0.738186240196228,
+      "learning_rate": 1.9024610997345872e-07,
+      "loss": 0.7499051094055176,
+      "mean_token_accuracy": 0.8100962191820145,
+      "num_tokens": 6352720.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.7548184543848038,
+      "epoch": 3.786729857819905,
+      "grad_norm": 0.8254923224449158,
+      "learning_rate": 1.7537457124423896e-07,
+      "loss": 0.7180478572845459,
+      "mean_token_accuracy": 0.8143037408590317,
+      "num_tokens": 6368238.0,
+      "step": 401
+    },
+    {
+      "entropy": 0.765844538807869,
+      "epoch": 3.7962085308056874,
+      "grad_norm": 0.765632152557373,
+      "learning_rate": 1.6110304657950714e-07,
+      "loss": 0.7474029660224915,
+      "mean_token_accuracy": 0.8108795136213303,
+      "num_tokens": 6384535.0,
+      "step": 402
+    },
+    {
+      "entropy": 0.8129818141460419,
+      "epoch": 3.8056872037914693,
+      "grad_norm": 0.8012934923171997,
+      "learning_rate": 1.474324075764111e-07,
+      "loss": 0.8102051019668579,
+      "mean_token_accuracy": 0.8035916984081268,
+      "num_tokens": 6399709.0,
+      "step": 403
+    },
+    {
+      "entropy": 0.7898041009902954,
+      "epoch": 3.815165876777251,
+      "grad_norm": 0.7556485533714294,
+      "learning_rate": 1.3436348913453578e-07,
+      "loss": 0.7623159885406494,
+      "mean_token_accuracy": 0.7990731745958328,
+      "num_tokens": 6415199.0,
+      "step": 404
+    },
+    {
+      "entropy": 0.7221266180276871,
+      "epoch": 3.824644549763033,
+      "grad_norm": 0.6958292126655579,
+      "learning_rate": 1.2189708940490653e-07,
+      "loss": 0.7062927484512329,
+      "mean_token_accuracy": 0.8177174478769302,
+      "num_tokens": 6431083.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.7521060258150101,
+      "epoch": 3.834123222748815,
+      "grad_norm": 0.6797523498535156,
+      "learning_rate": 1.1003396974124892e-07,
+      "loss": 0.7314427495002747,
+      "mean_token_accuracy": 0.821082592010498,
+      "num_tokens": 6447089.0,
+      "step": 406
+    },
+    {
+      "entropy": 0.7682420760393143,
+      "epoch": 3.843601895734597,
+      "grad_norm": 0.7217481136322021,
+      "learning_rate": 9.877485465349057e-08,
+      "loss": 0.7689416408538818,
+      "mean_token_accuracy": 0.8068108856678009,
+      "num_tokens": 6464283.0,
+      "step": 407
+    },
+    {
+      "entropy": 0.6823010891675949,
+      "epoch": 3.853080568720379,
+      "grad_norm": 0.8023832440376282,
+      "learning_rate": 8.812043176351093e-08,
+      "loss": 0.6633629202842712,
+      "mean_token_accuracy": 0.8254884779453278,
+      "num_tokens": 6481156.0,
+      "step": 408
+    },
+    {
+      "entropy": 0.7826178818941116,
+      "epoch": 3.862559241706161,
+      "grad_norm": 0.7779275178909302,
+      "learning_rate": 7.807135176314707e-08,
+      "loss": 0.761837363243103,
+      "mean_token_accuracy": 0.80552838742733,
+      "num_tokens": 6496497.0,
+      "step": 409
+    },
+    {
+      "entropy": 0.7546826750040054,
+      "epoch": 3.8720379146919433,
+      "grad_norm": 0.824887752532959,
+      "learning_rate": 6.862822837445882e-08,
+      "loss": 0.7290815114974976,
+      "mean_token_accuracy": 0.8155907243490219,
+      "num_tokens": 6511042.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8004759103059769,
+      "epoch": 3.881516587677725,
+      "grad_norm": 0.7989736795425415,
+      "learning_rate": 5.979163831223988e-08,
+      "loss": 0.7977555990219116,
+      "mean_token_accuracy": 0.8019419759511948,
+      "num_tokens": 6527738.0,
+      "step": 411
+    },
+    {
+      "entropy": 0.7018654048442841,
+      "epoch": 3.890995260663507,
+      "grad_norm": 0.7676022052764893,
+      "learning_rate": 5.1562121248803776e-08,
+      "loss": 0.684330940246582,
+      "mean_token_accuracy": 0.8249032348394394,
+      "num_tokens": 6543150.0,
