| { | |
| "best_metric": 41.47529142173956, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 7.112375533428165, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5e-06, | |
| "loss": 4.1975, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.353475781823255e-06, | |
| "loss": 2.1673, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 7.0909127620559225e-06, | |
| "loss": 1.5647, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 7.600607256866187e-06, | |
| "loss": 1.3195, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.99052002250922e-06, | |
| "loss": 1.1266, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.306372679485841e-06, | |
| "loss": 1.0249, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.571858177028285e-06, | |
| "loss": 0.9504, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.800852474557276e-06, | |
| "loss": 0.8944, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.002185532343359e-06, | |
| "loss": 0.8579, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.181825524111845e-06, | |
| "loss": 0.8082, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.343995804332475e-06, | |
| "loss": 0.7843, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.491795055492509e-06, | |
| "loss": 0.738, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.627564263195184e-06, | |
| "loss": 0.7352, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.753115237662518e-06, | |
| "loss": 0.689, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.869878990592806e-06, | |
| "loss": 0.6907, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.979005496722949e-06, | |
| "loss": 0.6499, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.958695652173914e-06, | |
| "loss": 0.6483, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.904347826086957e-06, | |
| "loss": 0.631, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.85e-06, | |
| "loss": 0.6239, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.795652173913044e-06, | |
| "loss": 0.6266, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.741304347826088e-06, | |
| "loss": 0.5798, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.686956521739131e-06, | |
| "loss": 0.5646, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.632608695652174e-06, | |
| "loss": 0.5941, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.578260869565218e-06, | |
| "loss": 0.5765, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.523913043478261e-06, | |
| "loss": 0.5629, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.469565217391305e-06, | |
| "loss": 0.56, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.415217391304348e-06, | |
| "loss": 0.5278, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.360869565217392e-06, | |
| "loss": 0.5222, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.306521739130435e-06, | |
| "loss": 0.454, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.252173913043479e-06, | |
| "loss": 0.4257, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.197826086956522e-06, | |
| "loss": 0.4147, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 9.143478260869566e-06, | |
| "loss": 0.4281, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.08913043478261e-06, | |
| "loss": 0.4128, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 9.034782608695653e-06, | |
| "loss": 0.4162, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 8.980434782608696e-06, | |
| "loss": 0.4048, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.92608695652174e-06, | |
| "loss": 0.412, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.871739130434783e-06, | |
| "loss": 0.4121, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 8.817391304347827e-06, | |
| "loss": 0.4114, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.76304347826087e-06, | |
| "loss": 0.4008, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.708695652173914e-06, | |
| "loss": 0.4105, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_loss": 0.55078125, | |
| "eval_runtime": 1444.4454, | |
| "eval_samples_per_second": 10.38, | |
| "eval_steps_per_second": 0.325, | |
| "eval_wer": 48.12443957357776, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.654347826086957e-06, | |
| "loss": 0.3946, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.6e-06, | |
| "loss": 0.4019, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 8.545652173913044e-06, | |
| "loss": 0.3976, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.491304347826088e-06, | |
| "loss": 0.3945, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.436956521739131e-06, | |
| "loss": 0.3922, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 8.382608695652175e-06, | |
| "loss": 0.3872, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.328260869565218e-06, | |
| "loss": 0.4003, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.273913043478262e-06, | |
| "loss": 0.3661, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.219565217391305e-06, | |
| "loss": 0.3727, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.165217391304349e-06, | |
| "loss": 0.362, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.110869565217392e-06, | |
| "loss": 0.3838, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.056521739130435e-06, | |
| "loss": 0.3769, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.002173913043479e-06, | |
| "loss": 0.3863, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.947826086956522e-06, | |
| "loss": 0.3736, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 7.893478260869566e-06, | |
| "loss": 0.3779, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.83913043478261e-06, | |
| "loss": 0.3683, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.784782608695653e-06, | |
| "loss": 0.2924, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.730434782608696e-06, | |
| "loss": 0.2672, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.67608695652174e-06, | |
| "loss": 0.2528, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 7.621739130434783e-06, | |
| "loss": 0.2698, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.567391304347827e-06, | |
| "loss": 0.2618, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.5130434782608695e-06, | |
| "loss": 0.2584, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 7.458695652173914e-06, | |
| "loss": 0.2564, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 7.404347826086957e-06, | |
| "loss": 0.2693, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 7.350000000000001e-06, | |
| "loss": 0.2577, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.295652173913043e-06, | |
| "loss": 0.2644, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.241304347826088e-06, | |
| "loss": 0.2523, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.186956521739131e-06, | |
| "loss": 0.2646, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 7.132608695652175e-06, | |
| "loss": 0.2574, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 7.078260869565217e-06, | |
| "loss": 0.2583, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.023913043478261e-06, | |
| "loss": 0.2638, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.969565217391305e-06, | |
| "loss": 0.2615, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.915217391304349e-06, | |
| "loss": 0.2583, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.860869565217392e-06, | |
| "loss": 0.2551, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 6.806521739130435e-06, | |
| "loss": 0.2624, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 6.752173913043479e-06, | |
| "loss": 0.2543, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 6.6978260869565226e-06, | |
| "loss": 0.2607, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 6.643478260869566e-06, | |
| "loss": 0.2592, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 6.589130434782609e-06, | |
| "loss": 0.2614, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 6.534782608695652e-06, | |
| "loss": 0.258, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "eval_loss": 0.490478515625, | |
| "eval_runtime": 1402.708, | |
| "eval_samples_per_second": 10.689, | |
| "eval_steps_per_second": 0.334, | |
| "eval_wer": 42.0294908837302, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 6.4804347826086965e-06, | |
| "loss": 0.2471, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 6.42608695652174e-06, | |
| "loss": 0.2553, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 6.3717391304347835e-06, | |
| "loss": 0.2505, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 6.317391304347826e-06, | |
| "loss": 0.2604, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.26304347826087e-06, | |
| "loss": 0.2013, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.208695652173914e-06, | |
| "loss": 0.1813, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.154347826086957e-06, | |
| "loss": 0.1682, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.1e-06, | |
| "loss": 0.1621, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.0456521739130435e-06, | |
| "loss": 0.168, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 5.991304347826088e-06, | |
| "loss": 0.1714, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 5.936956521739131e-06, | |
| "loss": 0.162, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 5.882608695652175e-06, | |
| "loss": 0.1642, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 5.8282608695652175e-06, | |
| "loss": 0.167, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 5.773913043478261e-06, | |
| "loss": 0.1697, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 5.719565217391305e-06, | |
| "loss": 0.1771, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 5.665217391304349e-06, | |
| "loss": 0.1719, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 5.610869565217391e-06, | |
| "loss": 0.1723, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 5.556521739130435e-06, | |
| "loss": 0.1744, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.502173913043479e-06, | |
| "loss": 0.1699, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.447826086956523e-06, | |
| "loss": 0.1741, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.393478260869565e-06, | |
| "loss": 0.1671, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 5.339130434782609e-06, | |
| "loss": 0.1669, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 5.284782608695652e-06, | |
| "loss": 0.1704, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.230434782608697e-06, | |
| "loss": 0.1628, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 5.17608695652174e-06, | |
| "loss": 0.1784, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 5.121739130434783e-06, | |
| "loss": 0.1724, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 5.067391304347826e-06, | |
| "loss": 0.1703, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 5.01304347826087e-06, | |
| "loss": 0.1694, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 4.958695652173913e-06, | |
| "loss": 0.1672, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 4.904347826086957e-06, | |
| "loss": 0.1721, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 4.85e-06, | |
| "loss": 0.1628, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 4.795652173913044e-06, | |
| "loss": 0.1669, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.741304347826087e-06, | |
| "loss": 0.1451, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.6869565217391314e-06, | |
| "loss": 0.1112, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.632608695652174e-06, | |
| "loss": 0.1013, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 4.5782608695652176e-06, | |
| "loss": 0.1037, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 4.523913043478261e-06, | |
| "loss": 0.102, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 4.4695652173913045e-06, | |
| "loss": 0.1056, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 4.415217391304348e-06, | |
| "loss": 0.1133, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 4.3608695652173915e-06, | |
| "loss": 0.0998, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "eval_loss": 0.52197265625, | |
| "eval_runtime": 1430.0538, | |
| "eval_samples_per_second": 10.485, | |
| "eval_steps_per_second": 0.328, | |
| "eval_wer": 41.680781109893395, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 4.306521739130435e-06, | |
| "loss": 0.1052, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 4.2521739130434785e-06, | |
| "loss": 0.1045, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 4.197826086956522e-06, | |
| "loss": 0.1025, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 4.1434782608695654e-06, | |
| "loss": 0.1068, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 4.089130434782609e-06, | |
| "loss": 0.1042, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 4.034782608695652e-06, | |
| "loss": 0.1107, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.980434782608696e-06, | |
| "loss": 0.111, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 3.926086956521739e-06, | |
| "loss": 0.1037, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 3.871739130434783e-06, | |
| "loss": 0.1072, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.817391304347826e-06, | |
| "loss": 0.1089, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.76304347826087e-06, | |
| "loss": 0.1086, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.7086956521739133e-06, | |
| "loss": 0.1072, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.6543478260869568e-06, | |
| "loss": 0.1089, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.1045, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 3.5456521739130437e-06, | |
| "loss": 0.1056, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 3.4913043478260872e-06, | |
| "loss": 0.1105, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 3.436956521739131e-06, | |
| "loss": 0.1078, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 3.382608695652174e-06, | |
| "loss": 0.1024, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 3.3282608695652173e-06, | |
| "loss": 0.1058, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.273913043478261e-06, | |
| "loss": 0.1114, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.2195652173913046e-06, | |
| "loss": 0.0885, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.165217391304348e-06, | |
| "loss": 0.0652, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 3.1108695652173916e-06, | |
| "loss": 0.0651, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.056521739130435e-06, | |
| "loss": 0.0642, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 3.0021739130434786e-06, | |
| "loss": 0.0648, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 2.9478260869565216e-06, | |
| "loss": 0.0639, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.8934782608695655e-06, | |
| "loss": 0.0675, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 2.8391304347826086e-06, | |
| "loss": 0.0644, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 2.7847826086956525e-06, | |
| "loss": 0.0658, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.7304347826086956e-06, | |
| "loss": 0.0628, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 2.6760869565217395e-06, | |
| "loss": 0.0643, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 2.621739130434783e-06, | |
| "loss": 0.0656, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.5673913043478264e-06, | |
| "loss": 0.0675, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 2.51304347826087e-06, | |
| "loss": 0.0631, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 2.4586956521739134e-06, | |
| "loss": 0.0664, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.404347826086957e-06, | |
| "loss": 0.0638, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 2.35e-06, | |
| "loss": 0.065, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.2956521739130434e-06, | |
| "loss": 0.0631, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 2.241304347826087e-06, | |
| "loss": 0.0659, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 2.1869565217391304e-06, | |
| "loss": 0.064, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "eval_loss": 0.5546875, | |
| "eval_runtime": 1414.7522, | |
| "eval_samples_per_second": 10.598, | |
| "eval_steps_per_second": 0.332, | |
| "eval_wer": 41.47529142173956, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.1369565217391305e-06, | |
| "loss": 0.0684, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 2.082608695652174e-06, | |
| "loss": 0.0638, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 2.0282608695652174e-06, | |
| "loss": 0.064, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 1.973913043478261e-06, | |
| "loss": 0.0667, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 1.9195652173913044e-06, | |
| "loss": 0.0651, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 1.865217391304348e-06, | |
| "loss": 0.066, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 1.8108695652173913e-06, | |
| "loss": 0.0625, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 1.7565217391304348e-06, | |
| "loss": 0.0629, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 1.7021739130434783e-06, | |
| "loss": 0.0599, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.6478260869565218e-06, | |
| "loss": 0.0407, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.5934782608695653e-06, | |
| "loss": 0.0412, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.539130434782609e-06, | |
| "loss": 0.0409, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 1.4847826086956525e-06, | |
| "loss": 0.0394, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 1.430434782608696e-06, | |
| "loss": 0.0425, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.3760869565217392e-06, | |
| "loss": 0.0442, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 1.3217391304347827e-06, | |
| "loss": 0.0412, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 1.2673913043478262e-06, | |
| "loss": 0.041, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 1.2130434782608697e-06, | |
| "loss": 0.0395, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 1.1586956521739131e-06, | |
| "loss": 0.0424, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 1.1043478260869566e-06, | |
| "loss": 0.0408, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 1.0500000000000001e-06, | |
| "loss": 0.0442, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 9.956521739130436e-07, | |
| "loss": 0.0424, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.51, | |
| "learning_rate": 9.413043478260871e-07, | |
| "loss": 0.0421, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 8.869565217391305e-07, | |
| "loss": 0.0435, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 8.326086956521739e-07, | |
| "loss": 0.0398, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 7.782608695652174e-07, | |
| "loss": 0.0404, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 7.23913043478261e-07, | |
| "loss": 0.0386, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 6.695652173913044e-07, | |
| "loss": 0.0422, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 6.152173913043479e-07, | |
| "loss": 0.0387, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 5.608695652173914e-07, | |
| "loss": 0.0413, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 5.065217391304348e-07, | |
| "loss": 0.0431, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 4.521739130434783e-07, | |
| "loss": 0.0423, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 3.9782608695652175e-07, | |
| "loss": 0.0432, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 3.434782608695653e-07, | |
| "loss": 0.0387, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 2.891304347826087e-07, | |
| "loss": 0.0388, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 2.3478260869565217e-07, | |
| "loss": 0.0372, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.8043478260869565e-07, | |
| "loss": 0.0361, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.2608695652173914e-07, | |
| "loss": 0.034, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.173913043478261e-08, | |
| "loss": 0.0302, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 1.739130434782609e-08, | |
| "loss": 0.0356, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "eval_loss": 0.5888671875, | |
| "eval_runtime": 1380.3134, | |
| "eval_samples_per_second": 10.863, | |
| "eval_steps_per_second": 0.34, | |
| "eval_wer": 41.57367739364352, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "step": 5000, | |
| "total_flos": 9.232510778931885e+19, | |
| "train_loss": 0.2779711181640625, | |
| "train_runtime": 22972.7584, | |
| "train_samples_per_second": 13.93, | |
| "train_steps_per_second": 0.218 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 9.232510778931885e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |