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": 1.827102976517607,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 0.1769,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 1.4900351957163886,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 0.1374,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 1.5414737247304657,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.1603,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 1.6168131777606338,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.1336,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 1.4017237291281952,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.1206,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 1.4562368217146875,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.1151,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 1.4571758092973182,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.1234,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 1.5410748639896168,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.1249,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 1.646427327961463,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.1354,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 1.7546307434959925,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.1227,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 1.4942883439251808,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.1296,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 1.569626020509824,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.1144,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 1.5695789720169004,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.142,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 1.5657073773866816,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.1339,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 1.5997577084572416,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.1208,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 1.4657267451608693,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.1286,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 1.1924476261927555,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.0945,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 1.3783860909739807,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.1334,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 1.7509745718596197,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.1462,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 1.4004831201605126,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.1375,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 1.387384879344789,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.1388,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 1.4108091172900128,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.1457,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 1.3128036257792943,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.1348,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 1.2579234506608277,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.1123,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 1.5135277878687383,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.1303,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 1.4882349889919755,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.126,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 1.3771160112417262,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.1366,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 1.3257814545341855,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.1327,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 1.5510329045346232,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.1344,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 1.2740589737209256,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.1282,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 1.1460948221705733,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.125,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 1.5765836785834393,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.1294,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 1.443059177605433,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.1117,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 1.4804039478298812,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.1414,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 1.371276421173181,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.1168,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 1.4979168893494963,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.1351,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 1.9576330464792342,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.1613,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 1.4465136405288177,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.1192,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 1.2352428817797585,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.1105,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 1.4996610899624405,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.1319,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 1.3631364481677741,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.1391,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 1.067164772844924,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.0927,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 1.3734962443860643,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.1159,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 1.279372593065496,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.1043,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 1.4190007057181424,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.1235,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 1.487507011279817,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.1287,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 1.2865071021447134,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.1147,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 1.4972457778402661,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.1358,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 1.4321136124582565,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.1475,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 1.3585707922460628,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.1182,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 1.462257099326064,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.1378,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 1.1818808861925425,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.1122,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 1.4480090564004418,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.1493,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 1.2044853521433176,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.117,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 1.3862314861336689,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.1227,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 1.3379778462076535,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.1291,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 1.4729587076949389,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.1348,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 1.5843065096721016,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.1928,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 1.4274397660051057,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.1285,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 1.3186924114475216,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.1261,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 1.6384484527725125,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.1576,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 1.3822817762555502,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.1227,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 1.2835568397131556,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.1011,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 1.2835452803812242,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.1339,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 1.3625203122755831,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.133,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 1.2595925859066146,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.1281,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 1.3112093359175865,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.1284,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 1.2508635632062837,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.1163,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 1.2407017759303838,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.1255,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 1.3224150583032743,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.1428,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 1.4724642848523672,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.1471,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 1.3568140158300623,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.0977,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 1.4847422362319815,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.1496,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 1.4947222374248261,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.1272,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 1.3943313801221096,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.1293,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 1.1904943855887953,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.0902,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 1.2563088692990063,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.1056,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 1.3135492202542542,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.1313,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 1.367839502636672,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.1256,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 1.713212221851054,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.1576,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 1.379544626429899,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.127,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 1.2320502579113293,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.1101,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 1.2317895633554645,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.118,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 1.2408095462553383,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.1071,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 1.1825505772955096,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.1049,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 1.3874370943477323,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.115,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 1.427916091299502,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.1238,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 1.5231201651716084,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.133,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 1.5057108219050812,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.1624,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 1.2103442002053861,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.1079,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 1.4012713662583858,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.1338,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 1.4258448060463964,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.1303,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 1.3050970932896557,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.1174,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 1.2721068381593104,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.1232,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 1.2122363489683141,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.1195,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 1.272688965614364,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.1373,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 1.1608508373179223,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.1133,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 1.2227691543432275,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.102,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 1.4448103216657495,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.133,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 1.3056797341013262,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.1091,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 1.2954225298377735,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.1063,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 1.2476554499065569,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.094,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 1.4876643486299068,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.1333,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 1.289932594566136,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.1222,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 1.2803463590427817,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.1237,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 1.303543244242864,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.1267,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 1.4704776262840162,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.1282,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 1.3966989936810759,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.136,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 1.5182556400633362,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.1263,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 1.426288811043197,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.1327,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 1.4330473684303955,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.153,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 1.518734602112281,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.1358,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 1.3353227082861636,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.146,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 1.3034887282579821,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.1248,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 1.3189515509141343,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.1252,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 1.238441009507851,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.1141,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 1.3819887561781907,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.1356,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 1.379897407057252,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.1178,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 1.325084518770577,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.1304,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 1.5465529707024976,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.1256,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 1.3769532645553098,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.1338,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 1.3569000782423215,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.1192,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 1.3515191702508695,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.1045,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 1.3961016118569598,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.1377,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 1.4281190964461064,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.1114,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 1.3556160791139888,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.1112,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 1.5072855399545235,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.1335,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 1.420897964549251,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.1404,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 1.407553589748203,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.1332,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 1.3669365973856973,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.1216,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 1.6793093757875397,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.1403,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 1.2646206096900132,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.1198,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 1.4866123875637047,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.1612,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 1.4439954786391624,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.1299,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 1.173455943092589,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.1082,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 1.2793917912310773,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.1087,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 1.6118422025200412,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.1683,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 2.0072119514619944,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.1247,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 1.365027057223322,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.1109,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 1.4180332004587342,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.1392,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 1.1487627783180734,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.111,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 1.2586689843164713,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.129,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 1.1990609072183336,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.1078,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 1.2259703125492027,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.1134,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 1.4924732772694669,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.1422,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 1.4199316308358614,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.1331,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 1.192555889955299,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.1063,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 1.3994224170990595,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.1341,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 1.9122572196396865,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.1213,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 1.2381123291685874,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.1284,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 1.3688652152360412,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.1293,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 1.411782317965158,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.1276,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 1.2639585880478126,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.1108,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 1.293666225876996,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.1234,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 1.338740545283697,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.1109,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 1.4284486454272576,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.1441,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 1.4382705051210485,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.1408,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 1.2135254671244131,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.1326,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 1.359762759073591,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.1382,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 1.2399091083753444,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.1081,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 1.4077295962176515,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.1336,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 1.2821608740834443,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.1149,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 1.4285235609505906,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.1231,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 1.4553378573796356,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.126,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 1.5087623264412504,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.1427,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 1.2723721161675168,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.1159,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.2350656639974504,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.1066,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 1.2031521640392466,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.1341,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 1.2853385166265359,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.0982,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 1.3183142946123474,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.1048,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 1.3628593335811692,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.1353,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 1.1896836439538132,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.1193,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 1.512670157148694,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.1259,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 1.3505254542304534,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.1156,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 1.3854715504848292,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.1199,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 1.300503752999626,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.1099,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 1.555164990359498,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.1334,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 1.727862267847299,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.1275,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 1.3949306691921821,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.1275,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 1.4587984882318334,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.1588,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 1.4862582792770003,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.1628,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 1.7456750531569738,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.1521,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 1.395792381764765,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.1334,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 1.283223711328585,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.1371,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 1.2998302518774747,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.1359,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 1.3685003966374945,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.1394,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 1.573910657577574,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.1611,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 1.2919101034020877,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.1336,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 1.2382183570095089,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.1338,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 1.1961438522291599,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.1151,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 1.2669776006611495,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.1168,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 1.535011146115232,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.1518,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 1.2478271570817967,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.1297,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 1.2307420985473116,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.1333,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 1.4671337397508504,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.1439,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 1.0950285569869143,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.0933,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 1.3540042453127357,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.113,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 1.4242721564024847,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.1305,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 1.3208104735851849,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.1246,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 1.1634198336022346,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.1128,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.14297430217266083,
+      "eval_runtime": 2.0044,
+      "eval_samples_per_second": 26.941,
+      "eval_steps_per_second": 6.985,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 1.109320048258132,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.1171,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 1.3663239028312661,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.1133,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 1.413312677670238,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.1301,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 1.7870659832009859,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.1653,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 1.3579978647451665,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.1317,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 1.3871245825301648,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.1222,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 1.2494878596786871,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.1124,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 1.336210175524145,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.1203,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 1.269646083754458,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.1195,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 1.4742339568956986,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.1315,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 1.3958142207648663,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.159,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 1.3093806502842773,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.1075,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 1.598004670462551,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.1595,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 1.2256473973220972,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.117,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 1.2806271879732039,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.1201,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 1.3621949929174726,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.1377,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 1.29685548466387,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.1187,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 1.3513075570515798,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.1318,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 1.1411939880770432,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.0977,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 1.1771016684217812,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.1155,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 1.3029619917857975,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.1334,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 0.9430597345704533,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.088,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 1.4896602146835947,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.1331,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 1.3191855731346847,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.1371,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 1.1609196374763002,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.099,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 1.3484940659705418,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.1439,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 1.2365706962862804,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.0982,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 1.0330594153069186,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.0932,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 1.3434995452184864,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.114,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 1.4401074637693385,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.1153,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 1.175104546742839,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.1152,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 1.4485388929293215,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.1337,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 1.3723597251128261,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.1322,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 1.0665247670846534,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.1049,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 1.2230258288585052,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.1115,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 1.4216976671597172,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.1214,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 1.353047224504761,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.1261,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 1.8099292942279634,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.1655,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 1.3384321749246484,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.1235,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 1.4610459813375962,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.144,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 1.3157990649806677,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.1215,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 1.1713637424596963,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.1061,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 1.3634719931958224,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.1591,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 1.3972352794564042,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.154,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 1.2833594463318787,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.1357,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 1.3502035380368025,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.1475,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 1.3776920758364728,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.1302,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 1.1621278044554224,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.1065,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 1.397452660620324,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.1343,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 1.3155846253018793,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.1275,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 1.3183550705098486,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.1321,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 1.224917894733324,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.1284,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 1.218331935334064,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.1122,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 1.2768054807233162,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.1176,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 1.1195908766466598,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.1138,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 1.2686117031298727,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.1204,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 1.4061981724908679,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.133,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 1.4330059777972,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.1282,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 1.1122942247003986,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.1051,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 1.4232549506510417,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.1304,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 1.1630929994247012,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.0985,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 1.356076556340411,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.1291,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 1.230492119460449,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.1129,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 1.4858926808094397,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.1299,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 1.6559686913140585,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.1497,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 1.267799299245616,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.1145,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 1.3598780337981418,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.1433,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 1.1478962659798289,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.1044,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 1.6986915130033347,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.158,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 1.1932599109406896,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.1145,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 1.2592062129958332,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.117,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 1.5924309951848903,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.122,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 1.359562103769234,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.1083,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 1.307291810505378,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.1477,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 1.2392919702946,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.1045,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 1.5284366178602025,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.1557,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 1.0216757015391396,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.1071,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 1.4463016884399356,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.1391,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 1.2813059234028101,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.1,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 1.2740167935275937,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.124,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 1.1459164752319326,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.1092,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 1.2332108196647178,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.114,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 1.3442317431647344,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.1247,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 1.2932384447975627,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.1352,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 1.2426438103067674,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.1119,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 1.3037351903379313,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.1002,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 1.2090427541566888,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.1179,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 1.3387257840755578,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.107,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 1.297034033227028,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.1227,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 1.358634930346158,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.1329,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 1.3679036744386044,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.1289,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 1.3089459026374033,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.1234,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 1.0926015166962475,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.117,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 1.2734725130779114,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.1088,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 1.2339151614301926,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.1152,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 1.3782146195515437,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.1323,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 1.2149573621573182,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.1184,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 1.125858012090774,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.0909,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 1.146297875204721,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.0835,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 1.378101851466363,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.1322,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 1.267016945947184,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.1035,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 1.474586234297009,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.1337,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 1.279395474681168,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.1021,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 1.3950878505963644,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.1262,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 1.4319211740409692,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.14,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 1.1055135412702288,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.0995,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 1.4363511462540821,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.1453,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 1.3403038278985364,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.1165,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 1.300420002905585,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.1286,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 1.3316531702575327,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.1144,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 1.3788425406563576,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.1218,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 1.1110715299647975,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.1076,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 1.1634228803167546,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.103,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 1.3397501884355985,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.122,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 1.5993829039322145,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.1436,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 1.4450763749856235,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.1332,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 1.2850882353246391,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.1031,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 1.4399103847373502,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.1467,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 1.2476042143545296,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.115,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 1.3275738170012203,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.1143,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 1.6642268682554573,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.1437,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 1.385058099781615,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.1255,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 1.3324694739758467,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.122,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 1.4372908790739205,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.1339,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 1.3294113203228832,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.1149,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 1.1929466160957578,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.1127,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 1.370945450514848,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.1207,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 1.366136576650825,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.1129,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 1.3417004420832874,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.1236,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 1.38737726700282,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.1201,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 1.2994492962324584,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.1146,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 1.1891068068331234,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.1012,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 1.67299707109756,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.1501,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9897327043686092,
+      "learning_rate": 5e-06,
+      "loss": 0.0581,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 0.9976953665357307,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.054,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 1.0415916258378042,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.064,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 0.9470684886589946,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.056,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 1.0198835296204702,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.0573,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 0.8666361437709743,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.0417,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 0.9443017223563431,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.0498,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 1.1449511337376854,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.0698,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 0.8914985355315801,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.0511,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 1.11285575939223,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.0509,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 0.9335266132576906,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.0507,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 0.9922989180860411,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.0409,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 1.287172838144389,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.057,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 1.4282487660235856,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.0639,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 1.1040591580380639,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.0455,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 1.0893791605522936,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.0399,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 1.2546148946446485,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.0488,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 1.0883234764895982,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.0481,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 1.8912048665251426,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.0526,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 1.4230914942529824,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.0678,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 1.2488115723543989,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.0587,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 1.2623957787918163,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.0617,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 1.2128361036543145,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.0344,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 1.609011929760213,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.0584,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 1.1545284882587372,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.0495,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 1.14185445151326,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.049,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 1.3918369330089968,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.0561,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 1.3267416876032634,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.053,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 1.2277113892475262,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.0503,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 1.3339401784861573,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.0458,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 1.4816306737370104,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.0524,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 1.2677432771857902,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.0516,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 1.1637591446365905,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.0515,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 1.2343586240798645,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.0442,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 1.1086407979371122,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.045,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 1.5176697748930295,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.0483,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 1.0831107837595422,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.0454,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 1.3838233636447876,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.0619,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 0.9690216151426481,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.0432,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 1.130515238263726,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.0403,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 1.0208873671732968,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.0444,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 1.3488942766579668,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.0583,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 1.2801046301358214,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.0613,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 1.429409737823447,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.0529,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 1.1100376933816927,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.059,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 1.2774578969736563,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.0644,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 1.2111603709288385,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.0571,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 1.014303020355431,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.0462,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 1.210592852832827,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.0642,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 0.9784520496128083,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.0381,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 1.1719832435633613,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.0707,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 