diff --git "a/checkpoint-1899/trainer_state.json" "b/checkpoint-1899/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1899/trainer_state.json"
@@ -0,0 +1,19035 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1899,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.5241050124168396,
+      "epoch": 0.000526592943654555,
+      "grad_norm": 111.6484146118164,
+      "learning_rate": 0.0,
+      "loss": 7.630620002746582,
+      "mean_token_accuracy": 0.16127368062734604,
+      "num_tokens": 12336.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.5055236518383026,
+      "epoch": 0.00105318588730911,
+      "grad_norm": 111.12712097167969,
+      "learning_rate": 5.012531328320802e-07,
+      "loss": 7.332417964935303,
+      "mean_token_accuracy": 0.17185717076063156,
+      "num_tokens": 24672.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.5026952624320984,
+      "epoch": 0.001579778830963665,
+      "grad_norm": 112.11380767822266,
+      "learning_rate": 1.0025062656641603e-06,
+      "loss": 7.473679542541504,
+      "mean_token_accuracy": 0.1643187664449215,
+      "num_tokens": 37008.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.5194158256053925,
+      "epoch": 0.00210637177461822,
+      "grad_norm": 114.23673248291016,
+      "learning_rate": 1.5037593984962406e-06,
+      "loss": 7.426502704620361,
+      "mean_token_accuracy": 0.1571561936289072,
+      "num_tokens": 49344.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.496931940317154,
+      "epoch": 0.0026329647182727752,
+      "grad_norm": 118.40357208251953,
+      "learning_rate": 2.0050125313283207e-06,
+      "loss": 7.894323348999023,
+      "mean_token_accuracy": 0.1548401154577732,
+      "num_tokens": 61680.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.5049863159656525,
+      "epoch": 0.00315955766192733,
+      "grad_norm": 104.21913146972656,
+      "learning_rate": 2.506265664160401e-06,
+      "loss": 6.593452453613281,
+      "mean_token_accuracy": 0.20564104989171028,
+      "num_tokens": 74016.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.4925725162029266,
+      "epoch": 0.003686150605581885,
+      "grad_norm": 114.10234069824219,
+      "learning_rate": 3.007518796992481e-06,
+      "loss": 7.3693318367004395,
+      "mean_token_accuracy": 0.16597539372742176,
+      "num_tokens": 86352.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.5191607773303986,
+      "epoch": 0.00421274354923644,
+      "grad_norm": 115.99723815917969,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 7.358461856842041,
+      "mean_token_accuracy": 0.16679511591792107,
+      "num_tokens": 98688.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.5183644592761993,
+      "epoch": 0.004739336492890996,
+      "grad_norm": 106.1409912109375,
+      "learning_rate": 4.010025062656641e-06,
+      "loss": 6.668917655944824,
+      "mean_token_accuracy": 0.18950679898262024,
+      "num_tokens": 111024.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.5116282999515533,
+      "epoch": 0.0052659294365455505,
+      "grad_norm": 110.57721710205078,
+      "learning_rate": 4.511278195488722e-06,
+      "loss": 6.99241304397583,
+      "mean_token_accuracy": 0.18210845813155174,
+      "num_tokens": 123360.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.5269342362880707,
+      "epoch": 0.005792522380200105,
+      "grad_norm": 120.43307495117188,
+      "learning_rate": 5.012531328320802e-06,
+      "loss": 7.586757659912109,
+      "mean_token_accuracy": 0.1462114453315735,
+      "num_tokens": 135696.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.5252384543418884,
+      "epoch": 0.00631911532385466,
+      "grad_norm": 111.8587875366211,
+      "learning_rate": 5.5137844611528826e-06,
+      "loss": 7.363627910614014,
+      "mean_token_accuracy": 0.17107644863426685,
+      "num_tokens": 148032.0,
+      "step": 12
+    },
+    {
+      "entropy": 1.5224651992321014,
+      "epoch": 0.006845708267509215,
+      "grad_norm": 107.8873291015625,
+      "learning_rate": 6.015037593984962e-06,
+      "loss": 6.356827735900879,
+      "mean_token_accuracy": 0.19671257957816124,
+      "num_tokens": 160368.0,
+      "step": 13
+    },
+    {
+      "entropy": 1.5416274964809418,
+      "epoch": 0.00737230121116377,
+      "grad_norm": 115.19085693359375,
+      "learning_rate": 6.516290726817042e-06,
+      "loss": 6.221941947937012,
+      "mean_token_accuracy": 0.2142542451620102,
+      "num_tokens": 172704.0,
+      "step": 14
+    },
+    {
+      "entropy": 1.5107212960720062,
+      "epoch": 0.007898894154818325,
+      "grad_norm": 110.73648834228516,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 6.397078514099121,
+      "mean_token_accuracy": 0.20885847136378288,
+      "num_tokens": 185040.0,
+      "step": 15
+    },
+    {
+      "entropy": 1.5597798526287079,
+      "epoch": 0.00842548709847288,
+      "grad_norm": 99.72173309326172,
+      "learning_rate": 7.518796992481203e-06,
+      "loss": 5.646764278411865,
+      "mean_token_accuracy": 0.24349310994148254,
+      "num_tokens": 197376.0,
+      "step": 16
+    },
+    {
+      "entropy": 1.5534575581550598,
+      "epoch": 0.008952080042127435,
+      "grad_norm": 104.43334197998047,
+      "learning_rate": 8.020050125313283e-06,
+      "loss": 5.892775535583496,
+      "mean_token_accuracy": 0.22073492035269737,
+      "num_tokens": 209712.0,
+      "step": 17
+    },
+    {
+      "entropy": 1.5676631033420563,
+      "epoch": 0.009478672985781991,
+      "grad_norm": 99.27777099609375,
+      "learning_rate": 8.521303258145363e-06,
+      "loss": 5.239416599273682,
+      "mean_token_accuracy": 0.26043422892689705,
+      "num_tokens": 222048.0,
+      "step": 18
+    },
+    {
+      "entropy": 1.5712612569332123,
+      "epoch": 0.010005265929436546,
+      "grad_norm": 94.56501770019531,
+      "learning_rate": 9.022556390977444e-06,
+      "loss": 5.018033981323242,
+      "mean_token_accuracy": 0.2874941825866699,
+      "num_tokens": 234384.0,
+      "step": 19
+    },
+    {
+      "entropy": 1.5856993794441223,
+      "epoch": 0.010531858873091101,
+      "grad_norm": 103.30012512207031,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 5.112401485443115,
+      "mean_token_accuracy": 0.26686083525419235,
+      "num_tokens": 246720.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.542406052350998,
+      "epoch": 0.011058451816745656,
+      "grad_norm": 73.81751251220703,
+      "learning_rate": 1.0025062656641604e-05,
+      "loss": 3.298715353012085,
+      "mean_token_accuracy": 0.42641731351614,
+      "num_tokens": 259056.0,
+      "step": 21
+    },
+    {
+      "entropy": 1.5376673340797424,
+      "epoch": 0.01158504476040021,
+      "grad_norm": 74.847900390625,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 3.5809030532836914,
+      "mean_token_accuracy": 0.40752222388982773,
+      "num_tokens": 271392.0,
+      "step": 22
+    },
+    {
+      "entropy": 1.548084020614624,
+      "epoch": 0.012111637704054766,
+      "grad_norm": 63.552825927734375,
+      "learning_rate": 1.1027568922305765e-05,
+      "loss": 3.100147247314453,
+      "mean_token_accuracy": 0.45311493426561356,
+      "num_tokens": 283728.0,
+      "step": 23
+    },
+    {
+      "entropy": 1.5953525304794312,
+      "epoch": 0.01263823064770932,
+      "grad_norm": 60.015655517578125,
+      "learning_rate": 1.1528822055137844e-05,
+      "loss": 2.783557176589966,
+      "mean_token_accuracy": 0.47465962916612625,
+      "num_tokens": 296064.0,
+      "step": 24
+    },
+    {
+      "entropy": 1.5308564603328705,
+      "epoch": 0.013164823591363875,
+      "grad_norm": 40.24797821044922,
+      "learning_rate": 1.2030075187969925e-05,
+      "loss": 2.171975612640381,
+      "mean_token_accuracy": 0.5734862834215164,
+      "num_tokens": 308400.0,
+      "step": 25
+    },
+    {
+      "entropy": 1.5149377882480621,
+      "epoch": 0.01369141653501843,
+      "grad_norm": 31.702476501464844,
+      "learning_rate": 1.2531328320802006e-05,
+      "loss": 1.907788634300232,
+      "mean_token_accuracy": 0.6134172528982162,
+      "num_tokens": 320736.0,
+      "step": 26
+    },
+    {
+      "entropy": 1.4917882978916168,
+      "epoch": 0.014218009478672985,
+      "grad_norm": 20.41485023498535,
+      "learning_rate": 1.3032581453634085e-05,
+      "loss": 1.6917057037353516,
+      "mean_token_accuracy": 0.6387842893600464,
+      "num_tokens": 333072.0,
+      "step": 27
+    },
+    {
+      "entropy": 1.4882407486438751,
+      "epoch": 0.01474460242232754,
+      "grad_norm": 15.065500259399414,
+      "learning_rate": 1.3533834586466165e-05,
+      "loss": 1.4968407154083252,
+      "mean_token_accuracy": 0.6677049100399017,
+      "num_tokens": 345408.0,
+      "step": 28
+    },
+    {
+      "entropy": 1.4370096921920776,
+      "epoch": 0.015271195365982097,
+      "grad_norm": 10.740882873535156,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 1.3545355796813965,
+      "mean_token_accuracy": 0.698340117931366,
+      "num_tokens": 357744.0,
+      "step": 29
+    },
+    {
+      "entropy": 1.341864824295044,
+      "epoch": 0.01579778830963665,
+      "grad_norm": 3.3069381713867188,
+      "learning_rate": 1.4536340852130325e-05,
+      "loss": 1.0850293636322021,
+      "mean_token_accuracy": 0.7386660128831863,
+      "num_tokens": 370080.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.4095737040042877,
+      "epoch": 0.016324381253291206,
+      "grad_norm": 2.9421474933624268,
+      "learning_rate": 1.5037593984962406e-05,
+      "loss": 1.1328743696212769,
+      "mean_token_accuracy": 0.7404871881008148,
+      "num_tokens": 382416.0,
+      "step": 31
+    },
+    {
+      "entropy": 1.4024023413658142,
+      "epoch": 0.01685097419694576,
+      "grad_norm": 2.5040762424468994,
+      "learning_rate": 1.5538847117794486e-05,
+      "loss": 1.1228951215744019,
+      "mean_token_accuracy": 0.7428316622972488,
+      "num_tokens": 394752.0,
+      "step": 32
+    },
+    {
+      "entropy": 1.3549567759037018,
+      "epoch": 0.017377567140600316,
+      "grad_norm": 1.7973337173461914,
+      "learning_rate": 1.6040100250626565e-05,
+      "loss": 1.0077903270721436,
+      "mean_token_accuracy": 0.766022652387619,
+      "num_tokens": 407088.0,
+      "step": 33
+    },
+    {
+      "entropy": 1.3836270570755005,
+      "epoch": 0.01790416008425487,
+      "grad_norm": 1.7406286001205444,
+      "learning_rate": 1.6541353383458648e-05,
+      "loss": 1.0222362279891968,
+      "mean_token_accuracy": 0.7626231610774994,
+      "num_tokens": 419424.0,
+      "step": 34
+    },
+    {
+      "entropy": 1.3947510421276093,
+      "epoch": 0.018430753027909426,
+      "grad_norm": 1.8865187168121338,
+      "learning_rate": 1.7042606516290727e-05,
+      "loss": 1.0371894836425781,
+      "mean_token_accuracy": 0.7562028616666794,
+      "num_tokens": 431760.0,
+      "step": 35
+    },
+    {
+      "entropy": 1.3465319573879242,
+      "epoch": 0.018957345971563982,
+      "grad_norm": 1.7011761665344238,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 0.9494946599006653,
+      "mean_token_accuracy": 0.7783531844615936,
+      "num_tokens": 444096.0,
+      "step": 36
+    },
+    {
+      "entropy": 1.4580606818199158,
+      "epoch": 0.019483938915218536,
+      "grad_norm": 1.9021601676940918,
+      "learning_rate": 1.8045112781954888e-05,
+      "loss": 1.0398352146148682,
+      "mean_token_accuracy": 0.7636931091547012,
+      "num_tokens": 456432.0,
+      "step": 37
+    },
+    {
+      "entropy": 1.4095810949802399,
+      "epoch": 0.020010531858873092,
+      "grad_norm": 1.6176142692565918,
+      "learning_rate": 1.8546365914786967e-05,
+      "loss": 0.9648343324661255,
+      "mean_token_accuracy": 0.763045608997345,
+      "num_tokens": 468768.0,
+      "step": 38
+    },
+    {
+      "entropy": 1.4290408492088318,
+      "epoch": 0.020537124802527645,
+      "grad_norm": 1.802105188369751,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.9583951234817505,
+      "mean_token_accuracy": 0.7706119865179062,
+      "num_tokens": 481104.0,
+      "step": 39
+    },
+    {
+      "entropy": 1.4769366383552551,
+      "epoch": 0.021063717746182202,
+      "grad_norm": 1.8512225151062012,
+      "learning_rate": 1.954887218045113e-05,
+      "loss": 1.0114065408706665,
+      "mean_token_accuracy": 0.7521056234836578,
+      "num_tokens": 493440.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.508185237646103,
+      "epoch": 0.021590310689836755,
+      "grad_norm": 2.076702356338501,
+      "learning_rate": 2.0050125313283208e-05,
+      "loss": 0.9636508226394653,
+      "mean_token_accuracy": 0.7661653459072113,
+      "num_tokens": 505776.0,
+      "step": 41
+    },
+    {
+      "entropy": 1.459254652261734,
+      "epoch": 0.022116903633491312,
+      "grad_norm": 2.349229335784912,
+      "learning_rate": 2.0551378446115287e-05,
+      "loss": 0.9206173419952393,
+      "mean_token_accuracy": 0.7662864923477173,
+      "num_tokens": 518112.0,
+      "step": 42
+    },
+    {
+      "entropy": 1.5238048434257507,
+      "epoch": 0.022643496577145865,
+      "grad_norm": 2.185512065887451,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.9314085245132446,
+      "mean_token_accuracy": 0.7649829983711243,
+      "num_tokens": 530448.0,
+      "step": 43
+    },
+    {
+      "entropy": 1.5003608763217926,
+      "epoch": 0.02317008952080042,
+      "grad_norm": 2.0811667442321777,
+      "learning_rate": 2.1553884711779448e-05,
+      "loss": 0.8671697974205017,
+      "mean_token_accuracy": 0.7754391133785248,
+      "num_tokens": 542784.0,
+      "step": 44
+    },
+    {
+      "entropy": 1.591546893119812,
+      "epoch": 0.023696682464454975,
+      "grad_norm": 2.0840373039245605,
+      "learning_rate": 2.205513784461153e-05,
+      "loss": 0.872833251953125,
+      "mean_token_accuracy": 0.7740741670131683,
+      "num_tokens": 555120.0,
+      "step": 45
+    },
+    {
+      "entropy": 1.6416102647781372,
+      "epoch": 0.02422327540810953,
+      "grad_norm": 2.033761501312256,
+      "learning_rate": 2.255639097744361e-05,
+      "loss": 0.8919705152511597,
+      "mean_token_accuracy": 0.7629313617944717,
+      "num_tokens": 567456.0,
+      "step": 46
+    },
+    {
+      "entropy": 1.6393465399742126,
+      "epoch": 0.024749868351764088,
+      "grad_norm": 1.7740504741668701,
+      "learning_rate": 2.3057644110275688e-05,
+      "loss": 0.8117689490318298,
+      "mean_token_accuracy": 0.7816901206970215,
+      "num_tokens": 579792.0,
+      "step": 47
+    },
+    {
+      "entropy": 1.6808819770812988,
+      "epoch": 0.02527646129541864,
+      "grad_norm": 1.4741381406784058,
+      "learning_rate": 2.355889724310777e-05,
+      "loss": 0.8110437393188477,
+      "mean_token_accuracy": 0.7762878388166428,
+      "num_tokens": 592128.0,
+      "step": 48
+    },
+    {
+      "entropy": 1.7813679873943329,
+      "epoch": 0.025803054239073198,
+      "grad_norm": 1.135923147201538,
+      "learning_rate": 2.406015037593985e-05,
+      "loss": 0.8597409129142761,
+      "mean_token_accuracy": 0.7689527124166489,
+      "num_tokens": 604464.0,
+      "step": 49
+    },
+    {
+      "entropy": 1.7125210165977478,
+      "epoch": 0.02632964718272775,
+      "grad_norm": 1.0037126541137695,
+      "learning_rate": 2.456140350877193e-05,
+      "loss": 0.7560303807258606,
+      "mean_token_accuracy": 0.7888963371515274,
+      "num_tokens": 616800.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.7505145072937012,
+      "epoch": 0.026856240126382307,
+      "grad_norm": 0.9906253218650818,
+      "learning_rate": 2.506265664160401e-05,
+      "loss": 0.7666164636611938,
+      "mean_token_accuracy": 0.7822892367839813,
+      "num_tokens": 629136.0,
+      "step": 51
+    },
+    {
+      "entropy": 1.8206515610218048,
+      "epoch": 0.02738283307003686,
+      "grad_norm": 1.0637407302856445,
+      "learning_rate": 2.556390977443609e-05,
+      "loss": 0.7570075988769531,
+      "mean_token_accuracy": 0.7857469767332077,
+      "num_tokens": 641472.0,
+      "step": 52
+    },
+    {
+      "entropy": 1.9179592430591583,
+      "epoch": 0.027909426013691417,
+      "grad_norm": 0.994309663772583,
+      "learning_rate": 2.606516290726817e-05,
+      "loss": 0.8264732956886292,
+      "mean_token_accuracy": 0.7659910470247269,
+      "num_tokens": 653808.0,
+      "step": 53
+    },
+    {
+      "entropy": 1.8192859292030334,
+      "epoch": 0.02843601895734597,
+      "grad_norm": 1.283566951751709,
+      "learning_rate": 2.656641604010025e-05,
+      "loss": 0.7167642712593079,
+      "mean_token_accuracy": 0.7935251444578171,
+      "num_tokens": 666144.0,
+      "step": 54
+    },
+    {
+      "entropy": 1.8640032410621643,
+      "epoch": 0.028962611901000527,
+      "grad_norm": 0.993376612663269,
+      "learning_rate": 2.706766917293233e-05,
+      "loss": 0.732735276222229,
+      "mean_token_accuracy": 0.7889553904533386,
+      "num_tokens": 678480.0,
+      "step": 55
+    },
+    {
+      "entropy": 1.9309577941894531,
+      "epoch": 0.02948920484465508,
+      "grad_norm": 0.952996551990509,
+      "learning_rate": 2.756892230576441e-05,
+      "loss": 0.7183045148849487,
+      "mean_token_accuracy": 0.7865996956825256,
+      "num_tokens": 690816.0,
+      "step": 56
+    },
+    {
+      "entropy": 1.9969252347946167,
+      "epoch": 0.030015797788309637,
+      "grad_norm": 0.9141530990600586,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.7127647399902344,
+      "mean_token_accuracy": 0.7965414077043533,
+      "num_tokens": 703152.0,
+      "step": 57
+    },
+    {
+      "entropy": 1.9162320494651794,
+      "epoch": 0.030542390731964193,
+      "grad_norm": 1.0311601161956787,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.6988133788108826,
+      "mean_token_accuracy": 0.7964022308588028,
+      "num_tokens": 715488.0,
+      "step": 58
+    },
+    {
+      "entropy": 1.9365825355052948,
+      "epoch": 0.031068983675618746,
+      "grad_norm": 0.8333176374435425,
+      "learning_rate": 2.907268170426065e-05,
+      "loss": 0.6844547986984253,
+      "mean_token_accuracy": 0.8021096438169479,
+      "num_tokens": 727824.0,
+      "step": 59
+    },
+    {
+      "entropy": 1.9246950447559357,
+      "epoch": 0.0315955766192733,
+      "grad_norm": 0.8664101362228394,
+      "learning_rate": 2.9573934837092732e-05,
+      "loss": 0.6528723239898682,
+      "mean_token_accuracy": 0.8053467720746994,
+      "num_tokens": 740160.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.929794728755951,
+      "epoch": 0.032122169562927856,
+      "grad_norm": 0.9577063918113708,
+      "learning_rate": 3.007518796992481e-05,
+      "loss": 0.678918719291687,
+      "mean_token_accuracy": 0.7989283800125122,
+      "num_tokens": 752496.0,
+      "step": 61
+    },
+    {
+      "entropy": 1.826391577720642,
+      "epoch": 0.03264876250658241,
+      "grad_norm": 0.8492108583450317,
+      "learning_rate": 3.0576441102756894e-05,
+      "loss": 0.6723799109458923,
+      "mean_token_accuracy": 0.7983379364013672,
+      "num_tokens": 764832.0,
+      "step": 62
+    },
+    {
+      "entropy": 1.9241991639137268,
+      "epoch": 0.03317535545023697,
+      "grad_norm": 0.928464412689209,
+      "learning_rate": 3.107769423558897e-05,
+      "loss": 0.6530962586402893,
+      "mean_token_accuracy": 0.8008216470479965,
+      "num_tokens": 777168.0,
+      "step": 63
+    },
+    {
+      "entropy": 1.8967448770999908,
+      "epoch": 0.03370194839389152,
+      "grad_norm": 1.0670896768569946,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.6292267441749573,
+      "mean_token_accuracy": 0.8003135621547699,
+      "num_tokens": 789504.0,
+      "step": 64
+    },
+    {
+      "entropy": 1.8847980499267578,
+      "epoch": 0.034228541337546076,
+      "grad_norm": 1.0110414028167725,
+      "learning_rate": 3.208020050125313e-05,
+      "loss": 0.6133747100830078,
+      "mean_token_accuracy": 0.8148374855518341,
+      "num_tokens": 801840.0,
+      "step": 65
+    },
+    {
+      "entropy": 1.930319905281067,
+      "epoch": 0.03475513428120063,
+      "grad_norm": 1.023224949836731,
+      "learning_rate": 3.258145363408521e-05,
+      "loss": 0.6123053431510925,
+      "mean_token_accuracy": 0.8167780637741089,
+      "num_tokens": 814176.0,
+      "step": 66
+    },
+    {
+      "entropy": 1.9471323192119598,
+      "epoch": 0.03528172722485519,
+      "grad_norm": 1.1810884475708008,
+      "learning_rate": 3.3082706766917295e-05,
+      "loss": 0.6039740443229675,
+      "mean_token_accuracy": 0.8179137706756592,
+      "num_tokens": 826512.0,
+      "step": 67
+    },
+    {
+      "entropy": 1.8890258967876434,
+      "epoch": 0.03580832016850974,
+      "grad_norm": 0.9907887578010559,
+      "learning_rate": 3.3583959899749374e-05,
+      "loss": 0.599247932434082,
+      "mean_token_accuracy": 0.8207112550735474,
+      "num_tokens": 838848.0,
+      "step": 68
+    },
+    {
+      "entropy": 1.9326116144657135,
+      "epoch": 0.036334913112164295,
+      "grad_norm": 1.0037294626235962,
+      "learning_rate": 3.4085213032581453e-05,
+      "loss": 0.5630755424499512,
+      "mean_token_accuracy": 0.8216072916984558,
+      "num_tokens": 851184.0,
+      "step": 69
+    },
+    {
+      "entropy": 1.925311028957367,
+      "epoch": 0.03686150605581885,
+      "grad_norm": 1.1875734329223633,
+      "learning_rate": 3.458646616541353e-05,
+      "loss": 0.5733011364936829,
+      "mean_token_accuracy": 0.8164849728345871,
+      "num_tokens": 863520.0,
+      "step": 70
+    },
+    {
+      "entropy": 2.051402121782303,
+      "epoch": 0.03738809899947341,
+      "grad_norm": 1.162881851196289,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.550703763961792,
+      "mean_token_accuracy": 0.8178133964538574,
+      "num_tokens": 875856.0,
+      "step": 71
+    },
+    {
+      "entropy": 1.9634678363800049,
+      "epoch": 0.037914691943127965,
+      "grad_norm": 0.9865695834159851,
+      "learning_rate": 3.55889724310777e-05,
+      "loss": 0.5769374966621399,
+      "mean_token_accuracy": 0.8114093840122223,
+      "num_tokens": 888192.0,
+      "step": 72
+    },
+    {
+      "entropy": 1.9920983016490936,
+      "epoch": 0.038441284886782515,
+      "grad_norm": 1.0162739753723145,
+      "learning_rate": 3.6090225563909776e-05,
+      "loss": 0.5818509459495544,
+      "mean_token_accuracy": 0.8134536445140839,
+      "num_tokens": 900528.0,
+      "step": 73
+    },
+    {
+      "entropy": 1.9029254913330078,
+      "epoch": 0.03896787783043707,
+      "grad_norm": 0.9538004398345947,
+      "learning_rate": 3.6591478696741855e-05,
+      "loss": 0.5178959369659424,
+      "mean_token_accuracy": 0.8257192969322205,
+      "num_tokens": 912864.0,
+      "step": 74
+    },
+    {
+      "entropy": 1.967795580625534,
+      "epoch": 0.03949447077409163,
+      "grad_norm": 0.9578227996826172,
+      "learning_rate": 3.7092731829573934e-05,
+      "loss": 0.5342862010002136,
+      "mean_token_accuracy": 0.8278465121984482,
+      "num_tokens": 925200.0,
+      "step": 75
+    },
+    {
+      "entropy": 1.9465965926647186,
+      "epoch": 0.040021063717746184,
+      "grad_norm": 1.0253666639328003,
+      "learning_rate": 3.759398496240601e-05,
+      "loss": 0.5030800104141235,
+      "mean_token_accuracy": 0.83804552257061,
+      "num_tokens": 937536.0,
+      "step": 76
+    },
+    {
+      "entropy": 1.8636894226074219,
+      "epoch": 0.040547656661400734,
+      "grad_norm": 0.8184434771537781,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.45680901408195496,
+      "mean_token_accuracy": 0.8485544174909592,
+      "num_tokens": 949872.0,
+      "step": 77
+    },
+    {
+      "entropy": 1.928781270980835,
+      "epoch": 0.04107424960505529,
+      "grad_norm": 0.8536883592605591,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.4830615818500519,
+      "mean_token_accuracy": 0.8397646248340607,
+      "num_tokens": 962208.0,
+      "step": 78
+    },
+    {
+      "entropy": 1.8684503734111786,
+      "epoch": 0.04160084254870985,
+      "grad_norm": 0.868567168712616,
+      "learning_rate": 3.909774436090226e-05,
+      "loss": 0.49555566906929016,
+      "mean_token_accuracy": 0.8393678069114685,
+      "num_tokens": 974544.0,
+      "step": 79
+    },
+    {
+      "entropy": 1.832389086484909,
+      "epoch": 0.042127435492364404,
+      "grad_norm": 0.7834892868995667,
+      "learning_rate": 3.9598997493734336e-05,
+      "loss": 0.4530882239341736,
+      "mean_token_accuracy": 0.8548341691493988,
+      "num_tokens": 986880.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.8249096870422363,
+      "epoch": 0.04265402843601896,
+      "grad_norm": 0.7752470374107361,
+      "learning_rate": 4.0100250626566415e-05,
+      "loss": 0.4578617215156555,
+      "mean_token_accuracy": 0.8508521616458893,
+      "num_tokens": 999216.0,
+      "step": 81
+    },
+    {
+      "entropy": 1.9078024625778198,
+      "epoch": 0.04318062137967351,
+      "grad_norm": 0.7781356573104858,
+      "learning_rate": 4.0601503759398494e-05,
+      "loss": 0.43578821420669556,
+      "mean_token_accuracy": 0.8600046187639236,
+      "num_tokens": 1011552.0,
+      "step": 82
+    },
+    {
+      "entropy": 1.8074560463428497,
+      "epoch": 0.04370721432332807,
+      "grad_norm": 0.7703754305839539,
+      "learning_rate": 4.110275689223057e-05,
+      "loss": 0.4543125629425049,
+      "mean_token_accuracy": 0.8561971783638,
+      "num_tokens": 1023888.0,
+      "step": 83
+    },
+    {
+      "entropy": 1.7470267415046692,
+      "epoch": 0.044233807266982623,
+      "grad_norm": 0.6839037537574768,
+      "learning_rate": 4.160401002506266e-05,
+      "loss": 0.4043883681297302,
+      "mean_token_accuracy": 0.8632047474384308,
+      "num_tokens": 1036224.0,
+      "step": 84
+    },
+    {
+      "entropy": 1.8056954443454742,
+      "epoch": 0.04476040021063718,
+      "grad_norm": 0.8745210766792297,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.4576418101787567,
+      "mean_token_accuracy": 0.8515569567680359,
+      "num_tokens": 1048560.0,
+      "step": 85
+    },
+    {
+      "entropy": 1.7271961271762848,
+      "epoch": 0.04528699315429173,
+      "grad_norm": 0.7987569570541382,
+      "learning_rate": 4.260651629072682e-05,
+      "loss": 0.4062059819698334,
+      "mean_token_accuracy": 0.8644601851701736,
+      "num_tokens": 1060896.0,
+      "step": 86
+    },
+    {
+      "entropy": 1.7829216122627258,
+      "epoch": 0.045813586097946286,
+      "grad_norm": 0.8260216116905212,
+      "learning_rate": 4.3107769423558896e-05,
+      "loss": 0.4285086989402771,
+      "mean_token_accuracy": 0.8619498312473297,
+      "num_tokens": 1073232.0,
+      "step": 87
+    },
+    {
+      "entropy": 1.720045566558838,
+      "epoch": 0.04634017904160084,
+      "grad_norm": 0.8638037443161011,
+      "learning_rate": 4.3609022556390975e-05,
+      "loss": 0.42514118552207947,
+      "mean_token_accuracy": 0.8556364923715591,
+      "num_tokens": 1085568.0,
+      "step": 88
+    },
+    {
+      "entropy": 1.749743402004242,
+      "epoch": 0.0468667719852554,
+      "grad_norm": 0.6750407218933105,
+      "learning_rate": 4.411027568922306e-05,
+      "loss": 0.3981553912162781,
+      "mean_token_accuracy": 0.8708555996417999,
+      "num_tokens": 1097904.0,
+      "step": 89
+    },
+    {
+      "entropy": 1.785173624753952,
+      "epoch": 0.04739336492890995,
+      "grad_norm": 0.7802124619483948,
+      "learning_rate": 4.461152882205514e-05,
+      "loss": 0.3845086097717285,
+      "mean_token_accuracy": 0.8747637271881104,
+      "num_tokens": 1110240.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.7087633907794952,
+      "epoch": 0.047919957872564506,
+      "grad_norm": 0.8488918542861938,
+      "learning_rate": 4.511278195488722e-05,
+      "loss": 0.3644571006298065,
+      "mean_token_accuracy": 0.8757548332214355,
+      "num_tokens": 1122576.0,
+      "step": 91
+    },
+    {
+      "entropy": 1.6821248829364777,
+      "epoch": 0.04844655081621906,
+      "grad_norm": 0.7744936943054199,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 0.40767285227775574,
+      "mean_token_accuracy": 0.8596755862236023,
+      "num_tokens": 1134912.0,
+      "step": 92
+    },
+    {
+      "entropy": 1.7099383771419525,
+      "epoch": 0.04897314375987362,
+      "grad_norm": 0.9390520453453064,
+      "learning_rate": 4.6115288220551377e-05,
+      "loss": 0.4450380802154541,
+      "mean_token_accuracy": 0.8517685830593109,
+      "num_tokens": 1147248.0,
+      "step": 93
+    },
+    {
+      "entropy": 1.6644540429115295,
+      "epoch": 0.049499736703528176,
+      "grad_norm": 0.9382246136665344,
+      "learning_rate": 4.6616541353383456e-05,
+      "loss": 0.37574899196624756,
+      "mean_token_accuracy": 0.8736520707607269,
+      "num_tokens": 1159584.0,
+      "step": 94
+    },
+    {
+      "entropy": 1.7045432031154633,
+      "epoch": 0.050026329647182725,
+      "grad_norm": 0.784248948097229,
+      "learning_rate": 4.711779448621554e-05,
+      "loss": 0.39656203985214233,
+      "mean_token_accuracy": 0.8673653900623322,
+      "num_tokens": 1171920.0,
+      "step": 95
+    },
+    {
+      "entropy": 1.6858630180358887,
+      "epoch": 0.05055292259083728,
+      "grad_norm": 0.7153828144073486,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.3821033835411072,
+      "mean_token_accuracy": 0.870804488658905,
+      "num_tokens": 1184256.0,
+      "step": 96
+    },
+    {
+      "entropy": 1.5798078775405884,
+      "epoch": 0.05107951553449184,
+      "grad_norm": 0.6857785582542419,
+      "learning_rate": 4.81203007518797e-05,
+      "loss": 0.3593214452266693,
+      "mean_token_accuracy": 0.8746844828128815,
+      "num_tokens": 1196592.0,
+      "step": 97
+    },
+    {
+      "entropy": 1.670957088470459,
+      "epoch": 0.051606108478146395,
+      "grad_norm": 0.7506043910980225,
+      "learning_rate": 4.862155388471178e-05,
+      "loss": 0.3727911412715912,
+      "mean_token_accuracy": 0.8756437748670578,
+      "num_tokens": 1208928.0,
+      "step": 98
+    },
+    {
+      "entropy": 1.5233455896377563,
+      "epoch": 0.052132701421800945,
+      "grad_norm": 0.7735759019851685,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 0.37093132734298706,
+      "mean_token_accuracy": 0.8704032748937607,
+      "num_tokens": 1221264.0,
+      "step": 99
+    },
+    {
+      "entropy": 1.6192973852157593,
+      "epoch": 0.0526592943654555,
+      "grad_norm": 0.7478036880493164,
+      "learning_rate": 4.9624060150375936e-05,
+      "loss": 0.3426092863082886,
+      "mean_token_accuracy": 0.8861435055732727,
+      "num_tokens": 1233600.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.493833750486374,
+      "epoch": 0.05318588730911006,
+      "grad_norm": 0.7044211030006409,
+      "learning_rate": 5.012531328320802e-05,
+      "loss": 0.35205507278442383,
+      "mean_token_accuracy": 0.8778510093688965,
+      "num_tokens": 1245936.0,
+      "step": 101
+    },
+    {
+      "entropy": 1.4939774870872498,
+      "epoch": 0.053712480252764615,
+      "grad_norm": 0.7720161080360413,
+      "learning_rate": 5.06265664160401e-05,
+      "loss": 0.3975054621696472,
+      "mean_token_accuracy": 0.8641460090875626,
+      "num_tokens": 1258272.0,
+      "step": 102
+    },
+    {
+      "entropy": 1.6101951897144318,
+      "epoch": 0.05423907319641917,
+      "grad_norm": 0.669315755367279,
+      "learning_rate": 5.112781954887218e-05,
+      "loss": 0.34713947772979736,
+      "mean_token_accuracy": 0.8790434300899506,
+      "num_tokens": 1270608.0,
+      "step": 103
+    },
+    {
+      "entropy": 1.5196483135223389,
+      "epoch": 0.05476566614007372,
+      "grad_norm": 0.8516085743904114,
+      "learning_rate": 5.162907268170426e-05,
+      "loss": 0.3601211607456207,
+      "mean_token_accuracy": 0.8756401091814041,
+      "num_tokens": 1282944.0,
+      "step": 104
+    },
+    {
+      "entropy": 1.4721749424934387,
+      "epoch": 0.05529225908372828,
+      "grad_norm": 0.8156314492225647,
+      "learning_rate": 5.213032581453634e-05,
+      "loss": 0.3697987198829651,
+      "mean_token_accuracy": 0.8728637546300888,
+      "num_tokens": 1295280.0,
+      "step": 105
+    },
+    {
+      "entropy": 1.4663786590099335,
+      "epoch": 0.055818852027382834,
+      "grad_norm": 0.6453195810317993,
+      "learning_rate": 5.2631578947368424e-05,
+      "loss": 0.3338157534599304,
+      "mean_token_accuracy": 0.8864734172821045,
+      "num_tokens": 1307616.0,
+      "step": 106
+    },
+    {
+      "entropy": 1.497151494026184,
+      "epoch": 0.05634544497103739,
+      "grad_norm": 0.7146667242050171,
+      "learning_rate": 5.31328320802005e-05,
+      "loss": 0.34923166036605835,
+      "mean_token_accuracy": 0.8819292187690735,
+      "num_tokens": 1319952.0,
+      "step": 107
+    },
+    {
+      "entropy": 1.5161625444889069,
+      "epoch": 0.05687203791469194,
+      "grad_norm": 0.8290141820907593,
+      "learning_rate": 5.363408521303258e-05,
+      "loss": 0.3371197581291199,
+      "mean_token_accuracy": 0.8877773582935333,
+      "num_tokens": 1332288.0,
+      "step": 108
+    },
+    {
+      "entropy": 1.4333459436893463,
+      "epoch": 0.0573986308583465,
+      "grad_norm": 0.8022440671920776,
+      "learning_rate": 5.413533834586466e-05,
+      "loss": 0.31702953577041626,
+      "mean_token_accuracy": 0.8909593671560287,
+      "num_tokens": 1344624.0,
+      "step": 109
+    },
+    {
+      "entropy": 1.4444147646427155,
+      "epoch": 0.057925223802001054,
+      "grad_norm": 0.7960955500602722,
+      "learning_rate": 5.463659147869674e-05,
+      "loss": 0.3525749742984772,
+      "mean_token_accuracy": 0.8755318969488144,
+      "num_tokens": 1356960.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.469768613576889,
+      "epoch": 0.05845181674565561,
+      "grad_norm": 0.8183742165565491,
+      "learning_rate": 5.513784461152882e-05,
+      "loss": 0.3398601710796356,
+      "mean_token_accuracy": 0.8840423971414566,
+      "num_tokens": 1369296.0,
+      "step": 111
+    },
+    {
+      "entropy": 1.4481310844421387,
+      "epoch": 0.05897840968931016,
+      "grad_norm": 0.6801962852478027,
+      "learning_rate": 5.5639097744360905e-05,
+      "loss": 0.34869900345802307,
+      "mean_token_accuracy": 0.877396434545517,
+      "num_tokens": 1381632.0,
+      "step": 112
+    },
+    {
+      "entropy": 1.4586880207061768,
+      "epoch": 0.05950500263296472,
+      "grad_norm": 0.7105273008346558,
+      "learning_rate": 5.6140350877192984e-05,
+      "loss": 0.32431477308273315,
+      "mean_token_accuracy": 0.8829907327890396,
+      "num_tokens": 1393968.0,
+      "step": 113
+    },
+    {
+      "entropy": 1.4191592037677765,
+      "epoch": 0.06003159557661927,
+      "grad_norm": 0.8621596097946167,
+      "learning_rate": 5.664160401002506e-05,
+      "loss": 0.29889625310897827,
+      "mean_token_accuracy": 0.8882533758878708,
+      "num_tokens": 1406304.0,
+      "step": 114
+    },
+    {
+      "entropy": 1.431228756904602,
+      "epoch": 0.06055818852027383,
+      "grad_norm": 0.662341833114624,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.34662240743637085,
+      "mean_token_accuracy": 0.8789393156766891,
+      "num_tokens": 1418640.0,
+      "step": 115
+    },
+    {
+      "entropy": 1.317075937986374,
+      "epoch": 0.061084781463928386,
+      "grad_norm": 0.6940162777900696,
+      "learning_rate": 5.764411027568922e-05,
+      "loss": 0.3154875636100769,
+      "mean_token_accuracy": 0.886825367808342,
+      "num_tokens": 1430976.0,
+      "step": 116
+    },
+    {
+      "entropy": 1.3747568726539612,
+      "epoch": 0.061611374407582936,
+      "grad_norm": 0.6974264979362488,
+      "learning_rate": 5.81453634085213e-05,
+      "loss": 0.36218538880348206,
+      "mean_token_accuracy": 0.8753509372472763,
+      "num_tokens": 1443312.0,
+      "step": 117
+    },
+    {
+      "entropy": 1.3549829423427582,
+      "epoch": 0.06213796735123749,
+      "grad_norm": 0.7157356142997742,
+      "learning_rate": 5.8646616541353386e-05,
+      "loss": 0.32229024171829224,
+      "mean_token_accuracy": 0.8855572938919067,
+      "num_tokens": 1455648.0,
+      "step": 118
+    },
+    {
+      "entropy": 1.4570170640945435,
+      "epoch": 0.06266456029489205,
+      "grad_norm": 0.7710041999816895,
+      "learning_rate": 5.9147869674185465e-05,
+      "loss": 0.3285849094390869,
+      "mean_token_accuracy": 0.8840549886226654,
+      "num_tokens": 1467984.0,
+      "step": 119
+    },
+    {
+      "entropy": 1.4027460813522339,
+      "epoch": 0.0631911532385466,
+      "grad_norm": 0.6999377012252808,
+      "learning_rate": 5.9649122807017544e-05,
+      "loss": 0.3464536666870117,
+      "mean_token_accuracy": 0.8799307644367218,
+      "num_tokens": 1480320.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.3301794826984406,
+      "epoch": 0.06371774618220116,
+      "grad_norm": 0.7836945056915283,
+      "learning_rate": 6.015037593984962e-05,
+      "loss": 0.3436964750289917,
+      "mean_token_accuracy": 0.8764877319335938,
+      "num_tokens": 1492656.0,
+      "step": 121
+    },
+    {
+      "entropy": 1.3585115671157837,
+      "epoch": 0.06424433912585571,
+      "grad_norm": 0.7850233316421509,
+      "learning_rate": 6.06516290726817e-05,
+      "loss": 0.35563230514526367,
+      "mean_token_accuracy": 0.8757744282484055,
+      "num_tokens": 1504992.0,
+      "step": 122
+    },
+    {
+      "entropy": 1.2973756194114685,
+      "epoch": 0.06477093206951026,
+      "grad_norm": 0.7351557612419128,
+      "learning_rate": 6.115288220551379e-05,
+      "loss": 0.3156384527683258,
+      "mean_token_accuracy": 0.886470839381218,
+      "num_tokens": 1517328.0,
+      "step": 123
+    },
+    {
+      "entropy": 1.3591448068618774,
+      "epoch": 0.06529752501316483,
+      "grad_norm": 0.7262402176856995,
+      "learning_rate": 6.165413533834587e-05,
+      "loss": 0.27420130372047424,
+      "mean_token_accuracy": 0.904089480638504,
+      "num_tokens": 1529664.0,
+      "step": 124
+    },
+    {
+      "entropy": 1.253787875175476,
+      "epoch": 0.06582411795681938,
+      "grad_norm": 0.6739073991775513,
+      "learning_rate": 6.215538847117795e-05,
+      "loss": 0.34957295656204224,
+      "mean_token_accuracy": 0.879968523979187,
+      "num_tokens": 1542000.0,
+      "step": 125
+    },
+    {
+      "entropy": 1.261105090379715,
+      "epoch": 0.06635071090047394,
+      "grad_norm": 0.7828674912452698,
+      "learning_rate": 6.265664160401002e-05,
+      "loss": 0.3064384460449219,
+      "mean_token_accuracy": 0.8853085935115814,
+      "num_tokens": 1554336.0,
+      "step": 126
+    },
+    {
+      "entropy": 1.2267013192176819,
+      "epoch": 0.06687730384412849,
+      "grad_norm": 0.9648973345756531,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 0.3406292498111725,
+      "mean_token_accuracy": 0.8785828351974487,
+      "num_tokens": 1566672.0,
+      "step": 127
+    },
+    {
+      "entropy": 1.2064116597175598,
+      "epoch": 0.06740389678778304,
+      "grad_norm": 0.8842466473579407,
+      "learning_rate": 6.365914786967418e-05,
+      "loss": 0.3578197658061981,
+      "mean_token_accuracy": 0.8797773271799088,
+      "num_tokens": 1579008.0,
+      "step": 128
+    },
+    {
+      "entropy": 1.1993357837200165,
+      "epoch": 0.0679304897314376,
+      "grad_norm": 0.8396037817001343,
+      "learning_rate": 6.416040100250626e-05,
+      "loss": 0.32377177476882935,
+      "mean_token_accuracy": 0.8842920362949371,
+      "num_tokens": 1591344.0,
+      "step": 129
+    },
+    {
+      "entropy": 1.2047476172447205,
+      "epoch": 0.06845708267509215,
+      "grad_norm": 0.6860671043395996,
+      "learning_rate": 6.466165413533834e-05,
+      "loss": 0.29577356576919556,
+      "mean_token_accuracy": 0.896445706486702,
+      "num_tokens": 1603680.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2046065032482147,
+      "epoch": 0.06898367561874671,
+      "grad_norm": 0.6543269157409668,
+      "learning_rate": 6.516290726817042e-05,
+      "loss": 0.2779792845249176,
+      "mean_token_accuracy": 0.9030982404947281,
+      "num_tokens": 1616016.0,
+      "step": 131
+    },
+    {
+      "entropy": 1.1669755578041077,
+      "epoch": 0.06951026856240126,
+      "grad_norm": 0.7457153797149658,
+      "learning_rate": 6.566416040100251e-05,
+      "loss": 0.3149452209472656,
+      "mean_token_accuracy": 0.8876487761735916,
+      "num_tokens": 1628352.0,
+      "step": 132
+    },
+    {
+      "entropy": 1.2609427571296692,
+      "epoch": 0.07003686150605581,
+      "grad_norm": 0.7485870122909546,
+      "learning_rate": 6.616541353383459e-05,
+      "loss": 0.31902027130126953,
+      "mean_token_accuracy": 0.8813460171222687,
+      "num_tokens": 1640688.0,
+      "step": 133
+    },
+    {
+      "entropy": 1.2575346231460571,
+      "epoch": 0.07056345444971038,
+      "grad_norm": 0.7213554382324219,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.2797701358795166,
+      "mean_token_accuracy": 0.8987373113632202,
+      "num_tokens": 1653024.0,
+      "step": 134
+    },
+    {
+      "entropy": 1.186894565820694,
+      "epoch": 0.07109004739336493,
+      "grad_norm": 0.7020873427391052,
+      "learning_rate": 6.716791979949875e-05,
+      "loss": 0.2995363771915436,
+      "mean_token_accuracy": 0.8897402882575989,
+      "num_tokens": 1665360.0,
+      "step": 135
+    },
+    {
+      "entropy": 1.161018431186676,
+      "epoch": 0.07161664033701948,
+      "grad_norm": 0.7155889868736267,
+      "learning_rate": 6.766917293233083e-05,
+      "loss": 0.2925366759300232,
+      "mean_token_accuracy": 0.8951611965894699,
+      "num_tokens": 1677696.0,
+      "step": 136
+    },
+    {
+      "entropy": 1.2650261223316193,
+      "epoch": 0.07214323328067404,
+      "grad_norm": 0.7642542123794556,
+      "learning_rate": 6.817042606516291e-05,
+      "loss": 0.2801384925842285,
+      "mean_token_accuracy": 0.8974052667617798,
+      "num_tokens": 1690032.0,
+      "step": 137
+    },
+    {
+      "entropy": 1.2530198991298676,
+      "epoch": 0.07266982622432859,
+      "grad_norm": 0.8588266372680664,
+      "learning_rate": 6.867167919799499e-05,
+      "loss": 0.3360725939273834,
+      "mean_token_accuracy": 0.8794126808643341,
+      "num_tokens": 1702368.0,
+      "step": 138
+    },
+    {
+      "entropy": 1.2518652081489563,
+      "epoch": 0.07319641916798315,
+      "grad_norm": 0.6655072569847107,
+      "learning_rate": 6.917293233082706e-05,
+      "loss": 0.2788892388343811,
+      "mean_token_accuracy": 0.8994075357913971,
+      "num_tokens": 1714704.0,
+      "step": 139
+    },
+    {
+      "entropy": 1.1829611957073212,
+      "epoch": 0.0737230121116377,
+      "grad_norm": 0.7172735333442688,
+      "learning_rate": 6.967418546365914e-05,
+      "loss": 0.2918061316013336,
+      "mean_token_accuracy": 0.8959014713764191,
+      "num_tokens": 1727040.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2773233950138092,
+      "epoch": 0.07424960505529225,
+      "grad_norm": 0.697880208492279,
+      "learning_rate": 7.017543859649122e-05,
+      "loss": 0.3217264711856842,
+      "mean_token_accuracy": 0.8819008469581604,
+      "num_tokens": 1739376.0,
+      "step": 141
+    },
+    {
+      "entropy": 1.2314070761203766,
+      "epoch": 0.07477619799894682,
+      "grad_norm": 0.701502799987793,
+      "learning_rate": 7.06766917293233e-05,
+      "loss": 0.29351311922073364,
+      "mean_token_accuracy": 0.8908087462186813,
+      "num_tokens": 1751712.0,
+      "step": 142
+    },
+    {
+      "entropy": 1.2544872462749481,
+      "epoch": 0.07530279094260137,
+      "grad_norm": 0.7028551697731018,
+      "learning_rate": 7.11779448621554e-05,
+      "loss": 0.3528595566749573,
+      "mean_token_accuracy": 0.8716742098331451,
+      "num_tokens": 1764048.0,
+      "step": 143
+    },
+    {
+      "entropy": 1.1984558403491974,
+      "epoch": 0.07582938388625593,
+      "grad_norm": 0.6910477876663208,
+      "learning_rate": 7.167919799498747e-05,
+      "loss": 0.3136734366416931,
+      "mean_token_accuracy": 0.8870812356472015,
+      "num_tokens": 1776384.0,
+      "step": 144
+    },
+    {
+      "entropy": 1.2033996284008026,
+      "epoch": 0.07635597682991048,
+      "grad_norm": 0.7239225506782532,
+      "learning_rate": 7.218045112781955e-05,
+      "loss": 0.2979337275028229,
+      "mean_token_accuracy": 0.8915670961141586,
+      "num_tokens": 1788720.0,
+      "step": 145
+    },
+    {
+      "entropy": 1.1562331318855286,
+      "epoch": 0.07688256977356503,
+      "grad_norm": 0.7481216192245483,
+      "learning_rate": 7.268170426065163e-05,
+      "loss": 0.2653358280658722,
+      "mean_token_accuracy": 0.8980085849761963,
+      "num_tokens": 1801056.0,
+      "step": 146
+    },
+    {
+      "entropy": 1.2064136862754822,
+      "epoch": 0.07740916271721959,
+      "grad_norm": 0.7776615023612976,
+      "learning_rate": 7.318295739348371e-05,
+      "loss": 0.2848292589187622,
+      "mean_token_accuracy": 0.8949488550424576,
+      "num_tokens": 1813392.0,
+      "step": 147
+    },
+    {
+      "entropy": 1.1387107968330383,
+      "epoch": 0.07793575566087414,
+      "grad_norm": 0.8243488669395447,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.31445616483688354,
+      "mean_token_accuracy": 0.887312263250351,
+      "num_tokens": 1825728.0,
+      "step": 148
+    },
+    {
+      "entropy": 1.0903229415416718,
+      "epoch": 0.0784623486045287,
+      "grad_norm": 0.775776207447052,
+      "learning_rate": 7.418546365914787e-05,
+      "loss": 0.2984585762023926,
+      "mean_token_accuracy": 0.8886806517839432,
+      "num_tokens": 1838064.0,
+      "step": 149
+    },
+    {
+      "entropy": 1.1340011656284332,
+      "epoch": 0.07898894154818326,
+      "grad_norm": 0.7941513657569885,
+      "learning_rate": 7.468671679197995e-05,
+      "loss": 0.27699777483940125,
+      "mean_token_accuracy": 0.9033427834510803,
+      "num_tokens": 1850400.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.084587723016739,
+      "epoch": 0.0795155344918378,
+      "grad_norm": 0.7044249773025513,
+      "learning_rate": 7.518796992481203e-05,
+      "loss": 0.2815166711807251,
+      "mean_token_accuracy": 0.8928717374801636,
+      "num_tokens": 1862736.0,
+      "step": 151
+    },
+    {
+      "entropy": 1.0816469937562943,
+      "epoch": 0.08004212743549237,
+      "grad_norm": 0.8506439328193665,
+      "learning_rate": 7.56892230576441e-05,
+      "loss": 0.3135736584663391,
+      "mean_token_accuracy": 0.8835454434156418,
+      "num_tokens": 1875072.0,
+      "step": 152
+    },
+    {
+      "entropy": 1.037431538105011,
+      "epoch": 0.08056872037914692,
+      "grad_norm": 0.6357259750366211,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 0.2715918719768524,
+      "mean_token_accuracy": 0.9003684967756271,
+      "num_tokens": 1887408.0,
+      "step": 153
+    },
+    {
+      "entropy": 1.101006418466568,
+      "epoch": 0.08109531332280147,
+      "grad_norm": 0.7176199555397034,
+      "learning_rate": 7.669172932330826e-05,
+      "loss": 0.3140455186367035,
+      "mean_token_accuracy": 0.8868316859006882,
+      "num_tokens": 1899744.0,
+      "step": 154
+    },
+    {
+      "entropy": 1.048825979232788,
+      "epoch": 0.08162190626645603,
+      "grad_norm": 0.8264813423156738,
+      "learning_rate": 7.719298245614036e-05,
+      "loss": 0.27318042516708374,
+      "mean_token_accuracy": 0.8995131105184555,
+      "num_tokens": 1912080.0,
+      "step": 155
+    },
+    {
+      "entropy": 1.0369156897068024,
+      "epoch": 0.08214849921011058,
+      "grad_norm": 0.7105727791786194,
+      "learning_rate": 7.769423558897244e-05,
+      "loss": 0.287497341632843,
+      "mean_token_accuracy": 0.8940722495317459,
+      "num_tokens": 1924416.0,
+      "step": 156
+    },
+    {
+      "entropy": 1.0527066588401794,
+      "epoch": 0.08267509215376515,
+      "grad_norm": 0.7094339728355408,
+      "learning_rate": 7.819548872180451e-05,
+      "loss": 0.31165987253189087,
+      "mean_token_accuracy": 0.8851522654294968,
+      "num_tokens": 1936752.0,
+      "step": 157
+    },
+    {
+      "entropy": 1.018677532672882,
+      "epoch": 0.0832016850974197,
+      "grad_norm": 0.6412932276725769,
+      "learning_rate": 7.869674185463659e-05,
+      "loss": 0.28470396995544434,
+      "mean_token_accuracy": 0.8961537182331085,
+      "num_tokens": 1949088.0,
+      "step": 158
+    },
+    {
+      "entropy": 1.011895701289177,
+      "epoch": 0.08372827804107424,
+      "grad_norm": 0.6732510924339294,
+      "learning_rate": 7.919799498746867e-05,
+      "loss": 0.28573673963546753,
+      "mean_token_accuracy": 0.8944952934980392,
+      "num_tokens": 1961424.0,
+      "step": 159
+    },
+    {
+      "entropy": 1.011293813586235,
+      "epoch": 0.08425487098472881,
+      "grad_norm": 0.6785159707069397,
+      "learning_rate": 7.969924812030075e-05,
+      "loss": 0.30068039894104004,
+      "mean_token_accuracy": 0.8911136984825134,
+      "num_tokens": 1973760.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.020560160279274,
+      "epoch": 0.08478146392838336,
+      "grad_norm": 0.7218092679977417,
+      "learning_rate": 8.020050125313283e-05,
+      "loss": 0.30739086866378784,
+      "mean_token_accuracy": 0.885330468416214,
+      "num_tokens": 1986096.0,
+      "step": 161
+    },
+    {
+      "entropy": 1.0597094595432281,
+      "epoch": 0.08530805687203792,
+      "grad_norm": 0.7374167442321777,
+      "learning_rate": 8.070175438596491e-05,
+      "loss": 0.31700241565704346,
+      "mean_token_accuracy": 0.8844025731086731,
+      "num_tokens": 1998432.0,
+      "step": 162
+    },
+    {
+      "entropy": 1.0140591859817505,
+      "epoch": 0.08583464981569247,
+      "grad_norm": 0.6328212022781372,
+      "learning_rate": 8.120300751879699e-05,
+      "loss": 0.282073050737381,
+      "mean_token_accuracy": 0.9006371796131134,
+      "num_tokens": 2010768.0,
+      "step": 163
+    },
+    {
+      "entropy": 1.0362461507320404,
+      "epoch": 0.08636124275934702,
+      "grad_norm": 0.6613935828208923,
+      "learning_rate": 8.170426065162907e-05,
+      "loss": 0.2826480567455292,
+      "mean_token_accuracy": 0.8936625272035599,
+      "num_tokens": 2023104.0,
+      "step": 164
+    },
+    {
+      "entropy": 1.0381528735160828,
+      "epoch": 0.08688783570300158,
+      "grad_norm": 0.6834626197814941,
+      "learning_rate": 8.220551378446115e-05,
+      "loss": 0.2852419912815094,
+      "mean_token_accuracy": 0.8987163454294205,
+      "num_tokens": 2035440.0,
+      "step": 165
+    },
+    {
+      "entropy": 1.0698944479227066,
+      "epoch": 0.08741442864665613,
+      "grad_norm": 0.7904614806175232,
+      "learning_rate": 8.270676691729324e-05,
+      "loss": 0.2937854826450348,
+      "mean_token_accuracy": 0.8909425586462021,
+      "num_tokens": 2047776.0,
+      "step": 166
+    },
+    {
+      "entropy": 1.040902853012085,
+      "epoch": 0.08794102159031068,
+      "grad_norm": 0.7362328767776489,
+      "learning_rate": 8.320802005012532e-05,
+      "loss": 0.2914920449256897,
+      "mean_token_accuracy": 0.898249089717865,
+      "num_tokens": 2060112.0,
+      "step": 167
+    },
+    {
+      "entropy": 1.028788536787033,
+      "epoch": 0.08846761453396525,
+      "grad_norm": 0.8054221868515015,
+      "learning_rate": 8.37092731829574e-05,
+      "loss": 0.3018620014190674,
+      "mean_token_accuracy": 0.8883562982082367,
+      "num_tokens": 2072448.0,
+      "step": 168
+    },
+    {
+      "entropy": 1.0401608049869537,
+      "epoch": 0.0889942074776198,
+      "grad_norm": 0.6363168954849243,
+      "learning_rate": 8.421052631578948e-05,
+      "loss": 0.2924948036670685,
+      "mean_token_accuracy": 0.8931814581155777,
+      "num_tokens": 2084784.0,
+      "step": 169
+    },
+    {
+      "entropy": 1.0388599783182144,
+      "epoch": 0.08952080042127436,
+      "grad_norm": 1.0536071062088013,
+      "learning_rate": 8.471177944862155e-05,
+      "loss": 0.3022274971008301,
+      "mean_token_accuracy": 0.8892710357904434,
+      "num_tokens": 2097120.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.0794771611690521,
+      "epoch": 0.09004739336492891,
+      "grad_norm": 0.7485163807868958,
+      "learning_rate": 8.521303258145363e-05,
+      "loss": 0.28198155760765076,
+      "mean_token_accuracy": 0.8907860517501831,
+      "num_tokens": 2109456.0,
+      "step": 171
+    },
+    {
+      "entropy": 1.0737251937389374,
+      "epoch": 0.09057398630858346,
+      "grad_norm": 0.7381237745285034,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.2523871660232544,
+      "mean_token_accuracy": 0.9074936509132385,
+      "num_tokens": 2121792.0,
+      "step": 172
+    },
+    {
+      "entropy": 1.003856584429741,
+      "epoch": 0.09110057925223802,
+      "grad_norm": 0.7226340770721436,
+      "learning_rate": 8.621553884711779e-05,
+      "loss": 0.27109643816947937,
+      "mean_token_accuracy": 0.9041073471307755,
+      "num_tokens": 2134128.0,
+      "step": 173
+    },
+    {
+      "entropy": 1.1186200976371765,
+      "epoch": 0.09162717219589257,
+      "grad_norm": 0.7579658627510071,
+      "learning_rate": 8.671679197994987e-05,
+      "loss": 0.3053061366081238,
+      "mean_token_accuracy": 0.8858212381601334,
+      "num_tokens": 2146464.0,
+      "step": 174
+    },
+    {
+      "entropy": 1.1462504863739014,
+      "epoch": 0.09215376513954714,
+      "grad_norm": 0.6947624087333679,
+      "learning_rate": 8.721804511278195e-05,
+      "loss": 0.3097134232521057,
+      "mean_token_accuracy": 0.8846383690834045,
+      "num_tokens": 2158800.0,
+      "step": 175
+    },
+    {
+      "entropy": 1.1312821209430695,
+      "epoch": 0.09268035808320169,
+      "grad_norm": 0.6850878596305847,
+      "learning_rate": 8.771929824561403e-05,
+      "loss": 0.26429426670074463,
+      "mean_token_accuracy": 0.8986905962228775,
+      "num_tokens": 2171136.0,
+      "step": 176
+    },
+    {
+      "entropy": 1.146384358406067,
+      "epoch": 0.09320695102685624,
+      "grad_norm": 0.7358205914497375,
+      "learning_rate": 8.822055137844612e-05,
+      "loss": 0.2753527760505676,
+      "mean_token_accuracy": 0.8983759731054306,
+      "num_tokens": 2183472.0,
+      "step": 177
+    },
+    {
+      "entropy": 1.2575857639312744,
+      "epoch": 0.0937335439705108,
+      "grad_norm": 0.6449548006057739,
+      "learning_rate": 8.87218045112782e-05,
+      "loss": 0.2840855121612549,
+      "mean_token_accuracy": 0.8939386010169983,
+      "num_tokens": 2195808.0,
+      "step": 178
+    },
+    {
+      "entropy": 1.190169095993042,
+      "epoch": 0.09426013691416535,
+      "grad_norm": 0.7093657851219177,
+      "learning_rate": 8.922305764411028e-05,
+      "loss": 0.2701612114906311,
+      "mean_token_accuracy": 0.9016707092523575,
+      "num_tokens": 2208144.0,
+      "step": 179
+    },
+    {
+      "entropy": 1.229299396276474,
+      "epoch": 0.0947867298578199,
+      "grad_norm": 0.7190297245979309,
+      "learning_rate": 8.972431077694236e-05,
+      "loss": 0.25798100233078003,
+      "mean_token_accuracy": 0.9037416577339172,
+      "num_tokens": 2220480.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.2668792605400085,
+      "epoch": 0.09531332280147446,
+      "grad_norm": 0.7405722737312317,
+      "learning_rate": 9.022556390977444e-05,
+      "loss": 0.2883971929550171,
+      "mean_token_accuracy": 0.8925597667694092,
+      "num_tokens": 2232816.0,
+      "step": 181
+    },
+    {
+      "entropy": 1.1805595457553864,
+      "epoch": 0.09583991574512901,
+      "grad_norm": 0.6785654425621033,
+      "learning_rate": 9.072681704260652e-05,
+      "loss": 0.2622534930706024,
+      "mean_token_accuracy": 0.9043945819139481,
+      "num_tokens": 2245152.0,
+      "step": 182
+    },
+    {
+      "entropy": 1.24412140250206,
+      "epoch": 0.09636650868878358,
+      "grad_norm": 0.6113075017929077,
+      "learning_rate": 9.12280701754386e-05,
+      "loss": 0.30303800106048584,
+      "mean_token_accuracy": 0.8807209581136703,
+      "num_tokens": 2257488.0,
+      "step": 183
+    },
+    {
+      "entropy": 1.1531569063663483,
+      "epoch": 0.09689310163243813,
+      "grad_norm": 0.6058000922203064,
+      "learning_rate": 9.172932330827067e-05,
+      "loss": 0.2888239622116089,
+      "mean_token_accuracy": 0.8989417850971222,
+      "num_tokens": 2269824.0,
+      "step": 184
+    },
+    {
+      "entropy": 1.1511635780334473,
+      "epoch": 0.09741969457609267,
+      "grad_norm": 0.6023234128952026,
+      "learning_rate": 9.223057644110275e-05,
+      "loss": 0.2515050172805786,
+      "mean_token_accuracy": 0.9097182005643845,
+      "num_tokens": 2282160.0,
+      "step": 185
+    },
+    {
+      "entropy": 1.1716180443763733,
+      "epoch": 0.09794628751974724,
+      "grad_norm": 0.6153926253318787,
+      "learning_rate": 9.273182957393483e-05,
+      "loss": 0.2679345905780792,
+      "mean_token_accuracy": 0.8987378627061844,
+      "num_tokens": 2294496.0,
+      "step": 186
+    },
+    {
+      "entropy": 1.1668370068073273,
+      "epoch": 0.09847288046340179,
+      "grad_norm": 0.6524267196655273,
+      "learning_rate": 9.323308270676691e-05,
+      "loss": 0.28671011328697205,
+      "mean_token_accuracy": 0.8961494415998459,
+      "num_tokens": 2306832.0,
+      "step": 187
+    },
+    {
+      "entropy": 1.1817428171634674,
+      "epoch": 0.09899947340705635,
+      "grad_norm": 0.6342366337776184,
+      "learning_rate": 9.373433583959899e-05,
+      "loss": 0.2956732511520386,
+      "mean_token_accuracy": 0.8947313278913498,
+      "num_tokens": 2319168.0,
+      "step": 188
+    },
+    {
+      "entropy": 1.118630200624466,
+      "epoch": 0.0995260663507109,
+      "grad_norm": 0.6613883972167969,
+      "learning_rate": 9.423558897243108e-05,
+      "loss": 0.24410809576511383,
+      "mean_token_accuracy": 0.9081003367900848,
+      "num_tokens": 2331504.0,
+      "step": 189
+    },
+    {
+      "entropy": 1.1118974089622498,
+      "epoch": 0.10005265929436545,
+      "grad_norm": 0.6532029509544373,
+      "learning_rate": 9.473684210526316e-05,
+      "loss": 0.2677920162677765,
+      "mean_token_accuracy": 0.8997665196657181,
+      "num_tokens": 2343840.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1150999963283539,
+      "epoch": 0.10057925223802001,
+      "grad_norm": 0.6919822692871094,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.2759622633457184,
+      "mean_token_accuracy": 0.896903395652771,
+      "num_tokens": 2356176.0,
+      "step": 191
+    },
+    {
+      "entropy": 1.0659538507461548,
+      "epoch": 0.10110584518167456,
+      "grad_norm": 0.6363855600357056,
+      "learning_rate": 9.573934837092732e-05,
+      "loss": 0.2608574628829956,
+      "mean_token_accuracy": 0.8995112329721451,
+      "num_tokens": 2368512.0,
+      "step": 192
+    },
+    {
+      "entropy": 1.0789634585380554,
+      "epoch": 0.10163243812532911,
+      "grad_norm": 0.7043361067771912,
+      "learning_rate": 9.62406015037594e-05,
+      "loss": 0.27143171429634094,
+      "mean_token_accuracy": 0.897380143404007,
+      "num_tokens": 2380848.0,
+      "step": 193
+    },
+    {
+      "entropy": 1.1381100416183472,
+      "epoch": 0.10215903106898368,
+      "grad_norm": 0.6552066206932068,
+      "learning_rate": 9.674185463659148e-05,
+      "loss": 0.26525720953941345,
+      "mean_token_accuracy": 0.9070263504981995,
+      "num_tokens": 2393184.0,
+      "step": 194
+    },
+    {
+      "entropy": 1.1865114271640778,
+      "epoch": 0.10268562401263823,
+      "grad_norm": 0.634785532951355,
+      "learning_rate": 9.724310776942356e-05,
+      "loss": 0.26738592982292175,
+      "mean_token_accuracy": 0.9013326019048691,
+      "num_tokens": 2405520.0,
+      "step": 195
+    },
+    {
+      "entropy": 1.1835360527038574,
+      "epoch": 0.10321221695629279,
+      "grad_norm": 0.7315617799758911,
+      "learning_rate": 9.774436090225564e-05,
+      "loss": 0.29927945137023926,
+      "mean_token_accuracy": 0.8894186317920685,
+      "num_tokens": 2417856.0,
+      "step": 196
+    },
+    {
+      "entropy": 1.1675716638565063,
+      "epoch": 0.10373880989994734,
+      "grad_norm": 0.5887401700019836,
+      "learning_rate": 9.824561403508771e-05,
+      "loss": 0.2657887935638428,
+      "mean_token_accuracy": 0.9044163674116135,
+      "num_tokens": 2430192.0,
+      "step": 197
+    },
+    {
+      "entropy": 1.2119200527668,
+      "epoch": 0.10426540284360189,
+      "grad_norm": 0.7186099290847778,
+      "learning_rate": 9.87468671679198e-05,
+      "loss": 0.24999147653579712,
+      "mean_token_accuracy": 0.9089571237564087,
+      "num_tokens": 2442528.0,
+      "step": 198
+    },
+    {
+      "entropy": 1.1587353944778442,
+      "epoch": 0.10479199578725645,
+      "grad_norm": 0.6930050253868103,
+      "learning_rate": 9.924812030075187e-05,
+      "loss": 0.28423306345939636,
+      "mean_token_accuracy": 0.8938881307840347,
+      "num_tokens": 2454864.0,
+      "step": 199
+    },
+    {
+      "entropy": 1.1999735534191132,
+      "epoch": 0.105318588730911,
+      "grad_norm": 0.6575484871864319,
+      "learning_rate": 9.974937343358397e-05,
+      "loss": 0.25892868638038635,
+      "mean_token_accuracy": 0.9065304547548294,
+      "num_tokens": 2467200.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.1664292514324188,
+      "epoch": 0.10584518167456557,
+      "grad_norm": 0.6501355767250061,
+      "learning_rate": 0.00010025062656641604,
+      "loss": 0.2730686068534851,
+      "mean_token_accuracy": 0.8991207629442215,
+      "num_tokens": 2479536.0,
+      "step": 201
+    },
+    {
+      "entropy": 1.2156886756420135,
+      "epoch": 0.10637177461822012,
+      "grad_norm": 1.1883974075317383,
+      "learning_rate": 0.00010075187969924814,
+      "loss": 0.2882535755634308,
+      "mean_token_accuracy": 0.8935256004333496,
+      "num_tokens": 2491872.0,
+      "step": 202
+    },
+    {
+      "entropy": 1.1884984374046326,
+      "epoch": 0.10689836756187467,
+      "grad_norm": 0.6555978655815125,
+      "learning_rate": 0.0001012531328320802,
+      "loss": 0.2524234354496002,
+      "mean_token_accuracy": 0.9072889387607574,
+      "num_tokens": 2504208.0,
+      "step": 203
+    },
+    {
+      "entropy": 1.204279750585556,
+      "epoch": 0.10742496050552923,
+      "grad_norm": 0.6422716379165649,
+      "learning_rate": 0.0001017543859649123,
+      "loss": 0.2662729322910309,
+      "mean_token_accuracy": 0.9040449559688568,
+      "num_tokens": 2516544.0,
+      "step": 204
+    },
+    {
+      "entropy": 1.2619583904743195,
+      "epoch": 0.10795155344918378,
+      "grad_norm": 0.7150806188583374,
+      "learning_rate": 0.00010225563909774436,
+      "loss": 0.2756538689136505,
+      "mean_token_accuracy": 0.8975429385900497,
+      "num_tokens": 2528880.0,
+      "step": 205
+    },
+    {
+      "entropy": 1.265720009803772,
+      "epoch": 0.10847814639283834,
+      "grad_norm": 0.6873537302017212,
+      "learning_rate": 0.00010275689223057645,
+      "loss": 0.25472116470336914,
+      "mean_token_accuracy": 0.9096198529005051,
+      "num_tokens": 2541216.0,
+      "step": 206
+    },
+    {
+      "entropy": 1.269091546535492,
+      "epoch": 0.10900473933649289,
+      "grad_norm": 0.6951980590820312,
+      "learning_rate": 0.00010325814536340852,
+      "loss": 0.26120638847351074,
+      "mean_token_accuracy": 0.9050850421190262,
+      "num_tokens": 2553552.0,
+      "step": 207
+    },
+    {
+      "entropy": 1.269478291273117,
+      "epoch": 0.10953133228014744,
+      "grad_norm": 0.556747555732727,
+      "learning_rate": 0.00010375939849624061,
+      "loss": 0.25462740659713745,
+      "mean_token_accuracy": 0.9035714715719223,
+      "num_tokens": 2565888.0,
+      "step": 208
+    },
+    {
+      "entropy": 1.261308342218399,
+      "epoch": 0.110057925223802,
+      "grad_norm": 0.6084737777709961,
+      "learning_rate": 0.00010426065162907268,
+      "loss": 0.28605493903160095,
+      "mean_token_accuracy": 0.8935003876686096,
+      "num_tokens": 2578224.0,
+      "step": 209
+    },
+    {
+      "entropy": 1.3189480602741241,
+      "epoch": 0.11058451816745656,
+      "grad_norm": 0.5886352062225342,
+      "learning_rate": 0.00010476190476190477,
+      "loss": 0.24048885703086853,
+      "mean_token_accuracy": 0.9100839495658875,
+      "num_tokens": 2590560.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.3389763832092285,
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.6794092059135437,
+      "learning_rate": 0.00010526315789473685,
+      "loss": 0.2771369516849518,
+      "mean_token_accuracy": 0.890377551317215,
+      "num_tokens": 2602896.0,
+      "step": 211
+    },
+    {
+      "entropy": 1.3012954592704773,
+      "epoch": 0.11163770405476567,
+      "grad_norm": 0.6500777006149292,
+      "learning_rate": 0.00010576441102756893,
+      "loss": 0.26194366812705994,
+      "mean_token_accuracy": 0.9035488367080688,
+      "num_tokens": 2615232.0,
+      "step": 212
+    },
+    {
+      "entropy": 1.3287411630153656,
+      "epoch": 0.11216429699842022,
+      "grad_norm": 0.7444258332252502,
+      "learning_rate": 0.000106265664160401,
+      "loss": 0.25705063343048096,
+      "mean_token_accuracy": 0.8985376209020615,
+      "num_tokens": 2627568.0,
+      "step": 213
+    },
+    {
+      "entropy": 1.3762761354446411,
+      "epoch": 0.11269088994207478,
+      "grad_norm": 0.6778383255004883,
+      "learning_rate": 0.0001067669172932331,
+      "loss": 0.26562660932540894,
+      "mean_token_accuracy": 0.8980493247509003,
+      "num_tokens": 2639904.0,
+      "step": 214
+    },
+    {
+      "entropy": 1.3341316282749176,
+      "epoch": 0.11321748288572933,
+      "grad_norm": 0.5709004998207092,
+      "learning_rate": 0.00010726817042606516,
+      "loss": 0.22839783132076263,
+      "mean_token_accuracy": 0.9137850105762482,
+      "num_tokens": 2652240.0,
+      "step": 215
+    },
+    {
+      "entropy": 1.397248774766922,
+      "epoch": 0.11374407582938388,
+      "grad_norm": 0.6649454236030579,
+      "learning_rate": 0.00010776942355889726,
+      "loss": 0.2758176624774933,
+      "mean_token_accuracy": 0.8951268941164017,
+      "num_tokens": 2664576.0,
+      "step": 216
+    },
+    {
+      "entropy": 1.3985490798950195,
+      "epoch": 0.11427066877303844,
+      "grad_norm": 0.6822545528411865,
+      "learning_rate": 0.00010827067669172932,
+      "loss": 0.2714245021343231,
+      "mean_token_accuracy": 0.9002304524183273,
+      "num_tokens": 2676912.0,
+      "step": 217
+    },
+    {
+      "entropy": 1.4322108924388885,
+      "epoch": 0.114797261716693,
+      "grad_norm": 0.6045425534248352,
+      "learning_rate": 0.00010877192982456141,
+      "loss": 0.27720269560813904,
+      "mean_token_accuracy": 0.8973477184772491,
+      "num_tokens": 2689248.0,
+      "step": 218
+    },
+    {
+      "entropy": 1.412218064069748,
+      "epoch": 0.11532385466034756,
+      "grad_norm": 0.5996334552764893,
+      "learning_rate": 0.00010927318295739348,
+      "loss": 0.2552666962146759,
+      "mean_token_accuracy": 0.9052979648113251,
+      "num_tokens": 2701584.0,
+      "step": 219
+    },
+    {
+      "entropy": 1.4080830514431,
+      "epoch": 0.11585044760400211,
+      "grad_norm": 0.5775550007820129,
+      "learning_rate": 0.00010977443609022557,
+      "loss": 0.26107677817344666,
+      "mean_token_accuracy": 0.902474120259285,
+      "num_tokens": 2713920.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.4139901399612427,
+      "epoch": 0.11637704054765666,
+      "grad_norm": 0.5853016972541809,
+      "learning_rate": 0.00011027568922305764,
+      "loss": 0.2802305519580841,
+      "mean_token_accuracy": 0.8949225097894669,
+      "num_tokens": 2726256.0,
+      "step": 221
+    },
+    {
+      "entropy": 1.354727566242218,
+      "epoch": 0.11690363349131122,
+      "grad_norm": 0.6593271493911743,
+      "learning_rate": 0.00011077694235588973,
+      "loss": 0.25806909799575806,
+      "mean_token_accuracy": 0.9009847342967987,
+      "num_tokens": 2738592.0,
+      "step": 222
+    },
+    {
+      "entropy": 1.363867849111557,
+      "epoch": 0.11743022643496577,
+      "grad_norm": 0.5569132566452026,
+      "learning_rate": 0.00011127819548872181,
+      "loss": 0.2619836926460266,
+      "mean_token_accuracy": 0.9014725536108017,
+      "num_tokens": 2750928.0,
+      "step": 223
+    },
+    {
+      "entropy": 1.4332265555858612,
+      "epoch": 0.11795681937862032,
+      "grad_norm": 0.6189314723014832,
+      "learning_rate": 0.00011177944862155389,
+      "loss": 0.26161426305770874,
+      "mean_token_accuracy": 0.8975115269422531,
+      "num_tokens": 2763264.0,
+      "step": 224
+    },
+    {
+      "entropy": 1.3454554677009583,
+      "epoch": 0.11848341232227488,
+      "grad_norm": 0.6003779768943787,
+      "learning_rate": 0.00011228070175438597,
+      "loss": 0.2651001214981079,
+      "mean_token_accuracy": 0.9023340046405792,
+      "num_tokens": 2775600.0,
+      "step": 225
+    },
+    {
+      "entropy": 1.3587769269943237,
+      "epoch": 0.11901000526592943,
+      "grad_norm": 0.7411830425262451,
+      "learning_rate": 0.00011278195488721806,
+      "loss": 0.2888566851615906,
+      "mean_token_accuracy": 0.895685225725174,
+      "num_tokens": 2787936.0,
+      "step": 226
+    },
+    {
+      "entropy": 1.4554640650749207,
+      "epoch": 0.119536598209584,
+      "grad_norm": 0.6714136004447937,
+      "learning_rate": 0.00011328320802005013,
+      "loss": 0.2898600101470947,
+      "mean_token_accuracy": 0.8905848562717438,
+      "num_tokens": 2800272.0,
+      "step": 227
+    },
+    {
+      "entropy": 1.4274349808692932,
+      "epoch": 0.12006319115323855,
+      "grad_norm": 0.5972358584403992,
+      "learning_rate": 0.00011378446115288222,
+      "loss": 0.2719231843948364,
+      "mean_token_accuracy": 0.8995204567909241,
+      "num_tokens": 2812608.0,
+      "step": 228
+    },
+    {
+      "entropy": 1.3975107669830322,
+      "epoch": 0.1205897840968931,
+      "grad_norm": 0.7388063073158264,
+      "learning_rate": 0.00011428571428571428,
+      "loss": 0.2811048924922943,
+      "mean_token_accuracy": 0.8961653858423233,
+      "num_tokens": 2824944.0,
+      "step": 229
+    },
+    {
+      "entropy": 1.41668900847435,
+      "epoch": 0.12111637704054766,
+      "grad_norm": 0.6408892273902893,
+      "learning_rate": 0.00011478696741854638,
+      "loss": 0.23900066316127777,
+      "mean_token_accuracy": 0.9131550490856171,
+      "num_tokens": 2837280.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.4389840960502625,
+      "epoch": 0.12164296998420221,
+      "grad_norm": 0.6279134154319763,
+      "learning_rate": 0.00011528822055137844,
+      "loss": 0.27471503615379333,
+      "mean_token_accuracy": 0.8994272202253342,
+      "num_tokens": 2849616.0,
+      "step": 231
+    },
+    {
+      "entropy": 1.3705064952373505,
+      "epoch": 0.12216956292785677,
+      "grad_norm": 0.7351282835006714,
+      "learning_rate": 0.00011578947368421053,
+      "loss": 0.2829359471797943,
+      "mean_token_accuracy": 0.8966951966285706,
+      "num_tokens": 2861952.0,
+      "step": 232
+    },
+    {
+      "entropy": 1.4880229830741882,
+      "epoch": 0.12269615587151132,
+      "grad_norm": 0.6098307371139526,
+      "learning_rate": 0.0001162907268170426,
+      "loss": 0.25372314453125,
+      "mean_token_accuracy": 0.901044175028801,
+      "num_tokens": 2874288.0,
+      "step": 233
+    },
+    {
+      "entropy": 1.4908549189567566,
+      "epoch": 0.12322274881516587,
+      "grad_norm": 0.6075619459152222,
+      "learning_rate": 0.00011679197994987469,
+      "loss": 0.2654881477355957,
+      "mean_token_accuracy": 0.8972955793142319,
+      "num_tokens": 2886624.0,
+      "step": 234
+    },
+    {
+      "entropy": 1.3783467411994934,
+      "epoch": 0.12374934175882044,
+      "grad_norm": 0.6030115485191345,
+      "learning_rate": 0.00011729323308270677,
+      "loss": 0.25111234188079834,
+      "mean_token_accuracy": 0.902543842792511,
+      "num_tokens": 2898960.0,
+      "step": 235
+    },
+    {
+      "entropy": 1.3683518171310425,
+      "epoch": 0.12427593470247499,
+      "grad_norm": 0.6671847105026245,
+      "learning_rate": 0.00011779448621553886,
+      "loss": 0.25656643509864807,
+      "mean_token_accuracy": 0.9067094326019287,
+      "num_tokens": 2911296.0,
+      "step": 236
+    },
+    {
+      "entropy": 1.4006175696849823,
+      "epoch": 0.12480252764612954,
+      "grad_norm": 0.5684525370597839,
+      "learning_rate": 0.00011829573934837093,
+      "loss": 0.2626839876174927,
+      "mean_token_accuracy": 0.9033264219760895,
+      "num_tokens": 2923632.0,
+      "step": 237
+    },
+    {
+      "entropy": 1.335418850183487,
+      "epoch": 0.1253291205897841,
+      "grad_norm": 0.5542851090431213,
+      "learning_rate": 0.00011879699248120302,
+      "loss": 0.24251079559326172,
+      "mean_token_accuracy": 0.9084768146276474,
+      "num_tokens": 2935968.0,
+      "step": 238
+    },
+    {
+      "entropy": 1.3918206691741943,
+      "epoch": 0.12585571353343866,
+      "grad_norm": 0.7893558740615845,
+      "learning_rate": 0.00011929824561403509,
+      "loss": 0.2818213999271393,
+      "mean_token_accuracy": 0.8929243236780167,
+      "num_tokens": 2948304.0,
+      "step": 239
+    },
+    {
+      "entropy": 1.3344398736953735,
