| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.833180568285976, | |
| "eval_steps": 1000, | |
| "global_step": 4000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.011457378551787351, | |
| "grad_norm": 3.879394769668579, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.5663, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.022914757103574702, | |
| "grad_norm": 4.039769649505615, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.5106, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.034372135655362054, | |
| "grad_norm": 4.186612129211426, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.4423, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.045829514207149404, | |
| "grad_norm": 2.756096839904785, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.3852, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05728689275893675, | |
| "grad_norm": 2.9499497413635254, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.3232, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06874427131072411, | |
| "grad_norm": 2.5358545780181885, | |
| "learning_rate": 3e-06, | |
| "loss": 0.3331, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08020164986251145, | |
| "grad_norm": 2.2754788398742676, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.3308, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09165902841429881, | |
| "grad_norm": 2.565856695175171, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.3264, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10311640696608616, | |
| "grad_norm": 2.5726332664489746, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.3006, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1145737855178735, | |
| "grad_norm": 2.47786021232605, | |
| "learning_rate": 5e-06, | |
| "loss": 0.2915, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.12603116406966086, | |
| "grad_norm": 2.316861391067505, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.293, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.13748854262144822, | |
| "grad_norm": 2.3276755809783936, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2825, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.14894592117323557, | |
| "grad_norm": 2.4386003017425537, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3054, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.1604032997250229, | |
| "grad_norm": 2.051398277282715, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2833, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.17186067827681026, | |
| "grad_norm": 2.4201934337615967, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2796, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.18331805682859761, | |
| "grad_norm": 1.9952783584594727, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.275, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.19477543538038497, | |
| "grad_norm": 2.1200835704803467, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.2898, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.20623281393217233, | |
| "grad_norm": 2.4650216102600098, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2645, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.21769019248395968, | |
| "grad_norm": 1.9929370880126953, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.278, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.229147571035747, | |
| "grad_norm": 2.1805336475372314, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2702, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.24060494958753437, | |
| "grad_norm": 2.0869078636169434, | |
| "learning_rate": 9.92857142857143e-06, | |
| "loss": 0.2454, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2520623281393217, | |
| "grad_norm": 1.7062550783157349, | |
| "learning_rate": 9.857142857142859e-06, | |
| "loss": 0.2534, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2635197066911091, | |
| "grad_norm": 1.6921783685684204, | |
| "learning_rate": 9.785714285714286e-06, | |
| "loss": 0.2398, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.27497708524289644, | |
| "grad_norm": 2.111154794692993, | |
| "learning_rate": 9.714285714285715e-06, | |
| "loss": 0.2582, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2864344637946838, | |
| "grad_norm": 2.4726927280426025, | |
| "learning_rate": 9.642857142857144e-06, | |
| "loss": 0.2365, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.29789184234647115, | |
| "grad_norm": 1.8263903856277466, | |
| "learning_rate": 9.571428571428573e-06, | |
| "loss": 0.2363, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3093492208982585, | |
| "grad_norm": 2.1429872512817383, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2355, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.3208065994500458, | |
| "grad_norm": 1.7079923152923584, | |
| "learning_rate": 9.42857142857143e-06, | |
| "loss": 0.2291, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.33226397800183316, | |
| "grad_norm": 2.1364455223083496, | |
| "learning_rate": 9.357142857142859e-06, | |
| "loss": 0.2284, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.3437213565536205, | |
| "grad_norm": 2.03523588180542, | |
| "learning_rate": 9.285714285714288e-06, | |
| "loss": 0.2439, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.3551787351054079, | |
| "grad_norm": 1.883085012435913, | |
| "learning_rate": 9.214285714285715e-06, | |
| "loss": 0.2238, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.36663611365719523, | |
| "grad_norm": 2.125384569168091, | |
| "learning_rate": 9.142857142857144e-06, | |
| "loss": 0.2335, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3780934922089826, | |
| "grad_norm": 2.0674521923065186, | |
| "learning_rate": 9.071428571428573e-06, | |
| "loss": 0.235, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.38955087076076994, | |
| "grad_norm": 1.971111536026001, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2343, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.4010082493125573, | |
| "grad_norm": 1.8734443187713623, | |
| "learning_rate": 8.92857142857143e-06, | |
| "loss": 0.2122, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.41246562786434465, | |
| "grad_norm": 1.9618253707885742, | |
| "learning_rate": 8.857142857142858e-06, | |
| "loss": 0.2189, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.423923006416132, | |
| "grad_norm": 2.4546687602996826, | |
| "learning_rate": 8.785714285714286e-06, | |
| "loss": 0.2137, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.43538038496791936, | |
| "grad_norm": 1.8629581928253174, | |
| "learning_rate": 8.714285714285715e-06, | |
| "loss": 0.214, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.44683776351970667, | |
| "grad_norm": 1.8550409078598022, | |
| "learning_rate": 8.642857142857144e-06, | |
| "loss": 0.2124, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.458295142071494, | |
| "grad_norm": 1.9905967712402344, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 0.23, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.458295142071494, | |
| "eval_loss": 0.2573724389076233, | |
| "eval_runtime": 2144.4724, | |
| "eval_samples_per_second": 2.509, | |
| "eval_steps_per_second": 0.157, | |
| "eval_wer": 0.19923257385979562, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4697525206232814, | |
| "grad_norm": 1.6646443605422974, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.2414, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.48120989917506873, | |
| "grad_norm": 2.01544189453125, | |
| "learning_rate": 8.428571428571429e-06, | |
| "loss": 0.2209, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4926672777268561, | |
| "grad_norm": 2.151685953140259, | |
| "learning_rate": 8.357142857142858e-06, | |
| "loss": 0.2155, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.5041246562786434, | |
| "grad_norm": 2.1384220123291016, | |
| "learning_rate": 8.285714285714287e-06, | |
| "loss": 0.212, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5155820348304307, | |
| "grad_norm": 1.722456932067871, | |
| "learning_rate": 8.214285714285714e-06, | |
| "loss": 0.1897, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.5270394133822182, | |
| "grad_norm": 1.6334404945373535, | |
| "learning_rate": 8.142857142857143e-06, | |
| "loss": 0.1983, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5384967919340055, | |
| "grad_norm": 1.665479063987732, | |
| "learning_rate": 8.071428571428572e-06, | |
| "loss": 0.1906, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.5499541704857929, | |
| "grad_norm": 1.7400776147842407, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2024, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.5614115490375802, | |
| "grad_norm": 1.9680901765823364, | |
| "learning_rate": 7.928571428571429e-06, | |
| "loss": 0.2043, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.5728689275893676, | |
| "grad_norm": 1.9159817695617676, | |
| "learning_rate": 7.857142857142858e-06, | |
| "loss": 0.2203, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5843263061411549, | |
| "grad_norm": 2.0622291564941406, | |
| "learning_rate": 7.785714285714287e-06, | |
| "loss": 0.2139, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5957836846929423, | |
| "grad_norm": 1.821271538734436, | |
| "learning_rate": 7.714285714285716e-06, | |
| "loss": 0.2035, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6072410632447296, | |
| "grad_norm": 2.25801944732666, | |
| "learning_rate": 7.642857142857143e-06, | |
| "loss": 0.2112, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.618698441796517, | |
| "grad_norm": 1.8023837804794312, | |
| "learning_rate": 7.571428571428572e-06, | |
| "loss": 0.1935, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.6301558203483043, | |
