functionary-medium-v3.1 / trainer_state.json
khaimai's picture
Upload folder using huggingface_hub
c22e342 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 169,
"global_step": 508,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.001968503937007874,
"grad_norm": 4.332395134133844,
"learning_rate": 3.125e-07,
"loss": 0.7038,
"step": 1
},
{
"epoch": 0.003937007874015748,
"grad_norm": 3.8032272638568987,
"learning_rate": 6.25e-07,
"loss": 0.6613,
"step": 2
},
{
"epoch": 0.005905511811023622,
"grad_norm": 3.8441769742686236,
"learning_rate": 9.375000000000001e-07,
"loss": 0.6784,
"step": 3
},
{
"epoch": 0.007874015748031496,
"grad_norm": 13.21086646658165,
"learning_rate": 1.25e-06,
"loss": 0.684,
"step": 4
},
{
"epoch": 0.00984251968503937,
"grad_norm": 4.69882563265012,
"learning_rate": 1.5625e-06,
"loss": 0.8,
"step": 5
},
{
"epoch": 0.011811023622047244,
"grad_norm": 2.9865736925889923,
"learning_rate": 1.8750000000000003e-06,
"loss": 0.6825,
"step": 6
},
{
"epoch": 0.013779527559055118,
"grad_norm": 2.2616602440416993,
"learning_rate": 2.1875000000000002e-06,
"loss": 0.6512,
"step": 7
},
{
"epoch": 0.015748031496062992,
"grad_norm": 1.7232429041769661,
"learning_rate": 2.5e-06,
"loss": 0.6967,
"step": 8
},
{
"epoch": 0.017716535433070866,
"grad_norm": 1.518727852616696,
"learning_rate": 2.8125e-06,
"loss": 0.7053,
"step": 9
},
{
"epoch": 0.01968503937007874,
"grad_norm": 1.5602105724827853,
"learning_rate": 3.125e-06,
"loss": 0.6418,
"step": 10
},
{
"epoch": 0.021653543307086614,
"grad_norm": 1.1712436246749327,
"learning_rate": 3.4375e-06,
"loss": 0.6534,
"step": 11
},
{
"epoch": 0.023622047244094488,
"grad_norm": 1.212415322561924,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.608,
"step": 12
},
{
"epoch": 0.025590551181102362,
"grad_norm": 2.380916453953302,
"learning_rate": 4.0625000000000005e-06,
"loss": 0.6561,
"step": 13
},
{
"epoch": 0.027559055118110236,
"grad_norm": 1.2987604338710539,
"learning_rate": 4.3750000000000005e-06,
"loss": 0.6322,
"step": 14
},
{
"epoch": 0.02952755905511811,
"grad_norm": 1.185019864587836,
"learning_rate": 4.6875000000000004e-06,
"loss": 0.4954,
"step": 15
},
{
"epoch": 0.031496062992125984,
"grad_norm": 1.2589412860556821,
"learning_rate": 5e-06,
"loss": 0.6023,
"step": 16
},
{
"epoch": 0.03346456692913386,
"grad_norm": 1.2322462534120997,
"learning_rate": 4.999949034290173e-06,
"loss": 0.681,
"step": 17
},
{
"epoch": 0.03543307086614173,
"grad_norm": 1.5064860621939775,
"learning_rate": 4.999796139238694e-06,
"loss": 0.5486,
"step": 18
},
{
"epoch": 0.03740157480314961,
"grad_norm": 1.3098881938046696,
"learning_rate": 4.999541321079486e-06,
"loss": 0.5843,
"step": 19
},
{
"epoch": 0.03937007874015748,
"grad_norm": 1.0503442131959184,
"learning_rate": 4.999184590202142e-06,
"loss": 0.4994,
"step": 20
},
{
"epoch": 0.04133858267716536,
"grad_norm": 1.2483896372838335,
"learning_rate": 4.998725961151493e-06,
"loss": 0.5296,
"step": 21
},
{
"epoch": 0.04330708661417323,
"grad_norm": 1.0327187684618095,
"learning_rate": 4.998165452627025e-06,
"loss": 0.5137,
"step": 22
},
{
"epoch": 0.045275590551181105,
"grad_norm": 1.1974834598681166,
"learning_rate": 4.99750308748211e-06,
"loss": 0.5128,
"step": 23
},
{
"epoch": 0.047244094488188976,
"grad_norm": 1.0642219502411139,
"learning_rate": 4.996738892723076e-06,
"loss": 0.5345,
"step": 24
},
{
"epoch": 0.04921259842519685,
"grad_norm": 1.0658394002157676,
"learning_rate": 4.995872899508103e-06,
"loss": 0.5282,
"step": 25
},
{
"epoch": 0.051181102362204724,
"grad_norm": 1.1719573437053,
"learning_rate": 4.9949051431459615e-06,
"loss": 0.5788,
"step": 26
},
{
"epoch": 0.0531496062992126,
"grad_norm": 0.9655899374808902,
"learning_rate": 4.993835663094562e-06,
"loss": 0.6245,
"step": 27
},
{
"epoch": 0.05511811023622047,
"grad_norm": 0.9796161059781048,
"learning_rate": 4.992664502959351e-06,
"loss": 0.5289,
"step": 28
},
{
"epoch": 0.05708661417322835,
"grad_norm": 1.71222119533011,
"learning_rate": 4.9913917104915374e-06,
"loss": 0.57,
"step": 29
},
{
"epoch": 0.05905511811023622,
"grad_norm": 1.0197717375686455,
"learning_rate": 4.990017337586137e-06,
"loss": 0.519,
"step": 30
},
{
"epoch": 0.0610236220472441,
"grad_norm": 1.1975231795964436,
"learning_rate": 4.988541440279862e-06,
"loss": 0.5343,
"step": 31
},
{
"epoch": 0.06299212598425197,
"grad_norm": 1.0891858921723208,
"learning_rate": 4.9869640787488375e-06,
"loss": 0.5278,
"step": 32
},
{
"epoch": 0.06496062992125984,
"grad_norm": 1.0524330061660225,
"learning_rate": 4.985285317306141e-06,
"loss": 0.4977,
"step": 33
},
{
"epoch": 0.06692913385826772,
"grad_norm": 0.8881056102807858,
"learning_rate": 4.983505224399188e-06,
"loss": 0.4628,
"step": 34
},
{
"epoch": 0.0688976377952756,
"grad_norm": 1.4276667421032012,
"learning_rate": 4.981623872606938e-06,
"loss": 0.5652,
"step": 35
},
{
"epoch": 0.07086614173228346,
"grad_norm": 1.5304505733171991,
"learning_rate": 4.9796413386369344e-06,
"loss": 0.472,
"step": 36
},
{
"epoch": 0.07283464566929133,
"grad_norm": 1.0115212501786546,
"learning_rate": 4.977557703322178e-06,
"loss": 0.5252,
"step": 37
},
{
"epoch": 0.07480314960629922,
"grad_norm": 0.8772818580054346,
"learning_rate": 4.975373051617831e-06,
"loss": 0.5178,
"step": 38
},
{
"epoch": 0.07677165354330709,
"grad_norm": 1.0066576706461172,
"learning_rate": 4.973087472597754e-06,
"loss": 0.5763,
"step": 39
},
{
"epoch": 0.07874015748031496,
"grad_norm": 1.0932413280744155,
"learning_rate": 4.970701059450872e-06,
"loss": 0.541,
"step": 40
},
{
"epoch": 0.08070866141732283,
"grad_norm": 0.8923882213553012,
"learning_rate": 4.968213909477376e-06,
"loss": 0.5698,
"step": 41
},
{
"epoch": 0.08267716535433071,
"grad_norm": 0.9066223497697822,
"learning_rate": 4.965626124084759e-06,
"loss": 0.5799,
"step": 42
},
{
"epoch": 0.08464566929133858,
"grad_norm": 1.818495666444202,
"learning_rate": 4.962937808783675e-06,
"loss": 0.4798,
"step": 43
},
{
"epoch": 0.08661417322834646,
"grad_norm": 0.8209739924387579,
"learning_rate": 4.960149073183643e-06,
"loss": 0.5458,
"step": 44
},
{
"epoch": 0.08858267716535433,
"grad_norm": 0.964480961227574,
"learning_rate": 4.957260030988575e-06,
"loss": 0.6135,
"step": 45
},
{
"epoch": 0.09055118110236221,
"grad_norm": 1.3029039264019875,
"learning_rate": 4.954270799992138e-06,
"loss": 0.5808,
"step": 46
},
{
"epoch": 0.09251968503937008,
"grad_norm": 1.3124173456233417,
"learning_rate": 4.951181502072957e-06,
"loss": 0.5146,
"step": 47
},
{
"epoch": 0.09448818897637795,
"grad_norm": 1.091556201810711,
"learning_rate": 4.947992263189641e-06,
"loss": 0.5056,
"step": 48
},
{
"epoch": 0.09645669291338582,
"grad_norm": 1.845554567847516,
"learning_rate": 4.944703213375648e-06,
"loss": 0.4985,
"step": 49
},
{
"epoch": 0.0984251968503937,
"grad_norm": 0.9595152711989854,
"learning_rate": 4.941314486733986e-06,
"loss": 0.5185,
"step": 50
},
{
"epoch": 0.10039370078740158,
"grad_norm": 1.052944693142675,
"learning_rate": 4.937826221431742e-06,
"loss": 0.541,
"step": 51
},
{
"epoch": 0.10236220472440945,
"grad_norm": 1.8359217216949988,
"learning_rate": 4.934238559694448e-06,
"loss": 0.4732,
"step": 52
},
{
"epoch": 0.10433070866141732,
"grad_norm": 0.7780724565161726,
"learning_rate": 4.930551647800287e-06,
"loss": 0.5362,
"step": 53
},
{
"epoch": 0.1062992125984252,
"grad_norm": 1.0943420987327674,
"learning_rate": 4.926765636074124e-06,
"loss": 0.5573,
"step": 54
},
{
"epoch": 0.10826771653543307,
"grad_norm": 0.8835792924826419,
"learning_rate": 4.92288067888138e-06,
"loss": 0.55,
"step": 55
},
{
"epoch": 0.11023622047244094,
"grad_norm": 2.9389315424393643,
"learning_rate": 4.918896934621734e-06,
"loss": 0.5186,
"step": 56
},
{
"epoch": 0.11220472440944881,
"grad_norm": 0.8470711864407996,
"learning_rate": 4.914814565722671e-06,
"loss": 0.5772,
"step": 57
},
{
"epoch": 0.1141732283464567,
"grad_norm": 2.683190905501092,
"learning_rate": 4.9106337386328524e-06,
"loss": 0.542,
"step": 58
},
{
"epoch": 0.11614173228346457,
"grad_norm": 1.0455114679897974,
"learning_rate": 4.906354623815336e-06,
"loss": 0.4742,
"step": 59
},
{
"epoch": 0.11811023622047244,
"grad_norm": 0.9509865582235187,
"learning_rate": 4.9019773957406194e-06,
"loss": 0.4711,
"step": 60
},
{
"epoch": 0.12007874015748031,
"grad_norm": 0.8665458092112138,
"learning_rate": 4.897502232879533e-06,
"loss": 0.5401,
"step": 61
},
{
"epoch": 0.1220472440944882,
"grad_norm": 0.8508175382415222,
"learning_rate": 4.892929317695957e-06,
"loss": 0.5181,
"step": 62
},
{
"epoch": 0.12401574803149606,
"grad_norm": 0.8821611955743189,
"learning_rate": 4.888258836639386e-06,
"loss": 0.5483,
"step": 63
},
{
"epoch": 0.12598425196850394,
"grad_norm": 1.3830991975814106,
"learning_rate": 4.883490980137327e-06,
"loss": 0.4886,
"step": 64
},
{
"epoch": 0.1279527559055118,
"grad_norm": 0.8883477594577199,
"learning_rate": 4.878625942587531e-06,
"loss": 0.5847,
"step": 65
},
{
"epoch": 0.12992125984251968,
"grad_norm": 0.9173712226341854,
"learning_rate": 4.8736639223500734e-06,
"loss": 0.4875,
"step": 66
},
{
"epoch": 0.13188976377952755,
"grad_norm": 1.429358689482403,
"learning_rate": 4.868605121739261e-06,
"loss": 0.5519,
"step": 67
},
{
"epoch": 0.13385826771653545,
"grad_norm": 1.1044409729056377,
"learning_rate": 4.863449747015384e-06,
"loss": 0.4856,
"step": 68
},
{
"epoch": 0.13582677165354332,
"grad_norm": 0.9315698624773923,
"learning_rate": 4.858198008376308e-06,
"loss": 0.47,
"step": 69
},
{
"epoch": 0.1377952755905512,
"grad_norm": 1.7217476799446472,
"learning_rate": 4.8528501199489045e-06,
"loss": 0.5075,
"step": 70
},
{
"epoch": 0.13976377952755906,
"grad_norm": 3.073555434911596,
"learning_rate": 4.847406299780316e-06,
"loss": 0.4722,
"step": 71
},
{
"epoch": 0.14173228346456693,
"grad_norm": 1.1660178741995686,
"learning_rate": 4.84186676982907e-06,
"loss": 0.5038,
"step": 72
},
{
"epoch": 0.1437007874015748,
"grad_norm": 1.0506730147991332,
"learning_rate": 4.836231755956028e-06,
"loss": 0.5016,
"step": 73
},
{
"epoch": 0.14566929133858267,
"grad_norm": 0.8677747645293913,
"learning_rate": 4.830501487915175e-06,
"loss": 0.5029,
"step": 74
},
{
"epoch": 0.14763779527559054,
"grad_norm": 0.8908161828150893,
"learning_rate": 4.824676199344253e-06,
"loss": 0.5234,
"step": 75
},
{
"epoch": 0.14960629921259844,
"grad_norm": 0.9393694358554296,
"learning_rate": 4.8187561277552376e-06,
"loss": 0.4979,
"step": 76
},
{
"epoch": 0.1515748031496063,
"grad_norm": 1.5061362639101046,
"learning_rate": 4.812741514524647e-06,
"loss": 0.6668,
"step": 77
},
{
"epoch": 0.15354330708661418,
"grad_norm": 0.8981888468432785,
"learning_rate": 4.806632604883709e-06,
"loss": 0.5162,
"step": 78
},
{
"epoch": 0.15551181102362205,
"grad_norm": 0.7689317410884454,
"learning_rate": 4.800429647908354e-06,
"loss": 0.5035,
"step": 79
},
{
"epoch": 0.15748031496062992,
"grad_norm": 0.9244306023248633,
"learning_rate": 4.7941328965090705e-06,
"loss": 0.4985,
"step": 80
},
{
"epoch": 0.1594488188976378,
"grad_norm": 1.0540613081580672,
"learning_rate": 4.787742607420579e-06,
"loss": 0.3994,
"step": 81
},
{
"epoch": 0.16141732283464566,
"grad_norm": 1.007287531899793,
"learning_rate": 4.7812590411913755e-06,
"loss": 0.4842,
"step": 82
},
{
"epoch": 0.16338582677165353,
"grad_norm": 0.8767045652213882,
"learning_rate": 4.774682462173105e-06,
"loss": 0.5551,
"step": 83
},
{
"epoch": 0.16535433070866143,
"grad_norm": 1.9260648630773072,
"learning_rate": 4.768013138509781e-06,
"loss": 0.5416,
"step": 84
},
{
"epoch": 0.1673228346456693,
"grad_norm": 0.9582175240132643,
"learning_rate": 4.7612513421268546e-06,
"loss": 0.5334,
"step": 85
},
{
"epoch": 0.16929133858267717,
"grad_norm": 0.8128204814045301,
"learning_rate": 4.754397348720129e-06,
"loss": 0.4635,
"step": 86
},
{
"epoch": 0.17125984251968504,
"grad_norm": 0.8980619226018841,
"learning_rate": 4.7474514377445155e-06,
"loss": 0.4183,
"step": 87
},
{
"epoch": 0.1732283464566929,
"grad_norm": 1.6857219087991773,
"learning_rate": 4.740413892402639e-06,
"loss": 0.4992,
"step": 88
},
{
"epoch": 0.17519685039370078,
"grad_norm": 0.8886797894135899,
"learning_rate": 4.733284999633297e-06,
"loss": 0.4769,
"step": 89
},
{
"epoch": 0.17716535433070865,
"grad_norm": 0.818408900835814,
"learning_rate": 4.726065050099752e-06,
"loss": 0.5397,
"step": 90
},
{
"epoch": 0.17913385826771652,
"grad_norm": 1.0812576227525792,
"learning_rate": 4.718754338177887e-06,
"loss": 0.4881,
"step": 91
},
{
"epoch": 0.18110236220472442,
"grad_norm": 1.4788693769546513,
"learning_rate": 4.711353161944199e-06,
"loss": 0.4761,
"step": 92
},
{
"epoch": 0.1830708661417323,
"grad_norm": 1.2325496456214284,
"learning_rate": 4.703861823163649e-06,
"loss": 0.4836,
"step": 93
},
{
"epoch": 0.18503937007874016,
"grad_norm": 0.8865488032350944,
"learning_rate": 4.696280627277356e-06,
"loss": 0.5754,
"step": 94
},
{
"epoch": 0.18700787401574803,
"grad_norm": 0.7276510882056314,
"learning_rate": 4.688609883390144e-06,
"loss": 0.4811,
"step": 95
},
{
"epoch": 0.1889763779527559,
"grad_norm": 1.379464962941928,
"learning_rate": 4.680849904257938e-06,
"loss": 0.5191,
"step": 96
},
{
"epoch": 0.19094488188976377,
"grad_norm": 1.0428272965764391,
"learning_rate": 4.673001006275013e-06,
"loss": 0.523,
"step": 97
},
{
"epoch": 0.19291338582677164,
"grad_norm": 0.8570935917785768,
"learning_rate": 4.665063509461098e-06,
"loss": 0.6276,
"step": 98
},
{
"epoch": 0.19488188976377951,
"grad_norm": 0.8005902318380953,
"learning_rate": 4.6570377374483155e-06,
"loss": 0.4075,
"step": 99
},
{
"epoch": 0.1968503937007874,
"grad_norm": 0.8977816634387835,
"learning_rate": 4.648924017468003e-06,
"loss": 0.5731,
"step": 100
},
{
"epoch": 0.19881889763779528,
"grad_norm": 0.9141364434484606,
"learning_rate": 4.640722680337358e-06,
"loss": 0.5193,
"step": 101
},
{
"epoch": 0.20078740157480315,
"grad_norm": 0.8901321318942949,
"learning_rate": 4.632434060445956e-06,
"loss": 0.522,
"step": 102
},
{
"epoch": 0.20275590551181102,
"grad_norm": 1.043927384339449,
"learning_rate": 4.624058495742115e-06,
"loss": 0.462,
"step": 103
},
{
"epoch": 0.2047244094488189,
"grad_norm": 0.8110104992888918,
"learning_rate": 4.615596327719111e-06,
"loss": 0.5066,
"step": 104
},
{
"epoch": 0.20669291338582677,
"grad_norm": 0.8047954894259877,
"learning_rate": 4.607047901401267e-06,
"loss": 0.5461,
"step": 105
},
{
"epoch": 0.20866141732283464,
"grad_norm": 1.071113008622335,
"learning_rate": 4.598413565329876e-06,
"loss": 0.4851,
"step": 106
},
{
"epoch": 0.2106299212598425,
"grad_norm": 1.0506203680713584,
"learning_rate": 4.589693671548989e-06,
"loss": 0.6077,
"step": 107
},
{
"epoch": 0.2125984251968504,
"grad_norm": 3.8529396922156187,
"learning_rate": 4.580888575591068e-06,
"loss": 0.4328,
"step": 108
},
{
"epoch": 0.21456692913385828,
"grad_norm": 0.9888673635049777,
"learning_rate": 4.571998636462487e-06,
"loss": 0.5187,
"step": 109
},
{
"epoch": 0.21653543307086615,
"grad_norm": 1.6344963772635281,
"learning_rate": 4.56302421662889e-06,
"loss": 0.539,
"step": 110
},
{
"epoch": 0.21850393700787402,
"grad_norm": 0.9247317260190662,
"learning_rate": 4.55396568200042e-06,
"loss": 0.5578,
"step": 111
},
{
"epoch": 0.2204724409448819,
"grad_norm": 0.9051549148747274,
"learning_rate": 4.544823401916794e-06,
"loss": 0.5248,
"step": 112
},
{
"epoch": 0.22244094488188976,
"grad_norm": 1.357378622779329,
"learning_rate": 4.5355977491322485e-06,
"loss": 0.5421,
"step": 113
},
{
"epoch": 0.22440944881889763,
"grad_norm": 0.9619565073321122,
"learning_rate": 4.5262890998003374e-06,
"loss": 0.431,
"step": 114
},
{
"epoch": 0.2263779527559055,
"grad_norm": 1.7617613793605524,
"learning_rate": 4.5168978334585955e-06,
"loss": 0.5252,
"step": 115
},
{
"epoch": 0.2283464566929134,
"grad_norm": 1.032437247985628,
"learning_rate": 4.507424333013069e-06,
"loss": 0.4954,
"step": 116
},
{
"epoch": 0.23031496062992127,
"grad_norm": 1.8092825383360753,
"learning_rate": 4.497868984722698e-06,
"loss": 0.4427,
"step": 117
},
{
"epoch": 0.23228346456692914,
"grad_norm": 0.8778261117229528,
"learning_rate": 4.4882321781835666e-06,
"loss": 0.5413,
"step": 118
},
{
"epoch": 0.234251968503937,
"grad_norm": 1.1025746342315803,
"learning_rate": 4.478514306313026e-06,
"loss": 0.4299,
"step": 119
},
{
"epoch": 0.23622047244094488,
"grad_norm": 2.8378126602332436,
"learning_rate": 4.4687157653336645e-06,
"loss": 0.5511,
"step": 120
},
{
"epoch": 0.23818897637795275,
"grad_norm": 0.9667061503063853,
"learning_rate": 4.458836954757161e-06,
"loss": 0.4908,
"step": 121
},
{
"epoch": 0.24015748031496062,
"grad_norm": 0.7831662518960953,
"learning_rate": 4.448878277367988e-06,
"loss": 0.4249,
"step": 122
},
{
"epoch": 0.2421259842519685,
"grad_norm": 1.0044477612786213,
"learning_rate": 4.438840139206998e-06,
"loss": 0.5353,
"step": 123
},
{
"epoch": 0.2440944881889764,
"grad_norm": 0.8674217801187197,
"learning_rate": 4.428722949554858e-06,
"loss": 0.5313,
"step": 124
},
{
"epoch": 0.24606299212598426,
"grad_norm": 0.9013952910070694,
"learning_rate": 4.41852712091537e-06,
"loss": 0.4605,
"step": 125
},
{
"epoch": 0.24803149606299213,
"grad_norm": 1.353639010795885,
"learning_rate": 4.40825306899865e-06,
"loss": 0.4818,
"step": 126
},
{
"epoch": 0.25,
"grad_norm": 1.732310535590409,
"learning_rate": 4.397901212704176e-06,
"loss": 0.4927,
"step": 127
},
{
"epoch": 0.25196850393700787,
"grad_norm": 0.8300534233456965,
"learning_rate": 4.387471974103713e-06,
"loss": 0.4265,
"step": 128
},
{
"epoch": 0.25393700787401574,
"grad_norm": 0.8246016132923696,
"learning_rate": 4.3769657784240975e-06,
"loss": 0.5417,
"step": 129
},
{
"epoch": 0.2559055118110236,
"grad_norm": 2.3420257329268197,
"learning_rate": 4.366383054029907e-06,
"loss": 0.455,
"step": 130
},
{
"epoch": 0.2578740157480315,
"grad_norm": 0.8435577950726181,
"learning_rate": 4.355724232405989e-06,
"loss": 0.5828,
"step": 131
},
{
"epoch": 0.25984251968503935,
"grad_norm": 0.9019738301923552,
"learning_rate": 4.3449897481398735e-06,
"loss": 0.4956,
"step": 132
},
{
"epoch": 0.2618110236220472,
"grad_norm": 1.358855180994632,
"learning_rate": 4.3341800389040465e-06,
"loss": 0.4645,
"step": 133
},
{
"epoch": 0.2637795275590551,
"grad_norm": 0.9505080270122251,
"learning_rate": 4.3232955454381126e-06,
"loss": 0.4748,
"step": 134
},
{
"epoch": 0.265748031496063,
"grad_norm": 1.2239916154381045,
"learning_rate": 4.31233671153082e-06,
"loss": 0.5063,
"step": 135
},
{
"epoch": 0.2677165354330709,
"grad_norm": 0.7921771152274636,
"learning_rate": 4.3013039840019675e-06,
"loss": 0.4813,
"step": 136
},
{
"epoch": 0.26968503937007876,
"grad_norm": 0.9993865396431013,
"learning_rate": 4.2901978126841885e-06,
"loss": 0.4572,
"step": 137
},
{
"epoch": 0.27165354330708663,
"grad_norm": 0.8513708686442735,
"learning_rate": 4.2790186504046045e-06,
"loss": 0.6056,
"step": 138
},
{
"epoch": 0.2736220472440945,
"grad_norm": 0.7314619767820716,
"learning_rate": 4.267766952966369e-06,
"loss": 0.4675,
"step": 139
},
{
"epoch": 0.2755905511811024,
"grad_norm": 0.8844513998291939,
"learning_rate": 4.256443179130081e-06,
"loss": 0.5111,
"step": 140
},
{
"epoch": 0.27755905511811024,
"grad_norm": 0.9156687716284092,
"learning_rate": 4.2450477905950745e-06,
"loss": 0.4682,
"step": 141
},
{
"epoch": 0.2795275590551181,
"grad_norm": 2.073718224317071,
"learning_rate": 4.233581251980604e-06,
"loss": 0.4546,
"step": 142
},
{
"epoch": 0.281496062992126,
"grad_norm": 1.0794774294141698,
"learning_rate": 4.222044030806894e-06,
"loss": 0.5123,
"step": 143
},
{
"epoch": 0.28346456692913385,
"grad_norm": 1.0016070803809438,
"learning_rate": 4.210436597476077e-06,
"loss": 0.5368,
"step": 144
},
{
"epoch": 0.2854330708661417,
"grad_norm": 0.9098780784046511,
"learning_rate": 4.198759425253015e-06,
"loss": 0.4818,
"step": 145
},
{
"epoch": 0.2874015748031496,
"grad_norm": 0.953137882333216,
"learning_rate": 4.187012990246005e-06,
"loss": 0.4783,
"step": 146
},
{
"epoch": 0.28937007874015747,
"grad_norm": 0.8310403789001214,
"learning_rate": 4.175197771387368e-06,
"loss": 0.4613,
"step": 147
},
{
"epoch": 0.29133858267716534,
"grad_norm": 1.1060295165436405,
"learning_rate": 4.163314250413913e-06,
"loss": 0.4841,
"step": 148
},
{
"epoch": 0.2933070866141732,
"grad_norm": 0.9471537101152984,
"learning_rate": 4.1513629118473095e-06,
"loss": 0.3994,
"step": 149
},
{
"epoch": 0.2952755905511811,
"grad_norm": 0.8986530874577185,
"learning_rate": 4.139344242974317e-06,
"loss": 0.5461,
"step": 150
},
{
"epoch": 0.297244094488189,
"grad_norm": 0.8896339993189858,
"learning_rate": 4.127258733826929e-06,
"loss": 0.5245,
"step": 151
},
{
"epoch": 0.2992125984251969,
"grad_norm": 1.3706641760505383,
"learning_rate": 4.1151068771623864e-06,
"loss": 0.5109,
"step": 152
},
{
"epoch": 0.30118110236220474,
"grad_norm": 1.0147467704170827,
"learning_rate": 4.102889168443091e-06,
"loss": 0.3745,
"step": 153
},
{
"epoch": 0.3031496062992126,
"grad_norm": 1.0264900634849563,
"learning_rate": 4.0906061058164e-06,
"loss": 0.4894,
"step": 154
},
{
"epoch": 0.3051181102362205,
"grad_norm": 1.1560527014582376,
"learning_rate": 4.078258190094318e-06,
"loss": 0.5614,
"step": 155
},
{
"epoch": 0.30708661417322836,
"grad_norm": 1.0261028613642564,
"learning_rate": 4.065845924733077e-06,
"loss": 0.4748,
"step": 156
},
{
"epoch": 0.3090551181102362,
"grad_norm": 1.3512728606768398,
"learning_rate": 4.053369815812608e-06,
"loss": 0.507,
"step": 157
},
{
"epoch": 0.3110236220472441,
"grad_norm": 1.3513792498067174,
"learning_rate": 4.040830372015909e-06,
"loss": 0.4525,
"step": 158
},
{
"epoch": 0.31299212598425197,
"grad_norm": 1.0720661582317261,
"learning_rate": 4.0282281046083045e-06,
"loss": 0.5038,
"step": 159
},
{
"epoch": 0.31496062992125984,
"grad_norm": 1.7686925251926875,
"learning_rate": 4.015563527416596e-06,
"loss": 0.5094,
"step": 160
},
{
"epoch": 0.3169291338582677,
"grad_norm": 0.9696692443073361,
"learning_rate": 4.002837156808116e-06,
"loss": 0.5307,
"step": 161
},
{
"epoch": 0.3188976377952756,
"grad_norm": 1.3990075425359831,
"learning_rate": 3.990049511669675e-06,
"loss": 0.4939,
"step": 162
},
{
"epoch": 0.32086614173228345,
"grad_norm": 1.225093666699612,
"learning_rate": 3.977201113386402e-06,
"loss": 0.6196,
"step": 163
},
{
"epoch": 0.3228346456692913,
"grad_norm": 1.2749819137188019,
"learning_rate": 3.964292485820487e-06,
"loss": 0.5102,
"step": 164
},
{
"epoch": 0.3248031496062992,
"grad_norm": 0.819264787351332,
"learning_rate": 3.951324155289825e-06,
"loss": 0.4787,
"step": 165
},
{
"epoch": 0.32677165354330706,
"grad_norm": 0.842964654581977,
"learning_rate": 3.938296650546552e-06,
"loss": 0.4691,
"step": 166
},
{
"epoch": 0.328740157480315,
"grad_norm": 1.0712152280027085,
"learning_rate": 3.9252105027554885e-06,
"loss": 0.4763,
"step": 167
},
{
"epoch": 0.33070866141732286,
"grad_norm": 1.05082390567464,
"learning_rate": 3.912066245472484e-06,
"loss": 0.509,
"step": 168
},
{
"epoch": 0.33267716535433073,
"grad_norm": 1.1903053575513274,
"learning_rate": 3.898864414622661e-06,
"loss": 0.5244,
"step": 169
},
{
"epoch": 0.33267716535433073,
"eval_accuracy": 0.8246661365010368,
"eval_accuracy_first_token": 0.7662962171404146,
"eval_accuracy_first_token_<": 0.9489204844655081,
"eval_accuracy_first_token_<_total": 1899,
"eval_accuracy_first_token_<|python_tag|>": 0.9,
"eval_accuracy_first_token_<|python_tag|>_total": 990,
"eval_accuracy_first_token_Here": 0.649546827794562,
"eval_accuracy_first_token_Here_total": 331,
"eval_accuracy_first_token_The": 0.8935108153078203,
"eval_accuracy_first_token_The_total": 2404,
"eval_accuracy_first_token_To": 0.7887931034482759,
"eval_accuracy_first_token_To_total": 928,
"eval_loss": 0.509745717048645,
"eval_perplexity": 1.199116424327827,
"eval_runtime": 552.2978,
"eval_samples_per_second": 1.248,
"eval_steps_per_second": 0.158,
"eval_total_number_first_token": 9358,
"step": 169
},
{
"epoch": 0.3346456692913386,
"grad_norm": 1.339307004668538,
"learning_rate": 3.8856055484785625e-06,
"loss": 0.5357,
"step": 170
},
{
"epoch": 0.33661417322834647,
"grad_norm": 0.8533902981924752,
"learning_rate": 3.872290187638208e-06,
"loss": 0.4445,
"step": 171
},
{
"epoch": 0.33858267716535434,
"grad_norm": 0.8902879713059768,
"learning_rate": 3.858918875003053e-06,
"loss": 0.5513,
"step": 172
},
{
"epoch": 0.3405511811023622,
"grad_norm": 0.834178892942214,
"learning_rate": 3.845492155755848e-06,
"loss": 0.5228,
"step": 173
},
{
"epoch": 0.3425196850393701,
"grad_norm": 1.3019761562742433,
"learning_rate": 3.832010577338414e-06,
"loss": 0.4312,
"step": 174
},
{
"epoch": 0.34448818897637795,
"grad_norm": 4.039845846275563,
"learning_rate": 3.818474689429324e-06,
"loss": 0.5169,
"step": 175
},
{
"epoch": 0.3464566929133858,
"grad_norm": 0.7754436253803598,
"learning_rate": 3.8048850439214844e-06,
"loss": 0.4312,
"step": 176
},
{
"epoch": 0.3484251968503937,
"grad_norm": 0.952028926293702,
"learning_rate": 3.7912421948996394e-06,
"loss": 0.5327,
"step": 177
},
{
"epoch": 0.35039370078740156,
"grad_norm": 1.3003893654273762,
"learning_rate": 3.7775466986177763e-06,
"loss": 0.5361,
"step": 178
},
{
"epoch": 0.35236220472440943,
"grad_norm": 0.9157436593409249,
"learning_rate": 3.7637991134764475e-06,
"loss": 0.4125,
"step": 179
},
{
"epoch": 0.3543307086614173,
"grad_norm": 1.35306351061652,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.4931,
"step": 180
},
{
"epoch": 0.3562992125984252,
"grad_norm": 0.7608013791131959,
"learning_rate": 3.736149920813726e-06,
"loss": 0.4712,
"step": 181
},
{
"epoch": 0.35826771653543305,
"grad_norm": 0.9464529110177089,
"learning_rate": 3.7222494406209174e-06,
"loss": 0.6079,
"step": 182
},
{
"epoch": 0.36023622047244097,
"grad_norm": 0.8940626357877456,
"learning_rate": 3.7082991261798473e-06,
"loss": 0.4839,
"step": 183
},
{
"epoch": 0.36220472440944884,
"grad_norm": 0.9203015169741534,
"learning_rate": 3.6942995462806574e-06,
"loss": 0.4931,
"step": 184
},
{
"epoch": 0.3641732283464567,
"grad_norm": 0.9322524391033973,
"learning_rate": 3.6802512717221694e-06,
"loss": 0.4785,
"step": 185
},
{
"epoch": 0.3661417322834646,
"grad_norm": 0.9190265161551326,
"learning_rate": 3.6661548752886113e-06,
"loss": 0.5132,
"step": 186
},
{
"epoch": 0.36811023622047245,
"grad_norm": 0.8294908667711738,
"learning_rate": 3.6520109317262624e-06,
"loss": 0.4042,
"step": 187
},
{
"epoch": 0.3700787401574803,
"grad_norm": 0.999949692414383,
"learning_rate": 3.6378200177200223e-06,
"loss": 0.4822,
"step": 188
},
{
"epoch": 0.3720472440944882,
"grad_norm": 3.014470961734112,
"learning_rate": 3.623582711869895e-06,
"loss": 0.3565,
"step": 189
},
{
"epoch": 0.37401574803149606,
"grad_norm": 1.1256555660488559,
"learning_rate": 3.6092995946673996e-06,
"loss": 0.442,
"step": 190
},
{
"epoch": 0.37598425196850394,
"grad_norm": 2.345863305053407,
"learning_rate": 3.5949712484719013e-06,
"loss": 0.4855,
"step": 191
},
{
"epoch": 0.3779527559055118,
"grad_norm": 0.9047629883258947,
"learning_rate": 3.5805982574868673e-06,
"loss": 0.4716,
"step": 192
},
{
"epoch": 0.3799212598425197,
"grad_norm": 0.9238635453577433,
"learning_rate": 3.5661812077360496e-06,
"loss": 0.4555,
"step": 193
},
{
"epoch": 0.38188976377952755,
"grad_norm": 0.8838668873279338,
"learning_rate": 3.5517206870395852e-06,
"loss": 0.5298,
"step": 194
},
{
"epoch": 0.3838582677165354,
"grad_norm": 0.9462930307216526,
"learning_rate": 3.5372172849900377e-06,
"loss": 0.4931,
"step": 195
},
{
"epoch": 0.3858267716535433,
"grad_norm": 0.9622124378606475,
"learning_rate": 3.5226715929283507e-06,
"loss": 0.497,
"step": 196
},
{
"epoch": 0.38779527559055116,
"grad_norm": 0.9079516318983362,
"learning_rate": 3.508084203919739e-06,
"loss": 0.4954,
"step": 197
},
{
"epoch": 0.38976377952755903,
"grad_norm": 1.1546647829704102,
"learning_rate": 3.493455712729514e-06,
"loss": 0.5305,
"step": 198
},
{
"epoch": 0.39173228346456695,
"grad_norm": 0.7798719077204135,
"learning_rate": 3.478786715798823e-06,
"loss": 0.451,
"step": 199
},
{
"epoch": 0.3937007874015748,
"grad_norm": 0.9459194095168413,
"learning_rate": 3.46407781122034e-06,
"loss": 0.5437,
"step": 200
},
{
"epoch": 0.3956692913385827,
"grad_norm": 0.9501953125,
"learning_rate": 3.449329598713874e-06,
"loss": 0.5366,
"step": 201
},
{
"epoch": 0.39763779527559057,
"grad_norm": 1.4056853114275554,
"learning_rate": 3.434542679601922e-06,
"loss": 0.505,
"step": 202
},
{
"epoch": 0.39960629921259844,
"grad_norm": 1.1059042675913875,
"learning_rate": 3.4197176567851463e-06,
"loss": 0.5734,
"step": 203
},
{
"epoch": 0.4015748031496063,
"grad_norm": 0.8360499368792941,
"learning_rate": 3.404855134717795e-06,
"loss": 0.5807,
"step": 204
},
{
"epoch": 0.4035433070866142,
"grad_norm": 1.5711715398240578,
"learning_rate": 3.3899557193830585e-06,
"loss": 0.4644,
"step": 205
},
{
"epoch": 0.40551181102362205,
"grad_norm": 0.8968395119800229,
"learning_rate": 3.375020018268359e-06,
"loss": 0.4902,
"step": 206
},
{
"epoch": 0.4074803149606299,
"grad_norm": 1.1447160591147387,
"learning_rate": 3.360048640340585e-06,
"loss": 0.4326,
"step": 207
},
{
"epoch": 0.4094488188976378,
"grad_norm": 0.904757092159541,
"learning_rate": 3.345042196021257e-06,
"loss": 0.4989,
"step": 208
},
{
"epoch": 0.41141732283464566,
"grad_norm": 1.1049980536396267,
"learning_rate": 3.3300012971616467e-06,
"loss": 0.4361,
"step": 209
},
{
"epoch": 0.41338582677165353,
"grad_norm": 0.83343430543165,
"learning_rate": 3.3149265570178215e-06,
"loss": 0.574,
"step": 210
},
{
"epoch": 0.4153543307086614,
"grad_norm": 1.0114446438073477,
"learning_rate": 3.2998185902256475e-06,
"loss": 0.5026,
"step": 211
},
{
"epoch": 0.41732283464566927,
"grad_norm": 0.8198824163776982,
"learning_rate": 3.284678012775727e-06,
"loss": 0.5051,
"step": 212
},
{
"epoch": 0.41929133858267714,
"grad_norm": 0.9127041105062259,
"learning_rate": 3.269505441988281e-06,
"loss": 0.4701,
"step": 213
},
{
"epoch": 0.421259842519685,
"grad_norm": 4.605482002737506,
"learning_rate": 3.2543014964879814e-06,
"loss": 0.5211,
"step": 214
},
{
"epoch": 0.42322834645669294,
"grad_norm": 0.9332310758215561,
"learning_rate": 3.2390667961787276e-06,
"loss": 0.4969,
"step": 215
},
{
"epoch": 0.4251968503937008,
"grad_norm": 1.4647155705637793,
"learning_rate": 3.223801962218372e-06,
"loss": 0.5109,
"step": 216
},
{
"epoch": 0.4271653543307087,
"grad_norm": 0.7928872277960117,
"learning_rate": 3.208507616993393e-06,
"loss": 0.5023,
"step": 217
},
{
"epoch": 0.42913385826771655,
"grad_norm": 1.4862552178807018,
"learning_rate": 3.19318438409352e-06,
"loss": 0.474,
"step": 218
},
{
"epoch": 0.4311023622047244,
"grad_norm": 0.8070313474908052,
"learning_rate": 3.1778328882863058e-06,
"loss": 0.4826,
"step": 219
},
{
"epoch": 0.4330708661417323,
"grad_norm": 1.066146986749072,
"learning_rate": 3.162453755491655e-06,
"loss": 0.4812,
"step": 220
},
{
"epoch": 0.43503937007874016,
"grad_norm": 1.1392211636109735,
"learning_rate": 3.147047612756302e-06,
"loss": 0.5323,
"step": 221
},
{
"epoch": 0.43700787401574803,
"grad_norm": 1.4312267251600133,
"learning_rate": 3.1316150882282486e-06,
"loss": 0.4592,
"step": 222
},
{
"epoch": 0.4389763779527559,
"grad_norm": 1.6780209812766753,
"learning_rate": 3.1161568111311487e-06,
"loss": 0.414,
"step": 223
},
{
"epoch": 0.4409448818897638,
"grad_norm": 0.9508842644408767,
"learning_rate": 3.1006734117386517e-06,
"loss": 0.4582,
"step": 224
},
{
"epoch": 0.44291338582677164,
"grad_norm": 0.9030605623165181,
"learning_rate": 3.0851655213487123e-06,
"loss": 0.5879,
"step": 225
},
{
"epoch": 0.4448818897637795,
"grad_norm": 0.8294458831301379,
"learning_rate": 3.0696337722578444e-06,
"loss": 0.5549,
"step": 226
},
{
"epoch": 0.4468503937007874,
"grad_norm": 0.883421459990203,
"learning_rate": 3.05407879773534e-06,
"loss": 0.462,
"step": 227
},
{
"epoch": 0.44881889763779526,
"grad_norm": 1.180508460557826,
"learning_rate": 3.038501231997454e-06,
"loss": 0.3841,
"step": 228
},
{
"epoch": 0.4507874015748031,
"grad_norm": 0.7962446898877688,
"learning_rate": 3.0229017101815424e-06,
"loss": 0.4295,
"step": 229
},
{
"epoch": 0.452755905511811,
"grad_norm": 1.0717622725485056,
"learning_rate": 3.007280868320167e-06,
"loss": 0.3735,
"step": 230
},
{
"epoch": 0.4547244094488189,
"grad_norm": 0.8086381614393529,
"learning_rate": 2.9916393433151636e-06,
"loss": 0.4798,
"step": 231
},
{
"epoch": 0.4566929133858268,
"grad_norm": 0.8936705133798069,
"learning_rate": 2.975977772911671e-06,
"loss": 0.4978,
"step": 232
},
{
"epoch": 0.45866141732283466,
"grad_norm": 0.8810316708162265,
"learning_rate": 2.960296795672132e-06,
"loss": 0.442,
"step": 233
},
{
"epoch": 0.46062992125984253,
"grad_norm": 5.897839891083801,
"learning_rate": 2.9445970509502547e-06,
"loss": 0.403,
"step": 234
},
{
"epoch": 0.4625984251968504,
"grad_norm": 0.9519402065973711,
"learning_rate": 2.928879178864946e-06,
"loss": 0.5554,
"step": 235
},
{
"epoch": 0.4645669291338583,
"grad_norm": 0.7844120883152197,
"learning_rate": 2.9131438202742123e-06,
"loss": 0.5294,
"step": 236
},
{
"epoch": 0.46653543307086615,
"grad_norm": 0.8063590456534869,
"learning_rate": 2.897391616749031e-06,
"loss": 0.4248,
"step": 237
},
{
"epoch": 0.468503937007874,
"grad_norm": 1.0564127057502695,
"learning_rate": 2.8816232105471864e-06,
"loss": 0.3977,
"step": 238
},
{
"epoch": 0.4704724409448819,
"grad_norm": 0.8741365668536474,
"learning_rate": 2.865839244587093e-06,
"loss": 0.5085,
"step": 239
},
{
"epoch": 0.47244094488188976,
"grad_norm": 0.8777418773756611,
"learning_rate": 2.8500403624215733e-06,
"loss": 0.4319,
"step": 240
},
{
"epoch": 0.4744094488188976,
"grad_norm": 1.3773613507232982,
"learning_rate": 2.8342272082116214e-06,
"loss": 0.4283,
"step": 241
},
{
"epoch": 0.4763779527559055,
"grad_norm": 0.7719588739949143,
"learning_rate": 2.8184004267001427e-06,
"loss": 0.4309,
"step": 242
},
{
"epoch": 0.47834645669291337,
"grad_norm": 0.9822729160842563,
"learning_rate": 2.802560663185658e-06,
"loss": 0.3833,
"step": 243
},
{
"epoch": 0.48031496062992124,
"grad_norm": 0.9917830897090999,
"learning_rate": 2.786708563496002e-06,
"loss": 0.4897,
"step": 244
},
{
"epoch": 0.4822834645669291,
"grad_norm": 0.8144457334187595,
"learning_rate": 2.7708447739619833e-06,
"loss": 0.4388,
"step": 245
},
{
"epoch": 0.484251968503937,
"grad_norm": 0.994242994512558,
"learning_rate": 2.7549699413910387e-06,
"loss": 0.371,
"step": 246
},
{
"epoch": 0.4862204724409449,
"grad_norm": 0.8932095220572448,
"learning_rate": 2.739084713040856e-06,
"loss": 0.4611,
"step": 247
},
{
"epoch": 0.4881889763779528,
"grad_norm": 1.1131293477611248,
"learning_rate": 2.723189736592986e-06,
"loss": 0.4332,
"step": 248
},
{
"epoch": 0.49015748031496065,
"grad_norm": 0.710443345927187,
"learning_rate": 2.707285660126435e-06,
"loss": 0.3728,
"step": 249
},
{
"epoch": 0.4921259842519685,
"grad_norm": 0.8696923722717679,
"learning_rate": 2.69137313209124e-06,
"loss": 0.4867,
"step": 250
},
{
"epoch": 0.4940944881889764,
"grad_norm": 0.9579603395355718,
"learning_rate": 2.6754528012820288e-06,
"loss": 0.4599,
"step": 251
},
{
"epoch": 0.49606299212598426,
"grad_norm": 0.8469557790483282,
"learning_rate": 2.6595253168115707e-06,
"loss": 0.5015,
"step": 252
},
{
"epoch": 0.49803149606299213,
"grad_norm": 0.9679659162207404,
"learning_rate": 2.643591328084309e-06,
"loss": 0.454,
"step": 253
},
{
"epoch": 0.5,
"grad_norm": 1.2026592195228134,
"learning_rate": 2.6276514847698763e-06,
"loss": 0.4387,
"step": 254
},
{
"epoch": 0.5019685039370079,
"grad_norm": 1.027519530136241,
"learning_rate": 2.61170643677662e-06,
"loss": 0.579,
"step": 255
},
{
"epoch": 0.5039370078740157,
"grad_norm": 0.8919743806127818,
"learning_rate": 2.595756834225089e-06,
"loss": 0.5206,
"step": 256
},
{
"epoch": 0.5059055118110236,
"grad_norm": 0.9565537176068634,
"learning_rate": 2.5798033274215363e-06,
"loss": 0.455,
"step": 257
},
{
"epoch": 0.5078740157480315,
"grad_norm": 0.8750677082568031,
"learning_rate": 2.5638465668314006e-06,
"loss": 0.4438,
"step": 258
},
{
"epoch": 0.5098425196850394,
"grad_norm": 4.156160052480647,
"learning_rate": 2.547887203052786e-06,
"loss": 0.4682,
"step": 259
},
{
"epoch": 0.5118110236220472,
"grad_norm": 0.8806621051495006,
"learning_rate": 2.531925886789935e-06,
"loss": 0.4055,
"step": 260
},
{
"epoch": 0.5137795275590551,
"grad_norm": 0.8166380192798925,
"learning_rate": 2.515963268826698e-06,
"loss": 0.4172,
"step": 261
},
{
"epoch": 0.515748031496063,
"grad_norm": 0.8474794194583648,
"learning_rate": 2.5e-06,
"loss": 0.4054,
"step": 262
},
{
"epoch": 0.5177165354330708,
"grad_norm": 0.8625258165448675,
"learning_rate": 2.4840367311733023e-06,
"loss": 0.3982,
"step": 263
},
{
"epoch": 0.5196850393700787,
"grad_norm": 0.8312498910086424,
"learning_rate": 2.4680741132100664e-06,
"loss": 0.4446,
"step": 264
},
{
"epoch": 0.5216535433070866,
"grad_norm": 1.2996938473325663,
"learning_rate": 2.452112796947215e-06,
"loss": 0.4721,
"step": 265
},
{
"epoch": 0.5236220472440944,
"grad_norm": 0.7749604922653435,
"learning_rate": 2.4361534331686002e-06,
"loss": 0.4437,
"step": 266
},
{
"epoch": 0.5255905511811023,
"grad_norm": 1.2900722363673192,
"learning_rate": 2.420196672578465e-06,
"loss": 0.4791,
"step": 267
},
{
"epoch": 0.5275590551181102,
"grad_norm": 2.3581594405299335,
"learning_rate": 2.404243165774912e-06,
"loss": 0.4225,
"step": 268
},
{
"epoch": 0.5295275590551181,
"grad_norm": 1.0362736267414994,
"learning_rate": 2.3882935632233805e-06,
"loss": 0.514,
"step": 269
},
{
"epoch": 0.531496062992126,
"grad_norm": 0.9331680987472524,
"learning_rate": 2.372348515230124e-06,
"loss": 0.4193,
"step": 270
},
{
"epoch": 0.5334645669291339,
"grad_norm": 1.1905250166916495,
"learning_rate": 2.356408671915692e-06,
"loss": 0.5042,
"step": 271
},
{
"epoch": 0.5354330708661418,
"grad_norm": 0.8572786945892718,
"learning_rate": 2.340474683188429e-06,
"loss": 0.4399,
"step": 272
},
{
"epoch": 0.5374015748031497,
"grad_norm": 1.0648664039670561,
"learning_rate": 2.324547198717972e-06,
"loss": 0.4741,
"step": 273
},
{
"epoch": 0.5393700787401575,
"grad_norm": 1.9327803933661767,
"learning_rate": 2.308626867908761e-06,
"loss": 0.4574,
"step": 274
},
{
"epoch": 0.5413385826771654,
"grad_norm": 0.9128616465522047,
"learning_rate": 2.2927143398735664e-06,
"loss": 0.4437,
"step": 275
},
{
"epoch": 0.5433070866141733,
"grad_norm": 0.873444229946243,
"learning_rate": 2.2768102634070146e-06,
"loss": 0.4382,
"step": 276
},
{
"epoch": 0.5452755905511811,
"grad_norm": 0.9487309713553738,
"learning_rate": 2.2609152869591445e-06,
"loss": 0.4964,
"step": 277
},
{
"epoch": 0.547244094488189,
"grad_norm": 1.5543510034516503,
"learning_rate": 2.2450300586089626e-06,
"loss": 0.458,
"step": 278
},
{
"epoch": 0.5492125984251969,
"grad_norm": 0.843153283370843,
"learning_rate": 2.229155226038017e-06,
"loss": 0.5173,
"step": 279
},
{
"epoch": 0.5511811023622047,
"grad_norm": 1.0206989602121845,
"learning_rate": 2.2132914365039993e-06,
"loss": 0.551,
"step": 280
},
{
"epoch": 0.5531496062992126,
"grad_norm": 0.8733293387877374,
"learning_rate": 2.1974393368143433e-06,
"loss": 0.4698,
"step": 281
},
{
"epoch": 0.5551181102362205,
"grad_norm": 0.9045357110695507,
"learning_rate": 2.1815995732998586e-06,
"loss": 0.4783,
"step": 282
},
{
"epoch": 0.5570866141732284,
"grad_norm": 1.1714298674446673,
"learning_rate": 2.165772791788379e-06,
"loss": 0.4505,
"step": 283
},
{
"epoch": 0.5590551181102362,
"grad_norm": 1.0449206200708,
"learning_rate": 2.1499596375784284e-06,
"loss": 0.4361,
"step": 284
},
{
"epoch": 0.5610236220472441,
"grad_norm": 0.9414076112602807,
"learning_rate": 2.1341607554129073e-06,
"loss": 0.4665,
"step": 285
},
{
"epoch": 0.562992125984252,
"grad_norm": 1.0364017694779857,
"learning_rate": 2.1183767894528135e-06,
"loss": 0.436,
"step": 286
},
{
"epoch": 0.5649606299212598,
"grad_norm": 0.9062612467100826,
"learning_rate": 2.1026083832509704e-06,
"loss": 0.526,
"step": 287
},
{
"epoch": 0.5669291338582677,
"grad_norm": 0.7845496876032904,
"learning_rate": 2.086856179725788e-06,
"loss": 0.4373,
"step": 288
},
{
"epoch": 0.5688976377952756,
"grad_norm": 1.5434192627506897,
"learning_rate": 2.0711208211350543e-06,
"loss": 0.495,
"step": 289
},
{
"epoch": 0.5708661417322834,
"grad_norm": 1.732032499885022,
"learning_rate": 2.055402949049746e-06,
"loss": 0.4232,
"step": 290
},
{
"epoch": 0.5728346456692913,
"grad_norm": 1.1791659079954764,
"learning_rate": 2.0397032043278686e-06,
"loss": 0.4766,
"step": 291
},
{
"epoch": 0.5748031496062992,
"grad_norm": 0.9826059946246523,
"learning_rate": 2.024022227088329e-06,
"loss": 0.4686,
"step": 292
},
{
"epoch": 0.5767716535433071,
"grad_norm": 1.1819752621296946,
"learning_rate": 2.008360656684837e-06,
"loss": 0.5234,
"step": 293
},
{
"epoch": 0.5787401574803149,
"grad_norm": 1.836095015890831,
"learning_rate": 1.9927191316798333e-06,
"loss": 0.3621,
"step": 294
},
{
"epoch": 0.5807086614173228,
"grad_norm": 1.0606463475234196,
"learning_rate": 1.977098289818459e-06,
"loss": 0.3665,
"step": 295
},
{
"epoch": 0.5826771653543307,
"grad_norm": 0.8708744885029884,
"learning_rate": 1.961498768002547e-06,
"loss": 0.4281,
"step": 296
},
{
"epoch": 0.5846456692913385,
"grad_norm": 0.9694572451225193,
"learning_rate": 1.9459212022646604e-06,
"loss": 0.4269,
"step": 297
},
{
"epoch": 0.5866141732283464,
"grad_norm": 1.52601054119704,
"learning_rate": 1.930366227742157e-06,
"loss": 0.4095,
"step": 298
},
{
"epoch": 0.5885826771653543,
"grad_norm": 0.8299536757583065,
"learning_rate": 1.914834478651288e-06,
"loss": 0.5123,
"step": 299
},
{
"epoch": 0.5905511811023622,
"grad_norm": 0.9288221226511972,
"learning_rate": 1.8993265882613483e-06,
"loss": 0.4786,
"step": 300
},
{
"epoch": 0.59251968503937,
"grad_norm": 1.9153123784076878,
"learning_rate": 1.8838431888688528e-06,
"loss": 0.442,
"step": 301
},
{
"epoch": 0.594488188976378,
"grad_norm": 0.8883998580295289,
"learning_rate": 1.8683849117717518e-06,
"loss": 0.4459,
"step": 302
},
{
"epoch": 0.5964566929133859,
"grad_norm": 0.9033764630985829,
"learning_rate": 1.852952387243698e-06,
"loss": 0.4801,
"step": 303
},
{
"epoch": 0.5984251968503937,
"grad_norm": 1.0771433466136495,
"learning_rate": 1.8375462445083464e-06,
"loss": 0.4227,
"step": 304
},
{
"epoch": 0.6003937007874016,
"grad_norm": 1.1264043096408438,
"learning_rate": 1.8221671117136953e-06,
"loss": 0.4932,
"step": 305
},
{
"epoch": 0.6023622047244095,
"grad_norm": 1.08738798244193,
"learning_rate": 1.80681561590648e-06,
"loss": 0.5191,
"step": 306
},
{
"epoch": 0.6043307086614174,
"grad_norm": 0.7987448502017881,
"learning_rate": 1.7914923830066074e-06,
"loss": 0.4618,
"step": 307
},
{
"epoch": 0.6062992125984252,
"grad_norm": 0.9944446151408849,
"learning_rate": 1.7761980377816287e-06,
"loss": 0.4358,
"step": 308
},
{
"epoch": 0.6082677165354331,
"grad_norm": 1.0742200261888726,
"learning_rate": 1.7609332038212728e-06,
"loss": 0.4193,
"step": 309
},
{
"epoch": 0.610236220472441,
"grad_norm": 1.0872956708805233,
"learning_rate": 1.7456985035120194e-06,
"loss": 0.4545,
"step": 310
},
{
"epoch": 0.6122047244094488,
"grad_norm": 1.0692325907893834,
"learning_rate": 1.7304945580117195e-06,
"loss": 0.4365,
"step": 311
},
{
"epoch": 0.6141732283464567,
"grad_norm": 1.4216199373627865,
"learning_rate": 1.7153219872242727e-06,
"loss": 0.4552,
"step": 312
},
{
"epoch": 0.6161417322834646,
"grad_norm": 0.9025617724171956,
"learning_rate": 1.700181409774353e-06,
"loss": 0.4142,
"step": 313
},
{
"epoch": 0.6181102362204725,
"grad_norm": 1.4701867077129656,
"learning_rate": 1.6850734429821792e-06,
"loss": 0.4939,
"step": 314
},
{
"epoch": 0.6200787401574803,
"grad_norm": 1.6211825794563106,
"learning_rate": 1.6699987028383547e-06,
"loss": 0.4243,
"step": 315
},
{
"epoch": 0.6220472440944882,
"grad_norm": 1.2528280691205667,
"learning_rate": 1.6549578039787436e-06,
"loss": 0.4472,
"step": 316
},
{
"epoch": 0.6240157480314961,
"grad_norm": 1.4234084359812396,
"learning_rate": 1.6399513596594158e-06,
"loss": 0.3996,
"step": 317
},
{
"epoch": 0.6259842519685039,
"grad_norm": 1.2903411072700746,
"learning_rate": 1.6249799817316415e-06,
"loss": 0.417,
"step": 318
},
{
"epoch": 0.6279527559055118,
"grad_norm": 1.0448194221823608,
"learning_rate": 1.6100442806169423e-06,
"loss": 0.4193,
"step": 319
},
{
"epoch": 0.6299212598425197,
"grad_norm": 0.7502474376681744,
"learning_rate": 1.595144865282205e-06,
"loss": 0.5476,
"step": 320
},
{
"epoch": 0.6318897637795275,
"grad_norm": 0.8490377322709838,
"learning_rate": 1.5802823432148545e-06,
"loss": 0.4885,
"step": 321
},
{
"epoch": 0.6338582677165354,
"grad_norm": 1.2648831300690013,
"learning_rate": 1.5654573203980782e-06,
"loss": 0.4387,
"step": 322
},
{
"epoch": 0.6358267716535433,
"grad_norm": 0.8514651671613641,
"learning_rate": 1.5506704012861257e-06,
"loss": 0.4465,
"step": 323
},
{
"epoch": 0.6377952755905512,
"grad_norm": 1.0222395663644381,
"learning_rate": 1.5359221887796616e-06,
"loss": 0.437,
"step": 324
},
{
"epoch": 0.639763779527559,