+      "epoch": 0.1263823064770932,
+      "grad_norm": 0.62665194272995,
+      "learning_rate": 0.00011979949874686718,
+      "loss": 0.24543407559394836,
+      "mean_token_accuracy": 0.9088149815797806,
+      "num_tokens": 2960640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.362634390592575,
+      "epoch": 0.12690889942074776,
+      "grad_norm": 0.6178820133209229,
+      "learning_rate": 0.00012030075187969925,
+      "loss": 0.24784636497497559,
+      "mean_token_accuracy": 0.9079759418964386,
+      "num_tokens": 2972976.0,
+      "step": 241
+    },
+    {
+      "entropy": 1.3410207033157349,
+      "epoch": 0.12743549236440233,
+      "grad_norm": 0.6127734780311584,
+      "learning_rate": 0.00012080200501253134,
+      "loss": 0.276973694562912,
+      "mean_token_accuracy": 0.900666818022728,
+      "num_tokens": 2985312.0,
+      "step": 242
+    },
+    {
+      "entropy": 1.3882217109203339,
+      "epoch": 0.12796208530805686,
+      "grad_norm": 0.6211519837379456,
+      "learning_rate": 0.0001213032581453634,
+      "loss": 0.29095229506492615,
+      "mean_token_accuracy": 0.8922038376331329,
+      "num_tokens": 2997648.0,
+      "step": 243
+    },
+    {
+      "entropy": 1.3064784407615662,
+      "epoch": 0.12848867825171142,
+      "grad_norm": 0.6020708084106445,
+      "learning_rate": 0.0001218045112781955,
+      "loss": 0.25750190019607544,
+      "mean_token_accuracy": 0.8979595750570297,
+      "num_tokens": 3009984.0,
+      "step": 244
+    },
+    {
+      "entropy": 1.3404588103294373,
+      "epoch": 0.129015271195366,
+      "grad_norm": 0.5311986804008484,
+      "learning_rate": 0.00012230576441102757,
+      "loss": 0.2417442798614502,
+      "mean_token_accuracy": 0.9036334156990051,
+      "num_tokens": 3022320.0,
+      "step": 245
+    },
+    {
+      "entropy": 1.3250301778316498,
+      "epoch": 0.12954186413902052,
+      "grad_norm": 0.6158598065376282,
+      "learning_rate": 0.00012280701754385965,
+      "loss": 0.28592783212661743,
+      "mean_token_accuracy": 0.8968044072389603,
+      "num_tokens": 3034656.0,
+      "step": 246
+    },
+    {
+      "entropy": 1.2895326018333435,
+      "epoch": 0.1300684570826751,
+      "grad_norm": 0.5133572816848755,
+      "learning_rate": 0.00012330827067669173,
+      "loss": 0.26722657680511475,
+      "mean_token_accuracy": 0.899601012468338,
+      "num_tokens": 3046992.0,
+      "step": 247
+    },
+    {
+      "entropy": 1.2834957540035248,
+      "epoch": 0.13059505002632965,
+      "grad_norm": 0.5774098038673401,
+      "learning_rate": 0.0001238095238095238,
+      "loss": 0.2552984654903412,
+      "mean_token_accuracy": 0.9021749496459961,
+      "num_tokens": 3059328.0,
+      "step": 248
+    },
+    {
+      "entropy": 1.28067284822464,
+      "epoch": 0.13112164296998421,
+      "grad_norm": 0.5813867449760437,
+      "learning_rate": 0.0001243107769423559,
+      "loss": 0.26484355330467224,
+      "mean_token_accuracy": 0.8958484381437302,
+      "num_tokens": 3071664.0,
+      "step": 249
+    },
+    {
+      "entropy": 1.3292741775512695,
+      "epoch": 0.13164823591363875,
+      "grad_norm": 0.5437096357345581,
+      "learning_rate": 0.00012481203007518797,
+      "loss": 0.25312983989715576,
+      "mean_token_accuracy": 0.9083140790462494,
+      "num_tokens": 3084000.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.217045933008194,
+      "epoch": 0.1321748288572933,
+      "grad_norm": 0.5642932057380676,
+      "learning_rate": 0.00012531328320802005,
+      "loss": 0.24098949134349823,
+      "mean_token_accuracy": 0.9082589000463486,
+      "num_tokens": 3096336.0,
+      "step": 251
+    },
+    {
+      "entropy": 1.3332138061523438,
+      "epoch": 0.13270142180094788,
+      "grad_norm": 0.5955415368080139,
+      "learning_rate": 0.00012581453634085213,
+      "loss": 0.2849401831626892,
+      "mean_token_accuracy": 0.8916681408882141,
+      "num_tokens": 3108672.0,
+      "step": 252
+    },
+    {
+      "entropy": 1.295847773551941,
+      "epoch": 0.1332280147446024,
+      "grad_norm": 0.6146616339683533,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 0.2735283076763153,
+      "mean_token_accuracy": 0.8989967852830887,
+      "num_tokens": 3121008.0,
+      "step": 253
+    },
+    {
+      "entropy": 1.2074873447418213,
+      "epoch": 0.13375460768825698,
+      "grad_norm": 0.5243288278579712,
+      "learning_rate": 0.0001268170426065163,
+      "loss": 0.24956879019737244,
+      "mean_token_accuracy": 0.9056222587823868,
+      "num_tokens": 3133344.0,
+      "step": 254
+    },
+    {
+      "entropy": 1.2322421371936798,
+      "epoch": 0.13428120063191154,
+      "grad_norm": 0.5780612230300903,
+      "learning_rate": 0.00012731829573934836,
+      "loss": 0.22142812609672546,
+      "mean_token_accuracy": 0.9111984372138977,
+      "num_tokens": 3145680.0,
+      "step": 255
+    },
+    {
+      "entropy": 1.2550228536128998,
+      "epoch": 0.13480779357556608,
+      "grad_norm": 0.5654560923576355,
+      "learning_rate": 0.00012781954887218047,
+      "loss": 0.2577667236328125,
+      "mean_token_accuracy": 0.9028536081314087,
+      "num_tokens": 3158016.0,
+      "step": 256
+    },
+    {
+      "entropy": 1.2815439105033875,
+      "epoch": 0.13533438651922064,
+      "grad_norm": 0.5860424041748047,
+      "learning_rate": 0.00012832080200501252,
+      "loss": 0.24961405992507935,
+      "mean_token_accuracy": 0.9073447585105896,
+      "num_tokens": 3170352.0,
+      "step": 257
+    },
+    {
+      "entropy": 1.3496994376182556,
+      "epoch": 0.1358609794628752,
+      "grad_norm": 0.5496608018875122,
+      "learning_rate": 0.00012882205513784463,
+      "loss": 0.23788216710090637,
+      "mean_token_accuracy": 0.913466677069664,
+      "num_tokens": 3182688.0,
+      "step": 258
+    },
+    {
+      "entropy": 1.2379789352416992,
+      "epoch": 0.13638757240652974,
+      "grad_norm": 0.5407847166061401,
+      "learning_rate": 0.00012932330827067668,
+      "loss": 0.242479607462883,
+      "mean_token_accuracy": 0.9110987186431885,
+      "num_tokens": 3195024.0,
+      "step": 259
+    },
+    {
+      "entropy": 1.2678188979625702,
+      "epoch": 0.1369141653501843,
+      "grad_norm": 0.6954227089881897,
+      "learning_rate": 0.0001298245614035088,
+      "loss": 0.3035907447338104,
+      "mean_token_accuracy": 0.8935103565454483,
+      "num_tokens": 3207360.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.2501021325588226,
+      "epoch": 0.13744075829383887,
+      "grad_norm": 0.6763582825660706,
+      "learning_rate": 0.00013032581453634084,
+      "loss": 0.2833873927593231,
+      "mean_token_accuracy": 0.8930742889642715,
+      "num_tokens": 3219696.0,
+      "step": 261
+    },
+    {
+      "entropy": 1.2093367278575897,
+      "epoch": 0.13796735123749343,
+      "grad_norm": 0.5583021640777588,
+      "learning_rate": 0.00013082706766917294,
+      "loss": 0.23031063377857208,
+      "mean_token_accuracy": 0.9110294431447983,
+      "num_tokens": 3232032.0,
+      "step": 262
+    },
+    {
+      "entropy": 1.2286772429943085,
+      "epoch": 0.13849394418114797,
+      "grad_norm": 0.5849270224571228,
+      "learning_rate": 0.00013132832080200502,
+      "loss": 0.27444857358932495,
+      "mean_token_accuracy": 0.8965310454368591,
+      "num_tokens": 3244368.0,
+      "step": 263
+    },
+    {
+      "entropy": 1.2541997134685516,
+      "epoch": 0.13902053712480253,
+      "grad_norm": 0.5210989713668823,
+      "learning_rate": 0.0001318295739348371,
+      "loss": 0.24464797973632812,
+      "mean_token_accuracy": 0.9072559326887131,
+      "num_tokens": 3256704.0,
+      "step": 264
+    },
+    {
+      "entropy": 1.2401838898658752,
+      "epoch": 0.1395471300684571,
+      "grad_norm": 0.5357401371002197,
+      "learning_rate": 0.00013233082706766918,
+      "loss": 0.24787980318069458,
+      "mean_token_accuracy": 0.9098693877458572,
+      "num_tokens": 3269040.0,
+      "step": 265
+    },
+    {
+      "entropy": 1.2660530507564545,
+      "epoch": 0.14007372301211163,
+      "grad_norm": 0.5940161943435669,
+      "learning_rate": 0.00013283208020050126,
+      "loss": 0.27411437034606934,
+      "mean_token_accuracy": 0.8980700373649597,
+      "num_tokens": 3281376.0,
+      "step": 266
+    },
+    {
+      "entropy": 1.241824746131897,
+      "epoch": 0.1406003159557662,
+      "grad_norm": 0.604097306728363,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.2866377830505371,
+      "mean_token_accuracy": 0.8919961899518967,
+      "num_tokens": 3293712.0,
+      "step": 267
+    },
+    {
+      "entropy": 1.1769486963748932,
+      "epoch": 0.14112690889942076,
+      "grad_norm": 0.7264558672904968,
+      "learning_rate": 0.00013383458646616542,
+      "loss": 0.26414263248443604,
+      "mean_token_accuracy": 0.8967715352773666,
+      "num_tokens": 3306048.0,
+      "step": 268
+    },
+    {
+      "entropy": 1.2283678948879242,
+      "epoch": 0.1416535018430753,
+      "grad_norm": 0.5465900301933289,
+      "learning_rate": 0.0001343358395989975,
+      "loss": 0.2366294264793396,
+      "mean_token_accuracy": 0.9092733561992645,
+      "num_tokens": 3318384.0,
+      "step": 269
+    },
+    {
+      "entropy": 1.262256383895874,
+      "epoch": 0.14218009478672985,
+      "grad_norm": 0.5315939784049988,
+      "learning_rate": 0.00013483709273182958,
+      "loss": 0.22708219289779663,
+      "mean_token_accuracy": 0.914520725607872,
+      "num_tokens": 3330720.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.175988882780075,
+      "epoch": 0.14270668773038442,
+      "grad_norm": 0.5613946318626404,
+      "learning_rate": 0.00013533834586466166,
+      "loss": 0.22835803031921387,
+      "mean_token_accuracy": 0.9081931263208389,
+      "num_tokens": 3343056.0,
+      "step": 271
+    },
+    {
+      "entropy": 1.2391490042209625,
+      "epoch": 0.14323328067403895,
+      "grad_norm": 0.5381202101707458,
+      "learning_rate": 0.00013583959899749373,
+      "loss": 0.25535333156585693,
+      "mean_token_accuracy": 0.9033481776714325,
+      "num_tokens": 3355392.0,
+      "step": 272
+    },
+    {
+      "entropy": 1.2381133437156677,
+      "epoch": 0.14375987361769352,
+      "grad_norm": 0.6545175909996033,
+      "learning_rate": 0.00013634085213032581,
+      "loss": 0.268245667219162,
+      "mean_token_accuracy": 0.9008642733097076,
+      "num_tokens": 3367728.0,
+      "step": 273
+    },
+    {
+      "entropy": 1.2534950375556946,
+      "epoch": 0.14428646656134808,
+      "grad_norm": 0.6084462404251099,
+      "learning_rate": 0.0001368421052631579,
+      "loss": 0.241957426071167,
+      "mean_token_accuracy": 0.9131230562925339,
+      "num_tokens": 3380064.0,
+      "step": 274
+    },
+    {
+      "entropy": 1.2566357254981995,
+      "epoch": 0.14481305950500264,
+      "grad_norm": 0.5803132057189941,
+      "learning_rate": 0.00013734335839598997,
+      "loss": 0.24542784690856934,
+      "mean_token_accuracy": 0.9083303511142731,
+      "num_tokens": 3392400.0,
+      "step": 275
+    },
+    {
+      "entropy": 1.3298097252845764,
+      "epoch": 0.14533965244865718,
+      "grad_norm": 0.5581626296043396,
+      "learning_rate": 0.00013784461152882208,
+      "loss": 0.2682327330112457,
+      "mean_token_accuracy": 0.8970459997653961,
+      "num_tokens": 3404736.0,
+      "step": 276
+    },
+    {
+      "entropy": 1.2984472513198853,
+      "epoch": 0.14586624539231174,
+      "grad_norm": 0.5797234177589417,
+      "learning_rate": 0.00013834586466165413,
+      "loss": 0.2551107406616211,
+      "mean_token_accuracy": 0.9025808274745941,
+      "num_tokens": 3417072.0,
+      "step": 277
+    },
+    {
+      "entropy": 1.232091337442398,
+      "epoch": 0.1463928383359663,
+      "grad_norm": 0.56398606300354,
+      "learning_rate": 0.00013884711779448624,
+      "loss": 0.2523190379142761,
+      "mean_token_accuracy": 0.9042512625455856,
+      "num_tokens": 3429408.0,
+      "step": 278
+    },
+    {
+      "entropy": 1.364453375339508,
+      "epoch": 0.14691943127962084,
+      "grad_norm": 0.5884234309196472,
+      "learning_rate": 0.0001393483709273183,
+      "loss": 0.2788117229938507,
+      "mean_token_accuracy": 0.90156190097332,
+      "num_tokens": 3441744.0,
+      "step": 279
+    },
+    {
+      "entropy": 1.2391767501831055,
+      "epoch": 0.1474460242232754,
+      "grad_norm": 0.5528289079666138,
+      "learning_rate": 0.0001398496240601504,
+      "loss": 0.2562858760356903,
+      "mean_token_accuracy": 0.905647337436676,
+      "num_tokens": 3454080.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.3013744056224823,
+      "epoch": 0.14797261716692997,
+      "grad_norm": 0.5353385806083679,
+      "learning_rate": 0.00014035087719298245,
+      "loss": 0.24675647914409637,
+      "mean_token_accuracy": 0.9100503474473953,
+      "num_tokens": 3466416.0,
+      "step": 281
+    },
+    {
+      "entropy": 1.2755855023860931,
+      "epoch": 0.1484992101105845,
+      "grad_norm": 0.5495297908782959,
+      "learning_rate": 0.00014085213032581455,
+      "loss": 0.2519223690032959,
+      "mean_token_accuracy": 0.8966221362352371,
+      "num_tokens": 3478752.0,
+      "step": 282
+    },
+    {
+      "entropy": 1.303219497203827,
+      "epoch": 0.14902580305423907,
+      "grad_norm": 0.5309876203536987,
+      "learning_rate": 0.0001413533834586466,
+      "loss": 0.24427109956741333,
+      "mean_token_accuracy": 0.9075532704591751,
+      "num_tokens": 3491088.0,
+      "step": 283
+    },
+    {
+      "entropy": 1.2614858448505402,
+      "epoch": 0.14955239599789363,
+      "grad_norm": 0.5807812809944153,
+      "learning_rate": 0.0001418546365914787,
+      "loss": 0.26809874176979065,
+      "mean_token_accuracy": 0.899653896689415,
+      "num_tokens": 3503424.0,
+      "step": 284
+    },
+    {
+      "entropy": 1.2611050605773926,
+      "epoch": 0.1500789889415482,
+      "grad_norm": 0.5592539310455322,
+      "learning_rate": 0.0001423558897243108,
+      "loss": 0.26588040590286255,
+      "mean_token_accuracy": 0.8970725983381271,
+      "num_tokens": 3515760.0,
+      "step": 285
+    },
+    {
+      "entropy": 1.2830645143985748,
+      "epoch": 0.15060558188520273,
+      "grad_norm": 0.5146520733833313,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 0.24322782456874847,
+      "mean_token_accuracy": 0.9055588096380234,
+      "num_tokens": 3528096.0,
+      "step": 286
+    },
+    {
+      "entropy": 1.2292780876159668,
+      "epoch": 0.1511321748288573,
+      "grad_norm": 0.588441789150238,
+      "learning_rate": 0.00014335839598997495,
+      "loss": 0.2558269202709198,
+      "mean_token_accuracy": 0.8964600563049316,
+      "num_tokens": 3540432.0,
+      "step": 287
+    },
+    {
+      "entropy": 1.2410956919193268,
+      "epoch": 0.15165876777251186,
+      "grad_norm": 0.5761001706123352,
+      "learning_rate": 0.00014385964912280703,
+      "loss": 0.2511615455150604,
+      "mean_token_accuracy": 0.9040048718452454,
+      "num_tokens": 3552768.0,
+      "step": 288
+    },
+    {
+      "entropy": 1.3154540359973907,
+      "epoch": 0.1521853607161664,
+      "grad_norm": 0.6528503894805908,
+      "learning_rate": 0.0001443609022556391,
+      "loss": 0.2512635290622711,
+      "mean_token_accuracy": 0.9024206399917603,
+      "num_tokens": 3565104.0,
+      "step": 289
+    },
+    {
+      "entropy": 1.2947738468647003,
+      "epoch": 0.15271195365982096,
+      "grad_norm": 0.6061785221099854,
+      "learning_rate": 0.00014486215538847118,
+      "loss": 0.25342220067977905,
+      "mean_token_accuracy": 0.9054669588804245,
+      "num_tokens": 3577440.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.2879831790924072,
+      "epoch": 0.15323854660347552,
+      "grad_norm": 0.5570182800292969,
+      "learning_rate": 0.00014536340852130326,
+      "loss": 0.27056393027305603,
+      "mean_token_accuracy": 0.8992442786693573,
+      "num_tokens": 3589776.0,
+      "step": 291
+    },
+    {
+      "entropy": 1.3449256420135498,
+      "epoch": 0.15376513954713006,
+      "grad_norm": 0.4950920045375824,
+      "learning_rate": 0.00014586466165413534,
+      "loss": 0.24097028374671936,
+      "mean_token_accuracy": 0.9014376401901245,
+      "num_tokens": 3602112.0,
+      "step": 292
+    },
+    {
+      "entropy": 1.4016221165657043,
+      "epoch": 0.15429173249078462,
+      "grad_norm": 0.554709792137146,
+      "learning_rate": 0.00014636591478696742,
+      "loss": 0.2270800769329071,
+      "mean_token_accuracy": 0.9162011295557022,
+      "num_tokens": 3614448.0,
+      "step": 293
+    },
+    {
+      "entropy": 1.514324426651001,
+      "epoch": 0.15481832543443919,
+      "grad_norm": 0.5531271696090698,
+      "learning_rate": 0.0001468671679197995,
+      "loss": 0.24683888256549835,
+      "mean_token_accuracy": 0.9074130207300186,
+      "num_tokens": 3626784.0,
+      "step": 294
+    },
+    {
+      "entropy": 1.471198707818985,
+      "epoch": 0.15534491837809372,
+      "grad_norm": 0.6712215542793274,
+      "learning_rate": 0.00014736842105263158,
+      "loss": 0.281759113073349,
+      "mean_token_accuracy": 0.8955343812704086,
+      "num_tokens": 3639120.0,
+      "step": 295
+    },
+    {
+      "entropy": 1.4630400836467743,
+      "epoch": 0.15587151132174829,
+      "grad_norm": 0.561416506767273,
+      "learning_rate": 0.00014786967418546366,
+      "loss": 0.275588721036911,
+      "mean_token_accuracy": 0.8987540155649185,
+      "num_tokens": 3651456.0,
+      "step": 296
+    },
+    {
+      "entropy": 1.491294264793396,
+      "epoch": 0.15639810426540285,
+      "grad_norm": 0.5332501530647278,
+      "learning_rate": 0.00014837092731829574,
+      "loss": 0.2502642273902893,
+      "mean_token_accuracy": 0.9047496467828751,
+      "num_tokens": 3663792.0,
+      "step": 297
+    },
+    {
+      "entropy": 1.3809879124164581,
+      "epoch": 0.1569246972090574,
+      "grad_norm": 0.499104768037796,
+      "learning_rate": 0.00014887218045112784,
+      "loss": 0.23079532384872437,
+      "mean_token_accuracy": 0.909051313996315,
+      "num_tokens": 3676128.0,
+      "step": 298
+    },
+    {
+      "entropy": 1.3202951848506927,
+      "epoch": 0.15745129015271195,
+      "grad_norm": 0.5114240646362305,
+      "learning_rate": 0.0001493734335839599,
+      "loss": 0.26147356629371643,
+      "mean_token_accuracy": 0.9033501297235489,
+      "num_tokens": 3688464.0,
+      "step": 299
+    },
+    {
+      "entropy": 1.3542028069496155,
+      "epoch": 0.1579778830963665,
+      "grad_norm": 0.492830753326416,
+      "learning_rate": 0.000149874686716792,
+      "loss": 0.2421473264694214,
+      "mean_token_accuracy": 0.9081345945596695,
+      "num_tokens": 3700800.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.2910203635692596,
+      "epoch": 0.15850447604002108,
+      "grad_norm": 0.4862864911556244,
+      "learning_rate": 0.00015037593984962405,
+      "loss": 0.24877643585205078,
+      "mean_token_accuracy": 0.9057342112064362,
+      "num_tokens": 3713136.0,
+      "step": 301
+    },
+    {
+      "entropy": 1.3562625646591187,
+      "epoch": 0.1590310689836756,
+      "grad_norm": 0.6064998507499695,
+      "learning_rate": 0.00015087719298245616,
+      "loss": 0.2678510546684265,
+      "mean_token_accuracy": 0.895992174744606,
+      "num_tokens": 3725472.0,
+      "step": 302
+    },
+    {
+      "entropy": 1.3357418477535248,
+      "epoch": 0.15955766192733017,
+      "grad_norm": 0.5504618883132935,
+      "learning_rate": 0.0001513784461152882,
+      "loss": 0.24523413181304932,
+      "mean_token_accuracy": 0.9103363454341888,
+      "num_tokens": 3737808.0,
+      "step": 303
+    },
+    {
+      "entropy": 1.3335883915424347,
+      "epoch": 0.16008425487098474,
+      "grad_norm": 0.73444664478302,
+      "learning_rate": 0.00015187969924812032,
+      "loss": 0.2613740563392639,
+      "mean_token_accuracy": 0.8996832072734833,
+      "num_tokens": 3750144.0,
+      "step": 304
+    },
+    {
+      "entropy": 1.3423579335212708,
+      "epoch": 0.16061084781463927,
+      "grad_norm": 0.5342377424240112,
+      "learning_rate": 0.00015238095238095237,
+      "loss": 0.27072861790657043,
+      "mean_token_accuracy": 0.8979348689317703,
+      "num_tokens": 3762480.0,
+      "step": 305
+    },
+    {
+      "entropy": 1.2893436253070831,
+      "epoch": 0.16113744075829384,
+      "grad_norm": 0.49320632219314575,
+      "learning_rate": 0.00015288220551378448,
+      "loss": 0.24137352406978607,
+      "mean_token_accuracy": 0.9123422801494598,
+      "num_tokens": 3774816.0,
+      "step": 306
+    },
+    {
+      "entropy": 1.3107283115386963,
+      "epoch": 0.1616640337019484,
+      "grad_norm": 0.5037325620651245,
+      "learning_rate": 0.00015338345864661653,
+      "loss": 0.28284943103790283,
+      "mean_token_accuracy": 0.8933527320623398,
+      "num_tokens": 3787152.0,
+      "step": 307
+    },
+    {
+      "entropy": 1.2858748137950897,
+      "epoch": 0.16219062664560294,
+      "grad_norm": 0.7596031427383423,
+      "learning_rate": 0.00015388471177944863,
+      "loss": 0.2364031970500946,
+      "mean_token_accuracy": 0.9070021212100983,
+      "num_tokens": 3799488.0,
+      "step": 308
+    },
+    {
+      "entropy": 1.2649638950824738,
+      "epoch": 0.1627172195892575,
+      "grad_norm": 0.4712359607219696,
+      "learning_rate": 0.0001543859649122807,
+      "loss": 0.25866541266441345,
+      "mean_token_accuracy": 0.8989972174167633,
+      "num_tokens": 3811824.0,
+      "step": 309
+    },
+    {
+      "entropy": 1.1775002479553223,
+      "epoch": 0.16324381253291206,
+      "grad_norm": 0.47527384757995605,
+      "learning_rate": 0.0001548872180451128,
+      "loss": 0.22686654329299927,
+      "mean_token_accuracy": 0.9112849235534668,
+      "num_tokens": 3824160.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1848091185092926,
+      "epoch": 0.16377040547656663,
+      "grad_norm": 0.5009036064147949,
+      "learning_rate": 0.00015538847117794487,
+      "loss": 0.23199157416820526,
+      "mean_token_accuracy": 0.9103835225105286,
+      "num_tokens": 3836496.0,
+      "step": 311
+    },
+    {
+      "entropy": 1.1879796385765076,
+      "epoch": 0.16429699842022116,
+      "grad_norm": 0.530559778213501,
+      "learning_rate": 0.00015588972431077695,
+      "loss": 0.2687099277973175,
+      "mean_token_accuracy": 0.9024157524108887,
+      "num_tokens": 3848832.0,
+      "step": 312
+    },
+    {
+      "entropy": 1.1956203877925873,
+      "epoch": 0.16482359136387573,
+      "grad_norm": 0.5160781145095825,
+      "learning_rate": 0.00015639097744360903,
+      "loss": 0.24907854199409485,
+      "mean_token_accuracy": 0.9046753495931625,
+      "num_tokens": 3861168.0,
+      "step": 313
+    },
+    {
+      "entropy": 1.1685654819011688,
+      "epoch": 0.1653501843075303,
+      "grad_norm": 0.5312115550041199,
+      "learning_rate": 0.0001568922305764411,
+      "loss": 0.2576726973056793,
+      "mean_token_accuracy": 0.9074427336454391,
+      "num_tokens": 3873504.0,
+      "step": 314
+    },
+    {
+      "entropy": 1.1273577511310577,
+      "epoch": 0.16587677725118483,
+      "grad_norm": 0.6181286573410034,
+      "learning_rate": 0.00015739348370927319,
+      "loss": 0.23891520500183105,
+      "mean_token_accuracy": 0.9119571298360825,
+      "num_tokens": 3885840.0,
+      "step": 315
+    },
+    {
+      "entropy": 1.2541460990905762,
+      "epoch": 0.1664033701948394,
+      "grad_norm": 0.4766935408115387,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 0.25483545660972595,
+      "mean_token_accuracy": 0.8961910307407379,
+      "num_tokens": 3898176.0,
+      "step": 316
+    },
+    {
+      "entropy": 1.1125807464122772,
+      "epoch": 0.16692996313849395,
+      "grad_norm": 0.5105007886886597,
+      "learning_rate": 0.00015839598997493734,
+      "loss": 0.2685958445072174,
+      "mean_token_accuracy": 0.8989451825618744,
+      "num_tokens": 3910512.0,
+      "step": 317
+    },
+    {
+      "entropy": 1.2121298015117645,
+      "epoch": 0.1674565560821485,
+      "grad_norm": 0.44646549224853516,
+      "learning_rate": 0.00015889724310776942,
+      "loss": 0.253686785697937,
+      "mean_token_accuracy": 0.9083440899848938,
+      "num_tokens": 3922848.0,
+      "step": 318
+    },
+    {
+      "entropy": 1.1011106967926025,
+      "epoch": 0.16798314902580305,
+      "grad_norm": 0.46656280755996704,
+      "learning_rate": 0.0001593984962406015,
+      "loss": 0.2927643060684204,
+      "mean_token_accuracy": 0.8931492418050766,
+      "num_tokens": 3935184.0,
+      "step": 319
+    },
+    {
+      "entropy": 1.1864506900310516,
+      "epoch": 0.16850974196945762,
+      "grad_norm": 0.4583514332771301,
+      "learning_rate": 0.00015989974937343358,
+      "loss": 0.22579318284988403,
+      "mean_token_accuracy": 0.9128856211900711,
+      "num_tokens": 3947520.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.113357663154602,
+      "epoch": 0.16903633491311215,
+      "grad_norm": 0.5008531212806702,
+      "learning_rate": 0.00016040100250626566,
+      "loss": 0.2566073536872864,
+      "mean_token_accuracy": 0.9067006707191467,
+      "num_tokens": 3959856.0,
+      "step": 321
+    },
+    {
+      "entropy": 1.1666322350502014,
+      "epoch": 0.16956292785676672,
+      "grad_norm": 0.5016852021217346,
+      "learning_rate": 0.00016090225563909777,
+      "loss": 0.24807997047901154,
+      "mean_token_accuracy": 0.9074774533510208,
+      "num_tokens": 3972192.0,
+      "step": 322
+    },
+    {
+      "entropy": 1.0860579013824463,
+      "epoch": 0.17008952080042128,
+      "grad_norm": 0.5153301358222961,
+      "learning_rate": 0.00016140350877192982,
+      "loss": 0.2500810921192169,
+      "mean_token_accuracy": 0.9049581736326218,
+      "num_tokens": 3984528.0,
+      "step": 323
+    },
+    {
+      "entropy": 1.0754130184650421,
+      "epoch": 0.17061611374407584,
+      "grad_norm": 0.5485661625862122,
+      "learning_rate": 0.00016190476190476192,
+      "loss": 0.3087775707244873,
+      "mean_token_accuracy": 0.8840746134519577,
+      "num_tokens": 3996864.0,
+      "step": 324
+    },
+    {
+      "entropy": 1.097093939781189,
+      "epoch": 0.17114270668773038,
+      "grad_norm": 0.4524535834789276,
+      "learning_rate": 0.00016240601503759398,
+      "loss": 0.2556874454021454,
+      "mean_token_accuracy": 0.9040686786174774,
+      "num_tokens": 4009200.0,
+      "step": 325
+    },
+    {
+      "entropy": 1.0161993354558945,
+      "epoch": 0.17166929963138494,
+      "grad_norm": 0.5920532941818237,
+      "learning_rate": 0.00016290726817042608,
+      "loss": 0.2545970678329468,
+      "mean_token_accuracy": 0.9038624912500381,
+      "num_tokens": 4021536.0,
+      "step": 326
+    },
+    {
+      "entropy": 1.045365571975708,
+      "epoch": 0.1721958925750395,
+      "grad_norm": 0.5128290057182312,
+      "learning_rate": 0.00016340852130325813,
+      "loss": 0.23849724233150482,
+      "mean_token_accuracy": 0.9063487350940704,
+      "num_tokens": 4033872.0,
+      "step": 327
+    },
+    {
+      "entropy": 1.0524345636367798,
+      "epoch": 0.17272248551869404,
+      "grad_norm": 0.5058865547180176,
+      "learning_rate": 0.00016390977443609024,
+      "loss": 0.25231170654296875,
+      "mean_token_accuracy": 0.9029942154884338,
+      "num_tokens": 4046208.0,
+      "step": 328
+    },
+    {
+      "entropy": 1.1274868845939636,
+      "epoch": 0.1732490784623486,
+      "grad_norm": 0.5103746652603149,
+      "learning_rate": 0.0001644110275689223,
+      "loss": 0.26119422912597656,
+      "mean_token_accuracy": 0.8970126211643219,
+      "num_tokens": 4058544.0,
+      "step": 329
+    },
+    {
+      "entropy": 1.13772451877594,
+      "epoch": 0.17377567140600317,
+      "grad_norm": 0.463789701461792,
+      "learning_rate": 0.0001649122807017544,
+      "loss": 0.23172324895858765,
+      "mean_token_accuracy": 0.9090371131896973,
+      "num_tokens": 4070880.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1093823909759521,
+      "epoch": 0.1743022643496577,
+      "grad_norm": 0.4940638542175293,
+      "learning_rate": 0.00016541353383458648,
+      "loss": 0.25032839179039,
+      "mean_token_accuracy": 0.9061925411224365,
+      "num_tokens": 4083216.0,
+      "step": 331
+    },
+    {
+      "entropy": 1.0906598567962646,
+      "epoch": 0.17482885729331227,
+      "grad_norm": 0.5244042873382568,
+      "learning_rate": 0.00016591478696741856,
+      "loss": 0.2559994161128998,
+      "mean_token_accuracy": 0.9056536555290222,
+      "num_tokens": 4095552.0,
+      "step": 332
+    },
+    {
+      "entropy": 1.0064781606197357,
+      "epoch": 0.17535545023696683,
+      "grad_norm": 0.5328090786933899,
+      "learning_rate": 0.00016641604010025064,
+      "loss": 0.2423868030309677,
+      "mean_token_accuracy": 0.9023227244615555,
+      "num_tokens": 4107888.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.9421130269765854,
+      "epoch": 0.17588204318062137,
+      "grad_norm": 0.4938746392726898,
+      "learning_rate": 0.00016691729323308271,
+      "loss": 0.2487686276435852,
+      "mean_token_accuracy": 0.9072201699018478,
+      "num_tokens": 4120224.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.8471245914697647,
+      "epoch": 0.17640863612427593,
+      "grad_norm": 0.4719039499759674,
+      "learning_rate": 0.0001674185463659148,
+      "loss": 0.2677307426929474,
+      "mean_token_accuracy": 0.89954973757267,
+      "num_tokens": 4132560.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.7532882988452911,
+      "epoch": 0.1769352290679305,
+      "grad_norm": 0.4836346209049225,
+      "learning_rate": 0.00016791979949874687,
+      "loss": 0.2508479356765747,
+      "mean_token_accuracy": 0.9021095633506775,
+      "num_tokens": 4144896.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.6489329487085342,
+      "epoch": 0.17746182201158506,
+      "grad_norm": 0.5742446184158325,
+      "learning_rate": 0.00016842105263157895,
+      "loss": 0.2582406997680664,
+      "mean_token_accuracy": 0.9013715237379074,
+      "num_tokens": 4157232.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.6537502855062485,
+      "epoch": 0.1779884149552396,
+      "grad_norm": 0.5390785336494446,
+      "learning_rate": 0.00016892230576441103,
+      "loss": 0.2649543881416321,
+      "mean_token_accuracy": 0.8998610377311707,
+      "num_tokens": 4169568.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.6827399134635925,
+      "epoch": 0.17851500789889416,
+      "grad_norm": 0.507554829120636,
+      "learning_rate": 0.0001694235588972431,
+      "loss": 0.27018067240715027,
+      "mean_token_accuracy": 0.8985839486122131,
+      "num_tokens": 4181904.0,
+      "step": 339
+    },
+    {
+      "entropy": 0.7333462089300156,
+      "epoch": 0.17904160084254872,
+      "grad_norm": 0.4551506042480469,
+      "learning_rate": 0.0001699248120300752,
+      "loss": 0.25013625621795654,
+      "mean_token_accuracy": 0.9111319333314896,
+      "num_tokens": 4194240.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7495439946651459,
+      "epoch": 0.17956819378620326,
+      "grad_norm": 0.48456624150276184,
+      "learning_rate": 0.00017042606516290727,
+      "loss": 0.2707773745059967,
+      "mean_token_accuracy": 0.8947316855192184,
+      "num_tokens": 4206576.0,
+      "step": 341
+    },
+    {
+      "entropy": 0.7812775671482086,
+      "epoch": 0.18009478672985782,
+      "grad_norm": 0.4550190269947052,
+      "learning_rate": 0.00017092731829573935,
+      "loss": 0.2396947741508484,
+      "mean_token_accuracy": 0.910267636179924,
+      "num_tokens": 4218912.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.9400374740362167,
+      "epoch": 0.18062137967351238,
+      "grad_norm": 0.5784738659858704,
+      "learning_rate": 0.00017142857142857143,
+      "loss": 0.26176217198371887,
+      "mean_token_accuracy": 0.9009139388799667,
+      "num_tokens": 4231248.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.937469407916069,
+      "epoch": 0.18114797261716692,
+      "grad_norm": 0.5101773738861084,
+      "learning_rate": 0.00017192982456140353,
+      "loss": 0.222784623503685,
+      "mean_token_accuracy": 0.9079811573028564,
+      "num_tokens": 4243584.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.9701919555664062,
+      "epoch": 0.18167456556082148,
+      "grad_norm": 0.4810556471347809,
+      "learning_rate": 0.00017243107769423558,
+      "loss": 0.25527724623680115,
+      "mean_token_accuracy": 0.8977989107370377,
+      "num_tokens": 4255920.0,
+      "step": 345
+    },
+    {
+      "entropy": 1.0421793162822723,
+      "epoch": 0.18220115850447605,
+      "grad_norm": 0.49446040391921997,
+      "learning_rate": 0.0001729323308270677,
+      "loss": 0.24790042638778687,
+      "mean_token_accuracy": 0.9076082557439804,
+      "num_tokens": 4268256.0,
+      "step": 346
+    },
+    {
+      "entropy": 1.0526159405708313,
+      "epoch": 0.18272775144813058,
+      "grad_norm": 0.5106586217880249,
+      "learning_rate": 0.00017343358395989974,
+      "loss": 0.258137047290802,
+      "mean_token_accuracy": 0.9013570845127106,
+      "num_tokens": 4280592.0,
+      "step": 347
+    },
+    {
+      "entropy": 1.0377429127693176,
+      "epoch": 0.18325434439178515,
+      "grad_norm": 0.5372883677482605,
+      "learning_rate": 0.00017393483709273185,
+      "loss": 0.2536155879497528,
+      "mean_token_accuracy": 0.9079491943120956,
+      "num_tokens": 4292928.0,
+      "step": 348
+    },
+    {
+      "entropy": 1.1226612329483032,
+      "epoch": 0.1837809373354397,
+      "grad_norm": 0.506536066532135,
+      "learning_rate": 0.0001744360902255639,
+      "loss": 0.2503521740436554,
+      "mean_token_accuracy": 0.9043506681919098,
+      "num_tokens": 4305264.0,
+      "step": 349
+    },
+    {
+      "entropy": 1.0955768823623657,
+      "epoch": 0.18430753027909427,
+      "grad_norm": 0.450717031955719,
+      "learning_rate": 0.000174937343358396,
+      "loss": 0.2611716687679291,
+      "mean_token_accuracy": 0.9029702991247177,
+      "num_tokens": 4317600.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1000354140996933,
+      "epoch": 0.1848341232227488,
+      "grad_norm": 0.5249627828598022,
+      "learning_rate": 0.00017543859649122806,
+      "loss": 0.23257189989089966,
+      "mean_token_accuracy": 0.9120175540447235,
+      "num_tokens": 4329936.0,
+      "step": 351
+    },
+    {
+      "entropy": 1.086414873600006,
+      "epoch": 0.18536071616640337,
+      "grad_norm": 0.488702654838562,
+      "learning_rate": 0.00017593984962406016,
+      "loss": 0.246709406375885,
+      "mean_token_accuracy": 0.9029638320207596,
+      "num_tokens": 4342272.0,
+      "step": 352
+    },
+    {
+      "entropy": 1.0042328536510468,
+      "epoch": 0.18588730911005794,
+      "grad_norm": 0.41849616169929504,
+      "learning_rate": 0.00017644110275689224,
+      "loss": 0.21574650704860687,
+      "mean_token_accuracy": 0.9175253808498383,
+      "num_tokens": 4354608.0,
+      "step": 353
+    },
+    {
+      "entropy": 1.0020455569028854,
+      "epoch": 0.18641390205371247,
+      "grad_norm": 0.46401116251945496,
+      "learning_rate": 0.00017694235588972432,
+      "loss": 0.2567673325538635,
+      "mean_token_accuracy": 0.9054214805364609,
+      "num_tokens": 4366944.0,
+      "step": 354
+    },
+    {
+      "entropy": 1.0092186033725739,
+      "epoch": 0.18694049499736703,
+      "grad_norm": 0.47048723697662354,
+      "learning_rate": 0.0001774436090225564,
+      "loss": 0.24711501598358154,
+      "mean_token_accuracy": 0.9056050777435303,
+      "num_tokens": 4379280.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.9878545552492142,
+      "epoch": 0.1874670879410216,
+      "grad_norm": 0.47208353877067566,
+      "learning_rate": 0.00017794486215538848,
+      "loss": 0.2748403549194336,
+      "mean_token_accuracy": 0.8964878469705582,
+      "num_tokens": 4391616.0,
+      "step": 356
+    },
+    {
+      "entropy": 0.9895420670509338,
+      "epoch": 0.18799368088467613,
+      "grad_norm": 0.503665566444397,
+      "learning_rate": 0.00017844611528822056,
+      "loss": 0.24393996596336365,
+      "mean_token_accuracy": 0.9081792533397675,
+      "num_tokens": 4403952.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.983001708984375,
+      "epoch": 0.1885202738283307,
+      "grad_norm": 0.5069933533668518,
+      "learning_rate": 0.00017894736842105264,
+      "loss": 0.2519129514694214,
+      "mean_token_accuracy": 0.9055497497320175,
+      "num_tokens": 4416288.0,
+      "step": 358
+    },
+    {
+      "entropy": 0.9950737804174423,
+      "epoch": 0.18904686677198526,
+      "grad_norm": 0.4974028766155243,
+      "learning_rate": 0.00017944862155388472,
+      "loss": 0.26243290305137634,
+      "mean_token_accuracy": 0.8986903131008148,
+      "num_tokens": 4428624.0,
+      "step": 359
+    },
+    {
+      "entropy": 1.0079210847616196,
+      "epoch": 0.1895734597156398,
+      "grad_norm": 0.5110583901405334,
+      "learning_rate": 0.0001799498746867168,
+      "loss": 0.2294146716594696,
+      "mean_token_accuracy": 0.9131651520729065,
+      "num_tokens": 4440960.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0026690363883972,
+      "epoch": 0.19010005265929436,
+      "grad_norm": 0.4598653316497803,
+      "learning_rate": 0.00018045112781954887,
+      "loss": 0.2722235321998596,
+      "mean_token_accuracy": 0.8976626545190811,
+      "num_tokens": 4453296.0,
+      "step": 361
+    },
+    {
+      "entropy": 1.0199629068374634,
+      "epoch": 0.19062664560294892,
+      "grad_norm": 0.4637596905231476,
+      "learning_rate": 0.00018095238095238095,
+      "loss": 0.2665867507457733,
+      "mean_token_accuracy": 0.9000826925039291,
+      "num_tokens": 4465632.0,
+      "step": 362
+    },
+    {
+      "entropy": 1.0425358265638351,
+      "epoch": 0.1911532385466035,
+      "grad_norm": 0.47371116280555725,
+      "learning_rate": 0.00018145363408521303,
+      "loss": 0.261046439409256,
+      "mean_token_accuracy": 0.8978727906942368,
+      "num_tokens": 4477968.0,
+      "step": 363
+    },
+    {
+      "entropy": 1.0688822865486145,
+      "epoch": 0.19167983149025802,
+      "grad_norm": 0.588598370552063,
+      "learning_rate": 0.0001819548872180451,
+      "loss": 0.2722393274307251,
+      "mean_token_accuracy": 0.8954607844352722,
+      "num_tokens": 4490304.0,
+      "step": 364
+    },
+    {
+      "entropy": 1.0311770141124725,
+      "epoch": 0.1922064244339126,
+      "grad_norm": 0.6127772331237793,
+      "learning_rate": 0.0001824561403508772,
+      "loss": 0.26557645201683044,
+      "mean_token_accuracy": 0.898269534111023,
+      "num_tokens": 4502640.0,
+      "step": 365
+    },
+    {
+      "entropy": 1.023191675543785,
+      "epoch": 0.19273301737756715,
+      "grad_norm": 0.4277988076210022,
+      "learning_rate": 0.0001829573934837093,
+      "loss": 0.25198060274124146,
+      "mean_token_accuracy": 0.9041500091552734,
+      "num_tokens": 4514976.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.992491751909256,
+      "epoch": 0.1932596103212217,
+      "grad_norm": 0.4713631272315979,
+      "learning_rate": 0.00018345864661654135,
+      "loss": 0.25175848603248596,
+      "mean_token_accuracy": 0.9065804481506348,
+      "num_tokens": 4527312.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.9969351142644882,
+      "epoch": 0.19378620326487625,
+      "grad_norm": 0.4729400873184204,
+      "learning_rate": 0.00018395989974937345,
+      "loss": 0.2526015341281891,
+      "mean_token_accuracy": 0.9010139107704163,
+      "num_tokens": 4539648.0,
+      "step": 368
+    },
+    {
+      "entropy": 1.0536664575338364,
+      "epoch": 0.1943127962085308,
+      "grad_norm": 0.5235168933868408,
+      "learning_rate": 0.0001844611528822055,
+      "loss": 0.292032927274704,
+      "mean_token_accuracy": 0.8931012600660324,
+      "num_tokens": 4551984.0,
+      "step": 369
+    },
+    {
+      "entropy": 1.042996197938919,
+      "epoch": 0.19483938915218535,
+      "grad_norm": 0.48083820939064026,
+      "learning_rate": 0.0001849624060150376,
+      "loss": 0.26578348875045776,
+      "mean_token_accuracy": 0.9036569446325302,
+      "num_tokens": 4564320.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.0278095752000809,
+      "epoch": 0.1953659820958399,
+      "grad_norm": 0.39684435725212097,
+      "learning_rate": 0.00018546365914786966,
+      "loss": 0.23184379935264587,
+      "mean_token_accuracy": 0.9090972989797592,
+      "num_tokens": 4576656.0,
+      "step": 371
+    },
+    {
+      "entropy": 1.0532077848911285,
+      "epoch": 0.19589257503949448,
+      "grad_norm": 0.40189671516418457,
+      "learning_rate": 0.00018596491228070177,
+      "loss": 0.22099974751472473,
+      "mean_token_accuracy": 0.9153455793857574,
+      "num_tokens": 4588992.0,
+      "step": 372
+    },
+    {
+      "entropy": 1.090534657239914,
+      "epoch": 0.196419167983149,
+      "grad_norm": 0.4149918258190155,
+      "learning_rate": 0.00018646616541353382,
+      "loss": 0.26646703481674194,
+      "mean_token_accuracy": 0.8960744440555573,
+      "num_tokens": 4601328.0,
+      "step": 373
+    },
+    {
+      "entropy": 1.1020061373710632,
+      "epoch": 0.19694576092680358,
+      "grad_norm": 0.4702380299568176,
+      "learning_rate": 0.00018696741854636593,
+      "loss": 0.221252903342247,
+      "mean_token_accuracy": 0.9114280194044113,
+      "num_tokens": 4613664.0,
+      "step": 374
+    },
+    {
+      "entropy": 1.0703049004077911,
+      "epoch": 0.19747235387045814,
+      "grad_norm": 0.3877512216567993,
+      "learning_rate": 0.00018746867167919798,
+      "loss": 0.2393355667591095,
+      "mean_token_accuracy": 0.9077306538820267,
+      "num_tokens": 4626000.0,
+      "step": 375
+    },
+    {
+      "entropy": 1.1242592930793762,
+      "epoch": 0.1979989468141127,
+      "grad_norm": 0.4995574355125427,
+      "learning_rate": 0.00018796992481203009,
+      "loss": 0.26316261291503906,
+      "mean_token_accuracy": 0.9027625322341919,
+      "num_tokens": 4638336.0,
+      "step": 376
+    },
+    {
+      "entropy": 1.1577873229980469,
+      "epoch": 0.19852553975776724,
+      "grad_norm": 0.5234032273292542,
+      "learning_rate": 0.00018847117794486217,
+      "loss": 0.27375704050064087,
+      "mean_token_accuracy": 0.8979371786117554,
+      "num_tokens": 4650672.0,
+      "step": 377
+    },
+    {
+      "entropy": 1.1344494223594666,
+      "epoch": 0.1990521327014218,
+      "grad_norm": 0.5000519752502441,
+      "learning_rate": 0.00018897243107769424,
+      "loss": 0.26018521189689636,
+      "mean_token_accuracy": 0.9034091979265213,
+      "num_tokens": 4663008.0,
+      "step": 378
+    },
+    {
+      "entropy": 1.2858175039291382,
+      "epoch": 0.19957872564507637,
+      "grad_norm": 0.4786364436149597,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 0.2609444260597229,
+      "mean_token_accuracy": 0.8963579833507538,
+      "num_tokens": 4675344.0,
+      "step": 379
+    },
+    {
+      "entropy": 1.205563873052597,
+      "epoch": 0.2001053185887309,
+      "grad_norm": 0.49225375056266785,
+      "learning_rate": 0.0001899749373433584,
+      "loss": 0.2702328562736511,
+      "mean_token_accuracy": 0.89518603682518,
+      "num_tokens": 4687680.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.2623814940452576,
+      "epoch": 0.20063191153238547,
+      "grad_norm": 0.4847429692745209,
+      "learning_rate": 0.00019047619047619048,
+      "loss": 0.23678088188171387,
+      "mean_token_accuracy": 0.9141177982091904,
+      "num_tokens": 4700016.0,
+      "step": 381
+    },
+    {
+      "entropy": 1.2621397376060486,
+      "epoch": 0.20115850447604003,
+      "grad_norm": 0.42388200759887695,
+      "learning_rate": 0.00019097744360902256,
+      "loss": 0.22780729830265045,
+      "mean_token_accuracy": 0.9121877998113632,
+      "num_tokens": 4712352.0,
+      "step": 382
+    },
+    {
+      "entropy": 1.292501300573349,
+      "epoch": 0.20168509741969456,
+      "grad_norm": 0.4779922366142273,
+      "learning_rate": 0.00019147869674185464,
+      "loss": 0.2594977915287018,
+      "mean_token_accuracy": 0.9015758037567139,
+      "num_tokens": 4724688.0,
+      "step": 383
+    },
+    {
+      "entropy": 1.2592476904392242,
+      "epoch": 0.20221169036334913,
+      "grad_norm": 0.5031237006187439,
+      "learning_rate": 0.00019197994987468672,
+      "loss": 0.2655055820941925,
+      "mean_token_accuracy": 0.8991453051567078,
+      "num_tokens": 4737024.0,
+      "step": 384
+    },
+    {
+      "entropy": 1.2670154571533203,
+      "epoch": 0.2027382833070037,
+      "grad_norm": 0.43786391615867615,
+      "learning_rate": 0.0001924812030075188,
+      "loss": 0.2638092637062073,
+      "mean_token_accuracy": 0.9021878391504288,
+      "num_tokens": 4749360.0,
+      "step": 385
+    },
+    {
+      "entropy": 1.2625199258327484,
+      "epoch": 0.20326487625065823,
+      "grad_norm": 0.6012351512908936,
+      "learning_rate": 0.00019298245614035088,
+      "loss": 0.25398504734039307,
+      "mean_token_accuracy": 0.9037431925535202,
+      "num_tokens": 4761696.0,
+      "step": 386
+    },
+    {
+      "entropy": 1.2534445226192474,
+      "epoch": 0.2037914691943128,
+      "grad_norm": 0.48004376888275146,
+      "learning_rate": 0.00019348370927318296,
+      "loss": 0.24518710374832153,
+      "mean_token_accuracy": 0.9088338762521744,
+      "num_tokens": 4774032.0,
+      "step": 387
+    },
+    {
+      "entropy": 1.2239902317523956,
+      "epoch": 0.20431806213796735,
+      "grad_norm": 0.40316712856292725,
+      "learning_rate": 0.00019398496240601503,
+      "loss": 0.22658605873584747,
+      "mean_token_accuracy": 0.9146187752485275,
+      "num_tokens": 4786368.0,
+      "step": 388
+    },
+    {
+      "entropy": 1.236211746931076,
+      "epoch": 0.20484465508162192,
+      "grad_norm": 0.4805733859539032,
+      "learning_rate": 0.0001944862155388471,
+      "loss": 0.2645362615585327,
+      "mean_token_accuracy": 0.899237260222435,
+      "num_tokens": 4798704.0,
+      "step": 389
+    },
+    {
+      "entropy": 1.2168347239494324,
+      "epoch": 0.20537124802527645,
+      "grad_norm": 0.4017006754875183,
+      "learning_rate": 0.00019498746867167922,
+      "loss": 0.23945499956607819,
+      "mean_token_accuracy": 0.9106440991163254,
+      "num_tokens": 4811040.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.2192648351192474,
+      "epoch": 0.20589784096893102,
+      "grad_norm": 0.4144308269023895,
+      "learning_rate": 0.00019548872180451127,
+      "loss": 0.24552568793296814,
+      "mean_token_accuracy": 0.9093390554189682,
+      "num_tokens": 4823376.0,
+      "step": 391
+    },
+    {
+      "entropy": 1.2436101734638214,
+      "epoch": 0.20642443391258558,
+      "grad_norm": 0.4212309420108795,
+      "learning_rate": 0.00019598997493734338,
+      "loss": 0.23153892159461975,
+      "mean_token_accuracy": 0.9111403673887253,
+      "num_tokens": 4835712.0,
+      "step": 392
+    },
+    {
+      "entropy": 1.255335807800293,
+      "epoch": 0.20695102685624012,
+      "grad_norm": 0.40789833664894104,
+      "learning_rate": 0.00019649122807017543,
+      "loss": 0.22183379530906677,
+      "mean_token_accuracy": 0.9136195629835129,
+      "num_tokens": 4848048.0,
+      "step": 393
+    },
+    {
+      "entropy": 1.2751134932041168,
+      "epoch": 0.20747761979989468,
+      "grad_norm": 0.4400269091129303,
+      "learning_rate": 0.00019699248120300754,
+      "loss": 0.24014121294021606,
+      "mean_token_accuracy": 0.9085260182619095,
+      "num_tokens": 4860384.0,
+      "step": 394
+    },
+    {
+      "entropy": 1.360796719789505,
+      "epoch": 0.20800421274354924,
+      "grad_norm": 0.4804284870624542,
+      "learning_rate": 0.0001974937343358396,
+      "loss": 0.25582095980644226,
+      "mean_token_accuracy": 0.9006912261247635,
+      "num_tokens": 4872720.0,
+      "step": 395
+    },
+    {
+      "entropy": 1.3161850571632385,
+      "epoch": 0.20853080568720378,
+      "grad_norm": 0.43244144320487976,
+      "learning_rate": 0.0001979949874686717,
+      "loss": 0.2349988967180252,
+      "mean_token_accuracy": 0.9116800725460052,
+      "num_tokens": 4885056.0,
+      "step": 396
+    },
+    {
+      "entropy": 1.3374731242656708,
+      "epoch": 0.20905739863085834,
+      "grad_norm": 0.4748205840587616,
+      "learning_rate": 0.00019849624060150375,
+      "loss": 0.22185081243515015,
+      "mean_token_accuracy": 0.9125295132398605,
+      "num_tokens": 4897392.0,
+      "step": 397
+    },
+    {
+      "entropy": 1.3775228261947632,
+      "epoch": 0.2095839915745129,
+      "grad_norm": 0.4581362307071686,
+      "learning_rate": 0.00019899749373433585,
+      "loss": 0.2567383646965027,
+      "mean_token_accuracy": 0.9047663658857346,
+      "num_tokens": 4909728.0,
+      "step": 398
+    },
+    {
+      "entropy": 1.4304566085338593,
+      "epoch": 0.21011058451816747,
+      "grad_norm": 0.5330026149749756,
+      "learning_rate": 0.00019949874686716793,
+      "loss": 0.24410131573677063,
+      "mean_token_accuracy": 0.9004727154970169,
+      "num_tokens": 4922064.0,
+      "step": 399
+    },
+    {
+      "entropy": 1.4176018834114075,
+      "epoch": 0.210637177461822,
+      "grad_norm": 0.5187347531318665,
+      "learning_rate": 0.0002,
+      "loss": 0.2622545063495636,
+      "mean_token_accuracy": 0.8981441259384155,
+      "num_tokens": 4934400.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.3980787992477417,
+      "epoch": 0.21116377040547657,
+      "grad_norm": 0.4053288400173187,
+      "learning_rate": 0.0001999999973286123,
+      "loss": 0.23980936408042908,
+      "mean_token_accuracy": 0.912782147526741,
+      "num_tokens": 4946736.0,
+      "step": 401
+    },
+    {
+      "entropy": 1.4483689367771149,
+      "epoch": 0.21169036334913113,
+      "grad_norm": 0.45620113611221313,
+      "learning_rate": 0.00019999998931444929,
+      "loss": 0.25808754563331604,
+      "mean_token_accuracy": 0.9030732810497284,
+      "num_tokens": 4959072.0,
+      "step": 402
+    },
+    {
+      "entropy": 1.4470272064208984,
+      "epoch": 0.21221695629278567,
+      "grad_norm": 0.45807719230651855,
+      "learning_rate": 0.0001999999759575115,
+      "loss": 0.258093923330307,
+      "mean_token_accuracy": 0.9026659429073334,
+      "num_tokens": 4971408.0,
+      "step": 403
+    },
+    {
+      "entropy": 1.425654649734497,
+      "epoch": 0.21274354923644023,
+      "grad_norm": 0.4397832751274109,
+      "learning_rate": 0.0001999999572577997,
+      "loss": 0.24665014445781708,
+      "mean_token_accuracy": 0.9062764793634415,
+      "num_tokens": 4983744.0,
+      "step": 404
+    },
+    {
+      "entropy": 1.5687190890312195,
+      "epoch": 0.2132701421800948,
+      "grad_norm": 0.5367022156715393,
+      "learning_rate": 0.00019999993321531494,
+      "loss": 0.24174389243125916,
+      "mean_token_accuracy": 0.9081769734621048,
+      "num_tokens": 4996080.0,
+      "step": 405
+    },
+    {
+      "entropy": 1.405859649181366,
+      "epoch": 0.21379673512374933,
+      "grad_norm": 0.4007917046546936,
+      "learning_rate": 0.00019999990383005872,
+      "loss": 0.25402745604515076,
+      "mean_token_accuracy": 0.9025277644395828,
+      "num_tokens": 5008416.0,
+      "step": 406
+    },
+    {
+      "entropy": 1.4313717782497406,
+      "epoch": 0.2143233280674039,
+      "grad_norm": 0.4160802662372589,
+      "learning_rate": 0.00019999986910203282,
+      "loss": 0.24924606084823608,
+      "mean_token_accuracy": 0.9043161869049072,
+      "num_tokens": 5020752.0,
+      "step": 407
+    },
+    {
+      "entropy": 1.4701331555843353,
+      "epoch": 0.21484992101105846,
+      "grad_norm": 0.4316672086715698,
+      "learning_rate": 0.00019999982903123921,
+      "loss": 0.2638508379459381,
+      "mean_token_accuracy": 0.8981801271438599,
+      "num_tokens": 5033088.0,
+      "step": 408
+    },
+    {
+      "entropy": 1.4392390549182892,
+      "epoch": 0.215376513954713,
+      "grad_norm": 0.4544552266597748,
+      "learning_rate": 0.00019999978361768031,
+      "loss": 0.23990359902381897,
+      "mean_token_accuracy": 0.9076667875051498,
+      "num_tokens": 5045424.0,
+      "step": 409
+    },
+    {
+      "entropy": 1.4876410961151123,
+      "epoch": 0.21590310689836756,
+      "grad_norm": 0.4295142889022827,
+      "learning_rate": 0.00019999973286135886,
+      "loss": 0.21840089559555054,
+      "mean_token_accuracy": 0.9214875847101212,
+      "num_tokens": 5057760.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.50434610247612,
+      "epoch": 0.21642969984202212,
+      "grad_norm": 0.4359295070171356,
+      "learning_rate": 0.00019999967676227775,
+      "loss": 0.2510741353034973,
+      "mean_token_accuracy": 0.9052848815917969,
+      "num_tokens": 5070096.0,
+      "step": 411
+    },
+    {
+      "entropy": 1.4829690754413605,
+      "epoch": 0.21695629278567669,
+      "grad_norm": 0.42792809009552,
+      "learning_rate": 0.00019999961532044045,
+      "loss": 0.2551060616970062,
+      "mean_token_accuracy": 0.9011600762605667,
+      "num_tokens": 5082432.0,
+      "step": 412
+    },
+    {
+      "entropy": 1.4852145910263062,
+      "epoch": 0.21748288572933122,
+      "grad_norm": 0.4182438850402832,
+      "learning_rate": 0.00019999954853585052,
+      "loss": 0.24789488315582275,
+      "mean_token_accuracy": 0.9069723039865494,
+      "num_tokens": 5094768.0,
+      "step": 413
+    },
+    {
+      "entropy": 1.4210355281829834,
+      "epoch": 0.21800947867298578,
+      "grad_norm": 0.42366713285446167,
+      "learning_rate": 0.00019999947640851195,
+      "loss": 0.24643118679523468,
+      "mean_token_accuracy": 0.9047895669937134,
+      "num_tokens": 5107104.0,
+      "step": 414
+    },
+    {
+      "entropy": 1.525164395570755,
+      "epoch": 0.21853607161664035,
+      "grad_norm": 0.4372459352016449,
+      "learning_rate": 0.00019999939893842905,
+      "loss": 0.2424328625202179,
+      "mean_token_accuracy": 0.9087924510240555,
+      "num_tokens": 5119440.0,
+      "step": 415
+    },
+    {
+      "entropy": 1.4822115898132324,
+      "epoch": 0.21906266456029488,
+      "grad_norm": 0.4659183919429779,
+      "learning_rate": 0.00019999931612560637,
+      "loss": 0.22431665658950806,
+      "mean_token_accuracy": 0.9165952354669571,
+      "num_tokens": 5131776.0,
+      "step": 416
+    },
+    {
+      "entropy": 1.5467821657657623,
+      "epoch": 0.21958925750394945,
+      "grad_norm": 0.44125568866729736,
+      "learning_rate": 0.00019999922797004884,
+      "loss": 0.23688746988773346,
+      "mean_token_accuracy": 0.9075507670640945,
+      "num_tokens": 5144112.0,
+      "step": 417
+    },
+    {
+      "entropy": 1.4780353605747223,
+      "epoch": 0.220115850447604,
+      "grad_norm": 0.43714895844459534,
+      "learning_rate": 0.00019999913447176174,
+      "loss": 0.24803104996681213,
+      "mean_token_accuracy": 0.9086460769176483,
+      "num_tokens": 5156448.0,
+      "step": 418
+    },
+    {
+      "entropy": 1.442844033241272,
+      "epoch": 0.22064244339125855,
+      "grad_norm": 0.5062248110771179,
+      "learning_rate": 0.00019999903563075051,
+      "loss": 0.2509956657886505,
+      "mean_token_accuracy": 0.9084739983081818,
+      "num_tokens": 5168784.0,
+      "step": 419
+    },
+    {
+      "entropy": 1.472544252872467,
+      "epoch": 0.2211690363349131,
+      "grad_norm": 0.44560346007347107,
+      "learning_rate": 0.00019999893144702116,
+      "loss": 0.2571001648902893,
+      "mean_token_accuracy": 0.9050741344690323,
+      "num_tokens": 5181120.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.4741449654102325,
+      "epoch": 0.22169562927856767,
+      "grad_norm": 0.468660444021225,
+      "learning_rate": 0.00019999882192057975,
+      "loss": 0.2654499113559723,
+      "mean_token_accuracy": 0.9030983746051788,
+      "num_tokens": 5193456.0,
+      "step": 421
+    },
+    {
+      "entropy": 1.4696997106075287,
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.42363062500953674,
+      "learning_rate": 0.0001999987070514329,
+      "loss": 0.2351139783859253,
+      "mean_token_accuracy": 0.9069055020809174,
+      "num_tokens": 5205792.0,
+      "step": 422
+    },
+    {
+      "entropy": 1.452900618314743,
+      "epoch": 0.22274881516587677,
+      "grad_norm": 0.45147648453712463,
+      "learning_rate": 0.00019999858683958726,
+      "loss": 0.236272394657135,
+      "mean_token_accuracy": 0.9084236770868301,
+      "num_tokens": 5218128.0,
+      "step": 423
+    },
+    {
+      "entropy": 1.5096050798892975,
+      "epoch": 0.22327540810953134,
+      "grad_norm": 0.4349573850631714,
+      "learning_rate": 0.00019999846128505015,
+      "loss": 0.21908816695213318,
+      "mean_token_accuracy": 0.9153110533952713,
+      "num_tokens": 5230464.0,
+      "step": 424
+    },
+    {
+      "entropy": 1.5504009425640106,
+      "epoch": 0.2238020010531859,
+      "grad_norm": 0.40080899000167847,
+      "learning_rate": 0.00019999833038782897,
+      "loss": 0.21589110791683197,
+      "mean_token_accuracy": 0.9171598702669144,
+      "num_tokens": 5242800.0,
+      "step": 425
+    },
+    {
+      "entropy": 1.4936735033988953,
+      "epoch": 0.22432859399684044,
+      "grad_norm": 0.4071311056613922,
+      "learning_rate": 0.0001999981941479314,
+      "loss": 0.2125779092311859,
+      "mean_token_accuracy": 0.9237974286079407,
+      "num_tokens": 5255136.0,
+      "step": 426
+    },
+    {
+      "entropy": 1.5081810355186462,
+      "epoch": 0.224855186940495,
+      "grad_norm": 0.4242808222770691,
+      "learning_rate": 0.0001999980525653656,
+      "loss": 0.2447715699672699,
+      "mean_token_accuracy": 0.9042525738477707,
+      "num_tokens": 5267472.0,
+      "step": 427
+    },
+    {
+      "entropy": 1.5077386796474457,
+      "epoch": 0.22538177988414956,
+      "grad_norm": 0.5070521235466003,
+      "learning_rate": 0.00019999790564014,
+      "loss": 0.26209956407546997,
+      "mean_token_accuracy": 0.9026446640491486,
+      "num_tokens": 5279808.0,
+      "step": 428
+    },
+    {
+      "entropy": 1.4962070286273956,
+      "epoch": 0.2259083728278041,
+      "grad_norm": 0.6017259955406189,
+      "learning_rate": 0.00019999775337226327,
+      "loss": 0.3165000081062317,
+      "mean_token_accuracy": 0.8868077397346497,
+      "num_tokens": 5292144.0,
+      "step": 429
+    },
+    {
+      "entropy": 1.4911207258701324,
+      "epoch": 0.22643496577145866,
+      "grad_norm": 0.47006961703300476,
+      "learning_rate": 0.00019999759576174448,
+      "loss": 0.22799812257289886,
+      "mean_token_accuracy": 0.9069435149431229,
+      "num_tokens": 5304480.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.5028018951416016,
+      "epoch": 0.22696155871511323,
+      "grad_norm": 0.4170990288257599,
+      "learning_rate": 0.00019999743280859296,
+      "loss": 0.21681472659111023,
+      "mean_token_accuracy": 0.9184076339006424,
+      "num_tokens": 5316816.0,
+      "step": 431
+    },
+    {
+      "entropy": 1.6698609590530396,
+      "epoch": 0.22748815165876776,
+      "grad_norm": 0.6547060012817383,
+      "learning_rate": 0.0001999972645128184,
+      "loss": 0.24768398702144623,
+      "mean_token_accuracy": 0.9032211750745773,
+      "num_tokens": 5329152.0,
+      "step": 432
+    },
+    {
+      "entropy": 1.549822211265564,
+      "epoch": 0.22801474460242233,
+      "grad_norm": 0.433370441198349,
+      "learning_rate": 0.00019999709087443083,
+      "loss": 0.24651046097278595,
+      "mean_token_accuracy": 0.9072823375463486,
+      "num_tokens": 5341488.0,
+      "step": 433
+    },
+    {
+      "entropy": 1.5724502205848694,
+      "epoch": 0.2285413375460769,
+      "grad_norm": 0.5081071853637695,
+      "learning_rate": 0.00019999691189344045,
+      "loss": 0.2539141774177551,
+      "mean_token_accuracy": 0.901251494884491,
+      "num_tokens": 5353824.0,
+      "step": 434
+    },
+    {
+      "entropy": 1.580730140209198,
+      "epoch": 0.22906793048973143,
+      "grad_norm": 0.4615112841129303,
+      "learning_rate": 0.000199996727569858,
+      "loss": 0.23179525136947632,
+      "mean_token_accuracy": 0.9096536040306091,
+      "num_tokens": 5366160.0,
+      "step": 435
+    },
+    {
+      "entropy": 1.6549272239208221,
+      "epoch": 0.229594523433386,
+      "grad_norm": 0.6388656497001648,
+      "learning_rate": 0.00019999653790369438,
+      "loss": 0.29721513390541077,
+      "mean_token_accuracy": 0.8857980072498322,
+      "num_tokens": 5378496.0,
+      "step": 436
+    },
+    {
+      "entropy": 1.6025878190994263,
+      "epoch": 0.23012111637704055,
+      "grad_norm": 0.49696317315101624,
+      "learning_rate": 0.0001999963428949608,
+      "loss": 0.2271474003791809,
+      "mean_token_accuracy": 0.9138942360877991,
+      "num_tokens": 5390832.0,
+      "step": 437
+    },
+    {
+      "entropy": 1.5836864113807678,
+      "epoch": 0.23064770932069512,
+      "grad_norm": 0.4876512289047241,
+      "learning_rate": 0.00019999614254366895,
+      "loss": 0.24330684542655945,
+      "mean_token_accuracy": 0.9093976020812988,
+      "num_tokens": 5403168.0,
+      "step": 438
+    },
+    {
+      "entropy": 1.5200332999229431,
+      "epoch": 0.23117430226434965,
+      "grad_norm": 0.6539078950881958,
+      "learning_rate": 0.00019999593684983058,
+      "loss": 0.2600172460079193,
+      "mean_token_accuracy": 0.8999185413122177,
+      "num_tokens": 5415504.0,
+      "step": 439
+    },
+    {
+      "entropy": 1.4733262360095978,
+      "epoch": 0.23170089520800422,
+      "grad_norm": 0.42358288168907166,
+      "learning_rate": 0.000199995725813458,
+      "loss": 0.22544077038764954,
+      "mean_token_accuracy": 0.9120991379022598,
+      "num_tokens": 5427840.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.414179265499115,
+      "epoch": 0.23222748815165878,
+      "grad_norm": 0.49307864904403687,
+      "learning_rate": 0.00019999550943456375,
+      "loss": 0.24960950016975403,
+      "mean_token_accuracy": 0.9004078656435013,
+      "num_tokens": 5440176.0,
+      "step": 441
+    },
+    {
+      "entropy": 1.429827481508255,
+      "epoch": 0.23275408109531331,
+      "grad_norm": 0.505660891532898,
+      "learning_rate": 0.00019999528771316057,
+      "loss": 0.25559940934181213,
+      "mean_token_accuracy": 0.9029601365327835,
+      "num_tokens": 5452512.0,
+      "step": 442
+    },
+    {
+      "entropy": 1.4452406466007233,
+      "epoch": 0.23328067403896788,
+      "grad_norm": 0.5408805012702942,
+      "learning_rate": 0.00019999506064926175,
+      "loss": 0.2864495515823364,
+      "mean_token_accuracy": 0.8919995576143265,
+      "num_tokens": 5464848.0,
+      "step": 443
+    },
+    {
+      "entropy": 1.441390186548233,
+      "epoch": 0.23380726698262244,
+      "grad_norm": 0.49373185634613037,
+      "learning_rate": 0.0001999948282428807,
+      "loss": 0.2585368752479553,
+      "mean_token_accuracy": 0.8959326446056366,
+      "num_tokens": 5477184.0,
+      "step": 444
+    },
+    {
+      "entropy": 1.386150062084198,
+      "epoch": 0.23433385992627698,
+      "grad_norm": 0.4148918390274048,
+      "learning_rate": 0.0001999945904940312,
+      "loss": 0.2401098757982254,
+      "mean_token_accuracy": 0.9103811830282211,
+      "num_tokens": 5489520.0,
+      "step": 445
+    },
+    {
+      "entropy": 1.4309166073799133,
+      "epoch": 0.23486045286993154,
+      "grad_norm": 0.46061956882476807,
+      "learning_rate": 0.00019999434740272743,
+      "loss": 0.2515195310115814,
+      "mean_token_accuracy": 0.9047931730747223,
+      "num_tokens": 5501856.0,
+      "step": 446
+    },
+    {
+      "entropy": 1.4778549373149872,
+      "epoch": 0.2353870458135861,
+      "grad_norm": 0.543989896774292,
+      "learning_rate": 0.00019999409896898372,
+      "loss": 0.23118984699249268,
+      "mean_token_accuracy": 0.9122479408979416,
+      "num_tokens": 5514192.0,
+      "step": 447
+    },
+    {
+      "entropy": 1.3631019592285156,
+      "epoch": 0.23591363875724064,
+      "grad_norm": 0.39872297644615173,
+      "learning_rate": 0.00019999384519281494,
+      "loss": 0.23500953614711761,
+      "mean_token_accuracy": 0.9080264270305634,
+      "num_tokens": 5526528.0,
+      "step": 448
+    },
+    {
+      "entropy": 1.4472846686840057,
+      "epoch": 0.2364402317008952,
+      "grad_norm": 0.4705425500869751,
+      "learning_rate": 0.00019999358607423608,
+      "loss": 0.2556760609149933,
+      "mean_token_accuracy": 0.903324156999588,
+      "num_tokens": 5538864.0,
+      "step": 449
+    },
+    {
+      "entropy": 1.4216105043888092,
+      "epoch": 0.23696682464454977,
+      "grad_norm": 0.42123982310295105,
+      "learning_rate": 0.00019999332161326253,
+      "loss": 0.24162599444389343,
+      "mean_token_accuracy": 0.906711995601654,
+      "num_tokens": 5551200.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.3214130699634552,
+      "epoch": 0.23749341758820433,
+      "grad_norm": 0.4052497148513794,
+      "learning_rate": 0.00019999305180991002,
+      "loss": 0.23021447658538818,
+      "mean_token_accuracy": 0.9166329354047775,
+      "num_tokens": 5563536.0,
+      "step": 451
+    },
+    {
+      "entropy": 1.2956161499023438,
+      "epoch": 0.23802001053185887,
+      "grad_norm": 0.4208316206932068,
+      "learning_rate": 0.0001999927766641945,
+      "loss": 0.25906720757484436,
+      "mean_token_accuracy": 0.8939131498336792,
+      "num_tokens": 5575872.0,
+      "step": 452
+    },
+    {
+      "entropy": 1.2186987400054932,
+      "epoch": 0.23854660347551343,
+      "grad_norm": 0.423001766204834,
+      "learning_rate": 0.00019999249617613237,
+      "loss": 0.2634647488594055,
+      "mean_token_accuracy": 0.8950765579938889,
+      "num_tokens": 5588208.0,
+      "step": 453
+    },
+    {
+      "entropy": 1.1696477234363556,
+      "epoch": 0.239073196419168,
+      "grad_norm": 0.7100384831428528,
+      "learning_rate": 0.00019999221034574028,
+      "loss": 0.2620895504951477,
+      "mean_token_accuracy": 0.901919424533844,
+      "num_tokens": 5600544.0,
+      "step": 454
+    },
+    {
+      "entropy": 1.1881844103336334,
+      "epoch": 0.23959978936282253,
+      "grad_norm": 0.40986204147338867,
+      "learning_rate": 0.00019999191917303513,
+      "loss": 0.23100420832633972,
+      "mean_token_accuracy": 0.9076221436262131,
+      "num_tokens": 5612880.0,
+      "step": 455
+    },
+    {
+      "entropy": 1.0672442317008972,
+      "epoch": 0.2401263823064771,
+      "grad_norm": 0.9633269309997559,
+      "learning_rate": 0.0001999916226580343,
+      "loss": 0.2286304533481598,
+      "mean_token_accuracy": 0.9094352871179581,
+      "num_tokens": 5625216.0,
+      "step": 456
+    },
+    {
+      "entropy": 1.0725638568401337,
+      "epoch": 0.24065297525013166,
+      "grad_norm": 0.5792625546455383,
+      "learning_rate": 0.0001999913208007553,
+      "loss": 0.21992720663547516,
+      "mean_token_accuracy": 0.9132383912801743,
+      "num_tokens": 5637552.0,
+      "step": 457
+    },
+    {
+      "entropy": 1.0756218880414963,
+      "epoch": 0.2411795681937862,
+      "grad_norm": 0.4889819622039795,
+      "learning_rate": 0.0001999910136012161,
+      "loss": 0.22047317028045654,
+      "mean_token_accuracy": 0.9096969217061996,
+      "num_tokens": 5649888.0,
+      "step": 458
+    },
+    {
+      "entropy": 1.025196298956871,
+      "epoch": 0.24170616113744076,
+      "grad_norm": 0.43335074186325073,
+      "learning_rate": 0.00019999070105943494,
+      "loss": 0.23299942910671234,
+      "mean_token_accuracy": 0.9110541045665741,
+      "num_tokens": 5662224.0,
+      "step": 459
+    },
+    {
+      "entropy": 1.0740464180707932,
+      "epoch": 0.24223275408109532,
+      "grad_norm": 0.4134856164455414,
+      "learning_rate": 0.00019999038317543036,
+      "loss": 0.2113601118326187,
+      "mean_token_accuracy": 0.9125392287969589,
+      "num_tokens": 5674560.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0400484502315521,
+      "epoch": 0.24275934702474986,
+      "grad_norm": 0.3933151364326477,
+      "learning_rate": 0.00019999005994922125,
+      "loss": 0.23023052513599396,
+      "mean_token_accuracy": 0.9128931611776352,
+      "num_tokens": 5686896.0,
+      "step": 461
+    },
+    {
+      "entropy": 1.033817708492279,
+      "epoch": 0.24328593996840442,
+      "grad_norm": 0.4203343093395233,
+      "learning_rate": 0.00019998973138082675,
+      "loss": 0.23599188029766083,
+      "mean_token_accuracy": 0.9099869877099991,
+      "num_tokens": 5699232.0,
+      "step": 462
+    },
+    {
+      "entropy": 1.0623035430908203,
+      "epoch": 0.24381253291205898,
+      "grad_norm": 0.41818881034851074,
+      "learning_rate": 0.00019998939747026644,
+      "loss": 0.23520024120807648,
+      "mean_token_accuracy": 0.9101691842079163,
+      "num_tokens": 5711568.0,
+      "step": 463
+    },
+    {
+      "entropy": 1.0741385221481323,
+      "epoch": 0.24433912585571355,
+      "grad_norm": 0.4350592792034149,
+      "learning_rate": 0.00019998905821756006,
+      "loss": 0.26694831252098083,
+      "mean_token_accuracy": 0.8980891853570938,
+      "num_tokens": 5723904.0,
+      "step": 464
+    },
+    {
+      "entropy": 1.1385951340198517,
+      "epoch": 0.24486571879936808,
+      "grad_norm": 0.4376484155654907,
+      "learning_rate": 0.00019998871362272784,
+      "loss": 0.25515180826187134,
+      "mean_token_accuracy": 0.9086296558380127,
+      "num_tokens": 5736240.0,
+      "step": 465
+    },
+    {
+      "entropy": 1.0884679853916168,
+      "epoch": 0.24539231174302265,
+      "grad_norm": 0.42011401057243347,
+      "learning_rate": 0.00019998836368579013,
+      "loss": 0.2314184159040451,
+      "mean_token_accuracy": 0.9099607318639755,
+      "num_tokens": 5748576.0,
+      "step": 466
+    },
+    {
+      "entropy": 1.09638212621212,
+      "epoch": 0.2459189046866772,
+      "grad_norm": 0.39627403020858765,
+      "learning_rate": 0.0001999880084067678,
+      "loss": 0.21417196094989777,
+      "mean_token_accuracy": 0.9153980165719986,
+      "num_tokens": 5760912.0,
+      "step": 467
+    },
+    {
+      "entropy": 1.091229110956192,
+      "epoch": 0.24644549763033174,
+      "grad_norm": 0.4235895872116089,
+      "learning_rate": 0.00019998764778568192,
+      "loss": 0.2682695686817169,
+      "mean_token_accuracy": 0.9011466205120087,
+      "num_tokens": 5773248.0,
+      "step": 468
+    },
+    {
+      "entropy": 1.1204252541065216,
+      "epoch": 0.2469720905739863,
+      "grad_norm": 0.43689507246017456,
+      "learning_rate": 0.00019998728182255381,
+      "loss": 0.24578194320201874,
+      "mean_token_accuracy": 0.9062277525663376,
+      "num_tokens": 5785584.0,
+      "step": 469
+    },
+    {
+      "entropy": 1.1320042610168457,
+      "epoch": 0.24749868351764087,
+      "grad_norm": 0.4219723641872406,
+      "learning_rate": 0.0001999869105174053,
+      "loss": 0.22462952136993408,
+      "mean_token_accuracy": 0.9171530902385712,
+      "num_tokens": 5797920.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.1340222358703613,
+      "epoch": 0.2480252764612954,
+      "grad_norm": 0.3906649351119995,
+      "learning_rate": 0.00019998653387025842,
+      "loss": 0.2314276546239853,
+      "mean_token_accuracy": 0.9133926779031754,
+      "num_tokens": 5810256.0,
+      "step": 471
+    },
+    {
+      "entropy": 1.148337960243225,
+      "epoch": 0.24855186940494997,
+      "grad_norm": 0.435470312833786,
+      "learning_rate": 0.00019998615188113547,
+      "loss": 0.27237382531166077,
+      "mean_token_accuracy": 0.8951744884252548,
+      "num_tokens": 5822592.0,
+      "step": 472
+    },
+    {
+      "entropy": 1.084697663784027,
+      "epoch": 0.24907846234860453,
+      "grad_norm": 0.3843922019004822,
+      "learning_rate": 0.00019998576455005918,
+      "loss": 0.21877625584602356,
+      "mean_token_accuracy": 0.9148665964603424,
+      "num_tokens": 5834928.0,
+      "step": 473
+    },
+    {
+      "entropy": 1.1236861050128937,
+      "epoch": 0.24960505529225907,
+      "grad_norm": 0.3693842887878418,
+      "learning_rate": 0.0001999853718770525,
+      "loss": 0.24262937903404236,
+      "mean_token_accuracy": 0.9089461863040924,
+      "num_tokens": 5847264.0,
+      "step": 474
+    },
+    {
+      "entropy": 1.0875613689422607,
+      "epoch": 0.25013164823591366,
+      "grad_norm": 0.36348292231559753,
+      "learning_rate": 0.0001999849738621388,
+      "loss": 0.22801700234413147,
+      "mean_token_accuracy": 0.9149875193834305,
+      "num_tokens": 5859600.0,
+      "step": 475
+    },
+    {
+      "entropy": 1.1611447632312775,
+      "epoch": 0.2506582411795682,
+      "grad_norm": 0.3937731683254242,
+      "learning_rate": 0.00019998457050534165,
+      "loss": 0.2227628082036972,
+      "mean_token_accuracy": 0.9130124449729919,
+      "num_tokens": 5871936.0,
+      "step": 476
+    },
+    {
+      "entropy": 1.18001389503479,
+      "epoch": 0.25118483412322273,
+      "grad_norm": 0.38412168622016907,
+      "learning_rate": 0.00019998416180668505,
+      "loss": 0.25140437483787537,
+      "mean_token_accuracy": 0.9021612256765366,
+      "num_tokens": 5884272.0,
+      "step": 477
+    },
+    {
+      "entropy": 1.1783344447612762,
+      "epoch": 0.2517114270668773,
+      "grad_norm": 0.4068759083747864,
+      "learning_rate": 0.00019998374776619316,
