Aura-MoEv2 / checkpoint-260 /trainer_state.json
jeiku's picture
Training in progress, step 260, checkpoint
1e68cd0 verified
raw
history blame
46.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9980806142034548,
"eval_steps": 130,
"global_step": 260,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.003838771593090211,
"grad_norm": 41.942832946777344,
"learning_rate": 5e-06,
"loss": 29.5342,
"step": 1
},
{
"epoch": 0.003838771593090211,
"eval_loss": 1.869257926940918,
"eval_runtime": 31.5851,
"eval_samples_per_second": 9.372,
"eval_steps_per_second": 4.686,
"step": 1
},
{
"epoch": 0.007677543186180422,
"grad_norm": 37.62504577636719,
"learning_rate": 1e-05,
"loss": 29.602,
"step": 2
},
{
"epoch": 0.011516314779270634,
"grad_norm": 27.579742431640625,
"learning_rate": 1.5e-05,
"loss": 28.6884,
"step": 3
},
{
"epoch": 0.015355086372360844,
"grad_norm": 36.99763870239258,
"learning_rate": 2e-05,
"loss": 29.9575,
"step": 4
},
{
"epoch": 0.019193857965451054,
"grad_norm": 37.934635162353516,
"learning_rate": 2.5e-05,
"loss": 28.0644,
"step": 5
},
{
"epoch": 0.023032629558541268,
"grad_norm": 45.45661163330078,
"learning_rate": 3e-05,
"loss": 29.0135,
"step": 6
},
{
"epoch": 0.026871401151631478,
"grad_norm": 31.74431800842285,
"learning_rate": 3.5e-05,
"loss": 29.0581,
"step": 7
},
{
"epoch": 0.030710172744721688,
"grad_norm": 32.59209060668945,
"learning_rate": 4e-05,
"loss": 28.5253,
"step": 8
},
{
"epoch": 0.0345489443378119,
"grad_norm": 26.19696617126465,
"learning_rate": 4.5e-05,
"loss": 27.7504,
"step": 9
},
{
"epoch": 0.03838771593090211,
"grad_norm": 30.916200637817383,
"learning_rate": 5e-05,
"loss": 29.3754,
"step": 10
},
{
"epoch": 0.04222648752399232,
"grad_norm": 25.58551597595215,
"learning_rate": 4.9999525683718174e-05,
"loss": 29.0332,
"step": 11
},
{
"epoch": 0.046065259117082535,
"grad_norm": 30.068437576293945,
"learning_rate": 4.9998102752870765e-05,
"loss": 29.2469,
"step": 12
},
{
"epoch": 0.04990403071017274,
"grad_norm": 25.17665672302246,
"learning_rate": 4.999573126145132e-05,
"loss": 28.2097,
"step": 13
},
{
"epoch": 0.053742802303262956,
"grad_norm": 27.615007400512695,
"learning_rate": 4.99924112994468e-05,
"loss": 29.7609,
"step": 14
},
{
"epoch": 0.05758157389635317,
"grad_norm": 26.473073959350586,
"learning_rate": 4.998814299283415e-05,
"loss": 29.3846,
"step": 15
},
{
"epoch": 0.061420345489443376,
"grad_norm": 22.732168197631836,
"learning_rate": 4.998292650357558e-05,
"loss": 28.2741,
"step": 16
},
{
"epoch": 0.06525911708253358,
"grad_norm": 26.874069213867188,
"learning_rate": 4.9976762029612335e-05,
"loss": 30.2697,
"step": 17
},
{
"epoch": 0.0690978886756238,
"grad_norm": 22.940404891967773,
"learning_rate": 4.9969649804857257e-05,
"loss": 29.0416,
"step": 18
},
{
"epoch": 0.07293666026871401,
"grad_norm": 22.93003273010254,
"learning_rate": 4.996159009918585e-05,
"loss": 30.0236,
"step": 19
},
{
"epoch": 0.07677543186180422,
"grad_norm": 21.159198760986328,
"learning_rate": 4.995258321842611e-05,
"loss": 29.995,
"step": 20
},
{
"epoch": 0.08061420345489444,
"grad_norm": 23.158321380615234,
"learning_rate": 4.9942629504346827e-05,
"loss": 30.1427,
"step": 21
},
{
"epoch": 0.08445297504798464,
"grad_norm": 22.745277404785156,
"learning_rate": 4.993172933464471e-05,
"loss": 29.0171,
"step": 22
},
{
"epoch": 0.08829174664107485,
"grad_norm": 21.473155975341797,
"learning_rate": 4.9919883122929976e-05,
"loss": 29.1784,
"step": 23
},
{
"epoch": 0.09213051823416507,
"grad_norm": 21.882991790771484,
"learning_rate": 4.990709131871074e-05,
"loss": 27.7198,
"step": 24
},
{
"epoch": 0.09596928982725528,
"grad_norm": 20.5384464263916,
"learning_rate": 4.989335440737586e-05,
"loss": 29.3608,
"step": 25
},
{
"epoch": 0.09980806142034548,
"grad_norm": 21.028078079223633,
"learning_rate": 4.9878672910176615e-05,
"loss": 30.7017,
"step": 26
},
{
"epoch": 0.1036468330134357,
"grad_norm": 20.912294387817383,
"learning_rate": 4.9863047384206835e-05,
"loss": 29.2454,
"step": 27
},
{
"epoch": 0.10748560460652591,
"grad_norm": 22.990535736083984,
"learning_rate": 4.984647842238185e-05,
"loss": 30.5442,
"step": 28
},
{
"epoch": 0.11132437619961612,
"grad_norm": 20.818967819213867,
"learning_rate": 4.982896665341591e-05,
"loss": 28.0971,
"step": 29
},
{
"epoch": 0.11516314779270634,
"grad_norm": 22.41204071044922,
"learning_rate": 4.98105127417984e-05,
"loss": 30.0213,
"step": 30
},
{
"epoch": 0.11900191938579655,
"grad_norm": 20.42460823059082,
"learning_rate": 4.979111738776857e-05,
"loss": 29.0625,
"step": 31
},
{
"epoch": 0.12284069097888675,
"grad_norm": 23.12067413330078,
"learning_rate": 4.977078132728901e-05,
"loss": 30.4828,
"step": 32
},
{
"epoch": 0.12667946257197696,
"grad_norm": 19.101734161376953,
"learning_rate": 4.974950533201767e-05,
"loss": 27.8432,
"step": 33
},
{
"epoch": 0.13051823416506717,
"grad_norm": 18.871788024902344,
"learning_rate": 4.972729020927865e-05,
"loss": 28.2289,
"step": 34
},
{
"epoch": 0.1343570057581574,
"grad_norm": 19.23995590209961,
"learning_rate": 4.9704136802031485e-05,
"loss": 28.7979,
"step": 35
},
{
"epoch": 0.1381957773512476,
"grad_norm": 19.988222122192383,
"learning_rate": 4.968004598883923e-05,
"loss": 29.0405,
"step": 36
},
{
"epoch": 0.1420345489443378,
"grad_norm": 17.621200561523438,
"learning_rate": 4.965501868383506e-05,
"loss": 28.8017,
"step": 37
},
{
"epoch": 0.14587332053742802,
"grad_norm": 19.57996368408203,
"learning_rate": 4.9629055836687665e-05,
"loss": 28.5596,
"step": 38
},
{
"epoch": 0.14971209213051823,
"grad_norm": 19.62120819091797,
"learning_rate": 4.960215843256511e-05,
"loss": 28.6677,
"step": 39
},
{
"epoch": 0.15355086372360843,
"grad_norm": 18.79466438293457,
"learning_rate": 4.957432749209755e-05,
"loss": 30.2908,
"step": 40
},
{
"epoch": 0.15738963531669867,
"grad_norm": 21.75792694091797,
"learning_rate": 4.9545564071338426e-05,
"loss": 28.7169,
"step": 41
},
{
"epoch": 0.16122840690978887,
"grad_norm": 21.684478759765625,
"learning_rate": 4.9515869261724446e-05,
"loss": 28.8571,
"step": 42
},
{
"epoch": 0.16506717850287908,
"grad_norm": 20.90673065185547,
"learning_rate": 4.948524419003415e-05,
"loss": 28.5535,
"step": 43
},
{
"epoch": 0.1689059500959693,
"grad_norm": 20.5564022064209,
"learning_rate": 4.9453690018345144e-05,
"loss": 29.5709,
"step": 44
},
{
"epoch": 0.1727447216890595,
"grad_norm": 21.395124435424805,
"learning_rate": 4.942120794399002e-05,
"loss": 28.016,
"step": 45
},
{
"epoch": 0.1765834932821497,
"grad_norm": 18.09257698059082,
"learning_rate": 4.938779919951092e-05,
"loss": 29.6793,
"step": 46
},
{
"epoch": 0.18042226487523993,
"grad_norm": 19.043846130371094,
"learning_rate": 4.935346505261276e-05,
"loss": 30.1972,
"step": 47
},
{
"epoch": 0.18426103646833014,
"grad_norm": 20.1807804107666,
"learning_rate": 4.931820680611512e-05,
"loss": 28.7825,
"step": 48
},
{
"epoch": 0.18809980806142035,
"grad_norm": 18.534324645996094,
"learning_rate": 4.928202579790285e-05,
"loss": 27.895,
"step": 49
},
{
"epoch": 0.19193857965451055,
"grad_norm": 19.662391662597656,
"learning_rate": 4.9244923400875245e-05,
"loss": 30.0118,
"step": 50
},
{
"epoch": 0.19577735124760076,
"grad_norm": 19.72050666809082,
"learning_rate": 4.920690102289397e-05,
"loss": 28.6979,
"step": 51
},
{
"epoch": 0.19961612284069097,
"grad_norm": 20.052833557128906,
"learning_rate": 4.916796010672969e-05,
"loss": 29.5174,
"step": 52
},
{
"epoch": 0.2034548944337812,
"grad_norm": 18.85672378540039,
"learning_rate": 4.9128102130007225e-05,
"loss": 28.3002,
"step": 53
},
{
"epoch": 0.2072936660268714,
"grad_norm": 18.159709930419922,
"learning_rate": 4.908732860514958e-05,
"loss": 28.7818,
"step": 54
},
{
"epoch": 0.21113243761996162,
"grad_norm": 20.378021240234375,
"learning_rate": 4.9045641079320484e-05,
"loss": 29.458,
"step": 55
},
{
"epoch": 0.21497120921305182,
"grad_norm": 21.459836959838867,
"learning_rate": 4.900304113436571e-05,
"loss": 27.3894,
"step": 56
},
{
"epoch": 0.21880998080614203,
"grad_norm": 19.784387588500977,
"learning_rate": 4.895953038675306e-05,
"loss": 29.3604,
"step": 57
},
{
"epoch": 0.22264875239923224,
"grad_norm": 19.705406188964844,
"learning_rate": 4.891511048751102e-05,
"loss": 27.4372,
"step": 58
},
{
"epoch": 0.22648752399232247,
"grad_norm": 19.9382381439209,
"learning_rate": 4.886978312216611e-05,
"loss": 28.2482,
"step": 59
},
{
"epoch": 0.23032629558541268,
"grad_norm": 19.08539581298828,
"learning_rate": 4.882355001067892e-05,
"loss": 28.4456,
"step": 60
},
{
"epoch": 0.23416506717850288,
"grad_norm": 18.285186767578125,
"learning_rate": 4.877641290737884e-05,
"loss": 28.745,
"step": 61
},
{
"epoch": 0.2380038387715931,
"grad_norm": 19.981107711791992,
"learning_rate": 4.872837360089754e-05,
"loss": 27.6915,
"step": 62
},
{
"epoch": 0.2418426103646833,
"grad_norm": 18.624874114990234,
"learning_rate": 4.8679433914101006e-05,
"loss": 28.0039,
"step": 63
},
{
"epoch": 0.2456813819577735,
"grad_norm": 22.298728942871094,
"learning_rate": 4.862959570402049e-05,
"loss": 28.3383,
"step": 64
},
{
"epoch": 0.2495201535508637,
"grad_norm": 20.383586883544922,
"learning_rate": 4.857886086178194e-05,
"loss": 28.918,
"step": 65
},
{
"epoch": 0.2533589251439539,
"grad_norm": 18.7249755859375,
"learning_rate": 4.85272313125343e-05,
"loss": 29.0783,
"step": 66
},
{
"epoch": 0.2571976967370441,
"grad_norm": 18.3210391998291,
"learning_rate": 4.8474709015376416e-05,
"loss": 27.3239,
"step": 67
},
{
"epoch": 0.26103646833013433,
"grad_norm": 19.459135055541992,
"learning_rate": 4.842129596328277e-05,
"loss": 29.2664,
"step": 68
},
{
"epoch": 0.2648752399232246,
"grad_norm": 18.80237579345703,
"learning_rate": 4.836699418302777e-05,
"loss": 28.8449,
"step": 69
},
{
"epoch": 0.2687140115163148,
"grad_norm": 18.05824851989746,
"learning_rate": 4.8311805735108894e-05,
"loss": 27.5182,
"step": 70
},
{
"epoch": 0.272552783109405,
"grad_norm": 18.784578323364258,
"learning_rate": 4.825573271366851e-05,
"loss": 28.4831,
"step": 71
},
{
"epoch": 0.2763915547024952,
"grad_norm": 18.89769172668457,
"learning_rate": 4.8198777246414373e-05,
"loss": 27.7881,
"step": 72
},
{
"epoch": 0.2802303262955854,
"grad_norm": 17.64800453186035,
"learning_rate": 4.814094149453891e-05,
"loss": 27.7358,
"step": 73
},
{
"epoch": 0.2840690978886756,
"grad_norm": 19.012086868286133,
"learning_rate": 4.808222765263724e-05,
"loss": 29.8711,
"step": 74
},
{
"epoch": 0.28790786948176583,
"grad_norm": 18.49270248413086,
"learning_rate": 4.802263794862385e-05,
"loss": 28.3761,
"step": 75
},
{
"epoch": 0.29174664107485604,
"grad_norm": 21.752199172973633,
"learning_rate": 4.796217464364808e-05,
"loss": 28.3796,
"step": 76
},
{
"epoch": 0.29558541266794625,
"grad_norm": 19.340524673461914,
"learning_rate": 4.790084003200834e-05,
"loss": 29.4554,
"step": 77
},
{
"epoch": 0.29942418426103645,
"grad_norm": 21.60015869140625,
"learning_rate": 4.783863644106502e-05,
"loss": 28.0435,
"step": 78
},
{
"epoch": 0.30326295585412666,
"grad_norm": 20.29178237915039,
"learning_rate": 4.777556623115221e-05,
"loss": 27.563,
"step": 79
},
{
"epoch": 0.30710172744721687,
"grad_norm": 18.13334846496582,
"learning_rate": 4.7711631795488096e-05,
"loss": 27.4931,
"step": 80
},
{
"epoch": 0.31094049904030713,
"grad_norm": 18.14622688293457,
"learning_rate": 4.764683556008418e-05,
"loss": 28.0506,
"step": 81
},
{
"epoch": 0.31477927063339733,
"grad_norm": 19.75128936767578,
"learning_rate": 4.758117998365322e-05,
"loss": 27.2963,
"step": 82
},
{
"epoch": 0.31861804222648754,
"grad_norm": 17.486196517944336,
"learning_rate": 4.751466755751594e-05,
"loss": 28.7996,
"step": 83
},
{
"epoch": 0.32245681381957775,
"grad_norm": 19.39826011657715,
"learning_rate": 4.7447300805506455e-05,
"loss": 27.6073,
"step": 84
},
{
"epoch": 0.32629558541266795,
"grad_norm": 17.352880477905273,
"learning_rate": 4.7379082283876566e-05,
"loss": 28.5087,
"step": 85
},
{
"epoch": 0.33013435700575816,
"grad_norm": 20.071483612060547,
"learning_rate": 4.7310014581198695e-05,
"loss": 27.1697,
"step": 86
},
{
"epoch": 0.33397312859884837,
"grad_norm": 19.04390525817871,
"learning_rate": 4.7240100318267746e-05,
"loss": 27.9563,
"step": 87
},
{
"epoch": 0.3378119001919386,
"grad_norm": 20.634225845336914,
"learning_rate": 4.716934214800155e-05,
"loss": 27.6816,
"step": 88
},
{
"epoch": 0.3416506717850288,
"grad_norm": 19.746788024902344,
"learning_rate": 4.70977427553403e-05,
"loss": 27.3921,
"step": 89
},
{
"epoch": 0.345489443378119,
"grad_norm": 19.04279899597168,
"learning_rate": 4.702530485714461e-05,
"loss": 27.2845,
"step": 90
},
{
"epoch": 0.3493282149712092,
"grad_norm": 18.288244247436523,
"learning_rate": 4.695203120209245e-05,
"loss": 28.8947,
"step": 91
},
{
"epoch": 0.3531669865642994,
"grad_norm": 18.553823471069336,
"learning_rate": 4.6877924570574817e-05,
"loss": 27.7966,
"step": 92
},
{
"epoch": 0.3570057581573896,
"grad_norm": 18.615053176879883,
"learning_rate": 4.6802987774590275e-05,
"loss": 27.8787,
"step": 93
},
{
"epoch": 0.36084452975047987,
"grad_norm": 18.30453109741211,
"learning_rate": 4.672722365763821e-05,
"loss": 29.0792,
"step": 94
},
{
"epoch": 0.3646833013435701,
"grad_norm": 18.134958267211914,
"learning_rate": 4.665063509461097e-05,
"loss": 29.07,
"step": 95
},
{
"epoch": 0.3685220729366603,
"grad_norm": 19.019006729125977,
"learning_rate": 4.6573224991684744e-05,
"loss": 27.708,
"step": 96
},
{
"epoch": 0.3723608445297505,
"grad_norm": 18.863723754882812,
"learning_rate": 4.649499628620931e-05,
"loss": 27.4551,
"step": 97
},
{
"epoch": 0.3761996161228407,
"grad_norm": 18.62997055053711,
"learning_rate": 4.641595194659657e-05,
"loss": 26.7616,
"step": 98
},
{
"epoch": 0.3800383877159309,
"grad_norm": 18.19691276550293,
"learning_rate": 4.63360949722079e-05,
"loss": 27.6041,
"step": 99
},
{
"epoch": 0.3838771593090211,
"grad_norm": 19.37999153137207,
"learning_rate": 4.625542839324036e-05,
"loss": 26.9613,
"step": 100
},
{
"epoch": 0.3877159309021113,
"grad_norm": 18.335630416870117,
"learning_rate": 4.617395527061168e-05,
"loss": 27.4261,
"step": 101
},
{
"epoch": 0.3915547024952015,
"grad_norm": 19.16864585876465,
"learning_rate": 4.6091678695844164e-05,
"loss": 28.8591,
"step": 102
},
{
"epoch": 0.39539347408829173,
"grad_norm": 20.256633758544922,
"learning_rate": 4.600860179094732e-05,
"loss": 26.8054,
"step": 103
},
{
"epoch": 0.39923224568138194,
"grad_norm": 18.64192008972168,
"learning_rate": 4.5924727708299445e-05,
"loss": 27.3249,
"step": 104
},
{
"epoch": 0.40307101727447214,
"grad_norm": 17.976713180541992,
"learning_rate": 4.584005963052799e-05,
"loss": 29.1977,
"step": 105
},
{
"epoch": 0.4069097888675624,
"grad_norm": 17.99713706970215,
"learning_rate": 4.575460077038877e-05,
"loss": 29.0127,
"step": 106
},
{
"epoch": 0.4107485604606526,
"grad_norm": 18.080854415893555,
"learning_rate": 4.5668354370644086e-05,
"loss": 29.716,
"step": 107
},
{
"epoch": 0.4145873320537428,
"grad_norm": 17.458669662475586,
"learning_rate": 4.558132370393968e-05,
"loss": 29.7465,
"step": 108
},
{
"epoch": 0.418426103646833,
"grad_norm": 18.444486618041992,
"learning_rate": 4.5493512072680536e-05,
"loss": 30.3556,
"step": 109
},
{
"epoch": 0.42226487523992323,
"grad_norm": 17.931936264038086,
"learning_rate": 4.540492280890555e-05,
"loss": 27.0296,
"step": 110
},
{
"epoch": 0.42610364683301344,
"grad_norm": 17.657825469970703,
"learning_rate": 4.5315559274161144e-05,
"loss": 28.3838,
"step": 111
},
{
"epoch": 0.42994241842610365,
"grad_norm": 17.2442626953125,
"learning_rate": 4.522542485937369e-05,
"loss": 26.963,
"step": 112
},
{
"epoch": 0.43378119001919385,
"grad_norm": 16.4385929107666,
"learning_rate": 4.5134522984720816e-05,
"loss": 28.0953,
"step": 113
},
{
"epoch": 0.43761996161228406,
"grad_norm": 16.927474975585938,
"learning_rate": 4.504285709950167e-05,
"loss": 29.0116,
"step": 114
},
{
"epoch": 0.44145873320537427,
"grad_norm": 18.508386611938477,
"learning_rate": 4.4950430682006e-05,
"loss": 27.5212,
"step": 115
},
{
"epoch": 0.44529750479846447,
"grad_norm": 17.00088882446289,
"learning_rate": 4.485724723938215e-05,
"loss": 27.2209,
"step": 116
},
{
"epoch": 0.4491362763915547,
"grad_norm": 17.35956573486328,
"learning_rate": 4.476331030750408e-05,
"loss": 28.9031,
"step": 117
},
{
"epoch": 0.45297504798464494,
"grad_norm": 17.835237503051758,
"learning_rate": 4.4668623450837085e-05,
"loss": 25.8466,
"step": 118
},
{
"epoch": 0.45681381957773515,
"grad_norm": 18.052587509155273,
"learning_rate": 4.457319026230257e-05,
"loss": 27.7288,
"step": 119
},
{
"epoch": 0.46065259117082535,
"grad_norm": 21.071945190429688,
"learning_rate": 4.447701436314176e-05,
"loss": 27.2769,
"step": 120
},
{
"epoch": 0.46449136276391556,
"grad_norm": 17.07927703857422,
"learning_rate": 4.4380099402778244e-05,
"loss": 27.3407,
"step": 121
},
{
"epoch": 0.46833013435700577,
"grad_norm": 17.430988311767578,
"learning_rate": 4.428244905867952e-05,
"loss": 27.7631,
"step": 122
},
{
"epoch": 0.472168905950096,
"grad_norm": 18.140281677246094,
"learning_rate": 4.418406703621743e-05,
"loss": 27.4634,
"step": 123
},
{
"epoch": 0.4760076775431862,
"grad_norm": 18.464765548706055,
"learning_rate": 4.408495706852758e-05,
"loss": 27.1883,
"step": 124
},
{
"epoch": 0.4798464491362764,
"grad_norm": 17.243370056152344,
"learning_rate": 4.398512291636768e-05,
"loss": 29.4247,
"step": 125
},
{
"epoch": 0.4836852207293666,
"grad_norm": 18.10504722595215,
"learning_rate": 4.3884568367974845e-05,
"loss": 27.3026,
"step": 126
},
{
"epoch": 0.4875239923224568,
"grad_norm": 18.39180564880371,
"learning_rate": 4.378329723892184e-05,
"loss": 28.0279,
"step": 127
},
{
"epoch": 0.491362763915547,
"grad_norm": 16.66558265686035,
"learning_rate": 4.3681313371972276e-05,
"loss": 27.5308,
"step": 128
},
{
"epoch": 0.4952015355086372,
"grad_norm": 16.756423950195312,
"learning_rate": 4.357862063693486e-05,
"loss": 28.4946,
"step": 129
},
{
"epoch": 0.4990403071017274,
"grad_norm": 17.056476593017578,
"learning_rate": 4.347522293051648e-05,
"loss": 27.8562,
"step": 130
},
{
"epoch": 0.4990403071017274,
"eval_loss": 1.7600558996200562,
"eval_runtime": 31.0188,
"eval_samples_per_second": 9.543,
"eval_steps_per_second": 4.771,
"step": 130
},
{
"epoch": 0.5028790786948176,
"grad_norm": 17.216785430908203,
"learning_rate": 4.337112417617439e-05,
"loss": 27.1427,
"step": 131
},
{
"epoch": 0.5067178502879078,
"grad_norm": 16.08450698852539,
"learning_rate": 4.3266328323967333e-05,
"loss": 27.4919,
"step": 132
},
{
"epoch": 0.510556621880998,
"grad_norm": 17.361270904541016,
"learning_rate": 4.3160839350405606e-05,
"loss": 26.1541,
"step": 133
},
{
"epoch": 0.5143953934740882,
"grad_norm": 17.303585052490234,
"learning_rate": 4.305466125830023e-05,
"loss": 26.0347,
"step": 134
},
{
"epoch": 0.5182341650671785,
"grad_norm": 17.440980911254883,
"learning_rate": 4.294779807661105e-05,
"loss": 27.2826,
"step": 135
},
{
"epoch": 0.5220729366602687,
"grad_norm": 19.435768127441406,
"learning_rate": 4.284025386029381e-05,
"loss": 27.5317,
"step": 136
},
{
"epoch": 0.525911708253359,
"grad_norm": 17.796663284301758,
"learning_rate": 4.273203269014634e-05,
"loss": 27.8947,
"step": 137
},
{
"epoch": 0.5297504798464492,
"grad_norm": 17.049524307250977,
"learning_rate": 4.2623138672653684e-05,
"loss": 29.3983,
"step": 138
},
{
"epoch": 0.5335892514395394,
"grad_norm": 17.124839782714844,
"learning_rate": 4.2513575939832275e-05,
"loss": 28.4112,
"step": 139
},
{
"epoch": 0.5374280230326296,
"grad_norm": 17.724750518798828,
"learning_rate": 4.2403348649073174e-05,
"loss": 29.1517,
"step": 140
},
{
"epoch": 0.5412667946257198,
"grad_norm": 17.679073333740234,
"learning_rate": 4.2292460982984255e-05,
"loss": 27.1786,
"step": 141
},
{
"epoch": 0.54510556621881,
"grad_norm": 18.939489364624023,
"learning_rate": 4.218091714923157e-05,
"loss": 28.6373,
"step": 142
},
{
"epoch": 0.5489443378119002,
"grad_norm": 17.122732162475586,
"learning_rate": 4.206872138037964e-05,
"loss": 27.252,
"step": 143
},
{
"epoch": 0.5527831094049904,
"grad_norm": 18.2526912689209,
"learning_rate": 4.1955877933730855e-05,
"loss": 29.9347,
"step": 144
},
{
"epoch": 0.5566218809980806,
"grad_norm": 18.54884147644043,
"learning_rate": 4.184239109116393e-05,
"loss": 27.988,
"step": 145
},
{
"epoch": 0.5604606525911708,
"grad_norm": 18.026559829711914,
"learning_rate": 4.172826515897146e-05,
"loss": 28.1473,
"step": 146
},
{
"epoch": 0.564299424184261,
"grad_norm": 17.47412109375,
"learning_rate": 4.161350446769645e-05,
"loss": 26.7597,
"step": 147
},
{
"epoch": 0.5681381957773513,
"grad_norm": 19.019777297973633,
"learning_rate": 4.149811337196807e-05,
"loss": 27.5097,
"step": 148
},
{
"epoch": 0.5719769673704415,
"grad_norm": 18.873889923095703,
"learning_rate": 4.1382096250336346e-05,
"loss": 26.3739,
"step": 149
},
{
"epoch": 0.5758157389635317,
"grad_norm": 17.524677276611328,
"learning_rate": 4.126545750510605e-05,
"loss": 26.7282,
"step": 150
},
{
"epoch": 0.5796545105566219,
"grad_norm": 17.283926010131836,
"learning_rate": 4.1148201562169685e-05,
"loss": 26.9749,
"step": 151
},
{
"epoch": 0.5834932821497121,
"grad_norm": 17.261751174926758,
"learning_rate": 4.103033287083946e-05,
"loss": 27.2926,
"step": 152
},
{
"epoch": 0.5873320537428023,
"grad_norm": 16.952129364013672,
"learning_rate": 4.0911855903678534e-05,
"loss": 29.1231,
"step": 153
},
{
"epoch": 0.5911708253358925,
"grad_norm": 16.523609161376953,
"learning_rate": 4.079277515633127e-05,
"loss": 27.9168,
"step": 154
},
{
"epoch": 0.5950095969289827,
"grad_norm": 17.648927688598633,
"learning_rate": 4.067309514735267e-05,
"loss": 26.9956,
"step": 155
},
{
"epoch": 0.5988483685220729,
"grad_norm": 18.750659942626953,
"learning_rate": 4.055282041803685e-05,
"loss": 27.5584,
"step": 156
},
{
"epoch": 0.6026871401151631,
"grad_norm": 17.183300018310547,
"learning_rate": 4.0431955532244827e-05,
"loss": 27.9265,
"step": 157
},
{
"epoch": 0.6065259117082533,
"grad_norm": 17.270681381225586,
"learning_rate": 4.031050507623125e-05,
"loss": 28.9735,
"step": 158
},
{
"epoch": 0.6103646833013435,
"grad_norm": 19.801315307617188,
"learning_rate": 4.0188473658470426e-05,
"loss": 26.7811,
"step": 159
},
{
"epoch": 0.6142034548944337,
"grad_norm": 16.823190689086914,
"learning_rate": 4.0065865909481417e-05,
"loss": 26.9071,
"step": 160
},
{
"epoch": 0.6180422264875239,
"grad_norm": 17.074491500854492,
"learning_rate": 3.9942686481652344e-05,
"loss": 27.9241,
"step": 161
},
{
"epoch": 0.6218809980806143,
"grad_norm": 18.242040634155273,
"learning_rate": 3.981894004906388e-05,
"loss": 28.2823,
"step": 162
},
{
"epoch": 0.6257197696737045,
"grad_norm": 16.200469970703125,
"learning_rate": 3.969463130731183e-05,
"loss": 27.2651,
"step": 163
},
{
"epoch": 0.6295585412667947,
"grad_norm": 17.807695388793945,
"learning_rate": 3.9569764973329026e-05,
"loss": 28.572,
"step": 164
},
{
"epoch": 0.6333973128598849,
"grad_norm": 16.356901168823242,
"learning_rate": 3.9444345785206285e-05,
"loss": 27.4942,
"step": 165
},
{
"epoch": 0.6372360844529751,
"grad_norm": 18.459415435791016,
"learning_rate": 3.931837850201263e-05,
"loss": 27.4934,
"step": 166
},
{
"epoch": 0.6410748560460653,
"grad_norm": 18.079626083374023,
"learning_rate": 3.919186790361475e-05,
"loss": 28.0261,
"step": 167
},
{
"epoch": 0.6449136276391555,
"grad_norm": 18.058738708496094,
"learning_rate": 3.906481879049558e-05,
"loss": 27.2611,
"step": 168
},
{
"epoch": 0.6487523992322457,
"grad_norm": 17.553970336914062,
"learning_rate": 3.893723598357214e-05,
"loss": 27.5768,
"step": 169
},
{
"epoch": 0.6525911708253359,
"grad_norm": 18.69593620300293,
"learning_rate": 3.880912432401265e-05,
"loss": 27.5784,
"step": 170
},
{
"epoch": 0.6564299424184261,
"grad_norm": 17.491533279418945,
"learning_rate": 3.8680488673052784e-05,
"loss": 29.1586,
"step": 171
},
{
"epoch": 0.6602687140115163,
"grad_norm": 18.12783432006836,
"learning_rate": 3.855133391181124e-05,
"loss": 27.3099,
"step": 172
},
{
"epoch": 0.6641074856046065,
"grad_norm": 18.322479248046875,
"learning_rate": 3.842166494110451e-05,
"loss": 29.5131,
"step": 173
},
{
"epoch": 0.6679462571976967,
"grad_norm": 17.041065216064453,
"learning_rate": 3.82914866812609e-05,
"loss": 28.5525,
"step": 174
},
{
"epoch": 0.6717850287907869,
"grad_norm": 17.609098434448242,
"learning_rate": 3.81608040719339e-05,
"loss": 28.8543,
"step": 175
},
{
"epoch": 0.6756238003838771,
"grad_norm": 17.111724853515625,
"learning_rate": 3.802962207191463e-05,
"loss": 25.6639,
"step": 176
},
{
"epoch": 0.6794625719769674,
"grad_norm": 17.328916549682617,
"learning_rate": 3.789794565894378e-05,
"loss": 26.0536,
"step": 177
},
{
"epoch": 0.6833013435700576,
"grad_norm": 17.837627410888672,
"learning_rate": 3.7765779829522675e-05,
"loss": 27.5278,
"step": 178
},
{
"epoch": 0.6871401151631478,
"grad_norm": 15.394762992858887,
"learning_rate": 3.7633129598723704e-05,
"loss": 27.7244,
"step": 179
},
{
"epoch": 0.690978886756238,
"grad_norm": 19.647441864013672,
"learning_rate": 3.7500000000000003e-05,
"loss": 25.729,
"step": 180
},
{
"epoch": 0.6948176583493282,
"grad_norm": 17.077699661254883,
"learning_rate": 3.7366396084994475e-05,
"loss": 26.2709,
"step": 181
},
{
"epoch": 0.6986564299424184,
"grad_norm": 16.858604431152344,
"learning_rate": 3.723232292334809e-05,
"loss": 27.8621,
"step": 182
},
{
"epoch": 0.7024952015355086,
"grad_norm": 17.455453872680664,
"learning_rate": 3.709778560250754e-05,
"loss": 27.0243,
"step": 183
},
{
"epoch": 0.7063339731285988,
"grad_norm": 16.5456600189209,
"learning_rate": 3.696278922753216e-05,
"loss": 27.6667,
"step": 184
},
{
"epoch": 0.710172744721689,
"grad_norm": 17.52065086364746,
"learning_rate": 3.6827338920900254e-05,
"loss": 27.4649,
"step": 185
},
{
"epoch": 0.7140115163147792,
"grad_norm": 15.55224323272705,
"learning_rate": 3.669143982231467e-05,
"loss": 26.4226,
"step": 186
},
{
"epoch": 0.7178502879078695,
"grad_norm": 16.189666748046875,
"learning_rate": 3.655509708850783e-05,
"loss": 26.3024,
"step": 187
},
{
"epoch": 0.7216890595009597,
"grad_norm": 17.70796775817871,
"learning_rate": 3.641831589304602e-05,
"loss": 27.1759,
"step": 188
},
{
"epoch": 0.72552783109405,
"grad_norm": 16.10495376586914,
"learning_rate": 3.628110142613308e-05,
"loss": 26.5754,
"step": 189
},
{
"epoch": 0.7293666026871402,
"grad_norm": 17.34556007385254,
"learning_rate": 3.6143458894413465e-05,
"loss": 27.8653,
"step": 190
},
{
"epoch": 0.7332053742802304,
"grad_norm": 18.321557998657227,
"learning_rate": 3.600539352077469e-05,
"loss": 27.7326,
"step": 191
},
{
"epoch": 0.7370441458733206,
"grad_norm": 15.874064445495605,
"learning_rate": 3.586691054414913e-05,
"loss": 26.683,
"step": 192
},
{
"epoch": 0.7408829174664108,
"grad_norm": 16.701417922973633,
"learning_rate": 3.572801521931522e-05,
"loss": 25.815,
"step": 193
},
{
"epoch": 0.744721689059501,
"grad_norm": 17.957345962524414,
"learning_rate": 3.558871281669811e-05,
"loss": 26.726,
"step": 194
},
{
"epoch": 0.7485604606525912,
"grad_norm": 17.00701141357422,
"learning_rate": 3.544900862216959e-05,
"loss": 26.8739,
"step": 195
},
{
"epoch": 0.7523992322456814,
"grad_norm": 19.22977638244629,
"learning_rate": 3.5308907936847594e-05,
"loss": 26.1091,
"step": 196
},
{
"epoch": 0.7562380038387716,
"grad_norm": 18.498205184936523,
"learning_rate": 3.516841607689501e-05,
"loss": 25.4755,
"step": 197
},
{
"epoch": 0.7600767754318618,
"grad_norm": 18.955081939697266,
"learning_rate": 3.502753837331797e-05,
"loss": 26.3684,
"step": 198
},
{
"epoch": 0.763915547024952,
"grad_norm": 18.34401512145996,
"learning_rate": 3.488628017176356e-05,
"loss": 28.767,
"step": 199
},
{
"epoch": 0.7677543186180422,
"grad_norm": 18.423044204711914,
"learning_rate": 3.474464683231698e-05,
"loss": 27.2117,
"step": 200
},
{
"epoch": 0.7715930902111324,
"grad_norm": 18.909847259521484,
"learning_rate": 3.460264372929815e-05,
"loss": 26.408,
"step": 201
},
{
"epoch": 0.7754318618042226,
"grad_norm": 16.31786346435547,
"learning_rate": 3.446027625105776e-05,
"loss": 26.7065,
"step": 202
},
{
"epoch": 0.7792706333973128,
"grad_norm": 18.030500411987305,
"learning_rate": 3.431754979977285e-05,
"loss": 27.028,
"step": 203
},
{
"epoch": 0.783109404990403,
"grad_norm": 16.70123863220215,
"learning_rate": 3.4174469791241806e-05,
"loss": 27.5544,
"step": 204
},
{
"epoch": 0.7869481765834933,
"grad_norm": 16.9814453125,
"learning_rate": 3.403104165467883e-05,
"loss": 27.5372,
"step": 205
},
{
"epoch": 0.7907869481765835,
"grad_norm": 17.28445053100586,
"learning_rate": 3.388727083250795e-05,
"loss": 27.4625,
"step": 206
},
{
"epoch": 0.7946257197696737,
"grad_norm": 17.412431716918945,
"learning_rate": 3.374316278015653e-05,
"loss": 27.897,
"step": 207
},
{
"epoch": 0.7984644913627639,
"grad_norm": 16.263408660888672,
"learning_rate": 3.3598722965848204e-05,
"loss": 27.2817,
"step": 208
},
{
"epoch": 0.8023032629558541,
"grad_norm": 16.9997615814209,
"learning_rate": 3.345395687039543e-05,
"loss": 26.4631,
"step": 209
},
{
"epoch": 0.8061420345489443,
"grad_norm": 17.816619873046875,
"learning_rate": 3.330886998699149e-05,
"loss": 25.6059,
"step": 210
},
{
"epoch": 0.8099808061420346,
"grad_norm": 17.235605239868164,
"learning_rate": 3.316346782100208e-05,
"loss": 27.4531,
"step": 211
},
{
"epoch": 0.8138195777351248,
"grad_norm": 15.711729049682617,
"learning_rate": 3.301775588975638e-05,
"loss": 27.0936,
"step": 212
},
{
"epoch": 0.817658349328215,
"grad_norm": 16.274784088134766,
"learning_rate": 3.28717397223377e-05,
"loss": 25.6661,
"step": 213
},
{
"epoch": 0.8214971209213052,
"grad_norm": 15.97848129272461,
"learning_rate": 3.272542485937369e-05,
"loss": 26.772,
"step": 214
},
{
"epoch": 0.8253358925143954,
"grad_norm": 16.191444396972656,
"learning_rate": 3.257881685282609e-05,
"loss": 25.8711,
"step": 215
},
{
"epoch": 0.8291746641074856,
"grad_norm": 16.814664840698242,
"learning_rate": 3.243192126578007e-05,
"loss": 25.5683,
"step": 216
},
{
"epoch": 0.8330134357005758,
"grad_norm": 15.929566383361816,
"learning_rate": 3.228474367223312e-05,
"loss": 27.5315,
"step": 217
},
{
"epoch": 0.836852207293666,
"grad_norm": 16.453189849853516,
"learning_rate": 3.2137289656883556e-05,
"loss": 27.1406,
"step": 218
},
{
"epoch": 0.8406909788867563,
"grad_norm": 15.205329895019531,
"learning_rate": 3.19895648149186e-05,
"loss": 26.2816,
"step": 219
},
{
"epoch": 0.8445297504798465,
"grad_norm": 16.378446578979492,
"learning_rate": 3.1841574751802076e-05,
"loss": 26.6438,
"step": 220
},
{
"epoch": 0.8483685220729367,
"grad_norm": 16.06751251220703,
"learning_rate": 3.16933250830617e-05,
"loss": 27.0177,
"step": 221
},
{
"epoch": 0.8522072936660269,
"grad_norm": 15.032289505004883,
"learning_rate": 3.1544821434076014e-05,
"loss": 27.5912,
"step": 222
},
{
"epoch": 0.8560460652591171,
"grad_norm": 16.624408721923828,
"learning_rate": 3.1396069439860894e-05,
"loss": 28.211,
"step": 223
},
{
"epoch": 0.8598848368522073,
"grad_norm": 17.585723876953125,
"learning_rate": 3.124707474485577e-05,
"loss": 26.8424,
"step": 224
},
{
"epoch": 0.8637236084452975,
"grad_norm": 15.791353225708008,
"learning_rate": 3.109784300270943e-05,
"loss": 26.7544,
"step": 225
},
{
"epoch": 0.8675623800383877,
"grad_norm": 16.265228271484375,
"learning_rate": 3.094837987606547e-05,
"loss": 26.9071,
"step": 226
},
{
"epoch": 0.8714011516314779,
"grad_norm": 15.219544410705566,
"learning_rate": 3.0798691036347453e-05,
"loss": 25.127,
"step": 227
},
{
"epoch": 0.8752399232245681,
"grad_norm": 16.0206241607666,
"learning_rate": 3.064878216354369e-05,
"loss": 27.3355,
"step": 228
},
{
"epoch": 0.8790786948176583,
"grad_norm": 15.717583656311035,
"learning_rate": 3.049865894599172e-05,
"loss": 27.7224,
"step": 229
},
{
"epoch": 0.8829174664107485,
"grad_norm": 16.433361053466797,
"learning_rate": 3.0348327080162435e-05,
"loss": 27.4405,
"step": 230
},
{
"epoch": 0.8867562380038387,
"grad_norm": 16.457277297973633,
"learning_rate": 3.0197792270443982e-05,
"loss": 26.5396,
"step": 231
},
{
"epoch": 0.8905950095969289,
"grad_norm": 15.61976146697998,
"learning_rate": 3.0047060228925256e-05,
"loss": 27.3457,
"step": 232
},
{
"epoch": 0.8944337811900192,
"grad_norm": 16.303308486938477,
"learning_rate": 2.9896136675179176e-05,
"loss": 26.2405,
"step": 233
},
{
"epoch": 0.8982725527831094,
"grad_norm": 15.62144660949707,
"learning_rate": 2.974502733604565e-05,
"loss": 26.1536,
"step": 234
},
{
"epoch": 0.9021113243761996,
"grad_norm": 16.30591583251953,
"learning_rate": 2.9593737945414264e-05,
"loss": 26.1949,
"step": 235
},
{
"epoch": 0.9059500959692899,
"grad_norm": 16.510095596313477,
"learning_rate": 2.9442274244006722e-05,
"loss": 26.6893,
"step": 236
},
{
"epoch": 0.9097888675623801,
"grad_norm": 16.378929138183594,
"learning_rate": 2.9290641979158993e-05,
"loss": 28.0321,
"step": 237
},
{
"epoch": 0.9136276391554703,
"grad_norm": 15.976604461669922,
"learning_rate": 2.913884690460325e-05,
"loss": 27.17,
"step": 238
},
{
"epoch": 0.9174664107485605,
"grad_norm": 16.334959030151367,
"learning_rate": 2.8986894780249524e-05,
"loss": 26.3522,
"step": 239
},
{
"epoch": 0.9213051823416507,
"grad_norm": 16.02887725830078,
"learning_rate": 2.8834791371967142e-05,
"loss": 28.4084,
"step": 240
},
{
"epoch": 0.9251439539347409,
"grad_norm": 17.058687210083008,
"learning_rate": 2.868254245136594e-05,
"loss": 26.801,
"step": 241
},
{
"epoch": 0.9289827255278311,
"grad_norm": 15.894112586975098,
"learning_rate": 2.8530153795577286e-05,
"loss": 25.7278,
"step": 242
},
{
"epoch": 0.9328214971209213,
"grad_norm": 16.269180297851562,
"learning_rate": 2.8377631187034825e-05,
"loss": 26.3214,
"step": 243
},
{
"epoch": 0.9366602687140115,
"grad_norm": 15.76980209350586,
"learning_rate": 2.8224980413255086e-05,
"loss": 27.0605,
"step": 244
},
{
"epoch": 0.9404990403071017,
"grad_norm": 15.396162986755371,
"learning_rate": 2.8072207266617855e-05,
"loss": 26.77,
"step": 245
},
{
"epoch": 0.944337811900192,
"grad_norm": 16.089946746826172,
"learning_rate": 2.7919317544146405e-05,
"loss": 26.3307,
"step": 246
},
{
"epoch": 0.9481765834932822,
"grad_norm": 15.087519645690918,
"learning_rate": 2.776631704728752e-05,
"loss": 25.4705,
"step": 247
},
{
"epoch": 0.9520153550863724,
"grad_norm": 16.0484619140625,
"learning_rate": 2.761321158169134e-05,
"loss": 25.8905,
"step": 248
},
{
"epoch": 0.9558541266794626,
"grad_norm": 16.0047607421875,
"learning_rate": 2.746000695699107e-05,
"loss": 26.0246,
"step": 249
},
{
"epoch": 0.9596928982725528,
"grad_norm": 16.781606674194336,
"learning_rate": 2.7306708986582553e-05,
"loss": 25.6385,
"step": 250
},
{
"epoch": 0.963531669865643,
"grad_norm": 15.778640747070312,
"learning_rate": 2.7153323487403655e-05,
"loss": 26.7195,
"step": 251
},
{
"epoch": 0.9673704414587332,
"grad_norm": 17.09006118774414,
"learning_rate": 2.6999856279713544e-05,
"loss": 25.475,
"step": 252
},
{
"epoch": 0.9712092130518234,
"grad_norm": 16.174684524536133,
"learning_rate": 2.6846313186871853e-05,
"loss": 24.9302,
"step": 253
},
{
"epoch": 0.9750479846449136,
"grad_norm": 15.83430290222168,
"learning_rate": 2.669270003511769e-05,
"loss": 25.4569,
"step": 254
},
{
"epoch": 0.9788867562380038,
"grad_norm": 16.559917449951172,
"learning_rate": 2.653902265334858e-05,
"loss": 26.9698,
"step": 255
},
{
"epoch": 0.982725527831094,
"grad_norm": 15.691744804382324,
"learning_rate": 2.638528687289925e-05,
"loss": 25.7697,
"step": 256
},
{
"epoch": 0.9865642994241842,
"grad_norm": 15.908937454223633,
"learning_rate": 2.6231498527320425e-05,
"loss": 26.0816,
"step": 257
},
{
"epoch": 0.9904030710172744,
"grad_norm": 15.256813049316406,
"learning_rate": 2.6077663452157397e-05,
"loss": 27.2372,
"step": 258
},
{
"epoch": 0.9942418426103646,
"grad_norm": 16.329204559326172,
"learning_rate": 2.592378748472863e-05,
"loss": 26.6932,
"step": 259
},
{
"epoch": 0.9980806142034548,
"grad_norm": 15.60101318359375,
"learning_rate": 2.5769876463904265e-05,
"loss": 26.632,
"step": 260
},
{
"epoch": 0.9980806142034548,
"eval_loss": 1.698987364768982,
"eval_runtime": 31.0703,
"eval_samples_per_second": 9.527,
"eval_steps_per_second": 4.763,
"step": 260
}
],
"logging_steps": 1,
"max_steps": 520,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 260,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 2.4952457875972424e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}