gemma-2b-trader-summarizer / trainer_state.json
Wezenite's picture
Upload folder using huggingface_hub
c474f27 verified
Invalid JSON: Unexpected token 'N', ..."ad_norm": NaN, "... is not valid JSON
{
"best_global_step": 1500,
"best_metric": 0.4265768229961395,
"best_model_checkpoint": "/content/drive/MyDrive/gemma2_trading_finetune/gemma-2b-trader-summarizer/checkpoint-1500",
"epoch": 3.0,
"eval_steps": 500,
"global_step": 1791,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.001675392670157068,
"grad_norm": NaN,
"learning_rate": 0.0,
"loss": 1.2713,
"step": 1
},
{
"epoch": 0.016753926701570682,
"grad_norm": 0.6294093728065491,
"learning_rate": 1.4000000000000001e-06,
"loss": 0.9665,
"step": 10
},
{
"epoch": 0.033507853403141365,
"grad_norm": 0.5912520885467529,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.8862,
"step": 20
},
{
"epoch": 0.050261780104712044,
"grad_norm": 0.5498993396759033,
"learning_rate": 5.400000000000001e-06,
"loss": 0.866,
"step": 30
},
{
"epoch": 0.06701570680628273,
"grad_norm": 0.47078797221183777,
"learning_rate": 7.4e-06,
"loss": 0.7701,
"step": 40
},
{
"epoch": 0.08376963350785341,
"grad_norm": 0.44182780385017395,
"learning_rate": 9.4e-06,
"loss": 0.7534,
"step": 50
},
{
"epoch": 0.10052356020942409,
"grad_norm": 0.4943735897541046,
"learning_rate": 1.14e-05,
"loss": 0.6793,
"step": 60
},
{
"epoch": 0.11727748691099477,
"grad_norm": 0.4289020001888275,
"learning_rate": 1.3400000000000002e-05,
"loss": 0.6589,
"step": 70
},
{
"epoch": 0.13403141361256546,
"grad_norm": 0.3783618211746216,
"learning_rate": 1.54e-05,
"loss": 0.5734,
"step": 80
},
{
"epoch": 0.15078534031413612,
"grad_norm": 0.43305572867393494,
"learning_rate": 1.7400000000000003e-05,
"loss": 0.6123,
"step": 90
},
{
"epoch": 0.16753926701570682,
"grad_norm": 0.42297738790512085,
"learning_rate": 1.94e-05,
"loss": 0.5813,
"step": 100
},
{
"epoch": 0.18429319371727748,
"grad_norm": 0.4422590732574463,
"learning_rate": 1.9999154385359104e-05,
"loss": 0.5432,
"step": 110
},
{
"epoch": 0.20104712041884817,
"grad_norm": 0.4152766764163971,
"learning_rate": 1.9995012943638305e-05,
"loss": 0.5422,
"step": 120
},
{
"epoch": 0.21780104712041884,
"grad_norm": 0.5250847339630127,
"learning_rate": 1.9987421785468592e-05,
"loss": 0.5781,
"step": 130
},
{
"epoch": 0.23455497382198953,
"grad_norm": 0.4559013843536377,
"learning_rate": 1.997638353089092e-05,
"loss": 0.5234,
"step": 140
},
{
"epoch": 0.2513089005235602,
"grad_norm": 0.49973589181900024,
"learning_rate": 1.996190198969009e-05,
"loss": 0.5465,
"step": 150
},
{
"epoch": 0.2680628272251309,
"grad_norm": 0.4770303964614868,
"learning_rate": 1.9943982160079823e-05,
"loss": 0.5019,
"step": 160
},
{
"epoch": 0.2848167539267016,
"grad_norm": 0.5174061059951782,
"learning_rate": 1.9922630226977667e-05,
"loss": 0.5366,
"step": 170
},
{
"epoch": 0.30157068062827225,
"grad_norm": 0.5140427947044373,
"learning_rate": 1.989785355987032e-05,
"loss": 0.5331,
"step": 180
},
{
"epoch": 0.3183246073298429,
"grad_norm": 0.49744996428489685,
"learning_rate": 1.986966071027007e-05,
"loss": 0.5043,
"step": 190
},
{
"epoch": 0.33507853403141363,
"grad_norm": 0.5526700019836426,
"learning_rate": 1.9838061408763314e-05,
"loss": 0.5658,
"step": 200
},
{
"epoch": 0.3518324607329843,
"grad_norm": 0.5416470170021057,
"learning_rate": 1.9803066561652108e-05,
"loss": 0.5356,
"step": 210
},
{
"epoch": 0.36858638743455496,
"grad_norm": 0.5986471772193909,
"learning_rate": 1.9764688247189918e-05,
"loss": 0.5746,
"step": 220
},
{
"epoch": 0.38534031413612563,
"grad_norm": 0.49069204926490784,
"learning_rate": 1.972293971141288e-05,
"loss": 0.4598,
"step": 230
},
{
"epoch": 0.40209424083769635,
"grad_norm": 0.5061984658241272,
"learning_rate": 1.9677835363568024e-05,
"loss": 0.5232,
"step": 240
},
{
"epoch": 0.418848167539267,
"grad_norm": 0.6716789603233337,
"learning_rate": 1.962939077114002e-05,
"loss": 0.5627,
"step": 250
},
{
"epoch": 0.4356020942408377,
"grad_norm": 0.6077855229377747,
"learning_rate": 1.9577622654478127e-05,
"loss": 0.4972,
"step": 260
},
{
"epoch": 0.4523560209424084,
"grad_norm": 0.5625795722007751,
"learning_rate": 1.9522548881025304e-05,
"loss": 0.4538,
"step": 270
},
{
"epoch": 0.46910994764397906,
"grad_norm": 0.5840628743171692,
"learning_rate": 1.9464188459151342e-05,
"loss": 0.4707,
"step": 280
},
{
"epoch": 0.48586387434554973,
"grad_norm": 0.6204021573066711,
"learning_rate": 1.9402561531592273e-05,
"loss": 0.4776,
"step": 290
},
{
"epoch": 0.5026178010471204,
"grad_norm": 0.6092849969863892,
"learning_rate": 1.9337689368498214e-05,
"loss": 0.454,
"step": 300
},
{
"epoch": 0.5193717277486911,
"grad_norm": 0.6331911087036133,
"learning_rate": 1.92695943600921e-05,
"loss": 0.5294,
"step": 310
},
{
"epoch": 0.5361256544502618,
"grad_norm": 0.5908372402191162,
"learning_rate": 1.919830000894185e-05,
"loss": 0.4953,
"step": 320
},
{
"epoch": 0.5528795811518324,
"grad_norm": 0.6011969447135925,
"learning_rate": 1.912383092184861e-05,
"loss": 0.5049,
"step": 330
},
{
"epoch": 0.5696335078534032,
"grad_norm": 0.6391215920448303,
"learning_rate": 1.9046212801353846e-05,
"loss": 0.5501,
"step": 340
},
{
"epoch": 0.5863874345549738,
"grad_norm": 0.587267279624939,
"learning_rate": 1.8965472436868288e-05,
"loss": 0.4655,
"step": 350
},
{
"epoch": 0.6031413612565445,
"grad_norm": 0.6430619359016418,
"learning_rate": 1.888163769542573e-05,
"loss": 0.4778,
"step": 360
},
{
"epoch": 0.6198952879581152,
"grad_norm": 0.6426651477813721,
"learning_rate": 1.879473751206489e-05,
"loss": 0.4926,
"step": 370
},
{
"epoch": 0.6366492146596858,
"grad_norm": 0.5685374736785889,
"learning_rate": 1.870480187984268e-05,
"loss": 0.5288,
"step": 380
},
{
"epoch": 0.6534031413612565,
"grad_norm": 0.6300281882286072,
"learning_rate": 1.8611861839482252e-05,
"loss": 0.4617,
"step": 390
},
{
"epoch": 0.6701570680628273,
"grad_norm": 0.5899410843849182,
"learning_rate": 1.8515949468659532e-05,
"loss": 0.4669,
"step": 400
},
{
"epoch": 0.6869109947643979,
"grad_norm": 0.6726074814796448,
"learning_rate": 1.8417097870931777e-05,
"loss": 0.5024,
"step": 410
},
{
"epoch": 0.7036649214659686,
"grad_norm": 0.6307454109191895,
"learning_rate": 1.8315341164312102e-05,
"loss": 0.4651,
"step": 420
},
{
"epoch": 0.7204188481675393,
"grad_norm": 0.6004384160041809,
"learning_rate": 1.8210714469493852e-05,
"loss": 0.4795,
"step": 430
},
{
"epoch": 0.7371727748691099,
"grad_norm": 0.6285229325294495,
"learning_rate": 1.810325389772892e-05,
"loss": 0.4493,
"step": 440
},
{
"epoch": 0.7539267015706806,
"grad_norm": 0.6102107167243958,
"learning_rate": 1.7992996538364185e-05,
"loss": 0.5082,
"step": 450
},
{
"epoch": 0.7706806282722513,
"grad_norm": 0.6182498931884766,
"learning_rate": 1.7879980446040327e-05,
"loss": 0.488,
"step": 460
},
{
"epoch": 0.787434554973822,
"grad_norm": 0.6647325754165649,
"learning_rate": 1.7764244627557524e-05,
"loss": 0.4537,
"step": 470
},
{
"epoch": 0.8041884816753927,
"grad_norm": 0.588350772857666,
"learning_rate": 1.7645829028412496e-05,
"loss": 0.4487,
"step": 480
},
{
"epoch": 0.8209424083769633,
"grad_norm": 0.7462045550346375,
"learning_rate": 1.7524774519011565e-05,
"loss": 0.4463,
"step": 490
},
{
"epoch": 0.837696335078534,
"grad_norm": 0.6237714886665344,
"learning_rate": 1.740112288056448e-05,
"loss": 0.4952,
"step": 500
},
{
"epoch": 0.837696335078534,
"eval_loss": 0.46747511625289917,
"eval_runtime": 72.2811,
"eval_samples_per_second": 14.679,
"eval_steps_per_second": 7.346,
"step": 500
},
{
"epoch": 0.8544502617801047,
"grad_norm": 0.6721609830856323,
"learning_rate": 1.7274916790663914e-05,
"loss": 0.4804,
"step": 510
},
{
"epoch": 0.8712041884816754,
"grad_norm": 0.6287488341331482,
"learning_rate": 1.714619980855553e-05,
"loss": 0.4275,
"step": 520
},
{
"epoch": 0.8879581151832461,
"grad_norm": 0.6959580183029175,
"learning_rate": 1.70150163601038e-05,
"loss": 0.4713,
"step": 530
},
{
"epoch": 0.9047120418848168,
"grad_norm": 0.6396615505218506,
"learning_rate": 1.6881411722458688e-05,
"loss": 0.4553,
"step": 540
},
{
"epoch": 0.9214659685863874,
"grad_norm": 0.6596575975418091,
"learning_rate": 1.674543200842853e-05,
"loss": 0.4765,
"step": 550
},
{
"epoch": 0.9382198952879581,
"grad_norm": 0.7078624367713928,
"learning_rate": 1.660712415056448e-05,
"loss": 0.5032,
"step": 560
},
{
"epoch": 0.9549738219895288,
"grad_norm": 0.7436346411705017,
"learning_rate": 1.646653588496201e-05,
"loss": 0.5238,
"step": 570
},
{
"epoch": 0.9717277486910995,
"grad_norm": 0.5996255874633789,
"learning_rate": 1.6323715734785143e-05,
"loss": 0.4802,
"step": 580
},
{
"epoch": 0.9884816753926702,
"grad_norm": 0.7176609039306641,
"learning_rate": 1.6178712993518945e-05,
"loss": 0.4705,
"step": 590
},
{
"epoch": 1.0050261780104712,
"grad_norm": 0.714697539806366,
"learning_rate": 1.6031577707956207e-05,
"loss": 0.4417,
"step": 600
},
{
"epoch": 1.0217801047120418,
"grad_norm": 0.7231177091598511,
"learning_rate": 1.5882360660924084e-05,
"loss": 0.4445,
"step": 610
},
{
"epoch": 1.0385340314136127,
"grad_norm": 0.6571698784828186,
"learning_rate": 1.5731113353756752e-05,
"loss": 0.4514,
"step": 620
},
{
"epoch": 1.0552879581151833,
"grad_norm": 0.6722269058227539,
"learning_rate": 1.557788798852001e-05,
"loss": 0.4953,
"step": 630
},
{
"epoch": 1.0720418848167539,
"grad_norm": 0.64736407995224,
"learning_rate": 1.5422737449994078e-05,
"loss": 0.4674,
"step": 640
},
{
"epoch": 1.0887958115183247,
"grad_norm": 0.7046222686767578,
"learning_rate": 1.5265715287420752e-05,
"loss": 0.4234,
"step": 650
},
{
"epoch": 1.1055497382198953,
"grad_norm": 0.755739688873291,
"learning_rate": 1.5106875696021192e-05,
"loss": 0.4162,
"step": 660
},
{
"epoch": 1.122303664921466,
"grad_norm": 0.7437321543693542,
"learning_rate": 1.4946273498290815e-05,
"loss": 0.4681,
"step": 670
},
{
"epoch": 1.1390575916230365,
"grad_norm": 0.7438350915908813,
"learning_rate": 1.4783964125077614e-05,
"loss": 0.4529,
"step": 680
},
{
"epoch": 1.1558115183246074,
"grad_norm": 0.9883790016174316,
"learning_rate": 1.4620003596450576e-05,
"loss": 0.4289,
"step": 690
},
{
"epoch": 1.172565445026178,
"grad_norm": 0.7670640349388123,
"learning_rate": 1.4454448502364687e-05,
"loss": 0.4249,
"step": 700
},
{
"epoch": 1.1893193717277486,
"grad_norm": 0.6871143579483032,
"learning_rate": 1.4287355983129265e-05,
"loss": 0.452,
"step": 710
},
{
"epoch": 1.2060732984293194,
"grad_norm": 0.7382912039756775,
"learning_rate": 1.411878370968635e-05,
"loss": 0.4538,
"step": 720
},
{
"epoch": 1.22282722513089,
"grad_norm": 0.819150984287262,
"learning_rate": 1.3948789863705914e-05,
"loss": 0.4031,
"step": 730
},
{
"epoch": 1.2395811518324606,
"grad_norm": 0.8083910942077637,
"learning_rate": 1.3777433117504849e-05,
"loss": 0.449,
"step": 740
},
{
"epoch": 1.2563350785340315,
"grad_norm": 0.7677626609802246,
"learning_rate": 1.3604772613796551e-05,
"loss": 0.4682,
"step": 750
},
{
"epoch": 1.273089005235602,
"grad_norm": 0.7691996097564697,
"learning_rate": 1.3430867945278212e-05,
"loss": 0.4604,
"step": 760
},
{
"epoch": 1.2898429319371727,
"grad_norm": 0.7367651462554932,
"learning_rate": 1.3255779134062747e-05,
"loss": 0.4282,
"step": 770
},
{
"epoch": 1.3065968586387435,
"grad_norm": 0.8763916492462158,
"learning_rate": 1.3079566610962557e-05,
"loss": 0.4554,
"step": 780
},
{
"epoch": 1.3233507853403141,
"grad_norm": 0.8387876749038696,
"learning_rate": 1.2902291194632187e-05,
"loss": 0.4407,
"step": 790
},
{
"epoch": 1.3401047120418847,
"grad_norm": 0.7726860642433167,
"learning_rate": 1.2724014070577159e-05,
"loss": 0.4261,
"step": 800
},
{
"epoch": 1.3568586387434554,
"grad_norm": 0.7063042521476746,
"learning_rate": 1.254479677003619e-05,
"loss": 0.4236,
"step": 810
},
{
"epoch": 1.3736125654450262,
"grad_norm": 0.8544300198554993,
"learning_rate": 1.2364701148744035e-05,
"loss": 0.4715,
"step": 820
},
{
"epoch": 1.3903664921465968,
"grad_norm": 0.7012007236480713,
"learning_rate": 1.2183789365582402e-05,
"loss": 0.4514,
"step": 830
},
{
"epoch": 1.4071204188481676,
"grad_norm": 0.7686796188354492,
"learning_rate": 1.2002123861126169e-05,
"loss": 0.4714,
"step": 840
},
{
"epoch": 1.4238743455497382,
"grad_norm": 0.8073607683181763,
"learning_rate": 1.1819767336092434e-05,
"loss": 0.4234,
"step": 850
},
{
"epoch": 1.4406282722513089,
"grad_norm": 0.8263911604881287,
"learning_rate": 1.163678272969973e-05,
"loss": 0.4742,
"step": 860
},
{
"epoch": 1.4573821989528795,
"grad_norm": 0.7900548577308655,
"learning_rate": 1.1453233197944974e-05,
"loss": 0.4329,
"step": 870
},
{
"epoch": 1.4741361256544503,
"grad_norm": 0.7620972394943237,
"learning_rate": 1.1269182091805537e-05,
"loss": 0.4254,
"step": 880
},
{
"epoch": 1.490890052356021,
"grad_norm": 0.7844494581222534,
"learning_rate": 1.1084692935374075e-05,
"loss": 0.4322,
"step": 890
},
{
"epoch": 1.5076439790575917,
"grad_norm": 0.8092195987701416,
"learning_rate": 1.0899829403933568e-05,
"loss": 0.4773,
"step": 900
},
{
"epoch": 1.5243979057591623,
"grad_norm": 0.7582876682281494,
"learning_rate": 1.0714655301980201e-05,
"loss": 0.4226,
"step": 910
},
{
"epoch": 1.541151832460733,
"grad_norm": 0.7948154211044312,
"learning_rate": 1.0529234541201631e-05,
"loss": 0.4539,
"step": 920
},
{
"epoch": 1.5579057591623036,
"grad_norm": 0.8422669172286987,
"learning_rate": 1.0343631118418283e-05,
"loss": 0.4429,
"step": 930
},
{
"epoch": 1.5746596858638744,
"grad_norm": 0.9496159553527832,
"learning_rate": 1.015790909349522e-05,
"loss": 0.4667,
"step": 940
},
{
"epoch": 1.591413612565445,
"grad_norm": 0.8045749068260193,
"learning_rate": 9.972132567232301e-06,
"loss": 0.4627,
"step": 950
},
{
"epoch": 1.6081675392670158,
"grad_norm": 0.7890786528587341,
"learning_rate": 9.786365659240194e-06,
"loss": 0.4682,
"step": 960
},
{
"epoch": 1.6249214659685864,
"grad_norm": 0.823998212814331,
"learning_rate": 9.600672485809868e-06,
"loss": 0.4823,
"step": 970
},
{
"epoch": 1.641675392670157,
"grad_norm": 0.8117020726203918,
"learning_rate": 9.41511713778326e-06,
"loss": 0.4565,
"step": 980
},
{
"epoch": 1.6584293193717277,
"grad_norm": 0.7982953786849976,
"learning_rate": 9.229763658432752e-06,
"loss": 0.4896,
"step": 990
},
{
"epoch": 1.6751832460732983,
"grad_norm": 0.8489425778388977,
"learning_rate": 9.044676021356988e-06,
"loss": 0.451,
"step": 1000
},
{
"epoch": 1.6751832460732983,
"eval_loss": 0.4372614920139313,
"eval_runtime": 72.7911,
"eval_samples_per_second": 14.576,
"eval_steps_per_second": 7.295,
"step": 1000
},
{
"epoch": 1.691937172774869,
"grad_norm": 0.8381561636924744,
"learning_rate": 8.859918108400817e-06,
"loss": 0.4762,
"step": 1010
},
{
"epoch": 1.70869109947644,
"grad_norm": 0.8479191660881042,
"learning_rate": 8.675553687606864e-06,
"loss": 0.4385,
"step": 1020
},
{
"epoch": 1.7254450261780105,
"grad_norm": 0.8599492311477661,
"learning_rate": 8.491646391206374e-06,
"loss": 0.4365,
"step": 1030
},
{
"epoch": 1.7421989528795812,
"grad_norm": 0.8813064098358154,
"learning_rate": 8.308259693656972e-06,
"loss": 0.4494,
"step": 1040
},
{
"epoch": 1.7589528795811518,
"grad_norm": 0.9143653512001038,
"learning_rate": 8.125456889734816e-06,
"loss": 0.4245,
"step": 1050
},
{
"epoch": 1.7757068062827224,
"grad_norm": 0.8547804951667786,
"learning_rate": 7.943301072688832e-06,
"loss": 0.4275,
"step": 1060
},
{
"epoch": 1.7924607329842932,
"grad_norm": 0.8245564699172974,
"learning_rate": 7.761855112464447e-06,
"loss": 0.4065,
"step": 1070
},
{
"epoch": 1.809214659685864,
"grad_norm": 0.9589866995811462,
"learning_rate": 7.581181634004423e-06,
"loss": 0.4725,
"step": 1080
},
{
"epoch": 1.8259685863874346,
"grad_norm": 0.8072324395179749,
"learning_rate": 7.401342995634275e-06,
"loss": 0.4779,
"step": 1090
},
{
"epoch": 1.8427225130890053,
"grad_norm": 0.8404145240783691,
"learning_rate": 7.222401267539634e-06,
"loss": 0.4746,
"step": 1100
},
{
"epoch": 1.8594764397905759,
"grad_norm": 0.8116699457168579,
"learning_rate": 7.044418210343161e-06,
"loss": 0.4744,
"step": 1110
},
{
"epoch": 1.8762303664921465,
"grad_norm": 0.8185380697250366,
"learning_rate": 6.867455253788214e-06,
"loss": 0.3999,
"step": 1120
},
{
"epoch": 1.8929842931937173,
"grad_norm": 0.9066283106803894,
"learning_rate": 6.691573475536774e-06,
"loss": 0.4411,
"step": 1130
},
{
"epoch": 1.909738219895288,
"grad_norm": 0.9375002980232239,
"learning_rate": 6.516833580088873e-06,
"loss": 0.4061,
"step": 1140
},
{
"epoch": 1.9264921465968587,
"grad_norm": 0.8629588484764099,
"learning_rate": 6.343295877830797e-06,
"loss": 0.4575,
"step": 1150
},
{
"epoch": 1.9432460732984294,
"grad_norm": 0.8223278522491455,
"learning_rate": 6.171020264219344e-06,
"loss": 0.4422,
"step": 1160
},
{
"epoch": 1.96,
"grad_norm": 0.8139267563819885,
"learning_rate": 6.000066199109287e-06,
"loss": 0.4382,
"step": 1170
},
{
"epoch": 1.9767539267015706,
"grad_norm": 0.8527988195419312,
"learning_rate": 5.830492686231151e-06,
"loss": 0.3895,
"step": 1180
},
{
"epoch": 1.9935078534031414,
"grad_norm": 0.9887182116508484,
"learning_rate": 5.662358252826483e-06,
"loss": 0.3828,
"step": 1190
},
{
"epoch": 2.0100523560209425,
"grad_norm": 0.8452683687210083,
"learning_rate": 5.49572092944749e-06,
"loss": 0.3828,
"step": 1200
},
{
"epoch": 2.026806282722513,
"grad_norm": 0.9050988554954529,
"learning_rate": 5.330638229928193e-06,
"loss": 0.4497,
"step": 1210
},
{
"epoch": 2.0435602094240837,
"grad_norm": 1.036914587020874,
"learning_rate": 5.167167131533877e-06,
"loss": 0.4706,
"step": 1220
},
{
"epoch": 2.0603141361256543,
"grad_norm": 0.8268985748291016,
"learning_rate": 5.005364055295727e-06,
"loss": 0.4536,
"step": 1230
},
{
"epoch": 2.0770680628272253,
"grad_norm": 0.9379895329475403,
"learning_rate": 4.84528484653748e-06,
"loss": 0.3812,
"step": 1240
},
{
"epoch": 2.093821989528796,
"grad_norm": 0.9189339876174927,
"learning_rate": 4.686984755600757e-06,
"loss": 0.3778,
"step": 1250
},
{
"epoch": 2.1105759162303666,
"grad_norm": 0.900139570236206,
"learning_rate": 4.530518418775734e-06,
"loss": 0.3908,
"step": 1260
},
{
"epoch": 2.127329842931937,
"grad_norm": 0.8743225336074829,
"learning_rate": 4.375939839443787e-06,
"loss": 0.4186,
"step": 1270
},
{
"epoch": 2.1440837696335078,
"grad_norm": 0.9526062607765198,
"learning_rate": 4.223302369438554e-06,
"loss": 0.46,
"step": 1280
},
{
"epoch": 2.1608376963350784,
"grad_norm": 0.8931829333305359,
"learning_rate": 4.072658690631887e-06,
"loss": 0.4238,
"step": 1290
},
{
"epoch": 2.1775916230366494,
"grad_norm": 0.9619183540344238,
"learning_rate": 3.924060796751012e-06,
"loss": 0.4357,
"step": 1300
},
{
"epoch": 2.19434554973822,
"grad_norm": 1.0413931608200073,
"learning_rate": 3.7775599754332457e-06,
"loss": 0.4443,
"step": 1310
},
{
"epoch": 2.2110994764397907,
"grad_norm": 0.9470687508583069,
"learning_rate": 3.6332067905243728e-06,
"loss": 0.4208,
"step": 1320
},
{
"epoch": 2.2278534031413613,
"grad_norm": 0.9725884199142456,
"learning_rate": 3.491051064626847e-06,
"loss": 0.4307,
"step": 1330
},
{
"epoch": 2.244607329842932,
"grad_norm": 0.9030916094779968,
"learning_rate": 3.3511418619038406e-06,
"loss": 0.4386,
"step": 1340
},
{
"epoch": 2.2613612565445025,
"grad_norm": 0.85715651512146,
"learning_rate": 3.213527471145037e-06,
"loss": 0.4329,
"step": 1350
},
{
"epoch": 2.278115183246073,
"grad_norm": 0.9860045909881592,
"learning_rate": 3.078255389100078e-06,
"loss": 0.3981,
"step": 1360
},
{
"epoch": 2.294869109947644,
"grad_norm": 0.9435902237892151,
"learning_rate": 2.9453723040853433e-06,
"loss": 0.3627,
"step": 1370
},
{
"epoch": 2.3116230366492148,
"grad_norm": 0.9470138549804688,
"learning_rate": 2.8149240798697632e-06,
"loss": 0.3991,
"step": 1380
},
{
"epoch": 2.3283769633507854,
"grad_norm": 0.8298546075820923,
"learning_rate": 2.6869557398452397e-06,
"loss": 0.3823,
"step": 1390
},
{
"epoch": 2.345130890052356,
"grad_norm": 0.8885378837585449,
"learning_rate": 2.5615114514870966e-06,
"loss": 0.4355,
"step": 1400
},
{
"epoch": 2.3618848167539266,
"grad_norm": 1.0156524181365967,
"learning_rate": 2.438634511109941e-06,
"loss": 0.4289,
"step": 1410
},
{
"epoch": 2.378638743455497,
"grad_norm": 0.9947994351387024,
"learning_rate": 2.318367328924234e-06,
"loss": 0.3928,
"step": 1420
},
{
"epoch": 2.3953926701570682,
"grad_norm": 0.8874132037162781,
"learning_rate": 2.2007514143986386e-06,
"loss": 0.374,
"step": 1430
},
{
"epoch": 2.412146596858639,
"grad_norm": 0.9219970703125,
"learning_rate": 2.0858273619333135e-06,
"loss": 0.4454,
"step": 1440
},
{
"epoch": 2.4289005235602095,
"grad_norm": 0.9621909856796265,
"learning_rate": 1.9736348368489966e-06,
"loss": 0.401,
"step": 1450
},
{
"epoch": 2.44565445026178,
"grad_norm": 1.1088659763336182,
"learning_rate": 1.8642125616967587e-06,
"loss": 0.4059,
"step": 1460
},
{
"epoch": 2.4624083769633507,
"grad_norm": 0.9601513147354126,
"learning_rate": 1.7575983028931775e-06,
"loss": 0.4326,
"step": 1470
},
{
"epoch": 2.4791623036649213,
"grad_norm": 0.936352550983429,
"learning_rate": 1.6538288576854721e-06,
"loss": 0.4374,
"step": 1480
},
{
"epoch": 2.4959162303664923,
"grad_norm": 1.0005784034729004,
"learning_rate": 1.5529400414511809e-06,
"loss": 0.4076,
"step": 1490
},
{
"epoch": 2.512670157068063,
"grad_norm": 0.990437388420105,
"learning_rate": 1.4549666753366875e-06,
"loss": 0.419,
"step": 1500
},
{
"epoch": 2.512670157068063,
"eval_loss": 0.4265768229961395,
"eval_runtime": 72.3773,
"eval_samples_per_second": 14.659,
"eval_steps_per_second": 7.337,
"step": 1500
},
{
"epoch": 2.5294240837696336,
"grad_norm": 1.0548429489135742,
"learning_rate": 1.3599425742389382e-06,
"loss": 0.4199,
"step": 1510
},
{
"epoch": 2.546178010471204,
"grad_norm": 0.9754648208618164,
"learning_rate": 1.2679005351344308e-06,
"loss": 0.4377,
"step": 1520
},
{
"epoch": 2.562931937172775,
"grad_norm": 1.028577208518982,
"learning_rate": 1.1788723257595403e-06,
"loss": 0.4506,
"step": 1530
},
{
"epoch": 2.5796858638743454,
"grad_norm": 0.9221422076225281,
"learning_rate": 1.0928886736460954e-06,
"loss": 0.3728,
"step": 1540
},
{
"epoch": 2.596439790575916,
"grad_norm": 0.9151352643966675,
"learning_rate": 1.0099792555159437e-06,
"loss": 0.4214,
"step": 1550
},
{
"epoch": 2.613193717277487,
"grad_norm": 1.017935872077942,
"learning_rate": 9.301726870382277e-07,
"loss": 0.4464,
"step": 1560
},
{
"epoch": 2.6299476439790577,
"grad_norm": 0.9608612060546875,
"learning_rate": 8.534965129528716e-07,
"loss": 0.4239,
"step": 1570
},
{
"epoch": 2.6467015706806283,
"grad_norm": 0.9908237457275391,
"learning_rate": 7.799771975636672e-07,
"loss": 0.393,
"step": 1580
},
{
"epoch": 2.663455497382199,
"grad_norm": 0.9740267992019653,
"learning_rate": 7.096401156042975e-07,
"loss": 0.3993,
"step": 1590
},
{
"epoch": 2.6802094240837695,
"grad_norm": 0.9356936812400818,
"learning_rate": 6.425095434804074e-07,
"loss": 0.3985,
"step": 1600
},
{
"epoch": 2.6969633507853406,
"grad_norm": 0.9094387888908386,
"learning_rate": 5.786086508907396e-07,
"loss": 0.4152,
"step": 1610
},
{
"epoch": 2.7137172774869107,
"grad_norm": 0.9659162759780884,
"learning_rate": 5.179594928302723e-07,
"loss": 0.4123,
"step": 1620
},
{
"epoch": 2.7304712041884818,
"grad_norm": 0.9622294902801514,
"learning_rate": 4.6058300197806994e-07,
"loss": 0.4131,
"step": 1630
},
{
"epoch": 2.7472251308900524,
"grad_norm": 1.0013716220855713,
"learning_rate": 4.064989814724951e-07,
"loss": 0.4437,
"step": 1640
},
{
"epoch": 2.763979057591623,
"grad_norm": 1.0195082426071167,
"learning_rate": 3.557260980762833e-07,
"loss": 0.3943,
"step": 1650
},
{
"epoch": 2.7807329842931936,
"grad_norm": 1.0120123624801636,
"learning_rate": 3.0828187573382575e-07,
"loss": 0.434,
"step": 1660
},
{
"epoch": 2.797486910994764,
"grad_norm": 0.9160566926002502,
"learning_rate": 2.641826895228894e-07,
"loss": 0.4643,
"step": 1670
},
{
"epoch": 2.8142408376963353,
"grad_norm": 0.9706994295120239,
"learning_rate": 2.2344376000285606e-07,
"loss": 0.4638,
"step": 1680
},
{
"epoch": 2.830994764397906,
"grad_norm": 1.0439672470092773,
"learning_rate": 1.8607914796144988e-07,
"loss": 0.4458,
"step": 1690
},
{
"epoch": 2.8477486910994765,
"grad_norm": 0.947834849357605,
"learning_rate": 1.5210174956173495e-07,
"loss": 0.4396,
"step": 1700
},
{
"epoch": 2.864502617801047,
"grad_norm": 1.0074108839035034,
"learning_rate": 1.2152329189109136e-07,
"loss": 0.4331,
"step": 1710
},
{
"epoch": 2.8812565445026177,
"grad_norm": 0.9114189147949219,
"learning_rate": 9.435432891368279e-08,
"loss": 0.4221,
"step": 1720
},
{
"epoch": 2.8980104712041888,
"grad_norm": 0.9793130159378052,
"learning_rate": 7.060423782781712e-08,
"loss": 0.3817,
"step": 1730
},
{
"epoch": 2.914764397905759,
"grad_norm": 0.9325648546218872,
"learning_rate": 5.028121582947076e-08,
"loss": 0.483,
"step": 1740
},
{
"epoch": 2.93151832460733,
"grad_norm": 0.9619839191436768,
"learning_rate": 3.339227728307637e-08,
"loss": 0.4234,
"step": 1750
},
{
"epoch": 2.9482722513089006,
"grad_norm": 0.9810802340507507,
"learning_rate": 1.9943251300553745e-08,
"loss": 0.4377,
"step": 1760
},
{
"epoch": 2.965026178010471,
"grad_norm": 0.981601357460022,
"learning_rate": 9.9387797294348e-09,
"loss": 0.4259,
"step": 1770
},
{
"epoch": 2.981780104712042,
"grad_norm": 0.981569766998291,
"learning_rate": 3.3823155507550775e-09,
"loss": 0.4402,
"step": 1780
},
{
"epoch": 2.9985340314136124,
"grad_norm": 1.0313425064086914,
"learning_rate": 2.761216872781969e-10,
"loss": 0.432,
"step": 1790
},
{
"epoch": 3.0,
"step": 1791,
"total_flos": 3.5998475160807014e+17,
"train_loss": 0.468212741586764,
"train_runtime": 7355.8498,
"train_samples_per_second": 3.894,
"train_steps_per_second": 0.243
}
],
"logging_steps": 10,
"max_steps": 1791,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3.5998475160807014e+17,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}