diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.03443437239052022,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 2.413330078125,
+      "epoch": 1.721718619526011e-05,
+      "grad_norm": 0.6972781419754028,
+      "learning_rate": 0.0,
+      "loss": 2.3998,
+      "mean_token_accuracy": 0.4832073478028178,
+      "num_tokens": 152243.0,
+      "step": 1
+    },
+    {
+      "entropy": 2.4014892578125,
+      "epoch": 3.443437239052022e-05,
+      "grad_norm": 0.703504204750061,
+      "learning_rate": 1e-08,
+      "loss": 2.3565,
+      "mean_token_accuracy": 0.48620754200965166,
+      "num_tokens": 302755.0,
+      "step": 2
+    },
+    {
+      "entropy": 2.3984375,
+      "epoch": 5.165155858578032e-05,
+      "grad_norm": 0.7933295369148254,
+      "learning_rate": 2e-08,
+      "loss": 2.3735,
+      "mean_token_accuracy": 0.4893745076842606,
+      "num_tokens": 430137.0,
+      "step": 3
+    },
+    {
+      "entropy": 2.4783935546875,
+      "epoch": 6.886874478104043e-05,
+      "grad_norm": 0.6833076477050781,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 2.4365,
+      "mean_token_accuracy": 0.4767027348279953,
+      "num_tokens": 575284.0,
+      "step": 4
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 8.608593097630054e-05,
+      "grad_norm": 0.7900882363319397,
+      "learning_rate": 4e-08,
+      "loss": 2.4671,
+      "mean_token_accuracy": 0.47980545135214925,
+      "num_tokens": 714418.0,
+      "step": 5
+    },
+    {
+      "entropy": 2.367431640625,
+      "epoch": 0.00010330311717156065,
+      "grad_norm": 0.6312107443809509,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.3371,
+      "mean_token_accuracy": 0.49468297231942415,
+      "num_tokens": 883298.0,
+      "step": 6
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.00012052030336682076,
+      "grad_norm": 0.6607802510261536,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 2.4411,
+      "mean_token_accuracy": 0.4778139302507043,
+      "num_tokens": 1034855.0,
+      "step": 7
+    },
+    {
+      "entropy": 2.482421875,
+      "epoch": 0.00013773748956208087,
+      "grad_norm": 0.6800277829170227,
+      "learning_rate": 7e-08,
+      "loss": 2.439,
+      "mean_token_accuracy": 0.4751331675797701,
+      "num_tokens": 1176490.0,
+      "step": 8
+    },
+    {
+      "entropy": 2.391357421875,
+      "epoch": 0.000154954675757341,
+      "grad_norm": 0.7818512916564941,
+      "learning_rate": 8e-08,
+      "loss": 2.3771,
+      "mean_token_accuracy": 0.48402665881440043,
+      "num_tokens": 1311861.0,
+      "step": 9
+    },
+    {
+      "entropy": 2.39453125,
+      "epoch": 0.00017217186195260108,
+      "grad_norm": 0.7182415127754211,
+      "learning_rate": 9e-08,
+      "loss": 2.3644,
+      "mean_token_accuracy": 0.4823771519586444,
+      "num_tokens": 1454916.0,
+      "step": 10
+    },
+    {
+      "entropy": 2.477783203125,
+      "epoch": 0.0001893890481478612,
+      "grad_norm": 0.7052203416824341,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.4729,
+      "mean_token_accuracy": 0.4706865563057363,
+      "num_tokens": 1587290.0,
+      "step": 11
+    },
+    {
+      "entropy": 2.4482421875,
+      "epoch": 0.0002066062343431213,
+      "grad_norm": 0.6999819278717041,
+      "learning_rate": 1.1e-07,
+      "loss": 2.4037,
+      "mean_token_accuracy": 0.47885329788550735,
+      "num_tokens": 1724163.0,
+      "step": 12
+    },
+    {
+      "entropy": 2.4818115234375,
+      "epoch": 0.0002238234205383814,
+      "grad_norm": 0.7237532734870911,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 2.5056,
+      "mean_token_accuracy": 0.46921027079224586,
+      "num_tokens": 1865666.0,
+      "step": 13
+    },
+    {
+      "entropy": 2.392333984375,
+      "epoch": 0.00024104060673364153,
+      "grad_norm": 0.6929540634155273,
+      "learning_rate": 1.3e-07,
+      "loss": 2.3854,
+      "mean_token_accuracy": 0.4831690890714526,
+      "num_tokens": 2020018.0,
+      "step": 14
+    },
+    {
+      "entropy": 2.4205322265625,
+      "epoch": 0.00025825779292890165,
+      "grad_norm": 0.6675299406051636,
+      "learning_rate": 1.4e-07,
+      "loss": 2.3398,
+      "mean_token_accuracy": 0.48172238236293197,
+      "num_tokens": 2158990.0,
+      "step": 15
+    },
+    {
+      "entropy": 2.385498046875,
+      "epoch": 0.00027547497912416174,
+      "grad_norm": 0.7503660321235657,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 2.3352,
+      "mean_token_accuracy": 0.4925760827027261,
+      "num_tokens": 2296401.0,
+      "step": 16
+    },
+    {
+      "entropy": 2.4197998046875,
+      "epoch": 0.00029269216531942183,
+      "grad_norm": 0.6798411011695862,
+      "learning_rate": 1.6e-07,
+      "loss": 2.4002,
+      "mean_token_accuracy": 0.4798423429019749,
+      "num_tokens": 2445270.0,
+      "step": 17
+    },
+    {
+      "entropy": 2.474609375,
+      "epoch": 0.000309909351514682,
+      "grad_norm": 0.6871540546417236,
+      "learning_rate": 1.7000000000000001e-07,
+      "loss": 2.4938,
+      "mean_token_accuracy": 0.47186582954600453,
+      "num_tokens": 2586867.0,
+      "step": 18
+    },
+    {
+      "entropy": 2.50732421875,
+      "epoch": 0.00032712653770994207,
+      "grad_norm": 0.717059314250946,
+      "learning_rate": 1.8e-07,
+      "loss": 2.525,
+      "mean_token_accuracy": 0.46879610791802406,
+      "num_tokens": 2720721.0,
+      "step": 19
+    },
+    {
+      "entropy": 2.424560546875,
+      "epoch": 0.00034434372390520216,
+      "grad_norm": 0.6196162700653076,
+      "learning_rate": 1.9e-07,
+      "loss": 2.4176,
+      "mean_token_accuracy": 0.4790610708296299,
+      "num_tokens": 2876790.0,
+      "step": 20
+    },
+    {
+      "entropy": 2.4759521484375,
+      "epoch": 0.0003615609101004623,
+      "grad_norm": 0.6419414281845093,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.4874,
+      "mean_token_accuracy": 0.471744445618242,
+      "num_tokens": 3033198.0,
+      "step": 21
+    },
+    {
+      "entropy": 2.383056640625,
+      "epoch": 0.0003787780962957224,
+      "grad_norm": 0.6465677618980408,
+      "learning_rate": 2.1000000000000003e-07,
+      "loss": 2.3503,
+      "mean_token_accuracy": 0.48874560045078397,
+      "num_tokens": 3181718.0,
+      "step": 22
+    },
+    {
+      "entropy": 2.3697509765625,
+      "epoch": 0.0003959952824909825,
+      "grad_norm": 0.7614319920539856,
+      "learning_rate": 2.2e-07,
+      "loss": 2.3713,
+      "mean_token_accuracy": 0.48694683285430074,
+      "num_tokens": 3328269.0,
+      "step": 23
+    },
+    {
+      "entropy": 2.3336181640625,
+      "epoch": 0.0004132124686862426,
+      "grad_norm": 0.7275253534317017,
+      "learning_rate": 2.3000000000000002e-07,
+      "loss": 2.3167,
+      "mean_token_accuracy": 0.4986211028881371,
+      "num_tokens": 3482113.0,
+      "step": 24
+    },
+    {
+      "entropy": 2.44775390625,
+      "epoch": 0.0004304296548815027,
+      "grad_norm": 0.6546138525009155,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 2.4172,
+      "mean_token_accuracy": 0.48186044162139297,
+      "num_tokens": 3634079.0,
+      "step": 25
+    },
+    {
+      "entropy": 2.495849609375,
+      "epoch": 0.0004476468410767628,
+      "grad_norm": 0.6819922924041748,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 2.4707,
+      "mean_token_accuracy": 0.47281224094331264,
+      "num_tokens": 3784902.0,
+      "step": 26
+    },
+    {
+      "entropy": 2.490966796875,
+      "epoch": 0.0004648640272720229,
+      "grad_norm": 0.6698691248893738,
+      "learning_rate": 2.6e-07,
+      "loss": 2.4847,
+      "mean_token_accuracy": 0.46762992488220334,
+      "num_tokens": 3923281.0,
+      "step": 27
+    },
+    {
+      "entropy": 2.4964599609375,
+      "epoch": 0.00048208121346728306,
+      "grad_norm": 0.6502507925033569,
+      "learning_rate": 2.7e-07,
+      "loss": 2.4646,
+      "mean_token_accuracy": 0.46740976348519325,
+      "num_tokens": 4074956.0,
+      "step": 28
+    },
+    {
+      "entropy": 2.4283447265625,
+      "epoch": 0.0004992983996625432,
+      "grad_norm": 0.7344250679016113,
+      "learning_rate": 2.8e-07,
+      "loss": 2.425,
+      "mean_token_accuracy": 0.47232619673013687,
+      "num_tokens": 4218108.0,
+      "step": 29
+    },
+    {
+      "entropy": 2.453857421875,
+      "epoch": 0.0005165155858578033,
+      "grad_norm": 0.7299566268920898,
+      "learning_rate": 2.9000000000000003e-07,
+      "loss": 2.4514,
+      "mean_token_accuracy": 0.477730430662632,
+      "num_tokens": 4352208.0,
+      "step": 30
+    },
+    {
+      "entropy": 2.39404296875,
+      "epoch": 0.0005337327720530634,
+      "grad_norm": 0.6726910471916199,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.3704,
+      "mean_token_accuracy": 0.4892354430630803,
+      "num_tokens": 4510862.0,
+      "step": 31
+    },
+    {
+      "entropy": 2.37158203125,
+      "epoch": 0.0005509499582483235,
+      "grad_norm": 0.6483345627784729,
+      "learning_rate": 3.1000000000000005e-07,
+      "loss": 2.3378,
+      "mean_token_accuracy": 0.4861680665053427,
+      "num_tokens": 4655812.0,
+      "step": 32
+    },
+    {
+      "entropy": 2.4970703125,
+      "epoch": 0.0005681671444435836,
+      "grad_norm": 0.7244667410850525,
+      "learning_rate": 3.2e-07,
+      "loss": 2.4361,
+      "mean_token_accuracy": 0.4727164036594331,
+      "num_tokens": 4796670.0,
+      "step": 33
+    },
+    {
+      "entropy": 2.4234619140625,
+      "epoch": 0.0005853843306388437,
+      "grad_norm": 0.6697008013725281,
+      "learning_rate": 3.3e-07,
+      "loss": 2.4176,
+      "mean_token_accuracy": 0.48080282052978873,
+      "num_tokens": 4946476.0,
+      "step": 34
+    },
+    {
+      "entropy": 2.4354248046875,
+      "epoch": 0.0006026015168341038,
+      "grad_norm": 0.6681280732154846,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 2.3967,
+      "mean_token_accuracy": 0.48242951929569244,
+      "num_tokens": 5097062.0,
+      "step": 35
+    },
+    {
+      "entropy": 2.4337158203125,
+      "epoch": 0.000619818703029364,
+      "grad_norm": 0.6784984469413757,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 2.4061,
+      "mean_token_accuracy": 0.4787406297400594,
+      "num_tokens": 5241121.0,
+      "step": 36
+    },
+    {
+      "entropy": 2.3486328125,
+      "epoch": 0.000637035889224624,
+      "grad_norm": 0.7190965414047241,
+      "learning_rate": 3.6e-07,
+      "loss": 2.3033,
+      "mean_token_accuracy": 0.4977131159976125,
+      "num_tokens": 5379636.0,
+      "step": 37
+    },
+    {
+      "entropy": 2.400390625,
+      "epoch": 0.0006542530754198841,
+      "grad_norm": 0.7516965866088867,
+      "learning_rate": 3.7e-07,
+      "loss": 2.4019,
+      "mean_token_accuracy": 0.48473000153899193,
+      "num_tokens": 5513259.0,
+      "step": 38
+    },
+    {
+      "entropy": 2.4234619140625,
+      "epoch": 0.0006714702616151442,
+      "grad_norm": 0.6816972494125366,
+      "learning_rate": 3.8e-07,
+      "loss": 2.4078,
+      "mean_token_accuracy": 0.4816413172520697,
+      "num_tokens": 5657625.0,
+      "step": 39
+    },
+    {
+      "entropy": 2.4327392578125,
+      "epoch": 0.0006886874478104043,
+      "grad_norm": 0.7355881929397583,
+      "learning_rate": 3.9e-07,
+      "loss": 2.4284,
+      "mean_token_accuracy": 0.4844762939028442,
+      "num_tokens": 5801203.0,
+      "step": 40
+    },
+    {
+      "entropy": 2.4520263671875,
+      "epoch": 0.0007059046340056644,
+      "grad_norm": 0.6229036450386047,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.4646,
+      "mean_token_accuracy": 0.47269141068682075,
+      "num_tokens": 5953949.0,
+      "step": 41
+    },
+    {
+      "entropy": 2.4000244140625,
+      "epoch": 0.0007231218202009246,
+      "grad_norm": 1.3853771686553955,
+      "learning_rate": 4.1000000000000004e-07,
+      "loss": 2.415,
+      "mean_token_accuracy": 0.48673709062859416,
+      "num_tokens": 6124847.0,
+      "step": 42
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.0007403390063961847,
+      "grad_norm": 0.6690321564674377,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 2.4083,
+      "mean_token_accuracy": 0.4795906525105238,
+      "num_tokens": 6277763.0,
+      "step": 43
+    },
+    {
+      "entropy": 2.38916015625,
+      "epoch": 0.0007575561925914448,
+      "grad_norm": 0.6857286691665649,
+      "learning_rate": 4.3e-07,
+      "loss": 2.3703,
+      "mean_token_accuracy": 0.4900118997320533,
+      "num_tokens": 6426281.0,
+      "step": 44
+    },
+    {
+      "entropy": 2.4334716796875,
+      "epoch": 0.0007747733787867049,
+      "grad_norm": 0.7215378284454346,
+      "learning_rate": 4.4e-07,
+      "loss": 2.4171,
+      "mean_token_accuracy": 0.48149433452636003,
+      "num_tokens": 6569929.0,
+      "step": 45
+    },
+    {
+      "entropy": 2.4718017578125,
+      "epoch": 0.000791990564981965,
+      "grad_norm": 0.6353131532669067,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 2.4404,
+      "mean_token_accuracy": 0.4716165652498603,
+      "num_tokens": 6715254.0,
+      "step": 46
+    },
+    {
+      "entropy": 2.4588623046875,
+      "epoch": 0.0008092077511772251,
+      "grad_norm": 0.7047929763793945,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 2.4363,
+      "mean_token_accuracy": 0.47845502756536007,
+      "num_tokens": 6858114.0,
+      "step": 47
+    },
+    {
+      "entropy": 2.4326171875,
+      "epoch": 0.0008264249373724852,
+      "grad_norm": 0.6554984450340271,
+      "learning_rate": 4.7000000000000005e-07,
+      "loss": 2.4103,
+      "mean_token_accuracy": 0.48302431078627706,
+      "num_tokens": 7003888.0,
+      "step": 48
+    },
+    {
+      "entropy": 2.424072265625,
+      "epoch": 0.0008436421235677454,
+      "grad_norm": 0.615967333316803,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 2.3426,
+      "mean_token_accuracy": 0.48605213360860944,
+      "num_tokens": 7178232.0,
+      "step": 49
+    },
+    {
+      "entropy": 2.3641357421875,
+      "epoch": 0.0008608593097630055,
+      "grad_norm": 0.7089155316352844,
+      "learning_rate": 4.900000000000001e-07,
+      "loss": 2.3231,
+      "mean_token_accuracy": 0.49523815233260393,
+      "num_tokens": 7323317.0,
+      "step": 50
+    },
+    {
+      "entropy": 2.355224609375,
+      "epoch": 0.0008780764959582655,
+      "grad_norm": 0.6622390151023865,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.328,
+      "mean_token_accuracy": 0.49393986631184816,
+      "num_tokens": 7473982.0,
+      "step": 51
+    },
+    {
+      "entropy": 2.41357421875,
+      "epoch": 0.0008952936821535256,
+      "grad_norm": 0.7162524461746216,
+      "learning_rate": 5.1e-07,
+      "loss": 2.3915,
+      "mean_token_accuracy": 0.4822638025507331,
+      "num_tokens": 7618445.0,
+      "step": 52
+    },
+    {
+      "entropy": 2.4310302734375,
+      "epoch": 0.0009125108683487857,
+      "grad_norm": 0.7647550106048584,
+      "learning_rate": 5.2e-07,
+      "loss": 2.3974,
+      "mean_token_accuracy": 0.4842324573546648,
+      "num_tokens": 7759148.0,
+      "step": 53
+    },
+    {
+      "entropy": 2.4063720703125,
+      "epoch": 0.0009297280545440458,
+      "grad_norm": 0.6380776166915894,
+      "learning_rate": 5.3e-07,
+      "loss": 2.4044,
+      "mean_token_accuracy": 0.48621372459456325,
+      "num_tokens": 7909476.0,
+      "step": 54
+    },
+    {
+      "entropy": 2.40234375,
+      "epoch": 0.000946945240739306,
+      "grad_norm": 0.6949607729911804,
+      "learning_rate": 5.4e-07,
+      "loss": 2.3508,
+      "mean_token_accuracy": 0.48716961592435837,
+      "num_tokens": 8052542.0,
+      "step": 55
+    },
+    {
+      "entropy": 2.361328125,
+      "epoch": 0.0009641624269345661,
+      "grad_norm": 0.6398380398750305,
+      "learning_rate": 5.5e-07,
+      "loss": 2.298,
+      "mean_token_accuracy": 0.49046063888818026,
+      "num_tokens": 8198359.0,
+      "step": 56
+    },
+    {
+      "entropy": 2.424560546875,
+      "epoch": 0.000981379613129826,
+      "grad_norm": 0.6661210060119629,
+      "learning_rate": 5.6e-07,
+      "loss": 2.3812,
+      "mean_token_accuracy": 0.4842396741732955,
+      "num_tokens": 8350561.0,
+      "step": 57
+    },
+    {
+      "entropy": 2.42333984375,
+      "epoch": 0.0009985967993250864,
+      "grad_norm": 0.6568598747253418,
+      "learning_rate": 5.7e-07,
+      "loss": 2.386,
+      "mean_token_accuracy": 0.4811908514238894,
+      "num_tokens": 8498421.0,
+      "step": 58
+    },
+    {
+      "entropy": 2.4390869140625,
+      "epoch": 0.0010158139855203465,
+      "grad_norm": 0.6521676778793335,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 2.4007,
+      "mean_token_accuracy": 0.47653205832466483,
+      "num_tokens": 8645838.0,
+      "step": 59
+    },
+    {
+      "entropy": 2.4642333984375,
+      "epoch": 0.0010330311717156066,
+      "grad_norm": 0.6818094849586487,
+      "learning_rate": 5.900000000000001e-07,
+      "loss": 2.4183,
+      "mean_token_accuracy": 0.47624633833765984,
+      "num_tokens": 8781213.0,
+      "step": 60
+    },
+    {
+      "entropy": 2.4544677734375,
+      "epoch": 0.0010502483579108667,
+      "grad_norm": 0.7099837064743042,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.429,
+      "mean_token_accuracy": 0.4788547111675143,
+      "num_tokens": 8932181.0,
+      "step": 61
+    },
+    {
+      "entropy": 2.475830078125,
+      "epoch": 0.0010674655441061268,
+      "grad_norm": 0.6400408744812012,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 2.452,
+      "mean_token_accuracy": 0.4709730679169297,
+      "num_tokens": 9069468.0,
+      "step": 62
+    },
+    {
+      "entropy": 2.4625244140625,
+      "epoch": 0.0010846827303013869,
+      "grad_norm": 0.6099865436553955,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 2.4152,
+      "mean_token_accuracy": 0.4845339651219547,
+      "num_tokens": 9238211.0,
+      "step": 63
+    },
+    {
+      "entropy": 2.4390869140625,
+      "epoch": 0.001101899916496647,
+      "grad_norm": 0.6318409442901611,
+      "learning_rate": 6.3e-07,
+      "loss": 2.402,
+      "mean_token_accuracy": 0.4825730072334409,
+      "num_tokens": 9393830.0,
+      "step": 64
+    },
+    {
+      "entropy": 2.3837890625,
+      "epoch": 0.001119117102691907,
+      "grad_norm": 0.6888275742530823,
+      "learning_rate": 6.4e-07,
+      "loss": 2.3289,
+      "mean_token_accuracy": 0.4893337092362344,
+      "num_tokens": 9536691.0,
+      "step": 65
+    },
+    {
+      "entropy": 2.476806640625,
+      "epoch": 0.0011363342888871671,
+      "grad_norm": 0.6647155284881592,
+      "learning_rate": 6.5e-07,
+      "loss": 2.4071,
+      "mean_token_accuracy": 0.47626868123188615,
+      "num_tokens": 9672776.0,
+      "step": 66
+    },
+    {
+      "entropy": 2.4517822265625,
+      "epoch": 0.0011535514750824272,
+      "grad_norm": 0.6950869560241699,
+      "learning_rate": 6.6e-07,
+      "loss": 2.4153,
+      "mean_token_accuracy": 0.4799360786564648,
+      "num_tokens": 9823719.0,
+      "step": 67
+    },
+    {
+      "entropy": 2.4583740234375,
+      "epoch": 0.0011707686612776873,
+      "grad_norm": 0.5642852783203125,
+      "learning_rate": 6.7e-07,
+      "loss": 2.4354,
+      "mean_token_accuracy": 0.47825055569410324,
+      "num_tokens": 9979166.0,
+      "step": 68
+    },
+    {
+      "entropy": 2.503173828125,
+      "epoch": 0.0011879858474729474,
+      "grad_norm": 0.6830074787139893,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 2.4759,
+      "mean_token_accuracy": 0.46751530282199383,
+      "num_tokens": 10123771.0,
+      "step": 69
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.0012052030336682075,
+      "grad_norm": 0.611301600933075,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 2.3415,
+      "mean_token_accuracy": 0.4828020860441029,
+      "num_tokens": 10266978.0,
+      "step": 70
+    },
+    {
+      "entropy": 2.393798828125,
+      "epoch": 0.0012224202198634678,
+      "grad_norm": 0.6329925060272217,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.3527,
+      "mean_token_accuracy": 0.4885860946960747,
+      "num_tokens": 10411005.0,
+      "step": 71
+    },
+    {
+      "entropy": 2.4456787109375,
+      "epoch": 0.001239637406058728,
+      "grad_norm": 0.6201856732368469,
+      "learning_rate": 7.1e-07,
+      "loss": 2.3725,
+      "mean_token_accuracy": 0.4878034754656255,
+      "num_tokens": 10560620.0,
+      "step": 72
+    },
+    {
+      "entropy": 2.50439453125,
+      "epoch": 0.001256854592253988,
+      "grad_norm": 0.6518511772155762,
+      "learning_rate": 7.2e-07,
+      "loss": 2.4772,
+      "mean_token_accuracy": 0.47327897092327476,
+      "num_tokens": 10692258.0,
+      "step": 73
+    },
+    {
+      "entropy": 2.5009765625,
+      "epoch": 0.001274071778449248,
+      "grad_norm": 0.6756422519683838,
+      "learning_rate": 7.3e-07,
+      "loss": 2.5074,
+      "mean_token_accuracy": 0.4726545801386237,
+      "num_tokens": 10841410.0,
+      "step": 74
+    },
+    {
+      "entropy": 2.47998046875,
+      "epoch": 0.0012912889646445082,
+      "grad_norm": 0.5928777456283569,
+      "learning_rate": 7.4e-07,
+      "loss": 2.4521,
+      "mean_token_accuracy": 0.4727823534049094,
+      "num_tokens": 10996060.0,
+      "step": 75
+    },
+    {
+      "entropy": 2.499755859375,
+      "epoch": 0.0013085061508397683,
+      "grad_norm": 0.6060748100280762,
+      "learning_rate": 7.5e-07,
+      "loss": 2.4569,
+      "mean_token_accuracy": 0.4745932733640075,
+      "num_tokens": 11146252.0,
+      "step": 76
+    },
+    {
+      "entropy": 2.429443359375,
+      "epoch": 0.0013257233370350284,
+      "grad_norm": 0.633307933807373,
+      "learning_rate": 7.6e-07,
+      "loss": 2.3987,
+      "mean_token_accuracy": 0.4819001527503133,
+      "num_tokens": 11287713.0,
+      "step": 77
+    },
+    {
+      "entropy": 2.38232421875,
+      "epoch": 0.0013429405232302885,
+      "grad_norm": 0.6448901295661926,
+      "learning_rate": 7.7e-07,
+      "loss": 2.3849,
+      "mean_token_accuracy": 0.4902081396430731,
+      "num_tokens": 11427731.0,
+      "step": 78
+    },
+    {
+      "entropy": 2.4320068359375,
+      "epoch": 0.0013601577094255485,
+      "grad_norm": 0.7015244364738464,
+      "learning_rate": 7.8e-07,
+      "loss": 2.3898,
+      "mean_token_accuracy": 0.48666849778965116,
+      "num_tokens": 11570107.0,
+      "step": 79
+    },
+    {
+      "entropy": 2.47998046875,
+      "epoch": 0.0013773748956208086,
+      "grad_norm": 0.6528029441833496,
+      "learning_rate": 7.900000000000001e-07,
+      "loss": 2.4978,
+      "mean_token_accuracy": 0.4753390052355826,
+      "num_tokens": 11704020.0,
+      "step": 80
+    },
+    {
+      "entropy": 2.433837890625,
+      "epoch": 0.0013945920818160687,
+      "grad_norm": 0.7266194820404053,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.3828,
+      "mean_token_accuracy": 0.479493273422122,
+      "num_tokens": 11849459.0,
+      "step": 81
+    },
+    {
+      "entropy": 2.4395751953125,
+      "epoch": 0.0014118092680113288,
+      "grad_norm": 0.6236125230789185,
+      "learning_rate": 8.100000000000001e-07,
+      "loss": 2.3505,
+      "mean_token_accuracy": 0.48662899900227785,
+      "num_tokens": 12010828.0,
+      "step": 82
+    },
+    {
+      "entropy": 2.475830078125,
+      "epoch": 0.001429026454206589,
+      "grad_norm": 0.7149572968482971,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 2.4566,
+      "mean_token_accuracy": 0.47693332051858306,
+      "num_tokens": 12157399.0,
+      "step": 83
+    },
+    {
+      "entropy": 2.438720703125,
+      "epoch": 0.0014462436404018492,
+      "grad_norm": 0.6647018790245056,
+      "learning_rate": 8.300000000000001e-07,
+      "loss": 2.3986,
+      "mean_token_accuracy": 0.4824375621974468,
+      "num_tokens": 12287842.0,
+      "step": 84
+    },
+    {
+      "entropy": 2.41455078125,
+      "epoch": 0.0014634608265971093,
+      "grad_norm": 0.6005454063415527,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 2.3891,
+      "mean_token_accuracy": 0.4884789031930268,
+      "num_tokens": 12436104.0,
+      "step": 85
+    },
+    {
+      "entropy": 2.4703369140625,
+      "epoch": 0.0014806780127923694,
+      "grad_norm": 0.6212813854217529,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 2.34,
+      "mean_token_accuracy": 0.4844899824820459,
+      "num_tokens": 12575067.0,
+      "step": 86
+    },
+    {
+      "entropy": 2.6005859375,
+      "epoch": 0.0014978951989876295,
+      "grad_norm": 0.8042929172515869,
+      "learning_rate": 8.6e-07,
+      "loss": 2.6196,
+      "mean_token_accuracy": 0.4688438312150538,
+      "num_tokens": 12730536.0,
+      "step": 87
+    },
+    {
+      "entropy": 2.4884033203125,
+      "epoch": 0.0015151123851828896,
+      "grad_norm": 0.5991901755332947,
+      "learning_rate": 8.7e-07,
+      "loss": 2.4611,
+      "mean_token_accuracy": 0.4766750931739807,
+      "num_tokens": 12882514.0,
+      "step": 88
+    },
+    {
+      "entropy": 2.4810791015625,
+      "epoch": 0.0015323295713781497,
+      "grad_norm": 0.6494613289833069,
+      "learning_rate": 8.8e-07,
+      "loss": 2.4435,
+      "mean_token_accuracy": 0.47679867735132575,
+      "num_tokens": 13023006.0,
+      "step": 89
+    },
+    {
+      "entropy": 2.4412841796875,
+      "epoch": 0.0015495467575734098,
+      "grad_norm": 0.6427425146102905,
+      "learning_rate": 8.900000000000001e-07,
+      "loss": 2.3905,
+      "mean_token_accuracy": 0.4838052117265761,
+      "num_tokens": 13156996.0,
+      "step": 90
+    },
+    {
+      "entropy": 2.394775390625,
+      "epoch": 0.0015667639437686699,
+      "grad_norm": 0.6279881596565247,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 2.3753,
+      "mean_token_accuracy": 0.48632694967091084,
+      "num_tokens": 13299631.0,
+      "step": 91
+    },
+    {
+      "entropy": 2.4176025390625,
+      "epoch": 0.00158398112996393,
+      "grad_norm": 0.6744757294654846,
+      "learning_rate": 9.100000000000001e-07,
+      "loss": 2.3854,
+      "mean_token_accuracy": 0.4846075074747205,
+      "num_tokens": 13440366.0,
+      "step": 92
+    },
+    {
+      "entropy": 2.4541015625,
+      "epoch": 0.00160119831615919,
+      "grad_norm": 0.6708775758743286,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.4069,
+      "mean_token_accuracy": 0.481810312718153,
+      "num_tokens": 13566615.0,
+      "step": 93
+    },
+    {
+      "entropy": 2.443115234375,
+      "epoch": 0.0016184155023544501,
+      "grad_norm": 0.5979477167129517,
+      "learning_rate": 9.300000000000001e-07,
+      "loss": 2.3846,
+      "mean_token_accuracy": 0.48044557217508554,
+      "num_tokens": 13719522.0,
+      "step": 94
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.0016356326885497102,
+      "grad_norm": 0.6026092171669006,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 2.4051,
+      "mean_token_accuracy": 0.4736237172037363,
+      "num_tokens": 13864674.0,
+      "step": 95
+    },
+    {
+      "entropy": 2.3526611328125,
+      "epoch": 0.0016528498747449703,
+      "grad_norm": 0.5873211622238159,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 2.2743,
+      "mean_token_accuracy": 0.49419589480385184,
+      "num_tokens": 14027711.0,
+      "step": 96
+    },
+    {
+      "entropy": 2.460693359375,
+      "epoch": 0.0016700670609402306,
+      "grad_norm": 0.5936735272407532,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.3992,
+      "mean_token_accuracy": 0.4806458824314177,
+      "num_tokens": 14172381.0,
+      "step": 97
+    },
+    {
+      "entropy": 2.45458984375,
+      "epoch": 0.0016872842471354907,
+      "grad_norm": 1.1826030015945435,
+      "learning_rate": 9.7e-07,
+      "loss": 2.4255,
+      "mean_token_accuracy": 0.4772787392139435,
+      "num_tokens": 14309500.0,
+      "step": 98
+    },
+    {
+      "entropy": 2.51806640625,
+      "epoch": 0.0017045014333307508,
+      "grad_norm": 0.6550461053848267,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.4692,
+      "mean_token_accuracy": 0.47163511207327247,
+      "num_tokens": 14449340.0,
+      "step": 99
+    },
+    {
+      "entropy": 2.446533203125,
+      "epoch": 0.001721718619526011,
+      "grad_norm": 0.5825348496437073,
+      "learning_rate": 9.9e-07,
+      "loss": 2.4099,
+      "mean_token_accuracy": 0.47821634402498603,
+      "num_tokens": 14600293.0,
+      "step": 100
+    },
+    {
+      "entropy": 2.454833984375,
+      "epoch": 0.001738935805721271,
+      "grad_norm": 0.5967729091644287,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.4471,
+      "mean_token_accuracy": 0.4838520339690149,
+      "num_tokens": 14754154.0,
+      "step": 101
+    },
+    {
+      "entropy": 2.470703125,
+      "epoch": 0.001756152991916531,
+      "grad_norm": 0.7432575225830078,
+      "learning_rate": 1.01e-06,
+      "loss": 2.4286,
+      "mean_token_accuracy": 0.48514684848487377,
+      "num_tokens": 14912580.0,
+      "step": 102
+    },
+    {
+      "entropy": 2.5068359375,
+      "epoch": 0.0017733701781117912,
+      "grad_norm": 0.6083717942237854,
+      "learning_rate": 1.02e-06,
+      "loss": 2.4618,
+      "mean_token_accuracy": 0.47291735280305147,
+      "num_tokens": 15054619.0,
+      "step": 103
+    },
+    {
+      "entropy": 2.37841796875,
+      "epoch": 0.0017905873643070513,
+      "grad_norm": 0.664107084274292,
+      "learning_rate": 1.03e-06,
+      "loss": 2.3513,
+      "mean_token_accuracy": 0.49140653293579817,
+      "num_tokens": 15184259.0,
+      "step": 104
+    },
+    {
+      "entropy": 2.408935546875,
+      "epoch": 0.0018078045505023114,
+      "grad_norm": 0.6227862238883972,
+      "learning_rate": 1.04e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.4819969036616385,
+      "num_tokens": 15328191.0,
+      "step": 105
+    },
+    {
+      "entropy": 2.4210205078125,
+      "epoch": 0.0018250217366975715,
+      "grad_norm": 0.5873227119445801,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 2.3571,
+      "mean_token_accuracy": 0.4860111135058105,
+      "num_tokens": 15484579.0,
+      "step": 106
+    },
+    {
+      "entropy": 2.4600830078125,
+      "epoch": 0.0018422389228928315,
+      "grad_norm": 0.7568690776824951,
+      "learning_rate": 1.06e-06,
+      "loss": 2.4228,
+      "mean_token_accuracy": 0.4871084992773831,
+      "num_tokens": 15621836.0,
+      "step": 107
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.0018594561090880916,
+      "grad_norm": 0.5804514288902283,
+      "learning_rate": 1.0700000000000001e-06,
+      "loss": 2.4028,
+      "mean_token_accuracy": 0.4824648411013186,
+      "num_tokens": 15778932.0,
+      "step": 108
+    },
+    {
+      "entropy": 2.455810546875,
+      "epoch": 0.001876673295283352,
+      "grad_norm": 0.5958721041679382,
+      "learning_rate": 1.08e-06,
+      "loss": 2.3833,
+      "mean_token_accuracy": 0.48080389108508825,
+      "num_tokens": 15924525.0,
+      "step": 109
+    },
+    {
+      "entropy": 2.3779296875,
+      "epoch": 0.001893890481478612,
+      "grad_norm": 0.6376572251319885,
+      "learning_rate": 1.0900000000000002e-06,
+      "loss": 2.3314,
+      "mean_token_accuracy": 0.4905442167073488,
+      "num_tokens": 16080275.0,
+      "step": 110
+    },
+    {
+      "entropy": 2.4273681640625,
+      "epoch": 0.0019111076676738721,
+      "grad_norm": 0.6554363369941711,
+      "learning_rate": 1.1e-06,
+      "loss": 2.4002,
+      "mean_token_accuracy": 0.4839071067981422,
+      "num_tokens": 16212132.0,
+      "step": 111
+    },
+    {
+      "entropy": 2.4451904296875,
+      "epoch": 0.0019283248538691322,
+      "grad_norm": 0.6245486736297607,
+      "learning_rate": 1.1100000000000002e-06,
+      "loss": 2.4013,
+      "mean_token_accuracy": 0.4785764031112194,
+      "num_tokens": 16346241.0,
+      "step": 112
+    },
+    {
+      "entropy": 2.44677734375,
+      "epoch": 0.0019455420400643923,
+      "grad_norm": 0.6356763243675232,
+      "learning_rate": 1.12e-06,
+      "loss": 2.3992,
+      "mean_token_accuracy": 0.4817821686156094,
+      "num_tokens": 16477490.0,
+      "step": 113
+    },
+    {
+      "entropy": 2.480224609375,
+      "epoch": 0.001962759226259652,
+      "grad_norm": 0.7369113564491272,
+      "learning_rate": 1.1300000000000002e-06,
+      "loss": 2.4604,
+      "mean_token_accuracy": 0.4817192433401942,
+      "num_tokens": 16635712.0,
+      "step": 114
+    },
+    {
+      "entropy": 2.446533203125,
+      "epoch": 0.0019799764124549123,
+      "grad_norm": 0.5834375023841858,
+      "learning_rate": 1.14e-06,
+      "loss": 2.3876,
+      "mean_token_accuracy": 0.48327695531770587,
+      "num_tokens": 16790029.0,
+      "step": 115
+    },
+    {
+      "entropy": 2.43994140625,
+      "epoch": 0.001997193598650173,
+      "grad_norm": 0.589958906173706,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 2.3869,
+      "mean_token_accuracy": 0.4807235752232373,
+      "num_tokens": 16939941.0,
+      "step": 116
+    },
+    {
+      "entropy": 2.4541015625,
+      "epoch": 0.002014410784845433,
+      "grad_norm": 0.5870490074157715,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.3979,
+      "mean_token_accuracy": 0.47932128235697746,
+      "num_tokens": 17090653.0,
+      "step": 117
+    },
+    {
+      "entropy": 2.4052734375,
+      "epoch": 0.002031627971040693,
+      "grad_norm": 0.6636145710945129,
+      "learning_rate": 1.1700000000000002e-06,
+      "loss": 2.3499,
+      "mean_token_accuracy": 0.48607511818408966,
+      "num_tokens": 17225388.0,
+      "step": 118
+    },
+    {
+      "entropy": 2.4976806640625,
+      "epoch": 0.002048845157235953,
+      "grad_norm": 0.8917343020439148,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.4271,
+      "mean_token_accuracy": 0.47680498752743006,
+      "num_tokens": 17365012.0,
+      "step": 119
+    },
+    {
+      "entropy": 2.4691162109375,
+      "epoch": 0.002066062343431213,
+      "grad_norm": 0.623699963092804,
+      "learning_rate": 1.19e-06,
+      "loss": 2.4145,
+      "mean_token_accuracy": 0.4758805222809315,
+      "num_tokens": 17515672.0,
+      "step": 120
+    },
+    {
+      "entropy": 2.508544921875,
+      "epoch": 0.0020832795296264733,
+      "grad_norm": 0.5833603143692017,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.4093,
+      "mean_token_accuracy": 0.48172463616356254,
+      "num_tokens": 17664187.0,
+      "step": 121
+    },
+    {
+      "entropy": 2.4619140625,
+      "epoch": 0.0021004967158217334,
+      "grad_norm": 0.6505944728851318,
+      "learning_rate": 1.21e-06,
+      "loss": 2.4608,
+      "mean_token_accuracy": 0.47443893272429705,
+      "num_tokens": 17806861.0,
+      "step": 122
+    },
+    {
+      "entropy": 2.5152587890625,
+      "epoch": 0.0021177139020169935,
+      "grad_norm": 0.6009451150894165,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.4738,
+      "mean_token_accuracy": 0.4736114493571222,
+      "num_tokens": 17949930.0,
+      "step": 123
+    },
+    {
+      "entropy": 2.415771484375,
+      "epoch": 0.0021349310882122535,
+      "grad_norm": 0.5856963992118835,
+      "learning_rate": 1.23e-06,
+      "loss": 2.3834,
+      "mean_token_accuracy": 0.4865727727301419,
+      "num_tokens": 18096555.0,
+      "step": 124
+    },
+    {
+      "entropy": 2.3616943359375,
+      "epoch": 0.0021521482744075136,
+      "grad_norm": 0.5739309787750244,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.3385,
+      "mean_token_accuracy": 0.49280355405062437,
+      "num_tokens": 18249345.0,
+      "step": 125
+    },
+    {
+      "entropy": 2.4732666015625,
+      "epoch": 0.0021693654606027737,
+      "grad_norm": 0.6043084859848022,
+      "learning_rate": 1.25e-06,
+      "loss": 2.416,
+      "mean_token_accuracy": 0.47766703460365534,
+      "num_tokens": 18385525.0,
+      "step": 126
+    },
+    {
+      "entropy": 2.5040283203125,
+      "epoch": 0.002186582646798034,
+      "grad_norm": 0.6954610347747803,
+      "learning_rate": 1.26e-06,
+      "loss": 2.4997,
+      "mean_token_accuracy": 0.4837381485849619,
+      "num_tokens": 18522666.0,
+      "step": 127
+    },
+    {
+      "entropy": 2.45166015625,
+      "epoch": 0.002203799832993294,
+      "grad_norm": 0.5906988382339478,
+      "learning_rate": 1.2700000000000001e-06,
+      "loss": 2.4099,
+      "mean_token_accuracy": 0.47865421138703823,
+      "num_tokens": 18662132.0,
+      "step": 128
+    },
+    {
+      "entropy": 2.395263671875,
+      "epoch": 0.002221017019188554,
+      "grad_norm": 0.5896300077438354,
+      "learning_rate": 1.28e-06,
+      "loss": 2.3787,
+      "mean_token_accuracy": 0.49431150034070015,
+      "num_tokens": 18803105.0,
+      "step": 129
+    },
+    {
+      "entropy": 2.435546875,
+      "epoch": 0.002238234205383814,
+      "grad_norm": 0.618240475654602,
+      "learning_rate": 1.2900000000000001e-06,
+      "loss": 2.3837,
+      "mean_token_accuracy": 0.48449931014329195,
+      "num_tokens": 18942086.0,
+      "step": 130
+    },
+    {
+      "entropy": 2.387451171875,
+      "epoch": 0.002255451391579074,
+      "grad_norm": 0.5633107423782349,
+      "learning_rate": 1.3e-06,
+      "loss": 2.3315,
+      "mean_token_accuracy": 0.4937907229177654,
+      "num_tokens": 19097977.0,
+      "step": 131
+    },
+    {
+      "entropy": 2.431640625,
+      "epoch": 0.0022726685777743343,
+      "grad_norm": 0.5887622833251953,
+      "learning_rate": 1.3100000000000002e-06,
+      "loss": 2.381,
+      "mean_token_accuracy": 0.48834379855543375,
+      "num_tokens": 19242582.0,
+      "step": 132
+    },
+    {
+      "entropy": 2.476806640625,
+      "epoch": 0.0022898857639695944,
+      "grad_norm": 0.633418083190918,
+      "learning_rate": 1.32e-06,
+      "loss": 2.4313,
+      "mean_token_accuracy": 0.4792938116006553,
+      "num_tokens": 19379757.0,
+      "step": 133
+    },
+    {
+      "entropy": 2.42333984375,
+      "epoch": 0.0023071029501648545,
+      "grad_norm": 0.6043598651885986,
+      "learning_rate": 1.3300000000000002e-06,
+      "loss": 2.343,
+      "mean_token_accuracy": 0.4908184530213475,
+      "num_tokens": 19524622.0,
+      "step": 134
+    },
+    {
+      "entropy": 2.46484375,
+      "epoch": 0.0023243201363601146,
+      "grad_norm": 0.6375739574432373,
+      "learning_rate": 1.34e-06,
+      "loss": 2.4083,
+      "mean_token_accuracy": 0.4828670499846339,
+      "num_tokens": 19663378.0,
+      "step": 135
+    },
+    {
+      "entropy": 2.428955078125,
+      "epoch": 0.0023415373225553746,
+      "grad_norm": 0.5688341856002808,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 2.3791,
+      "mean_token_accuracy": 0.48436517268419266,
+      "num_tokens": 19816124.0,
+      "step": 136
+    },
+    {
+      "entropy": 2.4317626953125,
+      "epoch": 0.0023587545087506347,
+      "grad_norm": 0.5994829535484314,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.3781,
+      "mean_token_accuracy": 0.4828225467354059,
+      "num_tokens": 19958536.0,
+      "step": 137
+    },
+    {
+      "entropy": 2.4664306640625,
+      "epoch": 0.002375971694945895,
+      "grad_norm": 0.5764389634132385,
+      "learning_rate": 1.3700000000000002e-06,
+      "loss": 2.4281,
+      "mean_token_accuracy": 0.47715017944574356,
+      "num_tokens": 20101195.0,
+      "step": 138
+    },
+    {
+      "entropy": 2.515869140625,
+      "epoch": 0.002393188881141155,
+      "grad_norm": 0.5928436517715454,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.4664,
+      "mean_token_accuracy": 0.4714601272717118,
+      "num_tokens": 20250847.0,
+      "step": 139
+    },
+    {
+      "entropy": 2.47509765625,
+      "epoch": 0.002410406067336415,
+      "grad_norm": 0.7260881662368774,
+      "learning_rate": 1.3900000000000002e-06,
+      "loss": 2.4494,
+      "mean_token_accuracy": 0.47532156156376004,
+      "num_tokens": 20391934.0,
+      "step": 140
+    },
+    {
+      "entropy": 2.42236328125,
+      "epoch": 0.002427623253531675,
+      "grad_norm": 0.5777806043624878,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.3889,
+      "mean_token_accuracy": 0.4810917223803699,
+      "num_tokens": 20542473.0,
+      "step": 141
+    },
+    {
+      "entropy": 2.426025390625,
+      "epoch": 0.0024448404397269356,
+      "grad_norm": 0.5795040130615234,
+      "learning_rate": 1.41e-06,
+      "loss": 2.357,
+      "mean_token_accuracy": 0.4965799758210778,
+      "num_tokens": 20692496.0,
+      "step": 142
+    },
+    {
+      "entropy": 2.446044921875,
+      "epoch": 0.0024620576259221957,
+      "grad_norm": 0.584563672542572,
+      "learning_rate": 1.42e-06,
+      "loss": 2.3614,
+      "mean_token_accuracy": 0.4806175325065851,
+      "num_tokens": 20829269.0,
+      "step": 143
+    },
+    {
+      "entropy": 2.431640625,
+      "epoch": 0.002479274812117456,
+      "grad_norm": 0.5803564190864563,
+      "learning_rate": 1.43e-06,
+      "loss": 2.3908,
+      "mean_token_accuracy": 0.4852483980357647,
+      "num_tokens": 20976041.0,
+      "step": 144
+    },
+    {
+      "entropy": 2.4169921875,
+      "epoch": 0.002496491998312716,
+      "grad_norm": 0.5736103057861328,
+      "learning_rate": 1.44e-06,
+      "loss": 2.407,
+      "mean_token_accuracy": 0.48149813804775476,
+      "num_tokens": 21123880.0,
+      "step": 145
+    },
+    {
+      "entropy": 2.501708984375,
+      "epoch": 0.002513709184507976,
+      "grad_norm": 0.5667904615402222,
+      "learning_rate": 1.45e-06,
+      "loss": 2.4569,
+      "mean_token_accuracy": 0.470580879598856,
+      "num_tokens": 21267685.0,
+      "step": 146
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.002530926370703236,
+      "grad_norm": 0.6545902490615845,
+      "learning_rate": 1.46e-06,
+      "loss": 2.368,
+      "mean_token_accuracy": 0.4863086869008839,
+      "num_tokens": 21415641.0,
+      "step": 147
+    },
+    {
+      "entropy": 2.4876708984375,
+      "epoch": 0.002548143556898496,
+      "grad_norm": 0.6045504808425903,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 2.4502,
+      "mean_token_accuracy": 0.47358084423467517,
+      "num_tokens": 21568579.0,
+      "step": 148
+    },
+    {
+      "entropy": 2.4735107421875,
+      "epoch": 0.0025653607430937563,
+      "grad_norm": 0.5392025113105774,
+      "learning_rate": 1.48e-06,
+      "loss": 2.3939,
+      "mean_token_accuracy": 0.47306955326348543,
+      "num_tokens": 21725147.0,
+      "step": 149
+    },
+    {
+      "entropy": 2.38037109375,
+      "epoch": 0.0025825779292890164,
+      "grad_norm": 0.6125035285949707,
+      "learning_rate": 1.4900000000000001e-06,
+      "loss": 2.3062,
+      "mean_token_accuracy": 0.49953836342319846,
+      "num_tokens": 21861243.0,
+      "step": 150
+    },
+    {
+      "entropy": 2.40234375,
+      "epoch": 0.0025997951154842765,
+      "grad_norm": 0.6005491614341736,
+      "learning_rate": 1.5e-06,
+      "loss": 2.3803,
+      "mean_token_accuracy": 0.483843975700438,
+      "num_tokens": 21996041.0,
+      "step": 151
+    },
+    {
+      "entropy": 2.51171875,
+      "epoch": 0.0026170123016795365,
+      "grad_norm": 0.6047178506851196,
+      "learning_rate": 1.5100000000000002e-06,
+      "loss": 2.4551,
+      "mean_token_accuracy": 0.4724902934394777,
+      "num_tokens": 22124115.0,
+      "step": 152
+    },
+    {
+      "entropy": 2.4654541015625,
+      "epoch": 0.0026342294878747966,
+      "grad_norm": 0.5984538197517395,
+      "learning_rate": 1.52e-06,
+      "loss": 2.4066,
+      "mean_token_accuracy": 0.47982197999954224,
+      "num_tokens": 22268397.0,
+      "step": 153
+    },
+    {
+      "entropy": 2.5594482421875,
+      "epoch": 0.0026514466740700567,
+      "grad_norm": 0.5999759435653687,
+      "learning_rate": 1.5300000000000002e-06,
+      "loss": 2.5766,
+      "mean_token_accuracy": 0.46560019347816706,
+      "num_tokens": 22417806.0,
+      "step": 154
+    },
+    {
+      "entropy": 2.468017578125,
+      "epoch": 0.002668663860265317,
+      "grad_norm": 0.5409292578697205,
+      "learning_rate": 1.54e-06,
+      "loss": 2.4253,
+      "mean_token_accuracy": 0.471966958604753,
+      "num_tokens": 22581097.0,
+      "step": 155
+    },
+    {
+      "entropy": 2.4951171875,
+      "epoch": 0.002685881046460577,
+      "grad_norm": 0.5459778904914856,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 2.4459,
+      "mean_token_accuracy": 0.47298973286524415,
+      "num_tokens": 22735616.0,
+      "step": 156
+    },
+    {
+      "entropy": 2.4017333984375,
+      "epoch": 0.002703098232655837,
+      "grad_norm": 0.5974356532096863,
+      "learning_rate": 1.56e-06,
+      "loss": 2.3332,
+      "mean_token_accuracy": 0.49170317640528083,
+      "num_tokens": 22876476.0,
+      "step": 157
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.002720315418851097,
+      "grad_norm": 0.6058854460716248,
+      "learning_rate": 1.5700000000000002e-06,
+      "loss": 2.3759,
+      "mean_token_accuracy": 0.48308438109233975,
+      "num_tokens": 23000697.0,
+      "step": 158
+    },
+    {
+      "entropy": 2.4227294921875,
+      "epoch": 0.002737532605046357,
+      "grad_norm": 0.6649252772331238,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.3364,
+      "mean_token_accuracy": 0.49115608306601644,
+      "num_tokens": 23128424.0,
+      "step": 159
+    },
+    {
+      "entropy": 2.459716796875,
+      "epoch": 0.0027547497912416173,
+      "grad_norm": 0.571075439453125,
+      "learning_rate": 1.5900000000000002e-06,
+      "loss": 2.4113,
+      "mean_token_accuracy": 0.4818022232502699,
+      "num_tokens": 23269190.0,
+      "step": 160
+    },
+    {
+      "entropy": 2.45654296875,
+      "epoch": 0.0027719669774368774,
+      "grad_norm": 0.5999649167060852,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.4138,
+      "mean_token_accuracy": 0.4814398717135191,
+      "num_tokens": 23423206.0,
+      "step": 161
+    },
+    {
+      "entropy": 2.4156494140625,
+      "epoch": 0.0027891841636321375,
+      "grad_norm": 0.5621162056922913,
+      "learning_rate": 1.6100000000000003e-06,
+      "loss": 2.3751,
+      "mean_token_accuracy": 0.48285360960289836,
+      "num_tokens": 23572561.0,
+      "step": 162
+    },
+    {
+      "entropy": 2.3802490234375,
+      "epoch": 0.0028064013498273976,
+      "grad_norm": 0.6032527089118958,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.3422,
+      "mean_token_accuracy": 0.4902229546569288,
+      "num_tokens": 23718051.0,
+      "step": 163
+    },
+    {
+      "entropy": 2.370849609375,
+      "epoch": 0.0028236185360226576,
+      "grad_norm": 0.55767822265625,
+      "learning_rate": 1.6300000000000003e-06,
+      "loss": 2.3473,
+      "mean_token_accuracy": 0.49248579889535904,
+      "num_tokens": 23877563.0,
+      "step": 164
+    },
+    {
+      "entropy": 2.5054931640625,
+      "epoch": 0.0028408357222179177,
+      "grad_norm": 0.5944464206695557,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.4374,
+      "mean_token_accuracy": 0.47586293099448085,
+      "num_tokens": 24013173.0,
+      "step": 165
+    },
+    {
+      "entropy": 2.4613037109375,
+      "epoch": 0.002858052908413178,
+      "grad_norm": 0.6158614754676819,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 2.4078,
+      "mean_token_accuracy": 0.48460292909294367,
+      "num_tokens": 24150366.0,
+      "step": 166
+    },
+    {
+      "entropy": 2.4307861328125,
+      "epoch": 0.0028752700946084384,
+      "grad_norm": 0.562500536441803,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.3867,
+      "mean_token_accuracy": 0.4834507182240486,
+      "num_tokens": 24300680.0,
+      "step": 167
+    },
+    {
+      "entropy": 2.4783935546875,
+      "epoch": 0.0028924872808036984,
+      "grad_norm": 0.5674195289611816,
+      "learning_rate": 1.6700000000000003e-06,
+      "loss": 2.4497,
+      "mean_token_accuracy": 0.47510121995583177,
+      "num_tokens": 24445534.0,
+      "step": 168
+    },
+    {
+      "entropy": 2.4825439453125,
+      "epoch": 0.0029097044669989585,
+      "grad_norm": 0.6282191276550293,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.4444,
+      "mean_token_accuracy": 0.4739877316169441,
+      "num_tokens": 24568649.0,
+      "step": 169
+    },
+    {
+      "entropy": 2.4176025390625,
+      "epoch": 0.0029269216531942186,
+      "grad_norm": 0.6034136414527893,
+      "learning_rate": 1.6900000000000003e-06,
+      "loss": 2.3804,
+      "mean_token_accuracy": 0.48952830489724874,
+      "num_tokens": 24710505.0,
+      "step": 170
+    },
+    {
+      "entropy": 2.441650390625,
+      "epoch": 0.0029441388393894787,
+      "grad_norm": 0.579363226890564,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.3715,
+      "mean_token_accuracy": 0.4852260472252965,
+      "num_tokens": 24860551.0,
+      "step": 171
+    },
+    {
+      "entropy": 2.4251708984375,
+      "epoch": 0.002961356025584739,
+      "grad_norm": 0.6410456895828247,
+      "learning_rate": 1.7100000000000004e-06,
+      "loss": 2.38,
+      "mean_token_accuracy": 0.4843108947388828,
+      "num_tokens": 24997729.0,
+      "step": 172
+    },
+    {
+      "entropy": 2.430419921875,
+      "epoch": 0.002978573211779999,
+      "grad_norm": 0.5755802392959595,
+      "learning_rate": 1.72e-06,
+      "loss": 2.3886,
+      "mean_token_accuracy": 0.4809368369169533,
+      "num_tokens": 25140050.0,
+      "step": 173
+    },
+    {
+      "entropy": 2.485595703125,
+      "epoch": 0.002995790397975259,
+      "grad_norm": 0.606526792049408,
+      "learning_rate": 1.73e-06,
+      "loss": 2.4497,
+      "mean_token_accuracy": 0.48244015080854297,
+      "num_tokens": 25280791.0,
+      "step": 174
+    },
+    {
+      "entropy": 2.48046875,
+      "epoch": 0.003013007584170519,
+      "grad_norm": 0.6201883554458618,
+      "learning_rate": 1.74e-06,
+      "loss": 2.4686,
+      "mean_token_accuracy": 0.4720367449335754,
+      "num_tokens": 25429327.0,
+      "step": 175
+    },
+    {
+      "entropy": 2.40380859375,
+      "epoch": 0.003030224770365779,
+      "grad_norm": 0.6322990655899048,
+      "learning_rate": 1.75e-06,
+      "loss": 2.3775,
+      "mean_token_accuracy": 0.48563892720267177,
+      "num_tokens": 25556902.0,
+      "step": 176
+    },
+    {
+      "entropy": 2.4178466796875,
+      "epoch": 0.0030474419565610393,
+      "grad_norm": 0.5754444003105164,
+      "learning_rate": 1.76e-06,
+      "loss": 2.3881,
+      "mean_token_accuracy": 0.4864646405912936,
+      "num_tokens": 25706350.0,
+      "step": 177
+    },
+    {
+      "entropy": 2.4537353515625,
+      "epoch": 0.0030646591427562994,
+      "grad_norm": 0.5913891792297363,
+      "learning_rate": 1.77e-06,
+      "loss": 2.4168,
+      "mean_token_accuracy": 0.47908906172960997,
+      "num_tokens": 25845686.0,
+      "step": 178
+    },
+    {
+      "entropy": 2.442138671875,
+      "epoch": 0.0030818763289515595,
+      "grad_norm": 0.5810282826423645,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.3687,
+      "mean_token_accuracy": 0.48843948962166905,
+      "num_tokens": 25991169.0,
+      "step": 179
+    },
+    {
+      "entropy": 2.430908203125,
+      "epoch": 0.0030990935151468195,
+      "grad_norm": 0.59124356508255,
+      "learning_rate": 1.79e-06,
+      "loss": 2.345,
+      "mean_token_accuracy": 0.49094062810763717,
+      "num_tokens": 26137613.0,
+      "step": 180
+    },
+    {
+      "entropy": 2.43798828125,
+      "epoch": 0.0031163107013420796,
+      "grad_norm": 0.5589233636856079,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.3647,
+      "mean_token_accuracy": 0.47995236283168197,
+      "num_tokens": 26304122.0,
+      "step": 181
+    },
+    {
+      "entropy": 2.43017578125,
+      "epoch": 0.0031335278875373397,
+      "grad_norm": 0.6091126203536987,
+      "learning_rate": 1.81e-06,
+      "loss": 2.3233,
+      "mean_token_accuracy": 0.49118568608537316,
+      "num_tokens": 26439950.0,
+      "step": 182
+    },
+    {
+      "entropy": 2.39697265625,
+      "epoch": 0.0031507450737326,
+      "grad_norm": 0.6027998328208923,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.332,
+      "mean_token_accuracy": 0.4913685843348503,
+      "num_tokens": 26579726.0,
+      "step": 183
+    },
+    {
+      "entropy": 2.404296875,
+      "epoch": 0.00316796225992786,
+      "grad_norm": 0.5764933228492737,
+      "learning_rate": 1.83e-06,
+      "loss": 2.3598,
+      "mean_token_accuracy": 0.48573345225304365,
+      "num_tokens": 26724853.0,
+      "step": 184
+    },
+    {
+      "entropy": 2.4344482421875,
+      "epoch": 0.00318517944612312,
+      "grad_norm": 0.5961573123931885,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.3914,
+      "mean_token_accuracy": 0.48558472096920013,
+      "num_tokens": 26863674.0,
+      "step": 185
+    },
+    {
+      "entropy": 2.446044921875,
+      "epoch": 0.00320239663231838,
+      "grad_norm": 0.570022702217102,
+      "learning_rate": 1.85e-06,
+      "loss": 2.4299,
+      "mean_token_accuracy": 0.48616287065669894,
+      "num_tokens": 27016137.0,
+      "step": 186
+    },
+    {
+      "entropy": 2.4267578125,
+      "epoch": 0.00321961381851364,
+      "grad_norm": 0.5620612502098083,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.3965,
+      "mean_token_accuracy": 0.4951572152785957,
+      "num_tokens": 27168668.0,
+      "step": 187
+    },
+    {
+      "entropy": 2.5091552734375,
+      "epoch": 0.0032368310047089003,
+      "grad_norm": 0.5910755395889282,
+      "learning_rate": 1.87e-06,
+      "loss": 2.4964,
+      "mean_token_accuracy": 0.47218859009444714,
+      "num_tokens": 27303758.0,
+      "step": 188
+    },
+    {
+      "entropy": 2.41259765625,
+      "epoch": 0.0032540481909041604,
+      "grad_norm": 0.5817851424217224,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.3692,
+      "mean_token_accuracy": 0.48497994616627693,
+      "num_tokens": 27442544.0,
+      "step": 189
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.0032712653770994205,
+      "grad_norm": 0.565247654914856,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 2.4424,
+      "mean_token_accuracy": 0.4795730533078313,
+      "num_tokens": 27596457.0,
+      "step": 190
+    },
+    {
+      "entropy": 2.4776611328125,
+      "epoch": 0.0032884825632946806,
+      "grad_norm": 0.5709455609321594,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.4314,
+      "mean_token_accuracy": 0.47365658916532993,
+      "num_tokens": 27743357.0,
+      "step": 191
+    },
+    {
+      "entropy": 2.5281982421875,
+      "epoch": 0.0033056997494899406,
+      "grad_norm": 0.629690945148468,
+      "learning_rate": 1.9100000000000003e-06,
+      "loss": 2.4736,
+      "mean_token_accuracy": 0.47410575672984123,
+      "num_tokens": 27876803.0,
+      "step": 192
+    },
+    {
+      "entropy": 2.4635009765625,
+      "epoch": 0.003322916935685201,
+      "grad_norm": 0.637241780757904,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.4301,
+      "mean_token_accuracy": 0.48486198624596,
+      "num_tokens": 28028469.0,
+      "step": 193
+    },
+    {
+      "entropy": 2.486083984375,
+      "epoch": 0.0033401341218804613,
+      "grad_norm": 0.5691868662834167,
+      "learning_rate": 1.93e-06,
+      "loss": 2.437,
+      "mean_token_accuracy": 0.4730471963994205,
+      "num_tokens": 28169802.0,
+      "step": 194
+    },
+    {
+      "entropy": 2.43017578125,
+      "epoch": 0.0033573513080757214,
+      "grad_norm": 0.6118927597999573,
+      "learning_rate": 1.94e-06,
+      "loss": 2.3835,
+      "mean_token_accuracy": 0.4860619972459972,
+      "num_tokens": 28299767.0,
+      "step": 195
+    },
+    {
+      "entropy": 2.4813232421875,
+      "epoch": 0.0033745684942709814,
+      "grad_norm": 0.564987063407898,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 2.4519,
+      "mean_token_accuracy": 0.4765043603256345,
+      "num_tokens": 28442455.0,
+      "step": 196
+    },
+    {
+      "entropy": 2.4971923828125,
+      "epoch": 0.0033917856804662415,
+      "grad_norm": 0.5519808530807495,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.4937,
+      "mean_token_accuracy": 0.4625990390777588,
+      "num_tokens": 28585079.0,
+      "step": 197
+    },
+    {
+      "entropy": 2.464599609375,
+      "epoch": 0.0034090028666615016,
+      "grad_norm": 0.6350486874580383,
+      "learning_rate": 1.97e-06,
+      "loss": 2.4561,
+      "mean_token_accuracy": 0.4764754744246602,
+      "num_tokens": 28722871.0,
+      "step": 198
+    },
+    {
+      "entropy": 2.4639892578125,
+      "epoch": 0.0034262200528567617,
+      "grad_norm": 0.5707213878631592,
+      "learning_rate": 1.98e-06,
+      "loss": 2.4378,
+      "mean_token_accuracy": 0.48255998734384775,
+      "num_tokens": 28868682.0,
+      "step": 199
+    },
+    {
+      "entropy": 2.415771484375,
+      "epoch": 0.003443437239052022,
+      "grad_norm": 0.593158483505249,
+      "learning_rate": 1.9900000000000004e-06,
+      "loss": 2.3004,
+      "mean_token_accuracy": 0.49145969236269593,
+      "num_tokens": 29016713.0,
+      "step": 200
+    },
+    {
+      "entropy": 2.462890625,
+      "epoch": 0.003460654425247282,
+      "grad_norm": 0.6505548357963562,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.3531,
+      "mean_token_accuracy": 0.48328409856185317,
+      "num_tokens": 29143923.0,
+      "step": 201
+    },
+    {
+      "entropy": 2.3577880859375,
+      "epoch": 0.003477871611442542,
+      "grad_norm": 0.6153655052185059,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 2.3246,
+      "mean_token_accuracy": 0.4954985845834017,
+      "num_tokens": 29285239.0,
+      "step": 202
+    },
+    {
+      "entropy": 2.51025390625,
+      "epoch": 0.003495088797637802,
+      "grad_norm": 0.5901440978050232,
+      "learning_rate": 2.02e-06,
+      "loss": 2.4444,
+      "mean_token_accuracy": 0.47322959266602993,
+      "num_tokens": 29423789.0,
+      "step": 203
+    },
+    {
+      "entropy": 2.473388671875,
+      "epoch": 0.003512305983833062,
+      "grad_norm": 0.6093935370445251,
+      "learning_rate": 2.0300000000000005e-06,
+      "loss": 2.4143,
+      "mean_token_accuracy": 0.4813940548337996,
+      "num_tokens": 29571521.0,
+      "step": 204
+    },
+    {
+      "entropy": 2.423828125,
+      "epoch": 0.0035295231700283223,
+      "grad_norm": 0.5486546754837036,
+      "learning_rate": 2.04e-06,
+      "loss": 2.3655,
+      "mean_token_accuracy": 0.4821597971022129,
+      "num_tokens": 29726314.0,
+      "step": 205
+    },
+    {
+      "entropy": 2.4498291015625,
+      "epoch": 0.0035467403562235824,
+      "grad_norm": 0.57234126329422,
+      "learning_rate": 2.05e-06,
+      "loss": 2.425,
+      "mean_token_accuracy": 0.47626253589987755,
+      "num_tokens": 29880000.0,
+      "step": 206
+    },
+    {
+      "entropy": 2.416015625,
+      "epoch": 0.0035639575424188425,
+      "grad_norm": 0.561396062374115,
+      "learning_rate": 2.06e-06,
+      "loss": 2.3439,
+      "mean_token_accuracy": 0.4864202947355807,
+      "num_tokens": 30027628.0,
+      "step": 207
+    },
+    {
+      "entropy": 2.4002685546875,
+      "epoch": 0.0035811747286141025,
+      "grad_norm": 0.569572389125824,
+      "learning_rate": 2.07e-06,
+      "loss": 2.3673,
+      "mean_token_accuracy": 0.4875691821798682,
+      "num_tokens": 30169974.0,
+      "step": 208
+    },
+    {
+      "entropy": 2.455078125,
+      "epoch": 0.0035983919148093626,
+      "grad_norm": 0.5601251721382141,
+      "learning_rate": 2.08e-06,
+      "loss": 2.4015,
+      "mean_token_accuracy": 0.4847674574702978,
+      "num_tokens": 30317198.0,
+      "step": 209
+    },
+    {
+      "entropy": 2.53271484375,
+      "epoch": 0.0036156091010046227,
+      "grad_norm": 0.5616832375526428,
+      "learning_rate": 2.09e-06,
+      "loss": 2.4877,
+      "mean_token_accuracy": 0.47258848743513227,
+      "num_tokens": 30460614.0,
+      "step": 210
+    },
+    {
+      "entropy": 2.454345703125,
+      "epoch": 0.003632826287199883,
+      "grad_norm": 0.5962870717048645,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.4077,
+      "mean_token_accuracy": 0.47984306002035737,
+      "num_tokens": 30595270.0,
+      "step": 211
+    },
+    {
+      "entropy": 2.414306640625,
+      "epoch": 0.003650043473395143,
+      "grad_norm": 0.5458055734634399,
+      "learning_rate": 2.11e-06,
+      "loss": 2.349,
+      "mean_token_accuracy": 0.48367989249527454,
+      "num_tokens": 30754164.0,
+      "step": 212
+    },
+    {
+      "entropy": 2.3519287109375,
+      "epoch": 0.003667260659590403,
+      "grad_norm": 0.6218618154525757,
+      "learning_rate": 2.12e-06,
+      "loss": 2.2976,
+      "mean_token_accuracy": 0.49590098252519965,
+      "num_tokens": 30910624.0,
+      "step": 213
+    },
+    {
+      "entropy": 2.4219970703125,
+      "epoch": 0.003684477845785663,
+      "grad_norm": 0.6083913445472717,
+      "learning_rate": 2.13e-06,
+      "loss": 2.3687,
+      "mean_token_accuracy": 0.4871506607159972,
+      "num_tokens": 31045412.0,
+      "step": 214
+    },
+    {
+      "entropy": 2.3848876953125,
+      "epoch": 0.003701695031980923,
+      "grad_norm": 0.5669545531272888,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.3481,
+      "mean_token_accuracy": 0.4880279768258333,
+      "num_tokens": 31198584.0,
+      "step": 215
+    },
+    {
+      "entropy": 2.497314453125,
+      "epoch": 0.0037189122181761833,
+      "grad_norm": 0.5737752318382263,
+      "learning_rate": 2.15e-06,
+      "loss": 2.4598,
+      "mean_token_accuracy": 0.47354970779269934,
+      "num_tokens": 31341345.0,
+      "step": 216
+    },
+    {
+      "entropy": 2.4847412109375,
+      "epoch": 0.0037361294043714434,
+      "grad_norm": 0.5978350639343262,
+      "learning_rate": 2.16e-06,
+      "loss": 2.4907,
+      "mean_token_accuracy": 0.4729965156875551,
+      "num_tokens": 31503528.0,
+      "step": 217
+    },
+    {
+      "entropy": 2.4783935546875,
+      "epoch": 0.003753346590566704,
+      "grad_norm": 0.5699813961982727,
+      "learning_rate": 2.17e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.4826196124777198,
+      "num_tokens": 31641779.0,
+      "step": 218
+    },
+    {
+      "entropy": 2.436279296875,
+      "epoch": 0.003770563776761964,
+      "grad_norm": 0.5500178337097168,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.4012,
+      "mean_token_accuracy": 0.4823254165239632,
+      "num_tokens": 31811688.0,
+      "step": 219
+    },
+    {
+      "entropy": 2.431884765625,
+      "epoch": 0.003787780962957224,
+      "grad_norm": 0.5497094392776489,
+      "learning_rate": 2.19e-06,
+      "loss": 2.4188,
+      "mean_token_accuracy": 0.48219793336465955,
+      "num_tokens": 31964463.0,
+      "step": 220
+    },
+    {
+      "entropy": 2.4559326171875,
+      "epoch": 0.003804998149152484,
+      "grad_norm": 0.5953599810600281,
+      "learning_rate": 2.2e-06,
+      "loss": 2.4074,
+      "mean_token_accuracy": 0.4810604937374592,
+      "num_tokens": 32106174.0,
+      "step": 221
+    },
+    {
+      "entropy": 2.4549560546875,
+      "epoch": 0.0038222153353477443,
+      "grad_norm": 0.5802156329154968,
+      "learning_rate": 2.21e-06,
+      "loss": 2.401,
+      "mean_token_accuracy": 0.4832776212133467,
+      "num_tokens": 32240256.0,
+      "step": 222
+    },
+    {
+      "entropy": 2.5172119140625,
+      "epoch": 0.0038394325215430044,
+      "grad_norm": 0.7379570007324219,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 2.4759,
+      "mean_token_accuracy": 0.47404406825080514,
+      "num_tokens": 32384376.0,
+      "step": 223
+    },
+    {
+      "entropy": 2.4268798828125,
+      "epoch": 0.0038566497077382644,
+      "grad_norm": 0.535229504108429,
+      "learning_rate": 2.2300000000000002e-06,
+      "loss": 2.3825,
+      "mean_token_accuracy": 0.48500062711536884,
+      "num_tokens": 32547276.0,
+      "step": 224
+    },
+    {
+      "entropy": 2.4552001953125,
+      "epoch": 0.0038738668939335245,
+      "grad_norm": 0.5618354082107544,
+      "learning_rate": 2.24e-06,
+      "loss": 2.4164,
+      "mean_token_accuracy": 0.4798359959386289,
+      "num_tokens": 32695574.0,
+      "step": 225
+    },
+    {
+      "entropy": 2.491455078125,
+      "epoch": 0.0038910840801287846,
+      "grad_norm": 0.5642921924591064,
+      "learning_rate": 2.25e-06,
+      "loss": 2.4474,
+      "mean_token_accuracy": 0.4762350879609585,
+      "num_tokens": 32836847.0,
+      "step": 226
+    },
+    {
+      "entropy": 2.415283203125,
+      "epoch": 0.003908301266324045,
+      "grad_norm": 0.614703893661499,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.3591,
+      "mean_token_accuracy": 0.48611282790079713,
+      "num_tokens": 32970328.0,
+      "step": 227
+    },
+    {
+      "entropy": 2.4599609375,
+      "epoch": 0.003925518452519304,
+      "grad_norm": 0.5725888013839722,
+      "learning_rate": 2.2700000000000003e-06,
+      "loss": 2.4113,
+      "mean_token_accuracy": 0.4802963142283261,
+      "num_tokens": 33112724.0,
+      "step": 228
+    },
+    {
+      "entropy": 2.4117431640625,
+      "epoch": 0.003942735638714565,
+      "grad_norm": 0.6599273681640625,
+      "learning_rate": 2.28e-06,
+      "loss": 2.3471,
+      "mean_token_accuracy": 0.4909554719924927,
+      "num_tokens": 33250365.0,
+      "step": 229
+    },
+    {
+      "entropy": 2.417236328125,
+      "epoch": 0.0039599528249098246,
+      "grad_norm": 0.5717114806175232,
+      "learning_rate": 2.29e-06,
+      "loss": 2.3673,
+      "mean_token_accuracy": 0.4863877324387431,
+      "num_tokens": 33402602.0,
+      "step": 230
+    },
+    {
+      "entropy": 2.4541015625,
+      "epoch": 0.003977170011105085,
+      "grad_norm": 0.5801478028297424,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.4395,
+      "mean_token_accuracy": 0.47902765218168497,
+      "num_tokens": 33544113.0,
+      "step": 231
+    },
+    {
+      "entropy": 2.454345703125,
+      "epoch": 0.003994387197300346,
+      "grad_norm": 0.6313830614089966,
+      "learning_rate": 2.3100000000000003e-06,
+      "loss": 2.368,
+      "mean_token_accuracy": 0.48253711173310876,
+      "num_tokens": 33693199.0,
+      "step": 232
+    },
+    {
+      "entropy": 2.411376953125,
+      "epoch": 0.004011604383495605,
+      "grad_norm": 0.5773042440414429,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.3673,
+      "mean_token_accuracy": 0.4857380697503686,
+      "num_tokens": 33834160.0,
+      "step": 233
+    },
+    {
+      "entropy": 2.4261474609375,
+      "epoch": 0.004028821569690866,
+      "grad_norm": 0.6097912788391113,
+      "learning_rate": 2.33e-06,
+      "loss": 2.4526,
+      "mean_token_accuracy": 0.4798373435623944,
+      "num_tokens": 33966216.0,
+      "step": 234
+    },
+    {
+      "entropy": 2.47802734375,
+      "epoch": 0.0040460387558861255,
+      "grad_norm": 0.5512007474899292,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.3984,
+      "mean_token_accuracy": 0.47808168828487396,
+      "num_tokens": 34113686.0,
+      "step": 235
+    },
+    {
+      "entropy": 2.4539794921875,
+      "epoch": 0.004063255942081386,
+      "grad_norm": 0.5286883115768433,
+      "learning_rate": 2.35e-06,
+      "loss": 2.412,
+      "mean_token_accuracy": 0.48627515137195587,
+      "num_tokens": 34282616.0,
+      "step": 236
+    },
+    {
+      "entropy": 2.4539794921875,
+      "epoch": 0.004080473128276646,
+      "grad_norm": 0.564912736415863,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.3459,
+      "mean_token_accuracy": 0.48682177206501365,
+      "num_tokens": 34436498.0,
+      "step": 237
+    },
+    {
+      "entropy": 2.4637451171875,
+      "epoch": 0.004097690314471906,
+      "grad_norm": 0.6081207990646362,
+      "learning_rate": 2.37e-06,
+      "loss": 2.4327,
+      "mean_token_accuracy": 0.4862173437140882,
+      "num_tokens": 34579113.0,
+      "step": 238
+    },
+    {
+      "entropy": 2.3974609375,
+      "epoch": 0.004114907500667166,
+      "grad_norm": 0.5634726285934448,
+      "learning_rate": 2.38e-06,
+      "loss": 2.3362,
+      "mean_token_accuracy": 0.49311843886971474,
+      "num_tokens": 34723287.0,
+      "step": 239
+    },
+    {
+      "entropy": 2.3797607421875,
+      "epoch": 0.004132124686862426,
+      "grad_norm": 0.5402041673660278,
+      "learning_rate": 2.39e-06,
+      "loss": 2.3051,
+      "mean_token_accuracy": 0.4971098625101149,
+      "num_tokens": 34879330.0,
+      "step": 240
+    },
+    {
+      "entropy": 2.4957275390625,
+      "epoch": 0.004149341873057686,
+      "grad_norm": 0.6134316325187683,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.4642,
+      "mean_token_accuracy": 0.4705090904608369,
+      "num_tokens": 35007660.0,
+      "step": 241
+    },
+    {
+      "entropy": 2.5213623046875,
+      "epoch": 0.0041665590592529465,
+      "grad_norm": 0.5660735368728638,
+      "learning_rate": 2.4100000000000002e-06,
+      "loss": 2.503,
+      "mean_token_accuracy": 0.4669585754163563,
+      "num_tokens": 35155284.0,
+      "step": 242
+    },
+    {
+      "entropy": 2.4287109375,
+      "epoch": 0.004183776245448206,
+      "grad_norm": 0.5641495585441589,
+      "learning_rate": 2.42e-06,
+      "loss": 2.411,
+      "mean_token_accuracy": 0.4803205137141049,
+      "num_tokens": 35306435.0,
+      "step": 243
+    },
+    {
+      "entropy": 2.47705078125,
+      "epoch": 0.004200993431643467,
+      "grad_norm": 0.5801253914833069,
+      "learning_rate": 2.43e-06,
+      "loss": 2.4781,
+      "mean_token_accuracy": 0.47467044508084655,
+      "num_tokens": 35458606.0,
+      "step": 244
+    },
+    {
+      "entropy": 2.3427734375,
+      "epoch": 0.004218210617838726,
+      "grad_norm": 0.5665391683578491,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.2603,
+      "mean_token_accuracy": 0.5001557641662657,
+      "num_tokens": 35599151.0,
+      "step": 245
+    },
+    {
+      "entropy": 2.4097900390625,
+      "epoch": 0.004235427804033987,
+      "grad_norm": 0.5967923998832703,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 2.3661,
+      "mean_token_accuracy": 0.4944393504410982,
+      "num_tokens": 35753307.0,
+      "step": 246
+    },
+    {
+      "entropy": 2.3837890625,
+      "epoch": 0.0042526449902292466,
+      "grad_norm": 0.5936510562896729,
+      "learning_rate": 2.46e-06,
+      "loss": 2.3321,
+      "mean_token_accuracy": 0.49646031484007835,
+      "num_tokens": 35892908.0,
+      "step": 247
+    },
+    {
+      "entropy": 2.4852294921875,
+      "epoch": 0.004269862176424507,
+      "grad_norm": 0.5804598927497864,
+      "learning_rate": 2.47e-06,
+      "loss": 2.3769,
+      "mean_token_accuracy": 0.47953970776870847,
+      "num_tokens": 36042857.0,
+      "step": 248
+    },
+    {
+      "entropy": 2.4976806640625,
+      "epoch": 0.004287079362619767,
+      "grad_norm": 0.6185864806175232,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.4387,
+      "mean_token_accuracy": 0.4731791294179857,
+      "num_tokens": 36194862.0,
+      "step": 249
+    },
+    {
+      "entropy": 2.4356689453125,
+      "epoch": 0.004304296548815027,
+      "grad_norm": 0.6290936470031738,
+      "learning_rate": 2.4900000000000003e-06,
+      "loss": 2.3616,
+      "mean_token_accuracy": 0.4869615500792861,
+      "num_tokens": 36347599.0,
+      "step": 250
+    },
+    {
+      "entropy": 2.4132080078125,
+      "epoch": 0.004321513735010287,
+      "grad_norm": 0.6308562755584717,
+      "learning_rate": 2.5e-06,
+      "loss": 2.355,
+      "mean_token_accuracy": 0.48826425010338426,
+      "num_tokens": 36506643.0,
+      "step": 251
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.0043387309212055475,
+      "grad_norm": 0.5634401440620422,
+      "learning_rate": 2.51e-06,
+      "loss": 2.4154,
+      "mean_token_accuracy": 0.4824108784087002,
+      "num_tokens": 36652239.0,
+      "step": 252
+    },
+    {
+      "entropy": 2.360595703125,
+      "epoch": 0.004355948107400807,
+      "grad_norm": 0.574517548084259,
+      "learning_rate": 2.52e-06,
+      "loss": 2.3125,
+      "mean_token_accuracy": 0.49683515494689345,
+      "num_tokens": 36803342.0,
+      "step": 253
+    },
+    {
+      "entropy": 2.440185546875,
+      "epoch": 0.004373165293596068,
+      "grad_norm": 0.6065932512283325,
+      "learning_rate": 2.5300000000000003e-06,
+      "loss": 2.4206,
+      "mean_token_accuracy": 0.48550984309986234,
+      "num_tokens": 36942721.0,
+      "step": 254
+    },
+    {
+      "entropy": 2.485107421875,
+      "epoch": 0.004390382479791327,
+      "grad_norm": 0.8562254905700684,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.4961,
+      "mean_token_accuracy": 0.47687816014513373,
+      "num_tokens": 37068241.0,
+      "step": 255
+    },
+    {
+      "entropy": 2.3944091796875,
+      "epoch": 0.004407599665986588,
+      "grad_norm": 0.547681987285614,
+      "learning_rate": 2.55e-06,
+      "loss": 2.3469,
+      "mean_token_accuracy": 0.49269043607637286,
+      "num_tokens": 37221499.0,
+      "step": 256
+    },
+    {
+      "entropy": 2.4814453125,
+      "epoch": 0.004424816852181848,
+      "grad_norm": 0.5839248299598694,
+      "learning_rate": 2.56e-06,
+      "loss": 2.4229,
+      "mean_token_accuracy": 0.47095031198114157,
+      "num_tokens": 37373368.0,
+      "step": 257
+    },
+    {
+      "entropy": 2.354736328125,
+      "epoch": 0.004442034038377108,
+      "grad_norm": 0.587020993232727,
+      "learning_rate": 2.5700000000000004e-06,
+      "loss": 2.3414,
+      "mean_token_accuracy": 0.4924391624517739,
+      "num_tokens": 37518300.0,
+      "step": 258
+    },
+    {
+      "entropy": 2.37255859375,
+      "epoch": 0.0044592512245723685,
+      "grad_norm": 0.5921294093132019,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.3288,
+      "mean_token_accuracy": 0.4891428491100669,
+      "num_tokens": 37668963.0,
+      "step": 259
+    },
+    {
+      "entropy": 2.461181640625,
+      "epoch": 0.004476468410767628,
+      "grad_norm": 0.6068063378334045,
+      "learning_rate": 2.59e-06,
+      "loss": 2.4455,
+      "mean_token_accuracy": 0.47976093366742134,
+      "num_tokens": 37805995.0,
+      "step": 260
+    },
+    {
+      "entropy": 2.4947509765625,
+      "epoch": 0.004493685596962889,
+      "grad_norm": 0.6007203459739685,
+      "learning_rate": 2.6e-06,
+      "loss": 2.4731,
+      "mean_token_accuracy": 0.4695629784837365,
+      "num_tokens": 37953374.0,
+      "step": 261
+    },
+    {
+      "entropy": 2.4515380859375,
+      "epoch": 0.004510902783158148,
+      "grad_norm": 0.6087055206298828,
+      "learning_rate": 2.6100000000000004e-06,
+      "loss": 2.3859,
+      "mean_token_accuracy": 0.48078236635774374,
+      "num_tokens": 38090268.0,
+      "step": 262
+    },
+    {
+      "entropy": 2.405029296875,
+      "epoch": 0.004528119969353409,
+      "grad_norm": 0.6398684978485107,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.3682,
+      "mean_token_accuracy": 0.4855854455381632,
+      "num_tokens": 38230536.0,
+      "step": 263
+    },
+    {
+      "entropy": 2.484375,
+      "epoch": 0.0045453371555486686,
+      "grad_norm": 0.5696126818656921,
+      "learning_rate": 2.6300000000000002e-06,
+      "loss": 2.4688,
+      "mean_token_accuracy": 0.47722396487370133,
+      "num_tokens": 38372741.0,
+      "step": 264
+    },
+    {
+      "entropy": 2.3275146484375,
+      "epoch": 0.004562554341743929,
+      "grad_norm": 0.5986506342887878,
+      "learning_rate": 2.64e-06,
+      "loss": 2.2601,
+      "mean_token_accuracy": 0.5040428307838738,
+      "num_tokens": 38510771.0,
+      "step": 265
+    },
+    {
+      "entropy": 2.4368896484375,
+      "epoch": 0.004579771527939189,
+      "grad_norm": 0.5909532904624939,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 2.3905,
+      "mean_token_accuracy": 0.48478930070996284,
+      "num_tokens": 38656604.0,
+      "step": 266
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.004596988714134449,
+      "grad_norm": 0.604336142539978,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.47,
+      "mean_token_accuracy": 0.47080791695043445,
+      "num_tokens": 38790002.0,
+      "step": 267
+    },
+    {
+      "entropy": 2.44287109375,
+      "epoch": 0.004614205900329709,
+      "grad_norm": 1.2783533334732056,
+      "learning_rate": 2.6700000000000003e-06,
+      "loss": 2.353,
+      "mean_token_accuracy": 0.4904564335010946,
+      "num_tokens": 38926583.0,
+      "step": 268
+    },
+    {
+      "entropy": 2.474365234375,
+      "epoch": 0.0046314230865249694,
+      "grad_norm": 0.6569477915763855,
+      "learning_rate": 2.68e-06,
+      "loss": 2.443,
+      "mean_token_accuracy": 0.47721053985878825,
+      "num_tokens": 39070295.0,
+      "step": 269
+    },
+    {
+      "entropy": 2.4146728515625,
+      "epoch": 0.004648640272720229,
+      "grad_norm": 0.6075790524482727,
+      "learning_rate": 2.6900000000000005e-06,
+      "loss": 2.3758,
+      "mean_token_accuracy": 0.4894418097101152,
+      "num_tokens": 39224913.0,
+      "step": 270
+    },
+    {
+      "entropy": 2.42333984375,
+      "epoch": 0.00466585745891549,
+      "grad_norm": 0.6399396657943726,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.3292,
+      "mean_token_accuracy": 0.4931036913767457,
+      "num_tokens": 39357252.0,
+      "step": 271
+    },
+    {
+      "entropy": 2.407958984375,
+      "epoch": 0.004683074645110749,
+      "grad_norm": 0.5720593333244324,
+      "learning_rate": 2.7100000000000003e-06,
+      "loss": 2.3628,
+      "mean_token_accuracy": 0.4841322088614106,
+      "num_tokens": 39493624.0,
+      "step": 272
+    },
+    {
+      "entropy": 2.5147705078125,
+      "epoch": 0.00470029183130601,
+      "grad_norm": 0.6194799542427063,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.4367,
+      "mean_token_accuracy": 0.4774662428535521,
+      "num_tokens": 39629017.0,
+      "step": 273
+    },
+    {
+      "entropy": 2.461181640625,
+      "epoch": 0.0047175090175012695,
+      "grad_norm": 0.5498484373092651,
+      "learning_rate": 2.7300000000000005e-06,
+      "loss": 2.4471,
+      "mean_token_accuracy": 0.4761566431261599,
+      "num_tokens": 39770313.0,
+      "step": 274
+    },
+    {
+      "entropy": 2.458984375,
+      "epoch": 0.00473472620369653,
+      "grad_norm": 0.6001895070075989,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.4156,
+      "mean_token_accuracy": 0.47761010052636266,
+      "num_tokens": 39921189.0,
+      "step": 275
+    },
+    {
+      "entropy": 2.3966064453125,
+      "epoch": 0.00475194338989179,
+      "grad_norm": 0.555188000202179,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 2.3301,
+      "mean_token_accuracy": 0.4860731796361506,
+      "num_tokens": 40076244.0,
+      "step": 276
+    },
+    {
+      "entropy": 2.404541015625,
+      "epoch": 0.00476916057608705,
+      "grad_norm": 0.5978058576583862,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.3391,
+      "mean_token_accuracy": 0.49592722998932004,
+      "num_tokens": 40212416.0,
+      "step": 277
+    },
+    {
+      "entropy": 2.437744140625,
+      "epoch": 0.00478637776228231,
+      "grad_norm": 0.6279250383377075,
+      "learning_rate": 2.7700000000000006e-06,
+      "loss": 2.3922,
+      "mean_token_accuracy": 0.48303127055987716,
+      "num_tokens": 40349633.0,
+      "step": 278
+    },
+    {
+      "entropy": 2.50244140625,
+      "epoch": 0.00480359494847757,
+      "grad_norm": 0.5651630759239197,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.4524,
+      "mean_token_accuracy": 0.47481566993519664,
+      "num_tokens": 40493744.0,
+      "step": 279
+    },
+    {
+      "entropy": 2.450439453125,
+      "epoch": 0.00482081213467283,
+      "grad_norm": 0.5598726272583008,
+      "learning_rate": 2.7900000000000004e-06,
+      "loss": 2.4624,
+      "mean_token_accuracy": 0.476602204144001,
+      "num_tokens": 40643755.0,
+      "step": 280
+    },
+    {
+      "entropy": 2.46826171875,
+      "epoch": 0.0048380293208680905,
+      "grad_norm": 0.5888033509254456,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.4159,
+      "mean_token_accuracy": 0.4830831168219447,
+      "num_tokens": 40784807.0,
+      "step": 281
+    },
+    {
+      "entropy": 2.379638671875,
+      "epoch": 0.00485524650706335,
+      "grad_norm": 0.5815398693084717,
+      "learning_rate": 2.8100000000000006e-06,
+      "loss": 2.351,
+      "mean_token_accuracy": 0.49111180379986763,
+      "num_tokens": 40934466.0,
+      "step": 282
+    },
+    {
+      "entropy": 2.4080810546875,
+      "epoch": 0.004872463693258611,
+      "grad_norm": 0.5862876176834106,
+      "learning_rate": 2.82e-06,
+      "loss": 2.3539,
+      "mean_token_accuracy": 0.49174999026581645,
+      "num_tokens": 41069116.0,
+      "step": 283
+    },
+    {
+      "entropy": 2.40478515625,
+      "epoch": 0.004889680879453871,
+      "grad_norm": 0.5889655947685242,
+      "learning_rate": 2.83e-06,
+      "loss": 2.3569,
+      "mean_token_accuracy": 0.4852382456883788,
+      "num_tokens": 41201155.0,
+      "step": 284
+    },
+    {
+      "entropy": 2.4091796875,
+      "epoch": 0.004906898065649131,
+      "grad_norm": 0.6074087619781494,
+      "learning_rate": 2.84e-06,
+      "loss": 2.3607,
+      "mean_token_accuracy": 0.48959261691197753,
+      "num_tokens": 41332480.0,
+      "step": 285
+    },
+    {
+      "entropy": 2.451416015625,
+      "epoch": 0.0049241152518443914,
+      "grad_norm": 0.6012029051780701,
+      "learning_rate": 2.85e-06,
+      "loss": 2.4236,
+      "mean_token_accuracy": 0.486750605981797,
+      "num_tokens": 41480685.0,
+      "step": 286
+    },
+    {
+      "entropy": 2.412841796875,
+      "epoch": 0.004941332438039651,
+      "grad_norm": 0.6248612999916077,
+      "learning_rate": 2.86e-06,
+      "loss": 2.3435,
+      "mean_token_accuracy": 0.49190296651795506,
+      "num_tokens": 41623635.0,
+      "step": 287
+    },
+    {
+      "entropy": 2.4788818359375,
+      "epoch": 0.004958549624234912,
+      "grad_norm": 0.5725988745689392,
+      "learning_rate": 2.87e-06,
+      "loss": 2.4732,
+      "mean_token_accuracy": 0.47983552562072873,
+      "num_tokens": 41781312.0,
+      "step": 288
+    },
+    {
+      "entropy": 2.446533203125,
+      "epoch": 0.004975766810430171,
+      "grad_norm": 0.6067764163017273,
+      "learning_rate": 2.88e-06,
+      "loss": 2.3832,
+      "mean_token_accuracy": 0.48594531044363976,
+      "num_tokens": 41913646.0,
+      "step": 289
+    },
+    {
+      "entropy": 2.4150390625,
+      "epoch": 0.004992983996625432,
+      "grad_norm": 0.6115426421165466,
+      "learning_rate": 2.89e-06,
+      "loss": 2.3762,
+      "mean_token_accuracy": 0.4852985511533916,
+      "num_tokens": 42048939.0,
+      "step": 290
+    },
+    {
+      "entropy": 2.5048828125,
+      "epoch": 0.0050102011828206915,
+      "grad_norm": 0.5820481777191162,
+      "learning_rate": 2.9e-06,
+      "loss": 2.4655,
+      "mean_token_accuracy": 0.473651675041765,
+      "num_tokens": 42187764.0,
+      "step": 291
+    },
+    {
+      "entropy": 2.4854736328125,
+      "epoch": 0.005027418369015952,
+      "grad_norm": 0.5613899827003479,
+      "learning_rate": 2.91e-06,
+      "loss": 2.4199,
+      "mean_token_accuracy": 0.48045311588793993,
+      "num_tokens": 42335423.0,
+      "step": 292
+    },
+    {
+      "entropy": 2.472412109375,
+      "epoch": 0.005044635555211212,
+      "grad_norm": 0.5480543375015259,
+      "learning_rate": 2.92e-06,
+      "loss": 2.4305,
+      "mean_token_accuracy": 0.47681946912780404,
+      "num_tokens": 42489078.0,
+      "step": 293
+    },
+    {
+      "entropy": 2.3948974609375,
+      "epoch": 0.005061852741406472,
+      "grad_norm": 0.554536759853363,
+      "learning_rate": 2.93e-06,
+      "loss": 2.3142,
+      "mean_token_accuracy": 0.4870352731086314,
+      "num_tokens": 42638769.0,
+      "step": 294
+    },
+    {
+      "entropy": 2.4532470703125,
+      "epoch": 0.005079069927601732,
+      "grad_norm": 0.5379908084869385,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.4019,
+      "mean_token_accuracy": 0.4769473564811051,
+      "num_tokens": 42784366.0,
+      "step": 295
+    },
+    {
+      "entropy": 2.442626953125,
+      "epoch": 0.005096287113796992,
+      "grad_norm": 0.5843129754066467,
+      "learning_rate": 2.95e-06,
+      "loss": 2.4068,
+      "mean_token_accuracy": 0.4839549297466874,
+      "num_tokens": 42931723.0,
+      "step": 296
+    },
+    {
+      "entropy": 2.4779052734375,
+      "epoch": 0.005113504299992252,
+      "grad_norm": 0.5816124081611633,
+      "learning_rate": 2.96e-06,
+      "loss": 2.4102,
+      "mean_token_accuracy": 0.47800721740350127,
+      "num_tokens": 43068668.0,
+      "step": 297
+    },
+    {
+      "entropy": 2.5050048828125,
+      "epoch": 0.0051307214861875125,
+      "grad_norm": 0.5763994455337524,
+      "learning_rate": 2.97e-06,
+      "loss": 2.4308,
+      "mean_token_accuracy": 0.47180112451314926,
+      "num_tokens": 43212264.0,
+      "step": 298
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.005147938672382772,
+      "grad_norm": 0.5906177759170532,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.44,
+      "mean_token_accuracy": 0.47890339232981205,
+      "num_tokens": 43358872.0,
+      "step": 299
+    },
+    {
+      "entropy": 2.361328125,
+      "epoch": 0.005165155858578033,
+      "grad_norm": 0.6285274028778076,
+      "learning_rate": 2.99e-06,
+      "loss": 2.3159,
+      "mean_token_accuracy": 0.49288657307624817,
+      "num_tokens": 43505769.0,
+      "step": 300
+    },
+    {
+      "entropy": 2.418701171875,
+      "epoch": 0.005182373044773292,
+      "grad_norm": 0.5737318396568298,
+      "learning_rate": 3e-06,
+      "loss": 2.3905,
+      "mean_token_accuracy": 0.48132931999862194,
+      "num_tokens": 43655664.0,
+      "step": 301
+    },
+    {
+      "entropy": 2.40673828125,
+      "epoch": 0.005199590230968553,
+      "grad_norm": 0.6073116660118103,
+      "learning_rate": 3.01e-06,
+      "loss": 2.3336,
+      "mean_token_accuracy": 0.4943325803615153,
+      "num_tokens": 43792452.0,
+      "step": 302
+    },
+    {
+      "entropy": 2.3916015625,
+      "epoch": 0.0052168074171638126,
+      "grad_norm": 0.5638276934623718,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.3604,
+      "mean_token_accuracy": 0.4889626274816692,
+      "num_tokens": 43950674.0,
+      "step": 303
+    },
+    {
+      "entropy": 2.45849609375,
+      "epoch": 0.005234024603359073,
+      "grad_norm": 0.5540956854820251,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 2.4143,
+      "mean_token_accuracy": 0.47294150246307254,
+      "num_tokens": 44104574.0,
+      "step": 304
+    },
+    {
+      "entropy": 2.4903564453125,
+      "epoch": 0.005251241789554333,
+      "grad_norm": 0.6057919263839722,
+      "learning_rate": 3.04e-06,
+      "loss": 2.4804,
+      "mean_token_accuracy": 0.4774297019466758,
+      "num_tokens": 44268375.0,
+      "step": 305
+    },
+    {
+      "entropy": 2.42431640625,
+      "epoch": 0.005268458975749593,
+      "grad_norm": 0.5463052988052368,
+      "learning_rate": 3.05e-06,
+      "loss": 2.3637,
+      "mean_token_accuracy": 0.4869398158043623,
+      "num_tokens": 44421122.0,
+      "step": 306
+    },
+    {
+      "entropy": 2.4371337890625,
+      "epoch": 0.005285676161944853,
+      "grad_norm": 0.581701934337616,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.3915,
+      "mean_token_accuracy": 0.4823682149872184,
+      "num_tokens": 44580278.0,
+      "step": 307
+    },
+    {
+      "entropy": 2.412841796875,
+      "epoch": 0.0053028933481401135,
+      "grad_norm": 0.6069723963737488,
+      "learning_rate": 3.0700000000000003e-06,
+      "loss": 2.3847,
+      "mean_token_accuracy": 0.4826384102925658,
+      "num_tokens": 44722347.0,
+      "step": 308
+    },
+    {
+      "entropy": 2.4246826171875,
+      "epoch": 0.005320110534335374,
+      "grad_norm": 0.6269805431365967,
+      "learning_rate": 3.08e-06,
+      "loss": 2.3817,
+      "mean_token_accuracy": 0.4870366188697517,
+      "num_tokens": 44859344.0,
+      "step": 309
+    },
+    {
+      "entropy": 2.5079345703125,
+      "epoch": 0.005337327720530634,
+      "grad_norm": 0.5604663491249084,
+      "learning_rate": 3.09e-06,
+      "loss": 2.4903,
+      "mean_token_accuracy": 0.4686946659348905,
+      "num_tokens": 45007467.0,
+      "step": 310
+    },
+    {
+      "entropy": 2.53173828125,
+      "epoch": 0.005354544906725894,
+      "grad_norm": 0.6524129509925842,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.5078,
+      "mean_token_accuracy": 0.4713136567734182,
+      "num_tokens": 45151205.0,
+      "step": 311
+    },
+    {
+      "entropy": 2.4884033203125,
+      "epoch": 0.005371762092921154,
+      "grad_norm": 0.6845608353614807,
+      "learning_rate": 3.1100000000000003e-06,
+      "loss": 2.4559,
+      "mean_token_accuracy": 0.47201906703412533,
+      "num_tokens": 45283726.0,
+      "step": 312
+    },
+    {
+      "entropy": 2.4666748046875,
+      "epoch": 0.005388979279116414,
+      "grad_norm": 0.5663162469863892,
+      "learning_rate": 3.12e-06,
+      "loss": 2.4598,
+      "mean_token_accuracy": 0.4775085118599236,
+      "num_tokens": 45431733.0,
+      "step": 313
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.005406196465311674,
+      "grad_norm": 0.6073246002197266,
+      "learning_rate": 3.13e-06,
+      "loss": 2.3789,
+      "mean_token_accuracy": 0.48635248839855194,
+      "num_tokens": 45564227.0,
+      "step": 314
+    },
+    {
+      "entropy": 2.403564453125,
+      "epoch": 0.0054234136515069345,
+      "grad_norm": 0.6272099018096924,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.3457,
+      "mean_token_accuracy": 0.49023490585386753,
+      "num_tokens": 45694763.0,
+      "step": 315
+    },
+    {
+      "entropy": 2.45947265625,
+      "epoch": 0.005440630837702194,
+      "grad_norm": 0.6056774258613586,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 2.4134,
+      "mean_token_accuracy": 0.48205672251060605,
+      "num_tokens": 45822589.0,
+      "step": 316
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.005457848023897455,
+      "grad_norm": 0.5762467384338379,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.4066,
+      "mean_token_accuracy": 0.474984189029783,
+      "num_tokens": 45963018.0,
+      "step": 317
+    },
+    {
+      "entropy": 2.45849609375,
+      "epoch": 0.005475065210092714,
+      "grad_norm": 0.5697450637817383,
+      "learning_rate": 3.17e-06,
+      "loss": 2.4098,
+      "mean_token_accuracy": 0.481595104560256,
+      "num_tokens": 46116597.0,
+      "step": 318
+    },
+    {
+      "entropy": 2.455322265625,
+      "epoch": 0.005492282396287975,
+      "grad_norm": 0.5741132497787476,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.4058,
+      "mean_token_accuracy": 0.48317222855985165,
+      "num_tokens": 46269799.0,
+      "step": 319
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.0055094995824832346,
+      "grad_norm": 0.5584312677383423,
+      "learning_rate": 3.1900000000000004e-06,
+      "loss": 2.4214,
+      "mean_token_accuracy": 0.47542993212118745,
+      "num_tokens": 46413009.0,
+      "step": 320
+    },
+    {
+      "entropy": 2.345703125,
+      "epoch": 0.005526716768678495,
+      "grad_norm": 0.585615336894989,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.3015,
+      "mean_token_accuracy": 0.5034409500658512,
+      "num_tokens": 46553428.0,
+      "step": 321
+    },
+    {
+      "entropy": 2.476318359375,
+      "epoch": 0.005543933954873755,
+      "grad_norm": 0.5882828235626221,
+      "learning_rate": 3.21e-06,
+      "loss": 2.4505,
+      "mean_token_accuracy": 0.4761179364286363,
+      "num_tokens": 46694050.0,
+      "step": 322
+    },
+    {
+      "entropy": 2.5,
+      "epoch": 0.005561151141069015,
+      "grad_norm": 0.5832546949386597,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.4788,
+      "mean_token_accuracy": 0.4718233407475054,
+      "num_tokens": 46834894.0,
+      "step": 323
+    },
+    {
+      "entropy": 2.4697265625,
+      "epoch": 0.005578368327264275,
+      "grad_norm": 0.5746425986289978,
+      "learning_rate": 3.2300000000000004e-06,
+      "loss": 2.437,
+      "mean_token_accuracy": 0.473955764900893,
+      "num_tokens": 46971269.0,
+      "step": 324
+    },
+    {
+      "entropy": 2.493896484375,
+      "epoch": 0.0055955855134595354,
+      "grad_norm": 0.5796445608139038,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.485,
+      "mean_token_accuracy": 0.4741158653050661,
+      "num_tokens": 47120409.0,
+      "step": 325
+    },
+    {
+      "entropy": 2.455078125,
+      "epoch": 0.005612802699654795,
+      "grad_norm": 0.5498818159103394,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 2.4014,
+      "mean_token_accuracy": 0.48121104296296835,
+      "num_tokens": 47268071.0,
+      "step": 326
+    },
+    {
+      "entropy": 2.36669921875,
+      "epoch": 0.005630019885850056,
+      "grad_norm": 0.5759862065315247,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.3247,
+      "mean_token_accuracy": 0.4973863451741636,
+      "num_tokens": 47417393.0,
+      "step": 327
+    },
+    {
+      "entropy": 2.4267578125,
+      "epoch": 0.005647237072045315,
+      "grad_norm": 0.5729436874389648,
+      "learning_rate": 3.2700000000000005e-06,
+      "loss": 2.3774,
+      "mean_token_accuracy": 0.48343247501179576,
+      "num_tokens": 47557540.0,
+      "step": 328
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.005664454258240576,
+      "grad_norm": 0.5836823582649231,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.4552,
+      "mean_token_accuracy": 0.4818938923999667,
+      "num_tokens": 47698283.0,
+      "step": 329
+    },
+    {
+      "entropy": 2.47900390625,
+      "epoch": 0.0056816714444358355,
+      "grad_norm": 0.5725101828575134,
+      "learning_rate": 3.2900000000000003e-06,
+      "loss": 2.4444,
+      "mean_token_accuracy": 0.4746799021959305,
+      "num_tokens": 47841703.0,
+      "step": 330
+    },
+    {
+      "entropy": 2.480712890625,
+      "epoch": 0.005698888630631096,
+      "grad_norm": 0.573144257068634,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.427,
+      "mean_token_accuracy": 0.47849088767543435,
+      "num_tokens": 47992998.0,
+      "step": 331
+    },
+    {
+      "entropy": 2.496337890625,
+      "epoch": 0.005716105816826356,
+      "grad_norm": 0.5374743938446045,
+      "learning_rate": 3.3100000000000005e-06,
+      "loss": 2.4603,
+      "mean_token_accuracy": 0.4686804707162082,
+      "num_tokens": 48155349.0,
+      "step": 332
+    },
+    {
+      "entropy": 2.3896484375,
+      "epoch": 0.005733323003021616,
+      "grad_norm": 0.5805611610412598,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.3428,
+      "mean_token_accuracy": 0.49433091934770346,
+      "num_tokens": 48301544.0,
+      "step": 333
+    },
+    {
+      "entropy": 2.4912109375,
+      "epoch": 0.005750540189216877,
+      "grad_norm": 0.5827764868736267,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 2.439,
+      "mean_token_accuracy": 0.4705883339047432,
+      "num_tokens": 48443261.0,
+      "step": 334
+    },
+    {
+      "entropy": 2.476318359375,
+      "epoch": 0.005767757375412136,
+      "grad_norm": 0.5864250063896179,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.4157,
+      "mean_token_accuracy": 0.4782019346021116,
+      "num_tokens": 48581146.0,
+      "step": 335
+    },
+    {
+      "entropy": 2.538818359375,
+      "epoch": 0.005784974561607397,
+      "grad_norm": 0.55199134349823,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 2.4757,
+      "mean_token_accuracy": 0.46869856445118785,
+      "num_tokens": 48736815.0,
+      "step": 336
+    },
+    {
+      "entropy": 2.4351806640625,
+      "epoch": 0.0058021917478026565,
+      "grad_norm": 0.574338436126709,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.3759,
+      "mean_token_accuracy": 0.4860638175159693,
+      "num_tokens": 48897927.0,
+      "step": 337
+    },
+    {
+      "entropy": 2.4539794921875,
+      "epoch": 0.005819408933997917,
+      "grad_norm": 0.5789890289306641,
+      "learning_rate": 3.3700000000000003e-06,
+      "loss": 2.3891,
+      "mean_token_accuracy": 0.4830233994871378,
+      "num_tokens": 49041318.0,
+      "step": 338
+    },
+    {
+      "entropy": 2.3941650390625,
+      "epoch": 0.005836626120193177,
+      "grad_norm": 0.6387662887573242,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.3505,
+      "mean_token_accuracy": 0.49095042794942856,
+      "num_tokens": 49176213.0,
+      "step": 339
+    },
+    {
+      "entropy": 2.441162109375,
+      "epoch": 0.005853843306388437,
+      "grad_norm": 0.6007640957832336,
+      "learning_rate": 3.3900000000000006e-06,
+      "loss": 2.3682,
+      "mean_token_accuracy": 0.4886678601615131,
+      "num_tokens": 49305639.0,
+      "step": 340
+    },
+    {
+      "entropy": 2.51904296875,
+      "epoch": 0.005871060492583697,
+      "grad_norm": 0.5937355160713196,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.4649,
+      "mean_token_accuracy": 0.47522739693522453,
+      "num_tokens": 49445884.0,
+      "step": 341
+    },
+    {
+      "entropy": 2.5255126953125,
+      "epoch": 0.0058882776787789574,
+      "grad_norm": 0.6529424786567688,
+      "learning_rate": 3.4100000000000004e-06,
+      "loss": 2.4718,
+      "mean_token_accuracy": 0.4746531555429101,
+      "num_tokens": 49585678.0,
+      "step": 342
+    },
+    {
+      "entropy": 2.4620361328125,
+      "epoch": 0.005905494864974217,
+      "grad_norm": 0.5634227991104126,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.4171,
+      "mean_token_accuracy": 0.47586130583658814,
+      "num_tokens": 49733283.0,
+      "step": 343
+    },
+    {
+      "entropy": 2.4451904296875,
+      "epoch": 0.005922712051169478,
+      "grad_norm": 0.5977884531021118,
+      "learning_rate": 3.4300000000000006e-06,
+      "loss": 2.3813,
+      "mean_token_accuracy": 0.47802019072696567,
+      "num_tokens": 49872559.0,
+      "step": 344
+    },
+    {
+      "entropy": 2.4005126953125,
+      "epoch": 0.005939929237364737,
+      "grad_norm": 0.5835461020469666,
+      "learning_rate": 3.44e-06,
+      "loss": 2.351,
+      "mean_token_accuracy": 0.4916755324229598,
+      "num_tokens": 50014503.0,
+      "step": 345
+    },
+    {
+      "entropy": 2.414306640625,
+      "epoch": 0.005957146423559998,
+      "grad_norm": 0.616673469543457,
+      "learning_rate": 3.45e-06,
+      "loss": 2.3877,
+      "mean_token_accuracy": 0.4860090627335012,
+      "num_tokens": 50167107.0,
+      "step": 346
+    },
+    {
+      "entropy": 2.537353515625,
+      "epoch": 0.0059743636097552575,
+      "grad_norm": 0.6396327018737793,
+      "learning_rate": 3.46e-06,
+      "loss": 2.4815,
+      "mean_token_accuracy": 0.4667340232990682,
+      "num_tokens": 50299354.0,
+      "step": 347
+    },
+    {
+      "entropy": 2.424560546875,
+      "epoch": 0.005991580795950518,
+      "grad_norm": 0.5635019540786743,
+      "learning_rate": 3.4700000000000002e-06,
+      "loss": 2.3938,
+      "mean_token_accuracy": 0.48614488868042827,
+      "num_tokens": 50448841.0,
+      "step": 348
+    },
+    {
+      "entropy": 2.446044921875,
+      "epoch": 0.006008797982145778,
+      "grad_norm": 0.5684136152267456,
+      "learning_rate": 3.48e-06,
+      "loss": 2.3895,
+      "mean_token_accuracy": 0.48185077449306846,
+      "num_tokens": 50591226.0,
+      "step": 349
+    },
+    {
+      "entropy": 2.4525146484375,
+      "epoch": 0.006026015168341038,
+      "grad_norm": 0.614551842212677,
+      "learning_rate": 3.49e-06,
+      "loss": 2.4292,
+      "mean_token_accuracy": 0.47555977012962103,
+      "num_tokens": 50736325.0,
+      "step": 350
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.006043232354536298,
+      "grad_norm": 0.5730347633361816,
+      "learning_rate": 3.5e-06,
+      "loss": 2.402,
+      "mean_token_accuracy": 0.4838436101563275,
+      "num_tokens": 50882008.0,
+      "step": 351
+    },
+    {
+      "entropy": 2.40478515625,
+      "epoch": 0.006060449540731558,
+      "grad_norm": 0.5943769812583923,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 2.3651,
+      "mean_token_accuracy": 0.4828978287987411,
+      "num_tokens": 51022802.0,
+      "step": 352
+    },
+    {
+      "entropy": 2.4420166015625,
+      "epoch": 0.006077666726926818,
+      "grad_norm": 0.5597745180130005,
+      "learning_rate": 3.52e-06,
+      "loss": 2.3646,
+      "mean_token_accuracy": 0.4837341457605362,
+      "num_tokens": 51165806.0,
+      "step": 353
+    },
+    {
+      "entropy": 2.487060546875,
+      "epoch": 0.0060948839131220785,
+      "grad_norm": 0.5304551124572754,
+      "learning_rate": 3.53e-06,
+      "loss": 2.4595,
+      "mean_token_accuracy": 0.4816332710906863,
+      "num_tokens": 51325353.0,
+      "step": 354
+    },
+    {
+      "entropy": 2.4630126953125,
+      "epoch": 0.006112101099317338,
+      "grad_norm": 0.6318809390068054,
+      "learning_rate": 3.54e-06,
+      "loss": 2.4102,
+      "mean_token_accuracy": 0.4823563848622143,
+      "num_tokens": 51447642.0,
+      "step": 355
+    },
+    {
+      "entropy": 2.4342041015625,
+      "epoch": 0.006129318285512599,
+      "grad_norm": 0.571685791015625,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 2.3455,
+      "mean_token_accuracy": 0.48439886793494225,
+      "num_tokens": 51593840.0,
+      "step": 356
+    },
+    {
+      "entropy": 2.42138671875,
+      "epoch": 0.006146535471707858,
+      "grad_norm": 0.6026922464370728,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.3275,
+      "mean_token_accuracy": 0.4946357752196491,
+      "num_tokens": 51727430.0,
+      "step": 357
+    },
+    {
+      "entropy": 2.3724365234375,
+      "epoch": 0.006163752657903119,
+      "grad_norm": 0.584361732006073,
+      "learning_rate": 3.57e-06,
+      "loss": 2.3214,
+      "mean_token_accuracy": 0.48994710063561797,
+      "num_tokens": 51873095.0,
+      "step": 358
+    },
+    {
+      "entropy": 2.47265625,
+      "epoch": 0.0061809698440983794,
+      "grad_norm": 0.5443347692489624,
+      "learning_rate": 3.58e-06,
+      "loss": 2.4081,
+      "mean_token_accuracy": 0.4771101255901158,
+      "num_tokens": 52022705.0,
+      "step": 359
+    },
+    {
+      "entropy": 2.44677734375,
+      "epoch": 0.006198187030293639,
+      "grad_norm": 0.5630244016647339,
+      "learning_rate": 3.5900000000000004e-06,
+      "loss": 2.4228,
+      "mean_token_accuracy": 0.4784181611612439,
+      "num_tokens": 52168321.0,
+      "step": 360
+    },
+    {
+      "entropy": 2.467529296875,
+      "epoch": 0.0062154042164889,
+      "grad_norm": 0.5848613977432251,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.4734,
+      "mean_token_accuracy": 0.47760110441595316,
+      "num_tokens": 52321812.0,
+      "step": 361
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.006232621402684159,
+      "grad_norm": 0.5731722712516785,
+      "learning_rate": 3.61e-06,
+      "loss": 2.3849,
+      "mean_token_accuracy": 0.48051671124994755,
+      "num_tokens": 52477181.0,
+      "step": 362
+    },
+    {
+      "entropy": 2.435546875,
+      "epoch": 0.00624983858887942,
+      "grad_norm": 0.5230873227119446,
+      "learning_rate": 3.62e-06,
+      "loss": 2.361,
+      "mean_token_accuracy": 0.4862850420176983,
+      "num_tokens": 52638308.0,
+      "step": 363
+    },
+    {
+      "entropy": 2.3856201171875,
+      "epoch": 0.0062670557750746795,
+      "grad_norm": 0.5921328067779541,
+      "learning_rate": 3.6300000000000004e-06,
+      "loss": 2.3105,
+      "mean_token_accuracy": 0.496675749309361,
+      "num_tokens": 52778098.0,
+      "step": 364
+    },
+    {
+      "entropy": 2.486083984375,
+      "epoch": 0.00628427296126994,
+      "grad_norm": 0.5676338076591492,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.4375,
+      "mean_token_accuracy": 0.4777340483851731,
+      "num_tokens": 52926007.0,
+      "step": 365
+    },
+    {
+      "entropy": 2.5103759765625,
+      "epoch": 0.0063014901474652,
+      "grad_norm": 0.5867769718170166,
+      "learning_rate": 3.65e-06,
+      "loss": 2.486,
+      "mean_token_accuracy": 0.47535310545936227,
+      "num_tokens": 53067977.0,
+      "step": 366
+    },
+    {
+      "entropy": 2.443115234375,
+      "epoch": 0.00631870733366046,
+      "grad_norm": 0.5718576312065125,
+      "learning_rate": 3.66e-06,
+      "loss": 2.4122,
+      "mean_token_accuracy": 0.48204088117927313,
+      "num_tokens": 53214899.0,
+      "step": 367
+    },
+    {
+      "entropy": 2.4334716796875,
+      "epoch": 0.00633592451985572,
+      "grad_norm": 0.5422816872596741,
+      "learning_rate": 3.6700000000000004e-06,
+      "loss": 2.4062,
+      "mean_token_accuracy": 0.48176093958318233,
+      "num_tokens": 53376856.0,
+      "step": 368
+    },
+    {
+      "entropy": 2.44580078125,
+      "epoch": 0.00635314170605098,
+      "grad_norm": 0.5711188912391663,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.393,
+      "mean_token_accuracy": 0.48084082640707493,
+      "num_tokens": 53512787.0,
+      "step": 369
+    },
+    {
+      "entropy": 2.394287109375,
+      "epoch": 0.00637035889224624,
+      "grad_norm": 0.5846443772315979,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 2.3919,
+      "mean_token_accuracy": 0.49098140047863126,
+      "num_tokens": 53658651.0,
+      "step": 370
+    },
+    {
+      "entropy": 2.4273681640625,
+      "epoch": 0.0063875760784415005,
+      "grad_norm": 0.5670855641365051,
+      "learning_rate": 3.7e-06,
+      "loss": 2.3645,
+      "mean_token_accuracy": 0.48496190132573247,
+      "num_tokens": 53799071.0,
+      "step": 371
+    },
+    {
+      "entropy": 2.559814453125,
+      "epoch": 0.00640479326463676,
+      "grad_norm": 0.5919318199157715,
+      "learning_rate": 3.7100000000000005e-06,
+      "loss": 2.5396,
+      "mean_token_accuracy": 0.46628884179517627,
+      "num_tokens": 53932932.0,
+      "step": 372
+    },
+    {
+      "entropy": 2.4007568359375,
+      "epoch": 0.006422010450832021,
+      "grad_norm": 0.5672026872634888,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.3322,
+      "mean_token_accuracy": 0.48820391669869423,
+      "num_tokens": 54086169.0,
+      "step": 373
+    },
+    {
+      "entropy": 2.3978271484375,
+      "epoch": 0.00643922763702728,
+      "grad_norm": 0.5757989883422852,
+      "learning_rate": 3.7300000000000003e-06,
+      "loss": 2.3164,
+      "mean_token_accuracy": 0.4853466097265482,
+      "num_tokens": 54241964.0,
+      "step": 374
+    },
+    {
+      "entropy": 2.44580078125,
+      "epoch": 0.006456444823222541,
+      "grad_norm": 0.562664806842804,
+      "learning_rate": 3.74e-06,
+      "loss": 2.4101,
+      "mean_token_accuracy": 0.47567897848784924,
+      "num_tokens": 54386084.0,
+      "step": 375
+    },
+    {
+      "entropy": 2.4031982421875,
+      "epoch": 0.0064736620094178006,
+      "grad_norm": 0.5849329829216003,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.3482,
+      "mean_token_accuracy": 0.48964470298960805,
+      "num_tokens": 54523676.0,
+      "step": 376
+    },
+    {
+      "entropy": 2.4454345703125,
+      "epoch": 0.006490879195613061,
+      "grad_norm": 0.6109877824783325,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.4117,
+      "mean_token_accuracy": 0.48503589117899537,
+      "num_tokens": 54662310.0,
+      "step": 377
+    },
+    {
+      "entropy": 2.4466552734375,
+      "epoch": 0.006508096381808321,
+      "grad_norm": 0.5700381398200989,
+      "learning_rate": 3.7700000000000003e-06,
+      "loss": 2.4555,
+      "mean_token_accuracy": 0.477643181104213,
+      "num_tokens": 54809183.0,
+      "step": 378
+    },
+    {
+      "entropy": 2.4498291015625,
+      "epoch": 0.006525313568003581,
+      "grad_norm": 0.5381008386611938,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.4346,
+      "mean_token_accuracy": 0.47657692804932594,
+      "num_tokens": 54968533.0,
+      "step": 379
+    },
+    {
+      "entropy": 2.4263916015625,
+      "epoch": 0.006542530754198841,
+      "grad_norm": 0.5504759550094604,
+      "learning_rate": 3.79e-06,
+      "loss": 2.3587,
+      "mean_token_accuracy": 0.48776144767180085,
+      "num_tokens": 55117824.0,
+      "step": 380
+    },
+    {
+      "entropy": 2.4163818359375,
+      "epoch": 0.0065597479403941015,
+      "grad_norm": 0.5289463400840759,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.3674,
+      "mean_token_accuracy": 0.48521781573072076,
+      "num_tokens": 55269929.0,
+      "step": 381
+    },
+    {
+      "entropy": 2.4097900390625,
+      "epoch": 0.006576965126589361,
+      "grad_norm": 0.55353182554245,
+      "learning_rate": 3.8100000000000004e-06,
+      "loss": 2.3691,
+      "mean_token_accuracy": 0.48239677073433995,
+      "num_tokens": 55423948.0,
+      "step": 382
+    },
+    {
+      "entropy": 2.5047607421875,
+      "epoch": 0.006594182312784622,
+      "grad_norm": 0.5586856603622437,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.4676,
+      "mean_token_accuracy": 0.47292688954621553,
+      "num_tokens": 55567887.0,
+      "step": 383
+    },
+    {
+      "entropy": 2.47802734375,
+      "epoch": 0.006611399498979881,
+      "grad_norm": 0.5976067781448364,
+      "learning_rate": 3.830000000000001e-06,
+      "loss": 2.4577,
+      "mean_token_accuracy": 0.4735474893823266,
+      "num_tokens": 55700170.0,
+      "step": 384
+    },
+    {
+      "entropy": 2.4603271484375,
+      "epoch": 0.006628616685175142,
+      "grad_norm": 0.5445942878723145,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.4459,
+      "mean_token_accuracy": 0.4747904692776501,
+      "num_tokens": 55852413.0,
+      "step": 385
+    },
+    {
+      "entropy": 2.494873046875,
+      "epoch": 0.006645833871370402,
+      "grad_norm": 0.5769667625427246,
+      "learning_rate": 3.85e-06,
+      "loss": 2.4465,
+      "mean_token_accuracy": 0.4768332834355533,
+      "num_tokens": 56010312.0,
+      "step": 386
+    },
+    {
+      "entropy": 2.4306640625,
+      "epoch": 0.006663051057565662,
+      "grad_norm": 0.5875312685966492,
+      "learning_rate": 3.86e-06,
+      "loss": 2.3771,
+      "mean_token_accuracy": 0.4805051935836673,
+      "num_tokens": 56149079.0,
+      "step": 387
+    },
+    {
+      "entropy": 2.406494140625,
+      "epoch": 0.0066802682437609225,
+      "grad_norm": 0.5804882049560547,
+      "learning_rate": 3.87e-06,
+      "loss": 2.38,
+      "mean_token_accuracy": 0.48818252328783274,
+      "num_tokens": 56292210.0,
+      "step": 388
+    },
+    {
+      "entropy": 2.4931640625,
+      "epoch": 0.006697485429956182,
+      "grad_norm": 0.6065754890441895,
+      "learning_rate": 3.88e-06,
+      "loss": 2.5048,
+      "mean_token_accuracy": 0.47188919549807906,
+      "num_tokens": 56425512.0,
+      "step": 389
+    },
+    {
+      "entropy": 2.528076171875,
+      "epoch": 0.006714702616151443,
+      "grad_norm": 0.5695645213127136,
+      "learning_rate": 3.89e-06,
+      "loss": 2.4927,
+      "mean_token_accuracy": 0.47105303034186363,
+      "num_tokens": 56567538.0,
+      "step": 390
+    },
+    {
+      "entropy": 2.4716796875,
+      "epoch": 0.006731919802346702,
+      "grad_norm": 0.5461863279342651,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.4607,
+      "mean_token_accuracy": 0.47544049797579646,
+      "num_tokens": 56723718.0,
+      "step": 391
+    },
+    {
+      "entropy": 2.42236328125,
+      "epoch": 0.006749136988541963,
+      "grad_norm": 0.7229442000389099,
+      "learning_rate": 3.910000000000001e-06,
+      "loss": 2.322,
+      "mean_token_accuracy": 0.49011519411578774,
+      "num_tokens": 56881653.0,
+      "step": 392
+    },
+    {
+      "entropy": 2.4364013671875,
+      "epoch": 0.0067663541747372226,
+      "grad_norm": 0.5672553181648254,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.418,
+      "mean_token_accuracy": 0.4815367963165045,
+      "num_tokens": 57026712.0,
+      "step": 393
+    },
+    {
+      "entropy": 2.4490966796875,
+      "epoch": 0.006783571360932483,
+      "grad_norm": 0.6193339824676514,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 2.324,
+      "mean_token_accuracy": 0.4879821529611945,
+      "num_tokens": 57154864.0,
+      "step": 394
+    },
+    {
+      "entropy": 2.424072265625,
+      "epoch": 0.006800788547127743,
+      "grad_norm": 0.6047675609588623,
+      "learning_rate": 3.94e-06,
+      "loss": 2.343,
+      "mean_token_accuracy": 0.49066451471298933,
+      "num_tokens": 57313059.0,
+      "step": 395
+    },
+    {
+      "entropy": 2.529052734375,
+      "epoch": 0.006818005733323003,
+      "grad_norm": 0.6238451600074768,
+      "learning_rate": 3.95e-06,
+      "loss": 2.5103,
+      "mean_token_accuracy": 0.4740439150482416,
+      "num_tokens": 57456573.0,
+      "step": 396
+    },
+    {
+      "entropy": 2.4873046875,
+      "epoch": 0.006835222919518263,
+      "grad_norm": 0.6354604363441467,
+      "learning_rate": 3.96e-06,
+      "loss": 2.4319,
+      "mean_token_accuracy": 0.475927259773016,
+      "num_tokens": 57601572.0,
+      "step": 397
+    },
+    {
+      "entropy": 2.527587890625,
+      "epoch": 0.0068524401057135234,
+      "grad_norm": 0.6518051624298096,
+      "learning_rate": 3.97e-06,
+      "loss": 2.4717,
+      "mean_token_accuracy": 0.4759511463344097,
+      "num_tokens": 57737985.0,
+      "step": 398
+    },
+    {
+      "entropy": 2.400146484375,
+      "epoch": 0.006869657291908783,
+      "grad_norm": 0.5659388899803162,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.3422,
+      "mean_token_accuracy": 0.4919633981771767,
+      "num_tokens": 57886268.0,
+      "step": 399
+    },
+    {
+      "entropy": 2.4190673828125,
+      "epoch": 0.006886874478104044,
+      "grad_norm": 0.6125981211662292,
+      "learning_rate": 3.990000000000001e-06,
+      "loss": 2.3588,
+      "mean_token_accuracy": 0.49498999677598476,
+      "num_tokens": 58020557.0,
+      "step": 400
+    },
+    {
+      "entropy": 2.4425048828125,
+      "epoch": 0.006904091664299303,
+      "grad_norm": 0.5755429267883301,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.4063,
+      "mean_token_accuracy": 0.4781268546357751,
+      "num_tokens": 58157132.0,
+      "step": 401
+    },
+    {
+      "entropy": 2.4521484375,
+      "epoch": 0.006921308850494564,
+      "grad_norm": 0.5530086159706116,
+      "learning_rate": 4.0100000000000006e-06,
+      "loss": 2.3836,
+      "mean_token_accuracy": 0.4830552595667541,
+      "num_tokens": 58306412.0,
+      "step": 402
+    },
+    {
+      "entropy": 2.4962158203125,
+      "epoch": 0.0069385260366898235,
+      "grad_norm": 0.6048458218574524,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.4414,
+      "mean_token_accuracy": 0.4728825897909701,
+      "num_tokens": 58441138.0,
+      "step": 403
+    },
+    {
+      "entropy": 2.456298828125,
+      "epoch": 0.006955743222885084,
+      "grad_norm": 0.5381326079368591,
+      "learning_rate": 4.03e-06,
+      "loss": 2.4554,
+      "mean_token_accuracy": 0.474265918135643,
+      "num_tokens": 58601535.0,
+      "step": 404
+    },
+    {
+      "entropy": 2.4443359375,
+      "epoch": 0.006972960409080344,
+      "grad_norm": 0.5713069438934326,
+      "learning_rate": 4.04e-06,
+      "loss": 2.3849,
+      "mean_token_accuracy": 0.4821310769766569,
+      "num_tokens": 58732975.0,
+      "step": 405
+    },
+    {
+      "entropy": 2.421142578125,
+      "epoch": 0.006990177595275604,
+      "grad_norm": 0.6063797473907471,
+      "learning_rate": 4.05e-06,
+      "loss": 2.3477,
+      "mean_token_accuracy": 0.4892558893188834,
+      "num_tokens": 58863667.0,
+      "step": 406
+    },
+    {
+      "entropy": 2.4732666015625,
+      "epoch": 0.007007394781470864,
+      "grad_norm": 0.5941247940063477,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.4404,
+      "mean_token_accuracy": 0.4769346718676388,
+      "num_tokens": 59001984.0,
+      "step": 407
+    },
+    {
+      "entropy": 2.4593505859375,
+      "epoch": 0.007024611967666124,
+      "grad_norm": 0.5566405653953552,
+      "learning_rate": 4.07e-06,
+      "loss": 2.4064,
+      "mean_token_accuracy": 0.480907566845417,
+      "num_tokens": 59155387.0,
+      "step": 408
+    },
+    {
+      "entropy": 2.40966796875,
+      "epoch": 0.007041829153861384,
+      "grad_norm": 0.5935348868370056,
+      "learning_rate": 4.08e-06,
+      "loss": 2.3289,
+      "mean_token_accuracy": 0.49064510269090533,
+      "num_tokens": 59311521.0,
+      "step": 409
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 0.0070590463400566445,
+      "grad_norm": 0.581543505191803,
+      "learning_rate": 4.09e-06,
+      "loss": 2.4144,
+      "mean_token_accuracy": 0.4777227705344558,
+      "num_tokens": 59461508.0,
+      "step": 410
+    },
+    {
+      "entropy": 2.46240234375,
+      "epoch": 0.007076263526251905,
+      "grad_norm": 0.5600920915603638,
+      "learning_rate": 4.1e-06,
+      "loss": 2.4242,
+      "mean_token_accuracy": 0.47735275235027075,
+      "num_tokens": 59611309.0,
+      "step": 411
+    },
+    {
+      "entropy": 2.3861083984375,
+      "epoch": 0.007093480712447165,
+      "grad_norm": 0.645194411277771,
+      "learning_rate": 4.1100000000000005e-06,
+      "loss": 2.3315,
+      "mean_token_accuracy": 0.4974592626094818,
+      "num_tokens": 59739837.0,
+      "step": 412
+    },
+    {
+      "entropy": 2.42822265625,
+      "epoch": 0.007110697898642425,
+      "grad_norm": 0.5792311429977417,
+      "learning_rate": 4.12e-06,
+      "loss": 2.3767,
+      "mean_token_accuracy": 0.4854038432240486,
+      "num_tokens": 59874636.0,
+      "step": 413
+    },
+    {
+      "entropy": 2.527587890625,
+      "epoch": 0.007127915084837685,
+      "grad_norm": 0.5748518109321594,
+      "learning_rate": 4.13e-06,
+      "loss": 2.4767,
+      "mean_token_accuracy": 0.4697956978343427,
+      "num_tokens": 60006386.0,
+      "step": 414
+    },
+    {
+      "entropy": 2.4110107421875,
+      "epoch": 0.0071451322710329454,
+      "grad_norm": 0.5458788871765137,
+      "learning_rate": 4.14e-06,
+      "loss": 2.3364,
+      "mean_token_accuracy": 0.4888159199617803,
+      "num_tokens": 60161574.0,
+      "step": 415
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.007162349457228205,
+      "grad_norm": 0.5602844953536987,
+      "learning_rate": 4.15e-06,
+      "loss": 2.4069,
+      "mean_token_accuracy": 0.4802040630020201,
+      "num_tokens": 60315301.0,
+      "step": 416
+    },
+    {
+      "entropy": 2.420166015625,
+      "epoch": 0.007179566643423466,
+      "grad_norm": 0.5277621150016785,
+      "learning_rate": 4.16e-06,
+      "loss": 2.3738,
+      "mean_token_accuracy": 0.4825376900844276,
+      "num_tokens": 60475723.0,
+      "step": 417
+    },
+    {
+      "entropy": 2.371337890625,
+      "epoch": 0.007196783829618725,
+      "grad_norm": 0.6119918823242188,
+      "learning_rate": 4.17e-06,
+      "loss": 2.3183,
+      "mean_token_accuracy": 0.49627504125237465,
+      "num_tokens": 60624809.0,
+      "step": 418
+    },
+    {
+      "entropy": 2.43359375,
+      "epoch": 0.007214001015813986,
+      "grad_norm": 0.5752723217010498,
+      "learning_rate": 4.18e-06,
+      "loss": 2.4064,
+      "mean_token_accuracy": 0.48068576911464334,
+      "num_tokens": 60766487.0,
+      "step": 419
+    },
+    {
+      "entropy": 2.55322265625,
+      "epoch": 0.0072312182020092455,
+      "grad_norm": 0.591503918170929,
+      "learning_rate": 4.1900000000000005e-06,
+      "loss": 2.5353,
+      "mean_token_accuracy": 0.46583034889772534,
+      "num_tokens": 60899696.0,
+      "step": 420
+    },
+    {
+      "entropy": 2.482666015625,
+      "epoch": 0.007248435388204506,
+      "grad_norm": 0.584340512752533,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.4115,
+      "mean_token_accuracy": 0.47717864625155926,
+      "num_tokens": 61036187.0,
+      "step": 421
+    },
+    {
+      "entropy": 2.4241943359375,
+      "epoch": 0.007265652574399766,
+      "grad_norm": 0.5605193376541138,
+      "learning_rate": 4.21e-06,
+      "loss": 2.4071,
+      "mean_token_accuracy": 0.4835399743169546,
+      "num_tokens": 61193032.0,
+      "step": 422
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.007282869760595026,
+      "grad_norm": 0.5565392374992371,
+      "learning_rate": 4.22e-06,
+      "loss": 2.3779,
+      "mean_token_accuracy": 0.479055879637599,
+      "num_tokens": 61336014.0,
+      "step": 423
+    },
+    {
+      "entropy": 2.4150390625,
+      "epoch": 0.007300086946790286,
+      "grad_norm": 0.5690925717353821,
+      "learning_rate": 4.23e-06,
+      "loss": 2.3403,
+      "mean_token_accuracy": 0.4891307670623064,
+      "num_tokens": 61482900.0,
+      "step": 424
+    },
+    {
+      "entropy": 2.3863525390625,
+      "epoch": 0.007317304132985546,
+      "grad_norm": 0.5693207383155823,
+      "learning_rate": 4.24e-06,
+      "loss": 2.3509,
+      "mean_token_accuracy": 0.4905586871318519,
+      "num_tokens": 61629105.0,
+      "step": 425
+    },
+    {
+      "entropy": 2.547119140625,
+      "epoch": 0.007334521319180806,
+      "grad_norm": 0.5351988077163696,
+      "learning_rate": 4.25e-06,
+      "loss": 2.4922,
+      "mean_token_accuracy": 0.4689330547116697,
+      "num_tokens": 61780626.0,
+      "step": 426
+    },
+    {
+      "entropy": 2.47998046875,
+      "epoch": 0.0073517385053760665,
+      "grad_norm": 0.5875465273857117,
+      "learning_rate": 4.26e-06,
+      "loss": 2.4184,
+      "mean_token_accuracy": 0.4823840647004545,
+      "num_tokens": 61915045.0,
+      "step": 427
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.007368955691571326,
+      "grad_norm": 0.5979102253913879,
+      "learning_rate": 4.270000000000001e-06,
+      "loss": 2.4622,
+      "mean_token_accuracy": 0.4727623569779098,
+      "num_tokens": 62050386.0,
+      "step": 428
+    },
+    {
+      "entropy": 2.4766845703125,
+      "epoch": 0.007386172877766587,
+      "grad_norm": 0.5433111190795898,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.3951,
+      "mean_token_accuracy": 0.4800742859952152,
+      "num_tokens": 62212530.0,
+      "step": 429
+    },
+    {
+      "entropy": 2.4046630859375,
+      "epoch": 0.007403390063961846,
+      "grad_norm": 0.5350104570388794,
+      "learning_rate": 4.2900000000000004e-06,
+      "loss": 2.3617,
+      "mean_token_accuracy": 0.4864892913028598,
+      "num_tokens": 62370301.0,
+      "step": 430
+    },
+    {
+      "entropy": 2.4168701171875,
+      "epoch": 0.007420607250157107,
+      "grad_norm": 0.5780488848686218,
+      "learning_rate": 4.3e-06,
+      "loss": 2.3724,
+      "mean_token_accuracy": 0.48195584304630756,
+      "num_tokens": 62520205.0,
+      "step": 431
+    },
+    {
+      "entropy": 2.5218505859375,
+      "epoch": 0.0074378244363523666,
+      "grad_norm": 0.5742896795272827,
+      "learning_rate": 4.31e-06,
+      "loss": 2.4678,
+      "mean_token_accuracy": 0.47272730339318514,
+      "num_tokens": 62661467.0,
+      "step": 432
+    },
+    {
+      "entropy": 2.519775390625,
+      "epoch": 0.007455041622547627,
+      "grad_norm": 0.5965112447738647,
+      "learning_rate": 4.32e-06,
+      "loss": 2.5211,
+      "mean_token_accuracy": 0.46558596193790436,
+      "num_tokens": 62798987.0,
+      "step": 433
+    },
+    {
+      "entropy": 2.4462890625,
+      "epoch": 0.007472258808742887,
+      "grad_norm": 0.5964140295982361,
+      "learning_rate": 4.33e-06,
+      "loss": 2.3445,
+      "mean_token_accuracy": 0.4845765414647758,
+      "num_tokens": 62941022.0,
+      "step": 434
+    },
+    {
+      "entropy": 2.403076171875,
+      "epoch": 0.007489475994938147,
+      "grad_norm": 0.5478980541229248,
+      "learning_rate": 4.34e-06,
+      "loss": 2.3149,
+      "mean_token_accuracy": 0.4858265924267471,
+      "num_tokens": 63106699.0,
+      "step": 435
+    },
+    {
+      "entropy": 2.486328125,
+      "epoch": 0.007506693181133408,
+      "grad_norm": 0.5869223475456238,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 2.4946,
+      "mean_token_accuracy": 0.472647019661963,
+      "num_tokens": 63247500.0,
+      "step": 436
+    },
+    {
+      "entropy": 2.389404296875,
+      "epoch": 0.0075239103673286675,
+      "grad_norm": 0.583699107170105,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.38,
+      "mean_token_accuracy": 0.4875432150438428,
+      "num_tokens": 63397717.0,
+      "step": 437
+    },
+    {
+      "entropy": 2.4312744140625,
+      "epoch": 0.007541127553523928,
+      "grad_norm": 0.6965104341506958,
+      "learning_rate": 4.3700000000000005e-06,
+      "loss": 2.3725,
+      "mean_token_accuracy": 0.4779693940654397,
+      "num_tokens": 63532729.0,
+      "step": 438
+    },
+    {
+      "entropy": 2.510498046875,
+      "epoch": 0.007558344739719188,
+      "grad_norm": 0.5973443984985352,
+      "learning_rate": 4.38e-06,
+      "loss": 2.5009,
+      "mean_token_accuracy": 0.469982345122844,
+      "num_tokens": 63673621.0,
+      "step": 439
+    },
+    {
+      "entropy": 2.414794921875,
+      "epoch": 0.007575561925914448,
+      "grad_norm": 0.6061626672744751,
+      "learning_rate": 4.39e-06,
+      "loss": 2.3686,
+      "mean_token_accuracy": 0.4842392741702497,
+      "num_tokens": 63811912.0,
+      "step": 440
+    },
+    {
+      "entropy": 2.4686279296875,
+      "epoch": 0.007592779112109708,
+      "grad_norm": 0.5435473322868347,
+      "learning_rate": 4.4e-06,
+      "loss": 2.4623,
+      "mean_token_accuracy": 0.47148632165044546,
+      "num_tokens": 63969367.0,
+      "step": 441
+    },
+    {
+      "entropy": 2.380859375,
+      "epoch": 0.007609996298304968,
+      "grad_norm": 0.6280810832977295,
+      "learning_rate": 4.41e-06,
+      "loss": 2.3327,
+      "mean_token_accuracy": 0.4927673670463264,
+      "num_tokens": 64106475.0,
+      "step": 442
+    },
+    {
+      "entropy": 2.431884765625,
+      "epoch": 0.007627213484500228,
+      "grad_norm": 0.60051429271698,
+      "learning_rate": 4.42e-06,
+      "loss": 2.3611,
+      "mean_token_accuracy": 0.48384140338748693,
+      "num_tokens": 64240694.0,
+      "step": 443
+    },
+    {
+      "entropy": 2.482421875,
+      "epoch": 0.0076444306706954885,
+      "grad_norm": 0.5727265477180481,
+      "learning_rate": 4.430000000000001e-06,
+      "loss": 2.4215,
+      "mean_token_accuracy": 0.4743316164240241,
+      "num_tokens": 64379463.0,
+      "step": 444
+    },
+    {
+      "entropy": 2.4876708984375,
+      "epoch": 0.007661647856890748,
+      "grad_norm": 0.6397287845611572,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.4762,
+      "mean_token_accuracy": 0.47825985960662365,
+      "num_tokens": 64533020.0,
+      "step": 445
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.007678865043086009,
+      "grad_norm": 0.5948851704597473,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 2.4374,
+      "mean_token_accuracy": 0.4723509643226862,
+      "num_tokens": 64669487.0,
+      "step": 446
+    },
+    {
+      "entropy": 2.45556640625,
+      "epoch": 0.007696082229281268,
+      "grad_norm": 0.5741977691650391,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.4201,
+      "mean_token_accuracy": 0.4786538486368954,
+      "num_tokens": 64821796.0,
+      "step": 447
+    },
+    {
+      "entropy": 2.421875,
+      "epoch": 0.007713299415476529,
+      "grad_norm": 0.5873377323150635,
+      "learning_rate": 4.47e-06,
+      "loss": 2.3954,
+      "mean_token_accuracy": 0.48342282278463244,
+      "num_tokens": 64959422.0,
+      "step": 448
+    },
+    {
+      "entropy": 2.4324951171875,
+      "epoch": 0.0077305166016717886,
+      "grad_norm": 0.6045364737510681,
+      "learning_rate": 4.48e-06,
+      "loss": 2.378,
+      "mean_token_accuracy": 0.48433305928483605,
+      "num_tokens": 65103972.0,
+      "step": 449
+    },
+    {
+      "entropy": 2.5223388671875,
+      "epoch": 0.007747733787867049,
+      "grad_norm": 0.5803834795951843,
+      "learning_rate": 4.49e-06,
+      "loss": 2.4753,
+      "mean_token_accuracy": 0.469526968896389,
+      "num_tokens": 65240634.0,
+      "step": 450
+    },
+    {
+      "entropy": 2.517822265625,
+      "epoch": 0.007764950974062309,
+      "grad_norm": 0.5813508033752441,
+      "learning_rate": 4.5e-06,
+      "loss": 2.5203,
+      "mean_token_accuracy": 0.472479164134711,
+      "num_tokens": 65396335.0,
+      "step": 451
+    },
+    {
+      "entropy": 2.4752197265625,
+      "epoch": 0.007782168160257569,
+      "grad_norm": 0.8954483270645142,
+      "learning_rate": 4.510000000000001e-06,
+      "loss": 2.428,
+      "mean_token_accuracy": 0.4781260257586837,
+      "num_tokens": 65551212.0,
+      "step": 452
+    },
+    {
+      "entropy": 2.38916015625,
+      "epoch": 0.007799385346452829,
+      "grad_norm": 0.5381391644477844,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.3428,
+      "mean_token_accuracy": 0.4869826496578753,
+      "num_tokens": 65706488.0,
+      "step": 453
+    },
+    {
+      "entropy": 2.3837890625,
+      "epoch": 0.00781660253264809,
+      "grad_norm": 0.6232318878173828,
+      "learning_rate": 4.530000000000001e-06,
+      "loss": 2.3453,
+      "mean_token_accuracy": 0.4916538861580193,
+      "num_tokens": 65848771.0,
+      "step": 454
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.00783381971884335,
+      "grad_norm": 0.5655352473258972,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.3439,
+      "mean_token_accuracy": 0.48888020450249314,
+      "num_tokens": 66002377.0,
+      "step": 455
+    },
+    {
+      "entropy": 2.50732421875,
+      "epoch": 0.007851036905038609,
+      "grad_norm": 0.5934751033782959,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 2.452,
+      "mean_token_accuracy": 0.4774050717242062,
+      "num_tokens": 66140992.0,
+      "step": 456
+    },
+    {
+      "entropy": 2.486572265625,
+      "epoch": 0.00786825409123387,
+      "grad_norm": 0.5821974873542786,
+      "learning_rate": 4.56e-06,
+      "loss": 2.4434,
+      "mean_token_accuracy": 0.47727885795757174,
+      "num_tokens": 66283307.0,
+      "step": 457
+    },
+    {
+      "entropy": 2.4403076171875,
+      "epoch": 0.00788547127742913,
+      "grad_norm": 0.6021990180015564,
+      "learning_rate": 4.57e-06,
+      "loss": 2.3891,
+      "mean_token_accuracy": 0.48405969655141234,
+      "num_tokens": 66430065.0,
+      "step": 458
+    },
+    {
+      "entropy": 2.4219970703125,
+      "epoch": 0.00790268846362439,
+      "grad_norm": 0.5609884858131409,
+      "learning_rate": 4.58e-06,
+      "loss": 2.3585,
+      "mean_token_accuracy": 0.4830049378797412,
+      "num_tokens": 66580625.0,
+      "step": 459
+    },
+    {
+      "entropy": 2.4365234375,
+      "epoch": 0.007919905649819649,
+      "grad_norm": 0.6351988911628723,
+      "learning_rate": 4.590000000000001e-06,
+      "loss": 2.3914,
+      "mean_token_accuracy": 0.4825670407153666,
+      "num_tokens": 66710819.0,
+      "step": 460
+    },
+    {
+      "entropy": 2.388916015625,
+      "epoch": 0.00793712283601491,
+      "grad_norm": 0.5797377824783325,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.3401,
+      "mean_token_accuracy": 0.49362843204289675,
+      "num_tokens": 66855860.0,
+      "step": 461
+    },
+    {
+      "entropy": 2.502685546875,
+      "epoch": 0.00795434002221017,
+      "grad_norm": 0.5734979510307312,
+      "learning_rate": 4.610000000000001e-06,
+      "loss": 2.475,
+      "mean_token_accuracy": 0.468968971632421,
+      "num_tokens": 66991905.0,
+      "step": 462
+    },
+    {
+      "entropy": 2.4248046875,
+      "epoch": 0.00797155720840543,
+      "grad_norm": 0.5948538184165955,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.3821,
+      "mean_token_accuracy": 0.49015933787450194,
+      "num_tokens": 67140600.0,
+      "step": 463
+    },
+    {
+      "entropy": 2.4041748046875,
+      "epoch": 0.007988774394600691,
+      "grad_norm": 0.5927978754043579,
+      "learning_rate": 4.6300000000000006e-06,
+      "loss": 2.3563,
+      "mean_token_accuracy": 0.4928512079641223,
+      "num_tokens": 67296282.0,
+      "step": 464
+    },
+    {
+      "entropy": 2.44189453125,
+      "epoch": 0.00800599158079595,
+      "grad_norm": 0.5638139247894287,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.4232,
+      "mean_token_accuracy": 0.47663756739348173,
+      "num_tokens": 67445364.0,
+      "step": 465
+    },
+    {
+      "entropy": 2.440673828125,
+      "epoch": 0.00802320876699121,
+      "grad_norm": 0.6092103123664856,
+      "learning_rate": 4.65e-06,
+      "loss": 2.3328,
+      "mean_token_accuracy": 0.487199897877872,
+      "num_tokens": 67574552.0,
+      "step": 466
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.008040425953186471,
+      "grad_norm": 0.5823683738708496,
+      "learning_rate": 4.66e-06,
+      "loss": 2.3507,
+      "mean_token_accuracy": 0.4836606332100928,
+      "num_tokens": 67715593.0,
+      "step": 467
+    },
+    {
+      "entropy": 2.4288330078125,
+      "epoch": 0.008057643139381732,
+      "grad_norm": 0.571380615234375,
+      "learning_rate": 4.670000000000001e-06,
+      "loss": 2.376,
+      "mean_token_accuracy": 0.4848336656577885,
+      "num_tokens": 67851959.0,
+      "step": 468
+    },
+    {
+      "entropy": 2.4405517578125,
+      "epoch": 0.00807486032557699,
+      "grad_norm": 0.5497409105300903,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.3331,
+      "mean_token_accuracy": 0.49043108662590384,
+      "num_tokens": 67999315.0,
+      "step": 469
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.008092077511772251,
+      "grad_norm": 0.6131438612937927,
+      "learning_rate": 4.69e-06,
+      "loss": 2.4471,
+      "mean_token_accuracy": 0.48354413686320186,
+      "num_tokens": 68151679.0,
+      "step": 470
+    },
+    {
+      "entropy": 2.45361328125,
+      "epoch": 0.008109294697967511,
+      "grad_norm": 0.5572087168693542,
+      "learning_rate": 4.7e-06,
+      "loss": 2.408,
+      "mean_token_accuracy": 0.4743777387775481,
+      "num_tokens": 68303962.0,
+      "step": 471
+    },
+    {
+      "entropy": 2.4327392578125,
+      "epoch": 0.008126511884162772,
+      "grad_norm": 0.6229270696640015,
+      "learning_rate": 4.71e-06,
+      "loss": 2.4372,
+      "mean_token_accuracy": 0.4852800811640918,
+      "num_tokens": 68435424.0,
+      "step": 472
+    },
+    {
+      "entropy": 2.44580078125,
+      "epoch": 0.00814372907035803,
+      "grad_norm": 0.5797957181930542,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.4088,
+      "mean_token_accuracy": 0.48404247080907226,
+      "num_tokens": 68579820.0,
+      "step": 473
+    },
+    {
+      "entropy": 2.49560546875,
+      "epoch": 0.008160946256553291,
+      "grad_norm": 0.5527267456054688,
+      "learning_rate": 4.7300000000000005e-06,
+      "loss": 2.3781,
+      "mean_token_accuracy": 0.47309215646237135,
+      "num_tokens": 68720632.0,
+      "step": 474
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.008178163442748552,
+      "grad_norm": 0.6314601302146912,
+      "learning_rate": 4.74e-06,
+      "loss": 2.4292,
+      "mean_token_accuracy": 0.47931129625067115,
+      "num_tokens": 68847385.0,
+      "step": 475
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.008195380628943812,
+      "grad_norm": 0.5509300827980042,
+      "learning_rate": 4.75e-06,
+      "loss": 2.4099,
+      "mean_token_accuracy": 0.4842258528806269,
+      "num_tokens": 68997389.0,
+      "step": 476
+    },
+    {
+      "entropy": 2.41748046875,
+      "epoch": 0.008212597815139071,
+      "grad_norm": 0.5300650000572205,
+      "learning_rate": 4.76e-06,
+      "loss": 2.3382,
+      "mean_token_accuracy": 0.48574389703571796,
+      "num_tokens": 69155404.0,
+      "step": 477
+    },
+    {
+      "entropy": 2.4625244140625,
+      "epoch": 0.008229815001334332,
+      "grad_norm": 0.6269882321357727,
+      "learning_rate": 4.77e-06,
+      "loss": 2.3776,
+      "mean_token_accuracy": 0.4838248719461262,
+      "num_tokens": 69290738.0,
+      "step": 478
+    },
+    {
+      "entropy": 2.437255859375,
+      "epoch": 0.008247032187529592,
+      "grad_norm": 0.6751134395599365,
+      "learning_rate": 4.78e-06,
+      "loss": 2.4007,
+      "mean_token_accuracy": 0.48909331811591983,
+      "num_tokens": 69408062.0,
+      "step": 479
+    },
+    {
+      "entropy": 2.4910888671875,
+      "epoch": 0.008264249373724853,
+      "grad_norm": 0.596396267414093,
+      "learning_rate": 4.79e-06,
+      "loss": 2.4679,
+      "mean_token_accuracy": 0.47261510556563735,
+      "num_tokens": 69539526.0,
+      "step": 480
+    },
+    {
+      "entropy": 2.4178466796875,
+      "epoch": 0.008281466559920111,
+      "grad_norm": 0.6338604688644409,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.3809,
+      "mean_token_accuracy": 0.4872106113471091,
+      "num_tokens": 69691415.0,
+      "step": 481
+    },
+    {
+      "entropy": 2.419189453125,
+      "epoch": 0.008298683746115372,
+      "grad_norm": 0.5384358167648315,
+      "learning_rate": 4.8100000000000005e-06,
+      "loss": 2.3549,
+      "mean_token_accuracy": 0.49019210133701563,
+      "num_tokens": 69848010.0,
+      "step": 482
+    },
+    {
+      "entropy": 2.4853515625,
+      "epoch": 0.008315900932310633,
+      "grad_norm": 0.5779227614402771,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.4459,
+      "mean_token_accuracy": 0.4744481286033988,
+      "num_tokens": 69997662.0,
+      "step": 483
+    },
+    {
+      "entropy": 2.41064453125,
+      "epoch": 0.008333118118505893,
+      "grad_norm": 0.5691106915473938,
+      "learning_rate": 4.83e-06,
+      "loss": 2.3619,
+      "mean_token_accuracy": 0.4926013760268688,
+      "num_tokens": 70147997.0,
+      "step": 484
+    },
+    {
+      "entropy": 2.382080078125,
+      "epoch": 0.008350335304701152,
+      "grad_norm": 0.5725884437561035,
+      "learning_rate": 4.84e-06,
+      "loss": 2.3575,
+      "mean_token_accuracy": 0.489425728097558,
+      "num_tokens": 70294184.0,
+      "step": 485
+    },
+    {
+      "entropy": 2.47314453125,
+      "epoch": 0.008367552490896412,
+      "grad_norm": 0.5658232569694519,
+      "learning_rate": 4.85e-06,
+      "loss": 2.426,
+      "mean_token_accuracy": 0.4749414478428662,
+      "num_tokens": 70442248.0,
+      "step": 486
+    },
+    {
+      "entropy": 2.483642578125,
+      "epoch": 0.008384769677091673,
+      "grad_norm": 0.6129222512245178,
+      "learning_rate": 4.86e-06,
+      "loss": 2.4917,
+      "mean_token_accuracy": 0.4728327440097928,
+      "num_tokens": 70579740.0,
+      "step": 487
+    },
+    {
+      "entropy": 2.4320068359375,
+      "epoch": 0.008401986863286933,
+      "grad_norm": 0.9568486213684082,
+      "learning_rate": 4.87e-06,
+      "loss": 2.3935,
+      "mean_token_accuracy": 0.4829396088607609,
+      "num_tokens": 70721524.0,
+      "step": 488
+    },
+    {
+      "entropy": 2.459228515625,
+      "epoch": 0.008419204049482194,
+      "grad_norm": 0.544103741645813,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.4176,
+      "mean_token_accuracy": 0.476782136131078,
+      "num_tokens": 70877087.0,
+      "step": 489
+    },
+    {
+      "entropy": 2.405517578125,
+      "epoch": 0.008436421235677453,
+      "grad_norm": 0.5833263397216797,
+      "learning_rate": 4.890000000000001e-06,
+      "loss": 2.3254,
+      "mean_token_accuracy": 0.4887930774129927,
+      "num_tokens": 71009717.0,
+      "step": 490
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.008453638421872713,
+      "grad_norm": 0.5471597909927368,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.4063,
+      "mean_token_accuracy": 0.48418434290215373,
+      "num_tokens": 71166661.0,
+      "step": 491
+    },
+    {
+      "entropy": 2.4708251953125,
+      "epoch": 0.008470855608067974,
+      "grad_norm": 0.5708027482032776,
+      "learning_rate": 4.9100000000000004e-06,
+      "loss": 2.3853,
+      "mean_token_accuracy": 0.4815034563653171,
+      "num_tokens": 71313337.0,
+      "step": 492
+    },
+    {
+      "entropy": 2.4544677734375,
+      "epoch": 0.008488072794263234,
+      "grad_norm": 0.5583227872848511,
+      "learning_rate": 4.92e-06,
+      "loss": 2.4402,
+      "mean_token_accuracy": 0.47839572792872787,
+      "num_tokens": 71472447.0,
+      "step": 493
+    },
+    {
+      "entropy": 2.5404052734375,
+      "epoch": 0.008505289980458493,
+      "grad_norm": 0.578138530254364,
+      "learning_rate": 4.93e-06,
+      "loss": 2.4715,
+      "mean_token_accuracy": 0.47347771003842354,
+      "num_tokens": 71615228.0,
+      "step": 494
+    },
+    {
+      "entropy": 2.3878173828125,
+      "epoch": 0.008522507166653754,
+      "grad_norm": 0.5488659143447876,
+      "learning_rate": 4.94e-06,
+      "loss": 2.3633,
+      "mean_token_accuracy": 0.49038203712552786,
+      "num_tokens": 71773695.0,
+      "step": 495
+    },
+    {
+      "entropy": 2.4010009765625,
+      "epoch": 0.008539724352849014,
+      "grad_norm": 0.61408531665802,
+      "learning_rate": 4.95e-06,
+      "loss": 2.3474,
+      "mean_token_accuracy": 0.48974387207999825,
+      "num_tokens": 71924349.0,
+      "step": 496
+    },
+    {
+      "entropy": 2.4000244140625,
+      "epoch": 0.008556941539044275,
+      "grad_norm": 0.5874980688095093,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.3611,
+      "mean_token_accuracy": 0.49134851479902864,
+      "num_tokens": 72064362.0,
+      "step": 497
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.008574158725239533,
+      "grad_norm": 0.6588267087936401,
+      "learning_rate": 4.970000000000001e-06,
+      "loss": 2.3615,
+      "mean_token_accuracy": 0.48920353641733527,
+      "num_tokens": 72195808.0,
+      "step": 498
+    },
+    {
+      "entropy": 2.4034423828125,
+      "epoch": 0.008591375911434794,
+      "grad_norm": 0.5799132585525513,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.3648,
+      "mean_token_accuracy": 0.4861590703949332,
+      "num_tokens": 72329610.0,
+      "step": 499
+    },
+    {
+      "entropy": 2.4595947265625,
+      "epoch": 0.008608593097630055,
+      "grad_norm": 0.5722876787185669,
+      "learning_rate": 4.9900000000000005e-06,
+      "loss": 2.4214,
+      "mean_token_accuracy": 0.47663868917152286,
+      "num_tokens": 72472581.0,
+      "step": 500
+    },
+    {
+      "entropy": 2.491943359375,
+      "epoch": 0.008625810283825315,
+      "grad_norm": 0.5580865740776062,
+      "learning_rate": 5e-06,
+      "loss": 2.4629,
+      "mean_token_accuracy": 0.4767118990421295,
+      "num_tokens": 72616609.0,
+      "step": 501
+    },
+    {
+      "entropy": 2.3966064453125,
+      "epoch": 0.008643027470020574,
+      "grad_norm": 0.5624485015869141,
+      "learning_rate": 5.01e-06,
+      "loss": 2.3162,
+      "mean_token_accuracy": 0.4926113812252879,
+      "num_tokens": 72772528.0,
+      "step": 502
+    },
+    {
+      "entropy": 2.401611328125,
+      "epoch": 0.008660244656215834,
+      "grad_norm": 0.5372104644775391,
+      "learning_rate": 5.02e-06,
+      "loss": 2.3705,
+      "mean_token_accuracy": 0.4896852090023458,
+      "num_tokens": 72936711.0,
+      "step": 503
+    },
+    {
+      "entropy": 2.5235595703125,
+      "epoch": 0.008677461842411095,
+      "grad_norm": 0.6084150075912476,
+      "learning_rate": 5.03e-06,
+      "loss": 2.4556,
+      "mean_token_accuracy": 0.47214356577023864,
+      "num_tokens": 73065595.0,
+      "step": 504
+    },
+    {
+      "entropy": 2.3858642578125,
+      "epoch": 0.008694679028606355,
+      "grad_norm": 0.555059015750885,
+      "learning_rate": 5.04e-06,
+      "loss": 2.3394,
+      "mean_token_accuracy": 0.4875369444489479,
+      "num_tokens": 73214645.0,
+      "step": 505
+    },
+    {
+      "entropy": 2.473876953125,
+      "epoch": 0.008711896214801614,
+      "grad_norm": 0.5269579887390137,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 2.3993,
+      "mean_token_accuracy": 0.4812088548205793,
+      "num_tokens": 73376592.0,
+      "step": 506
+    },
+    {
+      "entropy": 2.408935546875,
+      "epoch": 0.008729113400996875,
+      "grad_norm": 0.5673913955688477,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.3398,
+      "mean_token_accuracy": 0.4860687367618084,
+      "num_tokens": 73524487.0,
+      "step": 507
+    },
+    {
+      "entropy": 2.4530029296875,
+      "epoch": 0.008746330587192135,
+      "grad_norm": 0.6308383345603943,
+      "learning_rate": 5.070000000000001e-06,
+      "loss": 2.4261,
+      "mean_token_accuracy": 0.4800183614715934,
+      "num_tokens": 73662683.0,
+      "step": 508
+    },
+    {
+      "entropy": 2.451171875,
+      "epoch": 0.008763547773387396,
+      "grad_norm": 0.5882863998413086,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.3984,
+      "mean_token_accuracy": 0.48627386754378676,
+      "num_tokens": 73798089.0,
+      "step": 509
+    },
+    {
+      "entropy": 2.4722900390625,
+      "epoch": 0.008780764959582655,
+      "grad_norm": 0.5577517747879028,
+      "learning_rate": 5.09e-06,
+      "loss": 2.4251,
+      "mean_token_accuracy": 0.472780239302665,
