{ "best_metric": 10.977486814291872, "best_model_checkpoint": "/scratch/andoni.sudupe/checkpoints/sweep_new/large_5000/composite_eu+faktoria+Dialektoak_54378/checkpoint-3000", "epoch": 4.201680672268908, "eval_steps": 500, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "grad_norm": 6.300967115012794, "learning_rate": 9.301863326281162e-06, "loss": 4.1278, "step": 25 }, { "epoch": 0.04, "grad_norm": 1.5382915444172507, "learning_rate": 1.2104383916496624e-05, "loss": 0.6712, "step": 50 }, { "epoch": 0.06, "grad_norm": 1.3537379490957926, "learning_rate": 1.3579320331550385e-05, "loss": 0.3782, "step": 75 }, { "epoch": 0.08, "grad_norm": 1.3621261206194586, "learning_rate": 1.4586919790820036e-05, "loss": 0.3061, "step": 100 }, { "epoch": 0.11, "grad_norm": 1.5893966458519595, "learning_rate": 1.5353131038711454e-05, "loss": 0.2708, "step": 125 }, { "epoch": 0.13, "grad_norm": 1.086045129505703, "learning_rate": 1.5971544952351572e-05, "loss": 0.2463, "step": 150 }, { "epoch": 0.15, "grad_norm": 1.0672024280403232, "learning_rate": 1.6490063084441855e-05, "loss": 0.229, "step": 175 }, { "epoch": 0.17, "grad_norm": 1.256059408201239, "learning_rate": 1.6936515080463448e-05, "loss": 0.2162, "step": 200 }, { "epoch": 0.19, "grad_norm": 0.9993845416598218, "learning_rate": 1.7328510430629225e-05, "loss": 0.2107, "step": 225 }, { "epoch": 0.21, "grad_norm": 1.0505260739111444, "learning_rate": 1.7677901345815084e-05, "loss": 0.2009, "step": 250 }, { "epoch": 0.23, "grad_norm": 1.0119873632916145, "learning_rate": 1.7993047645048885e-05, "loss": 0.197, "step": 275 }, { "epoch": 0.25, "grad_norm": 1.139172975392001, "learning_rate": 1.8280066961774116e-05, "loss": 0.1907, "step": 300 }, { "epoch": 0.27, "grad_norm": 1.0104537183647218, "learning_rate": 1.8543571184297487e-05, "loss": 0.1839, "step": 325 }, { "epoch": 0.29, "grad_norm": 1.0867295704130835, "learning_rate": 1.8787123354240354e-05, "loss": 0.1774, "step": 350 }, { "epoch": 0.32, "grad_norm": 0.901420087491565, "learning_rate": 1.9013533418684656e-05, "loss": 0.1786, "step": 375 }, { "epoch": 0.34, "grad_norm": 0.9693540105262828, "learning_rate": 1.922505656756034e-05, "loss": 0.1717, "step": 400 }, { "epoch": 0.36, "grad_norm": 1.0701813443427979, "learning_rate": 1.942353029165938e-05, "loss": 0.1694, "step": 425 }, { "epoch": 0.38, "grad_norm": 0.9490474656276748, "learning_rate": 1.961047156486735e-05, "loss": 0.1643, "step": 450 }, { "epoch": 0.4, "grad_norm": 1.2264007170094673, "learning_rate": 1.9787147310153522e-05, "loss": 0.1648, "step": 475 }, { "epoch": 0.42, "grad_norm": 0.9006603441047074, "learning_rate": 1.9954626505304235e-05, "loss": 0.1619, "step": 500 }, { "epoch": 0.42, "eval_loss": 0.275390625, "eval_runtime": 254.5045, "eval_samples_per_second": 22.487, "eval_steps_per_second": 0.354, "eval_wer": 15.592997722002174, "step": 500 }, { "epoch": 0.44, "grad_norm": 0.8570219249791802, "learning_rate": 1.9924444444444448e-05, "loss": 0.1603, "step": 525 }, { "epoch": 0.46, "grad_norm": 0.9588558320641063, "learning_rate": 1.9813333333333336e-05, "loss": 0.1589, "step": 550 }, { "epoch": 0.48, "grad_norm": 0.9858102779857969, "learning_rate": 1.9702222222222224e-05, "loss": 0.151, "step": 575 }, { "epoch": 0.5, "grad_norm": 0.9483865614257107, "learning_rate": 1.9591111111111112e-05, "loss": 0.1534, "step": 600 }, { "epoch": 0.53, "grad_norm": 0.9868825430699558, "learning_rate": 1.948e-05, "loss": 0.1512, "step": 625 }, { "epoch": 0.55, "grad_norm": 1.0529786571393802, "learning_rate": 1.936888888888889e-05, "loss": 0.1512, "step": 650 }, { "epoch": 0.57, "grad_norm": 0.9385786526950042, "learning_rate": 1.925777777777778e-05, "loss": 0.1498, "step": 675 }, { "epoch": 0.59, "grad_norm": 0.9412097963356688, "learning_rate": 1.9146666666666667e-05, "loss": 0.1447, "step": 700 }, { "epoch": 0.61, "grad_norm": 0.9174465885778958, "learning_rate": 1.9035555555555556e-05, "loss": 0.1497, "step": 725 }, { "epoch": 0.63, "grad_norm": 0.8701008154017025, "learning_rate": 1.8924444444444447e-05, "loss": 0.143, "step": 750 }, { "epoch": 0.65, "grad_norm": 0.8965671476161902, "learning_rate": 1.8813333333333335e-05, "loss": 0.1411, "step": 775 }, { "epoch": 0.67, "grad_norm": 0.8082205975627236, "learning_rate": 1.8702222222222223e-05, "loss": 0.1443, "step": 800 }, { "epoch": 0.69, "grad_norm": 0.9168343860183532, "learning_rate": 1.8591111111111115e-05, "loss": 0.1377, "step": 825 }, { "epoch": 0.71, "grad_norm": 0.8621395034952818, "learning_rate": 1.8480000000000003e-05, "loss": 0.1416, "step": 850 }, { "epoch": 0.74, "grad_norm": 0.7851351075623371, "learning_rate": 1.836888888888889e-05, "loss": 0.1379, "step": 875 }, { "epoch": 0.76, "grad_norm": 0.7712587437064505, "learning_rate": 1.825777777777778e-05, "loss": 0.1364, "step": 900 }, { "epoch": 0.78, "grad_norm": 0.776808588038821, "learning_rate": 1.8146666666666667e-05, "loss": 0.1373, "step": 925 }, { "epoch": 0.8, "grad_norm": 0.9620381696389307, "learning_rate": 1.803555555555556e-05, "loss": 0.1347, "step": 950 }, { "epoch": 0.82, "grad_norm": 0.8161298293001353, "learning_rate": 1.7924444444444446e-05, "loss": 0.1366, "step": 975 }, { "epoch": 0.84, "grad_norm": 0.8477639336460413, "learning_rate": 1.7813333333333334e-05, "loss": 0.1307, "step": 1000 }, { "epoch": 0.84, "eval_loss": 0.23779296875, "eval_runtime": 265.5788, "eval_samples_per_second": 21.549, "eval_steps_per_second": 0.339, "eval_wer": 13.618732940669446, "step": 1000 }, { "epoch": 0.86, "grad_norm": 0.7732694809411719, "learning_rate": 1.7702222222222223e-05, "loss": 0.1296, "step": 1025 }, { "epoch": 0.88, "grad_norm": 0.7697617238940019, "learning_rate": 1.759111111111111e-05, "loss": 0.1309, "step": 1050 }, { "epoch": 0.9, "grad_norm": 0.7793826866229718, "learning_rate": 1.7480000000000002e-05, "loss": 0.132, "step": 1075 }, { "epoch": 0.92, "grad_norm": 0.9097979581204938, "learning_rate": 1.736888888888889e-05, "loss": 0.1294, "step": 1100 }, { "epoch": 0.95, "grad_norm": 0.7947900086298523, "learning_rate": 1.725777777777778e-05, "loss": 0.1243, "step": 1125 }, { "epoch": 0.97, "grad_norm": 0.7449775855001572, "learning_rate": 1.7146666666666666e-05, "loss": 0.1243, "step": 1150 }, { "epoch": 0.99, "grad_norm": 0.9088984299222161, "learning_rate": 1.7035555555555558e-05, "loss": 0.1244, "step": 1175 }, { "epoch": 1.01, "grad_norm": 0.650466685386898, "learning_rate": 1.6924444444444446e-05, "loss": 0.1144, "step": 1200 }, { "epoch": 1.03, "grad_norm": 0.6999200468723945, "learning_rate": 1.6813333333333334e-05, "loss": 0.0872, "step": 1225 }, { "epoch": 1.05, "grad_norm": 0.8130788941620081, "learning_rate": 1.6702222222222225e-05, "loss": 0.0872, "step": 1250 }, { "epoch": 1.07, "grad_norm": 0.7055087900891571, "learning_rate": 1.6591111111111113e-05, "loss": 0.0871, "step": 1275 }, { "epoch": 1.09, "grad_norm": 0.7467724973177869, "learning_rate": 1.648e-05, "loss": 0.0903, "step": 1300 }, { "epoch": 1.11, "grad_norm": 2.0092330241732705, "learning_rate": 1.636888888888889e-05, "loss": 0.0892, "step": 1325 }, { "epoch": 1.13, "grad_norm": 0.8003487139890444, "learning_rate": 1.6257777777777778e-05, "loss": 0.0894, "step": 1350 }, { "epoch": 1.16, "grad_norm": 0.8229895370346713, "learning_rate": 1.614666666666667e-05, "loss": 0.0888, "step": 1375 }, { "epoch": 1.18, "grad_norm": 0.6336447576459033, "learning_rate": 1.6035555555555557e-05, "loss": 0.09, "step": 1400 }, { "epoch": 1.2, "grad_norm": 0.6538764172018886, "learning_rate": 1.5924444444444445e-05, "loss": 0.0891, "step": 1425 }, { "epoch": 1.22, "grad_norm": 0.8798799334784856, "learning_rate": 1.5813333333333333e-05, "loss": 0.0893, "step": 1450 }, { "epoch": 1.24, "grad_norm": 0.722733282159126, "learning_rate": 1.5702222222222225e-05, "loss": 0.0889, "step": 1475 }, { "epoch": 1.26, "grad_norm": 0.6620220475573502, "learning_rate": 1.5591111111111113e-05, "loss": 0.089, "step": 1500 }, { "epoch": 1.26, "eval_loss": 0.2186279296875, "eval_runtime": 261.8264, "eval_samples_per_second": 21.858, "eval_steps_per_second": 0.344, "eval_wer": 12.200627988589488, "step": 1500 }, { "epoch": 1.28, "grad_norm": 0.6941021446505962, "learning_rate": 1.548e-05, "loss": 0.0918, "step": 1525 }, { "epoch": 1.3, "grad_norm": 0.5905994601505266, "learning_rate": 1.5368888888888892e-05, "loss": 0.0875, "step": 1550 }, { "epoch": 1.32, "grad_norm": 0.6232875490671203, "learning_rate": 1.5257777777777779e-05, "loss": 0.0861, "step": 1575 }, { "epoch": 1.34, "grad_norm": 0.677076256543859, "learning_rate": 1.5146666666666667e-05, "loss": 0.0871, "step": 1600 }, { "epoch": 1.37, "grad_norm": 0.6331596599630313, "learning_rate": 1.5035555555555557e-05, "loss": 0.0872, "step": 1625 }, { "epoch": 1.39, "grad_norm": 0.6997423821920836, "learning_rate": 1.4924444444444446e-05, "loss": 0.0867, "step": 1650 }, { "epoch": 1.41, "grad_norm": 0.6012767509409895, "learning_rate": 1.4813333333333334e-05, "loss": 0.0872, "step": 1675 }, { "epoch": 1.43, "grad_norm": 0.6889999240133302, "learning_rate": 1.4702222222222224e-05, "loss": 0.0876, "step": 1700 }, { "epoch": 1.45, "grad_norm": 0.6835162745577436, "learning_rate": 1.4591111111111112e-05, "loss": 0.0883, "step": 1725 }, { "epoch": 1.47, "grad_norm": 0.7487655653314806, "learning_rate": 1.448e-05, "loss": 0.0877, "step": 1750 }, { "epoch": 1.49, "grad_norm": 0.7284627286714441, "learning_rate": 1.436888888888889e-05, "loss": 0.0881, "step": 1775 }, { "epoch": 1.51, "grad_norm": 0.6367149820976872, "learning_rate": 1.425777777777778e-05, "loss": 0.0853, "step": 1800 }, { "epoch": 1.53, "grad_norm": 0.7863662278427825, "learning_rate": 1.4146666666666668e-05, "loss": 0.089, "step": 1825 }, { "epoch": 1.55, "grad_norm": 0.6563802998477202, "learning_rate": 1.4035555555555556e-05, "loss": 0.0867, "step": 1850 }, { "epoch": 1.58, "grad_norm": 0.6751589914560664, "learning_rate": 1.3924444444444446e-05, "loss": 0.0895, "step": 1875 }, { "epoch": 1.6, "grad_norm": 0.6369809973860161, "learning_rate": 1.3813333333333334e-05, "loss": 0.0894, "step": 1900 }, { "epoch": 1.62, "grad_norm": 0.645419762589393, "learning_rate": 1.3702222222222224e-05, "loss": 0.088, "step": 1925 }, { "epoch": 1.64, "grad_norm": 0.8727739833551272, "learning_rate": 1.3591111111111113e-05, "loss": 0.0853, "step": 1950 }, { "epoch": 1.66, "grad_norm": 0.924719737097865, "learning_rate": 1.3480000000000001e-05, "loss": 0.0885, "step": 1975 }, { "epoch": 1.68, "grad_norm": 0.5900362723725564, "learning_rate": 1.336888888888889e-05, "loss": 0.0867, "step": 2000 }, { "epoch": 1.68, "eval_loss": 0.216552734375, "eval_runtime": 257.0462, "eval_samples_per_second": 22.264, "eval_steps_per_second": 0.35, "eval_wer": 11.595214152318016, "step": 2000 }, { "epoch": 1.7, "grad_norm": 0.701196377891906, "learning_rate": 1.3257777777777778e-05, "loss": 0.0865, "step": 2025 }, { "epoch": 1.72, "grad_norm": 0.6686480007640652, "learning_rate": 1.3146666666666667e-05, "loss": 0.0843, "step": 2050 }, { "epoch": 1.74, "grad_norm": 0.6622810703792679, "learning_rate": 1.3035555555555557e-05, "loss": 0.0856, "step": 2075 }, { "epoch": 1.76, "grad_norm": 0.6123958684025059, "learning_rate": 1.2924444444444447e-05, "loss": 0.0838, "step": 2100 }, { "epoch": 1.79, "grad_norm": 0.664953688464529, "learning_rate": 1.2813333333333335e-05, "loss": 0.0857, "step": 2125 }, { "epoch": 1.81, "grad_norm": 0.6536867627910101, "learning_rate": 1.2702222222222223e-05, "loss": 0.0865, "step": 2150 }, { "epoch": 1.83, "grad_norm": 0.5988650594868448, "learning_rate": 1.2591111111111111e-05, "loss": 0.083, "step": 2175 }, { "epoch": 1.85, "grad_norm": 0.586993423148878, "learning_rate": 1.248e-05, "loss": 0.0845, "step": 2200 }, { "epoch": 1.87, "grad_norm": 0.7437532873642021, "learning_rate": 1.236888888888889e-05, "loss": 0.0833, "step": 2225 }, { "epoch": 1.89, "grad_norm": 0.62755674496958, "learning_rate": 1.225777777777778e-05, "loss": 0.0836, "step": 2250 }, { "epoch": 1.91, "grad_norm": 0.6624740226618913, "learning_rate": 1.2146666666666667e-05, "loss": 0.0846, "step": 2275 }, { "epoch": 1.93, "grad_norm": 0.6438709923084966, "learning_rate": 1.2035555555555556e-05, "loss": 0.086, "step": 2300 }, { "epoch": 1.95, "grad_norm": 0.7165576421254418, "learning_rate": 1.1924444444444445e-05, "loss": 0.0841, "step": 2325 }, { "epoch": 1.97, "grad_norm": 0.662096299123914, "learning_rate": 1.1813333333333334e-05, "loss": 0.0819, "step": 2350 }, { "epoch": 2.0, "grad_norm": 0.6543324475084017, "learning_rate": 1.1702222222222224e-05, "loss": 0.0872, "step": 2375 }, { "epoch": 2.02, "grad_norm": 0.5437283588356018, "learning_rate": 1.1591111111111114e-05, "loss": 0.0597, "step": 2400 }, { "epoch": 2.04, "grad_norm": 0.5484741263915869, "learning_rate": 1.148e-05, "loss": 0.0514, "step": 2425 }, { "epoch": 2.06, "grad_norm": 0.555239389444415, "learning_rate": 1.136888888888889e-05, "loss": 0.0525, "step": 2450 }, { "epoch": 2.08, "grad_norm": 0.5304195139624485, "learning_rate": 1.1257777777777778e-05, "loss": 0.0522, "step": 2475 }, { "epoch": 2.1, "grad_norm": 0.525869272014523, "learning_rate": 1.1146666666666668e-05, "loss": 0.0512, "step": 2500 }, { "epoch": 2.1, "eval_loss": 0.217041015625, "eval_runtime": 262.3433, "eval_samples_per_second": 21.815, "eval_steps_per_second": 0.343, "eval_wer": 11.000061567508773, "step": 2500 }, { "epoch": 2.12, "grad_norm": 0.5045125404118772, "learning_rate": 1.1035555555555558e-05, "loss": 0.0517, "step": 2525 }, { "epoch": 2.14, "grad_norm": 0.47949137949826903, "learning_rate": 1.0924444444444446e-05, "loss": 0.0505, "step": 2550 }, { "epoch": 2.16, "grad_norm": 0.6532578602408089, "learning_rate": 1.0813333333333334e-05, "loss": 0.0529, "step": 2575 }, { "epoch": 2.18, "grad_norm": 0.49071983222718196, "learning_rate": 1.0702222222222223e-05, "loss": 0.0506, "step": 2600 }, { "epoch": 2.21, "grad_norm": 0.4784105458007845, "learning_rate": 1.0591111111111112e-05, "loss": 0.0517, "step": 2625 }, { "epoch": 2.23, "grad_norm": 0.5758000095163379, "learning_rate": 1.0480000000000001e-05, "loss": 0.0521, "step": 2650 }, { "epoch": 2.25, "grad_norm": 0.5319566514739653, "learning_rate": 1.0368888888888891e-05, "loss": 0.0513, "step": 2675 }, { "epoch": 2.27, "grad_norm": 0.5459225671447665, "learning_rate": 1.0257777777777777e-05, "loss": 0.0522, "step": 2700 }, { "epoch": 2.29, "grad_norm": 0.5133886390417978, "learning_rate": 1.0146666666666667e-05, "loss": 0.0522, "step": 2725 }, { "epoch": 2.31, "grad_norm": 0.7459847735162664, "learning_rate": 1.0035555555555557e-05, "loss": 0.051, "step": 2750 }, { "epoch": 2.33, "grad_norm": 0.6816400128993043, "learning_rate": 9.924444444444445e-06, "loss": 0.0519, "step": 2775 }, { "epoch": 2.35, "grad_norm": 0.5777573576425261, "learning_rate": 9.813333333333333e-06, "loss": 0.052, "step": 2800 }, { "epoch": 2.37, "grad_norm": 0.6170633408447892, "learning_rate": 9.702222222222223e-06, "loss": 0.0519, "step": 2825 }, { "epoch": 2.39, "grad_norm": 0.5296607649366135, "learning_rate": 9.591111111111113e-06, "loss": 0.0527, "step": 2850 }, { "epoch": 2.42, "grad_norm": 0.5463919004411076, "learning_rate": 9.48e-06, "loss": 0.0526, "step": 2875 }, { "epoch": 2.44, "grad_norm": 0.6350401768777562, "learning_rate": 9.368888888888889e-06, "loss": 0.052, "step": 2900 }, { "epoch": 2.46, "grad_norm": 0.6226455687560415, "learning_rate": 9.257777777777779e-06, "loss": 0.0509, "step": 2925 }, { "epoch": 2.48, "grad_norm": 0.49602652266946257, "learning_rate": 9.146666666666667e-06, "loss": 0.0532, "step": 2950 }, { "epoch": 2.5, "grad_norm": 0.5354049550221535, "learning_rate": 9.035555555555556e-06, "loss": 0.0496, "step": 2975 }, { "epoch": 2.52, "grad_norm": 0.5464331340421218, "learning_rate": 8.924444444444446e-06, "loss": 0.0519, "step": 3000 }, { "epoch": 2.52, "eval_loss": 0.2164306640625, "eval_runtime": 256.1912, "eval_samples_per_second": 22.339, "eval_steps_per_second": 0.351, "eval_wer": 10.977486814291872, "step": 3000 }, { "epoch": 2.54, "grad_norm": 0.4673220979603608, "learning_rate": 8.813333333333334e-06, "loss": 0.0527, "step": 3025 }, { "epoch": 2.56, "grad_norm": 0.5184330052140134, "learning_rate": 8.702222222222222e-06, "loss": 0.0539, "step": 3050 }, { "epoch": 2.58, "grad_norm": 0.5791237941031989, "learning_rate": 8.591111111111112e-06, "loss": 0.0518, "step": 3075 }, { "epoch": 2.61, "grad_norm": 0.6132255091514093, "learning_rate": 8.48e-06, "loss": 0.0511, "step": 3100 }, { "epoch": 2.63, "grad_norm": 0.47390316198432736, "learning_rate": 8.36888888888889e-06, "loss": 0.0488, "step": 3125 }, { "epoch": 2.65, "grad_norm": 0.5868796340060679, "learning_rate": 8.25777777777778e-06, "loss": 0.0529, "step": 3150 }, { "epoch": 2.67, "grad_norm": 0.6708439967293209, "learning_rate": 8.146666666666668e-06, "loss": 0.0516, "step": 3175 }, { "epoch": 2.69, "grad_norm": 0.5844474410989317, "learning_rate": 8.035555555555556e-06, "loss": 0.0526, "step": 3200 }, { "epoch": 2.71, "grad_norm": 0.5750910044262906, "learning_rate": 7.924444444444444e-06, "loss": 0.051, "step": 3225 }, { "epoch": 2.73, "grad_norm": 0.5727131366447362, "learning_rate": 7.813333333333334e-06, "loss": 0.0495, "step": 3250 }, { "epoch": 2.75, "grad_norm": 0.573906950084281, "learning_rate": 7.702222222222223e-06, "loss": 0.0518, "step": 3275 }, { "epoch": 2.77, "grad_norm": 0.5680227954447488, "learning_rate": 7.5911111111111115e-06, "loss": 0.053, "step": 3300 }, { "epoch": 2.79, "grad_norm": 0.5505010013623186, "learning_rate": 7.48e-06, "loss": 0.05, "step": 3325 }, { "epoch": 2.82, "grad_norm": 0.5491474349494719, "learning_rate": 7.36888888888889e-06, "loss": 0.0515, "step": 3350 }, { "epoch": 2.84, "grad_norm": 0.578208015255246, "learning_rate": 7.257777777777778e-06, "loss": 0.05, "step": 3375 }, { "epoch": 2.86, "grad_norm": 0.5331859655728193, "learning_rate": 7.146666666666667e-06, "loss": 0.0508, "step": 3400 }, { "epoch": 2.88, "grad_norm": 0.5388653712641273, "learning_rate": 7.035555555555557e-06, "loss": 0.0497, "step": 3425 }, { "epoch": 2.9, "grad_norm": 0.6023270591075864, "learning_rate": 6.924444444444445e-06, "loss": 0.0534, "step": 3450 }, { "epoch": 2.92, "grad_norm": 0.5262576751874167, "learning_rate": 6.813333333333334e-06, "loss": 0.0498, "step": 3475 }, { "epoch": 2.94, "grad_norm": 0.5160078303681851, "learning_rate": 6.702222222222224e-06, "loss": 0.0511, "step": 3500 }, { "epoch": 2.94, "eval_loss": 0.217529296875, "eval_runtime": 261.2146, "eval_samples_per_second": 21.909, "eval_steps_per_second": 0.345, "eval_wer": 11.014427319555892, "step": 3500 }, { "epoch": 2.96, "grad_norm": 0.5462421570308692, "learning_rate": 6.591111111111112e-06, "loss": 0.0503, "step": 3525 }, { "epoch": 2.98, "grad_norm": 0.5492582977077912, "learning_rate": 6.480000000000001e-06, "loss": 0.0524, "step": 3550 }, { "epoch": 3.0, "grad_norm": 0.4126516634113135, "learning_rate": 6.368888888888889e-06, "loss": 0.0452, "step": 3575 }, { "epoch": 3.03, "grad_norm": 0.42807623731704686, "learning_rate": 6.2577777777777785e-06, "loss": 0.0292, "step": 3600 }, { "epoch": 3.05, "grad_norm": 0.38312185205283367, "learning_rate": 6.146666666666667e-06, "loss": 0.0283, "step": 3625 }, { "epoch": 3.07, "grad_norm": 0.39521126638291676, "learning_rate": 6.0355555555555555e-06, "loss": 0.0265, "step": 3650 }, { "epoch": 3.09, "grad_norm": 0.4079733023650659, "learning_rate": 5.924444444444445e-06, "loss": 0.0271, "step": 3675 }, { "epoch": 3.11, "grad_norm": 0.4335194257664806, "learning_rate": 5.813333333333334e-06, "loss": 0.0279, "step": 3700 }, { "epoch": 3.13, "grad_norm": 0.4014227984066138, "learning_rate": 5.702222222222222e-06, "loss": 0.0265, "step": 3725 }, { "epoch": 3.15, "grad_norm": 0.4205375416368579, "learning_rate": 5.591111111111112e-06, "loss": 0.0287, "step": 3750 }, { "epoch": 3.17, "grad_norm": 0.40781309120456916, "learning_rate": 5.480000000000001e-06, "loss": 0.0263, "step": 3775 }, { "epoch": 3.19, "grad_norm": 0.4715762133465575, "learning_rate": 5.368888888888889e-06, "loss": 0.0274, "step": 3800 }, { "epoch": 3.21, "grad_norm": 0.5080004417650299, "learning_rate": 5.257777777777779e-06, "loss": 0.0276, "step": 3825 }, { "epoch": 3.24, "grad_norm": 0.5878579520337655, "learning_rate": 5.146666666666668e-06, "loss": 0.0273, "step": 3850 }, { "epoch": 3.26, "grad_norm": 0.442150501280443, "learning_rate": 5.035555555555556e-06, "loss": 0.0272, "step": 3875 }, { "epoch": 3.28, "grad_norm": 0.5333202359210575, "learning_rate": 4.924444444444445e-06, "loss": 0.0283, "step": 3900 }, { "epoch": 3.3, "grad_norm": 0.4508059780802538, "learning_rate": 4.8133333333333336e-06, "loss": 0.0267, "step": 3925 }, { "epoch": 3.32, "grad_norm": 0.3602847319041051, "learning_rate": 4.7022222222222225e-06, "loss": 0.0264, "step": 3950 }, { "epoch": 3.34, "grad_norm": 0.4383791707038719, "learning_rate": 4.591111111111111e-06, "loss": 0.0273, "step": 3975 }, { "epoch": 3.36, "grad_norm": 0.42491548342981733, "learning_rate": 4.48e-06, "loss": 0.027, "step": 4000 }, { "epoch": 3.36, "eval_loss": 0.234375, "eval_runtime": 258.5498, "eval_samples_per_second": 22.135, "eval_steps_per_second": 0.348, "eval_wer": 11.071890327744372, "step": 4000 }, { "epoch": 3.38, "grad_norm": 0.4042003903460706, "learning_rate": 4.368888888888889e-06, "loss": 0.0275, "step": 4025 }, { "epoch": 3.4, "grad_norm": 0.42382139209374314, "learning_rate": 4.257777777777778e-06, "loss": 0.0276, "step": 4050 }, { "epoch": 3.42, "grad_norm": 0.40426989479155856, "learning_rate": 4.146666666666667e-06, "loss": 0.0265, "step": 4075 }, { "epoch": 3.45, "grad_norm": 0.4784873798364331, "learning_rate": 4.035555555555556e-06, "loss": 0.0264, "step": 4100 }, { "epoch": 3.47, "grad_norm": 0.4711675450109133, "learning_rate": 3.924444444444445e-06, "loss": 0.0276, "step": 4125 }, { "epoch": 3.49, "grad_norm": 0.49654117442408047, "learning_rate": 3.813333333333334e-06, "loss": 0.0272, "step": 4150 }, { "epoch": 3.51, "grad_norm": 0.4258565398678053, "learning_rate": 3.7022222222222227e-06, "loss": 0.0266, "step": 4175 }, { "epoch": 3.53, "grad_norm": 0.3767157286400597, "learning_rate": 3.5911111111111112e-06, "loss": 0.0263, "step": 4200 }, { "epoch": 3.55, "grad_norm": 0.4514847863455036, "learning_rate": 3.48e-06, "loss": 0.026, "step": 4225 }, { "epoch": 3.57, "grad_norm": 0.41051422119946, "learning_rate": 3.3688888888888895e-06, "loss": 0.0261, "step": 4250 }, { "epoch": 3.59, "grad_norm": 0.4041876161527076, "learning_rate": 3.257777777777778e-06, "loss": 0.0259, "step": 4275 }, { "epoch": 3.61, "grad_norm": 0.46742684857876726, "learning_rate": 3.146666666666667e-06, "loss": 0.0276, "step": 4300 }, { "epoch": 3.63, "grad_norm": 0.4193045758556068, "learning_rate": 3.0355555555555562e-06, "loss": 0.0268, "step": 4325 }, { "epoch": 3.66, "grad_norm": 0.4352545505627611, "learning_rate": 2.9244444444444447e-06, "loss": 0.0276, "step": 4350 }, { "epoch": 3.68, "grad_norm": 0.39115210255941957, "learning_rate": 2.8133333333333336e-06, "loss": 0.027, "step": 4375 }, { "epoch": 3.7, "grad_norm": 0.4268550980146758, "learning_rate": 2.702222222222222e-06, "loss": 0.0273, "step": 4400 }, { "epoch": 3.72, "grad_norm": 0.46681427224911076, "learning_rate": 2.5911111111111115e-06, "loss": 0.026, "step": 4425 }, { "epoch": 3.74, "grad_norm": 0.4106693872647045, "learning_rate": 2.4800000000000004e-06, "loss": 0.0268, "step": 4450 }, { "epoch": 3.76, "grad_norm": 0.42538670099303383, "learning_rate": 2.3688888888888893e-06, "loss": 0.0267, "step": 4475 }, { "epoch": 3.78, "grad_norm": 0.6429249485526561, "learning_rate": 2.257777777777778e-06, "loss": 0.0265, "step": 4500 }, { "epoch": 3.78, "eval_loss": 0.2364501953125, "eval_runtime": 264.6732, "eval_samples_per_second": 21.623, "eval_steps_per_second": 0.34, "eval_wer": 10.971330063414534, "step": 4500 }, { "epoch": 3.8, "grad_norm": 0.3830387264461008, "learning_rate": 2.1466666666666667e-06, "loss": 0.0269, "step": 4525 }, { "epoch": 3.82, "grad_norm": 0.4257490863375848, "learning_rate": 2.0355555555555556e-06, "loss": 0.0256, "step": 4550 }, { "epoch": 3.84, "grad_norm": 0.38631508692387634, "learning_rate": 1.9244444444444446e-06, "loss": 0.0266, "step": 4575 }, { "epoch": 3.87, "grad_norm": 0.4115019203792458, "learning_rate": 1.8133333333333337e-06, "loss": 0.0253, "step": 4600 }, { "epoch": 3.89, "grad_norm": 0.39780077827846405, "learning_rate": 1.7022222222222224e-06, "loss": 0.0267, "step": 4625 }, { "epoch": 3.91, "grad_norm": 0.41384273671054456, "learning_rate": 1.5911111111111113e-06, "loss": 0.0259, "step": 4650 }, { "epoch": 3.93, "grad_norm": 0.4461683949084482, "learning_rate": 1.48e-06, "loss": 0.0249, "step": 4675 }, { "epoch": 3.95, "grad_norm": 0.37108001683573033, "learning_rate": 1.3688888888888891e-06, "loss": 0.0257, "step": 4700 }, { "epoch": 3.97, "grad_norm": 0.47497418986017037, "learning_rate": 1.2577777777777779e-06, "loss": 0.0253, "step": 4725 }, { "epoch": 3.99, "grad_norm": 0.3993376306652589, "learning_rate": 1.1466666666666668e-06, "loss": 0.025, "step": 4750 }, { "epoch": 4.01, "grad_norm": 0.3507701846546068, "learning_rate": 1.0355555555555557e-06, "loss": 0.0198, "step": 4775 }, { "epoch": 4.03, "grad_norm": 0.34807084040650665, "learning_rate": 9.244444444444445e-07, "loss": 0.0151, "step": 4800 }, { "epoch": 4.05, "grad_norm": 0.2981778718141128, "learning_rate": 8.133333333333333e-07, "loss": 0.0156, "step": 4825 }, { "epoch": 4.08, "grad_norm": 0.29360191589660645, "learning_rate": 7.022222222222223e-07, "loss": 0.0143, "step": 4850 }, { "epoch": 4.1, "grad_norm": 0.34052887181343144, "learning_rate": 5.911111111111111e-07, "loss": 0.0151, "step": 4875 }, { "epoch": 4.12, "grad_norm": 0.32503785224710324, "learning_rate": 4.800000000000001e-07, "loss": 0.0146, "step": 4900 }, { "epoch": 4.14, "grad_norm": 0.30465948758589884, "learning_rate": 3.6888888888888893e-07, "loss": 0.0157, "step": 4925 }, { "epoch": 4.16, "grad_norm": 0.26092244464591186, "learning_rate": 2.577777777777778e-07, "loss": 0.0144, "step": 4950 }, { "epoch": 4.18, "grad_norm": 0.31357304404438757, "learning_rate": 1.4666666666666668e-07, "loss": 0.0147, "step": 4975 }, { "epoch": 4.2, "grad_norm": 0.31199949237586566, "learning_rate": 3.555555555555556e-08, "loss": 0.0144, "step": 5000 }, { "epoch": 4.2, "eval_loss": 0.2529296875, "eval_runtime": 262.8201, "eval_samples_per_second": 21.775, "eval_steps_per_second": 0.342, "eval_wer": 11.25248835347959, "step": 5000 }, { "epoch": 4.2, "step": 5000, "total_flos": 5.435317864264158e+21, "train_loss": 0.10249043579101562, "train_runtime": 84635.6038, "train_samples_per_second": 30.247, "train_steps_per_second": 0.059 } ], "logging_steps": 25, "max_steps": 5000, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 1000, "total_flos": 5.435317864264158e+21, "train_batch_size": 256, "trial_name": null, "trial_params": null }