| { | |
| "best_metric": 2.4780548434708303, | |
| "best_model_checkpoint": "./final-whisper-for-initial-publish-v2/checkpoint-8000", | |
| "epoch": 18.12191103789127, | |
| "eval_steps": 500, | |
| "global_step": 11000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04118616144975288, | |
| "grad_norm": 67.08155822753906, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 2.8368, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08237232289950576, | |
| "grad_norm": 17.44784164428711, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 2.0282, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12355848434925865, | |
| "grad_norm": 12.481098175048828, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.3152, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16474464579901152, | |
| "grad_norm": 8.900260925292969, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.7248, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20593080724876442, | |
| "grad_norm": 7.8956146240234375, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.5731, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2471169686985173, | |
| "grad_norm": 9.542287826538086, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.474, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2883031301482702, | |
| "grad_norm": 6.937229633331299, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.4301, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.32948929159802304, | |
| "grad_norm": 6.919783115386963, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.351, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.37067545304777594, | |
| "grad_norm": 6.427859783172607, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.2726, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.41186161449752884, | |
| "grad_norm": 5.83529806137085, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.1663, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.45304777594728174, | |
| "grad_norm": 2.4859721660614014, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.0893, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.4942339373970346, | |
| "grad_norm": 3.2526772022247314, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.076, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5354200988467874, | |
| "grad_norm": 4.417870044708252, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.0764, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5766062602965404, | |
| "grad_norm": 3.149733781814575, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.0679, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6177924217462932, | |
| "grad_norm": 3.8823468685150146, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.0525, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.6589785831960461, | |
| "grad_norm": 6.1009745597839355, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.0552, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.700164744645799, | |
| "grad_norm": 5.714720726013184, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.0529, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.7413509060955519, | |
| "grad_norm": 4.403444290161133, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.0472, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7825370675453048, | |
| "grad_norm": 3.7249884605407715, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.0424, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.8237232289950577, | |
| "grad_norm": 3.8490548133850098, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.0461, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8237232289950577, | |
| "eval_loss": 0.04240331053733826, | |
| "eval_runtime": 1653.3644, | |
| "eval_samples_per_second": 1.468, | |
| "eval_steps_per_second": 0.184, | |
| "eval_wer": 11.04637613609881, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8649093904448105, | |
| "grad_norm": 2.538515329360962, | |
| "learning_rate": 9.961818181818183e-06, | |
| "loss": 0.0384, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.9060955518945635, | |
| "grad_norm": 1.6598188877105713, | |
| "learning_rate": 9.916363636363637e-06, | |
| "loss": 0.0375, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.9472817133443163, | |
| "grad_norm": 3.256746292114258, | |
| "learning_rate": 9.870909090909092e-06, | |
| "loss": 0.0346, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.9884678747940692, | |
| "grad_norm": 3.04457426071167, | |
| "learning_rate": 9.825454545454546e-06, | |
| "loss": 0.034, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.029654036243822, | |
| "grad_norm": 0.6082770228385925, | |
| "learning_rate": 9.780000000000001e-06, | |
| "loss": 0.0272, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.0708401976935749, | |
| "grad_norm": 1.6969329118728638, | |
| "learning_rate": 9.734545454545455e-06, | |
| "loss": 0.0206, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.112026359143328, | |
| "grad_norm": 1.9652377367019653, | |
| "learning_rate": 9.68909090909091e-06, | |
| "loss": 0.0197, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.1532125205930808, | |
| "grad_norm": 2.402329206466675, | |
| "learning_rate": 9.643636363636364e-06, | |
| "loss": 0.0221, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.1943986820428336, | |
| "grad_norm": 3.167133092880249, | |
| "learning_rate": 9.598181818181818e-06, | |
| "loss": 0.0176, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.2355848434925865, | |
| "grad_norm": 0.4075870215892792, | |
| "learning_rate": 9.552727272727273e-06, | |
| "loss": 0.0165, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.2767710049423393, | |
| "grad_norm": 2.739176034927368, | |
| "learning_rate": 9.507272727272729e-06, | |
| "loss": 0.0156, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.3179571663920924, | |
| "grad_norm": 2.4662702083587646, | |
| "learning_rate": 9.461818181818183e-06, | |
| "loss": 0.0133, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.3591433278418452, | |
| "grad_norm": 2.955909252166748, | |
| "learning_rate": 9.416363636363636e-06, | |
| "loss": 0.0184, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.400329489291598, | |
| "grad_norm": 2.0838463306427, | |
| "learning_rate": 9.370909090909092e-06, | |
| "loss": 0.0138, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.441515650741351, | |
| "grad_norm": 2.2245779037475586, | |
| "learning_rate": 9.325454545454547e-06, | |
| "loss": 0.0132, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.4827018121911038, | |
| "grad_norm": 2.146961212158203, | |
| "learning_rate": 9.280000000000001e-06, | |
| "loss": 0.0126, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.5238879736408566, | |
| "grad_norm": 1.955930471420288, | |
| "learning_rate": 9.234545454545455e-06, | |
| "loss": 0.0111, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.5650741350906094, | |
| "grad_norm": 2.04664945602417, | |
| "learning_rate": 9.18909090909091e-06, | |
| "loss": 0.0118, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.6062602965403623, | |
| "grad_norm": 2.1349849700927734, | |
| "learning_rate": 9.143636363636364e-06, | |
| "loss": 0.0116, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.6474464579901154, | |
| "grad_norm": 2.336024761199951, | |
| "learning_rate": 9.09818181818182e-06, | |
| "loss": 0.0124, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.6474464579901154, | |
| "eval_loss": 0.02211674302816391, | |
| "eval_runtime": 1633.7086, | |
| "eval_samples_per_second": 1.486, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 5.11147362697118, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.6886326194398682, | |
| "grad_norm": 0.4564249813556671, | |
| "learning_rate": 9.052727272727273e-06, | |
| "loss": 0.0119, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.729818780889621, | |
| "grad_norm": 1.9494842290878296, | |
| "learning_rate": 9.007272727272729e-06, | |
| "loss": 0.0114, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.771004942339374, | |
| "grad_norm": 1.7129623889923096, | |
| "learning_rate": 8.961818181818182e-06, | |
| "loss": 0.0116, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.812191103789127, | |
| "grad_norm": 0.8837174773216248, | |
| "learning_rate": 8.916363636363638e-06, | |
| "loss": 0.01, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.8533772652388798, | |
| "grad_norm": 3.1779348850250244, | |
| "learning_rate": 8.870909090909091e-06, | |
| "loss": 0.0087, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.8945634266886326, | |
| "grad_norm": 1.1833815574645996, | |
| "learning_rate": 8.825454545454545e-06, | |
| "loss": 0.0122, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.9357495881383855, | |
| "grad_norm": 2.3456814289093018, | |
| "learning_rate": 8.78e-06, | |
| "loss": 0.0089, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.9769357495881383, | |
| "grad_norm": 1.5834710597991943, | |
| "learning_rate": 8.734545454545456e-06, | |
| "loss": 0.0111, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.018121911037891, | |
| "grad_norm": 1.1926366090774536, | |
| "learning_rate": 8.68909090909091e-06, | |
| "loss": 0.0069, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.059308072487644, | |
| "grad_norm": 0.5671884417533875, | |
| "learning_rate": 8.643636363636364e-06, | |
| "loss": 0.0043, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.100494233937397, | |
| "grad_norm": 0.8897690176963806, | |
| "learning_rate": 8.598181818181819e-06, | |
| "loss": 0.0034, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.1416803953871497, | |
| "grad_norm": 1.6989620923995972, | |
| "learning_rate": 8.552727272727274e-06, | |
| "loss": 0.003, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.182866556836903, | |
| "grad_norm": 2.0327303409576416, | |
| "learning_rate": 8.507272727272728e-06, | |
| "loss": 0.0037, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.224052718286656, | |
| "grad_norm": 0.6554267406463623, | |
| "learning_rate": 8.461818181818182e-06, | |
| "loss": 0.0034, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.2652388797364087, | |
| "grad_norm": 0.21176663041114807, | |
| "learning_rate": 8.416363636363637e-06, | |
| "loss": 0.0046, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.3064250411861615, | |
| "grad_norm": 1.2915725708007812, | |
| "learning_rate": 8.370909090909091e-06, | |
| "loss": 0.0026, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.3476112026359144, | |