+      "num_tokens": 73951527.0,
+      "step": 510
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.008797982145777915,
+      "grad_norm": 0.5598065257072449,
+      "learning_rate": 5.1e-06,
+      "loss": 2.3522,
+      "mean_token_accuracy": 0.4847842915914953,
+      "num_tokens": 74106602.0,
+      "step": 511
+    },
+    {
+      "entropy": 2.4395751953125,
+      "epoch": 0.008815199331973176,
+      "grad_norm": 0.6455389857292175,
+      "learning_rate": 5.11e-06,
+      "loss": 2.4265,
+      "mean_token_accuracy": 0.47678500413894653,
+      "num_tokens": 74248860.0,
+      "step": 512
+    },
+    {
+      "entropy": 2.429443359375,
+      "epoch": 0.008832416518168436,
+      "grad_norm": 0.5287806391716003,
+      "learning_rate": 5.12e-06,
+      "loss": 2.3812,
+      "mean_token_accuracy": 0.4817400835454464,
+      "num_tokens": 74406358.0,
+      "step": 513
+    },
+    {
+      "entropy": 2.409423828125,
+      "epoch": 0.008849633704363697,
+      "grad_norm": 0.5710276961326599,
+      "learning_rate": 5.130000000000001e-06,
+      "loss": 2.3643,
+      "mean_token_accuracy": 0.4914928264915943,
+      "num_tokens": 74556098.0,
+      "step": 514
+    },
+    {
+      "entropy": 2.4224853515625,
+      "epoch": 0.008866850890558955,
+      "grad_norm": 0.5766054391860962,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.372,
+      "mean_token_accuracy": 0.48493917658925056,
+      "num_tokens": 74692650.0,
+      "step": 515
+    },
+    {
+      "entropy": 2.424072265625,
+      "epoch": 0.008884068076754216,
+      "grad_norm": 0.6063751578330994,
+      "learning_rate": 5.150000000000001e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.48274309001863003,
+      "num_tokens": 74834650.0,
+      "step": 516
+    },
+    {
+      "entropy": 2.4222412109375,
+      "epoch": 0.008901285262949477,
+      "grad_norm": 0.5876448750495911,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.3979,
+      "mean_token_accuracy": 0.48454813193529844,
+      "num_tokens": 74975623.0,
+      "step": 517
+    },
+    {
+      "entropy": 2.4189453125,
+      "epoch": 0.008918502449144737,
+      "grad_norm": 0.5868141055107117,
+      "learning_rate": 5.1700000000000005e-06,
+      "loss": 2.3741,
+      "mean_token_accuracy": 0.48601944325491786,
+      "num_tokens": 75124642.0,
+      "step": 518
+    },
+    {
+      "entropy": 2.47705078125,
+      "epoch": 0.008935719635339996,
+      "grad_norm": 0.5537886023521423,
+      "learning_rate": 5.18e-06,
+      "loss": 2.4012,
+      "mean_token_accuracy": 0.47799683222547174,
+      "num_tokens": 75268718.0,
+      "step": 519
+    },
+    {
+      "entropy": 2.400390625,
+      "epoch": 0.008952936821535256,
+      "grad_norm": 0.573029637336731,
+      "learning_rate": 5.19e-06,
+      "loss": 2.3484,
+      "mean_token_accuracy": 0.48853049660101533,
+      "num_tokens": 75407285.0,
+      "step": 520
+    },
+    {
+      "entropy": 2.458251953125,
+      "epoch": 0.008970154007730517,
+      "grad_norm": 0.6153673529624939,
+      "learning_rate": 5.2e-06,
+      "loss": 2.4012,
+      "mean_token_accuracy": 0.4795549106784165,
+      "num_tokens": 75544196.0,
+      "step": 521
+    },
+    {
+      "entropy": 2.474853515625,
+      "epoch": 0.008987371193925777,
+      "grad_norm": 0.6100261807441711,
+      "learning_rate": 5.210000000000001e-06,
+      "loss": 2.4373,
+      "mean_token_accuracy": 0.47731659887358546,
+      "num_tokens": 75675964.0,
+      "step": 522
+    },
+    {
+      "entropy": 2.486328125,
+      "epoch": 0.009004588380121036,
+      "grad_norm": 0.5929141044616699,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.4467,
+      "mean_token_accuracy": 0.4785361369140446,
+      "num_tokens": 75820140.0,
+      "step": 523
+    },
+    {
+      "entropy": 2.472900390625,
+      "epoch": 0.009021805566316297,
+      "grad_norm": 0.5526888370513916,
+      "learning_rate": 5.230000000000001e-06,
+      "loss": 2.393,
+      "mean_token_accuracy": 0.4825805053114891,
+      "num_tokens": 75975261.0,
+      "step": 524
+    },
+    {
+      "entropy": 2.382080078125,
+      "epoch": 0.009039022752511557,
+      "grad_norm": 0.5640392303466797,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.3364,
+      "mean_token_accuracy": 0.48861693032085896,
+      "num_tokens": 76121140.0,
+      "step": 525
+    },
+    {
+      "entropy": 2.4207763671875,
+      "epoch": 0.009056239938706818,
+      "grad_norm": 0.626315712928772,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 2.3691,
+      "mean_token_accuracy": 0.4876149254851043,
+      "num_tokens": 76249826.0,
+      "step": 526
+    },
+    {
+      "entropy": 2.468505859375,
+      "epoch": 0.009073457124902077,
+      "grad_norm": 0.5884665846824646,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.4473,
+      "mean_token_accuracy": 0.479018023237586,
+      "num_tokens": 76382197.0,
+      "step": 527
+    },
+    {
+      "entropy": 2.551513671875,
+      "epoch": 0.009090674311097337,
+      "grad_norm": 0.6032463908195496,
+      "learning_rate": 5.27e-06,
+      "loss": 2.5197,
+      "mean_token_accuracy": 0.4695024830289185,
+      "num_tokens": 76523850.0,
+      "step": 528
+    },
+    {
+      "entropy": 2.5267333984375,
+      "epoch": 0.009107891497292598,
+      "grad_norm": 0.5680440068244934,
+      "learning_rate": 5.28e-06,
+      "loss": 2.4922,
+      "mean_token_accuracy": 0.4694391922093928,
+      "num_tokens": 76675341.0,
+      "step": 529
+    },
+    {
+      "entropy": 2.45654296875,
+      "epoch": 0.009125108683487858,
+      "grad_norm": 0.5584273338317871,
+      "learning_rate": 5.290000000000001e-06,
+      "loss": 2.4558,
+      "mean_token_accuracy": 0.4753922396339476,
+      "num_tokens": 76839148.0,
+      "step": 530
+    },
+    {
+      "entropy": 2.447998046875,
+      "epoch": 0.009142325869683117,
+      "grad_norm": 0.5700913667678833,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.3795,
+      "mean_token_accuracy": 0.4812642466276884,
+      "num_tokens": 76972890.0,
+      "step": 531
+    },
+    {
+      "entropy": 2.39208984375,
+      "epoch": 0.009159543055878377,
+      "grad_norm": 0.5814207196235657,
+      "learning_rate": 5.310000000000001e-06,
+      "loss": 2.3494,
+      "mean_token_accuracy": 0.4927141284570098,
+      "num_tokens": 77116305.0,
+      "step": 532
+    },
+    {
+      "entropy": 2.5166015625,
+      "epoch": 0.009176760242073638,
+      "grad_norm": 0.6363142728805542,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.4734,
+      "mean_token_accuracy": 0.46889354614540935,
+      "num_tokens": 77241586.0,
+      "step": 533
+    },
+    {
+      "entropy": 2.561279296875,
+      "epoch": 0.009193977428268899,
+      "grad_norm": 0.5870820879936218,
+      "learning_rate": 5.330000000000001e-06,
+      "loss": 2.4981,
+      "mean_token_accuracy": 0.4720300040207803,
+      "num_tokens": 77382449.0,
+      "step": 534
+    },
+    {
+      "entropy": 2.4193115234375,
+      "epoch": 0.009211194614464157,
+      "grad_norm": 0.5664909482002258,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.3611,
+      "mean_token_accuracy": 0.48326091514900327,
+      "num_tokens": 77519348.0,
+      "step": 535
+    },
+    {
+      "entropy": 2.3780517578125,
+      "epoch": 0.009228411800659418,
+      "grad_norm": 0.677797794342041,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 2.3674,
+      "mean_token_accuracy": 0.49126658914610744,
+      "num_tokens": 77672043.0,
+      "step": 536
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.009245628986854678,
+      "grad_norm": 0.5524245500564575,
+      "learning_rate": 5.36e-06,
+      "loss": 2.4252,
+      "mean_token_accuracy": 0.47922995453700423,
+      "num_tokens": 77815854.0,
+      "step": 537
+    },
+    {
+      "entropy": 2.3614501953125,
+      "epoch": 0.009262846173049939,
+      "grad_norm": 0.5834474563598633,
+      "learning_rate": 5.370000000000001e-06,
+      "loss": 2.3147,
+      "mean_token_accuracy": 0.4956310335546732,
+      "num_tokens": 77971110.0,
+      "step": 538
+    },
+    {
+      "entropy": 2.4951171875,
+      "epoch": 0.0092800633592452,
+      "grad_norm": 0.5798099040985107,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.4241,
+      "mean_token_accuracy": 0.4767613257281482,
+      "num_tokens": 78111794.0,
+      "step": 539
+    },
+    {
+      "entropy": 2.502197265625,
+      "epoch": 0.009297280545440458,
+      "grad_norm": 0.6156012415885925,
+      "learning_rate": 5.390000000000001e-06,
+      "loss": 2.4368,
+      "mean_token_accuracy": 0.4805396613664925,
+      "num_tokens": 78259210.0,
+      "step": 540
+    },
+    {
+      "entropy": 2.4141845703125,
+      "epoch": 0.009314497731635719,
+      "grad_norm": 0.5888620615005493,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.3621,
+      "mean_token_accuracy": 0.4862367841415107,
+      "num_tokens": 78400181.0,
+      "step": 541
+    },
+    {
+      "entropy": 2.41845703125,
+      "epoch": 0.00933171491783098,
+      "grad_norm": 0.5387586355209351,
+      "learning_rate": 5.410000000000001e-06,
+      "loss": 2.3808,
+      "mean_token_accuracy": 0.48267924739047885,
+      "num_tokens": 78558105.0,
+      "step": 542
+    },
+    {
+      "entropy": 2.4661865234375,
+      "epoch": 0.00934893210402624,
+      "grad_norm": 0.5678828954696655,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.4037,
+      "mean_token_accuracy": 0.4831194179132581,
+      "num_tokens": 78706162.0,
+      "step": 543
+    },
+    {
+      "entropy": 2.3319091796875,
+      "epoch": 0.009366149290221499,
+      "grad_norm": 0.5745298862457275,
+      "learning_rate": 5.4300000000000005e-06,
+      "loss": 2.2771,
+      "mean_token_accuracy": 0.5001829275861382,
+      "num_tokens": 78850606.0,
+      "step": 544
+    },
+    {
+      "entropy": 2.492919921875,
+      "epoch": 0.009383366476416759,
+      "grad_norm": 0.5902231931686401,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.4254,
+      "mean_token_accuracy": 0.4752844786271453,
+      "num_tokens": 78990742.0,
+      "step": 545
+    },
+    {
+      "entropy": 2.5008544921875,
+      "epoch": 0.00940058366261202,
+      "grad_norm": 0.6180260181427002,
+      "learning_rate": 5.450000000000001e-06,
+      "loss": 2.5212,
+      "mean_token_accuracy": 0.4779182830825448,
+      "num_tokens": 79147975.0,
+      "step": 546
+    },
+    {
+      "entropy": 2.451171875,
+      "epoch": 0.00941780084880728,
+      "grad_norm": 0.571796715259552,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.3884,
+      "mean_token_accuracy": 0.47992962412536144,
+      "num_tokens": 79295374.0,
+      "step": 547
+    },
+    {
+      "entropy": 2.451416015625,
+      "epoch": 0.009435018035002539,
+      "grad_norm": 0.6627357006072998,
+      "learning_rate": 5.470000000000001e-06,
+      "loss": 2.4386,
+      "mean_token_accuracy": 0.4789085192605853,
+      "num_tokens": 79435365.0,
+      "step": 548
+    },
+    {
+      "entropy": 2.44091796875,
+      "epoch": 0.0094522352211978,
+      "grad_norm": 0.5701068639755249,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.3711,
+      "mean_token_accuracy": 0.48519957158714533,
+      "num_tokens": 79583578.0,
+      "step": 549
+    },
+    {
+      "entropy": 2.4254150390625,
+      "epoch": 0.00946945240739306,
+      "grad_norm": 0.5793823599815369,
+      "learning_rate": 5.490000000000001e-06,
+      "loss": 2.3977,
+      "mean_token_accuracy": 0.4862471236847341,
+      "num_tokens": 79731666.0,
+      "step": 550
+    },
+    {
+      "entropy": 2.4296875,
+      "epoch": 0.00948666959358832,
+      "grad_norm": 0.538820743560791,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.3655,
+      "mean_token_accuracy": 0.487091563642025,
+      "num_tokens": 79879814.0,
+      "step": 551
+    },
+    {
+      "entropy": 2.5390625,
+      "epoch": 0.00950388677978358,
+      "grad_norm": 0.5948156714439392,
+      "learning_rate": 5.510000000000001e-06,
+      "loss": 2.4975,
+      "mean_token_accuracy": 0.46823509922251105,
+      "num_tokens": 80021770.0,
+      "step": 552
+    },
+    {
+      "entropy": 2.455322265625,
+      "epoch": 0.00952110396597884,
+      "grad_norm": 0.5883845686912537,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.3802,
+      "mean_token_accuracy": 0.47493234230205417,
+      "num_tokens": 80165630.0,
+      "step": 553
+    },
+    {
+      "entropy": 2.474365234375,
+      "epoch": 0.0095383211521741,
+      "grad_norm": 0.6061490774154663,
+      "learning_rate": 5.530000000000001e-06,
+      "loss": 2.4588,
+      "mean_token_accuracy": 0.4762619035318494,
+      "num_tokens": 80320803.0,
+      "step": 554
+    },
+    {
+      "entropy": 2.3839111328125,
+      "epoch": 0.009555538338369361,
+      "grad_norm": 0.5892252922058105,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.3663,
+      "mean_token_accuracy": 0.48988539073616266,
+      "num_tokens": 80455211.0,
+      "step": 555
+    },
+    {
+      "entropy": 2.3966064453125,
+      "epoch": 0.00957275552456462,
+      "grad_norm": 0.5676023960113525,
+      "learning_rate": 5.550000000000001e-06,
+      "loss": 2.3757,
+      "mean_token_accuracy": 0.4864460411481559,
+      "num_tokens": 80600140.0,
+      "step": 556
+    },
+    {
+      "entropy": 2.411865234375,
+      "epoch": 0.00958997271075988,
+      "grad_norm": 0.6370622515678406,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.4048,
+      "mean_token_accuracy": 0.4831241965293884,
+      "num_tokens": 80733228.0,
+      "step": 557
+    },
+    {
+      "entropy": 2.453369140625,
+      "epoch": 0.00960718989695514,
+      "grad_norm": 0.6127820611000061,
+      "learning_rate": 5.570000000000001e-06,
+      "loss": 2.4203,
+      "mean_token_accuracy": 0.48309980193153024,
+      "num_tokens": 80870174.0,
+      "step": 558
+    },
+    {
+      "entropy": 2.4246826171875,
+      "epoch": 0.009624407083150401,
+      "grad_norm": 0.6448102593421936,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.424,
+      "mean_token_accuracy": 0.4830983644351363,
+      "num_tokens": 81027247.0,
+      "step": 559
+    },
+    {
+      "entropy": 2.4490966796875,
+      "epoch": 0.00964162426934566,
+      "grad_norm": 0.6132236123085022,
+      "learning_rate": 5.590000000000001e-06,
+      "loss": 2.3908,
+      "mean_token_accuracy": 0.48710555862635374,
+      "num_tokens": 81181289.0,
+      "step": 560
+    },
+    {
+      "entropy": 2.4490966796875,
+      "epoch": 0.00965884145554092,
+      "grad_norm": 0.5913870334625244,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.4276,
+      "mean_token_accuracy": 0.47844883892685175,
+      "num_tokens": 81317554.0,
+      "step": 561
+    },
+    {
+      "entropy": 2.3934326171875,
+      "epoch": 0.009676058641736181,
+      "grad_norm": 0.5637186169624329,
+      "learning_rate": 5.610000000000001e-06,
+      "loss": 2.386,
+      "mean_token_accuracy": 0.49396452866494656,
+      "num_tokens": 81464385.0,
+      "step": 562
+    },
+    {
+      "entropy": 2.446044921875,
+      "epoch": 0.009693275827931442,
+      "grad_norm": 0.6086047887802124,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.3744,
+      "mean_token_accuracy": 0.4851002893410623,
+      "num_tokens": 81595954.0,
+      "step": 563
+    },
+    {
+      "entropy": 2.514404296875,
+      "epoch": 0.0097104930141267,
+      "grad_norm": 0.5994987487792969,
+      "learning_rate": 5.63e-06,
+      "loss": 2.445,
+      "mean_token_accuracy": 0.47357298992574215,
+      "num_tokens": 81731590.0,
+      "step": 564
+    },
+    {
+      "entropy": 2.452392578125,
+      "epoch": 0.009727710200321961,
+      "grad_norm": 0.5923967361450195,
+      "learning_rate": 5.64e-06,
+      "loss": 2.4243,
+      "mean_token_accuracy": 0.48147726710885763,
+      "num_tokens": 81874631.0,
+      "step": 565
+    },
+    {
+      "entropy": 2.49462890625,
+      "epoch": 0.009744927386517221,
+      "grad_norm": 0.5803849697113037,
+      "learning_rate": 5.65e-06,
+      "loss": 2.4918,
+      "mean_token_accuracy": 0.4728530729189515,
+      "num_tokens": 82032208.0,
+      "step": 566
+    },
+    {
+      "entropy": 2.5015869140625,
+      "epoch": 0.009762144572712482,
+      "grad_norm": 0.5941510200500488,
+      "learning_rate": 5.66e-06,
+      "loss": 2.4491,
+      "mean_token_accuracy": 0.4712064489722252,
+      "num_tokens": 82175645.0,
+      "step": 567
+    },
+    {
+      "entropy": 2.453125,
+      "epoch": 0.009779361758907743,
+      "grad_norm": 0.5803316831588745,
+      "learning_rate": 5.67e-06,
+      "loss": 2.407,
+      "mean_token_accuracy": 0.4795740279369056,
+      "num_tokens": 82314242.0,
+      "step": 568
+    },
+    {
+      "entropy": 2.4854736328125,
+      "epoch": 0.009796578945103001,
+      "grad_norm": 0.6026585698127747,
+      "learning_rate": 5.68e-06,
+      "loss": 2.4803,
+      "mean_token_accuracy": 0.4731423007324338,
+      "num_tokens": 82441744.0,
+      "step": 569
+    },
+    {
+      "entropy": 2.469482421875,
+      "epoch": 0.009813796131298262,
+      "grad_norm": 0.5523576140403748,
+      "learning_rate": 5.69e-06,
+      "loss": 2.4117,
+      "mean_token_accuracy": 0.4742597243748605,
+      "num_tokens": 82586104.0,
+      "step": 570
+    },
+    {
+      "entropy": 2.420166015625,
+      "epoch": 0.009831013317493522,
+      "grad_norm": 0.5471956133842468,
+      "learning_rate": 5.7e-06,
+      "loss": 2.3751,
+      "mean_token_accuracy": 0.48100833920761943,
+      "num_tokens": 82736506.0,
+      "step": 571
+    },
+    {
+      "entropy": 2.5037841796875,
+      "epoch": 0.009848230503688783,
+      "grad_norm": 0.57561856508255,
+      "learning_rate": 5.71e-06,
+      "loss": 2.4685,
+      "mean_token_accuracy": 0.46707925060763955,
+      "num_tokens": 82889738.0,
+      "step": 572
+    },
+    {
+      "entropy": 2.450927734375,
+      "epoch": 0.009865447689884042,
+      "grad_norm": 0.5832479596138,
+      "learning_rate": 5.72e-06,
+      "loss": 2.4341,
+      "mean_token_accuracy": 0.4823454241268337,
+      "num_tokens": 83033413.0,
+      "step": 573
+    },
+    {
+      "entropy": 2.460205078125,
+      "epoch": 0.009882664876079302,
+      "grad_norm": 0.546006977558136,
+      "learning_rate": 5.73e-06,
+      "loss": 2.4431,
+      "mean_token_accuracy": 0.4789656042121351,
+      "num_tokens": 83185297.0,
+      "step": 574
+    },
+    {
+      "entropy": 2.4942626953125,
+      "epoch": 0.009899882062274563,
+      "grad_norm": 0.5814852118492126,
+      "learning_rate": 5.74e-06,
+      "loss": 2.4655,
+      "mean_token_accuracy": 0.469060396309942,
+      "num_tokens": 83323076.0,
+      "step": 575
+    },
+    {
+      "entropy": 2.448486328125,
+      "epoch": 0.009917099248469823,
+      "grad_norm": 0.6251148581504822,
+      "learning_rate": 5.75e-06,
+      "loss": 2.3796,
+      "mean_token_accuracy": 0.48163892794400454,
+      "num_tokens": 83476832.0,
+      "step": 576
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.009934316434665082,
+      "grad_norm": 0.5571739077568054,
+      "learning_rate": 5.76e-06,
+      "loss": 2.4166,
+      "mean_token_accuracy": 0.47773123253136873,
+      "num_tokens": 83636406.0,
+      "step": 577
+    },
+    {
+      "entropy": 2.4129638671875,
+      "epoch": 0.009951533620860343,
+      "grad_norm": 0.5774782299995422,
+      "learning_rate": 5.77e-06,
+      "loss": 2.3531,
+      "mean_token_accuracy": 0.4910357999615371,
+      "num_tokens": 83784239.0,
+      "step": 578
+    },
+    {
+      "entropy": 2.395263671875,
+      "epoch": 0.009968750807055603,
+      "grad_norm": 0.5799167156219482,
+      "learning_rate": 5.78e-06,
+      "loss": 2.3157,
+      "mean_token_accuracy": 0.49086295487359166,
+      "num_tokens": 83930135.0,
+      "step": 579
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.009985967993250864,
+      "grad_norm": 0.9486396312713623,
+      "learning_rate": 5.7900000000000005e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.4829951082356274,
+      "num_tokens": 84069466.0,
+      "step": 580
+    },
+    {
+      "entropy": 2.456298828125,
+      "epoch": 0.010003185179446122,
+      "grad_norm": 0.5900285243988037,
+      "learning_rate": 5.8e-06,
+      "loss": 2.4223,
+      "mean_token_accuracy": 0.47999299177899957,
+      "num_tokens": 84204231.0,
+      "step": 581
+    },
+    {
+      "entropy": 2.4512939453125,
+      "epoch": 0.010020402365641383,
+      "grad_norm": 0.612040638923645,
+      "learning_rate": 5.81e-06,
+      "loss": 2.4226,
+      "mean_token_accuracy": 0.4809495103545487,
+      "num_tokens": 84341875.0,
+      "step": 582
+    },
+    {
+      "entropy": 2.44140625,
+      "epoch": 0.010037619551836643,
+      "grad_norm": 0.5587046146392822,
+      "learning_rate": 5.82e-06,
+      "loss": 2.3725,
+      "mean_token_accuracy": 0.4807244557887316,
+      "num_tokens": 84495749.0,
+      "step": 583
+    },
+    {
+      "entropy": 2.453125,
+      "epoch": 0.010054836738031904,
+      "grad_norm": 0.5574972629547119,
+      "learning_rate": 5.83e-06,
+      "loss": 2.4249,
+      "mean_token_accuracy": 0.47708367416635156,
+      "num_tokens": 84642208.0,
+      "step": 584
+    },
+    {
+      "entropy": 2.4827880859375,
+      "epoch": 0.010072053924227163,
+      "grad_norm": 0.5825822353363037,
+      "learning_rate": 5.84e-06,
+      "loss": 2.4421,
+      "mean_token_accuracy": 0.47600942524150014,
+      "num_tokens": 84798422.0,
+      "step": 585
+    },
+    {
+      "entropy": 2.4530029296875,
+      "epoch": 0.010089271110422423,
+      "grad_norm": 0.6338469386100769,
+      "learning_rate": 5.85e-06,
+      "loss": 2.3441,
+      "mean_token_accuracy": 0.48570866556838155,
+      "num_tokens": 84932951.0,
+      "step": 586
+    },
+    {
+      "entropy": 2.51806640625,
+      "epoch": 0.010106488296617684,
+      "grad_norm": 0.5964832305908203,
+      "learning_rate": 5.86e-06,
+      "loss": 2.4671,
+      "mean_token_accuracy": 0.47985436767339706,
+      "num_tokens": 85070544.0,
+      "step": 587
+    },
+    {
+      "entropy": 2.40576171875,
+      "epoch": 0.010123705482812944,
+      "grad_norm": 0.5808175206184387,
+      "learning_rate": 5.8700000000000005e-06,
+      "loss": 2.3715,
+      "mean_token_accuracy": 0.4833093252964318,
+      "num_tokens": 85216000.0,
+      "step": 588
+    },
+    {
+      "entropy": 2.3857421875,
+      "epoch": 0.010140922669008203,
+      "grad_norm": 0.5820797681808472,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.3318,
+      "mean_token_accuracy": 0.4918548730202019,
+      "num_tokens": 85367592.0,
+      "step": 589
+    },
+    {
+      "entropy": 2.4697265625,
+      "epoch": 0.010158139855203464,
+      "grad_norm": 0.5501261353492737,
+      "learning_rate": 5.89e-06,
+      "loss": 2.4305,
+      "mean_token_accuracy": 0.47398777306079865,
+      "num_tokens": 85507669.0,
+      "step": 590
+    },
+    {
+      "entropy": 2.3985595703125,
+      "epoch": 0.010175357041398724,
+      "grad_norm": 0.5922814607620239,
+      "learning_rate": 5.9e-06,
+      "loss": 2.3907,
+      "mean_token_accuracy": 0.49256287002936006,
+      "num_tokens": 85657583.0,
+      "step": 591
+    },
+    {
+      "entropy": 2.4813232421875,
+      "epoch": 0.010192574227593985,
+      "grad_norm": 0.6180179715156555,
+      "learning_rate": 5.91e-06,
+      "loss": 2.4081,
+      "mean_token_accuracy": 0.47511684615164995,
+      "num_tokens": 85796530.0,
+      "step": 592
+    },
+    {
+      "entropy": 2.548583984375,
+      "epoch": 0.010209791413789245,
+      "grad_norm": 0.586715817451477,
+      "learning_rate": 5.92e-06,
+      "loss": 2.4952,
+      "mean_token_accuracy": 0.4641613829880953,
+      "num_tokens": 85935315.0,
+      "step": 593
+    },
+    {
+      "entropy": 2.479736328125,
+      "epoch": 0.010227008599984504,
+      "grad_norm": 0.5589325428009033,
+      "learning_rate": 5.93e-06,
+      "loss": 2.4362,
+      "mean_token_accuracy": 0.4785456941463053,
+      "num_tokens": 86083670.0,
+      "step": 594
+    },
+    {
+      "entropy": 2.485107421875,
+      "epoch": 0.010244225786179765,
+      "grad_norm": 0.5777843594551086,
+      "learning_rate": 5.94e-06,
+      "loss": 2.4487,
+      "mean_token_accuracy": 0.47643798124045134,
+      "num_tokens": 86226759.0,
+      "step": 595
+    },
+    {
+      "entropy": 2.4775390625,
+      "epoch": 0.010261442972375025,
+      "grad_norm": 0.5679555535316467,
+      "learning_rate": 5.950000000000001e-06,
+      "loss": 2.4196,
+      "mean_token_accuracy": 0.48259661020711064,
+      "num_tokens": 86374974.0,
+      "step": 596
+    },
+    {
+      "entropy": 2.423095703125,
+      "epoch": 0.010278660158570286,
+      "grad_norm": 0.5793547034263611,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.418,
+      "mean_token_accuracy": 0.48791359877213836,
+      "num_tokens": 86517518.0,
+      "step": 597
+    },
+    {
+      "entropy": 2.4588623046875,
+      "epoch": 0.010295877344765544,
+      "grad_norm": 0.6030251979827881,
+      "learning_rate": 5.9700000000000004e-06,
+      "loss": 2.4074,
+      "mean_token_accuracy": 0.48480180045589805,
+      "num_tokens": 86647346.0,
+      "step": 598
+    },
+    {
+      "entropy": 2.570068359375,
+      "epoch": 0.010313094530960805,
+      "grad_norm": 0.5821578502655029,
+      "learning_rate": 5.98e-06,
+      "loss": 2.5287,
+      "mean_token_accuracy": 0.4660195717588067,
+      "num_tokens": 86790455.0,
+      "step": 599
+    },
+    {
+      "entropy": 2.4483642578125,
+      "epoch": 0.010330311717156065,
+      "grad_norm": 0.5719680786132812,
+      "learning_rate": 5.99e-06,
+      "loss": 2.357,
+      "mean_token_accuracy": 0.48361520702019334,
+      "num_tokens": 86939582.0,
+      "step": 600
+    },
+    {
+      "entropy": 2.5018310546875,
+      "epoch": 0.010347528903351326,
+      "grad_norm": 0.6109415888786316,
+      "learning_rate": 6e-06,
+      "loss": 2.4647,
+      "mean_token_accuracy": 0.47726114513352513,
+      "num_tokens": 87097332.0,
+      "step": 601
+    },
+    {
+      "entropy": 2.4161376953125,
+      "epoch": 0.010364746089546585,
+      "grad_norm": 0.565535306930542,
+      "learning_rate": 6.01e-06,
+      "loss": 2.3436,
+      "mean_token_accuracy": 0.4876682967878878,
+      "num_tokens": 87246475.0,
+      "step": 602
+    },
+    {
+      "entropy": 2.3807373046875,
+      "epoch": 0.010381963275741845,
+      "grad_norm": 0.594390332698822,
+      "learning_rate": 6.02e-06,
+      "loss": 2.3292,
+      "mean_token_accuracy": 0.49000673089176416,
+      "num_tokens": 87402167.0,
+      "step": 603
+    },
+    {
+      "entropy": 2.489501953125,
+      "epoch": 0.010399180461937106,
+      "grad_norm": 0.5601812601089478,
+      "learning_rate": 6.030000000000001e-06,
+      "loss": 2.4547,
+      "mean_token_accuracy": 0.470840523019433,
+      "num_tokens": 87546749.0,
+      "step": 604
+    },
+    {
+      "entropy": 2.491455078125,
+      "epoch": 0.010416397648132366,
+      "grad_norm": 0.6872953772544861,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.4646,
+      "mean_token_accuracy": 0.47835929365828633,
+      "num_tokens": 87680924.0,
+      "step": 605
+    },
+    {
+      "entropy": 2.499267578125,
+      "epoch": 0.010433614834327625,
+      "grad_norm": 0.5995067954063416,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 2.4872,
+      "mean_token_accuracy": 0.4722044039517641,
+      "num_tokens": 87818407.0,
+      "step": 606
+    },
+    {
+      "entropy": 2.4085693359375,
+      "epoch": 0.010450832020522886,
+      "grad_norm": 0.7130571007728577,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.3356,
+      "mean_token_accuracy": 0.48175712302327156,
+      "num_tokens": 87964354.0,
+      "step": 607
+    },
+    {
+      "entropy": 2.40869140625,
+      "epoch": 0.010468049206718146,
+      "grad_norm": 0.5531951189041138,
+      "learning_rate": 6.07e-06,
+      "loss": 2.396,
+      "mean_token_accuracy": 0.4883636045269668,
+      "num_tokens": 88108908.0,
+      "step": 608
+    },
+    {
+      "entropy": 2.4320068359375,
+      "epoch": 0.010485266392913407,
+      "grad_norm": 0.6004809737205505,
+      "learning_rate": 6.08e-06,
+      "loss": 2.3704,
+      "mean_token_accuracy": 0.47937911935150623,
+      "num_tokens": 88252541.0,
+      "step": 609
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.010502483579108665,
+      "grad_norm": 0.5746989846229553,
+      "learning_rate": 6.09e-06,
+      "loss": 2.3784,
+      "mean_token_accuracy": 0.4842778267338872,
+      "num_tokens": 88387113.0,
+      "step": 610
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.010519700765303926,
+      "grad_norm": 0.6954113245010376,
+      "learning_rate": 6.1e-06,
+      "loss": 2.4565,
+      "mean_token_accuracy": 0.4831416751258075,
+      "num_tokens": 88538988.0,
+      "step": 611
+    },
+    {
+      "entropy": 2.4976806640625,
+      "epoch": 0.010536917951499187,
+      "grad_norm": 0.5616413950920105,
+      "learning_rate": 6.110000000000001e-06,
+      "loss": 2.4658,
+      "mean_token_accuracy": 0.47561208764091134,
+      "num_tokens": 88697407.0,
+      "step": 612
+    },
+    {
+      "entropy": 2.39599609375,
+      "epoch": 0.010554135137694447,
+      "grad_norm": 0.5953220129013062,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.3488,
+      "mean_token_accuracy": 0.4882765398360789,
+      "num_tokens": 88843276.0,
+      "step": 613
+    },
+    {
+      "entropy": 2.5074462890625,
+      "epoch": 0.010571352323889706,
+      "grad_norm": 0.5962725281715393,
+      "learning_rate": 6.130000000000001e-06,
+      "loss": 2.4542,
+      "mean_token_accuracy": 0.4768844782374799,
+      "num_tokens": 88973725.0,
+      "step": 614
+    },
+    {
+      "entropy": 2.416748046875,
+      "epoch": 0.010588569510084966,
+      "grad_norm": 0.5971546769142151,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.384,
+      "mean_token_accuracy": 0.4932720325887203,
+      "num_tokens": 89112891.0,
+      "step": 615
+    },
+    {
+      "entropy": 2.4598388671875,
+      "epoch": 0.010605786696280227,
+      "grad_norm": 0.5331761837005615,
+      "learning_rate": 6.15e-06,
+      "loss": 2.3966,
+      "mean_token_accuracy": 0.4843857139348984,
+      "num_tokens": 89261669.0,
+      "step": 616
+    },
+    {
+      "entropy": 2.5048828125,
+      "epoch": 0.010623003882475487,
+      "grad_norm": 0.5298826694488525,
+      "learning_rate": 6.16e-06,
+      "loss": 2.4497,
+      "mean_token_accuracy": 0.47414906043559313,
+      "num_tokens": 89424640.0,
+      "step": 617
+    },
+    {
+      "entropy": 2.53369140625,
+      "epoch": 0.010640221068670748,
+      "grad_norm": 0.5918136835098267,
+      "learning_rate": 6.17e-06,
+      "loss": 2.4991,
+      "mean_token_accuracy": 0.4715537903830409,
+      "num_tokens": 89565111.0,
+      "step": 618
+    },
+    {
+      "entropy": 2.502685546875,
+      "epoch": 0.010657438254866007,
+      "grad_norm": 0.5514756441116333,
+      "learning_rate": 6.18e-06,
+      "loss": 2.4796,
+      "mean_token_accuracy": 0.47532700607553124,
+      "num_tokens": 89720834.0,
+      "step": 619
+    },
+    {
+      "entropy": 2.4874267578125,
+      "epoch": 0.010674655441061267,
+      "grad_norm": 0.6075375080108643,
+      "learning_rate": 6.190000000000001e-06,
+      "loss": 2.434,
+      "mean_token_accuracy": 0.48047855822369456,
+      "num_tokens": 89852579.0,
+      "step": 620
+    },
+    {
+      "entropy": 2.4002685546875,
+      "epoch": 0.010691872627256528,
+      "grad_norm": 0.5313543081283569,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.3702,
+      "mean_token_accuracy": 0.4890581676736474,
+      "num_tokens": 90014576.0,
+      "step": 621
+    },
+    {
+      "entropy": 2.5064697265625,
+      "epoch": 0.010709089813451788,
+      "grad_norm": 0.5929763317108154,
+      "learning_rate": 6.210000000000001e-06,
+      "loss": 2.4956,
+      "mean_token_accuracy": 0.47517322562634945,
+      "num_tokens": 90141327.0,
+      "step": 622
+    },
+    {
+      "entropy": 2.4847412109375,
+      "epoch": 0.010726306999647047,
+      "grad_norm": 0.6306544542312622,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.4664,
+      "mean_token_accuracy": 0.4766456396318972,
+      "num_tokens": 90289805.0,
+      "step": 623
+    },
+    {
+      "entropy": 2.434326171875,
+      "epoch": 0.010743524185842308,
+      "grad_norm": 0.6184057593345642,
+      "learning_rate": 6.2300000000000005e-06,
+      "loss": 2.4083,
+      "mean_token_accuracy": 0.48298388347029686,
+      "num_tokens": 90436897.0,
+      "step": 624
+    },
+    {
+      "entropy": 2.4671630859375,
+      "epoch": 0.010760741372037568,
+      "grad_norm": 2.9163689613342285,
+      "learning_rate": 6.24e-06,
+      "loss": 2.4082,
+      "mean_token_accuracy": 0.48466121684759855,
+      "num_tokens": 90587723.0,
+      "step": 625
+    },
+    {
+      "entropy": 2.4171142578125,
+      "epoch": 0.010777958558232829,
+      "grad_norm": 0.6109045743942261,
+      "learning_rate": 6.25e-06,
+      "loss": 2.3642,
+      "mean_token_accuracy": 0.48781463131308556,
+      "num_tokens": 90722550.0,
+      "step": 626
+    },
+    {
+      "entropy": 2.3975830078125,
+      "epoch": 0.010795175744428087,
+      "grad_norm": 0.5503095388412476,
+      "learning_rate": 6.26e-06,
+      "loss": 2.3723,
+      "mean_token_accuracy": 0.484878970310092,
+      "num_tokens": 90883612.0,
+      "step": 627
+    },
+    {
+      "entropy": 2.435302734375,
+      "epoch": 0.010812392930623348,
+      "grad_norm": 0.5557761192321777,
+      "learning_rate": 6.27e-06,
+      "loss": 2.4024,
+      "mean_token_accuracy": 0.482346317730844,
+      "num_tokens": 91020885.0,
+      "step": 628
+    },
+    {
+      "entropy": 2.454833984375,
+      "epoch": 0.010829610116818609,
+      "grad_norm": 0.5977523326873779,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.3842,
+      "mean_token_accuracy": 0.4845750341191888,
+      "num_tokens": 91183050.0,
+      "step": 629
+    },
+    {
+      "entropy": 2.536376953125,
+      "epoch": 0.010846827303013869,
+      "grad_norm": 0.6027834415435791,
+      "learning_rate": 6.290000000000001e-06,
+      "loss": 2.5204,
+      "mean_token_accuracy": 0.46805495163425803,
+      "num_tokens": 91320603.0,
+      "step": 630
+    },
+    {
+      "entropy": 2.42431640625,
+      "epoch": 0.010864044489209128,
+      "grad_norm": 0.569492518901825,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.3634,
+      "mean_token_accuracy": 0.484631123021245,
+      "num_tokens": 91466774.0,
+      "step": 631
+    },
+    {
+      "entropy": 2.501953125,
+      "epoch": 0.010881261675404388,
+      "grad_norm": 0.6085266470909119,
+      "learning_rate": 6.3100000000000006e-06,
+      "loss": 2.4384,
+      "mean_token_accuracy": 0.47073780838400126,
+      "num_tokens": 91599571.0,
+      "step": 632
+    },
+    {
+      "entropy": 2.454833984375,
+      "epoch": 0.010898478861599649,
+      "grad_norm": 0.5897052884101868,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.3718,
+      "mean_token_accuracy": 0.4813077808357775,
+      "num_tokens": 91736850.0,
+      "step": 633
+    },
+    {
+      "entropy": 2.5152587890625,
+      "epoch": 0.01091569604779491,
+      "grad_norm": 0.5574195384979248,
+      "learning_rate": 6.33e-06,
+      "loss": 2.5027,
+      "mean_token_accuracy": 0.4707440370693803,
+      "num_tokens": 91883052.0,
+      "step": 634
+    },
+    {
+      "entropy": 2.4981689453125,
+      "epoch": 0.010932913233990168,
+      "grad_norm": 0.5665111541748047,
+      "learning_rate": 6.34e-06,
+      "loss": 2.4967,
+      "mean_token_accuracy": 0.4747963696718216,
+      "num_tokens": 92038624.0,
+      "step": 635
+    },
+    {
+      "entropy": 2.5341796875,
+      "epoch": 0.010950130420185429,
+      "grad_norm": 0.6099963784217834,
+      "learning_rate": 6.35e-06,
+      "loss": 2.4812,
+      "mean_token_accuracy": 0.4711012118496001,
+      "num_tokens": 92172927.0,
+      "step": 636
+    },
+    {
+      "entropy": 2.4072265625,
+      "epoch": 0.01096734760638069,
+      "grad_norm": 0.5225919485092163,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.3695,
+      "mean_token_accuracy": 0.4833218730054796,
+      "num_tokens": 92340412.0,
+      "step": 637
+    },
+    {
+      "entropy": 2.373779296875,
+      "epoch": 0.01098456479257595,
+      "grad_norm": 0.5803242921829224,
+      "learning_rate": 6.370000000000001e-06,
+      "loss": 2.3518,
+      "mean_token_accuracy": 0.4897113349288702,
+      "num_tokens": 92491100.0,
+      "step": 638
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.011001781978771209,
+      "grad_norm": 0.6239297986030579,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.4732,
+      "mean_token_accuracy": 0.4763565128669143,
+      "num_tokens": 92624388.0,
+      "step": 639
+    },
+    {
+      "entropy": 2.521728515625,
+      "epoch": 0.011018999164966469,
+      "grad_norm": 0.6022891998291016,
+      "learning_rate": 6.390000000000001e-06,
+      "loss": 2.5116,
+      "mean_token_accuracy": 0.4692745329812169,
+      "num_tokens": 92749943.0,
+      "step": 640
+    },
+    {
+      "entropy": 2.4610595703125,
+      "epoch": 0.01103621635116173,
+      "grad_norm": 0.5882202386856079,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.4055,
+      "mean_token_accuracy": 0.479529797565192,
+      "num_tokens": 92901869.0,
+      "step": 641
+    },
+    {
+      "entropy": 2.42626953125,
+      "epoch": 0.01105343353735699,
+      "grad_norm": 0.6111239790916443,
+      "learning_rate": 6.4100000000000005e-06,
+      "loss": 2.337,
+      "mean_token_accuracy": 0.4874786385335028,
+      "num_tokens": 93049790.0,
+      "step": 642
+    },
+    {
+      "entropy": 2.459716796875,
+      "epoch": 0.01107065072355225,
+      "grad_norm": 0.5824291706085205,
+      "learning_rate": 6.42e-06,
+      "loss": 2.3972,
+      "mean_token_accuracy": 0.4820692026987672,
+      "num_tokens": 93190066.0,
+      "step": 643
+    },
+    {
+      "entropy": 2.393798828125,
+      "epoch": 0.01108786790974751,
+      "grad_norm": 0.6439885497093201,
+      "learning_rate": 6.43e-06,
+      "loss": 2.2805,
+      "mean_token_accuracy": 0.500021081417799,
+      "num_tokens": 93323041.0,
+      "step": 644
+    },
+    {
+      "entropy": 2.5235595703125,
+      "epoch": 0.01110508509594277,
+      "grad_norm": 0.5925695300102234,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.5175,
+      "mean_token_accuracy": 0.46626862324774265,
+      "num_tokens": 93454663.0,
+      "step": 645
+    },
+    {
+      "entropy": 2.5662841796875,
+      "epoch": 0.01112230228213803,
+      "grad_norm": 0.5767838954925537,
+      "learning_rate": 6.450000000000001e-06,
+      "loss": 2.5828,
+      "mean_token_accuracy": 0.46684390073642135,
+      "num_tokens": 93600987.0,
+      "step": 646
+    },
+    {
+      "entropy": 2.445068359375,
+      "epoch": 0.011139519468333291,
+      "grad_norm": 0.5589511394500732,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.3973,
+      "mean_token_accuracy": 0.48293518042191863,
+      "num_tokens": 93764681.0,
+      "step": 647
+    },
+    {
+      "entropy": 2.487548828125,
+      "epoch": 0.01115673665452855,
+      "grad_norm": 0.6247239112854004,
+      "learning_rate": 6.470000000000001e-06,
+      "loss": 2.4299,
+      "mean_token_accuracy": 0.47755016293376684,
+      "num_tokens": 93908163.0,
+      "step": 648
+    },
+    {
+      "entropy": 2.44873046875,
+      "epoch": 0.01117395384072381,
+      "grad_norm": 0.5852839350700378,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.3616,
+      "mean_token_accuracy": 0.48453861800953746,
+      "num_tokens": 94044102.0,
+      "step": 649
+    },
+    {
+      "entropy": 2.401611328125,
+      "epoch": 0.011191171026919071,
+      "grad_norm": 0.5801386833190918,
+      "learning_rate": 6.4900000000000005e-06,
+      "loss": 2.3961,
+      "mean_token_accuracy": 0.4849433288909495,
+      "num_tokens": 94186933.0,
+      "step": 650
+    },
+    {
+      "entropy": 2.474853515625,
+      "epoch": 0.011208388213114331,
+      "grad_norm": 0.5861256718635559,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.4425,
+      "mean_token_accuracy": 0.47974030720070004,
+      "num_tokens": 94318045.0,
+      "step": 651
+    },
+    {
+      "entropy": 2.49755859375,
+      "epoch": 0.01122560539930959,
+      "grad_norm": 0.5657753348350525,
+      "learning_rate": 6.51e-06,
+      "loss": 2.4787,
+      "mean_token_accuracy": 0.4704547990113497,
+      "num_tokens": 94470298.0,
+      "step": 652
+    },
+    {
+      "entropy": 2.4617919921875,
+      "epoch": 0.01124282258550485,
+      "grad_norm": 0.5523881912231445,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.4404,
+      "mean_token_accuracy": 0.47263912204653025,
+      "num_tokens": 94617182.0,
+      "step": 653
+    },
+    {
+      "entropy": 2.462646484375,
+      "epoch": 0.011260039771700111,
+      "grad_norm": 0.5891516804695129,
+      "learning_rate": 6.530000000000001e-06,
+      "loss": 2.4207,
+      "mean_token_accuracy": 0.4738515946082771,
+      "num_tokens": 94764738.0,
+      "step": 654
+    },
+    {
+      "entropy": 2.43310546875,
+      "epoch": 0.011277256957895372,
+      "grad_norm": 0.5879467129707336,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 2.4553,
+      "mean_token_accuracy": 0.4857351207174361,
+      "num_tokens": 94911680.0,
+      "step": 655
+    },
+    {
+      "entropy": 2.3919677734375,
+      "epoch": 0.01129447414409063,
+      "grad_norm": 0.5698356032371521,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 2.3241,
+      "mean_token_accuracy": 0.49458998907357454,
+      "num_tokens": 95063211.0,
+      "step": 656
+    },
+    {
+      "entropy": 2.4443359375,
+      "epoch": 0.011311691330285891,
+      "grad_norm": 0.5569403767585754,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.4115,
+      "mean_token_accuracy": 0.4799956767819822,
+      "num_tokens": 95205223.0,
+      "step": 657
+    },
+    {
+      "entropy": 2.4415283203125,
+      "epoch": 0.011328908516481152,
+      "grad_norm": 0.6046442985534668,
+      "learning_rate": 6.570000000000001e-06,
+      "loss": 2.3966,
+      "mean_token_accuracy": 0.4849221669137478,
+      "num_tokens": 95353465.0,
+      "step": 658
+    },
+    {
+      "entropy": 2.4443359375,
+      "epoch": 0.011346125702676412,
+      "grad_norm": 0.7441838979721069,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.4102,
+      "mean_token_accuracy": 0.47888694843277335,
+      "num_tokens": 95519881.0,
+      "step": 659
+    },
+    {
+      "entropy": 2.47998046875,
+      "epoch": 0.011363342888871671,
+      "grad_norm": 0.5905322432518005,
+      "learning_rate": 6.5900000000000004e-06,
+      "loss": 2.457,
+      "mean_token_accuracy": 0.47512120427563787,
+      "num_tokens": 95664332.0,
+      "step": 660
+    },
+    {
+      "entropy": 2.43408203125,
+      "epoch": 0.011380560075066931,
+      "grad_norm": 0.5880439281463623,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.3549,
+      "mean_token_accuracy": 0.4882641164585948,
+      "num_tokens": 95811290.0,
+      "step": 661
+    },
+    {
+      "entropy": 2.44677734375,
+      "epoch": 0.011397777261262192,
+      "grad_norm": 0.6042869091033936,
+      "learning_rate": 6.610000000000001e-06,
+      "loss": 2.4018,
+      "mean_token_accuracy": 0.48763177869841456,
+      "num_tokens": 95950564.0,
+      "step": 662
+    },
+    {
+      "entropy": 2.4495849609375,
+      "epoch": 0.011414994447457453,
+      "grad_norm": 0.6228047013282776,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.4301,
+      "mean_token_accuracy": 0.4780231602489948,
+      "num_tokens": 96088537.0,
+      "step": 663
+    },
+    {
+      "entropy": 2.398681640625,
+      "epoch": 0.011432211633652711,
+      "grad_norm": 0.6014442443847656,
+      "learning_rate": 6.630000000000001e-06,
+      "loss": 2.4022,
+      "mean_token_accuracy": 0.4849638855084777,
+      "num_tokens": 96222517.0,
+      "step": 664
+    },
+    {
+      "entropy": 2.395751953125,
+      "epoch": 0.011449428819847972,
+      "grad_norm": 0.5773961544036865,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.3727,
+      "mean_token_accuracy": 0.48979697469621897,
+      "num_tokens": 96371664.0,
+      "step": 665
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.011466646006043232,
+      "grad_norm": 0.5553280711174011,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 2.4178,
+      "mean_token_accuracy": 0.4752015918493271,
+      "num_tokens": 96533336.0,
+      "step": 666
+    },
+    {
+      "entropy": 2.4317626953125,
+      "epoch": 0.011483863192238493,
+      "grad_norm": 0.5981423258781433,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.3897,
+      "mean_token_accuracy": 0.48191826045513153,
+      "num_tokens": 96671339.0,
+      "step": 667
+    },
+    {
+      "entropy": 2.4635009765625,
+      "epoch": 0.011501080378433753,
+      "grad_norm": 0.5966627597808838,
+      "learning_rate": 6.6700000000000005e-06,
+      "loss": 2.4742,
+      "mean_token_accuracy": 0.4762999969534576,
+      "num_tokens": 96814601.0,
+      "step": 668
+    },
+    {
+      "entropy": 2.5224609375,
+      "epoch": 0.011518297564629012,
+      "grad_norm": 0.6274897456169128,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.499,
+      "mean_token_accuracy": 0.4678979804739356,
+      "num_tokens": 96956119.0,
+      "step": 669
+    },
+    {
+      "entropy": 2.422607421875,
+      "epoch": 0.011535514750824273,
+      "grad_norm": 0.5479171276092529,
+      "learning_rate": 6.690000000000001e-06,
+      "loss": 2.4028,
+      "mean_token_accuracy": 0.48214570991694927,
+      "num_tokens": 97106907.0,
+      "step": 670
+    },
+    {
+      "entropy": 2.450439453125,
+      "epoch": 0.011552731937019533,
+      "grad_norm": 0.5556632876396179,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.4311,
+      "mean_token_accuracy": 0.4770152415148914,
+      "num_tokens": 97251230.0,
+      "step": 671
+    },
+    {
+      "entropy": 2.4766845703125,
+      "epoch": 0.011569949123214794,
+      "grad_norm": 0.5444607734680176,
+      "learning_rate": 6.710000000000001e-06,
+      "loss": 2.4384,
+      "mean_token_accuracy": 0.48311482975259423,
+      "num_tokens": 97404310.0,
+      "step": 672
+    },
+    {
+      "entropy": 2.4534912109375,
+      "epoch": 0.011587166309410053,
+      "grad_norm": 0.5732424259185791,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.3888,
+      "mean_token_accuracy": 0.48798013804480433,
+      "num_tokens": 97557441.0,
+      "step": 673
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.011604383495605313,
+      "grad_norm": 0.5952773094177246,
+      "learning_rate": 6.730000000000001e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.47982571227476,
+      "num_tokens": 97696645.0,
+      "step": 674
+    },
+    {
+      "entropy": 2.4781494140625,
+      "epoch": 0.011621600681800574,
+      "grad_norm": 0.6089589595794678,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.3921,
+      "mean_token_accuracy": 0.48180749313905835,
+      "num_tokens": 97836331.0,
+      "step": 675
+    },
+    {
+      "entropy": 2.4674072265625,
+      "epoch": 0.011638817867995834,
+      "grad_norm": 0.5993421077728271,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 2.4598,
+      "mean_token_accuracy": 0.4802471627481282,
+      "num_tokens": 97988494.0,
+      "step": 676
+    },
+    {
+      "entropy": 2.42919921875,
+      "epoch": 0.011656035054191093,
+      "grad_norm": 0.5687212944030762,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.382,
+      "mean_token_accuracy": 0.48455388378351927,
+      "num_tokens": 98148161.0,
+      "step": 677
+    },
+    {
+      "entropy": 2.43896484375,
+      "epoch": 0.011673252240386353,
+      "grad_norm": 0.5404056906700134,
+      "learning_rate": 6.770000000000001e-06,
+      "loss": 2.378,
+      "mean_token_accuracy": 0.48297660844400525,
+      "num_tokens": 98303865.0,
+      "step": 678
+    },
+    {
+      "entropy": 2.41552734375,
+      "epoch": 0.011690469426581614,
+      "grad_norm": 0.6042897701263428,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.3817,
+      "mean_token_accuracy": 0.4834853089414537,
+      "num_tokens": 98442926.0,
+      "step": 679
+    },
+    {
+      "entropy": 2.4442138671875,
+      "epoch": 0.011707686612776875,
+      "grad_norm": 0.5709783434867859,
+      "learning_rate": 6.790000000000001e-06,
+      "loss": 2.389,
+      "mean_token_accuracy": 0.4857576950453222,
+      "num_tokens": 98585693.0,
+      "step": 680
+    },
+    {
+      "entropy": 2.452392578125,
+      "epoch": 0.011724903798972133,
+      "grad_norm": 0.5298141837120056,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.3874,
+      "mean_token_accuracy": 0.486653549131006,
+      "num_tokens": 98748357.0,
+      "step": 681
+    },
+    {
+      "entropy": 2.470703125,
+      "epoch": 0.011742120985167394,
+      "grad_norm": 0.5651002526283264,
+      "learning_rate": 6.810000000000001e-06,
+      "loss": 2.4578,
+      "mean_token_accuracy": 0.4745705393142998,
+      "num_tokens": 98894643.0,
+      "step": 682
+    },
+    {
+      "entropy": 2.4425048828125,
+      "epoch": 0.011759338171362654,
+      "grad_norm": 0.5605524182319641,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.4011,
+      "mean_token_accuracy": 0.4801498386077583,
+      "num_tokens": 99057594.0,
+      "step": 683
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 0.011776555357557915,
+      "grad_norm": 0.594336748123169,
+      "learning_rate": 6.830000000000001e-06,
+      "loss": 2.4241,
+      "mean_token_accuracy": 0.4828556412830949,
+      "num_tokens": 99191074.0,
+      "step": 684
+    },
+    {
+      "entropy": 2.4183349609375,
+      "epoch": 0.011793772543753174,
+      "grad_norm": 0.5907600522041321,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.3646,
+      "mean_token_accuracy": 0.4930391958914697,
+      "num_tokens": 99328908.0,
+      "step": 685
+    },
+    {
+      "entropy": 2.5035400390625,
+      "epoch": 0.011810989729948434,
+      "grad_norm": 0.5865106582641602,
+      "learning_rate": 6.850000000000001e-06,
+      "loss": 2.481,
+      "mean_token_accuracy": 0.47163935378193855,
+      "num_tokens": 99456705.0,
+      "step": 686
+    },
+    {
+      "entropy": 2.4417724609375,
+      "epoch": 0.011828206916143695,
+      "grad_norm": 0.677649736404419,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.4363,
+      "mean_token_accuracy": 0.48826620541512966,
+      "num_tokens": 99591855.0,
+      "step": 687
+    },
+    {
+      "entropy": 2.4449462890625,
+      "epoch": 0.011845424102338955,
+      "grad_norm": 0.5924775004386902,
+      "learning_rate": 6.870000000000001e-06,
+      "loss": 2.3789,
+      "mean_token_accuracy": 0.48261339543387294,
+      "num_tokens": 99733016.0,
+      "step": 688
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.011862641288534214,
+      "grad_norm": 0.5597655177116394,
+      "learning_rate": 6.88e-06,
+      "loss": 2.4079,
+      "mean_token_accuracy": 0.4812637008726597,
+      "num_tokens": 99884912.0,
+      "step": 689
+    },
+    {
+      "entropy": 2.4306640625,
+      "epoch": 0.011879858474729475,
+      "grad_norm": 0.5902450680732727,
+      "learning_rate": 6.89e-06,
+      "loss": 2.3469,
+      "mean_token_accuracy": 0.4851334313862026,
+      "num_tokens": 100033610.0,
+      "step": 690
+    },
+    {
+      "entropy": 2.40673828125,
+      "epoch": 0.011897075660924735,
+      "grad_norm": 0.5616133213043213,
+      "learning_rate": 6.9e-06,
+      "loss": 2.3481,
+      "mean_token_accuracy": 0.4861144246533513,
+      "num_tokens": 100175721.0,
+      "step": 691
+    },
+    {
+      "entropy": 2.4476318359375,
+      "epoch": 0.011914292847119996,
+      "grad_norm": 0.5511784553527832,
+      "learning_rate": 6.91e-06,
+      "loss": 2.4008,
+      "mean_token_accuracy": 0.4786239666864276,
+      "num_tokens": 100317239.0,
+      "step": 692
+    },
+    {
+      "entropy": 2.4410400390625,
+      "epoch": 0.011931510033315256,
+      "grad_norm": 0.5769883394241333,
+      "learning_rate": 6.92e-06,
+      "loss": 2.3883,
+      "mean_token_accuracy": 0.48180802492424846,
+      "num_tokens": 100472196.0,
+      "step": 693
+    },
+    {
+      "entropy": 2.423828125,
+      "epoch": 0.011948727219510515,
+      "grad_norm": 0.5468559265136719,
+      "learning_rate": 6.93e-06,
+      "loss": 2.3532,
+      "mean_token_accuracy": 0.4820295791141689,
+      "num_tokens": 100630584.0,
+      "step": 694
+    },
+    {
+      "entropy": 2.484619140625,
+      "epoch": 0.011965944405705775,
+      "grad_norm": 0.5945432782173157,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.4283,
+      "mean_token_accuracy": 0.4779118075966835,
+      "num_tokens": 100786824.0,
+      "step": 695
+    },
+    {
+      "entropy": 2.4324951171875,
+      "epoch": 0.011983161591901036,
+      "grad_norm": 0.5941588878631592,
+      "learning_rate": 6.95e-06,
+      "loss": 2.3637,
+      "mean_token_accuracy": 0.4826899361796677,
+      "num_tokens": 100927601.0,
+      "step": 696
+    },
+    {
+      "entropy": 2.46240234375,
+      "epoch": 0.012000378778096297,
+      "grad_norm": 0.5858972668647766,
+      "learning_rate": 6.96e-06,
+      "loss": 2.4147,
+      "mean_token_accuracy": 0.47940469440072775,
+      "num_tokens": 101064043.0,
+      "step": 697
+    },
+    {
+      "entropy": 2.44091796875,
+      "epoch": 0.012017595964291555,
+      "grad_norm": 0.5900692939758301,
+      "learning_rate": 6.97e-06,
+      "loss": 2.3959,
+      "mean_token_accuracy": 0.48975384049117565,
+      "num_tokens": 101209114.0,
+      "step": 698
+    },
+    {
+      "entropy": 2.400146484375,
+      "epoch": 0.012034813150486816,
+      "grad_norm": 0.5515392422676086,
+      "learning_rate": 6.98e-06,
+      "loss": 2.3642,
+      "mean_token_accuracy": 0.4900199566036463,
+      "num_tokens": 101358773.0,
+      "step": 699
+    },
+    {
+      "entropy": 2.3948974609375,
+      "epoch": 0.012052030336682076,
+      "grad_norm": 0.5985309481620789,
+      "learning_rate": 6.99e-06,
+      "loss": 2.3383,
+      "mean_token_accuracy": 0.49027787847444415,
+      "num_tokens": 101507432.0,
+      "step": 700
+    },
+    {
+      "entropy": 2.45751953125,
+      "epoch": 0.012069247522877337,
+      "grad_norm": 0.5618374943733215,
+      "learning_rate": 7e-06,
+      "loss": 2.434,
+      "mean_token_accuracy": 0.47642564633861184,
+      "num_tokens": 101645197.0,
+      "step": 701
+    },
+    {
+      "entropy": 2.405517578125,
+      "epoch": 0.012086464709072596,
+      "grad_norm": 0.5900906920433044,
+      "learning_rate": 7.01e-06,
+      "loss": 2.3832,
+      "mean_token_accuracy": 0.48692094907164574,
+      "num_tokens": 101786084.0,
+      "step": 702
+    },
+    {
+      "entropy": 2.4290771484375,
+      "epoch": 0.012103681895267856,
+      "grad_norm": 0.5847712755203247,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.4093,
+      "mean_token_accuracy": 0.48114079609513283,
+      "num_tokens": 101929991.0,
+      "step": 703
+    },
+    {
+      "entropy": 2.4827880859375,
+      "epoch": 0.012120899081463117,
+      "grad_norm": 0.6014482378959656,
+      "learning_rate": 7.0300000000000005e-06,
+      "loss": 2.4566,
+      "mean_token_accuracy": 0.4787940843962133,
+      "num_tokens": 102069231.0,
+      "step": 704
+    },
+    {
+      "entropy": 2.5064697265625,
+      "epoch": 0.012138116267658377,
+      "grad_norm": 0.5504627823829651,
+      "learning_rate": 7.04e-06,
+      "loss": 2.4534,
+      "mean_token_accuracy": 0.46901731938123703,
+      "num_tokens": 102218938.0,
+      "step": 705
+    },
+    {
+      "entropy": 2.401123046875,
+      "epoch": 0.012155333453853636,
+      "grad_norm": 0.6932315826416016,
+      "learning_rate": 7.05e-06,
+      "loss": 2.3427,
+      "mean_token_accuracy": 0.4926096093840897,
+      "num_tokens": 102353532.0,
+      "step": 706
+    },
+    {
+      "entropy": 2.46923828125,
+      "epoch": 0.012172550640048897,
+      "grad_norm": 0.5833364725112915,
+      "learning_rate": 7.06e-06,
+      "loss": 2.4615,
+      "mean_token_accuracy": 0.47761310590431094,
+      "num_tokens": 102490402.0,
+      "step": 707
+    },
+    {
+      "entropy": 2.439697265625,
+      "epoch": 0.012189767826244157,
+      "grad_norm": 0.61143559217453,
+      "learning_rate": 7.07e-06,
+      "loss": 2.3579,
+      "mean_token_accuracy": 0.4858525595627725,
+      "num_tokens": 102627917.0,
+      "step": 708
+    },
+    {
+      "entropy": 2.49169921875,
+      "epoch": 0.012206985012439418,
+      "grad_norm": 0.601938009262085,
+      "learning_rate": 7.08e-06,
+      "loss": 2.4313,
+      "mean_token_accuracy": 0.4729892536997795,
+      "num_tokens": 102759385.0,
+      "step": 709
+    },
+    {
+      "entropy": 2.4722900390625,
+      "epoch": 0.012224202198634676,
+      "grad_norm": 0.5767009854316711,
+      "learning_rate": 7.09e-06,
+      "loss": 2.4482,
+      "mean_token_accuracy": 0.4772743955254555,
+      "num_tokens": 102915430.0,
+      "step": 710
+    },
+    {
+      "entropy": 2.46142578125,
+      "epoch": 0.012241419384829937,
+      "grad_norm": 0.546136200428009,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.3961,
+      "mean_token_accuracy": 0.4819467253983021,
+      "num_tokens": 103066691.0,
+      "step": 711
+    },
+    {
+      "entropy": 2.48583984375,
+      "epoch": 0.012258636571025197,
+      "grad_norm": 0.5756231546401978,
+      "learning_rate": 7.1100000000000005e-06,
+      "loss": 2.4676,
+      "mean_token_accuracy": 0.4714622185565531,
+      "num_tokens": 103217176.0,
+      "step": 712
+    },
+    {
+      "entropy": 2.4246826171875,
+      "epoch": 0.012275853757220458,
+      "grad_norm": 0.5694965720176697,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.3603,
+      "mean_token_accuracy": 0.4863223168067634,
+      "num_tokens": 103368786.0,
+      "step": 713
+    },
+    {
+      "entropy": 2.447265625,
+      "epoch": 0.012293070943415717,
+      "grad_norm": 0.5844082236289978,
+      "learning_rate": 7.13e-06,
+      "loss": 2.4418,
+      "mean_token_accuracy": 0.482551914639771,
+      "num_tokens": 103515573.0,
+      "step": 714
+    },
+    {
+      "entropy": 2.509765625,
+      "epoch": 0.012310288129610977,
+      "grad_norm": 0.5859200358390808,
+      "learning_rate": 7.14e-06,
+      "loss": 2.4894,
+      "mean_token_accuracy": 0.4717716183513403,
+      "num_tokens": 103649277.0,
+      "step": 715
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.012327505315806238,
+      "grad_norm": 0.5378598570823669,
+      "learning_rate": 7.15e-06,
+      "loss": 2.4289,
+      "mean_token_accuracy": 0.47941537760198116,
+      "num_tokens": 103802169.0,
+      "step": 716
+    },
+    {
+      "entropy": 2.4493408203125,
+      "epoch": 0.012344722502001498,
+      "grad_norm": 0.6121527552604675,
+      "learning_rate": 7.16e-06,
+      "loss": 2.3747,
+      "mean_token_accuracy": 0.4810113995335996,
+      "num_tokens": 103939625.0,
+      "step": 717
+    },
+    {
+      "entropy": 2.4144287109375,
+      "epoch": 0.012361939688196759,
+      "grad_norm": 0.5776681900024414,
+      "learning_rate": 7.17e-06,
+      "loss": 2.3531,
+      "mean_token_accuracy": 0.4864069065079093,
+      "num_tokens": 104080617.0,
+      "step": 718
+    },
+    {
+      "entropy": 2.3646240234375,
+      "epoch": 0.012379156874392018,
+      "grad_norm": 0.5905261635780334,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.3327,
+      "mean_token_accuracy": 0.49934633634984493,
+      "num_tokens": 104220572.0,
+      "step": 719
+    },
+    {
+      "entropy": 2.4034423828125,
+      "epoch": 0.012396374060587278,
+      "grad_norm": 0.5707198977470398,
+      "learning_rate": 7.190000000000001e-06,
+      "loss": 2.3433,
+      "mean_token_accuracy": 0.49225129559636116,
+      "num_tokens": 104361519.0,
+      "step": 720
+    },
+    {
+      "entropy": 2.4696044921875,
+      "epoch": 0.012413591246782539,
+      "grad_norm": 0.5869008302688599,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.4365,
+      "mean_token_accuracy": 0.47933553624898195,
+      "num_tokens": 104506297.0,
+      "step": 721
+    },
+    {
+      "entropy": 2.4625244140625,
+      "epoch": 0.0124308084329778,
+      "grad_norm": 0.5947871208190918,
+      "learning_rate": 7.2100000000000004e-06,
+      "loss": 2.4067,
+      "mean_token_accuracy": 0.48690010188147426,
+      "num_tokens": 104647015.0,
+      "step": 722
+    },
+    {
+      "entropy": 2.450439453125,
+      "epoch": 0.012448025619173058,
+      "grad_norm": 0.6154801845550537,
+      "learning_rate": 7.22e-06,
+      "loss": 2.4108,
+      "mean_token_accuracy": 0.47685753647238016,
+      "num_tokens": 104802411.0,
+      "step": 723
+    },
+    {
+      "entropy": 2.419921875,
+      "epoch": 0.012465242805368319,
+      "grad_norm": 0.6330761313438416,
+      "learning_rate": 7.23e-06,
+      "loss": 2.3759,
+      "mean_token_accuracy": 0.4840613235719502,
+      "num_tokens": 104951305.0,
+      "step": 724
+    },
+    {
+      "entropy": 2.43359375,
+      "epoch": 0.012482459991563579,
+      "grad_norm": 0.6040731072425842,
+      "learning_rate": 7.24e-06,
+      "loss": 2.3609,
+      "mean_token_accuracy": 0.4850413934327662,
+      "num_tokens": 105091990.0,
+      "step": 725
+    },
+    {
+      "entropy": 2.44482421875,
+      "epoch": 0.01249967717775884,
+      "grad_norm": 0.6250449419021606,
+      "learning_rate": 7.25e-06,
+      "loss": 2.4137,
+      "mean_token_accuracy": 0.4755057515576482,
+      "num_tokens": 105212920.0,
+      "step": 726
+    },
+    {
+      "entropy": 2.4415283203125,
+      "epoch": 0.012516894363954098,
+      "grad_norm": 0.6397738456726074,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.4425,
+      "mean_token_accuracy": 0.4787881104275584,
+      "num_tokens": 105329271.0,
+      "step": 727
+    },
+    {
+      "entropy": 2.3997802734375,
+      "epoch": 0.012534111550149359,
+      "grad_norm": 0.5615900158882141,
+      "learning_rate": 7.270000000000001e-06,
+      "loss": 2.4034,
+      "mean_token_accuracy": 0.4808120485395193,
+      "num_tokens": 105487622.0,
+      "step": 728
+    },
+    {
+      "entropy": 2.464111328125,
+      "epoch": 0.01255132873634462,
+      "grad_norm": 0.5766611695289612,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.4341,
+      "mean_token_accuracy": 0.47967397794127464,
+      "num_tokens": 105633781.0,
+      "step": 729
+    },
+    {
+      "entropy": 2.482421875,
+      "epoch": 0.01256854592253988,
+      "grad_norm": 0.557733416557312,
+      "learning_rate": 7.2900000000000005e-06,
+      "loss": 2.4198,
+      "mean_token_accuracy": 0.4778105691075325,
+      "num_tokens": 105783888.0,
+      "step": 730
+    },
+    {
+      "entropy": 2.3619384765625,
+      "epoch": 0.012585763108735139,
+      "grad_norm": 0.5741136074066162,
+      "learning_rate": 7.3e-06,
+      "loss": 2.3078,
+      "mean_token_accuracy": 0.49396718852221966,
+      "num_tokens": 105927244.0,
+      "step": 731
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.0126029802949304,
+      "grad_norm": 0.5635918974876404,
+      "learning_rate": 7.31e-06,
+      "loss": 2.3871,
+      "mean_token_accuracy": 0.48639540281146765,
+      "num_tokens": 106070617.0,
+      "step": 732
+    },
+    {
+      "entropy": 2.4696044921875,
+      "epoch": 0.01262019748112566,
+      "grad_norm": 0.578965425491333,
+      "learning_rate": 7.32e-06,
+      "loss": 2.3938,
+      "mean_token_accuracy": 0.4847126523964107,
+      "num_tokens": 106230425.0,
+      "step": 733
+    },
+    {
+      "entropy": 2.458251953125,
+      "epoch": 0.01263741466732092,
+      "grad_norm": 0.9026923775672913,
+      "learning_rate": 7.33e-06,
+      "loss": 2.4071,
+      "mean_token_accuracy": 0.4807370454072952,
+      "num_tokens": 106375429.0,
+      "step": 734
+    },
+    {
+      "entropy": 2.4134521484375,
+      "epoch": 0.012654631853516179,
+      "grad_norm": 0.6249892711639404,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.3442,
+      "mean_token_accuracy": 0.48796508787199855,
+      "num_tokens": 106523980.0,
+      "step": 735
+    },
+    {
+      "entropy": 2.403076171875,
+      "epoch": 0.01267184903971144,
+      "grad_norm": 0.5887693166732788,
+      "learning_rate": 7.350000000000001e-06,
+      "loss": 2.3872,
+      "mean_token_accuracy": 0.48691826686263084,
+      "num_tokens": 106669149.0,
+      "step": 736
+    },
+    {
+      "entropy": 2.4166259765625,
+      "epoch": 0.0126890662259067,
+      "grad_norm": 0.617521345615387,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.3293,
+      "mean_token_accuracy": 0.48858933337032795,
+      "num_tokens": 106798139.0,
+      "step": 737
+    },
+    {
+      "entropy": 2.44921875,
+      "epoch": 0.01270628341210196,
+      "grad_norm": 0.5710563659667969,
+      "learning_rate": 7.370000000000001e-06,
+      "loss": 2.4245,
+      "mean_token_accuracy": 0.47817583242431283,
+      "num_tokens": 106937567.0,
+      "step": 738
+    },
+    {
+      "entropy": 2.4063720703125,
+      "epoch": 0.01272350059829722,
+      "grad_norm": 1.2993676662445068,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.3553,
+      "mean_token_accuracy": 0.48716708505526185,
+      "num_tokens": 107069215.0,
+      "step": 739
+    },
+    {
+      "entropy": 2.417236328125,
+      "epoch": 0.01274071778449248,
+      "grad_norm": 0.5727201104164124,
+      "learning_rate": 7.39e-06,
+      "loss": 2.3806,
+      "mean_token_accuracy": 0.4816734539344907,
+      "num_tokens": 107216177.0,
+      "step": 740
+    },
+    {
+      "entropy": 2.470947265625,
+      "epoch": 0.01275793497068774,
+      "grad_norm": 0.568335771560669,
+      "learning_rate": 7.4e-06,
+      "loss": 2.4227,
+      "mean_token_accuracy": 0.4798375847749412,
+      "num_tokens": 107367907.0,
+      "step": 741
+    },
+    {
+      "entropy": 2.4110107421875,
+      "epoch": 0.012775152156883001,
+      "grad_norm": 0.6011884808540344,
+      "learning_rate": 7.41e-06,
+      "loss": 2.3219,
+      "mean_token_accuracy": 0.49019240494817495,
+      "num_tokens": 107504131.0,
+      "step": 742
+    },
+    {
+      "entropy": 2.41845703125,
+      "epoch": 0.01279236934307826,
+      "grad_norm": 0.5796740055084229,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.3559,
+      "mean_token_accuracy": 0.48748622741550207,
+      "num_tokens": 107645959.0,
+      "step": 743
+    },
+    {
+      "entropy": 2.4940185546875,
+      "epoch": 0.01280958652927352,
+      "grad_norm": 0.5444216132164001,
+      "learning_rate": 7.430000000000001e-06,
+      "loss": 2.4531,
+      "mean_token_accuracy": 0.4708722811192274,
+      "num_tokens": 107795485.0,
+      "step": 744
+    },
+    {
+      "entropy": 2.413818359375,
+      "epoch": 0.012826803715468781,
+      "grad_norm": 0.5786703824996948,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.3548,
+      "mean_token_accuracy": 0.49165352433919907,
+      "num_tokens": 107937289.0,
+      "step": 745
+    },
+    {
+      "entropy": 2.411865234375,
+      "epoch": 0.012844020901664041,
+      "grad_norm": 0.6046903729438782,
+      "learning_rate": 7.450000000000001e-06,
+      "loss": 2.3753,
+      "mean_token_accuracy": 0.4823318342678249,
+      "num_tokens": 108093121.0,
+      "step": 746
+    },
+    {
+      "entropy": 2.4716796875,
+      "epoch": 0.012861238087859302,
+      "grad_norm": 0.536321222782135,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.4663,
+      "mean_token_accuracy": 0.4771149712614715,
+      "num_tokens": 108257267.0,
+      "step": 747
+    },
+    {
+      "entropy": 2.4007568359375,
+      "epoch": 0.01287845527405456,
+      "grad_norm": 0.6158970594406128,
+      "learning_rate": 7.4700000000000005e-06,
+      "loss": 2.3579,
+      "mean_token_accuracy": 0.4832296408712864,
+      "num_tokens": 108396635.0,
+      "step": 748
+    },
+    {
+      "entropy": 2.4298095703125,
+      "epoch": 0.012895672460249821,
+      "grad_norm": 0.5570418834686279,
+      "learning_rate": 7.48e-06,
+      "loss": 2.4086,
+      "mean_token_accuracy": 0.48002893943339586,
+      "num_tokens": 108544662.0,
+      "step": 749
+    },
+    {
+      "entropy": 2.54150390625,
+      "epoch": 0.012912889646445082,
+      "grad_norm": 0.6132979393005371,
+      "learning_rate": 7.49e-06,
+      "loss": 2.4981,
+      "mean_token_accuracy": 0.46807813877239823,
+      "num_tokens": 108709034.0,
+      "step": 750
+    },
+    {
+      "entropy": 2.438232421875,
+      "epoch": 0.012930106832640342,
+      "grad_norm": 0.5490818023681641,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.3753,
+      "mean_token_accuracy": 0.48799073603004217,
+      "num_tokens": 108859610.0,
+      "step": 751
+    },
+    {
+      "entropy": 2.4285888671875,
+      "epoch": 0.012947324018835601,
+      "grad_norm": 0.5972912311553955,
+      "learning_rate": 7.510000000000001e-06,
+      "loss": 2.4225,
+      "mean_token_accuracy": 0.48296895902603865,
+      "num_tokens": 109009475.0,
+      "step": 752
+    },
+    {
+      "entropy": 2.4571533203125,
+      "epoch": 0.012964541205030862,
+      "grad_norm": 0.5518878102302551,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.3751,
+      "mean_token_accuracy": 0.4834339157678187,
+      "num_tokens": 109158002.0,
+      "step": 753
+    },
+    {
+      "entropy": 2.41162109375,
+      "epoch": 0.012981758391226122,
+      "grad_norm": 0.602249801158905,
+      "learning_rate": 7.530000000000001e-06,
+      "loss": 2.3587,
+      "mean_token_accuracy": 0.4893317800015211,
+      "num_tokens": 109293690.0,
+      "step": 754
+    },
+    {
+      "entropy": 2.4407958984375,
+      "epoch": 0.012998975577421383,
+      "grad_norm": 0.5610491633415222,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.4023,
+      "mean_token_accuracy": 0.4784908280707896,
+      "num_tokens": 109439590.0,
+      "step": 755
+    },
+    {
+      "entropy": 2.4281005859375,
+      "epoch": 0.013016192763616641,
+      "grad_norm": 0.5914566516876221,
+      "learning_rate": 7.5500000000000006e-06,
+      "loss": 2.4052,
+      "mean_token_accuracy": 0.4888386274687946,
+      "num_tokens": 109582432.0,
+      "step": 756
+    },
+    {
+      "entropy": 2.504638671875,
+      "epoch": 0.013033409949811902,
+      "grad_norm": 0.5595380067825317,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.4702,
+      "mean_token_accuracy": 0.4686399414204061,
+      "num_tokens": 109735418.0,
+      "step": 757
+    },
+    {
+      "entropy": 2.4144287109375,
+      "epoch": 0.013050627136007163,
+      "grad_norm": 0.5793854594230652,
+      "learning_rate": 7.57e-06,
+      "loss": 2.3714,
+      "mean_token_accuracy": 0.4886330468580127,
+      "num_tokens": 109891701.0,
+      "step": 758
+    },
+    {
+      "entropy": 2.4320068359375,
+      "epoch": 0.013067844322202423,
+      "grad_norm": 0.6832471489906311,
+      "learning_rate": 7.58e-06,
+      "loss": 2.3694,
+      "mean_token_accuracy": 0.48566275043413043,
+      "num_tokens": 110039268.0,
+      "step": 759
+    },
+    {
+      "entropy": 2.4044189453125,
+      "epoch": 0.013085061508397682,
+      "grad_norm": 0.6120168566703796,
+      "learning_rate": 7.590000000000001e-06,
+      "loss": 2.3659,
+      "mean_token_accuracy": 0.4929888774640858,
+      "num_tokens": 110182050.0,
+      "step": 760
+    },
+    {
+      "entropy": 2.491943359375,
+      "epoch": 0.013102278694592942,
+      "grad_norm": 0.6201086044311523,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.4232,
+      "mean_token_accuracy": 0.4787399894557893,
+      "num_tokens": 110303657.0,
+      "step": 761
+    },
+    {
+      "entropy": 2.499755859375,
+      "epoch": 0.013119495880788203,
+      "grad_norm": 0.5904567837715149,
+      "learning_rate": 7.610000000000001e-06,
+      "loss": 2.4735,
+      "mean_token_accuracy": 0.47595866583287716,
+      "num_tokens": 110445376.0,
+      "step": 762
+    },
+    {
+      "entropy": 2.4344482421875,
+      "epoch": 0.013136713066983463,
+      "grad_norm": 0.7313216924667358,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.3898,
+      "mean_token_accuracy": 0.4856723416596651,
+      "num_tokens": 110599091.0,
+      "step": 763
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.013153930253178722,
+      "grad_norm": 0.582177460193634,
+      "learning_rate": 7.630000000000001e-06,
+      "loss": 2.3465,