1.2625916297991016,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.0527,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 1.3170727364565398,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.0735,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 1.1097596752701608,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.0467,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 1.0145673433454296,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.0441,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 1.3242962698920817,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.0557,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 1.1132636486546539,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.0504,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 1.2663745242083,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.0513,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 1.3808909559129485,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.0506,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 1.169040119446607,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.0514,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 1.3098414371197993,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.0554,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 1.0718349301790095,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.043,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 1.04727472377348,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.0398,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 0.995126407210452,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.0417,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 1.2553676809089644,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.0602,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 1.026457302269716,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.0395,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 0.9431981459105999,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.0424,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.149290531873703,
+      "eval_runtime": 1.9602,
+      "eval_samples_per_second": 27.549,
+      "eval_steps_per_second": 7.142,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 1.4092204366472383,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.0503,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 1.1669389869948825,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.0526,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 1.3821973350407255,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.0577,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 1.0372377482470008,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.0396,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 1.0612487090399547,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.0457,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 1.1746415167788675,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.0564,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 1.1128434225947956,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.0438,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 1.1926025665287554,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.058,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 1.3300449475121947,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.0546,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 1.07741748182952,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.0456,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 1.026238739692988,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.0401,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 1.2776975472954084,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.0479,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 1.0227515293253542,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.0394,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 0.9619245793531402,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.042,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 1.1097647186842046,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.0444,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 1.3724681304345978,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.0544,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 1.2786185531679677,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.0505,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 1.208230848843645,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.0517,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 1.032822996264697,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.0491,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 1.1358788661673513,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.0429,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 1.012404431909948,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.0497,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 1.3476275972851193,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.0547,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 1.120934681762057,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.0508,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 1.2358503737692217,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.0657,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 1.2905600904212207,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.0541,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 1.0862920179524578,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.0553,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 1.2897383431720169,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.0532,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 0.8676412832778311,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.0341,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 1.0070810750380685,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.0436,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 1.1683009001405902,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.0502,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 1.043980479409614,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.0457,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 1.0214915684450645,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.0419,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 1.2088578126380025,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.0432,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 1.0570131746245084,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.0415,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 1.1065715438511983,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.0447,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 1.0515327530651686,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.0478,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 1.080725851528956,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.0512,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 1.5069414156584842,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.0542,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 1.125286736937081,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.0449,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 1.1300051130239896,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.047,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 0.9848923094780573,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.0419,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 1.242376222642144,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.0498,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 1.203124506077074,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.0556,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 1.2373632429538983,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.0495,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 1.1015062758046326,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.0476,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 1.2407096310548562,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.0513,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 1.169452710104418,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.0481,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 1.242354320973221,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.0632,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 1.339047718155681,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.0579,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 1.427996809302124,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.0578,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 1.4849508009393295,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.0528,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 1.2447976871011308,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.0558,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 1.222163875205143,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.0475,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 1.2017743540961925,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.0491,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 1.0095706554643626,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.0355,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 0.9273039916236164,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.0397,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 1.200591192229917,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.0552,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 1.0280292927039731,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.0371,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 1.1933944187772714,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.0472,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 1.1878053727266924,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.0466,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 1.0342167480261764,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.042,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 1.2341385591172127,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.0547,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 1.0860786721919125,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.0372,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 0.9885587913914822,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.0406,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 1.4319089698843999,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.0564,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 1.0790115456190126,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.0354,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 1.0483477481084469,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.0408,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 1.039902775263006,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.0408,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 1.14239876119193,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.0425,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 1.1748778927777797,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.0451,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 1.2599511625637014,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.0425,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 0.9506039965786434,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.0317,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 1.2633062386226614,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.0534,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 1.3904315436738799,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.0558,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 1.107330463030352,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.0399,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 1.168130178816936,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.0509,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 1.5155722024708298,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.0556,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 1.179715134341272,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.0567,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 1.1393115683532504,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.0449,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 1.1550395233479744,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.0374,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 0.9455922267785334,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.0357,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 1.1152759151284144,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.0455,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 1.1987936928083571,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.0409,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 1.1802719007887685,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.0535,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 1.1288088192718482,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.0424,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 1.2762985504280422,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.0463,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 1.0068677208975914,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.0475,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 0.9567129102721482,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.0347,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 1.0660482938753442,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.0427,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 1.742885156678103,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.0653,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 1.3383714416798358,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.0493,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 1.2955159797477693,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.0554,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 1.1215701789819985,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.0532,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 1.2557443273198885,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.0521,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 1.201302306077298,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.0476,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 0.9957243694617502,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.0373,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 1.1971901178245348,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.0489,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 1.372062128143406,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.0502,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 1.30002061284066,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.0528,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 1.1248828551144325,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.0561,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2215053778619316,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.0519,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 1.400332475233556,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.0563,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 1.3975373826207649,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.0534,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 1.1767338996512713,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.0454,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 1.105583191131632,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.043,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 1.1338995496712656,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.0405,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 1.2010572939762567,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.0407,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 0.9848970741825713,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.0361,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 1.1500335536377977,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.0393,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 1.1285404843225988,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.0506,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 1.00714625153047,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.037,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 1.2520707988390285,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.0392,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 1.0883839883812527,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.0422,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 1.183754265409029,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.0516,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 1.1578309905021071,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.0543,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 1.085356063824633,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.0437,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 1.05468960428384,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.053,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 1.19559806291137,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.0395,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 1.1894024376760204,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.0464,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 1.3261500033401383,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.0597,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 1.0810523508570704,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.0431,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 1.4172021328353654,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.0509,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 1.1452718670380237,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.0404,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 1.0826039059752341,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.0402,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 1.2647662565281288,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.0488,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 1.185207623264876,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.0403,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 1.2619209228628094,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.0438,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 1.3053392292041395,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.048,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 1.2356510095256419,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.0509,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 1.0660616841452752,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.0393,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 0.971888364116127,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.0354,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 0.9967350959218624,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.0335,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 1.3817349598402762,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.0483,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 1.416899056597805,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.0554,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 1.1157691942763885,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.0504,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 1.2135328156852179,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.0476,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 1.1843133313498206,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.0487,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 1.1483443818535748,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.0482,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 1.1792968078135226,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.0611,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 0.9799695145596028,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.0378,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 1.0398479801118483,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.0511,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 1.1518364535580912,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.0392,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 1.2612913312995284,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.0547,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 1.4755757229597952,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.0677,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 1.1231859982771106,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.0481,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 1.5054238679806888,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.0492,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 1.0451705652605092,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.0458,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 1.0107141600031384,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.0425,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 1.3169901448503478,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.0485,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 1.2944639579592272,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.0513,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 1.332267209199536,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.0592,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 1.0858509392828761,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.0381,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 0.9336420846683142,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.0305,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 1.0141831957744358,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.0367,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 1.0750903205977453,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.0378,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 1.2486591610141904,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.0509,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 1.220366368344356,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.057,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 0.9612720166416755,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.0347,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 1.1354734458999889,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.0492,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 1.0450629490694525,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.0359,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 1.2147263683476315,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.0427,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 1.1570349713470343,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.0467,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 1.235551304989249,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.0433,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 1.4285503369923682,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.0552,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 1.3243924501000652,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.0434,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 1.149794681590079,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.0488,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 1.1131747717021003,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.0444,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 1.0847425381310065,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.0442,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 1.3505071122001504,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.0483,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 1.0785498303530907,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.0355,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 1.0920364629397974,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.0443,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 1.201435548336575,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.0496,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 1.3670816366047744,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.0441,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 1.244940033171781,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.0599,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 1.0839204008805932,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.041,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 1.2298583538941907,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.0402,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 1.3122928750912277,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.0511,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 1.1206233471110985,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.0367,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 1.019566942167285,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.0455,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 1.3305736651360114,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.0543,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 0.9428402773729421,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.0361,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 1.1745558133467238,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.0436,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 1.126114851123789,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.0498,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 0.9589301793058848,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.0376,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 1.0727070362608244,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.0515,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 1.0189538585823592,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.0465,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 1.3828429297424205,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.055,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 1.0034334730904557,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.0337,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 0.9705752329034909,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.0399,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 1.1189080408285355,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.0352,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 1.3774526791987014,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.0491,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 1.0586561558503529,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.0448,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 0.9335756724791543,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.0353,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 1.2168330888259584,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.0526,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 1.3077585624557646,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.052,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 1.047398320179946,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.051,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 1.1163489674378195,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.04,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 1.2454180858846498,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.0436,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 1.0340427846027735,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.04,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 0.9749792264132685,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.0438,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.14543114602565765,
+      "eval_runtime": 1.9558,
+      "eval_samples_per_second": 27.61,
+      "eval_steps_per_second": 7.158,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 1.047996015971046,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.0442,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 1.3030958640619468,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.0583,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 1.3270589919647697,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.0598,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 1.013251267830963,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.0391,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 1.1534262545305207,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.058,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 1.1802094090539037,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.039,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 0.9472523626592134,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.0387,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 1.4818128557218848,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.0508,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 1.0467961216564687,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.0376,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 1.0764032268836488,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.04,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 1.1060550923230614,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.04,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 0.9341280679120579,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.0394,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 1.2695782031152785,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.0449,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 1.4055472317071316,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.0562,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 1.1609370975777964,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.0471,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 1.3705462219042284,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.048,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 1.160873906029491,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.046,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 1.0254503554825711,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.0411,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 1.0385472193179268,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.0423,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 1.2255602112842634,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.0438,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 1.1655896262830023,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.0468,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 0.8407578267039898,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.0411,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 1.1544973218407946,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.0456,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 1.086352956864287,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.0377,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 1.1179831471438937,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.0435,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 1.3206751638177998,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.0481,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 1.0932854220378916,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.0523,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 1.0540986194623885,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.0417,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 1.0171417000481298,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.0356,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 0.9800841643222271,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.0296,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 1.0819185314215787,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.0489,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 1.0959832497603015,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.0527,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 1.217301817185727,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.0517,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 1.574138996873441,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.0654,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 1.0986199043994893,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.0397,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 1.1118235484465493,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.0361,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 1.2045637785003465,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.0479,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 0.9573678876555733,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.0326,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 1.3940778713848283,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.0475,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 1.1859603268260792,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.0418,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 1.0991880727957932,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.0478,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 1.3381955488973505,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.0522,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 1.348504135490946,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.0524,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 1.0099344548785596,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.0381,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 0.9947959260374297,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.0333,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 1.06348752752877,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.0454,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 0.981299298003997,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.0369,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 1.2749148288845744,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.0484,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 0.8187561336465383,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.0347,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 0.9974647287360009,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.0395,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 1.3259319672672527,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.0442,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 0.991639430244018,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.0346,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 1.1807221282537466,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.056,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 1.0482211133990704,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.0389,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 0.9150125394733329,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.0418,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 1.2263558139972717,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.0518,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 1.2063278282555905,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.0502,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 1.2934623938286378,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.0471,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 1.0284033480881405,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.0403,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 1.135097218354692,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.0472,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 1.2674967396350294,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.0585,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 1.0498640309274887,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.0488,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 1.321544179671229,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.053,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 1.410789822096634,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.0531,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 1.0454626732028163,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.0387,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 1.0295755666261253,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.0402,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 1.1139849051443895,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.0501,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6909971645457605,
+      "learning_rate": 0.0,
+      "loss": 0.0188,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 7302035079168.0,
+      "train_loss": 0.08638812355909073,
+      "train_runtime": 520.9989,
+      "train_samples_per_second": 20.457,
+      "train_steps_per_second": 1.282
+    }
+  ],
+  "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": 7302035079168.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}