adaption_crop_disease_reasoning / trainer_state.json
WhyParabola's picture
Add 11 files
fcd2dbb verified
Raw
History Blame Contribute Delete
49 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 52,
"global_step": 264,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.007575757575757576,
"grad_norm": 0.33429259061813354,
"learning_rate": 0.0,
"loss": 1.231689453125,
"step": 1
},
{
"epoch": 0.015151515151515152,
"grad_norm": 0.5006925463676453,
"learning_rate": 7.142857142857143e-06,
"loss": 1.685546875,
"step": 2
},
{
"epoch": 0.022727272727272728,
"grad_norm": 0.4349065124988556,
"learning_rate": 1.4285714285714285e-05,
"loss": 1.5908203125,
"step": 3
},
{
"epoch": 0.030303030303030304,
"grad_norm": 0.4520890414714813,
"learning_rate": 2.1428571428571428e-05,
"loss": 1.603515625,
"step": 4
},
{
"epoch": 0.03787878787878788,
"grad_norm": 0.4511120021343231,
"learning_rate": 2.857142857142857e-05,
"loss": 1.583984375,
"step": 5
},
{
"epoch": 0.045454545454545456,
"grad_norm": 0.45080456137657166,
"learning_rate": 3.571428571428572e-05,
"loss": 1.48828125,
"step": 6
},
{
"epoch": 0.05303030303030303,
"grad_norm": 0.5375288128852844,
"learning_rate": 4.2857142857142856e-05,
"loss": 1.6015625,
"step": 7
},
{
"epoch": 0.06060606060606061,
"grad_norm": 0.5009720325469971,
"learning_rate": 5e-05,
"loss": 1.51513671875,
"step": 8
},
{
"epoch": 0.06818181818181818,
"grad_norm": 0.47150489687919617,
"learning_rate": 5.714285714285714e-05,
"loss": 1.48046875,
"step": 9
},
{
"epoch": 0.07575757575757576,
"grad_norm": 0.42623651027679443,
"learning_rate": 6.428571428571429e-05,
"loss": 1.32366943359375,
"step": 10
},
{
"epoch": 0.08333333333333333,
"grad_norm": 0.4326189458370209,
"learning_rate": 7.142857142857143e-05,
"loss": 1.34423828125,
"step": 11
},
{
"epoch": 0.09090909090909091,
"grad_norm": 0.36637046933174133,
"learning_rate": 7.857142857142858e-05,
"loss": 1.233642578125,
"step": 12
},
{
"epoch": 0.09848484848484848,
"grad_norm": 0.331664502620697,
"learning_rate": 8.571428571428571e-05,
"loss": 1.287109375,
"step": 13
},
{
"epoch": 0.10606060606060606,
"grad_norm": 0.3909582197666168,
"learning_rate": 9.285714285714286e-05,
"loss": 1.4072265625,
"step": 14
},
{
"epoch": 0.11363636363636363,
"grad_norm": 0.28124600648880005,
"learning_rate": 0.0001,
"loss": 1.3623046875,
"step": 15
},
{
"epoch": 0.12121212121212122,
"grad_norm": 0.24927006661891937,
"learning_rate": 9.999644698917173e-05,
"loss": 1.35693359375,
"step": 16
},
{
"epoch": 0.12878787878787878,
"grad_norm": 0.1861179620027542,
"learning_rate": 9.99857885177485e-05,
"loss": 1.353515625,
"step": 17
},
{
"epoch": 0.13636363636363635,
"grad_norm": 0.17025507986545563,
"learning_rate": 9.996802626882653e-05,
"loss": 1.2822265625,
"step": 18
},
{
"epoch": 0.14393939393939395,
"grad_norm": 0.22721347212791443,
"learning_rate": 9.99431630472708e-05,
"loss": 1.3330078125,
"step": 19
},
{
"epoch": 0.15151515151515152,
"grad_norm": 0.23746272921562195,
"learning_rate": 9.991120277927223e-05,
"loss": 1.2744140625,
"step": 20
},
{
"epoch": 0.1590909090909091,
"grad_norm": 0.2563123404979706,
"learning_rate": 9.987215051172763e-05,
"loss": 1.29736328125,
"step": 21
},
{
"epoch": 0.16666666666666666,
"grad_norm": 0.25374919176101685,
"learning_rate": 9.982601241144277e-05,
"loss": 1.31103515625,
"step": 22
},
{
"epoch": 0.17424242424242425,
"grad_norm": 0.24282824993133545,
"learning_rate": 9.977279576415853e-05,
"loss": 1.3349609375,
"step": 23
},
{
"epoch": 0.18181818181818182,
"grad_norm": 0.2210780382156372,
"learning_rate": 9.971250897340038e-05,
"loss": 1.318359375,
"step": 24
},
{
"epoch": 0.1893939393939394,
"grad_norm": 0.18094603717327118,
"learning_rate": 9.964516155915151e-05,
"loss": 1.3046875,
"step": 25
},
{
"epoch": 0.19696969696969696,
"grad_norm": 0.17291805148124695,
"learning_rate": 9.95707641563493e-05,
"loss": 1.275390625,
"step": 26
},
{
"epoch": 0.20454545454545456,
"grad_norm": 0.12008367478847504,
"learning_rate": 9.948932851320614e-05,
"loss": 1.102142333984375,
"step": 27
},
{
"epoch": 0.21212121212121213,
"grad_norm": 0.11401014029979706,
"learning_rate": 9.940086748935406e-05,
"loss": 1.2431640625,
"step": 28
},
{
"epoch": 0.2196969696969697,
"grad_norm": 0.09813079237937927,
"learning_rate": 9.930539505381426e-05,
"loss": 1.18505859375,
"step": 29
},
{
"epoch": 0.22727272727272727,
"grad_norm": 0.09758817404508591,
"learning_rate": 9.920292628279099e-05,
"loss": 1.21484375,
"step": 30
},
{
"epoch": 0.23484848484848486,
"grad_norm": 0.10868855565786362,
"learning_rate": 9.909347735729111e-05,
"loss": 1.228515625,
"step": 31
},
{
"epoch": 0.24242424242424243,
"grad_norm": 0.09499826282262802,
"learning_rate": 9.897706556056872e-05,
"loss": 1.2041015625,
"step": 32
},
{
"epoch": 0.25,
"grad_norm": 0.09282124787569046,
"learning_rate": 9.885370927539598e-05,