"grad_norm": 0.9497855107455955,
"learning_rate": 1.5212132842011778e-06,
"loss": 0.5466,
"step": 325
},
{
"epoch": 0.6417322834645669,
"grad_norm": 0.8431905728595827,
"learning_rate": 1.5065442872704872e-06,
"loss": 0.5627,
"step": 326
},
{
"epoch": 0.6437007874015748,
"grad_norm": 1.082255095866634,
"learning_rate": 1.4919157960802619e-06,
"loss": 0.4315,
"step": 327
},
{
"epoch": 0.6456692913385826,
"grad_norm": 1.1418415531966362,
"learning_rate": 1.4773284070716504e-06,
"loss": 0.506,
"step": 328
},
{
"epoch": 0.6476377952755905,
"grad_norm": 1.0386895105443592,
"learning_rate": 1.4627827150099627e-06,
"loss": 0.4045,
"step": 329
},
{
"epoch": 0.6496062992125984,
"grad_norm": 0.7965031672604527,
"learning_rate": 1.4482793129604148e-06,
"loss": 0.4128,
"step": 330
},
{
"epoch": 0.6515748031496063,
"grad_norm": 1.0706828235481562,
"learning_rate": 1.4338187922639506e-06,
"loss": 0.4873,
"step": 331
},
{
"epoch": 0.6535433070866141,
"grad_norm": 1.010780223186052,
"learning_rate": 1.4194017425131324e-06,
"loss": 0.4529,
"step": 332
},
{
"epoch": 0.655511811023622,
"grad_norm": 0.9600940507276752,
"learning_rate": 1.405028751528099e-06,
"loss": 0.4724,
"step": 333
},
{
"epoch": 0.65748031496063,
"grad_norm": 0.9764309603793133,
"learning_rate": 1.3907004053326006e-06,
"loss": 0.3466,
"step": 334
},
{
"epoch": 0.6594488188976378,
"grad_norm": 1.1423974049119299,
"learning_rate": 1.3764172881301063e-06,
"loss": 0.4322,
"step": 335
},
{
"epoch": 0.6614173228346457,
"grad_norm": 1.2060374393851416,
"learning_rate": 1.362179982279979e-06,
"loss": 0.4461,
"step": 336
},
{
"epoch": 0.6633858267716536,
"grad_norm": 1.2086001572935363,
"learning_rate": 1.347989068273738e-06,
"loss": 0.3849,
"step": 337
},
{
"epoch": 0.6653543307086615,
"grad_norm": 0.8627331404783796,
"learning_rate": 1.3338451247113898e-06,
"loss": 0.4661,
"step": 338
},
{
"epoch": 0.6653543307086615,
"eval_accuracy": 0.8276808215111454,
"eval_accuracy_first_token": 0.7754862150032058,
"eval_accuracy_first_token_<": 0.9599789362822538,
"eval_accuracy_first_token_<_total": 1899,
"eval_accuracy_first_token_<|python_tag|>": 0.9202020202020202,
"eval_accuracy_first_token_<|python_tag|>_total": 990,
"eval_accuracy_first_token_Here": 0.6918429003021148,
"eval_accuracy_first_token_Here_total": 331,
"eval_accuracy_first_token_The": 0.8801996672212978,
"eval_accuracy_first_token_The_total": 2404,
"eval_accuracy_first_token_To": 0.8060344827586207,
"eval_accuracy_first_token_To_total": 928,
"eval_loss": 0.4993845224380493,
"eval_perplexity": 1.1952336899234497,
"eval_runtime": 550.918,
"eval_samples_per_second": 1.251,
"eval_steps_per_second": 0.158,
"eval_total_number_first_token": 9358,
"step": 338
},
{
"epoch": 0.6673228346456693,
"grad_norm": 0.8859890191170555,
"learning_rate": 1.3197487282778315e-06,
"loss": 0.4468,
"step": 339
},
{
"epoch": 0.6692913385826772,
"grad_norm": 1.0644157193112311,
"learning_rate": 1.3057004537193424e-06,
"loss": 0.4267,
"step": 340
},
{
"epoch": 0.6712598425196851,
"grad_norm": 1.1533288398188408,
"learning_rate": 1.2917008738201536e-06,
"loss": 0.4906,
"step": 341
},
{
"epoch": 0.6732283464566929,
"grad_norm": 0.8272350855889613,
"learning_rate": 1.2777505593790834e-06,
"loss": 0.4833,
"step": 342
},
{
"epoch": 0.6751968503937008,
"grad_norm": 1.32130782853451,
"learning_rate": 1.263850079186274e-06,
"loss": 0.4463,
"step": 343
},
{
"epoch": 0.6771653543307087,
"grad_norm": 0.8065185454495468,
"learning_rate": 1.2500000000000007e-06,
"loss": 0.3829,
"step": 344
},
{
"epoch": 0.6791338582677166,
"grad_norm": 0.7422824598097033,
"learning_rate": 1.2362008865235536e-06,
"loss": 0.3774,
"step": 345
},
{
"epoch": 0.6811023622047244,
"grad_norm": 1.0704521032466394,
"learning_rate": 1.2224533013822237e-06,
"loss": 0.4497,
"step": 346
},
{
"epoch": 0.6830708661417323,
"grad_norm": 1.4370970783293397,
"learning_rate": 1.2087578051003619e-06,
"loss": 0.3994,
"step": 347
},
{
"epoch": 0.6850393700787402,
"grad_norm": 1.0762816440657645,
"learning_rate": 1.1951149560785166e-06,
"loss": 0.4879,
"step": 348
},
{
"epoch": 0.687007874015748,
"grad_norm": 0.999638462276997,
"learning_rate": 1.181525310570677e-06,
"loss": 0.4049,
"step": 349
},
{
"epoch": 0.6889763779527559,
"grad_norm": 0.7125141978940648,
"learning_rate": 1.1679894226615861e-06,
"loss": 0.47,
"step": 350
},
{
"epoch": 0.6909448818897638,
"grad_norm": 2.3784943021939,
"learning_rate": 1.1545078442441526e-06,
"loss": 0.3979,
"step": 351
},
{
"epoch": 0.6929133858267716,
"grad_norm": 0.7438302581823832,
"learning_rate": 1.1410811249969475e-06,
"loss": 0.4839,
"step": 352
},
{
"epoch": 0.6948818897637795,
"grad_norm": 0.8945326825925858,
"learning_rate": 1.1277098123617922e-06,
"loss": 0.4467,
"step": 353
},
{
"epoch": 0.6968503937007874,
"grad_norm": 0.9603219580688872,
"learning_rate": 1.1143944515214385e-06,
"loss": 0.4495,
"step": 354
},
{
"epoch": 0.6988188976377953,
"grad_norm": 1.4664719141635538,
"learning_rate": 1.10113558537734e-06,
"loss": 0.4944,
"step": 355
},
{
"epoch": 0.7007874015748031,
"grad_norm": 1.2476331714652613,
"learning_rate": 1.0879337545275165e-06,
"loss": 0.5793,
"step": 356
},
{
"epoch": 0.702755905511811,
"grad_norm": 0.9156400646754094,
"learning_rate": 1.074789497244512e-06,
"loss": 0.4221,
"step": 357
},
{
"epoch": 0.7047244094488189,
"grad_norm": 2.5318845730546196,
"learning_rate": 1.0617033494534486e-06,
"loss": 0.4377,
"step": 358
},
{
"epoch": 0.7066929133858267,
"grad_norm": 1.2015844613438207,
"learning_rate": 1.0486758447101753e-06,
"loss": 0.4659,
"step": 359
},
{
"epoch": 0.7086614173228346,
"grad_norm": 0.9516842380567236,
"learning_rate": 1.035707514179513e-06,
"loss": 0.4184,
"step": 360
},
{
"epoch": 0.7106299212598425,
"grad_norm": 0.9525083829856291,
"learning_rate": 1.0227988866135995e-06,
"loss": 0.358,
"step": 361
},
{
"epoch": 0.7125984251968503,
"grad_norm": 1.4467162043063637,
"learning_rate": 1.0099504883303255e-06,
"loss": 0.3927,
"step": 362
},
{
"epoch": 0.7145669291338582,
"grad_norm": 1.186260329222654,
"learning_rate": 9.971628431918845e-07,
"loss": 0.4433,
"step": 363
},
{
"epoch": 0.7165354330708661,
"grad_norm": 0.8887871636025992,
"learning_rate": 9.844364725834058e-07,
"loss": 0.4411,
"step": 364
},
{
"epoch": 0.718503937007874,
"grad_norm": 1.0782965924363137,
"learning_rate": 9.71771895391696e-07,
"loss": 0.4019,
"step": 365
},
{
"epoch": 0.7204724409448819,
"grad_norm": 0.8292803711940244,
"learning_rate": 9.591696279840906e-07,
"loss": 0.4356,
"step": 366
},
{
"epoch": 0.7224409448818898,
"grad_norm": 1.1636402977394662,
"learning_rate": 9.466301841873929e-07,
"loss": 0.5021,
"step": 367
},
{
"epoch": 0.7244094488188977,
"grad_norm": 1.2022996028060275,
"learning_rate": 9.341540752669234e-07,
"loss": 0.5142,
"step": 368
},
{
"epoch": 0.7263779527559056,
"grad_norm": 1.0809958010299958,
"learning_rate": 9.217418099056821e-07,
"loss": 0.4273,
"step": 369
},
{
"epoch": 0.7283464566929134,
"grad_norm": 0.9929276956854678,
"learning_rate": 9.093938941836012e-07,
"loss": 0.4444,
"step": 370
},
{
"epoch": 0.7303149606299213,
"grad_norm": 0.7447267161683555,
"learning_rate": 8.971108315569094e-07,
"loss": 0.394,
"step": 371
},
{
"epoch": 0.7322834645669292,
"grad_norm": 0.8131869419909002,
"learning_rate": 8.848931228376137e-07,
"loss": 0.468,
"step": 372
},
{
"epoch": 0.734251968503937,
"grad_norm": 0.8078315201370335,
"learning_rate": 8.727412661730724e-07,
"loss": 0.5374,
"step": 373
},
{
"epoch": 0.7362204724409449,
"grad_norm": 0.8611154445074285,
"learning_rate": 8.606557570256844e-07,
"loss": 0.4585,
"step": 374
},
{
"epoch": 0.7381889763779528,
"grad_norm": 0.9620990091277316,
"learning_rate": 8.486370881526917e-07,
"loss": 0.4473,
"step": 375
},
{
"epoch": 0.7401574803149606,
"grad_norm": 1.072939801506809,
"learning_rate": 8.366857495860869e-07,
"loss": 0.4152,
"step": 376
},
{
"epoch": 0.7421259842519685,
"grad_norm": 1.0168662482449515,
"learning_rate": 8.24802228612633e-07,
"loss": 0.4174,
"step": 377
},
{
"epoch": 0.7440944881889764,
"grad_norm": 0.8387562652899657,
"learning_rate": 8.129870097539951e-07,
"loss": 0.5152,
"step": 378
},
{
"epoch": 0.7460629921259843,
"grad_norm": 1.1029564445037212,
"learning_rate": 8.012405747469861e-07,
"loss": 0.534,
"step": 379
},
{
"epoch": 0.7480314960629921,
"grad_norm": 0.7201029027810675,
"learning_rate": 7.895634025239243e-07,
"loss": 0.4516,
"step": 380
},
{
"epoch": 0.75,
"grad_norm": 0.9054127475907084,
"learning_rate": 7.779559691931065e-07,
"loss": 0.4897,
"step": 381
},
{
"epoch": 0.7519685039370079,
"grad_norm": 0.9229217986975267,
"learning_rate": 7.66418748019396e-07,
"loss": 0.4883,
"step": 382
},
{
"epoch": 0.7539370078740157,
"grad_norm": 0.9011840634292719,
"learning_rate": 7.54952209404926e-07,
"loss": 0.4866,
"step": 383
},
{
"epoch": 0.7559055118110236,
"grad_norm": 0.855848115039962,
"learning_rate": 7.435568208699203e-07,
"loss": 0.3808,
"step": 384
},
{
"epoch": 0.7578740157480315,
"grad_norm": 1.1287210701987485,
"learning_rate": 7.322330470336314e-07,
"loss": 0.3988,
"step": 385
},
{
"epoch": 0.7598425196850394,
"grad_norm": 0.9772116372811883,
"learning_rate": 7.209813495953963e-07,
"loss": 0.5117,
"step": 386
},
{
"epoch": 0.7618110236220472,
"grad_norm": 1.0255758644993986,
"learning_rate": 7.098021873158126e-07,
"loss": 0.438,
"step": 387
},
{
"epoch": 0.7637795275590551,
"grad_norm": 0.9832119201339524,
"learning_rate": 6.986960159980327e-07,
"loss": 0.5492,
"step": 388
},
{
"epoch": 0.765748031496063,
"grad_norm": 0.9391130557550369,
"learning_rate": 6.876632884691803e-07,
"loss": 0.395,
"step": 389
},
{
"epoch": 0.7677165354330708,
"grad_norm": 0.9181705334242035,
"learning_rate": 6.767044545618879e-07,
"loss": 0.4503,
"step": 390
},
{
"epoch": 0.7696850393700787,
"grad_norm": 1.0300761390988038,
"learning_rate": 6.658199610959537e-07,
"loss": 0.474,
"step": 391
},
{
"epoch": 0.7716535433070866,
"grad_norm": 0.8708437915946684,
"learning_rate": 6.55010251860127e-07,
"loss": 0.4441,
"step": 392
},
{
"epoch": 0.7736220472440944,
"grad_norm": 0.9670344205389225,
"learning_rate": 6.442757675940109e-07,
"loss": 0.4131,
"step": 393
},
{
"epoch": 0.7755905511811023,
"grad_norm": 0.9278345314974272,
"learning_rate": 6.336169459700933e-07,
"loss": 0.4111,
"step": 394
},
{
"epoch": 0.7775590551181102,
"grad_norm": 0.8525334343828224,
"learning_rate": 6.230342215759027e-07,
"loss": 0.3805,
"step": 395
},
{
"epoch": 0.7795275590551181,
"grad_norm": 0.8818820675161857,
"learning_rate": 6.125280258962874e-07,
"loss": 0.426,
"step": 396
},
{
"epoch": 0.781496062992126,
"grad_norm": 1.5469693241838445,
"learning_rate": 6.020987872958237e-07,
"loss": 0.5351,
"step": 397
},
{
"epoch": 0.7834645669291339,
"grad_norm": 0.8899842684673445,
"learning_rate": 5.91746931001351e-07,
"loss": 0.4157,
"step": 398
},
{
"epoch": 0.7854330708661418,
"grad_norm": 0.8232301686270491,
"learning_rate": 5.814728790846308e-07,
"loss": 0.3969,
"step": 399
},
{
"epoch": 0.7874015748031497,
"grad_norm": 1.0153791863734263,
"learning_rate": 5.712770504451426e-07,
"loss": 0.4635,
"step": 400
},
{
"epoch": 0.7893700787401575,
"grad_norm": 0.8407425085704395,
"learning_rate": 5.611598607930033e-07,
"loss": 0.4233,
"step": 401
},
{
"epoch": 0.7913385826771654,
"grad_norm": 0.8392579630637883,
"learning_rate": 5.511217226320126e-07,
"loss": 0.4487,
"step": 402
},
{
"epoch": 0.7933070866141733,
"grad_norm": 1.114866286158925,
"learning_rate": 5.411630452428396e-07,
"loss": 0.4769,
"step": 403
},
{
"epoch": 0.7952755905511811,
"grad_norm": 1.0781688957676823,
"learning_rate": 5.312842346663363e-07,
"loss": 0.3288,
"step": 404
},
{
"epoch": 0.797244094488189,
"grad_norm": 0.776231861916159,
"learning_rate": 5.214856936869753e-07,
"loss": 0.4266,
"step": 405
},
{
"epoch": 0.7992125984251969,
"grad_norm": 0.762343087232223,
"learning_rate": 5.117678218164337e-07,
"loss": 0.4505,
"step": 406
},
{
"epoch": 0.8011811023622047,
"grad_norm": 0.8033552900917362,
"learning_rate": 5.021310152773034e-07,
"loss": 0.4262,
"step": 407
},
{
"epoch": 0.8031496062992126,
"grad_norm": 0.9170707693341349,
"learning_rate": 4.925756669869313e-07,
"loss": 0.4283,
"step": 408
},
{
"epoch": 0.8051181102362205,
"grad_norm": 1.129653313885886,
"learning_rate": 4.831021665414043e-07,
"loss": 0.3817,
"step": 409
},
{
"epoch": 0.8070866141732284,
"grad_norm": 1.4424483504544063,
"learning_rate": 4.737109001996637e-07,
"loss": 0.4238,
"step": 410
},
{
"epoch": 0.8090551181102362,
"grad_norm": 1.0131174677216408,
"learning_rate": 4.6440225086775176e-07,
"loss": 0.4574,
"step": 411
},
{
"epoch": 0.8110236220472441,
"grad_norm": 1.0317767560783413,
"learning_rate": 4.55176598083206e-07,
"loss": 0.4272,
"step": 412
},
{
"epoch": 0.812992125984252,
"grad_norm": 1.0233220042513846,
"learning_rate": 4.4603431799958074e-07,
"loss": 0.4473,
"step": 413
},
{
"epoch": 0.8149606299212598,
"grad_norm": 0.8542754328879538,
"learning_rate": 4.3697578337111053e-07,
"loss": 0.4485,
"step": 414
},
{
"epoch": 0.8169291338582677,
"grad_norm": 1.0943473819404532,
"learning_rate": 4.280013635375138e-07,
"loss": 0.4701,
"step": 415
},
{
"epoch": 0.8188976377952756,
"grad_norm": 0.9268802927680222,
"learning_rate": 4.1911142440893197e-07,
"loss": 0.4236,
"step": 416
},
{
"epoch": 0.8208661417322834,
"grad_norm": 0.8831699289176582,
"learning_rate": 4.103063284510117e-07,
"loss": 0.4366,
"step": 417
},
{
"epoch": 0.8228346456692913,
"grad_norm": 1.4025172389257854,
"learning_rate": 4.015864346701251e-07,
"loss": 0.4431,
"step": 418
},
{
"epoch": 0.8248031496062992,
"grad_norm": 1.1860309850463138,
"learning_rate": 3.929520985987334e-07,
"loss": 0.4141,
"step": 419
},
{
"epoch": 0.8267716535433071,
"grad_norm": 0.8197597639637244,
"learning_rate": 3.8440367228089e-07,
"loss": 0.5017,
"step": 420
},
{
"epoch": 0.8287401574803149,
"grad_norm": 0.9362851536852345,
"learning_rate": 3.7594150425788677e-07,
"loss": 0.4179,
"step": 421
},
{
"epoch": 0.8307086614173228,
"grad_norm": 1.3750907694593544,
"learning_rate": 3.6756593955404403e-07,
"loss": 0.426,
"step": 422
},
{
"epoch": 0.8326771653543307,
"grad_norm": 0.8182694196765484,
"learning_rate": 3.5927731966264175e-07,
"loss": 0.3848,
"step": 423
},
{
"epoch": 0.8346456692913385,
"grad_norm": 1.1511162896225982,
"learning_rate": 3.510759825319976e-07,
"loss": 0.495,
"step": 424
},
{
"epoch": 0.8366141732283464,
"grad_norm": 1.1082154782574505,
"learning_rate": 3.4296226255168484e-07,
"loss": 0.4561,
"step": 425
},
{
"epoch": 0.8385826771653543,
"grad_norm": 1.352703252528523,
"learning_rate": 3.3493649053890325e-07,
"loss": 0.3687,
"step": 426
},
{
"epoch": 0.8405511811023622,
"grad_norm": 0.9424678467646581,
"learning_rate": 3.2699899372498736e-07,
"loss": 0.4757,
"step": 427
},
{
"epoch": 0.84251968503937,
"grad_norm": 1.1072236275545695,
"learning_rate": 3.191500957420626e-07,
"loss": 0.4473,
"step": 428
},
{
"epoch": 0.844488188976378,
"grad_norm": 1.8305124197035552,
"learning_rate": 3.1139011660985624e-07,
"loss": 0.4529,
"step": 429
},
{
"epoch": 0.8464566929133859,
"grad_norm": 1.093729672924255,
"learning_rate": 3.0371937272264454e-07,
"loss": 0.5216,
"step": 430
},
{
"epoch": 0.8484251968503937,
"grad_norm": 0.958833049325444,
"learning_rate": 2.961381768363511e-07,
"loss": 0.5062,
"step": 431
},
{
"epoch": 0.8503937007874016,
"grad_norm": 1.1796242400127814,
"learning_rate": 2.8864683805580136e-07,
"loss": 0.4664,
"step": 432
},
{
"epoch": 0.8523622047244095,
"grad_norm": 0.8060870545610721,
"learning_rate": 2.8124566182211434e-07,
"loss": 0.5173,
"step": 433
},
{
"epoch": 0.8543307086614174,
"grad_norm": 1.0949935927704386,
"learning_rate": 2.7393494990024835e-07,
"loss": 0.4514,
"step": 434
},
{
"epoch": 0.8562992125984252,
"grad_norm": 1.9213755509125885,
"learning_rate": 2.6671500036670324e-07,
"loss": 0.457,
"step": 435
},
{
"epoch": 0.8582677165354331,
"grad_norm": 0.7576072257543266,
"learning_rate": 2.5958610759736133e-07,
"loss": 0.3776,
"step": 436
},
{
"epoch": 0.860236220472441,
"grad_norm": 0.9698921361904715,
"learning_rate": 2.5254856225548546e-07,
"loss": 0.4309,
"step": 437
},
{
"epoch": 0.8622047244094488,
"grad_norm": 1.1235058717255908,
"learning_rate": 2.456026512798715e-07,
"loss": 0.4771,
"step": 438
},
{
"epoch": 0.8641732283464567,
"grad_norm": 0.976026525281605,
"learning_rate": 2.38748657873146e-07,
"loss": 0.4859,
"step": 439
},
{
"epoch": 0.8661417322834646,
"grad_norm": 0.7648607059270878,
"learning_rate": 2.3198686149022016e-07,
"loss": 0.4224,
"step": 440
},
{
"epoch": 0.8681102362204725,
"grad_norm": 1.2296857492126643,
"learning_rate": 2.2531753782689598e-07,
"loss": 0.3892,
"step": 441
},
{
"epoch": 0.8700787401574803,
"grad_norm": 1.0696201173297104,
"learning_rate": 2.1874095880862505e-07,
"loss": 0.4382,
"step": 442
},
{
"epoch": 0.8720472440944882,
"grad_norm": 0.8951240141423281,
"learning_rate": 2.122573925794219e-07,
"loss": 0.4117,
"step": 443
},
{
"epoch": 0.8740157480314961,
"grad_norm": 0.8173547373043929,
"learning_rate": 2.0586710349093013e-07,
"loss": 0.4907,
"step": 444
},
{
"epoch": 0.8759842519685039,
"grad_norm": 1.1001975619027133,
"learning_rate": 1.9957035209164562e-07,
"loss": 0.4993,
"step": 445
},
{
"epoch": 0.8779527559055118,
"grad_norm": 0.9491438659517182,
"learning_rate": 1.9336739511629232e-07,
"loss": 0.4663,
"step": 446
},
{
"epoch": 0.8799212598425197,
"grad_norm": 1.0810133349772146,
"learning_rate": 1.8725848547535368e-07,
"loss": 0.5156,
"step": 447
},
{
"epoch": 0.8818897637795275,
"grad_norm": 0.7618874778695859,
"learning_rate": 1.8124387224476347e-07,
"loss": 0.4335,
"step": 448
},
{
"epoch": 0.8838582677165354,
"grad_norm": 0.9430841395865759,
"learning_rate": 1.753238006557473e-07,
"loss": 0.383,
"step": 449
},
{
"epoch": 0.8858267716535433,
"grad_norm": 0.8657928111335921,
"learning_rate": 1.6949851208482583e-07,
"loss": 0.4255,
"step": 450
},
{
"epoch": 0.8877952755905512,
"grad_norm": 1.7161732171227453,
"learning_rate": 1.6376824404397252e-07,
"loss": 0.4448,
"step": 451
},
{
"epoch": 0.889763779527559,
"grad_norm": 1.270254639155913,
"learning_rate": 1.5813323017093042e-07,
"loss": 0.4279,
"step": 452
},
{
"epoch": 0.8917322834645669,
"grad_norm": 0.8123015748331465,
"learning_rate": 1.525937002196845e-07,
"loss": 0.403,
"step": 453
},
{
"epoch": 0.8937007874015748,
"grad_norm": 1.334895078629325,
"learning_rate": 1.471498800510962e-07,
"loss": 0.4821,
"step": 454
},
{
"epoch": 0.8956692913385826,
"grad_norm": 1.0041108869094808,
"learning_rate": 1.4180199162369207e-07,
"loss": 0.4335,
"step": 455
},
{
"epoch": 0.8976377952755905,
"grad_norm": 0.8554268743987077,
"learning_rate": 1.3655025298461662e-07,
"loss": 0.464,
"step": 456
},
{
"epoch": 0.8996062992125984,
"grad_norm": 0.8749648836446161,
"learning_rate": 1.3139487826073937e-07,
"loss": 0.4395,
"step": 457
},
{
"epoch": 0.9015748031496063,
"grad_norm": 1.033997084622728,
"learning_rate": 1.2633607764992673e-07,
"loss": 0.4063,
"step": 458
},
{
"epoch": 0.9035433070866141,
"grad_norm": 0.8497913384972722,
"learning_rate": 1.2137405741246916e-07,
"loss": 0.41,
"step": 459
},
{
"epoch": 0.905511811023622,
"grad_norm": 10.200487342112707,
"learning_rate": 1.1650901986267365e-07,
"loss": 0.4292,
"step": 460
},
{
"epoch": 0.90748031496063,
"grad_norm": 0.8025258853932543,
"learning_rate": 1.1174116336061469e-07,
"loss": 0.3503,
"step": 461
},
{
"epoch": 0.9094488188976378,
"grad_norm": 0.8215353071574096,
"learning_rate": 1.0707068230404406e-07,
"loss": 0.4633,
"step": 462
},
{
"epoch": 0.9114173228346457,
"grad_norm": 1.146375660721515,
"learning_rate": 1.0249776712046745e-07,
"loss": 0.4166,
"step": 463
},
{
"epoch": 0.9133858267716536,
"grad_norm": 0.8755933248602664,
"learning_rate": 9.802260425938098e-08,
"loss": 0.4614,
"step": 464
},
{
"epoch": 0.9153543307086615,
"grad_norm": 0.9696195760449576,
"learning_rate": 9.364537618466451e-08,
"loss": 0.4167,
"step": 465
},
{
"epoch": 0.9173228346456693,
"grad_norm": 1.0388943528957006,
"learning_rate": 8.936626136714754e-08,
"loss": 0.58,
"step": 466
},
{
"epoch": 0.9192913385826772,
"grad_norm": 0.8175192369303365,
"learning_rate": 8.518543427732951e-08,
"loss": 0.4174,
"step": 467
},
{
"epoch": 0.9212598425196851,
"grad_norm": 0.921738436654059,
"learning_rate": 8.110306537826601e-08,
"loss": 0.4204,
"step": 468
},
{
"epoch": 0.9232283464566929,
"grad_norm": 1.324496079833298,
"learning_rate": 7.711932111862024e-08,
"loss": 0.4078,
"step": 469
},
{
"epoch": 0.9251968503937008,
"grad_norm": 0.8739004378907311,
"learning_rate": 7.323436392587591e-08,
"loss": 0.4382,
"step": 470
},
{
"epoch": 0.9271653543307087,
"grad_norm": 1.4507666698538697,
"learning_rate": 6.94483521997133e-08,
"loss": 0.4327,
"step": 471
},
{
"epoch": 0.9291338582677166,
"grad_norm": 2.393429177400887,
"learning_rate": 6.576144030555259e-08,
"loss": 0.4487,
"step": 472
},
{
"epoch": 0.9311023622047244,
"grad_norm": 1.8599037252141386,
"learning_rate": 6.217377856825885e-08,
"loss": 0.4466,
"step": 473
},
{
"epoch": 0.9330708661417323,
"grad_norm": 14.522545485987576,
"learning_rate": 5.8685513266014125e-08,
"loss": 0.4628,
"step": 474
},
{
"epoch": 0.9350393700787402,
"grad_norm": 0.9566107624170317,
"learning_rate": 5.529678662435228e-08,
"loss": 0.4836,
"step": 475
},
{
"epoch": 0.937007874015748,
"grad_norm": 0.9657422719809673,
"learning_rate": 5.2007736810359696e-08,
"loss": 0.473,
"step": 476
},
{
"epoch": 0.9389763779527559,
"grad_norm": 1.2062447957929703,
"learning_rate": 4.881849792704368e-08,
"loss": 0.3763,
"step": 477
},
{
"epoch": 0.9409448818897638,
"grad_norm": 0.7659233057083479,
"learning_rate": 4.5729200007862686e-08,
"loss": 0.4329,
"step": 478
},
{
"epoch": 0.9429133858267716,
"grad_norm": 1.2608330514373431,
"learning_rate": 4.2739969011426075e-08,
"loss": 0.4357,
"step": 479
},
{
"epoch": 0.9448818897637795,
"grad_norm": 1.7397554811132858,
"learning_rate": 3.985092681635716e-08,
"loss": 0.429,
"step": 480
},
{
"epoch": 0.9468503937007874,
"grad_norm": 1.448927446479969,
"learning_rate": 3.70621912163252e-08,
"loss": 0.417,
"step": 481
},
{
"epoch": 0.9488188976377953,
"grad_norm": 1.1673579325201837,
"learning_rate": 3.43738759152415e-08,
"loss": 0.4629,
"step": 482
},
{
"epoch": 0.9507874015748031,
"grad_norm": 0.9845214613691381,
"learning_rate": 3.178609052262416e-08,
"loss": 0.3867,
"step": 483
},
{
"epoch": 0.952755905511811,
"grad_norm": 0.8723540832246429,
"learning_rate": 2.9298940549128962e-08,
"loss": 0.4598,
"step": 484
},
{
"epoch": 0.9547244094488189,
"grad_norm": 0.959905519952331,
"learning_rate": 2.691252740224637e-08,
"loss": 0.4659,
"step": 485
},
{
"epoch": 0.9566929133858267,
"grad_norm": 1.2404459615128476,
"learning_rate": 2.4626948382168725e-08,
"loss": 0.426,
"step": 486
},
{
"epoch": 0.9586614173228346,
"grad_norm": 0.893494717612766,
"learning_rate": 2.244229667782205e-08,
"loss": 0.4217,
"step": 487
},
{
"epoch": 0.9606299212598425,
"grad_norm": 1.1412736276476656,
"learning_rate": 2.0358661363065746e-08,
"loss": 0.5281,
"step": 488
},
{
"epoch": 0.9625984251968503,
"grad_norm": 1.2250423365692404,
"learning_rate": 1.837612739306216e-08,
"loss": 0.5264,
"step": 489
},
{
"epoch": 0.9645669291338582,
"grad_norm": 1.3877089712502941,
"learning_rate": 1.6494775600812418e-08,
"loss": 0.5014,
"step": 490
},
{
"epoch": 0.9665354330708661,
"grad_norm": 0.9974143156857421,
"learning_rate": 1.4714682693859616e-08,
"loss": 0.4364,
"step": 491
},
{
"epoch": 0.968503937007874,
"grad_norm": 1.3085113499447627,
"learning_rate": 1.3035921251163263e-08,
"loss": 0.5221,
"step": 492
},
{
"epoch": 0.9704724409448819,
"grad_norm": 0.9619634470389329,
"learning_rate": 1.1458559720137762e-08,
"loss": 0.4261,
"step": 493
},
{
"epoch": 0.9724409448818898,
"grad_norm": 1.2644328401227114,
"learning_rate": 9.982662413862975e-09,
"loss": 0.4717,
"step": 494
},
{
"epoch": 0.9744094488188977,
"grad_norm": 1.119378029772877,
"learning_rate": 8.608289508462708e-09,
"loss": 0.4448,
"step": 495
},
{
"epoch": 0.9763779527559056,
"grad_norm": 1.1725926045854247,
"learning_rate": 7.335497040648898e-09,
"loss": 0.4378,
"step": 496
},
{
"epoch": 0.9783464566929134,
"grad_norm": 0.8536423997109798,
"learning_rate": 6.164336905438994e-09,
"loss": 0.4371,
"step": 497
},
{
"epoch": 0.9803149606299213,
"grad_norm": 1.0090984449839402,
"learning_rate": 5.094856854039043e-09,
"loss": 0.3578,
"step": 498
},
{
"epoch": 0.9822834645669292,
"grad_norm": 1.0417737651445875,
"learning_rate": 4.1271004918971845e-09,
"loss": 0.361,
"step": 499
},
{
"epoch": 0.984251968503937,
"grad_norm": 1.01560610240081,
"learning_rate": 3.2611072769250797e-09,
"loss": 0.462,
"step": 500
},
{
"epoch": 0.9862204724409449,
"grad_norm": 0.9017494523501762,
"learning_rate": 2.4969125178903043e-09,
"loss": 0.4578,
"step": 501
},
{
"epoch": 0.9881889763779528,
"grad_norm": 0.837833251411561,
"learning_rate": 1.834547372975004e-09,
"loss": 0.4371,
"step": 502
},
{
"epoch": 0.9901574803149606,
"grad_norm": 0.9102649869347725,
"learning_rate": 1.2740388485071864e-09,
"loss": 0.5395,
"step": 503
},
{
"epoch": 0.9921259842519685,
"grad_norm": 0.8960123585568193,
"learning_rate": 8.154097978591014e-10,
"loss": 0.3902,
"step": 504
},
{
"epoch": 0.9940944881889764,
"grad_norm": 0.8174618918609772,
"learning_rate": 4.586789205140996e-10,
"loss": 0.387,
"step": 505
},
{
"epoch": 0.9960629921259843,
"grad_norm": 1.5873110230672518,
"learning_rate": 2.0386076130668453e-10,
"loss": 0.4675,
"step": 506
},
{
"epoch": 0.9980314960629921,
"grad_norm": 1.0566566451968318,
"learning_rate": 5.0965709827432984e-11,
"loss": 0.4223,
"step": 507
},
{
"epoch": 0.9980314960629921,
"eval_accuracy": 0.828535117694603,
"eval_accuracy_first_token": 0.7801880743748664,
"eval_accuracy_first_token_<": 0.9647182727751448,
"eval_accuracy_first_token_<_total": 1899,
"eval_accuracy_first_token_<|python_tag|>": 0.901010101010101,
"eval_accuracy_first_token_<|python_tag|>_total": 990,
"eval_accuracy_first_token_Here": 0.6132930513595166,
"eval_accuracy_first_token_Here_total": 331,
"eval_accuracy_first_token_The": 0.9009983361064892,
"eval_accuracy_first_token_The_total": 2404,
"eval_accuracy_first_token_To": 0.8125,
"eval_accuracy_first_token_To_total": 928,
"eval_loss": 0.4954877495765686,
"eval_perplexity": 1.1938056438441356,
"eval_runtime": 550.8232,
"eval_samples_per_second": 1.251,
"eval_steps_per_second": 0.158,
"eval_total_number_first_token": 9358,
"step": 507
},
{
"epoch": 1.0,
"grad_norm": 0.90365538285163,
"learning_rate": 0.0,
"loss": 0.4007,
"step": 508
},
{
"epoch": 1.0,
"step": 508,
"total_flos": 526396560506880.0,
"train_loss": 0.479396264325446,
"train_runtime": 60254.3679,
"train_samples_per_second": 0.135,
"train_steps_per_second": 0.008
}
],
"logging_steps": 1.0,
"max_steps": 508,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 5.0,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 526396560506880.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}