Qwen2.5-3B-Instruct_Long_CoT / trainer_state.json
TaiGary's picture
Add files using upload-large-folder tool
f5f8437 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": 4.91261143168503,
"learning_rate": 9.999944704978835e-06,
"loss": 0.6311,
"step": 1
},
{
"epoch": 0.005988023952095809,
"grad_norm": 3.079935758743856,
"learning_rate": 9.999778821138357e-06,
"loss": 0.5538,
"step": 2
},
{
"epoch": 0.008982035928143712,
"grad_norm": 2.4806703282030407,
"learning_rate": 9.999502352147583e-06,
"loss": 0.5501,
"step": 3
},
{
"epoch": 0.011976047904191617,
"grad_norm": 2.402200344613085,
"learning_rate": 9.999115304121459e-06,
"loss": 0.5719,
"step": 4
},
{
"epoch": 0.014970059880239521,
"grad_norm": 1.55389376038889,
"learning_rate": 9.998617685620715e-06,
"loss": 0.4352,
"step": 5
},
{
"epoch": 0.017964071856287425,
"grad_norm": 1.6003949600530698,
"learning_rate": 9.998009507651683e-06,
"loss": 0.4957,
"step": 6
},
{
"epoch": 0.020958083832335328,
"grad_norm": 1.4666803481379036,
"learning_rate": 9.997290783666048e-06,
"loss": 0.4431,
"step": 7
},
{
"epoch": 0.023952095808383235,
"grad_norm": 1.4652448166283036,
"learning_rate": 9.996461529560553e-06,
"loss": 0.4228,
"step": 8
},
{
"epoch": 0.02694610778443114,
"grad_norm": 1.5500236024494973,
"learning_rate": 9.995521763676645e-06,
"loss": 0.4749,
"step": 9
},
{
"epoch": 0.029940119760479042,
"grad_norm": 1.4562895323881662,
"learning_rate": 9.994471506800078e-06,
"loss": 0.432,
"step": 10
},
{
"epoch": 0.03293413173652695,
"grad_norm": 1.245903316284134,
"learning_rate": 9.993310782160439e-06,
"loss": 0.4651,
"step": 11
},
{
"epoch": 0.03592814371257485,
"grad_norm": 1.2831508715805897,
"learning_rate": 9.992039615430648e-06,
"loss": 0.4341,
"step": 12
},
{
"epoch": 0.038922155688622756,
"grad_norm": 1.3268731286255395,
"learning_rate": 9.99065803472638e-06,
"loss": 0.408,
"step": 13
},
{
"epoch": 0.041916167664670656,
"grad_norm": 1.2361753019228416,
"learning_rate": 9.989166070605447e-06,
"loss": 0.3972,
"step": 14
},
{
"epoch": 0.04491017964071856,
"grad_norm": 1.1216450793208277,
"learning_rate": 9.98756375606713e-06,
"loss": 0.3816,
"step": 15
},
{
"epoch": 0.04790419161676647,
"grad_norm": 1.2668706176835114,
"learning_rate": 9.985851126551428e-06,
"loss": 0.4194,
"step": 16
},
{
"epoch": 0.05089820359281437,
"grad_norm": 0.9512491812885194,
"learning_rate": 9.9840282199383e-06,
"loss": 0.3426,
"step": 17
},
{
"epoch": 0.05389221556886228,
"grad_norm": 1.039509843648798,
"learning_rate": 9.982095076546806e-06,
"loss": 0.3923,
"step": 18
},
{
"epoch": 0.05688622754491018,
"grad_norm": 1.19906673976129,
"learning_rate": 9.980051739134235e-06,
"loss": 0.4012,
"step": 19
},
{
"epoch": 0.059880239520958084,
"grad_norm": 1.2427130816226524,
"learning_rate": 9.977898252895133e-06,
"loss": 0.4049,
"step": 20
},
{
"epoch": 0.06287425149700598,
"grad_norm": 1.0442932060471062,
"learning_rate": 9.975634665460333e-06,
"loss": 0.3869,
"step": 21
},
{
"epoch": 0.0658682634730539,
"grad_norm": 1.158734120939648,
"learning_rate": 9.973261026895878e-06,
"loss": 0.3548,
"step": 22
},
{
"epoch": 0.0688622754491018,
"grad_norm": 1.0658889930864897,
"learning_rate": 9.970777389701927e-06,
"loss": 0.4334,
"step": 23
},
{
"epoch": 0.0718562874251497,
"grad_norm": 1.064193968007788,
"learning_rate": 9.968183808811586e-06,
"loss": 0.3431,
"step": 24
},
{
"epoch": 0.0748502994011976,
"grad_norm": 1.249133492058183,
"learning_rate": 9.965480341589702e-06,
"loss": 0.41,
"step": 25
},
{
"epoch": 0.07784431137724551,
"grad_norm": 1.1261218533693476,
"learning_rate": 9.962667047831585e-06,
"loss": 0.4098,
"step": 26
},
{
"epoch": 0.08083832335329341,
"grad_norm": 1.1132872190680627,
"learning_rate": 9.95974398976169e-06,
"loss": 0.3556,
"step": 27
},
{
"epoch": 0.08383233532934131,
"grad_norm": 1.1273818023064464,
"learning_rate": 9.95671123203224e-06,
"loss": 0.3973,
"step": 28
},
{
"epoch": 0.08682634730538923,
"grad_norm": 1.1634311044486243,
"learning_rate": 9.953568841721796e-06,
"loss": 0.4234,
"step": 29
},
{
"epoch": 0.08982035928143713,
"grad_norm": 0.9581166328946712,
"learning_rate": 9.950316888333775e-06,
"loss": 0.3456,
"step": 30
},
{
"epoch": 0.09281437125748503,
"grad_norm": 1.0350908991302805,
"learning_rate": 9.946955443794908e-06,
"loss": 0.4164,
"step": 31
},
{
"epoch": 0.09580838323353294,
"grad_norm": 1.2354900659793489,
"learning_rate": 9.943484582453653e-06,
"loss": 0.3919,
"step": 32
},
{
"epoch": 0.09880239520958084,
"grad_norm": 0.9485683661933684,
"learning_rate": 9.939904381078553e-06,
"loss": 0.3405,
"step": 33
},
{
"epoch": 0.10179640718562874,
"grad_norm": 0.897664616052441,
"learning_rate": 9.93621491885653e-06,
"loss": 0.3209,
"step": 34
},
{
"epoch": 0.10479041916167664,
"grad_norm": 1.1191710224059541,
"learning_rate": 9.932416277391144e-06,
"loss": 0.3682,
"step": 35
},
{
"epoch": 0.10778443113772455,
"grad_norm": 1.1245626606228858,
"learning_rate": 9.928508540700775e-06,
"loss": 0.3785,
"step": 36
},
{
"epoch": 0.11077844311377245,
"grad_norm": 1.1578039833064768,
"learning_rate": 9.924491795216777e-06,
"loss": 0.4176,
"step": 37
},
{
"epoch": 0.11377245508982035,
"grad_norm": 1.126559636716115,
"learning_rate": 9.920366129781564e-06,
"loss": 0.3853,
"step": 38
},
{
"epoch": 0.11676646706586827,
"grad_norm": 0.8909515320256748,
"learning_rate": 9.916131635646635e-06,
"loss": 0.3398,
"step": 39
},
{
"epoch": 0.11976047904191617,
"grad_norm": 1.1540262836420119,
"learning_rate": 9.91178840647057e-06,
"loss": 0.4112,
"step": 40
},
{
"epoch": 0.12275449101796407,
"grad_norm": 1.146975384229289,
"learning_rate": 9.907336538316946e-06,
"loss": 0.3665,
"step": 41
},
{
"epoch": 0.12574850299401197,
"grad_norm": 0.9372577232612972,
"learning_rate": 9.902776129652223e-06,
"loss": 0.3186,
"step": 42
},
{
"epoch": 0.12874251497005987,
"grad_norm": 1.0710871326759046,
"learning_rate": 9.898107281343557e-06,
"loss": 0.3258,
"step": 43
},
{
"epoch": 0.1317365269461078,
"grad_norm": 1.0451378044082003,
"learning_rate": 9.893330096656576e-06,
"loss": 0.3045,
"step": 44
},
{
"epoch": 0.1347305389221557,
"grad_norm": 1.0965195401318097,
"learning_rate": 9.888444681253087e-06,
"loss": 0.3812,
"step": 45
},
{
"epoch": 0.1377245508982036,
"grad_norm": 1.1950632141982052,
"learning_rate": 9.883451143188753e-06,
"loss": 0.3722,
"step": 46
},
{
"epoch": 0.1407185628742515,
"grad_norm": 0.9745884013764664,
"learning_rate": 9.878349592910694e-06,
"loss": 0.3091,
"step": 47
},
{
"epoch": 0.1437125748502994,
"grad_norm": 1.1037122550379987,
"learning_rate": 9.873140143255035e-06,
"loss": 0.3947,
"step": 48
},
{
"epoch": 0.1467065868263473,
"grad_norm": 1.056210524736011,
"learning_rate": 9.867822909444435e-06,
"loss": 0.3893,
"step": 49
},
{
"epoch": 0.1497005988023952,
"grad_norm": 1.0679680106300733,
"learning_rate": 9.862398009085511e-06,
"loss": 0.3635,
"step": 50
},
{
"epoch": 0.15269461077844312,
"grad_norm": 1.068096172291827,
"learning_rate": 9.856865562166256e-06,
"loss": 0.348,
"step": 51
},
{
"epoch": 0.15568862275449102,
"grad_norm": 0.9707150492343937,
"learning_rate": 9.851225691053382e-06,
"loss": 0.3171,
"step": 52
},
{
"epoch": 0.15868263473053892,
"grad_norm": 1.039032789279636,
"learning_rate": 9.8454785204896e-06,
"loss": 0.3816,
"step": 53
},
{
"epoch": 0.16167664670658682,
"grad_norm": 0.9152192117888228,
"learning_rate": 9.83962417759088e-06,
"loss": 0.3247,
"step": 54
},
{
"epoch": 0.16467065868263472,
"grad_norm": 1.2080523419571483,
"learning_rate": 9.833662791843628e-06,
"loss": 0.3263,
"step": 55
},
{
"epoch": 0.16766467065868262,
"grad_norm": 1.0331644184182198,
"learning_rate": 9.827594495101824e-06,
"loss": 0.3463,
"step": 56
},
{
"epoch": 0.17065868263473055,
"grad_norm": 1.086334967865921,
"learning_rate": 9.821419421584108e-06,
"loss": 0.3587,
"step": 57
},
{
"epoch": 0.17365269461077845,
"grad_norm": 1.0190532026723929,
"learning_rate": 9.815137707870806e-06,
"loss": 0.3913,
"step": 58
},
{
"epoch": 0.17664670658682635,
"grad_norm": 1.1332968591403225,
"learning_rate": 9.808749492900917e-06,
"loss": 0.3543,
"step": 59
},
{
"epoch": 0.17964071856287425,
"grad_norm": 0.9763217978195974,
"learning_rate": 9.802254917969033e-06,
"loss": 0.3621,
"step": 60
},
{
"epoch": 0.18263473053892215,
"grad_norm": 1.1504220571862578,
"learning_rate": 9.795654126722218e-06,
"loss": 0.3591,
"step": 61
},
{
"epoch": 0.18562874251497005,
"grad_norm": 0.9610745527405582,
"learning_rate": 9.788947265156828e-06,
"loss": 0.3414,
"step": 62
},
{
"epoch": 0.18862275449101795,
"grad_norm": 1.0290800255781967,
"learning_rate": 9.782134481615282e-06,
"loss": 0.3158,
"step": 63
},
{
"epoch": 0.19161676646706588,
"grad_norm": 1.0374248546763436,
"learning_rate": 9.775215926782788e-06,
"loss": 0.3797,
"step": 64
},
{
"epoch": 0.19461077844311378,
"grad_norm": 1.103902156800955,
"learning_rate": 9.768191753683997e-06,
"loss": 0.3414,
"step": 65
},
{
"epoch": 0.19760479041916168,
"grad_norm": 1.0496226358107943,
"learning_rate": 9.761062117679632e-06,
"loss": 0.339,
"step": 66
},
{
"epoch": 0.20059880239520958,
"grad_norm": 0.9813080505471711,
"learning_rate": 9.75382717646304e-06,
"loss": 0.321,
"step": 67
},
{
"epoch": 0.20359281437125748,
"grad_norm": 1.1031646383631348,
"learning_rate": 9.746487090056712e-06,
"loss": 0.4029,
"step": 68
},
{
"epoch": 0.20658682634730538,
"grad_norm": 0.9148799592866239,
"learning_rate": 9.739042020808746e-06,
"loss": 0.3044,
"step": 69
},
{
"epoch": 0.20958083832335328,
"grad_norm": 1.082892394526267,
"learning_rate": 9.73149213338924e-06,
"loss": 0.3855,
"step": 70
},
{
"epoch": 0.2125748502994012,
"grad_norm": 1.0160316179650355,
"learning_rate": 9.72383759478667e-06,
"loss": 0.3837,
"step": 71
},
{
"epoch": 0.2155688622754491,
"grad_norm": 1.0015944938864891,
"learning_rate": 9.71607857430419e-06,
"loss": 0.2923,
"step": 72
},
{
"epoch": 0.218562874251497,
"grad_norm": 1.1725178382839894,
"learning_rate": 9.708215243555875e-06,
"loss": 0.3739,
"step": 73
},
{
"epoch": 0.2215568862275449,
"grad_norm": 1.233827459046386,
"learning_rate": 9.700247776462944e-06,
"loss": 0.3588,
"step": 74
},
{
"epoch": 0.2245508982035928,
"grad_norm": 1.1348637863327076,
"learning_rate": 9.6921763492499e-06,
"loss": 0.3454,
"step": 75
},
{
"epoch": 0.2275449101796407,
"grad_norm": 0.9577445731509752,
"learning_rate": 9.68400114044064e-06,
"loss": 0.2805,
"step": 76
},
{
"epoch": 0.23053892215568864,
"grad_norm": 1.0549085314415414,
"learning_rate": 9.6757223308545e-06,
"loss": 0.3505,
"step": 77
},
{
"epoch": 0.23353293413173654,
"grad_norm": 0.9776363369616884,
"learning_rate": 9.667340103602263e-06,
"loss": 0.3961,
"step": 78
},
{
"epoch": 0.23652694610778444,
"grad_norm": 0.9877005565756038,
"learning_rate": 9.658854644082099e-06,
"loss": 0.3193,
"step": 79
},
{
"epoch": 0.23952095808383234,
"grad_norm": 1.3455819767065038,
"learning_rate": 9.650266139975474e-06,
"loss": 0.3822,
"step": 80
},
{
"epoch": 0.24251497005988024,
"grad_norm": 1.130723965906319,
"learning_rate": 9.641574781242999e-06,
"loss": 0.3366,
"step": 81
},
{
"epoch": 0.24550898203592814,
"grad_norm": 1.0437809706564194,
"learning_rate": 9.632780760120217e-06,
"loss": 0.3472,
"step": 82
},
{
"epoch": 0.24850299401197604,
"grad_norm": 1.0117093613939214,
"learning_rate": 9.62388427111336e-06,
"loss": 0.3222,
"step": 83
},
{
"epoch": 0.25149700598802394,
"grad_norm": 0.926822717528161,
"learning_rate": 9.614885510995047e-06,
"loss": 0.2907,
"step": 84
},
{
"epoch": 0.25449101796407186,
"grad_norm": 1.1364455920905385,
"learning_rate": 9.605784678799934e-06,
"loss": 0.3376,
"step": 85
},
{
"epoch": 0.25748502994011974,
"grad_norm": 1.0593146445840056,
"learning_rate": 9.596581975820304e-06,
"loss": 0.3422,
"step": 86
},
{
"epoch": 0.26047904191616766,
"grad_norm": 1.159159788496487,
"learning_rate": 9.587277605601617e-06,
"loss": 0.3661,
"step": 87
},
{
"epoch": 0.2634730538922156,
"grad_norm": 1.2008727080936834,
"learning_rate": 9.577871773938013e-06,
"loss": 0.3814,
"step": 88
},
{
"epoch": 0.26646706586826346,
"grad_norm": 1.0821098267893228,
"learning_rate": 9.568364688867757e-06,
"loss": 0.4292,
"step": 89
},
{
"epoch": 0.2694610778443114,
"grad_norm": 1.071232571082485,
"learning_rate": 9.558756560668637e-06,
"loss": 0.3247,
"step": 90
},
{
"epoch": 0.27245508982035926,
"grad_norm": 1.155397923217811,
"learning_rate": 9.549047601853313e-06,
"loss": 0.3392,
"step": 91
},
{
"epoch": 0.2754491017964072,
"grad_norm": 1.2172469728101094,
"learning_rate": 9.539238027164618e-06,
"loss": 0.362,
"step": 92
},
{
"epoch": 0.27844311377245506,
"grad_norm": 1.07947615730049,
"learning_rate": 9.52932805357081e-06,
"loss": 0.3433,
"step": 93
},
{
"epoch": 0.281437125748503,
"grad_norm": 1.1214966242158426,
"learning_rate": 9.519317900260769e-06,
"loss": 0.3754,
"step": 94
},
{
"epoch": 0.2844311377245509,
"grad_norm": 1.077053655515359,
"learning_rate": 9.509207788639148e-06,
"loss": 0.3699,
"step": 95
},
{
"epoch": 0.2874251497005988,
"grad_norm": 1.0321273316386002,
"learning_rate": 9.498997942321484e-06,
"loss": 0.3672,
"step": 96
},
{
"epoch": 0.2904191616766467,
"grad_norm": 0.9072455949755738,
"learning_rate": 9.488688587129243e-06,
"loss": 0.3791,
"step": 97
},
{
"epoch": 0.2934131736526946,
"grad_norm": 1.2127485055542233,
"learning_rate": 9.47827995108483e-06,
"loss": 0.3298,
"step": 98
},
{
"epoch": 0.2964071856287425,
"grad_norm": 1.161395597395874,
"learning_rate": 9.467772264406545e-06,
"loss": 0.3475,
"step": 99
},
{
"epoch": 0.2994011976047904,
"grad_norm": 0.942225229827603,
"learning_rate": 9.457165759503492e-06,
"loss": 0.3136,
"step": 100
},
{
"epoch": 0.3023952095808383,
"grad_norm": 1.0649013972998636,
"learning_rate": 9.446460670970436e-06,
"loss": 0.3252,
"step": 101
},
{
"epoch": 0.30538922155688625,
"grad_norm": 1.0737417454271467,
"learning_rate": 9.435657235582616e-06,
"loss": 0.2783,
"step": 102
},
{
"epoch": 0.3083832335329341,
"grad_norm": 1.1365571075806886,
"learning_rate": 9.424755692290507e-06,
"loss": 0.3469,
"step": 103
},
{
"epoch": 0.31137724550898205,
"grad_norm": 1.182051237376964,
"learning_rate": 9.413756282214538e-06,
"loss": 0.3785,
"step": 104
},
{
"epoch": 0.3143712574850299,
"grad_norm": 0.9895679323108517,
"learning_rate": 9.402659248639749e-06,
"loss": 0.3401,
"step": 105
},
{
"epoch": 0.31736526946107785,
"grad_norm": 1.0386951217106888,
"learning_rate": 9.391464837010428e-06,
"loss": 0.3319,
"step": 106
},
{
"epoch": 0.3203592814371258,
"grad_norm": 1.1503033016104387,
"learning_rate": 9.380173294924661e-06,
"loss": 0.3806,
"step": 107
},
{
"epoch": 0.32335329341317365,
"grad_norm": 1.0511544596958022,
"learning_rate": 9.368784872128877e-06,
"loss": 0.3414,
"step": 108
},
{
"epoch": 0.3263473053892216,
"grad_norm": 1.0043880055292935,
"learning_rate": 9.357299820512305e-06,
"loss": 0.328,
"step": 109
},
{
"epoch": 0.32934131736526945,
"grad_norm": 1.0401628794403335,
"learning_rate": 9.345718394101412e-06,
"loss": 0.327,
"step": 110
},
{
"epoch": 0.3323353293413174,
"grad_norm": 1.0459814040592879,
"learning_rate": 9.334040849054288e-06,
"loss": 0.368,
"step": 111
},
{
"epoch": 0.33532934131736525,
"grad_norm": 1.096034049623613,
"learning_rate": 9.322267443654974e-06,
"loss": 0.3404,
"step": 112
},
{
"epoch": 0.3383233532934132,
"grad_norm": 1.0649017355426917,
"learning_rate": 9.310398438307747e-06,
"loss": 0.3816,
"step": 113
},
{
"epoch": 0.3413173652694611,
"grad_norm": 1.0146503349030291,
"learning_rate": 9.29843409553137e-06,
"loss": 0.3157,
"step": 114
},
{
"epoch": 0.344311377245509,
"grad_norm": 0.9779937869507201,
"learning_rate": 9.286374679953278e-06,
"loss": 0.3211,
"step": 115
},
{
"epoch": 0.3473053892215569,
"grad_norm": 1.0157039181199519,
"learning_rate": 9.274220458303727e-06,
"loss": 0.2954,
"step": 116
},
{
"epoch": 0.3502994011976048,
"grad_norm": 1.029658161553897,
"learning_rate": 9.261971699409893e-06,
"loss": 0.3549,
"step": 117
},
{
"epoch": 0.3532934131736527,
"grad_norm": 1.1167885932808472,
"learning_rate": 9.249628674189928e-06,
"loss": 0.3393,
"step": 118
},
{
"epoch": 0.3562874251497006,
"grad_norm": 1.0575164285464933,
"learning_rate": 9.237191655646972e-06,
"loss": 0.3638,
"step": 119
},
{
"epoch": 0.3592814371257485,
"grad_norm": 1.1148178255174808,
"learning_rate": 9.224660918863104e-06,
"loss": 0.3956,
"step": 120
},
{
"epoch": 0.36227544910179643,
"grad_norm": 1.143949167554938,
"learning_rate": 9.212036740993265e-06,
"loss": 0.3479,
"step": 121
},
{
"epoch": 0.3652694610778443,
"grad_norm": 1.0870366101583933,
"learning_rate": 9.199319401259132e-06,
"loss": 0.3445,
"step": 122
},
{
"epoch": 0.36826347305389223,
"grad_norm": 1.1021468601811135,
"learning_rate": 9.186509180942928e-06,
"loss": 0.323,
"step": 123
},
{
"epoch": 0.3712574850299401,
"grad_norm": 0.9935888113526302,
"learning_rate": 9.173606363381218e-06,
"loss": 0.3802,
"step": 124
},
{
"epoch": 0.37425149700598803,
"grad_norm": 1.0651070646388379,
"learning_rate": 9.16061123395863e-06,
"loss": 0.3442,
"step": 125
},
{
"epoch": 0.3772455089820359,
"grad_norm": 1.0559416156034385,
"learning_rate": 9.147524080101543e-06,
"loss": 0.3397,
"step": 126
},
{
"epoch": 0.38023952095808383,
"grad_norm": 1.1605256685521508,
"learning_rate": 9.134345191271742e-06,
"loss": 0.3929,
"step": 127
},
{
"epoch": 0.38323353293413176,
"grad_norm": 1.0809721511662522,
"learning_rate": 9.121074858959997e-06,
"loss": 0.3736,
"step": 128
},
{
"epoch": 0.38622754491017963,
"grad_norm": 0.9692547638180142,
"learning_rate": 9.107713376679634e-06,
"loss": 0.3472,
"step": 129
},
{
"epoch": 0.38922155688622756,
"grad_norm": 1.1679545846686712,
"learning_rate": 9.094261039960028e-06,
"loss": 0.3537,
"step": 130
},
{
"epoch": 0.39221556886227543,
"grad_norm": 1.2370946226908197,
"learning_rate": 9.08071814634008e-06,
"loss": 0.4035,
"step": 131
},
{
"epoch": 0.39520958083832336,
"grad_norm": 1.1131347916712,
"learning_rate": 9.067084995361623e-06,
"loss": 0.3498,
"step": 132
},
{
"epoch": 0.39820359281437123,
"grad_norm": 1.1049485919923456,
"learning_rate": 9.053361888562807e-06,
"loss": 0.3834,
"step": 133
},
{
"epoch": 0.40119760479041916,
"grad_norm": 1.109256880719259,
"learning_rate": 9.039549129471423e-06,
"loss": 0.3339,
"step": 134
},
{
"epoch": 0.4041916167664671,
"grad_norm": 0.9207948685814955,
"learning_rate": 9.025647023598196e-06,
"loss": 0.3122,
"step": 135
},
{
"epoch": 0.40718562874251496,
"grad_norm": 1.03019012301144,
"learning_rate": 9.011655878430018e-06,
"loss": 0.32,
"step": 136
},
{
"epoch": 0.4101796407185629,
"grad_norm": 1.1491451207819028,
"learning_rate": 8.99757600342316e-06,
"loss": 0.3811,
"step": 137
},
{
"epoch": 0.41317365269461076,
"grad_norm": 1.11056828624651,
"learning_rate": 8.983407709996415e-06,
"loss": 0.3601,
"step": 138
},
{
"epoch": 0.4161676646706587,
"grad_norm": 1.094910199012047,
"learning_rate": 8.969151311524215e-06,
"loss": 0.3358,
"step": 139
},
{
"epoch": 0.41916167664670656,
"grad_norm": 1.1842978903623054,
"learning_rate": 8.954807123329703e-06,
"loss": 0.3657,
"step": 140
},
{
"epoch": 0.4221556886227545,
"grad_norm": 0.9280456153489471,
"learning_rate": 8.940375462677758e-06,
"loss": 0.2966,
"step": 141
},
{
"epoch": 0.4251497005988024,
"grad_norm": 1.0468297436062597,
"learning_rate": 8.92585664876797e-06,
"loss": 0.3345,
"step": 142
},
{
"epoch": 0.4281437125748503,
"grad_norm": 0.9548563799976826,
"learning_rate": 8.911251002727588e-06,
"loss": 0.3278,
"step": 143
},
{
"epoch": 0.4311377245508982,
"grad_norm": 0.9683579283478643,
"learning_rate": 8.896558847604414e-06,
"loss": 0.3141,
"step": 144
},
{
"epoch": 0.4341317365269461,
"grad_norm": 1.0219576053120933,
"learning_rate": 8.881780508359661e-06,
"loss": 0.4082,
"step": 145
},
{
"epoch": 0.437125748502994,
"grad_norm": 1.3094011242848036,
"learning_rate": 8.86691631186076e-06,
"loss": 0.3478,
"step": 146
},
{
"epoch": 0.44011976047904194,
"grad_norm": 0.8961118344156233,
"learning_rate": 8.851966586874138e-06,
"loss": 0.3001,
"step": 147
},
{
"epoch": 0.4431137724550898,
"grad_norm": 1.0047204821449822,
"learning_rate": 8.836931664057935e-06,
"loss": 0.3247,
"step": 148
},
{
"epoch": 0.44610778443113774,
"grad_norm": 1.1468341051542361,
"learning_rate": 8.821811875954705e-06,
"loss": 0.3474,
"step": 149
},
{
"epoch": 0.4491017964071856,
"grad_norm": 1.0057107776993626,
"learning_rate": 8.806607556984045e-06,
"loss": 0.2974,
"step": 150
},
{
"epoch": 0.45209580838323354,
"grad_norm": 1.0073623601237498,
"learning_rate": 8.791319043435213e-06,
"loss": 0.3361,
"step": 151
},
{
"epoch": 0.4550898203592814,
"grad_norm": 0.940551225988468,
"learning_rate": 8.775946673459682e-06,
"loss": 0.3374,
"step": 152
},
{
"epoch": 0.45808383233532934,
"grad_norm": 1.0119678442207494,
"learning_rate": 8.76049078706366e-06,
"loss": 0.3102,
"step": 153
},
{
"epoch": 0.46107784431137727,
"grad_norm": 1.0163499886944372,
"learning_rate": 8.744951726100572e-06,
"loss": 0.2988,
"step": 154
},
{
"epoch": 0.46407185628742514,
"grad_norm": 1.1224976891335088,
"learning_rate": 8.729329834263503e-06,
"loss": 0.3271,
"step": 155
},
{
"epoch": 0.46706586826347307,
"grad_norm": 1.1195231586617014,
"learning_rate": 8.713625457077585e-06,
"loss": 0.3529,
"step": 156
},
{
"epoch": 0.47005988023952094,
"grad_norm": 1.0532930175781698,
"learning_rate": 8.697838941892371e-06,
"loss": 0.3188,
"step": 157
},
{
"epoch": 0.47305389221556887,
"grad_norm": 1.0403301834877203,
"learning_rate": 8.681970637874131e-06,
"loss": 0.3681,
"step": 158
},
{
"epoch": 0.47604790419161674,
"grad_norm": 1.0968423395413482,
"learning_rate": 8.666020895998154e-06,
"loss": 0.3385,
"step": 159
},
{
"epoch": 0.47904191616766467,
"grad_norm": 1.1411418580606074,
"learning_rate": 8.64999006904096e-06,
"loss": 0.3552,
"step": 160
},
{
"epoch": 0.4820359281437126,
"grad_norm": 1.1284898897168192,
"learning_rate": 8.63387851157252e-06,
"loss": 0.3526,
"step": 161
},
{
"epoch": 0.48502994011976047,
"grad_norm": 1.0118870515654865,
"learning_rate": 8.617686579948396e-06,
"loss": 0.3043,
"step": 162
},
{
"epoch": 0.4880239520958084,
"grad_norm": 1.129436369714682,
"learning_rate": 8.60141463230187e-06,
"loss": 0.31,
"step": 163
},
{
"epoch": 0.49101796407185627,
"grad_norm": 1.169888912858332,
"learning_rate": 8.585063028536015e-06,
"loss": 0.3991,
"step": 164
},
{
"epoch": 0.4940119760479042,
"grad_norm": 1.0012678858420474,
"learning_rate": 8.568632130315747e-06,
"loss": 0.3401,
"step": 165
},
{
"epoch": 0.49700598802395207,
"grad_norm": 1.039062271716469,
"learning_rate": 8.552122301059807e-06,
"loss": 0.3312,
"step": 166
},
{
"epoch": 0.5,
"grad_norm": 1.0208370649023404,
"learning_rate": 8.535533905932739e-06,
"loss": 0.3174,
"step": 167
},
{
"epoch": 0.5029940119760479,
"grad_norm": 1.0815869317765068,
"learning_rate": 8.518867311836808e-06,
"loss": 0.3645,
"step": 168
},
{
"epoch": 0.5059880239520959,
"grad_norm": 1.0937634698755223,
"learning_rate": 8.502122887403882e-06,
"loss": 0.3723,
"step": 169
},
{
"epoch": 0.5089820359281437,
"grad_norm": 1.088050426988013,
"learning_rate": 8.485301002987285e-06,
"loss": 0.2992,
"step": 170
},
{
"epoch": 0.5119760479041916,
"grad_norm": 1.0883631371503362,
"learning_rate": 8.468402030653598e-06,
"loss": 0.3814,
"step": 171
},
{
"epoch": 0.5149700598802395,
"grad_norm": 0.9229260190200125,
"learning_rate": 8.451426344174433e-06,
"loss": 0.32,
"step": 172
},
{
"epoch": 0.5179640718562875,
"grad_norm": 1.0908774976253133,
"learning_rate": 8.434374319018165e-06,
"loss": 0.3906,
"step": 173
},
{
"epoch": 0.5209580838323353,
"grad_norm": 0.9855776743528881,
"learning_rate": 8.417246332341638e-06,
"loss": 0.322,
"step": 174
},
{
"epoch": 0.5239520958083832,
"grad_norm": 0.9849862892856149,
"learning_rate": 8.4000427629818e-06,
"loss": 0.3324,
"step": 175
},
{
"epoch": 0.5269461077844312,
"grad_norm": 1.0286691587141985,
"learning_rate": 8.382763991447344e-06,
"loss": 0.3637,
"step": 176
},
{
"epoch": 0.5299401197604791,
"grad_norm": 1.119187299723294,
"learning_rate": 8.365410399910287e-06,
"loss": 0.3393,
"step": 177
},
{
"epoch": 0.5329341317365269,
"grad_norm": 1.1570715940359464,
"learning_rate": 8.347982372197515e-06,
"loss": 0.3319,
"step": 178
},
{
"epoch": 0.5359281437125748,
"grad_norm": 1.0199789912426847,
"learning_rate": 8.33048029378229e-06,
"loss": 0.3487,
"step": 179
},
{
"epoch": 0.5389221556886228,
"grad_norm": 0.9385573606189407,
"learning_rate": 8.312904551775731e-06,
"loss": 0.301,
"step": 180
},
{
"epoch": 0.5419161676646707,
"grad_norm": 0.9538801663604035,
"learning_rate": 8.295255534918249e-06,
"loss": 0.3386,
"step": 181
},
{
"epoch": 0.5449101796407185,
"grad_norm": 1.2178915115208544,
"learning_rate": 8.277533633570948e-06,
"loss": 0.391,
"step": 182
},
{
"epoch": 0.5479041916167665,
"grad_norm": 1.0158036407505957,
"learning_rate": 8.25973923970699e-06,
"loss": 0.3634,
"step": 183
},
{
"epoch": 0.5508982035928144,
"grad_norm": 0.9749077215404242,
"learning_rate": 8.241872746902934e-06,
"loss": 0.3968,
"step": 184
},
{
"epoch": 0.5538922155688623,
"grad_norm": 1.028995666806137,
"learning_rate": 8.223934550330015e-06,
"loss": 0.3155,
"step": 185
},
{
"epoch": 0.5568862275449101,
"grad_norm": 1.0418658859660073,
"learning_rate": 8.20592504674542e-06,
"loss": 0.3681,
"step": 186
},
{
"epoch": 0.5598802395209581,
"grad_norm": 0.9194866364424841,
"learning_rate": 8.187844634483495e-06,
"loss": 0.3724,
"step": 187
},
{
"epoch": 0.562874251497006,
"grad_norm": 1.13731681206958,
"learning_rate": 8.16969371344696e-06,
"loss": 0.4257,
"step": 188
},
{
"epoch": 0.5658682634730539,
"grad_norm": 0.8948766112024334,
"learning_rate": 8.151472685098037e-06,
"loss": 0.3372,
"step": 189
},
{
"epoch": 0.5688622754491018,
"grad_norm": 0.9856820806546657,
"learning_rate": 8.13318195244958e-06,
"loss": 0.32,
"step": 190
},
{
"epoch": 0.5718562874251497,
"grad_norm": 0.971597768365592,
"learning_rate": 8.114821920056177e-06,
"loss": 0.344,
"step": 191
},
{
"epoch": 0.5748502994011976,
"grad_norm": 0.9544605195686608,
"learning_rate": 8.096392994005177e-06,
"loss": 0.3653,
"step": 192
},
{
"epoch": 0.5778443113772455,
"grad_norm": 0.8672689717252952,
"learning_rate": 8.077895581907719e-06,
"loss": 0.3368,
"step": 193
},
{
"epoch": 0.5808383233532934,
"grad_norm": 0.8726882750487673,
"learning_rate": 8.059330092889724e-06,
"loss": 0.2948,
"step": 194
},
{
"epoch": 0.5838323353293413,
"grad_norm": 0.9745845629683125,
"learning_rate": 8.040696937582833e-06,
"loss": 0.3329,
"step": 195
},
{
"epoch": 0.5868263473053892,
"grad_norm": 1.0416213313471583,
"learning_rate": 8.021996528115335e-06,
"loss": 0.3095,
"step": 196
},
{
"epoch": 0.5898203592814372,
"grad_norm": 1.0458717181934225,
"learning_rate": 8.003229278103044e-06,
"loss": 0.3215,
"step": 197
},
{
"epoch": 0.592814371257485,
"grad_norm": 1.014646911603464,
"learning_rate": 7.984395602640153e-06,
"loss": 0.3219,
"step": 198
},
{
"epoch": 0.5958083832335329,
"grad_norm": 1.1335825122426662,
"learning_rate": 7.96549591829006e-06,
"loss": 0.3352,
"step": 199
},
{
"epoch": 0.5988023952095808,
"grad_norm": 0.9318245011851041,
"learning_rate": 7.946530643076138e-06,
"loss": 0.2887,
"step": 200
},
{
"epoch": 0.5988023952095808,
"eval_loss": 0.3315368592739105,
"eval_runtime": 3.642,
"eval_samples_per_second": 14.827,
"eval_steps_per_second": 3.844,
"step": 200
},
{
"epoch": 0.6017964071856288,
"grad_norm": 0.9549062713352291,
"learning_rate": 7.927500196472506e-06,
"loss": 0.3361,
"step": 201
},
{
"epoch": 0.6047904191616766,
"grad_norm": 1.0832232915630333,
"learning_rate": 7.908404999394747e-06,
"loss": 0.2943,
"step": 202
},
{
"epoch": 0.6077844311377245,
"grad_norm": 1.007026023803311,
"learning_rate": 7.889245474190588e-06,
"loss": 0.3298,
"step": 203
},
{
"epoch": 0.6107784431137725,
"grad_norm": 1.1881765774876996,
"learning_rate": 7.870022044630569e-06,
"loss": 0.3626,
"step": 204
},
{
"epoch": 0.6137724550898204,
"grad_norm": 0.9228611722185958,
"learning_rate": 7.85073513589867e-06,
"loss": 0.334,
"step": 205
},
{
"epoch": 0.6167664670658682,
"grad_norm": 1.2275792244201782,
"learning_rate": 7.831385174582901e-06,
"loss": 0.3397,
"step": 206
},
{
"epoch": 0.6197604790419161,
"grad_norm": 1.0302795228637778,
"learning_rate": 7.81197258866587e-06,
"loss": 0.3411,
"step": 207
},
{
"epoch": 0.6227544910179641,
"grad_norm": 1.0141364018417613,
"learning_rate": 7.792497807515317e-06,
"loss": 0.3049,
"step": 208
},
{
"epoch": 0.625748502994012,
"grad_norm": 1.0329437028983652,
"learning_rate": 7.772961261874615e-06,
"loss": 0.2922,
"step": 209
},
{
"epoch": 0.6287425149700598,
"grad_norm": 1.0392874707429338,
"learning_rate": 7.75336338385325e-06,
"loss": 0.3234,
"step": 210
},
{
"epoch": 0.6317365269461078,
"grad_norm": 1.0752777683501904,
"learning_rate": 7.733704606917248e-06,
"loss": 0.3533,
"step": 211
},
{
"epoch": 0.6347305389221557,
"grad_norm": 1.0231863554114067,
"learning_rate": 7.713985365879607e-06,
"loss": 0.3034,
"step": 212
},
{
"epoch": 0.6377245508982036,
"grad_norm": 1.0060713571927344,
"learning_rate": 7.694206096890667e-06,
"loss": 0.3173,
"step": 213
},
{
"epoch": 0.6407185628742516,
"grad_norm": 0.9752992800529113,
"learning_rate": 7.674367237428467e-06,
"loss": 0.2954,
"step": 214
},
{
"epoch": 0.6437125748502994,
"grad_norm": 0.9494709162310689,
"learning_rate": 7.654469226289068e-06,
"loss": 0.2893,
"step": 215
},
{
"epoch": 0.6467065868263473,
"grad_norm": 1.0507350863150977,
"learning_rate": 7.63451250357685e-06,
"loss": 0.3448,
"step": 216
},
{
"epoch": 0.6497005988023952,
"grad_norm": 1.0707377970078802,
"learning_rate": 7.614497510694774e-06,
"loss": 0.3492,
"step": 217
},
{
"epoch": 0.6526946107784432,
"grad_norm": 1.0988365841741288,
"learning_rate": 7.5944246903346204e-06,
"loss": 0.3156,
"step": 218
},
{
"epoch": 0.655688622754491,
"grad_norm": 0.9528731784501043,
"learning_rate": 7.574294486467204e-06,
"loss": 0.3157,
"step": 219
},
{
"epoch": 0.6586826347305389,
"grad_norm": 1.0076712019887344,
"learning_rate": 7.55410734433254e-06,
"loss": 0.3538,
"step": 220
},
{
"epoch": 0.6616766467065869,
"grad_norm": 1.1512634861644306,
"learning_rate": 7.533863710430011e-06,
"loss": 0.3645,
"step": 221
},
{
"epoch": 0.6646706586826348,
"grad_norm": 1.0376102922485955,
"learning_rate": 7.513564032508484e-06,
"loss": 0.307,
"step": 222
},
{
"epoch": 0.6676646706586826,
"grad_norm": 1.0770447714155498,
"learning_rate": 7.493208759556406e-06,
"loss": 0.3582,
"step": 223
},
{
"epoch": 0.6706586826347305,
"grad_norm": 1.1443410825358713,
"learning_rate": 7.472798341791877e-06,
"loss": 0.3886,
"step": 224
},
{
"epoch": 0.6736526946107785,
"grad_norm": 0.9543160614166173,
"learning_rate": 7.452333230652688e-06,
"loss": 0.3209,
"step": 225
},
{
"epoch": 0.6766467065868264,
"grad_norm": 0.9211833334848217,
"learning_rate": 7.431813878786343e-06,
"loss": 0.3122,
"step": 226
},
{
"epoch": 0.6796407185628742,
"grad_norm": 1.0699830367463996,
"learning_rate": 7.4112407400400395e-06,
"loss": 0.3517,
"step": 227
},
{
"epoch": 0.6826347305389222,
"grad_norm": 0.96217365852473,
"learning_rate": 7.390614269450633e-06,
"loss": 0.3216,
"step": 228
},
{
"epoch": 0.6856287425149701,
"grad_norm": 0.8975229542312452,
"learning_rate": 7.369934923234577e-06,
"loss": 0.3053,
"step": 229
},
{
"epoch": 0.688622754491018,
"grad_norm": 1.1533413339048975,
"learning_rate": 7.349203158777826e-06,
"loss": 0.372,
"step": 230
},
{
"epoch": 0.6916167664670658,
"grad_norm": 1.074343263830326,
"learning_rate": 7.32841943462572e-06,
"loss": 0.3393,
"step": 231
},
{
"epoch": 0.6946107784431138,
"grad_norm": 1.0666239143441596,
"learning_rate": 7.3075842104728445e-06,
"loss": 0.3424,
"step": 232
},
{
"epoch": 0.6976047904191617,
"grad_norm": 1.094938617254385,
"learning_rate": 7.286697947152868e-06,
"loss": 0.3563,
"step": 233
},
{
"epoch": 0.7005988023952096,
"grad_norm": 1.0184949607267129,
"learning_rate": 7.265761106628338e-06,
"loss": 0.3187,
"step": 234
},
{
"epoch": 0.7035928143712575,
"grad_norm": 0.9477859620313964,
"learning_rate": 7.244774151980466e-06,
"loss": 0.3111,
"step": 235
},
{
"epoch": 0.7065868263473054,
"grad_norm": 1.2784955741205677,
"learning_rate": 7.223737547398898e-06,
"loss": 0.3233,
"step": 236
},
{
"epoch": 0.7095808383233533,
"grad_norm": 1.081005726353719,
"learning_rate": 7.20265175817143e-06,
"loss": 0.3377,
"step": 237
},
{
"epoch": 0.7125748502994012,
"grad_norm": 1.190150333675615,
"learning_rate": 7.181517250673729e-06,
"loss": 0.3629,
"step": 238
},
{
"epoch": 0.7155688622754491,
"grad_norm": 1.06007638134357,
"learning_rate": 7.1603344923590065e-06,
"loss": 0.3222,
"step": 239
},
{
"epoch": 0.718562874251497,
"grad_norm": 1.0039127397467644,
"learning_rate": 7.139103951747694e-06,
"loss": 0.3539,
"step": 240
},
{
"epoch": 0.7215568862275449,
"grad_norm": 0.9768503218553017,
"learning_rate": 7.1178260984170675e-06,
"loss": 0.2991,
"step": 241
},
{
"epoch": 0.7245508982035929,
"grad_norm": 1.0237836885016547,
"learning_rate": 7.0965014029908654e-06,
"loss": 0.2982,
"step": 242
},
{
"epoch": 0.7275449101796407,
"grad_norm": 0.9534223114899126,
"learning_rate": 7.075130337128883e-06,
"loss": 0.3333,
"step": 243
},
{
"epoch": 0.7305389221556886,
"grad_norm": 1.107575101192131,
"learning_rate": 7.053713373516538e-06,
"loss": 0.3514,
"step": 244
},
{
"epoch": 0.7335329341317365,
"grad_norm": 0.9913519197235001,
"learning_rate": 7.03225098585441e-06,
"loss": 0.3545,
"step": 245
},
{
"epoch": 0.7365269461077845,
"grad_norm": 0.8931507353854258,
"learning_rate": 7.0107436488477694e-06,
"loss": 0.3385,
"step": 246
},
{
"epoch": 0.7395209580838323,
"grad_norm": 1.0390223904251776,
"learning_rate": 6.989191838196083e-06,
"loss": 0.3545,
"step": 247
},
{
"epoch": 0.7425149700598802,
"grad_norm": 1.0256502618073768,
"learning_rate": 6.9675960305824785e-06,
"loss": 0.3178,
"step": 248
},
{
"epoch": 0.7455089820359282,
"grad_norm": 1.1423537363319498,
"learning_rate": 6.945956703663212e-06,
"loss": 0.3501,
"step": 249
},
{
"epoch": 0.7485029940119761,
"grad_norm": 0.9271944415517823,
"learning_rate": 6.9242743360570985e-06,
"loss": 0.3185,
"step": 250
},
{
"epoch": 0.7514970059880239,
"grad_norm": 1.19993585209742,
"learning_rate": 6.9025494073349284e-06,
"loss": 0.3687,
"step": 251
},
{
"epoch": 0.7544910179640718,
"grad_norm": 0.9270378572906929,
"learning_rate": 6.880782398008862e-06,
"loss": 0.3351,
"step": 252
},
{
"epoch": 0.7574850299401198,
"grad_norm": 1.0695698347277742,
"learning_rate": 6.858973789521792e-06,
"loss": 0.299,
"step": 253
},
{
"epoch": 0.7604790419161677,
"grad_norm": 1.0897573773044542,
"learning_rate": 6.837124064236709e-06,
"loss": 0.363,
"step": 254
},
{
"epoch": 0.7634730538922155,
"grad_norm": 0.9615682443275033,
"learning_rate": 6.815233705426019e-06,
"loss": 0.3344,
"step": 255
},
{
"epoch": 0.7664670658682635,
"grad_norm": 1.0073197101324296,
"learning_rate": 6.7933031972608644e-06,
"loss": 0.3543,
"step": 256
},
{
"epoch": 0.7694610778443114,
"grad_norm": 1.007467783613821,
"learning_rate": 6.771333024800411e-06,
"loss": 0.3392,
"step": 257
},
{
"epoch": 0.7724550898203593,
"grad_norm": 1.2469449963102903,
"learning_rate": 6.74932367398112e-06,
"loss": 0.333,
"step": 258
},
{
"epoch": 0.7754491017964071,
"grad_norm": 0.988775706371618,
"learning_rate": 6.727275631605996e-06,
"loss": 0.2922,
"step": 259
},
{
"epoch": 0.7784431137724551,
"grad_norm": 1.090281274776826,
"learning_rate": 6.70518938533383e-06,
"loss": 0.3276,
"step": 260
},
{
"epoch": 0.781437125748503,
"grad_norm": 1.1283264128240205,
"learning_rate": 6.683065423668403e-06,
"loss": 0.3489,
"step": 261
},
{
"epoch": 0.7844311377245509,
"grad_norm": 1.1247409764184755,
"learning_rate": 6.660904235947687e-06,
"loss": 0.3308,
"step": 262
},
{
"epoch": 0.7874251497005988,
"grad_norm": 1.0120648968805497,
"learning_rate": 6.638706312333018e-06,
"loss": 0.3533,
"step": 263
},
{
"epoch": 0.7904191616766467,
"grad_norm": 1.0558784177126312,
"learning_rate": 6.61647214379826e-06,
"loss": 0.3817,
"step": 264
},
{
"epoch": 0.7934131736526946,
"grad_norm": 1.1636090949104985,
"learning_rate": 6.594202222118941e-06,
"loss": 0.3487,
"step": 265
},
{
"epoch": 0.7964071856287425,
"grad_norm": 1.0568598755745766,
"learning_rate": 6.571897039861377e-06,
"loss": 0.3301,
"step": 266
},
{
"epoch": 0.7994011976047904,
"grad_norm": 1.0187699837001216,
"learning_rate": 6.549557090371775e-06,
"loss": 0.3436,
"step": 267
},
{
"epoch": 0.8023952095808383,
"grad_norm": 0.9004421112670595,
"learning_rate": 6.527182867765333e-06,
"loss": 0.252,
"step": 268
},
{
"epoch": 0.8053892215568862,
"grad_norm": 1.0723497786271863,
"learning_rate": 6.504774866915291e-06,
"loss": 0.3287,
"step": 269
},
{
"epoch": 0.8083832335329342,
"grad_norm": 1.1734805452823147,
"learning_rate": 6.482333583442002e-06,
"loss": 0.3022,
"step": 270
},
{
"epoch": 0.811377245508982,
"grad_norm": 0.8873530811813526,
"learning_rate": 6.459859513701967e-06,
"loss": 0.332,
"step": 271
},
{
"epoch": 0.8143712574850299,
"grad_norm": 1.1940411580417634,
"learning_rate": 6.437353154776848e-06,
"loss": 0.3351,
"step": 272
},
{
"epoch": 0.8173652694610778,
"grad_norm": 1.078022035336462,
"learning_rate": 6.414815004462483e-06,
"loss": 0.3595,
"step": 273
},
{
"epoch": 0.8203592814371258,
"grad_norm": 1.035205260590465,
"learning_rate": 6.3922455612578715e-06,
"loss": 0.3587,
"step": 274
},
{
"epoch": 0.8233532934131736,
"grad_norm": 1.015600774489993,
"learning_rate": 6.369645324354149e-06,
"loss": 0.2878,
"step": 275
},
{
"epoch": 0.8263473053892215,
"grad_norm": 0.9218732824049639,
"learning_rate": 6.3470147936235485e-06,
"loss": 0.3357,
"step": 276
},
{
"epoch": 0.8293413173652695,
"grad_norm": 0.8225883167794387,
"learning_rate": 6.3243544696083355e-06,
"loss": 0.299,
"step": 277
},
{
"epoch": 0.8323353293413174,
"grad_norm": 0.9527191384669422,
"learning_rate": 6.301664853509755e-06,
"loss": 0.2879,
"step": 278
},
{
"epoch": 0.8353293413173652,
"grad_norm": 0.9566605047180817,
"learning_rate": 6.278946447176924e-06,
"loss": 0.3232,
"step": 279
},
{
"epoch": 0.8383233532934131,
"grad_norm": 0.9389666222712573,
"learning_rate": 6.256199753095745e-06,
"loss": 0.3041,
"step": 280
},
{
"epoch": 0.8413173652694611,
"grad_norm": 0.9657496008836173,
"learning_rate": 6.233425274377793e-06,
"loss": 0.3199,
"step": 281
},
{
"epoch": 0.844311377245509,
"grad_norm": 1.1941292116139328,
"learning_rate": 6.21062351474918e-06,
"loss": 0.3693,
"step": 282
},
{
"epoch": 0.8473053892215568,
"grad_norm": 1.0340975228456097,
"learning_rate": 6.18779497853942e-06,
"loss": 0.3499,
"step": 283
},
{
"epoch": 0.8502994011976048,
"grad_norm": 0.9968492834253452,
"learning_rate": 6.164940170670266e-06,
"loss": 0.3511,
"step": 284
},
{
"epoch": 0.8532934131736527,
"grad_norm": 0.8765458996493468,
"learning_rate": 6.142059596644557e-06,
"loss": 0.2799,
"step": 285
},
{
"epoch": 0.8562874251497006,
"grad_norm": 1.0195650688617215,
"learning_rate": 6.11915376253502e-06,
"loss": 0.3082,
"step": 286
},
{
"epoch": 0.8592814371257484,
"grad_norm": 1.0558148200337274,
"learning_rate": 6.096223174973091e-06,
"loss": 0.3385,
"step": 287
},
{
"epoch": 0.8622754491017964,
"grad_norm": 0.8999169136719064,
"learning_rate": 6.073268341137694e-06,
"loss": 0.283,
"step": 288
},
{
"epoch": 0.8652694610778443,
"grad_norm": 1.064183642255837,
"learning_rate": 6.050289768744042e-06,
"loss": 0.2729,
"step": 289
},
{
"epoch": 0.8682634730538922,
"grad_norm": 1.143949133884989,
"learning_rate": 6.0272879660323936e-06,
"loss": 0.3613,
"step": 290
},
{
"epoch": 0.8712574850299402,
"grad_norm": 1.0023483859940223,
"learning_rate": 6.004263441756815e-06,
"loss": 0.3344,
"step": 291
},
{
"epoch": 0.874251497005988,
"grad_norm": 1.0322004046296314,
"learning_rate": 5.98121670517393e-06,
"loss": 0.3099,
"step": 292
},
{
"epoch": 0.8772455089820359,
"grad_norm": 0.8610742299394869,
"learning_rate": 5.958148266031654e-06,
"loss": 0.2855,
"step": 293
},
{
"epoch": 0.8802395209580839,
"grad_norm": 0.9400302075803765,
"learning_rate": 5.935058634557917e-06,
"loss": 0.2983,
"step": 294
},
{
"epoch": 0.8832335329341318,
"grad_norm": 0.8730200219256645,
"learning_rate": 5.911948321449384e-06,
"loss": 0.3269,
"step": 295
},
{
"epoch": 0.8862275449101796,
"grad_norm": 1.033306866481561,
"learning_rate": 5.8888178378601565e-06,
"loss": 0.3378,
"step": 296
},
{
"epoch": 0.8892215568862275,
"grad_norm": 1.041556460741408,
"learning_rate": 5.865667695390468e-06,
"loss": 0.3209,
"step": 297
},
{
"epoch": 0.8922155688622755,
"grad_norm": 0.9711853575723411,
"learning_rate": 5.842498406075363e-06,
"loss": 0.3151,
"step": 298
},
{
"epoch": 0.8952095808383234,
"grad_norm": 1.0551428006691412,
"learning_rate": 5.819310482373381e-06,
"loss": 0.324,
"step": 299
},
{
"epoch": 0.8982035928143712,
"grad_norm": 0.9716947439779557,
"learning_rate": 5.796104437155213e-06,
"loss": 0.3628,
"step": 300
},
{
"epoch": 0.9011976047904192,
"grad_norm": 1.0340025366141352,
"learning_rate": 5.772880783692363e-06,
"loss": 0.3061,
"step": 301
},
{
"epoch": 0.9041916167664671,
"grad_norm": 1.1205047048493808,
"learning_rate": 5.749640035645798e-06,
"loss": 0.2949,
"step": 302
},
{
"epoch": 0.907185628742515,
"grad_norm": 1.0301661962517938,
"learning_rate": 5.726382707054578e-06,
"loss": 0.3273,
"step": 303
},
{
"epoch": 0.9101796407185628,
"grad_norm": 0.9943112778437615,
"learning_rate": 5.703109312324493e-06,
"loss": 0.3195,
"step": 304
},
{
"epoch": 0.9131736526946108,
"grad_norm": 1.1431319568233977,
"learning_rate": 5.679820366216684e-06,
"loss": 0.3865,
"step": 305
},
{
"epoch": 0.9161676646706587,
"grad_norm": 0.9652478334795249,
"learning_rate": 5.656516383836263e-06,
"loss": 0.2859,
"step": 306
},
{
"epoch": 0.9191616766467066,
"grad_norm": 0.9534379341751311,
"learning_rate": 5.6331978806209044e-06,
"loss": 0.3344,
"step": 307
},
{
"epoch": 0.9221556886227545,
"grad_norm": 1.0015518778523254,
"learning_rate": 5.609865372329461e-06,
"loss": 0.3053,
"step": 308
},
{
"epoch": 0.9251497005988024,
"grad_norm": 0.9050683618593659,
"learning_rate": 5.586519375030549e-06,
"loss": 0.2759,
"step": 309
},
{
"epoch": 0.9281437125748503,
"grad_norm": 0.8835663867724614,
"learning_rate": 5.5631604050911354e-06,
"loss": 0.2986,
"step": 310
},
{
"epoch": 0.9311377245508982,
"grad_norm": 1.071755354339813,
"learning_rate": 5.539788979165115e-06,
"loss": 0.2835,
"step": 311
},
{
"epoch": 0.9341317365269461,
"grad_norm": 0.9246051209680475,
"learning_rate": 5.516405614181883e-06,
"loss": 0.2714,
"step": 312
},
{
"epoch": 0.937125748502994,
"grad_norm": 0.8702333494607535,
"learning_rate": 5.4930108273349034e-06,
"loss": 0.3063,
"step": 313
},
{
"epoch": 0.9401197604790419,
"grad_norm": 0.9957608999524614,
"learning_rate": 5.4696051360702725e-06,
"loss": 0.2927,
"step": 314
},
{
"epoch": 0.9431137724550899,
"grad_norm": 1.1036648669108162,
"learning_rate": 5.446189058075265e-06,
"loss": 0.3415,
"step": 315
},
{
"epoch": 0.9461077844311377,
"grad_norm": 0.9585454574342558,
"learning_rate": 5.4227631112668955e-06,
"loss": 0.3473,
"step": 316
},
{
"epoch": 0.9491017964071856,
"grad_norm": 1.0983162963710271,
"learning_rate": 5.39932781378045e-06,
"loss": 0.3461,
"step": 317
},
{
"epoch": 0.9520958083832335,
"grad_norm": 0.8783134587461954,
"learning_rate": 5.375883683958041e-06,
"loss": 0.3051,
"step": 318
},
{
"epoch": 0.9550898203592815,
"grad_norm": 1.01323208994916,
"learning_rate": 5.3524312403371255e-06,
"loss": 0.323,
"step": 319
},
{
"epoch": 0.9580838323353293,
"grad_norm": 1.1254019715241883,
"learning_rate": 5.328971001639054e-06,
"loss": 0.2867,
"step": 320
},
{
"epoch": 0.9610778443113772,
"grad_norm": 0.9909289230173806,
"learning_rate": 5.3055034867575825e-06,
"loss": 0.2934,
"step": 321
},
{
"epoch": 0.9640718562874252,
"grad_norm": 1.2052008980707514,
"learning_rate": 5.282029214747404e-06,
"loss": 0.3489,
"step": 322
},
{
"epoch": 0.9670658682634731,
"grad_norm": 1.023735489283037,
"learning_rate": 5.258548704812667e-06,
"loss": 0.3283,
"step": 323
},
{
"epoch": 0.9700598802395209,
"grad_norm": 1.0240362131412724,
"learning_rate": 5.235062476295488e-06,
"loss": 0.3145,
"step": 324
},
{
"epoch": 0.9730538922155688,
"grad_norm": 1.0051337574546766,
"learning_rate": 5.211571048664469e-06,
"loss": 0.2731,
"step": 325
},
{
"epoch": 0.9760479041916168,
"grad_norm": 1.0271715836908692,
"learning_rate": 5.188074941503203e-06,
"loss": 0.3354,
"step": 326
},
{
"epoch": 0.9790419161676647,
"grad_norm": 1.0140933788517197,
"learning_rate": 5.164574674498788e-06,
"loss": 0.3262,
"step": 327
},
{
"epoch": 0.9820359281437125,
"grad_norm": 1.1038022446431235,
"learning_rate": 5.141070767430331e-06,
"loss": 0.3116,
"step": 328
},
{
"epoch": 0.9850299401197605,
"grad_norm": 0.9737157184489322,
"learning_rate": 5.117563740157444e-06,
"loss": 0.3083,
"step": 329
},
{
"epoch": 0.9880239520958084,
"grad_norm": 1.0861065982351887,
"learning_rate": 5.094054112608758e-06,
"loss": 0.3704,
"step": 330
},
{
"epoch": 0.9910179640718563,
"grad_norm": 0.9154185324192624,
"learning_rate": 5.070542404770413e-06,
"loss": 0.3497,
"step": 331
},
{
"epoch": 0.9940119760479041,
"grad_norm": 0.8796184877045172,
"learning_rate": 5.047029136674563e-06,
"loss": 0.2869,
"step": 332
},
{
"epoch": 0.9970059880239521,
"grad_norm": 1.1272165605901225,
"learning_rate": 5.023514828387868e-06,
"loss": 0.4262,
"step": 333
},
{
"epoch": 1.0,
"grad_norm": 1.001983550199709,
"learning_rate": 5e-06,
"loss": 0.2557,
"step": 334
},
{
"epoch": 1.0029940119760479,
"grad_norm": 1.120780997325027,
"learning_rate": 4.976485171612134e-06,
"loss": 0.1994,
"step": 335
},
{
"epoch": 1.0059880239520957,
"grad_norm": 0.8916556850414936,
"learning_rate": 4.95297086332544e-06,
"loss": 0.2149,
"step": 336
},
{
"epoch": 1.0089820359281436,
"grad_norm": 1.0529869453649872,
"learning_rate": 4.9294575952295896e-06,
"loss": 0.222,
"step": 337
},
{
"epoch": 1.0119760479041917,
"grad_norm": 1.0805226116417181,
"learning_rate": 4.905945887391242e-06,
"loss": 0.2492,
"step": 338
},
{
"epoch": 1.0149700598802396,
"grad_norm": 1.0660566535181053,
"learning_rate": 4.882436259842556e-06,
"loss": 0.2263,
"step": 339
},
{
"epoch": 1.0179640718562875,
"grad_norm": 0.9478780510639341,
"learning_rate": 4.858929232569671e-06,
"loss": 0.2218,
"step": 340
},
{
"epoch": 1.0209580838323353,
"grad_norm": 1.0044816645468828,
"learning_rate": 4.835425325501214e-06,
"loss": 0.2313,
"step": 341
},
{
"epoch": 1.0239520958083832,
"grad_norm": 0.9538631959954335,
"learning_rate": 4.811925058496799e-06,
"loss": 0.1806,
"step": 342
},
{
"epoch": 1.026946107784431,
"grad_norm": 1.205111462154382,
"learning_rate": 4.788428951335534e-06,
"loss": 0.2397,
"step": 343
},
{
"epoch": 1.029940119760479,
"grad_norm": 0.9456920721104755,
"learning_rate": 4.7649375237045135e-06,
"loss": 0.212,
"step": 344
},
{
"epoch": 1.032934131736527,
"grad_norm": 1.0778522366484387,
"learning_rate": 4.741451295187333e-06,
"loss": 0.1953,
"step": 345
},
{
"epoch": 1.035928143712575,
"grad_norm": 0.9610523306514857,
"learning_rate": 4.717970785252596e-06,
"loss": 0.2519,
"step": 346
},
{
"epoch": 1.0389221556886228,
"grad_norm": 1.0513182126452563,
"learning_rate": 4.694496513242418e-06,
"loss": 0.248,
"step": 347
},
{
"epoch": 1.0419161676646707,
"grad_norm": 1.2276558319856183,
"learning_rate": 4.671028998360947e-06,
"loss": 0.2014,
"step": 348
},
{
"epoch": 1.0449101796407185,
"grad_norm": 1.0915388150245908,
"learning_rate": 4.647568759662876e-06,
"loss": 0.2474,
"step": 349
},
{
"epoch": 1.0479041916167664,
"grad_norm": 1.0473505301033634,
"learning_rate": 4.624116316041962e-06,
"loss": 0.2251,
"step": 350
},
{
"epoch": 1.0508982035928143,
"grad_norm": 0.9966004754305402,
"learning_rate": 4.600672186219551e-06,
"loss": 0.2344,
"step": 351
},
{
"epoch": 1.0538922155688624,
"grad_norm": 1.1589863372895899,
"learning_rate": 4.5772368887331044e-06,
"loss": 0.2614,
"step": 352
},
{
"epoch": 1.0568862275449102,
"grad_norm": 1.0058004616630787,
"learning_rate": 4.553810941924735e-06,
"loss": 0.2567,
"step": 353
},
{
"epoch": 1.0598802395209581,
"grad_norm": 1.0413209181041558,
"learning_rate": 4.530394863929728e-06,
"loss": 0.2266,
"step": 354
},
{
"epoch": 1.062874251497006,
"grad_norm": 1.0435012600160651,
"learning_rate": 4.506989172665097e-06,
"loss": 0.2126,
"step": 355
},
{
"epoch": 1.0658682634730539,
"grad_norm": 1.0667884818551385,
"learning_rate": 4.483594385818119e-06,
"loss": 0.2091,
"step": 356
},
{
"epoch": 1.0688622754491017,
"grad_norm": 1.1249026939269269,
"learning_rate": 4.460211020834887e-06,
"loss": 0.2273,
"step": 357
},
{
"epoch": 1.0718562874251496,
"grad_norm": 1.11459781794373,
"learning_rate": 4.436839594908866e-06,
"loss": 0.2776,
"step": 358
},
{
"epoch": 1.0748502994011977,
"grad_norm": 1.1463235445484568,
"learning_rate": 4.4134806249694514e-06,
"loss": 0.2436,
"step": 359
},
{
"epoch": 1.0778443113772456,
"grad_norm": 1.0948464644177056,
"learning_rate": 4.39013462767054e-06,
"loss": 0.2665,
"step": 360
},
{
"epoch": 1.0808383233532934,
"grad_norm": 0.9839018919086766,
"learning_rate": 4.366802119379098e-06,
"loss": 0.1785,
"step": 361
},
{
"epoch": 1.0838323353293413,
"grad_norm": 0.9293535244853764,
"learning_rate": 4.34348361616374e-06,
"loss": 0.2638,
"step": 362
},
{
"epoch": 1.0868263473053892,
"grad_norm": 1.1614689080414713,
"learning_rate": 4.3201796337833165e-06,
"loss": 0.2473,
"step": 363
},
{
"epoch": 1.089820359281437,
"grad_norm": 1.2354355378864978,
"learning_rate": 4.29689068767551e-06,
"loss": 0.2665,
"step": 364
},
{
"epoch": 1.092814371257485,
"grad_norm": 1.003922454947592,
"learning_rate": 4.273617292945425e-06,
"loss": 0.2302,
"step": 365
},
{
"epoch": 1.095808383233533,
"grad_norm": 0.9387178258529728,
"learning_rate": 4.250359964354203e-06,
"loss": 0.2343,
"step": 366
},
{
"epoch": 1.098802395209581,
"grad_norm": 1.1006861341375334,
"learning_rate": 4.227119216307637e-06,
"loss": 0.2342,
"step": 367
},
{
"epoch": 1.1017964071856288,
"grad_norm": 1.073596224662946,
"learning_rate": 4.203895562844789e-06,
"loss": 0.27,
"step": 368
},
{
"epoch": 1.1047904191616766,
"grad_norm": 1.1072394500329021,
"learning_rate": 4.18068951762662e-06,
"loss": 0.2676,
"step": 369
},
{
"epoch": 1.1077844311377245,
"grad_norm": 0.9900256592170503,
"learning_rate": 4.157501593924638e-06,
"loss": 0.2097,
"step": 370
},
{
"epoch": 1.1107784431137724,
"grad_norm": 0.983341273608456,
"learning_rate": 4.134332304609533e-06,
"loss": 0.244,
"step": 371
},
{
"epoch": 1.1137724550898203,
"grad_norm": 0.8155442010597642,
"learning_rate": 4.111182162139844e-06,
"loss": 0.2246,
"step": 372
},
{
"epoch": 1.1167664670658684,
"grad_norm": 1.0970038690218578,
"learning_rate": 4.088051678550617e-06,
"loss": 0.2184,
"step": 373
},
{
"epoch": 1.1197604790419162,
"grad_norm": 1.0391535550206905,
"learning_rate": 4.064941365442084e-06,
"loss": 0.2312,
"step": 374
},
{
"epoch": 1.122754491017964,
"grad_norm": 1.210328926363791,
"learning_rate": 4.041851733968348e-06,
"loss": 0.2649,
"step": 375
},
{
"epoch": 1.125748502994012,
"grad_norm": 1.1354077695930735,
"learning_rate": 4.018783294826071e-06,
"loss": 0.2445,
"step": 376
},
{
"epoch": 1.1287425149700598,
"grad_norm": 1.14220116338224,
"learning_rate": 3.995736558243186e-06,
"loss": 0.2637,
"step": 377
},
{
"epoch": 1.1317365269461077,
"grad_norm": 0.988509375139276,
"learning_rate": 3.972712033967608e-06,
"loss": 0.23,
"step": 378
},
{
"epoch": 1.1347305389221556,
"grad_norm": 1.0506781612844094,
"learning_rate": 3.949710231255961e-06,
"loss": 0.277,
"step": 379
},
{
"epoch": 1.1377245508982037,
"grad_norm": 1.2443171693011275,
"learning_rate": 3.926731658862307e-06,
"loss": 0.2314,
"step": 380
},
{
"epoch": 1.1407185628742516,
"grad_norm": 1.0343062250965522,
"learning_rate": 3.903776825026912e-06,
"loss": 0.2635,
"step": 381
},
{
"epoch": 1.1437125748502994,
"grad_norm": 1.0574681611431012,
"learning_rate": 3.8808462374649805e-06,
"loss": 0.2431,
"step": 382
},
{
"epoch": 1.1467065868263473,
"grad_norm": 1.0352789319458093,
"learning_rate": 3.857940403355444e-06,
"loss": 0.2152,
"step": 383
},
{
"epoch": 1.1497005988023952,
"grad_norm": 1.0907667330849988,
"learning_rate": 3.8350598293297345e-06,
"loss": 0.2601,
"step": 384
},
{
"epoch": 1.152694610778443,
"grad_norm": 1.072661199933113,
"learning_rate": 3.8122050214605822e-06,
"loss": 0.2171,
"step": 385
},
{
"epoch": 1.1556886227544911,
"grad_norm": 1.0224622971061967,
"learning_rate": 3.7893764852508207e-06,
"loss": 0.2445,
"step": 386
},
{
"epoch": 1.158682634730539,
"grad_norm": 0.9676923816780407,
"learning_rate": 3.766574725622208e-06,
"loss": 0.2124,
"step": 387
},
{
"epoch": 1.1616766467065869,
"grad_norm": 0.9809929713452926,
"learning_rate": 3.7438002469042567e-06,
"loss": 0.2297,
"step": 388
},
{
"epoch": 1.1646706586826348,
"grad_norm": 1.1255062768558792,
"learning_rate": 3.721053552823078e-06,
"loss": 0.2355,
"step": 389
},
{
"epoch": 1.1676646706586826,
"grad_norm": 1.0521872355661546,
"learning_rate": 3.698335146490246e-06,
"loss": 0.2253,
"step": 390
},
{
"epoch": 1.1706586826347305,
"grad_norm": 1.2158901114460023,
"learning_rate": 3.675645530391665e-06,
"loss": 0.2261,
"step": 391
},
{
"epoch": 1.1736526946107784,
"grad_norm": 1.0769757558699056,
"learning_rate": 3.652985206376455e-06,
"loss": 0.2101,
"step": 392
},
{
"epoch": 1.1766467065868262,
"grad_norm": 0.855439636520742,
"learning_rate": 3.630354675645853e-06,
"loss": 0.2014,
"step": 393
},
{
"epoch": 1.1796407185628743,
"grad_norm": 1.1261673928726597,
"learning_rate": 3.6077544387421293e-06,
"loss": 0.244,
"step": 394
},
{
"epoch": 1.1826347305389222,
"grad_norm": 1.18104716758802,
"learning_rate": 3.5851849955375177e-06,
"loss": 0.2015,
"step": 395
},
{
"epoch": 1.18562874251497,
"grad_norm": 1.1061500705091536,
"learning_rate": 3.5626468452231534e-06,
"loss": 0.2164,
"step": 396
},
{
"epoch": 1.188622754491018,
"grad_norm": 1.0836175743645666,
"learning_rate": 3.540140486298035e-06,
"loss": 0.1814,
"step": 397
},
{
"epoch": 1.1916167664670658,
"grad_norm": 0.9833294012403465,
"learning_rate": 3.517666416557999e-06,
"loss": 0.2652,
"step": 398
},
{
"epoch": 1.1946107784431137,
"grad_norm": 1.0550819609658846,
"learning_rate": 3.495225133084712e-06,
"loss": 0.2223,
"step": 399
},
{
"epoch": 1.1976047904191618,
"grad_norm": 0.9903612510218708,
"learning_rate": 3.472817132234669e-06,
"loss": 0.2536,
"step": 400
},
{
"epoch": 1.1976047904191618,
"eval_loss": 0.33238014578819275,
"eval_runtime": 3.6449,
"eval_samples_per_second": 14.815,
"eval_steps_per_second": 3.841,
"step": 400
},
{
"epoch": 1.2005988023952097,
"grad_norm": 1.0247937099750224,
"learning_rate": 3.4504429096282246e-06,
"loss": 0.2634,
"step": 401
},
{
"epoch": 1.2035928143712575,
"grad_norm": 1.0867958726835956,
"learning_rate": 3.428102960138625e-06,
"loss": 0.2551,
"step": 402
},
{
"epoch": 1.2065868263473054,
"grad_norm": 1.1064333537985727,
"learning_rate": 3.405797777881059e-06,
"loss": 0.2607,
"step": 403
},
{
"epoch": 1.2095808383233533,
"grad_norm": 1.0834821803134187,
"learning_rate": 3.3835278562017405e-06,
"loss": 0.2112,
"step": 404
},
{
"epoch": 1.2125748502994012,
"grad_norm": 0.9814388239856908,
"learning_rate": 3.3612936876669834e-06,
"loss": 0.2227,
"step": 405
},
{
"epoch": 1.215568862275449,
"grad_norm": 0.9407825836017719,
"learning_rate": 3.3390957640523147e-06,
"loss": 0.2144,
"step": 406
},
{
"epoch": 1.218562874251497,
"grad_norm": 0.9642302384460261,
"learning_rate": 3.3169345763315986e-06,
"loss": 0.2263,
"step": 407
},
{
"epoch": 1.221556886227545,
"grad_norm": 1.0553694814332681,
"learning_rate": 3.29481061466617e-06,
"loss": 0.2454,
"step": 408
},
{
"epoch": 1.2245508982035929,
"grad_norm": 1.2285862597884443,
"learning_rate": 3.2727243683940045e-06,
"loss": 0.2275,
"step": 409
},
{
"epoch": 1.2275449101796407,
"grad_norm": 1.1433244520952115,
"learning_rate": 3.2506763260188824e-06,
"loss": 0.2532,
"step": 410
},
{
"epoch": 1.2305389221556886,
"grad_norm": 0.9544476299298219,
"learning_rate": 3.2286669751995905e-06,
"loss": 0.2334,
"step": 411
},
{
"epoch": 1.2335329341317365,
"grad_norm": 1.1544553938710433,
"learning_rate": 3.2066968027391377e-06,
"loss": 0.2379,
"step": 412
},
{
"epoch": 1.2365269461077844,
"grad_norm": 0.9756670561566363,
"learning_rate": 3.1847662945739833e-06,
"loss": 0.2211,
"step": 413
},
{
"epoch": 1.2395209580838324,
"grad_norm": 1.079716744361595,
"learning_rate": 3.1628759357632943e-06,
"loss": 0.2457,
"step": 414
},
{
"epoch": 1.2425149700598803,
"grad_norm": 1.271911352296675,
"learning_rate": 3.1410262104782086e-06,
"loss": 0.2123,
"step": 415
},
{
"epoch": 1.2455089820359282,
"grad_norm": 1.1048345079691722,
"learning_rate": 3.119217601991139e-06,
"loss": 0.233,
"step": 416
},
{
"epoch": 1.248502994011976,
"grad_norm": 1.0929480158312508,
"learning_rate": 3.0974505926650724e-06,
"loss": 0.1997,
"step": 417
},
{
"epoch": 1.251497005988024,
"grad_norm": 0.9945212809458664,
"learning_rate": 3.0757256639429027e-06,
"loss": 0.2216,
"step": 418
},
{
"epoch": 1.2544910179640718,
"grad_norm": 1.0924750208722207,
"learning_rate": 3.0540432963367907e-06,
"loss": 0.2622,
"step": 419
},
{
"epoch": 1.2574850299401197,
"grad_norm": 1.122745683940641,
"learning_rate": 3.032403969417523e-06,
"loss": 0.2196,
"step": 420
},
{
"epoch": 1.2604790419161676,
"grad_norm": 0.9775844361922347,
"learning_rate": 3.010808161803917e-06,
"loss": 0.211,
"step": 421
},
{
"epoch": 1.2634730538922156,
"grad_norm": 1.048630697009518,
"learning_rate": 2.9892563511522305e-06,
"loss": 0.256,
"step": 422
},
{
"epoch": 1.2664670658682635,
"grad_norm": 1.024838020178679,
"learning_rate": 2.9677490141455915e-06,
"loss": 0.2112,
"step": 423
},
{
"epoch": 1.2694610778443114,
"grad_norm": 1.0724570620951142,
"learning_rate": 2.946286626483463e-06,
"loss": 0.2595,
"step": 424
},
{
"epoch": 1.2724550898203593,
"grad_norm": 1.1805157201672005,
"learning_rate": 2.924869662871117e-06,
"loss": 0.2353,
"step": 425
},
{
"epoch": 1.2754491017964071,
"grad_norm": 1.054921850904748,
"learning_rate": 2.903498597009136e-06,
"loss": 0.2428,
"step": 426
},
{
"epoch": 1.278443113772455,
"grad_norm": 1.078143176580592,
"learning_rate": 2.8821739015829338e-06,
"loss": 0.2413,
"step": 427
},
{
"epoch": 1.281437125748503,
"grad_norm": 1.0529775359979963,
"learning_rate": 2.8608960482523058e-06,
"loss": 0.2296,
"step": 428
},
{
"epoch": 1.284431137724551,
"grad_norm": 0.9909281380636389,
"learning_rate": 2.839665507640992e-06,
"loss": 0.2233,
"step": 429
},
{
"epoch": 1.2874251497005988,
"grad_norm": 0.9534280605862175,
"learning_rate": 2.818482749326272e-06,
"loss": 0.2654,
"step": 430
},
{
"epoch": 1.2904191616766467,
"grad_norm": 0.9837416059396413,
"learning_rate": 2.797348241828569e-06,
"loss": 0.2343,
"step": 431
},
{
"epoch": 1.2934131736526946,
"grad_norm": 0.9387845028581133,
"learning_rate": 2.776262452601104e-06,
"loss": 0.2522,
"step": 432
},
{
"epoch": 1.2964071856287425,
"grad_norm": 1.069463005144409,
"learning_rate": 2.7552258480195348e-06,
"loss": 0.2283,
"step": 433
},
{
"epoch": 1.2994011976047903,
"grad_norm": 0.9111041091668719,
"learning_rate": 2.734238893371667e-06,
"loss": 0.2219,
"step": 434
},
{
"epoch": 1.3023952095808382,
"grad_norm": 1.0089111000916127,
"learning_rate": 2.7133020528471322e-06,
"loss": 0.2245,
"step": 435
},
{
"epoch": 1.3053892215568863,
"grad_norm": 0.9650960141992857,
"learning_rate": 2.6924157895271563e-06,
"loss": 0.2402,
"step": 436
},
{
"epoch": 1.3083832335329342,
"grad_norm": 1.0149539090685251,
"learning_rate": 2.671580565374282e-06,
"loss": 0.2598,
"step": 437
},
{
"epoch": 1.311377245508982,
"grad_norm": 1.0771943650212874,
"learning_rate": 2.6507968412221763e-06,
"loss": 0.2014,
"step": 438
},
{
"epoch": 1.31437125748503,
"grad_norm": 1.2027170519521388,
"learning_rate": 2.6300650767654234e-06,
"loss": 0.2229,
"step": 439
},
{
"epoch": 1.3173652694610778,
"grad_norm": 0.9929191155022168,
"learning_rate": 2.6093857305493666e-06,
"loss": 0.2299,
"step": 440
},
{
"epoch": 1.3203592814371259,
"grad_norm": 1.0383176357442037,
"learning_rate": 2.588759259959962e-06,
"loss": 0.2141,
"step": 441
},
{
"epoch": 1.3233532934131738,
"grad_norm": 1.0303627717593251,
"learning_rate": 2.568186121213658e-06,
"loss": 0.2512,
"step": 442
},
{
"epoch": 1.3263473053892216,
"grad_norm": 1.0424212282482657,
"learning_rate": 2.547666769347312e-06,
"loss": 0.2414,
"step": 443
},
{
"epoch": 1.3293413173652695,
"grad_norm": 0.9929931437348049,
"learning_rate": 2.5272016582081236e-06,
"loss": 0.2929,
"step": 444
},
{
"epoch": 1.3323353293413174,
"grad_norm": 1.082910009096254,
"learning_rate": 2.5067912404435952e-06,
"loss": 0.2497,
"step": 445
},
{
"epoch": 1.3353293413173652,
"grad_norm": 1.0702295521244,
"learning_rate": 2.486435967491516e-06,
"loss": 0.2207,
"step": 446
},
{
"epoch": 1.3383233532934131,
"grad_norm": 0.9898950887226708,
"learning_rate": 2.4661362895699903e-06,
"loss": 0.2112,
"step": 447
},
{
"epoch": 1.341317365269461,
"grad_norm": 1.0539936236985832,
"learning_rate": 2.445892655667462e-06,
"loss": 0.2975,
"step": 448
},
{
"epoch": 1.3443113772455089,
"grad_norm": 1.188819406944188,
"learning_rate": 2.425705513532798e-06,
"loss": 0.2489,
"step": 449
},
{
"epoch": 1.347305389221557,
"grad_norm": 0.9828085705362084,
"learning_rate": 2.4055753096653795e-06,
"loss": 0.2566,
"step": 450
},
{
"epoch": 1.3502994011976048,
"grad_norm": 0.9610523967206707,
"learning_rate": 2.3855024893052286e-06,
"loss": 0.2106,
"step": 451
},
{
"epoch": 1.3532934131736527,
"grad_norm": 1.0045814690827872,
"learning_rate": 2.365487496423152e-06,
"loss": 0.2301,
"step": 452
},
{
"epoch": 1.3562874251497006,
"grad_norm": 0.912840718185143,
"learning_rate": 2.3455307737109338e-06,
"loss": 0.1917,
"step": 453
},
{
"epoch": 1.3592814371257484,
"grad_norm": 1.0069433274926745,
"learning_rate": 2.3256327625715345e-06,
"loss": 0.2,
"step": 454
},
{
"epoch": 1.3622754491017965,
"grad_norm": 1.1115575977035446,
"learning_rate": 2.3057939031093346e-06,
"loss": 0.2196,
"step": 455
},
{
"epoch": 1.3652694610778444,
"grad_norm": 0.9183562468891716,
"learning_rate": 2.2860146341203936e-06,
"loss": 0.212,
"step": 456
},
{
"epoch": 1.3682634730538923,
"grad_norm": 1.051460773668604,
"learning_rate": 2.2662953930827546e-06,
"loss": 0.2857,
"step": 457
},
{
"epoch": 1.3712574850299402,
"grad_norm": 1.069258490441479,
"learning_rate": 2.2466366161467528e-06,
"loss": 0.2297,
"step": 458
},
{
"epoch": 1.374251497005988,
"grad_norm": 1.1965839739725117,
"learning_rate": 2.227038738125385e-06,
"loss": 0.2099,
"step": 459
},
{
"epoch": 1.377245508982036,
"grad_norm": 1.1870579009373892,
"learning_rate": 2.207502192484685e-06,
"loss": 0.2214,
"step": 460
},
{
"epoch": 1.3802395209580838,
"grad_norm": 0.9970324035830678,
"learning_rate": 2.188027411334131e-06,
"loss": 0.2274,
"step": 461
},
{
"epoch": 1.3832335329341316,
"grad_norm": 1.101919632137053,
"learning_rate": 2.1686148254171012e-06,
"loss": 0.2176,
"step": 462
},
{
"epoch": 1.3862275449101795,
"grad_norm": 1.263607275636707,
"learning_rate": 2.1492648641013305e-06,
"loss": 0.211,
"step": 463
},
{
"epoch": 1.3892215568862276,
"grad_norm": 0.9060459950675255,
"learning_rate": 2.1299779553694323e-06,
"loss": 0.2135,
"step": 464
},
{
"epoch": 1.3922155688622755,
"grad_norm": 1.1480393649446123,
"learning_rate": 2.1107545258094135e-06,
"loss": 0.2216,
"step": 465
},
{
"epoch": 1.3952095808383234,
"grad_norm": 1.0467015918573908,
"learning_rate": 2.0915950006052555e-06,
"loss": 0.1804,
"step": 466
},
{
"epoch": 1.3982035928143712,
"grad_norm": 0.9350199782983745,
"learning_rate": 2.0724998035274947e-06,
"loss": 0.217,
"step": 467
},
{
"epoch": 1.401197604790419,
"grad_norm": 1.0302853629131765,
"learning_rate": 2.053469356923865e-06,
"loss": 0.1836,
"step": 468
},
{
"epoch": 1.4041916167664672,
"grad_norm": 1.0731569031640331,
"learning_rate": 2.0345040817099433e-06,
"loss": 0.2321,
"step": 469
},
{
"epoch": 1.407185628742515,
"grad_norm": 0.9524504459719505,
"learning_rate": 2.0156043973598475e-06,
"loss": 0.2063,
"step": 470
},
{
"epoch": 1.410179640718563,
"grad_norm": 1.1360036721328823,
"learning_rate": 1.996770721896957e-06,
"loss": 0.2352,
"step": 471
},
{
"epoch": 1.4131736526946108,
"grad_norm": 1.17890838626718,
"learning_rate": 1.9780034718846653e-06,
"loss": 0.2214,
"step": 472
},
{
"epoch": 1.4161676646706587,
"grad_norm": 0.997275780276146,
"learning_rate": 1.9593030624171683e-06,
"loss": 0.248,
"step": 473
},
{
"epoch": 1.4191616766467066,
"grad_norm": 1.1231967565109662,
"learning_rate": 1.9406699071102774e-06,
"loss": 0.255,
"step": 474
},
{
"epoch": 1.4221556886227544,
"grad_norm": 0.9609159454598649,
"learning_rate": 1.9221044180922833e-06,
"loss": 0.2537,
"step": 475
},
{
"epoch": 1.4251497005988023,
"grad_norm": 1.0708589069952428,
"learning_rate": 1.9036070059948253e-06,
"loss": 0.2363,
"step": 476
},
{
"epoch": 1.4281437125748502,
"grad_norm": 1.1302083020114193,
"learning_rate": 1.885178079943823e-06,
"loss": 0.2547,
"step": 477
},
{
"epoch": 1.4311377245508983,
"grad_norm": 1.0934539752776768,
"learning_rate": 1.866818047550419e-06,
"loss": 0.3015,
"step": 478
},
{
"epoch": 1.4341317365269461,
"grad_norm": 0.9579132671732616,
"learning_rate": 1.8485273149019655e-06,
"loss": 0.2161,
"step": 479
},
{
"epoch": 1.437125748502994,
"grad_norm": 1.0763312638956641,
"learning_rate": 1.8303062865530407e-06,
"loss": 0.21,
"step": 480
},
{
"epoch": 1.4401197604790419,
"grad_norm": 1.0765191221213615,
"learning_rate": 1.8121553655165058e-06,
"loss": 0.2126,
"step": 481
},
{
"epoch": 1.4431137724550898,
"grad_norm": 1.0273673628819773,
"learning_rate": 1.7940749532545832e-06,
"loss": 0.2637,
"step": 482
},
{
"epoch": 1.4461077844311379,
"grad_norm": 1.262828349173821,
"learning_rate": 1.7760654496699876e-06,
"loss": 0.2244,
"step": 483
},
{
"epoch": 1.4491017964071857,
"grad_norm": 1.0580669544068668,
"learning_rate": 1.7581272530970666e-06,
"loss": 0.2372,
"step": 484
},
{
"epoch": 1.4520958083832336,
"grad_norm": 0.9753331817270267,
"learning_rate": 1.7402607602930106e-06,
"loss": 0.2273,
"step": 485
},
{
"epoch": 1.4550898203592815,
"grad_norm": 1.2400674148085924,
"learning_rate": 1.7224663664290537e-06,
"loss": 0.2672,
"step": 486
},
{
"epoch": 1.4580838323353293,
"grad_norm": 0.9901382739329848,
"learning_rate": 1.7047444650817518e-06,
"loss": 0.2232,
"step": 487
},
{
"epoch": 1.4610778443113772,
"grad_norm": 1.1353857191012278,
"learning_rate": 1.6870954482242707e-06,
"loss": 0.2192,
"step": 488
},
{
"epoch": 1.464071856287425,
"grad_norm": 0.9750353982816822,
"learning_rate": 1.669519706217711e-06,
"loss": 0.1907,
"step": 489
},
{
"epoch": 1.467065868263473,
"grad_norm": 1.0723855942625553,
"learning_rate": 1.652017627802487e-06,
"loss": 0.2371,
"step": 490
},
{
"epoch": 1.4700598802395208,
"grad_norm": 0.9355182200974157,
"learning_rate": 1.6345896000897122e-06,
"loss": 0.2172,
"step": 491
},
{
"epoch": 1.473053892215569,
"grad_norm": 0.9265368141042264,
"learning_rate": 1.6172360085526567e-06,
"loss": 0.2432,
"step": 492
},
{
"epoch": 1.4760479041916168,
"grad_norm": 0.9227248335709498,
"learning_rate": 1.5999572370182016e-06,
"loss": 0.2409,
"step": 493
},
{
"epoch": 1.4790419161676647,
"grad_norm": 1.0790582411262901,
"learning_rate": 1.5827536676583643e-06,
"loss": 0.2436,
"step": 494
},
{
"epoch": 1.4820359281437125,
"grad_norm": 1.0264302029910064,
"learning_rate": 1.5656256809818343e-06,
"loss": 0.2249,
"step": 495
},
{
"epoch": 1.4850299401197604,
"grad_norm": 1.126308422203554,
"learning_rate": 1.54857365582557e-06,
"loss": 0.1977,
"step": 496
},
{
"epoch": 1.4880239520958085,
"grad_norm": 1.0732747224841288,
"learning_rate": 1.5315979693464039e-06,
"loss": 0.2398,
"step": 497
},
{
"epoch": 1.4910179640718564,
"grad_norm": 1.1445723389186533,
"learning_rate": 1.5146989970127158e-06,
"loss": 0.1985,
"step": 498
},
{
"epoch": 1.4940119760479043,
"grad_norm": 1.0676302353249558,
"learning_rate": 1.4978771125961177e-06,
"loss": 0.2363,
"step": 499
},
{
"epoch": 1.4970059880239521,
"grad_norm": 0.9173573812329417,
"learning_rate": 1.4811326881631937e-06,
"loss": 0.234,
"step": 500
},
{
"epoch": 1.5,
"grad_norm": 1.0193862521615822,
"learning_rate": 1.4644660940672628e-06,
"loss": 0.2336,
"step": 501
},
{
"epoch": 1.5029940119760479,
"grad_norm": 1.0602268713210259,
"learning_rate": 1.4478776989401949e-06,
"loss": 0.2394,
"step": 502
},
{
"epoch": 1.5059880239520957,
"grad_norm": 1.061438806266647,
"learning_rate": 1.4313678696842559e-06,
"loss": 0.2217,
"step": 503
},
{
"epoch": 1.5089820359281436,
"grad_norm": 1.0303078984767782,
"learning_rate": 1.4149369714639856e-06,
"loss": 0.2176,
"step": 504
},
{
"epoch": 1.5119760479041915,
"grad_norm": 1.0693557526929214,
"learning_rate": 1.3985853676981316e-06,
"loss": 0.2313,
"step": 505
},
{
"epoch": 1.5149700598802394,
"grad_norm": 1.0453253312488606,
"learning_rate": 1.3823134200516043e-06,
"loss": 0.2223,
"step": 506
},
{
"epoch": 1.5179640718562875,
"grad_norm": 1.161891677306505,
"learning_rate": 1.366121488427481e-06,
"loss": 0.2225,
"step": 507
},
{
"epoch": 1.5209580838323353,
"grad_norm": 1.1814680457414586,
"learning_rate": 1.3500099309590397e-06,
"loss": 0.2393,
"step": 508
},
{
"epoch": 1.5239520958083832,
"grad_norm": 1.0800401344324,
"learning_rate": 1.3339791040018479e-06,
"loss": 0.2701,
"step": 509
},
{
"epoch": 1.5269461077844313,
"grad_norm": 0.9627179711210931,
"learning_rate": 1.3180293621258694e-06,
"loss": 0.2258,
"step": 510
},
{
"epoch": 1.5299401197604792,
"grad_norm": 1.0751669284473526,
"learning_rate": 1.3021610581076316e-06,
"loss": 0.192,
"step": 511
},
{
"epoch": 1.532934131736527,
"grad_norm": 1.1799088510749032,
"learning_rate": 1.2863745429224145e-06,
"loss": 0.2271,
"step": 512
},
{
"epoch": 1.535928143712575,
"grad_norm": 1.0107788588672892,
"learning_rate": 1.270670165736499e-06,
"loss": 0.2331,
"step": 513
},
{
"epoch": 1.5389221556886228,
"grad_norm": 1.2423906374901523,
"learning_rate": 1.2550482738994284e-06,
"loss": 0.2765,
"step": 514
},
{
"epoch": 1.5419161676646707,
"grad_norm": 1.0498550698760971,
"learning_rate": 1.239509212936343e-06,
"loss": 0.2258,
"step": 515
},
{
"epoch": 1.5449101796407185,
"grad_norm": 1.0980432249980636,
"learning_rate": 1.22405332654032e-06,
"loss": 0.2112,
"step": 516
},
{
"epoch": 1.5479041916167664,
"grad_norm": 0.8956986912727961,
"learning_rate": 1.2086809565647877e-06,
"loss": 0.2108,
"step": 517
},
{
"epoch": 1.5508982035928143,
"grad_norm": 1.0341166644237898,
"learning_rate": 1.1933924430159571e-06,
"loss": 0.2238,
"step": 518
},
{
"epoch": 1.5538922155688621,
"grad_norm": 1.0921617558358852,
"learning_rate": 1.1781881240452958e-06,
"loss": 0.2204,
"step": 519
},
{
"epoch": 1.55688622754491,
"grad_norm": 1.0810072199637464,
"learning_rate": 1.1630683359420653e-06,
"loss": 0.2342,
"step": 520
},
{
"epoch": 1.5598802395209581,
"grad_norm": 1.066765875762098,
"learning_rate": 1.1480334131258626e-06,
"loss": 0.1822,
"step": 521
},
{
"epoch": 1.562874251497006,
"grad_norm": 1.1522250437523482,
"learning_rate": 1.1330836881392405e-06,
"loss": 0.2191,
"step": 522
},
{
"epoch": 1.5658682634730539,
"grad_norm": 1.0536092443548537,
"learning_rate": 1.11821949164034e-06,
"loss": 0.2369,
"step": 523
},
{
"epoch": 1.568862275449102,
"grad_norm": 1.2036067176879872,
"learning_rate": 1.103441152395588e-06,
"loss": 0.2817,
"step": 524
},
{
"epoch": 1.5718562874251498,
"grad_norm": 1.0845767895143383,
"learning_rate": 1.088748997272414e-06,
"loss": 0.2412,
"step": 525
},
{
"epoch": 1.5748502994011977,
"grad_norm": 1.0584018793413108,
"learning_rate": 1.0741433512320316e-06,
"loss": 0.1886,
"step": 526
},
{
"epoch": 1.5778443113772456,
"grad_norm": 1.212555864338793,
"learning_rate": 1.0596245373222424e-06,
"loss": 0.2155,
"step": 527
},
{
"epoch": 1.5808383233532934,
"grad_norm": 1.0050729915873335,
"learning_rate": 1.045192876670298e-06,
"loss": 0.2485,
"step": 528
},
{
"epoch": 1.5838323353293413,
"grad_norm": 1.0489415363001906,
"learning_rate": 1.0308486884757868e-06,
"loss": 0.2568,
"step": 529
},
{
"epoch": 1.5868263473053892,
"grad_norm": 0.9940582414418759,
"learning_rate": 1.0165922900035886e-06,
"loss": 0.2305,
"step": 530
},
{
"epoch": 1.589820359281437,
"grad_norm": 1.013684341805722,
"learning_rate": 1.0024239965768417e-06,
"loss": 0.2206,
"step": 531
},
{
"epoch": 1.592814371257485,
"grad_norm": 1.0978652424097113,
"learning_rate": 9.883441215699824e-07,
"loss": 0.1713,
"step": 532
},
{
"epoch": 1.5958083832335328,
"grad_norm": 1.2801543414358416,
"learning_rate": 9.74352976401805e-07,
"loss": 0.2775,
"step": 533
},
{
"epoch": 1.5988023952095807,
"grad_norm": 1.1797110487287925,
"learning_rate": 9.604508705285765e-07,
"loss": 0.2328,
"step": 534
},
{
"epoch": 1.6017964071856288,
"grad_norm": 1.0885741304401761,
"learning_rate": 9.466381114371942e-07,
"loss": 0.2411,
"step": 535
},
{
"epoch": 1.6047904191616766,
"grad_norm": 1.032210311703811,
"learning_rate": 9.329150046383773e-07,
"loss": 0.2268,
"step": 536
},
{
"epoch": 1.6077844311377245,
"grad_norm": 1.0979951516599142,
"learning_rate": 9.192818536599213e-07,
"loss": 0.2379,
"step": 537
},
{
"epoch": 1.6107784431137726,
"grad_norm": 1.0730570994602955,
"learning_rate": 9.057389600399719e-07,
"loss": 0.2423,
"step": 538
},
{
"epoch": 1.6137724550898205,
"grad_norm": 0.973145332860036,
"learning_rate": 8.922866233203681e-07,
"loss": 0.2169,
"step": 539
},
{
"epoch": 1.6167664670658684,
"grad_norm": 0.8567577170990717,
"learning_rate": 8.789251410400024e-07,
"loss": 0.1759,
"step": 540
},
{
"epoch": 1.6197604790419162,
"grad_norm": 0.8516744200262096,
"learning_rate": 8.65654808728259e-07,
"loss": 0.227,
"step": 541
},
{
"epoch": 1.622754491017964,
"grad_norm": 1.2651507489473401,
"learning_rate": 8.524759198984567e-07,
"loss": 0.2046,
"step": 542
},
{
"epoch": 1.625748502994012,
"grad_norm": 1.0369908383308433,
"learning_rate": 8.393887660413719e-07,
"loss": 0.2235,
"step": 543
},
{
"epoch": 1.6287425149700598,
"grad_norm": 1.2681100827644558,
"learning_rate": 8.263936366187825e-07,
"loss": 0.2745,
"step": 544
},
{
"epoch": 1.6317365269461077,
"grad_norm": 1.002040932603707,
"learning_rate": 8.134908190570723e-07,
"loss": 0.2424,
"step": 545
},
{
"epoch": 1.6347305389221556,
"grad_norm": 0.9278388457536483,
"learning_rate": 8.006805987408705e-07,
"loss": 0.1722,
"step": 546
},
{
"epoch": 1.6377245508982035,
"grad_norm": 1.0789532270464646,
"learning_rate": 7.879632590067354e-07,
"loss": 0.2342,
"step": 547
},
{
"epoch": 1.6407185628742516,
"grad_norm": 1.1474361548062308,
"learning_rate": 7.753390811368972e-07,
"loss": 0.2622,
"step": 548
},
{
"epoch": 1.6437125748502994,
"grad_norm": 1.013878947233111,
"learning_rate": 7.628083443530287e-07,
"loss": 0.2246,
"step": 549
},
{
"epoch": 1.6467065868263473,
"grad_norm": 1.2149814462829602,
"learning_rate": 7.503713258100726e-07,
"loss": 0.2476,
"step": 550
},
{
"epoch": 1.6497005988023952,
"grad_norm": 0.9862113045468833,
"learning_rate": 7.380283005901084e-07,
"loss": 0.2373,
"step": 551
},
{
"epoch": 1.6526946107784433,
"grad_norm": 0.9604866204400361,
"learning_rate": 7.257795416962754e-07,
"loss": 0.2084,
"step": 552
},
{
"epoch": 1.6556886227544911,
"grad_norm": 1.032851311213239,
"learning_rate": 7.136253200467231e-07,
"loss": 0.2309,
"step": 553
},
{
"epoch": 1.658682634730539,
"grad_norm": 1.0566214221929084,
"learning_rate": 7.015659044686307e-07,
"loss": 0.2247,
"step": 554
},
{
"epoch": 1.6616766467065869,
"grad_norm": 1.0780776230309663,
"learning_rate": 6.896015616922535e-07,
"loss": 0.2289,
"step": 555
},
{
"epoch": 1.6646706586826348,
"grad_norm": 0.9867261249598954,
"learning_rate": 6.777325563450282e-07,
"loss": 0.2592,
"step": 556
},
{
"epoch": 1.6676646706586826,
"grad_norm": 1.0304200753871176,
"learning_rate": 6.659591509457125e-07,
"loss": 0.2264,
"step": 557
},
{
"epoch": 1.6706586826347305,
"grad_norm": 1.1259463691795335,
"learning_rate": 6.542816058985896e-07,
"loss": 0.233,
"step": 558
},
{
"epoch": 1.6736526946107784,
"grad_norm": 1.0979094053677894,
"learning_rate": 6.427001794876974e-07,
"loss": 0.2282,
"step": 559
},
{
"epoch": 1.6766467065868262,
"grad_norm": 0.910623521870724,
"learning_rate": 6.312151278711237e-07,
"loss": 0.2421,
"step": 560
},
{
"epoch": 1.6796407185628741,
"grad_norm": 0.9493851025423784,
"learning_rate": 6.198267050753387e-07,
"loss": 0.2241,
"step": 561
},
{
"epoch": 1.6826347305389222,
"grad_norm": 1.0402727482099472,
"learning_rate": 6.085351629895736e-07,
"loss": 0.2216,
"step": 562
},
{
"epoch": 1.68562874251497,
"grad_norm": 1.1329599659665706,
"learning_rate": 5.973407513602514e-07,
"loss": 0.2061,
"step": 563
},
{
"epoch": 1.688622754491018,
"grad_norm": 1.092930128903981,
"learning_rate": 5.862437177854629e-07,
"loss": 0.2419,
"step": 564
},
{
"epoch": 1.6916167664670658,
"grad_norm": 1.2749426984934027,
"learning_rate": 5.752443077094927e-07,
"loss": 0.2091,
"step": 565
},
{
"epoch": 1.694610778443114,
"grad_norm": 1.0597332774652286,
"learning_rate": 5.643427644173838e-07,
"loss": 0.2326,
"step": 566
},
{
"epoch": 1.6976047904191618,
"grad_norm": 0.921481068070127,
"learning_rate": 5.535393290295643e-07,
"loss": 0.2107,
"step": 567
},
{
"epoch": 1.7005988023952097,
"grad_norm": 1.0610715933216073,
"learning_rate": 5.428342404965076e-07,
"loss": 0.203,
"step": 568
},
{
"epoch": 1.7035928143712575,
"grad_norm": 1.0361512425571644,
"learning_rate": 5.322277355934557e-07,
"loss": 0.2576,
"step": 569
},
{
"epoch": 1.7065868263473054,
"grad_norm": 0.988217839441027,
"learning_rate": 5.217200489151714e-07,
"loss": 0.2332,
"step": 570
},
{
"epoch": 1.7095808383233533,
"grad_norm": 0.9277442864497875,
"learning_rate": 5.113114128707592e-07,
"loss": 0.1855,
"step": 571
},
{
"epoch": 1.7125748502994012,
"grad_norm": 1.0594572003784557,
"learning_rate": 5.010020576785174e-07,
"loss": 0.2479,
"step": 572
},
{
"epoch": 1.715568862275449,
"grad_norm": 0.9656561178114307,
"learning_rate": 4.907922113608532e-07,
"loss": 0.1827,
"step": 573
},
{
"epoch": 1.718562874251497,
"grad_norm": 1.0819457958285057,
"learning_rate": 4.806820997392325e-07,
"loss": 0.2299,
"step": 574
},
{
"epoch": 1.7215568862275448,
"grad_norm": 0.9792246950473165,
"learning_rate": 4.7067194642919036e-07,
"loss": 0.2218,
"step": 575
},
{
"epoch": 1.7245508982035929,
"grad_norm": 1.1191812148197955,
"learning_rate": 4.607619728353818e-07,
"loss": 0.2127,
"step": 576
},
{
"epoch": 1.7275449101796407,
"grad_norm": 0.922379120615955,
"learning_rate": 4.50952398146689e-07,
"loss": 0.2244,
"step": 577
},
{
"epoch": 1.7305389221556886,
"grad_norm": 0.9813910722446134,
"learning_rate": 4.4124343933136525e-07,
"loss": 0.2182,
"step": 578
},
{
"epoch": 1.7335329341317365,
"grad_norm": 0.9767871904526764,
"learning_rate": 4.3163531113224466e-07,
"loss": 0.1978,
"step": 579
},
{
"epoch": 1.7365269461077846,
"grad_norm": 1.0619583014497225,
"learning_rate": 4.221282260619891e-07,
"loss": 0.2326,
"step": 580
},
{
"epoch": 1.7395209580838324,
"grad_norm": 0.9467428308927454,
"learning_rate": 4.127223943983849e-07,
"loss": 0.2128,
"step": 581
},
{
"epoch": 1.7425149700598803,
"grad_norm": 0.9392625203601921,
"learning_rate": 4.03418024179697e-07,
"loss": 0.2116,
"step": 582
},
{
"epoch": 1.7455089820359282,
"grad_norm": 0.9922657790166328,
"learning_rate": 3.9421532120006544e-07,
"loss": 0.2049,
"step": 583
},
{
"epoch": 1.748502994011976,
"grad_norm": 1.0066667669400309,
"learning_rate": 3.851144890049535e-07,
"loss": 0.2195,
"step": 584
},
{
"epoch": 1.751497005988024,
"grad_norm": 1.0766641048372028,
"learning_rate": 3.761157288866418e-07,
"loss": 0.2685,
"step": 585
},
{
"epoch": 1.7544910179640718,
"grad_norm": 1.0596149022464623,
"learning_rate": 3.672192398797858e-07,
"loss": 0.2227,
"step": 586
},
{
"epoch": 1.7574850299401197,
"grad_norm": 1.1025489138299882,
"learning_rate": 3.58425218757002e-07,
"loss": 0.2152,
"step": 587
},
{
"epoch": 1.7604790419161676,
"grad_norm": 1.0283706178017693,
"learning_rate": 3.497338600245254e-07,
"loss": 0.2134,
"step": 588
},
{
"epoch": 1.7634730538922154,
"grad_norm": 1.0178419856196608,
"learning_rate": 3.4114535591790233e-07,
"loss": 0.1973,
"step": 589
},
{
"epoch": 1.7664670658682635,
"grad_norm": 1.1267136724094033,
"learning_rate": 3.326598963977395e-07,
"loss": 0.2094,
"step": 590
},
{
"epoch": 1.7694610778443114,
"grad_norm": 1.109093713641339,
"learning_rate": 3.242776691455013e-07,
"loss": 0.2443,
"step": 591
},
{
"epoch": 1.7724550898203593,
"grad_norm": 0.8472231916676006,
"learning_rate": 3.159988595593616e-07,
"loss": 0.1973,
"step": 592
},
{
"epoch": 1.7754491017964071,
"grad_norm": 0.9949887612141531,
"learning_rate": 3.078236507501015e-07,
"loss": 0.1944,
"step": 593
},
{
"epoch": 1.7784431137724552,
"grad_norm": 0.9862996929874783,
"learning_rate": 2.9975222353705757e-07,
"loss": 0.2364,
"step": 594
},
{
"epoch": 1.781437125748503,
"grad_norm": 1.0051321912136495,
"learning_rate": 2.917847564441256e-07,
"loss": 0.2505,
"step": 595
},
{
"epoch": 1.784431137724551,
"grad_norm": 1.0411692409210203,
"learning_rate": 2.839214256958106e-07,
"loss": 0.2172,
"step": 596
},
{
"epoch": 1.7874251497005988,
"grad_norm": 1.085094682471637,
"learning_rate": 2.7616240521332884e-07,
"loss": 0.2127,
"step": 597
},
{
"epoch": 1.7904191616766467,
"grad_norm": 0.9916985433490433,
"learning_rate": 2.6850786661076047e-07,
"loss": 0.2319,
"step": 598
},
{
"epoch": 1.7934131736526946,
"grad_norm": 1.0280969247358485,
"learning_rate": 2.6095797919125533e-07,
"loss": 0.2088,
"step": 599
},
{
"epoch": 1.7964071856287425,
"grad_norm": 0.9994125851531264,
"learning_rate": 2.5351290994328703e-07,
"loss": 0.2392,
"step": 600
},
{
"epoch": 1.7964071856287425,
"eval_loss": 0.3272787630558014,
"eval_runtime": 3.6416,
"eval_samples_per_second": 14.829,
"eval_steps_per_second": 3.844,
"step": 600
},
{
"epoch": 1.7994011976047903,
"grad_norm": 0.9309879992857231,
"learning_rate": 2.4617282353696093e-07,
"loss": 0.2033,
"step": 601
},
{
"epoch": 1.8023952095808382,
"grad_norm": 1.1097529520379192,
"learning_rate": 2.3893788232036807e-07,
"loss": 0.2752,
"step": 602
},
{
"epoch": 1.805389221556886,
"grad_norm": 1.0721722881897398,
"learning_rate": 2.318082463160032e-07,
"loss": 0.2532,
"step": 603
},
{
"epoch": 1.8083832335329342,
"grad_norm": 0.9091579276007259,
"learning_rate": 2.2478407321721295e-07,
"loss": 0.2155,
"step": 604
},
{
"epoch": 1.811377245508982,
"grad_norm": 0.9389812519351679,
"learning_rate": 2.1786551838471892e-07,
"loss": 0.2188,
"step": 605
},
{
"epoch": 1.81437125748503,
"grad_norm": 1.161840331953347,
"learning_rate": 2.1105273484317402e-07,
"loss": 0.1872,
"step": 606
},
{
"epoch": 1.8173652694610778,
"grad_norm": 0.9504051205327076,
"learning_rate": 2.043458732777831e-07,
"loss": 0.1942,
"step": 607
},
{
"epoch": 1.8203592814371259,
"grad_norm": 1.1090438079209275,
"learning_rate": 1.9774508203096843e-07,
"loss": 0.2278,
"step": 608
},
{
"epoch": 1.8233532934131738,
"grad_norm": 0.9407274404888127,
"learning_rate": 1.9125050709908388e-07,
"loss": 0.1972,
"step": 609
},
{
"epoch": 1.8263473053892216,
"grad_norm": 1.0745812484507467,
"learning_rate": 1.8486229212919482e-07,
"loss": 0.2325,
"step": 610
},
{
"epoch": 1.8293413173652695,
"grad_norm": 0.9747379226894167,
"learning_rate": 1.7858057841589281e-07,
"loss": 0.2228,
"step": 611
},
{
"epoch": 1.8323353293413174,
"grad_norm": 1.179120253172105,
"learning_rate": 1.7240550489817652e-07,
"loss": 0.2431,
"step": 612
},
{
"epoch": 1.8353293413173652,
"grad_norm": 1.1744893526957478,
"learning_rate": 1.66337208156373e-07,
"loss": 0.2117,
"step": 613
},
{
"epoch": 1.8383233532934131,
"grad_norm": 1.0301396252652208,
"learning_rate": 1.6037582240912175e-07,
"loss": 0.23,
"step": 614
},
{
"epoch": 1.841317365269461,
"grad_norm": 1.003916501495437,
"learning_rate": 1.5452147951040165e-07,
"loss": 0.2266,
"step": 615
},
{
"epoch": 1.8443113772455089,
"grad_norm": 1.1935585401161681,
"learning_rate": 1.4877430894662037e-07,
"loss": 0.2544,
"step": 616
},
{
"epoch": 1.8473053892215567,
"grad_norm": 1.0375856037351014,
"learning_rate": 1.4313443783374405e-07,
"loss": 0.2107,
"step": 617
},
{
"epoch": 1.8502994011976048,
"grad_norm": 0.9715507711651011,
"learning_rate": 1.3760199091449045e-07,
"loss": 0.1866,
"step": 618
},
{
"epoch": 1.8532934131736527,
"grad_norm": 0.9444533009481978,
"learning_rate": 1.3217709055556638e-07,
"loss": 0.1941,
"step": 619
},
{
"epoch": 1.8562874251497006,
"grad_norm": 0.941076588422268,
"learning_rate": 1.268598567449647e-07,
"loss": 0.2222,
"step": 620
},
{
"epoch": 1.8592814371257484,
"grad_norm": 1.1397228624153752,
"learning_rate": 1.2165040708930763e-07,
"loss": 0.2164,
"step": 621
},
{
"epoch": 1.8622754491017965,
"grad_norm": 0.9107269942425813,
"learning_rate": 1.1654885681124661e-07,
"loss": 0.2223,
"step": 622
},
{
"epoch": 1.8652694610778444,
"grad_norm": 1.105601961255002,
"learning_rate": 1.1155531874691372e-07,
"loss": 0.2514,
"step": 623
},
{
"epoch": 1.8682634730538923,
"grad_norm": 0.9729981118143304,
"learning_rate": 1.0666990334342708e-07,
"loss": 0.1975,
"step": 624
},
{
"epoch": 1.8712574850299402,
"grad_norm": 0.9779988127830793,
"learning_rate": 1.0189271865644445e-07,
"loss": 0.2228,
"step": 625
},
{
"epoch": 1.874251497005988,
"grad_norm": 1.3915543597252602,
"learning_rate": 9.722387034777847e-08,
"loss": 0.2728,
"step": 626
},
{
"epoch": 1.877245508982036,
"grad_norm": 1.0526084744631556,
"learning_rate": 9.266346168305518e-08,
"loss": 0.2612,
"step": 627
},
{
"epoch": 1.8802395209580838,
"grad_norm": 0.9572611099771384,
"learning_rate": 8.821159352943142e-08,
"loss": 0.224,
"step": 628
},
{
"epoch": 1.8832335329341316,
"grad_norm": 1.0433683125286477,
"learning_rate": 8.38683643533661e-08,
"loss": 0.2298,
"step": 629
},
{
"epoch": 1.8862275449101795,
"grad_norm": 1.106063451906934,
"learning_rate": 7.963387021843683e-08,
"loss": 0.2294,
"step": 630
},
{
"epoch": 1.8892215568862274,
"grad_norm": 1.0546878106798963,
"learning_rate": 7.550820478322285e-08,
"loss": 0.2484,
"step": 631
},
{
"epoch": 1.8922155688622755,
"grad_norm": 0.9755414461598199,
"learning_rate": 7.149145929922607e-08,
"loss": 0.1925,
"step": 632
},
{
"epoch": 1.8952095808383234,
"grad_norm": 1.0501305741740294,
"learning_rate": 6.758372260885714e-08,
"loss": 0.2389,
"step": 633
},
{
"epoch": 1.8982035928143712,
"grad_norm": 1.1747388529854201,
"learning_rate": 6.378508114346982e-08,
"loss": 0.2145,
"step": 634
},
{
"epoch": 1.9011976047904193,
"grad_norm": 1.2217745025502478,
"learning_rate": 6.009561892144744e-08,
"loss": 0.2107,
"step": 635
},
{
"epoch": 1.9041916167664672,
"grad_norm": 1.0492158711037927,
"learning_rate": 5.651541754634726e-08,
"loss": 0.1997,
"step": 636
},
{
"epoch": 1.907185628742515,
"grad_norm": 1.114888732202594,
"learning_rate": 5.304455620509297e-08,
"loss": 0.2313,
"step": 637
},
{
"epoch": 1.910179640718563,
"grad_norm": 1.059750727632571,
"learning_rate": 4.968311166622553e-08,
"loss": 0.2255,
"step": 638
},
{
"epoch": 1.9131736526946108,
"grad_norm": 1.1533444205849184,
"learning_rate": 4.643115827820399e-08,
"loss": 0.1966,
"step": 639
},
{
"epoch": 1.9161676646706587,
"grad_norm": 1.1959739702031535,
"learning_rate": 4.328876796776071e-08,
"loss": 0.2544,
"step": 640
},
{
"epoch": 1.9191616766467066,
"grad_norm": 1.13642173517012,
"learning_rate": 4.0256010238310936e-08,
"loss": 0.2179,
"step": 641
},
{
"epoch": 1.9221556886227544,
"grad_norm": 0.9930081824703622,
"learning_rate": 3.733295216841626e-08,
"loss": 0.2369,
"step": 642
},
{
"epoch": 1.9251497005988023,
"grad_norm": 1.056398060172031,
"learning_rate": 3.451965841029914e-08,
"loss": 0.2207,
"step": 643
},
{
"epoch": 1.9281437125748502,
"grad_norm": 0.9816302895297371,
"learning_rate": 3.181619118841517e-08,
"loss": 0.2628,
"step": 644
},
{
"epoch": 1.931137724550898,
"grad_norm": 1.2293837371851895,
"learning_rate": 2.9222610298074717e-08,
"loss": 0.2087,
"step": 645
},
{
"epoch": 1.9341317365269461,
"grad_norm": 1.0566652454512948,
"learning_rate": 2.673897310412288e-08,
"loss": 0.2513,
"step": 646
},
{
"epoch": 1.937125748502994,
"grad_norm": 0.8697785390024771,
"learning_rate": 2.4365334539667717e-08,
"loss": 0.1904,
"step": 647
},
{
"epoch": 1.9401197604790419,
"grad_norm": 1.0077746831108416,
"learning_rate": 2.210174710486679e-08,
"loss": 0.2073,
"step": 648
},
{
"epoch": 1.94311377245509,
"grad_norm": 0.9678571513827157,
"learning_rate": 1.99482608657664e-08,
"loss": 0.2134,
"step": 649
},
{
"epoch": 1.9461077844311379,
"grad_norm": 1.0911916861875266,
"learning_rate": 1.7904923453193056e-08,
"loss": 0.2437,
"step": 650
},
{
"epoch": 1.9491017964071857,
"grad_norm": 1.0020991154802186,
"learning_rate": 1.5971780061701524e-08,
"loss": 0.2143,
"step": 651
},
{
"epoch": 1.9520958083832336,
"grad_norm": 1.0720613331511677,
"learning_rate": 1.4148873448573408e-08,
"loss": 0.1961,
"step": 652
},
{
"epoch": 1.9550898203592815,
"grad_norm": 0.9896227832191301,
"learning_rate": 1.2436243932872349e-08,
"loss": 0.2339,
"step": 653
},
{
"epoch": 1.9580838323353293,
"grad_norm": 1.0276684978864103,
"learning_rate": 1.0833929394552523e-08,
"loss": 0.1948,
"step": 654
},
{
"epoch": 1.9610778443113772,
"grad_norm": 0.9644187975916577,
"learning_rate": 9.341965273621522e-09,
"loss": 0.2357,
"step": 655
},
{
"epoch": 1.964071856287425,
"grad_norm": 1.0493622722132283,
"learning_rate": 7.96038456935322e-09,
"loss": 0.237,
"step": 656
},
{
"epoch": 1.967065868263473,
"grad_norm": 1.026325458950602,
"learning_rate": 6.6892178395611125e-09,
"loss": 0.2591,
"step": 657
},
{
"epoch": 1.9700598802395208,
"grad_norm": 1.1858650537616622,
"learning_rate": 5.528493199922769e-09,
"loss": 0.2451,
"step": 658
},
{
"epoch": 1.9730538922155687,
"grad_norm": 1.0799685397833174,
"learning_rate": 4.478236323355312e-09,
"loss": 0.2041,
"step": 659
},
{
"epoch": 1.9760479041916168,
"grad_norm": 1.1633677353722862,
"learning_rate": 3.538470439448105e-09,
"loss": 0.2775,
"step": 660
},
{
"epoch": 1.9790419161676647,
"grad_norm": 1.0133326860344347,
"learning_rate": 2.709216333952602e-09,
"loss": 0.239,
"step": 661
},
{
"epoch": 1.9820359281437125,
"grad_norm": 0.9558774908197417,
"learning_rate": 1.9904923483171632e-09,
"loss": 0.1923,
"step": 662
},
{
"epoch": 1.9850299401197606,
"grad_norm": 1.0797139878034046,
"learning_rate": 1.3823143792851545e-09,
"loss": 0.2689,
"step": 663
},
{
"epoch": 1.9880239520958085,
"grad_norm": 1.1383318374125293,
"learning_rate": 8.846958785418969e-10,
"loss": 0.2551,
"step": 664
},
{
"epoch": 1.9910179640718564,
"grad_norm": 1.0801836279778023,
"learning_rate": 4.97647852417682e-10,
"loss": 0.2022,
"step": 665
},
{
"epoch": 1.9940119760479043,
"grad_norm": 1.0048844263905563,
"learning_rate": 2.2117886164407797e-10,
"loss": 0.2086,
"step": 666
},
{
"epoch": 1.9970059880239521,
"grad_norm": 0.887091966500845,
"learning_rate": 5.529502116519148e-11,
"loss": 0.1915,
"step": 667
},
{
"epoch": 2.0,
"grad_norm": 0.990100110338295,
"learning_rate": 0.0,
"loss": 0.1454,
"step": 668
},
{
"epoch": 2.0,
"step": 668,
"total_flos": 35357291446272.0,
"train_loss": 0.2874097045131786,
"train_runtime": 2068.8682,
"train_samples_per_second": 5.152,
"train_steps_per_second": 0.323
}
],
"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": 35357291446272.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}