+      "loss": 0.2507692575454712,
+      "mean_token_accuracy": 0.9036175757646561,
+      "num_tokens": 5896608.0,
+      "step": 478
+    },
+    {
+      "entropy": 1.2171072959899902,
+      "epoch": 0.25223802001053186,
+      "grad_norm": 0.38274940848350525,
+      "learning_rate": 0.00019998332838389068,
+      "loss": 0.23562754690647125,
+      "mean_token_accuracy": 0.9149611443281174,
+      "num_tokens": 5908944.0,
+      "step": 479
+    },
+    {
+      "entropy": 1.2112184464931488,
+      "epoch": 0.2527646129541864,
+      "grad_norm": 0.4171972870826721,
+      "learning_rate": 0.00019998290365980247,
+      "loss": 0.26383018493652344,
+      "mean_token_accuracy": 0.8946961760520935,
+      "num_tokens": 5921280.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.1482765674591064,
+      "epoch": 0.253291205897841,
+      "grad_norm": 0.3705795407295227,
+      "learning_rate": 0.0001999824735939537,
+      "loss": 0.23539453744888306,
+      "mean_token_accuracy": 0.9124044477939606,
+      "num_tokens": 5933616.0,
+      "step": 481
+    },
+    {
+      "entropy": 1.1225126087665558,
+      "epoch": 0.2538177988414955,
+      "grad_norm": 0.38258257508277893,
+      "learning_rate": 0.00019998203818636997,
+      "loss": 0.24310466647148132,
+      "mean_token_accuracy": 0.9069766700267792,
+      "num_tokens": 5945952.0,
+      "step": 482
+    },
+    {
+      "entropy": 1.1831683218479156,
+      "epoch": 0.25434439178515006,
+      "grad_norm": 0.4023701548576355,
+      "learning_rate": 0.00019998159743707706,
+      "loss": 0.22924776375293732,
+      "mean_token_accuracy": 0.910895898938179,
+      "num_tokens": 5958288.0,
+      "step": 483
+    },
+    {
+      "entropy": 1.2125089168548584,
+      "epoch": 0.25487098472880465,
+      "grad_norm": 0.391679972410202,
+      "learning_rate": 0.0001999811513461012,
+      "loss": 0.23159608244895935,
+      "mean_token_accuracy": 0.91130131483078,
+      "num_tokens": 5970624.0,
+      "step": 484
+    },
+    {
+      "entropy": 1.1657118201255798,
+      "epoch": 0.2553975776724592,
+      "grad_norm": 0.36009886860847473,
+      "learning_rate": 0.00019998069991346878,
+      "loss": 0.22824552655220032,
+      "mean_token_accuracy": 0.9051208347082138,
+      "num_tokens": 5982960.0,
+      "step": 485
+    },
+    {
+      "entropy": 1.1666974127292633,
+      "epoch": 0.2559241706161137,
+      "grad_norm": 0.4200109839439392,
+      "learning_rate": 0.00019998024313920668,
+      "loss": 0.22612211108207703,
+      "mean_token_accuracy": 0.9098036587238312,
+      "num_tokens": 5995296.0,
+      "step": 486
+    },
+    {
+      "entropy": 1.1931614875793457,
+      "epoch": 0.2564507635597683,
+      "grad_norm": 0.3996848165988922,
+      "learning_rate": 0.00019997978102334195,
+      "loss": 0.23149043321609497,
+      "mean_token_accuracy": 0.915941908955574,
+      "num_tokens": 6007632.0,
+      "step": 487
+    },
+    {
+      "entropy": 1.2029556334018707,
+      "epoch": 0.25697735650342285,
+      "grad_norm": 0.381720632314682,
+      "learning_rate": 0.00019997931356590212,
+      "loss": 0.20601031184196472,
+      "mean_token_accuracy": 0.9207601994276047,
+      "num_tokens": 6019968.0,
+      "step": 488
+    },
+    {
+      "entropy": 1.286322921514511,
+      "epoch": 0.2575039494470774,
+      "grad_norm": 0.3947739601135254,
+      "learning_rate": 0.00019997884076691484,
+      "loss": 0.23165327310562134,
+      "mean_token_accuracy": 0.9127722233533859,
+      "num_tokens": 6032304.0,
+      "step": 489
+    },
+    {
+      "entropy": 1.1862512528896332,
+      "epoch": 0.258030542390732,
+      "grad_norm": 0.4733010232448578,
+      "learning_rate": 0.00019997836262640825,
+      "loss": 0.23455524444580078,
+      "mean_token_accuracy": 0.9086931645870209,
+      "num_tokens": 6044640.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.2432229220867157,
+      "epoch": 0.2585571353343865,
+      "grad_norm": 0.5269920229911804,
+      "learning_rate": 0.00019997787914441067,
+      "loss": 0.25063106417655945,
+      "mean_token_accuracy": 0.903441533446312,
+      "num_tokens": 6056976.0,
+      "step": 491
+    },
+    {
+      "entropy": 1.1587136387825012,
+      "epoch": 0.25908372827804105,
+      "grad_norm": 0.3660036325454712,
+      "learning_rate": 0.0001999773903209509,
+      "loss": 0.23804998397827148,
+      "mean_token_accuracy": 0.9118834733963013,
+      "num_tokens": 6069312.0,
+      "step": 492
+    },
+    {
+      "entropy": 1.2006062865257263,
+      "epoch": 0.25961032122169564,
+      "grad_norm": 0.4517950117588043,
+      "learning_rate": 0.00019997689615605785,
+      "loss": 0.24688124656677246,
+      "mean_token_accuracy": 0.9041164964437485,
+      "num_tokens": 6081648.0,
+      "step": 493
+    },
+    {
+      "entropy": 1.2588125467300415,
+      "epoch": 0.2601369141653502,
+      "grad_norm": 0.3569098114967346,
+      "learning_rate": 0.0001999763966497609,
+      "loss": 0.22678923606872559,
+      "mean_token_accuracy": 0.9117091447114944,
+      "num_tokens": 6093984.0,
+      "step": 494
+    },
+    {
+      "entropy": 1.3271512985229492,
+      "epoch": 0.26066350710900477,
+      "grad_norm": 0.3427906036376953,
+      "learning_rate": 0.00019997589180208972,
+      "loss": 0.2427246868610382,
+      "mean_token_accuracy": 0.91063092648983,
+      "num_tokens": 6106320.0,
+      "step": 495
+    },
+    {
+      "entropy": 1.320880800485611,
+      "epoch": 0.2611901000526593,
+      "grad_norm": 0.41021794080734253,
+      "learning_rate": 0.00019997538161307425,
+      "loss": 0.24322479963302612,
+      "mean_token_accuracy": 0.9047645032405853,
+      "num_tokens": 6118656.0,
+      "step": 496
+    },
+    {
+      "entropy": 1.2419660687446594,
+      "epoch": 0.26171669299631384,
+      "grad_norm": 0.4706597328186035,
+      "learning_rate": 0.00019997486608274478,
+      "loss": 0.24686744809150696,
+      "mean_token_accuracy": 0.9022967517375946,
+      "num_tokens": 6130992.0,
+      "step": 497
+    },
+    {
+      "entropy": 1.2451711893081665,
+      "epoch": 0.26224328593996843,
+      "grad_norm": 0.37942013144493103,
+      "learning_rate": 0.00019997434521113193,
+      "loss": 0.2456897497177124,
+      "mean_token_accuracy": 0.9084974527359009,
+      "num_tokens": 6143328.0,
+      "step": 498
+    },
+    {
+      "entropy": 1.363191157579422,
+      "epoch": 0.26276987888362296,
+      "grad_norm": 0.4037798047065735,
+      "learning_rate": 0.00019997381899826663,
+      "loss": 0.24723505973815918,
+      "mean_token_accuracy": 0.9025511145591736,
+      "num_tokens": 6155664.0,
+      "step": 499
+    },
+    {
+      "entropy": 1.3585219979286194,
+      "epoch": 0.2632964718272775,
+      "grad_norm": 0.4076964855194092,
+      "learning_rate": 0.0001999732874441801,
+      "loss": 0.231999009847641,
+      "mean_token_accuracy": 0.9121885448694229,
+      "num_tokens": 6168000.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.3322370648384094,
+      "epoch": 0.2638230647709321,
+      "grad_norm": 0.38084790110588074,
+      "learning_rate": 0.0001999727505489039,
+      "loss": 0.26456475257873535,
+      "mean_token_accuracy": 0.9003604352474213,
+      "num_tokens": 6180336.0,
+      "step": 501
+    },
+    {
+      "entropy": 1.3225562274456024,
+      "epoch": 0.2643496577145866,
+      "grad_norm": 0.3669094145298004,
+      "learning_rate": 0.00019997220831246987,
+      "loss": 0.22636903822422028,
+      "mean_token_accuracy": 0.9135940670967102,
+      "num_tokens": 6192672.0,
+      "step": 502
+    },
+    {
+      "entropy": 1.24677973985672,
+      "epoch": 0.26487625065824116,
+      "grad_norm": 0.36622145771980286,
+      "learning_rate": 0.00019997166073491027,
+      "loss": 0.2442116141319275,
+      "mean_token_accuracy": 0.903778612613678,
+      "num_tokens": 6205008.0,
+      "step": 503
+    },
+    {
+      "entropy": 1.2564740180969238,
+      "epoch": 0.26540284360189575,
+      "grad_norm": 0.37914055585861206,
+      "learning_rate": 0.00019997110781625754,
+      "loss": 0.250071257352829,
+      "mean_token_accuracy": 0.905722975730896,
+      "num_tokens": 6217344.0,
+      "step": 504
+    },
+    {
+      "entropy": 1.266231119632721,
+      "epoch": 0.2659294365455503,
+      "grad_norm": 0.4232172966003418,
+      "learning_rate": 0.00019997054955654456,
+      "loss": 0.22360444068908691,
+      "mean_token_accuracy": 0.9164858758449554,
+      "num_tokens": 6229680.0,
+      "step": 505
+    },
+    {
+      "entropy": 1.2842857539653778,
+      "epoch": 0.2664560294892048,
+      "grad_norm": 0.40517833828926086,
+      "learning_rate": 0.0001999699859558044,
+      "loss": 0.24367284774780273,
+      "mean_token_accuracy": 0.906918466091156,
+      "num_tokens": 6242016.0,
+      "step": 506
+    },
+    {
+      "entropy": 1.3325933814048767,
+      "epoch": 0.2669826224328594,
+      "grad_norm": 0.3894166648387909,
+      "learning_rate": 0.00019996941701407058,
+      "loss": 0.21899431943893433,
+      "mean_token_accuracy": 0.9159570783376694,
+      "num_tokens": 6254352.0,
+      "step": 507
+    },
+    {
+      "entropy": 1.2919807434082031,
+      "epoch": 0.26750921537651395,
+      "grad_norm": 0.3554909825325012,
+      "learning_rate": 0.00019996884273137686,
+      "loss": 0.22241441905498505,
+      "mean_token_accuracy": 0.9181459695100784,
+      "num_tokens": 6266688.0,
+      "step": 508
+    },
+    {
+      "entropy": 1.311550348997116,
+      "epoch": 0.2680358083201685,
+      "grad_norm": 0.36536744236946106,
+      "learning_rate": 0.0001999682631077573,
+      "loss": 0.2456498146057129,
+      "mean_token_accuracy": 0.9058470577001572,
+      "num_tokens": 6279024.0,
+      "step": 509
+    },
+    {
+      "entropy": 1.3708328008651733,
+      "epoch": 0.2685624012638231,
+      "grad_norm": 0.4042948782444,
+      "learning_rate": 0.00019996767814324637,
+      "loss": 0.266501247882843,
+      "mean_token_accuracy": 0.900126188993454,
+      "num_tokens": 6291360.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.3850271105766296,
+      "epoch": 0.2690889942074776,
+      "grad_norm": 0.39048540592193604,
+      "learning_rate": 0.00019996708783787874,
+      "loss": 0.24780599772930145,
+      "mean_token_accuracy": 0.9016763418912888,
+      "num_tokens": 6303696.0,
+      "step": 511
+    },
+    {
+      "entropy": 1.3151749074459076,
+      "epoch": 0.26961558715113215,
+      "grad_norm": 0.37497779726982117,
+      "learning_rate": 0.0001999664921916895,
+      "loss": 0.23127570748329163,
+      "mean_token_accuracy": 0.9075058400630951,
+      "num_tokens": 6316032.0,
+      "step": 512
+    },
+    {
+      "entropy": 1.3835597038269043,
+      "epoch": 0.27014218009478674,
+      "grad_norm": 0.3880425691604614,
+      "learning_rate": 0.00019996589120471392,
+      "loss": 0.23898069560527802,
+      "mean_token_accuracy": 0.9100230634212494,
+      "num_tokens": 6328368.0,
+      "step": 513
+    },
+    {
+      "entropy": 1.4238722920417786,
+      "epoch": 0.2706687730384413,
+      "grad_norm": 0.36198124289512634,
+      "learning_rate": 0.0001999652848769878,
+      "loss": 0.22968879342079163,
+      "mean_token_accuracy": 0.906474158167839,
+      "num_tokens": 6340704.0,
+      "step": 514
+    },
+    {
+      "entropy": 1.517054557800293,
+      "epoch": 0.2711953659820958,
+      "grad_norm": 0.3657878637313843,
+      "learning_rate": 0.00019996467320854703,
+      "loss": 0.22068339586257935,
+      "mean_token_accuracy": 0.9112901985645294,
+      "num_tokens": 6353040.0,
+      "step": 515
+    },
+    {
+      "entropy": 1.5027157366275787,
+      "epoch": 0.2717219589257504,
+      "grad_norm": 0.34945791959762573,
+      "learning_rate": 0.000199964056199428,
+      "loss": 0.25596681237220764,
+      "mean_token_accuracy": 0.9006428420543671,
+      "num_tokens": 6365376.0,
+      "step": 516
+    },
+    {
+      "entropy": 1.5874098539352417,
+      "epoch": 0.27224855186940494,
+      "grad_norm": 0.3800085783004761,
+      "learning_rate": 0.00019996343384966732,
+      "loss": 0.25539156794548035,
+      "mean_token_accuracy": 0.901904970407486,
+      "num_tokens": 6377712.0,
+      "step": 517
+    },
+    {
+      "entropy": 1.5508817732334137,
+      "epoch": 0.2727751448130595,
+      "grad_norm": 0.3190958797931671,
+      "learning_rate": 0.00019996280615930186,
+      "loss": 0.235137477517128,
+      "mean_token_accuracy": 0.9069496989250183,
+      "num_tokens": 6390048.0,
+      "step": 518
+    },
+    {
+      "entropy": 1.5596497654914856,
+      "epoch": 0.27330173775671407,
+      "grad_norm": 0.35837146639823914,
+      "learning_rate": 0.000199962173128369,
+      "loss": 0.24638012051582336,
+      "mean_token_accuracy": 0.9043673276901245,
+      "num_tokens": 6402384.0,
+      "step": 519
+    },
+    {
+      "entropy": 1.544301688671112,
+      "epoch": 0.2738283307003686,
+      "grad_norm": 0.3943609595298767,
+      "learning_rate": 0.00019996153475690623,
+      "loss": 0.24823732674121857,
+      "mean_token_accuracy": 0.9039914757013321,
+      "num_tokens": 6414720.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.4979600608348846,
+      "epoch": 0.2743549236440232,
+      "grad_norm": 0.3533165156841278,
+      "learning_rate": 0.00019996089104495148,
+      "loss": 0.21794329583644867,
+      "mean_token_accuracy": 0.9170633852481842,
+      "num_tokens": 6427056.0,
+      "step": 521
+    },
+    {
+      "entropy": 1.5348587334156036,
+      "epoch": 0.27488151658767773,
+      "grad_norm": 0.3516184091567993,
+      "learning_rate": 0.00019996024199254295,
+      "loss": 0.21856778860092163,
+      "mean_token_accuracy": 0.9159553945064545,
+      "num_tokens": 6439392.0,
+      "step": 522
+    },
+    {
+      "entropy": 1.5593447387218475,
+      "epoch": 0.27540810953133227,
+      "grad_norm": 0.3706720173358917,
+      "learning_rate": 0.0001999595875997192,
+      "loss": 0.24685126543045044,
+      "mean_token_accuracy": 0.9023942798376083,
+      "num_tokens": 6451728.0,
+      "step": 523
+    },
+    {
+      "entropy": 1.5938279330730438,
+      "epoch": 0.27593470247498686,
+      "grad_norm": 0.4374881088733673,
+      "learning_rate": 0.00019995892786651905,
+      "loss": 0.24188384413719177,
+      "mean_token_accuracy": 0.904677078127861,
+      "num_tokens": 6464064.0,
+      "step": 524
+    },
+    {
+      "entropy": 1.469886839389801,
+      "epoch": 0.2764612954186414,
+      "grad_norm": 0.46137046813964844,
+      "learning_rate": 0.00019995826279298166,
+      "loss": 0.23798640072345734,
+      "mean_token_accuracy": 0.9109133780002594,
+      "num_tokens": 6476400.0,
+      "step": 525
+    },
+    {
+      "entropy": 1.5873245596885681,
+      "epoch": 0.27698788836229593,
+      "grad_norm": 0.416920006275177,
+      "learning_rate": 0.00019995759237914656,
+      "loss": 0.25300133228302,
+      "mean_token_accuracy": 0.9008941799402237,
+      "num_tokens": 6488736.0,
+      "step": 526
+    },
+    {
+      "entropy": 1.5558778047561646,
+      "epoch": 0.2775144813059505,
+      "grad_norm": 0.36363670229911804,
+      "learning_rate": 0.0001999569166250535,
+      "loss": 0.22929418087005615,
+      "mean_token_accuracy": 0.9127305895090103,
+      "num_tokens": 6501072.0,
+      "step": 527
+    },
+    {
+      "entropy": 1.5871147811412811,
+      "epoch": 0.27804107424960506,
+      "grad_norm": 0.3830122947692871,
+      "learning_rate": 0.00019995623553074258,
+      "loss": 0.2472982108592987,
+      "mean_token_accuracy": 0.9040207862854004,
+      "num_tokens": 6513408.0,
+      "step": 528
+    },
+    {
+      "entropy": 1.6365769505500793,
+      "epoch": 0.2785676671932596,
+      "grad_norm": 0.3964959383010864,
+      "learning_rate": 0.00019995554909625428,
+      "loss": 0.25368693470954895,
+      "mean_token_accuracy": 0.9007250666618347,
+      "num_tokens": 6525744.0,
+      "step": 529
+    },
+    {
+      "entropy": 1.7017327845096588,
+      "epoch": 0.2790942601369142,
+      "grad_norm": 0.391290545463562,
+      "learning_rate": 0.00019995485732162934,
+      "loss": 0.2360502928495407,
+      "mean_token_accuracy": 0.9084432125091553,
+      "num_tokens": 6538080.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.6393355131149292,
+      "epoch": 0.2796208530805687,
+      "grad_norm": 0.38507914543151855,
+      "learning_rate": 0.0001999541602069088,
+      "loss": 0.23786881566047668,
+      "mean_token_accuracy": 0.9061485677957535,
+      "num_tokens": 6550416.0,
+      "step": 531
+    },
+    {
+      "entropy": 1.6275135576725006,
+      "epoch": 0.28014744602422326,
+      "grad_norm": 0.37895116209983826,
+      "learning_rate": 0.0001999534577521341,
+      "loss": 0.23145025968551636,
+      "mean_token_accuracy": 0.9090943783521652,
+      "num_tokens": 6562752.0,
+      "step": 532
+    },
+    {
+      "entropy": 1.598475605249405,
+      "epoch": 0.28067403896787785,
+      "grad_norm": 0.4161452054977417,
+      "learning_rate": 0.00019995274995734686,
+      "loss": 0.2727176547050476,
+      "mean_token_accuracy": 0.8956159949302673,
+      "num_tokens": 6575088.0,
+      "step": 533
+    },
+    {
+      "entropy": 1.5213307440280914,
+      "epoch": 0.2812006319115324,
+      "grad_norm": 0.43801331520080566,
+      "learning_rate": 0.00019995203682258915,
+      "loss": 0.22940769791603088,
+      "mean_token_accuracy": 0.9128211736679077,
+      "num_tokens": 6587424.0,
+      "step": 534
+    },
+    {
+      "entropy": 1.634105533361435,
+      "epoch": 0.2817272248551869,
+      "grad_norm": 0.4054886996746063,
+      "learning_rate": 0.0001999513183479033,
+      "loss": 0.23942533135414124,
+      "mean_token_accuracy": 0.9082587212324142,
+      "num_tokens": 6599760.0,
+      "step": 535
+    },
+    {
+      "entropy": 1.4907733798027039,
+      "epoch": 0.2822538177988415,
+      "grad_norm": 0.40736711025238037,
+      "learning_rate": 0.00019995059453333194,
+      "loss": 0.22832639515399933,
+      "mean_token_accuracy": 0.9135842621326447,
+      "num_tokens": 6612096.0,
+      "step": 536
+    },
+    {
+      "entropy": 1.5832161009311676,
+      "epoch": 0.28278041074249605,
+      "grad_norm": 0.3980075418949127,
+      "learning_rate": 0.0001999498653789181,
+      "loss": 0.24887777864933014,
+      "mean_token_accuracy": 0.9017034024000168,
+      "num_tokens": 6624432.0,
+      "step": 537
+    },
+    {
+      "entropy": 1.4983101785182953,
+      "epoch": 0.2833070036861506,
+      "grad_norm": 0.3781083822250366,
+      "learning_rate": 0.00019994913088470498,
+      "loss": 0.23180519044399261,
+      "mean_token_accuracy": 0.9052098244428635,
+      "num_tokens": 6636768.0,
+      "step": 538
+    },
+    {
+      "entropy": 1.5062950551509857,
+      "epoch": 0.2838335966298052,
+      "grad_norm": 0.4003005623817444,
+      "learning_rate": 0.00019994839105073623,
+      "loss": 0.22549505531787872,
+      "mean_token_accuracy": 0.9140617251396179,
+      "num_tokens": 6649104.0,
+      "step": 539
+    },
+    {
+      "entropy": 1.5416783392429352,
+      "epoch": 0.2843601895734597,
+      "grad_norm": 0.39962485432624817,
+      "learning_rate": 0.00019994764587705574,
+      "loss": 0.2179470807313919,
+      "mean_token_accuracy": 0.9161404520273209,
+      "num_tokens": 6661440.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.660192757844925,
+      "epoch": 0.28488678251711425,
+      "grad_norm": 0.42626917362213135,
+      "learning_rate": 0.0001999468953637078,
+      "loss": 0.26295173168182373,
+      "mean_token_accuracy": 0.8983173221349716,
+      "num_tokens": 6673776.0,
+      "step": 541
+    },
+    {
+      "entropy": 1.5149348974227905,
+      "epoch": 0.28541337546076884,
+      "grad_norm": 0.36926260590553284,
+      "learning_rate": 0.00019994613951073692,
+      "loss": 0.2277842015028,
+      "mean_token_accuracy": 0.9144963473081589,
+      "num_tokens": 6686112.0,
+      "step": 542
+    },
+    {
+      "entropy": 1.5601712763309479,
+      "epoch": 0.2859399684044234,
+      "grad_norm": 0.35751771926879883,
+      "learning_rate": 0.00019994537831818799,
+      "loss": 0.2134617567062378,
+      "mean_token_accuracy": 0.9172465950250626,
+      "num_tokens": 6698448.0,
+      "step": 543
+    },
+    {
+      "entropy": 1.629770964384079,
+      "epoch": 0.2864665613480779,
+      "grad_norm": 0.3708803951740265,
+      "learning_rate": 0.00019994461178610617,
+      "loss": 0.22911599278450012,
+      "mean_token_accuracy": 0.9093553721904755,
+      "num_tokens": 6710784.0,
+      "step": 544
+    },
+    {
+      "entropy": 1.5735207796096802,
+      "epoch": 0.2869931542917325,
+      "grad_norm": 0.3979873061180115,
+      "learning_rate": 0.00019994383991453702,
+      "loss": 0.2454529106616974,
+      "mean_token_accuracy": 0.9057244658470154,
+      "num_tokens": 6723120.0,
+      "step": 545
+    },
+    {
+      "entropy": 1.556952327489853,
+      "epoch": 0.28751974723538704,
+      "grad_norm": 0.36042284965515137,
+      "learning_rate": 0.0001999430627035263,
+      "loss": 0.22943763434886932,
+      "mean_token_accuracy": 0.9095467925071716,
+      "num_tokens": 6735456.0,
+      "step": 546
+    },
+    {
+      "entropy": 1.5784813165664673,
+      "epoch": 0.2880463401790416,
+      "grad_norm": 0.4354793131351471,
+      "learning_rate": 0.0001999422801531202,
+      "loss": 0.23101268708705902,
+      "mean_token_accuracy": 0.9098020642995834,
+      "num_tokens": 6747792.0,
+      "step": 547
+    },
+    {
+      "entropy": 1.6356432437896729,
+      "epoch": 0.28857293312269616,
+      "grad_norm": 0.38589903712272644,
+      "learning_rate": 0.00019994149226336514,
+      "loss": 0.2551157772541046,
+      "mean_token_accuracy": 0.9008017480373383,
+      "num_tokens": 6760128.0,
+      "step": 548
+    },
+    {
+      "entropy": 1.484334021806717,
+      "epoch": 0.2890995260663507,
+      "grad_norm": 0.3685605823993683,
+      "learning_rate": 0.0001999406990343079,
+      "loss": 0.21177145838737488,
+      "mean_token_accuracy": 0.9134591519832611,
+      "num_tokens": 6772464.0,
+      "step": 549
+    },
+    {
+      "entropy": 1.5352662205696106,
+      "epoch": 0.2896261190100053,
+      "grad_norm": 0.43439221382141113,
+      "learning_rate": 0.00019993990046599555,
+      "loss": 0.21738113462924957,
+      "mean_token_accuracy": 0.9168129861354828,
+      "num_tokens": 6784800.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.4901257157325745,
+      "epoch": 0.2901527119536598,
+      "grad_norm": 0.40765848755836487,
+      "learning_rate": 0.00019993909655847554,
+      "loss": 0.2299802005290985,
+      "mean_token_accuracy": 0.9086113125085831,
+      "num_tokens": 6797136.0,
+      "step": 551
+    },
+    {
+      "entropy": 1.5647250413894653,
+      "epoch": 0.29067930489731436,
+      "grad_norm": 0.3714927136898041,
+      "learning_rate": 0.0001999382873117956,
+      "loss": 0.23209913074970245,
+      "mean_token_accuracy": 0.912913978099823,
+      "num_tokens": 6809472.0,
+      "step": 552
+    },
+    {
+      "entropy": 1.5570427775382996,
+      "epoch": 0.29120589784096895,
+      "grad_norm": 0.3235602080821991,
+      "learning_rate": 0.0001999374727260037,
+      "loss": 0.20470383763313293,
+      "mean_token_accuracy": 0.9178998619318008,
+      "num_tokens": 6821808.0,
+      "step": 553
+    },
+    {
+      "entropy": 1.6096443831920624,
+      "epoch": 0.2917324907846235,
+      "grad_norm": 0.3791026473045349,
+      "learning_rate": 0.00019993665280114824,
+      "loss": 0.23357072472572327,
+      "mean_token_accuracy": 0.9100991189479828,
+      "num_tokens": 6834144.0,
+      "step": 554
+    },
+    {
+      "entropy": 1.7047458291053772,
+      "epoch": 0.292259083728278,
+      "grad_norm": 0.4588649272918701,
+      "learning_rate": 0.00019993582753727792,
+      "loss": 0.23484769463539124,
+      "mean_token_accuracy": 0.9089436531066895,
+      "num_tokens": 6846480.0,
+      "step": 555
+    },
+    {
+      "entropy": 1.6045478284358978,
+      "epoch": 0.2927856766719326,
+      "grad_norm": 0.38022348284721375,
+      "learning_rate": 0.00019993499693444168,
+      "loss": 0.24132560193538666,
+      "mean_token_accuracy": 0.9133987128734589,
+      "num_tokens": 6858816.0,
+      "step": 556
+    },
+    {
+      "entropy": 1.5578656792640686,
+      "epoch": 0.29331226961558715,
+      "grad_norm": 0.3959854543209076,
+      "learning_rate": 0.00019993416099268888,
+      "loss": 0.2288515269756317,
+      "mean_token_accuracy": 0.9094084948301315,
+      "num_tokens": 6871152.0,
+      "step": 557
+    },
+    {
+      "entropy": 1.5771779716014862,
+      "epoch": 0.2938388625592417,
+      "grad_norm": 0.41283491253852844,
+      "learning_rate": 0.00019993331971206911,
+      "loss": 0.25233519077301025,
+      "mean_token_accuracy": 0.8999712020158768,
+      "num_tokens": 6883488.0,
+      "step": 558
+    },
+    {
+      "entropy": 1.5294803977012634,
+      "epoch": 0.2943654555028963,
+      "grad_norm": 0.41696879267692566,
+      "learning_rate": 0.0001999324730926323,
+      "loss": 0.2409447431564331,
+      "mean_token_accuracy": 0.9075838178396225,
+      "num_tokens": 6895824.0,
+      "step": 559
+    },
+    {
+      "entropy": 1.557918757200241,
+      "epoch": 0.2948920484465508,
+      "grad_norm": 0.8365952372550964,
+      "learning_rate": 0.00019993162113442873,
+      "loss": 0.26256802678108215,
+      "mean_token_accuracy": 0.8970607817173004,
+      "num_tokens": 6908160.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.5486240088939667,
+      "epoch": 0.29541864139020535,
+      "grad_norm": 0.36394545435905457,
+      "learning_rate": 0.000199930763837509,
+      "loss": 0.23241116106510162,
+      "mean_token_accuracy": 0.9164627194404602,
+      "num_tokens": 6920496.0,
+      "step": 561
+    },
+    {
+      "entropy": 1.5366742312908173,
+      "epoch": 0.29594523433385994,
+      "grad_norm": 0.3717273771762848,
+      "learning_rate": 0.00019992990120192393,
+      "loss": 0.2316376268863678,
+      "mean_token_accuracy": 0.9116377085447311,
+      "num_tokens": 6932832.0,
+      "step": 562
+    },
+    {
+      "entropy": 1.6174998581409454,
+      "epoch": 0.2964718272775145,
+      "grad_norm": 0.3976577818393707,
+      "learning_rate": 0.00019992903322772483,
+      "loss": 0.2560444474220276,
+      "mean_token_accuracy": 0.8992763608694077,
+      "num_tokens": 6945168.0,
+      "step": 563
+    },
+    {
+      "entropy": 1.5432467758655548,
+      "epoch": 0.296998420221169,
+      "grad_norm": 0.32837730646133423,
+      "learning_rate": 0.00019992815991496318,
+      "loss": 0.2185848504304886,
+      "mean_token_accuracy": 0.9149428755044937,
+      "num_tokens": 6957504.0,
+      "step": 564
+    },
+    {
+      "entropy": 1.5321990847587585,
+      "epoch": 0.2975250131648236,
+      "grad_norm": 0.36741021275520325,
+      "learning_rate": 0.00019992728126369078,
+      "loss": 0.24631758034229279,
+      "mean_token_accuracy": 0.9008835554122925,
+      "num_tokens": 6969840.0,
+      "step": 565
+    },
+    {
+      "entropy": 1.6573955416679382,
+      "epoch": 0.29805160610847814,
+      "grad_norm": 0.40060925483703613,
+      "learning_rate": 0.00019992639727395983,
+      "loss": 0.24705615639686584,
+      "mean_token_accuracy": 0.9055600762367249,
+      "num_tokens": 6982176.0,
+      "step": 566
+    },
+    {
+      "entropy": 1.6146283447742462,
+      "epoch": 0.2985781990521327,
+      "grad_norm": 0.36694321036338806,
+      "learning_rate": 0.00019992550794582282,
+      "loss": 0.2382228970527649,
+      "mean_token_accuracy": 0.9077549129724503,
+      "num_tokens": 6994512.0,
+      "step": 567
+    },
+    {
+      "entropy": 1.608702838420868,
+      "epoch": 0.29910479199578727,
+      "grad_norm": 0.36138370633125305,
+      "learning_rate": 0.00019992461327933252,
+      "loss": 0.2497398406267166,
+      "mean_token_accuracy": 0.9065199494361877,
+      "num_tokens": 7006848.0,
+      "step": 568
+    },
+    {
+      "entropy": 1.5681868195533752,
+      "epoch": 0.2996313849394418,
+      "grad_norm": 0.3718649744987488,
+      "learning_rate": 0.00019992371327454204,
+      "loss": 0.22592690587043762,
+      "mean_token_accuracy": 0.9139988124370575,
+      "num_tokens": 7019184.0,
+      "step": 569
+    },
+    {
+      "entropy": 1.5541615784168243,
+      "epoch": 0.3001579778830964,
+      "grad_norm": 0.3475581407546997,
+      "learning_rate": 0.00019992280793150485,
+      "loss": 0.218974769115448,
+      "mean_token_accuracy": 0.9153032898902893,
+      "num_tokens": 7031520.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.5086607038974762,
+      "epoch": 0.30068457082675093,
+      "grad_norm": 0.336376428604126,
+      "learning_rate": 0.00019992189725027466,
+      "loss": 0.2033504694700241,
+      "mean_token_accuracy": 0.9149883091449738,
+      "num_tokens": 7043856.0,
+      "step": 571
+    },
+    {
+      "entropy": 1.5713791847229004,
+      "epoch": 0.30121116377040547,
+      "grad_norm": 0.4335557520389557,
+      "learning_rate": 0.0001999209812309055,
+      "loss": 0.2449672371149063,
+      "mean_token_accuracy": 0.9020196795463562,
+      "num_tokens": 7056192.0,
+      "step": 572
+    },
+    {
+      "entropy": 1.5503774285316467,
+      "epoch": 0.30173775671406006,
+      "grad_norm": 0.3847261071205139,
+      "learning_rate": 0.0001999200598734518,
+      "loss": 0.23619771003723145,
+      "mean_token_accuracy": 0.9090457707643509,
+      "num_tokens": 7068528.0,
+      "step": 573
+    },
+    {
+      "entropy": 1.5812987089157104,
+      "epoch": 0.3022643496577146,
+      "grad_norm": 0.4005545675754547,
+      "learning_rate": 0.00019991913317796825,
+      "loss": 0.22766467928886414,
+      "mean_token_accuracy": 0.9142022579908371,
+      "num_tokens": 7080864.0,
+      "step": 574
+    },
+    {
+      "entropy": 1.5593630969524384,
+      "epoch": 0.30279094260136913,
+      "grad_norm": 0.3692858815193176,
+      "learning_rate": 0.00019991820114450984,
+      "loss": 0.24361425638198853,
+      "mean_token_accuracy": 0.909025177359581,
+      "num_tokens": 7093200.0,
+      "step": 575
+    },
+    {
+      "entropy": 1.639871895313263,
+      "epoch": 0.3033175355450237,
+      "grad_norm": 0.3423769176006317,
+      "learning_rate": 0.0001999172637731319,
+      "loss": 0.1989578753709793,
+      "mean_token_accuracy": 0.9223005324602127,
+      "num_tokens": 7105536.0,
+      "step": 576
+    },
+    {
+      "entropy": 1.5683670938014984,
+      "epoch": 0.30384412848867826,
+      "grad_norm": 0.3434232771396637,
+      "learning_rate": 0.0001999163210638901,
+      "loss": 0.23291000723838806,
+      "mean_token_accuracy": 0.9105765074491501,
+      "num_tokens": 7117872.0,
+      "step": 577
+    },
+    {
+      "entropy": 1.6403064131736755,
+      "epoch": 0.3043707214323328,
+      "grad_norm": 0.3877623975276947,
+      "learning_rate": 0.00019991537301684037,
+      "loss": 0.2589385509490967,
+      "mean_token_accuracy": 0.9013812988996506,
+      "num_tokens": 7130208.0,
+      "step": 578
+    },
+    {
+      "entropy": 1.577894926071167,
+      "epoch": 0.3048973143759874,
+      "grad_norm": 0.3848702013492584,
+      "learning_rate": 0.00019991441963203906,
+      "loss": 0.2235189825296402,
+      "mean_token_accuracy": 0.9132665693759918,
+      "num_tokens": 7142544.0,
+      "step": 579
+    },
+    {
+      "entropy": 1.594813734292984,
+      "epoch": 0.3054239073196419,
+      "grad_norm": 0.37452247738838196,
+      "learning_rate": 0.00019991346090954268,
+      "loss": 0.2309045046567917,
+      "mean_token_accuracy": 0.9061081409454346,
+      "num_tokens": 7154880.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.6004383862018585,
+      "epoch": 0.30595050026329645,
+      "grad_norm": 0.36386990547180176,
+      "learning_rate": 0.0001999124968494082,
+      "loss": 0.22697995603084564,
+      "mean_token_accuracy": 0.9153416603803635,
+      "num_tokens": 7167216.0,
+      "step": 581
+    },
+    {
+      "entropy": 1.5263923108577728,
+      "epoch": 0.30647709320695105,
+      "grad_norm": 0.3861697316169739,
+      "learning_rate": 0.00019991152745169283,
+      "loss": 0.23612037301063538,
+      "mean_token_accuracy": 0.9100482612848282,
+      "num_tokens": 7179552.0,
+      "step": 582
+    },
+    {
+      "entropy": 1.4913224875926971,
+      "epoch": 0.3070036861506056,
+      "grad_norm": 0.37040355801582336,
+      "learning_rate": 0.00019991055271645412,
+      "loss": 0.2453731894493103,
+      "mean_token_accuracy": 0.9096489399671555,
+      "num_tokens": 7191888.0,
+      "step": 583
+    },
+    {
+      "entropy": 1.5870050489902496,
+      "epoch": 0.3075302790942601,
+      "grad_norm": 0.3563641309738159,
+      "learning_rate": 0.00019990957264374992,
+      "loss": 0.2088014781475067,
+      "mean_token_accuracy": 0.9178295880556107,
+      "num_tokens": 7204224.0,
+      "step": 584
+    },
+    {
+      "entropy": 1.5673131942749023,
+      "epoch": 0.3080568720379147,
+      "grad_norm": 0.42992061376571655,
+      "learning_rate": 0.00019990858723363842,
+      "loss": 0.22837956249713898,
+      "mean_token_accuracy": 0.9150240123271942,
+      "num_tokens": 7216560.0,
+      "step": 585
+    },
+    {
+      "entropy": 1.5697237253189087,
+      "epoch": 0.30858346498156924,
+      "grad_norm": 0.34907442331314087,
+      "learning_rate": 0.00019990759648617814,
+      "loss": 0.2061898112297058,
+      "mean_token_accuracy": 0.9186743050813675,
+      "num_tokens": 7228896.0,
+      "step": 586
+    },
+    {
+      "entropy": 1.5203896760940552,
+      "epoch": 0.3091100579252238,
+      "grad_norm": 0.3814384639263153,
+      "learning_rate": 0.00019990660040142787,
+      "loss": 0.23518475890159607,
+      "mean_token_accuracy": 0.9073454886674881,
+      "num_tokens": 7241232.0,
+      "step": 587
+    },
+    {
+      "entropy": 1.5959535539150238,
+      "epoch": 0.30963665086887837,
+      "grad_norm": 0.410187304019928,
+      "learning_rate": 0.00019990559897944675,
+      "loss": 0.2526550590991974,
+      "mean_token_accuracy": 0.9053095877170563,
+      "num_tokens": 7253568.0,
+      "step": 588
+    },
+    {
+      "entropy": 1.4929281175136566,
+      "epoch": 0.3101632438125329,
+      "grad_norm": 0.3563387989997864,
+      "learning_rate": 0.00019990459222029422,
+      "loss": 0.2284306138753891,
+      "mean_token_accuracy": 0.9153739362955093,
+      "num_tokens": 7265904.0,
+      "step": 589
+    },
+    {
+      "entropy": 1.5385482907295227,
+      "epoch": 0.31068983675618744,
+      "grad_norm": 0.3958924412727356,
+      "learning_rate": 0.00019990358012403009,
+      "loss": 0.25064951181411743,
+      "mean_token_accuracy": 0.9019092172384262,
+      "num_tokens": 7278240.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.5795462429523468,
+      "epoch": 0.31121642969984203,
+      "grad_norm": 0.4616306722164154,
+      "learning_rate": 0.00019990256269071435,
+      "loss": 0.25259649753570557,
+      "mean_token_accuracy": 0.9031090140342712,
+      "num_tokens": 7290576.0,
+      "step": 591
+    },
+    {
+      "entropy": 1.5900325179100037,
+      "epoch": 0.31174302264349657,
+      "grad_norm": 0.36789095401763916,
+      "learning_rate": 0.0001999015399204075,
+      "loss": 0.239149808883667,
+      "mean_token_accuracy": 0.9080104380846024,
+      "num_tokens": 7302912.0,
+      "step": 592
+    },
+    {
+      "entropy": 1.595876008272171,
+      "epoch": 0.3122696155871511,
+      "grad_norm": 0.3929153084754944,
+      "learning_rate": 0.0001999005118131702,
+      "loss": 0.2852938175201416,
+      "mean_token_accuracy": 0.8941920101642609,
+      "num_tokens": 7315248.0,
+      "step": 593
+    },
+    {
+      "entropy": 1.616676688194275,
+      "epoch": 0.3127962085308057,
+      "grad_norm": 0.3501856327056885,
+      "learning_rate": 0.00019989947836906346,
+      "loss": 0.24000589549541473,
+      "mean_token_accuracy": 0.9087683409452438,
+      "num_tokens": 7327584.0,
+      "step": 594
+    },
+    {
+      "entropy": 1.6581095457077026,
+      "epoch": 0.31332280147446023,
+      "grad_norm": 0.42188331484794617,
+      "learning_rate": 0.0001998984395881487,
+      "loss": 0.24285097420215607,
+      "mean_token_accuracy": 0.9076326787471771,
+      "num_tokens": 7339920.0,
+      "step": 595
+    },
+    {
+      "entropy": 1.6323899328708649,
+      "epoch": 0.3138493944181148,
+      "grad_norm": 0.38462045788764954,
+      "learning_rate": 0.00019989739547048753,
+      "loss": 0.25823602080345154,
+      "mean_token_accuracy": 0.9016465842723846,
+      "num_tokens": 7352256.0,
+      "step": 596
+    },
+    {
+      "entropy": 1.648387372493744,
+      "epoch": 0.31437598736176936,
+      "grad_norm": 0.3550112545490265,
+      "learning_rate": 0.00019989634601614198,
+      "loss": 0.23165376484394073,
+      "mean_token_accuracy": 0.9123150110244751,
+      "num_tokens": 7364592.0,
+      "step": 597
+    },
+    {
+      "entropy": 1.541102021932602,
+      "epoch": 0.3149025803054239,
+      "grad_norm": 0.32691195607185364,
+      "learning_rate": 0.0001998952912251743,
+      "loss": 0.2173168659210205,
+      "mean_token_accuracy": 0.9170237630605698,
+      "num_tokens": 7376928.0,
+      "step": 598
+    },
+    {
+      "entropy": 1.5581879019737244,
+      "epoch": 0.3154291732490785,
+      "grad_norm": 0.37725362181663513,
+      "learning_rate": 0.00019989423109764714,
+      "loss": 0.22197510302066803,
+      "mean_token_accuracy": 0.9138648211956024,
+      "num_tokens": 7389264.0,
+      "step": 599
+    },
+    {
+      "entropy": 1.5590221285820007,
+      "epoch": 0.315955766192733,
+      "grad_norm": 0.36627396941185,
+      "learning_rate": 0.0001998931656336234,
+      "loss": 0.23069526255130768,
+      "mean_token_accuracy": 0.9046057909727097,
+      "num_tokens": 7401600.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.617866188287735,
+      "epoch": 0.31648235913638756,
+      "grad_norm": 0.40031948685646057,
+      "learning_rate": 0.00019989209483316637,
+      "loss": 0.2336081862449646,
+      "mean_token_accuracy": 0.9123480021953583,
+      "num_tokens": 7413936.0,
+      "step": 601
+    },
+    {
+      "entropy": 1.53180992603302,
+      "epoch": 0.31700895208004215,
+      "grad_norm": 0.49535101652145386,
+      "learning_rate": 0.00019989101869633962,
+      "loss": 0.22568359971046448,
+      "mean_token_accuracy": 0.9089675545692444,
+      "num_tokens": 7426272.0,
+      "step": 602
+    },
+    {
+      "entropy": 1.5532157719135284,
+      "epoch": 0.3175355450236967,
+      "grad_norm": 0.4378284811973572,
+      "learning_rate": 0.00019988993722320703,
+      "loss": 0.2375514954328537,
+      "mean_token_accuracy": 0.9061621576547623,
+      "num_tokens": 7438608.0,
+      "step": 603
+    },
+    {
+      "entropy": 1.5265016555786133,
+      "epoch": 0.3180621379673512,
+      "grad_norm": 0.3522114157676697,
+      "learning_rate": 0.0001998888504138327,
+      "loss": 0.23949339985847473,
+      "mean_token_accuracy": 0.9073717594146729,
+      "num_tokens": 7450944.0,
+      "step": 604
+    },
+    {
+      "entropy": 1.61136794090271,
+      "epoch": 0.3185887309110058,
+      "grad_norm": 0.37977924942970276,
+      "learning_rate": 0.00019988775826828133,
+      "loss": 0.2230483740568161,
+      "mean_token_accuracy": 0.9096570760011673,
+      "num_tokens": 7463280.0,
+      "step": 605
+    },
+    {
+      "entropy": 1.593753159046173,
+      "epoch": 0.31911532385466035,
+      "grad_norm": 0.6494004130363464,
+      "learning_rate": 0.0001998866607866176,
+      "loss": 0.22851543128490448,
+      "mean_token_accuracy": 0.9144780784845352,
+      "num_tokens": 7475616.0,
+      "step": 606
+    },
+    {
+      "entropy": 1.5087667405605316,
+      "epoch": 0.3196419167983149,
+      "grad_norm": 0.3527071177959442,
+      "learning_rate": 0.00019988555796890674,
+      "loss": 0.20573341846466064,
+      "mean_token_accuracy": 0.9186414480209351,
+      "num_tokens": 7487952.0,
+      "step": 607
+    },
+    {
+      "entropy": 1.5721814334392548,
+      "epoch": 0.3201685097419695,
+      "grad_norm": 0.3536371886730194,
+      "learning_rate": 0.00019988444981521418,
+      "loss": 0.20877021551132202,
+      "mean_token_accuracy": 0.9166864901781082,
+      "num_tokens": 7500288.0,
+      "step": 608
+    },
+    {
+      "entropy": 1.5660102665424347,
+      "epoch": 0.320695102685624,
+      "grad_norm": 0.4063898026943207,
+      "learning_rate": 0.00019988333632560572,
+      "loss": 0.25388604402542114,
+      "mean_token_accuracy": 0.9036387950181961,
+      "num_tokens": 7512624.0,
+      "step": 609
+    },
+    {
+      "entropy": 1.5472122132778168,
+      "epoch": 0.32122169562927855,
+      "grad_norm": 0.3346962630748749,
+      "learning_rate": 0.00019988221750014747,
+      "loss": 0.2217264026403427,
+      "mean_token_accuracy": 0.9168782234191895,
+      "num_tokens": 7524960.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.557081937789917,
+      "epoch": 0.32174828857293314,
+      "grad_norm": 0.37367674708366394,
+      "learning_rate": 0.00019988109333890584,
+      "loss": 0.21202939748764038,
+      "mean_token_accuracy": 0.9218468219041824,
+      "num_tokens": 7537296.0,
+      "step": 611
+    },
+    {
+      "entropy": 1.4869562089443207,
+      "epoch": 0.3222748815165877,
+      "grad_norm": 0.4604549705982208,
+      "learning_rate": 0.00019987996384194755,
+      "loss": 0.2245841920375824,
+      "mean_token_accuracy": 0.9145530760288239,
+      "num_tokens": 7549632.0,
+      "step": 612
+    },
+    {
+      "entropy": 1.515323966741562,
+      "epoch": 0.3228014744602422,
+      "grad_norm": 0.39376160502433777,
+      "learning_rate": 0.0001998788290093397,
+      "loss": 0.26140767335891724,
+      "mean_token_accuracy": 0.8977334201335907,
+      "num_tokens": 7561968.0,
+      "step": 613
+    },
+    {
+      "entropy": 1.487097293138504,
+      "epoch": 0.3233280674038968,
+      "grad_norm": 0.40344923734664917,
+      "learning_rate": 0.00019987768884114962,
+      "loss": 0.21170933544635773,
+      "mean_token_accuracy": 0.9142201244831085,
+      "num_tokens": 7574304.0,
+      "step": 614
+    },
+    {
+      "entropy": 1.4696341156959534,
+      "epoch": 0.32385466034755134,
+      "grad_norm": 0.3404286801815033,
+      "learning_rate": 0.00019987654333744498,
+      "loss": 0.2107430100440979,
+      "mean_token_accuracy": 0.9169973582029343,
+      "num_tokens": 7586640.0,
+      "step": 615
+    },
+    {
+      "entropy": 1.4763599038124084,
+      "epoch": 0.3243812532912059,
+      "grad_norm": 0.4449232518672943,
+      "learning_rate": 0.00019987539249829381,
+      "loss": 0.22219976782798767,
+      "mean_token_accuracy": 0.9147885292768478,
+      "num_tokens": 7598976.0,
+      "step": 616
+    },
+    {
+      "entropy": 1.4189346730709076,
+      "epoch": 0.32490784623486046,
+      "grad_norm": 0.33330264687538147,
+      "learning_rate": 0.00019987423632376444,
+      "loss": 0.20641736686229706,
+      "mean_token_accuracy": 0.9154144078493118,
+      "num_tokens": 7611312.0,
+      "step": 617
+    },
+    {
+      "entropy": 1.5400767922401428,
+      "epoch": 0.325434439178515,
+      "grad_norm": 0.3576372563838959,
+      "learning_rate": 0.00019987307481392544,
+      "loss": 0.21838723123073578,
+      "mean_token_accuracy": 0.9165307581424713,
+      "num_tokens": 7623648.0,
+      "step": 618
+    },
+    {
+      "entropy": 1.4934936463832855,
+      "epoch": 0.32596103212216954,
+      "grad_norm": 0.3858436942100525,
+      "learning_rate": 0.00019987190796884582,
+      "loss": 0.2230948656797409,
+      "mean_token_accuracy": 0.9111486077308655,
+      "num_tokens": 7635984.0,
+      "step": 619
+    },
+    {
+      "entropy": 1.5231453776359558,
+      "epoch": 0.3264876250658241,
+      "grad_norm": 0.3916998505592346,
+      "learning_rate": 0.0001998707357885949,
+      "loss": 0.2513001561164856,
+      "mean_token_accuracy": 0.9021778851747513,
+      "num_tokens": 7648320.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.529598742723465,
+      "epoch": 0.32701421800947866,
+      "grad_norm": 0.768818736076355,
+      "learning_rate": 0.00019986955827324215,
+      "loss": 0.23995141685009003,
+      "mean_token_accuracy": 0.9082267880439758,
+      "num_tokens": 7660656.0,
+      "step": 621
+    },
+    {
+      "entropy": 1.56441992521286,
+      "epoch": 0.32754081095313325,
+      "grad_norm": 0.3836771249771118,
+      "learning_rate": 0.0001998683754228575,
+      "loss": 0.2571730613708496,
+      "mean_token_accuracy": 0.905523344874382,
+      "num_tokens": 7672992.0,
+      "step": 622
+    },
+    {
+      "entropy": 1.5032837986946106,
+      "epoch": 0.3280674038967878,
+      "grad_norm": 0.3536396920681,
+      "learning_rate": 0.00019986718723751122,
+      "loss": 0.2316400706768036,
+      "mean_token_accuracy": 0.9121131896972656,
+      "num_tokens": 7685328.0,
+      "step": 623
+    },
+    {
+      "entropy": 1.5374043881893158,
+      "epoch": 0.3285939968404423,
+      "grad_norm": 0.43927958607673645,
+      "learning_rate": 0.00019986599371727383,
+      "loss": 0.2222973108291626,
+      "mean_token_accuracy": 0.9122229516506195,
+      "num_tokens": 7697664.0,
+      "step": 624
+    },
+    {
+      "entropy": 1.5948379039764404,
+      "epoch": 0.3291205897840969,
+      "grad_norm": 0.3739503026008606,
+      "learning_rate": 0.00019986479486221618,
+      "loss": 0.23531857132911682,
+      "mean_token_accuracy": 0.9108096957206726,
+      "num_tokens": 7710000.0,
+      "step": 625
+    },
+    {
+      "entropy": 1.5133698880672455,
+      "epoch": 0.32964718272775145,
+      "grad_norm": 0.37042292952537537,
+      "learning_rate": 0.00019986359067240942,
+      "loss": 0.2119014859199524,
+      "mean_token_accuracy": 0.9210023730993271,
+      "num_tokens": 7722336.0,
+      "step": 626
+    },
+    {
+      "entropy": 1.565012902021408,
+      "epoch": 0.330173775671406,
+      "grad_norm": 0.36645272374153137,
+      "learning_rate": 0.00019986238114792505,
+      "loss": 0.22745972871780396,
+      "mean_token_accuracy": 0.9089792966842651,
+      "num_tokens": 7734672.0,
+      "step": 627
+    },
+    {
+      "entropy": 1.5374813377857208,
+      "epoch": 0.3307003686150606,
+      "grad_norm": 0.3918452858924866,
+      "learning_rate": 0.00019986116628883485,
+      "loss": 0.22233808040618896,
+      "mean_token_accuracy": 0.91575026512146,
+      "num_tokens": 7747008.0,
+      "step": 628
+    },
+    {
+      "entropy": 1.5814277529716492,
+      "epoch": 0.3312269615587151,
+      "grad_norm": 0.3956620395183563,
+      "learning_rate": 0.00019985994609521098,
+      "loss": 0.23237353563308716,
+      "mean_token_accuracy": 0.9120265543460846,
+      "num_tokens": 7759344.0,
+      "step": 629
+    },
+    {
+      "entropy": 1.607151210308075,
+      "epoch": 0.33175355450236965,
+      "grad_norm": 0.4154573380947113,
+      "learning_rate": 0.00019985872056712585,
+      "loss": 0.23763325810432434,
+      "mean_token_accuracy": 0.9039438813924789,
+      "num_tokens": 7771680.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.5939337015151978,
+      "epoch": 0.33228014744602424,
+      "grad_norm": 0.5367919206619263,
+      "learning_rate": 0.0001998574897046522,
+      "loss": 0.22427549958229065,
+      "mean_token_accuracy": 0.9124085754156113,
+      "num_tokens": 7784016.0,
+      "step": 631
+    },
+    {
+      "entropy": 1.6544562876224518,
+      "epoch": 0.3328067403896788,
+      "grad_norm": 0.49991947412490845,
+      "learning_rate": 0.00019985625350786313,
+      "loss": 0.26143935322761536,
+      "mean_token_accuracy": 0.8970800191164017,
+      "num_tokens": 7796352.0,
+      "step": 632
+    },
+    {
+      "entropy": 1.567628562450409,
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.43659552931785583,
+      "learning_rate": 0.00019985501197683202,
+      "loss": 0.22153861820697784,
+      "mean_token_accuracy": 0.9091014564037323,
+      "num_tokens": 7808688.0,
+      "step": 633
+    },
+    {
+      "entropy": 1.6360341310501099,
+      "epoch": 0.3338599262769879,
+      "grad_norm": 0.406440794467926,
+      "learning_rate": 0.00019985376511163255,
+      "loss": 0.23250769078731537,
+      "mean_token_accuracy": 0.9082053154706955,
+      "num_tokens": 7821024.0,
+      "step": 634
+    },
+    {
+      "entropy": 1.617436408996582,
+      "epoch": 0.33438651922064244,
+      "grad_norm": 0.4746139347553253,
+      "learning_rate": 0.00019985251291233872,
+      "loss": 0.2448023557662964,
+      "mean_token_accuracy": 0.9019272774457932,
+      "num_tokens": 7833360.0,
+      "step": 635
+    },
+    {
+      "entropy": 1.5576602518558502,
+      "epoch": 0.334913112164297,
+      "grad_norm": 0.37426477670669556,
+      "learning_rate": 0.00019985125537902497,
+      "loss": 0.24818755686283112,
+      "mean_token_accuracy": 0.9056387394666672,
+      "num_tokens": 7845696.0,
+      "step": 636
+    },
+    {
+      "entropy": 1.5108153223991394,
+      "epoch": 0.33543970510795157,
+      "grad_norm": 0.36447635293006897,
+      "learning_rate": 0.00019984999251176582,
+      "loss": 0.23007383942604065,
+      "mean_token_accuracy": 0.9102648496627808,
+      "num_tokens": 7858032.0,
+      "step": 637
+    },
+    {
+      "entropy": 1.6435940563678741,
+      "epoch": 0.3359662980516061,
+      "grad_norm": 0.4201553165912628,
+      "learning_rate": 0.00019984872431063632,
+      "loss": 0.2524194121360779,
+      "mean_token_accuracy": 0.9026057571172714,
+      "num_tokens": 7870368.0,
+      "step": 638
+    },
+    {
+      "entropy": 1.5785358846187592,
+      "epoch": 0.33649289099526064,
+      "grad_norm": 0.34197092056274414,
+      "learning_rate": 0.00019984745077571175,
+      "loss": 0.22064879536628723,
+      "mean_token_accuracy": 0.9168655872344971,
+      "num_tokens": 7882704.0,
+      "step": 639
+    },
+    {
+      "entropy": 1.5749069154262543,
+      "epoch": 0.33701948393891523,
+      "grad_norm": 0.35670214891433716,
+      "learning_rate": 0.00019984617190706768,
+      "loss": 0.210345059633255,
+      "mean_token_accuracy": 0.9159483909606934,
+      "num_tokens": 7895040.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.648300975561142,
+      "epoch": 0.33754607688256977,
+      "grad_norm": 0.5585328936576843,
+      "learning_rate": 0.00019984488770478004,
+      "loss": 0.26494815945625305,
+      "mean_token_accuracy": 0.9005215167999268,
+      "num_tokens": 7907376.0,
+      "step": 641
+    },
+    {
+      "entropy": 1.5516612827777863,
+      "epoch": 0.3380726698262243,
+      "grad_norm": 0.3812706172466278,
+      "learning_rate": 0.0001998435981689251,
+      "loss": 0.23725152015686035,
+      "mean_token_accuracy": 0.9066549986600876,
+      "num_tokens": 7919712.0,
+      "step": 642
+    },
+    {
+      "entropy": 1.538433849811554,
+      "epoch": 0.3385992627698789,
+      "grad_norm": 0.345907598733902,
+      "learning_rate": 0.00019984230329957937,
+      "loss": 0.21600767970085144,
+      "mean_token_accuracy": 0.913017600774765,
+      "num_tokens": 7932048.0,
+      "step": 643
+    },
+    {
+      "entropy": 1.5793551802635193,
+      "epoch": 0.33912585571353343,
+      "grad_norm": 0.375407874584198,
+      "learning_rate": 0.00019984100309681973,
+      "loss": 0.2297327220439911,
+      "mean_token_accuracy": 0.9100670218467712,
+      "num_tokens": 7944384.0,
+      "step": 644
+    },
+    {
+      "entropy": 1.6768275797367096,
+      "epoch": 0.33965244865718797,
+      "grad_norm": 0.3743014931678772,
+      "learning_rate": 0.00019983969756072337,
+      "loss": 0.23133614659309387,
+      "mean_token_accuracy": 0.9111513495445251,
+      "num_tokens": 7956720.0,
+      "step": 645
+    },
+    {
+      "entropy": 1.5642622113227844,
+      "epoch": 0.34017904160084256,
+      "grad_norm": 0.3964219093322754,
+      "learning_rate": 0.00019983838669136782,
+      "loss": 0.25138217210769653,
+      "mean_token_accuracy": 0.9010957926511765,
+      "num_tokens": 7969056.0,
+      "step": 646
+    },
+    {
+      "entropy": 1.562336266040802,
+      "epoch": 0.3407056345444971,
+      "grad_norm": 0.33525609970092773,
+      "learning_rate": 0.00019983707048883083,
+      "loss": 0.21662499010562897,
+      "mean_token_accuracy": 0.9179873615503311,
+      "num_tokens": 7981392.0,
+      "step": 647
+    },
+    {
+      "entropy": 1.5184325873851776,
+      "epoch": 0.3412322274881517,
+      "grad_norm": 0.36040404438972473,
+      "learning_rate": 0.0001998357489531906,
+      "loss": 0.21142053604125977,
+      "mean_token_accuracy": 0.9129342883825302,
+      "num_tokens": 7993728.0,
+      "step": 648
+    },
+    {
+      "entropy": 1.525155395269394,
+      "epoch": 0.3417588204318062,
+      "grad_norm": 0.37644830346107483,
+      "learning_rate": 0.00019983442208452553,
+      "loss": 0.24553003907203674,
+      "mean_token_accuracy": 0.9037658125162125,
+      "num_tokens": 8006064.0,
+      "step": 649
+    },
+    {
+      "entropy": 1.5811420381069183,
+      "epoch": 0.34228541337546076,
+      "grad_norm": 0.4055664837360382,
+      "learning_rate": 0.00019983308988291446,
+      "loss": 0.24890564382076263,
+      "mean_token_accuracy": 0.9027184396982193,
+      "num_tokens": 8018400.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.5781408250331879,
+      "epoch": 0.34281200631911535,
+      "grad_norm": 0.37236547470092773,
+      "learning_rate": 0.0001998317523484364,
+      "loss": 0.2316565066576004,
+      "mean_token_accuracy": 0.9119908362627029,
+      "num_tokens": 8030736.0,
+      "step": 651
+    },
+    {
+      "entropy": 1.470947027206421,
+      "epoch": 0.3433385992627699,
+      "grad_norm": 0.3868812918663025,
+      "learning_rate": 0.00019983040948117078,
+      "loss": 0.2534410059452057,
+      "mean_token_accuracy": 0.9021570980548859,
+      "num_tokens": 8043072.0,
+      "step": 652
+    },
+    {
+      "entropy": 1.5348678827285767,
+      "epoch": 0.3438651922064244,
+      "grad_norm": 0.4387357234954834,
+      "learning_rate": 0.00019982906128119732,
+      "loss": 0.22310803830623627,
+      "mean_token_accuracy": 0.9145376235246658,
+      "num_tokens": 8055408.0,
+      "step": 653
+    },
+    {
+      "entropy": 1.6187977194786072,
+      "epoch": 0.344391785150079,
+      "grad_norm": 0.3973976671695709,
+      "learning_rate": 0.00019982770774859608,
+      "loss": 0.24628473818302155,
+      "mean_token_accuracy": 0.9026829153299332,
+      "num_tokens": 8067744.0,
+      "step": 654
+    },
+    {
+      "entropy": 1.552239865064621,
+      "epoch": 0.34491837809373355,
+      "grad_norm": 0.35203489661216736,
+      "learning_rate": 0.00019982634888344737,
+      "loss": 0.2246260643005371,
+      "mean_token_accuracy": 0.9107431769371033,
+      "num_tokens": 8080080.0,
+      "step": 655
+    },
+    {
+      "entropy": 1.5582615733146667,
+      "epoch": 0.3454449710373881,
+      "grad_norm": 0.35882458090782166,
+      "learning_rate": 0.00019982498468583187,
+      "loss": 0.24929207563400269,
+      "mean_token_accuracy": 0.9079926460981369,
+      "num_tokens": 8092416.0,
+      "step": 656
+    },
+    {
+      "entropy": 1.5417989492416382,
+      "epoch": 0.3459715639810427,
+      "grad_norm": 0.3812078833580017,
+      "learning_rate": 0.00019982361515583056,
+      "loss": 0.23627105355262756,
+      "mean_token_accuracy": 0.9105822741985321,
+      "num_tokens": 8104752.0,
+      "step": 657
+    },
+    {
+      "entropy": 1.5216515362262726,
+      "epoch": 0.3464981569246972,
+      "grad_norm": 0.425227552652359,
+      "learning_rate": 0.00019982224029352477,
+      "loss": 0.24528373777866364,
+      "mean_token_accuracy": 0.9078731536865234,
+      "num_tokens": 8117088.0,
+      "step": 658
+    },
+    {
+      "entropy": 1.5206952691078186,
+      "epoch": 0.34702474986835175,
+      "grad_norm": 0.33637750148773193,
+      "learning_rate": 0.0001998208600989961,
+      "loss": 0.2107471525669098,
+      "mean_token_accuracy": 0.9135650843381882,
+      "num_tokens": 8129424.0,
+      "step": 659
+    },
+    {
+      "entropy": 1.4702115952968597,
+      "epoch": 0.34755134281200634,
+      "grad_norm": 0.39711427688598633,
+      "learning_rate": 0.00019981947457232646,
+      "loss": 0.2600536048412323,
+      "mean_token_accuracy": 0.900258481502533,
+      "num_tokens": 8141760.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.4251591563224792,
+      "epoch": 0.34807793575566087,
+      "grad_norm": 0.3307092487812042,
+      "learning_rate": 0.00019981808371359816,
+      "loss": 0.20669502019882202,
+      "mean_token_accuracy": 0.9150898307561874,
+      "num_tokens": 8154096.0,
+      "step": 661
+    },
+    {
+      "entropy": 1.4800917506217957,
+      "epoch": 0.3486045286993154,
+      "grad_norm": 0.3357905447483063,
+      "learning_rate": 0.00019981668752289372,
+      "loss": 0.21784240007400513,
+      "mean_token_accuracy": 0.9145694226026535,
+      "num_tokens": 8166432.0,
+      "step": 662
+    },
+    {
+      "entropy": 1.4939378798007965,
+      "epoch": 0.34913112164297,
+      "grad_norm": 0.3817315101623535,
+      "learning_rate": 0.00019981528600029598,
+      "loss": 0.23391678929328918,
+      "mean_token_accuracy": 0.9120414704084396,
+      "num_tokens": 8178768.0,
+      "step": 663
+    },
+    {
+      "entropy": 1.46476212143898,
+      "epoch": 0.34965771458662454,
+      "grad_norm": 0.38629287481307983,
+      "learning_rate": 0.00019981387914588822,
+      "loss": 0.2348746955394745,
+      "mean_token_accuracy": 0.9141973257064819,
+      "num_tokens": 8191104.0,
+      "step": 664
+    },
+    {
+      "entropy": 1.488828182220459,
+      "epoch": 0.35018430753027907,
+      "grad_norm": 0.43263551592826843,
+      "learning_rate": 0.00019981246695975396,
+      "loss": 0.24639199674129486,
+      "mean_token_accuracy": 0.9060526490211487,
+      "num_tokens": 8203440.0,
+      "step": 665
+    },
+    {
+      "entropy": 1.4127525091171265,
+      "epoch": 0.35071090047393366,
+      "grad_norm": 0.3648802936077118,
+      "learning_rate": 0.00019981104944197698,
+      "loss": 0.21946357190608978,
+      "mean_token_accuracy": 0.9159405082464218,
+      "num_tokens": 8215776.0,
+      "step": 666
+    },
+    {
+      "entropy": 1.4903503954410553,
+      "epoch": 0.3512374934175882,
+      "grad_norm": 0.34730279445648193,
+      "learning_rate": 0.00019980962659264144,
+      "loss": 0.2281692773103714,
+      "mean_token_accuracy": 0.9098831713199615,
+      "num_tokens": 8228112.0,
+      "step": 667
+    },
+    {
+      "entropy": 1.5247552990913391,
+      "epoch": 0.35176408636124273,
+      "grad_norm": 0.6796702742576599,
+      "learning_rate": 0.00019980819841183185,
+      "loss": 0.2541208267211914,
+      "mean_token_accuracy": 0.899626687169075,
+      "num_tokens": 8240448.0,
+      "step": 668
+    },
+    {
+      "entropy": 1.489711880683899,
+      "epoch": 0.3522906793048973,
+      "grad_norm": 0.33805036544799805,
+      "learning_rate": 0.00019980676489963294,
+      "loss": 0.2260848581790924,
+      "mean_token_accuracy": 0.9087798148393631,
+      "num_tokens": 8252784.0,
+      "step": 669
+    },
+    {
+      "entropy": 1.5027905106544495,
+      "epoch": 0.35281727224855186,
+      "grad_norm": 0.350353479385376,
+      "learning_rate": 0.00019980532605612985,
+      "loss": 0.21086406707763672,
+      "mean_token_accuracy": 0.9210447818040848,
+      "num_tokens": 8265120.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.5674294233322144,
+      "epoch": 0.35334386519220645,
+      "grad_norm": 0.3340439200401306,
+      "learning_rate": 0.00019980388188140798,
+      "loss": 0.20525240898132324,
+      "mean_token_accuracy": 0.922586515545845,
+      "num_tokens": 8277456.0,
+      "step": 671
+    },
+    {
+      "entropy": 1.5054379403591156,
+      "epoch": 0.353870458135861,
+      "grad_norm": 0.3567858934402466,
+      "learning_rate": 0.00019980243237555304,
+      "loss": 0.2413492202758789,
+      "mean_token_accuracy": 0.9065290242433548,
+      "num_tokens": 8289792.0,
+      "step": 672
+    },
+    {
+      "entropy": 1.458409994840622,
+      "epoch": 0.3543970510795155,
+      "grad_norm": 0.3392413854598999,
+      "learning_rate": 0.00019980097753865108,
+      "loss": 0.21458403766155243,
+      "mean_token_accuracy": 0.915942057967186,
+      "num_tokens": 8302128.0,
+      "step": 673
+    },
+    {
+      "entropy": 1.5474618673324585,
+      "epoch": 0.3549236440231701,
+      "grad_norm": 0.3782627582550049,
+      "learning_rate": 0.00019979951737078853,
+      "loss": 0.24051563441753387,
+      "mean_token_accuracy": 0.905216172337532,
+      "num_tokens": 8314464.0,
+      "step": 674
+    },
+    {
+      "entropy": 1.453640639781952,
+      "epoch": 0.35545023696682465,
+      "grad_norm": 0.3473234474658966,
+      "learning_rate": 0.000199798051872052,
+      "loss": 0.22346456348896027,
+      "mean_token_accuracy": 0.9117260426282883,
+      "num_tokens": 8326800.0,
+      "step": 675
+    },
+    {
+      "entropy": 1.4625252783298492,
+      "epoch": 0.3559768299104792,
+      "grad_norm": 0.3387875258922577,
+      "learning_rate": 0.0001997965810425285,
+      "loss": 0.23459070920944214,
+      "mean_token_accuracy": 0.9097730964422226,
+      "num_tokens": 8339136.0,
+      "step": 676
+    },
+    {
+      "entropy": 1.4122467339038849,
+      "epoch": 0.3565034228541338,
+      "grad_norm": 0.3719231188297272,
+      "learning_rate": 0.0001997951048823054,
+      "loss": 0.21170935034751892,
+      "mean_token_accuracy": 0.9177156686782837,
+      "num_tokens": 8351472.0,
+      "step": 677
+    },
+    {
+      "entropy": 1.3950088024139404,
+      "epoch": 0.3570300157977883,
+      "grad_norm": 0.346206933259964,
+      "learning_rate": 0.00019979362339147026,
+      "loss": 0.24381937086582184,
+      "mean_token_accuracy": 0.9096541851758957,
+      "num_tokens": 8363808.0,
+      "step": 678
+    },
+    {
+      "entropy": 1.392151266336441,
+      "epoch": 0.35755660874144285,
+      "grad_norm": 0.4163087010383606,
+      "learning_rate": 0.00019979213657011106,
+      "loss": 0.23680301010608673,
+      "mean_token_accuracy": 0.9078295677900314,
+      "num_tokens": 8376144.0,
+      "step": 679
+    },
+    {
+      "entropy": 1.272907316684723,
+      "epoch": 0.35808320168509744,
+      "grad_norm": 0.4227026104927063,
+      "learning_rate": 0.00019979064441831606,
+      "loss": 0.2319648414850235,
+      "mean_token_accuracy": 0.9085201323032379,
+      "num_tokens": 8388480.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.3540238738059998,
+      "epoch": 0.358609794628752,
+      "grad_norm": 0.41346514225006104,
+      "learning_rate": 0.00019978914693617383,
+      "loss": 0.228549063205719,
+      "mean_token_accuracy": 0.9105959385633469,
+      "num_tokens": 8400816.0,
+      "step": 681
+    },
+    {
+      "entropy": 1.360448271036148,
+      "epoch": 0.3591363875724065,
+      "grad_norm": 0.3442930281162262,
+      "learning_rate": 0.0001997876441237733,
+      "loss": 0.23922371864318848,
+      "mean_token_accuracy": 0.9045034646987915,
+      "num_tokens": 8413152.0,
+      "step": 682
+    },
+    {
+      "entropy": 1.2645371556282043,
+      "epoch": 0.3596629805160611,
+      "grad_norm": 0.3163891136646271,
+      "learning_rate": 0.00019978613598120362,
+      "loss": 0.23533323407173157,
+      "mean_token_accuracy": 0.9094915986061096,
+      "num_tokens": 8425488.0,
+      "step": 683
+    },
+    {
+      "entropy": 1.23646479845047,
+      "epoch": 0.36018957345971564,
+      "grad_norm": 0.3051406741142273,
+      "learning_rate": 0.00019978462250855439,
+      "loss": 0.23106315732002258,
+      "mean_token_accuracy": 0.908999964594841,
+      "num_tokens": 8437824.0,
+      "step": 684
+    },
+    {
+      "entropy": 1.2108670771121979,
+      "epoch": 0.3607161664033702,
+      "grad_norm": 0.3296745717525482,
+      "learning_rate": 0.00019978310370591545,
+      "loss": 0.21918885409832,
+      "mean_token_accuracy": 0.9144444465637207,
+      "num_tokens": 8450160.0,
+      "step": 685
+    },
+    {
+      "entropy": 1.2031162977218628,
+      "epoch": 0.36124275934702477,
+      "grad_norm": 0.34492868185043335,
+      "learning_rate": 0.0001997815795733769,
+      "loss": 0.23954619467258453,
+      "mean_token_accuracy": 0.9077229052782059,
+      "num_tokens": 8462496.0,
+      "step": 686
+    },
+    {
+      "entropy": 1.186975747346878,
+      "epoch": 0.3617693522906793,
+      "grad_norm": 0.3709777593612671,
+      "learning_rate": 0.00019978005011102925,
+      "loss": 0.2437351793050766,
+      "mean_token_accuracy": 0.908468559384346,
+      "num_tokens": 8474832.0,
+      "step": 687
+    },
+    {
+      "entropy": 1.2189928591251373,
+      "epoch": 0.36229594523433384,
+      "grad_norm": 0.39078283309936523,
+      "learning_rate": 0.00019977851531896335,
+      "loss": 0.23512142896652222,
+      "mean_token_accuracy": 0.9079535007476807,
+      "num_tokens": 8487168.0,
+      "step": 688
+    },
+    {
+      "entropy": 1.1402603685855865,
+      "epoch": 0.36282253817798843,
+      "grad_norm": 0.44031041860580444,
+      "learning_rate": 0.00019977697519727025,
+      "loss": 0.22278329730033875,
+      "mean_token_accuracy": 0.9095389991998672,
+      "num_tokens": 8499504.0,
+      "step": 689
+    },
+    {
+      "entropy": 1.2126906514167786,
+      "epoch": 0.36334913112164297,
+      "grad_norm": 0.4217199385166168,
+      "learning_rate": 0.00019977542974604137,
+      "loss": 0.2309587299823761,
+      "mean_token_accuracy": 0.9086143374443054,
+      "num_tokens": 8511840.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.2279771864414215,
+      "epoch": 0.3638757240652975,
+      "grad_norm": 0.38159534335136414,
+      "learning_rate": 0.00019977387896536847,
+      "loss": 0.22851325571537018,
+      "mean_token_accuracy": 0.9110317975282669,
+      "num_tokens": 8524176.0,
+      "step": 691
+    },
+    {
+      "entropy": 1.1451427340507507,
+      "epoch": 0.3644023170089521,
+      "grad_norm": 0.38951390981674194,
+      "learning_rate": 0.00019977232285534364,
+      "loss": 0.22088123857975006,
+      "mean_token_accuracy": 0.9204992055892944,
+      "num_tokens": 8536512.0,
+      "step": 692
+    },
+    {
+      "entropy": 1.2615329325199127,
+      "epoch": 0.36492890995260663,
+      "grad_norm": 0.4139882028102875,
+      "learning_rate": 0.0001997707614160592,
+      "loss": 0.24705028533935547,
+      "mean_token_accuracy": 0.9034076929092407,
+      "num_tokens": 8548848.0,
+      "step": 693
+    },
+    {
+      "entropy": 1.283850759267807,
+      "epoch": 0.36545550289626116,
+      "grad_norm": 0.369973361492157,
+      "learning_rate": 0.00019976919464760793,
+      "loss": 0.22415512800216675,
+      "mean_token_accuracy": 0.9107316434383392,
+      "num_tokens": 8561184.0,
+      "step": 694
+    },
+    {
+      "entropy": 1.2292048037052155,
+      "epoch": 0.36598209583991576,
+      "grad_norm": 0.38250577449798584,
+      "learning_rate": 0.00019976762255008277,
+      "loss": 0.23568174242973328,
+      "mean_token_accuracy": 0.9063956588506699,
+      "num_tokens": 8573520.0,
+      "step": 695
+    },
+    {
+      "entropy": 1.2365350425243378,
+      "epoch": 0.3665086887835703,
+      "grad_norm": 0.3148071765899658,
+      "learning_rate": 0.00019976604512357706,
+      "loss": 0.1988360434770584,
+      "mean_token_accuracy": 0.9224959909915924,
+      "num_tokens": 8585856.0,
+      "step": 696
+    },
+    {
+      "entropy": 1.3029758036136627,
+      "epoch": 0.3670352817272249,
+      "grad_norm": 0.3744061589241028,
+      "learning_rate": 0.00019976446236818444,
+      "loss": 0.20495107769966125,
+      "mean_token_accuracy": 0.9228791147470474,
+      "num_tokens": 8598192.0,
+      "step": 697
+    },
+    {
+      "entropy": 1.2882287502288818,
+      "epoch": 0.3675618746708794,
+      "grad_norm": 0.36938560009002686,
+      "learning_rate": 0.00019976287428399888,
+      "loss": 0.21769095957279205,
+      "mean_token_accuracy": 0.9142359048128128,
+      "num_tokens": 8610528.0,
+      "step": 698
+    },
+    {
+      "entropy": 1.3306386470794678,
+      "epoch": 0.36808846761453395,
+      "grad_norm": 0.34892189502716064,
+      "learning_rate": 0.00019976128087111468,
+      "loss": 0.22379228472709656,
+      "mean_token_accuracy": 0.910999983549118,
+      "num_tokens": 8622864.0,
+      "step": 699
+    },
+    {
+      "entropy": 1.422240823507309,
+      "epoch": 0.36861506055818855,
+      "grad_norm": 0.4013206660747528,
+      "learning_rate": 0.00019975968212962637,
+      "loss": 0.24196377396583557,
+      "mean_token_accuracy": 0.9052925109863281,
+      "num_tokens": 8635200.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.4155401587486267,
+      "epoch": 0.3691416535018431,
+      "grad_norm": 0.353211373090744,
+      "learning_rate": 0.0001997580780596289,
+      "loss": 0.23057395219802856,
+      "mean_token_accuracy": 0.9112053066492081,
+      "num_tokens": 8647536.0,
+      "step": 701
+    },
+    {
+      "entropy": 1.3654289841651917,
+      "epoch": 0.3696682464454976,
+      "grad_norm": 0.3298124372959137,
+      "learning_rate": 0.0001997564686612175,
+      "loss": 0.23379994928836823,
+      "mean_token_accuracy": 0.9082982391119003,
+      "num_tokens": 8659872.0,
+      "step": 702
+    },
+    {
+      "entropy": 1.4584643244743347,
+      "epoch": 0.3701948393891522,
+      "grad_norm": 0.3759617805480957,
+      "learning_rate": 0.00019975485393448764,
+      "loss": 0.23104792833328247,
+      "mean_token_accuracy": 0.9065845310688019,
+      "num_tokens": 8672208.0,
+      "step": 703
+    },
+    {
+      "entropy": 1.444229632616043,
+      "epoch": 0.37072143233280674,
+      "grad_norm": 0.3501303791999817,
+      "learning_rate": 0.00019975323387953528,
+      "loss": 0.2252783626317978,
+      "mean_token_accuracy": 0.9078841656446457,
+      "num_tokens": 8684544.0,
+      "step": 704
+    },
+    {
+      "entropy": 1.4961676001548767,
+      "epoch": 0.3712480252764613,
+      "grad_norm": 0.37864333391189575,
+      "learning_rate": 0.00019975160849645656,
+      "loss": 0.248760387301445,
+      "mean_token_accuracy": 0.9001511335372925,
+      "num_tokens": 8696880.0,
+      "step": 705
+    },
+    {
+      "entropy": 1.42547208070755,
+      "epoch": 0.37177461822011587,
+      "grad_norm": 0.34873253107070923,
+      "learning_rate": 0.00019974997778534793,
+      "loss": 0.21211083233356476,
+      "mean_token_accuracy": 0.9166723936796188,
+      "num_tokens": 8709216.0,
+      "step": 706
+    },
+    {
+      "entropy": 1.5042538046836853,
+      "epoch": 0.3723012111637704,
+      "grad_norm": 0.3550424873828888,
+      "learning_rate": 0.00019974834174630622,
+      "loss": 0.23886847496032715,
+      "mean_token_accuracy": 0.907848909497261,
+      "num_tokens": 8721552.0,
+      "step": 707
+    },
+    {
+      "entropy": 1.5081537961959839,
+      "epoch": 0.37282780410742494,
+      "grad_norm": 0.41367873549461365,
+      "learning_rate": 0.00019974670037942855,
+      "loss": 0.22889745235443115,
+      "mean_token_accuracy": 0.9153233766555786,
+      "num_tokens": 8733888.0,
+      "step": 708
+    },
+    {
+      "entropy": 1.3966407477855682,
+      "epoch": 0.37335439705107953,
+      "grad_norm": 0.3816620707511902,
+      "learning_rate": 0.00019974505368481238,
+      "loss": 0.22411897778511047,
+      "mean_token_accuracy": 0.9170472025871277,
+      "num_tokens": 8746224.0,
+      "step": 709
+    },
+    {
+      "entropy": 1.3528437912464142,
+      "epoch": 0.37388098999473407,
+      "grad_norm": 0.3909305930137634,
+      "learning_rate": 0.00019974340166255543,
+      "loss": 0.24883610010147095,
+      "mean_token_accuracy": 0.9069965332746506,
+      "num_tokens": 8758560.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.3574591279029846,
+      "epoch": 0.3744075829383886,
+      "grad_norm": 0.3753831088542938,
+      "learning_rate": 0.00019974174431275582,
+      "loss": 0.21758557856082916,
+      "mean_token_accuracy": 0.9138903617858887,
+      "num_tokens": 8770896.0,
+      "step": 711
+    },
+    {
+      "entropy": 1.3783180713653564,
+      "epoch": 0.3749341758820432,
+      "grad_norm": 0.3737315833568573,
+      "learning_rate": 0.0001997400816355119,
+      "loss": 0.21359983086585999,
+      "mean_token_accuracy": 0.9189057648181915,
+      "num_tokens": 8783232.0,
+      "step": 712
+    },
+    {
+      "entropy": 1.3974202573299408,
+      "epoch": 0.37546076882569773,
+      "grad_norm": 0.349575012922287,
+      "learning_rate": 0.00019973841363092232,
+      "loss": 0.2197628617286682,
+      "mean_token_accuracy": 0.910584107041359,
+      "num_tokens": 8795568.0,
+      "step": 713
+    },
+    {
+      "entropy": 1.4738838374614716,
+      "epoch": 0.37598736176935227,
+      "grad_norm": 0.39151832461357117,
+      "learning_rate": 0.0001997367402990862,
+      "loss": 0.22688592970371246,
+      "mean_token_accuracy": 0.9129604697227478,
+      "num_tokens": 8807904.0,
+      "step": 714
+    },
+    {
+      "entropy": 1.4208093285560608,
+      "epoch": 0.37651395471300686,
+      "grad_norm": 0.41211023926734924,
+      "learning_rate": 0.00019973506164010284,
+      "loss": 0.20385678112506866,
+      "mean_token_accuracy": 0.9227319210767746,
+      "num_tokens": 8820240.0,
+      "step": 715
+    },
+    {
+      "entropy": 1.4047819375991821,
+      "epoch": 0.3770405476566614,
+      "grad_norm": 0.4299006164073944,
+      "learning_rate": 0.00019973337765407187,
+      "loss": 0.23750200867652893,
+      "mean_token_accuracy": 0.9077501893043518,
+      "num_tokens": 8832576.0,
+      "step": 716
+    },
+    {
+      "entropy": 1.374906837940216,
+      "epoch": 0.37756714060031593,
+      "grad_norm": 0.36846232414245605,
+      "learning_rate": 0.00019973168834109327,
+      "loss": 0.23285596072673798,
+      "mean_token_accuracy": 0.9094402939081192,
+      "num_tokens": 8844912.0,
+      "step": 717
+    },
+    {
+      "entropy": 1.4372902512550354,
+      "epoch": 0.3780937335439705,
+      "grad_norm": 0.3780461549758911,
+      "learning_rate": 0.00019972999370126737,
+      "loss": 0.23158109188079834,
+      "mean_token_accuracy": 0.9099433869123459,
+      "num_tokens": 8857248.0,
+      "step": 718
+    },
+    {
+      "entropy": 1.3874932825565338,
+      "epoch": 0.37862032648762506,
+      "grad_norm": 0.34537607431411743,
+      "learning_rate": 0.00019972829373469467,
+      "loss": 0.21864044666290283,
+      "mean_token_accuracy": 0.9146191030740738,
+      "num_tokens": 8869584.0,
+      "step": 719
+    },
+    {
+      "entropy": 1.374584138393402,
+      "epoch": 0.3791469194312796,
+      "grad_norm": 0.33857598900794983,
+      "learning_rate": 0.00019972658844147621,
+      "loss": 0.20933406054973602,
+      "mean_token_accuracy": 0.917707160115242,
+      "num_tokens": 8881920.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.3695291578769684,
+      "epoch": 0.3796735123749342,
+      "grad_norm": 0.5717244148254395,
+      "learning_rate": 0.0001997248778217131,
+      "loss": 0.22078999876976013,
+      "mean_token_accuracy": 0.9167369604110718,
+      "num_tokens": 8894256.0,
+      "step": 721
+    },
+    {
+      "entropy": 1.26867213845253,
+      "epoch": 0.3802001053185887,
+      "grad_norm": 0.36367470026016235,
+      "learning_rate": 0.000199723161875507,
+      "loss": 0.23508650064468384,
+      "mean_token_accuracy": 0.90509033203125,
+      "num_tokens": 8906592.0,
+      "step": 722
+    },
+    {
+      "entropy": 1.3050034940242767,
+      "epoch": 0.3807266982622433,
+      "grad_norm": 0.3631988763809204,
+      "learning_rate": 0.0001997214406029597,
+      "loss": 0.22164851427078247,
+      "mean_token_accuracy": 0.9132580459117889,
+      "num_tokens": 8918928.0,
+      "step": 723
+    },
+    {
+      "entropy": 1.3181822001934052,
+      "epoch": 0.38125329120589785,
+      "grad_norm": 0.4014921486377716,
+      "learning_rate": 0.00019971971400417342,
+      "loss": 0.220194011926651,
+      "mean_token_accuracy": 0.9143774807453156,
+      "num_tokens": 8931264.0,
+      "step": 724
+    },
+    {
+      "entropy": 1.2655711770057678,
+      "epoch": 0.3817798841495524,
+      "grad_norm": 0.34297892451286316,
+      "learning_rate": 0.00019971798207925063,
+      "loss": 0.21592603623867035,
+      "mean_token_accuracy": 0.9114037603139877,
+      "num_tokens": 8943600.0,
+      "step": 725
+    },
+    {
+      "entropy": 1.3559162318706512,
+      "epoch": 0.382306477093207,
+      "grad_norm": 0.39595136046409607,
+      "learning_rate": 0.00019971624482829415,
+      "loss": 0.2135222852230072,
+      "mean_token_accuracy": 0.9156046807765961,
+      "num_tokens": 8955936.0,
+      "step": 726
+    },
+    {
+      "entropy": 1.4213075637817383,
+      "epoch": 0.3828330700368615,
+      "grad_norm": 0.35875996947288513,
+      "learning_rate": 0.00019971450225140716,
+      "loss": 0.22025166451931,
+      "mean_token_accuracy": 0.9119539260864258,
+      "num_tokens": 8968272.0,
+      "step": 727
+    },
+    {
+      "entropy": 1.2975660860538483,
+      "epoch": 0.38335966298051605,
+      "grad_norm": 0.33709245920181274,
+      "learning_rate": 0.00019971275434869302,
+      "loss": 0.2429068237543106,
+      "mean_token_accuracy": 0.9056646823883057,
+      "num_tokens": 8980608.0,
+      "step": 728
+    },
+    {
+      "entropy": 1.3944336771965027,
+      "epoch": 0.38388625592417064,
+      "grad_norm": 0.3580012321472168,
+      "learning_rate": 0.0001997110011202556,
+      "loss": 0.228425070643425,
+      "mean_token_accuracy": 0.9088682681322098,
+      "num_tokens": 8992944.0,
+      "step": 729
+    },
+    {
+      "entropy": 1.386294186115265,
+      "epoch": 0.3844128488678252,
+      "grad_norm": 0.5559288859367371,
+      "learning_rate": 0.00019970924256619888,
+      "loss": 0.22206491231918335,
+      "mean_token_accuracy": 0.9095857888460159,
+      "num_tokens": 9005280.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.3968450129032135,
+      "epoch": 0.3849394418114797,
+      "grad_norm": 0.38346561789512634,
+      "learning_rate": 0.0001997074786866273,
+      "loss": 0.2516261637210846,
+      "mean_token_accuracy": 0.9072631448507309,
+      "num_tokens": 9017616.0,
+      "step": 731
+    },
+    {
+      "entropy": 1.3758391439914703,
+      "epoch": 0.3854660347551343,
+      "grad_norm": 0.4184466302394867,
+      "learning_rate": 0.00019970570948164555,
+      "loss": 0.2083946019411087,
+      "mean_token_accuracy": 0.915515810251236,
+      "num_tokens": 9029952.0,
+      "step": 732
+    },
+    {
+      "entropy": 1.3424640595912933,
+      "epoch": 0.38599262769878884,
+      "grad_norm": 0.3482263684272766,
+      "learning_rate": 0.00019970393495135868,
+      "loss": 0.23488564789295197,
+      "mean_token_accuracy": 0.9109048992395401,
+      "num_tokens": 9042288.0,
+      "step": 733
+    },
+    {
+      "entropy": 1.3400058150291443,
+      "epoch": 0.3865192206424434,
+      "grad_norm": 0.4521651268005371,
+      "learning_rate": 0.00019970215509587205,
+      "loss": 0.24672454595565796,
+      "mean_token_accuracy": 0.9039757996797562,
+      "num_tokens": 9054624.0,
+      "step": 734
+    },
+    {
+      "entropy": 1.3971166908740997,
+      "epoch": 0.38704581358609796,
+      "grad_norm": 0.3671362102031708,
+      "learning_rate": 0.0001997003699152913,
+      "loss": 0.22332459688186646,
+      "mean_token_accuracy": 0.9150716066360474,
+      "num_tokens": 9066960.0,
+      "step": 735
+    },
+    {
+      "entropy": 1.3570407629013062,
+      "epoch": 0.3875724065297525,
+      "grad_norm": 0.4007357358932495,
+      "learning_rate": 0.00019969857940972235,
+      "loss": 0.22975003719329834,
+      "mean_token_accuracy": 0.9176786690950394,
+      "num_tokens": 9079296.0,
+      "step": 736
+    },
+    {
+      "entropy": 1.3730257153511047,
+      "epoch": 0.38809899947340704,
+      "grad_norm": 0.3905254602432251,
+      "learning_rate": 0.00019969678357927158,
+      "loss": 0.22929853200912476,
+      "mean_token_accuracy": 0.909871906042099,
+      "num_tokens": 9091632.0,
+      "step": 737
+    },
+    {
+      "entropy": 1.3547758758068085,
+      "epoch": 0.3886255924170616,
+      "grad_norm": 0.36366337537765503,
+      "learning_rate": 0.00019969498242404556,
+      "loss": 0.22672298550605774,
+      "mean_token_accuracy": 0.9105400443077087,
+      "num_tokens": 9103968.0,
+      "step": 738
+    },
+    {
+      "entropy": 1.322388231754303,
+      "epoch": 0.38915218536071616,
+      "grad_norm": 0.3948267698287964,
+      "learning_rate": 0.0001996931759441512,
+      "loss": 0.21981076896190643,
+      "mean_token_accuracy": 0.9127620607614517,
+      "num_tokens": 9116304.0,
+      "step": 739
+    },
+    {
+      "entropy": 1.3535179495811462,
+      "epoch": 0.3896787783043707,
+      "grad_norm": 0.39007002115249634,
+      "learning_rate": 0.00019969136413969577,
+      "loss": 0.23180383443832397,
+      "mean_token_accuracy": 0.9031165540218353,
+      "num_tokens": 9128640.0,
+      "step": 740
+    },
+    {
+      "entropy": 1.395143300294876,
+      "epoch": 0.3902053712480253,
+      "grad_norm": 0.39624056220054626,
+      "learning_rate": 0.00019968954701078678,
+      "loss": 0.21380290389060974,
+      "mean_token_accuracy": 0.9165746420621872,
+      "num_tokens": 9140976.0,
+      "step": 741
+    },
+    {
+      "entropy": 1.3894523680210114,
+      "epoch": 0.3907319641916798,
+      "grad_norm": 0.37505805492401123,
+      "learning_rate": 0.00019968772455753218,
+      "loss": 0.21238526701927185,
+      "mean_token_accuracy": 0.9176094681024551,
+      "num_tokens": 9153312.0,
+      "step": 742
+    },
+    {
+      "entropy": 1.3369868099689484,
+      "epoch": 0.39125855713533436,
+      "grad_norm": 0.364109605550766,
+      "learning_rate": 0.00019968589678004008,
+      "loss": 0.2217993289232254,
+      "mean_token_accuracy": 0.9145223498344421,
+      "num_tokens": 9165648.0,
+      "step": 743
+    },
+    {
+      "entropy": 1.4128968715667725,
+      "epoch": 0.39178515007898895,
+      "grad_norm": 0.37613528966903687,
+      "learning_rate": 0.00019968406367841903,
+      "loss": 0.2087160050868988,
+      "mean_token_accuracy": 0.9174602627754211,
+      "num_tokens": 9177984.0,
+      "step": 744
+    },
+    {
+      "entropy": 1.3632179498672485,
+      "epoch": 0.3923117430226435,
+      "grad_norm": 0.3908347189426422,
+      "learning_rate": 0.00019968222525277785,
+      "loss": 0.24387134611606598,
+      "mean_token_accuracy": 0.9096127450466156,
+      "num_tokens": 9190320.0,
+      "step": 745
+    },
+    {
+      "entropy": 1.3598384857177734,
+      "epoch": 0.392838335966298,
+      "grad_norm": 0.4070599377155304,
+      "learning_rate": 0.00019968038150322564,
+      "loss": 0.22589103877544403,
+      "mean_token_accuracy": 0.9176203459501266,
+      "num_tokens": 9202656.0,
+      "step": 746
+    },
+    {
+      "entropy": 1.2700338661670685,
+      "epoch": 0.3933649289099526,
+      "grad_norm": 0.3903157711029053,
+      "learning_rate": 0.0001996785324298719,
+      "loss": 0.27620363235473633,
+      "mean_token_accuracy": 0.89641934633255,
+      "num_tokens": 9214992.0,
+      "step": 747
+    },
+    {
+      "entropy": 1.3561755120754242,
+      "epoch": 0.39389152185360715,
+      "grad_norm": 0.3912152647972107,
+      "learning_rate": 0.00019967667803282637,
+      "loss": 0.20860803127288818,
+      "mean_token_accuracy": 0.918526291847229,
+      "num_tokens": 9227328.0,
+      "step": 748
+    },
+    {
+      "entropy": 1.2694224119186401,
+      "epoch": 0.39441811479726174,
+      "grad_norm": 0.36671656370162964,
+      "learning_rate": 0.00019967481831219917,
+      "loss": 0.23243974149227142,
+      "mean_token_accuracy": 0.9083072245121002,
+      "num_tokens": 9239664.0,
+      "step": 749
+    },
+    {
+      "entropy": 1.2892687618732452,
+      "epoch": 0.3949447077409163,
+      "grad_norm": 0.33213678002357483,
+      "learning_rate": 0.0001996729532681006,
+      "loss": 0.22482578456401825,
+      "mean_token_accuracy": 0.9108980298042297,
+      "num_tokens": 9252000.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.2784943878650665,
+      "epoch": 0.3954713006845708,
+      "grad_norm": 0.311139851808548,
+      "learning_rate": 0.0001996710829006415,
+      "loss": 0.22491638362407684,
+      "mean_token_accuracy": 0.9146853238344193,
+      "num_tokens": 9264336.0,
+      "step": 751
+    },
+    {
+      "entropy": 1.3392630815505981,
+      "epoch": 0.3959978936282254,
+      "grad_norm": 0.31522372364997864,
+      "learning_rate": 0.00019966920720993286,
+      "loss": 0.19653388857841492,
+      "mean_token_accuracy": 0.9204227477312088,
+      "num_tokens": 9276672.0,
+      "step": 752
+    },
+    {
+      "entropy": 1.3383756875991821,
+      "epoch": 0.39652448657187994,
+      "grad_norm": 0.353919118642807,
+      "learning_rate": 0.00019966732619608598,
+      "loss": 0.23490512371063232,
+      "mean_token_accuracy": 0.9106432646512985,
+      "num_tokens": 9289008.0,
+      "step": 753
+    },
+    {
+      "entropy": 1.3619754612445831,
+      "epoch": 0.3970510795155345,
+      "grad_norm": 0.40393969416618347,
+      "learning_rate": 0.00019966543985921258,
+      "loss": 0.25288036465644836,
+      "mean_token_accuracy": 0.8984902203083038,
+      "num_tokens": 9301344.0,
+      "step": 754
+    },
+    {
+      "entropy": 1.2913717925548553,
+      "epoch": 0.39757767245918907,
+      "grad_norm": 0.3196253478527069,
+      "learning_rate": 0.00019966354819942462,
+      "loss": 0.20618736743927002,
+      "mean_token_accuracy": 0.9167482256889343,
+      "num_tokens": 9313680.0,
+      "step": 755
+    },
+    {
+      "entropy": 1.3224665522575378,
+      "epoch": 0.3981042654028436,
+      "grad_norm": 0.32193470001220703,
+      "learning_rate": 0.0001996616512168344,
+      "loss": 0.21416081488132477,
+      "mean_token_accuracy": 0.9158345758914948,
+      "num_tokens": 9326016.0,
+      "step": 756
+    },
+    {
+      "entropy": 1.3358683288097382,
+      "epoch": 0.39863085834649814,
+      "grad_norm": 0.3324636220932007,
+      "learning_rate": 0.00019965974891155458,
+      "loss": 0.23043948411941528,
+      "mean_token_accuracy": 0.9085040837526321,
+      "num_tokens": 9338352.0,
+      "step": 757
+    },
+    {
+      "entropy": 1.3305124342441559,
+      "epoch": 0.39915745129015273,
+      "grad_norm": 0.34178489446640015,
+      "learning_rate": 0.00019965784128369798,
+      "loss": 0.22156858444213867,
+      "mean_token_accuracy": 0.9111177027225494,
+      "num_tokens": 9350688.0,
+      "step": 758
+    },
+    {
+      "entropy": 1.3993423879146576,
+      "epoch": 0.39968404423380727,
+      "grad_norm": 0.35117653012275696,
+      "learning_rate": 0.00019965592833337797,
+      "loss": 0.24540282785892487,
+      "mean_token_accuracy": 0.9031884074211121,
+      "num_tokens": 9363024.0,
+      "step": 759
+    },
+    {
+      "entropy": 1.363684356212616,
+      "epoch": 0.4002106371774618,
+      "grad_norm": 0.3588322699069977,
+      "learning_rate": 0.000199654010060708,
+      "loss": 0.23427176475524902,
+      "mean_token_accuracy": 0.9089349061250687,
+      "num_tokens": 9375360.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.3599910140037537,
+      "epoch": 0.4007372301211164,
+      "grad_norm": 0.3444473147392273,
+      "learning_rate": 0.000199652086465802,
+      "loss": 0.2288491129875183,
+      "mean_token_accuracy": 0.909742534160614,
+      "num_tokens": 9387696.0,
+      "step": 761
+    },
+    {
+      "entropy": 1.3465708196163177,
+      "epoch": 0.40126382306477093,
+      "grad_norm": 0.3881551921367645,
+      "learning_rate": 0.00019965015754877419,
+      "loss": 0.21851959824562073,
+      "mean_token_accuracy": 0.911541685461998,
+      "num_tokens": 9400032.0,
+      "step": 762
+    },
+    {
+      "entropy": 1.3606258928775787,
+      "epoch": 0.40179041600842547,
+      "grad_norm": 0.3825187683105469,
+      "learning_rate": 0.00019964822330973902,
+      "loss": 0.24844150245189667,
+      "mean_token_accuracy": 0.905233234167099,
+      "num_tokens": 9412368.0,
+      "step": 763
+    },
+    {
+      "entropy": 1.4477742910385132,
+      "epoch": 0.40231700895208006,
+      "grad_norm": 0.4269307553768158,
+      "learning_rate": 0.00019964628374881133,
+      "loss": 0.21143238246440887,
+      "mean_token_accuracy": 0.914307177066803,
+      "num_tokens": 9424704.0,
+      "step": 764
+    },
+    {
+      "entropy": 1.4273262321949005,
+      "epoch": 0.4028436018957346,
+      "grad_norm": 0.477578341960907,
+      "learning_rate": 0.0001996443388661063,
+      "loss": 0.21041172742843628,
+      "mean_token_accuracy": 0.9194318652153015,
+      "num_tokens": 9437040.0,
+      "step": 765
+    },
+    {
+      "entropy": 1.42348974943161,
+      "epoch": 0.40337019483938913,
+      "grad_norm": 0.42482423782348633,
+      "learning_rate": 0.00019964238866173933,
+      "loss": 0.22475981712341309,
+      "mean_token_accuracy": 0.9095583707094193,
+      "num_tokens": 9449376.0,
+      "step": 766
+    },
+    {
+      "entropy": 1.3782070577144623,
+      "epoch": 0.4038967877830437,
+      "grad_norm": 0.3717379570007324,
+      "learning_rate": 0.00019964043313582626,
+      "loss": 0.2339317947626114,
+      "mean_token_accuracy": 0.9064943641424179,
+      "num_tokens": 9461712.0,
+      "step": 767
+    },
+    {
+      "entropy": 1.3753153681755066,
+      "epoch": 0.40442338072669826,
+      "grad_norm": 0.32700079679489136,
+      "learning_rate": 0.00019963847228848312,
+      "loss": 0.2289605587720871,
+      "mean_token_accuracy": 0.9136441200971603,
+      "num_tokens": 9474048.0,
+      "step": 768
+    },
+    {
+      "entropy": 1.3423367142677307,
+      "epoch": 0.4049499736703528,
+      "grad_norm": 0.4840228855609894,
+      "learning_rate": 0.00019963650611982636,
+      "loss": 0.24808013439178467,
+      "mean_token_accuracy": 0.9013350456953049,
+      "num_tokens": 9486384.0,
+      "step": 769
+    },
+    {
+      "entropy": 1.388725221157074,
+      "epoch": 0.4054765666140074,
+      "grad_norm": 0.3298131823539734,
+      "learning_rate": 0.0001996345346299726,
+      "loss": 0.21849092841148376,
+      "mean_token_accuracy": 0.9153172671794891,
+      "num_tokens": 9498720.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.362036794424057,
+      "epoch": 0.4060031595576619,
+      "grad_norm": 0.423976331949234,
+      "learning_rate": 0.00019963255781903902,
+      "loss": 0.2337607443332672,
+      "mean_token_accuracy": 0.9124665558338165,
+      "num_tokens": 9511056.0,
+      "step": 771
+    },
+    {
+      "entropy": 1.321341723203659,
+      "epoch": 0.40652975250131645,
+      "grad_norm": 0.39548537135124207,
+      "learning_rate": 0.00019963057568714288,
+      "loss": 0.2241300493478775,
+      "mean_token_accuracy": 0.9098886847496033,
+      "num_tokens": 9523392.0,
+      "step": 772
+    },
+    {
+      "entropy": 1.2956216037273407,
+      "epoch": 0.40705634544497105,
+      "grad_norm": 0.36130490899086,
+      "learning_rate": 0.00019962858823440184,
+      "loss": 0.22631655633449554,
+      "mean_token_accuracy": 0.9111677557229996,
+      "num_tokens": 9535728.0,
+      "step": 773
+    },
+    {
+      "entropy": 1.3697873950004578,
+      "epoch": 0.4075829383886256,
+      "grad_norm": 0.3786010444164276,
+      "learning_rate": 0.00019962659546093395,
+      "loss": 0.22168497741222382,
+      "mean_token_accuracy": 0.9110663533210754,
+      "num_tokens": 9548064.0,
+      "step": 774
+    },
+    {
+      "entropy": 1.3025212287902832,
+      "epoch": 0.4081095313322802,
+      "grad_norm": 0.3891861140727997,
+      "learning_rate": 0.00019962459736685745,
+      "loss": 0.2645792067050934,
+      "mean_token_accuracy": 0.8977373838424683,
+      "num_tokens": 9560400.0,
+      "step": 775
+    },
+    {
+      "entropy": 1.3207987844944,
+      "epoch": 0.4086361242759347,
+      "grad_norm": 0.40660449862480164,
+      "learning_rate": 0.000199622593952291,
+      "loss": 0.23059073090553284,
+      "mean_token_accuracy": 0.9096915870904922,
+      "num_tokens": 9572736.0,
+      "step": 776
+    },
+    {
+      "entropy": 1.2878498435020447,
+      "epoch": 0.40916271721958924,
+      "grad_norm": 0.35745811462402344,
+      "learning_rate": 0.00019962058521735346,
+      "loss": 0.21263521909713745,
+      "mean_token_accuracy": 0.9142482280731201,
+      "num_tokens": 9585072.0,
+      "step": 777
+    },
+    {
+      "entropy": 1.2785497307777405,
+      "epoch": 0.40968931016324384,
+      "grad_norm": 0.3724184036254883,
+      "learning_rate": 0.00019961857116216415,
+      "loss": 0.2524154484272003,
+      "mean_token_accuracy": 0.8991471230983734,
+      "num_tokens": 9597408.0,
+      "step": 778
+    },
+    {
+      "entropy": 1.2408149540424347,
+      "epoch": 0.41021590310689837,
+      "grad_norm": 0.3448520600795746,
+      "learning_rate": 0.00019961655178684263,
+      "loss": 0.20971742272377014,
+      "mean_token_accuracy": 0.9130227416753769,
+      "num_tokens": 9609744.0,
+      "step": 779
+    },
+    {
+      "entropy": 1.2202830016613007,
+      "epoch": 0.4107424960505529,
+      "grad_norm": 0.37864333391189575,
+      "learning_rate": 0.0001996145270915087,
+      "loss": 0.23903977870941162,
+      "mean_token_accuracy": 0.9029916822910309,
+      "num_tokens": 9622080.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.1748075485229492,
+      "epoch": 0.4112690889942075,
+      "grad_norm": 0.36790093779563904,
+      "learning_rate": 0.0001996124970762827,
+      "loss": 0.24326302111148834,
+      "mean_token_accuracy": 0.9089493155479431,
+      "num_tokens": 9634416.0,
+      "step": 781
+    },
+    {
+      "entropy": 1.1609198153018951,
+      "epoch": 0.41179568193786203,
+      "grad_norm": 0.3445703685283661,
+      "learning_rate": 0.00019961046174128496,
+      "loss": 0.22028180956840515,
+      "mean_token_accuracy": 0.9107068032026291,
+      "num_tokens": 9646752.0,
+      "step": 782
+    },
+    {
+      "entropy": 1.1026230156421661,
+      "epoch": 0.41232227488151657,
+      "grad_norm": 0.3440577983856201,
+      "learning_rate": 0.00019960842108663644,
+      "loss": 0.19753555953502655,
+      "mean_token_accuracy": 0.9187215715646744,
+      "num_tokens": 9659088.0,
+      "step": 783
+    },
+    {
+      "entropy": 1.1320343911647797,
+      "epoch": 0.41284886782517116,
+      "grad_norm": 0.38988834619522095,
+      "learning_rate": 0.00019960637511245823,
+      "loss": 0.24359619617462158,
+      "mean_token_accuracy": 0.9055552929639816,
+      "num_tokens": 9671424.0,
+      "step": 784
+    },
+    {
+      "entropy": 1.1358483731746674,
+      "epoch": 0.4133754607688257,
+      "grad_norm": 0.34981632232666016,
+      "learning_rate": 0.0001996043238188718,
+      "loss": 0.24487480521202087,
+      "mean_token_accuracy": 0.9070784449577332,
+      "num_tokens": 9683760.0,
+      "step": 785
+    },
+    {
+      "entropy": 1.1212266087532043,
+      "epoch": 0.41390205371248023,
+      "grad_norm": 0.3241615891456604,
+      "learning_rate": 0.0001996022672059989,
+      "loss": 0.2086082100868225,
+      "mean_token_accuracy": 0.9190200418233871,
+      "num_tokens": 9696096.0,
+      "step": 786
+    },
+    {
+      "entropy": 1.1631585955619812,
+      "epoch": 0.4144286466561348,
+      "grad_norm": 0.36856940388679504,
+      "learning_rate": 0.00019960020527396165,
+      "loss": 0.20457178354263306,
+      "mean_token_accuracy": 0.9173056781291962,
+      "num_tokens": 9708432.0,
+      "step": 787
+    },
+    {
+      "entropy": 1.1695051193237305,
+      "epoch": 0.41495523959978936,
+      "grad_norm": 0.3899996876716614,
+      "learning_rate": 0.00019959813802288245,
+      "loss": 0.24216774106025696,
+      "mean_token_accuracy": 0.9103662818670273,
+      "num_tokens": 9720768.0,
+      "step": 788
+    },
+    {
+      "entropy": 1.1804447770118713,
+      "epoch": 0.4154818325434439,
+      "grad_norm": 0.3940189778804779,
+      "learning_rate": 0.000199596065452884,
+      "loss": 0.2604479491710663,
+      "mean_token_accuracy": 0.8987935930490494,
+      "num_tokens": 9733104.0,
+      "step": 789
+    },
+    {
+      "entropy": 1.2298611104488373,
+      "epoch": 0.4160084254870985,
+      "grad_norm": 0.3992246687412262,
+      "learning_rate": 0.00019959398756408937,
+      "loss": 0.24308446049690247,
+      "mean_token_accuracy": 0.9046371877193451,
+      "num_tokens": 9745440.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.130595326423645,
+      "epoch": 0.416535018430753,
+      "grad_norm": 0.32028424739837646,
+      "learning_rate": 0.00019959190435662186,
+      "loss": 0.21361656486988068,
+      "mean_token_accuracy": 0.9175485223531723,
+      "num_tokens": 9757776.0,
+      "step": 791
+    },
+    {
+      "entropy": 1.2031766474246979,
+      "epoch": 0.41706161137440756,
+      "grad_norm": 0.3323441445827484,
+      "learning_rate": 0.00019958981583060518,
+      "loss": 0.22529007494449615,
+      "mean_token_accuracy": 0.9116253554821014,
+      "num_tokens": 9770112.0,
+      "step": 792
+    },
+    {
+      "entropy": 1.1709198653697968,
+      "epoch": 0.41758820431806215,
+      "grad_norm": 0.3305797874927521,
+      "learning_rate": 0.0001995877219861633,
+      "loss": 0.21258684992790222,
+      "mean_token_accuracy": 0.9117163866758347,
+      "num_tokens": 9782448.0,
+      "step": 793
+    },
+    {
+      "entropy": 1.2462036609649658,
+      "epoch": 0.4181147972617167,
+      "grad_norm": 0.32563477754592896,
+      "learning_rate": 0.00019958562282342056,
+      "loss": 0.22679030895233154,
+      "mean_token_accuracy": 0.9114914387464523,
+      "num_tokens": 9794784.0,
+      "step": 794
+    },
+    {
+      "entropy": 1.1364922225475311,
+      "epoch": 0.4186413902053712,
+      "grad_norm": 0.3199447989463806,
+      "learning_rate": 0.0001995835183425015,
+      "loss": 0.22316469252109528,
+      "mean_token_accuracy": 0.9091953933238983,
+      "num_tokens": 9807120.0,
+      "step": 795
+    },
+    {
+      "entropy": 1.2300072014331818,
+      "epoch": 0.4191679831490258,
+      "grad_norm": 0.3554495871067047,
+      "learning_rate": 0.0001995814085435311,
+      "loss": 0.2148018628358841,
+      "mean_token_accuracy": 0.9172432273626328,
+      "num_tokens": 9819456.0,
+      "step": 796
+    },
+    {
+      "entropy": 1.2751358449459076,
+      "epoch": 0.41969457609268035,
+      "grad_norm": 0.3625534772872925,
+      "learning_rate": 0.00019957929342663457,
+      "loss": 0.217167928814888,
+      "mean_token_accuracy": 0.9148837774991989,
+      "num_tokens": 9831792.0,
+      "step": 797
+    },
+    {
+      "entropy": 1.2123389542102814,
+      "epoch": 0.42022116903633494,
+      "grad_norm": 0.3511140048503876,
+      "learning_rate": 0.00019957717299193752,
+      "loss": 0.21881210803985596,
+      "mean_token_accuracy": 0.9147087782621384,
+      "num_tokens": 9844128.0,
+      "step": 798
+    },
+    {
+      "entropy": 1.2617943584918976,
+      "epoch": 0.4207477619799895,
+      "grad_norm": 0.3321590721607208,
+      "learning_rate": 0.0001995750472395658,
+      "loss": 0.22508296370506287,
+      "mean_token_accuracy": 0.9119736105203629,
+      "num_tokens": 9856464.0,
+      "step": 799
+    },
+    {
+      "entropy": 1.2528738677501678,
+      "epoch": 0.421274354923644,
+      "grad_norm": 0.3493482768535614,
+      "learning_rate": 0.00019957291616964565,
+      "loss": 0.234869584441185,
+      "mean_token_accuracy": 0.9126180708408356,
+      "num_tokens": 9868800.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.263050615787506,
+      "epoch": 0.4218009478672986,
+      "grad_norm": 0.33217278122901917,
+      "learning_rate": 0.00019957077978230347,
+      "loss": 0.22601178288459778,
+      "mean_token_accuracy": 0.9101445376873016,
+      "num_tokens": 9881136.0,
+      "step": 801
+    },
+    {
+      "entropy": 1.279297262430191,
+      "epoch": 0.42232754081095314,
+      "grad_norm": 0.3346133232116699,
+      "learning_rate": 0.00019956863807766618,
+      "loss": 0.22423239052295685,
+      "mean_token_accuracy": 0.9134156703948975,
+      "num_tokens": 9893472.0,
+      "step": 802
+    },
+    {
+      "entropy": 1.342239260673523,
+      "epoch": 0.4228541337546077,
+      "grad_norm": 0.32531458139419556,
+      "learning_rate": 0.00019956649105586088,
+      "loss": 0.23583942651748657,
+      "mean_token_accuracy": 0.9080929756164551,
+      "num_tokens": 9905808.0,
+      "step": 803
+    },
+    {
+      "entropy": 1.301451414823532,
+      "epoch": 0.42338072669826227,
+      "grad_norm": 0.3336106240749359,
+      "learning_rate": 0.00019956433871701508,
+      "loss": 0.23021520674228668,
+      "mean_token_accuracy": 0.9132036864757538,
+      "num_tokens": 9918144.0,
+      "step": 804
+    },
+    {
+      "entropy": 1.294978141784668,
+      "epoch": 0.4239073196419168,
+      "grad_norm": 0.36648303270339966,
+      "learning_rate": 0.00019956218106125648,
+      "loss": 0.216038778424263,
+      "mean_token_accuracy": 0.9193701446056366,
+      "num_tokens": 9930480.0,
+      "step": 805
+    },
+    {
+      "entropy": 1.3323890268802643,
+      "epoch": 0.42443391258557134,
+      "grad_norm": 0.5710987448692322,
+      "learning_rate": 0.00019956001808871322,
+      "loss": 0.21820521354675293,
+      "mean_token_accuracy": 0.9192979484796524,
+      "num_tokens": 9942816.0,
+      "step": 806
+    },
+    {
+      "entropy": 1.3704762756824493,
+      "epoch": 0.42496050552922593,
+      "grad_norm": 0.3389364778995514,
+      "learning_rate": 0.00019955784979951366,
+      "loss": 0.20125524699687958,
+      "mean_token_accuracy": 0.9225403815507889,
+      "num_tokens": 9955152.0,
+      "step": 807
+    },
+    {
+      "entropy": 1.361533671617508,
+      "epoch": 0.42548709847288047,
+      "grad_norm": 0.36687371134757996,
+      "learning_rate": 0.00019955567619378653,
+      "loss": 0.22097808122634888,
+      "mean_token_accuracy": 0.9130203276872635,
+      "num_tokens": 9967488.0,
+      "step": 808
+    },
+    {
+      "entropy": 1.3404709100723267,
+      "epoch": 0.426013691416535,
+      "grad_norm": 0.3457012474536896,
+      "learning_rate": 0.00019955349727166088,
+      "loss": 0.21268591284751892,
+      "mean_token_accuracy": 0.9194072186946869,
+      "num_tokens": 9979824.0,
+      "step": 809
+    },
+    {
+      "entropy": 1.3375678360462189,
+      "epoch": 0.4265402843601896,
+      "grad_norm": 0.3357281982898712,
+      "learning_rate": 0.0001995513130332661,
+      "loss": 0.21844780445098877,
+      "mean_token_accuracy": 0.9134986102581024,
+      "num_tokens": 9992160.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.373364269733429,
+      "epoch": 0.42706687730384413,
+      "grad_norm": 0.3841971457004547,
+      "learning_rate": 0.00019954912347873175,
+      "loss": 0.2078622281551361,
+      "mean_token_accuracy": 0.9202176630496979,
+      "num_tokens": 10004496.0,
+      "step": 811
+    },
+    {
+      "entropy": 1.3305988311767578,
+      "epoch": 0.42759347024749866,
+      "grad_norm": 0.3852640986442566,
+      "learning_rate": 0.0001995469286081879,
+      "loss": 0.2230588048696518,
+      "mean_token_accuracy": 0.9157514870166779,
+      "num_tokens": 10016832.0,
+      "step": 812
+    },
+    {
+      "entropy": 1.3437947630882263,
+      "epoch": 0.42812006319115326,
+      "grad_norm": 0.390927791595459,
+      "learning_rate": 0.0001995447284217648,
+      "loss": 0.22504432499408722,
+      "mean_token_accuracy": 0.911404550075531,
+      "num_tokens": 10029168.0,
+      "step": 813
+    },
+    {
+      "entropy": 1.4292974174022675,
+      "epoch": 0.4286466561348078,
+      "grad_norm": 0.3649219274520874,
+      "learning_rate": 0.00019954252291959313,
+      "loss": 0.22576412558555603,
+      "mean_token_accuracy": 0.9123170077800751,
+      "num_tokens": 10041504.0,
+      "step": 814
+    },
+    {
+      "entropy": 1.3408824503421783,
+      "epoch": 0.4291732490784623,
+      "grad_norm": 0.4012393355369568,
+      "learning_rate": 0.00019954031210180371,
+      "loss": 0.2239188551902771,
+      "mean_token_accuracy": 0.9098076522350311,
+      "num_tokens": 10053840.0,
+      "step": 815
+    },
+    {
+      "entropy": 1.3955635130405426,
+      "epoch": 0.4296998420221169,
+      "grad_norm": 0.31125685572624207,
+      "learning_rate": 0.00019953809596852786,
+      "loss": 0.21482022106647491,
+      "mean_token_accuracy": 0.9152773320674896,
+      "num_tokens": 10066176.0,
+      "step": 816
+    },
+    {
+      "entropy": 1.372936338186264,
+      "epoch": 0.43022643496577145,
+      "grad_norm": 0.3509393632411957,
+      "learning_rate": 0.00019953587451989712,
+      "loss": 0.2275184839963913,
+      "mean_token_accuracy": 0.9097452610731125,
+      "num_tokens": 10078512.0,
+      "step": 817
+    },
+    {
+      "entropy": 1.2875484228134155,
+      "epoch": 0.430753027909426,
+      "grad_norm": 0.32160601019859314,
+      "learning_rate": 0.00019953364775604336,
+      "loss": 0.22542208433151245,
+      "mean_token_accuracy": 0.9094801992177963,
+      "num_tokens": 10090848.0,
+      "step": 818
+    },
+    {
+      "entropy": 1.387698233127594,
+      "epoch": 0.4312796208530806,
+      "grad_norm": 0.3713604807853699,
+      "learning_rate": 0.00019953141567709878,
+      "loss": 0.2560008466243744,
+      "mean_token_accuracy": 0.9061192125082016,
+      "num_tokens": 10103184.0,
+      "step": 819
+    },
+    {
+      "entropy": 1.3468024134635925,
+      "epoch": 0.4318062137967351,
+      "grad_norm": 0.30053895711898804,
+      "learning_rate": 0.00019952917828319587,
+      "loss": 0.21902483701705933,
+      "mean_token_accuracy": 0.9183452129364014,
+      "num_tokens": 10115520.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.2196392714977264,
+      "epoch": 0.43233280674038965,
+      "grad_norm": 0.3135424256324768,
+      "learning_rate": 0.00019952693557446748,
+      "loss": 0.2191144973039627,
+      "mean_token_accuracy": 0.9174018204212189,
+      "num_tokens": 10127856.0,
+      "step": 821
+    },
+    {
+      "entropy": 1.341955155134201,
+      "epoch": 0.43285939968404424,
+      "grad_norm": 0.31046387553215027,
+      "learning_rate": 0.00019952468755104672,
+      "loss": 0.19744142889976501,
+      "mean_token_accuracy": 0.9225213825702667,
+      "num_tokens": 10140192.0,
+      "step": 822
+    },
+    {
+      "entropy": 1.2434613406658173,
+      "epoch": 0.4333859926276988,
+      "grad_norm": 0.3251499831676483,
+      "learning_rate": 0.00019952243421306708,
+      "loss": 0.22867801785469055,
+      "mean_token_accuracy": 0.9065830409526825,
+      "num_tokens": 10152528.0,
+      "step": 823
+    },
+    {
+      "entropy": 1.2479578852653503,
+      "epoch": 0.43391258557135337,
+      "grad_norm": 0.33372238278388977,
+      "learning_rate": 0.0001995201755606622,
+      "loss": 0.24028503894805908,
+      "mean_token_accuracy": 0.9034052640199661,
+      "num_tokens": 10164864.0,
+      "step": 824
+    },
+    {
+      "entropy": 1.2929136455059052,
+      "epoch": 0.4344391785150079,
+      "grad_norm": 0.3748015761375427,
+      "learning_rate": 0.00019951791159396635,
+      "loss": 0.22330275177955627,
+      "mean_token_accuracy": 0.9107150882482529,
+      "num_tokens": 10177200.0,
+      "step": 825
+    },
+    {
+      "entropy": 1.1717173159122467,
+      "epoch": 0.43496577145866244,
+      "grad_norm": 0.32337212562561035,
+      "learning_rate": 0.00019951564231311382,
+      "loss": 0.20805582404136658,
+      "mean_token_accuracy": 0.9200474917888641,
+      "num_tokens": 10189536.0,
+      "step": 826
+    },
+    {
+      "entropy": 1.2089639604091644,
+      "epoch": 0.43549236440231703,
+      "grad_norm": 0.3620331883430481,
+      "learning_rate": 0.00019951336771823936,
+      "loss": 0.24102292954921722,
+      "mean_token_accuracy": 0.9084760844707489,
+      "num_tokens": 10201872.0,
+      "step": 827
+    },
+    {
+      "entropy": 1.2690958082675934,
+      "epoch": 0.43601895734597157,
+      "grad_norm": 0.3436319828033447,
+      "learning_rate": 0.00019951108780947793,
+      "loss": 0.2396576702594757,
+      "mean_token_accuracy": 0.9085531234741211,
+      "num_tokens": 10214208.0,
+      "step": 828
+    },
+    {
+      "entropy": 1.1895755529403687,
+      "epoch": 0.4365455502896261,
+      "grad_norm": 0.3198072612285614,
+      "learning_rate": 0.00019950880258696495,
+      "loss": 0.21371476352214813,
+      "mean_token_accuracy": 0.9214041531085968,
+      "num_tokens": 10226544.0,
+      "step": 829
+    },
+    {
+      "entropy": 1.2306865453720093,
+      "epoch": 0.4370721432332807,
+      "grad_norm": 0.33646130561828613,
+      "learning_rate": 0.00019950651205083607,
+      "loss": 0.22665852308273315,
+      "mean_token_accuracy": 0.9108365774154663,
+      "num_tokens": 10238880.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.2199313044548035,
+      "epoch": 0.43759873617693523,
+      "grad_norm": 0.2950493395328522,
+      "learning_rate": 0.00019950421620122726,
+      "loss": 0.19776280224323273,
+      "mean_token_accuracy": 0.922358974814415,
+      "num_tokens": 10251216.0,
+      "step": 831
+    },
+    {
+      "entropy": 1.2803155779838562,
+      "epoch": 0.43812532912058977,
+      "grad_norm": 0.3180263638496399,
+      "learning_rate": 0.00019950191503827477,
+      "loss": 0.2287098914384842,
+      "mean_token_accuracy": 0.9091289192438126,
+      "num_tokens": 10263552.0,
+      "step": 832
+    },
+    {
+      "entropy": 1.278805285692215,
+      "epoch": 0.43865192206424436,
+      "grad_norm": 0.30920109152793884,
+      "learning_rate": 0.00019949960856211526,
+      "loss": 0.22083017230033875,
+      "mean_token_accuracy": 0.915079653263092,
+      "num_tokens": 10275888.0,
+      "step": 833
+    },
+    {
+      "entropy": 1.255694329738617,
+      "epoch": 0.4391785150078989,
+      "grad_norm": 0.3112378716468811,
+      "learning_rate": 0.00019949729677288568,
+      "loss": 0.22883734107017517,
+      "mean_token_accuracy": 0.9063798487186432,
+      "num_tokens": 10288224.0,
+      "step": 834
+    },
+    {
+      "entropy": 1.2695356905460358,
+      "epoch": 0.43970510795155343,
+      "grad_norm": 0.36405643820762634,
+      "learning_rate": 0.00019949497967072314,
+      "loss": 0.25639811158180237,
+      "mean_token_accuracy": 0.8983906358480453,
+      "num_tokens": 10300560.0,
+      "step": 835
+    },
+    {
+      "entropy": 1.1852473616600037,
+      "epoch": 0.440231700895208,
+      "grad_norm": 0.3179355561733246,
+      "learning_rate": 0.00019949265725576534,
+      "loss": 0.2258102297782898,
+      "mean_token_accuracy": 0.911256417632103,
+      "num_tokens": 10312896.0,
+      "step": 836
+    },
+    {
+      "entropy": 1.2247830629348755,
+      "epoch": 0.44075829383886256,
+      "grad_norm": 0.30152472853660583,
+      "learning_rate": 0.00019949032952815002,
+      "loss": 0.22182312607765198,
+      "mean_token_accuracy": 0.9133886694908142,
+      "num_tokens": 10325232.0,
+      "step": 837
+    },
+    {
+      "entropy": 1.2358119487762451,
+      "epoch": 0.4412848867825171,
+      "grad_norm": 0.3537450134754181,
+      "learning_rate": 0.00019948799648801546,
+      "loss": 0.23890867829322815,
+      "mean_token_accuracy": 0.9061432033777237,
+      "num_tokens": 10337568.0,
+      "step": 838
+    },
+    {
+      "entropy": 1.255542904138565,
+      "epoch": 0.4418114797261717,
+      "grad_norm": 0.3185861110687256,
+      "learning_rate": 0.00019948565813550012,
+      "loss": 0.2323358654975891,
+      "mean_token_accuracy": 0.910455733537674,
+      "num_tokens": 10349904.0,
+      "step": 839
+    },
+    {
+      "entropy": 1.2060716450214386,
+      "epoch": 0.4423380726698262,
+      "grad_norm": 0.2943599820137024,
+      "learning_rate": 0.00019948331447074282,
+      "loss": 0.1973709762096405,
+      "mean_token_accuracy": 0.9198663532733917,
+      "num_tokens": 10362240.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.23261758685112,
+      "epoch": 0.44286466561348076,
+      "grad_norm": 0.3526243269443512,
+      "learning_rate": 0.00019948096549388269,
+      "loss": 0.2500985562801361,
+      "mean_token_accuracy": 0.9066997915506363,
+      "num_tokens": 10374576.0,
+      "step": 841
+    },
+    {
+      "entropy": 1.237447440624237,
+      "epoch": 0.44339125855713535,
+      "grad_norm": 0.3191603422164917,
+      "learning_rate": 0.00019947861120505914,
+      "loss": 0.21682989597320557,
+      "mean_token_accuracy": 0.914885938167572,
+      "num_tokens": 10386912.0,
+      "step": 842
+    },
+    {
+      "entropy": 1.2385708093643188,
+      "epoch": 0.4439178515007899,
+      "grad_norm": 0.31065261363983154,
+      "learning_rate": 0.000199476251604412,
+      "loss": 0.22177095711231232,
+      "mean_token_accuracy": 0.9108449369668961,
+      "num_tokens": 10399248.0,
+      "step": 843
+    },
+    {
+      "entropy": 1.209405928850174,
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.3202918767929077,
+      "learning_rate": 0.0001994738866920813,
+      "loss": 0.23262163996696472,
+      "mean_token_accuracy": 0.906676635146141,
+      "num_tokens": 10411584.0,
+      "step": 844
+    },
+    {
+      "entropy": 1.2493412792682648,
+      "epoch": 0.444971037388099,
+      "grad_norm": 0.3377321660518646,
+      "learning_rate": 0.00019947151646820745,
+      "loss": 0.25681400299072266,
+      "mean_token_accuracy": 0.9044294059276581,
+      "num_tokens": 10423920.0,
+      "step": 845
+    },
+    {
+      "entropy": 1.2441616654396057,
+      "epoch": 0.44549763033175355,
+      "grad_norm": 0.3140577971935272,
+      "learning_rate": 0.00019946914093293115,
+      "loss": 0.23581147193908691,
+      "mean_token_accuracy": 0.9107285887002945,
+      "num_tokens": 10436256.0,
+      "step": 846
+    },
+    {
+      "entropy": 1.2595540583133698,
+      "epoch": 0.4460242232754081,
+      "grad_norm": 0.32932135462760925,
+      "learning_rate": 0.00019946676008639343,
+      "loss": 0.2249414175748825,
+      "mean_token_accuracy": 0.9159797579050064,
+      "num_tokens": 10448592.0,
+      "step": 847
+    },
+    {
+      "entropy": 1.2226338982582092,
+      "epoch": 0.4465508162190627,
+      "grad_norm": 0.3085028827190399,
+      "learning_rate": 0.00019946437392873555,
+      "loss": 0.22729279100894928,
+      "mean_token_accuracy": 0.914769247174263,
+      "num_tokens": 10460928.0,
+      "step": 848
+    },
+    {
+      "entropy": 1.2365905344486237,
+      "epoch": 0.4470774091627172,
+      "grad_norm": 0.33299946784973145,
+      "learning_rate": 0.00019946198246009928,
+      "loss": 0.23490425944328308,
+      "mean_token_accuracy": 0.9082431495189667,
+      "num_tokens": 10473264.0,
+      "step": 849
+    },
+    {
+      "entropy": 1.4061111807823181,
+      "epoch": 0.4476040021063718,
+      "grad_norm": 0.3621409833431244,
+      "learning_rate": 0.00019945958568062656,
+      "loss": 0.22818249464035034,
+      "mean_token_accuracy": 0.9102335572242737,
+      "num_tokens": 10485600.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.2550583481788635,
+      "epoch": 0.44813059505002634,
+      "grad_norm": 0.33583348989486694,
+      "learning_rate": 0.0001994571835904596,
+      "loss": 0.2311660647392273,
+      "mean_token_accuracy": 0.9055789560079575,
+      "num_tokens": 10497936.0,
+      "step": 851
+    },
+    {
+      "entropy": 1.306765854358673,
+      "epoch": 0.4486571879936809,
+      "grad_norm": 0.34734323620796204,
+      "learning_rate": 0.00019945477618974104,
+      "loss": 0.24055415391921997,
+      "mean_token_accuracy": 0.9080438315868378,
+      "num_tokens": 10510272.0,
+      "step": 852
+    },
+    {
+      "entropy": 1.3021958768367767,
+      "epoch": 0.44918378093733546,
+      "grad_norm": 0.35933780670166016,
+      "learning_rate": 0.00019945236347861383,
+      "loss": 0.22972102463245392,
+      "mean_token_accuracy": 0.9099899232387543,
+      "num_tokens": 10522608.0,
+      "step": 853
+    },
+    {
+      "entropy": 1.2393949925899506,
+      "epoch": 0.44971037388099,
+      "grad_norm": 0.33583781123161316,
+      "learning_rate": 0.00019944994545722115,
+      "loss": 0.19537097215652466,
+      "mean_token_accuracy": 0.9247924536466599,
+      "num_tokens": 10534944.0,
+      "step": 854
+    },
+    {
+      "entropy": 1.2775700390338898,
+      "epoch": 0.45023696682464454,
+      "grad_norm": 0.33388838171958923,
+      "learning_rate": 0.00019944752212570658,
+      "loss": 0.20106476545333862,
+      "mean_token_accuracy": 0.9226871877908707,
+      "num_tokens": 10547280.0,
+      "step": 855
+    },
+    {
+      "entropy": 1.2714512348175049,
+      "epoch": 0.4507635597682991,
+      "grad_norm": 0.3518841862678528,
+      "learning_rate": 0.00019944509348421394,
+      "loss": 0.24091394245624542,
+      "mean_token_accuracy": 0.9107283055782318,
+      "num_tokens": 10559616.0,
+      "step": 856
+    },
+    {
+      "entropy": 1.3004609048366547,
+      "epoch": 0.45129015271195366,
+      "grad_norm": 0.31591442227363586,
+      "learning_rate": 0.00019944265953288746,
+      "loss": 0.22405283153057098,
+      "mean_token_accuracy": 0.9140247255563736,
+      "num_tokens": 10571952.0,
+      "step": 857
+    },
+    {
+      "entropy": 1.2649115324020386,
+      "epoch": 0.4518167456556082,
+      "grad_norm": 0.34270599484443665,
+      "learning_rate": 0.00019944022027187157,
+      "loss": 0.2242221236228943,
+      "mean_token_accuracy": 0.9118231236934662,
+      "num_tokens": 10584288.0,
+      "step": 858
+    },
+    {
+      "entropy": 1.20332869887352,
+      "epoch": 0.4523433385992628,
+      "grad_norm": 0.33847615122795105,
+      "learning_rate": 0.0001994377757013111,
+      "loss": 0.22582295536994934,
+      "mean_token_accuracy": 0.9115125685930252,
+      "num_tokens": 10596624.0,
+      "step": 859
+    },
+    {
+      "entropy": 1.2661672830581665,
+      "epoch": 0.4528699315429173,
+      "grad_norm": 0.34885501861572266,
+      "learning_rate": 0.00019943532582135116,
+      "loss": 0.2379041314125061,
+      "mean_token_accuracy": 0.9074203372001648,
+      "num_tokens": 10608960.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.2014594674110413,
+      "epoch": 0.45339652448657186,
+      "grad_norm": 0.29767417907714844,
+      "learning_rate": 0.0001994328706321372,
+      "loss": 0.20825329422950745,
+      "mean_token_accuracy": 0.9220836907625198,
+      "num_tokens": 10621296.0,
+      "step": 861
+    },
+    {
+      "entropy": 1.1979358792304993,
+      "epoch": 0.45392311743022645,
+      "grad_norm": 0.33858057856559753,
+      "learning_rate": 0.000199430410133815,
+      "loss": 0.21688413619995117,
+      "mean_token_accuracy": 0.9154618680477142,
+      "num_tokens": 10633632.0,
+      "step": 862
+    },
+    {
+      "entropy": 1.1113716661930084,
+      "epoch": 0.454449710373881,
+      "grad_norm": 0.32101956009864807,
+      "learning_rate": 0.00019942794432653053,
+      "loss": 0.20722433924674988,
+      "mean_token_accuracy": 0.9159047454595566,
+      "num_tokens": 10645968.0,
+      "step": 863
+    },
+    {
+      "entropy": 1.1962774693965912,
+      "epoch": 0.4549763033175355,
+      "grad_norm": 0.34402385354042053,
+      "learning_rate": 0.00019942547321043027,
+      "loss": 0.2361735999584198,
+      "mean_token_accuracy": 0.9091803878545761,
+      "num_tokens": 10658304.0,
+      "step": 864
+    },
+    {
+      "entropy": 1.2290304005146027,
+      "epoch": 0.4555028962611901,
+      "grad_norm": 0.35236313939094543,
+      "learning_rate": 0.00019942299678566086,
+      "loss": 0.21090444922447205,
+      "mean_token_accuracy": 0.9149055033922195,
+      "num_tokens": 10670640.0,
+      "step": 865
+    },
+    {
+      "entropy": 1.1575745642185211,
+      "epoch": 0.45602948920484465,
+      "grad_norm": 0.34065473079681396,
+      "learning_rate": 0.00019942051505236936,
+      "loss": 0.2252214103937149,
+      "mean_token_accuracy": 0.9156748950481415,
+      "num_tokens": 10682976.0,
+      "step": 866
+    },
+    {
+      "entropy": 1.216041624546051,
+      "epoch": 0.4565560821484992,
+      "grad_norm": 0.34139883518218994,
+      "learning_rate": 0.00019941802801070303,
+      "loss": 0.2223672717809677,
+      "mean_token_accuracy": 0.9103764742612839,
+      "num_tokens": 10695312.0,
+      "step": 867
+    },
+    {
+      "entropy": 1.210105150938034,
+      "epoch": 0.4570826750921538,
+      "grad_norm": 0.33792680501937866,
+      "learning_rate": 0.00019941553566080956,
+      "loss": 0.21836510300636292,
+      "mean_token_accuracy": 0.9169131815433502,
+      "num_tokens": 10707648.0,
+      "step": 868
+    },
+    {
+      "entropy": 1.2123220264911652,
+      "epoch": 0.4576092680358083,
+      "grad_norm": 0.33836740255355835,
+      "learning_rate": 0.0001994130380028369,
+      "loss": 0.22877393662929535,
+      "mean_token_accuracy": 0.9094105064868927,
+      "num_tokens": 10719984.0,
+      "step": 869
+    },
+    {
+      "entropy": 1.2113963067531586,
+      "epoch": 0.45813586097946285,
+      "grad_norm": 0.3214447796344757,
+      "learning_rate": 0.00019941053503693329,
+      "loss": 0.22401154041290283,
+      "mean_token_accuracy": 0.9101571440696716,
+      "num_tokens": 10732320.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.2003886699676514,
+      "epoch": 0.45866245392311744,
+      "grad_norm": 0.32083162665367126,
+      "learning_rate": 0.00019940802676324734,
+      "loss": 0.22345268726348877,
+      "mean_token_accuracy": 0.909741148352623,
+      "num_tokens": 10744656.0,
+      "step": 871
+    },
+    {
+      "entropy": 1.2443215548992157,
+      "epoch": 0.459189046866772,
+      "grad_norm": 0.32810378074645996,
+      "learning_rate": 0.00019940551318192796,
+      "loss": 0.22558192908763885,
+      "mean_token_accuracy": 0.911412313580513,
+      "num_tokens": 10756992.0,
+      "step": 872
+    },
+    {
+      "entropy": 1.2517528235912323,
+      "epoch": 0.4597156398104265,
+      "grad_norm": 0.2983804941177368,
+      "learning_rate": 0.00019940299429312437,
+      "loss": 0.22902357578277588,
+      "mean_token_accuracy": 0.9084698557853699,
+      "num_tokens": 10769328.0,
+      "step": 873
+    },
+    {
+      "entropy": 1.331667423248291,
+      "epoch": 0.4602422327540811,
+      "grad_norm": 0.33570387959480286,
+      "learning_rate": 0.00019940047009698605,
+      "loss": 0.21841317415237427,
+      "mean_token_accuracy": 0.9151860028505325,
+      "num_tokens": 10781664.0,
+      "step": 874
+    },
+    {
+      "entropy": 1.2408846318721771,
+      "epoch": 0.46076882569773564,
+      "grad_norm": 0.3269551992416382,
+      "learning_rate": 0.00019939794059366294,
+      "loss": 0.21816705167293549,
+      "mean_token_accuracy": 0.9179367125034332,
+      "num_tokens": 10794000.0,
+      "step": 875
+    },
+    {
+      "entropy": 1.2641226053237915,
+      "epoch": 0.46129541864139023,
+      "grad_norm": 0.31645506620407104,
+      "learning_rate": 0.0001993954057833051,
+      "loss": 0.22259601950645447,
+      "mean_token_accuracy": 0.9116221219301224,
+      "num_tokens": 10806336.0,
+      "step": 876
+    },
+    {
+      "entropy": 1.2864770293235779,
+      "epoch": 0.46182201158504477,
+      "grad_norm": 0.36219751834869385,
+      "learning_rate": 0.0001993928656660631,
+      "loss": 0.23597069084644318,
+      "mean_token_accuracy": 0.9075523316860199,
+      "num_tokens": 10818672.0,
+      "step": 877
+    },
+    {
+      "entropy": 1.284227877855301,
+      "epoch": 0.4623486045286993,
+      "grad_norm": 0.33684641122817993,
+      "learning_rate": 0.00019939032024208762,
+      "loss": 0.23688027262687683,
+      "mean_token_accuracy": 0.90504090487957,
+      "num_tokens": 10831008.0,
+      "step": 878
+    },
+    {
+      "entropy": 1.2863526940345764,
+      "epoch": 0.4628751974723539,
+      "grad_norm": 0.3741551637649536,
+      "learning_rate": 0.0001993877695115299,
+      "loss": 0.26197415590286255,
+      "mean_token_accuracy": 0.8967071026563644,
+      "num_tokens": 10843344.0,
+      "step": 879
+    },
+    {
+      "entropy": 1.3029287159442902,
+      "epoch": 0.46340179041600843,
+      "grad_norm": 0.33679425716400146,
+      "learning_rate": 0.00019938521347454127,
+      "loss": 0.2334819734096527,
+      "mean_token_accuracy": 0.9123462438583374,
+      "num_tokens": 10855680.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.2353507280349731,
+      "epoch": 0.46392838335966297,
+      "grad_norm": 0.3251015543937683,
+      "learning_rate": 0.00019938265213127344,
+      "loss": 0.24127048254013062,
+      "mean_token_accuracy": 0.9032625257968903,
+      "num_tokens": 10868016.0,
+      "step": 881
+    },
+    {
+      "entropy": 1.2318426966667175,
+      "epoch": 0.46445497630331756,
+      "grad_norm": 0.3128981292247772,
+      "learning_rate": 0.0001993800854818786,
+      "loss": 0.23091557621955872,
+      "mean_token_accuracy": 0.910790741443634,
+      "num_tokens": 10880352.0,
+      "step": 882
+    },
+    {
+      "entropy": 1.1969189643859863,
+      "epoch": 0.4649815692469721,
+      "grad_norm": 0.2763598561286926,
+      "learning_rate": 0.000199377513526509,
+      "loss": 0.19723600149154663,
+      "mean_token_accuracy": 0.9164901375770569,
+      "num_tokens": 10892688.0,
+      "step": 883
+    },
+    {
+      "entropy": 1.3742352426052094,
+      "epoch": 0.46550816219062663,
+      "grad_norm": 0.36182069778442383,
+      "learning_rate": 0.00019937493626531733,
+      "loss": 0.23250778019428253,
+      "mean_token_accuracy": 0.910800889134407,
+      "num_tokens": 10905024.0,
+      "step": 884
+    },
+    {
+      "entropy": 1.287173181772232,
+      "epoch": 0.4660347551342812,
+      "grad_norm": 0.37185028195381165,
+      "learning_rate": 0.00019937235369845666,
+      "loss": 0.2612641453742981,
+      "mean_token_accuracy": 0.8980638682842255,
+      "num_tokens": 10917360.0,
+      "step": 885
+    },
+    {
+      "entropy": 1.256176471710205,
+      "epoch": 0.46656134807793576,
+      "grad_norm": 0.3193638026714325,
+      "learning_rate": 0.00019936976582608023,
+      "loss": 0.2168237864971161,
+      "mean_token_accuracy": 0.9117200374603271,
+      "num_tokens": 10929696.0,
+      "step": 886
+    },
+    {
+      "entropy": 1.3229191601276398,
+      "epoch": 0.4670879410215903,
+      "grad_norm": 0.3389144241809845,
+      "learning_rate": 0.00019936717264834166,
+      "loss": 0.22236081957817078,
+      "mean_token_accuracy": 0.9158815294504166,
+      "num_tokens": 10942032.0,
+      "step": 887
+    },
+    {
+      "entropy": 1.3444121778011322,
+      "epoch": 0.4676145339652449,
+      "grad_norm": 0.30865800380706787,
+      "learning_rate": 0.00019936457416539497,
+      "loss": 0.22498130798339844,
+      "mean_token_accuracy": 0.9109358042478561,
+      "num_tokens": 10954368.0,
+      "step": 888
+    },
+    {
+      "entropy": 1.3391337096691132,
+      "epoch": 0.4681411269088994,
+      "grad_norm": 0.34072619676589966,
+      "learning_rate": 0.00019936197037739433,
+      "loss": 0.25583821535110474,
+      "mean_token_accuracy": 0.9030615091323853,
+      "num_tokens": 10966704.0,
+      "step": 889
+    },
+    {
+      "entropy": 1.3236074447631836,
+      "epoch": 0.46866771985255395,
+      "grad_norm": 0.30708205699920654,
+      "learning_rate": 0.00019935936128449437,
+      "loss": 0.2282828986644745,
+      "mean_token_accuracy": 0.9107722789049149,
+      "num_tokens": 10979040.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.3548866510391235,
+      "epoch": 0.46919431279620855,
+      "grad_norm": 0.3132648766040802,
+      "learning_rate": 0.00019935674688684993,
+      "loss": 0.21266606450080872,
+      "mean_token_accuracy": 0.9191526621580124,
+      "num_tokens": 10991376.0,
+      "step": 891
+    },
+    {
+      "entropy": 1.3176174461841583,
+      "epoch": 0.4697209057398631,
+      "grad_norm": 0.2955908477306366,
+      "learning_rate": 0.00019935412718461625,
+      "loss": 0.21996596455574036,
+      "mean_token_accuracy": 0.9133886098861694,
+      "num_tokens": 11003712.0,
+      "step": 892
+    },
+    {
+      "entropy": 1.3978124856948853,
+      "epoch": 0.4702474986835176,
+      "grad_norm": 0.29386526346206665,
+      "learning_rate": 0.00019935150217794887,
+      "loss": 0.21586540341377258,
+      "mean_token_accuracy": 0.913035660982132,
+      "num_tokens": 11016048.0,
+      "step": 893
+    },
+    {
+      "entropy": 1.3344171047210693,
+      "epoch": 0.4707740916271722,
+      "grad_norm": 0.307434618473053,
+      "learning_rate": 0.00019934887186700352,
+      "loss": 0.19684919714927673,
+      "mean_token_accuracy": 0.9249100238084793,
+      "num_tokens": 11028384.0,
+      "step": 894
+    },
+    {
+      "entropy": 1.2738686203956604,
+      "epoch": 0.47130068457082674,
+      "grad_norm": 0.30784371495246887,
+      "learning_rate": 0.00019934623625193646,
+      "loss": 0.23357078433036804,
+      "mean_token_accuracy": 0.906443789601326,
+      "num_tokens": 11040720.0,
+      "step": 895
+    },
+    {
+      "entropy": 1.3004032373428345,
+      "epoch": 0.4718272775144813,
+      "grad_norm": 0.3604377806186676,
+      "learning_rate": 0.0001993435953329041,
+      "loss": 0.23538318276405334,
+      "mean_token_accuracy": 0.910987451672554,
+      "num_tokens": 11053056.0,
+      "step": 896
+    },
+    {
+      "entropy": 1.349770426750183,
+      "epoch": 0.47235387045813587,
+      "grad_norm": 0.371415913105011,
+      "learning_rate": 0.00019934094911006322,
+      "loss": 0.23594705760478973,
+      "mean_token_accuracy": 0.9082933962345123,
+      "num_tokens": 11065392.0,
+      "step": 897
+    },
+    {
+      "entropy": 1.3280587792396545,
+      "epoch": 0.4728804634017904,
+      "grad_norm": 0.3518844544887543,
+      "learning_rate": 0.0001993382975835709,
+      "loss": 0.22077634930610657,
+      "mean_token_accuracy": 0.9151550680398941,
+      "num_tokens": 11077728.0,
+      "step": 898
+    },
+    {
+      "entropy": 1.2467398941516876,
+      "epoch": 0.473407056345445,
+      "grad_norm": 0.3485780358314514,
+      "learning_rate": 0.00019933564075358455,
+      "loss": 0.21681727468967438,
+      "mean_token_accuracy": 0.917348250746727,
+      "num_tokens": 11090064.0,
+      "step": 899
+    },
+    {
+      "entropy": 1.2497498095035553,
+      "epoch": 0.47393364928909953,
+      "grad_norm": 0.2795222997665405,
+      "learning_rate": 0.0001993329786202619,
+      "loss": 0.19592691957950592,
+      "mean_token_accuracy": 0.92547307908535,
+      "num_tokens": 11102400.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.3175421953201294,
+      "epoch": 0.47446024223275407,
+      "grad_norm": 0.34323444962501526,
+      "learning_rate": 0.00019933031118376097,
+      "loss": 0.23341751098632812,
+      "mean_token_accuracy": 0.9069218635559082,
+      "num_tokens": 11114736.0,
+      "step": 901
+    },
+    {
+      "entropy": 1.2457618713378906,
+      "epoch": 0.47498683517640866,
+      "grad_norm": 0.3164939284324646,
+      "learning_rate": 0.00019932763844424017,
+      "loss": 0.2065175175666809,
+      "mean_token_accuracy": 0.9177182614803314,
+      "num_tokens": 11127072.0,
+      "step": 902
+    },
+    {
+      "entropy": 1.2637942731380463,
+      "epoch": 0.4755134281200632,
+      "grad_norm": 0.3504999279975891,
+      "learning_rate": 0.00019932496040185808,
+      "loss": 0.21458998322486877,
+      "mean_token_accuracy": 0.9221720099449158,
+      "num_tokens": 11139408.0,
+      "step": 903
+    },
+    {
+      "entropy": 1.2935825884342194,
+      "epoch": 0.47604002106371773,
+      "grad_norm": 0.37840691208839417,
+      "learning_rate": 0.00019932227705677372,
+      "loss": 0.227766752243042,
+      "mean_token_accuracy": 0.9137209206819534,
+      "num_tokens": 11151744.0,
+      "step": 904
+    },
+    {
+      "entropy": 1.3187288045883179,
+      "epoch": 0.4765666140073723,
+      "grad_norm": 0.35070836544036865,
+      "learning_rate": 0.00019931958840914635,
+      "loss": 0.23313553631305695,
+      "mean_token_accuracy": 0.9072595983743668,
+      "num_tokens": 11164080.0,
+      "step": 905
+    },
+    {
+      "entropy": 1.2109759449958801,
+      "epoch": 0.47709320695102686,
+      "grad_norm": 0.3268450200557709,
+      "learning_rate": 0.00019931689445913564,
+      "loss": 0.21801263093948364,
+      "mean_token_accuracy": 0.9163070619106293,
+      "num_tokens": 11176416.0,
+      "step": 906
+    },
+    {
+      "entropy": 1.2274765968322754,
+      "epoch": 0.4776197998946814,
+      "grad_norm": 0.3167557120323181,
+      "learning_rate": 0.00019931419520690147,
+      "loss": 0.21107861399650574,
+      "mean_token_accuracy": 0.9160606116056442,
+      "num_tokens": 11188752.0,
+      "step": 907
+    },
+    {
+      "entropy": 1.1903901398181915,
+      "epoch": 0.478146392838336,
+      "grad_norm": 0.3267144560813904,
+      "learning_rate": 0.00019931149065260413,
+      "loss": 0.20752331614494324,
+      "mean_token_accuracy": 0.9146323353052139,
+      "num_tokens": 11201088.0,
+      "step": 908
+    },
+    {
+      "entropy": 1.2301645576953888,
+      "epoch": 0.4786729857819905,
+      "grad_norm": 0.3323666751384735,
+      "learning_rate": 0.0001993087807964041,
+      "loss": 0.20132845640182495,
+      "mean_token_accuracy": 0.9229601621627808,
+      "num_tokens": 11213424.0,
+      "step": 909
+    },
+    {
+      "entropy": 1.2413881123065948,
+      "epoch": 0.47919957872564506,
+      "grad_norm": 0.3245013356208801,
+      "learning_rate": 0.00019930606563846234,
+      "loss": 0.19781675934791565,
+      "mean_token_accuracy": 0.9200431853532791,
+      "num_tokens": 11225760.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.180581659078598,
+      "epoch": 0.47972617166929965,
+      "grad_norm": 0.31535807251930237,
+      "learning_rate": 0.00019930334517893991,
+      "loss": 0.2060268521308899,
+      "mean_token_accuracy": 0.9165100157260895,
+      "num_tokens": 11238096.0,
+      "step": 911
+    },
+    {
+      "entropy": 1.233038753271103,
+      "epoch": 0.4802527646129542,
+      "grad_norm": 0.32888907194137573,
+      "learning_rate": 0.0001993006194179984,
+      "loss": 0.22726945579051971,
+      "mean_token_accuracy": 0.9112953692674637,
+      "num_tokens": 11250432.0,
+      "step": 912
+    },
+    {
+      "entropy": 1.262090653181076,
+      "epoch": 0.4807793575566087,
+      "grad_norm": 0.3833845555782318,
+      "learning_rate": 0.00019929788835579962,
+      "loss": 0.23385398089885712,
+      "mean_token_accuracy": 0.9076380431652069,
+      "num_tokens": 11262768.0,
+      "step": 913
+    },
+    {
+      "entropy": 1.2252440452575684,
+      "epoch": 0.4813059505002633,
+      "grad_norm": 0.3054254949092865,
+      "learning_rate": 0.00019929515199250565,
+      "loss": 0.21175538003444672,
+      "mean_token_accuracy": 0.91593237221241,
+      "num_tokens": 11275104.0,
+      "step": 914
+    },
+    {
+      "entropy": 1.2347098588943481,
+      "epoch": 0.48183254344391785,
+      "grad_norm": 0.32731181383132935,
+      "learning_rate": 0.00019929241032827898,
+      "loss": 0.21469083428382874,
+      "mean_token_accuracy": 0.9144894182682037,
+      "num_tokens": 11287440.0,
+      "step": 915
+    },
+    {
+      "entropy": 1.2054485082626343,
+      "epoch": 0.4823591363875724,
+      "grad_norm": 0.3010466694831848,
+      "learning_rate": 0.0001992896633632823,
+      "loss": 0.2065753936767578,
+      "mean_token_accuracy": 0.9242510497570038,
+      "num_tokens": 11299776.0,
+      "step": 916
+    },
+    {
+      "entropy": 1.2271296083927155,
+      "epoch": 0.482885729331227,
+      "grad_norm": 0.42679160833358765,
+      "learning_rate": 0.00019928691109767876,
+      "loss": 0.23648393154144287,
+      "mean_token_accuracy": 0.9071210771799088,
+      "num_tokens": 11312112.0,
+      "step": 917
+    },
+    {
+      "entropy": 1.233128011226654,
+      "epoch": 0.4834123222748815,
+      "grad_norm": 0.3018331229686737,
+      "learning_rate": 0.00019928415353163173,
+      "loss": 0.20864740014076233,
+      "mean_token_accuracy": 0.9222661405801773,
+      "num_tokens": 11324448.0,
+      "step": 918
+    },
+    {
+      "entropy": 1.2823042571544647,
+      "epoch": 0.48393891521853605,
+      "grad_norm": 0.44017431139945984,
+      "learning_rate": 0.00019928139066530487,
+      "loss": 0.23257723450660706,
+      "mean_token_accuracy": 0.9054065048694611,
+      "num_tokens": 11336784.0,
+      "step": 919
+    },
+    {
+      "entropy": 1.2760326564311981,
+      "epoch": 0.48446550816219064,
+      "grad_norm": 0.3077127933502197,
+      "learning_rate": 0.00019927862249886222,
+      "loss": 0.22436192631721497,
+      "mean_token_accuracy": 0.9136722385883331,
+      "num_tokens": 11349120.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.3157328069210052,
+      "epoch": 0.4849921011058452,
+      "grad_norm": 0.29968273639678955,
+      "learning_rate": 0.0001992758490324681,
+      "loss": 0.20738902688026428,
+      "mean_token_accuracy": 0.9168381989002228,
+      "num_tokens": 11361456.0,
+      "step": 921
+    },
+    {
+      "entropy": 1.3066685795783997,
+      "epoch": 0.4855186940494997,
+      "grad_norm": 0.31207534670829773,
+      "learning_rate": 0.00019927307026628715,
+      "loss": 0.2138848602771759,
+      "mean_token_accuracy": 0.9126945734024048,
+      "num_tokens": 11373792.0,
+      "step": 922
+    },
+    {
+      "entropy": 1.3030335009098053,
+      "epoch": 0.4860452869931543,
+      "grad_norm": 0.3280918598175049,
+      "learning_rate": 0.00019927028620048438,
+      "loss": 0.23367798328399658,
+      "mean_token_accuracy": 0.9077742546796799,
+      "num_tokens": 11386128.0,
+      "step": 923
+    },
+    {
+      "entropy": 1.2811558842658997,
+      "epoch": 0.48657187993680884,
+      "grad_norm": 0.376886248588562,
+      "learning_rate": 0.00019926749683522502,
+      "loss": 0.20860403776168823,
+      "mean_token_accuracy": 0.9099331349134445,
+      "num_tokens": 11398464.0,
+      "step": 924
+    },
+    {
+      "entropy": 1.2956900000572205,
+      "epoch": 0.48709847288046343,
+      "grad_norm": 0.2849150598049164,
+      "learning_rate": 0.00019926470217067463,
+      "loss": 0.20377135276794434,
+      "mean_token_accuracy": 0.9249068945646286,
+      "num_tokens": 11410800.0,
+      "step": 925
+    },
+    {
+      "entropy": 1.3335639834403992,
+      "epoch": 0.48762506582411796,
+      "grad_norm": 0.30451253056526184,
+      "learning_rate": 0.00019926190220699916,
+      "loss": 0.2137700468301773,
+      "mean_token_accuracy": 0.9165246337652206,
+      "num_tokens": 11423136.0,
+      "step": 926
+    },
+    {
+      "entropy": 1.3315201103687286,
+      "epoch": 0.4881516587677725,
+      "grad_norm": 0.31507113575935364,
+      "learning_rate": 0.0001992590969443648,
+      "loss": 0.22226491570472717,
+      "mean_token_accuracy": 0.9110458046197891,
+      "num_tokens": 11435472.0,
+      "step": 927
+    },
+    {
+      "entropy": 1.3439573049545288,
+      "epoch": 0.4886782517114271,
+      "grad_norm": 0.3123377561569214,
+      "learning_rate": 0.00019925628638293815,
+      "loss": 0.2124030739068985,
+      "mean_token_accuracy": 0.9143670797348022,
+      "num_tokens": 11447808.0,
+      "step": 928
+    },
+    {
+      "entropy": 1.373394101858139,
+      "epoch": 0.4892048446550816,
+      "grad_norm": 0.38396134972572327,
+      "learning_rate": 0.00019925347052288594,
+      "loss": 0.21086710691452026,
+      "mean_token_accuracy": 0.9174065738916397,
+      "num_tokens": 11460144.0,
+      "step": 929
+    },
+    {
+      "entropy": 1.3038064241409302,
+      "epoch": 0.48973143759873616,
+      "grad_norm": 0.3282279074192047,
+      "learning_rate": 0.00019925064936437544,
+      "loss": 0.19842374324798584,
+      "mean_token_accuracy": 0.9203700721263885,
+      "num_tokens": 11472480.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.39743971824646,
+      "epoch": 0.49025803054239075,
+      "grad_norm": 0.3458907902240753,
+      "learning_rate": 0.00019924782290757403,
+      "loss": 0.21870574355125427,
+      "mean_token_accuracy": 0.9133100211620331,
+      "num_tokens": 11484816.0,
+      "step": 931
+    },
+    {
+      "entropy": 1.3737529814243317,
+      "epoch": 0.4907846234860453,
+      "grad_norm": 0.35361558198928833,
+      "learning_rate": 0.0001992449911526496,
+      "loss": 0.2228592187166214,
+      "mean_token_accuracy": 0.908146470785141,
+      "num_tokens": 11497152.0,
+      "step": 932
+    },
+    {
+      "entropy": 1.373944640159607,
+      "epoch": 0.4913112164296998,
+      "grad_norm": 0.3263792097568512,
+      "learning_rate": 0.0001992421540997702,
+      "loss": 0.19889512658119202,
+      "mean_token_accuracy": 0.9241671562194824,
+      "num_tokens": 11509488.0,
+      "step": 933
+    },
+    {
+      "entropy": 1.3812199532985687,
+      "epoch": 0.4918378093733544,
+      "grad_norm": 0.30469757318496704,
+      "learning_rate": 0.00019923931174910421,
+      "loss": 0.20764592289924622,
+      "mean_token_accuracy": 0.9166074395179749,
+      "num_tokens": 11521824.0,
+      "step": 934
+    },
+    {
+      "entropy": 1.3702083230018616,
+      "epoch": 0.49236440231700895,
+      "grad_norm": 0.3094841539859772,
+      "learning_rate": 0.00019923646410082045,
+      "loss": 0.22583043575286865,
+      "mean_token_accuracy": 0.9111629873514175,
+      "num_tokens": 11534160.0,
+      "step": 935
+    },
+    {
+      "entropy": 1.4550175070762634,
+      "epoch": 0.4928909952606635,
+      "grad_norm": 0.3479791581630707,
+      "learning_rate": 0.0001992336111550879,
+      "loss": 0.22404024004936218,
+      "mean_token_accuracy": 0.9125335812568665,
+      "num_tokens": 11546496.0,
+      "step": 936
+    },
+    {
+      "entropy": 1.369890809059143,
+      "epoch": 0.4934175882043181,
+      "grad_norm": 0.33549702167510986,
+      "learning_rate": 0.00019923075291207595,
+      "loss": 0.22361691296100616,
+      "mean_token_accuracy": 0.9108781665563583,
+      "num_tokens": 11558832.0,
+      "step": 937
+    },
+    {
+      "entropy": 1.4506123065948486,
+      "epoch": 0.4939441811479726,
+      "grad_norm": 0.31795117259025574,
+      "learning_rate": 0.00019922788937195432,
+      "loss": 0.20908354222774506,
+      "mean_token_accuracy": 0.916689082980156,
+      "num_tokens": 11571168.0,
+      "step": 938
+    },
+    {
+      "entropy": 1.4594894349575043,
+      "epoch": 0.49447077409162715,
+      "grad_norm": 0.3958785831928253,
+      "learning_rate": 0.0001992250205348929,
+      "loss": 0.22291752696037292,
+      "mean_token_accuracy": 0.9057322442531586,
+      "num_tokens": 11583504.0,
+      "step": 939
+    },
+    {
+      "entropy": 1.4367193281650543,
+      "epoch": 0.49499736703528174,
+      "grad_norm": 0.38530445098876953,
+      "learning_rate": 0.00019922214640106207,
+      "loss": 0.21757692098617554,
+      "mean_token_accuracy": 0.9140817075967789,
+      "num_tokens": 11595840.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.4827219545841217,
+      "epoch": 0.4955239599789363,
+      "grad_norm": 0.36136746406555176,
+      "learning_rate": 0.00019921926697063244,
+      "loss": 0.245650514960289,
+      "mean_token_accuracy": 0.9077811390161514,
+      "num_tokens": 11608176.0,
+      "step": 941
+    },
+    {
+      "entropy": 1.4725195467472076,
+      "epoch": 0.4960505529225908,
+      "grad_norm": 0.3393801748752594,
+      "learning_rate": 0.00019921638224377493,
+      "loss": 0.22296787798404694,
+      "mean_token_accuracy": 0.912161722779274,
+      "num_tokens": 11620512.0,
+      "step": 942
+    },
+    {
+      "entropy": 1.5998758971691132,
+      "epoch": 0.4965771458662454,
+      "grad_norm": 0.31578320264816284,
+      "learning_rate": 0.00019921349222066083,
+      "loss": 0.21946173906326294,
+      "mean_token_accuracy": 0.9090907126665115,
+      "num_tokens": 11632848.0,
+      "step": 943
+    },
+    {
+      "entropy": 1.5493124723434448,
+      "epoch": 0.49710373880989994,
+      "grad_norm": 0.3594050705432892,
+      "learning_rate": 0.00019921059690146165,
+      "loss": 0.25090494751930237,
+      "mean_token_accuracy": 0.899459958076477,
+      "num_tokens": 11645184.0,
+      "step": 944
+    },
+    {
+      "entropy": 1.4591304063796997,
+      "epoch": 0.4976303317535545,
+      "grad_norm": 0.3398474454879761,
+      "learning_rate": 0.0001992076962863493,
+      "loss": 0.22868260741233826,
+      "mean_token_accuracy": 0.9120910316705704,
+      "num_tokens": 11657520.0,
+      "step": 945
+    },
+    {
+      "entropy": 1.5574268996715546,
+      "epoch": 0.49815692469720907,
+      "grad_norm": 0.37178367376327515,
+      "learning_rate": 0.00019920479037549595,
+      "loss": 0.24399921298027039,
+      "mean_token_accuracy": 0.9063628762960434,
+      "num_tokens": 11669856.0,
+      "step": 946
+    },
+    {
+      "entropy": 1.4919456243515015,
+      "epoch": 0.4986835176408636,
+      "grad_norm": 0.344330757856369,
+      "learning_rate": 0.0001992018791690741,
+      "loss": 0.2234281301498413,
+      "mean_token_accuracy": 0.9139427691698074,
+      "num_tokens": 11682192.0,
+      "step": 947
+    },
+    {
+      "entropy": 1.486481934785843,
+      "epoch": 0.49921011058451814,
+      "grad_norm": 0.3283190429210663,
+      "learning_rate": 0.00019919896266725663,
+      "loss": 0.21686026453971863,
+      "mean_token_accuracy": 0.9156758487224579,
+      "num_tokens": 11694528.0,
+      "step": 948
+    },
+    {
+      "entropy": 1.4495719075202942,
+      "epoch": 0.49973670352817273,
+      "grad_norm": 0.3260519802570343,
+      "learning_rate": 0.00019919604087021664,
+      "loss": 0.21722926199436188,
+      "mean_token_accuracy": 0.9189446866512299,
+      "num_tokens": 11706864.0,
+      "step": 949
+    },
+    {
+      "entropy": 1.5058691203594208,
+      "epoch": 0.5002632964718273,
+      "grad_norm": 0.34286144375801086,
+      "learning_rate": 0.00019919311377812756,
+      "loss": 0.2165110558271408,
+      "mean_token_accuracy": 0.9119018763303757,
+      "num_tokens": 11719200.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.4864198863506317,
+      "epoch": 0.5007898894154819,
+      "grad_norm": 0.3339075446128845,
+      "learning_rate": 0.00019919018139116318,
+      "loss": 0.22437581419944763,
+      "mean_token_accuracy": 0.9123266935348511,
+      "num_tokens": 11731536.0,
+      "step": 951
+    },
+    {
+      "entropy": 1.5294865667819977,
+      "epoch": 0.5013164823591364,
+      "grad_norm": 0.3746489882469177,
+      "learning_rate": 0.00019918724370949754,
+      "loss": 0.2426229566335678,
+      "mean_token_accuracy": 0.9112533032894135,
+      "num_tokens": 11743872.0,
+      "step": 952
+    },
+    {
+      "entropy": 1.4867128133773804,
+      "epoch": 0.5018430753027909,
+      "grad_norm": 0.3308849036693573,
+      "learning_rate": 0.00019918430073330513,
+      "loss": 0.2087199091911316,
+      "mean_token_accuracy": 0.919797733426094,
+      "num_tokens": 11756208.0,
+      "step": 953
+    },
+    {
+      "entropy": 1.5136871337890625,
