{ "best_metric": 9.899702823179792, "best_model_checkpoint": "./checkpoint-8000", "epoch": 646.8235294117648, "global_step": 11000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.47, "learning_rate": 1.5136083400296205e-06, "loss": 0.0024, "step": 25 }, { "epoch": 2.94, "learning_rate": 1.8687587131475301e-06, "loss": 0.0024, "step": 50 }, { "epoch": 4.41, "learning_rate": 2.0711488350670174e-06, "loss": 0.0023, "step": 75 }, { "epoch": 5.88, "learning_rate": 2.213317753617305e-06, "loss": 0.0023, "step": 100 }, { "epoch": 7.35, "learning_rate": 2.3230029693718747e-06, "loss": 0.002, "step": 125 }, { "epoch": 8.82, "learning_rate": 2.412322158351148e-06, "loss": 0.002, "step": 150 }, { "epoch": 10.29, "learning_rate": 2.4876668872198717e-06, "loss": 0.0022, "step": 175 }, { "epoch": 11.76, "learning_rate": 2.552824062407326e-06, "loss": 0.0021, "step": 200 }, { "epoch": 13.24, "learning_rate": 2.610223373296667e-06, "loss": 0.0034, "step": 225 }, { "epoch": 14.71, "learning_rate": 2.661517182828361e-06, "loss": 0.0019, "step": 250 }, { "epoch": 16.18, "learning_rate": 2.7078803874740543e-06, "loss": 0.0018, "step": 275 }, { "epoch": 17.65, "learning_rate": 2.750178319990197e-06, "loss": 0.0023, "step": 300 }, { "epoch": 19.12, "learning_rate": 2.7890667754365044e-06, "loss": 0.0019, "step": 325 }, { "epoch": 20.59, "learning_rate": 2.8250546392106077e-06, "loss": 0.0021, "step": 350 }, { "epoch": 22.06, "learning_rate": 2.8585447348549113e-06, "loss": 0.0023, "step": 375 }, { "epoch": 23.53, "learning_rate": 2.889861392935294e-06, "loss": 0.0021, "step": 400 }, { "epoch": 25.0, "learning_rate": 2.9192696063561725e-06, "loss": 0.0016, "step": 425 }, { "epoch": 26.47, "learning_rate": 2.946988676871634e-06, "loss": 0.0018, "step": 450 }, { "epoch": 27.94, "learning_rate": 2.973202150939645e-06, "loss": 0.0022, "step": 475 }, { "epoch": 29.41, "learning_rate": 2.998065193492142e-06, "loss": 0.0018, "step": 500 }, { "epoch": 30.88, "learning_rate": 2.9811428571428573e-06, "loss": 0.0018, "step": 525 }, { "epoch": 32.35, "learning_rate": 2.959714285714286e-06, "loss": 0.0021, "step": 550 }, { "epoch": 33.82, "learning_rate": 2.9382857142857143e-06, "loss": 0.0017, "step": 575 }, { "epoch": 35.29, "learning_rate": 2.916857142857143e-06, "loss": 0.0018, "step": 600 }, { "epoch": 36.76, "learning_rate": 2.895428571428572e-06, "loss": 0.0016, "step": 625 }, { "epoch": 38.24, "learning_rate": 2.874e-06, "loss": 0.0017, "step": 650 }, { "epoch": 39.71, "learning_rate": 2.8525714285714285e-06, "loss": 0.0016, "step": 675 }, { "epoch": 41.18, "learning_rate": 2.8311428571428574e-06, "loss": 0.0013, "step": 700 }, { "epoch": 42.65, "learning_rate": 2.809714285714286e-06, "loss": 0.0014, "step": 725 }, { "epoch": 44.12, "learning_rate": 2.788285714285714e-06, "loss": 0.0012, "step": 750 }, { "epoch": 45.59, "learning_rate": 2.766857142857143e-06, "loss": 0.0014, "step": 775 }, { "epoch": 47.06, "learning_rate": 2.7454285714285716e-06, "loss": 0.0014, "step": 800 }, { "epoch": 48.53, "learning_rate": 2.724e-06, "loss": 0.0013, "step": 825 }, { "epoch": 50.0, "learning_rate": 2.7025714285714286e-06, "loss": 0.0011, "step": 850 }, { "epoch": 51.47, "learning_rate": 2.681142857142857e-06, "loss": 0.0013, "step": 875 }, { "epoch": 52.94, "learning_rate": 2.6597142857142857e-06, "loss": 0.0014, "step": 900 }, { "epoch": 54.41, "learning_rate": 2.6382857142857142e-06, "loss": 0.0011, "step": 925 }, { "epoch": 55.88, "learning_rate": 2.616857142857143e-06, "loss": 0.0021, "step": 950 }, { "epoch": 57.35, "learning_rate": 2.5954285714285713e-06, "loss": 0.0014, "step": 975 }, { "epoch": 58.82, "learning_rate": 2.574e-06, "loss": 0.0014, "step": 1000 }, { "epoch": 58.82, "eval_loss": 0.4951171875, "eval_runtime": 168.7172, "eval_samples_per_second": 1.612, "eval_steps_per_second": 0.101, "eval_wer": 10.36404160475483, "step": 1000 }, { "epoch": 60.29, "learning_rate": 2.5542857142857143e-06, "loss": 0.0013, "step": 1025 }, { "epoch": 61.76, "learning_rate": 2.532857142857143e-06, "loss": 0.0009, "step": 1050 }, { "epoch": 63.24, "learning_rate": 2.5114285714285718e-06, "loss": 0.0012, "step": 1075 }, { "epoch": 64.71, "learning_rate": 2.49e-06, "loss": 0.0009, "step": 1100 }, { "epoch": 66.18, "learning_rate": 2.4685714285714284e-06, "loss": 0.0014, "step": 1125 }, { "epoch": 67.65, "learning_rate": 2.4471428571428574e-06, "loss": 0.0012, "step": 1150 }, { "epoch": 69.12, "learning_rate": 2.425714285714286e-06, "loss": 0.0009, "step": 1175 }, { "epoch": 70.59, "learning_rate": 2.404285714285714e-06, "loss": 0.0008, "step": 1200 }, { "epoch": 72.06, "learning_rate": 2.382857142857143e-06, "loss": 0.0011, "step": 1225 }, { "epoch": 73.53, "learning_rate": 2.3614285714285715e-06, "loss": 0.0015, "step": 1250 }, { "epoch": 75.0, "learning_rate": 2.34e-06, "loss": 0.0008, "step": 1275 }, { "epoch": 76.47, "learning_rate": 2.3185714285714286e-06, "loss": 0.001, "step": 1300 }, { "epoch": 77.94, "learning_rate": 2.297142857142857e-06, "loss": 0.0009, "step": 1325 }, { "epoch": 79.41, "learning_rate": 2.2757142857142856e-06, "loss": 0.0009, "step": 1350 }, { "epoch": 80.88, "learning_rate": 2.2542857142857146e-06, "loss": 0.0011, "step": 1375 }, { "epoch": 82.35, "learning_rate": 2.232857142857143e-06, "loss": 0.0007, "step": 1400 }, { "epoch": 83.82, "learning_rate": 2.2114285714285712e-06, "loss": 0.001, "step": 1425 }, { "epoch": 85.29, "learning_rate": 2.19e-06, "loss": 0.001, "step": 1450 }, { "epoch": 86.76, "learning_rate": 2.1685714285714287e-06, "loss": 0.0008, "step": 1475 }, { "epoch": 88.24, "learning_rate": 2.1471428571428573e-06, "loss": 0.001, "step": 1500 }, { "epoch": 89.71, "learning_rate": 2.125714285714286e-06, "loss": 0.0008, "step": 1525 }, { "epoch": 91.18, "learning_rate": 2.1042857142857143e-06, "loss": 0.0008, "step": 1550 }, { "epoch": 92.65, "learning_rate": 2.082857142857143e-06, "loss": 0.0006, "step": 1575 }, { "epoch": 94.12, "learning_rate": 2.0614285714285714e-06, "loss": 0.0008, "step": 1600 }, { "epoch": 95.59, "learning_rate": 2.0400000000000004e-06, "loss": 0.0009, "step": 1625 }, { "epoch": 97.06, "learning_rate": 2.0185714285714285e-06, "loss": 0.0009, "step": 1650 }, { "epoch": 98.53, "learning_rate": 1.997142857142857e-06, "loss": 0.0008, "step": 1675 }, { "epoch": 100.0, "learning_rate": 1.975714285714286e-06, "loss": 0.0008, "step": 1700 }, { "epoch": 101.47, "learning_rate": 1.9542857142857145e-06, "loss": 0.0007, "step": 1725 }, { "epoch": 102.94, "learning_rate": 1.9328571428571426e-06, "loss": 0.0005, "step": 1750 }, { "epoch": 104.41, "learning_rate": 1.9114285714285716e-06, "loss": 0.0008, "step": 1775 }, { "epoch": 105.88, "learning_rate": 1.8900000000000001e-06, "loss": 0.0006, "step": 1800 }, { "epoch": 107.35, "learning_rate": 1.8685714285714287e-06, "loss": 0.0007, "step": 1825 }, { "epoch": 108.82, "learning_rate": 1.8471428571428574e-06, "loss": 0.0008, "step": 1850 }, { "epoch": 110.29, "learning_rate": 1.8257142857142857e-06, "loss": 0.0008, "step": 1875 }, { "epoch": 111.76, "learning_rate": 1.8042857142857143e-06, "loss": 0.0005, "step": 1900 }, { "epoch": 113.24, "learning_rate": 1.782857142857143e-06, "loss": 0.0007, "step": 1925 }, { "epoch": 114.71, "learning_rate": 1.7614285714285715e-06, "loss": 0.0006, "step": 1950 }, { "epoch": 116.18, "learning_rate": 1.7399999999999999e-06, "loss": 0.0005, "step": 1975 }, { "epoch": 117.65, "learning_rate": 1.7185714285714286e-06, "loss": 0.0006, "step": 2000 }, { "epoch": 117.65, "eval_loss": 0.51806640625, "eval_runtime": 168.3992, "eval_samples_per_second": 1.615, "eval_steps_per_second": 0.101, "eval_wer": 10.280460624071322, "step": 2000 }, { "epoch": 118.47, "learning_rate": 2.4642352941176473e-06, "loss": 0.0007, "step": 2025 }, { "epoch": 119.94, "learning_rate": 2.4554117647058827e-06, "loss": 0.0006, "step": 2050 }, { "epoch": 121.41, "learning_rate": 2.4465882352941176e-06, "loss": 0.0003, "step": 2075 }, { "epoch": 122.88, "learning_rate": 2.437764705882353e-06, "loss": 0.0007, "step": 2100 }, { "epoch": 124.35, "learning_rate": 2.4289411764705882e-06, "loss": 0.0005, "step": 2125 }, { "epoch": 125.82, "learning_rate": 2.4201176470588236e-06, "loss": 0.0006, "step": 2150 }, { "epoch": 127.29, "learning_rate": 2.411294117647059e-06, "loss": 0.0007, "step": 2175 }, { "epoch": 128.76, "learning_rate": 2.4024705882352942e-06, "loss": 0.0006, "step": 2200 }, { "epoch": 130.24, "learning_rate": 2.3936470588235295e-06, "loss": 0.0011, "step": 2225 }, { "epoch": 131.71, "learning_rate": 2.384823529411765e-06, "loss": 0.0007, "step": 2250 }, { "epoch": 133.18, "learning_rate": 2.376e-06, "loss": 0.0006, "step": 2275 }, { "epoch": 134.65, "learning_rate": 2.3671764705882355e-06, "loss": 0.0005, "step": 2300 }, { "epoch": 136.12, "learning_rate": 2.3583529411764704e-06, "loss": 0.0007, "step": 2325 }, { "epoch": 137.59, "learning_rate": 2.3495294117647058e-06, "loss": 0.0008, "step": 2350 }, { "epoch": 139.06, "learning_rate": 2.340705882352941e-06, "loss": 0.0036, "step": 2375 }, { "epoch": 140.53, "learning_rate": 2.3318823529411764e-06, "loss": 0.0006, "step": 2400 }, { "epoch": 142.0, "learning_rate": 2.3230588235294118e-06, "loss": 0.0006, "step": 2425 }, { "epoch": 143.47, "learning_rate": 2.314235294117647e-06, "loss": 0.0004, "step": 2450 }, { "epoch": 144.94, "learning_rate": 2.3054117647058824e-06, "loss": 0.0018, "step": 2475 }, { "epoch": 146.41, "learning_rate": 2.2965882352941177e-06, "loss": 0.0007, "step": 2500 }, { "epoch": 147.88, "learning_rate": 2.287764705882353e-06, "loss": 0.0005, "step": 2525 }, { "epoch": 149.35, "learning_rate": 2.2789411764705884e-06, "loss": 0.0006, "step": 2550 }, { "epoch": 150.82, "learning_rate": 2.2701176470588237e-06, "loss": 0.0007, "step": 2575 }, { "epoch": 152.29, "learning_rate": 2.2612941176470586e-06, "loss": 0.0007, "step": 2600 }, { "epoch": 153.76, "learning_rate": 2.252470588235294e-06, "loss": 0.0008, "step": 2625 }, { "epoch": 155.24, "learning_rate": 2.2436470588235293e-06, "loss": 0.0005, "step": 2650 }, { "epoch": 156.71, "learning_rate": 2.2348235294117646e-06, "loss": 0.0005, "step": 2675 }, { "epoch": 158.18, "learning_rate": 2.226e-06, "loss": 0.0004, "step": 2700 }, { "epoch": 159.65, "learning_rate": 2.2171764705882353e-06, "loss": 0.0008, "step": 2725 }, { "epoch": 161.12, "learning_rate": 2.2083529411764706e-06, "loss": 0.0005, "step": 2750 }, { "epoch": 162.59, "learning_rate": 2.199529411764706e-06, "loss": 0.0005, "step": 2775 }, { "epoch": 164.06, "learning_rate": 2.1907058823529413e-06, "loss": 0.0005, "step": 2800 }, { "epoch": 165.53, "learning_rate": 2.1818823529411766e-06, "loss": 0.0006, "step": 2825 }, { "epoch": 167.0, "learning_rate": 2.1730588235294115e-06, "loss": 0.0006, "step": 2850 }, { "epoch": 168.47, "learning_rate": 2.164235294117647e-06, "loss": 0.0006, "step": 2875 }, { "epoch": 169.94, "learning_rate": 2.155411764705882e-06, "loss": 0.0008, "step": 2900 }, { "epoch": 171.41, "learning_rate": 2.1465882352941175e-06, "loss": 0.0005, "step": 2925 }, { "epoch": 172.88, "learning_rate": 2.137764705882353e-06, "loss": 0.0005, "step": 2950 }, { "epoch": 174.35, "learning_rate": 2.128941176470588e-06, "loss": 0.0004, "step": 2975 }, { "epoch": 175.82, "learning_rate": 2.1201176470588235e-06, "loss": 0.0007, "step": 3000 }, { "epoch": 175.82, "eval_loss": 0.53173828125, "eval_runtime": 153.6387, "eval_samples_per_second": 1.77, "eval_steps_per_second": 0.111, "eval_wer": 10.11329866270431, "step": 3000 }, { "epoch": 177.29, "learning_rate": 2.112e-06, "loss": 0.0005, "step": 3025 }, { "epoch": 178.76, "learning_rate": 2.103176470588235e-06, "loss": 0.0005, "step": 3050 }, { "epoch": 180.24, "learning_rate": 2.0943529411764705e-06, "loss": 0.0007, "step": 3075 }, { "epoch": 181.71, "learning_rate": 2.085529411764706e-06, "loss": 0.0004, "step": 3100 }, { "epoch": 183.18, "learning_rate": 2.076705882352941e-06, "loss": 0.0004, "step": 3125 }, { "epoch": 184.65, "learning_rate": 2.0678823529411765e-06, "loss": 0.0005, "step": 3150 }, { "epoch": 186.12, "learning_rate": 2.059058823529412e-06, "loss": 0.0005, "step": 3175 }, { "epoch": 187.59, "learning_rate": 2.050235294117647e-06, "loss": 0.0004, "step": 3200 }, { "epoch": 189.06, "learning_rate": 2.0414117647058825e-06, "loss": 0.0005, "step": 3225 }, { "epoch": 190.53, "learning_rate": 2.032588235294118e-06, "loss": 0.0003, "step": 3250 }, { "epoch": 192.0, "learning_rate": 2.0237647058823527e-06, "loss": 0.0006, "step": 3275 }, { "epoch": 193.47, "learning_rate": 2.014941176470588e-06, "loss": 0.0005, "step": 3300 }, { "epoch": 194.94, "learning_rate": 2.0061176470588234e-06, "loss": 0.0004, "step": 3325 }, { "epoch": 196.41, "learning_rate": 1.9972941176470587e-06, "loss": 0.0004, "step": 3350 }, { "epoch": 197.88, "learning_rate": 1.988470588235294e-06, "loss": 0.0004, "step": 3375 }, { "epoch": 199.35, "learning_rate": 1.9796470588235294e-06, "loss": 0.0006, "step": 3400 }, { "epoch": 200.82, "learning_rate": 1.9708235294117647e-06, "loss": 0.0005, "step": 3425 }, { "epoch": 202.29, "learning_rate": 1.962e-06, "loss": 0.0004, "step": 3450 }, { "epoch": 203.76, "learning_rate": 1.9531764705882353e-06, "loss": 0.0005, "step": 3475 }, { "epoch": 205.24, "learning_rate": 1.9443529411764707e-06, "loss": 0.0006, "step": 3500 }, { "epoch": 206.71, "learning_rate": 1.9355294117647056e-06, "loss": 0.0003, "step": 3525 }, { "epoch": 208.18, "learning_rate": 1.926705882352941e-06, "loss": 0.0004, "step": 3550 }, { "epoch": 209.65, "learning_rate": 1.9178823529411762e-06, "loss": 0.0005, "step": 3575 }, { "epoch": 211.12, "learning_rate": 1.9090588235294116e-06, "loss": 0.0004, "step": 3600 }, { "epoch": 212.59, "learning_rate": 1.9002352941176471e-06, "loss": 0.0004, "step": 3625 }, { "epoch": 214.06, "learning_rate": 1.8914117647058824e-06, "loss": 0.0003, "step": 3650 }, { "epoch": 215.53, "learning_rate": 1.8825882352941178e-06, "loss": 0.0004, "step": 3675 }, { "epoch": 217.0, "learning_rate": 1.873764705882353e-06, "loss": 0.0005, "step": 3700 }, { "epoch": 218.47, "learning_rate": 1.8649411764705884e-06, "loss": 0.0004, "step": 3725 }, { "epoch": 219.94, "learning_rate": 1.8561176470588238e-06, "loss": 0.0005, "step": 3750 }, { "epoch": 221.41, "learning_rate": 1.847294117647059e-06, "loss": 0.0003, "step": 3775 }, { "epoch": 222.88, "learning_rate": 1.838470588235294e-06, "loss": 0.0004, "step": 3800 }, { "epoch": 224.35, "learning_rate": 1.8296470588235293e-06, "loss": 0.0003, "step": 3825 }, { "epoch": 225.82, "learning_rate": 1.8208235294117646e-06, "loss": 0.0003, "step": 3850 }, { "epoch": 227.29, "learning_rate": 1.812e-06, "loss": 0.0003, "step": 3875 }, { "epoch": 228.76, "learning_rate": 1.8031764705882353e-06, "loss": 0.0005, "step": 3900 }, { "epoch": 230.24, "learning_rate": 1.7943529411764706e-06, "loss": 0.0004, "step": 3925 }, { "epoch": 231.71, "learning_rate": 1.785529411764706e-06, "loss": 0.0004, "step": 3950 }, { "epoch": 233.18, "learning_rate": 1.7767058823529413e-06, "loss": 0.0003, "step": 3975 }, { "epoch": 234.65, "learning_rate": 1.7678823529411766e-06, "loss": 0.0004, "step": 4000 }, { "epoch": 234.65, "eval_loss": 0.53955078125, "eval_runtime": 161.7321, "eval_samples_per_second": 1.682, "eval_steps_per_second": 0.105, "eval_wer": 10.12258543833581, "step": 4000 }, { "epoch": 236.12, "learning_rate": 1.759764705882353e-06, "loss": 0.0003, "step": 4025 }, { "epoch": 237.59, "learning_rate": 1.7509411764705883e-06, "loss": 0.0003, "step": 4050 }, { "epoch": 239.06, "learning_rate": 1.7421176470588237e-06, "loss": 0.0002, "step": 4075 }, { "epoch": 240.53, "learning_rate": 1.733294117647059e-06, "loss": 0.0004, "step": 4100 }, { "epoch": 242.0, "learning_rate": 1.7244705882352943e-06, "loss": 0.0005, "step": 4125 }, { "epoch": 243.47, "learning_rate": 1.7156470588235296e-06, "loss": 0.0003, "step": 4150 }, { "epoch": 244.94, "learning_rate": 1.706823529411765e-06, "loss": 0.0003, "step": 4175 }, { "epoch": 246.41, "learning_rate": 1.6979999999999999e-06, "loss": 0.0004, "step": 4200 }, { "epoch": 247.88, "learning_rate": 1.6891764705882352e-06, "loss": 0.0003, "step": 4225 }, { "epoch": 249.35, "learning_rate": 1.6803529411764705e-06, "loss": 0.0003, "step": 4250 }, { "epoch": 250.82, "learning_rate": 1.6715294117647059e-06, "loss": 0.0003, "step": 4275 }, { "epoch": 252.29, "learning_rate": 1.6627058823529412e-06, "loss": 0.0004, "step": 4300 }, { "epoch": 253.76, "learning_rate": 1.6538823529411765e-06, "loss": 0.0004, "step": 4325 }, { "epoch": 255.24, "learning_rate": 1.6450588235294119e-06, "loss": 0.0003, "step": 4350 }, { "epoch": 256.71, "learning_rate": 1.6362352941176472e-06, "loss": 0.0003, "step": 4375 }, { "epoch": 258.18, "learning_rate": 1.6274117647058825e-06, "loss": 0.0003, "step": 4400 }, { "epoch": 259.65, "learning_rate": 1.6185882352941178e-06, "loss": 0.0007, "step": 4425 }, { "epoch": 261.12, "learning_rate": 1.6097647058823532e-06, "loss": 0.0003, "step": 4450 }, { "epoch": 262.59, "learning_rate": 1.600941176470588e-06, "loss": 0.0004, "step": 4475 }, { "epoch": 264.06, "learning_rate": 1.5921176470588234e-06, "loss": 0.0003, "step": 4500 }, { "epoch": 265.53, "learning_rate": 1.5832941176470587e-06, "loss": 0.0004, "step": 4525 }, { "epoch": 267.0, "learning_rate": 1.574470588235294e-06, "loss": 0.0002, "step": 4550 }, { "epoch": 268.47, "learning_rate": 1.5656470588235294e-06, "loss": 0.0004, "step": 4575 }, { "epoch": 269.94, "learning_rate": 1.5568235294117647e-06, "loss": 0.0003, "step": 4600 }, { "epoch": 271.41, "learning_rate": 1.548e-06, "loss": 0.0002, "step": 4625 }, { "epoch": 272.88, "learning_rate": 1.5391764705882354e-06, "loss": 0.0002, "step": 4650 }, { "epoch": 274.35, "learning_rate": 1.5303529411764707e-06, "loss": 0.0003, "step": 4675 }, { "epoch": 275.82, "learning_rate": 1.521529411764706e-06, "loss": 0.0003, "step": 4700 }, { "epoch": 277.29, "learning_rate": 1.5127058823529411e-06, "loss": 0.0003, "step": 4725 }, { "epoch": 278.76, "learning_rate": 1.5038823529411765e-06, "loss": 0.0002, "step": 4750 }, { "epoch": 280.24, "learning_rate": 1.4950588235294118e-06, "loss": 0.0002, "step": 4775 }, { "epoch": 281.71, "learning_rate": 1.486235294117647e-06, "loss": 0.0003, "step": 4800 }, { "epoch": 283.18, "learning_rate": 1.4774117647058823e-06, "loss": 0.0003, "step": 4825 }, { "epoch": 284.65, "learning_rate": 1.4685882352941176e-06, "loss": 0.0004, "step": 4850 }, { "epoch": 286.12, "learning_rate": 1.459764705882353e-06, "loss": 0.0002, "step": 4875 }, { "epoch": 287.59, "learning_rate": 1.4509411764705882e-06, "loss": 0.0002, "step": 4900 }, { "epoch": 289.06, "learning_rate": 1.4421176470588236e-06, "loss": 0.0012, "step": 4925 }, { "epoch": 290.53, "learning_rate": 1.433294117647059e-06, "loss": 0.0003, "step": 4950 }, { "epoch": 292.0, "learning_rate": 1.4244705882352942e-06, "loss": 0.0002, "step": 4975 }, { "epoch": 293.47, "learning_rate": 1.4156470588235296e-06, "loss": 0.0004, "step": 5000 }, { "epoch": 293.47, "eval_loss": 0.55322265625, "eval_runtime": 164.7504, "eval_samples_per_second": 1.651, "eval_steps_per_second": 0.103, "eval_wer": 10.104011887072808, "step": 5000 }, { "epoch": 294.94, "learning_rate": 1.407529411764706e-06, "loss": 0.0002, "step": 5025 }, { "epoch": 296.41, "learning_rate": 1.3987058823529413e-06, "loss": 0.0004, "step": 5050 }, { "epoch": 297.88, "learning_rate": 1.3898823529411764e-06, "loss": 0.0003, "step": 5075 }, { "epoch": 299.35, "learning_rate": 1.3810588235294117e-06, "loss": 0.0003, "step": 5100 }, { "epoch": 300.82, "learning_rate": 1.372235294117647e-06, "loss": 0.0004, "step": 5125 }, { "epoch": 302.29, "learning_rate": 1.3634117647058824e-06, "loss": 0.0004, "step": 5150 }, { "epoch": 303.76, "learning_rate": 1.3545882352941177e-06, "loss": 0.0002, "step": 5175 }, { "epoch": 305.24, "learning_rate": 1.3457647058823528e-06, "loss": 0.0003, "step": 5200 }, { "epoch": 306.71, "learning_rate": 1.3369411764705881e-06, "loss": 0.0005, "step": 5225 }, { "epoch": 308.18, "learning_rate": 1.3281176470588235e-06, "loss": 0.0002, "step": 5250 }, { "epoch": 309.65, "learning_rate": 1.3192941176470588e-06, "loss": 0.0002, "step": 5275 }, { "epoch": 311.12, "learning_rate": 1.3104705882352941e-06, "loss": 0.0002, "step": 5300 }, { "epoch": 312.59, "learning_rate": 1.3016470588235295e-06, "loss": 0.0003, "step": 5325 }, { "epoch": 314.06, "learning_rate": 1.2928235294117648e-06, "loss": 0.0003, "step": 5350 }, { "epoch": 315.53, "learning_rate": 1.284e-06, "loss": 0.0003, "step": 5375 }, { "epoch": 317.0, "learning_rate": 1.2751764705882352e-06, "loss": 0.0003, "step": 5400 }, { "epoch": 318.47, "learning_rate": 1.2663529411764706e-06, "loss": 0.0003, "step": 5425 }, { "epoch": 319.94, "learning_rate": 1.2575294117647059e-06, "loss": 0.0002, "step": 5450 }, { "epoch": 321.41, "learning_rate": 1.2487058823529412e-06, "loss": 0.0002, "step": 5475 }, { "epoch": 322.88, "learning_rate": 1.2398823529411765e-06, "loss": 0.0002, "step": 5500 }, { "epoch": 324.35, "learning_rate": 1.2310588235294119e-06, "loss": 0.0001, "step": 5525 }, { "epoch": 325.82, "learning_rate": 1.2222352941176472e-06, "loss": 0.0002, "step": 5550 }, { "epoch": 327.29, "learning_rate": 1.2134117647058825e-06, "loss": 0.0002, "step": 5575 }, { "epoch": 328.76, "learning_rate": 1.2045882352941177e-06, "loss": 0.0003, "step": 5600 }, { "epoch": 330.24, "learning_rate": 1.195764705882353e-06, "loss": 0.0003, "step": 5625 }, { "epoch": 331.71, "learning_rate": 1.1869411764705883e-06, "loss": 0.0003, "step": 5650 }, { "epoch": 333.18, "learning_rate": 1.1781176470588236e-06, "loss": 0.0003, "step": 5675 }, { "epoch": 334.65, "learning_rate": 1.169294117647059e-06, "loss": 0.0004, "step": 5700 }, { "epoch": 336.12, "learning_rate": 1.1604705882352943e-06, "loss": 0.0002, "step": 5725 }, { "epoch": 337.59, "learning_rate": 1.1516470588235294e-06, "loss": 0.0003, "step": 5750 }, { "epoch": 339.06, "learning_rate": 1.1428235294117647e-06, "loss": 0.0002, "step": 5775 }, { "epoch": 340.53, "learning_rate": 1.134e-06, "loss": 0.0002, "step": 5800 }, { "epoch": 342.0, "learning_rate": 1.1251764705882354e-06, "loss": 0.0003, "step": 5825 }, { "epoch": 343.47, "learning_rate": 1.1163529411764707e-06, "loss": 0.0002, "step": 5850 }, { "epoch": 344.94, "learning_rate": 1.1075294117647058e-06, "loss": 0.0002, "step": 5875 }, { "epoch": 346.41, "learning_rate": 1.0987058823529412e-06, "loss": 0.0002, "step": 5900 }, { "epoch": 347.88, "learning_rate": 1.0898823529411765e-06, "loss": 0.0002, "step": 5925 }, { "epoch": 349.35, "learning_rate": 1.0810588235294118e-06, "loss": 0.0003, "step": 5950 }, { "epoch": 350.82, "learning_rate": 1.0722352941176472e-06, "loss": 0.0002, "step": 5975 }, { "epoch": 352.29, "learning_rate": 1.0634117647058823e-06, "loss": 0.0013, "step": 6000 }, { "epoch": 352.29, "eval_loss": 0.564453125, "eval_runtime": 167.4651, "eval_samples_per_second": 1.624, "eval_steps_per_second": 0.102, "eval_wer": 10.085438335809807, "step": 6000 }, { "epoch": 353.76, "learning_rate": 1.0552941176470589e-06, "loss": 0.0002, "step": 6025 }, { "epoch": 355.24, "learning_rate": 1.0464705882352942e-06, "loss": 0.0002, "step": 6050 }, { "epoch": 356.71, "learning_rate": 1.0376470588235295e-06, "loss": 0.0002, "step": 6075 }, { "epoch": 358.18, "learning_rate": 1.0288235294117649e-06, "loss": 0.0002, "step": 6100 }, { "epoch": 359.65, "learning_rate": 1.0200000000000002e-06, "loss": 0.0002, "step": 6125 }, { "epoch": 361.12, "learning_rate": 1.0111764705882353e-06, "loss": 0.0002, "step": 6150 }, { "epoch": 362.59, "learning_rate": 1.0023529411764706e-06, "loss": 0.0002, "step": 6175 }, { "epoch": 364.06, "learning_rate": 9.93529411764706e-07, "loss": 0.0004, "step": 6200 }, { "epoch": 365.53, "learning_rate": 9.847058823529413e-07, "loss": 0.0002, "step": 6225 }, { "epoch": 367.0, "learning_rate": 9.758823529411766e-07, "loss": 0.0002, "step": 6250 }, { "epoch": 368.47, "learning_rate": 9.670588235294117e-07, "loss": 0.0002, "step": 6275 }, { "epoch": 369.94, "learning_rate": 9.58235294117647e-07, "loss": 0.0002, "step": 6300 }, { "epoch": 371.41, "learning_rate": 9.494117647058824e-07, "loss": 0.0002, "step": 6325 }, { "epoch": 372.88, "learning_rate": 9.405882352941177e-07, "loss": 0.0003, "step": 6350 }, { "epoch": 374.35, "learning_rate": 9.31764705882353e-07, "loss": 0.0002, "step": 6375 }, { "epoch": 375.82, "learning_rate": 9.229411764705882e-07, "loss": 0.0002, "step": 6400 }, { "epoch": 377.29, "learning_rate": 9.141176470588235e-07, "loss": 0.0002, "step": 6425 }, { "epoch": 378.76, "learning_rate": 9.052941176470588e-07, "loss": 0.0002, "step": 6450 }, { "epoch": 380.24, "learning_rate": 8.964705882352942e-07, "loss": 0.0002, "step": 6475 }, { "epoch": 381.71, "learning_rate": 8.876470588235295e-07, "loss": 0.0002, "step": 6500 }, { "epoch": 383.18, "learning_rate": 8.788235294117648e-07, "loss": 0.0002, "step": 6525 }, { "epoch": 384.65, "learning_rate": 8.699999999999999e-07, "loss": 0.0003, "step": 6550 }, { "epoch": 386.12, "learning_rate": 8.611764705882353e-07, "loss": 0.0003, "step": 6575 }, { "epoch": 387.59, "learning_rate": 8.523529411764706e-07, "loss": 0.0002, "step": 6600 }, { "epoch": 389.06, "learning_rate": 8.435294117647059e-07, "loss": 0.0002, "step": 6625 }, { "epoch": 390.53, "learning_rate": 8.347058823529412e-07, "loss": 0.0001, "step": 6650 }, { "epoch": 392.0, "learning_rate": 8.258823529411765e-07, "loss": 0.0002, "step": 6675 }, { "epoch": 393.47, "learning_rate": 8.170588235294118e-07, "loss": 0.0001, "step": 6700 }, { "epoch": 394.94, "learning_rate": 8.082352941176471e-07, "loss": 0.0001, "step": 6725 }, { "epoch": 396.41, "learning_rate": 7.994117647058823e-07, "loss": 0.0002, "step": 6750 }, { "epoch": 397.88, "learning_rate": 7.905882352941177e-07, "loss": 0.0003, "step": 6775 }, { "epoch": 399.35, "learning_rate": 7.817647058823529e-07, "loss": 0.0004, "step": 6800 }, { "epoch": 400.82, "learning_rate": 7.729411764705882e-07, "loss": 0.0003, "step": 6825 }, { "epoch": 402.29, "learning_rate": 7.641176470588236e-07, "loss": 0.0002, "step": 6850 }, { "epoch": 403.76, "learning_rate": 7.552941176470589e-07, "loss": 0.0002, "step": 6875 }, { "epoch": 405.24, "learning_rate": 7.464705882352941e-07, "loss": 0.0002, "step": 6900 }, { "epoch": 406.71, "learning_rate": 7.376470588235294e-07, "loss": 0.0002, "step": 6925 }, { "epoch": 408.18, "learning_rate": 7.288235294117648e-07, "loss": 0.0001, "step": 6950 }, { "epoch": 409.65, "learning_rate": 7.2e-07, "loss": 0.0001, "step": 6975 }, { "epoch": 411.12, "learning_rate": 7.111764705882353e-07, "loss": 0.0002, "step": 7000 }, { "epoch": 411.12, "eval_loss": 0.56689453125, "eval_runtime": 168.4232, "eval_samples_per_second": 1.615, "eval_steps_per_second": 0.101, "eval_wer": 10.11329866270431, "step": 7000 }, { "epoch": 412.59, "learning_rate": 7.030588235294118e-07, "loss": 0.0002, "step": 7025 }, { "epoch": 414.06, "learning_rate": 6.94235294117647e-07, "loss": 0.0003, "step": 7050 }, { "epoch": 415.53, "learning_rate": 6.854117647058824e-07, "loss": 0.0002, "step": 7075 }, { "epoch": 417.0, "learning_rate": 6.765882352941177e-07, "loss": 0.0001, "step": 7100 }, { "epoch": 418.47, "learning_rate": 6.677647058823529e-07, "loss": 0.0003, "step": 7125 }, { "epoch": 419.94, "learning_rate": 6.589411764705882e-07, "loss": 0.0002, "step": 7150 }, { "epoch": 421.41, "learning_rate": 6.501176470588235e-07, "loss": 0.0003, "step": 7175 }, { "epoch": 422.88, "learning_rate": 6.412941176470588e-07, "loss": 0.0003, "step": 7200 }, { "epoch": 424.35, "learning_rate": 6.324705882352941e-07, "loss": 0.0003, "step": 7225 }, { "epoch": 425.82, "learning_rate": 6.236470588235294e-07, "loss": 0.0002, "step": 7250 }, { "epoch": 427.29, "learning_rate": 6.148235294117648e-07, "loss": 0.0002, "step": 7275 }, { "epoch": 428.76, "learning_rate": 6.060000000000001e-07, "loss": 0.0003, "step": 7300 }, { "epoch": 430.24, "learning_rate": 5.971764705882353e-07, "loss": 0.0002, "step": 7325 }, { "epoch": 431.71, "learning_rate": 5.883529411764707e-07, "loss": 0.0003, "step": 7350 }, { "epoch": 433.18, "learning_rate": 5.795294117647059e-07, "loss": 0.0004, "step": 7375 }, { "epoch": 434.65, "learning_rate": 5.707058823529412e-07, "loss": 0.0002, "step": 7400 }, { "epoch": 436.12, "learning_rate": 5.618823529411765e-07, "loss": 0.0001, "step": 7425 }, { "epoch": 437.59, "learning_rate": 5.530588235294118e-07, "loss": 0.0002, "step": 7450 }, { "epoch": 439.06, "learning_rate": 5.442352941176471e-07, "loss": 0.0002, "step": 7475 }, { "epoch": 440.53, "learning_rate": 5.354117647058824e-07, "loss": 0.0001, "step": 7500 }, { "epoch": 442.0, "learning_rate": 5.265882352941176e-07, "loss": 0.0003, "step": 7525 }, { "epoch": 443.47, "learning_rate": 5.17764705882353e-07, "loss": 0.0008, "step": 7550 }, { "epoch": 444.94, "learning_rate": 5.089411764705882e-07, "loss": 0.0002, "step": 7575 }, { "epoch": 446.41, "learning_rate": 5.001176470588235e-07, "loss": 0.0001, "step": 7600 }, { "epoch": 447.88, "learning_rate": 4.912941176470589e-07, "loss": 0.0001, "step": 7625 }, { "epoch": 449.35, "learning_rate": 4.824705882352941e-07, "loss": 0.0001, "step": 7650 }, { "epoch": 450.82, "learning_rate": 4.7364705882352946e-07, "loss": 0.0001, "step": 7675 }, { "epoch": 452.29, "learning_rate": 4.648235294117648e-07, "loss": 0.0001, "step": 7700 }, { "epoch": 453.76, "learning_rate": 4.56e-07, "loss": 0.0003, "step": 7725 }, { "epoch": 455.24, "learning_rate": 4.4717647058823534e-07, "loss": 0.0001, "step": 7750 }, { "epoch": 456.71, "learning_rate": 4.3835294117647056e-07, "loss": 0.0003, "step": 7775 }, { "epoch": 458.18, "learning_rate": 4.295294117647059e-07, "loss": 0.0001, "step": 7800 }, { "epoch": 459.65, "learning_rate": 4.207058823529412e-07, "loss": 0.0002, "step": 7825 }, { "epoch": 461.12, "learning_rate": 4.1188235294117644e-07, "loss": 0.0002, "step": 7850 }, { "epoch": 462.59, "learning_rate": 4.0305882352941177e-07, "loss": 0.0001, "step": 7875 }, { "epoch": 464.06, "learning_rate": 3.942352941176471e-07, "loss": 0.0002, "step": 7900 }, { "epoch": 465.53, "learning_rate": 3.8541176470588237e-07, "loss": 0.0002, "step": 7925 }, { "epoch": 467.0, "learning_rate": 3.765882352941177e-07, "loss": 0.0002, "step": 7950 }, { "epoch": 468.47, "learning_rate": 3.67764705882353e-07, "loss": 0.0003, "step": 7975 }, { "epoch": 469.94, "learning_rate": 3.5894117647058825e-07, "loss": 0.0001, "step": 8000 }, { "epoch": 469.94, "eval_loss": 0.56689453125, "eval_runtime": 163.3464, "eval_samples_per_second": 1.665, "eval_steps_per_second": 0.104, "eval_wer": 9.899702823179792, "step": 8000 }, { "epoch": 471.41, "learning_rate": 3.508235294117647e-07, "loss": 0.0002, "step": 8025 }, { "epoch": 472.88, "learning_rate": 3.42e-07, "loss": 0.0001, "step": 8050 }, { "epoch": 474.35, "learning_rate": 3.331764705882353e-07, "loss": 0.0001, "step": 8075 }, { "epoch": 475.82, "learning_rate": 3.2435294117647057e-07, "loss": 0.0002, "step": 8100 }, { "epoch": 477.29, "learning_rate": 3.1552941176470585e-07, "loss": 0.0002, "step": 8125 }, { "epoch": 478.76, "learning_rate": 3.067058823529412e-07, "loss": 0.0001, "step": 8150 }, { "epoch": 480.24, "learning_rate": 2.978823529411765e-07, "loss": 0.0001, "step": 8175 }, { "epoch": 481.71, "learning_rate": 2.890588235294118e-07, "loss": 0.0002, "step": 8200 }, { "epoch": 483.18, "learning_rate": 2.8023529411764706e-07, "loss": 0.0002, "step": 8225 }, { "epoch": 484.65, "learning_rate": 2.714117647058824e-07, "loss": 0.0001, "step": 8250 }, { "epoch": 486.12, "learning_rate": 2.6258823529411766e-07, "loss": 0.0003, "step": 8275 }, { "epoch": 487.59, "learning_rate": 2.5376470588235294e-07, "loss": 0.0001, "step": 8300 }, { "epoch": 489.06, "learning_rate": 2.449411764705882e-07, "loss": 0.0002, "step": 8325 }, { "epoch": 490.53, "learning_rate": 2.3611764705882354e-07, "loss": 0.0001, "step": 8350 }, { "epoch": 492.0, "learning_rate": 2.2729411764705884e-07, "loss": 0.0002, "step": 8375 }, { "epoch": 493.47, "learning_rate": 2.1847058823529412e-07, "loss": 0.0002, "step": 8400 }, { "epoch": 494.94, "learning_rate": 2.096470588235294e-07, "loss": 0.0002, "step": 8425 }, { "epoch": 496.41, "learning_rate": 2.0082352941176472e-07, "loss": 0.0003, "step": 8450 }, { "epoch": 497.88, "learning_rate": 1.92e-07, "loss": 0.0002, "step": 8475 }, { "epoch": 499.35, "learning_rate": 1.831764705882353e-07, "loss": 0.0002, "step": 8500 }, { "epoch": 500.82, "learning_rate": 1.743529411764706e-07, "loss": 0.0001, "step": 8525 }, { "epoch": 502.29, "learning_rate": 1.6552941176470588e-07, "loss": 0.0002, "step": 8550 }, { "epoch": 503.76, "learning_rate": 1.5670588235294118e-07, "loss": 0.0001, "step": 8575 }, { "epoch": 505.24, "learning_rate": 1.4788235294117646e-07, "loss": 0.0001, "step": 8600 }, { "epoch": 506.71, "learning_rate": 1.3905882352941179e-07, "loss": 0.0002, "step": 8625 }, { "epoch": 508.18, "learning_rate": 1.3023529411764706e-07, "loss": 0.0002, "step": 8650 }, { "epoch": 509.65, "learning_rate": 1.2141176470588236e-07, "loss": 0.0001, "step": 8675 }, { "epoch": 511.12, "learning_rate": 1.1258823529411764e-07, "loss": 0.0001, "step": 8700 }, { "epoch": 512.59, "learning_rate": 1.0376470588235296e-07, "loss": 0.0001, "step": 8725 }, { "epoch": 514.06, "learning_rate": 9.494117647058823e-08, "loss": 0.0001, "step": 8750 }, { "epoch": 515.53, "learning_rate": 8.611764705882352e-08, "loss": 0.0001, "step": 8775 }, { "epoch": 517.0, "learning_rate": 7.729411764705882e-08, "loss": 0.0001, "step": 8800 }, { "epoch": 518.47, "learning_rate": 6.847058823529411e-08, "loss": 0.0001, "step": 8825 }, { "epoch": 519.94, "learning_rate": 5.964705882352941e-08, "loss": 0.0001, "step": 8850 }, { "epoch": 521.41, "learning_rate": 5.08235294117647e-08, "loss": 0.0001, "step": 8875 }, { "epoch": 522.88, "learning_rate": 4.2e-08, "loss": 0.0001, "step": 8900 }, { "epoch": 524.35, "learning_rate": 3.3176470588235295e-08, "loss": 0.0001, "step": 8925 }, { "epoch": 525.82, "learning_rate": 2.435294117647059e-08, "loss": 0.0001, "step": 8950 }, { "epoch": 527.29, "learning_rate": 1.5529411764705883e-08, "loss": 0.0001, "step": 8975 }, { "epoch": 528.76, "learning_rate": 6.705882352941177e-09, "loss": 0.0001, "step": 9000 }, { "epoch": 528.76, "eval_loss": 0.564453125, "eval_runtime": 154.1953, "eval_samples_per_second": 1.764, "eval_steps_per_second": 0.11, "eval_wer": 9.927563150074294, "step": 9000 }, { "epoch": 530.47, "learning_rate": 3.145263157894737e-07, "loss": 0.0001, "step": 9025 }, { "epoch": 531.94, "learning_rate": 3.0663157894736844e-07, "loss": 0.0002, "step": 9050 }, { "epoch": 533.41, "learning_rate": 2.987368421052632e-07, "loss": 0.0002, "step": 9075 }, { "epoch": 534.88, "learning_rate": 2.908421052631579e-07, "loss": 0.0001, "step": 9100 }, { "epoch": 536.35, "learning_rate": 2.829473684210526e-07, "loss": 0.0001, "step": 9125 }, { "epoch": 537.82, "learning_rate": 2.7505263157894737e-07, "loss": 0.0003, "step": 9150 }, { "epoch": 539.29, "learning_rate": 2.671578947368421e-07, "loss": 0.0002, "step": 9175 }, { "epoch": 540.76, "learning_rate": 2.5926315789473686e-07, "loss": 0.0002, "step": 9200 }, { "epoch": 542.24, "learning_rate": 2.513684210526316e-07, "loss": 0.0001, "step": 9225 }, { "epoch": 543.71, "learning_rate": 2.4347368421052635e-07, "loss": 0.0002, "step": 9250 }, { "epoch": 545.18, "learning_rate": 2.3557894736842104e-07, "loss": 0.0001, "step": 9275 }, { "epoch": 546.65, "learning_rate": 2.276842105263158e-07, "loss": 0.0002, "step": 9300 }, { "epoch": 548.12, "learning_rate": 2.197894736842105e-07, "loss": 0.0002, "step": 9325 }, { "epoch": 549.59, "learning_rate": 2.1189473684210525e-07, "loss": 0.0001, "step": 9350 }, { "epoch": 551.06, "learning_rate": 2.0400000000000003e-07, "loss": 0.0001, "step": 9375 }, { "epoch": 552.53, "learning_rate": 1.9610526315789472e-07, "loss": 0.0001, "step": 9400 }, { "epoch": 554.0, "learning_rate": 1.882105263157895e-07, "loss": 0.0002, "step": 9425 }, { "epoch": 555.47, "learning_rate": 1.803157894736842e-07, "loss": 0.0002, "step": 9450 }, { "epoch": 556.94, "learning_rate": 1.7242105263157896e-07, "loss": 0.0001, "step": 9475 }, { "epoch": 558.41, "learning_rate": 1.6452631578947368e-07, "loss": 0.0002, "step": 9500 }, { "epoch": 559.88, "learning_rate": 1.5663157894736842e-07, "loss": 0.0001, "step": 9525 }, { "epoch": 561.35, "learning_rate": 1.4873684210526317e-07, "loss": 0.0001, "step": 9550 }, { "epoch": 562.82, "learning_rate": 1.408421052631579e-07, "loss": 0.0001, "step": 9575 }, { "epoch": 564.29, "learning_rate": 1.3294736842105263e-07, "loss": 0.0002, "step": 9600 }, { "epoch": 565.76, "learning_rate": 1.2505263157894738e-07, "loss": 0.0001, "step": 9625 }, { "epoch": 567.24, "learning_rate": 1.1715789473684211e-07, "loss": 0.0002, "step": 9650 }, { "epoch": 568.71, "learning_rate": 1.0926315789473684e-07, "loss": 0.0001, "step": 9675 }, { "epoch": 570.18, "learning_rate": 1.0136842105263159e-07, "loss": 0.0001, "step": 9700 }, { "epoch": 571.65, "learning_rate": 9.347368421052632e-08, "loss": 0.0001, "step": 9725 }, { "epoch": 573.12, "learning_rate": 8.557894736842105e-08, "loss": 0.0001, "step": 9750 }, { "epoch": 574.59, "learning_rate": 7.768421052631578e-08, "loss": 0.0002, "step": 9775 }, { "epoch": 576.06, "learning_rate": 6.978947368421053e-08, "loss": 0.0001, "step": 9800 }, { "epoch": 577.53, "learning_rate": 6.189473684210526e-08, "loss": 0.0002, "step": 9825 }, { "epoch": 579.0, "learning_rate": 5.4e-08, "loss": 0.0001, "step": 9850 }, { "epoch": 580.47, "learning_rate": 4.610526315789473e-08, "loss": 0.0001, "step": 9875 }, { "epoch": 581.94, "learning_rate": 3.821052631578947e-08, "loss": 0.0002, "step": 9900 }, { "epoch": 583.41, "learning_rate": 3.031578947368421e-08, "loss": 0.0001, "step": 9925 }, { "epoch": 584.88, "learning_rate": 2.2421052631578946e-08, "loss": 0.0002, "step": 9950 }, { "epoch": 586.35, "learning_rate": 1.4526315789473685e-08, "loss": 0.0001, "step": 9975 }, { "epoch": 587.82, "learning_rate": 6.631578947368421e-09, "loss": 0.0001, "step": 10000 }, { "epoch": 587.82, "eval_loss": 0.5673828125, "eval_runtime": 153.8197, "eval_samples_per_second": 1.768, "eval_steps_per_second": 0.111, "eval_wer": 9.964710252600298, "step": 10000 }, { "epoch": 589.47, "learning_rate": 2.8514285714285715e-07, "loss": 0.0002, "step": 10025 }, { "epoch": 590.94, "learning_rate": 2.78e-07, "loss": 0.0002, "step": 10050 }, { "epoch": 592.41, "learning_rate": 2.7085714285714285e-07, "loss": 0.0001, "step": 10075 }, { "epoch": 593.88, "learning_rate": 2.6371428571428574e-07, "loss": 0.0001, "step": 10100 }, { "epoch": 595.35, "learning_rate": 2.565714285714286e-07, "loss": 0.0001, "step": 10125 }, { "epoch": 596.82, "learning_rate": 2.4942857142857144e-07, "loss": 0.0001, "step": 10150 }, { "epoch": 598.29, "learning_rate": 2.4228571428571427e-07, "loss": 0.0002, "step": 10175 }, { "epoch": 599.76, "learning_rate": 2.3514285714285715e-07, "loss": 0.0002, "step": 10200 }, { "epoch": 601.24, "learning_rate": 2.28e-07, "loss": 0.0001, "step": 10225 }, { "epoch": 602.71, "learning_rate": 2.2085714285714288e-07, "loss": 0.0001, "step": 10250 }, { "epoch": 604.18, "learning_rate": 2.137142857142857e-07, "loss": 0.0001, "step": 10275 }, { "epoch": 605.65, "learning_rate": 2.0657142857142856e-07, "loss": 0.0001, "step": 10300 }, { "epoch": 607.12, "learning_rate": 1.9942857142857144e-07, "loss": 0.0001, "step": 10325 }, { "epoch": 608.59, "learning_rate": 1.922857142857143e-07, "loss": 0.0001, "step": 10350 }, { "epoch": 610.06, "learning_rate": 1.8514285714285715e-07, "loss": 0.0002, "step": 10375 }, { "epoch": 611.53, "learning_rate": 1.78e-07, "loss": 0.0001, "step": 10400 }, { "epoch": 613.0, "learning_rate": 1.7085714285714286e-07, "loss": 0.0003, "step": 10425 }, { "epoch": 614.47, "learning_rate": 1.637142857142857e-07, "loss": 0.0002, "step": 10450 }, { "epoch": 615.94, "learning_rate": 1.5657142857142857e-07, "loss": 0.0001, "step": 10475 }, { "epoch": 617.41, "learning_rate": 1.4942857142857145e-07, "loss": 0.0001, "step": 10500 }, { "epoch": 618.88, "learning_rate": 1.422857142857143e-07, "loss": 0.0002, "step": 10525 }, { "epoch": 620.35, "learning_rate": 1.3514285714285713e-07, "loss": 0.0001, "step": 10550 }, { "epoch": 621.82, "learning_rate": 1.28e-07, "loss": 0.0002, "step": 10575 }, { "epoch": 623.29, "learning_rate": 1.2085714285714286e-07, "loss": 0.0002, "step": 10600 }, { "epoch": 624.76, "learning_rate": 1.1371428571428573e-07, "loss": 0.0002, "step": 10625 }, { "epoch": 626.24, "learning_rate": 1.0657142857142858e-07, "loss": 0.0002, "step": 10650 }, { "epoch": 627.71, "learning_rate": 9.942857142857142e-08, "loss": 0.0002, "step": 10675 }, { "epoch": 629.18, "learning_rate": 9.228571428571429e-08, "loss": 0.0001, "step": 10700 }, { "epoch": 630.65, "learning_rate": 8.514285714285714e-08, "loss": 0.0001, "step": 10725 }, { "epoch": 632.12, "learning_rate": 7.8e-08, "loss": 0.0001, "step": 10750 }, { "epoch": 633.59, "learning_rate": 7.085714285714286e-08, "loss": 0.0002, "step": 10775 }, { "epoch": 635.06, "learning_rate": 6.371428571428572e-08, "loss": 0.0001, "step": 10800 }, { "epoch": 636.53, "learning_rate": 5.657142857142857e-08, "loss": 0.0001, "step": 10825 }, { "epoch": 638.0, "learning_rate": 4.942857142857143e-08, "loss": 0.0001, "step": 10850 }, { "epoch": 639.47, "learning_rate": 4.2285714285714285e-08, "loss": 0.0002, "step": 10875 }, { "epoch": 640.94, "learning_rate": 3.514285714285714e-08, "loss": 0.0001, "step": 10900 }, { "epoch": 642.41, "learning_rate": 2.8000000000000003e-08, "loss": 0.0001, "step": 10925 }, { "epoch": 643.88, "learning_rate": 2.0857142857142856e-08, "loss": 0.0001, "step": 10950 }, { "epoch": 645.35, "learning_rate": 1.3714285714285715e-08, "loss": 0.0001, "step": 10975 }, { "epoch": 646.82, "learning_rate": 6.571428571428572e-09, "loss": 0.0003, "step": 11000 }, { "epoch": 646.82, "eval_loss": 0.56689453125, "eval_runtime": 153.4429, "eval_samples_per_second": 1.773, "eval_steps_per_second": 0.111, "eval_wer": 9.946136701337295, "step": 11000 }, { "epoch": 646.82, "step": 11000, "total_flos": 3.388775639915157e+20, "train_loss": 1.3152035799893466e-05, "train_runtime": 3841.6705, "train_samples_per_second": 91.627, "train_steps_per_second": 2.863 } ], "max_steps": 11000, "num_train_epochs": 648, "total_flos": 3.388775639915157e+20, "trial_name": null, "trial_params": null }