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": 4.4065612373252705,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 0.1955,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 2.2197780064252863,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 0.1285,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 2.2015930263947965,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.1469,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 1.827578701804767,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.1295,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 1.5758942990982774,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.1119,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 1.6507182925853439,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.1046,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 1.6014264894572352,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.1073,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 1.6485754378952608,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.1051,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 1.7127751361842105,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.1225,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 1.8736881390995126,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.1146,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 1.6302994704387683,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.1166,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 1.658412635750142,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.1014,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 1.586452780263754,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.118,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 1.7171140701630243,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.1178,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 1.7723022023043875,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.103,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 1.747252877677027,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.1219,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 1.2212678874534266,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.0826,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 1.4358004224119256,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.1166,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 1.9089852190398344,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.1353,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 1.4649193375690892,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.1211,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 1.3834175334015157,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.1165,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 1.603839212867631,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.1261,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 1.365937596034718,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.1176,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 1.4712574839082266,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.0985,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 1.5637703180515312,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.1154,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 1.6047280834400812,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.1183,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 1.4223195245469609,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.1209,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 1.4422259287641548,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.1171,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 1.6437613969311677,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.1185,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 1.2788386000967815,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.1143,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 1.107897899069559,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.1089,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 1.5033062659803353,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.1103,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 1.461127270216455,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.1052,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 1.8403673533317306,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.1243,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 1.3464233486305548,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.1035,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 1.3739823687493635,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.1159,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 1.8591637895028221,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.143,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 1.4813643061936836,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.1115,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 1.2715724988816555,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.0959,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 1.5161750497979416,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.1217,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 1.3357866014257982,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.1252,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 1.0555110672378796,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.0788,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 1.3033241364052839,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.1046,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 1.243376614868187,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.0904,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 1.4283771892195505,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.1152,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 1.5585806874359363,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.1119,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 1.2477939713464385,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.1071,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 1.5457745827412808,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.1272,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 1.4665216416878972,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.1262,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 1.3833224404650704,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.109,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 1.3833491944550003,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.1227,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 1.1770028073174121,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.1014,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 1.4053845858068004,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.1353,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 1.3135447059237153,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.1064,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 1.303961509612364,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.1098,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 1.270463091283257,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.1141,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 1.4425588095318302,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.1202,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 1.5325943373307436,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.157,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 1.4089413920953124,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.1211,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 1.5027451836890469,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.1154,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 1.5936350194990287,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.1334,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 1.3535105736747741,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.1102,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 1.3399865743083306,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.0954,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 1.244114406986766,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.12,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 1.405981538961236,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.1213,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 1.2857754483219834,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.1115,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 1.1915696382457759,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.1046,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 1.2661856633793638,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.1025,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 1.2652995993843252,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.1099,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 1.3000308896771475,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.1205,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 1.4185176226721756,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.127,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 1.2834093553886297,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.0868,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 1.5188201997430797,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.135,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 1.4299996949345777,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.1102,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 1.4731106307354767,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.1149,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 1.2304901647514455,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.0823,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 1.4962679210014527,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.0995,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 1.371336778613948,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.1257,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 1.291715492348199,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.1102,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 1.685429592404092,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.144,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 1.2590782327456564,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.1123,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 1.2405784333505532,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.1072,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 1.1877293441253545,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.1028,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 1.124460976300959,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.0949,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 1.0852435168028043,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.0909,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 1.4141342286823342,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.1062,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 1.3165917686726092,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.1109,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 1.7031143024622464,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.1263,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 1.4419372097813596,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.1409,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 1.2218915759103905,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.0958,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 1.282219907459581,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.1208,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 1.3616070807195635,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.1167,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 1.1658746078687818,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.1054,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 1.1927249055935787,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.1083,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 1.2042239684186842,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.109,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 1.2610816799773084,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.1219,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 1.1587254798304683,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.103,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 1.1191143026890922,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.0923,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 1.4422182598249509,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.1191,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 1.195681880483653,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.0997,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 1.242571886060057,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.0963,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 1.1694848264614026,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.087,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 1.5970622897796025,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.1185,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 1.240721577565901,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.1098,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 1.3293986022978173,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.1146,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 1.3418678794848522,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.1168,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 1.4460999757907682,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.1193,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 1.2279990682970239,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.1207,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 1.258779034721427,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.1074,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 1.4733245534702113,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.1228,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 1.2979836375325358,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.1277,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 1.359400606687236,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.1184,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 1.3633955553561836,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.1398,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 1.2763616658449906,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.1166,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 1.2437145898051003,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.113,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 1.2228193698347471,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.1054,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 1.3340933406777158,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.1211,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 1.3477395403858667,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.1075,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 1.2041373533847899,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.1155,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 1.4006931287313713,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.1197,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 1.195039927487423,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.113,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 1.2667464119020673,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.1013,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 1.3282888614241355,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.1021,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 1.2815260282670005,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.1183,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 1.279074654194517,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.1037,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 1.2929899930310056,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.1052,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 1.419258942958804,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.1266,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 1.3765106811519057,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.1266,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 1.362462815548897,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.1148,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 1.3632987389486526,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.1122,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 1.5034925676620612,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.1259,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 1.3544611887622489,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.1155,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 1.4492980529100679,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.1354,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 1.3661588478255717,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.1198,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 1.1640306434381218,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.1,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 1.1634026680103602,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.1067,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 1.7322204335361062,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.1474,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 1.355845489112246,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.1098,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 1.2624916483965816,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.1026,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 1.3328692742780148,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.1222,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 1.1530058254033686,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.1075,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 1.253988989798005,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.1243,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 1.120279212793384,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.099,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 1.0930123314565456,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.1037,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 1.4153555311188872,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.1354,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 1.348336318270794,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.1214,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 1.1281949837582366,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.099,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 1.4464084649993627,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.1314,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 1.362240115089838,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.1138,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 1.1666016390181924,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.1157,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 1.2166297978233047,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.1162,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 1.1863090290455607,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.1138,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 1.2184510368554387,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.1056,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 1.2509477525397605,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.114,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 1.1144592757525316,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.0927,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 1.472806268371992,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.1372,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 1.4079333024405298,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.1249,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 1.107333212534039,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.1148,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 1.3051134189131481,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.1233,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 1.2321113176130036,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.0971,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 1.4545466378655705,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.125,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 1.241570892156678,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.1046,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 1.3862223510600216,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.1143,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 1.3737319937521504,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.1152,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 1.5659401820346006,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.1308,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 1.2498371163755475,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.1021,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1525546668603834,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.0938,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 1.2042542991262981,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.1231,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 1.1878923084167112,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.0931,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 1.3398512326729737,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.101,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 1.398339779914391,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.1311,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 1.183076300483656,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.1043,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 1.4417209898208314,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.1185,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 1.3263894409471795,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.1054,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 1.3030005612313973,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.1134,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 1.2037650274107088,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.1028,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 1.45882996192205,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.1284,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 1.420515198725602,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.1179,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 1.2715290453689616,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.1047,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 1.3467385725782712,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.1348,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 1.4096066097090032,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.1502,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 1.592413225891031,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.1306,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 1.378077618972582,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.1249,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 1.230860642502367,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.1191,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 1.3357396620842799,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.1283,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 1.277753376860637,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.1224,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 1.4928256305526444,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.1444,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 1.293801974961885,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.1201,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 1.2732861672408446,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.1236,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 1.1806430205143141,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.105,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 1.2335244613886174,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.1043,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 1.4943148016459196,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.1332,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 1.1287088166015442,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.1128,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 1.2907241747139357,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.1178,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 1.2490062247611726,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.1273,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 1.0754655229921946,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.0917,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 1.3129072649314582,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.1081,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 1.301387582143134,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.1206,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 1.2626386705712753,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.1099,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 1.1553731825862477,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.103,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.12967930734157562,
+      "eval_runtime": 3.0895,
+      "eval_samples_per_second": 17.479,
+      "eval_steps_per_second": 4.531,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 1.065580455906977,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.1069,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 1.327028003529359,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.107,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 1.37254525876708,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.1205,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 1.6929232722416325,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.1479,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 1.3985479543185082,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.1227,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 1.3758539900260853,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.1103,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 1.1178595855352185,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.0953,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 1.2957256622861373,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.0984,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 1.310467173353057,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.1054,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 1.377290467586534,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.114,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 1.5076421711434547,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.1435,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 1.3388348636928986,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.1035,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 1.4799603945613076,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.1429,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 1.2704026231514247,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.1061,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 1.1862406734398296,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.1071,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 1.406620345172403,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.1288,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 1.1878596676373772,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.1081,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 1.3641067758728074,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.1191,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 1.1106154494603295,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.0892,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 1.0816983427822033,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.0998,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 1.35883750999472,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.1261,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 0.9674965388082327,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.0864,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 1.4072462856509256,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.1125,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 1.284713147462951,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.1191,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 1.0947335684056199,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.0825,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 1.2042865179091564,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.1224,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 1.2047509721027605,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.0933,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 1.0383323401537132,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.0904,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 1.1996995241190327,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.1015,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 1.4384127114598833,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.1102,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 1.219530877147753,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.1077,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 1.450202457263629,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.1202,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 1.3836989557717199,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.1254,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 1.0597558251054786,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.0896,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 1.2911742155883357,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.1029,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 1.3720320814291684,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.1126,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 1.2479977184248061,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.1133,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 1.5290869247113479,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.1499,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 1.3132235581165603,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.1168,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 1.445614202104451,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.1348,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 1.2752382556832138,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.1153,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 1.1414541424472615,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.0945,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 1.300063322456091,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.1407,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 1.2987493093777953,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.1319,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 1.1902433956342071,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.115,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 1.2154609264223468,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.1311,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 1.319752199567177,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.1192,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 1.1624874260647156,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.0995,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 1.375822484596771,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.1242,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 1.2555502955005289,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.1151,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 1.312315646579941,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.1133,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 1.1709248874273042,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.1138,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 1.18842956079299,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.0974,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 1.1898669858310484,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.1037,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 1.0921788806360353,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.0947,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 1.2673382939672257,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.1074,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 1.5770917405981864,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.1273,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 1.4304614292027829,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.1169,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 1.0682578240286473,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.094,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 1.2158273784641755,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.109,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 1.1700547879968635,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.0915,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 1.4049543868813235,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.1231,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 1.1993613508104815,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.1047,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 1.4161154069156883,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.1211,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 1.5850870408361393,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.1303,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 1.1673487183383158,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.1045,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 1.2377064120966121,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.1189,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 1.070217103159899,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.0878,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 1.4467313684093264,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.1317,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 1.2098061669036835,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.0993,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 1.209007854178232,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.1089,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 1.6577102347405146,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.1152,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 1.1370792975487856,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.0878,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 1.2459900141839657,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.1222,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 1.166386386374148,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.0966,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 1.4592436747130215,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.1375,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 0.9058767706140431,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.0918,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 1.3636159946780668,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.1241,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 1.2656568926330312,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.0934,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 1.1998177762208129,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.1111,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 1.0685242966043493,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.0923,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 1.2420200768503447,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.1064,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 1.2231781433768538,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.1091,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 1.2286919567332113,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.1222,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 1.1699345759133055,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.0956,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 1.2056415822013686,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.0883,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 1.2386725315331428,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.1058,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 1.2066238987372266,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.1039,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 1.2650391555707365,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.1056,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 1.3555033868634538,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.1198,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 1.2763945939860821,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.113,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 1.267229285416109,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.1123,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 1.0571428184665541,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.1055,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 1.1644333364684327,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.0956,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 1.120803764871421,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.0992,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 1.3173415932050718,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.1211,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 1.2844048417685456,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.108,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 1.09761099611606,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.0873,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 1.1118798939959478,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.0801,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 1.260614478420532,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.1159,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 1.2212406882958637,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.0946,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 1.388222319247571,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.114,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 1.1781232188314859,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.0978,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 1.3596961475986553,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.1103,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 1.30934773971009,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.1289,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 1.0489116521766797,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.0869,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 1.3105450781964374,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.1251,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 1.1473018082471196,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.0939,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 1.2825455199084834,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.1117,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 1.284830506924204,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.1018,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 1.3042723321366816,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.1113,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 1.1561436130343512,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.0988,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 0.9679997087771331,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.0857,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 1.2411481535932931,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.1013,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 1.515372494033565,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.1322,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 1.4041642641020642,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.1196,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 1.2852246821319842,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.0929,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 1.4603455796680176,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.1223,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 1.2983355592368993,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.0987,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 1.3388890473946482,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.104,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 1.510871691089383,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.1218,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 1.4201242894147899,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.1176,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 1.3426630725757096,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.1082,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 1.3554095725651663,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.1107,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 1.2939458915390958,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.0962,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 1.2472643242597599,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.1061,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 1.3955089142605492,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.1081,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 1.3827646276718795,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.0998,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 1.3495699494205038,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.1121,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 1.422888480285114,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.1086,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 1.2069473481126878,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.1031,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 1.2465858227765634,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.0929,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 1.6773302268949093,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.1351,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1764333640634264,
+      "learning_rate": 5e-06,
+      "loss": 0.0656,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 0.9591773748961174,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.0485,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 0.9374046331333385,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.0535,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 0.8992445154687344,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.0462,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 0.9256237488836899,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.0466,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 0.8579816906266063,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.0363,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 0.9507349905736202,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.043,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 1.1574912336011127,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.061,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 0.9081158119985024,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.045,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 0.9943292693505907,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.0415,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 1.067137437862889,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.0426,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 0.9241069374461985,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.032,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 1.002590227303186,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.0435,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 1.0569475878654684,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.0442,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 1.034179354964817,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.0371,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 1.076898531791512,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.0318,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 1.378424733584512,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.0435,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 1.0080973497294576,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.031,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 1.461126872044986,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.0404,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 1.5214252786477502,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.0527,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 1.3467607596535511,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.0422,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 1.4740106324946352,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.0504,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 1.6357048795815177,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.0394,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 1.6677379705737272,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.0442,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 1.3637113291003893,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.0401,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 1.3428611336015295,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.0396,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 1.5840072766229634,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.0443,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 1.2916216009158379,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.0395,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 1.1694922421649272,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.0352,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 1.5598804595456244,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.0407,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 1.3520992927329525,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.0371,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 1.2607297449298212,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.0458,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 1.166724380593442,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.039,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 1.2345154832909562,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.0334,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 1.2420526184895606,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.0387,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 1.2327920658462928,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.0361,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 1.063450052569685,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.0376,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 1.2947639210219768,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.0464,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 0.9238295221153128,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.0328,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 1.07872896566137,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.0323,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 1.0645865372483225,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.0334,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 1.5309460143263094,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.0563,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 1.4305930283555823,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.0528,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 1.4952239916758296,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.0492,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 1.0472196028321745,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.0424,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 1.219435171969885,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.0454,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 1.179228695236265,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.0413,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 1.0726147730353988,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.0334,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 1.1871253780534199,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.0485,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 1.071206826714153,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.0328,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 1.148935166874607,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.0507,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 1.2768326425661778,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.0431,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 1.358267277686262,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.0527,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 1.056470434261961,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.0338,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 0.9456036834502879,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.0343,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 1.2373214620221722,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.0397,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 1.1049929403115635,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.0391,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 1.2211355905533328,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.0374,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 1.3281332286868575,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.0398,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 1.117064675718309,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.0438,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 1.2588865962481668,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.0419,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 1.047720522578905,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.0372,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 1.0198165289117536,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.0312,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 1.060946422387745,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.0345,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 1.1842316093877692,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.0443,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 1.0790452144413636,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.0325,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 0.9699560636684093,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.0295,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.13798944652080536,
+      "eval_runtime": 3.0003,
+      "eval_samples_per_second": 17.998,
+      "eval_steps_per_second": 4.666,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 1.2157901040016188,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.0388,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 1.1011569164981818,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.0418,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 1.3453983385960346,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.0425,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 0.9693367976098909,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.0291,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 1.1072393977245265,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.0368,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 1.1163705025975095,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.043,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 1.1474924081920497,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.0325,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 1.2177136093467429,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.0426,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 1.245825590269506,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.0423,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 1.0896827719347248,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.0339,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 1.0748167654963068,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.0327,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 1.2767252212918998,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.0382,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 1.0593689405289175,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.0351,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 1.0060911132798365,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.0341,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 1.0485894682724721,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.0374,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 1.112140510006677,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.0382,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 1.2356105633868555,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.0432,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 1.0693380658690446,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.0374,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 1.2266894995380884,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.0441,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 1.0334522926841525,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.0305,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 1.0044106171554246,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.0368,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 1.343075814675232,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.0493,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 1.0317706086830971,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.0354,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 1.0870462734104525,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.0427,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 1.2170346848236773,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.0411,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 1.112693619580994,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.0413,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 1.2146298323353943,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.0398,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 0.8361344449981266,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.0285,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 1.1015889573265605,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.0357,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 1.2083039828301432,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.0384,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 1.0436869874894652,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.0374,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 0.9982037545357402,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.0314,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 1.212377500536702,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.0355,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 1.1042534962758057,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.034,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 1.0988308100919006,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.0401,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 1.127879318877056,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.0362,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 1.1316571251440897,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.0481,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 1.4175433924312808,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.0441,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 1.2986700350012201,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.0399,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 1.1744660709965484,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.0449,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 1.0435506061683533,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.036,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 1.0672593300562783,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.0354,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 1.4632231966586535,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.0529,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 1.2739563204729878,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.0407,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 1.1721021342265219,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.0371,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 1.2704216459181281,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.046,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 1.1313625830454526,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.0361,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 1.2762991803360553,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.0488,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 1.1931406600656533,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.0463,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 1.2450015568406423,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.0481,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 1.596045947715624,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.0449,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 1.0964552738168878,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.0387,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 1.0230267909692887,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.0331,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 1.267947294526304,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.0436,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 0.9880183539432023,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.0322,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 0.8499951755281159,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.0301,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 1.1321955798598409,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.045,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 1.0337261269945672,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.0286,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 1.1966282076273769,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.0394,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 1.3504526102786338,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.0431,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 0.9250041744316801,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.0289,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 1.2231716617683348,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.0455,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 1.1708480586055061,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.0321,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 0.9579304660731348,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.0302,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 1.1583497393984776,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.0386,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 1.0025981554625683,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.0311,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 0.9681565825541392,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.0297,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 0.9732086719618618,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.0334,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 0.92117447258435,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.0254,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 1.0118946433644136,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.0299,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 1.2588404567213798,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.0353,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 0.9652767354648498,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.0257,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 1.2515765537637877,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.0431,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 1.168209684138088,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.0402,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 1.0300541914389187,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.0312,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 1.0531284632401012,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.0342,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 1.6277060182809362,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.0465,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 1.240369965483913,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.0473,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 1.0699895492447167,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.0324,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 0.977858486141618,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.0295,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 1.0476617605687077,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.0347,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 1.2919080065235906,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.0453,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 1.3126205748045743,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.0346,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 1.0498189253473171,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.0354,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 1.0386278025247024,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.0319,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 1.1519159992620087,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.0315,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 1.1592488315860394,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.034,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 1.0538284120421586,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.0313,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 1.0806832756516112,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.0321,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 1.4803970186446034,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.0446,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 0.9880629983123569,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.0303,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 1.3755678554613855,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.0445,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 1.2625940965764664,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.044,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 1.1334851067129665,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.0374,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 1.0673911446501338,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.0337,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 1.0134770989566584,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.0299,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 1.2711301207684582,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.0385,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 1.1024823361044216,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.0317,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 1.18705787317824,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.0344,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 1.0772570959843681,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.0403,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.1519404779429157,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.0394,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 1.575015101719698,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.0456,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 1.3361695430449418,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.0424,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 1.240006381856814,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.0357,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 0.9930051715536725,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.0293,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 1.0658321420361776,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.0296,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 1.1384997672106798,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.0334,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 1.1070162609179075,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.0295,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 1.2444977193878448,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.0352,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 1.1169607530262595,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.0392,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 1.1649042282237652,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.0339,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 3.413943083316713,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.0336,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 1.0026981434455031,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.0333,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 1.3118300005142567,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.0397,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 1.094099412297827,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.0396,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 1.1243988340159612,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.0368,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 0.9453994274327575,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.0346,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 1.1456827546312829,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.0339,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 1.0993373318375055,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.0313,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 1.3596704594564506,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.0443,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 1.1256649906521585,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.0352,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 1.333650064151969,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.04,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 0.9238036992945391,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.0289,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 0.9940584021668427,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.027,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 1.1706020027938457,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.0385,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 1.077765664104097,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.029,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 1.2937777148391536,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.0353,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 1.2895264888389124,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.0352,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 1.1347729271688634,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.0346,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 1.2056270012734296,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.0378,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 1.0623659938818895,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.0305,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 0.882252657138589,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.0236,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 1.3067948742028912,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.0384,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 1.3707377896515653,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.037,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 0.9879631966154709,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.0344,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 1.1455241013062318,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.0405,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 0.9966090470151359,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.0357,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 1.2631107884304436,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.0405,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 1.1365167972365535,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.0434,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 0.9869311814053642,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.0313,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 0.9633815500078319,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.0346,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 1.2415262468948147,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.0338,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 1.421669014497495,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.0423,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 1.342378684249825,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.0469,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 1.01744324466632,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.0333,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 1.237578503688093,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.0376,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 1.0648078673748254,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.0388,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 1.018205366667148,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.0325,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 1.398970003520616,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.0404,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 1.3286906737030206,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.0385,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 1.2913225106365636,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.0431,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 1.2557236785723636,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.0343,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 0.8489824328703612,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.0207,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 0.9560226281698921,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.027,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 1.1802609388787295,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.0342,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 1.1967680682094646,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.038,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 1.1321881389818194,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.0415,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 0.8909193374542834,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.0262,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 1.0360458473473435,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.0317,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 0.9822816556321587,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.0263,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 0.9205261123241077,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.0302,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 1.1747018405922898,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.033,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 1.1256499028526628,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.0308,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 1.3467843294327857,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.0459,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 1.2455735489942554,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.0329,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 1.0317324785476452,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.0324,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 1.0812572334491712,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.0343,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 1.0436513983249827,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.0348,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 1.0660809947972594,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.0314,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 0.9999043993899599,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.0278,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 1.007381844598099,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.0305,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 1.1068484801951521,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.0336,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 1.156166239873616,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.0329,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 1.1852460948932435,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.0435,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 0.9657539259584565,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.0254,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 