{ "best_metric": 29.92358146984869, "best_model_checkpoint": "./checkpoint-8500", "epoch": 29.159519725557463, "eval_steps": 500, "global_step": 8500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.09, "learning_rate": 5e-08, "loss": 0.2499, "step": 25 }, { "epoch": 0.17, "learning_rate": 1e-07, "loss": 0.2433, "step": 50 }, { "epoch": 0.26, "learning_rate": 1.5e-07, "loss": 0.2512, "step": 75 }, { "epoch": 0.34, "learning_rate": 2e-07, "loss": 0.2503, "step": 100 }, { "epoch": 0.43, "learning_rate": 2.5e-07, "loss": 0.2498, "step": 125 }, { "epoch": 0.51, "learning_rate": 3e-07, "loss": 0.2501, "step": 150 }, { "epoch": 0.6, "learning_rate": 3.5e-07, "loss": 0.2468, "step": 175 }, { "epoch": 0.69, "learning_rate": 4e-07, "loss": 0.2462, "step": 200 }, { "epoch": 0.77, "learning_rate": 4.5e-07, "loss": 0.2495, "step": 225 }, { "epoch": 0.86, "learning_rate": 5e-07, "loss": 0.2472, "step": 250 }, { "epoch": 0.94, "learning_rate": 5.5e-07, "loss": 0.2449, "step": 275 }, { "epoch": 1.03, "learning_rate": 6e-07, "loss": 0.2475, "step": 300 }, { "epoch": 1.11, "learning_rate": 6.5e-07, "loss": 0.2452, "step": 325 }, { "epoch": 1.2, "learning_rate": 7e-07, "loss": 0.2441, "step": 350 }, { "epoch": 1.29, "learning_rate": 7.5e-07, "loss": 0.2422, "step": 375 }, { "epoch": 1.37, "learning_rate": 8e-07, "loss": 0.2422, "step": 400 }, { "epoch": 1.46, "learning_rate": 8.499999999999999e-07, "loss": 0.2445, "step": 425 }, { "epoch": 1.54, "learning_rate": 9e-07, "loss": 0.2449, "step": 450 }, { "epoch": 1.63, "learning_rate": 9.499999999999999e-07, "loss": 0.2442, "step": 475 }, { "epoch": 1.72, "learning_rate": 1e-06, "loss": 0.2476, "step": 500 }, { "epoch": 1.72, "eval_loss": 0.2750862240791321, "eval_runtime": 1902.3234, "eval_samples_per_second": 4.896, "eval_steps_per_second": 0.153, "eval_wer": 36.1694988054012, "step": 500 }, { "epoch": 1.8, "learning_rate": 9.968749999999999e-07, "loss": 0.2424, "step": 525 }, { "epoch": 1.89, "learning_rate": 9.9375e-07, "loss": 0.2374, "step": 550 }, { "epoch": 1.97, "learning_rate": 9.90625e-07, "loss": 0.2401, "step": 575 }, { "epoch": 2.06, "learning_rate": 9.875e-07, "loss": 0.2373, "step": 600 }, { "epoch": 2.14, "learning_rate": 9.84375e-07, "loss": 0.2344, "step": 625 }, { "epoch": 2.23, "learning_rate": 9.8125e-07, "loss": 0.2384, "step": 650 }, { "epoch": 2.32, "learning_rate": 9.78125e-07, "loss": 0.2382, "step": 675 }, { "epoch": 2.4, "learning_rate": 9.75e-07, "loss": 0.2404, "step": 700 }, { "epoch": 2.49, "learning_rate": 9.71875e-07, "loss": 0.2312, "step": 725 }, { "epoch": 2.57, "learning_rate": 9.6875e-07, "loss": 0.2337, "step": 750 }, { "epoch": 2.66, "learning_rate": 9.65625e-07, "loss": 0.2314, "step": 775 }, { "epoch": 2.74, "learning_rate": 9.624999999999999e-07, "loss": 0.2262, "step": 800 }, { "epoch": 2.83, "learning_rate": 9.59375e-07, "loss": 0.2342, "step": 825 }, { "epoch": 2.92, "learning_rate": 9.5625e-07, "loss": 0.2358, "step": 850 }, { "epoch": 3.0, "learning_rate": 9.53125e-07, "loss": 0.2316, "step": 875 }, { "epoch": 3.09, "learning_rate": 9.499999999999999e-07, "loss": 0.2344, "step": 900 }, { "epoch": 3.17, "learning_rate": 9.468749999999999e-07, "loss": 0.2224, "step": 925 }, { "epoch": 3.26, "learning_rate": 9.4375e-07, "loss": 0.2219, "step": 950 }, { "epoch": 3.34, "learning_rate": 9.40625e-07, "loss": 0.2302, "step": 975 }, { "epoch": 3.43, "learning_rate": 9.374999999999999e-07, "loss": 0.2284, "step": 1000 }, { "epoch": 3.43, "eval_loss": 0.26221010088920593, "eval_runtime": 1890.4325, "eval_samples_per_second": 4.927, "eval_steps_per_second": 0.154, "eval_wer": 35.16684115862662, "step": 1000 }, { "epoch": 3.52, "learning_rate": 9.34375e-07, "loss": 0.2283, "step": 1025 }, { "epoch": 3.6, "learning_rate": 9.3125e-07, "loss": 0.2219, "step": 1050 }, { "epoch": 3.69, "learning_rate": 9.281249999999999e-07, "loss": 0.2241, "step": 1075 }, { "epoch": 3.77, "learning_rate": 9.25e-07, "loss": 0.2259, "step": 1100 }, { "epoch": 3.86, "learning_rate": 9.21875e-07, "loss": 0.2214, "step": 1125 }, { "epoch": 3.95, "learning_rate": 9.187499999999999e-07, "loss": 0.2217, "step": 1150 }, { "epoch": 4.03, "learning_rate": 9.15625e-07, "loss": 0.2208, "step": 1175 }, { "epoch": 4.12, "learning_rate": 9.124999999999999e-07, "loss": 0.2191, "step": 1200 }, { "epoch": 4.2, "learning_rate": 9.09375e-07, "loss": 0.2163, "step": 1225 }, { "epoch": 4.29, "learning_rate": 9.0625e-07, "loss": 0.2222, "step": 1250 }, { "epoch": 4.37, "learning_rate": 9.031249999999999e-07, "loss": 0.2178, "step": 1275 }, { "epoch": 4.46, "learning_rate": 9e-07, "loss": 0.2179, "step": 1300 }, { "epoch": 4.55, "learning_rate": 8.96875e-07, "loss": 0.2214, "step": 1325 }, { "epoch": 4.63, "learning_rate": 8.9375e-07, "loss": 0.2147, "step": 1350 }, { "epoch": 4.72, "learning_rate": 8.906249999999999e-07, "loss": 0.217, "step": 1375 }, { "epoch": 4.8, "learning_rate": 8.874999999999999e-07, "loss": 0.2167, "step": 1400 }, { "epoch": 4.89, "learning_rate": 8.84375e-07, "loss": 0.2188, "step": 1425 }, { "epoch": 4.97, "learning_rate": 8.812499999999999e-07, "loss": 0.2154, "step": 1450 }, { "epoch": 5.06, "learning_rate": 8.78125e-07, "loss": 0.2173, "step": 1475 }, { "epoch": 5.15, "learning_rate": 8.75e-07, "loss": 0.213, "step": 1500 }, { "epoch": 5.15, "eval_loss": 0.252381831407547, "eval_runtime": 1875.2145, "eval_samples_per_second": 4.967, "eval_steps_per_second": 0.156, "eval_wer": 34.202213810814925, "step": 1500 }, { "epoch": 5.23, "learning_rate": 8.718749999999999e-07, "loss": 0.2121, "step": 1525 }, { "epoch": 5.32, "learning_rate": 8.687499999999999e-07, "loss": 0.2112, "step": 1550 }, { "epoch": 5.4, "learning_rate": 8.65625e-07, "loss": 0.2109, "step": 1575 }, { "epoch": 5.49, "learning_rate": 8.625e-07, "loss": 0.2106, "step": 1600 }, { "epoch": 5.57, "learning_rate": 8.593749999999999e-07, "loss": 0.2081, "step": 1625 }, { "epoch": 5.66, "learning_rate": 8.5625e-07, "loss": 0.2122, "step": 1650 }, { "epoch": 5.75, "learning_rate": 8.53125e-07, "loss": 0.211, "step": 1675 }, { "epoch": 5.83, "learning_rate": 8.499999999999999e-07, "loss": 0.2072, "step": 1700 }, { "epoch": 5.92, "learning_rate": 8.46875e-07, "loss": 0.2063, "step": 1725 }, { "epoch": 6.0, "learning_rate": 8.4375e-07, "loss": 0.2137, "step": 1750 }, { "epoch": 6.09, "learning_rate": 8.406249999999999e-07, "loss": 0.2095, "step": 1775 }, { "epoch": 6.17, "learning_rate": 8.375e-07, "loss": 0.2096, "step": 1800 }, { "epoch": 6.26, "learning_rate": 8.34375e-07, "loss": 0.2087, "step": 1825 }, { "epoch": 6.35, "learning_rate": 8.3125e-07, "loss": 0.2021, "step": 1850 }, { "epoch": 6.43, "learning_rate": 8.28125e-07, "loss": 0.2076, "step": 1875 }, { "epoch": 6.52, "learning_rate": 8.249999999999999e-07, "loss": 0.2024, "step": 1900 }, { "epoch": 6.6, "learning_rate": 8.21875e-07, "loss": 0.2013, "step": 1925 }, { "epoch": 6.69, "learning_rate": 8.187499999999999e-07, "loss": 0.2043, "step": 1950 }, { "epoch": 6.78, "learning_rate": 8.15625e-07, "loss": 0.1991, "step": 1975 }, { "epoch": 6.86, "learning_rate": 8.125e-07, "loss": 0.2048, "step": 2000 }, { "epoch": 6.86, "eval_loss": 0.2446550577878952, "eval_runtime": 1880.3132, "eval_samples_per_second": 4.953, "eval_steps_per_second": 0.155, "eval_wer": 33.52661673512121, "step": 2000 }, { "epoch": 6.95, "learning_rate": 8.093749999999999e-07, "loss": 0.2022, "step": 2025 }, { "epoch": 7.03, "learning_rate": 8.0625e-07, "loss": 0.208, "step": 2050 }, { "epoch": 7.12, "learning_rate": 8.031249999999999e-07, "loss": 0.2016, "step": 2075 }, { "epoch": 7.2, "learning_rate": 8e-07, "loss": 0.1955, "step": 2100 }, { "epoch": 7.29, "learning_rate": 7.96875e-07, "loss": 0.2013, "step": 2125 }, { "epoch": 7.38, "learning_rate": 7.937499999999999e-07, "loss": 0.1993, "step": 2150 }, { "epoch": 7.46, "learning_rate": 7.90625e-07, "loss": 0.201, "step": 2175 }, { "epoch": 7.55, "learning_rate": 7.875e-07, "loss": 0.1976, "step": 2200 }, { "epoch": 7.63, "learning_rate": 7.84375e-07, "loss": 0.194, "step": 2225 }, { "epoch": 7.72, "learning_rate": 7.812499999999999e-07, "loss": 0.1999, "step": 2250 }, { "epoch": 7.8, "learning_rate": 7.78125e-07, "loss": 0.1991, "step": 2275 }, { "epoch": 7.89, "learning_rate": 7.75e-07, "loss": 0.2026, "step": 2300 }, { "epoch": 7.98, "learning_rate": 7.718749999999999e-07, "loss": 0.1987, "step": 2325 }, { "epoch": 8.06, "learning_rate": 7.688749999999999e-07, "loss": 0.1996, "step": 2350 }, { "epoch": 8.15, "learning_rate": 7.6575e-07, "loss": 0.1903, "step": 2375 }, { "epoch": 8.23, "learning_rate": 7.62625e-07, "loss": 0.1945, "step": 2400 }, { "epoch": 8.32, "learning_rate": 7.594999999999999e-07, "loss": 0.1972, "step": 2425 }, { "epoch": 8.4, "learning_rate": 7.56375e-07, "loss": 0.1971, "step": 2450 }, { "epoch": 8.49, "learning_rate": 7.532499999999999e-07, "loss": 0.1988, "step": 2475 }, { "epoch": 8.58, "learning_rate": 7.50125e-07, "loss": 0.1948, "step": 2500 }, { "epoch": 8.58, "eval_loss": 0.2382451444864273, "eval_runtime": 1871.3425, "eval_samples_per_second": 4.977, "eval_steps_per_second": 0.156, "eval_wer": 32.74945639043247, "step": 2500 }, { "epoch": 8.66, "learning_rate": 7.47e-07, "loss": 0.1951, "step": 2525 }, { "epoch": 8.75, "learning_rate": 7.438749999999999e-07, "loss": 0.1897, "step": 2550 }, { "epoch": 8.83, "learning_rate": 7.4075e-07, "loss": 0.1887, "step": 2575 }, { "epoch": 8.92, "learning_rate": 7.37625e-07, "loss": 0.1966, "step": 2600 }, { "epoch": 9.01, "learning_rate": 7.345e-07, "loss": 0.1936, "step": 2625 }, { "epoch": 9.09, "learning_rate": 7.313749999999999e-07, "loss": 0.1939, "step": 2650 }, { "epoch": 9.18, "learning_rate": 7.282499999999999e-07, "loss": 0.1931, "step": 2675 }, { "epoch": 9.26, "learning_rate": 7.25125e-07, "loss": 0.1903, "step": 2700 }, { "epoch": 9.35, "learning_rate": 7.219999999999999e-07, "loss": 0.1852, "step": 2725 }, { "epoch": 9.43, "learning_rate": 7.18875e-07, "loss": 0.1916, "step": 2750 }, { "epoch": 9.52, "learning_rate": 7.1575e-07, "loss": 0.1882, "step": 2775 }, { "epoch": 9.61, "learning_rate": 7.126249999999999e-07, "loss": 0.1872, "step": 2800 }, { "epoch": 9.69, "learning_rate": 7.094999999999999e-07, "loss": 0.1915, "step": 2825 }, { "epoch": 9.78, "learning_rate": 7.06375e-07, "loss": 0.191, "step": 2850 }, { "epoch": 9.86, "learning_rate": 7.0325e-07, "loss": 0.1873, "step": 2875 }, { "epoch": 9.95, "learning_rate": 7.001249999999999e-07, "loss": 0.192, "step": 2900 }, { "epoch": 10.03, "learning_rate": 6.97e-07, "loss": 0.1874, "step": 2925 }, { "epoch": 10.12, "learning_rate": 6.93875e-07, "loss": 0.1803, "step": 2950 }, { "epoch": 10.21, "learning_rate": 6.907499999999999e-07, "loss": 0.1874, "step": 2975 }, { "epoch": 10.29, "learning_rate": 6.87625e-07, "loss": 0.1852, "step": 3000 }, { "epoch": 10.29, "eval_loss": 0.2334369719028473, "eval_runtime": 1876.9245, "eval_samples_per_second": 4.962, "eval_steps_per_second": 0.156, "eval_wer": 32.23224432453715, "step": 3000 }, { "epoch": 10.38, "learning_rate": 6.845e-07, "loss": 0.1835, "step": 3025 }, { "epoch": 10.46, "learning_rate": 6.813749999999999e-07, "loss": 0.1887, "step": 3050 }, { "epoch": 10.55, "learning_rate": 6.7825e-07, "loss": 0.1837, "step": 3075 }, { "epoch": 10.63, "learning_rate": 6.75125e-07, "loss": 0.1831, "step": 3100 }, { "epoch": 10.72, "learning_rate": 6.72e-07, "loss": 0.1879, "step": 3125 }, { "epoch": 10.81, "learning_rate": 6.68875e-07, "loss": 0.1852, "step": 3150 }, { "epoch": 10.89, "learning_rate": 6.657499999999999e-07, "loss": 0.1908, "step": 3175 }, { "epoch": 10.98, "learning_rate": 6.62625e-07, "loss": 0.1892, "step": 3200 }, { "epoch": 11.06, "learning_rate": 6.595e-07, "loss": 0.1819, "step": 3225 }, { "epoch": 11.15, "learning_rate": 6.56375e-07, "loss": 0.1827, "step": 3250 }, { "epoch": 11.23, "learning_rate": 6.5325e-07, "loss": 0.1814, "step": 3275 }, { "epoch": 11.32, "learning_rate": 6.501249999999999e-07, "loss": 0.1789, "step": 3300 }, { "epoch": 11.41, "learning_rate": 6.47e-07, "loss": 0.1778, "step": 3325 }, { "epoch": 11.49, "learning_rate": 6.438749999999999e-07, "loss": 0.1814, "step": 3350 }, { "epoch": 11.58, "learning_rate": 6.4075e-07, "loss": 0.1875, "step": 3375 }, { "epoch": 11.66, "learning_rate": 6.37625e-07, "loss": 0.1886, "step": 3400 }, { "epoch": 11.75, "learning_rate": 6.344999999999999e-07, "loss": 0.1816, "step": 3425 }, { "epoch": 11.84, "learning_rate": 6.31375e-07, "loss": 0.1822, "step": 3450 }, { "epoch": 11.92, "learning_rate": 6.2825e-07, "loss": 0.1798, "step": 3475 }, { "epoch": 12.01, "learning_rate": 6.25125e-07, "loss": 0.1789, "step": 3500 }, { "epoch": 12.01, "eval_loss": 0.22954809665679932, "eval_runtime": 1876.8523, "eval_samples_per_second": 4.963, "eval_steps_per_second": 0.156, "eval_wer": 31.72442797956207, "step": 3500 }, { "epoch": 12.09, "learning_rate": 6.219999999999999e-07, "loss": 0.1792, "step": 3525 }, { "epoch": 12.18, "learning_rate": 6.18875e-07, "loss": 0.1805, "step": 3550 }, { "epoch": 12.26, "learning_rate": 6.1575e-07, "loss": 0.1803, "step": 3575 }, { "epoch": 12.35, "learning_rate": 6.126249999999999e-07, "loss": 0.1808, "step": 3600 }, { "epoch": 12.44, "learning_rate": 6.095e-07, "loss": 0.1789, "step": 3625 }, { "epoch": 12.52, "learning_rate": 6.06375e-07, "loss": 0.178, "step": 3650 }, { "epoch": 12.61, "learning_rate": 6.032499999999999e-07, "loss": 0.1808, "step": 3675 }, { "epoch": 12.69, "learning_rate": 6.00125e-07, "loss": 0.1766, "step": 3700 }, { "epoch": 12.78, "learning_rate": 5.97e-07, "loss": 0.176, "step": 3725 }, { "epoch": 12.86, "learning_rate": 5.93875e-07, "loss": 0.1804, "step": 3750 }, { "epoch": 12.95, "learning_rate": 5.9075e-07, "loss": 0.1758, "step": 3775 }, { "epoch": 13.04, "learning_rate": 5.876249999999999e-07, "loss": 0.1754, "step": 3800 }, { "epoch": 13.12, "learning_rate": 5.845e-07, "loss": 0.1783, "step": 3825 }, { "epoch": 13.21, "learning_rate": 5.813749999999999e-07, "loss": 0.1763, "step": 3850 }, { "epoch": 13.29, "learning_rate": 5.7825e-07, "loss": 0.1768, "step": 3875 }, { "epoch": 13.38, "learning_rate": 5.75125e-07, "loss": 0.1724, "step": 3900 }, { "epoch": 13.46, "learning_rate": 5.719999999999999e-07, "loss": 0.1776, "step": 3925 }, { "epoch": 13.55, "learning_rate": 5.68875e-07, "loss": 0.1742, "step": 3950 }, { "epoch": 13.64, "learning_rate": 5.657499999999999e-07, "loss": 0.1732, "step": 3975 }, { "epoch": 13.72, "learning_rate": 5.62625e-07, "loss": 0.1738, "step": 4000 }, { "epoch": 13.72, "eval_loss": 0.2260207086801529, "eval_runtime": 1879.4952, "eval_samples_per_second": 4.956, "eval_steps_per_second": 0.155, "eval_wer": 31.23406083058173, "step": 4000 }, { "epoch": 13.81, "learning_rate": 5.595e-07, "loss": 0.1756, "step": 4025 }, { "epoch": 13.89, "learning_rate": 5.563749999999999e-07, "loss": 0.1778, "step": 4050 }, { "epoch": 13.98, "learning_rate": 5.5325e-07, "loss": 0.1756, "step": 4075 }, { "epoch": 14.07, "learning_rate": 5.50125e-07, "loss": 0.1748, "step": 4100 }, { "epoch": 14.15, "learning_rate": 5.47e-07, "loss": 0.1738, "step": 4125 }, { "epoch": 14.24, "learning_rate": 5.438749999999999e-07, "loss": 0.1706, "step": 4150 }, { "epoch": 14.32, "learning_rate": 5.4075e-07, "loss": 0.1764, "step": 4175 }, { "epoch": 14.41, "learning_rate": 5.37625e-07, "loss": 0.1734, "step": 4200 }, { "epoch": 14.49, "learning_rate": 5.344999999999999e-07, "loss": 0.1697, "step": 4225 }, { "epoch": 14.58, "learning_rate": 5.313750000000001e-07, "loss": 0.178, "step": 4250 }, { "epoch": 14.67, "learning_rate": 5.2825e-07, "loss": 0.1714, "step": 4275 }, { "epoch": 14.75, "learning_rate": 5.251249999999999e-07, "loss": 0.172, "step": 4300 }, { "epoch": 14.84, "learning_rate": 5.22e-07, "loss": 0.1714, "step": 4325 }, { "epoch": 14.92, "learning_rate": 5.18875e-07, "loss": 0.1701, "step": 4350 }, { "epoch": 15.01, "learning_rate": 5.15875e-07, "loss": 0.1714, "step": 4375 }, { "epoch": 15.09, "learning_rate": 5.1275e-07, "loss": 0.17, "step": 4400 }, { "epoch": 15.18, "learning_rate": 5.09625e-07, "loss": 0.1709, "step": 4425 }, { "epoch": 15.27, "learning_rate": 5.064999999999999e-07, "loss": 0.1742, "step": 4450 }, { "epoch": 15.35, "learning_rate": 5.03375e-07, "loss": 0.1708, "step": 4475 }, { "epoch": 15.44, "learning_rate": 5.0025e-07, "loss": 0.166, "step": 4500 }, { "epoch": 15.44, "eval_loss": 0.2236023247241974, "eval_runtime": 1888.9296, "eval_samples_per_second": 4.931, "eval_steps_per_second": 0.155, "eval_wer": 30.956215940511665, "step": 4500 }, { "epoch": 15.52, "learning_rate": 4.971249999999999e-07, "loss": 0.1727, "step": 4525 }, { "epoch": 15.61, "learning_rate": 4.94e-07, "loss": 0.1689, "step": 4550 }, { "epoch": 15.69, "learning_rate": 4.90875e-07, "loss": 0.1686, "step": 4575 }, { "epoch": 15.78, "learning_rate": 4.8775e-07, "loss": 0.1733, "step": 4600 }, { "epoch": 15.87, "learning_rate": 4.846249999999999e-07, "loss": 0.1704, "step": 4625 }, { "epoch": 15.95, "learning_rate": 4.815e-07, "loss": 0.1681, "step": 4650 }, { "epoch": 16.04, "learning_rate": 4.78375e-07, "loss": 0.1659, "step": 4675 }, { "epoch": 16.12, "learning_rate": 4.7524999999999997e-07, "loss": 0.1722, "step": 4700 }, { "epoch": 16.21, "learning_rate": 4.72125e-07, "loss": 0.1662, "step": 4725 }, { "epoch": 16.3, "learning_rate": 4.689999999999999e-07, "loss": 0.1679, "step": 4750 }, { "epoch": 16.38, "learning_rate": 4.6587499999999995e-07, "loss": 0.1675, "step": 4775 }, { "epoch": 16.47, "learning_rate": 4.6275e-07, "loss": 0.1672, "step": 4800 }, { "epoch": 16.55, "learning_rate": 4.5962499999999996e-07, "loss": 0.1671, "step": 4825 }, { "epoch": 16.64, "learning_rate": 4.565e-07, "loss": 0.1686, "step": 4850 }, { "epoch": 16.72, "learning_rate": 4.5337499999999997e-07, "loss": 0.1732, "step": 4875 }, { "epoch": 16.81, "learning_rate": 4.5024999999999995e-07, "loss": 0.1645, "step": 4900 }, { "epoch": 16.9, "learning_rate": 4.47125e-07, "loss": 0.1688, "step": 4925 }, { "epoch": 16.98, "learning_rate": 4.44e-07, "loss": 0.1627, "step": 4950 }, { "epoch": 17.07, "learning_rate": 4.40875e-07, "loss": 0.1642, "step": 4975 }, { "epoch": 17.15, "learning_rate": 4.3774999999999996e-07, "loss": 0.1629, "step": 5000 }, { "epoch": 17.15, "eval_loss": 0.2214428037405014, "eval_runtime": 1901.3668, "eval_samples_per_second": 4.899, "eval_steps_per_second": 0.154, "eval_wer": 30.81706978783567, "step": 5000 }, { "epoch": 17.24, "learning_rate": 4.34625e-07, "loss": 0.1681, "step": 5025 }, { "epoch": 17.32, "learning_rate": 4.3149999999999997e-07, "loss": 0.1691, "step": 5050 }, { "epoch": 17.41, "learning_rate": 4.28375e-07, "loss": 0.1663, "step": 5075 }, { "epoch": 17.5, "learning_rate": 4.2524999999999997e-07, "loss": 0.1643, "step": 5100 }, { "epoch": 17.58, "learning_rate": 4.2212499999999995e-07, "loss": 0.1619, "step": 5125 }, { "epoch": 17.67, "learning_rate": 4.19e-07, "loss": 0.1639, "step": 5150 }, { "epoch": 17.75, "learning_rate": 4.1587499999999996e-07, "loss": 0.1662, "step": 5175 }, { "epoch": 17.84, "learning_rate": 4.1275e-07, "loss": 0.1692, "step": 5200 }, { "epoch": 17.92, "learning_rate": 4.09625e-07, "loss": 0.1642, "step": 5225 }, { "epoch": 18.01, "learning_rate": 4.0649999999999994e-07, "loss": 0.1653, "step": 5250 }, { "epoch": 18.1, "learning_rate": 4.0337499999999997e-07, "loss": 0.1674, "step": 5275 }, { "epoch": 18.18, "learning_rate": 4.0025e-07, "loss": 0.1631, "step": 5300 }, { "epoch": 18.27, "learning_rate": 3.97125e-07, "loss": 0.1627, "step": 5325 }, { "epoch": 18.35, "learning_rate": 3.94e-07, "loss": 0.1633, "step": 5350 }, { "epoch": 18.44, "learning_rate": 3.9087499999999993e-07, "loss": 0.1612, "step": 5375 }, { "epoch": 18.52, "learning_rate": 3.8774999999999996e-07, "loss": 0.1591, "step": 5400 }, { "epoch": 18.61, "learning_rate": 3.84625e-07, "loss": 0.166, "step": 5425 }, { "epoch": 18.7, "learning_rate": 3.8149999999999997e-07, "loss": 0.1656, "step": 5450 }, { "epoch": 18.78, "learning_rate": 3.78375e-07, "loss": 0.1649, "step": 5475 }, { "epoch": 18.87, "learning_rate": 3.7525e-07, "loss": 0.1636, "step": 5500 }, { "epoch": 18.87, "eval_loss": 0.21942444145679474, "eval_runtime": 1897.9841, "eval_samples_per_second": 4.907, "eval_steps_per_second": 0.154, "eval_wer": 30.43676679820676, "step": 5500 }, { "epoch": 18.95, "learning_rate": 3.7212499999999995e-07, "loss": 0.1647, "step": 5525 }, { "epoch": 19.04, "learning_rate": 3.69e-07, "loss": 0.1601, "step": 5550 }, { "epoch": 19.13, "learning_rate": 3.65875e-07, "loss": 0.1658, "step": 5575 }, { "epoch": 19.21, "learning_rate": 3.6275e-07, "loss": 0.1625, "step": 5600 }, { "epoch": 19.3, "learning_rate": 3.5962499999999997e-07, "loss": 0.1625, "step": 5625 }, { "epoch": 19.38, "learning_rate": 3.5649999999999994e-07, "loss": 0.1634, "step": 5650 }, { "epoch": 19.47, "learning_rate": 3.5337499999999997e-07, "loss": 0.1606, "step": 5675 }, { "epoch": 19.55, "learning_rate": 3.5025e-07, "loss": 0.1659, "step": 5700 }, { "epoch": 19.64, "learning_rate": 3.47125e-07, "loss": 0.163, "step": 5725 }, { "epoch": 19.73, "learning_rate": 3.4399999999999996e-07, "loss": 0.1614, "step": 5750 }, { "epoch": 19.81, "learning_rate": 3.40875e-07, "loss": 0.1583, "step": 5775 }, { "epoch": 19.9, "learning_rate": 3.3774999999999996e-07, "loss": 0.157, "step": 5800 }, { "epoch": 19.98, "learning_rate": 3.34625e-07, "loss": 0.1601, "step": 5825 }, { "epoch": 20.07, "learning_rate": 3.315e-07, "loss": 0.1654, "step": 5850 }, { "epoch": 20.15, "learning_rate": 3.2837499999999995e-07, "loss": 0.1626, "step": 5875 }, { "epoch": 20.24, "learning_rate": 3.2525e-07, "loss": 0.1648, "step": 5900 }, { "epoch": 20.33, "learning_rate": 3.2212499999999995e-07, "loss": 0.1621, "step": 5925 }, { "epoch": 20.41, "learning_rate": 3.19e-07, "loss": 0.1621, "step": 5950 }, { "epoch": 20.5, "learning_rate": 3.15875e-07, "loss": 0.159, "step": 5975 }, { "epoch": 20.58, "learning_rate": 3.1274999999999994e-07, "loss": 0.1578, "step": 6000 }, { "epoch": 20.58, "eval_loss": 0.21812666952610016, "eval_runtime": 1902.7257, "eval_samples_per_second": 4.895, "eval_steps_per_second": 0.153, "eval_wer": 30.251984286775297, "step": 6000 }, { "epoch": 20.67, "learning_rate": 3.0962499999999997e-07, "loss": 0.1593, "step": 6025 }, { "epoch": 20.75, "learning_rate": 3.065e-07, "loss": 0.158, "step": 6050 }, { "epoch": 20.84, "learning_rate": 3.03375e-07, "loss": 0.1573, "step": 6075 }, { "epoch": 20.93, "learning_rate": 3.0025e-07, "loss": 0.1623, "step": 6100 }, { "epoch": 21.01, "learning_rate": 2.97125e-07, "loss": 0.1556, "step": 6125 }, { "epoch": 21.1, "learning_rate": 2.9399999999999996e-07, "loss": 0.1574, "step": 6150 }, { "epoch": 21.18, "learning_rate": 2.90875e-07, "loss": 0.1611, "step": 6175 }, { "epoch": 21.27, "learning_rate": 2.8775e-07, "loss": 0.1555, "step": 6200 }, { "epoch": 21.36, "learning_rate": 2.84625e-07, "loss": 0.1558, "step": 6225 }, { "epoch": 21.44, "learning_rate": 2.8149999999999997e-07, "loss": 0.1598, "step": 6250 }, { "epoch": 21.53, "learning_rate": 2.7837499999999995e-07, "loss": 0.159, "step": 6275 }, { "epoch": 21.61, "learning_rate": 2.7525e-07, "loss": 0.1595, "step": 6300 }, { "epoch": 21.7, "learning_rate": 2.72125e-07, "loss": 0.1579, "step": 6325 }, { "epoch": 21.78, "learning_rate": 2.69e-07, "loss": 0.1603, "step": 6350 }, { "epoch": 21.87, "learning_rate": 2.6587499999999996e-07, "loss": 0.1621, "step": 6375 }, { "epoch": 21.96, "learning_rate": 2.6275e-07, "loss": 0.1603, "step": 6400 }, { "epoch": 22.04, "learning_rate": 2.5962499999999997e-07, "loss": 0.1609, "step": 6425 }, { "epoch": 22.13, "learning_rate": 2.565e-07, "loss": 0.1567, "step": 6450 }, { "epoch": 22.21, "learning_rate": 2.5337500000000003e-07, "loss": 0.1555, "step": 6475 }, { "epoch": 22.3, "learning_rate": 2.5024999999999995e-07, "loss": 0.1628, "step": 6500 }, { "epoch": 22.3, "eval_loss": 0.2169613540172577, "eval_runtime": 1901.0014, "eval_samples_per_second": 4.9, "eval_steps_per_second": 0.154, "eval_wer": 30.185766825051672, "step": 6500 }, { "epoch": 22.38, "learning_rate": 2.47125e-07, "loss": 0.159, "step": 6525 }, { "epoch": 22.47, "learning_rate": 2.4399999999999996e-07, "loss": 0.1545, "step": 6550 }, { "epoch": 22.56, "learning_rate": 2.40875e-07, "loss": 0.1562, "step": 6575 }, { "epoch": 22.64, "learning_rate": 2.3774999999999997e-07, "loss": 0.156, "step": 6600 }, { "epoch": 22.73, "learning_rate": 2.34625e-07, "loss": 0.1561, "step": 6625 }, { "epoch": 22.81, "learning_rate": 2.315e-07, "loss": 0.16, "step": 6650 }, { "epoch": 22.9, "learning_rate": 2.2837499999999998e-07, "loss": 0.1557, "step": 6675 }, { "epoch": 22.98, "learning_rate": 2.2525e-07, "loss": 0.1569, "step": 6700 }, { "epoch": 23.07, "learning_rate": 2.2212499999999999e-07, "loss": 0.1572, "step": 6725 }, { "epoch": 23.16, "learning_rate": 2.19e-07, "loss": 0.1592, "step": 6750 }, { "epoch": 23.24, "learning_rate": 2.15875e-07, "loss": 0.1583, "step": 6775 }, { "epoch": 23.33, "learning_rate": 2.1275e-07, "loss": 0.1581, "step": 6800 }, { "epoch": 23.41, "learning_rate": 2.09625e-07, "loss": 0.1548, "step": 6825 }, { "epoch": 23.5, "learning_rate": 2.0649999999999998e-07, "loss": 0.1534, "step": 6850 }, { "epoch": 23.58, "learning_rate": 2.0337499999999998e-07, "loss": 0.1575, "step": 6875 }, { "epoch": 23.67, "learning_rate": 2.0025e-07, "loss": 0.1576, "step": 6900 }, { "epoch": 23.76, "learning_rate": 1.9712499999999999e-07, "loss": 0.1559, "step": 6925 }, { "epoch": 23.84, "learning_rate": 1.94e-07, "loss": 0.1553, "step": 6950 }, { "epoch": 23.93, "learning_rate": 1.90875e-07, "loss": 0.1548, "step": 6975 }, { "epoch": 24.01, "learning_rate": 1.8775e-07, "loss": 0.1566, "step": 7000 }, { "epoch": 24.01, "eval_loss": 0.21614298224449158, "eval_runtime": 1900.8837, "eval_samples_per_second": 4.9, "eval_steps_per_second": 0.154, "eval_wer": 30.06943885175342, "step": 7000 }, { "epoch": 24.1, "learning_rate": 1.84625e-07, "loss": 0.1538, "step": 7025 }, { "epoch": 24.19, "learning_rate": 1.8149999999999998e-07, "loss": 0.1601, "step": 7050 }, { "epoch": 24.27, "learning_rate": 1.78375e-07, "loss": 0.1577, "step": 7075 }, { "epoch": 24.36, "learning_rate": 1.7524999999999998e-07, "loss": 0.1533, "step": 7100 }, { "epoch": 24.44, "learning_rate": 1.72125e-07, "loss": 0.1566, "step": 7125 }, { "epoch": 24.53, "learning_rate": 1.69e-07, "loss": 0.155, "step": 7150 }, { "epoch": 24.61, "learning_rate": 1.65875e-07, "loss": 0.1541, "step": 7175 }, { "epoch": 24.7, "learning_rate": 1.6275e-07, "loss": 0.1536, "step": 7200 }, { "epoch": 24.79, "learning_rate": 1.5962499999999998e-07, "loss": 0.1555, "step": 7225 }, { "epoch": 24.87, "learning_rate": 1.565e-07, "loss": 0.1552, "step": 7250 }, { "epoch": 24.96, "learning_rate": 1.53375e-07, "loss": 0.1571, "step": 7275 }, { "epoch": 25.04, "learning_rate": 1.5024999999999999e-07, "loss": 0.1535, "step": 7300 }, { "epoch": 25.13, "learning_rate": 1.47125e-07, "loss": 0.1548, "step": 7325 }, { "epoch": 25.21, "learning_rate": 1.44e-07, "loss": 0.1504, "step": 7350 }, { "epoch": 25.3, "learning_rate": 1.40875e-07, "loss": 0.1586, "step": 7375 }, { "epoch": 25.39, "learning_rate": 1.3775e-07, "loss": 0.1535, "step": 7400 }, { "epoch": 25.47, "learning_rate": 1.3462499999999998e-07, "loss": 0.1532, "step": 7425 }, { "epoch": 25.56, "learning_rate": 1.315e-07, "loss": 0.1553, "step": 7450 }, { "epoch": 25.64, "learning_rate": 1.2837499999999998e-07, "loss": 0.1562, "step": 7475 }, { "epoch": 25.73, "learning_rate": 1.2524999999999999e-07, "loss": 0.1564, "step": 7500 }, { "epoch": 25.73, "eval_loss": 0.215592622756958, "eval_runtime": 1900.3797, "eval_samples_per_second": 4.901, "eval_steps_per_second": 0.154, "eval_wer": 29.99427308439147, "step": 7500 }, { "epoch": 25.81, "learning_rate": 1.22125e-07, "loss": 0.1585, "step": 7525 }, { "epoch": 25.9, "learning_rate": 1.19e-07, "loss": 0.1532, "step": 7550 }, { "epoch": 25.99, "learning_rate": 1.15875e-07, "loss": 0.1551, "step": 7575 }, { "epoch": 26.07, "learning_rate": 1.12875e-07, "loss": 0.1599, "step": 7600 }, { "epoch": 26.16, "learning_rate": 1.0975e-07, "loss": 0.1545, "step": 7625 }, { "epoch": 26.24, "learning_rate": 1.0662499999999999e-07, "loss": 0.1543, "step": 7650 }, { "epoch": 26.33, "learning_rate": 1.035e-07, "loss": 0.1563, "step": 7675 }, { "epoch": 26.42, "learning_rate": 1.00375e-07, "loss": 0.152, "step": 7700 }, { "epoch": 26.5, "learning_rate": 9.725e-08, "loss": 0.1549, "step": 7725 }, { "epoch": 26.59, "learning_rate": 9.412499999999999e-08, "loss": 0.1559, "step": 7750 }, { "epoch": 26.67, "learning_rate": 9.1e-08, "loss": 0.1545, "step": 7775 }, { "epoch": 26.76, "learning_rate": 8.787499999999998e-08, "loss": 0.1531, "step": 7800 }, { "epoch": 26.84, "learning_rate": 8.475e-08, "loss": 0.1542, "step": 7825 }, { "epoch": 26.93, "learning_rate": 8.1625e-08, "loss": 0.1513, "step": 7850 }, { "epoch": 27.02, "learning_rate": 7.85e-08, "loss": 0.1542, "step": 7875 }, { "epoch": 27.1, "learning_rate": 7.5375e-08, "loss": 0.1529, "step": 7900 }, { "epoch": 27.19, "learning_rate": 7.224999999999999e-08, "loss": 0.1547, "step": 7925 }, { "epoch": 27.27, "learning_rate": 6.9125e-08, "loss": 0.1507, "step": 7950 }, { "epoch": 27.36, "learning_rate": 6.6e-08, "loss": 0.1542, "step": 7975 }, { "epoch": 27.44, "learning_rate": 6.2875e-08, "loss": 0.1545, "step": 8000 }, { "epoch": 27.44, "eval_loss": 0.21525579690933228, "eval_runtime": 1895.3141, "eval_samples_per_second": 4.914, "eval_steps_per_second": 0.154, "eval_wer": 29.929397868513597, "step": 8000 }, { "epoch": 27.53, "learning_rate": 5.974999999999999e-08, "loss": 0.1534, "step": 8025 }, { "epoch": 27.62, "learning_rate": 5.6625e-08, "loss": 0.15, "step": 8050 }, { "epoch": 27.7, "learning_rate": 5.3499999999999996e-08, "loss": 0.1524, "step": 8075 }, { "epoch": 27.79, "learning_rate": 5.0375e-08, "loss": 0.1538, "step": 8100 }, { "epoch": 27.87, "learning_rate": 4.7249999999999996e-08, "loss": 0.1539, "step": 8125 }, { "epoch": 27.96, "learning_rate": 4.412499999999999e-08, "loss": 0.1562, "step": 8150 }, { "epoch": 28.04, "learning_rate": 4.1e-08, "loss": 0.1569, "step": 8175 }, { "epoch": 28.13, "learning_rate": 3.7875e-08, "loss": 0.1517, "step": 8200 }, { "epoch": 28.22, "learning_rate": 3.4750000000000003e-08, "loss": 0.1527, "step": 8225 }, { "epoch": 28.3, "learning_rate": 3.1625e-08, "loss": 0.1486, "step": 8250 }, { "epoch": 28.39, "learning_rate": 2.85e-08, "loss": 0.1544, "step": 8275 }, { "epoch": 28.47, "learning_rate": 2.5375e-08, "loss": 0.1554, "step": 8300 }, { "epoch": 28.56, "learning_rate": 2.2249999999999997e-08, "loss": 0.1557, "step": 8325 }, { "epoch": 28.64, "learning_rate": 1.9124999999999998e-08, "loss": 0.1559, "step": 8350 }, { "epoch": 28.73, "learning_rate": 1.6e-08, "loss": 0.1523, "step": 8375 }, { "epoch": 28.82, "learning_rate": 1.2875e-08, "loss": 0.1539, "step": 8400 }, { "epoch": 28.9, "learning_rate": 9.75e-09, "loss": 0.153, "step": 8425 }, { "epoch": 28.99, "learning_rate": 6.625e-09, "loss": 0.1547, "step": 8450 }, { "epoch": 29.07, "learning_rate": 3.5e-09, "loss": 0.1551, "step": 8475 }, { "epoch": 29.16, "learning_rate": 3.75e-10, "loss": 0.1548, "step": 8500 }, { "epoch": 29.16, "eval_loss": 0.21509824693202972, "eval_runtime": 1891.6612, "eval_samples_per_second": 4.924, "eval_steps_per_second": 0.154, "eval_wer": 29.92358146984869, "step": 8500 }, { "epoch": 29.16, "step": 8500, "total_flos": 3.523117159415808e+19, "train_loss": 0.1825357694906347, "train_runtime": 58163.2136, "train_samples_per_second": 9.353, "train_steps_per_second": 0.146 } ], "logging_steps": 25, "max_steps": 8500, "num_input_tokens_seen": 0, "num_train_epochs": 30, "save_steps": 500, "total_flos": 3.523117159415808e+19, "train_batch_size": 32, "trial_name": null, "trial_params": null }