"loss": 1.1943359375,
"step": 33
},
{
"epoch": 0.25757575757575757,
"grad_norm": 0.09949434548616409,
"learning_rate": 9.872342798116033e-05,
"loss": 1.07257080078125,
"step": 34
},
{
"epoch": 0.26515151515151514,
"grad_norm": 0.12807920575141907,
"learning_rate": 9.858624225078841e-05,
"loss": 1.26171875,
"step": 35
},
{
"epoch": 0.2727272727272727,
"grad_norm": 0.10821354389190674,
"learning_rate": 9.844217374749732e-05,
"loss": 1.2041015625,
"step": 36
},
{
"epoch": 0.2803030303030303,
"grad_norm": 0.1073274165391922,
"learning_rate": 9.829124522137386e-05,
"loss": 1.09130859375,
"step": 37
},
{
"epoch": 0.2878787878787879,
"grad_norm": 0.10323720425367355,
"learning_rate": 9.813348050578191e-05,
"loss": 1.14697265625,
"step": 38
},
{
"epoch": 0.29545454545454547,
"grad_norm": 0.08878322690725327,
"learning_rate": 9.796890451359894e-05,
"loss": 1.05615234375,
"step": 39
},
{
"epoch": 0.30303030303030304,
"grad_norm": 0.10852427035570145,
"learning_rate": 9.779754323328192e-05,
"loss": 1.15966796875,
"step": 40
},
{
"epoch": 0.3106060606060606,
"grad_norm": 0.09462536871433258,
"learning_rate": 9.76194237247635e-05,
"loss": 1.154296875,
"step": 41
},
{
"epoch": 0.3181818181818182,
"grad_norm": 0.09913163632154465,
"learning_rate": 9.743457411517892e-05,
"loss": 1.19384765625,
"step": 42
},
{
"epoch": 0.32575757575757575,
"grad_norm": 0.080966055393219,
"learning_rate": 9.724302359442434e-05,
"loss": 1.1201171875,
"step": 43
},
{
"epoch": 0.3333333333333333,
"grad_norm": 0.07970893383026123,
"learning_rate": 9.704480241054755e-05,
"loss": 1.212890625,
"step": 44
},
{
"epoch": 0.3409090909090909,
"grad_norm": 0.07800300419330597,
"learning_rate": 9.683994186497132e-05,
"loss": 1.1826171875,
"step": 45
},
{
"epoch": 0.3484848484848485,
"grad_norm": 0.08582798391580582,
"learning_rate": 9.66284743075506e-05,
"loss": 1.2060546875,
"step": 46
},
{
"epoch": 0.3560606060606061,
"grad_norm": 0.07266346365213394,
"learning_rate": 9.641043313146417e-05,
"loss": 1.1083984375,
"step": 47
},
{
"epoch": 0.36363636363636365,
"grad_norm": 0.09038397669792175,
"learning_rate": 9.618585276794129e-05,
"loss": 1.1845703125,
"step": 48
},
{
"epoch": 0.3712121212121212,
"grad_norm": 0.09399913996458054,
"learning_rate": 9.595476868082481e-05,
"loss": 1.1875,
"step": 49
},
{
"epoch": 0.3787878787878788,
"grad_norm": 0.07513436675071716,
"learning_rate": 9.571721736097089e-05,
"loss": 1.16357421875,
"step": 50
},
{
"epoch": 0.38636363636363635,
"grad_norm": 0.07933320850133896,
"learning_rate": 9.54732363204867e-05,
"loss": 1.1796875,
"step": 51
},
{
"epoch": 0.3939393939393939,
"grad_norm": 0.07782892882823944,
"learning_rate": 9.522286408680687e-05,
"loss": 1.2119140625,
"step": 52
},
{
"epoch": 0.4015151515151515,
"grad_norm": 0.07191629707813263,
"learning_rate": 9.496614019660951e-05,
"loss": 1.185546875,
"step": 53
},
{
"epoch": 0.4090909090909091,
"grad_norm": 0.07715420424938202,
"learning_rate": 9.47031051895729e-05,
"loss": 1.14306640625,
"step": 54
},
{
"epoch": 0.4166666666666667,
"grad_norm": 0.07936612516641617,
"learning_rate": 9.443380060197387e-05,
"loss": 1.1943359375,
"step": 55
},
{
"epoch": 0.42424242424242425,
"grad_norm": 0.08630044013261795,
"learning_rate": 9.415826896012865e-05,
"loss": 1.1494140625,
"step": 56
},
{
"epoch": 0.42424242424242425,
"eval_loss": 1.1767578125,
"eval_runtime": 2.2352,
"eval_samples_per_second": 3.579,
"eval_steps_per_second": 0.447,
"step": 56
},
{
"epoch": 0.4318181818181818,
"grad_norm": 0.07743937522172928,
"learning_rate": 9.387655377367758e-05,
"loss": 1.14404296875,
"step": 57
},
{
"epoch": 0.4393939393939394,
"grad_norm": 0.06784118711948395,
"learning_rate": 9.358869952871436e-05,
"loss": 1.13134765625,
"step": 58
},
{
"epoch": 0.44696969696969696,
"grad_norm": 0.06836315244436264,
"learning_rate": 9.329475168076114e-05,
"loss": 1.11572265625,
"step": 59
},
{
"epoch": 0.45454545454545453,
"grad_norm": 0.07465013861656189,
"learning_rate": 9.299475664759069e-05,
"loss": 1.1376953125,
"step": 60
},
{
"epoch": 0.4621212121212121,
"grad_norm": 0.0738844946026802,
"learning_rate": 9.268876180189639e-05,
"loss": 1.1220703125,
"step": 61
},
{
"epoch": 0.4696969696969697,
"grad_norm": 0.07709069550037384,
"learning_rate": 9.237681546381157e-05,
"loss": 1.14453125,
"step": 62
},
{
"epoch": 0.4772727272727273,
"grad_norm": 0.07183508574962616,
"learning_rate": 9.205896689327923e-05,
"loss": 1.125,
"step": 63
},
{
"epoch": 0.48484848484848486,
"grad_norm": 0.07055969536304474,
"learning_rate": 9.173526628227329e-05,
"loss": 0.9150390625,
"step": 64
},
{
"epoch": 0.49242424242424243,
"grad_norm": 0.08102235943078995,
"learning_rate": 9.140576474687264e-05,
"loss": 1.1328125,
"step": 65
},
{
"epoch": 0.5,
"grad_norm": 0.07835587114095688,
"learning_rate": 9.107051431918944e-05,