1.217490605949571,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.0317,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 1.1246364865043605,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.0379,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 1.2669234012939086,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.0315,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 1.0774280888759886,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.0338,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 1.5048788937131414,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.0453,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 0.8798234433135917,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.03,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 1.0117273348921811,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.0286,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 1.226880515666306,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.041,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 1.0252838919928586,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.0353,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 1.0345973106751614,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.0355,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 1.0825092529776146,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.0336,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 1.3325555611612934,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.0379,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 1.0327210436772234,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.0283,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 0.9324717443374351,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.0266,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 1.1509840971480578,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.0285,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 1.1825825855126257,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.0336,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 1.0257676308637427,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.0321,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 0.8772805468243204,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.0312,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 1.1136249396818934,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.0416,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 1.1915920594277698,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.0375,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 1.0540008806526575,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.0399,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 1.088092405259284,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.0325,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 1.2421564761475292,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.0366,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 0.9843605102201873,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.0285,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 1.0611917676640774,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.0344,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.13630160689353943,
+      "eval_runtime": 3.0002,
+      "eval_samples_per_second": 17.999,
+      "eval_steps_per_second": 4.666,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 1.0254837671255463,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.0342,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 1.1185488013289164,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.0343,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 1.2165819567501461,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.0452,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 0.921514028260834,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.0295,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 1.1056877389483954,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.042,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 1.0842683090275778,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.0295,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 0.856004830299506,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.0255,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 1.2474827283991294,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.0351,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 0.8869188649866818,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.0264,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 1.087622261136807,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.0318,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 1.036886822247305,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.0307,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 0.8537492480060034,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.0266,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 0.9686903545696238,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.0262,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 1.2351093797551782,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.034,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 1.1359880114967853,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.0336,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 1.2546811268543112,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.041,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 1.0501435059460567,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.0337,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 1.0755762920997425,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.0304,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 0.9738445137165452,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.0317,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 1.1273565531079817,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.0312,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 1.1179764678350104,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.0381,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 0.8181898886855192,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.0296,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 1.2725591022482674,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.038,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 0.9577876624114283,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.0264,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 1.1427466994384856,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.0351,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 1.4278708709453982,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.0427,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 1.1155974357725371,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.0422,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 0.9464724047693186,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.0312,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 0.9352230331700522,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.0244,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 1.0271798722033365,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.027,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 1.1810520303163827,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.0371,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 1.072238146175478,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.0373,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 1.3077889631432282,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.0406,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 1.277786502775004,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.0467,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 1.1634586819915176,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.0305,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 1.0757768567081973,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.0265,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 1.1254750019458777,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.0328,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 1.12887702440565,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.0328,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 1.1089707127823831,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.034,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 1.1876665133385695,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.032,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 1.146708602119369,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.0387,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 1.3772982736443893,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.0401,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 1.4794421652526726,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.0411,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 0.8897873694561148,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.024,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 0.9671367122589538,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.0256,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 1.0745363771029608,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.0384,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 0.8922087795410587,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.0248,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 1.1503843633339363,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.0367,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 0.8190186679872811,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.028,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 0.9993180663513629,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.0296,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 1.248192895053933,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.0326,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 0.8362414001174541,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.0261,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 1.0057233898567999,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.0358,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 0.9592211775157188,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.0284,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 0.9789515762933076,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.0274,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 1.3287949107658452,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.0414,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 1.3357408849690622,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.0416,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 1.1525511079203234,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.0365,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 0.9761955983213899,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.0309,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 1.0964663350135253,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.0337,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 1.2339221436866648,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.0403,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 1.0618030136504184,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.037,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 1.4468559417880995,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.0408,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 1.2004533443353893,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.0358,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 1.0867612954659296,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.0257,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 1.0257866085203768,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.0281,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 1.0268089485525802,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.0341,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5715097361829105,
+      "learning_rate": 0.0,
+      "loss": 0.0116,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 12178659213312.0,
+      "train_loss": 0.07465152751130169,
+      "train_runtime": 886.8941,
+      "train_samples_per_second": 12.017,
+      "train_steps_per_second": 0.753
+    }
+  ],
+  "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": 12178659213312.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}