| "grad_norm": 2.021291971206665, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1928, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.6416131989000916, | |
| "grad_norm": 1.789577841758728, | |
| "learning_rate": 7.428571428571429e-06, | |
| "loss": 0.1967, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.653070577451879, | |
| "grad_norm": 1.4411484003067017, | |
| "learning_rate": 7.357142857142858e-06, | |
| "loss": 0.1806, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.6645279560036663, | |
| "grad_norm": 1.5292807817459106, | |
| "learning_rate": 7.285714285714286e-06, | |
| "loss": 0.2042, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.6759853345554537, | |
| "grad_norm": 1.9637993574142456, | |
| "learning_rate": 7.2142857142857145e-06, | |
| "loss": 0.1944, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.687442713107241, | |
| "grad_norm": 1.4942073822021484, | |
| "learning_rate": 7.1428571428571436e-06, | |
| "loss": 0.1786, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6989000916590284, | |
| "grad_norm": 2.0111465454101562, | |
| "learning_rate": 7.0714285714285726e-06, | |
| "loss": 0.1955, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.7103574702108157, | |
| "grad_norm": 1.8509060144424438, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1878, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.7218148487626032, | |
| "grad_norm": 1.9552288055419922, | |
| "learning_rate": 6.928571428571429e-06, | |
| "loss": 0.192, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.7332722273143905, | |
| "grad_norm": 1.8190360069274902, | |
| "learning_rate": 6.857142857142858e-06, | |
| "loss": 0.1906, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.7447296058661779, | |
| "grad_norm": 1.8309643268585205, | |
| "learning_rate": 6.785714285714287e-06, | |
| "loss": 0.1727, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.7561869844179652, | |
| "grad_norm": 1.8337007761001587, | |
| "learning_rate": 6.714285714285714e-06, | |
| "loss": 0.1783, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.7676443629697525, | |
| "grad_norm": 1.7660903930664062, | |
| "learning_rate": 6.642857142857143e-06, | |
| "loss": 0.1997, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.7791017415215399, | |
| "grad_norm": 2.344595193862915, | |
| "learning_rate": 6.571428571428572e-06, | |
| "loss": 0.1944, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7905591200733272, | |
| "grad_norm": 1.631115436553955, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.181, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.8020164986251146, | |
| "grad_norm": 1.8305103778839111, | |
| "learning_rate": 6.4285714285714295e-06, | |
| "loss": 0.1685, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.8134738771769019, | |
| "grad_norm": 1.9758498668670654, | |
| "learning_rate": 6.357142857142858e-06, | |
| "loss": 0.1714, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.8249312557286893, | |
| "grad_norm": 1.4304888248443604, | |
| "learning_rate": 6.285714285714286e-06, | |
| "loss": 0.1621, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.8363886342804766, | |
| "grad_norm": 2.101116418838501, | |
| "learning_rate": 6.214285714285715e-06, | |
| "loss": 0.1754, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.847846012832264, | |
| "grad_norm": 1.6567695140838623, | |
| "learning_rate": 6.142857142857144e-06, | |
| "loss": 0.1659, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.8593033913840513, | |
| "grad_norm": 1.8299853801727295, | |
| "learning_rate": 6.071428571428571e-06, | |
| "loss": 0.1863, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.8707607699358387, | |
| "grad_norm": 1.789523720741272, | |
| "learning_rate": 6e-06, | |
| "loss": 0.1756, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.882218148487626, | |
| "grad_norm": 1.8699467182159424, | |
| "learning_rate": 5.928571428571429e-06, | |
| "loss": 0.1937, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.8936755270394133, | |
| "grad_norm": 1.700272560119629, | |
| "learning_rate": 5.857142857142858e-06, | |
| "loss": 0.186, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.9051329055912007, | |
| "grad_norm": 2.002411127090454, | |
| "learning_rate": 5.785714285714286e-06, | |
| "loss": 0.1635, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.916590284142988, | |
| "grad_norm": 1.8900529146194458, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 0.1775, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.916590284142988, | |
| "eval_loss": 0.25274595618247986, | |
| "eval_runtime": 2153.2081, | |
| "eval_samples_per_second": 2.499, | |
| "eval_steps_per_second": 0.157, | |
| "eval_wer": 0.20147918406298462, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.9280476626947755, | |
| "grad_norm": 2.0428643226623535, | |
| "learning_rate": 5.6428571428571435e-06, | |
| "loss": 0.1785, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.9395050412465628, | |
| "grad_norm": 1.6772222518920898, | |
| "learning_rate": 5.571428571428572e-06, | |
| "loss": 0.1748, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.9509624197983502, | |
| "grad_norm": 1.6048625707626343, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.1819, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.9624197983501375, | |
| "grad_norm": 1.6015398502349854, | |
| "learning_rate": 5.428571428571429e-06, | |
| "loss": 0.1642, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.9738771769019249, | |
| "grad_norm": 1.4918720722198486, | |
| "learning_rate": 5.357142857142857e-06, | |
| "loss": 0.1746, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.9853345554537122, | |
| "grad_norm": 1.2924649715423584, | |
| "learning_rate": 5.285714285714286e-06, | |
| "loss": 0.171, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.9967919340054996, | |
| "grad_norm": 2.293318748474121, | |
| "learning_rate": 5.214285714285715e-06, | |
| "loss": 0.1677, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.008249312557287, | |
| "grad_norm": 1.5060619115829468, | |
| "learning_rate": 5.142857142857142e-06, | |
| "loss": 0.1174, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.0197066911090742, | |
| "grad_norm": 1.4803202152252197, | |
| "learning_rate": 5.071428571428571e-06, | |
| "loss": 0.1145, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.0311640696608615, | |
| "grad_norm": 1.4294520616531372, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1055, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.042621448212649, | |
| "grad_norm": 1.2599146366119385, | |
| "learning_rate": 4.928571428571429e-06, | |
| "loss": 0.0992, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.0540788267644363, | |
| "grad_norm": 1.4009997844696045, | |
| "learning_rate": 4.857142857142858e-06, | |
| "loss": 0.0975, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.0655362053162236, | |
| "grad_norm": 1.5712811946868896, | |
| "learning_rate": 4.785714285714287e-06, | |
| "loss": 0.1084, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.076993583868011, | |
| "grad_norm": 1.3722585439682007, | |
| "learning_rate": 4.714285714285715e-06, | |
| "loss": 0.1118, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.0884509624197984, | |
| "grad_norm": 1.5365544557571411, | |
| "learning_rate": 4.642857142857144e-06, | |
| "loss": 0.1082, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0999083409715857, | |
| "grad_norm": 1.2618945837020874, | |
| "learning_rate": 4.571428571428572e-06, | |
| "loss": 0.1173, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.111365719523373, | |
| "grad_norm": 1.70241379737854, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1015, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.1228230980751603, | |
| "grad_norm": 1.8518885374069214, | |
| "learning_rate": 4.428571428571429e-06, | |
| "loss": 0.1122, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.1342804766269476, | |
| "grad_norm": 1.4387401342391968, | |
| "learning_rate": 4.357142857142857e-06, | |
| "loss": 0.1139, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.1457378551787352, | |
| "grad_norm": 1.3494459390640259, | |
| "learning_rate": 4.2857142857142855e-06, | |
| "loss": 0.106, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.1571952337305225, | |
| "grad_norm": 1.4629216194152832, | |
| "learning_rate": 4.2142857142857145e-06, | |
| "loss": 0.1073, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.1686526122823098, | |
| "grad_norm": 1.5630906820297241, | |
| "learning_rate": 4.1428571428571435e-06, | |
| "loss": 0.0988, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.180109990834097, | |
| "grad_norm": 1.3475199937820435, | |
| "learning_rate": 4.071428571428572e-06, | |
| "loss": 0.1094, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.1915673693858846, | |
| "grad_norm": 1.1417912244796753, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.104, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.203024747937672, | |
| "grad_norm": 1.4655263423919678, | |
| "learning_rate": 3.928571428571429e-06, | |
| "loss": 0.113, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.2144821264894592, | |
| "grad_norm": 1.585945963859558, | |
| "learning_rate": 3.857142857142858e-06, | |
| "loss": 0.1079, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.2259395050412465, | |
| "grad_norm": 1.3465677499771118, | |
| "learning_rate": 3.785714285714286e-06, | |
| "loss": 0.1121, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.2373968835930338, | |
| "grad_norm": 1.6796523332595825, | |
| "learning_rate": 3.7142857142857146e-06, | |
| "loss": 0.1181, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.2488542621448213, | |
| "grad_norm": 1.2698748111724854, | |
| "learning_rate": 3.642857142857143e-06, | |
| "loss": 0.1105, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.2603116406966086, | |
| "grad_norm": 1.820233941078186, | |
| "learning_rate": 3.5714285714285718e-06, | |
| "loss": 0.1118, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.271769019248396, | |
| "grad_norm": 1.2463735342025757, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.1026, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.2832263978001834, | |
| "grad_norm": 2.136754274368286, | |
| "learning_rate": 3.428571428571429e-06, | |
| "loss": 0.1173, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.2946837763519707, | |
| "grad_norm": 1.8479666709899902, | |
| "learning_rate": 3.357142857142857e-06, | |
| "loss": 0.1104, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.306141154903758, | |
| "grad_norm": 1.870710849761963, | |
| "learning_rate": 3.285714285714286e-06, | |
| "loss": 0.1151, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.3175985334555453, | |
| "grad_norm": 1.6915826797485352, | |
| "learning_rate": 3.2142857142857147e-06, | |
| "loss": 0.1092, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.3290559120073326, | |
| "grad_norm": 1.5454246997833252, | |
| "learning_rate": 3.142857142857143e-06, | |
| "loss": 0.1056, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.34051329055912, | |
| "grad_norm": 1.4232581853866577, | |
| "learning_rate": 3.071428571428572e-06, | |
| "loss": 0.0981, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.3519706691109075, | |
| "grad_norm": 1.5464751720428467, | |
| "learning_rate": 3e-06, | |
| "loss": 0.1051, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.3634280476626948, | |
| "grad_norm": 1.617953896522522, | |
| "learning_rate": 2.928571428571429e-06, | |
| "loss": 0.1012, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.374885426214482, | |
| "grad_norm": 1.522476315498352, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 0.0978, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.374885426214482, | |
| "eval_loss": 0.25593996047973633, | |
| "eval_runtime": 2144.4647, | |
| "eval_samples_per_second": 2.509, | |
| "eval_steps_per_second": 0.157, | |
| "eval_wer": 0.19509722056543002, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.3863428047662696, | |
| "grad_norm": 1.7234809398651123, | |
| "learning_rate": 2.785714285714286e-06, | |
| "loss": 0.1026, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.3978001833180569, | |
| "grad_norm": 1.527714490890503, | |
| "learning_rate": 2.7142857142857144e-06, | |
| "loss": 0.1024, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.4092575618698442, | |
| "grad_norm": 1.8621338605880737, | |
| "learning_rate": 2.642857142857143e-06, | |
| "loss": 0.1087, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.4207149404216315, | |
| "grad_norm": 1.4294075965881348, | |
| "learning_rate": 2.571428571428571e-06, | |
| "loss": 0.0975, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.4321723189734188, | |
| "grad_norm": 1.9539209604263306, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0971, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.4436296975252063, | |
| "grad_norm": 1.8456083536148071, | |
| "learning_rate": 2.428571428571429e-06, | |
| "loss": 0.1057, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.4550870760769936, | |
| "grad_norm": 1.2178508043289185, | |
| "learning_rate": 2.3571428571428574e-06, | |
| "loss": 0.0884, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.466544454628781, | |
| "grad_norm": 1.3775935173034668, | |
| "learning_rate": 2.285714285714286e-06, | |
| "loss": 0.1056, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.4780018331805682, | |
| "grad_norm": 1.5138427019119263, | |
| "learning_rate": 2.2142857142857146e-06, | |
| "loss": 0.1042, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.4894592117323557, | |
| "grad_norm": 1.603317141532898, | |
| "learning_rate": 2.1428571428571427e-06, | |
| "loss": 0.0981, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.500916590284143, | |
| "grad_norm": 1.2418969869613647, | |
| "learning_rate": 2.0714285714285717e-06, | |
| "loss": 0.0986, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.5123739688359303, | |
| "grad_norm": 1.4893618822097778, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1015, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.5238313473877176, | |
| "grad_norm": 1.6855746507644653, | |
| "learning_rate": 1.928571428571429e-06, | |
| "loss": 0.1074, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.535288725939505, | |
| "grad_norm": 1.8536219596862793, | |
| "learning_rate": 1.8571428571428573e-06, | |
| "loss": 0.1139, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.5467461044912922, | |
| "grad_norm": 1.944806456565857, | |
| "learning_rate": 1.7857142857142859e-06, | |
| "loss": 0.1083, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.5582034830430798, | |
| "grad_norm": 1.5871174335479736, | |
| "learning_rate": 1.7142857142857145e-06, | |
