loreweaver-rp-v1 / trainer_state.json
Darksp33d's picture
Upload folder using huggingface_hub
cba3312 verified
Raw
History Blame Contribute Delete
58.1 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 318,
"global_step": 318,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0031446540880503146,
"grad_norm": 1.4494149684906006,
"learning_rate": 0.0,
"loss": 2.1591796875,
"step": 1
},
{
"epoch": 0.006289308176100629,
"grad_norm": 1.472015619277954,
"learning_rate": 1.0000000000000002e-06,
"loss": 1.9853515625,
"step": 2
},
{
"epoch": 0.009433962264150943,
"grad_norm": 1.2991633415222168,
"learning_rate": 2.0000000000000003e-06,
"loss": 1.94287109375,
"step": 3
},
{
"epoch": 0.012578616352201259,
"grad_norm": 1.3701603412628174,
"learning_rate": 3e-06,
"loss": 1.96337890625,
"step": 4
},
{
"epoch": 0.015723270440251572,
"grad_norm": 1.668572187423706,
"learning_rate": 4.000000000000001e-06,
"loss": 2.1435546875,
"step": 5
},
{
"epoch": 0.018867924528301886,
"grad_norm": 1.2014765739440918,
"learning_rate": 5e-06,
"loss": 1.95751953125,
"step": 6
},
{
"epoch": 0.0220125786163522,
"grad_norm": 1.3996855020523071,
"learning_rate": 6e-06,
"loss": 2.08203125,
"step": 7
},
{
"epoch": 0.025157232704402517,
"grad_norm": 1.3427141904830933,
"learning_rate": 7e-06,
"loss": 2.046142578125,
"step": 8
},
{
"epoch": 0.02830188679245283,
"grad_norm": 1.6375218629837036,
"learning_rate": 8.000000000000001e-06,
"loss": 2.1396484375,
"step": 9
},
{
"epoch": 0.031446540880503145,
"grad_norm": 0.7162802815437317,
"learning_rate": 9e-06,
"loss": 1.611328125,
"step": 10
},
{
"epoch": 0.03459119496855346,
"grad_norm": 1.6569421291351318,
"learning_rate": 1e-05,
"loss": 2.0888671875,
"step": 11
},
{
"epoch": 0.03773584905660377,
"grad_norm": 1.2393434047698975,
"learning_rate": 9.999739903471488e-06,
"loss": 1.94384765625,
"step": 12
},
{
"epoch": 0.040880503144654086,
"grad_norm": 1.2549704313278198,
"learning_rate": 9.998959640946033e-06,
"loss": 1.9033203125,
"step": 13
},
{
"epoch": 0.0440251572327044,
"grad_norm": 1.196319580078125,
"learning_rate": 9.997659293601066e-06,
"loss": 1.92578125,
"step": 14
},
{
"epoch": 0.04716981132075472,
"grad_norm": 0.961357831954956,
"learning_rate": 9.995838996722916e-06,
"loss": 1.71826171875,
"step": 15
},
{
"epoch": 0.050314465408805034,
"grad_norm": 0.7349967956542969,
"learning_rate": 9.993498939692744e-06,
"loss": 1.650146484375,
"step": 16
},
{
"epoch": 0.05345911949685535,
"grad_norm": 0.8328500986099243,
"learning_rate": 9.990639365966835e-06,
"loss": 1.73095703125,
"step": 17
},
{
"epoch": 0.05660377358490566,
"grad_norm": 0.7642322778701782,
"learning_rate": 9.987260573051268e-06,
"loss": 1.638671875,
"step": 18
},
{
"epoch": 0.059748427672955975,
"grad_norm": 0.6029979586601257,
"learning_rate": 9.983362912470967e-06,
"loss": 1.62841796875,
"step": 19
},
{
"epoch": 0.06289308176100629,
"grad_norm": 0.81263667345047,
"learning_rate": 9.978946789733126e-06,
"loss": 1.64306640625,
"step": 20
},
{
"epoch": 0.0660377358490566,
"grad_norm": 0.6555175185203552,
"learning_rate": 9.97401266428502e-06,
"loss": 1.5634765625,
"step": 21
},
{
"epoch": 0.06918238993710692,
"grad_norm": 0.5500423908233643,
"learning_rate": 9.968561049466214e-06,
"loss": 1.6044921875,
"step": 22
},
{
"epoch": 0.07232704402515723,
"grad_norm": 0.4210081100463867,
"learning_rate": 9.96259251245514e-06,
"loss": 1.506103515625,
"step": 23
},
{
"epoch": 0.07547169811320754,
"grad_norm": 0.4597223401069641,
"learning_rate": 9.9561076742101e-06,
"loss": 1.5380859375,
"step": 24
},
{
"epoch": 0.07861635220125786,
"grad_norm": 0.5838366150856018,
"learning_rate": 9.949107209404664e-06,
"loss": 1.58203125,
"step": 25
},
{
"epoch": 0.08176100628930817,
"grad_norm": 0.5811012387275696,
"learning_rate": 9.941591846357467e-06,
"loss": 1.4892578125,
"step": 26
},
{
"epoch": 0.08490566037735849,
"grad_norm": 0.5185905694961548,
"learning_rate": 9.933562366956445e-06,
"loss": 1.463134765625,
"step": 27
},
{
"epoch": 0.0880503144654088,
"grad_norm": 0.4009871184825897,
"learning_rate": 9.925019606577486e-06,
"loss": 1.4609375,
"step": 28
},
{
"epoch": 0.09119496855345911,
"grad_norm": 0.4346908926963806,
"learning_rate": 9.915964453997516e-06,
"loss": 1.47900390625,
"step": 29
},
{
"epoch": 0.09433962264150944,
"grad_norm": 0.4531169831752777,
"learning_rate": 9.906397851302036e-06,
"loss": 1.435546875,
"step": 30
},
{
"epoch": 0.09748427672955975,
"grad_norm": 0.32488882541656494,
"learning_rate": 9.896320793787106e-06,
"loss": 1.35546875,
"step": 31
},
{
"epoch": 0.10062893081761007,
"grad_norm": 0.37860193848609924,
"learning_rate": 9.885734329855798e-06,
"loss": 1.43505859375,
"step": 32
},
{
"epoch": 0.10377358490566038,
"grad_norm": 0.4280271530151367,
"learning_rate": 9.874639560909118e-06,
"loss": 1.4111328125,
"step": 33
},
{
"epoch": 0.1069182389937107,
"grad_norm": 0.3741612732410431,
"learning_rate": 9.863037641231424e-06,
"loss": 1.3740234375,
"step": 34
},
{
"epoch": 0.11006289308176101,
"grad_norm": 0.2793862521648407,
"learning_rate": 9.850929777870324e-06,
"loss": 1.33642578125,
"step": 35
},
{
"epoch": 0.11320754716981132,
"grad_norm": 0.3218320310115814,
"learning_rate": 9.838317230511111e-06,
"loss": 1.393310546875,
"step": 36
},
{
"epoch": 0.11635220125786164,
"grad_norm": 0.3182618319988251,
"learning_rate": 9.8252013113457e-06,
"loss": 1.422607421875,
"step": 37
},
{
"epoch": 0.11949685534591195,
"grad_norm": 0.3342706561088562,
"learning_rate": 9.811583384936108e-06,
"loss": 1.34912109375,
"step": 38
},
{
"epoch": 0.12264150943396226,
"grad_norm": 0.23300714790821075,
"learning_rate": 9.797464868072489e-06,
"loss": 1.3857421875,
"step": 39
},
{
"epoch": 0.12578616352201258,
"grad_norm": 0.26877281069755554,