+      "epoch": 0.5023696682464455,
+      "grad_norm": 0.34608909487724304,
+      "learning_rate": 0.00019918135246276052,
+      "loss": 0.22718098759651184,
+      "mean_token_accuracy": 0.9114474952220917,
+      "num_tokens": 11768544.0,
+      "step": 954
+    },
+    {
+      "entropy": 1.5160588324069977,
+      "epoch": 0.5028962611901,
+      "grad_norm": 0.3665941655635834,
+      "learning_rate": 0.00019917839889803884,
+      "loss": 0.21780604124069214,
+      "mean_token_accuracy": 0.9177397042512894,
+      "num_tokens": 11780880.0,
+      "step": 955
+    },
+    {
+      "entropy": 1.4711733162403107,
+      "epoch": 0.5034228541337546,
+      "grad_norm": 0.3232526183128357,
+      "learning_rate": 0.00019917544003931537,
+      "loss": 0.20783714950084686,
+      "mean_token_accuracy": 0.9170488715171814,
+      "num_tokens": 11793216.0,
+      "step": 956
+    },
+    {
+      "entropy": 1.5061227977275848,
+      "epoch": 0.5039494470774092,
+      "grad_norm": 0.3460804522037506,
+      "learning_rate": 0.00019917247588676582,
+      "loss": 0.22913217544555664,
+      "mean_token_accuracy": 0.9122664779424667,
+      "num_tokens": 11805552.0,
+      "step": 957
+    },
+    {
+      "entropy": 1.4881491363048553,
+      "epoch": 0.5044760400210637,
+      "grad_norm": 0.38591158390045166,
+      "learning_rate": 0.00019916950644056607,
+      "loss": 0.2197885662317276,
+      "mean_token_accuracy": 0.9077084809541702,
+      "num_tokens": 11817888.0,
+      "step": 958
+    },
+    {
+      "entropy": 1.4673264026641846,
+      "epoch": 0.5050026329647183,
+      "grad_norm": 0.33065465092658997,
+      "learning_rate": 0.00019916653170089246,
+      "loss": 0.2319866120815277,
+      "mean_token_accuracy": 0.9126220494508743,
+      "num_tokens": 11830224.0,
+      "step": 959
+    },
+    {
+      "entropy": 1.4641340970993042,
+      "epoch": 0.5055292259083728,
+      "grad_norm": 0.36219480633735657,
+      "learning_rate": 0.00019916355166792155,
+      "loss": 0.23807652294635773,
+      "mean_token_accuracy": 0.9052844792604446,
+      "num_tokens": 11842560.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.4770829379558563,
+      "epoch": 0.5060558188520273,
+      "grad_norm": 0.3402878940105438,
+      "learning_rate": 0.00019916056634183027,
+      "loss": 0.23622629046440125,
+      "mean_token_accuracy": 0.9105163961648941,
+      "num_tokens": 11854896.0,
+      "step": 961
+    },
+    {
+      "entropy": 1.510101705789566,
+      "epoch": 0.506582411795682,
+      "grad_norm": 0.38479122519493103,
+      "learning_rate": 0.00019915757572279584,
+      "loss": 0.26114726066589355,
+      "mean_token_accuracy": 0.901023805141449,
+      "num_tokens": 11867232.0,
+      "step": 962
+    },
+    {
+      "entropy": 1.462478369474411,
+      "epoch": 0.5071090047393365,
+      "grad_norm": 0.30232900381088257,
+      "learning_rate": 0.0001991545798109958,
+      "loss": 0.21362073719501495,
+      "mean_token_accuracy": 0.9150588363409042,
+      "num_tokens": 11879568.0,
+      "step": 963
+    },
+    {
+      "entropy": 1.4567637145519257,
+      "epoch": 0.507635597682991,
+      "grad_norm": 0.3392339050769806,
+      "learning_rate": 0.00019915157860660797,
+      "loss": 0.229543536901474,
+      "mean_token_accuracy": 0.9123556017875671,
+      "num_tokens": 11891904.0,
+      "step": 964
+    },
+    {
+      "entropy": 1.4373153448104858,
+      "epoch": 0.5081621906266456,
+      "grad_norm": 0.3237869441509247,
+      "learning_rate": 0.00019914857210981052,
+      "loss": 0.222116619348526,
+      "mean_token_accuracy": 0.9098070710897446,
+      "num_tokens": 11904240.0,
+      "step": 965
+    },
+    {
+      "entropy": 1.4922161400318146,
+      "epoch": 0.5086887835703001,
+      "grad_norm": 0.3090606927871704,
+      "learning_rate": 0.00019914556032078198,
+      "loss": 0.22452594339847565,
+      "mean_token_accuracy": 0.9133640974760056,
+      "num_tokens": 11916576.0,
+      "step": 966
+    },
+    {
+      "entropy": 1.4847496449947357,
+      "epoch": 0.5092153765139548,
+      "grad_norm": 0.31530138850212097,
+      "learning_rate": 0.00019914254323970108,
+      "loss": 0.21454353630542755,
+      "mean_token_accuracy": 0.9175532162189484,
+      "num_tokens": 11928912.0,
+      "step": 967
+    },
+    {
+      "entropy": 1.481870412826538,
+      "epoch": 0.5097419694576093,
+      "grad_norm": 0.3098122775554657,
+      "learning_rate": 0.00019913952086674696,
+      "loss": 0.21168866753578186,
+      "mean_token_accuracy": 0.9122080057859421,
+      "num_tokens": 11941248.0,
+      "step": 968
+    },
+    {
+      "entropy": 1.5705283880233765,
+      "epoch": 0.5102685624012638,
+      "grad_norm": 0.32009753584861755,
+      "learning_rate": 0.000199136493202099,
+      "loss": 0.20256565511226654,
+      "mean_token_accuracy": 0.9237568825483322,
+      "num_tokens": 11953584.0,
+      "step": 969
+    },
+    {
+      "entropy": 1.5321417450904846,
+      "epoch": 0.5107951553449184,
+      "grad_norm": 0.33614546060562134,
+      "learning_rate": 0.000199133460245937,
+      "loss": 0.2249448001384735,
+      "mean_token_accuracy": 0.9105825424194336,
+      "num_tokens": 11965920.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.5060743689537048,
+      "epoch": 0.5113217482885729,
+      "grad_norm": 0.3201083540916443,
+      "learning_rate": 0.00019913042199844097,
+      "loss": 0.22582277655601501,
+      "mean_token_accuracy": 0.9086000770330429,
+      "num_tokens": 11978256.0,
+      "step": 971
+    },
+    {
+      "entropy": 1.5098620057106018,
+      "epoch": 0.5118483412322274,
+      "grad_norm": 0.739696204662323,
+      "learning_rate": 0.00019912737845979126,
+      "loss": 0.20600268244743347,
+      "mean_token_accuracy": 0.9227887839078903,
+      "num_tokens": 11990592.0,
+      "step": 972
+    },
+    {
+      "entropy": 1.5500882267951965,
+      "epoch": 0.5123749341758821,
+      "grad_norm": 0.3458639681339264,
+      "learning_rate": 0.00019912432963016855,
+      "loss": 0.23240959644317627,
+      "mean_token_accuracy": 0.9070769846439362,
+      "num_tokens": 12002928.0,
+      "step": 973
+    },
+    {
+      "entropy": 1.5514563918113708,
+      "epoch": 0.5129015271195366,
+      "grad_norm": 0.33823850750923157,
+      "learning_rate": 0.0001991212755097539,
+      "loss": 0.21927812695503235,
+      "mean_token_accuracy": 0.9085335284471512,
+      "num_tokens": 12015264.0,
+      "step": 974
+    },
+    {
+      "entropy": 1.4905177056789398,
+      "epoch": 0.5134281200631912,
+      "grad_norm": 0.3308093547821045,
+      "learning_rate": 0.0001991182160987285,
+      "loss": 0.22249096632003784,
+      "mean_token_accuracy": 0.9114507138729095,
+      "num_tokens": 12027600.0,
+      "step": 975
+    },
+    {
+      "entropy": 1.5092821419239044,
+      "epoch": 0.5139547130068457,
+      "grad_norm": 0.35571059584617615,
+      "learning_rate": 0.0001991151513972741,
+      "loss": 0.23242917656898499,
+      "mean_token_accuracy": 0.9118809998035431,
+      "num_tokens": 12039936.0,
+      "step": 976
+    },
+    {
+      "entropy": 1.426015704870224,
+      "epoch": 0.5144813059505002,
+      "grad_norm": 0.2942444086074829,
+      "learning_rate": 0.0001991120814055725,
+      "loss": 0.2060171216726303,
+      "mean_token_accuracy": 0.9149444848299026,
+      "num_tokens": 12052272.0,
+      "step": 977
+    },
+    {
+      "entropy": 1.4646040201187134,
+      "epoch": 0.5150078988941548,
+      "grad_norm": 0.36998897790908813,
+      "learning_rate": 0.00019910900612380605,
+      "loss": 0.23340070247650146,
+      "mean_token_accuracy": 0.9063846617937088,
+      "num_tokens": 12064608.0,
+      "step": 978
+    },
+    {
+      "entropy": 1.4659819900989532,
+      "epoch": 0.5155344918378094,
+      "grad_norm": 0.3255058526992798,
+      "learning_rate": 0.00019910592555215725,
+      "loss": 0.20911140739917755,
+      "mean_token_accuracy": 0.9192948192358017,
+      "num_tokens": 12076944.0,
+      "step": 979
+    },
+    {
+      "entropy": 1.5005815625190735,
+      "epoch": 0.516061084781464,
+      "grad_norm": 0.35735374689102173,
+      "learning_rate": 0.000199102839690809,
+      "loss": 0.2287176251411438,
+      "mean_token_accuracy": 0.9081791937351227,
+      "num_tokens": 12089280.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.4197439849376678,
+      "epoch": 0.5165876777251185,
+      "grad_norm": 0.30643951892852783,
+      "learning_rate": 0.00019909974853994452,
+      "loss": 0.19968025386333466,
+      "mean_token_accuracy": 0.9234030842781067,
+      "num_tokens": 12101616.0,
+      "step": 981
+    },
+    {
+      "entropy": 1.4192424714565277,
+      "epoch": 0.517114270668773,
+      "grad_norm": 0.337433785200119,
+      "learning_rate": 0.00019909665209974723,
+      "loss": 0.23955072462558746,
+      "mean_token_accuracy": 0.9109057635068893,
+      "num_tokens": 12113952.0,
+      "step": 982
+    },
+    {
+      "entropy": 1.4805451035499573,
+      "epoch": 0.5176408636124276,
+      "grad_norm": 0.34595102071762085,
+      "learning_rate": 0.00019909355037040104,
+      "loss": 0.23975268006324768,
+      "mean_token_accuracy": 0.9099203199148178,
+      "num_tokens": 12126288.0,
+      "step": 983
+    },
+    {
+      "entropy": 1.4859175980091095,
+      "epoch": 0.5181674565560821,
+      "grad_norm": 0.33112627267837524,
+      "learning_rate": 0.00019909044335209003,
+      "loss": 0.23115341365337372,
+      "mean_token_accuracy": 0.9144714772701263,
+      "num_tokens": 12138624.0,
+      "step": 984
+    },
+    {
+      "entropy": 1.4870480000972748,
+      "epoch": 0.5186940494997367,
+      "grad_norm": 0.2882140576839447,
+      "learning_rate": 0.00019908733104499868,
+      "loss": 0.2003549188375473,
+      "mean_token_accuracy": 0.9193922579288483,
+      "num_tokens": 12150960.0,
+      "step": 985
+    },
+    {
+      "entropy": 1.5195731818675995,
+      "epoch": 0.5192206424433913,
+      "grad_norm": 0.33627068996429443,
+      "learning_rate": 0.00019908421344931173,
+      "loss": 0.2529386878013611,
+      "mean_token_accuracy": 0.8942808210849762,
+      "num_tokens": 12163296.0,
+      "step": 986
+    },
+    {
+      "entropy": 1.480097621679306,
+      "epoch": 0.5197472353870458,
+      "grad_norm": 1.469023585319519,
+      "learning_rate": 0.0001990810905652142,
+      "loss": 0.24130460619926453,
+      "mean_token_accuracy": 0.9022433310747147,
+      "num_tokens": 12175632.0,
+      "step": 987
+    },
+    {
+      "entropy": 1.498535007238388,
+      "epoch": 0.5202738283307003,
+      "grad_norm": 0.29503491520881653,
+      "learning_rate": 0.00019907796239289154,
+      "loss": 0.21136939525604248,
+      "mean_token_accuracy": 0.9171615391969681,
+      "num_tokens": 12187968.0,
+      "step": 988
+    },
+    {
+      "entropy": 1.5154505968093872,
+      "epoch": 0.5208004212743549,
+      "grad_norm": 0.29515334963798523,
+      "learning_rate": 0.00019907482893252945,
+      "loss": 0.22373729944229126,
+      "mean_token_accuracy": 0.9104124307632446,
+      "num_tokens": 12200304.0,
+      "step": 989
+    },
+    {
+      "entropy": 1.5455197095870972,
+      "epoch": 0.5213270142180095,
+      "grad_norm": 0.36879390478134155,
+      "learning_rate": 0.00019907169018431394,
+      "loss": 0.23778440058231354,
+      "mean_token_accuracy": 0.9108938276767731,
+      "num_tokens": 12212640.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.5229522287845612,
+      "epoch": 0.5218536071616641,
+      "grad_norm": 0.33947938680648804,
+      "learning_rate": 0.0001990685461484313,
+      "loss": 0.22162161767482758,
+      "mean_token_accuracy": 0.9117240607738495,
+      "num_tokens": 12224976.0,
+      "step": 991
+    },
+    {
+      "entropy": 1.5108350217342377,
+      "epoch": 0.5223802001053186,
+      "grad_norm": 0.3344981074333191,
+      "learning_rate": 0.00019906539682506823,
+      "loss": 0.20931261777877808,
+      "mean_token_accuracy": 0.9169653207063675,
+      "num_tokens": 12237312.0,
+      "step": 992
+    },
+    {
+      "entropy": 1.508901059627533,
+      "epoch": 0.5229067930489731,
+      "grad_norm": 0.3136521279811859,
+      "learning_rate": 0.00019906224221441168,
+      "loss": 0.22339576482772827,
+      "mean_token_accuracy": 0.9082980453968048,
+      "num_tokens": 12249648.0,
+      "step": 993
+    },
+    {
+      "entropy": 1.5110598504543304,
+      "epoch": 0.5234333859926277,
+      "grad_norm": 0.3550313413143158,
+      "learning_rate": 0.0001990590823166489,
+      "loss": 0.2181922197341919,
+      "mean_token_accuracy": 0.913140058517456,
+      "num_tokens": 12261984.0,
+      "step": 994
+    },
+    {
+      "entropy": 1.4990539252758026,
+      "epoch": 0.5239599789362822,
+      "grad_norm": 0.3551686108112335,
+      "learning_rate": 0.00019905591713196747,
+      "loss": 0.21492283046245575,
+      "mean_token_accuracy": 0.917353168129921,
+      "num_tokens": 12274320.0,
+      "step": 995
+    },
+    {
+      "entropy": 1.6208187341690063,
+      "epoch": 0.5244865718799369,
+      "grad_norm": 0.3526499271392822,
+      "learning_rate": 0.0001990527466605553,
+      "loss": 0.2129640281200409,
+      "mean_token_accuracy": 0.913350835442543,
+      "num_tokens": 12286656.0,
+      "step": 996
+    },
+    {
+      "entropy": 1.5281141996383667,
+      "epoch": 0.5250131648235914,
+      "grad_norm": 0.33227595686912537,
+      "learning_rate": 0.00019904957090260056,
+      "loss": 0.21114130318164825,
+      "mean_token_accuracy": 0.9172855019569397,
+      "num_tokens": 12298992.0,
+      "step": 997
+    },
+    {
+      "entropy": 1.4690798819065094,
+      "epoch": 0.5255397577672459,
+      "grad_norm": 0.30048811435699463,
+      "learning_rate": 0.00019904638985829187,
+      "loss": 0.20892660319805145,
+      "mean_token_accuracy": 0.9182561188936234,
+      "num_tokens": 12311328.0,
+      "step": 998
+    },
+    {
+      "entropy": 1.6034342348575592,
+      "epoch": 0.5260663507109005,
+      "grad_norm": 0.3628803491592407,
+      "learning_rate": 0.000199043203527818,
+      "loss": 0.21608318388462067,
+      "mean_token_accuracy": 0.913134828209877,
+      "num_tokens": 12323664.0,
+      "step": 999
+    },
+    {
+      "entropy": 1.4999005496501923,
+      "epoch": 0.526592943654555,
+      "grad_norm": 0.3294629454612732,
+      "learning_rate": 0.0001990400119113681,
+      "loss": 0.21117916703224182,
+      "mean_token_accuracy": 0.917900949716568,
+      "num_tokens": 12336000.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.596118301153183,
+      "epoch": 0.5271195365982095,
+      "grad_norm": 0.3512589931488037,
+      "learning_rate": 0.00019903681500913167,
+      "loss": 0.20125192403793335,
+      "mean_token_accuracy": 0.921380877494812,
+      "num_tokens": 12348336.0,
+      "step": 1001
+    },
+    {
+      "entropy": 1.4794040620326996,
+      "epoch": 0.5276461295418642,
+      "grad_norm": 0.3046097457408905,
+      "learning_rate": 0.0001990336128212985,
+      "loss": 0.2075495421886444,
+      "mean_token_accuracy": 0.9140082001686096,
+      "num_tokens": 12360672.0,
+      "step": 1002
+    },
+    {
+      "entropy": 1.5259474217891693,
+      "epoch": 0.5281727224855187,
+      "grad_norm": 0.3376280963420868,
+      "learning_rate": 0.00019903040534805866,
+      "loss": 0.22057998180389404,
+      "mean_token_accuracy": 0.9114288538694382,
+      "num_tokens": 12373008.0,
+      "step": 1003
+    },
+    {
+      "entropy": 1.5479835271835327,
+      "epoch": 0.5286993154291733,
+      "grad_norm": 0.3668357729911804,
+      "learning_rate": 0.00019902719258960253,
+      "loss": 0.24602656066417694,
+      "mean_token_accuracy": 0.9003023505210876,
+      "num_tokens": 12385344.0,
+      "step": 1004
+    },
+    {
+      "entropy": 1.5563693940639496,
+      "epoch": 0.5292259083728278,
+      "grad_norm": 0.33542895317077637,
+      "learning_rate": 0.0001990239745461209,
+      "loss": 0.22432827949523926,
+      "mean_token_accuracy": 0.9152787178754807,
+      "num_tokens": 12397680.0,
+      "step": 1005
+    },
+    {
+      "entropy": 1.5360921025276184,
+      "epoch": 0.5297525013164823,
+      "grad_norm": 0.34227633476257324,
+      "learning_rate": 0.00019902075121780473,
+      "loss": 0.24711929261684418,
+      "mean_token_accuracy": 0.9043864160776138,
+      "num_tokens": 12410016.0,
+      "step": 1006
+    },
+    {
+      "entropy": 1.5392629504203796,
+      "epoch": 0.5302790942601369,
+      "grad_norm": 0.31668519973754883,
+      "learning_rate": 0.00019901752260484545,
+      "loss": 0.24319474399089813,
+      "mean_token_accuracy": 0.9046953171491623,
+      "num_tokens": 12422352.0,
+      "step": 1007
+    },
+    {
+      "entropy": 1.608776479959488,
+      "epoch": 0.5308056872037915,
+      "grad_norm": 0.3281152844429016,
+      "learning_rate": 0.00019901428870743466,
+      "loss": 0.2047930210828781,
+      "mean_token_accuracy": 0.9202082604169846,
+      "num_tokens": 12434688.0,
+      "step": 1008
+    },
+    {
+      "entropy": 1.6915300488471985,
+      "epoch": 0.531332280147446,
+      "grad_norm": 0.38227638602256775,
+      "learning_rate": 0.0001990110495257644,
+      "loss": 0.23451298475265503,
+      "mean_token_accuracy": 0.9067147076129913,
+      "num_tokens": 12447024.0,
+      "step": 1009
+    },
+    {
+      "entropy": 1.5676094889640808,
+      "epoch": 0.5318588730911006,
+      "grad_norm": 0.3091145157814026,
+      "learning_rate": 0.0001990078050600269,
+      "loss": 0.21618413925170898,
+      "mean_token_accuracy": 0.9121405184268951,
+      "num_tokens": 12459360.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.6247548460960388,
+      "epoch": 0.5323854660347551,
+      "grad_norm": 0.3261072337627411,
+      "learning_rate": 0.0001990045553104148,
+      "loss": 0.2148750126361847,
+      "mean_token_accuracy": 0.9210677593946457,
+      "num_tokens": 12471696.0,
+      "step": 1011
+    },
+    {
+      "entropy": 1.5738478302955627,
+      "epoch": 0.5329120589784097,
+      "grad_norm": 0.32214832305908203,
+      "learning_rate": 0.00019900130027712099,
+      "loss": 0.23918862640857697,
+      "mean_token_accuracy": 0.9062842279672623,
+      "num_tokens": 12484032.0,
+      "step": 1012
+    },
+    {
+      "entropy": 1.474539339542389,
+      "epoch": 0.5334386519220643,
+      "grad_norm": 0.346766859292984,
+      "learning_rate": 0.00019899803996033876,
+      "loss": 0.21119186282157898,
+      "mean_token_accuracy": 0.918285146355629,
+      "num_tokens": 12496368.0,
+      "step": 1013
+    },
+    {
+      "entropy": 1.5115889310836792,
+      "epoch": 0.5339652448657188,
+      "grad_norm": 0.3021303117275238,
+      "learning_rate": 0.00019899477436026157,
+      "loss": 0.19445011019706726,
+      "mean_token_accuracy": 0.9237413257360458,
+      "num_tokens": 12508704.0,
+      "step": 1014
+    },
+    {
+      "entropy": 1.492755115032196,
+      "epoch": 0.5344918378093734,
+      "grad_norm": 0.3711804449558258,
+      "learning_rate": 0.00019899150347708335,
+      "loss": 0.21977633237838745,
+      "mean_token_accuracy": 0.9105686545372009,
+      "num_tokens": 12521040.0,
+      "step": 1015
+    },
+    {
+      "entropy": 1.4154399931430817,
+      "epoch": 0.5350184307530279,
+      "grad_norm": 0.34001612663269043,
+      "learning_rate": 0.00019898822731099827,
+      "loss": 0.1881294697523117,
+      "mean_token_accuracy": 0.924776166677475,
+      "num_tokens": 12533376.0,
+      "step": 1016
+    },
+    {
+      "entropy": 1.41116601228714,
+      "epoch": 0.5355450236966824,
+      "grad_norm": 0.33592158555984497,
+      "learning_rate": 0.00019898494586220077,
+      "loss": 0.22032563388347626,
+      "mean_token_accuracy": 0.9117150008678436,
+      "num_tokens": 12545712.0,
+      "step": 1017
+    },
+    {
+      "entropy": 1.416001707315445,
+      "epoch": 0.536071616640337,
+      "grad_norm": 0.3635656535625458,
+      "learning_rate": 0.00019898165913088568,
+      "loss": 0.2379489541053772,
+      "mean_token_accuracy": 0.9045498371124268,
+      "num_tokens": 12558048.0,
+      "step": 1018
+    },
+    {
+      "entropy": 1.346794456243515,
+      "epoch": 0.5365982095839916,
+      "grad_norm": 0.41112419962882996,
+      "learning_rate": 0.0001989783671172481,
+      "loss": 0.24667346477508545,
+      "mean_token_accuracy": 0.9044399708509445,
+      "num_tokens": 12570384.0,
+      "step": 1019
+    },
+    {
+      "entropy": 1.417405605316162,
+      "epoch": 0.5371248025276462,
+      "grad_norm": 0.3971119821071625,
+      "learning_rate": 0.00019897506982148353,
+      "loss": 0.19713318347930908,
+      "mean_token_accuracy": 0.9267271012067795,
+      "num_tokens": 12582720.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.4534864723682404,
+      "epoch": 0.5376513954713007,
+      "grad_norm": 0.3351125717163086,
+      "learning_rate": 0.00019897176724378762,
+      "loss": 0.22281871736049652,
+      "mean_token_accuracy": 0.9098447114229202,
+      "num_tokens": 12595056.0,
+      "step": 1021
+    },
+    {
+      "entropy": 1.3609624803066254,
+      "epoch": 0.5381779884149552,
+      "grad_norm": 0.3119547665119171,
+      "learning_rate": 0.00019896845938435642,
+      "loss": 0.2145942896604538,
+      "mean_token_accuracy": 0.9176490157842636,
+      "num_tokens": 12607392.0,
+      "step": 1022
+    },
+    {
+      "entropy": 1.373078614473343,
+      "epoch": 0.5387045813586098,
+      "grad_norm": 0.3335905075073242,
+      "learning_rate": 0.00019896514624338633,
+      "loss": 0.20780502259731293,
+      "mean_token_accuracy": 0.9145664125680923,
+      "num_tokens": 12619728.0,
+      "step": 1023
+    },
+    {
+      "entropy": 1.45356023311615,
+      "epoch": 0.5392311743022643,
+      "grad_norm": 0.321328729391098,
+      "learning_rate": 0.00019896182782107408,
+      "loss": 0.21474064886569977,
+      "mean_token_accuracy": 0.914635494351387,
+      "num_tokens": 12632064.0,
+      "step": 1024
+    },
+    {
+      "entropy": 1.3756296038627625,
+      "epoch": 0.539757767245919,
+      "grad_norm": 0.3236009180545807,
+      "learning_rate": 0.0001989585041176166,
+      "loss": 0.19574275612831116,
+      "mean_token_accuracy": 0.9214889109134674,
+      "num_tokens": 12644400.0,
+      "step": 1025
+    },
+    {
+      "entropy": 1.374951332807541,
+      "epoch": 0.5402843601895735,
+      "grad_norm": 0.33436882495880127,
+      "learning_rate": 0.00019895517513321125,
+      "loss": 0.2343987077474594,
+      "mean_token_accuracy": 0.9015287011861801,
+      "num_tokens": 12656736.0,
+      "step": 1026
+    },
+    {
+      "entropy": 1.3473423719406128,
+      "epoch": 0.540810953133228,
+      "grad_norm": 0.30872446298599243,
+      "learning_rate": 0.0001989518408680556,
+      "loss": 0.23760074377059937,
+      "mean_token_accuracy": 0.9077365547418594,
+      "num_tokens": 12669072.0,
+      "step": 1027
+    },
+    {
+      "entropy": 1.4245812594890594,
+      "epoch": 0.5413375460768826,
+      "grad_norm": 0.32098084688186646,
+      "learning_rate": 0.0001989485013223476,
+      "loss": 0.22173723578453064,
+      "mean_token_accuracy": 0.9116310328245163,
+      "num_tokens": 12681408.0,
+      "step": 1028
+    },
+    {
+      "entropy": 1.357390284538269,
+      "epoch": 0.5418641390205371,
+      "grad_norm": 0.3540545701980591,
+      "learning_rate": 0.0001989451564962855,
+      "loss": 0.22398629784584045,
+      "mean_token_accuracy": 0.914269283413887,
+      "num_tokens": 12693744.0,
+      "step": 1029
+    },
+    {
+      "entropy": 1.4113155901432037,
+      "epoch": 0.5423907319641916,
+      "grad_norm": 0.3349270224571228,
+      "learning_rate": 0.00019894180639006787,
+      "loss": 0.23276042938232422,
+      "mean_token_accuracy": 0.905420333147049,
+      "num_tokens": 12706080.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.3748076260089874,
+      "epoch": 0.5429173249078463,
+      "grad_norm": 0.31247857213020325,
+      "learning_rate": 0.0001989384510038936,
+      "loss": 0.2479267716407776,
+      "mean_token_accuracy": 0.8999170660972595,
+      "num_tokens": 12718416.0,
+      "step": 1031
+    },
+    {
+      "entropy": 1.4230112135410309,
+      "epoch": 0.5434439178515008,
+      "grad_norm": 0.31677430868148804,
+      "learning_rate": 0.0001989350903379619,
+      "loss": 0.2388826310634613,
+      "mean_token_accuracy": 0.9079335629940033,
+      "num_tokens": 12730752.0,
+      "step": 1032
+    },
+    {
+      "entropy": 1.341791033744812,
+      "epoch": 0.5439705107951553,
+      "grad_norm": 0.30199915170669556,
+      "learning_rate": 0.0001989317243924722,
+      "loss": 0.21240849792957306,
+      "mean_token_accuracy": 0.9131926447153091,
+      "num_tokens": 12743088.0,
+      "step": 1033
+    },
+    {
+      "entropy": 1.3770472407341003,
+      "epoch": 0.5444971037388099,
+      "grad_norm": 0.28729885816574097,
+      "learning_rate": 0.00019892835316762437,
+      "loss": 0.20151323080062866,
+      "mean_token_accuracy": 0.9193004071712494,
+      "num_tokens": 12755424.0,
+      "step": 1034
+    },
+    {
+      "entropy": 1.3323063850402832,
+      "epoch": 0.5450236966824644,
+      "grad_norm": 0.3478204607963562,
+      "learning_rate": 0.00019892497666361848,
+      "loss": 0.22956590354442596,
+      "mean_token_accuracy": 0.910034716129303,
+      "num_tokens": 12767760.0,
+      "step": 1035
+    },
+    {
+      "entropy": 1.2777645885944366,
+      "epoch": 0.545550289626119,
+      "grad_norm": 0.32218146324157715,
+      "learning_rate": 0.00019892159488065506,
+      "loss": 0.2523505985736847,
+      "mean_token_accuracy": 0.8999349474906921,
+      "num_tokens": 12780096.0,
+      "step": 1036
+    },
+    {
+      "entropy": 1.3476167619228363,
+      "epoch": 0.5460768825697736,
+      "grad_norm": 0.3297399580478668,
+      "learning_rate": 0.00019891820781893485,
+      "loss": 0.24944739043712616,
+      "mean_token_accuracy": 0.8990609049797058,
+      "num_tokens": 12792432.0,
+      "step": 1037
+    },
+    {
+      "entropy": 1.3680584728717804,
+      "epoch": 0.5466034755134281,
+      "grad_norm": 0.31413063406944275,
+      "learning_rate": 0.00019891481547865887,
+      "loss": 0.23303711414337158,
+      "mean_token_accuracy": 0.9072806537151337,
+      "num_tokens": 12804768.0,
+      "step": 1038
+    },
+    {
+      "entropy": 1.2887639105319977,
+      "epoch": 0.5471300684570827,
+      "grad_norm": 0.30452004075050354,
+      "learning_rate": 0.00019891141786002853,
+      "loss": 0.20885013043880463,
+      "mean_token_accuracy": 0.91634900867939,
+      "num_tokens": 12817104.0,
+      "step": 1039
+    },
+    {
+      "entropy": 1.427742451429367,
+      "epoch": 0.5476566614007372,
+      "grad_norm": 0.3391040563583374,
+      "learning_rate": 0.00019890801496324554,
+      "loss": 0.21363744139671326,
+      "mean_token_accuracy": 0.9168877154588699,
+      "num_tokens": 12829440.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.4079940021038055,
+      "epoch": 0.5481832543443917,
+      "grad_norm": 0.34792712330818176,
+      "learning_rate": 0.00019890460678851187,
+      "loss": 0.23039647936820984,
+      "mean_token_accuracy": 0.9067210853099823,
+      "num_tokens": 12841776.0,
+      "step": 1041
+    },
+    {
+      "entropy": 1.362496793270111,
+      "epoch": 0.5487098472880464,
+      "grad_norm": 0.3340260684490204,
+      "learning_rate": 0.00019890119333602988,
+      "loss": 0.21935522556304932,
+      "mean_token_accuracy": 0.9152266681194305,
+      "num_tokens": 12854112.0,
+      "step": 1042
+    },
+    {
+      "entropy": 1.3945021033287048,
+      "epoch": 0.5492364402317009,
+      "grad_norm": 0.31123343110084534,
+      "learning_rate": 0.00019889777460600221,
+      "loss": 0.2165759801864624,
+      "mean_token_accuracy": 0.9149927645921707,
+      "num_tokens": 12866448.0,
+      "step": 1043
+    },
+    {
+      "entropy": 1.3361455798149109,
+      "epoch": 0.5497630331753555,
+      "grad_norm": 0.30849552154541016,
+      "learning_rate": 0.0001988943505986318,
+      "loss": 0.23334844410419464,
+      "mean_token_accuracy": 0.9073198139667511,
+      "num_tokens": 12878784.0,
+      "step": 1044
+    },
+    {
+      "entropy": 1.353363960981369,
+      "epoch": 0.55028962611901,
+      "grad_norm": 0.30394431948661804,
+      "learning_rate": 0.00019889092131412188,
+      "loss": 0.21873678267002106,
+      "mean_token_accuracy": 0.9092230349779129,
+      "num_tokens": 12891120.0,
+      "step": 1045
+    },
+    {
+      "entropy": 1.3730317652225494,
+      "epoch": 0.5508162190626645,
+      "grad_norm": 0.3051231801509857,
+      "learning_rate": 0.0001988874867526761,
+      "loss": 0.2074829638004303,
+      "mean_token_accuracy": 0.9159383773803711,
+      "num_tokens": 12903456.0,
+      "step": 1046
+    },
+    {
+      "entropy": 1.348529189825058,
+      "epoch": 0.5513428120063191,
+      "grad_norm": 0.34162041544914246,
+      "learning_rate": 0.00019888404691449828,
+      "loss": 0.20376741886138916,
+      "mean_token_accuracy": 0.9213349372148514,
+      "num_tokens": 12915792.0,
+      "step": 1047
+    },
+    {
+      "entropy": 1.3659497201442719,
+      "epoch": 0.5518694049499737,
+      "grad_norm": 0.3287021219730377,
+      "learning_rate": 0.00019888060179979266,
+      "loss": 0.22284625470638275,
+      "mean_token_accuracy": 0.9160720854997635,
+      "num_tokens": 12928128.0,
+      "step": 1048
+    },
+    {
+      "entropy": 1.39521923661232,
+      "epoch": 0.5523959978936283,
+      "grad_norm": 0.33031636476516724,
+      "learning_rate": 0.00019887715140876373,
+      "loss": 0.2400185465812683,
+      "mean_token_accuracy": 0.9030187875032425,
+      "num_tokens": 12940464.0,
+      "step": 1049
+    },
+    {
+      "entropy": 1.4470847249031067,
+      "epoch": 0.5529225908372828,
+      "grad_norm": 0.37168604135513306,
+      "learning_rate": 0.00019887369574161633,
+      "loss": 0.22982707619667053,
+      "mean_token_accuracy": 0.9084116071462631,
+      "num_tokens": 12952800.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.3384085595607758,
+      "epoch": 0.5534491837809373,
+      "grad_norm": 0.32396450638771057,
+      "learning_rate": 0.00019887023479855565,
+      "loss": 0.24000367522239685,
+      "mean_token_accuracy": 0.9041632562875748,
+      "num_tokens": 12965136.0,
+      "step": 1051
+    },
+    {
+      "entropy": 1.3531986474990845,
+      "epoch": 0.5539757767245919,
+      "grad_norm": 0.319603830575943,
+      "learning_rate": 0.00019886676857978707,
+      "loss": 0.23166413605213165,
+      "mean_token_accuracy": 0.9047738164663315,
+      "num_tokens": 12977472.0,
+      "step": 1052
+    },
+    {
+      "entropy": 1.3648565709590912,
+      "epoch": 0.5545023696682464,
+      "grad_norm": 0.3087165057659149,
+      "learning_rate": 0.00019886329708551642,
+      "loss": 0.21422043442726135,
+      "mean_token_accuracy": 0.9170399606227875,
+      "num_tokens": 12989808.0,
+      "step": 1053
+    },
+    {
+      "entropy": 1.4327314496040344,
+      "epoch": 0.555028962611901,
+      "grad_norm": 0.31154927611351013,
+      "learning_rate": 0.00019885982031594973,
+      "loss": 0.23706527054309845,
+      "mean_token_accuracy": 0.9066223204135895,
+      "num_tokens": 13002144.0,
+      "step": 1054
+    },
+    {
+      "entropy": 1.401264727115631,
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.30120137333869934,
+      "learning_rate": 0.00019885633827129343,
+      "loss": 0.21149972081184387,
+      "mean_token_accuracy": 0.9169722944498062,
+      "num_tokens": 13014480.0,
+      "step": 1055
+    },
+    {
+      "entropy": 1.4713716506958008,
+      "epoch": 0.5560821484992101,
+      "grad_norm": 0.39476147294044495,
+      "learning_rate": 0.00019885285095175422,
+      "loss": 0.26564526557922363,
+      "mean_token_accuracy": 0.8999374806880951,
+      "num_tokens": 13026816.0,
+      "step": 1056
+    },
+    {
+      "entropy": 1.4001291990280151,
+      "epoch": 0.5566087414428647,
+      "grad_norm": 0.3243412971496582,
+      "learning_rate": 0.00019884935835753914,
+      "loss": 0.2307104468345642,
+      "mean_token_accuracy": 0.9069186598062515,
+      "num_tokens": 13039152.0,
+      "step": 1057
+    },
+    {
+      "entropy": 1.4063750207424164,
+      "epoch": 0.5571353343865192,
+      "grad_norm": 0.3205982446670532,
+      "learning_rate": 0.0001988458604888555,
+      "loss": 0.2200152724981308,
+      "mean_token_accuracy": 0.9117906540632248,
+      "num_tokens": 13051488.0,
+      "step": 1058
+    },
+    {
+      "entropy": 1.4639129042625427,
+      "epoch": 0.5576619273301737,
+      "grad_norm": 0.35051432251930237,
+      "learning_rate": 0.00019884235734591096,
+      "loss": 0.21262915432453156,
+      "mean_token_accuracy": 0.9155555963516235,
+      "num_tokens": 13063824.0,
+      "step": 1059
+    },
+    {
+      "entropy": 1.3807141780853271,
+      "epoch": 0.5581885202738284,
+      "grad_norm": 0.31436532735824585,
+      "learning_rate": 0.00019883884892891348,
+      "loss": 0.23192916810512543,
+      "mean_token_accuracy": 0.9132644683122635,
+      "num_tokens": 13076160.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.4159002602100372,
+      "epoch": 0.5587151132174829,
+      "grad_norm": 0.35557860136032104,
+      "learning_rate": 0.00019883533523807131,
+      "loss": 0.2557951807975769,
+      "mean_token_accuracy": 0.9019583463668823,
+      "num_tokens": 13088496.0,
+      "step": 1061
+    },
+    {
+      "entropy": 1.4077826738357544,
+      "epoch": 0.5592417061611374,
+      "grad_norm": 0.35081014037132263,
+      "learning_rate": 0.00019883181627359305,
+      "loss": 0.21151600778102875,
+      "mean_token_accuracy": 0.914755642414093,
+      "num_tokens": 13100832.0,
+      "step": 1062
+    },
+    {
+      "entropy": 1.3742648661136627,
+      "epoch": 0.559768299104792,
+      "grad_norm": 0.28598442673683167,
+      "learning_rate": 0.00019882829203568764,
+      "loss": 0.21076026558876038,
+      "mean_token_accuracy": 0.9148049652576447,
+      "num_tokens": 13113168.0,
+      "step": 1063
+    },
+    {
+      "entropy": 1.4464651346206665,
+      "epoch": 0.5602948920484465,
+      "grad_norm": 0.29035648703575134,
+      "learning_rate": 0.00019882476252456428,
+      "loss": 0.20042768120765686,
+      "mean_token_accuracy": 0.9182227253913879,
+      "num_tokens": 13125504.0,
+      "step": 1064
+    },
+    {
+      "entropy": 1.3639024496078491,
+      "epoch": 0.5608214849921012,
+      "grad_norm": 0.3004502058029175,
+      "learning_rate": 0.00019882122774043243,
+      "loss": 0.2036202847957611,
+      "mean_token_accuracy": 0.9217555373907089,
+      "num_tokens": 13137840.0,
+      "step": 1065
+    },
+    {
+      "entropy": 1.3562072217464447,
+      "epoch": 0.5613480779357557,
+      "grad_norm": 0.29948773980140686,
+      "learning_rate": 0.000198817687683502,
+      "loss": 0.20943708717823029,
+      "mean_token_accuracy": 0.9172091037034988,
+      "num_tokens": 13150176.0,
+      "step": 1066
+    },
+    {
+      "entropy": 1.4171293675899506,
+      "epoch": 0.5618746708794102,
+      "grad_norm": 0.32766589522361755,
+      "learning_rate": 0.00019881414235398313,
+      "loss": 0.22548557817935944,
+      "mean_token_accuracy": 0.9138119965791702,
+      "num_tokens": 13162512.0,
+      "step": 1067
+    },
+    {
+      "entropy": 1.494034618139267,
+      "epoch": 0.5624012638230648,
+      "grad_norm": 0.3495429754257202,
+      "learning_rate": 0.00019881059175208624,
+      "loss": 0.24128882586956024,
+      "mean_token_accuracy": 0.9064372777938843,
+      "num_tokens": 13174848.0,
+      "step": 1068
+    },
+    {
+      "entropy": 1.4479554295539856,
+      "epoch": 0.5629278567667193,
+      "grad_norm": 0.3285626769065857,
+      "learning_rate": 0.0001988070358780222,
+      "loss": 0.1992790549993515,
+      "mean_token_accuracy": 0.922189861536026,
+      "num_tokens": 13187184.0,
+      "step": 1069
+    },
+    {
+      "entropy": 1.36091148853302,
+      "epoch": 0.5634544497103738,
+      "grad_norm": 0.29997125267982483,
+      "learning_rate": 0.00019880347473200197,
+      "loss": 0.19809836149215698,
+      "mean_token_accuracy": 0.9212192296981812,
+      "num_tokens": 13199520.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.4168705940246582,
+      "epoch": 0.5639810426540285,
+      "grad_norm": 0.320639044046402,
+      "learning_rate": 0.0001987999083142371,
+      "loss": 0.21112670004367828,
+      "mean_token_accuracy": 0.9181769639253616,
+      "num_tokens": 13211856.0,
+      "step": 1071
+    },
+    {
+      "entropy": 1.3809196650981903,
+      "epoch": 0.564507635597683,
+      "grad_norm": 0.36394959688186646,
+      "learning_rate": 0.0001987963366249392,
+      "loss": 0.23442226648330688,
+      "mean_token_accuracy": 0.9071120172739029,
+      "num_tokens": 13224192.0,
+      "step": 1072
+    },
+    {
+      "entropy": 1.3890978693962097,
+      "epoch": 0.5650342285413376,
+      "grad_norm": 0.34244871139526367,
+      "learning_rate": 0.00019879275966432037,
+      "loss": 0.22117844223976135,
+      "mean_token_accuracy": 0.9115606546401978,
+      "num_tokens": 13236528.0,
+      "step": 1073
+    },
+    {
+      "entropy": 1.3886411488056183,
+      "epoch": 0.5655608214849921,
+      "grad_norm": 0.35859042406082153,
+      "learning_rate": 0.00019878917743259292,
+      "loss": 0.21675634384155273,
+      "mean_token_accuracy": 0.9194170236587524,
+      "num_tokens": 13248864.0,
+      "step": 1074
+    },
+    {
+      "entropy": 1.4300493896007538,
+      "epoch": 0.5660874144286466,
+      "grad_norm": 0.34859365224838257,
+      "learning_rate": 0.00019878558992996947,
+      "loss": 0.24318671226501465,
+      "mean_token_accuracy": 0.9037031382322311,
+      "num_tokens": 13261200.0,
+      "step": 1075
+    },
+    {
+      "entropy": 1.3681704699993134,
+      "epoch": 0.5666140073723012,
+      "grad_norm": 0.3271603286266327,
+      "learning_rate": 0.00019878199715666305,
+      "loss": 0.2319377213716507,
+      "mean_token_accuracy": 0.9129595458507538,
+      "num_tokens": 13273536.0,
+      "step": 1076
+    },
+    {
+      "entropy": 1.395903766155243,
+      "epoch": 0.5671406003159558,
+      "grad_norm": 0.29173579812049866,
+      "learning_rate": 0.00019877839911288693,
+      "loss": 0.20177659392356873,
+      "mean_token_accuracy": 0.9242661744356155,
+      "num_tokens": 13285872.0,
+      "step": 1077
+    },
+    {
+      "entropy": 1.4401375949382782,
+      "epoch": 0.5676671932596103,
+      "grad_norm": 0.3181249499320984,
+      "learning_rate": 0.0001987747957988547,
+      "loss": 0.22763216495513916,
+      "mean_token_accuracy": 0.907039076089859,
+      "num_tokens": 13298208.0,
+      "step": 1078
+    },
+    {
+      "entropy": 1.3585200309753418,
+      "epoch": 0.5681937862032649,
+      "grad_norm": 0.2902856171131134,
+      "learning_rate": 0.00019877118721478026,
+      "loss": 0.19510802626609802,
+      "mean_token_accuracy": 0.9210563600063324,
+      "num_tokens": 13310544.0,
+      "step": 1079
+    },
+    {
+      "entropy": 1.3690218329429626,
+      "epoch": 0.5687203791469194,
+      "grad_norm": 0.31650200486183167,
+      "learning_rate": 0.00019876757336087782,
+      "loss": 0.23386812210083008,
+      "mean_token_accuracy": 0.9065526723861694,
+      "num_tokens": 13322880.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.4170444905757904,
+      "epoch": 0.569246972090574,
+      "grad_norm": 0.39458808302879333,
+      "learning_rate": 0.00019876395423736192,
+      "loss": 0.21256215870380402,
+      "mean_token_accuracy": 0.9154652059078217,
+      "num_tokens": 13335216.0,
+      "step": 1081
+    },
+    {
+      "entropy": 1.3711142838001251,
+      "epoch": 0.5697735650342285,
+      "grad_norm": 0.3102315366268158,
+      "learning_rate": 0.00019876032984444744,
+      "loss": 0.2107260674238205,
+      "mean_token_accuracy": 0.9209572970867157,
+      "num_tokens": 13347552.0,
+      "step": 1082
+    },
+    {
+      "entropy": 1.3468361794948578,
+      "epoch": 0.5703001579778831,
+      "grad_norm": 0.3204945921897888,
+      "learning_rate": 0.00019875670018234946,
+      "loss": 0.21677452325820923,
+      "mean_token_accuracy": 0.9137840569019318,
+      "num_tokens": 13359888.0,
+      "step": 1083
+    },
+    {
+      "entropy": 1.3397773206233978,
+      "epoch": 0.5708267509215377,
+      "grad_norm": 0.30542871356010437,
+      "learning_rate": 0.00019875306525128354,
+      "loss": 0.20801898837089539,
+      "mean_token_accuracy": 0.9187730103731155,
+      "num_tokens": 13372224.0,
+      "step": 1084
+    },
+    {
+      "entropy": 1.3542706668376923,
+      "epoch": 0.5713533438651922,
+      "grad_norm": 0.39070987701416016,
+      "learning_rate": 0.00019874942505146542,
+      "loss": 0.23829345405101776,
+      "mean_token_accuracy": 0.9077901542186737,
+      "num_tokens": 13384560.0,
+      "step": 1085
+    },
+    {
+      "entropy": 1.3789608180522919,
+      "epoch": 0.5718799368088467,
+      "grad_norm": 0.31589949131011963,
+      "learning_rate": 0.00019874577958311124,
+      "loss": 0.23594669997692108,
+      "mean_token_accuracy": 0.9101516306400299,
+      "num_tokens": 13396896.0,
+      "step": 1086
+    },
+    {
+      "entropy": 1.2831343114376068,
+      "epoch": 0.5724065297525013,
+      "grad_norm": 0.31056761741638184,
+      "learning_rate": 0.00019874212884643733,
+      "loss": 0.22809851169586182,
+      "mean_token_accuracy": 0.9048083126544952,
+      "num_tokens": 13409232.0,
+      "step": 1087
+    },
+    {
+      "entropy": 1.343775451183319,
+      "epoch": 0.5729331226961558,
+      "grad_norm": 0.35355332493782043,
+      "learning_rate": 0.00019873847284166047,
+      "loss": 0.24591763317584991,
+      "mean_token_accuracy": 0.9045094102621078,
+      "num_tokens": 13421568.0,
+      "step": 1088
+    },
+    {
+      "entropy": 1.3326263129711151,
+      "epoch": 0.5734597156398105,
+      "grad_norm": 0.30252182483673096,
+      "learning_rate": 0.00019873481156899769,
+      "loss": 0.2264258861541748,
+      "mean_token_accuracy": 0.9123017936944962,
+      "num_tokens": 13433904.0,
+      "step": 1089
+    },
+    {
+      "entropy": 1.4111762046813965,
+      "epoch": 0.573986308583465,
+      "grad_norm": 0.30441656708717346,
+      "learning_rate": 0.00019873114502866633,
+      "loss": 0.20945101976394653,
+      "mean_token_accuracy": 0.9133919775485992,
+      "num_tokens": 13446240.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.3507026731967926,
+      "epoch": 0.5745129015271195,
+      "grad_norm": 0.29725486040115356,
+      "learning_rate": 0.00019872747322088405,
+      "loss": 0.22033721208572388,
+      "mean_token_accuracy": 0.9093347936868668,
+      "num_tokens": 13458576.0,
+      "step": 1091
+    },
+    {
+      "entropy": 1.3604736626148224,
+      "epoch": 0.5750394944707741,
+      "grad_norm": 0.2988698482513428,
+      "learning_rate": 0.00019872379614586884,
+      "loss": 0.20934537053108215,
+      "mean_token_accuracy": 0.919558510184288,
+      "num_tokens": 13470912.0,
+      "step": 1092
+    },
+    {
+      "entropy": 1.2718960344791412,
+      "epoch": 0.5755660874144286,
+      "grad_norm": 0.296278178691864,
+      "learning_rate": 0.00019872011380383896,
+      "loss": 0.2263902723789215,
+      "mean_token_accuracy": 0.9120451807975769,
+      "num_tokens": 13483248.0,
+      "step": 1093
+    },
+    {
+      "entropy": 1.3046118319034576,
+      "epoch": 0.5760926803580833,
+      "grad_norm": 0.2931486666202545,
+      "learning_rate": 0.00019871642619501302,
+      "loss": 0.22945842146873474,
+      "mean_token_accuracy": 0.9100485295057297,
+      "num_tokens": 13495584.0,
+      "step": 1094
+    },
+    {
+      "entropy": 1.3241084516048431,
+      "epoch": 0.5766192733017378,
+      "grad_norm": 0.3063081204891205,
+      "learning_rate": 0.00019871273331960996,
+      "loss": 0.2523494362831116,
+      "mean_token_accuracy": 0.9037816673517227,
+      "num_tokens": 13507920.0,
+      "step": 1095
+    },
+    {
+      "entropy": 1.3798290491104126,
+      "epoch": 0.5771458662453923,
+      "grad_norm": 0.3451617956161499,
+      "learning_rate": 0.00019870903517784898,
+      "loss": 0.2181454300880432,
+      "mean_token_accuracy": 0.9167459905147552,
+      "num_tokens": 13520256.0,
+      "step": 1096
+    },
+    {
+      "entropy": 1.292254477739334,
+      "epoch": 0.5776724591890469,
+      "grad_norm": 0.29267409443855286,
+      "learning_rate": 0.0001987053317699496,
+      "loss": 0.2220621407032013,
+      "mean_token_accuracy": 0.9128179252147675,
+      "num_tokens": 13532592.0,
+      "step": 1097
+    },
+    {
+      "entropy": 1.2694779932498932,
+      "epoch": 0.5781990521327014,
+      "grad_norm": 0.2939811646938324,
+      "learning_rate": 0.0001987016230961317,
+      "loss": 0.21204796433448792,
+      "mean_token_accuracy": 0.9153830707073212,
+      "num_tokens": 13544928.0,
+      "step": 1098
+    },
+    {
+      "entropy": 1.301151692867279,
+      "epoch": 0.5787256450763559,
+      "grad_norm": 0.3276446461677551,
+      "learning_rate": 0.0001986979091566154,
+      "loss": 0.20669840276241302,
+      "mean_token_accuracy": 0.9193602800369263,
+      "num_tokens": 13557264.0,
+      "step": 1099
+    },
+    {
+      "entropy": 1.2780292332172394,
+      "epoch": 0.5792522380200106,
+      "grad_norm": 0.31635478138923645,
+      "learning_rate": 0.00019869418995162124,
+      "loss": 0.22475430369377136,
+      "mean_token_accuracy": 0.9149479269981384,
+      "num_tokens": 13569600.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.3376207053661346,
+      "epoch": 0.5797788309636651,
+      "grad_norm": 0.3281361162662506,
+      "learning_rate": 0.00019869046548136996,
+      "loss": 0.2165728509426117,
+      "mean_token_accuracy": 0.9121999144554138,
+      "num_tokens": 13581936.0,
+      "step": 1101
+    },
+    {
+      "entropy": 1.2792271375656128,
+      "epoch": 0.5803054239073197,
+      "grad_norm": 0.34059932827949524,
+      "learning_rate": 0.00019868673574608266,
+      "loss": 0.2282489687204361,
+      "mean_token_accuracy": 0.9086829572916031,
+      "num_tokens": 13594272.0,
+      "step": 1102
+    },
+    {
+      "entropy": 1.312345266342163,
+      "epoch": 0.5808320168509742,
+      "grad_norm": 0.3228605389595032,
+      "learning_rate": 0.00019868300074598076,
+      "loss": 0.22286418080329895,
+      "mean_token_accuracy": 0.912465900182724,
+      "num_tokens": 13606608.0,
+      "step": 1103
+    },
+    {
+      "entropy": 1.3179580569267273,
+      "epoch": 0.5813586097946287,
+      "grad_norm": 0.3180788457393646,
+      "learning_rate": 0.000198679260481286,
+      "loss": 0.2119407206773758,
+      "mean_token_accuracy": 0.9113909304141998,
+      "num_tokens": 13618944.0,
+      "step": 1104
+    },
+    {
+      "entropy": 1.3403850197792053,
+      "epoch": 0.5818852027382833,
+      "grad_norm": 0.3467418849468231,
+      "learning_rate": 0.00019867551495222042,
+      "loss": 0.23494362831115723,
+      "mean_token_accuracy": 0.9045893847942352,
+      "num_tokens": 13631280.0,
+      "step": 1105
+    },
+    {
+      "entropy": 1.3972817361354828,
+      "epoch": 0.5824117956819379,
+      "grad_norm": 0.289737343788147,
+      "learning_rate": 0.00019867176415900635,
+      "loss": 0.22767332196235657,
+      "mean_token_accuracy": 0.9065064936876297,
+      "num_tokens": 13643616.0,
+      "step": 1106
+    },
+    {
+      "entropy": 1.3646803498268127,
+      "epoch": 0.5829383886255924,
+      "grad_norm": 0.30822867155075073,
+      "learning_rate": 0.00019866800810186644,
+      "loss": 0.22304248809814453,
+      "mean_token_accuracy": 0.9127557724714279,
+      "num_tokens": 13655952.0,
+      "step": 1107
+    },
+    {
+      "entropy": 1.3917404413223267,
+      "epoch": 0.583464981569247,
+      "grad_norm": 0.3238813579082489,
+      "learning_rate": 0.0001986642467810237,
+      "loss": 0.23602177202701569,
+      "mean_token_accuracy": 0.9076020121574402,
+      "num_tokens": 13668288.0,
+      "step": 1108
+    },
+    {
+      "entropy": 1.4159584939479828,
+      "epoch": 0.5839915745129015,
+      "grad_norm": 0.3269672989845276,
+      "learning_rate": 0.00019866048019670144,
+      "loss": 0.22922644019126892,
+      "mean_token_accuracy": 0.9118731915950775,
+      "num_tokens": 13680624.0,
+      "step": 1109
+    },
+    {
+      "entropy": 1.426487773656845,
+      "epoch": 0.584518167456556,
+      "grad_norm": 0.33397728204727173,
+      "learning_rate": 0.00019865670834912316,
+      "loss": 0.21286353468894958,
+      "mean_token_accuracy": 0.921152800321579,
+      "num_tokens": 13692960.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.4119893908500671,
+      "epoch": 0.5850447604002106,
+      "grad_norm": 0.29753443598747253,
+      "learning_rate": 0.00019865293123851288,
+      "loss": 0.2373121976852417,
+      "mean_token_accuracy": 0.9035847336053848,
+      "num_tokens": 13705296.0,
+      "step": 1111
+    },
+    {
+      "entropy": 1.380676507949829,
+      "epoch": 0.5855713533438652,
+      "grad_norm": 0.2954482436180115,
+      "learning_rate": 0.00019864914886509475,
+      "loss": 0.21925240755081177,
+      "mean_token_accuracy": 0.9149321764707565,
+      "num_tokens": 13717632.0,
+      "step": 1112
+    },
+    {
+      "entropy": 1.382420837879181,
+      "epoch": 0.5860979462875198,
+      "grad_norm": 0.3165317177772522,
+      "learning_rate": 0.00019864536122909336,
+      "loss": 0.20962990820407867,
+      "mean_token_accuracy": 0.9176686257123947,
+      "num_tokens": 13729968.0,
+      "step": 1113
+    },
+    {
+      "entropy": 1.4016212224960327,
+      "epoch": 0.5866245392311743,
+      "grad_norm": 0.3741922974586487,
+      "learning_rate": 0.00019864156833073352,
+      "loss": 0.21471966803073883,
+      "mean_token_accuracy": 0.9138137102127075,
+      "num_tokens": 13742304.0,
+      "step": 1114
+    },
+    {
+      "entropy": 1.330253154039383,
+      "epoch": 0.5871511321748288,
+      "grad_norm": 0.30213749408721924,
+      "learning_rate": 0.00019863777017024042,
+      "loss": 0.21769596636295319,
+      "mean_token_accuracy": 0.9100621193647385,
+      "num_tokens": 13754640.0,
+      "step": 1115
+    },
+    {
+      "entropy": 1.353569656610489,
+      "epoch": 0.5876777251184834,
+      "grad_norm": 0.37241899967193604,
+      "learning_rate": 0.00019863396674783953,
+      "loss": 0.2180408537387848,
+      "mean_token_accuracy": 0.9123483598232269,
+      "num_tokens": 13766976.0,
+      "step": 1116
+    },
+    {
+      "entropy": 1.3220522999763489,
+      "epoch": 0.588204318062138,
+      "grad_norm": 0.33360689878463745,
+      "learning_rate": 0.00019863015806375662,
+      "loss": 0.21166455745697021,
+      "mean_token_accuracy": 0.9151635468006134,
+      "num_tokens": 13779312.0,
+      "step": 1117
+    },
+    {
+      "entropy": 1.4124476611614227,
+      "epoch": 0.5887309110057926,
+      "grad_norm": 0.35278213024139404,
+      "learning_rate": 0.0001986263441182178,
+      "loss": 0.24385729432106018,
+      "mean_token_accuracy": 0.9091434925794601,
+      "num_tokens": 13791648.0,
+      "step": 1118
+    },
+    {
+      "entropy": 1.3624746203422546,
+      "epoch": 0.5892575039494471,
+      "grad_norm": 0.303867906332016,
+      "learning_rate": 0.00019862252491144954,
+      "loss": 0.20682811737060547,
+      "mean_token_accuracy": 0.9137058258056641,
+      "num_tokens": 13803984.0,
+      "step": 1119
+    },
+    {
+      "entropy": 1.4214110970497131,
+      "epoch": 0.5897840968931016,
+      "grad_norm": 0.29519587755203247,
+      "learning_rate": 0.00019861870044367844,
+      "loss": 0.1955740600824356,
+      "mean_token_accuracy": 0.9222785979509354,
+      "num_tokens": 13816320.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.332068532705307,
+      "epoch": 0.5903106898367562,
+      "grad_norm": 0.31455355882644653,
+      "learning_rate": 0.0001986148707151316,
+      "loss": 0.22069820761680603,
+      "mean_token_accuracy": 0.9075255244970322,
+      "num_tokens": 13828656.0,
+      "step": 1121
+    },
+    {
+      "entropy": 1.288136750459671,
+      "epoch": 0.5908372827804107,
+      "grad_norm": 0.3047003448009491,
+      "learning_rate": 0.0001986110357260364,
+      "loss": 0.23481610417366028,
+      "mean_token_accuracy": 0.9039241820573807,
+      "num_tokens": 13840992.0,
+      "step": 1122
+    },
+    {
+      "entropy": 1.3215467631816864,
+      "epoch": 0.5913638757240653,
+      "grad_norm": 0.2980950176715851,
+      "learning_rate": 0.0001986071954766205,
+      "loss": 0.20938631892204285,
+      "mean_token_accuracy": 0.9155066460371017,
+      "num_tokens": 13853328.0,
+      "step": 1123
+    },
+    {
+      "entropy": 1.3901076912879944,
+      "epoch": 0.5918904686677199,
+      "grad_norm": 0.3429953455924988,
+      "learning_rate": 0.0001986033499671118,
+      "loss": 0.24023953080177307,
+      "mean_token_accuracy": 0.9067379236221313,
+      "num_tokens": 13865664.0,
+      "step": 1124
+    },
+    {
+      "entropy": 1.3358671963214874,
+      "epoch": 0.5924170616113744,
+      "grad_norm": 0.2808769643306732,
+      "learning_rate": 0.00019859949919773866,
+      "loss": 0.20975762605667114,
+      "mean_token_accuracy": 0.9166984409093857,
+      "num_tokens": 13878000.0,
+      "step": 1125
+    },
+    {
+      "entropy": 1.3196144104003906,
+      "epoch": 0.592943654555029,
+      "grad_norm": 0.3139532506465912,
+      "learning_rate": 0.0001985956431687296,
+      "loss": 0.23033276200294495,
+      "mean_token_accuracy": 0.9116202145814896,
+      "num_tokens": 13890336.0,
+      "step": 1126
+    },
+    {
+      "entropy": 1.3117577731609344,
+      "epoch": 0.5934702474986835,
+      "grad_norm": 0.27732840180397034,
+      "learning_rate": 0.0001985917818803136,
+      "loss": 0.20888479053974152,
+      "mean_token_accuracy": 0.9136437624692917,
+      "num_tokens": 13902672.0,
+      "step": 1127
+    },
+    {
+      "entropy": 1.3565605282783508,
+      "epoch": 0.593996840442338,
+      "grad_norm": 0.3072391152381897,
+      "learning_rate": 0.00019858791533271993,
+      "loss": 0.2347339540719986,
+      "mean_token_accuracy": 0.9031410962343216,
+      "num_tokens": 13915008.0,
+      "step": 1128
+    },
+    {
+      "entropy": 1.3418332636356354,
+      "epoch": 0.5945234333859927,
+      "grad_norm": 0.306152880191803,
+      "learning_rate": 0.000198584043526178,
+      "loss": 0.21512483060359955,
+      "mean_token_accuracy": 0.913850799202919,
+      "num_tokens": 13927344.0,
+      "step": 1129
+    },
+    {
+      "entropy": 1.2864646911621094,
+      "epoch": 0.5950500263296472,
+      "grad_norm": 0.272443026304245,
+      "learning_rate": 0.0001985801664609177,
+      "loss": 0.19553211331367493,
+      "mean_token_accuracy": 0.9215691983699799,
+      "num_tokens": 13939680.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.3621837496757507,
+      "epoch": 0.5955766192733017,
+      "grad_norm": 0.30602437257766724,
+      "learning_rate": 0.00019857628413716923,
+      "loss": 0.2275530993938446,
+      "mean_token_accuracy": 0.9101060777902603,
+      "num_tokens": 13952016.0,
+      "step": 1131
+    },
+    {
+      "entropy": 1.3566639721393585,
+      "epoch": 0.5961032122169563,
+      "grad_norm": 0.3048968017101288,
+      "learning_rate": 0.00019857239655516302,
+      "loss": 0.20348000526428223,
+      "mean_token_accuracy": 0.9176989495754242,
+      "num_tokens": 13964352.0,
+      "step": 1132
+    },
+    {
+      "entropy": 1.3097732663154602,
+      "epoch": 0.5966298051606108,
+      "grad_norm": 0.2780429720878601,
+      "learning_rate": 0.00019856850371512987,
+      "loss": 0.21352221071720123,
+      "mean_token_accuracy": 0.9180653989315033,
+      "num_tokens": 13976688.0,
+      "step": 1133
+    },
+    {
+      "entropy": 1.3601540923118591,
+      "epoch": 0.5971563981042654,
+      "grad_norm": 0.28900468349456787,
+      "learning_rate": 0.00019856460561730086,
+      "loss": 0.21711350977420807,
+      "mean_token_accuracy": 0.9169211685657501,
+      "num_tokens": 13989024.0,
+      "step": 1134
+    },
+    {
+      "entropy": 1.3538919687271118,
+      "epoch": 0.59768299104792,
+      "grad_norm": 0.34999901056289673,
+      "learning_rate": 0.00019856070226190745,
+      "loss": 0.24132667481899261,
+      "mean_token_accuracy": 0.9076129645109177,
+      "num_tokens": 14001360.0,
+      "step": 1135
+    },
+    {
+      "entropy": 1.3285188376903534,
+      "epoch": 0.5982095839915745,
+      "grad_norm": 0.3087072968482971,
+      "learning_rate": 0.00019855679364918125,
+      "loss": 0.21430730819702148,
+      "mean_token_accuracy": 0.911715641617775,
+      "num_tokens": 14013696.0,
+      "step": 1136
+    },
+    {
+      "entropy": 1.3584640622138977,
+      "epoch": 0.5987361769352291,
+      "grad_norm": 0.34012794494628906,
+      "learning_rate": 0.00019855287977935438,
+      "loss": 0.22795352339744568,
+      "mean_token_accuracy": 0.9065137207508087,
+      "num_tokens": 14026032.0,
+      "step": 1137
+    },
+    {
+      "entropy": 1.337429255247116,
+      "epoch": 0.5992627698788836,
+      "grad_norm": 0.30005964636802673,
+      "learning_rate": 0.0001985489606526592,
+      "loss": 0.22258292138576508,
+      "mean_token_accuracy": 0.9104499816894531,
+      "num_tokens": 14038368.0,
+      "step": 1138
+    },
+    {
+      "entropy": 1.3470830917358398,
+      "epoch": 0.5997893628225381,
+      "grad_norm": 0.33669254183769226,
+      "learning_rate": 0.0001985450362693283,
+      "loss": 0.24716174602508545,
+      "mean_token_accuracy": 0.9071019142866135,
+      "num_tokens": 14050704.0,
+      "step": 1139
+    },
+    {
+      "entropy": 1.3219377994537354,
+      "epoch": 0.6003159557661928,
+      "grad_norm": 0.28407081961631775,
+      "learning_rate": 0.0001985411066295947,
+      "loss": 0.2014906108379364,
+      "mean_token_accuracy": 0.9218366146087646,
+      "num_tokens": 14063040.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.3444631695747375,
+      "epoch": 0.6008425487098473,
+      "grad_norm": 0.3870190978050232,
+      "learning_rate": 0.00019853717173369163,
+      "loss": 0.231156125664711,
+      "mean_token_accuracy": 0.9088256508111954,
+      "num_tokens": 14075376.0,
+      "step": 1141
+    },
+    {
+      "entropy": 1.308059424161911,
+      "epoch": 0.6013691416535019,
+      "grad_norm": 0.2819710373878479,
+      "learning_rate": 0.0001985332315818527,
+      "loss": 0.20217713713645935,
+      "mean_token_accuracy": 0.9211951047182083,
+      "num_tokens": 14087712.0,
+      "step": 1142
+    },
+    {
+      "entropy": 1.3603175282478333,
+      "epoch": 0.6018957345971564,
+      "grad_norm": 0.2916105389595032,
+      "learning_rate": 0.00019852928617431186,
+      "loss": 0.21151602268218994,
+      "mean_token_accuracy": 0.9143334627151489,
+      "num_tokens": 14100048.0,
+      "step": 1143
+    },
+    {
+      "entropy": 1.4045506119728088,
+      "epoch": 0.6024223275408109,
+      "grad_norm": 0.33014437556266785,
+      "learning_rate": 0.00019852533551130324,
+      "loss": 0.2289315164089203,
+      "mean_token_accuracy": 0.9064499735832214,
+      "num_tokens": 14112384.0,
+      "step": 1144
+    },
+    {
+      "entropy": 1.3394192457199097,
+      "epoch": 0.6029489204844655,
+      "grad_norm": 0.28447818756103516,
+      "learning_rate": 0.00019852137959306144,
+      "loss": 0.21860811114311218,
+      "mean_token_accuracy": 0.9082520753145218,
+      "num_tokens": 14124720.0,
+      "step": 1145
+    },
+    {
+      "entropy": 1.391217827796936,
+      "epoch": 0.6034755134281201,
+      "grad_norm": 0.28422555327415466,
+      "learning_rate": 0.00019851741841982128,
+      "loss": 0.19729138910770416,
+      "mean_token_accuracy": 0.9245949536561966,
+      "num_tokens": 14137056.0,
+      "step": 1146
+    },
+    {
+      "entropy": 1.3858129382133484,
+      "epoch": 0.6040021063717746,
+      "grad_norm": 0.29647180438041687,
+      "learning_rate": 0.00019851345199181793,
+      "loss": 0.22420856356620789,
+      "mean_token_accuracy": 0.9095530211925507,
+      "num_tokens": 14149392.0,
+      "step": 1147
+    },
+    {
+      "entropy": 1.3382080793380737,
+      "epoch": 0.6045286993154292,
+      "grad_norm": 0.3809289336204529,
+      "learning_rate": 0.0001985094803092868,
+      "loss": 0.23012962937355042,
+      "mean_token_accuracy": 0.9103907644748688,
+      "num_tokens": 14161728.0,
+      "step": 1148
+    },
+    {
+      "entropy": 1.252241313457489,
+      "epoch": 0.6050552922590837,
+      "grad_norm": 0.28908881545066833,
+      "learning_rate": 0.00019850550337246366,
+      "loss": 0.21694348752498627,
+      "mean_token_accuracy": 0.9180784374475479,
+      "num_tokens": 14174064.0,
+      "step": 1149
+    },
+    {
+      "entropy": 1.3115107417106628,
+      "epoch": 0.6055818852027383,
+      "grad_norm": 0.30919113755226135,
+      "learning_rate": 0.00019850152118158472,
+      "loss": 0.22491689026355743,
+      "mean_token_accuracy": 0.9089210480451584,
+      "num_tokens": 14186400.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.2686559855937958,
+      "epoch": 0.6061084781463928,
+      "grad_norm": 0.27780571579933167,
+      "learning_rate": 0.0001984975337368862,
+      "loss": 0.18563629686832428,
+      "mean_token_accuracy": 0.9245605766773224,
+      "num_tokens": 14198736.0,
+      "step": 1151
+    },
+    {
+      "entropy": 1.272880494594574,
+      "epoch": 0.6066350710900474,
+      "grad_norm": 0.31458860635757446,
+      "learning_rate": 0.00019849354103860495,
+      "loss": 0.21442924439907074,
+      "mean_token_accuracy": 0.9128790646791458,
+      "num_tokens": 14211072.0,
+      "step": 1152
+    },
+    {
+      "entropy": 1.2540790140628815,
+      "epoch": 0.607161664033702,
+      "grad_norm": 0.32044321298599243,
+      "learning_rate": 0.000198489543086978,
+      "loss": 0.20413605868816376,
+      "mean_token_accuracy": 0.9179537445306778,
+      "num_tokens": 14223408.0,
+      "step": 1153
+    },
+    {
+      "entropy": 1.3018059134483337,
+      "epoch": 0.6076882569773565,
+      "grad_norm": 0.3418496251106262,
+      "learning_rate": 0.00019848553988224254,
+      "loss": 0.2165253758430481,
+      "mean_token_accuracy": 0.9125253260135651,
+      "num_tokens": 14235744.0,
+      "step": 1154
+    },
+    {
+      "entropy": 1.2327319979667664,
+      "epoch": 0.608214849921011,
+      "grad_norm": 0.3597237467765808,
+      "learning_rate": 0.0001984815314246364,
+      "loss": 0.235613614320755,
+      "mean_token_accuracy": 0.9087760597467422,
+      "num_tokens": 14248080.0,
+      "step": 1155
+    },
+    {
+      "entropy": 1.3383413255214691,
+      "epoch": 0.6087414428646656,
+      "grad_norm": 0.36206018924713135,
+      "learning_rate": 0.00019847751771439738,
+      "loss": 0.2203393429517746,
+      "mean_token_accuracy": 0.9124666899442673,
+      "num_tokens": 14260416.0,
+      "step": 1156
+    },
+    {
+      "entropy": 1.2537713646888733,
+      "epoch": 0.6092680358083201,
+      "grad_norm": 0.314043253660202,
+      "learning_rate": 0.00019847349875176383,
+      "loss": 0.18635031580924988,
+      "mean_token_accuracy": 0.924136608839035,
+      "num_tokens": 14272752.0,
+      "step": 1157
+    },
+    {
+      "entropy": 1.2804813086986542,
+      "epoch": 0.6097946287519748,
+      "grad_norm": 0.3333822786808014,
+      "learning_rate": 0.00019846947453697436,
+      "loss": 0.24032224714756012,
+      "mean_token_accuracy": 0.9084908664226532,
+      "num_tokens": 14285088.0,
+      "step": 1158
+    },
+    {
+      "entropy": 1.2433983385562897,
+      "epoch": 0.6103212216956293,
+      "grad_norm": 0.3367704451084137,
+      "learning_rate": 0.0001984654450702678,
+      "loss": 0.21996420621871948,
+      "mean_token_accuracy": 0.9129030555486679,
+      "num_tokens": 14297424.0,
+      "step": 1159
+    },
+    {
+      "entropy": 1.2872591018676758,
+      "epoch": 0.6108478146392838,
+      "grad_norm": 0.3126697838306427,
+      "learning_rate": 0.00019846141035188334,
+      "loss": 0.22938281297683716,
+      "mean_token_accuracy": 0.9084669649600983,
+      "num_tokens": 14309760.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.2282376289367676,
+      "epoch": 0.6113744075829384,
+      "grad_norm": 0.3045027554035187,
+      "learning_rate": 0.0001984573703820606,
+      "loss": 0.21839705109596252,
+      "mean_token_accuracy": 0.9129580110311508,
+      "num_tokens": 14322096.0,
+      "step": 1161
+    },
+    {
+      "entropy": 1.296276867389679,
+      "epoch": 0.6119010005265929,
+      "grad_norm": 0.3050879240036011,
+      "learning_rate": 0.00019845332516103933,
+      "loss": 0.21910154819488525,
+      "mean_token_accuracy": 0.9160716086626053,
+      "num_tokens": 14334432.0,
+      "step": 1162
+    },
+    {
+      "entropy": 1.2552114129066467,
+      "epoch": 0.6124275934702474,
+      "grad_norm": 0.3606356978416443,
+      "learning_rate": 0.0001984492746890597,
+      "loss": 0.23291565477848053,
+      "mean_token_accuracy": 0.9024920612573624,
+      "num_tokens": 14346768.0,
+      "step": 1163
+    },
+    {
+      "entropy": 1.2350418269634247,
+      "epoch": 0.6129541864139021,
+      "grad_norm": 0.30588603019714355,
+      "learning_rate": 0.00019844521896636213,
+      "loss": 0.22299546003341675,
+      "mean_token_accuracy": 0.9098654091358185,
+      "num_tokens": 14359104.0,
+      "step": 1164
+    },
+    {
+      "entropy": 1.2168787717819214,
+      "epoch": 0.6134807793575566,
+      "grad_norm": 0.30036792159080505,
+      "learning_rate": 0.00019844115799318743,
+      "loss": 0.2024330198764801,
+      "mean_token_accuracy": 0.9177137911319733,
+      "num_tokens": 14371440.0,
+      "step": 1165
+    },
+    {
+      "entropy": 1.2549520134925842,
+      "epoch": 0.6140073723012112,
+      "grad_norm": 0.38305899500846863,
+      "learning_rate": 0.00019843709176977666,
+      "loss": 0.2291533648967743,
+      "mean_token_accuracy": 0.912639394402504,
+      "num_tokens": 14383776.0,
+      "step": 1166
+    },
+    {
+      "entropy": 1.235708236694336,
+      "epoch": 0.6145339652448657,
+      "grad_norm": 0.34467607736587524,
+      "learning_rate": 0.0001984330202963712,
+      "loss": 0.22410570085048676,
+      "mean_token_accuracy": 0.9121714979410172,
+      "num_tokens": 14396112.0,
+      "step": 1167
+    },
+    {
+      "entropy": 1.1345283687114716,
+      "epoch": 0.6150605581885202,
+      "grad_norm": 0.3024910092353821,
+      "learning_rate": 0.0001984289435732127,
+      "loss": 0.22202812135219574,
+      "mean_token_accuracy": 0.913068488240242,
+      "num_tokens": 14408448.0,
+      "step": 1168
+    },
+    {
+      "entropy": 1.154352843761444,
+      "epoch": 0.6155871511321749,
+      "grad_norm": 0.3543850779533386,
+      "learning_rate": 0.00019842486160054327,
+      "loss": 0.24754247069358826,
+      "mean_token_accuracy": 0.9041948169469833,
+      "num_tokens": 14420784.0,
+      "step": 1169
+    },
+    {
+      "entropy": 1.1256769597530365,
+      "epoch": 0.6161137440758294,
+      "grad_norm": 0.31246209144592285,
+      "learning_rate": 0.00019842077437860517,
+      "loss": 0.23024949431419373,
+      "mean_token_accuracy": 0.9097237884998322,
+      "num_tokens": 14433120.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.2304538488388062,
+      "epoch": 0.616640337019484,
+      "grad_norm": 0.35418546199798584,
+      "learning_rate": 0.00019841668190764106,
+      "loss": 0.23257943987846375,
+      "mean_token_accuracy": 0.9058277457952499,
+      "num_tokens": 14445456.0,
+      "step": 1171
+    },
+    {
+      "entropy": 1.1736994087696075,
+      "epoch": 0.6171669299631385,
+      "grad_norm": 0.29421883821487427,
+      "learning_rate": 0.0001984125841878939,
+      "loss": 0.2265184223651886,
+      "mean_token_accuracy": 0.9127582311630249,
+      "num_tokens": 14457792.0,
+      "step": 1172
+    },
+    {
+      "entropy": 1.210179328918457,
+      "epoch": 0.617693522906793,
+      "grad_norm": 0.3278239667415619,
+      "learning_rate": 0.0001984084812196069,
+      "loss": 0.2220735400915146,
+      "mean_token_accuracy": 0.9109156280755997,
+      "num_tokens": 14470128.0,
+      "step": 1173
+    },
+    {
+      "entropy": 1.1588793694972992,
+      "epoch": 0.6182201158504476,
+      "grad_norm": 0.27815452218055725,
+      "learning_rate": 0.00019840437300302366,
+      "loss": 0.22251342236995697,
+      "mean_token_accuracy": 0.9095259606838226,
+      "num_tokens": 14482464.0,
+      "step": 1174
+    },
+    {
+      "entropy": 1.1791687309741974,
+      "epoch": 0.6187467087941022,
+      "grad_norm": 0.2694774270057678,
+      "learning_rate": 0.00019840025953838804,
+      "loss": 0.2061459869146347,
+      "mean_token_accuracy": 0.9203435033559799,
+      "num_tokens": 14494800.0,
+      "step": 1175
+    },
+    {
+      "entropy": 1.1984158754348755,
+      "epoch": 0.6192733017377567,
+      "grad_norm": 0.29139313101768494,
+      "learning_rate": 0.00019839614082594424,
+      "loss": 0.22982501983642578,
+      "mean_token_accuracy": 0.911991760134697,
+      "num_tokens": 14507136.0,
+      "step": 1176
+    },
+    {
+      "entropy": 1.1957634687423706,
+      "epoch": 0.6197998946814113,
+      "grad_norm": 0.2952525019645691,
+      "learning_rate": 0.0001983920168659368,
+      "loss": 0.20150825381278992,
+      "mean_token_accuracy": 0.9190460294485092,
+      "num_tokens": 14519472.0,
+      "step": 1177
+    },
+    {
+      "entropy": 1.2356001436710358,
+      "epoch": 0.6203264876250658,
+      "grad_norm": 0.3485545516014099,
+      "learning_rate": 0.0001983878876586105,
+      "loss": 0.22730165719985962,
+      "mean_token_accuracy": 0.9079895615577698,
+      "num_tokens": 14531808.0,
+      "step": 1178
+    },
+    {
+      "entropy": 1.2191669344902039,
+      "epoch": 0.6208530805687204,
+      "grad_norm": 0.31337442994117737,
+      "learning_rate": 0.00019838375320421046,
+      "loss": 0.2090841382741928,
+      "mean_token_accuracy": 0.9179579317569733,
+      "num_tokens": 14544144.0,
+      "step": 1179
+    },
+    {
+      "entropy": 1.1627593338489532,
+      "epoch": 0.6213796735123749,
+      "grad_norm": 0.3373863995075226,
+      "learning_rate": 0.00019837961350298213,
+      "loss": 0.22159543633460999,
+      "mean_token_accuracy": 0.9035765677690506,
+      "num_tokens": 14556480.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.244479089975357,
+      "epoch": 0.6219062664560295,
+      "grad_norm": 0.2939426898956299,
+      "learning_rate": 0.0001983754685551713,
+      "loss": 0.20940116047859192,
+      "mean_token_accuracy": 0.9177102446556091,
+      "num_tokens": 14568816.0,
+      "step": 1181
+    },
+    {
+      "entropy": 1.2316089570522308,
+      "epoch": 0.6224328593996841,
+      "grad_norm": 0.3053775727748871,
+      "learning_rate": 0.00019837131836102396,
+      "loss": 0.21764273941516876,
+      "mean_token_accuracy": 0.9150272458791733,
+      "num_tokens": 14581152.0,
+      "step": 1182
+    },
+    {
+      "entropy": 1.2752362787723541,
+      "epoch": 0.6229594523433386,
+      "grad_norm": 0.3304329812526703,
+      "learning_rate": 0.00019836716292078647,
+      "loss": 0.23420026898384094,
+      "mean_token_accuracy": 0.9085260033607483,
+      "num_tokens": 14593488.0,
+      "step": 1183
+    },
+    {
+      "entropy": 1.2659085094928741,
+      "epoch": 0.6234860452869931,
+      "grad_norm": 0.3893950879573822,
+      "learning_rate": 0.00019836300223470562,
+      "loss": 0.2608282268047333,
+      "mean_token_accuracy": 0.9021032303571701,
+      "num_tokens": 14605824.0,
+      "step": 1184
+    },
+    {
+      "entropy": 1.3032302856445312,
+      "epoch": 0.6240126382306477,
+      "grad_norm": 0.3007699251174927,
+      "learning_rate": 0.00019835883630302836,
+      "loss": 0.20315539836883545,
+      "mean_token_accuracy": 0.9167050868272781,
+      "num_tokens": 14618160.0,
+      "step": 1185
+    },
+    {
+      "entropy": 1.2084311544895172,
+      "epoch": 0.6245392311743022,
+      "grad_norm": 0.32948681712150574,
+      "learning_rate": 0.00019835466512600197,
+      "loss": 0.23484407365322113,
+      "mean_token_accuracy": 0.9104520976543427,
+      "num_tokens": 14630496.0,
+      "step": 1186
+    },
+    {
+      "entropy": 1.2006751894950867,
+      "epoch": 0.6250658241179569,
+      "grad_norm": 0.2981775104999542,
+      "learning_rate": 0.00019835048870387405,
+      "loss": 0.22286085784435272,
+      "mean_token_accuracy": 0.9095976203680038,
+      "num_tokens": 14642832.0,
+      "step": 1187
+    },
+    {
+      "entropy": 1.3096867799758911,
+      "epoch": 0.6255924170616114,
+      "grad_norm": 0.3300210237503052,
+      "learning_rate": 0.0001983463070368926,
+      "loss": 0.22102200984954834,
+      "mean_token_accuracy": 0.9119229018688202,
+      "num_tokens": 14655168.0,
+      "step": 1188
+    },
+    {
+      "entropy": 1.2840961813926697,
+      "epoch": 0.6261190100052659,
+      "grad_norm": 0.31997406482696533,
+      "learning_rate": 0.00019834212012530579,
+      "loss": 0.23797491192817688,
+      "mean_token_accuracy": 0.9043692797422409,
+      "num_tokens": 14667504.0,
+      "step": 1189
+    },
+    {
+      "entropy": 1.2393836975097656,
+      "epoch": 0.6266456029489205,
+      "grad_norm": 0.30698251724243164,
+      "learning_rate": 0.00019833792796936222,
+      "loss": 0.20287004113197327,
+      "mean_token_accuracy": 0.9166252613067627,
+      "num_tokens": 14679840.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.3439719676971436,
+      "epoch": 0.627172195892575,
+      "grad_norm": 0.32229262590408325,
+      "learning_rate": 0.00019833373056931072,
+      "loss": 0.23332646489143372,
+      "mean_token_accuracy": 0.9090562462806702,
+      "num_tokens": 14692176.0,
+      "step": 1191
+    },
+    {
+      "entropy": 1.2390953600406647,
+      "epoch": 0.6276987888362296,
+      "grad_norm": 0.3261661231517792,
+      "learning_rate": 0.00019832952792540054,
+      "loss": 0.22384727001190186,
+      "mean_token_accuracy": 0.9121635258197784,
+      "num_tokens": 14704512.0,
+      "step": 1192
+    },
+    {
+      "entropy": 1.3028371036052704,
+      "epoch": 0.6282253817798842,
+      "grad_norm": 0.31324324011802673,
+      "learning_rate": 0.00019832532003788108,
+      "loss": 0.2185819447040558,
+      "mean_token_accuracy": 0.9167534857988358,
+      "num_tokens": 14716848.0,
+      "step": 1193
+    },
+    {
+      "entropy": 1.1885470747947693,
+      "epoch": 0.6287519747235387,
+      "grad_norm": 0.46663784980773926,
+      "learning_rate": 0.00019832110690700213,
+      "loss": 0.21501871943473816,
+      "mean_token_accuracy": 0.9109776020050049,
+      "num_tokens": 14729184.0,
+      "step": 1194
+    },
+    {
+      "entropy": 1.2453204691410065,
+      "epoch": 0.6292785676671933,
+      "grad_norm": 0.3183780014514923,
+      "learning_rate": 0.00019831688853301388,
+      "loss": 0.22543299198150635,
+      "mean_token_accuracy": 0.910496860742569,
+      "num_tokens": 14741520.0,
+      "step": 1195
+    },
+    {
+      "entropy": 1.2583057284355164,
+      "epoch": 0.6298051606108478,
+      "grad_norm": 0.3381344676017761,
+      "learning_rate": 0.0001983126649161667,
+      "loss": 0.214624285697937,
+      "mean_token_accuracy": 0.9137195497751236,
+      "num_tokens": 14753856.0,
+      "step": 1196
+    },
+    {
+      "entropy": 1.2429248094558716,
+      "epoch": 0.6303317535545023,
+      "grad_norm": 0.30227339267730713,
+      "learning_rate": 0.00019830843605671131,
+      "loss": 0.2062695324420929,
+      "mean_token_accuracy": 0.9234135448932648,
+      "num_tokens": 14766192.0,
+      "step": 1197
+    },
+    {
+      "entropy": 1.220680981874466,
+      "epoch": 0.630858346498157,
+      "grad_norm": 0.3083505630493164,
+      "learning_rate": 0.00019830420195489877,
+      "loss": 0.2090727537870407,
+      "mean_token_accuracy": 0.9170975387096405,
+      "num_tokens": 14778528.0,
+      "step": 1198
+    },
+    {
+      "entropy": 1.248794138431549,
+      "epoch": 0.6313849394418115,
+      "grad_norm": 0.3802277743816376,
+      "learning_rate": 0.00019829996261098046,
+      "loss": 0.2367159128189087,
+      "mean_token_accuracy": 0.9018937945365906,
+      "num_tokens": 14790864.0,
+      "step": 1199
+    },
+    {
+      "entropy": 1.3259137272834778,
+      "epoch": 0.631911532385466,
+      "grad_norm": 0.414235919713974,
+      "learning_rate": 0.00019829571802520803,
+      "loss": 0.2376767247915268,
+      "mean_token_accuracy": 0.9016633927822113,
+      "num_tokens": 14803200.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.3644030392169952,
+      "epoch": 0.6324381253291206,
+      "grad_norm": 0.3521367013454437,
+      "learning_rate": 0.0001982914681978334,
+      "loss": 0.23597122728824615,
+      "mean_token_accuracy": 0.9043200761079788,
+      "num_tokens": 14815536.0,
+      "step": 1201
+    },
+    {
+      "entropy": 1.3481078147888184,
+      "epoch": 0.6329647182727751,
+      "grad_norm": 0.2994113266468048,
+      "learning_rate": 0.000198287213129109,
+      "loss": 0.21383647620677948,
+      "mean_token_accuracy": 0.9209251701831818,
+      "num_tokens": 14827872.0,
+      "step": 1202
+    },
+    {
+      "entropy": 1.3808801472187042,
+      "epoch": 0.6334913112164297,
+      "grad_norm": 0.2982345223426819,
+      "learning_rate": 0.00019828295281928727,
+      "loss": 0.2155206948518753,
+      "mean_token_accuracy": 0.9116237312555313,
+      "num_tokens": 14840208.0,
+      "step": 1203
+    },
+    {
+      "entropy": 1.418733924627304,
+      "epoch": 0.6340179041600843,
+      "grad_norm": 0.3023967742919922,
+      "learning_rate": 0.00019827868726862117,
+      "loss": 0.23393575847148895,
+      "mean_token_accuracy": 0.9095605909824371,
+      "num_tokens": 14852544.0,
+      "step": 1204
+    },
+    {
+      "entropy": 1.5514957010746002,
+      "epoch": 0.6345444971037388,
+      "grad_norm": 0.40304750204086304,
+      "learning_rate": 0.00019827441647736393,
+      "loss": 0.26255497336387634,
+      "mean_token_accuracy": 0.9011336117982864,
+      "num_tokens": 14864880.0,
+      "step": 1205
+    },
+    {
+      "entropy": 1.4219785034656525,
+      "epoch": 0.6350710900473934,
+      "grad_norm": 0.291746586561203,
+      "learning_rate": 0.00019827014044576912,
+      "loss": 0.21620029211044312,
+      "mean_token_accuracy": 0.9115499705076218,
+      "num_tokens": 14877216.0,
+      "step": 1206
+    },
+    {
+      "entropy": 1.3420428931713104,
+      "epoch": 0.6355976829910479,
+      "grad_norm": 0.3108786940574646,
+      "learning_rate": 0.0001982658591740905,
+      "loss": 0.23703286051750183,
+      "mean_token_accuracy": 0.9016107618808746,
+      "num_tokens": 14889552.0,
+      "step": 1207
+    },
+    {
+      "entropy": 1.4392370581626892,
+      "epoch": 0.6361242759347024,
+      "grad_norm": 0.27034372091293335,
+      "learning_rate": 0.00019826157266258232,
+      "loss": 0.21097946166992188,
+      "mean_token_accuracy": 0.9190766364336014,
+      "num_tokens": 14901888.0,
+      "step": 1208
+    },
+    {
+      "entropy": 1.4509558379650116,
+      "epoch": 0.636650868878357,
+      "grad_norm": 0.29659944772720337,
+      "learning_rate": 0.000198257280911499,
+      "loss": 0.22469229996204376,
+      "mean_token_accuracy": 0.913391038775444,
+      "num_tokens": 14914224.0,
+      "step": 1209
+    },
+    {
+      "entropy": 1.3904311954975128,
+      "epoch": 0.6371774618220116,
+      "grad_norm": 0.2719598412513733,
+      "learning_rate": 0.00019825298392109529,
+      "loss": 0.18848726153373718,
+      "mean_token_accuracy": 0.9262688905000687,
+      "num_tokens": 14926560.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.4359396994113922,
+      "epoch": 0.6377040547656662,
+      "grad_norm": 0.3138201832771301,
+      "learning_rate": 0.00019824868169162631,
+      "loss": 0.24162685871124268,
+      "mean_token_accuracy": 0.9020742177963257,
+      "num_tokens": 14938896.0,
+      "step": 1211
+    },
+    {
+      "entropy": 1.3890084326267242,
+      "epoch": 0.6382306477093207,
+      "grad_norm": 0.2986134886741638,
+      "learning_rate": 0.00019824437422334744,
+      "loss": 0.22484566271305084,
+      "mean_token_accuracy": 0.9077206701040268,
+      "num_tokens": 14951232.0,
+      "step": 1212
+    },
+    {
+      "entropy": 1.378806233406067,
+      "epoch": 0.6387572406529752,
+      "grad_norm": 0.29134130477905273,
+      "learning_rate": 0.00019824006151651443,
+      "loss": 0.2171209454536438,
+      "mean_token_accuracy": 0.9133529961109161,
+      "num_tokens": 14963568.0,
+      "step": 1213
+    },
+    {
+      "entropy": 1.363432914018631,
+      "epoch": 0.6392838335966298,
+      "grad_norm": 0.3041214346885681,
+      "learning_rate": 0.00019823574357138323,
+      "loss": 0.24093735218048096,
+      "mean_token_accuracy": 0.903191551566124,
+      "num_tokens": 14975904.0,
+      "step": 1214
+    },
+    {
+      "entropy": 1.432945430278778,
+      "epoch": 0.6398104265402843,
+      "grad_norm": 0.30195045471191406,
+      "learning_rate": 0.00019823142038821027,
+      "loss": 0.20907755196094513,
+      "mean_token_accuracy": 0.9161103218793869,
+      "num_tokens": 14988240.0,
+      "step": 1215
+    },
+    {
+      "entropy": 1.3210029900074005,
+      "epoch": 0.640337019483939,
+      "grad_norm": 0.29575979709625244,
+      "learning_rate": 0.00019822709196725208,
+      "loss": 0.21090644598007202,
+      "mean_token_accuracy": 0.9149498790502548,
+      "num_tokens": 15000576.0,
+      "step": 1216
+    },
+    {
+      "entropy": 1.3699910044670105,
+      "epoch": 0.6408636124275935,
+      "grad_norm": 0.2982396185398102,
+      "learning_rate": 0.00019822275830876568,
+      "loss": 0.2054290473461151,
+      "mean_token_accuracy": 0.9217698276042938,
+      "num_tokens": 15012912.0,
+      "step": 1217
+    },
+    {
+      "entropy": 1.41866934299469,
+      "epoch": 0.641390205371248,
+      "grad_norm": 0.3208104968070984,
+      "learning_rate": 0.00019821841941300834,
+      "loss": 0.24424780905246735,
+      "mean_token_accuracy": 0.9118457287549973,
+      "num_tokens": 15025248.0,
+      "step": 1218
+    },
+    {
+      "entropy": 1.399036020040512,
+      "epoch": 0.6419167983149026,
+      "grad_norm": 0.3094969689846039,
+      "learning_rate": 0.00019821407528023758,
+      "loss": 0.23681235313415527,
+      "mean_token_accuracy": 0.9031997323036194,
+      "num_tokens": 15037584.0,
+      "step": 1219
+    },
+    {
+      "entropy": 1.3915925920009613,
+      "epoch": 0.6424433912585571,
+      "grad_norm": 0.2973760962486267,
+      "learning_rate": 0.00019820972591071131,
+      "loss": 0.19640693068504333,
+      "mean_token_accuracy": 0.9181762486696243,
+      "num_tokens": 15049920.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.4664425551891327,
+      "epoch": 0.6429699842022117,
+      "grad_norm": 0.7199912667274475,
+      "learning_rate": 0.0001982053713046878,
+      "loss": 0.2480940818786621,
+      "mean_token_accuracy": 0.9039055407047272,
+      "num_tokens": 15062256.0,
+      "step": 1221
+    },
+    {
+      "entropy": 1.4178178012371063,
+      "epoch": 0.6434965771458663,
+      "grad_norm": 0.3677431643009186,
+      "learning_rate": 0.00019820101146242547,
+      "loss": 0.22347916662693024,
+      "mean_token_accuracy": 0.9104437232017517,
+      "num_tokens": 15074592.0,
+      "step": 1222
+    },
+    {
+      "entropy": 1.3942118883132935,
+      "epoch": 0.6440231700895208,
+      "grad_norm": 0.3363230228424072,
+      "learning_rate": 0.00019819664638418314,
+      "loss": 0.23758456110954285,
+      "mean_token_accuracy": 0.9113318026065826,
+      "num_tokens": 15086928.0,
+      "step": 1223
+    },
+    {
+      "entropy": 1.3070717453956604,
+      "epoch": 0.6445497630331753,
+      "grad_norm": 0.3119392693042755,
+      "learning_rate": 0.00019819227607021996,
+      "loss": 0.1808784306049347,
+      "mean_token_accuracy": 0.9277928322553635,
+      "num_tokens": 15099264.0,
+      "step": 1224
+    },
+    {
+      "entropy": 1.3485881388187408,
+      "epoch": 0.6450763559768299,
+      "grad_norm": 0.2953029274940491,
+      "learning_rate": 0.00019818790052079542,
+      "loss": 0.21690060198307037,
+      "mean_token_accuracy": 0.9107033312320709,
+      "num_tokens": 15111600.0,
+      "step": 1225
+    },
+    {
+      "entropy": 1.3310377597808838,
+      "epoch": 0.6456029489204844,
+      "grad_norm": 0.30317792296409607,
+      "learning_rate": 0.0001981835197361692,
+      "loss": 0.21382620930671692,
+      "mean_token_accuracy": 0.9134241938591003,
+      "num_tokens": 15123936.0,
+      "step": 1226
+    },
+    {
+      "entropy": 1.3547466397285461,
+      "epoch": 0.6461295418641391,
+      "grad_norm": 0.31380537152290344,
+      "learning_rate": 0.00019817913371660136,
+      "loss": 0.23309272527694702,
+      "mean_token_accuracy": 0.9060655981302261,
+      "num_tokens": 15136272.0,
+      "step": 1227
+    },
+    {
+      "entropy": 1.359354555606842,
+      "epoch": 0.6466561348077936,
+      "grad_norm": 0.3240332007408142,
+      "learning_rate": 0.00019817474246235233,
+      "loss": 0.2206256240606308,
+      "mean_token_accuracy": 0.9170451760292053,
+      "num_tokens": 15148608.0,
+      "step": 1228
+    },
+    {
+      "entropy": 1.3810060322284698,
+      "epoch": 0.6471827277514481,
+      "grad_norm": 0.28998929262161255,
+      "learning_rate": 0.00019817034597368276,
+      "loss": 0.20593759417533875,
+      "mean_token_accuracy": 0.9143964052200317,
+      "num_tokens": 15160944.0,
+      "step": 1229
+    },
+    {
+      "entropy": 1.3853793740272522,
+      "epoch": 0.6477093206951027,
+      "grad_norm": 0.3167085349559784,
+      "learning_rate": 0.00019816594425085365,
+      "loss": 0.214578315615654,
+      "mean_token_accuracy": 0.9154220819473267,
+      "num_tokens": 15173280.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.3485651314258575,
+      "epoch": 0.6482359136387572,
+      "grad_norm": 0.32996588945388794,
+      "learning_rate": 0.00019816153729412627,
+      "loss": 0.227751225233078,
+      "mean_token_accuracy": 0.9076825082302094,
+      "num_tokens": 15185616.0,
+      "step": 1231
+    },
+    {
+      "entropy": 1.2957139611244202,
+      "epoch": 0.6487625065824117,
+      "grad_norm": 0.2826898396015167,
+      "learning_rate": 0.0001981571251037623,
+      "loss": 0.2006268948316574,
+      "mean_token_accuracy": 0.9202533811330795,
+      "num_tokens": 15197952.0,
+      "step": 1232
+    },
+    {
+      "entropy": 1.42320716381073,
+      "epoch": 0.6492890995260664,
+      "grad_norm": 0.3413523733615875,
+      "learning_rate": 0.00019815270768002357,
+      "loss": 0.2385195791721344,
+      "mean_token_accuracy": 0.9027537554502487,
+      "num_tokens": 15210288.0,
+      "step": 1233
+    },
+    {
+      "entropy": 1.3331094086170197,
+      "epoch": 0.6498156924697209,
+      "grad_norm": 0.3069290220737457,
+      "learning_rate": 0.00019814828502317245,
+      "loss": 0.2170781046152115,
+      "mean_token_accuracy": 0.9153832644224167,
+      "num_tokens": 15222624.0,
+      "step": 1234
+    },
+    {
+      "entropy": 1.365331083536148,
+      "epoch": 0.6503422854133755,
+      "grad_norm": 0.2763271629810333,
+      "learning_rate": 0.0001981438571334714,
+      "loss": 0.21132177114486694,
+      "mean_token_accuracy": 0.9158383458852768,
+      "num_tokens": 15234960.0,
+      "step": 1235
+    },
+    {
+      "entropy": 1.3776979446411133,
+      "epoch": 0.65086887835703,
+      "grad_norm": 0.3232009708881378,
+      "learning_rate": 0.0001981394240111833,
+      "loss": 0.23016487061977386,
+      "mean_token_accuracy": 0.9127575606107712,
+      "num_tokens": 15247296.0,
+      "step": 1236
+    },
+    {
+      "entropy": 1.42231023311615,
+      "epoch": 0.6513954713006845,
+      "grad_norm": 0.2981938421726227,
+      "learning_rate": 0.00019813498565657127,
+      "loss": 0.22353443503379822,
+      "mean_token_accuracy": 0.9140705019235611,
+      "num_tokens": 15259632.0,
+      "step": 1237
+    },
+    {
+      "entropy": 1.4192762076854706,
+      "epoch": 0.6519220642443391,
+      "grad_norm": 0.30204325914382935,
+      "learning_rate": 0.00019813054206989884,
+      "loss": 0.22629618644714355,
+      "mean_token_accuracy": 0.9069414585828781,
+      "num_tokens": 15271968.0,
+      "step": 1238
+    },
+    {
+      "entropy": 1.4479241073131561,
+      "epoch": 0.6524486571879937,
+      "grad_norm": 0.2993687689304352,
+      "learning_rate": 0.00019812609325142982,
+      "loss": 0.22675678133964539,
+      "mean_token_accuracy": 0.9094991534948349,
+      "num_tokens": 15284304.0,
+      "step": 1239
+    },
+    {
+      "entropy": 1.4620944261550903,
+      "epoch": 0.6529752501316483,
+      "grad_norm": 0.3134925961494446,
+      "learning_rate": 0.00019812163920142827,
+      "loss": 0.22805798053741455,
+      "mean_token_accuracy": 0.9107698351144791,
+      "num_tokens": 15296640.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.495926171541214,
+      "epoch": 0.6535018430753028,
+      "grad_norm": 0.3140438199043274,
+      "learning_rate": 0.00019811717992015862,
+      "loss": 0.2086748480796814,
+      "mean_token_accuracy": 0.9102585017681122,
+      "num_tokens": 15308976.0,
+      "step": 1241
+    },
+    {
+      "entropy": 1.4446808993816376,
+      "epoch": 0.6540284360189573,
+      "grad_norm": 0.30744868516921997,
+      "learning_rate": 0.00019811271540788556,
+      "loss": 0.22943398356437683,
+      "mean_token_accuracy": 0.9074350297451019,
+      "num_tokens": 15321312.0,
+      "step": 1242
+    },
+    {
+      "entropy": 1.5100261270999908,
+      "epoch": 0.6545550289626119,
+      "grad_norm": 0.3204497694969177,
+      "learning_rate": 0.00019810824566487417,
+      "loss": 0.22844550013542175,
+      "mean_token_accuracy": 0.9073818922042847,
+      "num_tokens": 15333648.0,
+      "step": 1243
+    },
+    {
+      "entropy": 1.4075362086296082,
+      "epoch": 0.6550816219062665,
+      "grad_norm": 0.32697567343711853,
+      "learning_rate": 0.00019810377069138976,
+      "loss": 0.23277214169502258,
+      "mean_token_accuracy": 0.9124523550271988,
+      "num_tokens": 15345984.0,
+      "step": 1244
+    },
+    {
+      "entropy": 1.474784404039383,
+      "epoch": 0.655608214849921,
+      "grad_norm": 0.31703367829322815,
+      "learning_rate": 0.000198099290487698,
+      "loss": 0.2128317952156067,
+      "mean_token_accuracy": 0.9103437811136246,
+      "num_tokens": 15358320.0,
+      "step": 1245
+    },
+    {
+      "entropy": 1.3723844587802887,
+      "epoch": 0.6561348077935756,
+      "grad_norm": 0.3136475384235382,
+      "learning_rate": 0.0001980948050540648,
+      "loss": 0.2106158286333084,
+      "mean_token_accuracy": 0.9139569699764252,
+      "num_tokens": 15370656.0,
+      "step": 1246
+    },
+    {
+      "entropy": 1.4297202229499817,
+      "epoch": 0.6566614007372301,
+      "grad_norm": 0.31566086411476135,
+      "learning_rate": 0.00019809031439075653,
+      "loss": 0.22323954105377197,
+      "mean_token_accuracy": 0.912299782037735,
+      "num_tokens": 15382992.0,
+      "step": 1247
+    },
+    {
+      "entropy": 1.404792070388794,
+      "epoch": 0.6571879936808847,
+      "grad_norm": 0.30582812428474426,
+      "learning_rate": 0.00019808581849803972,
+      "loss": 0.19535206258296967,
+      "mean_token_accuracy": 0.926054835319519,
+      "num_tokens": 15395328.0,
+      "step": 1248
+    },
+    {
+      "entropy": 1.365869253873825,
+      "epoch": 0.6577145866245392,
+      "grad_norm": 0.32905054092407227,
+      "learning_rate": 0.00019808131737618126,
+      "loss": 0.206635981798172,
+      "mean_token_accuracy": 0.9179205745458603,
+      "num_tokens": 15407664.0,
+      "step": 1249
+    },
+    {
+      "entropy": 1.4380927085876465,
+      "epoch": 0.6582411795681938,
+      "grad_norm": 0.3363628685474396,
+      "learning_rate": 0.00019807681102544834,
+      "loss": 0.23559972643852234,
+      "mean_token_accuracy": 0.9073036760091782,
+      "num_tokens": 15420000.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.4528064727783203,
+      "epoch": 0.6587677725118484,
+      "grad_norm": 0.3616321384906769,
+      "learning_rate": 0.00019807229944610854,
+      "loss": 0.23041245341300964,
+      "mean_token_accuracy": 0.9012827575206757,
+      "num_tokens": 15432336.0,
+      "step": 1251
+    },
+    {
+      "entropy": 1.4460172951221466,
+      "epoch": 0.6592943654555029,
+      "grad_norm": 0.305955171585083,
+      "learning_rate": 0.00019806778263842964,
+      "loss": 0.21666109561920166,
+      "mean_token_accuracy": 0.9198573529720306,
+      "num_tokens": 15444672.0,
+      "step": 1252
+    },
+    {
+      "entropy": 1.500756323337555,
+      "epoch": 0.6598209583991574,
+      "grad_norm": 0.29015445709228516,
+      "learning_rate": 0.00019806326060267977,
+      "loss": 0.21642570197582245,
+      "mean_token_accuracy": 0.9154843240976334,
+      "num_tokens": 15457008.0,
+      "step": 1253
+    },
+    {
+      "entropy": 1.4731870293617249,
+      "epoch": 0.660347551342812,
+      "grad_norm": 0.3006648123264313,
+      "learning_rate": 0.00019805873333912739,
+      "loss": 0.20880235731601715,
+      "mean_token_accuracy": 0.924048438668251,
+      "num_tokens": 15469344.0,
+      "step": 1254
+    },
+    {
+      "entropy": 1.5191810727119446,
+      "epoch": 0.6608741442864665,
+      "grad_norm": 0.290721595287323,
+      "learning_rate": 0.00019805420084804124,
+      "loss": 0.21848249435424805,
+      "mean_token_accuracy": 0.9141736477613449,
+      "num_tokens": 15481680.0,
+      "step": 1255
+    },
+    {
+      "entropy": 1.521433562040329,
+      "epoch": 0.6614007372301212,
+      "grad_norm": 0.3009195327758789,
+      "learning_rate": 0.00019804966312969046,
+      "loss": 0.20721524953842163,
+      "mean_token_accuracy": 0.9165780544281006,
+      "num_tokens": 15494016.0,
+      "step": 1256
+    },
+    {
+      "entropy": 1.5230248272418976,
+      "epoch": 0.6619273301737757,
+      "grad_norm": 0.2907300293445587,
+      "learning_rate": 0.00019804512018434432,
+      "loss": 0.2220079004764557,
+      "mean_token_accuracy": 0.9127085506916046,
+      "num_tokens": 15506352.0,
+      "step": 1257
+    },
+    {
+      "entropy": 1.4415357112884521,
+      "epoch": 0.6624539231174302,
+      "grad_norm": 0.27609995007514954,
+      "learning_rate": 0.00019804057201227259,
+      "loss": 0.19916732609272003,
+      "mean_token_accuracy": 0.9211323857307434,
+      "num_tokens": 15518688.0,
+      "step": 1258
+    },
+    {
+      "entropy": 1.5071959793567657,
+      "epoch": 0.6629805160610848,
+      "grad_norm": 0.3255380690097809,
+      "learning_rate": 0.0001980360186137452,
+      "loss": 0.23832152783870697,
+      "mean_token_accuracy": 0.9030174165964127,
+      "num_tokens": 15531024.0,
+      "step": 1259
+    },
+    {
+      "entropy": 1.544082134962082,
+      "epoch": 0.6635071090047393,
+      "grad_norm": 0.31669536232948303,
+      "learning_rate": 0.00019803145998903253,
+      "loss": 0.2193489968776703,
+      "mean_token_accuracy": 0.9140938073396683,
+      "num_tokens": 15543360.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.5186286568641663,
+      "epoch": 0.6640337019483938,
+      "grad_norm": 0.3087945282459259,
+      "learning_rate": 0.00019802689613840514,
+      "loss": 0.2145516723394394,
+      "mean_token_accuracy": 0.9134328663349152,
+      "num_tokens": 15555696.0,
+      "step": 1261
+    },
+    {
+      "entropy": 1.4552082121372223,
+      "epoch": 0.6645602948920485,
+      "grad_norm": 0.3144809901714325,
+      "learning_rate": 0.000198022327062134,
+      "loss": 0.22542014718055725,
+      "mean_token_accuracy": 0.9080185741186142,
+      "num_tokens": 15568032.0,
+      "step": 1262
+    },
+    {
+      "entropy": 1.4759101569652557,
+      "epoch": 0.665086887835703,
+      "grad_norm": 0.28055956959724426,
+      "learning_rate": 0.00019801775276049035,
+      "loss": 0.19411805272102356,
+      "mean_token_accuracy": 0.9219205677509308,
+      "num_tokens": 15580368.0,
+      "step": 1263
+    },
+    {
+      "entropy": 1.5172317326068878,
+      "epoch": 0.6656134807793576,
+      "grad_norm": 0.28847187757492065,
+      "learning_rate": 0.00019801317323374574,
+      "loss": 0.1965804398059845,
+      "mean_token_accuracy": 0.9217233210802078,
+      "num_tokens": 15592704.0,
+      "step": 1264
+    },
+    {
+      "entropy": 1.4631735682487488,
+      "epoch": 0.6661400737230121,
+      "grad_norm": 0.3277401924133301,
+      "learning_rate": 0.000198008588482172,
+      "loss": 0.23770709335803986,
+      "mean_token_accuracy": 0.9039840698242188,
+      "num_tokens": 15605040.0,
+      "step": 1265
+    },
+    {
+      "entropy": 1.5213152170181274,
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.31430521607398987,
+      "learning_rate": 0.00019800399850604129,
+      "loss": 0.23058706521987915,
+      "mean_token_accuracy": 0.9074577987194061,
+      "num_tokens": 15617376.0,
+      "step": 1266
+    },
+    {
+      "entropy": 1.4705300331115723,
+      "epoch": 0.6671932596103213,
+      "grad_norm": 0.3256817162036896,
+      "learning_rate": 0.0001979994033056261,
+      "loss": 0.25320321321487427,
+      "mean_token_accuracy": 0.9007576107978821,
+      "num_tokens": 15629712.0,
+      "step": 1267
+    },
+    {
+      "entropy": 1.530697613954544,
+      "epoch": 0.6677198525539758,
+      "grad_norm": 0.30975228548049927,
+      "learning_rate": 0.00019799480288119927,
+      "loss": 0.23081722855567932,
+      "mean_token_accuracy": 0.9080140292644501,
+      "num_tokens": 15642048.0,
+      "step": 1268
+    },
+    {
+      "entropy": 1.495768964290619,
+      "epoch": 0.6682464454976303,
+      "grad_norm": 0.29344475269317627,
+      "learning_rate": 0.00019799019723303388,
+      "loss": 0.2015610784292221,
+      "mean_token_accuracy": 0.9178611040115356,
+      "num_tokens": 15654384.0,
+      "step": 1269
+    },
+    {
+      "entropy": 1.5803856551647186,
+      "epoch": 0.6687730384412849,
+      "grad_norm": 0.3023964464664459,
+      "learning_rate": 0.00019798558636140333,
+      "loss": 0.2174011915922165,
+      "mean_token_accuracy": 0.9140703827142715,
+      "num_tokens": 15666720.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.4897505939006805,
+      "epoch": 0.6692996313849394,
+      "grad_norm": 0.2858726978302002,
+      "learning_rate": 0.0001979809702665813,
+      "loss": 0.20024171471595764,
+      "mean_token_accuracy": 0.9191962629556656,
+      "num_tokens": 15679056.0,
+      "step": 1271
+    },
+    {
+      "entropy": 1.5215353667736053,
+      "epoch": 0.669826224328594,
+      "grad_norm": 0.27068790793418884,
+      "learning_rate": 0.00019797634894884184,
+      "loss": 0.21988946199417114,
+      "mean_token_accuracy": 0.9098561853170395,
+      "num_tokens": 15691392.0,
+      "step": 1272
+    },
+    {
+      "entropy": 1.5361709892749786,
+      "epoch": 0.6703528172722486,
+      "grad_norm": 0.3324735164642334,
+      "learning_rate": 0.00019797172240845935,
+      "loss": 0.2674825191497803,
+      "mean_token_accuracy": 0.8955648243427277,
+      "num_tokens": 15703728.0,
+      "step": 1273
+    },
+    {
+      "entropy": 1.5240019261837006,
+      "epoch": 0.6708794102159031,
+      "grad_norm": 0.3050387501716614,
+      "learning_rate": 0.00019796709064570842,
+      "loss": 0.22774918377399445,
+      "mean_token_accuracy": 0.9072832018136978,
+      "num_tokens": 15716064.0,
+      "step": 1274
+    },
+    {
+      "entropy": 1.4996045529842377,
+      "epoch": 0.6714060031595577,
+      "grad_norm": 0.2832599878311157,
+      "learning_rate": 0.00019796245366086404,
+      "loss": 0.21256539225578308,
+      "mean_token_accuracy": 0.9113875329494476,
+      "num_tokens": 15728400.0,
+      "step": 1275
+    },
+    {
+      "entropy": 1.537344366312027,
+      "epoch": 0.6719325961032122,
+      "grad_norm": 0.28976351022720337,
+      "learning_rate": 0.00019795781145420148,
+      "loss": 0.2061801552772522,
+      "mean_token_accuracy": 0.9169371575117111,
+      "num_tokens": 15740736.0,
+      "step": 1276
+    },
+    {
+      "entropy": 1.5299006402492523,
+      "epoch": 0.6724591890468667,
+      "grad_norm": 0.2772320806980133,
+      "learning_rate": 0.0001979531640259963,
+      "loss": 0.20862804353237152,
+      "mean_token_accuracy": 0.9104678928852081,
+      "num_tokens": 15753072.0,
+      "step": 1277
+    },
+    {
+      "entropy": 1.5044286847114563,
+      "epoch": 0.6729857819905213,
+      "grad_norm": 0.3096766769886017,
+      "learning_rate": 0.0001979485113765244,
+      "loss": 0.23148494958877563,
+      "mean_token_accuracy": 0.9055089354515076,
+      "num_tokens": 15765408.0,
+      "step": 1278
+    },
+    {
+      "entropy": 1.492220789194107,
+      "epoch": 0.6735123749341759,
+      "grad_norm": 0.2892164885997772,
+      "learning_rate": 0.00019794385350606198,
+      "loss": 0.2099100798368454,
+      "mean_token_accuracy": 0.9140574485063553,
+      "num_tokens": 15777744.0,
+      "step": 1279
+    },
+    {
+      "entropy": 1.5607218742370605,
+      "epoch": 0.6740389678778305,
+      "grad_norm": 0.3277181088924408,
+      "learning_rate": 0.00019793919041488555,
+      "loss": 0.22673308849334717,
+      "mean_token_accuracy": 0.9092708230018616,
+      "num_tokens": 15790080.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.5162492990493774,
+      "epoch": 0.674565560821485,
+      "grad_norm": 0.3172418475151062,
+      "learning_rate": 0.00019793452210327194,
+      "loss": 0.23533746600151062,
+      "mean_token_accuracy": 0.9022314250469208,
+      "num_tokens": 15802416.0,
+      "step": 1281
+    },
+    {
+      "entropy": 1.4783676862716675,
+      "epoch": 0.6750921537651395,
+      "grad_norm": 0.2982933521270752,
+      "learning_rate": 0.00019792984857149826,
+      "loss": 0.21463748812675476,
+      "mean_token_accuracy": 0.9117985665798187,
+      "num_tokens": 15814752.0,
+      "step": 1282
+    },
+    {
+      "entropy": 1.5162720084190369,
+      "epoch": 0.6756187467087941,
+      "grad_norm": 0.2976648211479187,
+      "learning_rate": 0.000197925169819842,
+      "loss": 0.21266265213489532,
+      "mean_token_accuracy": 0.9171696603298187,
+      "num_tokens": 15827088.0,
+      "step": 1283
+    },
+    {
+      "entropy": 1.5761696994304657,
+      "epoch": 0.6761453396524486,
+      "grad_norm": 0.32240986824035645,
+      "learning_rate": 0.00019792048584858082,
+      "loss": 0.22523874044418335,
+      "mean_token_accuracy": 0.9125792235136032,
+      "num_tokens": 15839424.0,
+      "step": 1284
+    },
+    {
+      "entropy": 1.4458520412445068,
+      "epoch": 0.6766719325961033,
+      "grad_norm": 0.2936963140964508,
+      "learning_rate": 0.00019791579665799286,
+      "loss": 0.20535065233707428,
+      "mean_token_accuracy": 0.9165598750114441,
+      "num_tokens": 15851760.0,
+      "step": 1285
+    },
+    {
+      "entropy": 1.4983311593532562,
+      "epoch": 0.6771985255397578,
+      "grad_norm": 0.34449514746665955,
+      "learning_rate": 0.0001979111022483565,
+      "loss": 0.21712952852249146,
+      "mean_token_accuracy": 0.9153823703527451,
+      "num_tokens": 15864096.0,
+      "step": 1286
+    },
+    {
+      "entropy": 1.562139093875885,
+      "epoch": 0.6777251184834123,
+      "grad_norm": 0.31743091344833374,
+      "learning_rate": 0.00019790640261995034,
+      "loss": 0.219345360994339,
+      "mean_token_accuracy": 0.9084994047880173,
+      "num_tokens": 15876432.0,
+      "step": 1287
+    },
+    {
+      "entropy": 1.5375787019729614,
+      "epoch": 0.6782517114270669,
+      "grad_norm": 0.27115342020988464,
+      "learning_rate": 0.00019790169777305345,
+      "loss": 0.19260238111019135,
+      "mean_token_accuracy": 0.9227229058742523,
+      "num_tokens": 15888768.0,
+      "step": 1288
+    },
+    {
+      "entropy": 1.4687424898147583,
+      "epoch": 0.6787783043707214,
+      "grad_norm": 0.302287220954895,
+      "learning_rate": 0.00019789698770794506,
+      "loss": 0.24450846016407013,
+      "mean_token_accuracy": 0.9040905237197876,
+      "num_tokens": 15901104.0,
+      "step": 1289
+    },
+    {
+      "entropy": 1.5125727355480194,
+      "epoch": 0.6793048973143759,
+      "grad_norm": 0.3316330313682556,
+      "learning_rate": 0.00019789227242490481,
+      "loss": 0.23443403840065002,
+      "mean_token_accuracy": 0.9077270179986954,
+      "num_tokens": 15913440.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.457822471857071,
+      "epoch": 0.6798314902580306,
+      "grad_norm": 0.3082807958126068,
+      "learning_rate": 0.00019788755192421266,
+      "loss": 0.2276977300643921,
+      "mean_token_accuracy": 0.907697319984436,
+      "num_tokens": 15925776.0,
+      "step": 1291
+    },
+    {
+      "entropy": 1.5687705874443054,
+      "epoch": 0.6803580832016851,
+      "grad_norm": 0.3347495198249817,
+      "learning_rate": 0.00019788282620614877,
+      "loss": 0.23658336699008942,
+      "mean_token_accuracy": 0.9071597903966904,
+      "num_tokens": 15938112.0,
+      "step": 1292
+    },
+    {
+      "entropy": 1.4747570753097534,
+      "epoch": 0.6808846761453397,
+      "grad_norm": 0.31832975149154663,
+      "learning_rate": 0.0001978780952709937,
+      "loss": 0.19530466198921204,
+      "mean_token_accuracy": 0.9262362122535706,
+      "num_tokens": 15950448.0,
+      "step": 1293
+    },
+    {
+      "entropy": 1.5592786967754364,
+      "epoch": 0.6814112690889942,
+      "grad_norm": 0.29860836267471313,
+      "learning_rate": 0.00019787335911902835,
+      "loss": 0.21259617805480957,
+      "mean_token_accuracy": 0.9156482219696045,
+      "num_tokens": 15962784.0,
+      "step": 1294
+    },
+    {
+      "entropy": 1.590408980846405,
+      "epoch": 0.6819378620326487,
+      "grad_norm": 0.33426377177238464,
+      "learning_rate": 0.0001978686177505338,
+      "loss": 0.250851035118103,
+      "mean_token_accuracy": 0.9034130126237869,
+      "num_tokens": 15975120.0,
+      "step": 1295
+    },
+    {
+      "entropy": 1.520596295595169,
+      "epoch": 0.6824644549763034,
+      "grad_norm": 0.31754234433174133,
+      "learning_rate": 0.00019786387116579155,
+      "loss": 0.23741202056407928,
+      "mean_token_accuracy": 0.9074528068304062,
+      "num_tokens": 15987456.0,
+      "step": 1296
+    },
+    {
+      "entropy": 1.589228630065918,
+      "epoch": 0.6829910479199579,
+      "grad_norm": 0.334957480430603,
+      "learning_rate": 0.0001978591193650834,
+      "loss": 0.2431160807609558,
+      "mean_token_accuracy": 0.9039677083492279,
+      "num_tokens": 15999792.0,
+      "step": 1297
+    },
+    {
+      "entropy": 1.5251929759979248,
+      "epoch": 0.6835176408636124,
+      "grad_norm": 0.26925790309906006,
+      "learning_rate": 0.00019785436234869138,
+      "loss": 0.18518362939357758,
+      "mean_token_accuracy": 0.9277700483798981,
+      "num_tokens": 16012128.0,
+      "step": 1298
+    },
+    {
+      "entropy": 1.49588942527771,
+      "epoch": 0.684044233807267,
+      "grad_norm": 0.2995278835296631,
+      "learning_rate": 0.00019784960011689793,
+      "loss": 0.21347451210021973,
+      "mean_token_accuracy": 0.9186176806688309,
+      "num_tokens": 16024464.0,
+      "step": 1299
+    },
+    {
+      "entropy": 1.5090012848377228,
+      "epoch": 0.6845708267509215,
+      "grad_norm": 0.2900152802467346,
+      "learning_rate": 0.00019784483266998575,
+      "loss": 0.20770123600959778,
+      "mean_token_accuracy": 0.9164092987775803,
+      "num_tokens": 16036800.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.524385005235672,
+      "epoch": 0.685097419694576,
+      "grad_norm": 0.2801123857498169,
+      "learning_rate": 0.00019784006000823787,
+      "loss": 0.1997857242822647,
+      "mean_token_accuracy": 0.9173883497714996,
+      "num_tokens": 16049136.0,
+      "step": 1301
+    },
+    {
+      "entropy": 1.5324156284332275,
+      "epoch": 0.6856240126382307,
+      "grad_norm": 0.30303382873535156,
+      "learning_rate": 0.0001978352821319376,
+      "loss": 0.21537630259990692,
+      "mean_token_accuracy": 0.9123150855302811,
+      "num_tokens": 16061472.0,
+      "step": 1302
+    },
+    {
+      "entropy": 1.538516879081726,
+      "epoch": 0.6861506055818852,
+      "grad_norm": 0.2965938448905945,
+      "learning_rate": 0.00019783049904136855,
+      "loss": 0.20010440051555634,
+      "mean_token_accuracy": 0.9172840118408203,
+      "num_tokens": 16073808.0,
+      "step": 1303
+    },
+    {
+      "entropy": 1.4958137571811676,
+      "epoch": 0.6866771985255398,
+      "grad_norm": 0.29906049370765686,
+      "learning_rate": 0.0001978257107368147,
+      "loss": 0.20351183414459229,
+      "mean_token_accuracy": 0.921799048781395,
+      "num_tokens": 16086144.0,
+      "step": 1304
+    },
+    {
+      "entropy": 1.4531968832015991,
+      "epoch": 0.6872037914691943,
+      "grad_norm": 0.31886446475982666,
+      "learning_rate": 0.0001978209172185603,
+      "loss": 0.21320727467536926,
+      "mean_token_accuracy": 0.9105631560087204,
+      "num_tokens": 16098480.0,
+      "step": 1305
+    },
+    {
+      "entropy": 1.3787581324577332,
+      "epoch": 0.6877303844128488,
+      "grad_norm": 0.2898115813732147,
+      "learning_rate": 0.0001978161184868899,
+      "loss": 0.2124798595905304,
+      "mean_token_accuracy": 0.9114013463258743,
+      "num_tokens": 16110816.0,
+      "step": 1306
+    },
+    {
+      "entropy": 1.4378518164157867,
+      "epoch": 0.6882569773565034,
+      "grad_norm": 0.31470921635627747,
+      "learning_rate": 0.00019781131454208833,
+      "loss": 0.19701991975307465,
+      "mean_token_accuracy": 0.9186731576919556,
+      "num_tokens": 16123152.0,
+      "step": 1307
+    },
+    {
+      "entropy": 1.4289041459560394,
+      "epoch": 0.688783570300158,
+      "grad_norm": 0.3303619921207428,
+      "learning_rate": 0.0001978065053844409,
+      "loss": 0.23253998160362244,
+      "mean_token_accuracy": 0.9104226678609848,
+      "num_tokens": 16135488.0,
+      "step": 1308
+    },
+    {
+      "entropy": 1.458989679813385,
+      "epoch": 0.6893101632438126,
+      "grad_norm": 0.30710119009017944,
+      "learning_rate": 0.00019780169101423295,
+      "loss": 0.2202616184949875,
+      "mean_token_accuracy": 0.9115623980760574,
+      "num_tokens": 16147824.0,
+      "step": 1309
+    },
+    {
+      "entropy": 1.5141521096229553,
+      "epoch": 0.6898367561874671,
+      "grad_norm": 0.3152952194213867,
+      "learning_rate": 0.0001977968714317504,
+      "loss": 0.19381463527679443,
+      "mean_token_accuracy": 0.9204765856266022,
+      "num_tokens": 16160160.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.4804642498493195,
+      "epoch": 0.6903633491311216,
+      "grad_norm": 0.3526984751224518,
+      "learning_rate": 0.0001977920466372793,
+      "loss": 0.22401507198810577,
+      "mean_token_accuracy": 0.9142041653394699,
+      "num_tokens": 16172496.0,
+      "step": 1311
+    },
+    {
+      "entropy": 1.4126333594322205,
+      "epoch": 0.6908899420747762,
+      "grad_norm": 0.2999390661716461,
+      "learning_rate": 0.00019778721663110603,
+      "loss": 0.212639719247818,
+      "mean_token_accuracy": 0.9140689373016357,
+      "num_tokens": 16184832.0,
+      "step": 1312
+    },
+    {
+      "entropy": 1.4407996237277985,
+      "epoch": 0.6914165350184307,
+      "grad_norm": 0.3175979554653168,
+      "learning_rate": 0.00019778238141351744,
+      "loss": 0.23567378520965576,
+      "mean_token_accuracy": 0.9058562517166138,
+      "num_tokens": 16197168.0,
+      "step": 1313
+    },
+    {
+      "entropy": 1.4304053783416748,
+      "epoch": 0.6919431279620853,
+      "grad_norm": 0.32357415556907654,
+      "learning_rate": 0.00019777754098480047,
+      "loss": 0.23323741555213928,
+      "mean_token_accuracy": 0.9068403393030167,
+      "num_tokens": 16209504.0,
+      "step": 1314
+    },
+    {
+      "entropy": 1.4463219940662384,
+      "epoch": 0.6924697209057399,
+      "grad_norm": 0.32085898518562317,
+      "learning_rate": 0.0001977726953452425,
+      "loss": 0.25303539633750916,
+      "mean_token_accuracy": 0.9010388404130936,
+      "num_tokens": 16221840.0,
+      "step": 1315
+    },
+    {
+      "entropy": 1.5001619458198547,
+      "epoch": 0.6929963138493944,
+      "grad_norm": 0.30218154191970825,
+      "learning_rate": 0.0001977678444951312,
+      "loss": 0.20458348095417023,
+      "mean_token_accuracy": 0.9198221117258072,
+      "num_tokens": 16234176.0,
+      "step": 1316
+    },
+    {
+      "entropy": 1.4169287383556366,
+      "epoch": 0.693522906793049,
+      "grad_norm": 0.2987150549888611,
+      "learning_rate": 0.0001977629884347545,
+      "loss": 0.20915895700454712,
+      "mean_token_accuracy": 0.914911612868309,
+      "num_tokens": 16246512.0,
+      "step": 1317
+    },
+    {
+      "entropy": 1.4708910584449768,
+      "epoch": 0.6940494997367035,
+      "grad_norm": 0.2975226938724518,
+      "learning_rate": 0.00019775812716440073,
+      "loss": 0.21582108736038208,
+      "mean_token_accuracy": 0.916588231921196,
+      "num_tokens": 16258848.0,
+      "step": 1318
+    },
+    {
+      "entropy": 1.395564466714859,
+      "epoch": 0.6945760926803581,
+      "grad_norm": 0.26800790429115295,
+      "learning_rate": 0.00019775326068435843,
+      "loss": 0.19701655209064484,
+      "mean_token_accuracy": 0.9204675257205963,
+      "num_tokens": 16271184.0,
+      "step": 1319
+    },
+    {
+      "entropy": 1.4954661428928375,
+      "epoch": 0.6951026856240127,
+      "grad_norm": 0.3218332529067993,
+      "learning_rate": 0.00019774838899491647,
+      "loss": 0.20654654502868652,
+      "mean_token_accuracy": 0.9205126464366913,
+      "num_tokens": 16283520.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.475353330373764,
+      "epoch": 0.6956292785676672,
+      "grad_norm": 0.29199621081352234,
+      "learning_rate": 0.00019774351209636413,
+      "loss": 0.22376279532909393,
+      "mean_token_accuracy": 0.9100230783224106,
+      "num_tokens": 16295856.0,
+      "step": 1321
+    },
+    {
+      "entropy": 1.5148694515228271,
+      "epoch": 0.6961558715113217,
+      "grad_norm": 0.30520734190940857,
+      "learning_rate": 0.00019773862998899086,
+      "loss": 0.2176750898361206,
+      "mean_token_accuracy": 0.9201380461454391,
+      "num_tokens": 16308192.0,
+      "step": 1322
+    },
+    {
+      "entropy": 1.5539270341396332,
+      "epoch": 0.6966824644549763,
+      "grad_norm": 0.32190483808517456,
+      "learning_rate": 0.00019773374267308653,
+      "loss": 0.2056635320186615,
+      "mean_token_accuracy": 0.914083868265152,
+      "num_tokens": 16320528.0,
+      "step": 1323
+    },
+    {
+      "entropy": 1.4479952156543732,
+      "epoch": 0.6972090573986308,
+      "grad_norm": 0.3086356222629547,
+      "learning_rate": 0.00019772885014894125,
+      "loss": 0.19541175663471222,
+      "mean_token_accuracy": 0.9208300113677979,
+      "num_tokens": 16332864.0,
+      "step": 1324
+    },
+    {
+      "entropy": 1.4255986511707306,
+      "epoch": 0.6977356503422855,
+      "grad_norm": 0.29759716987609863,
+      "learning_rate": 0.00019772395241684547,
+      "loss": 0.22268646955490112,
+      "mean_token_accuracy": 0.9130487442016602,
+      "num_tokens": 16345200.0,
+      "step": 1325
+    },
+    {
+      "entropy": 1.417414665222168,
+      "epoch": 0.69826224328594,
+      "grad_norm": 0.30235084891319275,
+      "learning_rate": 0.00019771904947708991,
+      "loss": 0.19872839748859406,
+      "mean_token_accuracy": 0.9202670305967331,
+      "num_tokens": 16357536.0,
+      "step": 1326
+    },
+    {
+      "entropy": 1.462871015071869,
+      "epoch": 0.6987888362295945,
+      "grad_norm": 0.330930233001709,
+      "learning_rate": 0.00019771414132996565,
+      "loss": 0.21214503049850464,
+      "mean_token_accuracy": 0.9157495498657227,
+      "num_tokens": 16369872.0,
+      "step": 1327
+    },
+    {
+      "entropy": 1.4870843291282654,
+      "epoch": 0.6993154291732491,
+      "grad_norm": 0.33824852108955383,
+      "learning_rate": 0.00019770922797576407,
+      "loss": 0.23521599173545837,
+      "mean_token_accuracy": 0.9104333966970444,
+      "num_tokens": 16382208.0,
+      "step": 1328
+    },
+    {
+      "entropy": 1.4692321121692657,
+      "epoch": 0.6998420221169036,
+      "grad_norm": 0.2853015959262848,
+      "learning_rate": 0.0001977043094147768,
+      "loss": 0.21163907647132874,
+      "mean_token_accuracy": 0.9183786511421204,
+      "num_tokens": 16394544.0,
+      "step": 1329
+    },
+    {
+      "entropy": 1.4160318076610565,
+      "epoch": 0.7003686150605581,
+      "grad_norm": 0.3194182217121124,
+      "learning_rate": 0.00019769938564729585,
+      "loss": 0.22332850098609924,
+      "mean_token_accuracy": 0.9086870551109314,
+      "num_tokens": 16406880.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.5249937772750854,
+      "epoch": 0.7008952080042128,
+      "grad_norm": 0.2979271709918976,
+      "learning_rate": 0.00019769445667361358,
+      "loss": 0.20148411393165588,
+      "mean_token_accuracy": 0.919349730014801,
+      "num_tokens": 16419216.0,
+      "step": 1331
+    },
+    {
+      "entropy": 1.475309818983078,
+      "epoch": 0.7014218009478673,
+      "grad_norm": 0.31436359882354736,
+      "learning_rate": 0.00019768952249402252,
+      "loss": 0.2226470410823822,
+      "mean_token_accuracy": 0.9143447577953339,
+      "num_tokens": 16431552.0,
+      "step": 1332
+    },
+    {
+      "entropy": 1.45591801404953,
+      "epoch": 0.7019483938915219,
+      "grad_norm": 0.3134812116622925,
+      "learning_rate": 0.00019768458310881557,
+      "loss": 0.239861398935318,
+      "mean_token_accuracy": 0.9030255079269409,
+      "num_tokens": 16443888.0,
+      "step": 1333
+    },
+    {
+      "entropy": 1.4746840596199036,
+      "epoch": 0.7024749868351764,
+      "grad_norm": 0.2945224642753601,
+      "learning_rate": 0.00019767963851828604,
+      "loss": 0.2150000035762787,
+      "mean_token_accuracy": 0.9179409146308899,
+      "num_tokens": 16456224.0,
+      "step": 1334
+    },
+    {
+      "entropy": 1.4820148646831512,
+      "epoch": 0.7030015797788309,
+      "grad_norm": 0.30797162652015686,
+      "learning_rate": 0.00019767468872272737,
+      "loss": 0.22350315749645233,
+      "mean_token_accuracy": 0.9135662913322449,
+      "num_tokens": 16468560.0,
+      "step": 1335
+    },
+    {
+      "entropy": 1.5698617100715637,
+      "epoch": 0.7035281727224855,
+      "grad_norm": 0.3179170489311218,
+      "learning_rate": 0.00019766973372243343,
+      "loss": 0.20203739404678345,
+      "mean_token_accuracy": 0.9202521592378616,
+      "num_tokens": 16480896.0,
+      "step": 1336
+    },
+    {
+      "entropy": 1.5165068209171295,
+      "epoch": 0.7040547656661401,
+      "grad_norm": 0.282023549079895,
+      "learning_rate": 0.0001976647735176984,
+      "loss": 0.2072068601846695,
+      "mean_token_accuracy": 0.9170822501182556,
+      "num_tokens": 16493232.0,
+      "step": 1337
+    },
+    {
+      "entropy": 1.5599988996982574,
+      "epoch": 0.7045813586097947,
+      "grad_norm": 0.32171323895454407,
+      "learning_rate": 0.0001976598081088167,
+      "loss": 0.2177916020154953,
+      "mean_token_accuracy": 0.9149684607982635,
+      "num_tokens": 16505568.0,
+      "step": 1338
+    },
+    {
+      "entropy": 1.566039115190506,
+      "epoch": 0.7051079515534492,
+      "grad_norm": 0.2976767420768738,
+      "learning_rate": 0.00019765483749608313,
+      "loss": 0.20423297584056854,
+      "mean_token_accuracy": 0.9218804687261581,
+      "num_tokens": 16517904.0,
+      "step": 1339
+    },
+    {
+      "entropy": 1.5926516354084015,
+      "epoch": 0.7056345444971037,
+      "grad_norm": 0.3847099244594574,
+      "learning_rate": 0.00019764986167979275,
+      "loss": 0.22431136667728424,
+      "mean_token_accuracy": 0.9091393053531647,
+      "num_tokens": 16530240.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.5083985030651093,
+      "epoch": 0.7061611374407583,
+      "grad_norm": 0.3162590563297272,
+      "learning_rate": 0.0001976448806602409,
+      "loss": 0.2287883162498474,
+      "mean_token_accuracy": 0.9064637273550034,
+      "num_tokens": 16542576.0,
+      "step": 1341
+    },
+    {
+      "entropy": 1.4989595413208008,
+      "epoch": 0.7066877303844129,
+      "grad_norm": 0.3000263571739197,
+      "learning_rate": 0.00019763989443772337,
+      "loss": 0.21327601373195648,
+      "mean_token_accuracy": 0.9155562520027161,
+      "num_tokens": 16554912.0,
+      "step": 1342
+    },
+    {
+      "entropy": 1.5128290951251984,
+      "epoch": 0.7072143233280674,
+      "grad_norm": 0.2836988568305969,
+      "learning_rate": 0.00019763490301253608,
+      "loss": 0.19906045496463776,
+      "mean_token_accuracy": 0.9238555133342743,
+      "num_tokens": 16567248.0,
+      "step": 1343
+    },
+    {
+      "entropy": 1.5032199919223785,
+      "epoch": 0.707740916271722,
+      "grad_norm": 0.2904132306575775,
+      "learning_rate": 0.0001976299063849754,
+      "loss": 0.21531090140342712,
+      "mean_token_accuracy": 0.9095672369003296,
+      "num_tokens": 16579584.0,
+      "step": 1344
+    },
+    {
+      "entropy": 1.587888479232788,
+      "epoch": 0.7082675092153765,
+      "grad_norm": 0.3187963664531708,
+      "learning_rate": 0.00019762490455533792,
+      "loss": 0.21344661712646484,
+      "mean_token_accuracy": 0.9137663394212723,
+      "num_tokens": 16591920.0,
+      "step": 1345
+    },
+    {
+      "entropy": 1.5427050590515137,
+      "epoch": 0.708794102159031,
+      "grad_norm": 0.2940306067466736,
+      "learning_rate": 0.00019761989752392053,
+      "loss": 0.1942479908466339,
+      "mean_token_accuracy": 0.9190978854894638,
+      "num_tokens": 16604256.0,
+      "step": 1346
+    },
+    {
+      "entropy": 1.5427431166172028,
+      "epoch": 0.7093206951026856,
+      "grad_norm": 0.3436744809150696,
+      "learning_rate": 0.00019761488529102057,
+      "loss": 0.2245483100414276,
+      "mean_token_accuracy": 0.909973606467247,
+      "num_tokens": 16616592.0,
+      "step": 1347
+    },
+    {
+      "entropy": 1.4897418022155762,
+      "epoch": 0.7098472880463402,
+      "grad_norm": 0.34564313292503357,
+      "learning_rate": 0.0001976098678569355,
+      "loss": 0.23059901595115662,
+      "mean_token_accuracy": 0.9069952964782715,
+      "num_tokens": 16628928.0,
+      "step": 1348
+    },
+    {
+      "entropy": 1.5078385174274445,
+      "epoch": 0.7103738809899948,
+      "grad_norm": 0.32199203968048096,
+      "learning_rate": 0.0001976048452219632,
+      "loss": 0.22921977937221527,
+      "mean_token_accuracy": 0.9082883149385452,
+      "num_tokens": 16641264.0,
+      "step": 1349
+    },
+    {
+      "entropy": 1.5083224773406982,
+      "epoch": 0.7109004739336493,
+      "grad_norm": 0.3183170258998871,
+      "learning_rate": 0.00019759981738640185,
+      "loss": 0.22415608167648315,
+      "mean_token_accuracy": 0.9100949168205261,
+      "num_tokens": 16653600.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.5398178398609161,
+      "epoch": 0.7114270668773038,
+      "grad_norm": 0.30399900674819946,
+      "learning_rate": 0.0001975947843505499,
+      "loss": 0.22306974232196808,
+      "mean_token_accuracy": 0.9138671904802322,
+      "num_tokens": 16665936.0,
+      "step": 1351
+    },
+    {
+      "entropy": 1.535412222146988,
+      "epoch": 0.7119536598209584,
+      "grad_norm": 0.2907712757587433,
+      "learning_rate": 0.00019758974611470615,
+      "loss": 0.2197205275297165,
+      "mean_token_accuracy": 0.9130839556455612,
+      "num_tokens": 16678272.0,
+      "step": 1352
+    },
+    {
+      "entropy": 1.4855974614620209,
+      "epoch": 0.7124802527646129,
+      "grad_norm": 0.28864648938179016,
+      "learning_rate": 0.0001975847026791697,
+      "loss": 0.21561571955680847,
+      "mean_token_accuracy": 0.9168848842382431,
+      "num_tokens": 16690608.0,
+      "step": 1353
+    },
+    {
+      "entropy": 1.453218251466751,
+      "epoch": 0.7130068457082676,
+      "grad_norm": 0.3135017454624176,
+      "learning_rate": 0.00019757965404423994,
+      "loss": 0.2011297047138214,
+      "mean_token_accuracy": 0.9190320074558258,
+      "num_tokens": 16702944.0,
+      "step": 1354
+    },
+    {
+      "entropy": 1.4790661036968231,
+      "epoch": 0.7135334386519221,
+      "grad_norm": 0.3001917004585266,
+      "learning_rate": 0.00019757460021021655,
+      "loss": 0.23639211058616638,
+      "mean_token_accuracy": 0.9105001240968704,
+      "num_tokens": 16715280.0,
+      "step": 1355
+    },
+    {
+      "entropy": 1.496882289648056,
+      "epoch": 0.7140600315955766,
+      "grad_norm": 0.31785574555397034,
+      "learning_rate": 0.00019756954117739956,
+      "loss": 0.2306048721075058,
+      "mean_token_accuracy": 0.9150114953517914,
+      "num_tokens": 16727616.0,
+      "step": 1356
+    },
+    {
+      "entropy": 1.4668599963188171,
+      "epoch": 0.7145866245392312,
+      "grad_norm": 0.29470139741897583,
+      "learning_rate": 0.00019756447694608932,
+      "loss": 0.21168065071105957,
+      "mean_token_accuracy": 0.9137952625751495,
+      "num_tokens": 16739952.0,
+      "step": 1357
+    },
+    {
+      "entropy": 1.422877013683319,
+      "epoch": 0.7151132174828857,
+      "grad_norm": 0.27277153730392456,
+      "learning_rate": 0.00019755940751658643,
+      "loss": 0.20774061977863312,
+      "mean_token_accuracy": 0.9166775196790695,
+      "num_tokens": 16752288.0,
+      "step": 1358
+    },
+    {
+      "entropy": 1.4244271516799927,
+      "epoch": 0.7156398104265402,
+      "grad_norm": 0.31778132915496826,
+      "learning_rate": 0.00019755433288919187,
+      "loss": 0.231976717710495,
+      "mean_token_accuracy": 0.9075390994548798,
+      "num_tokens": 16764624.0,
+      "step": 1359
+    },
+    {
+      "entropy": 1.4513772428035736,
+      "epoch": 0.7161664033701949,
+      "grad_norm": 0.30516210198402405,
+      "learning_rate": 0.0001975492530642069,
+      "loss": 0.20754879713058472,
+      "mean_token_accuracy": 0.9216670095920563,
+      "num_tokens": 16776960.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.4767080545425415,
+      "epoch": 0.7166929963138494,
+      "grad_norm": 0.3156355917453766,
+      "learning_rate": 0.00019754416804193298,
+      "loss": 0.20898932218551636,
+      "mean_token_accuracy": 0.9173450469970703,
+      "num_tokens": 16789296.0,
+      "step": 1361
+    },
+    {
+      "entropy": 1.4257088005542755,
+      "epoch": 0.717219589257504,
+      "grad_norm": 0.33443206548690796,
+      "learning_rate": 0.0001975390778226721,
+      "loss": 0.237446129322052,
+      "mean_token_accuracy": 0.9077172726392746,
+      "num_tokens": 16801632.0,
+      "step": 1362
+    },
+    {
+      "entropy": 1.393272042274475,
+      "epoch": 0.7177461822011585,
+      "grad_norm": 0.30247074365615845,
+      "learning_rate": 0.00019753398240672637,
+      "loss": 0.21317076683044434,
+      "mean_token_accuracy": 0.9131140112876892,
+      "num_tokens": 16813968.0,
+      "step": 1363
+    },
+    {
+      "entropy": 1.431013286113739,
+      "epoch": 0.718272775144813,
+      "grad_norm": 0.29114142060279846,
+      "learning_rate": 0.0001975288817943983,
+      "loss": 0.19287830591201782,
+      "mean_token_accuracy": 0.9247073382139206,
+      "num_tokens": 16826304.0,
+      "step": 1364
+    },
+    {
+      "entropy": 1.3813497424125671,
+      "epoch": 0.7187993680884676,
+      "grad_norm": 0.2997731566429138,
+      "learning_rate": 0.00019752377598599067,
+      "loss": 0.2021138221025467,
+      "mean_token_accuracy": 0.9203190356492996,
+      "num_tokens": 16838640.0,
+      "step": 1365
+    },
+    {
+      "entropy": 1.4252038598060608,
+      "epoch": 0.7193259610321222,
+      "grad_norm": 0.3252887725830078,
+      "learning_rate": 0.0001975186649818066,
+      "loss": 0.20873992145061493,
+      "mean_token_accuracy": 0.912327915430069,
+      "num_tokens": 16850976.0,
+      "step": 1366
+    },
+    {
+      "entropy": 1.4511545896530151,
+      "epoch": 0.7198525539757767,
+      "grad_norm": 0.34324178099632263,
+      "learning_rate": 0.00019751354878214946,
+      "loss": 0.21222266554832458,
+      "mean_token_accuracy": 0.9157299846410751,
+      "num_tokens": 16863312.0,
+      "step": 1367
+    },
+    {
+      "entropy": 1.3630057871341705,
+      "epoch": 0.7203791469194313,
+      "grad_norm": 0.37623706459999084,
+      "learning_rate": 0.00019750842738732304,
+      "loss": 0.2201295793056488,
+      "mean_token_accuracy": 0.9165831059217453,
+      "num_tokens": 16875648.0,
+      "step": 1368
+    },
+    {
+      "entropy": 1.3505005836486816,
+      "epoch": 0.7209057398630858,
+      "grad_norm": 0.30142372846603394,
+      "learning_rate": 0.0001975033007976313,
+      "loss": 0.22212043404579163,
+      "mean_token_accuracy": 0.9122855514287949,
+      "num_tokens": 16887984.0,
+      "step": 1369
+    },
+    {
+      "entropy": 1.4169504642486572,
+      "epoch": 0.7214323328067404,
+      "grad_norm": 0.3326164782047272,
+      "learning_rate": 0.0001974981690133786,
+      "loss": 0.22192654013633728,
+      "mean_token_accuracy": 0.9108520150184631,
+      "num_tokens": 16900320.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.367185264825821,
+      "epoch": 0.721958925750395,
+      "grad_norm": 0.33256617188453674,
+      "learning_rate": 0.00019749303203486958,
+      "loss": 0.24168333411216736,
+      "mean_token_accuracy": 0.9047919362783432,
+      "num_tokens": 16912656.0,
+      "step": 1371
+    },
+    {
+      "entropy": 1.3614123463630676,
+      "epoch": 0.7224855186940495,
+      "grad_norm": 0.3197641670703888,
+      "learning_rate": 0.00019748788986240917,
+      "loss": 0.23665747046470642,
+      "mean_token_accuracy": 0.9026706665754318,
+      "num_tokens": 16924992.0,
+      "step": 1372
+    },
+    {
+      "entropy": 1.3906526267528534,
+      "epoch": 0.7230121116377041,
+      "grad_norm": 0.29753339290618896,
+      "learning_rate": 0.00019748274249630269,
+      "loss": 0.19765980541706085,
+      "mean_token_accuracy": 0.9235783219337463,
+      "num_tokens": 16937328.0,
+      "step": 1373
+    },
+    {
+      "entropy": 1.3462652564048767,
+      "epoch": 0.7235387045813586,
+      "grad_norm": 0.2804873585700989,
+      "learning_rate": 0.00019747758993685568,
+      "loss": 0.22064946591854095,
+      "mean_token_accuracy": 0.9110289812088013,
+      "num_tokens": 16949664.0,
+      "step": 1374
+    },
+    {
+      "entropy": 1.4082630276679993,
+      "epoch": 0.7240652975250131,
+      "grad_norm": 0.3095674216747284,
+      "learning_rate": 0.000197472432184374,
+      "loss": 0.20987483859062195,
+      "mean_token_accuracy": 0.9188564568758011,
+      "num_tokens": 16962000.0,
+      "step": 1375
+    },
+    {
+      "entropy": 1.304635614156723,
+      "epoch": 0.7245918904686677,
+      "grad_norm": 0.2778051197528839,
+      "learning_rate": 0.00019746726923916387,
+      "loss": 0.21651524305343628,
+      "mean_token_accuracy": 0.9125857651233673,
+      "num_tokens": 16974336.0,
+      "step": 1376
+    },
+    {
+      "entropy": 1.3015822768211365,
+      "epoch": 0.7251184834123223,
+      "grad_norm": 0.2806200683116913,
+      "learning_rate": 0.00019746210110153172,
+      "loss": 0.22142334282398224,
+      "mean_token_accuracy": 0.911768451333046,
+      "num_tokens": 16986672.0,
+      "step": 1377
+    },
+    {
+      "entropy": 1.3526243567466736,
+      "epoch": 0.7256450763559769,
+      "grad_norm": 0.3183799684047699,
+      "learning_rate": 0.0001974569277717844,
+      "loss": 0.2256852686405182,
+      "mean_token_accuracy": 0.9073309451341629,
+      "num_tokens": 16999008.0,
+      "step": 1378
+    },
+    {
+      "entropy": 1.3906278908252716,
+      "epoch": 0.7261716692996314,
+      "grad_norm": 0.2876151204109192,
+      "learning_rate": 0.00019745174925022904,
+      "loss": 0.2028118073940277,
+      "mean_token_accuracy": 0.923457607626915,
+      "num_tokens": 17011344.0,
+      "step": 1379
+    },
+    {
+      "entropy": 1.350665271282196,
+      "epoch": 0.7266982622432859,
+      "grad_norm": 0.33748704195022583,
+      "learning_rate": 0.00019744656553717305,
+      "loss": 0.2175031304359436,
+      "mean_token_accuracy": 0.9140133857727051,
+      "num_tokens": 17023680.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.3705949485301971,
+      "epoch": 0.7272248551869405,
+      "grad_norm": 0.3174329400062561,
+      "learning_rate": 0.00019744137663292412,
+      "loss": 0.20505183935165405,
+      "mean_token_accuracy": 0.916661724448204,
+      "num_tokens": 17036016.0,
+      "step": 1381
+    },
+    {
+      "entropy": 1.3908927738666534,
+      "epoch": 0.727751448130595,
+      "grad_norm": 0.31065794825553894,
+      "learning_rate": 0.00019743618253779033,
+      "loss": 0.23529362678527832,
+      "mean_token_accuracy": 0.910541832447052,
+      "num_tokens": 17048352.0,
+      "step": 1382
+    },
+    {
+      "entropy": 1.3258641958236694,
+      "epoch": 0.7282780410742496,
+      "grad_norm": 0.26997655630111694,
+      "learning_rate": 0.00019743098325207995,
+      "loss": 0.18447992205619812,
+      "mean_token_accuracy": 0.9224408715963364,
+      "num_tokens": 17060688.0,
+      "step": 1383
+    },
+    {
+      "entropy": 1.3685748875141144,
+      "epoch": 0.7288046340179042,
+      "grad_norm": 0.3111379146575928,
+      "learning_rate": 0.00019742577877610173,
+      "loss": 0.22453925013542175,
+      "mean_token_accuracy": 0.9133676290512085,
+      "num_tokens": 17073024.0,
+      "step": 1384
+    },
+    {
+      "entropy": 1.42083340883255,
+      "epoch": 0.7293312269615587,
+      "grad_norm": 0.331272691488266,
+      "learning_rate": 0.00019742056911016456,
+      "loss": 0.23528197407722473,
+      "mean_token_accuracy": 0.9075358361005783,
+      "num_tokens": 17085360.0,
+      "step": 1385
+    },
+    {
+      "entropy": 1.4835380911827087,
+      "epoch": 0.7298578199052133,
+      "grad_norm": 0.30057939887046814,
+      "learning_rate": 0.0001974153542545777,
+      "loss": 0.19702467322349548,
+      "mean_token_accuracy": 0.9221197068691254,
+      "num_tokens": 17097696.0,
+      "step": 1386
+    },
+    {
+      "entropy": 1.412318229675293,
+      "epoch": 0.7303844128488678,
+      "grad_norm": 0.314919650554657,
+      "learning_rate": 0.00019741013420965077,
+      "loss": 0.22496280074119568,
+      "mean_token_accuracy": 0.9107683897018433,
+      "num_tokens": 17110032.0,
+      "step": 1387
+    },
+    {
+      "entropy": 1.5203742980957031,
+      "epoch": 0.7309110057925223,
+      "grad_norm": 0.4578418433666229,
+      "learning_rate": 0.00019740490897569366,
+      "loss": 0.22902803122997284,
+      "mean_token_accuracy": 0.9130762368440628,
+      "num_tokens": 17122368.0,
+      "step": 1388
+    },
+    {
+      "entropy": 1.496030032634735,
+      "epoch": 0.731437598736177,
+      "grad_norm": 0.29055219888687134,
+      "learning_rate": 0.00019739967855301652,
+      "loss": 0.21085382997989655,
+      "mean_token_accuracy": 0.9122315496206284,
+      "num_tokens": 17134704.0,
+      "step": 1389
+    },
+    {
+      "entropy": 1.4665674567222595,
+      "epoch": 0.7319641916798315,
+      "grad_norm": 0.2788514494895935,
+      "learning_rate": 0.0001973944429419299,
+      "loss": 0.20301197469234467,
+      "mean_token_accuracy": 0.9215195178985596,
+      "num_tokens": 17147040.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.5158916115760803,
+      "epoch": 0.732490784623486,
+      "grad_norm": 0.3139953315258026,
+      "learning_rate": 0.00019738920214274454,
+      "loss": 0.20764051377773285,
+      "mean_token_accuracy": 0.915527880191803,
+      "num_tokens": 17159376.0,
+      "step": 1391
+    },
+    {
+      "entropy": 1.5249950289726257,
+      "epoch": 0.7330173775671406,
+      "grad_norm": 0.30590322613716125,
+      "learning_rate": 0.00019738395615577161,
+      "loss": 0.22601057589054108,
+      "mean_token_accuracy": 0.9111659973859787,
+      "num_tokens": 17171712.0,
+      "step": 1392
+    },
+    {
+      "entropy": 1.468850463628769,
+      "epoch": 0.7335439705107951,
+      "grad_norm": 0.311515212059021,
+      "learning_rate": 0.00019737870498132248,
+      "loss": 0.2210465669631958,
+      "mean_token_accuracy": 0.9097635447978973,
+      "num_tokens": 17184048.0,
+      "step": 1393
+    },
+    {
+      "entropy": 1.5013999938964844,
+      "epoch": 0.7340705634544498,
+      "grad_norm": 0.3101951479911804,
+      "learning_rate": 0.00019737344861970896,
+      "loss": 0.21400384604930878,
+      "mean_token_accuracy": 0.9109934121370316,
+      "num_tokens": 17196384.0,
+      "step": 1394
+    },
+    {
+      "entropy": 1.4603233933448792,
+      "epoch": 0.7345971563981043,
+      "grad_norm": 0.26839688420295715,
+      "learning_rate": 0.00019736818707124303,
+      "loss": 0.1959075927734375,
+      "mean_token_accuracy": 0.9210042357444763,
+      "num_tokens": 17208720.0,
+      "step": 1395
+    },
+    {
+      "entropy": 1.513748586177826,
+      "epoch": 0.7351237493417588,
+      "grad_norm": 0.3097459673881531,
+      "learning_rate": 0.00019736292033623704,
+      "loss": 0.2226242572069168,
+      "mean_token_accuracy": 0.9143490940332413,
+      "num_tokens": 17221056.0,
+      "step": 1396
+    },
+    {
+      "entropy": 1.4970153272151947,
+      "epoch": 0.7356503422854134,
+      "grad_norm": 0.3178339898586273,
+      "learning_rate": 0.00019735764841500368,
+      "loss": 0.22419294714927673,
+      "mean_token_accuracy": 0.9115694165229797,
+      "num_tokens": 17233392.0,
+      "step": 1397
+    },
+    {
+      "entropy": 1.4637455344200134,
+      "epoch": 0.7361769352290679,
+      "grad_norm": 0.3103068768978119,
+      "learning_rate": 0.0001973523713078559,
+      "loss": 0.2213975489139557,
+      "mean_token_accuracy": 0.908700242638588,
+      "num_tokens": 17245728.0,
+      "step": 1398
+    },
+    {
+      "entropy": 1.4048950970172882,
+      "epoch": 0.7367035281727224,
+      "grad_norm": 0.3316826820373535,
+      "learning_rate": 0.00019734708901510695,
+      "loss": 0.22181765735149384,
+      "mean_token_accuracy": 0.9124202877283096,
+      "num_tokens": 17258064.0,
+      "step": 1399
+    },
+    {
+      "entropy": 1.400584727525711,
+      "epoch": 0.7372301211163771,
+      "grad_norm": 0.30288171768188477,
+      "learning_rate": 0.0001973418015370704,
+      "loss": 0.23632709681987762,
+      "mean_token_accuracy": 0.903136819601059,
+      "num_tokens": 17270400.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.4175618290901184,
+      "epoch": 0.7377567140600316,
+      "grad_norm": 0.2811354696750641,
+      "learning_rate": 0.00019733650887406018,
+      "loss": 0.21318787336349487,
+      "mean_token_accuracy": 0.9126076847314835,
+      "num_tokens": 17282736.0,
+      "step": 1401
+    },
+    {
+      "entropy": 1.4151119589805603,
+      "epoch": 0.7382833070036862,
+      "grad_norm": 0.2701869010925293,
+      "learning_rate": 0.00019733121102639048,
+      "loss": 0.21690967679023743,
+      "mean_token_accuracy": 0.9093466997146606,
+      "num_tokens": 17295072.0,
+      "step": 1402
+    },
+    {
+      "entropy": 1.3918836414813995,
+      "epoch": 0.7388098999473407,
+      "grad_norm": 0.27023303508758545,
+      "learning_rate": 0.00019732590799437577,
+      "loss": 0.18338678777217865,
+      "mean_token_accuracy": 0.9283169507980347,
+      "num_tokens": 17307408.0,
+      "step": 1403
+    },
+    {
+      "entropy": 1.4320692718029022,
+      "epoch": 0.7393364928909952,
+      "grad_norm": 0.26482948660850525,
+      "learning_rate": 0.00019732059977833088,
+      "loss": 0.2143803983926773,
+      "mean_token_accuracy": 0.9123003780841827,
+      "num_tokens": 17319744.0,
+      "step": 1404
+    },
+    {
+      "entropy": 1.4552482962608337,
+      "epoch": 0.7398630858346498,
+      "grad_norm": 0.3084104359149933,
+      "learning_rate": 0.00019731528637857094,
+      "loss": 0.21455544233322144,
+      "mean_token_accuracy": 0.9187665581703186,
+      "num_tokens": 17332080.0,
+      "step": 1405
+    },
+    {
+      "entropy": 1.362955391407013,
+      "epoch": 0.7403896787783044,
+      "grad_norm": 0.2885385751724243,
+      "learning_rate": 0.00019730996779541133,
+      "loss": 0.2053847312927246,
+      "mean_token_accuracy": 0.9195720255374908,
+      "num_tokens": 17344416.0,
+      "step": 1406
+    },
+    {
+      "entropy": 1.3902620077133179,
+      "epoch": 0.740916271721959,
+      "grad_norm": 0.31122514605522156,
+      "learning_rate": 0.0001973046440291678,
+      "loss": 0.24145708978176117,
+      "mean_token_accuracy": 0.9063299000263214,
+      "num_tokens": 17356752.0,
+      "step": 1407
+    },
+    {
+      "entropy": 1.3320231139659882,
+      "epoch": 0.7414428646656135,
+      "grad_norm": 0.285426527261734,
+      "learning_rate": 0.00019729931508015647,
+      "loss": 0.202953040599823,
+      "mean_token_accuracy": 0.9166394174098969,
+      "num_tokens": 17369088.0,
+      "step": 1408
+    },
+    {
+      "entropy": 1.3167757391929626,
+      "epoch": 0.741969457609268,
+      "grad_norm": 0.3052743375301361,
+      "learning_rate": 0.00019729398094869358,
+      "loss": 0.23250755667686462,
+      "mean_token_accuracy": 0.909424439072609,
+      "num_tokens": 17381424.0,
+      "step": 1409
+    },
+    {
+      "entropy": 1.3000437319278717,
+      "epoch": 0.7424960505529226,
+      "grad_norm": 0.33441463112831116,
+      "learning_rate": 0.00019728864163509583,
+      "loss": 0.2160688042640686,
+      "mean_token_accuracy": 0.9094254523515701,
+      "num_tokens": 17393760.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.3271216750144958,
+      "epoch": 0.7430226434965771,
+      "grad_norm": 0.2913570702075958,
+      "learning_rate": 0.00019728329713968016,
+      "loss": 0.1972869336605072,
+      "mean_token_accuracy": 0.9217556267976761,
+      "num_tokens": 17406096.0,
+      "step": 1411
+    },
+    {
+      "entropy": 1.3546923398971558,
+      "epoch": 0.7435492364402317,
+      "grad_norm": 0.31364765763282776,
+      "learning_rate": 0.00019727794746276391,
+      "loss": 0.21945282816886902,
+      "mean_token_accuracy": 0.9129824340343475,
+      "num_tokens": 17418432.0,
+      "step": 1412
+    },
+    {
+      "entropy": 1.3115905821323395,
+      "epoch": 0.7440758293838863,
+      "grad_norm": 0.29262346029281616,
+      "learning_rate": 0.00019727259260466458,
+      "loss": 0.22911304235458374,
+      "mean_token_accuracy": 0.9053215533494949,
+      "num_tokens": 17430768.0,
+      "step": 1413
+    },
+    {
+      "entropy": 1.2379891574382782,
+      "epoch": 0.7446024223275408,
+      "grad_norm": 0.3259003162384033,
+      "learning_rate": 0.0001972672325657001,
+      "loss": 0.19887758791446686,
+      "mean_token_accuracy": 0.9181624054908752,
+      "num_tokens": 17443104.0,
+      "step": 1414
+    },
+    {
+      "entropy": 1.3647005259990692,
+      "epoch": 0.7451290152711953,
+      "grad_norm": 0.4454919695854187,
+      "learning_rate": 0.00019726186734618867,
+      "loss": 0.22842802107334137,
+      "mean_token_accuracy": 0.9110278934240341,
+      "num_tokens": 17455440.0,
+      "step": 1415
+    },
+    {
+      "entropy": 1.3383563160896301,
+      "epoch": 0.7456556082148499,
+      "grad_norm": 0.30020466446876526,
+      "learning_rate": 0.00019725649694644876,
+      "loss": 0.22231580317020416,
+      "mean_token_accuracy": 0.9075456112623215,
+      "num_tokens": 17467776.0,
+      "step": 1416
+    },
+    {
+      "entropy": 1.3714047372341156,
+      "epoch": 0.7461822011585044,
+      "grad_norm": 0.29304370284080505,
+      "learning_rate": 0.0001972511213667992,
+      "loss": 0.21095791459083557,
+      "mean_token_accuracy": 0.9166263341903687,
+      "num_tokens": 17480112.0,
+      "step": 1417
+    },
+    {
+      "entropy": 1.3512816727161407,
+      "epoch": 0.7467087941021591,
+      "grad_norm": 0.3256070613861084,
+      "learning_rate": 0.0001972457406075591,
+      "loss": 0.20153497159481049,
+      "mean_token_accuracy": 0.9194765239953995,
+      "num_tokens": 17492448.0,
+      "step": 1418
+    },
+    {
+      "entropy": 1.3885486125946045,
+      "epoch": 0.7472353870458136,
+      "grad_norm": 0.3231737017631531,
+      "learning_rate": 0.00019724035466904788,
+      "loss": 0.2227013111114502,
+      "mean_token_accuracy": 0.9109441936016083,
+      "num_tokens": 17504784.0,
+      "step": 1419
+    },
+    {
+      "entropy": 1.3755838871002197,
+      "epoch": 0.7477619799894681,
+      "grad_norm": 0.29787546396255493,
+      "learning_rate": 0.0001972349635515853,
+      "loss": 0.216201514005661,
+      "mean_token_accuracy": 0.9155356884002686,
+      "num_tokens": 17517120.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.437903732061386,
+      "epoch": 0.7482885729331227,
+      "grad_norm": 0.33993855118751526,
+      "learning_rate": 0.00019722956725549138,
+      "loss": 0.21523010730743408,
+      "mean_token_accuracy": 0.9125061631202698,
+      "num_tokens": 17529456.0,
+      "step": 1421
+    },
+    {
+      "entropy": 1.333466112613678,
+      "epoch": 0.7488151658767772,
+      "grad_norm": 0.33743515610694885,
+      "learning_rate": 0.00019722416578108642,
+      "loss": 0.22355730831623077,
+      "mean_token_accuracy": 0.9136557281017303,
+      "num_tokens": 17541792.0,
+      "step": 1422
+    },
+    {
+      "entropy": 1.360717535018921,
+      "epoch": 0.7493417588204319,
+      "grad_norm": 0.2994014024734497,
+      "learning_rate": 0.00019721875912869118,
+      "loss": 0.19619853794574738,
+      "mean_token_accuracy": 0.9230263978242874,
+      "num_tokens": 17554128.0,
+      "step": 1423
+    },
+    {
+      "entropy": 1.2954363822937012,
+      "epoch": 0.7498683517640864,
+      "grad_norm": 0.29989591240882874,
+      "learning_rate": 0.0001972133472986265,
+      "loss": 0.2181311547756195,
+      "mean_token_accuracy": 0.9170015305280685,
+      "num_tokens": 17566464.0,
+      "step": 1424
+    },
+    {
+      "entropy": 1.380856215953827,
+      "epoch": 0.7503949447077409,
+      "grad_norm": 0.28764021396636963,
+      "learning_rate": 0.00019720793029121377,
+      "loss": 0.21031011641025543,
+      "mean_token_accuracy": 0.9177572429180145,
+      "num_tokens": 17578800.0,
+      "step": 1425
+    },
+    {
+      "entropy": 1.3035545349121094,
+      "epoch": 0.7509215376513955,
+      "grad_norm": 0.2988784611225128,
+      "learning_rate": 0.00019720250810677446,
+      "loss": 0.20998632907867432,
+      "mean_token_accuracy": 0.9130989611148834,
+      "num_tokens": 17591136.0,
+      "step": 1426
+    },
+    {
+      "entropy": 1.289101094007492,
+      "epoch": 0.75144813059505,
+      "grad_norm": 0.49617183208465576,
+      "learning_rate": 0.0001971970807456305,
+      "loss": 0.21026289463043213,
+      "mean_token_accuracy": 0.9138355553150177,
+      "num_tokens": 17603472.0,
+      "step": 1427
+    },
+    {
+      "entropy": 1.3593553006649017,
+      "epoch": 0.7519747235387045,
+      "grad_norm": 0.3344785273075104,
+      "learning_rate": 0.0001971916482081041,
+      "loss": 0.20339743793010712,
+      "mean_token_accuracy": 0.9177296459674835,
+      "num_tokens": 17615808.0,
+      "step": 1428
+    },
+    {
+      "entropy": 1.3161483705043793,
+      "epoch": 0.7525013164823592,
+      "grad_norm": 0.3118573725223541,
+      "learning_rate": 0.00019718621049451772,
+      "loss": 0.2201179563999176,
+      "mean_token_accuracy": 0.9167029410600662,
+      "num_tokens": 17628144.0,
+      "step": 1429
+    },
+    {
+      "entropy": 1.3809407949447632,
+      "epoch": 0.7530279094260137,
+      "grad_norm": 0.3208526074886322,
+      "learning_rate": 0.00019718076760519423,
+      "loss": 0.23530805110931396,
+      "mean_token_accuracy": 0.9093427956104279,
+      "num_tokens": 17640480.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.3892112076282501,
+      "epoch": 0.7535545023696683,
+      "grad_norm": 0.2703491747379303,
+      "learning_rate": 0.00019717531954045663,
+      "loss": 0.2121361941099167,
+      "mean_token_accuracy": 0.916591003537178,
+      "num_tokens": 17652816.0,
+      "step": 1431
+    },
+    {
+      "entropy": 1.464809536933899,
+      "epoch": 0.7540810953133228,
+      "grad_norm": 0.2980828285217285,
+      "learning_rate": 0.00019716986630062842,
+      "loss": 0.21547271311283112,
+      "mean_token_accuracy": 0.9138219356536865,
+      "num_tokens": 17665152.0,
+      "step": 1432
+    },
+    {
+      "entropy": 1.4611221253871918,
+      "epoch": 0.7546076882569773,
+      "grad_norm": 0.3190062642097473,
+      "learning_rate": 0.00019716440788603332,
+      "loss": 0.22262822091579437,
+      "mean_token_accuracy": 0.9171289652585983,
+      "num_tokens": 17677488.0,
+      "step": 1433
+    },
+    {
+      "entropy": 1.4677778780460358,
+      "epoch": 0.7551342812006319,
+      "grad_norm": 0.2803551256656647,
+      "learning_rate": 0.00019715894429699537,
+      "loss": 0.2030690461397171,
+      "mean_token_accuracy": 0.9211814850568771,
+      "num_tokens": 17689824.0,
+      "step": 1434
+    },
+    {
+      "entropy": 1.546930730342865,
+      "epoch": 0.7556608741442865,
+      "grad_norm": 0.3004758954048157,
+      "learning_rate": 0.00019715347553383888,
+      "loss": 0.21929655969142914,
+      "mean_token_accuracy": 0.9147268533706665,
+      "num_tokens": 17702160.0,
+      "step": 1435
+    },
+    {
+      "entropy": 1.458657145500183,
+      "epoch": 0.756187467087941,
+      "grad_norm": 0.2918381094932556,
+      "learning_rate": 0.00019714800159688852,
+      "loss": 0.20412471890449524,
+      "mean_token_accuracy": 0.9208769351243973,
+      "num_tokens": 17714496.0,
+      "step": 1436
+    },
+    {
+      "entropy": 1.4661605954170227,
+      "epoch": 0.7567140600315956,
+      "grad_norm": 0.28824299573898315,
+      "learning_rate": 0.00019714252248646924,
+      "loss": 0.19737893342971802,
+      "mean_token_accuracy": 0.9271862953901291,
+      "num_tokens": 17726832.0,
+      "step": 1437
+    },
+    {
+      "entropy": 1.4540847837924957,
+      "epoch": 0.7572406529752501,
+      "grad_norm": 0.2962500751018524,
+      "learning_rate": 0.00019713703820290634,
+      "loss": 0.21638396382331848,
+      "mean_token_accuracy": 0.9159590601921082,
+      "num_tokens": 17739168.0,
+      "step": 1438
+    },
+    {
+      "entropy": 1.4231859147548676,
+      "epoch": 0.7577672459189047,
+      "grad_norm": 0.29495999217033386,
+      "learning_rate": 0.00019713154874652532,
+      "loss": 0.2159111201763153,
+      "mean_token_accuracy": 0.9145191609859467,
+      "num_tokens": 17751504.0,
+      "step": 1439
+    },
+    {
+      "entropy": 1.3984394371509552,
+      "epoch": 0.7582938388625592,
+      "grad_norm": 0.3085009753704071,
+      "learning_rate": 0.0001971260541176521,
+      "loss": 0.2231639325618744,
+      "mean_token_accuracy": 0.9079516977071762,
+      "num_tokens": 17763840.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.4092283248901367,
+      "epoch": 0.7588204318062138,
+      "grad_norm": 0.31046876311302185,
+      "learning_rate": 0.00019712055431661286,
+      "loss": 0.21453410387039185,
+      "mean_token_accuracy": 0.916040763258934,
+      "num_tokens": 17776176.0,
+      "step": 1441
+    },
+    {
+      "entropy": 1.3918389678001404,
+      "epoch": 0.7593470247498684,
+      "grad_norm": 0.3046187460422516,
+      "learning_rate": 0.00019711504934373408,
+      "loss": 0.20514801144599915,
+      "mean_token_accuracy": 0.9181930124759674,
+      "num_tokens": 17788512.0,
+      "step": 1442
+    },
+    {
+      "entropy": 1.3869245946407318,
+      "epoch": 0.7598736176935229,
+      "grad_norm": 0.31358590722084045,
+      "learning_rate": 0.00019710953919934256,
+      "loss": 0.22413912415504456,
+      "mean_token_accuracy": 0.9102539569139481,
+      "num_tokens": 17800848.0,
+      "step": 1443
+    },
+    {
+      "entropy": 1.3643006086349487,
+      "epoch": 0.7604002106371774,
+      "grad_norm": 0.27224990725517273,
+      "learning_rate": 0.00019710402388376544,
+      "loss": 0.20147213339805603,
+      "mean_token_accuracy": 0.9196710139513016,
+      "num_tokens": 17813184.0,
+      "step": 1444
+    },
+    {
+      "entropy": 1.39336559176445,
+      "epoch": 0.760926803580832,
+      "grad_norm": 0.3104258179664612,
+      "learning_rate": 0.00019709850339733008,
+      "loss": 0.21504773199558258,
+      "mean_token_accuracy": 0.9135936945676804,
+      "num_tokens": 17825520.0,
+      "step": 1445
+    },
+    {
+      "entropy": 1.3767728209495544,
+      "epoch": 0.7614533965244866,
+      "grad_norm": 0.3152048587799072,
+      "learning_rate": 0.0001970929777403642,
+      "loss": 0.221971333026886,
+      "mean_token_accuracy": 0.911506250500679,
+      "num_tokens": 17837856.0,
+      "step": 1446
+    },
+    {
+      "entropy": 1.4271937906742096,
+      "epoch": 0.7619799894681412,
+      "grad_norm": 0.3253970444202423,
+      "learning_rate": 0.00019708744691319588,
+      "loss": 0.2253606915473938,
+      "mean_token_accuracy": 0.9094466418027878,
+      "num_tokens": 17850192.0,
+      "step": 1447
+    },
+    {
+      "entropy": 1.3933376669883728,
+      "epoch": 0.7625065824117957,
+      "grad_norm": 0.3199850022792816,
+      "learning_rate": 0.00019708191091615343,
+      "loss": 0.21662366390228271,
+      "mean_token_accuracy": 0.9175620973110199,
+      "num_tokens": 17862528.0,
+      "step": 1448
+    },
+    {
+      "entropy": 1.433942288160324,
+      "epoch": 0.7630331753554502,
+      "grad_norm": 0.2839816212654114,
+      "learning_rate": 0.00019707636974956546,
+      "loss": 0.20464417338371277,
+      "mean_token_accuracy": 0.9186283946037292,
+      "num_tokens": 17874864.0,
+      "step": 1449
+    },
+    {
+      "entropy": 1.3883062601089478,
+      "epoch": 0.7635597682991048,
+      "grad_norm": 0.3059125542640686,
+      "learning_rate": 0.00019707082341376093,
+      "loss": 0.21790815889835358,
+      "mean_token_accuracy": 0.9127152860164642,
+      "num_tokens": 17887200.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.4840424954891205,
+      "epoch": 0.7640863612427593,
+      "grad_norm": 0.317634642124176,
+      "learning_rate": 0.0001970652719090691,
+      "loss": 0.22078271210193634,
+      "mean_token_accuracy": 0.9099593609571457,
+      "num_tokens": 17899536.0,
+      "step": 1451
+    },
+    {
+      "entropy": 1.373614490032196,
+      "epoch": 0.764612954186414,
+      "grad_norm": 0.2764304280281067,
+      "learning_rate": 0.00019705971523581957,
+      "loss": 0.20861467719078064,
+      "mean_token_accuracy": 0.9210799187421799,
+      "num_tokens": 17911872.0,
+      "step": 1452
+    },
+    {
+      "entropy": 1.4546839594841003,
+      "epoch": 0.7651395471300685,
+      "grad_norm": 0.30617302656173706,
+      "learning_rate": 0.00019705415339434212,
+      "loss": 0.21506822109222412,
+      "mean_token_accuracy": 0.9104336351156235,
+      "num_tokens": 17924208.0,
+      "step": 1453
+    },
+    {
+      "entropy": 1.46762016415596,
+      "epoch": 0.765666140073723,
+      "grad_norm": 0.33141979575157166,
+      "learning_rate": 0.000197048586384967,
+      "loss": 0.23449833691120148,
+      "mean_token_accuracy": 0.909214973449707,
+      "num_tokens": 17936544.0,
+      "step": 1454
+    },
+    {
+      "entropy": 1.3759736120700836,
+      "epoch": 0.7661927330173776,
+      "grad_norm": 0.28817251324653625,
+      "learning_rate": 0.00019704301420802467,
+      "loss": 0.2031758576631546,
+      "mean_token_accuracy": 0.9166503101587296,
+      "num_tokens": 17948880.0,
+      "step": 1455
+    },
+    {
+      "entropy": 1.4604673385620117,
+      "epoch": 0.7667193259610321,
+      "grad_norm": 0.28507548570632935,
+      "learning_rate": 0.0001970374368638459,
+      "loss": 0.21905720233917236,
+      "mean_token_accuracy": 0.9121448248624802,
+      "num_tokens": 17961216.0,
+      "step": 1456
+    },
+    {
+      "entropy": 1.4368546605110168,
+      "epoch": 0.7672459189046866,
+      "grad_norm": 0.3244202435016632,
+      "learning_rate": 0.00019703185435276179,
+      "loss": 0.21986661851406097,
+      "mean_token_accuracy": 0.9103575944900513,
+      "num_tokens": 17973552.0,
+      "step": 1457
+    },
+    {
+      "entropy": 1.4183049201965332,
+      "epoch": 0.7677725118483413,
+      "grad_norm": 0.3483145833015442,
+      "learning_rate": 0.00019702626667510376,
+      "loss": 0.2258511334657669,
+      "mean_token_accuracy": 0.9121571332216263,
+      "num_tokens": 17985888.0,
+      "step": 1458
+    },
+    {
+      "entropy": 1.4387763142585754,
+      "epoch": 0.7682991047919958,
+      "grad_norm": 0.30231183767318726,
+      "learning_rate": 0.00019702067383120352,
+      "loss": 0.20727157592773438,
+      "mean_token_accuracy": 0.9170671254396439,
+      "num_tokens": 17998224.0,
+      "step": 1459
+    },
+    {
+      "entropy": 1.3991906344890594,
+      "epoch": 0.7688256977356503,
+      "grad_norm": 0.3081129789352417,
+      "learning_rate": 0.00019701507582139304,
+      "loss": 0.2180902510881424,
+      "mean_token_accuracy": 0.916274681687355,
+      "num_tokens": 18010560.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.4099707305431366,
+      "epoch": 0.7693522906793049,
+      "grad_norm": 0.27477627992630005,
+      "learning_rate": 0.00019700947264600466,
+      "loss": 0.19322465360164642,
+      "mean_token_accuracy": 0.9191512167453766,
+      "num_tokens": 18022896.0,
+      "step": 1461
+    },
+    {
+      "entropy": 1.4558897018432617,
+      "epoch": 0.7698788836229594,
+      "grad_norm": 0.2677430212497711,
+      "learning_rate": 0.00019700386430537105,
+      "loss": 0.21823228895664215,
+      "mean_token_accuracy": 0.9123710244894028,
+      "num_tokens": 18035232.0,
+      "step": 1462
+    },
+    {
+      "entropy": 1.4174329936504364,
+      "epoch": 0.770405476566614,
+      "grad_norm": 0.2713851034641266,
+      "learning_rate": 0.00019699825079982512,
+      "loss": 0.19724853336811066,
+      "mean_token_accuracy": 0.9200254529714584,
+      "num_tokens": 18047568.0,
+      "step": 1463
+    },
+    {
+      "entropy": 1.4575957953929901,
+      "epoch": 0.7709320695102686,
+      "grad_norm": 0.30220046639442444,
+      "learning_rate": 0.00019699263212970007,
+      "loss": 0.20754371583461761,
+      "mean_token_accuracy": 0.9198533892631531,
+      "num_tokens": 18059904.0,
+      "step": 1464
+    },
+    {
+      "entropy": 1.450849562883377,
+      "epoch": 0.7714586624539231,
+      "grad_norm": 0.28346338868141174,
+      "learning_rate": 0.0001969870082953295,
+      "loss": 0.19828161597251892,
+      "mean_token_accuracy": 0.9172644019126892,
+      "num_tokens": 18072240.0,
+      "step": 1465
+    },
+    {
+      "entropy": 1.4658414423465729,
+      "epoch": 0.7719852553975777,
+      "grad_norm": 0.3004949986934662,
+      "learning_rate": 0.00019698137929704724,
+      "loss": 0.20767343044281006,
+      "mean_token_accuracy": 0.9210254102945328,
+      "num_tokens": 18084576.0,
+      "step": 1466
+    },
+    {
+      "entropy": 1.4050607681274414,
+      "epoch": 0.7725118483412322,
+      "grad_norm": 0.2896452248096466,
+      "learning_rate": 0.00019697574513518747,
+      "loss": 0.20795601606369019,
+      "mean_token_accuracy": 0.914929062128067,
+      "num_tokens": 18096912.0,
+      "step": 1467
+    },
+    {
+      "entropy": 1.4351738393306732,
+      "epoch": 0.7730384412848867,
+      "grad_norm": 0.34425756335258484,
+      "learning_rate": 0.00019697010581008463,
+      "loss": 0.21601206064224243,
+      "mean_token_accuracy": 0.9149073958396912,
+      "num_tokens": 18109248.0,
+      "step": 1468
+    },
+    {
+      "entropy": 1.3905524611473083,
+      "epoch": 0.7735650342285414,
+      "grad_norm": 0.3009505271911621,
+      "learning_rate": 0.00019696446132207353,
+      "loss": 0.19988638162612915,
+      "mean_token_accuracy": 0.9213133901357651,
+      "num_tokens": 18121584.0,
+      "step": 1469
+    },
+    {
+      "entropy": 1.4411927461624146,
+      "epoch": 0.7740916271721959,
+      "grad_norm": 0.3239229619503021,
+      "learning_rate": 0.0001969588116714892,
+      "loss": 0.2117571234703064,
+      "mean_token_accuracy": 0.9106622785329819,
+      "num_tokens": 18133920.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.4064708650112152,
+      "epoch": 0.7746182201158505,
+      "grad_norm": 0.3362599015235901,
+      "learning_rate": 0.00019695315685866707,
+      "loss": 0.23721855878829956,
+      "mean_token_accuracy": 0.9070688337087631,
+      "num_tokens": 18146256.0,
+      "step": 1471
+    },
+    {
+      "entropy": 1.4237781167030334,
+      "epoch": 0.775144813059505,
+      "grad_norm": 0.3095615804195404,
+      "learning_rate": 0.00019694749688394281,
+      "loss": 0.2411169707775116,
+      "mean_token_accuracy": 0.9083186686038971,
+      "num_tokens": 18158592.0,
+      "step": 1472
+    },
+    {
+      "entropy": 1.3879849016666412,
+      "epoch": 0.7756714060031595,
+      "grad_norm": 0.28509050607681274,
+      "learning_rate": 0.00019694183174765245,
+      "loss": 0.22019074857234955,
+      "mean_token_accuracy": 0.911543145775795,
+      "num_tokens": 18170928.0,
+      "step": 1473
+    },
+    {
+      "entropy": 1.3992516100406647,
+      "epoch": 0.7761979989468141,
+      "grad_norm": 0.3135905861854553,
+      "learning_rate": 0.00019693616145013227,
+      "loss": 0.2088148593902588,
+      "mean_token_accuracy": 0.9130579680204391,
+      "num_tokens": 18183264.0,
+      "step": 1474
+    },
+    {
+      "entropy": 1.3637616336345673,
+      "epoch": 0.7767245918904687,
+      "grad_norm": 0.26418524980545044,
+      "learning_rate": 0.00019693048599171887,
+      "loss": 0.20479831099510193,
+      "mean_token_accuracy": 0.916004553437233,
+      "num_tokens": 18195600.0,
+      "step": 1475
+    },
+    {
+      "entropy": 1.4390030205249786,
+      "epoch": 0.7772511848341233,
+      "grad_norm": 0.290937215089798,
+      "learning_rate": 0.0001969248053727492,
+      "loss": 0.21481530368328094,
+      "mean_token_accuracy": 0.9162030220031738,
+      "num_tokens": 18207936.0,
+      "step": 1476
+    },
+    {
+      "entropy": 1.4204131364822388,
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.3285602331161499,
+      "learning_rate": 0.00019691911959356044,
+      "loss": 0.19588018953800201,
+      "mean_token_accuracy": 0.9232761710882187,
+      "num_tokens": 18220272.0,
+      "step": 1477
+    },
+    {
+      "entropy": 1.4693542420864105,
+      "epoch": 0.7783043707214323,
+      "grad_norm": 0.3064950704574585,
+      "learning_rate": 0.0001969134286544902,
+      "loss": 0.20760740339756012,
+      "mean_token_accuracy": 0.9132848978042603,
+      "num_tokens": 18232608.0,
+      "step": 1478
+    },
+    {
+      "entropy": 1.3828904628753662,
+      "epoch": 0.7788309636650869,
+      "grad_norm": 0.3061995506286621,
+      "learning_rate": 0.00019690773255587625,
+      "loss": 0.2265746295452118,
+      "mean_token_accuracy": 0.9108180850744247,
+      "num_tokens": 18244944.0,
+      "step": 1479
+    },
+    {
+      "entropy": 1.3849403858184814,
+      "epoch": 0.7793575566087414,
+      "grad_norm": 0.3251616656780243,
+      "learning_rate": 0.00019690203129805672,
+      "loss": 0.2233261615037918,
+      "mean_token_accuracy": 0.9125779420137405,
+      "num_tokens": 18257280.0,
+      "step": 1480
+    },
+    {
+      "entropy": 1.3147023022174835,
+      "epoch": 0.779884149552396,
+      "grad_norm": 0.31116747856140137,
+      "learning_rate": 0.00019689632488137016,
+      "loss": 0.24036768078804016,
+      "mean_token_accuracy": 0.906961664557457,
+      "num_tokens": 18269616.0,
+      "step": 1481
+    },
+    {
+      "entropy": 1.3186087906360626,
+      "epoch": 0.7804107424960506,
+      "grad_norm": 0.30934205651283264,
+      "learning_rate": 0.0001968906133061552,
+      "loss": 0.21348203718662262,
+      "mean_token_accuracy": 0.9163684844970703,
+      "num_tokens": 18281952.0,
+      "step": 1482
+    },
+    {
+      "entropy": 1.3171938359737396,
+      "epoch": 0.7809373354397051,
+      "grad_norm": 0.3076079487800598,
+      "learning_rate": 0.00019688489657275103,
+      "loss": 0.22806760668754578,
+      "mean_token_accuracy": 0.9050341248512268,
+      "num_tokens": 18294288.0,
+      "step": 1483
+    },
+    {
+      "entropy": 1.331948697566986,
+      "epoch": 0.7814639283833597,
+      "grad_norm": 0.3263510465621948,
+      "learning_rate": 0.0001968791746814969,
+      "loss": 0.2278282344341278,
+      "mean_token_accuracy": 0.9075683802366257,
+      "num_tokens": 18306624.0,
+      "step": 1484
+    },
+    {
+      "entropy": 1.3975997269153595,
+      "epoch": 0.7819905213270142,
+      "grad_norm": 0.37345680594444275,
+      "learning_rate": 0.00019687344763273255,
+      "loss": 0.22596792876720428,
+      "mean_token_accuracy": 0.913162499666214,
+      "num_tokens": 18318960.0,
+      "step": 1485
+    },
+    {
+      "entropy": 1.3315426707267761,
+      "epoch": 0.7825171142706687,
+      "grad_norm": 0.28193777799606323,
+      "learning_rate": 0.00019686771542679797,
+      "loss": 0.1983056217432022,
+      "mean_token_accuracy": 0.9222036451101303,
+      "num_tokens": 18331296.0,
+      "step": 1486
+    },
+    {
+      "entropy": 1.4003748893737793,
+      "epoch": 0.7830437072143234,
+      "grad_norm": 0.30775102972984314,
+      "learning_rate": 0.00019686197806403343,
+      "loss": 0.22822785377502441,
+      "mean_token_accuracy": 0.9109926074743271,
+      "num_tokens": 18343632.0,
+      "step": 1487
+    },
+    {
+      "entropy": 1.3931090831756592,
+      "epoch": 0.7835703001579779,
+      "grad_norm": 0.2824244201183319,
+      "learning_rate": 0.0001968562355447795,
+      "loss": 0.20619836449623108,
+      "mean_token_accuracy": 0.9165951460599899,
+      "num_tokens": 18355968.0,
+      "step": 1488
+    },
+    {
+      "entropy": 1.3029468953609467,
+      "epoch": 0.7840968931016324,
+      "grad_norm": 0.26663464307785034,
+      "learning_rate": 0.00019685048786937713,
+      "loss": 0.2070493847131729,
+      "mean_token_accuracy": 0.91603122651577,
+      "num_tokens": 18368304.0,
+      "step": 1489
+    },
+    {
+      "entropy": 1.4118830561637878,
+      "epoch": 0.784623486045287,
+      "grad_norm": 0.2948478162288666,
+      "learning_rate": 0.00019684473503816746,
+      "loss": 0.2121935933828354,
+      "mean_token_accuracy": 0.9169209152460098,
+      "num_tokens": 18380640.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.5040743947029114,
+      "epoch": 0.7851500789889415,
+      "grad_norm": 0.33927392959594727,
+      "learning_rate": 0.00019683897705149208,
+      "loss": 0.23608259856700897,
+      "mean_token_accuracy": 0.9112606197595596,
+      "num_tokens": 18392976.0,
+      "step": 1491
+    },
+    {
+      "entropy": 1.388082355260849,
+      "epoch": 0.785676671932596,
+      "grad_norm": 0.2748982012271881,
+      "learning_rate": 0.0001968332139096927,
+      "loss": 0.21956929564476013,
+      "mean_token_accuracy": 0.9111993908882141,
+      "num_tokens": 18405312.0,
+      "step": 1492
+    },
+    {
+      "entropy": 1.419054538011551,
+      "epoch": 0.7862032648762507,
+      "grad_norm": 0.2975718379020691,
+      "learning_rate": 0.00019682744561311163,
+      "loss": 0.20622003078460693,
+      "mean_token_accuracy": 0.91482774913311,
+      "num_tokens": 18417648.0,
+      "step": 1493
+    },
+    {
+      "entropy": 1.3681580722332,
+      "epoch": 0.7867298578199052,
+      "grad_norm": 0.30167657136917114,
+      "learning_rate": 0.00019682167216209106,
+      "loss": 0.22527143359184265,
+      "mean_token_accuracy": 0.9118600189685822,
+      "num_tokens": 18429984.0,
+      "step": 1494
+    },
+    {
+      "entropy": 1.3992347717285156,
+      "epoch": 0.7872564507635598,
+      "grad_norm": 0.3023481070995331,
+      "learning_rate": 0.00019681589355697394,
+      "loss": 0.20631316304206848,
+      "mean_token_accuracy": 0.9144344329833984,
+      "num_tokens": 18442320.0,
+      "step": 1495
+    },
+    {
+      "entropy": 1.2978556156158447,
+      "epoch": 0.7877830437072143,
+      "grad_norm": 0.275492399930954,
+      "learning_rate": 0.00019681010979810318,
+      "loss": 0.2021515965461731,
+      "mean_token_accuracy": 0.9172556847333908,
+      "num_tokens": 18454656.0,
+      "step": 1496
+    },
+    {
+      "entropy": 1.2966987490653992,
+      "epoch": 0.7883096366508688,
+      "grad_norm": 0.3461536169052124,
+      "learning_rate": 0.00019680432088582217,
+      "loss": 0.22693704068660736,
+      "mean_token_accuracy": 0.9091721177101135,
+      "num_tokens": 18466992.0,
+      "step": 1497
+    },
+    {
+      "entropy": 1.304772973060608,
+      "epoch": 0.7888362295945235,
+      "grad_norm": 0.32779547572135925,
+      "learning_rate": 0.00019679852682047457,
+      "loss": 0.21427524089813232,
+      "mean_token_accuracy": 0.9151846766471863,
+      "num_tokens": 18479328.0,
+      "step": 1498
+    },
+    {
+      "entropy": 1.2882267534732819,
+      "epoch": 0.789362822538178,
+      "grad_norm": 0.3116947114467621,
+      "learning_rate": 0.00019679272760240435,
+      "loss": 0.20784853398799896,
+      "mean_token_accuracy": 0.9149093627929688,
+      "num_tokens": 18491664.0,
+      "step": 1499
+    },
+    {
+      "entropy": 1.263739138841629,
+      "epoch": 0.7898894154818326,
+      "grad_norm": 0.443194717168808,
+      "learning_rate": 0.0001967869232319557,
+      "loss": 0.23977030813694,
+      "mean_token_accuracy": 0.9084160029888153,
+      "num_tokens": 18504000.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.2093487679958344,
+      "epoch": 0.7904160084254871,
+      "grad_norm": 0.29420503973960876,
+      "learning_rate": 0.0001967811137094733,
+      "loss": 0.22335118055343628,
+      "mean_token_accuracy": 0.909618929028511,
+      "num_tokens": 18516336.0,
+      "step": 1501
+    },
+    {
+      "entropy": 1.271395206451416,
+      "epoch": 0.7909426013691416,
+      "grad_norm": 0.5013836026191711,
+      "learning_rate": 0.00019677529903530197,
+      "loss": 0.2110762596130371,
+      "mean_token_accuracy": 0.9183898121118546,
+      "num_tokens": 18528672.0,
+      "step": 1502
+    },
+    {
+      "entropy": 1.2578312754631042,
+      "epoch": 0.7914691943127962,
+      "grad_norm": 0.37272414565086365,
+      "learning_rate": 0.0001967694792097869,
+      "loss": 0.22164121270179749,
+      "mean_token_accuracy": 0.9116562008857727,
+      "num_tokens": 18541008.0,
+      "step": 1503
+    },
+    {
+      "entropy": 1.2144514918327332,
+      "epoch": 0.7919957872564508,
+      "grad_norm": 0.3145962059497833,
+      "learning_rate": 0.0001967636542332736,
+      "loss": 0.22305795550346375,
+      "mean_token_accuracy": 0.9156023114919662,
+      "num_tokens": 18553344.0,
+      "step": 1504
+    },
+    {
+      "entropy": 1.2401383221149445,
+      "epoch": 0.7925223802001053,
+      "grad_norm": 0.2835533022880554,
+      "learning_rate": 0.00019675782410610778,
+      "loss": 0.19279327988624573,
+      "mean_token_accuracy": 0.926255002617836,
+      "num_tokens": 18565680.0,
+      "step": 1505
+    },
+    {
+      "entropy": 1.2049389481544495,
+      "epoch": 0.7930489731437599,
+      "grad_norm": 0.2632310092449188,
+      "learning_rate": 0.00019675198882863567,
+      "loss": 0.2010316103696823,
+      "mean_token_accuracy": 0.91452956199646,
+      "num_tokens": 18578016.0,
+      "step": 1506
+    },
+    {
+      "entropy": 1.249726414680481,
+      "epoch": 0.7935755660874144,
+      "grad_norm": 0.30521342158317566,
+      "learning_rate": 0.0001967461484012036,
+      "loss": 0.23277787864208221,
+      "mean_token_accuracy": 0.9076357632875443,
+      "num_tokens": 18590352.0,
+      "step": 1507
+    },
+    {
+      "entropy": 1.1978009939193726,
+      "epoch": 0.794102159031069,
+      "grad_norm": 0.2853034734725952,
+      "learning_rate": 0.00019674030282415826,
+      "loss": 0.2127259522676468,
+      "mean_token_accuracy": 0.9155232310295105,
+      "num_tokens": 18602688.0,
+      "step": 1508
+    },
+    {
+      "entropy": 1.25815749168396,
+      "epoch": 0.7946287519747235,
+      "grad_norm": 0.28645238280296326,
+      "learning_rate": 0.00019673445209784677,
+      "loss": 0.22002597153186798,
+      "mean_token_accuracy": 0.9130334705114365,
+      "num_tokens": 18615024.0,
+      "step": 1509
+    },
+    {
+      "entropy": 1.3180667161941528,
+      "epoch": 0.7951553449183781,
+      "grad_norm": 0.2906380891799927,
+      "learning_rate": 0.00019672859622261633,
+      "loss": 0.22877947986125946,
+      "mean_token_accuracy": 0.912013441324234,
+      "num_tokens": 18627360.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.2853264808654785,
+      "epoch": 0.7956819378620327,
+      "grad_norm": 0.28856080770492554,
+      "learning_rate": 0.00019672273519881465,
+      "loss": 0.2222394347190857,
+      "mean_token_accuracy": 0.9078328162431717,
+      "num_tokens": 18639696.0,
+      "step": 1511
+    },
+    {
+      "entropy": 1.2696054577827454,
+      "epoch": 0.7962085308056872,
+      "grad_norm": 0.30683693289756775,
+      "learning_rate": 0.0001967168690267896,
+      "loss": 0.22363853454589844,
+      "mean_token_accuracy": 0.9149999022483826,
+      "num_tokens": 18652032.0,
+      "step": 1512
+    },
+    {
+      "entropy": 1.272040843963623,
+      "epoch": 0.7967351237493417,
+      "grad_norm": 0.3236723840236664,
+      "learning_rate": 0.0001967109977068895,
+      "loss": 0.21476420760154724,
+      "mean_token_accuracy": 0.9179616123437881,
+      "num_tokens": 18664368.0,
+      "step": 1513
+    },
+    {
+      "entropy": 1.260970801115036,
+      "epoch": 0.7972617166929963,
+      "grad_norm": 0.30413469672203064,
+      "learning_rate": 0.00019670512123946284,
+      "loss": 0.22492626309394836,
+      "mean_token_accuracy": 0.9107203483581543,
+      "num_tokens": 18676704.0,
+      "step": 1514
+    },
+    {
+      "entropy": 1.2300290167331696,
+      "epoch": 0.7977883096366508,
+      "grad_norm": 0.2895817458629608,
+      "learning_rate": 0.00019669923962485847,
+      "loss": 0.2138168066740036,
+      "mean_token_accuracy": 0.9164409637451172,
+      "num_tokens": 18689040.0,
+      "step": 1515
+    },
+    {
+      "entropy": 1.2246987521648407,
+      "epoch": 0.7983149025803055,
+      "grad_norm": 0.30868765711784363,
+      "learning_rate": 0.0001966933528634256,
+      "loss": 0.2062627673149109,
+      "mean_token_accuracy": 0.9158836156129837,
+      "num_tokens": 18701376.0,
+      "step": 1516
+    },
+    {
+      "entropy": 1.24353489279747,
+      "epoch": 0.79884149552396,
+      "grad_norm": 0.2896019518375397,
+      "learning_rate": 0.00019668746095551364,
+      "loss": 0.2127014845609665,
+      "mean_token_accuracy": 0.9154282063245773,
+      "num_tokens": 18713712.0,
+      "step": 1517
+    },
+    {
+      "entropy": 1.219551146030426,
+      "epoch": 0.7993680884676145,
+      "grad_norm": 0.5301105380058289,
+      "learning_rate": 0.0001966815639014724,
+      "loss": 0.20765289664268494,
+      "mean_token_accuracy": 0.9193731993436813,
+      "num_tokens": 18726048.0,
+      "step": 1518
+    },
+    {
+      "entropy": 1.2230626046657562,
+      "epoch": 0.7998946814112691,
+      "grad_norm": 0.2891480624675751,
+      "learning_rate": 0.00019667566170165194,
+      "loss": 0.20567816495895386,
+      "mean_token_accuracy": 0.9149514883756638,
+      "num_tokens": 18738384.0,
+      "step": 1519
+    },
+    {
+      "entropy": 1.1955215632915497,
+      "epoch": 0.8004212743549236,
+      "grad_norm": 0.3952074646949768,
+      "learning_rate": 0.00019666975435640258,
+      "loss": 0.23236501216888428,
+      "mean_token_accuracy": 0.910706028342247,
+      "num_tokens": 18750720.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.1208319664001465,
+      "epoch": 0.8009478672985783,
+      "grad_norm": 0.3569393455982208,
+      "learning_rate": 0.00019666384186607506,
+      "loss": 0.2112501859664917,
+      "mean_token_accuracy": 0.9126569777727127,
+      "num_tokens": 18763056.0,
+      "step": 1521
+    },
+    {
+      "entropy": 1.1377290189266205,
+      "epoch": 0.8014744602422328,
+      "grad_norm": 0.2929931879043579,
+      "learning_rate": 0.00019665792423102037,
+      "loss": 0.22051754593849182,
+      "mean_token_accuracy": 0.9091715216636658,
+      "num_tokens": 18775392.0,
+      "step": 1522
+    },
+    {
+      "entropy": 1.1271924078464508,
+      "epoch": 0.8020010531858873,
+      "grad_norm": 0.3465351164340973,
+      "learning_rate": 0.0001966520014515898,
+      "loss": 0.22917094826698303,
+      "mean_token_accuracy": 0.9075794517993927,
+      "num_tokens": 18787728.0,
+      "step": 1523
+    },
+    {
+      "entropy": 1.1986294984817505,
+      "epoch": 0.8025276461295419,
+      "grad_norm": 0.32447630167007446,
+      "learning_rate": 0.00019664607352813494,
+      "loss": 0.23350387811660767,
+      "mean_token_accuracy": 0.9076796770095825,
+      "num_tokens": 18800064.0,
+      "step": 1524
+    },
+    {
+      "entropy": 1.1514316499233246,
+      "epoch": 0.8030542390731964,
+      "grad_norm": 0.3304944932460785,
+      "learning_rate": 0.00019664014046100775,
+      "loss": 0.21476618945598602,
+      "mean_token_accuracy": 0.913004919886589,
+      "num_tokens": 18812400.0,
+      "step": 1525
+    },
+    {
+      "entropy": 1.1340425908565521,
+      "epoch": 0.8035808320168509,
+      "grad_norm": 0.5441511869430542,
+      "learning_rate": 0.00019663420225056035,
+      "loss": 0.22214046120643616,
+      "mean_token_accuracy": 0.9116765111684799,
+      "num_tokens": 18824736.0,
+      "step": 1526
+    },
+    {
+      "entropy": 1.0562350451946259,
+      "epoch": 0.8041074249605056,
+      "grad_norm": 0.3077342212200165,
+      "learning_rate": 0.00019662825889714533,
+      "loss": 0.21319684386253357,
+      "mean_token_accuracy": 0.918745294213295,
+      "num_tokens": 18837072.0,
+      "step": 1527
+    },
+    {
+      "entropy": 1.0274389684200287,
+      "epoch": 0.8046340179041601,
+      "grad_norm": 0.3354491889476776,
+      "learning_rate": 0.0001966223104011155,
+      "loss": 0.2296486645936966,
+      "mean_token_accuracy": 0.9057023078203201,
+      "num_tokens": 18849408.0,
+      "step": 1528
+    },
+    {
+      "entropy": 1.0815512537956238,
+      "epoch": 0.8051606108478147,
+      "grad_norm": 0.3144749402999878,
+      "learning_rate": 0.00019661635676282394,
+      "loss": 0.229911670088768,
+      "mean_token_accuracy": 0.9100077450275421,
+      "num_tokens": 18861744.0,
+      "step": 1529
+    },
+    {
+      "entropy": 1.0930194556713104,
+      "epoch": 0.8056872037914692,
+      "grad_norm": 0.300818532705307,
+      "learning_rate": 0.00019661039798262415,
+      "loss": 0.20994170010089874,
+      "mean_token_accuracy": 0.9187563508749008,
+      "num_tokens": 18874080.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0869486331939697,
+      "epoch": 0.8062137967351237,
+      "grad_norm": 0.28593873977661133,
+      "learning_rate": 0.00019660443406086984,
+      "loss": 0.20231741666793823,
+      "mean_token_accuracy": 0.918257012963295,
+      "num_tokens": 18886416.0,
+      "step": 1531
+    },
+    {
+      "entropy": 1.0835871398448944,
+      "epoch": 0.8067403896787783,
+      "grad_norm": 0.7419033050537109,
+      "learning_rate": 0.000196598464997915,
+      "loss": 0.2157197892665863,
+      "mean_token_accuracy": 0.9135694205760956,
+      "num_tokens": 18898752.0,
+      "step": 1532
+    },
+    {
+      "entropy": 1.1274356842041016,
+      "epoch": 0.8072669826224329,
+      "grad_norm": 0.31080278754234314,
+      "learning_rate": 0.00019659249079411405,
+      "loss": 0.22466787695884705,
+      "mean_token_accuracy": 0.9107064604759216,
+      "num_tokens": 18911088.0,
+      "step": 1533
+    },
+    {
+      "entropy": 1.0774615406990051,
+      "epoch": 0.8077935755660874,
+      "grad_norm": 0.28806376457214355,
+      "learning_rate": 0.00019658651144982163,
+      "loss": 0.2336619794368744,
+      "mean_token_accuracy": 0.9125310033559799,
+      "num_tokens": 18923424.0,
+      "step": 1534
+    },
+    {
+      "entropy": 1.0493130534887314,
+      "epoch": 0.808320168509742,
+      "grad_norm": 0.2861652672290802,
+      "learning_rate": 0.0001965805269653927,
+      "loss": 0.20115122199058533,
+      "mean_token_accuracy": 0.919718474149704,
+      "num_tokens": 18935760.0,
+      "step": 1535
+    },
+    {
+      "entropy": 1.0706775784492493,
+      "epoch": 0.8088467614533965,
+      "grad_norm": 0.29678940773010254,
+      "learning_rate": 0.0001965745373411825,
+      "loss": 0.21643559634685516,
+      "mean_token_accuracy": 0.9111261665821075,
+      "num_tokens": 18948096.0,
+      "step": 1536
+    },
+    {
+      "entropy": 1.0705525279045105,
+      "epoch": 0.809373354397051,
+      "grad_norm": 0.3051356077194214,
+      "learning_rate": 0.0001965685425775466,
+      "loss": 0.2151341289281845,
+      "mean_token_accuracy": 0.9130971729755402,
+      "num_tokens": 18960432.0,
+      "step": 1537
+    },
+    {
+      "entropy": 1.0797742903232574,
+      "epoch": 0.8098999473407056,
+      "grad_norm": 0.29696813225746155,
+      "learning_rate": 0.0001965625426748409,
+      "loss": 0.22983068227767944,
+      "mean_token_accuracy": 0.9089944660663605,
+      "num_tokens": 18972768.0,
+      "step": 1538
+    },
+    {
+      "entropy": 1.0420788824558258,
+      "epoch": 0.8104265402843602,
+      "grad_norm": 0.29141098260879517,
+      "learning_rate": 0.00019655653763342155,
+      "loss": 0.20703990757465363,
+      "mean_token_accuracy": 0.9171910136938095,
+      "num_tokens": 18985104.0,
+      "step": 1539
+    },
+    {
+      "entropy": 1.0945038497447968,
+      "epoch": 0.8109531332280148,
+      "grad_norm": 0.30959945917129517,
+      "learning_rate": 0.00019655052745364509,
+      "loss": 0.24617335200309753,
+      "mean_token_accuracy": 0.9053667187690735,
+      "num_tokens": 18997440.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0674882233142853,
+      "epoch": 0.8114797261716693,
+      "grad_norm": 0.29151442646980286,
+      "learning_rate": 0.00019654451213586824,
+      "loss": 0.202132448554039,
+      "mean_token_accuracy": 0.9202746897935867,
+      "num_tokens": 19009776.0,
+      "step": 1541
+    },
+    {
+      "entropy": 1.1482883095741272,
+      "epoch": 0.8120063191153238,
+      "grad_norm": 0.3008717894554138,
+      "learning_rate": 0.00019653849168044815,
+      "loss": 0.2195252776145935,
+      "mean_token_accuracy": 0.9127016216516495,
+      "num_tokens": 19022112.0,
+      "step": 1542
+    },
+    {
+      "entropy": 1.0573867559432983,
+      "epoch": 0.8125329120589784,
+      "grad_norm": 0.2779742479324341,
+      "learning_rate": 0.00019653246608774215,
+      "loss": 0.20425541698932648,
+      "mean_token_accuracy": 0.9176970571279526,
+      "num_tokens": 19034448.0,
+      "step": 1543
+    },
+    {
+      "entropy": 1.1284627318382263,
+      "epoch": 0.8130595050026329,
+      "grad_norm": 0.3272855281829834,
+      "learning_rate": 0.00019652643535810803,
+      "loss": 0.2418060004711151,
+      "mean_token_accuracy": 0.9060062915086746,
+      "num_tokens": 19046784.0,
+      "step": 1544
+    },
+    {
+      "entropy": 1.160320371389389,
+      "epoch": 0.8135860979462876,
+      "grad_norm": 0.28571900725364685,
+      "learning_rate": 0.00019652039949190372,
+      "loss": 0.19945061206817627,
+      "mean_token_accuracy": 0.9213592410087585,
+      "num_tokens": 19059120.0,
+      "step": 1545
+    },
+    {
+      "entropy": 1.0708126425743103,
+      "epoch": 0.8141126908899421,
+      "grad_norm": 0.28214171528816223,
+      "learning_rate": 0.00019651435848948762,
+      "loss": 0.21407712996006012,
+      "mean_token_accuracy": 0.9169910848140717,
+      "num_tokens": 19071456.0,
+      "step": 1546
+    },
+    {
+      "entropy": 1.071442037820816,
+      "epoch": 0.8146392838335966,
+      "grad_norm": 0.27851834893226624,
+      "learning_rate": 0.00019650831235121824,
+      "loss": 0.20925593376159668,
+      "mean_token_accuracy": 0.9159252196550369,
+      "num_tokens": 19083792.0,
+      "step": 1547
+    },
+    {
+      "entropy": 1.1155974566936493,
+      "epoch": 0.8151658767772512,
+      "grad_norm": 0.3253529369831085,
+      "learning_rate": 0.00019650226107745461,
+      "loss": 0.21849502623081207,
+      "mean_token_accuracy": 0.9110303372144699,
+      "num_tokens": 19096128.0,
+      "step": 1548
+    },
+    {
+      "entropy": 1.0896004736423492,
+      "epoch": 0.8156924697209057,
+      "grad_norm": 0.28845205903053284,
+      "learning_rate": 0.0001964962046685559,
+      "loss": 0.22848929464817047,
+      "mean_token_accuracy": 0.9086662530899048,
+      "num_tokens": 19108464.0,
+      "step": 1549
+    },
+    {
+      "entropy": 1.0635098963975906,
+      "epoch": 0.8162190626645603,
+      "grad_norm": 0.3507218658924103,
+      "learning_rate": 0.00019649014312488164,
+      "loss": 0.22886288166046143,
+      "mean_token_accuracy": 0.9107942432165146,
+      "num_tokens": 19120800.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0771049559116364,
+      "epoch": 0.8167456556082149,
+      "grad_norm": 0.3061995506286621,
+      "learning_rate": 0.0001964840764467917,
+      "loss": 0.22587770223617554,
+      "mean_token_accuracy": 0.907668799161911,
+      "num_tokens": 19133136.0,
+      "step": 1551
+    },
+    {
+      "entropy": 1.0604461431503296,
+      "epoch": 0.8172722485518694,
+      "grad_norm": 0.29044315218925476,
+      "learning_rate": 0.00019647800463464622,
+      "loss": 0.2082800567150116,
+      "mean_token_accuracy": 0.9161125719547272,
+      "num_tokens": 19145472.0,
+      "step": 1552
+    },
+    {
+      "entropy": 1.0522086769342422,
+      "epoch": 0.817798841495524,
+      "grad_norm": 0.301567405462265,
+      "learning_rate": 0.0001964719276888056,
+      "loss": 0.20523683726787567,
+      "mean_token_accuracy": 0.9130432307720184,
+      "num_tokens": 19157808.0,
+      "step": 1553
+    },
+    {
+      "entropy": 1.02540722489357,
+      "epoch": 0.8183254344391785,
+      "grad_norm": 0.29618892073631287,
+      "learning_rate": 0.00019646584560963065,
+      "loss": 0.22347863018512726,
+      "mean_token_accuracy": 0.9131524860858917,
+      "num_tokens": 19170144.0,
+      "step": 1554
+    },
+    {
+      "entropy": 1.0311861038208008,
+      "epoch": 0.818852027382833,
+      "grad_norm": 0.27861911058425903,
+      "learning_rate": 0.00019645975839748244,
+      "loss": 0.182900071144104,
+      "mean_token_accuracy": 0.9252242147922516,
+      "num_tokens": 19182480.0,
+      "step": 1555
+    },
+    {
+      "entropy": 1.0627757012844086,
+      "epoch": 0.8193786203264877,
+      "grad_norm": 0.3110608458518982,
+      "learning_rate": 0.00019645366605272228,
+      "loss": 0.23385684192180634,
+      "mean_token_accuracy": 0.9053423404693604,
+      "num_tokens": 19194816.0,
+      "step": 1556
+    },
+    {
+      "entropy": 1.0807601511478424,
+      "epoch": 0.8199052132701422,
+      "grad_norm": 0.31477105617523193,
+      "learning_rate": 0.00019644756857571186,
+      "loss": 0.22126808762550354,
+      "mean_token_accuracy": 0.9116723388433456,
+      "num_tokens": 19207152.0,
+      "step": 1557
+    },
+    {
+      "entropy": 1.079503744840622,
+      "epoch": 0.8204318062137967,
+      "grad_norm": 0.314552366733551,
+      "learning_rate": 0.00019644146596681312,
+      "loss": 0.22594426572322845,
+      "mean_token_accuracy": 0.9112756550312042,
+      "num_tokens": 19219488.0,
+      "step": 1558
+    },
+    {
+      "entropy": 1.0542688369750977,
+      "epoch": 0.8209583991574513,
+      "grad_norm": 0.30733972787857056,
+      "learning_rate": 0.0001964353582263884,
+      "loss": 0.22227591276168823,
+      "mean_token_accuracy": 0.9121880233287811,
+      "num_tokens": 19231824.0,
+      "step": 1559
+    },
+    {
+      "entropy": 1.0992242991924286,
+      "epoch": 0.8214849921011058,
+      "grad_norm": 0.29605984687805176,
+      "learning_rate": 0.0001964292453548002,
+      "loss": 0.22202306985855103,
+      "mean_token_accuracy": 0.9160446226596832,
+      "num_tokens": 19244160.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.1573190689086914,
+      "epoch": 0.8220115850447604,
+      "grad_norm": 0.3093453347682953,
+      "learning_rate": 0.00019642312735241148,
+      "loss": 0.22188301384449005,
+      "mean_token_accuracy": 0.9117791652679443,
+      "num_tokens": 19256496.0,
+      "step": 1561
+    },
+    {
+      "entropy": 1.0756475329399109,
+      "epoch": 0.822538177988415,
+      "grad_norm": 0.29175224900245667,
+      "learning_rate": 0.00019641700421958543,
+      "loss": 0.20454362034797668,
+      "mean_token_accuracy": 0.9208622127771378,
+      "num_tokens": 19268832.0,
+      "step": 1562
+    },
+    {
+      "entropy": 1.1309452950954437,
+      "epoch": 0.8230647709320695,
+      "grad_norm": 0.3055620789527893,
+      "learning_rate": 0.00019641087595668547,
+      "loss": 0.24093538522720337,
+      "mean_token_accuracy": 0.9080805480480194,
+      "num_tokens": 19281168.0,
+      "step": 1563
+    },
+    {
+      "entropy": 1.0666328370571136,
+      "epoch": 0.8235913638757241,
+      "grad_norm": 0.26273226737976074,
+      "learning_rate": 0.00019640474256407545,
+      "loss": 0.18816737830638885,
+      "mean_token_accuracy": 0.9207744300365448,
+      "num_tokens": 19293504.0,
+      "step": 1564
+    },
+    {
+      "entropy": 1.1276142299175262,
+      "epoch": 0.8241179568193786,
+      "grad_norm": 0.3050276041030884,
+      "learning_rate": 0.0001963986040421195,
+      "loss": 0.21849283576011658,
+      "mean_token_accuracy": 0.9142372459173203,
+      "num_tokens": 19305840.0,
+      "step": 1565
+    },
+    {
+      "entropy": 1.0887978076934814,
+      "epoch": 0.8246445497630331,
+      "grad_norm": 0.2735901176929474,
+      "learning_rate": 0.00019639246039118198,
+      "loss": 0.20344410836696625,
+      "mean_token_accuracy": 0.9216662347316742,
+      "num_tokens": 19318176.0,
+      "step": 1566
+    },
+    {
+      "entropy": 1.0880299508571625,
+      "epoch": 0.8251711427066877,
+      "grad_norm": 0.2992630898952484,
+      "learning_rate": 0.00019638631161162761,
+      "loss": 0.20155635476112366,
+      "mean_token_accuracy": 0.9177606403827667,
+      "num_tokens": 19330512.0,
+      "step": 1567
+    },
+    {
+      "entropy": 1.0990647375583649,
+      "epoch": 0.8256977356503423,
+      "grad_norm": 0.29473257064819336,
+      "learning_rate": 0.00019638015770382142,
+      "loss": 0.20519070327281952,
+      "mean_token_accuracy": 0.9206506013870239,
+      "num_tokens": 19342848.0,
+      "step": 1568
+    },
+    {
+      "entropy": 1.0590737760066986,
+      "epoch": 0.8262243285939969,
+      "grad_norm": 0.28194659948349,
+      "learning_rate": 0.00019637399866812873,
+      "loss": 0.20845578610897064,
+      "mean_token_accuracy": 0.9180682003498077,
+      "num_tokens": 19355184.0,
+      "step": 1569
+    },
+    {
+      "entropy": 1.1114600002765656,
+      "epoch": 0.8267509215376514,
+      "grad_norm": 0.3110455870628357,
+      "learning_rate": 0.00019636783450491517,
+      "loss": 0.2090432196855545,
+      "mean_token_accuracy": 0.9182581901550293,
+      "num_tokens": 19367520.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.159915953874588,
+      "epoch": 0.8272775144813059,
+      "grad_norm": 0.3254052400588989,
+      "learning_rate": 0.00019636166521454668,
+      "loss": 0.21491852402687073,
+      "mean_token_accuracy": 0.9105751514434814,
+      "num_tokens": 19379856.0,
+      "step": 1571
+    },
+    {
+      "entropy": 1.1465847194194794,
+      "epoch": 0.8278041074249605,
+      "grad_norm": 0.28690120577812195,
+      "learning_rate": 0.00019635549079738946,
+      "loss": 0.21204861998558044,
+      "mean_token_accuracy": 0.9141225218772888,
+      "num_tokens": 19392192.0,
+      "step": 1572
+    },
+    {
+      "entropy": 1.1238433420658112,
+      "epoch": 0.8283307003686151,
+      "grad_norm": 0.3004522919654846,
+      "learning_rate": 0.0001963493112538101,
+      "loss": 0.22797521948814392,
+      "mean_token_accuracy": 0.9102497845888138,
+      "num_tokens": 19404528.0,
+      "step": 1573
+    },
+    {
+      "entropy": 1.1517022848129272,
+      "epoch": 0.8288572933122696,
+      "grad_norm": 0.3072613477706909,
+      "learning_rate": 0.00019634312658417538,
+      "loss": 0.2027738243341446,
+      "mean_token_accuracy": 0.9213471561670303,
+      "num_tokens": 19416864.0,
+      "step": 1574
+    },
+    {
+      "entropy": 1.1321575045585632,
+      "epoch": 0.8293838862559242,
+      "grad_norm": 0.33203771710395813,
+      "learning_rate": 0.0001963369367888525,
+      "loss": 0.23440833389759064,
+      "mean_token_accuracy": 0.9052640497684479,
+      "num_tokens": 19429200.0,
+      "step": 1575
+    },
+    {
+      "entropy": 1.1507805287837982,
+      "epoch": 0.8299104791995787,
+      "grad_norm": 0.28893306851387024,
+      "learning_rate": 0.00019633074186820886,
+      "loss": 0.21041327714920044,
+      "mean_token_accuracy": 0.9186779856681824,
+      "num_tokens": 19441536.0,
+      "step": 1576
+    },
+    {
+      "entropy": 1.1217812597751617,
+      "epoch": 0.8304370721432333,
+      "grad_norm": 0.286567747592926,
+      "learning_rate": 0.00019632454182261228,
+      "loss": 0.20225471258163452,
+      "mean_token_accuracy": 0.9235644489526749,
+      "num_tokens": 19453872.0,
+      "step": 1577
+    },
+    {
+      "entropy": 1.1518227756023407,
+      "epoch": 0.8309636650868878,
+      "grad_norm": 0.30317631363868713,
+      "learning_rate": 0.00019631833665243075,
+      "loss": 0.22191138565540314,
+      "mean_token_accuracy": 0.9106118381023407,
+      "num_tokens": 19466208.0,
+      "step": 1578
+    },
+    {
+      "entropy": 1.164818674325943,
+      "epoch": 0.8314902580305424,
+      "grad_norm": 0.3264894187450409,
+      "learning_rate": 0.0001963121263580327,
+      "loss": 0.22096458077430725,
+      "mean_token_accuracy": 0.9108817130327225,
+      "num_tokens": 19478544.0,
+      "step": 1579
+    },
+    {
+      "entropy": 1.0784786641597748,
+      "epoch": 0.832016850974197,
+      "grad_norm": 0.2952364683151245,
+      "learning_rate": 0.0001963059109397868,
+      "loss": 0.2409459352493286,
+      "mean_token_accuracy": 0.9058417230844498,
+      "num_tokens": 19490880.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0973447263240814,
+      "epoch": 0.8325434439178515,
+      "grad_norm": 0.2781769931316376,
+      "learning_rate": 0.00019629969039806195,
+      "loss": 0.2172403782606125,
+      "mean_token_accuracy": 0.9110274314880371,
+      "num_tokens": 19503216.0,
+      "step": 1581
+    },
+    {
+      "entropy": 1.1422053277492523,
+      "epoch": 0.833070036861506,
+      "grad_norm": 0.2971686124801636,
+      "learning_rate": 0.0001962934647332275,
+      "loss": 0.21011781692504883,
+      "mean_token_accuracy": 0.9153971076011658,
+      "num_tokens": 19515552.0,
+      "step": 1582
+    },
+    {
+      "entropy": 1.1796616911888123,
+      "epoch": 0.8335966298051606,
+      "grad_norm": 0.29507434368133545,
+      "learning_rate": 0.000196287233945653,
+      "loss": 0.19983485341072083,
+      "mean_token_accuracy": 0.9158152490854263,
+      "num_tokens": 19527888.0,
+      "step": 1583
+    },
+    {
+      "entropy": 1.1146294176578522,
+      "epoch": 0.8341232227488151,
+      "grad_norm": 0.275584876537323,
+      "learning_rate": 0.00019628099803570837,
+      "loss": 0.2035273015499115,
+      "mean_token_accuracy": 0.9163656979799271,
+      "num_tokens": 19540224.0,
+      "step": 1584
+    },
+    {
+      "entropy": 1.1470642983913422,
+      "epoch": 0.8346498156924698,
+      "grad_norm": 0.2765231430530548,
+      "learning_rate": 0.00019627475700376374,
+      "loss": 0.22747623920440674,
+      "mean_token_accuracy": 0.9137639552354813,
+      "num_tokens": 19552560.0,
+      "step": 1585
+    },
+    {
+      "entropy": 1.1012212336063385,
+      "epoch": 0.8351764086361243,
+      "grad_norm": 0.28456148505210876,
+      "learning_rate": 0.00019626851085018964,
+      "loss": 0.2105616182088852,
+      "mean_token_accuracy": 0.9195501506328583,
+      "num_tokens": 19564896.0,
+      "step": 1586
+    },
+    {
+      "entropy": 1.106357455253601,
+      "epoch": 0.8357030015797788,
+      "grad_norm": 0.32695838809013367,
+      "learning_rate": 0.00019626225957535684,
+      "loss": 0.22337135672569275,
+      "mean_token_accuracy": 0.9134061336517334,
+      "num_tokens": 19577232.0,
+      "step": 1587
+    },
+    {
+      "entropy": 1.169274091720581,
+      "epoch": 0.8362295945234334,
+      "grad_norm": 0.336927205324173,
+      "learning_rate": 0.0001962560031796365,
+      "loss": 0.2262648195028305,
+      "mean_token_accuracy": 0.9084394127130508,
+      "num_tokens": 19589568.0,
+      "step": 1588
+    },
+    {
+      "entropy": 1.1865193247795105,
+      "epoch": 0.8367561874670879,
+      "grad_norm": 0.3452165722846985,
+      "learning_rate": 0.00019624974166339998,
+      "loss": 0.24824325740337372,
+      "mean_token_accuracy": 0.9022375047206879,
+      "num_tokens": 19601904.0,
+      "step": 1589
+    },
+    {
+      "entropy": 1.1062462031841278,
+      "epoch": 0.8372827804107424,
+      "grad_norm": 0.3023000955581665,
+      "learning_rate": 0.000196243475027019,
+      "loss": 0.23354491591453552,
+      "mean_token_accuracy": 0.9097914397716522,
+      "num_tokens": 19614240.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.1293524503707886,
+      "epoch": 0.8378093733543971,
+      "grad_norm": 0.2942156195640564,
+      "learning_rate": 0.00019623720327086557,
+      "loss": 0.22122958302497864,
+      "mean_token_accuracy": 0.9109482020139694,
+      "num_tokens": 19626576.0,
+      "step": 1591
+    },
+    {
+      "entropy": 1.124775916337967,
+      "epoch": 0.8383359662980516,
+      "grad_norm": 0.315075546503067,
+      "learning_rate": 0.00019623092639531198,
+      "loss": 0.22557595372200012,
+      "mean_token_accuracy": 0.9101526886224747,
+      "num_tokens": 19638912.0,
+      "step": 1592
+    },
+    {
+      "entropy": 1.148577243089676,
+      "epoch": 0.8388625592417062,
+      "grad_norm": 0.43819931149482727,
+      "learning_rate": 0.0001962246444007309,
+      "loss": 0.22852541506290436,
+      "mean_token_accuracy": 0.9057890176773071,
+      "num_tokens": 19651248.0,
+      "step": 1593
+    },
+    {
+      "entropy": 1.14993816614151,
+      "epoch": 0.8393891521853607,
+      "grad_norm": 0.31310200691223145,
+      "learning_rate": 0.00019621835728749525,
+      "loss": 0.2385086715221405,
+      "mean_token_accuracy": 0.9071990847587585,
+      "num_tokens": 19663584.0,
+      "step": 1594
+    },
+    {
+      "entropy": 1.1644853949546814,
+      "epoch": 0.8399157451290152,
+      "grad_norm": 0.2970283031463623,
+      "learning_rate": 0.00019621206505597823,
+      "loss": 0.2200639396905899,
+      "mean_token_accuracy": 0.9170584827661514,
+      "num_tokens": 19675920.0,
+      "step": 1595
+    },
+    {
+      "entropy": 1.1328399777412415,
+      "epoch": 0.8404423380726699,
+      "grad_norm": 0.3323354125022888,
+      "learning_rate": 0.0001962057677065534,
+      "loss": 0.22121144831180573,
+      "mean_token_accuracy": 0.9114271998405457,
+      "num_tokens": 19688256.0,
+      "step": 1596
+    },
+    {
+      "entropy": 1.0829459726810455,
+      "epoch": 0.8409689310163244,
+      "grad_norm": 0.28420135378837585,
+      "learning_rate": 0.0001961994652395946,
+      "loss": 0.2405356764793396,
+      "mean_token_accuracy": 0.9100402593612671,
+      "num_tokens": 19700592.0,
+      "step": 1597
+    },
+    {
+      "entropy": 1.0924118757247925,
+      "epoch": 0.841495523959979,
+      "grad_norm": 0.25400787591934204,
+      "learning_rate": 0.00019619315765547594,
+      "loss": 0.18612799048423767,
+      "mean_token_accuracy": 0.9215864688158035,
+      "num_tokens": 19712928.0,
+      "step": 1598
+    },
+    {
+      "entropy": 1.1357889473438263,
+      "epoch": 0.8420221169036335,
+      "grad_norm": 0.2727976143360138,
+      "learning_rate": 0.00019618684495457188,
+      "loss": 0.22136437892913818,
+      "mean_token_accuracy": 0.9126606434583664,
+      "num_tokens": 19725264.0,
+      "step": 1599
+    },
+    {
+      "entropy": 1.0817617177963257,
+      "epoch": 0.842548709847288,
+      "grad_norm": 0.2677607238292694,
+      "learning_rate": 0.0001961805271372572,
+      "loss": 0.20533126592636108,
+      "mean_token_accuracy": 0.9148357510566711,
+      "num_tokens": 19737600.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.1030884683132172,
+      "epoch": 0.8430753027909426,
+      "grad_norm": 0.2927640378475189,
+      "learning_rate": 0.0001961742042039069,
+      "loss": 0.22252412140369415,
+      "mean_token_accuracy": 0.9111676663160324,
+      "num_tokens": 19749936.0,
+      "step": 1601
+    },
+    {
+      "entropy": 1.1529828608036041,
+      "epoch": 0.8436018957345972,
+      "grad_norm": 0.30727458000183105,
+      "learning_rate": 0.00019616787615489635,
+      "loss": 0.2029101848602295,
+      "mean_token_accuracy": 0.9227928221225739,
+      "num_tokens": 19762272.0,
+      "step": 1602
+    },
+    {
+      "entropy": 1.0316384136676788,
+      "epoch": 0.8441284886782517,
+      "grad_norm": 0.27791184186935425,
+      "learning_rate": 0.00019616154299060122,
+      "loss": 0.21758908033370972,
+      "mean_token_accuracy": 0.9157789200544357,
+      "num_tokens": 19774608.0,
+      "step": 1603
+    },
+    {
+      "entropy": 1.0185693204402924,
+      "epoch": 0.8446550816219063,
+      "grad_norm": 0.2891453504562378,
+      "learning_rate": 0.00019615520471139749,
+      "loss": 0.19542157649993896,
+      "mean_token_accuracy": 0.9184477776288986,
+      "num_tokens": 19786944.0,
+      "step": 1604
+    },
+    {
+      "entropy": 1.010437250137329,
+      "epoch": 0.8451816745655608,
+      "grad_norm": 0.29489409923553467,
+      "learning_rate": 0.00019614886131766137,
+      "loss": 0.2418351173400879,
+      "mean_token_accuracy": 0.9019056260585785,
+      "num_tokens": 19799280.0,
+      "step": 1605
+    },
+    {
+      "entropy": 1.0607281923294067,
+      "epoch": 0.8457082675092154,
+      "grad_norm": 0.2774660587310791,
+      "learning_rate": 0.00019614251280976948,
+      "loss": 0.1955130398273468,
+      "mean_token_accuracy": 0.9208610653877258,
+      "num_tokens": 19811616.0,
+      "step": 1606
+    },
+    {
+      "entropy": 1.0406393706798553,
+      "epoch": 0.8462348604528699,
+      "grad_norm": 0.3048405647277832,
+      "learning_rate": 0.00019613615918809868,
+      "loss": 0.24297615885734558,
+      "mean_token_accuracy": 0.9006519466638565,
+      "num_tokens": 19823952.0,
+      "step": 1607
+    },
+    {
+      "entropy": 0.9977871924638748,
+      "epoch": 0.8467614533965245,
+      "grad_norm": 0.29823529720306396,
+      "learning_rate": 0.00019612980045302617,
+      "loss": 0.22520115971565247,
+      "mean_token_accuracy": 0.9101865142583847,
+      "num_tokens": 19836288.0,
+      "step": 1608
+    },
+    {
+      "entropy": 0.9967030137777328,
+      "epoch": 0.8472880463401791,
+      "grad_norm": 0.30385443568229675,
+      "learning_rate": 0.0001961234366049294,
+      "loss": 0.2113213837146759,
+      "mean_token_accuracy": 0.9132890105247498,
+      "num_tokens": 19848624.0,
+      "step": 1609
+    },
+    {
+      "entropy": 0.9831344783306122,
+      "epoch": 0.8478146392838336,
+      "grad_norm": 0.2813466787338257,
+      "learning_rate": 0.00019611706764418617,
+      "loss": 0.20107880234718323,
+      "mean_token_accuracy": 0.9161145985126495,
+      "num_tokens": 19860960.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0240805000066757,
+      "epoch": 0.8483412322274881,
+      "grad_norm": 0.3142811954021454,
+      "learning_rate": 0.00019611069357117455,
+      "loss": 0.20909224450588226,
+      "mean_token_accuracy": 0.9144964218139648,
+      "num_tokens": 19873296.0,
+      "step": 1611
+    },
+    {
+      "entropy": 0.9786220341920853,
+      "epoch": 0.8488678251711427,
+      "grad_norm": 0.2860144376754761,
+      "learning_rate": 0.00019610431438627296,
+      "loss": 0.20540790259838104,
+      "mean_token_accuracy": 0.9215816557407379,
+      "num_tokens": 19885632.0,
+      "step": 1612
+    },
+    {
+      "entropy": 1.0295808017253876,
+      "epoch": 0.8493944181147972,
+      "grad_norm": 0.2819706201553345,
+      "learning_rate": 0.00019609793008986003,
+      "loss": 0.2058688849210739,
+      "mean_token_accuracy": 0.9158887714147568,
+      "num_tokens": 19897968.0,
+      "step": 1613
+    },
+    {
+      "entropy": 1.0835894644260406,
+      "epoch": 0.8499210110584519,
+      "grad_norm": 0.2932409346103668,
+      "learning_rate": 0.00019609154068231486,
+      "loss": 0.2193165421485901,
+      "mean_token_accuracy": 0.9130483567714691,
+      "num_tokens": 19910304.0,
+      "step": 1614
+    },
+    {
+      "entropy": 1.0591089725494385,
+      "epoch": 0.8504476040021064,
+      "grad_norm": 0.3336801528930664,
+      "learning_rate": 0.00019608514616401668,
+      "loss": 0.2534202039241791,
+      "mean_token_accuracy": 0.9020056575536728,
+      "num_tokens": 19922640.0,
+      "step": 1615
+    },
+    {
+      "entropy": 1.0837358832359314,
+      "epoch": 0.8509741969457609,
+      "grad_norm": 0.2906700372695923,
+      "learning_rate": 0.00019607874653534513,
+      "loss": 0.21131698787212372,
+      "mean_token_accuracy": 0.9210411906242371,
+      "num_tokens": 19934976.0,
+      "step": 1616
+    },
+    {
+      "entropy": 1.0396023094654083,
+      "epoch": 0.8515007898894155,
+      "grad_norm": 0.29053011536598206,
+      "learning_rate": 0.0001960723417966801,
+      "loss": 0.21710500121116638,
+      "mean_token_accuracy": 0.914475828409195,
+      "num_tokens": 19947312.0,
+      "step": 1617
+    },
+    {
+      "entropy": 1.0233092904090881,
+      "epoch": 0.85202738283307,
+      "grad_norm": 0.26935815811157227,
+      "learning_rate": 0.00019606593194840177,
+      "loss": 0.22018460929393768,
+      "mean_token_accuracy": 0.9134867042303085,
+      "num_tokens": 19959648.0,
+      "step": 1618
+    },
+    {
+      "entropy": 1.0701540112495422,
+      "epoch": 0.8525539757767245,
+      "grad_norm": 0.28057801723480225,
+      "learning_rate": 0.00019605951699089075,
+      "loss": 0.21612004935741425,
+      "mean_token_accuracy": 0.9129879325628281,
+      "num_tokens": 19971984.0,
+      "step": 1619
+    },
+    {
+      "entropy": 1.012107402086258,
+      "epoch": 0.8530805687203792,
+      "grad_norm": 0.2693917155265808,
+      "learning_rate": 0.00019605309692452774,
+      "loss": 0.2114534080028534,
+      "mean_token_accuracy": 0.9149720519781113,
+      "num_tokens": 19984320.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0413274019956589,
+      "epoch": 0.8536071616640337,
+      "grad_norm": 0.27927207946777344,
+      "learning_rate": 0.00019604667174969394,
+      "loss": 0.21135683357715607,
+      "mean_token_accuracy": 0.9143811762332916,
+      "num_tokens": 19996656.0,
+      "step": 1621
+    },
+    {
+      "entropy": 0.9798619896173477,
+      "epoch": 0.8541337546076883,
+      "grad_norm": 0.2772925794124603,
+      "learning_rate": 0.00019604024146677079,
+      "loss": 0.19996821880340576,
+      "mean_token_accuracy": 0.9175093472003937,
+      "num_tokens": 20008992.0,
+      "step": 1622
+    },
+    {
+      "entropy": 0.9713831841945648,
+      "epoch": 0.8546603475513428,
+      "grad_norm": 0.2909282445907593,
+      "learning_rate": 0.00019603380607613994,
+      "loss": 0.21484866738319397,
+      "mean_token_accuracy": 0.9112430512905121,
+      "num_tokens": 20021328.0,
+      "step": 1623
+    },
+    {
+      "entropy": 0.9760904163122177,
+      "epoch": 0.8551869404949973,
+      "grad_norm": 0.2880821228027344,
+      "learning_rate": 0.0001960273655781835,
+      "loss": 0.21467523276805878,
+      "mean_token_accuracy": 0.9168452471494675,
+      "num_tokens": 20033664.0,
+      "step": 1624
+    },
+    {
+      "entropy": 1.0048719644546509,
+      "epoch": 0.855713533438652,
+      "grad_norm": 0.3039538860321045,
+      "learning_rate": 0.00019602091997328376,
+      "loss": 0.21055804193019867,
+      "mean_token_accuracy": 0.9170923382043839,
+      "num_tokens": 20046000.0,
+      "step": 1625
+    },
+    {
+      "entropy": 0.95078045129776,
+      "epoch": 0.8562401263823065,
+      "grad_norm": 0.28427428007125854,
+      "learning_rate": 0.00019601446926182335,
+      "loss": 0.2352895438671112,
+      "mean_token_accuracy": 0.9064240008592606,
+      "num_tokens": 20058336.0,
+      "step": 1626
+    },
+    {
+      "entropy": 0.9472018629312515,
+      "epoch": 0.856766719325961,
+      "grad_norm": 0.2894728183746338,
+      "learning_rate": 0.00019600801344418526,
+      "loss": 0.2225104421377182,
+      "mean_token_accuracy": 0.913215383887291,
+      "num_tokens": 20070672.0,
+      "step": 1627
+    },
+    {
+      "entropy": 1.0047412514686584,
+      "epoch": 0.8572933122696156,
+      "grad_norm": 0.30747485160827637,
+      "learning_rate": 0.00019600155252075268,
+      "loss": 0.2196165919303894,
+      "mean_token_accuracy": 0.9105159193277359,
+      "num_tokens": 20083008.0,
+      "step": 1628
+    },
+    {
+      "entropy": 1.0216589123010635,
+      "epoch": 0.8578199052132701,
+      "grad_norm": 0.2717617452144623,
+      "learning_rate": 0.0001959950864919092,
+      "loss": 0.20522895455360413,
+      "mean_token_accuracy": 0.9180753976106644,
+      "num_tokens": 20095344.0,
+      "step": 1629
+    },
+    {
+      "entropy": 1.0321899056434631,
+      "epoch": 0.8583464981569247,
+      "grad_norm": 0.30837494134902954,
+      "learning_rate": 0.00019598861535803863,
+      "loss": 0.21629376709461212,
+      "mean_token_accuracy": 0.9151955991983414,
+      "num_tokens": 20107680.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0596579015254974,
+      "epoch": 0.8588730911005793,
+      "grad_norm": 0.3152981996536255,
+      "learning_rate": 0.00019598213911952515,
+      "loss": 0.24829894304275513,
+      "mean_token_accuracy": 0.9071104824542999,
+      "num_tokens": 20120016.0,
+      "step": 1631
+    },
+    {
+      "entropy": 1.027356281876564,
+      "epoch": 0.8593996840442338,
+      "grad_norm": 0.2758905589580536,
+      "learning_rate": 0.00019597565777675324,
+      "loss": 0.20423409342765808,
+      "mean_token_accuracy": 0.919173076748848,
+      "num_tokens": 20132352.0,
+      "step": 1632
+    },
+    {
+      "entropy": 1.0339451730251312,
+      "epoch": 0.8599262769878884,
+      "grad_norm": 0.2790839672088623,
+      "learning_rate": 0.0001959691713301076,
+      "loss": 0.20338216423988342,
+      "mean_token_accuracy": 0.916516900062561,
+      "num_tokens": 20144688.0,
+      "step": 1633
+    },
+    {
+      "entropy": 1.0149255990982056,
+      "epoch": 0.8604528699315429,
+      "grad_norm": 0.2958633005619049,
+      "learning_rate": 0.00019596267977997332,
+      "loss": 0.22986331582069397,
+      "mean_token_accuracy": 0.9077225029468536,
+      "num_tokens": 20157024.0,
+      "step": 1634
+    },
+    {
+      "entropy": 0.9824042469263077,
+      "epoch": 0.8609794628751974,
+      "grad_norm": 0.2640792429447174,
+      "learning_rate": 0.0001959561831267358,
+      "loss": 0.20994797348976135,
+      "mean_token_accuracy": 0.9186595976352692,
+      "num_tokens": 20169360.0,
+      "step": 1635
+    },
+    {
+      "entropy": 0.9684455096721649,
+      "epoch": 0.861506055818852,
+      "grad_norm": 0.2797081768512726,
+      "learning_rate": 0.00019594968137078068,
+      "loss": 0.21638056635856628,
+      "mean_token_accuracy": 0.9145815968513489,
+      "num_tokens": 20181696.0,
+      "step": 1636
+    },
+    {
+      "entropy": 0.9729262888431549,
+      "epoch": 0.8620326487625066,
+      "grad_norm": 0.2565400004386902,
+      "learning_rate": 0.0001959431745124939,
+      "loss": 0.20372052490711212,
+      "mean_token_accuracy": 0.9224633723497391,
+      "num_tokens": 20194032.0,
+      "step": 1637
+    },
+    {
+      "entropy": 0.9809018820524216,
+      "epoch": 0.8625592417061612,
+      "grad_norm": 0.3174954354763031,
+      "learning_rate": 0.0001959366625522618,
+      "loss": 0.2331572026014328,
+      "mean_token_accuracy": 0.9117663353681564,
+      "num_tokens": 20206368.0,
+      "step": 1638
+    },
+    {
+      "entropy": 0.9628456979990005,
+      "epoch": 0.8630858346498157,
+      "grad_norm": 0.2894734740257263,
+      "learning_rate": 0.0001959301454904709,
+      "loss": 0.2223999798297882,
+      "mean_token_accuracy": 0.9094413071870804,
+      "num_tokens": 20218704.0,
+      "step": 1639
+    },
+    {
+      "entropy": 1.0126699656248093,
+      "epoch": 0.8636124275934702,
+      "grad_norm": 0.2930248975753784,
+      "learning_rate": 0.0001959236233275081,
+      "loss": 0.21124093234539032,
+      "mean_token_accuracy": 0.9162924438714981,
+      "num_tokens": 20231040.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.9787211865186691,
+      "epoch": 0.8641390205371248,
+      "grad_norm": 0.30070269107818604,
+      "learning_rate": 0.00019591709606376059,
+      "loss": 0.21121284365653992,
+      "mean_token_accuracy": 0.9159329831600189,
+      "num_tokens": 20243376.0,
+      "step": 1641
+    },
+    {
+      "entropy": 1.010118991136551,
+      "epoch": 0.8646656134807793,
+      "grad_norm": 0.28400495648384094,
+      "learning_rate": 0.00019591056369961586,
+      "loss": 0.21099403500556946,
+      "mean_token_accuracy": 0.9190017580986023,
+      "num_tokens": 20255712.0,
+      "step": 1642
+    },
+    {
+      "entropy": 0.9496536999940872,
+      "epoch": 0.865192206424434,
+      "grad_norm": 0.2675454914569855,
+      "learning_rate": 0.00019590402623546167,
+      "loss": 0.2049626260995865,
+      "mean_token_accuracy": 0.9156690835952759,
+      "num_tokens": 20268048.0,
+      "step": 1643
+    },
+    {
+      "entropy": 0.9466981887817383,
+      "epoch": 0.8657187993680885,
+      "grad_norm": 0.30046844482421875,
+      "learning_rate": 0.00019589748367168612,
+      "loss": 0.21190345287322998,
+      "mean_token_accuracy": 0.9161060154438019,
+      "num_tokens": 20280384.0,
+      "step": 1644
+    },
+    {
+      "entropy": 1.0232085585594177,
+      "epoch": 0.866245392311743,
+      "grad_norm": 0.3621319532394409,
+      "learning_rate": 0.00019589093600867763,
+      "loss": 0.2193715125322342,
+      "mean_token_accuracy": 0.9074915945529938,
+      "num_tokens": 20292720.0,
+      "step": 1645
+    },
+    {
+      "entropy": 0.993518516421318,
+      "epoch": 0.8667719852553976,
+      "grad_norm": 0.28956446051597595,
+      "learning_rate": 0.00019588438324682488,
+      "loss": 0.21250395476818085,
+      "mean_token_accuracy": 0.9168424904346466,
+      "num_tokens": 20305056.0,
+      "step": 1646
+    },
+    {
+      "entropy": 0.9817381650209427,
+      "epoch": 0.8672985781990521,
+      "grad_norm": 0.34931275248527527,
+      "learning_rate": 0.00019587782538651687,
+      "loss": 0.2343052327632904,
+      "mean_token_accuracy": 0.9059536755084991,
+      "num_tokens": 20317392.0,
+      "step": 1647
+    },
+    {
+      "entropy": 0.9840113967657089,
+      "epoch": 0.8678251711427067,
+      "grad_norm": 0.304169237613678,
+      "learning_rate": 0.00019587126242814288,
+      "loss": 0.20760950446128845,
+      "mean_token_accuracy": 0.9196920245885849,
+      "num_tokens": 20329728.0,
+      "step": 1648
+    },
+    {
+      "entropy": 0.99486044049263,
+      "epoch": 0.8683517640863613,
+      "grad_norm": 0.3464246094226837,
+      "learning_rate": 0.00019586469437209256,
+      "loss": 0.25210103392601013,
+      "mean_token_accuracy": 0.9036094397306442,
+      "num_tokens": 20342064.0,
+      "step": 1649
+    },
+    {
+      "entropy": 0.9775729477405548,
+      "epoch": 0.8688783570300158,
+      "grad_norm": 0.27859246730804443,
+      "learning_rate": 0.00019585812121875577,
+      "loss": 0.21410948038101196,
+      "mean_token_accuracy": 0.9136857688426971,
+      "num_tokens": 20354400.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.0246230065822601,
+      "epoch": 0.8694049499736703,
+      "grad_norm": 0.28361114859580994,
+      "learning_rate": 0.00019585154296852277,
+      "loss": 0.19681544601917267,
+      "mean_token_accuracy": 0.9178670197725296,
+      "num_tokens": 20366736.0,
+      "step": 1651
+    },
+    {
+      "entropy": 0.9918327033519745,
+      "epoch": 0.8699315429173249,
+      "grad_norm": 0.2778252065181732,
+      "learning_rate": 0.00019584495962178403,
+      "loss": 0.20562246441841125,
+      "mean_token_accuracy": 0.9203788787126541,
+      "num_tokens": 20379072.0,
+      "step": 1652
+    },
+    {
+      "entropy": 1.0197946578264236,
+      "epoch": 0.8704581358609794,
+      "grad_norm": 0.2764199674129486,
+      "learning_rate": 0.0001958383711789304,
+      "loss": 0.2005048245191574,
+      "mean_token_accuracy": 0.9199100285768509,
+      "num_tokens": 20391408.0,
+      "step": 1653
+    },
+    {
+      "entropy": 0.946838304400444,
+      "epoch": 0.8709847288046341,
+      "grad_norm": 0.318099707365036,
+      "learning_rate": 0.00019583177764035295,
+      "loss": 0.18645739555358887,
+      "mean_token_accuracy": 0.9281531125307083,
+      "num_tokens": 20403744.0,
+      "step": 1654
+    },
+    {
+      "entropy": 0.9969734102487564,
+      "epoch": 0.8715113217482886,
+      "grad_norm": 0.330425888299942,
+      "learning_rate": 0.00019582517900644313,
+      "loss": 0.2349887192249298,
+      "mean_token_accuracy": 0.9082667678594589,
+      "num_tokens": 20416080.0,
+      "step": 1655
+    },
+    {
+      "entropy": 0.9850341528654099,
+      "epoch": 0.8720379146919431,
+      "grad_norm": 0.30549484491348267,
+      "learning_rate": 0.00019581857527759265,
+      "loss": 0.22454282641410828,
+      "mean_token_accuracy": 0.9096111208200455,
+      "num_tokens": 20428416.0,
+      "step": 1656
+    },
+    {
+      "entropy": 0.9726726412773132,
+      "epoch": 0.8725645076355977,
+      "grad_norm": 0.26101022958755493,
+      "learning_rate": 0.00019581196645419358,
+      "loss": 0.18027372658252716,
+      "mean_token_accuracy": 0.9280071556568146,
+      "num_tokens": 20440752.0,
+      "step": 1657
+    },
+    {
+      "entropy": 1.045421838760376,
+      "epoch": 0.8730911005792522,
+      "grad_norm": 0.31144458055496216,
+      "learning_rate": 0.0001958053525366382,
+      "loss": 0.22132211923599243,
+      "mean_token_accuracy": 0.9151443690061569,
+      "num_tokens": 20453088.0,
+      "step": 1658
+    },
+    {
+      "entropy": 1.0683381259441376,
+      "epoch": 0.8736176935229067,
+      "grad_norm": 0.34451693296432495,
+      "learning_rate": 0.0001957987335253191,
+      "loss": 0.2497258186340332,
+      "mean_token_accuracy": 0.9017534852027893,
+      "num_tokens": 20465424.0,
+      "step": 1659
+    },
+    {
+      "entropy": 1.0424925982952118,
+      "epoch": 0.8741442864665614,
+      "grad_norm": 0.305476576089859,
+      "learning_rate": 0.00019579210942062932,
+      "loss": 0.21568341553211212,
+      "mean_token_accuracy": 0.913795217871666,
+      "num_tokens": 20477760.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077263206243515,
+      "epoch": 0.8746708794102159,
+      "grad_norm": 0.28648725152015686,
+      "learning_rate": 0.00019578548022296203,
+      "loss": 0.18887600302696228,
+      "mean_token_accuracy": 0.9262412637472153,
+      "num_tokens": 20490096.0,
+      "step": 1661
+    },
+    {
+      "entropy": 1.0266597121953964,
+      "epoch": 0.8751974723538705,
+      "grad_norm": 0.31522682309150696,
+      "learning_rate": 0.00019577884593271076,
+      "loss": 0.21652202308177948,
+      "mean_token_accuracy": 0.907884418964386,
+      "num_tokens": 20502432.0,
+      "step": 1662
+    },
+    {
+      "entropy": 1.0532887279987335,
+      "epoch": 0.875724065297525,
+      "grad_norm": 0.25786280632019043,
+      "learning_rate": 0.00019577220655026938,
+      "loss": 0.18014520406723022,
+      "mean_token_accuracy": 0.9304206520318985,
+      "num_tokens": 20514768.0,
+      "step": 1663
+    },
+    {
+      "entropy": 1.0302300453186035,
+      "epoch": 0.8762506582411795,
+      "grad_norm": 0.27384263277053833,
+      "learning_rate": 0.000195765562076032,
+      "loss": 0.19573301076889038,
+      "mean_token_accuracy": 0.9229860901832581,
+      "num_tokens": 20527104.0,
+      "step": 1664
+    },
+    {
+      "entropy": 1.0142173171043396,
+      "epoch": 0.8767772511848341,
+      "grad_norm": 0.2829652428627014,
+      "learning_rate": 0.0001957589125103931,
+      "loss": 0.19770190119743347,
+      "mean_token_accuracy": 0.9212857335805893,
+      "num_tokens": 20539440.0,
+      "step": 1665
+    },
+    {
+      "entropy": 1.030619502067566,
+      "epoch": 0.8773038441284887,
+      "grad_norm": 0.31889650225639343,
+      "learning_rate": 0.0001957522578537474,
+      "loss": 0.21395039558410645,
+      "mean_token_accuracy": 0.9180440455675125,
+      "num_tokens": 20551776.0,
+      "step": 1666
+    },
+    {
+      "entropy": 1.0783992409706116,
+      "epoch": 0.8778304370721433,
+      "grad_norm": 0.31347182393074036,
+      "learning_rate": 0.00019574559810648994,
+      "loss": 0.23077845573425293,
+      "mean_token_accuracy": 0.9107988774776459,
+      "num_tokens": 20564112.0,
+      "step": 1667
+    },
+    {
+      "entropy": 1.0769988298416138,
+      "epoch": 0.8783570300157978,
+      "grad_norm": 0.30257898569107056,
+      "learning_rate": 0.00019573893326901607,
+      "loss": 0.18793995678424835,
+      "mean_token_accuracy": 0.9216104596853256,
+      "num_tokens": 20576448.0,
+      "step": 1668
+    },
+    {
+      "entropy": 1.041060984134674,
+      "epoch": 0.8788836229594523,
+      "grad_norm": 0.29523128271102905,
+      "learning_rate": 0.0001957322633417215,
+      "loss": 0.20431648194789886,
+      "mean_token_accuracy": 0.9192978888750076,
+      "num_tokens": 20588784.0,
+      "step": 1669
+    },
+    {
+      "entropy": 1.0586915016174316,
+      "epoch": 0.8794102159031069,
+      "grad_norm": 0.29364022612571716,
+      "learning_rate": 0.0001957255883250021,
+      "loss": 0.22029076516628265,
+      "mean_token_accuracy": 0.9125624299049377,
+      "num_tokens": 20601120.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0633411705493927,
+      "epoch": 0.8799368088467614,
+      "grad_norm": 0.2905363440513611,
+      "learning_rate": 0.00019571890821925417,
+      "loss": 0.19489197432994843,
+      "mean_token_accuracy": 0.9242742508649826,
+      "num_tokens": 20613456.0,
+      "step": 1671
+    },
+    {
+      "entropy": 1.135309487581253,
+      "epoch": 0.880463401790416,
+      "grad_norm": 0.3044449985027313,
+      "learning_rate": 0.0001957122230248743,
+      "loss": 0.19300708174705505,
+      "mean_token_accuracy": 0.9245816022157669,
+      "num_tokens": 20625792.0,
+      "step": 1672
+    },
+    {
+      "entropy": 1.0445529222488403,
+      "epoch": 0.8809899947340706,
+      "grad_norm": 0.28844696283340454,
+      "learning_rate": 0.00019570553274225928,
+      "loss": 0.20883029699325562,
+      "mean_token_accuracy": 0.9140314012765884,
+      "num_tokens": 20638128.0,
+      "step": 1673
+    },
+    {
+      "entropy": 1.1107747852802277,
+      "epoch": 0.8815165876777251,
+      "grad_norm": 0.32846325635910034,
+      "learning_rate": 0.00019569883737180634,
+      "loss": 0.24370679259300232,
+      "mean_token_accuracy": 0.901743620634079,
+      "num_tokens": 20650464.0,
+      "step": 1674
+    },
+    {
+      "entropy": 1.1209677755832672,
+      "epoch": 0.8820431806213797,
+      "grad_norm": 0.2889225482940674,
+      "learning_rate": 0.0001956921369139129,
+      "loss": 0.21621105074882507,
+      "mean_token_accuracy": 0.9143183678388596,
+      "num_tokens": 20662800.0,
+      "step": 1675
+    },
+    {
+      "entropy": 1.0856188535690308,
+      "epoch": 0.8825697735650342,
+      "grad_norm": 0.27882513403892517,
+      "learning_rate": 0.00019568543136897675,
+      "loss": 0.21793431043624878,
+      "mean_token_accuracy": 0.9142801910638809,
+      "num_tokens": 20675136.0,
+      "step": 1676
+    },
+    {
+      "entropy": 1.0983311533927917,
+      "epoch": 0.8830963665086888,
+      "grad_norm": 0.272402822971344,
+      "learning_rate": 0.00019567872073739595,
+      "loss": 0.20574277639389038,
+      "mean_token_accuracy": 0.9181714653968811,
+      "num_tokens": 20687472.0,
+      "step": 1677
+    },
+    {
+      "entropy": 1.0844581425189972,
+      "epoch": 0.8836229594523434,
+      "grad_norm": 0.27611178159713745,
+      "learning_rate": 0.0001956720050195689,
+      "loss": 0.2087741196155548,
+      "mean_token_accuracy": 0.9139288067817688,
+      "num_tokens": 20699808.0,
+      "step": 1678
+    },
+    {
+      "entropy": 1.12508624792099,
+      "epoch": 0.8841495523959979,
+      "grad_norm": 0.29753515124320984,
+      "learning_rate": 0.0001956652842158942,
+      "loss": 0.20983164012432098,
+      "mean_token_accuracy": 0.9156063199043274,
+      "num_tokens": 20712144.0,
+      "step": 1679
+    },
+    {
+      "entropy": 1.1237779259681702,
+      "epoch": 0.8846761453396524,
+      "grad_norm": 0.2853649854660034,
+      "learning_rate": 0.00019565855832677086,
+      "loss": 0.20436659455299377,
+      "mean_token_accuracy": 0.9203456342220306,
+      "num_tokens": 20724480.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.1044580936431885,
+      "epoch": 0.885202738283307,
+      "grad_norm": 0.2854003608226776,
+      "learning_rate": 0.00019565182735259822,
+      "loss": 0.2195129096508026,
+      "mean_token_accuracy": 0.9125242233276367,
+      "num_tokens": 20736816.0,
+      "step": 1681
+    },
+    {
+      "entropy": 1.084385484457016,
+      "epoch": 0.8857293312269615,
+      "grad_norm": 0.29433006048202515,
+      "learning_rate": 0.0001956450912937758,
+      "loss": 0.2089007943868637,
+      "mean_token_accuracy": 0.9135125875473022,
+      "num_tokens": 20749152.0,
+      "step": 1682
+    },
+    {
+      "entropy": 1.0872920453548431,
+      "epoch": 0.8862559241706162,
+      "grad_norm": 0.2716350555419922,
+      "learning_rate": 0.00019563835015070348,
+      "loss": 0.20034612715244293,
+      "mean_token_accuracy": 0.9140070527791977,
+      "num_tokens": 20761488.0,
+      "step": 1683
+    },
+    {
+      "entropy": 1.0698755085468292,
+      "epoch": 0.8867825171142707,
+      "grad_norm": 0.263379842042923,
+      "learning_rate": 0.00019563160392378144,
+      "loss": 0.2013968825340271,
+      "mean_token_accuracy": 0.920297846198082,
+      "num_tokens": 20773824.0,
+      "step": 1684
+    },
+    {
+      "entropy": 1.0688921809196472,
+      "epoch": 0.8873091100579252,
+      "grad_norm": 0.2941676676273346,
+      "learning_rate": 0.00019562485261341017,
+      "loss": 0.222881019115448,
+      "mean_token_accuracy": 0.9110155403614044,
+      "num_tokens": 20786160.0,
+      "step": 1685
+    },
+    {
+      "entropy": 1.0462908446788788,
+      "epoch": 0.8878357030015798,
+      "grad_norm": 0.2833232581615448,
+      "learning_rate": 0.00019561809621999047,
+      "loss": 0.2037080079317093,
+      "mean_token_accuracy": 0.9212964028120041,
+      "num_tokens": 20798496.0,
+      "step": 1686
+    },
+    {
+      "entropy": 1.0370090901851654,
+      "epoch": 0.8883622959452343,
+      "grad_norm": 0.299941748380661,
+      "learning_rate": 0.0001956113347439234,
+      "loss": 0.21904562413692474,
+      "mean_token_accuracy": 0.9075719863176346,
+      "num_tokens": 20810832.0,
+      "step": 1687
+    },
+    {
+      "entropy": 1.0448128283023834,
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.2878952920436859,
+      "learning_rate": 0.0001956045681856104,
+      "loss": 0.2038077563047409,
+      "mean_token_accuracy": 0.9187713265419006,
+      "num_tokens": 20823168.0,
+      "step": 1688
+    },
+    {
+      "entropy": 1.0190322399139404,
+      "epoch": 0.8894154818325435,
+      "grad_norm": 0.34769734740257263,
+      "learning_rate": 0.0001955977965454531,
+      "loss": 0.213621124625206,
+      "mean_token_accuracy": 0.9181492775678635,
+      "num_tokens": 20835504.0,
+      "step": 1689
+    },
+    {
+      "entropy": 1.0234404504299164,
+      "epoch": 0.889942074776198,
+      "grad_norm": 0.2829277813434601,
+      "learning_rate": 0.00019559101982385356,
+      "loss": 0.20716841518878937,
+      "mean_token_accuracy": 0.9201241731643677,
+      "num_tokens": 20847840.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0015476644039154,
+      "epoch": 0.8904686677198526,
+      "grad_norm": 0.26036950945854187,
+      "learning_rate": 0.000195584238021214,
+      "loss": 0.18766967952251434,
+      "mean_token_accuracy": 0.9250665605068207,
+      "num_tokens": 20860176.0,
+      "step": 1691
+    },
+    {
+      "entropy": 1.032362923026085,
+      "epoch": 0.8909952606635071,
+      "grad_norm": 0.4818808138370514,
+      "learning_rate": 0.00019557745113793704,
+      "loss": 0.19411087036132812,
+      "mean_token_accuracy": 0.921613797545433,
+      "num_tokens": 20872512.0,
+      "step": 1692
+    },
+    {
+      "entropy": 0.9709571003913879,
+      "epoch": 0.8915218536071616,
+      "grad_norm": 0.2804003655910492,
+      "learning_rate": 0.0001955706591744256,
+      "loss": 0.21858839690685272,
+      "mean_token_accuracy": 0.9120522737503052,
+      "num_tokens": 20884848.0,
+      "step": 1693
+    },
+    {
+      "entropy": 1.0739005506038666,
+      "epoch": 0.8920484465508162,
+      "grad_norm": 0.2946876585483551,
+      "learning_rate": 0.00019556386213108288,
+      "loss": 0.19359217584133148,
+      "mean_token_accuracy": 0.925596609711647,
+      "num_tokens": 20897184.0,
+      "step": 1694
+    },
+    {
+      "entropy": 1.0181286334991455,
+      "epoch": 0.8925750394944708,
+      "grad_norm": 0.3197228014469147,
+      "learning_rate": 0.00019555706000831235,
+      "loss": 0.2175355702638626,
+      "mean_token_accuracy": 0.9139997512102127,
+      "num_tokens": 20909520.0,
+      "step": 1695
+    },
+    {
+      "entropy": 1.0345196276903152,
+      "epoch": 0.8931016324381253,
+      "grad_norm": 0.29833653569221497,
+      "learning_rate": 0.00019555025280651786,
+      "loss": 0.2255219966173172,
+      "mean_token_accuracy": 0.9102495014667511,
+      "num_tokens": 20921856.0,
+      "step": 1696
+    },
+    {
+      "entropy": 0.9587903320789337,
+      "epoch": 0.8936282253817799,
+      "grad_norm": 0.2722357511520386,
+      "learning_rate": 0.00019554344052610348,
+      "loss": 0.19479139149188995,
+      "mean_token_accuracy": 0.9188268184661865,
+      "num_tokens": 20934192.0,
+      "step": 1697
+    },
+    {
+      "entropy": 0.9713253229856491,
+      "epoch": 0.8941548183254344,
+      "grad_norm": 0.2900646924972534,
+      "learning_rate": 0.0001955366231674736,
+      "loss": 0.20535482466220856,
+      "mean_token_accuracy": 0.9150903820991516,
+      "num_tokens": 20946528.0,
+      "step": 1698
+    },
+    {
+      "entropy": 1.020512416958809,
+      "epoch": 0.894681411269089,
+      "grad_norm": 0.2782817780971527,
+      "learning_rate": 0.00019552980073103297,
+      "loss": 0.20126226544380188,
+      "mean_token_accuracy": 0.9161560386419296,
+      "num_tokens": 20958864.0,
+      "step": 1699
+    },
+    {
+      "entropy": 0.9827070981264114,
+      "epoch": 0.8952080042127436,
+      "grad_norm": 0.2738768458366394,
+      "learning_rate": 0.00019552297321718658,
+      "loss": 0.19939669966697693,
+      "mean_token_accuracy": 0.9180780202150345,
+      "num_tokens": 20971200.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.9992471039295197,
+      "epoch": 0.8957345971563981,
+      "grad_norm": 0.31194329261779785,
+      "learning_rate": 0.0001955161406263397,
+      "loss": 0.23202218115329742,
+      "mean_token_accuracy": 0.9068114459514618,
+      "num_tokens": 20983536.0,
+      "step": 1701
+    },
+    {
+      "entropy": 0.9976187497377396,
+      "epoch": 0.8962611901000527,
+      "grad_norm": 0.29019710421562195,
+      "learning_rate": 0.00019550930295889803,
+      "loss": 0.21963413059711456,
+      "mean_token_accuracy": 0.9133173823356628,
+      "num_tokens": 20995872.0,
+      "step": 1702
+    },
+    {
+      "entropy": 1.0274995416402817,
+      "epoch": 0.8967877830437072,
+      "grad_norm": 0.2910099923610687,
+      "learning_rate": 0.00019550246021526736,
+      "loss": 0.20295658707618713,
+      "mean_token_accuracy": 0.9164696484804153,
+      "num_tokens": 21008208.0,
+      "step": 1703
+    },
+    {
+      "entropy": 1.0305428206920624,
+      "epoch": 0.8973143759873617,
+      "grad_norm": 0.3525075912475586,
+      "learning_rate": 0.000195495612395854,
+      "loss": 0.20155680179595947,
+      "mean_token_accuracy": 0.9174475073814392,
+      "num_tokens": 21020544.0,
+      "step": 1704
+    },
+    {
+      "entropy": 1.0068664401769638,
+      "epoch": 0.8978409689310163,
+      "grad_norm": 0.33528581261634827,
+      "learning_rate": 0.00019548875950106448,
+      "loss": 0.22776372730731964,
+      "mean_token_accuracy": 0.906059592962265,
+      "num_tokens": 21032880.0,
+      "step": 1705
+    },
+    {
+      "entropy": 1.0565652251243591,
+      "epoch": 0.8983675618746709,
+      "grad_norm": 0.29654526710510254,
+      "learning_rate": 0.00019548190153130553,
+      "loss": 0.23316293954849243,
+      "mean_token_accuracy": 0.9070921689271927,
+      "num_tokens": 21045216.0,
+      "step": 1706
+    },
+    {
+      "entropy": 0.9845771491527557,
+      "epoch": 0.8988941548183255,
+      "grad_norm": 0.3172268271446228,
+      "learning_rate": 0.00019547503848698435,
+      "loss": 0.22646287083625793,
+      "mean_token_accuracy": 0.9135229587554932,
+      "num_tokens": 21057552.0,
+      "step": 1707
+    },
+    {
+      "entropy": 1.014034628868103,
+      "epoch": 0.89942074776198,
+      "grad_norm": 0.27241358160972595,
+      "learning_rate": 0.00019546817036850827,
+      "loss": 0.21421730518341064,
+      "mean_token_accuracy": 0.9127627313137054,
+      "num_tokens": 21069888.0,
+      "step": 1708
+    },
+    {
+      "entropy": 1.044843077659607,
+      "epoch": 0.8999473407056345,
+      "grad_norm": 0.27422401309013367,
+      "learning_rate": 0.00019546129717628512,
+      "loss": 0.22259175777435303,
+      "mean_token_accuracy": 0.90959133207798,
+      "num_tokens": 21082224.0,
+      "step": 1709
+    },
+    {
+      "entropy": 1.0896000117063522,
+      "epoch": 0.9004739336492891,
+      "grad_norm": 0.31072503328323364,
+      "learning_rate": 0.00019545441891072283,
+      "loss": 0.21754461526870728,
+      "mean_token_accuracy": 0.9161411225795746,
+      "num_tokens": 21094560.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0415829718112946,
+      "epoch": 0.9010005265929436,
+      "grad_norm": 0.261515736579895,
+      "learning_rate": 0.00019544753557222975,
+      "loss": 0.1926898956298828,
+      "mean_token_accuracy": 0.9213140159845352,
+      "num_tokens": 21106896.0,
+      "step": 1711
+    },
+    {
+      "entropy": 1.0076445639133453,
+      "epoch": 0.9015271195365983,
+      "grad_norm": 0.26134979724884033,
+      "learning_rate": 0.00019544064716121452,
+      "loss": 0.20023104548454285,
+      "mean_token_accuracy": 0.9194119721651077,
+      "num_tokens": 21119232.0,
+      "step": 1712
+    },
+    {
+      "entropy": 1.007362276315689,
+      "epoch": 0.9020537124802528,
+      "grad_norm": 0.2691597044467926,
+      "learning_rate": 0.00019543375367808604,
+      "loss": 0.19764183461666107,
+      "mean_token_accuracy": 0.9207281470298767,
+      "num_tokens": 21131568.0,
+      "step": 1713
+    },
+    {
+      "entropy": 1.0264256298542023,
+      "epoch": 0.9025803054239073,
+      "grad_norm": 0.2841010093688965,
+      "learning_rate": 0.00019542685512325357,
+      "loss": 0.19019414484500885,
+      "mean_token_accuracy": 0.9279275834560394,
+      "num_tokens": 21143904.0,
+      "step": 1714
+    },
+    {
+      "entropy": 1.0494853258132935,
+      "epoch": 0.9031068983675619,
+      "grad_norm": 0.2881733477115631,
+      "learning_rate": 0.00019541995149712662,
+      "loss": 0.19694221019744873,
+      "mean_token_accuracy": 0.9203800857067108,
+      "num_tokens": 21156240.0,
+      "step": 1715
+    },
+    {
+      "entropy": 1.029950574040413,
+      "epoch": 0.9036334913112164,
+      "grad_norm": 0.2856560945510864,
+      "learning_rate": 0.00019541304280011498,
+      "loss": 0.208457350730896,
+      "mean_token_accuracy": 0.9130570143461227,
+      "num_tokens": 21168576.0,
+      "step": 1716
+    },
+    {
+      "entropy": 1.036653459072113,
+      "epoch": 0.9041600842548709,
+      "grad_norm": 0.35959112644195557,
+      "learning_rate": 0.00019540612903262887,
+      "loss": 0.2388800084590912,
+      "mean_token_accuracy": 0.909120500087738,
+      "num_tokens": 21180912.0,
+      "step": 1717
+    },
+    {
+      "entropy": 1.00953871011734,
+      "epoch": 0.9046866771985256,
+      "grad_norm": 0.3067711293697357,
+      "learning_rate": 0.00019539921019507862,
+      "loss": 0.2145732194185257,
+      "mean_token_accuracy": 0.9115498811006546,
+      "num_tokens": 21193248.0,
+      "step": 1718
+    },
+    {
+      "entropy": 0.9901564866304398,
+      "epoch": 0.9052132701421801,
+      "grad_norm": 0.3036840856075287,
+      "learning_rate": 0.00019539228628787501,
+      "loss": 0.2193184792995453,
+      "mean_token_accuracy": 0.9104121774435043,
+      "num_tokens": 21205584.0,
+      "step": 1719
+    },
+    {
+      "entropy": 0.9771833568811417,
+      "epoch": 0.9057398630858347,
+      "grad_norm": 0.32093900442123413,
+      "learning_rate": 0.00019538535731142907,
+      "loss": 0.23529593646526337,
+      "mean_token_accuracy": 0.9093421250581741,
+      "num_tokens": 21217920.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.011559009552002,
+      "epoch": 0.9062664560294892,
+      "grad_norm": 0.2692308723926544,
+      "learning_rate": 0.00019537842326615215,
+      "loss": 0.20593476295471191,
+      "mean_token_accuracy": 0.9199935644865036,
+      "num_tokens": 21230256.0,
+      "step": 1721
+    },
+    {
+      "entropy": 1.073755070567131,
+      "epoch": 0.9067930489731437,
+      "grad_norm": 0.3065417408943176,
+      "learning_rate": 0.00019537148415245582,
+      "loss": 0.23454323410987854,
+      "mean_token_accuracy": 0.9069335609674454,
+      "num_tokens": 21242592.0,
+      "step": 1722
+    },
+    {
+      "entropy": 1.0648082047700882,
+      "epoch": 0.9073196419167984,
+      "grad_norm": 0.32037290930747986,
+      "learning_rate": 0.0001953645399707521,
+      "loss": 0.2138725221157074,
+      "mean_token_accuracy": 0.9131258726119995,
+      "num_tokens": 21254928.0,
+      "step": 1723
+    },
+    {
+      "entropy": 1.0663966238498688,
+      "epoch": 0.9078462348604529,
+      "grad_norm": 0.2899954915046692,
+      "learning_rate": 0.00019535759072145318,
+      "loss": 0.204481303691864,
+      "mean_token_accuracy": 0.9170624017715454,
+      "num_tokens": 21267264.0,
+      "step": 1724
+    },
+    {
+      "entropy": 1.020171880722046,
+      "epoch": 0.9083728278041074,
+      "grad_norm": 0.29145675897598267,
+      "learning_rate": 0.0001953506364049716,
+      "loss": 0.21512804925441742,
+      "mean_token_accuracy": 0.9173563122749329,
+      "num_tokens": 21279600.0,
+      "step": 1725
+    },
+    {
+      "entropy": 1.0235641449689865,
+      "epoch": 0.908899420747762,
+      "grad_norm": 0.27262571454048157,
+      "learning_rate": 0.00019534367702172016,
+      "loss": 0.22479593753814697,
+      "mean_token_accuracy": 0.9110193848609924,
+      "num_tokens": 21291936.0,
+      "step": 1726
+    },
+    {
+      "entropy": 0.9845803081989288,
+      "epoch": 0.9094260136914165,
+      "grad_norm": 0.2786709666252136,
+      "learning_rate": 0.00019533671257211205,
+      "loss": 0.2018449455499649,
+      "mean_token_accuracy": 0.9166059494018555,
+      "num_tokens": 21304272.0,
+      "step": 1727
+    },
+    {
+      "entropy": 0.9907389134168625,
+      "epoch": 0.909952606635071,
+      "grad_norm": 0.2637747526168823,
+      "learning_rate": 0.00019532974305656075,
+      "loss": 0.211765855550766,
+      "mean_token_accuracy": 0.9151495099067688,
+      "num_tokens": 21316608.0,
+      "step": 1728
+    },
+    {
+      "entropy": 1.0142314583063126,
+      "epoch": 0.9104791995787257,
+      "grad_norm": 0.2817028760910034,
+      "learning_rate": 0.0001953227684754799,
+      "loss": 0.19784370064735413,
+      "mean_token_accuracy": 0.9207914024591446,
+      "num_tokens": 21328944.0,
+      "step": 1729
+    },
+    {
+      "entropy": 1.0088147670030594,
+      "epoch": 0.9110057925223802,
+      "grad_norm": 0.2688204050064087,
+      "learning_rate": 0.00019531578882928357,
+      "loss": 0.20449689030647278,
+      "mean_token_accuracy": 0.9207023978233337,
+      "num_tokens": 21341280.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.9528838694095612,
+      "epoch": 0.9115323854660348,
+      "grad_norm": 0.27442365884780884,
+      "learning_rate": 0.00019530880411838616,
+      "loss": 0.21661211550235748,
+      "mean_token_accuracy": 0.9121255576610565,
+      "num_tokens": 21353616.0,
+      "step": 1731
+    },
+    {
+      "entropy": 0.9663679003715515,
+      "epoch": 0.9120589784096893,
+      "grad_norm": 0.2818276584148407,
+      "learning_rate": 0.00019530181434320224,
+      "loss": 0.21009200811386108,
+      "mean_token_accuracy": 0.9155483096837997,
+      "num_tokens": 21365952.0,
+      "step": 1732
+    },
+    {
+      "entropy": 0.9394736588001251,
+      "epoch": 0.9125855713533438,
+      "grad_norm": 0.3315548598766327,
+      "learning_rate": 0.0001952948195041468,
+      "loss": 0.21075570583343506,
+      "mean_token_accuracy": 0.9181785732507706,
+      "num_tokens": 21378288.0,
+      "step": 1733
+    },
+    {
+      "entropy": 0.9442544430494308,
+      "epoch": 0.9131121642969984,
+      "grad_norm": 0.3284794092178345,
+      "learning_rate": 0.000195287819601635,
+      "loss": 0.23087412118911743,
+      "mean_token_accuracy": 0.9092219769954681,
+      "num_tokens": 21390624.0,
+      "step": 1734
+    },
+    {
+      "entropy": 0.9372933954000473,
+      "epoch": 0.913638757240653,
+      "grad_norm": 0.29973915219306946,
+      "learning_rate": 0.0001952808146360825,
+      "loss": 0.2058921754360199,
+      "mean_token_accuracy": 0.9160513281822205,
+      "num_tokens": 21402960.0,
+      "step": 1735
+    },
+    {
+      "entropy": 0.9260263890028,
+      "epoch": 0.9141653501843076,
+      "grad_norm": 0.2917967438697815,
+      "learning_rate": 0.00019527380460790508,
+      "loss": 0.2268896996974945,
+      "mean_token_accuracy": 0.9107523709535599,
+      "num_tokens": 21415296.0,
+      "step": 1736
+    },
+    {
+      "entropy": 0.9555350244045258,
+      "epoch": 0.9146919431279621,
+      "grad_norm": 0.2836836874485016,
+      "learning_rate": 0.0001952667895175189,
+      "loss": 0.21083997189998627,
+      "mean_token_accuracy": 0.9108854085206985,
+      "num_tokens": 21427632.0,
+      "step": 1737
+    },
+    {
+      "entropy": 0.9710684269666672,
+      "epoch": 0.9152185360716166,
+      "grad_norm": 0.3263842761516571,
+      "learning_rate": 0.00019525976936534035,
+      "loss": 0.19704914093017578,
+      "mean_token_accuracy": 0.9213050603866577,
+      "num_tokens": 21439968.0,
+      "step": 1738
+    },
+    {
+      "entropy": 0.920185998082161,
+      "epoch": 0.9157451290152712,
+      "grad_norm": 0.3038593828678131,
+      "learning_rate": 0.00019525274415178626,
+      "loss": 0.2302832156419754,
+      "mean_token_accuracy": 0.9091469496488571,
+      "num_tokens": 21452304.0,
+      "step": 1739
+    },
+    {
+      "entropy": 0.9555569142103195,
+      "epoch": 0.9162717219589257,
+      "grad_norm": 0.3209346532821655,
+      "learning_rate": 0.00019524571387727365,
+      "loss": 0.2268507480621338,
+      "mean_token_accuracy": 0.9097347557544708,
+      "num_tokens": 21464640.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.9613929092884064,
+      "epoch": 0.9167983149025803,
+      "grad_norm": 0.28639957308769226,
+      "learning_rate": 0.0001952386785422199,
+      "loss": 0.22456036508083344,
+      "mean_token_accuracy": 0.9144959151744843,
+      "num_tokens": 21476976.0,
+      "step": 1741
+    },
+    {
+      "entropy": 0.9676374346017838,
+      "epoch": 0.9173249078462349,
+      "grad_norm": 0.29118600487709045,
+      "learning_rate": 0.00019523163814704253,
+      "loss": 0.21779820322990417,
+      "mean_token_accuracy": 0.9170442819595337,
+      "num_tokens": 21489312.0,
+      "step": 1742
+    },
+    {
+      "entropy": 0.9580910950899124,
+      "epoch": 0.9178515007898894,
+      "grad_norm": 0.28892767429351807,
+      "learning_rate": 0.0001952245926921596,
+      "loss": 0.1965646892786026,
+      "mean_token_accuracy": 0.9236303865909576,
+      "num_tokens": 21501648.0,
+      "step": 1743
+    },
+    {
+      "entropy": 1.0211364477872849,
+      "epoch": 0.918378093733544,
+      "grad_norm": 0.2978745996952057,
+      "learning_rate": 0.00019521754217798935,
+      "loss": 0.2231602966785431,
+      "mean_token_accuracy": 0.9130188077688217,
+      "num_tokens": 21513984.0,
+      "step": 1744
+    },
+    {
+      "entropy": 0.9940465688705444,
+      "epoch": 0.9189046866771985,
+      "grad_norm": 0.2930096685886383,
+      "learning_rate": 0.00019521048660495026,
+      "loss": 0.19159580767154694,
+      "mean_token_accuracy": 0.9195211082696915,
+      "num_tokens": 21526320.0,
+      "step": 1745
+    },
+    {
+      "entropy": 0.9247337281703949,
+      "epoch": 0.919431279620853,
+      "grad_norm": 0.2832118570804596,
+      "learning_rate": 0.00019520342597346125,
+      "loss": 0.21378204226493835,
+      "mean_token_accuracy": 0.9141468554735184,
+      "num_tokens": 21538656.0,
+      "step": 1746
+    },
+    {
+      "entropy": 0.9416433870792389,
+      "epoch": 0.9199578725645077,
+      "grad_norm": 0.3025250732898712,
+      "learning_rate": 0.00019519636028394148,
+      "loss": 0.21570643782615662,
+      "mean_token_accuracy": 0.9150727838277817,
+      "num_tokens": 21550992.0,
+      "step": 1747
+    },
+    {
+      "entropy": 0.9537800550460815,
+      "epoch": 0.9204844655081622,
+      "grad_norm": 0.29714229702949524,
+      "learning_rate": 0.00019518928953681032,
+      "loss": 0.22764158248901367,
+      "mean_token_accuracy": 0.9034122675657272,
+      "num_tokens": 21563328.0,
+      "step": 1748
+    },
+    {
+      "entropy": 0.9719788879156113,
+      "epoch": 0.9210110584518167,
+      "grad_norm": 0.2960829734802246,
+      "learning_rate": 0.00019518221373248756,
+      "loss": 0.2084151953458786,
+      "mean_token_accuracy": 0.9151151925325394,
+      "num_tokens": 21575664.0,
+      "step": 1749
+    },
+    {
+      "entropy": 0.9682374745607376,
+      "epoch": 0.9215376513954713,
+      "grad_norm": 0.2902142405509949,
+      "learning_rate": 0.00019517513287139326,
+      "loss": 0.19477756321430206,
+      "mean_token_accuracy": 0.9197756201028824,
+      "num_tokens": 21588000.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.9519971162080765,
+      "epoch": 0.9220642443391258,
+      "grad_norm": 0.28657373785972595,
+      "learning_rate": 0.00019516804695394777,
+      "loss": 0.23340506851673126,
+      "mean_token_accuracy": 0.9101651012897491,
+      "num_tokens": 21600336.0,
+      "step": 1751
+    },
+    {
+      "entropy": 0.9670501947402954,
+      "epoch": 0.9225908372827805,
+      "grad_norm": 0.2798555791378021,
+      "learning_rate": 0.00019516095598057174,
+      "loss": 0.2065662145614624,
+      "mean_token_accuracy": 0.9186369478702545,
+      "num_tokens": 21612672.0,
+      "step": 1752
+    },
+    {
+      "entropy": 0.9457984417676926,
+      "epoch": 0.923117430226435,
+      "grad_norm": 0.2589358985424042,
+      "learning_rate": 0.00019515385995168608,
+      "loss": 0.1857452392578125,
+      "mean_token_accuracy": 0.9278716593980789,
+      "num_tokens": 21625008.0,
+      "step": 1753
+    },
+    {
+      "entropy": 0.923014685511589,
+      "epoch": 0.9236440231700895,
+      "grad_norm": 0.29148226976394653,
+      "learning_rate": 0.00019514675886771207,
+      "loss": 0.21651901304721832,
+      "mean_token_accuracy": 0.9152826964855194,
+      "num_tokens": 21637344.0,
+      "step": 1754
+    },
+    {
+      "entropy": 0.95480976998806,
+      "epoch": 0.9241706161137441,
+      "grad_norm": 0.28343141078948975,
+      "learning_rate": 0.00019513965272907128,
+      "loss": 0.2175363004207611,
+      "mean_token_accuracy": 0.9101854413747787,
+      "num_tokens": 21649680.0,
+      "step": 1755
+    },
+    {
+      "entropy": 0.9613447189331055,
+      "epoch": 0.9246972090573986,
+      "grad_norm": 0.3009392023086548,
+      "learning_rate": 0.0001951325415361855,
+      "loss": 0.22403854131698608,
+      "mean_token_accuracy": 0.9168571084737778,
+      "num_tokens": 21662016.0,
+      "step": 1756
+    },
+    {
+      "entropy": 0.9929066747426987,
+      "epoch": 0.9252238020010531,
+      "grad_norm": 0.3036598563194275,
+      "learning_rate": 0.00019512542528947696,
+      "loss": 0.22976119816303253,
+      "mean_token_accuracy": 0.9101551920175552,
+      "num_tokens": 21674352.0,
+      "step": 1757
+    },
+    {
+      "entropy": 0.968064159154892,
+      "epoch": 0.9257503949447078,
+      "grad_norm": 0.28782951831817627,
+      "learning_rate": 0.00019511830398936809,
+      "loss": 0.21798846125602722,
+      "mean_token_accuracy": 0.9074330925941467,
+      "num_tokens": 21686688.0,
+      "step": 1758
+    },
+    {
+      "entropy": 0.9647489339113235,
+      "epoch": 0.9262769878883623,
+      "grad_norm": 0.2734207212924957,
+      "learning_rate": 0.00019511117763628156,
+      "loss": 0.209343820810318,
+      "mean_token_accuracy": 0.9190388172864914,
+      "num_tokens": 21699024.0,
+      "step": 1759
+    },
+    {
+      "entropy": 0.9990984350442886,
+      "epoch": 0.9268035808320169,
+      "grad_norm": 0.26921236515045166,
+      "learning_rate": 0.00019510404623064053,
+      "loss": 0.21170470118522644,
+      "mean_token_accuracy": 0.9162604361772537,
+      "num_tokens": 21711360.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.9623449593782425,
+      "epoch": 0.9273301737756714,
+      "grad_norm": 0.2714099884033203,
+      "learning_rate": 0.00019509690977286825,
+      "loss": 0.21800348162651062,
+      "mean_token_accuracy": 0.9147970378398895,
+      "num_tokens": 21723696.0,
+      "step": 1761
+    },
+    {
+      "entropy": 1.0402387380599976,
+      "epoch": 0.9278567667193259,
+      "grad_norm": 0.2736240327358246,
+      "learning_rate": 0.00019508976826338844,
+      "loss": 0.22173887491226196,
+      "mean_token_accuracy": 0.9075192958116531,
+      "num_tokens": 21736032.0,
+      "step": 1762
+    },
+    {
+      "entropy": 1.0271132439374924,
+      "epoch": 0.9283833596629805,
+      "grad_norm": 0.2877287268638611,
+      "learning_rate": 0.00019508262170262502,
+      "loss": 0.21932274103164673,
+      "mean_token_accuracy": 0.915712833404541,
+      "num_tokens": 21748368.0,
+      "step": 1763
+    },
+    {
+      "entropy": 1.034477338194847,
+      "epoch": 0.9289099526066351,
+      "grad_norm": 0.2779938280582428,
+      "learning_rate": 0.0001950754700910023,
+      "loss": 0.207020103931427,
+      "mean_token_accuracy": 0.9147273898124695,
+      "num_tokens": 21760704.0,
+      "step": 1764
+    },
+    {
+      "entropy": 0.9738915860652924,
+      "epoch": 0.9294365455502897,
+      "grad_norm": 0.25146281719207764,
+      "learning_rate": 0.00019506831342894473,
+      "loss": 0.1815354973077774,
+      "mean_token_accuracy": 0.9243048876523972,
+      "num_tokens": 21773040.0,
+      "step": 1765
+    },
+    {
+      "entropy": 0.9640511274337769,
+      "epoch": 0.9299631384939442,
+      "grad_norm": 0.2794640362262726,
+      "learning_rate": 0.0001950611517168772,
+      "loss": 0.222852423787117,
+      "mean_token_accuracy": 0.9066295772790909,
+      "num_tokens": 21785376.0,
+      "step": 1766
+    },
+    {
+      "entropy": 0.9703959077596664,
+      "epoch": 0.9304897314375987,
+      "grad_norm": 0.2606351971626282,
+      "learning_rate": 0.00019505398495522487,
+      "loss": 0.1913529932498932,
+      "mean_token_accuracy": 0.9248643219470978,
+      "num_tokens": 21797712.0,
+      "step": 1767
+    },
+    {
+      "entropy": 0.9996339529752731,
+      "epoch": 0.9310163243812533,
+      "grad_norm": 0.310641884803772,
+      "learning_rate": 0.00019504681314441323,
+      "loss": 0.22644834220409393,
+      "mean_token_accuracy": 0.9118537902832031,
+      "num_tokens": 21810048.0,
+      "step": 1768
+    },
+    {
+      "entropy": 0.9964617937803268,
+      "epoch": 0.9315429173249078,
+      "grad_norm": 0.2875900864601135,
+      "learning_rate": 0.00019503963628486793,
+      "loss": 0.2038813978433609,
+      "mean_token_accuracy": 0.9186841398477554,
+      "num_tokens": 21822384.0,
+      "step": 1769
+    },
+    {
+      "entropy": 0.9495928287506104,
+      "epoch": 0.9320695102685624,
+      "grad_norm": 0.2897108495235443,
+      "learning_rate": 0.00019503245437701508,
+      "loss": 0.20874524116516113,
+      "mean_token_accuracy": 0.9164410084486008,
+      "num_tokens": 21834720.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.0277469605207443,
+      "epoch": 0.932596103212217,
+      "grad_norm": 0.31645241379737854,
+      "learning_rate": 0.00019502526742128104,
+      "loss": 0.2254372090101242,
+      "mean_token_accuracy": 0.9089788496494293,
+      "num_tokens": 21847056.0,
+      "step": 1771
+    },
+    {
+      "entropy": 1.041675180196762,
+      "epoch": 0.9331226961558715,
+      "grad_norm": 0.3042741119861603,
+      "learning_rate": 0.00019501807541809243,
+      "loss": 0.21151864528656006,
+      "mean_token_accuracy": 0.9152505397796631,
+      "num_tokens": 21859392.0,
+      "step": 1772
+    },
+    {
+      "entropy": 1.0295332968235016,
+      "epoch": 0.933649289099526,
+      "grad_norm": 0.32914313673973083,
+      "learning_rate": 0.00019501087836787623,
+      "loss": 0.2500017285346985,
+      "mean_token_accuracy": 0.8985483199357986,
+      "num_tokens": 21871728.0,
+      "step": 1773
+    },
+    {
+      "entropy": 1.0453828126192093,
+      "epoch": 0.9341758820431806,
+      "grad_norm": 0.2983553111553192,
+      "learning_rate": 0.00019500367627105965,
+      "loss": 0.21561893820762634,
+      "mean_token_accuracy": 0.9130748361349106,
+      "num_tokens": 21884064.0,
+      "step": 1774
+    },
+    {
+      "entropy": 1.002024069428444,
+      "epoch": 0.9347024749868352,
+      "grad_norm": 0.28427883982658386,
+      "learning_rate": 0.00019499646912807023,
+      "loss": 0.22381159663200378,
+      "mean_token_accuracy": 0.9057275652885437,
+      "num_tokens": 21896400.0,
+      "step": 1775
+    },
+    {
+      "entropy": 1.0608263909816742,
+      "epoch": 0.9352290679304898,
+      "grad_norm": 0.27167659997940063,
+      "learning_rate": 0.00019498925693933584,
+      "loss": 0.1875414401292801,
+      "mean_token_accuracy": 0.9234858453273773,
+      "num_tokens": 21908736.0,
+      "step": 1776
+    },
+    {
+      "entropy": 1.0062582045793533,
+      "epoch": 0.9357556608741443,
+      "grad_norm": 0.2636836767196655,
+      "learning_rate": 0.00019498203970528462,
+      "loss": 0.21502971649169922,
+      "mean_token_accuracy": 0.919221356511116,
+      "num_tokens": 21921072.0,
+      "step": 1777
+    },
+    {
+      "entropy": 1.006988525390625,
+      "epoch": 0.9362822538177988,
+      "grad_norm": 0.26056134700775146,
+      "learning_rate": 0.00019497481742634503,
+      "loss": 0.201970174908638,
+      "mean_token_accuracy": 0.9196654409170151,
+      "num_tokens": 21933408.0,
+      "step": 1778
+    },
+    {
+      "entropy": 1.0629092454910278,
+      "epoch": 0.9368088467614534,
+      "grad_norm": 0.29083991050720215,
+      "learning_rate": 0.00019496759010294577,
+      "loss": 0.22228607535362244,
+      "mean_token_accuracy": 0.9100936055183411,
+      "num_tokens": 21945744.0,
+      "step": 1779
+    },
+    {
+      "entropy": 1.0019027441740036,
+      "epoch": 0.9373354397051079,
+      "grad_norm": 0.2699200212955475,
+      "learning_rate": 0.00019496035773551592,
+      "loss": 0.22549045085906982,
+      "mean_token_accuracy": 0.9094872325658798,
+      "num_tokens": 21958080.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0057822614908218,
+      "epoch": 0.9378620326487626,
+      "grad_norm": 0.28440120816230774,
+      "learning_rate": 0.00019495312032448485,
+      "loss": 0.22345775365829468,
+      "mean_token_accuracy": 0.9118607193231583,
+      "num_tokens": 21970416.0,
+      "step": 1781
+    },
+    {
+      "entropy": 1.0094308406114578,
+      "epoch": 0.9383886255924171,
+      "grad_norm": 0.2834523320198059,
+      "learning_rate": 0.00019494587787028216,
+      "loss": 0.22185727953910828,
+      "mean_token_accuracy": 0.9113683104515076,
+      "num_tokens": 21982752.0,
+      "step": 1782
+    },
+    {
+      "entropy": 0.9964211732149124,
+      "epoch": 0.9389152185360716,
+      "grad_norm": 0.2954987585544586,
+      "learning_rate": 0.00019493863037333778,
+      "loss": 0.20775803923606873,
+      "mean_token_accuracy": 0.9154636561870575,
+      "num_tokens": 21995088.0,
+      "step": 1783
+    },
+    {
+      "entropy": 1.0346784442663193,
+      "epoch": 0.9394418114797262,
+      "grad_norm": 0.2964785695075989,
+      "learning_rate": 0.000194931377834082,
+      "loss": 0.2272047996520996,
+      "mean_token_accuracy": 0.9083570241928101,
+      "num_tokens": 22007424.0,
+      "step": 1784
+    },
+    {
+      "entropy": 1.0281512439250946,
+      "epoch": 0.9399684044233807,
+      "grad_norm": 0.26846325397491455,
+      "learning_rate": 0.00019492412025294534,
+      "loss": 0.20067930221557617,
+      "mean_token_accuracy": 0.9195989966392517,
+      "num_tokens": 22019760.0,
+      "step": 1785
+    },
+    {
+      "entropy": 0.9764789193868637,
+      "epoch": 0.9404949973670352,
+      "grad_norm": 0.26383689045906067,
+      "learning_rate": 0.0001949168576303586,
+      "loss": 0.2051767259836197,
+      "mean_token_accuracy": 0.92010198533535,
+      "num_tokens": 22032096.0,
+      "step": 1786
+    },
+    {
+      "entropy": 1.078786700963974,
+      "epoch": 0.9410215903106899,
+      "grad_norm": 0.3011668920516968,
+      "learning_rate": 0.000194909589966753,
+      "loss": 0.221777081489563,
+      "mean_token_accuracy": 0.9121461063623428,
+      "num_tokens": 22044432.0,
+      "step": 1787
+    },
+    {
+      "entropy": 0.9719114750623703,
+      "epoch": 0.9415481832543444,
+      "grad_norm": 0.27153825759887695,
+      "learning_rate": 0.0001949023172625599,
+      "loss": 0.21883930265903473,
+      "mean_token_accuracy": 0.913441613316536,
+      "num_tokens": 22056768.0,
+      "step": 1788
+    },
+    {
+      "entropy": 0.9923989027738571,
+      "epoch": 0.942074776197999,
+      "grad_norm": 0.3212338984012604,
+      "learning_rate": 0.00019489503951821107,
+      "loss": 0.22175338864326477,
+      "mean_token_accuracy": 0.9168316125869751,
+      "num_tokens": 22069104.0,
+      "step": 1789
+    },
+    {
+      "entropy": 0.9783156663179398,
+      "epoch": 0.9426013691416535,
+      "grad_norm": 0.273110568523407,
+      "learning_rate": 0.00019488775673413857,
+      "loss": 0.21147552132606506,
+      "mean_token_accuracy": 0.9130014330148697,
+      "num_tokens": 22081440.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.9893038272857666,
+      "epoch": 0.943127962085308,
+      "grad_norm": 0.2790793478488922,
+      "learning_rate": 0.00019488046891077474,
+      "loss": 0.22615540027618408,
+      "mean_token_accuracy": 0.9109261482954025,
+      "num_tokens": 22093776.0,
+      "step": 1791
+    },
+    {
+      "entropy": 1.0380553901195526,
+      "epoch": 0.9436545550289626,
+      "grad_norm": 0.27198460698127747,
+      "learning_rate": 0.00019487317604855212,
+      "loss": 0.205999955534935,
+      "mean_token_accuracy": 0.9175222963094711,
+      "num_tokens": 22106112.0,
+      "step": 1792
+    },
+    {
+      "entropy": 1.0116278380155563,
+      "epoch": 0.9441811479726172,
+      "grad_norm": 0.2635519504547119,
+      "learning_rate": 0.00019486587814790377,
+      "loss": 0.21772751212120056,
+      "mean_token_accuracy": 0.9136025160551071,
+      "num_tokens": 22118448.0,
+      "step": 1793
+    },
+    {
+      "entropy": 1.0237055569887161,
+      "epoch": 0.9447077409162717,
+      "grad_norm": 0.2741040289402008,
+      "learning_rate": 0.00019485857520926288,
+      "loss": 0.20760180056095123,
+      "mean_token_accuracy": 0.9132292419672012,
+      "num_tokens": 22130784.0,
+      "step": 1794
+    },
+    {
+      "entropy": 1.0196455419063568,
+      "epoch": 0.9452343338599263,
+      "grad_norm": 0.27313318848609924,
+      "learning_rate": 0.00019485126723306292,
+      "loss": 0.18969658017158508,
+      "mean_token_accuracy": 0.9258754402399063,
+      "num_tokens": 22143120.0,
+      "step": 1795
+    },
+    {
+      "entropy": 1.0403377711772919,
+      "epoch": 0.9457609268035808,
+      "grad_norm": 0.2604946494102478,
+      "learning_rate": 0.00019484395421973783,
+      "loss": 0.2132149338722229,
+      "mean_token_accuracy": 0.9197222590446472,
+      "num_tokens": 22155456.0,
+      "step": 1796
+    },
+    {
+      "entropy": 0.9670400768518448,
+      "epoch": 0.9462875197472354,
+      "grad_norm": 0.24395477771759033,
+      "learning_rate": 0.00019483663616972165,
+      "loss": 0.19094271957874298,
+      "mean_token_accuracy": 0.9200141131877899,
+      "num_tokens": 22167792.0,
+      "step": 1797
+    },
+    {
+      "entropy": 1.0197023451328278,
+      "epoch": 0.94681411269089,
+      "grad_norm": 0.2558318078517914,
+      "learning_rate": 0.00019482931308344888,
+      "loss": 0.19818465411663055,
+      "mean_token_accuracy": 0.9231192022562027,
+      "num_tokens": 22180128.0,
+      "step": 1798
+    },
+    {
+      "entropy": 1.0280962586402893,
+      "epoch": 0.9473407056345445,
+      "grad_norm": 0.28250062465667725,
+      "learning_rate": 0.00019482198496135417,
+      "loss": 0.22149549424648285,
+      "mean_token_accuracy": 0.9105774611234665,
+      "num_tokens": 22192464.0,
+      "step": 1799
+    },
+    {
+      "entropy": 1.004937931895256,
+      "epoch": 0.9478672985781991,
+      "grad_norm": 0.2890641391277313,
+      "learning_rate": 0.00019481465180387264,
+      "loss": 0.2117823362350464,
+      "mean_token_accuracy": 0.9159420728683472,
+      "num_tokens": 22204800.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.9912999272346497,
+      "epoch": 0.9483938915218536,
+      "grad_norm": 0.2721498906612396,
+      "learning_rate": 0.00019480731361143955,
+      "loss": 0.19752754271030426,
+      "mean_token_accuracy": 0.9188252538442612,
+      "num_tokens": 22217136.0,
+      "step": 1801
+    },
+    {
+      "entropy": 0.9911512136459351,
+      "epoch": 0.9489204844655081,
+      "grad_norm": 0.2967390716075897,
+      "learning_rate": 0.00019479997038449054,
+      "loss": 0.2183246612548828,
+      "mean_token_accuracy": 0.908792570233345,
+      "num_tokens": 22229472.0,
+      "step": 1802
+    },
+    {
+      "entropy": 1.0003108382225037,
+      "epoch": 0.9494470774091627,
+      "grad_norm": 0.281955748796463,
+      "learning_rate": 0.00019479262212346153,
+      "loss": 0.20015354454517365,
+      "mean_token_accuracy": 0.9192506521940231,
+      "num_tokens": 22241808.0,
+      "step": 1803
+    },
+    {
+      "entropy": 1.005830705165863,
+      "epoch": 0.9499736703528173,
+      "grad_norm": 0.29906165599823,
+      "learning_rate": 0.00019478526882878876,
+      "loss": 0.2205047309398651,
+      "mean_token_accuracy": 0.9138208776712418,
+      "num_tokens": 22254144.0,
+      "step": 1804
+    },
+    {
+      "entropy": 0.9975222796201706,
+      "epoch": 0.9505002632964719,
+      "grad_norm": 0.297961950302124,
+      "learning_rate": 0.00019477791050090875,
+      "loss": 0.21634206175804138,
+      "mean_token_accuracy": 0.9120276570320129,
+      "num_tokens": 22266480.0,
+      "step": 1805
+    },
+    {
+      "entropy": 0.9912213832139969,
+      "epoch": 0.9510268562401264,
+      "grad_norm": 0.3019638955593109,
+      "learning_rate": 0.00019477054714025832,
+      "loss": 0.21199895441532135,
+      "mean_token_accuracy": 0.9165389537811279,
+      "num_tokens": 22278816.0,
+      "step": 1806
+    },
+    {
+      "entropy": 0.9327168315649033,
+      "epoch": 0.9515534491837809,
+      "grad_norm": 0.288906991481781,
+      "learning_rate": 0.00019476317874727456,
+      "loss": 0.23043318092823029,
+      "mean_token_accuracy": 0.9077083468437195,
+      "num_tokens": 22291152.0,
+      "step": 1807
+    },
+    {
+      "entropy": 0.9759026318788528,
+      "epoch": 0.9520800421274355,
+      "grad_norm": 0.2820807993412018,
+      "learning_rate": 0.00019475580532239497,
+      "loss": 0.21610046923160553,
+      "mean_token_accuracy": 0.914526030421257,
+      "num_tokens": 22303488.0,
+      "step": 1808
+    },
+    {
+      "entropy": 1.01944038271904,
+      "epoch": 0.95260663507109,
+      "grad_norm": 0.29718121886253357,
+      "learning_rate": 0.00019474842686605716,
+      "loss": 0.2213006466627121,
+      "mean_token_accuracy": 0.9078305512666702,
+      "num_tokens": 22315824.0,
+      "step": 1809
+    },
+    {
+      "entropy": 0.9788577258586884,
+      "epoch": 0.9531332280147446,
+      "grad_norm": 0.2710515558719635,
+      "learning_rate": 0.00019474104337869924,
+      "loss": 0.19855360686779022,
+      "mean_token_accuracy": 0.9186171144247055,
+      "num_tokens": 22328160.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.9749794155359268,
+      "epoch": 0.9536598209583992,
+      "grad_norm": 0.2699875235557556,
+      "learning_rate": 0.00019473365486075944,
+      "loss": 0.20002411305904388,
+      "mean_token_accuracy": 0.917415052652359,
+      "num_tokens": 22340496.0,
+      "step": 1811
+    },
+    {
+      "entropy": 1.0002324134111404,
+      "epoch": 0.9541864139020537,
+      "grad_norm": 0.2768368422985077,
+      "learning_rate": 0.00019472626131267646,
+      "loss": 0.2179412692785263,
+      "mean_token_accuracy": 0.9109683334827423,
+      "num_tokens": 22352832.0,
+      "step": 1812
+    },
+    {
+      "entropy": 1.0184978395700455,
+      "epoch": 0.9547130068457083,
+      "grad_norm": 0.2990346848964691,
+      "learning_rate": 0.00019471886273488915,
+      "loss": 0.22559918463230133,
+      "mean_token_accuracy": 0.911268338561058,
+      "num_tokens": 22365168.0,
+      "step": 1813
+    },
+    {
+      "entropy": 0.9672580063343048,
+      "epoch": 0.9552395997893628,
+      "grad_norm": 0.258120059967041,
+      "learning_rate": 0.00019471145912783671,
+      "loss": 0.19848020374774933,
+      "mean_token_accuracy": 0.9207747876644135,
+      "num_tokens": 22377504.0,
+      "step": 1814
+    },
+    {
+      "entropy": 1.0332013070583344,
+      "epoch": 0.9557661927330173,
+      "grad_norm": 0.27658164501190186,
+      "learning_rate": 0.00019470405049195876,
+      "loss": 0.22540630400180817,
+      "mean_token_accuracy": 0.909706637263298,
+      "num_tokens": 22389840.0,
+      "step": 1815
+    },
+    {
+      "entropy": 1.0286743193864822,
+      "epoch": 0.956292785676672,
+      "grad_norm": 0.30504515767097473,
+      "learning_rate": 0.00019469663682769491,
+      "loss": 0.22958287596702576,
+      "mean_token_accuracy": 0.90970279276371,
+      "num_tokens": 22402176.0,
+      "step": 1816
+    },
+    {
+      "entropy": 1.0396344661712646,
+      "epoch": 0.9568193786203265,
+      "grad_norm": 0.3122229278087616,
+      "learning_rate": 0.00019468921813548546,
+      "loss": 0.21201902627944946,
+      "mean_token_accuracy": 0.912709191441536,
+      "num_tokens": 22414512.0,
+      "step": 1817
+    },
+    {
+      "entropy": 0.9904589354991913,
+      "epoch": 0.957345971563981,
+      "grad_norm": 0.2821761667728424,
+      "learning_rate": 0.00019468179441577073,
+      "loss": 0.21789240837097168,
+      "mean_token_accuracy": 0.9137752056121826,
+      "num_tokens": 22426848.0,
+      "step": 1818
+    },
+    {
+      "entropy": 0.9968847781419754,
+      "epoch": 0.9578725645076356,
+      "grad_norm": 0.2896675765514374,
+      "learning_rate": 0.0001946743656689914,
+      "loss": 0.19366274774074554,
+      "mean_token_accuracy": 0.9199696481227875,
+      "num_tokens": 22439184.0,
+      "step": 1819
+    },
+    {
+      "entropy": 0.9972859472036362,
+      "epoch": 0.9583991574512901,
+      "grad_norm": 0.2833546996116638,
+      "learning_rate": 0.00019466693189558852,
+      "loss": 0.20901280641555786,
+      "mean_token_accuracy": 0.9163765460252762,
+      "num_tokens": 22451520.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.0221098363399506,
+      "epoch": 0.9589257503949447,
+      "grad_norm": 0.30042803287506104,
+      "learning_rate": 0.00019465949309600334,
+      "loss": 0.20722414553165436,
+      "mean_token_accuracy": 0.9166164398193359,
+      "num_tokens": 22463856.0,
+      "step": 1821
+    },
+    {
+      "entropy": 0.9847268313169479,
+      "epoch": 0.9594523433385993,
+      "grad_norm": 0.29877185821533203,
+      "learning_rate": 0.00019465204927067754,
+      "loss": 0.22702287137508392,
+      "mean_token_accuracy": 0.910624086856842,
+      "num_tokens": 22476192.0,
+      "step": 1822
+    },
+    {
+      "entropy": 1.0212393552064896,
+      "epoch": 0.9599789362822538,
+      "grad_norm": 0.3099231421947479,
+      "learning_rate": 0.00019464460042005293,
+      "loss": 0.23366589844226837,
+      "mean_token_accuracy": 0.9104152172803879,
+      "num_tokens": 22488528.0,
+      "step": 1823
+    },
+    {
+      "entropy": 1.0240531265735626,
+      "epoch": 0.9605055292259084,
+      "grad_norm": 0.2815970182418823,
+      "learning_rate": 0.00019463714654457172,
+      "loss": 0.19396552443504333,
+      "mean_token_accuracy": 0.9191655963659286,
+      "num_tokens": 22500864.0,
+      "step": 1824
+    },
+    {
+      "entropy": 0.9908776879310608,
+      "epoch": 0.9610321221695629,
+      "grad_norm": 0.25329092144966125,
+      "learning_rate": 0.00019462968764467648,
+      "loss": 0.19574449956417084,
+      "mean_token_accuracy": 0.9221374094486237,
+      "num_tokens": 22513200.0,
+      "step": 1825
+    },
+    {
+      "entropy": 1.0275763273239136,
+      "epoch": 0.9615587151132174,
+      "grad_norm": 0.29698899388313293,
+      "learning_rate": 0.00019462222372080992,
+      "loss": 0.23368221521377563,
+      "mean_token_accuracy": 0.9036354571580887,
+      "num_tokens": 22525536.0,
+      "step": 1826
+    },
+    {
+      "entropy": 1.0337569862604141,
+      "epoch": 0.9620853080568721,
+      "grad_norm": 0.3343221843242645,
+      "learning_rate": 0.00019461475477341512,
+      "loss": 0.21144059300422668,
+      "mean_token_accuracy": 0.9137226939201355,
+      "num_tokens": 22537872.0,
+      "step": 1827
+    },
+    {
+      "entropy": 1.0447833985090256,
+      "epoch": 0.9626119010005266,
+      "grad_norm": 0.3130665421485901,
+      "learning_rate": 0.0001946072808029355,
+      "loss": 0.2065916359424591,
+      "mean_token_accuracy": 0.9174522012472153,
+      "num_tokens": 22550208.0,
+      "step": 1828
+    },
+    {
+      "entropy": 1.0395597666502,
+      "epoch": 0.9631384939441812,
+      "grad_norm": 0.26665323972702026,
+      "learning_rate": 0.00019459980180981478,
+      "loss": 0.21482327580451965,
+      "mean_token_accuracy": 0.9130977690219879,
+      "num_tokens": 22562544.0,
+      "step": 1829
+    },
+    {
+      "entropy": 1.0779733955860138,
+      "epoch": 0.9636650868878357,
+      "grad_norm": 0.28825950622558594,
+      "learning_rate": 0.0001945923177944969,
+      "loss": 0.21779173612594604,
+      "mean_token_accuracy": 0.9156653732061386,
+      "num_tokens": 22574880.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.0036635249853134,
+      "epoch": 0.9641916798314902,
+      "grad_norm": 0.2941460907459259,
+      "learning_rate": 0.0001945848287574262,
+      "loss": 0.2330627739429474,
+      "mean_token_accuracy": 0.9059428721666336,
+      "num_tokens": 22587216.0,
+      "step": 1831
+    },
+    {
+      "entropy": 0.9797258973121643,
+      "epoch": 0.9647182727751448,
+      "grad_norm": 0.25932562351226807,
+      "learning_rate": 0.00019457733469904715,
+      "loss": 0.2051321119070053,
+      "mean_token_accuracy": 0.9192904084920883,
+      "num_tokens": 22599552.0,
+      "step": 1832
+    },
+    {
+      "entropy": 1.039179965853691,
+      "epoch": 0.9652448657187994,
+      "grad_norm": 0.26872527599334717,
+      "learning_rate": 0.0001945698356198047,
+      "loss": 0.21343429386615753,
+      "mean_token_accuracy": 0.9149598628282547,
+      "num_tokens": 22611888.0,
+      "step": 1833
+    },
+    {
+      "entropy": 1.0027782917022705,
+      "epoch": 0.965771458662454,
+      "grad_norm": 0.27663007378578186,
+      "learning_rate": 0.00019456233152014406,
+      "loss": 0.19677159190177917,
+      "mean_token_accuracy": 0.9209509938955307,
+      "num_tokens": 22624224.0,
+      "step": 1834
+    },
+    {
+      "entropy": 1.0312834680080414,
+      "epoch": 0.9662980516061085,
+      "grad_norm": 0.28490519523620605,
+      "learning_rate": 0.00019455482240051064,
+      "loss": 0.21659842133522034,
+      "mean_token_accuracy": 0.9108777791261673,
+      "num_tokens": 22636560.0,
+      "step": 1835
+    },
+    {
+      "entropy": 1.012413963675499,
+      "epoch": 0.966824644549763,
+      "grad_norm": 0.27517637610435486,
+      "learning_rate": 0.00019454730826135022,
+      "loss": 0.22000275552272797,
+      "mean_token_accuracy": 0.9186523258686066,
+      "num_tokens": 22648896.0,
+      "step": 1836
+    },
+    {
+      "entropy": 0.9954932332038879,
+      "epoch": 0.9673512374934176,
+      "grad_norm": 0.27069538831710815,
+      "learning_rate": 0.0001945397891031089,
+      "loss": 0.19973544776439667,
+      "mean_token_accuracy": 0.9211115539073944,
+      "num_tokens": 22661232.0,
+      "step": 1837
+    },
+    {
+      "entropy": 1.0200046300888062,
+      "epoch": 0.9678778304370721,
+      "grad_norm": 0.2899874150753021,
+      "learning_rate": 0.00019453226492623305,
+      "loss": 0.22812676429748535,
+      "mean_token_accuracy": 0.9122350364923477,
+      "num_tokens": 22673568.0,
+      "step": 1838
+    },
+    {
+      "entropy": 0.9972085207700729,
+      "epoch": 0.9684044233807267,
+      "grad_norm": 0.29899537563323975,
+      "learning_rate": 0.0001945247357311693,
+      "loss": 0.2259579598903656,
+      "mean_token_accuracy": 0.9079723209142685,
+      "num_tokens": 22685904.0,
+      "step": 1839
+    },
+    {
+      "entropy": 1.0059181898832321,
+      "epoch": 0.9689310163243813,
+      "grad_norm": 0.29635098576545715,
+      "learning_rate": 0.00019451720151836467,
+      "loss": 0.23771284520626068,
+      "mean_token_accuracy": 0.9140820056200027,
+      "num_tokens": 22698240.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0503579378128052,
+      "epoch": 0.9694576092680358,
+      "grad_norm": 0.29462575912475586,
+      "learning_rate": 0.00019450966228826635,
+      "loss": 0.19687220454216003,
+      "mean_token_accuracy": 0.9218025654554367,
+      "num_tokens": 22710576.0,
+      "step": 1841
+    },
+    {
+      "entropy": 0.9612717479467392,
+      "epoch": 0.9699842022116903,
+      "grad_norm": 0.27696606516838074,
+      "learning_rate": 0.00019450211804132196,
+      "loss": 0.19762445986270905,
+      "mean_token_accuracy": 0.9190530776977539,
+      "num_tokens": 22722912.0,
+      "step": 1842
+    },
+    {
+      "entropy": 0.9952682256698608,
+      "epoch": 0.9705107951553449,
+      "grad_norm": 0.2652910053730011,
+      "learning_rate": 0.00019449456877797933,
+      "loss": 0.21299602091312408,
+      "mean_token_accuracy": 0.9156957715749741,
+      "num_tokens": 22735248.0,
+      "step": 1843
+    },
+    {
+      "entropy": 0.9986753612756729,
+      "epoch": 0.9710373880989994,
+      "grad_norm": 0.2620997726917267,
+      "learning_rate": 0.00019448701449868662,
+      "loss": 0.1993142068386078,
+      "mean_token_accuracy": 0.9197477102279663,
+      "num_tokens": 22747584.0,
+      "step": 1844
+    },
+    {
+      "entropy": 0.9899931252002716,
+      "epoch": 0.9715639810426541,
+      "grad_norm": 0.29060348868370056,
+      "learning_rate": 0.0001944794552038923,
+      "loss": 0.2212766706943512,
+      "mean_token_accuracy": 0.9078968465328217,
+      "num_tokens": 22759920.0,
+      "step": 1845
+    },
+    {
+      "entropy": 1.0022216588258743,
+      "epoch": 0.9720905739863086,
+      "grad_norm": 0.30724501609802246,
+      "learning_rate": 0.00019447189089404513,
+      "loss": 0.2398059368133545,
+      "mean_token_accuracy": 0.9064589291810989,
+      "num_tokens": 22772256.0,
+      "step": 1846
+    },
+    {
+      "entropy": 0.997131809592247,
+      "epoch": 0.9726171669299631,
+      "grad_norm": 0.286475270986557,
+      "learning_rate": 0.0001944643215695941,
+      "loss": 0.21739161014556885,
+      "mean_token_accuracy": 0.9086156636476517,
+      "num_tokens": 22784592.0,
+      "step": 1847
+    },
+    {
+      "entropy": 0.9620841443538666,
+      "epoch": 0.9731437598736177,
+      "grad_norm": 0.2802620828151703,
+      "learning_rate": 0.00019445674723098864,
+      "loss": 0.18609827756881714,
+      "mean_token_accuracy": 0.9273043125867844,
+      "num_tokens": 22796928.0,
+      "step": 1848
+    },
+    {
+      "entropy": 0.9614392817020416,
+      "epoch": 0.9736703528172722,
+      "grad_norm": 0.2547474205493927,
+      "learning_rate": 0.0001944491678786783,
+      "loss": 0.20111055672168732,
+      "mean_token_accuracy": 0.9191482216119766,
+      "num_tokens": 22809264.0,
+      "step": 1849
+    },
+    {
+      "entropy": 0.9734077900648117,
+      "epoch": 0.9741969457609269,
+      "grad_norm": 0.2975533902645111,
+      "learning_rate": 0.00019444158351311308,
+      "loss": 0.20715618133544922,
+      "mean_token_accuracy": 0.9150333106517792,
+      "num_tokens": 22821600.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.9761214107275009,
+      "epoch": 0.9747235387045814,
+      "grad_norm": 0.28768840432167053,
+      "learning_rate": 0.0001944339941347432,
+      "loss": 0.18795141577720642,
+      "mean_token_accuracy": 0.9215870052576065,
+      "num_tokens": 22833936.0,
+      "step": 1851
+    },
+    {
+      "entropy": 0.9937604367733002,
+      "epoch": 0.9752501316482359,
+      "grad_norm": 0.29926928877830505,
+      "learning_rate": 0.00019442639974401923,
+      "loss": 0.21389541029930115,
+      "mean_token_accuracy": 0.917677566409111,
+      "num_tokens": 22846272.0,
+      "step": 1852
+    },
+    {
+      "entropy": 0.9756381958723068,
+      "epoch": 0.9757767245918905,
+      "grad_norm": 0.298100084066391,
+      "learning_rate": 0.00019441880034139198,
+      "loss": 0.21645361185073853,
+      "mean_token_accuracy": 0.91536945104599,
+      "num_tokens": 22858608.0,
+      "step": 1853
+    },
+    {
+      "entropy": 0.9328114837408066,
+      "epoch": 0.976303317535545,
+      "grad_norm": 0.27972036600112915,
+      "learning_rate": 0.0001944111959273126,
+      "loss": 0.20870962738990784,
+      "mean_token_accuracy": 0.9177060127258301,
+      "num_tokens": 22870944.0,
+      "step": 1854
+    },
+    {
+      "entropy": 0.9690214693546295,
+      "epoch": 0.9768299104791995,
+      "grad_norm": 0.28668034076690674,
+      "learning_rate": 0.00019440358650223244,
+      "loss": 0.20322152972221375,
+      "mean_token_accuracy": 0.9198893010616302,
+      "num_tokens": 22883280.0,
+      "step": 1855
+    },
+    {
+      "entropy": 0.9374023228883743,
+      "epoch": 0.9773565034228542,
+      "grad_norm": 0.2798213064670563,
+      "learning_rate": 0.00019439597206660336,
+      "loss": 0.20974165201187134,
+      "mean_token_accuracy": 0.9153610616922379,
+      "num_tokens": 22895616.0,
+      "step": 1856
+    },
+    {
+      "entropy": 0.988965705037117,
+      "epoch": 0.9778830963665087,
+      "grad_norm": 0.2775600254535675,
+      "learning_rate": 0.00019438835262087728,
+      "loss": 0.19997943937778473,
+      "mean_token_accuracy": 0.9216296821832657,
+      "num_tokens": 22907952.0,
+      "step": 1857
+    },
+    {
+      "entropy": 1.0289342552423477,
+      "epoch": 0.9784096893101633,
+      "grad_norm": 0.3432585299015045,
+      "learning_rate": 0.00019438072816550654,
+      "loss": 0.22643734514713287,
+      "mean_token_accuracy": 0.9054546356201172,
+      "num_tokens": 22920288.0,
+      "step": 1858
+    },
+    {
+      "entropy": 0.9867872446775436,
+      "epoch": 0.9789362822538178,
+      "grad_norm": 0.29411035776138306,
+      "learning_rate": 0.0001943730987009438,
+      "loss": 0.22154730558395386,
+      "mean_token_accuracy": 0.9093334376811981,
+      "num_tokens": 22932624.0,
+      "step": 1859
+    },
+    {
+      "entropy": 1.0116321295499802,
+      "epoch": 0.9794628751974723,
+      "grad_norm": 0.25875324010849,
+      "learning_rate": 0.00019436546422764198,
+      "loss": 0.19839854538440704,
+      "mean_token_accuracy": 0.9156231582164764,
+      "num_tokens": 22944960.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0039302557706833,
+      "epoch": 0.9799894681411269,
+      "grad_norm": 0.29284337162971497,
+      "learning_rate": 0.00019435782474605423,
+      "loss": 0.2374696135520935,
+      "mean_token_accuracy": 0.905433177947998,
+      "num_tokens": 22957296.0,
+      "step": 1861
+    },
+    {
+      "entropy": 1.0000396072864532,
+      "epoch": 0.9805160610847815,
+      "grad_norm": 0.24808640778064728,
+      "learning_rate": 0.00019435018025663412,
+      "loss": 0.18588939309120178,
+      "mean_token_accuracy": 0.9234057366847992,
+      "num_tokens": 22969632.0,
+      "step": 1862
+    },
+    {
+      "entropy": 1.0562945902347565,
+      "epoch": 0.981042654028436,
+      "grad_norm": 0.2809024453163147,
+      "learning_rate": 0.00019434253075983543,
+      "loss": 0.19566890597343445,
+      "mean_token_accuracy": 0.9169919043779373,
+      "num_tokens": 22981968.0,
+      "step": 1863
+    },
+    {
+      "entropy": 1.066685050725937,
+      "epoch": 0.9815692469720906,
+      "grad_norm": 0.2744297385215759,
+      "learning_rate": 0.0001943348762561123,
+      "loss": 0.21286125481128693,
+      "mean_token_accuracy": 0.911152720451355,
+      "num_tokens": 22994304.0,
+      "step": 1864
+    },
+    {
+      "entropy": 1.0672977268695831,
+      "epoch": 0.9820958399157451,
+      "grad_norm": 0.30773451924324036,
+      "learning_rate": 0.00019432721674591906,
+      "loss": 0.22466342151165009,
+      "mean_token_accuracy": 0.9027170091867447,
+      "num_tokens": 23006640.0,
+      "step": 1865
+    },
+    {
+      "entropy": 0.9817476272583008,
+      "epoch": 0.9826224328593997,
+      "grad_norm": 0.25929689407348633,
+      "learning_rate": 0.00019431955222971048,
+      "loss": 0.20428617298603058,
+      "mean_token_accuracy": 0.9160020500421524,
+      "num_tokens": 23018976.0,
+      "step": 1866
+    },
+    {
+      "entropy": 1.0060690939426422,
+      "epoch": 0.9831490258030542,
+      "grad_norm": 0.3303799629211426,
+      "learning_rate": 0.00019431188270794153,
+      "loss": 0.22543492913246155,
+      "mean_token_accuracy": 0.9112392216920853,
+      "num_tokens": 23031312.0,
+      "step": 1867
+    },
+    {
+      "entropy": 0.9948325008153915,
+      "epoch": 0.9836756187467088,
+      "grad_norm": 0.2678799331188202,
+      "learning_rate": 0.00019430420818106753,
+      "loss": 0.19793486595153809,
+      "mean_token_accuracy": 0.9236938953399658,
+      "num_tokens": 23043648.0,
+      "step": 1868
+    },
+    {
+      "entropy": 1.0155895799398422,
+      "epoch": 0.9842022116903634,
+      "grad_norm": 0.32058584690093994,
+      "learning_rate": 0.00019429652864954403,
+      "loss": 0.22886891663074493,
+      "mean_token_accuracy": 0.9127451777458191,
+      "num_tokens": 23055984.0,
+      "step": 1869
+    },
+    {
+      "entropy": 1.0170871764421463,
+      "epoch": 0.9847288046340179,
+      "grad_norm": 0.2957039177417755,
+      "learning_rate": 0.00019428884411382694,
+      "loss": 0.21584120392799377,
+      "mean_token_accuracy": 0.9092356264591217,
+      "num_tokens": 23068320.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0231945216655731,
+      "epoch": 0.9852553975776724,
+      "grad_norm": 0.2651207149028778,
+      "learning_rate": 0.00019428115457437249,
+      "loss": 0.18167884647846222,
+      "mean_token_accuracy": 0.9309497475624084,
+      "num_tokens": 23080656.0,
+      "step": 1871
+    },
+    {
+      "entropy": 0.9718558192253113,
+      "epoch": 0.985781990521327,
+      "grad_norm": 0.29894301295280457,
+      "learning_rate": 0.0001942734600316371,
+      "loss": 0.20839130878448486,
+      "mean_token_accuracy": 0.9109772592782974,
+      "num_tokens": 23092992.0,
+      "step": 1872
+    },
+    {
+      "entropy": 1.0036133825778961,
+      "epoch": 0.9863085834649815,
+      "grad_norm": 0.28074634075164795,
+      "learning_rate": 0.00019426576048607754,
+      "loss": 0.21850639581680298,
+      "mean_token_accuracy": 0.9151745438575745,
+      "num_tokens": 23105328.0,
+      "step": 1873
+    },
+    {
+      "entropy": 1.0208822190761566,
+      "epoch": 0.9868351764086362,
+      "grad_norm": 0.27776047587394714,
+      "learning_rate": 0.00019425805593815094,
+      "loss": 0.1931566298007965,
+      "mean_token_accuracy": 0.9207221418619156,
+      "num_tokens": 23117664.0,
+      "step": 1874
+    },
+    {
+      "entropy": 0.9643638432025909,
+      "epoch": 0.9873617693522907,
+      "grad_norm": 0.28227487206459045,
+      "learning_rate": 0.00019425034638831466,
+      "loss": 0.2174389511346817,
+      "mean_token_accuracy": 0.9123106747865677,
+      "num_tokens": 23130000.0,
+      "step": 1875
+    },
+    {
+      "entropy": 1.0151402354240417,
+      "epoch": 0.9878883622959452,
+      "grad_norm": 0.3007744252681732,
+      "learning_rate": 0.00019424263183702634,
+      "loss": 0.23054783046245575,
+      "mean_token_accuracy": 0.9074221700429916,
+      "num_tokens": 23142336.0,
+      "step": 1876
+    },
+    {
+      "entropy": 1.023296281695366,
+      "epoch": 0.9884149552395998,
+      "grad_norm": 0.2875635623931885,
+      "learning_rate": 0.000194234912284744,
+      "loss": 0.2183448076248169,
+      "mean_token_accuracy": 0.9146157503128052,
+      "num_tokens": 23154672.0,
+      "step": 1877
+    },
+    {
+      "entropy": 1.025291085243225,
+      "epoch": 0.9889415481832543,
+      "grad_norm": 0.2833808660507202,
+      "learning_rate": 0.00019422718773192584,
+      "loss": 0.2020779550075531,
+      "mean_token_accuracy": 0.9207105487585068,
+      "num_tokens": 23167008.0,
+      "step": 1878
+    },
+    {
+      "entropy": 1.0463321805000305,
+      "epoch": 0.989468141126909,
+      "grad_norm": 0.27166518568992615,
+      "learning_rate": 0.0001942194581790305,
+      "loss": 0.21112844347953796,
+      "mean_token_accuracy": 0.9151988327503204,
+      "num_tokens": 23179344.0,
+      "step": 1879
+    },
+    {
+      "entropy": 1.0174481570720673,
+      "epoch": 0.9899947340705635,
+      "grad_norm": 0.28479474782943726,
+      "learning_rate": 0.00019421172362651677,
+      "loss": 0.2219792902469635,
+      "mean_token_accuracy": 0.9093788266181946,
+      "num_tokens": 23191680.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.068659245967865,
+      "epoch": 0.990521327014218,
+      "grad_norm": 0.27825888991355896,
+      "learning_rate": 0.00019420398407484383,
+      "loss": 0.2115180492401123,
+      "mean_token_accuracy": 0.913215160369873,
+      "num_tokens": 23204016.0,
+      "step": 1881
+    },
+    {
+      "entropy": 1.0102509111166,
+      "epoch": 0.9910479199578726,
+      "grad_norm": 0.2567676901817322,
+      "learning_rate": 0.00019419623952447113,
+      "loss": 0.1962418258190155,
+      "mean_token_accuracy": 0.9181914627552032,
+      "num_tokens": 23216352.0,
+      "step": 1882
+    },
+    {
+      "entropy": 1.0041260570287704,
+      "epoch": 0.9915745129015271,
+      "grad_norm": 0.2607569992542267,
+      "learning_rate": 0.00019418848997585842,
+      "loss": 0.18007442355155945,
+      "mean_token_accuracy": 0.9283247590065002,
+      "num_tokens": 23228688.0,
+      "step": 1883
+    },
+    {
+      "entropy": 1.0119262784719467,
+      "epoch": 0.9921011058451816,
+      "grad_norm": 0.2691285312175751,
+      "learning_rate": 0.00019418073542946577,
+      "loss": 0.2185424566268921,
+      "mean_token_accuracy": 0.9108522534370422,
+      "num_tokens": 23241024.0,
+      "step": 1884
+    },
+    {
+      "entropy": 1.0200906693935394,
+      "epoch": 0.9926276987888363,
+      "grad_norm": 0.2762972414493561,
+      "learning_rate": 0.0001941729758857535,
+      "loss": 0.21047905087471008,
+      "mean_token_accuracy": 0.921312689781189,
+      "num_tokens": 23253360.0,
+      "step": 1885
+    },
+    {
+      "entropy": 1.0015190541744232,
+      "epoch": 0.9931542917324908,
+      "grad_norm": 0.2789957523345947,
+      "learning_rate": 0.0001941652113451822,
+      "loss": 0.22564658522605896,
+      "mean_token_accuracy": 0.9052274376153946,
+      "num_tokens": 23265696.0,
+      "step": 1886
+    },
+    {
+      "entropy": 1.0212842226028442,
+      "epoch": 0.9936808846761453,
+      "grad_norm": 0.2787906527519226,
+      "learning_rate": 0.00019415744180821293,
+      "loss": 0.19954261183738708,
+      "mean_token_accuracy": 0.9225759655237198,
+      "num_tokens": 23278032.0,
+      "step": 1887
+    },
+    {
+      "entropy": 1.0256837606430054,
+      "epoch": 0.9942074776197999,
+      "grad_norm": 0.3047766089439392,
+      "learning_rate": 0.0001941496672753068,
+      "loss": 0.22101764380931854,
+      "mean_token_accuracy": 0.9100571721792221,
+      "num_tokens": 23290368.0,
+      "step": 1888
+    },
+    {
+      "entropy": 1.0017095357179642,
+      "epoch": 0.9947340705634544,
+      "grad_norm": 0.3079693913459778,
+      "learning_rate": 0.00019414188774692542,
+      "loss": 0.23877866566181183,
+      "mean_token_accuracy": 0.9044640958309174,
+      "num_tokens": 23302704.0,
+      "step": 1889
+    },
+    {
+      "entropy": 0.9979955106973648,
+      "epoch": 0.995260663507109,
+      "grad_norm": 0.28942957520484924,
+      "learning_rate": 0.00019413410322353055,
+      "loss": 0.21703916788101196,
+      "mean_token_accuracy": 0.911283403635025,
+      "num_tokens": 23315040.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.000881314277649,
+      "epoch": 0.9957872564507636,
+      "grad_norm": 0.2723085582256317,
+      "learning_rate": 0.00019412631370558433,
+      "loss": 0.2083275318145752,
+      "mean_token_accuracy": 0.9137948602437973,
+      "num_tokens": 23327376.0,
+      "step": 1891
+    },
+    {
+      "entropy": 0.9760307371616364,
+      "epoch": 0.9963138493944181,
+      "grad_norm": 0.28710952401161194,
+      "learning_rate": 0.00019411851919354925,
+      "loss": 0.2206597775220871,
+      "mean_token_accuracy": 0.913771316409111,
+      "num_tokens": 23339712.0,
+      "step": 1892
+    },
+    {
+      "entropy": 1.0313685238361359,
+      "epoch": 0.9968404423380727,
+      "grad_norm": 0.2784387767314911,
+      "learning_rate": 0.00019411071968788792,
+      "loss": 0.20836740732192993,
+      "mean_token_accuracy": 0.9158921986818314,
+      "num_tokens": 23352048.0,
+      "step": 1893
+    },
+    {
+      "entropy": 0.9972709268331528,
+      "epoch": 0.9973670352817272,
+      "grad_norm": 0.2760670781135559,
+      "learning_rate": 0.00019410291518906337,
+      "loss": 0.215082049369812,
+      "mean_token_accuracy": 0.9125083088874817,
+      "num_tokens": 23364384.0,
+      "step": 1894
+    },
+    {
+      "entropy": 0.9749780297279358,
+      "epoch": 0.9978936282253817,
+      "grad_norm": 0.2756492793560028,
+      "learning_rate": 0.00019409510569753897,
+      "loss": 0.19914628565311432,
+      "mean_token_accuracy": 0.9158760160207748,
+      "num_tokens": 23376720.0,
+      "step": 1895
+    },
+    {
+      "entropy": 0.9882865399122238,
+      "epoch": 0.9984202211690363,
+      "grad_norm": 0.3229084610939026,
+      "learning_rate": 0.00019408729121377828,
+      "loss": 0.21740438044071198,
+      "mean_token_accuracy": 0.9149737656116486,
+      "num_tokens": 23389056.0,
+      "step": 1896
+    },
+    {
+      "entropy": 0.979081466794014,
+      "epoch": 0.9989468141126909,
+      "grad_norm": 0.2839930057525635,
+      "learning_rate": 0.00019407947173824522,
+      "loss": 0.20839889347553253,
+      "mean_token_accuracy": 0.9204780906438828,
+      "num_tokens": 23401392.0,
+      "step": 1897
+    },
+    {
+      "entropy": 0.9407005459070206,
+      "epoch": 0.9994734070563455,
+      "grad_norm": 0.28035539388656616,
+      "learning_rate": 0.00019407164727140396,
+      "loss": 0.2062290906906128,
+      "mean_token_accuracy": 0.9152461290359497,
+      "num_tokens": 23413728.0,
+      "step": 1898
+    },
+    {
+      "entropy": 0.9624819904565811,
+      "epoch": 1.0,
+      "grad_norm": 0.29015156626701355,
+      "learning_rate": 0.00019406381781371902,
+      "loss": 0.21500550210475922,
+      "mean_token_accuracy": 0.915779635310173,
+      "num_tokens": 23425036.0,
+      "step": 1899
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.925739836008565,
+      "eval_loss": 0.21148517727851868,
+      "eval_mean_token_accuracy": 0.9153817107703826,
+      "eval_num_tokens": 23425036.0,
+      "eval_runtime": 664.4914,
+      "eval_samples_per_second": 8.573,
+      "eval_steps_per_second": 2.858,
+      "step": 1899
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 13293,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 7,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6346022728450111e+19,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}