| "loss": 0.097, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.569660861594867, | |
| "grad_norm": 1.466895580291748, | |
| "learning_rate": 1.642857142857143e-06, | |
| "loss": 0.0901, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.5811182401466546, | |
| "grad_norm": 1.374851942062378, | |
| "learning_rate": 1.5714285714285714e-06, | |
| "loss": 0.0989, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.5925756186984419, | |
| "grad_norm": 1.267625093460083, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.1102, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.6040329972502292, | |
| "grad_norm": 1.3605536222457886, | |
| "learning_rate": 1.4285714285714286e-06, | |
| "loss": 0.0922, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.6154903758020165, | |
| "grad_norm": 1.5604383945465088, | |
| "learning_rate": 1.3571428571428572e-06, | |
| "loss": 0.0935, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.6269477543538038, | |
| "grad_norm": 1.552208423614502, | |
| "learning_rate": 1.2857142857142856e-06, | |
| "loss": 0.0976, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.638405132905591, | |
| "grad_norm": 1.723597526550293, | |
| "learning_rate": 1.2142857142857144e-06, | |
| "loss": 0.0903, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.6498625114573784, | |
| "grad_norm": 1.799728512763977, | |
| "learning_rate": 1.142857142857143e-06, | |
| "loss": 0.1024, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.661319890009166, | |
| "grad_norm": 1.9864362478256226, | |
| "learning_rate": 1.0714285714285714e-06, | |
| "loss": 0.1041, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.6727772685609532, | |
| "grad_norm": 1.837119460105896, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.1042, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.6842346471127407, | |
| "grad_norm": 1.6304919719696045, | |
| "learning_rate": 9.285714285714287e-07, | |
| "loss": 0.0969, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.695692025664528, | |
| "grad_norm": 1.7215403318405151, | |
| "learning_rate": 8.571428571428572e-07, | |
| "loss": 0.0958, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.7071494042163153, | |
| "grad_norm": 1.7634841203689575, | |
| "learning_rate": 7.857142857142857e-07, | |
| "loss": 0.1047, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.7186067827681026, | |
| "grad_norm": 1.5024609565734863, | |
| "learning_rate": 7.142857142857143e-07, | |
| "loss": 0.0857, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.73006416131989, | |
| "grad_norm": 1.2845063209533691, | |
| "learning_rate": 6.428571428571428e-07, | |
| "loss": 0.0883, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.7415215398716772, | |
| "grad_norm": 1.503865361213684, | |
| "learning_rate": 5.714285714285715e-07, | |
| "loss": 0.093, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.7529789184234648, | |
| "grad_norm": 1.4570015668869019, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0984, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.764436296975252, | |
| "grad_norm": 1.2372715473175049, | |
| "learning_rate": 4.285714285714286e-07, | |
| "loss": 0.0959, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.7758936755270394, | |
| "grad_norm": 1.4279190301895142, | |
| "learning_rate": 3.5714285714285716e-07, | |
| "loss": 0.0978, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.7873510540788269, | |
| "grad_norm": 2.094740629196167, | |
| "learning_rate": 2.8571428571428575e-07, | |
| "loss": 0.0989, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.7988084326306142, | |
| "grad_norm": 1.5156452655792236, | |
| "learning_rate": 2.142857142857143e-07, | |
| "loss": 0.0983, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.8102658111824015, | |
| "grad_norm": 1.3175643682479858, | |
| "learning_rate": 1.4285714285714287e-07, | |
| "loss": 0.1031, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.8217231897341888, | |
| "grad_norm": 1.2365864515304565, | |
| "learning_rate": 7.142857142857144e-08, | |
| "loss": 0.096, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.833180568285976, | |
| "grad_norm": 1.4035123586654663, | |
| "learning_rate": 0.0, | |
| "loss": 0.0902, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.833180568285976, | |
| "eval_loss": 0.25563082098960876, | |
| "eval_runtime": 2173.621, | |
| "eval_samples_per_second": 2.476, | |
| "eval_steps_per_second": 0.155, | |
| "eval_wer": 0.19340729253648256, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.833180568285976, | |
| "step": 4000, | |
| "total_flos": 4.348390132600013e+20, | |
| "train_loss": 0.1730844228863716, | |
| "train_runtime": 33932.8776, | |
| "train_samples_per_second": 3.772, | |
| "train_steps_per_second": 0.118 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 4000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.348390132600013e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |