diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8074 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 804,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.136738121509552,
+      "epoch": 0.003738317757009346,
+      "grad_norm": 0.42455214262008667,
+      "learning_rate": 0.0002,
+      "loss": 2.479450225830078,
+      "mean_token_accuracy": 0.5328910201787949,
+      "num_tokens": 16274.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.2458838522434235,
+      "epoch": 0.007476635514018692,
+      "grad_norm": 0.37535905838012695,
+      "learning_rate": 0.0002,
+      "loss": 2.177677869796753,
+      "mean_token_accuracy": 0.5608605146408081,
+      "num_tokens": 32608.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.4101199209690094,
+      "epoch": 0.011214953271028037,
+      "grad_norm": 0.2836460769176483,
+      "learning_rate": 0.0002,
+      "loss": 1.7183680534362793,
+      "mean_token_accuracy": 0.5886513292789459,
+      "num_tokens": 49069.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.3778190314769745,
+      "epoch": 0.014953271028037384,
+      "grad_norm": 0.234278604388237,
+      "learning_rate": 0.0002,
+      "loss": 1.4014493227005005,
+      "mean_token_accuracy": 0.6345515549182892,
+      "num_tokens": 65545.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.3580558598041534,
+      "epoch": 0.018691588785046728,
+      "grad_norm": 0.25148847699165344,
+      "learning_rate": 0.0002,
+      "loss": 1.2977453470230103,
+      "mean_token_accuracy": 0.6410679519176483,
+      "num_tokens": 81930.0,
+      "step": 5
+    },
+    {
+      "entropy": 1.2881037592887878,
+      "epoch": 0.022429906542056073,
+      "grad_norm": 0.1429636925458908,
+      "learning_rate": 0.0002,
+      "loss": 1.200700283050537,
+      "mean_token_accuracy": 0.6533316373825073,
+      "num_tokens": 97974.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.2034580409526825,
+      "epoch": 0.026168224299065422,
+      "grad_norm": 0.10399583727121353,
+      "learning_rate": 0.0002,
+      "loss": 1.099328637123108,
+      "mean_token_accuracy": 0.667132779955864,
+      "num_tokens": 114155.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.1255775392055511,
+      "epoch": 0.029906542056074768,
+      "grad_norm": 0.11593594402074814,
+      "learning_rate": 0.0002,
+      "loss": 1.0287823677062988,
+      "mean_token_accuracy": 0.676301121711731,
+      "num_tokens": 130551.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.035294532775879,
+      "epoch": 0.03364485981308411,
+      "grad_norm": 0.1378099024295807,
+      "learning_rate": 0.0002,
+      "loss": 0.9908202290534973,
+      "mean_token_accuracy": 0.6857471317052841,
+      "num_tokens": 146573.0,
+      "step": 9
+    },
+    {
+      "entropy": 0.9605472087860107,
+      "epoch": 0.037383177570093455,
+      "grad_norm": 0.14775174856185913,
+      "learning_rate": 0.0002,
+      "loss": 0.9240690469741821,
+      "mean_token_accuracy": 0.6948150098323822,
+      "num_tokens": 162866.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.9512569606304169,
+      "epoch": 0.041121495327102804,
+      "grad_norm": 0.11748532950878143,
+      "learning_rate": 0.0002,
+      "loss": 0.8622909784317017,
+      "mean_token_accuracy": 0.7091559618711472,
+      "num_tokens": 179220.0,
+      "step": 11
+    },
+    {
+      "entropy": 0.9225088059902191,
+      "epoch": 0.044859813084112146,
+      "grad_norm": 0.12350241094827652,
+      "learning_rate": 0.0002,
+      "loss": 0.822359025478363,
+      "mean_token_accuracy": 0.7142422199249268,
+      "num_tokens": 195296.0,
+      "step": 12
+    },
+    {
+      "entropy": 0.8315692692995071,
+      "epoch": 0.048598130841121495,
+      "grad_norm": 1.498412013053894,
+      "learning_rate": 0.0002,
+      "loss": 0.857573390007019,
+      "mean_token_accuracy": 0.7114518880844116,
+      "num_tokens": 211304.0,
+      "step": 13
+    },
+    {
+      "entropy": 0.8135845959186554,
+      "epoch": 0.052336448598130844,
+      "grad_norm": 0.24739782512187958,
+      "learning_rate": 0.0002,
+      "loss": 0.7810372114181519,
+      "mean_token_accuracy": 0.7160922139883041,
+      "num_tokens": 227628.0,
+      "step": 14
+    },
+    {
+      "entropy": 0.7939814329147339,
+      "epoch": 0.056074766355140186,
+      "grad_norm": 0.09629958122968674,
+      "learning_rate": 0.0002,
+      "loss": 0.7519910335540771,
+      "mean_token_accuracy": 0.7263199836015701,
+      "num_tokens": 243754.0,
+      "step": 15
+    },
+    {
+      "entropy": 0.7665929198265076,
+      "epoch": 0.059813084112149535,
+      "grad_norm": 0.097053661942482,
+      "learning_rate": 0.0002,
+      "loss": 0.7232163548469543,
+      "mean_token_accuracy": 0.7332944422960281,
+      "num_tokens": 260330.0,
+      "step": 16
+    },
+    {
+      "entropy": 0.725548505783081,
+      "epoch": 0.06355140186915888,
+      "grad_norm": 0.07506023347377777,
+      "learning_rate": 0.0002,
+      "loss": 0.7072620987892151,
+      "mean_token_accuracy": 0.7361972779035568,
+      "num_tokens": 276558.0,
+      "step": 17
+    },
+    {
+      "entropy": 0.6787595748901367,
+      "epoch": 0.06728971962616823,
+      "grad_norm": 0.09339383244514465,
+      "learning_rate": 0.0002,
+      "loss": 0.6842125654220581,
+      "mean_token_accuracy": 0.7451010048389435,
+      "num_tokens": 293015.0,
+      "step": 18
+    },
+    {
+      "entropy": 0.6606214493513107,
+      "epoch": 0.07102803738317758,
+      "grad_norm": 0.1005033329129219,
+      "learning_rate": 0.0002,
+      "loss": 0.67589271068573,
+      "mean_token_accuracy": 0.7439846247434616,
+      "num_tokens": 309125.0,
+      "step": 19
+    },
+    {
+      "entropy": 0.6459693312644958,
+      "epoch": 0.07476635514018691,
+      "grad_norm": 0.08065377175807953,
+      "learning_rate": 0.0002,
+      "loss": 0.6449049711227417,
+      "mean_token_accuracy": 0.7484360784292221,
+      "num_tokens": 325340.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6585913002490997,
+      "epoch": 0.07850467289719626,
+      "grad_norm": 0.0826243907213211,
+      "learning_rate": 0.0002,
+      "loss": 0.6479358673095703,
+      "mean_token_accuracy": 0.7465608268976212,
+      "num_tokens": 341526.0,
+      "step": 21
+    },
+    {
+      "entropy": 0.6610395163297653,
+      "epoch": 0.08224299065420561,
+      "grad_norm": 0.11156535893678665,
+      "learning_rate": 0.0002,
+      "loss": 0.6428132057189941,
+      "mean_token_accuracy": 0.7519424259662628,
+      "num_tokens": 357723.0,
+      "step": 22
+    },
+    {
+      "entropy": 0.6467725187540054,
+      "epoch": 0.08598130841121496,
+      "grad_norm": 0.07777491956949234,
+      "learning_rate": 0.0002,
+      "loss": 0.6323320269584656,
+      "mean_token_accuracy": 0.7518283724784851,
+      "num_tokens": 374088.0,
+      "step": 23
+    },
+    {
+      "entropy": 0.6338784694671631,
+      "epoch": 0.08971962616822429,
+      "grad_norm": 0.067124143242836,
+      "learning_rate": 0.0002,
+      "loss": 0.6243886351585388,
+      "mean_token_accuracy": 0.7597910910844803,
+      "num_tokens": 390225.0,
+      "step": 24
+    },
+    {
+      "entropy": 0.6436891853809357,
+      "epoch": 0.09345794392523364,
+      "grad_norm": 0.07950877398252487,
+      "learning_rate": 0.0002,
+      "loss": 0.6342618465423584,
+      "mean_token_accuracy": 0.7521715611219406,
+      "num_tokens": 406768.0,
+      "step": 25
+    },
+    {
+      "entropy": 0.6160638332366943,
+      "epoch": 0.09719626168224299,
+      "grad_norm": 0.07700544595718384,
+      "learning_rate": 0.0002,
+      "loss": 0.6131497621536255,
+      "mean_token_accuracy": 0.7559348195791245,
+      "num_tokens": 422877.0,
+      "step": 26
+    },
+    {
+      "entropy": 0.628534808754921,
+      "epoch": 0.10093457943925234,
+      "grad_norm": 0.06708905845880508,
+      "learning_rate": 0.0002,
+      "loss": 0.6215195059776306,
+      "mean_token_accuracy": 0.7550520598888397,
+      "num_tokens": 439151.0,
+      "step": 27
+    },
+    {
+      "entropy": 0.6049717515707016,
+      "epoch": 0.10467289719626169,
+      "grad_norm": 0.06930189579725266,
+      "learning_rate": 0.0002,
+      "loss": 0.6006554961204529,
+      "mean_token_accuracy": 0.7655697911977768,
+      "num_tokens": 455376.0,
+      "step": 28
+    },
+    {
+      "entropy": 0.6179223954677582,
+      "epoch": 0.10841121495327102,
+      "grad_norm": 0.07139614224433899,
+      "learning_rate": 0.0002,
+      "loss": 0.6130858659744263,
+      "mean_token_accuracy": 0.757579579949379,
+      "num_tokens": 471522.0,
+      "step": 29
+    },
+    {
+      "entropy": 0.5762445032596588,
+      "epoch": 0.11214953271028037,
+      "grad_norm": 0.06326469033956528,
+      "learning_rate": 0.0002,
+      "loss": 0.5740499496459961,
+      "mean_token_accuracy": 0.7769385576248169,
+      "num_tokens": 487795.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.5995292961597443,
+      "epoch": 0.11588785046728972,
+      "grad_norm": 0.05662618950009346,
+      "learning_rate": 0.0002,
+      "loss": 0.5942716002464294,
+      "mean_token_accuracy": 0.762609601020813,
+      "num_tokens": 503903.0,
+      "step": 31
+    },
+    {
+      "entropy": 0.564709946513176,
+      "epoch": 0.11962616822429907,
+      "grad_norm": 0.05710163712501526,
+      "learning_rate": 0.0002,
+      "loss": 0.5685403347015381,
+      "mean_token_accuracy": 0.7732207477092743,
+      "num_tokens": 520130.0,
+      "step": 32
+    },
+    {
+      "entropy": 0.5898115634918213,
+      "epoch": 0.1233644859813084,
+      "grad_norm": 0.06266085058450699,
+      "learning_rate": 0.0002,
+      "loss": 0.5901201963424683,
+      "mean_token_accuracy": 0.7677827328443527,
+      "num_tokens": 536535.0,
+      "step": 33
+    },
+    {
+      "entropy": 0.5797764509916306,
+      "epoch": 0.12710280373831775,
+      "grad_norm": 0.053195368498563766,
+      "learning_rate": 0.0002,
+      "loss": 0.5766412615776062,
+      "mean_token_accuracy": 0.7718744575977325,
+      "num_tokens": 552762.0,
+      "step": 34
+    },
+    {
+      "entropy": 0.5923163294792175,
+      "epoch": 0.1308411214953271,
+      "grad_norm": 0.058649640530347824,
+      "learning_rate": 0.0002,
+      "loss": 0.588531494140625,
+      "mean_token_accuracy": 0.7659970670938492,
+      "num_tokens": 569081.0,
+      "step": 35
+    },
+    {
+      "entropy": 0.598005622625351,
+      "epoch": 0.13457943925233645,
+      "grad_norm": 0.05565391853451729,
+      "learning_rate": 0.0002,
+      "loss": 0.5939058661460876,
+      "mean_token_accuracy": 0.7645552158355713,
+      "num_tokens": 585378.0,
+      "step": 36
+    },
+    {
+      "entropy": 0.568940594792366,
+      "epoch": 0.1383177570093458,
+      "grad_norm": 0.05046789348125458,
+      "learning_rate": 0.0002,
+      "loss": 0.5695131421089172,
+      "mean_token_accuracy": 0.76941779255867,
+      "num_tokens": 601818.0,
+      "step": 37
+    },
+    {
+      "entropy": 0.5911677330732346,
+      "epoch": 0.14205607476635515,
+      "grad_norm": 0.05749228596687317,
+      "learning_rate": 0.0002,
+      "loss": 0.595350444316864,
+      "mean_token_accuracy": 0.7632527947425842,
+      "num_tokens": 617971.0,
+      "step": 38
+    },
+    {
+      "entropy": 0.578888863325119,
+      "epoch": 0.14579439252336449,
+      "grad_norm": 0.05733935162425041,
+      "learning_rate": 0.0002,
+      "loss": 0.5821264982223511,
+      "mean_token_accuracy": 0.7635951936244965,
+      "num_tokens": 634366.0,
+      "step": 39
+    },
+    {
+      "entropy": 0.580509215593338,
+      "epoch": 0.14953271028037382,
+      "grad_norm": 0.04922669008374214,
+      "learning_rate": 0.0002,
+      "loss": 0.5817846059799194,
+      "mean_token_accuracy": 0.7654395997524261,
+      "num_tokens": 650944.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.581391915678978,
+      "epoch": 0.15327102803738318,
+      "grad_norm": 0.05423928052186966,
+      "learning_rate": 0.0002,
+      "loss": 0.5844112634658813,
+      "mean_token_accuracy": 0.7646835446357727,
+      "num_tokens": 667338.0,
+      "step": 41
+    },
+    {
+      "entropy": 0.5903676152229309,
+      "epoch": 0.15700934579439252,
+      "grad_norm": 0.04594385251402855,
+      "learning_rate": 0.0002,
+      "loss": 0.5891512036323547,
+      "mean_token_accuracy": 0.7634948194026947,
+      "num_tokens": 683821.0,
+      "step": 42
+    },
+    {
+      "entropy": 0.5767852067947388,
+      "epoch": 0.16074766355140188,
+      "grad_norm": 0.04914949834346771,
+      "learning_rate": 0.0002,
+      "loss": 0.5805380344390869,
+      "mean_token_accuracy": 0.7673552930355072,
+      "num_tokens": 700040.0,
+      "step": 43
+    },
+    {
+      "entropy": 0.5833797305822372,
+      "epoch": 0.16448598130841122,
+      "grad_norm": 0.044859349727630615,
+      "learning_rate": 0.0002,
+      "loss": 0.5833380818367004,
+      "mean_token_accuracy": 0.7672955095767975,
+      "num_tokens": 716503.0,
+      "step": 44
+    },
+    {
+      "entropy": 0.5525845289230347,
+      "epoch": 0.16822429906542055,
+      "grad_norm": 0.04274710640311241,
+      "learning_rate": 0.0002,
+      "loss": 0.5535459518432617,
+      "mean_token_accuracy": 0.7772536426782608,
+      "num_tokens": 732556.0,
+      "step": 45
+    },
+    {
+      "entropy": 0.591767281293869,
+      "epoch": 0.17196261682242991,
+      "grad_norm": 0.047692377120256424,
+      "learning_rate": 0.0002,
+      "loss": 0.5856772661209106,
+      "mean_token_accuracy": 0.7700498551130295,
+      "num_tokens": 749181.0,
+      "step": 46
+    },
+    {
+      "entropy": 0.5679502487182617,
+      "epoch": 0.17570093457943925,
+      "grad_norm": 0.04246673360466957,
+      "learning_rate": 0.0002,
+      "loss": 0.5660921931266785,
+      "mean_token_accuracy": 0.7716551572084427,
+      "num_tokens": 765518.0,
+      "step": 47
+    },
+    {
+      "entropy": 0.5837526619434357,
+      "epoch": 0.17943925233644858,
+      "grad_norm": 0.04593057557940483,
+      "learning_rate": 0.0002,
+      "loss": 0.584723174571991,
+      "mean_token_accuracy": 0.7659307718276978,
+      "num_tokens": 782078.0,
+      "step": 48
+    },
+    {
+      "entropy": 0.5635000020265579,
+      "epoch": 0.18317757009345795,
+      "grad_norm": 0.04482231289148331,
+      "learning_rate": 0.0002,
+      "loss": 0.5647243857383728,
+      "mean_token_accuracy": 0.7718922346830368,
+      "num_tokens": 798351.0,
+      "step": 49
+    },
+    {
+      "entropy": 0.5844053775072098,
+      "epoch": 0.18691588785046728,
+      "grad_norm": 0.037112969905138016,
+      "learning_rate": 0.0002,
+      "loss": 0.583613395690918,
+      "mean_token_accuracy": 0.7636143714189529,
+      "num_tokens": 814901.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.5725089907646179,
+      "epoch": 0.19065420560747665,
+      "grad_norm": 0.04044761508703232,
+      "learning_rate": 0.0002,
+      "loss": 0.572070300579071,
+      "mean_token_accuracy": 0.7681350558996201,
+      "num_tokens": 831260.0,
+      "step": 51
+    },
+    {
+      "entropy": 0.5514933615922928,
+      "epoch": 0.19439252336448598,
+      "grad_norm": 0.048450976610183716,
+      "learning_rate": 0.0002,
+      "loss": 0.5633155107498169,
+      "mean_token_accuracy": 0.7741293609142303,
+      "num_tokens": 847432.0,
+      "step": 52
+    },
+    {
+      "entropy": 0.5666041970252991,
+      "epoch": 0.19813084112149532,
+      "grad_norm": 0.037479840219020844,
+      "learning_rate": 0.0002,
+      "loss": 0.5648237466812134,
+      "mean_token_accuracy": 0.7711822837591171,
+      "num_tokens": 863773.0,
+      "step": 53
+    },
+    {
+      "entropy": 0.5806857347488403,
+      "epoch": 0.20186915887850468,
+      "grad_norm": 0.041213128715753555,
+      "learning_rate": 0.0002,
+      "loss": 0.5815138816833496,
+      "mean_token_accuracy": 0.7682100683450699,
+      "num_tokens": 880158.0,
+      "step": 54
+    },
+    {
+      "entropy": 0.5745600759983063,
+      "epoch": 0.205607476635514,
+      "grad_norm": 0.043696995824575424,
+      "learning_rate": 0.0002,
+      "loss": 0.568037211894989,
+      "mean_token_accuracy": 0.7706614434719086,
+      "num_tokens": 896711.0,
+      "step": 55
+    },
+    {
+      "entropy": 0.5852600038051605,
+      "epoch": 0.20934579439252338,
+      "grad_norm": 0.04500531032681465,
+      "learning_rate": 0.0002,
+      "loss": 0.5849421620368958,
+      "mean_token_accuracy": 0.7663431912660599,
+      "num_tokens": 913375.0,
+      "step": 56
+    },
+    {
+      "entropy": 0.5551716238260269,
+      "epoch": 0.2130841121495327,
+      "grad_norm": 0.04126116633415222,
+      "learning_rate": 0.0002,
+      "loss": 0.5583643913269043,
+      "mean_token_accuracy": 0.7752658873796463,
+      "num_tokens": 929701.0,
+      "step": 57
+    },
+    {
+      "entropy": 0.5559567362070084,
+      "epoch": 0.21682242990654205,
+      "grad_norm": 0.03906504437327385,
+      "learning_rate": 0.0002,
+      "loss": 0.5586991310119629,
+      "mean_token_accuracy": 0.7782035171985626,
+      "num_tokens": 945982.0,
+      "step": 58
+    },
+    {
+      "entropy": 0.5741323977708817,
+      "epoch": 0.2205607476635514,
+      "grad_norm": 0.044159967452287674,
+      "learning_rate": 0.0002,
+      "loss": 0.5750066041946411,
+      "mean_token_accuracy": 0.7701967805624008,
+      "num_tokens": 962111.0,
+      "step": 59
+    },
+    {
+      "entropy": 0.585385873913765,
+      "epoch": 0.22429906542056074,
+      "grad_norm": 0.03850168734788895,
+      "learning_rate": 0.0002,
+      "loss": 0.5819919109344482,
+      "mean_token_accuracy": 0.7681078165769577,
+      "num_tokens": 978456.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.5729155838489532,
+      "epoch": 0.22803738317757008,
+      "grad_norm": 0.04180893301963806,
+      "learning_rate": 0.0002,
+      "loss": 0.5612373948097229,
+      "mean_token_accuracy": 0.7728097885847092,
+      "num_tokens": 994829.0,
+      "step": 61
+    },
+    {
+      "entropy": 0.5742225646972656,
+      "epoch": 0.23177570093457944,
+      "grad_norm": 0.0462227538228035,
+      "learning_rate": 0.0002,
+      "loss": 0.5630587339401245,
+      "mean_token_accuracy": 0.7737539410591125,
+      "num_tokens": 1011036.0,
+      "step": 62
+    },
+    {
+      "entropy": 0.5770351439714432,
+      "epoch": 0.23551401869158878,
+      "grad_norm": 0.03771355003118515,
+      "learning_rate": 0.0002,
+      "loss": 0.5766980051994324,
+      "mean_token_accuracy": 0.7672136276960373,
+      "num_tokens": 1027230.0,
+      "step": 63
+    },
+    {
+      "entropy": 0.5691997706890106,
+      "epoch": 0.23925233644859814,
+      "grad_norm": 0.04262315854430199,
+      "learning_rate": 0.0002,
+      "loss": 0.574401319026947,
+      "mean_token_accuracy": 0.768265426158905,
+      "num_tokens": 1043682.0,
+      "step": 64
+    },
+    {
+      "entropy": 0.5537833720445633,
+      "epoch": 0.24299065420560748,
+      "grad_norm": 0.04550046846270561,
+      "learning_rate": 0.0002,
+      "loss": 0.562879741191864,
+      "mean_token_accuracy": 0.7726933360099792,
+      "num_tokens": 1059653.0,
+      "step": 65
+    },
+    {
+      "entropy": 0.5562547147274017,
+      "epoch": 0.2467289719626168,
+      "grad_norm": 0.039495520293712616,
+      "learning_rate": 0.0002,
+      "loss": 0.5617212653160095,
+      "mean_token_accuracy": 0.7734608948230743,
+      "num_tokens": 1076229.0,
+      "step": 66
+    },
+    {
+      "entropy": 0.555762991309166,
+      "epoch": 0.2504672897196262,
+      "grad_norm": 0.04136459901928902,
+      "learning_rate": 0.0002,
+      "loss": 0.5561961531639099,
+      "mean_token_accuracy": 0.7754806876182556,
+      "num_tokens": 1092742.0,
+      "step": 67
+    },
+    {
+      "entropy": 0.560298278927803,
+      "epoch": 0.2542056074766355,
+      "grad_norm": 0.036972060799598694,
+      "learning_rate": 0.0002,
+      "loss": 0.5573002696037292,
+      "mean_token_accuracy": 0.7789498865604401,
+      "num_tokens": 1109145.0,
+      "step": 68
+    },
+    {
+      "entropy": 0.5759327560663223,
+      "epoch": 0.25794392523364484,
+      "grad_norm": 0.03908069431781769,
+      "learning_rate": 0.0002,
+      "loss": 0.5715816617012024,
+      "mean_token_accuracy": 0.7693368047475815,
+      "num_tokens": 1125662.0,
+      "step": 69
+    },
+    {
+      "entropy": 0.570976510643959,
+      "epoch": 0.2616822429906542,
+      "grad_norm": 0.036351628601551056,
+      "learning_rate": 0.0002,
+      "loss": 0.5671893358230591,
+      "mean_token_accuracy": 0.7697141170501709,
+      "num_tokens": 1141777.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.5589511841535568,
+      "epoch": 0.26542056074766357,
+      "grad_norm": 0.042053237557411194,
+      "learning_rate": 0.0002,
+      "loss": 0.5577408075332642,
+      "mean_token_accuracy": 0.774786502122879,
+      "num_tokens": 1158296.0,
+      "step": 71
+    },
+    {
+      "entropy": 0.5530008673667908,
+      "epoch": 0.2691588785046729,
+      "grad_norm": 0.03951295465230942,
+      "learning_rate": 0.0002,
+      "loss": 0.557781994342804,
+      "mean_token_accuracy": 0.7735090106725693,
+      "num_tokens": 1174440.0,
+      "step": 72
+    },
+    {
+      "entropy": 0.5726254433393478,
+      "epoch": 0.27289719626168224,
+      "grad_norm": 0.0380430705845356,
+      "learning_rate": 0.0002,
+      "loss": 0.5711798667907715,
+      "mean_token_accuracy": 0.7688230872154236,
+      "num_tokens": 1190746.0,
+      "step": 73
+    },
+    {
+      "entropy": 0.5716652125120163,
+      "epoch": 0.2766355140186916,
+      "grad_norm": 0.04010101035237312,
+      "learning_rate": 0.0002,
+      "loss": 0.5822720527648926,
+      "mean_token_accuracy": 0.7654355019330978,
+      "num_tokens": 1207069.0,
+      "step": 74
+    },
+    {
+      "entropy": 0.5660725682973862,
+      "epoch": 0.2803738317757009,
+      "grad_norm": 0.03960711508989334,
+      "learning_rate": 0.0002,
+      "loss": 0.5682789087295532,
+      "mean_token_accuracy": 0.7714986652135849,
+      "num_tokens": 1223466.0,
+      "step": 75
+    },
+    {
+      "entropy": 0.5635482668876648,
+      "epoch": 0.2841121495327103,
+      "grad_norm": 0.033172570168972015,
+      "learning_rate": 0.0002,
+      "loss": 0.5611684918403625,
+      "mean_token_accuracy": 0.7727522701025009,
+      "num_tokens": 1239902.0,
+      "step": 76
+    },
+    {
+      "entropy": 0.5699751675128937,
+      "epoch": 0.28785046728971964,
+      "grad_norm": 0.04123734310269356,
+      "learning_rate": 0.0002,
+      "loss": 0.5646215081214905,
+      "mean_token_accuracy": 0.7738035619258881,
+      "num_tokens": 1256190.0,
+      "step": 77
+    },
+    {
+      "entropy": 0.5798220336437225,
+      "epoch": 0.29158878504672897,
+      "grad_norm": 0.03779308497905731,
+      "learning_rate": 0.0002,
+      "loss": 0.5753127932548523,
+      "mean_token_accuracy": 0.769858792424202,
+      "num_tokens": 1272670.0,
+      "step": 78
+    },
+    {
+      "entropy": 0.5637505650520325,
+      "epoch": 0.2953271028037383,
+      "grad_norm": 0.036078162491321564,
+      "learning_rate": 0.0002,
+      "loss": 0.5640622973442078,
+      "mean_token_accuracy": 0.7736664712429047,
+      "num_tokens": 1288842.0,
+      "step": 79
+    },
+    {
+      "entropy": 0.5597875267267227,
+      "epoch": 0.29906542056074764,
+      "grad_norm": 0.032451264560222626,
+      "learning_rate": 0.0002,
+      "loss": 0.563526451587677,
+      "mean_token_accuracy": 0.7725878953933716,
+      "num_tokens": 1305088.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.5821113288402557,
+      "epoch": 0.30280373831775703,
+      "grad_norm": 0.038235142827034,
+      "learning_rate": 0.0002,
+      "loss": 0.5853811502456665,
+      "mean_token_accuracy": 0.7640060931444168,
+      "num_tokens": 1321504.0,
+      "step": 81
+    },
+    {
+      "entropy": 0.5438373982906342,
+      "epoch": 0.30654205607476637,
+      "grad_norm": 0.03746205195784569,
+      "learning_rate": 0.0002,
+      "loss": 0.5479013919830322,
+      "mean_token_accuracy": 0.7800798416137695,
+      "num_tokens": 1337868.0,
+      "step": 82
+    },
+    {
+      "entropy": 0.5653156489133835,
+      "epoch": 0.3102803738317757,
+      "grad_norm": 0.03490284085273743,
+      "learning_rate": 0.0002,
+      "loss": 0.5649343132972717,
+      "mean_token_accuracy": 0.7717723995447159,
+      "num_tokens": 1354004.0,
+      "step": 83
+    },
+    {
+      "entropy": 0.5832313299179077,
+      "epoch": 0.31401869158878504,
+      "grad_norm": 0.03734894096851349,
+      "learning_rate": 0.0002,
+      "loss": 0.5807749032974243,
+      "mean_token_accuracy": 0.7655426859855652,
+      "num_tokens": 1370558.0,
+      "step": 84
+    },
+    {
+      "entropy": 0.5402187407016754,
+      "epoch": 0.3177570093457944,
+      "grad_norm": 0.03854204714298248,
+      "learning_rate": 0.0002,
+      "loss": 0.5464577078819275,
+      "mean_token_accuracy": 0.7792229354381561,
+      "num_tokens": 1386854.0,
+      "step": 85
+    },
+    {
+      "entropy": 0.5570175051689148,
+      "epoch": 0.32149532710280376,
+      "grad_norm": 0.03692445531487465,
+      "learning_rate": 0.0002,
+      "loss": 0.5628695487976074,
+      "mean_token_accuracy": 0.7712178230285645,
+      "num_tokens": 1403119.0,
+      "step": 86
+    },
+    {
+      "entropy": 0.544977530837059,
+      "epoch": 0.3252336448598131,
+      "grad_norm": 0.0364147312939167,
+      "learning_rate": 0.0002,
+      "loss": 0.5444334149360657,
+      "mean_token_accuracy": 0.7836048603057861,
+      "num_tokens": 1419561.0,
+      "step": 87
+    },
+    {
+      "entropy": 0.5631776601076126,
+      "epoch": 0.32897196261682243,
+      "grad_norm": 0.03956674784421921,
+      "learning_rate": 0.0002,
+      "loss": 0.5600812435150146,
+      "mean_token_accuracy": 0.7744863480329514,
+      "num_tokens": 1435885.0,
+      "step": 88
+    },
+    {
+      "entropy": 0.5635630786418915,
+      "epoch": 0.33271028037383177,
+      "grad_norm": 0.033574752509593964,
+      "learning_rate": 0.0002,
+      "loss": 0.559590756893158,
+      "mean_token_accuracy": 0.7763379067182541,
+      "num_tokens": 1452239.0,
+      "step": 89
+    },
+    {
+      "entropy": 0.5845580548048019,
+      "epoch": 0.3364485981308411,
+      "grad_norm": 0.04166650027036667,
+      "learning_rate": 0.0002,
+      "loss": 0.5896680355072021,
+      "mean_token_accuracy": 0.7613022774457932,
+      "num_tokens": 1469012.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.5831678807735443,
+      "epoch": 0.3401869158878505,
+      "grad_norm": 0.03430558741092682,
+      "learning_rate": 0.0002,
+      "loss": 0.5816903114318848,
+      "mean_token_accuracy": 0.7649097442626953,
+      "num_tokens": 1485787.0,
+      "step": 91
+    },
+    {
+      "entropy": 0.5729665160179138,
+      "epoch": 0.34392523364485983,
+      "grad_norm": 0.035805486142635345,
+      "learning_rate": 0.0002,
+      "loss": 0.5738450884819031,
+      "mean_token_accuracy": 0.769588515162468,
+      "num_tokens": 1502079.0,
+      "step": 92
+    },
+    {
+      "entropy": 0.5653799027204514,
+      "epoch": 0.34766355140186916,
+      "grad_norm": 0.03914204239845276,
+      "learning_rate": 0.0002,
+      "loss": 0.5674072504043579,
+      "mean_token_accuracy": 0.7691812217235565,
+      "num_tokens": 1518311.0,
+      "step": 93
+    },
+    {
+      "entropy": 0.5616943389177322,
+      "epoch": 0.3514018691588785,
+      "grad_norm": 0.03653113916516304,
+      "learning_rate": 0.0002,
+      "loss": 0.5629323720932007,
+      "mean_token_accuracy": 0.7708428353071213,
+      "num_tokens": 1534540.0,
+      "step": 94
+    },
+    {
+      "entropy": 0.5549721717834473,
+      "epoch": 0.35514018691588783,
+      "grad_norm": 0.03406484052538872,
+      "learning_rate": 0.0002,
+      "loss": 0.5544697046279907,
+      "mean_token_accuracy": 0.7769979536533356,
+      "num_tokens": 1550933.0,
+      "step": 95
+    },
+    {
+      "entropy": 0.5604167878627777,
+      "epoch": 0.35887850467289717,
+      "grad_norm": 0.03333809971809387,
+      "learning_rate": 0.0002,
+      "loss": 0.5594492554664612,
+      "mean_token_accuracy": 0.7748987078666687,
+      "num_tokens": 1567207.0,
+      "step": 96
+    },
+    {
+      "entropy": 0.546283945441246,
+      "epoch": 0.36261682242990656,
+      "grad_norm": 0.03933919966220856,
+      "learning_rate": 0.0002,
+      "loss": 0.5514240264892578,
+      "mean_token_accuracy": 0.7776303142309189,
+      "num_tokens": 1583311.0,
+      "step": 97
+    },
+    {
+      "entropy": 0.5612452626228333,
+      "epoch": 0.3663551401869159,
+      "grad_norm": 0.04003525897860527,
+      "learning_rate": 0.0002,
+      "loss": 0.5653887987136841,
+      "mean_token_accuracy": 0.7751908302307129,
+      "num_tokens": 1599670.0,
+      "step": 98
+    },
+    {
+      "entropy": 0.5733049809932709,
+      "epoch": 0.37009345794392523,
+      "grad_norm": 0.038050130009651184,
+      "learning_rate": 0.0002,
+      "loss": 0.5774834752082825,
+      "mean_token_accuracy": 0.7684330493211746,
+      "num_tokens": 1616055.0,
+      "step": 99
+    },
+    {
+      "entropy": 0.5716951340436935,
+      "epoch": 0.37383177570093457,
+      "grad_norm": 0.03439132496714592,
+      "learning_rate": 0.0002,
+      "loss": 0.5752108097076416,
+      "mean_token_accuracy": 0.7674903124570847,
+      "num_tokens": 1632632.0,
+      "step": 100
+    },
+    {
+      "entropy": 0.5594859421253204,
+      "epoch": 0.3775700934579439,
+      "grad_norm": 0.03533682972192764,
