| { | |
| "best_metric": 33.562978427279056, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 476.0004, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.1499999999999999e-08, | |
| "loss": 0.729, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.4e-08, | |
| "loss": 0.7262, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.6499999999999996e-08, | |
| "loss": 0.7071, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 4.9e-08, | |
| "loss": 0.7016, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 6.15e-08, | |
| "loss": 0.6788, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 7.399999999999999e-08, | |
| "loss": 0.654, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 8.649999999999999e-08, | |
| "loss": 0.6294, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 9.9e-08, | |
| "loss": 0.5968, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "learning_rate": 1.115e-07, | |
| "loss": 0.5725, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "learning_rate": 1.24e-07, | |
| "loss": 0.5483, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 1.365e-07, | |
| "loss": 0.522, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 1.49e-07, | |
| "loss": 0.5107, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "learning_rate": 1.615e-07, | |
| "loss": 0.4823, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "learning_rate": 1.7399999999999997e-07, | |
| "loss": 0.4793, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "learning_rate": 1.8649999999999998e-07, | |
| "loss": 0.4597, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "learning_rate": 1.99e-07, | |
| "loss": 0.4533, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 2.1149999999999998e-07, | |
| "loss": 0.4347, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "learning_rate": 2.24e-07, | |
| "loss": 0.4293, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "learning_rate": 2.3649999999999998e-07, | |
| "loss": 0.4133, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "learning_rate": 2.4899999999999997e-07, | |
| "loss": 0.406, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "learning_rate": 2.487222222222222e-07, | |
| "loss": 0.3914, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "learning_rate": 2.473333333333333e-07, | |
| "loss": 0.3806, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "learning_rate": 2.4594444444444445e-07, | |
| "loss": 0.3753, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "learning_rate": 2.4455555555555555e-07, | |
| "loss": 0.3656, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "learning_rate": 2.4316666666666665e-07, | |
| "loss": 0.359, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "learning_rate": 2.4177777777777775e-07, | |
| "loss": 0.3518, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "learning_rate": 2.403888888888889e-07, | |
| "loss": 0.3407, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "learning_rate": 2.3899999999999996e-07, | |
| "loss": 0.3351, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "learning_rate": 2.3761111111111108e-07, | |
| "loss": 0.3315, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "learning_rate": 2.362222222222222e-07, | |
| "loss": 0.3223, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "learning_rate": 2.3483333333333334e-07, | |
| "loss": 0.3137, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "learning_rate": 2.3344444444444444e-07, | |
| "loss": 0.3102, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "learning_rate": 2.3205555555555554e-07, | |
| "loss": 0.3054, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 2.3066666666666664e-07, | |
| "loss": 0.2968, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 83.0, | |
| "learning_rate": 2.2927777777777777e-07, | |
| "loss": 0.289, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 85.0, | |
| "learning_rate": 2.2788888888888887e-07, | |
| "loss": 0.2893, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 88.0, | |
| "learning_rate": 2.265e-07, | |
| "loss": 0.2802, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 90.0, | |
| "learning_rate": 2.2511111111111112e-07, | |
| "loss": 0.273, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 92.0, | |
| "learning_rate": 2.237222222222222e-07, | |
| "loss": 0.2741, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "learning_rate": 2.2233333333333333e-07, | |
| "loss": 0.2683, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 95.0, | |
| "eval_loss": 0.47103559970855713, | |
| "eval_runtime": 117.5444, | |
| "eval_samples_per_second": 6.381, | |
| "eval_steps_per_second": 0.204, | |
| "eval_wer": 33.562978427279056, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 97.0, | |
| "learning_rate": 2.2094444444444443e-07, | |
| "loss": 0.2598, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 99.0, | |
| "learning_rate": 2.1955555555555555e-07, | |
| "loss": 0.2568, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 102.0, | |
| "learning_rate": 2.1816666666666665e-07, | |
| "loss": 0.2498, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 104.0, | |
| "learning_rate": 2.1677777777777778e-07, | |
| "loss": 0.2462, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 107.0, | |
| "learning_rate": 2.1538888888888888e-07, | |
| "loss": 0.2401, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 109.0, | |
| "learning_rate": 2.1399999999999998e-07, | |
| "loss": 0.2404, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 111.0, | |
| "learning_rate": 2.126111111111111e-07, | |
| "loss": 0.2348, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 114.0, | |
| "learning_rate": 2.112222222222222e-07, | |
| "loss": 0.2308, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 116.0, | |
| "learning_rate": 2.0983333333333334e-07, | |
| "loss": 0.2281, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 119.0, | |
| "learning_rate": 2.0844444444444441e-07, | |
| "loss": 0.2232, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 121.0, | |
| "learning_rate": 2.0705555555555554e-07, | |
| "loss": 0.2216, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 123.0, | |
| "learning_rate": 2.0566666666666664e-07, | |
| "loss": 0.2144, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 126.0, | |
| "learning_rate": 2.0427777777777777e-07, | |
| "loss": 0.2125, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 128.0, | |
| "learning_rate": 2.028888888888889e-07, | |
| "loss": 0.2103, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 130.0, | |
| "learning_rate": 2.015e-07, | |
| "loss": 0.2039, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 133.0, | |
| "learning_rate": 2.001111111111111e-07, | |
| "loss": 0.2002, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 135.0, | |
| "learning_rate": 1.987222222222222e-07, | |
| "loss": 0.2015, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 138.0, | |
| "learning_rate": 1.9733333333333333e-07, | |
| "loss": 0.1954, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 140.0, | |
| "learning_rate": 1.9594444444444443e-07, | |
| "loss": 0.1929, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 142.0, | |
| "learning_rate": 1.9455555555555556e-07, | |
| "loss": 0.1893, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 145.0, | |
| "learning_rate": 1.9316666666666666e-07, | |
| "loss": 0.1873, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 147.0, | |
| "learning_rate": 1.9177777777777776e-07, | |
| "loss": 0.1841, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 149.0, | |
| "learning_rate": 1.9038888888888888e-07, | |
| "loss": 0.1817, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 152.0, | |
| "learning_rate": 1.8899999999999999e-07, | |
| "loss": 0.179, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 154.0, | |
| "learning_rate": 1.876111111111111e-07, | |
| "loss": 0.1749, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 157.0, | |
| "learning_rate": 1.8622222222222221e-07, | |
| "loss": 0.1722, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 159.0, | |
| "learning_rate": 1.8483333333333331e-07, | |
| "loss": 0.1696, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 161.0, | |
| "learning_rate": 1.8344444444444444e-07, | |
| "loss": 0.169, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 164.0, | |
| "learning_rate": 1.8205555555555554e-07, | |
| "loss": 0.1656, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 166.0, | |
| "learning_rate": 1.8066666666666667e-07, | |
| "loss": 0.1605, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 169.0, | |
| "learning_rate": 1.7927777777777777e-07, | |
| "loss": 0.1627, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 171.0, | |
| "learning_rate": 1.778888888888889e-07, | |
| "loss": 0.1574, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 173.0, | |
| "learning_rate": 1.7649999999999997e-07, | |
| "loss": 0.1552, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 176.0, | |
| "learning_rate": 1.751111111111111e-07, | |
| "loss": 0.1532, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 178.0, | |
| "learning_rate": 1.7372222222222223e-07, | |
| "loss": 0.1517, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 180.0, | |
| "learning_rate": 1.7233333333333333e-07, | |
| "loss": 0.1483, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 183.0, | |
| "learning_rate": 1.7094444444444446e-07, | |
| "loss": 0.147, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 185.0, | |
| "learning_rate": 1.6955555555555553e-07, | |
| "loss": 0.1458, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 188.0, | |
| "learning_rate": 1.6816666666666666e-07, | |
| "loss": 0.142, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "learning_rate": 1.6677777777777776e-07, | |
| "loss": 0.142, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 190.0, | |
| "eval_loss": 0.4714232385158539, | |
| "eval_runtime": 99.0664, | |
| "eval_samples_per_second": 7.571, | |
| "eval_steps_per_second": 0.242, | |
| "eval_wer": 33.83437717466945, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 192.0, | |
| "learning_rate": 1.6538888888888889e-07, | |
| "loss": 0.1394, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 195.0, | |
| "learning_rate": 1.64e-07, | |
| "loss": 0.1373, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 197.0, | |
| "learning_rate": 1.6261111111111111e-07, | |
| "loss": 0.1349, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 199.0, | |
| "learning_rate": 1.6122222222222221e-07, | |
| "loss": 0.1337, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 202.0, | |
| "learning_rate": 1.5983333333333332e-07, | |
| "loss": 0.1336, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 204.0, | |
| "learning_rate": 1.5844444444444444e-07, | |
| "loss": 0.1306, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 207.0, | |
| "learning_rate": 1.5705555555555554e-07, | |
| "loss": 0.1286, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 209.0, | |
| "learning_rate": 1.5566666666666667e-07, | |
| "loss": 0.1279, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 211.0, | |
| "learning_rate": 1.5427777777777777e-07, | |
| "loss": 0.124, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 214.0, | |
| "learning_rate": 1.5288888888888887e-07, | |
| "loss": 0.1244, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 216.0, | |
| "learning_rate": 1.515e-07, | |
| "loss": 0.1208, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 219.0, | |
| "learning_rate": 1.501111111111111e-07, | |
| "loss": 0.1203, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 221.0, | |
| "learning_rate": 1.4872222222222223e-07, | |
| "loss": 0.1183, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 223.0, | |
| "learning_rate": 1.4733333333333333e-07, | |
| "loss": 0.1177, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 226.0, | |
| "learning_rate": 1.4594444444444443e-07, | |
| "loss": 0.1175, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 228.0, | |
| "learning_rate": 1.4455555555555553e-07, | |
| "loss": 0.1135, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 230.0, | |
| "learning_rate": 1.4316666666666666e-07, | |
| "loss": 0.1138, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 233.0, | |
| "learning_rate": 1.4177777777777779e-07, | |
| "loss": 0.1115, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 235.0, | |
| "learning_rate": 1.4038888888888889e-07, | |
| "loss": 0.1111, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 238.0, | |
| "learning_rate": 1.3900000000000001e-07, | |
| "loss": 0.1094, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 240.0, | |
| "learning_rate": 1.376111111111111e-07, | |
| "loss": 0.1076, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 242.0, | |
| "learning_rate": 1.3622222222222222e-07, | |
| "loss": 0.1066, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 245.0, | |
| "learning_rate": 1.3483333333333332e-07, | |
| "loss": 0.1047, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 247.0, | |
| "learning_rate": 1.3344444444444444e-07, | |
| "loss": 0.1041, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 249.0, | |
| "learning_rate": 1.3205555555555557e-07, | |
| "loss": 0.1037, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 252.0, | |
| "learning_rate": 1.3066666666666665e-07, | |
| "loss": 0.1009, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 254.0, | |
| "learning_rate": 1.2927777777777777e-07, | |
| "loss": 0.1007, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 257.0, | |
| "learning_rate": 1.2788888888888887e-07, | |
| "loss": 0.0996, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 259.0, | |
| "learning_rate": 1.265e-07, | |
| "loss": 0.0979, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 261.0, | |
| "learning_rate": 1.251111111111111e-07, | |
| "loss": 0.0969, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 264.0, | |
| "learning_rate": 1.237222222222222e-07, | |
| "loss": 0.0966, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 266.0, | |
| "learning_rate": 1.2233333333333333e-07, | |
| "loss": 0.096, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 269.0, | |
| "learning_rate": 1.2094444444444443e-07, | |
| "loss": 0.0939, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 271.0, | |
| "learning_rate": 1.1955555555555556e-07, | |
| "loss": 0.0922, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 273.0, | |
| "learning_rate": 1.1816666666666666e-07, | |
| "loss": 0.0932, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 276.0, | |
| "learning_rate": 1.1677777777777776e-07, | |
| "loss": 0.0914, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 278.0, | |
| "learning_rate": 1.1538888888888889e-07, | |
| "loss": 0.0894, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 280.0, | |
| "learning_rate": 1.14e-07, | |
| "loss": 0.09, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 283.0, | |
| "learning_rate": 1.126111111111111e-07, | |
| "loss": 0.0887, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 285.0, | |
| "learning_rate": 1.1122222222222222e-07, | |
| "loss": 0.0871, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 285.0, | |
| "eval_loss": 0.4781545102596283, | |
| "eval_runtime": 100.2584, | |
| "eval_samples_per_second": 7.481, | |
| "eval_steps_per_second": 0.239, | |
| "eval_wer": 33.95963813500348, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 288.0, | |
| "learning_rate": 1.0983333333333333e-07, | |
| "loss": 0.0874, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 290.0, | |
| "learning_rate": 1.0844444444444443e-07, | |
| "loss": 0.0862, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 292.0, | |
| "learning_rate": 1.0705555555555555e-07, | |
| "loss": 0.0849, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 295.0, | |
| "learning_rate": 1.0566666666666667e-07, | |
| "loss": 0.0841, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 297.0, | |
| "learning_rate": 1.0427777777777777e-07, | |
| "loss": 0.0836, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 299.0, | |
| "learning_rate": 1.0288888888888889e-07, | |
| "loss": 0.0824, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 302.0, | |
| "learning_rate": 1.015e-07, | |
| "loss": 0.0825, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 304.0, | |
| "learning_rate": 1.001111111111111e-07, | |
| "loss": 0.0811, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 307.0, | |
| "learning_rate": 9.872222222222222e-08, | |
| "loss": 0.0804, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 309.0, | |
| "learning_rate": 9.733333333333332e-08, | |
| "loss": 0.0804, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 311.0, | |
| "learning_rate": 9.594444444444445e-08, | |
| "loss": 0.0799, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 314.0, | |
| "learning_rate": 9.455555555555556e-08, | |
| "loss": 0.0784, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 316.0, | |
| "learning_rate": 9.316666666666666e-08, | |
| "loss": 0.0778, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 319.0, | |
| "learning_rate": 9.177777777777777e-08, | |
| "loss": 0.0762, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 321.0, | |
| "learning_rate": 9.038888888888889e-08, | |
| "loss": 0.077, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 323.0, | |
| "learning_rate": 8.899999999999999e-08, | |
| "loss": 0.0761, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 326.0, | |
| "learning_rate": 8.76111111111111e-08, | |
| "loss": 0.0755, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 328.0, | |
| "learning_rate": 8.622222222222222e-08, | |
| "loss": 0.0753, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 330.0, | |
| "learning_rate": 8.483333333333333e-08, | |
| "loss": 0.074, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 333.0, | |
| "learning_rate": 8.344444444444445e-08, | |
| "loss": 0.0736, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 335.0, | |
| "learning_rate": 8.205555555555556e-08, | |
| "loss": 0.0731, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 338.0, | |
| "learning_rate": 8.066666666666666e-08, | |
| "loss": 0.0728, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 340.0, | |
| "learning_rate": 7.927777777777778e-08, | |
| "loss": 0.0719, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 342.0, | |
| "learning_rate": 7.788888888888888e-08, | |
| "loss": 0.0721, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 345.0, | |
| "learning_rate": 7.649999999999999e-08, | |
| "loss": 0.0719, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 347.0, | |
| "learning_rate": 7.511111111111112e-08, | |
| "loss": 0.0701, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 349.0, | |
| "learning_rate": 7.372222222222222e-08, | |
| "loss": 0.0705, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 352.0, | |
| "learning_rate": 7.233333333333333e-08, | |
| "loss": 0.0703, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 354.0, | |
| "learning_rate": 7.094444444444445e-08, | |
| "loss": 0.0689, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 357.0, | |
| "learning_rate": 6.955555555555555e-08, | |
| "loss": 0.0694, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 359.0, | |
| "learning_rate": 6.816666666666666e-08, | |
| "loss": 0.0685, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 361.0, | |
| "learning_rate": 6.677777777777778e-08, | |
| "loss": 0.0681, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 364.0, | |
| "learning_rate": 6.538888888888888e-08, | |
| "loss": 0.0674, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 366.0, | |
| "learning_rate": 6.4e-08, | |
| "loss": 0.0671, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 369.0, | |
| "learning_rate": 6.261111111111112e-08, | |
| "loss": 0.0677, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 371.0, | |
| "learning_rate": 6.122222222222222e-08, | |
| "loss": 0.0666, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 373.0, | |
| "learning_rate": 5.983333333333333e-08, | |
| "loss": 0.0658, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 376.0, | |
| "learning_rate": 5.844444444444444e-08, | |
| "loss": 0.0656, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 378.0, | |
| "learning_rate": 5.705555555555555e-08, | |
| "loss": 0.0652, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "learning_rate": 5.566666666666667e-08, | |
| "loss": 0.0656, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 380.0, | |
| "eval_loss": 0.4829937517642975, | |
| "eval_runtime": 96.1124, | |
| "eval_samples_per_second": 7.803, | |
| "eval_steps_per_second": 0.25, | |
| "eval_wer": 33.72303409881698, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 383.0, | |
| "learning_rate": 5.4277777777777776e-08, | |
| "loss": 0.0645, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 385.0, | |
| "learning_rate": 5.2888888888888884e-08, | |
| "loss": 0.0647, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 388.0, | |
| "learning_rate": 5.15e-08, | |
| "loss": 0.0651, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 390.0, | |
| "learning_rate": 5.011111111111111e-08, | |
| "loss": 0.0642, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 392.0, | |
| "learning_rate": 4.872222222222222e-08, | |
| "loss": 0.0641, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 395.0, | |
| "learning_rate": 4.733333333333333e-08, | |
| "loss": 0.0634, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 397.0, | |
| "learning_rate": 4.594444444444445e-08, | |
| "loss": 0.063, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 399.0, | |
| "learning_rate": 4.4555555555555555e-08, | |
| "loss": 0.0628, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 402.0, | |
| "learning_rate": 4.316666666666666e-08, | |
| "loss": 0.0631, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 404.0, | |
| "learning_rate": 4.177777777777777e-08, | |
| "loss": 0.0622, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 407.0, | |
| "learning_rate": 4.038888888888889e-08, | |
| "loss": 0.0617, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 409.0, | |
| "learning_rate": 3.9e-08, | |
| "loss": 0.0619, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 411.0, | |
| "learning_rate": 3.7611111111111106e-08, | |
| "loss": 0.0622, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 414.0, | |
| "learning_rate": 3.622222222222223e-08, | |
| "loss": 0.0617, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 416.0, | |
| "learning_rate": 3.4833333333333334e-08, | |
| "loss": 0.0615, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 419.0, | |
| "learning_rate": 3.344444444444444e-08, | |
| "loss": 0.0611, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 421.0, | |
| "learning_rate": 3.205555555555555e-08, | |
| "loss": 0.0615, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 423.0, | |
| "learning_rate": 3.0666666666666663e-08, | |
| "loss": 0.0609, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 426.0, | |
| "learning_rate": 2.9277777777777777e-08, | |
| "loss": 0.061, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 428.0, | |
| "learning_rate": 2.7888888888888888e-08, | |
| "loss": 0.0602, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 430.0, | |
| "learning_rate": 2.65e-08, | |
| "loss": 0.0606, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 433.0, | |
| "learning_rate": 2.511111111111111e-08, | |
| "loss": 0.0595, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 435.0, | |
| "learning_rate": 2.372222222222222e-08, | |
| "loss": 0.0608, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 438.0, | |
| "learning_rate": 2.233333333333333e-08, | |
| "loss": 0.0599, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 440.0, | |
| "learning_rate": 2.0944444444444446e-08, | |
| "loss": 0.0602, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 442.0, | |
| "learning_rate": 1.9555555555555553e-08, | |
| "loss": 0.0593, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 445.0, | |
| "learning_rate": 1.8166666666666667e-08, | |
| "loss": 0.0597, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 447.0, | |
| "learning_rate": 1.6777777777777775e-08, | |
| "loss": 0.0592, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 449.0, | |
| "learning_rate": 1.538888888888889e-08, | |
| "loss": 0.059, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 452.0, | |
| "learning_rate": 1.4e-08, | |
| "loss": 0.0591, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 454.0, | |
| "learning_rate": 1.261111111111111e-08, | |
| "loss": 0.0598, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 457.0, | |
| "learning_rate": 1.1222222222222221e-08, | |
| "loss": 0.0597, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 459.0, | |
| "learning_rate": 9.833333333333332e-09, | |
| "loss": 0.0588, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 461.0, | |
| "learning_rate": 8.444444444444443e-09, | |
| "loss": 0.0598, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 464.0, | |
| "learning_rate": 7.055555555555555e-09, | |
| "loss": 0.0588, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 466.0, | |
| "learning_rate": 5.666666666666667e-09, | |
| "loss": 0.0596, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 469.0, | |
| "learning_rate": 4.277777777777778e-09, | |
| "loss": 0.0579, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 471.0, | |
| "learning_rate": 2.8888888888888886e-09, | |
| "loss": 0.0596, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 473.0, | |
| "learning_rate": 1.5e-09, | |
| "loss": 0.0588, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "learning_rate": 1.1111111111111111e-10, | |
| "loss": 0.0595, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "eval_loss": 0.4846940040588379, | |
| "eval_runtime": 110.8316, | |
| "eval_samples_per_second": 6.767, | |
| "eval_steps_per_second": 0.217, | |
| "eval_wer": 33.7160751565762, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 476.0, | |
| "step": 5000, | |
| "total_flos": 2.004511153324032e+19, | |
| "train_loss": 0.17557046995162964, | |
| "train_runtime": 18026.516, | |
| "train_samples_per_second": 17.752, | |
| "train_steps_per_second": 0.277 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 2.004511153324032e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |