es2 / trainer_state.json
huranokuma's picture
End of training
275dd30
raw
history blame
25.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 101208,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9752983953837645e-05,
"loss": 1.339,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 4.950596790767528e-05,
"loss": 1.3788,
"step": 1000
},
{
"epoch": 0.04,
"learning_rate": 4.9258951861512924e-05,
"loss": 1.378,
"step": 1500
},
{
"epoch": 0.06,
"learning_rate": 4.901193581535057e-05,
"loss": 1.3805,
"step": 2000
},
{
"epoch": 0.07,
"learning_rate": 4.876491976918821e-05,
"loss": 1.3945,
"step": 2500
},
{
"epoch": 0.09,
"learning_rate": 4.8517903723025847e-05,
"loss": 1.3919,
"step": 3000
},
{
"epoch": 0.1,
"learning_rate": 4.827088767686349e-05,
"loss": 1.4032,
"step": 3500
},
{
"epoch": 0.12,
"learning_rate": 4.802387163070113e-05,
"loss": 1.4197,
"step": 4000
},
{
"epoch": 0.13,
"learning_rate": 4.7776855584538775e-05,
"loss": 1.4222,
"step": 4500
},
{
"epoch": 0.15,
"learning_rate": 4.752983953837641e-05,
"loss": 1.4057,
"step": 5000
},
{
"epoch": 0.16,
"learning_rate": 4.7282823492214054e-05,
"loss": 1.4156,
"step": 5500
},
{
"epoch": 0.18,
"learning_rate": 4.70358074460517e-05,
"loss": 1.4243,
"step": 6000
},
{
"epoch": 0.19,
"learning_rate": 4.6788791399889334e-05,
"loss": 1.4176,
"step": 6500
},
{
"epoch": 0.21,
"learning_rate": 4.6541775353726977e-05,
"loss": 1.4254,
"step": 7000
},
{
"epoch": 0.22,
"learning_rate": 4.629475930756462e-05,
"loss": 1.4088,
"step": 7500
},
{
"epoch": 0.24,
"learning_rate": 4.604774326140226e-05,
"loss": 1.4307,
"step": 8000
},
{
"epoch": 0.25,
"learning_rate": 4.5800727215239905e-05,
"loss": 1.4459,
"step": 8500
},
{
"epoch": 0.27,
"learning_rate": 4.555371116907755e-05,
"loss": 1.4486,
"step": 9000
},
{
"epoch": 0.28,
"learning_rate": 4.530669512291519e-05,
"loss": 1.4432,
"step": 9500
},
{
"epoch": 0.3,
"learning_rate": 4.505967907675283e-05,
"loss": 1.4432,
"step": 10000
},
{
"epoch": 0.31,
"learning_rate": 4.481266303059047e-05,
"loss": 1.4424,
"step": 10500
},
{
"epoch": 0.33,
"learning_rate": 4.456564698442811e-05,
"loss": 1.4498,
"step": 11000
},
{
"epoch": 0.34,
"learning_rate": 4.4318630938265756e-05,
"loss": 1.4754,
"step": 11500
},
{
"epoch": 0.36,
"learning_rate": 4.407161489210339e-05,
"loss": 1.4712,
"step": 12000
},
{
"epoch": 0.37,
"learning_rate": 4.3824598845941035e-05,
"loss": 1.4552,
"step": 12500
},
{
"epoch": 0.39,
"learning_rate": 4.357758279977868e-05,
"loss": 1.4759,
"step": 13000
},
{
"epoch": 0.4,
"learning_rate": 4.333056675361632e-05,
"loss": 1.4877,
"step": 13500
},
{
"epoch": 0.41,
"learning_rate": 4.308355070745396e-05,
"loss": 1.4775,
"step": 14000