+      "learning_rate": 0.0002,
+      "loss": 0.568997323513031,
+      "mean_token_accuracy": 0.7706844657659531,
+      "num_tokens": 1648909.0,
+      "step": 101
+    },
+    {
+      "entropy": 0.5699178278446198,
+      "epoch": 0.3813084112149533,
+      "grad_norm": 0.03436512500047684,
+      "learning_rate": 0.0002,
+      "loss": 0.5691055059432983,
+      "mean_token_accuracy": 0.7695006430149078,
+      "num_tokens": 1665325.0,
+      "step": 102
+    },
+    {
+      "entropy": 0.540950670838356,
+      "epoch": 0.3850467289719626,
+      "grad_norm": 0.03944160416722298,
+      "learning_rate": 0.0002,
+      "loss": 0.5404485464096069,
+      "mean_token_accuracy": 0.781610980629921,
+      "num_tokens": 1681319.0,
+      "step": 103
+    },
+    {
+      "entropy": 0.5614301711320877,
+      "epoch": 0.38878504672897196,
+      "grad_norm": 0.03688221052289009,
+      "learning_rate": 0.0002,
+      "loss": 0.5563839077949524,
+      "mean_token_accuracy": 0.7734142690896988,
+      "num_tokens": 1697654.0,
+      "step": 104
+    },
+    {
+      "entropy": 0.5428010523319244,
+      "epoch": 0.3925233644859813,
+      "grad_norm": 0.03438030928373337,
+      "learning_rate": 0.0002,
+      "loss": 0.5420697331428528,
+      "mean_token_accuracy": 0.7806366533041,
+      "num_tokens": 1713877.0,
+      "step": 105
+    },
+    {
+      "entropy": 0.5636697560548782,
+      "epoch": 0.39626168224299063,
+      "grad_norm": 0.039323464035987854,
+      "learning_rate": 0.0002,
+      "loss": 0.5671778321266174,
+      "mean_token_accuracy": 0.7705285251140594,
+      "num_tokens": 1730348.0,
+      "step": 106
+    },
+    {
+      "entropy": 0.5588130354881287,
+      "epoch": 0.4,
+      "grad_norm": 0.038590364158153534,
+      "learning_rate": 0.0002,
+      "loss": 0.5618171095848083,
+      "mean_token_accuracy": 0.772266685962677,
+      "num_tokens": 1746757.0,
+      "step": 107
+    },
+    {
+      "entropy": 0.561880424618721,
+      "epoch": 0.40373831775700936,
+      "grad_norm": 0.03582950308918953,
+      "learning_rate": 0.0002,
+      "loss": 0.5647022128105164,
+      "mean_token_accuracy": 0.771363690495491,
+      "num_tokens": 1763220.0,
+      "step": 108
+    },
+    {
+      "entropy": 0.5577187687158585,
+      "epoch": 0.4074766355140187,
+      "grad_norm": 0.0341983400285244,
+      "learning_rate": 0.0002,
+      "loss": 0.5510934591293335,
+      "mean_token_accuracy": 0.7795996069908142,
+      "num_tokens": 1779600.0,
+      "step": 109
+    },
+    {
+      "entropy": 0.5766544491052628,
+      "epoch": 0.411214953271028,
+      "grad_norm": 0.03449225798249245,
+      "learning_rate": 0.0002,
+      "loss": 0.567363977432251,
+      "mean_token_accuracy": 0.7708986699581146,
+      "num_tokens": 1795971.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.5705956965684891,
+      "epoch": 0.41495327102803736,
+      "grad_norm": 0.033156272023916245,
+      "learning_rate": 0.0002,
+      "loss": 0.5640801191329956,
+      "mean_token_accuracy": 0.7707958519458771,
+      "num_tokens": 1812546.0,
+      "step": 111
+    },
+    {
+      "entropy": 0.5591244697570801,
+      "epoch": 0.41869158878504675,
+      "grad_norm": 0.029204925522208214,
+      "learning_rate": 0.0002,
+      "loss": 0.5617936849594116,
+      "mean_token_accuracy": 0.7732467949390411,
+      "num_tokens": 1828919.0,
+      "step": 112
+    },
+    {
+      "entropy": 0.553697258234024,
+      "epoch": 0.4224299065420561,
+      "grad_norm": 0.03737286105751991,
+      "learning_rate": 0.0002,
+      "loss": 0.558631420135498,
+      "mean_token_accuracy": 0.777028888463974,
+      "num_tokens": 1845572.0,
+      "step": 113
+    },
+    {
+      "entropy": 0.5312661975622177,
+      "epoch": 0.4261682242990654,
+      "grad_norm": 0.03722711279988289,
+      "learning_rate": 0.0002,
+      "loss": 0.5394684672355652,
+      "mean_token_accuracy": 0.7818541079759598,
+      "num_tokens": 1861653.0,
+      "step": 114
+    },
+    {
+      "entropy": 0.546062633395195,
+      "epoch": 0.42990654205607476,
+      "grad_norm": 0.03108621947467327,
+      "learning_rate": 0.0002,
+      "loss": 0.5509571433067322,
+      "mean_token_accuracy": 0.7786260843276978,
+      "num_tokens": 1877945.0,
+      "step": 115
+    },
+    {
+      "entropy": 0.5667106062173843,
+      "epoch": 0.4336448598130841,
+      "grad_norm": 0.029234400019049644,
+      "learning_rate": 0.0002,
+      "loss": 0.5700245499610901,
+      "mean_token_accuracy": 0.7707543075084686,
+      "num_tokens": 1894419.0,
+      "step": 116
+    },
+    {
+      "entropy": 0.5690203160047531,
+      "epoch": 0.4373831775700935,
+      "grad_norm": 0.03232111781835556,
+      "learning_rate": 0.0002,
+      "loss": 0.5682043433189392,
+      "mean_token_accuracy": 0.7720295935869217,
+      "num_tokens": 1910917.0,
+      "step": 117
+    },
+    {
+      "entropy": 0.5568649768829346,
+      "epoch": 0.4411214953271028,
+      "grad_norm": 0.031728681176900864,
+      "learning_rate": 0.0002,
+      "loss": 0.5546939969062805,
+      "mean_token_accuracy": 0.7736632823944092,
+      "num_tokens": 1927308.0,
+      "step": 118
+    },
+    {
+      "entropy": 0.5677070319652557,
+      "epoch": 0.44485981308411215,
+      "grad_norm": 0.038115937262773514,
+      "learning_rate": 0.0002,
+      "loss": 0.569874107837677,
+      "mean_token_accuracy": 0.7713787406682968,
+      "num_tokens": 1943291.0,
+      "step": 119
+    },
+    {
+      "entropy": 0.5715564787387848,
+      "epoch": 0.4485981308411215,
+      "grad_norm": 0.031203260645270348,
+      "learning_rate": 0.0002,
+      "loss": 0.5690625309944153,
+      "mean_token_accuracy": 0.7683161348104477,
+      "num_tokens": 1959628.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.5510662794113159,
+      "epoch": 0.4523364485981308,
+      "grad_norm": 0.034137893468141556,
+      "learning_rate": 0.0002,
+      "loss": 0.552444338798523,
+      "mean_token_accuracy": 0.7768227756023407,
+      "num_tokens": 1976103.0,
+      "step": 121
+    },
+    {
+      "entropy": 0.5565275996923447,
+      "epoch": 0.45607476635514016,
+      "grad_norm": 0.039453648030757904,
+      "learning_rate": 0.0002,
+      "loss": 0.5519005656242371,
+      "mean_token_accuracy": 0.7769535332918167,
+      "num_tokens": 1992472.0,
+      "step": 122
+    },
+    {
+      "entropy": 0.5421037673950195,
+      "epoch": 0.45981308411214955,
+      "grad_norm": 0.03410222753882408,
+      "learning_rate": 0.0002,
+      "loss": 0.5471655130386353,
+      "mean_token_accuracy": 0.7754515707492828,
+      "num_tokens": 2008555.0,
+      "step": 123
+    },
+    {
+      "entropy": 0.5602420270442963,
+      "epoch": 0.4635514018691589,
+      "grad_norm": 0.034221749752759933,
+      "learning_rate": 0.0002,
+      "loss": 0.561537504196167,
+      "mean_token_accuracy": 0.7700601816177368,
+      "num_tokens": 2025061.0,
+      "step": 124
+    },
+    {
+      "entropy": 0.5571303218603134,
+      "epoch": 0.4672897196261682,
+      "grad_norm": 0.036211200058460236,
+      "learning_rate": 0.0002,
+      "loss": 0.5621545910835266,
+      "mean_token_accuracy": 0.7712312340736389,
+      "num_tokens": 2041431.0,
+      "step": 125
+    },
+    {
+      "entropy": 0.5687634646892548,
+      "epoch": 0.47102803738317756,
+      "grad_norm": 0.03822026401758194,
+      "learning_rate": 0.0002,
+      "loss": 0.5733909010887146,
+      "mean_token_accuracy": 0.7690061926841736,
+      "num_tokens": 2057835.0,
+      "step": 126
+    },
+    {
+      "entropy": 0.561365082859993,
+      "epoch": 0.4747663551401869,
+      "grad_norm": 0.03744380176067352,
+      "learning_rate": 0.0002,
+      "loss": 0.561904788017273,
+      "mean_token_accuracy": 0.7733899652957916,
+      "num_tokens": 2073971.0,
+      "step": 127
+    },
+    {
+      "entropy": 0.5848346501588821,
+      "epoch": 0.4785046728971963,
+      "grad_norm": 0.04521637409925461,
+      "learning_rate": 0.0002,
+      "loss": 0.5786125063896179,
+      "mean_token_accuracy": 0.7650021761655807,
+      "num_tokens": 2090449.0,
+      "step": 128
+    },
+    {
+      "entropy": 0.547502264380455,
+      "epoch": 0.4822429906542056,
+      "grad_norm": 0.030477922409772873,
+      "learning_rate": 0.0002,
+      "loss": 0.5489711761474609,
+      "mean_token_accuracy": 0.7773880660533905,
+      "num_tokens": 2106721.0,
+      "step": 129
+    },
+    {
+      "entropy": 0.5484167486429214,
+      "epoch": 0.48598130841121495,
+      "grad_norm": 0.031105147674679756,
+      "learning_rate": 0.0002,
+      "loss": 0.5544927716255188,
+      "mean_token_accuracy": 0.7739315629005432,
+      "num_tokens": 2123019.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.5533222556114197,
+      "epoch": 0.4897196261682243,
+      "grad_norm": 0.0377504825592041,
+      "learning_rate": 0.0002,
+      "loss": 0.5580374002456665,
+      "mean_token_accuracy": 0.7739338725805283,
+      "num_tokens": 2139430.0,
+      "step": 131
+    },
+    {
+      "entropy": 0.5609411746263504,
+      "epoch": 0.4934579439252336,
+      "grad_norm": 0.04023468494415283,
+      "learning_rate": 0.0002,
+      "loss": 0.566258430480957,
+      "mean_token_accuracy": 0.7712711840867996,
+      "num_tokens": 2155890.0,
+      "step": 132
+    },
+    {
+      "entropy": 0.5359190106391907,
+      "epoch": 0.497196261682243,
+      "grad_norm": 0.03230973705649376,
+      "learning_rate": 0.0002,
+      "loss": 0.5397756695747375,
+      "mean_token_accuracy": 0.7778284996747971,
+      "num_tokens": 2172146.0,
+      "step": 133
+    },
+    {
+      "entropy": 0.5826548039913177,
+      "epoch": 0.5009345794392523,
+      "grad_norm": 0.03292697295546532,
+      "learning_rate": 0.0002,
+      "loss": 0.5766130685806274,
+      "mean_token_accuracy": 0.7655939012765884,
+      "num_tokens": 2188654.0,
+      "step": 134
+    },
+    {
+      "entropy": 0.5718280375003815,
+      "epoch": 0.5046728971962616,
+      "grad_norm": 0.03372567892074585,
+      "learning_rate": 0.0002,
+      "loss": 0.5630422830581665,
+      "mean_token_accuracy": 0.7710609585046768,
+      "num_tokens": 2205019.0,
+      "step": 135
+    },
+    {
+      "entropy": 0.5586309432983398,
+      "epoch": 0.508411214953271,
+      "grad_norm": 0.033095572143793106,
+      "learning_rate": 0.0002,
+      "loss": 0.5580739974975586,
+      "mean_token_accuracy": 0.7730830013751984,
+      "num_tokens": 2221412.0,
+      "step": 136
+    },
+    {
+      "entropy": 0.555738240480423,
+      "epoch": 0.5121495327102804,
+      "grad_norm": 0.03156888484954834,
+      "learning_rate": 0.0002,
+      "loss": 0.5593985319137573,
+      "mean_token_accuracy": 0.7678060680627823,
+      "num_tokens": 2237712.0,
+      "step": 137
+    },
+    {
+      "entropy": 0.5371007025241852,
+      "epoch": 0.5158878504672897,
+      "grad_norm": 0.03322545811533928,
+      "learning_rate": 0.0002,
+      "loss": 0.5458611845970154,
+      "mean_token_accuracy": 0.7752078175544739,
+      "num_tokens": 2253827.0,
+      "step": 138
+    },
+    {
+      "entropy": 0.5338983088731766,
+      "epoch": 0.5196261682242991,
+      "grad_norm": 0.03455954045057297,
+      "learning_rate": 0.0002,
+      "loss": 0.545717179775238,
+      "mean_token_accuracy": 0.7804258912801743,
+      "num_tokens": 2270009.0,
+      "step": 139
+    },
+    {
+      "entropy": 0.5622379928827286,
+      "epoch": 0.5233644859813084,
+      "grad_norm": 0.030741458758711815,
+      "learning_rate": 0.0002,
+      "loss": 0.5627423524856567,
+      "mean_token_accuracy": 0.771199107170105,
+      "num_tokens": 2286480.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.5630074888467789,
+      "epoch": 0.5271028037383177,
+      "grad_norm": 0.03147517144680023,
+      "learning_rate": 0.0002,
+      "loss": 0.5579048991203308,
+      "mean_token_accuracy": 0.7733362317085266,
+      "num_tokens": 2302838.0,
+      "step": 141
+    },
+    {
+      "entropy": 0.5669132769107819,
+      "epoch": 0.5308411214953271,
+      "grad_norm": 0.03270615637302399,
+      "learning_rate": 0.0002,
+      "loss": 0.5602433085441589,
+      "mean_token_accuracy": 0.7726781070232391,
+      "num_tokens": 2318805.0,
+      "step": 142
+    },
+    {
+      "entropy": 0.5551939755678177,
+      "epoch": 0.5345794392523364,
+      "grad_norm": 0.03414132818579674,
+      "learning_rate": 0.0002,
+      "loss": 0.5483931303024292,
+      "mean_token_accuracy": 0.7783452570438385,
+      "num_tokens": 2335157.0,
+      "step": 143
+    },
+    {
+      "entropy": 0.5692463368177414,
+      "epoch": 0.5383177570093458,
+      "grad_norm": 0.037064556032419205,
+      "learning_rate": 0.0002,
+      "loss": 0.5773426294326782,
+      "mean_token_accuracy": 0.7641823440790176,
+      "num_tokens": 2351384.0,
+      "step": 144
+    },
+    {
+      "entropy": 0.5659217685461044,
+      "epoch": 0.5420560747663551,
+      "grad_norm": 0.037828508764505386,
+      "learning_rate": 0.0002,
+      "loss": 0.5750471353530884,
+      "mean_token_accuracy": 0.765160858631134,
+      "num_tokens": 2367801.0,
+      "step": 145
+    },
+    {
+      "entropy": 0.561379611492157,
+      "epoch": 0.5457943925233645,
+      "grad_norm": 0.03333268687129021,
+      "learning_rate": 0.0002,
+      "loss": 0.5598861575126648,
+      "mean_token_accuracy": 0.7742918580770493,
+      "num_tokens": 2384482.0,
+      "step": 146
+    },
+    {
+      "entropy": 0.5676955878734589,
+      "epoch": 0.5495327102803739,
+      "grad_norm": 0.0296424962580204,
+      "learning_rate": 0.0002,
+      "loss": 0.560563325881958,
+      "mean_token_accuracy": 0.774119034409523,
+      "num_tokens": 2400736.0,
+      "step": 147
+    },
+    {
+      "entropy": 0.5658537149429321,
+      "epoch": 0.5532710280373832,
+      "grad_norm": 0.0320475734770298,
+      "learning_rate": 0.0002,
+      "loss": 0.5580253601074219,
+      "mean_token_accuracy": 0.7784629613161087,
+      "num_tokens": 2417222.0,
+      "step": 148
+    },
+    {
+      "entropy": 0.5581035912036896,
+      "epoch": 0.5570093457943925,
+      "grad_norm": 0.03125445544719696,
+      "learning_rate": 0.0002,
+      "loss": 0.5552960634231567,
+      "mean_token_accuracy": 0.7758258730173111,
+      "num_tokens": 2433798.0,
+      "step": 149
+    },
+    {
+      "entropy": 0.5418982058763504,
+      "epoch": 0.5607476635514018,
+      "grad_norm": 0.0312193613499403,
+      "learning_rate": 0.0002,
+      "loss": 0.5517758727073669,
+      "mean_token_accuracy": 0.775468572974205,
+      "num_tokens": 2450186.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.5608511418104172,
+      "epoch": 0.5644859813084112,
+      "grad_norm": 0.032521311193704605,
+      "learning_rate": 0.0002,
+      "loss": 0.574313759803772,
+      "mean_token_accuracy": 0.7643774896860123,
+      "num_tokens": 2466450.0,
+      "step": 151
+    },
+    {
+      "entropy": 0.5575774312019348,
+      "epoch": 0.5682242990654206,
+      "grad_norm": 0.02971738576889038,
+      "learning_rate": 0.0002,
+      "loss": 0.5584424138069153,
+      "mean_token_accuracy": 0.7738354802131653,
+      "num_tokens": 2482922.0,
+      "step": 152
+    },
+    {
+      "entropy": 0.5496693998575211,
+      "epoch": 0.5719626168224299,
+      "grad_norm": 0.0331539623439312,
+      "learning_rate": 0.0002,
+      "loss": 0.5571150183677673,
+      "mean_token_accuracy": 0.7751006782054901,
+      "num_tokens": 2499215.0,
+      "step": 153
+    },
+    {
+      "entropy": 0.5449398458003998,
+      "epoch": 0.5757009345794393,
+      "grad_norm": 0.028877289965748787,
+      "learning_rate": 0.0002,
+      "loss": 0.5457977652549744,
+      "mean_token_accuracy": 0.7815146297216415,
+      "num_tokens": 2515544.0,
+      "step": 154
+    },
+    {
+      "entropy": 0.5479637384414673,
+      "epoch": 0.5794392523364486,
+      "grad_norm": 0.03473028913140297,
+      "learning_rate": 0.0002,
+      "loss": 0.5433490872383118,
+      "mean_token_accuracy": 0.7794239521026611,
+      "num_tokens": 2531678.0,
+      "step": 155
+    },
+    {
+      "entropy": 0.5400601774454117,
+      "epoch": 0.5831775700934579,
+      "grad_norm": 0.030446792021393776,
+      "learning_rate": 0.0002,
+      "loss": 0.5424179434776306,
+      "mean_token_accuracy": 0.7802795916795731,
+      "num_tokens": 2547759.0,
+      "step": 156
+    },
+    {
+      "entropy": 0.5537837445735931,
+      "epoch": 0.5869158878504673,
+      "grad_norm": 0.030434170737862587,
+      "learning_rate": 0.0002,
+      "loss": 0.5575878024101257,
+      "mean_token_accuracy": 0.7747874706983566,
+      "num_tokens": 2564234.0,
+      "step": 157
+    },
+    {
+      "entropy": 0.5445020198822021,
+      "epoch": 0.5906542056074766,
+      "grad_norm": 0.031060006469488144,
+      "learning_rate": 0.0002,
+      "loss": 0.5391767024993896,
+      "mean_token_accuracy": 0.7812654078006744,
+      "num_tokens": 2580415.0,
+      "step": 158
+    },
+    {
+      "entropy": 0.5522216707468033,
+      "epoch": 0.594392523364486,
+      "grad_norm": 0.03142917901277542,
+      "learning_rate": 0.0002,
+      "loss": 0.5536507964134216,
+      "mean_token_accuracy": 0.7752956748008728,
+      "num_tokens": 2596612.0,
+      "step": 159
+    },
+    {
+      "entropy": 0.5727219432592392,
+      "epoch": 0.5981308411214953,
+      "grad_norm": 0.033186040818691254,
+      "learning_rate": 0.0002,
+      "loss": 0.5693776607513428,
+      "mean_token_accuracy": 0.7679739594459534,
+      "num_tokens": 2613027.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.5479543507099152,
+      "epoch": 0.6018691588785047,
+      "grad_norm": 0.029086332768201828,
+      "learning_rate": 0.0002,
+      "loss": 0.5436608195304871,
+      "mean_token_accuracy": 0.7767124027013779,
+      "num_tokens": 2629513.0,
+      "step": 161
+    },
+    {
+      "entropy": 0.5365437120199203,
+      "epoch": 0.6056074766355141,
+      "grad_norm": 0.03190625086426735,
+      "learning_rate": 0.0002,
+      "loss": 0.539345383644104,
+      "mean_token_accuracy": 0.7788415402173996,
+      "num_tokens": 2645517.0,
+      "step": 162
+    },
+    {
+      "entropy": 0.5434722602367401,
+      "epoch": 0.6093457943925233,
+      "grad_norm": 0.03164217248558998,
+      "learning_rate": 0.0002,
+      "loss": 0.5468231439590454,
+      "mean_token_accuracy": 0.7775093168020248,
+      "num_tokens": 2661809.0,
+      "step": 163
+    },
+    {
+      "entropy": 0.5627670586109161,
+      "epoch": 0.6130841121495327,
+      "grad_norm": 0.030292298644781113,
+      "learning_rate": 0.0002,
+      "loss": 0.5634962916374207,
+      "mean_token_accuracy": 0.77353835105896,
+      "num_tokens": 2678296.0,
+      "step": 164
+    },
+    {
+      "entropy": 0.5599590986967087,
+      "epoch": 0.616822429906542,
+      "grad_norm": 0.029385516420006752,
+      "learning_rate": 0.0002,
+      "loss": 0.5621803402900696,
+      "mean_token_accuracy": 0.7726689726114273,
+      "num_tokens": 2694312.0,
+      "step": 165
+    },
+    {
+      "entropy": 0.5674590021371841,
+      "epoch": 0.6205607476635514,
+      "grad_norm": 0.03131924942135811,
+      "learning_rate": 0.0002,
+      "loss": 0.5683683753013611,
+      "mean_token_accuracy": 0.7695939242839813,
+      "num_tokens": 2710615.0,
+      "step": 166
+    },
+    {
+      "entropy": 0.5440984219312668,
+      "epoch": 0.6242990654205608,
+      "grad_norm": 0.030983613803982735,
+      "learning_rate": 0.0002,
+      "loss": 0.5437349081039429,
+      "mean_token_accuracy": 0.7795680165290833,
+      "num_tokens": 2726995.0,
+      "step": 167
+    },
+    {
+      "entropy": 0.5603994876146317,
+      "epoch": 0.6280373831775701,
+      "grad_norm": 0.028368143364787102,
+      "learning_rate": 0.0002,
+      "loss": 0.5631406903266907,
+      "mean_token_accuracy": 0.7677181214094162,
+      "num_tokens": 2743576.0,
+      "step": 168
+    },
+    {
+      "entropy": 0.5566483587026596,
+      "epoch": 0.6317757009345795,
+      "grad_norm": 0.03626226633787155,
+      "learning_rate": 0.0002,
+      "loss": 0.5647379159927368,
+      "mean_token_accuracy": 0.7732031494379044,
+      "num_tokens": 2760174.0,
+      "step": 169
+    },
+    {
+      "entropy": 0.564967080950737,
+      "epoch": 0.6355140186915887,
+      "grad_norm": 0.03088950738310814,
+      "learning_rate": 0.0002,
+      "loss": 0.5664903521537781,
+      "mean_token_accuracy": 0.7681966125965118,
+      "num_tokens": 2776574.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.5679657459259033,
+      "epoch": 0.6392523364485981,
+      "grad_norm": 0.03709281235933304,
+      "learning_rate": 0.0002,
+      "loss": 0.5604699850082397,
+      "mean_token_accuracy": 0.7735682725906372,
+      "num_tokens": 2792828.0,
+      "step": 171
+    },
+    {
+      "entropy": 0.581437885761261,
+      "epoch": 0.6429906542056075,
+      "grad_norm": 0.026090197265148163,
+      "learning_rate": 0.0002,
+      "loss": 0.5791311264038086,
+      "mean_token_accuracy": 0.762902557849884,
+      "num_tokens": 2809131.0,
+      "step": 172
+    },
+    {
+      "entropy": 0.5521832853555679,
+      "epoch": 0.6467289719626168,
+      "grad_norm": 0.029906727373600006,
+      "learning_rate": 0.0002,
+      "loss": 0.5577026605606079,
+      "mean_token_accuracy": 0.7733243852853775,
+      "num_tokens": 2825112.0,
+      "step": 173
+    },
+    {
+      "entropy": 0.555216521024704,
+      "epoch": 0.6504672897196262,
+      "grad_norm": 0.033849943429231644,
+      "learning_rate": 0.0002,
+      "loss": 0.5657702684402466,
+      "mean_token_accuracy": 0.7707085460424423,
+      "num_tokens": 2841339.0,
+      "step": 174
+    },
+    {
+      "entropy": 0.551365852355957,
+      "epoch": 0.6542056074766355,
+      "grad_norm": 0.031996551901102066,
+      "learning_rate": 0.0002,
+      "loss": 0.5525310039520264,
+      "mean_token_accuracy": 0.7775240540504456,
+      "num_tokens": 2857512.0,
+      "step": 175
+    },
+    {
+      "entropy": 0.549150824546814,
+      "epoch": 0.6579439252336449,
+      "grad_norm": 0.029290124773979187,
+      "learning_rate": 0.0002,
+      "loss": 0.5493086576461792,
+      "mean_token_accuracy": 0.7772240936756134,
+      "num_tokens": 2873797.0,
+      "step": 176
+    },
+    {
+      "entropy": 0.5804181694984436,
+      "epoch": 0.6616822429906543,
+      "grad_norm": 0.02999035269021988,
+      "learning_rate": 0.0002,
+      "loss": 0.5758557319641113,
+      "mean_token_accuracy": 0.766997441649437,
+      "num_tokens": 2890362.0,
+      "step": 177
+    },
+    {
+      "entropy": 0.5632698237895966,
+      "epoch": 0.6654205607476635,
+      "grad_norm": 0.02878536470234394,
+      "learning_rate": 0.0002,
+      "loss": 0.559673011302948,
+      "mean_token_accuracy": 0.7719457000494003,
+      "num_tokens": 2906508.0,
+      "step": 178
+    },
+    {
+      "entropy": 0.5700209885835648,
+      "epoch": 0.6691588785046729,
+      "grad_norm": 0.028609320521354675,
+      "learning_rate": 0.0002,
+      "loss": 0.5671200752258301,
+      "mean_token_accuracy": 0.7697011083364487,
+      "num_tokens": 2922954.0,
+      "step": 179
+    },
+    {
+      "entropy": 0.5542242079973221,
+      "epoch": 0.6728971962616822,
+      "grad_norm": 0.028964772820472717,
+      "learning_rate": 0.0002,
+      "loss": 0.5530304908752441,
+      "mean_token_accuracy": 0.776089608669281,
+      "num_tokens": 2939138.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.570984736084938,
+      "epoch": 0.6766355140186916,
+      "grad_norm": 0.030573254451155663,
+      "learning_rate": 0.0002,
+      "loss": 0.5777382850646973,
+      "mean_token_accuracy": 0.7673922181129456,
+      "num_tokens": 2955869.0,
+      "step": 181
+    },
+    {
+      "entropy": 0.5505239069461823,
+      "epoch": 0.680373831775701,
+      "grad_norm": 0.02849324606359005,
+      "learning_rate": 0.0002,
+      "loss": 0.5563762784004211,
+      "mean_token_accuracy": 0.7741904556751251,
+      "num_tokens": 2972440.0,
+      "step": 182
+    },
+    {
+      "entropy": 0.5634328275918961,
+      "epoch": 0.6841121495327103,
+      "grad_norm": 0.029679251834750175,
+      "learning_rate": 0.0002,
+      "loss": 0.5653568506240845,
+      "mean_token_accuracy": 0.7715763002634048,
+      "num_tokens": 2988878.0,
+      "step": 183
+    },
+    {
+      "entropy": 0.5458139479160309,
+      "epoch": 0.6878504672897197,
+      "grad_norm": 0.028634849935770035,
+      "learning_rate": 0.0002,
+      "loss": 0.5491960048675537,
+      "mean_token_accuracy": 0.7766706198453903,
+      "num_tokens": 3004892.0,
+      "step": 184
+    },
+    {
+      "entropy": 0.566186711192131,
+      "epoch": 0.6915887850467289,
+      "grad_norm": 0.03155568614602089,
+      "learning_rate": 0.0002,
+      "loss": 0.5609148740768433,
+      "mean_token_accuracy": 0.77189701795578,
+      "num_tokens": 3021081.0,
+      "step": 185
+    },
+    {
+      "entropy": 0.5414428561925888,
+      "epoch": 0.6953271028037383,
+      "grad_norm": 0.027838880196213722,
+      "learning_rate": 0.0002,
+      "loss": 0.5334644317626953,
+      "mean_token_accuracy": 0.7823236584663391,
+      "num_tokens": 3037358.0,
+      "step": 186
+    },
+    {
+      "entropy": 0.5581627041101456,
+      "epoch": 0.6990654205607477,
+      "grad_norm": 0.028703616932034492,
+      "learning_rate": 0.0002,
+      "loss": 0.5619620680809021,
+      "mean_token_accuracy": 0.7744630128145218,
+      "num_tokens": 3053550.0,
+      "step": 187
+    },
+    {
+      "entropy": 0.5253149345517159,
+      "epoch": 0.702803738317757,
+      "grad_norm": 0.036868445575237274,
+      "learning_rate": 0.0002,
+      "loss": 0.5375533103942871,
+      "mean_token_accuracy": 0.7792882919311523,
+      "num_tokens": 3069739.0,
+      "step": 188
+    },
+    {
+      "entropy": 0.5389144420623779,
+      "epoch": 0.7065420560747664,
+      "grad_norm": 0.03131941705942154,
+      "learning_rate": 0.0002,
+      "loss": 0.5459954738616943,
+      "mean_token_accuracy": 0.7803291976451874,
+      "num_tokens": 3085960.0,
+      "step": 189
+    },
+    {
+      "entropy": 0.5600542724132538,
+      "epoch": 0.7102803738317757,
+      "grad_norm": 0.029567614197731018,
+      "learning_rate": 0.0002,
+      "loss": 0.5498960614204407,
+      "mean_token_accuracy": 0.7771395742893219,
+      "num_tokens": 3102054.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.5614524185657501,
+      "epoch": 0.7140186915887851,
+      "grad_norm": 0.02720000222325325,
+      "learning_rate": 0.0002,
+      "loss": 0.5563923716545105,
+      "mean_token_accuracy": 0.7744134813547134,
+      "num_tokens": 3118447.0,
+      "step": 191
+    },
+    {
+      "entropy": 0.5513639450073242,
+      "epoch": 0.7177570093457943,
+      "grad_norm": 0.03140437230467796,
+      "learning_rate": 0.0002,
+      "loss": 0.5556260347366333,
+      "mean_token_accuracy": 0.7725710570812225,
+      "num_tokens": 3134824.0,
+      "step": 192
+    },
+    {
+      "entropy": 0.5523516684770584,
+      "epoch": 0.7214953271028037,
+      "grad_norm": 0.027633987367153168,
+      "learning_rate": 0.0002,
+      "loss": 0.5527896285057068,
+      "mean_token_accuracy": 0.775657445192337,
+      "num_tokens": 3151098.0,
+      "step": 193
+    },
+    {
+      "entropy": 0.5488898605108261,
+      "epoch": 0.7252336448598131,
+      "grad_norm": 0.02825162373483181,
+      "learning_rate": 0.0002,
+      "loss": 0.5509235858917236,
+      "mean_token_accuracy": 0.7764269262552261,
+      "num_tokens": 3167503.0,
+      "step": 194
+    },
+    {
+      "entropy": 0.5719494670629501,
+      "epoch": 0.7289719626168224,
+      "grad_norm": 0.03426945582032204,
+      "learning_rate": 0.0002,
+      "loss": 0.581325888633728,
+      "mean_token_accuracy": 0.7639684081077576,
+      "num_tokens": 3183932.0,
+      "step": 195
+    },
+    {
+      "entropy": 0.5686557441949844,
+      "epoch": 0.7327102803738318,
+      "grad_norm": 0.0296348724514246,
+      "learning_rate": 0.0002,
+      "loss": 0.5693853497505188,
+      "mean_token_accuracy": 0.7690740376710892,
+      "num_tokens": 3200339.0,
+      "step": 196
+    },
+    {
+      "entropy": 0.545837864279747,
+      "epoch": 0.7364485981308411,
+      "grad_norm": 0.03093736432492733,
+      "learning_rate": 0.0002,
+      "loss": 0.5460183024406433,
+      "mean_token_accuracy": 0.7779112309217453,
+      "num_tokens": 3216460.0,
+      "step": 197
+    },
+    {
+      "entropy": 0.5601817220449448,
+      "epoch": 0.7401869158878505,
+      "grad_norm": 0.02910701371729374,
+      "learning_rate": 0.0002,
+      "loss": 0.5535531044006348,
+      "mean_token_accuracy": 0.7790419608354568,
+      "num_tokens": 3232857.0,
+      "step": 198
+    },
+    {
+      "entropy": 0.5590776205062866,
+      "epoch": 0.7439252336448599,
+      "grad_norm": 0.02963181957602501,
+      "learning_rate": 0.0002,
+      "loss": 0.5586906671524048,
+      "mean_token_accuracy": 0.771443635225296,
+      "num_tokens": 3249103.0,
+      "step": 199
+    },
+    {
+      "entropy": 0.5511395633220673,
+      "epoch": 0.7476635514018691,
+      "grad_norm": 0.031875208020210266,
+      "learning_rate": 0.0002,
+      "loss": 0.5533637404441833,
+      "mean_token_accuracy": 0.7755965292453766,
+      "num_tokens": 3265378.0,
+      "step": 200
+    },
+    {
+      "entropy": 0.5485802739858627,
+      "epoch": 0.7514018691588785,
+      "grad_norm": 0.027761714532971382,
+      "learning_rate": 0.0002,
+      "loss": 0.5528791546821594,
+      "mean_token_accuracy": 0.7763257622718811,
+      "num_tokens": 3281575.0,
+      "step": 201
+    },
+    {
