|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 9.999915498738517, |
|
"global_step": 51770, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 5.791505791505791e-07, |
|
"loss": 2.2583, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 0.00014478764478764478, |
|
"loss": 1.4754, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 0.00028957528957528956, |
|
"loss": 1.2393, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 0.0002999848327088156, |
|
"loss": 1.2015, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 0.0002999345361024936, |
|
"loss": 1.1787, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 0.0002998490306892274, |
|
"loss": 1.1622, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 0.0002997283365480707, |
|
"loss": 1.1493, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 0.0002995724820213708, |
|
"loss": 1.1407, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 0.00029938150370811345, |
|
"loss": 1.1306, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 0.0002991554464553282, |
|
"loss": 1.1228, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 0.0002988943633475569, |
|
"loss": 1.1142, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 0.0002985983156943883, |
|
"loss": 1.1082, |
|
"step": 2750 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 0.0002982673730160606, |
|
"loss": 1.1035, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 0.0002979016130271363, |
|
"loss": 1.0949, |
|
"step": 3250 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 0.00029750112161825245, |
|
"loss": 1.0909, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 0.00029706599283595153, |
|
"loss": 1.0862, |
|
"step": 3750 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 0.0002965963288605962, |
|
"loss": 1.0799, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 0.0002960922399823749, |
|
"loss": 1.0752, |
|
"step": 4250 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 0.0002955538445754026, |
|
"loss": 1.0717, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 0.00029498126906992285, |
|
"loss": 1.0675, |
|
"step": 4750 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 0.00029437464792261906, |
|
"loss": 1.0629, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_alliteration_score": 0.39176506206478956, |
|
"eval_harmonic_meter_score": 0.10270844382058368, |
|
"eval_harmonic_rhyme_score": 0.13433666525497734, |
|
"eval_meter_score": 0.28353772535274874, |
|
"eval_rhyme_score": 0.6209502922727462, |
|
"eval_runtime": 2477.4307, |
|
"eval_samples_per_second": 1.09, |
|
"eval_steps_per_second": 0.136, |
|
"step": 5177 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 0.0002937341235850398, |
|
"loss": 1.0558, |
|
"step": 5250 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 0.0002930598464701476, |
|
"loss": 1.0356, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 0.00029235197491699733, |
|
"loss": 1.0344, |
|
"step": 5750 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 0.0002916106751535544, |
|
"loss": 1.0315, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 0.0002908361212576593, |
|
"loss": 1.0306, |
|
"step": 6250 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 0.00029002849511614926, |
|
"loss": 1.027, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 0.0002891879863821466, |
|
"loss": 1.0232, |
|
"step": 6750 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 0.00028831479243052253, |
|
"loss": 1.021, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 0.0002874091183115481, |
|
"loss": 1.0189, |
|
"step": 7250 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 0.00028647117670274266, |
|
"loss": 1.0161, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 0.00028550118785893134, |
|
"loss": 1.0132, |
|
"step": 7750 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 0.000284499379560523, |
|
"loss": 1.0102, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 0.0002834659870600212, |
|
"loss": 1.0088, |
|
"step": 8250 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 0.0002824012530267801, |
|
"loss": 1.0055, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 0.00028130542749001906, |
|
"loss": 1.0033, |
|
"step": 8750 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 0.0002801787677801088, |
|
"loss": 1.001, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 0.00027902153846814335, |
|
"loss": 0.9979, |
|
"step": 9250 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 0.0002778340113038107, |
|
"loss": 0.994, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 0.00027661646515157913, |
|
"loss": 0.9919, |
|
"step": 9750 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 0.0002753691859252116, |
|
"loss": 0.9895, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 0.0002740924665206254, |
|
"loss": 0.9881, |
|
"step": 10250 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_alliteration_score": 0.4102959309494451, |
|
"eval_harmonic_meter_score": 0.11239968726166898, |
|
"eval_harmonic_rhyme_score": 0.2758249186922835, |
|
"eval_meter_score": 0.30497402024728615, |
|
"eval_rhyme_score": 0.7041029209046453, |
|
"eval_runtime": 2601.7347, |
|
"eval_samples_per_second": 1.038, |
|
"eval_steps_per_second": 0.13, |
|
"step": 10354 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 0.000272786606747112, |
|
"loss": 0.9675, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 0.00027145191325693354, |
|
"loss": 0.9528, |
|
"step": 10750 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 0.0002700886994733122, |
|
"loss": 0.9521, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 0.0002686972855168298, |
|
"loss": 0.952, |
|
"step": 11250 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 0.00026727799813025446, |
|
"loss": 0.9508, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 0.0002658311706018126, |
|
"loss": 0.9499, |
|
"step": 11750 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 0.000264357142686923, |
|
"loss": 0.947, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 2.37, |
|
"learning_rate": 0.0002628562605284134, |
|
"loss": 0.9458, |
|
"step": 12250 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 0.0002613288765752361, |
|
"loss": 0.9463, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 0.0002597753494997032, |
|
"loss": 0.9415, |
|
"step": 12750 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 0.00025819604411326055, |
|
"loss": 0.9412, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 0.0002565913312808196, |
|
"loss": 0.9381, |
|
"step": 13250 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 0.00025496158783366803, |
|
"loss": 0.9364, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 0.00025330719648097947, |
|
"loss": 0.9346, |
|
"step": 13750 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 0.0002516285457199425, |
|
"loss": 0.9339, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 0.00024992602974453066, |
|
"loss": 0.9323, |
|
"step": 14250 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 0.000248200048352935, |
|
"loss": 0.9284, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 2.85, |
|
"learning_rate": 0.00024645100685368013, |
|
"loss": 0.9269, |
|
"step": 14750 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 0.0002446793159704463, |
|
"loss": 0.9252, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 0.00024288539174562046, |
|
"loss": 0.9243, |
|
"step": 15250 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 0.00024106965544259755, |
|
"loss": 0.9211, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_alliteration_score": 0.4366551398931154, |
|
"eval_harmonic_meter_score": 0.11132001638651683, |
|
"eval_harmonic_rhyme_score": 0.3526620304363614, |
|
"eval_meter_score": 0.3150168760675491, |
|
"eval_rhyme_score": 0.7517596634968854, |
|
"eval_runtime": 2670.3356, |
|
"eval_samples_per_second": 1.011, |
|
"eval_steps_per_second": 0.127, |
|
"step": 15531 |
|
}, |
|
{ |
|
"epoch": 3.04, |
|
"learning_rate": 0.00023923253344685635, |
|
"loss": 0.8862, |
|
"step": 15750 |
|
}, |
|
{ |
|
"epoch": 3.09, |
|
"learning_rate": 0.0002373744571658321, |
|
"loss": 0.8819, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 3.14, |
|
"learning_rate": 0.00023549586292761015, |
|
"loss": 0.8832, |
|
"step": 16250 |
|
}, |
|
{ |
|
"epoch": 3.19, |
|
"learning_rate": 0.00023359719187846398, |
|
"loss": 0.88, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 3.24, |
|
"learning_rate": 0.00023167888987926164, |
|
"loss": 0.881, |
|
"step": 16750 |
|
}, |
|
{ |
|
"epoch": 3.28, |
|
"learning_rate": 0.00022974140740076524, |
|
"loss": 0.8792, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 3.33, |
|
"learning_rate": 0.00022778519941784797, |
|
"loss": 0.8803, |
|
"step": 17250 |
|
}, |
|
{ |
|
"epoch": 3.38, |
|
"learning_rate": 0.00022581072530265315, |
|
"loss": 0.8774, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 3.43, |
|
"learning_rate": 0.000223818448716721, |
|
"loss": 0.8759, |
|
"step": 17750 |
|
}, |
|
{ |
|
"epoch": 3.48, |
|
"learning_rate": 0.000221808837502108, |
|
"loss": 0.8737, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 3.53, |
|
"learning_rate": 0.00021978236357152417, |
|
"loss": 0.8717, |
|
"step": 18250 |
|
}, |
|
{ |
|
"epoch": 3.57, |
|
"learning_rate": 0.00021773950279751545, |
|
"loss": 0.871, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 3.62, |
|
"learning_rate": 0.0002156807349007151, |
|
"loss": 0.8702, |
|
"step": 18750 |
|
}, |
|
{ |
|
"epoch": 3.67, |
|
"learning_rate": 0.00021360654333719197, |
|
"loss": 0.8682, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 3.72, |
|
"learning_rate": 0.00021151741518492183, |
|
"loss": 0.8659, |
|
"step": 19250 |
|
}, |
|
{ |
|
"epoch": 3.77, |
|
"learning_rate": 0.00020941384102940748, |
|
"loss": 0.8658, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 3.81, |
|
"learning_rate": 0.00020729631484847592, |
|
"loss": 0.8625, |
|
"step": 19750 |
|
}, |
|
{ |
|
"epoch": 3.86, |
|
"learning_rate": 0.00020516533389627854, |
|
"loss": 0.8607, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 3.91, |
|
"learning_rate": 0.00020302139858652179, |
|
"loss": 0.8604, |
|
"step": 20250 |
|
}, |
|
{ |
|
"epoch": 3.96, |
|
"learning_rate": 0.00020086501237495636, |
|
"loss": 0.8578, |
|
"step": 20500 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_alliteration_score": 0.4113367687177933, |
|
"eval_harmonic_meter_score": 0.11086907691273673, |
|
"eval_harmonic_rhyme_score": 0.4120432314176195, |
|
"eval_meter_score": 0.30988666333117193, |
|
"eval_rhyme_score": 0.7734423562173335, |
|
"eval_runtime": 2384.5447, |
|
"eval_samples_per_second": 1.132, |
|
"eval_steps_per_second": 0.142, |
|
"step": 20708 |
|
}, |
|
{ |
|
"epoch": 4.01, |
|
"learning_rate": 0.00019869668164115161, |
|
"loss": 0.8501, |
|
"step": 20750 |
|
}, |
|
{ |
|
"epoch": 4.06, |
|
"learning_rate": 0.00019651691556958343, |
|
"loss": 0.8109, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 4.1, |
|
"learning_rate": 0.00019432622603006387, |
|
"loss": 0.8126, |
|
"step": 21250 |
|
}, |
|
{ |
|
"epoch": 4.15, |
|
"learning_rate": 0.00019212512745753942, |
|
"loss": 0.8134, |
|
"step": 21500 |
|
}, |
|
{ |
|
"epoch": 4.2, |
|
"learning_rate": 0.00018991413673128776, |
|
"loss": 0.8149, |
|
"step": 21750 |
|
}, |
|
{ |
|
"epoch": 4.25, |
|
"learning_rate": 0.0001876937730535397, |
|
"loss": 0.8136, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 4.3, |
|
"learning_rate": 0.00018546455782755646, |
|
"loss": 0.8118, |
|
"step": 22250 |
|
}, |
|
{ |
|
"epoch": 4.35, |
|
"learning_rate": 0.0001832270145351892, |
|
"loss": 0.811, |
|
"step": 22500 |
|
}, |
|
{ |
|
"epoch": 4.39, |
|
"learning_rate": 0.00018098166861395132, |
|
"loss": 0.8101, |
|
"step": 22750 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 0.00017872904733363072, |
|
"loss": 0.8102, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 4.49, |
|
"learning_rate": 0.00017646967967247255, |
|
"loss": 0.8072, |
|
"step": 23250 |
|
}, |
|
{ |
|
"epoch": 4.54, |
|
"learning_rate": 0.00017420409619295997, |
|
"loss": 0.8071, |
|
"step": 23500 |
|
}, |
|
{ |
|
"epoch": 4.59, |
|
"learning_rate": 0.00017193282891722365, |
|
"loss": 0.8049, |
|
"step": 23750 |
|
}, |
|
{ |
|
"epoch": 4.64, |
|
"learning_rate": 0.00016965641120210775, |
|
"loss": 0.8035, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 4.68, |
|
"learning_rate": 0.00016737537761392346, |
|
"loss": 0.8033, |
|
"step": 24250 |
|
}, |
|
{ |
|
"epoch": 4.73, |
|
"learning_rate": 0.00016509026380291723, |
|
"loss": 0.7998, |
|
"step": 24500 |
|
}, |
|
{ |
|
"epoch": 4.78, |
|
"learning_rate": 0.00016280160637748584, |
|
"loss": 0.8009, |
|
"step": 24750 |
|
}, |
|
{ |
|
"epoch": 4.83, |
|
"learning_rate": 0.0001605099427781652, |
|
"loss": 0.7986, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 4.88, |
|
"learning_rate": 0.00015821581115142477, |
|
"loss": 0.7954, |
|
"step": 25250 |
|
}, |
|
{ |
|
"epoch": 4.93, |
|
"learning_rate": 0.00015591975022329567, |
|
"loss": 0.7945, |
|
"step": 25500 |
|
}, |
|
{ |
|
"epoch": 4.97, |
|
"learning_rate": 0.0001536222991728626, |
|
"loss": 0.7938, |
|
"step": 25750 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_alliteration_score": 0.41413515189088657, |
|
"eval_harmonic_meter_score": 0.1026455246284712, |
|
"eval_harmonic_rhyme_score": 0.34333090390331705, |
|
"eval_meter_score": 0.2882500313162144, |
|
"eval_rhyme_score": 0.6791661142839286, |
|
"eval_runtime": 2270.9559, |
|
"eval_samples_per_second": 1.189, |
|
"eval_steps_per_second": 0.149, |
|
"step": 25885 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 0.0001513239975056498, |
|
"loss": 0.772, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 0.00014902538492693, |
|
"loss": 0.747, |
|
"step": 26250 |
|
}, |
|
{ |
|
"epoch": 5.12, |
|
"learning_rate": 0.00014672700121498661, |
|
"loss": 0.7489, |
|
"step": 26500 |
|
}, |
|
{ |
|
"epoch": 5.17, |
|
"learning_rate": 0.00014442938609435875, |
|
"loss": 0.748, |
|
"step": 26750 |
|
}, |
|
{ |
|
"epoch": 5.22, |