},
{
"epoch": 0.43,
"learning_rate": 4.28365346612916e-05,
"loss": 1.4701,
"step": 14500
},
{
"epoch": 0.44,
"learning_rate": 4.258951861512924e-05,
"loss": 1.4771,
"step": 15000
},
{
"epoch": 0.46,
"learning_rate": 4.234250256896688e-05,
"loss": 1.4856,
"step": 15500
},
{
"epoch": 0.47,
"learning_rate": 4.209548652280452e-05,
"loss": 1.4878,
"step": 16000
},
{
"epoch": 0.49,
"learning_rate": 4.1848470476642165e-05,
"loss": 1.5109,
"step": 16500
},
{
"epoch": 0.5,
"learning_rate": 4.160145443047981e-05,
"loss": 1.5056,
"step": 17000
},
{
"epoch": 0.52,
"learning_rate": 4.1354438384317445e-05,
"loss": 1.4853,
"step": 17500
},
{
"epoch": 0.53,
"learning_rate": 4.110742233815509e-05,
"loss": 1.5015,
"step": 18000
},
{
"epoch": 0.55,
"learning_rate": 4.086040629199273e-05,
"loss": 1.4883,
"step": 18500
},
{
"epoch": 0.56,
"learning_rate": 4.061339024583037e-05,
"loss": 1.4959,
"step": 19000
},
{
"epoch": 0.58,
"learning_rate": 4.036637419966801e-05,
"loss": 1.4887,
"step": 19500
},
{
"epoch": 0.59,
"learning_rate": 4.011935815350565e-05,
"loss": 1.5087,
"step": 20000
},
{
"epoch": 0.61,
"learning_rate": 3.9872342107343295e-05,
"loss": 1.5022,
"step": 20500
},
{
"epoch": 0.62,
"learning_rate": 3.962532606118093e-05,
"loss": 1.5063,
"step": 21000
},
{
"epoch": 0.64,
"learning_rate": 3.9378310015018575e-05,
"loss": 1.5054,
"step": 21500
},
{
"epoch": 0.65,
"learning_rate": 3.913129396885622e-05,
"loss": 1.5237,
"step": 22000
},
{
"epoch": 0.67,
"learning_rate": 3.888427792269386e-05,
"loss": 1.4989,
"step": 22500
},
{
"epoch": 0.68,
"learning_rate": 3.86372618765315e-05,
"loss": 1.521,
"step": 23000
},
{
"epoch": 0.7,
"learning_rate": 3.839024583036914e-05,
"loss": 1.5334,
"step": 23500
},
{
"epoch": 0.71,
"learning_rate": 3.814322978420678e-05,
"loss": 1.5102,
"step": 24000
},
{
"epoch": 0.73,
"learning_rate": 3.7896213738044425e-05,
"loss": 1.5112,
"step": 24500
},
{
"epoch": 0.74,
"learning_rate": 3.764919769188207e-05,
"loss": 1.534,
"step": 25000
},
{
"epoch": 0.76,
"learning_rate": 3.740218164571971e-05,
"loss": 1.5219,
"step": 25500
},
{
"epoch": 0.77,
"learning_rate": 3.7155165599557354e-05,
"loss": 1.551,
"step": 26000
},
{
"epoch": 0.79,
"learning_rate": 3.690814955339499e-05,
"loss": 1.5412,
"step": 26500
},
{
"epoch": 0.8,
"learning_rate": 3.6661133507232633e-05,
"loss": 1.539,
"step": 27000
},
{
"epoch": 0.82,
"learning_rate": 3.6414117461070276e-05,
"loss": 1.5338,
"step": 27500
},
{
"epoch": 0.83,
"learning_rate": 3.616710141490792e-05,
"loss": 1.531,
"step": 28000
},
{
"epoch": 0.84,
"learning_rate": 3.5920085368745556e-05,
"loss": 1.5448,
"step": 28500
},
{
"epoch": 0.86,
"learning_rate": 3.56730693225832e-05,
"loss": 1.5478,
"step": 29000
},
{
"epoch": 0.87,
"learning_rate": 3.542605327642084e-05,
"loss": 1.5441,
"step": 29500
},
{
"epoch": 0.89,
"learning_rate": 3.517903723025848e-05,
"loss": 1.5415,
"step": 30000
},
{
"epoch": 0.9,
"learning_rate": 3.493202118409612e-05,
"loss": 1.5513,
"step": 30500
},
{
"epoch": 0.92,
"learning_rate": 3.4685005137933763e-05,
"loss": 1.554,
"step": 31000
},
{
"epoch": 0.93,
"learning_rate": 3.4437989091771406e-05,
"loss": 1.5526,
"step": 31500
},
{
"epoch": 0.95,
"learning_rate": 3.419097304560904e-05,
"loss": 1.5608,
"step": 32000
},
{
"epoch": 0.96,
"learning_rate": 3.3943956999446686e-05,
"loss": 1.5627,
"step": 32500
},
{
"epoch": 0.98,
"learning_rate": 3.369694095328433e-05,
"loss": 1.5581,
"step": 33000
},
{
"epoch": 0.99,
"learning_rate": 3.3449924907121965e-05,
"loss": 1.5707,
"step": 33500
},
{
"epoch": 1.01,
"learning_rate": 3.320290886095961e-05,
"loss": 1.4586,
"step": 34000
},
{
"epoch": 1.02,
"learning_rate": 3.295589281479725e-05,
"loss": 1.3769,
"step": 34500
},
{
"epoch": 1.04,
"learning_rate": 3.2708876768634893e-05,
"loss": 1.3723,
"step": 35000
},
{
"epoch": 1.05,
"learning_rate": 3.246186072247253e-05,
"loss": 1.3776,
"step": 35500
},
{
"epoch": 1.07,
"learning_rate": 3.221484467631017e-05,
"loss": 1.391,
"step": 36000
},
{
"epoch": 1.08,
"learning_rate": 3.1967828630147816e-05,
"loss": 1.3848,
"step": 36500
},
{
"epoch": 1.1,
"learning_rate": 3.172081258398546e-05,
"loss": 1.3889,
"step": 37000
},
{
"epoch": 1.11,
"learning_rate": 3.1473796537823095e-05,
"loss": 1.4039,
"step": 37500
},
{
"epoch": 1.13,
"learning_rate": 3.122678049166074e-05,
"loss": 1.3999,
"step": 38000
},
{
"epoch": 1.14,
"learning_rate": 3.097976444549838e-05,
"loss": 1.4151,
"step": 38500
},
{
"epoch": 1.16,
"learning_rate": 3.073274839933602e-05,
"loss": 1.4157,
"step": 39000
},
{
"epoch": 1.17,
"learning_rate": 3.0485732353173663e-05,
"loss": 1.4274,
"step": 39500
},
{
"epoch": 1.19,
"learning_rate": 3.0238716307011306e-05,
"loss": 1.4204,
"step": 40000
},
{
"epoch": 1.2,
"learning_rate": 2.999170026084895e-05,
"loss": 1.4113,
"step": 40500
},
{
"epoch": 1.22,
"learning_rate": 2.9744684214686585e-05,
"loss": 1.4203,
"step": 41000
},
{
"epoch": 1.23,
"learning_rate": 2.9497668168524228e-05,
"loss": 1.3833,
"step": 41500
},
{
"epoch": 1.24,
"learning_rate": 2.925065212236187e-05,
"loss": 1.3336,
"step": 42000
},
{
"epoch": 1.26,
"learning_rate": 2.9003636076199514e-05,
"loss": 1.3271,
"step": 42500
},
{
"epoch": 1.27,
"learning_rate": 2.875662003003715e-05,
"loss": 1.3437,
"step": 43000
},
{
"epoch": 1.29,
"learning_rate": 2.8509603983874793e-05,
"loss": 1.3378,
"step": 43500
},
{
"epoch": 1.3,
"learning_rate": 2.8262587937712436e-05,
"loss": 1.3275,
"step": 44000
},
{
"epoch": 1.32,
"learning_rate": 2.8015571891550076e-05,
"loss": 1.352,
"step": 44500
},
{
"epoch": 1.33,
"learning_rate": 2.776855584538772e-05,
"loss": 1.3527,
"step": 45000
},
{
"epoch": 1.35,
"learning_rate": 2.752153979922536e-05,
"loss": 1.4116,
"step": 45500
},
{
"epoch": 1.36,
"learning_rate": 2.7274523753063e-05,
"loss": 1.3997,
"step": 46000
},
{
"epoch": 1.38,
"learning_rate": 2.702750770690064e-05,
"loss": 1.4157,
"step": 46500
},
{
"epoch": 1.39,
"learning_rate": 2.6780491660738284e-05,
"loss": 1.4121,
"step": 47000
},
{
"epoch": 1.41,
"learning_rate": 2.6533475614575926e-05,
"loss": 1.4252,
"step": 47500
},
{
"epoch": 1.42,
"learning_rate": 2.6286459568413563e-05,
"loss": 1.4227,
"step": 48000
},
{
"epoch": 1.44,
"learning_rate": 2.6039443522251206e-05,
"loss": 1.429,
"step": 48500
},
{
"epoch": 1.45,
"learning_rate": 2.579242747608885e-05,
"loss": 1.4173,
"step": 49000
},
{
"epoch": 1.47,
"learning_rate": 2.554541142992649e-05,
"loss": 1.4398,
"step": 49500
},
{
"epoch": 1.48,
"learning_rate": 2.5298395383764128e-05,
"loss": 1.4437,
"step": 50000
},
{
"epoch": 1.5,
"learning_rate": 2.505137933760177e-05,
"loss": 1.4467,
"step": 50500
},
{
"epoch": 1.51,
"learning_rate": 2.4804363291439414e-05,
"loss": 1.4396,
"step": 51000
},
{
"epoch": 1.53,
"learning_rate": 2.4557347245277053e-05,
"loss": 1.444,
"step": 51500
},
{
"epoch": 1.54,
"learning_rate": 2.4310331199114696e-05,
"loss": 1.4635,
"step": 52000
},
{
"epoch": 1.56,
"learning_rate": 2.4063315152952336e-05,
"loss": 1.4614,
"step": 52500
},
{
"epoch": 1.57,
"learning_rate": 2.381629910678998e-05,
"loss": 1.4783,
"step": 53000
},
{
"epoch": 1.59,
"learning_rate": 2.356928306062762e-05,
"loss": 1.4699,
"step": 53500
},
{
"epoch": 1.6,
"learning_rate": 2.332226701446526e-05,
"loss": 1.4788,
"step": 54000
},
{
"epoch": 1.62,
"learning_rate": 2.3075250968302904e-05,
"loss": 1.4732,
"step": 54500
},
{
"epoch": 1.63,
"learning_rate": 2.2828234922140544e-05,
"loss": 1.4794,
"step": 55000
},
{
"epoch": 1.65,
"learning_rate": 2.2581218875978187e-05,
"loss": 1.4708,
"step": 55500
},
{
"epoch": 1.66,
"learning_rate": 2.2334202829815826e-05,
"loss": 1.4666,
"step": 56000
},
{
"epoch": 1.67,
"learning_rate": 2.208718678365347e-05,
"loss": 1.4993,
"step": 56500
},
{
"epoch": 1.69,
"learning_rate": 2.184017073749111e-05,
"loss": 1.4921,
"step": 57000
},
{
"epoch": 1.7,
"learning_rate": 2.1593154691328748e-05,
"loss": 1.482,
"step": 57500
},
{
"epoch": 1.72,
"learning_rate": 2.134613864516639e-05,
"loss": 1.4862,
"step": 58000
},
{
"epoch": 1.73,
"learning_rate": 2.109912259900403e-05,
"loss": 1.4985,
"step": 58500
},
{
"epoch": 1.75,
"learning_rate": 2.0852106552841674e-05,
"loss": 1.5143,
"step": 59000
},
{
"epoch": 1.76,
"learning_rate": 2.0605090506679313e-05,
"loss": 1.5043,
"step": 59500
},
{
"epoch": 1.78,
"learning_rate": 2.0358074460516956e-05,
"loss": 1.5227,
"step": 60000
},
{
"epoch": 1.79,
"learning_rate": 2.0111058414354596e-05,
"loss": 1.5202,
"step": 60500
},
{
"epoch": 1.81,
"learning_rate": 1.986404236819224e-05,
"loss": 1.4978,
"step": 61000
},
{
"epoch": 1.82,
"learning_rate": 1.961702632202988e-05,
"loss": 1.5271,
"step": 61500
},
{
"epoch": 1.84,
"learning_rate": 1.937001027586752e-05,
"loss": 1.5253,
"step": 62000
},
{
"epoch": 1.85,
"learning_rate": 1.9122994229705164e-05,
"loss": 1.5317,
"step": 62500
},
{
"epoch": 1.87,
"learning_rate": 1.8875978183542804e-05,
"loss": 1.5253,
"step": 63000
},
{
"epoch": 1.88,
"learning_rate": 1.8628962137380447e-05,
"loss": 1.5434,
"step": 63500
},
{
"epoch": 1.9,
"learning_rate": 1.8381946091218086e-05,
"loss": 1.5219,
"step": 64000
},
{
"epoch": 1.91,
"learning_rate": 1.813493004505573e-05,
"loss": 1.537,
"step": 64500
},
{
"epoch": 1.93,
"learning_rate": 1.788791399889337e-05,
"loss": 1.5622,
"step": 65000
},
{
"epoch": 1.94,
"learning_rate": 1.764089795273101e-05,
"loss": 1.5391,
"step": 65500
},
{
"epoch": 1.96,
"learning_rate": 1.739388190656865e-05,
"loss": 1.573,
"step": 66000
},
{
"epoch": 1.97,
"learning_rate": 1.714686586040629e-05,
"loss": 1.566,
"step": 66500
},
{
"epoch": 1.99,
"learning_rate": 1.6899849814243934e-05,
"loss": 1.5783,
"step": 67000
},
{
"epoch": 2.0,
"learning_rate": 1.6652833768081573e-05,
"loss": 1.5639,
"step": 67500
},
{
"epoch": 2.02,
"learning_rate": 1.6405817721919216e-05,
"loss": 1.3644,
"step": 68000
},
{
"epoch": 2.03,
"learning_rate": 1.6158801675756856e-05,
"loss": 1.3529,
"step": 68500
},
{
"epoch": 2.05,
"learning_rate": 1.59117856295945e-05,
"loss": 1.3552,
"step": 69000
},
{
"epoch": 2.06,
"learning_rate": 1.566476958343214e-05,
"loss": 1.3508,
"step": 69500
},
{
"epoch": 2.07,
"learning_rate": 1.5417753537269785e-05,
"loss": 1.3663,
"step": 70000
},
{
"epoch": 2.09,
"learning_rate": 1.5170737491107422e-05,
"loss": 1.3622,
"step": 70500
},
{
"epoch": 2.1,
"learning_rate": 1.4923721444945064e-05,
"loss": 1.3698,
"step": 71000
},
{
"epoch": 2.12,
"learning_rate": 1.4676705398782705e-05,
"loss": 1.3664,
"step": 71500
},
{
"epoch": 2.13,
"learning_rate": 1.4429689352620346e-05,
"loss": 1.3803,
"step": 72000
},
{
"epoch": 2.15,
"learning_rate": 1.4182673306457989e-05,
"loss": 1.3713,
"step": 72500
},
{
"epoch": 2.16,
"learning_rate": 1.3935657260295629e-05,
"loss": 1.3897,
"step": 73000
},
{
"epoch": 2.18,
"learning_rate": 1.3688641214133272e-05,
"loss": 1.3869,
"step": 73500
},
{
"epoch": 2.19,
"learning_rate": 1.3441625167970911e-05,
"loss": 1.3919,
"step": 74000
},
{
"epoch": 2.21,
"learning_rate": 1.3194609121808554e-05,
"loss": 1.4013,
"step": 74500
},
{
"epoch": 2.22,
"learning_rate": 1.2947593075646194e-05,
"loss": 1.3853,
"step": 75000
},
{
"epoch": 2.24,
"learning_rate": 1.2700577029483837e-05,
"loss": 1.3776,
"step": 75500
},
{
"epoch": 2.25,
"learning_rate": 1.2453560983321478e-05,
"loss": 1.3928,
"step": 76000
},
{
"epoch": 2.27,
"learning_rate": 1.220654493715912e-05,
"loss": 1.4051,
"step": 76500
},
{
"epoch": 2.28,
"learning_rate": 1.195952889099676e-05,
"loss": 1.3934,
"step": 77000
},
{
"epoch": 2.3,
"learning_rate": 1.1712512844834402e-05,
"loss": 1.4057,
"step": 77500
},
{
"epoch": 2.31,
"learning_rate": 1.1465496798672043e-05,
"loss": 1.3998,
"step": 78000
},
{
"epoch": 2.33,
"learning_rate": 1.1218480752509682e-05,
"loss": 1.4068,
"step": 78500
},
{
"epoch": 2.34,
"learning_rate": 1.0971464706347324e-05,
"loss": 1.409,
"step": 79000
},
{
"epoch": 2.36,
"learning_rate": 1.0724448660184967e-05,
"loss": 1.395,
"step": 79500
},
{
"epoch": 2.37,
"learning_rate": 1.0477432614022608e-05,
"loss": 1.4237,
"step": 80000
},
{
"epoch": 2.39,
"learning_rate": 1.023041656786025e-05,
"loss": 1.4202,
"step": 80500
},
{
"epoch": 2.4,
"learning_rate": 9.98340052169789e-06,
"loss": 1.3998,
"step": 81000
},
{
"epoch": 2.42,
"learning_rate": 9.736384475535532e-06,
"loss": 1.4187,
"step": 81500
},
{
"epoch": 2.43,
"learning_rate": 9.489368429373173e-06,
"loss": 1.4312,
"step": 82000
},
{
"epoch": 2.45,
"learning_rate": 9.242352383210814e-06,
"loss": 1.4199,
"step": 82500
},
{
"epoch": 2.46,
"learning_rate": 8.995336337048454e-06,
"loss": 1.4032,
"step": 83000
},
{
"epoch": 2.48,
"learning_rate": 8.748320290886097e-06,
"loss": 1.4152,
"step": 83500
},
{
"epoch": 2.49,
"learning_rate": 8.501304244723738e-06,
"loss": 1.428,
"step": 84000
},
{
"epoch": 2.5,
"learning_rate": 8.25428819856138e-06,
"loss": 1.4191,
"step": 84500
},
{
"epoch": 2.52,
"learning_rate": 8.00727215239902e-06,
"loss": 1.4356,
"step": 85000
},
{
"epoch": 2.53,
"learning_rate": 7.760256106236662e-06,
"loss": 1.443,
"step": 85500
},
{
"epoch": 2.55,
"learning_rate": 7.513240060074303e-06,
"loss": 1.4406,
"step": 86000
},
{
"epoch": 2.56,
"learning_rate": 7.266224013911944e-06,
"loss": 1.4256,
"step": 86500
},
{
"epoch": 2.58,
"learning_rate": 7.0192079677495855e-06,
"loss": 1.4408,
"step": 87000
},
{
"epoch": 2.59,
"learning_rate": 6.7721919215872276e-06,
"loss": 1.4476,
"step": 87500
},
{
"epoch": 2.61,
"learning_rate": 6.525175875424867e-06,
"loss": 1.434,
"step": 88000
},
{
"epoch": 2.62,
"learning_rate": 6.278159829262508e-06,
"loss": 1.4378,
"step": 88500
},
{
"epoch": 2.64,
"learning_rate": 6.0311437831001505e-06,
"loss": 1.4276,
"step": 89000
},
{
"epoch": 2.65,
"learning_rate": 5.784127736937792e-06,
"loss": 1.4491,
"step": 89500
},
{
"epoch": 2.67,
"learning_rate": 5.537111690775433e-06,
"loss": 1.4583,
"step": 90000
},
{
"epoch": 2.68,
"learning_rate": 5.290095644613074e-06,
"loss": 1.4546,
"step": 90500
},
{
"epoch": 2.7,
"learning_rate": 5.043079598450716e-06,
"loss": 1.4338,
"step": 91000
},
{
"epoch": 2.71,
"learning_rate": 4.796063552288357e-06,
"loss": 1.4628,
"step": 91500
},
{
"epoch": 2.73,
"learning_rate": 4.549047506125998e-06,
"loss": 1.4546,
"step": 92000
},
{
"epoch": 2.74,
"learning_rate": 4.302031459963639e-06,
"loss": 1.449,
"step": 92500
},
{
"epoch": 2.76,
"learning_rate": 4.055015413801281e-06,
"loss": 1.4489,
"step": 93000
},
{
"epoch": 2.77,
"learning_rate": 3.8079993676389218e-06,
"loss": 1.4632,
"step": 93500
},
{
"epoch": 2.79,
"learning_rate": 3.560983321476563e-06,
"loss": 1.4739,
"step": 94000
},
{
"epoch": 2.8,
"learning_rate": 3.3139672753142047e-06,
"loss": 1.4662,
"step": 94500
},
{
"epoch": 2.82,
"learning_rate": 3.066951229151846e-06,
"loss": 1.4496,
"step": 95000
},
{
"epoch": 2.83,
"learning_rate": 2.819935182989487e-06,
"loss": 1.4544,
"step": 95500
},
{
"epoch": 2.85,
"learning_rate": 2.5729191368271284e-06,
"loss": 1.4656,
"step": 96000
},
{
"epoch": 2.86,
"learning_rate": 2.3259030906647697e-06,
"loss": 1.4679,
"step": 96500
},
{
"epoch": 2.88,
"learning_rate": 2.078887044502411e-06,
"loss": 1.4869,
"step": 97000
},
{
"epoch": 2.89,
"learning_rate": 1.8318709983400524e-06,
"loss": 1.4645,
"step": 97500
},
{
"epoch": 2.9,
"learning_rate": 1.5848549521776935e-06,
"loss": 1.4838,
"step": 98000
},
{
"epoch": 2.92,
"learning_rate": 1.3378389060153347e-06,
"loss": 1.4858,
"step": 98500
},
{
"epoch": 2.93,
"learning_rate": 1.0908228598529762e-06,
"loss": 1.4788,
"step": 99000
},
{
"epoch": 2.95,
"learning_rate": 8.438068136906174e-07,
"loss": 1.4936,
"step": 99500
},
{
"epoch": 2.96,
"learning_rate": 5.967907675282587e-07,
"loss": 1.4874,
"step": 100000
},
{
"epoch": 2.98,
"learning_rate": 3.4977472136589993e-07,
"loss": 1.4835,
"step": 100500
},
{
"epoch": 2.99,
"learning_rate": 1.0275867520354124e-07,
"loss": 1.5014,
"step": 101000
},
{
"epoch": 3.0,
"step": 101208,
"total_flos": 3.759621873629921e+17,
"train_loss": 1.4532578269084653,
"train_runtime": 40248.475,
"train_samples_per_second": 10.058,
"train_steps_per_second": 2.515
}
],
"max_steps": 101208,
"num_train_epochs": 3,
"total_flos": 3.759621873629921e+17,
"trial_name": null,
"trial_params": null
}