"learning_rate": 9.782847229625729e-06,
"loss": 1.34716796875,
"step": 40
},
{
"epoch": 0.1289308176100629,
"grad_norm": 0.20799706876277924,
"learning_rate": 9.767731990394638e-06,
"loss": 1.3388671875,
"step": 41
},
{
"epoch": 0.1320754716981132,
"grad_norm": 0.20554538071155548,
"learning_rate": 9.752120722947717e-06,
"loss": 1.3515625,
"step": 42
},
{
"epoch": 0.13522012578616352,
"grad_norm": 0.3131389915943146,
"learning_rate": 9.736015051459551e-06,
"loss": 1.32373046875,
"step": 43
},
{
"epoch": 0.13836477987421383,
"grad_norm": 0.15417590737342834,
"learning_rate": 9.719416651541839e-06,
"loss": 1.302734375,
"step": 44
},
{
"epoch": 0.14150943396226415,
"grad_norm": 0.22120970487594604,
"learning_rate": 9.702327250069058e-06,
"loss": 1.27978515625,
"step": 45
},
{
"epoch": 0.14465408805031446,
"grad_norm": 0.16401228308677673,
"learning_rate": 9.68474862499881e-06,
"loss": 1.24658203125,
"step": 46
},
{
"epoch": 0.14779874213836477,
"grad_norm": 0.17903250455856323,
"learning_rate": 9.666682605186834e-06,
"loss": 1.299560546875,
"step": 47
},
{
"epoch": 0.1509433962264151,
"grad_norm": 0.18163374066352844,
"learning_rate": 9.648131070196749e-06,
"loss": 1.261474609375,
"step": 48
},
{
"epoch": 0.1540880503144654,
"grad_norm": 0.14828112721443176,
"learning_rate": 9.62909595010449e-06,
"loss": 1.33056640625,
"step": 49
},
{
"epoch": 0.15723270440251572,
"grad_norm": 0.1566939502954483,
"learning_rate": 9.609579225297524e-06,
"loss": 1.2666015625,
"step": 50
},
{
"epoch": 0.16037735849056603,
"grad_norm": 0.17874041199684143,
"learning_rate": 9.589582926268798e-06,
"loss": 1.36767578125,
"step": 51
},
{
"epoch": 0.16352201257861634,
"grad_norm": 0.16513372957706451,
"learning_rate": 9.569109133405495e-06,
"loss": 1.3505859375,
"step": 52
},
{
"epoch": 0.16666666666666666,
"grad_norm": 0.14548315107822418,
"learning_rate": 9.548159976772593e-06,
"loss": 1.29736328125,
"step": 53
},
{
"epoch": 0.16981132075471697,
"grad_norm": 0.15943686664104462,
"learning_rate": 9.526737635891262e-06,
"loss": 1.25634765625,
"step": 54
},
{
"epoch": 0.17295597484276728,
"grad_norm": 0.13931553065776825,
"learning_rate": 9.504844339512096e-06,
"loss": 1.2578125,
"step": 55
},
{
"epoch": 0.1761006289308176,
"grad_norm": 0.14830654859542847,
"learning_rate": 9.482482365383254e-06,
"loss": 1.315673828125,
"step": 56
},
{
"epoch": 0.1792452830188679,
"grad_norm": 0.12663020193576813,
"learning_rate": 9.45965404001347e-06,
"loss": 1.24853515625,
"step": 57
},
{
"epoch": 0.18238993710691823,
"grad_norm": 0.1341799944639206,
"learning_rate": 9.436361738430016e-06,
"loss": 1.26416015625,
"step": 58
},
{
"epoch": 0.18553459119496854,
"grad_norm": 0.12654203176498413,
"learning_rate": 9.412607883931608e-06,
"loss": 1.26416015625,
"step": 59
},
{
"epoch": 0.18867924528301888,
"grad_norm": 0.12026432156562805,
"learning_rate": 9.388394947836278e-06,
"loss": 1.2392578125,
"step": 60
},
{
"epoch": 0.1918238993710692,
"grad_norm": 0.11050467193126678,
"learning_rate": 9.363725449224281e-06,
"loss": 1.21728515625,
"step": 61
},
{
"epoch": 0.1949685534591195,
"grad_norm": 0.11030342429876328,
"learning_rate": 9.338601954675995e-06,
"loss": 1.35595703125,
"step": 62
},
{
"epoch": 0.19811320754716982,
"grad_norm": 0.11015181988477707,
"learning_rate": 9.313027078004903e-06,
"loss": 1.197509765625,
"step": 63
},
{
"epoch": 0.20125786163522014,
"grad_norm": 0.11742514371871948,
"learning_rate": 9.287003479985667e-06,
"loss": 1.2265625,
"step": 64
},
{
"epoch": 0.20440251572327045,
"grad_norm": 0.12025771290063858,
"learning_rate": 9.260533868077283e-06,
"loss": 1.24658203125,
"step": 65
},
{
"epoch": 0.20754716981132076,
"grad_norm": 0.10835342109203339,
"learning_rate": 9.233620996141421e-06,
"loss": 1.20751953125,
"step": 66
},
{
"epoch": 0.21069182389937108,
"grad_norm": 0.10320407897233963,
"learning_rate": 9.206267664155906e-06,
"loss": 1.16796875,
"step": 67
},
{
"epoch": 0.2138364779874214,
"grad_norm": 0.10821900516748428,
"learning_rate": 9.178476717923415e-06,
"loss": 1.2431640625,
"step": 68
},
{
"epoch": 0.2169811320754717,
"grad_norm": 0.10739477723836899,
"learning_rate": 9.150251048775403e-06,
"loss": 1.251220703125,
"step": 69
},
{
"epoch": 0.22012578616352202,
"grad_norm": 0.108885757625103,
"learning_rate": 9.121593593271297e-06,
"loss": 1.35107421875,
"step": 70
},
{
"epoch": 0.22327044025157233,
"grad_norm": 0.10561460256576538,
"learning_rate": 9.092507332892968e-06,
"loss": 1.27880859375,
"step": 71
},
{
"epoch": 0.22641509433962265,
"grad_norm": 0.09498227387666702,
"learning_rate": 9.062995293734562e-06,
"loss": 1.2900390625,
"step": 72
},
{
"epoch": 0.22955974842767296,
"grad_norm": 0.10343296080827713,
"learning_rate": 9.033060546187651e-06,
"loss": 1.164794921875,
"step": 73
},
{
"epoch": 0.23270440251572327,
"grad_norm": 0.10015164315700531,
"learning_rate": 9.002706204621802e-06,
"loss": 1.28173828125,
"step": 74
},
{
"epoch": 0.2358490566037736,
"grad_norm": 0.10706444084644318,
"learning_rate": 8.971935427060563e-06,
"loss": 1.15380859375,
"step": 75
},
{
"epoch": 0.2389937106918239,
"grad_norm": 0.10046739131212234,
"learning_rate": 8.940751414852904e-06,
"loss": 1.25927734375,
"step": 76
},
{
"epoch": 0.24213836477987422,
"grad_norm": 0.09688299149274826,
"learning_rate": 8.90915741234015e-06,
"loss": 1.32666015625,
"step": 77
},
{
"epoch": 0.24528301886792453,
"grad_norm": 0.1397327482700348,
"learning_rate": 8.877156706518453e-06,
"loss": 1.2939453125,
"step": 78
},
{
"epoch": 0.24842767295597484,
"grad_norm": 0.10606326162815094,
"learning_rate": 8.84475262669681e-06,
"loss": 1.298828125,
"step": 79
},
{
"epoch": 0.25157232704402516,
"grad_norm": 0.10689855366945267,
"learning_rate": 8.811948544150693e-06,
"loss": 1.2236328125,
"step": 80
},
{
"epoch": 0.25471698113207547,
"grad_norm": 0.10143591463565826,
"learning_rate": 8.778747871771293e-06,
"loss": 1.173828125,
"step": 81
},
{
"epoch": 0.2578616352201258,
"grad_norm": 0.09580762684345245,
"learning_rate": 8.745154063710464e-06,
"loss": 1.23779296875,
"step": 82
},
{
"epoch": 0.2610062893081761,
"grad_norm": 0.09986191987991333,
"learning_rate": 8.71117061502135e-06,
"loss": 1.2890625,
"step": 83
},
{
"epoch": 0.2641509433962264,
"grad_norm": 0.09059023857116699,
"learning_rate": 8.676801061294764e-06,
"loss": 1.255859375,
"step": 84
},
{
"epoch": 0.2672955974842767,
"grad_norm": 0.09870278090238571,
"learning_rate": 8.642048978291347e-06,
"loss": 1.247802734375,
"step": 85
},
{
"epoch": 0.27044025157232704,
"grad_norm": 0.09271863102912903,
"learning_rate": 8.60691798156956e-06,
"loss": 1.2841796875,
"step": 86
},
{
"epoch": 0.27358490566037735,
"grad_norm": 0.10226648300886154,
"learning_rate": 8.571411726109518e-06,
"loss": 1.173583984375,
"step": 87
},
{
"epoch": 0.27672955974842767,
"grad_norm": 0.08855801820755005,
"learning_rate": 8.535533905932739e-06,
"loss": 1.27197265625,
"step": 88
},
{
"epoch": 0.279874213836478,
"grad_norm": 0.09819382429122925,
"learning_rate": 8.49928825371781e-06,
"loss": 1.2265625,
"step": 89
},
{
"epoch": 0.2830188679245283,
"grad_norm": 0.0931096225976944,
"learning_rate": 8.46267854041206e-06,
"loss": 1.2666015625,
"step": 90
},
{
"epoch": 0.2861635220125786,
"grad_norm": 0.09582605212926865,
"learning_rate": 8.425708574839221e-06,
"loss": 1.2294921875,
"step": 91
},
{
"epoch": 0.2893081761006289,
"grad_norm": 0.09605991095304489,
"learning_rate": 8.388382203303181e-06,
"loss": 1.26953125,
"step": 92
},
{
"epoch": 0.29245283018867924,
"grad_norm": 0.08945699781179428,
"learning_rate": 8.3507033091878e-06,
"loss": 1.26611328125,
"step": 93
},
{
"epoch": 0.29559748427672955,
"grad_norm": 0.09489528834819794,
"learning_rate": 8.312675812552898e-06,
"loss": 1.22314453125,
"step": 94
},
{
"epoch": 0.29874213836477986,
"grad_norm": 0.0868961364030838,
"learning_rate": 8.274303669726427e-06,
"loss": 1.24072265625,
"step": 95
},
{
"epoch": 0.3018867924528302,
"grad_norm": 0.09084431827068329,
"learning_rate": 8.235590872892837e-06,
"loss": 1.2958984375,
"step": 96
},
{
"epoch": 0.3050314465408805,
"grad_norm": 0.09020113945007324,
"learning_rate": 8.196541449677758e-06,
"loss": 1.23828125,
"step": 97
},
{
"epoch": 0.3081761006289308,
"grad_norm": 0.08751130104064941,
"learning_rate": 8.157159462728956e-06,
"loss": 1.2392578125,
"step": 98
},
{
"epoch": 0.3113207547169811,
"grad_norm": 0.09124515950679779,
"learning_rate": 8.117449009293668e-06,
"loss": 1.230712890625,
"step": 99
},
{
"epoch": 0.31446540880503143,
"grad_norm": 0.08627773076295853,
"learning_rate": 8.077414220792328e-06,
"loss": 1.222900390625,
"step": 100
},
{
"epoch": 0.31761006289308175,
"grad_norm": 0.10408709198236465,
"learning_rate": 8.03705926238874e-06,
"loss": 1.2490234375,
"step": 101
},
{
"epoch": 0.32075471698113206,
"grad_norm": 0.0939742848277092,
"learning_rate": 7.996388332556735e-06,
"loss": 1.19482421875,
"step": 102
},
{
"epoch": 0.3238993710691824,
"grad_norm": 0.19658835232257843,
"learning_rate": 7.955405662643384e-06,
"loss": 1.249267578125,
"step": 103
},
{
"epoch": 0.3270440251572327,
"grad_norm": 0.08333238214254379,
"learning_rate": 7.914115516428751e-06,
"loss": 1.2451171875,
"step": 104
},
{
"epoch": 0.330188679245283,
"grad_norm": 0.08837057650089264,
"learning_rate": 7.872522189682318e-06,
"loss": 1.27587890625,
"step": 105
},
{
"epoch": 0.3333333333333333,
"grad_norm": 0.09254736453294754,
"learning_rate": 7.830630009716038e-06,
"loss": 1.1962890625,
"step": 106
},
{
"epoch": 0.33647798742138363,
"grad_norm": 0.10213299095630646,
"learning_rate": 7.788443334934148e-06,
"loss": 1.24658203125,
"step": 107
},
{
"epoch": 0.33962264150943394,
"grad_norm": 0.08779237419366837,
"learning_rate": 7.745966554379708e-06,
"loss": 1.24365234375,
"step": 108
},
{
"epoch": 0.34276729559748426,
"grad_norm": 0.09156819432973862,
"learning_rate": 7.703204087277989e-06,
"loss": 1.26806640625,
"step": 109
},
{
"epoch": 0.34591194968553457,
"grad_norm": 0.0824931263923645,
"learning_rate": 7.660160382576683e-06,
"loss": 1.237060546875,
"step": 110
},
{
"epoch": 0.3490566037735849,
"grad_norm": 0.0844307392835617,
"learning_rate": 7.616839918483061e-06,
"loss": 1.21240234375,
"step": 111
},
{
"epoch": 0.3522012578616352,
"grad_norm": 0.0942101925611496,
"learning_rate": 7.573247201998051e-06,
"loss": 1.19580078125,
"step": 112
},
{
"epoch": 0.3553459119496855,
"grad_norm": 0.09428957849740982,
"learning_rate": 7.529386768447342e-06,
"loss": 1.212890625,
"step": 113
},
{
"epoch": 0.3584905660377358,
"grad_norm": 0.10116898268461227,
"learning_rate": 7.485263181009539e-06,
"loss": 1.2587890625,
"step": 114
},
{
"epoch": 0.36163522012578614,
"grad_norm": 0.08661499619483948,
"learning_rate": 7.440881030241407e-06,
"loss": 1.26025390625,
"step": 115
},
{
"epoch": 0.36477987421383645,
"grad_norm": 0.0860162153840065,
"learning_rate": 7.396244933600285e-06,
"loss": 1.236328125,
"step": 116
},
{
"epoch": 0.36792452830188677,
"grad_norm": 0.0877000018954277,
"learning_rate": 7.351359534963684e-06,
"loss": 1.26806640625,
"step": 117
},
{
"epoch": 0.3710691823899371,
"grad_norm": 0.08836546540260315,
"learning_rate": 7.306229504146154e-06,
"loss": 1.240234375,
"step": 118
},
{
"epoch": 0.3742138364779874,
"grad_norm": 0.0874360054731369,
"learning_rate": 7.260859536413429e-06,
"loss": 1.25146484375,
"step": 119
},
{
"epoch": 0.37735849056603776,
"grad_norm": 0.09354016184806824,
"learning_rate": 7.215254351993957e-06,
"loss": 1.267578125,