+      "mean_token_accuracy": 0.49040036741644144,
+      "num_tokens": 110767359.0,
+      "step": 764
+    },
+    {
+      "entropy": 2.4423828125,
+      "epoch": 0.013171147439373983,
+      "grad_norm": 0.5971955060958862,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.4243,
+      "mean_token_accuracy": 0.47924549924209714,
+      "num_tokens": 110904749.0,
+      "step": 765
+    },
+    {
+      "entropy": 2.370361328125,
+      "epoch": 0.013188364625569243,
+      "grad_norm": 0.5825552940368652,
+      "learning_rate": 7.650000000000001e-06,
+      "loss": 2.337,
+      "mean_token_accuracy": 0.48746788455173373,
+      "num_tokens": 111041416.0,
+      "step": 766
+    },
+    {
+      "entropy": 2.485107421875,
+      "epoch": 0.013205581811764504,
+      "grad_norm": 0.5713107585906982,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.4491,
+      "mean_token_accuracy": 0.4754057708196342,
+      "num_tokens": 111179069.0,
+      "step": 767
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.013222798997959763,
+      "grad_norm": 0.5602453947067261,
+      "learning_rate": 7.670000000000001e-06,
+      "loss": 2.3753,
+      "mean_token_accuracy": 0.49099403340369463,
+      "num_tokens": 111333973.0,
+      "step": 768
+    },
+    {
+      "entropy": 2.40283203125,
+      "epoch": 0.013240016184155023,
+      "grad_norm": 0.5362405180931091,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.3705,
+      "mean_token_accuracy": 0.48479648493230343,
+      "num_tokens": 111483914.0,
+      "step": 769
+    },
+    {
+      "entropy": 2.4000244140625,
+      "epoch": 0.013257233370350284,
+      "grad_norm": 0.5498298406600952,
+      "learning_rate": 7.690000000000001e-06,
+      "loss": 2.332,
+      "mean_token_accuracy": 0.49140041740611196,
+      "num_tokens": 111637688.0,
+      "step": 770
+    },
+    {
+      "entropy": 2.3616943359375,
+      "epoch": 0.013274450556545544,
+      "grad_norm": 0.5781952142715454,
+      "learning_rate": 7.7e-06,
+      "loss": 2.3117,
+      "mean_token_accuracy": 0.49226083187386394,
+      "num_tokens": 111777176.0,
+      "step": 771
+    },
+    {
+      "entropy": 2.45654296875,
+      "epoch": 0.013291667742740805,
+      "grad_norm": 0.5963894724845886,
+      "learning_rate": 7.71e-06,
+      "loss": 2.4077,
+      "mean_token_accuracy": 0.4808903872035444,
+      "num_tokens": 111907633.0,
+      "step": 772
+    },
+    {
+      "entropy": 2.47265625,
+      "epoch": 0.013308884928936063,
+      "grad_norm": 0.5909532904624939,
+      "learning_rate": 7.72e-06,
+      "loss": 2.4344,
+      "mean_token_accuracy": 0.48116163862869143,
+      "num_tokens": 112047691.0,
+      "step": 773
+    },
+    {
+      "entropy": 2.4112548828125,
+      "epoch": 0.013326102115131324,
+      "grad_norm": 0.6000514626502991,
+      "learning_rate": 7.73e-06,
+      "loss": 2.3594,
+      "mean_token_accuracy": 0.4857400543987751,
+      "num_tokens": 112185378.0,
+      "step": 774
+    },
+    {
+      "entropy": 2.41943359375,
+      "epoch": 0.013343319301326585,
+      "grad_norm": 0.586081326007843,
+      "learning_rate": 7.74e-06,
+      "loss": 2.3978,
+      "mean_token_accuracy": 0.4904695344157517,
+      "num_tokens": 112341351.0,
+      "step": 775
+    },
+    {
+      "entropy": 2.4359130859375,
+      "epoch": 0.013360536487521845,
+      "grad_norm": 0.5683102607727051,
+      "learning_rate": 7.75e-06,
+      "loss": 2.3903,
+      "mean_token_accuracy": 0.4875342110171914,
+      "num_tokens": 112487164.0,
+      "step": 776
+    },
+    {
+      "entropy": 2.4697265625,
+      "epoch": 0.013377753673717104,
+      "grad_norm": 0.5665966272354126,
+      "learning_rate": 7.76e-06,
+      "loss": 2.4298,
+      "mean_token_accuracy": 0.47403282299637794,
+      "num_tokens": 112629741.0,
+      "step": 777
+    },
+    {
+      "entropy": 2.45703125,
+      "epoch": 0.013394970859912364,
+      "grad_norm": 0.5655020475387573,
+      "learning_rate": 7.77e-06,
+      "loss": 2.402,
+      "mean_token_accuracy": 0.48069310747087,
+      "num_tokens": 112767619.0,
+      "step": 778
+    },
+    {
+      "entropy": 2.48095703125,
+      "epoch": 0.013412188046107625,
+      "grad_norm": 0.5934699177742004,
+      "learning_rate": 7.78e-06,
+      "loss": 2.4576,
+      "mean_token_accuracy": 0.4850818943232298,
+      "num_tokens": 112913870.0,
+      "step": 779
+    },
+    {
+      "entropy": 2.4310302734375,
+      "epoch": 0.013429405232302885,
+      "grad_norm": 0.7897031307220459,
+      "learning_rate": 7.790000000000002e-06,
+      "loss": 2.3135,
+      "mean_token_accuracy": 0.4851987957954407,
+      "num_tokens": 113047948.0,
+      "step": 780
+    },
+    {
+      "entropy": 2.3919677734375,
+      "epoch": 0.013446622418498144,
+      "grad_norm": 0.5672735571861267,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.3579,
+      "mean_token_accuracy": 0.48478852584958076,
+      "num_tokens": 113210948.0,
+      "step": 781
+    },
+    {
+      "entropy": 2.4381103515625,
+      "epoch": 0.013463839604693405,
+      "grad_norm": 0.6065455079078674,
+      "learning_rate": 7.810000000000001e-06,
+      "loss": 2.3854,
+      "mean_token_accuracy": 0.4848202792927623,
+      "num_tokens": 113361592.0,
+      "step": 782
+    },
+    {
+      "entropy": 2.4210205078125,
+      "epoch": 0.013481056790888665,
+      "grad_norm": 0.5307328701019287,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.3612,
+      "mean_token_accuracy": 0.4887157790362835,
+      "num_tokens": 113524797.0,
+      "step": 783
+    },
+    {
+      "entropy": 2.4393310546875,
+      "epoch": 0.013498273977083926,
+      "grad_norm": 0.6069521307945251,
+      "learning_rate": 7.830000000000001e-06,
+      "loss": 2.4034,
+      "mean_token_accuracy": 0.4795961854979396,
+      "num_tokens": 113659620.0,
+      "step": 784
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.013515491163279185,
+      "grad_norm": 0.5770717263221741,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.4369,
+      "mean_token_accuracy": 0.48013802990317345,
+      "num_tokens": 113803890.0,
+      "step": 785
+    },
+    {
+      "entropy": 2.4228515625,
+      "epoch": 0.013532708349474445,
+      "grad_norm": 0.6027282476425171,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 2.406,
+      "mean_token_accuracy": 0.48077098093926907,
+      "num_tokens": 113948731.0,
+      "step": 786
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.013549925535669706,
+      "grad_norm": 0.6242732405662537,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.3891,
+      "mean_token_accuracy": 0.4767256425693631,
+      "num_tokens": 114095401.0,
+      "step": 787
+    },
+    {
+      "entropy": 2.507568359375,
+      "epoch": 0.013567142721864966,
+      "grad_norm": 0.5736910104751587,
+      "learning_rate": 7.870000000000001e-06,
+      "loss": 2.4509,
+      "mean_token_accuracy": 0.4732041200622916,
+      "num_tokens": 114245869.0,
+      "step": 788
+    },
+    {
+      "entropy": 2.485595703125,
+      "epoch": 0.013584359908060225,
+      "grad_norm": 0.5609222054481506,
+      "learning_rate": 7.88e-06,
+      "loss": 2.4312,
+      "mean_token_accuracy": 0.4742095875553787,
+      "num_tokens": 114388858.0,
+      "step": 789
+    },
+    {
+      "entropy": 2.56689453125,
+      "epoch": 0.013601577094255485,
+      "grad_norm": 0.5870925784111023,
+      "learning_rate": 7.89e-06,
+      "loss": 2.5389,
+      "mean_token_accuracy": 0.46493083937093616,
+      "num_tokens": 114523322.0,
+      "step": 790
+    },
+    {
+      "entropy": 2.49267578125,
+      "epoch": 0.013618794280450746,
+      "grad_norm": 0.5677596926689148,
+      "learning_rate": 7.9e-06,
+      "loss": 2.4385,
+      "mean_token_accuracy": 0.4769080653786659,
+      "num_tokens": 114663505.0,
+      "step": 791
+    },
+    {
+      "entropy": 2.485595703125,
+      "epoch": 0.013636011466646007,
+      "grad_norm": 0.5597982406616211,
+      "learning_rate": 7.91e-06,
+      "loss": 2.4263,
+      "mean_token_accuracy": 0.4761881032027304,
+      "num_tokens": 114813922.0,
+      "step": 792
+    },
+    {
+      "entropy": 2.416259765625,
+      "epoch": 0.013653228652841265,
+      "grad_norm": 0.5396745204925537,
+      "learning_rate": 7.92e-06,
+      "loss": 2.3336,
+      "mean_token_accuracy": 0.49072846584022045,
+      "num_tokens": 114962105.0,
+      "step": 793
+    },
+    {
+      "entropy": 2.3951416015625,
+      "epoch": 0.013670445839036526,
+      "grad_norm": 0.6175518035888672,
+      "learning_rate": 7.93e-06,
+      "loss": 2.339,
+      "mean_token_accuracy": 0.4950829269364476,
+      "num_tokens": 115103571.0,
+      "step": 794
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.013687663025231786,
+      "grad_norm": 0.5724807977676392,
+      "learning_rate": 7.94e-06,
+      "loss": 2.3744,
+      "mean_token_accuracy": 0.47975975926965475,
+      "num_tokens": 115243719.0,
+      "step": 795
+    },
+    {
+      "entropy": 2.5086669921875,
+      "epoch": 0.013704880211427047,
+      "grad_norm": 0.5798308849334717,
+      "learning_rate": 7.950000000000002e-06,
+      "loss": 2.4594,
+      "mean_token_accuracy": 0.4762448235414922,
+      "num_tokens": 115383378.0,
+      "step": 796
+    },
+    {
+      "entropy": 2.4212646484375,
+      "epoch": 0.013722097397622307,
+      "grad_norm": 0.5839712023735046,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.3884,
+      "mean_token_accuracy": 0.48001448903232813,
+      "num_tokens": 115527648.0,
+      "step": 797
+    },
+    {
+      "entropy": 2.388671875,
+      "epoch": 0.013739314583817566,
+      "grad_norm": 0.630893349647522,
+      "learning_rate": 7.970000000000002e-06,
+      "loss": 2.3391,
+      "mean_token_accuracy": 0.48810708662495017,
+      "num_tokens": 115673489.0,
+      "step": 798
+    },
+    {
+      "entropy": 2.475341796875,
+      "epoch": 0.013756531770012827,
+      "grad_norm": 0.590587854385376,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.4425,
+      "mean_token_accuracy": 0.4739443711005151,
+      "num_tokens": 115814268.0,
+      "step": 799
+    },
+    {
+      "entropy": 2.380615234375,
+      "epoch": 0.013773748956208087,
+      "grad_norm": 0.566112220287323,
+      "learning_rate": 7.990000000000001e-06,
+      "loss": 2.3345,
+      "mean_token_accuracy": 0.49479513335973024,
+      "num_tokens": 115962087.0,
+      "step": 800
+    },
+    {
+      "entropy": 2.53515625,
+      "epoch": 0.013790966142403348,
+      "grad_norm": 0.5729132890701294,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.478,
+      "mean_token_accuracy": 0.46622643573209643,
+      "num_tokens": 116096894.0,
+      "step": 801
+    },
+    {
+      "entropy": 2.43701171875,
+      "epoch": 0.013808183328598607,
+      "grad_norm": 0.5780159831047058,
+      "learning_rate": 8.010000000000001e-06,
+      "loss": 2.4245,
+      "mean_token_accuracy": 0.48138847574591637,
+      "num_tokens": 116248222.0,
+      "step": 802
+    },
+    {
+      "entropy": 2.44482421875,
+      "epoch": 0.013825400514793867,
+      "grad_norm": 0.5970706343650818,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.4229,
+      "mean_token_accuracy": 0.4804395758546889,
+      "num_tokens": 116405644.0,
+      "step": 803
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.013842617700989128,
+      "grad_norm": 0.5705693364143372,
+      "learning_rate": 8.030000000000001e-06,
+      "loss": 2.3441,
+      "mean_token_accuracy": 0.48369065998122096,
+      "num_tokens": 116548738.0,
+      "step": 804
+    },
+    {
+      "entropy": 2.4342041015625,
+      "epoch": 0.013859834887184388,
+      "grad_norm": 0.5916334390640259,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.3664,
+      "mean_token_accuracy": 0.4856993416324258,
+      "num_tokens": 116691888.0,
+      "step": 805
+    },
+    {
+      "entropy": 2.4691162109375,
+      "epoch": 0.013877052073379647,
+      "grad_norm": 0.579868495464325,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 2.3724,
+      "mean_token_accuracy": 0.4843080313876271,
+      "num_tokens": 116830242.0,
+      "step": 806
+    },
+    {
+      "entropy": 2.40966796875,
+      "epoch": 0.013894269259574907,
+      "grad_norm": 0.5866817831993103,
+      "learning_rate": 8.06e-06,
+      "loss": 2.3801,
+      "mean_token_accuracy": 0.48936720937490463,
+      "num_tokens": 116965900.0,
+      "step": 807
+    },
+    {
+      "entropy": 2.4056396484375,
+      "epoch": 0.013911486445770168,
+      "grad_norm": 0.5471423864364624,
+      "learning_rate": 8.07e-06,
+      "loss": 2.3545,
+      "mean_token_accuracy": 0.4835878200829029,
+      "num_tokens": 117118489.0,
+      "step": 808
+    },
+    {
+      "entropy": 2.42529296875,
+      "epoch": 0.013928703631965429,
+      "grad_norm": 0.577364981174469,
+      "learning_rate": 8.08e-06,
+      "loss": 2.3581,
+      "mean_token_accuracy": 0.48418712290003896,
+      "num_tokens": 117264043.0,
+      "step": 809
+    },
+    {
+      "entropy": 2.43310546875,
+      "epoch": 0.013945920818160687,
+      "grad_norm": 0.5444287657737732,
+      "learning_rate": 8.09e-06,
+      "loss": 2.4178,
+      "mean_token_accuracy": 0.476511531509459,
+      "num_tokens": 117419082.0,
+      "step": 810
+    },
+    {
+      "entropy": 2.435302734375,
+      "epoch": 0.013963138004355948,
+      "grad_norm": 0.6165118217468262,
+      "learning_rate": 8.1e-06,
+      "loss": 2.4043,
+      "mean_token_accuracy": 0.48205558583140373,
+      "num_tokens": 117552496.0,
+      "step": 811
+    },
+    {
+      "entropy": 2.462158203125,
+      "epoch": 0.013980355190551208,
+      "grad_norm": 0.5636385679244995,
+      "learning_rate": 8.110000000000002e-06,
+      "loss": 2.4252,
+      "mean_token_accuracy": 0.4745056303218007,
+      "num_tokens": 117702578.0,
+      "step": 812
+    },
+    {
+      "entropy": 2.542724609375,
+      "epoch": 0.013997572376746469,
+      "grad_norm": 0.5832167863845825,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.5385,
+      "mean_token_accuracy": 0.4675266365520656,
+      "num_tokens": 117850701.0,
+      "step": 813
+    },
+    {
+      "entropy": 2.489990234375,
+      "epoch": 0.014014789562941728,
+      "grad_norm": 0.5839970111846924,
+      "learning_rate": 8.13e-06,
+      "loss": 2.4495,
+      "mean_token_accuracy": 0.47388132382184267,
+      "num_tokens": 118001878.0,
+      "step": 814
+    },
+    {
+      "entropy": 2.429931640625,
+      "epoch": 0.014032006749136988,
+      "grad_norm": 0.6004050970077515,
+      "learning_rate": 8.14e-06,
+      "loss": 2.3689,
+      "mean_token_accuracy": 0.48473711824044585,
+      "num_tokens": 118146781.0,
+      "step": 815
+    },
+    {
+      "entropy": 2.517333984375,
+      "epoch": 0.014049223935332249,
+      "grad_norm": 0.5721548795700073,
+      "learning_rate": 8.15e-06,
+      "loss": 2.5005,
+      "mean_token_accuracy": 0.47145770117640495,
+      "num_tokens": 118281715.0,
+      "step": 816
+    },
+    {
+      "entropy": 2.41259765625,
+      "epoch": 0.01406644112152751,
+      "grad_norm": 0.6043224930763245,
+      "learning_rate": 8.16e-06,
+      "loss": 2.3785,
+      "mean_token_accuracy": 0.48464843491092324,
+      "num_tokens": 118420852.0,
+      "step": 817
+    },
+    {
+      "entropy": 2.46435546875,
+      "epoch": 0.014083658307722768,
+      "grad_norm": 0.5578818321228027,
+      "learning_rate": 8.17e-06,
+      "loss": 2.4389,
+      "mean_token_accuracy": 0.4739728611893952,
+      "num_tokens": 118561047.0,
+      "step": 818
+    },
+    {
+      "entropy": 2.46435546875,
+      "epoch": 0.014100875493918029,
+      "grad_norm": 0.5339409112930298,
+      "learning_rate": 8.18e-06,
+      "loss": 2.4303,
+      "mean_token_accuracy": 0.478867762722075,
+      "num_tokens": 118724226.0,
+      "step": 819
+    },
+    {
+      "entropy": 2.4332275390625,
+      "epoch": 0.014118092680113289,
+      "grad_norm": 0.5819604396820068,
+      "learning_rate": 8.19e-06,
+      "loss": 2.3808,
+      "mean_token_accuracy": 0.4880368346348405,
+      "num_tokens": 118869937.0,
+      "step": 820
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.01413530986630855,
+      "grad_norm": 0.571780800819397,
+      "learning_rate": 8.2e-06,
+      "loss": 2.4314,
+      "mean_token_accuracy": 0.4772975808009505,
+      "num_tokens": 119022262.0,
+      "step": 821
+    },
+    {
+      "entropy": 2.4794921875,
+      "epoch": 0.01415252705250381,
+      "grad_norm": 0.5831077694892883,
+      "learning_rate": 8.210000000000001e-06,
+      "loss": 2.4701,
+      "mean_token_accuracy": 0.47361723706126213,
+      "num_tokens": 119165669.0,
+      "step": 822
+    },
+    {
+      "entropy": 2.4208984375,
+      "epoch": 0.014169744238699069,
+      "grad_norm": 0.5717747211456299,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.3679,
+      "mean_token_accuracy": 0.48527460684999824,
+      "num_tokens": 119301111.0,
+      "step": 823
+    },
+    {
+      "entropy": 2.45947265625,
+      "epoch": 0.01418696142489433,
+      "grad_norm": 0.6002344489097595,
+      "learning_rate": 8.23e-06,
+      "loss": 2.3749,
+      "mean_token_accuracy": 0.48156877839937806,
+      "num_tokens": 119443072.0,
+      "step": 824
+    },
+    {
+      "entropy": 2.4619140625,
+      "epoch": 0.01420417861108959,
+      "grad_norm": 0.5510353446006775,
+      "learning_rate": 8.24e-06,
+      "loss": 2.4012,
+      "mean_token_accuracy": 0.47862795926630497,
+      "num_tokens": 119599588.0,
+      "step": 825
+    },
+    {
+      "entropy": 2.426513671875,
+      "epoch": 0.01422139579728485,
+      "grad_norm": 0.6013565063476562,
+      "learning_rate": 8.25e-06,
+      "loss": 2.345,
+      "mean_token_accuracy": 0.48915665224194527,
+      "num_tokens": 119745982.0,
+      "step": 826
+    },
+    {
+      "entropy": 2.506103515625,
+      "epoch": 0.01423861298348011,
+      "grad_norm": 0.5756415724754333,
+      "learning_rate": 8.26e-06,
+      "loss": 2.4707,
+      "mean_token_accuracy": 0.4742961646988988,
+      "num_tokens": 119883211.0,
+      "step": 827
+    },
+    {
+      "entropy": 2.427734375,
+      "epoch": 0.01425583016967537,
+      "grad_norm": 0.6262539625167847,
+      "learning_rate": 8.27e-06,
+      "loss": 2.3734,
+      "mean_token_accuracy": 0.4855869854800403,
+      "num_tokens": 120007564.0,
+      "step": 828
+    },
+    {
+      "entropy": 2.447998046875,
+      "epoch": 0.01427304735587063,
+      "grad_norm": 0.577958345413208,
+      "learning_rate": 8.28e-06,
+      "loss": 2.4136,
+      "mean_token_accuracy": 0.4838542784564197,
+      "num_tokens": 120147199.0,
+      "step": 829
+    },
+    {
+      "entropy": 2.4615478515625,
+      "epoch": 0.014290264542065891,
+      "grad_norm": 0.6253990530967712,
+      "learning_rate": 8.29e-06,
+      "loss": 2.4558,
+      "mean_token_accuracy": 0.4775591907091439,
+      "num_tokens": 120294626.0,
+      "step": 830
+    },
+    {
+      "entropy": 2.3575439453125,
+      "epoch": 0.01430748172826115,
+      "grad_norm": 0.5368080139160156,
+      "learning_rate": 8.3e-06,
+      "loss": 2.3092,
+      "mean_token_accuracy": 0.4943057978525758,
+      "num_tokens": 120451476.0,
+      "step": 831
+    },
+    {
+      "entropy": 2.447265625,
+      "epoch": 0.01432469891445641,
+      "grad_norm": 0.6116092801094055,
+      "learning_rate": 8.31e-06,
+      "loss": 2.4052,
+      "mean_token_accuracy": 0.47798394318670034,
+      "num_tokens": 120605020.0,
+      "step": 832
+    },
+    {
+      "entropy": 2.3782958984375,
+      "epoch": 0.01434191610065167,
+      "grad_norm": 0.5678116083145142,
+      "learning_rate": 8.32e-06,
+      "loss": 2.3034,
+      "mean_token_accuracy": 0.4946139776147902,
+      "num_tokens": 120752495.0,
+      "step": 833
+    },
+    {
+      "entropy": 2.363037109375,
+      "epoch": 0.014359133286846931,
+      "grad_norm": 0.5678343772888184,
+      "learning_rate": 8.33e-06,
+      "loss": 2.3141,
+      "mean_token_accuracy": 0.49313395330682397,
+      "num_tokens": 120900694.0,
+      "step": 834
+    },
+    {
+      "entropy": 2.4739990234375,
+      "epoch": 0.01437635047304219,
+      "grad_norm": 0.5975309014320374,
+      "learning_rate": 8.34e-06,
+      "loss": 2.4388,
+      "mean_token_accuracy": 0.4748355788178742,
+      "num_tokens": 121032159.0,
+      "step": 835
+    },
+    {
+      "entropy": 2.51806640625,
+      "epoch": 0.01439356765923745,
+      "grad_norm": 0.5592584013938904,
+      "learning_rate": 8.35e-06,
+      "loss": 2.4939,
+      "mean_token_accuracy": 0.4737435169517994,
+      "num_tokens": 121177988.0,
+      "step": 836
+    },
+    {
+      "entropy": 2.433837890625,
+      "epoch": 0.014410784845432711,
+      "grad_norm": 0.5620052218437195,
+      "learning_rate": 8.36e-06,
+      "loss": 2.3528,
+      "mean_token_accuracy": 0.484672705642879,
+      "num_tokens": 121333599.0,
+      "step": 837
+    },
+    {
+      "entropy": 2.4949951171875,
+      "epoch": 0.014428002031627972,
+      "grad_norm": 0.6018322706222534,
+      "learning_rate": 8.370000000000001e-06,
+      "loss": 2.4874,
+      "mean_token_accuracy": 0.471315645147115,
+      "num_tokens": 121483935.0,
+      "step": 838
+    },
+    {
+      "entropy": 2.4498291015625,
+      "epoch": 0.01444521921782323,
+      "grad_norm": 0.5906838774681091,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.4314,
+      "mean_token_accuracy": 0.48000403633341193,
+      "num_tokens": 121626405.0,
+      "step": 839
+    },
+    {
+      "entropy": 2.3984375,
+      "epoch": 0.014462436404018491,
+      "grad_norm": 0.6163339614868164,
+      "learning_rate": 8.390000000000001e-06,
+      "loss": 2.3408,
+      "mean_token_accuracy": 0.4879218554124236,
+      "num_tokens": 121755719.0,
+      "step": 840
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.014479653590213751,
+      "grad_norm": 0.6231170892715454,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.4589,
+      "mean_token_accuracy": 0.47292218124493957,
+      "num_tokens": 121876833.0,
+      "step": 841
+    },
+    {
+      "entropy": 2.4288330078125,
+      "epoch": 0.014496870776409012,
+      "grad_norm": 0.6095725893974304,
+      "learning_rate": 8.41e-06,
+      "loss": 2.3584,
+      "mean_token_accuracy": 0.4825339512899518,
+      "num_tokens": 122029110.0,
+      "step": 842
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.01451408796260427,
+      "grad_norm": 0.5682320594787598,
+      "learning_rate": 8.42e-06,
+      "loss": 2.4085,
+      "mean_token_accuracy": 0.47804021881893277,
+      "num_tokens": 122177885.0,
+      "step": 843
+    },
+    {
+      "entropy": 2.4327392578125,
+      "epoch": 0.014531305148799531,
+      "grad_norm": 0.5729749202728271,
+      "learning_rate": 8.43e-06,
+      "loss": 2.365,
+      "mean_token_accuracy": 0.4850345575250685,
+      "num_tokens": 122326681.0,
+      "step": 844
+    },
+    {
+      "entropy": 2.3778076171875,
+      "epoch": 0.014548522334994792,
+      "grad_norm": 0.5983383059501648,
+      "learning_rate": 8.44e-06,
+      "loss": 2.3205,
+      "mean_token_accuracy": 0.4961178773082793,
+      "num_tokens": 122468050.0,
+      "step": 845
+    },
+    {
+      "entropy": 2.498779296875,
+      "epoch": 0.014565739521190052,
+      "grad_norm": 0.5855973958969116,
+      "learning_rate": 8.45e-06,
+      "loss": 2.4899,
+      "mean_token_accuracy": 0.4722827118821442,
+      "num_tokens": 122605022.0,
+      "step": 846
+    },
+    {
+      "entropy": 2.51318359375,
+      "epoch": 0.014582956707385313,
+      "grad_norm": 0.7350115180015564,
+      "learning_rate": 8.46e-06,
+      "loss": 2.4897,
+      "mean_token_accuracy": 0.4719462259672582,
+      "num_tokens": 122756763.0,
+      "step": 847
+    },
+    {
+      "entropy": 2.453125,
+      "epoch": 0.014600173893580572,
+      "grad_norm": 0.5812132954597473,
+      "learning_rate": 8.47e-06,
+      "loss": 2.3954,
+      "mean_token_accuracy": 0.48041255166754127,
+      "num_tokens": 122903879.0,
+      "step": 848
+    },
+    {
+      "entropy": 2.479736328125,
+      "epoch": 0.014617391079775832,
+      "grad_norm": 0.5762138962745667,
+      "learning_rate": 8.48e-06,
+      "loss": 2.4433,
+      "mean_token_accuracy": 0.4698569756001234,
+      "num_tokens": 123041198.0,
+      "step": 849
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.014634608265971093,
+      "grad_norm": 0.5930407643318176,
+      "learning_rate": 8.49e-06,
+      "loss": 2.3483,
+      "mean_token_accuracy": 0.4922699723392725,
+      "num_tokens": 123183506.0,
+      "step": 850
+    },
+    {
+      "entropy": 2.3505859375,
+      "epoch": 0.014651825452166353,
+      "grad_norm": 0.5760442614555359,
+      "learning_rate": 8.5e-06,
+      "loss": 2.2918,
+      "mean_token_accuracy": 0.4991521080955863,
+      "num_tokens": 123324526.0,
+      "step": 851
+    },
+    {
+      "entropy": 2.53662109375,
+      "epoch": 0.014669042638361612,
+      "grad_norm": 0.6048057079315186,
+      "learning_rate": 8.51e-06,
+      "loss": 2.4955,
+      "mean_token_accuracy": 0.46434704307466745,
+      "num_tokens": 123453650.0,
+      "step": 852
+    },
+    {
+      "entropy": 2.4481201171875,
+      "epoch": 0.014686259824556873,
+      "grad_norm": 0.5807772874832153,
+      "learning_rate": 8.52e-06,
+      "loss": 2.4092,
+      "mean_token_accuracy": 0.4818004462867975,
+      "num_tokens": 123600203.0,
+      "step": 853
+    },
+    {
+      "entropy": 2.42041015625,
+      "epoch": 0.014703477010752133,
+      "grad_norm": 0.5653759837150574,
+      "learning_rate": 8.530000000000001e-06,
+      "loss": 2.3675,
+      "mean_token_accuracy": 0.4863137351348996,
+      "num_tokens": 123750776.0,
+      "step": 854
+    },
+    {
+      "entropy": 2.3917236328125,
+      "epoch": 0.014720694196947394,
+      "grad_norm": 0.5585445761680603,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.36,
+      "mean_token_accuracy": 0.4850917439907789,
+      "num_tokens": 123908995.0,
+      "step": 855
+    },
+    {
+      "entropy": 2.408203125,
+      "epoch": 0.014737911383142652,
+      "grad_norm": 0.6110696196556091,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 2.3376,
+      "mean_token_accuracy": 0.4934307490475476,
+      "num_tokens": 124048260.0,
+      "step": 856
+    },
+    {
+      "entropy": 2.4285888671875,
+      "epoch": 0.014755128569337913,
+      "grad_norm": 0.5494508147239685,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.3685,
+      "mean_token_accuracy": 0.485230874735862,
+      "num_tokens": 124195711.0,
+      "step": 857
+    },
+    {
+      "entropy": 2.49462890625,
+      "epoch": 0.014772345755533173,
+      "grad_norm": 0.5444778203964233,
+      "learning_rate": 8.570000000000001e-06,
+      "loss": 2.4763,
+      "mean_token_accuracy": 0.4732412826269865,
+      "num_tokens": 124345089.0,
+      "step": 858
+    },
+    {
+      "entropy": 2.4862060546875,
+      "epoch": 0.014789562941728434,
+      "grad_norm": 0.5632426738739014,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.447,
+      "mean_token_accuracy": 0.47163996985182166,
+      "num_tokens": 124491147.0,
+      "step": 859
+    },
+    {
+      "entropy": 2.492919921875,
+      "epoch": 0.014806780127923693,
+      "grad_norm": 0.5277729034423828,
+      "learning_rate": 8.59e-06,
+      "loss": 2.4818,
+      "mean_token_accuracy": 0.47168840002268553,
+      "num_tokens": 124653263.0,
+      "step": 860
+    },
+    {
+      "entropy": 2.5084228515625,
+      "epoch": 0.014823997314118953,
+      "grad_norm": 0.5971653461456299,
+      "learning_rate": 8.6e-06,
+      "loss": 2.4695,
+      "mean_token_accuracy": 0.4728867751546204,
+      "num_tokens": 124790799.0,
+      "step": 861
+    },
+    {
+      "entropy": 2.415771484375,
+      "epoch": 0.014841214500314214,
+      "grad_norm": 0.6263266801834106,
+      "learning_rate": 8.61e-06,
+      "loss": 2.3963,
+      "mean_token_accuracy": 0.4838090669363737,
+      "num_tokens": 124936295.0,
+      "step": 862
+    },
+    {
+      "entropy": 2.46142578125,
+      "epoch": 0.014858431686509474,
+      "grad_norm": 0.592215895652771,
+      "learning_rate": 8.62e-06,
+      "loss": 2.4003,
+      "mean_token_accuracy": 0.4809744218364358,
+      "num_tokens": 125067834.0,
+      "step": 863
+    },
+    {
+      "entropy": 2.4461669921875,
+      "epoch": 0.014875648872704733,
+      "grad_norm": 0.5886532664299011,
+      "learning_rate": 8.63e-06,
+      "loss": 2.4247,
+      "mean_token_accuracy": 0.48370860423892736,
+      "num_tokens": 125208751.0,
+      "step": 864
+    },
+    {
+      "entropy": 2.451904296875,
+      "epoch": 0.014892866058899994,
+      "grad_norm": 0.544989824295044,
+      "learning_rate": 8.64e-06,
+      "loss": 2.4477,
+      "mean_token_accuracy": 0.4746030508540571,
+      "num_tokens": 125362283.0,
+      "step": 865
+    },
+    {
+      "entropy": 2.47802734375,
+      "epoch": 0.014910083245095254,
+      "grad_norm": 0.5210621356964111,
+      "learning_rate": 8.65e-06,
+      "loss": 2.4684,
+      "mean_token_accuracy": 0.4717202326282859,
+      "num_tokens": 125535484.0,
+      "step": 866
+    },
+    {
+      "entropy": 2.4478759765625,
+      "epoch": 0.014927300431290515,
+      "grad_norm": 0.5993431806564331,
+      "learning_rate": 8.66e-06,
+      "loss": 2.3869,
+      "mean_token_accuracy": 0.47974208323284984,
+      "num_tokens": 125670144.0,
+      "step": 867
+    },
+    {
+      "entropy": 2.3865966796875,
+      "epoch": 0.014944517617485773,
+      "grad_norm": 0.5767236948013306,
+      "learning_rate": 8.67e-06,
+      "loss": 2.3517,
+      "mean_token_accuracy": 0.4869615617208183,
+      "num_tokens": 125807478.0,
+      "step": 868
+    },
+    {
+      "entropy": 2.526611328125,
+      "epoch": 0.014961734803681034,
+      "grad_norm": 0.6212364435195923,
+      "learning_rate": 8.68e-06,
+      "loss": 2.5197,
+      "mean_token_accuracy": 0.4605614240281284,
+      "num_tokens": 125927918.0,
+      "step": 869
+    },
+    {
+      "entropy": 2.4381103515625,
+      "epoch": 0.014978951989876295,
+      "grad_norm": 0.6073437929153442,
+      "learning_rate": 8.690000000000002e-06,
+      "loss": 2.3433,
+      "mean_token_accuracy": 0.4832576513290405,
+      "num_tokens": 126070123.0,
+      "step": 870
+    },
+    {
+      "entropy": 2.5057373046875,
+      "epoch": 0.014996169176071555,
+      "grad_norm": 0.5757139325141907,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.4574,
+      "mean_token_accuracy": 0.4752332870848477,
+      "num_tokens": 126221119.0,
+      "step": 871
+    },
+    {
+      "entropy": 2.47509765625,
+      "epoch": 0.015013386362266816,
+      "grad_norm": 0.5421658754348755,
+      "learning_rate": 8.710000000000001e-06,
+      "loss": 2.4344,
+      "mean_token_accuracy": 0.48030671663582325,
+      "num_tokens": 126376014.0,
+      "step": 872
+    },
+    {
+      "entropy": 2.404052734375,
+      "epoch": 0.015030603548462074,
+      "grad_norm": 0.6064536571502686,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.3335,
+      "mean_token_accuracy": 0.4924787334166467,
+      "num_tokens": 126511198.0,
+      "step": 873
+    },
+    {
+      "entropy": 2.507080078125,
+      "epoch": 0.015047820734657335,
+      "grad_norm": 0.5994257926940918,
+      "learning_rate": 8.730000000000001e-06,
+      "loss": 2.4807,
+      "mean_token_accuracy": 0.4721558247692883,
+      "num_tokens": 126647001.0,
+      "step": 874
+    },
+    {
+      "entropy": 2.5140380859375,
+      "epoch": 0.015065037920852595,
+      "grad_norm": 0.5920009016990662,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.479,
+      "mean_token_accuracy": 0.47505279863253236,
+      "num_tokens": 126795132.0,
+      "step": 875
+    },
+    {
+      "entropy": 2.4464111328125,
+      "epoch": 0.015082255107047856,
+      "grad_norm": 0.6077152490615845,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.4088,
+      "mean_token_accuracy": 0.48258367320522666,
+      "num_tokens": 126943201.0,
+      "step": 876
+    },
+    {
+      "entropy": 2.421630859375,
+      "epoch": 0.015099472293243115,
+      "grad_norm": 0.5878798365592957,
+      "learning_rate": 8.76e-06,
+      "loss": 2.3334,
+      "mean_token_accuracy": 0.48565139481797814,
+      "num_tokens": 127089107.0,
+      "step": 877
+    },
+    {
+      "entropy": 2.4451904296875,
+      "epoch": 0.015116689479438375,
+      "grad_norm": 0.6210505962371826,
+      "learning_rate": 8.77e-06,
+      "loss": 2.3817,
+      "mean_token_accuracy": 0.48540265718474984,
+      "num_tokens": 127229750.0,
+      "step": 878
+    },
+    {
+      "entropy": 2.404541015625,
+      "epoch": 0.015133906665633636,
+      "grad_norm": 0.6188786029815674,
+      "learning_rate": 8.78e-06,
+      "loss": 2.3565,
+      "mean_token_accuracy": 0.4887020909227431,
+      "num_tokens": 127362276.0,
+      "step": 879
+    },
+    {
+      "entropy": 2.43212890625,
+      "epoch": 0.015151123851828896,
+      "grad_norm": 0.587872326374054,
+      "learning_rate": 8.79e-06,
+      "loss": 2.4133,
+      "mean_token_accuracy": 0.48197965091094375,
+      "num_tokens": 127499990.0,
+      "step": 880
+    },
+    {
+      "entropy": 2.486572265625,
+      "epoch": 0.015168341038024155,
+      "grad_norm": 0.6304271817207336,
+      "learning_rate": 8.8e-06,
+      "loss": 2.4624,
+      "mean_token_accuracy": 0.47782522393390536,
+      "num_tokens": 127642780.0,
+      "step": 881
+    },
+    {
+      "entropy": 2.51318359375,
+      "epoch": 0.015185558224219416,
+      "grad_norm": 0.5970308780670166,
+      "learning_rate": 8.81e-06,
+      "loss": 2.4563,
+      "mean_token_accuracy": 0.47266413597390056,
+      "num_tokens": 127792589.0,
+      "step": 882
+    },
+    {
+      "entropy": 2.486083984375,
+      "epoch": 0.015202775410414676,
+      "grad_norm": 0.558076024055481,
+      "learning_rate": 8.82e-06,
+      "loss": 2.4215,
+      "mean_token_accuracy": 0.4797168541699648,
+      "num_tokens": 127940654.0,
+      "step": 883
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.015219992596609937,
+      "grad_norm": 0.569231390953064,
+      "learning_rate": 8.83e-06,
+      "loss": 2.3439,
+      "mean_token_accuracy": 0.48988029221072793,
+      "num_tokens": 128082621.0,
+      "step": 884
+    },
+    {
+      "entropy": 2.4755859375,
+      "epoch": 0.015237209782805195,
+      "grad_norm": 0.5827202796936035,
+      "learning_rate": 8.84e-06,
+      "loss": 2.4142,
+      "mean_token_accuracy": 0.4810488768853247,
+      "num_tokens": 128215498.0,
+      "step": 885
+    },
+    {
+      "entropy": 2.5076904296875,
+      "epoch": 0.015254426969000456,
+      "grad_norm": 0.5866194367408752,
+      "learning_rate": 8.85e-06,
+      "loss": 2.4625,
+      "mean_token_accuracy": 0.4744989378377795,
+      "num_tokens": 128353838.0,
+      "step": 886
+    },
+    {
+      "entropy": 2.479736328125,
+      "epoch": 0.015271644155195717,
+      "grad_norm": 0.5743751525878906,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 2.4037,
+      "mean_token_accuracy": 0.4828067999333143,
+      "num_tokens": 128499102.0,
+      "step": 887
+    },
+    {
+      "entropy": 2.485595703125,
+      "epoch": 0.015288861341390977,
+      "grad_norm": 0.5429300665855408,
+      "learning_rate": 8.870000000000001e-06,
+      "loss": 2.4312,
+      "mean_token_accuracy": 0.4705997440032661,
+      "num_tokens": 128658894.0,
+      "step": 888
+    },
+    {
+      "entropy": 2.4361572265625,
+      "epoch": 0.015306078527586236,
+      "grad_norm": 0.5508224368095398,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.4386,
+      "mean_token_accuracy": 0.48232552874833345,
+      "num_tokens": 128803595.0,
+      "step": 889
+    },
+    {
+      "entropy": 2.453369140625,
+      "epoch": 0.015323295713781496,
+      "grad_norm": 0.5834963917732239,
+      "learning_rate": 8.890000000000001e-06,
+      "loss": 2.4298,
+      "mean_token_accuracy": 0.4826282048597932,
+      "num_tokens": 128939226.0,
+      "step": 890
+    },
+    {
+      "entropy": 2.508056640625,
+      "epoch": 0.015340512899976757,
+      "grad_norm": 0.5773516297340393,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.4476,
+      "mean_token_accuracy": 0.4731542756780982,
+      "num_tokens": 129073231.0,
+      "step": 891
+    },
+    {
+      "entropy": 2.526611328125,
+      "epoch": 0.015357730086172017,
+      "grad_norm": 0.6158662438392639,
+      "learning_rate": 8.910000000000001e-06,
+      "loss": 2.5126,
+      "mean_token_accuracy": 0.47055464377626777,
+      "num_tokens": 129212971.0,
+      "step": 892
+    },
+    {
+      "entropy": 2.49365234375,
+      "epoch": 0.015374947272367276,
+      "grad_norm": 0.571234941482544,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.4735,
+      "mean_token_accuracy": 0.47626718133687973,
+      "num_tokens": 129364837.0,
+      "step": 893
+    },
+    {
+      "entropy": 2.50927734375,
+      "epoch": 0.015392164458562537,
+      "grad_norm": 0.5566508173942566,
+      "learning_rate": 8.930000000000001e-06,
+      "loss": 2.4622,
+      "mean_token_accuracy": 0.475893325638026,
+      "num_tokens": 129513090.0,
+      "step": 894
+    },
+    {
+      "entropy": 2.4493408203125,
+      "epoch": 0.015409381644757797,
+      "grad_norm": 0.6352636814117432,
+      "learning_rate": 8.94e-06,
+      "loss": 2.4011,
+      "mean_token_accuracy": 0.48302224883809686,
+      "num_tokens": 129639016.0,
+      "step": 895
+    },
+    {
+      "entropy": 2.4149169921875,
+      "epoch": 0.015426598830953058,
+      "grad_norm": 0.6071867942810059,
+      "learning_rate": 8.95e-06,
+      "loss": 2.3388,
+      "mean_token_accuracy": 0.48970347130671144,
+      "num_tokens": 129775367.0,
+      "step": 896
+    },
+    {
+      "entropy": 2.477783203125,
+      "epoch": 0.015443816017148317,
+      "grad_norm": 0.6278470754623413,
+      "learning_rate": 8.96e-06,
+      "loss": 2.475,
+      "mean_token_accuracy": 0.4722928828559816,
+      "num_tokens": 129899393.0,
+      "step": 897
+    },
+    {
+      "entropy": 2.461669921875,
+      "epoch": 0.015461033203343577,
+      "grad_norm": 0.5411078929901123,
+      "learning_rate": 8.97e-06,
+      "loss": 2.4289,
+      "mean_token_accuracy": 0.4794709859415889,
+      "num_tokens": 130049335.0,
+      "step": 898
+    },
+    {
+      "entropy": 2.46728515625,
+      "epoch": 0.015478250389538838,
+      "grad_norm": 0.6227301955223083,
+      "learning_rate": 8.98e-06,
+      "loss": 2.4675,
+      "mean_token_accuracy": 0.48002155777066946,
+      "num_tokens": 130191241.0,
+      "step": 899
+    },
+    {
+      "entropy": 2.513916015625,
+      "epoch": 0.015495467575734098,
+      "grad_norm": 0.553110659122467,
+      "learning_rate": 8.99e-06,
+      "loss": 2.4838,
+      "mean_token_accuracy": 0.47137136245146394,
+      "num_tokens": 130348446.0,
+      "step": 900
+    },
+    {
+      "entropy": 2.462890625,
+      "epoch": 0.015512684761929359,
+      "grad_norm": 0.596051037311554,
+      "learning_rate": 9e-06,
+      "loss": 2.4365,
+      "mean_token_accuracy": 0.48188950633630157,
+      "num_tokens": 130487415.0,
+      "step": 901
+    },
+    {
+      "entropy": 2.442138671875,
+      "epoch": 0.015529901948124617,
+      "grad_norm": 0.6585226058959961,
+      "learning_rate": 9.01e-06,
+      "loss": 2.4141,
+      "mean_token_accuracy": 0.4807861549779773,
+      "num_tokens": 130614941.0,
+      "step": 902
+    },
+    {
+      "entropy": 2.4937744140625,
+      "epoch": 0.015547119134319878,
+      "grad_norm": 0.6055911779403687,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.4229,
+      "mean_token_accuracy": 0.4837539931759238,
+      "num_tokens": 130754310.0,
+      "step": 903
+    },
+    {
+      "entropy": 2.43115234375,
+      "epoch": 0.015564336320515139,
+      "grad_norm": 0.5959157347679138,
+      "learning_rate": 9.030000000000002e-06,
+      "loss": 2.3802,
+      "mean_token_accuracy": 0.4859010446816683,
+      "num_tokens": 130906157.0,
+      "step": 904
+    },
+    {
+      "entropy": 2.461181640625,
+      "epoch": 0.015581553506710399,
+      "grad_norm": 0.5708449482917786,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.4052,
+      "mean_token_accuracy": 0.4815267431549728,
+      "num_tokens": 131059316.0,
+      "step": 905
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.015598770692905658,
+      "grad_norm": 0.577296257019043,
+      "learning_rate": 9.050000000000001e-06,
+      "loss": 2.3894,
+      "mean_token_accuracy": 0.4845569171011448,
+      "num_tokens": 131205522.0,
+      "step": 906
+    },
+    {
+      "entropy": 2.41845703125,
+      "epoch": 0.015615987879100918,
+      "grad_norm": 0.60146164894104,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.3876,
+      "mean_token_accuracy": 0.4812913998030126,
+      "num_tokens": 131352873.0,
+      "step": 907
+    },
+    {
+      "entropy": 2.4771728515625,
+      "epoch": 0.01563320506529618,
+      "grad_norm": 0.6240298748016357,
+      "learning_rate": 9.070000000000001e-06,
+      "loss": 2.4327,
+      "mean_token_accuracy": 0.4764832267537713,
+      "num_tokens": 131486068.0,
+      "step": 908
+    },
+    {
+      "entropy": 2.4599609375,
+      "epoch": 0.01565042225149144,
+      "grad_norm": 1.2479300498962402,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.4198,
+      "mean_token_accuracy": 0.47274223202839494,
+      "num_tokens": 131625641.0,
+      "step": 909
+    },
+    {
+      "entropy": 2.379638671875,
+      "epoch": 0.0156676394376867,
+      "grad_norm": 0.6223964095115662,
+      "learning_rate": 9.090000000000001e-06,
+      "loss": 2.3421,
+      "mean_token_accuracy": 0.4910487150773406,
+      "num_tokens": 131774897.0,
+      "step": 910
+    },
+    {
+      "entropy": 2.4305419921875,
+      "epoch": 0.01568485662388196,
+      "grad_norm": 0.6342669725418091,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.4116,
+      "mean_token_accuracy": 0.48323542159050703,
+      "num_tokens": 131918941.0,
+      "step": 911
+    },
+    {
+      "entropy": 2.3966064453125,
+      "epoch": 0.015702073810077218,
+      "grad_norm": 0.5977234244346619,
+      "learning_rate": 9.110000000000001e-06,
+      "loss": 2.3407,
+      "mean_token_accuracy": 0.4932427750900388,
+      "num_tokens": 132058255.0,
+      "step": 912
+    },
+    {
+      "entropy": 2.4349365234375,
+      "epoch": 0.015719290996272478,
+      "grad_norm": 0.5736342072486877,
+      "learning_rate": 9.12e-06,
+      "loss": 2.3943,
+      "mean_token_accuracy": 0.4837222811765969,
+      "num_tokens": 132197276.0,
+      "step": 913
+    },
+    {
+      "entropy": 2.4132080078125,
+      "epoch": 0.01573650818246774,
+      "grad_norm": 0.6241025924682617,
+      "learning_rate": 9.13e-06,
+      "loss": 2.3856,
+      "mean_token_accuracy": 0.48743872344493866,
+      "num_tokens": 132337325.0,
+      "step": 914
+    },
+    {
+      "entropy": 2.41748046875,
+      "epoch": 0.015753725368663,
+      "grad_norm": 0.6098312735557556,
+      "learning_rate": 9.14e-06,
+      "loss": 2.374,
+      "mean_token_accuracy": 0.4855101592838764,
+      "num_tokens": 132486321.0,
+      "step": 915
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.01577094255485826,
+      "grad_norm": 0.5596804022789001,
+      "learning_rate": 9.15e-06,
+      "loss": 2.3876,
+      "mean_token_accuracy": 0.47576976707205176,
+      "num_tokens": 132647149.0,
+      "step": 916
+    },
+    {
+      "entropy": 2.4208984375,
+      "epoch": 0.01578815974105352,
+      "grad_norm": 0.5681569576263428,
+      "learning_rate": 9.16e-06,
+      "loss": 2.3797,
+      "mean_token_accuracy": 0.48612832371145487,
+      "num_tokens": 132795123.0,
+      "step": 917
+    },
+    {
+      "entropy": 2.4420166015625,
+      "epoch": 0.01580537692724878,
+      "grad_norm": 0.6165553331375122,
+      "learning_rate": 9.17e-06,
+      "loss": 2.4223,
+      "mean_token_accuracy": 0.4839365719817579,
+      "num_tokens": 132930289.0,
+      "step": 918
+    },
+    {
+      "entropy": 2.458740234375,
+      "epoch": 0.01582259411344404,
+      "grad_norm": 0.6469439268112183,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.4158,
+      "mean_token_accuracy": 0.48115426022559404,
+      "num_tokens": 133057834.0,
+      "step": 919
+    },
+    {
+      "entropy": 2.4097900390625,
+      "epoch": 0.015839811299639298,
+      "grad_norm": 0.5866466164588928,
+      "learning_rate": 9.190000000000002e-06,
+      "loss": 2.3269,
+      "mean_token_accuracy": 0.48917456716299057,
+      "num_tokens": 133200644.0,
+      "step": 920
+    },
+    {
+      "entropy": 2.463623046875,
+      "epoch": 0.01585702848583456,
+      "grad_norm": 0.5503515601158142,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.4136,
+      "mean_token_accuracy": 0.478064242284745,
+      "num_tokens": 133352710.0,
+      "step": 921
+    },
+    {
+      "entropy": 2.4775390625,
+      "epoch": 0.01587424567202982,
+      "grad_norm": 0.5962862968444824,
+      "learning_rate": 9.210000000000002e-06,
+      "loss": 2.454,
+      "mean_token_accuracy": 0.47544111032038927,
+      "num_tokens": 133501140.0,
+      "step": 922
+    },
+    {
+      "entropy": 2.502197265625,
+      "epoch": 0.01589146285822508,
+      "grad_norm": 0.6619735360145569,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.4128,
+      "mean_token_accuracy": 0.47879257379099727,
+      "num_tokens": 133618064.0,
+      "step": 923
+    },
+    {
+      "entropy": 2.3917236328125,
+      "epoch": 0.01590868004442034,
+      "grad_norm": 0.6243993639945984,
+      "learning_rate": 9.230000000000001e-06,
+      "loss": 2.3601,
+      "mean_token_accuracy": 0.49083237862214446,
+      "num_tokens": 133746355.0,
+      "step": 924
+    },
+    {
+      "entropy": 2.4593505859375,
+      "epoch": 0.0159258972306156,
+      "grad_norm": 0.5959250926971436,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.3893,
+      "mean_token_accuracy": 0.487882892601192,
+      "num_tokens": 133887812.0,
+      "step": 925
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.01594311441681086,
+      "grad_norm": 0.6122800707817078,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 2.432,
+      "mean_token_accuracy": 0.47797348722815514,
+      "num_tokens": 134038997.0,
+      "step": 926
+    },
+    {
+      "entropy": 2.46875,
+      "epoch": 0.015960331603006122,
+      "grad_norm": 0.547160267829895,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 2.4407,
+      "mean_token_accuracy": 0.47520409850403666,
+      "num_tokens": 134198302.0,
+      "step": 927
+    },
+    {
+      "entropy": 2.48974609375,
+      "epoch": 0.015977548789201382,
+      "grad_norm": 0.5570476651191711,
+      "learning_rate": 9.270000000000001e-06,
+      "loss": 2.4886,
+      "mean_token_accuracy": 0.47539124358445406,
+      "num_tokens": 134342027.0,
+      "step": 928
+    },
+    {
+      "entropy": 2.41064453125,
+      "epoch": 0.01599476597539664,
+      "grad_norm": 0.5551185011863708,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.3801,
+      "mean_token_accuracy": 0.48781104385852814,
+      "num_tokens": 134495434.0,
+      "step": 929
+    },
+    {
+      "entropy": 2.4661865234375,
+      "epoch": 0.0160119831615919,
+      "grad_norm": 0.6195770502090454,
+      "learning_rate": 9.29e-06,
+      "loss": 2.4388,
+      "mean_token_accuracy": 0.4786299457773566,
+      "num_tokens": 134626002.0,
+      "step": 930
+    },
+    {
+      "entropy": 2.4622802734375,
+      "epoch": 0.01602920034778716,
+      "grad_norm": 0.6120036244392395,
+      "learning_rate": 9.3e-06,
+      "loss": 2.4435,
+      "mean_token_accuracy": 0.47908638790249825,
+      "num_tokens": 134778634.0,
+      "step": 931
+    },
+    {
+      "entropy": 2.5621337890625,
+      "epoch": 0.01604641753398242,
+      "grad_norm": 0.5915412306785583,
+      "learning_rate": 9.31e-06,
+      "loss": 2.5866,
+      "mean_token_accuracy": 0.462941222358495,
+      "num_tokens": 134929621.0,
+      "step": 932
+    },
+    {
+      "entropy": 2.41796875,
+      "epoch": 0.01606363472017768,
+      "grad_norm": 0.5830768346786499,
+      "learning_rate": 9.32e-06,
+      "loss": 2.355,
+      "mean_token_accuracy": 0.4878848767839372,
+      "num_tokens": 135086585.0,
+      "step": 933
+    },
+    {
+      "entropy": 2.353759765625,
+      "epoch": 0.016080851906372942,
+      "grad_norm": 0.6241956949234009,
+      "learning_rate": 9.33e-06,
+      "loss": 2.3049,
+      "mean_token_accuracy": 0.49892251333221793,
+      "num_tokens": 135231782.0,
+      "step": 934
+    },
+    {
+      "entropy": 2.406005859375,
+      "epoch": 0.016098069092568203,
+      "grad_norm": 0.5853713750839233,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.4124,
+      "mean_token_accuracy": 0.4829076291061938,
+      "num_tokens": 135378790.0,
+      "step": 935
+    },
+    {
+      "entropy": 2.5213623046875,
+      "epoch": 0.016115286278763463,
+      "grad_norm": 0.5981717109680176,
+      "learning_rate": 9.350000000000002e-06,
+      "loss": 2.5113,
+      "mean_token_accuracy": 0.4728419524617493,
+      "num_tokens": 135520311.0,
+      "step": 936
+    },
+    {
+      "entropy": 2.5010986328125,
+      "epoch": 0.01613250346495872,
+      "grad_norm": 0.5917856693267822,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.4518,
+      "mean_token_accuracy": 0.46924112644046545,
+      "num_tokens": 135669776.0,
+      "step": 937
+    },
+    {
+      "entropy": 2.42724609375,
+      "epoch": 0.01614972065115398,
+      "grad_norm": 0.5660749673843384,
+      "learning_rate": 9.370000000000002e-06,
+      "loss": 2.3848,
+      "mean_token_accuracy": 0.4782256758771837,
+      "num_tokens": 135821760.0,
+      "step": 938
+    },
+    {
+      "entropy": 2.498779296875,
+      "epoch": 0.01616693783734924,
+      "grad_norm": 0.5981243252754211,
+      "learning_rate": 9.38e-06,
+      "loss": 2.4628,
+      "mean_token_accuracy": 0.4730893294326961,
+      "num_tokens": 135956326.0,
+      "step": 939
+    },
+    {
+      "entropy": 2.3973388671875,
+      "epoch": 0.016184155023544502,
+      "grad_norm": 0.580276608467102,
+      "learning_rate": 9.39e-06,
+      "loss": 2.3663,
+      "mean_token_accuracy": 0.4926355588249862,
+      "num_tokens": 136101164.0,
+      "step": 940
+    },
+    {
+      "entropy": 2.470947265625,
+      "epoch": 0.016201372209739762,
+      "grad_norm": 0.5627194046974182,
+      "learning_rate": 9.4e-06,
+      "loss": 2.416,
+      "mean_token_accuracy": 0.4773054295219481,
+      "num_tokens": 136251505.0,
+      "step": 941
+    },
+    {
+      "entropy": 2.519775390625,
+      "epoch": 0.016218589395935023,
+      "grad_norm": 0.6778742671012878,
+      "learning_rate": 9.41e-06,
+      "loss": 2.4862,
+      "mean_token_accuracy": 0.4745273180305958,
+      "num_tokens": 136391079.0,
+      "step": 942
+    },
+    {
+      "entropy": 2.4498291015625,
+      "epoch": 0.016235806582130283,
+      "grad_norm": 0.582788348197937,
+      "learning_rate": 9.42e-06,
+      "loss": 2.3946,
+      "mean_token_accuracy": 0.48375819250941277,
+      "num_tokens": 136540299.0,
+      "step": 943
+    },
+    {
+      "entropy": 2.4471435546875,
+      "epoch": 0.016253023768325544,
+      "grad_norm": 0.5528161525726318,
+      "learning_rate": 9.43e-06,
+      "loss": 2.3405,
+      "mean_token_accuracy": 0.47900286270305514,
+      "num_tokens": 136688027.0,
+      "step": 944
+    },
+    {
+      "entropy": 2.429931640625,
+      "epoch": 0.0162702409545208,
+      "grad_norm": 0.6193938851356506,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.3861,
+      "mean_token_accuracy": 0.4864586624316871,
+      "num_tokens": 136823997.0,
+      "step": 945
+    },
+    {
+      "entropy": 2.4169921875,
+      "epoch": 0.01628745814071606,
+      "grad_norm": 0.5639959573745728,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 2.3965,
+      "mean_token_accuracy": 0.4871432662475854,
+      "num_tokens": 136967987.0,
+      "step": 946
+    },
+    {
+      "entropy": 2.4097900390625,
+      "epoch": 0.016304675326911322,
+      "grad_norm": 0.593999981880188,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.48685387754812837,
+      "num_tokens": 137120260.0,
+      "step": 947
+    },
+    {
+      "entropy": 2.4072265625,
+      "epoch": 0.016321892513106583,
+      "grad_norm": 0.5643085837364197,
+      "learning_rate": 9.47e-06,
+      "loss": 2.387,
+      "mean_token_accuracy": 0.4881442333571613,
+      "num_tokens": 137280017.0,
+      "step": 948
+    },
+    {
+      "entropy": 2.474853515625,
+      "epoch": 0.016339109699301843,
+      "grad_norm": 0.602178156375885,
+      "learning_rate": 9.48e-06,
+      "loss": 2.4429,
+      "mean_token_accuracy": 0.4794820128008723,
+      "num_tokens": 137431493.0,
+      "step": 949
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.016356326885497104,
+      "grad_norm": 0.5669793486595154,
+      "learning_rate": 9.49e-06,
+      "loss": 2.4303,
+      "mean_token_accuracy": 0.4830532097257674,
+      "num_tokens": 137575150.0,
+      "step": 950
+    },
+    {
+      "entropy": 2.4486083984375,
+      "epoch": 0.016373544071692364,
+      "grad_norm": 0.5908523201942444,
+      "learning_rate": 9.5e-06,
+      "loss": 2.4052,
+      "mean_token_accuracy": 0.4807011899538338,
+      "num_tokens": 137718243.0,
+      "step": 951
+    },
+    {
+      "entropy": 2.466552734375,
+      "epoch": 0.016390761257887625,
+      "grad_norm": 0.5942707061767578,
+      "learning_rate": 9.51e-06,
+      "loss": 2.4347,
+      "mean_token_accuracy": 0.47939926059916615,
+      "num_tokens": 137855181.0,
+      "step": 952
+    },
+    {
+      "entropy": 2.43359375,
+      "epoch": 0.016407978444082885,
+      "grad_norm": 0.6280327439308167,
+      "learning_rate": 9.52e-06,
+      "loss": 2.4106,
+      "mean_token_accuracy": 0.48289193166419864,
+      "num_tokens": 137984833.0,
+      "step": 953
+    },
+    {
+      "entropy": 2.4107666015625,
+      "epoch": 0.016425195630278142,
+      "grad_norm": 0.6055025458335876,
+      "learning_rate": 9.53e-06,
+      "loss": 2.3538,
+      "mean_token_accuracy": 0.48937113396823406,
+      "num_tokens": 138132043.0,
+      "step": 954
+    },
+    {
+      "entropy": 2.403076171875,
+      "epoch": 0.016442412816473403,
+      "grad_norm": 0.557295024394989,
+      "learning_rate": 9.54e-06,
+      "loss": 2.3975,
+      "mean_token_accuracy": 0.48123540729284286,
+      "num_tokens": 138295818.0,
+      "step": 955
+    },
+    {
+      "entropy": 2.475830078125,
+      "epoch": 0.016459630002668663,
+      "grad_norm": 0.6141370534896851,
+      "learning_rate": 9.55e-06,
+      "loss": 2.4164,
+      "mean_token_accuracy": 0.47790815867483616,
+      "num_tokens": 138437598.0,
+      "step": 956
+    },
+    {
+      "entropy": 2.3380126953125,
+      "epoch": 0.016476847188863924,
+      "grad_norm": 0.5739216208457947,
+      "learning_rate": 9.56e-06,
+      "loss": 2.277,
+      "mean_token_accuracy": 0.5020720390602946,
+      "num_tokens": 138579286.0,
+      "step": 957
+    },
+    {
+      "entropy": 2.482177734375,
+      "epoch": 0.016494064375059184,
+      "grad_norm": 0.5679910778999329,
+      "learning_rate": 9.57e-06,
+      "loss": 2.4916,
+      "mean_token_accuracy": 0.4723518299870193,
+      "num_tokens": 138737250.0,
+      "step": 958
+    },
+    {
+      "entropy": 2.428466796875,
+      "epoch": 0.016511281561254445,
+      "grad_norm": 0.5608857274055481,
+      "learning_rate": 9.58e-06,
+      "loss": 2.3651,
+      "mean_token_accuracy": 0.48081721225753427,
+      "num_tokens": 138887535.0,
+      "step": 959
+    },
+    {
+      "entropy": 2.45068359375,
+      "epoch": 0.016528498747449705,
+      "grad_norm": 0.5733979344367981,
+      "learning_rate": 9.59e-06,
+      "loss": 2.3793,
+      "mean_token_accuracy": 0.4807218159548938,
+      "num_tokens": 139046347.0,
+      "step": 960
+    },
+    {
+      "entropy": 2.4381103515625,
+      "epoch": 0.016545715933644966,
+      "grad_norm": 0.5601542592048645,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.3687,
+      "mean_token_accuracy": 0.48160395165905356,
+      "num_tokens": 139201178.0,
+      "step": 961
+    },
+    {
+      "entropy": 2.4107666015625,
+      "epoch": 0.016562933119840223,
+      "grad_norm": 0.5981694459915161,
+      "learning_rate": 9.610000000000001e-06,
+      "loss": 2.3668,
+      "mean_token_accuracy": 0.48300402937456965,
+      "num_tokens": 139337844.0,
+      "step": 962
+    },
+    {
+      "entropy": 2.4951171875,
+      "epoch": 0.016580150306035484,
+      "grad_norm": 0.5836125016212463,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.4706,
+      "mean_token_accuracy": 0.467432489618659,
+      "num_tokens": 139469993.0,
+      "step": 963
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.016597367492230744,
+      "grad_norm": 0.5777595043182373,
+      "learning_rate": 9.630000000000001e-06,
+      "loss": 2.3732,
+      "mean_token_accuracy": 0.48406707495450974,
+      "num_tokens": 139628918.0,
+      "step": 964
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.016614584678426005,
+      "grad_norm": 0.5910548567771912,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.4025,
+      "mean_token_accuracy": 0.48476587794721127,
+      "num_tokens": 139779064.0,
+      "step": 965
+    },
+    {
+      "entropy": 2.4569091796875,
+      "epoch": 0.016631801864621265,
+      "grad_norm": 0.6000712513923645,
+      "learning_rate": 9.65e-06,
+      "loss": 2.4165,
+      "mean_token_accuracy": 0.4791805609129369,
+      "num_tokens": 139944530.0,
+      "step": 966
+    },
+    {
+      "entropy": 2.45849609375,
+      "epoch": 0.016649019050816526,
+      "grad_norm": 0.5988768935203552,
+      "learning_rate": 9.66e-06,
+      "loss": 2.3822,
+      "mean_token_accuracy": 0.4866671049967408,
+      "num_tokens": 140083921.0,
+      "step": 967
+    },
+    {
+      "entropy": 2.46728515625,
+      "epoch": 0.016666236237011786,
+      "grad_norm": 0.5854329466819763,
+      "learning_rate": 9.67e-06,
+      "loss": 2.3907,
+      "mean_token_accuracy": 0.47991981683298945,
+      "num_tokens": 140224971.0,
+      "step": 968
+    },
+    {
+      "entropy": 2.533447265625,
+      "epoch": 0.016683453423207047,
+      "grad_norm": 0.6449822187423706,
+      "learning_rate": 9.68e-06,
+      "loss": 2.5346,
+      "mean_token_accuracy": 0.4715711669996381,
+      "num_tokens": 140353765.0,
+      "step": 969
+    },
+    {
+      "entropy": 2.4517822265625,
+      "epoch": 0.016700670609402304,
+      "grad_norm": 0.562483549118042,
+      "learning_rate": 9.69e-06,
+      "loss": 2.4227,
+      "mean_token_accuracy": 0.47965225437656045,
+      "num_tokens": 140498830.0,
+      "step": 970
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.016717887795597564,
+      "grad_norm": 0.5308300852775574,
+      "learning_rate": 9.7e-06,
+      "loss": 2.4248,
+      "mean_token_accuracy": 0.47759495256468654,
+      "num_tokens": 140665634.0,
+      "step": 971
+    },
+    {
+      "entropy": 2.492919921875,
+      "epoch": 0.016735104981792825,
+      "grad_norm": 0.9326145052909851,
+      "learning_rate": 9.71e-06,
+      "loss": 2.4872,
+      "mean_token_accuracy": 0.468785522505641,
+      "num_tokens": 140802529.0,
+      "step": 972
+    },
+    {
+      "entropy": 2.37744140625,
+      "epoch": 0.016752322167988085,
+      "grad_norm": 0.608925998210907,
+      "learning_rate": 9.72e-06,
+      "loss": 2.3317,
+      "mean_token_accuracy": 0.4957911465317011,
+      "num_tokens": 140941381.0,
+      "step": 973
+    },
+    {
+      "entropy": 2.4649658203125,
+      "epoch": 0.016769539354183346,
+      "grad_norm": 0.5866920948028564,
+      "learning_rate": 9.73e-06,
+      "loss": 2.4661,
+      "mean_token_accuracy": 0.4737902185879648,
+      "num_tokens": 141074013.0,
+      "step": 974
+    },
+    {
+      "entropy": 2.52001953125,
+      "epoch": 0.016786756540378606,
+      "grad_norm": 0.6548328399658203,
+      "learning_rate": 9.74e-06,
+      "loss": 2.5289,
+      "mean_token_accuracy": 0.4689825112000108,
+      "num_tokens": 141211732.0,
+      "step": 975
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 0.016803973726573867,
+      "grad_norm": 0.5622747540473938,
+      "learning_rate": 9.75e-06,
+      "loss": 2.4627,
+      "mean_token_accuracy": 0.474064817186445,
+      "num_tokens": 141361499.0,
+      "step": 976
+    },
+    {
+      "entropy": 2.4368896484375,
+      "epoch": 0.016821190912769127,
+      "grad_norm": 0.5637742280960083,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.4061,
+      "mean_token_accuracy": 0.4773513269610703,
+      "num_tokens": 141516957.0,
+      "step": 977
+    },
+    {
+      "entropy": 2.357666015625,
+      "epoch": 0.016838408098964388,
+      "grad_norm": 0.6168224215507507,
+      "learning_rate": 9.770000000000001e-06,
+      "loss": 2.3441,
+      "mean_token_accuracy": 0.4971775123849511,
+      "num_tokens": 141659725.0,
+      "step": 978
+    },
+    {
+      "entropy": 2.482666015625,
+      "epoch": 0.016855625285159645,
+      "grad_norm": 0.5854248404502869,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.4517,
+      "mean_token_accuracy": 0.4778465088456869,
+      "num_tokens": 141812273.0,
+      "step": 979
+    },
+    {
+      "entropy": 2.5281982421875,
+      "epoch": 0.016872842471354906,
+      "grad_norm": 0.5742073059082031,
+      "learning_rate": 9.790000000000001e-06,
+      "loss": 2.5098,
+      "mean_token_accuracy": 0.47218647226691246,
+      "num_tokens": 141952803.0,
+      "step": 980
+    },
+    {
+      "entropy": 2.4508056640625,
+      "epoch": 0.016890059657550166,
+      "grad_norm": 0.7346315383911133,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.3992,
+      "mean_token_accuracy": 0.4813456032425165,
+      "num_tokens": 142105077.0,
+      "step": 981
+    },
+    {
+      "entropy": 2.3946533203125,
+      "epoch": 0.016907276843745427,
+      "grad_norm": 0.613284170627594,
+      "learning_rate": 9.810000000000001e-06,
+      "loss": 2.3475,
+      "mean_token_accuracy": 0.49409126583486795,
+      "num_tokens": 142252173.0,
+      "step": 982
+    },
+    {
+      "entropy": 2.473876953125,
+      "epoch": 0.016924494029940687,
+      "grad_norm": 0.5658174753189087,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.459,
+      "mean_token_accuracy": 0.47720590187236667,
+      "num_tokens": 142396038.0,
+      "step": 983
+    },
+    {
+      "entropy": 2.395751953125,
+      "epoch": 0.016941711216135948,
+      "grad_norm": 0.5382007956504822,
+      "learning_rate": 9.83e-06,
+      "loss": 2.373,
+      "mean_token_accuracy": 0.4882864346727729,
+      "num_tokens": 142564149.0,
+      "step": 984
+    },
+    {
+      "entropy": 2.482666015625,
+      "epoch": 0.016958928402331208,
+      "grad_norm": 0.5813350677490234,
+      "learning_rate": 9.84e-06,
+      "loss": 2.4836,
+      "mean_token_accuracy": 0.4725424610078335,
+      "num_tokens": 142701037.0,
+      "step": 985
+    },
+    {
+      "entropy": 2.487060546875,
+      "epoch": 0.01697614558852647,
+      "grad_norm": 0.5810776948928833,
+      "learning_rate": 9.85e-06,
+      "loss": 2.4691,
+      "mean_token_accuracy": 0.48016467317938805,
+      "num_tokens": 142855073.0,
+      "step": 986
+    },
+    {
+      "entropy": 2.460693359375,
+      "epoch": 0.016993362774721726,
+      "grad_norm": 0.5684233903884888,
+      "learning_rate": 9.86e-06,
+      "loss": 2.4004,
+      "mean_token_accuracy": 0.47402910608798265,
+      "num_tokens": 143006703.0,
+      "step": 987
+    },
+    {
+      "entropy": 2.426513671875,
+      "epoch": 0.017010579960916986,
+      "grad_norm": 0.5875155925750732,
+      "learning_rate": 9.87e-06,
+      "loss": 2.4032,
+      "mean_token_accuracy": 0.4833544669672847,
+      "num_tokens": 143158319.0,
+      "step": 988
+    },
+    {
+      "entropy": 2.3924560546875,
+      "epoch": 0.017027797147112247,
+      "grad_norm": 0.9405050873756409,
+      "learning_rate": 9.88e-06,
+      "loss": 2.3455,
+      "mean_token_accuracy": 0.4880230622366071,
+      "num_tokens": 143281917.0,
+      "step": 989
+    },
+    {
+      "entropy": 2.43017578125,
+      "epoch": 0.017045014333307507,
+      "grad_norm": 0.593075156211853,
+      "learning_rate": 9.89e-06,
+      "loss": 2.3675,
+      "mean_token_accuracy": 0.48371881106868386,
+      "num_tokens": 143424058.0,
+      "step": 990
+    },
+    {
+      "entropy": 2.4437255859375,
+      "epoch": 0.017062231519502768,
+      "grad_norm": 0.5774204134941101,
+      "learning_rate": 9.9e-06,
+      "loss": 2.4004,
+      "mean_token_accuracy": 0.47859969455748796,
+      "num_tokens": 143567199.0,
+      "step": 991
+    },
+    {
+      "entropy": 2.468505859375,
+      "epoch": 0.01707944870569803,
+      "grad_norm": 0.6035756468772888,
+      "learning_rate": 9.91e-06,
+      "loss": 2.3898,
+      "mean_token_accuracy": 0.48759705713018775,
+      "num_tokens": 143712988.0,
+      "step": 992
+    },
+    {
+      "entropy": 2.4527587890625,
+      "epoch": 0.01709666589189329,
+      "grad_norm": 0.5580205321311951,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 2.4491,
+      "mean_token_accuracy": 0.47088390588760376,
+      "num_tokens": 143861050.0,
+      "step": 993
+    },
+    {
+      "entropy": 2.4556884765625,
+      "epoch": 0.01711388307808855,
+      "grad_norm": 0.5672869086265564,
+      "learning_rate": 9.930000000000001e-06,
+      "loss": 2.4092,
+      "mean_token_accuracy": 0.48082907358184457,
+      "num_tokens": 144017646.0,
+      "step": 994
+    },
+    {
+      "entropy": 2.445556640625,
+      "epoch": 0.017131100264283806,
+      "grad_norm": 0.6213098764419556,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.4035,
+      "mean_token_accuracy": 0.48345539439469576,
+      "num_tokens": 144150604.0,
+      "step": 995
+    },
+    {
+      "entropy": 2.489990234375,
+      "epoch": 0.017148317450479067,
+      "grad_norm": 0.5691782832145691,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 2.4575,
+      "mean_token_accuracy": 0.4700373010709882,
+      "num_tokens": 144292661.0,
+      "step": 996
+    },
+    {
+      "entropy": 2.454833984375,
+      "epoch": 0.017165534636674327,
+      "grad_norm": 0.5717594027519226,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.418,
+      "mean_token_accuracy": 0.48314423533156514,
+      "num_tokens": 144445246.0,
+      "step": 997
+    },
+    {
+      "entropy": 2.4835205078125,
+      "epoch": 0.017182751822869588,
+      "grad_norm": 0.5863724946975708,
+      "learning_rate": 9.970000000000001e-06,
+      "loss": 2.4627,
+      "mean_token_accuracy": 0.4698511818423867,
+      "num_tokens": 144589498.0,
+      "step": 998
+    },
+    {
+      "entropy": 2.441650390625,
+      "epoch": 0.01719996900906485,
+      "grad_norm": 0.770044207572937,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 2.3511,
+      "mean_token_accuracy": 0.4848445672541857,
+      "num_tokens": 144744013.0,
+      "step": 999
+    },
+    {
+      "entropy": 2.443115234375,
+      "epoch": 0.01721718619526011,
+      "grad_norm": 0.6044683456420898,
+      "learning_rate": 9.990000000000001e-06,
+      "loss": 2.411,
+      "mean_token_accuracy": 0.4802071452140808,
+      "num_tokens": 144878720.0,
+      "step": 1000
+    },
+    {
+      "entropy": 2.4459228515625,
+      "epoch": 0.01723440338145537,
+      "grad_norm": 0.5749707221984863,
+      "learning_rate": 1e-05,
+      "loss": 2.3881,
+      "mean_token_accuracy": 0.480140985455364,
+      "num_tokens": 145028328.0,
+      "step": 1001
+    },
+    {
+      "entropy": 2.55908203125,
+      "epoch": 0.01725162056765063,
+      "grad_norm": 0.6056835651397705,
+      "learning_rate": 9.999999992427464e-06,
+      "loss": 2.5261,
+      "mean_token_accuracy": 0.46429841918870807,
+      "num_tokens": 145177909.0,
+      "step": 1002
+    },
+    {
+      "entropy": 2.475830078125,
+      "epoch": 0.01726883775384589,
+      "grad_norm": 0.6159217357635498,
+      "learning_rate": 9.999999969709854e-06,
+      "loss": 2.4367,
+      "mean_token_accuracy": 0.4752111411653459,
+      "num_tokens": 145331073.0,
+      "step": 1003
+    },
+    {
+      "entropy": 2.4053955078125,
+      "epoch": 0.017286054940041148,
+      "grad_norm": 0.5820682644844055,
+      "learning_rate": 9.999999931847169e-06,
+      "loss": 2.3818,
+      "mean_token_accuracy": 0.4844290711916983,
+      "num_tokens": 145480948.0,
+      "step": 1004
+    },
+    {
+      "entropy": 2.432861328125,
+      "epoch": 0.017303272126236408,
+      "grad_norm": 0.5615769624710083,
+      "learning_rate": 9.999999878839412e-06,
+      "loss": 2.3797,
+      "mean_token_accuracy": 0.4817391145043075,
+      "num_tokens": 145615367.0,
+      "step": 1005
+    },
+    {
+      "entropy": 2.3687744140625,
+      "epoch": 0.01732048931243167,
+      "grad_norm": 0.5944392085075378,
+      "learning_rate": 9.999999810686582e-06,
+      "loss": 2.3464,
+      "mean_token_accuracy": 0.4950461401604116,
+      "num_tokens": 145751257.0,
+      "step": 1006
+    },
+    {
+      "entropy": 2.491455078125,
+      "epoch": 0.01733770649862693,
+      "grad_norm": 0.5785252451896667,
+      "learning_rate": 9.99999972738868e-06,
+      "loss": 2.4419,
+      "mean_token_accuracy": 0.4759130119346082,
+      "num_tokens": 145881989.0,
+      "step": 1007
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.01735492368482219,
+      "grad_norm": 0.573020339012146,
+      "learning_rate": 9.999999628945702e-06,
+      "loss": 2.4126,
+      "mean_token_accuracy": 0.47999184019863605,
+      "num_tokens": 146033406.0,
+      "step": 1008
+    },
+    {
+      "entropy": 2.474365234375,
+      "epoch": 0.01737214087101745,
+      "grad_norm": 0.5616711378097534,
+      "learning_rate": 9.999999515357654e-06,
+      "loss": 2.4179,
+      "mean_token_accuracy": 0.47331848414614797,
+      "num_tokens": 146185163.0,
+      "step": 1009
+    },
+    {
+      "entropy": 2.3839111328125,
+      "epoch": 0.01738935805721271,
+      "grad_norm": 0.5724861025810242,
+      "learning_rate": 9.999999386624534e-06,
+      "loss": 2.338,
+      "mean_token_accuracy": 0.4900155235081911,
+      "num_tokens": 146335979.0,
+      "step": 1010
+    },
+    {
+      "entropy": 2.397216796875,
+      "epoch": 0.01740657524340797,
+      "grad_norm": 0.566051185131073,
+      "learning_rate": 9.99999924274634e-06,
+      "loss": 2.3623,
+      "mean_token_accuracy": 0.49173881486058235,
+      "num_tokens": 146492297.0,
+      "step": 1011
+    },
+    {
+      "entropy": 2.425537109375,
+      "epoch": 0.01742379242960323,
+      "grad_norm": 0.5698406100273132,
+      "learning_rate": 9.999999083723077e-06,
+      "loss": 2.3803,
+      "mean_token_accuracy": 0.4822213868610561,
+      "num_tokens": 146638961.0,
+      "step": 1012
+    },
+    {
+      "entropy": 2.457763671875,
+      "epoch": 0.01744100961579849,
+      "grad_norm": 0.5576664805412292,
+      "learning_rate": 9.999998909554743e-06,
+      "loss": 2.4175,
+      "mean_token_accuracy": 0.47678716061636806,
+      "num_tokens": 146785199.0,
+      "step": 1013
+    },
+    {
+      "entropy": 2.45947265625,
+      "epoch": 0.01745822680199375,
+      "grad_norm": 0.5309212803840637,
+      "learning_rate": 9.99999872024134e-06,
+      "loss": 2.4368,
+      "mean_token_accuracy": 0.4784322748892009,
+      "num_tokens": 146940920.0,
+      "step": 1014
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.01747544398818901,
+      "grad_norm": 0.6436007022857666,
+      "learning_rate": 9.999998515782865e-06,
+      "loss": 2.3972,
+      "mean_token_accuracy": 0.4809753801673651,
+      "num_tokens": 147096034.0,
+      "step": 1015
+    },
+    {
+      "entropy": 2.4503173828125,
+      "epoch": 0.01749266117438427,
+      "grad_norm": 0.5794939398765564,
+      "learning_rate": 9.99999829617932e-06,
+      "loss": 2.4099,
+      "mean_token_accuracy": 0.47518764482811093,
+      "num_tokens": 147241716.0,
+      "step": 1016
+    },
+    {
+      "entropy": 2.4049072265625,
+      "epoch": 0.01750987836057953,
+      "grad_norm": 0.6116197109222412,
+      "learning_rate": 9.999998061430709e-06,
+      "loss": 2.3468,
+      "mean_token_accuracy": 0.4926842013373971,
+      "num_tokens": 147378764.0,
+      "step": 1017
+    },
+    {
+      "entropy": 2.47314453125,
+      "epoch": 0.01752709554677479,
+      "grad_norm": 0.5858633518218994,
+      "learning_rate": 9.99999781153703e-06,
+      "loss": 2.3979,
+      "mean_token_accuracy": 0.47995937149971724,
+      "num_tokens": 147517319.0,
+      "step": 1018
+    },
+    {
+      "entropy": 2.5513916015625,