+      "step": 412
+    },
+    {
+      "entropy": 0.7496612817049026,
+      "epoch": 3.9004739336492893,
+      "grad_norm": 0.7940666675567627,
+      "learning_rate": 4.394017978101905e-08,
+      "loss": 0.7147549390792847,
+      "mean_token_accuracy": 0.8174268007278442,
+      "num_tokens": 6559353.0,
+      "step": 413
+    },
+    {
+      "entropy": 0.7699935585260391,
+      "epoch": 3.909952606635071,
+      "grad_norm": 0.8074795007705688,
+      "learning_rate": 3.6926279399617236e-08,
+      "loss": 0.7725629210472107,
+      "mean_token_accuracy": 0.8059059679508209,
+      "num_tokens": 6575235.0,
+      "step": 414
+    },
+    {
+      "entropy": 0.6986105889081955,
+      "epoch": 3.919431279620853,
+      "grad_norm": 0.701176106929779,
+      "learning_rate": 3.0520848460765525e-08,
+      "loss": 0.686797022819519,
+      "mean_token_accuracy": 0.8240027725696564,
+      "num_tokens": 6593020.0,
+      "step": 415
+    },
+    {
+      "entropy": 0.7832924127578735,
+      "epoch": 3.9289099526066353,
+      "grad_norm": 0.7602893710136414,
+      "learning_rate": 2.4724278159898863e-08,
+      "loss": 0.778874397277832,
+      "mean_token_accuracy": 0.8027657568454742,
+      "num_tokens": 6610854.0,
+      "step": 416
+    },
+    {
+      "entropy": 0.7910645306110382,
+      "epoch": 3.938388625592417,
+      "grad_norm": 0.759920597076416,
+      "learning_rate": 1.9536922507841227e-08,
+      "loss": 0.7716935873031616,
+      "mean_token_accuracy": 0.8021509498357773,
+      "num_tokens": 6626142.0,
+      "step": 417
+    },
+    {
+      "entropy": 0.8243566155433655,
+      "epoch": 3.947867298578199,
+      "grad_norm": 0.7890955805778503,
+      "learning_rate": 1.495909830917075e-08,
+      "loss": 0.8134855031967163,
+      "mean_token_accuracy": 0.7995802313089371,
+      "num_tokens": 6642353.0,
+      "step": 418
+    },
+    {
+      "entropy": 0.7705054581165314,
+      "epoch": 3.957345971563981,
+      "grad_norm": 0.7859002351760864,
+      "learning_rate": 1.099108514288627e-08,
+      "loss": 0.7462877035140991,
+      "mean_token_accuracy": 0.8144356906414032,
+      "num_tokens": 6658562.0,
+      "step": 419
+    },
+    {
+      "entropy": 0.7886055111885071,
+      "epoch": 3.966824644549763,
+      "grad_norm": 0.7850826382637024,
+      "learning_rate": 7.633125345317682e-09,
+      "loss": 0.8090062737464905,
+      "mean_token_accuracy": 0.798252746462822,
+      "num_tokens": 6674374.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7322007268667221,
+      "epoch": 3.976303317535545,
+      "grad_norm": 0.7463184595108032,
+      "learning_rate": 4.885423995341088e-09,
+      "loss": 0.7309263944625854,
+      "mean_token_accuracy": 0.8136301338672638,
+      "num_tokens": 6690686.0,
+      "step": 421
+    },
+    {
+      "entropy": 0.8167396634817123,
+      "epoch": 3.985781990521327,
+      "grad_norm": 0.8257619738578796,
+      "learning_rate": 2.7481489018410525e-09,
+      "loss": 0.8237687945365906,
+      "mean_token_accuracy": 0.7974095642566681,
+      "num_tokens": 6705420.0,
+      "step": 422
+    },
+    {
+      "entropy": 0.7956383675336838,
+      "epoch": 3.995260663507109,
+      "grad_norm": 0.8490376472473145,
+      "learning_rate": 1.2214305934699078e-09,
+      "loss": 0.7691541910171509,
+      "mean_token_accuracy": 0.8052787631750107,
+      "num_tokens": 6720459.0,
+      "step": 423
+    },
+    {
+      "entropy": 0.7545907199382782,
+      "epoch": 4.0,
+      "grad_norm": 1.2672629356384277,
+      "learning_rate": 3.053623106741288e-10,
+      "loss": 0.7667921781539917,
+      "mean_token_accuracy": 0.7996481955051422,
+      "num_tokens": 6728853.0,
+      "step": 424
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 424,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.7217785105730765e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}