"loss": 1.12646484375,
"step": 66
},
{
"epoch": 0.5075757575757576,
"grad_norm": 0.06946656852960587,
"learning_rate": 9.07295679391526e-05,
"loss": 1.1455078125,
"step": 67
},
{
"epoch": 0.5151515151515151,
"grad_norm": 0.08474933356046677,
"learning_rate": 9.038297944614785e-05,
"loss": 1.1357421875,
"step": 68
},
{
"epoch": 0.5227272727272727,
"grad_norm": 0.0714707151055336,
"learning_rate": 9.003080357051607e-05,
"loss": 1.13037109375,
"step": 69
},
{
"epoch": 0.5303030303030303,
"grad_norm": 0.06972786784172058,
"learning_rate": 8.967309592491052e-05,
"loss": 1.15185546875,
"step": 70
},
{
"epoch": 0.5378787878787878,
"grad_norm": 0.07356660068035126,
"learning_rate": 8.930991299551515e-05,
"loss": 1.123046875,
"step": 71
},
{
"epoch": 0.5454545454545454,
"grad_norm": 0.08069396018981934,
"learning_rate": 8.894131213312467e-05,
"loss": 1.134765625,
"step": 72
},
{
"epoch": 0.553030303030303,
"grad_norm": 0.0700419619679451,
"learning_rate": 8.85673515440882e-05,
"loss": 1.11279296875,
"step": 73
},
{
"epoch": 0.5606060606060606,
"grad_norm": 0.07328810542821884,
"learning_rate": 8.818809028111783e-05,
"loss": 1.08984375,
"step": 74
},
{
"epoch": 0.5681818181818182,
"grad_norm": 0.07241823524236679,
"learning_rate": 8.780358823396352e-05,
"loss": 1.041015625,
"step": 75
},
{
"epoch": 0.5757575757575758,
"grad_norm": 0.06917116045951843,
"learning_rate": 8.741390611995581e-05,
"loss": 1.115234375,
"step": 76
},
{
"epoch": 0.5833333333333334,
"grad_norm": 0.07732143253087997,
"learning_rate": 8.701910547441786e-05,
"loss": 1.11376953125,
"step": 77
},
{
"epoch": 0.5909090909090909,
"grad_norm": 0.08793466538190842,
"learning_rate": 8.661924864094822e-05,
"loss": 1.12060546875,
"step": 78
},
{
"epoch": 0.5984848484848485,
"grad_norm": 0.07058653235435486,
"learning_rate": 8.621439876157622e-05,
"loss": 1.119140625,
"step": 79
},
{
"epoch": 0.6060606060606061,
"grad_norm": 0.07628446072340012,
"learning_rate": 8.5804619766791e-05,
"loss": 1.11083984375,
"step": 80
},
{
"epoch": 0.6136363636363636,
"grad_norm": 0.07461079210042953,
"learning_rate": 8.53899763654461e-05,
"loss": 1.09912109375,
"step": 81
},
{
"epoch": 0.6212121212121212,
"grad_norm": 0.07395722717046738,
"learning_rate": 8.497053403454133e-05,
"loss": 1.09228515625,
"step": 82
},
{
"epoch": 0.6287878787878788,
"grad_norm": 0.07566767185926437,
"learning_rate": 8.454635900888305e-05,
"loss": 1.13232421875,
"step": 83
},
{
"epoch": 0.6363636363636364,
"grad_norm": 0.07291608303785324,
"learning_rate": 8.4117518270625e-05,
"loss": 1.111328125,
"step": 84
},
{
"epoch": 0.6439393939393939,
"grad_norm": 0.07590695470571518,
"learning_rate": 8.368407953869104e-05,
"loss": 1.0595703125,
"step": 85
},
{
"epoch": 0.6515151515151515,
"grad_norm": 0.07056854665279388,
"learning_rate": 8.324611125808153e-05,
"loss": 1.12353515625,
"step": 86
},
{
"epoch": 0.6590909090909091,
"grad_norm": 0.08102933317422867,
"learning_rate": 8.280368258906505e-05,
"loss": 1.0751953125,
"step": 87
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.0732758417725563,
"learning_rate": 8.235686339625725e-05,
"loss": 1.08251953125,
"step": 88
},
{
"epoch": 0.6742424242424242,
"grad_norm": 0.07417816668748856,
"learning_rate": 8.190572423758835e-05,
"loss": 1.1328125,
"step": 89
},
{
"epoch": 0.6818181818181818,
"grad_norm": 0.07961579412221909,
"learning_rate": 8.14503363531613e-05,
"loss": 1.107421875,
"step": 90
},
{
"epoch": 0.6893939393939394,
"grad_norm": 0.07082675397396088,
"learning_rate": 8.099077165400204e-05,
"loss": 1.04638671875,
"step": 91
},
{
"epoch": 0.696969696969697,
"grad_norm": 0.07951372116804123,
"learning_rate": 8.052710271070405e-05,
"loss": 1.1015625,
"step": 92
},
{
"epoch": 0.7045454545454546,
"grad_norm": 0.07731400430202484,
"learning_rate": 8.005940274196846e-05,
"loss": 1.05712890625,
"step": 93
},
{
"epoch": 0.7121212121212122,
"grad_norm": 0.0780235081911087,
"learning_rate": 7.958774560304213e-05,
"loss": 1.1123046875,
"step": 94
},
{
"epoch": 0.7196969696969697,
"grad_norm": 0.07907724380493164,
"learning_rate": 7.911220577405484e-05,
"loss": 1.1240234375,
"step": 95
},
{
"epoch": 0.7272727272727273,
"grad_norm": 0.08562244474887848,
"learning_rate": 7.863285834825832e-05,
"loss": 1.11279296875,
"step": 96
},
{
"epoch": 0.7348484848484849,
"grad_norm": 0.08156926184892654,
"learning_rate": 7.814977902016779e-05,
"loss": 1.09765625,
"step": 97
},
{
"epoch": 0.7424242424242424,
"grad_norm": 0.08556552231311798,
"learning_rate": 7.766304407360924e-05,
"loss": 1.06201171875,
"step": 98
},
{
"epoch": 0.75,
"grad_norm": 0.07410125434398651,
"learning_rate": 7.717273036967312e-05,
"loss": 1.1123046875,
"step": 99
},
{
"epoch": 0.7575757575757576,
"grad_norm": 0.07887564599514008,
"learning_rate": 7.667891533457719e-05,
"loss": 1.0986328125,
"step": 100
},
{
"epoch": 0.7651515151515151,
"grad_norm": 0.08241984993219376,
"learning_rate": 7.618167694743998e-05,
"loss": 1.0625,
"step": 101
},
{
"epoch": 0.7727272727272727,
"grad_norm": 0.075218066573143,
"learning_rate": 7.568109372796697e-05,
"loss": 1.06787109375,
"step": 102
},
{
"epoch": 0.7803030303030303,
"grad_norm": 0.0772901326417923,
"learning_rate": 7.517724472405146e-05,
"loss": 1.1064453125,
"step": 103
},
{
"epoch": 0.7878787878787878,
"grad_norm": 0.08121981471776962,
"learning_rate": 7.4670209499292e-05,
"loss": 1.0888671875,
"step": 104
},
{
"epoch": 0.7954545454545454,
"grad_norm": 0.08568460494279861,
"learning_rate": 7.416006812042828e-05,
"loss": 1.04150390625,
"step": 105
},
{
"epoch": 0.803030303030303,
"grad_norm": 0.08072242885828018,
"learning_rate": 7.36469011446978e-05,
"loss": 1.0478515625,
"step": 106
},
{
"epoch": 0.8106060606060606,
"grad_norm": 0.07596985250711441,
"learning_rate": 7.313078960711483e-05,
"loss": 1.10546875,
"step": 107
},
{
"epoch": 0.8181818181818182,
"grad_norm": 0.10650064051151276,
"learning_rate": 7.261181500767413e-05,
"loss": 1.09326171875,
"step": 108
},
{
"epoch": 0.8181818181818182,
"eval_loss": 1.1123046875,
"eval_runtime": 2.2043,
"eval_samples_per_second": 3.629,
"eval_steps_per_second": 0.454,
"step": 108
},
{
"epoch": 0.8257575757575758,
"grad_norm": 0.08208151906728745,
"learning_rate": 7.209005929848107e-05,
"loss": 1.10546875,
"step": 109
},
{
"epoch": 0.8333333333333334,
"grad_norm": 0.08447117358446121,
"learning_rate": 7.156560487081053e-05,
"loss": 1.13671875,
"step": 110
},
{
"epoch": 0.8409090909090909,
"grad_norm": 0.08980458229780197,
"learning_rate": 7.103853454209628e-05,
"loss": 1.1298828125,
"step": 111
},
{
"epoch": 0.8484848484848485,
"grad_norm": 0.08742181211709976,
"learning_rate": 7.050893154285327e-05,
"loss": 1.1005859375,
"step": 112
},
{
"epoch": 0.8560606060606061,
"grad_norm": 0.07789453119039536,
"learning_rate": 6.99768795035344e-05,
"loss": 1.07958984375,
"step": 113
},
{
"epoch": 0.8636363636363636,
"grad_norm": 0.096565380692482,
"learning_rate": 6.944246244132443e-05,
"loss": 1.0458984375,
"step": 114
},
{
"epoch": 0.8712121212121212,
"grad_norm": 0.08536340296268463,
"learning_rate": 6.890576474687263e-05,
"loss": 1.08447265625,
"step": 115
},
{
"epoch": 0.8787878787878788,
"grad_norm": 0.09130439907312393,
"learning_rate": 6.836687117096657e-05,
"loss": 1.068359375,
"step": 116
},
{
"epoch": 0.8863636363636364,
"grad_norm": 0.0798850879073143,
"learning_rate": 6.782586681114894e-05,
"loss": 0.9716796875,
"step": 117
},
{
"epoch": 0.8939393939393939,
"grad_norm": 0.09666863083839417,
"learning_rate": 6.728283709827963e-05,
"loss": 1.0810546875,
"step": 118
},
{
"epoch": 0.9015151515151515,
"grad_norm": 0.081708624958992,
"learning_rate": 6.673786778304537e-05,
"loss": 0.946533203125,
"step": 119
},
{
"epoch": 0.9090909090909091,
"grad_norm": 0.09238912165164948,
"learning_rate": 6.619104492241848e-05,
"loss": 1.09375,
"step": 120
},
{
"epoch": 0.9166666666666666,
"grad_norm": 0.09275467693805695,
"learning_rate": 6.564245486606762e-05,
"loss": 0.96533203125,
"step": 121
},
{
"epoch": 0.9242424242424242,
"grad_norm": 0.10164269059896469,
"learning_rate": 6.509218424272216e-05,
"loss": 1.09130859375,
"step": 122
},
{
"epoch": 0.9318181818181818,
"grad_norm": 0.08422201871871948,
"learning_rate": 6.454031994649247e-05,
"loss": 1.03857421875,
"step": 123
},
{
"epoch": 0.9393939393939394,
"grad_norm": 0.09149625152349472,
"learning_rate": 6.398694912314831e-05,
"loss": 1.05517578125,
"step": 124
},
{
"epoch": 0.946969696969697,
"grad_norm": 0.09097972512245178,
"learning_rate": 6.343215915635762e-05,
"loss": 1.05322265625,
"step": 125
},
{
"epoch": 0.9545454545454546,
"grad_norm": 0.09669139236211777,
"learning_rate": 6.287603765388743e-05,
"loss": 1.0791015625,
"step": 126
},
{
"epoch": 0.9621212121212122,
"grad_norm": 0.08664537966251373,
"learning_rate": 6.231867243376977e-05,
"loss": 1.10498046875,
"step": 127
},
{
"epoch": 0.9696969696969697,
"grad_norm": 0.12009935826063156,
"learning_rate": 6.176015151043407e-05,
"loss": 0.974609375,
"step": 128
},
{
"epoch": 0.9772727272727273,
"grad_norm": 0.09793458133935928,
"learning_rate": 6.120056308080872e-05,
"loss": 1.07763671875,
"step": 129
},
{
"epoch": 0.9848484848484849,
"grad_norm": 0.09152775257825851,
"learning_rate": 6.06399955103937e-05,
"loss": 1.06884765625,
"step": 130
},
{
"epoch": 0.9924242424242424,
"grad_norm": 0.09448117762804031,
"learning_rate": 6.007853731930667e-05,
"loss": 1.11083984375,
"step": 131
},
{
"epoch": 1.0,
"grad_norm": 0.0885840430855751,
"learning_rate": 5.951627716830467e-05,
"loss": 0.959716796875,
"step": 132
},
{
"epoch": 1.0075757575757576,
"grad_norm": 0.07837975770235062,
"learning_rate": 5.8953303844783456e-05,
"loss": 0.858642578125,
"step": 133
},
{
"epoch": 1.0151515151515151,
"grad_norm": 0.10188230127096176,
"learning_rate": 5.838970624875698e-05,
"loss": 1.08935546875,
"step": 134
},
{
"epoch": 1.0227272727272727,
"grad_norm": 0.09374918043613434,
"learning_rate": 5.782557337881911e-05,
"loss": 1.0625,
"step": 135
},
{
"epoch": 1.0303030303030303,
"grad_norm": 0.08964771032333374,
"learning_rate": 5.726099431808963e-05,
"loss": 1.03955078125,
"step": 136
},
{
"epoch": 1.0378787878787878,
"grad_norm": 0.08666251599788666,
"learning_rate": 5.669605822014706e-05,
"loss": 1.048828125,
"step": 137
},
{
"epoch": 1.0454545454545454,
"grad_norm": 0.0832890048623085,
"learning_rate": 5.61308542949502e-05,
"loss": 1.02880859375,
"step": 138
},
{
"epoch": 1.053030303030303,
"grad_norm": 0.09309335798025131,
"learning_rate": 5.556547179475088e-05,
"loss": 1.0673828125,
"step": 139
},
{
"epoch": 1.0606060606060606,
"grad_norm": 0.0930396094918251,
"learning_rate": 5.500000000000001e-05,
"loss": 0.99951171875,
"step": 140
},
{
"epoch": 1.0681818181818181,
"grad_norm": 0.0936756506562233,
"learning_rate": 5.443452820524913e-05,
"loss": 1.033203125,
"step": 141
},
{
"epoch": 1.0757575757575757,
"grad_norm": 0.08000978827476501,
"learning_rate": 5.3869145705049814e-05,
"loss": 0.945709228515625,
"step": 142
},
{
"epoch": 1.0833333333333333,
"grad_norm": 0.08305247128009796,
"learning_rate": 5.330394177985295e-05,
"loss": 1.0087890625,
"step": 143
},
{
"epoch": 1.0909090909090908,
"grad_norm": 0.08222263306379318,
"learning_rate": 5.273900568191038e-05,
"loss": 0.947265625,
"step": 144
},
{
"epoch": 1.0984848484848484,
"grad_norm": 0.08853105455636978,
"learning_rate": 5.2174426621180906e-05,
"loss": 0.990234375,
"step": 145
},
{
"epoch": 1.106060606060606,
"grad_norm": 0.11007614433765411,
"learning_rate": 5.161029375124303e-05,
"loss": 1.0546875,
"step": 146
},
{
"epoch": 1.1136363636363635,
"grad_norm": 0.08950755000114441,
"learning_rate": 5.104669615521657e-05,
"loss": 1.07275390625,
"step": 147
},
{
"epoch": 1.121212121212121,
"grad_norm": 0.10570712387561798,
"learning_rate": 5.048372283169532e-05,
"loss": 1.056640625,
"step": 148
},
{
"epoch": 1.128787878787879,
"grad_norm": 0.10939672589302063,
"learning_rate": 4.992146268069333e-05,
"loss": 1.05615234375,
"step": 149
},
{
"epoch": 1.1363636363636362,
"grad_norm": 0.09200646728277206,
"learning_rate": 4.936000448960631e-05,
"loss": 1.0712890625,
"step": 150
},
{
"epoch": 1.143939393939394,
"grad_norm": 0.09504993259906769,
"learning_rate": 4.87994369191913e-05,
"loss": 1.11328125,
"step": 151
},
{
"epoch": 1.1515151515151516,
"grad_norm": 0.09119915962219238,
"learning_rate": 4.823984848956593e-05,
"loss": 1.03515625,
"step": 152
},
{
"epoch": 1.1590909090909092,
"grad_norm": 0.10821505635976791,
"learning_rate": 4.768132756623024e-05,
"loss": 1.03955078125,
"step": 153
},
{
"epoch": 1.1666666666666667,
"grad_norm": 0.10071408748626709,
"learning_rate": 4.7123962346112584e-05,
"loss": 1.06884765625,
"step": 154
},
{
"epoch": 1.1742424242424243,
"grad_norm": 0.09807620942592621,
"learning_rate": 4.6567840843642384e-05,
"loss": 1.091796875,
"step": 155
},
{
"epoch": 1.1818181818181819,
"grad_norm": 0.10361678898334503,
"learning_rate": 4.601305087685169e-05,
"loss": 1.0732421875,
"step": 156
},
{
"epoch": 1.1893939393939394,
"grad_norm": 0.09949921071529388,
"learning_rate": 4.545968005350756e-05,
"loss": 1.0869140625,
"step": 157
},
{
"epoch": 1.196969696969697,
"grad_norm": 0.1019904762506485,
"learning_rate": 4.490781575727786e-05,
"loss": 1.0546875,
"step": 158
},
{
"epoch": 1.2045454545454546,
"grad_norm": 0.09665553271770477,
"learning_rate": 4.43575451339324e-05,
"loss": 0.9061279296875,
"step": 159
},
{
"epoch": 1.2121212121212122,
"grad_norm": 0.09645150601863861,
"learning_rate": 4.380895507758155e-05,
"loss": 1.05322265625,
"step": 160
},
{
"epoch": 1.2121212121212122,
"eval_loss": 1.083984375,
"eval_runtime": 2.2092,
"eval_samples_per_second": 3.621,
"eval_steps_per_second": 0.453,
"step": 160
},
{
"epoch": 1.2196969696969697,
"grad_norm": 0.0945630818605423,
"learning_rate": 4.3262132216954656e-05,
"loss": 1.00634765625,
"step": 161
},
{
"epoch": 1.2272727272727273,
"grad_norm": 0.10197978466749191,
"learning_rate": 4.271716290172038e-05,
"loss": 1.025390625,
"step": 162
},
{
"epoch": 1.2348484848484849,
"grad_norm": 0.10562216490507126,
"learning_rate": 4.217413318885108e-05,
"loss": 1.025390625,
"step": 163
},
{
"epoch": 1.2424242424242424,
"grad_norm": 0.09081273525953293,
"learning_rate": 4.163312882903344e-05,
"loss": 1.0380859375,
"step": 164
},
{
"epoch": 1.25,
"grad_norm": 0.08740173280239105,
"learning_rate": 4.109423525312738e-05,
"loss": 1.0419921875,
"step": 165
},
{
"epoch": 1.2575757575757576,
"grad_norm": 0.09267672151327133,
"learning_rate": 4.0557537558675583e-05,
"loss": 0.90081787109375,
"step": 166
},
{
"epoch": 1.2651515151515151,
"grad_norm": 0.1051904633641243,
"learning_rate": 4.00231204964656e-05,
"loss": 1.0615234375,
"step": 167
},
{
"epoch": 1.2727272727272727,
"grad_norm": 0.10390182584524155,
"learning_rate": 3.949106845714674e-05,
"loss": 1.04150390625,
"step": 168
},
{
"epoch": 1.2803030303030303,
"grad_norm": 0.09526224434375763,
"learning_rate": 3.896146545790372e-05,
"loss": 0.94580078125,
"step": 169
},
{
"epoch": 1.2878787878787878,
"grad_norm": 0.0959937572479248,
"learning_rate": 3.843439512918949e-05,
"loss": 1.009765625,
"step": 170
},
{
"epoch": 1.2954545454545454,
"grad_norm": 0.10648560523986816,
"learning_rate": 3.790994070151895e-05,
"loss": 0.97216796875,
"step": 171
},
{
"epoch": 1.303030303030303,
"grad_norm": 0.09567048400640488,
"learning_rate": 3.738818499232589e-05,
"loss": 1.0234375,
"step": 172
},
{
"epoch": 1.3106060606060606,
"grad_norm": 0.09656581282615662,
"learning_rate": 3.686921039288519e-05,
"loss": 1.033203125,
"step": 173
},
{
"epoch": 1.3181818181818181,
"grad_norm": 0.11271104216575623,
"learning_rate": 3.6353098855302215e-05,
"loss": 1.05029296875,
"step": 174
},
{
"epoch": 1.3257575757575757,
"grad_norm": 0.09784029424190521,
"learning_rate": 3.583993187957173e-05,
"loss": 0.9931640625,
"step": 175
},
{
"epoch": 1.3333333333333333,
"grad_norm": 0.10976437479257584,
"learning_rate": 3.532979050070804e-05,
"loss": 1.07421875,
"step": 176
},
{
"epoch": 1.3409090909090908,
"grad_norm": 0.0987778902053833,
"learning_rate": 3.482275527594856e-05,
"loss": 1.03466796875,
"step": 177
},
{
"epoch": 1.3484848484848486,
"grad_norm": 0.10711042582988739,
"learning_rate": 3.431890627203305e-05,
"loss": 1.0654296875,
"step": 178
},
{
"epoch": 1.356060606060606,
"grad_norm": 0.0991782397031784,
"learning_rate": 3.381832305256004e-05,
"loss": 1.00244140625,
"step": 179
},
{
"epoch": 1.3636363636363638,
"grad_norm": 0.12253911793231964,
"learning_rate": 3.3321084665422807e-05,
"loss": 1.02294921875,
"step": 180
},
{
"epoch": 1.371212121212121,
"grad_norm": 0.12328967452049255,
"learning_rate": 3.2827269630326885e-05,
"loss": 1.02734375,
"step": 181
},
{
"epoch": 1.378787878787879,
"grad_norm": 0.09768681973218918,
"learning_rate": 3.233695592639077e-05,
"loss": 1.05419921875,
"step": 182
},
{
"epoch": 1.3863636363636362,
"grad_norm": 0.11941851675510406,
"learning_rate": 3.185022097983221e-05,
"loss": 1.0625,
"step": 183
},
{
"epoch": 1.393939393939394,
"grad_norm": 0.11135486513376236,
"learning_rate": 3.1367141651741694e-05,
"loss": 1.09814453125,
"step": 184
},
{
"epoch": 1.4015151515151514,
"grad_norm": 0.09698060154914856,
"learning_rate": 3.088779422594514e-05,
"loss": 1.078125,
"step": 185
},
{
"epoch": 1.4090909090909092,
"grad_norm": 0.10321120172739029,
"learning_rate": 3.0412254396957896e-05,
"loss": 1.0263671875,
"step": 186
},
{
"epoch": 1.4166666666666667,
"grad_norm": 0.11265275627374649,
"learning_rate": 2.994059725803156e-05,
"loss": 1.07373046875,
"step": 187
},
{
"epoch": 1.4242424242424243,
"grad_norm": 0.12025588750839233,
"learning_rate": 2.947289728929597e-05,
"loss": 1.0380859375,
"step": 188
},
{
"epoch": 1.4318181818181819,
"grad_norm": 0.10383759438991547,
"learning_rate": 2.900922834599797e-05,
"loss": 1.03662109375,
"step": 189
},
{
"epoch": 1.4393939393939394,
"grad_norm": 0.10661465674638748,
"learning_rate": 2.854966364683872e-05,
"loss": 1.03515625,
"step": 190
},
{
"epoch": 1.446969696969697,
"grad_norm": 0.09079551696777344,
"learning_rate": 2.809427576241167e-05,
"loss": 1.03564453125,
"step": 191
},
{
"epoch": 1.4545454545454546,
"grad_norm": 0.10627592355012894,
"learning_rate": 2.764313660374277e-05,
"loss": 1.0234375,
"step": 192
},
{
"epoch": 1.4621212121212122,
"grad_norm": 0.10409808158874512,
"learning_rate": 2.7196317410934964e-05,
"loss": 1.0361328125,
"step": 193
},
{
"epoch": 1.4696969696969697,
"grad_norm": 0.09841874986886978,
"learning_rate": 2.6753888741918488e-05,
"loss": 1.04736328125,
"step": 194
},
{
"epoch": 1.4772727272727273,
"grad_norm": 0.10632629692554474,
"learning_rate": 2.6315920461308964e-05,
"loss": 1.037109375,
"step": 195
},
{
"epoch": 1.4848484848484849,
"grad_norm": 0.0952998474240303,
"learning_rate": 2.588248172937502e-05,
"loss": 0.846923828125,
"step": 196
},
{
"epoch": 1.4924242424242424,
"grad_norm": 0.11421612650156021,
"learning_rate": 2.5453640991116967e-05,
"loss": 1.02978515625,
"step": 197
},
{
"epoch": 1.5,
"grad_norm": 0.10957960784435272,
"learning_rate": 2.5029465965458683e-05,
"loss": 1.02587890625,
"step": 198
},
{
"epoch": 1.5075757575757576,
"grad_norm": 0.09530032426118851,
"learning_rate": 2.46100236345539e-05,
"loss": 1.05908203125,
"step": 199
},
{
"epoch": 1.5151515151515151,
"grad_norm": 0.11528428643941879,
"learning_rate": 2.4195380233209008e-05,
"loss": 1.033203125,
"step": 200
},
{
"epoch": 1.5227272727272727,
"grad_norm": 0.1050548404455185,
"learning_rate": 2.3785601238423787e-05,
"loss": 1.03564453125,
"step": 201
},
{
"epoch": 1.5303030303030303,
"grad_norm": 0.10518501698970795,
"learning_rate": 2.3380751359051795e-05,
"loss": 1.05615234375,
"step": 202
},
{
"epoch": 1.5378787878787878,
"grad_norm": 0.10230522602796555,
"learning_rate": 2.298089452558216e-05,
"loss": 1.0546875,
"step": 203
},
{
"epoch": 1.5454545454545454,
"grad_norm": 0.11399275809526443,
"learning_rate": 2.258609388004419e-05,
"loss": 1.0419921875,
"step": 204
},
{
"epoch": 1.553030303030303,
"grad_norm": 0.10160615295171738,
"learning_rate": 2.219641176603649e-05,
"loss": 1.0361328125,
"step": 205
},
{
"epoch": 1.5606060606060606,
"grad_norm": 0.10528803616762161,
"learning_rate": 2.181190971888218e-05,
"loss": 1.00537109375,
"step": 206
},
{
"epoch": 1.5681818181818183,
"grad_norm": 0.0914541482925415,
"learning_rate": 2.1432648455911808e-05,
"loss": 0.9873046875,
"step": 207
},
{
"epoch": 1.5757575757575757,
"grad_norm": 0.09788926690816879,
"learning_rate": 2.1058687866875328e-05,
"loss": 1.04248046875,
"step": 208
},
{
"epoch": 1.5833333333333335,
"grad_norm": 0.10901805758476257,
"learning_rate": 2.0690087004484844e-05,
"loss": 1.03125,
"step": 209
},
{
"epoch": 1.5909090909090908,
"grad_norm": 0.12249114364385605,
"learning_rate": 2.0326904075089492e-05,
"loss": 1.02685546875,
"step": 210
},
{
"epoch": 1.5984848484848486,
"grad_norm": 0.1073046326637268,
"learning_rate": 1.996919642948395e-05,
"loss": 1.03955078125,
"step": 211
},
{
"epoch": 1.606060606060606,
"grad_norm": 0.10490427166223526,
"learning_rate": 1.961702055385215e-05,
"loss": 1.046875,
"step": 212
},
{
"epoch": 1.606060606060606,
"eval_loss": 1.072265625,
"eval_runtime": 2.2194,
"eval_samples_per_second": 3.605,
"eval_steps_per_second": 0.451,
"step": 212
},
{
"epoch": 1.6136363636363638,
"grad_norm": 0.10367228835821152,
"learning_rate": 1.927043206084741e-05,
"loss": 1.020263671875,
"step": 213
},
{
"epoch": 1.621212121212121,
"grad_norm": 0.10339723527431488,
"learning_rate": 1.892948568081055e-05,
"loss": 1.01513671875,
"step": 214
},
{
"epoch": 1.628787878787879,
"grad_norm": 0.09706207364797592,
"learning_rate": 1.8594235253127375e-05,
"loss": 1.078125,
"step": 215
},
{
"epoch": 1.6363636363636362,
"grad_norm": 0.1056489497423172,
"learning_rate": 1.8264733717726722e-05,
"loss": 1.04931640625,
"step": 216
},
{
"epoch": 1.643939393939394,
"grad_norm": 0.10254838317632675,
"learning_rate": 1.7941033106720768e-05,
"loss": 0.98974609375,
"step": 217
},
{
"epoch": 1.6515151515151514,
"grad_norm": 0.09696169197559357,
"learning_rate": 1.7623184536188424e-05,
"loss": 1.06884765625,
"step": 218
},
{
"epoch": 1.6590909090909092,
"grad_norm": 0.10388782620429993,
"learning_rate": 1.7311238198103627e-05,
"loss": 1.00830078125,
"step": 219
},
{
"epoch": 1.6666666666666665,
"grad_norm": 0.10069543868303299,
"learning_rate": 1.7005243352409334e-05,
"loss": 1.01708984375,
"step": 220
},
{
"epoch": 1.6742424242424243,
"grad_norm": 0.0987059697508812,
"learning_rate": 1.6705248319238876e-05,
"loss": 1.06982421875,
"step": 221
},
{
"epoch": 1.6818181818181817,
"grad_norm": 0.11282015591859818,
"learning_rate": 1.6411300471285656e-05,
"loss": 1.0341796875,
"step": 222
},
{
"epoch": 1.6893939393939394,
"grad_norm": 0.10136653482913971,
"learning_rate": 1.6123446226322414e-05,
"loss": 0.98681640625,
"step": 223
},
{
"epoch": 1.696969696969697,
"grad_norm": 0.10275169461965561,
"learning_rate": 1.5841731039871348e-05,
"loss": 1.03662109375,
"step": 224
},
{
"epoch": 1.7045454545454546,
"grad_norm": 0.10096168518066406,
"learning_rate": 1.556619939802615e-05,
"loss": 0.99609375,
"step": 225
},
{
"epoch": 1.7121212121212122,
"grad_norm": 0.10797571390867233,
"learning_rate": 1.529689481042711e-05,
"loss": 1.0498046875,
"step": 226
},
{
"epoch": 1.7196969696969697,
"grad_norm": 0.09793099015951157,
"learning_rate": 1.50338598033905e-05,
"loss": 1.0771484375,
"step": 227
},
{
"epoch": 1.7272727272727273,
"grad_norm": 0.10779029130935669,
"learning_rate": 1.4777135913193132e-05,
"loss": 1.04248046875,
"step": 228
},
{
"epoch": 1.7348484848484849,
"grad_norm": 0.1068890392780304,
"learning_rate": 1.4526763679513303e-05,
"loss": 1.03271484375,
"step": 229
},
{
"epoch": 1.7424242424242424,
"grad_norm": 0.10784820467233658,
"learning_rate": 1.428278263902913e-05,
"loss": 1.0029296875,
"step": 230
},
{
"epoch": 1.75,
"grad_norm": 0.10273246467113495,
"learning_rate": 1.4045231319175198e-05,
"loss": 1.05859375,
"step": 231
},
{
"epoch": 1.7575757575757576,
"grad_norm": 0.10326556116342545,
"learning_rate": 1.3814147232058714e-05,
"loss": 1.0439453125,
"step": 232
},
{
"epoch": 1.7651515151515151,
"grad_norm": 0.10401753336191177,
"learning_rate": 1.3589566868535836e-05,
"loss": 1.0126953125,
"step": 233
},
{
"epoch": 1.7727272727272727,
"grad_norm": 0.09665694832801819,
"learning_rate": 1.3371525692449394e-05,
"loss": 1.0205078125,
"step": 234
},
{
"epoch": 1.7803030303030303,
"grad_norm": 0.10136399418115616,
"learning_rate": 1.3160058135028691e-05,
"loss": 1.0517578125,
"step": 235
},
{
"epoch": 1.7878787878787878,
"grad_norm": 0.10064416378736496,
"learning_rate": 1.2955197589452462e-05,
"loss": 1.03759765625,
"step": 236
},
{
"epoch": 1.7954545454545454,
"grad_norm": 0.1058633103966713,
"learning_rate": 1.2756976405575668e-05,
"loss": 0.98779296875,
"step": 237
},
{
"epoch": 1.803030303030303,
"grad_norm": 0.1052682176232338,
"learning_rate": 1.2565425884821096e-05,
"loss": 0.9951171875,
"step": 238
},
{
"epoch": 1.8106060606060606,
"grad_norm": 0.09581825882196426,
"learning_rate": 1.2380576275236511e-05,
"loss": 1.064453125,
"step": 239
},
{
"epoch": 1.8181818181818183,
"grad_norm": 0.12475554645061493,
"learning_rate": 1.2202456766718093e-05,
"loss": 1.0390625,
"step": 240
},
{
"epoch": 1.8257575757575757,
"grad_norm": 0.10554253309965134,
"learning_rate": 1.2031095486401069e-05,
"loss": 1.05712890625,
"step": 241
},
{
"epoch": 1.8333333333333335,
"grad_norm": 0.0999523252248764,
"learning_rate": 1.1866519494218084e-05,
"loss": 1.0908203125,
"step": 242
},
{
"epoch": 1.8409090909090908,
"grad_norm": 0.1016223132610321,
"learning_rate": 1.1708754778626134e-05,
"loss": 1.0869140625,
"step": 243
},
{
"epoch": 1.8484848484848486,
"grad_norm": 0.10487925261259079,
"learning_rate": 1.1557826252502677e-05,
"loss": 1.046875,
"step": 244
},
{
"epoch": 1.856060606060606,
"grad_norm": 0.10017699748277664,
"learning_rate": 1.1413757749211602e-05,
"loss": 1.03759765625,
"step": 245
},
{
"epoch": 1.8636363636363638,
"grad_norm": 0.10680767148733139,
"learning_rate": 1.1276572018839673e-05,
"loss": 1.009765625,
"step": 246
},
{
"epoch": 1.871212121212121,
"grad_norm": 0.09949818253517151,
"learning_rate": 1.1146290724604024e-05,
"loss": 1.04443359375,
"step": 247
},
{
"epoch": 1.878787878787879,
"grad_norm": 0.1087232455611229,
"learning_rate": 1.1022934439431295e-05,
"loss": 1.02294921875,
"step": 248
},
{
"epoch": 1.8863636363636362,
"grad_norm": 0.09736225754022598,
"learning_rate": 1.0906522642708893e-05,
"loss": 0.931884765625,
"step": 249
},
{
"epoch": 1.893939393939394,
"grad_norm": 0.11243616044521332,
"learning_rate": 1.0797073717209014e-05,
"loss": 1.03466796875,
"step": 250
},
{
"epoch": 1.9015151515151514,
"grad_norm": 0.0957452729344368,
"learning_rate": 1.0694604946185762e-05,
"loss": 0.90380859375,
"step": 251
},
{
"epoch": 1.9090909090909092,
"grad_norm": 0.11622810363769531,
"learning_rate": 1.0599132510645939e-05,
"loss": 1.0478515625,
"step": 252
},
{
"epoch": 1.9166666666666665,
"grad_norm": 0.09894511848688126,
"learning_rate": 1.0510671486793873e-05,
"loss": 0.94091796875,
"step": 253
},
{
"epoch": 1.9242424242424243,
"grad_norm": 0.11916989833116531,
"learning_rate": 1.0429235843650698e-05,
"loss": 1.03564453125,
"step": 254
},
{
"epoch": 1.9318181818181817,
"grad_norm": 0.10247177630662918,
"learning_rate": 1.0354838440848503e-05,
"loss": 0.99951171875,
"step": 255
},
{
"epoch": 1.9393939393939394,
"grad_norm": 0.10714147984981537,
"learning_rate": 1.0287491026599623e-05,
"loss": 1.02294921875,
"step": 256
},
{
"epoch": 1.946969696969697,
"grad_norm": 0.10434214025735855,
"learning_rate": 1.0227204235841493e-05,
"loss": 1.0146484375,
"step": 257
},
{
"epoch": 1.9545454545454546,
"grad_norm": 0.11075802892446518,
"learning_rate": 1.0173987588557237e-05,
"loss": 1.0380859375,
"step": 258
},
{
"epoch": 1.9621212121212122,
"grad_norm": 0.1054508239030838,
"learning_rate": 1.0127849488272375e-05,
"loss": 1.07275390625,
"step": 259
},
{
"epoch": 1.9696969696969697,
"grad_norm": 0.12777963280677795,
"learning_rate": 1.008879722072778e-05,
"loss": 0.939453125,
"step": 260
},
{
"epoch": 1.9772727272727273,
"grad_norm": 0.10826599597930908,
"learning_rate": 1.0056836952729215e-05,
"loss": 1.04638671875,
"step": 261
},
{
"epoch": 1.9848484848484849,
"grad_norm": 0.10503465682268143,
"learning_rate": 1.0031973731173486e-05,
"loss": 1.03564453125,
"step": 262
},
{
"epoch": 1.9924242424242424,
"grad_norm": 0.10461784154176712,
"learning_rate": 1.0014211482251503e-05,
"loss": 1.07080078125,
"step": 263
},
{
"epoch": 2.0,
"grad_norm": 0.10553426295518875,
"learning_rate": 1.0003553010828276e-05,
"loss": 0.9228515625,
"step": 264
},
{
"epoch": 2.0,
"eval_loss": 1.0654296875,
"eval_runtime": 2.2193,
"eval_samples_per_second": 3.605,
"eval_steps_per_second": 0.451,
"step": 264
}
],
"logging_steps": 1.0,
"max_steps": 264,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 0,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.930198800069034e+19,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}