Qwen2.5-7B-Instruct_Short_CoT / trainer_state.json
TaiGary's picture
Add files using upload-large-folder tool
3ccac74 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 200,
"global_step": 668,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0029940119760479044,
"grad_norm": 2.694822704587142,
"learning_rate": 9.999944704978835e-06,
"loss": 0.174,
"step": 1
},
{
"epoch": 0.005988023952095809,
"grad_norm": 1.540694377428953,
"learning_rate": 9.999778821138357e-06,
"loss": 0.1031,
"step": 2
},
{
"epoch": 0.008982035928143712,
"grad_norm": 1.6650818674344259,
"learning_rate": 9.999502352147583e-06,
"loss": 0.1121,
"step": 3
},
{
"epoch": 0.011976047904191617,
"grad_norm": 1.7580445000808085,
"learning_rate": 9.999115304121459e-06,
"loss": 0.1051,
"step": 4
},
{
"epoch": 0.014970059880239521,
"grad_norm": 1.7432661703660053,
"learning_rate": 9.998617685620715e-06,
"loss": 0.1052,
"step": 5
},
{
"epoch": 0.017964071856287425,
"grad_norm": 1.5622913387439772,
"learning_rate": 9.998009507651683e-06,
"loss": 0.1033,
"step": 6
},
{
"epoch": 0.020958083832335328,
"grad_norm": 1.534172674280163,
"learning_rate": 9.997290783666048e-06,
"loss": 0.1035,
"step": 7
},
{
"epoch": 0.023952095808383235,
"grad_norm": 1.4876532660064046,
"learning_rate": 9.996461529560553e-06,
"loss": 0.0952,
"step": 8
},
{
"epoch": 0.02694610778443114,
"grad_norm": 1.5065957449770506,
"learning_rate": 9.995521763676645e-06,
"loss": 0.1145,
"step": 9
},
{
"epoch": 0.029940119760479042,
"grad_norm": 1.3513837861157794,
"learning_rate": 9.994471506800078e-06,
"loss": 0.0975,
"step": 10
},
{
"epoch": 0.03293413173652695,
"grad_norm": 1.4058269709713151,
"learning_rate": 9.993310782160439e-06,
"loss": 0.1068,
"step": 11
},
{
"epoch": 0.03592814371257485,
"grad_norm": 1.3803877704214274,
"learning_rate": 9.992039615430648e-06,
"loss": 0.0921,
"step": 12
},
{
"epoch": 0.038922155688622756,
"grad_norm": 1.3945715898711406,
"learning_rate": 9.99065803472638e-06,
"loss": 0.1114,
"step": 13
},
{
"epoch": 0.041916167664670656,
"grad_norm": 1.4546816139508025,
"learning_rate": 9.989166070605447e-06,
"loss": 0.1072,
"step": 14
},
{
"epoch": 0.04491017964071856,
"grad_norm": 1.480351599059349,
"learning_rate": 9.98756375606713e-06,
"loss": 0.1,
"step": 15
},
{
"epoch": 0.04790419161676647,
"grad_norm": 1.46275915665189,
"learning_rate": 9.985851126551428e-06,
"loss": 0.113,
"step": 16
},
{
"epoch": 0.05089820359281437,
"grad_norm": 1.0722369900557316,
"learning_rate": 9.9840282199383e-06,
"loss": 0.0803,
"step": 17
},
{
"epoch": 0.05389221556886228,
"grad_norm": 1.3430666265845805,
"learning_rate": 9.982095076546806e-06,
"loss": 0.1182,
"step": 18
},
{
"epoch": 0.05688622754491018,
"grad_norm": 1.561360979226197,
"learning_rate": 9.980051739134235e-06,
"loss": 0.1272,
"step": 19
},
{
"epoch": 0.059880239520958084,
"grad_norm": 1.2716064252529893,
"learning_rate": 9.977898252895133e-06,
"loss": 0.1182,
"step": 20
},
{
"epoch": 0.06287425149700598,
"grad_norm": 1.2433276467135879,
"learning_rate": 9.975634665460333e-06,
"loss": 0.1136,
"step": 21
},
{
"epoch": 0.0658682634730539,
"grad_norm": 1.4562587683694241,
"learning_rate": 9.973261026895878e-06,
"loss": 0.1193,
"step": 22
},
{
"epoch": 0.0688622754491018,
"grad_norm": 1.2046318508865133,
"learning_rate": 9.970777389701927e-06,
"loss": 0.1127,
"step": 23
},
{
"epoch": 0.0718562874251497,
"grad_norm": 1.1209438570730352,
"learning_rate": 9.968183808811586e-06,
"loss": 0.0894,
"step": 24
},
{
"epoch": 0.0748502994011976,
"grad_norm": 1.4258740274999953,
"learning_rate": 9.965480341589702e-06,
"loss": 0.1086,
"step": 25
},
{
"epoch": 0.07784431137724551,
"grad_norm": 1.348457039378542,
"learning_rate": 9.962667047831585e-06,
"loss": 0.1076,
"step": 26
},
{
"epoch": 0.08083832335329341,
"grad_norm": 1.3067044062679876,
"learning_rate": 9.95974398976169e-06,
"loss": 0.1162,
"step": 27
},
{
"epoch": 0.08383233532934131,
"grad_norm": 1.280806605368862,
"learning_rate": 9.95671123203224e-06,
"loss": 0.1143,
"step": 28
},
{
"epoch": 0.08682634730538923,
"grad_norm": 1.4042111584668944,
"learning_rate": 9.953568841721796e-06,
"loss": 0.1154,
"step": 29
},
{
"epoch": 0.08982035928143713,
"grad_norm": 1.2535340127099879,
"learning_rate": 9.950316888333775e-06,
"loss": 0.1125,
"step": 30
},
{
"epoch": 0.09281437125748503,
"grad_norm": 1.0560186599040056,
"learning_rate": 9.946955443794908e-06,
"loss": 0.0976,
"step": 31
},
{
"epoch": 0.09580838323353294,
"grad_norm": 1.4146974136560293,
"learning_rate": 9.943484582453653e-06,
"loss": 0.1072,
"step": 32
},
{
"epoch": 0.09880239520958084,
"grad_norm": 1.4941301730987362,
"learning_rate": 9.939904381078553e-06,
"loss": 0.102,
"step": 33
},
{
"epoch": 0.10179640718562874,
"grad_norm": 1.413313808359942,
"learning_rate": 9.93621491885653e-06,
"loss": 0.1153,
"step": 34
},
{
"epoch": 0.10479041916167664,
"grad_norm": 1.2850277969498978,
"learning_rate": 9.932416277391144e-06,
"loss": 0.103,
"step": 35
},
{
"epoch": 0.10778443113772455,
"grad_norm": 1.2903830996306178,
"learning_rate": 9.928508540700775e-06,
"loss": 0.1235,
"step": 36
},
{
"epoch": 0.11077844311377245,
"grad_norm": 1.5326088245895257,
"learning_rate": 9.924491795216777e-06,
"loss": 0.1329,
"step": 37
},
{
"epoch": 0.11377245508982035,
"grad_norm": 1.3215649330381873,
"learning_rate": 9.920366129781564e-06,
"loss": 0.1159,
"step": 38
},
{
"epoch": 0.11676646706586827,
"grad_norm": 1.042094511765466,
"learning_rate": 9.916131635646635e-06,
"loss": 0.0936,
"step": 39
},
{
"epoch": 0.11976047904191617,
"grad_norm": 1.3770014683229508,
"learning_rate": 9.91178840647057e-06,
"loss": 0.1272,
"step": 40
},
{
"epoch": 0.12275449101796407,
"grad_norm": 1.1931812346620854,
"learning_rate": 9.907336538316946e-06,
"loss": 0.1146,
"step": 41
},
{
"epoch": 0.12574850299401197,
"grad_norm": 0.9277316058880373,
"learning_rate": 9.902776129652223e-06,
"loss": 0.0784,
"step": 42
},
{
"epoch": 0.12874251497005987,
"grad_norm": 1.0765649656661282,
"learning_rate": 9.898107281343557e-06,
"loss": 0.1004,
"step": 43
},
{
"epoch": 0.1317365269461078,
"grad_norm": 0.9926609945457437,
"learning_rate": 9.893330096656576e-06,
"loss": 0.0881,
"step": 44
},
{
"epoch": 0.1347305389221557,
"grad_norm": 1.222816772091786,
"learning_rate": 9.888444681253087e-06,
"loss": 0.1075,
"step": 45
},
{
"epoch": 0.1377245508982036,
"grad_norm": 1.2894461166853382,
"learning_rate": 9.883451143188753e-06,
"loss": 0.113,
"step": 46
},
{
"epoch": 0.1407185628742515,
"grad_norm": 1.121235366858226,
"learning_rate": 9.878349592910694e-06,
"loss": 0.0981,
"step": 47
},
{
"epoch": 0.1437125748502994,
"grad_norm": 1.273548714881534,
"learning_rate": 9.873140143255035e-06,
"loss": 0.1205,
"step": 48
},
{
"epoch": 0.1467065868263473,
"grad_norm": 1.2866045456682342,
"learning_rate": 9.867822909444435e-06,
"loss": 0.1119,
"step": 49
},
{
"epoch": 0.1497005988023952,
"grad_norm": 1.2783141532135376,
"learning_rate": 9.862398009085511e-06,
"loss": 0.1023,
"step": 50
},
{
"epoch": 0.15269461077844312,
"grad_norm": 1.3325991624769549,
"learning_rate": 9.856865562166256e-06,
"loss": 0.1155,
"step": 51
},
{
"epoch": 0.15568862275449102,
"grad_norm": 1.0915358634542538,
"learning_rate": 9.851225691053382e-06,
"loss": 0.0946,
"step": 52
},
{
"epoch": 0.15868263473053892,
"grad_norm": 1.3812817768812469,
"learning_rate": 9.8454785204896e-06,
"loss": 0.1246,
"step": 53
},
{
"epoch": 0.16167664670658682,
"grad_norm": 1.1711427977372355,
"learning_rate": 9.83962417759088e-06,
"loss": 0.1038,
"step": 54
},
{
"epoch": 0.16467065868263472,
"grad_norm": 1.3270361072749957,
"learning_rate": 9.833662791843628e-06,
"loss": 0.1166,
"step": 55
},
{
"epoch": 0.16766467065868262,
"grad_norm": 1.1961374353611072,
"learning_rate": 9.827594495101824e-06,
"loss": 0.1082,
"step": 56
},
{
"epoch": 0.17065868263473055,
"grad_norm": 1.2508456378463604,
"learning_rate": 9.821419421584108e-06,
"loss": 0.109,
"step": 57
},
{
"epoch": 0.17365269461077845,
"grad_norm": 1.5853611337924214,
"learning_rate": 9.815137707870806e-06,
"loss": 0.1481,
"step": 58
},
{
"epoch": 0.17664670658682635,
"grad_norm": 1.3533294342793396,
"learning_rate": 9.808749492900917e-06,
"loss": 0.1257,
"step": 59
},
{
"epoch": 0.17964071856287425,
"grad_norm": 1.3361681014706468,
"learning_rate": 9.802254917969033e-06,
"loss": 0.1062,
"step": 60
},
{
"epoch": 0.18263473053892215,
"grad_norm": 1.5134202209845473,
"learning_rate": 9.795654126722218e-06,
"loss": 0.1307,
"step": 61
},
{
"epoch": 0.18562874251497005,
"grad_norm": 1.1706733413997306,
"learning_rate": 9.788947265156828e-06,
"loss": 0.1035,
"step": 62
},
{
"epoch": 0.18862275449101795,
"grad_norm": 1.0889556046235287,
"learning_rate": 9.782134481615282e-06,
"loss": 0.0987,
"step": 63
},
{
"epoch": 0.19161676646706588,
"grad_norm": 1.2398715014649333,
"learning_rate": 9.775215926782788e-06,
"loss": 0.1188,
"step": 64
},
{
"epoch": 0.19461077844311378,
"grad_norm": 1.3618169558692828,
"learning_rate": 9.768191753683997e-06,
"loss": 0.1125,
"step": 65
},
{
"epoch": 0.19760479041916168,
"grad_norm": 1.1704468214057708,
"learning_rate": 9.761062117679632e-06,
"loss": 0.1136,
"step": 66
},
{
"epoch": 0.20059880239520958,
"grad_norm": 1.1182850488025071,
"learning_rate": 9.75382717646304e-06,
"loss": 0.1044,
"step": 67
},
{
"epoch": 0.20359281437125748,
"grad_norm": 1.0346184105202574,
"learning_rate": 9.746487090056712e-06,
"loss": 0.0991,
"step": 68
},
{
"epoch": 0.20658682634730538,
"grad_norm": 1.062900689169356,
"learning_rate": 9.739042020808746e-06,
"loss": 0.1024,
"step": 69
},
{
"epoch": 0.20958083832335328,
"grad_norm": 1.0911542026965957,
"learning_rate": 9.73149213338924e-06,
"loss": 0.1136,
"step": 70
},
{
"epoch": 0.2125748502994012,
"grad_norm": 1.2496717021456991,
"learning_rate": 9.72383759478667e-06,
"loss": 0.1207,
"step": 71
},
{
"epoch": 0.2155688622754491,
"grad_norm": 1.080099379875001,
"learning_rate": 9.71607857430419e-06,
"loss": 0.0899,
"step": 72
},
{
"epoch": 0.218562874251497,
"grad_norm": 1.4072414605443209,
"learning_rate": 9.708215243555875e-06,
"loss": 0.1291,
"step": 73
},
{
"epoch": 0.2215568862275449,
"grad_norm": 1.1700638225577527,
"learning_rate": 9.700247776462944e-06,
"loss": 0.103,
"step": 74
},
{
"epoch": 0.2245508982035928,
"grad_norm": 1.2565934320808063,
"learning_rate": 9.6921763492499e-06,
"loss": 0.1099,
"step": 75
},
{
"epoch": 0.2275449101796407,
"grad_norm": 0.9546525733169299,
"learning_rate": 9.68400114044064e-06,
"loss": 0.0804,
"step": 76
},
{
"epoch": 0.23053892215568864,
"grad_norm": 1.08478464224836,
"learning_rate": 9.6757223308545e-06,
"loss": 0.0939,
"step": 77
},
{
"epoch": 0.23353293413173654,
"grad_norm": 1.3415590730099887,
"learning_rate": 9.667340103602263e-06,
"loss": 0.1208,
"step": 78
},
{
"epoch": 0.23652694610778444,
"grad_norm": 1.1986328154053258,
"learning_rate": 9.658854644082099e-06,
"loss": 0.1046,
"step": 79
},
{
"epoch": 0.23952095808383234,
"grad_norm": 1.7151828135588658,
"learning_rate": 9.650266139975474e-06,
"loss": 0.1399,
"step": 80
},
{
"epoch": 0.24251497005988024,
"grad_norm": 1.1444934755253402,
"learning_rate": 9.641574781242999e-06,
"loss": 0.106,
"step": 81
},
{
"epoch": 0.24550898203592814,
"grad_norm": 1.0474417551763626,
"learning_rate": 9.632780760120217e-06,
"loss": 0.0957,
"step": 82
},
{
"epoch": 0.24850299401197604,
"grad_norm": 1.0019186144881402,
"learning_rate": 9.62388427111336e-06,
"loss": 0.0952,
"step": 83
},
{
"epoch": 0.25149700598802394,
"grad_norm": 1.045782599298007,
"learning_rate": 9.614885510995047e-06,
"loss": 0.098,
"step": 84
},
{
"epoch": 0.25449101796407186,
"grad_norm": 1.0596517738810545,
"learning_rate": 9.605784678799934e-06,
"loss": 0.0939,
"step": 85
},
{
"epoch": 0.25748502994011974,
"grad_norm": 1.3158738779872337,
"learning_rate": 9.596581975820304e-06,
"loss": 0.107,
"step": 86
},
{
"epoch": 0.26047904191616766,
"grad_norm": 1.1584081147238108,
"learning_rate": 9.587277605601617e-06,
"loss": 0.1091,
"step": 87
},
{
"epoch": 0.2634730538922156,
"grad_norm": 1.3742865320327458,
"learning_rate": 9.577871773938013e-06,
"loss": 0.1253,
"step": 88
},
{
"epoch": 0.26646706586826346,
"grad_norm": 1.3133549998241743,
"learning_rate": 9.568364688867757e-06,
"loss": 0.1304,
"step": 89
},
{
"epoch": 0.2694610778443114,
"grad_norm": 0.9687727348118643,
"learning_rate": 9.558756560668637e-06,
"loss": 0.0938,
"step": 90
},
{
"epoch": 0.27245508982035926,
"grad_norm": 1.1880519099821893,
"learning_rate": 9.549047601853313e-06,
"loss": 0.1184,
"step": 91
},
{
"epoch": 0.2754491017964072,
"grad_norm": 1.1641235380588761,
"learning_rate": 9.539238027164618e-06,
"loss": 0.1211,
"step": 92
},
{
"epoch": 0.27844311377245506,
"grad_norm": 0.993560219725402,
"learning_rate": 9.52932805357081e-06,
"loss": 0.0986,
"step": 93
},
{
"epoch": 0.281437125748503,
"grad_norm": 1.0050842703408378,
"learning_rate": 9.519317900260769e-06,
"loss": 0.1076,
"step": 94
},
{
"epoch": 0.2844311377245509,
"grad_norm": 1.0082192441585922,
"learning_rate": 9.509207788639148e-06,
"loss": 0.1075,
"step": 95
},
{
"epoch": 0.2874251497005988,
"grad_norm": 1.0364990627964445,
"learning_rate": 9.498997942321484e-06,
"loss": 0.1163,
"step": 96
},
{
"epoch": 0.2904191616766467,
"grad_norm": 1.0083859812100648,
"learning_rate": 9.488688587129243e-06,
"loss": 0.1046,
"step": 97
},
{
"epoch": 0.2934131736526946,
"grad_norm": 0.9802364376711591,
"learning_rate": 9.47827995108483e-06,
"loss": 0.0908,
"step": 98
},
{
"epoch": 0.2964071856287425,
"grad_norm": 1.2296567986616285,
"learning_rate": 9.467772264406545e-06,
"loss": 0.1142,
"step": 99
},
{
"epoch": 0.2994011976047904,
"grad_norm": 0.9944404277217507,
"learning_rate": 9.457165759503492e-06,
"loss": 0.0933,
"step": 100
},
{
"epoch": 0.3023952095808383,
"grad_norm": 1.1318248455399238,
"learning_rate": 9.446460670970436e-06,
"loss": 0.0995,
"step": 101
},
{
"epoch": 0.30538922155688625,
"grad_norm": 1.0856325033453054,
"learning_rate": 9.435657235582616e-06,
"loss": 0.0926,
"step": 102
},
{
"epoch": 0.3083832335329341,
"grad_norm": 1.2935359839355358,
"learning_rate": 9.424755692290507e-06,
"loss": 0.1179,
"step": 103
},
{
"epoch": 0.31137724550898205,
"grad_norm": 1.1697314711229891,
"learning_rate": 9.413756282214538e-06,
"loss": 0.1096,
"step": 104
},
{
"epoch": 0.3143712574850299,
"grad_norm": 1.039674731853863,
"learning_rate": 9.402659248639749e-06,
"loss": 0.1061,
"step": 105
},
{
"epoch": 0.31736526946107785,
"grad_norm": 1.21813072366842,
"learning_rate": 9.391464837010428e-06,
"loss": 0.1128,
"step": 106
},
{
"epoch": 0.3203592814371258,
"grad_norm": 1.2347699420503069,
"learning_rate": 9.380173294924661e-06,
"loss": 0.114,
"step": 107
},
{
"epoch": 0.32335329341317365,
"grad_norm": 1.0952193810495732,
"learning_rate": 9.368784872128877e-06,
"loss": 0.1197,
"step": 108
},
{
"epoch": 0.3263473053892216,
"grad_norm": 1.1988107428883443,
"learning_rate": 9.357299820512305e-06,
"loss": 0.1097,
"step": 109
},
{
"epoch": 0.32934131736526945,
"grad_norm": 1.3238374098387835,
"learning_rate": 9.345718394101412e-06,
"loss": 0.1168,
"step": 110
},
{
"epoch": 0.3323353293413174,
"grad_norm": 1.154675757029323,
"learning_rate": 9.334040849054288e-06,
"loss": 0.1221,
"step": 111
},
{
"epoch": 0.33532934131736525,
"grad_norm": 1.1765490559607166,
"learning_rate": 9.322267443654974e-06,
"loss": 0.1179,
"step": 112
},
{
"epoch": 0.3383233532934132,
"grad_norm": 1.2072662482322083,
"learning_rate": 9.310398438307747e-06,
"loss": 0.1337,
"step": 113
},
{
"epoch": 0.3413173652694611,
"grad_norm": 1.1674159812190554,
"learning_rate": 9.29843409553137e-06,
"loss": 0.1142,
"step": 114
},
{
"epoch": 0.344311377245509,
"grad_norm": 1.0999888356981997,
"learning_rate": 9.286374679953278e-06,
"loss": 0.1058,
"step": 115
},
{
"epoch": 0.3473053892215569,
"grad_norm": 0.9399577245375,
"learning_rate": 9.274220458303727e-06,
"loss": 0.0989,
"step": 116
},
{
"epoch": 0.3502994011976048,
"grad_norm": 1.140609139219963,
"learning_rate": 9.261971699409893e-06,
"loss": 0.1144,
"step": 117
},
{
"epoch": 0.3532934131736527,
"grad_norm": 1.2038276135727357,
"learning_rate": 9.249628674189928e-06,
"loss": 0.1099,
"step": 118
},
{
"epoch": 0.3562874251497006,
"grad_norm": 1.1233255402091744,
"learning_rate": 9.237191655646972e-06,
"loss": 0.1142,
"step": 119
},
{
"epoch": 0.3592814371257485,
"grad_norm": 1.1927948048157466,
"learning_rate": 9.224660918863104e-06,
"loss": 0.111,
"step": 120
},
{
"epoch": 0.36227544910179643,
"grad_norm": 1.1248820119091534,
"learning_rate": 9.212036740993265e-06,
"loss": 0.1167,
"step": 121
},
{
"epoch": 0.3652694610778443,
"grad_norm": 1.0535201928601414,
"learning_rate": 9.199319401259132e-06,
"loss": 0.0988,
"step": 122
},
{
"epoch": 0.36826347305389223,
"grad_norm": 1.0549521131159716,
"learning_rate": 9.186509180942928e-06,
"loss": 0.0921,
"step": 123
},
{
"epoch": 0.3712574850299401,
"grad_norm": 1.1359154866076464,
"learning_rate": 9.173606363381218e-06,
"loss": 0.1157,
"step": 124
},
{
"epoch": 0.37425149700598803,
"grad_norm": 1.1451306175535456,
"learning_rate": 9.16061123395863e-06,
"loss": 0.1043,
"step": 125
},
{
"epoch": 0.3772455089820359,
"grad_norm": 1.1181962731375765,
"learning_rate": 9.147524080101543e-06,
"loss": 0.0981,
"step": 126
},
{
"epoch": 0.38023952095808383,
"grad_norm": 1.2551946563935505,
"learning_rate": 9.134345191271742e-06,
"loss": 0.1197,
"step": 127
},
{
"epoch": 0.38323353293413176,
"grad_norm": 1.1583239194903672,
"learning_rate": 9.121074858959997e-06,
"loss": 0.124,
"step": 128
},
{
"epoch": 0.38622754491017963,
"grad_norm": 1.196120284188902,
"learning_rate": 9.107713376679634e-06,
"loss": 0.121,
"step": 129
},
{
"epoch": 0.38922155688622756,
"grad_norm": 1.1742439607970117,
"learning_rate": 9.094261039960028e-06,
"loss": 0.1,
"step": 130
},
{
"epoch": 0.39221556886227543,
"grad_norm": 1.292026851324083,
"learning_rate": 9.08071814634008e-06,
"loss": 0.1176,
"step": 131
},
{
"epoch": 0.39520958083832336,
"grad_norm": 1.1451954562828406,
"learning_rate": 9.067084995361623e-06,
"loss": 0.1097,
"step": 132
},
{
"epoch": 0.39820359281437123,
"grad_norm": 1.207998049376194,
"learning_rate": 9.053361888562807e-06,
"loss": 0.1213,
"step": 133
},
{
"epoch": 0.40119760479041916,
"grad_norm": 1.2132625284976668,
"learning_rate": 9.039549129471423e-06,
"loss": 0.1186,
"step": 134
},
{
"epoch": 0.4041916167664671,
"grad_norm": 0.9639736738335528,
"learning_rate": 9.025647023598196e-06,
"loss": 0.0944,
"step": 135
},
{
"epoch": 0.40718562874251496,
"grad_norm": 1.032185303357511,
"learning_rate": 9.011655878430018e-06,
"loss": 0.1012,
"step": 136
},
{
"epoch": 0.4101796407185629,
"grad_norm": 1.3933864089957346,
"learning_rate": 8.99757600342316e-06,
"loss": 0.1462,
"step": 137
},
{
"epoch": 0.41317365269461076,
"grad_norm": 1.0900174358692771,
"learning_rate": 8.983407709996415e-06,
"loss": 0.0983,
"step": 138
},
{
"epoch": 0.4161676646706587,
"grad_norm": 0.9682241109471825,
"learning_rate": 8.969151311524215e-06,
"loss": 0.0947,
"step": 139
},
{
"epoch": 0.41916167664670656,
"grad_norm": 1.2153257975847114,
"learning_rate": 8.954807123329703e-06,
"loss": 0.1226,
"step": 140
},
{
"epoch": 0.4221556886227545,
"grad_norm": 1.046922805146823,
"learning_rate": 8.940375462677758e-06,
"loss": 0.1069,
"step": 141
},
{
"epoch": 0.4251497005988024,
"grad_norm": 1.1072835807231411,
"learning_rate": 8.92585664876797e-06,
"loss": 0.1233,
"step": 142
},
{
"epoch": 0.4281437125748503,
"grad_norm": 1.0130173393868498,
"learning_rate": 8.911251002727588e-06,
"loss": 0.0983,
"step": 143
},
{
"epoch": 0.4311377245508982,
"grad_norm": 0.9712494160291322,
"learning_rate": 8.896558847604414e-06,
"loss": 0.1033,
"step": 144
},
{
"epoch": 0.4341317365269461,
"grad_norm": 1.1440674743071744,
"learning_rate": 8.881780508359661e-06,
"loss": 0.1181,
"step": 145
},
{
"epoch": 0.437125748502994,
"grad_norm": 1.2084807313190982,
"learning_rate": 8.86691631186076e-06,
"loss": 0.1205,
"step": 146
},
{
"epoch": 0.44011976047904194,
"grad_norm": 0.9339544529717132,
"learning_rate": 8.851966586874138e-06,
"loss": 0.092,
"step": 147
},
{
"epoch": 0.4431137724550898,
"grad_norm": 1.195893147677474,
"learning_rate": 8.836931664057935e-06,
"loss": 0.1252,
"step": 148
},
{
"epoch": 0.44610778443113774,
"grad_norm": 1.1554746188228662,
"learning_rate": 8.821811875954705e-06,
"loss": 0.109,
"step": 149
},
{
"epoch": 0.4491017964071856,
"grad_norm": 1.0358188700192996,
"learning_rate": 8.806607556984045e-06,
"loss": 0.1147,
"step": 150
},
{
"epoch": 0.45209580838323354,
"grad_norm": 1.1495619704118192,
"learning_rate": 8.791319043435213e-06,
"loss": 0.1108,
"step": 151
},
{
"epoch": 0.4550898203592814,
"grad_norm": 1.0767571878157336,
"learning_rate": 8.775946673459682e-06,
"loss": 0.1042,
"step": 152
},
{
"epoch": 0.45808383233532934,
"grad_norm": 1.033094443704525,
"learning_rate": 8.76049078706366e-06,
"loss": 0.0989,
"step": 153
},
{
"epoch": 0.46107784431137727,
"grad_norm": 1.0818141713509735,
"learning_rate": 8.744951726100572e-06,
"loss": 0.109,
"step": 154
},
{
"epoch": 0.46407185628742514,
"grad_norm": 1.4560036834415162,
"learning_rate": 8.729329834263503e-06,
"loss": 0.0919,
"step": 155
},
{
"epoch": 0.46706586826347307,
"grad_norm": 1.3249320979346864,
"learning_rate": 8.713625457077585e-06,
"loss": 0.1275,
"step": 156
},
{
"epoch": 0.47005988023952094,
"grad_norm": 1.1823259804921005,
"learning_rate": 8.697838941892371e-06,
"loss": 0.116,
"step": 157
},
{
"epoch": 0.47305389221556887,
"grad_norm": 1.0577662488118387,
"learning_rate": 8.681970637874131e-06,
"loss": 0.1123,
"step": 158
},
{
"epoch": 0.47604790419161674,
"grad_norm": 1.223312576774771,
"learning_rate": 8.666020895998154e-06,
"loss": 0.1189,
"step": 159
},
{
"epoch": 0.47904191616766467,
"grad_norm": 1.093078965639497,
"learning_rate": 8.64999006904096e-06,
"loss": 0.1024,
"step": 160
},
{
"epoch": 0.4820359281437126,
"grad_norm": 1.3025555910186162,
"learning_rate": 8.63387851157252e-06,
"loss": 0.1246,
"step": 161
},
{
"epoch": 0.48502994011976047,
"grad_norm": 1.0704130612666423,
"learning_rate": 8.617686579948396e-06,
"loss": 0.0976,
"step": 162
},
{
"epoch": 0.4880239520958084,
"grad_norm": 1.2594827510216382,
"learning_rate": 8.60141463230187e-06,
"loss": 0.1138,
"step": 163
},
{
"epoch": 0.49101796407185627,
"grad_norm": 1.229833818844591,
"learning_rate": 8.585063028536015e-06,
"loss": 0.1204,
"step": 164
},
{
"epoch": 0.4940119760479042,
"grad_norm": 1.2596948575451763,
"learning_rate": 8.568632130315747e-06,
"loss": 0.1263,
"step": 165
},
{
"epoch": 0.49700598802395207,
"grad_norm": 1.0330515879808106,
"learning_rate": 8.552122301059807e-06,
"loss": 0.1024,
"step": 166
},
{
"epoch": 0.5,
"grad_norm": 1.0937944524544967,
"learning_rate": 8.535533905932739e-06,
"loss": 0.0979,
"step": 167
},
{
"epoch": 0.5029940119760479,
"grad_norm": 1.0781310185766044,
"learning_rate": 8.518867311836808e-06,
"loss": 0.1194,
"step": 168
},
{
"epoch": 0.5059880239520959,
"grad_norm": 0.9939367851701493,
"learning_rate": 8.502122887403882e-06,
"loss": 0.088,
"step": 169
},
{
"epoch": 0.5089820359281437,
"grad_norm": 1.020133013152269,
"learning_rate": 8.485301002987285e-06,
"loss": 0.0938,
"step": 170
},
{
"epoch": 0.5119760479041916,
"grad_norm": 1.166186467385633,
"learning_rate": 8.468402030653598e-06,
"loss": 0.1204,
"step": 171
},
{
"epoch": 0.5149700598802395,
"grad_norm": 0.9708486555429043,
"learning_rate": 8.451426344174433e-06,
"loss": 0.1002,
"step": 172
},
{
"epoch": 0.5179640718562875,
"grad_norm": 1.1595729468396334,
"learning_rate": 8.434374319018165e-06,
"loss": 0.1107,
"step": 173
},
{
"epoch": 0.5209580838323353,
"grad_norm": 1.091133620520665,
"learning_rate": 8.417246332341638e-06,
"loss": 0.1061,
"step": 174
},
{
"epoch": 0.5239520958083832,
"grad_norm": 1.1501502000269177,
"learning_rate": 8.4000427629818e-06,
"loss": 0.1127,
"step": 175
},
{
"epoch": 0.5269461077844312,
"grad_norm": 0.9818832434869366,
"learning_rate": 8.382763991447344e-06,
"loss": 0.0985,
"step": 176
},
{
"epoch": 0.5299401197604791,
"grad_norm": 1.285223024185485,
"learning_rate": 8.365410399910287e-06,
"loss": 0.1178,
"step": 177
},
{
"epoch": 0.5329341317365269,
"grad_norm": 1.2746897920320701,
"learning_rate": 8.347982372197515e-06,
"loss": 0.1166,
"step": 178
},
{
"epoch": 0.5359281437125748,
"grad_norm": 1.0945045183262572,
"learning_rate": 8.33048029378229e-06,
"loss": 0.1095,
"step": 179
},
{
"epoch": 0.5389221556886228,
"grad_norm": 1.1594313777901462,
"learning_rate": 8.312904551775731e-06,
"loss": 0.1232,
"step": 180
},
{
"epoch": 0.5419161676646707,
"grad_norm": 1.2108016364395127,
"learning_rate": 8.295255534918249e-06,
"loss": 0.1429,
"step": 181
},
{
"epoch": 0.5449101796407185,
"grad_norm": 1.3055603548974295,
"learning_rate": 8.277533633570948e-06,
"loss": 0.1317,
"step": 182
},
{
"epoch": 0.5479041916167665,
"grad_norm": 1.1438874937764232,
"learning_rate": 8.25973923970699e-06,
"loss": 0.1173,
"step": 183
},
{
"epoch": 0.5508982035928144,
"grad_norm": 1.1476605891455036,
"learning_rate": 8.241872746902934e-06,
"loss": 0.1215,
"step": 184
},
{
"epoch": 0.5538922155688623,
"grad_norm": 1.1394979568405286,
"learning_rate": 8.223934550330015e-06,
"loss": 0.1157,
"step": 185
},
{
"epoch": 0.5568862275449101,
"grad_norm": 1.0943696575087467,
"learning_rate": 8.20592504674542e-06,
"loss": 0.1135,
"step": 186
},
{
"epoch": 0.5598802395209581,
"grad_norm": 1.3105972553683807,
"learning_rate": 8.187844634483495e-06,
"loss": 0.136,
"step": 187
},
{
"epoch": 0.562874251497006,
"grad_norm": 1.2477435850517815,
"learning_rate": 8.16969371344696e-06,
"loss": 0.118,
"step": 188
},
{
"epoch": 0.5658682634730539,
"grad_norm": 1.0440441044548117,
"learning_rate": 8.151472685098037e-06,
"loss": 0.1186,
"step": 189
},
{
"epoch": 0.5688622754491018,
"grad_norm": 0.9937620218088092,
"learning_rate": 8.13318195244958e-06,
"loss": 0.1061,
"step": 190
},
{
"epoch": 0.5718562874251497,
"grad_norm": 1.0553284676258374,
"learning_rate": 8.114821920056177e-06,
"loss": 0.1086,
"step": 191
},
{
"epoch": 0.5748502994011976,
"grad_norm": 1.2465752458529162,
"learning_rate": 8.096392994005177e-06,
"loss": 0.1213,
"step": 192
},
{
"epoch": 0.5778443113772455,
"grad_norm": 1.0274519007784997,
"learning_rate": 8.077895581907719e-06,
"loss": 0.1034,
"step": 193
},
{
"epoch": 0.5808383233532934,
"grad_norm": 1.130587784163485,
"learning_rate": 8.059330092889724e-06,
"loss": 0.1132,
"step": 194
},
{
"epoch": 0.5838323353293413,
"grad_norm": 1.114546483447677,
"learning_rate": 8.040696937582833e-06,
"loss": 0.1137,
"step": 195
},
{
"epoch": 0.5868263473053892,
"grad_norm": 0.873781031172648,
"learning_rate": 8.021996528115335e-06,
"loss": 0.0838,
"step": 196
},
{
"epoch": 0.5898203592814372,
"grad_norm": 1.2321249744806237,
"learning_rate": 8.003229278103044e-06,
"loss": 0.1137,
"step": 197
},
{
"epoch": 0.592814371257485,
"grad_norm": 1.1358395333676212,
"learning_rate": 7.984395602640153e-06,
"loss": 0.1179,
"step": 198
},
{
"epoch": 0.5958083832335329,
"grad_norm": 1.0348332164450986,
"learning_rate": 7.96549591829006e-06,
"loss": 0.1058,
"step": 199
},
{
"epoch": 0.5988023952095808,
"grad_norm": 0.9547236919983346,
"learning_rate": 7.946530643076138e-06,
"loss": 0.0921,
"step": 200
},
{
"epoch": 0.5988023952095808,
"eval_loss": 0.12016214430332184,
"eval_runtime": 8.3266,
"eval_samples_per_second": 6.485,
"eval_steps_per_second": 1.681,
"step": 200
},
{
"epoch": 0.6017964071856288,
"grad_norm": 0.9344428491844222,
"learning_rate": 7.927500196472506e-06,
"loss": 0.0988,
"step": 201
},
{
"epoch": 0.6047904191616766,
"grad_norm": 1.0777875715216008,
"learning_rate": 7.908404999394747e-06,
"loss": 0.1019,
"step": 202
},
{
"epoch": 0.6077844311377245,
"grad_norm": 1.237958694976887,
"learning_rate": 7.889245474190588e-06,
"loss": 0.12,
"step": 203
},
{
"epoch": 0.6107784431137725,
"grad_norm": 1.4421618806920444,
"learning_rate": 7.870022044630569e-06,
"loss": 0.1405,
"step": 204
},
{
"epoch": 0.6137724550898204,
"grad_norm": 1.1144924473329114,
"learning_rate": 7.85073513589867e-06,
"loss": 0.1151,
"step": 205
},
{
"epoch": 0.6167664670658682,
"grad_norm": 1.0921040138700113,
"learning_rate": 7.831385174582901e-06,
"loss": 0.1098,
"step": 206
},
{
"epoch": 0.6197604790419161,
"grad_norm": 1.0662690709409506,
"learning_rate": 7.81197258866587e-06,
"loss": 0.0906,
"step": 207
},
{
"epoch": 0.6227544910179641,
"grad_norm": 1.055975543135569,
"learning_rate": 7.792497807515317e-06,
"loss": 0.1022,
"step": 208
},
{
"epoch": 0.625748502994012,
"grad_norm": 0.972964899017892,
"learning_rate": 7.772961261874615e-06,
"loss": 0.1007,
"step": 209
},
{
"epoch": 0.6287425149700598,
"grad_norm": 1.1345049478890776,
"learning_rate": 7.75336338385325e-06,
"loss": 0.1112,
"step": 210
},
{
"epoch": 0.6317365269461078,
"grad_norm": 1.1918299507322672,
"learning_rate": 7.733704606917248e-06,
"loss": 0.1281,
"step": 211
},
{
"epoch": 0.6347305389221557,
"grad_norm": 1.1044275348689654,
"learning_rate": 7.713985365879607e-06,
"loss": 0.1035,
"step": 212
},
{
"epoch": 0.6377245508982036,
"grad_norm": 1.2196447175508784,
"learning_rate": 7.694206096890667e-06,
"loss": 0.1354,
"step": 213
},
{
"epoch": 0.6407185628742516,
"grad_norm": 0.9858226926171154,
"learning_rate": 7.674367237428467e-06,
"loss": 0.1005,
"step": 214
},
{
"epoch": 0.6437125748502994,
"grad_norm": 1.0313314743245496,
"learning_rate": 7.654469226289068e-06,
"loss": 0.1035,
"step": 215
},
{
"epoch": 0.6467065868263473,
"grad_norm": 1.2073723687580094,
"learning_rate": 7.63451250357685e-06,
"loss": 0.1215,
"step": 216
},
{
"epoch": 0.6497005988023952,
"grad_norm": 1.1575984459704936,
"learning_rate": 7.614497510694774e-06,
"loss": 0.103,
"step": 217
},
{
"epoch": 0.6526946107784432,
"grad_norm": 1.1928147558872801,
"learning_rate": 7.5944246903346204e-06,
"loss": 0.1138,
"step": 218
},
{
"epoch": 0.655688622754491,
"grad_norm": 0.8982426003073268,
"learning_rate": 7.574294486467204e-06,
"loss": 0.0831,
"step": 219
},
{
"epoch": 0.6586826347305389,
"grad_norm": 0.9394252234084429,
"learning_rate": 7.55410734433254e-06,
"loss": 0.0959,
"step": 220
},
{
"epoch": 0.6616766467065869,
"grad_norm": 1.137723030712234,
"learning_rate": 7.533863710430011e-06,
"loss": 0.1109,
"step": 221
},
{
"epoch": 0.6646706586826348,
"grad_norm": 0.8267045625199,
"learning_rate": 7.513564032508484e-06,
"loss": 0.0842,
"step": 222
},
{
"epoch": 0.6676646706586826,
"grad_norm": 1.1847186477673504,
"learning_rate": 7.493208759556406e-06,
"loss": 0.1151,
"step": 223
},
{
"epoch": 0.6706586826347305,
"grad_norm": 1.0731347002786096,
"learning_rate": 7.472798341791877e-06,
"loss": 0.1087,
"step": 224
},
{
"epoch": 0.6736526946107785,
"grad_norm": 0.9153166575730822,
"learning_rate": 7.452333230652688e-06,
"loss": 0.0784,
"step": 225
},
{
"epoch": 0.6766467065868264,
"grad_norm": 1.6389325670271402,
"learning_rate": 7.431813878786343e-06,
"loss": 0.1158,
"step": 226
},
{
"epoch": 0.6796407185628742,
"grad_norm": 0.94424634794741,
"learning_rate": 7.4112407400400395e-06,
"loss": 0.0857,
"step": 227
},
{
"epoch": 0.6826347305389222,
"grad_norm": 0.89702175760304,
"learning_rate": 7.390614269450633e-06,
"loss": 0.0831,
"step": 228
},
{
"epoch": 0.6856287425149701,
"grad_norm": 1.016468144740291,
"learning_rate": 7.369934923234577e-06,
"loss": 0.0991,
"step": 229
},
{
"epoch": 0.688622754491018,
"grad_norm": 1.182973315725974,
"learning_rate": 7.349203158777826e-06,
"loss": 0.1118,
"step": 230
},
{
"epoch": 0.6916167664670658,
"grad_norm": 1.0092567790353997,
"learning_rate": 7.32841943462572e-06,
"loss": 0.1024,
"step": 231
},
{
"epoch": 0.6946107784431138,
"grad_norm": 1.1411601927497255,
"learning_rate": 7.3075842104728445e-06,
"loss": 0.1146,
"step": 232
},
{
"epoch": 0.6976047904191617,
"grad_norm": 1.1078945128986997,
"learning_rate": 7.286697947152868e-06,
"loss": 0.1164,
"step": 233
},
{
"epoch": 0.7005988023952096,
"grad_norm": 0.8662144303520954,
"learning_rate": 7.265761106628338e-06,
"loss": 0.0829,
"step": 234
},
{
"epoch": 0.7035928143712575,
"grad_norm": 0.9752397406169193,
"learning_rate": 7.244774151980466e-06,
"loss": 0.0943,
"step": 235
},
{
"epoch": 0.7065868263473054,
"grad_norm": 1.1117357986281629,
"learning_rate": 7.223737547398898e-06,
"loss": 0.1087,
"step": 236
},
{
"epoch": 0.7095808383233533,
"grad_norm": 1.0339611771677366,
"learning_rate": 7.20265175817143e-06,
"loss": 0.1092,
"step": 237
},
{
"epoch": 0.7125748502994012,
"grad_norm": 1.3500709005637261,
"learning_rate": 7.181517250673729e-06,
"loss": 0.1507,
"step": 238
},
{
"epoch": 0.7155688622754491,
"grad_norm": 1.1511251305711105,
"learning_rate": 7.1603344923590065e-06,
"loss": 0.1098,
"step": 239
},
{
"epoch": 0.718562874251497,
"grad_norm": 1.2192955758731912,
"learning_rate": 7.139103951747694e-06,
"loss": 0.1272,
"step": 240
},
{
"epoch": 0.7215568862275449,
"grad_norm": 1.0776006747271616,
"learning_rate": 7.1178260984170675e-06,
"loss": 0.1067,
"step": 241
},
{
"epoch": 0.7245508982035929,
"grad_norm": 0.9530867829023002,
"learning_rate": 7.0965014029908654e-06,
"loss": 0.0968,
"step": 242
},
{
"epoch": 0.7275449101796407,
"grad_norm": 1.1096466841531403,
"learning_rate": 7.075130337128883e-06,
"loss": 0.1299,
"step": 243
},
{
"epoch": 0.7305389221556886,
"grad_norm": 1.2165381641728317,
"learning_rate": 7.053713373516538e-06,
"loss": 0.1349,
"step": 244
},
{
"epoch": 0.7335329341317365,
"grad_norm": 1.1146967815470412,
"learning_rate": 7.03225098585441e-06,
"loss": 0.1135,
"step": 245
},
{
"epoch": 0.7365269461077845,
"grad_norm": 1.1148754947833899,
"learning_rate": 7.0107436488477694e-06,
"loss": 0.1266,
"step": 246
},
{
"epoch": 0.7395209580838323,
"grad_norm": 1.1161083256305881,
"learning_rate": 6.989191838196083e-06,
"loss": 0.1194,
"step": 247
},
{
"epoch": 0.7425149700598802,
"grad_norm": 0.9469562698402685,
"learning_rate": 6.9675960305824785e-06,
"loss": 0.099,
"step": 248
},
{
"epoch": 0.7455089820359282,
"grad_norm": 1.151194749997882,
"learning_rate": 6.945956703663212e-06,
"loss": 0.1153,
"step": 249
},
{
"epoch": 0.7485029940119761,
"grad_norm": 1.0882092511428272,
"learning_rate": 6.9242743360570985e-06,
"loss": 0.1078,
"step": 250
},
{
"epoch": 0.7514970059880239,
"grad_norm": 1.028388896286411,
"learning_rate": 6.9025494073349284e-06,
"loss": 0.1096,
"step": 251
},
{
"epoch": 0.7544910179640718,
"grad_norm": 0.9972504617324368,
"learning_rate": 6.880782398008862e-06,
"loss": 0.1087,
"step": 252
},
{
"epoch": 0.7574850299401198,
"grad_norm": 0.9256438231559054,
"learning_rate": 6.858973789521792e-06,
"loss": 0.0922,
"step": 253
},
{
"epoch": 0.7604790419161677,
"grad_norm": 1.0098118238043352,
"learning_rate": 6.837124064236709e-06,
"loss": 0.0967,
"step": 254
},
{
"epoch": 0.7634730538922155,
"grad_norm": 0.9171644357934685,
"learning_rate": 6.815233705426019e-06,
"loss": 0.095,
"step": 255
},
{
"epoch": 0.7664670658682635,
"grad_norm": 1.0311629178083461,
"learning_rate": 6.7933031972608644e-06,
"loss": 0.1009,
"step": 256
},
{
"epoch": 0.7694610778443114,
"grad_norm": 1.2065768311047647,
"learning_rate": 6.771333024800411e-06,
"loss": 0.1127,
"step": 257
},
{
"epoch": 0.7724550898203593,
"grad_norm": 1.1649667504142824,
"learning_rate": 6.74932367398112e-06,
"loss": 0.112,
"step": 258
},
{
"epoch": 0.7754491017964071,
"grad_norm": 0.9308015487490596,
"learning_rate": 6.727275631605996e-06,
"loss": 0.0889,
"step": 259
},
{
"epoch": 0.7784431137724551,
"grad_norm": 1.2121108383798198,
"learning_rate": 6.70518938533383e-06,
"loss": 0.1105,
"step": 260
},
{
"epoch": 0.781437125748503,
"grad_norm": 0.9872567997673177,
"learning_rate": 6.683065423668403e-06,
"loss": 0.091,
"step": 261
},
{
"epoch": 0.7844311377245509,
"grad_norm": 1.1434398705082174,
"learning_rate": 6.660904235947687e-06,
"loss": 0.1084,
"step": 262
},
{
"epoch": 0.7874251497005988,
"grad_norm": 1.1135729967157757,
"learning_rate": 6.638706312333018e-06,
"loss": 0.1036,
"step": 263
},
{
"epoch": 0.7904191616766467,
"grad_norm": 1.2172285636766491,
"learning_rate": 6.61647214379826e-06,
"loss": 0.1184,
"step": 264
},
{
"epoch": 0.7934131736526946,
"grad_norm": 1.3488738740905057,
"learning_rate": 6.594202222118941e-06,
"loss": 0.1222,
"step": 265
},
{
"epoch": 0.7964071856287425,
"grad_norm": 1.0269053596938786,
"learning_rate": 6.571897039861377e-06,
"loss": 0.1028,
"step": 266
},
{
"epoch": 0.7994011976047904,
"grad_norm": 1.1129746486220893,
"learning_rate": 6.549557090371775e-06,
"loss": 0.1154,
"step": 267
},
{
"epoch": 0.8023952095808383,
"grad_norm": 0.842333383910561,
"learning_rate": 6.527182867765333e-06,
"loss": 0.0771,
"step": 268
},
{
"epoch": 0.8053892215568862,
"grad_norm": 1.3078259478030483,
"learning_rate": 6.504774866915291e-06,
"loss": 0.13,
"step": 269
},
{
"epoch": 0.8083832335329342,
"grad_norm": 1.0001135093091333,
"learning_rate": 6.482333583442002e-06,
"loss": 0.0997,
"step": 270
},
{
"epoch": 0.811377245508982,
"grad_norm": 1.1205663397366659,
"learning_rate": 6.459859513701967e-06,
"loss": 0.1063,
"step": 271
},
{
"epoch": 0.8143712574850299,
"grad_norm": 1.2779493257817607,
"learning_rate": 6.437353154776848e-06,
"loss": 0.1157,
"step": 272
},
{
"epoch": 0.8173652694610778,
"grad_norm": 1.0352680525112001,
"learning_rate": 6.414815004462483e-06,
"loss": 0.0886,
"step": 273
},
{
"epoch": 0.8203592814371258,
"grad_norm": 1.0436596456460765,
"learning_rate": 6.3922455612578715e-06,
"loss": 0.113,
"step": 274
},
{
"epoch": 0.8233532934131736,
"grad_norm": 0.9993307292108748,
"learning_rate": 6.369645324354149e-06,
"loss": 0.0914,
"step": 275
},
{
"epoch": 0.8263473053892215,
"grad_norm": 1.2497821802624005,
"learning_rate": 6.3470147936235485e-06,
"loss": 0.13,
"step": 276
},
{
"epoch": 0.8293413173652695,
"grad_norm": 0.8174766850501531,
"learning_rate": 6.3243544696083355e-06,
"loss": 0.0876,
"step": 277
},
{
"epoch": 0.8323353293413174,
"grad_norm": 1.1314296946186437,
"learning_rate": 6.301664853509755e-06,
"loss": 0.1181,
"step": 278
},
{
"epoch": 0.8353293413173652,
"grad_norm": 1.0004348069946,
"learning_rate": 6.278946447176924e-06,
"loss": 0.09,
"step": 279
},
{
"epoch": 0.8383233532934131,
"grad_norm": 1.0671378195782413,
"learning_rate": 6.256199753095745e-06,
"loss": 0.1056,
"step": 280
},
{
"epoch": 0.8413173652694611,
"grad_norm": 0.8364701680716429,
"learning_rate": 6.233425274377793e-06,
"loss": 0.0824,
"step": 281
},
{
"epoch": 0.844311377245509,
"grad_norm": 1.1223141953145617,
"learning_rate": 6.21062351474918e-06,
"loss": 0.1032,
"step": 282
},
{
"epoch": 0.8473053892215568,
"grad_norm": 1.0524119154471643,
"learning_rate": 6.18779497853942e-06,
"loss": 0.1081,
"step": 283
},
{
"epoch": 0.8502994011976048,
"grad_norm": 1.0264788336701436,
"learning_rate": 6.164940170670266e-06,
"loss": 0.1069,
"step": 284
},
{
"epoch": 0.8532934131736527,
"grad_norm": 0.981192757287711,
"learning_rate": 6.142059596644557e-06,
"loss": 0.0894,
"step": 285
},
{
"epoch": 0.8562874251497006,
"grad_norm": 0.9615230917131988,
"learning_rate": 6.11915376253502e-06,
"loss": 0.0897,
"step": 286
},
{
"epoch": 0.8592814371257484,
"grad_norm": 1.0337209438384687,
"learning_rate": 6.096223174973091e-06,
"loss": 0.1006,
"step": 287
},
{
"epoch": 0.8622754491017964,
"grad_norm": 0.9905876169014615,
"learning_rate": 6.073268341137694e-06,
"loss": 0.095,
"step": 288
},
{
"epoch": 0.8652694610778443,
"grad_norm": 1.1027961277676517,
"learning_rate": 6.050289768744042e-06,
"loss": 0.0998,
"step": 289
},
{
"epoch": 0.8682634730538922,
"grad_norm": 1.083233411983881,
"learning_rate": 6.0272879660323936e-06,
"loss": 0.1096,
"step": 290
},
{
"epoch": 0.8712574850299402,
"grad_norm": 1.0913128739322395,
"learning_rate": 6.004263441756815e-06,
"loss": 0.1059,
"step": 291
},
{
"epoch": 0.874251497005988,
"grad_norm": 1.0909830526364266,
"learning_rate": 5.98121670517393e-06,
"loss": 0.1078,
"step": 292
},
{
"epoch": 0.8772455089820359,
"grad_norm": 0.8883708109490787,
"learning_rate": 5.958148266031654e-06,
"loss": 0.0961,
"step": 293
},
{
"epoch": 0.8802395209580839,
"grad_norm": 1.0038270303162207,
"learning_rate": 5.935058634557917e-06,
"loss": 0.0939,
"step": 294
},
{
"epoch": 0.8832335329341318,
"grad_norm": 0.9312159317082542,
"learning_rate": 5.911948321449384e-06,
"loss": 0.0924,
"step": 295
},
{
"epoch": 0.8862275449101796,
"grad_norm": 1.1127876140747484,
"learning_rate": 5.8888178378601565e-06,
"loss": 0.1104,
"step": 296
},
{
"epoch": 0.8892215568862275,
"grad_norm": 1.0620371276229927,
"learning_rate": 5.865667695390468e-06,
"loss": 0.0993,
"step": 297
},
{
"epoch": 0.8922155688622755,
"grad_norm": 0.8791295012960114,
"learning_rate": 5.842498406075363e-06,
"loss": 0.0843,
"step": 298
},
{
"epoch": 0.8952095808383234,
"grad_norm": 0.9421170351009726,
"learning_rate": 5.819310482373381e-06,
"loss": 0.0815,
"step": 299
},
{
"epoch": 0.8982035928143712,
"grad_norm": 1.1078125505045537,
"learning_rate": 5.796104437155213e-06,
"loss": 0.1117,
"step": 300
},
{
"epoch": 0.9011976047904192,
"grad_norm": 0.9854231317801608,
"learning_rate": 5.772880783692363e-06,
"loss": 0.0887,
"step": 301
},
{
"epoch": 0.9041916167664671,
"grad_norm": 1.1279967736862624,
"learning_rate": 5.749640035645798e-06,
"loss": 0.1063,
"step": 302
},
{
"epoch": 0.907185628742515,
"grad_norm": 0.9960776335710316,
"learning_rate": 5.726382707054578e-06,
"loss": 0.0921,
"step": 303
},
{
"epoch": 0.9101796407185628,
"grad_norm": 1.064755038701587,
"learning_rate": 5.703109312324493e-06,
"loss": 0.0961,
"step": 304
},
{
"epoch": 0.9131736526946108,
"grad_norm": 1.1225010610453825,
"learning_rate": 5.679820366216684e-06,
"loss": 0.1181,
"step": 305
},
{
"epoch": 0.9161676646706587,
"grad_norm": 0.9015413967576371,
"learning_rate": 5.656516383836263e-06,
"loss": 0.0787,
"step": 306
},
{
"epoch": 0.9191616766467066,
"grad_norm": 1.1259253163412393,
"learning_rate": 5.6331978806209044e-06,
"loss": 0.1139,
"step": 307
},
{
"epoch": 0.9221556886227545,
"grad_norm": 1.0731440571023292,
"learning_rate": 5.609865372329461e-06,
"loss": 0.0968,
"step": 308
},
{
"epoch": 0.9251497005988024,
"grad_norm": 1.0478169134439905,
"learning_rate": 5.586519375030549e-06,
"loss": 0.1082,
"step": 309
},
{
"epoch": 0.9281437125748503,
"grad_norm": 1.0407654799348711,
"learning_rate": 5.5631604050911354e-06,
"loss": 0.0989,
"step": 310
},
{
"epoch": 0.9311377245508982,
"grad_norm": 1.0508551890563849,
"learning_rate": 5.539788979165115e-06,
"loss": 0.1013,
"step": 311
},
{
"epoch": 0.9341317365269461,
"grad_norm": 0.923954532579174,
"learning_rate": 5.516405614181883e-06,
"loss": 0.0973,
"step": 312
},
{
"epoch": 0.937125748502994,
"grad_norm": 0.8966214747374445,
"learning_rate": 5.4930108273349034e-06,
"loss": 0.0842,
"step": 313
},
{
"epoch": 0.9401197604790419,
"grad_norm": 1.0545356732072688,
"learning_rate": 5.4696051360702725e-06,
"loss": 0.1036,
"step": 314
},
{
"epoch": 0.9431137724550899,
"grad_norm": 1.2592838120797554,
"learning_rate": 5.446189058075265e-06,
"loss": 0.1175,
"step": 315
},
{
"epoch": 0.9461077844311377,
"grad_norm": 1.147053941416306,
"learning_rate": 5.4227631112668955e-06,
"loss": 0.1093,
"step": 316
},
{
"epoch": 0.9491017964071856,
"grad_norm": 0.9646520479241986,
"learning_rate": 5.39932781378045e-06,
"loss": 0.0841,
"step": 317
},
{
"epoch": 0.9520958083832335,
"grad_norm": 1.0575832107417094,
"learning_rate": 5.375883683958041e-06,
"loss": 0.1116,
"step": 318
},
{
"epoch": 0.9550898203592815,
"grad_norm": 0.9810315231097588,
"learning_rate": 5.3524312403371255e-06,
"loss": 0.0939,
"step": 319
},
{
"epoch": 0.9580838323353293,
"grad_norm": 1.0279268894390523,
"learning_rate": 5.328971001639054e-06,
"loss": 0.0998,
"step": 320
},
{
"epoch": 0.9610778443113772,
"grad_norm": 1.136417758736507,
"learning_rate": 5.3055034867575825e-06,
"loss": 0.1095,
"step": 321
},
{
"epoch": 0.9640718562874252,
"grad_norm": 1.153488132682876,
"learning_rate": 5.282029214747404e-06,
"loss": 0.1081,
"step": 322
},
{
"epoch": 0.9670658682634731,
"grad_norm": 1.2048728627853904,
"learning_rate": 5.258548704812667e-06,
"loss": 0.111,
"step": 323
},
{
"epoch": 0.9700598802395209,
"grad_norm": 1.1484151553985187,
"learning_rate": 5.235062476295488e-06,
"loss": 0.1125,
"step": 324
},
{
"epoch": 0.9730538922155688,
"grad_norm": 1.1343182106364535,
"learning_rate": 5.211571048664469e-06,
"loss": 0.093,
"step": 325
},
{
"epoch": 0.9760479041916168,
"grad_norm": 0.9845672359142281,
"learning_rate": 5.188074941503203e-06,
"loss": 0.1015,
"step": 326
},
{
"epoch": 0.9790419161676647,
"grad_norm": 1.165160638091069,
"learning_rate": 5.164574674498788e-06,
"loss": 0.1028,
"step": 327
},
{
"epoch": 0.9820359281437125,
"grad_norm": 1.0164063834652908,
"learning_rate": 5.141070767430331e-06,
"loss": 0.0986,
"step": 328
},
{
"epoch": 0.9850299401197605,
"grad_norm": 1.1404188476330233,
"learning_rate": 5.117563740157444e-06,
"loss": 0.1035,
"step": 329
},
{
"epoch": 0.9880239520958084,
"grad_norm": 1.1740883832722306,
"learning_rate": 5.094054112608758e-06,
"loss": 0.1081,
"step": 330
},
{
"epoch": 0.9910179640718563,
"grad_norm": 1.0738025982568447,
"learning_rate": 5.070542404770413e-06,
"loss": 0.1061,
"step": 331
},
{
"epoch": 0.9940119760479041,
"grad_norm": 0.926438169582587,
"learning_rate": 5.047029136674563e-06,
"loss": 0.088,
"step": 332
},
{
"epoch": 0.9970059880239521,
"grad_norm": 1.418334686291164,
"learning_rate": 5.023514828387868e-06,
"loss": 0.1334,
"step": 333
},
{
"epoch": 1.0,
"grad_norm": 0.9103765676053909,
"learning_rate": 5e-06,
"loss": 0.0752,
"step": 334
},
{
"epoch": 1.0029940119760479,
"grad_norm": 0.7920123120809723,
"learning_rate": 4.976485171612134e-06,
"loss": 0.0498,
"step": 335
},
{
"epoch": 1.0059880239520957,
"grad_norm": 0.7645844304656827,
"learning_rate": 4.95297086332544e-06,
"loss": 0.049,
"step": 336
},
{
"epoch": 1.0089820359281436,
"grad_norm": 0.6769348281909342,
"learning_rate": 4.9294575952295896e-06,
"loss": 0.0428,
"step": 337
},
{
"epoch": 1.0119760479041917,
"grad_norm": 0.7200135230361624,
"learning_rate": 4.905945887391242e-06,
"loss": 0.0456,
"step": 338
},
{
"epoch": 1.0149700598802396,
"grad_norm": 0.6729442176900161,
"learning_rate": 4.882436259842556e-06,
"loss": 0.0373,
"step": 339
},
{
"epoch": 1.0179640718562875,
"grad_norm": 0.6971669698075075,
"learning_rate": 4.858929232569671e-06,
"loss": 0.0422,
"step": 340
},
{
"epoch": 1.0209580838323353,
"grad_norm": 0.7495540742531779,
"learning_rate": 4.835425325501214e-06,
"loss": 0.0498,
"step": 341
},
{
"epoch": 1.0239520958083832,
"grad_norm": 0.6366168809607379,
"learning_rate": 4.811925058496799e-06,
"loss": 0.0401,
"step": 342
},
{
"epoch": 1.026946107784431,
"grad_norm": 0.7222098169912883,
"learning_rate": 4.788428951335534e-06,
"loss": 0.035,
"step": 343
},
{
"epoch": 1.029940119760479,
"grad_norm": 0.7836392148736803,
"learning_rate": 4.7649375237045135e-06,
"loss": 0.044,
"step": 344
},
{
"epoch": 1.032934131736527,
"grad_norm": 0.5574391641162871,
"learning_rate": 4.741451295187333e-06,
"loss": 0.0264,
"step": 345
},
{
"epoch": 1.035928143712575,
"grad_norm": 0.7628970798704946,
"learning_rate": 4.717970785252596e-06,
"loss": 0.0415,
"step": 346
},
{
"epoch": 1.0389221556886228,
"grad_norm": 0.8667402571860164,
"learning_rate": 4.694496513242418e-06,
"loss": 0.0375,
"step": 347
},
{
"epoch": 1.0419161676646707,
"grad_norm": 0.6670803258208319,
"learning_rate": 4.671028998360947e-06,
"loss": 0.0288,
"step": 348
},
{
"epoch": 1.0449101796407185,
"grad_norm": 0.6991849065877793,
"learning_rate": 4.647568759662876e-06,
"loss": 0.0341,
"step": 349
},
{
"epoch": 1.0479041916167664,
"grad_norm": 0.9326177137319185,
"learning_rate": 4.624116316041962e-06,
"loss": 0.0427,
"step": 350
},
{
"epoch": 1.0508982035928143,
"grad_norm": 0.7317921993459834,
"learning_rate": 4.600672186219551e-06,
"loss": 0.0279,
"step": 351
},
{
"epoch": 1.0538922155688624,
"grad_norm": 0.936307056378493,
"learning_rate": 4.5772368887331044e-06,
"loss": 0.0333,
"step": 352
},
{
"epoch": 1.0568862275449102,
"grad_norm": 0.9429289119624542,
"learning_rate": 4.553810941924735e-06,
"loss": 0.044,
"step": 353
},
{
"epoch": 1.0598802395209581,
"grad_norm": 0.929022486839685,
"learning_rate": 4.530394863929728e-06,
"loss": 0.0404,
"step": 354
},
{
"epoch": 1.062874251497006,
"grad_norm": 0.920787814386675,
"learning_rate": 4.506989172665097e-06,
"loss": 0.0411,
"step": 355
},
{
"epoch": 1.0658682634730539,
"grad_norm": 0.9822944109810009,
"learning_rate": 4.483594385818119e-06,
"loss": 0.0411,
"step": 356
},
{
"epoch": 1.0688622754491017,
"grad_norm": 1.0776428732380365,
"learning_rate": 4.460211020834887e-06,
"loss": 0.0443,
"step": 357
},
{
"epoch": 1.0718562874251496,
"grad_norm": 0.9242680121191666,
"learning_rate": 4.436839594908866e-06,
"loss": 0.037,
"step": 358
},
{
"epoch": 1.0748502994011977,
"grad_norm": 1.102736384384949,
"learning_rate": 4.4134806249694514e-06,
"loss": 0.0314,
"step": 359
},
{
"epoch": 1.0778443113772456,
"grad_norm": 0.9586978115919587,
"learning_rate": 4.39013462767054e-06,
"loss": 0.0379,
"step": 360
},
{
"epoch": 1.0808383233532934,
"grad_norm": 0.8744433388997015,
"learning_rate": 4.366802119379098e-06,
"loss": 0.0407,
"step": 361
},
{
"epoch": 1.0838323353293413,
"grad_norm": 0.8245451265781271,
"learning_rate": 4.34348361616374e-06,
"loss": 0.034,
"step": 362
},
{
"epoch": 1.0868263473053892,
"grad_norm": 1.1068030098172257,
"learning_rate": 4.3201796337833165e-06,
"loss": 0.0416,
"step": 363
},
{
"epoch": 1.089820359281437,
"grad_norm": 0.7777816793906646,
"learning_rate": 4.29689068767551e-06,
"loss": 0.0323,
"step": 364
},
{
"epoch": 1.092814371257485,
"grad_norm": 1.0250067844793325,
"learning_rate": 4.273617292945425e-06,
"loss": 0.0513,
"step": 365
},
{
"epoch": 1.095808383233533,
"grad_norm": 0.845654890529388,
"learning_rate": 4.250359964354203e-06,
"loss": 0.0374,
"step": 366
},
{
"epoch": 1.098802395209581,
"grad_norm": 0.8964258556108368,
"learning_rate": 4.227119216307637e-06,
"loss": 0.0367,
"step": 367
},
{
"epoch": 1.1017964071856288,
"grad_norm": 0.7868435665620686,
"learning_rate": 4.203895562844789e-06,
"loss": 0.0336,
"step": 368
},
{
"epoch": 1.1047904191616766,
"grad_norm": 0.8457116061447782,
"learning_rate": 4.18068951762662e-06,
"loss": 0.0293,
"step": 369
},
{
"epoch": 1.1077844311377245,
"grad_norm": 0.7530274777405538,
"learning_rate": 4.157501593924638e-06,
"loss": 0.0367,
"step": 370
},
{
"epoch": 1.1107784431137724,
"grad_norm": 0.9776857778183234,
"learning_rate": 4.134332304609533e-06,
"loss": 0.0443,
"step": 371
},
{
"epoch": 1.1137724550898203,
"grad_norm": 0.6793113738947543,
"learning_rate": 4.111182162139844e-06,
"loss": 0.0319,
"step": 372
},
{
"epoch": 1.1167664670658684,
"grad_norm": 0.7881838904063722,
"learning_rate": 4.088051678550617e-06,
"loss": 0.0315,
"step": 373
},
{
"epoch": 1.1197604790419162,
"grad_norm": 0.862318278207734,
"learning_rate": 4.064941365442084e-06,
"loss": 0.0371,
"step": 374
},
{
"epoch": 1.122754491017964,
"grad_norm": 1.0744855612366662,
"learning_rate": 4.041851733968348e-06,
"loss": 0.0513,
"step": 375
},
{
"epoch": 1.125748502994012,
"grad_norm": 0.914922903066341,
"learning_rate": 4.018783294826071e-06,
"loss": 0.042,
"step": 376
},
{
"epoch": 1.1287425149700598,
"grad_norm": 0.933269550307153,
"learning_rate": 3.995736558243186e-06,
"loss": 0.0392,
"step": 377
},
{
"epoch": 1.1317365269461077,
"grad_norm": 0.9029916694162012,
"learning_rate": 3.972712033967608e-06,
"loss": 0.0465,
"step": 378
},
{
"epoch": 1.1347305389221556,
"grad_norm": 0.859954495431855,
"learning_rate": 3.949710231255961e-06,
"loss": 0.0484,
"step": 379
},
{
"epoch": 1.1377245508982037,
"grad_norm": 0.7531777410274714,
"learning_rate": 3.926731658862307e-06,
"loss": 0.0329,
"step": 380
},
{
"epoch": 1.1407185628742516,
"grad_norm": 0.7087909296607872,
"learning_rate": 3.903776825026912e-06,
"loss": 0.0337,
"step": 381
},
{
"epoch": 1.1437125748502994,
"grad_norm": 0.7330388221118903,
"learning_rate": 3.8808462374649805e-06,
"loss": 0.0372,
"step": 382
},
{
"epoch": 1.1467065868263473,
"grad_norm": 0.7368243151394032,
"learning_rate": 3.857940403355444e-06,
"loss": 0.0373,
"step": 383
},
{
"epoch": 1.1497005988023952,
"grad_norm": 0.727158340191875,
"learning_rate": 3.8350598293297345e-06,
"loss": 0.0371,
"step": 384
},
{
"epoch": 1.152694610778443,
"grad_norm": 0.9195286471944959,
"learning_rate": 3.8122050214605822e-06,
"loss": 0.0421,
"step": 385
},
{
"epoch": 1.1556886227544911,
"grad_norm": 0.8716150110798513,
"learning_rate": 3.7893764852508207e-06,
"loss": 0.0431,
"step": 386
},
{
"epoch": 1.158682634730539,
"grad_norm": 0.7216158638005744,
"learning_rate": 3.766574725622208e-06,
"loss": 0.0307,
"step": 387
},
{
"epoch": 1.1616766467065869,
"grad_norm": 0.7001316533601489,
"learning_rate": 3.7438002469042567e-06,
"loss": 0.0287,
"step": 388
},
{
"epoch": 1.1646706586826348,
"grad_norm": 0.7904658156314911,
"learning_rate": 3.721053552823078e-06,
"loss": 0.0335,
"step": 389
},
{
"epoch": 1.1676646706586826,
"grad_norm": 0.7175422431712003,
"learning_rate": 3.698335146490246e-06,
"loss": 0.0324,
"step": 390
},
{
"epoch": 1.1706586826347305,
"grad_norm": 0.7842272991353707,
"learning_rate": 3.675645530391665e-06,
"loss": 0.0294,
"step": 391
},
{
"epoch": 1.1736526946107784,
"grad_norm": 0.961378079601426,
"learning_rate": 3.652985206376455e-06,
"loss": 0.0361,
"step": 392
},
{
"epoch": 1.1766467065868262,
"grad_norm": 0.9744359586687764,
"learning_rate": 3.630354675645853e-06,
"loss": 0.0412,
"step": 393
},
{
"epoch": 1.1796407185628743,
"grad_norm": 0.8939009027709087,
"learning_rate": 3.6077544387421293e-06,
"loss": 0.0377,
"step": 394
},
{
"epoch": 1.1826347305389222,
"grad_norm": 0.6647428987195185,
"learning_rate": 3.5851849955375177e-06,
"loss": 0.0297,
"step": 395
},
{
"epoch": 1.18562874251497,
"grad_norm": 0.7347697572783929,
"learning_rate": 3.5626468452231534e-06,
"loss": 0.0272,
"step": 396
},
{
"epoch": 1.188622754491018,
"grad_norm": 0.8121052793901488,
"learning_rate": 3.540140486298035e-06,
"loss": 0.0326,
"step": 397
},
{
"epoch": 1.1916167664670658,
"grad_norm": 0.9567500286690479,
"learning_rate": 3.517666416557999e-06,
"loss": 0.0451,
"step": 398
},
{
"epoch": 1.1946107784431137,
"grad_norm": 0.661783630411826,
"learning_rate": 3.495225133084712e-06,
"loss": 0.0252,
"step": 399
},
{
"epoch": 1.1976047904191618,
"grad_norm": 0.6375626484149183,
"learning_rate": 3.472817132234669e-06,
"loss": 0.0223,
"step": 400
},
{
"epoch": 1.1976047904191618,
"eval_loss": 0.12274425476789474,
"eval_runtime": 8.3056,
"eval_samples_per_second": 6.502,
"eval_steps_per_second": 1.686,
"step": 400
},
{
"epoch": 1.2005988023952097,
"grad_norm": 0.8482845892477424,
"learning_rate": 3.4504429096282246e-06,
"loss": 0.0356,
"step": 401
},
{
"epoch": 1.2035928143712575,
"grad_norm": 0.8532014024276183,
"learning_rate": 3.428102960138625e-06,
"loss": 0.0387,
"step": 402
},
{
"epoch": 1.2065868263473054,
"grad_norm": 0.95367515220183,
"learning_rate": 3.405797777881059e-06,
"loss": 0.0432,
"step": 403
},
{
"epoch": 1.2095808383233533,
"grad_norm": 0.7228544990492038,
"learning_rate": 3.3835278562017405e-06,
"loss": 0.028,
"step": 404
},
{
"epoch": 1.2125748502994012,
"grad_norm": 0.8264683521593323,
"learning_rate": 3.3612936876669834e-06,
"loss": 0.0376,
"step": 405
},
{
"epoch": 1.215568862275449,
"grad_norm": 0.8962344956030094,
"learning_rate": 3.3390957640523147e-06,
"loss": 0.0397,
"step": 406
},
{
"epoch": 1.218562874251497,
"grad_norm": 0.7516331406177484,
"learning_rate": 3.3169345763315986e-06,
"loss": 0.0298,
"step": 407
},
{
"epoch": 1.221556886227545,
"grad_norm": 0.8997824255484812,
"learning_rate": 3.29481061466617e-06,
"loss": 0.043,
"step": 408
},
{
"epoch": 1.2245508982035929,
"grad_norm": 0.9155261698983177,
"learning_rate": 3.2727243683940045e-06,
"loss": 0.0342,
"step": 409
},
{
"epoch": 1.2275449101796407,
"grad_norm": 0.7441360133603632,
"learning_rate": 3.2506763260188824e-06,
"loss": 0.0299,
"step": 410
},
{
"epoch": 1.2305389221556886,
"grad_norm": 0.6227423061521378,
"learning_rate": 3.2286669751995905e-06,
"loss": 0.0272,
"step": 411
},
{
"epoch": 1.2335329341317365,
"grad_norm": 0.8789385458345906,
"learning_rate": 3.2066968027391377e-06,
"loss": 0.0317,
"step": 412
},
{
"epoch": 1.2365269461077844,
"grad_norm": 0.6498772967966782,
"learning_rate": 3.1847662945739833e-06,
"loss": 0.0251,
"step": 413
},
{
"epoch": 1.2395209580838324,
"grad_norm": 0.6806058535848892,
"learning_rate": 3.1628759357632943e-06,
"loss": 0.0272,
"step": 414
},
{
"epoch": 1.2425149700598803,
"grad_norm": 0.7405883946592593,
"learning_rate": 3.1410262104782086e-06,
"loss": 0.0312,
"step": 415
},
{
"epoch": 1.2455089820359282,
"grad_norm": 0.8289539441670357,
"learning_rate": 3.119217601991139e-06,
"loss": 0.0326,
"step": 416
},
{
"epoch": 1.248502994011976,
"grad_norm": 0.8212534436158723,
"learning_rate": 3.0974505926650724e-06,
"loss": 0.0374,
"step": 417
},
{
"epoch": 1.251497005988024,
"grad_norm": 0.9186883114815179,
"learning_rate": 3.0757256639429027e-06,
"loss": 0.0477,
"step": 418
},
{
"epoch": 1.2544910179640718,
"grad_norm": 0.7857164567317839,
"learning_rate": 3.0540432963367907e-06,
"loss": 0.0338,
"step": 419
},
{
"epoch": 1.2574850299401197,
"grad_norm": 0.7416241333181448,
"learning_rate": 3.032403969417523e-06,
"loss": 0.0295,
"step": 420
},
{
"epoch": 1.2604790419161676,
"grad_norm": 0.6296875419732539,
"learning_rate": 3.010808161803917e-06,
"loss": 0.0279,
"step": 421
},
{
"epoch": 1.2634730538922156,
"grad_norm": 0.858388360398613,
"learning_rate": 2.9892563511522305e-06,
"loss": 0.0411,
"step": 422
},
{
"epoch": 1.2664670658682635,
"grad_norm": 0.7311365145504828,
"learning_rate": 2.9677490141455915e-06,
"loss": 0.0289,
"step": 423
},
{
"epoch": 1.2694610778443114,
"grad_norm": 0.8611729516176496,
"learning_rate": 2.946286626483463e-06,
"loss": 0.0369,
"step": 424
},
{
"epoch": 1.2724550898203593,
"grad_norm": 0.8638938444837622,
"learning_rate": 2.924869662871117e-06,
"loss": 0.0331,
"step": 425
},
{
"epoch": 1.2754491017964071,
"grad_norm": 0.7477554937360593,
"learning_rate": 2.903498597009136e-06,
"loss": 0.0363,
"step": 426
},
{
"epoch": 1.278443113772455,
"grad_norm": 0.7671241329830223,
"learning_rate": 2.8821739015829338e-06,
"loss": 0.0333,
"step": 427
},
{
"epoch": 1.281437125748503,
"grad_norm": 0.6121159777799046,
"learning_rate": 2.8608960482523058e-06,
"loss": 0.0242,
"step": 428
},
{
"epoch": 1.284431137724551,
"grad_norm": 0.7573160710717566,
"learning_rate": 2.839665507640992e-06,
"loss": 0.0328,
"step": 429
},
{
"epoch": 1.2874251497005988,
"grad_norm": 0.7164317510192734,
"learning_rate": 2.818482749326272e-06,
"loss": 0.0305,
"step": 430
},
{
"epoch": 1.2904191616766467,
"grad_norm": 0.8103144727787737,
"learning_rate": 2.797348241828569e-06,
"loss": 0.0292,
"step": 431
},
{
"epoch": 1.2934131736526946,
"grad_norm": 0.749983408990304,
"learning_rate": 2.776262452601104e-06,
"loss": 0.0243,
"step": 432
},
{
"epoch": 1.2964071856287425,
"grad_norm": 0.7860892667640146,
"learning_rate": 2.7552258480195348e-06,
"loss": 0.0316,
"step": 433
},
{
"epoch": 1.2994011976047903,
"grad_norm": 0.6696413883450865,
"learning_rate": 2.734238893371667e-06,
"loss": 0.0257,
"step": 434
},
{
"epoch": 1.3023952095808382,
"grad_norm": 0.8456737952611747,
"learning_rate": 2.7133020528471322e-06,
"loss": 0.0363,
"step": 435
},
{
"epoch": 1.3053892215568863,
"grad_norm": 0.7666486409896474,
"learning_rate": 2.6924157895271563e-06,
"loss": 0.0334,
"step": 436
},
{
"epoch": 1.3083832335329342,
"grad_norm": 0.8264949977798605,
"learning_rate": 2.671580565374282e-06,
"loss": 0.0378,
"step": 437
},
{
"epoch": 1.311377245508982,
"grad_norm": 0.8727137273398754,
"learning_rate": 2.6507968412221763e-06,
"loss": 0.0403,
"step": 438
},
{
"epoch": 1.31437125748503,
"grad_norm": 0.8345367260510321,
"learning_rate": 2.6300650767654234e-06,
"loss": 0.031,
"step": 439
},
{
"epoch": 1.3173652694610778,
"grad_norm": 0.8998558702359915,
"learning_rate": 2.6093857305493666e-06,
"loss": 0.0409,
"step": 440
},
{
"epoch": 1.3203592814371259,
"grad_norm": 0.7572176445110357,
"learning_rate": 2.588759259959962e-06,
"loss": 0.0289,
"step": 441
},
{
"epoch": 1.3233532934131738,
"grad_norm": 0.8788703593742704,
"learning_rate": 2.568186121213658e-06,
"loss": 0.0358,
"step": 442
},
{
"epoch": 1.3263473053892216,
"grad_norm": 0.8886184474502267,
"learning_rate": 2.547666769347312e-06,
"loss": 0.0333,
"step": 443
},
{
"epoch": 1.3293413173652695,
"grad_norm": 0.8850229903761628,
"learning_rate": 2.5272016582081236e-06,
"loss": 0.0321,
"step": 444
},
{
"epoch": 1.3323353293413174,
"grad_norm": 0.8534261499675657,
"learning_rate": 2.5067912404435952e-06,
"loss": 0.0357,
"step": 445
},
{
"epoch": 1.3353293413173652,
"grad_norm": 0.967076925657212,
"learning_rate": 2.486435967491516e-06,
"loss": 0.041,
"step": 446
},
{
"epoch": 1.3383233532934131,
"grad_norm": 0.86808042150452,
"learning_rate": 2.4661362895699903e-06,
"loss": 0.0356,
"step": 447
},
{
"epoch": 1.341317365269461,
"grad_norm": 0.9139637793505679,
"learning_rate": 2.445892655667462e-06,
"loss": 0.0368,
"step": 448
},
{
"epoch": 1.3443113772455089,
"grad_norm": 0.9330373787227488,
"learning_rate": 2.425705513532798e-06,
"loss": 0.0438,
"step": 449
},
{
"epoch": 1.347305389221557,
"grad_norm": 0.8994773466836672,
"learning_rate": 2.4055753096653795e-06,
"loss": 0.0445,
"step": 450
},
{
"epoch": 1.3502994011976048,
"grad_norm": 1.0517765784734638,
"learning_rate": 2.3855024893052286e-06,
"loss": 0.0429,
"step": 451
},
{
"epoch": 1.3532934131736527,
"grad_norm": 0.8178120057042947,
"learning_rate": 2.365487496423152e-06,
"loss": 0.035,
"step": 452
},
{
"epoch": 1.3562874251497006,
"grad_norm": 0.8039252618164829,
"learning_rate": 2.3455307737109338e-06,
"loss": 0.0293,
"step": 453
},
{
"epoch": 1.3592814371257484,
"grad_norm": 0.8098474809654538,
"learning_rate": 2.3256327625715345e-06,
"loss": 0.038,
"step": 454
},
{
"epoch": 1.3622754491017965,
"grad_norm": 0.6834159216563565,
"learning_rate": 2.3057939031093346e-06,
"loss": 0.0283,
"step": 455
},
{
"epoch": 1.3652694610778444,
"grad_norm": 0.6657612928843127,
"learning_rate": 2.2860146341203936e-06,
"loss": 0.0287,
"step": 456
},
{
"epoch": 1.3682634730538923,
"grad_norm": 0.8131771805140413,
"learning_rate": 2.2662953930827546e-06,
"loss": 0.0387,
"step": 457
},
{
"epoch": 1.3712574850299402,
"grad_norm": 0.6817770464408984,
"learning_rate": 2.2466366161467528e-06,
"loss": 0.0252,
"step": 458
},
{
"epoch": 1.374251497005988,
"grad_norm": 0.7793573256649142,
"learning_rate": 2.227038738125385e-06,
"loss": 0.0307,
"step": 459
},
{
"epoch": 1.377245508982036,
"grad_norm": 1.0261060760433192,
"learning_rate": 2.207502192484685e-06,
"loss": 0.0383,
"step": 460
},
{
"epoch": 1.3802395209580838,
"grad_norm": 0.6348635641826605,
"learning_rate": 2.188027411334131e-06,
"loss": 0.0228,
"step": 461
},
{
"epoch": 1.3832335329341316,
"grad_norm": 0.8573130652179038,
"learning_rate": 2.1686148254171012e-06,
"loss": 0.0423,
"step": 462
},
{
"epoch": 1.3862275449101795,
"grad_norm": 0.7806582324146677,
"learning_rate": 2.1492648641013305e-06,
"loss": 0.0316,
"step": 463
},
{
"epoch": 1.3892215568862276,
"grad_norm": 0.5894221825824172,
"learning_rate": 2.1299779553694323e-06,
"loss": 0.023,
"step": 464
},
{
"epoch": 1.3922155688622755,
"grad_norm": 0.826535206087488,
"learning_rate": 2.1107545258094135e-06,
"loss": 0.0366,
"step": 465
},
{
"epoch": 1.3952095808383234,
"grad_norm": 0.9616742775987399,
"learning_rate": 2.0915950006052555e-06,
"loss": 0.0301,
"step": 466
},
{
"epoch": 1.3982035928143712,
"grad_norm": 0.729862632476751,
"learning_rate": 2.0724998035274947e-06,
"loss": 0.0267,
"step": 467
},
{
"epoch": 1.401197604790419,
"grad_norm": 0.6637369110872577,
"learning_rate": 2.053469356923865e-06,
"loss": 0.0263,
"step": 468
},
{
"epoch": 1.4041916167664672,
"grad_norm": 0.6619961912456589,
"learning_rate": 2.0345040817099433e-06,
"loss": 0.023,
"step": 469
},
{
"epoch": 1.407185628742515,
"grad_norm": 0.729297579432835,
"learning_rate": 2.0156043973598475e-06,
"loss": 0.0301,
"step": 470
},
{
"epoch": 1.410179640718563,
"grad_norm": 0.7886765442748123,
"learning_rate": 1.996770721896957e-06,
"loss": 0.0257,
"step": 471
},
{
"epoch": 1.4131736526946108,
"grad_norm": 0.6330705774442724,
"learning_rate": 1.9780034718846653e-06,
"loss": 0.0223,
"step": 472
},
{
"epoch": 1.4161676646706587,
"grad_norm": 0.9521481226689832,
"learning_rate": 1.9593030624171683e-06,
"loss": 0.0412,
"step": 473
},
{
"epoch": 1.4191616766467066,
"grad_norm": 0.8013957788507334,
"learning_rate": 1.9406699071102774e-06,
"loss": 0.0349,
"step": 474
},
{
"epoch": 1.4221556886227544,
"grad_norm": 0.611011430738466,
"learning_rate": 1.9221044180922833e-06,
"loss": 0.0225,
"step": 475
},
{
"epoch": 1.4251497005988023,
"grad_norm": 0.847951402483942,
"learning_rate": 1.9036070059948253e-06,
"loss": 0.0337,
"step": 476
},
{
"epoch": 1.4281437125748502,
"grad_norm": 0.914452142873737,
"learning_rate": 1.885178079943823e-06,
"loss": 0.0364,
"step": 477
},
{
"epoch": 1.4311377245508983,
"grad_norm": 0.9317726028752528,
"learning_rate": 1.866818047550419e-06,
"loss": 0.0399,
"step": 478
},
{
"epoch": 1.4341317365269461,
"grad_norm": 0.6902701669651811,
"learning_rate": 1.8485273149019655e-06,
"loss": 0.0278,
"step": 479
},
{
"epoch": 1.437125748502994,
"grad_norm": 0.7767045065124477,
"learning_rate": 1.8303062865530407e-06,
"loss": 0.026,
"step": 480
},
{
"epoch": 1.4401197604790419,
"grad_norm": 0.7623917999011538,
"learning_rate": 1.8121553655165058e-06,
"loss": 0.0286,
"step": 481
},
{
"epoch": 1.4431137724550898,
"grad_norm": 0.717065207465707,
"learning_rate": 1.7940749532545832e-06,
"loss": 0.0293,
"step": 482
},
{
"epoch": 1.4461077844311379,
"grad_norm": 0.8990399187307896,
"learning_rate": 1.7760654496699876e-06,
"loss": 0.0314,
"step": 483
},
{
"epoch": 1.4491017964071857,
"grad_norm": 0.6593536905788724,
"learning_rate": 1.7581272530970666e-06,
"loss": 0.0288,
"step": 484
},
{
"epoch": 1.4520958083832336,
"grad_norm": 0.7110876686323473,
"learning_rate": 1.7402607602930106e-06,
"loss": 0.0315,
"step": 485
},
{
"epoch": 1.4550898203592815,
"grad_norm": 0.855571531618953,
"learning_rate": 1.7224663664290537e-06,
"loss": 0.0328,
"step": 486
},
{
"epoch": 1.4580838323353293,
"grad_norm": 0.822138534764874,
"learning_rate": 1.7047444650817518e-06,
"loss": 0.029,
"step": 487
},
{
"epoch": 1.4610778443113772,
"grad_norm": 0.7985007308245561,
"learning_rate": 1.6870954482242707e-06,
"loss": 0.0278,
"step": 488
},
{
"epoch": 1.464071856287425,
"grad_norm": 0.6529053912134282,
"learning_rate": 1.669519706217711e-06,
"loss": 0.0231,
"step": 489
},
{
"epoch": 1.467065868263473,
"grad_norm": 1.0775658669244046,
"learning_rate": 1.652017627802487e-06,
"loss": 0.0404,
"step": 490
},
{
"epoch": 1.4700598802395208,
"grad_norm": 0.7566624267062195,
"learning_rate": 1.6345896000897122e-06,
"loss": 0.0264,
"step": 491
},
{
"epoch": 1.473053892215569,
"grad_norm": 0.889393201746035,
"learning_rate": 1.6172360085526567e-06,
"loss": 0.0364,
"step": 492
},
{
"epoch": 1.4760479041916168,
"grad_norm": 0.7896130226159147,
"learning_rate": 1.5999572370182016e-06,
"loss": 0.0318,
"step": 493
},
{
"epoch": 1.4790419161676647,
"grad_norm": 0.7833912705448801,
"learning_rate": 1.5827536676583643e-06,
"loss": 0.0321,
"step": 494
},
{
"epoch": 1.4820359281437125,
"grad_norm": 0.7765840124546369,
"learning_rate": 1.5656256809818343e-06,
"loss": 0.0286,
"step": 495
},
{
"epoch": 1.4850299401197604,
"grad_norm": 0.7209927182279341,
"learning_rate": 1.54857365582557e-06,
"loss": 0.0282,
"step": 496
},
{
"epoch": 1.4880239520958085,
"grad_norm": 0.9815242098107917,
"learning_rate": 1.5315979693464039e-06,
"loss": 0.0462,
"step": 497
},
{
"epoch": 1.4910179640718564,
"grad_norm": 0.7869024478324081,
"learning_rate": 1.5146989970127158e-06,
"loss": 0.0263,
"step": 498
},
{
"epoch": 1.4940119760479043,
"grad_norm": 0.8160274706586781,
"learning_rate": 1.4978771125961177e-06,
"loss": 0.0297,
"step": 499
},
{
"epoch": 1.4970059880239521,
"grad_norm": 0.7805482932790326,
"learning_rate": 1.4811326881631937e-06,
"loss": 0.0333,
"step": 500
},
{
"epoch": 1.5,
"grad_norm": 0.9558030047904206,
"learning_rate": 1.4644660940672628e-06,
"loss": 0.0363,
"step": 501
},
{
"epoch": 1.5029940119760479,
"grad_norm": 1.0100842926631917,
"learning_rate": 1.4478776989401949e-06,
"loss": 0.0305,
"step": 502
},
{
"epoch": 1.5059880239520957,
"grad_norm": 0.8435445692139016,
"learning_rate": 1.4313678696842559e-06,
"loss": 0.0316,
"step": 503
},
{
"epoch": 1.5089820359281436,
"grad_norm": 0.7106837250190794,
"learning_rate": 1.4149369714639856e-06,
"loss": 0.0296,
"step": 504
},
{
"epoch": 1.5119760479041915,
"grad_norm": 0.7739948521366974,
"learning_rate": 1.3985853676981316e-06,
"loss": 0.0279,
"step": 505
},
{
"epoch": 1.5149700598802394,
"grad_norm": 0.8117653947864281,
"learning_rate": 1.3823134200516043e-06,
"loss": 0.0309,
"step": 506
},
{
"epoch": 1.5179640718562875,
"grad_norm": 0.9087591195077263,
"learning_rate": 1.366121488427481e-06,
"loss": 0.0384,
"step": 507
},
{
"epoch": 1.5209580838323353,
"grad_norm": 0.7686143275900593,
"learning_rate": 1.3500099309590397e-06,
"loss": 0.0294,
"step": 508
},
{
"epoch": 1.5239520958083832,
"grad_norm": 0.8931968006552831,
"learning_rate": 1.3339791040018479e-06,
"loss": 0.0308,
"step": 509
},
{
"epoch": 1.5269461077844313,
"grad_norm": 0.8868404871926984,
"learning_rate": 1.3180293621258694e-06,
"loss": 0.0353,
"step": 510
},
{
"epoch": 1.5299401197604792,
"grad_norm": 0.758248005845863,
"learning_rate": 1.3021610581076316e-06,
"loss": 0.0297,
"step": 511
},
{
"epoch": 1.532934131736527,
"grad_norm": 0.8672445860829112,
"learning_rate": 1.2863745429224145e-06,
"loss": 0.0288,
"step": 512
},
{
"epoch": 1.535928143712575,
"grad_norm": 0.7675559253844609,
"learning_rate": 1.270670165736499e-06,
"loss": 0.0322,
"step": 513
},
{
"epoch": 1.5389221556886228,
"grad_norm": 0.9548904782356388,
"learning_rate": 1.2550482738994284e-06,
"loss": 0.0402,
"step": 514
},
{
"epoch": 1.5419161676646707,
"grad_norm": 0.8462083353048224,
"learning_rate": 1.239509212936343e-06,
"loss": 0.0404,
"step": 515
},
{
"epoch": 1.5449101796407185,
"grad_norm": 0.7254937332397977,
"learning_rate": 1.22405332654032e-06,
"loss": 0.0283,
"step": 516
},
{
"epoch": 1.5479041916167664,
"grad_norm": 0.8570120599090103,
"learning_rate": 1.2086809565647877e-06,
"loss": 0.0341,
"step": 517
},
{
"epoch": 1.5508982035928143,
"grad_norm": 0.7360277266798015,
"learning_rate": 1.1933924430159571e-06,
"loss": 0.0292,
"step": 518
},
{
"epoch": 1.5538922155688621,
"grad_norm": 0.8117403429158275,
"learning_rate": 1.1781881240452958e-06,
"loss": 0.0287,
"step": 519
},
{
"epoch": 1.55688622754491,
"grad_norm": 0.7776108690426635,
"learning_rate": 1.1630683359420653e-06,
"loss": 0.0336,
"step": 520
},
{
"epoch": 1.5598802395209581,
"grad_norm": 0.7674741930868076,
"learning_rate": 1.1480334131258626e-06,
"loss": 0.0322,
"step": 521
},
{
"epoch": 1.562874251497006,
"grad_norm": 0.9943192478363255,
"learning_rate": 1.1330836881392405e-06,
"loss": 0.0386,
"step": 522
},
{
"epoch": 1.5658682634730539,
"grad_norm": 0.7389979799147285,
"learning_rate": 1.11821949164034e-06,
"loss": 0.0268,
"step": 523
},
{
"epoch": 1.568862275449102,
"grad_norm": 0.6633213086550153,
"learning_rate": 1.103441152395588e-06,
"loss": 0.0237,
"step": 524
},
{
"epoch": 1.5718562874251498,
"grad_norm": 0.7593865064786751,
"learning_rate": 1.088748997272414e-06,
"loss": 0.0289,
"step": 525
},
{
"epoch": 1.5748502994011977,
"grad_norm": 0.9226523367651599,
"learning_rate": 1.0741433512320316e-06,
"loss": 0.034,
"step": 526
},
{
"epoch": 1.5778443113772456,
"grad_norm": 0.73814752667089,
"learning_rate": 1.0596245373222424e-06,
"loss": 0.0248,
"step": 527
},
{
"epoch": 1.5808383233532934,
"grad_norm": 0.8591782256330072,
"learning_rate": 1.045192876670298e-06,
"loss": 0.036,
"step": 528
},
{
"epoch": 1.5838323353293413,
"grad_norm": 0.8078932915516085,
"learning_rate": 1.0308486884757868e-06,
"loss": 0.0288,
"step": 529
},
{
"epoch": 1.5868263473053892,
"grad_norm": 0.8422027954576564,
"learning_rate": 1.0165922900035886e-06,
"loss": 0.032,
"step": 530
},
{
"epoch": 1.589820359281437,
"grad_norm": 0.6402676123892606,
"learning_rate": 1.0024239965768417e-06,
"loss": 0.0284,
"step": 531
},
{
"epoch": 1.592814371257485,
"grad_norm": 0.6932599188526581,
"learning_rate": 9.883441215699824e-07,
"loss": 0.0225,
"step": 532
},
{
"epoch": 1.5958083832335328,
"grad_norm": 0.8265428761855518,
"learning_rate": 9.74352976401805e-07,
"loss": 0.0288,
"step": 533
},
{
"epoch": 1.5988023952095807,
"grad_norm": 0.8598560545298901,
"learning_rate": 9.604508705285765e-07,
"loss": 0.0338,
"step": 534
},
{
"epoch": 1.6017964071856288,
"grad_norm": 0.7342467342475081,
"learning_rate": 9.466381114371942e-07,
"loss": 0.03,
"step": 535
},
{
"epoch": 1.6047904191616766,
"grad_norm": 0.8819889873555795,
"learning_rate": 9.329150046383773e-07,
"loss": 0.037,
"step": 536
},
{
"epoch": 1.6077844311377245,
"grad_norm": 0.6933969738630927,
"learning_rate": 9.192818536599213e-07,
"loss": 0.0266,
"step": 537
},
{
"epoch": 1.6107784431137726,
"grad_norm": 0.7635103122063983,
"learning_rate": 9.057389600399719e-07,
"loss": 0.032,
"step": 538
},
{
"epoch": 1.6137724550898205,
"grad_norm": 0.8093499940109913,
"learning_rate": 8.922866233203681e-07,
"loss": 0.0373,
"step": 539
},
{
"epoch": 1.6167664670658684,
"grad_norm": 0.670879961016856,
"learning_rate": 8.789251410400024e-07,
"loss": 0.0275,
"step": 540
},
{
"epoch": 1.6197604790419162,
"grad_norm": 0.7502571831965572,
"learning_rate": 8.65654808728259e-07,
"loss": 0.0311,
"step": 541
},
{
"epoch": 1.622754491017964,
"grad_norm": 0.7501339998587506,
"learning_rate": 8.524759198984567e-07,
"loss": 0.0265,
"step": 542
},
{
"epoch": 1.625748502994012,
"grad_norm": 0.9965584980768845,
"learning_rate": 8.393887660413719e-07,
"loss": 0.0382,
"step": 543
},
{
"epoch": 1.6287425149700598,
"grad_norm": 1.0608902045933304,
"learning_rate": 8.263936366187825e-07,
"loss": 0.0457,
"step": 544
},
{
"epoch": 1.6317365269461077,
"grad_norm": 0.8491007047520747,
"learning_rate": 8.134908190570723e-07,
"loss": 0.0398,
"step": 545
},
{
"epoch": 1.6347305389221556,
"grad_norm": 0.8087918069570493,
"learning_rate": 8.006805987408705e-07,
"loss": 0.0317,
"step": 546
},
{
"epoch": 1.6377245508982035,
"grad_norm": 0.7380314097200338,
"learning_rate": 7.879632590067354e-07,
"loss": 0.0315,
"step": 547
},
{
"epoch": 1.6407185628742516,
"grad_norm": 0.7152291509501483,
"learning_rate": 7.753390811368972e-07,
"loss": 0.0237,
"step": 548
},
{
"epoch": 1.6437125748502994,
"grad_norm": 0.9032206741121168,
"learning_rate": 7.628083443530287e-07,
"loss": 0.0341,
"step": 549
},
{
"epoch": 1.6467065868263473,
"grad_norm": 0.791470160856038,
"learning_rate": 7.503713258100726e-07,
"loss": 0.0295,
"step": 550
},
{
"epoch": 1.6497005988023952,
"grad_norm": 0.8982642421455997,
"learning_rate": 7.380283005901084e-07,
"loss": 0.0405,
"step": 551
},
{
"epoch": 1.6526946107784433,
"grad_norm": 0.7479241952338301,
"learning_rate": 7.257795416962754e-07,
"loss": 0.0289,
"step": 552
},
{
"epoch": 1.6556886227544911,
"grad_norm": 0.5945504219676864,
"learning_rate": 7.136253200467231e-07,
"loss": 0.0177,
"step": 553
},
{
"epoch": 1.658682634730539,
"grad_norm": 0.7374110072286232,
"learning_rate": 7.015659044686307e-07,
"loss": 0.0244,
"step": 554
},
{
"epoch": 1.6616766467065869,
"grad_norm": 0.7182229108008051,
"learning_rate": 6.896015616922535e-07,
"loss": 0.0271,
"step": 555
},
{
"epoch": 1.6646706586826348,
"grad_norm": 0.8175851271732705,
"learning_rate": 6.777325563450282e-07,
"loss": 0.0325,
"step": 556
},
{
"epoch": 1.6676646706586826,
"grad_norm": 0.8060955271645127,
"learning_rate": 6.659591509457125e-07,
"loss": 0.0309,
"step": 557
},
{
"epoch": 1.6706586826347305,
"grad_norm": 0.749762273978849,
"learning_rate": 6.542816058985896e-07,
"loss": 0.0263,
"step": 558
},
{
"epoch": 1.6736526946107784,
"grad_norm": 0.8326712833404715,
"learning_rate": 6.427001794876974e-07,
"loss": 0.0309,
"step": 559
},
{
"epoch": 1.6766467065868262,
"grad_norm": 0.7299755313881022,
"learning_rate": 6.312151278711237e-07,
"loss": 0.0278,
"step": 560
},
{
"epoch": 1.6796407185628741,
"grad_norm": 0.8931889418235481,
"learning_rate": 6.198267050753387e-07,
"loss": 0.0331,
"step": 561
},
{
"epoch": 1.6826347305389222,
"grad_norm": 0.7047277203420241,
"learning_rate": 6.085351629895736e-07,
"loss": 0.0268,
"step": 562
},
{
"epoch": 1.68562874251497,
"grad_norm": 0.7998674663222014,
"learning_rate": 5.973407513602514e-07,
"loss": 0.0286,
"step": 563
},
{
"epoch": 1.688622754491018,
"grad_norm": 0.9024059485355096,
"learning_rate": 5.862437177854629e-07,
"loss": 0.0367,
"step": 564
},
{
"epoch": 1.6916167664670658,
"grad_norm": 0.8339021894551686,
"learning_rate": 5.752443077094927e-07,
"loss": 0.0339,
"step": 565
},
{
"epoch": 1.694610778443114,
"grad_norm": 0.6966678288107665,
"learning_rate": 5.643427644173838e-07,
"loss": 0.0267,
"step": 566
},
{
"epoch": 1.6976047904191618,
"grad_norm": 0.877469773656579,
"learning_rate": 5.535393290295643e-07,
"loss": 0.0321,
"step": 567
},
{
"epoch": 1.7005988023952097,
"grad_norm": 0.7395916979542717,
"learning_rate": 5.428342404965076e-07,
"loss": 0.0269,
"step": 568
},
{
"epoch": 1.7035928143712575,
"grad_norm": 0.6522625029162095,
"learning_rate": 5.322277355934557e-07,
"loss": 0.0233,
"step": 569
},
{
"epoch": 1.7065868263473054,
"grad_norm": 0.673364217725514,
"learning_rate": 5.217200489151714e-07,
"loss": 0.022,
"step": 570
},
{
"epoch": 1.7095808383233533,
"grad_norm": 0.6746275844631606,
"learning_rate": 5.113114128707592e-07,
"loss": 0.0249,
"step": 571
},
{
"epoch": 1.7125748502994012,
"grad_norm": 0.8565551734630578,
"learning_rate": 5.010020576785174e-07,
"loss": 0.0363,
"step": 572
},
{
"epoch": 1.715568862275449,
"grad_norm": 0.8164502086251159,
"learning_rate": 4.907922113608532e-07,
"loss": 0.0272,
"step": 573
},
{
"epoch": 1.718562874251497,
"grad_norm": 0.8938217759994574,
"learning_rate": 4.806820997392325e-07,
"loss": 0.0342,
"step": 574
},
{
"epoch": 1.7215568862275448,
"grad_norm": 0.6524794927709131,
"learning_rate": 4.7067194642919036e-07,
"loss": 0.0209,
"step": 575
},
{
"epoch": 1.7245508982035929,
"grad_norm": 0.7259599873949754,
"learning_rate": 4.607619728353818e-07,
"loss": 0.026,
"step": 576
},
{
"epoch": 1.7275449101796407,
"grad_norm": 0.8063411056617391,
"learning_rate": 4.50952398146689e-07,
"loss": 0.0306,
"step": 577
},
{
"epoch": 1.7305389221556886,
"grad_norm": 0.7265239467576244,
"learning_rate": 4.4124343933136525e-07,
"loss": 0.0256,
"step": 578
},
{
"epoch": 1.7335329341317365,
"grad_norm": 0.9376866641359151,
"learning_rate": 4.3163531113224466e-07,
"loss": 0.0288,
"step": 579
},
{
"epoch": 1.7365269461077846,
"grad_norm": 0.9201568401188499,
"learning_rate": 4.221282260619891e-07,
"loss": 0.0346,
"step": 580
},
{
"epoch": 1.7395209580838324,
"grad_norm": 0.6363801799371435,
"learning_rate": 4.127223943983849e-07,
"loss": 0.0267,
"step": 581
},
{
"epoch": 1.7425149700598803,
"grad_norm": 0.764412207543806,
"learning_rate": 4.03418024179697e-07,
"loss": 0.0256,
"step": 582
},
{
"epoch": 1.7455089820359282,
"grad_norm": 0.9655544882389034,
"learning_rate": 3.9421532120006544e-07,
"loss": 0.0373,
"step": 583
},
{
"epoch": 1.748502994011976,
"grad_norm": 0.6673030026053577,
"learning_rate": 3.851144890049535e-07,
"loss": 0.0262,
"step": 584
},
{
"epoch": 1.751497005988024,
"grad_norm": 0.6839874633904656,
"learning_rate": 3.761157288866418e-07,
"loss": 0.0252,
"step": 585
},
{
"epoch": 1.7544910179640718,
"grad_norm": 0.7786676439718657,
"learning_rate": 3.672192398797858e-07,
"loss": 0.0308,
"step": 586
},
{
"epoch": 1.7574850299401197,
"grad_norm": 0.7584560601242782,
"learning_rate": 3.58425218757002e-07,
"loss": 0.024,
"step": 587
},
{
"epoch": 1.7604790419161676,
"grad_norm": 0.7510318107946398,
"learning_rate": 3.497338600245254e-07,
"loss": 0.0222,
"step": 588
},
{
"epoch": 1.7634730538922154,
"grad_norm": 0.7250295300427998,
"learning_rate": 3.4114535591790233e-07,
"loss": 0.0251,
"step": 589
},
{
"epoch": 1.7664670658682635,
"grad_norm": 0.8390580850015997,
"learning_rate": 3.326598963977395e-07,
"loss": 0.0239,
"step": 590
},
{
"epoch": 1.7694610778443114,
"grad_norm": 0.6886547360013243,
"learning_rate": 3.242776691455013e-07,
"loss": 0.0242,
"step": 591
},
{
"epoch": 1.7724550898203593,
"grad_norm": 0.978824922086262,
"learning_rate": 3.159988595593616e-07,
"loss": 0.0271,
"step": 592
},
{
"epoch": 1.7754491017964071,
"grad_norm": 0.5475118355720677,
"learning_rate": 3.078236507501015e-07,
"loss": 0.0233,
"step": 593
},
{
"epoch": 1.7784431137724552,
"grad_norm": 0.9053987470674856,
"learning_rate": 2.9975222353705757e-07,
"loss": 0.0411,
"step": 594
},
{
"epoch": 1.781437125748503,
"grad_norm": 0.8040444816314111,
"learning_rate": 2.917847564441256e-07,
"loss": 0.0272,
"step": 595
},
{
"epoch": 1.784431137724551,
"grad_norm": 0.7560893065684454,
"learning_rate": 2.839214256958106e-07,
"loss": 0.0286,
"step": 596
},
{
"epoch": 1.7874251497005988,
"grad_norm": 0.6813116340105736,
"learning_rate": 2.7616240521332884e-07,
"loss": 0.0279,
"step": 597
},
{
"epoch": 1.7904191616766467,
"grad_norm": 0.7818721757749574,
"learning_rate": 2.6850786661076047e-07,
"loss": 0.028,
"step": 598
},
{
"epoch": 1.7934131736526946,
"grad_norm": 0.7833483284532683,
"learning_rate": 2.6095797919125533e-07,
"loss": 0.0301,
"step": 599
},
{
"epoch": 1.7964071856287425,
"grad_norm": 0.8074948988559361,
"learning_rate": 2.5351290994328703e-07,
"loss": 0.032,
"step": 600
},
{
"epoch": 1.7964071856287425,
"eval_loss": 0.11569614708423615,
"eval_runtime": 8.3105,
"eval_samples_per_second": 6.498,
"eval_steps_per_second": 1.685,
"step": 600
},
{
"epoch": 1.7994011976047903,
"grad_norm": 0.7627295967716242,
"learning_rate": 2.4617282353696093e-07,
"loss": 0.0316,
"step": 601
},
{
"epoch": 1.8023952095808382,
"grad_norm": 1.0351380615872117,
"learning_rate": 2.3893788232036807e-07,
"loss": 0.0414,
"step": 602
},
{
"epoch": 1.805389221556886,
"grad_norm": 0.8254307879341827,
"learning_rate": 2.318082463160032e-07,
"loss": 0.0357,
"step": 603
},
{
"epoch": 1.8083832335329342,
"grad_norm": 0.621299796743885,
"learning_rate": 2.2478407321721295e-07,
"loss": 0.0217,
"step": 604
},
{
"epoch": 1.811377245508982,
"grad_norm": 0.7997985341671922,
"learning_rate": 2.1786551838471892e-07,
"loss": 0.0302,
"step": 605
},
{
"epoch": 1.81437125748503,
"grad_norm": 0.7000903319246224,
"learning_rate": 2.1105273484317402e-07,
"loss": 0.0265,
"step": 606
},
{
"epoch": 1.8173652694610778,
"grad_norm": 0.6705282788765942,
"learning_rate": 2.043458732777831e-07,
"loss": 0.0273,
"step": 607
},
{
"epoch": 1.8203592814371259,
"grad_norm": 0.9406181937130907,
"learning_rate": 1.9774508203096843e-07,
"loss": 0.0363,
"step": 608
},
{
"epoch": 1.8233532934131738,
"grad_norm": 0.6881737824493238,
"learning_rate": 1.9125050709908388e-07,
"loss": 0.0212,
"step": 609
},
{
"epoch": 1.8263473053892216,
"grad_norm": 0.7261710446004571,
"learning_rate": 1.8486229212919482e-07,
"loss": 0.0271,
"step": 610
},
{
"epoch": 1.8293413173652695,
"grad_norm": 0.6753904490937951,
"learning_rate": 1.7858057841589281e-07,
"loss": 0.0246,
"step": 611
},
{
"epoch": 1.8323353293413174,
"grad_norm": 0.800972682165409,
"learning_rate": 1.7240550489817652e-07,
"loss": 0.0273,
"step": 612
},
{
"epoch": 1.8353293413173652,
"grad_norm": 0.8856638741023288,
"learning_rate": 1.66337208156373e-07,
"loss": 0.0257,
"step": 613
},
{
"epoch": 1.8383233532934131,
"grad_norm": 0.848123138869985,
"learning_rate": 1.6037582240912175e-07,
"loss": 0.0283,
"step": 614
},
{
"epoch": 1.841317365269461,
"grad_norm": 0.8153695174447103,
"learning_rate": 1.5452147951040165e-07,
"loss": 0.0321,
"step": 615
},
{
"epoch": 1.8443113772455089,
"grad_norm": 0.8507387540239293,
"learning_rate": 1.4877430894662037e-07,
"loss": 0.0319,
"step": 616
},
{
"epoch": 1.8473053892215567,
"grad_norm": 0.7469927587716587,
"learning_rate": 1.4313443783374405e-07,
"loss": 0.0257,
"step": 617
},
{
"epoch": 1.8502994011976048,
"grad_norm": 0.8319162549138669,
"learning_rate": 1.3760199091449045e-07,
"loss": 0.0286,
"step": 618
},
{
"epoch": 1.8532934131736527,
"grad_norm": 0.7497255249934415,
"learning_rate": 1.3217709055556638e-07,
"loss": 0.0293,
"step": 619
},
{
"epoch": 1.8562874251497006,
"grad_norm": 0.7557654251140918,
"learning_rate": 1.268598567449647e-07,
"loss": 0.0294,
"step": 620
},
{
"epoch": 1.8592814371257484,
"grad_norm": 0.7737192110553284,
"learning_rate": 1.2165040708930763e-07,
"loss": 0.0297,
"step": 621
},
{
"epoch": 1.8622754491017965,
"grad_norm": 0.5870476070829563,
"learning_rate": 1.1654885681124661e-07,
"loss": 0.0236,
"step": 622
},
{
"epoch": 1.8652694610778444,
"grad_norm": 0.7595469833939291,
"learning_rate": 1.1155531874691372e-07,
"loss": 0.0263,
"step": 623
},
{
"epoch": 1.8682634730538923,
"grad_norm": 0.7912993151046831,
"learning_rate": 1.0666990334342708e-07,
"loss": 0.0258,
"step": 624
},
{
"epoch": 1.8712574850299402,
"grad_norm": 0.7059884315250772,
"learning_rate": 1.0189271865644445e-07,
"loss": 0.0232,
"step": 625
},
{
"epoch": 1.874251497005988,
"grad_norm": 0.7830213909548226,
"learning_rate": 9.722387034777847e-08,
"loss": 0.0257,
"step": 626
},
{
"epoch": 1.877245508982036,
"grad_norm": 0.8042684494196181,
"learning_rate": 9.266346168305518e-08,
"loss": 0.0327,
"step": 627
},
{
"epoch": 1.8802395209580838,
"grad_norm": 0.7598214971995989,
"learning_rate": 8.821159352943142e-08,
"loss": 0.0247,
"step": 628
},
{
"epoch": 1.8832335329341316,
"grad_norm": 0.6847376918437824,
"learning_rate": 8.38683643533661e-08,
"loss": 0.0228,
"step": 629
},
{
"epoch": 1.8862275449101795,
"grad_norm": 0.7885075330216371,
"learning_rate": 7.963387021843683e-08,
"loss": 0.0225,
"step": 630
},
{
"epoch": 1.8892215568862274,
"grad_norm": 0.8403888633325877,
"learning_rate": 7.550820478322285e-08,
"loss": 0.03,
"step": 631
},
{
"epoch": 1.8922155688622755,
"grad_norm": 0.7839221265900145,
"learning_rate": 7.149145929922607e-08,
"loss": 0.0346,
"step": 632
},
{
"epoch": 1.8952095808383234,
"grad_norm": 0.8825142678915027,
"learning_rate": 6.758372260885714e-08,
"loss": 0.036,
"step": 633
},
{
"epoch": 1.8982035928143712,
"grad_norm": 0.8820324482632153,
"learning_rate": 6.378508114346982e-08,
"loss": 0.0361,
"step": 634
},
{
"epoch": 1.9011976047904193,
"grad_norm": 0.8295764701865402,
"learning_rate": 6.009561892144744e-08,
"loss": 0.0296,
"step": 635
},
{
"epoch": 1.9041916167664672,
"grad_norm": 0.7587867160827685,
"learning_rate": 5.651541754634726e-08,
"loss": 0.0243,
"step": 636
},
{
"epoch": 1.907185628742515,
"grad_norm": 1.1572414445874655,
"learning_rate": 5.304455620509297e-08,
"loss": 0.0407,
"step": 637
},
{
"epoch": 1.910179640718563,
"grad_norm": 0.8213888631978954,
"learning_rate": 4.968311166622553e-08,
"loss": 0.0301,
"step": 638
},
{
"epoch": 1.9131736526946108,
"grad_norm": 0.7684142991073891,
"learning_rate": 4.643115827820399e-08,
"loss": 0.028,
"step": 639
},
{
"epoch": 1.9161676646706587,
"grad_norm": 0.7261099591746644,
"learning_rate": 4.328876796776071e-08,
"loss": 0.0253,
"step": 640
},
{
"epoch": 1.9191616766467066,
"grad_norm": 0.8085083822365345,
"learning_rate": 4.0256010238310936e-08,
"loss": 0.0305,
"step": 641
},
{
"epoch": 1.9221556886227544,
"grad_norm": 0.7885260127800128,
"learning_rate": 3.733295216841626e-08,
"loss": 0.0293,
"step": 642
},
{
"epoch": 1.9251497005988023,
"grad_norm": 0.9966837288767681,
"learning_rate": 3.451965841029914e-08,
"loss": 0.0409,
"step": 643
},
{
"epoch": 1.9281437125748502,
"grad_norm": 0.6527694593103184,
"learning_rate": 3.181619118841517e-08,
"loss": 0.0208,
"step": 644
},
{
"epoch": 1.931137724550898,
"grad_norm": 0.6434419204390688,
"learning_rate": 2.9222610298074717e-08,
"loss": 0.0205,
"step": 645
},
{
"epoch": 1.9341317365269461,
"grad_norm": 0.725497894132429,
"learning_rate": 2.673897310412288e-08,
"loss": 0.0259,
"step": 646
},
{
"epoch": 1.937125748502994,
"grad_norm": 0.6852583371487462,
"learning_rate": 2.4365334539667717e-08,
"loss": 0.0244,
"step": 647
},
{
"epoch": 1.9401197604790419,
"grad_norm": 0.8260510760174733,
"learning_rate": 2.210174710486679e-08,
"loss": 0.0353,
"step": 648
},
{
"epoch": 1.94311377245509,
"grad_norm": 0.5411743202112663,
"learning_rate": 1.99482608657664e-08,
"loss": 0.0201,
"step": 649
},
{
"epoch": 1.9461077844311379,
"grad_norm": 0.6786876049644537,
"learning_rate": 1.7904923453193056e-08,
"loss": 0.0235,
"step": 650
},
{
"epoch": 1.9491017964071857,
"grad_norm": 0.791955901981055,
"learning_rate": 1.5971780061701524e-08,
"loss": 0.0246,
"step": 651
},
{
"epoch": 1.9520958083832336,
"grad_norm": 0.6738215438769756,
"learning_rate": 1.4148873448573408e-08,
"loss": 0.0244,
"step": 652
},
{
"epoch": 1.9550898203592815,
"grad_norm": 0.7286329828267174,
"learning_rate": 1.2436243932872349e-08,
"loss": 0.0262,
"step": 653
},
{
"epoch": 1.9580838323353293,
"grad_norm": 0.7316191284999307,
"learning_rate": 1.0833929394552523e-08,
"loss": 0.0226,
"step": 654
},
{
"epoch": 1.9610778443113772,
"grad_norm": 0.7336140222712861,
"learning_rate": 9.341965273621522e-09,
"loss": 0.0247,
"step": 655
},
{
"epoch": 1.964071856287425,
"grad_norm": 1.0225380383840057,
"learning_rate": 7.96038456935322e-09,
"loss": 0.0316,
"step": 656
},
{
"epoch": 1.967065868263473,
"grad_norm": 0.9926980469443579,
"learning_rate": 6.6892178395611125e-09,
"loss": 0.0371,
"step": 657
},
{
"epoch": 1.9700598802395208,
"grad_norm": 0.8122232095047586,
"learning_rate": 5.528493199922769e-09,
"loss": 0.0296,
"step": 658
},
{
"epoch": 1.9730538922155687,
"grad_norm": 0.6964096373843873,
"learning_rate": 4.478236323355312e-09,
"loss": 0.0269,
"step": 659
},
{
"epoch": 1.9760479041916168,
"grad_norm": 0.6747332635033737,
"learning_rate": 3.538470439448105e-09,
"loss": 0.0235,
"step": 660
},
{
"epoch": 1.9790419161676647,
"grad_norm": 0.9678315204729531,
"learning_rate": 2.709216333952602e-09,
"loss": 0.0336,
"step": 661
},
{
"epoch": 1.9820359281437125,
"grad_norm": 0.7122196382626427,
"learning_rate": 1.9904923483171632e-09,
"loss": 0.0276,
"step": 662
},
{
"epoch": 1.9850299401197606,
"grad_norm": 0.9035083163114916,
"learning_rate": 1.3823143792851545e-09,
"loss": 0.033,
"step": 663
},
{
"epoch": 1.9880239520958085,
"grad_norm": 1.0076961376070062,
"learning_rate": 8.846958785418969e-10,
"loss": 0.0416,
"step": 664
},
{
"epoch": 1.9910179640718564,
"grad_norm": 0.6443600818514333,
"learning_rate": 4.97647852417682e-10,
"loss": 0.0225,
"step": 665
},
{
"epoch": 1.9940119760479043,
"grad_norm": 0.7873480944429055,
"learning_rate": 2.2117886164407797e-10,
"loss": 0.0264,
"step": 666
},
{
"epoch": 1.9970059880239521,
"grad_norm": 0.7450213849077284,
"learning_rate": 5.529502116519148e-11,
"loss": 0.0285,
"step": 667
},
{
"epoch": 2.0,
"grad_norm": 0.4334108045966172,
"learning_rate": 0.0,
"loss": 0.0101,
"step": 668
},
{
"epoch": 2.0,
"step": 668,
"total_flos": 16797260906496.0,
"train_loss": 0.06999007535507222,
"train_runtime": 1799.694,
"train_samples_per_second": 5.922,
"train_steps_per_second": 0.371
}
],
"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": 16797260906496.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}