+      "epoch": 0.017544312732970052,
+      "grad_norm": 0.6136896014213562,
+      "learning_rate": 9.999997546498284e-06,
+      "loss": 2.5054,
+      "mean_token_accuracy": 0.46979639306664467,
+      "num_tokens": 147657307.0,
+      "step": 1019
+    },
+    {
+      "entropy": 2.4195556640625,
+      "epoch": 0.01756152991916531,
+      "grad_norm": 0.593510091304779,
+      "learning_rate": 9.999997266314471e-06,
+      "loss": 2.3881,
+      "mean_token_accuracy": 0.48458396503701806,
+      "num_tokens": 147806252.0,
+      "step": 1020
+    },
+    {
+      "entropy": 2.41064453125,
+      "epoch": 0.01757874710536057,
+      "grad_norm": 0.675787091255188,
+      "learning_rate": 9.999996970985592e-06,
+      "loss": 2.3654,
+      "mean_token_accuracy": 0.48551547806710005,
+      "num_tokens": 147979427.0,
+      "step": 1021
+    },
+    {
+      "entropy": 2.46533203125,
+      "epoch": 0.01759596429155583,
+      "grad_norm": 0.6314930319786072,
+      "learning_rate": 9.99999666051165e-06,
+      "loss": 2.4691,
+      "mean_token_accuracy": 0.4804490995593369,
+      "num_tokens": 148114782.0,
+      "step": 1022
+    },
+    {
+      "entropy": 2.4417724609375,
+      "epoch": 0.01761318147775109,
+      "grad_norm": 0.579401969909668,
+      "learning_rate": 9.999996334892646e-06,
+      "loss": 2.4204,
+      "mean_token_accuracy": 0.482807548251003,
+      "num_tokens": 148259581.0,
+      "step": 1023
+    },
+    {
+      "entropy": 2.452880859375,
+      "epoch": 0.01763039866394635,
+      "grad_norm": 0.5902701616287231,
+      "learning_rate": 9.999995994128578e-06,
+      "loss": 2.4434,
+      "mean_token_accuracy": 0.47913903277367353,
+      "num_tokens": 148400114.0,
+      "step": 1024
+    },
+    {
+      "entropy": 2.482666015625,
+      "epoch": 0.017647615850141612,
+      "grad_norm": 0.5950527787208557,
+      "learning_rate": 9.999995638219448e-06,
+      "loss": 2.4566,
+      "mean_token_accuracy": 0.4763036798685789,
+      "num_tokens": 148540440.0,
+      "step": 1025
+    },
+    {
+      "entropy": 2.446533203125,
+      "epoch": 0.017664833036336872,
+      "grad_norm": 0.6330975294113159,
+      "learning_rate": 9.999995267165256e-06,
+      "loss": 2.3904,
+      "mean_token_accuracy": 0.4806701922789216,
+      "num_tokens": 148692789.0,
+      "step": 1026
+    },
+    {
+      "entropy": 2.427978515625,
+      "epoch": 0.017682050222532133,
+      "grad_norm": 0.617744505405426,
+      "learning_rate": 9.999994880966008e-06,
+      "loss": 2.369,
+      "mean_token_accuracy": 0.4830102613195777,
+      "num_tokens": 148836617.0,
+      "step": 1027
+    },
+    {
+      "entropy": 2.42822265625,
+      "epoch": 0.017699267408727393,
+      "grad_norm": 0.6210005879402161,
+      "learning_rate": 9.999994479621703e-06,
+      "loss": 2.3569,
+      "mean_token_accuracy": 0.49285168340429664,
+      "num_tokens": 148982058.0,
+      "step": 1028
+    },
+    {
+      "entropy": 2.52392578125,
+      "epoch": 0.01771648459492265,
+      "grad_norm": 0.5944798588752747,
+      "learning_rate": 9.999994063132336e-06,
+      "loss": 2.4881,
+      "mean_token_accuracy": 0.46862922469154,
+      "num_tokens": 149117257.0,
+      "step": 1029
+    },
+    {
+      "entropy": 2.4248046875,
+      "epoch": 0.01773370178111791,
+      "grad_norm": 0.603406548500061,
+      "learning_rate": 9.999993631497918e-06,
+      "loss": 2.3968,
+      "mean_token_accuracy": 0.4805483613163233,
+      "num_tokens": 149253194.0,
+      "step": 1030
+    },
+    {
+      "entropy": 2.48681640625,
+      "epoch": 0.01775091896731317,
+      "grad_norm": 0.5487124919891357,
+      "learning_rate": 9.999993184718445e-06,
+      "loss": 2.4613,
+      "mean_token_accuracy": 0.4695818484760821,
+      "num_tokens": 149418815.0,
+      "step": 1031
+    },
+    {
+      "entropy": 2.3785400390625,
+      "epoch": 0.017768136153508432,
+      "grad_norm": 0.5934000015258789,
+      "learning_rate": 9.999992722793916e-06,
+      "loss": 2.3197,
+      "mean_token_accuracy": 0.4929351657629013,
+      "num_tokens": 149565302.0,
+      "step": 1032
+    },
+    {
+      "entropy": 2.41943359375,
+      "epoch": 0.017785353339703693,
+      "grad_norm": 0.5832322835922241,
+      "learning_rate": 9.999992245724338e-06,
+      "loss": 2.3851,
+      "mean_token_accuracy": 0.4843961731530726,
+      "num_tokens": 149701523.0,
+      "step": 1033
+    },
+    {
+      "entropy": 2.477294921875,
+      "epoch": 0.017802570525898953,
+      "grad_norm": 0.578316330909729,
+      "learning_rate": 9.99999175350971e-06,
+      "loss": 2.4705,
+      "mean_token_accuracy": 0.4734199936501682,
+      "num_tokens": 149841173.0,
+      "step": 1034
+    },
+    {
+      "entropy": 2.496337890625,
+      "epoch": 0.017819787712094214,
+      "grad_norm": 1.1719509363174438,
+      "learning_rate": 9.999991246150034e-06,
+      "loss": 2.469,
+      "mean_token_accuracy": 0.4775820942595601,
+      "num_tokens": 149977715.0,
+      "step": 1035
+    },
+    {
+      "entropy": 2.458251953125,
+      "epoch": 0.017837004898289474,
+      "grad_norm": 0.5495033860206604,
+      "learning_rate": 9.999990723645309e-06,
+      "loss": 2.4407,
+      "mean_token_accuracy": 0.47903554420918226,
+      "num_tokens": 150124141.0,
+      "step": 1036
+    },
+    {
+      "entropy": 2.453857421875,
+      "epoch": 0.01785422208448473,
+      "grad_norm": 0.6319074034690857,
+      "learning_rate": 9.99999018599554e-06,
+      "loss": 2.4007,
+      "mean_token_accuracy": 0.4762126957066357,
+      "num_tokens": 150278398.0,
+      "step": 1037
+    },
+    {
+      "entropy": 2.435791015625,
+      "epoch": 0.01787143927067999,
+      "grad_norm": 0.6012912392616272,
+      "learning_rate": 9.999989633200726e-06,
+      "loss": 2.3693,
+      "mean_token_accuracy": 0.490336746443063,
+      "num_tokens": 150415689.0,
+      "step": 1038
+    },
+    {
+      "entropy": 2.431640625,
+      "epoch": 0.017888656456875252,
+      "grad_norm": 0.5937034487724304,
+      "learning_rate": 9.99998906526087e-06,
+      "loss": 2.381,
+      "mean_token_accuracy": 0.48564466880634427,
+      "num_tokens": 150565533.0,
+      "step": 1039
+    },
+    {
+      "entropy": 2.541259765625,
+      "epoch": 0.017905873643070513,
+      "grad_norm": 0.5868061780929565,
+      "learning_rate": 9.999988482175975e-06,
+      "loss": 2.485,
+      "mean_token_accuracy": 0.4694917304441333,
+      "num_tokens": 150703215.0,
+      "step": 1040
+    },
+    {
+      "entropy": 2.51025390625,
+      "epoch": 0.017923090829265773,
+      "grad_norm": 0.5923919677734375,
+      "learning_rate": 9.99998788394604e-06,
+      "loss": 2.5,
+      "mean_token_accuracy": 0.47092792578041553,
+      "num_tokens": 150846937.0,
+      "step": 1041
+    },
+    {
+      "entropy": 2.44384765625,
+      "epoch": 0.017940308015461034,
+      "grad_norm": 0.6263927817344666,
+      "learning_rate": 9.999987270571067e-06,
+      "loss": 2.4557,
+      "mean_token_accuracy": 0.4760148832574487,
+      "num_tokens": 150988381.0,
+      "step": 1042
+    },
+    {
+      "entropy": 2.388427734375,
+      "epoch": 0.017957525201656294,
+      "grad_norm": 0.5355521440505981,
+      "learning_rate": 9.999986642051061e-06,
+      "loss": 2.3634,
+      "mean_token_accuracy": 0.4859662549570203,
+      "num_tokens": 151137551.0,
+      "step": 1043
+    },
+    {
+      "entropy": 2.41552734375,
+      "epoch": 0.017974742387851555,
+      "grad_norm": 0.615512490272522,
+      "learning_rate": 9.999985998386022e-06,
+      "loss": 2.4254,
+      "mean_token_accuracy": 0.4817372509278357,
+      "num_tokens": 151280397.0,
+      "step": 1044
+    },
+    {
+      "entropy": 2.496826171875,
+      "epoch": 0.017991959574046812,
+      "grad_norm": 0.543178379535675,
+      "learning_rate": 9.99998533957595e-06,
+      "loss": 2.4526,
+      "mean_token_accuracy": 0.4679850875400007,
+      "num_tokens": 151423217.0,
+      "step": 1045
+    },
+    {
+      "entropy": 2.487548828125,
+      "epoch": 0.018009176760242072,
+      "grad_norm": 0.6349004507064819,
+      "learning_rate": 9.999984665620852e-06,
+      "loss": 2.4205,
+      "mean_token_accuracy": 0.4785635150037706,
+      "num_tokens": 151566681.0,
+      "step": 1046
+    },
+    {
+      "entropy": 2.467041015625,
+      "epoch": 0.018026393946437333,
+      "grad_norm": 0.5757213234901428,
+      "learning_rate": 9.999983976520725e-06,
+      "loss": 2.3992,
+      "mean_token_accuracy": 0.48239869019016623,
+      "num_tokens": 151722733.0,
+      "step": 1047
+    },
+    {
+      "entropy": 2.3770751953125,
+      "epoch": 0.018043611132632593,
+      "grad_norm": 0.5888438820838928,
+      "learning_rate": 9.999983272275572e-06,
+      "loss": 2.3367,
+      "mean_token_accuracy": 0.488128668628633,
+      "num_tokens": 151878057.0,
+      "step": 1048
+    },
+    {
+      "entropy": 2.40380859375,
+      "epoch": 0.018060828318827854,
+      "grad_norm": 0.6084235906600952,
+      "learning_rate": 9.999982552885396e-06,
+      "loss": 2.4194,
+      "mean_token_accuracy": 0.48174417577683926,
+      "num_tokens": 152013813.0,
+      "step": 1049
+    },
+    {
+      "entropy": 2.525390625,
+      "epoch": 0.018078045505023115,
+      "grad_norm": 0.6073904633522034,
+      "learning_rate": 9.999981818350201e-06,
+      "loss": 2.4913,
+      "mean_token_accuracy": 0.470255627296865,
+      "num_tokens": 152145871.0,
+      "step": 1050
+    },
+    {
+      "entropy": 2.4578857421875,
+      "epoch": 0.018095262691218375,
+      "grad_norm": 0.5444018840789795,
+      "learning_rate": 9.999981068669986e-06,
+      "loss": 2.4527,
+      "mean_token_accuracy": 0.47973277885466814,
+      "num_tokens": 152304784.0,
+      "step": 1051
+    },
+    {
+      "entropy": 2.47509765625,
+      "epoch": 0.018112479877413636,
+      "grad_norm": 0.5837482213973999,
+      "learning_rate": 9.999980303844756e-06,
+      "loss": 2.4489,
+      "mean_token_accuracy": 0.47514383727684617,
+      "num_tokens": 152453018.0,
+      "step": 1052
+    },
+    {
+      "entropy": 2.4283447265625,
+      "epoch": 0.018129697063608896,
+      "grad_norm": 2.452045202255249,
+      "learning_rate": 9.999979523874513e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.48897374235093594,
+      "num_tokens": 152612683.0,
+      "step": 1053
+    },
+    {
+      "entropy": 2.404541015625,
+      "epoch": 0.018146914249804153,
+      "grad_norm": 0.5417919754981995,
+      "learning_rate": 9.999978728759256e-06,
+      "loss": 2.3388,
+      "mean_token_accuracy": 0.4870688715018332,
+      "num_tokens": 152762264.0,
+      "step": 1054
+    },
+    {
+      "entropy": 2.4012451171875,
+      "epoch": 0.018164131435999414,
+      "grad_norm": 0.5944674611091614,
+      "learning_rate": 9.999977918498992e-06,
+      "loss": 2.3714,
+      "mean_token_accuracy": 0.4950359701178968,
+      "num_tokens": 152906647.0,
+      "step": 1055
+    },
+    {
+      "entropy": 2.453857421875,
+      "epoch": 0.018181348622194674,
+      "grad_norm": 0.5601978302001953,
+      "learning_rate": 9.99997709309372e-06,
+      "loss": 2.3945,
+      "mean_token_accuracy": 0.4819331760518253,
+      "num_tokens": 153047278.0,
+      "step": 1056
+    },
+    {
+      "entropy": 2.5194091796875,
+      "epoch": 0.018198565808389935,
+      "grad_norm": 0.6144431233406067,
+      "learning_rate": 9.999976252543444e-06,
+      "loss": 2.4835,
+      "mean_token_accuracy": 0.4768884154036641,
+      "num_tokens": 153186932.0,
+      "step": 1057
+    },
+    {
+      "entropy": 2.403564453125,
+      "epoch": 0.018215782994585195,
+      "grad_norm": 0.9560090899467468,
+      "learning_rate": 9.999975396848167e-06,
+      "loss": 2.3291,
+      "mean_token_accuracy": 0.4874680512584746,
+      "num_tokens": 153341046.0,
+      "step": 1058
+    },
+    {
+      "entropy": 2.431884765625,
+      "epoch": 0.018233000180780456,
+      "grad_norm": 0.6547354459762573,
+      "learning_rate": 9.99997452600789e-06,
+      "loss": 2.4077,
+      "mean_token_accuracy": 0.48102515283972025,
+      "num_tokens": 153486117.0,
+      "step": 1059
+    },
+    {
+      "entropy": 2.381103515625,
+      "epoch": 0.018250217366975716,
+      "grad_norm": 0.5701848268508911,
+      "learning_rate": 9.999973640022618e-06,
+      "loss": 2.3703,
+      "mean_token_accuracy": 0.4909746255725622,
+      "num_tokens": 153635642.0,
+      "step": 1060
+    },
+    {
+      "entropy": 2.427978515625,
+      "epoch": 0.018267434553170977,
+      "grad_norm": 0.5640613436698914,
+      "learning_rate": 9.99997273889235e-06,
+      "loss": 2.3833,
+      "mean_token_accuracy": 0.483435302041471,
+      "num_tokens": 153778267.0,
+      "step": 1061
+    },
+    {
+      "entropy": 2.54736328125,
+      "epoch": 0.018284651739366234,
+      "grad_norm": 0.5851112008094788,
+      "learning_rate": 9.999971822617094e-06,
+      "loss": 2.5209,
+      "mean_token_accuracy": 0.4660449046641588,
+      "num_tokens": 153920352.0,
+      "step": 1062
+    },
+    {
+      "entropy": 2.47998046875,
+      "epoch": 0.018301868925561494,
+      "grad_norm": 0.5928635001182556,
+      "learning_rate": 9.999970891196847e-06,
+      "loss": 2.4422,
+      "mean_token_accuracy": 0.47202530410140753,
+      "num_tokens": 154064892.0,
+      "step": 1063
+    },
+    {
+      "entropy": 2.507080078125,
+      "epoch": 0.018319086111756755,
+      "grad_norm": 0.6197124719619751,
+      "learning_rate": 9.999969944631615e-06,
+      "loss": 2.4772,
+      "mean_token_accuracy": 0.4731879197061062,
+      "num_tokens": 154215041.0,
+      "step": 1064
+    },
+    {
+      "entropy": 2.46923828125,
+      "epoch": 0.018336303297952015,
+      "grad_norm": 0.5558657646179199,
+      "learning_rate": 9.999968982921403e-06,
+      "loss": 2.4352,
+      "mean_token_accuracy": 0.47735925391316414,
+      "num_tokens": 154367730.0,
+      "step": 1065
+    },
+    {
+      "entropy": 2.430419921875,
+      "epoch": 0.018353520484147276,
+      "grad_norm": 0.6059717535972595,
+      "learning_rate": 9.99996800606621e-06,
+      "loss": 2.3824,
+      "mean_token_accuracy": 0.4844053997658193,
+      "num_tokens": 154505107.0,
+      "step": 1066
+    },
+    {
+      "entropy": 2.4879150390625,
+      "epoch": 0.018370737670342537,
+      "grad_norm": 0.5660464763641357,
+      "learning_rate": 9.999967014066038e-06,
+      "loss": 2.4352,
+      "mean_token_accuracy": 0.476345878560096,
+      "num_tokens": 154643818.0,
+      "step": 1067
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.018387954856537797,
+      "grad_norm": 0.701174795627594,
+      "learning_rate": 9.999966006920896e-06,
+      "loss": 2.416,
+      "mean_token_accuracy": 0.48350341338664293,
+      "num_tokens": 154785623.0,
+      "step": 1068
+    },
+    {
+      "entropy": 2.48486328125,
+      "epoch": 0.018405172042733058,
+      "grad_norm": 0.6571546196937561,
+      "learning_rate": 9.99996498463078e-06,
+      "loss": 2.4509,
+      "mean_token_accuracy": 0.47686226665973663,
+      "num_tokens": 154947202.0,
+      "step": 1069
+    },
+    {
+      "entropy": 2.4830322265625,
+      "epoch": 0.018422389228928315,
+      "grad_norm": 0.6001805067062378,
+      "learning_rate": 9.999963947195699e-06,
+      "loss": 2.4316,
+      "mean_token_accuracy": 0.47862283093854785,
+      "num_tokens": 155097645.0,
+      "step": 1070
+    },
+    {
+      "entropy": 2.4422607421875,
+      "epoch": 0.018439606415123575,
+      "grad_norm": 0.6199787259101868,
+      "learning_rate": 9.999962894615653e-06,
+      "loss": 2.4132,
+      "mean_token_accuracy": 0.48402009485289454,
+      "num_tokens": 155229538.0,
+      "step": 1071
+    },
+    {
+      "entropy": 2.408203125,
+      "epoch": 0.018456823601318836,
+      "grad_norm": 0.5521693825721741,
+      "learning_rate": 9.999961826890645e-06,
+      "loss": 2.3365,
+      "mean_token_accuracy": 0.4847987382672727,
+      "num_tokens": 155376732.0,
+      "step": 1072
+    },
+    {
+      "entropy": 2.4617919921875,
+      "epoch": 0.018474040787514096,
+      "grad_norm": 0.6697084903717041,
+      "learning_rate": 9.999960744020681e-06,
+      "loss": 2.4263,
+      "mean_token_accuracy": 0.48609263403341174,
+      "num_tokens": 155514355.0,
+      "step": 1073
+    },
+    {
+      "entropy": 2.443603515625,
+      "epoch": 0.018491257973709357,
+      "grad_norm": 0.6446971297264099,
+      "learning_rate": 9.999959646005761e-06,
+      "loss": 2.448,
+      "mean_token_accuracy": 0.4798265271820128,
+      "num_tokens": 155652470.0,
+      "step": 1074
+    },
+    {
+      "entropy": 2.47412109375,
+      "epoch": 0.018508475159904617,
+      "grad_norm": 0.597260057926178,
+      "learning_rate": 9.999958532845889e-06,
+      "loss": 2.4575,
+      "mean_token_accuracy": 0.4776256578043103,
+      "num_tokens": 155809404.0,
+      "step": 1075
+    },
+    {
+      "entropy": 2.388916015625,
+      "epoch": 0.018525692346099878,
+      "grad_norm": 0.5906361937522888,
+      "learning_rate": 9.99995740454107e-06,
+      "loss": 2.3603,
+      "mean_token_accuracy": 0.4867819882929325,
+      "num_tokens": 155949820.0,
+      "step": 1076
+    },
+    {
+      "entropy": 2.4091796875,
+      "epoch": 0.01854290953229514,
+      "grad_norm": 0.5962374806404114,
+      "learning_rate": 9.999956261091306e-06,
+      "loss": 2.3754,
+      "mean_token_accuracy": 0.486242544837296,
+      "num_tokens": 156083865.0,
+      "step": 1077
+    },
+    {
+      "entropy": 2.466064453125,
+      "epoch": 0.0185601267184904,
+      "grad_norm": 0.5743368864059448,
+      "learning_rate": 9.999955102496602e-06,
+      "loss": 2.4522,
+      "mean_token_accuracy": 0.4817331531085074,
+      "num_tokens": 156230750.0,
+      "step": 1078
+    },
+    {
+      "entropy": 2.4454345703125,
+      "epoch": 0.018577343904685656,
+      "grad_norm": 0.5727860927581787,
+      "learning_rate": 9.999953928756958e-06,
+      "loss": 2.424,
+      "mean_token_accuracy": 0.47826589411124587,
+      "num_tokens": 156377631.0,
+      "step": 1079
+    },
+    {
+      "entropy": 2.433837890625,
+      "epoch": 0.018594561090880916,
+      "grad_norm": 0.6403164863586426,
+      "learning_rate": 9.999952739872383e-06,
+      "loss": 2.379,
+      "mean_token_accuracy": 0.4838769477792084,
+      "num_tokens": 156535199.0,
+      "step": 1080
+    },
+    {
+      "entropy": 2.504150390625,
+      "epoch": 0.018611778277076177,
+      "grad_norm": 0.5877081751823425,
+      "learning_rate": 9.999951535842875e-06,
+      "loss": 2.4479,
+      "mean_token_accuracy": 0.4734333767555654,
+      "num_tokens": 156667471.0,
+      "step": 1081
+    },
+    {
+      "entropy": 2.421630859375,
+      "epoch": 0.018628995463271437,
+      "grad_norm": 0.5537156462669373,
+      "learning_rate": 9.999950316668443e-06,
+      "loss": 2.3707,
+      "mean_token_accuracy": 0.48230436397716403,
+      "num_tokens": 156819504.0,
+      "step": 1082
+    },
+    {
+      "entropy": 2.5322265625,
+      "epoch": 0.018646212649466698,
+      "grad_norm": 0.603832483291626,
+      "learning_rate": 9.999949082349085e-06,
+      "loss": 2.4499,
+      "mean_token_accuracy": 0.4676863308995962,
+      "num_tokens": 156944926.0,
+      "step": 1083
+    },
+    {
+      "entropy": 2.473388671875,
+      "epoch": 0.01866342983566196,
+      "grad_norm": 0.5896692872047424,
+      "learning_rate": 9.99994783288481e-06,
+      "loss": 2.4796,
+      "mean_token_accuracy": 0.4778019469231367,
+      "num_tokens": 157090650.0,
+      "step": 1084
+    },
+    {
+      "entropy": 2.4200439453125,
+      "epoch": 0.01868064702185722,
+      "grad_norm": 0.5518651008605957,
+      "learning_rate": 9.999946568275619e-06,
+      "loss": 2.4286,
+      "mean_token_accuracy": 0.47666720813140273,
+      "num_tokens": 157252639.0,
+      "step": 1085
+    },
+    {
+      "entropy": 2.4140625,
+      "epoch": 0.01869786420805248,
+      "grad_norm": 0.5450928211212158,
+      "learning_rate": 9.999945288521516e-06,
+      "loss": 2.3944,
+      "mean_token_accuracy": 0.48317190259695053,
+      "num_tokens": 157409995.0,
+      "step": 1086
+    },
+    {
+      "entropy": 2.4267578125,
+      "epoch": 0.018715081394247737,
+      "grad_norm": 0.5875595211982727,
+      "learning_rate": 9.999943993622504e-06,
+      "loss": 2.4209,
+      "mean_token_accuracy": 0.4857109053991735,
+      "num_tokens": 157551609.0,
+      "step": 1087
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.018732298580442997,
+      "grad_norm": 0.5621595978736877,
+      "learning_rate": 9.99994268357859e-06,
+      "loss": 2.4107,
+      "mean_token_accuracy": 0.4831027192994952,
+      "num_tokens": 157699503.0,
+      "step": 1088
+    },
+    {
+      "entropy": 2.497802734375,
+      "epoch": 0.018749515766638258,
+      "grad_norm": 0.6029961109161377,
+      "learning_rate": 9.999941358389775e-06,
+      "loss": 2.4699,
+      "mean_token_accuracy": 0.4773255423642695,
+      "num_tokens": 157839741.0,
+      "step": 1089
+    },
+    {
+      "entropy": 2.4298095703125,
+      "epoch": 0.018766732952833518,
+      "grad_norm": 0.5884754061698914,
+      "learning_rate": 9.999940018056062e-06,
+      "loss": 2.4153,
+      "mean_token_accuracy": 0.4817211125046015,
+      "num_tokens": 157988371.0,
+      "step": 1090
+    },
+    {
+      "entropy": 2.417724609375,
+      "epoch": 0.01878395013902878,
+      "grad_norm": 0.5795974135398865,
+      "learning_rate": 9.99993866257746e-06,
+      "loss": 2.3917,
+      "mean_token_accuracy": 0.4811216425150633,
+      "num_tokens": 158133019.0,
+      "step": 1091
+    },
+    {
+      "entropy": 2.404541015625,
+      "epoch": 0.01880116732522404,
+      "grad_norm": 0.5938680768013,
+      "learning_rate": 9.99993729195397e-06,
+      "loss": 2.3589,
+      "mean_token_accuracy": 0.4899957957677543,
+      "num_tokens": 158276428.0,
+      "step": 1092
+    },
+    {
+      "entropy": 2.453857421875,
+      "epoch": 0.0188183845114193,
+      "grad_norm": 0.6011793613433838,
+      "learning_rate": 9.999935906185596e-06,
+      "loss": 2.3976,
+      "mean_token_accuracy": 0.48242352809756994,
+      "num_tokens": 158418256.0,
+      "step": 1093
+    },
+    {
+      "entropy": 2.4844970703125,
+      "epoch": 0.01883560169761456,
+      "grad_norm": 0.6211831569671631,
+      "learning_rate": 9.999934505272341e-06,
+      "loss": 2.4357,
+      "mean_token_accuracy": 0.47488651471212506,
+      "num_tokens": 158538103.0,
+      "step": 1094
+    },
+    {
+      "entropy": 2.4769287109375,
+      "epoch": 0.018852818883809817,
+      "grad_norm": 0.5956041812896729,
+      "learning_rate": 9.999933089214214e-06,
+      "loss": 2.4364,
+      "mean_token_accuracy": 0.4817122952081263,
+      "num_tokens": 158684837.0,
+      "step": 1095
+    },
+    {
+      "entropy": 2.4451904296875,
+      "epoch": 0.018870036070005078,
+      "grad_norm": 0.5776230692863464,
+      "learning_rate": 9.999931658011213e-06,
+      "loss": 2.4465,
+      "mean_token_accuracy": 0.48071191366761923,
+      "num_tokens": 158841114.0,
+      "step": 1096
+    },
+    {
+      "entropy": 2.4998779296875,
+      "epoch": 0.01888725325620034,
+      "grad_norm": 0.5656998753547668,
+      "learning_rate": 9.999930211663346e-06,
+      "loss": 2.4445,
+      "mean_token_accuracy": 0.47156142350286245,
+      "num_tokens": 158982072.0,
+      "step": 1097
+    },
+    {
+      "entropy": 2.410400390625,
+      "epoch": 0.0189044704423956,
+      "grad_norm": 0.5946214199066162,
+      "learning_rate": 9.999928750170619e-06,
+      "loss": 2.3455,
+      "mean_token_accuracy": 0.4895977326668799,
+      "num_tokens": 159115070.0,
+      "step": 1098
+    },
+    {
+      "entropy": 2.4327392578125,
+      "epoch": 0.01892168762859086,
+      "grad_norm": 0.594409167766571,
+      "learning_rate": 9.999927273533032e-06,
+      "loss": 2.3842,
+      "mean_token_accuracy": 0.48081815196201205,
+      "num_tokens": 159253866.0,
+      "step": 1099
+    },
+    {
+      "entropy": 2.450927734375,
+      "epoch": 0.01893890481478612,
+      "grad_norm": 0.5916756987571716,
+      "learning_rate": 9.999925781750594e-06,
+      "loss": 2.3838,
+      "mean_token_accuracy": 0.4844925357028842,
+      "num_tokens": 159415802.0,
+      "step": 1100
+    },
+    {
+      "entropy": 2.5107421875,
+      "epoch": 0.01895612200098138,
+      "grad_norm": 0.6757313013076782,
+      "learning_rate": 9.999924274823305e-06,
+      "loss": 2.4928,
+      "mean_token_accuracy": 0.4722679229453206,
+      "num_tokens": 159567360.0,
+      "step": 1101
+    },
+    {
+      "entropy": 2.44482421875,
+      "epoch": 0.01897333918717664,
+      "grad_norm": 0.5762059092521667,
+      "learning_rate": 9.999922752751173e-06,
+      "loss": 2.4003,
+      "mean_token_accuracy": 0.4811086105182767,
+      "num_tokens": 159709631.0,
+      "step": 1102
+    },
+    {
+      "entropy": 2.425537109375,
+      "epoch": 0.0189905563733719,
+      "grad_norm": 0.5835311412811279,
+      "learning_rate": 9.999921215534201e-06,
+      "loss": 2.377,
+      "mean_token_accuracy": 0.4846574803814292,
+      "num_tokens": 159847869.0,
+      "step": 1103
+    },
+    {
+      "entropy": 2.3896484375,
+      "epoch": 0.01900777355956716,
+      "grad_norm": 0.610913872718811,
+      "learning_rate": 9.999919663172394e-06,
+      "loss": 2.3431,
+      "mean_token_accuracy": 0.4912629318423569,
+      "num_tokens": 159976268.0,
+      "step": 1104
+    },
+    {
+      "entropy": 2.4576416015625,
+      "epoch": 0.01902499074576242,
+      "grad_norm": 0.6417136192321777,
+      "learning_rate": 9.999918095665758e-06,
+      "loss": 2.4618,
+      "mean_token_accuracy": 0.4749236977659166,
+      "num_tokens": 160114362.0,
+      "step": 1105
+    },
+    {
+      "entropy": 2.4779052734375,
+      "epoch": 0.01904220793195768,
+      "grad_norm": 0.5750072598457336,
+      "learning_rate": 9.999916513014294e-06,
+      "loss": 2.4414,
+      "mean_token_accuracy": 0.4833888350985944,
+      "num_tokens": 160272635.0,
+      "step": 1106
+    },
+    {
+      "entropy": 2.4361572265625,
+      "epoch": 0.01905942511815294,
+      "grad_norm": 0.6341116428375244,
+      "learning_rate": 9.999914915218012e-06,
+      "loss": 2.3878,
+      "mean_token_accuracy": 0.47810999071225524,
+      "num_tokens": 160430906.0,
+      "step": 1107
+    },
+    {
+      "entropy": 2.3680419921875,
+      "epoch": 0.0190766423043482,
+      "grad_norm": 0.575920581817627,
+      "learning_rate": 9.999913302276912e-06,
+      "loss": 2.3544,
+      "mean_token_accuracy": 0.4951640688814223,
+      "num_tokens": 160569906.0,
+      "step": 1108
+    },
+    {
+      "entropy": 2.431884765625,
+      "epoch": 0.01909385949054346,
+      "grad_norm": 0.5743590593338013,
+      "learning_rate": 9.999911674191001e-06,
+      "loss": 2.382,
+      "mean_token_accuracy": 0.48963714949786663,
+      "num_tokens": 160710393.0,
+      "step": 1109
+    },
+    {
+      "entropy": 2.5234375,
+      "epoch": 0.019111076676738722,
+      "grad_norm": 0.6044170260429382,
+      "learning_rate": 9.999910030960285e-06,
+      "loss": 2.467,
+      "mean_token_accuracy": 0.4708155104890466,
+      "num_tokens": 160837634.0,
+      "step": 1110
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.019128293862933982,
+      "grad_norm": 0.641613245010376,
+      "learning_rate": 9.99990837258477e-06,
+      "loss": 2.4,
+      "mean_token_accuracy": 0.474128358066082,
+      "num_tokens": 160981820.0,
+      "step": 1111
+    },
+    {
+      "entropy": 2.4156494140625,
+      "epoch": 0.01914551104912924,
+      "grad_norm": 0.7928540110588074,
+      "learning_rate": 9.999906699064455e-06,
+      "loss": 2.369,
+      "mean_token_accuracy": 0.4871662328951061,
+      "num_tokens": 161128688.0,
+      "step": 1112
+    },
+    {
+      "entropy": 2.45849609375,
+      "epoch": 0.0191627282353245,
+      "grad_norm": 0.6063829064369202,
+      "learning_rate": 9.999905010399351e-06,
+      "loss": 2.3988,
+      "mean_token_accuracy": 0.4813544964417815,
+      "num_tokens": 161265040.0,
+      "step": 1113
+    },
+    {
+      "entropy": 2.40625,
+      "epoch": 0.01917994542151976,
+      "grad_norm": 0.5939781069755554,
+      "learning_rate": 9.999903306589463e-06,
+      "loss": 2.3918,
+      "mean_token_accuracy": 0.48077729996293783,
+      "num_tokens": 161418528.0,
+      "step": 1114
+    },
+    {
+      "entropy": 2.444580078125,
+      "epoch": 0.01919716260771502,
+      "grad_norm": 0.6092802882194519,
+      "learning_rate": 9.99990158763479e-06,
+      "loss": 2.4441,
+      "mean_token_accuracy": 0.4797166772186756,
+      "num_tokens": 161557877.0,
+      "step": 1115
+    },
+    {
+      "entropy": 2.457763671875,
+      "epoch": 0.01921437979391028,
+      "grad_norm": 0.6434066295623779,
+      "learning_rate": 9.999899853535344e-06,
+      "loss": 2.4275,
+      "mean_token_accuracy": 0.48657548474147916,
+      "num_tokens": 161698007.0,
+      "step": 1116
+    },
+    {
+      "entropy": 2.4796142578125,
+      "epoch": 0.019231596980105542,
+      "grad_norm": 0.5764423608779907,
+      "learning_rate": 9.999898104291128e-06,
+      "loss": 2.4679,
+      "mean_token_accuracy": 0.4779613367281854,
+      "num_tokens": 161848916.0,
+      "step": 1117
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.019248814166300802,
+      "grad_norm": 0.8059520721435547,
+      "learning_rate": 9.999896339902148e-06,
+      "loss": 2.4523,
+      "mean_token_accuracy": 0.4732037871144712,
+      "num_tokens": 162016077.0,
+      "step": 1118
+    },
+    {
+      "entropy": 2.46240234375,
+      "epoch": 0.019266031352496063,
+      "grad_norm": 0.5723803639411926,
+      "learning_rate": 9.999894560368406e-06,
+      "loss": 2.4095,
+      "mean_token_accuracy": 0.48319359589368105,
+      "num_tokens": 162166218.0,
+      "step": 1119
+    },
+    {
+      "entropy": 2.4815673828125,
+      "epoch": 0.01928324853869132,
+      "grad_norm": 0.5882191061973572,
+      "learning_rate": 9.999892765689912e-06,
+      "loss": 2.4345,
+      "mean_token_accuracy": 0.47986175399273634,
+      "num_tokens": 162305491.0,
+      "step": 1120
+    },
+    {
+      "entropy": 2.4931640625,
+      "epoch": 0.01930046572488658,
+      "grad_norm": 0.6278371214866638,
+      "learning_rate": 9.999890955866667e-06,
+      "loss": 2.3912,
+      "mean_token_accuracy": 0.47920875577256083,
+      "num_tokens": 162426796.0,
+      "step": 1121
+    },
+    {
+      "entropy": 2.485595703125,
+      "epoch": 0.01931768291108184,
+      "grad_norm": 0.6226006746292114,
+      "learning_rate": 9.999889130898682e-06,
+      "loss": 2.4591,
+      "mean_token_accuracy": 0.47889810614287853,
+      "num_tokens": 162575000.0,
+      "step": 1122
+    },
+    {
+      "entropy": 2.508544921875,
+      "epoch": 0.0193349000972771,
+      "grad_norm": 0.675370454788208,
+      "learning_rate": 9.999887290785957e-06,
+      "loss": 2.4308,
+      "mean_token_accuracy": 0.47410700796172023,
+      "num_tokens": 162713512.0,
+      "step": 1123
+    },
+    {
+      "entropy": 2.420166015625,
+      "epoch": 0.019352117283472362,
+      "grad_norm": 0.5244258642196655,
+      "learning_rate": 9.9998854355285e-06,
+      "loss": 2.3653,
+      "mean_token_accuracy": 0.48809623531997204,
+      "num_tokens": 162866946.0,
+      "step": 1124
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.019369334469667623,
+      "grad_norm": 0.5886077284812927,
+      "learning_rate": 9.999883565126316e-06,
+      "loss": 2.381,
+      "mean_token_accuracy": 0.4897596640512347,
+      "num_tokens": 163012916.0,
+      "step": 1125
+    },
+    {
+      "entropy": 2.4671630859375,
+      "epoch": 0.019386551655862883,
+      "grad_norm": 0.6227119565010071,
+      "learning_rate": 9.999881679579411e-06,
+      "loss": 2.4095,
+      "mean_token_accuracy": 0.4798518419265747,
+      "num_tokens": 163152736.0,
+      "step": 1126
+    },
+    {
+      "entropy": 2.4190673828125,
+      "epoch": 0.019403768842058144,
+      "grad_norm": 0.6482838988304138,
+      "learning_rate": 9.999879778887793e-06,
+      "loss": 2.3961,
+      "mean_token_accuracy": 0.48625363502651453,
+      "num_tokens": 163284418.0,
+      "step": 1127
+    },
+    {
+      "entropy": 2.404296875,
+      "epoch": 0.0194209860282534,
+      "grad_norm": 0.7466304302215576,
+      "learning_rate": 9.999877863051464e-06,
+      "loss": 2.3752,
+      "mean_token_accuracy": 0.4825763679109514,
+      "num_tokens": 163435314.0,
+      "step": 1128
+    },
+    {
+      "entropy": 2.39111328125,
+      "epoch": 0.01943820321444866,
+      "grad_norm": 0.5774152278900146,
+      "learning_rate": 9.999875932070431e-06,
+      "loss": 2.3536,
+      "mean_token_accuracy": 0.4831749680452049,
+      "num_tokens": 163590841.0,
+      "step": 1129
+    },
+    {
+      "entropy": 2.47607421875,
+      "epoch": 0.019455420400643922,
+      "grad_norm": 0.6369538307189941,
+      "learning_rate": 9.9998739859447e-06,
+      "loss": 2.4949,
+      "mean_token_accuracy": 0.47335345949977636,
+      "num_tokens": 163724787.0,
+      "step": 1130
+    },
+    {
+      "entropy": 2.528076171875,
+      "epoch": 0.019472637586839182,
+      "grad_norm": 0.5823352932929993,
+      "learning_rate": 9.999872024674277e-06,
+      "loss": 2.5034,
+      "mean_token_accuracy": 0.4663214525207877,
+      "num_tokens": 163854370.0,
+      "step": 1131
+    },
+    {
+      "entropy": 2.518310546875,
+      "epoch": 0.019489854773034443,
+      "grad_norm": 0.6219056844711304,
+      "learning_rate": 9.99987004825917e-06,
+      "loss": 2.5005,
+      "mean_token_accuracy": 0.47132935794070363,
+      "num_tokens": 163999338.0,
+      "step": 1132
+    },
+    {
+      "entropy": 2.46630859375,
+      "epoch": 0.019507071959229703,
+      "grad_norm": 0.5910077691078186,
+      "learning_rate": 9.999868056699382e-06,
+      "loss": 2.4295,
+      "mean_token_accuracy": 0.4800408352166414,
+      "num_tokens": 164145130.0,
+      "step": 1133
+    },
+    {
+      "entropy": 2.49267578125,
+      "epoch": 0.019524289145424964,
+      "grad_norm": 0.6001760959625244,
+      "learning_rate": 9.99986604999492e-06,
+      "loss": 2.4397,
+      "mean_token_accuracy": 0.4783216747455299,
+      "num_tokens": 164283826.0,
+      "step": 1134
+    },
+    {
+      "entropy": 2.46240234375,
+      "epoch": 0.019541506331620224,
+      "grad_norm": 0.6204310655593872,
+      "learning_rate": 9.99986402814579e-06,
+      "loss": 2.4157,
+      "mean_token_accuracy": 0.48227256163954735,
+      "num_tokens": 164413337.0,
+      "step": 1135
+    },
+    {
+      "entropy": 2.3477783203125,
+      "epoch": 0.019558723517815485,
+      "grad_norm": 0.5665948987007141,
+      "learning_rate": 9.999861991151999e-06,
+      "loss": 2.2835,
+      "mean_token_accuracy": 0.5003114375285804,
+      "num_tokens": 164568720.0,
+      "step": 1136
+    },
+    {
+      "entropy": 2.46435546875,
+      "epoch": 0.019575940704010742,
+      "grad_norm": 0.5802145004272461,
+      "learning_rate": 9.999859939013553e-06,
+      "loss": 2.3899,
+      "mean_token_accuracy": 0.484941388014704,
+      "num_tokens": 164705235.0,
+      "step": 1137
+    },
+    {
+      "entropy": 2.4754638671875,
+      "epoch": 0.019593157890206003,
+      "grad_norm": 0.6445546746253967,
+      "learning_rate": 9.999857871730456e-06,
+      "loss": 2.4595,
+      "mean_token_accuracy": 0.4826706210151315,
+      "num_tokens": 164834818.0,
+      "step": 1138
+    },
+    {
+      "entropy": 2.490478515625,
+      "epoch": 0.019610375076401263,
+      "grad_norm": 0.6412522196769714,
+      "learning_rate": 9.999855789302716e-06,
+      "loss": 2.4606,
+      "mean_token_accuracy": 0.473591239657253,
+      "num_tokens": 164978552.0,
+      "step": 1139
+    },
+    {
+      "entropy": 2.3907470703125,
+      "epoch": 0.019627592262596524,
+      "grad_norm": 0.5629065036773682,
+      "learning_rate": 9.99985369173034e-06,
+      "loss": 2.3848,
+      "mean_token_accuracy": 0.4844861035235226,
+      "num_tokens": 165121299.0,
+      "step": 1140
+    },
+    {
+      "entropy": 2.4473876953125,
+      "epoch": 0.019644809448791784,
+      "grad_norm": 0.6042686104774475,
+      "learning_rate": 9.999851579013334e-06,
+      "loss": 2.4534,
+      "mean_token_accuracy": 0.48099326388910413,
+      "num_tokens": 165256223.0,
+      "step": 1141
+    },
+    {
+      "entropy": 2.4193115234375,
+      "epoch": 0.019662026634987045,
+      "grad_norm": 0.5502400994300842,
+      "learning_rate": 9.999849451151702e-06,
+      "loss": 2.4046,
+      "mean_token_accuracy": 0.4827784230001271,
+      "num_tokens": 165406097.0,
+      "step": 1142
+    },
+    {
+      "entropy": 2.4688720703125,
+      "epoch": 0.019679243821182305,
+      "grad_norm": 0.5841612815856934,
+      "learning_rate": 9.999847308145456e-06,
+      "loss": 2.4319,
+      "mean_token_accuracy": 0.48387880716472864,
+      "num_tokens": 165549274.0,
+      "step": 1143
+    },
+    {
+      "entropy": 2.417724609375,
+      "epoch": 0.019696461007377566,
+      "grad_norm": 0.5570199489593506,
+      "learning_rate": 9.999845149994595e-06,
+      "loss": 2.3172,
+      "mean_token_accuracy": 0.4893231294117868,
+      "num_tokens": 165709510.0,
+      "step": 1144
+    },
+    {
+      "entropy": 2.4190673828125,
+      "epoch": 0.019713678193572823,
+      "grad_norm": 0.6449766159057617,
+      "learning_rate": 9.999842976699133e-06,
+      "loss": 2.3729,
+      "mean_token_accuracy": 0.48674700502306223,
+      "num_tokens": 165858965.0,
+      "step": 1145
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.019730895379768083,
+      "grad_norm": 0.5931810736656189,
+      "learning_rate": 9.99984078825907e-06,
+      "loss": 2.4001,
+      "mean_token_accuracy": 0.4783625756390393,
+      "num_tokens": 165996304.0,
+      "step": 1146
+    },
+    {
+      "entropy": 2.46533203125,
+      "epoch": 0.019748112565963344,
+      "grad_norm": 0.543803870677948,
+      "learning_rate": 9.999838584674417e-06,
+      "loss": 2.4305,
+      "mean_token_accuracy": 0.4785708379931748,
+      "num_tokens": 166150884.0,
+      "step": 1147
+    },
+    {
+      "entropy": 2.55517578125,
+      "epoch": 0.019765329752158604,
+      "grad_norm": 0.59067302942276,
+      "learning_rate": 9.99983636594518e-06,
+      "loss": 2.5251,
+      "mean_token_accuracy": 0.4674039273522794,
+      "num_tokens": 166283288.0,
+      "step": 1148
+    },
+    {
+      "entropy": 2.45947265625,
+      "epoch": 0.019782546938353865,
+      "grad_norm": 0.56581711769104,
+      "learning_rate": 9.999834132071364e-06,
+      "loss": 2.4188,
+      "mean_token_accuracy": 0.48025199631229043,
+      "num_tokens": 166428168.0,
+      "step": 1149
+    },
+    {
+      "entropy": 2.4490966796875,
+      "epoch": 0.019799764124549125,
+      "grad_norm": 0.6398937106132507,
+      "learning_rate": 9.999831883052978e-06,
+      "loss": 2.4001,
+      "mean_token_accuracy": 0.48107053339481354,
+      "num_tokens": 166579098.0,
+      "step": 1150
+    },
+    {
+      "entropy": 2.396484375,
+      "epoch": 0.019816981310744386,
+      "grad_norm": 0.5707962512969971,
+      "learning_rate": 9.999829618890028e-06,
+      "loss": 2.3725,
+      "mean_token_accuracy": 0.4916608249768615,
+      "num_tokens": 166724897.0,
+      "step": 1151
+    },
+    {
+      "entropy": 2.3883056640625,
+      "epoch": 0.019834198496939646,
+      "grad_norm": 0.5890077352523804,
+      "learning_rate": 9.99982733958252e-06,
+      "loss": 2.3406,
+      "mean_token_accuracy": 0.49726316425949335,
+      "num_tokens": 166869333.0,
+      "step": 1152
+    },
+    {
+      "entropy": 2.41796875,
+      "epoch": 0.019851415683134904,
+      "grad_norm": 0.5820266604423523,
+      "learning_rate": 9.99982504513046e-06,
+      "loss": 2.3695,
+      "mean_token_accuracy": 0.4875667071901262,
+      "num_tokens": 167009538.0,
+      "step": 1153
+    },
+    {
+      "entropy": 2.4931640625,
+      "epoch": 0.019868632869330164,
+      "grad_norm": 0.5474282503128052,
+      "learning_rate": 9.999822735533857e-06,
+      "loss": 2.4836,
+      "mean_token_accuracy": 0.47462243027985096,
+      "num_tokens": 167165134.0,
+      "step": 1154
+    },
+    {
+      "entropy": 2.48193359375,
+      "epoch": 0.019885850055525425,
+      "grad_norm": 0.5690717101097107,
+      "learning_rate": 9.999820410792717e-06,
+      "loss": 2.4011,
+      "mean_token_accuracy": 0.47985804779455066,
+      "num_tokens": 167308042.0,
+      "step": 1155
+    },
+    {
+      "entropy": 2.46240234375,
+      "epoch": 0.019903067241720685,
+      "grad_norm": 0.6280803680419922,
+      "learning_rate": 9.99981807090705e-06,
+      "loss": 2.4093,
+      "mean_token_accuracy": 0.4808740792796016,
+      "num_tokens": 167463555.0,
+      "step": 1156
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.019920284427915946,
+      "grad_norm": 0.5713448524475098,
+      "learning_rate": 9.999815715876858e-06,
+      "loss": 2.4033,
+      "mean_token_accuracy": 0.48702497407794,
+      "num_tokens": 167617300.0,
+      "step": 1157
+    },
+    {
+      "entropy": 2.41650390625,
+      "epoch": 0.019937501614111206,
+      "grad_norm": 0.5802963376045227,
+      "learning_rate": 9.999813345702151e-06,
+      "loss": 2.3546,
+      "mean_token_accuracy": 0.4853116972371936,
+      "num_tokens": 167751762.0,
+      "step": 1158
+    },
+    {
+      "entropy": 2.54541015625,
+      "epoch": 0.019954718800306467,
+      "grad_norm": 0.5789119005203247,
+      "learning_rate": 9.999810960382936e-06,
+      "loss": 2.5738,
+      "mean_token_accuracy": 0.46945408172905445,
+      "num_tokens": 167894886.0,
+      "step": 1159
+    },
+    {
+      "entropy": 2.4666748046875,
+      "epoch": 0.019971935986501727,
+      "grad_norm": 0.5464440584182739,
+      "learning_rate": 9.999808559919221e-06,
+      "loss": 2.4415,
+      "mean_token_accuracy": 0.4792809300124645,
+      "num_tokens": 168052147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 2.450927734375,
+      "epoch": 0.019989153172696988,
+      "grad_norm": 0.5954328179359436,
+      "learning_rate": 9.999806144311013e-06,
+      "loss": 2.3982,
+      "mean_token_accuracy": 0.4799165716394782,
+      "num_tokens": 168188013.0,
+      "step": 1161
+    },
+    {
+      "entropy": 2.483154296875,
+      "epoch": 0.020006370358892245,
+      "grad_norm": 0.6072090268135071,
+      "learning_rate": 9.999803713558316e-06,
+      "loss": 2.4523,
+      "mean_token_accuracy": 0.48105273954570293,
+      "num_tokens": 168330752.0,
+      "step": 1162
+    },
+    {
+      "entropy": 2.472900390625,
+      "epoch": 0.020023587545087505,
+      "grad_norm": 0.5980669260025024,
+      "learning_rate": 9.99980126766114e-06,
+      "loss": 2.4257,
+      "mean_token_accuracy": 0.47724353708326817,
+      "num_tokens": 168466830.0,
+      "step": 1163
+    },
+    {
+      "entropy": 2.526123046875,
+      "epoch": 0.020040804731282766,
+      "grad_norm": 0.6172068119049072,
+      "learning_rate": 9.999798806619494e-06,
+      "loss": 2.5131,
+      "mean_token_accuracy": 0.47623227164149284,
+      "num_tokens": 168600305.0,
+      "step": 1164
+    },
+    {
+      "entropy": 2.515380859375,
+      "epoch": 0.020058021917478026,
+      "grad_norm": 0.6749062538146973,
+      "learning_rate": 9.999796330433385e-06,
+      "loss": 2.4829,
+      "mean_token_accuracy": 0.47597271809354424,
+      "num_tokens": 168758522.0,
+      "step": 1165
+    },
+    {
+      "entropy": 2.50341796875,
+      "epoch": 0.020075239103673287,
+      "grad_norm": 0.6174373626708984,
+      "learning_rate": 9.999793839102816e-06,
+      "loss": 2.4201,
+      "mean_token_accuracy": 0.47675873059779406,
+      "num_tokens": 168908687.0,
+      "step": 1166
+    },
+    {
+      "entropy": 2.4525146484375,
+      "epoch": 0.020092456289868547,
+      "grad_norm": 0.5919591188430786,
+      "learning_rate": 9.9997913326278e-06,
+      "loss": 2.4458,
+      "mean_token_accuracy": 0.48330248473212123,
+      "num_tokens": 169051058.0,
+      "step": 1167
+    },
+    {
+      "entropy": 2.4176025390625,
+      "epoch": 0.020109673476063808,
+      "grad_norm": 0.6027262210845947,
+      "learning_rate": 9.999788811008342e-06,
+      "loss": 2.3923,
+      "mean_token_accuracy": 0.48441468458622694,
+      "num_tokens": 169197102.0,
+      "step": 1168
+    },
+    {
+      "entropy": 2.500244140625,
+      "epoch": 0.02012689066225907,
+      "grad_norm": 0.5818685293197632,
+      "learning_rate": 9.99978627424445e-06,
+      "loss": 2.4626,
+      "mean_token_accuracy": 0.47057256614789367,
+      "num_tokens": 169324225.0,
+      "step": 1169
+    },
+    {
+      "entropy": 2.364501953125,
+      "epoch": 0.020144107848454326,
+      "grad_norm": 0.6910699605941772,
+      "learning_rate": 9.999783722336132e-06,
+      "loss": 2.3512,
+      "mean_token_accuracy": 0.48974316706880927,
+      "num_tokens": 169484740.0,
+      "step": 1170
+    },
+    {
+      "entropy": 2.423828125,
+      "epoch": 0.020161325034649586,
+      "grad_norm": 0.583166778087616,
+      "learning_rate": 9.999781155283396e-06,
+      "loss": 2.3721,
+      "mean_token_accuracy": 0.4812298850156367,
+      "num_tokens": 169629618.0,
+      "step": 1171
+    },
+    {
+      "entropy": 2.403076171875,
+      "epoch": 0.020178542220844847,
+      "grad_norm": 0.5646472573280334,
+      "learning_rate": 9.99977857308625e-06,
+      "loss": 2.3559,
+      "mean_token_accuracy": 0.4837103271856904,
+      "num_tokens": 169781342.0,
+      "step": 1172
+    },
+    {
+      "entropy": 2.3782958984375,
+      "epoch": 0.020195759407040107,
+      "grad_norm": 0.5860679149627686,
+      "learning_rate": 9.9997759757447e-06,
+      "loss": 2.3599,
+      "mean_token_accuracy": 0.49117877380922437,
+      "num_tokens": 169922222.0,
+      "step": 1173
+    },
+    {
+      "entropy": 2.416748046875,
+      "epoch": 0.020212976593235368,
+      "grad_norm": 0.6000187993049622,
+      "learning_rate": 9.999773363258755e-06,
+      "loss": 2.4166,
+      "mean_token_accuracy": 0.4860631856136024,
+      "num_tokens": 170059673.0,
+      "step": 1174
+    },
+    {
+      "entropy": 2.488037109375,
+      "epoch": 0.020230193779430628,
+      "grad_norm": 0.5900315046310425,
+      "learning_rate": 9.999770735628423e-06,
+      "loss": 2.4818,
+      "mean_token_accuracy": 0.47089498909190297,
+      "num_tokens": 170199601.0,
+      "step": 1175
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.02024741096562589,
+      "grad_norm": 0.7553943991661072,
+      "learning_rate": 9.999768092853711e-06,
+      "loss": 2.3982,
+      "mean_token_accuracy": 0.48237905837595463,
+      "num_tokens": 170339217.0,
+      "step": 1176
+    },
+    {
+      "entropy": 2.489013671875,
+      "epoch": 0.02026462815182115,
+      "grad_norm": 0.5692632794380188,
+      "learning_rate": 9.99976543493463e-06,
+      "loss": 2.4907,
+      "mean_token_accuracy": 0.4684693585149944,
+      "num_tokens": 170487890.0,
+      "step": 1177
+    },
+    {
+      "entropy": 2.477294921875,
+      "epoch": 0.020281845338016406,
+      "grad_norm": 0.5685938000679016,
+      "learning_rate": 9.999762761871184e-06,
+      "loss": 2.403,
+      "mean_token_accuracy": 0.48181371157988906,
+      "num_tokens": 170641505.0,
+      "step": 1178
+    },
+    {
+      "entropy": 2.43310546875,
+      "epoch": 0.020299062524211667,
+      "grad_norm": 0.6160794496536255,
+      "learning_rate": 9.999760073663383e-06,
+      "loss": 2.3919,
+      "mean_token_accuracy": 0.48729625064879656,
+      "num_tokens": 170769357.0,
+      "step": 1179
+    },
+    {
+      "entropy": 2.5234375,
+      "epoch": 0.020316279710406927,
+      "grad_norm": 0.6029751300811768,
+      "learning_rate": 9.999757370311237e-06,
+      "loss": 2.5087,
+      "mean_token_accuracy": 0.46955224964767694,
+      "num_tokens": 170893993.0,
+      "step": 1180
+    },
+    {
+      "entropy": 2.463134765625,
+      "epoch": 0.020333496896602188,
+      "grad_norm": 0.635032057762146,
+      "learning_rate": 9.999754651814751e-06,
+      "loss": 2.4153,
+      "mean_token_accuracy": 0.4809435624629259,
+      "num_tokens": 171043577.0,
+      "step": 1181
+    },
+    {
+      "entropy": 2.3951416015625,
+      "epoch": 0.02035071408279745,
+      "grad_norm": 0.5716371536254883,
+      "learning_rate": 9.999751918173935e-06,
+      "loss": 2.3335,
+      "mean_token_accuracy": 0.49141963897272944,
+      "num_tokens": 171198846.0,
+      "step": 1182
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.02036793126899271,
+      "grad_norm": 0.5712882876396179,
+      "learning_rate": 9.999749169388798e-06,
+      "loss": 2.3149,
+      "mean_token_accuracy": 0.49085350101813674,
+      "num_tokens": 171351472.0,
+      "step": 1183
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.02038514845518797,
+      "grad_norm": 0.5964887142181396,
+      "learning_rate": 9.999746405459345e-06,
+      "loss": 2.3524,
+      "mean_token_accuracy": 0.4884929973632097,
+      "num_tokens": 171488905.0,
+      "step": 1184
+    },
+    {
+      "entropy": 2.5108642578125,
+      "epoch": 0.02040236564138323,
+      "grad_norm": 0.5682933330535889,
+      "learning_rate": 9.999743626385587e-06,
+      "loss": 2.4582,
+      "mean_token_accuracy": 0.4702299786731601,
+      "num_tokens": 171634563.0,
+      "step": 1185
+    },
+    {
+      "entropy": 2.4671630859375,
+      "epoch": 0.02041958282757849,
+      "grad_norm": 0.6103932857513428,
+      "learning_rate": 9.999740832167532e-06,
+      "loss": 2.4564,
+      "mean_token_accuracy": 0.4751875218935311,
+      "num_tokens": 171777207.0,
+      "step": 1186
+    },
+    {
+      "entropy": 2.49267578125,
+      "epoch": 0.020436800013773748,
+      "grad_norm": 1.0117425918579102,
+      "learning_rate": 9.99973802280519e-06,
+      "loss": 2.4774,
+      "mean_token_accuracy": 0.4714944367296994,
+      "num_tokens": 171934632.0,
+      "step": 1187
+    },
+    {
+      "entropy": 2.45458984375,
+      "epoch": 0.020454017199969008,
+      "grad_norm": 0.5842525959014893,
+      "learning_rate": 9.999735198298566e-06,
+      "loss": 2.4146,
+      "mean_token_accuracy": 0.47794078243896365,
+      "num_tokens": 172076052.0,
+      "step": 1188
+    },
+    {
+      "entropy": 2.43310546875,
+      "epoch": 0.02047123438616427,
+      "grad_norm": 0.594760000705719,
+      "learning_rate": 9.999732358647671e-06,
+      "loss": 2.3923,
+      "mean_token_accuracy": 0.4838462113402784,
+      "num_tokens": 172214297.0,
+      "step": 1189
+    },
+    {
+      "entropy": 2.46826171875,
+      "epoch": 0.02048845157235953,
+      "grad_norm": 0.5831640958786011,
+      "learning_rate": 9.999729503852515e-06,
+      "loss": 2.3907,
+      "mean_token_accuracy": 0.48116701701655984,
+      "num_tokens": 172354024.0,
+      "step": 1190
+    },
+    {
+      "entropy": 2.4698486328125,
+      "epoch": 0.02050566875855479,
+      "grad_norm": 0.612289547920227,
+      "learning_rate": 9.999726633913103e-06,
+      "loss": 2.4473,
+      "mean_token_accuracy": 0.48014617431908846,
+      "num_tokens": 172496977.0,
+      "step": 1191
+    },
+    {
+      "entropy": 2.462890625,
+      "epoch": 0.02052288594475005,
+      "grad_norm": 0.5079575777053833,
+      "learning_rate": 9.999723748829446e-06,
+      "loss": 2.4118,
+      "mean_token_accuracy": 0.4765968224965036,
+      "num_tokens": 172664155.0,
+      "step": 1192
+    },
+    {
+      "entropy": 2.4237060546875,
+      "epoch": 0.02054010313094531,
+      "grad_norm": 0.5485002398490906,
+      "learning_rate": 9.999720848601552e-06,
+      "loss": 2.374,
+      "mean_token_accuracy": 0.4809511392377317,
+      "num_tokens": 172824727.0,
+      "step": 1193
+    },
+    {
+      "entropy": 2.4146728515625,
+      "epoch": 0.02055732031714057,
+      "grad_norm": 0.5650107264518738,
+      "learning_rate": 9.999717933229429e-06,
+      "loss": 2.3705,
+      "mean_token_accuracy": 0.49346699519082904,
+      "num_tokens": 172981026.0,
+      "step": 1194
+    },
+    {
+      "entropy": 2.47900390625,
+      "epoch": 0.020574537503335828,
+      "grad_norm": 0.5556767582893372,
+      "learning_rate": 9.999715002713088e-06,
+      "loss": 2.4417,
+      "mean_token_accuracy": 0.4759158226661384,
+      "num_tokens": 173123752.0,
+      "step": 1195
+    },
+    {
+      "entropy": 2.505859375,
+      "epoch": 0.02059175468953109,
+      "grad_norm": 0.5984048843383789,
+      "learning_rate": 9.999712057052537e-06,
+      "loss": 2.4571,
+      "mean_token_accuracy": 0.4764223378151655,
+      "num_tokens": 173259038.0,
+      "step": 1196
+    },
+    {
+      "entropy": 2.4449462890625,
+      "epoch": 0.02060897187572635,
+      "grad_norm": 0.6215763688087463,
+      "learning_rate": 9.999709096247784e-06,
+      "loss": 2.4247,
+      "mean_token_accuracy": 0.48310248367488384,
+      "num_tokens": 173401014.0,
+      "step": 1197
+    },
+    {
+      "entropy": 2.391845703125,
+      "epoch": 0.02062618906192161,
+      "grad_norm": 0.6615683436393738,
+      "learning_rate": 9.99970612029884e-06,
+      "loss": 2.368,
+      "mean_token_accuracy": 0.49283183040097356,
+      "num_tokens": 173553153.0,
+      "step": 1198
+    },
+    {
+      "entropy": 2.4296875,
+      "epoch": 0.02064340624811687,
+      "grad_norm": 0.5707510113716125,
+      "learning_rate": 9.999703129205711e-06,
+      "loss": 2.3997,
+      "mean_token_accuracy": 0.4811532562598586,
+      "num_tokens": 173705473.0,
+      "step": 1199
+    },
+    {
+      "entropy": 2.3585205078125,
+      "epoch": 0.02066062343431213,
+      "grad_norm": 0.5650596618652344,
+      "learning_rate": 9.999700122968408e-06,
+      "loss": 2.2966,
+      "mean_token_accuracy": 0.4981447677128017,
+      "num_tokens": 173849665.0,
+      "step": 1200
+    },
+    {
+      "entropy": 2.46142578125,
+      "epoch": 0.02067784062050739,
+      "grad_norm": 0.5815020203590393,
+      "learning_rate": 9.99969710158694e-06,
+      "loss": 2.4193,
+      "mean_token_accuracy": 0.4771373653784394,
+      "num_tokens": 173988108.0,
+      "step": 1201
+    },
+    {
+      "entropy": 2.4542236328125,
+      "epoch": 0.020695057806702652,
+      "grad_norm": 0.5876851677894592,
+      "learning_rate": 9.999694065061316e-06,
+      "loss": 2.4786,
+      "mean_token_accuracy": 0.4749302426353097,
+      "num_tokens": 174126294.0,
+      "step": 1202
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.02071227499289791,
+      "grad_norm": 0.562313437461853,
+      "learning_rate": 9.999691013391544e-06,
+      "loss": 2.4206,
+      "mean_token_accuracy": 0.4778692005202174,
+      "num_tokens": 174269124.0,
+      "step": 1203
+    },
+    {
+      "entropy": 2.443115234375,
+      "epoch": 0.02072949217909317,
+      "grad_norm": 0.5638776421546936,
+      "learning_rate": 9.999687946577636e-06,
+      "loss": 2.3994,
+      "mean_token_accuracy": 0.48797021247446537,
+      "num_tokens": 174427015.0,
+      "step": 1204
+    },
+    {
+      "entropy": 2.4337158203125,
+      "epoch": 0.02074670936528843,
+      "grad_norm": 0.5831981301307678,
+      "learning_rate": 9.999684864619599e-06,
+      "loss": 2.3989,
+      "mean_token_accuracy": 0.4862150545231998,
+      "num_tokens": 174569995.0,
+      "step": 1205
+    },
+    {
+      "entropy": 2.449951171875,
+      "epoch": 0.02076392655148369,
+      "grad_norm": 0.5815305709838867,
+      "learning_rate": 9.999681767517441e-06,
+      "loss": 2.3913,
+      "mean_token_accuracy": 0.4784506266005337,
+      "num_tokens": 174706593.0,
+      "step": 1206
+    },
+    {
+      "entropy": 2.487060546875,
+      "epoch": 0.02078114373767895,
+      "grad_norm": 0.5909605026245117,
+      "learning_rate": 9.999678655271176e-06,
+      "loss": 2.4765,
+      "mean_token_accuracy": 0.47097450820729136,
+      "num_tokens": 174846238.0,
+      "step": 1207
+    },
+    {
+      "entropy": 2.44287109375,
+      "epoch": 0.02079836092387421,
+      "grad_norm": 0.5870411992073059,
+      "learning_rate": 9.99967552788081e-06,
+      "loss": 2.3842,
+      "mean_token_accuracy": 0.48391013080254197,
+      "num_tokens": 174992749.0,
+      "step": 1208
+    },
+    {
+      "entropy": 2.4306640625,
+      "epoch": 0.020815578110069472,
+      "grad_norm": 0.5437948703765869,
+      "learning_rate": 9.999672385346352e-06,
+      "loss": 2.3978,
+      "mean_token_accuracy": 0.4776941388845444,
+      "num_tokens": 175147217.0,
+      "step": 1209
+    },
+    {
+      "entropy": 2.428955078125,
+      "epoch": 0.020832795296264733,
+      "grad_norm": 0.5859825611114502,
+      "learning_rate": 9.999669227667815e-06,
+      "loss": 2.3921,
+      "mean_token_accuracy": 0.4817708395421505,
+      "num_tokens": 175293722.0,
+      "step": 1210
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.020850012482459993,
+      "grad_norm": 0.5771918892860413,
+      "learning_rate": 9.999666054845206e-06,
+      "loss": 2.3826,
+      "mean_token_accuracy": 0.4837156576104462,
+      "num_tokens": 175434663.0,
+      "step": 1211
+    },
+    {
+      "entropy": 2.4981689453125,
+      "epoch": 0.02086722966865525,
+      "grad_norm": 0.6018497943878174,
+      "learning_rate": 9.999662866878534e-06,
+      "loss": 2.4908,
+      "mean_token_accuracy": 0.4697205852717161,
+      "num_tokens": 175582504.0,
+      "step": 1212
+    },
+    {
+      "entropy": 2.51611328125,
+      "epoch": 0.02088444685485051,
+      "grad_norm": 0.5939236283302307,
+      "learning_rate": 9.999659663767811e-06,
+      "loss": 2.4614,
+      "mean_token_accuracy": 0.47295166924595833,
+      "num_tokens": 175721585.0,
+      "step": 1213
+    },
+    {
+      "entropy": 2.399169921875,
+      "epoch": 0.02090166404104577,
+      "grad_norm": 0.5899982452392578,
+      "learning_rate": 9.999656445513043e-06,
+      "loss": 2.3798,
+      "mean_token_accuracy": 0.4900959865190089,
+      "num_tokens": 175867778.0,
+      "step": 1214
+    },
+    {
+      "entropy": 2.4320068359375,
+      "epoch": 0.020918881227241032,
+      "grad_norm": 0.5718798041343689,
+      "learning_rate": 9.999653212114245e-06,
+      "loss": 2.395,
+      "mean_token_accuracy": 0.48120944295078516,
+      "num_tokens": 176013170.0,
+      "step": 1215
+    },
+    {
+      "entropy": 2.39794921875,
+      "epoch": 0.020936098413436292,
+      "grad_norm": 0.603789746761322,
+      "learning_rate": 9.99964996357142e-06,
+      "loss": 2.3539,
+      "mean_token_accuracy": 0.4898811914026737,
+      "num_tokens": 176154005.0,
+      "step": 1216
+    },
+    {
+      "entropy": 2.54248046875,
+      "epoch": 0.020953315599631553,
+      "grad_norm": 0.5883952379226685,
+      "learning_rate": 9.999646699884585e-06,
+      "loss": 2.4972,
+      "mean_token_accuracy": 0.46428748965263367,
+      "num_tokens": 176294756.0,
+      "step": 1217
+    },
+    {
+      "entropy": 2.4794921875,
+      "epoch": 0.020970532785826813,
+      "grad_norm": 0.5553483963012695,
+      "learning_rate": 9.999643421053747e-06,
+      "loss": 2.4403,
+      "mean_token_accuracy": 0.4781416282057762,
+      "num_tokens": 176450378.0,
+      "step": 1218
+    },
+    {
+      "entropy": 2.402099609375,
+      "epoch": 0.020987749972022074,
+      "grad_norm": 0.5453605055809021,
+      "learning_rate": 9.999640127078914e-06,
+      "loss": 2.3609,
+      "mean_token_accuracy": 0.4892904283478856,
+      "num_tokens": 176605732.0,
+      "step": 1219
+    },
+    {
+      "entropy": 2.425048828125,
+      "epoch": 0.02100496715821733,
+      "grad_norm": 0.5747688412666321,
+      "learning_rate": 9.999636817960097e-06,
+      "loss": 2.4142,
+      "mean_token_accuracy": 0.4863899489864707,
+      "num_tokens": 176749939.0,
+      "step": 1220
+    },
+    {
+      "entropy": 2.408203125,
+      "epoch": 0.02102218434441259,
+      "grad_norm": 0.5680667757987976,
+      "learning_rate": 9.999633493697307e-06,
+      "loss": 2.3843,
+      "mean_token_accuracy": 0.4866964789107442,
+      "num_tokens": 176896720.0,
+      "step": 1221
+    },
+    {
+      "entropy": 2.4571533203125,
+      "epoch": 0.021039401530607852,
+      "grad_norm": 0.5690086483955383,
+      "learning_rate": 9.999630154290553e-06,
+      "loss": 2.3909,
+      "mean_token_accuracy": 0.48753285547718406,
+      "num_tokens": 177039495.0,
+      "step": 1222
+    },
+    {
+      "entropy": 2.43115234375,
+      "epoch": 0.021056618716803113,
+      "grad_norm": 0.6062940359115601,
+      "learning_rate": 9.999626799739846e-06,
+      "loss": 2.4407,
+      "mean_token_accuracy": 0.4770724857226014,
+      "num_tokens": 177188199.0,
+      "step": 1223
+    },
+    {
+      "entropy": 2.490478515625,
+      "epoch": 0.021073835902998373,
+      "grad_norm": 0.6283055543899536,
+      "learning_rate": 9.999623430045196e-06,
+      "loss": 2.4805,
+      "mean_token_accuracy": 0.4742564079351723,
+      "num_tokens": 177323708.0,
+      "step": 1224
+    },
+    {
+      "entropy": 2.48046875,
+      "epoch": 0.021091053089193634,
+      "grad_norm": 0.5583751201629639,
+      "learning_rate": 9.999620045206614e-06,
+      "loss": 2.4941,
+      "mean_token_accuracy": 0.47078192001208663,
+      "num_tokens": 177479672.0,
+      "step": 1225
+    },
+    {
+      "entropy": 2.4630126953125,
+      "epoch": 0.021108270275388894,
+      "grad_norm": 0.5786680579185486,
+      "learning_rate": 9.999616645224109e-06,
+      "loss": 2.4421,
+      "mean_token_accuracy": 0.4788076733238995,
+      "num_tokens": 177624243.0,
+      "step": 1226
+    },
+    {
+      "entropy": 2.41796875,
+      "epoch": 0.021125487461584155,
+      "grad_norm": 0.5827528834342957,
+      "learning_rate": 9.999613230097692e-06,
+      "loss": 2.3804,
+      "mean_token_accuracy": 0.48857234325259924,
+      "num_tokens": 177764756.0,
+      "step": 1227
+    },
+    {
+      "entropy": 2.4376220703125,
+      "epoch": 0.02114270464777941,
+      "grad_norm": 0.5747877955436707,
+      "learning_rate": 9.99960979982737e-06,
+      "loss": 2.4097,
+      "mean_token_accuracy": 0.4837777316570282,
+      "num_tokens": 177898426.0,
+      "step": 1228
+    },
+    {
+      "entropy": 2.466064453125,
+      "epoch": 0.021159921833974672,
+      "grad_norm": 0.5510834455490112,
+      "learning_rate": 9.999606354413159e-06,
+      "loss": 2.398,
+      "mean_token_accuracy": 0.4762152610346675,
+      "num_tokens": 178050984.0,
+      "step": 1229
+    },
+    {
+      "entropy": 2.40283203125,
+      "epoch": 0.021177139020169933,
+      "grad_norm": 0.5941503643989563,
+      "learning_rate": 9.999602893855067e-06,
+      "loss": 2.3841,
+      "mean_token_accuracy": 0.48818005435168743,
+      "num_tokens": 178189354.0,
+      "step": 1230
+    },
+    {
+      "entropy": 2.438720703125,
+      "epoch": 0.021194356206365193,
+      "grad_norm": 0.5585294961929321,
+      "learning_rate": 9.999599418153104e-06,
+      "loss": 2.3762,
+      "mean_token_accuracy": 0.4838878009468317,
+      "num_tokens": 178331962.0,
+      "step": 1231
+    },
+    {
+      "entropy": 2.4259033203125,
+      "epoch": 0.021211573392560454,
+      "grad_norm": 0.6388170123100281,
+      "learning_rate": 9.999595927307279e-06,
+      "loss": 2.3748,
+      "mean_token_accuracy": 0.48712681280449033,
+      "num_tokens": 178482059.0,
+      "step": 1232
+    },
+    {
+      "entropy": 2.4521484375,
+      "epoch": 0.021228790578755714,
+      "grad_norm": 0.5800935626029968,
+      "learning_rate": 9.999592421317606e-06,
+      "loss": 2.4034,
+      "mean_token_accuracy": 0.48077769484370947,
+      "num_tokens": 178614560.0,
+      "step": 1233
+    },
+    {
+      "entropy": 2.4222412109375,
+      "epoch": 0.021246007764950975,
+      "grad_norm": 0.5996948480606079,
+      "learning_rate": 9.999588900184094e-06,
+      "loss": 2.3708,
+      "mean_token_accuracy": 0.4873070912435651,
+      "num_tokens": 178766038.0,
+      "step": 1234
+    },
+    {
+      "entropy": 2.4306640625,
+      "epoch": 0.021263224951146235,
+      "grad_norm": 0.5760281682014465,
+      "learning_rate": 9.999585363906754e-06,
+      "loss": 2.3985,
+      "mean_token_accuracy": 0.4887649049051106,
+      "num_tokens": 178911786.0,
+      "step": 1235
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.021280442137341496,
+      "grad_norm": 0.5570631623268127,
+      "learning_rate": 9.999581812485595e-06,
+      "loss": 2.4024,
+      "mean_token_accuracy": 0.4858241407200694,
+      "num_tokens": 179059130.0,
+      "step": 1236
+    },
+    {
+      "entropy": 2.43017578125,
+      "epoch": 0.021297659323536753,
+      "grad_norm": 0.6132498979568481,
+      "learning_rate": 9.999578245920632e-06,
+      "loss": 2.413,
+      "mean_token_accuracy": 0.4828817122615874,
+      "num_tokens": 179193557.0,
+      "step": 1237
+    },
+    {
+      "entropy": 2.518310546875,
+      "epoch": 0.021314876509732014,
+      "grad_norm": 0.6096269488334656,
+      "learning_rate": 9.99957466421187e-06,
+      "loss": 2.4572,
+      "mean_token_accuracy": 0.4700988312251866,
+      "num_tokens": 179318893.0,
+      "step": 1238
+    },
+    {
+      "entropy": 2.4744873046875,
+      "epoch": 0.021332093695927274,
+      "grad_norm": 0.5955691337585449,
+      "learning_rate": 9.999571067359323e-06,
+      "loss": 2.4467,
+      "mean_token_accuracy": 0.475006652995944,
+      "num_tokens": 179459850.0,
+      "step": 1239
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.021349310882122535,
+      "grad_norm": 0.5774043798446655,
+      "learning_rate": 9.999567455363003e-06,
+      "loss": 2.4671,
+      "mean_token_accuracy": 0.4733973373658955,
+      "num_tokens": 179610624.0,
+      "step": 1240
+    },
+    {
+      "entropy": 2.450439453125,
+      "epoch": 0.021366528068317795,
+      "grad_norm": 0.5889162421226501,
+      "learning_rate": 9.99956382822292e-06,
+      "loss": 2.4152,
+      "mean_token_accuracy": 0.4829295091331005,
+      "num_tokens": 179743518.0,
+      "step": 1241
+    },
+    {
+      "entropy": 2.4434814453125,
+      "epoch": 0.021383745254513056,
+      "grad_norm": 0.5916315913200378,
+      "learning_rate": 9.999560185939082e-06,
+      "loss": 2.3796,
+      "mean_token_accuracy": 0.48776215221732855,
+      "num_tokens": 179888613.0,
+      "step": 1242
+    },
+    {
+      "entropy": 2.44091796875,
+      "epoch": 0.021400962440708316,
+      "grad_norm": 0.6239886283874512,
+      "learning_rate": 9.999556528511504e-06,
+      "loss": 2.4542,
+      "mean_token_accuracy": 0.48557718843221664,
+      "num_tokens": 180029182.0,
+      "step": 1243
+    },
+    {
+      "entropy": 2.452880859375,
+      "epoch": 0.021418179626903577,
+      "grad_norm": 0.5450658202171326,
+      "learning_rate": 9.999552855940197e-06,
+      "loss": 2.4064,
+      "mean_token_accuracy": 0.48464760929346085,
+      "num_tokens": 180184347.0,
+      "step": 1244
+    },
+    {
+      "entropy": 2.46923828125,
+      "epoch": 0.021435396813098834,
+      "grad_norm": 0.5909526944160461,
+      "learning_rate": 9.999549168225169e-06,
+      "loss": 2.3883,
+      "mean_token_accuracy": 0.47910354332998395,
+      "num_tokens": 180333887.0,
+      "step": 1245
+    },
+    {
+      "entropy": 2.4539794921875,
+      "epoch": 0.021452613999294094,
+      "grad_norm": 0.581224799156189,
+      "learning_rate": 9.999545465366433e-06,
+      "loss": 2.4123,
+      "mean_token_accuracy": 0.4825686193071306,
+      "num_tokens": 180490238.0,
+      "step": 1246
+    },
+    {
+      "entropy": 2.439697265625,
+      "epoch": 0.021469831185489355,
+      "grad_norm": 0.6227947473526001,
+      "learning_rate": 9.999541747364002e-06,
+      "loss": 2.4026,
+      "mean_token_accuracy": 0.4787580128759146,
+      "num_tokens": 180625744.0,
+      "step": 1247
+    },
+    {
+      "entropy": 2.4227294921875,
+      "epoch": 0.021487048371684615,
+      "grad_norm": 0.5937827825546265,
+      "learning_rate": 9.999538014217884e-06,
+      "loss": 2.359,
+      "mean_token_accuracy": 0.4844051315449178,
+      "num_tokens": 180761591.0,
+      "step": 1248
+    },
+    {
+      "entropy": 2.46337890625,
+      "epoch": 0.021504265557879876,
+      "grad_norm": 0.5774495601654053,
+      "learning_rate": 9.999534265928092e-06,
+      "loss": 2.4378,
+      "mean_token_accuracy": 0.4757684310898185,
+      "num_tokens": 180903465.0,
+      "step": 1249
+    },
+    {
+      "entropy": 2.5347900390625,
+      "epoch": 0.021521482744075136,
+      "grad_norm": 0.5800737738609314,
+      "learning_rate": 9.999530502494639e-06,
+      "loss": 2.4917,
+      "mean_token_accuracy": 0.4692853162996471,
+      "num_tokens": 181053233.0,
+      "step": 1250
+    },
+    {
+      "entropy": 2.4166259765625,
+      "epoch": 0.021538699930270397,
+      "grad_norm": 0.5999194383621216,
+      "learning_rate": 9.999526723917533e-06,
+      "loss": 2.372,
+      "mean_token_accuracy": 0.4842418390326202,
+      "num_tokens": 181202187.0,
+      "step": 1251
+    },
+    {
+      "entropy": 2.3966064453125,
+      "epoch": 0.021555917116465657,
+      "grad_norm": 0.561681866645813,
+      "learning_rate": 9.999522930196787e-06,
+      "loss": 2.3211,
+      "mean_token_accuracy": 0.48686323687434196,
+      "num_tokens": 181345964.0,
+      "step": 1252
+    },
+    {
+      "entropy": 2.350830078125,
+      "epoch": 0.021573134302660914,
+      "grad_norm": 0.6096113920211792,
+      "learning_rate": 9.999519121332413e-06,
+      "loss": 2.3158,
+      "mean_token_accuracy": 0.49966981168836355,
+      "num_tokens": 181493986.0,
+      "step": 1253
+    },
+    {
+      "entropy": 2.466064453125,
+      "epoch": 0.021590351488856175,
+      "grad_norm": 0.5756585001945496,
+      "learning_rate": 9.99951529732442e-06,
+      "loss": 2.4099,
+      "mean_token_accuracy": 0.4760406082496047,
+      "num_tokens": 181643446.0,
+      "step": 1254
+    },
+    {
+      "entropy": 2.4566650390625,
+      "epoch": 0.021607568675051435,
+      "grad_norm": 0.6152567267417908,
+      "learning_rate": 9.999511458172823e-06,
+      "loss": 2.4127,
+      "mean_token_accuracy": 0.48100624280050397,
+      "num_tokens": 181798482.0,
+      "step": 1255
+    },
+    {
+      "entropy": 2.42578125,
+      "epoch": 0.021624785861246696,
+      "grad_norm": 0.5952988266944885,
+      "learning_rate": 9.999507603877634e-06,
+      "loss": 2.3729,
+      "mean_token_accuracy": 0.48960635205730796,
+      "num_tokens": 181947283.0,
+      "step": 1256
+    },
+    {
+      "entropy": 2.489013671875,
+      "epoch": 0.021642003047441957,
+      "grad_norm": 0.5729377865791321,
+      "learning_rate": 9.99950373443886e-06,
+      "loss": 2.4575,
+      "mean_token_accuracy": 0.47360460739582777,
+      "num_tokens": 182103948.0,
+      "step": 1257
+    },
+    {
+      "entropy": 2.367919921875,
+      "epoch": 0.021659220233637217,
+      "grad_norm": 0.5920243859291077,
+      "learning_rate": 9.999499849856518e-06,
+      "loss": 2.3522,
+      "mean_token_accuracy": 0.4938758905045688,
+      "num_tokens": 182244306.0,
+      "step": 1258
+    },
+    {
+      "entropy": 2.4716796875,
+      "epoch": 0.021676437419832478,
+      "grad_norm": 0.6242178678512573,
+      "learning_rate": 9.999495950130616e-06,
+      "loss": 2.4071,
+      "mean_token_accuracy": 0.4799223360605538,
+      "num_tokens": 182370840.0,
+      "step": 1259
+    },
+    {
+      "entropy": 2.4951171875,
+      "epoch": 0.021693654606027738,
+      "grad_norm": 0.7572094798088074,
+      "learning_rate": 9.999492035261166e-06,
+      "loss": 2.466,
+      "mean_token_accuracy": 0.48017187928780913,
+      "num_tokens": 182517478.0,
+      "step": 1260
+    },
+    {
+      "entropy": 2.4541015625,
+      "epoch": 0.021710871792223,
+      "grad_norm": 0.6314413547515869,
+      "learning_rate": 9.999488105248184e-06,
+      "loss": 2.4002,
+      "mean_token_accuracy": 0.4775316468439996,
+      "num_tokens": 182661585.0,
+      "step": 1261
+    },
+    {
+      "entropy": 2.4586181640625,
+      "epoch": 0.021728088978418256,
+      "grad_norm": 0.6284291744232178,
+      "learning_rate": 9.999484160091678e-06,
+      "loss": 2.4529,
+      "mean_token_accuracy": 0.4805305516347289,
+      "num_tokens": 182804645.0,
+      "step": 1262
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.021745306164613516,
+      "grad_norm": 0.6105698347091675,
+      "learning_rate": 9.99948019979166e-06,
+      "loss": 2.4449,
+      "mean_token_accuracy": 0.4798229462467134,
+      "num_tokens": 182947450.0,
+      "step": 1263
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.021762523350808777,
+      "grad_norm": 0.5790930390357971,
+      "learning_rate": 9.999476224348144e-06,
+      "loss": 2.4185,
+      "mean_token_accuracy": 0.4775314489379525,
+      "num_tokens": 183089455.0,
+      "step": 1264
+    },
+    {
+      "entropy": 2.5020751953125,
+      "epoch": 0.021779740537004037,
+      "grad_norm": 0.6524637341499329,
+      "learning_rate": 9.99947223376114e-06,
+      "loss": 2.4855,
+      "mean_token_accuracy": 0.480673139449209,
+      "num_tokens": 183241120.0,
+      "step": 1265
+    },
+    {
+      "entropy": 2.4835205078125,
+      "epoch": 0.021796957723199298,
+      "grad_norm": 0.5516358613967896,
+      "learning_rate": 9.99946822803066e-06,
+      "loss": 2.414,
+      "mean_token_accuracy": 0.47934063244611025,
+      "num_tokens": 183390072.0,
+      "step": 1266
+    },
+    {
+      "entropy": 2.506591796875,
+      "epoch": 0.02181417490939456,
+      "grad_norm": 0.5939943194389343,
+      "learning_rate": 9.99946420715672e-06,
+      "loss": 2.4367,
+      "mean_token_accuracy": 0.47747283428907394,
+      "num_tokens": 183539529.0,
+      "step": 1267
+    },
+    {
+      "entropy": 2.4515380859375,
+      "epoch": 0.02183139209558982,
+      "grad_norm": 0.5539537072181702,
+      "learning_rate": 9.999460171139328e-06,
+      "loss": 2.3763,
+      "mean_token_accuracy": 0.4820182635448873,
+      "num_tokens": 183688585.0,
+      "step": 1268
+    },
+    {
+      "entropy": 2.404541015625,
+      "epoch": 0.02184860928178508,
+      "grad_norm": 0.5919788479804993,
+      "learning_rate": 9.999456119978496e-06,
+      "loss": 2.4045,
+      "mean_token_accuracy": 0.4871440161950886,
+      "num_tokens": 183821506.0,
+      "step": 1269
+    },
+    {
+      "entropy": 2.472412109375,
+      "epoch": 0.021865826467980336,
+      "grad_norm": 0.6141915917396545,
+      "learning_rate": 9.999452053674242e-06,
+      "loss": 2.438,
+      "mean_token_accuracy": 0.4749964135698974,
+      "num_tokens": 183954051.0,
+      "step": 1270
+    },
+    {
+      "entropy": 2.4482421875,
+      "epoch": 0.021883043654175597,
+      "grad_norm": 0.5740827322006226,
+      "learning_rate": 9.99944797222657e-06,
+      "loss": 2.4203,
+      "mean_token_accuracy": 0.47882386669516563,
+      "num_tokens": 184097023.0,
+      "step": 1271
+    },
+    {
+      "entropy": 2.4649658203125,
+      "epoch": 0.021900260840370857,
+      "grad_norm": 0.6000681519508362,
+      "learning_rate": 9.999443875635499e-06,
+      "loss": 2.3876,
+      "mean_token_accuracy": 0.48048424860462546,
+      "num_tokens": 184244536.0,
+      "step": 1272
+    },
+    {
+      "entropy": 2.448486328125,
+      "epoch": 0.021917478026566118,
+      "grad_norm": 0.5612481236457825,
+      "learning_rate": 9.999439763901037e-06,
+      "loss": 2.375,
+      "mean_token_accuracy": 0.48185514099895954,
+      "num_tokens": 184392474.0,
+      "step": 1273
+    },
+    {
+      "entropy": 2.397705078125,
+      "epoch": 0.02193469521276138,
+      "grad_norm": 0.593385636806488,
+      "learning_rate": 9.9994356370232e-06,
+      "loss": 2.3472,
+      "mean_token_accuracy": 0.4966060775332153,
+      "num_tokens": 184541344.0,
+      "step": 1274
+    },
+    {
+      "entropy": 2.4896240234375,
+      "epoch": 0.02195191239895664,
+      "grad_norm": 0.5971876978874207,
+      "learning_rate": 9.999431495001998e-06,
+      "loss": 2.4195,
+      "mean_token_accuracy": 0.4794482118450105,
+      "num_tokens": 184672977.0,
+      "step": 1275
+    },
+    {
+      "entropy": 2.4857177734375,
+      "epoch": 0.0219691295851519,
+      "grad_norm": 0.6552586555480957,
+      "learning_rate": 9.999427337837444e-06,
+      "loss": 2.4895,
+      "mean_token_accuracy": 0.47428874857723713,
+      "num_tokens": 184826101.0,
+      "step": 1276
+    },
+    {
+      "entropy": 2.42529296875,
+      "epoch": 0.02198634677134716,
+      "grad_norm": 0.5778135061264038,
+      "learning_rate": 9.999423165529554e-06,
+      "loss": 2.4422,
+      "mean_token_accuracy": 0.4804799151606858,
+      "num_tokens": 184970321.0,
+      "step": 1277
+    },
+    {
+      "entropy": 2.4859619140625,
+      "epoch": 0.022003563957542417,
+      "grad_norm": 0.537326455116272,
+      "learning_rate": 9.999418978078335e-06,
+      "loss": 2.4407,
+      "mean_token_accuracy": 0.47214187448844314,
+      "num_tokens": 185128581.0,
+      "step": 1278
+    },
+    {
+      "entropy": 2.39697265625,
+      "epoch": 0.022020781143737678,
+      "grad_norm": 0.6056032776832581,
+      "learning_rate": 9.999414775483803e-06,
+      "loss": 2.381,
+      "mean_token_accuracy": 0.4875910747796297,
+      "num_tokens": 185266968.0,
+      "step": 1279
+    },
+    {
+      "entropy": 2.50537109375,
+      "epoch": 0.022037998329932938,
+      "grad_norm": 0.5535710453987122,
+      "learning_rate": 9.99941055774597e-06,
+      "loss": 2.4915,
+      "mean_token_accuracy": 0.4735433543100953,
+      "num_tokens": 185416068.0,
+      "step": 1280
+    },
+    {
+      "entropy": 2.4454345703125,
+      "epoch": 0.0220552155161282,
+      "grad_norm": 0.6215351223945618,
+      "learning_rate": 9.99940632486485e-06,
+      "loss": 2.3778,
+      "mean_token_accuracy": 0.4865972097031772,
+      "num_tokens": 185558293.0,
+      "step": 1281
+    },
+    {
+      "entropy": 2.510498046875,
+      "epoch": 0.02207243270232346,
+      "grad_norm": 0.5621281862258911,
+      "learning_rate": 9.999402076840452e-06,
+      "loss": 2.4886,
+      "mean_token_accuracy": 0.4751009796746075,
+      "num_tokens": 185707903.0,
+      "step": 1282
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.02208964988851872,
+      "grad_norm": 0.5782500505447388,
+      "learning_rate": 9.999397813672793e-06,
+      "loss": 2.4012,
+      "mean_token_accuracy": 0.47963299648836255,
+      "num_tokens": 185864974.0,
+      "step": 1283
+    },
+    {
+      "entropy": 2.452880859375,
+      "epoch": 0.02210686707471398,
+      "grad_norm": 0.584717333316803,
+      "learning_rate": 9.999393535361884e-06,
+      "loss": 2.4252,
+      "mean_token_accuracy": 0.4816917534917593,
+      "num_tokens": 186005584.0,
+      "step": 1284
+    },
+    {
+      "entropy": 2.43310546875,
+      "epoch": 0.02212408426090924,
+      "grad_norm": 0.5956825613975525,
+      "learning_rate": 9.99938924190774e-06,
+      "loss": 2.3782,
+      "mean_token_accuracy": 0.48892744118347764,
+      "num_tokens": 186143411.0,
+      "step": 1285
+    },
+    {
+      "entropy": 2.4464111328125,
+      "epoch": 0.0221413014471045,
+      "grad_norm": 0.5437039136886597,
+      "learning_rate": 9.99938493331037e-06,
+      "loss": 2.4353,
+      "mean_token_accuracy": 0.4839400313794613,
+      "num_tokens": 186299707.0,
+      "step": 1286
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.02215851863329976,
+      "grad_norm": 0.6034572720527649,
+      "learning_rate": 9.999380609569791e-06,
+      "loss": 2.4795,
+      "mean_token_accuracy": 0.47284478275105357,
+      "num_tokens": 186447393.0,
+      "step": 1287
+    },
+    {
+      "entropy": 2.4296875,
+      "epoch": 0.02217573581949502,
+      "grad_norm": 0.5920705795288086,
+      "learning_rate": 9.999376270686015e-06,
+      "loss": 2.4214,
+      "mean_token_accuracy": 0.4788710060529411,
+      "num_tokens": 186590786.0,
+      "step": 1288
+    },
+    {
+      "entropy": 2.429443359375,
+      "epoch": 0.02219295300569028,
+      "grad_norm": 0.5723591446876526,
+      "learning_rate": 9.999371916659054e-06,
+      "loss": 2.385,
+      "mean_token_accuracy": 0.48299159901216626,
+      "num_tokens": 186739554.0,
+      "step": 1289
+    },
+    {
+      "entropy": 2.39111328125,
+      "epoch": 0.02221017019188554,
+      "grad_norm": 0.5848551988601685,
+      "learning_rate": 9.999367547488923e-06,
+      "loss": 2.2912,
+      "mean_token_accuracy": 0.49383825389668345,
+      "num_tokens": 186887370.0,
+      "step": 1290
+    },
+    {
+      "entropy": 2.467529296875,
+      "epoch": 0.0222273873780808,
+      "grad_norm": 0.5705557465553284,
+      "learning_rate": 9.999363163175632e-06,
+      "loss": 2.4211,
+      "mean_token_accuracy": 0.4749067425727844,
+      "num_tokens": 187028206.0,
+      "step": 1291
+    },
+    {
+      "entropy": 2.5201416015625,
+      "epoch": 0.02224460456427606,
+      "grad_norm": 0.6075615882873535,
+      "learning_rate": 9.999358763719198e-06,
+      "loss": 2.4732,
+      "mean_token_accuracy": 0.4724921630695462,
+      "num_tokens": 187178625.0,
+      "step": 1292
+    },
+    {
+      "entropy": 2.46142578125,
+      "epoch": 0.02226182175047132,
+      "grad_norm": 0.5848605036735535,
+      "learning_rate": 9.999354349119632e-06,
+      "loss": 2.4225,
+      "mean_token_accuracy": 0.47803614335134625,
+      "num_tokens": 187317792.0,
+      "step": 1293
+    },
+    {
+      "entropy": 2.443359375,
+      "epoch": 0.022279038936666582,
+      "grad_norm": 0.5687076449394226,
+      "learning_rate": 9.999349919376949e-06,
+      "loss": 2.3935,
+      "mean_token_accuracy": 0.48158882977440953,
+      "num_tokens": 187468854.0,
+      "step": 1294
+    },
+    {
+      "entropy": 2.4393310546875,
+      "epoch": 0.02229625612286184,
+      "grad_norm": 0.5710064768791199,
+      "learning_rate": 9.999345474491161e-06,
+      "loss": 2.3597,
+      "mean_token_accuracy": 0.4866549982689321,
+      "num_tokens": 187617503.0,
+      "step": 1295
+    },
+    {
+      "entropy": 2.4757080078125,
+      "epoch": 0.0223134733090571,
+      "grad_norm": 0.6155255436897278,
+      "learning_rate": 9.999341014462283e-06,
+      "loss": 2.4421,
+      "mean_token_accuracy": 0.47693472215905786,
+      "num_tokens": 187745930.0,
+      "step": 1296
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.02233069049525236,
+      "grad_norm": 0.5629665851593018,
+      "learning_rate": 9.999336539290325e-06,
+      "loss": 2.4156,
+      "mean_token_accuracy": 0.4816053519025445,
+      "num_tokens": 187892581.0,
+      "step": 1297
+    },
+    {
+      "entropy": 2.4129638671875,
+      "epoch": 0.02234790768144762,
+      "grad_norm": 0.5968115329742432,
+      "learning_rate": 9.999332048975305e-06,
+      "loss": 2.3635,
+      "mean_token_accuracy": 0.48733530612662435,
+      "num_tokens": 188044827.0,
+      "step": 1298
+    },
+    {
+      "entropy": 2.4227294921875,
+      "epoch": 0.02236512486764288,
+      "grad_norm": 0.5534366965293884,
+      "learning_rate": 9.999327543517234e-06,
+      "loss": 2.4033,
+      "mean_token_accuracy": 0.4784287307411432,
+      "num_tokens": 188208828.0,
+      "step": 1299
+    },
+    {
+      "entropy": 2.42578125,
+      "epoch": 0.022382342053838142,
+      "grad_norm": 0.5828860402107239,
+      "learning_rate": 9.999323022916128e-06,
+      "loss": 2.3879,
+      "mean_token_accuracy": 0.4846663847565651,
+      "num_tokens": 188357779.0,
+      "step": 1300
+    },
+    {
+      "entropy": 2.4833984375,
+      "epoch": 0.022399559240033402,
+      "grad_norm": 0.6039474606513977,
+      "learning_rate": 9.999318487171998e-06,
+      "loss": 2.4643,
+      "mean_token_accuracy": 0.47219090024009347,
+      "num_tokens": 188488042.0,
+      "step": 1301
+    },
+    {
+      "entropy": 2.443359375,
+      "epoch": 0.022416776426228663,
+      "grad_norm": 0.6493925452232361,
+      "learning_rate": 9.999313936284858e-06,
+      "loss": 2.4156,
+      "mean_token_accuracy": 0.4812625157646835,
+      "num_tokens": 188631539.0,
+      "step": 1302
+    },
+    {
+      "entropy": 2.439453125,
+      "epoch": 0.02243399361242392,
+      "grad_norm": 0.590706467628479,
+      "learning_rate": 9.999309370254722e-06,
+      "loss": 2.4059,
+      "mean_token_accuracy": 0.482134644407779,
+      "num_tokens": 188778873.0,
+      "step": 1303
+    },
+    {
+      "entropy": 2.376708984375,
+      "epoch": 0.02245121079861918,
+      "grad_norm": 0.7171503305435181,
+      "learning_rate": 9.999304789081604e-06,
+      "loss": 2.3167,
+      "mean_token_accuracy": 0.49248863477259874,
+      "num_tokens": 188929372.0,
+      "step": 1304
+    },
+    {
+      "entropy": 2.47119140625,
+      "epoch": 0.02246842798481444,
+      "grad_norm": 0.603790283203125,
+      "learning_rate": 9.999300192765521e-06,
+      "loss": 2.4484,
+      "mean_token_accuracy": 0.4733037226833403,
+      "num_tokens": 189068047.0,
+      "step": 1305
+    },
+    {
+      "entropy": 2.41162109375,
+      "epoch": 0.0224856451710097,
+      "grad_norm": 0.5595695376396179,
+      "learning_rate": 9.999295581306483e-06,
+      "loss": 2.3642,
+      "mean_token_accuracy": 0.4862880017608404,
+      "num_tokens": 189213830.0,
+      "step": 1306
+    },
+    {
+      "entropy": 2.4998779296875,
+      "epoch": 0.022502862357204962,
+      "grad_norm": 0.616247296333313,
+      "learning_rate": 9.999290954704505e-06,
+      "loss": 2.4518,
+      "mean_token_accuracy": 0.48123760567978024,
+      "num_tokens": 189351112.0,
+      "step": 1307
+    },
+    {
+      "entropy": 2.48583984375,
+      "epoch": 0.022520079543400223,
+      "grad_norm": 0.5927817821502686,
+      "learning_rate": 9.999286312959602e-06,
+      "loss": 2.4738,
+      "mean_token_accuracy": 0.476222672034055,
+      "num_tokens": 189488488.0,
+      "step": 1308
+    },
+    {
+      "entropy": 2.479736328125,
+      "epoch": 0.022537296729595483,
+      "grad_norm": 0.5709447860717773,
+      "learning_rate": 9.999281656071784e-06,
+      "loss": 2.4563,
+      "mean_token_accuracy": 0.47250251518562436,
+      "num_tokens": 189630709.0,
+      "step": 1309
+    },
+    {
+      "entropy": 2.4727783203125,
+      "epoch": 0.022554513915790744,
+      "grad_norm": 0.5687850713729858,
+      "learning_rate": 9.99927698404107e-06,
+      "loss": 2.4368,
+      "mean_token_accuracy": 0.4784710453823209,
+      "num_tokens": 189772213.0,
+      "step": 1310
+    },
+    {
+      "entropy": 2.3721923828125,
+      "epoch": 0.022571731101986004,
+      "grad_norm": 0.6119619607925415,
+      "learning_rate": 9.999272296867475e-06,
+      "loss": 2.2997,
+      "mean_token_accuracy": 0.49369402416050434,
+      "num_tokens": 189913660.0,
+      "step": 1311
+    },
+    {
+      "entropy": 2.4296875,
+      "epoch": 0.02258894828818126,
+      "grad_norm": 0.61922687292099,
+      "learning_rate": 9.999267594551007e-06,
+      "loss": 2.4161,
+      "mean_token_accuracy": 0.4793846160173416,
+      "num_tokens": 190047956.0,
+      "step": 1312
+    },
+    {
+      "entropy": 2.5008544921875,
+      "epoch": 0.02260616547437652,
+      "grad_norm": 0.8448692560195923,
+      "learning_rate": 9.999262877091687e-06,
+      "loss": 2.4671,
+      "mean_token_accuracy": 0.4724547420628369,
+      "num_tokens": 190193511.0,
+      "step": 1313
+    },
+    {
+      "entropy": 2.43505859375,
+      "epoch": 0.022623382660571782,
+      "grad_norm": 0.592045247554779,
+      "learning_rate": 9.999258144489524e-06,
+      "loss": 2.3928,
+      "mean_token_accuracy": 0.47932835714891553,
+      "num_tokens": 190339476.0,
+      "step": 1314
+    },
+    {
+      "entropy": 2.49169921875,
+      "epoch": 0.022640599846767043,
+      "grad_norm": 0.6020418405532837,
+      "learning_rate": 9.999253396744534e-06,
+      "loss": 2.4387,
+      "mean_token_accuracy": 0.4741725055500865,
+      "num_tokens": 190473944.0,
+      "step": 1315
+    },
+    {
+      "entropy": 2.437744140625,
+      "epoch": 0.022657817032962303,
+      "grad_norm": 0.5783674120903015,
+      "learning_rate": 9.999248633856735e-06,
+      "loss": 2.3754,
+      "mean_token_accuracy": 0.48019342171028256,
+      "num_tokens": 190621619.0,
+      "step": 1316
+    },
+    {
+      "entropy": 2.5028076171875,
+      "epoch": 0.022675034219157564,
+      "grad_norm": 0.6043087840080261,
+      "learning_rate": 9.999243855826137e-06,
+      "loss": 2.4531,
+      "mean_token_accuracy": 0.4730870760977268,
+      "num_tokens": 190753516.0,
+      "step": 1317
+    },
+    {
+      "entropy": 2.45654296875,
+      "epoch": 0.022692251405352824,
+      "grad_norm": 0.646530270576477,
+      "learning_rate": 9.999239062652754e-06,
+      "loss": 2.3895,
+      "mean_token_accuracy": 0.48305864399299026,
+      "num_tokens": 190898210.0,
+      "step": 1318
+    },
+    {
+      "entropy": 2.519775390625,
+      "epoch": 0.022709468591548085,
+      "grad_norm": 0.5441105961799622,
+      "learning_rate": 9.999234254336603e-06,
+      "loss": 2.5006,
+      "mean_token_accuracy": 0.465965838637203,
+      "num_tokens": 191072262.0,
+      "step": 1319
+    },
+    {
+      "entropy": 2.4124755859375,
+      "epoch": 0.022726685777743342,
+      "grad_norm": 0.5883780121803284,
+      "learning_rate": 9.9992294308777e-06,
+      "loss": 2.3936,
+      "mean_token_accuracy": 0.4880047831684351,
+      "num_tokens": 191216422.0,
+      "step": 1320
+    },
+    {
+      "entropy": 2.4327392578125,
+      "epoch": 0.022743902963938602,
+      "grad_norm": 0.5934004783630371,
+      "learning_rate": 9.999224592276055e-06,
+      "loss": 2.3924,
+      "mean_token_accuracy": 0.48111487692222,
+      "num_tokens": 191352211.0,
+      "step": 1321
+    },
+    {
+      "entropy": 2.497314453125,
+      "epoch": 0.022761120150133863,
+      "grad_norm": 0.5587040781974792,
+      "learning_rate": 9.999219738531687e-06,
+      "loss": 2.4294,
+      "mean_token_accuracy": 0.4762568627484143,
+      "num_tokens": 191496965.0,
+      "step": 1322
+    },
+    {
+      "entropy": 2.4332275390625,
+      "epoch": 0.022778337336329123,
+      "grad_norm": 0.5939592719078064,
+      "learning_rate": 9.999214869644608e-06,
+      "loss": 2.3724,
+      "mean_token_accuracy": 0.4876466835848987,
+      "num_tokens": 191642816.0,
+      "step": 1323
+    },
+    {
+      "entropy": 2.3785400390625,
+      "epoch": 0.022795554522524384,
+      "grad_norm": 0.5794064402580261,
+      "learning_rate": 9.999209985614832e-06,
+      "loss": 2.4016,
+      "mean_token_accuracy": 0.48678439343348145,
+      "num_tokens": 191792676.0,
+      "step": 1324
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.022812771708719645,
+      "grad_norm": 0.5635095238685608,
+      "learning_rate": 9.999205086442378e-06,
+      "loss": 2.4489,
+      "mean_token_accuracy": 0.47591904271394014,
+      "num_tokens": 191943634.0,
+      "step": 1325
+    },
+    {
+      "entropy": 2.454833984375,
+      "epoch": 0.022829988894914905,
+      "grad_norm": 0.5465742349624634,
+      "learning_rate": 9.999200172127257e-06,
+      "loss": 2.4428,
+      "mean_token_accuracy": 0.4766951338388026,
+      "num_tokens": 192097281.0,
+      "step": 1326
+    },
+    {
+      "entropy": 2.490478515625,
+      "epoch": 0.022847206081110166,
+      "grad_norm": 0.5686846375465393,
+      "learning_rate": 9.999195242669487e-06,
+      "loss": 2.4461,
+      "mean_token_accuracy": 0.4775790839921683,
+      "num_tokens": 192238225.0,
+      "step": 1327
+    },
+    {
+      "entropy": 2.546630859375,
+      "epoch": 0.022864423267305423,
+      "grad_norm": 0.5859907865524292,
+      "learning_rate": 9.99919029806908e-06,
+      "loss": 2.5229,
+      "mean_token_accuracy": 0.4657089659012854,
+      "num_tokens": 192380707.0,
+      "step": 1328
+    },
+    {
+      "entropy": 2.48095703125,
+      "epoch": 0.022881640453500683,
+      "grad_norm": 0.5355405807495117,
+      "learning_rate": 9.99918533832605e-06,
+      "loss": 2.4508,
+      "mean_token_accuracy": 0.47027599764987826,
+      "num_tokens": 192539763.0,
+      "step": 1329
+    },
+    {
+      "entropy": 2.449462890625,
+      "epoch": 0.022898857639695944,
+      "grad_norm": 0.5785843133926392,
+      "learning_rate": 9.999180363440416e-06,
+      "loss": 2.4106,
+      "mean_token_accuracy": 0.48218734934926033,
+      "num_tokens": 192687897.0,
+      "step": 1330
+    },
+    {
+      "entropy": 2.4329833984375,
+      "epoch": 0.022916074825891204,
+      "grad_norm": 0.5569692850112915,
+      "learning_rate": 9.99917537341219e-06,
+      "loss": 2.4042,
+      "mean_token_accuracy": 0.4766626372002065,
+      "num_tokens": 192834922.0,
+      "step": 1331
+    },
+    {
+      "entropy": 2.3797607421875,
+      "epoch": 0.022933292012086465,
+      "grad_norm": 0.5599532723426819,
+      "learning_rate": 9.999170368241389e-06,
+      "loss": 2.3352,
+      "mean_token_accuracy": 0.4913885644637048,
+      "num_tokens": 192977065.0,
+      "step": 1332
+    },
+    {
+      "entropy": 2.492431640625,
+      "epoch": 0.022950509198281725,
+      "grad_norm": 0.5502541661262512,
+      "learning_rate": 9.999165347928028e-06,
+      "loss": 2.4414,
+      "mean_token_accuracy": 0.4727879990823567,
+      "num_tokens": 193116265.0,
+      "step": 1333
+    },
+    {
+      "entropy": 2.4232177734375,
+      "epoch": 0.022967726384476986,
+      "grad_norm": 0.5547073483467102,
+      "learning_rate": 9.999160312472121e-06,
+      "loss": 2.3541,
+      "mean_token_accuracy": 0.48737787129357457,
+      "num_tokens": 193276114.0,
+      "step": 1334
+    },
+    {
+      "entropy": 2.4656982421875,
+      "epoch": 0.022984943570672246,
+      "grad_norm": 0.5852370858192444,
+      "learning_rate": 9.999155261873682e-06,
+      "loss": 2.446,
+      "mean_token_accuracy": 0.475621965713799,
+      "num_tokens": 193440455.0,
+      "step": 1335
+    },
+    {
+      "entropy": 2.421630859375,
+      "epoch": 0.023002160756867507,
+      "grad_norm": 0.5449553728103638,
+      "learning_rate": 9.999150196132731e-06,
+      "loss": 2.3845,
+      "mean_token_accuracy": 0.4841674976050854,
+      "num_tokens": 193607915.0,
+      "step": 1336
+    },
+    {
+      "entropy": 2.4388427734375,
+      "epoch": 0.023019377943062764,
+      "grad_norm": 0.5830256342887878,
+      "learning_rate": 9.999145115249278e-06,
+      "loss": 2.4171,
+      "mean_token_accuracy": 0.4808310712687671,
+      "num_tokens": 193747035.0,
+      "step": 1337
+    },
+    {
+      "entropy": 2.4093017578125,
+      "epoch": 0.023036595129258024,
+      "grad_norm": 0.5849303603172302,
+      "learning_rate": 9.999140019223343e-06,
+      "loss": 2.3965,
+      "mean_token_accuracy": 0.48324496299028397,
+      "num_tokens": 193882217.0,
+      "step": 1338
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 0.023053812315453285,
+      "grad_norm": 0.5599417686462402,
+      "learning_rate": 9.999134908054939e-06,
+      "loss": 2.4355,
+      "mean_token_accuracy": 0.47678020503371954,
+      "num_tokens": 194037678.0,
+      "step": 1339
+    },
+    {
+      "entropy": 2.40771484375,
+      "epoch": 0.023071029501648545,
+      "grad_norm": 0.635410726070404,
+      "learning_rate": 9.999129781744081e-06,
+      "loss": 2.3913,
+      "mean_token_accuracy": 0.48584868013858795,
+      "num_tokens": 194181984.0,
+      "step": 1340
+    },
+    {
+      "entropy": 2.4732666015625,
+      "epoch": 0.023088246687843806,
+      "grad_norm": 0.6021797060966492,
+      "learning_rate": 9.999124640290787e-06,
+      "loss": 2.4765,
+      "mean_token_accuracy": 0.47223041532561183,
+      "num_tokens": 194330357.0,
+      "step": 1341
+    },
+    {
+      "entropy": 2.4163818359375,
+      "epoch": 0.023105463874039067,
+      "grad_norm": 0.6050888895988464,
+      "learning_rate": 9.99911948369507e-06,
+      "loss": 2.4004,
+      "mean_token_accuracy": 0.4854390062391758,
+      "num_tokens": 194466254.0,
+      "step": 1342
+    },
+    {
+      "entropy": 2.4708251953125,
+      "epoch": 0.023122681060234327,
+      "grad_norm": 0.5668588876724243,
+      "learning_rate": 9.999114311956946e-06,
+      "loss": 2.4037,
+      "mean_token_accuracy": 0.48454291047528386,
+      "num_tokens": 194616186.0,
+      "step": 1343
+    },
+    {
+      "entropy": 2.50146484375,
+      "epoch": 0.023139898246429588,
+      "grad_norm": 0.5625669956207275,
+      "learning_rate": 9.99910912507643e-06,
+      "loss": 2.5026,
+      "mean_token_accuracy": 0.4760167011991143,
+      "num_tokens": 194772091.0,
+      "step": 1344
+    },
+    {
+      "entropy": 2.4669189453125,
+      "epoch": 0.023157115432624845,
+      "grad_norm": 0.6059160828590393,
+      "learning_rate": 9.999103923053541e-06,
+      "loss": 2.4043,
+      "mean_token_accuracy": 0.4812000300735235,
+      "num_tokens": 194918670.0,
+      "step": 1345
+    },
+    {
+      "entropy": 2.41552734375,
+      "epoch": 0.023174332618820105,
+      "grad_norm": 0.5639379620552063,
+      "learning_rate": 9.999098705888293e-06,
+      "loss": 2.3719,
+      "mean_token_accuracy": 0.4815536108799279,
+      "num_tokens": 195073659.0,
+      "step": 1346
+    },
+    {
+      "entropy": 2.4085693359375,
+      "epoch": 0.023191549805015366,
+      "grad_norm": 0.5673343539237976,
+      "learning_rate": 9.9990934735807e-06,
+      "loss": 2.3906,
+      "mean_token_accuracy": 0.4839661535806954,
+      "num_tokens": 195215626.0,
+      "step": 1347
+    },
+    {
+      "entropy": 2.357421875,
+      "epoch": 0.023208766991210626,
+      "grad_norm": 0.54830002784729,
+      "learning_rate": 9.99908822613078e-06,
+      "loss": 2.3375,
+      "mean_token_accuracy": 0.49586250027641654,
+      "num_tokens": 195371848.0,
+      "step": 1348
+    },
+    {
+      "entropy": 2.517578125,
+      "epoch": 0.023225984177405887,
+      "grad_norm": 0.5844534039497375,
+      "learning_rate": 9.999082963538548e-06,
+      "loss": 2.4989,
+      "mean_token_accuracy": 0.46372106671333313,
+      "num_tokens": 195524114.0,
+      "step": 1349
+    },
+    {
+      "entropy": 2.4599609375,
+      "epoch": 0.023243201363601147,
+      "grad_norm": 0.5597494840621948,
+      "learning_rate": 9.99907768580402e-06,
+      "loss": 2.4202,
+      "mean_token_accuracy": 0.4795773741789162,
+      "num_tokens": 195668084.0,
+      "step": 1350
+    },
+    {
+      "entropy": 2.4263916015625,
+      "epoch": 0.023260418549796408,
+      "grad_norm": 0.5759093165397644,
+      "learning_rate": 9.999072392927213e-06,
+      "loss": 2.3863,
+      "mean_token_accuracy": 0.48264182545244694,
+      "num_tokens": 195819421.0,
+      "step": 1351
+    },
+    {
+      "entropy": 2.4315185546875,
+      "epoch": 0.02327763573599167,
+      "grad_norm": 0.5471842288970947,
+      "learning_rate": 9.999067084908141e-06,
+      "loss": 2.3874,
+      "mean_token_accuracy": 0.48090588115155697,
+      "num_tokens": 195965275.0,
+      "step": 1352
+    },
+    {
+      "entropy": 2.430908203125,
+      "epoch": 0.023294852922186925,
+      "grad_norm": 0.5952867865562439,
+      "learning_rate": 9.999061761746822e-06,
+      "loss": 2.3695,
+      "mean_token_accuracy": 0.48542801523581147,
+      "num_tokens": 196100273.0,
+      "step": 1353
+    },
+    {
+      "entropy": 2.4898681640625,
+      "epoch": 0.023312070108382186,
+      "grad_norm": 0.5692234635353088,
+      "learning_rate": 9.999056423443272e-06,
+      "loss": 2.4315,
+      "mean_token_accuracy": 0.47685773856937885,
+      "num_tokens": 196241976.0,
+      "step": 1354
+    },
+    {
+      "entropy": 2.411376953125,
+      "epoch": 0.023329287294577446,
+      "grad_norm": 0.5885135531425476,
+      "learning_rate": 9.999051069997505e-06,
+      "loss": 2.3328,
+      "mean_token_accuracy": 0.48680604388937354,
+      "num_tokens": 196397855.0,
+      "step": 1355
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.023346504480772707,
+      "grad_norm": 0.5604063272476196,
+      "learning_rate": 9.999045701409539e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.48355547012761235,
+      "num_tokens": 196557337.0,
+      "step": 1356
+    },
+    {
+      "entropy": 2.4569091796875,
+      "epoch": 0.023363721666967967,
+      "grad_norm": 0.5709223747253418,
+      "learning_rate": 9.99904031767939e-06,
+      "loss": 2.4283,
+      "mean_token_accuracy": 0.47822416480630636,
+      "num_tokens": 196698366.0,
+      "step": 1357
+    },
+    {
+      "entropy": 2.3804931640625,
+      "epoch": 0.023380938853163228,
+      "grad_norm": 0.5742472410202026,
+      "learning_rate": 9.999034918807075e-06,
+      "loss": 2.3166,
+      "mean_token_accuracy": 0.49618813674896955,
+      "num_tokens": 196847854.0,
+      "step": 1358
+    },
+    {
+      "entropy": 2.4959716796875,
+      "epoch": 0.02339815603935849,
+      "grad_norm": 0.5980573892593384,
+      "learning_rate": 9.99902950479261e-06,
+      "loss": 2.4141,
+      "mean_token_accuracy": 0.4758835444226861,
+      "num_tokens": 196971667.0,
+      "step": 1359
+    },
+    {
+      "entropy": 2.532470703125,
+      "epoch": 0.02341537322555375,
+      "grad_norm": 0.5753028988838196,
+      "learning_rate": 9.99902407563601e-06,
+      "loss": 2.5277,
+      "mean_token_accuracy": 0.473021827172488,
+      "num_tokens": 197117509.0,
+      "step": 1360
+    },
+    {
+      "entropy": 2.3953857421875,
+      "epoch": 0.02343259041174901,
+      "grad_norm": 0.6161779761314392,
+      "learning_rate": 9.999018631337294e-06,
+      "loss": 2.3512,
+      "mean_token_accuracy": 0.4856503955088556,
+      "num_tokens": 197251978.0,
+      "step": 1361
+    },
+    {
+      "entropy": 2.5050048828125,
+      "epoch": 0.023449807597944267,
+      "grad_norm": 0.5542529225349426,
+      "learning_rate": 9.999013171896476e-06,
+      "loss": 2.5021,
+      "mean_token_accuracy": 0.46956229070201516,
+      "num_tokens": 197391278.0,
+      "step": 1362
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.023467024784139527,
+      "grad_norm": 0.5800600051879883,
+      "learning_rate": 9.999007697313573e-06,
+      "loss": 2.3979,
+      "mean_token_accuracy": 0.48410065239295363,
+      "num_tokens": 197531015.0,
+      "step": 1363
+    },
+    {
+      "entropy": 2.4874267578125,
+      "epoch": 0.023484241970334788,
+      "grad_norm": 0.6007982492446899,
+      "learning_rate": 9.999002207588603e-06,
+      "loss": 2.423,
+      "mean_token_accuracy": 0.4762545581907034,
+      "num_tokens": 197661883.0,
+      "step": 1364
+    },
+    {
+      "entropy": 2.4798583984375,
+      "epoch": 0.023501459156530048,
+      "grad_norm": 0.6128295660018921,
+      "learning_rate": 9.998996702721582e-06,
+      "loss": 2.4135,
+      "mean_token_accuracy": 0.4799572662450373,
+      "num_tokens": 197799526.0,
+      "step": 1365
+    },
+    {
+      "entropy": 2.499267578125,
+      "epoch": 0.02351867634272531,
+      "grad_norm": 0.574564516544342,
+      "learning_rate": 9.998991182712526e-06,
+      "loss": 2.4331,
+      "mean_token_accuracy": 0.4757503070868552,
+      "num_tokens": 197966743.0,
+      "step": 1366
+    },
+    {
+      "entropy": 2.4637451171875,
+      "epoch": 0.02353589352892057,
+      "grad_norm": 0.5752230882644653,
+      "learning_rate": 9.998985647561453e-06,
+      "loss": 2.438,
+      "mean_token_accuracy": 0.4770435835234821,
+      "num_tokens": 198124743.0,
+      "step": 1367
+    },
+    {
+      "entropy": 2.42333984375,
+      "epoch": 0.02355311071511583,
+      "grad_norm": 0.5870989561080933,
+      "learning_rate": 9.998980097268375e-06,
+      "loss": 2.3783,
+      "mean_token_accuracy": 0.487036875449121,
+      "num_tokens": 198265217.0,
+      "step": 1368
+    },
+    {
+      "entropy": 2.399169921875,
+      "epoch": 0.02357032790131109,
+      "grad_norm": 0.5645302534103394,
+      "learning_rate": 9.998974531833316e-06,
+      "loss": 2.3897,
+      "mean_token_accuracy": 0.487074441742152,
+      "num_tokens": 198402224.0,
+      "step": 1369
+    },
+    {
+      "entropy": 2.40234375,
+      "epoch": 0.023587545087506347,
+      "grad_norm": 0.6568628549575806,
+      "learning_rate": 9.99896895125629e-06,
+      "loss": 2.3445,
+      "mean_token_accuracy": 0.4943762863986194,
+      "num_tokens": 198536773.0,
+      "step": 1370
+    },
+    {
+      "entropy": 2.462890625,
+      "epoch": 0.023604762273701608,
+      "grad_norm": 0.5660800337791443,
+      "learning_rate": 9.998963355537313e-06,
+      "loss": 2.4593,
+      "mean_token_accuracy": 0.4756892016157508,
+      "num_tokens": 198690958.0,
+      "step": 1371
+    },
+    {
+      "entropy": 2.48193359375,
+      "epoch": 0.02362197945989687,
+      "grad_norm": 0.6144444346427917,
+      "learning_rate": 9.998957744676403e-06,
+      "loss": 2.4622,
+      "mean_token_accuracy": 0.4751414889469743,
+      "num_tokens": 198845530.0,
+      "step": 1372
+    },
+    {
+      "entropy": 2.463623046875,
+      "epoch": 0.02363919664609213,
+      "grad_norm": 0.6004127264022827,
+      "learning_rate": 9.998952118673575e-06,
+      "loss": 2.4163,
+      "mean_token_accuracy": 0.4740082244388759,
+      "num_tokens": 198973843.0,
+      "step": 1373
+    },
+    {
+      "entropy": 2.4368896484375,
+      "epoch": 0.02365641383228739,
+      "grad_norm": 0.5763023495674133,
+      "learning_rate": 9.998946477528848e-06,
+      "loss": 2.4106,
+      "mean_token_accuracy": 0.4760644896887243,
+      "num_tokens": 199120869.0,
+      "step": 1374
+    },
+    {
+      "entropy": 2.455078125,
+      "epoch": 0.02367363101848265,
+      "grad_norm": 0.5937883257865906,
+      "learning_rate": 9.99894082124224e-06,
+      "loss": 2.4061,
+      "mean_token_accuracy": 0.47505834326148033,
+      "num_tokens": 199271029.0,
+      "step": 1375
+    },
+    {
+      "entropy": 2.4976806640625,
+      "epoch": 0.02369084820467791,
+      "grad_norm": 0.5722872614860535,
+      "learning_rate": 9.998935149813766e-06,
+      "loss": 2.4279,
+      "mean_token_accuracy": 0.4803337692283094,
+      "num_tokens": 199421781.0,
+      "step": 1376
+    },
+    {
+      "entropy": 2.4547119140625,
+      "epoch": 0.02370806539087317,
+      "grad_norm": 0.5956291556358337,
+      "learning_rate": 9.998929463243443e-06,
+      "loss": 2.4151,
+      "mean_token_accuracy": 0.478096229955554,
+      "num_tokens": 199575267.0,
+      "step": 1377
+    },
+    {
+      "entropy": 2.51513671875,
+      "epoch": 0.023725282577068428,
+      "grad_norm": 0.5729405879974365,
+      "learning_rate": 9.99892376153129e-06,
+      "loss": 2.4828,
+      "mean_token_accuracy": 0.4753085137344897,
+      "num_tokens": 199718432.0,
+      "step": 1378
+    },
+    {
+      "entropy": 2.4134521484375,
+      "epoch": 0.02374249976326369,
+      "grad_norm": 0.5974603295326233,
+      "learning_rate": 9.998918044677321e-06,
+      "loss": 2.3474,
+      "mean_token_accuracy": 0.4921872243285179,
+      "num_tokens": 199860520.0,
+      "step": 1379
+    },
+    {
+      "entropy": 2.4600830078125,
+      "epoch": 0.02375971694945895,
+      "grad_norm": 0.5969933271408081,
+      "learning_rate": 9.99891231268156e-06,
+      "loss": 2.4926,
+      "mean_token_accuracy": 0.4793005189858377,
+      "num_tokens": 200000679.0,
+      "step": 1380
+    },
+    {
+      "entropy": 2.45556640625,
+      "epoch": 0.02377693413565421,
+      "grad_norm": 0.5951429605484009,
+      "learning_rate": 9.998906565544017e-06,
+      "loss": 2.4085,
+      "mean_token_accuracy": 0.4794029123149812,
+      "num_tokens": 200145478.0,
+      "step": 1381
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.02379415132184947,
+      "grad_norm": 0.6001779437065125,
+      "learning_rate": 9.998900803264714e-06,
+      "loss": 2.4392,
+      "mean_token_accuracy": 0.4742898163385689,
+      "num_tokens": 200297504.0,
+      "step": 1382
+    },
+    {
+      "entropy": 2.4384765625,
+      "epoch": 0.02381136850804473,
+      "grad_norm": 0.5844825506210327,
+      "learning_rate": 9.998895025843664e-06,
+      "loss": 2.3799,
+      "mean_token_accuracy": 0.4815810131840408,
+      "num_tokens": 200438128.0,
+      "step": 1383
+    },
+    {
+      "entropy": 2.439453125,
+      "epoch": 0.02382858569423999,
+      "grad_norm": 0.5872356295585632,
+      "learning_rate": 9.99888923328089e-06,
+      "loss": 2.375,
+      "mean_token_accuracy": 0.48686960944905877,
+      "num_tokens": 200576616.0,
+      "step": 1384
+    },
+    {
+      "entropy": 2.4556884765625,
+      "epoch": 0.023845802880435252,
+      "grad_norm": 0.5676760077476501,
+      "learning_rate": 9.998883425576407e-06,
+      "loss": 2.4051,
+      "mean_token_accuracy": 0.47929678950458765,
+      "num_tokens": 200715807.0,
+      "step": 1385
+    },
+    {
+      "entropy": 2.5262451171875,
+      "epoch": 0.023863020066630512,
+      "grad_norm": 0.6212140321731567,
+      "learning_rate": 9.998877602730231e-06,
+      "loss": 2.5013,
+      "mean_token_accuracy": 0.46635109139606357,
+      "num_tokens": 200852036.0,
+      "step": 1386
+    },
+    {
+      "entropy": 2.466796875,
+      "epoch": 0.02388023725282577,
+      "grad_norm": 0.5637140870094299,
+      "learning_rate": 9.998871764742383e-06,
+      "loss": 2.4031,
+      "mean_token_accuracy": 0.4789075516164303,
+      "num_tokens": 201005894.0,
+      "step": 1387
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.02389745443902103,
+      "grad_norm": 0.554351806640625,
+      "learning_rate": 9.998865911612878e-06,
+      "loss": 2.334,
+      "mean_token_accuracy": 0.48602917743846774,
+      "num_tokens": 201163254.0,
+      "step": 1388
+    },
+    {
+      "entropy": 2.4622802734375,
+      "epoch": 0.02391467162521629,
+      "grad_norm": 0.5873886346817017,
+      "learning_rate": 9.998860043341733e-06,
+      "loss": 2.4122,
+      "mean_token_accuracy": 0.4796907198615372,
+      "num_tokens": 201311034.0,
+      "step": 1389
+    },
+    {
+      "entropy": 2.5067138671875,
+      "epoch": 0.02393188881141155,
+      "grad_norm": 0.589946985244751,
+      "learning_rate": 9.99885415992897e-06,
+      "loss": 2.4737,
+      "mean_token_accuracy": 0.46914456551894546,
+      "num_tokens": 201439256.0,
+      "step": 1390
+    },
+    {
+      "entropy": 2.5040283203125,
+      "epoch": 0.02394910599760681,
+      "grad_norm": 0.5907305479049683,
+      "learning_rate": 9.998848261374602e-06,
+      "loss": 2.5056,
+      "mean_token_accuracy": 0.4672775506041944,
+      "num_tokens": 201575176.0,
+      "step": 1391
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.023966323183802072,
+      "grad_norm": 0.5816705822944641,
+      "learning_rate": 9.998842347678652e-06,
+      "loss": 2.4009,
+      "mean_token_accuracy": 0.48051235545426607,
+      "num_tokens": 201741972.0,
+      "step": 1392
+    },
+    {
+      "entropy": 2.532470703125,
+      "epoch": 0.023983540369997332,
+      "grad_norm": 0.5787369012832642,
+      "learning_rate": 9.998836418841133e-06,
+      "loss": 2.5186,
+      "mean_token_accuracy": 0.46860673651099205,
+      "num_tokens": 201887196.0,
+      "step": 1393
+    },
+    {
+      "entropy": 2.3990478515625,
+      "epoch": 0.024000757556192593,
+      "grad_norm": 0.6086745262145996,
+      "learning_rate": 9.998830474862064e-06,
+      "loss": 2.3506,
+      "mean_token_accuracy": 0.4886458469554782,
+      "num_tokens": 202021347.0,
+      "step": 1394
+    },
+    {
+      "entropy": 2.41748046875,
+      "epoch": 0.02401797474238785,
+      "grad_norm": 0.5796844363212585,
+      "learning_rate": 9.998824515741467e-06,
+      "loss": 2.3866,
+      "mean_token_accuracy": 0.4884433811530471,
+      "num_tokens": 202160896.0,
+      "step": 1395
+    },
+    {
+      "entropy": 2.5753173828125,
+      "epoch": 0.02403519192858311,
+      "grad_norm": 0.5776175856590271,
+      "learning_rate": 9.998818541479355e-06,
+      "loss": 2.5233,
+      "mean_token_accuracy": 0.4616355444304645,
+      "num_tokens": 202292953.0,
+      "step": 1396
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.02405240911477837,
+      "grad_norm": 0.5363594889640808,
+      "learning_rate": 9.998812552075747e-06,
+      "loss": 2.3689,
+      "mean_token_accuracy": 0.4840633156709373,
+      "num_tokens": 202460982.0,
+      "step": 1397
+    },
+    {
+      "entropy": 2.39208984375,
+      "epoch": 0.02406962630097363,
+      "grad_norm": 0.5426232218742371,
+      "learning_rate": 9.998806547530664e-06,
+      "loss": 2.3481,
+      "mean_token_accuracy": 0.48746701143682003,
+      "num_tokens": 202621458.0,
+      "step": 1398
+    },
+    {
+      "entropy": 2.4622802734375,
+      "epoch": 0.024086843487168892,
+      "grad_norm": 0.5721414089202881,
+      "learning_rate": 9.998800527844122e-06,
+      "loss": 2.4094,
+      "mean_token_accuracy": 0.4796240641735494,
+      "num_tokens": 202756740.0,
+      "step": 1399
+    },
+    {
+      "entropy": 2.363037109375,
+      "epoch": 0.024104060673364153,
+      "grad_norm": 0.5773559212684631,
+      "learning_rate": 9.99879449301614e-06,
+      "loss": 2.3559,
+      "mean_token_accuracy": 0.49749248987063766,
+      "num_tokens": 202900096.0,
+      "step": 1400
+    },
+    {
+      "entropy": 2.53125,
+      "epoch": 0.024121277859559413,
+      "grad_norm": 0.611416220664978,
+      "learning_rate": 9.998788443046735e-06,
+      "loss": 2.5107,
+      "mean_token_accuracy": 0.4666637545451522,
+      "num_tokens": 203038362.0,
+      "step": 1401
+    },
+    {
+      "entropy": 2.4573974609375,
+      "epoch": 0.024138495045754674,
+      "grad_norm": 0.6409509778022766,
+      "learning_rate": 9.998782377935927e-06,
+      "loss": 2.423,
+      "mean_token_accuracy": 0.4764982839114964,
+      "num_tokens": 203176454.0,
+      "step": 1402
+    },
+    {
+      "entropy": 2.419921875,
+      "epoch": 0.02415571223194993,
+      "grad_norm": 0.5751169323921204,
+      "learning_rate": 9.998776297683733e-06,
+      "loss": 2.3627,
+      "mean_token_accuracy": 0.48058157647028565,
+      "num_tokens": 203311141.0,
+      "step": 1403
+    },
+    {
+      "entropy": 2.4736328125,
+      "epoch": 0.02417292941814519,
+      "grad_norm": 0.580534040927887,
+      "learning_rate": 9.998770202290173e-06,
+      "loss": 2.449,
+      "mean_token_accuracy": 0.4753856221213937,
+      "num_tokens": 203452951.0,
+      "step": 1404
+    },
+    {
+      "entropy": 2.426513671875,
+      "epoch": 0.024190146604340452,
+      "grad_norm": 0.5762086510658264,
+      "learning_rate": 9.998764091755264e-06,
+      "loss": 2.3869,
+      "mean_token_accuracy": 0.4834116967394948,
+      "num_tokens": 203600387.0,
+      "step": 1405
+    },
+    {
+      "entropy": 2.509521484375,
+      "epoch": 0.024207363790535712,
+      "grad_norm": 0.5431259870529175,
+      "learning_rate": 9.998757966079024e-06,
+      "loss": 2.486,
+      "mean_token_accuracy": 0.4661334496922791,
+      "num_tokens": 203751350.0,
+      "step": 1406
+    },
+    {
+      "entropy": 2.499267578125,
+      "epoch": 0.024224580976730973,
+      "grad_norm": 0.6089497804641724,
+      "learning_rate": 9.998751825261474e-06,
+      "loss": 2.4389,
+      "mean_token_accuracy": 0.4734781333245337,
+      "num_tokens": 203902498.0,
+      "step": 1407
+    },
+    {
+      "entropy": 2.48583984375,
+      "epoch": 0.024241798162926233,
+      "grad_norm": 0.5681830048561096,
+      "learning_rate": 9.998745669302632e-06,
+      "loss": 2.4445,
+      "mean_token_accuracy": 0.4708978096023202,
+      "num_tokens": 204044975.0,
+      "step": 1408
+    },
+    {
+      "entropy": 2.4814453125,
+      "epoch": 0.024259015349121494,
+      "grad_norm": 0.5554722547531128,
+      "learning_rate": 9.998739498202514e-06,
+      "loss": 2.4522,
+      "mean_token_accuracy": 0.47262357315048575,
+      "num_tokens": 204205348.0,
+      "step": 1409
+    },
+    {
+      "entropy": 2.444580078125,
+      "epoch": 0.024276232535316754,
+      "grad_norm": 0.5820813775062561,
+      "learning_rate": 9.99873331196114e-06,
+      "loss": 2.3785,
+      "mean_token_accuracy": 0.4880591509863734,
+      "num_tokens": 204345568.0,
+      "step": 1410
+    },
+    {
+      "entropy": 2.45556640625,
+      "epoch": 0.024293449721512015,
+      "grad_norm": 0.5614075660705566,
+      "learning_rate": 9.99872711057853e-06,
+      "loss": 2.4034,
+      "mean_token_accuracy": 0.4788000574335456,
+      "num_tokens": 204494358.0,
+      "step": 1411
+    },
+    {
+      "entropy": 2.4716796875,
+      "epoch": 0.024310666907707272,
+      "grad_norm": 0.5851388573646545,
+      "learning_rate": 9.998720894054703e-06,
+      "loss": 2.4773,
+      "mean_token_accuracy": 0.4750785259529948,
+      "num_tokens": 204626789.0,
+      "step": 1412
+    },
+    {
+      "entropy": 2.45458984375,
+      "epoch": 0.024327884093902533,
+      "grad_norm": 0.7529696822166443,
+      "learning_rate": 9.998714662389676e-06,
+      "loss": 2.4149,
+      "mean_token_accuracy": 0.4763760110363364,
+      "num_tokens": 204774398.0,
+      "step": 1413
+    },
+    {
+      "entropy": 2.4195556640625,
+      "epoch": 0.024345101280097793,
+      "grad_norm": 0.6168255805969238,
+      "learning_rate": 9.99870841558347e-06,
+      "loss": 2.3811,
+      "mean_token_accuracy": 0.4846136327832937,
+      "num_tokens": 204924934.0,
+      "step": 1414
+    },
+    {
+      "entropy": 2.4605712890625,
+      "epoch": 0.024362318466293054,
+      "grad_norm": 0.6109985113143921,
+      "learning_rate": 9.998702153636102e-06,
+      "loss": 2.4489,
+      "mean_token_accuracy": 0.4767269352450967,
+      "num_tokens": 205059906.0,
+      "step": 1415
+    },
+    {
+      "entropy": 2.436767578125,
+      "epoch": 0.024379535652488314,
+      "grad_norm": 0.5300309658050537,
+      "learning_rate": 9.998695876547591e-06,
+      "loss": 2.3666,
+      "mean_token_accuracy": 0.4840262867510319,
+      "num_tokens": 205220596.0,
+      "step": 1416
+    },
+    {
+      "entropy": 2.429931640625,
+      "epoch": 0.024396752838683575,
+      "grad_norm": 0.5640124678611755,
+      "learning_rate": 9.998689584317955e-06,
+      "loss": 2.3631,
+      "mean_token_accuracy": 0.49288097163662314,
+      "num_tokens": 205386977.0,
+      "step": 1417
+    },
+    {
+      "entropy": 2.46435546875,
+      "epoch": 0.024413970024878835,
+      "grad_norm": 0.5903874635696411,
+      "learning_rate": 9.998683276947219e-06,
+      "loss": 2.4291,
+      "mean_token_accuracy": 0.47703268751502037,
+      "num_tokens": 205526617.0,
+      "step": 1418
+    },
+    {
+      "entropy": 2.4453125,
+      "epoch": 0.024431187211074096,
+      "grad_norm": 0.567484438419342,
+      "learning_rate": 9.998676954435394e-06,
+      "loss": 2.3968,
+      "mean_token_accuracy": 0.4830042072571814,
+      "num_tokens": 205669600.0,
+      "step": 1419
+    },
+    {
+      "entropy": 2.4515380859375,
+      "epoch": 0.024448404397269353,
+      "grad_norm": 0.5405850410461426,
+      "learning_rate": 9.998670616782505e-06,
+      "loss": 2.3772,
+      "mean_token_accuracy": 0.4838403551839292,
+      "num_tokens": 205821251.0,
+      "step": 1420
+    },
+    {
+      "entropy": 2.4334716796875,
+      "epoch": 0.024465621583464613,
+      "grad_norm": 0.5751288533210754,
+      "learning_rate": 9.99866426398857e-06,
+      "loss": 2.3644,
+      "mean_token_accuracy": 0.48748740553855896,
+      "num_tokens": 205964418.0,
+      "step": 1421
+    },
+    {
+      "entropy": 2.474853515625,
+      "epoch": 0.024482838769659874,
+      "grad_norm": 0.5722588896751404,
+      "learning_rate": 9.998657896053604e-06,
+      "loss": 2.4274,
+      "mean_token_accuracy": 0.47551520401611924,
+      "num_tokens": 206095207.0,
+      "step": 1422
+    },
+    {
+      "entropy": 2.470458984375,
+      "epoch": 0.024500055955855134,
+      "grad_norm": 0.6179440021514893,
+      "learning_rate": 9.998651512977631e-06,
+      "loss": 2.4267,
+      "mean_token_accuracy": 0.47643746761605144,
+      "num_tokens": 206240116.0,
+      "step": 1423
+    },
+    {
+      "entropy": 2.4114990234375,
+      "epoch": 0.024517273142050395,
+      "grad_norm": 0.608542799949646,
+      "learning_rate": 9.99864511476067e-06,
+      "loss": 2.3299,
+      "mean_token_accuracy": 0.4850447475910187,
+      "num_tokens": 206389712.0,
+      "step": 1424
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.024534490328245655,
+      "grad_norm": 0.5260599851608276,
+      "learning_rate": 9.998638701402739e-06,
+      "loss": 2.4158,
+      "mean_token_accuracy": 0.4785124696791172,
+      "num_tokens": 206548123.0,
+      "step": 1425
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.024551707514440916,
+      "grad_norm": 0.6076765060424805,
+      "learning_rate": 9.998632272903858e-06,
+      "loss": 2.3619,
+      "mean_token_accuracy": 0.49168099416419864,
+      "num_tokens": 206687271.0,
+      "step": 1426
+    },
+    {
+      "entropy": 2.499755859375,
+      "epoch": 0.024568924700636176,
+      "grad_norm": 0.6022489070892334,
+      "learning_rate": 9.998625829264045e-06,
+      "loss": 2.487,
+      "mean_token_accuracy": 0.46968710515648127,
+      "num_tokens": 206824438.0,
+      "step": 1427
+    },
+    {
+      "entropy": 2.4072265625,
+      "epoch": 0.024586141886831434,
+      "grad_norm": 0.5461668372154236,
+      "learning_rate": 9.998619370483324e-06,
+      "loss": 2.3735,
+      "mean_token_accuracy": 0.49147024331614375,
+      "num_tokens": 206981519.0,
+      "step": 1428
+    },
+    {
+      "entropy": 2.4703369140625,
+      "epoch": 0.024603359073026694,
+      "grad_norm": 0.5643055438995361,
+      "learning_rate": 9.998612896561709e-06,
+      "loss": 2.4402,
+      "mean_token_accuracy": 0.47540956176817417,
+      "num_tokens": 207129316.0,
+      "step": 1429
+    },
+    {
+      "entropy": 2.430908203125,
+      "epoch": 0.024620576259221955,
+      "grad_norm": 0.5822505950927734,
+      "learning_rate": 9.998606407499222e-06,
+      "loss": 2.3844,
+      "mean_token_accuracy": 0.4864038904197514,
+      "num_tokens": 207277904.0,
+      "step": 1430
+    },
+    {
+      "entropy": 2.41015625,
+      "epoch": 0.024637793445417215,
+      "grad_norm": 0.5898188352584839,
+      "learning_rate": 9.998599903295883e-06,
+      "loss": 2.3588,
+      "mean_token_accuracy": 0.4909462621435523,
+      "num_tokens": 207422167.0,
+      "step": 1431
+    },
+    {
+      "entropy": 2.4686279296875,
+      "epoch": 0.024655010631612476,
+      "grad_norm": 0.5768987536430359,
+      "learning_rate": 9.99859338395171e-06,
+      "loss": 2.4193,
+      "mean_token_accuracy": 0.47896814439445734,
+      "num_tokens": 207560662.0,
+      "step": 1432
+    },
+    {
+      "entropy": 2.5447998046875,
+      "epoch": 0.024672227817807736,
+      "grad_norm": 0.582497239112854,
+      "learning_rate": 9.998586849466727e-06,
+      "loss": 2.5383,
+      "mean_token_accuracy": 0.46700617065653205,
+      "num_tokens": 207702226.0,
+      "step": 1433
+    },
+    {
+      "entropy": 2.4814453125,
+      "epoch": 0.024689445004002997,
+      "grad_norm": 0.6393965482711792,
+      "learning_rate": 9.99858029984095e-06,
+      "loss": 2.438,
+      "mean_token_accuracy": 0.47804489312693477,
+      "num_tokens": 207827846.0,
+      "step": 1434
+    },
+    {
+      "entropy": 2.3828125,
+      "epoch": 0.024706662190198257,
+      "grad_norm": 0.618248701095581,
+      "learning_rate": 9.998573735074398e-06,
+      "loss": 2.3306,
+      "mean_token_accuracy": 0.48714121896773577,
+      "num_tokens": 207996679.0,
+      "step": 1435
+    },
+    {
+      "entropy": 2.452880859375,
+      "epoch": 0.024723879376393518,
+      "grad_norm": 0.6095260381698608,
+      "learning_rate": 9.998567155167094e-06,
+      "loss": 2.3982,
+      "mean_token_accuracy": 0.48710576351732016,
+      "num_tokens": 208129844.0,
+      "step": 1436
+    },
+    {
+      "entropy": 2.4764404296875,
+      "epoch": 0.024741096562588775,
+      "grad_norm": 0.5430065393447876,
+      "learning_rate": 9.998560560119058e-06,
+      "loss": 2.4219,
+      "mean_token_accuracy": 0.47808440774679184,
+      "num_tokens": 208273997.0,
+      "step": 1437
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.024758313748784035,
+      "grad_norm": 0.6290302276611328,
+      "learning_rate": 9.998553949930306e-06,
+      "loss": 2.3994,
+      "mean_token_accuracy": 0.484294559340924,
+      "num_tokens": 208429606.0,
+      "step": 1438
+    },
+    {
+      "entropy": 2.431640625,
+      "epoch": 0.024775530934979296,
+      "grad_norm": 0.5840831398963928,
+      "learning_rate": 9.99854732460086e-06,
+      "loss": 2.3615,
+      "mean_token_accuracy": 0.4878291576169431,
+      "num_tokens": 208577096.0,
+      "step": 1439
+    },
+    {
+      "entropy": 2.4404296875,
+      "epoch": 0.024792748121174556,
+      "grad_norm": 0.5737127065658569,
+      "learning_rate": 9.998540684130743e-06,
+      "loss": 2.4082,
+      "mean_token_accuracy": 0.48507948173210025,
+      "num_tokens": 208744507.0,
+      "step": 1440
+    },
+    {
+      "entropy": 2.380126953125,
+      "epoch": 0.024809965307369817,
+      "grad_norm": 0.5842684507369995,
+      "learning_rate": 9.998534028519971e-06,
+      "loss": 2.3265,
+      "mean_token_accuracy": 0.49405472399666905,
+      "num_tokens": 208887356.0,
+      "step": 1441
+    },
+    {
+      "entropy": 2.447265625,
+      "epoch": 0.024827182493565077,
+      "grad_norm": 0.5664902329444885,
+      "learning_rate": 9.998527357768566e-06,
+      "loss": 2.4049,
+      "mean_token_accuracy": 0.4856785982847214,
+      "num_tokens": 209035653.0,
+      "step": 1442
+    },
+    {
+      "entropy": 2.5068359375,
+      "epoch": 0.024844399679760338,
+      "grad_norm": 0.5968948602676392,
+      "learning_rate": 9.99852067187655e-06,
+      "loss": 2.4616,
+      "mean_token_accuracy": 0.4748954540118575,
+      "num_tokens": 209174761.0,
+      "step": 1443
+    },
+    {
+      "entropy": 2.4140625,
+      "epoch": 0.0248616168659556,
+      "grad_norm": 0.5464600920677185,
+      "learning_rate": 9.99851397084394e-06,
+      "loss": 2.391,
+      "mean_token_accuracy": 0.4841625513508916,
+      "num_tokens": 209335128.0,
+      "step": 1444
+    },
+    {
+      "entropy": 2.406494140625,
+      "epoch": 0.024878834052150856,
+      "grad_norm": 0.5924455523490906,
+      "learning_rate": 9.998507254670757e-06,
+      "loss": 2.3687,
+      "mean_token_accuracy": 0.4873992637731135,
+      "num_tokens": 209494921.0,
+      "step": 1445
+    },
+    {
+      "entropy": 2.4461669921875,
+      "epoch": 0.024896051238346116,
+      "grad_norm": 0.5587389469146729,
+      "learning_rate": 9.998500523357022e-06,
+      "loss": 2.4399,
+      "mean_token_accuracy": 0.4791180179454386,
+      "num_tokens": 209636823.0,
+      "step": 1446
+    },
+    {
+      "entropy": 2.4124755859375,
+      "epoch": 0.024913268424541377,
+      "grad_norm": 0.5826576948165894,
+      "learning_rate": 9.998493776902756e-06,
+      "loss": 2.3668,
+      "mean_token_accuracy": 0.4858208396472037,
+      "num_tokens": 209782197.0,
+      "step": 1447
+    },
+    {
+      "entropy": 2.4521484375,
+      "epoch": 0.024930485610736637,
+      "grad_norm": 0.6207579374313354,
+      "learning_rate": 9.998487015307978e-06,
+      "loss": 2.392,
+      "mean_token_accuracy": 0.4869683226570487,
+      "num_tokens": 209916075.0,
+      "step": 1448
+    },
+    {
+      "entropy": 2.4453125,
+      "epoch": 0.024947702796931898,
+      "grad_norm": 0.640599250793457,
+      "learning_rate": 9.998480238572711e-06,
+      "loss": 2.383,
+      "mean_token_accuracy": 0.480794875882566,
+      "num_tokens": 210062565.0,
+      "step": 1449
+    },
+    {
+      "entropy": 2.4508056640625,
+      "epoch": 0.024964919983127158,
+      "grad_norm": 0.5667426586151123,
+      "learning_rate": 9.998473446696972e-06,
+      "loss": 2.4082,
+      "mean_token_accuracy": 0.4857399258762598,
+      "num_tokens": 210214070.0,
+      "step": 1450
+    },
+    {
+      "entropy": 2.4837646484375,
+      "epoch": 0.02498213716932242,
+      "grad_norm": 0.5633426904678345,
+      "learning_rate": 9.998466639680786e-06,
+      "loss": 2.4637,
+      "mean_token_accuracy": 0.47439560387283564,
+      "num_tokens": 210356264.0,
+      "step": 1451
+    },
+    {
+      "entropy": 2.3912353515625,
+      "epoch": 0.02499935435551768,
+      "grad_norm": 0.621870219707489,
+      "learning_rate": 9.998459817524168e-06,
+      "loss": 2.3623,
+      "mean_token_accuracy": 0.4901335104368627,
+      "num_tokens": 210509728.0,
+      "step": 1452
+    },
+    {
+      "entropy": 2.4949951171875,
+      "epoch": 0.025016571541712936,
+      "grad_norm": 0.5852912664413452,
+      "learning_rate": 9.998452980227143e-06,
+      "loss": 2.4597,
+      "mean_token_accuracy": 0.4733400009572506,
+      "num_tokens": 210644493.0,
+      "step": 1453
+    },
+    {
+      "entropy": 2.434326171875,
+      "epoch": 0.025033788727908197,
+      "grad_norm": 0.5888333320617676,
+      "learning_rate": 9.998446127789731e-06,
+      "loss": 2.4475,
+      "mean_token_accuracy": 0.4807679671794176,
+      "num_tokens": 210791477.0,
+      "step": 1454
+    },
+    {
+      "entropy": 2.44189453125,
+      "epoch": 0.025051005914103457,
+      "grad_norm": 0.5960559248924255,
+      "learning_rate": 9.998439260211953e-06,
+      "loss": 2.4365,
+      "mean_token_accuracy": 0.4799190401099622,
+      "num_tokens": 210939578.0,
+      "step": 1455
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.025068223100298718,
+      "grad_norm": 0.5698879957199097,
+      "learning_rate": 9.998432377493826e-06,
+      "loss": 2.3899,
+      "mean_token_accuracy": 0.4848698596470058,
+      "num_tokens": 211096780.0,
+      "step": 1456
+    },
+    {
+      "entropy": 2.3997802734375,
+      "epoch": 0.02508544028649398,
+      "grad_norm": 0.5995359420776367,
+      "learning_rate": 9.998425479635373e-06,
+      "loss": 2.383,
+      "mean_token_accuracy": 0.4919665399938822,
+      "num_tokens": 211234392.0,
+      "step": 1457
+    },
+    {
+      "entropy": 2.497802734375,
+      "epoch": 0.02510265747268924,
+      "grad_norm": 0.5807369351387024,
+      "learning_rate": 9.99841856663662e-06,
+      "loss": 2.4674,
+      "mean_token_accuracy": 0.47135368920862675,
+      "num_tokens": 211366377.0,
+      "step": 1458
+    },
+    {
+      "entropy": 2.4554443359375,
+      "epoch": 0.0251198746588845,
+      "grad_norm": 0.5541589260101318,
+      "learning_rate": 9.99841163849758e-06,
+      "loss": 2.4065,
+      "mean_token_accuracy": 0.48336717346683145,
+      "num_tokens": 211517256.0,
+      "step": 1459
+    },
+    {
+      "entropy": 2.473876953125,
+      "epoch": 0.02513709184507976,
+      "grad_norm": 0.597100555896759,
+      "learning_rate": 9.99840469521828e-06,
+      "loss": 2.4374,
+      "mean_token_accuracy": 0.4758867691271007,
+      "num_tokens": 211656203.0,
+      "step": 1460
+    },
+    {
+      "entropy": 2.459228515625,
+      "epoch": 0.025154309031275017,
+      "grad_norm": 0.5845798850059509,
+      "learning_rate": 9.998397736798737e-06,
+      "loss": 2.4206,
+      "mean_token_accuracy": 0.4736228142865002,
+      "num_tokens": 211798384.0,
+      "step": 1461
+    },
+    {
+      "entropy": 2.455078125,
+      "epoch": 0.025171526217470278,
+      "grad_norm": 0.6050572395324707,
+      "learning_rate": 9.998390763238975e-06,
+      "loss": 2.4037,
+      "mean_token_accuracy": 0.48261626344174147,
+      "num_tokens": 211932728.0,
+      "step": 1462
+    },
+    {
+      "entropy": 2.4322509765625,
+      "epoch": 0.025188743403665538,
+      "grad_norm": 0.5519930124282837,
+      "learning_rate": 9.998383774539013e-06,
+      "loss": 2.3796,
+      "mean_token_accuracy": 0.48328032717108727,
+      "num_tokens": 212075313.0,
+      "step": 1463
+    },
+    {
+      "entropy": 2.4688720703125,
+      "epoch": 0.0252059605898608,
+      "grad_norm": 0.5802918672561646,
+      "learning_rate": 9.998376770698875e-06,
+      "loss": 2.4487,
+      "mean_token_accuracy": 0.4819605387747288,
+      "num_tokens": 212220989.0,
+      "step": 1464
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.02522317777605606,
+      "grad_norm": 0.5660965442657471,
+      "learning_rate": 9.998369751718577e-06,
+      "loss": 2.3826,
+      "mean_token_accuracy": 0.48062465619295835,
+      "num_tokens": 212377908.0,
+      "step": 1465
+    },
+    {
+      "entropy": 2.5208740234375,
+      "epoch": 0.02524039496225132,
+      "grad_norm": 0.6206576228141785,
+      "learning_rate": 9.998362717598144e-06,
+      "loss": 2.5238,
+      "mean_token_accuracy": 0.46971421781927347,
+      "num_tokens": 212521169.0,
+      "step": 1466
+    },
+    {
+      "entropy": 2.45263671875,
+      "epoch": 0.02525761214844658,
+      "grad_norm": 0.5654458403587341,
+      "learning_rate": 9.998355668337599e-06,
+      "loss": 2.4136,
+      "mean_token_accuracy": 0.4812098955735564,
+      "num_tokens": 212661516.0,
+      "step": 1467
+    },
+    {
+      "entropy": 2.4288330078125,
+      "epoch": 0.02527482933464184,
+      "grad_norm": 0.5742049217224121,
+      "learning_rate": 9.99834860393696e-06,
+      "loss": 2.3939,
+      "mean_token_accuracy": 0.4802519157528877,
+      "num_tokens": 212802767.0,
+      "step": 1468
+    },
+    {
+      "entropy": 2.47021484375,
+      "epoch": 0.0252920465208371,
+      "grad_norm": 0.5584784150123596,
+      "learning_rate": 9.998341524396249e-06,
+      "loss": 2.4361,
+      "mean_token_accuracy": 0.47756797447800636,
+      "num_tokens": 212937639.0,
+      "step": 1469
+    },
+    {
+      "entropy": 2.4468994140625,
+      "epoch": 0.025309263707032358,
+      "grad_norm": 0.7078640460968018,
+      "learning_rate": 9.998334429715488e-06,
+      "loss": 2.3877,
+      "mean_token_accuracy": 0.4793696654960513,
+      "num_tokens": 213079562.0,
+      "step": 1470
+    },
+    {
+      "entropy": 2.39892578125,
+      "epoch": 0.02532648089322762,
+      "grad_norm": 0.5251427292823792,
+      "learning_rate": 9.998327319894699e-06,
+      "loss": 2.3489,
+      "mean_token_accuracy": 0.4894181271083653,
+      "num_tokens": 213245711.0,
+      "step": 1471
+    },
+    {
+      "entropy": 2.3956298828125,
+      "epoch": 0.02534369807942288,
+      "grad_norm": 0.5619721412658691,
+      "learning_rate": 9.998320194933904e-06,
+      "loss": 2.3521,
+      "mean_token_accuracy": 0.49081948725506663,
+      "num_tokens": 213387544.0,
+      "step": 1472
+    },
+    {
+      "entropy": 2.41552734375,
+      "epoch": 0.02536091526561814,
+      "grad_norm": 0.5867946147918701,
+      "learning_rate": 9.998313054833123e-06,
+      "loss": 2.3628,
+      "mean_token_accuracy": 0.48890325892716646,
+      "num_tokens": 213525323.0,
+      "step": 1473
+    },
+    {
+      "entropy": 2.5234375,
+      "epoch": 0.0253781324518134,
+      "grad_norm": 0.5271015763282776,
+      "learning_rate": 9.998305899592378e-06,
+      "loss": 2.4971,
+      "mean_token_accuracy": 0.4719184576533735,
+      "num_tokens": 213676017.0,
+      "step": 1474
+    },
+    {
+      "entropy": 2.4503173828125,
+      "epoch": 0.02539534963800866,
+      "grad_norm": 0.5921691060066223,
+      "learning_rate": 9.99829872921169e-06,
+      "loss": 2.4328,
+      "mean_token_accuracy": 0.4846241660416126,
+      "num_tokens": 213819688.0,
+      "step": 1475
+    },
+    {
+      "entropy": 2.4017333984375,
+      "epoch": 0.02541256682420392,
+      "grad_norm": 0.5926265716552734,
+      "learning_rate": 9.998291543691081e-06,
+      "loss": 2.3141,
+      "mean_token_accuracy": 0.4890713025815785,
+      "num_tokens": 213966213.0,
+      "step": 1476
+    },
+    {
+      "entropy": 2.418212890625,
+      "epoch": 0.025429784010399182,
+      "grad_norm": 0.5591694116592407,
+      "learning_rate": 9.998284343030575e-06,
+      "loss": 2.3925,
+      "mean_token_accuracy": 0.4819628051482141,
+      "num_tokens": 214097650.0,
+      "step": 1477
+    },
+    {
+      "entropy": 2.39453125,
+      "epoch": 0.02544700119659444,
+      "grad_norm": 0.5935323238372803,
+      "learning_rate": 9.99827712723019e-06,
+      "loss": 2.3999,
+      "mean_token_accuracy": 0.48563890950754285,
+      "num_tokens": 214233345.0,
+      "step": 1478
+    },
+    {
+      "entropy": 2.5748291015625,
+      "epoch": 0.0254642183827897,
+      "grad_norm": 0.5952027440071106,
+      "learning_rate": 9.998269896289953e-06,
+      "loss": 2.531,
+      "mean_token_accuracy": 0.46888946322724223,
+      "num_tokens": 214371139.0,
+      "step": 1479
+    },
+    {
+      "entropy": 2.436767578125,
+      "epoch": 0.02548143556898496,
+      "grad_norm": 0.5706344842910767,
+      "learning_rate": 9.998262650209882e-06,
+      "loss": 2.4313,
+      "mean_token_accuracy": 0.4844216462224722,
+      "num_tokens": 214524344.0,
+      "step": 1480
+    },
+    {
+      "entropy": 2.444580078125,
+      "epoch": 0.02549865275518022,
+      "grad_norm": 0.6236752271652222,
+      "learning_rate": 9.99825538899e-06,
+      "loss": 2.4011,
+      "mean_token_accuracy": 0.48665000684559345,
+      "num_tokens": 214685683.0,
+      "step": 1481
+    },
+    {
+      "entropy": 2.4854736328125,
+      "epoch": 0.02551586994137548,
+      "grad_norm": 0.56282639503479,
+      "learning_rate": 9.998248112630328e-06,
+      "loss": 2.414,
+      "mean_token_accuracy": 0.47536311158910394,
+      "num_tokens": 214833325.0,
+      "step": 1482
+    },
+    {
+      "entropy": 2.44873046875,
+      "epoch": 0.02553308712757074,
+      "grad_norm": 0.5949015617370605,
+      "learning_rate": 9.99824082113089e-06,
+      "loss": 2.4348,
+      "mean_token_accuracy": 0.4832215076312423,
+      "num_tokens": 214974294.0,
+      "step": 1483
+    },
+    {
+      "entropy": 2.509765625,
+      "epoch": 0.025550304313766002,
+      "grad_norm": 0.6240081191062927,
+      "learning_rate": 9.998233514491706e-06,
+      "loss": 2.4456,
+      "mean_token_accuracy": 0.4775039851665497,
+      "num_tokens": 215107075.0,
+      "step": 1484
+    },
+    {
+      "entropy": 2.48486328125,
+      "epoch": 0.025567521499961263,
+      "grad_norm": 0.5603934526443481,
+      "learning_rate": 9.9982261927128e-06,
+      "loss": 2.4404,
+      "mean_token_accuracy": 0.4763817982748151,
+      "num_tokens": 215257991.0,
+      "step": 1485
+    },
+    {
+      "entropy": 2.472412109375,
+      "epoch": 0.02558473868615652,
+      "grad_norm": 0.5318784117698669,
+      "learning_rate": 9.998218855794193e-06,
+      "loss": 2.4512,
+      "mean_token_accuracy": 0.4735376350581646,
+      "num_tokens": 215424684.0,
+      "step": 1486
+    },
+    {
+      "entropy": 2.4873046875,
+      "epoch": 0.02560195587235178,
+      "grad_norm": 0.5659030079841614,
+      "learning_rate": 9.998211503735908e-06,
+      "loss": 2.4298,
+      "mean_token_accuracy": 0.47843836853280663,
+      "num_tokens": 215573454.0,
+      "step": 1487
+    },
+    {
+      "entropy": 2.3856201171875,
+      "epoch": 0.02561917305854704,
+      "grad_norm": 0.6163880228996277,
+      "learning_rate": 9.998204136537965e-06,
+      "loss": 2.3667,
+      "mean_token_accuracy": 0.49307617312297225,
+      "num_tokens": 215709724.0,
+      "step": 1488
+    },
+    {
+      "entropy": 2.4710693359375,
+      "epoch": 0.0256363902447423,
+      "grad_norm": 0.5732681751251221,
+      "learning_rate": 9.99819675420039e-06,
+      "loss": 2.4201,
+      "mean_token_accuracy": 0.4794590980745852,
+      "num_tokens": 215844176.0,
+      "step": 1489
+    },
+    {
+      "entropy": 2.4678955078125,
+      "epoch": 0.025653607430937562,
+      "grad_norm": 0.8923822045326233,
+      "learning_rate": 9.998189356723203e-06,
+      "loss": 2.4845,
+      "mean_token_accuracy": 0.47123232623562217,
+      "num_tokens": 215988104.0,
+      "step": 1490
+    },
+    {
+      "entropy": 2.5465087890625,
+      "epoch": 0.025670824617132822,
+      "grad_norm": 0.5661553740501404,
+      "learning_rate": 9.998181944106428e-06,
+      "loss": 2.5037,
+      "mean_token_accuracy": 0.4655986800789833,
+      "num_tokens": 216131658.0,
+      "step": 1491
+    },
+    {
+      "entropy": 2.486572265625,
+      "epoch": 0.025688041803328083,
+      "grad_norm": 0.5814360976219177,
+      "learning_rate": 9.998174516350086e-06,
+      "loss": 2.4732,
+      "mean_token_accuracy": 0.4705640790052712,
+      "num_tokens": 216274055.0,
+      "step": 1492
+    },
+    {
+      "entropy": 2.3916015625,
+      "epoch": 0.025705258989523343,
+      "grad_norm": 0.5932287573814392,
+      "learning_rate": 9.998167073454197e-06,
+      "loss": 2.3721,
+      "mean_token_accuracy": 0.49608813878148794,
+      "num_tokens": 216414995.0,
+      "step": 1493
+    },
+    {
+      "entropy": 2.4781494140625,
+      "epoch": 0.025722476175718604,
+      "grad_norm": 0.5934662222862244,
+      "learning_rate": 9.99815961541879e-06,
+      "loss": 2.4183,
+      "mean_token_accuracy": 0.4743193374015391,
+      "num_tokens": 216559579.0,
+      "step": 1494
+    },
+    {
+      "entropy": 2.4559326171875,
+      "epoch": 0.02573969336191386,
+      "grad_norm": 0.5569154620170593,
+      "learning_rate": 9.998152142243882e-06,
+      "loss": 2.3922,
+      "mean_token_accuracy": 0.48210062831640244,
+      "num_tokens": 216706806.0,
+      "step": 1495
+    },
+    {
+      "entropy": 2.4129638671875,
+      "epoch": 0.02575691054810912,
+      "grad_norm": 0.5600568056106567,
+      "learning_rate": 9.998144653929499e-06,
+      "loss": 2.371,
+      "mean_token_accuracy": 0.4889267375692725,
+      "num_tokens": 216861317.0,
+      "step": 1496
+    },
+    {
+      "entropy": 2.4239501953125,
+      "epoch": 0.025774127734304382,
+      "grad_norm": 0.5837308168411255,
+      "learning_rate": 9.99813715047566e-06,
+      "loss": 2.3703,
+      "mean_token_accuracy": 0.48130458453670144,
+      "num_tokens": 217013439.0,
+      "step": 1497
+    },
+    {
+      "entropy": 2.4178466796875,
+      "epoch": 0.025791344920499643,
+      "grad_norm": 0.5723254084587097,
+      "learning_rate": 9.998129631882392e-06,
+      "loss": 2.3257,
+      "mean_token_accuracy": 0.4927904959768057,
+      "num_tokens": 217161046.0,
+      "step": 1498
+    },
+    {
+      "entropy": 2.4444580078125,
+      "epoch": 0.025808562106694903,
+      "grad_norm": 0.6308748126029968,
+      "learning_rate": 9.998122098149714e-06,
+      "loss": 2.3964,
+      "mean_token_accuracy": 0.4866189956665039,
+      "num_tokens": 217301151.0,
+      "step": 1499
+    },
+    {
+      "entropy": 2.422607421875,
+      "epoch": 0.025825779292890164,
+      "grad_norm": 0.5735398530960083,
+      "learning_rate": 9.998114549277653e-06,
+      "loss": 2.4083,
+      "mean_token_accuracy": 0.4805009118281305,
+      "num_tokens": 217444559.0,
+      "step": 1500
+    },
+    {
+      "entropy": 2.4912109375,
+      "epoch": 0.025842996479085424,
+      "grad_norm": 0.5808705687522888,
+      "learning_rate": 9.998106985266229e-06,
+      "loss": 2.4692,
+      "mean_token_accuracy": 0.47182623157277703,
+      "num_tokens": 217596025.0,
+      "step": 1501
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.025860213665280685,
+      "grad_norm": 0.5593245625495911,
+      "learning_rate": 9.998099406115465e-06,
+      "loss": 2.4285,
+      "mean_token_accuracy": 0.4782219841144979,
+      "num_tokens": 217747354.0,
+      "step": 1502
+    },
+    {
+      "entropy": 2.429931640625,
+      "epoch": 0.02587743085147594,
+      "grad_norm": 0.5433899760246277,
+      "learning_rate": 9.998091811825383e-06,
+      "loss": 2.4056,
+      "mean_token_accuracy": 0.4807021920569241,
+      "num_tokens": 217895004.0,
+      "step": 1503
+    },
+    {
+      "entropy": 2.4007568359375,
+      "epoch": 0.025894648037671202,
+      "grad_norm": 0.5361343026161194,
+      "learning_rate": 9.998084202396008e-06,
+      "loss": 2.3451,
+      "mean_token_accuracy": 0.4870217302814126,
+      "num_tokens": 218041797.0,
+      "step": 1504
+    },
+    {
+      "entropy": 2.5289306640625,
+      "epoch": 0.025911865223866463,
+      "grad_norm": 0.5812287330627441,
+      "learning_rate": 9.998076577827363e-06,
+      "loss": 2.466,
+      "mean_token_accuracy": 0.4721836093813181,
+      "num_tokens": 218184823.0,
+      "step": 1505
+    },
+    {
+      "entropy": 2.4560546875,
+      "epoch": 0.025929082410061723,
+      "grad_norm": 0.667839527130127,
+      "learning_rate": 9.998068938119471e-06,
+      "loss": 2.4359,
+      "mean_token_accuracy": 0.4753602985292673,
+      "num_tokens": 218322885.0,
+      "step": 1506
+    },
+    {
+      "entropy": 2.56201171875,
+      "epoch": 0.025946299596256984,
+      "grad_norm": 0.5627098083496094,
+      "learning_rate": 9.998061283272353e-06,
+      "loss": 2.4754,
+      "mean_token_accuracy": 0.4772064401768148,
+      "num_tokens": 218485130.0,
+      "step": 1507
+    },
+    {
+      "entropy": 2.4844970703125,
+      "epoch": 0.025963516782452244,
+      "grad_norm": 0.5547756552696228,
+      "learning_rate": 9.998053613286036e-06,
+      "loss": 2.4126,
+      "mean_token_accuracy": 0.4796820618212223,
+      "num_tokens": 218631229.0,
+      "step": 1508
+    },
+    {
+      "entropy": 2.4619140625,
+      "epoch": 0.025980733968647505,
+      "grad_norm": 0.5786131024360657,
+      "learning_rate": 9.99804592816054e-06,
+      "loss": 2.435,
+      "mean_token_accuracy": 0.48094206769019365,
+      "num_tokens": 218784443.0,
+      "step": 1509
+    },
+    {
+      "entropy": 2.4591064453125,
+      "epoch": 0.025997951154842765,
+      "grad_norm": 0.5810285806655884,
+      "learning_rate": 9.99803822789589e-06,
+      "loss": 2.419,
+      "mean_token_accuracy": 0.47504253312945366,
+      "num_tokens": 218923714.0,
+      "step": 1510
+    },
+    {
+      "entropy": 2.465087890625,
+      "epoch": 0.026015168341038022,
+      "grad_norm": 0.570573091506958,
+      "learning_rate": 9.998030512492108e-06,
+      "loss": 2.4312,
+      "mean_token_accuracy": 0.4744260283187032,
+      "num_tokens": 219076633.0,
+      "step": 1511
+    },
+    {
+      "entropy": 2.4339599609375,
+      "epoch": 0.026032385527233283,
+      "grad_norm": 0.5941885709762573,
+      "learning_rate": 9.998022781949217e-06,
+      "loss": 2.4043,
+      "mean_token_accuracy": 0.48197115818038583,
+      "num_tokens": 219219462.0,
+      "step": 1512
+    },
+    {
+      "entropy": 2.4404296875,
+      "epoch": 0.026049602713428543,
+      "grad_norm": 0.6012705564498901,
+      "learning_rate": 9.998015036267243e-06,
+      "loss": 2.4462,
+      "mean_token_accuracy": 0.4794664392247796,
+      "num_tokens": 219366857.0,
+      "step": 1513
+    },
+    {
+      "entropy": 2.44775390625,
+      "epoch": 0.026066819899623804,
+      "grad_norm": 0.6487138271331787,
+      "learning_rate": 9.998007275446206e-06,
+      "loss": 2.4562,
+      "mean_token_accuracy": 0.48240925278514624,
+      "num_tokens": 219501713.0,
+      "step": 1514
+    },
+    {
+      "entropy": 2.4669189453125,
+      "epoch": 0.026084037085819065,
+      "grad_norm": 0.5475379228591919,
+      "learning_rate": 9.997999499486134e-06,
+      "loss": 2.4614,
+      "mean_token_accuracy": 0.4733142964541912,
+      "num_tokens": 219651285.0,
+      "step": 1515
+    },
+    {
+      "entropy": 2.4892578125,
+      "epoch": 0.026101254272014325,
+      "grad_norm": 0.5989498496055603,
+      "learning_rate": 9.997991708387047e-06,
+      "loss": 2.4537,
+      "mean_token_accuracy": 0.4730882477015257,
+      "num_tokens": 219781732.0,
+      "step": 1516
+    },
+    {
+      "entropy": 2.48681640625,
+      "epoch": 0.026118471458209586,
+      "grad_norm": 0.6295292973518372,
+      "learning_rate": 9.99798390214897e-06,
+      "loss": 2.5009,
+      "mean_token_accuracy": 0.4708400582894683,
+      "num_tokens": 219923239.0,
+      "step": 1517
+    },
+    {
+      "entropy": 2.525390625,
+      "epoch": 0.026135688644404846,
+      "grad_norm": 0.5331100225448608,
+      "learning_rate": 9.997976080771924e-06,
+      "loss": 2.5217,
+      "mean_token_accuracy": 0.4696351978927851,
+      "num_tokens": 220085911.0,
+      "step": 1518
+    },
+    {
+      "entropy": 2.3858642578125,
+      "epoch": 0.026152905830600107,
+      "grad_norm": 0.5755723118782043,
+      "learning_rate": 9.997968244255937e-06,
+      "loss": 2.3168,
+      "mean_token_accuracy": 0.4918542858213186,
+      "num_tokens": 220225909.0,
+      "step": 1519
+    },
+    {
+      "entropy": 2.41943359375,
+      "epoch": 0.026170123016795364,
+      "grad_norm": 0.583797812461853,
+      "learning_rate": 9.99796039260103e-06,
+      "loss": 2.4,
+      "mean_token_accuracy": 0.4865543758496642,
+      "num_tokens": 220360506.0,
+      "step": 1520
+    },
+    {
+      "entropy": 2.4930419921875,
+      "epoch": 0.026187340202990624,
+      "grad_norm": 0.5480349659919739,
+      "learning_rate": 9.997952525807229e-06,
+      "loss": 2.4718,
+      "mean_token_accuracy": 0.4783789166249335,
+      "num_tokens": 220518785.0,
+      "step": 1521
+    },
+    {
+      "entropy": 2.446533203125,
+      "epoch": 0.026204557389185885,
+      "grad_norm": 0.565557062625885,
+      "learning_rate": 9.997944643874553e-06,
+      "loss": 2.4001,
+      "mean_token_accuracy": 0.48282787948846817,
+      "num_tokens": 220669933.0,
+      "step": 1522
+    },
+    {
+      "entropy": 2.398681640625,
+      "epoch": 0.026221774575381145,
+      "grad_norm": 0.5465447902679443,
+      "learning_rate": 9.997936746803032e-06,
+      "loss": 2.3459,
+      "mean_token_accuracy": 0.4924741433933377,
+      "num_tokens": 220817234.0,
+      "step": 1523
+    },
+    {
+      "entropy": 2.4630126953125,
+      "epoch": 0.026238991761576406,
+      "grad_norm": 0.6321006417274475,
+      "learning_rate": 9.997928834592686e-06,
+      "loss": 2.4352,
+      "mean_token_accuracy": 0.47752982610836625,
+      "num_tokens": 220954570.0,
+      "step": 1524
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.026256208947771666,
+      "grad_norm": 0.56505286693573,
+      "learning_rate": 9.99792090724354e-06,
+      "loss": 2.4511,
+      "mean_token_accuracy": 0.48103819927200675,
+      "num_tokens": 221097668.0,
+      "step": 1525
+    },
+    {
+      "entropy": 2.50634765625,
+      "epoch": 0.026273426133966927,
+      "grad_norm": 0.5791621208190918,
+      "learning_rate": 9.997912964755618e-06,
+      "loss": 2.4866,
+      "mean_token_accuracy": 0.46793483989313245,
+      "num_tokens": 221236494.0,
+      "step": 1526
+    },
+    {
+      "entropy": 2.4517822265625,
+      "epoch": 0.026290643320162187,
+      "grad_norm": 0.6186292171478271,
+      "learning_rate": 9.997905007128946e-06,
+      "loss": 2.4148,
+      "mean_token_accuracy": 0.47660065814852715,
+      "num_tokens": 221371798.0,
+      "step": 1527
+    },
+    {
+      "entropy": 2.457763671875,
+      "epoch": 0.026307860506357444,
+      "grad_norm": 0.5822423100471497,
+      "learning_rate": 9.997897034363544e-06,
+      "loss": 2.4246,
+      "mean_token_accuracy": 0.4802300548180938,
+      "num_tokens": 221514663.0,
+      "step": 1528
+    },
+    {
+      "entropy": 2.4844970703125,
+      "epoch": 0.026325077692552705,
+      "grad_norm": 0.5579072833061218,
+      "learning_rate": 9.997889046459438e-06,
+      "loss": 2.4567,
+      "mean_token_accuracy": 0.47267806623131037,
+      "num_tokens": 221654782.0,
+      "step": 1529
+    },
+    {
+      "entropy": 2.482666015625,
+      "epoch": 0.026342294878747965,
+      "grad_norm": 0.5542824864387512,
+      "learning_rate": 9.997881043416653e-06,
+      "loss": 2.4276,
+      "mean_token_accuracy": 0.47997074108570814,
+      "num_tokens": 221806130.0,
+      "step": 1530
+    },
+    {
+      "entropy": 2.4971923828125,
+      "epoch": 0.026359512064943226,
+      "grad_norm": 0.5979059934616089,
+      "learning_rate": 9.997873025235215e-06,
+      "loss": 2.4411,
+      "mean_token_accuracy": 0.4823180711828172,
+      "num_tokens": 221951263.0,
+      "step": 1531
+    },
+    {
+      "entropy": 2.483154296875,
+      "epoch": 0.026376729251138487,
+      "grad_norm": 1.1579712629318237,
+      "learning_rate": 9.997864991915142e-06,
+      "loss": 2.4429,
+      "mean_token_accuracy": 0.4803687371313572,
+      "num_tokens": 222106221.0,
+      "step": 1532
+    },
+    {
+      "entropy": 2.437255859375,
+      "epoch": 0.026393946437333747,
+      "grad_norm": 0.585996687412262,
+      "learning_rate": 9.997856943456465e-06,
+      "loss": 2.4413,
+      "mean_token_accuracy": 0.4799509192816913,
+      "num_tokens": 222278140.0,
+      "step": 1533
+    },
+    {
+      "entropy": 2.478759765625,
+      "epoch": 0.026411163623529008,
+      "grad_norm": 0.5700007081031799,
+      "learning_rate": 9.997848879859205e-06,
+      "loss": 2.3974,
+      "mean_token_accuracy": 0.48377181869000196,
+      "num_tokens": 222420219.0,
+      "step": 1534
+    },
+    {
+      "entropy": 2.4041748046875,
+      "epoch": 0.026428380809724268,
+      "grad_norm": 0.5667886137962341,
+      "learning_rate": 9.99784080112339e-06,
+      "loss": 2.387,
+      "mean_token_accuracy": 0.48956062365323305,
+      "num_tokens": 222562430.0,
+      "step": 1535
+    },
+    {
+      "entropy": 2.5068359375,
+      "epoch": 0.026445597995919525,
+      "grad_norm": 0.5823205709457397,
+      "learning_rate": 9.997832707249038e-06,
+      "loss": 2.4595,
+      "mean_token_accuracy": 0.47320169396698475,
+      "num_tokens": 222700414.0,
+      "step": 1536
+    },
+    {
+      "entropy": 2.4405517578125,
+      "epoch": 0.026462815182114786,
+      "grad_norm": 0.546593427658081,
+      "learning_rate": 9.997824598236179e-06,
+      "loss": 2.3965,
+      "mean_token_accuracy": 0.48351220693439245,
+      "num_tokens": 222856133.0,
+      "step": 1537
+    },
+    {
+      "entropy": 2.4632568359375,
+      "epoch": 0.026480032368310046,
+      "grad_norm": 0.5839020013809204,
+      "learning_rate": 9.997816474084833e-06,
+      "loss": 2.4677,
+      "mean_token_accuracy": 0.4769261754117906,
+      "num_tokens": 223003734.0,
+      "step": 1538
+    },
+    {
+      "entropy": 2.3848876953125,
+      "epoch": 0.026497249554505307,
+      "grad_norm": 0.5713815093040466,
+      "learning_rate": 9.997808334795032e-06,
+      "loss": 2.3267,
+      "mean_token_accuracy": 0.4928825213573873,
+      "num_tokens": 223150812.0,
+      "step": 1539
+    },
+    {
+      "entropy": 2.4129638671875,
+      "epoch": 0.026514466740700567,
+      "grad_norm": 0.5621561408042908,
+      "learning_rate": 9.997800180366792e-06,
+      "loss": 2.369,
+      "mean_token_accuracy": 0.486449159681797,
+      "num_tokens": 223302261.0,
+      "step": 1540
+    },
+    {
+      "entropy": 2.4468994140625,
+      "epoch": 0.026531683926895828,
+      "grad_norm": 0.5728362202644348,
+      "learning_rate": 9.997792010800144e-06,
+      "loss": 2.3988,
+      "mean_token_accuracy": 0.4837730205617845,
+      "num_tokens": 223448432.0,
+      "step": 1541
+    },
+    {
+      "entropy": 2.4149169921875,
+      "epoch": 0.02654890111309109,
+      "grad_norm": 0.6072005033493042,
+      "learning_rate": 9.99778382609511e-06,
+      "loss": 2.3772,
+      "mean_token_accuracy": 0.48883628472685814,
+      "num_tokens": 223586866.0,
+      "step": 1542
+    },
+    {
+      "entropy": 2.4757080078125,
+      "epoch": 0.02656611829928635,
+      "grad_norm": 0.5731198191642761,
+      "learning_rate": 9.997775626251715e-06,
+      "loss": 2.4887,
+      "mean_token_accuracy": 0.4729891321621835,
+      "num_tokens": 223739874.0,
+      "step": 1543
+    },
+    {
+      "entropy": 2.4337158203125,
+      "epoch": 0.02658333548548161,
+      "grad_norm": 0.5194607377052307,
+      "learning_rate": 9.997767411269984e-06,
+      "loss": 2.4039,
+      "mean_token_accuracy": 0.47655164077878,
+      "num_tokens": 223889106.0,
+      "step": 1544
+    },
+    {
+      "entropy": 2.372314453125,
+      "epoch": 0.026600552671676866,
+      "grad_norm": 0.5741761922836304,
+      "learning_rate": 9.997759181149941e-06,
+      "loss": 2.3325,
+      "mean_token_accuracy": 0.49444987904280424,
+      "num_tokens": 224029055.0,
+      "step": 1545
+    },
+    {
+      "entropy": 2.479736328125,
+      "epoch": 0.026617769857872127,
+      "grad_norm": 0.5601452589035034,
+      "learning_rate": 9.997750935891615e-06,
+      "loss": 2.4167,
+      "mean_token_accuracy": 0.47284849267452955,
+      "num_tokens": 224166339.0,
+      "step": 1546
+    },
+    {
+      "entropy": 2.44091796875,
+      "epoch": 0.026634987044067387,
+      "grad_norm": 0.6984487771987915,
+      "learning_rate": 9.997742675495025e-06,
+      "loss": 2.3943,
+      "mean_token_accuracy": 0.4849527506157756,
+      "num_tokens": 224314144.0,
+      "step": 1547
+    },
+    {
+      "entropy": 2.42138671875,
+      "epoch": 0.026652204230262648,
+      "grad_norm": 0.6234527230262756,
+      "learning_rate": 9.9977343999602e-06,
+      "loss": 2.3438,
+      "mean_token_accuracy": 0.49563004495576024,
+      "num_tokens": 224465327.0,
+      "step": 1548
+    },
+    {
+      "entropy": 2.462158203125,
+      "epoch": 0.02666942141645791,
+      "grad_norm": 0.5658462047576904,
+      "learning_rate": 9.997726109287164e-06,
+      "loss": 2.4079,
+      "mean_token_accuracy": 0.48139845160767436,
+      "num_tokens": 224608026.0,
+      "step": 1549
+    },
+    {
+      "entropy": 2.40087890625,
+      "epoch": 0.02668663860265317,
+      "grad_norm": 0.57075035572052,
+      "learning_rate": 9.997717803475942e-06,
+      "loss": 2.3665,
+      "mean_token_accuracy": 0.4883747296407819,
+      "num_tokens": 224758446.0,
+      "step": 1550
+    },
+    {
+      "entropy": 2.45947265625,
+      "epoch": 0.02670385578884843,
+      "grad_norm": 0.5546119809150696,
+      "learning_rate": 9.99770948252656e-06,
+      "loss": 2.4301,
+      "mean_token_accuracy": 0.47884270129725337,
+      "num_tokens": 224900091.0,
+      "step": 1551
+    },
+    {
+      "entropy": 2.4818115234375,
+      "epoch": 0.02672107297504369,
+      "grad_norm": 0.584319531917572,
+      "learning_rate": 9.99770114643904e-06,
+      "loss": 2.444,
+      "mean_token_accuracy": 0.4730917243286967,
+      "num_tokens": 225040740.0,
+      "step": 1552
+    },
+    {
+      "entropy": 2.4044189453125,
+      "epoch": 0.026738290161238947,
+      "grad_norm": 0.6092365980148315,
+      "learning_rate": 9.997692795213412e-06,
+      "loss": 2.3605,
+      "mean_token_accuracy": 0.4851328986696899,
+      "num_tokens": 225186755.0,
+      "step": 1553
+    },
+    {
+      "entropy": 2.4434814453125,
+      "epoch": 0.026755507347434208,
+      "grad_norm": 0.6220741868019104,
+      "learning_rate": 9.997684428849698e-06,
+      "loss": 2.4281,
+      "mean_token_accuracy": 0.4839205415919423,
+      "num_tokens": 225334853.0,
+      "step": 1554
+    },
+    {
+      "entropy": 2.4833984375,
+      "epoch": 0.026772724533629468,
+      "grad_norm": 0.6230896711349487,
+      "learning_rate": 9.997676047347926e-06,
+      "loss": 2.4525,
+      "mean_token_accuracy": 0.4806727943941951,
+      "num_tokens": 225463554.0,
+      "step": 1555
+    },
+    {
+      "entropy": 2.4664306640625,
+      "epoch": 0.02678994171982473,
+      "grad_norm": 0.5725734233856201,
+      "learning_rate": 9.997667650708117e-06,
+      "loss": 2.4501,
+      "mean_token_accuracy": 0.4741324451752007,
+      "num_tokens": 225606325.0,
+      "step": 1556
+    },
+    {
+      "entropy": 2.511962890625,
+      "epoch": 0.02680715890601999,
+      "grad_norm": 0.5844495296478271,
+      "learning_rate": 9.9976592389303e-06,
+      "loss": 2.4653,
+      "mean_token_accuracy": 0.47366986563429236,
+      "num_tokens": 225755308.0,
+      "step": 1557
+    },
+    {
+      "entropy": 2.454345703125,
+      "epoch": 0.02682437609221525,
+      "grad_norm": 0.5523918867111206,
+      "learning_rate": 9.9976508120145e-06,
+      "loss": 2.4185,
+      "mean_token_accuracy": 0.4829486352391541,
+      "num_tokens": 225916539.0,
+      "step": 1558
+    },
+    {
+      "entropy": 2.47607421875,
+      "epoch": 0.02684159327841051,
+      "grad_norm": 0.5556900501251221,
+      "learning_rate": 9.997642369960743e-06,
+      "loss": 2.4678,
+      "mean_token_accuracy": 0.47704968182370067,
+      "num_tokens": 226065388.0,
+      "step": 1559
+    },
+    {
+      "entropy": 2.4385986328125,
+      "epoch": 0.02685881046460577,
+      "grad_norm": 0.5409045219421387,
+      "learning_rate": 9.997633912769054e-06,
+      "loss": 2.3615,
+      "mean_token_accuracy": 0.4902169401757419,
+      "num_tokens": 226221599.0,
+      "step": 1560
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.026876027650801028,
+      "grad_norm": 0.612868070602417,
+      "learning_rate": 9.997625440439457e-06,
+      "loss": 2.4594,
+      "mean_token_accuracy": 0.47914552967995405,
+      "num_tokens": 226362147.0,
+      "step": 1561
+    },
+    {
+      "entropy": 2.46142578125,
+      "epoch": 0.02689324483699629,
+      "grad_norm": 0.5608747601509094,
+      "learning_rate": 9.997616952971979e-06,
+      "loss": 2.4424,
+      "mean_token_accuracy": 0.4790024384856224,
+      "num_tokens": 226510293.0,
+      "step": 1562
+    },
+    {
+      "entropy": 2.47900390625,
+      "epoch": 0.02691046202319155,
+      "grad_norm": 0.5777642726898193,
+      "learning_rate": 9.997608450366646e-06,
+      "loss": 2.4461,
+      "mean_token_accuracy": 0.4724403969012201,
+      "num_tokens": 226654656.0,
+      "step": 1563
+    },
+    {
+      "entropy": 2.42529296875,
+      "epoch": 0.02692767920938681,
+      "grad_norm": 0.5585452914237976,
+      "learning_rate": 9.997599932623485e-06,
+      "loss": 2.3716,
+      "mean_token_accuracy": 0.4799959072843194,
+      "num_tokens": 226810305.0,
+      "step": 1564
+    },
+    {
+      "entropy": 2.391357421875,
+      "epoch": 0.02694489639558207,
+      "grad_norm": 0.6446172595024109,
+      "learning_rate": 9.997591399742518e-06,
+      "loss": 2.3486,
+      "mean_token_accuracy": 0.4916059426032007,
+      "num_tokens": 226969167.0,
+      "step": 1565
+    },
+    {
+      "entropy": 2.4498291015625,
+      "epoch": 0.02696211358177733,
+      "grad_norm": 0.5548785924911499,
+      "learning_rate": 9.997582851723773e-06,
+      "loss": 2.4116,
+      "mean_token_accuracy": 0.4788562678731978,
+      "num_tokens": 227115813.0,
+      "step": 1566
+    },
+    {
+      "entropy": 2.5760498046875,
+      "epoch": 0.02697933076797259,
+      "grad_norm": 0.5751841068267822,
+      "learning_rate": 9.997574288567277e-06,
+      "loss": 2.5083,
+      "mean_token_accuracy": 0.4620812046341598,
+      "num_tokens": 227265495.0,
+      "step": 1567
+    },
+    {
+      "entropy": 2.470947265625,
+      "epoch": 0.02699654795416785,
+      "grad_norm": 1.4221575260162354,
+      "learning_rate": 9.997565710273056e-06,
+      "loss": 2.4116,
+      "mean_token_accuracy": 0.4814313040114939,
+      "num_tokens": 227406243.0,
+      "step": 1568
+    },
+    {
+      "entropy": 2.461669921875,
+      "epoch": 0.027013765140363112,
+      "grad_norm": 0.5775115489959717,
+      "learning_rate": 9.997557116841134e-06,
+      "loss": 2.4235,
+      "mean_token_accuracy": 0.4807606884278357,
+      "num_tokens": 227566411.0,
+      "step": 1569
+    },
+    {
+      "entropy": 2.447509765625,
+      "epoch": 0.02703098232655837,
+      "grad_norm": 0.5913234353065491,
+      "learning_rate": 9.997548508271537e-06,
+      "loss": 2.3974,
+      "mean_token_accuracy": 0.48274747421965003,
+      "num_tokens": 227695596.0,
+      "step": 1570
+    },
+    {
+      "entropy": 2.4384765625,
+      "epoch": 0.02704819951275363,
+      "grad_norm": 0.5852504968643188,
+      "learning_rate": 9.997539884564293e-06,
+      "loss": 2.3895,
+      "mean_token_accuracy": 0.4838076075538993,
+      "num_tokens": 227834644.0,
+      "step": 1571
+    },
+    {
+      "entropy": 2.4619140625,
+      "epoch": 0.02706541669894889,
+      "grad_norm": 0.5741419196128845,
+      "learning_rate": 9.997531245719427e-06,
+      "loss": 2.4267,
+      "mean_token_accuracy": 0.4829978961497545,
+      "num_tokens": 227981607.0,
+      "step": 1572
+    },
+    {
+      "entropy": 2.4593505859375,
+      "epoch": 0.02708263388514415,
+      "grad_norm": 0.5879804491996765,
+      "learning_rate": 9.997522591736965e-06,
+      "loss": 2.4387,
+      "mean_token_accuracy": 0.47719143191352487,
+      "num_tokens": 228129360.0,
+      "step": 1573
+    },
+    {
+      "entropy": 2.425048828125,
+      "epoch": 0.02709985107133941,
+      "grad_norm": 0.6101236343383789,
+      "learning_rate": 9.997513922616935e-06,
+      "loss": 2.3956,
+      "mean_token_accuracy": 0.4810835770331323,
+      "num_tokens": 228271608.0,
+      "step": 1574
+    },
+    {
+      "entropy": 2.447021484375,
+      "epoch": 0.027117068257534672,
+      "grad_norm": 0.5525056719779968,
+      "learning_rate": 9.997505238359362e-06,
+      "loss": 2.4221,
+      "mean_token_accuracy": 0.47596909245476127,
+      "num_tokens": 228425263.0,
+      "step": 1575
+    },
+    {
+      "entropy": 2.4608154296875,
+      "epoch": 0.027134285443729932,
+      "grad_norm": 0.5748025178909302,
+      "learning_rate": 9.997496538964269e-06,
+      "loss": 2.4484,
+      "mean_token_accuracy": 0.4767248681746423,
+      "num_tokens": 228577817.0,
+      "step": 1576
+    },
+    {
+      "entropy": 2.52978515625,
+      "epoch": 0.027151502629925193,
+      "grad_norm": 0.556822657585144,
+      "learning_rate": 9.997487824431687e-06,
+      "loss": 2.4771,
+      "mean_token_accuracy": 0.4696510974317789,
+      "num_tokens": 228731761.0,
+      "step": 1577
+    },
+    {
+      "entropy": 2.517578125,
+      "epoch": 0.02716871981612045,
+      "grad_norm": 0.6445668339729309,
+      "learning_rate": 9.997479094761641e-06,
+      "loss": 2.4696,
+      "mean_token_accuracy": 0.47059842152521014,
+      "num_tokens": 228882447.0,
+      "step": 1578
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.02718593700231571,
+      "grad_norm": 0.5520902276039124,
+      "learning_rate": 9.997470349954158e-06,
+      "loss": 2.4349,
+      "mean_token_accuracy": 0.4763860795646906,
+      "num_tokens": 229032651.0,
+      "step": 1579
+    },
+    {
+      "entropy": 2.477294921875,
+      "epoch": 0.02720315418851097,
+      "grad_norm": 0.6009482145309448,
+      "learning_rate": 9.997461590009263e-06,
+      "loss": 2.4098,
+      "mean_token_accuracy": 0.47861299896612763,
+      "num_tokens": 229180126.0,
+      "step": 1580
+    },
+    {
+      "entropy": 2.380615234375,
+      "epoch": 0.02722037137470623,
+      "grad_norm": 0.5625278353691101,
+      "learning_rate": 9.997452814926984e-06,
+      "loss": 2.3496,
+      "mean_token_accuracy": 0.4916512775234878,
+      "num_tokens": 229322174.0,
+      "step": 1581
+    },
+    {
+      "entropy": 2.5185546875,
+      "epoch": 0.027237588560901492,
+      "grad_norm": 0.6114506721496582,
+      "learning_rate": 9.997444024707345e-06,
+      "loss": 2.4779,
+      "mean_token_accuracy": 0.4617612757720053,
+      "num_tokens": 229454480.0,
+      "step": 1582
+    },
+    {
+      "entropy": 2.508544921875,
+      "epoch": 0.027254805747096753,
+      "grad_norm": 0.6254023909568787,
+      "learning_rate": 9.997435219350377e-06,
+      "loss": 2.4818,
+      "mean_token_accuracy": 0.47184726875275373,
+      "num_tokens": 229588770.0,
+      "step": 1583
+    },
+    {
+      "entropy": 2.37744140625,
+      "epoch": 0.027272022933292013,
+      "grad_norm": 0.5333112478256226,
+      "learning_rate": 9.997426398856103e-06,
+      "loss": 2.3411,
+      "mean_token_accuracy": 0.4888959531672299,
+      "num_tokens": 229754120.0,
+      "step": 1584
+    },
+    {
+      "entropy": 2.42529296875,
+      "epoch": 0.027289240119487274,
+      "grad_norm": 0.5652498602867126,
+      "learning_rate": 9.997417563224551e-06,
+      "loss": 2.3861,
+      "mean_token_accuracy": 0.4848038759082556,
+      "num_tokens": 229894455.0,
+      "step": 1585
+    },
+    {
+      "entropy": 2.4168701171875,
+      "epoch": 0.02730645730568253,
+      "grad_norm": 0.5742363929748535,
+      "learning_rate": 9.997408712455748e-06,
+      "loss": 2.3838,
+      "mean_token_accuracy": 0.4861548813059926,
+      "num_tokens": 230044890.0,
+      "step": 1586
+    },
+    {
+      "entropy": 2.4842529296875,
+      "epoch": 0.02732367449187779,
+      "grad_norm": 0.5573915839195251,
+      "learning_rate": 9.99739984654972e-06,
+      "loss": 2.4365,
+      "mean_token_accuracy": 0.4722730196081102,
+      "num_tokens": 230180808.0,
+      "step": 1587
+    },
+    {
+      "entropy": 2.497802734375,
+      "epoch": 0.02734089167807305,
+      "grad_norm": 0.6639450192451477,
+      "learning_rate": 9.997390965506495e-06,
+      "loss": 2.4367,
+      "mean_token_accuracy": 0.47327509289607406,
+      "num_tokens": 230325006.0,
+      "step": 1588
+    },
+    {
+      "entropy": 2.514404296875,
+      "epoch": 0.027358108864268312,
+      "grad_norm": 0.5561137795448303,
+      "learning_rate": 9.997382069326099e-06,
+      "loss": 2.4852,
+      "mean_token_accuracy": 0.4662863416597247,
+      "num_tokens": 230483748.0,
+      "step": 1589
+    },
+    {
+      "entropy": 2.4310302734375,
+      "epoch": 0.027375326050463573,
+      "grad_norm": 0.5391974449157715,
+      "learning_rate": 9.997373158008558e-06,
+      "loss": 2.4237,
+      "mean_token_accuracy": 0.4804062177427113,
+      "num_tokens": 230645992.0,
+      "step": 1590
+    },
+    {
+      "entropy": 2.4122314453125,
+      "epoch": 0.027392543236658833,
+      "grad_norm": 0.577002763748169,
+      "learning_rate": 9.997364231553902e-06,
+      "loss": 2.3422,
+      "mean_token_accuracy": 0.4947265670634806,
+      "num_tokens": 230788584.0,
+      "step": 1591
+    },
+    {
+      "entropy": 2.4708251953125,
+      "epoch": 0.027409760422854094,
+      "grad_norm": 0.5783287286758423,
+      "learning_rate": 9.997355289962157e-06,
+      "loss": 2.4332,
+      "mean_token_accuracy": 0.47268676944077015,
+      "num_tokens": 230920875.0,
+      "step": 1592
+    },
+    {
+      "entropy": 2.495361328125,
+      "epoch": 0.027426977609049354,
+      "grad_norm": 0.5736862421035767,
+      "learning_rate": 9.997346333233347e-06,
+      "loss": 2.4839,
+      "mean_token_accuracy": 0.47372210351750255,
+      "num_tokens": 231066994.0,
+      "step": 1593
+    },
+    {
+      "entropy": 2.414306640625,
+      "epoch": 0.027444194795244615,
+      "grad_norm": 0.6175742149353027,
+      "learning_rate": 9.9973373613675e-06,
+      "loss": 2.3682,
+      "mean_token_accuracy": 0.481872939504683,
+      "num_tokens": 231204186.0,
+      "step": 1594
+    },
+    {
+      "entropy": 2.45068359375,
+      "epoch": 0.027461411981439872,
+      "grad_norm": 0.6151363253593445,
+      "learning_rate": 9.997328374364647e-06,
+      "loss": 2.4043,
+      "mean_token_accuracy": 0.4820513529703021,
+      "num_tokens": 231352849.0,
+      "step": 1595
+    },
+    {
+      "entropy": 2.406005859375,
+      "epoch": 0.027478629167635132,
+      "grad_norm": 0.591463565826416,
+      "learning_rate": 9.997319372224814e-06,
+      "loss": 2.3365,
+      "mean_token_accuracy": 0.4920554677955806,
+      "num_tokens": 231492437.0,
+      "step": 1596
+    },
+    {
+      "entropy": 2.4417724609375,
+      "epoch": 0.027495846353830393,
+      "grad_norm": 0.531035304069519,
+      "learning_rate": 9.997310354948026e-06,
+      "loss": 2.4092,
+      "mean_token_accuracy": 0.4780901288613677,
+      "num_tokens": 231644457.0,
+      "step": 1597
+    },
+    {
+      "entropy": 2.4251708984375,
+      "epoch": 0.027513063540025653,
+      "grad_norm": 0.5585000514984131,
+      "learning_rate": 9.99730132253431e-06,
+      "loss": 2.3906,
+      "mean_token_accuracy": 0.4829976833425462,
+      "num_tokens": 231800143.0,
+      "step": 1598
+    },
+    {
+      "entropy": 2.36767578125,
+      "epoch": 0.027530280726220914,
+      "grad_norm": 0.5619704127311707,
+      "learning_rate": 9.997292274983696e-06,
+      "loss": 2.3234,
+      "mean_token_accuracy": 0.49107312550768256,
+      "num_tokens": 231959373.0,
+      "step": 1599
+    },
+    {
+      "entropy": 2.4547119140625,
+      "epoch": 0.027547497912416175,
+      "grad_norm": 0.6032162308692932,
+      "learning_rate": 9.997283212296211e-06,
+      "loss": 2.4341,
+      "mean_token_accuracy": 0.4722294365055859,
+      "num_tokens": 232110415.0,
+      "step": 1600
+    },
+    {
+      "entropy": 2.40576171875,
+      "epoch": 0.027564715098611435,
+      "grad_norm": 0.6166012287139893,
+      "learning_rate": 9.99727413447188e-06,
+      "loss": 2.3535,
+      "mean_token_accuracy": 0.489782630931586,
+      "num_tokens": 232247894.0,
+      "step": 1601
+    },
+    {
+      "entropy": 2.4820556640625,
+      "epoch": 0.027581932284806696,
+      "grad_norm": 0.5851189494132996,
+      "learning_rate": 9.997265041510733e-06,
+      "loss": 2.4721,
+      "mean_token_accuracy": 0.4751852685585618,
+      "num_tokens": 232388779.0,
+      "step": 1602
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.027599149471001953,
+      "grad_norm": 0.5721387267112732,
+      "learning_rate": 9.997255933412797e-06,
+      "loss": 2.4279,
+      "mean_token_accuracy": 0.48017449537292123,
+      "num_tokens": 232534833.0,
+      "step": 1603
+    },
+    {
+      "entropy": 2.4854736328125,
+      "epoch": 0.027616366657197213,
+      "grad_norm": 0.5649420022964478,
+      "learning_rate": 9.997246810178099e-06,
+      "loss": 2.4592,
+      "mean_token_accuracy": 0.4723159852437675,
+      "num_tokens": 232685772.0,
+      "step": 1604
+    },
+    {
+      "entropy": 2.4073486328125,
+      "epoch": 0.027633583843392474,
+      "grad_norm": 1.3038853406906128,
+      "learning_rate": 9.997237671806665e-06,
+      "loss": 2.3341,
+      "mean_token_accuracy": 0.4938134755939245,
+      "num_tokens": 232844671.0,
+      "step": 1605
+    },
+    {
+      "entropy": 2.388671875,
+      "epoch": 0.027650801029587734,
+      "grad_norm": 0.5900284647941589,
+      "learning_rate": 9.997228518298527e-06,
+      "loss": 2.338,
+      "mean_token_accuracy": 0.4900298644788563,
+      "num_tokens": 232985544.0,
+      "step": 1606
+    },
+    {
+      "entropy": 2.411376953125,
+      "epoch": 0.027668018215782995,
+      "grad_norm": 0.5831287503242493,
+      "learning_rate": 9.997219349653708e-06,
+      "loss": 2.3888,
+      "mean_token_accuracy": 0.48680589720606804,
+      "num_tokens": 233120888.0,
+      "step": 1607
+    },
+    {
+      "entropy": 2.5101318359375,
+      "epoch": 0.027685235401978255,
+      "grad_norm": 0.5937960743904114,
+      "learning_rate": 9.99721016587224e-06,
+      "loss": 2.4551,
+      "mean_token_accuracy": 0.47399421967566013,
+      "num_tokens": 233260349.0,
+      "step": 1608
+    },
+    {
+      "entropy": 2.4505615234375,
+      "epoch": 0.027702452588173516,
+      "grad_norm": 0.5459933876991272,
+      "learning_rate": 9.997200966954149e-06,
+      "loss": 2.4176,
+      "mean_token_accuracy": 0.47501937579363585,
+      "num_tokens": 233405845.0,
+      "step": 1609
+    },
+    {
+      "entropy": 2.4483642578125,
+      "epoch": 0.027719669774368776,
+      "grad_norm": 0.5332619547843933,
+      "learning_rate": 9.997191752899462e-06,
+      "loss": 2.3866,
+      "mean_token_accuracy": 0.47997900983318686,
+      "num_tokens": 233570774.0,
+      "step": 1610
+    },
+    {
+      "entropy": 2.4091796875,
+      "epoch": 0.027736886960564033,
+      "grad_norm": 0.57257479429245,
+      "learning_rate": 9.997182523708208e-06,
+      "loss": 2.3707,
+      "mean_token_accuracy": 0.4883900200948119,
+      "num_tokens": 233720314.0,
+      "step": 1611
+    },
+    {
+      "entropy": 2.4427490234375,
+      "epoch": 0.027754104146759294,
+      "grad_norm": 0.5707916617393494,
+      "learning_rate": 9.997173279380415e-06,
+      "loss": 2.4081,
+      "mean_token_accuracy": 0.48100787540897727,
+      "num_tokens": 233865434.0,
+      "step": 1612
+    },
+    {
+      "entropy": 2.4808349609375,
+      "epoch": 0.027771321332954554,
+      "grad_norm": 0.6137810945510864,
+      "learning_rate": 9.997164019916109e-06,
+      "loss": 2.4036,
+      "mean_token_accuracy": 0.4781500454992056,
+      "num_tokens": 234000908.0,
+      "step": 1613
+    },
+    {
+      "entropy": 2.44140625,
+      "epoch": 0.027788538519149815,
+      "grad_norm": 0.569161593914032,
+      "learning_rate": 9.997154745315321e-06,
+      "loss": 2.4136,
+      "mean_token_accuracy": 0.48161053331568837,
+      "num_tokens": 234154080.0,
+      "step": 1614
+    },
+    {
+      "entropy": 2.43212890625,
+      "epoch": 0.027805755705345075,
+      "grad_norm": 0.5593316555023193,
+      "learning_rate": 9.997145455578076e-06,
+      "loss": 2.3662,
+      "mean_token_accuracy": 0.4847466191276908,
+      "num_tokens": 234304655.0,
+      "step": 1615
+    },
+    {
+      "entropy": 2.4293212890625,
+      "epoch": 0.027822972891540336,
+      "grad_norm": 0.6409513354301453,
+      "learning_rate": 9.997136150704407e-06,
+      "loss": 2.3821,
+      "mean_token_accuracy": 0.48784798802807927,
+      "num_tokens": 234428782.0,
+      "step": 1616
+    },
+    {
+      "entropy": 2.4359130859375,
+      "epoch": 0.027840190077735597,
+      "grad_norm": 0.5953328013420105,
+      "learning_rate": 9.997126830694336e-06,
+      "loss": 2.3689,
+      "mean_token_accuracy": 0.48606695234775543,
+      "num_tokens": 234563589.0,
+      "step": 1617
+    },
+    {
+      "entropy": 2.4755859375,
+      "epoch": 0.027857407263930857,
+      "grad_norm": 1.202374815940857,
+      "learning_rate": 9.997117495547897e-06,
+      "loss": 2.4366,
+      "mean_token_accuracy": 0.47866149339824915,
+      "num_tokens": 234713984.0,
+      "step": 1618
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.027874624450126118,
+      "grad_norm": 0.6060045957565308,
+      "learning_rate": 9.997108145265113e-06,
+      "loss": 2.4131,
+      "mean_token_accuracy": 0.48174894973635674,
+      "num_tokens": 234856641.0,
+      "step": 1619
+    },
+    {
+      "entropy": 2.4287109375,
+      "epoch": 0.027891841636321375,
+      "grad_norm": 0.5465239882469177,
+      "learning_rate": 9.997098779846017e-06,
+      "loss": 2.3764,
+      "mean_token_accuracy": 0.48386607645079494,
+      "num_tokens": 235015161.0,
+      "step": 1620
+    },
+    {
+      "entropy": 2.4818115234375,
+      "epoch": 0.027909058822516635,
+      "grad_norm": 0.5660187602043152,
+      "learning_rate": 9.997089399290635e-06,
+      "loss": 2.4657,
+      "mean_token_accuracy": 0.47468016063794494,
+      "num_tokens": 235151596.0,
+      "step": 1621
+    },
+    {
+      "entropy": 2.471435546875,
+      "epoch": 0.027926276008711896,
+      "grad_norm": 0.572928786277771,
+      "learning_rate": 9.997080003598995e-06,
+      "loss": 2.4231,
+      "mean_token_accuracy": 0.47549346927553415,
+      "num_tokens": 235291268.0,
+      "step": 1622
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.027943493194907156,
+      "grad_norm": 0.566953182220459,
+      "learning_rate": 9.997070592771127e-06,
+      "loss": 2.4507,
+      "mean_token_accuracy": 0.47433658177033067,
+      "num_tokens": 235435246.0,
+      "step": 1623
+    },
+    {
+      "entropy": 2.448486328125,
+      "epoch": 0.027960710381102417,
+      "grad_norm": 0.6662017703056335,
+      "learning_rate": 9.997061166807057e-06,
+      "loss": 2.4008,
+      "mean_token_accuracy": 0.48391238413751125,
+      "num_tokens": 235570223.0,
+      "step": 1624
+    },
+    {
+      "entropy": 2.416015625,
+      "epoch": 0.027977927567297677,
+      "grad_norm": 0.5774509906768799,
+      "learning_rate": 9.997051725706816e-06,
+      "loss": 2.3538,
+      "mean_token_accuracy": 0.48118720296770334,
+      "num_tokens": 235725981.0,
+      "step": 1625
+    },
+    {
+      "entropy": 2.4285888671875,
+      "epoch": 0.027995144753492938,
+      "grad_norm": 0.5478702783584595,
+      "learning_rate": 9.997042269470431e-06,
+      "loss": 2.3242,
+      "mean_token_accuracy": 0.49387524649500847,
+      "num_tokens": 235876462.0,
+      "step": 1626
+    },
+    {
+      "entropy": 2.462158203125,
+      "epoch": 0.0280123619396882,
+      "grad_norm": 0.573614776134491,
+      "learning_rate": 9.997032798097935e-06,
+      "loss": 2.425,
+      "mean_token_accuracy": 0.48535694507882,
+      "num_tokens": 236015336.0,
+      "step": 1627
+    },
+    {
+      "entropy": 2.50634765625,
+      "epoch": 0.028029579125883455,
+      "grad_norm": 0.6228254437446594,
+      "learning_rate": 9.997023311589349e-06,
+      "loss": 2.4627,
+      "mean_token_accuracy": 0.47959386138245463,
+      "num_tokens": 236155028.0,
+      "step": 1628
+    },
+    {
+      "entropy": 2.5274658203125,
+      "epoch": 0.028046796312078716,
+      "grad_norm": 0.6030795574188232,
+      "learning_rate": 9.997013809944708e-06,
+      "loss": 2.4979,
+      "mean_token_accuracy": 0.4653917569667101,
+      "num_tokens": 236309148.0,
+      "step": 1629
+    },
+    {
+      "entropy": 2.388916015625,
+      "epoch": 0.028064013498273976,
+      "grad_norm": 0.558562695980072,
+      "learning_rate": 9.997004293164038e-06,
+      "loss": 2.3689,
+      "mean_token_accuracy": 0.48823840310797095,
+      "num_tokens": 236475648.0,
+      "step": 1630
+    },
+    {
+      "entropy": 2.43798828125,
+      "epoch": 0.028081230684469237,
+      "grad_norm": 0.5895214080810547,
+      "learning_rate": 9.996994761247368e-06,
+      "loss": 2.4171,
+      "mean_token_accuracy": 0.4803608972579241,
+      "num_tokens": 236615733.0,
+      "step": 1631
+    },
+    {
+      "entropy": 2.421875,
+      "epoch": 0.028098447870664497,
+      "grad_norm": 0.5724160075187683,
+      "learning_rate": 9.996985214194727e-06,
+      "loss": 2.3814,
+      "mean_token_accuracy": 0.483953685965389,
+      "num_tokens": 236756289.0,
+      "step": 1632
+    },
+    {
+      "entropy": 2.47021484375,
+      "epoch": 0.028115665056859758,
+      "grad_norm": 0.5840807557106018,
+      "learning_rate": 9.996975652006146e-06,
+      "loss": 2.4387,
+      "mean_token_accuracy": 0.4743587113916874,
+      "num_tokens": 236887859.0,
+      "step": 1633
+    },
+    {
+      "entropy": 2.46826171875,
+      "epoch": 0.02813288224305502,
+      "grad_norm": 0.5316832065582275,
+      "learning_rate": 9.996966074681651e-06,
+      "loss": 2.4542,
+      "mean_token_accuracy": 0.4724106374196708,
+      "num_tokens": 237042538.0,
+      "step": 1634
+    },
+    {
+      "entropy": 2.4212646484375,
+      "epoch": 0.02815009942925028,
+      "grad_norm": 0.5952966809272766,
+      "learning_rate": 9.996956482221273e-06,
+      "loss": 2.4116,
+      "mean_token_accuracy": 0.4807650912553072,
+      "num_tokens": 237190659.0,
+      "step": 1635
+    },
+    {
+      "entropy": 2.4124755859375,
+      "epoch": 0.028167316615445536,
+      "grad_norm": 0.5247029066085815,
+      "learning_rate": 9.99694687462504e-06,
+      "loss": 2.359,
+      "mean_token_accuracy": 0.48549360083416104,
+      "num_tokens": 237358115.0,
+      "step": 1636
+    },
+    {
+      "entropy": 2.387939453125,
+      "epoch": 0.028184533801640797,
+      "grad_norm": 0.5859695672988892,
+      "learning_rate": 9.996937251892982e-06,
+      "loss": 2.3148,
+      "mean_token_accuracy": 0.49537815153598785,
+      "num_tokens": 237496922.0,
+      "step": 1637
+    },
+    {
+      "entropy": 2.4844970703125,
+      "epoch": 0.028201750987836057,
+      "grad_norm": 0.6078894734382629,
+      "learning_rate": 9.996927614025127e-06,
+      "loss": 2.443,
+      "mean_token_accuracy": 0.4838123098015785,
+      "num_tokens": 237623117.0,
+      "step": 1638
+    },
+    {
+      "entropy": 2.533203125,
+      "epoch": 0.028218968174031318,
+      "grad_norm": 0.5760785341262817,
+      "learning_rate": 9.996917961021504e-06,
+      "loss": 2.4848,
+      "mean_token_accuracy": 0.4731389367952943,
+      "num_tokens": 237763918.0,
+      "step": 1639
+    },
+    {
+      "entropy": 2.46826171875,
+      "epoch": 0.028236185360226578,
+      "grad_norm": 0.539117157459259,
+      "learning_rate": 9.996908292882144e-06,
+      "loss": 2.4377,
+      "mean_token_accuracy": 0.47789577953517437,
+      "num_tokens": 237926377.0,
+      "step": 1640
+    },
+    {
+      "entropy": 2.465087890625,
+      "epoch": 0.02825340254642184,
+      "grad_norm": 0.5144519209861755,
+      "learning_rate": 9.996898609607075e-06,
+      "loss": 2.4543,
+      "mean_token_accuracy": 0.474421392660588,
+      "num_tokens": 238089528.0,
+      "step": 1641
+    },
+    {
+      "entropy": 2.4031982421875,
+      "epoch": 0.0282706197326171,
+      "grad_norm": 0.5754570960998535,
+      "learning_rate": 9.996888911196326e-06,
+      "loss": 2.3766,
+      "mean_token_accuracy": 0.4901612875983119,
+      "num_tokens": 238231433.0,
+      "step": 1642
+    },
+    {
+      "entropy": 2.4429931640625,
+      "epoch": 0.02828783691881236,
+      "grad_norm": 0.6072561740875244,
+      "learning_rate": 9.996879197649927e-06,
+      "loss": 2.394,
+      "mean_token_accuracy": 0.4848616924136877,
+      "num_tokens": 238354749.0,
+      "step": 1643
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.02830505410500762,
+      "grad_norm": 0.70046466588974,
+      "learning_rate": 9.996869468967909e-06,
+      "loss": 2.3639,
+      "mean_token_accuracy": 0.48499562544748187,
+      "num_tokens": 238506100.0,
+      "step": 1644
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.028322271291202877,
+      "grad_norm": 0.6182661652565002,
+      "learning_rate": 9.996859725150299e-06,
+      "loss": 2.3903,
+      "mean_token_accuracy": 0.4828766668215394,
+      "num_tokens": 238651208.0,
+      "step": 1645
+    },
+    {
+      "entropy": 2.39111328125,
+      "epoch": 0.028339488477398138,
+      "grad_norm": 0.5814700126647949,
+      "learning_rate": 9.996849966197127e-06,
+      "loss": 2.3526,
+      "mean_token_accuracy": 0.48803521366789937,
+      "num_tokens": 238805878.0,
+      "step": 1646
+    },
+    {
+      "entropy": 2.37841796875,
+      "epoch": 0.0283567056635934,
+      "grad_norm": 0.5634573698043823,
+      "learning_rate": 9.996840192108421e-06,
+      "loss": 2.3943,
+      "mean_token_accuracy": 0.48795266123488545,
+      "num_tokens": 238958546.0,
+      "step": 1647
+    },
+    {
+      "entropy": 2.5108642578125,
+      "epoch": 0.02837392284978866,
+      "grad_norm": 0.6036639213562012,
+      "learning_rate": 9.996830402884216e-06,
+      "loss": 2.4941,
+      "mean_token_accuracy": 0.4679912384599447,
+      "num_tokens": 239087208.0,
+      "step": 1648
+    },
+    {
+      "entropy": 2.4544677734375,
+      "epoch": 0.02839114003598392,
+      "grad_norm": 0.5786367654800415,
+      "learning_rate": 9.996820598524536e-06,
+      "loss": 2.4108,
+      "mean_token_accuracy": 0.4794202446937561,
+      "num_tokens": 239230718.0,
+      "step": 1649
+    },
+    {
+      "entropy": 2.427490234375,
+      "epoch": 0.02840835722217918,
+      "grad_norm": 0.5587934255599976,
+      "learning_rate": 9.996810779029413e-06,
+      "loss": 2.4071,
+      "mean_token_accuracy": 0.48648894764482975,
+      "num_tokens": 239388616.0,
+      "step": 1650
+    },
+    {
+      "entropy": 2.488037109375,
+      "epoch": 0.02842557440837444,
+      "grad_norm": 0.5868260860443115,
+      "learning_rate": 9.996800944398879e-06,
+      "loss": 2.4666,
+      "mean_token_accuracy": 0.47458827355876565,
+      "num_tokens": 239532563.0,
+      "step": 1651
+    },
+    {
+      "entropy": 2.3829345703125,
+      "epoch": 0.0284427915945697,
+      "grad_norm": 0.5640127658843994,
+      "learning_rate": 9.996791094632958e-06,
+      "loss": 2.3417,
+      "mean_token_accuracy": 0.4937547417357564,
+      "num_tokens": 239675698.0,
+      "step": 1652
+    },
+    {
+      "entropy": 2.542724609375,
+      "epoch": 0.028460008780764958,
+      "grad_norm": 0.5434319972991943,
+      "learning_rate": 9.996781229731685e-06,
+      "loss": 2.4758,
+      "mean_token_accuracy": 0.4686996676027775,
+      "num_tokens": 239821208.0,
+      "step": 1653
+    },
+    {
+      "entropy": 2.4918212890625,
+      "epoch": 0.02847722596696022,
+      "grad_norm": 0.5542386174201965,
+      "learning_rate": 9.996771349695089e-06,
+      "loss": 2.4299,
+      "mean_token_accuracy": 0.47332891169935465,
+      "num_tokens": 239965661.0,
+      "step": 1654
+    },
+    {
+      "entropy": 2.4259033203125,
+      "epoch": 0.02849444315315548,
+      "grad_norm": 0.5533227324485779,
+      "learning_rate": 9.996761454523198e-06,
+      "loss": 2.4194,
+      "mean_token_accuracy": 0.48256733594462276,
+      "num_tokens": 240122225.0,
+      "step": 1655
+    },
+    {
+      "entropy": 2.398193359375,
+      "epoch": 0.02851166033935074,
+      "grad_norm": 0.5755099058151245,
+      "learning_rate": 9.996751544216043e-06,
+      "loss": 2.3681,
+      "mean_token_accuracy": 0.48602569615468383,
+      "num_tokens": 240255685.0,
+      "step": 1656
+    },
+    {
+      "entropy": 2.336669921875,
+      "epoch": 0.028528877525546,
+      "grad_norm": 0.573204517364502,
+      "learning_rate": 9.996741618773654e-06,
+      "loss": 2.2821,
+      "mean_token_accuracy": 0.5057984632439911,
+      "num_tokens": 240412754.0,
+      "step": 1657
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.02854609471174126,
+      "grad_norm": 0.6060733199119568,
+      "learning_rate": 9.99673167819606e-06,
+      "loss": 2.4604,
+      "mean_token_accuracy": 0.47678601182997227,
+      "num_tokens": 240540902.0,
+      "step": 1658
+    },
+    {
+      "entropy": 2.4752197265625,
+      "epoch": 0.02856331189793652,
+      "grad_norm": 0.5993353724479675,
+      "learning_rate": 9.996721722483296e-06,
+      "loss": 2.4133,
+      "mean_token_accuracy": 0.483869846444577,
+      "num_tokens": 240689582.0,
+      "step": 1659
+    },
+    {
+      "entropy": 2.38134765625,
+      "epoch": 0.028580529084131782,
+      "grad_norm": 0.5666155815124512,
+      "learning_rate": 9.996711751635384e-06,
+      "loss": 2.3604,
+      "mean_token_accuracy": 0.491028075106442,
+      "num_tokens": 240844782.0,
+      "step": 1660
+    },
+    {
+      "entropy": 2.441650390625,
+      "epoch": 0.02859774627032704,
+      "grad_norm": 0.5586746335029602,
+      "learning_rate": 9.996701765652361e-06,
+      "loss": 2.413,
+      "mean_token_accuracy": 0.4783767703920603,
+      "num_tokens": 240991388.0,
+      "step": 1661
+    },
+    {
+      "entropy": 2.455322265625,
+      "epoch": 0.0286149634565223,
+      "grad_norm": 0.5903511047363281,
+      "learning_rate": 9.996691764534255e-06,
+      "loss": 2.4062,
+      "mean_token_accuracy": 0.48205051850527525,
+      "num_tokens": 241124952.0,
+      "step": 1662
+    },
+    {
+      "entropy": 2.498046875,
+      "epoch": 0.02863218064271756,
+      "grad_norm": 0.5668911933898926,
+      "learning_rate": 9.996681748281094e-06,
+      "loss": 2.4549,
+      "mean_token_accuracy": 0.4747361782938242,
+      "num_tokens": 241262036.0,
+      "step": 1663
+    },
+    {
+      "entropy": 2.5220947265625,
+      "epoch": 0.02864939782891282,
+      "grad_norm": 0.6033514738082886,
+      "learning_rate": 9.996671716892914e-06,
+      "loss": 2.5395,
+      "mean_token_accuracy": 0.46681966073811054,
+      "num_tokens": 241416917.0,
+      "step": 1664
+    },
+    {
+      "entropy": 2.4417724609375,
+      "epoch": 0.02866661501510808,
+      "grad_norm": 0.6681655645370483,
+      "learning_rate": 9.996661670369739e-06,
+      "loss": 2.3974,
+      "mean_token_accuracy": 0.48373745242133737,
+      "num_tokens": 241554068.0,
+      "step": 1665
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.02868383220130334,
+      "grad_norm": 0.5739055275917053,
+      "learning_rate": 9.996651608711603e-06,
+      "loss": 2.361,
+      "mean_token_accuracy": 0.48040412552654743,
+      "num_tokens": 241714285.0,
+      "step": 1666
+    },
+    {
+      "entropy": 2.4683837890625,
+      "epoch": 0.028701049387498602,
+      "grad_norm": 0.6406212449073792,
+      "learning_rate": 9.996641531918536e-06,
+      "loss": 2.43,
+      "mean_token_accuracy": 0.48026127368211746,
+      "num_tokens": 241841614.0,
+      "step": 1667
+    },
+    {
+      "entropy": 2.5107421875,
+      "epoch": 0.028718266573693862,
+      "grad_norm": 0.5852003693580627,
+      "learning_rate": 9.996631439990568e-06,
+      "loss": 2.4992,
+      "mean_token_accuracy": 0.4704901375807822,
+      "num_tokens": 241980442.0,
+      "step": 1668
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.028735483759889123,
+      "grad_norm": 0.6459202170372009,
+      "learning_rate": 9.996621332927729e-06,
+      "loss": 2.4125,
+      "mean_token_accuracy": 0.4840882923454046,
+      "num_tokens": 242131950.0,
+      "step": 1669
+    },
+    {
+      "entropy": 2.4384765625,
+      "epoch": 0.02875270094608438,
+      "grad_norm": 0.6423004269599915,
+      "learning_rate": 9.996611210730053e-06,
+      "loss": 2.4433,
+      "mean_token_accuracy": 0.48514924151822925,
+      "num_tokens": 242274852.0,
+      "step": 1670
+    },
+    {
+      "entropy": 2.502685546875,
+      "epoch": 0.02876991813227964,
+      "grad_norm": 0.5465385913848877,
+      "learning_rate": 9.996601073397568e-06,
+      "loss": 2.4691,
+      "mean_token_accuracy": 0.4714298346079886,
+      "num_tokens": 242426188.0,
+      "step": 1671
+    },
+    {
+      "entropy": 2.405517578125,
+      "epoch": 0.0287871353184749,
+      "grad_norm": 0.5957438945770264,
+      "learning_rate": 9.996590920930303e-06,
+      "loss": 2.3951,
+      "mean_token_accuracy": 0.48868590872734785,
+      "num_tokens": 242571018.0,
+      "step": 1672
+    },
+    {
+      "entropy": 2.550537109375,
+      "epoch": 0.02880435250467016,
+      "grad_norm": 0.5585090517997742,
+      "learning_rate": 9.99658075332829e-06,
+      "loss": 2.5181,
+      "mean_token_accuracy": 0.46476193610578775,
+      "num_tokens": 242721700.0,
+      "step": 1673
+    },
+    {
+      "entropy": 2.4349365234375,
+      "epoch": 0.028821569690865422,
+      "grad_norm": 0.6010538935661316,
+      "learning_rate": 9.996570570591561e-06,
+      "loss": 2.3535,
+      "mean_token_accuracy": 0.48848696844652295,
+      "num_tokens": 242858693.0,
+      "step": 1674
+    },
+    {
+      "entropy": 2.4791259765625,
+      "epoch": 0.028838786877060683,
+      "grad_norm": 0.5733229517936707,
+      "learning_rate": 9.996560372720147e-06,
+      "loss": 2.4825,
+      "mean_token_accuracy": 0.4683424327522516,
+      "num_tokens": 243006868.0,
+      "step": 1675
+    },
+    {
+      "entropy": 2.4215087890625,
+      "epoch": 0.028856004063255943,
+      "grad_norm": 0.5834994316101074,
+      "learning_rate": 9.996550159714078e-06,
+      "loss": 2.3763,
+      "mean_token_accuracy": 0.4828924615867436,
+      "num_tokens": 243154886.0,
+      "step": 1676
+    },
+    {
+      "entropy": 2.44775390625,
+      "epoch": 0.028873221249451204,
+      "grad_norm": 0.6166858077049255,
+      "learning_rate": 9.996539931573385e-06,
+      "loss": 2.3781,
+      "mean_token_accuracy": 0.4847164931707084,
+      "num_tokens": 243288516.0,
+      "step": 1677
+    },
+    {
+      "entropy": 2.49072265625,
+      "epoch": 0.02889043843564646,
+      "grad_norm": 0.5641864538192749,
+      "learning_rate": 9.996529688298098e-06,
+      "loss": 2.4465,
+      "mean_token_accuracy": 0.473747827578336,
+      "num_tokens": 243441352.0,
+      "step": 1678
+    },
+    {
+      "entropy": 2.451171875,
+      "epoch": 0.02890765562184172,
+      "grad_norm": 0.6030487418174744,
+      "learning_rate": 9.996519429888249e-06,
+      "loss": 2.4481,
+      "mean_token_accuracy": 0.4830109137110412,
+      "num_tokens": 243588150.0,
+      "step": 1679
+    },
+    {
+      "entropy": 2.4708251953125,
+      "epoch": 0.028924872808036982,
+      "grad_norm": 0.5804004073143005,
+      "learning_rate": 9.99650915634387e-06,
+      "loss": 2.4297,
+      "mean_token_accuracy": 0.47798701329156756,
+      "num_tokens": 243737262.0,
+      "step": 1680
+    },
+    {
+      "entropy": 2.4835205078125,
+      "epoch": 0.028942089994232242,
+      "grad_norm": 0.5946245789527893,
+      "learning_rate": 9.996498867664992e-06,
+      "loss": 2.4639,
+      "mean_token_accuracy": 0.47278458066284657,
+      "num_tokens": 243872204.0,
+      "step": 1681
+    },
+    {
+      "entropy": 2.450439453125,
+      "epoch": 0.028959307180427503,
+      "grad_norm": 0.6240355372428894,
+      "learning_rate": 9.996488563851646e-06,
+      "loss": 2.3891,
+      "mean_token_accuracy": 0.48081753915175796,
+      "num_tokens": 243998841.0,
+      "step": 1682
+    },
+    {
+      "entropy": 2.385009765625,
+      "epoch": 0.028976524366622763,
+      "grad_norm": 0.5424374341964722,
+      "learning_rate": 9.99647824490386e-06,
+      "loss": 2.3166,
+      "mean_token_accuracy": 0.490779772400856,
+      "num_tokens": 244149478.0,
+      "step": 1683
+    },
+    {
+      "entropy": 2.4439697265625,
+      "epoch": 0.028993741552818024,
+      "grad_norm": 0.718917727470398,
+      "learning_rate": 9.99646791082167e-06,
+      "loss": 2.4061,
+      "mean_token_accuracy": 0.4811778524890542,
+      "num_tokens": 244291993.0,
+      "step": 1684
+    },
+    {
+      "entropy": 2.394287109375,
+      "epoch": 0.029010958739013284,
+      "grad_norm": 0.6151554584503174,
+      "learning_rate": 9.996457561605105e-06,
+      "loss": 2.3617,
+      "mean_token_accuracy": 0.48714847723022103,
+      "num_tokens": 244435849.0,
+      "step": 1685
+    },
+    {
+      "entropy": 2.4591064453125,
+      "epoch": 0.02902817592520854,
+      "grad_norm": 0.5844502449035645,
+      "learning_rate": 9.996447197254195e-06,
+      "loss": 2.4036,
+      "mean_token_accuracy": 0.48159506963565946,
+      "num_tokens": 244580376.0,
+      "step": 1686
+    },
+    {
+      "entropy": 2.40625,
+      "epoch": 0.029045393111403802,
+      "grad_norm": 0.5591570734977722,
+      "learning_rate": 9.996436817768974e-06,
+      "loss": 2.3592,
+      "mean_token_accuracy": 0.49357689963653684,
+      "num_tokens": 244729260.0,
+      "step": 1687
+    },
+    {
+      "entropy": 2.4224853515625,
+      "epoch": 0.029062610297599063,
+      "grad_norm": 0.5616257190704346,
+      "learning_rate": 9.996426423149472e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.482600009534508,
+      "num_tokens": 244872812.0,
+      "step": 1688
+    },
+    {
+      "entropy": 2.500244140625,
+      "epoch": 0.029079827483794323,
+      "grad_norm": 0.5827319622039795,
+      "learning_rate": 9.996416013395721e-06,
+      "loss": 2.4213,
+      "mean_token_accuracy": 0.47733526350930333,
+      "num_tokens": 245013545.0,
+      "step": 1689
+    },
+    {
+      "entropy": 2.4273681640625,
+      "epoch": 0.029097044669989584,
+      "grad_norm": 0.5836253762245178,
+      "learning_rate": 9.996405588507753e-06,
+      "loss": 2.4129,
+      "mean_token_accuracy": 0.48227705620229244,
+      "num_tokens": 245158448.0,
+      "step": 1690
+    },
+    {
+      "entropy": 2.41015625,
+      "epoch": 0.029114261856184844,
+      "grad_norm": 0.5588595271110535,
+      "learning_rate": 9.996395148485598e-06,
+      "loss": 2.3954,
+      "mean_token_accuracy": 0.48199733812361956,
+      "num_tokens": 245323743.0,
+      "step": 1691
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.029131479042380105,
+      "grad_norm": 0.585265576839447,
+      "learning_rate": 9.996384693329289e-06,
+      "loss": 2.3807,
+      "mean_token_accuracy": 0.4879599893465638,
+      "num_tokens": 245479102.0,
+      "step": 1692
+    },
+    {
+      "entropy": 2.47900390625,
+      "epoch": 0.029148696228575365,
+      "grad_norm": 0.5371046662330627,
+      "learning_rate": 9.996374223038858e-06,
+      "loss": 2.4331,
+      "mean_token_accuracy": 0.47392465313896537,
+      "num_tokens": 245627290.0,
+      "step": 1693
+    },
+    {
+      "entropy": 2.4324951171875,
+      "epoch": 0.029165913414770626,
+      "grad_norm": 0.608191967010498,
+      "learning_rate": 9.996363737614334e-06,
+      "loss": 2.3786,
+      "mean_token_accuracy": 0.4849713812582195,
+      "num_tokens": 245761478.0,
+      "step": 1694
+    },
+    {
+      "entropy": 2.5067138671875,
+      "epoch": 0.029183130600965883,
+      "grad_norm": 0.5639045834541321,
+      "learning_rate": 9.996353237055753e-06,
+      "loss": 2.4584,
+      "mean_token_accuracy": 0.47028431948274374,
+      "num_tokens": 245900244.0,
+      "step": 1695
+    },
+    {
+      "entropy": 2.4661865234375,
+      "epoch": 0.029200347787161143,
+      "grad_norm": 0.5640761852264404,
+      "learning_rate": 9.996342721363144e-06,
+      "loss": 2.4086,
+      "mean_token_accuracy": 0.47613369347527623,
+      "num_tokens": 246061900.0,
+      "step": 1696
+    },
+    {
+      "entropy": 2.413818359375,
+      "epoch": 0.029217564973356404,
+      "grad_norm": 0.5837756395339966,
+      "learning_rate": 9.996332190536538e-06,
+      "loss": 2.3655,
+      "mean_token_accuracy": 0.48909167293459177,
+      "num_tokens": 246203266.0,
+      "step": 1697
+    },
+    {
+      "entropy": 2.466064453125,
+      "epoch": 0.029234782159551664,
+      "grad_norm": 0.6410163044929504,
+      "learning_rate": 9.99632164457597e-06,
+      "loss": 2.4183,
+      "mean_token_accuracy": 0.47384981345385313,
+      "num_tokens": 246345295.0,
+      "step": 1698
+    },
+    {
+      "entropy": 2.474365234375,
+      "epoch": 0.029251999345746925,
+      "grad_norm": 0.5565051436424255,
+      "learning_rate": 9.99631108348147e-06,
+      "loss": 2.452,
+      "mean_token_accuracy": 0.4777741697616875,
+      "num_tokens": 246501317.0,
+      "step": 1699
+    },
+    {
+      "entropy": 2.4288330078125,
+      "epoch": 0.029269216531942185,
+      "grad_norm": 0.5428564548492432,
+      "learning_rate": 9.996300507253068e-06,
+      "loss": 2.3949,
+      "mean_token_accuracy": 0.48448564764112234,
+      "num_tokens": 246654551.0,
+      "step": 1700
+    },
+    {
+      "entropy": 2.4349365234375,
+      "epoch": 0.029286433718137446,
+      "grad_norm": 0.5566284656524658,
+      "learning_rate": 9.9962899158908e-06,
+      "loss": 2.4152,
+      "mean_token_accuracy": 0.4823923450894654,
+      "num_tokens": 246816439.0,
+      "step": 1701
+    },
+    {
+      "entropy": 2.48681640625,
+      "epoch": 0.029303650904332706,
+      "grad_norm": 0.5799583792686462,
+      "learning_rate": 9.996279309394697e-06,
+      "loss": 2.4778,
+      "mean_token_accuracy": 0.4676981116645038,
+      "num_tokens": 246961368.0,
+      "step": 1702
+    },
+    {
+      "entropy": 2.494140625,
+      "epoch": 0.029320868090527964,
+      "grad_norm": 0.5828601121902466,
+      "learning_rate": 9.99626868776479e-06,
+      "loss": 2.491,
+      "mean_token_accuracy": 0.4712865217588842,
+      "num_tokens": 247098743.0,
+      "step": 1703
+    },
+    {
+      "entropy": 2.4791259765625,
+      "epoch": 0.029338085276723224,
+      "grad_norm": 0.6346420645713806,
+      "learning_rate": 9.99625805100111e-06,
+      "loss": 2.462,
+      "mean_token_accuracy": 0.4806337282061577,
+      "num_tokens": 247245594.0,
+      "step": 1704
+    },
+    {
+      "entropy": 2.4534912109375,
+      "epoch": 0.029355302462918485,
+      "grad_norm": 0.5871755480766296,
+      "learning_rate": 9.996247399103693e-06,
+      "loss": 2.4171,
+      "mean_token_accuracy": 0.4820070252753794,
+      "num_tokens": 247385625.0,
+      "step": 1705
+    },
+    {
+      "entropy": 2.427001953125,
+      "epoch": 0.029372519649113745,
+      "grad_norm": 0.570570170879364,
+      "learning_rate": 9.996236732072568e-06,
+      "loss": 2.4239,
+      "mean_token_accuracy": 0.4824762148782611,
+      "num_tokens": 247535659.0,
+      "step": 1706
+    },
+    {
+      "entropy": 2.4500732421875,
+      "epoch": 0.029389736835309006,
+      "grad_norm": 0.5755301713943481,
+      "learning_rate": 9.99622604990777e-06,
+      "loss": 2.4254,
+      "mean_token_accuracy": 0.48341484228149056,
+      "num_tokens": 247679680.0,
+      "step": 1707
+    },
+    {
+      "entropy": 2.4388427734375,
+      "epoch": 0.029406954021504266,
+      "grad_norm": 0.5620051622390747,
+      "learning_rate": 9.996215352609327e-06,
+      "loss": 2.4294,
+      "mean_token_accuracy": 0.4801658443175256,
+      "num_tokens": 247823282.0,
+      "step": 1708
+    },
+    {
+      "entropy": 2.495361328125,
+      "epoch": 0.029424171207699527,
+      "grad_norm": 0.5927258133888245,
+      "learning_rate": 9.996204640177276e-06,
+      "loss": 2.4746,
+      "mean_token_accuracy": 0.47673043003305793,
+      "num_tokens": 247963838.0,
+      "step": 1709
+    },
+    {
+      "entropy": 2.48583984375,
+      "epoch": 0.029441388393894787,
+      "grad_norm": 0.620323657989502,
+      "learning_rate": 9.996193912611648e-06,
+      "loss": 2.4201,
+      "mean_token_accuracy": 0.4740843581967056,
+      "num_tokens": 248101252.0,
+      "step": 1710
+    },
+    {
+      "entropy": 2.49658203125,
+      "epoch": 0.029458605580090044,
+      "grad_norm": 0.5656887888908386,
+      "learning_rate": 9.996183169912474e-06,
+      "loss": 2.4453,
+      "mean_token_accuracy": 0.4755016630515456,
+      "num_tokens": 248251659.0,
+      "step": 1711
+    },
+    {
+      "entropy": 2.44482421875,
+      "epoch": 0.029475822766285305,
+      "grad_norm": 0.5744235515594482,
+      "learning_rate": 9.996172412079788e-06,
+      "loss": 2.4092,
+      "mean_token_accuracy": 0.4831536212004721,
+      "num_tokens": 248394514.0,
+      "step": 1712
+    },
+    {
+      "entropy": 2.4427490234375,
+      "epoch": 0.029493039952480565,
+      "grad_norm": 0.5855416059494019,
+      "learning_rate": 9.996161639113622e-06,
+      "loss": 2.399,
+      "mean_token_accuracy": 0.48080876702442765,
+      "num_tokens": 248535417.0,
+      "step": 1713
+    },
+    {
+      "entropy": 2.4171142578125,
+      "epoch": 0.029510257138675826,
+      "grad_norm": 0.6481823325157166,
+      "learning_rate": 9.99615085101401e-06,
+      "loss": 2.3865,
+      "mean_token_accuracy": 0.48705816362053156,
+      "num_tokens": 248672896.0,
+      "step": 1714
+    },
+    {
+      "entropy": 2.397216796875,
+      "epoch": 0.029527474324871086,
+      "grad_norm": 0.525701642036438,
+      "learning_rate": 9.996140047780983e-06,
+      "loss": 2.3877,
+      "mean_token_accuracy": 0.48106732219457626,
+      "num_tokens": 248838851.0,
+      "step": 1715
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.029544691511066347,
+      "grad_norm": 0.5457813143730164,
+      "learning_rate": 9.996129229414573e-06,
+      "loss": 2.362,
+      "mean_token_accuracy": 0.4910212284885347,
+      "num_tokens": 249000921.0,
+      "step": 1716
+    },
+    {
+      "entropy": 2.3861083984375,
+      "epoch": 0.029561908697261607,
+      "grad_norm": 0.5792080163955688,
+      "learning_rate": 9.996118395914816e-06,
+      "loss": 2.3344,
+      "mean_token_accuracy": 0.4931281958706677,
+      "num_tokens": 249142323.0,
+      "step": 1717
+    },
+    {
+      "entropy": 2.4461669921875,
+      "epoch": 0.029579125883456868,
+      "grad_norm": 0.5751327872276306,
+      "learning_rate": 9.996107547281741e-06,
+      "loss": 2.4195,
+      "mean_token_accuracy": 0.4737996533513069,
+      "num_tokens": 249273907.0,
+      "step": 1718
+    },
+    {
+      "entropy": 2.50390625,
+      "epoch": 0.02959634306965213,
+      "grad_norm": 0.5305837988853455,
+      "learning_rate": 9.996096683515385e-06,
+      "loss": 2.4582,
+      "mean_token_accuracy": 0.47129624895751476,
+      "num_tokens": 249420011.0,
+      "step": 1719
+    },
+    {
+      "entropy": 2.50146484375,
+      "epoch": 0.029613560255847386,
+      "grad_norm": 0.5615363121032715,
+      "learning_rate": 9.996085804615776e-06,
+      "loss": 2.4453,
+      "mean_token_accuracy": 0.4726092144846916,
+      "num_tokens": 249577874.0,
+      "step": 1720
+    },
+    {
+      "entropy": 2.4356689453125,
+      "epoch": 0.029630777442042646,
+      "grad_norm": 0.584879457950592,
+      "learning_rate": 9.99607491058295e-06,
+      "loss": 2.3874,
+      "mean_token_accuracy": 0.4838989768177271,
+      "num_tokens": 249733995.0,
+      "step": 1721
+    },
+    {
+      "entropy": 2.4276123046875,
+      "epoch": 0.029647994628237907,
+      "grad_norm": 0.6611423492431641,
+      "learning_rate": 9.996064001416943e-06,
+      "loss": 2.3991,
+      "mean_token_accuracy": 0.4866844713687897,
+      "num_tokens": 249863834.0,
+      "step": 1722
+    },
+    {
+      "entropy": 2.387939453125,
+      "epoch": 0.029665211814433167,
+      "grad_norm": 0.5828907489776611,
+      "learning_rate": 9.996053077117781e-06,
+      "loss": 2.35,
+      "mean_token_accuracy": 0.48920594388619065,
+      "num_tokens": 249999433.0,
+      "step": 1723
+    },
+    {
+      "entropy": 2.5263671875,
+      "epoch": 0.029682429000628428,
+      "grad_norm": 0.6122536063194275,
+      "learning_rate": 9.996042137685502e-06,
+      "loss": 2.5203,
+      "mean_token_accuracy": 0.4699005661532283,
+      "num_tokens": 250134401.0,
+      "step": 1724
+    },
+    {
+      "entropy": 2.44482421875,
+      "epoch": 0.029699646186823688,
+      "grad_norm": 0.6737586259841919,
+      "learning_rate": 9.996031183120138e-06,
+      "loss": 2.3536,
+      "mean_token_accuracy": 0.48844635486602783,
+      "num_tokens": 250292079.0,
+      "step": 1725
+    },
+    {
+      "entropy": 2.47119140625,
+      "epoch": 0.02971686337301895,
+      "grad_norm": 0.5378682613372803,
+      "learning_rate": 9.996020213421722e-06,
+      "loss": 2.4471,
+      "mean_token_accuracy": 0.48123562429100275,
+      "num_tokens": 250449143.0,
+      "step": 1726
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.02973408055921421,
+      "grad_norm": 0.5978697538375854,
+      "learning_rate": 9.996009228590286e-06,
+      "loss": 2.4195,
+      "mean_token_accuracy": 0.48353995755314827,
+      "num_tokens": 250581816.0,
+      "step": 1727
+    },
+    {
+      "entropy": 2.515869140625,
+      "epoch": 0.029751297745409466,
+      "grad_norm": 0.5631060004234314,
+      "learning_rate": 9.995998228625868e-06,
+      "loss": 2.4653,
+      "mean_token_accuracy": 0.473801807500422,
+      "num_tokens": 250717401.0,
+      "step": 1728
+    },
+    {
+      "entropy": 2.482421875,
+      "epoch": 0.029768514931604727,
+      "grad_norm": 0.5445279479026794,
+      "learning_rate": 9.995987213528493e-06,
+      "loss": 2.4938,
+      "mean_token_accuracy": 0.47592513309791684,
+      "num_tokens": 250867240.0,
+      "step": 1729
+    },
+    {
+      "entropy": 2.5250244140625,
+      "epoch": 0.029785732117799987,
+      "grad_norm": 0.5898825526237488,
+      "learning_rate": 9.995976183298204e-06,
+      "loss": 2.4803,
+      "mean_token_accuracy": 0.4719364200718701,
+      "num_tokens": 251016870.0,
+      "step": 1730
+    },
+    {
+      "entropy": 2.4178466796875,
+      "epoch": 0.029802949303995248,
+      "grad_norm": 0.5892136096954346,
+      "learning_rate": 9.995965137935026e-06,
+      "loss": 2.3497,
+      "mean_token_accuracy": 0.48639260046184063,
+      "num_tokens": 251159320.0,
+      "step": 1731
+    },
+    {
+      "entropy": 2.41259765625,
+      "epoch": 0.02982016649019051,
+      "grad_norm": 0.6107653379440308,
+      "learning_rate": 9.995954077439e-06,
+      "loss": 2.3769,
+      "mean_token_accuracy": 0.4900634833611548,
+      "num_tokens": 251294730.0,
+      "step": 1732
+    },
+    {
+      "entropy": 2.41552734375,
+      "epoch": 0.02983738367638577,
+      "grad_norm": 0.5585425496101379,
+      "learning_rate": 9.995943001810151e-06,
+      "loss": 2.3654,
+      "mean_token_accuracy": 0.48663356387987733,
+      "num_tokens": 251438812.0,
+      "step": 1733
+    },
+    {
+      "entropy": 2.45751953125,
+      "epoch": 0.02985460086258103,
+      "grad_norm": 0.5406437516212463,
+      "learning_rate": 9.99593191104852e-06,
+      "loss": 2.4438,
+      "mean_token_accuracy": 0.4822356994263828,
+      "num_tokens": 251597411.0,
+      "step": 1734
+    },
+    {
+      "entropy": 2.44091796875,
+      "epoch": 0.02987181804877629,
+      "grad_norm": 0.5275185704231262,
+      "learning_rate": 9.995920805154136e-06,
+      "loss": 2.3614,
+      "mean_token_accuracy": 0.4821869912557304,
+      "num_tokens": 251751924.0,
+      "step": 1735
+    },
+    {
+      "entropy": 2.424560546875,
+      "epoch": 0.029889035234971547,
+      "grad_norm": 0.5398510694503784,
+      "learning_rate": 9.995909684127036e-06,
+      "loss": 2.3199,
+      "mean_token_accuracy": 0.4816825478337705,
+      "num_tokens": 251897737.0,
+      "step": 1736
+    },
+    {
+      "entropy": 2.421142578125,
+      "epoch": 0.029906252421166808,
+      "grad_norm": 0.5590670704841614,
+      "learning_rate": 9.995898547967252e-06,
+      "loss": 2.3626,
+      "mean_token_accuracy": 0.49043199233710766,
+      "num_tokens": 252047294.0,
+      "step": 1737
+    },
+    {
+      "entropy": 2.443359375,
+      "epoch": 0.029923469607362068,
+      "grad_norm": 0.5771318078041077,
+      "learning_rate": 9.995887396674816e-06,
+      "loss": 2.4432,
+      "mean_token_accuracy": 0.4781417637132108,
+      "num_tokens": 252192301.0,
+      "step": 1738
+    },
+    {
+      "entropy": 2.5240478515625,
+      "epoch": 0.02994068679355733,
+      "grad_norm": 0.5523571968078613,
+      "learning_rate": 9.995876230249765e-06,
+      "loss": 2.4913,
+      "mean_token_accuracy": 0.4711290653795004,
+      "num_tokens": 252343478.0,
+      "step": 1739
+    },
+    {
+      "entropy": 2.466796875,
+      "epoch": 0.02995790397975259,
+      "grad_norm": 0.5873834490776062,
+      "learning_rate": 9.99586504869213e-06,
+      "loss": 2.4351,
+      "mean_token_accuracy": 0.4716539951041341,
+      "num_tokens": 252494029.0,
+      "step": 1740
+    },
+    {
+      "entropy": 2.42822265625,
+      "epoch": 0.02997512116594785,
+      "grad_norm": 0.6091635227203369,
+      "learning_rate": 9.99585385200195e-06,
+      "loss": 2.3937,
+      "mean_token_accuracy": 0.4848588118329644,
+      "num_tokens": 252644102.0,
+      "step": 1741
+    },
+    {
+      "entropy": 2.36083984375,
+      "epoch": 0.02999233835214311,
+      "grad_norm": 0.6105889678001404,
+      "learning_rate": 9.995842640179251e-06,
+      "loss": 2.3327,
+      "mean_token_accuracy": 0.4923085803166032,
+      "num_tokens": 252785353.0,
+      "step": 1742
+    },
+    {
+      "entropy": 2.4542236328125,
+      "epoch": 0.03000955553833837,
+      "grad_norm": 0.5387972593307495,
+      "learning_rate": 9.995831413224073e-06,
+      "loss": 2.3902,
+      "mean_token_accuracy": 0.48074496537446976,
+      "num_tokens": 252931252.0,
+      "step": 1743
+    },
+    {
+      "entropy": 2.4627685546875,
+      "epoch": 0.03002677272453363,
+      "grad_norm": 0.588346540927887,
+      "learning_rate": 9.995820171136447e-06,
+      "loss": 2.4444,
+      "mean_token_accuracy": 0.4807250229641795,
+      "num_tokens": 253081891.0,
+      "step": 1744
+    },
+    {
+      "entropy": 2.404052734375,
+      "epoch": 0.030043989910728888,
+      "grad_norm": 0.6005964279174805,
+      "learning_rate": 9.995808913916409e-06,
+      "loss": 2.3296,
+      "mean_token_accuracy": 0.4894480253569782,
+      "num_tokens": 253227000.0,
+      "step": 1745
+    },
+    {
+      "entropy": 2.4176025390625,
+      "epoch": 0.03006120709692415,
+      "grad_norm": 0.5213177800178528,
+      "learning_rate": 9.995797641563993e-06,
+      "loss": 2.3511,
+      "mean_token_accuracy": 0.4866582774557173,
+      "num_tokens": 253382573.0,
+      "step": 1746
+    },
+    {
+      "entropy": 2.4058837890625,
+      "epoch": 0.03007842428311941,
+      "grad_norm": 0.6120131015777588,
+      "learning_rate": 9.995786354079232e-06,
+      "loss": 2.3611,
+      "mean_token_accuracy": 0.48763601342216134,
+      "num_tokens": 253516221.0,
+      "step": 1747
+    },
+    {
+      "entropy": 2.4691162109375,
+      "epoch": 0.03009564146931467,
+      "grad_norm": 0.5826076865196228,
+      "learning_rate": 9.99577505146216e-06,
+      "loss": 2.4718,
+      "mean_token_accuracy": 0.4789294474758208,
+      "num_tokens": 253656220.0,
+      "step": 1748
+    },
+    {
+      "entropy": 2.5205078125,
+      "epoch": 0.03011285865550993,
+      "grad_norm": 0.6555241346359253,
+      "learning_rate": 9.995763733712813e-06,
+      "loss": 2.4926,
+      "mean_token_accuracy": 0.47109587537124753,
+      "num_tokens": 253792424.0,
+      "step": 1749
+    },
+    {
+      "entropy": 2.447998046875,
+      "epoch": 0.03013007584170519,
+      "grad_norm": 1.2172026634216309,
+      "learning_rate": 9.995752400831224e-06,
+      "loss": 2.3486,
+      "mean_token_accuracy": 0.4864461640827358,
+      "num_tokens": 253936774.0,
+      "step": 1750
+    },
+    {
+      "entropy": 2.4464111328125,
+      "epoch": 0.03014729302790045,
+      "grad_norm": 0.5619203448295593,
+      "learning_rate": 9.995741052817426e-06,
+      "loss": 2.4002,
+      "mean_token_accuracy": 0.4801822490990162,
+      "num_tokens": 254079519.0,
+      "step": 1751
+    },
+    {
+      "entropy": 2.40087890625,
+      "epoch": 0.030164510214095712,
+      "grad_norm": 0.6061223745346069,
+      "learning_rate": 9.995729689671457e-06,
+      "loss": 2.3576,
+      "mean_token_accuracy": 0.49237477174028754,
+      "num_tokens": 254218671.0,
+      "step": 1752
+    },
+    {
+      "entropy": 2.4599609375,
+      "epoch": 0.03018172740029097,
+      "grad_norm": 0.5428354740142822,
+      "learning_rate": 9.995718311393348e-06,
+      "loss": 2.4407,
+      "mean_token_accuracy": 0.47417072020471096,
+      "num_tokens": 254366027.0,
+      "step": 1753
+    },
+    {
+      "entropy": 2.444580078125,
+      "epoch": 0.03019894458648623,
+      "grad_norm": 0.6137836575508118,
+      "learning_rate": 9.995706917983135e-06,
+      "loss": 2.3982,
+      "mean_token_accuracy": 0.4815092282369733,
+      "num_tokens": 254500348.0,
+      "step": 1754
+    },
+    {
+      "entropy": 2.4688720703125,
+      "epoch": 0.03021616177268149,
+      "grad_norm": 0.5226655602455139,
+      "learning_rate": 9.99569550944085e-06,
+      "loss": 2.4489,
+      "mean_token_accuracy": 0.47875166358426213,
+      "num_tokens": 254665949.0,
+      "step": 1755
+    },
+    {
+      "entropy": 2.4661865234375,
+      "epoch": 0.03023337895887675,
+      "grad_norm": 0.5874465107917786,
+      "learning_rate": 9.995684085766532e-06,
+      "loss": 2.4685,
+      "mean_token_accuracy": 0.47104707127436996,
+      "num_tokens": 254821526.0,
+      "step": 1756
+    },
+    {
+      "entropy": 2.3651123046875,
+      "epoch": 0.03025059614507201,
+      "grad_norm": 0.5636343955993652,
+      "learning_rate": 9.995672646960214e-06,
+      "loss": 2.3245,
+      "mean_token_accuracy": 0.49339101929217577,
+      "num_tokens": 254965888.0,
+      "step": 1757
+    },
+    {
+      "entropy": 2.449951171875,
+      "epoch": 0.03026781333126727,
+      "grad_norm": 0.5968552231788635,
+      "learning_rate": 9.995661193021929e-06,
+      "loss": 2.4269,
+      "mean_token_accuracy": 0.47896120324730873,
+      "num_tokens": 255097221.0,
+      "step": 1758
+    },
+    {
+      "entropy": 2.50341796875,
+      "epoch": 0.030285030517462532,
+      "grad_norm": 0.6104995012283325,
+      "learning_rate": 9.995649723951713e-06,
+      "loss": 2.4343,
+      "mean_token_accuracy": 0.48089546570554376,
+      "num_tokens": 255240193.0,
+      "step": 1759
+    },
+    {
+      "entropy": 2.516357421875,
+      "epoch": 0.030302247703657793,
+      "grad_norm": 0.6157956719398499,
+      "learning_rate": 9.995638239749601e-06,
+      "loss": 2.4065,
+      "mean_token_accuracy": 0.47861207462847233,
+      "num_tokens": 255377884.0,
+      "step": 1760
+    },
+    {
+      "entropy": 2.451416015625,
+      "epoch": 0.03031946488985305,
+      "grad_norm": 0.5781038403511047,
+      "learning_rate": 9.995626740415627e-06,
+      "loss": 2.3953,
+      "mean_token_accuracy": 0.4827661495655775,
+      "num_tokens": 255545815.0,
+      "step": 1761
+    },
+    {
+      "entropy": 2.41162109375,
+      "epoch": 0.03033668207604831,
+      "grad_norm": 0.5962122678756714,
+      "learning_rate": 9.995615225949826e-06,
+      "loss": 2.3384,
+      "mean_token_accuracy": 0.48970350436866283,
+      "num_tokens": 255687102.0,
+      "step": 1762
+    },
+    {
+      "entropy": 2.56103515625,
+      "epoch": 0.03035389926224357,
+      "grad_norm": 0.556058943271637,
+      "learning_rate": 9.995603696352231e-06,
+      "loss": 2.5805,
+      "mean_token_accuracy": 0.46257754508405924,
+      "num_tokens": 255832845.0,
+      "step": 1763
+    },
+    {
+      "entropy": 2.5029296875,
+      "epoch": 0.03037111644843883,
+      "grad_norm": 0.5889034867286682,
+      "learning_rate": 9.995592151622881e-06,
+      "loss": 2.4863,
+      "mean_token_accuracy": 0.4708904637955129,
+      "num_tokens": 255966812.0,
+      "step": 1764
+    },
+    {
+      "entropy": 2.4873046875,
+      "epoch": 0.030388333634634092,
+      "grad_norm": 0.564799964427948,
+      "learning_rate": 9.99558059176181e-06,
+      "loss": 2.4345,
+      "mean_token_accuracy": 0.47310470743104815,
+      "num_tokens": 256115280.0,
+      "step": 1765
+    },
+    {
+      "entropy": 2.46533203125,
+      "epoch": 0.030405550820829352,
+      "grad_norm": 0.5584237575531006,
+      "learning_rate": 9.995569016769052e-06,
+      "loss": 2.4093,
+      "mean_token_accuracy": 0.48134846799075603,
+      "num_tokens": 256262835.0,
+      "step": 1766
+    },
+    {
+      "entropy": 2.4100341796875,
+      "epoch": 0.030422768007024613,
+      "grad_norm": 0.6054108142852783,
+      "learning_rate": 9.99555742664464e-06,
+      "loss": 2.3842,
+      "mean_token_accuracy": 0.4840676193125546,
+      "num_tokens": 256412903.0,
+      "step": 1767
+    },
+    {
+      "entropy": 2.4122314453125,
+      "epoch": 0.030439985193219873,
+      "grad_norm": 0.5872666835784912,
+      "learning_rate": 9.99554582138861e-06,
+      "loss": 2.4073,
+      "mean_token_accuracy": 0.4819220928475261,
+      "num_tokens": 256556383.0,
+      "step": 1768
+    },
+    {
+      "entropy": 2.4541015625,
+      "epoch": 0.030457202379415134,
+      "grad_norm": 0.6415526270866394,
+      "learning_rate": 9.995534201001e-06,
+      "loss": 2.4326,
+      "mean_token_accuracy": 0.48358411993831396,
+      "num_tokens": 256701332.0,
+      "step": 1769
+    },
+    {
+      "entropy": 2.5523681640625,
+      "epoch": 0.03047441956561039,
+      "grad_norm": 0.5595957040786743,
+      "learning_rate": 9.995522565481843e-06,
+      "loss": 2.5369,
+      "mean_token_accuracy": 0.46628848975524306,
+      "num_tokens": 256846217.0,
+      "step": 1770
+    },
+    {
+      "entropy": 2.4697265625,
+      "epoch": 0.03049163675180565,
+      "grad_norm": 0.5512428879737854,
+      "learning_rate": 9.995510914831175e-06,
+      "loss": 2.4261,
+      "mean_token_accuracy": 0.4768143934197724,
+      "num_tokens": 256989853.0,
+      "step": 1771
+    },
+    {
+      "entropy": 2.4036865234375,
+      "epoch": 0.030508853938000912,
+      "grad_norm": 0.5823377370834351,
+      "learning_rate": 9.995499249049032e-06,
+      "loss": 2.3555,
+      "mean_token_accuracy": 0.4956060196273029,
+      "num_tokens": 257135041.0,
+      "step": 1772
+    },
+    {
+      "entropy": 2.4488525390625,
+      "epoch": 0.030526071124196173,
+      "grad_norm": 0.555253803730011,
+      "learning_rate": 9.995487568135447e-06,
+      "loss": 2.4463,
+      "mean_token_accuracy": 0.4722318626008928,
+      "num_tokens": 257277086.0,
+      "step": 1773
+    },
+    {
+      "entropy": 2.419921875,
+      "epoch": 0.030543288310391433,
+      "grad_norm": 0.6303548216819763,
+      "learning_rate": 9.995475872090456e-06,
+      "loss": 2.3485,
+      "mean_token_accuracy": 0.48873788258060813,
+      "num_tokens": 257419673.0,
+      "step": 1774
+    },
+    {
+      "entropy": 2.552490234375,
+      "epoch": 0.030560505496586694,
+      "grad_norm": 0.5881249904632568,
+      "learning_rate": 9.995464160914096e-06,
+      "loss": 2.5034,
+      "mean_token_accuracy": 0.4676412958651781,
+      "num_tokens": 257552483.0,
+      "step": 1775
+    },
+    {
+      "entropy": 2.4627685546875,
+      "epoch": 0.030577722682781954,
+      "grad_norm": 0.5897058248519897,
+      "learning_rate": 9.995452434606401e-06,
+      "loss": 2.3792,
+      "mean_token_accuracy": 0.4818587265908718,
+      "num_tokens": 257696825.0,
+      "step": 1776
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.030594939868977215,
+      "grad_norm": 0.5501959919929504,
+      "learning_rate": 9.995440693167407e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.4862223519012332,
+      "num_tokens": 257852558.0,
+      "step": 1777
+    },
+    {
+      "entropy": 2.48681640625,
+      "epoch": 0.03061215705517247,
+      "grad_norm": 0.5968089699745178,
+      "learning_rate": 9.99542893659715e-06,
+      "loss": 2.4541,
+      "mean_token_accuracy": 0.4725658977404237,
+      "num_tokens": 257972949.0,
+      "step": 1778
+    },
+    {
+      "entropy": 2.490966796875,
+      "epoch": 0.030629374241367732,
+      "grad_norm": 0.5689330101013184,
+      "learning_rate": 9.995417164895665e-06,
+      "loss": 2.4556,
+      "mean_token_accuracy": 0.4732924662530422,
+      "num_tokens": 258118477.0,
+      "step": 1779
+    },
+    {
+      "entropy": 2.4151611328125,
+      "epoch": 0.030646591427562993,
+      "grad_norm": 0.6031442880630493,
+      "learning_rate": 9.99540537806299e-06,
+      "loss": 2.4075,
+      "mean_token_accuracy": 0.47733619809150696,
+      "num_tokens": 258262583.0,
+      "step": 1780
+    },
+    {
+      "entropy": 2.418701171875,
+      "epoch": 0.030663808613758253,
+      "grad_norm": 0.5307676792144775,
+      "learning_rate": 9.995393576099156e-06,
+      "loss": 2.4052,
+      "mean_token_accuracy": 0.48428564984351397,
+      "num_tokens": 258415703.0,
+      "step": 1781
+    },
+    {
+      "entropy": 2.466552734375,
+      "epoch": 0.030681025799953514,
+      "grad_norm": 0.613979697227478,
+      "learning_rate": 9.995381759004202e-06,
+      "loss": 2.4284,
+      "mean_token_accuracy": 0.48056693421676755,
+      "num_tokens": 258546742.0,
+      "step": 1782
+    },
+    {
+      "entropy": 2.5028076171875,
+      "epoch": 0.030698242986148774,
+      "grad_norm": 0.6007286906242371,
+      "learning_rate": 9.995369926778163e-06,
+      "loss": 2.4515,
+      "mean_token_accuracy": 0.4713759426958859,
+      "num_tokens": 258667720.0,
+      "step": 1783
+    },
+    {
+      "entropy": 2.459228515625,
+      "epoch": 0.030715460172344035,
+      "grad_norm": 0.5262390375137329,
+      "learning_rate": 9.995358079421076e-06,
+      "loss": 2.4149,
+      "mean_token_accuracy": 0.476927958894521,
+      "num_tokens": 258818608.0,
+      "step": 1784
+    },
+    {
+      "entropy": 2.476318359375,
+      "epoch": 0.030732677358539295,
+      "grad_norm": 0.6184121370315552,
+      "learning_rate": 9.995346216932974e-06,
+      "loss": 2.4888,
+      "mean_token_accuracy": 0.47189527517184615,
+      "num_tokens": 258947687.0,
+      "step": 1785
+    },
+    {
+      "entropy": 2.421630859375,
+      "epoch": 0.030749894544734552,
+      "grad_norm": 0.641552209854126,
+      "learning_rate": 9.995334339313898e-06,
+      "loss": 2.3801,
+      "mean_token_accuracy": 0.48330466262996197,
+      "num_tokens": 259089154.0,
+      "step": 1786
+    },
+    {
+      "entropy": 2.3883056640625,
+      "epoch": 0.030767111730929813,
+      "grad_norm": 0.573010265827179,
+      "learning_rate": 9.995322446563878e-06,
+      "loss": 2.3846,
+      "mean_token_accuracy": 0.48760603182017803,
+      "num_tokens": 259231925.0,
+      "step": 1787
+    },
+    {
+      "entropy": 2.4847412109375,
+      "epoch": 0.030784328917125073,
+      "grad_norm": 0.5725507736206055,
+      "learning_rate": 9.995310538682954e-06,
+      "loss": 2.4799,
+      "mean_token_accuracy": 0.4728474370203912,
+      "num_tokens": 259374698.0,
+      "step": 1788
+    },
+    {
+      "entropy": 2.4398193359375,
+      "epoch": 0.030801546103320334,
+      "grad_norm": 0.681101381778717,
+      "learning_rate": 9.995298615671161e-06,
+      "loss": 2.4197,
+      "mean_token_accuracy": 0.4821738447062671,
+      "num_tokens": 259524367.0,
+      "step": 1789
+    },
+    {
+      "entropy": 2.400146484375,
+      "epoch": 0.030818763289515595,
+      "grad_norm": 0.5721313953399658,
+      "learning_rate": 9.995286677528533e-06,
+      "loss": 2.363,
+      "mean_token_accuracy": 0.48788571590557694,
+      "num_tokens": 259678466.0,
+      "step": 1790
+    },
+    {
+      "entropy": 2.4361572265625,
+      "epoch": 0.030835980475710855,
+      "grad_norm": 0.6100559234619141,
+      "learning_rate": 9.995274724255111e-06,
+      "loss": 2.4359,
+      "mean_token_accuracy": 0.48224303452298045,
+      "num_tokens": 259822604.0,
+      "step": 1791
+    },
+    {
+      "entropy": 2.4278564453125,
+      "epoch": 0.030853197661906116,
+      "grad_norm": 0.5751356482505798,
+      "learning_rate": 9.995262755850926e-06,
+      "loss": 2.4072,
+      "mean_token_accuracy": 0.48218475840985775,
+      "num_tokens": 259965240.0,
+      "step": 1792
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.030870414848101376,
+      "grad_norm": 0.6143491864204407,
+      "learning_rate": 9.995250772316019e-06,
+      "loss": 2.4289,
+      "mean_token_accuracy": 0.4808902507647872,
+      "num_tokens": 260105881.0,
+      "step": 1793
+    },
+    {
+      "entropy": 2.53466796875,
+      "epoch": 0.030887632034296633,
+      "grad_norm": 0.6435692310333252,
+      "learning_rate": 9.995238773650422e-06,
+      "loss": 2.5116,
+      "mean_token_accuracy": 0.46682596672326326,
+      "num_tokens": 260232986.0,
+      "step": 1794
+    },
+    {
+      "entropy": 2.459716796875,
+      "epoch": 0.030904849220491894,
+      "grad_norm": 0.5735113620758057,
+      "learning_rate": 9.995226759854173e-06,
+      "loss": 2.4054,
+      "mean_token_accuracy": 0.4811546648852527,
+      "num_tokens": 260366767.0,
+      "step": 1795
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.030922066406687154,
+      "grad_norm": 1.1088193655014038,
+      "learning_rate": 9.995214730927309e-06,
+      "loss": 2.4319,
+      "mean_token_accuracy": 0.47814410272985697,
+      "num_tokens": 260518664.0,
+      "step": 1796
+    },
+    {
+      "entropy": 2.4351806640625,
+      "epoch": 0.030939283592882415,
+      "grad_norm": 0.6060841083526611,
+      "learning_rate": 9.995202686869867e-06,
+      "loss": 2.4261,
+      "mean_token_accuracy": 0.48954142862930894,
+      "num_tokens": 260665493.0,
+      "step": 1797
+    },
+    {
+      "entropy": 2.4580078125,
+      "epoch": 0.030956500779077675,
+      "grad_norm": 0.5547589659690857,
+      "learning_rate": 9.995190627681883e-06,
+      "loss": 2.4379,
+      "mean_token_accuracy": 0.476859278511256,
+      "num_tokens": 260813391.0,
+      "step": 1798
+    },
+    {
+      "entropy": 2.43603515625,
+      "epoch": 0.030973717965272936,
+      "grad_norm": 0.6191805601119995,
+      "learning_rate": 9.995178553363392e-06,
+      "loss": 2.4344,
+      "mean_token_accuracy": 0.47961972700431943,
+      "num_tokens": 260955219.0,
+      "step": 1799
+    },
+    {
+      "entropy": 2.4512939453125,
+      "epoch": 0.030990935151468196,
+      "grad_norm": 0.5580746531486511,
+      "learning_rate": 9.995166463914432e-06,
+      "loss": 2.3881,
+      "mean_token_accuracy": 0.4819506905041635,
+      "num_tokens": 261107398.0,
+      "step": 1800
+    },
+    {
+      "entropy": 2.4354248046875,
+      "epoch": 0.031008152337663457,
+      "grad_norm": 0.5572928190231323,
+      "learning_rate": 9.99515435933504e-06,
+      "loss": 2.4291,
+      "mean_token_accuracy": 0.47748745791614056,
+      "num_tokens": 261258885.0,
+      "step": 1801
+    },
+    {
+      "entropy": 2.4228515625,
+      "epoch": 0.031025369523858717,
+      "grad_norm": 0.5744470953941345,
+      "learning_rate": 9.99514223962525e-06,
+      "loss": 2.4143,
+      "mean_token_accuracy": 0.4853782169520855,
+      "num_tokens": 261408577.0,
+      "step": 1802
+    },
+    {
+      "entropy": 2.5040283203125,
+      "epoch": 0.031042586710053974,
+      "grad_norm": 0.5782204866409302,
+      "learning_rate": 9.995130104785103e-06,
+      "loss": 2.4765,
+      "mean_token_accuracy": 0.47027752734720707,
+      "num_tokens": 261547340.0,
+      "step": 1803
+    },
+    {
+      "entropy": 2.4500732421875,
+      "epoch": 0.031059803896249235,
+      "grad_norm": 0.6056774258613586,
+      "learning_rate": 9.995117954814632e-06,
+      "loss": 2.4059,
+      "mean_token_accuracy": 0.47932664630934596,
+      "num_tokens": 261680759.0,
+      "step": 1804
+    },
+    {
+      "entropy": 2.4530029296875,
+      "epoch": 0.031077021082444495,
+      "grad_norm": 0.5747044682502747,
+      "learning_rate": 9.995105789713874e-06,
+      "loss": 2.407,
+      "mean_token_accuracy": 0.4791347337886691,
+      "num_tokens": 261839318.0,
+      "step": 1805
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.031094238268639756,
+      "grad_norm": 0.5714244246482849,
+      "learning_rate": 9.995093609482868e-06,
+      "loss": 2.3137,
+      "mean_token_accuracy": 0.4866795940324664,
+      "num_tokens": 261983899.0,
+      "step": 1806
+    },
+    {
+      "entropy": 2.45361328125,
+      "epoch": 0.031111455454835017,
+      "grad_norm": 0.6132124066352844,
+      "learning_rate": 9.995081414121652e-06,
+      "loss": 2.3875,
+      "mean_token_accuracy": 0.4830195610411465,
+      "num_tokens": 262123814.0,
+      "step": 1807
+    },
+    {
+      "entropy": 2.4935302734375,
+      "epoch": 0.031128672641030277,
+      "grad_norm": 0.5878351330757141,
+      "learning_rate": 9.995069203630258e-06,
+      "loss": 2.4513,
+      "mean_token_accuracy": 0.4727471759542823,
+      "num_tokens": 262264324.0,
+      "step": 1808
+    },
+    {
+      "entropy": 2.50146484375,
+      "epoch": 0.031145889827225538,
+      "grad_norm": 0.8043968081474304,
+      "learning_rate": 9.995056978008728e-06,
+      "loss": 2.4545,
+      "mean_token_accuracy": 0.47300353879109025,
+      "num_tokens": 262404423.0,
+      "step": 1809
+    },
+    {
+      "entropy": 2.419677734375,
+      "epoch": 0.031163107013420798,
+      "grad_norm": 0.5692616105079651,
+      "learning_rate": 9.995044737257097e-06,
+      "loss": 2.3788,
+      "mean_token_accuracy": 0.4819628787226975,
+      "num_tokens": 262554058.0,
+      "step": 1810
+    },
+    {
+      "entropy": 2.4549560546875,
+      "epoch": 0.031180324199616055,
+      "grad_norm": 0.6038571000099182,
+      "learning_rate": 9.995032481375401e-06,
+      "loss": 2.4808,
+      "mean_token_accuracy": 0.4799831509590149,
+      "num_tokens": 262687962.0,
+      "step": 1811
+    },
+    {
+      "entropy": 2.389404296875,
+      "epoch": 0.031197541385811316,
+      "grad_norm": 0.6061134338378906,
+      "learning_rate": 9.995020210363678e-06,
+      "loss": 2.3422,
+      "mean_token_accuracy": 0.49405871424824,
+      "num_tokens": 262821281.0,
+      "step": 1812
+    },
+    {
+      "entropy": 2.488037109375,
+      "epoch": 0.031214758572006576,
+      "grad_norm": 0.5849748849868774,
+      "learning_rate": 9.995007924221966e-06,
+      "loss": 2.4825,
+      "mean_token_accuracy": 0.4694899981841445,
+      "num_tokens": 262963950.0,
+      "step": 1813
+    },
+    {
+      "entropy": 2.513671875,
+      "epoch": 0.031231975758201837,
+      "grad_norm": 0.5811004638671875,
+      "learning_rate": 9.994995622950301e-06,
+      "loss": 2.4641,
+      "mean_token_accuracy": 0.4680064986459911,
+      "num_tokens": 263089602.0,
+      "step": 1814
+    },
+    {
+      "entropy": 2.4034423828125,
+      "epoch": 0.031249192944397097,
+      "grad_norm": 0.5289223790168762,
+      "learning_rate": 9.994983306548722e-06,
+      "loss": 2.3548,
+      "mean_token_accuracy": 0.48990146769210696,
+      "num_tokens": 263262419.0,
+      "step": 1815
+    },
+    {
+      "entropy": 2.536865234375,
+      "epoch": 0.03126641013059236,
+      "grad_norm": 0.5391739010810852,
+      "learning_rate": 9.994970975017264e-06,
+      "loss": 2.5102,
+      "mean_token_accuracy": 0.46705460641533136,
+      "num_tokens": 263415410.0,
+      "step": 1816
+    },
+    {
+      "entropy": 2.4627685546875,
+      "epoch": 0.03128362731678762,
+      "grad_norm": 0.5562330484390259,
+      "learning_rate": 9.994958628355964e-06,
+      "loss": 2.4007,
+      "mean_token_accuracy": 0.4822265561670065,
+      "num_tokens": 263560708.0,
+      "step": 1817
+    },
+    {
+      "entropy": 2.5174560546875,
+      "epoch": 0.03130084450298288,
+      "grad_norm": 0.589061975479126,
+      "learning_rate": 9.994946266564862e-06,
+      "loss": 2.4517,
+      "mean_token_accuracy": 0.4671993814408779,
+      "num_tokens": 263706338.0,
+      "step": 1818
+    },
+    {
+      "entropy": 2.431396484375,
+      "epoch": 0.03131806168917814,
+      "grad_norm": 0.5534610152244568,
+      "learning_rate": 9.994933889643995e-06,
+      "loss": 2.4091,
+      "mean_token_accuracy": 0.4794508134946227,
+      "num_tokens": 263846694.0,
+      "step": 1819
+    },
+    {
+      "entropy": 2.5390625,
+      "epoch": 0.0313352788753734,
+      "grad_norm": 0.691927969455719,
+      "learning_rate": 9.9949214975934e-06,
+      "loss": 2.5067,
+      "mean_token_accuracy": 0.46535237692296505,
+      "num_tokens": 263993842.0,
+      "step": 1820
+    },
+    {
+      "entropy": 2.4317626953125,
+      "epoch": 0.03135249606156866,
+      "grad_norm": 0.5267771482467651,
+      "learning_rate": 9.994909090413113e-06,
+      "loss": 2.3885,
+      "mean_token_accuracy": 0.48383715003728867,
+      "num_tokens": 264149235.0,
+      "step": 1821
+    },
+    {
+      "entropy": 2.4241943359375,
+      "epoch": 0.03136971324776392,
+      "grad_norm": 0.5582144856452942,
+      "learning_rate": 9.994896668103174e-06,
+      "loss": 2.4032,
+      "mean_token_accuracy": 0.48607371421530843,
+      "num_tokens": 264305264.0,
+      "step": 1822
+    },
+    {
+      "entropy": 2.4718017578125,
+      "epoch": 0.03138693043395918,
+      "grad_norm": 0.5681629180908203,
+      "learning_rate": 9.994884230663619e-06,
+      "loss": 2.4183,
+      "mean_token_accuracy": 0.47495387122035027,
+      "num_tokens": 264445168.0,
+      "step": 1823
+    },
+    {
+      "entropy": 2.498779296875,
+      "epoch": 0.031404147620154435,
+      "grad_norm": 0.5814370512962341,
+      "learning_rate": 9.994871778094486e-06,
+      "loss": 2.4998,
+      "mean_token_accuracy": 0.475694193970412,
+      "num_tokens": 264584998.0,
+      "step": 1824
+    },
+    {
+      "entropy": 2.4818115234375,
+      "epoch": 0.031421364806349696,
+      "grad_norm": 0.5599236488342285,
+      "learning_rate": 9.994859310395812e-06,
+      "loss": 2.4159,
+      "mean_token_accuracy": 0.47769313864409924,
+      "num_tokens": 264735369.0,
+      "step": 1825
+    },
+    {
+      "entropy": 2.4952392578125,
+      "epoch": 0.031438581992544956,
+      "grad_norm": 0.5620182156562805,
+      "learning_rate": 9.994846827567638e-06,
+      "loss": 2.481,
+      "mean_token_accuracy": 0.4710625964216888,
+      "num_tokens": 264876244.0,
+      "step": 1826
+    },
+    {
+      "entropy": 2.4227294921875,
+      "epoch": 0.03145579917874022,
+      "grad_norm": 0.5782105326652527,
+      "learning_rate": 9.994834329609997e-06,
+      "loss": 2.3754,
+      "mean_token_accuracy": 0.4855298362672329,
+      "num_tokens": 265015068.0,
+      "step": 1827
+    },
+    {
+      "entropy": 2.49755859375,
+      "epoch": 0.03147301636493548,
+      "grad_norm": 0.578761637210846,
+      "learning_rate": 9.994821816522931e-06,
+      "loss": 2.4221,
+      "mean_token_accuracy": 0.47399685718119144,
+      "num_tokens": 265142077.0,
+      "step": 1828
+    },
+    {
+      "entropy": 2.4862060546875,
+      "epoch": 0.03149023355113074,
+      "grad_norm": 0.5534428954124451,
+      "learning_rate": 9.994809288306475e-06,
+      "loss": 2.4853,
+      "mean_token_accuracy": 0.47513916390016675,
+      "num_tokens": 265286452.0,
+      "step": 1829
+    },
+    {
+      "entropy": 2.52734375,
+      "epoch": 0.031507450737326,
+      "grad_norm": 0.5408220887184143,
+      "learning_rate": 9.99479674496067e-06,
+      "loss": 2.4523,
+      "mean_token_accuracy": 0.47395798983052373,
+      "num_tokens": 265441215.0,
+      "step": 1830
+    },
+    {
+      "entropy": 2.4600830078125,
+      "epoch": 0.03152466792352126,
+      "grad_norm": 0.5792752504348755,
+      "learning_rate": 9.994784186485551e-06,
+      "loss": 2.4759,
+      "mean_token_accuracy": 0.47542623011395335,
+      "num_tokens": 265586593.0,
+      "step": 1831
+    },
+    {
+      "entropy": 2.4188232421875,
+      "epoch": 0.03154188510971652,
+      "grad_norm": 0.5839810371398926,
+      "learning_rate": 9.994771612881157e-06,
+      "loss": 2.3782,
+      "mean_token_accuracy": 0.48431128449738026,
+      "num_tokens": 265725310.0,
+      "step": 1832
+    },
+    {
+      "entropy": 2.454345703125,
+      "epoch": 0.03155910229591178,
+      "grad_norm": 0.5647464394569397,
+      "learning_rate": 9.994759024147526e-06,
+      "loss": 2.4133,
+      "mean_token_accuracy": 0.47814659122377634,
+      "num_tokens": 265858415.0,
+      "step": 1833
+    },
+    {
+      "entropy": 2.45654296875,
+      "epoch": 0.03157631948210704,
+      "grad_norm": 0.5568596124649048,
+      "learning_rate": 9.994746420284698e-06,
+      "loss": 2.4259,
+      "mean_token_accuracy": 0.4771571112796664,
+      "num_tokens": 266011950.0,
+      "step": 1834
+    },
+    {
+      "entropy": 2.413330078125,
+      "epoch": 0.0315935366683023,
+      "grad_norm": 0.5861586332321167,
+      "learning_rate": 9.994733801292709e-06,
+      "loss": 2.3792,
+      "mean_token_accuracy": 0.4801113181747496,
+      "num_tokens": 266157685.0,
+      "step": 1835
+    },
+    {
+      "entropy": 2.4085693359375,
+      "epoch": 0.03161075385449756,
+      "grad_norm": 0.5990592837333679,
+      "learning_rate": 9.994721167171597e-06,
+      "loss": 2.3922,
+      "mean_token_accuracy": 0.48092158418148756,
+      "num_tokens": 266293266.0,
+      "step": 1836
+    },
+    {
+      "entropy": 2.432373046875,
+      "epoch": 0.03162797104069282,
+      "grad_norm": 0.5652367472648621,
+      "learning_rate": 9.9947085179214e-06,
+      "loss": 2.3796,
+      "mean_token_accuracy": 0.48524676635861397,
+      "num_tokens": 266448458.0,
+      "step": 1837
+    },
+    {
+      "entropy": 2.5150146484375,
+      "epoch": 0.03164518822688808,
+      "grad_norm": 0.581610918045044,
+      "learning_rate": 9.99469585354216e-06,
+      "loss": 2.4421,
+      "mean_token_accuracy": 0.4794493457302451,
+      "num_tokens": 266588403.0,
+      "step": 1838
+    },
+    {
+      "entropy": 2.479248046875,
+      "epoch": 0.03166240541308334,
+      "grad_norm": 0.5524604320526123,
+      "learning_rate": 9.994683174033913e-06,
+      "loss": 2.409,
+      "mean_token_accuracy": 0.4818028402514756,
+      "num_tokens": 266748122.0,
+      "step": 1839
+    },
+    {
+      "entropy": 2.4130859375,
+      "epoch": 0.031679622599278597,
+      "grad_norm": 0.5744349360466003,
+      "learning_rate": 9.994670479396695e-06,
+      "loss": 2.3541,
+      "mean_token_accuracy": 0.4899553945288062,
+      "num_tokens": 266889574.0,
+      "step": 1840
+    },
+    {
+      "entropy": 2.50341796875,
+      "epoch": 0.03169683978547386,
+      "grad_norm": 0.5297801494598389,
+      "learning_rate": 9.994657769630546e-06,
+      "loss": 2.4784,
+      "mean_token_accuracy": 0.4729268723167479,
+      "num_tokens": 267038916.0,
+      "step": 1841
+    },
+    {
+      "entropy": 2.42041015625,
+      "epoch": 0.03171405697166912,
+      "grad_norm": 0.570783257484436,
+      "learning_rate": 9.994645044735507e-06,
+      "loss": 2.377,
+      "mean_token_accuracy": 0.4859691094607115,
+      "num_tokens": 267183760.0,
+      "step": 1842
+    },
+    {
+      "entropy": 2.4842529296875,
+      "epoch": 0.03173127415786438,
+      "grad_norm": 0.5840896368026733,
+      "learning_rate": 9.994632304711616e-06,
+      "loss": 2.4273,
+      "mean_token_accuracy": 0.4811637466773391,
+      "num_tokens": 267324455.0,
+      "step": 1843
+    },
+    {
+      "entropy": 2.4864501953125,
+      "epoch": 0.03174849134405964,
+      "grad_norm": 0.5744589567184448,
+      "learning_rate": 9.994619549558908e-06,
+      "loss": 2.4713,
+      "mean_token_accuracy": 0.4727841066196561,
+      "num_tokens": 267472269.0,
+      "step": 1844
+    },
+    {
+      "entropy": 2.46923828125,
+      "epoch": 0.0317657085302549,
+      "grad_norm": 0.5857048630714417,
+      "learning_rate": 9.994606779277425e-06,
+      "loss": 2.4301,
+      "mean_token_accuracy": 0.48071943665854633,
+      "num_tokens": 267624717.0,
+      "step": 1845
+    },
+    {
+      "entropy": 2.420654296875,
+      "epoch": 0.03178292571645016,
+      "grad_norm": 0.5835974812507629,
+      "learning_rate": 9.994593993867203e-06,
+      "loss": 2.3782,
+      "mean_token_accuracy": 0.4824722218327224,
+      "num_tokens": 267762034.0,
+      "step": 1846
+    },
+    {
+      "entropy": 2.503662109375,
+      "epoch": 0.03180014290264542,
+      "grad_norm": 0.5819142460823059,
+      "learning_rate": 9.994581193328283e-06,
+      "loss": 2.5042,
+      "mean_token_accuracy": 0.47309408662840724,
+      "num_tokens": 267897855.0,
+      "step": 1847
+    },
+    {
+      "entropy": 2.4564208984375,
+      "epoch": 0.03181736008884068,
+      "grad_norm": 0.6381955742835999,
+      "learning_rate": 9.994568377660703e-06,
+      "loss": 2.4308,
+      "mean_token_accuracy": 0.48583648912608624,
+      "num_tokens": 268025165.0,
+      "step": 1848
+    },
+    {
+      "entropy": 2.439453125,
+      "epoch": 0.03183457727503594,
+      "grad_norm": 0.5733355283737183,
+      "learning_rate": 9.994555546864504e-06,
+      "loss": 2.418,
+      "mean_token_accuracy": 0.48078050184994936,
+      "num_tokens": 268172403.0,
+      "step": 1849
+    },
+    {
+      "entropy": 2.3912353515625,
+      "epoch": 0.0318517944612312,
+      "grad_norm": 0.5635650157928467,
+      "learning_rate": 9.99454270093972e-06,
+      "loss": 2.3733,
+      "mean_token_accuracy": 0.4842658434063196,
+      "num_tokens": 268337839.0,
+      "step": 1850
+    },
+    {
+      "entropy": 2.4051513671875,
+      "epoch": 0.03186901164742646,
+      "grad_norm": 0.5738821625709534,
+      "learning_rate": 9.994529839886395e-06,
+      "loss": 2.332,
+      "mean_token_accuracy": 0.49162986455485225,
+      "num_tokens": 268490293.0,
+      "step": 1851
+    },
+    {
+      "entropy": 2.4752197265625,
+      "epoch": 0.03188622883362172,
+      "grad_norm": 0.5878849029541016,
+      "learning_rate": 9.994516963704564e-06,
+      "loss": 2.4238,
+      "mean_token_accuracy": 0.47905471734702587,
+      "num_tokens": 268627952.0,
+      "step": 1852
+    },
+    {
+      "entropy": 2.494384765625,
+      "epoch": 0.03190344601981698,
+      "grad_norm": 0.5611072182655334,
+      "learning_rate": 9.994504072394268e-06,
+      "loss": 2.4512,
+      "mean_token_accuracy": 0.4697739346884191,
+      "num_tokens": 268771756.0,
+      "step": 1853
+    },
+    {
+      "entropy": 2.4912109375,
+      "epoch": 0.031920663206012244,
+      "grad_norm": 0.5786696672439575,
+      "learning_rate": 9.994491165955546e-06,
+      "loss": 2.4426,
+      "mean_token_accuracy": 0.47656662855297327,
+      "num_tokens": 268906129.0,
+      "step": 1854
+    },
+    {
+      "entropy": 2.48193359375,
+      "epoch": 0.031937880392207504,
+      "grad_norm": 0.5850675106048584,
+      "learning_rate": 9.994478244388437e-06,
+      "loss": 2.4469,
+      "mean_token_accuracy": 0.47550681745633483,
+      "num_tokens": 269056386.0,
+      "step": 1855
+    },
+    {
+      "entropy": 2.499755859375,
+      "epoch": 0.031955097578402765,
+      "grad_norm": 0.5520582795143127,
+      "learning_rate": 9.994465307692978e-06,
+      "loss": 2.4417,
+      "mean_token_accuracy": 0.47765777353197336,
+      "num_tokens": 269195599.0,
+      "step": 1856
+    },
+    {
+      "entropy": 2.4632568359375,
+      "epoch": 0.03197231476459802,
+      "grad_norm": 0.5770227313041687,
+      "learning_rate": 9.994452355869211e-06,
+      "loss": 2.4842,
+      "mean_token_accuracy": 0.4763926393352449,
+      "num_tokens": 269336161.0,
+      "step": 1857
+    },
+    {
+      "entropy": 2.4757080078125,
+      "epoch": 0.03198953195079328,
+      "grad_norm": 0.6584348678588867,
+      "learning_rate": 9.994439388917177e-06,
+      "loss": 2.3984,
+      "mean_token_accuracy": 0.4819053406827152,
+      "num_tokens": 269461648.0,
+      "step": 1858
+    },
+    {
+      "entropy": 2.427001953125,
+      "epoch": 0.03200674913698854,
+      "grad_norm": 0.5825558304786682,
+      "learning_rate": 9.99442640683691e-06,
+      "loss": 2.3939,
+      "mean_token_accuracy": 0.4860147815197706,
+      "num_tokens": 269614043.0,
+      "step": 1859
+    },
+    {
+      "entropy": 2.5079345703125,
+      "epoch": 0.0320239663231838,
+      "grad_norm": 0.5787304043769836,
+      "learning_rate": 9.994413409628451e-06,
+      "loss": 2.526,
+      "mean_token_accuracy": 0.4718983927741647,
+      "num_tokens": 269763008.0,
+      "step": 1860
+    },
+    {
+      "entropy": 2.4483642578125,
+      "epoch": 0.03204118350937906,
+      "grad_norm": 0.5473356246948242,
+      "learning_rate": 9.994400397291843e-06,
+      "loss": 2.4192,
+      "mean_token_accuracy": 0.4761442271992564,
+      "num_tokens": 269916961.0,
+      "step": 1861
+    },
+    {
+      "entropy": 2.347412109375,
+      "epoch": 0.03205840069557432,
+      "grad_norm": 0.5818413496017456,
+      "learning_rate": 9.99438736982712e-06,
+      "loss": 2.3176,
+      "mean_token_accuracy": 0.49792582634836435,
+      "num_tokens": 270053323.0,
+      "step": 1862
+    },
+    {
+      "entropy": 2.522216796875,
+      "epoch": 0.03207561788176958,
+      "grad_norm": 0.5837598443031311,
+      "learning_rate": 9.994374327234326e-06,
+      "loss": 2.5254,
+      "mean_token_accuracy": 0.4689773670397699,
+      "num_tokens": 270183004.0,
+      "step": 1863
+    },
+    {
+      "entropy": 2.424560546875,
+      "epoch": 0.03209283506796484,
+      "grad_norm": 0.5816727876663208,
+      "learning_rate": 9.994361269513499e-06,
+      "loss": 2.3983,
+      "mean_token_accuracy": 0.4842468095012009,
+      "num_tokens": 270337766.0,
+      "step": 1864
+    },
+    {
+      "entropy": 2.4515380859375,
+      "epoch": 0.0321100522541601,
+      "grad_norm": 0.5595372319221497,
+      "learning_rate": 9.994348196664679e-06,
+      "loss": 2.4535,
+      "mean_token_accuracy": 0.4758255537599325,
+      "num_tokens": 270501411.0,
+      "step": 1865
+    },
+    {
+      "entropy": 2.4014892578125,
+      "epoch": 0.03212726944035536,
+      "grad_norm": 0.5381937623023987,
+      "learning_rate": 9.994335108687903e-06,
+      "loss": 2.3243,
+      "mean_token_accuracy": 0.4866001163609326,
+      "num_tokens": 270652865.0,
+      "step": 1866
+    },
+    {
+      "entropy": 2.427734375,
+      "epoch": 0.032144486626550624,
+      "grad_norm": 0.5453839302062988,
+      "learning_rate": 9.994322005583213e-06,
+      "loss": 2.3798,
+      "mean_token_accuracy": 0.4782120902091265,
+      "num_tokens": 270798390.0,
+      "step": 1867
+    },
+    {
+      "entropy": 2.5419921875,
+      "epoch": 0.032161703812745884,
+      "grad_norm": 0.6079609394073486,
+      "learning_rate": 9.994308887350647e-06,
+      "loss": 2.5727,
+      "mean_token_accuracy": 0.46313366387039423,
+      "num_tokens": 270943300.0,
+      "step": 1868
+    },
+    {
+      "entropy": 2.486083984375,
+      "epoch": 0.032178920998941145,
+      "grad_norm": 0.597481906414032,
+      "learning_rate": 9.994295753990247e-06,
+      "loss": 2.4698,
+      "mean_token_accuracy": 0.46938649471849203,
+      "num_tokens": 271077719.0,
+      "step": 1869
+    },
+    {
+      "entropy": 2.40771484375,
+      "epoch": 0.032196138185136405,
+      "grad_norm": 1.105528473854065,
+      "learning_rate": 9.99428260550205e-06,
+      "loss": 2.3737,
+      "mean_token_accuracy": 0.4825026970356703,
+      "num_tokens": 271237964.0,
+      "step": 1870
+    },
+    {
+      "entropy": 2.4666748046875,
+      "epoch": 0.032213355371331666,
+      "grad_norm": 0.5856791138648987,
+      "learning_rate": 9.9942694418861e-06,
+      "loss": 2.4256,
+      "mean_token_accuracy": 0.47253927774727345,
+      "num_tokens": 271380338.0,
+      "step": 1871
+    },
+    {
+      "entropy": 2.3770751953125,
+      "epoch": 0.032230572557526926,
+      "grad_norm": 0.5685958862304688,
+      "learning_rate": 9.994256263142433e-06,
+      "loss": 2.3337,
+      "mean_token_accuracy": 0.4951976570300758,
+      "num_tokens": 271518289.0,
+      "step": 1872
+    },
+    {
+      "entropy": 2.5079345703125,
+      "epoch": 0.03224778974372219,
+      "grad_norm": 0.5702466368675232,
+      "learning_rate": 9.99424306927109e-06,
+      "loss": 2.4513,
+      "mean_token_accuracy": 0.4732716833241284,
+      "num_tokens": 271660591.0,
+      "step": 1873
+    },
+    {
+      "entropy": 2.501953125,
+      "epoch": 0.03226500692991744,
+      "grad_norm": 0.5536970496177673,
+      "learning_rate": 9.994229860272114e-06,
+      "loss": 2.4598,
+      "mean_token_accuracy": 0.47078709537163377,
+      "num_tokens": 271815738.0,
+      "step": 1874
+    },
+    {
+      "entropy": 2.433349609375,
+      "epoch": 0.0322822241161127,
+      "grad_norm": 0.5575986504554749,
+      "learning_rate": 9.99421663614554e-06,
+      "loss": 2.3716,
+      "mean_token_accuracy": 0.4846403314732015,
+      "num_tokens": 271964724.0,
+      "step": 1875
+    },
+    {
+      "entropy": 2.439208984375,
+      "epoch": 0.03229944130230796,
+      "grad_norm": 0.594307541847229,
+      "learning_rate": 9.99420339689141e-06,
+      "loss": 2.3674,
+      "mean_token_accuracy": 0.48272377625107765,
+      "num_tokens": 272103280.0,
+      "step": 1876
+    },
+    {
+      "entropy": 2.382080078125,
+      "epoch": 0.03231665848850322,
+      "grad_norm": 0.5934799909591675,
+      "learning_rate": 9.994190142509766e-06,
+      "loss": 2.3689,
+      "mean_token_accuracy": 0.4859785996377468,
+      "num_tokens": 272237744.0,
+      "step": 1877
+    },
+    {
+      "entropy": 2.453369140625,
+      "epoch": 0.03233387567469848,
+      "grad_norm": 0.5809702277183533,
+      "learning_rate": 9.994176873000646e-06,
+      "loss": 2.4275,
+      "mean_token_accuracy": 0.4772698413580656,
+      "num_tokens": 272384323.0,
+      "step": 1878
+    },
+    {
+      "entropy": 2.457275390625,
+      "epoch": 0.03235109286089374,
+      "grad_norm": 0.5598036646842957,
+      "learning_rate": 9.99416358836409e-06,
+      "loss": 2.4166,
+      "mean_token_accuracy": 0.47385495621711016,
+      "num_tokens": 272530813.0,
+      "step": 1879
+    },
+    {
+      "entropy": 2.51318359375,
+      "epoch": 0.032368310047089004,
+      "grad_norm": 0.567221462726593,
+      "learning_rate": 9.994150288600139e-06,
+      "loss": 2.5107,
+      "mean_token_accuracy": 0.4726970442570746,
+      "num_tokens": 272679093.0,
+      "step": 1880
+    },
+    {
+      "entropy": 2.4515380859375,
+      "epoch": 0.032385527233284264,
+      "grad_norm": 0.6015093922615051,
+      "learning_rate": 9.994136973708834e-06,
+      "loss": 2.399,
+      "mean_token_accuracy": 0.47881753370165825,
+      "num_tokens": 272819409.0,
+      "step": 1881
+    },
+    {
+      "entropy": 2.424072265625,
+      "epoch": 0.032402744419479525,
+      "grad_norm": 0.5528237819671631,
+      "learning_rate": 9.994123643690214e-06,
+      "loss": 2.3889,
+      "mean_token_accuracy": 0.4852414163760841,
+      "num_tokens": 272964763.0,
+      "step": 1882
+    },
+    {
+      "entropy": 2.4576416015625,
+      "epoch": 0.032419961605674785,
+      "grad_norm": 0.6007446050643921,
+      "learning_rate": 9.99411029854432e-06,
+      "loss": 2.4264,
+      "mean_token_accuracy": 0.4797611120156944,
+      "num_tokens": 273090453.0,
+      "step": 1883
+    },
+    {
+      "entropy": 2.463623046875,
+      "epoch": 0.032437178791870046,
+      "grad_norm": 0.5631487369537354,
+      "learning_rate": 9.994096938271193e-06,
+      "loss": 2.4681,
+      "mean_token_accuracy": 0.4756158501841128,
+      "num_tokens": 273239046.0,
+      "step": 1884
+    },
+    {
+      "entropy": 2.4407958984375,
+      "epoch": 0.032454395978065306,
+      "grad_norm": 0.5506601929664612,
+      "learning_rate": 9.994083562870873e-06,
+      "loss": 2.3948,
+      "mean_token_accuracy": 0.47984826657921076,
+      "num_tokens": 273396483.0,
+      "step": 1885
+    },
+    {
+      "entropy": 2.4609375,
+      "epoch": 0.03247161316426057,
+      "grad_norm": 0.5861914157867432,
+      "learning_rate": 9.9940701723434e-06,
+      "loss": 2.4491,
+      "mean_token_accuracy": 0.47965921740978956,
+      "num_tokens": 273539136.0,
+      "step": 1886
+    },
+    {
+      "entropy": 2.4434814453125,
+      "epoch": 0.03248883035045583,
+      "grad_norm": 0.5634111166000366,
+      "learning_rate": 9.994056766688815e-06,
+      "loss": 2.4001,
+      "mean_token_accuracy": 0.4822332635521889,
+      "num_tokens": 273683327.0,
+      "step": 1887
+    },
+    {
+      "entropy": 2.4342041015625,
+      "epoch": 0.03250604753665109,
+      "grad_norm": 0.6063517332077026,
+      "learning_rate": 9.994043345907158e-06,
+      "loss": 2.3837,
+      "mean_token_accuracy": 0.49188648257404566,
+      "num_tokens": 273812912.0,
+      "step": 1888
+    },
+    {
+      "entropy": 2.454345703125,
+      "epoch": 0.03252326472284635,
+      "grad_norm": 0.6061086058616638,
+      "learning_rate": 9.99402990999847e-06,
+      "loss": 2.3902,
+      "mean_token_accuracy": 0.4804657520726323,
+      "num_tokens": 273953195.0,
+      "step": 1889
+    },
+    {
+      "entropy": 2.3865966796875,
+      "epoch": 0.0325404819090416,
+      "grad_norm": 0.5631823539733887,
+      "learning_rate": 9.994016458962795e-06,
+      "loss": 2.3116,
+      "mean_token_accuracy": 0.4911833154037595,
+      "num_tokens": 274105327.0,
+      "step": 1890
+    },
+    {
+      "entropy": 2.44384765625,
+      "epoch": 0.03255769909523686,
+      "grad_norm": 0.5700172185897827,
+      "learning_rate": 9.994002992800167e-06,
+      "loss": 2.4395,
+      "mean_token_accuracy": 0.4804955665022135,
+      "num_tokens": 274245805.0,
+      "step": 1891
+    },
+    {
+      "entropy": 2.508056640625,
+      "epoch": 0.03257491628143212,
+      "grad_norm": 0.5785242915153503,
+      "learning_rate": 9.993989511510633e-06,
+      "loss": 2.4574,
+      "mean_token_accuracy": 0.47288175392895937,
+      "num_tokens": 274390327.0,
+      "step": 1892
+    },
+    {
+      "entropy": 2.5166015625,
+      "epoch": 0.032592133467627384,
+      "grad_norm": 0.5731043815612793,
+      "learning_rate": 9.99397601509423e-06,
+      "loss": 2.4638,
+      "mean_token_accuracy": 0.47718239948153496,
+      "num_tokens": 274536899.0,
+      "step": 1893
+    },
+    {
+      "entropy": 2.3447265625,
+      "epoch": 0.032609350653822644,
+      "grad_norm": 0.6108200550079346,
+      "learning_rate": 9.993962503551e-06,
+      "loss": 2.2998,
+      "mean_token_accuracy": 0.504625148139894,
+      "num_tokens": 274690625.0,
+      "step": 1894
+    },
+    {
+      "entropy": 2.4359130859375,
+      "epoch": 0.032626567840017905,
+      "grad_norm": 0.5686972141265869,
+      "learning_rate": 9.993948976880985e-06,
+      "loss": 2.3772,
+      "mean_token_accuracy": 0.4784249165095389,
+      "num_tokens": 274857384.0,
+      "step": 1895
+    },
+    {
+      "entropy": 2.465576171875,
+      "epoch": 0.032643785026213165,
+      "grad_norm": 0.5697457194328308,
+      "learning_rate": 9.993935435084225e-06,
+      "loss": 2.3909,
+      "mean_token_accuracy": 0.483038027305156,
+      "num_tokens": 275007281.0,
+      "step": 1896
+    },
+    {
+      "entropy": 2.4967041015625,
+      "epoch": 0.032661002212408426,
+      "grad_norm": 0.5822637677192688,
+      "learning_rate": 9.99392187816076e-06,
+      "loss": 2.4637,
+      "mean_token_accuracy": 0.46831735828891397,
+      "num_tokens": 275145804.0,
+      "step": 1897
+    },
+    {
+      "entropy": 2.4696044921875,
+      "epoch": 0.032678219398603686,
+      "grad_norm": 0.5560740828514099,
+      "learning_rate": 9.993908306110632e-06,
+      "loss": 2.4337,
+      "mean_token_accuracy": 0.4820185494609177,
+      "num_tokens": 275303743.0,
+      "step": 1898
+    },
+    {
+      "entropy": 2.4893798828125,
+      "epoch": 0.03269543658479895,
+      "grad_norm": 0.5658345222473145,
+      "learning_rate": 9.993894718933882e-06,
+      "loss": 2.4622,
+      "mean_token_accuracy": 0.4734317073598504,
+      "num_tokens": 275455938.0,
+      "step": 1899
+    },
+    {
+      "entropy": 2.367431640625,
+      "epoch": 0.03271265377099421,
+      "grad_norm": 0.5981510281562805,
+      "learning_rate": 9.993881116630556e-06,
+      "loss": 2.2978,
+      "mean_token_accuracy": 0.495200231205672,
+      "num_tokens": 275607789.0,
+      "step": 1900
+    },
+    {
+      "entropy": 2.46435546875,
+      "epoch": 0.03272987095718947,
+      "grad_norm": 0.5610291361808777,
+      "learning_rate": 9.993867499200684e-06,
+      "loss": 2.4479,
+      "mean_token_accuracy": 0.4744836580939591,
+      "num_tokens": 275761693.0,
+      "step": 1901
+    },
+    {
+      "entropy": 2.5068359375,
+      "epoch": 0.03274708814338473,
+      "grad_norm": 0.6446607708930969,
+      "learning_rate": 9.993853866644319e-06,
+      "loss": 2.4878,
+      "mean_token_accuracy": 0.46817948622629046,
+      "num_tokens": 275887793.0,
+      "step": 1902
+    },
+    {
+      "entropy": 2.503662109375,
+      "epoch": 0.03276430532957999,
+      "grad_norm": 0.6273615956306458,
+      "learning_rate": 9.993840218961495e-06,
+      "loss": 2.4879,
+      "mean_token_accuracy": 0.4687243062071502,
+      "num_tokens": 276018500.0,
+      "step": 1903
+    },
+    {
+      "entropy": 2.498779296875,
+      "epoch": 0.03278152251577525,
+      "grad_norm": 0.5124494433403015,
+      "learning_rate": 9.993826556152255e-06,
+      "loss": 2.4705,
+      "mean_token_accuracy": 0.4730762280523777,
+      "num_tokens": 276183672.0,
+      "step": 1904
+    },
+    {
+      "entropy": 2.518798828125,
+      "epoch": 0.03279873970197051,
+      "grad_norm": 0.5735337138175964,
+      "learning_rate": 9.993812878216642e-06,
+      "loss": 2.4894,
+      "mean_token_accuracy": 0.47411146527156234,
+      "num_tokens": 276336067.0,
+      "step": 1905
+    },
+    {
+      "entropy": 2.498291015625,
+      "epoch": 0.03281595688816577,
+      "grad_norm": 0.6396939158439636,
+      "learning_rate": 9.993799185154695e-06,
+      "loss": 2.4308,
+      "mean_token_accuracy": 0.4775810223072767,
+      "num_tokens": 276454633.0,
+      "step": 1906
+    },
+    {
+      "entropy": 2.43798828125,
+      "epoch": 0.032833174074361024,
+      "grad_norm": 0.5700352191925049,
+      "learning_rate": 9.993785476966458e-06,
+      "loss": 2.4091,
+      "mean_token_accuracy": 0.47954965522512794,
+      "num_tokens": 276592071.0,
+      "step": 1907
+    },
+    {
+      "entropy": 2.555419921875,
+      "epoch": 0.032850391260556285,
+      "grad_norm": 0.5614842772483826,
+      "learning_rate": 9.993771753651971e-06,
+      "loss": 2.5613,
+      "mean_token_accuracy": 0.4670336083509028,
+      "num_tokens": 276746077.0,
+      "step": 1908
+    },
+    {
+      "entropy": 2.423095703125,
+      "epoch": 0.032867608446751545,
+      "grad_norm": 0.5809895396232605,
+      "learning_rate": 9.993758015211276e-06,
+      "loss": 2.3669,
+      "mean_token_accuracy": 0.48658125940710306,
+      "num_tokens": 276900397.0,
+      "step": 1909
+    },
+    {
+      "entropy": 2.384765625,
+      "epoch": 0.032884825632946806,
+      "grad_norm": 0.5184022784233093,
+      "learning_rate": 9.993744261644414e-06,
+      "loss": 2.3631,
+      "mean_token_accuracy": 0.4841692647896707,
+      "num_tokens": 277080897.0,
+      "step": 1910
+    },
+    {
+      "entropy": 2.4111328125,
+      "epoch": 0.032902042819142066,
+      "grad_norm": 0.575218915939331,
+      "learning_rate": 9.993730492951428e-06,
+      "loss": 2.3794,
+      "mean_token_accuracy": 0.48858750332146883,
+      "num_tokens": 277229579.0,
+      "step": 1911
+    },
+    {
+      "entropy": 2.4375,
+      "epoch": 0.03291926000533733,
+      "grad_norm": 0.8869200944900513,
+      "learning_rate": 9.993716709132359e-06,
+      "loss": 2.3684,
+      "mean_token_accuracy": 0.48189750453457236,
+      "num_tokens": 277362179.0,
+      "step": 1912
+    },
+    {
+      "entropy": 2.4571533203125,
+      "epoch": 0.03293647719153259,
+      "grad_norm": 0.5700924396514893,
+      "learning_rate": 9.993702910187247e-06,
+      "loss": 2.3851,
+      "mean_token_accuracy": 0.48076847614720464,
+      "num_tokens": 277520460.0,
+      "step": 1913
+    },
+    {
+      "entropy": 2.51025390625,
+      "epoch": 0.03295369437772785,
+      "grad_norm": 0.6675810813903809,
+      "learning_rate": 9.993689096116138e-06,
+      "loss": 2.4804,
+      "mean_token_accuracy": 0.47350738383829594,
+      "num_tokens": 277667131.0,
+      "step": 1914
+    },
+    {
+      "entropy": 2.4278564453125,
+      "epoch": 0.03297091156392311,
+      "grad_norm": 0.6161773204803467,
+      "learning_rate": 9.993675266919068e-06,
+      "loss": 2.3634,
+      "mean_token_accuracy": 0.48864840995520353,
+      "num_tokens": 277798538.0,
+      "step": 1915
+    },
+    {
+      "entropy": 2.42724609375,
+      "epoch": 0.03298812875011837,
+      "grad_norm": 0.6031524538993835,
+      "learning_rate": 9.993661422596083e-06,
+      "loss": 2.3735,
+      "mean_token_accuracy": 0.4872046741656959,
+      "num_tokens": 277938299.0,
+      "step": 1916
+    },
+    {
+      "entropy": 2.459716796875,
+      "epoch": 0.03300534593631363,
+      "grad_norm": 0.5436295866966248,
+      "learning_rate": 9.993647563147226e-06,
+      "loss": 2.4323,
+      "mean_token_accuracy": 0.47822323255240917,
+      "num_tokens": 278080906.0,
+      "step": 1917
+    },
+    {
+      "entropy": 2.482177734375,
+      "epoch": 0.03302256312250889,
+      "grad_norm": 0.5724844336509705,
+      "learning_rate": 9.993633688572536e-06,
+      "loss": 2.4886,
+      "mean_token_accuracy": 0.47133151395246387,
+      "num_tokens": 278220379.0,
+      "step": 1918
+    },
+    {
+      "entropy": 2.4127197265625,
+      "epoch": 0.03303978030870415,
+      "grad_norm": 0.583770751953125,
+      "learning_rate": 9.993619798872056e-06,
+      "loss": 2.3904,
+      "mean_token_accuracy": 0.48571639275178313,
+      "num_tokens": 278372188.0,
+      "step": 1919
+    },
+    {
+      "entropy": 2.45849609375,
+      "epoch": 0.03305699749489941,
+      "grad_norm": 0.5579352378845215,
+      "learning_rate": 9.993605894045828e-06,
+      "loss": 2.4178,
+      "mean_token_accuracy": 0.4752850723452866,
+      "num_tokens": 278515572.0,
+      "step": 1920
+    },
+    {
+      "entropy": 2.46728515625,
+      "epoch": 0.03307421468109467,
+      "grad_norm": 0.5899266600608826,
+      "learning_rate": 9.993591974093896e-06,
+      "loss": 2.4453,
+      "mean_token_accuracy": 0.4791210265830159,
+      "num_tokens": 278659949.0,
+      "step": 1921
+    },
+    {
+      "entropy": 2.431640625,
+      "epoch": 0.03309143186728993,
+      "grad_norm": 0.5982487797737122,
+      "learning_rate": 9.993578039016298e-06,
+      "loss": 2.4021,
+      "mean_token_accuracy": 0.4784347270615399,
+      "num_tokens": 278784019.0,
+      "step": 1922
+    },
+    {
+      "entropy": 2.448486328125,
+      "epoch": 0.03310864905348519,
+      "grad_norm": 0.7279030680656433,
+      "learning_rate": 9.993564088813078e-06,
+      "loss": 2.4092,
+      "mean_token_accuracy": 0.47610331838950515,
+      "num_tokens": 278941353.0,
+      "step": 1923
+    },
+    {
+      "entropy": 2.5162353515625,
+      "epoch": 0.033125866239680446,
+      "grad_norm": 0.5260778665542603,
+      "learning_rate": 9.993550123484282e-06,
+      "loss": 2.4839,
+      "mean_token_accuracy": 0.4682774804532528,
+      "num_tokens": 279101143.0,
+      "step": 1924
+    },
+    {
+      "entropy": 2.448974609375,
+      "epoch": 0.033143083425875706,
+      "grad_norm": 0.5392525792121887,
+      "learning_rate": 9.993536143029946e-06,
+      "loss": 2.3976,
+      "mean_token_accuracy": 0.479318049736321,
+      "num_tokens": 279249944.0,
+      "step": 1925
+    },
+    {
+      "entropy": 2.4193115234375,
+      "epoch": 0.03316030061207097,
+      "grad_norm": 0.5925559401512146,
+      "learning_rate": 9.993522147450118e-06,
+      "loss": 2.3594,
+      "mean_token_accuracy": 0.49047204852104187,
+      "num_tokens": 279387621.0,
+      "step": 1926
+    },
+    {
+      "entropy": 2.51318359375,
+      "epoch": 0.03317751779826623,
+      "grad_norm": 0.6154163479804993,
+      "learning_rate": 9.993508136744838e-06,
+      "loss": 2.4331,
+      "mean_token_accuracy": 0.47704603895545006,
+      "num_tokens": 279525188.0,
+      "step": 1927
+    },
+    {
+      "entropy": 2.469482421875,
+      "epoch": 0.03319473498446149,
+      "grad_norm": 0.5946447253227234,
+      "learning_rate": 9.993494110914149e-06,
+      "loss": 2.4433,
+      "mean_token_accuracy": 0.4768550405278802,
+      "num_tokens": 279675297.0,
+      "step": 1928
+    },
+    {
+      "entropy": 2.4324951171875,
+      "epoch": 0.03321195217065675,
+      "grad_norm": 0.5541896820068359,
+      "learning_rate": 9.99348006995809e-06,
+      "loss": 2.4008,
+      "mean_token_accuracy": 0.4792257468216121,
+      "num_tokens": 279826165.0,
+      "step": 1929
+    },
+    {
+      "entropy": 2.430908203125,
+      "epoch": 0.03322916935685201,
+      "grad_norm": 0.5395227670669556,
+      "learning_rate": 9.993466013876707e-06,
+      "loss": 2.3978,
+      "mean_token_accuracy": 0.4806228969246149,
+      "num_tokens": 279980558.0,
+      "step": 1930
+    },
+    {
+      "entropy": 2.423828125,
+      "epoch": 0.03324638654304727,
+      "grad_norm": 0.5909799933433533,
+      "learning_rate": 9.993451942670044e-06,
+      "loss": 2.393,
+      "mean_token_accuracy": 0.48585162637755275,
+      "num_tokens": 280119585.0,
+      "step": 1931
+    },
+    {
+      "entropy": 2.512939453125,
+      "epoch": 0.03326360372924253,
+      "grad_norm": 0.6035844683647156,
+      "learning_rate": 9.993437856338139e-06,
+      "loss": 2.5345,
+      "mean_token_accuracy": 0.4715218679048121,
+      "num_tokens": 280244876.0,
+      "step": 1932
+    },
+    {
+      "entropy": 2.4185791015625,
+      "epoch": 0.03328082091543779,
+      "grad_norm": 0.5301707983016968,
+      "learning_rate": 9.993423754881039e-06,
+      "loss": 2.4086,
+      "mean_token_accuracy": 0.48241995088756084,
+      "num_tokens": 280401852.0,
+      "step": 1933
+    },
+    {
+      "entropy": 2.4444580078125,
+      "epoch": 0.03329803810163305,
+      "grad_norm": 0.5419909358024597,
+      "learning_rate": 9.993409638298785e-06,
+      "loss": 2.4023,
+      "mean_token_accuracy": 0.4783248957246542,
+      "num_tokens": 280561735.0,
+      "step": 1934
+    },
+    {
+      "entropy": 2.4560546875,
+      "epoch": 0.03331525528782831,
+      "grad_norm": 0.5301235914230347,
+      "learning_rate": 9.99339550659142e-06,
+      "loss": 2.4201,
+      "mean_token_accuracy": 0.4791613514535129,
+      "num_tokens": 280719202.0,
+      "step": 1935
+    },
+    {
+      "entropy": 2.3726806640625,
+      "epoch": 0.03333247247402357,
+      "grad_norm": 0.5838637948036194,
+      "learning_rate": 9.993381359758985e-06,
+      "loss": 2.347,
+      "mean_token_accuracy": 0.4987304233945906,
+      "num_tokens": 280866306.0,
+      "step": 1936
+    },
+    {
+      "entropy": 2.45068359375,
+      "epoch": 0.03334968966021883,
+      "grad_norm": 0.5787121057510376,
+      "learning_rate": 9.993367197801527e-06,
+      "loss": 2.3823,
+      "mean_token_accuracy": 0.47824144922196865,
+      "num_tokens": 280996610.0,
+      "step": 1937
+    },
+    {
+      "entropy": 2.5333251953125,
+      "epoch": 0.03336690684641409,
+      "grad_norm": 0.5946321487426758,
+      "learning_rate": 9.993353020719083e-06,
+      "loss": 2.5036,
+      "mean_token_accuracy": 0.4754126025363803,
+      "num_tokens": 281133369.0,
+      "step": 1938
+    },
+    {
+      "entropy": 2.444580078125,
+      "epoch": 0.033384124032609354,
+      "grad_norm": 0.5681336522102356,
+      "learning_rate": 9.993338828511701e-06,
+      "loss": 2.419,
+      "mean_token_accuracy": 0.480977026745677,
+      "num_tokens": 281293453.0,
+      "step": 1939
+    },
+    {
+      "entropy": 2.469970703125,
+      "epoch": 0.03340134121880461,
+      "grad_norm": 0.5857387185096741,
+      "learning_rate": 9.993324621179424e-06,
+      "loss": 2.446,
+      "mean_token_accuracy": 0.48483802145347,
+      "num_tokens": 281434754.0,
+      "step": 1940
+    },
+    {
+      "entropy": 2.4569091796875,
+      "epoch": 0.03341855840499987,
+      "grad_norm": 0.6091550588607788,
+      "learning_rate": 9.99331039872229e-06,
+      "loss": 2.4002,
+      "mean_token_accuracy": 0.4800125630572438,
+      "num_tokens": 281571983.0,
+      "step": 1941
+    },
+    {
+      "entropy": 2.4578857421875,
+      "epoch": 0.03343577559119513,
+      "grad_norm": 0.6192805171012878,
+      "learning_rate": 9.993296161140346e-06,
+      "loss": 2.3834,
+      "mean_token_accuracy": 0.4805147326551378,
+      "num_tokens": 281716426.0,
+      "step": 1942
+    },
+    {
+      "entropy": 2.5145263671875,
+      "epoch": 0.03345299277739039,
+      "grad_norm": 0.5912413001060486,
+      "learning_rate": 9.993281908433637e-06,
+      "loss": 2.4856,
+      "mean_token_accuracy": 0.47100443998351693,
+      "num_tokens": 281859208.0,
+      "step": 1943
+    },
+    {
+      "entropy": 2.4542236328125,
+      "epoch": 0.03347020996358565,
+      "grad_norm": 0.5443603992462158,
+      "learning_rate": 9.9932676406022e-06,
+      "loss": 2.4321,
+      "mean_token_accuracy": 0.4776157597079873,
+      "num_tokens": 282011533.0,
+      "step": 1944
+    },
+    {
+      "entropy": 2.4461669921875,
+      "epoch": 0.03348742714978091,
+      "grad_norm": 0.5912041068077087,
+      "learning_rate": 9.993253357646085e-06,
+      "loss": 2.4245,
+      "mean_token_accuracy": 0.4797707684338093,
+      "num_tokens": 282142801.0,
+      "step": 1945
+    },
+    {
+      "entropy": 2.4661865234375,
+      "epoch": 0.03350464433597617,
+      "grad_norm": 0.5891805291175842,
+      "learning_rate": 9.99323905956533e-06,
+      "loss": 2.408,
+      "mean_token_accuracy": 0.4835380604490638,
+      "num_tokens": 282282290.0,
+      "step": 1946
+    },
+    {
+      "entropy": 2.505859375,
+      "epoch": 0.03352186152217143,
+      "grad_norm": 0.6140239238739014,
+      "learning_rate": 9.993224746359981e-06,
+      "loss": 2.4778,
+      "mean_token_accuracy": 0.47382217831909657,
+      "num_tokens": 282421711.0,
+      "step": 1947
+    },
+    {
+      "entropy": 2.464111328125,
+      "epoch": 0.03353907870836669,
+      "grad_norm": 0.5682819485664368,
+      "learning_rate": 9.993210418030082e-06,
+      "loss": 2.4156,
+      "mean_token_accuracy": 0.47630415530875325,
+      "num_tokens": 282576394.0,
+      "step": 1948
+    },
+    {
+      "entropy": 2.3934326171875,
+      "epoch": 0.03355629589456195,
+      "grad_norm": 0.5857619643211365,
+      "learning_rate": 9.993196074575673e-06,
+      "loss": 2.3721,
+      "mean_token_accuracy": 0.49081736570224166,
+      "num_tokens": 282729923.0,
+      "step": 1949
+    },
+    {
+      "entropy": 2.4425048828125,
+      "epoch": 0.03357351308075721,
+      "grad_norm": 0.6023845076560974,
+      "learning_rate": 9.9931817159968e-06,
+      "loss": 2.3967,
+      "mean_token_accuracy": 0.4840724840760231,
+      "num_tokens": 282862205.0,
+      "step": 1950
+    },
+    {
+      "entropy": 2.423095703125,
+      "epoch": 0.03359073026695247,
+      "grad_norm": 0.5758981108665466,
+      "learning_rate": 9.993167342293508e-06,
+      "loss": 2.3915,
+      "mean_token_accuracy": 0.48772860085591674,
+      "num_tokens": 282997486.0,
+      "step": 1951
+    },
+    {
+      "entropy": 2.5380859375,
+      "epoch": 0.033607947453147734,
+      "grad_norm": 0.550911545753479,
+      "learning_rate": 9.993152953465837e-06,
+      "loss": 2.5101,
+      "mean_token_accuracy": 0.4724648226983845,
+      "num_tokens": 283148960.0,
+      "step": 1952
+    },
+    {
+      "entropy": 2.501708984375,
+      "epoch": 0.033625164639342994,
+      "grad_norm": 0.5853079557418823,
+      "learning_rate": 9.993138549513834e-06,
+      "loss": 2.46,
+      "mean_token_accuracy": 0.4766088160686195,
+      "num_tokens": 283290181.0,
+      "step": 1953
+    },
+    {
+      "entropy": 2.412353515625,
+      "epoch": 0.033642381825538255,
+      "grad_norm": 0.5772018432617188,
+      "learning_rate": 9.99312413043754e-06,
+      "loss": 2.3818,
+      "mean_token_accuracy": 0.4874812732450664,
+      "num_tokens": 283446157.0,
+      "step": 1954
+    },
+    {
+      "entropy": 2.4298095703125,
+      "epoch": 0.033659599011733515,
+      "grad_norm": 0.5674806833267212,
+      "learning_rate": 9.993109696236997e-06,
+      "loss": 2.4043,
+      "mean_token_accuracy": 0.48742856411263347,
+      "num_tokens": 283583169.0,
+      "step": 1955
+    },
+    {
+      "entropy": 2.4627685546875,
+      "epoch": 0.033676816197928776,
+      "grad_norm": 0.5692137479782104,
+      "learning_rate": 9.993095246912254e-06,
+      "loss": 2.4209,
+      "mean_token_accuracy": 0.4784468016587198,
+      "num_tokens": 283726808.0,
+      "step": 1956
+    },
+    {
+      "entropy": 2.411865234375,
+      "epoch": 0.03369403338412403,
+      "grad_norm": 0.5745928287506104,
+      "learning_rate": 9.99308078246335e-06,
+      "loss": 2.3845,
+      "mean_token_accuracy": 0.48880593106150627,
+      "num_tokens": 283884790.0,
+      "step": 1957
+    },
+    {
+      "entropy": 2.4476318359375,
+      "epoch": 0.03371125057031929,
+      "grad_norm": 0.5757923722267151,
+      "learning_rate": 9.993066302890335e-06,
+      "loss": 2.4339,
+      "mean_token_accuracy": 0.47860253555700183,
+      "num_tokens": 284026807.0,
+      "step": 1958
+    },
+    {
+      "entropy": 2.4261474609375,
+      "epoch": 0.03372846775651455,
+      "grad_norm": 0.509971559047699,
+      "learning_rate": 9.993051808193246e-06,
+      "loss": 2.39,
+      "mean_token_accuracy": 0.4785323585383594,
+      "num_tokens": 284189392.0,
+      "step": 1959
+    },
+    {
+      "entropy": 2.4044189453125,
+      "epoch": 0.03374568494270981,
+      "grad_norm": 0.575952410697937,
+      "learning_rate": 9.99303729837213e-06,
+      "loss": 2.3911,
+      "mean_token_accuracy": 0.48549577174708247,
+      "num_tokens": 284324103.0,
+      "step": 1960
+    },
+    {
+      "entropy": 2.46044921875,
+      "epoch": 0.03376290212890507,
+      "grad_norm": 0.5873697996139526,
+      "learning_rate": 9.993022773427028e-06,
+      "loss": 2.419,
+      "mean_token_accuracy": 0.4775766748934984,
+      "num_tokens": 284466158.0,
+      "step": 1961
+    },
+    {
+      "entropy": 2.533935546875,
+      "epoch": 0.03378011931510033,
+      "grad_norm": 0.5654860734939575,
+      "learning_rate": 9.99300823335799e-06,
+      "loss": 2.4545,
+      "mean_token_accuracy": 0.47519396571442485,
+      "num_tokens": 284602979.0,
+      "step": 1962
+    },
+    {
+      "entropy": 2.472412109375,
+      "epoch": 0.03379733650129559,
+      "grad_norm": 0.7695214748382568,
+      "learning_rate": 9.992993678165055e-06,
+      "loss": 2.4068,
+      "mean_token_accuracy": 0.4817899586632848,
+      "num_tokens": 284740325.0,
+      "step": 1963
+    },
+    {
+      "entropy": 2.4033203125,
+      "epoch": 0.03381455368749085,
+      "grad_norm": 0.5517217516899109,
+      "learning_rate": 9.99297910784827e-06,
+      "loss": 2.3849,
+      "mean_token_accuracy": 0.48449106933549047,
+      "num_tokens": 284885765.0,
+      "step": 1964
+    },
+    {
+      "entropy": 2.4239501953125,
+      "epoch": 0.033831770873686114,
+      "grad_norm": 0.6454329490661621,
+      "learning_rate": 9.992964522407676e-06,
+      "loss": 2.4082,
+      "mean_token_accuracy": 0.48670034017413855,
+      "num_tokens": 285020125.0,
+      "step": 1965
+    },
+    {
+      "entropy": 2.4947509765625,
+      "epoch": 0.033848988059881374,
+      "grad_norm": 0.5675487518310547,
+      "learning_rate": 9.99294992184332e-06,
+      "loss": 2.4896,
+      "mean_token_accuracy": 0.4730933913961053,
+      "num_tokens": 285165664.0,
+      "step": 1966
+    },
+    {
+      "entropy": 2.4700927734375,
+      "epoch": 0.033866205246076635,
+      "grad_norm": 0.581511914730072,
+      "learning_rate": 9.992935306155246e-06,
+      "loss": 2.4402,
+      "mean_token_accuracy": 0.4833594365045428,
+      "num_tokens": 285304514.0,
+      "step": 1967
+    },
+    {
+      "entropy": 2.4307861328125,
+      "epoch": 0.033883422432271895,
+      "grad_norm": 0.5686525702476501,
+      "learning_rate": 9.992920675343496e-06,
+      "loss": 2.412,
+      "mean_token_accuracy": 0.4830741249024868,
+      "num_tokens": 285451178.0,
+      "step": 1968
+    },
+    {
+      "entropy": 2.44921875,
+      "epoch": 0.033900639618467156,
+      "grad_norm": 0.5576603412628174,
+      "learning_rate": 9.992906029408115e-06,
+      "loss": 2.3827,
+      "mean_token_accuracy": 0.4801848717033863,
+      "num_tokens": 285595105.0,
+      "step": 1969
+    },
+    {
+      "entropy": 2.415771484375,
+      "epoch": 0.033917856804662416,
+      "grad_norm": 0.5764795541763306,
+      "learning_rate": 9.992891368349151e-06,
+      "loss": 2.3845,
+      "mean_token_accuracy": 0.48759686667472124,
+      "num_tokens": 285726774.0,
+      "step": 1970
+    },
+    {
+      "entropy": 2.4775390625,
+      "epoch": 0.03393507399085768,
+      "grad_norm": 0.6182502508163452,
+      "learning_rate": 9.992876692166644e-06,
+      "loss": 2.491,
+      "mean_token_accuracy": 0.4704209272749722,
+      "num_tokens": 285846737.0,
+      "step": 1971
+    },
+    {
+      "entropy": 2.4888916015625,
+      "epoch": 0.03395229117705294,
+      "grad_norm": 0.5855472683906555,
+      "learning_rate": 9.99286200086064e-06,
+      "loss": 2.4541,
+      "mean_token_accuracy": 0.4759668963961303,
+      "num_tokens": 285986773.0,
+      "step": 1972
+    },
+    {
+      "entropy": 2.473388671875,
+      "epoch": 0.03396950836324819,
+      "grad_norm": 0.5262144804000854,
+      "learning_rate": 9.992847294431186e-06,
+      "loss": 2.473,
+      "mean_token_accuracy": 0.4791461192071438,
+      "num_tokens": 286147251.0,
+      "step": 1973
+    },
+    {
+      "entropy": 2.4501953125,
+      "epoch": 0.03398672554944345,
+      "grad_norm": 0.558925211429596,
+      "learning_rate": 9.99283257287832e-06,
+      "loss": 2.4154,
+      "mean_token_accuracy": 0.47716324497014284,
+      "num_tokens": 286310437.0,
+      "step": 1974
+    },
+    {
+      "entropy": 2.426025390625,
+      "epoch": 0.03400394273563871,
+      "grad_norm": 0.5795169472694397,
+      "learning_rate": 9.992817836202093e-06,
+      "loss": 2.3785,
+      "mean_token_accuracy": 0.4854766642674804,
+      "num_tokens": 286448769.0,
+      "step": 1975
+    },
+    {
+      "entropy": 2.52880859375,
+      "epoch": 0.03402115992183397,
+      "grad_norm": 0.5628653764724731,
+      "learning_rate": 9.992803084402547e-06,
+      "loss": 2.5193,
+      "mean_token_accuracy": 0.4707297068089247,
+      "num_tokens": 286588873.0,
+      "step": 1976
+    },
+    {
+      "entropy": 2.4481201171875,
+      "epoch": 0.03403837710802923,
+      "grad_norm": 0.5932325124740601,
+      "learning_rate": 9.992788317479727e-06,
+      "loss": 2.4434,
+      "mean_token_accuracy": 0.47935432521626353,
+      "num_tokens": 286733286.0,
+      "step": 1977
+    },
+    {
+      "entropy": 2.45361328125,
+      "epoch": 0.034055594294224494,
+      "grad_norm": 0.576995313167572,
+      "learning_rate": 9.992773535433678e-06,
+      "loss": 2.4079,
+      "mean_token_accuracy": 0.47960167936980724,
+      "num_tokens": 286875771.0,
+      "step": 1978
+    },
+    {
+      "entropy": 2.4423828125,
+      "epoch": 0.034072811480419754,
+      "grad_norm": 0.5512486100196838,
+      "learning_rate": 9.992758738264442e-06,
+      "loss": 2.4155,
+      "mean_token_accuracy": 0.480304847471416,
+      "num_tokens": 287031003.0,
+      "step": 1979
+    },
+    {
+      "entropy": 2.3990478515625,
+      "epoch": 0.034090028666615015,
+      "grad_norm": 0.5462455153465271,
+      "learning_rate": 9.992743925972069e-06,
+      "loss": 2.3775,
+      "mean_token_accuracy": 0.48557318560779095,
+      "num_tokens": 287180788.0,
+      "step": 1980
+    },
+    {
+      "entropy": 2.5538330078125,
+      "epoch": 0.034107245852810275,
+      "grad_norm": 0.5789005756378174,
+      "learning_rate": 9.992729098556601e-06,
+      "loss": 2.5184,
+      "mean_token_accuracy": 0.4639392225071788,
+      "num_tokens": 287318088.0,
+      "step": 1981
+    },
+    {
+      "entropy": 2.4896240234375,
+      "epoch": 0.034124463039005536,
+      "grad_norm": 0.66868656873703,
+      "learning_rate": 9.992714256018082e-06,
+      "loss": 2.4807,
+      "mean_token_accuracy": 0.4695613798685372,
+      "num_tokens": 287466895.0,
+      "step": 1982
+    },
+    {
+      "entropy": 2.4345703125,
+      "epoch": 0.034141680225200796,
+      "grad_norm": 0.5860902667045593,
+      "learning_rate": 9.99269939835656e-06,
+      "loss": 2.3898,
+      "mean_token_accuracy": 0.4838330140337348,
+      "num_tokens": 287612255.0,
+      "step": 1983
+    },
+    {
+      "entropy": 2.507080078125,
+      "epoch": 0.03415889741139606,
+      "grad_norm": 0.5566897988319397,
+      "learning_rate": 9.992684525572076e-06,
+      "loss": 2.498,
+      "mean_token_accuracy": 0.47418344393372536,
+      "num_tokens": 287765365.0,
+      "step": 1984
+    },
+    {
+      "entropy": 2.367431640625,
+      "epoch": 0.03417611459759132,
+      "grad_norm": 0.5585086941719055,
+      "learning_rate": 9.992669637664679e-06,
+      "loss": 2.3324,
+      "mean_token_accuracy": 0.49284734204411507,
+      "num_tokens": 287923606.0,
+      "step": 1985
+    },
+    {
+      "entropy": 2.4853515625,
+      "epoch": 0.03419333178378658,
+      "grad_norm": 0.5721833109855652,
+      "learning_rate": 9.99265473463441e-06,
+      "loss": 2.4174,
+      "mean_token_accuracy": 0.47697839234024286,
+      "num_tokens": 288065850.0,
+      "step": 1986
+    },
+    {
+      "entropy": 2.4268798828125,
+      "epoch": 0.03421054896998184,
+      "grad_norm": 0.5846219062805176,
+      "learning_rate": 9.992639816481317e-06,
+      "loss": 2.4077,
+      "mean_token_accuracy": 0.4877488249912858,
+      "num_tokens": 288219901.0,
+      "step": 1987
+    },
+    {
+      "entropy": 2.494140625,
+      "epoch": 0.0342277661561771,
+      "grad_norm": 0.7576902508735657,
+      "learning_rate": 9.992624883205446e-06,
+      "loss": 2.4919,
+      "mean_token_accuracy": 0.4697610507719219,
+      "num_tokens": 288368125.0,
+      "step": 1988
+    },
+    {
+      "entropy": 2.4490966796875,
+      "epoch": 0.03424498334237236,
+      "grad_norm": 0.5778993368148804,
+      "learning_rate": 9.99260993480684e-06,
+      "loss": 2.3865,
+      "mean_token_accuracy": 0.48272112663835287,
+      "num_tokens": 288502228.0,
+      "step": 1989
+    },
+    {
+      "entropy": 2.4388427734375,
+      "epoch": 0.03426220052856761,
+      "grad_norm": 0.5564126372337341,
+      "learning_rate": 9.992594971285545e-06,
+      "loss": 2.4208,
+      "mean_token_accuracy": 0.479803703725338,
+      "num_tokens": 288650170.0,
+      "step": 1990
+    },
+    {
+      "entropy": 2.4871826171875,
+      "epoch": 0.03427941771476287,
+      "grad_norm": 0.5940274596214294,
+      "learning_rate": 9.992579992641606e-06,
+      "loss": 2.4353,
+      "mean_token_accuracy": 0.4715647897683084,
+      "num_tokens": 288780528.0,
+      "step": 1991
+    },
+    {
+      "entropy": 2.4788818359375,
+      "epoch": 0.034296634900958134,
+      "grad_norm": 0.6397010087966919,
+      "learning_rate": 9.99256499887507e-06,
+      "loss": 2.4715,
+      "mean_token_accuracy": 0.4812845438718796,
+      "num_tokens": 288936902.0,
+      "step": 1992
+    },
+    {
+      "entropy": 2.417724609375,
+      "epoch": 0.034313852087153394,
+      "grad_norm": 0.5475616455078125,
+      "learning_rate": 9.99254998998598e-06,
+      "loss": 2.3702,
+      "mean_token_accuracy": 0.4824229711666703,
+      "num_tokens": 289091883.0,
+      "step": 1993
+    },
+    {
+      "entropy": 2.4339599609375,
+      "epoch": 0.034331069273348655,
+      "grad_norm": 0.5338613390922546,
+      "learning_rate": 9.99253496597438e-06,
+      "loss": 2.417,
+      "mean_token_accuracy": 0.479421756695956,
+      "num_tokens": 289241366.0,
+      "step": 1994
+    },
+    {
+      "entropy": 2.406005859375,
+      "epoch": 0.034348286459543916,
+      "grad_norm": 0.6116018295288086,
+      "learning_rate": 9.992519926840322e-06,
+      "loss": 2.3794,
+      "mean_token_accuracy": 0.4870318342000246,
+      "num_tokens": 289380641.0,
+      "step": 1995
+    },
+    {
+      "entropy": 2.4573974609375,
+      "epoch": 0.034365503645739176,
+      "grad_norm": 0.6071226596832275,
+      "learning_rate": 9.992504872583847e-06,
+      "loss": 2.4109,
+      "mean_token_accuracy": 0.47875087382271886,
+      "num_tokens": 289521937.0,
+      "step": 1996
+    },
+    {
+      "entropy": 2.4119873046875,
+      "epoch": 0.03438272083193444,
+      "grad_norm": 0.5995868444442749,
+      "learning_rate": 9.992489803205e-06,
+      "loss": 2.3732,
+      "mean_token_accuracy": 0.4848559848032892,
+      "num_tokens": 289670647.0,
+      "step": 1997
+    },
+    {
+      "entropy": 2.4595947265625,
+      "epoch": 0.0343999380181297,
+      "grad_norm": 0.5530126094818115,
+      "learning_rate": 9.992474718703829e-06,
+      "loss": 2.4787,
+      "mean_token_accuracy": 0.47666746005415916,
+      "num_tokens": 289833608.0,
+      "step": 1998
+    },
+    {
+      "entropy": 2.4443359375,
+      "epoch": 0.03441715520432496,
+      "grad_norm": 0.5142894387245178,
+      "learning_rate": 9.99245961908038e-06,
+      "loss": 2.3763,
+      "mean_token_accuracy": 0.48295469116419554,
+      "num_tokens": 289992658.0,
+      "step": 1999
+    },
+    {
+      "entropy": 2.42138671875,
+      "epoch": 0.03443437239052022,
+      "grad_norm": 0.5748444199562073,
+      "learning_rate": 9.992444504334696e-06,
+      "loss": 2.4077,
+      "mean_token_accuracy": 0.48649854911491275,
+      "num_tokens": 290134205.0,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 58082,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.7734025191882752e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}