"step": 120
},
{
"epoch": 0.3805031446540881,
"grad_norm": 0.0862109363079071,
"learning_rate": 7.169418695587791e-06,
"loss": 1.191650390625,
"step": 121
},
{
"epoch": 0.3836477987421384,
"grad_norm": 0.08533702045679092,
"learning_rate": 7.123357335872982e-06,
"loss": 1.214111328125,
"step": 122
},
{
"epoch": 0.3867924528301887,
"grad_norm": 0.08776259422302246,
"learning_rate": 7.0770750650094335e-06,
"loss": 1.22802734375,
"step": 123
},
{
"epoch": 0.389937106918239,
"grad_norm": 0.09998597204685211,
"learning_rate": 7.0305766981403365e-06,
"loss": 1.2080078125,
"step": 124
},
{
"epoch": 0.39308176100628933,
"grad_norm": 0.0824577659368515,
"learning_rate": 6.983867072891213e-06,
"loss": 1.23583984375,
"step": 125
},
{
"epoch": 0.39622641509433965,
"grad_norm": 0.08219949156045914,
"learning_rate": 6.936951048866616e-06,
"loss": 1.22607421875,
"step": 126
},
{
"epoch": 0.39937106918238996,
"grad_norm": 0.08929724246263504,
"learning_rate": 6.889833507144534e-06,
"loss": 1.220458984375,
"step": 127
},
{
"epoch": 0.4025157232704403,
"grad_norm": 0.0950295552611351,
"learning_rate": 6.842519349768582e-06,
"loss": 1.1494140625,
"step": 128
},
{
"epoch": 0.4056603773584906,
"grad_norm": 0.09173066914081573,
"learning_rate": 6.7950134992379935e-06,
"loss": 1.1640625,
"step": 129
},
{
"epoch": 0.4088050314465409,
"grad_norm": 0.08214280754327774,
"learning_rate": 6.747320897995493e-06,
"loss": 1.228515625,
"step": 130
},
{
"epoch": 0.4119496855345912,
"grad_norm": 0.08603708446025848,
"learning_rate": 6.699446507913083e-06,
"loss": 1.25390625,
"step": 131
},
{
"epoch": 0.41509433962264153,
"grad_norm": 0.0935707688331604,
"learning_rate": 6.651395309775837e-06,
"loss": 1.18408203125,
"step": 132
},
{
"epoch": 0.41823899371069184,
"grad_norm": 0.08640452474355698,
"learning_rate": 6.6031723027636775e-06,
"loss": 1.2451171875,
"step": 133
},
{
"epoch": 0.42138364779874216,
"grad_norm": 0.08143847435712814,
"learning_rate": 6.554782503931298e-06,
"loss": 1.24267578125,
"step": 134
},
{
"epoch": 0.42452830188679247,
"grad_norm": 0.08352352678775787,
"learning_rate": 6.5062309476861714e-06,
"loss": 1.22216796875,
"step": 135
},
{
"epoch": 0.4276729559748428,
"grad_norm": 0.08868604898452759,
"learning_rate": 6.457522685264793e-06,
"loss": 1.2099609375,
"step": 136
},
{
"epoch": 0.4308176100628931,
"grad_norm": 0.08718673884868622,
"learning_rate": 6.408662784207149e-06,
"loss": 1.2470703125,
"step": 137
},
{
"epoch": 0.4339622641509434,
"grad_norm": 0.08005645871162415,
"learning_rate": 6.359656327829498e-06,
"loss": 1.241455078125,
"step": 138
},
{
"epoch": 0.4371069182389937,
"grad_norm": 0.09011710435152054,
"learning_rate": 6.310508414695511e-06,
"loss": 1.21142578125,
"step": 139
},
{
"epoch": 0.44025157232704404,
"grad_norm": 0.10179577022790909,
"learning_rate": 6.261224158085826e-06,
"loss": 1.26611328125,
"step": 140
},
{
"epoch": 0.44339622641509435,
"grad_norm": 0.08692909777164459,
"learning_rate": 6.211808685466063e-06,
"loss": 1.23388671875,
"step": 141
},
{
"epoch": 0.44654088050314467,
"grad_norm": 0.09201560169458389,
"learning_rate": 6.162267137953374e-06,
"loss": 1.24853515625,
"step": 142
},
{
"epoch": 0.449685534591195,
"grad_norm": 0.08598781377077103,
"learning_rate": 6.112604669781572e-06,
"loss": 1.2255859375,
"step": 143
},
{
"epoch": 0.4528301886792453,
"grad_norm": 0.08363979309797287,
"learning_rate": 6.062826447764883e-06,
"loss": 1.217041015625,
"step": 144
},
{
"epoch": 0.4559748427672956,
"grad_norm": 0.0804029256105423,
"learning_rate": 6.012937650760406e-06,
"loss": 1.2509765625,
"step": 145
},
{
"epoch": 0.4591194968553459,
"grad_norm": 0.08158931881189346,
"learning_rate": 5.962943469129303e-06,
"loss": 1.234375,
"step": 146
},
{
"epoch": 0.46226415094339623,
"grad_norm": 0.08835593611001968,
"learning_rate": 5.91284910419681e-06,
"loss": 1.27587890625,
"step": 147
},
{
"epoch": 0.46540880503144655,
"grad_norm": 0.10376080870628357,
"learning_rate": 5.862659767711094e-06,
"loss": 1.1826171875,
"step": 148
},
{
"epoch": 0.46855345911949686,
"grad_norm": 0.09770874679088593,
"learning_rate": 5.812380681301031e-06,
"loss": 1.1572265625,
"step": 149
},
{
"epoch": 0.4716981132075472,
"grad_norm": 0.10530182719230652,
"learning_rate": 5.762017075932952e-06,
"loss": 1.2587890625,
"step": 150
},
{
"epoch": 0.4748427672955975,
"grad_norm": 0.08281303942203522,
"learning_rate": 5.711574191366427e-06,
"loss": 1.1708984375,
"step": 151
},
{
"epoch": 0.4779874213836478,
"grad_norm": 0.08968009054660797,
"learning_rate": 5.66105727560912e-06,
"loss": 1.269775390625,
"step": 152
},
{
"epoch": 0.4811320754716981,
"grad_norm": 0.07928328961133957,
"learning_rate": 5.6104715843708e-06,
"loss": 1.21484375,
"step": 153
},
{
"epoch": 0.48427672955974843,
"grad_norm": 0.08762849122285843,
"learning_rate": 5.559822380516539e-06,
"loss": 1.220458984375,
"step": 154
},
{
"epoch": 0.48742138364779874,
"grad_norm": 0.08722134679555893,
"learning_rate": 5.509114933519179e-06,
"loss": 1.25634765625,
"step": 155
},
{
"epoch": 0.49056603773584906,
"grad_norm": 0.0900566354393959,
"learning_rate": 5.458354518911086e-06,
"loss": 1.236328125,
"step": 156
},
{
"epoch": 0.4937106918238994,
"grad_norm": 0.08496430516242981,
"learning_rate": 5.4075464177353165e-06,
"loss": 1.21630859375,
"step": 157
},
{
"epoch": 0.4968553459119497,
"grad_norm": 0.08240823447704315,
"learning_rate": 5.356695915996162e-06,
"loss": 1.22607421875,
"step": 158
},
{
"epoch": 0.5,
"grad_norm": 0.08470958471298218,
"learning_rate": 5.3058083041092145e-06,
"loss": 1.22705078125,
"step": 159
},
{
"epoch": 0.5031446540880503,
"grad_norm": 0.08934212476015091,
"learning_rate": 5.25488887635095e-06,
"loss": 1.285400390625,
"step": 160
},
{
"epoch": 0.5062893081761006,
"grad_norm": 0.09710770845413208,
"learning_rate": 5.2039429303079294e-06,
"loss": 1.232421875,
"step": 161
},
{
"epoch": 0.5094339622641509,
"grad_norm": 0.08722967654466629,
"learning_rate": 5.152975766325631e-06,
"loss": 1.263427734375,
"step": 162
},
{
"epoch": 0.5125786163522013,
"grad_norm": 0.07902610301971436,
"learning_rate": 5.101992686957028e-06,
"loss": 1.25732421875,
"step": 163
},
{
"epoch": 0.5157232704402516,
"grad_norm": 0.08694876730442047,
"learning_rate": 5.050998996410899e-06,
"loss": 1.298095703125,
"step": 164
},
{
"epoch": 0.5188679245283019,
"grad_norm": 0.08544755727052689,
"learning_rate": 5e-06,
"loss": 1.208984375,
"step": 165
},
{
"epoch": 0.5220125786163522,
"grad_norm": 0.0834759846329689,
"learning_rate": 4.949001003589102e-06,
"loss": 1.23779296875,
"step": 166
},
{
"epoch": 0.5251572327044025,
"grad_norm": 0.08612226694822311,
"learning_rate": 4.898007313042975e-06,
"loss": 1.283935546875,
"step": 167
},
{
"epoch": 0.5283018867924528,
"grad_norm": 0.09189831465482712,
"learning_rate": 4.84702423367437e-06,
"loss": 1.17431640625,
"step": 168
},
{
"epoch": 0.5314465408805031,
"grad_norm": 0.09417322278022766,
"learning_rate": 4.796057069692073e-06,
"loss": 1.21875,
"step": 169
},
{
"epoch": 0.5345911949685535,
"grad_norm": 0.0859440341591835,
"learning_rate": 4.74511112364905e-06,
"loss": 1.237548828125,
"step": 170
},
{
"epoch": 0.5377358490566038,
"grad_norm": 0.09315012395381927,
"learning_rate": 4.694191695890788e-06,
"loss": 1.21923828125,
"step": 171
},
{
"epoch": 0.5408805031446541,
"grad_norm": 0.08770447224378586,
"learning_rate": 4.643304084003839e-06,
"loss": 1.22705078125,
"step": 172
},
{
"epoch": 0.5440251572327044,
"grad_norm": 0.08003748953342438,
"learning_rate": 4.592453582264684e-06,
"loss": 1.28173828125,
"step": 173
},
{
"epoch": 0.5471698113207547,
"grad_norm": 0.09152880311012268,
"learning_rate": 4.541645481088914e-06,
"loss": 1.24169921875,
"step": 174
},
{
"epoch": 0.550314465408805,
"grad_norm": 0.07619733363389969,
"learning_rate": 4.4908850664808245e-06,
"loss": 1.2568359375,
"step": 175
},
{
"epoch": 0.5534591194968553,
"grad_norm": 0.08338797092437744,
"learning_rate": 4.4401776194834615e-06,
"loss": 1.16650390625,
"step": 176
},
{
"epoch": 0.5566037735849056,
"grad_norm": 0.07812988758087158,
"learning_rate": 4.389528415629201e-06,
"loss": 1.24755859375,
"step": 177
},
{
"epoch": 0.559748427672956,
"grad_norm": 0.0926586166024208,
"learning_rate": 4.33894272439088e-06,
"loss": 1.214599609375,
"step": 178
},
{
"epoch": 0.5628930817610063,
"grad_norm": 0.10608647018671036,
"learning_rate": 4.2884258086335755e-06,
"loss": 1.16162109375,
"step": 179
},
{
"epoch": 0.5660377358490566,
"grad_norm": 0.0896821841597557,
"learning_rate": 4.23798292406705e-06,
"loss": 1.21923828125,
"step": 180
},
{
"epoch": 0.5691823899371069,
"grad_norm": 0.10245795547962189,
"learning_rate": 4.187619318698971e-06,
"loss": 1.207763671875,
"step": 181
},
{
"epoch": 0.5723270440251572,
"grad_norm": 0.0842718854546547,
"learning_rate": 4.137340232288908e-06,
"loss": 1.30224609375,
"step": 182
},
{
"epoch": 0.5754716981132075,
"grad_norm": 0.09193960577249527,
"learning_rate": 4.087150895803192e-06,
"loss": 1.218994140625,
"step": 183
},
{
"epoch": 0.5786163522012578,
"grad_norm": 0.08665217459201813,
"learning_rate": 4.0370565308706986e-06,
"loss": 1.1982421875,
"step": 184
},
{
"epoch": 0.5817610062893082,
"grad_norm": 0.0908685177564621,
"learning_rate": 3.987062349239596e-06,
"loss": 1.21240234375,
"step": 185
},
{
"epoch": 0.5849056603773585,
"grad_norm": 0.08689988404512405,
"learning_rate": 3.937173552235117e-06,
"loss": 1.296875,
"step": 186
},
{
"epoch": 0.5880503144654088,
"grad_norm": 0.11394862085580826,
"learning_rate": 3.887395330218429e-06,
"loss": 1.20361328125,
"step": 187
},
{
"epoch": 0.5911949685534591,
"grad_norm": 0.0949663296341896,
"learning_rate": 3.837732862046627e-06,
"loss": 1.263671875,
"step": 188
},
{
"epoch": 0.5943396226415094,
"grad_norm": 0.08149112015962601,
"learning_rate": 3.7881913145339387e-06,
"loss": 1.25390625,
"step": 189
},
{
"epoch": 0.5974842767295597,
"grad_norm": 0.09439687430858612,
"learning_rate": 3.738775841914175e-06,
"loss": 1.2255859375,
"step": 190
},
{
"epoch": 0.60062893081761,
"grad_norm": 0.0807260051369667,
"learning_rate": 3.689491585304491e-06,
"loss": 1.22998046875,
"step": 191
},
{
"epoch": 0.6037735849056604,
"grad_norm": 0.08422300219535828,
"learning_rate": 3.640343672170503e-06,
"loss": 1.23828125,
"step": 192
},
{
"epoch": 0.6069182389937107,
"grad_norm": 0.08983633667230606,
"learning_rate": 3.5913372157928515e-06,
"loss": 1.21337890625,
"step": 193
},
{
"epoch": 0.610062893081761,
"grad_norm": 0.09498164057731628,
"learning_rate": 3.5424773147352085e-06,
"loss": 1.17724609375,
"step": 194
},
{
"epoch": 0.6132075471698113,
"grad_norm": 0.0914149209856987,
"learning_rate": 3.4937690523138302e-06,
"loss": 1.232421875,
"step": 195
},
{
"epoch": 0.6163522012578616,
"grad_norm": 0.0758446678519249,
"learning_rate": 3.4452174960687033e-06,
"loss": 1.1982421875,
"step": 196
},
{
"epoch": 0.6194968553459119,
"grad_norm": 0.10985689610242844,
"learning_rate": 3.3968276972363224e-06,
"loss": 1.2041015625,
"step": 197
},
{
"epoch": 0.6226415094339622,
"grad_norm": 0.09818878769874573,
"learning_rate": 3.3486046902241663e-06,
"loss": 1.19287109375,
"step": 198
},
{
"epoch": 0.6257861635220126,
"grad_norm": 0.09397387504577637,
"learning_rate": 3.3005534920869175e-06,
"loss": 1.267578125,
"step": 199
},
{
"epoch": 0.6289308176100629,
"grad_norm": 0.09300831705331802,
"learning_rate": 3.252679102004509e-06,
"loss": 1.24462890625,
"step": 200
},
{
"epoch": 0.6320754716981132,
"grad_norm": 0.08829600363969803,
"learning_rate": 3.204986500762006e-06,
"loss": 1.25390625,
"step": 201
},
{
"epoch": 0.6352201257861635,
"grad_norm": 0.0840512067079544,
"learning_rate": 3.1574806502314206e-06,
"loss": 1.11474609375,
"step": 202
},
{
"epoch": 0.6383647798742138,
"grad_norm": 0.08670662343502045,
"learning_rate": 3.110166492855468e-06,
"loss": 1.203857421875,
"step": 203
},
{
"epoch": 0.6415094339622641,
"grad_norm": 0.08876918256282806,
"learning_rate": 3.063048951133386e-06,
"loss": 1.237060546875,
"step": 204
},
{
"epoch": 0.6446540880503144,
"grad_norm": 0.0912042111158371,
"learning_rate": 3.016132927108787e-06,
"loss": 1.20458984375,
"step": 205
},
{
"epoch": 0.6477987421383647,
"grad_norm": 0.09555654227733612,
"learning_rate": 2.9694233018596665e-06,
"loss": 1.240478515625,
"step": 206
},
{
"epoch": 0.6509433962264151,
"grad_norm": 0.10727142542600632,
"learning_rate": 2.9229249349905686e-06,
"loss": 1.22119140625,
"step": 207
},
{
"epoch": 0.6540880503144654,
"grad_norm": 0.08861692994832993,
"learning_rate": 2.8766426641270197e-06,
"loss": 1.1953125,
"step": 208
},
{
"epoch": 0.6572327044025157,
"grad_norm": 0.09431105107069016,
"learning_rate": 2.83058130441221e-06,
"loss": 1.26220703125,
"step": 209
},
{
"epoch": 0.660377358490566,
"grad_norm": 0.08846162259578705,
"learning_rate": 2.7847456480060476e-06,
"loss": 1.22021484375,
"step": 210
},
{
"epoch": 0.6635220125786163,
"grad_norm": 0.07520295679569244,
"learning_rate": 2.7391404635865725e-06,
"loss": 1.20068359375,
"step": 211
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.08306454122066498,
"learning_rate": 2.6937704958538483e-06,
"loss": 1.154296875,
"step": 212
},
{
"epoch": 0.6698113207547169,
"grad_norm": 0.08136342465877533,
"learning_rate": 2.648640465036316e-06,
"loss": 1.1875,
"step": 213
},
{
"epoch": 0.6729559748427673,
"grad_norm": 0.0894201323390007,
"learning_rate": 2.603755066399718e-06,
"loss": 1.272705078125,
"step": 214
},
{
"epoch": 0.6761006289308176,
"grad_norm": 0.10022783279418945,
"learning_rate": 2.559118969758595e-06,
"loss": 1.2265625,
"step": 215
},
{
"epoch": 0.6792452830188679,
"grad_norm": 0.0807405486702919,
"learning_rate": 2.514736818990463e-06,
"loss": 1.18994140625,
"step": 216
},
{
"epoch": 0.6823899371069182,
"grad_norm": 0.08904239535331726,
"learning_rate": 2.470613231552661e-06,
"loss": 1.11181640625,
"step": 217
},
{
"epoch": 0.6855345911949685,
"grad_norm": 0.08609310537576675,
"learning_rate": 2.4267527980019523e-06,
"loss": 1.216796875,
"step": 218
},
{
"epoch": 0.6886792452830188,
"grad_norm": 0.08292370289564133,
"learning_rate": 2.383160081516941e-06,
"loss": 1.2783203125,
"step": 219
},
{
"epoch": 0.6918238993710691,
"grad_norm": 0.10215222090482712,
"learning_rate": 2.339839617423318e-06,
"loss": 1.248046875,
"step": 220
},
{
"epoch": 0.6949685534591195,
"grad_norm": 0.08919369429349899,
"learning_rate": 2.296795912722014e-06,
"loss": 1.26953125,
"step": 221
},
{
"epoch": 0.6981132075471698,
"grad_norm": 0.09696005284786224,
"learning_rate": 2.254033445620293e-06,
"loss": 1.19677734375,
"step": 222
},
{
"epoch": 0.7012578616352201,
"grad_norm": 0.09767638891935349,
"learning_rate": 2.211556665065854e-06,
"loss": 1.23193359375,
"step": 223
},
{
"epoch": 0.7044025157232704,
"grad_norm": 0.08746252954006195,
"learning_rate": 2.169369990283963e-06,
"loss": 1.228515625,
"step": 224
},
{
"epoch": 0.7075471698113207,
"grad_norm": 0.0967244878411293,
"learning_rate": 2.1274778103176854e-06,
"loss": 1.240234375,
"step": 225
},
{
"epoch": 0.710691823899371,
"grad_norm": 0.08571658283472061,
"learning_rate": 2.08588448357125e-06,
"loss": 1.265625,
"step": 226
},
{
"epoch": 0.7138364779874213,
"grad_norm": 0.10917211323976517,
"learning_rate": 2.0445943373566178e-06,
"loss": 1.21875,
"step": 227
},
{
"epoch": 0.7169811320754716,
"grad_norm": 0.08573350310325623,
"learning_rate": 2.0036116674432653e-06,
"loss": 1.181884765625,
"step": 228
},
{
"epoch": 0.720125786163522,
"grad_norm": 0.09475314617156982,
"learning_rate": 1.962940737611264e-06,
"loss": 1.20849609375,
"step": 229
},
{
"epoch": 0.7232704402515723,
"grad_norm": 0.08965165168046951,
"learning_rate": 1.922585779207674e-06,
"loss": 1.17822265625,
"step": 230
},
{
"epoch": 0.7264150943396226,
"grad_norm": 0.08729615807533264,
"learning_rate": 1.8825509907063328e-06,
"loss": 1.27294921875,
"step": 231
},
{
"epoch": 0.7295597484276729,
"grad_norm": 0.09284593164920807,
"learning_rate": 1.8428405372710446e-06,
"loss": 1.2294921875,
"step": 232
},
{
"epoch": 0.7327044025157232,
"grad_norm": 0.0874868854880333,
"learning_rate": 1.8034585503222441e-06,
"loss": 1.11474609375,
"step": 233
},
{
"epoch": 0.7358490566037735,
"grad_norm": 0.08022027462720871,
"learning_rate": 1.7644091271071645e-06,
"loss": 1.214111328125,
"step": 234
},
{
"epoch": 0.7389937106918238,
"grad_norm": 0.0947025790810585,
"learning_rate": 1.7256963302735752e-06,
"loss": 1.19580078125,
"step": 235
},
{
"epoch": 0.7421383647798742,
"grad_norm": 0.07570719718933105,
"learning_rate": 1.687324187447102e-06,
"loss": 1.265625,
"step": 236
},
{
"epoch": 0.7452830188679245,
"grad_norm": 0.08627166599035263,
"learning_rate": 1.6492966908122033e-06,
"loss": 1.240234375,
"step": 237
},
{
"epoch": 0.7484276729559748,
"grad_norm": 0.09424252808094025,
"learning_rate": 1.611617796696821e-06,
"loss": 1.22412109375,
"step": 238
},
{
"epoch": 0.7515723270440252,
"grad_norm": 0.08505000174045563,
"learning_rate": 1.5742914251607794e-06,
"loss": 1.22509765625,
"step": 239
},
{
"epoch": 0.7547169811320755,
"grad_norm": 0.0981978178024292,
"learning_rate": 1.5373214595879416e-06,
"loss": 1.228515625,
"step": 240
},
{
"epoch": 0.7578616352201258,
"grad_norm": 0.08118585497140884,
"learning_rate": 1.500711746282192e-06,
"loss": 1.25537109375,
"step": 241
},
{
"epoch": 0.7610062893081762,
"grad_norm": 0.08585264533758163,
"learning_rate": 1.4644660940672628e-06,
"loss": 1.310791015625,
"step": 242
},
{
"epoch": 0.7641509433962265,
"grad_norm": 0.08260893076658249,
"learning_rate": 1.4285882738904822e-06,
"loss": 1.175048828125,
"step": 243
},
{
"epoch": 0.7672955974842768,
"grad_norm": 0.09124623984098434,
"learning_rate": 1.3930820184304423e-06,
"loss": 1.22314453125,
"step": 244
},
{
"epoch": 0.7704402515723271,
"grad_norm": 0.0838477611541748,
"learning_rate": 1.357951021708655e-06,
"loss": 1.238525390625,
"step": 245
},
{
"epoch": 0.7735849056603774,
"grad_norm": 0.10469858348369598,
"learning_rate": 1.323198938705238e-06,
"loss": 1.23486328125,
"step": 246
},
{
"epoch": 0.7767295597484277,
"grad_norm": 0.09047719091176987,
"learning_rate": 1.2888293849786503e-06,
"loss": 1.189697265625,
"step": 247
},
{
"epoch": 0.779874213836478,
"grad_norm": 0.09436509758234024,
"learning_rate": 1.2548459362895377e-06,
"loss": 1.19091796875,
"step": 248
},
{
"epoch": 0.7830188679245284,
"grad_norm": 0.09370195865631104,
"learning_rate": 1.2212521282287093e-06,
"loss": 1.20654296875,
"step": 249
},
{
"epoch": 0.7861635220125787,
"grad_norm": 0.09401333332061768,
"learning_rate": 1.188051455849309e-06,
"loss": 1.1904296875,
"step": 250
},
{
"epoch": 0.789308176100629,
"grad_norm": 0.08302223682403564,
"learning_rate": 1.1552473733031893e-06,
"loss": 1.224853515625,
"step": 251
},
{
"epoch": 0.7924528301886793,
"grad_norm": 0.08327268809080124,
"learning_rate": 1.1228432934815487e-06,
"loss": 1.2490234375,
"step": 252
},
{
"epoch": 0.7955974842767296,
"grad_norm": 0.09186363220214844,
"learning_rate": 1.0908425876598512e-06,
"loss": 1.2255859375,
"step": 253
},
{
"epoch": 0.7987421383647799,
"grad_norm": 0.10745052248239517,
"learning_rate": 1.0592485851470973e-06,
"loss": 1.23876953125,
"step": 254
},
{
"epoch": 0.8018867924528302,
"grad_norm": 0.09070909023284912,
"learning_rate": 1.0280645729394368e-06,
"loss": 1.21923828125,
"step": 255
},
{
"epoch": 0.8050314465408805,
"grad_norm": 0.09789802879095078,
"learning_rate": 9.972937953781985e-07,
"loss": 1.210693359375,
"step": 256
},
{
"epoch": 0.8081761006289309,
"grad_norm": 0.09813636541366577,
"learning_rate": 9.6693945381235e-07,
"loss": 1.248046875,
"step": 257
},
{
"epoch": 0.8113207547169812,
"grad_norm": 0.0862487405538559,
"learning_rate": 9.370047062654386e-07,
"loss": 1.1884765625,
"step": 258
},
{
"epoch": 0.8144654088050315,
"grad_norm": 0.08022520691156387,
"learning_rate": 9.074926671070322e-07,
"loss": 1.191650390625,
"step": 259
},
{
"epoch": 0.8176100628930818,
"grad_norm": 0.08337897062301636,
"learning_rate": 8.784064067287057e-07,
"loss": 1.1689453125,
"step": 260
},
{
"epoch": 0.8207547169811321,
"grad_norm": 0.09539865702390671,
"learning_rate": 8.497489512245971e-07,
"loss": 1.1767578125,
"step": 261
},
{
"epoch": 0.8238993710691824,
"grad_norm": 0.09431402385234833,
"learning_rate": 8.215232820765851e-07,
"loss": 1.2138671875,
"step": 262
},
{
"epoch": 0.8270440251572327,
"grad_norm": 0.10074929147958755,
"learning_rate": 7.937323358440935e-07,
"loss": 1.26904296875,
"step": 263
},
{
"epoch": 0.8301886792452831,
"grad_norm": 0.08571036159992218,
"learning_rate": 7.663790038585794e-07,
"loss": 1.23046875,
"step": 264
},
{
"epoch": 0.8333333333333334,
"grad_norm": 0.08719199895858765,
"learning_rate": 7.394661319227175e-07,
"loss": 1.23876953125,
"step": 265
},
{
"epoch": 0.8364779874213837,
"grad_norm": 0.08014900237321854,
"learning_rate": 7.129965200143335e-07,
"loss": 1.227294921875,
"step": 266
},
{
"epoch": 0.839622641509434,
"grad_norm": 0.1006547287106514,
"learning_rate": 6.86972921995096e-07,
"loss": 1.232421875,
"step": 267
},
{
"epoch": 0.8427672955974843,
"grad_norm": 0.09184549003839493,
"learning_rate": 6.613980453240065e-07,
"loss": 1.17724609375,
"step": 268
},
{
"epoch": 0.8459119496855346,
"grad_norm": 0.08853388577699661,
"learning_rate": 6.36274550775719e-07,
"loss": 1.205322265625,
"step": 269
},
{
"epoch": 0.8490566037735849,
"grad_norm": 0.08986873179674149,
"learning_rate": 6.116050521637218e-07,
"loss": 1.197265625,
"step": 270
},
{
"epoch": 0.8522012578616353,
"grad_norm": 0.09413354843854904,
"learning_rate": 5.873921160683943e-07,
"loss": 1.21044921875,
"step": 271
},
{
"epoch": 0.8553459119496856,
"grad_norm": 0.10209821909666061,
"learning_rate": 5.636382615699842e-07,
"loss": 1.228759765625,
"step": 272
},
{
"epoch": 0.8584905660377359,
"grad_norm": 0.09260933846235275,
"learning_rate": 5.403459599865307e-07,
"loss": 1.19091796875,
"step": 273
},
{
"epoch": 0.8616352201257862,
"grad_norm": 0.0983046144247055,
"learning_rate": 5.175176346167465e-07,
"loss": 1.22705078125,
"step": 274
},
{
"epoch": 0.8647798742138365,
"grad_norm": 0.08732131868600845,
"learning_rate": 4.951556604879049e-07,
"loss": 1.2109375,
"step": 275
},
{
"epoch": 0.8679245283018868,
"grad_norm": 0.09008695185184479,
"learning_rate": 4.732623641087403e-07,
"loss": 1.22607421875,
"step": 276
},
{
"epoch": 0.8710691823899371,
"grad_norm": 0.09183256328105927,
"learning_rate": 4.5184002322740784e-07,
"loss": 1.27099609375,
"step": 277
},
{
"epoch": 0.8742138364779874,
"grad_norm": 0.10036151111125946,
"learning_rate": 4.3089086659450774e-07,
"loss": 1.24609375,
"step": 278
},
{
"epoch": 0.8773584905660378,
"grad_norm": 0.08253729343414307,
"learning_rate": 4.1041707373120354e-07,
"loss": 1.240234375,
"step": 279
},
{
"epoch": 0.8805031446540881,
"grad_norm": 0.08661764860153198,
"learning_rate": 3.9042077470247574e-07,
"loss": 1.2265625,
"step": 280
},
{
"epoch": 0.8836477987421384,
"grad_norm": 0.11636807769536972,
"learning_rate": 3.709040498955102e-07,
"loss": 1.21630859375,
"step": 281
},
{
"epoch": 0.8867924528301887,
"grad_norm": 0.09158287197351456,
"learning_rate": 3.518689298032524e-07,
"loss": 1.1962890625,
"step": 282
},
{
"epoch": 0.889937106918239,
"grad_norm": 0.08876530826091766,
"learning_rate": 3.333173948131663e-07,
"loss": 1.27099609375,
"step": 283
},
{
"epoch": 0.8930817610062893,
"grad_norm": 0.08798261731863022,
"learning_rate": 3.1525137500119207e-07,
"loss": 1.24951171875,
"step": 284
},
{
"epoch": 0.8962264150943396,
"grad_norm": 0.08027750253677368,
"learning_rate": 2.9767274993094285e-07,
"loss": 1.24560546875,
"step": 285
},
{
"epoch": 0.89937106918239,
"grad_norm": 0.09272100776433945,
"learning_rate": 2.8058334845816214e-07,
"loss": 1.199462890625,
"step": 286
},
{
"epoch": 0.9025157232704403,
"grad_norm": 0.10138490796089172,
"learning_rate": 2.6398494854045055e-07,
"loss": 1.2041015625,
"step": 287
},
{
"epoch": 0.9056603773584906,
"grad_norm": 0.10046400874853134,
"learning_rate": 2.478792770522842e-07,
"loss": 1.17041015625,
"step": 288
},
{
"epoch": 0.9088050314465409,
"grad_norm": 0.08769627660512924,
"learning_rate": 2.32268009605362e-07,
"loss": 1.277587890625,
"step": 289
},
{
"epoch": 0.9119496855345912,
"grad_norm": 0.08884772658348083,
"learning_rate": 2.171527703742715e-07,
"loss": 1.254638671875,
"step": 290
},
{
"epoch": 0.9150943396226415,
"grad_norm": 0.07322319597005844,
"learning_rate": 2.0253513192751374e-07,
"loss": 1.20361328125,
"step": 291
},
{
"epoch": 0.9182389937106918,
"grad_norm": 0.08685608953237534,
"learning_rate": 1.884166150638933e-07,
"loss": 1.252685546875,
"step": 292
},
{
"epoch": 0.9213836477987422,
"grad_norm": 0.08980607241392136,
"learning_rate": 1.7479868865430072e-07,
"loss": 1.177734375,
"step": 293
},
{
"epoch": 0.9245283018867925,
"grad_norm": 0.0856226310133934,
"learning_rate": 1.6168276948889007e-07,
"loss": 1.2509765625,
"step": 294
},
{
"epoch": 0.9276729559748428,
"grad_norm": 0.0801997110247612,
"learning_rate": 1.4907022212967803e-07,
"loss": 1.2158203125,
"step": 295
},
{
"epoch": 0.9308176100628931,
"grad_norm": 0.08406123518943787,
"learning_rate": 1.3696235876857812e-07,
"loss": 1.2265625,
"step": 296
},
{
"epoch": 0.9339622641509434,
"grad_norm": 0.08864335715770721,
"learning_rate": 1.253604390908819e-07,
"loss": 1.2197265625,
"step": 297
},
{
"epoch": 0.9371069182389937,
"grad_norm": 0.10364779084920883,
"learning_rate": 1.1426567014420297e-07,
"loss": 1.198486328125,
"step": 298
},
{
"epoch": 0.940251572327044,
"grad_norm": 0.0869453102350235,
"learning_rate": 1.0367920621289496e-07,
"loss": 1.20751953125,
"step": 299
},
{
"epoch": 0.9433962264150944,
"grad_norm": 0.08727333694696426,
"learning_rate": 9.360214869796492e-08,
"loss": 1.207275390625,
"step": 300
},
{
"epoch": 0.9465408805031447,
"grad_norm": 0.09867426753044128,
"learning_rate": 8.403554600248498e-08,
"loss": 1.198974609375,
"step": 301
},
{
"epoch": 0.949685534591195,
"grad_norm": 0.10300202667713165,
"learning_rate": 7.498039342251573e-08,
"loss": 1.2158203125,
"step": 302
},
{
"epoch": 0.9528301886792453,
"grad_norm": 0.09961962699890137,
"learning_rate": 6.643763304355566e-08,
"loss": 1.24267578125,
"step": 303
},
{
"epoch": 0.9559748427672956,
"grad_norm": 0.08277103304862976,
"learning_rate": 5.8408153642533493e-08,
"loss": 1.27392578125,
"step": 304
},
{
"epoch": 0.9591194968553459,
"grad_norm": 0.09768108278512955,
"learning_rate": 5.089279059533658e-08,
"loss": 1.2109375,
"step": 305
},
{
"epoch": 0.9622641509433962,
"grad_norm": 0.08892697095870972,
"learning_rate": 4.389232578989988e-08,
"loss": 1.17431640625,
"step": 306
},
{
"epoch": 0.9654088050314465,
"grad_norm": 0.08239562809467316,
"learning_rate": 3.7407487544861565e-08,
"loss": 1.228759765625,
"step": 307
},
{
"epoch": 0.9685534591194969,
"grad_norm": 0.09559322148561478,
"learning_rate": 3.143895053378698e-08,
"loss": 1.22509765625,
"step": 308
},
{
"epoch": 0.9716981132075472,
"grad_norm": 0.09237507730722427,
"learning_rate": 2.59873357149798e-08,
"loss": 1.183837890625,
"step": 309
},
{
"epoch": 0.9748427672955975,
"grad_norm": 0.08496708422899246,
"learning_rate": 2.1053210266875346e-08,
"loss": 1.21435546875,
"step": 310
},
{
"epoch": 0.9779874213836478,
"grad_norm": 0.0905926525592804,
"learning_rate": 1.6637087529033925e-08,
"loss": 1.20458984375,
"step": 311
},
{
"epoch": 0.9811320754716981,
"grad_norm": 0.09607579559087753,
"learning_rate": 1.2739426948732426e-08,
"loss": 1.19677734375,
"step": 312
},
{
"epoch": 0.9842767295597484,
"grad_norm": 0.08372277021408081,
"learning_rate": 9.36063403316534e-09,
"loss": 1.118896484375,
"step": 313
},
{
"epoch": 0.9874213836477987,
"grad_norm": 0.08898620307445526,
"learning_rate": 6.501060307256835e-09,
"loss": 1.208984375,
"step": 314
},
{
"epoch": 0.9905660377358491,
"grad_norm": 0.0899147167801857,
"learning_rate": 4.161003277085574e-09,
"loss": 1.23486328125,
"step": 315
},
{
"epoch": 0.9937106918238994,
"grad_norm": 0.09119053184986115,
"learning_rate": 2.3407063989361324e-09,
"loss": 1.259521484375,
"step": 316
},
{
"epoch": 0.9968553459119497,
"grad_norm": 0.08343734592199326,
"learning_rate": 1.040359053967599e-09,
"loss": 1.231689453125,
"step": 317
},
{
"epoch": 1.0,
"grad_norm": 0.0955134704709053,
"learning_rate": 2.6009652851211044e-10,
"loss": 1.25927734375,
"step": 318
},
{
"epoch": 1.0,
"eval_loss": 1.35986328125,
"eval_runtime": 10.9098,
"eval_samples_per_second": 0.55,
"eval_steps_per_second": 0.092,
"step": 318
}
],
"logging_steps": 1.0,
"max_steps": 318,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"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": 8.935569965579239e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}