{ "best_metric": 0.149032992036405, "best_model_checkpoint": "./impaired-v3-independent-mild/checkpoint-2000", "epoch": 294.11764705882354, "eval_steps": 1000, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.47, "learning_rate": 5.000000000000001e-07, "loss": 0.4262, "step": 25 }, { "epoch": 2.94, "learning_rate": 1.0000000000000002e-06, "loss": 0.3162, "step": 50 }, { "epoch": 4.41, "learning_rate": 1.5e-06, "loss": 0.154, "step": 75 }, { "epoch": 5.88, "learning_rate": 2.0000000000000003e-06, "loss": 0.0959, "step": 100 }, { "epoch": 7.35, "learning_rate": 2.5e-06, "loss": 0.0457, "step": 125 }, { "epoch": 8.82, "learning_rate": 3e-06, "loss": 0.019, "step": 150 }, { "epoch": 10.29, "learning_rate": 3.5e-06, "loss": 0.0061, "step": 175 }, { "epoch": 11.76, "learning_rate": 4.000000000000001e-06, "loss": 0.0039, "step": 200 }, { "epoch": 13.24, "learning_rate": 4.5e-06, "loss": 0.0054, "step": 225 }, { "epoch": 14.71, "learning_rate": 5e-06, "loss": 0.0048, "step": 250 }, { "epoch": 16.18, "learning_rate": 5.500000000000001e-06, "loss": 0.0037, "step": 275 }, { "epoch": 17.65, "learning_rate": 6e-06, "loss": 0.0039, "step": 300 }, { "epoch": 19.12, "learning_rate": 6.5000000000000004e-06, "loss": 0.0038, "step": 325 }, { "epoch": 20.59, "learning_rate": 7e-06, "loss": 0.0034, "step": 350 }, { "epoch": 22.06, "learning_rate": 7.500000000000001e-06, "loss": 0.0043, "step": 375 }, { "epoch": 23.53, "learning_rate": 8.000000000000001e-06, "loss": 0.0035, "step": 400 }, { "epoch": 25.0, "learning_rate": 8.5e-06, "loss": 0.0061, "step": 425 }, { "epoch": 26.47, "learning_rate": 9e-06, "loss": 0.0056, "step": 450 }, { "epoch": 27.94, "learning_rate": 9.5e-06, "loss": 0.0063, "step": 475 }, { "epoch": 29.41, "learning_rate": 1e-05, "loss": 0.0073, "step": 500 }, { "epoch": 30.88, "learning_rate": 9.944444444444445e-06, "loss": 0.0079, "step": 525 }, { "epoch": 32.35, "learning_rate": 9.88888888888889e-06, "loss": 0.005, "step": 550 }, { "epoch": 33.82, "learning_rate": 9.833333333333333e-06, "loss": 0.0038, "step": 575 }, { "epoch": 35.29, "learning_rate": 9.777777777777779e-06, "loss": 0.0027, "step": 600 }, { "epoch": 36.76, "learning_rate": 9.722222222222223e-06, "loss": 0.0024, "step": 625 }, { "epoch": 38.24, "learning_rate": 9.666666666666667e-06, "loss": 0.0012, "step": 650 }, { "epoch": 39.71, "learning_rate": 9.611111111111112e-06, "loss": 0.0003, "step": 675 }, { "epoch": 41.18, "learning_rate": 9.555555555555556e-06, "loss": 0.0004, "step": 700 }, { "epoch": 42.65, "learning_rate": 9.5e-06, "loss": 0.0003, "step": 725 }, { "epoch": 44.12, "learning_rate": 9.444444444444445e-06, "loss": 0.0002, "step": 750 }, { "epoch": 45.59, "learning_rate": 9.38888888888889e-06, "loss": 0.0002, "step": 775 }, { "epoch": 47.06, "learning_rate": 9.333333333333334e-06, "loss": 0.0002, "step": 800 }, { "epoch": 48.53, "learning_rate": 9.277777777777778e-06, "loss": 0.0002, "step": 825 }, { "epoch": 50.0, "learning_rate": 9.222222222222224e-06, "loss": 0.0001, "step": 850 }, { "epoch": 51.47, "learning_rate": 9.166666666666666e-06, "loss": 0.0001, "step": 875 }, { "epoch": 52.94, "learning_rate": 9.111111111111112e-06, "loss": 0.0001, "step": 900 }, { "epoch": 54.41, "learning_rate": 9.055555555555556e-06, "loss": 0.0001, "step": 925 }, { "epoch": 55.88, "learning_rate": 9e-06, "loss": 0.0001, "step": 950 }, { "epoch": 57.35, "learning_rate": 8.944444444444446e-06, "loss": 0.0001, "step": 975 }, { "epoch": 58.82, "learning_rate": 8.888888888888888e-06, "loss": 0.0001, "step": 1000 }, { "epoch": 58.82, "eval_loss": 0.438310444355011, "eval_runtime": 63.0739, "eval_samples_per_second": 1.015, "eval_steps_per_second": 0.127, "eval_wer": 0.14789533560864618, "step": 1000 }, { "epoch": 60.29, "learning_rate": 8.833333333333334e-06, "loss": 0.0001, "step": 1025 }, { "epoch": 61.76, "learning_rate": 8.777777777777778e-06, "loss": 0.0001, "step": 1050 }, { "epoch": 63.24, "learning_rate": 8.722222222222224e-06, "loss": 0.0001, "step": 1075 }, { "epoch": 64.71, "learning_rate": 8.666666666666668e-06, "loss": 0.0001, "step": 1100 }, { "epoch": 66.18, "learning_rate": 8.611111111111112e-06, "loss": 0.0001, "step": 1125 }, { "epoch": 67.65, "learning_rate": 8.555555555555556e-06, "loss": 0.0001, "step": 1150 }, { "epoch": 69.12, "learning_rate": 8.5e-06, "loss": 0.0001, "step": 1175 }, { "epoch": 70.59, "learning_rate": 8.444444444444446e-06, "loss": 0.0001, "step": 1200 }, { "epoch": 72.06, "learning_rate": 8.38888888888889e-06, "loss": 0.0001, "step": 1225 }, { "epoch": 73.53, "learning_rate": 8.333333333333334e-06, "loss": 0.0001, "step": 1250 }, { "epoch": 75.0, "learning_rate": 8.277777777777778e-06, "loss": 0.0001, "step": 1275 }, { "epoch": 76.47, "learning_rate": 8.222222222222222e-06, "loss": 0.0001, "step": 1300 }, { "epoch": 77.94, "learning_rate": 8.166666666666668e-06, "loss": 0.0001, "step": 1325 }, { "epoch": 79.41, "learning_rate": 8.111111111111112e-06, "loss": 0.0001, "step": 1350 }, { "epoch": 80.88, "learning_rate": 8.055555555555557e-06, "loss": 0.0001, "step": 1375 }, { "epoch": 82.35, "learning_rate": 8.000000000000001e-06, "loss": 0.0, "step": 1400 }, { "epoch": 83.82, "learning_rate": 7.944444444444445e-06, "loss": 0.0, "step": 1425 }, { "epoch": 85.29, "learning_rate": 7.88888888888889e-06, "loss": 0.0, "step": 1450 }, { "epoch": 86.76, "learning_rate": 7.833333333333333e-06, "loss": 0.0, "step": 1475 }, { "epoch": 88.24, "learning_rate": 7.77777777777778e-06, "loss": 0.0, "step": 1500 }, { "epoch": 89.71, "learning_rate": 7.722222222222223e-06, "loss": 0.0, "step": 1525 }, { "epoch": 91.18, "learning_rate": 7.666666666666667e-06, "loss": 0.0, "step": 1550 }, { "epoch": 92.65, "learning_rate": 7.611111111111111e-06, "loss": 0.0, "step": 1575 }, { "epoch": 94.12, "learning_rate": 7.555555555555556e-06, "loss": 0.0, "step": 1600 }, { "epoch": 95.59, "learning_rate": 7.500000000000001e-06, "loss": 0.0, "step": 1625 }, { "epoch": 97.06, "learning_rate": 7.444444444444445e-06, "loss": 0.0, "step": 1650 }, { "epoch": 98.53, "learning_rate": 7.38888888888889e-06, "loss": 0.0, "step": 1675 }, { "epoch": 100.0, "learning_rate": 7.333333333333333e-06, "loss": 0.0, "step": 1700 }, { "epoch": 101.47, "learning_rate": 7.277777777777778e-06, "loss": 0.0, "step": 1725 }, { "epoch": 102.94, "learning_rate": 7.222222222222223e-06, "loss": 0.0, "step": 1750 }, { "epoch": 104.41, "learning_rate": 7.166666666666667e-06, "loss": 0.0, "step": 1775 }, { "epoch": 105.88, "learning_rate": 7.111111111111112e-06, "loss": 0.0, "step": 1800 }, { "epoch": 107.35, "learning_rate": 7.055555555555557e-06, "loss": 0.0, "step": 1825 }, { "epoch": 108.82, "learning_rate": 7e-06, "loss": 0.0, "step": 1850 }, { "epoch": 110.29, "learning_rate": 6.944444444444445e-06, "loss": 0.0, "step": 1875 }, { "epoch": 111.76, "learning_rate": 6.88888888888889e-06, "loss": 0.0, "step": 1900 }, { "epoch": 113.24, "learning_rate": 6.833333333333334e-06, "loss": 0.0, "step": 1925 }, { "epoch": 114.71, "learning_rate": 6.777777777777779e-06, "loss": 0.0, "step": 1950 }, { "epoch": 116.18, "learning_rate": 6.7222222222222235e-06, "loss": 0.0, "step": 1975 }, { "epoch": 117.65, "learning_rate": 6.666666666666667e-06, "loss": 0.0, "step": 2000 }, { "epoch": 117.65, "eval_loss": 0.48648372292518616, "eval_runtime": 61.9265, "eval_samples_per_second": 1.033, "eval_steps_per_second": 0.129, "eval_wer": 0.149032992036405, "step": 2000 }, { "epoch": 119.12, "learning_rate": 6.6111111111111115e-06, "loss": 0.0, "step": 2025 }, { "epoch": 120.59, "learning_rate": 6.555555555555556e-06, "loss": 0.0, "step": 2050 }, { "epoch": 122.06, "learning_rate": 6.5000000000000004e-06, "loss": 0.0, "step": 2075 }, { "epoch": 123.53, "learning_rate": 6.444444444444445e-06, "loss": 0.0, "step": 2100 }, { "epoch": 125.0, "learning_rate": 6.3888888888888885e-06, "loss": 0.0, "step": 2125 }, { "epoch": 126.47, "learning_rate": 6.333333333333333e-06, "loss": 0.0, "step": 2150 }, { "epoch": 127.94, "learning_rate": 6.277777777777778e-06, "loss": 0.0, "step": 2175 }, { "epoch": 129.41, "learning_rate": 6.222222222222223e-06, "loss": 0.0, "step": 2200 }, { "epoch": 130.88, "learning_rate": 6.166666666666667e-06, "loss": 0.0, "step": 2225 }, { "epoch": 132.35, "learning_rate": 6.111111111111112e-06, "loss": 0.0, "step": 2250 }, { "epoch": 133.82, "learning_rate": 6.055555555555555e-06, "loss": 0.0, "step": 2275 }, { "epoch": 135.29, "learning_rate": 6e-06, "loss": 0.0, "step": 2300 }, { "epoch": 136.76, "learning_rate": 5.944444444444445e-06, "loss": 0.0, "step": 2325 }, { "epoch": 138.24, "learning_rate": 5.88888888888889e-06, "loss": 0.0, "step": 2350 }, { "epoch": 139.71, "learning_rate": 5.833333333333334e-06, "loss": 0.0, "step": 2375 }, { "epoch": 141.18, "learning_rate": 5.777777777777778e-06, "loss": 0.0, "step": 2400 }, { "epoch": 142.65, "learning_rate": 5.722222222222222e-06, "loss": 0.0, "step": 2425 }, { "epoch": 144.12, "learning_rate": 5.666666666666667e-06, "loss": 0.0, "step": 2450 }, { "epoch": 145.59, "learning_rate": 5.611111111111112e-06, "loss": 0.0, "step": 2475 }, { "epoch": 147.06, "learning_rate": 5.555555555555557e-06, "loss": 0.0, "step": 2500 }, { "epoch": 148.53, "learning_rate": 5.500000000000001e-06, "loss": 0.0, "step": 2525 }, { "epoch": 150.0, "learning_rate": 5.444444444444445e-06, "loss": 0.0, "step": 2550 }, { "epoch": 151.47, "learning_rate": 5.388888888888889e-06, "loss": 0.0, "step": 2575 }, { "epoch": 152.94, "learning_rate": 5.333333333333334e-06, "loss": 0.0, "step": 2600 }, { "epoch": 154.41, "learning_rate": 5.2777777777777785e-06, "loss": 0.0, "step": 2625 }, { "epoch": 155.88, "learning_rate": 5.2222222222222226e-06, "loss": 0.0, "step": 2650 }, { "epoch": 157.35, "learning_rate": 5.1666666666666675e-06, "loss": 0.0, "step": 2675 }, { "epoch": 158.82, "learning_rate": 5.1111111111111115e-06, "loss": 0.0, "step": 2700 }, { "epoch": 160.29, "learning_rate": 5.0555555555555555e-06, "loss": 0.0, "step": 2725 }, { "epoch": 161.76, "learning_rate": 5e-06, "loss": 0.0, "step": 2750 }, { "epoch": 163.24, "learning_rate": 4.944444444444445e-06, "loss": 0.0, "step": 2775 }, { "epoch": 164.71, "learning_rate": 4.888888888888889e-06, "loss": 0.0, "step": 2800 }, { "epoch": 166.18, "learning_rate": 4.833333333333333e-06, "loss": 0.0, "step": 2825 }, { "epoch": 167.65, "learning_rate": 4.777777777777778e-06, "loss": 0.0, "step": 2850 }, { "epoch": 169.12, "learning_rate": 4.722222222222222e-06, "loss": 0.0, "step": 2875 }, { "epoch": 170.59, "learning_rate": 4.666666666666667e-06, "loss": 0.0, "step": 2900 }, { "epoch": 172.06, "learning_rate": 4.611111111111112e-06, "loss": 0.0, "step": 2925 }, { "epoch": 173.53, "learning_rate": 4.555555555555556e-06, "loss": 0.0, "step": 2950 }, { "epoch": 175.0, "learning_rate": 4.5e-06, "loss": 0.0, "step": 2975 }, { "epoch": 176.47, "learning_rate": 4.444444444444444e-06, "loss": 0.0, "step": 3000 }, { "epoch": 176.47, "eval_loss": 0.5102497935295105, "eval_runtime": 62.0332, "eval_samples_per_second": 1.032, "eval_steps_per_second": 0.129, "eval_wer": 0.14789533560864618, "step": 3000 }, { "epoch": 177.94, "learning_rate": 4.388888888888889e-06, "loss": 0.0, "step": 3025 }, { "epoch": 179.41, "learning_rate": 4.333333333333334e-06, "loss": 0.0, "step": 3050 }, { "epoch": 180.88, "learning_rate": 4.277777777777778e-06, "loss": 0.0, "step": 3075 }, { "epoch": 182.35, "learning_rate": 4.222222222222223e-06, "loss": 0.0, "step": 3100 }, { "epoch": 183.82, "learning_rate": 4.166666666666667e-06, "loss": 0.0, "step": 3125 }, { "epoch": 185.29, "learning_rate": 4.111111111111111e-06, "loss": 0.0, "step": 3150 }, { "epoch": 186.76, "learning_rate": 4.055555555555556e-06, "loss": 0.0, "step": 3175 }, { "epoch": 188.24, "learning_rate": 4.000000000000001e-06, "loss": 0.0, "step": 3200 }, { "epoch": 189.71, "learning_rate": 3.944444444444445e-06, "loss": 0.0, "step": 3225 }, { "epoch": 191.18, "learning_rate": 3.88888888888889e-06, "loss": 0.0, "step": 3250 }, { "epoch": 192.65, "learning_rate": 3.833333333333334e-06, "loss": 0.0, "step": 3275 }, { "epoch": 194.12, "learning_rate": 3.777777777777778e-06, "loss": 0.0, "step": 3300 }, { "epoch": 195.59, "learning_rate": 3.7222222222222225e-06, "loss": 0.0, "step": 3325 }, { "epoch": 197.06, "learning_rate": 3.6666666666666666e-06, "loss": 0.0, "step": 3350 }, { "epoch": 198.53, "learning_rate": 3.6111111111111115e-06, "loss": 0.0, "step": 3375 }, { "epoch": 200.0, "learning_rate": 3.555555555555556e-06, "loss": 0.0, "step": 3400 }, { "epoch": 201.47, "learning_rate": 3.5e-06, "loss": 0.0, "step": 3425 }, { "epoch": 202.94, "learning_rate": 3.444444444444445e-06, "loss": 0.0, "step": 3450 }, { "epoch": 204.41, "learning_rate": 3.3888888888888893e-06, "loss": 0.0, "step": 3475 }, { "epoch": 205.88, "learning_rate": 3.3333333333333333e-06, "loss": 0.0, "step": 3500 }, { "epoch": 207.35, "learning_rate": 3.277777777777778e-06, "loss": 0.0, "step": 3525 }, { "epoch": 208.82, "learning_rate": 3.2222222222222227e-06, "loss": 0.0, "step": 3550 }, { "epoch": 210.29, "learning_rate": 3.1666666666666667e-06, "loss": 0.0, "step": 3575 }, { "epoch": 211.76, "learning_rate": 3.1111111111111116e-06, "loss": 0.0, "step": 3600 }, { "epoch": 213.24, "learning_rate": 3.055555555555556e-06, "loss": 0.0, "step": 3625 }, { "epoch": 214.71, "learning_rate": 3e-06, "loss": 0.0, "step": 3650 }, { "epoch": 216.18, "learning_rate": 2.944444444444445e-06, "loss": 0.0, "step": 3675 }, { "epoch": 217.65, "learning_rate": 2.888888888888889e-06, "loss": 0.0, "step": 3700 }, { "epoch": 219.12, "learning_rate": 2.8333333333333335e-06, "loss": 0.0, "step": 3725 }, { "epoch": 220.59, "learning_rate": 2.7777777777777783e-06, "loss": 0.0, "step": 3750 }, { "epoch": 222.06, "learning_rate": 2.7222222222222224e-06, "loss": 0.0, "step": 3775 }, { "epoch": 223.53, "learning_rate": 2.666666666666667e-06, "loss": 0.0, "step": 3800 }, { "epoch": 225.0, "learning_rate": 2.6111111111111113e-06, "loss": 0.0, "step": 3825 }, { "epoch": 226.47, "learning_rate": 2.5555555555555557e-06, "loss": 0.0, "step": 3850 }, { "epoch": 227.94, "learning_rate": 2.5e-06, "loss": 0.0, "step": 3875 }, { "epoch": 229.41, "learning_rate": 2.4444444444444447e-06, "loss": 0.0, "step": 3900 }, { "epoch": 230.88, "learning_rate": 2.388888888888889e-06, "loss": 0.0, "step": 3925 }, { "epoch": 232.35, "learning_rate": 2.3333333333333336e-06, "loss": 0.0, "step": 3950 }, { "epoch": 233.82, "learning_rate": 2.277777777777778e-06, "loss": 0.0, "step": 3975 }, { "epoch": 235.29, "learning_rate": 2.222222222222222e-06, "loss": 0.0, "step": 4000 }, { "epoch": 235.29, "eval_loss": 0.5246536135673523, "eval_runtime": 62.0964, "eval_samples_per_second": 1.031, "eval_steps_per_second": 0.129, "eval_wer": 0.14562002275312855, "step": 4000 }, { "epoch": 236.76, "learning_rate": 2.166666666666667e-06, "loss": 0.0, "step": 4025 }, { "epoch": 238.24, "learning_rate": 2.1111111111111114e-06, "loss": 0.0, "step": 4050 }, { "epoch": 239.71, "learning_rate": 2.0555555555555555e-06, "loss": 0.0, "step": 4075 }, { "epoch": 241.18, "learning_rate": 2.0000000000000003e-06, "loss": 0.0, "step": 4100 }, { "epoch": 242.65, "learning_rate": 1.944444444444445e-06, "loss": 0.0, "step": 4125 }, { "epoch": 244.12, "learning_rate": 1.888888888888889e-06, "loss": 0.0, "step": 4150 }, { "epoch": 245.59, "learning_rate": 1.8333333333333333e-06, "loss": 0.0, "step": 4175 }, { "epoch": 247.06, "learning_rate": 1.777777777777778e-06, "loss": 0.0, "step": 4200 }, { "epoch": 248.53, "learning_rate": 1.7222222222222224e-06, "loss": 0.0, "step": 4225 }, { "epoch": 250.0, "learning_rate": 1.6666666666666667e-06, "loss": 0.0, "step": 4250 }, { "epoch": 251.47, "learning_rate": 1.6111111111111113e-06, "loss": 0.0, "step": 4275 }, { "epoch": 252.94, "learning_rate": 1.5555555555555558e-06, "loss": 0.0, "step": 4300 }, { "epoch": 254.41, "learning_rate": 1.5e-06, "loss": 0.0, "step": 4325 }, { "epoch": 255.88, "learning_rate": 1.4444444444444445e-06, "loss": 0.0, "step": 4350 }, { "epoch": 257.35, "learning_rate": 1.3888888888888892e-06, "loss": 0.0, "step": 4375 }, { "epoch": 258.82, "learning_rate": 1.3333333333333334e-06, "loss": 0.0, "step": 4400 }, { "epoch": 260.29, "learning_rate": 1.2777777777777779e-06, "loss": 0.0, "step": 4425 }, { "epoch": 261.76, "learning_rate": 1.2222222222222223e-06, "loss": 0.0, "step": 4450 }, { "epoch": 263.24, "learning_rate": 1.1666666666666668e-06, "loss": 0.0, "step": 4475 }, { "epoch": 264.71, "learning_rate": 1.111111111111111e-06, "loss": 0.0, "step": 4500 }, { "epoch": 266.18, "learning_rate": 1.0555555555555557e-06, "loss": 0.0, "step": 4525 }, { "epoch": 267.65, "learning_rate": 1.0000000000000002e-06, "loss": 0.0, "step": 4550 }, { "epoch": 269.12, "learning_rate": 9.444444444444445e-07, "loss": 0.0, "step": 4575 }, { "epoch": 270.59, "learning_rate": 8.88888888888889e-07, "loss": 0.0, "step": 4600 }, { "epoch": 272.06, "learning_rate": 8.333333333333333e-07, "loss": 0.0, "step": 4625 }, { "epoch": 273.53, "learning_rate": 7.777777777777779e-07, "loss": 0.0, "step": 4650 }, { "epoch": 275.0, "learning_rate": 7.222222222222222e-07, "loss": 0.0, "step": 4675 }, { "epoch": 276.47, "learning_rate": 6.666666666666667e-07, "loss": 0.0, "step": 4700 }, { "epoch": 277.94, "learning_rate": 6.111111111111112e-07, "loss": 0.0, "step": 4725 }, { "epoch": 279.41, "learning_rate": 5.555555555555555e-07, "loss": 0.0, "step": 4750 }, { "epoch": 280.88, "learning_rate": 5.000000000000001e-07, "loss": 0.0, "step": 4775 }, { "epoch": 282.35, "learning_rate": 4.444444444444445e-07, "loss": 0.0, "step": 4800 }, { "epoch": 283.82, "learning_rate": 3.8888888888888895e-07, "loss": 0.0, "step": 4825 }, { "epoch": 285.29, "learning_rate": 3.3333333333333335e-07, "loss": 0.0, "step": 4850 }, { "epoch": 286.76, "learning_rate": 2.7777777777777776e-07, "loss": 0.0, "step": 4875 }, { "epoch": 288.24, "learning_rate": 2.2222222222222224e-07, "loss": 0.0, "step": 4900 }, { "epoch": 289.71, "learning_rate": 1.6666666666666668e-07, "loss": 0.0, "step": 4925 }, { "epoch": 291.18, "learning_rate": 1.1111111111111112e-07, "loss": 0.0, "step": 4950 }, { "epoch": 292.65, "learning_rate": 5.555555555555556e-08, "loss": 0.0, "step": 4975 }, { "epoch": 294.12, "learning_rate": 0.0, "loss": 0.0, "step": 5000 }, { "epoch": 294.12, "eval_loss": 0.5301908254623413, "eval_runtime": 68.1161, "eval_samples_per_second": 0.94, "eval_steps_per_second": 0.117, "eval_wer": 0.14732650739476677, "step": 5000 }, { "epoch": 294.12, "step": 5000, "total_flos": 2.717998645248e+20, "train_loss": 0.005774005842686165, "train_runtime": 28774.7965, "train_samples_per_second": 2.78, "train_steps_per_second": 0.174 } ], "logging_steps": 25, "max_steps": 5000, "num_input_tokens_seen": 0, "num_train_epochs": 295, "save_steps": 1000, "total_flos": 2.717998645248e+20, "train_batch_size": 8, "trial_name": null, "trial_params": null }