diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4742 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 668,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0029940119760479044,
+      "grad_norm": 2.5684986427415533,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 0.6759,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 2.1870078024423028,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 0.6117,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 1.659885142945663,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.5765,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 1.7108034589922276,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.608,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 1.6862143432864258,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.4876,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 1.6428818519073503,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.5406,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 1.4915404124545144,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.4928,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 1.481552884135472,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.4914,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 1.4981861535283378,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.5378,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 1.4580052958877623,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.4876,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 1.3582660261355135,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.5228,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 1.3926113145853165,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.4881,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 1.29568649454872,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.4574,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 1.2267607590673537,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.4503,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 1.1963629473598174,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.4464,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 1.3831607897839804,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.4966,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 1.0481654548154904,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.3979,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 1.091606495839716,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.4524,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 1.2507495459115079,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.4724,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 1.1990598982855212,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.4587,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 1.155342503208866,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.4394,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 1.2390772484090544,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.4174,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 1.164457967553235,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.4944,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 1.1351881596611175,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.4104,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 1.3998377368521713,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.495,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 1.1505745999363388,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.4722,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 1.1585864328091309,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.418,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 1.2092346616695684,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.4564,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 1.298972760103769,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.4944,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 1.0084602154365117,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.4062,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 1.0974357647667319,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.4733,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 1.2745067083305746,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.448,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 1.0747468075170226,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.3934,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 0.9333090085311501,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.3725,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 1.14159174643903,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.4145,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 1.248637492178584,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.4352,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 1.2129032217921345,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.4793,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 1.198699958171089,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.4368,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 1.0124757599941567,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.4018,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 1.2119767898213634,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.4818,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 1.1662555350625896,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.4078,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 1.051168755596803,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.3668,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 1.1521618630128962,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.3786,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 1.0997399670953272,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.3653,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 1.1429955555759315,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.4367,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 1.2392350262563248,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.438,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 1.098349991075592,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.3631,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 1.2085404419425987,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.4569,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 1.212310820747002,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.4616,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 1.1284546373103213,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.4174,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 1.0903620002295922,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.4028,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 1.0761941109453603,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.381,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 1.1556026759584297,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.4319,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 0.9849307021976944,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.382,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 1.1912271006745863,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.3853,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 1.1898494397924877,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.4196,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 1.1227163351073903,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.4133,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 1.1066482810815517,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.4436,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 1.1810694338768457,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.4072,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 1.1609313976088875,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.4261,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 1.2363968036252764,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.4119,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 1.0975052342560239,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.4089,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 1.1487799393925107,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.3675,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 1.1359381147922636,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.4341,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 1.2415358939443493,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.4041,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 1.1414873276872581,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.3864,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 1.1609884036002451,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.3654,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 1.1096140337573228,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.453,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 0.9347086540882115,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.3529,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 1.1142307223616525,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.4463,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 1.1477956644252436,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.4428,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 1.1472219277512563,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.349,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 1.2221276827006464,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.4309,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 1.2400041912538753,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.4038,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 1.251134248671204,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.4168,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 1.0803914383597526,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.3291,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 1.1611919573165284,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.396,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 1.0471821847843046,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.4507,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 1.0426396720282607,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.3615,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 1.3386653875399963,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.4381,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 1.2277656193526383,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.388,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 1.1459232654853235,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.4028,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 1.1540093406988936,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.3774,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 1.0992611835852355,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.3488,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 1.184796952752682,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.3834,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 1.1080115599881006,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.3914,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 1.2238744327888202,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.4424,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 1.2270530425779878,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.4365,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 1.1736574761494094,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.4835,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 1.1411742203539919,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.3901,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 1.2236797734889129,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.3921,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 1.251489658085733,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.4274,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 1.135473638867319,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.3992,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 1.1668727258894909,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.4333,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 1.165562593332114,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.4136,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 1.0687396544986907,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.4193,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 1.0701193730054788,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.429,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 1.2519083257468975,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.3866,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 1.312712236583964,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.4041,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 1.0050239827324903,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.3695,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 1.1153471500696392,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.3708,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 1.179381167893325,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.3327,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 1.254304738107218,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.4038,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 1.2621893242184667,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.4245,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 0.9978635094350298,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.3882,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 1.1286368317354771,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.3759,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 1.220523130521007,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.4442,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 1.146478925830045,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.4,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 1.114819392412841,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.3767,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 1.2553976069040254,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.3865,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 1.1492094209562156,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.4121,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 1.1404283765191363,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.3832,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 1.1789709325395334,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.4558,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 1.0893115011437613,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.3612,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 1.0854846371061804,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.3731,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 1.1316564595426168,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.3558,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 1.1971310916956936,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.4249,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 1.3357806414170021,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.3997,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 1.2511206589495913,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.4261,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 1.3179733475060855,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.4638,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 1.2381890827244346,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.3963,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 1.2440786950013163,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.4058,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 1.2160973819099832,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.3846,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 1.145322701275584,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.4363,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 1.179641996064665,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.3902,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 1.1466304935908986,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.388,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 1.1782709285272193,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.4452,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 1.1071071364791731,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.4255,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 1.083811012640481,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.3975,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 1.3332943668034665,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.4212,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 1.2569468907272243,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.449,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 1.248560947876821,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.4125,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 1.149353671899829,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.4254,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 1.1613800574162654,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.3835,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 0.9988646415476073,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.359,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 1.0927876402089238,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.3559,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 1.2560186364161712,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.4326,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 1.2457156993500715,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.4053,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 1.184326394223682,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.37,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 1.2795811190191488,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.4177,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 0.9899712392839162,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.3422,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 1.1754225465612913,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.3889,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 1.0167887098583426,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.369,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 1.0846276754633213,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.3682,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 1.1214165030061098,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.4625,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 1.4105225494381062,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.4012,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 0.9799729049828462,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.3535,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 1.2039339501889776,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.3844,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 1.3159630886577631,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.4047,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 1.1499245892271774,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.358,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 1.0999162512050258,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.3844,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 1.013634367003106,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.392,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 1.1512044575884284,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.3605,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 1.1318735047721669,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.3382,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 1.2135497935515223,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.3698,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 1.2369182724013155,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.4206,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 1.1804738248995306,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.3708,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 1.1265364765470836,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.4285,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 1.2776129776015892,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.405,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 1.180048393091318,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.3977,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 1.2269982294399184,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.3974,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 1.0130140420096416,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.3369,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 1.2063996523783755,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.3549,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 1.2456942652390874,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.4476,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 1.0955780451392119,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.3778,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 1.1466811521539844,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.3757,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1496206710899008,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.3729,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 1.1423132867983192,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.4135,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 1.190506415309705,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.4116,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 1.2879604150787622,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.3567,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 1.1251296874959529,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.4254,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 1.0632336835362381,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.3651,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 1.3109571106658262,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.4546,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 1.062925456915242,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.3664,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 1.1181420059524119,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.3958,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 1.1681647473612264,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.4186,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 1.1907855008978772,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.3895,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 1.2391247464404112,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.38,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 1.171624220790207,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.4196,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 1.0525592177810936,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.3487,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 1.1520734943096,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.4011,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 1.3006950049032056,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.4467,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 1.1245071889562643,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.4218,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 1.066169550139698,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.4514,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 1.162360163754938,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.3696,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 1.1882540005679727,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.4263,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 1.0745205378770677,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.4334,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 1.2974008762520666,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.4834,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 1.046926791239543,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.3987,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 1.0806052824075587,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.3658,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 1.0894265799252698,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.3902,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 1.1170923269610191,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.4286,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 0.9552857351676135,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.3803,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 0.972660924386459,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.3431,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 1.1326423674540689,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.3979,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 1.1013918617515985,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.3595,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 1.200696062870925,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.3678,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 1.1229205922335992,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.37,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 1.173685360698167,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.3785,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 1.0354247658085591,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.3483,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.3796332776546478,
+      "eval_runtime": 3.7053,
+      "eval_samples_per_second": 14.574,
+      "eval_steps_per_second": 3.778,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 1.0574773190112456,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.3892,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 1.1627078648264697,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.3364,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 1.1360940220384608,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.3774,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 1.3192022944846828,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.41,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 1.0942381063837077,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.383,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 1.3614868436190495,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.3939,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 1.1206766826096806,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.3845,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 1.1189993345245735,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.3496,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 1.1650115590509722,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.3395,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 1.275616649735328,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.3836,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 1.1616563636709636,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.4019,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 1.1535616499523835,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.3536,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 1.1038379901849984,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.3666,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 1.1007088332293462,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.3433,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 1.0743865607569583,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.337,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 1.1374602968466345,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.3986,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 1.1118633395830146,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.3911,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 1.1970869644694837,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.3668,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 1.0828803324659033,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.3637,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 1.1109487848401283,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.4,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 1.2280215530273493,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.4228,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 1.0776092634988133,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.3572,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 1.2234694461071112,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.4112,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 1.2042101548321122,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.432,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 1.0243439862909391,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.3641,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 1.0078877275806968,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.3582,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 1.1495218126332543,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.4047,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 1.0928476831713445,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.3681,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 0.9735351055021317,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.3354,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 1.3281545946844366,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.4231,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 1.134679194197466,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.3883,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 1.1797357418084604,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.3865,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 1.1679497477474623,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.4002,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 1.2120989325893021,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.3743,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 1.081674963630625,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.3677,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 1.4561780928336705,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.3804,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 1.1453773968147631,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.3869,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 1.2629534360010732,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.4191,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 1.2440272178213474,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.3823,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 1.0953456009902518,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.4018,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 1.0549037266542864,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.3442,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 1.1417886011522977,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.3544,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 0.972964717456768,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.3767,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 1.0937648333982815,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.3908,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 1.0800529707297977,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.4148,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 1.0802451648884879,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.3906,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 1.138450309538047,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.4087,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 1.1077009560466826,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.3623,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 1.3026237036590287,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.4026,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 1.069643066214063,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.3702,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 1.2541965589911064,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.4132,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 1.0366143269231232,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.3903,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 1.187038076762178,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.3433,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 1.150551500226316,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.4123,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 1.0762471472963122,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.3842,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 1.1576992406553228,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.4085,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 1.0454618845223356,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.3821,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 1.3416982784512352,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.3773,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 1.1051718444526994,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.3257,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 1.1737632090794308,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.358,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 1.1449183444471438,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.3866,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 1.1263524138678083,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.3702,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 1.1218013494296515,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.3905,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 1.174713138126168,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.4293,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 1.2539509334304397,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.3837,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 1.2033742603281803,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.3758,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 1.1076585293735763,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.3972,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 1.0548977793227818,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.297,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 1.2276991791363598,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.3784,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 1.288170266299013,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.3511,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 1.0061235960933745,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.3713,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 1.3030029720769087,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.3837,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 1.2248625969334517,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.4131,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 1.1918714843907305,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.4198,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 1.099590838483367,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.3299,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 1.0536203266496,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.3869,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 0.927704018973631,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.3389,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 1.0434985624136586,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.3366,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 1.0405061516221041,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.37,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 1.0424347085964958,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.3473,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 1.129766907877991,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.3846,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 1.3039910864481177,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.4177,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 1.148760959535974,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.3945,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 1.1080447701890759,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.4027,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 1.0023362785937817,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.3362,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 1.0901616095446836,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.3521,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 1.0766067914320732,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.3729,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 1.0101397539789119,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.3241,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 1.139863043537744,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.3082,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 1.2091712419553207,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.4053,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 1.149061562688958,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.3782,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 1.169722746436091,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.3558,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 0.9643842755004689,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.3362,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 1.0518258348351994,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.3478,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 0.946031588136223,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.3711,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 1.2205594745152815,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.3856,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 1.2306317554276809,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.3839,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 1.126437835719368,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.3572,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 1.1495100730640169,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.3711,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 1.1191286649997845,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.4146,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 1.140294144290007,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.3524,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 1.1639276508597438,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.3336,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 1.149470100240248,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.3765,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 1.1565973147002475,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.3685,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 1.2523468315631388,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.4365,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 1.0350966794044707,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.3167,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 1.1189736382283255,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.3869,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 1.1050512048488856,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.3463,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 1.0768572485726693,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.329,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 1.0309839929253064,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.3498,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 1.2579179338388875,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.3263,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 1.0263819321527083,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.309,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 1.0430261552272444,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.3521,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 1.0961591062626004,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.3308,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 1.2292224007452555,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.3851,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 1.1827469161608375,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.4088,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 1.1796059985890623,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.3877,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 1.0219712666391143,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.3573,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 1.114783481529202,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.3734,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 1.2563564812348642,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.3413,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 1.0719443117396734,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.3316,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 1.353725591949236,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.3942,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 1.1031868778772997,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.3809,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 1.1474447568600847,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.3626,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 1.116564029163105,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.3126,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 1.0860383713922228,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.3767,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 1.110075445686529,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.3702,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 1.2033900803395718,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.3502,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 1.1246494849930384,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.3506,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 1.2179443769838192,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.4214,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 1.0010973104342133,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.3786,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 1.049066224432196,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.334,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 1.2462964500590836,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.4756,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0476271838489304,
+      "learning_rate": 5e-06,
+      "loss": 0.2556,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 1.168425489550037,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.2101,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 1.068017815315684,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.2324,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 1.1810357236226157,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.2305,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 1.2232547041108326,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.2718,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 1.164955136722702,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.2346,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 1.0221714873395247,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.2241,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 1.1459158510550242,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.2519,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 1.060560887372203,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.1956,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 1.3294272599658636,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.2426,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 1.0702693452730088,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.2124,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 1.2066034342766916,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.2043,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 1.2431056638420028,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.2747,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 1.2945024399322367,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.2636,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 1.2998085647473765,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.2067,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 1.3219620709888988,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.2617,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 1.2156946632430135,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.2451,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 1.1594061830162097,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.2427,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 1.2497466533708288,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.2665,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 1.1743896523999087,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.2687,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 1.1271819462842865,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.2287,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 1.4384318948208028,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.2165,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 1.2453042186279306,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.2285,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 1.2530657103788965,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.2359,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 1.2816127592696887,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.2729,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 1.2479174508342532,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.2489,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 1.2931935678589295,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.2802,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 1.1173831084173216,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.1855,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 1.206288574865383,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.2815,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 1.3050156701445272,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.2576,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 1.4745885866224457,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.2757,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 1.1661333094276236,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.2416,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 1.1205615286038384,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.2431,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 1.2652484953661443,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.2396,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 1.213872201441164,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.2688,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 1.3237545879753974,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.2866,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 1.1899301010649346,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.2177,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 1.1914780493750283,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.2644,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 1.0125957502275225,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.2413,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 1.2421091026319004,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.2191,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 1.2142784097553314,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.2313,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 1.3968228397824856,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.2642,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 1.264837103559239,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.2599,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 1.2630449446020016,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.2657,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 1.2042554378585877,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.2361,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 1.1797078128489729,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.2842,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 1.415726950657126,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.2359,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 1.1003039995307402,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.2671,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 1.1851580873021248,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.2515,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 1.2308936239295925,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.2295,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 1.249402388317944,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.2721,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 1.1727121345263098,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.217,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 1.1747382795998378,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.2633,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 1.0817252925267244,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.2125,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 1.1004174381946845,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.2284,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 1.277215608655292,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.25,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 1.1917015049994824,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.2295,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 1.3269288844800287,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.2282,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 1.1713378526146192,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.2162,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 0.9444509537521274,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.2165,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 1.2130288816639867,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.2543,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 1.244386581386584,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.2113,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 1.1574614970547168,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.2282,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 1.1605942729242582,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.1841,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 1.150704897532228,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.2789,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 1.2001676027683796,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.235,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 1.1285698556511736,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.2544,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.37970176339149475,
+      "eval_runtime": 3.7051,
+      "eval_samples_per_second": 14.574,
+      "eval_steps_per_second": 3.779,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 1.1625715676776498,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.2639,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 1.189480455950514,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.2674,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 1.262065554188991,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.263,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 1.2761504655555476,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.2172,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 1.1558017366715925,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.2334,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 1.0919956423106152,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.2274,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 1.1160379899877206,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.239,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 1.179058854026378,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.2545,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 1.3250081090583692,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.2342,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 1.254266629618469,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.2615,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 1.1577472869715493,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.2392,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 1.3130641624616446,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.2475,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 1.1211787874921293,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.2363,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 1.2088790269114458,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.2459,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 1.2589597737733156,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.205,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 1.3119593400526566,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.2383,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 1.3373388352139062,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.21,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 1.1238655359143708,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.2303,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 1.2629963455592639,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.2579,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 1.2692853722990318,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.2254,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 1.11113546714538,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.2191,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 1.1797548235214954,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.2612,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 1.2213714279437202,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.2197,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 1.177859280561155,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.2578,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 1.3660531446198467,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.2402,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 1.268938550833524,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.2503,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 1.2616660013024945,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.2455,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 1.2154668523504337,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.2385,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 1.2001124991304555,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.2296,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 1.0763594812393562,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.2683,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 1.178197448209806,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.2473,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 1.0747155223914846,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.2627,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 1.2216513682257195,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.2329,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 1.0758255583261227,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.2245,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 1.209473326219844,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.2361,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 1.1407325753544735,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.2487,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 1.187618645013766,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.2718,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 1.1972288192645222,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.2043,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 1.2955689031695745,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.2255,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 1.094700584348356,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.2408,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 1.159343966280313,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.2202,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 1.2428177669168883,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.2612,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 1.279890395622471,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.2489,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 1.169737606490123,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.2932,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 1.21758341892203,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.2518,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 1.249110711268284,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.2369,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 1.0980594238862944,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.2211,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 1.2926941061296717,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.2992,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 1.2942387014735002,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.2515,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 1.1708050591374326,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.2723,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 1.0968957266733412,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.2216,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 1.1680681073380084,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.2363,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 1.0458970035518702,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.1962,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 1.114453810542386,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.2052,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 1.2620140006517826,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.2328,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 1.0403708257571322,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.2182,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 1.2133474850509236,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.2932,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 1.204241316848595,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.2374,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 1.289272263963612,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.2146,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 1.4704790293257122,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.2264,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 1.123096530253908,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.2282,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 1.2897844893368526,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.2261,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 1.3434458295110552,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.2102,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 1.0157562714610826,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.2195,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 1.2819643738917095,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.2281,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 1.2121225258393904,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.19,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 1.1344215880388435,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.2305,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 1.1522591681141334,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.1892,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 1.2878630912696023,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.2386,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 1.0706926166705846,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.2102,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 1.3304419910806877,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.2468,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 1.3204218703560082,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.2334,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 1.1541318181242202,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.2585,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 1.2976760525804758,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.2606,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 1.1049120525438758,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.2522,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 1.1791965891303355,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.2401,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 1.3529032802253063,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.2713,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 1.3127992167795512,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.3185,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 1.0799207016412442,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.2219,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 1.1738823085766839,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.2137,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 1.2037235548728373,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.2187,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 1.1926206769559855,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.2672,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 1.4023615277813126,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.2269,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 1.2128575283501126,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.2475,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 1.070669635525351,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.2326,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 1.2938227445885073,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.2728,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 1.1121747938498743,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.2201,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 1.3124066628620017,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.2318,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 1.1801763035249164,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.1978,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 1.2020638598739795,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.2477,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 1.092189746185797,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.2215,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 1.0818581448729352,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.2552,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 1.0358416012327907,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.2449,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 1.2402944265744058,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.2413,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 1.1750533945768598,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.231,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 1.2978110732531931,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.1995,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 1.2154585480023996,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.2481,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 1.3120624416719264,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.2026,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 1.2972401593799026,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.2554,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 1.1027634508400548,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.2419,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.1835273237650885,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.2463,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 1.2284384416023544,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.2421,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 1.183488824462398,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.2271,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 1.1596493715456915,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.2226,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 1.300757782466068,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.241,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 1.2364875673306517,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.2311,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 1.816354201272432,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.2335,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 1.2854111379233553,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.2439,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 1.2228317077935926,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.2718,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 1.1165357883160134,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.2377,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 1.2294473021925143,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.1981,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 1.3263960366019032,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.2363,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 1.1941111376486677,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.2383,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 1.3528430599244223,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.2656,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 1.2238270253510002,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.2341,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 1.2409862553676885,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.2158,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 1.0890426630176158,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.2197,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 1.2191842075805623,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.2293,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 1.227028499984413,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.2295,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 1.2005905306903297,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.2434,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 1.2519430035660841,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.19,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 1.2730746814840446,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.2176,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 1.2360215809676898,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.2514,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 1.2838706070758128,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.2767,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 1.3255335556707444,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.25,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 1.2639985454718012,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.1869,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 1.309043769350664,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.2159,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 1.2134451271153457,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.261,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 1.1486765709843596,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.2596,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 1.1072959876131667,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.243,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 1.207308336311589,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.2241,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 1.2364123773541555,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.1793,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 1.3905301448522334,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.2686,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 1.3656965269730603,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.248,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 1.2617615223880487,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.2422,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 1.2204211590326455,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.2359,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 1.294891714190642,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.2442,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 1.1525507873442862,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.2393,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 1.1584177268431417,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.2234,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 1.0055322122599197,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.1824,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 1.0098507157684533,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.2292,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 1.4344411116678193,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.2131,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 1.1826565146400527,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.2239,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 1.4176191282805275,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.2779,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 1.2076644450982323,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.2521,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 1.0744721747653352,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.1774,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 1.3058780249397233,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.2437,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 1.3423023073413152,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.2668,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 1.1665751455734443,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.227,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 1.3532514708402759,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.2631,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 1.2352662285799931,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.2538,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 1.1756218677153056,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.2158,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 1.182674050367567,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.2342,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 1.220203545939254,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.2369,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 1.269533465738827,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.2363,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 1.122871483191075,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.256,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 1.152873452816829,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.2296,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 1.2953037923277637,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.2357,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 1.3367690295549244,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.2405,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 1.0721948205613723,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.2441,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 1.1362758598396672,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.2283,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 1.2945375334386646,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.2368,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 1.3164551228807921,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.2168,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 1.274431063796111,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.2455,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 1.4038274188616755,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.2134,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 1.259735373753407,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.2494,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 1.0494031147303466,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.2091,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 1.2440169735545998,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.2105,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 1.2566805807138788,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.2628,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 1.153470819511567,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.2339,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 1.129604040357155,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.1943,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 1.2368833376852846,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.2548,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 1.1145094060997698,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.1916,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 1.2006187285348597,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.235,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 1.0749497055177248,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.2196,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 1.3061263390616138,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.2185,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 1.0806882790011505,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.2259,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 1.1407626427181037,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.2248,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 1.0679297158261194,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.2038,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 1.2605937763440398,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.2491,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 1.1920753626539569,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.2247,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 1.1085287162448896,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.2162,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 1.231648781460144,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.219,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 1.1960171355811111,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.2254,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 1.2347747656559855,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.2733,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 1.2687097360599482,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.2254,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 1.2353963474413097,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.2186,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 1.2277333923199956,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.2244,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 1.1449690714218823,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.2019,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 1.276195929163353,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.2104,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 1.184162567747019,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.2419,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 0.9444262899615484,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.2043,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 1.168244170419642,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.2069,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 1.1500658928489604,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.2423,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 1.222558521904561,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.2619,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 1.1904232721020807,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.224,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 1.221722551610529,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.2209,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 1.1357016410917007,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.2353,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 1.2442717635682137,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.222,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 1.174462302042287,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.2508,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.3729483187198639,
+      "eval_runtime": 3.7058,
+      "eval_samples_per_second": 14.572,
+      "eval_steps_per_second": 3.778,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 1.097604685127189,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.2069,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 1.3271527818206188,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.2799,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 1.1515639279275545,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.2582,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 1.0476033326439913,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.2239,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 1.0842282843605353,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.2214,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 1.2735967756779112,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.1866,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 1.1247710327086144,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.2044,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 1.3453392552474546,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.2352,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 1.0584823797687992,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.2015,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 1.2279827401542938,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.2386,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 1.1087453471963389,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.2243,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 1.346469723716929,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.2379,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 1.3409753618185365,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.2114,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 1.2062006488860182,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.2283,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 3.1526948795129135,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.2423,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 1.3471553297914798,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.2569,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 1.2106352530323428,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.2115,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 1.126757868861992,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.1921,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 1.1145921573101967,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.202,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 1.1476260942876484,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.2382,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 1.3297050664495638,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.2267,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 1.0595032173421388,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.2238,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 1.2339372348965452,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.2627,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 1.1540838458616327,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.1973,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 1.1791562191719591,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.2301,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 1.5372906995398794,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.2778,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 1.2684762478547584,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.2661,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 1.1064831078128279,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.2254,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 1.2390074876358401,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.2356,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 1.2303601796316272,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.2302,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 1.2246836564295691,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.2545,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 1.1106316426739853,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.1957,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 1.2321477349954442,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.2514,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 1.3382202728511319,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.2185,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 1.4262040782038419,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.2221,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 1.2172306411050249,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.202,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 1.2466982683338215,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.2312,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 1.1909265103858955,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.2275,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 1.320999347660595,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.2135,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 1.3295842392935364,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.2595,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 1.279345661783718,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.2197,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 1.1571114162922322,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.2372,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 1.1833556326780292,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.2163,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 1.1176901483983197,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.268,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 1.3611823229112894,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.213,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 1.19091766759133,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.2549,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 1.0353035425264436,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.1971,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 1.214811795559626,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.2183,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 1.1223584417179708,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.2186,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 1.2429482850784426,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.2471,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 1.1529142422396113,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.2157,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 1.2144522523278902,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.2083,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 1.1534076182407542,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.247,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 1.2177196804105526,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.1981,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 1.0871194334548366,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.2399,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 1.2289883996581026,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.2361,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 1.2023316566201565,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.2644,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 1.3348824304695317,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.2501,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 1.1979732558640979,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.2048,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 1.2833136488145622,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.2771,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 1.1435148917456563,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.2363,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 1.0953281262889996,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.201,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 1.2579015794523474,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.2729,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 1.33930507305549,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.2619,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 1.2027290821840704,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.2092,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 1.1235327042688426,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.2148,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 1.0290897656495457,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.2012,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.053965460699182,
+      "learning_rate": 0.0,
+      "loss": 0.1366,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 24252133736448.0,
+      "train_loss": 0.31642159471611775,
+      "train_runtime": 2061.4612,
+      "train_samples_per_second": 5.17,
+      "train_steps_per_second": 0.324
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 668,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 24252133736448.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}