+      "entropy": 0.5456139892339706,
+      "epoch": 0.7551401869158878,
+      "grad_norm": 0.030668726190924644,
+      "learning_rate": 0.0002,
+      "loss": 0.5525423884391785,
+      "mean_token_accuracy": 0.7762203961610794,
+      "num_tokens": 3297643.0,
+      "step": 202
+    },
+    {
+      "entropy": 0.5445601046085358,
+      "epoch": 0.7588785046728972,
+      "grad_norm": 0.029748164117336273,
+      "learning_rate": 0.0002,
+      "loss": 0.5478560924530029,
+      "mean_token_accuracy": 0.7770299166440964,
+      "num_tokens": 3314080.0,
+      "step": 203
+    },
+    {
+      "entropy": 0.5609561204910278,
+      "epoch": 0.7626168224299066,
+      "grad_norm": 0.03164554014801979,
+      "learning_rate": 0.0002,
+      "loss": 0.5568655729293823,
+      "mean_token_accuracy": 0.7736087292432785,
+      "num_tokens": 3330614.0,
+      "step": 204
+    },
+    {
+      "entropy": 0.5710489153862,
+      "epoch": 0.7663551401869159,
+      "grad_norm": 0.028819117695093155,
+      "learning_rate": 0.0002,
+      "loss": 0.5648672580718994,
+      "mean_token_accuracy": 0.7728124856948853,
+      "num_tokens": 3347098.0,
+      "step": 205
+    },
+    {
+      "entropy": 0.5394908934831619,
+      "epoch": 0.7700934579439253,
+      "grad_norm": 0.027154654264450073,
+      "learning_rate": 0.0002,
+      "loss": 0.5411190986633301,
+      "mean_token_accuracy": 0.7803514152765274,
+      "num_tokens": 3363492.0,
+      "step": 206
+    },
+    {
+      "entropy": 0.5685155987739563,
+      "epoch": 0.7738317757009345,
+      "grad_norm": 0.029519978910684586,
+      "learning_rate": 0.0002,
+      "loss": 0.5696613192558289,
+      "mean_token_accuracy": 0.7668858170509338,
+      "num_tokens": 3380119.0,
+      "step": 207
+    },
+    {
+      "entropy": 0.5677367150783539,
+      "epoch": 0.7775700934579439,
+      "grad_norm": 0.030474061146378517,
+      "learning_rate": 0.0002,
+      "loss": 0.5646864771842957,
+      "mean_token_accuracy": 0.770299568772316,
+      "num_tokens": 3396414.0,
+      "step": 208
+    },
+    {
+      "entropy": 0.550197958946228,
+      "epoch": 0.7813084112149533,
+      "grad_norm": 0.03337560594081879,
+      "learning_rate": 0.0002,
+      "loss": 0.5529907941818237,
+      "mean_token_accuracy": 0.7742650210857391,
+      "num_tokens": 3412594.0,
+      "step": 209
+    },
+    {
+      "entropy": 0.5500561147928238,
+      "epoch": 0.7850467289719626,
+      "grad_norm": 0.029891593381762505,
+      "learning_rate": 0.0002,
+      "loss": 0.5457524657249451,
+      "mean_token_accuracy": 0.7752271890640259,
+      "num_tokens": 3428832.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.5407024472951889,
+      "epoch": 0.788785046728972,
+      "grad_norm": 0.03455657884478569,
+      "learning_rate": 0.0002,
+      "loss": 0.5474290251731873,
+      "mean_token_accuracy": 0.7768412232398987,
+      "num_tokens": 3444963.0,
+      "step": 211
+    },
+    {
+      "entropy": 0.5501608401536942,
+      "epoch": 0.7925233644859813,
+      "grad_norm": 0.03011375665664673,
+      "learning_rate": 0.0002,
+      "loss": 0.5565037131309509,
+      "mean_token_accuracy": 0.775830090045929,
+      "num_tokens": 3461474.0,
+      "step": 212
+    },
+    {
+      "entropy": 0.5669280886650085,
+      "epoch": 0.7962616822429907,
+      "grad_norm": 0.026809366419911385,
+      "learning_rate": 0.0002,
+      "loss": 0.5677840113639832,
+      "mean_token_accuracy": 0.7668448388576508,
+      "num_tokens": 3477897.0,
+      "step": 213
+    },
+    {
+      "entropy": 0.5445934683084488,
+      "epoch": 0.8,
+      "grad_norm": 0.029105374589562416,
+      "learning_rate": 0.0002,
+      "loss": 0.5453011989593506,
+      "mean_token_accuracy": 0.7774604558944702,
+      "num_tokens": 3494383.0,
+      "step": 214
+    },
+    {
+      "entropy": 0.5411728769540787,
+      "epoch": 0.8037383177570093,
+      "grad_norm": 0.030234847217798233,
+      "learning_rate": 0.0002,
+      "loss": 0.5382598042488098,
+      "mean_token_accuracy": 0.7800557911396027,
+      "num_tokens": 3510715.0,
+      "step": 215
+    },
+    {
+      "entropy": 0.5646664798259735,
+      "epoch": 0.8074766355140187,
+      "grad_norm": 0.026832353323698044,
+      "learning_rate": 0.0002,
+      "loss": 0.5600794553756714,
+      "mean_token_accuracy": 0.7739932984113693,
+      "num_tokens": 3527178.0,
+      "step": 216
+    },
+    {
+      "entropy": 0.5487450510263443,
+      "epoch": 0.811214953271028,
+      "grad_norm": 0.03036404214799404,
+      "learning_rate": 0.0002,
+      "loss": 0.5507562160491943,
+      "mean_token_accuracy": 0.7767736315727234,
+      "num_tokens": 3543310.0,
+      "step": 217
+    },
+    {
+      "entropy": 0.5374595075845718,
+      "epoch": 0.8149532710280374,
+      "grad_norm": 0.02880716696381569,
+      "learning_rate": 0.0002,
+      "loss": 0.5401508808135986,
+      "mean_token_accuracy": 0.7788828462362289,
+      "num_tokens": 3559560.0,
+      "step": 218
+    },
+    {
+      "entropy": 0.5644234865903854,
+      "epoch": 0.8186915887850468,
+      "grad_norm": 0.028643809258937836,
+      "learning_rate": 0.0002,
+      "loss": 0.5721843242645264,
+      "mean_token_accuracy": 0.766772523522377,
+      "num_tokens": 3575647.0,
+      "step": 219
+    },
+    {
+      "entropy": 0.5402477234601974,
+      "epoch": 0.822429906542056,
+      "grad_norm": 0.029598018154501915,
+      "learning_rate": 0.0002,
+      "loss": 0.544965922832489,
+      "mean_token_accuracy": 0.7768107801675797,
+      "num_tokens": 3591846.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.5643706917762756,
+      "epoch": 0.8261682242990654,
+      "grad_norm": 0.026736166328191757,
+      "learning_rate": 0.0002,
+      "loss": 0.5665444731712341,
+      "mean_token_accuracy": 0.7706874907016754,
+      "num_tokens": 3608162.0,
+      "step": 221
+    },
+    {
+      "entropy": 0.5756243169307709,
+      "epoch": 0.8299065420560747,
+      "grad_norm": 0.03206360712647438,
+      "learning_rate": 0.0002,
+      "loss": 0.5674804449081421,
+      "mean_token_accuracy": 0.7733075469732285,
+      "num_tokens": 3624379.0,
+      "step": 222
+    },
+    {
+      "entropy": 0.5734997987747192,
+      "epoch": 0.8336448598130841,
+      "grad_norm": 0.03715579956769943,
+      "learning_rate": 0.0002,
+      "loss": 0.5701056718826294,
+      "mean_token_accuracy": 0.768852710723877,
+      "num_tokens": 3640765.0,
+      "step": 223
+    },
+    {
+      "entropy": 0.5725346952676773,
+      "epoch": 0.8373831775700935,
+      "grad_norm": 0.030887214466929436,
+      "learning_rate": 0.0002,
+      "loss": 0.573249340057373,
+      "mean_token_accuracy": 0.767242893576622,
+      "num_tokens": 3657061.0,
+      "step": 224
+    },
+    {
+      "entropy": 0.5410794317722321,
+      "epoch": 0.8411214953271028,
+      "grad_norm": 0.035770442336797714,
+      "learning_rate": 0.0002,
+      "loss": 0.5480395555496216,
+      "mean_token_accuracy": 0.7784460484981537,
+      "num_tokens": 3673262.0,
+      "step": 225
+    },
+    {
+      "entropy": 0.5552464425563812,
+      "epoch": 0.8448598130841122,
+      "grad_norm": 0.02601957693696022,
+      "learning_rate": 0.0002,
+      "loss": 0.5526980757713318,
+      "mean_token_accuracy": 0.7752740979194641,
+      "num_tokens": 3689556.0,
+      "step": 226
+    },
+    {
+      "entropy": 0.5603229254484177,
+      "epoch": 0.8485981308411215,
+      "grad_norm": 0.02927876077592373,
+      "learning_rate": 0.0002,
+      "loss": 0.5548710823059082,
+      "mean_token_accuracy": 0.7742596417665482,
+      "num_tokens": 3705909.0,
+      "step": 227
+    },
+    {
+      "entropy": 0.5540427714586258,
+      "epoch": 0.8523364485981308,
+      "grad_norm": 0.027498876675963402,
+      "learning_rate": 0.0002,
+      "loss": 0.551406741142273,
+      "mean_token_accuracy": 0.7762021422386169,
+      "num_tokens": 3722220.0,
+      "step": 228
+    },
+    {
+      "entropy": 0.547235295176506,
+      "epoch": 0.8560747663551402,
+      "grad_norm": 0.03035261482000351,
+      "learning_rate": 0.0002,
+      "loss": 0.5486667156219482,
+      "mean_token_accuracy": 0.7744798958301544,
+      "num_tokens": 3738434.0,
+      "step": 229
+    },
+    {
+      "entropy": 0.5418660938739777,
+      "epoch": 0.8598130841121495,
+      "grad_norm": 0.027413956820964813,
+      "learning_rate": 0.0002,
+      "loss": 0.5479095578193665,
+      "mean_token_accuracy": 0.7781544178724289,
+      "num_tokens": 3755054.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.5522046238183975,
+      "epoch": 0.8635514018691589,
+      "grad_norm": 0.030631685629487038,
+      "learning_rate": 0.0002,
+      "loss": 0.5644456148147583,
+      "mean_token_accuracy": 0.7745120227336884,
+      "num_tokens": 3771586.0,
+      "step": 231
+    },
+    {
+      "entropy": 0.551509827375412,
+      "epoch": 0.8672897196261682,
+      "grad_norm": 0.026732027530670166,
+      "learning_rate": 0.0002,
+      "loss": 0.5578428506851196,
+      "mean_token_accuracy": 0.7758689671754837,
+      "num_tokens": 3787722.0,
+      "step": 232
+    },
+    {
+      "entropy": 0.5497246235609055,
+      "epoch": 0.8710280373831776,
+      "grad_norm": 0.026703782379627228,
+      "learning_rate": 0.0002,
+      "loss": 0.5488728284835815,
+      "mean_token_accuracy": 0.7787783890962601,
+      "num_tokens": 3804140.0,
+      "step": 233
+    },
+    {
+      "entropy": 0.5564501583576202,
+      "epoch": 0.874766355140187,
+      "grad_norm": 0.03062787838280201,
+      "learning_rate": 0.0002,
+      "loss": 0.5527218580245972,
+      "mean_token_accuracy": 0.773425742983818,
+      "num_tokens": 3820337.0,
+      "step": 234
+    },
+    {
+      "entropy": 0.5788596719503403,
+      "epoch": 0.8785046728971962,
+      "grad_norm": 0.02753940224647522,
+      "learning_rate": 0.0002,
+      "loss": 0.5711097121238708,
+      "mean_token_accuracy": 0.7668430060148239,
+      "num_tokens": 3836808.0,
+      "step": 235
+    },
+    {
+      "entropy": 0.5476598590612411,
+      "epoch": 0.8822429906542056,
+      "grad_norm": 0.027400346472859383,
+      "learning_rate": 0.0002,
+      "loss": 0.5463334918022156,
+      "mean_token_accuracy": 0.7781798541545868,
+      "num_tokens": 3853299.0,
+      "step": 236
+    },
+    {
+      "entropy": 0.5600839555263519,
+      "epoch": 0.8859813084112149,
+      "grad_norm": 0.028775498270988464,
+      "learning_rate": 0.0002,
+      "loss": 0.5644516348838806,
+      "mean_token_accuracy": 0.7703653573989868,
+      "num_tokens": 3869775.0,
+      "step": 237
+    },
+    {
+      "entropy": 0.5513971447944641,
+      "epoch": 0.8897196261682243,
+      "grad_norm": 0.03031282313168049,
+      "learning_rate": 0.0002,
+      "loss": 0.5573993921279907,
+      "mean_token_accuracy": 0.7723653614521027,
+      "num_tokens": 3885744.0,
+      "step": 238
+    },
+    {
+      "entropy": 0.5429326295852661,
+      "epoch": 0.8934579439252337,
+      "grad_norm": 0.033718034625053406,
+      "learning_rate": 0.0002,
+      "loss": 0.5453519821166992,
+      "mean_token_accuracy": 0.7795106470584869,
+      "num_tokens": 3902195.0,
+      "step": 239
+    },
+    {
+      "entropy": 0.5441034436225891,
+      "epoch": 0.897196261682243,
+      "grad_norm": 0.02875681221485138,
+      "learning_rate": 0.0002,
+      "loss": 0.5451952219009399,
+      "mean_token_accuracy": 0.7779187709093094,
+      "num_tokens": 3918423.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.5620324909687042,
+      "epoch": 0.9009345794392524,
+      "grad_norm": 0.03552895039319992,
+      "learning_rate": 0.0002,
+      "loss": 0.5609052181243896,
+      "mean_token_accuracy": 0.7701629996299744,
+      "num_tokens": 3934679.0,
+      "step": 241
+    },
+    {
+      "entropy": 0.5460164248943329,
+      "epoch": 0.9046728971962616,
+      "grad_norm": 0.029886359348893166,
+      "learning_rate": 0.0002,
+      "loss": 0.5491501688957214,
+      "mean_token_accuracy": 0.7752289175987244,
+      "num_tokens": 3950961.0,
+      "step": 242
+    },
+    {
+      "entropy": 0.546056404709816,
+      "epoch": 0.908411214953271,
+      "grad_norm": 0.030852871015667915,
+      "learning_rate": 0.0002,
+      "loss": 0.5487393140792847,
+      "mean_token_accuracy": 0.7780929207801819,
+      "num_tokens": 3966886.0,
+      "step": 243
+    },
+    {
+      "entropy": 0.5609732866287231,
+      "epoch": 0.9121495327102803,
+      "grad_norm": 0.030140889808535576,
+      "learning_rate": 0.0002,
+      "loss": 0.5606030225753784,
+      "mean_token_accuracy": 0.7716473340988159,
+      "num_tokens": 3983151.0,
+      "step": 244
+    },
+    {
+      "entropy": 0.5604758709669113,
+      "epoch": 0.9158878504672897,
+      "grad_norm": 0.029578525573015213,
+      "learning_rate": 0.0002,
+      "loss": 0.5585839748382568,
+      "mean_token_accuracy": 0.7732879519462585,
+      "num_tokens": 3999795.0,
+      "step": 245
+    },
+    {
+      "entropy": 0.5532268136739731,
+      "epoch": 0.9196261682242991,
+      "grad_norm": 0.027891919016838074,
+      "learning_rate": 0.0002,
+      "loss": 0.5495057106018066,
+      "mean_token_accuracy": 0.7766385078430176,
+      "num_tokens": 4016175.0,
+      "step": 246
+    },
+    {
+      "entropy": 0.5554448813199997,
+      "epoch": 0.9233644859813084,
+      "grad_norm": 0.02818435989320278,
+      "learning_rate": 0.0002,
+      "loss": 0.5564701557159424,
+      "mean_token_accuracy": 0.7740474194288254,
+      "num_tokens": 4032347.0,
+      "step": 247
+    },
+    {
+      "entropy": 0.5707967579364777,
+      "epoch": 0.9271028037383178,
+      "grad_norm": 0.03303643688559532,
+      "learning_rate": 0.0002,
+      "loss": 0.5724941492080688,
+      "mean_token_accuracy": 0.7658564150333405,
+      "num_tokens": 4048630.0,
+      "step": 248
+    },
+    {
+      "entropy": 0.5627187788486481,
+      "epoch": 0.930841121495327,
+      "grad_norm": 0.028818685561418533,
+      "learning_rate": 0.0002,
+      "loss": 0.558162271976471,
+      "mean_token_accuracy": 0.7722532004117966,
+      "num_tokens": 4065016.0,
+      "step": 249
+    },
+    {
+      "entropy": 0.559718668460846,
+      "epoch": 0.9345794392523364,
+      "grad_norm": 0.028685420751571655,
+      "learning_rate": 0.0002,
+      "loss": 0.5579565763473511,
+      "mean_token_accuracy": 0.7721400260925293,
+      "num_tokens": 4081500.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.5709707289934158,
+      "epoch": 0.9383177570093458,
+      "grad_norm": 0.02683587744832039,
+      "learning_rate": 0.0002,
+      "loss": 0.5758322477340698,
+      "mean_token_accuracy": 0.767963632941246,
+      "num_tokens": 4097612.0,
+      "step": 251
+    },
+    {
+      "entropy": 0.5344300121068954,
+      "epoch": 0.9420560747663551,
+      "grad_norm": 0.03321561962366104,
+      "learning_rate": 0.0002,
+      "loss": 0.5469601154327393,
+      "mean_token_accuracy": 0.7776601910591125,
+      "num_tokens": 4113985.0,
+      "step": 252
+    },
+    {
+      "entropy": 0.5629133135080338,
+      "epoch": 0.9457943925233645,
+      "grad_norm": 0.029224324971437454,
+      "learning_rate": 0.0002,
+      "loss": 0.5662878751754761,
+      "mean_token_accuracy": 0.7696039378643036,
+      "num_tokens": 4130584.0,
+      "step": 253
+    },
+    {
+      "entropy": 0.5419919043779373,
+      "epoch": 0.9495327102803738,
+      "grad_norm": 0.02663436345756054,
+      "learning_rate": 0.0002,
+      "loss": 0.5382669568061829,
+      "mean_token_accuracy": 0.7805986851453781,
+      "num_tokens": 4146982.0,
+      "step": 254
+    },
+    {
+      "entropy": 0.5534703135490417,
+      "epoch": 0.9532710280373832,
+      "grad_norm": 0.031037772074341774,
+      "learning_rate": 0.0002,
+      "loss": 0.5461369752883911,
+      "mean_token_accuracy": 0.7764857709407806,
+      "num_tokens": 4163123.0,
+      "step": 255
+    },
+    {
+      "entropy": 0.5555615872144699,
+      "epoch": 0.9570093457943926,
+      "grad_norm": 0.030655430629849434,
+      "learning_rate": 0.0002,
+      "loss": 0.547662615776062,
+      "mean_token_accuracy": 0.7764341235160828,
+      "num_tokens": 4179406.0,
+      "step": 256
+    },
+    {
+      "entropy": 0.5540354400873184,
+      "epoch": 0.9607476635514018,
+      "grad_norm": 0.026620658114552498,
+      "learning_rate": 0.0002,
+      "loss": 0.5554948449134827,
+      "mean_token_accuracy": 0.7734782248735428,
+      "num_tokens": 4195892.0,
+      "step": 257
+    },
+    {
+      "entropy": 0.5525697767734528,
+      "epoch": 0.9644859813084112,
+      "grad_norm": 0.034786809235811234,
+      "learning_rate": 0.0002,
+      "loss": 0.5611132383346558,
+      "mean_token_accuracy": 0.7743758261203766,
+      "num_tokens": 4212488.0,
+      "step": 258
+    },
+    {
+      "entropy": 0.5380009859800339,
+      "epoch": 0.9682242990654205,
+      "grad_norm": 0.030664261430501938,
+      "learning_rate": 0.0002,
+      "loss": 0.5456718802452087,
+      "mean_token_accuracy": 0.7777111083269119,
+      "num_tokens": 4228671.0,
+      "step": 259
+    },
+    {
+      "entropy": 0.5611264109611511,
+      "epoch": 0.9719626168224299,
+      "grad_norm": 0.03532060608267784,
+      "learning_rate": 0.0002,
+      "loss": 0.5694712400436401,
+      "mean_token_accuracy": 0.7701623737812042,
+      "num_tokens": 4244945.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.5638745427131653,
+      "epoch": 0.9757009345794393,
+      "grad_norm": 0.030665291473269463,
+      "learning_rate": 0.0002,
+      "loss": 0.5605924129486084,
+      "mean_token_accuracy": 0.773237332701683,
+      "num_tokens": 4261322.0,
+      "step": 261
+    },
+    {
+      "entropy": 0.5428778678178787,
+      "epoch": 0.9794392523364486,
+      "grad_norm": 0.02648600935935974,
+      "learning_rate": 0.0002,
+      "loss": 0.5397819876670837,
+      "mean_token_accuracy": 0.7809075862169266,
+      "num_tokens": 4277549.0,
+      "step": 262
+    },
+    {
+      "entropy": 0.5510172247886658,
+      "epoch": 0.983177570093458,
+      "grad_norm": 0.03357943519949913,
+      "learning_rate": 0.0002,
+      "loss": 0.5478790998458862,
+      "mean_token_accuracy": 0.776640459895134,
+      "num_tokens": 4293936.0,
+      "step": 263
+    },
+    {
+      "entropy": 0.574288085103035,
+      "epoch": 0.9869158878504672,
+      "grad_norm": 0.034169699996709824,
+      "learning_rate": 0.0002,
+      "loss": 0.5746400952339172,
+      "mean_token_accuracy": 0.7657656222581863,
+      "num_tokens": 4310409.0,
+      "step": 264
+    },
+    {
+      "entropy": 0.5594803243875504,
+      "epoch": 0.9906542056074766,
+      "grad_norm": 0.02862885594367981,
+      "learning_rate": 0.0002,
+      "loss": 0.5636025071144104,
+      "mean_token_accuracy": 0.7713442891836166,
+      "num_tokens": 4326712.0,
+      "step": 265
+    },
+    {
+      "entropy": 0.539836198091507,
+      "epoch": 0.994392523364486,
+      "grad_norm": 0.029507551342248917,
+      "learning_rate": 0.0002,
+      "loss": 0.5382612943649292,
+      "mean_token_accuracy": 0.7807378023862839,
+      "num_tokens": 4343219.0,
+      "step": 266
+    },
+    {
+      "entropy": 0.5551609694957733,
+      "epoch": 0.9981308411214953,
+      "grad_norm": 0.03200124204158783,
+      "learning_rate": 0.0002,
+      "loss": 0.5617719888687134,
+      "mean_token_accuracy": 0.773328885436058,
+      "num_tokens": 4359527.0,
+      "step": 267
+    },
+    {
+      "entropy": 0.5970158576965332,
+      "epoch": 1.0,
+      "grad_norm": 0.037066467106342316,
+      "learning_rate": 0.0002,
+      "loss": 0.5663654208183289,
+      "mean_token_accuracy": 0.7600691318511963,
+      "num_tokens": 4365381.0,
+      "step": 268
+    },
+    {
+      "entropy": 0.5525570958852768,
+      "epoch": 1.0037383177570094,
+      "grad_norm": 0.03900527581572533,
+      "learning_rate": 0.0002,
+      "loss": 0.5462143421173096,
+      "mean_token_accuracy": 0.778084933757782,
+      "num_tokens": 4381629.0,
+      "step": 269
+    },
+    {
+      "entropy": 0.54682257771492,
+      "epoch": 1.0074766355140188,
+      "grad_norm": 0.035536471754312515,
+      "learning_rate": 0.0002,
+      "loss": 0.545343816280365,
+      "mean_token_accuracy": 0.7801588326692581,
+      "num_tokens": 4398070.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.5365859419107437,
+      "epoch": 1.011214953271028,
+      "grad_norm": 0.034878168255090714,
+      "learning_rate": 0.0002,
+      "loss": 0.5464543104171753,
+      "mean_token_accuracy": 0.7776919007301331,
+      "num_tokens": 4414453.0,
+      "step": 271
+    },
+    {
+      "entropy": 0.5360404700040817,
+      "epoch": 1.0149532710280373,
+      "grad_norm": 0.048785679042339325,
+      "learning_rate": 0.0002,
+      "loss": 0.5453898906707764,
+      "mean_token_accuracy": 0.7778858244419098,
+      "num_tokens": 4430869.0,
+      "step": 272
+    },
+    {
+      "entropy": 0.5594298243522644,
+      "epoch": 1.0186915887850467,
+      "grad_norm": 0.03072672337293625,
+      "learning_rate": 0.0002,
+      "loss": 0.5592078566551208,
+      "mean_token_accuracy": 0.7696069329977036,
+      "num_tokens": 4447216.0,
+      "step": 273
+    },
+    {
+      "entropy": 0.5412543565034866,
+      "epoch": 1.0224299065420561,
+      "grad_norm": 0.02756349742412567,
+      "learning_rate": 0.0002,
+      "loss": 0.531460702419281,
+      "mean_token_accuracy": 0.7837116718292236,
+      "num_tokens": 4463360.0,
+      "step": 274
+    },
+    {
+      "entropy": 0.5926365852355957,
+      "epoch": 1.0261682242990655,
+      "grad_norm": 0.03326818346977234,
+      "learning_rate": 0.0002,
+      "loss": 0.5784746408462524,
+      "mean_token_accuracy": 0.766831248998642,
+      "num_tokens": 4479796.0,
+      "step": 275
+    },
+    {
+      "entropy": 0.5474714189767838,
+      "epoch": 1.0299065420560747,
+      "grad_norm": 0.03355902433395386,
+      "learning_rate": 0.0002,
+      "loss": 0.5428712368011475,
+      "mean_token_accuracy": 0.7803516983985901,
+      "num_tokens": 4495955.0,
+      "step": 276
+    },
+    {
+      "entropy": 0.5455677509307861,
+      "epoch": 1.033644859813084,
+      "grad_norm": 0.030776510015130043,
+      "learning_rate": 0.0002,
+      "loss": 0.5518208742141724,
+      "mean_token_accuracy": 0.7767827361822128,
+      "num_tokens": 4512316.0,
+      "step": 277
+    },
+    {
+      "entropy": 0.526500478386879,
+      "epoch": 1.0373831775700935,
+      "grad_norm": 0.033801671117544174,
+      "learning_rate": 0.0002,
+      "loss": 0.5366923809051514,
+      "mean_token_accuracy": 0.7806129455566406,
+      "num_tokens": 4528811.0,
+      "step": 278
+    },
+    {
+      "entropy": 0.541542574763298,
+      "epoch": 1.0411214953271029,
+      "grad_norm": 0.02768930047750473,
+      "learning_rate": 0.0002,
+      "loss": 0.5472083687782288,
+      "mean_token_accuracy": 0.7802270948886871,
+      "num_tokens": 4545459.0,
+      "step": 279
+    },
+    {
+      "entropy": 0.5428692698478699,
+      "epoch": 1.0448598130841122,
+      "grad_norm": 0.026535486802458763,
+      "learning_rate": 0.0002,
+      "loss": 0.5472521185874939,
+      "mean_token_accuracy": 0.7790202796459198,
+      "num_tokens": 4561743.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.5543521493673325,
+      "epoch": 1.0485981308411214,
+      "grad_norm": 0.02853293903172016,
+      "learning_rate": 0.0002,
+      "loss": 0.5497039556503296,
+      "mean_token_accuracy": 0.7779083847999573,
+      "num_tokens": 4577971.0,
+      "step": 281
+    },
+    {
+      "entropy": 0.5532562881708145,
+      "epoch": 1.0523364485981308,
+      "grad_norm": 0.032453302294015884,
+      "learning_rate": 0.0002,
+      "loss": 0.5507426261901855,
+      "mean_token_accuracy": 0.7767385840415955,
+      "num_tokens": 4594320.0,
+      "step": 282
+    },
+    {
+      "entropy": 0.5443340390920639,
+      "epoch": 1.0560747663551402,
+      "grad_norm": 0.02682569809257984,
+      "learning_rate": 0.0002,
+      "loss": 0.5412614345550537,
+      "mean_token_accuracy": 0.77817003428936,
+      "num_tokens": 4610559.0,
+      "step": 283
+    },
+    {
+      "entropy": 0.5440877228975296,
+      "epoch": 1.0598130841121496,
+      "grad_norm": 0.0319623202085495,
+      "learning_rate": 0.0002,
+      "loss": 0.5511168241500854,
+      "mean_token_accuracy": 0.7755201905965805,
+      "num_tokens": 4627054.0,
+      "step": 284
+    },
+    {
+      "entropy": 0.5398871749639511,
+      "epoch": 1.063551401869159,
+      "grad_norm": 0.03265109285712242,
+      "learning_rate": 0.0002,
+      "loss": 0.5518860220909119,
+      "mean_token_accuracy": 0.7781514823436737,
+      "num_tokens": 4643292.0,
+      "step": 285
+    },
+    {
+      "entropy": 0.5381411910057068,
+      "epoch": 1.0672897196261681,
+      "grad_norm": 0.03156265616416931,
+      "learning_rate": 0.0002,
+      "loss": 0.5409814119338989,
+      "mean_token_accuracy": 0.778824657201767,
+      "num_tokens": 4659607.0,
+      "step": 286
+    },
+    {
+      "entropy": 0.5548661798238754,
+      "epoch": 1.0710280373831775,
+      "grad_norm": 0.02918156050145626,
+      "learning_rate": 0.0002,
+      "loss": 0.5548707246780396,
+      "mean_token_accuracy": 0.7780612111091614,
+      "num_tokens": 4676232.0,
+      "step": 287
+    },
+    {
+      "entropy": 0.5623743981122971,
+      "epoch": 1.074766355140187,
+      "grad_norm": 0.03226201981306076,
+      "learning_rate": 0.0002,
+      "loss": 0.554677426815033,
+      "mean_token_accuracy": 0.7718137502670288,
+      "num_tokens": 4692488.0,
+      "step": 288
+    },
+    {
+      "entropy": 0.5557957291603088,
+      "epoch": 1.0785046728971963,
+      "grad_norm": 0.03029857762157917,
+      "learning_rate": 0.0002,
+      "loss": 0.5494679808616638,
+      "mean_token_accuracy": 0.7741122990846634,
+      "num_tokens": 4709044.0,
+      "step": 289
+    },
+    {
+      "entropy": 0.5503045320510864,
+      "epoch": 1.0822429906542057,
+      "grad_norm": 0.0307212695479393,
+      "learning_rate": 0.0002,
+      "loss": 0.5489369630813599,
+      "mean_token_accuracy": 0.7767912298440933,
+      "num_tokens": 4725364.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.539084866642952,
+      "epoch": 1.0859813084112149,
+      "grad_norm": 0.03190942481160164,
+      "learning_rate": 0.0002,
+      "loss": 0.5426544547080994,
+      "mean_token_accuracy": 0.7790433466434479,
+      "num_tokens": 4741941.0,
+      "step": 291
+    },
+    {
+      "entropy": 0.5557795614004135,
+      "epoch": 1.0897196261682243,
+      "grad_norm": 0.029419884085655212,
+      "learning_rate": 0.0002,
+      "loss": 0.5554101467132568,
+      "mean_token_accuracy": 0.7740319818258286,
+      "num_tokens": 4758205.0,
+      "step": 292
+    },
+    {
+      "entropy": 0.5505965203046799,
+      "epoch": 1.0934579439252337,
+      "grad_norm": 0.03429512679576874,
+      "learning_rate": 0.0002,
+      "loss": 0.5480504035949707,
+      "mean_token_accuracy": 0.7799728959798813,
+      "num_tokens": 4774531.0,
+      "step": 293
+    },
+    {
+      "entropy": 0.5202708393335342,
+      "epoch": 1.097196261682243,
+      "grad_norm": 0.03249349445104599,
+      "learning_rate": 0.0002,
+      "loss": 0.5263637900352478,
+      "mean_token_accuracy": 0.7847210466861725,
+      "num_tokens": 4790852.0,
+      "step": 294
+    },
+    {
+      "entropy": 0.5314037203788757,
+      "epoch": 1.1009345794392524,
+      "grad_norm": 0.03021075204014778,
+      "learning_rate": 0.0002,
+      "loss": 0.5385116934776306,
+      "mean_token_accuracy": 0.7801961302757263,
+      "num_tokens": 4807025.0,
+      "step": 295
+    },
+    {
+      "entropy": 0.5293320417404175,
+      "epoch": 1.1046728971962616,
+      "grad_norm": 0.028836827725172043,
+      "learning_rate": 0.0002,
+      "loss": 0.5360400676727295,
+      "mean_token_accuracy": 0.7814829498529434,
+      "num_tokens": 4823214.0,
+      "step": 296
+    },
+    {
+      "entropy": 0.5449055731296539,
+      "epoch": 1.108411214953271,
+      "grad_norm": 0.03287169337272644,
+      "learning_rate": 0.0002,
+      "loss": 0.5433195233345032,
+      "mean_token_accuracy": 0.7783774435520172,
+      "num_tokens": 4839406.0,
+      "step": 297
+    },
+    {
+      "entropy": 0.5480804145336151,
+      "epoch": 1.1121495327102804,
+      "grad_norm": 0.02733924239873886,
+      "learning_rate": 0.0002,
+      "loss": 0.5476067066192627,
+      "mean_token_accuracy": 0.7756602764129639,
+      "num_tokens": 4855715.0,
+      "step": 298
+    },
+    {
+      "entropy": 0.5223022475838661,
+      "epoch": 1.1158878504672898,
+      "grad_norm": 0.030849482864141464,
+      "learning_rate": 0.0002,
+      "loss": 0.5228248834609985,
+      "mean_token_accuracy": 0.7864928692579269,
+      "num_tokens": 4871905.0,
+      "step": 299
+    },
+    {
+      "entropy": 0.5403697192668915,
+      "epoch": 1.1196261682242992,
+      "grad_norm": 0.04067312180995941,
+      "learning_rate": 0.0002,
+      "loss": 0.5488058924674988,
+      "mean_token_accuracy": 0.7756592333316803,
+      "num_tokens": 4888255.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.5412193536758423,
+      "epoch": 1.1233644859813083,
+      "grad_norm": 0.0385926179587841,
+      "learning_rate": 0.0002,
+      "loss": 0.5371171832084656,
+      "mean_token_accuracy": 0.7830539792776108,
+      "num_tokens": 4904401.0,
+      "step": 301
+    },
+    {
+      "entropy": 0.5478123128414154,
+      "epoch": 1.1271028037383177,
+      "grad_norm": 0.034694232046604156,
+      "learning_rate": 0.0002,
+      "loss": 0.5448811054229736,
+      "mean_token_accuracy": 0.7775193750858307,
+      "num_tokens": 4920569.0,
+      "step": 302
+    },
+    {
+      "entropy": 0.5659691095352173,
+      "epoch": 1.1308411214953271,
+      "grad_norm": 0.0340200699865818,
+      "learning_rate": 0.0002,
+      "loss": 0.5552433729171753,
+      "mean_token_accuracy": 0.7739417254924774,
+      "num_tokens": 4936990.0,
+      "step": 303
+    },
+    {
+      "entropy": 0.5497810244560242,
+      "epoch": 1.1345794392523365,
+      "grad_norm": 0.03331321105360985,
+      "learning_rate": 0.0002,
+      "loss": 0.5419542193412781,
+      "mean_token_accuracy": 0.7797157317399979,
+      "num_tokens": 4953234.0,
+      "step": 304
+    },
+    {
+      "entropy": 0.5460351407527924,
+      "epoch": 1.1383177570093457,
+      "grad_norm": 0.028563540428876877,
+      "learning_rate": 0.0002,
+      "loss": 0.5521276593208313,
+      "mean_token_accuracy": 0.7758445143699646,
+      "num_tokens": 4969315.0,
+      "step": 305
+    },
+    {
+      "entropy": 0.5569003522396088,
+      "epoch": 1.142056074766355,
+      "grad_norm": 0.037444427609443665,
+      "learning_rate": 0.0002,
+      "loss": 0.5681867003440857,
+      "mean_token_accuracy": 0.768034890294075,
+      "num_tokens": 4985515.0,
+      "step": 306
+    },
+    {
+      "entropy": 0.5470724403858185,
+      "epoch": 1.1457943925233645,
+      "grad_norm": 0.03636348247528076,
+      "learning_rate": 0.0002,
+      "loss": 0.557773768901825,
+      "mean_token_accuracy": 0.7707736641168594,
+      "num_tokens": 5001866.0,
+      "step": 307
+    },
+    {
+      "entropy": 0.5416572839021683,
+      "epoch": 1.1495327102803738,
+      "grad_norm": 0.03566127270460129,
+      "learning_rate": 0.0002,
+      "loss": 0.5432877540588379,
+      "mean_token_accuracy": 0.7794886082410812,
+      "num_tokens": 5018262.0,
+      "step": 308
+    },
+    {
+      "entropy": 0.5430650115013123,
+      "epoch": 1.1532710280373832,
+      "grad_norm": 0.03256046026945114,
+      "learning_rate": 0.0002,
+      "loss": 0.5351021885871887,
+      "mean_token_accuracy": 0.7833839505910873,
+      "num_tokens": 5034628.0,
+      "step": 309
+    },
+    {
+      "entropy": 0.5461114794015884,
+      "epoch": 1.1570093457943926,
+      "grad_norm": 0.029036300256848335,
+      "learning_rate": 0.0002,
+      "loss": 0.5413210391998291,
+      "mean_token_accuracy": 0.7808763384819031,
+      "num_tokens": 5050585.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.554639920592308,
+      "epoch": 1.1607476635514018,
+      "grad_norm": 0.03361903131008148,
+      "learning_rate": 0.0002,
+      "loss": 0.5511503219604492,
+      "mean_token_accuracy": 0.7750130593776703,
+      "num_tokens": 5067128.0,
+      "step": 311
+    },
+    {
+      "entropy": 0.548187181353569,
+      "epoch": 1.1644859813084112,
+      "grad_norm": 0.031710732728242874,
+      "learning_rate": 0.0002,
+      "loss": 0.5527835488319397,
+      "mean_token_accuracy": 0.7745499759912491,
+      "num_tokens": 5083711.0,
+      "step": 312
+    },
+    {
+      "entropy": 0.5482472777366638,
+      "epoch": 1.1682242990654206,
+      "grad_norm": 0.02857045829296112,
+      "learning_rate": 0.0002,
+      "loss": 0.5483166575431824,
+      "mean_token_accuracy": 0.7769645005464554,
+      "num_tokens": 5100016.0,
+      "step": 313
+    },
+    {
+      "entropy": 0.5292567014694214,
+      "epoch": 1.17196261682243,
+      "grad_norm": 0.03421966731548309,
+      "learning_rate": 0.0002,
+      "loss": 0.5375955104827881,
+      "mean_token_accuracy": 0.7807260155677795,
+      "num_tokens": 5116491.0,
+      "step": 314
+    },
+    {
+      "entropy": 0.5369534343481064,
+      "epoch": 1.1757009345794391,
+      "grad_norm": 0.03103984147310257,
+      "learning_rate": 0.0002,
+      "loss": 0.5436944365501404,
+      "mean_token_accuracy": 0.7778706103563309,
+      "num_tokens": 5132827.0,
+      "step": 315
+    },
+    {
+      "entropy": 0.5408389419317245,
+      "epoch": 1.1794392523364485,
+      "grad_norm": 0.030202720314264297,
+      "learning_rate": 0.0002,
+      "loss": 0.5370794534683228,
+      "mean_token_accuracy": 0.7810905873775482,
+      "num_tokens": 5149320.0,
+      "step": 316
+    },
+    {
+      "entropy": 0.5496339946985245,
+      "epoch": 1.183177570093458,
+      "grad_norm": 0.051472123712301254,
+      "learning_rate": 0.0002,
+      "loss": 0.5469900965690613,
+      "mean_token_accuracy": 0.7799471020698547,
+      "num_tokens": 5165783.0,
+      "step": 317
+    },
+    {
+      "entropy": 0.5685789883136749,
+      "epoch": 1.1869158878504673,
+      "grad_norm": 0.02587290294468403,
+      "learning_rate": 0.0002,
+      "loss": 0.5606150031089783,
+      "mean_token_accuracy": 0.7720179110765457,
+      "num_tokens": 5182220.0,
+      "step": 318
+    },
+    {
+      "entropy": 0.562404528260231,
+      "epoch": 1.1906542056074767,
+      "grad_norm": 0.02697838842868805,
+      "learning_rate": 0.0002,
+      "loss": 0.5576499700546265,
+      "mean_token_accuracy": 0.773972287774086,
+      "num_tokens": 5198785.0,
+      "step": 319
+    },
+    {
+      "entropy": 0.5579852014780045,
+      "epoch": 1.194392523364486,
+      "grad_norm": 0.03318994492292404,
+      "learning_rate": 0.0002,
+      "loss": 0.5587575435638428,
+      "mean_token_accuracy": 0.7752004414796829,
+      "num_tokens": 5215306.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.5371066629886627,
+      "epoch": 1.1981308411214953,
+      "grad_norm": 0.03372027724981308,
+      "learning_rate": 0.0002,
+      "loss": 0.5434669852256775,
+      "mean_token_accuracy": 0.7805782556533813,
+      "num_tokens": 5231543.0,
+      "step": 321
+    },
+    {
+      "entropy": 0.5399051606655121,
+      "epoch": 1.2018691588785047,
+      "grad_norm": 0.02985556237399578,
+      "learning_rate": 0.0002,
+      "loss": 0.535866379737854,
+      "mean_token_accuracy": 0.7829108238220215,
+      "num_tokens": 5248118.0,
+      "step": 322
+    },
+    {
+      "entropy": 0.5591253787279129,
+      "epoch": 1.205607476635514,
+      "grad_norm": 0.036681629717350006,
+      "learning_rate": 0.0002,
+      "loss": 0.5629308223724365,
+      "mean_token_accuracy": 0.7753039449453354,
+      "num_tokens": 5264354.0,
+      "step": 323
+    },
+    {
+      "entropy": 0.530854269862175,
+      "epoch": 1.2093457943925234,
+      "grad_norm": 0.03271258994936943,
+      "learning_rate": 0.0002,
+      "loss": 0.5393017530441284,
+      "mean_token_accuracy": 0.779604122042656,
+      "num_tokens": 5280420.0,
+      "step": 324
+    },
+    {
+      "entropy": 0.5350240170955658,
+      "epoch": 1.2130841121495326,
+      "grad_norm": 0.03653952106833458,
+      "learning_rate": 0.0002,
+      "loss": 0.5409165620803833,
+      "mean_token_accuracy": 0.7822408974170685,
+      "num_tokens": 5296606.0,
+      "step": 325
+    },
+    {
+      "entropy": 0.5648403018712997,
+      "epoch": 1.216822429906542,
+      "grad_norm": 0.03994280472397804,
+      "learning_rate": 0.0002,
+      "loss": 0.5743881464004517,
+      "mean_token_accuracy": 0.7659939229488373,
+      "num_tokens": 5313007.0,
+      "step": 326
+    },
+    {
+      "entropy": 0.5432536900043488,
+      "epoch": 1.2205607476635514,
+      "grad_norm": 0.02927422896027565,
+      "learning_rate": 0.0002,
+      "loss": 0.5374885201454163,
+      "mean_token_accuracy": 0.7827716916799545,
+      "num_tokens": 5329527.0,
+      "step": 327
+    },
+    {
+      "entropy": 0.5514259338378906,
+      "epoch": 1.2242990654205608,
+      "grad_norm": 0.0344291552901268,
+      "learning_rate": 0.0002,
+      "loss": 0.5449932217597961,
+      "mean_token_accuracy": 0.7758236229419708,
+      "num_tokens": 5345773.0,
+      "step": 328
+    },
+    {
+      "entropy": 0.5402681678533554,
+      "epoch": 1.2280373831775702,
+      "grad_norm": 0.030332623049616814,
+      "learning_rate": 0.0002,
+      "loss": 0.5353883504867554,
+      "mean_token_accuracy": 0.7829276025295258,
+      "num_tokens": 5362127.0,
+      "step": 329
+    },
+    {
+      "entropy": 0.5448068082332611,
+      "epoch": 1.2317757009345796,
+      "grad_norm": 0.03531905263662338,
+      "learning_rate": 0.0002,
+      "loss": 0.5491797924041748,
+      "mean_token_accuracy": 0.7790112793445587,
+      "num_tokens": 5378581.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.5472903549671173,
+      "epoch": 1.2355140186915887,
+      "grad_norm": 0.031015828251838684,
+      "learning_rate": 0.0002,
+      "loss": 0.5477415323257446,
+      "mean_token_accuracy": 0.7792595028877258,
+      "num_tokens": 5395212.0,
+      "step": 331
+    },
+    {
+      "entropy": 0.5454255491495132,
+      "epoch": 1.2392523364485981,
+      "grad_norm": 0.029113125056028366,
+      "learning_rate": 0.0002,
+      "loss": 0.5511353611946106,
+      "mean_token_accuracy": 0.7783665657043457,
+      "num_tokens": 5411639.0,
+      "step": 332
+    },
+    {
+      "entropy": 0.540988489985466,
+      "epoch": 1.2429906542056075,
+      "grad_norm": 0.03799515590071678,
+      "learning_rate": 0.0002,
+      "loss": 0.5437558889389038,
+      "mean_token_accuracy": 0.7799660414457321,
+      "num_tokens": 5428149.0,
+      "step": 333
+    },
+    {
+      "entropy": 0.5526563823223114,
+      "epoch": 1.246728971962617,
+      "grad_norm": 0.031356047838926315,
+      "learning_rate": 0.0002,
+      "loss": 0.5517259836196899,
+      "mean_token_accuracy": 0.7754609137773514,
+      "num_tokens": 5444823.0,
+      "step": 334
+    },
+    {
+      "entropy": 0.5555935055017471,
+      "epoch": 1.250467289719626,
+      "grad_norm": 0.029837489128112793,
+      "learning_rate": 0.0002,
+      "loss": 0.5508607029914856,
+      "mean_token_accuracy": 0.7758114188909531,
+      "num_tokens": 5461092.0,
+      "step": 335
+    },
+    {
+      "entropy": 0.5559129118919373,
+      "epoch": 1.2542056074766355,
+      "grad_norm": 0.033403050154447556,
+      "learning_rate": 0.0002,
+      "loss": 0.5595272183418274,
+      "mean_token_accuracy": 0.7721016854047775,
+      "num_tokens": 5477562.0,
+      "step": 336
+    },
+    {
+      "entropy": 0.5547732263803482,
+      "epoch": 1.2579439252336448,
+      "grad_norm": 0.030496301129460335,
+      "learning_rate": 0.0002,
+      "loss": 0.5562547445297241,
+      "mean_token_accuracy": 0.7715321332216263,
+      "num_tokens": 5493726.0,
+      "step": 337
+    },
+    {
+      "entropy": 0.5420037060976028,
+      "epoch": 1.2616822429906542,
+      "grad_norm": 0.029967380687594414,
+      "learning_rate": 0.0002,
+      "loss": 0.5383363962173462,
+      "mean_token_accuracy": 0.7796787321567535,
+      "num_tokens": 5510133.0,
+      "step": 338
+    },
+    {
+      "entropy": 0.5514421463012695,
+      "epoch": 1.2654205607476636,
+      "grad_norm": 0.028999267145991325,
+      "learning_rate": 0.0002,
+      "loss": 0.5496451258659363,
+      "mean_token_accuracy": 0.775850385427475,
+      "num_tokens": 5526472.0,
+      "step": 339
+    },
+    {
+      "entropy": 0.5502642691135406,
+      "epoch": 1.269158878504673,
+      "grad_norm": 0.031769949942827225,
+      "learning_rate": 0.0002,
+      "loss": 0.5571942925453186,
+      "mean_token_accuracy": 0.7752085626125336,
+      "num_tokens": 5542630.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.5570239722728729,
+      "epoch": 1.2728971962616822,
+      "grad_norm": 0.027039222419261932,
+      "learning_rate": 0.0002,
+      "loss": 0.5579078197479248,
+      "mean_token_accuracy": 0.771331325173378,
+      "num_tokens": 5559029.0,
+      "step": 341
+    },
+    {
+      "entropy": 0.5515656173229218,
+      "epoch": 1.2766355140186916,
+      "grad_norm": 0.029703624546527863,
+      "learning_rate": 0.0002,
+      "loss": 0.5461840033531189,
+      "mean_token_accuracy": 0.7807486355304718,
+      "num_tokens": 5575356.0,
+      "step": 342
+    },
+    {
+      "entropy": 0.5342657566070557,
+      "epoch": 1.280373831775701,
+      "grad_norm": 0.03180944547057152,
+      "learning_rate": 0.0002,
+      "loss": 0.5324912071228027,
+      "mean_token_accuracy": 0.7840546667575836,
+      "num_tokens": 5591449.0,
+      "step": 343
+    },
+    {
+      "entropy": 0.5342217683792114,
+      "epoch": 1.2841121495327104,
+      "grad_norm": 0.029021048918366432,
+      "learning_rate": 0.0002,
+      "loss": 0.5414597392082214,
+      "mean_token_accuracy": 0.780342310667038,
+      "num_tokens": 5607833.0,
+      "step": 344
+    },
+    {
+      "entropy": 0.55648373067379,
+      "epoch": 1.2878504672897195,
+      "grad_norm": 0.046803366392850876,
+      "learning_rate": 0.0002,
+      "loss": 0.5738747119903564,
+      "mean_token_accuracy": 0.7661998569965363,
+      "num_tokens": 5624322.0,
+      "step": 345
+    },
+    {
+      "entropy": 0.5502487570047379,
+      "epoch": 1.291588785046729,
+      "grad_norm": 0.03315203636884689,
+      "learning_rate": 0.0002,
+      "loss": 0.5500799417495728,
+      "mean_token_accuracy": 0.7736621350049973,
+      "num_tokens": 5640389.0,
+      "step": 346
+    },
+    {
+      "entropy": 0.5614677965641022,
+      "epoch": 1.2953271028037383,
+      "grad_norm": 0.03268338739871979,
+      "learning_rate": 0.0002,
+      "loss": 0.5521507263183594,
+      "mean_token_accuracy": 0.7767487168312073,
+      "num_tokens": 5657037.0,
+      "step": 347
+    },
+    {
+      "entropy": 0.5565991103649139,
+      "epoch": 1.2990654205607477,
+      "grad_norm": 0.03224696218967438,
+      "learning_rate": 0.0002,
+      "loss": 0.5461902022361755,
+      "mean_token_accuracy": 0.7800005227327347,
+      "num_tokens": 5673177.0,
+      "step": 348
+    },
+    {
+      "entropy": 0.5575346350669861,
+      "epoch": 1.302803738317757,
+      "grad_norm": 0.03355446085333824,
+      "learning_rate": 0.0002,
+      "loss": 0.5535120368003845,
+      "mean_token_accuracy": 0.7763462960720062,
+      "num_tokens": 5689487.0,
+      "step": 349
+    },
+    {
+      "entropy": 0.5268712043762207,
+      "epoch": 1.3065420560747665,
+      "grad_norm": 0.03238300606608391,
+      "learning_rate": 0.0002,
+      "loss": 0.5330650806427002,
+      "mean_token_accuracy": 0.7824077904224396,
+      "num_tokens": 5705730.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.5374382436275482,
+      "epoch": 1.3102803738317756,
+      "grad_norm": 0.03740353882312775,
+      "learning_rate": 0.0002,
+      "loss": 0.547337532043457,
+      "mean_token_accuracy": 0.7766657322645187,
+      "num_tokens": 5722106.0,
+      "step": 351
+    },
+    {
+      "entropy": 0.5559621900320053,
+      "epoch": 1.314018691588785,
+      "grad_norm": 0.030790183693170547,
+      "learning_rate": 0.0002,
+      "loss": 0.5596389770507812,
+      "mean_token_accuracy": 0.7734881490468979,
+      "num_tokens": 5738765.0,
+      "step": 352
+    },
+    {
+      "entropy": 0.5731736719608307,
+      "epoch": 1.3177570093457944,
+      "grad_norm": 0.029810264706611633,
+      "learning_rate": 0.0002,
+      "loss": 0.5712341070175171,
+      "mean_token_accuracy": 0.7706846296787262,
+      "num_tokens": 5755390.0,
+      "step": 353
+    },
+    {
+      "entropy": 0.5471203178167343,
+      "epoch": 1.3214953271028038,
+      "grad_norm": 0.031591080129146576,
+      "learning_rate": 0.0002,
+      "loss": 0.5398964881896973,
+      "mean_token_accuracy": 0.7803937494754791,
+      "num_tokens": 5771544.0,
+      "step": 354
+    },
+    {
+      "entropy": 0.54648657143116,
+      "epoch": 1.325233644859813,
+      "grad_norm": 0.02947516180574894,
+      "learning_rate": 0.0002,
+      "loss": 0.5480178594589233,
+      "mean_token_accuracy": 0.7766480296850204,
+      "num_tokens": 5787883.0,
+      "step": 355
+    },
+    {
+      "entropy": 0.547360748052597,
+      "epoch": 1.3289719626168224,
+      "grad_norm": 0.03545290604233742,
+      "learning_rate": 0.0002,
+      "loss": 0.5540775656700134,
+      "mean_token_accuracy": 0.7739741057157516,
+      "num_tokens": 5804479.0,
+      "step": 356
+    },
+    {
+      "entropy": 0.5666938722133636,
+      "epoch": 1.3327102803738318,
+      "grad_norm": 0.030594807118177414,
+      "learning_rate": 0.0002,
+      "loss": 0.5684746503829956,
+      "mean_token_accuracy": 0.7727518826723099,
+      "num_tokens": 5820828.0,
+      "step": 357
+    },
+    {
+      "entropy": 0.5427974164485931,
+      "epoch": 1.3364485981308412,
+      "grad_norm": 0.028014780953526497,
+      "learning_rate": 0.0002,
+      "loss": 0.5475256443023682,
+      "mean_token_accuracy": 0.7783348560333252,
+      "num_tokens": 5837279.0,
+      "step": 358
+    },
+    {
+      "entropy": 0.5367537513375282,
+      "epoch": 1.3401869158878505,
+      "grad_norm": 0.02776091918349266,
+      "learning_rate": 0.0002,
+      "loss": 0.5400804877281189,
+      "mean_token_accuracy": 0.7791878134012222,
+      "num_tokens": 5853587.0,
+      "step": 359
+    },
+    {
+      "entropy": 0.5417804718017578,
+      "epoch": 1.34392523364486,
+      "grad_norm": 0.030707573518157005,
+      "learning_rate": 0.0002,
+      "loss": 0.5384262204170227,
+      "mean_token_accuracy": 0.7809765785932541,
+      "num_tokens": 5869870.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.5336113572120667,
+      "epoch": 1.347663551401869,
+      "grad_norm": 0.02800143137574196,
+      "learning_rate": 0.0002,
+      "loss": 0.5343897342681885,
+      "mean_token_accuracy": 0.7851298898458481,
+      "num_tokens": 5886161.0,
+      "step": 361
+    },
+    {
+      "entropy": 0.5595630407333374,
+      "epoch": 1.3514018691588785,
+      "grad_norm": 0.033487141132354736,
+      "learning_rate": 0.0002,
+      "loss": 0.5617246031761169,
+      "mean_token_accuracy": 0.7718270570039749,
+      "num_tokens": 5902403.0,
+      "step": 362
+    },
+    {
+      "entropy": 0.5564123690128326,
+      "epoch": 1.355140186915888,
+      "grad_norm": 0.030184373259544373,
+      "learning_rate": 0.0002,
+      "loss": 0.5559881925582886,
+      "mean_token_accuracy": 0.7759525775909424,
+      "num_tokens": 5918682.0,
+      "step": 363
+    },
+    {
+      "entropy": 0.5504587888717651,
+      "epoch": 1.358878504672897,
+      "grad_norm": 0.030958138406276703,
+      "learning_rate": 0.0002,
+      "loss": 0.5547590255737305,
+      "mean_token_accuracy": 0.7714252471923828,
+      "num_tokens": 5935040.0,
+      "step": 364
+    },
+    {
+      "entropy": 0.5442556142807007,
+      "epoch": 1.3626168224299064,
+      "grad_norm": 0.02721753716468811,
+      "learning_rate": 0.0002,
+      "loss": 0.5465304851531982,
+      "mean_token_accuracy": 0.7787513434886932,
+      "num_tokens": 5951319.0,
+      "step": 365
+    },
+    {
+      "entropy": 0.5471279621124268,
+      "epoch": 1.3663551401869158,
+      "grad_norm": 0.02792942337691784,
+      "learning_rate": 0.0002,
+      "loss": 0.5474483966827393,
+      "mean_token_accuracy": 0.7781872451305389,
+      "num_tokens": 5967748.0,
+      "step": 366
+    },
+    {
+      "entropy": 0.5490291863679886,
+      "epoch": 1.3700934579439252,
+      "grad_norm": 0.028123004361987114,
+      "learning_rate": 0.0002,
+      "loss": 0.543886125087738,
+      "mean_token_accuracy": 0.7788564115762711,
+      "num_tokens": 5984234.0,
+      "step": 367
+    },
+    {
+      "entropy": 0.5531343668699265,
+      "epoch": 1.3738317757009346,
+      "grad_norm": 0.028855035081505775,
+      "learning_rate": 0.0002,
+      "loss": 0.5493640303611755,
+      "mean_token_accuracy": 0.7753023356199265,
+      "num_tokens": 6000373.0,
+      "step": 368
+    },
+    {
+      "entropy": 0.5437296479940414,
+      "epoch": 1.377570093457944,
+      "grad_norm": 0.03395684063434601,
+      "learning_rate": 0.0002,
+      "loss": 0.5439614057540894,
+      "mean_token_accuracy": 0.7777044326066971,
+      "num_tokens": 6016881.0,
+      "step": 369
+    },
+    {
+      "entropy": 0.5404936373233795,
+      "epoch": 1.3813084112149534,
+      "grad_norm": 0.032147135585546494,
+      "learning_rate": 0.0002,
+      "loss": 0.545538604259491,
+      "mean_token_accuracy": 0.7780790776014328,
+      "num_tokens": 6033034.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.5291552245616913,
+      "epoch": 1.3850467289719626,
+      "grad_norm": 0.033786509186029434,
+      "learning_rate": 0.0002,
+      "loss": 0.5318939685821533,
+      "mean_token_accuracy": 0.7833190411329269,
+      "num_tokens": 6049080.0,
+      "step": 371
+    },
+    {
+      "entropy": 0.567609116435051,
+      "epoch": 1.388785046728972,
+      "grad_norm": 0.028979122638702393,
+      "learning_rate": 0.0002,
+      "loss": 0.5664098858833313,
+      "mean_token_accuracy": 0.7676924616098404,
+      "num_tokens": 6065561.0,
+      "step": 372
+    },
+    {
+      "entropy": 0.5492920726537704,
+      "epoch": 1.3925233644859814,
+      "grad_norm": 0.03035721927881241,
+      "learning_rate": 0.0002,
+      "loss": 0.5452470779418945,
+      "mean_token_accuracy": 0.7780449539422989,
+      "num_tokens": 6081826.0,
+      "step": 373
+    },
+    {
+      "entropy": 0.5530673414468765,
+      "epoch": 1.3962616822429905,
+      "grad_norm": 0.028820699080824852,
+      "learning_rate": 0.0002,
+      "loss": 0.54978346824646,
+      "mean_token_accuracy": 0.7757824063301086,
+      "num_tokens": 6098304.0,
+      "step": 374
+    },
+    {
+      "entropy": 0.5440555065870285,
+      "epoch": 1.4,
+      "grad_norm": 0.032084014266729355,
+      "learning_rate": 0.0002,
+      "loss": 0.5414038896560669,
+      "mean_token_accuracy": 0.7783534377813339,
+      "num_tokens": 6114548.0,
+      "step": 375
+    },
+    {
+      "entropy": 0.5330861955881119,
+      "epoch": 1.4037383177570093,
+      "grad_norm": 0.03771211951971054,
+      "learning_rate": 0.0002,
+      "loss": 0.5359234809875488,
+      "mean_token_accuracy": 0.7809047400951385,
+      "num_tokens": 6130775.0,
+      "step": 376
+    },
+    {
+      "entropy": 0.5356210172176361,
+      "epoch": 1.4074766355140187,
+      "grad_norm": 0.031670473515987396,
+      "learning_rate": 0.0002,
+      "loss": 0.539705753326416,
+      "mean_token_accuracy": 0.7816208004951477,
+      "num_tokens": 6147119.0,
+      "step": 377
+    },
+    {
+      "entropy": 0.5264314264059067,
+      "epoch": 1.411214953271028,
+      "grad_norm": 0.03143749758601189,
+      "learning_rate": 0.0002,
+      "loss": 0.5309686660766602,
+      "mean_token_accuracy": 0.7844567745923996,
+      "num_tokens": 6163184.0,
+      "step": 378
+    },
+    {
+      "entropy": 0.5401338934898376,
+      "epoch": 1.4149532710280375,
+      "grad_norm": 0.03694392740726471,
+      "learning_rate": 0.0002,
+      "loss": 0.5471729040145874,
+      "mean_token_accuracy": 0.7767902612686157,
+      "num_tokens": 6179419.0,
+      "step": 379
+    },
+    {
+      "entropy": 0.5529890209436417,
+      "epoch": 1.4186915887850469,
+      "grad_norm": 0.033812835812568665,
+      "learning_rate": 0.0002,
+      "loss": 0.565970778465271,
+      "mean_token_accuracy": 0.7665126621723175,
+      "num_tokens": 6195649.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.5637273043394089,
+      "epoch": 1.422429906542056,
+      "grad_norm": 0.02983681485056877,
+      "learning_rate": 0.0002,
+      "loss": 0.5626575350761414,
+      "mean_token_accuracy": 0.7695245891809464,
+      "num_tokens": 6212102.0,
+      "step": 381
+    },
+    {
+      "entropy": 0.5482936352491379,
+      "epoch": 1.4261682242990654,
+      "grad_norm": 0.026917651295661926,
+      "learning_rate": 0.0002,
+      "loss": 0.54402095079422,
+      "mean_token_accuracy": 0.7796863317489624,
+      "num_tokens": 6228396.0,
+      "step": 382
+    },
+    {
+      "entropy": 0.5501932799816132,
+      "epoch": 1.4299065420560748,
+      "grad_norm": 0.029674382880330086,
+      "learning_rate": 0.0002,
+      "loss": 0.5445118546485901,
+      "mean_token_accuracy": 0.7807586491107941,
+      "num_tokens": 6244754.0,
+      "step": 383
+    },
+    {
+      "entropy": 0.5524531006813049,
+      "epoch": 1.433644859813084,
+      "grad_norm": 0.031190212815999985,
+      "learning_rate": 0.0002,
+      "loss": 0.5460107922554016,
+      "mean_token_accuracy": 0.7775425314903259,
+      "num_tokens": 6261129.0,
+      "step": 384
+    },
+    {
+      "entropy": 0.5559191405773163,
+      "epoch": 1.4373831775700934,
+      "grad_norm": 0.03291701897978783,
+      "learning_rate": 0.0002,
+      "loss": 0.5565997362136841,
+      "mean_token_accuracy": 0.7729406207799911,
+      "num_tokens": 6277295.0,
+      "step": 385
+    },
+    {
+      "entropy": 0.5387507379055023,
+      "epoch": 1.4411214953271028,
+      "grad_norm": 0.029946424067020416,
+      "learning_rate": 0.0002,
+      "loss": 0.5400161147117615,
+      "mean_token_accuracy": 0.7815364450216293,
+      "num_tokens": 6293740.0,
+      "step": 386
+    },
+    {
+      "entropy": 0.5453538447618484,
+      "epoch": 1.4448598130841122,
+      "grad_norm": 0.03343931958079338,
+      "learning_rate": 0.0002,
+      "loss": 0.5513398051261902,
+      "mean_token_accuracy": 0.7751688063144684,
+      "num_tokens": 6309953.0,
+      "step": 387
+    },
+    {
+      "entropy": 0.5543301105499268,
+      "epoch": 1.4485981308411215,
+      "grad_norm": 0.02918226271867752,
+      "learning_rate": 0.0002,
+      "loss": 0.5551657676696777,
+      "mean_token_accuracy": 0.7708156853914261,
+      "num_tokens": 6326235.0,
+      "step": 388
+    },
+    {
+      "entropy": 0.5655284225940704,
+      "epoch": 1.452336448598131,
+      "grad_norm": 0.031265538185834885,
+      "learning_rate": 0.0002,
+      "loss": 0.5654095411300659,
+      "mean_token_accuracy": 0.7689371109008789,
+      "num_tokens": 6342382.0,
+      "step": 389
+    },
+    {
+      "entropy": 0.543906643986702,
+      "epoch": 1.45607476635514,
+      "grad_norm": 0.0303683839738369,
+      "learning_rate": 0.0002,
+      "loss": 0.5439735651016235,
+      "mean_token_accuracy": 0.7809619605541229,
+      "num_tokens": 6358481.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.5395262986421585,
+      "epoch": 1.4598130841121495,
+      "grad_norm": 0.029894739389419556,
+      "learning_rate": 0.0002,
+      "loss": 0.5406848192214966,
+      "mean_token_accuracy": 0.7806549370288849,
+      "num_tokens": 6374676.0,
+      "step": 391
+    },
+    {
+      "entropy": 0.5415726751089096,
+      "epoch": 1.4635514018691589,
+      "grad_norm": 0.03674378991127014,
+      "learning_rate": 0.0002,
+      "loss": 0.5463482141494751,
+      "mean_token_accuracy": 0.777136966586113,
+      "num_tokens": 6390831.0,
+      "step": 392
+    },
+    {
+      "entropy": 0.5493590235710144,
+      "epoch": 1.4672897196261683,
+      "grad_norm": 0.028417574241757393,
+      "learning_rate": 0.0002,
+      "loss": 0.5471122860908508,
+      "mean_token_accuracy": 0.7766950279474258,
+      "num_tokens": 6407316.0,
+      "step": 393
+    },
+    {
+      "entropy": 0.5292122960090637,
+      "epoch": 1.4710280373831774,
+      "grad_norm": 0.03298041224479675,
+      "learning_rate": 0.0002,
+      "loss": 0.5297983884811401,
+      "mean_token_accuracy": 0.7847895622253418,
+      "num_tokens": 6423557.0,
+      "step": 394
+    },
+    {
+      "entropy": 0.5528764873743057,
+      "epoch": 1.4747663551401868,
+      "grad_norm": 0.02988980896770954,
+      "learning_rate": 0.0002,
+      "loss": 0.5556773543357849,
+      "mean_token_accuracy": 0.7742319852113724,
+      "num_tokens": 6439634.0,
+      "step": 395
+    },
+    {
+      "entropy": 0.551476001739502,
+      "epoch": 1.4785046728971962,
+      "grad_norm": 0.030642125755548477,
+      "learning_rate": 0.0002,
+      "loss": 0.550857663154602,
+      "mean_token_accuracy": 0.7747069448232651,
+      "num_tokens": 6455686.0,
+      "step": 396
+    },
+    {
+      "entropy": 0.5380145758390427,
+      "epoch": 1.4822429906542056,
+      "grad_norm": 0.02931920997798443,
+      "learning_rate": 0.0002,
+      "loss": 0.5356473326683044,
+      "mean_token_accuracy": 0.7809695601463318,
+      "num_tokens": 6471935.0,
+      "step": 397
+    },
+    {
+      "entropy": 0.5300455242395401,
+      "epoch": 1.485981308411215,
+      "grad_norm": 0.035832446068525314,
+      "learning_rate": 0.0002,
+      "loss": 0.5344862937927246,
+      "mean_token_accuracy": 0.783536359667778,
+      "num_tokens": 6488278.0,
+      "step": 398
+    },
+    {
+      "entropy": 0.5476836711168289,
+      "epoch": 1.4897196261682244,
+      "grad_norm": 0.0333210714161396,
+      "learning_rate": 0.0002,
+      "loss": 0.5524132251739502,
+      "mean_token_accuracy": 0.7725930511951447,
+      "num_tokens": 6504288.0,
+      "step": 399
+    },
+    {
+      "entropy": 0.535921260714531,
+      "epoch": 1.4934579439252336,
+      "grad_norm": 0.03744916617870331,
+      "learning_rate": 0.0002,
+      "loss": 0.5385454297065735,
+      "mean_token_accuracy": 0.7788458317518234,
+      "num_tokens": 6520498.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.5604094713926315,
+      "epoch": 1.497196261682243,
+      "grad_norm": 0.029392311349511147,
+      "learning_rate": 0.0002,
+      "loss": 0.5529255867004395,
+      "mean_token_accuracy": 0.7746162861585617,
+      "num_tokens": 6536660.0,
+      "step": 401
+    },
+    {
+      "entropy": 0.5519481748342514,
+      "epoch": 1.5009345794392523,
+      "grad_norm": 0.031238850206136703,
+      "learning_rate": 0.0002,
+      "loss": 0.5494250655174255,
+      "mean_token_accuracy": 0.7768653184175491,
+      "num_tokens": 6553126.0,
+      "step": 402
+    },
+    {
+      "entropy": 0.5527535974979401,
+      "epoch": 1.5046728971962615,
+      "grad_norm": 0.03839221969246864,
+      "learning_rate": 0.0002,
+      "loss": 0.5625566840171814,
+      "mean_token_accuracy": 0.7720162123441696,
+      "num_tokens": 6569143.0,
+      "step": 403
+    },
+    {
+      "entropy": 0.5520975738763809,
+      "epoch": 1.508411214953271,
+      "grad_norm": 0.03356786444783211,
+      "learning_rate": 0.0002,
+      "loss": 0.5538414716720581,
+      "mean_token_accuracy": 0.7732953429222107,
+      "num_tokens": 6585541.0,
+      "step": 404
+    },
+    {
+      "entropy": 0.5360509008169174,
+      "epoch": 1.5121495327102803,
+      "grad_norm": 0.03467942029237747,
+      "learning_rate": 0.0002,
+      "loss": 0.538615345954895,
+      "mean_token_accuracy": 0.781312882900238,
+      "num_tokens": 6602274.0,
+      "step": 405
+    },
+    {
+      "entropy": 0.5343838483095169,
+      "epoch": 1.5158878504672897,
+      "grad_norm": 0.031671129167079926,
+      "learning_rate": 0.0002,
+      "loss": 0.5385384559631348,
+      "mean_token_accuracy": 0.7813232839107513,
+      "num_tokens": 6618593.0,
+      "step": 406
+    },
+    {
+      "entropy": 0.5469245910644531,
+      "epoch": 1.519626168224299,
+      "grad_norm": 0.032783299684524536,
+      "learning_rate": 0.0002,
+      "loss": 0.5508422255516052,
+      "mean_token_accuracy": 0.7776025533676147,
+      "num_tokens": 6634867.0,
+      "step": 407
+    },
+    {
+      "entropy": 0.5412425100803375,
+      "epoch": 1.5233644859813085,
+      "grad_norm": 0.031013811007142067,
+      "learning_rate": 0.0002,
+      "loss": 0.5350072383880615,
+      "mean_token_accuracy": 0.7846681326627731,
+      "num_tokens": 6651176.0,
+      "step": 408
+    },
+    {
+      "entropy": 0.5564695447683334,
+      "epoch": 1.5271028037383179,
+      "grad_norm": 0.029242711141705513,
+      "learning_rate": 0.0002,
+      "loss": 0.5514513254165649,
+      "mean_token_accuracy": 0.7769511342048645,
+      "num_tokens": 6667349.0,
+      "step": 409
+    },
+    {
+      "entropy": 0.5430487841367722,
+      "epoch": 1.5308411214953273,
+      "grad_norm": 0.034237101674079895,
+      "learning_rate": 0.0002,
+      "loss": 0.5410158038139343,
+      "mean_token_accuracy": 0.7824828028678894,
+      "num_tokens": 6683579.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.540764793753624,
+      "epoch": 1.5345794392523364,
+      "grad_norm": 0.033801931887865067,
+      "learning_rate": 0.0002,
+      "loss": 0.5481416583061218,
+      "mean_token_accuracy": 0.7755506336688995,
+      "num_tokens": 6700088.0,
+      "step": 411
+    },
+    {
+      "entropy": 0.5592468529939651,
+      "epoch": 1.5383177570093458,
+      "grad_norm": 0.03298957645893097,
+      "learning_rate": 0.0002,
+      "loss": 0.568514347076416,
+      "mean_token_accuracy": 0.768169105052948,
+      "num_tokens": 6716293.0,
+      "step": 412
+    },
+    {
+      "entropy": 0.5506596267223358,
+      "epoch": 1.542056074766355,
+      "grad_norm": 0.02886568009853363,
+      "learning_rate": 0.0002,
+      "loss": 0.5483353137969971,
+      "mean_token_accuracy": 0.7779833972454071,
+      "num_tokens": 6732868.0,
+      "step": 413
+    },
+    {
+      "entropy": 0.5599238723516464,
+      "epoch": 1.5457943925233644,
+      "grad_norm": 0.030115319415926933,
+      "learning_rate": 0.0002,
+      "loss": 0.5546934008598328,
+      "mean_token_accuracy": 0.7749194204807281,
+      "num_tokens": 6749345.0,
+      "step": 414
+    },
+    {
+      "entropy": 0.5543588250875473,
+      "epoch": 1.5495327102803738,
+      "grad_norm": 0.0297012310475111,
+      "learning_rate": 0.0002,
+      "loss": 0.5521113276481628,
+      "mean_token_accuracy": 0.7790784388780594,
+      "num_tokens": 6765790.0,
+      "step": 415
+    },
+    {
+      "entropy": 0.5414643287658691,
+      "epoch": 1.5532710280373832,
+      "grad_norm": 0.031022896990180016,
+      "learning_rate": 0.0002,
+      "loss": 0.5424535870552063,
+      "mean_token_accuracy": 0.7811086177825928,
+      "num_tokens": 6781960.0,
+      "step": 416
+    },
+    {
+      "entropy": 0.5437638312578201,
+      "epoch": 1.5570093457943925,
+      "grad_norm": 0.03145059570670128,
+      "learning_rate": 0.0002,
+      "loss": 0.5443586707115173,
+      "mean_token_accuracy": 0.7771809101104736,
+      "num_tokens": 6798095.0,
+      "step": 417
+    },
+    {
+      "entropy": 0.5528093725442886,
+      "epoch": 1.560747663551402,
+      "grad_norm": 0.03482895717024803,
+      "learning_rate": 0.0002,
+      "loss": 0.5612707138061523,
+      "mean_token_accuracy": 0.769527867436409,
+      "num_tokens": 6814364.0,
+      "step": 418
+    },
+    {
+      "entropy": 0.5420306473970413,
+      "epoch": 1.5644859813084113,
+      "grad_norm": 0.028292428702116013,
+      "learning_rate": 0.0002,
+      "loss": 0.5386024117469788,
+      "mean_token_accuracy": 0.779121145606041,
+      "num_tokens": 6830638.0,
+      "step": 419
+    },
+    {
+      "entropy": 0.5656863749027252,
+      "epoch": 1.5682242990654207,
+      "grad_norm": 0.03678108751773834,
+      "learning_rate": 0.0002,
+      "loss": 0.556216835975647,
+      "mean_token_accuracy": 0.7773893773555756,
+      "num_tokens": 6846999.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.5444925278425217,
+      "epoch": 1.5719626168224299,
+      "grad_norm": 0.027032526209950447,
+      "learning_rate": 0.0002,
+      "loss": 0.5408859848976135,
+      "mean_token_accuracy": 0.779949277639389,
+      "num_tokens": 6863623.0,
+      "step": 421
+    },
+    {
+      "entropy": 0.5538263916969299,
+      "epoch": 1.5757009345794393,
+      "grad_norm": 0.030989233404397964,
+      "learning_rate": 0.0002,
+      "loss": 0.5511465072631836,
+      "mean_token_accuracy": 0.7773707062005997,
+      "num_tokens": 6880063.0,
+      "step": 422
+    },
+    {
+      "entropy": 0.5507351458072662,
+      "epoch": 1.5794392523364484,
+      "grad_norm": 0.031385209411382675,
+      "learning_rate": 0.0002,
+      "loss": 0.5568031668663025,
+      "mean_token_accuracy": 0.7725684642791748,
+      "num_tokens": 6896314.0,
+      "step": 423
+    },
+    {
+      "entropy": 0.5659656077623367,
+      "epoch": 1.5831775700934578,
+      "grad_norm": 0.03111201897263527,
+      "learning_rate": 0.0002,
+      "loss": 0.5699700713157654,
+      "mean_token_accuracy": 0.7707180082798004,
+      "num_tokens": 6912699.0,
+      "step": 424
+    },
+    {
+      "entropy": 0.544001892209053,
+      "epoch": 1.5869158878504672,
+      "grad_norm": 0.028569631278514862,
+      "learning_rate": 0.0002,
+      "loss": 0.5452186465263367,
+      "mean_token_accuracy": 0.7768339067697525,
+      "num_tokens": 6928999.0,
+      "step": 425
+    },
+    {
+      "entropy": 0.5471579134464264,
+      "epoch": 1.5906542056074766,
+      "grad_norm": 0.0320747047662735,
+      "learning_rate": 0.0002,
+      "loss": 0.5472685694694519,
+      "mean_token_accuracy": 0.7775517106056213,
+      "num_tokens": 6945430.0,
+      "step": 426
+    },
+    {
+      "entropy": 0.5361240059137344,
+      "epoch": 1.594392523364486,
+      "grad_norm": 0.030863573774695396,
+      "learning_rate": 0.0002,
+      "loss": 0.5366886258125305,
+      "mean_token_accuracy": 0.7811662554740906,
+      "num_tokens": 6961737.0,
+      "step": 427
+    },
+    {
+      "entropy": 0.5404235273599625,
+      "epoch": 1.5981308411214954,
+      "grad_norm": 0.03127690777182579,
+      "learning_rate": 0.0002,
+      "loss": 0.542918860912323,
+      "mean_token_accuracy": 0.7802147716283798,
+      "num_tokens": 6978071.0,
+      "step": 428
+    },
+    {
+      "entropy": 0.5443604737520218,
+      "epoch": 1.6018691588785048,
+      "grad_norm": 0.0312102772295475,
+      "learning_rate": 0.0002,
+      "loss": 0.5430488586425781,
+      "mean_token_accuracy": 0.7779473662376404,
+      "num_tokens": 6994482.0,
+      "step": 429
+    },
+    {
+      "entropy": 0.5391067713499069,
+      "epoch": 1.6056074766355142,
+      "grad_norm": 0.030182059854269028,
+      "learning_rate": 0.0002,
+      "loss": 0.5479601621627808,
+      "mean_token_accuracy": 0.7749738842248917,
+      "num_tokens": 7010796.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.5425884425640106,
+      "epoch": 1.6093457943925233,
+      "grad_norm": 0.03449074178934097,
+      "learning_rate": 0.0002,
+      "loss": 0.5474182367324829,
+      "mean_token_accuracy": 0.7784157991409302,
+      "num_tokens": 7026912.0,
+      "step": 431
+    },
+    {
+      "entropy": 0.5485401153564453,
+      "epoch": 1.6130841121495327,
+      "grad_norm": 0.028809968382120132,
+      "learning_rate": 0.0002,
+      "loss": 0.545950710773468,
+      "mean_token_accuracy": 0.7787201404571533,
+      "num_tokens": 7043303.0,
+      "step": 432
+    },
+    {
+      "entropy": 0.5581236183643341,
+      "epoch": 1.616822429906542,
+      "grad_norm": 0.03381406515836716,
+      "learning_rate": 0.0002,
+      "loss": 0.551964521408081,
+      "mean_token_accuracy": 0.7750234007835388,
+      "num_tokens": 7059825.0,
+      "step": 433
+    },
+    {
+      "entropy": 0.5686533004045486,
+      "epoch": 1.6205607476635513,
+      "grad_norm": 0.03113553300499916,
+      "learning_rate": 0.0002,
+      "loss": 0.565824031829834,
+      "mean_token_accuracy": 0.7702126502990723,
+      "num_tokens": 7076166.0,
+      "step": 434
+    },
+    {
+      "entropy": 0.541614830493927,
+      "epoch": 1.6242990654205607,
+      "grad_norm": 0.02829281985759735,
+      "learning_rate": 0.0002,
+      "loss": 0.5373748540878296,
+      "mean_token_accuracy": 0.7802039682865143,
+      "num_tokens": 7092453.0,
+      "step": 435
+    },
+    {
+      "entropy": 0.5370374619960785,
+      "epoch": 1.62803738317757,
+      "grad_norm": 0.038676466792821884,
+      "learning_rate": 0.0002,
+      "loss": 0.5463578701019287,
+      "mean_token_accuracy": 0.777619257569313,
+      "num_tokens": 7108489.0,
+      "step": 436
+    },
+    {
+      "entropy": 0.5232041031122208,
+      "epoch": 1.6317757009345795,
+      "grad_norm": 0.03186805173754692,
+      "learning_rate": 0.0002,
+      "loss": 0.5275849103927612,
+      "mean_token_accuracy": 0.7832721769809723,
+      "num_tokens": 7124721.0,
+      "step": 437
+    },
+    {
+      "entropy": 0.553793340921402,
+      "epoch": 1.6355140186915889,
+      "grad_norm": 0.034783560782670975,
+      "learning_rate": 0.0002,
+      "loss": 0.5545384287834167,
+      "mean_token_accuracy": 0.7748959958553314,
+      "num_tokens": 7141085.0,
+      "step": 438
+    },
+    {
+      "entropy": 0.5291094183921814,
+      "epoch": 1.6392523364485982,
+      "grad_norm": 0.032007865607738495,
+      "learning_rate": 0.0002,
+      "loss": 0.5233455896377563,
+      "mean_token_accuracy": 0.7880895286798477,
+      "num_tokens": 7157324.0,
+      "step": 439
+    },
+    {
+      "entropy": 0.5520578771829605,
+      "epoch": 1.6429906542056076,
+      "grad_norm": 0.03354523330926895,
+      "learning_rate": 0.0002,
+      "loss": 0.546679675579071,
+      "mean_token_accuracy": 0.7785679250955582,
+      "num_tokens": 7173748.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.556230291724205,
+      "epoch": 1.6467289719626168,
+      "grad_norm": 0.029538605362176895,
+      "learning_rate": 0.0002,
+      "loss": 0.5578120946884155,
+      "mean_token_accuracy": 0.7732451260089874,
+      "num_tokens": 7190279.0,
+      "step": 441
+    },
+    {
+      "entropy": 0.5558031052350998,
+      "epoch": 1.6504672897196262,
+      "grad_norm": 0.032964110374450684,
+      "learning_rate": 0.0002,
+      "loss": 0.5642295479774475,
+      "mean_token_accuracy": 0.7702266126871109,
+      "num_tokens": 7206573.0,
+      "step": 442
+    },
+    {
+      "entropy": 0.5462649762630463,
+      "epoch": 1.6542056074766354,
+      "grad_norm": 0.03030145727097988,
+      "learning_rate": 0.0002,
+      "loss": 0.5472926497459412,
+      "mean_token_accuracy": 0.7779358178377151,
+      "num_tokens": 7222876.0,
+      "step": 443
+    },
+    {
+      "entropy": 0.5564112961292267,
+      "epoch": 1.6579439252336448,
+      "grad_norm": 0.03184789791703224,
+      "learning_rate": 0.0002,
+      "loss": 0.5510616898536682,
+      "mean_token_accuracy": 0.778466135263443,
+      "num_tokens": 7239236.0,
+      "step": 444
+    },
+    {
+      "entropy": 0.5630057454109192,
+      "epoch": 1.6616822429906541,
+      "grad_norm": 0.029552895575761795,
+      "learning_rate": 0.0002,
+      "loss": 0.5598853826522827,
+      "mean_token_accuracy": 0.7684319913387299,
+      "num_tokens": 7255686.0,
+      "step": 445
+    },
+    {
+      "entropy": 0.5392823964357376,
+      "epoch": 1.6654205607476635,
+      "grad_norm": 0.03244130313396454,
+      "learning_rate": 0.0002,
+      "loss": 0.535351574420929,
+      "mean_token_accuracy": 0.782013326883316,
+      "num_tokens": 7271954.0,
+      "step": 446
+    },
+    {
+      "entropy": 0.5588325262069702,
+      "epoch": 1.669158878504673,
+      "grad_norm": 0.03356410935521126,
+      "learning_rate": 0.0002,
+      "loss": 0.5625864267349243,
+      "mean_token_accuracy": 0.7675231099128723,
+      "num_tokens": 7288380.0,
+      "step": 447
+    },
+    {
+      "entropy": 0.5461699217557907,
+      "epoch": 1.6728971962616823,
+      "grad_norm": 0.03438694775104523,
+      "learning_rate": 0.0002,
+      "loss": 0.5539066195487976,
+      "mean_token_accuracy": 0.7729915380477905,
+      "num_tokens": 7304727.0,
+      "step": 448
+    },
+    {
+      "entropy": 0.5323803722858429,
+      "epoch": 1.6766355140186917,
+      "grad_norm": 0.031167054548859596,
+      "learning_rate": 0.0002,
+      "loss": 0.5316429138183594,
+      "mean_token_accuracy": 0.7824341952800751,
+      "num_tokens": 7321168.0,
+      "step": 449
+    },
+    {
+      "entropy": 0.5499107241630554,
+      "epoch": 1.680373831775701,
+      "grad_norm": 0.02761506475508213,
+      "learning_rate": 0.0002,
+      "loss": 0.5486537218093872,
+      "mean_token_accuracy": 0.7810203582048416,
+      "num_tokens": 7337539.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.5529857277870178,
+      "epoch": 1.6841121495327103,
+      "grad_norm": 0.027268333360552788,
+      "learning_rate": 0.0002,
+      "loss": 0.5520575642585754,
+      "mean_token_accuracy": 0.7759742587804794,
+      "num_tokens": 7353841.0,
+      "step": 451
+    },
+    {
+      "entropy": 0.5497375279664993,
+      "epoch": 1.6878504672897197,
+      "grad_norm": 0.03056921809911728,
+      "learning_rate": 0.0002,
+      "loss": 0.5536624789237976,
+      "mean_token_accuracy": 0.7728510946035385,
+      "num_tokens": 7370087.0,
+      "step": 452
+    },
+    {
+      "entropy": 0.5587235540151596,
+      "epoch": 1.6915887850467288,
+      "grad_norm": 0.029484033584594727,
+      "learning_rate": 0.0002,
+      "loss": 0.5550922751426697,
+      "mean_token_accuracy": 0.7721101641654968,
+      "num_tokens": 7386564.0,
+      "step": 453
+    },
+    {
+      "entropy": 0.5490273237228394,
+      "epoch": 1.6953271028037382,
+      "grad_norm": 0.030445117503404617,
+      "learning_rate": 0.0002,
+      "loss": 0.5525585412979126,
+      "mean_token_accuracy": 0.7748779058456421,
+      "num_tokens": 7402899.0,
+      "step": 454
+    },
+    {
+      "entropy": 0.5428823530673981,
+      "epoch": 1.6990654205607476,
+      "grad_norm": 0.03236573934555054,
+      "learning_rate": 0.0002,
+      "loss": 0.5434162020683289,
+      "mean_token_accuracy": 0.7772506028413773,
+      "num_tokens": 7419333.0,
+      "step": 455
+    },
+    {
+      "entropy": 0.5431548655033112,
+      "epoch": 1.702803738317757,
+      "grad_norm": 0.02851344272494316,
+      "learning_rate": 0.0002,
+      "loss": 0.548797607421875,
+      "mean_token_accuracy": 0.7761964499950409,
+      "num_tokens": 7435551.0,
+      "step": 456
+    },
+    {
+      "entropy": 0.5437126010656357,
+      "epoch": 1.7065420560747664,
+      "grad_norm": 0.028895648196339607,
+      "learning_rate": 0.0002,
+      "loss": 0.5451984405517578,
+      "mean_token_accuracy": 0.7779933959245682,
+      "num_tokens": 7451825.0,
+      "step": 457
+    },
+    {
+      "entropy": 0.5545783340930939,
+      "epoch": 1.7102803738317758,
+      "grad_norm": 0.029966024681925774,
+      "learning_rate": 0.0002,
+      "loss": 0.5465367436408997,
+      "mean_token_accuracy": 0.7784044593572617,
+      "num_tokens": 7468353.0,
+      "step": 458
+    },
+    {
+      "entropy": 0.5588552802801132,
+      "epoch": 1.7140186915887852,
+      "grad_norm": 0.03131483122706413,
+      "learning_rate": 0.0002,
+      "loss": 0.5527791976928711,
+      "mean_token_accuracy": 0.7737509310245514,
+      "num_tokens": 7484697.0,
+      "step": 459
+    },
+    {
+      "entropy": 0.544754758477211,
+      "epoch": 1.7177570093457943,
+      "grad_norm": 0.032681871205568314,
+      "learning_rate": 0.0002,
+      "loss": 0.544547975063324,
+      "mean_token_accuracy": 0.7755615264177322,
+      "num_tokens": 7500910.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.5452838689088821,
+      "epoch": 1.7214953271028037,
+      "grad_norm": 0.03389411419630051,
+      "learning_rate": 0.0002,
+      "loss": 0.5545741319656372,
+      "mean_token_accuracy": 0.7773826718330383,
+      "num_tokens": 7517229.0,
+      "step": 461
+    },
+    {
+      "entropy": 0.5361888408660889,
+      "epoch": 1.7252336448598131,
+      "grad_norm": 0.03197981417179108,
+      "learning_rate": 0.0002,
+      "loss": 0.5412315726280212,
+      "mean_token_accuracy": 0.7807294726371765,
+      "num_tokens": 7533496.0,
+      "step": 462
+    },
+    {
+      "entropy": 0.5551300644874573,
+      "epoch": 1.7289719626168223,
+      "grad_norm": 0.029581068083643913,
+      "learning_rate": 0.0002,
+      "loss": 0.5573985576629639,
+      "mean_token_accuracy": 0.7714389115571976,
+      "num_tokens": 7549945.0,
+      "step": 463
+    },
+    {
+      "entropy": 0.561688169836998,
+      "epoch": 1.7327102803738317,
+      "grad_norm": 0.06545328348875046,
+      "learning_rate": 0.0002,
+      "loss": 0.5578382611274719,
+      "mean_token_accuracy": 0.7731240689754486,
+      "num_tokens": 7566295.0,
+      "step": 464
+    },
+    {
+      "entropy": 0.546341672539711,
+      "epoch": 1.736448598130841,
+      "grad_norm": 0.033630188554525375,
+      "learning_rate": 0.0002,
+      "loss": 0.5406599044799805,
+      "mean_token_accuracy": 0.7778750658035278,
+      "num_tokens": 7582430.0,
+      "step": 465
+    },
+    {
+      "entropy": 0.5360140800476074,
+      "epoch": 1.7401869158878505,
+      "grad_norm": 0.0722091943025589,
+      "learning_rate": 0.0002,
+      "loss": 0.5544889569282532,
+      "mean_token_accuracy": 0.7800410985946655,
+      "num_tokens": 7598724.0,
+      "step": 466
+    },
+    {
+      "entropy": 0.5511822551488876,
+      "epoch": 1.7439252336448599,
+      "grad_norm": 0.0453655868768692,
+      "learning_rate": 0.0002,
+      "loss": 0.5510369539260864,
+      "mean_token_accuracy": 0.7753463685512543,
+      "num_tokens": 7615066.0,
+      "step": 467
+    },
+    {
+      "entropy": 0.5692480951547623,
+      "epoch": 1.7476635514018692,
+      "grad_norm": 0.03852085396647453,
+      "learning_rate": 0.0002,
+      "loss": 0.5623142123222351,
+      "mean_token_accuracy": 0.7731426060199738,
+      "num_tokens": 7631470.0,
+      "step": 468
+    },
+    {
+      "entropy": 0.5454059541225433,
+      "epoch": 1.7514018691588786,
+      "grad_norm": 0.03272124007344246,
+      "learning_rate": 0.0002,
+      "loss": 0.5412269830703735,
+      "mean_token_accuracy": 0.7835355550050735,
+      "num_tokens": 7647653.0,
+      "step": 469
+    },
+    {
+      "entropy": 0.5629090666770935,
+      "epoch": 1.7551401869158878,
+      "grad_norm": 0.03476416692137718,
+      "learning_rate": 0.0002,
+      "loss": 0.5631899833679199,
+      "mean_token_accuracy": 0.7717723101377487,
+      "num_tokens": 7664168.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.5542246699333191,
+      "epoch": 1.7588785046728972,
+      "grad_norm": 0.04324596747756004,
+      "learning_rate": 0.0002,
+      "loss": 0.5525742769241333,
+      "mean_token_accuracy": 0.7750708311796188,
+      "num_tokens": 7680515.0,
+      "step": 471
+    },
+    {
+      "entropy": 0.5454518496990204,
+      "epoch": 1.7626168224299066,
+      "grad_norm": 0.029530908912420273,
+      "learning_rate": 0.0002,
+      "loss": 0.5429126024246216,
+      "mean_token_accuracy": 0.7794791609048843,
+      "num_tokens": 7696701.0,
+      "step": 472
+    },
+    {
+      "entropy": 0.537155881524086,
+      "epoch": 1.7663551401869158,
+      "grad_norm": 0.03453315421938896,
+      "learning_rate": 0.0002,
+      "loss": 0.5391392111778259,
+      "mean_token_accuracy": 0.7809355705976486,
+      "num_tokens": 7712858.0,
+      "step": 473
+    },
+    {
+      "entropy": 0.5679727643728256,
+      "epoch": 1.7700934579439251,
+      "grad_norm": 0.033509086817502975,
+      "learning_rate": 0.0002,
+      "loss": 0.5633649826049805,
+      "mean_token_accuracy": 0.7738876938819885,
+      "num_tokens": 7729346.0,
+      "step": 474
+    },
+    {
+      "entropy": 0.5437932014465332,
+      "epoch": 1.7738317757009345,
+      "grad_norm": 0.03146712854504585,
+      "learning_rate": 0.0002,
+      "loss": 0.5468876361846924,
+      "mean_token_accuracy": 0.7783123552799225,
+      "num_tokens": 7745581.0,
+      "step": 475
+    },
+    {
+      "entropy": 0.5403594225645065,
+      "epoch": 1.777570093457944,
+      "grad_norm": 0.03930533304810524,
+      "learning_rate": 0.0002,
+      "loss": 0.5513323545455933,
+      "mean_token_accuracy": 0.7714991271495819,
+      "num_tokens": 7761661.0,
+      "step": 476
+    },
+    {
+      "entropy": 0.5275924056768417,
+      "epoch": 1.7813084112149533,
+      "grad_norm": 0.029612882062792778,
+      "learning_rate": 0.0002,
+      "loss": 0.5271086096763611,
+      "mean_token_accuracy": 0.7869246900081635,
+      "num_tokens": 7777776.0,
+      "step": 477
+    },
+    {
+      "entropy": 0.5542174875736237,
+      "epoch": 1.7850467289719627,
+      "grad_norm": 0.03353600203990936,
+      "learning_rate": 0.0002,
+      "loss": 0.5515888333320618,
+      "mean_token_accuracy": 0.7757923156023026,
+      "num_tokens": 7794208.0,
+      "step": 478
+    },
+    {
+      "entropy": 0.5686136484146118,
+      "epoch": 1.788785046728972,
+      "grad_norm": 0.0359848290681839,
+      "learning_rate": 0.0002,
+      "loss": 0.5625109672546387,
+      "mean_token_accuracy": 0.770489439368248,
+      "num_tokens": 7810470.0,
+      "step": 479
+    },
+    {
+      "entropy": 0.5591467618942261,
+      "epoch": 1.7925233644859813,
+      "grad_norm": 0.03523039445281029,
+      "learning_rate": 0.0002,
+      "loss": 0.5659374594688416,
+      "mean_token_accuracy": 0.7701008170843124,
+      "num_tokens": 7826783.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.5457279831171036,
+      "epoch": 1.7962616822429907,
+      "grad_norm": 0.03323599696159363,
+      "learning_rate": 0.0002,
+      "loss": 0.5534178614616394,
+      "mean_token_accuracy": 0.7753042280673981,
+      "num_tokens": 7843052.0,
+      "step": 481
+    },
+    {
+      "entropy": 0.5539824962615967,
+      "epoch": 1.8,
+      "grad_norm": 0.03477081283926964,
+      "learning_rate": 0.0002,
+      "loss": 0.5533367395401001,
+      "mean_token_accuracy": 0.7745995223522186,
+      "num_tokens": 7859471.0,
+      "step": 482
+    },
+    {
+      "entropy": 0.5645311027765274,
+      "epoch": 1.8037383177570092,
+      "grad_norm": 0.03863651677966118,
+      "learning_rate": 0.0002,
+      "loss": 0.5656211376190186,
+      "mean_token_accuracy": 0.7659904807806015,
+      "num_tokens": 7875935.0,
+      "step": 483
+    },
+    {
+      "entropy": 0.5520930737257004,
+      "epoch": 1.8074766355140186,
+      "grad_norm": 0.030585726723074913,
+      "learning_rate": 0.0002,
+      "loss": 0.5455970168113708,
+      "mean_token_accuracy": 0.7777505069971085,
+      "num_tokens": 7892341.0,
+      "step": 484
+    },
+    {
+      "entropy": 0.5476150959730148,
+      "epoch": 1.811214953271028,
+      "grad_norm": 0.03229457512497902,
+      "learning_rate": 0.0002,
+      "loss": 0.5434688329696655,
+      "mean_token_accuracy": 0.7792231887578964,
+      "num_tokens": 7908634.0,
+      "step": 485
+    },
+    {
+      "entropy": 0.5417618900537491,
+      "epoch": 1.8149532710280374,
+      "grad_norm": 0.036823540925979614,
+      "learning_rate": 0.0002,
+      "loss": 0.5497291684150696,
+      "mean_token_accuracy": 0.7776121497154236,
+      "num_tokens": 7924838.0,
+      "step": 486
+    },
+    {
+      "entropy": 0.5513382405042648,
+      "epoch": 1.8186915887850468,
+      "grad_norm": 0.03352766111493111,
+      "learning_rate": 0.0002,
+      "loss": 0.5549485683441162,
+      "mean_token_accuracy": 0.7709111571311951,
+      "num_tokens": 7941280.0,
+      "step": 487
+    },
+    {
+      "entropy": 0.5313259810209274,
+      "epoch": 1.8224299065420562,
+      "grad_norm": 0.03318050131201744,
+      "learning_rate": 0.0002,
+      "loss": 0.5327860116958618,
+      "mean_token_accuracy": 0.7831867635250092,
+      "num_tokens": 7957629.0,
+      "step": 488
+    },
+    {
+      "entropy": 0.5544183105230331,
+      "epoch": 1.8261682242990656,
+      "grad_norm": 0.031777020543813705,
+      "learning_rate": 0.0002,
+      "loss": 0.5501781702041626,
+      "mean_token_accuracy": 0.7757555842399597,
+      "num_tokens": 7974007.0,
+      "step": 489
+    },
+    {
+      "entropy": 0.5466530621051788,
+      "epoch": 1.8299065420560747,
+      "grad_norm": 0.03351164236664772,
+      "learning_rate": 0.0002,
+      "loss": 0.5517051815986633,
+      "mean_token_accuracy": 0.7755817025899887,
+      "num_tokens": 7990357.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.5637551099061966,
+      "epoch": 1.8336448598130841,
+      "grad_norm": 0.029702039435505867,
+      "learning_rate": 0.0002,
+      "loss": 0.5635112524032593,
+      "mean_token_accuracy": 0.7732028961181641,
+      "num_tokens": 8006595.0,
+      "step": 491
+    },
+    {
+      "entropy": 0.5468285977840424,
+      "epoch": 1.8373831775700935,
+      "grad_norm": 0.03702588006854057,
+      "learning_rate": 0.0002,
+      "loss": 0.5488995909690857,
+      "mean_token_accuracy": 0.7771299928426743,
+      "num_tokens": 8022970.0,
+      "step": 492
+    },
+    {
+      "entropy": 0.5305586457252502,
+      "epoch": 1.8411214953271027,
+      "grad_norm": 0.03400636091828346,
+      "learning_rate": 0.0002,
+      "loss": 0.5370339155197144,
+      "mean_token_accuracy": 0.7821398675441742,
+      "num_tokens": 8038831.0,
+      "step": 493
+    },
+    {
+      "entropy": 0.5582829564809799,
+      "epoch": 1.844859813084112,
+      "grad_norm": 0.034549374133348465,
+      "learning_rate": 0.0002,
+      "loss": 0.5647696852684021,
+      "mean_token_accuracy": 0.7711978554725647,
+      "num_tokens": 8055116.0,
+      "step": 494
+    },
+    {
+      "entropy": 0.5546866208314896,
+      "epoch": 1.8485981308411215,
+      "grad_norm": 0.030837178230285645,
+      "learning_rate": 0.0002,
+      "loss": 0.5502550601959229,
+      "mean_token_accuracy": 0.7736955285072327,
+      "num_tokens": 8071294.0,
+      "step": 495
+    },
+    {
+      "entropy": 0.5573176741600037,
+      "epoch": 1.8523364485981308,
+      "grad_norm": 0.030723506584763527,
+      "learning_rate": 0.0002,
+      "loss": 0.5504124164581299,
+      "mean_token_accuracy": 0.7768528610467911,
+      "num_tokens": 8087507.0,
+      "step": 496
+    },
+    {
+      "entropy": 0.5464990586042404,
+      "epoch": 1.8560747663551402,
+      "grad_norm": 0.031615249812603,
+      "learning_rate": 0.0002,
+      "loss": 0.5419277548789978,
+      "mean_token_accuracy": 0.7811318039894104,
+      "num_tokens": 8103739.0,
+      "step": 497
+    },
+    {
+      "entropy": 0.5507507473230362,
+      "epoch": 1.8598130841121496,
+      "grad_norm": 0.031241271644830704,
+      "learning_rate": 0.0002,
+      "loss": 0.5549343824386597,
+      "mean_token_accuracy": 0.7766365706920624,
+      "num_tokens": 8120195.0,
+      "step": 498
+    },
+    {
+      "entropy": 0.528255894780159,
+      "epoch": 1.863551401869159,
+      "grad_norm": 0.03131619840860367,
+      "learning_rate": 0.0002,
+      "loss": 0.5327234268188477,
+      "mean_token_accuracy": 0.7834469825029373,
+      "num_tokens": 8136237.0,
+      "step": 499
+    },
+    {
+      "entropy": 0.5606546849012375,
+      "epoch": 1.8672897196261682,
+      "grad_norm": 0.029506012797355652,
+      "learning_rate": 0.0002,
+      "loss": 0.5612553954124451,
+      "mean_token_accuracy": 0.7706436961889267,
+      "num_tokens": 8152593.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.5505183190107346,
+      "epoch": 1.8710280373831776,
+      "grad_norm": 0.027479592710733414,
+      "learning_rate": 0.0002,
+      "loss": 0.5496048331260681,
+      "mean_token_accuracy": 0.7758743315935135,
+      "num_tokens": 8168904.0,
+      "step": 501
+    },
+    {
+      "entropy": 0.5521711856126785,
+      "epoch": 1.874766355140187,
+      "grad_norm": 0.029544338583946228,
+      "learning_rate": 0.0002,
+      "loss": 0.5500134825706482,
+      "mean_token_accuracy": 0.7751173824071884,
+      "num_tokens": 8185076.0,
+      "step": 502
+    },
+    {
+      "entropy": 0.5535303503274918,
+      "epoch": 1.8785046728971961,
+      "grad_norm": 0.03169948607683182,
+      "learning_rate": 0.0002,
+      "loss": 0.5551643967628479,
+      "mean_token_accuracy": 0.7735145539045334,
+      "num_tokens": 8201211.0,
+      "step": 503
+    },
+    {
+      "entropy": 0.5526206642389297,
+      "epoch": 1.8822429906542055,
+      "grad_norm": 0.029640454798936844,
+      "learning_rate": 0.0002,
+      "loss": 0.5438597202301025,
+      "mean_token_accuracy": 0.7798311114311218,
+      "num_tokens": 8217307.0,
+      "step": 504
+    },
+    {
+      "entropy": 0.5524756759405136,
+      "epoch": 1.885981308411215,
+      "grad_norm": 0.036887627094984055,
+      "learning_rate": 0.0002,
+      "loss": 0.5504904985427856,
+      "mean_token_accuracy": 0.7746451199054718,
+      "num_tokens": 8233547.0,
+      "step": 505
+    },
+    {
+      "entropy": 0.5518786162137985,
+      "epoch": 1.8897196261682243,
+      "grad_norm": 0.03335992246866226,
+      "learning_rate": 0.0002,
+      "loss": 0.5517621040344238,
+      "mean_token_accuracy": 0.7776313275098801,
+      "num_tokens": 8249917.0,
+      "step": 506
+    },
+    {
+      "entropy": 0.5617891550064087,
+      "epoch": 1.8934579439252337,
+      "grad_norm": 0.032016802579164505,
+      "learning_rate": 0.0002,
+      "loss": 0.5628353357315063,
+      "mean_token_accuracy": 0.7710386514663696,
+      "num_tokens": 8266109.0,
+      "step": 507
+    },
+    {
+      "entropy": 0.542267769575119,
+      "epoch": 1.897196261682243,
+      "grad_norm": 0.030068468302488327,
+      "learning_rate": 0.0002,
+      "loss": 0.5358614921569824,
+      "mean_token_accuracy": 0.7818213999271393,
+      "num_tokens": 8282620.0,
+      "step": 508
+    },
+    {
+      "entropy": 0.5325959101319313,
+      "epoch": 1.9009345794392525,
+      "grad_norm": 0.026692470535635948,
+      "learning_rate": 0.0002,
+      "loss": 0.5332909822463989,
+      "mean_token_accuracy": 0.7836195677518845,
+      "num_tokens": 8299102.0,
+      "step": 509
+    },
+    {
+      "entropy": 0.5525154024362564,
+      "epoch": 1.9046728971962616,
+      "grad_norm": 0.029202645644545555,
+      "learning_rate": 0.0002,
+      "loss": 0.5564330816268921,
+      "mean_token_accuracy": 0.7734202891588211,
+      "num_tokens": 8315257.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.5515794008970261,
+      "epoch": 1.908411214953271,
+      "grad_norm": 0.03252064809203148,
+      "learning_rate": 0.0002,
+      "loss": 0.5536593198776245,
+      "mean_token_accuracy": 0.7732348889112473,
+      "num_tokens": 8331546.0,
+      "step": 511
+    },
+    {
+      "entropy": 0.548397034406662,
+      "epoch": 1.9121495327102802,
+      "grad_norm": 0.02674887888133526,
+      "learning_rate": 0.0002,
+      "loss": 0.5501974821090698,
+      "mean_token_accuracy": 0.7732282876968384,
+      "num_tokens": 8347949.0,
+      "step": 512
+    },
+    {
+      "entropy": 0.5453347116708755,
+      "epoch": 1.9158878504672896,
+      "grad_norm": 0.031093353405594826,
+      "learning_rate": 0.0002,
+      "loss": 0.5438382625579834,
+      "mean_token_accuracy": 0.7786152511835098,
+      "num_tokens": 8364391.0,
+      "step": 513
+    },
+    {
+      "entropy": 0.5397911220788956,
+      "epoch": 1.919626168224299,
+      "grad_norm": 0.03151440620422363,
+      "learning_rate": 0.0002,
+      "loss": 0.5433223247528076,
+      "mean_token_accuracy": 0.782426580786705,
+      "num_tokens": 8380708.0,
+      "step": 514
+    },
+    {
+      "entropy": 0.5369860827922821,
+      "epoch": 1.9233644859813084,
+      "grad_norm": 0.035960424691438675,
+      "learning_rate": 0.0002,
+      "loss": 0.5434173345565796,
+      "mean_token_accuracy": 0.7791251987218857,
+      "num_tokens": 8397064.0,
+      "step": 515
+    },
+    {
+      "entropy": 0.5706264823675156,
+      "epoch": 1.9271028037383178,
+      "grad_norm": 0.032570187002420425,
+      "learning_rate": 0.0002,
+      "loss": 0.5725105404853821,
+      "mean_token_accuracy": 0.7670982778072357,
+      "num_tokens": 8413343.0,
+      "step": 516
+    },
+    {
+      "entropy": 0.5545293837785721,
+      "epoch": 1.9308411214953272,
+      "grad_norm": 0.02958519756793976,
+      "learning_rate": 0.0002,
+      "loss": 0.5506482124328613,
+      "mean_token_accuracy": 0.7761605083942413,
+      "num_tokens": 8429612.0,
+      "step": 517
+    },
+    {
+      "entropy": 0.5615670382976532,
+      "epoch": 1.9345794392523366,
+      "grad_norm": 0.033964622765779495,
+      "learning_rate": 0.0002,
+      "loss": 0.5493220686912537,
+      "mean_token_accuracy": 0.7761300206184387,
+      "num_tokens": 8446212.0,
+      "step": 518
+    },
+    {
+      "entropy": 0.5410606861114502,
+      "epoch": 1.938317757009346,
+      "grad_norm": 0.03190803527832031,
+      "learning_rate": 0.0002,
+      "loss": 0.5400397181510925,
+      "mean_token_accuracy": 0.7820769846439362,
+      "num_tokens": 8462334.0,
+      "step": 519
+    },
+    {
+      "entropy": 0.5581596940755844,
+      "epoch": 1.9420560747663551,
+      "grad_norm": 0.03331439197063446,
+      "learning_rate": 0.0002,
+      "loss": 0.5643476843833923,
+      "mean_token_accuracy": 0.7706414759159088,
+      "num_tokens": 8478855.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.5354294329881668,
+      "epoch": 1.9457943925233645,
+      "grad_norm": 0.03460956737399101,
+      "learning_rate": 0.0002,
+      "loss": 0.5405067801475525,
+      "mean_token_accuracy": 0.7803503423929214,
+      "num_tokens": 8495299.0,
+      "step": 521
+    },
+    {
+      "entropy": 0.5444121658802032,
+      "epoch": 1.9495327102803737,
+      "grad_norm": 0.03110821172595024,
+      "learning_rate": 0.0002,
+      "loss": 0.5491779446601868,
+      "mean_token_accuracy": 0.7754928916692734,
+      "num_tokens": 8512049.0,
+      "step": 522
+    },
+    {
+      "entropy": 0.5544265806674957,
+      "epoch": 1.953271028037383,
+      "grad_norm": 0.03355475887656212,
+      "learning_rate": 0.0002,
+      "loss": 0.5517175197601318,
+      "mean_token_accuracy": 0.7744346410036087,
+      "num_tokens": 8528391.0,
+      "step": 523
+    },
+    {
+      "entropy": 0.5565292239189148,
+      "epoch": 1.9570093457943925,
+      "grad_norm": 0.029388394206762314,
+      "learning_rate": 0.0002,
+      "loss": 0.5525711178779602,
+      "mean_token_accuracy": 0.7771426290273666,
+      "num_tokens": 8544951.0,
+      "step": 524
+    },
+    {
+      "entropy": 0.5562525689601898,
+      "epoch": 1.9607476635514018,
+      "grad_norm": 0.03225899115204811,
+      "learning_rate": 0.0002,
+      "loss": 0.5521144866943359,
+      "mean_token_accuracy": 0.7747148424386978,
+      "num_tokens": 8561319.0,
+      "step": 525
+    },
+    {
+      "entropy": 0.5570588409900665,
+      "epoch": 1.9644859813084112,
+      "grad_norm": 0.03741025924682617,
+      "learning_rate": 0.0002,
+      "loss": 0.557031512260437,
+      "mean_token_accuracy": 0.7722350209951401,
+      "num_tokens": 8577515.0,
+      "step": 526
+    },
+    {
+      "entropy": 0.5351222157478333,
+      "epoch": 1.9682242990654206,
+      "grad_norm": 0.029640430584549904,
+      "learning_rate": 0.0002,
+      "loss": 0.5402463674545288,
+      "mean_token_accuracy": 0.7800589948892593,
+      "num_tokens": 8593870.0,
+      "step": 527
+    },
+    {
+      "entropy": 0.5618735998868942,
+      "epoch": 1.97196261682243,
+      "grad_norm": 0.03617403656244278,
+      "learning_rate": 0.0002,
+      "loss": 0.5702189803123474,
+      "mean_token_accuracy": 0.7701396048069,
+      "num_tokens": 8610566.0,
+      "step": 528
+    },
+    {
+      "entropy": 0.5421914011240005,
+      "epoch": 1.9757009345794394,
+      "grad_norm": 0.0348467081785202,
+      "learning_rate": 0.0002,
+      "loss": 0.5410540699958801,
+      "mean_token_accuracy": 0.7805227339267731,
+      "num_tokens": 8626999.0,
+      "step": 529
+    },
+    {
+      "entropy": 0.5460681766271591,
+      "epoch": 1.9794392523364486,
+      "grad_norm": 0.03002713993191719,
+      "learning_rate": 0.0002,
+      "loss": 0.5420626997947693,
+      "mean_token_accuracy": 0.7775750756263733,
+      "num_tokens": 8643379.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.5560031235218048,
+      "epoch": 1.983177570093458,
+      "grad_norm": 0.02867712639272213,
+      "learning_rate": 0.0002,
+      "loss": 0.5502682328224182,
+      "mean_token_accuracy": 0.7767003029584885,
+      "num_tokens": 8659814.0,
+      "step": 531
+    },
+    {
+      "entropy": 0.5697564780712128,
+      "epoch": 1.9869158878504671,
+      "grad_norm": 0.034316740930080414,
+      "learning_rate": 0.0002,
+      "loss": 0.5676635503768921,
+      "mean_token_accuracy": 0.7674341350793839,
+      "num_tokens": 8676132.0,
+      "step": 532
+    },
+    {
+      "entropy": 0.5442752093076706,
+      "epoch": 1.9906542056074765,
+      "grad_norm": 0.032586514949798584,
+      "learning_rate": 0.0002,
+      "loss": 0.5461133718490601,
+      "mean_token_accuracy": 0.7766181975603104,
+      "num_tokens": 8692605.0,
+      "step": 533
+    },
+    {
+      "entropy": 0.5408132523298264,
+      "epoch": 1.994392523364486,
+      "grad_norm": 0.028621118515729904,
+      "learning_rate": 0.0002,
+      "loss": 0.5469282865524292,
+      "mean_token_accuracy": 0.7724520564079285,
+      "num_tokens": 8708747.0,
+      "step": 534
+    },
+    {
+      "entropy": 0.5602825433015823,
+      "epoch": 1.9981308411214953,
+      "grad_norm": 0.030758850276470184,
+      "learning_rate": 0.0002,
+      "loss": 0.561021625995636,
+      "mean_token_accuracy": 0.7700503617525101,
+      "num_tokens": 8725013.0,
+      "step": 535
+    },
+    {
+      "entropy": 0.5262570530176163,
+      "epoch": 2.0,
+      "grad_norm": 0.041106510907411575,
+      "learning_rate": 0.0002,
+      "loss": 0.5266788601875305,
+      "mean_token_accuracy": 0.7866178452968597,
+      "num_tokens": 8730698.0,
+      "step": 536
+    },
+    {
+      "entropy": 0.5550900399684906,
+      "epoch": 2.0037383177570094,
+      "grad_norm": 0.031247610226273537,
+      "learning_rate": 0.0002,
+      "loss": 0.5427648425102234,
+      "mean_token_accuracy": 0.7791824787855148,
+      "num_tokens": 8747042.0,
+      "step": 537
+    },
+    {
+      "entropy": 0.5459851175546646,
+      "epoch": 2.007476635514019,
+      "grad_norm": 0.03372490033507347,
+      "learning_rate": 0.0002,
+      "loss": 0.5406405925750732,
+      "mean_token_accuracy": 0.7772052884101868,
+      "num_tokens": 8763334.0,
+      "step": 538
+    },
+    {
+      "entropy": 0.5324701964855194,
+      "epoch": 2.011214953271028,
+      "grad_norm": 0.035976309329271317,
+      "learning_rate": 0.0002,
+      "loss": 0.5377114415168762,
+      "mean_token_accuracy": 0.7791852504014969,
+      "num_tokens": 8779449.0,
+      "step": 539
+    },
+    {
+      "entropy": 0.5410480201244354,
+      "epoch": 2.0149532710280376,
+      "grad_norm": 0.03491639345884323,
+      "learning_rate": 0.0002,
+      "loss": 0.546131432056427,
+      "mean_token_accuracy": 0.7801298946142197,
+      "num_tokens": 8795796.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.5370665192604065,
+      "epoch": 2.0186915887850465,
+      "grad_norm": 0.04541780799627304,
+      "learning_rate": 0.0002,
+      "loss": 0.5394240617752075,
+      "mean_token_accuracy": 0.7802438586950302,
+      "num_tokens": 8812239.0,
+      "step": 541
+    },
+    {
+      "entropy": 0.5596486777067184,
+      "epoch": 2.022429906542056,
+      "grad_norm": 0.03784856200218201,
+      "learning_rate": 0.0002,
+      "loss": 0.5519159436225891,
+      "mean_token_accuracy": 0.7757291346788406,
+      "num_tokens": 8828405.0,
+      "step": 542
+    },
+    {
+      "entropy": 0.5454950630664825,
+      "epoch": 2.0261682242990653,
+      "grad_norm": 0.044418152421712875,
+      "learning_rate": 0.0002,
+      "loss": 0.5402126908302307,
+      "mean_token_accuracy": 0.7813611626625061,
+      "num_tokens": 8844636.0,
+      "step": 543
+    },
+    {
+      "entropy": 0.5182391032576561,
+      "epoch": 2.0299065420560747,
+      "grad_norm": 0.03811662271618843,
+      "learning_rate": 0.0002,
+      "loss": 0.5167620778083801,
+      "mean_token_accuracy": 0.7919700294733047,
+      "num_tokens": 8860774.0,
+      "step": 544
+    },
+    {
+      "entropy": 0.5337931364774704,
+      "epoch": 2.033644859813084,
+      "grad_norm": 0.04127589613199234,
+      "learning_rate": 0.0002,
+      "loss": 0.5322132110595703,
+      "mean_token_accuracy": 0.78387551009655,
+      "num_tokens": 8876930.0,
+      "step": 545
+    },
+    {
+      "entropy": 0.5331794023513794,
+      "epoch": 2.0373831775700935,
+      "grad_norm": 0.03687538579106331,
+      "learning_rate": 0.0002,
+      "loss": 0.5393625497817993,
+      "mean_token_accuracy": 0.7809632271528244,
+      "num_tokens": 8893177.0,
+      "step": 546
+    },
+    {
+      "entropy": 0.5235881805419922,
+      "epoch": 2.041121495327103,
+      "grad_norm": 0.04371653124690056,
+      "learning_rate": 0.0002,
+      "loss": 0.5326187610626221,
+      "mean_token_accuracy": 0.7816834598779678,
+      "num_tokens": 8909522.0,
+      "step": 547
+    },
+    {
+      "entropy": 0.5280764102935791,
+      "epoch": 2.0448598130841122,
+      "grad_norm": 0.03337106481194496,
+      "learning_rate": 0.0002,
+      "loss": 0.531906247138977,
+      "mean_token_accuracy": 0.7849632948637009,
+      "num_tokens": 8925662.0,
+      "step": 548
+    },
+    {
+      "entropy": 0.5238441824913025,
+      "epoch": 2.0485981308411216,
+      "grad_norm": 0.04188013821840286,
+      "learning_rate": 0.0002,
+      "loss": 0.5119596123695374,
+      "mean_token_accuracy": 0.7923747897148132,
+      "num_tokens": 8942074.0,
+      "step": 549
+    },
+    {
+      "entropy": 0.5489809960126877,
+      "epoch": 2.052336448598131,
+      "grad_norm": 0.03609222173690796,
+      "learning_rate": 0.0002,
+      "loss": 0.5403028726577759,
+      "mean_token_accuracy": 0.7791068702936172,
+      "num_tokens": 8958329.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.5437890142202377,
+      "epoch": 2.05607476635514,
+      "grad_norm": 0.03745613619685173,
+      "learning_rate": 0.0002,
+      "loss": 0.5450887084007263,
+      "mean_token_accuracy": 0.7779961079359055,
+      "num_tokens": 8974422.0,
+      "step": 551
+    },
+    {
+      "entropy": 0.5333922803401947,
+      "epoch": 2.0598130841121494,
+      "grad_norm": 0.042939819395542145,
+      "learning_rate": 0.0002,
+      "loss": 0.5377407670021057,
+      "mean_token_accuracy": 0.7828031182289124,
+      "num_tokens": 8990622.0,
+      "step": 552
+    },
+    {
+      "entropy": 0.5367026776075363,
+      "epoch": 2.0635514018691588,
+      "grad_norm": 0.03213382139801979,
+      "learning_rate": 0.0002,
+      "loss": 0.5339721441268921,
+      "mean_token_accuracy": 0.7822674959897995,
+      "num_tokens": 9006957.0,
+      "step": 553
+    },
+    {
+      "entropy": 0.5521921962499619,
+      "epoch": 2.067289719626168,
+      "grad_norm": 0.043271131813526154,
+      "learning_rate": 0.0002,
+      "loss": 0.5421251654624939,
+      "mean_token_accuracy": 0.7799997925758362,
+      "num_tokens": 9023385.0,
+      "step": 554
+    },
+    {
+      "entropy": 0.536955714225769,
+      "epoch": 2.0710280373831775,
+      "grad_norm": 0.036944299936294556,
+      "learning_rate": 0.0002,
+      "loss": 0.53342604637146,
+      "mean_token_accuracy": 0.7826140820980072,
+      "num_tokens": 9039979.0,
+      "step": 555
+    },
+    {
+      "entropy": 0.5285215377807617,
+      "epoch": 2.074766355140187,
+      "grad_norm": 0.03607345372438431,
+      "learning_rate": 0.0002,
+      "loss": 0.5354245901107788,
+      "mean_token_accuracy": 0.7834082543849945,
+      "num_tokens": 9056411.0,
+      "step": 556
+    },
+    {
+      "entropy": 0.5265194773674011,
+      "epoch": 2.0785046728971963,
+      "grad_norm": 0.03231218829751015,
+      "learning_rate": 0.0002,
+      "loss": 0.5345092415809631,
+      "mean_token_accuracy": 0.7815968543291092,
+      "num_tokens": 9072925.0,
+      "step": 557
+    },
+    {
+      "entropy": 0.528659924864769,
+      "epoch": 2.0822429906542057,
+      "grad_norm": 0.0367112010717392,
+      "learning_rate": 0.0002,
+      "loss": 0.5276532173156738,
+      "mean_token_accuracy": 0.7867192178964615,
+      "num_tokens": 9089043.0,
+      "step": 558
+    },
+    {
+      "entropy": 0.5355315059423447,
+      "epoch": 2.085981308411215,
+      "grad_norm": 0.03834950551390648,
+      "learning_rate": 0.0002,
+      "loss": 0.5311352014541626,
+      "mean_token_accuracy": 0.7842549979686737,
+      "num_tokens": 9105540.0,
+      "step": 559
+    },
+    {
+      "entropy": 0.5209446102380753,
+      "epoch": 2.0897196261682245,
+      "grad_norm": 0.03739333152770996,
+      "learning_rate": 0.0002,
+      "loss": 0.5185686945915222,
+      "mean_token_accuracy": 0.7894220799207687,
+      "num_tokens": 9121845.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.521641194820404,
+      "epoch": 2.0934579439252334,
+      "grad_norm": 0.04100764915347099,
+      "learning_rate": 0.0002,
+      "loss": 0.5293187499046326,
+      "mean_token_accuracy": 0.7825994938611984,
+      "num_tokens": 9138058.0,
+      "step": 561
+    },
+    {
+      "entropy": 0.5365670919418335,
+      "epoch": 2.097196261682243,
+      "grad_norm": 0.036154504865407944,
+      "learning_rate": 0.0002,
+      "loss": 0.5414365530014038,
+      "mean_token_accuracy": 0.780719056725502,
+      "num_tokens": 9154400.0,
+      "step": 562
+    },
+    {
+      "entropy": 0.5273338481783867,
+      "epoch": 2.100934579439252,
+      "grad_norm": 0.04028065875172615,
+      "learning_rate": 0.0002,
+      "loss": 0.529691755771637,
+      "mean_token_accuracy": 0.7834713310003281,
+      "num_tokens": 9170963.0,
+      "step": 563
+    },
+    {
+      "entropy": 0.5385439097881317,
+      "epoch": 2.1046728971962616,
+      "grad_norm": 0.03361035883426666,
+      "learning_rate": 0.0002,
+      "loss": 0.5270369052886963,
+      "mean_token_accuracy": 0.7854906022548676,
+      "num_tokens": 9187292.0,
+      "step": 564
+    },
+    {
+      "entropy": 0.5487733483314514,
+      "epoch": 2.108411214953271,
+      "grad_norm": 0.03783544525504112,
+      "learning_rate": 0.0002,
+      "loss": 0.5383955240249634,
+      "mean_token_accuracy": 0.7841213643550873,
+      "num_tokens": 9203564.0,
+      "step": 565
+    },
+    {
+      "entropy": 0.5291211605072021,
+      "epoch": 2.1121495327102804,
+      "grad_norm": 0.037037670612335205,
+      "learning_rate": 0.0002,
+      "loss": 0.5292587280273438,
+      "mean_token_accuracy": 0.7856062203645706,
+      "num_tokens": 9219726.0,
+      "step": 566
+    },
+    {
+      "entropy": 0.5350653678178787,
+      "epoch": 2.1158878504672898,
+      "grad_norm": 0.045786142349243164,
+      "learning_rate": 0.0002,
+      "loss": 0.5461608171463013,
+      "mean_token_accuracy": 0.7790984213352203,
+      "num_tokens": 9235906.0,
+      "step": 567
+    },
+    {
+      "entropy": 0.5272955000400543,
+      "epoch": 2.119626168224299,
+      "grad_norm": 0.03238094225525856,
+      "learning_rate": 0.0002,
+      "loss": 0.5275669097900391,
+      "mean_token_accuracy": 0.7858364135026932,
+      "num_tokens": 9252138.0,
+      "step": 568
+    },
+    {
+      "entropy": 0.5282166749238968,
+      "epoch": 2.1233644859813086,
+      "grad_norm": 0.040294334292411804,
+      "learning_rate": 0.0002,
+      "loss": 0.5235229730606079,
+      "mean_token_accuracy": 0.7853135764598846,
+      "num_tokens": 9268479.0,
+      "step": 569
+    },
+    {
+      "entropy": 0.5460007339715958,
+      "epoch": 2.127102803738318,
+      "grad_norm": 0.033559415489435196,
+      "learning_rate": 0.0002,
+      "loss": 0.543584406375885,
+      "mean_token_accuracy": 0.780407503247261,
+      "num_tokens": 9284881.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.5487277060747147,
+      "epoch": 2.130841121495327,
+      "grad_norm": 0.04158215597271919,
+      "learning_rate": 0.0002,
+      "loss": 0.5564124584197998,
+      "mean_token_accuracy": 0.7759140133857727,
+      "num_tokens": 9301347.0,
+      "step": 571
+    },
+    {
+      "entropy": 0.5200594365596771,
+      "epoch": 2.1345794392523363,
+      "grad_norm": 0.03448987007141113,
+      "learning_rate": 0.0002,
+      "loss": 0.5191783905029297,
+      "mean_token_accuracy": 0.7882550060749054,
+      "num_tokens": 9317467.0,
+      "step": 572
+    },
+    {
+      "entropy": 0.5137363150715828,
+      "epoch": 2.1383177570093457,
+      "grad_norm": 0.04049127548933029,
+      "learning_rate": 0.0002,
+      "loss": 0.5194275975227356,
+      "mean_token_accuracy": 0.7879694402217865,
+      "num_tokens": 9333343.0,
+      "step": 573
+    },
+    {
+      "entropy": 0.5200390294194221,
+      "epoch": 2.142056074766355,
+      "grad_norm": 0.036729056388139725,
+      "learning_rate": 0.0002,
+      "loss": 0.5214616060256958,
+      "mean_token_accuracy": 0.788168340921402,
+      "num_tokens": 9349842.0,
+      "step": 574
+    },
+    {
+      "entropy": 0.5292311161756516,
+      "epoch": 2.1457943925233645,
+      "grad_norm": 0.0415414534509182,
+      "learning_rate": 0.0002,
+      "loss": 0.5297327041625977,
+      "mean_token_accuracy": 0.7849121540784836,
+      "num_tokens": 9366012.0,
+      "step": 575
+    },
+    {
+      "entropy": 0.5386081859469414,
+      "epoch": 2.149532710280374,
+      "grad_norm": 0.03256456181406975,
+      "learning_rate": 0.0002,
+      "loss": 0.5289144515991211,
+      "mean_token_accuracy": 0.7840749174356461,
+      "num_tokens": 9382392.0,
+      "step": 576
+    },
+    {
+      "entropy": 0.5395008027553558,
+      "epoch": 2.1532710280373832,
+      "grad_norm": 0.04578758776187897,
+      "learning_rate": 0.0002,
+      "loss": 0.5399198532104492,
+      "mean_token_accuracy": 0.7795116752386093,
+      "num_tokens": 9398608.0,
+      "step": 577
+    },
+    {
+      "entropy": 0.5388733297586441,
+      "epoch": 2.1570093457943926,
+      "grad_norm": 0.04348094016313553,
+      "learning_rate": 0.0002,
+      "loss": 0.5477968454360962,
+      "mean_token_accuracy": 0.7775707393884659,
+      "num_tokens": 9414864.0,
+      "step": 578
+    },
+    {
+      "entropy": 0.5495359748601913,
+      "epoch": 2.160747663551402,
+      "grad_norm": 0.04143773764371872,
+      "learning_rate": 0.0002,
+      "loss": 0.5492925643920898,
+      "mean_token_accuracy": 0.7751609981060028,
+      "num_tokens": 9431018.0,
+      "step": 579
+    },
+    {
+      "entropy": 0.5342540368437767,
+      "epoch": 2.1644859813084114,
+      "grad_norm": 0.0339248850941658,
+      "learning_rate": 0.0002,
+      "loss": 0.5319566130638123,
+      "mean_token_accuracy": 0.7832492738962173,
+      "num_tokens": 9447324.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.5374208092689514,
+      "epoch": 2.1682242990654204,
+      "grad_norm": 0.04187169671058655,
+      "learning_rate": 0.0002,
+      "loss": 0.532558023929596,
+      "mean_token_accuracy": 0.7825674563646317,
+      "num_tokens": 9463630.0,
+      "step": 581
+    },
+    {
+      "entropy": 0.5422970801591873,
+      "epoch": 2.1719626168224297,
+      "grad_norm": 0.035590704530477524,
+      "learning_rate": 0.0002,
+      "loss": 0.5381810665130615,
+      "mean_token_accuracy": 0.7824431657791138,
+      "num_tokens": 9480262.0,
+      "step": 582
+    },
+    {
+      "entropy": 0.5083005726337433,
+      "epoch": 2.175700934579439,
+      "grad_norm": 0.04746020957827568,
+      "learning_rate": 0.0002,
+      "loss": 0.5185438990592957,
+      "mean_token_accuracy": 0.7896369099617004,
+      "num_tokens": 9496413.0,
+      "step": 583
+    },
+    {
+      "entropy": 0.538268506526947,
+      "epoch": 2.1794392523364485,
+      "grad_norm": 0.036502279341220856,
+      "learning_rate": 0.0002,
+      "loss": 0.5346280336380005,
+      "mean_token_accuracy": 0.7869013249874115,
+      "num_tokens": 9512790.0,
+      "step": 584
+    },
+    {
+      "entropy": 0.5377851128578186,
+      "epoch": 2.183177570093458,
+      "grad_norm": 0.04309968277812004,
+      "learning_rate": 0.0002,
+      "loss": 0.5360051989555359,
+      "mean_token_accuracy": 0.7825826555490494,
+      "num_tokens": 9529078.0,
+      "step": 585
+    },
+    {
+      "entropy": 0.5475880205631256,
+      "epoch": 2.1869158878504673,
+      "grad_norm": 0.035980913788080215,
+      "learning_rate": 0.0002,
+      "loss": 0.5418691635131836,
+      "mean_token_accuracy": 0.7812779098749161,
+      "num_tokens": 9545620.0,
+      "step": 586
+    },
+    {
+      "entropy": 0.5415029525756836,
+      "epoch": 2.1906542056074767,
+      "grad_norm": 0.03402319550514221,
+      "learning_rate": 0.0002,
+      "loss": 0.5380699038505554,
+      "mean_token_accuracy": 0.7777233719825745,
+      "num_tokens": 9561847.0,
+      "step": 587
+    },
+    {
+      "entropy": 0.5402775406837463,
+      "epoch": 2.194392523364486,
+      "grad_norm": 0.03904155641794205,
+      "learning_rate": 0.0002,
+      "loss": 0.5427509546279907,
+      "mean_token_accuracy": 0.7772547751665115,
+      "num_tokens": 9578299.0,
+      "step": 588
+    },
+    {
+      "entropy": 0.5400314331054688,
+      "epoch": 2.1981308411214955,
+      "grad_norm": 0.040362391620874405,
+      "learning_rate": 0.0002,
+      "loss": 0.5459818840026855,
+      "mean_token_accuracy": 0.7782857865095139,
+      "num_tokens": 9594690.0,
+      "step": 589
+    },
+    {
+      "entropy": 0.5383201092481613,
+      "epoch": 2.201869158878505,
+      "grad_norm": 0.03448455408215523,
+      "learning_rate": 0.0002,
+      "loss": 0.5378321409225464,
+      "mean_token_accuracy": 0.7824592739343643,
+      "num_tokens": 9611075.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.5350049138069153,
+      "epoch": 2.205607476635514,
+      "grad_norm": 0.03910663723945618,
+      "learning_rate": 0.0002,
+      "loss": 0.5314739346504211,
+      "mean_token_accuracy": 0.7833352535963058,
+      "num_tokens": 9627390.0,
+      "step": 591
+    },
+    {
+      "entropy": 0.5416655540466309,
+      "epoch": 2.209345794392523,
+      "grad_norm": 0.035276249051094055,
+      "learning_rate": 0.0002,
+      "loss": 0.5419362187385559,
+      "mean_token_accuracy": 0.7800017446279526,
+      "num_tokens": 9643588.0,
+      "step": 592
+    },
+    {
+      "entropy": 0.5288861393928528,
+      "epoch": 2.2130841121495326,
+      "grad_norm": 0.03930996358394623,
+      "learning_rate": 0.0002,
+      "loss": 0.5276063680648804,
+      "mean_token_accuracy": 0.7874087691307068,
+      "num_tokens": 9660028.0,
+      "step": 593
+    },
+    {
+      "entropy": 0.5260862559080124,
+      "epoch": 2.216822429906542,
+      "grad_norm": 0.0427854023873806,
+      "learning_rate": 0.0002,
+      "loss": 0.5307914614677429,
+      "mean_token_accuracy": 0.7824005782604218,
+      "num_tokens": 9676398.0,
+      "step": 594
+    },
+    {
+      "entropy": 0.5367477387189865,
+      "epoch": 2.2205607476635514,
+      "grad_norm": 0.04066091775894165,
+      "learning_rate": 0.0002,
+      "loss": 0.5374104380607605,
+      "mean_token_accuracy": 0.7815362513065338,
+      "num_tokens": 9692902.0,
+      "step": 595
+    },
+    {
+      "entropy": 0.5327900499105453,
+      "epoch": 2.2242990654205608,
+      "grad_norm": 0.03865550830960274,
+      "learning_rate": 0.0002,
+      "loss": 0.5303129553794861,
+      "mean_token_accuracy": 0.7837819904088974,
+      "num_tokens": 9709300.0,
+      "step": 596
+    },
+    {
+      "entropy": 0.5307036936283112,
+      "epoch": 2.22803738317757,
+      "grad_norm": 0.03896172344684601,
+      "learning_rate": 0.0002,
+      "loss": 0.5255040526390076,
+      "mean_token_accuracy": 0.7894146144390106,
+      "num_tokens": 9725629.0,
+      "step": 597
+    },
+    {
+      "entropy": 0.555603951215744,
+      "epoch": 2.2317757009345796,
+      "grad_norm": 0.042428482323884964,
+      "learning_rate": 0.0002,
+      "loss": 0.5537799000740051,
+      "mean_token_accuracy": 0.7759463936090469,
+      "num_tokens": 9742301.0,
+      "step": 598
+    },
+    {
+      "entropy": 0.5576256364583969,
+      "epoch": 2.235514018691589,
+      "grad_norm": 0.0402006059885025,
+      "learning_rate": 0.0002,
+      "loss": 0.554506242275238,
+      "mean_token_accuracy": 0.7767222672700882,
+      "num_tokens": 9758260.0,
+      "step": 599
+    },
+    {
+      "entropy": 0.5121647417545319,
+      "epoch": 2.2392523364485983,
+      "grad_norm": 0.043138571083545685,
+      "learning_rate": 0.0002,
+      "loss": 0.5151476860046387,
+      "mean_token_accuracy": 0.790720209479332,
+      "num_tokens": 9774531.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.5213710218667984,
+      "epoch": 2.2429906542056073,
+      "grad_norm": 0.04871455207467079,
+      "learning_rate": 0.0002,
+      "loss": 0.5321851968765259,
+      "mean_token_accuracy": 0.785635232925415,
+      "num_tokens": 9790856.0,
+      "step": 601
+    },
+    {
+      "entropy": 0.5272484123706818,
+      "epoch": 2.2467289719626167,
+      "grad_norm": 0.040508877485990524,
+      "learning_rate": 0.0002,
+      "loss": 0.5326333045959473,
+      "mean_token_accuracy": 0.7852048426866531,
+      "num_tokens": 9807182.0,
+      "step": 602
+    },
+    {
+      "entropy": 0.5392047762870789,
+      "epoch": 2.250467289719626,
+      "grad_norm": 0.041053518652915955,
+      "learning_rate": 0.0002,
+      "loss": 0.5345852375030518,
+      "mean_token_accuracy": 0.7831785529851913,
+      "num_tokens": 9823474.0,
+      "step": 603
+    },
+    {
+      "entropy": 0.5716013610363007,
+      "epoch": 2.2542056074766355,
+      "grad_norm": 0.03574617952108383,
+      "learning_rate": 0.0002,
+      "loss": 0.5612589120864868,
+      "mean_token_accuracy": 0.7706786692142487,
+      "num_tokens": 9840032.0,
+      "step": 604
+    },
+    {
+      "entropy": 0.5285372212529182,
+      "epoch": 2.257943925233645,
+      "grad_norm": 0.03970465064048767,
+      "learning_rate": 0.0002,
+      "loss": 0.5292190909385681,
+      "mean_token_accuracy": 0.7823670506477356,
+      "num_tokens": 9856156.0,
+      "step": 605
+    },
+    {
+      "entropy": 0.5217142850160599,
+      "epoch": 2.2616822429906542,
+      "grad_norm": 0.04674631357192993,
+      "learning_rate": 0.0002,
+      "loss": 0.5359308123588562,
+      "mean_token_accuracy": 0.7826663255691528,
+      "num_tokens": 9872455.0,
+      "step": 606
+    },
+    {
+      "entropy": 0.5451616048812866,
+      "epoch": 2.2654205607476636,
+      "grad_norm": 0.03408370912075043,
+      "learning_rate": 0.0002,
+      "loss": 0.5420879125595093,
+      "mean_token_accuracy": 0.7793671786785126,
+      "num_tokens": 9888926.0,
+      "step": 607
+    },
+    {
+      "entropy": 0.5499323606491089,
+      "epoch": 2.269158878504673,
+      "grad_norm": 0.042992595583200455,
+      "learning_rate": 0.0002,
+      "loss": 0.5445036292076111,
+      "mean_token_accuracy": 0.7789521962404251,
+      "num_tokens": 9905440.0,
+      "step": 608
+    },
+    {
+      "entropy": 0.5252562239766121,
+      "epoch": 2.2728971962616824,
+      "grad_norm": 0.035366203635931015,
+      "learning_rate": 0.0002,
+      "loss": 0.5255159139633179,
+      "mean_token_accuracy": 0.7839477211236954,
+      "num_tokens": 9921478.0,
+      "step": 609
+    },
+    {
+      "entropy": 0.544396385550499,
+      "epoch": 2.2766355140186914,
+      "grad_norm": 0.03851408511400223,
+      "learning_rate": 0.0002,
+      "loss": 0.5386646389961243,
+      "mean_token_accuracy": 0.783714771270752,
+      "num_tokens": 9937700.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.5345716625452042,
+      "epoch": 2.2803738317757007,
+      "grad_norm": 0.03978222236037254,
+      "learning_rate": 0.0002,
+      "loss": 0.5340750813484192,
+      "mean_token_accuracy": 0.7818202525377274,
+      "num_tokens": 9953970.0,
+      "step": 611
+    },
+    {
+      "entropy": 0.5345568954944611,
+      "epoch": 2.28411214953271,
+      "grad_norm": 0.0438290499150753,
+      "learning_rate": 0.0002,
+      "loss": 0.5399565100669861,
+      "mean_token_accuracy": 0.7790825515985489,
+      "num_tokens": 9970196.0,
+      "step": 612
+    },
+    {
+      "entropy": 0.5261162966489792,
+      "epoch": 2.2878504672897195,
+      "grad_norm": 0.0412151962518692,
+      "learning_rate": 0.0002,
+      "loss": 0.5271809101104736,
+      "mean_token_accuracy": 0.7853071689605713,
+      "num_tokens": 9986604.0,
+      "step": 613
+    },
+    {
+      "entropy": 0.5332745313644409,
+      "epoch": 2.291588785046729,
+      "grad_norm": 0.03914888948202133,
+      "learning_rate": 0.0002,
+      "loss": 0.5373774766921997,
+      "mean_token_accuracy": 0.7806718051433563,
+      "num_tokens": 10003320.0,
+      "step": 614
+    },
+    {
+      "entropy": 0.5465565472841263,
+      "epoch": 2.2953271028037383,
+      "grad_norm": 0.036169324070215225,
+      "learning_rate": 0.0002,
+      "loss": 0.5407333970069885,
+      "mean_token_accuracy": 0.7835270166397095,
+      "num_tokens": 10019823.0,
+      "step": 615
+    },
+    {
+      "entropy": 0.5345954746007919,
+      "epoch": 2.2990654205607477,
+      "grad_norm": 0.0409700982272625,
+      "learning_rate": 0.0002,
+      "loss": 0.5321113467216492,
+      "mean_token_accuracy": 0.7842907607555389,
+      "num_tokens": 10036092.0,
+      "step": 616
+    },
+    {
+      "entropy": 0.5495668053627014,
+      "epoch": 2.302803738317757,
+      "grad_norm": 0.03986911475658417,
+      "learning_rate": 0.0002,
+      "loss": 0.5555334687232971,
+      "mean_token_accuracy": 0.7753788381814957,
+      "num_tokens": 10052693.0,
+      "step": 617
+    },
+    {
+      "entropy": 0.5311184674501419,
+      "epoch": 2.3065420560747665,
+      "grad_norm": 0.04703551530838013,
+      "learning_rate": 0.0002,
+      "loss": 0.5360814929008484,
+      "mean_token_accuracy": 0.7818758338689804,
+      "num_tokens": 10069207.0,
+      "step": 618
+    },
+    {
+      "entropy": 0.5312536582350731,
+      "epoch": 2.310280373831776,
+      "grad_norm": 0.042258135974407196,
+      "learning_rate": 0.0002,
+      "loss": 0.5263288021087646,
+      "mean_token_accuracy": 0.7859921306371689,
+      "num_tokens": 10085277.0,
+      "step": 619
+    },
+    {
+      "entropy": 0.5470355451107025,
+      "epoch": 2.3140186915887853,
+      "grad_norm": 0.03973059356212616,
+      "learning_rate": 0.0002,
+      "loss": 0.545122504234314,
+      "mean_token_accuracy": 0.7791093438863754,
+      "num_tokens": 10101813.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.5233904868364334,
+      "epoch": 2.317757009345794,
+      "grad_norm": 0.042754027992486954,
+      "learning_rate": 0.0002,
+      "loss": 0.5272444486618042,
+      "mean_token_accuracy": 0.7844137996435165,
+      "num_tokens": 10118006.0,
+      "step": 621
+    },
+    {
+      "entropy": 0.5061093345284462,
+      "epoch": 2.3214953271028036,
+      "grad_norm": 0.038454048335552216,
+      "learning_rate": 0.0002,
+      "loss": 0.5092731714248657,
+      "mean_token_accuracy": 0.7937048375606537,
+      "num_tokens": 10134010.0,
+      "step": 622
+    },
+    {
+      "entropy": 0.5244115591049194,
+      "epoch": 2.325233644859813,
+      "grad_norm": 0.04292070120573044,
+      "learning_rate": 0.0002,
+      "loss": 0.531722366809845,
+      "mean_token_accuracy": 0.7844948768615723,
+      "num_tokens": 10150394.0,
+      "step": 623
+    },
+    {
+      "entropy": 0.5242590308189392,
+      "epoch": 2.3289719626168224,
+      "grad_norm": 0.041382841765880585,
+      "learning_rate": 0.0002,
+      "loss": 0.5197622776031494,
+      "mean_token_accuracy": 0.7893139272928238,
+      "num_tokens": 10166971.0,
+      "step": 624
+    },
+    {
+      "entropy": 0.5171145796775818,
+      "epoch": 2.3327102803738318,
+      "grad_norm": 0.04326708987355232,
+      "learning_rate": 0.0002,
+      "loss": 0.5186761021614075,
+      "mean_token_accuracy": 0.7864463478326797,
+      "num_tokens": 10183110.0,
+      "step": 625
+    },
+    {
+      "entropy": 0.537827268242836,
+      "epoch": 2.336448598130841,
+      "grad_norm": 0.03789723291993141,
+      "learning_rate": 0.0002,
+      "loss": 0.533248245716095,
+      "mean_token_accuracy": 0.7839608788490295,
+      "num_tokens": 10199350.0,
+      "step": 626
+    },
+    {
+      "entropy": 0.5234524011611938,
+      "epoch": 2.3401869158878505,
+      "grad_norm": 0.04052973911166191,
+      "learning_rate": 0.0002,
+      "loss": 0.5173730254173279,
+      "mean_token_accuracy": 0.7916721105575562,
+      "num_tokens": 10215685.0,
+      "step": 627
+    },
+    {
+      "entropy": 0.5470897704362869,
+      "epoch": 2.34392523364486,
+      "grad_norm": 0.03785538300871849,
+      "learning_rate": 0.0002,
+      "loss": 0.5449895262718201,
+      "mean_token_accuracy": 0.7797056883573532,
+      "num_tokens": 10231984.0,
+      "step": 628
+    },
+    {
+      "entropy": 0.5209166556596756,
+      "epoch": 2.3476635514018693,
+      "grad_norm": 0.04008970037102699,
+      "learning_rate": 0.0002,
+      "loss": 0.5221812129020691,
+      "mean_token_accuracy": 0.7847004383802414,
+      "num_tokens": 10248210.0,
+      "step": 629
+    },
+    {
+      "entropy": 0.5186321288347244,
+      "epoch": 2.3514018691588783,
+      "grad_norm": 0.04004783183336258,
+      "learning_rate": 0.0002,
+      "loss": 0.5238394141197205,
+      "mean_token_accuracy": 0.7874705046415329,
+      "num_tokens": 10264273.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.5370597541332245,
+      "epoch": 2.3551401869158877,
+      "grad_norm": 0.04393518716096878,
+      "learning_rate": 0.0002,
+      "loss": 0.5417079925537109,
+      "mean_token_accuracy": 0.7808088809251785,
+      "num_tokens": 10280306.0,
+      "step": 631
+    },
+    {
+      "entropy": 0.5319319814443588,
+      "epoch": 2.358878504672897,
+      "grad_norm": 0.041104018688201904,
+      "learning_rate": 0.0002,
+      "loss": 0.5323211550712585,
+      "mean_token_accuracy": 0.7829127311706543,
+      "num_tokens": 10296761.0,
+      "step": 632
+    },
+    {
+      "entropy": 0.5280259251594543,
+      "epoch": 2.3626168224299064,
+      "grad_norm": 0.04328769072890282,
+      "learning_rate": 0.0002,
+      "loss": 0.5300282835960388,
+      "mean_token_accuracy": 0.7860163003206253,
+      "num_tokens": 10312760.0,
+      "step": 633
+    },
+    {
+      "entropy": 0.5409206449985504,
+      "epoch": 2.366355140186916,
+      "grad_norm": 0.03652770817279816,
+      "learning_rate": 0.0002,
+      "loss": 0.5389151573181152,
+      "mean_token_accuracy": 0.7825797498226166,
+      "num_tokens": 10328967.0,
+      "step": 634
+    },
+    {
+      "entropy": 0.5524943023920059,
+      "epoch": 2.3700934579439252,
+      "grad_norm": 0.03295084461569786,
+      "learning_rate": 0.0002,
+      "loss": 0.5457990169525146,
+      "mean_token_accuracy": 0.7791374921798706,
+      "num_tokens": 10345465.0,
+      "step": 635
+    },
+    {
+      "entropy": 0.5282382369041443,
+      "epoch": 2.3738317757009346,
+      "grad_norm": 0.04183012619614601,
+      "learning_rate": 0.0002,
+      "loss": 0.5293847322463989,
+      "mean_token_accuracy": 0.7866223007440567,
+      "num_tokens": 10361693.0,
+      "step": 636
+    },
+    {
+      "entropy": 0.5491131991147995,
+      "epoch": 2.377570093457944,
+      "grad_norm": 0.04285868629813194,
+      "learning_rate": 0.0002,
+      "loss": 0.5491060614585876,
+      "mean_token_accuracy": 0.7811785042285919,
+      "num_tokens": 10377893.0,
+      "step": 637
+    },
+    {
+      "entropy": 0.5452415496110916,
+      "epoch": 2.3813084112149534,
+      "grad_norm": 0.03759608045220375,
+      "learning_rate": 0.0002,
+      "loss": 0.542472779750824,
+      "mean_token_accuracy": 0.7795874774456024,
+      "num_tokens": 10394225.0,
+      "step": 638
+    },
+    {
+      "entropy": 0.5412954837083817,
+      "epoch": 2.385046728971963,
+      "grad_norm": 0.04048043116927147,
+      "learning_rate": 0.0002,
+      "loss": 0.5419274568557739,
+      "mean_token_accuracy": 0.7792576551437378,
+      "num_tokens": 10410592.0,
+      "step": 639
+    },
+    {
+      "entropy": 0.5429834425449371,
+      "epoch": 2.388785046728972,
+      "grad_norm": 0.03450307622551918,
+      "learning_rate": 0.0002,
+      "loss": 0.5450208187103271,
+      "mean_token_accuracy": 0.7783068269491196,
+      "num_tokens": 10427270.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.5542936772108078,
+      "epoch": 2.392523364485981,
+      "grad_norm": 0.038160763680934906,
+      "learning_rate": 0.0002,
+      "loss": 0.5524436831474304,
+      "mean_token_accuracy": 0.7759971767663956,
+      "num_tokens": 10443620.0,
+      "step": 641
+    },
+    {
+      "entropy": 0.5481744855642319,
+      "epoch": 2.3962616822429905,
+      "grad_norm": 0.045667361468076706,
+      "learning_rate": 0.0002,
+      "loss": 0.5538838505744934,
+      "mean_token_accuracy": 0.7768992632627487,
+      "num_tokens": 10459970.0,
+      "step": 642
+    },
+    {
+      "entropy": 0.5398150980472565,
+      "epoch": 2.4,
+      "grad_norm": 0.03911864385008812,
+      "learning_rate": 0.0002,
+      "loss": 0.5386427044868469,
+      "mean_token_accuracy": 0.7801983505487442,
+      "num_tokens": 10476045.0,
+      "step": 643
+    },
+    {
+      "entropy": 0.539577066898346,
+      "epoch": 2.4037383177570093,
+      "grad_norm": 0.038023967295885086,
+      "learning_rate": 0.0002,
+      "loss": 0.5344752669334412,
+      "mean_token_accuracy": 0.7824069559574127,
+      "num_tokens": 10492232.0,
+      "step": 644
+    },
+    {
+      "entropy": 0.5577429980039597,
+      "epoch": 2.4074766355140187,
+      "grad_norm": 0.04197937622666359,
+      "learning_rate": 0.0002,
+      "loss": 0.5572798252105713,
+      "mean_token_accuracy": 0.7743511646986008,
+      "num_tokens": 10508664.0,
+      "step": 645
+    },
+    {
+      "entropy": 0.5277721136808395,
+      "epoch": 2.411214953271028,
+      "grad_norm": 0.04219021648168564,
+      "learning_rate": 0.0002,
+      "loss": 0.5340980887413025,
+      "mean_token_accuracy": 0.784222811460495,
+      "num_tokens": 10524893.0,
+      "step": 646
+    },
+    {
+      "entropy": 0.5434385687112808,
+      "epoch": 2.4149532710280375,
+      "grad_norm": 0.04173292592167854,
+      "learning_rate": 0.0002,
+      "loss": 0.5448347330093384,
+      "mean_token_accuracy": 0.7814051806926727,
+      "num_tokens": 10541450.0,
+      "step": 647
+    },
+    {
+      "entropy": 0.542335107922554,
+      "epoch": 2.418691588785047,
+      "grad_norm": 0.04221229627728462,
+      "learning_rate": 0.0002,
+      "loss": 0.539567232131958,
+      "mean_token_accuracy": 0.778713047504425,
+      "num_tokens": 10557626.0,
+      "step": 648
+    },
+    {
+      "entropy": 0.5323275178670883,
+      "epoch": 2.4224299065420563,
+      "grad_norm": 0.04274986311793327,
+      "learning_rate": 0.0002,
+      "loss": 0.5353638529777527,
+      "mean_token_accuracy": 0.7840659618377686,
+      "num_tokens": 10573896.0,
+      "step": 649
+    },
+    {
+      "entropy": 0.53394415974617,
+      "epoch": 2.426168224299065,
+      "grad_norm": 0.03369283676147461,
+      "learning_rate": 0.0002,
+      "loss": 0.5374135971069336,
+      "mean_token_accuracy": 0.7838425934314728,
+      "num_tokens": 10590327.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.5415229946374893,
+      "epoch": 2.4299065420560746,
+      "grad_norm": 0.041201673448085785,
+      "learning_rate": 0.0002,
+      "loss": 0.5444363355636597,
+      "mean_token_accuracy": 0.7790240347385406,
+      "num_tokens": 10606679.0,
+      "step": 651
+    },
+    {
+      "entropy": 0.5405219197273254,
+      "epoch": 2.433644859813084,
+      "grad_norm": 0.03498642519116402,
+      "learning_rate": 0.0002,
+      "loss": 0.5373417139053345,
+      "mean_token_accuracy": 0.780651792883873,
+      "num_tokens": 10622823.0,
+      "step": 652
+    },
+    {
+      "entropy": 0.552605539560318,
+      "epoch": 2.4373831775700934,
+      "grad_norm": 0.035641010850667953,
+      "learning_rate": 0.0002,
+      "loss": 0.5453448295593262,
+      "mean_token_accuracy": 0.778374508023262,
+      "num_tokens": 10639286.0,
+      "step": 653
+    },
+    {
+      "entropy": 0.5386696010828018,
+      "epoch": 2.4411214953271028,
+      "grad_norm": 0.03597128391265869,
+      "learning_rate": 0.0002,
+      "loss": 0.5370036959648132,
+      "mean_token_accuracy": 0.7805056869983673,
+      "num_tokens": 10655659.0,
+      "step": 654
+    },
+    {
+      "entropy": 0.5298606008291245,
+      "epoch": 2.444859813084112,
+      "grad_norm": 0.044264063239097595,
+      "learning_rate": 0.0002,
+      "loss": 0.5362837314605713,
+      "mean_token_accuracy": 0.783015176653862,
+      "num_tokens": 10672206.0,
+      "step": 655
+    },
+    {
+      "entropy": 0.5373793393373489,
+      "epoch": 2.4485981308411215,
+      "grad_norm": 0.03741737827658653,
+      "learning_rate": 0.0002,
+      "loss": 0.5384023189544678,
+      "mean_token_accuracy": 0.7820392400026321,
+      "num_tokens": 10688480.0,
+      "step": 656
+    },
+    {
+      "entropy": 0.5373315960168839,
+      "epoch": 2.452336448598131,
+      "grad_norm": 0.03572740778326988,
+      "learning_rate": 0.0002,
+      "loss": 0.5341760516166687,
+      "mean_token_accuracy": 0.7831654995679855,
+      "num_tokens": 10704724.0,
+      "step": 657
+    },
+    {
+      "entropy": 0.5349675416946411,
+      "epoch": 2.4560747663551403,
+      "grad_norm": 0.033913351595401764,
+      "learning_rate": 0.0002,
+      "loss": 0.5286991000175476,
+      "mean_token_accuracy": 0.7848687022924423,
+      "num_tokens": 10720999.0,
+      "step": 658
+    },
+    {
+      "entropy": 0.5428586453199387,
+      "epoch": 2.4598130841121497,
+      "grad_norm": 0.03939831256866455,
+      "learning_rate": 0.0002,
+      "loss": 0.5347580313682556,
+      "mean_token_accuracy": 0.7854909747838974,
+      "num_tokens": 10737393.0,
+      "step": 659
+    },
+    {
+      "entropy": 0.5376365929841995,
+      "epoch": 2.463551401869159,
+      "grad_norm": 0.040565796196460724,
+      "learning_rate": 0.0002,
+      "loss": 0.5377070903778076,
+      "mean_token_accuracy": 0.7797142714262009,
+      "num_tokens": 10753719.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.5290745496749878,
+      "epoch": 2.467289719626168,
+      "grad_norm": 0.042844053357839584,
+      "learning_rate": 0.0002,
+      "loss": 0.5402673482894897,
+      "mean_token_accuracy": 0.7787514328956604,
+      "num_tokens": 10770023.0,
+      "step": 661
+    },
+    {
+      "entropy": 0.5171011686325073,
+      "epoch": 2.4710280373831774,
+      "grad_norm": 0.03879300504922867,
+      "learning_rate": 0.0002,
+      "loss": 0.5276088714599609,
+      "mean_token_accuracy": 0.7832213789224625,
+      "num_tokens": 10786225.0,
+      "step": 662
+    },
+    {
+      "entropy": 0.5425639152526855,
+      "epoch": 2.474766355140187,
+      "grad_norm": 0.039429496973752975,
+      "learning_rate": 0.0002,
+      "loss": 0.5366618633270264,
+      "mean_token_accuracy": 0.7817561626434326,
+      "num_tokens": 10802664.0,
+      "step": 663
+    },
+    {
+      "entropy": 0.5487925857305527,
+      "epoch": 2.4785046728971962,
+      "grad_norm": 0.037153951823711395,
+      "learning_rate": 0.0002,
+      "loss": 0.5378848314285278,
+      "mean_token_accuracy": 0.7842334508895874,
+      "num_tokens": 10818791.0,
+      "step": 664
+    },
+    {
+      "entropy": 0.5683897584676743,
+      "epoch": 2.4822429906542056,
+      "grad_norm": 0.037182942032814026,
+      "learning_rate": 0.0002,
+      "loss": 0.558560848236084,
+      "mean_token_accuracy": 0.7739170640707016,
+      "num_tokens": 10835365.0,
+      "step": 665
+    },
+    {
+      "entropy": 0.522308886051178,
+      "epoch": 2.485981308411215,
+      "grad_norm": 0.038330383598804474,
+      "learning_rate": 0.0002,
+      "loss": 0.5248692035675049,
+      "mean_token_accuracy": 0.786837100982666,
+      "num_tokens": 10851372.0,
+      "step": 666
+    },
+    {
+      "entropy": 0.5347889512777328,
+      "epoch": 2.4897196261682244,
+      "grad_norm": 0.0479077473282814,
+      "learning_rate": 0.0002,
+      "loss": 0.5449220538139343,
+      "mean_token_accuracy": 0.7782158553600311,
+      "num_tokens": 10867792.0,
+      "step": 667
+    },
+    {
+      "entropy": 0.5346285253763199,
+      "epoch": 2.493457943925234,
+      "grad_norm": 0.03586270660161972,
+      "learning_rate": 0.0002,
+      "loss": 0.5363122224807739,
+      "mean_token_accuracy": 0.7790896743535995,
+      "num_tokens": 10883867.0,
+      "step": 668
+    },
+    {
+      "entropy": 0.543928936123848,
+      "epoch": 2.497196261682243,
+      "grad_norm": 0.034230347722768784,
+      "learning_rate": 0.0002,
+      "loss": 0.5394496321678162,
+      "mean_token_accuracy": 0.7808743715286255,
+      "num_tokens": 10900252.0,
+      "step": 669
+    },
+    {
+      "entropy": 0.5318520069122314,
+      "epoch": 2.500934579439252,
+      "grad_norm": 0.03817044571042061,
+      "learning_rate": 0.0002,
+      "loss": 0.5254085063934326,
+      "mean_token_accuracy": 0.7874796390533447,
+      "num_tokens": 10916581.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.5420865267515182,
+      "epoch": 2.5046728971962615,
+      "grad_norm": 0.040148667991161346,
+      "learning_rate": 0.0002,
+      "loss": 0.5419960021972656,
+      "mean_token_accuracy": 0.7817145586013794,
+      "num_tokens": 10932966.0,
+      "step": 671
+    },
+    {
+      "entropy": 0.5323408842086792,
+      "epoch": 2.508411214953271,
+      "grad_norm": 0.03530094772577286,
+      "learning_rate": 0.0002,
+      "loss": 0.5375533103942871,
+      "mean_token_accuracy": 0.7799390703439713,
+      "num_tokens": 10949355.0,
+      "step": 672
+    },
+    {
+      "entropy": 0.5461747795343399,
+      "epoch": 2.5121495327102803,
+      "grad_norm": 0.044817496091127396,
+      "learning_rate": 0.0002,
+      "loss": 0.5525781512260437,
+      "mean_token_accuracy": 0.7765567153692245,
+      "num_tokens": 10965822.0,
+      "step": 673
+    },
+    {
+      "entropy": 0.5261489972472191,
+      "epoch": 2.5158878504672897,
+      "grad_norm": 0.03769567608833313,
+      "learning_rate": 0.0002,
+      "loss": 0.5308763980865479,
+      "mean_token_accuracy": 0.7855723053216934,
+      "num_tokens": 10982147.0,
+      "step": 674
+    },
+    {
+      "entropy": 0.5603571683168411,
+      "epoch": 2.519626168224299,
+      "grad_norm": 0.03521028161048889,
+      "learning_rate": 0.0002,
+      "loss": 0.5556987524032593,
+      "mean_token_accuracy": 0.7740853279829025,
+      "num_tokens": 10998452.0,
+      "step": 675
+    },
+    {
+      "entropy": 0.5460142344236374,
+      "epoch": 2.5233644859813085,
+      "grad_norm": 0.04217759519815445,
+      "learning_rate": 0.0002,
+      "loss": 0.537194013595581,
+      "mean_token_accuracy": 0.7822094410657883,
+      "num_tokens": 11014939.0,
+      "step": 676
+    },
+    {
+      "entropy": 0.5280861929059029,
+      "epoch": 2.527102803738318,
+      "grad_norm": 0.0387798547744751,
+      "learning_rate": 0.0002,
+      "loss": 0.5307282209396362,
+      "mean_token_accuracy": 0.7829688042402267,
+      "num_tokens": 11031382.0,
+      "step": 677
+    },
+    {
+      "entropy": 0.540501594543457,
+      "epoch": 2.5308411214953273,
+      "grad_norm": 0.04039468243718147,
+      "learning_rate": 0.0002,
+      "loss": 0.5451690554618835,
+      "mean_token_accuracy": 0.7794826477766037,
+      "num_tokens": 11047827.0,
+      "step": 678
+    },
+    {
+      "entropy": 0.5370010584592819,
+      "epoch": 2.5345794392523366,
+      "grad_norm": 0.04067126661539078,
+      "learning_rate": 0.0002,
+      "loss": 0.5422821044921875,
+      "mean_token_accuracy": 0.7765499353408813,
+      "num_tokens": 11063880.0,
+      "step": 679
+    },
+    {
+      "entropy": 0.5253852158784866,
+      "epoch": 2.538317757009346,
+      "grad_norm": 0.0398363396525383,
+      "learning_rate": 0.0002,
+      "loss": 0.5209527611732483,
+      "mean_token_accuracy": 0.7851253598928452,
+      "num_tokens": 11079962.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.5486203581094742,
+      "epoch": 2.542056074766355,
+      "grad_norm": 0.036812763661146164,
+      "learning_rate": 0.0002,
+      "loss": 0.5471652746200562,
+      "mean_token_accuracy": 0.7765830755233765,
+      "num_tokens": 11096121.0,
+      "step": 681
+    },
+    {
+      "entropy": 0.5477603673934937,
+      "epoch": 2.5457943925233644,
+      "grad_norm": 0.036881882697343826,
+      "learning_rate": 0.0002,
+      "loss": 0.5459142327308655,
+      "mean_token_accuracy": 0.7773014456033707,
+      "num_tokens": 11112554.0,
+      "step": 682
+    },
+    {
+      "entropy": 0.5485733300447464,
+      "epoch": 2.5495327102803738,
+      "grad_norm": 0.038766611367464066,
+      "learning_rate": 0.0002,
+      "loss": 0.5489373803138733,
+      "mean_token_accuracy": 0.7768855541944504,
+      "num_tokens": 11128948.0,
+      "step": 683
+    },
+    {
+      "entropy": 0.5454789996147156,
+      "epoch": 2.553271028037383,
+      "grad_norm": 0.03894040733575821,
+      "learning_rate": 0.0002,
+      "loss": 0.5461348295211792,
+      "mean_token_accuracy": 0.7770956754684448,
+      "num_tokens": 11145269.0,
+      "step": 684
+    },
+    {
+      "entropy": 0.5280539393424988,
+      "epoch": 2.5570093457943925,
+      "grad_norm": 0.03855814412236214,
+      "learning_rate": 0.0002,
+      "loss": 0.5288204550743103,
+      "mean_token_accuracy": 0.7839067131280899,
+      "num_tokens": 11161868.0,
+      "step": 685
+    },
+    {
+      "entropy": 0.5194002389907837,
+      "epoch": 2.560747663551402,
+      "grad_norm": 0.04192323610186577,
+      "learning_rate": 0.0002,
+      "loss": 0.5202341079711914,
+      "mean_token_accuracy": 0.7879376262426376,
+      "num_tokens": 11178075.0,
+      "step": 686
+    },
+    {
+      "entropy": 0.530809260904789,
+      "epoch": 2.5644859813084113,
+      "grad_norm": 0.0389430895447731,
+      "learning_rate": 0.0002,
+      "loss": 0.5309376120567322,
+      "mean_token_accuracy": 0.7828936278820038,
+      "num_tokens": 11194469.0,
+      "step": 687
+    },
+    {
+      "entropy": 0.5557933151721954,
+      "epoch": 2.5682242990654207,
+      "grad_norm": 0.04179178923368454,
+      "learning_rate": 0.0002,
+      "loss": 0.5567797422409058,
+      "mean_token_accuracy": 0.77239590883255,
+      "num_tokens": 11210862.0,
+      "step": 688
+    },
+    {
+      "entropy": 0.5501526147127151,
+      "epoch": 2.5719626168224297,
+      "grad_norm": 0.037524376064538956,
+      "learning_rate": 0.0002,
+      "loss": 0.5423179864883423,
+      "mean_token_accuracy": 0.7853735685348511,
+      "num_tokens": 11227307.0,
+      "step": 689
+    },
+    {
+      "entropy": 0.5341353267431259,
+      "epoch": 2.575700934579439,
+      "grad_norm": 0.03998028114438057,
+      "learning_rate": 0.0002,
+      "loss": 0.5339105129241943,
+      "mean_token_accuracy": 0.7838831543922424,
+      "num_tokens": 11243496.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.5265656411647797,
+      "epoch": 2.5794392523364484,
+      "grad_norm": 0.037673015147447586,
+      "learning_rate": 0.0002,
+      "loss": 0.532894492149353,
+      "mean_token_accuracy": 0.7829709053039551,
+      "num_tokens": 11259619.0,
+      "step": 691
+    },
+    {
+      "entropy": 0.5275788754224777,
+      "epoch": 2.583177570093458,
+      "grad_norm": 0.042666979134082794,
+      "learning_rate": 0.0002,
+      "loss": 0.5304786562919617,
+      "mean_token_accuracy": 0.786617174744606,
+      "num_tokens": 11275881.0,
+      "step": 692
+    },
+    {
+      "entropy": 0.5386812686920166,
+      "epoch": 2.586915887850467,
+      "grad_norm": 0.03469294682145119,
+      "learning_rate": 0.0002,
+      "loss": 0.5371942520141602,
+      "mean_token_accuracy": 0.7833464443683624,
+      "num_tokens": 11292531.0,
+      "step": 693
+    },
+    {
+      "entropy": 0.5397353172302246,
+      "epoch": 2.5906542056074766,
+      "grad_norm": 0.03771025687456131,
+      "learning_rate": 0.0002,
+      "loss": 0.5416728854179382,
+      "mean_token_accuracy": 0.7781106233596802,
+      "num_tokens": 11308718.0,
+      "step": 694
+    },
+    {
+      "entropy": 0.5285734534263611,
+      "epoch": 2.594392523364486,
+      "grad_norm": 0.03823580965399742,
+      "learning_rate": 0.0002,
+      "loss": 0.5330255627632141,
+      "mean_token_accuracy": 0.7819734811782837,
+      "num_tokens": 11324754.0,
+      "step": 695
+    },
+    {
+      "entropy": 0.5268194079399109,
+      "epoch": 2.5981308411214954,
+      "grad_norm": 0.04260648787021637,
+      "learning_rate": 0.0002,
+      "loss": 0.5239391326904297,
+      "mean_token_accuracy": 0.7846912294626236,
+      "num_tokens": 11340968.0,
+      "step": 696
+    },
+    {
+      "entropy": 0.5480171293020248,
+      "epoch": 2.601869158878505,
+      "grad_norm": 0.036302078515291214,
+      "learning_rate": 0.0002,
+      "loss": 0.5469081997871399,
+      "mean_token_accuracy": 0.7779969573020935,
+      "num_tokens": 11357426.0,
+      "step": 697
+    },
+    {
+      "entropy": 0.5506609529256821,
+      "epoch": 2.605607476635514,
+      "grad_norm": 0.03509664908051491,
+      "learning_rate": 0.0002,
+      "loss": 0.5489600300788879,
+      "mean_token_accuracy": 0.7774904668331146,
+      "num_tokens": 11373998.0,
+      "step": 698
+    },
+    {
+      "entropy": 0.5326560884714127,
+      "epoch": 2.6093457943925236,
+      "grad_norm": 0.04218761622905731,
+      "learning_rate": 0.0002,
+      "loss": 0.5333682894706726,
+      "mean_token_accuracy": 0.7844813168048859,
+      "num_tokens": 11390325.0,
+      "step": 699
+    },
+    {
+      "entropy": 0.5372011959552765,
+      "epoch": 2.613084112149533,
+      "grad_norm": 0.033543672412633896,
+      "learning_rate": 0.0002,
+      "loss": 0.5339676737785339,
+      "mean_token_accuracy": 0.7814666479825974,
+      "num_tokens": 11406819.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.5389137715101242,
+      "epoch": 2.616822429906542,
+      "grad_norm": 0.041952140629291534,
+      "learning_rate": 0.0002,
+      "loss": 0.544516921043396,
+      "mean_token_accuracy": 0.777948647737503,
+      "num_tokens": 11423114.0,
+      "step": 701
+    },
+    {
+      "entropy": 0.5339553654193878,
+      "epoch": 2.6205607476635513,
+      "grad_norm": 0.04165533185005188,
+      "learning_rate": 0.0002,
+      "loss": 0.5357546210289001,
+      "mean_token_accuracy": 0.7815313339233398,
+      "num_tokens": 11439501.0,
+      "step": 702
+    },
+    {
+      "entropy": 0.5327703803777695,
+      "epoch": 2.6242990654205607,
+      "grad_norm": 0.04676680266857147,
+      "learning_rate": 0.0002,
+      "loss": 0.5317723155021667,
+      "mean_token_accuracy": 0.7842179238796234,
+      "num_tokens": 11455968.0,
+      "step": 703
+    },
+    {
+      "entropy": 0.5289216041564941,
+      "epoch": 2.62803738317757,
+      "grad_norm": 0.04229076951742172,
+      "learning_rate": 0.0002,
+      "loss": 0.5229822397232056,
+      "mean_token_accuracy": 0.7897761762142181,
+      "num_tokens": 11472189.0,
+      "step": 704
+    },
+    {
+      "entropy": 0.5446136146783829,
+      "epoch": 2.6317757009345795,
+      "grad_norm": 0.045102689415216446,
+      "learning_rate": 0.0002,
+      "loss": 0.5427557229995728,
+      "mean_token_accuracy": 0.780061662197113,
+      "num_tokens": 11488564.0,
+      "step": 705
+    },
+    {
+      "entropy": 0.5221227407455444,
+      "epoch": 2.635514018691589,
+      "grad_norm": 0.048957664519548416,
+      "learning_rate": 0.0002,
+      "loss": 0.5266565680503845,
+      "mean_token_accuracy": 0.7874798774719238,
+      "num_tokens": 11505006.0,
+      "step": 706
+    },
+    {
+      "entropy": 0.5404656529426575,
+      "epoch": 2.6392523364485982,
+      "grad_norm": 0.04282474145293236,
+      "learning_rate": 0.0002,
+      "loss": 0.5407752394676208,
+      "mean_token_accuracy": 0.7841375470161438,
+      "num_tokens": 11521778.0,
+      "step": 707
+    },
+    {
+      "entropy": 0.5307196229696274,
+      "epoch": 2.6429906542056076,
+      "grad_norm": 0.03666854277253151,
+      "learning_rate": 0.0002,
+      "loss": 0.5308942198753357,
+      "mean_token_accuracy": 0.7844790369272232,
+      "num_tokens": 11538182.0,
+      "step": 708
+    },
+    {
+      "entropy": 0.5377398878335953,
+      "epoch": 2.6467289719626166,
+      "grad_norm": 0.039178650826215744,
+      "learning_rate": 0.0002,
+      "loss": 0.5345498919487,
+      "mean_token_accuracy": 0.7794453948736191,
+      "num_tokens": 11554684.0,
+      "step": 709
+    },
+    {
+      "entropy": 0.5536644905805588,
+      "epoch": 2.650467289719626,
+      "grad_norm": 0.03832559287548065,
+      "learning_rate": 0.0002,
+      "loss": 0.5499898195266724,
+      "mean_token_accuracy": 0.7778642326593399,
+      "num_tokens": 11571201.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.5420532822608948,
+      "epoch": 2.6542056074766354,
+      "grad_norm": 0.03739701583981514,
+      "learning_rate": 0.0002,
+      "loss": 0.5493305921554565,
+      "mean_token_accuracy": 0.7773284465074539,
+      "num_tokens": 11587368.0,
+      "step": 711
+    },
+    {
+      "entropy": 0.5395905524492264,
+      "epoch": 2.6579439252336448,
+      "grad_norm": 0.04595664143562317,
+      "learning_rate": 0.0002,
+      "loss": 0.544034481048584,
+      "mean_token_accuracy": 0.7786346226930618,
+      "num_tokens": 11603772.0,
+      "step": 712
+    },
+    {
+      "entropy": 0.5372798144817352,
+      "epoch": 2.661682242990654,
+      "grad_norm": 0.04127715900540352,
+      "learning_rate": 0.0002,
+      "loss": 0.5426503419876099,
+      "mean_token_accuracy": 0.7785855233669281,
+      "num_tokens": 11620154.0,
+      "step": 713
+    },
+    {
+      "entropy": 0.5510691106319427,
+      "epoch": 2.6654205607476635,
+      "grad_norm": 0.03742792084813118,
+      "learning_rate": 0.0002,
+      "loss": 0.5547200441360474,
+      "mean_token_accuracy": 0.7739700675010681,
+      "num_tokens": 11636423.0,
+      "step": 714
+    },
+    {
+      "entropy": 0.5472855418920517,
+      "epoch": 2.669158878504673,
+      "grad_norm": 0.04350201040506363,
+      "learning_rate": 0.0002,
+      "loss": 0.5393113493919373,
+      "mean_token_accuracy": 0.7822133600711823,
+      "num_tokens": 11652714.0,
+      "step": 715
+    },
+    {
+      "entropy": 0.5539711713790894,
+      "epoch": 2.6728971962616823,
+      "grad_norm": 0.03917175903916359,
+      "learning_rate": 0.0002,
+      "loss": 0.5469767451286316,
+      "mean_token_accuracy": 0.7737536281347275,
+      "num_tokens": 11668901.0,
+      "step": 716
+    },
+    {
+      "entropy": 0.532858207821846,
+      "epoch": 2.6766355140186917,
+      "grad_norm": 0.04417381435632706,
+      "learning_rate": 0.0002,
+      "loss": 0.5299221873283386,
+      "mean_token_accuracy": 0.7851613610982895,
+      "num_tokens": 11685245.0,
+      "step": 717
+    },
+    {
+      "entropy": 0.5360149517655373,
+      "epoch": 2.680373831775701,
+      "grad_norm": 0.045524608343839645,
+      "learning_rate": 0.0002,
+      "loss": 0.5396956205368042,
+      "mean_token_accuracy": 0.7826401442289352,
+      "num_tokens": 11701586.0,
+      "step": 718
+    },
+    {
+      "entropy": 0.5417494922876358,
+      "epoch": 2.6841121495327105,
+      "grad_norm": 0.04371016472578049,
+      "learning_rate": 0.0002,
+      "loss": 0.547548770904541,
+      "mean_token_accuracy": 0.7796981632709503,
+      "num_tokens": 11717869.0,
+      "step": 719
+    },
+    {
+      "entropy": 0.5364665389060974,
+      "epoch": 2.68785046728972,
+      "grad_norm": 0.04112569987773895,
+      "learning_rate": 0.0002,
+      "loss": 0.5360340476036072,
+      "mean_token_accuracy": 0.7829563021659851,
+      "num_tokens": 11733931.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.5393244326114655,
+      "epoch": 2.691588785046729,
+      "grad_norm": 0.03485904261469841,
+      "learning_rate": 0.0002,
+      "loss": 0.5365700721740723,
+      "mean_token_accuracy": 0.7832135111093521,
+      "num_tokens": 11750251.0,
+      "step": 721
+    },
+    {
+      "entropy": 0.5402418822050095,
+      "epoch": 2.695327102803738,
+      "grad_norm": 0.035655390471220016,
+      "learning_rate": 0.0002,
+      "loss": 0.5399341583251953,
+      "mean_token_accuracy": 0.7821518778800964,
+      "num_tokens": 11766402.0,
+      "step": 722
+    },
+    {
+      "entropy": 0.5645921379327774,
+      "epoch": 2.6990654205607476,
+      "grad_norm": 0.03827064484357834,
+      "learning_rate": 0.0002,
+      "loss": 0.5571991205215454,
+      "mean_token_accuracy": 0.7733338475227356,
+      "num_tokens": 11783003.0,
+      "step": 723
+    },
+    {
+      "entropy": 0.5261956825852394,
+      "epoch": 2.702803738317757,
+      "grad_norm": 0.03958537429571152,
+      "learning_rate": 0.0002,
+      "loss": 0.5266394019126892,
+      "mean_token_accuracy": 0.7840381115674973,
+      "num_tokens": 11799129.0,
+      "step": 724
+    },
+    {
+      "entropy": 0.5451592206954956,
+      "epoch": 2.7065420560747664,
+      "grad_norm": 0.03517312929034233,
+      "learning_rate": 0.0002,
+      "loss": 0.5486286878585815,
+      "mean_token_accuracy": 0.7751190662384033,
+      "num_tokens": 11815476.0,
+      "step": 725
+    },
+    {
+      "entropy": 0.5393543839454651,
+      "epoch": 2.710280373831776,
+      "grad_norm": 0.0371185727417469,
+      "learning_rate": 0.0002,
+      "loss": 0.5353912115097046,
+      "mean_token_accuracy": 0.7856109440326691,
+      "num_tokens": 11831984.0,
+      "step": 726
+    },
+    {
+      "entropy": 0.5295817404985428,
+      "epoch": 2.714018691588785,
+      "grad_norm": 0.038498733192682266,
+      "learning_rate": 0.0002,
+      "loss": 0.5308818221092224,
+      "mean_token_accuracy": 0.7821252197027206,
+      "num_tokens": 11848466.0,
+      "step": 727
+    },
+    {
+      "entropy": 0.5493554025888443,
+      "epoch": 2.717757009345794,
+      "grad_norm": 0.039601247757673264,
+      "learning_rate": 0.0002,
+      "loss": 0.547062873840332,
+      "mean_token_accuracy": 0.7771877348423004,
+      "num_tokens": 11864918.0,
+      "step": 728
+    },
+    {
+      "entropy": 0.5261873155832291,
+      "epoch": 2.7214953271028035,
+      "grad_norm": 0.038875121623277664,
+      "learning_rate": 0.0002,
+      "loss": 0.5303913950920105,
+      "mean_token_accuracy": 0.7834369987249374,
+      "num_tokens": 11881212.0,
+      "step": 729
+    },
+    {
+      "entropy": 0.5295650064945221,
+      "epoch": 2.725233644859813,
+      "grad_norm": 0.03564633056521416,
+      "learning_rate": 0.0002,
+      "loss": 0.5287110805511475,
+      "mean_token_accuracy": 0.7855493873357773,
+      "num_tokens": 11897285.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.5161215513944626,
+      "epoch": 2.7289719626168223,
+      "grad_norm": 0.03593486547470093,
+      "learning_rate": 0.0002,
+      "loss": 0.5092670321464539,
+      "mean_token_accuracy": 0.7932390123605728,
+      "num_tokens": 11913860.0,
+      "step": 731
+    },
+    {
+      "entropy": 0.5438321828842163,
+      "epoch": 2.7327102803738317,
+      "grad_norm": 0.03703992813825607,
+      "learning_rate": 0.0002,
+      "loss": 0.5476446151733398,
+      "mean_token_accuracy": 0.7795855104923248,
+      "num_tokens": 11930261.0,
+      "step": 732
+    },
+    {
+      "entropy": 0.5311425775289536,
+      "epoch": 2.736448598130841,
+      "grad_norm": 0.037695612758398056,
+      "learning_rate": 0.0002,
+      "loss": 0.534140408039093,
+      "mean_token_accuracy": 0.7818547487258911,
+      "num_tokens": 11946560.0,
+      "step": 733
+    },
+    {
+      "entropy": 0.5209683775901794,
+      "epoch": 2.7401869158878505,
+      "grad_norm": 0.03890440613031387,
+      "learning_rate": 0.0002,
+      "loss": 0.5241660475730896,
+      "mean_token_accuracy": 0.787934735417366,
+      "num_tokens": 11962956.0,
+      "step": 734
+    },
+    {
+      "entropy": 0.5443778336048126,
+      "epoch": 2.74392523364486,
+      "grad_norm": 0.03788384422659874,
+      "learning_rate": 0.0002,
+      "loss": 0.5421298742294312,
+      "mean_token_accuracy": 0.7814332842826843,
+      "num_tokens": 11979290.0,
+      "step": 735
+    },
+    {
+      "entropy": 0.5518642365932465,
+      "epoch": 2.7476635514018692,
+      "grad_norm": 0.040383294224739075,
+      "learning_rate": 0.0002,
+      "loss": 0.555992603302002,
+      "mean_token_accuracy": 0.7762507796287537,
+      "num_tokens": 11995487.0,
+      "step": 736
+    },
+    {
+      "entropy": 0.5241853296756744,
+      "epoch": 2.7514018691588786,
+      "grad_norm": 0.039098046720027924,
+      "learning_rate": 0.0002,
+      "loss": 0.52538001537323,
+      "mean_token_accuracy": 0.7879568636417389,
+      "num_tokens": 12011811.0,
+      "step": 737
+    },
+    {
+      "entropy": 0.5365500450134277,
+      "epoch": 2.755140186915888,
+      "grad_norm": 0.041129641234874725,
+      "learning_rate": 0.0002,
+      "loss": 0.5400822162628174,
+      "mean_token_accuracy": 0.7799027115106583,
+      "num_tokens": 12028077.0,
+      "step": 738
+    },
+    {
+      "entropy": 0.5531625598669052,
+      "epoch": 2.7588785046728974,
+      "grad_norm": 0.040683627128601074,
+      "learning_rate": 0.0002,
+      "loss": 0.5487805604934692,
+      "mean_token_accuracy": 0.7768402397632599,
+      "num_tokens": 12044438.0,
+      "step": 739
+    },
+    {
+      "entropy": 0.5442669987678528,
+      "epoch": 2.762616822429907,
+      "grad_norm": 0.040092833340168,
+      "learning_rate": 0.0002,
+      "loss": 0.547822117805481,
+      "mean_token_accuracy": 0.7773446589708328,
+      "num_tokens": 12060591.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.5772512406110764,
+      "epoch": 2.7663551401869158,
+      "grad_norm": 0.04026317968964577,
+      "learning_rate": 0.0002,
+      "loss": 0.5646360516548157,
+      "mean_token_accuracy": 0.771668940782547,
+      "num_tokens": 12076995.0,
+      "step": 741
+    },
+    {
+      "entropy": 0.550237238407135,
+      "epoch": 2.770093457943925,
+      "grad_norm": 0.03907431662082672,
+      "learning_rate": 0.0002,
+      "loss": 0.5490835905075073,
+      "mean_token_accuracy": 0.7739729285240173,
+      "num_tokens": 12093381.0,
+      "step": 742
+    },
+    {
+      "entropy": 0.5247115790843964,
+      "epoch": 2.7738317757009345,
+      "grad_norm": 0.038372889161109924,
+      "learning_rate": 0.0002,
+      "loss": 0.5327771902084351,
+      "mean_token_accuracy": 0.7830088138580322,
+      "num_tokens": 12109606.0,
+      "step": 743
+    },
+    {
+      "entropy": 0.5228423178195953,
+      "epoch": 2.777570093457944,
+      "grad_norm": 0.044808950275182724,
+      "learning_rate": 0.0002,
+      "loss": 0.5320113301277161,
+      "mean_token_accuracy": 0.7844864279031754,
+      "num_tokens": 12125852.0,
+      "step": 744
+    },
+    {
+      "entropy": 0.530124694108963,
+      "epoch": 2.7813084112149533,
+      "grad_norm": 0.03803320229053497,
+      "learning_rate": 0.0002,
+      "loss": 0.5259105563163757,
+      "mean_token_accuracy": 0.784927174448967,
+      "num_tokens": 12142284.0,
+      "step": 745
+    },
+    {
+      "entropy": 0.5511182993650436,
+      "epoch": 2.7850467289719627,
+      "grad_norm": 0.03542547672986984,
+      "learning_rate": 0.0002,
+      "loss": 0.5479939579963684,
+      "mean_token_accuracy": 0.7765052765607834,
+      "num_tokens": 12158595.0,
+      "step": 746
+    },
+    {
+      "entropy": 0.5446942672133446,
+      "epoch": 2.788785046728972,
+      "grad_norm": 0.04419056326150894,
+      "learning_rate": 0.0002,
+      "loss": 0.5443319082260132,
+      "mean_token_accuracy": 0.7793864756822586,
+      "num_tokens": 12174701.0,
+      "step": 747
+    },
+    {
+      "entropy": 0.5594473034143448,
+      "epoch": 2.792523364485981,
+      "grad_norm": 0.034595683217048645,
+      "learning_rate": 0.0002,
+      "loss": 0.5563086271286011,
+      "mean_token_accuracy": 0.7736584842205048,
+      "num_tokens": 12191251.0,
+      "step": 748
+    },
+    {
+      "entropy": 0.5360135138034821,
+      "epoch": 2.7962616822429904,
+      "grad_norm": 0.037246719002723694,
+      "learning_rate": 0.0002,
+      "loss": 0.53227299451828,
+      "mean_token_accuracy": 0.7829789221286774,
+      "num_tokens": 12207768.0,
+      "step": 749
+    },
+    {
+      "entropy": 0.5398263484239578,
+      "epoch": 2.8,
+      "grad_norm": 0.040100134909152985,
+      "learning_rate": 0.0002,
+      "loss": 0.5421459078788757,
+      "mean_token_accuracy": 0.7809022068977356,
+      "num_tokens": 12224318.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.5231769531965256,
+      "epoch": 2.803738317757009,
+      "grad_norm": 0.04384174942970276,
+      "learning_rate": 0.0002,
+      "loss": 0.532194972038269,
+      "mean_token_accuracy": 0.7854811102151871,
+      "num_tokens": 12240560.0,
+      "step": 751
+    },
+    {
+      "entropy": 0.5372914969921112,
+      "epoch": 2.8074766355140186,
+      "grad_norm": 0.041690364480018616,
+      "learning_rate": 0.0002,
+      "loss": 0.5358984470367432,
+      "mean_token_accuracy": 0.7798957526683807,
+      "num_tokens": 12256698.0,
+      "step": 752
+    },
+    {
+      "entropy": 0.5490759164094925,
+      "epoch": 2.811214953271028,
+      "grad_norm": 0.035977087914943695,
+      "learning_rate": 0.0002,
+      "loss": 0.5457599759101868,
+      "mean_token_accuracy": 0.7804635167121887,
+      "num_tokens": 12273196.0,
+      "step": 753
+    },
+    {
+      "entropy": 0.5454043298959732,
+      "epoch": 2.8149532710280374,
+      "grad_norm": 0.038070011883974075,
+      "learning_rate": 0.0002,
+      "loss": 0.5440967082977295,
+      "mean_token_accuracy": 0.7805526405572891,
+      "num_tokens": 12289448.0,
+      "step": 754
+    },
+    {
+      "entropy": 0.5171327888965607,
+      "epoch": 2.8186915887850468,
+      "grad_norm": 0.043416958302259445,
+      "learning_rate": 0.0002,
+      "loss": 0.5235368609428406,
+      "mean_token_accuracy": 0.7856585532426834,
+      "num_tokens": 12305728.0,
+      "step": 755
+    },
+    {
+      "entropy": 0.5519092679023743,
+      "epoch": 2.822429906542056,
+      "grad_norm": 0.04086895287036896,
+      "learning_rate": 0.0002,
+      "loss": 0.5498278737068176,
+      "mean_token_accuracy": 0.776623860001564,
+      "num_tokens": 12322183.0,
+      "step": 756
+    },
+    {
+      "entropy": 0.5254665613174438,
+      "epoch": 2.8261682242990656,
+      "grad_norm": 0.03841882944107056,
+      "learning_rate": 0.0002,
+      "loss": 0.5266950130462646,
+      "mean_token_accuracy": 0.786298468708992,
+      "num_tokens": 12338635.0,
+      "step": 757
+    },
+    {
+      "entropy": 0.5411494970321655,
+      "epoch": 2.829906542056075,
+      "grad_norm": 0.0401926264166832,
+      "learning_rate": 0.0002,
+      "loss": 0.5370864272117615,
+      "mean_token_accuracy": 0.7828312367200851,
+      "num_tokens": 12355089.0,
+      "step": 758
+    },
+    {
+      "entropy": 0.5513712912797928,
+      "epoch": 2.8336448598130843,
+      "grad_norm": 0.0333644300699234,
+      "learning_rate": 0.0002,
+      "loss": 0.5467706918716431,
+      "mean_token_accuracy": 0.7782435566186905,
+      "num_tokens": 12371887.0,
+      "step": 759
+    },
+    {
+      "entropy": 0.5323017537593842,
+      "epoch": 2.8373831775700937,
+      "grad_norm": 0.04663057625293732,
+      "learning_rate": 0.0002,
+      "loss": 0.5380391478538513,
+      "mean_token_accuracy": 0.7797932773828506,
+      "num_tokens": 12388141.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.5405762195587158,
+      "epoch": 2.8411214953271027,
+      "grad_norm": 0.03440416231751442,
+      "learning_rate": 0.0002,
+      "loss": 0.5396413207054138,
+      "mean_token_accuracy": 0.7808928042650223,
+      "num_tokens": 12404651.0,
+      "step": 761
+    },
+    {
+      "entropy": 0.5396543145179749,
+      "epoch": 2.844859813084112,
+      "grad_norm": 0.035920772701501846,
+      "learning_rate": 0.0002,
+      "loss": 0.5427154898643494,
+      "mean_token_accuracy": 0.7765078693628311,
+      "num_tokens": 12420963.0,
+      "step": 762
+    },
+    {
+      "entropy": 0.5548029094934464,
+      "epoch": 2.8485981308411215,
+      "grad_norm": 0.039103955030441284,
+      "learning_rate": 0.0002,
+      "loss": 0.5556834936141968,
+      "mean_token_accuracy": 0.7767738401889801,
+      "num_tokens": 12437531.0,
+      "step": 763
+    },
+    {
+      "entropy": 0.5415983200073242,
+      "epoch": 2.852336448598131,
+      "grad_norm": 0.042144615203142166,
+      "learning_rate": 0.0002,
+      "loss": 0.5447718501091003,
+      "mean_token_accuracy": 0.7775606215000153,
+      "num_tokens": 12454010.0,
+      "step": 764
+    },
+    {
+      "entropy": 0.5518344044685364,
+      "epoch": 2.8560747663551402,
+      "grad_norm": 0.04103001952171326,
+      "learning_rate": 0.0002,
+      "loss": 0.5544916391372681,
+      "mean_token_accuracy": 0.7738406360149384,
+      "num_tokens": 12470274.0,
+      "step": 765
+    },
+    {
+      "entropy": 0.5313066393136978,
+      "epoch": 2.8598130841121496,
+      "grad_norm": 0.032009996473789215,
+      "learning_rate": 0.0002,
+      "loss": 0.5202063918113708,
+      "mean_token_accuracy": 0.7903139889240265,
+      "num_tokens": 12486731.0,
+      "step": 766
+    },
+    {
+      "entropy": 0.5384960919618607,
+      "epoch": 2.863551401869159,
+      "grad_norm": 0.03478989377617836,
+      "learning_rate": 0.0002,
+      "loss": 0.5340580344200134,
+      "mean_token_accuracy": 0.7825128734111786,
+      "num_tokens": 12503030.0,
+      "step": 767
+    },
+    {
+      "entropy": 0.5224665403366089,
+      "epoch": 2.867289719626168,
+      "grad_norm": 0.04253165051341057,
+      "learning_rate": 0.0002,
+      "loss": 0.5197573900222778,
+      "mean_token_accuracy": 0.7862243801355362,
+      "num_tokens": 12519459.0,
+      "step": 768
+    },
+    {
+      "entropy": 0.5370153039693832,
+      "epoch": 2.8710280373831774,
+      "grad_norm": 0.043792687356472015,
+      "learning_rate": 0.0002,
+      "loss": 0.5484573245048523,
+      "mean_token_accuracy": 0.7781466245651245,
+      "num_tokens": 12535908.0,
+      "step": 769
+    },
+    {
+      "entropy": 0.5293686985969543,
+      "epoch": 2.8747663551401867,
+      "grad_norm": 0.040938977152109146,
+      "learning_rate": 0.0002,
+      "loss": 0.536544680595398,
+      "mean_token_accuracy": 0.7815448939800262,
+      "num_tokens": 12552278.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.5587695837020874,
+      "epoch": 2.878504672897196,
+      "grad_norm": 0.03950825333595276,
+      "learning_rate": 0.0002,
+      "loss": 0.5560339689254761,
+      "mean_token_accuracy": 0.7734224796295166,
+      "num_tokens": 12568976.0,
+      "step": 771
+    },
+    {
+      "entropy": 0.5375785976648331,
+      "epoch": 2.8822429906542055,
+      "grad_norm": 0.039299093186855316,
+      "learning_rate": 0.0002,
+      "loss": 0.5312560796737671,
+      "mean_token_accuracy": 0.7845937460660934,
+      "num_tokens": 12585119.0,
+      "step": 772
+    },
+    {
+      "entropy": 0.5369968712329865,
+      "epoch": 2.885981308411215,
+      "grad_norm": 0.03820907697081566,
+      "learning_rate": 0.0002,
+      "loss": 0.5376330614089966,
+      "mean_token_accuracy": 0.7842772305011749,
+      "num_tokens": 12601276.0,
+      "step": 773
+    },
+    {
+      "entropy": 0.544855386018753,
+      "epoch": 2.8897196261682243,
+      "grad_norm": 0.03554603457450867,
+      "learning_rate": 0.0002,
+      "loss": 0.5421188473701477,
+      "mean_token_accuracy": 0.777969166636467,
+      "num_tokens": 12617527.0,
+      "step": 774
+    },
+    {
+      "entropy": 0.5378414988517761,
+      "epoch": 2.8934579439252337,
+      "grad_norm": 0.040478140115737915,
+      "learning_rate": 0.0002,
+      "loss": 0.5390060544013977,
+      "mean_token_accuracy": 0.7802563160657883,
+      "num_tokens": 12633807.0,
+      "step": 775
+    },
+    {
+      "entropy": 0.5441332012414932,
+      "epoch": 2.897196261682243,
+      "grad_norm": 0.043334268033504486,
+      "learning_rate": 0.0002,
+      "loss": 0.5497444272041321,
+      "mean_token_accuracy": 0.7784465253353119,
+      "num_tokens": 12650362.0,
+      "step": 776
+    },
+    {
+      "entropy": 0.5244786143302917,
+      "epoch": 2.9009345794392525,
+      "grad_norm": 0.04438484460115433,
+      "learning_rate": 0.0002,
+      "loss": 0.5322083830833435,
+      "mean_token_accuracy": 0.7820873856544495,
+      "num_tokens": 12666671.0,
+      "step": 777
+    },
+    {
+      "entropy": 0.5577758997678757,
+      "epoch": 2.904672897196262,
+      "grad_norm": 0.0378209687769413,
+      "learning_rate": 0.0002,
+      "loss": 0.5586643815040588,
+      "mean_token_accuracy": 0.7750083953142166,
+      "num_tokens": 12682921.0,
+      "step": 778
+    },
+    {
+      "entropy": 0.5417321473360062,
+      "epoch": 2.9084112149532713,
+      "grad_norm": 0.04056469723582268,
+      "learning_rate": 0.0002,
+      "loss": 0.5351645350456238,
+      "mean_token_accuracy": 0.7857696861028671,
+      "num_tokens": 12699271.0,
+      "step": 779
+    },
+    {
+      "entropy": 0.5629414021968842,
+      "epoch": 2.91214953271028,
+      "grad_norm": 0.03783043846487999,
+      "learning_rate": 0.0002,
+      "loss": 0.5545192956924438,
+      "mean_token_accuracy": 0.7746420204639435,
+      "num_tokens": 12715410.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.5410288572311401,
+      "epoch": 2.9158878504672896,
+      "grad_norm": 0.03987415134906769,
+      "learning_rate": 0.0002,
+      "loss": 0.5423367619514465,
+      "mean_token_accuracy": 0.778704360127449,
+      "num_tokens": 12731587.0,
+      "step": 781
+    },
+    {
+      "entropy": 0.5282145440578461,
+      "epoch": 2.919626168224299,
+      "grad_norm": 0.03972848504781723,
+      "learning_rate": 0.0002,
+      "loss": 0.5305967330932617,
+      "mean_token_accuracy": 0.7833620458841324,
+      "num_tokens": 12747966.0,
+      "step": 782
+    },
+    {
+      "entropy": 0.5176219195127487,
+      "epoch": 2.9233644859813084,
+      "grad_norm": 0.0456930510699749,
+      "learning_rate": 0.0002,
+      "loss": 0.5257712602615356,
+      "mean_token_accuracy": 0.7853413671255112,
+      "num_tokens": 12764231.0,
+      "step": 783
+    },
+    {
+      "entropy": 0.5378240942955017,
+      "epoch": 2.9271028037383178,
+      "grad_norm": 0.04101483151316643,
+      "learning_rate": 0.0002,
+      "loss": 0.5421459674835205,
+      "mean_token_accuracy": 0.7807480245828629,
+      "num_tokens": 12780553.0,
+      "step": 784
+    },
+    {
+      "entropy": 0.557637631893158,
+      "epoch": 2.930841121495327,
+      "grad_norm": 0.04411512240767479,
+      "learning_rate": 0.0002,
+      "loss": 0.554862380027771,
+      "mean_token_accuracy": 0.7744756042957306,
+      "num_tokens": 12796795.0,
+      "step": 785
+    },
+    {
+      "entropy": 0.5379236787557602,
+      "epoch": 2.9345794392523366,
+      "grad_norm": 0.03611760586500168,
+      "learning_rate": 0.0002,
+      "loss": 0.533431887626648,
+      "mean_token_accuracy": 0.7804027199745178,
+      "num_tokens": 12813009.0,
+      "step": 786
+    },
+    {
+      "entropy": 0.5566578209400177,
+      "epoch": 2.938317757009346,
+      "grad_norm": 0.03715398907661438,
+      "learning_rate": 0.0002,
+      "loss": 0.5433220267295837,
+      "mean_token_accuracy": 0.7798871845006943,
+      "num_tokens": 12829448.0,
+      "step": 787
+    },
+    {
+      "entropy": 0.5410407036542892,
+      "epoch": 2.942056074766355,
+      "grad_norm": 0.03546489402651787,
+      "learning_rate": 0.0002,
+      "loss": 0.5368283987045288,
+      "mean_token_accuracy": 0.780596062541008,
+      "num_tokens": 12845740.0,
+      "step": 788
+    },
+    {
+      "entropy": 0.5211344361305237,
+      "epoch": 2.9457943925233643,
+      "grad_norm": 0.04108017310500145,
+      "learning_rate": 0.0002,
+      "loss": 0.5253114104270935,
+      "mean_token_accuracy": 0.7879950255155563,
+      "num_tokens": 12861833.0,
+      "step": 789
+    },
+    {
+      "entropy": 0.5430938005447388,
+      "epoch": 2.9495327102803737,
+      "grad_norm": 0.04749761149287224,
+      "learning_rate": 0.0002,
+      "loss": 0.5581343770027161,
+      "mean_token_accuracy": 0.7737741619348526,
+      "num_tokens": 12878195.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.5290612280368805,
+      "epoch": 2.953271028037383,
+      "grad_norm": 0.04016729071736336,
+      "learning_rate": 0.0002,
+      "loss": 0.5378360152244568,
+      "mean_token_accuracy": 0.782782569527626,
+      "num_tokens": 12894573.0,
+      "step": 791
+    },
+    {
+      "entropy": 0.5361242145299911,
+      "epoch": 2.9570093457943925,
+      "grad_norm": 0.03678933158516884,
+      "learning_rate": 0.0002,
+      "loss": 0.5257135629653931,
+      "mean_token_accuracy": 0.7893773168325424,
+      "num_tokens": 12910863.0,
+      "step": 792
+    },
+    {
+      "entropy": 0.5259503424167633,
+      "epoch": 2.960747663551402,
+      "grad_norm": 0.03752489387989044,
+      "learning_rate": 0.0002,
+      "loss": 0.5143717527389526,
+      "mean_token_accuracy": 0.7932046353816986,
+      "num_tokens": 12927225.0,
+      "step": 793
+    },
+    {
+      "entropy": 0.5639631152153015,
+      "epoch": 2.9644859813084112,
+      "grad_norm": 0.040728434920310974,
+      "learning_rate": 0.0002,
+      "loss": 0.5542000532150269,
+      "mean_token_accuracy": 0.7755206525325775,
+      "num_tokens": 12943436.0,
+      "step": 794
+    },
+    {
+      "entropy": 0.5300319939851761,
+      "epoch": 2.9682242990654206,
+      "grad_norm": 0.03866998106241226,
+      "learning_rate": 0.0002,
+      "loss": 0.536708652973175,
+      "mean_token_accuracy": 0.7822965830564499,
+      "num_tokens": 12959780.0,
+      "step": 795
+    },
+    {
+      "entropy": 0.5109827667474747,
+      "epoch": 2.97196261682243,
+      "grad_norm": 0.05049736052751541,
+      "learning_rate": 0.0002,
+      "loss": 0.5270419716835022,
+      "mean_token_accuracy": 0.785243034362793,
+      "num_tokens": 12976091.0,
+      "step": 796
+    },
+    {
+      "entropy": 0.5446041822433472,
+      "epoch": 2.9757009345794394,
+      "grad_norm": 0.03969959914684296,
+      "learning_rate": 0.0002,
+      "loss": 0.5448250770568848,
+      "mean_token_accuracy": 0.7787258327007294,
+      "num_tokens": 12992521.0,
+      "step": 797
+    },
+    {
+      "entropy": 0.5244547426700592,
+      "epoch": 2.979439252336449,
+      "grad_norm": 0.03810025751590729,
+      "learning_rate": 0.0002,
+      "loss": 0.5291001796722412,
+      "mean_token_accuracy": 0.784002810716629,
+      "num_tokens": 13008703.0,
+      "step": 798
+    },
+    {
+      "entropy": 0.5672779381275177,
+      "epoch": 2.983177570093458,
+      "grad_norm": 0.039007145911455154,
+      "learning_rate": 0.0002,
+      "loss": 0.5631389021873474,
+      "mean_token_accuracy": 0.7706998288631439,
+      "num_tokens": 13025057.0,
+      "step": 799
+    },
+    {
+      "entropy": 0.5293317213654518,
+      "epoch": 2.986915887850467,
+      "grad_norm": 0.037154458463191986,
+      "learning_rate": 0.0002,
+      "loss": 0.5227472186088562,
+      "mean_token_accuracy": 0.7874658554792404,
+      "num_tokens": 13041252.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.537149466574192,
+      "epoch": 2.9906542056074765,
+      "grad_norm": 0.03552801162004471,
+      "learning_rate": 0.0002,
+      "loss": 0.5343438982963562,
+      "mean_token_accuracy": 0.7848320156335831,
+      "num_tokens": 13057526.0,
+      "step": 801
+    },
+    {
+      "entropy": 0.530741959810257,
+      "epoch": 2.994392523364486,
+      "grad_norm": 0.043317440897226334,
+      "learning_rate": 0.0002,
+      "loss": 0.5383322238922119,
+      "mean_token_accuracy": 0.7800941169261932,
+      "num_tokens": 13073756.0,
+      "step": 802
+    },
+    {
+      "entropy": 0.5432422608137131,
+      "epoch": 2.9981308411214953,
+      "grad_norm": 0.044908758252859116,
+      "learning_rate": 0.0002,
+      "loss": 0.5521294474601746,
+      "mean_token_accuracy": 0.7733398079872131,
+      "num_tokens": 13090164.0,
+      "step": 803
+    },
+    {
+      "entropy": 0.5437783300876617,
+      "epoch": 3.0,
+      "grad_norm": 0.04327746853232384,
+      "learning_rate": 0.0002,
+      "loss": 0.5394353270530701,
+      "mean_token_accuracy": 0.7844643592834473,
+      "num_tokens": 13096259.0,
+      "step": 804
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 804,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.2208281383809843e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}