|
"learning_rate": 0.00014213307910909895, |
|
"loss": 0.7485, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 5.26, |
|
"learning_rate": 0.000139838619496073, |
|
"loss": 0.7493, |
|
"step": 27250 |
|
}, |
|
{ |
|
"epoch": 5.31, |
|
"learning_rate": 0.00013754654605833231, |
|
"loss": 0.7471, |
|
"step": 27500 |
|
}, |
|
{ |
|
"epoch": 5.36, |
|
"learning_rate": 0.0001352573970385877, |
|
"loss": 0.7466, |
|
"step": 27750 |
|
}, |
|
{ |
|
"epoch": 5.41, |
|
"learning_rate": 0.00013297170999281534, |
|
"loss": 0.7462, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 5.46, |
|
"learning_rate": 0.0001306900216640233, |
|
"loss": 0.7451, |
|
"step": 28250 |
|
}, |
|
{ |
|
"epoch": 5.51, |
|
"learning_rate": 0.0001284128678562096, |
|
"loss": 0.7454, |
|
"step": 28500 |
|
}, |
|
{ |
|
"epoch": 5.55, |
|
"learning_rate": 0.0001261407833085402, |
|
"loss": 0.7424, |
|
"step": 28750 |
|
}, |
|
{ |
|
"epoch": 5.6, |
|
"learning_rate": 0.00012387430156977778, |
|
"loss": 0.7414, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 5.65, |
|
"learning_rate": 0.00012161395487298911, |
|
"loss": 0.7405, |
|
"step": 29250 |
|
}, |
|
{ |
|
"epoch": 5.7, |
|
"learning_rate": 0.00011936027401056238, |
|
"loss": 0.7379, |
|
"step": 29500 |
|
}, |
|
{ |
|
"epoch": 5.75, |
|
"learning_rate": 0.00011711378820956193, |
|
"loss": 0.738, |
|
"step": 29750 |
|
}, |
|
{ |
|
"epoch": 5.79, |
|
"learning_rate": 0.00011487502500745135, |
|
"loss": 0.7363, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 5.84, |
|
"learning_rate": 0.00011264451012821272, |
|
"loss": 0.7347, |
|
"step": 30250 |
|
}, |
|
{ |
|
"epoch": 5.89, |
|
"learning_rate": 0.00011042276735889216, |
|
"loss": 0.7321, |
|
"step": 30500 |
|
}, |
|
{ |
|
"epoch": 5.94, |
|
"learning_rate": 0.00010821031842659992, |
|
"loss": 0.7332, |
|
"step": 30750 |
|
}, |
|
{ |
|
"epoch": 5.99, |
|
"learning_rate": 0.00010600768287599417, |
|
"loss": 0.7313, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_alliteration_score": 0.431023911710607, |
|
"eval_harmonic_meter_score": 0.10282598031904473, |
|
"eval_harmonic_rhyme_score": 0.43195083122804273, |
|
"eval_meter_score": 0.3169839487583337, |
|
"eval_rhyme_score": 0.7959113397342684, |
|
"eval_runtime": 2406.9132, |
|
"eval_samples_per_second": 1.122, |
|
"eval_steps_per_second": 0.14, |
|
"step": 31062 |
|
}, |
|
{ |
|
"epoch": 6.04, |
|
"learning_rate": 0.00010381537794727796, |
|
"loss": 0.6992, |
|
"step": 31250 |
|
}, |
|
{ |
|
"epoch": 6.08, |
|
"learning_rate": 0.00010163391845473613, |
|
"loss": 0.6893, |
|
"step": 31500 |
|
}, |
|
{ |
|
"epoch": 6.13, |
|
"learning_rate": 9.946381666584314e-05, |
|
"loss": 0.6898, |
|
"step": 31750 |
|
}, |
|
{ |
|
"epoch": 6.18, |
|
"learning_rate": 9.7305582180968e-05, |
|
"loss": 0.69, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 6.23, |
|
"learning_rate": 9.51597218137062e-05, |
|
"loss": 0.6907, |
|
"step": 32250 |
|
}, |
|
{ |
|
"epoch": 6.28, |
|
"learning_rate": 9.302673947186545e-05, |
|
"loss": 0.6903, |
|
"step": 32500 |
|
}, |
|
{ |
|
"epoch": 6.33, |
|
"learning_rate": 9.090713603913437e-05, |
|
"loss": 0.69, |
|
"step": 32750 |
|
}, |
|
{ |
|
"epoch": 6.37, |
|
"learning_rate": 8.88014092574609e-05, |
|
"loss": 0.688, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 6.42, |
|
"learning_rate": 8.671005361016862e-05, |
|
"loss": 0.687, |
|
"step": 33250 |
|
}, |
|
{ |
|
"epoch": 6.47, |
|
"learning_rate": 8.463356020583833e-05, |
|
"loss": 0.6859, |
|
"step": 33500 |
|
}, |
|
{ |
|
"epoch": 6.52, |
|
"learning_rate": 8.257241666298198e-05, |
|
"loss": 0.6874, |
|
"step": 33750 |
|
}, |
|
{ |
|
"epoch": 6.57, |
|
"learning_rate": 8.05271069955365e-05, |
|
"loss": 0.6849, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 6.62, |
|
"learning_rate": 7.849811149920353e-05, |
|
"loss": 0.6831, |
|
"step": 34250 |
|
}, |
|
{ |
|
"epoch": 6.66, |
|
"learning_rate": 7.648590663866274e-05, |
|
"loss": 0.6843, |
|
"step": 34500 |
|
}, |
|
{ |
|
"epoch": 6.71, |
|
"learning_rate": 7.449096493568466e-05, |
|
"loss": 0.6832, |
|
"step": 34750 |
|
}, |
|
{ |
|
"epoch": 6.76, |
|
"learning_rate": 7.25137548581695e-05, |
|
"loss": 0.6816, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 6.81, |
|
"learning_rate": 7.055474071013765e-05, |
|
"loss": 0.6814, |
|
"step": 35250 |
|
}, |
|
{ |
|
"epoch": 6.86, |
|
"learning_rate": 6.861438252269817e-05, |
|
"loss": 0.6788, |
|
"step": 35500 |
|
}, |
|
{ |
|
"epoch": 6.91, |
|
"learning_rate": 6.669313594602076e-05, |
|
"loss": 0.681, |
|
"step": 35750 |
|
}, |
|
{ |
|
"epoch": 6.95, |
|
"learning_rate": 6.479145214233655e-05, |
|
"loss": 0.6787, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"eval_alliteration_score": 0.419920073778051, |
|
"eval_harmonic_meter_score": 0.10124233945058883, |
|
"eval_harmonic_rhyme_score": 0.4258480329916534, |
|
"eval_meter_score": 0.3190283478146982, |
|
"eval_rhyme_score": 0.7980873008918292, |
|
"eval_runtime": 2453.1033, |
|
"eval_samples_per_second": 1.101, |
|
"eval_steps_per_second": 0.138, |
|
"step": 36239 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"learning_rate": 6.290977767999224e-05, |
|
"loss": 0.676, |
|
"step": 36250 |
|
}, |
|
{ |
|
"epoch": 7.05, |
|
"learning_rate": 6.10485544285837e-05, |
|
"loss": 0.6422, |
|
"step": 36500 |
|
}, |
|
{ |
|
"epoch": 7.1, |
|
"learning_rate": 5.920821945519259e-05, |
|
"loss": 0.6424, |
|
"step": 36750 |
|
}, |
|
{ |
|
"epoch": 7.15, |
|
"learning_rate": 5.7389204921750964e-05, |
|
"loss": 0.6446, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 7.2, |
|
"learning_rate": 5.5591937983557386e-05, |
|
"loss": 0.6426, |
|
"step": 37250 |
|
}, |
|
{ |
|
"epoch": 7.24, |
|
"learning_rate": 5.381684068896904e-05, |
|
"loss": 0.6429, |
|
"step": 37500 |
|
}, |
|
{ |
|
"epoch": 7.29, |
|
"learning_rate": 5.2064329880293066e-05, |
|
"loss": 0.6415, |
|
"step": 37750 |
|
}, |
|
{ |
|
"epoch": 7.34, |
|
"learning_rate": 5.0334817095900465e-05, |
|
"loss": 0.6416, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 7.39, |
|
"learning_rate": 4.862870847358517e-05, |
|
"loss": 0.6414, |
|
"step": 38250 |
|
}, |
|
{ |
|
"epoch": 7.44, |
|
"learning_rate": 4.694640465519168e-05, |
|
"loss": 0.6424, |
|
"step": 38500 |
|
}, |
|
{ |
|
"epoch": 7.48, |
|
"learning_rate": 4.528830069253341e-05, |
|
"loss": 0.6407, |
|
"step": 38750 |
|
}, |
|
{ |
|
"epoch": 7.53, |
|
"learning_rate": 4.365478595462316e-05, |
|
"loss": 0.6414, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 7.58, |
|
"learning_rate": 4.2046244036238635e-05, |
|
"loss": 0.6394, |
|
"step": 39250 |
|
}, |
|
{ |
|
"epoch": 7.63, |
|
"learning_rate": 4.0463052667843626e-05, |
|
"loss": 0.6377, |
|
"step": 39500 |
|
}, |
|
{ |
|
"epoch": 7.68, |
|
"learning_rate": 3.8905583626886546e-05, |
|
"loss": 0.6384, |
|
"step": 39750 |
|
}, |
|
{ |
|
"epoch": 7.73, |
|
"learning_rate": 3.7374202650496385e-05, |
|
"loss": 0.6387, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 7.77, |
|
"learning_rate": 3.586926934959779e-05, |
|
"loss": 0.6388, |
|
"step": 40250 |
|
}, |
|
{ |
|
"epoch": 7.82, |
|
"learning_rate": 3.439113712446411e-05, |
|
"loss": 0.6369, |
|
"step": 40500 |
|
}, |
|
{ |
|
"epoch": 7.87, |
|
"learning_rate": 3.294015308172949e-05, |
|
"loss": 0.637, |
|
"step": 40750 |
|
}, |
|
{ |
|
"epoch": 7.92, |
|
"learning_rate": 3.151665795287821e-05, |
|
"loss": 0.6353, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 7.97, |
|
"learning_rate": 3.0120986014231658e-05, |
|
"loss": 0.636, |
|
"step": 41250 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_alliteration_score": 0.44240757714634893, |
|
"eval_harmonic_meter_score": 0.11214598745179055, |
|
"eval_harmonic_rhyme_score": 0.4118503741406305, |
|
"eval_meter_score": 0.3255113242746067, |
|
"eval_rhyme_score": 0.7902240726132346, |
|
"eval_runtime": 2501.3058, |
|
"eval_samples_per_second": 1.079, |
|
"eval_steps_per_second": 0.135, |
|
"step": 41416 |
|
}, |
|
{ |
|
"epoch": 8.02, |
|
"learning_rate": 2.8753465008450878e-05, |
|
"loss": 0.613, |
|
"step": 41500 |
|
}, |
|
{ |
|
"epoch": 8.06, |
|
"learning_rate": 2.7414416067573497e-05, |
|
"loss": 0.6104, |
|
"step": 41750 |
|
}, |
|
{ |
|
"epoch": 8.11, |
|
"learning_rate": 2.6104153637602887e-05, |
|
"loss": 0.6121, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 8.16, |
|
"learning_rate": 2.4822985404667573e-05, |
|
"loss": 0.6116, |
|
"step": 42250 |
|
}, |
|
{ |
|
"epoch": 8.21, |
|
"learning_rate": 2.3571212222767847e-05, |
|
"loss": 0.6112, |
|
"step": 42500 |
|
}, |
|
{ |
|
"epoch": 8.26, |
|
"learning_rate": 2.2349128043127078e-05, |
|
"loss": 0.61, |
|
"step": 42750 |
|
}, |
|
{ |
|
"epoch": 8.31, |
|
"learning_rate": 2.115701984516349e-05, |
|
"loss": 0.611, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 8.35, |
|
"learning_rate": 1.999516756909961e-05, |
|
"loss": 0.6095, |
|
"step": 43250 |
|
}, |
|
{ |
|
"epoch": 8.4, |
|
"learning_rate": 1.886384405022437e-05, |
|
"loss": 0.6114, |
|
"step": 43500 |
|
}, |
|
{ |
|
"epoch": 8.45, |
|
"learning_rate": 1.7763314954823966e-05, |
|
"loss": 0.6113, |
|
"step": 43750 |
|
}, |
|
{ |
|
"epoch": 8.5, |
|
"learning_rate": 1.669383871779581e-05, |
|
"loss": 0.6121, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 8.55, |
|
"learning_rate": 1.5655666481960795e-05, |
|
"loss": 0.6104, |
|
"step": 44250 |
|
}, |
|
{ |
|
"epoch": 8.6, |
|
"learning_rate": 1.4649042039088254e-05, |
|
"loss": 0.6114, |
|
"step": 44500 |
|
}, |
|
{ |
|
"epoch": 8.64, |
|
"learning_rate": 1.3674201772646525e-05, |
|
"loss": 0.6089, |
|
"step": 44750 |
|
}, |
|
{ |
|
"epoch": 8.69, |
|
"learning_rate": 1.2731374602293753e-05, |
|
"loss": 0.6079, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 8.74, |
|
"learning_rate": 1.1820781930120982e-05, |
|
"loss": 0.6103, |
|
"step": 45250 |
|
}, |
|
{ |
|
"epoch": 8.79, |
|
"learning_rate": 1.0942637588661024e-05, |
|
"loss": 0.6087, |
|
"step": 45500 |
|
}, |
|
{ |
|
"epoch": 8.84, |
|
"learning_rate": 1.0097147790674365e-05, |
|
"loss": 0.6091, |
|
"step": 45750 |
|
}, |
|
{ |
|
"epoch": 8.89, |
|
"learning_rate": 9.28451108072481e-06, |
|
"loss": 0.6088, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 8.93, |
|
"learning_rate": 8.504918288555651e-06, |
|
"loss": 0.6089, |
|
"step": 46250 |
|
}, |
|
{ |
|
"epoch": 8.98, |
|
"learning_rate": 7.758552484277575e-06, |
|
"loss": 0.6089, |
|
"step": 46500 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"eval_alliteration_score": 0.4335748792270531, |
|
"eval_harmonic_meter_score": 0.1091721650313209, |
|
"eval_harmonic_rhyme_score": 0.4075133294099665, |
|
"eval_meter_score": 0.3241651625589595, |
|
"eval_rhyme_score": 0.7938227877904658, |
|
"eval_runtime": 2386.2391, |
|
"eval_samples_per_second": 1.131, |
|
"eval_steps_per_second": 0.142, |
|
"step": 46593 |
|
}, |
|
{ |
|
"epoch": 9.03, |
|
"learning_rate": 7.045588935378604e-06, |
|
"loss": 0.6036, |
|
"step": 46750 |
|
}, |
|
{ |
|
"epoch": 9.08, |
|
"learning_rate": 6.366195065566454e-06, |
|
"loss": 0.5958, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 9.13, |
|
"learning_rate": 5.7205304154527955e-06, |
|
"loss": 0.5959, |
|
"step": 47250 |
|
}, |
|
{ |
|
"epoch": 9.18, |
|
"learning_rate": 5.108746605088709e-06, |
|
"loss": 0.5972, |
|
"step": 47500 |
|
}, |
|
{ |
|
"epoch": 9.22, |
|
"learning_rate": 4.530987298359967e-06, |
|
"loss": 0.5972, |
|
"step": 47750 |
|
}, |
|
{ |
|
"epoch": 9.27, |
|
"learning_rate": 3.987388169250816e-06, |
|
"loss": 0.599, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 9.32, |
|
"learning_rate": 3.4780768699839313e-06, |
|
"loss": 0.5964, |
|
"step": 48250 |
|
}, |
|
{ |
|
"epoch": 9.37, |
|
"learning_rate": 3.0031730010441957e-06, |
|
"loss": 0.5978, |
|
"step": 48500 |
|
}, |
|
{ |
|
"epoch": 9.42, |
|
"learning_rate": 2.562788083092998e-06, |
|
"loss": 0.5971, |
|
"step": 48750 |
|
}, |
|
{ |
|
"epoch": 9.46, |
|
"learning_rate": 2.1570255307802486e-06, |
|
"loss": 0.5957, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 9.51, |
|
"learning_rate": 1.785980628459638e-06, |
|
"loss": 0.5962, |
|
"step": 49250 |
|
}, |
|
{ |
|
"epoch": 9.56, |
|
"learning_rate": 1.4497405078132517e-06, |
|
"loss": 0.5976, |
|
"step": 49500 |
|
}, |
|
{ |
|
"epoch": 9.61, |
|
"learning_rate": 1.1483841273906391e-06, |
|
"loss": 0.5977, |
|
"step": 49750 |
|
}, |
|
{ |
|
"epoch": 9.66, |
|
"learning_rate": 8.819822540670762e-07, |
|
"loss": 0.596, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 9.71, |
|
"learning_rate": 6.505974464255936e-07, |
|
"loss": 0.5967, |
|
"step": 50250 |
|
}, |
|
{ |
|
"epoch": 9.75, |
|
"learning_rate": 4.542840400665104e-07, |
|
"loss": 0.5965, |
|
"step": 50500 |
|
}, |
|
{ |
|
"epoch": 9.8, |
|
"learning_rate": 2.9308813484780755e-07, |
|
"loss": 0.5977, |
|
"step": 50750 |
|
}, |
|
{ |
|
"epoch": 9.85, |
|
"learning_rate": 1.670475840597041e-07, |
|
"loss": 0.5965, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 9.9, |
|
"learning_rate": 7.619198553558414e-08, |
|
"loss": 0.5948, |
|
"step": 51250 |
|
}, |
|
{ |
|
"epoch": 9.95, |
|
"learning_rate": 2.0542674701656824e-08, |
|
"loss": 0.598, |
|
"step": 51500 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"learning_rate": 1.1271956673053472e-10, |
|
"loss": 0.5966, |
|
"step": 51750 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"eval_alliteration_score": 0.43619909502262444, |
|
"eval_harmonic_meter_score": 0.10679777933598275, |
|
"eval_harmonic_rhyme_score": 0.4260903757100796, |
|
"eval_meter_score": 0.31873257964794116, |
|
"eval_rhyme_score": 0.8030446845582035, |
|
"eval_runtime": 2387.7503, |
|
"eval_samples_per_second": 1.131, |
|
"eval_steps_per_second": 0.142, |
|
"step": 51770 |
|
}, |
|
{ |
|
"epoch": 10.0, |
|
"step": 51770, |
|
"total_flos": 2.202815059032277e+18, |
|
"train_loss": 0.12071779469207651, |
|
"train_runtime": 12056.8595, |
|
"train_samples_per_second": 549.652, |
|
"train_steps_per_second": 4.294 |
|
} |
|
], |
|
"max_steps": 51770, |
|
"num_train_epochs": 10, |
|
"total_flos": 2.202815059032277e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|