| "grad_norm": 0.15533868968486786, | |
| "learning_rate": 8.325454545454547e-06, | |
| "loss": 0.0043, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.3887973640856672, | |
| "grad_norm": 0.19825848937034607, | |
| "learning_rate": 8.28e-06, | |
| "loss": 0.0042, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.42998352553542, | |
| "grad_norm": 0.14249640703201294, | |
| "learning_rate": 8.234545454545456e-06, | |
| "loss": 0.0023, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.471169686985173, | |
| "grad_norm": 0.12349322438240051, | |
| "learning_rate": 8.18909090909091e-06, | |
| "loss": 0.0026, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.471169686985173, | |
| "eval_loss": 0.015178721398115158, | |
| "eval_runtime": 1647.8725, | |
| "eval_samples_per_second": 1.473, | |
| "eval_steps_per_second": 0.184, | |
| "eval_wer": 3.5966752116833685, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.5123558484349258, | |
| "grad_norm": 0.24572087824344635, | |
| "learning_rate": 8.143636363636365e-06, | |
| "loss": 0.0032, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.5535420098846786, | |
| "grad_norm": 0.2075294703245163, | |
| "learning_rate": 8.098181818181819e-06, | |
| "loss": 0.0023, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.594728171334432, | |
| "grad_norm": 0.264713317155838, | |
| "learning_rate": 8.052727272727272e-06, | |
| "loss": 0.0044, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.6359143327841847, | |
| "grad_norm": 1.9720451831817627, | |
| "learning_rate": 8.007272727272728e-06, | |
| "loss": 0.0048, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.6771004942339376, | |
| "grad_norm": 0.7824451327323914, | |
| "learning_rate": 7.961818181818183e-06, | |
| "loss": 0.0048, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.7182866556836904, | |
| "grad_norm": 0.681707501411438, | |
| "learning_rate": 7.916363636363637e-06, | |
| "loss": 0.0036, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.7594728171334433, | |
| "grad_norm": 0.39950481057167053, | |
| "learning_rate": 7.870909090909091e-06, | |
| "loss": 0.0047, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.800658978583196, | |
| "grad_norm": 0.7777596712112427, | |
| "learning_rate": 7.825454545454546e-06, | |
| "loss": 0.0029, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.841845140032949, | |
| "grad_norm": 0.06510231643915176, | |
| "learning_rate": 7.78e-06, | |
| "loss": 0.0067, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.883031301482702, | |
| "grad_norm": 2.2729947566986084, | |
| "learning_rate": 7.734545454545455e-06, | |
| "loss": 0.0024, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.9242174629324547, | |
| "grad_norm": 1.6635749340057373, | |
| "learning_rate": 7.68909090909091e-06, | |
| "loss": 0.0035, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.9654036243822075, | |
| "grad_norm": 0.06815456598997116, | |
| "learning_rate": 7.643636363636365e-06, | |
| "loss": 0.004, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.0065897858319603, | |
| "grad_norm": 1.5607892274856567, | |
| "learning_rate": 7.598181818181819e-06, | |
| "loss": 0.0046, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.047775947281713, | |
| "grad_norm": 0.8395469784736633, | |
| "learning_rate": 7.552727272727274e-06, | |
| "loss": 0.0018, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.088962108731466, | |
| "grad_norm": 0.04032081738114357, | |
| "learning_rate": 7.507272727272728e-06, | |
| "loss": 0.0011, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.130148270181219, | |
| "grad_norm": 0.3644678294658661, | |
| "learning_rate": 7.461818181818182e-06, | |
| "loss": 0.0014, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.171334431630972, | |
| "grad_norm": 0.35729554295539856, | |
| "learning_rate": 7.416363636363637e-06, | |
| "loss": 0.0009, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.212520593080725, | |
| "grad_norm": 0.05964287370443344, | |
| "learning_rate": 7.370909090909092e-06, | |
| "loss": 0.0014, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.253706754530478, | |
| "grad_norm": 0.799541711807251, | |
| "learning_rate": 7.325454545454546e-06, | |
| "loss": 0.0015, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.2948929159802307, | |
| "grad_norm": 0.058581918478012085, | |
| "learning_rate": 7.280000000000001e-06, | |
| "loss": 0.0008, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.2948929159802307, | |
| "eval_loss": 0.015332825481891632, | |
| "eval_runtime": 1629.67, | |
| "eval_samples_per_second": 1.489, | |
| "eval_steps_per_second": 0.187, | |
| "eval_wer": 3.24710634661695, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.3360790774299836, | |
| "grad_norm": 0.20925483107566833, | |
| "learning_rate": 7.234545454545455e-06, | |
| "loss": 0.001, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.3772652388797364, | |
| "grad_norm": 0.5220603346824646, | |
| "learning_rate": 7.18909090909091e-06, | |
| "loss": 0.0006, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.4184514003294892, | |
| "grad_norm": 0.33813387155532837, | |
| "learning_rate": 7.1436363636363635e-06, | |
| "loss": 0.0008, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.459637561779242, | |
| "grad_norm": 0.5609467625617981, | |
| "learning_rate": 7.098181818181818e-06, | |
| "loss": 0.0006, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.500823723228995, | |
| "grad_norm": 0.028300393372774124, | |
| "learning_rate": 7.052727272727274e-06, | |
| "loss": 0.0008, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.5420098846787478, | |
| "grad_norm": 0.056196194142103195, | |
| "learning_rate": 7.007272727272728e-06, | |
| "loss": 0.0009, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.583196046128501, | |
| "grad_norm": 0.202606201171875, | |
| "learning_rate": 6.961818181818183e-06, | |
| "loss": 0.0019, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.624382207578254, | |
| "grad_norm": 0.014030909165740013, | |
| "learning_rate": 6.9163636363636365e-06, | |
| "loss": 0.001, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.6655683690280068, | |
| "grad_norm": 0.08081363141536713, | |
| "learning_rate": 6.870909090909091e-06, | |
| "loss": 0.002, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.7067545304777596, | |
| "grad_norm": 0.6110771894454956, | |
| "learning_rate": 6.8254545454545466e-06, | |
| "loss": 0.0005, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.7479406919275124, | |
| "grad_norm": 0.30081290006637573, | |
| "learning_rate": 6.780000000000001e-06, | |
| "loss": 0.0021, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.7891268533772653, | |
| "grad_norm": 0.042301613837480545, | |
| "learning_rate": 6.734545454545455e-06, | |
| "loss": 0.0012, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.830313014827018, | |
| "grad_norm": 1.5658293962478638, | |
| "learning_rate": 6.6890909090909095e-06, | |
| "loss": 0.0018, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.871499176276771, | |
| "grad_norm": 0.20856991410255432, | |
| "learning_rate": 6.643636363636364e-06, | |
| "loss": 0.0017, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.912685337726524, | |
| "grad_norm": 0.19910617172718048, | |
| "learning_rate": 6.5981818181818195e-06, | |
| "loss": 0.0011, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.9538714991762767, | |
| "grad_norm": 0.1468815952539444, | |
| "learning_rate": 6.5527272727272724e-06, | |
| "loss": 0.0011, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.9950576606260295, | |
| "grad_norm": 0.06176592409610748, | |
| "learning_rate": 6.507272727272728e-06, | |
| "loss": 0.0009, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.036243822075782, | |
| "grad_norm": 0.013582455925643444, | |
| "learning_rate": 6.4618181818181825e-06, | |
| "loss": 0.0004, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.077429983525535, | |
| "grad_norm": 0.013000385835766792, | |
| "learning_rate": 6.416363636363637e-06, | |
| "loss": 0.0003, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.118616144975288, | |
| "grad_norm": 0.04700480401515961, | |
| "learning_rate": 6.370909090909091e-06, | |
| "loss": 0.0004, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.118616144975288, | |
| "eval_loss": 0.013498911634087563, | |
| "eval_runtime": 1637.8085, | |
| "eval_samples_per_second": 1.482, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 2.718868950516585, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.159802306425041, | |
| "grad_norm": 0.049776773899793625, | |
| "learning_rate": 6.325454545454545e-06, | |
| "loss": 0.0007, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.200988467874794, | |
| "grad_norm": 0.047507863491773605, | |
| "learning_rate": 6.280000000000001e-06, | |
| "loss": 0.001, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.242174629324547, | |
| "grad_norm": 0.058455199003219604, | |
| "learning_rate": 6.2345454545454555e-06, | |
| "loss": 0.0004, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.283360790774299, | |
| "grad_norm": 0.03676324710249901, | |
| "learning_rate": 6.18909090909091e-06, | |
| "loss": 0.0008, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.324546952224052, | |
| "grad_norm": 0.08864962309598923, | |
| "learning_rate": 6.143636363636364e-06, | |
| "loss": 0.0008, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.365733113673806, | |
| "grad_norm": 0.25547558069229126, | |
| "learning_rate": 6.098181818181818e-06, | |
| "loss": 0.0014, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.406919275123559, | |
| "grad_norm": 0.014656963758170605, | |
| "learning_rate": 6.052727272727274e-06, | |
| "loss": 0.0004, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.448105436573312, | |
| "grad_norm": 0.01910262741148472, | |
| "learning_rate": 6.0072727272727284e-06, | |
| "loss": 0.0004, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.4892915980230645, | |
| "grad_norm": 0.021646469831466675, | |
| "learning_rate": 5.961818181818182e-06, | |
| "loss": 0.0005, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.530477759472817, | |
| "grad_norm": 0.024704741314053535, | |
| "learning_rate": 5.916363636363637e-06, | |
| "loss": 0.0005, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.57166392092257, | |
| "grad_norm": 0.024691058322787285, | |
| "learning_rate": 5.870909090909091e-06, | |
| "loss": 0.0004, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.612850082372323, | |
| "grad_norm": 0.023056305944919586, | |
| "learning_rate": 5.825454545454546e-06, | |
| "loss": 0.0002, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.654036243822076, | |
| "grad_norm": 0.02144230715930462, | |
| "learning_rate": 5.78e-06, | |
| "loss": 0.0006, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.695222405271829, | |
| "grad_norm": 0.056895911693573, | |
| "learning_rate": 5.734545454545455e-06, | |
| "loss": 0.0006, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.736408566721582, | |
| "grad_norm": 0.026411665603518486, | |
| "learning_rate": 5.68909090909091e-06, | |
| "loss": 0.0002, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.7775947281713345, | |
| "grad_norm": 0.6490919589996338, | |
| "learning_rate": 5.643636363636364e-06, | |
| "loss": 0.0005, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.818780889621087, | |
| "grad_norm": 0.01870059408247471, | |
| "learning_rate": 5.598181818181818e-06, | |
| "loss": 0.0004, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.85996705107084, | |
| "grad_norm": 0.36998653411865234, | |
| "learning_rate": 5.552727272727273e-06, | |
| "loss": 0.001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.901153212520593, | |
| "grad_norm": 0.09682920575141907, | |
| "learning_rate": 5.507272727272728e-06, | |
| "loss": 0.0008, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.942339373970346, | |
| "grad_norm": 0.12028121948242188, | |
| "learning_rate": 5.461818181818183e-06, | |
| "loss": 0.0005, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.942339373970346, | |
| "eval_loss": 0.013804786838591099, | |
| "eval_runtime": 1636.262, | |
| "eval_samples_per_second": 1.483, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 2.8043191175328204, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.983525535420099, | |
| "grad_norm": 0.36403629183769226, | |
| "learning_rate": 5.4163636363636365e-06, | |
| "loss": 0.0002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.0247116968698515, | |
| "grad_norm": 0.020813830196857452, | |
| "learning_rate": 5.370909090909091e-06, | |
| "loss": 0.0007, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.065897858319604, | |
| "grad_norm": 0.0156087102368474, | |
| "learning_rate": 5.325454545454546e-06, | |
| "loss": 0.0002, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 5.107084019769357, | |
| "grad_norm": 0.08841024339199066, | |
| "learning_rate": 5.28e-06, | |
| "loss": 0.0003, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.14827018121911, | |
| "grad_norm": 0.09709523618221283, | |
| "learning_rate": 5.234545454545456e-06, | |
| "loss": 0.0002, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 5.189456342668863, | |
| "grad_norm": 0.006491546519100666, | |
| "learning_rate": 5.1890909090909095e-06, | |
| "loss": 0.0004, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.230642504118616, | |
| "grad_norm": 0.03339486941695213, | |
| "learning_rate": 5.143636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.2718286655683695, | |
| "grad_norm": 0.019623372703790665, | |
| "learning_rate": 5.098181818181819e-06, | |
| "loss": 0.0002, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.313014827018122, | |
| "grad_norm": 0.012536031194031239, | |
| "learning_rate": 5.052727272727273e-06, | |
| "loss": 0.0004, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.354200988467875, | |
| "grad_norm": 0.00885727908462286, | |
| "learning_rate": 5.007272727272727e-06, | |
| "loss": 0.0001, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.395387149917628, | |
| "grad_norm": 0.012449279427528381, | |
| "learning_rate": 4.9618181818181824e-06, | |
| "loss": 0.0003, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.436573311367381, | |
| "grad_norm": 0.04439382627606392, | |
| "learning_rate": 4.916363636363637e-06, | |
| "loss": 0.0003, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.477759472817134, | |
| "grad_norm": 0.08744444698095322, | |
| "learning_rate": 4.870909090909091e-06, | |
| "loss": 0.0006, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.518945634266887, | |
| "grad_norm": 0.015908684581518173, | |
| "learning_rate": 4.825454545454546e-06, | |
| "loss": 0.0003, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.560131795716639, | |
| "grad_norm": 0.04740598052740097, | |
| "learning_rate": 4.78e-06, | |
| "loss": 0.0002, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.601317957166392, | |
| "grad_norm": 0.011928090825676918, | |
| "learning_rate": 4.734545454545455e-06, | |
| "loss": 0.0001, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.642504118616145, | |
| "grad_norm": 0.044589392840862274, | |
| "learning_rate": 4.689090909090909e-06, | |
| "loss": 0.0002, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.683690280065898, | |
| "grad_norm": 0.010786905884742737, | |
| "learning_rate": 4.643636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.724876441515651, | |
| "grad_norm": 0.006343629211187363, | |
| "learning_rate": 4.598181818181818e-06, | |
| "loss": 0.0003, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.766062602965404, | |
| "grad_norm": 0.006056615151464939, | |
| "learning_rate": 4.552727272727273e-06, | |
| "loss": 0.0003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.766062602965404, | |
| "eval_loss": 0.013986092992126942, | |
| "eval_runtime": 1723.588, | |
| "eval_samples_per_second": 1.408, | |
| "eval_steps_per_second": 0.176, | |
| "eval_wer": 2.734405344519537, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.8072487644151565, | |
| "grad_norm": 0.043298039585351944, | |
| "learning_rate": 4.5072727272727275e-06, | |
| "loss": 0.0002, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.848434925864909, | |
| "grad_norm": 0.007331861648708582, | |
| "learning_rate": 4.461818181818182e-06, | |
| "loss": 0.0001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.889621087314662, | |
| "grad_norm": 0.012765350751578808, | |
| "learning_rate": 4.416363636363637e-06, | |
| "loss": 0.0002, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.930807248764415, | |
| "grad_norm": 0.033175356686115265, | |
| "learning_rate": 4.370909090909091e-06, | |
| "loss": 0.0004, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.971993410214168, | |
| "grad_norm": 0.005575309973210096, | |
| "learning_rate": 4.325454545454546e-06, | |
| "loss": 0.0001, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 6.013179571663921, | |
| "grad_norm": 0.005467335693538189, | |
| "learning_rate": 4.2800000000000005e-06, | |
| "loss": 0.0003, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 6.0543657331136735, | |
| "grad_norm": 0.0050591155886650085, | |
| "learning_rate": 4.234545454545455e-06, | |
| "loss": 0.0001, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 6.095551894563426, | |
| "grad_norm": 0.004481570329517126, | |
| "learning_rate": 4.18909090909091e-06, | |
| "loss": 0.0001, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.136738056013179, | |
| "grad_norm": 0.002914627082645893, | |
| "learning_rate": 4.143636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 6.177924217462932, | |
| "grad_norm": 0.004947616718709469, | |
| "learning_rate": 4.098181818181818e-06, | |
| "loss": 0.0001, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 6.219110378912685, | |
| "grad_norm": 0.003795625874772668, | |
| "learning_rate": 4.0527272727272735e-06, | |
| "loss": 0.0001, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.260296540362438, | |
| "grad_norm": 0.025668440386652946, | |
| "learning_rate": 4.007272727272727e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.3014827018121915, | |
| "grad_norm": 0.01409939769655466, | |
| "learning_rate": 3.961818181818182e-06, | |
| "loss": 0.0001, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.342668863261944, | |
| "grad_norm": 0.0054998998530209064, | |
| "learning_rate": 3.9163636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.383855024711697, | |
| "grad_norm": 0.0049223750829696655, | |
| "learning_rate": 3.870909090909091e-06, | |
| "loss": 0.0001, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.42504118616145, | |
| "grad_norm": 0.0028136519249528646, | |
| "learning_rate": 3.825454545454546e-06, | |
| "loss": 0.0001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.466227347611203, | |
| "grad_norm": 0.005505116190761328, | |
| "learning_rate": 3.7800000000000002e-06, | |
| "loss": 0.0001, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.507413509060956, | |
| "grad_norm": 0.007019472308456898, | |
| "learning_rate": 3.7345454545454544e-06, | |
| "loss": 0.0001, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.548599670510709, | |
| "grad_norm": 0.006632586009800434, | |
| "learning_rate": 3.6890909090909094e-06, | |
| "loss": 0.0001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.589785831960461, | |
| "grad_norm": 0.002231697551906109, | |
| "learning_rate": 3.643636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.589785831960461, | |
| "eval_loss": 0.013741781935095787, | |
| "eval_runtime": 1710.3123, | |
| "eval_samples_per_second": 1.419, | |
| "eval_steps_per_second": 0.178, | |
| "eval_wer": 2.532432222481162, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.630971993410214, | |
| "grad_norm": 0.008344273082911968, | |
| "learning_rate": 3.5981818181818186e-06, | |
| "loss": 0.0, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.672158154859967, | |
| "grad_norm": 0.0035063326358795166, | |
| "learning_rate": 3.552727272727273e-06, | |
| "loss": 0.0, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.71334431630972, | |
| "grad_norm": 0.007493205368518829, | |
| "learning_rate": 3.5072727272727274e-06, | |
| "loss": 0.0002, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.754530477759473, | |
| "grad_norm": 0.013130983337759972, | |
| "learning_rate": 3.4618181818181824e-06, | |
| "loss": 0.0001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.795716639209226, | |
| "grad_norm": 0.022349372506141663, | |
| "learning_rate": 3.4163636363636366e-06, | |
| "loss": 0.0001, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.8369028006589785, | |
| "grad_norm": 0.004155050031840801, | |
| "learning_rate": 3.370909090909091e-06, | |
| "loss": 0.0001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.878088962108731, | |
| "grad_norm": 0.004423918202519417, | |
| "learning_rate": 3.3254545454545458e-06, | |
| "loss": 0.0001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.919275123558484, | |
| "grad_norm": 0.010945485904812813, | |
| "learning_rate": 3.2800000000000004e-06, | |
| "loss": 0.0001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.960461285008237, | |
| "grad_norm": 0.00635559344664216, | |
| "learning_rate": 3.2345454545454545e-06, | |
| "loss": 0.0001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 7.00164744645799, | |
| "grad_norm": 0.0034311157651245594, | |
| "learning_rate": 3.1890909090909096e-06, | |
| "loss": 0.0003, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 7.042833607907743, | |
| "grad_norm": 0.01130327396094799, | |
| "learning_rate": 3.1436363636363637e-06, | |
| "loss": 0.0, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 7.0840197693574956, | |
| "grad_norm": 0.004722132347524166, | |
| "learning_rate": 3.0981818181818183e-06, | |
| "loss": 0.0, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 7.125205930807248, | |
| "grad_norm": 0.004732927773147821, | |
| "learning_rate": 3.052727272727273e-06, | |
| "loss": 0.0002, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 7.166392092257001, | |
| "grad_norm": 0.0030134401749819517, | |
| "learning_rate": 3.0072727272727275e-06, | |
| "loss": 0.0, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 7.207578253706755, | |
| "grad_norm": 0.002724016085267067, | |
| "learning_rate": 2.9618181818181817e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 7.248764415156508, | |
| "grad_norm": 0.004742628429085016, | |
| "learning_rate": 2.9163636363636367e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.289950576606261, | |
| "grad_norm": 0.009233110584318638, | |
| "learning_rate": 2.870909090909091e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.3311367380560135, | |
| "grad_norm": 0.004101255908608437, | |
| "learning_rate": 2.8254545454545455e-06, | |
| "loss": 0.0, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.372322899505766, | |
| "grad_norm": 0.0037873326800763607, | |
| "learning_rate": 2.7800000000000005e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.413509060955519, | |
| "grad_norm": 0.015564778819680214, | |
| "learning_rate": 2.7345454545454547e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.413509060955519, | |
| "eval_loss": 0.013747855089604855, | |
| "eval_runtime": 1717.6068, | |
| "eval_samples_per_second": 1.413, | |
| "eval_steps_per_second": 0.177, | |
| "eval_wer": 2.5013594344752583, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.454695222405272, | |
| "grad_norm": 0.0023361605126410723, | |
| "learning_rate": 2.6890909090909097e-06, | |
| "loss": 0.0002, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.495881383855025, | |
| "grad_norm": 0.0026130119804292917, | |
| "learning_rate": 2.643636363636364e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.537067545304778, | |
| "grad_norm": 0.005539346020668745, | |
| "learning_rate": 2.5981818181818184e-06, | |
| "loss": 0.0001, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.578253706754531, | |
| "grad_norm": 0.00382342585362494, | |
| "learning_rate": 2.552727272727273e-06, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 7.619439868204283, | |
| "grad_norm": 0.0032592774368822575, | |
| "learning_rate": 2.5072727272727276e-06, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.660626029654036, | |
| "grad_norm": 0.003578683128580451, | |
| "learning_rate": 2.461818181818182e-06, | |
| "loss": 0.0001, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.701812191103789, | |
| "grad_norm": 0.006414057221263647, | |
| "learning_rate": 2.416363636363637e-06, | |
| "loss": 0.0, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.742998352553542, | |
| "grad_norm": 0.0032208007760345936, | |
| "learning_rate": 2.3709090909090914e-06, | |
| "loss": 0.0, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.784184514003295, | |
| "grad_norm": 0.0016529737040400505, | |
| "learning_rate": 2.3254545454545456e-06, | |
| "loss": 0.0, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.825370675453048, | |
| "grad_norm": 0.002902130363509059, | |
| "learning_rate": 2.28e-06, | |
| "loss": 0.0001, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.8665568369028005, | |
| "grad_norm": 0.002918258076533675, | |
| "learning_rate": 2.234545454545455e-06, | |
| "loss": 0.0, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.907742998352553, | |
| "grad_norm": 0.0021400710102170706, | |
| "learning_rate": 2.1890909090909094e-06, | |
| "loss": 0.0, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.948929159802306, | |
| "grad_norm": 0.0016904472140595317, | |
| "learning_rate": 2.143636363636364e-06, | |
| "loss": 0.0, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.990115321252059, | |
| "grad_norm": 0.006286232266575098, | |
| "learning_rate": 2.0981818181818186e-06, | |
| "loss": 0.0002, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 8.031301482701812, | |
| "grad_norm": 0.004598398692905903, | |
| "learning_rate": 2.0527272727272727e-06, | |
| "loss": 0.0, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 8.072487644151565, | |
| "grad_norm": 0.0023243261966854334, | |
| "learning_rate": 2.0072727272727273e-06, | |
| "loss": 0.0, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 8.113673805601318, | |
| "grad_norm": 0.0025131104048341513, | |
| "learning_rate": 1.961818181818182e-06, | |
| "loss": 0.0, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 8.15485996705107, | |
| "grad_norm": 0.005834953393787146, | |
| "learning_rate": 1.9163636363636365e-06, | |
| "loss": 0.0003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 8.196046128500823, | |
| "grad_norm": 0.005762421526014805, | |
| "learning_rate": 1.870909090909091e-06, | |
| "loss": 0.0001, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 8.237232289950576, | |
| "grad_norm": 0.002938208868727088, | |
| "learning_rate": 1.8254545454545455e-06, | |
| "loss": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.237232289950576, | |
| "eval_loss": 0.013800903223454952, | |
| "eval_runtime": 1718.2773, | |
| "eval_samples_per_second": 1.412, | |
| "eval_steps_per_second": 0.177, | |
| "eval_wer": 2.602345995494446, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.278418451400329, | |
| "grad_norm": 0.0034450343810021877, | |
| "learning_rate": 1.7800000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 8.319604612850082, | |
| "grad_norm": 0.002541518537327647, | |
| "learning_rate": 1.7345454545454545e-06, | |
| "loss": 0.0, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 8.360790774299835, | |
| "grad_norm": 0.002959892386570573, | |
| "learning_rate": 1.689090909090909e-06, | |
| "loss": 0.0, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 8.401976935749587, | |
| "grad_norm": 0.004116767086088657, | |
| "learning_rate": 1.6436363636363637e-06, | |
| "loss": 0.0001, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 8.44316309719934, | |
| "grad_norm": 0.01786639727652073, | |
| "learning_rate": 1.5981818181818185e-06, | |
| "loss": 0.0, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 8.484349258649093, | |
| "grad_norm": 0.008570323698222637, | |
| "learning_rate": 1.5527272727272729e-06, | |
| "loss": 0.0, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 8.525535420098846, | |
| "grad_norm": 0.0028420346789062023, | |
| "learning_rate": 1.5072727272727275e-06, | |
| "loss": 0.0001, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 8.566721581548599, | |
| "grad_norm": 0.002692845184355974, | |
| "learning_rate": 1.461818181818182e-06, | |
| "loss": 0.0001, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 8.607907742998353, | |
| "grad_norm": 0.0014716355362907052, | |
| "learning_rate": 1.4163636363636365e-06, | |
| "loss": 0.0, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 8.649093904448105, | |
| "grad_norm": 0.004192767199128866, | |
| "learning_rate": 1.370909090909091e-06, | |
| "loss": 0.0, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 8.69028006589786, | |
| "grad_norm": 0.0037771447096019983, | |
| "learning_rate": 1.3254545454545456e-06, | |
| "loss": 0.0, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 8.731466227347612, | |
| "grad_norm": 0.0033924980089068413, | |
| "learning_rate": 1.28e-06, | |
| "loss": 0.0, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 8.772652388797365, | |
| "grad_norm": 0.007148527540266514, | |
| "learning_rate": 1.2345454545454546e-06, | |
| "loss": 0.0, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 8.813838550247118, | |
| "grad_norm": 0.0036525556351989508, | |
| "learning_rate": 1.1890909090909092e-06, | |
| "loss": 0.0001, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 8.85502471169687, | |
| "grad_norm": 0.004420670215040445, | |
| "learning_rate": 1.1436363636363638e-06, | |
| "loss": 0.0, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 8.896210873146623, | |
| "grad_norm": 0.007836336269974709, | |
| "learning_rate": 1.0981818181818182e-06, | |
| "loss": 0.0, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 8.937397034596376, | |
| "grad_norm": 0.0029930309392511845, | |
| "learning_rate": 1.0527272727272728e-06, | |
| "loss": 0.0, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 8.978583196046129, | |
| "grad_norm": 0.004859840031713247, | |
| "learning_rate": 1.0072727272727274e-06, | |
| "loss": 0.0001, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 9.019769357495882, | |
| "grad_norm": 0.0022968496195971966, | |
| "learning_rate": 9.61818181818182e-07, | |
| "loss": 0.0, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 9.060955518945635, | |
| "grad_norm": 0.0016871189000084996, | |
| "learning_rate": 9.163636363636365e-07, | |
| "loss": 0.0001, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.060955518945635, | |
| "eval_loss": 0.013956129550933838, | |
| "eval_runtime": 1736.8144, | |
| "eval_samples_per_second": 1.397, | |
| "eval_steps_per_second": 0.175, | |
| "eval_wer": 2.5091276314767343, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.102141680395388, | |
| "grad_norm": 0.003995977807790041, | |
| "learning_rate": 8.70909090909091e-07, | |
| "loss": 0.0001, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 9.14332784184514, | |
| "grad_norm": 0.0030109367799013853, | |
| "learning_rate": 8.254545454545456e-07, | |
| "loss": 0.0, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 9.184514003294893, | |
| "grad_norm": 0.004546226933598518, | |
| "learning_rate": 7.8e-07, | |
| "loss": 0.0, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 9.225700164744646, | |
| "grad_norm": 0.003448527306318283, | |
| "learning_rate": 7.345454545454545e-07, | |
| "loss": 0.0, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 9.266886326194399, | |
| "grad_norm": 0.003833542810752988, | |
| "learning_rate": 6.890909090909091e-07, | |
| "loss": 0.0, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 9.308072487644152, | |
| "grad_norm": 0.0035726006608456373, | |
| "learning_rate": 6.436363636363636e-07, | |
| "loss": 0.0001, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 9.349258649093905, | |
| "grad_norm": 0.0038840766064822674, | |
| "learning_rate": 5.981818181818182e-07, | |
| "loss": 0.0, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 9.390444810543658, | |
| "grad_norm": 0.001967709045857191, | |
| "learning_rate": 5.527272727272728e-07, | |
| "loss": 0.0, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 9.43163097199341, | |
| "grad_norm": 0.0022351674269884825, | |
| "learning_rate": 5.072727272727273e-07, | |
| "loss": 0.0, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 9.472817133443163, | |
| "grad_norm": 0.004692086949944496, | |
| "learning_rate": 4.618181818181819e-07, | |
| "loss": 0.0, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 9.514003294892916, | |
| "grad_norm": 0.0022310977801680565, | |
| "learning_rate": 4.163636363636364e-07, | |
| "loss": 0.0, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 9.555189456342669, | |
| "grad_norm": 0.0022611399181187153, | |
| "learning_rate": 3.7090909090909093e-07, | |
| "loss": 0.0, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 9.596375617792422, | |
| "grad_norm": 0.0022751970682293177, | |
| "learning_rate": 3.254545454545455e-07, | |
| "loss": 0.0, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 9.637561779242175, | |
| "grad_norm": 0.003461811225861311, | |
| "learning_rate": 2.8e-07, | |
| "loss": 0.0, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 9.678747940691927, | |
| "grad_norm": 0.002553915837779641, | |
| "learning_rate": 2.3454545454545456e-07, | |
| "loss": 0.0, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 9.71993410214168, | |
| "grad_norm": 0.002379871206358075, | |
| "learning_rate": 1.890909090909091e-07, | |
| "loss": 0.0, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 9.761120263591433, | |
| "grad_norm": 0.0019515318563207984, | |
| "learning_rate": 1.4363636363636365e-07, | |
| "loss": 0.0, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 9.802306425041186, | |
| "grad_norm": 0.0017870984738692641, | |
| "learning_rate": 9.818181818181819e-08, | |
| "loss": 0.0, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 9.843492586490939, | |
| "grad_norm": 0.0032659450080245733, | |
| "learning_rate": 5.272727272727273e-08, | |
| "loss": 0.0, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 9.884678747940692, | |
| "grad_norm": 0.01834402047097683, | |
| "learning_rate": 7.272727272727273e-09, | |
| "loss": 0.0, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 9.884678747940692, | |
| "eval_loss": 0.013986977748572826, | |
| "eval_runtime": 1723.1255, | |
| "eval_samples_per_second": 1.408, | |
| "eval_steps_per_second": 0.176, | |
| "eval_wer": 2.5013594344752583, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 9.925864909390445, | |
| "grad_norm": 0.0242526326328516, | |
| "learning_rate": 3.5047058823529417e-06, | |
| "loss": 0.0, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 9.967051070840197, | |
| "grad_norm": 0.002037163358181715, | |
| "learning_rate": 3.4752941176470588e-06, | |
| "loss": 0.0, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 10.00823723228995, | |
| "grad_norm": 0.0032065610866993666, | |
| "learning_rate": 3.4458823529411767e-06, | |
| "loss": 0.0, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 10.049423393739703, | |
| "grad_norm": 0.0026907401625066996, | |
| "learning_rate": 3.4164705882352946e-06, | |
| "loss": 0.0, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 10.090609555189456, | |
| "grad_norm": 0.0022176543716341257, | |
| "learning_rate": 3.387058823529412e-06, | |
| "loss": 0.0, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 10.131795716639209, | |
| "grad_norm": 0.0032068961299955845, | |
| "learning_rate": 3.35764705882353e-06, | |
| "loss": 0.0, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 10.172981878088962, | |
| "grad_norm": 0.0039983587339520454, | |
| "learning_rate": 3.328235294117647e-06, | |
| "loss": 0.0, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 10.214168039538714, | |
| "grad_norm": 0.0018802180420607328, | |
| "learning_rate": 3.298823529411765e-06, | |
| "loss": 0.0, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 10.255354200988467, | |
| "grad_norm": 0.0026977595407515764, | |
| "learning_rate": 3.2694117647058824e-06, | |
| "loss": 0.0, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 10.29654036243822, | |
| "grad_norm": 0.0030683802906423807, | |
| "learning_rate": 3.2400000000000003e-06, | |
| "loss": 0.0, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 10.337726523887973, | |
| "grad_norm": 0.002406976418569684, | |
| "learning_rate": 3.2105882352941182e-06, | |
| "loss": 0.0, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 10.378912685337726, | |
| "grad_norm": 0.004232148639857769, | |
| "learning_rate": 3.1811764705882353e-06, | |
| "loss": 0.0, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 10.420098846787479, | |
| "grad_norm": 0.0028111624997109175, | |
| "learning_rate": 3.151764705882353e-06, | |
| "loss": 0.0, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 10.461285008237232, | |
| "grad_norm": 0.0027185629587620497, | |
| "learning_rate": 3.1223529411764707e-06, | |
| "loss": 0.0, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 10.502471169686984, | |
| "grad_norm": 0.002079539466649294, | |
| "learning_rate": 3.0929411764705886e-06, | |
| "loss": 0.0, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 10.543657331136739, | |
| "grad_norm": 0.0021090833470225334, | |
| "learning_rate": 3.0635294117647065e-06, | |
| "loss": 0.0, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 10.58484349258649, | |
| "grad_norm": 0.0029999890830367804, | |
| "learning_rate": 3.0341176470588235e-06, | |
| "loss": 0.0, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 10.626029654036245, | |
| "grad_norm": 0.0023287280928343534, | |
| "learning_rate": 3.0047058823529414e-06, | |
| "loss": 0.0, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 10.667215815485998, | |
| "grad_norm": 0.0014742282219231129, | |
| "learning_rate": 2.975294117647059e-06, | |
| "loss": 0.0, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 10.70840197693575, | |
| "grad_norm": 0.002694872207939625, | |
| "learning_rate": 2.945882352941177e-06, | |
| "loss": 0.0001, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 10.70840197693575, | |
| "eval_loss": 0.014553220942616463, | |
| "eval_runtime": 1723.0905, | |
| "eval_samples_per_second": 1.409, | |
| "eval_steps_per_second": 0.176, | |
| "eval_wer": 2.5945777984929697, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 10.749588138385503, | |
| "grad_norm": 0.001027889782562852, | |
| "learning_rate": 2.9164705882352943e-06, | |
| "loss": 0.0, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 10.790774299835256, | |
| "grad_norm": 0.0017091067275032401, | |
| "learning_rate": 2.887058823529412e-06, | |
| "loss": 0.0, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 10.831960461285009, | |
| "grad_norm": 0.001180932391434908, | |
| "learning_rate": 2.8576470588235293e-06, | |
| "loss": 0.0, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 10.873146622734762, | |
| "grad_norm": 0.0035964970011264086, | |
| "learning_rate": 2.828235294117647e-06, | |
| "loss": 0.0005, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 10.914332784184515, | |
| "grad_norm": 0.0023179524578154087, | |
| "learning_rate": 2.798823529411765e-06, | |
| "loss": 0.0, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 10.955518945634267, | |
| "grad_norm": 0.10292612761259079, | |
| "learning_rate": 2.7694117647058825e-06, | |
| "loss": 0.0001, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 10.99670510708402, | |
| "grad_norm": 1.5822137594223022, | |
| "learning_rate": 2.7400000000000004e-06, | |
| "loss": 0.0007, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 11.037891268533773, | |
| "grad_norm": 0.09981144219636917, | |
| "learning_rate": 2.7105882352941175e-06, | |
| "loss": 0.0028, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 11.079077429983526, | |
| "grad_norm": 1.2205625772476196, | |
| "learning_rate": 2.6811764705882354e-06, | |
| "loss": 0.0018, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 11.120263591433279, | |
| "grad_norm": 2.4946441650390625, | |
| "learning_rate": 2.6517647058823533e-06, | |
| "loss": 0.0037, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 11.161449752883032, | |
| "grad_norm": 0.01300705224275589, | |
| "learning_rate": 2.622352941176471e-06, | |
| "loss": 0.0007, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 11.202635914332784, | |
| "grad_norm": 0.050090231001377106, | |
| "learning_rate": 2.5929411764705887e-06, | |
| "loss": 0.0004, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 11.243822075782537, | |
| "grad_norm": 0.1064329594373703, | |
| "learning_rate": 2.5635294117647058e-06, | |
| "loss": 0.0003, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 11.28500823723229, | |
| "grad_norm": 0.00640516122803092, | |
| "learning_rate": 2.5341176470588237e-06, | |
| "loss": 0.0002, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 11.326194398682043, | |
| "grad_norm": 0.011775782331824303, | |
| "learning_rate": 2.504705882352941e-06, | |
| "loss": 0.0001, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 11.367380560131796, | |
| "grad_norm": 0.00485497061163187, | |
| "learning_rate": 2.475294117647059e-06, | |
| "loss": 0.0011, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 11.408566721581549, | |
| "grad_norm": 1.4145103693008423, | |
| "learning_rate": 2.4458823529411765e-06, | |
| "loss": 0.0005, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 11.449752883031302, | |
| "grad_norm": 0.01526624895632267, | |
| "learning_rate": 2.4164705882352944e-06, | |
| "loss": 0.0002, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 11.490939044481054, | |
| "grad_norm": 0.02947794646024704, | |
| "learning_rate": 2.387058823529412e-06, | |
| "loss": 0.0001, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 11.532125205930807, | |
| "grad_norm": 0.008897418156266212, | |
| "learning_rate": 2.3576470588235294e-06, | |
| "loss": 0.0002, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 11.532125205930807, | |
| "eval_loss": 0.016228729858994484, | |
| "eval_runtime": 1688.3118, | |
| "eval_samples_per_second": 1.438, | |
| "eval_steps_per_second": 0.18, | |
| "eval_wer": 3.029596830575623, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 11.57331136738056, | |
| "grad_norm": 0.1610272079706192, | |
| "learning_rate": 2.3282352941176473e-06, | |
| "loss": 0.0004, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 11.614497528830313, | |
| "grad_norm": 1.4256123304367065, | |
| "learning_rate": 2.298823529411765e-06, | |
| "loss": 0.0002, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 11.655683690280066, | |
| "grad_norm": 0.004848022013902664, | |
| "learning_rate": 2.2694117647058827e-06, | |
| "loss": 0.0003, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 11.696869851729819, | |
| "grad_norm": 0.007711484562605619, | |
| "learning_rate": 2.24e-06, | |
| "loss": 0.0006, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 11.738056013179571, | |
| "grad_norm": 0.11259625852108002, | |
| "learning_rate": 2.2105882352941176e-06, | |
| "loss": 0.0002, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 11.779242174629324, | |
| "grad_norm": 0.004204327240586281, | |
| "learning_rate": 2.1811764705882355e-06, | |
| "loss": 0.0001, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 11.820428336079077, | |
| "grad_norm": 0.007289988454431295, | |
| "learning_rate": 2.151764705882353e-06, | |
| "loss": 0.0001, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 11.86161449752883, | |
| "grad_norm": 0.02570285275578499, | |
| "learning_rate": 2.1223529411764705e-06, | |
| "loss": 0.0003, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 11.902800658978583, | |
| "grad_norm": 0.009201893582940102, | |
| "learning_rate": 2.0929411764705884e-06, | |
| "loss": 0.0002, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 11.943986820428336, | |
| "grad_norm": 0.003304847748950124, | |
| "learning_rate": 2.0635294117647063e-06, | |
| "loss": 0.0002, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 11.985172981878089, | |
| "grad_norm": 0.07221189141273499, | |
| "learning_rate": 2.0341176470588238e-06, | |
| "loss": 0.0001, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 12.026359143327841, | |
| "grad_norm": 0.01542985625565052, | |
| "learning_rate": 2.0058823529411765e-06, | |
| "loss": 0.0002, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 12.067545304777594, | |
| "grad_norm": 0.009480349719524384, | |
| "learning_rate": 1.976470588235294e-06, | |
| "loss": 0.0, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 12.108731466227347, | |
| "grad_norm": 0.0024922864977270365, | |
| "learning_rate": 1.947058823529412e-06, | |
| "loss": 0.0002, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 12.1499176276771, | |
| "grad_norm": 0.0071571036241948605, | |
| "learning_rate": 1.9176470588235298e-06, | |
| "loss": 0.0001, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 12.191103789126853, | |
| "grad_norm": 0.008463573642075062, | |
| "learning_rate": 1.888235294117647e-06, | |
| "loss": 0.0001, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 12.232289950576606, | |
| "grad_norm": 0.03403167426586151, | |
| "learning_rate": 1.858823529411765e-06, | |
| "loss": 0.0001, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 12.273476112026358, | |
| "grad_norm": 0.0034997081384062767, | |
| "learning_rate": 1.8294117647058827e-06, | |
| "loss": 0.0, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 12.314662273476111, | |
| "grad_norm": 0.0037921613547950983, | |
| "learning_rate": 1.8011764705882354e-06, | |
| "loss": 0.0004, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 12.355848434925864, | |
| "grad_norm": 0.0021253039594739676, | |
| "learning_rate": 1.771764705882353e-06, | |
| "loss": 0.0001, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 12.355848434925864, | |
| "eval_loss": 0.012992938980460167, | |
| "eval_runtime": 1716.627, | |
| "eval_samples_per_second": 1.414, | |
| "eval_steps_per_second": 0.177, | |
| "eval_wer": 2.3382272974442633, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 12.397034596375617, | |
| "grad_norm": 0.023194992914795876, | |
| "learning_rate": 1.742352941176471e-06, | |
| "loss": 0.0, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 12.43822075782537, | |
| "grad_norm": 0.15046222507953644, | |
| "learning_rate": 1.7129411764705882e-06, | |
| "loss": 0.0001, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 12.479406919275124, | |
| "grad_norm": 0.003174582729116082, | |
| "learning_rate": 1.6835294117647061e-06, | |
| "loss": 0.0001, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 12.520593080724876, | |
| "grad_norm": 0.0067563788034021854, | |
| "learning_rate": 1.6541176470588236e-06, | |
| "loss": 0.0, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 12.56177924217463, | |
| "grad_norm": 0.002005406189709902, | |
| "learning_rate": 1.6247058823529413e-06, | |
| "loss": 0.0, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 12.602965403624383, | |
| "grad_norm": 0.002191119594499469, | |
| "learning_rate": 1.595294117647059e-06, | |
| "loss": 0.0001, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 12.644151565074136, | |
| "grad_norm": 0.0046321190893650055, | |
| "learning_rate": 1.5658823529411765e-06, | |
| "loss": 0.0001, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 12.685337726523889, | |
| "grad_norm": 0.006401429884135723, | |
| "learning_rate": 1.5364705882352942e-06, | |
| "loss": 0.0001, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 12.726523887973642, | |
| "grad_norm": 0.0027000552508980036, | |
| "learning_rate": 1.507058823529412e-06, | |
| "loss": 0.0002, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 12.767710049423394, | |
| "grad_norm": 0.19640517234802246, | |
| "learning_rate": 1.4776470588235296e-06, | |
| "loss": 0.0001, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 12.808896210873147, | |
| "grad_norm": 0.0019183914409950376, | |
| "learning_rate": 1.4482352941176472e-06, | |
| "loss": 0.0001, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 12.8500823723229, | |
| "grad_norm": 0.012867034412920475, | |
| "learning_rate": 1.4188235294117647e-06, | |
| "loss": 0.0001, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 12.891268533772653, | |
| "grad_norm": 0.0031524612568318844, | |
| "learning_rate": 1.3894117647058824e-06, | |
| "loss": 0.0001, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 12.932454695222406, | |
| "grad_norm": 0.004168114624917507, | |
| "learning_rate": 1.3600000000000001e-06, | |
| "loss": 0.0, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 12.973640856672159, | |
| "grad_norm": 0.007319289725273848, | |
| "learning_rate": 1.3305882352941176e-06, | |
| "loss": 0.0001, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 13.014827018121911, | |
| "grad_norm": 0.007646168116480112, | |
| "learning_rate": 1.3011764705882355e-06, | |
| "loss": 0.0, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 13.056013179571664, | |
| "grad_norm": 0.0034050345420837402, | |
| "learning_rate": 1.2717647058823532e-06, | |
| "loss": 0.0, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 13.097199341021417, | |
| "grad_norm": 0.0018609551480039954, | |
| "learning_rate": 1.2423529411764707e-06, | |
| "loss": 0.0, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 13.13838550247117, | |
| "grad_norm": 0.005138488952070475, | |
| "learning_rate": 1.2129411764705884e-06, | |
| "loss": 0.0, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 13.179571663920923, | |
| "grad_norm": 0.003146812552586198, | |
| "learning_rate": 1.183529411764706e-06, | |
| "loss": 0.0, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 13.179571663920923, | |
| "eval_loss": 0.013023543171584606, | |
| "eval_runtime": 1630.2411, | |
| "eval_samples_per_second": 1.489, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 2.4780548434708303, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 13.220757825370676, | |
| "grad_norm": 0.0029420964419841766, | |
| "learning_rate": 1.1541176470588235e-06, | |
| "loss": 0.0, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 13.261943986820429, | |
| "grad_norm": 0.006158372852951288, | |
| "learning_rate": 1.1247058823529412e-06, | |
| "loss": 0.0001, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 13.303130148270181, | |
| "grad_norm": 0.00477724289521575, | |
| "learning_rate": 1.095294117647059e-06, | |
| "loss": 0.0, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 13.344316309719934, | |
| "grad_norm": 0.012528087943792343, | |
| "learning_rate": 1.0658823529411766e-06, | |
| "loss": 0.0, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 13.385502471169687, | |
| "grad_norm": 0.0019438378512859344, | |
| "learning_rate": 1.0364705882352943e-06, | |
| "loss": 0.0001, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 13.42668863261944, | |
| "grad_norm": 0.003603472374379635, | |
| "learning_rate": 1.0070588235294118e-06, | |
| "loss": 0.0, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 13.467874794069193, | |
| "grad_norm": 0.00542394258081913, | |
| "learning_rate": 9.776470588235295e-07, | |
| "loss": 0.0, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 13.509060955518946, | |
| "grad_norm": 0.001990295248106122, | |
| "learning_rate": 9.482352941176472e-07, | |
| "loss": 0.0, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 13.550247116968698, | |
| "grad_norm": 0.009004420600831509, | |
| "learning_rate": 9.188235294117647e-07, | |
| "loss": 0.0, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 13.591433278418451, | |
| "grad_norm": 0.0023559145629405975, | |
| "learning_rate": 8.894117647058824e-07, | |
| "loss": 0.0, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 13.632619439868204, | |
| "grad_norm": 0.006258898414671421, | |
| "learning_rate": 8.6e-07, | |
| "loss": 0.0, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 13.673805601317957, | |
| "grad_norm": 0.0016879294998943806, | |
| "learning_rate": 8.305882352941177e-07, | |
| "loss": 0.0, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 13.71499176276771, | |
| "grad_norm": 0.006166738457977772, | |
| "learning_rate": 8.011764705882354e-07, | |
| "loss": 0.0, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 13.756177924217463, | |
| "grad_norm": 0.002089226385578513, | |
| "learning_rate": 7.71764705882353e-07, | |
| "loss": 0.0, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 13.797364085667216, | |
| "grad_norm": 0.006312618497759104, | |
| "learning_rate": 7.423529411764706e-07, | |
| "loss": 0.0, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 13.838550247116968, | |
| "grad_norm": 0.002466730307787657, | |
| "learning_rate": 7.129411764705884e-07, | |
| "loss": 0.0, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 13.879736408566721, | |
| "grad_norm": 0.0038565557915717363, | |
| "learning_rate": 6.83529411764706e-07, | |
| "loss": 0.0, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 13.920922570016474, | |
| "grad_norm": 0.0016545254038646817, | |
| "learning_rate": 6.541176470588236e-07, | |
| "loss": 0.0, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 13.962108731466227, | |
| "grad_norm": 0.005063951946794987, | |
| "learning_rate": 6.247058823529412e-07, | |
| "loss": 0.0, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 14.00329489291598, | |
| "grad_norm": 0.006067285314202309, | |
| "learning_rate": 5.952941176470588e-07, | |
| "loss": 0.0, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 14.00329489291598, | |
| "eval_loss": 0.013567977584898472, | |
| "eval_runtime": 1612.6489, | |
| "eval_samples_per_second": 1.505, | |
| "eval_steps_per_second": 0.189, | |
| "eval_wer": 2.4469820554649266, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 14.044481054365733, | |
| "grad_norm": 0.0043179988861083984, | |
| "learning_rate": 5.658823529411765e-07, | |
| "loss": 0.0, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 14.085667215815485, | |
| "grad_norm": 0.0059325359761714935, | |
| "learning_rate": 5.364705882352941e-07, | |
| "loss": 0.0, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 14.126853377265238, | |
| "grad_norm": 0.002210423117503524, | |
| "learning_rate": 5.070588235294118e-07, | |
| "loss": 0.0, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 14.168039538714991, | |
| "grad_norm": 0.0023594272788614035, | |
| "learning_rate": 4.776470588235294e-07, | |
| "loss": 0.0, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 14.209225700164744, | |
| "grad_norm": 0.001723596709780395, | |
| "learning_rate": 4.482352941176471e-07, | |
| "loss": 0.0, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 14.250411861614497, | |
| "grad_norm": 0.0022638991940766573, | |
| "learning_rate": 4.188235294117647e-07, | |
| "loss": 0.0, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 14.29159802306425, | |
| "grad_norm": 0.0020570564083755016, | |
| "learning_rate": 3.894117647058824e-07, | |
| "loss": 0.0, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 14.332784184514002, | |
| "grad_norm": 0.017068659886717796, | |
| "learning_rate": 3.6e-07, | |
| "loss": 0.0, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 14.373970345963755, | |
| "grad_norm": 0.006104827858507633, | |
| "learning_rate": 3.305882352941177e-07, | |
| "loss": 0.0, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 14.41515650741351, | |
| "grad_norm": 0.004056777339428663, | |
| "learning_rate": 3.011764705882353e-07, | |
| "loss": 0.0, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 14.456342668863263, | |
| "grad_norm": 0.002557105151936412, | |
| "learning_rate": 2.7176470588235296e-07, | |
| "loss": 0.0, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 14.497528830313016, | |
| "grad_norm": 0.0033158042933791876, | |
| "learning_rate": 2.423529411764706e-07, | |
| "loss": 0.0, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 14.538714991762768, | |
| "grad_norm": 0.0037535373121500015, | |
| "learning_rate": 2.1294117647058827e-07, | |
| "loss": 0.0, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 14.579901153212521, | |
| "grad_norm": 0.0012597900349646807, | |
| "learning_rate": 1.835294117647059e-07, | |
| "loss": 0.0, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 14.621087314662274, | |
| "grad_norm": 0.006313154939562082, | |
| "learning_rate": 1.5411764705882355e-07, | |
| "loss": 0.0, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 14.662273476112027, | |
| "grad_norm": 0.0067338962107896805, | |
| "learning_rate": 1.2470588235294119e-07, | |
| "loss": 0.0, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 14.70345963756178, | |
| "grad_norm": 0.001957102445885539, | |
| "learning_rate": 9.529411764705883e-08, | |
| "loss": 0.0, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 14.744645799011533, | |
| "grad_norm": 0.0035015118774026632, | |
| "learning_rate": 6.588235294117648e-08, | |
| "loss": 0.0, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 14.785831960461286, | |
| "grad_norm": 0.0019214386120438576, | |
| "learning_rate": 3.6470588235294116e-08, | |
| "loss": 0.0, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 14.827018121911038, | |
| "grad_norm": 0.005048139486461878, | |
| "learning_rate": 7.0588235294117654e-09, | |
| "loss": 0.0, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 14.827018121911038, | |
| "eval_loss": 0.013581627048552036, | |
| "eval_runtime": 1622.8495, | |
| "eval_samples_per_second": 1.496, | |
| "eval_steps_per_second": 0.187, | |
| "eval_wer": 2.4858230404723063, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 14.868204283360791, | |
| "grad_norm": 0.002532125450670719, | |
| "learning_rate": 2.5921739130434787e-06, | |
| "loss": 0.0, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 14.909390444810544, | |
| "grad_norm": 0.0009551341645419598, | |
| "learning_rate": 2.5704347826086954e-06, | |
| "loss": 0.0, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 14.950576606260297, | |
| "grad_norm": 0.0017027213471010327, | |
| "learning_rate": 2.548695652173913e-06, | |
| "loss": 0.0, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 14.99176276771005, | |
| "grad_norm": 0.0021833304781466722, | |
| "learning_rate": 2.5269565217391305e-06, | |
| "loss": 0.0, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 15.032948929159803, | |
| "grad_norm": 0.0017440480878576636, | |
| "learning_rate": 2.505217391304348e-06, | |
| "loss": 0.0, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 15.074135090609555, | |
| "grad_norm": 0.005614432040601969, | |
| "learning_rate": 2.4834782608695652e-06, | |
| "loss": 0.0, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 15.115321252059308, | |
| "grad_norm": 0.0023146765306591988, | |
| "learning_rate": 2.461739130434783e-06, | |
| "loss": 0.0, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 15.156507413509061, | |
| "grad_norm": 0.00141927320510149, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.0, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 15.197693574958814, | |
| "grad_norm": 0.0027971493545919657, | |
| "learning_rate": 2.4182608695652175e-06, | |
| "loss": 0.0, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 15.238879736408567, | |
| "grad_norm": 0.0009957117727026343, | |
| "learning_rate": 2.396521739130435e-06, | |
| "loss": 0.0, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 15.28006589785832, | |
| "grad_norm": 0.0014513045316562057, | |
| "learning_rate": 2.3747826086956526e-06, | |
| "loss": 0.0, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 15.321252059308073, | |
| "grad_norm": 0.0019499498885124922, | |
| "learning_rate": 2.3530434782608698e-06, | |
| "loss": 0.0, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 15.362438220757825, | |
| "grad_norm": 0.0013381906319409609, | |
| "learning_rate": 2.3313043478260873e-06, | |
| "loss": 0.0, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 15.403624382207578, | |
| "grad_norm": 0.0021802387200295925, | |
| "learning_rate": 2.3095652173913045e-06, | |
| "loss": 0.0, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 15.444810543657331, | |
| "grad_norm": 0.001607355079613626, | |
| "learning_rate": 2.287826086956522e-06, | |
| "loss": 0.0, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 15.485996705107084, | |
| "grad_norm": 0.0017373546725139022, | |
| "learning_rate": 2.2660869565217396e-06, | |
| "loss": 0.0, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 15.527182866556837, | |
| "grad_norm": 0.0009414468659088016, | |
| "learning_rate": 2.2443478260869567e-06, | |
| "loss": 0.0, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 15.56836902800659, | |
| "grad_norm": 0.0011722907656803727, | |
| "learning_rate": 2.2226086956521743e-06, | |
| "loss": 0.0, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 15.609555189456342, | |
| "grad_norm": 0.0012756260111927986, | |
| "learning_rate": 2.2008695652173914e-06, | |
| "loss": 0.0, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 15.650741350906095, | |
| "grad_norm": 0.003932903986424208, | |
| "learning_rate": 2.179130434782609e-06, | |
| "loss": 0.0, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 15.650741350906095, | |
| "eval_loss": 0.013979827053844929, | |
| "eval_runtime": 1641.4283, | |
| "eval_samples_per_second": 1.479, | |
| "eval_steps_per_second": 0.185, | |
| "eval_wer": 2.532432222481162, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 15.691927512355848, | |
| "grad_norm": 0.0014398075873032212, | |
| "learning_rate": 2.157391304347826e-06, | |
| "loss": 0.0, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 15.733113673805601, | |
| "grad_norm": 0.0019754846580326557, | |
| "learning_rate": 2.1356521739130437e-06, | |
| "loss": 0.0, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 15.774299835255354, | |
| "grad_norm": 0.001919199014082551, | |
| "learning_rate": 2.113913043478261e-06, | |
| "loss": 0.0, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 15.815485996705107, | |
| "grad_norm": 0.005105574149638414, | |
| "learning_rate": 2.0921739130434784e-06, | |
| "loss": 0.0, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 15.85667215815486, | |
| "grad_norm": 0.0087418919429183, | |
| "learning_rate": 2.070434782608696e-06, | |
| "loss": 0.0, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 15.897858319604612, | |
| "grad_norm": 0.0017288941890001297, | |
| "learning_rate": 2.048695652173913e-06, | |
| "loss": 0.0, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 15.939044481054365, | |
| "grad_norm": 0.0016419363673776388, | |
| "learning_rate": 2.0269565217391307e-06, | |
| "loss": 0.0, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 15.980230642504118, | |
| "grad_norm": 0.0014526849845424294, | |
| "learning_rate": 2.005217391304348e-06, | |
| "loss": 0.0, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 16.021416803953873, | |
| "grad_norm": 0.0014665353810414672, | |
| "learning_rate": 1.9834782608695654e-06, | |
| "loss": 0.0, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 16.062602965403624, | |
| "grad_norm": 0.003292278153821826, | |
| "learning_rate": 1.9617391304347825e-06, | |
| "loss": 0.0, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 16.10378912685338, | |
| "grad_norm": 0.0014827033737674356, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.0, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 16.14497528830313, | |
| "grad_norm": 0.001969423145055771, | |
| "learning_rate": 1.9182608695652176e-06, | |
| "loss": 0.0, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 16.186161449752884, | |
| "grad_norm": 0.0010216091759502888, | |
| "learning_rate": 1.896521739130435e-06, | |
| "loss": 0.0, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 16.227347611202635, | |
| "grad_norm": 0.002770212711766362, | |
| "learning_rate": 1.8747826086956521e-06, | |
| "loss": 0.0, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 16.26853377265239, | |
| "grad_norm": 0.002100354991853237, | |
| "learning_rate": 1.8530434782608697e-06, | |
| "loss": 0.0, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 16.30971993410214, | |
| "grad_norm": 0.001876773894764483, | |
| "learning_rate": 1.8313043478260872e-06, | |
| "loss": 0.0, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 16.350906095551895, | |
| "grad_norm": 0.001118077663704753, | |
| "learning_rate": 1.8095652173913044e-06, | |
| "loss": 0.0, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 16.392092257001647, | |
| "grad_norm": 0.0008371293661184609, | |
| "learning_rate": 1.787826086956522e-06, | |
| "loss": 0.0, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 16.4332784184514, | |
| "grad_norm": 0.0007290031644515693, | |
| "learning_rate": 1.7660869565217393e-06, | |
| "loss": 0.0, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 16.474464579901152, | |
| "grad_norm": 0.0012829427141696215, | |
| "learning_rate": 1.7443478260869566e-06, | |
| "loss": 0.0, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 16.474464579901152, | |
| "eval_loss": 0.014236521907150745, | |
| "eval_runtime": 1633.2029, | |
| "eval_samples_per_second": 1.486, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 2.5013594344752583, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 16.515650741350907, | |
| "grad_norm": 0.0031344795133918524, | |
| "learning_rate": 1.722608695652174e-06, | |
| "loss": 0.0, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 16.556836902800658, | |
| "grad_norm": 0.0019432102562859654, | |
| "learning_rate": 1.7008695652173916e-06, | |
| "loss": 0.0, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 16.598023064250413, | |
| "grad_norm": 0.001211246708407998, | |
| "learning_rate": 1.6791304347826087e-06, | |
| "loss": 0.0, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 16.639209225700164, | |
| "grad_norm": 0.001109111588448286, | |
| "learning_rate": 1.6573913043478263e-06, | |
| "loss": 0.0, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 16.680395387149918, | |
| "grad_norm": 0.0025855598505586386, | |
| "learning_rate": 1.6356521739130434e-06, | |
| "loss": 0.0, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 16.72158154859967, | |
| "grad_norm": 0.0008468018495477736, | |
| "learning_rate": 1.613913043478261e-06, | |
| "loss": 0.0, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 16.762767710049424, | |
| "grad_norm": 0.0015540047315880656, | |
| "learning_rate": 1.5921739130434785e-06, | |
| "loss": 0.0, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 16.803953871499175, | |
| "grad_norm": 0.0022469875402748585, | |
| "learning_rate": 1.5704347826086957e-06, | |
| "loss": 0.0, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 16.84514003294893, | |
| "grad_norm": 0.00045295292511582375, | |
| "learning_rate": 1.5486956521739132e-06, | |
| "loss": 0.0, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 16.88632619439868, | |
| "grad_norm": 0.0018261218210682273, | |
| "learning_rate": 1.5269565217391306e-06, | |
| "loss": 0.0, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 16.927512355848435, | |
| "grad_norm": 0.0018533602124080062, | |
| "learning_rate": 1.505217391304348e-06, | |
| "loss": 0.0, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 16.968698517298186, | |
| "grad_norm": 0.0009767924202606082, | |
| "learning_rate": 1.4834782608695653e-06, | |
| "loss": 0.0, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 17.00988467874794, | |
| "grad_norm": 0.0009673904860392213, | |
| "learning_rate": 1.4617391304347828e-06, | |
| "loss": 0.0, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 17.051070840197692, | |
| "grad_norm": 0.002202032832428813, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.0, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 17.092257001647447, | |
| "grad_norm": 0.001030725659802556, | |
| "learning_rate": 1.4182608695652175e-06, | |
| "loss": 0.0, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 17.133443163097198, | |
| "grad_norm": 0.0008200545562431216, | |
| "learning_rate": 1.3965217391304347e-06, | |
| "loss": 0.0, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 17.174629324546952, | |
| "grad_norm": 0.0010224173311144114, | |
| "learning_rate": 1.3747826086956522e-06, | |
| "loss": 0.0, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 17.215815485996703, | |
| "grad_norm": 0.000996377901174128, | |
| "learning_rate": 1.3530434782608698e-06, | |
| "loss": 0.0, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 17.257001647446458, | |
| "grad_norm": 0.0017713250126689672, | |
| "learning_rate": 1.331304347826087e-06, | |
| "loss": 0.0, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 17.29818780889621, | |
| "grad_norm": 0.0008379866485483944, | |
| "learning_rate": 1.3095652173913045e-06, | |
| "loss": 0.0, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 17.29818780889621, | |
| "eval_loss": 0.01441162545233965, | |
| "eval_runtime": 1632.1765, | |
| "eval_samples_per_second": 1.487, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 2.55573681348559, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 17.339373970345964, | |
| "grad_norm": 0.001256891991943121, | |
| "learning_rate": 1.2878260869565219e-06, | |
| "loss": 0.0, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 17.38056013179572, | |
| "grad_norm": 0.003084406489506364, | |
| "learning_rate": 1.2660869565217392e-06, | |
| "loss": 0.0, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 17.42174629324547, | |
| "grad_norm": 0.002908531576395035, | |
| "learning_rate": 1.2443478260869568e-06, | |
| "loss": 0.0, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 17.462932454695224, | |
| "grad_norm": 0.0008340766653418541, | |
| "learning_rate": 1.2226086956521741e-06, | |
| "loss": 0.0, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 17.504118616144975, | |
| "grad_norm": 0.0014115035301074386, | |
| "learning_rate": 1.2008695652173915e-06, | |
| "loss": 0.0, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 17.54530477759473, | |
| "grad_norm": 0.001977501902729273, | |
| "learning_rate": 1.1791304347826088e-06, | |
| "loss": 0.0, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 17.58649093904448, | |
| "grad_norm": 0.0015209122793748975, | |
| "learning_rate": 1.1573913043478262e-06, | |
| "loss": 0.0, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 17.627677100494235, | |
| "grad_norm": 0.0009397733956575394, | |
| "learning_rate": 1.1356521739130435e-06, | |
| "loss": 0.0, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 17.668863261943986, | |
| "grad_norm": 0.0006780756521038711, | |
| "learning_rate": 1.1139130434782609e-06, | |
| "loss": 0.0, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 17.71004942339374, | |
| "grad_norm": 0.001231124042533338, | |
| "learning_rate": 1.0921739130434782e-06, | |
| "loss": 0.0, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 17.751235584843492, | |
| "grad_norm": 0.001504486077465117, | |
| "learning_rate": 1.0704347826086956e-06, | |
| "loss": 0.0, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 17.792421746293247, | |
| "grad_norm": 0.0006280929665081203, | |
| "learning_rate": 1.0486956521739131e-06, | |
| "loss": 0.0, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 17.833607907742998, | |
| "grad_norm": 0.0007227340247482061, | |
| "learning_rate": 1.0269565217391305e-06, | |
| "loss": 0.0, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 17.874794069192752, | |
| "grad_norm": 0.0014692394761368632, | |
| "learning_rate": 1.005217391304348e-06, | |
| "loss": 0.0, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 17.915980230642504, | |
| "grad_norm": 0.0010748576605692506, | |
| "learning_rate": 9.834782608695654e-07, | |
| "loss": 0.0, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 17.957166392092258, | |
| "grad_norm": 0.0012865326134487987, | |
| "learning_rate": 9.617391304347828e-07, | |
| "loss": 0.0, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 17.99835255354201, | |
| "grad_norm": 0.0034256454091519117, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.0, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 18.039538714991764, | |
| "grad_norm": 0.001147790695540607, | |
| "learning_rate": 9.182608695652175e-07, | |
| "loss": 0.0, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 18.080724876441515, | |
| "grad_norm": 0.0010212146444246173, | |
| "learning_rate": 8.965217391304348e-07, | |
| "loss": 0.0, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 18.12191103789127, | |
| "grad_norm": 0.0022493095602840185, | |
| "learning_rate": 8.747826086956522e-07, | |
| "loss": 0.0, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 18.12191103789127, | |
| "eval_loss": 0.014553242363035679, | |
| "eval_runtime": 1629.5576, | |
| "eval_samples_per_second": 1.489, | |
| "eval_steps_per_second": 0.187, | |
| "eval_wer": 2.5168958284782104, | |
| "step": 11000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 12000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.077025230749696e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |